Embed Size (px)
Citation preview
ConstructionMaterialsThird edition
ConstructionMaterials
Their nature andbehaviour
Third edition
Edited by
J.M. Illston andP.L.J. Domone
London and New York
First published 2001 by Spon Press11 New Fetter Lane, London EC4P 4EE
Simultaneously published in the USA and Canadaby Spon Press29 West 35th Street, New York, NY 10001
Spon Press is an imprint of the Taylor & Francis Group
© 2001 Spon Press
All rights reserved. No part of this book may be reprinted or reproduced or utilised inany form or by any electronic, mechanical, or other means, now known or hereafterinvented, including photocopying and recording, or in any information storage orretrieval system, without permission in writing from the publishers.
The publisher makes no representation, express or implied, with regard to theaccuracy of the information contained in this book and cannot accept any legalresponsibility or liability for any errors or omissions that may be made.
British Library Cataloguing in Publication DataA catalogue record for this book is available from the British Library
Library of Congress Cataloging in Publication DataConstruction materials : their nature and behaviour/edited by J.M. Illston and P.L.J.Domone, – 3rd ed.
p. cm.Includes bibliographical references and index.1. Building materials. I. Illston, J.M. II. Domone, P.L.J.
TA403 .C636 2001624.1'8–dc21
2001020094
ISBN 0-419-25860-4 (pbk)ISBN 0-419-25850-7 (hb)
This edition published in the Taylor & Francis e-Library, 2002.
ISBN 0-203-47898-3 Master e-book ISBNISBN 0-203-78722-6 (Glassbook Format)
The biggest thing university taught me was thatwith ambition, perseverance and a book you cando anything you want to.
It doesn’t matter what the subject is; once you’velearnt how to study, you can do anything youwant.
George Laurer, inventor of the bar code
Contents
Contributors xv
Acknowledgements xviii
Preface P.L.J. Domone and J.M. Illston xix
Part One Fundamentals W.D. Biggs, revised and updated by I.R. McColl and J.R. Moon 1
Introduction 3
1 States of matter 51.1 Fluids 51.2 Solids 71.3 Intermediate behaviour 12
2 Energy and equilibrium 152.1 Mixing 162.2 Entropy 162.3 Free energy 172.4 Equilibrium and equilibrium diagrams 17
3 Atomic structure and interatomic bonding 253.1 Ionic bonding 263.2 Covalent bonding 273.3 Metallic bonding 283.4 Van der Waals bonds 29
4 Elasticity and plasticity 314.1 Linear elasticity 314.2 Consequences of the theory 324.3 Long-range elasticity 334.4 Viscoelasticity 344.5 Plasticity 37
5 Surfaces 395.1 Surface energy 39
v
5.2 Water of crystallisation 405.3 Wetting 405.4 Adhesives 415.5 Adsorption 42
6 Fracture and fatigue 446.1 Brittle fracture 456.2 Ductile fracture 466.3 Fracture mechanics 476.4 Fatigue 48
7 Electrical and thermal conductivity 49
Further reading 51
Part Two Metals and alloys W.D. Biggs, revised and updated by I.R. McColl and J.R. Moon 53
Introduction 55
8 Physical metallurgy 578.1 Grain structure 578.2 Crystal structures of metals 588.3 Solutions and compounds 60
9 Mechanical properties of metals 629.1 Stress–strain behaviour 629.2 Tensile strength 639.3 Ductility 649.4 Plasticity 649.5 Dislocation energy 669.6 Strengthening of metals 669.7 Unstable microstructures 68
10 Forming of metals 6910.1 Castings 6910.2 Hot working 7010.3 Cold working 7010.4 Joining 71
11 Oxidation and corrosion 7311.1 Dry oxidation 7311.2 Wet corrosion 7311.3 Control of corrosion 7611.4 Protection against corrosion 76
12 Metals, their differences and uses 7812.1 The extraction of iron 7812.2 Cast irons 78
Contents
vi
12.3 Steel 7912.4 Aluminium and alloys 8512.5 Copper and alloys 87
Further reading 88
Part Three Concrete P.L.J. Domone 89
Introduction 91
13 Portland cements 9513.1 Manufacture 9513.2 Physical properties 9613.3 Chemical composition 9713.4 Hydration 9813.5 Structure and strength of hardened cement paste 10413.6 Water in hcp and drying shrinkage 10513.7 Modifications of Portland cement 10613.8 Cement standards and nomenclature 10813.9 References 109
14 Admixtures 11014.1 Plasticisers 11014.2 Superplasticisers 11114.3 Accelerators 11214.4 Retarders 11314.5 Air entraining agents 11414.6 Classification of admixtures 11514.7 References 116
15 Cement replacement materials 11715.1 Pozzolanic behaviour 11715.2 Types of material 11815.3 Chemical composition and physical properties 11915.4 Supply and specification 120
16 Aggregates for concrete 12116.1 Types of aggregate 12116.2 Aggregate classification: shape and size 12316.3 Other properties of aggregates 12616.4 Reference 127
17 Properties of fresh concrete 12817.1 General behaviour 12817.2 Measurement of workability 12917.3 Factors affecting workability 132
Contents
vii
17.4 Loss of workability 13417.5 References 135
18 Early age properties of concrete 13618.1 Behaviour after placing 13618.2 Curing 13818.3 Strength gain and temperature effects 13918.4 References 142
19 Deformation of concrete 14319.1 Drying shrinkage 14319.2 Autogenous shrinkage 15019.3 Carbonation shrinkage 15019.4 Thermal expansion 15019.5 Stress–strain behaviour 15219.6 Creep 15619.7 References 159
20 Strength and failure of concrete 16120.1 Strength tests 16120.2 Factors influencing strength of Portland cement concrete 16520.3 Strength of concrete containing CRMs 17020.4 Cracking and fracture in concrete 17120.5 Strength under multiaxial loading 17320.6 References 175
21 Concrete mix design 17621.1 The mix design process 17621.2 UK method of ‘Design of normal concrete mixes’ 17821.3 Mix design with cement replacement materials (CRMs) 18121.4 Design of mixes containing admixtures 18221.5 References 183
22 Non-destructive testing of hardened concrete 18422.1 Surface hardness – rebound (or Schmidt) hammer test 18422.2 Resonant frequency test 18522.3 Ultrasonic pulse velocity test (upv) 18722.4 Near-to-surface tests 19022.5 References 191
23 Durability of concrete 19223.1 Transport mechanisms through concrete 19223.2 Measurements of flow constants for cement paste and concrete 19423.3 Degradation of concrete 19923.4 Durability of steel in concrete 21023.5 Recommendations for durable concrete construction 21523.6 References 216
Contents
viii
24 High performance concrete 21724.1 High strength concrete 21724.2 Self-compacting concrete 21924.3 References 220
Further reading 221
Part Four Bituminous materials D.G. Bonner 225
Introduction 227
25 Structure of bituminous materials 22925.1 Constituents of bituminous materials 22925.2 Bitumen 22925.3 Types of bitumen 23125.4 Aggregates 23425.5 Reference 236
26 Viscosity and deformation of bituminous materials 23726.1 Viscosity and rheology of binders 23726.2 Measurement of viscosity 23826.3 Influence of temperature on viscosity 23926.4 Resistance of bitumens to deformation 24026.5 Determination of permanent deformation 24126.6 Factors affecting permanent deformation 24226.7 References 243
27 Strength and failure of bituminous materials 24427.1 The road structure 24427.2 Modes of failure in a bituminous structure 24427.3 Fatigue characteristics 24627.4 References 250
28 Durability of bituminous materials 25128.1 Introduction 25128.2 Ageing of bitumen 25128.3 Permeability 25228.4 Adhesion 25328.5 References 257
29 Practice and processing of bituminous materials 25929.1 Bituminous mixtures 25929.2 Recipe and designed mixes 26229.3 Methods of production 26429.4 References 265
Further reading 266
Contents
ix
Part Five Brickwork and Blockwork R.C. de Vekey 267
Introduction 269
30 Materials and components for brickwork and blockwork 27330.1 Materials used for manufacture of units and mortars 27330.2 Other constituents and additives 27730.3 Mortar 27830.4 Fired clay bricks and blocks 28230.5 Calcium silicate units 29030.6 Concrete units 29030.7 Natural stone 29330.8 Ancillary devices – ties and other fixings/connectors 29430.9 References 294
31 Masonry construction and forms 29731.1 Introduction 29731.2 Mortar 29731.3 Walls and other masonry forms 29831.4 Bond patterns 29931.5 Specials 30131.6 Joint-style 30231.7 Workmanship and accuracy 30231.8 Buildability, site efficiency and productivity 30231.9 Appearance 30331.10 References 303
32 Structural behaviour and movement of masonry 30432.1 Introduction 30432.2 Compressive loading 30532.3 Shear load in the bed plane 31032.4 Flexure (bending) 31132.5 Tension 31332.6 Elastic modulus 31432.7 Movement and creep of masonry materials 31532.8 References 316
33 Durability and non-structural properties of masonry 31733.1 Introduction 31733.2 Durability 31733.3 Chemical attack 31833.4 Erosion 32133.5 Staining 32233.6 Thermal conductivity 32333.7 Rain resistance 32433.8 Sound transmission 32533.9 Fire resistance 325
Contents
x
33.10 Sustainability issues 32633.11 References 328
Further reading 329
Part Six Polymers L. Hollaway 333
Introduction 335
34 Polymers: types, properties and applications 33734.1 Polymeric materials 33734.2 Processing of thermoplastic polymers 33834.3 Polymer properties 33934.4 Applications and uses of polymers 34134.5 References 345
Part Seven Fibre composites 347
Introduction 349
Section 1 Polymer composites L. Hollaway 351Introduction 351
35 Fibres for polymer composites 35235.1 Fibre manufacture 35235.2 Fibre properties 35535.3 References 356
36 Analysis of the behaviour of polymer composites 35736.1 Characterisation and definition of composite materials 35736.2 Elastic properties of continuous unidirectional laminae 35836.3 Elastic properties of in-plane random long-fibre laminae 35936.4 Macro-analysis of stress distribution in a fibre/matrix composite 35936.5 Elastic properties of short-fibre composite materials 36036.6 Laminate theory 36036.7 Isotropic lamina 36236.8 Orthotropic lamina 36236.9 The strength characteristics and failure criteria of composite laminae 36436.10 References 368
37 Manufacturing techniques for polymer composites 36937.1 Manufacture of fibre-reinforced thermosetting composites 36937.2 Manufacture of fibre-reinforced thermoplastic composites 37437.3 References 374
Contents
xi
38 Durability and design of polymer composites 37538.1 Temperature 37538.2 Fire 37638.3 Moisture 37638.4 Solution and solvent action 37738.5 Weather 37738.6 Design with composites 37838.7 References 378
39 Uses of polymer composites 37939.1 Marine applications 38039.2 Applications in truck and automobile systems 38039.3 Aircraft, space and civil applications 38039.4 Pipes and tanks for chemicals 38039.5 Development of uses in civil engineering structures 38039.6 Composite bridges 38139.7 Retrofitting bonded composite plates to concrete beams 38139.8 Composite rebars 38439.9 References 384
Section 2 Fibre-reinforced cements and concrete D.J. Hannant 385
40 Properties of fibre and matrices 38640.1 Physical properties 38640.2 Structure of the fibre–matrix interface 388
41 Structure and post-cracking composite theory 38941.1 Theoretical stress–strain curves in uniaxial tension 38941.2 Uniaxial tension – fracture mechanics approach 39741.3 Principles of fibre reinforcement in flexure 39841.4 References 402
42 Fibre-reinforced cements 40342.1 Asbestos cement 40342.2 Glass-reinforced cement (GRC) 40542.3 Natural fibres in cement 40942.4 Polymer fibre-reinforced cement 41142.5 References 414
43 Fibre-reinforced concrete 41543.1 Steel fibre concrete 41543.2 Polypropylene fibre-reinforced concrete 41943.3 Glass fibre-reinforced concrete 42143.4 Reference 421
Further reading 422
Contents
xii
Part Eight Timber J.M. Dinwoodie 423
Introduction 425
44 Structure of timber and the presence of moisture 42744.1 Structure at the macroscopic level 42744.2 Structure at the microscopic level 42944.3 Molecular structure and ultrastructure 43444.4 Variability in structure 44244.5 Appearance of timber in relation to its structure 44344.6 Mass–volume relationships 44744.7 Moisture in timber 45044.8 Flow in timber 45444.9 References 461
45 Deformation in timber 46345.1 Introduction 46345.2 Dimensional change due to moisture 46345.3 Thermal movement 46645.4 Deformation under load 46845.5 References 486
46 Strength and failure in timber 48946.1 Introduction 48946.2 Determination of strength 48946.3 Strength values 49146.4 Variability in strength values 49146.5 Inter-relationships among the strength properties 49446.6 Factors affecting strength 49446.7 Strength, toughness, failure and fracture morphology 50346.8 Design stresses for timber 51046.9 References 513
47 Durability of timber 51547.1 Introduction 51547.2 Chemical, physical and mechanical agencies affecting durability and
causing degradation 51547.3 Natural durability and attack by fungi and insects 51747.4 Performance of timber in fire 52047.5 References 523
48 Processing of timber 52548.1 Introduction 52548.2 Mechanical processing 52548.3 Chemical processing 53448.4 Finishes 53948.5 References 541
Further reading 542
Index 543
Contents
xiii
Contributors
Professor D.G. BonnerDepartment of Aerospace, Civil andEnvironmental EngineeringUniversity of HertfordshireHatfield CampusCollege LaneHatfieldHerts AL10 9AB(Bituminous materials)
Professor J.M. Dinwoodie OBE16 Stratton RoadPrinces RisboroughNr AylesburyBucks HP17 9BH(Timber)
Dr P.L.J. DomoneDepartment of Civil and EnvironmentalEngineeringUniversity College LondonGower Street, London WC1E 6BT(Editor and concrete)
Professor D.J. HannantDepartment of Civil EngineeringUniversity of SurreyGuildfordSurrey GU2 5XH(Fibre reinforced cements and concrete)
Professor L. HollawayDepartment of Civil EngineeringUniversity of SurreyGuildfordSurrey GU2 5XH(Polymers and polymer composites)
Professor J.M. Illston10 Merrifield RoadFordSalisburyWiltshire SP4 6DF(Previous editor)
Dr I.R. McCollSchool of Mechanical, Materials, ManufacturingEngineering and ManagementUniversity of NottinghamUniversity ParkNottingham NG7 2RD(Fundamentals and metals)
Dr J.R. MoonSchool of Mechanical, Materials, ManufacturingEngineering and ManagementUniversity of NottinghamUniversity ParkNottingham NG7 2RD(Fundamentals and metals)
Dr R.C. de VekeyBuilding Research EstablishmentGarstonWatfordHerts WD2 7JR(Brickwork and blockwork)
xv
D.G. BonnerProfessor David Bonner graduated in Civil Engi-neering from the University of Leeds where he latergained a PhD in Traffic Engineering. Following aperiod working in the Highways Department ofLincolnshire County Council, he joined the Uni-versity of Hertfordshire (then Hatfield Polytech-nic) where he became Reader in ConstructionMaterials. He subsequently became Head of CivilEngineering and is presently Associate Dean(Academic Quality) for the Faculty of Engi-neering and Information Sciences.
J.M. DinwoodieProfessor John Dinwoodie graduated in Forestryfrom Aberdeen University, and was subsequentlyawarded his MTech in Non-Metallic Materialsfrom Brunel University, and both his PhD andDSc in Wood Science subjects from AberdeenUniversity. He carried out research at the UKBuilding Research Establishment for a period of35 years on timber and wood-based panels with aspecial interest in the rheological behaviour ofthese materials. For this work he was awardedwith a special merit promotion to Senior Prin-cipal Scientific Officer. Since his retirement fromBRE in 1995, he has been employed as a consul-tant to BRE to represent the UK in the prepara-tion of European standards for wood-basedpanels. In 1985 he was awarded the Sir StuartMallinson Gold Medal for research on creep inparticleboard and was for many years a Fellow ofthe Royal Microscopical Society. In 1994 he wasappointed an Honorary Professor in the Depart-ment of Forest Sciences, University of Wales,Bangor, and in the same year was awarded anOBE. He is author, or co-author, of over onehundred and thirty technical papers and authorof three text books on wood science and techno-logy.
P.L.J. DomoneDr Peter Domone graduated in civil engineeringfrom University College London, where he sub-
Contributors
xvi
sequently completed a PhD in concrete techno-logy. After a period in industrial research withTaylor Woodrow Construction Ltd, he wasappointed to the academic staff at UCL, first aslecturer and then as senior lecturer in concretetechnology. He teaches all aspects of civil engi-neering materials to undergraduate students, andhis principle research interests have included non-destructive testing, the rheology of fresh concrete,high-strength concrete and most recently, self-compacting concrete.
D.J. HannantProfessor David Hannant is a Professor in Con-struction Materials at the University of Surreyand has been researching and teaching in the fieldof fibre-reinforced cement and concrete since1968. He has authored a book, patents and manypublications on steel, glass and polypropylenefibres and has been active in commercialising thinsheet products to replace asbestos cement.
L.C. HollawayProfessor Len Hollaway is Professor of Compos-ite Structures at the University of Surrey and hasbeen engaged in the research and teaching ofcomposites for more than 30 years. He is theauthor and editor of a number of books and haswritten many research papers on the structuraland material aspects of fibre matrix composites.
J.M. IllstonProfessor John Illston spent the early part of hiscareer as a practising civil engineer before enter-ing higher education. He was involved in teachingand researching into concrete technology andstructural engineering as Lecturer and Reader atKing’s College, London and then Head ofDepartment of Civil Engineering at Hatfield Poly-technic. His interest in his discipline took secondplace when he became Director of the Polytech-nic, but, on retirement he undertook preparationof the second edition of this book. He has actedas passive editor of this edition.
I.R. McCollDr Ian McColl is a senior lecturer at the Univer-sity of Nottingham. His first degree is in physicsand after receiving his PhD from Nottingham hewas involved in industrially sponsored researchand development work at the university beforetaking up a lecturing post in 1988. He teachesmainly in the areas of engineering materials andengineering design. His research interests centrearound the fretting and fatigue properties ofengineering materials, components and assem-blies, and the use of surface engineering toimprove these properties.
J.R. MoonDr Bob Moon is a metallurgist married to thefirst woman to study civil engineering at the Uni-versity of Nottingham. His PhD was awarded bythe University of Wales (Cardiff) in 1960. He hasworked in the steel industry in South Wales,researched into titanium and other new metals atIMI in Birmingham and into superconductors,magnetic materials and materials for steam tur-bines at C.A. Parsons on Tyneside. He joined thestaff of the University of Nottingham over 30
Contributors
xvii
years ago and has taught materials to civil engi-neers for the majority of that time. He is nowreader in materials science and researches into thetriangle connecting processing, microstructureand properties of materials made from powders.
R.C. de VekeyDr Bob de Vekey studied chemistry at HatfieldPolytechnic and graduated with the Royal Societyof Chemistry. He subsequently gained a DIC inmaterials science and a PhD from ImperialCollege, London on the results of his work at theBuilding Research Establishment (BRE) on mater-ials. At BRE he has worked on many aspects ofbuilding materials research and development andbetween 1978 and 2000 led a section concernedmainly with structural behaviour, testing, dura-bility and safety of brick and block masonrybuildings. From September 2000 he has relin-quished his previous role and become an associ-ate to the BRE. He has written many papers andadvisory publications and has contributed toseveral books on building materials and masonryand has been involved in the development of UK and international standards and codes forbuilding.
xviii
Acknowledgements
I first of all want to express my thanks to JohnIllston for his tremendous work and vision thatresulted in the first two editions of this book, andfor his encouragement and helpful advice in theearly stages of preparing this edition. My thanksalso go to all the contributors who have willinglyrevised and updated their text despite many othercommitments. They have all done an excellentjob, and any shortfalls in the book are entirelymy responsibility. I greatly appreciate the adviceand inspiration provided by my students at UCLand my colleagues at UCL and elsewhere, whohave suffered due to my neglect of other dutieswhilst preparing and editing the manuscript.
Finally, but most importantly, I must acknow-ledge the support given to me by my wife andchildren, who I have neglected most of all, butwho have borne with good grace the many hoursI have spent in my study.
P.L.J. Domone
I wish to express my appreciation to the BuildingResearch Establishment (BRE) and in particularto Dr A.F. Bravery, Director of the Centre forTimber Technology and Construction, not only
for permission to use many plates and figuresfrom the BRE collection, but also for providinglaboratory support for the production of thefigures from both existing negatives and fromnew material.
Thanks are also due to several publishers forpermission to reproduce figures.
To the many colleagues who have so willinglyhelped me in some form or other in the produc-tion of this revised text I would like to record myvery grateful thanks. In particular, I would like torecord my appreciation to Dr P.W. Bonfield andDr Hilary Derbyshire, BRE; Dr J.A. Petty, Uni-versity of Aberdeen; Dr D.G. Hunt, University ofthe South Bank for all their valuable and helpfuladvice. I would also like to thank colleagues atBRE for assistance on specific topics: C.A.Benham, J. Boxall, Dr J.K. Carey, Dr V. Enjily,C. Holland, J.S. Mundy, Dr R.J. Orsler, J.F.Russell, E.D. Suttie and P.P. White. Lastly, mydeep appreciation to both my daughter, who didmuch of the word processing, and my wife forher willing assistance in editing my text and inproof reading.
J.M. Dinwoodie
Preface
This book is an updated and extended version ofthe second edition, which was published in 1994.This has been extremely popular and success-ful, but the continuing recent advances in manyareas of construction materials technology haveresulted in the need for this new edition.
The first edition was published under the titleConcrete, Timber and Metals in 1979. Its scope,content and form were significantly changed forthe second edition, with the addition of threefurther materials – bituminous materials,masonry and fibre composites – with a separatepart of the book devoted to each material,following a general introductory part on ‘Funda-mentals’.
This overall format has been retained. One newsmall part has been added, on polymers, whichwas previously subsumed within the section onpolymer composites. The other significantchanges are, first, in the section on concrete,where Portland cement, admixtures, cementreplacement materials and aggregates now havetheir own chapters, and new chapters on mixdesign, non-destructive testing and high perform-ance concrete have been added; and, second, inthe section on fibre reinforced cement and con-crete, which has been rearranged so that eachtype of composite is considered in full in turn.
All of the contributors to the second editionwere able and willing to contribute again, withtwo exceptions. First, the co-author of the firstedition and editor and inspiration for the secondedition, John Illston, is enjoying a well-earnedretirement from all professional engineering andacademic activities, and did not wish to continueas editor for this edition. This role was taken overby Peter Domone, with considerable apprehen-
sion about following such an illustrious predeces-sor and the magnitude of the task. Fortunately,John Illston provided much needed encourage-ment, advice and comments, particularly in theearly stages,
Second, one of the contributors, Bill Biggs, hadsadly died in the intervening period, but two newcontributors have revised, extended and updatedhis ‘Fundamentals’ and ‘Metals’ sections. Themost significant addition is a consideration ofequilibrium phase diagrams.
Objectives and scopeAs before, the book is addressed primarily to stu-dents taking courses in civil or structural engi-neering, where there is a continuing need for theunified treatment of the kind that we have againattempted. We believe that the book providesmost if not all of the information required by stu-dents for at least the first two years of three- orfour-year degree programmes. More specialistproject work in the third or fourth years mayrequire recourse to the more detailed texts thatare listed in ‘Further reading’ at the end of eachsection. We also believe that our approach willcontinue to provide a valuable source of interestand stimulation to both undergraduates andgraduates in engineering generally, materialsscience, building, architecture and related disci-plines.
The objective of developing an understandingof the behaviour of materials from a knowledgeof their structure remains paramount. Only inthis way can information from mechanicaltesting, experience in processing, handling andplacing, and materials science, i.e. empiricism,
xix
craft and science, be brought together to give thesound foundation for materials technologyrequired by the practitioner.
The ‘Fundamentals’ section provides the neces-sary basis for this. Within each of the subsequentsections on individual materials, their structureand composition from the molecular levelupwards is discussed, and then the topics ofdeformation, strength and failure, durability, andmanufacture and processing are considered. Acompletely unified treatment for each material isnot possible due to their different nature and thedifferent requirements for manufacture, process-ing and handling, but a look at the contents listwill show how each topic has been covered andhow the materials can be compared and con-trasted. Cross-references are given throughout thetext to aid this, from which it will also be appar-ent that there are several cases of overlap betweenmaterials, for example concrete and bituminouscomposites use similar aggregates, and Portlandcement is a component of masonry, some fibrecomposites and concrete.
It is impossible in a single book to cover thefield of construction materials in a fully compre-hensive manner. Not all such materials of con-struction are included, nor has the attempt beenmade to introduce design criteria or to provide acompendium of materials data. Neither is thisbook a manual of good practice. Nevertheless wehope that we have provided a firm foundation forthe application and practice of materials techno-logy.
Levels of informationThe structure of materials can be described ondimensional scales varying from the smallest,atomic or molecular, through materials structuralto the largest, engineering.
The molecular level
This considers the material at the smallest scale,in terms of atoms or molecules or aggregations ofmolecules. It is very much the realm of materialsscience and a general introduction for all mater-
Preface
xx
ials is given in Part One of the book. The sizes ofthe particles are in the range of 10�7 to 10�3 mm.Examples occurring in this book include thecrystal structure of metals, cellulose molecules intimber, calcium silicate hydrates in hardenedcement paste and the variety of polymers, such aspolyvinyl chloride, included in fibre composites.
As shown in Part One consideration of estab-lished atomic models leads to useful descriptionsof the forms of physical structure, both regularand disordered, and of the ways in which mater-ials are held together. Chemical composition is offundamental importance in determining thisstructure. This may develop with time as chemi-cal reactions continue; for example, the hydrationof cement is a very slow process and the structureand properties show correspondingly significantchanges with time. Chemical composition is ofespecial significance for durability which is oftendetermined, as in the cases of timber and metals,by the rate at which external substances such asoxygen or acids react with the chemicals of whichthe material is made.
Chemical and physical factors also cometogether in determining whether or not the mater-ial is porous, and what degree of porosity ispresent. In materials such as bricks, timber andconcrete, important properties such as strengthand rigidity are inversely related to their porosi-ties. Similarly, there is often a direct connectionbetween permeability and porosity.
Some structural phenomena such as dislocationsin metals are directly observable by microscopicand diffractometer techniques, but more oftenmathematical and geometrical models areemployed to deduce both the structure of thematerial and the way in which it is likely tobehave. Some engineering analyses, like fracturemechanics, come straight from molecular scaleconsiderations, but they are the exception. Muchmore often the information from the molecularlevel serves to provide mental pictures which aidthe engineers’ understanding so that they candeduce likely behaviour under anticipated con-ditions. In the hands of specialists, knowledge ofthe chemical and physical structure may well offera route to the development of better materials.
Materials structural level
This level is a step up in size from the molecularlevel, and the material is considered as a compos-ite of different phases, which interact to realisethe behaviour of the total material. This may be amatter of separately identifiable entities withinthe material structure as in cells in timber orgrains in metals; alternatively, it may result fromthe deliberate mixing of disparate parts, in arandom manner in concrete or asphalt or somefibre composites, or in a regular way in masonry.Often the material consists of particles such asaggregates distributed in a matrix like hydratedcement or bitumen. The dimensions of the parti-cles differ enormously from the wall thickness ofa wood cell at 5�10�3 mm to the length of abrick at 225mm. Size itself is not an issue; whatmatters is that the individual phases can be recog-nised independently.
The significance of the materials structural levellies in the possibility of developing a more generaltreatment of the materials than is provided fromknowledge derived from examination of the totalmaterial. The behaviours of the individual phasescan be combined in the form of multiphasemodels which allow the prediction of behaviouroutside the range of normal experimental obser-vation. In formulating the models considerationmust be given to three aspects:
1. Geometry: the shape, size and concentrationof the particles and their distribution in thematrix or continuous phase.
2. State and properties: the chemical and physi-cal states and properties of the individualphases influence the structure and behaviourof the total material.
3. Interfacial effects. The information under (1)and (2) may not be sufficient because theinterfaces between the phases may introduceadditional modes of behaviour that cannot berelated to the individual properties of thephases. This is especially true of strength, thebreakdown of the material often being con-trolled by the bond strength at an interface.
To operate at the materials structural level
Preface
xxi
requires a considerable knowledge of the threeaspects described above. This must he derivedfrom testing the phases themselves, and addition-ally from interface tests. While the use of themultiphase models is often confined to research inthe interest of improving understanding, it issometimes taken through into practice, albeitmostly in simplified form. Examples include theestimation of the elastic modulus of concrete, andthe strength of fibre composites.
The engineering level
At the engineering level the total material is con-sidered; it is normally taken as continuous andhomogeneous and average properties are assumedthroughout the whole volume of the materialbody. The materials at this level are thosetraditionally recognised by construction practi-tioners, and it is the behaviour of these materialsthat is the endpoint of this book.
The minimum scale that must be considered isgoverned by the size of the representative cell,which is the minimum volume of the materialthat represents the entire material system, includ-ing its regions of disorder. The linear dimensionsof this cell varies considerably from, say, 10�3 mmfor metals to 100mm for concrete and 1000mmfor masonry. Properties measured over volumesgreater than the unit cell can be taken to apply tothe material at large. If the properties are thesame in all directions then the material isisotropic and the representative cell is a cube,while if the properties can only be described withreference to orientation, the material isanisotropic, and the representative cell may beregarded as a parallelepiped.
Most of the technical information on materialsused in practice comes from tests on specimens ofthe total material, which are prepared to repre-sent the condition of the material in the engin-eering structure. The range of tests, which can beidentified under the headings used throughoutthis book, includes strength and failure, deforma-tion, and durability. The test data is often pre-sented either in graphical or mathematical form,but the graphs and equations may neither express
the physical and chemical processes within thematerials, nor provide a high order of accuracy ofprediction. However, the graphs or equationsusually give an indication of how the propertyvalues are affected by significant variables; suchas the carbon content of steel, the moisturecontent of timber, the fibre content and orienta-tion in composites or the temperature of asphalt.It is extremely important to recognise that thequality of information is satisfactory only withinthe ranges of the variables used in the tests.Extrapolation beyond those ranges is very risky;this is a common mistake made not only by stu-dents, but also often by more experienced engi-neers and technologists who should know better.
Comparability and variabilityThroughout this book we have tried to excitecomparison of one material with another. Atten-tion has been given to the structure of eachmaterial, and although a similar scientific founda-tion applies to all, the variety of physical andchemical compositions gives rise to great differ-ences in behaviour. The differences are carriedthrough to the engineering level of informationand have to be considered by practitionersengaged in the design of structures who first haveto select which material(s) to use, and then en-sure that they are used efficiently, safely andeconomically.
Selection of materials
The engineer must consider the fitness of thematerial for the purpose of the structure beingdesigned. This essential fitness-for-purpose is amatter of ensuring that the material will performadequately both during construction and in sub-sequent service. Strength, deformation and dura-bility are likely to be the principal criteria thatmust be satisfied, but other aspects of behaviourwill be important for particular applications, forexample water-tightness or speed of construction.In addition, aesthetics and environmental impactshould not be forgotten.
Table 0.1 gives some properties of a number of
Preface
xxii
individual and groups of materials. These aremainly structural materials, with some othersadded for comparison. The mechanical propertieslisted – stiffness, strength (or limiting stress) andwork to fracture (or toughness) – are all definedand discussed in this book. It is immediatelyapparent that there is a wide variety of eachproperty that the engineer can select from (orcope with, if it is not ideal). It is also interestingto note the overall range of each property.Density varies by about two orders of magnitudefrom the least to the most dense (timber tometals). Stiffness varies by nearly three orders ofmagnitude (nylon to diamond), strength by aboutfour orders (concrete to diamond) and work tofracture by the greatest of all, five orders (glass toductile metals). The great range of the last prop-erty is perhaps the most significant of all. It is ameasure of how easy it is to break a material,particularly under impact loading, and how wellit copes with minor flaws, cracks, etc.; it shouldnot be confused with strength. Low values areextremely difficult for engineers to deal with –low strength and stiffness can be accommodatedby bigger section sizes and structural arrange-ments (within limits), but low toughness is muchmore difficult to handle. It is one reason whyfibre composites have become so popular.
Clearly, in many circumstances more than onematerial satisfies the criterion of fitness-for-purpose; for instance, members carrying tensioncan be made of steel or timber, facing panels canbe fabricated from fibre composite, metal, timberor masonry. The matter may be resolved by theengineer making a choice based on his or herjudgement, with often some help from calcula-tions. For example, comparisons of minimumweight or minimum cost options for a simplestructure with different materials, obtained withsome fairly elementary structural mechanics, withdifferent materials gives some interesting results.
Consider the cantilever shown in Figure 0.1.For linear elastic behaviour, the deflection at thefree end is given by:
� �Fl3/3EI
where E is the elastic modulus of the material and
TA
BL
E 0
.1Se
lect
ed p
rope
rtie
s of
a r
ange
of
mat
eria
ls
Mat
eria
lD
ensi
tySt
iffn
ess
(E)
Stre
ngth
or
limit
ing
Wor
k to
fra
ctur
eR
elat
ive
cost
(£)
(ton
ne/m
3 )(G
Pa)
stre
ss (
MP
a)*
(tou
ghne
ss)
(kJ/
m2 )
per
unit
:m
ass
volu
me
Dia
mon
d3.
510
0050
000
–2
�10
68.
9�
105
Com
mon
pur
e m
etal
s5–
1920
–200
20–8
010
0–10
00–
–St
ruct
ural
ste
el7.
8519
5–20
523
5–45
010
0–13
01.
01.
0H
igh
stre
ngth
ste
el7.
8520
526
0–13
0015
–120
––
Cas
t ir
on6.
9–7.
817
0–19
022
0–10
000.
2–0.
3–
–Si
lica
glas
s2.
694
50–2
000.
013.
41.
1T
itan
ium
and
allo
ys4.
580
–130
180–
1320
25–1
1527
.516
Alu
min
ium
and
allo
ys2.
769
–79
40–6
308–
305.
01.
7T
imbe
r0.
17–0
.98
0.6–
1.0
perp
gra
in90
–200
(te
ns)
8–20
Cra
ck p
erp
grai
n–
–(d
ry)
9–16
par
gra
in15
–90
(com
p)**
0.5–
2 cr
ack
par
grai
nSp
ruce
(pa
r. t
o gr
ain)
0.5
1340
–80
0.5–
11.
00.
06C
oncr
ete
1.8–
2.5
20–4
54–
10 (
tens
)0.
030.
70.
1220
–150
(co
mp)
Epo
xy r
esin
1.1–
1.4
2.6–
330
–100
0.1–
0.3
3.8
0.53
Gla
ss fi
bre
com
p (g
rp)
1.4–
2.2
35–4
510
0–30
010
–100
––
Car
bon
fibre
com
posi
te18
0–20
060
0–70
05–
30–
–N
ylon
1.1–
1.2
2–4
50–9
02–
47.
51.
1R
ubbe
r0.
95–1
.15
2–10
15–3
0–
––
*in
tens
ion
unle
ss s
tate
d; y
ield
str
ess
for
met
als,
oth
erw
ise
ulti
mat
e st
ress
.**
on c
lear
spe
cim
ens.
Preface
xxiv
l
F
FIGURE 0.1 A simple cantilever.
I is the second moment of area. For a rectangularbeam I�bd3/12, where b is the width of thebeam and d is its depth. Usually, the width is adesign constant, e.g. a bridge for four carriage-ways, but the depth can be altered to meet theloading and deflection requirements. The depthis:
d� (4Fl3/Eb�)1/3
The weight is:
W� �lA� �lbd� �lb(4Fl3/Eb�)1/3
where � is the density.Rearranging, we have:
W� l 2b2/3(4F/�)1/3(�/E1/3) (0.1)
so that, for a given set of design conditions (F, l,b, �), W is minimised by maximising E1/3/�. Itturns out that the same condition applies for anytype of beam, of any shape or of any loading con-figuration. The parameter E1/3/� is therefore theselection criterion for stiffness at minimumweight.
A similar approach can be taken to derive aselection criterion for the strength of the beam.For the cantilever in Figure 0.1, the maximumtension stress generated is:
�max �6Fl/bd 2
Following the argument through as before gives:
W� l3/2b1/2(6F)1/2(�/�1l/i2mit) (0.2)
So, for strength at minimum weight, we want tomaximise �1
l/i2mit /�. Note the use of �limit, the
maximum useful stress that the material can tol-erate in tension. In defining this, we must takeinto account the variability of materials, dis-cussed below.
The cost of a beam is simply W£M, where £M isthe cost of the material per unit mass. Workingthe arguments through again give criteria forstiffness and for strength at minimum cost. Theseturn out to be similar to those used before, i.e.E1/3/£V and �1/2/£V, where £V is the material costper unit volume�density�cost per unit mass.
Table 0.2 shows how these four factors relativeto those for structural steel work out for some ofthe materials listed in Table 0.1. Where appropri-ate the mid-range properties have been used,together with the prices given in Table 0.2.Remembering that we are looking for maximumvalues, the most obvious result is that timber isoutstanding in terms of stiffness and strength atboth minimum weight and minimum cost –nature clearly got things right without any helpfrom us. Thinking only about weight and neglect-ing cost points to the efficiency of diamond (not arealistic option!), titanium, aluminium, epoxyresin and nylon. But when it comes to minimisingcosts, the highest scoring materials other thantimber are steel and concrete.
Table 0.2 does not, of course, give a completepicture. Prices will vary from time to time, andother properties, such as ease of construction,durability and toughness have not been takeninto account. Also, composite systems, such asreinforced concrete and fibre reinforced systemshave not been considered. Nevertheless, it doesgive food for thought.
The cost of the energy used in manufacturingthe material and its transport and fabricationmust also not be overlooked, either in simple eco-nomic or in environmental terms. High-strengthversions of all the materials covered in this bookare often sought with the intention of reducingthe volume of material for a given structure, nor-mally with a commensurate saving in energy;and, conversely, lower grade materials are some-times introduced to replace higher grades as, forexample, in the partial replacement of cement bythe waste material, pulverised fuel ash (fly ash).
Variability and characteristic strength
An important issue facing engineers is the vari-ability of the properties of the material itself,which clearly depends on the uniformity of itsstructure and composition.
The strength or maximum allowable stress is ofparticular concern. In tension, for ductile mater-ials this can be taken as the yield strength or theproof stress (see Chapter 9), but for brittle mater-ials it may have to be chosen on the basis of theapproach described in Chapter 6. In compression,brittle materials have a well defined maximumstress, but ductile materials may not fail at all,they just continue being squashed.
The variability can be assessed by a series oftests on nominally similar specimens from eitherthe same or successive batches of a material; thisusually gives a variation of strength or maximumstress with a normal or Gaussian distribution asshown in Figure 0.2. This can be represented bythe bell-shaped mean curve given by the equation
y � exp �� � (0.3)
where y is the probability density, and � is thevariate, in this case the strength. The strengthresults are therefore expressed in terms of twonumbers:
1. the mean strength, �m, where for n results:
�m � ��/n (0.4)
(���m)2
�2s2
1�s�2��
Preface
xxv
Strength �N
umbe
r of
res
ults
or
prob
abili
ty d
ensi
ty�m
FIGURE 0.2 Histogram and normal distribution ofstrength.
2. the range or variability, expressed as the stan-dard deviation, s, given by
s2 � �[(� � �m)2/(n�1)] (0.5)
The standard deviation has the same units as thevariate.
For comparison of different materials or differentkinds of the same material the non-dimensionalcoefficient of variation cv is used, given by
cv � s/�m (0.6)
cv is often expressed as a percentage.
TABLE 0.2 Weight and cost comparisons for the use of alternative materials for the cantilever of Figure 0.1(all figures relative to mild steel�1, material properties as in Table 0.1)
Material Cost Minimum weight Minimum cost(£/tonne) Stiffness criterion Strength criterion Stiffness criterion Strength criterion
E1/3/� �max1/2/� E1/3/� .£/tonne �max
1/2/� .£/tonne
Diamond 2�106 3.8 31.8 1.9�10�6 1.6�10�5
Structural steel 1.0 1.0 1.0 1.0 1.0Silica glass 3.4 2.4 2.1 0.8 0.7Titanium and alloys 27.5 1.4 3.0 0.05 0.1Aluminium and alloys 5.0 2.1 3.4 0.4 0.7Spruce (par. to grain) 1.0 6.4 7.7 6.4 7.7Concrete 0.7 1.8 0.6 2.6 0.8Epoxy resin 3.8 1.5 3.2 0.4 0.8Nylon 7.5 1.7 3.6 0.2 0.5
Preface
xxvi
Values of typical mean strengths and their coeffi-cients of variation for materials in this book aregiven in Table 0.3. Steel and ungraded timber atthe two ends of the scale. The manufacture of steelis a well developed and closely controlled processso that a particular steel can be readily reproducedand the variability of properties such as strength issmall; conversely, ungraded or unprocessed timber,which in its natural form is full of defects such asknots, and inevitably exhibits a wide variation inproperty values. The variability can, however, bereduced by processing so that the coefficient of
TABLE 0.3 Comparison of strengths of construction materials and their coefficients of variation. c�compres-sion, t� tension
Material Mean strength Coefficient of CommentMPa variation (%)
Steel 460t 2 Structural mild steelConcrete 40c 15 Typical concrete cube strength at 28 daysTimber 30t 35 Ungraded softwood
120t 18 Knot free, straight grained softwood11t 10 Structural grade chipboard
Fibre cement composites 18t 10 Continuous polypropylene fibre with 6% volume fraction in stress direction
Masonry 20c 10 Small walls, brick on bed
Area �% failure
� failure rate
Strength �Margin� ks
Num
ber
of r
esul
ts
Characteristicstrength
�m
FIGURE 0.3 Failure rate, margin and characteristicstrength.
1
10
100
0 0.5 1 1.5 2 2.5k
Failu
re r
ate
(%)
FIGURE 0.4 Relationship between k and failurerate for normally distributed results
variation for chipboard or fibreboard is appreciablylower than that for ungraded timber.
Engineers need to take both the mean strengthand the strength variation into account in deter-mining a ‘safe’ strength at which the possibility offailure is reduced to acceptable levels. If the meanstress alone is used, then by definition half of thematerial will fail to meet this criterion, which isclearly unacceptable. The statistical nature of thedistribution of results means that a minimumstrength below which no specimen will ever failcannot be defined, and therefore a value knownas the characteristic strength (�char) is used, whichis set at a distance below the mean, called the
Preface
xxvii
margin, below which an acceptably small numberof results will fall.
The total area under the normal distributioncurve of Figure 0.2 represents 100 per cent of theresults, and the area below any particularstrength is the number of results that will occurbelow that strength, or in other words the failurerate (Figure 0.3). The greater the margin, thelower the failure rate. Clearly, if the margin iszero, then the failure rate is 50 per cent. It is aproperty of the normal distribution curve that ifthe value of the margin is expressed as ks where kis a constant, then k is related to the failure rateas in Figure 0.4. Hence
�char � �m �ks (0.7)
Engineering judgement and consensus is used tochoose an acceptable failure rate. In practice, thisis not always the same in all circumstances; forexample, 5 per cent is typical for concrete(i.e. k�1.64), and 2 per cent for timber (i.e.k�1.96).
As we have said, this analysis uses values of sobtained from testing prepared specimens of thematerial concerned. In practice it is also necessary
to consider likely differences between these andthe bulk properties of material, which may varydue to size effects, manufacturing inconsistencies,etc., and therefore further materials safety factorswill need to be applied to the characteristicstrength. It is beyond the scope of this book toconsider these in any detail – guidance can beobtained from relevant design codes and designtext books.
Concluding remarksWe hope that this preface has encouraged you toread on. It has described the general content,nature and approach of the book, and sets thescene for the level and type of information on thevarious materials that is provided. It has alsogiven an introduction to some ways of comparingmaterials, and discussed how the unavoidablevariability of properties can be taken into accountby engineers. You will probably find it useful torefer back to these latter two sections from timeto time.
Enjoy the book.
A note on unitsIn common with all international publications, and with national practice in many countries, the SIsystem of units has been used throughout this text. Practice does however vary between different partsof the engineering profession and between individuals over whether to express quantities which havethe dimensions of [force]/[length]2 in the units of its constituent parts, e.g. N/m2, or with the interna-tionally recognised combined unit of the Pascal (Pa). In this book, the latter is used in Parts 1 to 7, andthe former in Part 8, which reflects the general practice in other publications on the materials con-cerned.
The following relationships may be useful whilst reading:1 Pa � 1 N/m2
1kPa � 103 Pa � 103 N/m2 � 1 kN/m2
1MPa � 106 Pa � 106 N/m2 � 1 N/mm2
1GPa � 109 Pa � 109 N/m2 � 1 kN/mm2
The magnitude of the unit for a particular property is normally chosen such that convenient numbersare obtained e.g. MPa (or N/mm2) for strength and GPa (or kN/mm2) for modulus of elasticity ofstructural materials.
Part One
Fundamentals
W.D. Biggs
Revised and updatedby I.R. McColl andJ.R. Moon
Introduction
As engineers our job is to design, but any designremains just that and no more until we start touse materials to convert it into a working arte-fact. There are, basically, three things we need toknow about materials:
1. How do they behave in service?2. Why do they behave in the way that they do?3. Is there anything we can do to alter their
behaviour?
This first section of the book is primarily con-cerned with (2). We include consideration of thefundamental elements of which all matter is com-posed and the forces which hold it together. Theconcept of ‘atomistics’ is not new. The ancientGreeks – and especially Democritus (c.460 BC) –had the idea of a single elementary particle buttheir science did not extend to observation andexperiment. For that we had to wait nearly 22centuries until Dalton, Avogadro and Cannizzaroformulated atomic theory as we know it today.Even so, very many mysteries remain unresolved,a fact which is as pleasing as it is provoking. Soin treating the subject in this way we are reachinga long way back into the development of thoughtabout the universe and the way in which it is puttogether.
One other important concept is more recent.Engineering is much concerned with change – thechange from the unloaded to the loaded state, theconsequences of changing temperature, environ-ment, etc. The first scientific studies of change canbe attributed to Sadi Carnot (1824), laterextended by such giants as Clausius, Joule andothers to produce such ideas as the conservationof energy, momentum, etc. Since the early studieswere carried out on heat engines it became
known as the science of thermodynamics, but ifwe take a broader view it is really the art andscience of managing, controlling and using thetransfer of energy – whether the energy of theatom, the energy of the tides or the energy of,say, a lifting rig.
In many engineering courses thermodynamicsis treated as a separate topic, or, in many civilengineering courses, not considered at all. But,because its applications set rules which no engi-neer can ignore, a brief discussion is included.What are these rules? In summary (and ratherjocularly) they are:
1. You cannot win, i.e. you cannot get more outof a system than you put in.
2. You cannot break even – in any change some-thing will be lost or, to be more precise, it willbe useless for the purpose you have in mind.
All this may be unfamiliar ground so you may skippast the sections relating to this on first reading.But come back to it because it is important.
We shall not, in the present part, deal specifi-cally with item (3) above; later parts will dealwith this in more specific terms. But what everyengineer should remember is that engineering isall about compromise and trade-off. Some pro-perties can be varied – strength is one such – butsome, such as density, cannot be varied. If wewere designing aircraft we could, in principle,choose between maximum strength and minimumweight, and minimum weight would win becauseit would ensure a higher payload (we must notforget that engineering is also about money*).
3
*An engineer has been defined as a person who can do for£100 what any fool could do for £1000.
But, of course, compromise must be sought andengineering is about finding optimal solutions,not necessarily a ‘best’ solution.
So good luck with your reading. If you reallyunderstand the principles much of what followswill be clearer to you. But, in a few short chapters
we can do little more than describe the highlightsof materials science, and a list of suggestions forfurther reading is given at the end of this part.
We should add a simple philosophy – engi-neering is much too important to be left toengineers.
Introduction
4
Chapter one
States of matter
1.1 Fluids1.2 Solids1.3 Intermediate behaviour
We conventionally think of matter (material) asbeing in one of two states, fluid or solid. Werecognise these ‘states of matter’ based on theresponse of the matter to an applied force. Fluids,i.e. gases and liquids, flow easily when a force isapplied, whereas solids resist an applied force.
1.1 FluidsConsider first the observable characteristics of a gas.
1. Gases are of low mass density, expand to fill acontainer, are easily compressed and have lowviscosity.
2. Gases exert a uniform pressure on the walls oftheir container.
3. Gases diffuse readily into each other.
The first set of observations suggests that the gasparticles are not in direct contact with each other,the second and third that they are in constantmotion in random directions and at high speeds.
Liquids exhibit most of the above properties;most diffuse into each other readily and their vis-cosities, although some orders of magnitudehigher than those of gases, are still low, so thatthey flow irreversibly under small forces.However, there are two significant differences.The first is that liquids are a condensed state andhave much higher mass densities, which are notvery different from solids. The second is that theyare almost incompressible (hydraulic brakingsystems depend on this).
All this suggests that the particles of fluids arein contact but are free to move relative to eachother. We shall consider the nature of the inter-action between the particles in Chapter 3. Mean-while, we can think of a liquid as being similar topeople at a well-run party in which the guestskeep circulating from one group to another.
Since the particles are free to move, the appli-cation of even a small force causes irreversibleflow. Strictly, this is a shear deformation in whichparts of the material slide past other parts (seeFigure 1.3(c)). In an ideal (Newtonian) liquid therate of shear flow, d/dt, is proportional to theapplied shear stress:
d/dt� �� (1.1)
where � �fluidity.This is generally written in the inverse form:
� � d/dt (1.2)
where is the coefficient of dynamic viscosity.As might be expected, the viscosities of gasesand liquids differ markedly. For air at 20°C, �18�10�5 Pa.s, for water at 20°C, �1.5�10�3 Pa.s. At higher temperatures theparticles possess more energy of their own andthe stress required to move them is reduced, i.e.viscosity reduces rapidly as temperature isincreased (e.g. lubricating oil, treacle, asphalt).
1.1.1 Diffusion
We noted above that both gases and liquids diffuseinto each other. Here, for simplicity, we considerthe diffusion of one gas into another, but the samearguments apply broadly to liquids and even tosolids under the appropriate conditions.
5
Diffusion is caused by countless haphazardwanderings of individual atoms or moleculeswhich continually bump into each other. Theatoms or molecules can rebound in any directionand the path of any individual atom or moleculeis unpredictable but, if enough make such move-ments, the result is a steady and systematic flow.
Imagine a box containing a partition (plane),on one side there are C1 particles of gas/unitvolume and on the other C2, where C1 �C2. Theparticles are moving randomly and exerting pres-sures of P1 and P2, respectively, on the partition,with P1 �P2. Now remove the partition. Both setsof particles will move across the plane into thespace formerly occupied solely by the other.However, because of the higher concentrationmore will move in the direction C1 →C2 and,eventually, a totally random mixture is formedand pressure is equalised.
Provided certain simplifying assumptions aremade it is easy to show that the total flux J ofparticles (i.e. the number of particles flowing inthe x-direction per unit time per unit cross-sectional area) through the plane is:
J� �DdC/dx (1.3)
which is Fick’s first law of diffusion. The constantD is known as the diffusion coefficient and is pro-portional to the average distance travelled by aparticle before colliding with another and the fre-quency at which the collisions occur. Thus, therate of diffusion is proportional to the concentra-tion gradient dC/dx, and is directed down thegradient, hence the minus sign above. Both ofthese follow directly from the statistical nature ofthe process.
It does not take much imagination to see thattemperature will play an important role. Athigher temperatures the particles are more activeand jump more frequently so that the rate of dif-fusion increases exponentially as the temperaturerises and vice versa. This has many importanttechnological consequences, e.g. in modifying theproperties of metals and alloys by heat treatment.
1.1.2 Vapour–liquid transition
Understanding of the vapour–liquid transitionderives from the work of Van der Waals. Hepointed out that the perfect gas law PV�RTwhich relates pressure P, volume V and tempera-ture T, via a constant R, neglected two importantfactors:
1. the volume of the particles themselves;2. the forces of attraction between the particles.
The first is fairly obvious and can be corrected bydeducting from the volume a term representingthe volume of the particles. Thus:
P(V�b)�RT (1.4)
The second is less obvious, the forces of attractionbetween the particles will have the effect ofdrawing them closer together, just as if additionalpressure were applied. So a correction must beadded to P. Consider a single particle about tostrike the wall of the chamber containing the gas. Itwill be subjected to an attractive force due to theadjacent particles, proportional to the number ofparticles n per unit volume. Furthermore, thenumber of particles striking the wall is also propor-tional to n. So that the total attractive force experi-enced by the particles about to strike the wall isproportional to n2. Now, for a chamber containinga fixed number of particles n is inversely propor-tional to the volume of the chamber V, so that theattractive force is proportional to 1/V 2. Thus, theVan der Waals equation is:
(P�a/V 2)(V�b)�RT (1.5)
where a is a constant. The consequences of thisare noteworthy. If P is plotted against V for dif-ferent values of T we obtain a set of isotherms asshown in Figure 1.1. Note that at low tempera-tures each curve has a maximum and minimum,at higher temperatures they become smooth.
The minimum and maximum on each curve areconnected by the dotted line. The point C wherethe minimum and maximum coincide at a pointof inflection is the critical point and denotes theconditions where liquid and vapour can coexist inequilibrium.
States of matter
6
Real materials, however, behave as shown inFigure 1.2 where we note that starting with aliquid at X at some constant temperature T1 thevolume increases as the pressure is reduced. Even-tually the liquid becomes saturated with vapourand the volume increases with no change of pres-sure, as a saturated vapour, i.e. a vapour satu-rated with liquid, forms. With further decreasesin pressure a true vapour is formed and the liquidis said to have evaporated. The reverse happens ifwe start with a gas and increase the pressurewhile keeping temperature constant.
The particles within a liquid (at constant tem-perature and pressure) may or may not have suffi-cient thermal energy to escape. Those whichescape constitute the vapour and the pressurethey exert is the vapour pressure. In a closed con-tainer equilibrium is set up between those parti-cles escaping from the liquid and those returning,this is the saturation vapour pressure. When the
Solids
7
FIGURE 1.1 Isotherms from Van der Waals equa-tion. The values for P, V and T are expressed in termsof their values at the critical point C. In this form theyare dimensionless and independent of their actualvalues.
FIGURE 1.2 P–V–T curves for water vapour.
saturation vapour pressure is equal to or greaterthan the external pressure the liquid boils. Thusthe boiling point is conventionally defined as thetemperature to which the liquid must be raised inorder for this to occur at atmospheric pressure.As the atmospheric pressure decreases so does theboiling point.
1.2 SolidsIntuitively the first thing that we notice about asolid is that it resists the application of a force. Aliquid will also resist a force provided that it isapplied equally in all directions. The essential dif-ference is that, unlike gases and liquids, a soliddoes not appear to change shape except by a tinyamount when a modest load is applied in just onedirection. We shall see later that this is notstrictly true; under appropriate conditions allmaterials will flow, but the definition is goodenough for our purpose and accords with normalexperience.
The reason for no apparent shape change issimple, the mechanism quite complex. First, wemust note what mechanics tells us. Any combina-tion of forces can be thought of as a combinationof a hydrostatic pressure (Figure 1.3 (d)) and ashear (Figure 1.3 (c)). Flow in any circumstances
States of matter
8
p
p
p
(d) Hydrostatic loading
�
G � shear modulus � �/
(c) Behaviour under shear stress
�
Strain
Str
ess
E � Young’s modulus ofelasticity (or stiffness)� tan �
(a) Linear behaviour under direct stress
Strain
(b) Non-linear behaviour under direct stress
��1
�3
�2
�
Tan � � tangent modulusat �1
Tan � � secant modulusbetween �2 and �3
�
Str
ess
FIGURE 1.3 Elastic stress–strain behaviour.
other than under pure hydrostatic pressure isreally a response to the shear component. Theatoms of which the solid is composed are physi-cally bonded together into patterns that must bemaintained. When we first apply a force wedisplace all the atoms of which the solid is com-
posed in a way that nearest neighbour atoms inthe pattern remain nearest neighbours (provided,of course, that the force is not sufficient to breakany bonds). This is quite unlike a liquid in whichthe individual particles move past each other.
1.2.1 The elastic constants
Before discussing the mechanics and con-sequences of bonding we should digress a little todefine some of the properties that are used tocharacterise solids.
When a solid is extended or compressed by amodest external force the dimensions change alittle. Conventionally we express the force F interms of the area A over which it acts, this isknown as stress �:
� �F/A (1.6)
Under simple uniaxial tension or compression thematerial either extends or shortens. Convention-ally we express this as an increment of strain�� ��l/l, i.e. change in length per unit length.Strain is therefore a ratio and has no dimensions.Integrating the expression for strain between theoriginal length, l0 and the final length, l1, gives�� ln(l1/l0)� ln(1� �l/l0), where �l� l1 � l0. Thisrather cumbersome expression for strain is usedfor analytical purposes when thinking about largestrains, but for most engineering purposes where�l/l0 �1, it reduces to the much more convenientand familiar definition:
� � �l/l0 (1.7)
When solids are not strained very much, say upto about � �0.001, or 0.1 per cent, the material
reverts to its original dimensions when the load isreleased, i.e. the strain is reversible. The relation-ship between stress and strain is often linear or,at least, linear to a first approximation (Figure1.3a). This is known as Hooke’s law, i.e.:
��E� (1.8)
where E is a material constant known as Young’smodulus of elasticity, or simply the elasticmodulus, and has units of stress, e.g. N/m2 (orPa) or, more commonly, GPa. Table 1.1 includessome typical values.
In many materials the stress–strain relationshipis not sufficiently linear for this definition. Forsuch materials, a modulus value can be defined aseither the slope of the tangent to the stress–straincurve at a particular value of stress – giving thetangent modulus – or as the slope of the linejoining two points on the curve at two values ofstress – giving the secant modulus (Figure 1.3b).When quoting either value, clearly the corre-sponding level of stress must also be quoted, e.g.the tangent modulus at the origin. Values ofsecant modulus are often given for a zero lowervalue of stress, i.e. it is the slope of the line fromthe origin to a point on the curve.
Modulus values arising from similar relation-ships apply for other forms of stress. Thus, inshear (Figure 1.3c) the shear stress � produces a
Solids
9
TABLE 1.1 Typical mass densities, elastic constants and Poisson’s ratios for a range of materials
Material Mass density E G K �Mgm�3 GPa GPa GPa
Diamond 3.5 1000 417 556 0.20Alumina 4.0 530 217 315 0.22Tungsten carbide 15.8 720 300 400 0.20Iron 7.9 210 79 206 0.33Silica glass 2.5 70 30 35 0.17Titanium 4.5 120 46 105 0.31Spruce (parallel to grain) 0.5 13 500 – 0.38Aluminium 2.7 69 26 68 0.33Concrete (lightweight) 1.4 15 6 8 0.20Concrete (normal weight) 2.4 25 10 14 0.20Epoxy resin 1.1 4.5 – – –Nylon 1.14 3 1.1 4.5 0.39
shear strain which, when small, is convenientlymeasured as an angle (in radians). For linearbehaviour
� �G (1.9)
where G is known as the shear modulus.Under hydrostatic pressure (Figure 1.3d) the
volume changes. As before this is expressed non-dimensionally as �V/V and
p� �K(�V/V) (1.10)
where p is the hydrostatic pressure and K is thebulk modulus. The minus sign arises as a con-sequence of the fact that as pressure increases,volume decreases.
Young’s modulus is a measure of the stressrequired to produce a given deformation, andengineers often refer to it as the stiffness of thematerial, for that is what it is. It will be apparentthat whereas a liquid can display a bulk modulusit does not possess a shear modulus since itcannot resist shear. Solids, on the other hand,display a bulk modulus and a shear modulus. It isthe resistance to shear that is the fundamental dif-ference between solids and liquids.
1.2.2 Elastic resilience – stored energy
We noted that a linearly elastic (Hookean) mater-ial deforms reversibly, i.e. it returns to its originallength when the load is removed. When a bar ofmaterial experiences an extension of �l under anapplied force F, mechanical work F�l is done.This work is stored in the material as a slightstretching of the atomic bonds. The total energystored is obtained by integrating this equation.The energy per unit volume of material U, whichis called the strain energy density, is thenobtained by dividing by the volume of materialAl0:
U�� (1.11)
Now, as previously defined, ��F/A and
� � ���
l0
l� � �
�
l0
l�, and therefore, for elastic deforma-
tion:
U����2� �2/2E� �2E/2 (1.12)
We can use this to estimate, in a few simple cases,the stresses generated by impact. Figure 1.4shows a simple wire suspended from a support
F�l�Al0
States of matter
10
l0l0 � ∆lS
l0 � ∆lD
Staticallyloaded
Dynamicallyloaded
Unloaded
h
M
FIGURE 1.4 Simple arrangement for comparing static and dynamic loads.
and carrying a small tray at its lower end. If aweight of mass M were to be gently lowered ontothe tray, the wire would experience a static forceMg, where g is the acceleration due to gravity,and a stress of �s �Mg/A, and would strain by�s/E�Mg/AE. If now, the weight were lifted andallowed to fall onto the tray from a height h, theimpact would cause the wire to stretch by morethan the static strain, after which it would oscil-late until it settled at the static strain. Taking themaximum dynamic stretch to be �lD, the weightwould lose potential energy Mg(h� �lD), all ofwhich would be converted into strain energy ofthe wire. Now the wire has a volume of Al0 andso the strain energy at maximum dynamic strainis (�D�D/2)(Al0). Equating the two energies andremembering that �D � �lD/l0, gives after somemanipulation:
�D ��s ����2s ����s�� (1.13)
Typically, E/�s 103 and if h� l0/2, we have�D/�s �34. No wonder things can snap when hitwith a hammer!
1.2.3 Poisson’s ratio
Clearly the application of a force displaces theatoms from their normal positions. Thus, in anelastic solid, the application of a longitudinaltension force causes a transverse contraction, asin Figure 1.5.
Poisson’s ratio � relates the longitudinal andtransverse strains:
� � �transverse strain/longitudinal strain
1.2.4 Relation between elastic constants
We now have all four elastic constants, andtypical values for these are given in Table 1.1. Itis not too difficult using linear elastic analysis toshow that they are not independent, and that theyare related to one another as follows:
K��3(1�
E
2�)� (1.14)
2hE�
l0
Solids
11
a0 � ∆aa0
b � ∆b
b
Longitudinal strain�l � ∆a/a0
Transverse strain�t � ∆b/b0
Poisson’s ratio� � �t/�l
FIGURE 1.5 Poisson effect under a uniaxial tensileload.
G��2(1
E
� �)� (1.15)
Thus if any two are known, or measured, theother two can be calculated, which makes theengineer’s life easier.
It is interesting and instructive to calculate thechange of volume that is brought about by a uni-axial stress. If a cube of side a is stressed in the zdirection, it will extend by:
�z �a�z
and in accordance with Poisson’s ratio:
�x � �y � �a��z
The final volume is:
V� (a��x)(a� �y)(a��z)
Expanding and neglecting second and higherorder terms in � which are vanishingly small:
V�a3 �a2(�x � �y � �z)�a3 �a3(�x ��y � �z)
the change of volume is:
�V�a3(�x � �y ��z)�a3�z(1�2�)
and:
�V/V��z(1�2�) (1.16)
For a liquid, the volume remains constant duringdeformation, i.e. the material is incompressible,and �V�0. K is therefore infinite, and, fromequation 1.14, � �0.5. In most other materialsthere is a small volume increase when stretchedelastically and a volume decrease when com-pressed. This means that K is finite and positive –if it were negative the volume would increasewhen the material is compressed, which does notbear thinking about. Since E is also a positivenumber, then equation 1.14 tells us that 1�2�must be positive, and hence � �0.5. Combiningthe two conditions gives
� �0.5 (1.17)
Table 1.1 shows that for many common materials� lies between 0.2 and 0.35. Soft rubber is unusualin having a Poisson’s ratio close to 0.5, whichmeans that it behaves very like a stiffish fluid.
1.3 Intermediate behaviourWe have already emphasised the lack of anydividing line between solids and liquids, thereason being that the response of materials toshear can vary over a wide range. Solids may beeither crystalline or amorphous (Figure 1.6). In acrystalline solid (and metals are an importantgroup) the atoms are arranged in a regular, three-dimensional array or lattice. Amorphous solidsdo not possess this regularity of structure and wecan further subdivide them into glassy and molecu-lar solids.
True glasses are obtained by cooling liquidswhich, because of their elaborate molecular con-figuration, lack the necessary activation energy(see Chapter 2) at the melting temperature torearrange themselves in an ordered crystallinearray.
The material then retains, even when rigid, atypically glassy structure involving short rangeorder only and resembling, except for its immo-bility, the structure of the liquid form. Silica givesrise to the commonest form of glass. If cooledsufficiently slowly from the melt it can beobtained in the crystalline form which is morestable thermodynamically than the vitreous state.In nature, cooling over long time spans gives riseto beautiful crystals of quartz and other crys-talline forms.
At ordinary cooling rates crystallisation doesnot take place, and the amorphous nature of thesolid gives rise to its most important physicalproperty, its transparency. Most non-metallicsingle crystals are transparent, but in the poly-crystalline state light is scattered from internalreflections at flaws and internal boundaries, thematerial then becomes translucent in thin sectionsand completely opaque in the mass, as with natu-rally occurring rocks. If it is amorphous, there areno internal boundaries. Polymers may be totallyamorphous and transparent, but most containsmall crystalline regions (spherulites) within theirstructure that act as light scatterers, giving therather milky appearance that we are familiarwith, that is if they do not contain deliberatelyadded colouring matter.
Molecular solids are of two types. So-calledthermoplastic polymers are best exemplified bysoft polymers such as polyethylene, which consistof long, highly convoluted molecules. The bulkstrength depends upon the entanglement of themolecular chains rather than upon three-dimensional bonding. Thermosets, such as Bake-lite, have genuine extended three-dimensionalmolecules. Mostly, these are harder and strongerthan thermoplastics but tend to be rather brittle.
There are, however, more complex structuresinvolving both solid and liquid phases. One of themost familiar of these is the gel, known to mostof us from childhood in the form of jellies andlozenges.
Gels are formed when a liquid contains a fairlyconcentrated suspension of very fine particles,usually of colloidal dimensions (�1�m). The par-ticles bond into a loose structure, trapping liquid
States of matter
12
in its interstices. Depending on the number oflinks formed, gels can vary from very nearly fluidstructures to almost rigid solids. If the links arefew or weak, the individual particles have consid-erable freedom of movement around their points
of contact, and the gel deforms easily. A highdegree of linkage gives a structure that is hardand rigid in spite of all its internal pores. Themost important engineering gel is undoubtedlycement, which develops a highly rigid structure
Intermediate behaviour
13
FIGURE 1.6 Schematic structures of some covalent materials: (a) vitreous silica (glass); (b) crystalline silica; (c)thermoplastic polymer showing crystalline regions (spherulites); (d) thermosetting polymer; (e) gel – by compari-son with (d), the links between the chains are weak and easily broken.
(which we will discuss in some detail in Chapter13). When water is added to the cement powder,the individual particles take up water of hydra-tion, swell and link up with each other to giverise to a high-strength but permeable gel ofcomplex calcium silicates. A feature of many gelsis their very high specific surface area; if the gel ispermeable as well as porous, the surface is avail-able for adsorbing large amounts of watervapour, and such a gel is an effective dryingagent. Adsorption is a reversible process (seepages 42–3); when the gel is saturated it may beheated to drive off the water and its dryingpowers regained. Silica gel is an example of this.
If a gel sets by the formation of rather weaklinks, the linkages may be broken by vigorousstirring so that the gel liquefies again. When thestirring ceases, the bonds will gradually link upand the gel will thicken and return to its originalset. Behaviour of this sort, in which an increase inthe applied shear rate causes the material to act ina more fluid manner, is known as thixotropy.Not all gels behave in this manner or, if they do,it is only at a certain stage of their setting pro-cedure. Hardened concrete, alas, will not healitself spontaneously after it has cracked, althoughit exhibits a marked degree of thixotropy at anearly state of setting.
The most familiar application of thixotropy isin non-drip paints, which liquefy when stirred
and spread easily when being brushed on, butwhich set as a gel as soon as brushing is com-pleted so that dripping or streaks on vertical sur-faces are avoided. Clays can also exhibitthixotropy. This is turned to advantage on apotter’s wheel and in the mixing of drilling mudsfor oil exploration. The thixotropic mud serves toline the shaft with an impermeable layer, whilstin the centre it is kept fluid by the movement ofthe drill and acts as a medium for removing therock drillings. On the other hand, a thixotropicclay underlying major civil engineering workscould be highly hazardous.
The reverse effect to thixotropy occurs whenan increase in the applied shear rate causes aviscous material to behave more in the manner ofa solid, and is known as dilatancy. It is a lessfamiliar but rather more spectacular phenome-non. Cornflour–water mixtures demonstrate theeffect over a rather narrow range of composition,when the viscous liquid will fracture if stirred vig-orously. It is of short duration, however, sincefracture relieves the stress, and the fractured sur-faces immediately liquefy and run together again.
Silicone putty is also a dilatant; it flows veryslowly if left to itself, but fractures if pulled sud-denly, and will bounce like a rubber ball ifthrown against a hard surface. So far, no engi-neering applications of dilatancy have beendeveloped.
States of matter
14
Chapter two
Energy andequilibrium
2.1 Mixing2.2 Entropy2.3 Free energy2.4 Equilibrium and equilibrium diagrams
Although the engineer conventionally expresseshis findings in terms of force, deflection, stress,strain and so on, these are simply the convention.Fundamentally, what he is really dealing with isenergy. Any change, no matter how simple,involves an exchange of energy. The mere act oflifting a beam involves a change in the potentialenergy of the beam, a change in the strain energyheld in the lifting cables, an input of mechanicalenergy from the lifting device which is itself trans-forming electrical or other energy into kineticenergy. The harnessing and control of energy is atthe heart of all engineering.
Thermodynamics teaches us about energy, anddraws attention to the fact that every materialpossesses an internal energy associated with itsstructure. In this section we examine some of thethermodynamic principles which are of import-ance to understanding the behaviour patterns.
We begin by recognising that all systems arealways seeking to minimise their energy, i.e. tobecome more stable. We also note that althoughthermodynamically correct, some changestowards a more stable condition proceed soslowly that the system appears to be stable eventhough it is not. For example, a small ball sittingin a hollow at the top of a hill will remain thereuntil it is lifted out and rolled down the hill. The
ball is in a metastable state and requires a smallinput of energy to start it on its way down themain slope.
Figure 2.1 shows a ball sitting in a depression,its potential energy is P1. It will roll to a lowerenergy state P2, but only if it is first lifted to thetop of the hump between the two hollows. Someenergy has to be lent to the ball to do this. Ofcourse, the ball returns the energy when it rollsdown the hump to its new position. This bor-rowed energy is known as the activation energyfor the process. Thereafter it possesses free energyas it rolls down to P2. However, it is losing poten-tial energy all the time and eventually (say, at sealevel) it will achieve a stable equilibrium.However, note two things. At P1, P2, etc. it is‘stable’, actually metastable because there areother more stable states available to it, given thenecessary activation energy. Where does the acti-vation energy come from? In materials science itis extracted mostly (but not exclusively) from
15
FIGURE 2.1 Schematic illustration of activationenergy and free energy.
heat. As things are heated to higher temperaturesthe atomic particles react more rapidly and canbreak out of their metastable state into one wherethey can now lose energy.
2.1 MixingIf whisky and water are placed in the same con-tainer, they mix spontaneously. The internalenergy of the solution so formed is less than thesum of the two internal energies before they weremixed. There is no way that we can separatethem except by distillation, i.e. by heating themup and collecting the vapours and separatingthese into alcohol and water. We must, in fact,put in energy to separate them. But, since energycan neither be created nor destroyed, the fact thatwe must use energy, and quite a lot of it, torestore the status quo must surely pose the ques-tion, ‘Where does the energy come from initially?’The answer is by no means simple but, as weshall see, every particle, whether water or whisky,possesses kinetic energies of motion and of inter-action.
When a system such as a liquid is left to itself,its internal energy remains constant, but when itinteracts with another system it will either lose orgain energy. The transfer may involve work orheat or both and the first law of thermodynamics,‘the conservation of energy and heat’, requiresthat:
dE�dQ�dW (2.1)
where E� internal energy, Q�heat andW�work done by the system on the surround-ings.
What this tells us is that if we raise a cupful ofwater from 20°C to 30°C it does not matter howwe do it. We can heat it, stir it with paddles oreven put in a whole army of gnomes eachequipped with a hot water bottle, but the internalenergy at 30°C will always be above that at 20°Cby exactly the same amount. The first law saysnothing about the sequences of changes that arenecessary to bring about a change in internalenergy.
2.2 EntropyClassical thermodynamics, as normally taught toengineers, regards entropy S as a capacity pro-perty of a system which increases in proportionto the heat absorbed (dQ) at a given temperature(T). Hence the well-known relationship:
dS�dQ/T
which is a perfectly good definition but does notgive any sort of picture of the meaning of entropyand how it is defined. To a materials scientistentropy has a real physical meaning, it is ameasure of the state of disorder in the system.Whisky and water combine; this simply says that,statistically, there are many ways that the atomscan get mixed up and only one possible way inwhich the whisky can stay on top of, or, depend-ing on how you pour it, at the bottom of, thewater. Boltzmann showed that the entropy of asystem could be represented by:
S�klnN (2.2)
where N is the number of ways in which the par-ticles could be distributed and k is a constant(Boltzmann’s constant k�1.38�10�23 J/K). Thelogarithmic relationship is important; if the mole-cules of water can adopt N1 configurations andthose of whisky N2 the number of possible config-urations open to the mixture is not N1 �N2 butN1 �N2. It follows from this that the entropy ofany closed system will tend to a maximum sincethis represents the most probable array of config-urations. You should be grateful for this. As youread these words, you are keeping alive bybreathing a randomly distributed mixture ofoxygen and nitrogen. Now it is statistically pos-sible that at some instant all the oxygen atomswill collect in one corner of the room while youtry to exist on pure nitrogen, but only statisticallypossible. There are so many other possible distri-butions involving a random arrangement of thetwo gases that it is most likely that you will con-tinue to breathe the normal random mixture.
Energy and equilibrium
16
2.3 Free energyIt must be clear that the fundamental tendencyfor entropy to increase, that is, for systems tobecome more randomised, must be stopped some-where and somehow. For, if not, the entire uni-verse would break down into chaos. As we shallsee, the reason for the existence of liquids andsolids is that atoms and molecules are not totallyindifferent to each other and, under certain con-ditions and with certain limitations, will associatewith each other in a non-random way.
As we stated above, from the first law ofthermodynamics the change in internal energy isgiven by:
dE�dQ�dW
From the second law of thermodynamics theentropy change in a reversible process is:
TdS�dQ
Hence:
dE�TdS�dW (2.3)
In discussing a system subject to change, it is conve-nient to use the concept of free energy. For irre-versible changes, the change in free energy is alwaysnegative and is a measure of the driving forceleading to equilibrium. Since a spontaneous changemust lead to a more probable state (or else it wouldnot happen) it follows that, at equilibrium, energy isminimised while entropy is maximised.
The Helmholtz free energy is defined as:
H�E�TS (2.4)
and the Gibbs free energy as:
G�pV�E�TS (2.5)
and, at equilibrium, both must be a minimum.We shall later in Chapter 4 apply these ideas tothe elastic deformation of materials.
2.4 Equilibrium and equilibriumdiagramsSo far, we have seen that materials can exist asgases, liquids and crystalline or amorphous
solids. We need a scheme that allows us to sum-marise the influences of temperature and pressureon the relative stability of each state and on thetransitions that can occur between these. Ther-modynamics tells us that we must seek theinternal condition of a material that minimises itsinternal energy.
The time-honoured approach is through themedium of the equilibrium diagram. Note theword ‘equilibrium’. As we have seen at the startof this chapter, it takes a finite time for a transi-tion to occur from one state to another or for achemical reaction to take place. Sometimes, thistime is vanishingly small, as when dynamiteexplodes. At other times, it can be a few seconds,days or even centuries. Glass made in the MiddleAges is still glass and shows no sign of crystallis-ing. Not every substance or mixture has reacheda state of thermodynamic equilibrium, i.e. whereits internal energy is lowest. Nevertheless, wehave to begin from a knowledge of what the equi-librium condition would be.
Although there is a great wealth of fundamen-tal theory underlying the forms of equilibriumdiagrams, we shall only dip our toes into it. Wewill think of them as useful maps. Until recently,these maps were established by experiment, butthe theory and the computational power to dothe calculations have now reached the stagewhere the diagrams can be predicted if the neces-sary basic data is in place.
2.4.1 Single component diagrams
There are two commonly used types of diagram.We have met one already, the P–V–T diagram forwater shown in Figure 1.2. An alternative andmore helpful way of looking at the same data isshown in Figure 2.2. We need to look at this insome detail in order to establish some groundrules and language for use later.
The diagram is in ‘temperature–pressure space’,i.e. the axes are of temperature and pressure. Anumber of lines are marked which representboundary conditions between differing ‘phases’,i.e. states of H2O. The line AD represents combi-nations of temperature and pressure at which
Equilibrium and equilibrium diagrams
17
liquid water and solid ice are in equilibrium, i.e.can coexist. A small heat input will alter the pro-portions of ice and water by melting some of theice. However, it is absorbed as a change ininternal energy of the mixture, the latent heat ofmelting. The temperature is not altered.
What happens if we put in large amounts ofheat, so that all the ice is melted and we havesome heat left over? Of course, the temperaturerises and now we have just slightly warmedwater.
By the same token, line AB represents the equi-librium between liquid water and gaseous steam,and line AC the equilibrium between solid ice andrather cold water vapour.
It is helpful to go on one or two journeyswithin the diagram. To begin with, let us start atpoint X, representing �5°C at atmospheric pres-sure. We know we should have ice and, indeed,the point X lies within the phase field labelled ice.Adding heat at constant pressure takes the tem-perature along the broken line. This crosses thephase boundary, AD, at 0°C and the ice begins tomelt as further heat is added. Not until all the icehas melted does the temperature continue to rise.We now have liquid water until we reach 100°C.Now, again, heat has to be added to boil thewater but there is no temperature increase until
Energy and equilibrium
18
0.001
0.01
0.1
1.0
10
100
1000
�20 0 20 40 60 80 100 120
Temperature (°C)
B
D
C
Liquid
Vapour
xX Y
Ice
A
Pres
sure
(ba
r)
FIGURE 2.2 Pressure–temperature diagram forwater (from W.D. Kingery (1976) Introduction toceramics, Wiley, New York).
all the liquid water has gone. We now have steamand its temperature can be increased by furtherheat input.
Now think of keeping temperature constantand increasing pressure, again starting at point X.If the pressure is raised enough, to about 100atmospheres (�10MPa) we reach the ice waterequilibrium and the ice can begin to melt. It isthought in some quarters that this accounts forthe low friction between, for example, an iceskate and the ice itself: local pressures cause localmelting. It is a factor which no doubt enters theminds of engineers when contemplating the use oflocally refrigerated and frozen ground as cofferdams or as foundations for oil rigs in Alaska.
Now consider starting with steam at say 200°Cand 1 atmosphere. Increasing pressure to theliquid–vapour equilibrium will cause it to con-dense to water. Wet steam at high pressures is afamiliar bugbear of mechanical engines which usesteam as their working fluid. It should be notedthough that when the temperature is taken abovea critical temperature, 374.14°C, no amount ofpressure will cause the steam to condense. Thepressure at the critical point is �221 atmospheres(22.1MPa) and the mass density of the steam is�0.32Mgm�3, about 1/3 of that of liquid waterat 0°C!
There is a formal way of summarising the rela-tionship between the number of phases (P) thatcan coexist at any given point in the diagram andthe changes brought about by small changes intemperature or pressure. It is the famous phaserule, first enunciated by J.W. Gibbs:
P�F�C�2 (2.6)
Here, C is the number of components in thesystem; in this case we have only H2O so C�1. Fis the number of degrees of freedom allowed tochange. To illustrate, at point X (Figure 2.2)there is just one phase, ice, so P�1 and F�2.This means that both temperature and pressurecan be changed independently without bringingabout a significant change to the material. At Y,both ice and liquid can coexist, P�2 and F�1.To maintain the equilibrium, temperature andpressure must be changed in a co-ordinated way
so that the point Y moves along the boundaryAD. At A, all three phases can coexist and F�0,i.e. any change at all will disturb the equilibrium.
One further point needs to be made beforemoving on. The slope of each phase boundarycurve is given by the Clausius–Clapeyron equa-tion:
dp/dT� �H/T�V (2.7)
where �H is the heat of transformation (fusion,boiling) per mole of the substance, T is theabsolute temperature and �V is the volumechange accompanying the transformation (permole). �H and �V refer to changes from the lowtemperature condition to the higher temperatureone. Now, �H is always positive and �V onboiling is large and positive. Thus, AB has a lowslope. Usually, �V on melting is also positive, butwe know that water is unusual in that it contractswhen it melts. In this case, �V is small and negat-ive, giving AD a steep and negative slope.
Transformations can also occur between solidphases. For example, silica can exist in four crys-talline forms, each stable over a different temper-ature range. �V on going from one form toanother is very small and so dp/dT is steep. Theequilibrium diagram is shown in Figure 2.3. Insilica the transformation rates are extremely slow
Equilibrium and equilibrium diagrams
19
573°C 870°C 1470°C 1713°C
Temperature
Pres
sure �
2-T
ridym
ite
Liqu
id
�-Q
uart
z
Vapour
1 bar
�-Q
uart
z
�-C
risto
balit
e
FIGURE 2.3 Equilibrium phase diagram for silica(SiO2) (from W.D. Kingery (1976) Introduction toceramics, Wiley, New York).
and so high temperature phases can be retained atlow temperatures even though they are not ther-modynamically stable. Extensions of each curvein the manner shown in Figure 2.4 can be usefulin thinking about such circumstances. The phasewith the lowest vapour pressure is the moststable. Figure 2.4 shows clearly that (amorphous)silica glass is unstable with respect to cristobalite,tridymite and quartz at low temperatures. Silicaglass remains technologically useful only becausethe transformation rates are extremely slow.
2.4.2 Two component diagrams
We now go on to look at two component dia-grams, such as we get when carbon is alloyed intoiron to make steels and cast irons. We now have anextra variable, composition, and, strictly, weshould consider the joint influences of this variablein addition to temperature and pressure. We needthree-dimensional diagrams. To simplify things weusually take pressure to be constant. After all, mostengineering materials are prepared and used atatmospheric pressure, unless you work for NASA!This leaves us with a composition–temperaturediagram, the lifeblood of materials scientists.
Temperature
Pres
sure
Liquid
�-Quartz
�-Quartz
�-Cristobalite
�2-Tridymite
Silica glass
FIGURE 2.4 Equilibrium phase diagram fromFigure 2.3 extended to include the metastable (amor-phous) silica glass phase of SiO2 (from W.D. Kingery(1976) Introduction to ceramics, Wiley, New York).
We begin with the simplest example, Figure2.5. This is for alloys formed between copper(Cu) and nickel (Ni). The diagram is drawn withcomposition as the horizontal axis, one end rep-resenting pure Cu, the other pure Ni. The verticalaxis is temperature.
Let us think about an alloy which is50%Cu:50%Ni by mass. At high temperatures,e.g. at A, the alloy is totally molten. On cooling,we move down the composition line until wearrive at B. At this temperature, a tiny number ofsmall crystals begin to form. Further reduction intemperature brings about an increase in theamount of solid in equilibrium with a diminishingamount of liquid. On arriving at C, all the liquidhas gone and the material is totally solid. Furthercooling brings no further changes. Before movingon, note the important difference between thisalloy and the pure metals of which it is com-posed. Both Cu and Ni have well-defined uniquemelting (or freezing) temperatures. The alloysolidifies over the temperature range BC; metal-lurgists often speak of the ‘pasty range’.
We now need to examine several matters inmore detail. First, the solid crystals that form arewhat is known as a ‘solid solution’. Cu and Niare chemically similar elements and both, whenpure, form face-centred cubic crystals (seeChapter 8). In this case, a 50:50 alloy is alsocomposed of face-centred cubic crystals but eachlattice site has a 50:50 chance of being occupiedby a Cu atom or an Ni atom.
Energy and equilibrium
20
Cu NiCu/Ni:50/50
Tem
pera
ture
Y Z
C
X
B
D
A
Composition (mass%)
Liquid
Solid
1453°C
1083°C
FIGURE 2.5 Equilibrium phase diagram for Cu–Ni.
How does the phase rule work? Consider pointA. Here we have two components (Cu and Ni)and one phase (liquid). This leaves us with threedegrees of freedom. We can independently altercomposition, temperature and pressure and thestructure remains liquid. But remember, we havetaken pressure to be constant and so we are leftwith two practical degrees of freedom, composi-tion and temperature. The same argument holdsat point D, but, of course, the structure here isthe crystalline solid solution of Cu and Ni.
At a point between B and C we have liquid andsolid phases coexisting, P�2 and F�2. Asbefore we must discount one degree of freedombecause pressure is taken as constant. This leavesus with F�1, which means that the status quocan be maintained only by a coupled change incomposition and temperature. What is the statusquo in this case? It is not only that the structure istwo phase, but also that the proportions of liquidand solid phases remain unaltered.
The next trick is to find the proportions ofliquid and solid corresponding to any point in thetwo-phase field. To do this we draw the constanttemperature line through the point X, Figure 2.5.This intersects the phase boundaries at Y and Z.The solid line containing Y represents the lowerlimit of 100 per cent liquid, and is known as theliquidus. The solid line containing Z is the upperlimit of 100 per cent solid and is known as thesolidus.
Neither the liquid nor solid phases correspond-ing to point X have a composition identical withthat of the alloy as a whole. The liquid containsmore Cu and less Ni, the solid less Cu and moreNi. The composition of each phase is given by thepoints Y and Z respectively. The proportions ofthe phases balance so that the weighted average isthe same as the overall composition of the alloy.It is easy to show that:
(Weight of liquid of composition Y) � YX�(Weight of solid of composition Z) � XZ
This is similar to what would be expected of amechanical lever balanced about X and con-sequently is known as the Lever rule.
One consequence of all this can be seen by
re-examining the cooling of the 50:50 alloy fromthe liquid phase. Consider Figure 2.6. At point X1
on the liquidus, solidification is about to begin.At a temperature infinitesimally below X1 therewill be some crystals solidifying out of the liquid;their composition is given by Z1. At a tempera-ture about halfway between solidus and liquidus(X2), we have a mixture of solid and liquid ofcompositions Z2 and Y2. In general, the propor-tion of liquid to solid halfway through the freez-ing range need not be �50:50, but in this case itis. Finally, at a temperature infinitesimally aboveX3, which is on the solidus, we have nearly 100per cent solid of composition Z3 together with avanishingly small amount of liquid of composi-tion Y3. When the temperature falls to just belowX3, the alloy is totally solid and Z3 has becomeidentical with X3.
Note two important features. First, Z3 is thesame as the average composition we started with,X1. Second, solidification takes place over a rangeof temperatures, and as it occurs the composi-tions of liquid and solid phases change continu-ously. For this to actually happen, substantialamounts of diffusion must occur in both liquidand solid. That in solids is very much slower thanthat in liquids, and is the source of some practicaldifficulty. Either solidification must occur slowlyenough for diffusion in the solid to keep up or
Equilibrium and equilibrium diagrams
21
Z1Y2
Y3
Z2X3
X1
Composition
Tem
pera
ture
X2
Cu Ni
Liquid
Solid
Z3
FIGURE 2.6 Equilibrium phase diagram for Cu–Ni(Figure 2.5) redrawn to emphasise composition varia-tions with temperature.
Al � Si
Liquid
Al 40 60 8020 Si
Liquid � Si
Al
Si
400
200
600
800
1000
1200
1400
Silicon (mass%)
Tem
pera
ture
(°C
) Liquid + Al
TE
Y
A
BLB
CE
X
FIGURE 2.7 Equilibrium phase diagram for Al–Si.
strict equilibrium conditions are not met. Thekinetics of phases transformations is something towhich we will return. However, for the moment,we continue to confine attention to very slowlyformed, equilibrium or near equilibrium struc-tures.
2.4.3 Eutectic systems
Let us now examine another diagram, that foraluminium–silicon (Al–Si) alloys (Figure 2.7.)Pure Al forms face-centred crystals, but Si has thesame crystal structure as diamond. These areincompatible and extensive solid solutions likethose for Cu:Ni cannot be formed. Si crystals candissolve only tiny amounts of Al. For our pur-poses, we can ignore this solubility, although wemight recognise that the semiconductor industrymakes great use of it, small as it is. Al crystals candissolve a little Si, but again not very much, andwe will ignore it. Thus, two solid phases are pos-sible, Al and Si.
When liquid, the elements dissolve readily inthe melt in any proportions.
Consider the composition Y. On cooling to theliquidus line at A, pure (or nearly pure) crystals
of Si begin to form. At B we have solid Si coexist-ing with liquid of composition LB in proportionsgiven by the Lever rule. At C we have solid Si inequilibrium with liquid of composition nearly atE.
Now consider alloy X. The sequence is muchthe same except the first solid to form is nowcrystals of Al. When the temperature has fallen toalmost TE we have solid Al in equilibrium withliquid of composition close to E.
Note that both alloy X and alloy Y, whencooled to TE, contain substantial amounts ofliquid of composition E. An infinitesimal drop oftemperature below TE causes this liquid to solid-ify into a mixture of solid Al and solid Si. At Ewe have three phases which can coexist: liquid,solid Al and solid Si. The system has two com-ponents and thus the phase rule gives us nodegrees of freedom once we have discounted pres-sure. E is an invariant point; any change in tem-perature or composition will disturb theequilibrium.
The point E is known as the eutectic point andwe speak of the eutectic composition and theeutectic temperature, TE. This is the lowest tem-perature at which liquid can exist and the eutecticalloy is that which remains liquid down to TE. Itsolidifies at a unique temperature, quite unlikeCu–Ni or Al–Si alloys of other compositions.Alloys close to the eutectic composition (Al�13%Si) are widely used because they can beeasily cast into complex shapes, and the Si dis-persed in the Al strengthens it. Eutectic alloys inother systems find similar uses (cast-iron is ofnear eutectic composition) as well as uses asbrazing alloys, etc.
2.4.4 Intermediate compounds
Often, the basic components of a system canform compounds. For example, lime, CaO,and silica, SiO2, can form the compounds2(CaO)SiO2, 3(CaO)SiO2 and others, which havegreat technological significance as active ingredi-ents in Portland cement (to be discussed in detailin Chapter 13). Similarly, SiO2 and corundum,Al2O3, form mullite, 3(Al2O3)2(SiO2), an import-
Energy and equilibrium
22
1400
1600
1800
2000
2200
20 40 60 80
Tem
pera
ture
(°C
)
Mullite � Liquid
Crystobalite � Mullite
Liquid
Corundum� liquid
Corundum� mullite
Crystobalite� liquid
Mullite � Liquid
Al2O3SiO2
Mullite 3Al2O3:2SiO2
Al2O3 (mass%)
FIGURE 2.8 Equilibrium phase diagram forSiO2–Al2O3.
ant constituent of fired clays, pottery and bricks.In metals we can have CuAl2, Fe3C and manymore.
Figure 2.8 shows the SiO2–Al2O3 diagram. Itcan be thought of as two diagrams, one for‘SiO2–mullite’ and the other for ‘mullite–Al2O3’,joined together. Each part diagram is a simpleeutectic system like Al–Si. In a similar way, thelime (CaO) silica (SiO2) diagram, Figure 2.9, can
600
1000
1400
1800
2200
2600
SiO2 20 40 60 80 CaO
CaO (mass%)
Tem
pera
ture
(°C
) Liquid
C � CaOS � SiO2
CS
C3S2
C2S
C3S
FIGURE 2.9 Equilibrium phase diagram for lime(CaO) – silica (SiO2).
be thought of as a series of joined together eutec-tic systems.
In many cases we do not have to think aboutthe whole diagram. Figure 2.10 shows theAl–CuAl2 diagram, again a simple eutecticsystem. A notable feature is the so-called solvusline, AB, which represents the solubility of CuAl2
in solid crystals of Al. This curves sharply, so thatvery much less CuAl2 will dissolve in Al at lowtemperatures than will at high temperatures. This
Equilibrium and equilibrium diagrams
23
20Al 30 40 5010
Copper (mass%)
200
300
400
500
600
700
Tem
pera
ture
(°C
)
Liquid
� � liquid
� � CuAl2
�
CuAl2
CuAl2 � liquid
A
B
FIGURE 2.10 Equilibrium phase diagram forAl–CuAl2.
Ferrite � Fe3C
Austenite
1 2 3 4 50
Carbon (mass%)
200
400
600
800
1000
1200
1400
1600
Liquid �austenite Li
quid
� F
e 3C
Liquid
0.007%C
Austenite � Fe3C
4.3%C
0.8%CTem
pera
ture
(°C
)
0.035%C
X
B A
Y
C
Ferrite + austenite
Ferrite
is a fortunate fact that underlies our ability toalter the microstructures of some alloys by suit-able heat treatments. Details will be taken uplater.
2.4.5 The Fe–C diagram
When I was a student, my tutor told me that Icould consider myself a true metallurgist when Ihad dreamed about the Fe–C diagram! Such is itsimportance to the metallurgy of steels and castirons, both of which are of prime interest to civilengineers.
A section of the diagram for carbon contentsup to �5 per cent is shown in Figure 2.11. It hasa eutectic at �4.3%C and 1150°C. This lowmelting temperature allows such alloys to bemelted with relative ease and to be cast intocomplex shapes. Cast irons have carbon contentsusually above 2 per cent. Sometimes the productsof solidification are Fe and Fe3C, sometimes Feand graphite, and sometimes Fe, Fe3C andgraphite. Graphite containing cast-irons are relat-ively soft and easy to machine. Cast-irons con-taining no graphite but with all the carbon asFe3C are very hard and wear resistant. They oftenturn up as wear resistant facings on earth moving
FIGURE 2.11 Equilibrium phase diagramfor iron–carbon.
machinery. Both types are easily cast intocomplex shapes.
At lower carbon contents, less than �1 per cent,the liquidus climbs to temperatures which makemelting more tricky. But by far the most importantfeature of the diagram arises from the allotropy ofiron. At temperatures below 910°C, pure Fe formsinto body-centred cubic crystals, known as ferrite.At higher temperatures, the crystals have face-centred cubic structures and are known as austen-ite. Up to 2%C can dissolve in austenite at1150°C, but this rapidly reduces and Fe3C isprecipitated as the temperature falls to 723°C, atwhich a maximum of 0.8 per cent carbon can dis-solve in austenite. Almost no carbon will dissolvein ferrite, but that which does has very profoundeffects. Transitions from austenite to lower tem-perature forms of the alloys give rise to a part ofthe diagram that is reminiscent of a eutecticdiagram. To emphasise the fact that it representstransitions from one solid condition to anothersolid condition, it is referred to as a eutectoid.
Consider the alloy X at 1000°C (Figure 2.11).It is fully austenitic; on cooling to the point Asome Fe3C is precipitated and the composition ofthe austenite in equilibrium with the Fe3C is givenby point B. Similarly, on cooling alloy Y from1000°C to say 750°C, ferrite is precipitated andthe composition of the austenite in equilibriumwith it is C. At 723°C, we have the eutectoidpoint and the austenite contains 0.8%C. Furthercooling causes the austenite to decompose into amixture of ferrite and Fe3C. It consists of alter-nating lamellae (lathes) of Fe and Fe3C, arrangedin colonies within which the lamellae are nearlyparallel. The scale is such that the structure actsas a diffraction grating to light and gives themicrostructures an iridescent and pearly appear-ance. Consequently, the mixture is known aspearlite. The consequences of all this to the met-allurgy of steels and cast irons will be dealt within Chapter 12.
Energy and equilibrium
24
Chapter three
Atomicstructure andinteratomicbonding
3.1 Ionic bonding3.2 Covalent bonding3.3 Metallic bonding3.4 Van der Waals bonds
It will be clear that the differences between solidsand liquids, indicated earlier, are a consequenceof the bonding forces between the atoms of whichthey are composed. In order to discuss theseforces some understanding of the structure of theatom is required. The following uses a simplifiedmodel.
The atoms of all elements can be visualised as apositively charged nucleus, comprising of one ormore positively charged protons and electricallyneutral neutrons, surrounded by orbiting nega-tively charged electrons. The electrical charges onthe proton (�e) and electron (�e) are equal andopposite. All atoms, in their normal state, containequal numbers of protons and electrons and areelectrically neutral. The mass of the proton isabout 2000 times that of the electron. Except inthe case of hydrogen, the incorporation of neu-trons in the nucleus is essential for stability;however, for our purposes we can disregard theirpresence. The nucleus of hydrogen, the simplestatom, comprises a single proton, about which a
single electron orbits (Figure 3.1 (a)). Helium hasa nucleus containing two protons, with two elec-trons in orbit about it (Figure 3.2 (b)). The nextelement, lithium, has a nucleus containing threeprotons, with three electrons arranged in orbitsabout it, the third electron being in a separate
25
�1e �2e
�3e �26e
(d) Iron(c) Lithium
(a) Hydrogen (b) Helium
FIGURE 3.1 Schematic of the atomic structures ofthe first three elements of the periodic table and iron.
shell from the first two (Figure 3.1 (c)). Heavieratoms, of course, contain increasing numbers ofprotons (and neutrons) and electrons (Figure3.1 (d)).
In order to emphasise how the electronic struc-ture of the elements affects their properties, list-ings of all the known elements are usuallyarranged in a form called the periodic table. Thistable enables predictions to be made regardingwhich elements will form compounds and whatthe properties of the compounds will be. Anatomic structure of particular stability ariseswhenever the outermost shell contains eight elec-trons. These octets, as they are known, are foundin neon, argon, krypton, xenon, etc.; that is, theinert gases, so-called because they are just that.
Consider now the atom of sodium. It containseleven electrons arranged in shells as 2–8–1(Figure 3.2). Clearly it can realise the octet con-figuration by losing an electron. In doing so it isleft with a net positive charge and is known as anion, in this case a positive ion. Chlorine, on theother hand, contains seventeen electrons arrangedas 2–8–7 (Figure 3.2) and the octet configurationcan be achieved if it accepts an electron tobecome a negative ion. The electrons in the outershell of an atom are responsible for bonding withother atoms and are called valence electrons. Thenumber of these outermost electrons determinesthe valency, i.e. the number of bonds which canbe formed with other atoms. Sodium has avalency of 1 and the chemical formula for sodium
Atomic structure and interatomic bonding
26
�11e
Sodium2–8–1
�17e
Chlorine2–8–7
Electrontransfer
FIGURE 3.2 Schematic of ionic bonding.
chloride is NaC1, since chlorine, being able toaccept one electron, also has a valency of 1.Oxygen, however, has six valence electrons andneeds to borrow two. Since sodium can onlydonate one electron, the chemical formula forsodium oxide is Na2O. Magnesium has twovalence electrons and so the chemical formula formagnesium chloride is MgCl2 and for magnesiumoxide MgO. Hence, it can be seen how thenumber of valence electrons determines howatoms combine to form compounds. The numberof valence electrons also strongly influences thenature of the interatomic bonds.
3.1 Ionic bondingIf an atom (A) with one electron in the outermostshell reacts with an atom (B) with seven electronsin the outermost shell, then both can attain theoctet structure if atom A donates its valence elec-tron to atom B. However, the electrical neutralityof the atoms is disturbed and B, with an extraelectron, becomes a negatively charged ion,whereas A becomes a positively charged ion. Thetwo ions are then attracted to each other by theelectrostatic force between them and an ioniccompound results, the strength of the bond isproportional to eAeB/r where eA and eB are thecharges on the ions and r is the interatomic sepa-ration. The bond is strong, as shown by the highmelting point of ionic compounds, and itsstrength increases, as might be expected, wheretwo or more electrons are donated. Thus themelting point of sodium chloride, NaCl, is801°C; that of magnesium oxide, MgO, wheretwo electrons are involved, is 2640°C; and that ofzirconium carbide, ZrC, where four electrons areinvolved, is 3500°C. Although ionic bondinginvolves the transfer of electrons between differ-ent atoms, the overall neutrality of the material ismaintained.
The ionic bond is always non-directional; thatis, when a crystal is built up of large numbers ofions, the electrostatic charges are arranged sym-metrically around each ion, with the result that Aions tend to surround themselves with B ions andvice versa (Figure 3.3). The pattern adopted
depends on the relative sizes of the A and B ions,i.e. how many B ions can be comfortably accom-modated around A ions whilst preserving thecorrect ratio of A to B ions.
3.2 Covalent bondingAn obvious limitation of the ionic bond is that itcan only occur between atoms of different ele-ments, and therefore it cannot be responsible forthe bonding of any of the solid elements. Whereboth atoms are of the electron acceptor type, i.e.with close to eight outermost electrons, octetstructures can be built up by the sharing of twoor more valence electrons between the atoms.
Thus, two chlorine atoms can bond togetherand achieve the octet structure by each contribut-ing one electron to share with the other atom.The interaction may be written:
Covalent bonding
27
FIGURE 3.3 Non-directional nature of the ionicbond: A� ions surrounded by B� ions, B� ions sur-rounded by A� ions. This particular arrangement,where each species of ion taken by itself lies on the sitesof a face-centred cubic structure, is known as thesodium chloride (NaCl) or rock salt structure.
as in one form of sulphur, where there is anobvious tendency for the atoms to form up inlong chains.
Structurally, the elements showing covalentbonding obey what is known as the (8�N) Rule.This states that the number of nearest neighboursto each atom is given by 8�N where N is thenumber of electrons in the outermost shell. Thuschlorine (N�7) has only one nearest neighbourand the atoms pair off as diatomic molecules;with sulphur, selenium and tellurium (N�6) longchains occur; with arsenic, antimony and bismuth(N�5) sheets of atoms occur, and with carbon(N�4) a three-dimensional network can form, asin diamond. Atoms with values of N less than 4cannot show covalent bonding with their ownspecies. They would require at least five nearestneighbours, whilst only having three or fewerelectrons available for making up the sharedbonds.
The possible arrangements of atoms are shownschematically in Figure 3.4, and it will be seenthat only carbon can achieve a three-dimensionalpattern in which all the bonds are covalent. Thisreveals the major structural difference betweenionic and covalent bonding: the covalent bond,unlike the ionic bond, is saturated by the indi-vidual atoms participating in it.
The chlorine molecule, which consists of twochlorine atoms, is structurally self-sufficient andthere is no extension of the covalent bondingbetween molecules. Similarly, the chains ofsulphur and the sheets of bismuth are really largemolecules but there is little bonding between thechains or sheets. The result is that the covalent
Cl� → ClCl Cl
O � O → O O
�S S � �S S � . . . → S S S S (3.3)
Where two electrons are required to make up theoctet, the interaction may either be:
as in the oxygen molecule, or:
(3.1)
(3.2)
elements (which effectively means the non-metal-lic elements) have poor physical strength, notbecause the covalent bond itself is weak, butbecause, with the exception of diamond (oneform of carbon), they do not form a three-dimensional lattice. In fact, the hardness and highmelting point of diamond (3500°C) show that thecovalent bond is extremely strong. Covalentbonding is not limited to elements; many com-pounds are covalent, some simple examples beingHCl, H2O, CH4 and NH3. A large number ofcompounds show partly ionic and partly covalentbonding, e.g. sulphates such as Na2SO4 in whichthe sulphate ion is covalently bonded whilst thebond to the sodium is ionic.
The vast field of industrial polymers is also pre-dominantly concerned with covalent compounds.Individual bonds may exhibit hybrid qualitiesakin to both ionic and covalent bonding;although in the pure form the two bond typesrepresent different modes of linkage, there is infact no sharp line of demarcation between them.When elements of high valency combine to formionic compounds (e.g. the nitrides and carbides ofthe transition elements), the donor ion may lose
Atomic structure and interatomic bonding
28
FIGURE 3.4 Schematic structures of elements con-forming to the 8�N rule. (a) Chlorine (N�7): indi-vidual molecules; (b) tellurium (N�6): spiral chains;(c) antimony (N�5): corrugated sheets; (d) diamond(N�4): three-dimensional crystal.
three, or even four, electrons to the acceptor ion.The result is a strong polarising pull (electrostaticforce) exerted by the donor ion on the electronsof the acceptor ion, so that the electrons aresucked back towards the donor and spend moretime between the two ions than circling each indi-vidually. The bond thus acquires some of thecharacteristics of the covalent linkage. Com-pounds of this sort are usually extremely hardand have very high melting points, since theycombine to some extent the strength of the cova-lent bond and the non-directionality of the ionicbond.
3.3 Metallic bondingMetallic atoms possess few valence electrons andthus cannot bond with themselves covalently,rather, they obey what is termed the free-electrontheory. In a metallic crystal the valence electronsare detached from their atoms and can movefreely between the positive metallic ions (Figure3.5). The positive ions are arranged regularly in acrystal lattice, and the electrostatic attractionbetween the positive ions and the free negativeelectrons provides the cohesive strength of themetal. The metallic bond may thus be regarded as
FIGURE 3.5 Schematic ‘free-electron’ structure fora monovalent metal. In the absence of an electric fieldthe electrons are in ceaseless random motion, but theoverall distribution remains uniform over any period oftime.
a very special case of covalent bonding, in whichthe octet structure is satisfied by a generaliseddonation of the valence electrons to form a‘cloud’ which permeates the whole crystal lattice,rather than by electron sharing between specificatoms (true covalent bonding) or by donation toanother atom (ionic bonding).
Since the electrostatic attraction between ionsand electrons is non-directional, i.e. the bondingis not localised between individual pairs orgroups of atoms, metallic crystals can grow easilyin three dimensions, and the ions can approachall neighbours equally to give maximum struc-tural density.
Crystallographically, these structures areknown as ‘close-packed’ (see Section 8.2). Theyare geometrically simple by comparison with thestructures of ionic compounds and naturallyoccurring minerals, and it is this simplicity thataccounts in part for the ductility (ability todeform non-reversibly) of the metallic elements.
Metallic bonding explains the high thermal andelectrical conductivity of metals. Since the valenceelectrons are not bound to any particular atom,they can move through the lattice under theapplication of an electric potential, causing acurrent flow, and can also, by a series of colli-sions with neighbouring electrons, transmitthermal energy rapidly through the lattice (seeChapter 7). Optical properties are also explainedby the theory. If a ray of light falls on a metal, theelectrons (being free) can absorb the energy of thelight beam, thus preventing it from passingthrough the crystal and rendering the metalopaque. The electrons which have absorbed theenergy are excited to high energy levels and sub-sequently fall back to their original values withthe emission of the light energy. In other words,the light is reflected back from the surface of themetal, and the high reflectivity of metals can beaccounted for.
The ability of metals to form alloys (of extremeimportance to engineers) is also explained by thefree-electron theory. Alloys are discussed inChapter 8.
Since the electrons are not bound, when twometals are alloyed there is no question of electron
exchange or sharing between atoms in ionic orcovalent bonding, and hence the ordinary valencylaws of combination do not apply. The principallimitation then becomes one of atomic size, andproviding there is no great size difference (seeSection 2.4.2), two metals may be able to form acontinuous series of alloys or solid solutions from100 per cent A to 100 per cent B.
3.4 Van der Waals bondsThe three strong primary types of atomic bond(ionic, covalent and metallic) all occur because ofthe need for atoms to achieve a stable electronconfiguration. However, even when a stable con-figuration exists already, as in the case of theinert gases, some form of bonding force betweenthe molecules must be present since these ele-ments will all liquefy and ultimately solidify atsufficiently low temperatures.
Bonds of this nature are universal to all atomsand molecules, but are normally so weak thattheir effect is overwhelmed when primary bondsare present. They are known as Van der Waalsbonds, and are one reason why real gases deviatefrom the ideal gas laws. They arise as follows.Although in Figure 3.1 we represented the orbit-ing electrons in discrete shells, the true picture isthat of a cloud, the density of the cloud at anypoint being related to the probability of findingan electron there. Such a picture implies that theelectron charge is ‘spread’ around the atom, and,over a period of time, the charge may be thoughtof as symmetrically distributed within its particu-lar cloud.
However, the electronic charge is moving, andthis means that on a scale of nanoseconds theelectrostatic field around the atom is continuouslyfluctuating, resulting in the formation of adynamic electric dipole, i.e. the centres of positivecharge and negative charge are no longer coinci-dent. When another atom is brought into proxim-ity, the dipoles of the two atoms may interactco-operatively with one another (Figure 3.6) andthe result is a weak non-directional electrostaticbond.
Van der Waals bonds
29
The attractive force between two atoms isgiven by the expression:
F� (3.4)
where �1 and �2 are the polarisabilities of the twoatoms, i.e. the ease with which their electronicfields can be distorted by a unit electric field toform a dipole, and r is the distance between them.Polarisability increases with atomic number, sincethe outermost electrons, which are responsible forthe effect, are further removed from the nucleusand therefore more easily pulled towards neigh-bouring atoms. This is shown by a comparison ofthe freezing points of the inert gases He (1K) andXe (133K), or the halogens, fluorine (51K) andiodine (387K).
As well as this fluctuating dipole, many mole-cules have permanent dipoles as a result ofbonding between different species of atoms.These can play a considerable part in the struc-ture of polymers and organic compounds, whereside chains and radical groups of ions can lead topoints of predominantly positive or negativecharges. These will exert an electrostatic attrac-tion to other oppositely charged groups.
The strongest and most important example ofdipole interaction occurs in compounds betweenhydrogen and nitrogen, oxygen or fluorine. Itoccurs because of the small and simple structureof the hydrogen atom and is known as the hydro-gen bond. When, for example, hydrogen links
�1�2�
r6
Atomic structure and interatomic bonding
30
FIGURE 3.6 Weak Van der Waals linkage betweenatoms arising from fluctuating electronic fields.
FIGURE 3.7 The hydrogen bond: (a) individualwater molecule showing dipole resulting from bondangle; (b) structure of water.
covalently with oxygen to form water, the elec-tron contributed by the hydrogen atom spendsthe greater part of its time between the twoatoms. The bond acquires a definite dipole withhydrogen becoming virtually a positively chargedion.
Since the hydrogen nucleus is not screened byany other electron shells, it can attract to itselfother negative ends of dipoles, and the result isthe hydrogen bond. It is considerably stronger(roughly �10) than other Van der Waals link-ages, but is much weaker (by 10 to 20 times) thanany of the primary bonds. Figure 3.7 shows theresultant structure of water, where the hydrogenbond forms a secondary link between the watermolecules, and acts as a bridge between two elec-tronegative oxygen ions. Thus, this relativelyinsignificant bond is one of the most vital factorsfor the benefit and survival of mankind. It isresponsible for the abnormally high melting andboiling points of water and for its high specificheat, which affords an essential global tempera-ture control. In the absence of the hydrogenbond, water might well be gaseous at ambienttemperatures like ammonia or hydrogen sulphide.
Chapter four
Elasticity andplasticity
4.1 Linear elasticity4.2 Consequences of the theory4.3 Long-range elasticity4.4 Viscoelasticity4.5 Plasticity
We noted in Chapter 1 that the product of stressand strain has the dimensions of energy per unitvolume. Since thermodynamics teaches us aboutenergy it is convenient to start with the thermo-dynamics of a deformation process. From the firstlaw, the internal energy is (equation (2.1)):
dE�dQ�dW
We can write the second law of thermodynamicsas:
dQ�TdS
and hence:
dE�TdS�dW
Let us consider a bar in which an applied tensileforce F results in an extension dL. The work doneon the bar is FdL and so the work done by thebar in resisting the force is
dW��FdL
� dE�TdS�FdL
The Helmholtz free energy (equation (2.4)) is:
H�E�TS
so that the change in free energy is:
dH�dE�TdS�SdT
However, if the conditions are such that tempera-ture remains constant, then SdT�0 and thechange is:
dH�dE�TdS�FdL
so that:
F� (�H/�L)T � (�E/�L)T �T(�S/�L)T (4.1)
where the subscript T indicates that the equationis valid for constant temperature.
This is the thermodynamic equation of state fordeformation and shows that the force is distrib-uted between a change in internal energy and achange in entropy. We now examine thesechanges.
4.1 Linear elasticityThe first term of equation (4.1) implies that atleast part of the response to an applied forceinvolves a change in internal energy. In crystallinesolids, where the atoms are arranged in orderlyranks in space, the capacity for independentmovement is limited and the bulk of the internalenergy is to be found in the interaction energiesbetween the atoms.
To keep things simple, let us consider the ener-gies involving a pair of atoms. There are bothattractive and repulsive forces which balance oneanother when the atoms are in equilibrium.Whatever the cause of the energies, they tend tovary as the inverse of the distance between theatoms, raised to some power. So, if the distancebetween the atoms is r, and A, B, m and n are
31
constants that vary with the material and itsstructure, then:
The attractive energy is Ar�n
The repulsive energy is Br�m
The resultant energy is U�Br�m �Ar�n
These energies, as a function of interatomicspacing, are sketched in Figure 4.1 (a). Figure4.1 (b) presents the same information, but interms of the force between adjacent atoms, F.There are three things to note:
1. The bond energy U is a continuous functionof r. Thus we can express the energy as aseries:
U(r) �U(r0) � r(dU/dr)r0� (r2/2)(d2U/dr2)r0
� . . . . . .
where U(r0) is the energy at r� r0, i.e. the inter-atomic separation at which the attractive andrepulsive forces balance, and the differential istaken at r� r0.
2. The minimum in the curve at r0 allows thesecond term to be eliminated, since(dU/dr)�0 at a minimum.
3. The displacement from r0 is small, so ignoreterms higher than r2. Then we find that:
U(r) �U(r0) � (r2/2)(d2U/dr2)r0
hence:
F�dU/dr� r(d2U/dr2)r0
i.e. force is proportional to displacement via aconstant (d2U/dr2).
In other words, the constant of proportion-ality is the slope of the F:r graph at the equi-librium position where r� r0.
4.2 Consequences of the theoryHaving established some mathematical factsabout the graphs of force and energy against dis-tance, we can use them to predict some con-sequences and tie them up with the real world.To a materials scientist there are a great manyconsequences but only eight of the most note-worthy are given below.
1. When a material is extended or compressed a
Elasticity and plasticity
32
Ener
gy U
(a)
r0
Ar�m
Br�m
(b)
Forc
e F
Attraction
Repulsion
Interatomicdistance r
r0
FIGURE 4.1 (a) Condon Morse curves of energyversus interatomic spacing, r (b) the force acting on anion as a function of r. r0 is the interatomic spacing.
little, the force is proportional to the exten-sion (from equation (1.8)). This is Hooke’slaw. The slope of the F:r curve at r� r0 is thefundamental origin of the elastic constant (orstiffness).
2. Since the F:r curve is nearly symmetricalabout the equilibrium position, the stiffness ofa material will be nearly the same in tensionand compression. This is, in fact, the case.
3. At large strains, greater than about 10 percent, the F:r curve can no longer be con-sidered straight and so Hooke’s law shouldbreak down. It does.
4. There should be a limit to the tensile strength,since the attractive force between the atoms hasa maximum value. This is so. It turns out thatthe maximum strength corresponding to themaximum in Figure 4.1(b) is about E/10. Forsteel this would be at �20000MPa, but weknow only too well that steels are nowherenear this strong. We will see why later. For themoment we note that the strain correspondingto maximum load is �1 (100 per cent). Some-thing else happens before we get there.
5. There should be no possibility of failure incompression since the repulsive force betweenthe atoms increases ad infinitum. This is so. Auniaxial compressive stress always gives riseto shear stresses which are maximised onplanes at 45° to the compression direction. Itis these shear stresses that cause failure.
6. If the atoms vibrate about their equilibriumpositions, then their mean separation willincrease the more they vibrate. This followsfrom the trough of the U:r curve being asym-metrical. In fact, in materials at any tempera-ture above absolute zero (�273°C), the atomsdo vibrate in proportion to the temperature. Itfollows then that as a material is heated itshould expand in all directions. This is so.
7. If the heating process is continued the atomsshould reach a degree of vibration that causesthem to separate, i.e. the bond is broken. Thisis what happens when a solid evaporates.
8. Any degree of vibration by the atoms shouldreduce the extra amount of energy required tobreak the bond. In other words the tensilestrength should decrease as the temperatureincreases. This does happen.
There is another point to make, which the readermay have noticed already. Points 4 and 5 both
Long-range elasticity
33
FIGURE 4.2 Polyethylene. (a) Diagrammatic. In (b)X represents two hydrogen atoms, one above and onebelow the plane of the carbon chain.
contain intimations that other things are going onin materials when they are strained. Later, weshall find out what.
4.3 Long-range elasticityWe now consider the second term in equation(4.1). Specific polymers will be discussed inChapter 34, but meanwhile we will consider thebasic nature of polymer structure. In its simplestform this consists of a chain of atoms, oftenmany units long and generally, though notnecessarily, of carbon, each of which is covalentlybonded to another atom. The structure of poly-ethylene is shown diagrammatically in Figure4.2 (a).
However, things are not quite so simple. Twoadjacent carbon bonds must be at a fixed angle toone another (Figure 4.2 (b)), but there is no con-straint about which plane the angle lies in. Theconsequence is that over a distance of tens orhundreds of carbon bonds the chain tends to gettwisted. Second, the long chain is a complete mol-ecular entity and, in the simplest case, is notbonded to another chain. We therefore have astructure which looks rather like a tangle of wetspaghetti, or a can of worms. In addition, at tem-peratures above a certain level, which is a charac-teristic property of the polymer, the chains are inconstant motion as a result of thermal fluctua-tions. It is intuitively obvious that the applicationof a force will cause changes in the internal con-figuration. A given atom in the molecular chainwill not change its neighbours within the chain
but it will change its neighbours in adjacentchains several times during the loading–unloadingcycle. Think of the worms wriggling about.
Expressed in thermodynamic terms we may saythat the unloaded configuration is, or is close tobeing, the most probable one, in other words thisis the equilibrium configuration of maximumentropy. Under load the configuration changes toone that is the most probable for the loaded con-dition and the entropy changes. A straight chainis only one of many possible configurations, andit can be shown that the number of possible con-figurations, whether convoluted or not, is a func-tion of its end-to-end length, r. The number ofpossible configurations Nc decreases as rincreases. This is known as the configurationalentropy Sc and is given by Boltzmann’s equation(equation (2.2)) here rewritten as:
Sc �klnNc (2.2a)
i.e. the entropy decreases as r increases.We shall not pursue here the statistical argu-
ments describing the probability of a particularconfiguration being obtained. This is done fully instandard texts (Treloar, 1949) and here we stateonly the principles.
1. If the ends of the molecule are moved from
Elasticity and plasticity
34
FIGURE 4.3 Schematic representation of a ran-domly convoluted molecular chain.
their most probable positions, the change inentropy causes a force to act along the linejoining the ends of the molecule (Figure 4.3).
2. The entropy force is proportional to theabsolute temperature. The magnitude of theforce is a function of the end-to-end length asrepresented by a straight line joining the ends.In this it behaves as though it were an elasticspring.
The spring analogy should not, however, betaken too literally. Being statistical in origin thetension in a chain whose endpoints are fixed willbe subject to continual fluctuations like the pres-sure exerted by a gas on the walls of a container.Similarly, if the molecule is subjected to a con-stant tension its length will fluctuate. But we seethat rubber elasticity is long range and is quite adifferent thing from the short-range forces whichdetermine Hookean elasticity. For rubbers weoften speak of the entropy spring.
4.4 ViscoelasticityWe have already indicated that it is not possible todraw a sharp dividing line between the mechanicalbehaviour of liquids and solids. There is a largegroup of materials, known as viscoelastics, whosebehaviour is part liquid and part solid. Manynatural materials, tendon, plant fibres and woodall behave in this way. Of engineering materials,rubbers are the archetype and so are many softpolymers as well as substances like tar and asphalt.
Two cases at least are of importance to engi-neers. If the strain remains constant, as in atension bolt clamping two girders together, stressrelaxation may occur without dimensionalchange. When the material is under constantstress it will respond by steadily increasing strainand, under prolonged loading, many normallyrigid materials such as cement and concrete areprone to this process, i.e. to creep.
If an instantaneous stress is applied to aHookean material (Figure 4.4 (a)), a correspond-ing strain is produced. As long as the load ismaintained the strain remains constant (Figure4.4 (b)) and is fully recovered when the load is
Viscoelasticity
35
Time
(a) Load history: constant applied stress
Load Unload
Str
ess
Time
(c) Viscoelastic (creep) response
Time
(b) Elastic (Hookean) strain response
Load Unload
Str
ain
Time
(d) Strain history: constant applied strain
Stretch Relax
Str
ain
Time
(f) Viscoelastic response – stress relaxation
(e) Elastic (Hookean) stress response
Time
Str
ess Stretch Relax
Str
ain
Str
ess
FIGURE 4.4 Types of materials’ response to constant stress (a–c) and constant strain (d–f).
removed. The corresponding strain behaviour ofa viscoelastic material is illustrated in Figure4.4 (c). The strain builds up over a period of timeand, when the load is removed, the strain takestime to relax. This behaviour is known as creep.A similar phenomenon occurs if a material is sub-jected to a fixed strain (Figure 4.4 (d)). In aHookean material stress remains constant for as
long as the strain is maintained (Figure 4.4 (e)); ina viscoelastic material the stress diminishes over aperiod of time, as shown in Figure 4.4 (f). This isstress relaxation.
The structural reasons for this behaviour neednot delay us here as they will be discussed in laterchapters dealing with specific materials, cements,polymers and timber especially. Meanwhile we
Elasticity and plasticity
36
�
(a) Maxwell
�
(b) Voigt
sS � EeS
sF � HdeF
dt
sS � EeS sF � HdeF
dt
FIGURE 4.5 Viscoelastic models.
note that the behaviour is usually modelled usingmechanical analogues consisting of arrays ofsprings which behave according to Hooke’s lawand viscous elements which behave as an idealNewtonian liquid.
One such array is shown in Figure 4.5(a). It con-sists of an elastic spring s of modulus E in serieswith a dash pot, i.e. a piston moving in a fluid F ofviscosity contained in a cylinder. Now think ofsuddenly applying a constant strain. At first all thestrain is taken up by stretching of the spring andthe load required to do this is calculated from thestrain of the spring. Later, the spring shortens bypulling the piston up through the fluid in the dashpot. Some of the total strain is now taken up by themovement of the piston and less by the stretch inthe spring. The load required is now less thanbefore, and thus the system is exhibiting stressrelaxation. Mathematical analysis gives:
�t � �0exp(�t/�) (4.3)
where �0 is the stress sustained at time� 0,�t isthe stress sustained at time t and � �E/ is the so-called relaxation time. Under constant strain thestress decays exponentially.
Now think of doing things differently. Apply aconstant stress. The spring stretches and remainsat that strain as long as the load remains. At thesame time the dash pot slowly extends as thepiston is pulled through the fluid in it. The totalextension increases linearly with time. This model
represents stress relaxation very well, but is lesssuccessful at representing creep.
To get a good representation of creep, we use amodel (Figure 4.5 (b)) in which the spring anddash pot are arranged in parallel. Both elementsmust experience the same strain but load can betransferred over time from one element to theother. Unfortunately, this gives a poor representa-tion of stress relaxation.
To get out of these difficulties, the two typesare combined into what is known as the fourelement model. This and its responses are shownin Figure 4.6. In this, �M is the relaxation time forstress at constant strain and �V is the relaxationtime for strain at constant stress.
The concept of relaxation times is importantfor two reasons. First because it helps us to dis-tinguish between solids and liquids. A perfectsolid will support the stress indefinitely, i.e. � � �,but for a liquid, relaxation is virtually instanta-neous (for water � ��10�11s). In between there isa grey area where stress relaxation may occurover a few seconds or centuries.
Then we have the relationship between therelaxation time and the time scale of the loadingt. If the load is applied so fast that the relaxationcannot occur (t� �) the material will effectivelybehave elastically, but under slow loading (t� �)it will flow. This is well shown by ‘potty putty’which bounces under impact loading but col-lapses into a puddle under its own weight whenleft alone. Many polymeric materials show asimilar sensitivity to loading speed.
To model a real material, we would need tocombine large numbers of simple elements, eachhaving its own relaxation time, to achieve arelaxation spectrum in which different elementsrelax at different times.
There are a number of important consequencesof viscoelasticity. The first is that the stress–strainrelationship is non-linear. We noted that in aHookean solid the strain energy stored onloading is completely recovered when we unload.Figure 4.7 shows that for a viscoelastic materialthe energy recovered on unloading is less thanthat stored during loading. Where has the energygone? It has gone into heat, which explains why,
after a few miles in which they are repeatedlyloaded and unloaded, car tyres get hot.
The second consequence is known as Boltz-mann’s superposition theory. This states that eachincrement of load makes an independent and addi-tive contribution to the total deformation. Thus,under the loading programme shown in Figure 4.8,the creep response is additive and the total creep isthe sum of all the units of incremental creep. Thishas certain consequences when we apply it to thecreep behaviour of concrete and soils.
4.5 PlasticityMany materials display plasticity; that is to say,when they are deformed the change in shape ispermanent and the original dimensions can onlybe restored by applying a force in the opposite
Plasticity
37
Time
ec
Total
Maxwell
Constant stress
stra
in e
Time
Constant strain
Str
ess
se0 � s/EM
e1
e0
e1 � e0 � eV∞ � s/EV
slopet ∞ � sEMtM
�shM
�
s0 � s∞ � EMe
s∞ � EV e
s � s∞
s0 � s∞� exp �t
tM
sEV
eC � 1 � exp �ttV
s0
s∞
Max
wel
l
�
Voig
t
tM �hM
EM
M
eM EM
eT
VEVeV
tV �hV
EV
FIGURE 4.6 Fourelement model of vis-coelasticity and itsresponse.
FIGURE 4.7 Loading/unloading curves for a visco-elastic solid.
direction. This is a particularly valuable attributeof metals and we shall discuss the reasons inChapter 9. Meanwhile we consider the moregeneral case as exhibited, for instance, by clays.
We have noted already that the ideal (Newton-ian) fluid flows according to:
� � (�/�t) (4.4)
where is the viscosity. When the fluid containsdispersed particles these perturb the pattern offlow and, for dispersed spheres, the viscosity isnow given by Einstein’s equation:
� 0[1�5Vf /2] (4.5)
where 0 is the viscosity of the pure fluid and Vf
the volume fraction of particles.The actual viscosities of suspensions are usually
much higher than this because the particles arenon-spherical. Thus, for randomly orienteddumb-bells,
� 0(1�3p2Vf /2) (4.6)
where p is the ratio of dumb-bell length to dia-meter.
The above equations break down still furtherwhen the volume fraction of the particlesincreases to the point where the perturbed regionsin the liquid begin to overlap and we find termsin V f
2 appearing. We are now in the region ofpastes, clays, sand, gravel, etc., where the volumefraction of grains is high. Typically, Vf �0.65 for
Elasticity and plasticity
38
FIGURE 4.8 Boltzmann’s superposition principle.S
hear
str
ess
�
Rate of strain (�/�t)
�yB
Bingham behaviour:intercept � yield stress (�y)slope � plastic viscosity (�)
Newtonian behaviour:slope � viscosity ( )
A
B
X
�yA
FIGURE 4.9 Newtonian and Bingham fluid behavi-our.
a sand bed and Vf �0.95 for a mixed sand/gravelbed where the sand can fill the spaces between thegrains of gravel.
Clearly the mechanical properties are sensitiveto the volume fraction of grains, dry claybecomes plastic when water is added to it. At 30per cent water content, each clay particle is sepa-rated by a film of water and it can be modelledinto shapes. With more water added the grainsfloat apart to form a slurry.
Pastes and modelling clays, however, areplastic solids rather than viscous fluids. Theydeform, more or less elastically, up to a certainyield stress and can preserve their shape againstgravity. This is important to the moulding ofbricks and clay pottery. Above this stress,however, they behave like liquids and deformrapidly. This behaviour is described, approxi-mately, by the Bingham equation:
(�/�t)� (���y)/ (4.7)
where �y is the yield stress as shown in Figure 4.9. Tattersall and Banfill (1983) have shown thatfresh concrete can be considered as a so-called‘Bingham solid’ and instance the differencebetween two concrete mixes having the flowcurves shown in Figure 4.9. At low rates of flow,B is less workable than A; the reverse is true athigh rates of flow. At X both would be classed ashaving the same workability.
Chapter five
Surfaces
5.1 Surface energy5.2 Water of crystallisation5.3 Wetting5.4 Adhesives5.5 Adsorption
All materials are bounded by interfaces of varyingnature. For the engineer the most important arethe liquid–vapour, solid–vapour, solid–liquid andsolid–solid interfaces. The latter exists as theboundary between two differing solid phases in amaterial, e.g. cement gel and aggregate in con-crete, or between two similar crystals which differonly in orientation, e.g. the grain boundaries in apure metal, or, at the macroscopic level, as theinterface between structural components, e.g.concrete and steel. Surfaces owe their interest andimportance to two simple features: they are areasof abnormality in relation to the structure thatthey bound, and they are the only part of thematerial accessible to chemical change, i.e. allchemical change and, for that matter, most tem-perature changes take place through the media ofsurfaces. The importance of surfaces in determin-ing the bulk behaviour of materials naturallydepends on the ratio of surface area to the totalmass. This, in turn, depends partly on the sizeand partly on the shape of the individual particlesmaking up the bulk material.
Surface influence on bulk behaviour may reachits zenith with clays, which are composed ofplatelets that may be as small as 0.01�m thick by0.1�m across. These platelets have a high surfacearea/volume ratio by comparison with spheres ofequal volume. One gram of montmorillonite clay,rather smaller than a sugar cube, may contain a
total surface area of over 800m2! Porous struc-tures such as cement and wood also containenormous internal surface areas that exercise aconsiderable effect on their engineering proper-ties.
5.1 Surface energyAll surfaces have one thing in common: the atomsor ions in the surface are subjected to asymmetricor unsaturated bonding forces; this is particularlythe case with solid–vapour and liquid–vapourinterfaces. Since bonding is taken to lower theirenergy (Figure 4.1) the surface atoms or ions willbe in a state of higher energy than interior ones.This excess energy is known as the surface energyof the material. In solids the presence of thesurface energy is not immediately apparent, sincethe atoms in the surface are held firmly in posi-tion. However, with liquids the mobile structurepermits the individual atoms to respond, and theresult is the well-known surface tension effect.Since surfaces are high-energy regions they willalways act to minimise their area, and thus lowertheir energy, when possible. If a soap film isstretched across a frame with a movable wire asin Figure 5.1, the force required to hold the wirein place is:
F�2l (5.1)
where l is the length of the wire, is the surfacetension of the soap film/air interface and thefactor 2 is introduced because the film has twosurfaces. The surface tension is in the plane of thesoap film/air interface and has units of force perunit length (N/m). It is important to note that
39
surface tension differs from an elastic force actingbetween the surface atoms in that it remains con-stant whether the film is forced to expand orallowed to contract. This is because the workdone in expanding the film is used to bring addi-tional atoms to the surface rather than to increasethe interatomic spacing in the surface. Only whenthe film has become so thin that the two surfacesinteract with each other will the force showpartial elastic behaviour, by which time the film ison the point of rupture.
If the film is stretched by pulling the movablewire through a distance d, the work done on it2ld is stored as surface energy of the newlycreated surface of area 2ld, therefore the surfaceenergy per unit area � is:
� ��2
2
l
ld
d�� (5.2)
Surface tension and surface energy are numeri-cally equal. The unit of surface energy is Jm�2
(�Nm�1). It should be noted that tends to beused interchangeably for both surface tension andenergy.
5.2 Water of crystallisationIt is well known that many ionic crystals containwater molecules locked up in their structure aswater of crystallisation; such crystals are knownin general as hydrates, and their formation can bevery important in the development of bulkstrength. Both cement and ‘plaster of Paris’ owetheir commercial importance to their ability totake up water and form a rigid mass of interlock-
Surfaces
40
FIGURE 5.1 Equilibrium between soap film andapplied force F.
ing crystals. However, in the case of cement thecrystals are so small that it is difficult to decidewhether to classify the structure as a crystallinehydrate or a hydrated gel.
Water and ammonia are the only two mole-cules that can be taken up as a structural part ofcrystals, and both for the same reasons: they aresmall molecules which are strongly polar. Thesmall size permits them to penetrate into theinterstices of crystal structures where closepacking of ions is not possible. This is particu-larly the case where the negative ion is large, suchas SO4 (sulphate), SiO4 (silicate) or B4O7 (borate).It must be emphasised that this process is not tobe thought of as a capillary action, analogous tothe take-up of large amounts of water by clays.The water molecules are bonded into definite siteswithin the crystal structure, and the crystal willonly form a stable hydrate if ions of the appropri-ate signs are available and correctly placed toform bonds with the positively and negativelycharged regions of the water molecule. We havealready mentioned (page 30) the abnormal prop-erties of water arising from the ability of themolecules to link up by means of hydrogenbonds; the formation of crystalline hydrates is anextension of the same behaviour.
Normally the water molecules cluster roundthe positive ions in the crystal-forming hydratedions. This has the effect of making the smallpositive ion behave as if it were a good deallarger. As a result, the size difference between thepositive and negative ions is effectively reduced,thus making possible simpler and more closelypacked crystal structures. Water bonded in thismanner is very firmly held, in many instances, sothat hydration becomes virtually irreversible.Cement, for example, retains its water of crys-tallisation up to temperatures of �900°C.
5.3 WettingThere are other ways besides reduction in area bywhich a system of interfaces may reduce its totalsurface energy. One of the most importanttechnological aspects concerns the behaviour ofliquids on solids.
If a droplet of liquid is placed on a solid, theimmediate behaviour of the liquid depends on therelative magnitudes of three surface tensions(energies): liquid–solid ls, liquid–vapour lv, andsolid–vapour sv. At the periphery of the dropletthe three tensions operate as depicted in Figure5.2 and the final contour of the droplet will resultwhen they are in equilibrium, resolved parallel tothe solid surface, that is when:
sv � ls � lv cos � (5.3)
The difference:
sv � (ls � lv) (5.4)
is termed the spreading force and if it is positivethen � �0°, and complete wetting of the solidsurface occurs. The energy of such a system isobviously lowered when the solid–vapour inter-face is replaced by a solid–liquid and aliquid–vapour interface.
When sv � (ls � lv) is negative then � �0° andpartial or no wetting occurs. If sv �ls then lv cos � (equation 5.3) is positive and � �90°giving partial wetting. If sv � ls (this is compara-tively rare, provided the surfaces are clean) thenlv cos � (equation 5.3) is negative and � �90°giving little or no tendency to wetting.
The rise of water in a capillary tube is a con-sequence of the ability of water to wet glass. If, inFigure 5.3, � is the angle of contact betweenwater and glass, the water is drawn up the tubeby a circumferential force 2rlv cos �, so that:
2rlv cos � � r2h� (5.5)
where r2h� is the mass of water in the capillary,neglecting the mass of water contained in the
Adhesives
41
Vapour
Liquid
Solid
�
sv ls
lv
FIGURE 5.2 Surface forces acting at the peripheryof a droplet.
curve of the meniscus. It follows that the heightof the water in the capillary is:
h��2lv
�
c
r
os �� (5.6)
If r is small, h can be very large. If all the pores inbrick or concrete were continuous, h�10m ispossible. Rising damp indeed! Fortunately thepores are not continuous and evaporation keepsthe level lower than this.
5.4 AdhesivesThe ability of adhesives to spread and thoroughlywet surfaces is highly important, e.g. where glue,molten solder or brazing alloy has to penetrateinto a thin joint. Furthermore, the adhesion of aliquid to a solid surface is relevant to theperformance of adhesives. The work to breakaway the adhesive (which may be considered as aviscous liquid) from the solid is the work requiredto create a liquid–vapour and a solid–vapourinterface from an equivalent area of liquid–solidinterface, i.e. it is the work to totally ‘de-wet’ thesolid surface. Hence the work to cause breakageat the interface, per unit area, is given by:
W� lv � sv � ls (5.7)
But from equation (5.3):
sv � ls � lv cos � (5.8)
FIGURE 5.3 Capillary rise of liquid up a tube.
and therefore:
W� lv(1�cos �) (5.9)
Thus, the liquid–solid adhesion increases with theability of the adhesive to wet the solid, reaching amaximum, when � �0° and wetting is complete,given by:
W�2lv (5.10)
For this to be the case sv �lv (equation 5.8) andunder these conditions fracture will occur withinthe adhesive, since the energy necessary to formtwo liquid–vapour interfaces is less than that toform a liquid–vapour and a solid–vapour interface.
Surface tension is also the cause of the adhe-sion between two flat surfaces separated by a thinfilm of fluid. Where the surface of a liquid iscurved (as for example in Figure 5.4) there will bea pressure difference p across it; if the curvatureis spherical of radius r, then:
p� �2
r
� (5.11)
In the case of two circular discs, however, thesurface of the film has two radii of curvature, asshown in Figure 5.4; r1 is approximately equal tothe radius of the discs and presents a convexsurface to the atmosphere whilst r2 �d/2, where dis the thickness of the film between the plates,and presents a concave surface to the atmosphere.The pressure difference between the liquid filmand its surroundings is now given by:
p� � �� � � (5.12)
If d� r1 then:
p� ��2
d
� (5.13)
the negative sign indicating that the pressure islower within the liquid than outside it.
Since the pressure acts over the whole surfaceof the discs, the force to overcome it and separatethe discs is given by:
F��2
d
r21
� (5.14)
2�d
1�r1
1�r2
1�r1
Surfaces
42
FIGURE 5.4 Adhesive effect of a thin film of liquidbetween two flat plates. If the liquid wets the surfacesof the discs, then r2 �d/2.
The magnitude of F thus depends on the factorr2
1/d, and it is therefore important that surfaces tobe joined should be as flat and closely spaced aspossible. Any reader who has tried to pull aparttwo wet beer glasses will confirm how tenaciouslythey cling to each other; by contrast, however,they can easily be slid apart since liquid filmshave little resistance to shear. If d�0.01mm,r�100mm and (water)�0.073Nm�1 thenF�460N. This value of F for a liquid film givessome idea of the potential of adhesives that gainadditional strength and rigidity by setting tohighly viscous materials on polymerisation orsolvent evaporation.
5.5 AdsorptionThe ability of liquids to wet solids depends verymuch on the cleanliness of the solid, as anyonewith any experience of soldering will appreciate.The presence of any dirt, such as oxide or greasefilms, will totally alter the balance of surface ten-sions discussed above and usually preventswetting.
Clean surfaces, in fact, are so rare as to be vir-tually non-existent, since the broken surfacebonds will readily attract to themselves anyforeign atoms or molecules that have a slightaffinity for the surface material. This effect isknown as adsorption, and by satisfying or par-tially satisfying the unsaturated surface bonds itserves to lower surface energy. Adsorption is a
dynamic process, i.e. molecules are constantlyalighting on and taking off from the surface.
Different molecules adsorb with varyingdegrees of intensity, depending on the nature ofthe bond that is able to form at the interface, andthe strength of the bond may be expressed interms of �a, the energy of adsorption. As in thecase of interatomic bonds, a negative value of �a
is taken to indicate positive adsorption, i.e. themolecules are attracted to the interface, and thesurface energy (tension) is lowered thereby. Apositive value of �a indicates a repulsive inter-action and the molecules avoid the surface.Typical plots of �a against the distance of theadsorbed layer from the surface are given inFigure 5.5. They closely resemble theCondon–Morse curves (Figure 4.1) and theirshape is due to the same circumstance of equilib-rium between attractive and repulsive forces,although the attraction is far weaker than that ofthe principal interatomic bonds.
If the molecule being adsorbed is non-polarand does not react chemically with the surface,adsorption, if it occurs, will be by Van der Waalsbonds, and the minimum value of �a is small(curve 2 in Figure 5.5). If on the other hand themolecule is strongly polar, as is the case withwater or ammonia, the electrostatic forcesbetween the surface and the charged portion ofthe molecule give rise to stronger bonding. If achemical reaction occurs as part of the bondingmechanism, e.g. when fatty acid in a lubricantforms an adsorbed layer of metallic soap on ametal surface, the bonding is still stronger (curve1) and the effect is referred to as chemisorption.
The behaviour of water is of particular import-ance in this context. Because of its ability to formhydrogen bonds with neighbouring molecules,water adsorbs rapidly and strongly on most solidsurfaces. Despite the tenacity with which such alayer is held (clay does not lose all its adsorbedwater until heated to 300°C), the interactioncannot be thought of as chemisorption; rather itis, in a sense, a halfway stage to solution or alter-
Adsorption
43
FIGURE 5.5 Energies of adsorption for differentadsorption mechanisms: curve 1, chemisorption; curve2, physical adsorption.
natively to the taking up of crystalline water ofhydration. Bonding is strong enough to maintaina surface layer perhaps several molecules thick,but the affinity is not sufficient for the moleculesto penetrate into the interstices of the structure.
The physical nature of such a film is difficult tovisualise; it cannot be thought of as a fluid in theaccepted sense of the term even when more thanone molecule thick, as in the case of the clays andcements. Yet the molecules are mobile in this situ-ation. They will not desorb readily, but they candiffuse along the surface under the impetus ofpressure gradients.
Such movements, occurring over the vastinternal surface area of cement gels, are primarilyresponsible for the slow creep of concrete understress. The ability of water molecules to penetratesolid–solid interfaces in clays and build up thickadsorbed layers results in the swelling of claysand has caused, and may well continue to cause,considerable structural damage to buildingserected on clays that are liable to behave in thismanner. The readiness with which water willadsorb on surfaces is turned to account in the useof porous silica gel and molecular sieves as dryingagents.
44
Chapter six
Fracture andfatigue
6.1 Brittle fracture6.2 Ductile fracture6.3 Fracture mechanics6.4 Fatigue
The term ‘fracture’ is often used synonymouslywith the term ‘failure’ but the two are notnecessarily describing the same process. Failure,in the sense that a component is rendered unfitfor further service, can occur without fracture; as,for instance, by excessive deformation or byreduction of load-supporting area due to wear orcorrosion. Fracture is the separation of a compo-nent into two or more pieces under the action ofa static or slowly changing imposed load, at tem-peratures that are low compared with the meltingtemperature of the material.
Separation of a component into two or morepieces can also occur under the action of animpact (suddenly applied) load; a fluctuatingload, fatigue; a static load with the component ata temperature above one-half of its melting tem-perature, creep rupture; or the combined actionof an imposed load and a chemical reaction,stress corrosion cracking.
Brittle fracture occurs when a material absorbslittle or no energy prior to fracture. Ductile frac-ture requires a material that can experienceappreciable plastic (i.e. irreversible) deformationand energy absorption prior to fracture. Thetensile stress–strain behaviour for brittle andductile fractures is shown in Figure 6.1. The areaunder the stress–strain curve up to fracture is ameasure of the energy absorbed per unit volume
of material, and is termed the toughness of thematerial. Ductile materials exhibit much highervalues of toughness than do brittle materials.Figure 6.2 illustrates the extent of deformationassociated with brittle and ductile fractures. Inthe case of ductile fracture, the specimen canexperience a considerable change in shape priorto fracture.
We shall confine our attention here to ductileand brittle fractures, and to the process offatigue. Furthermore, we shall only consider uni-axial stressing, although the theory can beextended to cover the case of multiaxial stressing.
The strength of a material is expected to be afunction of the interatomic bonding forces. As wehave seen in Chapter 4, various approaches based
Strain
Str
ess
Ductile
Brittle
Frac
tureToughness
Yield stressFr
actu
re
FIGURE 6.1 Schematic of tensile stress–straincurves for brittle and ductile materials taken to frac-ture.
on the Condon–Morse curves (Figure 4.1) predictthat the strengths of materials should be of theorder of 10 per cent of their elastic modulusvalues, i.e. �E/10. However, real materialsexhibit very much lower values than this. Brittlematerials, i.e. glasses and ceramics, under mostconditions, fracture at stresses in the region ofE/1000, with little deformation. Ductile mater-ials, i.e. most metallic alloys and many polymers,start to deform irreversibly at stresses aroundE/1000 and fracture at stresses between �E/1000and E/100.
6.1 Brittle fractureIn order to explain the low observed strengths ofbrittle materials, A.A. Griffith, working onglasses, proposed that the presence of smallcracks in such materials controlled their strength.The effect of the cracks is to concentrate stress inthe regions around their crack tips so that,although the average stress is very much lowerthan E/10, the stress at the tips is �E/10 whenfracture occurs.
Griffith adopted a thermodynamics approach,i.e. for a crack to grow under a remotely applieduniform static stress, the rate of release of strain
Brittle fracture
45
��
� �
Beforeloading
Ductilefracture
Necking aftermax load
Brittlefracture
FIGURE 6.2 Schematic of brittle and ductile frac-tures.
energy accompanying crack growth must exceedthe rate of absorption of energy. The inferencebeing that when the thermodynamic criterion issatisfied for crack growth the forces carried bythe atomic bonds at the crack tips are sufficientfor them to rupture. For brittle materials, energyis absorbed in the creation of additional fracturesurfaces.
Consider a thin plate of unit thickness contain-ing a central crack of length 2a under an applieduniform stress � (Figure 6.3). The strain energyreleased by an elliptical region of semi-axes a and2a (Figure 6.3) if completely relaxed (unstressed)due to the presence of the crack is:
UR � (2a2) (6.1)
and the energy absorbed by a crack of length 2ais:
UA �4a (6.2)
where is the surface energy. The factor of 4takes account of the two crack tips and the twofracture faces. Although the concept of an ellipti-cal region being completely relaxed by the pres-ence of the crack is unrealistic, it turns out thatthe values predicted by equation (6.1) are quitereasonable.
�2
�2E
�
�
4a
2a
Crack
Unstressedregion due to
crack
FIGURE 6.3 Schematic of a crack in a uniformlystressed plate.
For crack growth to be possible requires:
� (6.3)
i.e. the rate of release of strain energy to exceedthe rate of absorption. Differentiating equations(6.1) and (6.2) and substituting in equation (6.3)gives:
�4 (6.4)
and
� ��� (6.5)
Inspection of equation (6.4) shows that for agiven applied stress there is a critical value ofcrack length for crack growth to commence. Fur-thermore, once crack growth commences the rateof release of strain energy increases continuouslywith increasing crack length, whereas the rate ofabsorption of energy remains unchanged. Con-sequently, crack growth can become very rapidapproaching the speed of sound in the material.The large excess of energy released may alsoresult in crack splitting and branching. Theminimum value of applied stress required for acrack of length 2a to grow is given by equation(6.5). As might have been expected, this predictsthat the ‘brittle tensile fracture strength’ decreaseswith increasing crack length.
Brittle materials do not exhibit a single value oftensile strength. Real materials contain manyflaws (cracks), the size and distribution of whichare random, so that the fracture stress is deter-mined by the size and location of the largestdefect in the sample. Since no two samples areexactly alike, if fifty samples are tested the meas-ured strengths will show an approximate Gauss-ian distribution (Figure 6.4 (a)). The concept of amean strength is sometimes useful, but it must beremembered what it actually means. If the datais replotted as a cumulative probability curve(Figure 6.4 (b)) the mean stress simply means thestress at which 50 per cent of the samples may beexpected to fail. Not a very good design parame-ter! There are ways of dealing with the problem
2E�a
2�2a�
E
dUA�da
dUR�da
Fracture and fatigue
46
FIGURE 6.4 Strength of brittle solids: (a) distribu-tion of measured strengths; (b) cumulative probabilityof failure under load.
but, of course, the safest thing to do would be tonever subject these materials to tension forces.
Fracture in the way described above cannotoccur in compression since the applied forceswould tend to close the cracks rather than openthem. The mechanism is much more complicatedand many materials such as glass, stone, brickand concrete are strong in compression, althoughthey are weak in tension. The classical arched anddomed styles of architecture testify to the strengthof brittle materials in compression.
6.2 Ductile fractureAs indicated earlier, ductile fracture involvesappreciable plastic (irreversible) deformationbefore fracture occurs. Plastic deformation com-mences at stresses around E/1000, or greater, andfailure occurs at stresses between �E/1000 andE/100, i.e. at stresses still appreciably less than
E/10. Why is this? The crystalline, i.e. periodic,structure of ductile crystalline metals and poly-mers exhibit many local disruptions at the atomicscale, known formally as dislocations (seeChapter 8). These dislocations are able to moverelatively easily under applied stresses of�E/1000. This movement allows the shape of amaterial, not its volume, to change relativelyeasily, i.e. it enables plastic deformation. Thestress at which plastic deformation first occurs istermed the yield stress or yield strength (Figure6.1).
Ductile fracture occurs when a particularregion of a material suffers excessive plasticdeformation and necks down sufficiently for thenecked region to no longer be able to sustain theapplied force (Figure 6.2). The process of neckingis aided by the formation of internal necksaround impurity particles, which lead to the typeof fracture surface indicated schematically inFigure 6.2.
6.3 Fracture mechanicsIt was generally thought at one time that the Grif-fith approach applied only to elastic materialscontaining flaws, i.e. glass, concrete, fired clayproducts, ceramics, etc. However, doubts startedto arise when structures, including welded steelships, oil rigs, pressure vessels and aircraft, manu-factured from ductile materials, fractured in abrittle manner.
For ductile materials the work involved withplastic deformation as the crack grows is consid-erably greater than the surface energy term, i.e.2 per crack tip. Hence, it is necessary to replace in equations (6.2), (6.4) and (6.5) by anotherterm Gc, the fracture energy, also known simplyas the toughness. This is the total energyabsorbed per unit increment of crack growth andhas units of J/m2. Table 6.1 gives some typicalvalues and it will at once be apparent whycopper, say, is tough and glass is not.
Equation (6.5) becomes:
� ���EGc�a
Fracture mechanics
47
TABLE 6.1 Fracture energy Gc and fracture tough-ness Kc for common construction materials
Material Gc(J/m2) Kc(MN/m3/2)
Pure, ductile metals 100–1000 100–350High-strength steels 15–120 50–1500.2% C plain carbon steel 100 150Glass fibre reinforced polymers 10–100 20–50Polyethylene 6–8 1Reinforced concrete 0.2–4.0 10–15Common timbers* 0.5–2.0 0.5–1.0Concrete 0.03 0.2Glass 0.01 0.8
*Crack parallel to grain.
or, as it is more generally written:
�E�G�c� ����a� (6.6)
The left-hand side of this equation is materialsdependent and the right-hand side includes theengineering factors. It is clear that the criticalcombination of crack length and stress at whichfast fracture occurs is a material constant.
The term ���a� is abbreviated to a singlesymbol K, having units of N/m3/2, and is calledthe stress intensity factor. Fast fracture occurswhen K reaches a critical value Kc, known as thefracture toughness.
In the case of ductile materials, Gc is tens oftimes larger than surface energy , so that thecrack sizes necessary for the brittle fracture ofductile materials to occur are generally very muchlarger than for brittle materials. However, suchsizes can be realised due to the presence of largeimpurity particles, holes, notches and, as we willsee in the next section, fatigue cracks. Also,service conditions and manufacturing processescan embrittle otherwise ductile materials. Carbonsteels are ductile at room temperatures andabove, but as the temperature is lowered theyexperience a transition from ductile to brittlefracture behaviour. The temperature at which thistransition occurs depends on carbon content and grain size, but can occur at temperaturesabove 0°C, so it can have serious engineering
consequences. The welding of steels can altertheir structure at the grain level and increase thetemperature of the ductile to brittle transitiontemperature.
6.4 FatigueFailure by fatigue may occur when a componentis subjected to repetitive or cyclically varyingstresses. It may occur after the stress has beenapplied for a thousand cycles or after many mil-lions of cycles, at stresses appreciably below thosewhich the component could sustain under a staticstress. Indeed, in the case of a metal, failure mayoccur at stresses below the yield strength.
Fatigue failure usually originates from a regionof concentrated stress, such as an impurity parti-cle in the material or at an engineering feature,such as a hole or notch. Once a fatigue crack hasinitiated it grows under the action of the cyclicstress until its length is such that its associatedstress intensity factor equals the fracture tough-ness of the material (equation 6.6) when fast frac-ture occurs. The crack growth region of thefatigue fracture face is relatively smooth whereas,in the case of a ductile metal, the fast fractureregion is rougher and characteristic of localplastic deformation. Fatigue failure does not nor-mally involve gross plastic deformation and henceshows similarities with brittle fracture. Ductilemetals are particularly prone to fatigue failure,but ceramics and polymers can also suffer fromthe same phenomenon.
Fatigue performance, in its simplest form, ischaracterised in the laboratory by plotting experi-
Fracture and fatigue
48
Number of cycles to failure(logarithmic scale)
Stress range
Scatterband
FIGURE 6.5 Schematic of stress range versusfatigue life (number of cycles to failure).
mental stress range versus number of cycles tofailure (Figure 6.5). As the stress range isincreased the number of cycles to failuredecreases. Such curves show considerable scatterand are affected by the environment and surfacefinish of the test pieces (or components).
With some reservation it may be stated that, inmost buildings, the designer has little to fear fromfatigue in the structure itself, though such cases asthe collapse of the oil rig Alexander Keilland in1980, due to fatigue failure in one of the legs,cannot be ignored. Welded joints, if not carefullymade, are particularly at risk because of theadventitious introduction of stress raisers in the form of unwelded regions and changes to thematerial structure. Certainly individual com-ponents such as bolts may be at risk under cyclicloading conditions. Fatigue is an importantconsideration in bridge design.
Chapter seven
Electrical andthermalconductivity
Although electrical conductivity is not a con-straint in structural design, thermal conductivityis important and perhaps nowhere more so thanin the shell of a building. Nonetheless, we shallbriefly consider electrical conductivity since itprovides a basis for the more complex ideas ofthermal conductivity.
Some typical values of electrical and thermalconductivity are given in Table 7.1. Electricalconductivity is defined as the current per unitcross-sectional area of conductor, per unit voltagegradient along the conductor. Thermal conductiv-ity is defined similarly. It can be seen that:
• metals are good conductors of heat and elec-tricity;
• non-metals are poor conductors;• there is an approximate relationship between
electrical and thermal conductivity in metals.This is not true of non-metals.
The structure of the metallic bond was describedearlier as an ‘electron cloud’ and as long as noelectrostatic field is applied, the free electrons in ametal behave in much the same way as the parti-cles in a gas, i.e. they move around randomly andcollide elastically with each other and with the
49
TABLE 7.1 Electrical and thermal conductivities of a range of materials. Gases are at STP, other substances areat 0ºC (273K)
Electrical Thermal Wiedemann–Franz ratioconductivity conductivity ��W/K2
(�m)�1 W/m�K
Air 24.1�10�3 –Oxygen 24.4�10�3 –Nitrogen 24.3�10�3 –Brick 5�10�7 1 7.3�103
Concrete (dry) 1�10�9 1.4 5.1�106
Glass 1�10�10 1.0 3.7�107
Polyethylene 1�10�16 0.5 1.8�1013
Aluminium alloys 2.1�107 150 2.6�10�8
Copper 6.0�107 390 2.4�10�8
Lead 0.6�107 35 2.1�10�8
Silver 6.8�107 420 2.3�10�8
Steel 0.9�107 50 2.0�10�8
stationary ions in the lattice. However, when anelectrostatic field is applied, the electrons driftpreferentially and, being negatively charged, they‘drift’ towards the positive ‘pole’, though therandom motion still remains. In drifting, theycollide more frequently with the ‘stationary’ ions.This provides a resistance to drift, whichincreases with temperature.
The force F tending to accelerate each electronis:
F�Ee (7.1)
where e is the charge on the electron and E is theelectrostatic field.
The bodily movement of the electrons J, i.e.current per unit cross-sectional area, can beexpressed as:
J�neVa (7.2)
where n is the concentration of free electrons andVa is the drift velocity. Since metals contain alarge concentration of free electrons the flux ishigh. Thus, as Ohm’s law states, current isdirectly proportional to voltage (potential dif-ference) and inversely proportional to resistance,I�V/R. Since non-metals are either ionically orcovalently bonded there are few free electrons,hence their low conductivities.
Thermal conductivity in metals follows muchthe same general argument, though here it is notthe bodily movement of electrons but rather thetransference of heat energy by collision. Theanalysis is, however, much more complicated butit is intuitively obvious that the higher the tem-perature, the greater the excitation of the elec-trons and the larger the number of collisions.Since, in metals, both thermal and electrical con-ductivities have their origins in the same struc-tural features the proportionality apparent inTable 7.1 is not surprising. It is known as theWiedemann–Franz ratio and has a theoreticalconstant value of 2.44�10�8 � �W/K2, which
Electrical and thermal conductivity
50
0 1 2 3
Mass density (Mgm�3)
Hea
t ca
paci
ty (J m
�3°C
�1)
0
1
2
3
Wool
Cork
Timber
Concrete
Nylon
Perspex
Quartz
Brick
Glasses
Bitumen
Asbestos
GraniteSandstone
FIGURE 7.1 Thermal capacities of various sub-stances.
should be independent of temperature if heat flowis entirely by free electrons.
The thermal conductivity of non-metals is morecomplex still, since it involves energy transferbetween the atoms which make up the material.Heating of the material manifests itself as increas-ing vibration of the nominally stationary atoms.What is important to note, however, is that thethermal conductivity is controlled largely by theproduct of the mass density and the specific heatat constant volume Cv. This is called the heatcapacity. Specific heat is the heat energy requiredto increase the temperature of unit mass of thematerial by one degree. Figure 7.1 shows thatlight materials with an open structure are betterinsulators than heavy, compact materials.
Moisture has a significant effect on the thermalconductivity of porous materials. If the pores arefilled, the water acts as a bridge and since theconductivity of water is many times greater thanthat of air, the resulting conductivity is greater.At the same time, water is denser and has ahigher specific heat so that the heat capacity alsoincreases.
Further reading
GeneralAshby, M.F. and Jones, D.R.H. Engineering Materials,
Pergamon Press, Oxford.• Vol. 1 (2nd edn, 1996) An introduction to their
properties and applications.• Vol. 2 (1986) An introduction to microstructures,
processing and design.• Vol. 3 (1993) Materials failure analysis.Certainly books to be dipped into. Perhaps a little
detailed for easy reading but with a set of excellentcase studies which illustrate the engineer’s approach.
Callister, Jr, W.D. (2000) Materials Science and Engi-neering, An introduction, 5th edn, John Wiley &Sons Inc., New York.
A comprehensive introduction to the science and engi-neering of materials.
Gordon, J.E. (1976) The New Science of Strong Mater-ials – or Why You Don’t Fall Through the Floor.Penguin Books, Harmondsworth, Middlesex.
Gordon, J.E. (1978) Structures – or Why Things Don’tFall Down. Penguin Books, Harmondsworth, Middlesex.
Excellent and very readable. Read them on the bus orin the bath. They will tell you more than many hoursof library study.
Petrowski, H. (1982) To Engineer is Human, Macmil-lan, London.
Petrowski considers what it is like to be an engineer inthe twentieth century and lays some emphasis on thethings that have gone wrong. Not a book for thoselacking in self-confidence but good (and easy)reading.
Van Vlack, L.H. (1989) Elements of Materials Scienceand Engineering, 6th edn, Addison Wesley, New York.
Fairly elementary but expands upon matters dealt within this part.
The symposium Design Life of Buildings (Institution ofCivil Engineers, 1985, Thomas Telford, London).
Scarcely a textbook but the papers cover many aspectsof the engineer’s profession. Well worth a glance atthose papers that may interest you.
SpecialistAngrist, S.W. and Hepler, L. (1973) Order and Chaos,
Harmondsworth, Middlesex.A helpful (and non-mathematical) introduction to ther-
modynamics. Recommended as bedtime reading.
Cottrell, A.H. (1964) The Mechanical Properties ofMatter, John Wiley, New York.
First class, scientific and of much wider coverage thanthe title suggests. Essential reading for any studentwishing to follow up the concepts herein and highlydesirable reading for all students of all branches ofengineering.
Freudenthal, A.M. (1950) The Inelastic Behaviour ofEngineering Materials and Structures, John Wiley,New York.
Old now but still very readable in the way it deals withmaterials in relation to their structures. Not easy toread, but good.
Houwink, R. and De Decker, H.K. (eds) (1973) Elas-ticity, Plasticity and Structure of Matter, 3rd edn,Cambridge University Press, Cambridge.
You need to be pretty strong-minded for this, butwhere else could you find out about the rheology ofbaker’s dough or of liquid paints, gelatins and gluesor soaps? They may not be ‘engineering’ as definedby your average university lecturer but they are allmaterials and it never does to assume that the find-ings of one discipline are inapplicable to another.
Treloar, L.R.G. (1949) The Physics of Rubber Elastic-ity, Clarendon Press, Oxford.
Old now and pretty hard going, but the classic work onthe subject. Perhaps overtaken by Aklonis, J.J. andMacKnight, W.J.L. (1983) Introduction to PolymerViscoelasticity, 2nd edn, John Wiley, New York.
51
Part Two
Metals andalloys
W.D. Biggs
Revised and updatedby I.R. McColl andJ.R. Moon
Introduction
Useful metals have been known to mankind for along time and probably came into service verygradually. Stone tools were adequate for manypurposes in early cultures, but when metalsbecame available they offered many advantages.Metals could be strong and hard, but their chiefadvantages were in their ductility. This enabledthem to withstand a blow and the range ofshaping procedures and products that becamepossible. The significance to us all is encapsulatedin the ideas of the Stone Age, Bronze Age andIron Age. Metals and their differences have evenbeen the subject of poetry. Thus, from RudyardKipling:
Gold is for the mistress – silver for the maid,Copper for the craftsman – cunning at his
trade,‘Good!’ said the Baron, sitting in his hall,‘But iron – cold iron – is master of them all.’
Most metals are found in nature as ores, oxides,sulphides, carbonates, etc. Extraction metallurgyis that aspect of the art concerned with extractingthe metals from these compounds. The basicchemistry is generally fairly simple. The industrialproblem is to do the job on a big enough scale tomake it economically worthwhile. The converseproblem also exists. When a metal is exposed to aworking environment it will tend to revert to theappropriate compound, i.e. it corrodes. Rust onsteel is almost the same as the ore from whichiron is extracted. Thus, the metallurgist has twotasks, to get the metal extracted from its ore andthen to keep it that way.
The origins of extraction metallurgy are in pre-history and it is supposed that early discoverieswere made accidentally: a piece of rather special
rock when heated in the reducing atmosphere ofa fire gave up some metal. Copper, lead and tinwere amongst the earliest to be produced thisway, and alloys such as bronze (copper and tin)and brass (copper and zinc) followed. Bronze wasmuch prized for its ability to be cast into shapesor mechanically formed, as well as for its combi-nation of strength, hardness and toughness.
Unaided fires can reach temperatures of about1100–1200°C on a good day. This is sufficient tomelt all the metals referred to so far, but cannotcope with iron, which melts at about 1550°C.The early history of iron involved extractionprocesses which gave rise to solid lumps of veryporous and friable metal, mixed with a glassyslag. This was formed into useful articles by ham-mering at temperatures high enough to melt theslag. The slag was partially squeezed out and thepores closed. The product is wrought iron, a tech-nical term with a specific meaning in metallurgy.There were later improvements, particularly inthe eighteenth century, but iron remained anexpensive and precious material until the techno-logy for reaching higher temperatures evolved.
The eighteenth and nineteenth centuries sawthe development of a forced air blast to raise thetemperatures in blast furnaces enough to makeliquid iron possible. In the furnace, this picked upabout 4 mass per cent of carbon from the fuel,which had the advantage of also lowering themelting temperature of the alloy. The product,pig iron or cast iron, was brittle but easily castinto moulds and found a wide range of uses inengineering and as household articles.
The problem was that to make tough iron orsteel required a much lower carbon content and,consequently, a higher melting temperature. It
55
was Henry Bessemer who worked out how to dothis on a large scale. A Bessemer converter blewcold air through a molten bath of pig iron tooxidise the impurities, including carbon, to leaverelatively pure liquid iron. The oxidation processgenerated enough heat to raise the melt tempera-ture enough to keep the material molten as itbecame purer. The whole process, from start tofinish, took about an hour and produced about10 to 20 tonnes of steel, not much by today’sstandards, but revolutionary at the time. Aslower, but more easily controlled process alsocame into use. Large pools of metal were meltedin shallow hearths and oxidation achieved byreactions with a covering slag and controlledadditions of iron oxide.
Both processes have been superseded by moreefficient methods. The basic oxygen converter isbased on Bessemer’s principle, but uses oxygenrather than air. No heat is wasted in heating the80 per cent of air that is nitrogen. The oxygen isblown in from the top through a lance. Severalhundred tonnes per hour are possible frommodern furnaces. Much steel is also producedfrom remelted scrap, for which purpose electricarc furnaces are often used: good control overmaterial quality is possible using well-characterised scrap.
The development of freely available supplies ofelectricity in the late nineteenth century allowedaluminium to be extracted on a commercial scale:
this metal has to be extracted electrolyticallyfrom molten salts.
Many other metals, magnesium being one, areproduced electrolytically and the process isexploited in the refinement of still more metals,copper for instance. Electrical power is alsoessential to the production of the ‘modernmetals’, such as titanium.
Physical metallurgy (Chapter 8) deals with thestructure of metals with the aim of designing andproducing internal structures to meet the proper-ties desired. This provides a basis for understand-ing the properties of metals and alloys which aresubsequently described.
We should emphasise that metals need to beshaped into a great range of artifacts if they areto fulfil the needs of the engineer. This is the areaof mechanical metallurgy, dealt with in Chapter10, involving understanding and control ofprocesses such as hot working, cold working andjoining. Many of these processes remained smallscale until the advent of steam power in theIndustrial Revolution. The rolling mill, developedin the eighteenth century, revolutionised thespeed of production of long shapes. One of themost important features of these processes andperhaps the least understood is that they allowmetallurgists to control the internal microstruc-tures of the materials and through that theirproperties.
Introduction
56
Chapter eight
Physicalmetallurgy
8.1 Grain structure8.2 Crystal structures of metals8.3 Solutions and compounds
8.1 Grain structureProbably the greatest single breakthrough in ourunderstanding of the properties of metals wasmade by Henry Clifton Sorby in the latter part ofthe nineteenth century. He developed a techniqueby which the structure of metals could be exam-ined using an optical microscope. Despite manyadvances since that time, metallography, as it isknown, remains the most widely used tool of themetallurgist. Figure 8.1 shows the structure of apure metal. It consists of an assembly of minute,
interlocked grains. There are a number of waysby which such a structure can develop. Here weconsider one of them, the solidification of a puremetal.
We noted in Chapter 1 that, in a liquid, theatoms are in a state of constant motion andchange positions frequently. As a liquid cools themotion of the atoms becomes more sluggish and,sooner or later, the atoms arrange themselves intoa pattern that forms the nucleus of the solidmaterial. The kinetics of nucleation are quitecomplex. We note here only that nucleation ofthe solid almost always begins from an impurityparticle in the melt. Let us not worry too muchabout the conditions for nucleation of a solidcrystal, but rather concern ourselves with thegrowth thereafter. Suppose we assume that asolid nucleus has formed within a liquid melt andthat it has the geometrical configuration of an ele-mentary cube of atoms. As the metal solidifies itmust give up the latent heat of solidification. Thecorners of the cube lose heat faster than the edgesso that atoms from the melt attach themselves tothe corners first, then to the edges and last of allto the faces of the elementary cube. Thus abranching pattern or dendrite is built up fromeach nucleation site (Figure 8.2) and dendriteswill grow from each site until they are stopped byinterference from other dendrites.
Eventually the liquid is used up by infillingbetween the arms of the dendrites and the charac-teristic polycrystalline structure results (Figure8.2 (f)). There are three important facts to notehere:
57
FIGURE 8.1 The microstructure of Armco iron.
1. within each grain the atoms are arranged in aregular lattice;
2. the orientation of the crystal lattice differsfrom grain to grain;
3. at each grain boundary there is a line of mis-match in the atomic arrangement.
In practice few metals are really pure and itwould be prohibitively expensive to make themso. Most, of course, are deliberately alloyed. Inthese cases compositional changes accompany thesolidification process, as described in Chapter 2,and the microstructural developments are morecomplex.
8.2 Crystal structures of metalsIt will be recalled from Chapter 3 that the natureof the metallic bond allows development of crys-talline structures in which each metal ion isbound in a non-directional way to all its imme-diate neighbours. Metal ions, therefore, tend topack as closely as possible in order to achieve ashigh a density as possible. This requirementmeans that most metals have, at any given tem-perature, one of only three crystal structures.Compare this with the enormous range of struc-tures met in ionic and covalent solids. Whatstructures are most likely? We will think of the
Physical metallurgy
58
FIGURE 8.2 Solidification of a metal by dendriticgrowth (schematic).
FIGURE 8.3 Close-packed directions in a close-packed plane of spheres.
ions as being equally sized and incompressible,like billiard balls, and find how these might bepacked together to achieve maximum density. Webegin with one ball. The maximum number ofothers that can be packed around it in one planeis six (Figure 8.3). By continuing the plane, aclose-packed plane is built up.
There are now two possible arrangements forthe next plane provided that each sphere is incontact with three spheres below (Figure 8.4). Thesecond layer is placed in the dimples at BBBB. . .,etc. But putting a third plane in place offers twooptions about where to put it. The spheres may beplaced either in positions vertically above the ori-ginal spheres to give the sequence ABABAB. . ., orthey may be placed in the unoccupied dimplesCCCC. . ., which are vertically aligned withneither the spheres in the first or the second layer,to give ABCABC. . . (Figure 8.4).
At first sight these differences in stacking mayappear to be trifling but they have important con-sequences. We define these structures in terms ofa unit cell and visualise the metal as composed ofan array of such cells. The ABABAB . . . arrange-ment gives us the hexagonal close-packed (hcp)structure (Figure 8.5 (a)). The ABCABC . . .arrangement gives us the face-centred cubic (fcc)
arrangement (Figure 8.5 (b)). At first sight thisdoes not appear to be close packed, but a view onthe diagonal plane XYZ (Figure 8.6) shows thatit is. These two structures represent the closestpossible packing of spheres of equal size. In eachcase the spheres occupy 74 per cent of the volumeavailable.
Crystal structures of metals
59
FIGURE 8.4 The B spheres (shaded) rest on the Aspheres. A third close-packed layer may then rest onthe B spheres either directly above the A spheres or inC positions.
FIGURE 8.5 (a) The close-packed hexagonal struc-ture. (b) The face-centred cubic structure (from W.Hume-Rothery and G.E. Raynor (1956) The Structureof Metals and Alloys, by permission of the Institute ofMetals).
FIGURE 8.6 Face-centred cubic structure: (a) loca-tion of atomic sites and (b) view on XYZ planeshowing close packing.
There is a third common structure in which theatoms are not quite close packed. This is thebody-centred cubic cell (bcc) shown in Figure 8.7.
Metals crystallising in the above forms are:
fcc – aluminium, copper, nickel, iron (above910°C), lead, silver, gold;
hcp – magnesium, zinc, cadmium, cobalt, tita-nium;
bcc – iron (below 910°C), chromium, molybde-num, niobium, vanadium.
Why do these differences occur? Why is alu-minium fcc and magnesium hcp? We expect thatthe structure adopted is the one that gives thecrystal the least internal energy, although this
FIGURE 8.7 The body-centred cubic structure: (a)expanded unit cell and (b) sphere model of unit cell(from W. Hume-Rothery and G.E. Raynor (1956) TheStructure of Metals and Alloys, by permission of theInstitute of Metals).
structure need not necessarily be close packed.The energy difference between different structuresis often very small and the crystal structure thathas the lowest energy at one temperature may nothave the lowest energy at another. These variousforms are called allotropes. Changes from onestructure to another, brought about by changes oftemperature, are of fundamental importance tometallurgical practice. For example, the changefrom fcc to bcc as the temperature of iron isreduced through 910°C forms the foundation ofthe metallurgy of steel.
8.3 Solutions and compoundsVery few metals are used in the pure state, theyare nearly always alloyed with other elements toobtain better mechanical properties. The phasesthat can form in any given alloy system and theranges of composition and temperature for whichthey are stable are summarised in equilibriumdiagrams, which we discussed in some detail inChapter 2. We saw there that some alloying ele-ments can dissolve in the basis metal to formsolid solutions. Some elements dissolve readily,some with more difficulty. Iron can only dissolve0.007 mass per cent carbon at room temperature,but copper can dissolve 30 per cent zinc to makebrass. In the copper–nickel system, includingMonel metal, cupronickel, etc., there is completesolubility of nickel in copper and of copper innickel.
There are two principal classes of solid solu-tion. In one, small atoms, such as carbon andnitrogen, fit into the spaces between the largeratoms. This is an interstitial solid solution (Figure8.8 (a)). The solubility of the small atoms isgenerally limited, but the effects on properties canbe dramatic. In the other class, the dissolvedatoms are similar in size and chemistry to thoseof the host metal and they simply replace some ofthe host atoms to produce a substitutional solidsolution (Figure 8.8 (b)). The classic example iscopper–nickel, which are similar chemically andwhere the sizes of the ions are 0.127nm and0.125nm, respectively. Substitutional solid solu-tions may be random (they mostly are) or ordered
Physical metallurgy
60
FIGURE 8.8 Solid solutions: (a) interstitial and (b)substitutional.
depending upon what sort of neighbours the ionsfind most comfortable in energy terms.
However, some atoms do not fit comfortablyinto the host lattice. At some limit, the energy ofthe lattice is increased to a critical value and thesurplus atoms are rejected, usually as an interme-
FIGURE 8.9 Relatively coarse pearlite in a plaincarbon steel (�200) (from A.R. Bailey (1967) TheStructure and Strength of Metals, by permission ofMetallurgical Services Laboratories Ltd).
diate compound, such as iron carbide (Fe3C) insteel, or CuAl2 in some of the heat treatable alu-minium alloys (see Chapter 12). Generally suchcompounds are hard and brittle and exert a ‘rein-forcing’ effect upon the matrix of softer metal inwhich they are dispersed. Figure 8.9 shows Fe3Cin carbon steel and it will be noted that the ‘rein-forcement’ occurs at the microscopic level of reso-lution. Some intermediate compounds are used intheir own right, such as tungsten carbide forcutting tools.
The metallurgist refers to these solid solutionsand compounds as ‘phases’. A phase is simply amaterial, or part of a material, which has thesame structure and properties. Much of the
science and technology of physical metallurgy isconcerned with organising manufacturing sched-ules to get the right phases together in the mostappropriate geometrical dispersion. This micro-structure of the material is what controls itsstrength and the properties related to strength.The interplay between manufacturing, micro-structure and properties is what modern materialsscience and engineering is all about. In doingthese things, the equilibrium diagrams for theparticular alloy systems are uppermost in themind. A full discussion is beyond the scope of thischapter and the reader is referred to the standardtexts such as Cottrell (1967).
Solutions and compounds
61
62
Chapter nine
Mechanicalproperties ofmetals
9.1 Stress–-strain behaviour9.2 Tensile strength9.3 Ductility9.4 Plasticity9.5 Dislocation energy9.6 Strengthening of metals9.7 Unstable microstructures
Materials scientists tend to divide properties intotwo groups – structure insensitive and structuresensitive. Structure insensitive properties arewholly invariant, they are associated with theproperties of the atoms themselves and theprimary forces between them. They do notdepend on the microstructure. The principalproperties here are mass, density, elastic modulus,specific heat, thermal expansion coefficient and,to a lesser extent, some chemical and electricalcharacteristics. On the other hand, structure sen-sitive properties are wholly dependent upon themicrostructure of the material, which in turn isgenerated by its past history; whether hot rolledor cold rolled, whether heated and cooled, and, ifso, how. All of these processes influence the finalmicrostructure, and, through that, can be used tomanipulate the properties of the material. Fromthe designer’s point of view, the most importantstructure sensitive properties are the yieldstrength, fracture strength (or energy), ductilityand the performance in fatigue.
9.1 Stress–strain behaviourThe most commonly used form of tension testuses an arrangement in which the test piece iselongated in series with a device to measure theload generated by the stretching. It gives rise tobehaviour like that in Figure 9.1.
Up to A, the behaviour is essentially elastic (seeChapter 1) but at A it changes. This point is vari-ously called the yield point (YP), the yield stressor strength (YS), the limit of proportionality (LP)or the elastic limit (EL). We shall generally usethe first, yield point, not because it is the mostdescriptive but because it (or yield stress) is mostwidely used in specifications. Below the yieldpoint the material returns to its original dimen-sions on unloading. Beyond the yield point it ispermanently deformed. Try squeezing a beer can,first lightly, and it takes no notice, then squeeze ithard and it stays deformed.
The yield stress is of particular importance forstructural designers as it is often chosen to be thelevel which the stresses in service should notexceed, or even approach. In Figure 9.1, the yieldstress is easily defined as there is a distinct changeof slope at point A, but this is not the case withmany materials. An equivalent, the proof stress, isthen used, which is obtained as shown in Figure9.2. A small strain is chosen, typically 0.1 or 0.2per cent, and a line drawn through this value onthe strain axis parallel to the initial linear part of
the stress–strain curve (or to a tangent to thecurve at the origin if it is entirely non-linear). Thepoint of intersection with the stress–strain curvethen gives the 0.1 per cent or 0.2 per cent proofstress.
Once the yield stress is passed, reversing thedirection of stretch gives an unloading line likeBC. There is a permanent deformation OC,known as plasticity. But here is the interesting
Tensile strength
63
FIGURE 9.1 Tensile stress–strain curve (schematic). From O to A the bar remains substantially constant inshape; from A to E it lengthens but reduces uniformly; at E it forms a ‘neck’ at which fracture eventually occurs.
0.2% proof stress
0.1% proof stress
0.1% 0.2%
Str
ess
�
Strain �
FIGURE 9.2 Determination of proof stress whenthe yield point is not well defined.
fact. Once you reload it to B it goes on as before.In other words, if we took a new beer can itwould start to deform at stress A. If we takeanother, loaded it to B and then unloaded it, itwill not start to deform again until the stress B isreached. What an interesting phenomenon! Over-load a metal and it gets stronger. We call thisstrain or work hardening. It was the principalmethod used by early blacksmiths to produceharder, stronger bronze and iron by cold ham-mering. It is of importance (and comfort) to thestructural engineer since it forms the basis ofwhat is known as ‘plastic design’ of structuralsteelwork. If parts of structural steelwork are‘overloaded’, they deform plastically and theimposed stresses are redistributed away from theoverloaded parts.
9.2 Tensile strengthAt point E in Figure 9.1 the load falls away. Thisoccurs because, as the material starts to deformlocally, a waist or neck is produced. In this neckthe material is continuing to get stronger but itscross-sectional area is decreasing faster, and theload that can be sustained decreases. Fracture will
eventually occur in the necked region. This doesnot always happen. Sometimes, failure occursbefore the necking process can begin. Servicefractures in such materials can be sudden andunexpected (Chapter 6).
In the early days of testing, the change inbehaviour at E was readily observed and wasdescribed as the ultimate tensile strength, a termwhich is still in use. Its value is of little concern tothe designers of structures and artifacts, but is ofappreciable interest to those who have to shapethe parts by deforming them.
If the reduction in area is measured in the neckregion when it has formed, and the true stress,that is, load/actual area, is calculated, the dashedcurve is obtained (Figure 9.1), i.e. increasingstress up to failure, which is a much more logicalbehaviour than reducing stress. This curve can bedescribed by simple mathematics, although this israrely done except for research purposes.
9.3 DuctilityThe reduction in cross-sectional area of the testpiece at the site of fracture, and the total elonga-tion at fracture, are both conventionally used asmeasures of ductility. Unfortunately, elongationto fracture is a function not only of materialbehaviour but also of test piece geometry. Longtest pieces will appear to be less ductile than shortones of the same diameter. This is a consequenceof necking, which is concentrated into a lengthequal to about two or three diameters. Reductionin area does not suffer as markedly from thisproblem, although round and square section testpieces may give differing results. It is for theseand other reasons that test piece geometries aresubject to rigorous standards.
As yield strength and tensile strength increase,ductility and fracture toughness decrease (Figure9.3). For example, cold working a material willraise its yield strength ever closer to the tensilestrength, but at the same time the reserve of duc-tility is diminished and, in the limit, the materialwill snap under heavy cold working. This isfamiliar to anyone who has broken a piece ofwire by continually bending and rebending it.
Mechanical properties of metals
64
Yield strength (MPa)
50
100
150
200
Frac
ture
tou
ghne
ss (M
Nm
�3
/2)
Al limit line
Ti limit line
Steel limit line
0 500 1000 1500 25002000
Maraging steelsLow alloy quenchedand tempered steelsHardened stainlesssteels
Ti 6%Al 4%
V
FIGURE 9.3 Relationship between fracture tough-ness and yield strength for a range of alloys (afterDieter 1988, Mechanical Metallurgy, McGraw-Hill,London)
9.4 PlasticityThis is one of the most important properties ofmetals and one that we should all be grateful for.If you run your car into a tree, the wing, etc., willbe seriously deformed, but in doing so willabsorb energy. Careful design, based on this idea,underlies the use of crumple zones to improvesafety. It also makes possible a wide range ofshaping methods.
Clearly, at the yield stress (see Figure 9.1), thebehaviour changes from recoverable (elastic)deformation to one which adds permanent defor-mation. Early observations on specially preparedsamples showed that at the yield stress, wholerows of atoms slide bodily past each other ratherlike a pack of cards (Figure 9.4). This is known asslip. It is a shear deformation produced by ashear stress. Further theoretical considerationshowed that the energy required to shear a wholeplane of atomic bonds was far greater than thestrain energy applied at yield, and an alternativemechanism was sought. Suppose you want tomove a heavy carpet lying on a rough floor. Theway not to do it is to try to move it bodily; thefrictional forces over the whole area are consider-able. The more economical way is to create asmall ripple in the carpet and then to work itacross to the other edge. The force required is
small because the friction between the carpet andfloor is easily overcome locally (Figure 9.5).
This is a good analogy for the mechanism thatexists in metals, and which provides the answerto a puzzle met in Chapter 6. There, the theo-retical strength of a material was estimated to be�E/10, but we know the practical strengths to be�1/100 of this. Think about Figure 9.6. We wantto shear the top half of the metal over the bottomhalf along the shear plane XX. We hold thebottom half rigidly and push hard over thesurface YY so that the material is displaced byone atom spacing (Figure 9.6 (b)). The oppositesurface at the other end of the slip plane has notmoved. This means that we have more atoms perunit length along the slip plane in the top halfthan in the bottom. The atoms do not squeeze upevenly all along the plane.
Somewhere along the plane the picture lookslike Figure 9.7. On either side the crystal appearsto be perfect, even across the slip plane. But wehave generated a dislocation. Because only oneinteratomic bond, that at X, is unsatisfied, ittakes only a small shear stress to move the site of
Plasticity
65
FIGURE 9.4 (a) Crystal under stress. (b) Freeunconstrained yielding results in considerable lateraldisplacement. (c) Rotation of slip planes during defor-mation permits elongation with very little overalllateral movement.
FIGURE 9.5 How to move a heavy carpet.
the dislocation to the left or the right. If it con-tinues on its way from left to right it will eventu-ally reach the opposite surface and produce oneunit of slip (Figure 9.6 (c)). You will see that thedislocation represents a boundary; to the left, thematerial has slipped, to the right it has not.Three-dimensionally, the dislocation can bethought of as a small core which extends perpen-dicular to the plane of Figure 9.7. Looked at froma distance it looks like a line and for this reason itis known as a line defect.
The slip produced by one dislocation is aninteratomic distance, typically �0.2nm. Imaginenow taking a cube of metal, 50mm edge, and
X
YY
X
(c)
b
b
(b)
(a)
b
FIGURE 9.6 Illustration of the slip process.
deforming it between two parallel plates so thatits height is halved. A simple calculation con-cludes that this requires the movement of �100million dislocations. If it is all done in about onesecond, as in drop forging, the average disloca-tion speed has to be about 90km/h. We need lotsof them, moving fast, to give us plasticity on apractical scale. They arise in the mannerdescribed above and also from errors in assem-bling crystals during solidification and otherways.
It is important to note that dislocations canmove only along certain atomic planes within thelattice. These are the slip planes and generally arethe close-packed planes in the crystal. The hcpmetals have a fairly restricted number of slipplanes; the bcc and fcc metals have many.
9.5 Dislocation energyThe atoms at the core of the dislocation are dis-placed from their proper positions. We canexpress this in terms of a strain. It is approxi-mately 1/2 so that the corresponding stress is ofthe order of G/2, where G is the shear modulus,and the strain energy per unit volume is aboutG/8. If we now assume that the radius of the coreis about the size of the atom b, the cross-sectionalarea of the core is b2 and its total volume isb2l, where l is the length of the dislocation line.
Mechanical properties of metals
66
X
FIGURE 9.7 Illustration of one type of dislocation –an edge dislocation.
Hence, the total dislocation energy per unitlength is:
U/l� Gb2/8�Gb2/2 (9.1)
which has units of J/m or N.In order to minimise the energy, the dislocation
line tries to be as short as possible. It behaves asif it were an elastic band under a tension T, thevalue of which is identical with U/l.
T� Gb2/2 is actually very small, but, relativeto the size of the dislocation, it is large and itplays an important role in determining the way inwhich obstacles of one sort or another canobstruct the movement of dislocations.
9.6 Strengthening of metalsAlthough we must take care not to lose too muchductility and toughness, the critical property indesign is the yield stress. If the material requires ahigher stress to produce yield then the safeworking stress is correspondingly increased. Inusing the term ‘strengthening’, we are concernedwith the various ways by which we can make thestart of slip more difficult. We now consider someof the ways in which this is done.
9.6.1 Grain boundaries
In a single crystal of a pure metal the shear stressrequired to move a dislocation is small, in somecases maybe only �1MPa. However, most mater-ials are polycrystalline. It was noted above that,within a single grain, the atomic arrangement isregular but there is a discontinuity at a grainboundary. A dislocation that has reached a grainboundary cannot produce a slip step there, unlessthe neighbouring grain also deforms to accommo-date the shape change. A dislocation in thesecond grain cannot move until the shear stress,resolved onto the new slip plane and in the newslip direction, reaches the value needed to con-tinue movement. Back in the first grain, the dislo-cation is stuck and other dislocations will pile upbehind it in a traffic jam, exerting a force on it,until, ultimately, the push is too great and it isforced to the grain boundary. The stress on the
leading dislocation is a simple function of thenumber of dislocations in the pile-up. In a coarse-grained structure many dislocations can pile upand the critical stress is reached early, whereas ina fine-grained structure the length of the pile-up issmaller and more stress must be applied fromexternal forces, i.e. the yield point is raised. Theoutcome is summarised in the famous Hall–Petchequation:
�y ��0 �kd�1/2 (9.2)
where �y is the yield strength of our polycrys-talline material, �0 is the yield strength of onecrystal on its own, k is a proportionality constantand d is the grain size of the material. Mild steelwith a grain size of 250�m has �y �100MPa,but when d�2.5�m, �y �500MPa. The incen-tive for making fine-grained steels is clear.
Control of grain size in castings is generallyachieved by ‘inoculating’ the liquid metal withsubstances which can react with ingredients in themetal to form small solid particles that act asnucleation sites for crystal growth. In wroughtproducts, the thermal and mechanical history ofthe working process can be controlled to give finegrains. Rolling and forging are used not only toshape materials but also, perhaps more import-
Strengthening of metals
67
Fe – 0.007%C
0
1000
1500
500
0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
True strain
True
str
ess
(MPa
)
0 50 80 90 95 98 99 99.8 99.95
Reduction in area (%)
FIGURE 9.8 Work hardening in wire drawing tostrains beyond those available in tension testing.
antly, to control their microstructures and hencetheir properties.
9.6.2 Strain hardening
In a tension test, we can stretch a reasonablyductile metal by only about 30 per cent or sobefore it becomes unstable and begins to form aneck. But when we roll the same metal or form itinto wire by drawing, the deformation in thelocal area being worked is essentially compres-sive. This allows us, for example, to draw a wireto many times its original length with relativeease. The work hardening is extended beyondwhat can be achieved in a tension test (Figure9.8).
Metals, especially those in the fcc and bccsystems, have many different planes on which dis-locations can move to produce slip. But none ofthese are markedly different from the others and,under increasing stress, all dislocations try tomove at once. If the slip planes intersect eachother, as indeed they do, the dislocations on oneslip plane act as a barrier to dislocations trying tomove across them. As we have seen, many mil-lions of dislocations are on the move, the trafficpile-up is considerable, and the dislocations getjammed. Very much more stress needs to beapplied to get things moving again, and strainhardening is the result. It is one of the mosteffective ways of raising the yield strength of ametal, though, if carried too far, it results in frac-ture as we have seen.
9.6.3 Annealing
Since each dislocation is a region of high strain inthe lattice, they are not thermodynamically stableand comparatively little energy is required tocause a redistribution and cancellation of thetrapped dislocation arrays. The energy is mostconveniently supplied in the form of heat, whichgives the atoms enough energy to move sponta-neously, and to form small areas which are relat-ively free of dislocations. This is called recoverybut, since the dislocation density is only slightlyreduced, the yield strength and ductility remain
almost unchanged. The major change involvesrecrystallisation. New grains nucleate and grow,the material is restored to its original dislocationdensity and the yield point returns to its originalvalue. This process (Figure 9.9) involves heatingto temperatures of around 0.6 Tm (where Tm isthe melting point in degrees K) and is known asannealing. In addition to overcoming the effectsof work hardening, this is an important way ofcontrolling grain size. We shall later see (Section10.3 pp. 70 and 71) the importance of annealingin the preparation of metals and alloys for com-mercial use.
9.6.4 Dispersion hardening
One of the most powerful ways of impeding dis-location movement, and hence of increasing theyield strength, is to disperse hard and often brittleparticles throughout the structure, so that the dis-locations in the softer matrix are stopped andthen must find a way round. But there are limita-tions and perhaps the following analogy will helpto explain what these are.
Bitumen is a soft substance which, if left alone,will flow under its own weight and will most cer-tainly flow under load. And yet it is used for
Mechanical properties of metals
68
FIGURE 9.9 Load–extension curves for heavily coldworked copper reheated to various temperatures (afterK.J. Pascoe (1978) An Introduction to the Properties ofEngineering Materials, Van Nostrand Reinhold).
roads. This is possible only because it is mixedwith a hard aggregate. Let us suppose that theaggregate is supplied as a fine ‘flour’. This willoffer little or no resistance to flow since it will becarried along by the bitumen matrix, rather asfine sand is carried along by water. At the otherextreme, assume that the same amount of aggreg-ate is supplied in the form of a few large boul-ders. The bitumen will simply flow in betweenthem. Between these two extremes there lies aregion where the particles are small enough andsufficiently close together to restrict flow.
So it is in metals. By various means, the metal-lurgist is able to produce dispersions of hard par-ticles of such a size and of such a spacing that themovement of dislocations is impeded with a con-sequent increase in yield strength. This facility isused in certain aluminium alloys, where the dis-persion is brought about by suitable heat treat-ment, and increasingly in steel metallurgy.
9.7 Unstable microstructuresMany of us know that if you take a piece of steelcontaining, say, 0.5%C, heat it to glowing red(�900°C) and then quench it by placing it in a bathof water, the outcome is a very hard but brittle sub-stance. Indeed, it could be used to cut a piece ofsteel that had not been so treated. The quenchingof steel is an example in which an unstablemicrostructure is generated when there is no timefor diffusion to keep up with the requirements ofthermodynamic equilibrium. The procedure gener-ates a new and unexpected structure, and in whichthere are large, internal locked-in stresses. It iscalled martensite. In the as-quenched condition it istoo brittle to be useful, but if heated to just a fewhundred degrees C a number of subtle changescome about. The steel is softened a little, not much,but a useful degree of toughness is restored. Thissecond heating is called tempering, and gives ustempered steel. In this state it finds many uses ascomponents in machinery, gears, cranks, etc.
One difficulty with this method of obtainingstrength is that it works well only for certaintypes of steel!
Chapter ten
Forming ofmetals
10.1 Castings10.2 Hot working10.3 Cold working10.4 Joining
We now look at the various ways in which metalsand alloys may be prepared for industrial use.There are many ways available to the metal pro-ducer and the treatment here is of necessity brief,and will concentrate on only a few of the moreimportant ones. Further information may befound in, for example, Thomas (1970).
Before looking at some of the processes inmore detail, it must be emphasised that metallur-gists look on them not only as ways of shapingmaterials but also as ways of controllingmicrostructures and, consequently, properties.
Figure 10.1 outlines the processing routes formost of the more common metals and alloys usedin structural engineering.
10.1 CastingsMost common metals can be produced bymelting and casting into moulds. By casting, themetal may be shaped and dimensioned accordingto the required component geometry, or a prismof material may be produced for further process-ing.
The general process of solidification of amolten pure metal has already been described.Solidification of alloys is more complicated andoften gives rise to compositional variations fromplace to place in a casting and on a microstruc-
tural scale within dendrites. When intended forfurther processing, little attempt is made tocontrol grain size and the metal often solidifies toa rather coarse grain structure and will contain anumber of casting defects, such as porosity, com-positional variations and shrinkage. These are notdisastrous because further processing will rectifythem. Shaped castings need more care. They arenormally degassed, the grain size carefully con-trolled by means available to the foundryman,and compositional variations minimised by atten-tion to solidification patterns within the mould, ifthe desired mechanical properties are to beachieved.
69
Melt
Cast as ingot
Mechanically work and shapee.g.– rolling
– forging– extrusion– drawing
Cast to shapee.g. – sand casting
– die casting– investment casting
Finishe.g.– machining
– joining
FIGURE 10.1 Processing of the more commonengineering materials.
10.2 Hot working
Forming of metals
70
Drawing
Die
Rolling
ForgingExtrusion
DieRam
BendingDeep drawing
Clamp Clamp
Roller
Roller
Shearing Stretch forming
Die
FIGURE 10.2 Some metal-forming methods.
The working of metals and alloys by rolling,forging, extrusion, etc., depends upon plasticitywhich is usually much greater at high tempera-tures, i.e. temperatures above their recrystallisa-tion temperature. This allows all the commonmetals to be heavily deformed, especially incompression, without breaking. For steel struc-tural members, the most usual method is by hotrolling between simple cylindrical or shaped rollsat temperatures around 1000°C or more. Afterrolling, the members are left to cool naturally and
end up with annealed microstructures and grainsizes which depend on how heavy was the defor-mation, at what temperature, and how fast wasthe cooling rate. It follows that all these processvariables need to be controlled to give productswith consistent properties. Another feature of hotworking processes is that the exposure to air athigh temperatures causes a heavy film of oxide toform on the surface. Thus steel sections delivered‘as rolled’ are covered with iron oxide (mill scale)and need to be shot blasted or sand blastedbefore receiving any protective coating.
Many familiar articles, e.g. engine crankshafts,are forged into shape. This involves placing a hotblank into one half of a shaped mould and thenimpressing the other half of the mould onto theblank. This is done under impact using suchmethods as drop forging, die stamping, etc., ormore slowly using large hydraulic presses.
Many metals can be extruded. This has theadvantage that very long lengths with complexsections can be produced. Aluminium glazingbars are a familiar example.
One disadvantage of hot forming arises fromthe contraction of dimensions on cooling andfrom such problems as oxidation. These andother factors conspire to limit the precision of theproduct. In some cases the tolerances of hotformed parts are acceptable. In other cases,further cold forming or machining may beemployed to meet more demanding tolerances.
10.3 Cold workingBecause of their cold ductility, many metals andalloys can be cold worked, that is to say, shapedat temperatures below the recrystallisation tem-perature of the alloy. This creates an immensenumber of dislocations and, as a consequence, themetal work hardens and its yield point is raised.Indeed, for pure metals and some alloys it is theonly way of increasing the yield strength.
There are many processes. Cold rolling isextensively used to produce sheet material, whilehigh-strength wire, as used for cables, is colddrawn by pulling through a tapered die. Metalsheets can be shaped into cups, bowls or motor
car body panels by deep drawing or stretchforming (Figure 10.2).
Clearly there comes a limit beyond which theductility is exhausted and the metal will fracture.If further cold work is required the metal must beannealed by heating it to a temperature whererecrystallisation occurs when the original ductilityis restored and further working is possible. Somemetals can be cold extruded.
Cold working using well-designed tools and alittle thought about how to go about it is capableof delivering to demanding tolerances. From themetallurgist’s point of view, control over rollingand annealing schedules is a very effective way ofcontrolling the grain size of the product.
10.4 JoiningAlthough adhesive bonding is being used increas-ingly for joining metal parts, the commonestmethods are still welding, brazing, soldering orby mechanical fasteners, such as rivets and bolts.
10.4.1 Welding
It is beyond the scope of this book to list thevarious welding processes that are available (see,for example, Houldcroft (1977)). All weldinginvolves essentially the same sequence of opera-tions at the joint. The material is heated locally toits melting temperature, additional metal may ormay not be added and the joint is then allowed tocool naturally. Whatever the material or processall welds should comply with the two followingideal requirements:
1. There should be complete continuity betweenthe parts to be joined and every part of thejoint should be indistinguishable from theparent metal. In practice this is not alwaysachieved, although welds giving satisfactoryperformance can be made.
2. The joining material should have metallurgi-cal properties that are no worse than those ofthe parent metal. This is largely the concern ofthe supplier of welding consumables thoughpoor welding practice can significantly affectthe final product.
The weld itself is a small and rapidly formedcasting, but is surrounded by a heat-affected zone(HAZ). During welding, a temperature gradient iscreated in the parent material. This gradientranges from the melting temperature at the fusionzone to ambient temperature at some distantpoint. In the regions that have been exposed tohigh temperature and fast cooling rates, metallur-gical changes can occur. The quality of the jointis affected by:
1. the structure and properties of the weld metal;2. the structure and properties of that part of the
parent material which has undergone asignificant thermal cycle (the HAZ).
Both of these are significantly affected by the rateof cooling after welding; the slower the rate ofcooling, the closer the structure is to equilibrium.Cooling occurs, principally, by conduction in theparent material and since the thermal conductiv-ity of a metal is a fixed property, the controllingfactor is the thermal mass, i.e. the thickness andsize of the material to be welded. The greater thethermal mass the faster the cooling rate.Responses to rapid cooling differ markedly frommetal to metal, not only from, say, aluminium tosteel but also from one steel to another. Struc-tural steels are designed to be weldable, i.e. theyare able to be welded without serious loss ofperformance in the weld and heat-affected zone.Nevertheless, the job must be carried out withthought, care and skill. The best way of copingwith these sorts of problems is to get advice froma good welding specialist.
10.4.2 Brazing, soldering and gluing
Brazing and soldering, and in some cases gluing,involve joining by means of a thin film of amaterial which has a melting temperature lowerthan that of the parent material and which, whenmelted, flows into the joint, often by capillaryaction, to form a thin film which subsequentlysolidifies. A good brazed or soldered joint shouldhave a strength that is not too different from thatof the parent material. Quite high forces areneeded to break a film of liquid provided the film
Joining
71
is thin enough (Chapter 5) and the same appliesto thin solid films. This is not quite the wholeexplanation but is a very significant part of it.The rest is associated with the behaviour ofmaterials under complex stress conditions, biaxialand triaxial, and is beyond the scope of thischapter. See, for instance, Cottrell (1964).
Although it may seem strange to say so, gluingworks in a very similar way. Thin layers ofmodern adhesives bond well to the substratematerial and are strong in shear.
Design of joints to be made by gluing, solder-ing or brazing should avoid potential failure bypeeling and aim to use the adhesive in shear.
10.4.3 Pinning
Some materials (such as cast iron) do not lendthemselves to joining by welding. Gluing andbrazing are valid options but require thoughtabout the joint design. Bolting or riveting are byfar the most common ways of making jointsinvolving such materials. Both bolting and rivet-ing rely on friction. A tightened bolt forces thetwo members together and the friction betweennut and bolt at the threads holds it in place. Inriveting, the hot rivet is hammered into preparedholes. As it cools it contracts and develops atensile stress which effectively locks the memberstogether. High-strength friction grip (HSFG) boltsused in structural steelwork combine bothaspects, the nut is tightened to place the bolt intotension, and this tensile pre-stress acts in the sameway as the tensile stress in a rivet.
Forming of metals
72
Chapter eleven
Oxidation andcorrosion
11.1 Dry oxidation11.2 Wet corrosion11.3 Control of corrosion11.4 Protection against corrosion
Corrosion can be divided into two areas: dry oxi-dation and wet corrosion.
11.1 Dry oxidationThe earth’s atmosphere is an oxidising one, andof the metals, only gold and silver are widelyfound in the native, i.e. unoxidised, state. Thegeneral oxidising reaction is often written:
M�O→MO (11.1)
where M is the metal and O is oxygen. Mostusually, this takes place in two steps. First, themetal forms an ion, releasing electrons, and theelectrons are accepted by oxygen to form an ion.Second, the ions attract one another to form thecompound:
M→M�� �2e & O�2e→O�� and thenM�� �O�� →MO (11.2)
At the surface, the oxygen ions attach themselvesto the surface to form a thin layer of oxide.Thereafter the metal ions M�� and electrons mustdiffuse outwards through the oxide to meet O��
at the outer surface, or the oxygen ions mustdiffuse inwards. The rate of oxidation is deter-mined by whichever reaction can proceed thefaster and, largely, this is controlled by the thick-ness and structure of the oxide skin.
On some metals the oxide occupies less volumethan the metal from which it was formed. If it isbrittle (and oxides usually are), it will crack andsplit, exposing fresh metal. On other metals, theoxide occupies more volume and it will tend towrinkle and spring away, again exposing freshmetal.
In some cases, however, the oxide volumematches the metal volume and thin adherent filmsform that act as barriers to further oxidation.This is true of aluminium, chromium and nickel,the last two metals being essential components ofso-called ‘stainless steel’ for that very reason.
11.2 Wet corrosionIn the presence of moisture, the situation changesdrastically and the loss of metal by corrosionbecomes appreciable. It has been estimated that,in the UK, the annual corrosion bill for eitherreplacement or prevention is around £5000million.
As in dry oxidation, the basic mechanisminvolves ionisation but, if the ions are soluble inthe corroding medium, which is usually, but notnecessarily, water, the metal steadily corrodes.Consider a simple cell in which two electrodes,both of the same metal, are connected via abattery and placed in a suitable electrolyte (Figure11.1). The battery pumps out electrons whichincrease the negativity of the cathode while, atthe same time, electrons are leaving the anode inorder to maintain the flow of current. The elec-trolyte completes the circuit and the metal ions go
73
into solution and flow across to try to neutralisethe surplus negative charge on the cathode. Thusthe anode corrodes and the metal ions collect on the cathode. This is, in fact, a simple electro-plating cell and is used to produce metals of ahigh degree of purity. ‘Electrolytic’ copper usedfor its high electrical conductivity is one example.
However, in some cases, the metal ions reactwith the electrolyte. Iron is one example, and ifwe were to set up a cell similar to that in Figure11.1 we would obtain the following reactions:
at the anode:
Fe→Fe�� �2e� or Fe→Fe��� �3e�
H2O→H� � (OH)�
and 2H� �2e� →H2 (11.3)
and, at the cathode:
Fe�� �2(OH)� →Fe(OH)2 or Fe��� �3(OH)� →Fe(OH)3 (11.4)
Iron hydroxides Fe(OH)2 and Fe(OH)3 eitherdeposit away from the cathode or, if they depositon the cathode, they do so as a loose depositgiving no protection. It is, in fact, the orangeslimy oxide often seen on, say, sheet piling instagnant water. But, since, for example:
2Fe(OH)3 →Fe2O3 �3H2O (11.5)
evaporation of water to the air leaves the moreconventionally recognised rust, a mixture of ironhydroxides and, principally, Fe2O3.
Oxidation and corrosion
74
FIGURE 11.1 A simple cell.
TABLE 11.1 Standard electrode potentials
Electrode Voltage
Na�� �2.71 base, anodic, corrodesAl�� �1.66Zn�� �0.76Fe�� �0.46Ni�� �0.25Sn�� �0.14H� �0.0 referenceCu�� �0.34Ag�� �0.80Pt�� �1.20Au�� �1.50 noble, cathodic
Note: Conventions differ as to which are negative and whichare positive in the series. This does not matter too much sinceit is the relative position that is important. The conventionused here is more directly associated with the text.
For the above a battery was used to drive thereactions along; however, the reinforcing bars instructural concrete are not connected to a batteryand yet they rust. Why?
In these circumstances, the whole subtlety ofthe process often makes it difficult to establishwhat is happening, and many large books havebeen written about it. This is not surprising, asnearly every case of corrosion involves somedepartures from the ideal but here we can onlyconsider a few of the more general phenomena.
11.2.1 Electrochemical series
The process of ionisation itself produces a changein electric potential and this can be measured rel-ative to some convenient reference value. The ref-erence value usually selected is that of theionisation of hydrogen:
H→H� �e� (11.6)
and the emf produced by a given metal relative tothis reaction gives the electromotive series. Anabbreviated version is given in Table 11.1. Metalswhich are more positive than the reference valueare anodic and will corrode, metals which arerelatively cathodic will not.
One of the first batteries for generating electric-
ity was the Daniell cell (Figure 11.2) containing azinc anode which corroded and a copper cathodewhich did not. With zinc at �0.76V and copperat �0.47V, the cell produced an output of1.23V.
The electromotive series given in Table 11.1was developed under laboratory conditions. Inthe real world the nature of the electrolyte, seawater, boiler water, etc., may change the order ofmerit. Table 11.2 gives an example. When pairsof metals are immersed in sea water one willbecome the cathode and unreactive, the other willbecome the anode and react (i.e. corrode). Table
Wet corrosion
75
TABLE 11.2 Galvanic series in sea water
magnesium anodic, corrodeszincaluminiummild steelcast ironleadtin60–40 brass70–30 brasscoppernickelsilverstainless steel (18Cr 8Ni)monel metal (70Ni 30Cu) cathodic, noble
FIGURE 11.2 A Daniell cell.
11.2 indicates that brass and stainless steel fit-tings would form a cell in which the brass wouldcorrode. Other situations which are more com-monly found in construction are listed in Blyth(1990). These also include metal–non-metal cells,such as steel–concrete, etc.
11.2.2 Cells
The above is an extreme example of a composi-tion cell, i.e. a galvanic cell which arises as aresult of the energy difference between twometals. But there are ways in which a composi-tion cell can be used to advantage.
The most familiar is the use of zinc to galvanisesteel. Table 11.2 shows that zinc is more anodicthan iron and on exposure to the atmosphere thezinc corrodes first, thereby protecting the steel.The same system is used for buried pipelineswhere zinc slabs are connected to the steel pipeand are periodically replaced when they corrode.The converse occurs with chromium plating.Chromium is more cathodic than iron so that, if asmall pit appears in the chromium, the steelunderneath rusts away quite rapidly. The phe-nomenon will be familiar to the owners of elderlycars.
There are, however, some variations whoseorigins are, at first sight, less obvious. In the so-called concentration cell preferential corrosion isa consequence of a difference in the constitutionof the electrolyte itself. For example, considerwater containing dissolved oxygen to differingconcentrations in different regions. The reaction:
2H2O�O2 �4e� →4(OH)� (11.7)
removes electrons and these must be suppliedfrom adjacent areas which are deficient inoxygen. These act as the anode and hencecorrode. Thus, in a riveted or bolted connection,corrosion will occur in the inaccessible (i.e.oxygen-poor) areas as indicated in Figure 11.3. Aclassic case is the ‘waterline’ corrosion of steelpiling in stagnant water. Here the surface layersof the water are richer in oxygen and become thecathode. The lower, oxygen-deficient layers areanodic and corrosion occurs locally. Much the
same mechanism applies to pitting corrosionwhere the oxygen-poor region at the bottom ofthe pit is anodic and the pit therefore tends todeepen, often leading to premature failure byfatigue or brittle fracture.
Note that water line corrosion can be confusedwith a different phenomenon. We are all familiarwith the enhanced corrosion that is seen on steelsupports of seaside piers, etc. Here we have aregion that is washed by wave and tidal action.The region is alternately wetted and dried and itis this that accelerates corrosion.
11.3 Control of corrosionOf all the problems facing the design engineer,the management and control of corrosion is oneof the most difficult. But it cannot be stressed toofirmly that the problems start at, and must betackled at, the design stage. There are threerequirements, all easy in theory but difficult inpractice.
1. Understand the environment in which themetal must work, whether polluted or not,whether facing or away from pervadingsources of corrosion, whether wet and/orhumid or dry, whether these conditions arestable or variable. Thus, given adequate cover
Oxidation and corrosion
76
FIGURE 11.3 Concentration cells: (a) at rivetedconnection; (b) waterline corrosion.
and ‘normal’ working conditions, steel rein-forcement in concrete should not corrode. Butit does, often because of sloppy workmanshipand poor inspection leading to inadequate andover-porous cover. Furthermore, if it is amotorway bridge, a combination of the exces-sive use of de-icing salts with inadequatedrainage will speed up corrosion.
2. Consider the ‘design life’. How long beforethe first major maintenance? Are you design-ing a throw-away structure like a modernmotor car or are you designing a bridge for acentury of service? If the component is notexpected to outlive the structure as a whole,how easy is it to inspect and replace?
3. Select the most appropriate method ofcontrol. You will, of course, be excused forimagining that the ‘most appropriate’ methodis that one which involves the longest life.You will, of course, be wrong. In the commer-cial world the most appropriate means ofcontrol is that one which produces the longestlife at the least annual cost. So, on the whole,you would be best to master such matters aspayback, rate of return, and discounted cashflow before deciding upon an appropriatetechnology.
11.4 Protection against corrosion
11.4.1 Design
At ambient temperatures corrosion occurs only ifmoisture is present. Thus, surfaces should beexposed as little as possible to moisture andarranged so that they dry out quickly afterwetting. In practice all surfaces are at risk, verti-cal surfaces suffer ‘run off’, flat surfaces retainmoisture on their top side and can attract dewand condensation on the under side. Water reten-tion by ‘V’, ‘H’ and other channel sections isobvious and drain holes should be provided, ifmechanically acceptable. Overlaps and jointsshould be arranged to avoid the formation ofwater channels (Figure 11.4). Porous materials,which can retain moisture, should not be incontact with metals.
Steel which has been allowed to rust on siteposes a threat, for the methods available to cleanthe steel are usually less than adequate and somerust will inevitably remain at the bottom of thepits formed during rusting. These will containsufficient active material for rusting to continuebelow any paint film. The only real remedy is notto let rusting start by protecting the steel bypriming coats as an integral part of the manufac-turing process and, if these are damaged duringerecting, to repair the damage as soon as possible.
Remember also that structures need mainte-nance and the design should ensure that this canbe carried out effectively.
11.4.2 Isolation from the environment
This is done by applying one or more protectivecoatings to a suitably prepared surface. Somemetallic coatings simply form a protective barrier,e.g. nickel or chromium on steel. Other coatings,e.g. zinc, cadmium and aluminium, are anodic
Protection against corrosion
77
FIGURE 11.4 Design of joints to minimise corro-sion risks. Note that small relative movements at jointscan quickly abrade away any protective coatings ofprimer or paint.
with respect to steel and provide sacrificial pro-tection.
Organic coatings, such as paints, pitch, tar,etc., form a protective barrier and are commonlyused, often in conjunction with a metallic primer.There are now so many different types of paintavailable that it would be impossible to sum-marise the virtues and limitations here. For steelwork the publications of the British Con-structional Steelwork Association provide excel-lent guidance for the practising engineer.
It must be emphasised that all paint coatingseven of the highest quality and meticulous appli-cation are only as good as the quality of the pre-ceding surface preparation. Application, whetherby brushing or spraying, should always be carriedout on dry surfaces and in conditions of lowhumidity.
11.4.3 Cathodic protection
In addition to the use of sacrificial anodes,cathodic protection can be achieved by the use ofan external power source to make the metalcathodic to its surroundings. Inert anodes areused, commonly carbon, titanium, lead or plati-num. The procedure is not without its problems.Thus, in many cases, the cost of replacementanodes is greater than the cost of the impressedpower supply.
In buried structures, secondary reactions withother nearby buried structures may enhance,rather than control, corrosion and there is thepossibility of hydrogen evolution at the cathode.This can diffuse into the metal and embrittle it.Nonetheless, this method of cathodic protectionhas been quite widely used in marine environ-ments, especially on offshore oil rigs. Similarly,considerable attention has also been paid to theapplication of cathodic protection to concretereinforcement.
78
Chapter twelve
Metals, theirdifferences anduses
12.1 The extraction of iron12.2 Cast irons12.3 Steel12.4 Aluminium and alloys12.5 Copper and alloys
Metals are divided into ferrous (based on iron)and non-ferrous (based on metals other thaniron). The principal ferrous alloys in use todayare those of iron and carbon, i.e. cast iron andsteel. The cast irons contain more than 1.7%C,steels contain up to about 1.5%C with structuralsteels containing only about 0.25%C or less. Themain non-ferrous alloys of interest in civil andstructural engineering are those based on alu-minium and copper.
12.1 The extraction of ironLike all metals, iron is extracted from naturallyoccurring ores. These are actually quite complexchemical compounds but, for simplicity, we shallassume that the starting material is iron oxide.High temperatures (�1600°C) are needed toallow the reaction:
FeO�C→Fe�CO (12.1)
The CO is collected and used as a fuel in thesteelworks. The pig iron, which contains about 4per cent of carbon, is industrially useless unlessprocessed further. It is remelted, often with scrap
iron or steel, and by controlled oxidation usingan air blast, the carbon content may be reducedto 2–4 per cent. This is cast iron which, as itsname implies, is cast directly into sand moulds. Itcan be produced in a tough form (SG iron), butusually it is a brittle material and is best used incompression rather than in tension.
12.2 Cast ironsThese are used in a variety of applications, themajor consumption being in pipes and fittings forservices. In civil engineering, an important use is fortunnel segments and mine shaft tubing. The engi-neer should, however, be aware that cast iron wasone of the dominant structural materials in thenineteenth century and will be found as beams,columns and arches in many rehabilitation andrefurbishment projects. As we shall see, it should betreated with respect. There are four principal types,whose properties are summarised in Table 12.1.
12.2.1 White cast iron
The carbon here is combined as hard brittle ironcarbide, Fe3C, and the cast iron overall is hardand very brittle. This makes it structurally unde-sirable and its main use is for applications wherehigh resistance to wear and abrasion is required.Typical applications in the civil engineering fieldinclude hard facings to earth-moving machinery.
12.2.2 Grey cast iron
Old cast iron structures are almost certainly ofthis type and it is still the form of cast iron mostcommonly met. Most of the carbon is present inthe form of free graphite flakes, which can bequite large. The graphite makes it softer thanwhite cast iron and allows it to be machined.Nevertheless, it is still a brittle material and itsuse in tension is inadvisable, but it has goodstrength in compression.
12.2.3 Spheroidal graphite (SG), nodularor ductile irons
The flakes in grey cast iron act as internal notchesand the metal is brittle in tension. Graphite maybe induced to form spherulites by the addition ofcertain alloying elements and correct casting pro-cedure. These irons have good strength, tough-ness and ductility. Modern tunnel linings andmany other applications are of this type of iron.
12.2.4 Malleable irons
These used to be produced from white irons byannealing which results, again, in nodules ofgraphite. Depending on the process used, and thecomposition, they may be described as whiteheartor blackheart (this derives from the appearance ofthe fracture surface). Both have good strengthand resistance to impact. The procedure has beensuperseded in recent years by the production ofSG irons.
12.2.5 Joining of cast irons
All cast irons are extremely difficult to weld and,all too often, the welds are unreliable and of poorstrength. Unless specialist advice is available theengineer should use brazing rather than welding.Riveting is possible, but again needs care for fearof cracking the iron. Bolting is the safest methodof joining.
12.3 SteelSteel making involves some very complex thermo-chemistry, but the basic reaction is simply that ofreducing the carbon content still further (forstructural steels down to about 0.2 per cent orlower) by a process of controlled oxidation:
2Fe[C]�O2 ↔ 2Fe�2CO (12.2)
To keep the reaction moving to the right a con-siderable amount of oxygen must be used. Somedissolves in the liquid steel. If not removed thiswould form hard, brittle iron oxide, FeO, and wewould be back at square one. Thus, when thedesired carbon content is reached the residualoxygen is ‘fixed’ as an oxide which, after a periodof resting, rises to the surface and is removed asslag. The metals commonly used to fix the oxygenin this way are manganese and silicon, and steelstreated in this way are known as killed steels.Manganese is added for another reason also. Oneof the most persistent impurities, and one that isdifficult to eliminate economically, is sulphur.This could form iron sulphide, FeS. If FeS is
Steel
79
TABLE 12.1 Representative properties of cast irons
Material Young’s Compressive Yield Tensile %E Brinell Impactmodulus strength strength strength hardness energy
(GPa) (MPa) (MPa) (MPa) (kgm�2) (J)
White 170 – – 275 0 500 4 CUGrey 100–145 600–1200 – 150–400 0.2–0.7 130–300 8–50 IUFerritic SG 165 – 240–400 400–600 18–6 115–215 5–15 CNPearlitic SG 170 – 270–530 450–700 7–2 140–300 2–10 CN
%E�% Elongation to fracture, CU�Charpy unnotched, CN�Charpy notched, IU� Izod unnotched.
present, even in small quantities, it can cause adefect known as hot shortness, in which the steelcracks disastrously if it is stressed when hot. Evencontraction stresses on cooling can lead to irre-deemable cracking. The manganese additioncounteracts this by forming relatively innocuousMnS, and so even the simplest steels containsilicon and manganese.
The terminology used to describe steels is tradi-tional and far from exact. With up to about 0.25per cent carbon they are ‘mild steel’ or ‘low carbonsteel’ (structural steels come into this category).Between about 0.3 per cent and 0.6 per centcarbon they are ‘medium carbon steels’ or often‘carbon steels’. Above about 0.6 per cent they are‘high carbon steels’. If elements, other than thenormal Mn and Si, are added they become ‘alloysteels’ and, if certain elements, notably chromiumand nickel, are added in quantity, ‘stainless steels’.
We discussed the iron–carbon equilibriumphase diagram in Chapter 2 (Figure 2.11) and itis worthwhile having a look at this and re-readingthe accompanying text before reading on.
12.3.1 Structural steels
Structural steels are processed into the requiredsection shapes and lengths by hot rolling and theirmicrostructures are effectively those of a nor-malised steel. They consist of two phases, ferriteand Fe3C. The Fe3C does not appear as large iso-lated particles, but combines with ferrite intolaminar regions of alternating layers of ferrite andFe3C (see Figure 8.9). These layers are generallyabout 0.5�m thick and act as a diffraction gratingfor natural light. Thus, a steel containing only suchregions and properly prepared has a colouredpearly appearance and is known throughout theworld as pearlite. The overall composition ofpearlite is close to 0.8 per cent carbon. Steels con-taining less carbon are mixtures of ferrite andregions of pearlite, as illustrated in Figure 12.1. Toa working approximation, ferrite contains nocarbon and so the proportions of ferrite:pearlitedepend linearly on the carbon content between 100per cent ferrite at 0 per cent carbon to 100 per centpearlite at 0.8 per cent carbon. It follows that at
Metals, their differences and uses
80
FIGURE 12.1 Typical microstructure of a low-carbon (structural) steel.
400
300
200
100
00
20
40
60900
600
300
0
Tens
ile s
tren
gth
(MPa
)
Elon
gatio
n (%
)
Har
dnes
sTensile strength
Hardness
Elongation
1.40.8
Carbon (mass%)
0
50
100
0
Pear
lite
(mas
s%)
1.40.8
FIGURE 12.2 The influence pearlite content, orcarbon content, on the strength and ductility of steels(adapted from Rollason, Metallurgy for Engineers,1961, Arnold, by permission Butterworth-Heinemann,a division of Reed Educational & Professional Publish-ing Ltd.).
high carbon contents (say 0.7 per cent) the proper-ties of the steel are dominated by those of pearlite,high hardness, high strength and poor ductility andtoughness. Conversely, at say 0.15 per cent carbon,the pro-perties are dominated by those of theferrite. This is a typical metallic substance and itsproperties are governed by its grain size andwhether or not it has been strain (work) hardened.
From all of this it might be expected that theproperties of steels are affected strongly by howmuch pearlite they contain, or to put it anotherway, by their carbon content. Figure 12.2 illus-trates this.
The tensile strength increases approximatelylinearly from about 300MPa at 0 per cent carbonto about 900MPa at 100 per cent pearlite (0.8per cent carbon). Over the same range, the elon-gation to fracture decreases from about 40 percent to nearly zero. This is a bit of an oversimpli-fication, because the properties of low carbonsteels are affected by the grain size of the ferritethat occupies the greater part of the microstruc-ture. However, before going into this, we need tonote another very important feature of thebehaviour of steels.
Steels, especially structural steels, can gothrough a ductile to brittle transition as the tem-perature of use changes over ranges which aretypical of those due to variations in weather,season and climate. This phenomenon is usuallyshown up most clearly by impact tests, whichgive results like those in Figure 12.3. The methodnormally used is the Charpy test in which anotched bar is struck by a hammer of knownmomentum and the energy used up in causingfracture is deduced from the follow through ofthe pendulum. Even though this phenomenonappears to have been first noted by I.K. Brunel in1847, it still brings about its fair share of failures.
The trick is to formulate steels for which theductile to brittle transition temperature (DBTT) islow and which can also be joined successfully bywelding. To do this, we want the carbon contentto be low, a high ratio of manganese to carbonand a small grain size of the ferrite. When werecall that the yield strength is increased by redu-
Steel
81
FIGURE 12.3 Energy absorption of structural steel(schematic).
cing grain size, we have a powerful argument infavour of fine-grained steels. These are summar-ised in Table 12.2. To produce and maintain finegrain sizes, careful control must be exercised overthe temperatures of hot rolling, the amounts ofdeformation imposed and cooling rates. As mightbe expected, so-called controlled rolled steels aremore expensive than less carefully controlledproducts.
Prior to 1990, the relevant British standard wasBS 4360 ‘Weldable structural steels’ and muchsteel made to that specification is still in use. It isnow superseded by BS EN 10025: 1993 ‘Hotrolled products of non alloy structural steels’ andBS EN 10113: 1993 ‘Hot rolled products in weld-able fine grain structural steels’. This division intotwo standards emphasises once again the import-ance of grain size. Generally these steels contain�0.16% C, 0.1 to 0.5% Si and 1.5 to 1.6% Mn.They are all designed to be weldable. Extractsfrom these standards and equivalents betweenthem are given in Table 12.3.
In essence, there are four strength grades. The1993 standards identify them by numbers thatare the same as the minimum yield strength of thesteel: the older schemes referred to the tensilestrength. A number of points arise from thesestandards:
1. The specified minimum yield strength ingrades 235 and 275 (40 and 43) can beobtained by various combinations of carbonand manganese.
2. Sulphur and phosphorous are both deleteri-ous. Maximum permissible values are alwaysspecified.
TABLE 12.2 Illustrating the importance of grainsize to the properties of low carbon structural steels
Grain size �y Ductile to brittle(�m) (MPa) transition temperature
(°C)
25 255 010 400 �405 560 �60
TA
BL
E 1
2.3
Ext
ract
s an
d ap
prox
imat
e eq
uiva
lent
s fr
om B
riti
sh S
tand
ards
for
str
uctu
ral s
teel
s
BS
EN
BS
EN
Pre
viou
sB
S 43
60:
%C
%M
n%
SiO
ther
(%
)Y
ield
str
engt
hT
ensi
le s
tren
gth
1002
5: 1
993
1011
3: 1
993
desc
ript
ion
1990
(Max
)(M
ax)
(MP
a)(M
Pa)
S 23
5Fe
360
400.
06–0
.22
1.5–
1.6
0.5
235
340–
500
S 27
5Fe
430
430.
06–0
.22
1.5–
1.6
0.5
275
430–
580
S 35
5Fe
510
500.
06–0
.22
1.5–
1.6
0.5
0.1N
b, 0
.1V
355
490–
640
S 46
055
0.06
–0.2
21.
5–1.
60.
50.
1Nb,
0.1
V45
055
0–70
0
BS
EN
BS
EN
Pre
viou
sB
S 43
60:
Tem
pera
ture
(°C
) at
%
P%
SE
xam
ples
1002
5: 1
993
1011
3: 1
993
desc
ript
ion
1990
whi
ch im
pact
ene
rgy
is:
BS
EN
100
25B
S 43
6027
J40
JS2
75JO
43C
JRB
200.
050.
05�
y�
275
MPa
JOC
00.
040.
04�
max
�43
0–58
0M
PaJ2
D�
200.
040.
0427
J at
0°C
K2
N o
r M
*D
D*
�20
E*
�40
BS
EN
100
25B
S 43
60N
L o
r M
L*
EE
*�
50�
300.
040.
03S3
55K
250
DD
F*�
600.
025
0.02
5�
y�
355
MPa
�m
ax�
490
�64
0M
Pa40
J at
�20
°C
N�
norm
alis
ed o
r no
rmal
ised
rol
led;
M�
ther
mom
echa
nica
l rol
led;
L�
low
tem
pera
ture
(�
50°C
) im
pact
s; *
�a
grai
n-re
fined
ste
el; %
m
ass%
.
3. The lower grade structural steels are normallydelivered normalised (cooled in air). The term‘as rolled’ implies no further treatment. Fre-quently, a steel which had been rolled at hightemperatures, above �950°C and just left tocool naturally, is effectively a normalised steel.
4. The specified yield strength decreases with sizeof section. This size effect arises from the factthat thick sections cool more slowly than thinones and, consequently, the grain size is largerand the Fe3C ends up differently distributed.
5. The higher grade steels, 355 and 460 (50 and55) achieve their enhanced strengths not byincreasing the carbon content but by attentionto their grain sizes. This is done by addingniobium and/or vanadium to the steel, insmall amounts, and by careful control ofrolling schedules, i.e. thermomechanical treat-ment. The small grain size is also required togive a low ductile to brittle transition temper-ature. To give information about the low tem-perature performance of steels, a letter code isadded according to Table 12.3.
6. All structural steels are designed to be readilyweldable, although some care is needed withthe higher, grain refined grades if the low tem-perature performance is to be retained (seebelow).
12.3.2 Cold rolled steels
Many lightweight sections are produced fromcold rolled steel of very low carbon content.Strength is derived from work hardening of theferrite, and good control over section sizes andshapes is possible. Examples of applicationsinclude lightweight lintels, angle sections, hollowsquare sections made by welding two anglestogether and so forth. Welding, of course, willlocally anneal the material, with consequentchanges to properties in the heat-affected zone.
Sheet and other sections are available in aspecial steel (Cor Ten) which contains a smallquantity of copper, in addition to the usual alloy-ing elements. When exposed to rainwater it ruststo produce a hard, adherent and subsequentlyprotective oxide layer of an attractive brown
colour. It is finding an increasing range of archi-tectural applications as cladding for buildings.
12.3.3 Heat treated steels
With a carbon content greater than about 0.3 percent, the properties of steel can be varied by heattreatment, that is to say, by fast cooling (gener-ally by quenching in oil or water) from a hightemperature, followed by reheating to tempera-tures not exceeding about 650°C (tempering).The fast cooling produces a hard brittle micro-structure, known as martensite, in which all thecarbon is trapped and which is of little use,except for a few applications such as cutting toolsor cutlery. The structure is, however, metastableand the reheating causes the carbon to beprecipitated as tiny particles of carbides through-out the matrix. The loss of carbon from themartensite in this way allows it to become softerand more ductile. By varying the tempering tem-perature and controlling the amount of carbonleft in the martensite, great control over the pro-perties can be achieved (Figure 12.4).
Such steels do not, however, find great applica-tion in structural engineering. High strength fric-tion grip (HSFG) bolts are one. These aresupplied in the hardened and tempered condition.It follows that these should not be reheated (e.g.by welding or flame cutting) or the effects of theheat treatment may well be cancelled. The boltcould easily be embrittled.
It is, however, important to note that thehigher strength structural steels can also hardenin the same way, not to the same extent butenough to compromise their properties. Welding,and especially flame cutting, of these gradesshould be undertaken with caution and duerespect for the specification and any recommen-dations by the manufacturer of the welding con-sumables. Better still, consult an expert weldingengineer.
12.3.4 Stainless steel
The term ‘stainless steel’ covers a wide range offerrous alloys, all of which contain at least 12 per
Steel
83
cent of chromium which produces a stablepassive oxide film. Other alloying elements,notably nickel and molybdenum, may also bepresent. There are three basic types, groupedaccording to metallurgical structure (Table 12.4):
Metals, their differences and uses
84
FIGURE 12.4 Variation in properties of 1 per centNi steel with varying tempering temperatures. TS:tensile strength; RA: reduction in area; YP: yield point(from O.H. Wyatt and D. Dew-Hughes (1974) Metals,Ceramics and Polymers, by permission of CambridgeUniversity Press).
TABLE 12.4 Representative properties of stainless steels
Type Yield strength Tensile strength % elongation Hardness Izod impact(MPa) (MPa) to fracture energy/J
Martensitic 350–1360 600–1800 33–3 170–450 7–130Ferritic 280 420 20 170 –Austenitic 200 510 40 185 –
Note: the wide range of properties for the martensitic type reflects its ability to be heat treated.
1. Martensitic. The 410 series are low carbonsteels containing 13 per cent chromium. Theyare heat treatable and can be made very hard.Since they retain a keen cutting edge they areparticularly used for cutlery.
2. Ferritic. These (the 430 series) also contain 13per cent chromium but with very low carbon.They are not heat treatable. They are reason-ably ductile middle strength steels.
3. Austenitic. The 300 series are low carbonwith a basic composition of 18 per cent Crand 8 per cent Ni though other additions maybe made. Like ferritic steels, they are not heattreatable, are reasonably ductile and havegood strength.
All these steels offer good resistance to corrosionas long as the passive film can be maintained. Allwill corrode in solutions low in oxygen and thishas been the cause of some embarrassing disas-ters. The austenitic steels are the most resistant topitting corrosion, though they may suffer fromstress corrosion cracking in chloride solutions atslightly elevated temperatures. Type 316 (18 Cr10 Ni 3 Mo) is recommended for all externalapplications. Ferritic steels should be limited tointernal uses.
For all practical purposes, martensitic and fer-ritic stainless steels should be regarded as unweld-able, since both undergo significant changes instructure and properties as a result of the thermalcycle. Ordinarily, austenitic stainless steels can bewelded (BS 2926: 1984). However, they cansuffer from a form of intergranular attack (welddecay), and grades recommended for welding, i.e.stabilised by the use of titanium, should be speci-fied.
12.4 Aluminium and alloys
Aluminium and its alloys are used both struc-turally and decoratively. Pure aluminium andsome alloys can be readily rolled and extrudedinto long lengths with complex cross sectionalshapes. Typical applications include cladding,roofing, window frames, window and door furni-ture, etc. Other alloys are designed to be used ascastings and are generally fairly strong but nottough. Some alloys can be heat-treated to givehigh strengths.
The high cost, however, is still a limiting factorand aluminium can compete with steel onlywhere the inherent properties of the material, i.e.lightness, strength, durability and appearance,can be exploited. The most economical use oflightweight materials is in structures that have ahigh ratio of self-weight to live-load, i.e. roofs,footbridges and long span structures, and wherethe lightness of the material offers advantages intransport, handling and erection. The compara-tively good durability of aluminium makes itattractive in polluted and coastal areas where itshigh initial cost may be offset by reduced mainte-nance. Aluminium alloys may be cast or wroughtwhile structural sections are produced almostexclusively by extrusion.
Because the modulus is significantly lower thanthat of steel, 70GPa compared with 210GPa, the
deflection under a given load is correspondinglygreater and, for deflection limited designs, deeperbeams must be used. At the same time, of course,the density of aluminium is significantly lowerthan that of steel (2.7 against 7.9Mgm�3) sothat specific moduli E/� are very similar,�26GPa/Mgm�3.
The thermal expansion of aluminium and itsalloys is nearly twice that of steel, but the lowermodulus means that the thermal stress developedby a given rise in temperature is less in aluminiumthan in steel.
Unlike mild steel, aluminium and its alloys donot show a definitive yield stress and, for designpurposes, the working stress is usually defined asthat stress at which a small, but acceptable,amount of plastic deformation has occurred. Thisis known as the proof stress, generally defined asthe stress corresponding to a plastic deformationof 0.1 or 0.2 per cent.
The term ‘aluminium’ is normally used toinclude aluminium alloys. There are three generaltypes of alloy, those designed for casting,wrought non-heat-treatable alloys and age hard-enable alloys. Examples of each are given inTable 12.5.
Casting alloys are generally based on a eutecticalloy system, aluminium–silicon being a widelyused example. Solidification is over a narrowtemperature range, which makes them very
Aluminium and alloys
85
TABLE 12.5 Representative properties of aluminium alloy types
Alloy and treatment Yield Tensile % elongation to hardnessstrength strength fracture kgm�2
(MPa) (MPa)
Pure aluminium:annealed 28 70 43 19work-hardened 125 130 6 35
Al–11.5%Si, as-cast, modified – 190 7 –Al–2.5%Mg, non-heat-treatable, work 270 330 10 80hardenedAl–5.5%Cu, naturally aged 310 405 15 100Al–5.5%Zn, 2.5%Mg, 1.6%Cu, 505 570 11 150artificially aged to best strength
suitable for casting into moulds that cause rapidsolidification.
Wrought, non-heat-treatable alloys are avail-able in a wide range of forms: sheet, rod,extruded sections, etc. Their properties are con-trolled by how much they have been strain(work) hardened.
Age hardenable alloys are those whose proper-ties can be changed by heat treatment. Earlier wenoted that plastic flow, by dislocation movement,can be impeded by suitable barriers. The classicexample here is the original alloy, first developedin 1906, Al–4% Cu, Duralumin, upon which thewhole of the aircraft industry depended, and, inmore sophisticated form, still depends.
When heated to around 550°C the copper dis-solves into solid solution in the aluminium andremains in solution when the alloy is rapidlycooled. Thereafter, even at room temperature, afine dispersion of a hard, intermediate com-pound CuAl2 forms slowly. Because the parti-cles are small, actually sub-microscopic, andevenly dispersed throughout the matrix, theyoffer maximum resistance to dislocation move-ment and the yield stress is consequently consid-erably higher than that of pure aluminium(Figure 12.5).
This process, known as ageing, can be speededup by reheating to temperatures of about 150°C.But, if reheated to too high a temperature(�250°C) the minute particles of CuAl2 coalesceand clump together. They are then more widelyseparated, the dislocations can pass easilythrough the matrix and the yield strength is cor-respondingly reduced. This is known as overage-ing. This phenomenon places a restriction on thetemperatures the alloy can be used at if the pro-perties are not to deteriorate during use. Modernalloys are more sophisticated and capable of useat higher temperatures, but the same principlesapply.
The durability of aluminium alloys is, gener-ally, greater than that of steel, but their corrosionresistance depends upon their composition andheat treatment. The fully heat treated alloysgenerally being the most susceptible to corrosionand needing, therefore, some protection.
Metals, their differences and uses
86
FIGURE 12.5 Effects of ageing on the strength ofan Al–Cu alloy: (a) naturally aged at room tempera-ture; (b) accelerated ageing at 150°C; (c) over-ageingat higher temperature.
It will be clear that welding of casting alloysand non-heat-treatable alloys is possible with theusual care. However, the welding of heat treatedaluminium alloys is not without its problems,since the thermal cycle will, inevitably, producean over-aged structure in the parent metal.Although techniques for welding are now wellestablished, bolting, and to a lesser extent rivet-ing, are preferred, especially for joints made onsite.
Steel bolts may be used but should be protectedby a zinc coating. Cold riveting, using rivets up to22mm diameter, is often used and there are manyvarieties, some solid, some hollow, but limita-tions of size and the lower shear strength of alu-minium require that, compared with steel, morerivets are used. In both bolting and riveting,arrangements must be made to keep the bolts orrivets (if not themselves aluminium) electricallyisolated from the aluminium. Otherwise the phe-nomenon of bimetallic corrosion can lead torapid attack of the aluminium.
Most normal structural forms are available andspecial sections can be produced by extrusionmore readily than in steel, though at some cost,and these are really only justified when large
quantities are needed. On the other hand, the useof special sections perhaps allows the designermore freedom and scope, the classic example isextruded window frame sections.
12.5 Copper and alloysCopper finds its main uses in applications whereits compatibility with water, high thermal con-ductivity and high electrical conductivity areimportant, e.g. domestic water services, heating,sanitation, etc. Also, its high resistance to corro-sion combined with the pleasing colour of itsoxide film has seen much demand for roofs,cladding and flashing. Decorative schemes makeconsiderable use of the wide range and variety ofcolours available in copper alloys. In addition topure copper, two alloy families are widely used.
12.5.1 Brasses
These are alloys of copper and zinc with otheradditions to produce enhanced strength. Two
main classes are used: alpha brasses, nomin-ally 70Cu:30Zn, and alpha-beta, nominally60Cu:40Zn. Both are stronger than pure copperand 70:30 brass is extremely ductile. Neither areheat treatable, both are difficult to weld and arebetter soldered or brazed. Neither is as resistantas pure copper to corrosion. Brasses are, how-ever, generally cheaper than pure copper.
12.5.2 Bronzes
These are basically alloys of copper and tin, butwith a whole range of possible other additions toproduce alloys with specific properties, e.g. phos-phor–bronze, aluminium–bronze, silicon–bronze,etc. Gunmetal is bronze containing zinc. Many ofthe compositions are tractable to forming only bycasting, although some of the less highly alloyedbronzes are ductile enough to enable sheet metalworking. All are stronger and harder than copperand brasses, have high corrosion resistance, andmany are weldable by inert gas processes. All areexpensive.
Copper and alloys
87
88
Further reading
Properties of a wide range of engineering materials andtheir codings according to both British and Amer-ican Standards are given in:
Bolton, W. (1996) Engineering Materials Pocket Book,Newnes, Oxford.
Other information which expands on the subjectmatter of this part can be found in:
Blyth, A. (ed.) (1990) Specification, Architectural Pressand Building Publications, London.
Very straightforward and oversimplified but extremelyuseful as reference.
Brandon, D. and Kaplan, W.D. (1997) JoiningProcesses – An Introduction, John Wiley and Sons,Chichester.
Aimed at students, this includes more detailed sectionson surface science and welding than we have roomfor in this book.
British Constructional Steelwork Association (1986)Guides for Protection Against Corrosion in Steel-work, Brit Const Steelwork Ass.
Brunel, I.K., in evidence to the Royal Commissionappointed to inquire into the application of iron torailway structures. Report, HMSO, 1849.
Chilton, J.P. (1963) Principles of Metallic Corrosion,Royal Institute of Chemistry, London.
Cottrell, A.H. (1964) The Mechanical Properties ofMatter, John Wiley, London.
Particularly good for the treatment of multiaxial stress-ing and for plastic design.
Cottrell, A.H. (1967) An Introduction to Metallurgy,Edward Arnold, London.
Thorough but probably more detailed than required.Worth dipping into.
Evans, U.R. (1960) The Corrosion and Oxidation ofMetals, Edward Arnold, London.
Size and cost also make this a reference book, but it isthe classic textbook on the subject.
Honeycomb, R.W.K. (1981) Steels, Microstructure andProperties, Edward Arnold, London.
Houldcroft, P.T. (1977) Welding Process Technology,Cambridge University Press, Cambridge.
Llewellyn, D.T. (1994) Steels, Metallurgy and Applica-tions, Butterworth–Heinemann, Oxford.
Pascoe, K.J. (1978) An Introduction to the Propertiesof Engineering Materials, Van Nostrand Reinhold,New York.
Very much a beginner’s text.
Polmear, I.J. (1981) Light Alloys, Metallurgy of theLight Metals, Edward Arnold, London.
Thomas, G.G. (1970) Production Technology, OxfordUniversity Press, Oxford.
Part Three
Concrete
P.L.J. Domone
Introduction
Concrete is a ubiquitous material and its versatility,comparative cheapness and energy efficiency haveensured that it is of great and increasing import-ance for all types of construction throughout theworld. Many structures have concrete as their prin-cipal material, albeit as a composite with steel togive either reinforced or prestressed concrete, buteven in those structures where other materials suchas steel or timber form the principal structural ele-ments, concrete will normally still have an import-ant role, for example in the foundations.
In its simplest form, concrete is a mixture ofcement, water and aggregates in which thecement and water have combined to bind theaggregate particles together to form a monolithicwhole. The hardened properties are obviously ofparamount importance, and depend on a verycomplex structure. However, as we shall see, thestudy of concrete is further complicated by otherfactors, including:
• Many concretes also contain other materials,in small or large quantities, which modify theproperties, generally to advantage, but oftenwith undesirable side effects.
• The properties in the newly mixed, fresh (orfluid) state must be such that the concrete canbe transported from the mixer, handled,placed in the moulds or formwork and com-pacted satisfactorily. This requirement can bedemanding, for example with in-situ concretebeing placed in extreme weather conditions inparts of a structure with difficult access. Theresponsibility for ensuring that these opera-tions are carried out satisfactorily rests withcivil engineers; in this respect concrete is dif-ferent to most other structural materialswhich are supplied in a ready-to-use state.
• Even when hardened, the concrete’s structureand properties are not static, but continue tochange with time. For example, about 50–60per cent of the ultimate strength may bedeveloped in 7 days, 80–85 per cent in 28 days,and small but measurable increases in strengthhave been found in 30-year-old concrete.
• The concrete, and any steel contained withinit, can deteriorate for a variety of reasons, andso ensuring adequate durability as well asmechanical properties such as strength andstiffness is a major consideration.
A look at the contents list will show that all of theseissues are covered in this part of the book. We startby describing the constituent materials of concrete,and then discuss the fresh and early propertiesbefore going on to consider the hardened propertiesof deformation and strength. The principles of mixdesign, the process of selecting the relative propor-tions of the constituents to give the required proper-ties, are then presented. We then discuss somenon-destructive test methods, following whichvarious aspects of durability are then considered.Finally, we come right up-to-date by describingsome aspects of recent developments in highperformance concrete which are extending theproperties and uses of the material in exciting ways.This is a logical sequence of presentation, but not allcourses in concrete technology follow this order,and the chapters and sections within them arewritten so that they need not be read consecutively.
Historical backgroundEven though our knowledge and understandingof the material is far from complete, and researchcontinues apace, concrete has been successfully
91
used in many cultures and in many civilisations.It is not just a modern material; various forms ofit have been used for several millennia. The oldestconcrete discovered so far is in southern Israel,and dates from about 7000 BC. It was used forflooring, and consists of quicklime, made byburning limestone, mixed with water and stonewhich set into a hardened material. Mortars andconcretes made from lime, sand and gravelsdating from about 5000 BC have been found inEastern Europe, and similar mixtures were usedby the ancient Egyptians and Greeks some threeto four thousand years later. Early concretes pro-duced by the Romans were also of this type, butduring the second century BC, it was the Romanswho first made concrete with a hydraulic cement,i.e. one which reacts chemically with the mixwater, and is therefore capable of hardeningunder water and is subsequently insoluble. Thecement was a mixture of lime and volcanic ashfrom a source near Pozzuoli. This ash containedsilica and alumina in an active form which com-bined chemically with the lime; the term poz-zolana is still used to describe such materials.Concretes produced by combining this cementwith aggregates were used in many of the greatRoman structures, for example in the foundationsand columns of aqueducts, and, in combinationwith pumice, a lightweight aggregate, in thearches of the Colosseum and in the dome of thePantheon in Rome.
Lime concretes were used in some structures inthe Middle Ages and after, particularly in thickwalls of castles and other fortifications, but it wasnot until the early stages of the Industrial Revolu-tion in the second half of the eighteenth centurythat a revival of interest in the material led to anysignificant developments. In 1756, John Smeatonrequired a mortar for use in the foundations andmasonry of the Eddystone Lighthouse, and, aftermany experiments, he found that a mixture ofburnt Aberthaw blue lias, a clay-bearing lime-stone from South Wales, and an Italian pozzolanaproduced a suitable hydraulic cement.
In the 1790s, James Parker developed andpatented ‘Roman cement’ (a confusing namesince it bore little resemblance to the cement of
Roman times). This was made from nodules of acalcareous clay from North Kent, which werebroken up, burnt in a kiln or furnace, and thenground to a powder to produce the cement.Alternative sources of suitable clay were soonidentified, and production of significant quantitiescontinued until the 1860s. The cement was usedin many of the pioneering civil engineering struc-tures of the period, such as Brunel’s ThamesTunnel and the foundation of Stephenson’s Bri-tannia Bridge over the Menai Straits.
Roman cement, and some others of a similartype developed at about the same time, relied onusing a raw material which was a natural mixtureof clay and calcareous minerals. Methods of pro-ducing an ‘artificial’ cement from separate clay-and lime-bearing materials were therefore sought,resulting in the patenting by Joseph Aspdin in1824 of Portland cement. A mixture of clay andcalcined (or burnt) limestone was further calcineduntil carbon dioxide was expelled, and theproduct was then ground into the fine cementpowder. This had hydraulic cementitious proper-ties when mixed with water, and was called Port-land cement because Aspdin considered thehardened product to have a resemblance to Port-land stone. In 1828, Brunel found the hardenedmortar to be three times stronger than that madefrom Roman cement, and he used it for repairs inthe Thames Tunnel. However, Portland cementwas relatively expensive, and it did not becomewidespread in use until larger scale productionprocesses were developed. In particular, thereplacement of single-shaft kilns by continuous-process rotary kilns in the 1890s was critical.Increasingly larger capacity kilns have met theenormous world-wide demand of the twentiethcentury. A measure of the importance of Portlandcement is that it was the subject of one of the firstBritish Standards (BS 12) in 1904, subsequentlyrevised several times. Although the constituentmaterials have remained essentially the same,refinements in the production processes, inparticular higher burning temperatures and finergrinding, and a greater knowledge of cementchemistry and physics have led to steadilyincreasing quality and uniformity of the cement.
Introduction
92
For at least the last 100 years, the vast majorityof concrete has been made with Portland cement.However, as we will see in the next chapter, this isnot a single material, and there are a considerablenumber of varieties and types, with an ever moreincreasing number of international standards.
Over the last 50 years or so, there has alsobeen increasing use of other materials incorpo-rated either to replace some of the Portlandcement (cement replacement materials) or addedto enhance the fresh and/or hardened properties(admixtures). These are described in their ownchapters.
DefinitionsConcrete is a mixture of cement, water, fineaggregate (sand) and coarse aggregate (gravel orcrushed rocks) in which the cement and waterhave hardened by a chemical reaction – hydration– to form a binder for the (nearly) non-reactingaggregate. Other materials in addition to theabove are often incorporated, such as cementreplacement materials and admixture mentionedabove.
Grout is a mixture of cement and water only; itwill hydrate and gain strength, but it is rarelyused for structural purposes since it is subject tomuch higher dimensional changes than concreteunder loading or in different environments, and itis more expensive. Mortar, a mixture of cement,water and fine aggregate or sand, is more com-monly used for small volume applications, forexample in brickwork (see Chapter 30).
Mix proportionsThe aggregates form the bulk of the concretevolume, typically 70–80 per cent. Most of theremainder is the hydrated cement and waterbinder, often called the hardened cement paste(hcp). There is also a small quantity of air voids(typically 1–3 per cent of the volume) which hasnot been expelled when the concrete was placed.
The aggregate is divided at a particle size of5mm, all particles with a diameter smaller thanthis being the fine aggregate, and all particles
larger being the coarse aggregate. The maximumparticle size of the coarse aggregate can be either10, 20 or 40mm. In most concrete, the fineaggregate is somewhere between 30 and 45 percent of the total aggregate. On mixing, thevolume of water is normally in the range of50–75 per cent of the cement and therefore,assuming no air, the freshly mixed concrete com-prises, by volume:
6–16% cement12–20% water20–30% fine aggregate40–55% coarse aggregate.
We will see that nearly all of the properties of theconcrete are affected by the amounts of the con-stituents, i.e. the mix proportions. Therefore, toensure that satisfactory properties are achieved,the mix proportions must be carefully chosen andcontrolled. Measuring exact volumes of thematerials is difficult, and the weights required arenormally specified and used for concrete produc-tion; the mix proportions are therefore expressedas the weight of each material required in a unitvolume of the concrete, e.g. in kg/m3. The relativedensity of Portland cement is about 3.15, andmost aggregates used for concrete have a relativedensity of 2.55 to 2.65 (the exceptions beinglightweight and high density aggregate used formore specialised concrete). A few calculationsusing these figures and the volumes given aboveshow that the ranges of the mix proportions byweight are:
cement 150–600kg/m3
water 110–250kg/m3
aggregates (coarse�fine) 1600–2000kg/m3
The total of these for any particular mix gives, ofcourse, the concrete density, which can vary from2200–2450kg/m3. As we will see, the ratio of theweights of water to cement, normally referred tojust as the water/cement ratio, is an importantfactor influencing many of the concrete’s proper-ties. Values are typically in the range 0.3 to 1.0.
Mix proportions
93
Chapter thirteen
Portlandcements
13.1 Manufacture13.2 Physical properties13.3 Chemical composition13.4 Hydration13.5 Structure and strength of hardened cement paste13.6 Water in hcp and drying shrinkage13.7 Modifications of Portland cement13.8 Cement standards and nomenclature13.9 References
13.1 ManufactureThe crucial ingredients of Portland cement arecalcium oxide (CaO) and silicon dioxide or silica(SiO2). Both of these occur in large quantities, theformer in various forms of calcareous calciumcarbonate (CaCO3), e.g. chalk and limestone, andthe latter in a variety of mineral forms in argilla-ceous clay or shale. Cement production is a large-scale operation requiring huge quantities of theraw materials, and the plants are therefore nor-mally sited close to a suitable source of one orboth of these, which occasionally even occur in asingle source such as marl. The raw materials allcontain some other components, and in particularclays contain oxides of aluminium, iron, magne-sium, sodium and potassium. These cannot beavoided; the first two have a significant effect onthe manufacture and composition of the resultingcement, and as we shall see when discussing dura-bility, some of the others can have significanteffects even though they are present in smallquantities.
The manufacturing process is basically simple,although high temperatures are involved. Initially
the chalk and clay are reduced to particle sizes of75µm or less, and then intimately mixed in therequired proportions. The composition of themixture is critical, and it may be necessary to addsmall quantities of other materials such as groundsand or iron oxide. The mixture can be eithermixed with water to form a slurry (for the wetprocess) or dried and transported as a powder (inthe dry process). The most important part of con-verting the raw material mixture into Portlandcement takes place in a rotary kiln. This is aninclined steel cylinder lined with refractorybricks, it can be up to 200m long and 6m in dia-meter, and it is rotated about its longitudinal axiswhich is set at a slope of about 3 degrees (Figure13.1). The kiln is heated at its lower end to about1500°C by the combustion of a fuel–air mixture;the most common fuel is powdered coal, but oil,natural gas and organic waste materials are alsoused. In the wet process, the slurry is fed into thehigher end of the kiln, which is at about ambienttemperature, and it is heated as it passes downunder the combined action of rotation andgravity. The mixture takes a few hours to passthrough the kiln, but the process is continuous,and kilns are only stopped for maintenance fromtime to time. The mixture experiences increasingtemperature as it passes down the kiln, and fourphysical and chemical changes occur at progres-sively higher temperatures – drying, preheating,calcining and burning or clinkering; the basic ele-ments of each stage are included in Figure 13.1.The dry process requires less energy than the
wet process since the raw materials do not need
95
drying, and the heat transfer to the powderedmaterials is more efficient. The general arrange-ment of the plant is somewhat different, with thepreheating and some calcining being carried outin heat exchangers before the materials enter thekiln about 800°C. A dry process kiln is thereforeshorter than a wet process kiln. There are alsointermediate semi-dry and semi-wet processes,but in all processes great attention is given to effi-ciency in terms of energy use and to the environ-mental issues of exhaust gas discharge.
The heart of all of the processes is the forma-tion of the calcium silicates and aluminates andsmaller amounts of other compounds in theburning zone. The reactions take place in a semi-molten state, and the oxides of iron, aluminiumand magnesium from the clay assist by acting as aflux enabling the calcium silicates to be formed atconsiderably lower temperatures than wouldotherwise be possible. The material emerges fromthe kiln as a clinker with particle sizes of theorder of a few millimetres. This is cooled and afew per cent of gypsum (calcium sulfate dihy-drate, CaSO4.2H2O) is added before the mixtureis then ground to a fine powder. The groundclinker and gypsum are often the sole constituentsof the final Portland cement, although in many
countries the addition of up to 5 per cent of aninert filler such as limestone dust is allowed.
13.2 Physical propertiesPortland cements are fine grey powders. The par-ticles have a specific gravity of about 3.14, andmost have a size between 2 and 80�m. The parti-cle size is, of course, dependent on the grindingprocess, and it can be and is varied depending onthe requirements of the cement, as will be dis-cussed in Section 13.3. The particles are too smallfor their distribution to be measured by sieveanalysis (as used for aggregates), and instead thespecific surface (SSA), the surface area per unitweight, is normally used as an alternative mea-surement. This increases as the particle sizereduces, i.e. a higher value means smaller averageparticle size. There are a number of ways of mea-suring this, but unfortunately they all give some-what different values. It is therefore necessary todefine the method of measurement when specify-ing, quoting or using a value. The Blaine method,which is the most commonly used, is based onmeasuring the flow rate of air under a constantpressure through a small compacted sample ofthe cement. Values of SSA measured with this
Portland cements
96
Burning or clinkeringcombination of oxides to
produce calcium silicates,calcium aluminates andcalcium aluminoferrites
CalciningCaCO3 →
CaO � CO2
Preheatingdecomposition
of clayminerals
Drying
20°C
Rawmaterials
Fuel � air
Clinker250°C 650°C 950°C 1250°C 1500°C
FIGURE 13.1 The processes taking place in a Portland cement kiln in the wet process.
method range from about 300 to 500m2/kg formost cements in common use.
13.3 Chemical compositionWe have seen that Portland cement consists of amixture of compounds formed from a number ofoxides at the high temperatures in the burningzone of the kiln. For convenience, a shorthandnotation for the principal oxides present is oftenused:
CaO (lime)�CSiO2 (silica)�S
Al2O3 (alumina)�AFe2O3 (iron oxide)�F.
The four main compounds, sometimes calledphases, in the cement are:
Tricalcium silicate 3CaO.SiO2 in short C3S
Dicalcium silicate 2CaO.SiO2 in short C2S
Tricalcium aluminate 3CaO.Al2O3 in short C3A
Tetracalcium aluminoferrite 4CaO.Al2O3.Fe2O3 in short C4AF
Strictly, C4AF is not a true compound, but repre-sents the average composition of a solid solution.
Each grain of cement consists of an intimatemixture of these compounds, but it is difficult todetermine the amounts of each by direct analysis;instead the oxide proportions are determined,and the compound composition then calculatedfrom these using a set of equations developed byBogue (1955), which, in the shorthand form, are:
(C3S)�4.07(C)�7.60(S)�6.72(A)�1.43(F)�2.85(S̄) (13.1)
(C2S)�2.87(S)�0.754(C3S) (13.2)
(C3A)�2.65(A)�1.69(F) (13.3)
(C4AF)�3.04(F) (13.4)
Where S̄�SO3, and (C3S), (C2S) etc. are the per-centages by weight of the various compounds,and (C), (S) are the percentages by weight of theoxides from the oxide analysis. The value of (C)should be the total from the oxide analysis lessthe free lime, i.e. that not compounded.
The Bogue equations do not give exact values
of the compound composition, mainly becausethese do not occur in a chemically pure form, butcontain some of the minor oxides in solid solu-tion. (The impure forms of tricalcium and dical-cium silicates, C3S and C2S, are called alite andbelite respectively). However, the values obtainedare sufficiently accurate for most purposes,including consideration of the variations in thecomposition for different types of Portlandcement, and their effect on its behaviour.
The approximate range of oxide compositionthat can be expected for Portland cements isgiven in the first column of figures in Table 13.1.It can be seen that CaO and SiO2 are the prin-cipal oxides, with the ratio of CaO and SiO2 nor-mally being about three to one by weight. Thecomposition of any one cement will depend onthe quality and proportions of the raw materials,and will therefore vary from one cement works toanother, and even with time from a single works.All four cements A, B, C and D are typical indi-vidual cements, for which the compound compo-sition has been calculated with the above Bogueformulae. In each case, the principal compoundsare C3S and C2S, and these together normallyamount to about three-quarters of the cement.However, it is immediately clear that the relativeproportions of each compound vary considerably,and that these large differences are caused by verysmall variations in the oxide composition. As wewill see, such variations have considerable effectson the hydration process and hardened cementproperties, and therefore careful control of theraw materials and manufacturing processes isvital if cement of uniform quality is to be pro-duced. Cement A can be considered to have a‘typical’ or ‘average’ composition for Portlandcement; cements B, C and D are common anduseful variations of this, i.e. higher early strength,low heat and sulfate-resisting properties respec-tively, all of which are discussed in more detail inSection 13.3. (Note: the compound compositionsin Table 13.1 do not add up to 100 per cent – theremainder comprises the minor compounds,which include the gypsum added to the clinkerbefore grinding)
Chemical composition
97
13.4 Hydration
For an initial period after mixing, the fluidity orconsistency of a paste of cement and waterappears to remain relatively constant. In fact, asmall but gradual loss of fluidity occurs, whichcan be partially recovered on remixing. At a timecalled the initial set, normally between two andfour hours after mixing at normal temperatures,the mix starts to stiffen at a much faster rate.However, it still has little or no strength, andhardening, or strength gain, does not start untilafter the final set, which occurs some hours later.The rate of strength gain is rapid for the next fewdays, and continues, but at a steadily decreasingrate, for at least a few months.
The cement paste also gets warm, particularlyduring the setting and early hardening period. Inother words, the hydration reactions are exother-mic, and measurement of the rate of heat outputat constant temperature is a direct indication ofthe rate of reaction. Figure 13.2 shows a typicalvariation of this with time after mixing. Immedi-ately on mixing, there is a high but very shortpeak (A), lasting only a few minutes. This quicklydeclines to a low constant value for the so-calleddormant period, when the cement is relatively
inactive; this may last for up to two or threehours. The rate then starts to increase rapidly, ata time corresponding roughly to the initial set,and reaches a broad peak (B), some time after thefinal set. The reactions then gradually slow down,with sometimes a short spurt after one or twodays giving the further narrow peak (C)
The hydration reactions causing this behaviourinvolve all four main compounds simultaneously.The physical and chemical processes which resultin the formation of the solid products of thehardened cement paste are complex, but thefollowing simplified description, starting by con-sidering the chemical reactions of each of thecompounds individually, is nevertheless valuable.
The main contribution to the short, intense firstpeak (A) is rehydration of calcium sulphate hemi-hydrate, which arises from the decomposition ofthe gypsum in the grinding process. Gypsum isreformed:
2CS̄0.5H�3H�2CS̄H2
[H�H2O in shorthand form] (13.5)
Additional contributions to this peak come fromthe hydration of the free lime, the heat of wetting,heat of solution and the initial reactions of thealuminate phases.
Portland cements
98
TABLE 13.1 Compositions of Portland cements
Cement
Range A B C D
Oxides (% by weight)CaO 60–67 66 67 64 64SiO2 17–25 21 21 22 23Al2O3 3–8 7 5 7 4Fe2O3 0.5–6 3 3 4 5Na2O�K2O 0.2–1.3MgO 0.1–4Free CaO 0–2 1 1 1 1SO3 1–3 2 2 2 2
Compounds (% by weight)C3S 48 65 31 42C2S 24 11 40 34C3A 13 8 12 2C4AF 9 9 12 15
The behaviour of the aluminates is particularlyimportant in the early stages of hydration. In apure form, C3A reacts very violently with water,resulting in an immediate stiffening of the paste.This must be prevented, which is why the gypsumis added to the clinker. The initial reaction of thegypsum and C3A is
C3A�3CS̄H2 �26H→C3A3CS̄H32 (13.6)
The product, calcium sulfoaluminate, is alsoknown as ettringite. Although the ettringite isinsoluble and crystallises out, the reaction ismuch slower than that of the C3A alone, and sothe problems of flash set are avoided.
Usually about 5–6 per cent of gypsum byweight of the cement is added and, as this is con-sumed, the ettringite is gradually transformed tocalcium monosulfoaluminate, which has a lowersulfate content, and eventually, if all the gypsumis consumed before all the C3A, the directhydrate, C3AH6 is formed. This causes the shortthird peak C, which can occur some 2 or 3 daysafter hydration starts. Whether this peak occursat all depends on the relative amount of gypsumand C3A in the unhydrated cement, and it follows
that it tends to be a feature of high C3A contentcements.
The C4AF phase reacts over similar timescales,and the reaction also involves an intermediatecompound with the gypsum. The products havean imprecise and variable composition, butinclude high and low sulfate forms approximat-ing to C3(A.F).3CS̄.H32 and C3(A.F).CS̄.H16
respectively, i.e. similar to the C3A products. Thereactions of products contribute little of signific-ance to the overall cement behaviour.
As we have seen, the two calcium silicates C3Sand C2S form the bulk of unhydrated cement, andit is their hydration products that give the hard-ened cement most of its significant engineeringproperties such as strength and stiffness; theirreactions and reaction rates therefore dominatethe properties of the hcp (and concrete), and areextremely important.
The C3S (or, more accurately, the alite) is thefaster to react, producing a tricalcium disilicatehydrate and calcium hydroxide:
2C3S�6H→C3S2H3 �3CH (13.7)
Most of the main peak B in the heat evolution
Hydration
99
Dormant period
Rat
e of
hea
t ou
tput
, or
rate
of re
actio
n (a
rbitr
ary
units
)
Time after mixing (hours, log scale)
0.1 10 1.0 100
A
B
C
FIGURE 13.2 Typical rate of hydrating cement paste at constant temperature (Forester, 1970).
curve (Figure 13.2) results from this reaction, andit is the calcium silicate hydrate (often simplyreferred to as CBSBH) that is responsible forthe strength of the hardened cement paste (hcp).
The C2S (or, strictly, the belite) reacts muchmore slowly, but produces identical products:
2C2S�4H→C3S2H3 �CH (13.8)
This reaction contributes little heat in thetimescales of Figure 13.2, but it does make animportant contribution to the long-term strengthof hcp.
The cumulative amounts of individual productsformed over timescales a few days longer thanthose of Figure 13.2 are shown in Figure 13.3.The dominance of the CBSBH after a day or sois readily apparent; this is accompanied by anincrease in calcium hydroxide, which, togetherwith some of the minor oxides, results in the hcpbeing highly alkaline with a pH between 12.5 and13. As we will see in Chapter 23, this alkalinityhas significant influences on some aspects of thedurability of concrete construction.
The timescales and the contributions of thereactions of the individual compounds to thedevelopment of the cement strength are shown inFigure 13.4. This further emphasises the long-
Portland cements
100
FIGURE 13.3 Typical hydration product develop-ment in Portland cement paste (Soroka, 1979).
FIGURE 13.4 Development of strength of purecompounds from Portland cement (Bogue, 1955).
term nature of the strength-giving reactions of thecalcium silicates, particularly of the C2S (or, morecorrectly, the belite). In fact the reactions cannever be regarded as complete, and the extent oftheir completeness is called the degree of hydra-tion.
In common with most chemical processes,increasing temperature accelerates all of theabove reactions. With decreasing temperature,hydration will continue even below 0°C, butstops completely at about �10°C. We will be dis-cussing the effect of temperature in relation tocement and concrete strength development inChapter 17.
The physical processes during hydration andthe resulting microstructure of the hardenedcement paste are equally, if not more, importantthan the chemical reactions, and numerousstudies have been made of these by scanning,transmission and analytical electron microscopy.Figure 13.5 illustrates the hydration of a singlegrain of cement in a large volume of water. Theimportant features are:
• the processes take place at the solid–liquidinterface, with solid products being depositedin the region around a diminishing core ofunhydrated cement in each cement grain;
• the very early products form a surface layer
on the cement grain which acts as a barrier tofurther reactions during the dormant period;
• the dormant period ends when this layer isbroken down by either a build-up of internalpressure by osmosis, or by calcium hydroxidecrystals, or both, enabling hydration toproceed more rapidly;
• the hydration products (known as the gel)consist of:– an amorphous mass, mainly CBSBH, of
small, irregular fibrous particles, somesolid, some hollow and some flattened, typi-cally 0.5–2�m long and less than 0.2�mdiameter, with a very high surface area esti-mated to be of the order of 200000m2/kg,i.e. approaching a thousand times greaterthan the fresh cement grains from which ithas been formed;
– large hexagonal crystals of calcium hydrox-ide interspersed in the fibrous matrix;
• the gel contains many small gel pores, typi-cally between 0.5 and 0.5nm wide, inbetween the fibrous particles, and as hydra-tion continues, new product is depositedwithin the existing matrix, decreasing the gelporosity;
• the rate of hydration reduces over a longperiod after peak B due to the increased diffi-culty of diffusion of water through the hydra-
tion products to the unhydrated cement. It hasbeen estimated that, for this reason, completehydration is not possible for cement grains ofmore than 50�m in diameter – even aftermany years there is a residual core of theunhydrated cement;
• the products deposited near the freshcement/water interface (inner product) aredenser than those deposited further away(outer product);
• at complete hydration:– the gel porosity reaches a lower limit of
about 28 per cent;– the volume of the products of hydration is a
little more than twice that of the unhy-drated cement, but about two-thirds of thecombined initial volume of the unhydratedcement and the water which it consumes.
In reality, of course, the hydration is occurringsimultaneously in a mass of cement grains in themix water, and so the hydration productionsinteract and compete for the same space. Animportant and vital feature of hydration is that itoccurs at a (nearly) constant overall volume, i.e.the mixture does not swell or contract and thehardened cement paste or concrete is the samesize and shape when hardened as the mould inwhich was placed after mixing. Using this fact,
Hydration
101
Unhydratedcement
Hydrates(mainly C–S–H)
Calcitecrystals
Mixwater
Fresh After 1 hr After afew days
After severalhours
After afew weeks
FIGURE 13.5 Illustration of the hydration of a single cement grain.
and the measured properties of the fresh andhydrated materials1 it can be shown that:
• at a water/cement ratio of about 0.43, there isjust sufficient mix water to hydrate all thecement and fill all of the resulting gel pores.Therefore at water/cement ratios lower thanthis, full hydration can never occur unlessthere is an available external source of water,for example if the cement or concrete isimmersed in water. This is the condition ofinsufficient water, and the paste is subject toself-desiccation. In practice, in a sealed speci-men the hydration will cease somewhat beforeall of the available water is consumed, and aninitial water/cement ratio of about 0.5 isrequired for full hydration. As we will see inChapter 19, self-desiccation can also haveother effects.
• at a water/cement ratio of about 0.38, thevolume of hydration products, i.e. the gel,exactly matches that of the fresh cement andwater. At values lower than this, hydrationwill be stopped before completion, even if anexternal source of water is available. This iscalled the condition of insufficient volume. Atwater/cement ratios higher than this there isan increasing amount of unfilled spacebetween the original grains in the form of capillary pores, which are on average about ahundred times larger than gel pores within the gel itself. Calculations give the relativevolumes of unhydrated cement, gel and capil-lary pores at complete hydration shown inFigure 13.6. In reality, for the reasons dis-cussed above, hydration is never complete andtherefore the volumes in Figure 13.6 are neverachieved, but may be approached. However,at any stage of hydration, the volume of capil-lary pores will increase with water/cementratio.
Portland cements
102
1 Relative densities: unhydrated cement 3.15, gel solids 2.61,saturated gel 2.16, unsaturated gel 1.88.
Gel porosity 28 per cent.1gm of cement chemically combines with 0.23gm of water
during hydration.The analysis derives from the work of Powers in the 1950s;
a full treatment can be found in Neville (1995).
FIGURE 13.6 Volumetric composition of fully-hydrated hardened cement paste after storage in water(Hansen, 1970).
The diagrams in Figure 13.7 provide a visualillustration of this. These show an idealised struc-ture of two cement pastes with high and lowwater/cement ratios, say of the order of 0.8 and0.4 respectively, on mixing and when mature, sayafter several months. In the high water/cementratio paste the grains are initially fairly widelydispersed in the mix water and, when mature,there is still a significant capillary pore volume.On the other hand, in the low water/cement ratiopaste, the grains are initially much more closelypacked, and the hydrates occupy a greatervolume of the mature paste, which therefore hasa greater volume of capillary pores (but which, ifthe water/cement ratio is low enough, may even-tually disappear altogether).
Although it is important to distinguish betweenthe capillary and gel pores, in practice there is anear continuous distribution of pore sizes. Figure13.8 shows typical measurements which illustratethis, and also provides direct evidence of thereduction in both overall pore volume and pore
Hydration
103
High water/cement ratio
Several months old
Low water/cement ratio
Several months old
Unhydratedcement
Mix water incapillary pores
Hydrates(gel)
Fresh Fresh
FIGURE 13.7 Structure of cement pastes.
FIGURE 13.8 Pore size distribution in 28-day-old hydrated cement paste (Mehta, 1986).
size with reducing water/cement ratio for pastesof similar age, in this case 28 days.
13.5 Structure and strength ofhardened cement pasteWe have therefore seen that, at any stage ofhydration, the hardened cement paste (hcp) con-sists of:
1. a residue of unhydrated cement, at the centreof the original grains;
2. the hydrates, chiefly calcium silicates(CBSBH) but also some calcium aluminates,sulfoaluminates and ferrites, which have acomplex fibrous form containing gel pores;
3. crystals of calcium hydroxide (sometimescalled calcite);
4. the unfilled residues of the spaces between thecement grains – the capillary pores.
The strength of hcp derives from Van der Waalstype forces between the hydrate fibres. Althoughthese forces are of relatively low magnitude, theintegrated effect over the enormous surface areais considerable. The unhydrated cement is, initself, strong and is not detrimental to strength.Indeed it is even beneficial in that it is exposed if
the paste or concrete is subsequently cracked orfractured, and can therefore form new hydratesto seal the crack and restore some structuralintegrity provided, of course, some water ispresent. No other common structural materialshave this self-healing property.
For any particular cement, the compressivestrength of specimens stored at constant tempera-ture and humidity increases with age and decreas-ing water/cement ratio; Figure 13.9 shows typicalbehaviour. We should note here that the strengthcontinues to increase at water/cement ratiosbelow 0.38, even though Figure 13.6 shows thatthere is an increasing volume of unhydatredcement in the ‘end state’. This is further evidencethat unhydrated cement is not detrimental tostrength – it is the quality of the hydrates that isthe governing factor (there are, however, lowerpractical limits to water/cement ratio, which wewill discuss in Chapter 20).
We have seen that both the size and volume ofthe capillary pores are also influenced by age andwater/cement ratio (Figures 13.6 and 13.8) and itis therefore not surprising that the strength andporosity are closely linked. In simple terms – lessporosity, due to either increasing age or lowerwater/cement ratio or both, means higher
Portland cements
104
0
20
40
60
80
100
120
140
0.25 0.35 0.45 0.55
Water/cement ratio
Age
1 year
28 days
7 days
3 days
1 day
Com
pres
sive
str
engt
h (M
Pa)
FIGURE 13.9 Compressive strength development of Portland cement paste stored in water at 20°C (Domoneand Thurairatnam, 1986).
strength. The relationship between the two wasshown by Powers (1958) to be of the form
� �k.(1�P3) (13.9)
where k is a constant� �compressive strength
and P�porosity�pore volume/total pastevolume
Note that in this expression the porosity is raisedto power three, showing its great significance.
Powers’ experiments were on ‘normally’ curedpastes, i.e. kept in water at ambient temperaturesand pressures, with variations in porosityobtained by varying the water/cement ratio. Thisresulted in porosities ranging from about 25 to50 per cent. Porosities down to about 2 per centwere obtained by Roy and Gouda (1975) bycuring pastes with water/cement ratios down to0.093 at higher temperatures (up to 250°C) andpressures (up to 350MPa). Figure 13.10 showsthat, at these very low porosities, they achievedcompressive strengths of more than 600MPa,with Powers’ results being consistent with theiroverall relationship of the form
��A log (P/Pcrit) (13.10)
where A is a constant and Pcrit is a critical poros-ity giving zero strength, shown by Figure 13.10 tobe about 55 per cent.
The size of the pores rather than their totalvolume has also been shown to be an importantfactor. Birchall et al. (1981) reduced the volumeof the larger pores (greater than about 15�m dia-meter) by incorporating a polymer in pastes ofwater/cement ratios of about 0.2 and curing ini-tially under pressure. The resulting ‘macrodefect-free’ (MDF) cement had compressive strengths of200MPa and above, with flexural strengths of70MPa, a much higher fraction of compressivestrength than in ‘normal’ pastes or concrete.
Clearly, the extremes of low porosities or highstrengths cannot be achieved in concretes pro-duced on a large scale by conventional civilengineering practice, but they have possible appli-cations in factory-produced small units as analternative to extruded plastics, etc. However,
Water in hcp and drying shrinkage
105
FIGURE 13.10 The dependence of strength of hard-ened cement paste on porosity (Roy and Gouda, 1975).
results such as those shown in Figure 13.10 areuseful per se in helping to understand the behavi-our of hardened cement paste.
We will discuss concrete strength in detail inChapter 20, and in Chapter 23 we will see thatporosity is also a significant factor influencing thedurability of concrete.
13.6 Water in hcp and dryingshrinkageThe large surface areas in the gel give the hcp aconsiderable affinity for water, and make itsoverall dimensions water-sensitive, i.e. loss ofwater results in shrinkage, which is largely recov-erable on regain of water. We will discuss themagnitude of these effects and their consequencesin Chapter 19, but for the moment we will con-sider the various ways in which the water is con-tained in the paste and how its loss can lead toshrinkage. The possible sites of the water areillustrated in the diagram of the gel structureshown in Figure 13.11.
1. Water vapour. The larger voids may only bepartially filled with water, and the remainingspace will contain water vapour at a pressurein equilibrium with the relative humidity andtemperature of the surrounding environment.
2. Capillary water. This is located in the capil-lary and larger gel pores (wider than about5nm). Water in the voids larger than about
50nm can be considered as free water as it isbeyond the reach of any surface forces (seeChapter 5), and its removal does not result inany overall shrinkage; however, the water inpores smaller than about 50nm is subject tocapillary tension forces, and its removal atnormal temperatures and humidities mayresult in some shrinkage.
3. Adsorbed water. This is the water that is closeto the solid surfaces, and under the influenceof surface attractive forces. Up to five molecu-lar layers of water can be held, giving amaximum total thickness of about 1.3nm. Alarge proportion of this water can be lost ondrying to 30 per cent relative humidity, andthis loss is the main contributing factor todrying shrinkage.
4. Interlayer water. This is the water in gel poresnarrower than about 2.6nm; it follows from(3) that such water will be under the influenceof attractive forces from two surfaces, andwill therefore be more strongly held. It can beremoved only by strong drying, for example,at elevated temperatures and/or relativehumidities less than 10 per cent, but its lossresults in considerable shrinkage, the Van derWaals forces being able to pull the solid sur-faces closer together.
Portland cements
106
FIGURE 13.11 Schematic of types of water withincalcium silicate hydrate (Feldman and Sereda, 1970).
5. Chemically combined water. This is the waterthat has combined with the fresh cement inthe hydration reactions discussed in Section13.1.3. This is not lost on drying, but is onlyevolved when the paste is decomposed byheating to high temperatures in excess of900–1000°C.
The above divisions should not be thought of ashaving distinct boundaries, but the removal of thewater does become progressively more difficult.An arbitrary but often useful division is some-times made between evaporable and non-evaporable water. There are a number of ways ofdefining this, the simplest being that the evap-orable water is that lost on drying at 105°C. Thisencompasses all the water in (1) to (3) above, andsome of (4). The non-evaporable water includesthe rest of (4) and all of (5); its amount expressedas a proportion of the total water contentincreases as hydration proceeds, and this can beused to assess the progress of the hydration reac-tions.
13.7 Modifications of PortlandcementWhen discussing the properties and compositionof cements in Sections 13.2 and 13.3, we pointedout that these can be altered either by variationsin the composition of the raw material or bychanges in the manufacturing process. In thissection we will discuss ways in which the cementcan be altered from ‘average’ or ‘normal’ toobtain properties that are more useful for specificpurposes.
13.7.1 Setting, strength gain and heatoutput
The relative timescales of the dormant period,setting and strength gain govern some of the criti-cal operations in concrete practice, for examplethe transport and placing of the concrete, and thetime at which formwork can safely be removed.One way of modifying these properties is to alterthe compound composition by varying the type
and relative proportions of the raw materials usedin the cement manufacture. For example, if acement with a higher C3S and lower C2S content isproduced, as in cement B in Table 13.1, this willhave a higher rate of strength gain than cement A(but it should be noted that this does not meanrapid setting). Figure 13.12 shows that rapid hard-ening properties can also be achieved by finergrinding of the cement, which gives an increasedsurface area exposed to the mix water, and there-fore faster hydration reactions.
Since the hydration reactions are exothermic, aconsequence of the rapid hardening is a higherrate of heat output in the early stages of hydra-tion, which will increase the risk of thermalcracking in large pours from substantial tempera-ture differentials at early ages, i.e. during the firstfew days after casting. To reduce the rate of heatof hydration output a low-heat cement with alower C3S and higher C2S content may be used,i.e. as in cement C in Table 13.1. The disadvan-tage is a lower rate of gain of strength.
13.7.2 Sulfate resistance
If sulfates from external sources, such as ground-water, come into contact with the hcp, reactionscan take place with the hydration products of thecalcium aluminate phases, forming calcium sul-
foaluminate, ettringite, or, strictly, reformingsince it was also formed very early in the hydra-tion process (as described in Section 13.4). Cru-cially the reaction is expansive and can thereforelead to disruption, cracking and loss of strengthin the relatively brittle, low tensile strength hcp.(Its earlier formation did not have this effect asthe paste was still fluid, or at least soft.) The solu-tion is a low C3A content cement such as cementD in Table 13.1 which is, therefore, an exampleof a sulfate resisting cement. We will return tothis when discussing sulfate attack in more detailin Chapter 23.
13.7.3 White cement
The grey colour of most Portland cements islargely due to ferrite in the C4AF phase, whichderives from the ferrite compounds in the clay orshale used in the cement manufacture. The use ofnon-ferrite-containing material, such as chinaclay, results in a near zero C4AF content cement,which is almost pure white, and therefore attract-ive to architects for exposed finishes. Whitecement is significantly more expensive thannormal Portland cements due to the increasedcost of the raw materials, and the greater care needed during manufacture to avoid discoloration.
Modifications of Portland cement
107
0
25
50
75
0.1 1 10 100 1000Age (days)
Com
pres
sive
str
engt
h (M
Pa)
SSA (m2kg)
742
490
277
FIGURE 13.12 Effect of specific surface area on the strength gain of Portland cement concrete with awater/cement ratio of 0.4 (Bennett and Collings, 1969).
As we shall see in the next two chapters, it isalso possible to modify the concrete properties byother means, involving the use of admixturesand/or cement replacement materials.
13.8 Cement standards andnomenclatureThe first edition of the UK standard for ordinaryPortland cement was issued in 1904, since whenthere have been a further fourteen editions, thelast being in 1996. A unified European standard,BS–EN 197–1: 2000,2 has now replaced this.Each edition of BS 12 was increasingly complexand rigorous in its detailed requirements forchemical composition and properties such assetting time and strength production. The 1989edition was a move towards the European stan-dard, the two most significant changes introducedbeing:
• the cement could contain up to 5 per cent ofmaterial additional to the Portland cementclinker and the gypsum. This additionalmaterial is of similar particle size to thecement, and is often an inert filler such aslimestone powder. This had been a commonpractice in many countries before 1989
• the classifications of Ordinary PortlandCement (opc), with the composition, proper-ties and behaviour corresponding to thosedescribed as ‘normal’ throughout this chapter,and Rapid Hardening Portland Cement(rhpc), with a higher early strength gain, asmore briefly described above, were dropped infavour of a more extensive strength classifica-tion system. The strength characteristics of acement are determined by measuring the com-pressive strength of prisms made of a standardmortar with a sand:cement:water ratio of3:1:0.5 by weight, which has been mixed, castand stored under defined and carefully con-trolled conditions. A cement is then given anumber, either 32.5, 42.5, 52.5 or 62.5,
Portland cements
108
2 A list of all standards referred to in the text is included in‘Further reading’ on p. 222.
which corresponds to the strength in MPaachieved at 28 days, and a letter, either N orR (N for normal and R for rapid), dependingon the strength at either 2 or 7 days. Therequirements of the strength classes are set outin Table 13.2. The previous Ordinary Port-land Cement roughly corresponds to class42.5N, and although it is strictly not nowcorrect to use the term ‘opc’, it will take along time before it dies out.
The complexity of the subject is illustrated by thefact that BS-EN 197–1: 2000 includes five maincement types, denoted CEM I to V. CEM I isPortland cement, and CEM II to V are compositecements (i.e. blends with other cementitiousmaterials such as those described in Chapter 15),with, in all, 26 subtypes, only five of which arecurrently manufactured in the UK.
Sulfate Resisting Portland Cement has a separ-ate standard (BS 4027) for which there is noEuropean equivalent; the most significant dif-ference to other Portland cements is the require-ment for a C3A content of less than 3.5 per cent.There is no standard for white cement.
Many other countries have their own stand-ards. For example, in the USA, the AmericanSociety for Testing and Materials (ASTM) classi-fies Portland cement in their specificationC–150–94 by type number:
TABLE 13.2 BS 12: 1996 strength requirements forPortland cement
Class Compressive strength from mortar prisms (MPa)
2 days 7 days 28 days
32.5N �16�32.5 �52.5
32.5R �10
42.5N �10�42.5 �62.5
42.5R �20
52.5N �20 �52.5 �72.5
62.5N �20 �62.5
Type I is ordinary Portland cement;Type II is moderate sulfate-resistant or moderate-heat cement;Type III is a rapid-hardening cement;Type IV is low-heat cement;Type V is sulfate-resistant cement.
It is beyond the scope of this book, and veryboring for its readers, to go into further detailsabout these standards. They can be found in mostlibraries when necessary.
13.9 ReferencesBennett, E.W. and Collings, B.C. (1969) High early
strength concrete by means of very fine Portlandcement. Proc. Inst. Civil Engineers July, 1–10.
Birchall, J.D., Howard, A.J. and Kendall, K. (1981)Flexural strength and porosity of cements. Nature289, 5796, Jan., 388–90.
Bogue, R.H. (1955) Chemistry of Portland Cement,Van Nostrand Reinhold, New York.
Domone, P.L. and Thurairatnam, H. (1986) Develop-ment of mechanical properties of ordinary Portlandand Oilwell B cement grouts. Magazine of ConcreteResearch 38, 136, Sept., 129–38.
Feldman, R.F. and Sereda, P.J. (1970) A new model forhydrated Portland cement paste and its practicalimplications. Eng. J. (Canada) 53, 8/9, 53–9.
Forester, J. (1970) A conduction calorimeter for thestudy of cement hydration. Cement Technology 1, 3,May/June, 95–9.
Hansen, T.C. (1970) Physical composition of hardenedPortland cement paste. Proc. Amer. Conc. Inst. 67,404.
Mehta, P.K. (1986) Concrete: Structure, Properties andMaterials, Prentice-Hall, New Jersey, p. 450
Neville, A.M. (1995) Properties of Concrete, 4th edn,Pearson Education, Harlow, p. 844.
Powers, T.C. (1958) Structure and physical propertiesof hardened cement paste. J. Amer. Ceramic Soc. 41,1, 1–6.
Roy, D.M. and Gouda, G.R. (1975) Optimization ofstrength in cement pastes. Cement and ConcreteResearch 5, 2, March, 153–62.
Soroka, I. (1979) Portland Cement Paste and Concrete,Macmillan, London.
References
109
110
Chapter fourteen
Admixtures
14.1 Plasticisers14.2 Superplasticisers14.3 Accelerators14.4 Retarders14.5 Air entraining agents14.6 Classification of admixtures14.7 References
Admixtures are chemicals that are added to theconcrete immediately before or during mixingand significantly change its fresh, early age orhardened state to economic or physical advant-age. Only small quantities are required, typically1 to 2 per cent by weight of the cement. Theirpopularity and use have increased considerably inrecent years; estimates for the UK are that about12 per cent of all concrete produced in 1975 con-tained an admixture, and that this increased to 50per cent by 1991. In some places, notably parts ofEurope, North America, Australia and Japan, theproportion is even higher.
An extremely large number of commercialproducts are available, which are usually classi-fied or grouped according to mode of actionrather than chemical type. We shall consider thefive distinct types, namely plasticisers, superplas-ticisers, accelerators, retarders and air entrainingagents, and briefly mention others.
14.1 PlasticisersPlasticisers, also called workability aids, increasethe fluidity or workability of a cement paste orconcrete. They are long chain polymers, the maintypes being based on either lignosulfonic acids orhydroxycarboxylic acids and their salts. They arerelatively inexpensive but can contain significant
impurities. Their plasticising action is due to thesurface active nature of the component polymermolecules, which are adsorbed onto the surfaceof the cement grains with an ionic group pointingoutwards. This gives the surface a uniform negat-ive charge of the order of a few millivolts, causingmutual repulsion of the cement particles, therebybreaking up the flocs of particles formed in a neatwater/cement system, and releasing entrappedwater, as illustrated in Figure 14.1. The particlesalso become surrounded by a sheath of orientedwater molecules which prevent close approach ofthe cement grains, a phenomenon known as sterichindrance. The overall effect is one of greaterlubrication and hence increased fluidity of thepaste or concrete.
Plasticisers are also known as water-reducers,since they can produce a concrete with the sameworkability at a lower water/cement ratio, henceincreasing the strength or durability with thesame cement content. Their use has been increas-ingly widespread since their development in the1950s, and the quantitative benefits that can beobtained are discussed when considering mixdesign in Chapter 21.
Significant, and sometimes undesirable, sec-ondary effects with some plasticisers are that theyact as retarders, delaying the set and decreasingthe early strength gain, and/or that they entrainair in the form of small bubbles. They may alsocontain impurities which have other undesirableside-effects at increasing doses, and therefore themagnitude of the primary effects that can be sat-isfactorily achieved with plasticisers are relativelymodest, though nevertheless useful and costeffective.
14.2 SuperplasticisersAs the name implies superplasticisers are morepowerful than plasticisers and are used to achieveincreases in fluidity and workability of a muchgreater magnitude than those obtainable withplasticisers. They are also known as high-rangewater reducers. They were first marketed in the1960s, since when they have been continuallydeveloped and increasingly widely used. Theyhave higher molecular weights and are manufac-tured to higher standards of purity than plasticis-ers, and can therefore be used to achievesubstantially greater primary effects withoutsignificant undesirable side-effects. They are acrucial ingredient of many of the so-called ‘highperformance’ concretes, which are discussed inChapter 24.
Currently four main types can be distinguished(Edmeades and Hewlett, 1998; Ramachandran etal., 1998):
1. modified lignosulfonates (MLSs), essentiallypurified lignosulfonate plasticisers with thehigher molecular weight fractions selected togive greater efficiency;
2. sulfonated melamine formaldehyde conden-sates (SMFs), normally a sodium salt;
3. sulfonated naphthalene formaldehyde conden-sates (SNFs), again normally a sodium salt;
4. polymers containing carboxylic acid groups,e.g. polycarboxylates (PCLs). These includepolyacrylates, acrylic esters and sulfonated
polystyrenes. These have been the mostrecently developed, and are sometimes referredto as ‘new generation’ superplasticisers.
The mode of action of superplasticisers is similarto that of plasticisers, i.e. they cause a combina-tion of mutual repulsion and steric hindrancebetween the cement particles. Opinions differabout the relative magnitude and importance ofthese two effects with different superplasticisers,but the general consensus (Collepardi, 1998;Edmeades and Hewlett, 1998) is that:
• with MLSs, SMFs and SNFs, electrostaticrepulsion is the dominant mechanism;
• with PCLs, steric hindrance is equally if notmore important. This is due to a high densityof polymer side chains on the polymer back-bone which protrude from the cement particlesurface (Figure 14.2). This leads to greaterefficiency, i.e. similar increases in fluidityrequire lower admixture dosages. The term‘comb polymer’ has been used to describe thismolecular structure.
Some typical fluidity effects of admixtures of dif-ferent types, measured by spread tests on amortar, are shown in Figure 14.3. The limitedrange and effectiveness of a lignosulfonate-basedplasticiser and the greater efficiency of a PCLsuperplasticiser (in this case a polyacrylate) com-pared to an NFL-based material are apparent.
Some of the more important features of the
Superplasticisers
111
(a) Adsorption on to cementparticle surface
(b) Dispersion of particles and release ofentrapped water to give greater fluidity
FIGURE 14.1 Mode of action of plasticisers.
behaviour of superplasticisers which directlyeffect their use in concrete are:
• The behaviour of any particular combinationof superplasticiser and cement will depend onseveral factors other than the admixture type,including the cement composition, the cementfineness and the water/cement ratio (Aitcin etal., 1994)
• Substantially increased performance can beobtained if the superplasticiser is added ashort time (1–2 minutes) after the first contactof the mix water with the cement. It appearsthat if the superplasticiser is added at thesame time as the mix water, a significant
Admixtures
112
FIGURE 14.2 ‘Comb’ molecules of polycarboxylicsuperplasticisers on the surface of a cement grainleading to steric hindrance between grains.
Mortar: s/c/w 2.2/.1/0.47 Flow
Lignosulfonateplasticiser
Naphthalene formaldehydesuperplasticiser
Polyacrylatesuperplasticiser
Admixture dosage (% solids by weight of cement)
0 0.2 0.4 0.6 0.8160
180
200
220
240
Mor
tar
flow
(m
m)
FIGURE 14.3 Typical effects of plasticising andsuperplasticising admixtures on flow of mortars (Jek-navarian et al., 1997 by permission, American Con-crete Institute).
amount is incorporated into the rapidC3A/gypsum reaction, hence reducing thatavailable for workability increase. This effecthas been clearly demonstrated for MLS-,SMF- and SNF-based admixtures, but hasbeen reported as being less significant for atleast some PCLs, which are therefore moretolerant of variations in mixing procedures.
• The superplasticising action occurs for only alimited time, which may be less than thatrequired if, for example, the concrete has tobe transported by road from mixing plant tosite. Methods of overcoming this include:– blending a retarder with the superplasti-
ciser;– addition of the superplasticiser on site just
before discharge from the mixer truck;– repeated additions of small extra doses of
the admixture.The losses with some PCLs have been shownto be lower than with other types, at leastover the critical first hour after mixing.
• For any particular binder/superplasticisercombination there is a ‘saturation point’ oroptimum dosage beyond which no furtherincreases in fluidity occur (Figure 14.4). Atdosages higher than this, not only is there noincrease in fluidity, but detrimental effectssuch as segregation, excessive retardation orair entrapment during mixing which is subse-quently released, can occur.
We will comment on the quantitative benefitsthat can be obtained in Chapter 21.
14.3 AcceleratorsAn accelerator is used to increase the rate ofhardening of the cement paste, thus enhancingthe early strength, perhaps thereby allowing earlyremoval of formwork, or reducing the curingtime for concrete placed in cold weather. Theymay also reduce the setting time.
Calcium chloride (CaCl2) has historically beenvery popular as it is readily available and veryeffective. Figure 14.5 (a) shows that it acceleratesboth the initial and final set, and Figure 14.5 (b)
shows that a 2 per cent addition by weight ofcement can result in very significant early strengthincreases. This effect diminishes with time, andthe long-term strength is similar to that of non-accelerated concrete.
The CaCl2 becomes involved in the hydrationreactions involving C3A, gypsum and C4AF, butthe acceleration is caused by it acting as a catalyst
Retarders
113
Flow time � time for agiven volume of paste ormortar to flow out of a funnel
Flow
tim
e (s
ecs)
120
100
80
60
400 0.5 1 1.5 2 2.5 3
Superplasticiser dosage (% solids by weight of cement)
Saturationpoint
FIGURE 14.4 The saturation point for acement/superplasticiser combination (Aitcin, et al.,1994 by permission, American Concrete Institute).
FIGURE 14.5 Typical effects of calcium chloride admixture on (a) setting times, and (b) early strength of con-crete (Dransfield and Egan, 1988).
in the C3S and C2S reactions (Edmeades andHewlett, 1998). There is also some modificationto the structure of the CBSBH produced.
Of great significance is the increased vulnera-bility of embedded steel to corrosion due to thepresence of the chloride ions. This has led to theuse of CaCl2 being prohibited in reinforced andprestressed concrete, and to the development of anumber of alternative chloride-free accelerators,most commonly based on either calcium formate,sodium aluminate or triethanolamine. However,as with plasticisers and superplasticisers the mag-nitude of the effects of these depends on thecement composition and fineness and cannot bepredicted with certainty, and so should be estab-lished by testing. We shall discuss the corrosionof steel in concrete in some detail when consider-ing durability in Chapter 23.
14.4 RetardersRetarders delay the setting time of a mix, andexamples of use include:
1. counteracting the accelerating effect of hotweather, particularly if the concrete has to betransported over a long distance;
2. controlling the set in large pours, where con-creting may take several hours, to achieveconcurrent setting of all the concrete, henceavoiding cold joints and discontinuities, andachieving uniform strength development.
The retardations resulting from varying doses ofthree different retarding chemicals are shown inFigure 14.6. Sucrose and citric acid are veryeffective retarders, but it is difficult to controltheir effects, and lignosulfonates, often with asignificant sugar content, are preferred. Theretarding action of normal plasticisers, such assome lignosulfonates and carboxylic acids, hasalready been mentioned; most commercialretarders are based on these compounds, andtherefore have some plasticising action as well.
The mode of action of retarders involves modi-fication of the formation of the early hydrationproducts, including the calcite crystals. As withother admixtures, temperature, mix proportions,fineness and composition of the cement and timeof addition of the admixture all affect the degreeof retardation, and it is therefore difficult to gen-eralise.
Admixtures
114
FIGURE 14.6 Influence of retarders on the settingtimes of cement paste (Ramachandran et al., 1981).
FIGURE 14.7 Schematic of air entrainment bysurface-active molecules: (a) surface-active molecule;(b) stabilised air bubble (Mindess and Young, 1981).
14.5 Air entraining agentsAir entraining agents (AEAs) are organic mater-ials which, when added to the mix water, entraina controlled quantity of air in the form of micro-scopic bubbles in the cement paste component ofthe concrete. The bubble diameters are generallyin the range 0.02–0.1mm, with an averagespacing of about 0.25mm. They are sufficientlystable to be unchanged during the placing, com-paction, setting and hardening of the concrete.Entrained air should not be confused withentrapped air which is normally present as theresult of incomplete compaction of the concrete,and usually occurs in the form of larger irregularcavities.
AEAs are powerful surfactants which act at theair–water interface within the cement paste. Theirmolecules have a hydrocarbon chain or backboneterminated by a hydrophilic polar group, typi-cally of a carboxylic or sulfonic acid. Thisbecomes orientated into the aqueous phase, withthe hydrocarbon backbone pointing inwardstowards the air, thus forming stable, negativelycharged bubbles which become uniformly dis-persed. This is illustrated in Figure 14.7. Only alimited number of materials are suitable, includ-ing vinsol resins extracted from pinewood andsynthetic alkylsulfonates and alkylsulfates.
The major reason for entraining air is toprovide freeze–thaw resistance to the concrete.Moist concrete contains free water in entrappedand capillary voids, which expands on freezing,setting up disruptive internal bursting stresses.
Successive freeze–thaw cycles, say, over a winter,may lead to a progressive deterioration.Entrained air voids, uniformly dispersed through-out the hcp, provide a reservoir for the water toexpand into when it freezes, thus reducing thedisruptive stresses. Entrained air volumes of onlyabout 4–7 per cent by volume of the concrete arerequired to provide effective protection, but thebubble diameter and spacing are importantfactors. We will consider freeze–thaw damage inmore detail when discussing durability of con-crete in Chapter 23.
Air entrainment has two important secondaryeffects:
1. There is a general increase in the workabilityof the mix, the bubbles seeming to act likesmall ball-bearings. The bubbles’ size meansthat they can compensate for the lack of finematerial in a coarse sand, which would other-wise produce a concrete with poor cohesion.(Aggregate gradings will be discussed inChapter 16.)
2. The increase in porosity results in a drop instrength, by a factor of about 6 per cent foreach 1 per cent of air. This must therefore betaken into account in mix design, but theimprovement in workability means the losscan at least be partly offset by reducing thewater content and hence the water/cementratio.
Air entraining agents have little influence on thehydration reactions, at least at normal dosages,and therefore have no effect on the resulting con-crete properties other than those resulting fromthe physical presence of the voids, as describedabove.
14.6 Classification of admixturesWe said at the start of this section that admix-tures are grouped by mode of action, and as wellas those described, there are a number that havedual actions, such as acceleration or retardationtogether with either plasticising or superplasticis-ing. These can be either single materials, or, moreoften, blends of two or more materials. They arerecognised in standards in several countries, butsometimes with differing terminology; the Britishand ASTM (American) Standards classificationgiven in Table 14.1 provides a good example ofthis.
Other admixtures include pumping aids, water-proofers, anti-bacterial agents, bonding agents,viscosity agents or thickeners, foaming agents,corrosion inhibitors, wash water systems and pig-ments for producing coloured concrete; someselected texts that contain information on these,and give a more detailed treatment of the admix-tures we have described, are included in ‘Furtherreading’ on page 221.
Classification of admixtures
115
TABLE 14.1 Classification of admixtures in British and American standards
British Standards ASTM Standards
BS 5075–1: 1982 ASTM C-494Accelerating Type A: Water-reducingRetarding Type B: RetardingWater-reducing: normal, accelerating, retarding Type C: Accelerating
BS 5075–2: 1982 Type D: Water-reducing and retardingAir-entraining Type E: Water-reducing and accelerating
BS 5075–3: 1985 Type F: Water-reducing, high rangeSuperplasticising Type G: Water-reducing, high range and retarding
ASTM C-260Air-entraining admixtures
A list of relevant standards is included in ‘Further reading’ on page 222.
14.7 ReferencesAitcin, P.-C., Jolicoeur, C. and MacGregor, J.G. (1994)
Superplasticisers: how they work and why they occa-sionally don’t. Concrete International 16, 5, May,45–52.
Collepardi, M. (1998) Admixtures used to enhanceplacing characteristics of concrete. Cement and Con-crete Composites 20, 103–12.
Dransfield, J.M. and Egan, P. (1988) Accelerators. InCement Admixtures: Use and Applications (ed. P.C.Hewlett), Longman, Essex (2nd edn), pp. 102–29.
Edmeades, R.M. and Hewlett, P.C. (1998) Cementadmixtures. In Lea’s Chemistry of Cement and Con-crete (ed. P.C. Hewlett), Arnold, London, pp. 837–901.
Jeknavorian, A.A., Roberts, L.R., Jardine, L., Koyata,H. and Darwin, D.C. (1997) Condensed polyacrylicacid-aminated polyether polymers as superplasticis-ers for concrete. ACI SP – 173. Proceedings of FifthCANMET/ACI International Conference on Super-plasticisers and Other Chemical Admixtures in Con-crete. Rome, Italy (ed. V.M. Malhotra), AmericanConcrete Institute, Detroit, USA, 55–81.
Mindess, S. and Young, J.F. (1981) Concrete, Prentice-Hall, New Jersey.
Ramachandran, V.S., Feldman, R.F. and Beaudoin, J.J.(1981) Concrete Science, Heyden and Sons, London.
Ramachandran, V.S., Malhotra, V.M., Jolicoeur, C.and Spiratos, N. (1998) Superplasticisers: Propertiesand Applications in Concrete, CANMET, Ottawa, p. 404.
Admixtures
116
Chapter fifteen
Cementreplacementmaterials
15.1 Pozzolanic behaviour15.2 Types of material15.3 Chemical composition and physical properties15.4 Supply and specification
Cement replacement materials (CRMs) are usedas a substitute for some of the Portland cement ina concrete; partial cement replacement materialsis therefore a more accurate but less convenientname. Confusingly, there are also a number ofother names for this group of materials, includingsupplementary cementitious materials, cementextenders, mineral admixtures, mineral additives,latent hydraulic materials or, simply, cementi-tious materials.
Several types of materials are in common use,some of which are by-products from other indus-trial processes, and hence their use may have eco-nomic advantages. However, the main reason fortheir use is that they can give a variety of usefulenhancements of or modifications to the concreteproperties. All the materials have two commonfeatures:
• their particle size range is similar to or smallerthan that of Portland cement;
• they become involved in the hydration reac-tions.
They can be supplied either as individual mater-ials and added to the concrete at mixing, or aspre-blended mixtures with the Portland cement.
The former case allows choice of the rate of addi-tion, but means that an extra material must behandled at the batching plant; a pre-blendedmixture overcomes the handling problem but theaddition rate is fixed. Pre-blended mixtures havethe alternative names of extended cements, Port-land composite cements or blended Portlandcements. Generally, only one material is used inconjunction with the Portland cement, but thereare an increasing number of examples of the com-bined use of two or even three materials forparticular applications.
The incorporation of CRMs leads to a rethinkabout the definition and use of the water/cementratio, which is an important controlling factor formany properties of hardened cement paste andconcrete. It is generally accepted that this shouldremain as the ratio of the amount of mix water tothat of the Portland cement, and the termwater/binder ratio should be used for the ratio ofthe amount of mix water to the sum of theamounts of all of the cementitious materials, i.e.the Portland cement plus the CRMs.
15.1 Pozzolanic behaviourA common feature of nearly all CRMs is thatthey exhibit pozzolanic behaviour to a greater orlesser extent, and so we will define this before dis-cussing the individual materials. A pozzolanicmaterial is one which contains active silica (SiO2)
117
and is not cementitious in itself but will, in afinely divided form and in the presence of mois-ture, chemically react with calcium hydroxide atordinary temperatures to form cementitious com-pounds. The key to the pozzolanic behaviour isthe structure of the silica; this must be in a glassyor amorphous form with a disordered structure,which is formed in rapid cooling from a moltenstate. A uniform crystalline structure which isformed in slower cooling, such as is found insilica sand, is not chemically active.
Naturally occurring pozzolanic materials wereused in early concretes, as mentioned in the Intro-duction to this part of the book, but when a poz-zolanic material is used in conjunction with aPortland cement, the calcium hydroxide thattakes part in the pozzolanic reaction is that pro-duced from the cement hydration (see equations(13.7) and (13.8)). Further quantities of calciumsilicate hydrate are produced:
2S�3CH→C3S2H3 (15.1)
The reaction is clearly secondary to the hydrationof the Portland cement, which has lead to thename ‘latent hydraulic material’ in the list ofalternatives above. The products of the poz-zolanic reaction cannot be distinguished fromthose of the primary cement hydration, and there-fore make their own contribution to the strengthand other properties of the hardened cementpaste and concrete.
15.2 Types of materialThe main cement replacement materials in useworld-wide are:
1. pulverised fuel ash (pfa), called fly ash inseveral countries:the ash from pulverised coal used to firepower stations, collected from the exhaustgases before discharge to the atmosphere; onlyselected ashes have a suitable composition andparticle size range for use in concrete;
2. ground granulated blast furnace slag (ggbs):slag from the ‘scum’ formed in iron smeltingin a blast furnace, which is rapidly cooled in
water and ground to a similar fineness toPortland cement;
3. condensed silica fume (csf), often calledmicrosilica:extremely fine particles of silica condensedfrom the waste gases given off in the produc-tion of silicon metal;
4. calcined clay or shale:a clay or shale heated, rapidly cooled andground;
5. rice husk ash:ash from the controlled burning of rice husksafter the rice grains have been separated;
6. natural pozzolans; some volcanic ashes anddiatomaceous earth.
We should also mention fine fillers. These are alsomaterials which are ground to a similar finenessto that of the Portland cement before adding tothe cement and/or concrete, but they are chemi-cally inert, or nearly so, and so do not conform tothe definition of CRMs given above. Limestonepowder is a common example and, as we men-tioned in Chapter 13, up to 5 per cent addition ofthis to Portland cement is permitted in manycountries. Higher additions are also used incement for specific applications. There is somereaction between the calcium carbonate in thelimestone with the aluminate phases in thecement, but the main enhancement of propertiesis physical – the fine powder particles canimprove the workability and cohesiveness of thefresh paste or concrete, and can act as nucleationsites for the cement hydration products.
We will now discuss the first four of the mater-ials in the above list in more detail, usingmetakaolin (also known as HRM – high reactiv-ity metakaolin) as the example of a calcined clay.All these four are somewhat different in theircomposition and mode of action, and therefore intheir uses in concrete. Rice husk ash has simi-larities with microsilica, and natural pozzolansare not extensively used.
Cement replacement materials
118
15.3 Chemical composition andphysical propertiesTypical chemical compositions and physicalproperties are given in Table 15.1, together withtypical equivalent properties of Portland cementfor comparison. Two types of pfa are included,high- and low-lime, which result from burningdifferent types of coal. High-lime pfa is not avail-able in many countries, where only the low-limeform is marketed. It is normally safe to assumethat when pfa is referred to in text books, papersetc, it is the low-lime version unless specificallystated otherwise.
The following features arise from the table.
• All of the materials contain substantiallygreater quantities of silica than does Portlandcement, but crucially, most of this is in theactive amorphous or glassy form required forthe pozzolanic action.
• The csf is almost entirely active silica.• The alumina in the pfa, ggbs and metakaolin
are also in an active form, and becomeinvolved in the pozzolanic reactions, formingcomplex products. The metakaolin comprisesnearly all active silica and alumina.
• Two of the materials, high-lime pfa and ggbs,
also contain significant quantities of CaO.This also takes part in the hydration reac-tions, and therefore neither material is a truepozzolan, and both are to a certain extentself-cementing. The reactions are very slow inthe neat material, but they are much quickerin the presence of the cement hydration,which seems to act as a form of catalyst.
• The above considerations lead to maximumeffective Portland cement replacement levels ofabout 90 per cent for high-lime pfa and ggbs,40 per cent for low-lime pfa and metakaolinand 25 per cent for csf. At higher levels thanthese, there is insufficient Portland cement toproduce the required quantities of calciumhydroxide for the secondary reactions.However, as we will see, levels less than thesemaxima are normally used, to good effect.
• Pfa and ggbs have particle sizes similar tothose of the Portland cement, whereas themetakaolin particles are, on average, nearlyten times smaller and the csf particles 100times smaller (although the ggbs andmetakaolin are both ground specifically foruse in concrete, and so their fineness can bevaried). The consequences of the associateddifferences in surface area are:
Chemical composition and physical properties
119
TABLE 15.1 Typical composition and properties of cement replacement materials
pfa ggbs csf metakaolin Portland cement
Low lime High lime(class F) (class C)
Oxides (% by wt)SiO2 48 40 36 97 52 20CaO 3 20 40 �1 �1 64Al2O3 27 18 9 �1 41 5Fe2O3 9 8 1 �1 5 4MgO 2 4 11 �1 �1 1
Pozzolan reactivity (mg CH/gm) 875 400 427 1050Particle size range (microns) 1–100 3–100 0.03–0.3 0.2–15 0.5–100Specific surface area (m2/kg) 350 400 20000 12000 350Particle relative density 2.3 2.9 2.2 2.5 3.15Particle shape spherical irregular spherical irregular angular
Materials with these properties will conform to the relevant standards for use in concrete, where these exist. A list of these isincluded in ‘Further reading’ on page 222.
– The rate of reaction of the metakaolin ishigher than that of pfa and ggbs, and thatof csf highest of all (but remember that allare still secondary to that of the Portlandcement).
– Both metakaolin and csf result in a loss offluidity of the cement paste and concrete ifno other changes are made to the mix, withagain the effect of csf being greater thanthat of metakaolin. To maintain fluidity,either the water content must be increased,or a plasticiser or superplasticiser added(see admixtures above). The latter is thepreferred option, since other propertiessuch as strength are not compromised. Witha sufficient dosage of superplasticiser to dis-perse the fine particles, a combination ofexcellent workability with good cohesionand low bleed can be obtained.
• The spherical shape of the pfa particles leadsto an increase in fluidity if no other changesare made to the mix. Some increase is alsoobtained with ggbs.
• All the materials have lower specific gravities
than Portland cement, and therefore substitu-tion of the cement on a weight-for-weightbasis will result in a greater volume of paste.
All the above effects have led to an increasing useof the various CRMs in all types of concrete inthe latter part of the twentieth century, and wewill discuss their effect on several aspects of con-crete behaviour in the subsequent chapters.
15.4 Supply and specificationAs we said earlier, CRMs can be supplied asseparate materials or preblended with Portlandcement. Blends with ggbs are called PortlandBlast Furnace cement, and blends with pfa Port-land Pozzolanic cement. There are a bewilderingarray of relevant standards throughout the world,made all the more complicated by the recentemergence of unified European standards. Aswith Portland cements, it is beyond the scope ofthis book to list and discuss these. Standards forthe individual materials are listed in ‘Furtherreading’ on page 222.
Cement replacement materials
120
Chapter sixteen
Aggregates forconcrete
16.1 Types of aggregate16.2 Aggregate classification: shape and size16.3 Other properties of aggregates16.4 Reference
In Chapter 13 we saw that hardened cementpaste (hcp) has strength and other properties thatcould make it suitable for use as a constructionmaterial in its own right, but it suffers from twomain drawbacks – high dimensional changes, inparticular low modulus, high creep and shrink-age, and cost. Both of these disadvantages areovercome, or at least modified, by adding aggre-gates to the cement paste, thus producing con-crete. The objective is to use as much aggregate aspossible, binding the particles together with thehcp. This means that the largest possible aggre-gate size should be used, with a continuous rangeof particle sizes from the fine sand up to thecoarse stones; this minimises the void content ofthe aggregate mixture and therefore the amountof hcp required, and helps the fresh concrete toflow more easily. Normally the aggregates occupyabout 65–80 per cent of the total concretevolume.
With one or two notable exceptions, the aggre-gates can be thought of as being inert fillers; forexample, they do not hydrate, and they do notswell or shrink. They are distributed throughoutthe hcp, and it is sometimes useful to regard con-crete as a two-phase material of either coarseaggregate dispersed in a mortar matrix, or coarseand fine aggregate dispersed in an hcp matrix.
Models based on this two-phase material are ofvalue in describing deformation behaviour, as dis-cussed in Chapter 15, but, when cracking andstrength are being considered, a three-phasemodel of aggregate, hcp and the transition orinterfacial zone between the two (about30–50�m wide) is required, since the transitionzone is often the weakest phase, and cracking ini-tiates within it. This will be discussed in moredetail when considering concrete strength inChapter 20. In this section we shall first describethe various types of most commonly used aggre-gates and how they are classified, and then con-sider some of their most important propertieswhen used in concrete.
16.1 Types of aggregateAggregates can be obtained either from naturalsources, such as gravel deposits and crushedrocks, or specifically manufactured for use in con-crete. It is convenient to group them in terms oftheir density.
16.1.1 Normal density aggregates
Many different natural materials are used formaking concrete, including gravels, igneous rockssuch as basalt and granite and the stronger sedi-mentary rocks such as limestone and sandstone.They should have sufficient integrity to maintaintheir shape during the concrete mixing, and besufficiently strong to withstand the stresses
121
imposed on the concrete. The latter becomes aparticular consideration with high-strength con-crete (Chapter 24). The mineral constituents arenot generally of great importance, the notableexceptions being those that can participate inalkali–silica reaction and thaumasite damage,both of which will be discussed in Chapter 23.
All of the above rock types have relative densitieswithin a limited range of approximately 2.55–2.75,and therefore all produce concretes with similardensities, normally in the range 2250–2450kg/m3,depending on the mix proportions.
Gravels from suitable deposits in river valleysor shallow coastal waters have particles which,for the most part, are of a suitable size for directuse in concrete, and therefore only requirewashing and grading, i.e. subdividing into varioussizes, before use. Bulk rock sources, e.g. granite,require crushing to produce suitable size material.The particles are therefore sharp and angular, dis-tinctly different from the naturally rounded parti-cles in a gravel; we will see later that particleshape has a significant effect on fresh and hardened concrete properties.
16.1.2 Lightweight aggregates
Lightweight aggregates are used to produce lowerdensity concretes, which are advantageous inreducing the self-weight of structures and alsohave better thermal insulation than normalweight concrete. The reduced relative density isobtained from air voids within the aggregate par-ticles. Pumice, a naturally occurring volcanic rockof low density, has been used since Roman times,but it is only available at a few locations, andartificial lightweight aggregates are now widelyavailable. They are of three main types:
1. sintered pulverised fuel ash, formed byheating the pelletised ash from pulverised coalused in power stations until partial fusion andhence binding occurs;
2. expanded clay or shale, formed by heatingsuitable sources of clay or shale until gas isgiven off and trapped in the semi-moltenmass;
Aggregates for concrete
122
FIGURE 16.1 Typical strength versus density rela-tionships for concretes containing selected lightweightaggregates (from aggregate supplier’s information).
3. foamed slag, formed by directing jets ofwater, steam and compressed air on to themolten slag from blast furnaces.
In each case, both fine and coarse aggregates canbe produced, and many different products areavailable, particularly in industrialised countries.Because they all achieve lower specific gravity byincreased porosity, they are all weaker than thenormal density aggregates, and their use results inan overall lowering in the concrete strength – thepenalty to be paid for the lower density. Thequality and properties of different aggregates varyconsiderably, and therefore produce differentstrength/density relationships, illustrated inFigure 16.1.
Lightweight aggregates are not as rigid asnormal weight aggregates, and therefore produceconcrete with a lower elastic modulus and highercreep and shrinkage. As with strength, the pro-perties depend on the lightweight aggregate typeand source, and also whether lightweight fines ornatural sands are used.
16.1.3 Heavyweight aggregates
Where concrete of high density is required, forexample in radiation shielding, heavyweight
aggregates can be used. Densities of 3500 to4500kg/m3 are obtained by using barytes (abarium sulphate ore), and about 7000kg/m3 byusing steel shot.
16.2 Aggregate classification: shapeand sizeWithin each of the types described above, aggre-gates are classified principally by shape and parti-cle size.
Normal-density aggregates in particular maycontain a range of particle shapes, from wellrounded to angular, but it is usually consideredsufficient to classify the aggregate as uncrushed,i.e. coming from a natural gravel deposit, withmost particles rounded or irregular, or crushed,i.e. coming from a bulk source, with all particlessharp and angular.
The principal size division is that between fineand coarse aggregate at a particle size of 5mm(although some countries use 4, 6 or 8mm). Thecoarse aggregate can have a maximum size of 10,14, 20 or 40mm (although, again some countriesuse different values). The distribution of particlesizes within each of these major divisions is alsoimportant both for classification and for deter-mining the optimum combination for a particularmix (a part of the mix design process, discussedin Chapter 21). To measure this, a sieve analysisis carried out using a series of standard sieveswith apertures ranging from 75�m to 37.5mm,each sieve having approximately twice the aper-ture size of the previous one, i.e. in a logarithmicprogression, with the exception of a 14mm sieve.
The analysis starts with drying and weighing arepresentative sample of the aggregate, and thenpassing this through a stack or nest of the sieves,starting with that with the largest aperture. Theweights of aggregate retained on each sieve arethen measured. These are converted first to per-centage retained and then to cumulative, i.e.total, per cent passing, which are then plottedagainst the sieve size to give a grading curve.
Standards for aggregate for use in concretecontain limits inside which the grading curves forcoarse and fine aggregate must fall. In the UK
Aggregate classification: shape and size
123
standard (BS 8821), the coarse aggregate can beeither single size where all (or, strictly, nearly all)of the particles are within two successive sizes,e.g. 5 to 10mm, 10 to 20mm or 20 to 40mm, orgraded, where the smallest size is 5mm in eachcase, e.g. 5 to 14mm, 5 to 20mm or 5 to 40mm(10mm graded material is obviously the same as10mm single size). The range for suitable fineaggregate is wide, and the UK standard subdi-vides this into three overlapping divisions of fine,medium and coarse fine aggregate. The variouslimits are given in Table 16.1, and the gradingcurves for the mid-points of the ranges for eachaggregate division are plotted in Figure 16.2.
A single number, the fineness modulus, issometimes also calculated from the sieve analysisresults. The cumulative per cent passing figuresare converted to cumulative per cent retained,and the fineness modulus is defined as the sum ofall of these starting with that for the 150�msieve, divided by 100. A higher fineness modulusindicates a coarser material; the values for thegrading curves in Figure 16.2 are given in Table16.2. It is important to remember that the defini-tion applies to the standard sieve sizes only (withthe 14mm sieve excluded), and that for coarseaggregate with all particles larger than, say, 5mmthe cumulative per cent retained on all sievessmaller than 5mm should be entered as 100.
During the process of mix design, the individualsubdivisions or fractions of aggregate are combinedin proportions to give a suitable overall grading forgood workability and stability. This should be con-tinuous and uniform. Examples for maximumcoarse aggregates sizes of 10, 20 and 40mm thatare produced by the mix design process to be dis-cussed in Chapter 21 are shown in Figure 16.3.These result from using aggregates with ideal grad-ings, and in practice it is normally not possible toachieve these exactly, but they are good ‘targets’.
Sieve analysis and grading curves take noaccount of particle shape, but this does influencethe voids content of the aggregate sample – more
1 A list of all standards referred to in the text is included in‘Further reading’ on page 222.
TA
BL
E 1
6.1
Bri
tish
Sta
ndar
d gr
adin
g re
quir
emen
ts f
or a
ggre
gate
s fo
r co
ncre
te (
BS
882:
199
2)
Siev
e
Coa
rse
aggr
egat
eFi
ne a
ggre
gate
size
Sing
le s
ize
Gra
ded
Ove
rall
Add
itio
nal l
imit
s
(mm
)40
mm
20m
m14
mm
10m
m40
–5m
m20
–5m
m14
–5m
mlim
its
Coa
rse
Med
ium
Fine
Per
cen
t by
wei
ght
pass
ing
37.5
85–1
0010
0–
–90
–100
100
–20
0–25
85–1
0010
0–
35–7
090
–100
100
14–
0–70
85–1
0010
025
–55
40–8
090
–100
100–
50–
250–
5085
–100
10–4
030
–60
50–8
510
05
–0–
50–
100–
250–
50–
100–
1089
–100
2.36
––
0–5
––
–60
–100
60–1
0065
–100
80–1
001.
18–
30–1
0030
–90
45–1
0070
–100
0.6
15–1
0015
–54
25–8
055
–100
0.3
5–70
5–40
5–48
5–70
0.15
0–15
Aggregate classification: shape and size
125
0
10
20
30
40
50
60
70
80
90
100
0.15 0.3 0.6 1.18 2.36 5 10 14 20 37.5Sieve size (mm)
Fine aggregateoverall rangecoarsemediumfine
Coarseaggregate
single size:10 mm14 mm20 mm40 mmgraded:14–5 mm20–5 mm40–5 mm
Per
cent
pas
sing
by
wei
ght
FIGURE 16.2 Grading curves of aggregates at mid-range of BS 882 limits.
Per
cent
pas
sing
by
wei
ght
0
10
20
30
40
50
60
70
80
90
100
Sieve size (mm)
Max. aggregate size:10 mm20 mm40 mm
0.15 0.3 0.6 1.18 2.36 5 10 20 37.5
FIGURE 16.3 Examples of good combined aggregate grading curves for use in concrete.
rounded particles will pack more efficiently andwill therefore have a lower voids content.According to Dewar (1983) it is sufficient to useonly three numbers to characterise an aggregatefor mix design purposes – specific gravity (or par-ticle relative density), mean particle size and voidscontent in the loosely compacted state.
We should also mention here the bulk density.This is the weight of aggregate occupying a unit
overall volume of both the particles and the airvoids between them. It is measured by weighing acontainer of known volume filled with aggregate.The value will clearly depend on the gradingwhich will govern how well the particles fittogether, and also on how well the aggregate iscompacted. Unlike the particle relative density,which is more useful, it is not therefore a con-stant for any particular aggregate type.
16.3 Other properties of aggregatesOther properties of aggregates that effect theiruse in concrete include porosity and absorption,elasticity, strength and surface characteristics.
16.3.1 Porosity and absorption
We have mentioned the high porosity of light-weight aggregates, but normal weight aggregatesalso contain some pores (typically 1–3 per cent byvolume) which can absorb and hold water. Beforeconcrete mixing, the aggregate can therefore be inone of the four moisture conditions shown inFigure 16.4. In the fresh concrete, aggregate thatis in either of conditions (1) or (2) will absorbsome of the mix water, and aggregate in con-dition (4) will contribute water to it. Condition(3), saturated surface dry, is perhaps most desir-able, but is difficult to achieve except in the labo-ratory. Of prime importance to the subsequentconcrete properties is the amount of water avail-able for cement hydration, i.e. the amount that is
non-absorbed or ‘free’; therefore, to ensure thatthe required free water/cement ratio is obtained,it is necessary to allow for the aggregate moisturecondition when calculating the amount of mixwater. If the aggregate is drier than saturatedsurface dry, extra water must be added; if it iswetter, then less mix water is required.
16.3.2 Elastic properties and strength
Since the aggregate occupies most of the concretevolume, its elastic properties have a major influ-ence on the elastic properties of the concrete, aswe shall discuss in Chapter 19. The reduction instrength of the concrete resulting from the use ofthe porous, lower strength lightweight aggregatehas been discussed above. Normal weight aggre-gates are generally considerably stronger than thehcp, and therefore do not have a major influenceon the strength of most concretes. However, inhigh-strength concrete (with strengths in excessof, say, 80MPa – see Chapter 24) more carefulaggregate selection is important.
Aggregates for concrete
126
TABLE 16.2 Fineness modulus (fm) values for aggregates with the gradings in Figure 16.2
Coarse aggregate Fine aggregate
Single size fm Graded fm Grading fm
40mm 7.93 40–5mm 7.25 coarse 3.0320mm 6.93 20–5mm 6.55 medium 2.6514mm 6.70 10–5mm 6.28 fine 2.1010mm 5.93
1. Completely dry,all pores empty
4. Wet,excess water
2. Air dry,pores partially filled
3. Saturated and surface dry,
all pores full but noexcess water
FIGURE 16.4 Possible moisture conditions of aggregate.
16.3.3 Surface characteristics
The surface texture of the aggregate depends on themineral constituents and the degree to which theparticles have been polished or abraded. It seems tohave a greater influence on the flexural strengththan on the compressive strength of the concrete,probably because a rougher texture results in abetter adhesion to the hardened cement paste. Thisadhesion is also greatly affected by the cleanlinessof the surface – which must therefore not be conta-
minated by mud, clay or other similar materials.The interface or transition zone between theaggregate surface and the hardened cement pastehas a major influence on the properties of the con-crete, particularly strength, and is discussed in somedetail in Chapter 20.
16.4 ReferenceDewar J.D. (1999) Computer Modelling of Concrete
Mixtures, E & FN Spon, London, p. 272.
Reference
127
128
Chapter seventeen
Properties offresh concrete
17.1 General behaviour17.2 Measurement of workability17.3 Factors affecting workability17.4 Loss of workability17.5 References
Civil engineers are responsible for the production,transport, placing, compacting and curing of freshconcrete. Without adequate attention to all of thesethe potential hardened properties of the mix, such asstrength and durability, will not be achieved in thefinished structural element. It is important to recog-nise that it is not sufficient simply to ensure that theconcrete is placed correctly; the behaviour and treat-ment of the concrete during the period beforesetting, typically some six to ten hours after casting,and during the first few days of hardening, have asignificant effect on the long-term performance.
It is beyond the scope of this book to discuss theoperations and equipment used to batch, mix,handle, compact and finish the concrete. The aimof these practices is to produce a homogeneousstructure with minimum air voids as efficiently aspossible; it is also necessary to ensure that the con-crete is then stable and achieves its full, matureproperties. We therefore need to consider theproperties when freshly mixed, between placingand setting, and during the early stages of hydra-tion. We will discuss the former in this chapter,and the latter two in the next chapter. We will seethat much of the technology is based on empiri-cism, but that an understanding can be achievedby applying the principles of rheology – the scienceof the deformation and flow of materials.
17.1 General behaviourExperience in mixing, handling and placing freshconcrete quickly gives concrete workers (and stu-dents) a subjective understanding of the behavi-our, and an ability to recognise ‘good’ and ‘bad’concrete. A major problem is that a wide varietyof subjective terms are used to describe the con-crete, e.g. harsh, cohesive, lean, stiff, rich, whichcan mean different things to different people, anddo not quantify the behaviour in any way.However, the main properties of interest can bedescribed as follows:
1. Fluidity or consistency. It must be capable ofbeing handled and of flowing into the form-work and around any reinforcement, with theassistance of whatever equipment is available.For example, concrete for a lightly reinforced,shallow floor slab need not be as fluid as thatfor a narrow, heavily reinforced column.
2. Compactability. All, or nearly all, of the airentrapped during mixing and handling shouldbe capable of being removed by the compact-ing system being used, such as poker vibra-tors.
3. Stability or cohesiveness. The concrete shouldremain as a homogeneous uniform mass. Forexample, the mortar should not be so fluidthat it segregates from the coarse aggregate.
The first two of these properties, fluidity andcompactability, are generally combined into theproperty called workability. Although this mightseem a fairly obvious property, engineers and
concrete technologists have struggled since con-crete construction became popular early in thelast century to produce an adequate definition.Two fairly recent examples illustrate the diffi-culty. These are:
• ‘that property of freshly mixed concrete ormortar which determines the ease and homo-geneity with which it can be mixed, placed,consolidated and finished’ (ACI Committee,116, 1990).
• ‘that property determining the effort requiredto manipulate a freshly mixed quantity ofconcrete with minimum loss of homogeneity’(American Society for Testing and Materials,1993).
These relate to the requirements in very generalterms only, but the biggest problem is that neithermakes any reference to a quantitative measurableproperty, which engineers need and have for mostother properties, e.g. elastic modulus, fracturetoughness, etc. As we will see, measurement ofworkability is by no means straightforward.
In general, higher workability concretes(however defined or measured) are easier to placeand handle, but if this high workability isobtained, for example, by an increased watercontent, then a lower strength and/or durabilitywill result. Plasticisers and superplasticisers(Chapter 14) will help, but in most cases it isnormal practice to use the lowest workabilityconsistent with efficient handling and placing inthe particular application, thereby increasing theneed for a proper understanding of the freshproperties and the factors that effect them.(Achieving the balance between workability andstrength is part of the mix design process, whichwe will be discussing in Chapter 21.)
As mentioned in the Introduction, most of theconcrete (about 65–80 per cent of the volume)consists of fine and coarse aggregate. Theremainder is cement paste, which in turn con-sists of 30–50 per cent by volume of cement, therest being water. Cement paste, mortar and con-crete are all therefore concentrated suspensionsof particles of varying sizes, but all considerablydenser than the mix water. Surface-attractive
forces are significant in relation to gravitationalforces for the cement particles, but less so forthe aggregate particles, and the main resistanceto flow comes from interference and frictionbetween them. The behaviour is therefore farfrom simple.
17.2 Measurement of workability
17.2.1 Rheological properties
Rigorous definition of the flow behaviour of anyfluid requires the application of rheological prin-ciples, and in particular the measurement of therelationship between shear stress and rate ofshear strain (or shear rate), normally called theflow curve. This is normally carried out in arheometer or viscometer of some sort. There areseveral types of these, perhaps the most commonbeing a concentric cylinder viscometer in which aninner cylinder or bob is rotated in an outer cylin-der or cup of the fluid. Any respectable under-graduate textbook in fluid mechanics will describesuch instruments. Several such tests have beendeveloped for concrete, involving either a mixingor shearing action, and which are of sufficient sizeto cope with coarse aggregate particles of up to20mm (RILEM, 2000). There is general agree-ment that the behaviour of fresh paste, mortarand concrete all approximate reasonably closelyto the Bingham model illustrated in Figure 17.1.Flow only starts when the applied shear stressreaches a yield stress (�y) sufficient to overcomethe interparticle interference effects, and at higherstresses the shear rate varies approximately lin-early with shear stress, the slope defining theplastic viscosity (�). Thus the two Bingham con-stants, �y and �, are required to define the behavi-our, unlike the simpler and very common case of aNewtonian fluid which does not have a yieldstress, and which therefore requires only a singleconstant, viscosity (Chapter 4). Because at leasttwo data points are required to define the flowcurve, the first satisfactory test that was devised tomeasure this on concrete was called the two- pointworkability test (Tattersall and Banfill, 1983;Tattersall, 1991; Domone et al., 1999).
Measurement of workability
129
17.2.2 Single-point tests
Measurement of the two Bingham constants forconcrete is a relatively recent development andinvolves the use of sophisticated and expensiveapparatus. A large number of simpler and moreconvenient tests have been devised over manyyears, some only being used by their inventors.These all give only one number or test value, andcan therefore be called single point tests; theyclearly give a limited, but neverthless useful,description of the concrete. Four have beenadopted as ‘standard tests’ in many countries,and we will now consider these in some detail.1
The simplest, and crudest, is the slump test(Figure 17.2). The concrete is placed in thefrustum of a steel cone and hand tamped in threelayers. The cone is lifted off, and the slump isdefined as the downward movement of the con-crete. A true slump in which the concrete retainsthe overall shape of the cone and does not col-lapse is preferred, which gives a limit to theslump measurement of about 180mm. A shearslump invalidates the test, and may indicate a mixprone to segregation due to lack of cement orpaste. A collapsed slump is not ideal, but therehas been a trend to the increasing use of super-plasticised high workability mixes which producecollapsed slumps with little or no segregation,and slump values up to, and even above, 250mm
Properties of fresh concrete
130
Bingham behaviour: � � �y � �.•
�y = yield stress� = plastic viscosity
She
ar s
tres
s (�
)
tan�1�
Rate of shear strain (•)
�y
FIGURE 17.1 Flow curve of fresh concrete and thedefinitions of yield stress and plastic viscosity.
1 A list of relevant standards is included in ‘Further reading’on page 222.
are regularly measured and quoted. For such veryhigh workability mixes an alternative is tomeasure the final diameter or ‘flow’ of the con-crete, which is more sensitive to changes in themix than the change in height. Indeed, for self-compacting mixes (see Chapter 24) the slump-flow test is carried out without any initialcompaction when filling the cone.
As a general guide, mixes with slumps rangingfrom about 10mm upwards can be handled withconventional site equipment, with higher slumps(100mm and above) being required to ensure fullcompaction of the concrete in areas with limitedaccess or congested reinforcement. However,some zero slump mixes have sufficient workabil-ity for some applications. A recent Europeanstandard (ENV 206) classifies slump in fourclasses:
S1: slump�10–40mmS2: slump�50–90mmS3: slump�100–150mmS4: slump�160mm
and recommends using a mix in class S3 unlessthere is particular reason otherwise.
The compacting factor test (Figure 17.3) is ableto distinguish between low slump mixes. Con-crete is placed in an upper hopper, dropped into alower hopper to bring it to a standard state, andthen allowed to fall into a cylinder. The resultingdegree of compaction of the concrete is thenmeasured by comparing its weight with theweight of concrete in the cylinder when fullycompacted. The compacting factor is defined asthe former divided by the latter; values in therange 0.7 to 1 are obtained. The closer the valueapproaches 1, the higher the workability.
In the Vebe test (Figure 17.4), the response ofthe concrete to vibration is determined. The Vebetime is defined as the time taken to completelyremould a sample of concrete from a slump testcarried out in a cylindrical container. Standardvibration is applied, and remoulding times from 1to about 25secs are obtained, with higher valuesindicating decreased workability. It is often diffi-cult to define the end point of complete remould-ing with a sufficient degree of accuracy.
To differentiate between high workabilitymixes, the flow table test has been devised (Figure17.5). It is essentially a slump test with a lowervolume of concrete in which, after lifting thecone, some extra work is done on the concrete bylifting and dropping one edge of the board (ortable) on which the test is carried out. A flow orspread of 400mm indicates medium workability,and 500mm high workability.
Apart from only giving a single test value, thesefour tests (or five if we consider the slump–flowtest to be distinct from the slump test) all measurethe response of the concrete to specific, but arbi-trary and different, test conditions. The slump,slump–flow, Vebe and flow table tests provide ameasure of the consistency or mobility of the con-crete; the slump test after a standard amount ofcompaction work has been done on the concrete,
the slump–flow test after the minimal amount ofwork of pouring into the cone, the Vebe testduring a standard energy input, and the flowtable with a combination of compaction workand energy input. The compacting factor testcomes closest to assessing the compactability ofthe concrete, but the amount of work done on theconcrete in falling into the cylinder is much lessthan the energy input from practical compactionequipment such as a poker vibrator.
There is some degree of correlation betweenthe results of these tests, as illustrated in Figure17.6, but as each of the tests measures theresponse to different conditions the correlation isquite broad. It is even possible for the results tobe conflicting – e.g. for, say, the slump test torank Mix A to be more workable than Mix B,and for the Vebe test to give the opposite
Measurement of workability
131
Types of slump
1. The cone is filled withconcrete in three equallayers, and each layer iscompacted with twenty-fivetamps of the tamping rod.
2. The cone is slowlyraised and theconcrete is allowed toslump under its ownweight.
3. The slump is measuredto the nearest 5 (orsometimes 10) mm usingthe upturned cone andslump rod as a guide
200
100
300
True Shear Collapse
Slump
FIGURE 17.2 The slump test.
Properties of fresh concrete
132
1. Concrete is loaded into the upper hopper.2. The trap door is opened, and the concrete falls into the lower hopper.3. The trap door is opened, and the concrete falls into the cylinder.4. The concrete is struck off level with the top of the cylinder.5. The cylinder + concete is weighed, to give the partially compacted
weight of concrete.6. The concrete is fully compacted, and extra concrete added to fill the
cylinder.7. The cylinder + concrete is weighed, to give the fully compacted weight of
concrete.
Compacting factor � weight of partially compact concreteweight of fully compact concrete
Approx. 1 metre
Upperhopper
Lowerhopper
300 � 150 mm �cylinder
FIGURE 17.3 The compacting factor test.
300 mm
1. A slump test is performed in acontainer.
2. A clear perspex disc, free to movevertically, is lowered onto the concretesurface.
3. Vibration at a standard rate is applied.
Vibration
Clear perspexplate
Vebe degrees is the time(in seconds) to completecovering of the underside ofthe disc with concrete.
FIGURE 17.4 The Vebe test.
ranking. The result therefore depends on thechoice of test, which is far from satisfactory.
The slump and slump–flow tests clearly involvevery low shear rates, and therefore, not surpris-ingly, reasonable correlations are obtained withyield stress. No correlation is obtained withplastic viscosity.
Despite there limitations, single point tests,particularly the slump test and, to a somewhatlesser extent, the flow table and slump–flow tests,are popular and in regular use. Perhaps the mainreason for this is their simplicity and ease of useboth in the laboratory and on site, but specifiersand users must be aware of the potential pitfallsof over-reliance on the results.
17.3 Factors affecting workabilityLower values of yield stress (�y) and plastic viscos-ity (�) indicate a more fluid mix; in particular,reducing �y lowers the resistance to flow at lowshear stresses, e.g. under self-weight when beingpoured, and reducing � results in less cohesive orless ‘sticky’ mixes and increased response duringcompaction by vibration, when the localised shearrates can be high. Some of the more important
effects of variation of mix proportions and con-stituents on �y and �, shown schematically inFigure 17.7, are as follows:
• Increasing the water content whilst keepingthe proportions of the other constituents con-stant decreases �y and � in approximatelysimilar proportions.
• Adding a plasticiser or superplasticiserdecreases �y but leaves � relatively constant.In essence, the admixtures allow the particlesto flow more easily but in the same volume ofwater. The effect is more marked with super-plasticisers, which can even increase �, andcan therefore be used to give greatly increasedflow properties under self-weight, whilstmaintaining the cohesion of the mix. This isthe basis for whole range of high workabilityor flowing concretes.
• Increasing the paste content will normallyincrease � and decrease �y, i.e. the mix maystart to flow more easily but will be stickier,and vice versa.
• Replacing some of the cement with pfa orggbs will generally decrease �y, but may eitherincrease or decrease �, depending on the
Factors affecting workability
133
‘Flow’
Hinge
40
200
130
200
700
1. A conical mould (smaller than that of the standard slump test) is usedto produce a sample of concrete in the centre of a 700 mm squareboard, hinged along one edge.
2. The free edge of the board is lifted against the stop and dropped 15 times.
Flow � final diameter of the concrete(mean of two measurements at right angles)
Stop
Dimensions in mm
FIGURE 17.5 The flow table test.
Vebe
(se
cs)
30
25
20
15
10
5
00 50 100 150 200
Slump (mm)
Com
pact
ing
fact
or
1
0.9
0.8
0.7
0.60 50 100 150 200
Slump (mm)
FIGURE 17.6 Typical relationship between resultsfrom single point workability tests (data from Ellis,1977).
nature of the CRM and its interaction withthe cement.
• The small bubbles of air produced by air-entraining agents provide lubrication toreduce the plastic viscosity, but at relativelyconstant yield stress.
There is a great deal of information available onthe effect of mix constituents and proportions onworkability measurements using single pointtests, particularly slump. Since the specificationand achievement of a slump value is the basis forthe mix design method that we will describe inChapter 21, to avoid unnecessary duplication wewill leave discussion of these relationships untilthen.
17.4 Loss of workabilityAlthough cement remains sufficiently workablefor handling and placing for some time after ithas been mixed, its workability continuallydecreases. This is due to:
• mix water being absorbed by the aggregate ifthis not in a saturated state before mixing;
• evaporation of the mix water;• early hydration reactions (but this should not
be confused with cement setting);
Properties of fresh concrete
134
Plastic viscosity
Yiel
d st
ress
More water
Less water
pfa
Plasticiser
ggbs
More paste
Less paste
Superplasticiser
Air-entrainingagent
FIGURE 17.7 Summary of the effect of varying theproportions of concrete constituents on the Binghamconstants.
0
50
100
150
200
0 20 40 60 80 100 120Time (mins)
Slu
mp
(m
m)
Mix 120° 29°C
Cement � 300 kg/m3
no admixtures 20° 29°CMix 2
FIGURE 17.8 Typical slump loss behaviour of mixes differing in water content only (Previte, 1977).
• interactions between admixtures (particularlyplasticisers and superplasticisers) and thecementitious constituents of the mix.
Absorption of water by the aggregate can beavoided by ensuring that saturated aggregate isused, for example by spraying aggregate stock-piles with water and keeping them covered inhot/dry weather, although this may be difficult insome regions. It is also difficult, and perhapsundesirable, with lightweight aggregates. Evapo-ration of mix water can be reduced by keepingthe concrete covered during transport and han-dling as far as possible.
Most available data relates to loss of slump,which increases with higher temperatures, higherinitial slump, higher cement content and higheralkali and lower sulfate content of the cement.Figure 17.8 shows data from two mixes differingin water content only which illustrate the firsttwo factors.
The rate of loss of workability can be reducedby continued agitation of the concrete, e.g. in areadymix truck, or modified by admixtures(Chapter 14). In principle, retempering, i.e.adding water to compensate for slump loss,should not have a significant effect on strength ifonly that water which has been lost by evapora-tion is replaced. Also, studies have shown thatwater can be added during retempering toincrease the initial water/cement ratio by up to 5per cent without any loss in 28-day strength
(Cheong and Lee, 1993). However, except in verycontrolled circumstances, retempering can lead tounacceptably increased water/cement ratio andhence lower strength, and is therefore bestavoided.
17.5 ReferencesACI (1990) Cement and Concrete Terminology. ACI
116R-90. American Concrete Institute, Detroit,USA.
ASTM (1993) Specification C 125–93: Standard defini-tions and terms relating to concrete and concreteaggregates.
Cheong, H.K. and Lee, S.C. (1993) Strength of retem-pered concrete. ACI Materials Journal 90, 3, 203–6.
Domone, P.L., Xu, Y. and Banfill, P.F.G. (1999) Devel-opments of the two-point workability test for high-performance concrete. Magazine of ConcreteResearch 51, 171–9.
Ellis, C. (1977) Some Aspects of pfa in Concrete,MPhil thesis, Sheffield City Polytechnic.
Previte, R.W. (1977) Concrete slump loss. ACI Mater-ials Journal 74, 8, 361–7.
RILEM (2000) TC 145-WSM. Compendium of Con-crete Workability Tests, RILEM, Paris.
Tattersall, G.H. (1991) Workability and QualityControl of Concrete, E & FN Spon, London.
Tattersall, G.H. and Banfill, P.F.G. (1983), The Rheol-ogy of Fresh Concrete, Pitman, London.
References
135
136
Chapter eighteen
Early ageproperties ofconcrete
18.1 Behaviour after placing18.2 Curing18.3 Strength gain and temperature effects18.4 References
Successful placing of concrete is not enough. It isnecessary to ensure that it comes through the firstfew days of its life without mishap, so that it goeson to have the required mature properties.
Immediately after placing, the concrete is stillin a plastic and at least semi-fluid state, and thecomponent materials are relatively free to move.During setting, it changes into a material which isstiff and unable to flow, but which has nostrength. Clearly it must not be disturbed duringthis period. At some stage, hardening starts andthe concrete develops strength, initially quiterapidly.
In this chapter we will discuss the behaviour ofthe concrete during each of these stages and howthey affect construction practice. The hydrationprocesses and the timescales involved have beendescribed in some detail in Chapter 14, and theirmodification by admixtures and cement replace-ment materials in Chapter 15. In particular, wediscussed the exothermic nature of the hydrationreactions, and we will see that this has someimportant consequences.
18.1 Behaviour after placingThe constituent materials of the concrete are ofdiffering relative density (cement 3.15, normalaggregates approx. 2.6, etc.) and therefore whilstthe concrete is in its semi-fluid, plastic state theaggregate and cement particles tend to settle andthe mix water has a tendency to migrateupwards. This may continue for several hours,until the time of final set and the onset of strengthgain. Interparticle interference reduces the move-ment, but the effects can be significant. There arefour inter-related phenomena – bleeding, segrega-tion, plastic settlement and plastic shrinkage.
18.1.1 Segregation and bleeding
Segregation involves the larger aggregate particlesfalling towards the lower parts of the pour, andbleeding is the process of the upward migrationor upward displacement of water. They oftenoccur simultaneously (Figure 18.1).
The most obvious manifestation of bleeding isthe appearance of a layer of water on the topsurface of concrete shortly after it has beenplaced; in extreme cases this can amount to 2 percent or more of the total depth of the concrete. In time this water either evaporates or is re-absorbed into the concrete with continuinghydration, thus resulting in a net reduction of theoriginal concrete volume. This in itself may not
be of concern, but there are two other effects ofbleeding that can give greater problems, illus-trated in Figure 18.1. Firstly, the cement paste ator just below the top surface of the concretebecomes water rich and therefore hydrates to aweak structure, a phenomenon known as surfacelaitance. This is a problem in, for example, floorslabs which are required to have a hard wearingsurface. Secondly, the upward migrating water
Behaviour after placing
137
Cement andaggregatesWater
Water richpockets
Surfacelaitance
FIGURE 18.1 Segregation and bleed in freshlyplaced concrete.
0
100
200
300
400
500
20 40 60
Compressive strength (MPa)
Dis
tanc
e fr
om b
ase
of c
olum
n (m
m)
/c 0.7 0.6 0.5
FIGURE 18.2 Variation ofstrength in a concrete column afterfull compaction (Hoshino, 1989).
can be trapped under aggregate particles, causinga local enhanced weakening of the transition zonebetween the paste and the aggregate, which mayalready be a relatively weak part of the concrete,and hence an overall loss of concrete strength.However, in most concrete, some bleed may beunavoidable, and may not be harmful. We willdiscuss the transition zone and its effects in moredetail in Chapter 20.
The combined effects of bleed and particle set-tlement are that after hardening the concrete inthe lower part of a pour of any significant depthis stronger than that in the upper part. This effectis illustrated in Figure 18.2, which shows testdata from trial columns made with three differentmixes. Even though these are of a modest heightof 500mm, the strength differences between thetop and bottom are of the order of 10 per cent
18.1.2 Plastic settlement
Overall settlement of the concrete will result ingreater movement in the fresh concrete near thetop surface of a deep pour. If there is any localrestraint to this movement from, say, horizontalreinforcing bars, then plastic settlement crackingcan occur, in which vertical cracks form along theline of the bars, penetrating from the surface tothe bars (Figure 18.3 (a)).
18.1.3 Plastic shrinkage
Bleed water arriving at an unprotected concretesurface will be subject to evaporation; if the rateof evaporation is greater than the rate of arrivalof water at the surface, then there will be a netreduction in water content of the surface con-crete, and plastic shrinkage, i.e. drying shrinkagewhilst the concrete is still plastic, will occur. Therestraint of the mass of concrete will cause tensilestrains to be set up in the near-surface region, andas the concrete has near zero tensile strength,plastic shrinkage cracking may result. The crack-ing pattern, illustrated in Figure 18.3 (b) is a fairlyregular ‘crazing’, and is therefore distinctly differ-ent from that resulting from plastic settlement.
Any tendency to plastic shrinkage cracking willbe encouraged by greater evaporation rates of thesurface water which occurs, for example, withhigher concrete or ambient temperatures, or if theconcrete is exposed to wind.
18.1.4 Methods of reducing segregationand bleed and their effects
A major cause of excessive bleed is the use of apoorly graded aggregate, with a lack of finematerial below a particle size of 300�m beingmost critical. This can be remedied by increasing
Early age properties of concrete
138
FIGURE 18.3 Causes of cracking in freshly placedconcrete.
the sand content, but if this is not feasible forother reasons, or if a particularly coarse sand hasto be used, then air entrainment can be an effect-ive substitute for the fine particles.
Higher bleeds may also occur with higherworkability mixes, and if very high workabilitiesare required, it is preferable to use superplasticis-ers rather than high water contents, as discussedin Chapter 14. Condensed silica fume, with itsvery high surface area, is also an effective bleedcontrol agent.
Bleed, however, cannot be entirely eliminated,and so measures must be taken in practice toreduce its effects if these are critical. Plastic settle-ment and plastic shrinkage cracks that occur soonafter placing the concrete can be eliminated byrevibrating the surface region, particularly inlarge flat slabs.
18.2 CuringAll concretes, no matter how great or small theirtendency to bleed, must be protected from mois-ture loss from as soon after placing as possible,and for the first few days of hardening. This willnot only reduce or eliminate plastic shrinkagecracking, but also ensure that there is an ade-quate supply of water for continued hydrationand strength gain. This protection is calledcuring, and is an essential part of any successfulconcreting operation, although often overlooked.Curing methods include:
• spraying or ponding the surface of the con-crete with water;
• protecting exposed surfaces from wind andsun by windbreaks and sunshades;
• covering surfaces with wet hessian and/orpolythene sheets;
• applying a curing membrane, usually a spray-applied resin seal, to the exposed surface; thisprevents moisture loss, and weathers away ina few weeks.
Extended periods of curing are required for mixesthat gain strength slowly, such as those contain-ing CRMs, particularly pfa and ggbs, and in con-ditions of low ambient temperatures.
18.3 Strength gain and temperatureeffects
18.3.1 Effect of curing temperature
We mentioned in Chapter 13 that the hydrationreactions between cement and water are tempera-ture-dependent and their rate increases withcuring temperature. Typical effects of curing tem-perature on concrete strength development areshown in Figure 18.4, in which the following fea-tures can be seen.
1. At early ages the rate of strength gainincreases with curing temperature, as wouldbe expected.
2. At later ages, however, higher strengths areobtained from the concrete cured at lowertemperatures. Explanations of this behaviourhave been conflicting, but it would seem that,as similar behaviour is obtained with cementpaste, the CBSBH gel more rapidly pro-duced at the higher temperatures is lessuniform and hence weaker than that producedat the lower temperatures. There also appearsto be an optimum temperature for maximumlong-term strength, in this case 13°C; this
Strength gain and temperature effects
139
FIGURE 18.4 Effect of curing temperature on con-crete strength (Portland cement, w/c�0.4) (Klieger,1958 by permission, American Concrete Institute).
optimum has been found to vary for differenttypes of cement.
3. The hydration reactions proceed at tempera-tures below the freezing point of water, 0°C.In fact they only cease completely at about�10°C. However, the concrete must only beexposed to such temperatures after a signific-ant amount of the mix water has been incor-porated in the hydration reactions, since theexpansion of free water on freezing willdisrupt the immature, weak concrete. Adegree of hydration equivalent to a strength of3.5MPa is considered sufficient to give protec-tion against this effect.
18.3.2 Maturity
Since the degree of cement hydration depends onboth time and temperature, it might be expectedthat maturity, defined as the product of time andcuring temperature, would show some correlationwith strength. For the reasons given above, –10°Cis taken as the datum point for temperature, andhence
maturity� �t(T�10)
where t and T are the time (normally in hours)and curing temperature (in °C) respectively.Figure 18.5 shows maturity plotted against
FIGURE 18.5 Strength versus maturity relationshipfor a concrete cured at three different temperatures(Carino et al., 1983).
strength for three different but constant curingtemperatures, �1, 21 and 43°C. It can be seenthat concrete at all three temperatures has similarstrength/maturity values at low maturities, butsubsequently the maturities for the lowest tem-perature give higher strength values. In addition,initial curing for even a few hours at tempera-tures higher or lower than that for subsequentcuring alters the strength/maturity relationship.
The maturity concept is useful when estimatingthe strength of concrete in a structure from thestrength of laboratory samples cured at a differ-ent temperature. However, the limitations of thetemperature range, as illustrated in Figure 18.5,must be borne in mind.
18.3.3 Heat of hydration effects
As well as being temperature-dependent, cementhydration is exothermic, and in Chapter 13 wediscussed in some detail the rate of heat output atconstant temperature (i.e. isothermal) conditionsin relation to the various hydration reactions. Theopposite extreme to the isothermal condition isadiabatic (i.e. perfect insulation or no heat loss),and in this condition the exothermic reactionsresult in heating of a cement paste, mortar orconcrete. This leads to progressively faster hydra-tion, heat output rate and temperature rise, theresult being substantial temperature rises in relat-ively short timescales (Figure 18.6). The tempera-ture rise in concrete is less than that in cementpaste as the aggregate acts as a heat sink andthere is less cement to react. An average rise of13°C per 100kg of cement per m3 of concrete hasbeen suggested for typical structural concretes.
When placed in a structure, the concrete willlose heat to its surrounding environment eitherdirectly or through formwork, and it will there-fore not be under either truly adiabatic or isother-mal conditions, but in some intermediate state.This results in some rise in temperature withinthe pour followed by cooling to ambient. Typicaltemperature/time profiles for the centre of poursof varying depths are shown in Figure 18.7; it canbe seen that the central regions of a pour with anoverall thickness in excess of about 1.5 to 2m
Early age properties of concrete
140
FIGURE 18.6 Typical temperature rise duringcuring under adiabatic conditions for a neat cementpaste and concrete with varying cement content (Bam-forth, 1988).
will behave adiabatically for the first few daysafter casting.
Such behaviour has two important effects.
1. The cool down from the peak temperature con-ditions will result in thermal contraction of theconcrete, which if restrained will result intensile stresses which may be sufficiently largeto crack the concrete. Restraint can result fromthe structure surrounding the concrete, e.g. the
FIGURE 18.7 The effect of depth of a concretepour on temperature rise at mid-depth during hydra-tion (placing temperature�20°C, Portland cementcontent � 400kg/m3) (Browne and Blundell, 1973 bypermission, Concrete Society).
soil underneath a foundation, or from the outerregions of the concrete pour itself, which willhave been subject to greater heat losses, andtherefore will not have reached the same peaktemperatures, or from reinforcement within theconcrete. The amount of restraint will obvi-ously vary in different structural situations.
As an example, a typical coefficient ofthermal expansion for concrete is 10�10�6
per °C, and therefore a thermal shrinkagestrain of 300�10�6 would result from a cooldown of 30°C. Taking a typical elasticmodulus for the concrete of 30GPa, andassuming complete restraint with no relax-ation of the stresses due to creep, the resultingtensile stress would be 9MPa, well in excessof the tensile strength of the concrete, whichwould therefore have cracked.
In structural concrete, a reinforcementsystem can be designed to control and reducethe effects of any cracking, but it is also neces-sary in pours of any substantial size to limitthe temperature differentials. Insulation byway of increased formwork thickness orthermal blankets will have some beneficialeffect, but, more commonly, or in addition,low-heat mixes are used. If strength or dura-bility criteria mean that a sufficiently low
Strength gain and temperature effects
141
FIGURE 18.8 Temperature variation at mid-heightof a 2.5m deep pour during hydration of concretemade with 100 per cent Portland cement, 70 per cent�30 per cent pfa and 25 per cent Portlandcement�75 per cent ggbs (Bamforth, 1980 by permis-sion, Institute of Civil Engineers).
cement content cannot be used, then either alow-heat Portland cement (mentioned inChapter 13) can be used, or, more conve-niently, partial cement replacement by pul-verised fuel ash (pfa) or ground granulatedblast furnace slag (ggbs) is an effective solu-tion, as shown in Figure 18.8. (These mater-ials were described in Chapter 14.)
FIGURE 18.9 Strengthdevelopment of concretemade with 100 per centPortland cement, 70 percent Portland cement�30per cent pfa and 25 percent Portland cement�75per cent ggbs when sub-jected to: (a) standardcuring at 20°C; and (b) thecuring cycle shown inFigure 18.8 (Bamforth,1980 by permission, Insti-tute of Civil Engineers).
As alternative or additional measures, thetemperature of the fresh concrete can bereduced by pre-cooling the mix water or theaggregates, or by injecting liquid nitrogen.
2. Much of the concrete will have hydrated forat least a few days after casting at tempera-tures higher than ambient, and the long-termstrength may therefore be reduced, due to theeffects described above. Typical effects of thison strength development are shown in Figure18.9. By comparing Figure 18.9 (a) and (b) itcan be seen that pfa and ggbs mixes do notsuffer the same strength losses as the 100 percent Portland cement mixes. Measurement ofthe concrete properties after being subjectedto such ‘temperature matched curing’ is there-fore extremely important if a full picture ofthe in-situ behaviour is to be achieved.
18.4 ReferencesBamforth, P.B. (1980) In-situ measurement of the effect
of partial Portland cement replacement using eitherfly ash or ground granulated blast-furnace slag onthe performance of mass concrete. Proc. Instn Civ.Engnrs, Part 2, 69, Sept., 777–800.
Bamforth, P.B. (1988) Early age thermal cracking inlarge sections: towards a design approach. Institutefor International Research, Proceedings of AsiaPacific Conference on Roads, Highways andBridges, Hong Kong, Sept.
Browne, R.D. and Blundell, R. (1973) Hong Kong.Proceedings of Symposium on Large Pours for RCStructures, Concrete Society University of Birming-ham, Sept., pp. 42–65.
Carino, N.J., Lew, H.S. and Volz, C.K. (1983) Temper-ature effect on strength maturity relationships ofmortar. J. Amer. Concrete Inst., 80, 2, March–April,93-101.
Hoshino, J. (1989) Relationship between bleeding,coarse aggregate and specimen height of concrete.ACI Materials Journal 86, 2, 185–90.
Klieger, P. (1958) Effect of mixing and curing tempera-ture on concrete strength. Journ. Amer. ConcreteInst. 54, 12, June, 1063–81.
Early age properties of concrete
142
Chapter nineteen
Deformation ofconcrete
19.1 Drying shrinkage19.2 Autogenous shrinkage19.3 Carbonation shrinkage19.4 Thermal expansion19.5 Stress–strain behaviour19.6 Creep19.7 References
Deformation of concrete results both fromenvironmental effects, such as moisture gain orloss movement and heat, and from applied stress,both short- and long-term. A general view of thenature of the behaviour is given in Figure 19.1,which shows the strain arising from a uniaxialcompressive stress applied to the concrete in adrying environment. The load or stress is appliedat a time t1, and held constant until removal attime t2. Before applying the stress, there is a netcontraction in volume of the concrete, or shrink-age, associated with the drying. The dotted exten-sion in this curve beyond time t1 would be thesubsequent behaviour without stress, and theeffects of the stress are therefore the differencesbetween this curve and the solid curves. Immedi-ately on loading there is an instantaneous strainresponse, which for low levels of stress isapproximately proportional to the stress, andhence an elastic modulus can be defined. Withtime, the strain continues to increase at a decreas-ing rate. This increase, after allowing for shrink-age, represents the creep strain. Althoughreducing in rate with time, the creep does nottend to a limiting value.
On unloading, at time t2, there is an immediate
(elastic) strain recovery, which is often less thanthe initial strain on loading. This is followed by atime-dependent creep recovery, which is less thanthe preceding creep, i.e. there is a permanentdeformation, but, unlike creep, this reaches com-pletion in due course.
In this chapter we shall discuss the mechanismsand the factors influencing the magnitude of all ofthe components of this behaviour, i.e. shrinkage,elastic response and creep, and also consider ther-mally induced strains. We shall for the most partbe concerned with the behaviour of hardenedcement paste and concrete when mature, butsome mention of age effects will be made.
19.1 Drying shrinkage
19.1.1 Drying shrinkage of hardenedcement paste
In Section 13.6 we described the broad divisionsof water in hardened cement paste (hcp) and howtheir removal led to a net volumetric contraction,or drying shrinkage of the paste. Even thoughshrinkage is a volumetric effect, it is normallymeasured in the laboratory or on structural ele-ments by determination of length change, and itis therefore expressed as a linear strain.
A considerable complication in interpretingand comparing drying shrinkage measurements isthat specimen size will affect the result. Watercan only be lost from the surface and, therefore,the inner core of a specimen will act as a restraint
143
against overall movement; the amount ofrestraint and hence the measured shrinkage willtherefore vary with specimen size. In addition, therate of moisture loss, and hence the rate ofshrinkage, will depend on the rate of transfer ofwater from the core to the surface. The behaviour
of hcp discussed in this section is therefore basedon experimental data from specimens with arelatively small cross-section.
A schematic illustration of typical shrinkagebehaviour is shown in Figure 19.2. Maximumshrinkage occurs on the first drying, and a consid-
Deformation of concrete
144
Str
ain
(con
trac
tion)
Time
Elasticstrain
Creepstrain
Creeprecovery
Shrinkagestrain
Loadapplication
Elasticrecovery
Loadremoval
t1 t2
Irreversibleshrinkage
Firstdrying
Drying
Wetting
Drying
Wetting
Swellingwithcontinuousimmersion
Swelling
Shrinkage
Strain
Reversibleshrinkage
Time
FIGURE 19.1 The response of concrete to a compressive stress applied in drying environment.
FIGURE 19.2 Strain response of cement paste or concrete to alternate cycles of drying and wetting.
erable part of this is irreversible, i.e. is not recov-ered on subsequent rewetting. Further drying andwetting cycles result in more or less completelyreversible shrinkage; hence there is an importantdistinction between reversible and irreversibleshrinkage.
Also shown in Figure 19.2 is a continuous, butrelatively small, swelling of the hcp on continu-ous immersion in water. The water content firstincreases to make up for the self-desiccationduring hydration (see Section 13.4), and to keepthe paste saturated. Second, additional water isdrawn into the CBSBH structure to cause thenet increase in volume. This is a characteristic ofmany gels, but in hcp the expansion is resisted bythe skeletal structure so that the swelling is smallcompared to the drying shrinkage strains.
In principle, the stronger the hcp structure, theless it will respond to the forces of swelling orshrinkage. This is confirmed by the results shownin Figure 19.3, in which the increasing totalporosity of the paste is, in effect, decreasingstrength. It is interesting that the reversibleshrinkage appears independent of porosity, andthe overall trend of increased shrinkage on first
drying is entirely due to the irreversible shrink-age.
The variations in porosity shown in Figure19.3 were obtained by testing pastes of differentwater/cement ratios, and in general, increasedwater/cement ratio will result in increasingshrinkage. As we have seen in Chapter 13, reduc-tion in porosity also results from greater degreesof hydration of pastes with the same water/cement ratio, but the effect of the degree ofhydration on shrinkage is not so simple. Theobvious effect should be that of reduced shrink-age with age of paste if properly cured; however,the unhydrated cement grains provide somerestraint to the shrinkage, and as their volumedecreases with hydration, an increase in shrink-age would result. Another argument is that amore mature paste contains more water of thetype whose loss causes greater shrinkage, e.g. lesscapillary water, and so loss of the same amountof water from such a paste would cause moreshrinkage. It is thus difficult to predict the neteffect of age on the shrinkage of any particularpaste.
Since shrinkage results from water loss, the
Drying shrinkage
145
�0.7
�0.6
�0.5
�0.4
�0.3
�0.2
�0.1
0
0.25 0.3 0.35 0.4 0.45 0.5
Porosity of hcp
Irreversibleshrinkage
Str
ain
(%)
First shrinkage
Reversibleshrinkage
FIGURE 19.3 Reversible and irreversible shrinkage of hcp after drying at 47 per cent relative humidity(Helmuth and Turk, 1967).
relationship between the two is of interest.Typical data are given in Figure 19.4, whichshows that there is a distinct change of slope withincreased moisture losses, in this case about a 17per cent loss. This implies that there is more thanone mechanism of shrinkage; as other tests haveshown two or even three changes of slope, it islikely that, in fact, several mechanisms areinvolved.
19.1.2 Mechanisms of shrinkage andswelling
Four principal mechanisms have been proposedfor shrinkage and swelling in cement pastes,which are now summarised.
Capillary tension
Free water surfaces in the capillary and larger gelpores (see Section 13.6) will be in surface tension,and when water starts to evaporate due to a low-ering of the ambient vapour pressure, the freesurface becomes more concave and the surfacetension increases (Figure 19.5). The relationshipbetween the radius of curvature, r, of the menis-cus and the corresponding vapour pressure, p, isgiven by Kelvin’s equation:
ln(p/p0)�2T/R��r (19.1)
Deformation of concrete
146
�1.4
�1.2
�1
�0.8
�0.6
�0.4
�0.2
0
0 5 10 15 20 25
Water loss (% by weight)
w/c � 0.5
Str
ain
(%)
FIGURE 19.4 The effect of water loss on the dryingshrinkage of hardened cement paste (Verbeck andHelmuth, 1968).
FIGURE 19.5 Relationship between the radius ofcurvature and vapour pressure for water in a capillary(Soroka, 1979).
where p0 is the vapour pressure over a planesurface, T is the surface tension of the liquid, R isthe gas constant, � is the absolute temperatureand � the density of the liquid.
The tension within the water near the meniscuscan be shown to be 2T/r, and this tensile stressmust be balanced by compressive stresses in thesurrounding solid. Hence the evaporation whichcauses an increase in the tensile stress will subjectthe hcp solid to increased compressive stresswhich will result in a decrease in volume, i.e.shrinkage. The diameter of the meniscus cannotbe smaller than the diameter of the capillary, andthe pore therefore empties at the correspondingvapour pressure, p1. Hence on exposing a cementpaste to a steadily decreasing vapour pressure, thepores gradually empty according to their size, thewidest first. Higher water/cement ratio pasteswith higher porosities will therefore shrink more,thus explaining the general form of Figure 19.3.As a pore empties, the imposed stresses on thesurrounding solid reduce to zero, and so fullrecovery of shrinkage would be expected on com-plete drying. Since this does not occur, it is gener-ally accepted that other mechanisms becomeoperative at low humidities, and that thismechanism only applies at relative humiditiesabove about 50 per cent.
Surface tension or surface energy
The surface of both solid and liquid materials willbe in a state of tension due to the net attractiveforces of the molecules within the material.Work, therefore, has to be done against this forceto increase the surface area, and the surfaceenergy is defined as the work required to increasethe surface by unit area.
Surface tension forces induce compressivestresses in the material of value 2T/r (see above)and in the hcp solids, whose average particle sizeis very small, these stresses are significant.Adsorption of water molecules onto the surfaceof the particles reduces the surface energy, hencereducing the balancing internal compressivestresses, leading to an overall volume increase, i.e.swelling. This process is also reversible.
Disjoining pressure
Figure 19.6 shows a typical gel pore, narrowingfrom a wider section containing free water incontact with vapour to a much narrower spacebetween the solid in which all the water is underthe influence of surface forces. The two layers areprevented from moving apart by an interparticleVan der Waals type bond force. The adsorbedwater forms a layer about five molecules or1.3nm thick on the solid surface at saturation,which is under pressure from the surface-attract-ive forces. In regions narrower than twice thisthickness, i.e. about 2.6nm, the interlayer waterwill be in an area of hindered adsorption. Thisresults in the development of a swelling or dis-
Drying shrinkage
147
FIGURE 19.6 Water forces in a gel pore in hard-ened cement paste (Bazant, 1972).
joining pressure, which is balanced by a tensionin the interparticle bond.
On drying, the thickness of the adsorbed waterlayer reduces, as does the area of hinderedadsorption, hence reducing the disjoining pres-sure. This results in an overall shrinkage.
Movement of interlayer water
The mechanisms described above concern the freeand adsorbed water. The third type of evaporablewater, the interlayer water, may also have a role.Its intimate contact with the solid surfaces andthe tortuosity of its path to the open air suggestthat a steep hygrometric energy gradient isneeded to move it, but also that such movementis likely to result in significantly higher shrinkagethan the movement of an equal amount of free oradsorbed water. It is likely that this mechanism isassociated with the steeper slope of the graph inFigure 19.4 at the higher values of water loss.
The above discussions apply to the reversibleshrinkage only, but the reversibility depends onthe assumption that there is no change in struc-ture during the humidity cycle. This is highlyunlikely, at least during the first cycle, because:
1. the first cycle opens up interconnectionsbetween previously unconnected capillaries,thereby reducing the area for action of sub-sequent capillary tension effects;
2. some new interparticle bonds will formbetween surfaces that move closer together asa result of movement of adsorbed or inter-layer water, resulting in a more consolidatedstructure and a decreased total system energy.
Opinion is divided on the relative importance ofthe above mechanisms and their relative contribu-tion to the total shrinkage. These differences ofopinion are clear from Table 19.1, which showsthe mechanisms proposed by four main authors,and the suggested humidity levels over which theyact.
19.1.3 Drying shrinkage of concrete
Effect of mix constituents andproportions
The drying shrinkage of concrete is less than thatof neat cement paste because of the restraininginfluence of the aggregate, which, apart from afew exceptions, is dimensionally stable underchanging moisture states.
The effect of aggregate content is shown inFigure 19.7. It is apparent that normal concreteshave a shrinkage of some 10 to 20 per cent ofthat of neat paste. Aggregate stiffness will alsohave an effect. Normal density aggregates arestiffer and therefore give more restraint thanlightweight aggregates, and therefore lightweightaggregate concretes will tend to have a highershrinkage than normal density concretes ofsimilar volumetric mix proportions.
The combined effect of aggregate content andstiffness are contained in the empirical equation
�c/�p � (1�g)n (19.2)
where �c and �p are the shrinkage strains of theconcrete and paste respectively, g is the aggregatevolume content, and n is a constant whichdepends on the aggregate stiffness, and has beenfound to vary between 1.2 and 1.7.
Deformation of concrete
148
TABLE 19.1 Summary of suggested shrinkage mechanisms (Soroka, 1979)
SourceRelative humidity (%)
0 10 20 30 40 50 60 70 80 90 100
←→Powers (1965) disjoining pressure
←→capillary tension
←→← →Ishai (1965) surface energy capillary tension
←→← →Feldman and Sereda (1970) interlayer water capillary tension and surface energy
←→← →Wittman (1968) surface energy disjoining pressure
FIGURE 19.7 Influence of aggregate content in con-crete on the ratio of the shrinkage of concrete to that ofneat cement paste (Pickett, 1956).
The overall pattern of the effect of mix propor-tions on the shrinkage of concrete is shown inFigure 19.8; the separate effects of increasedshrinkage with increasing water content andincreasing water/cement ratio can be identified.
The properties and composition of the cementand the incorporation of pfa, ggbs and micro-silica all have little effect on the drying shrinkageof concrete, although interpretation of the data issometimes difficult. Admixtures do not in them-selves have a significant effect, but if their use
results in changes in the mix proportions then, asshown in Figure 19.8, the shrinkage will beaffected.
Effect of specimen geometry
The size and shape of a concrete specimen willinfluence the rate of moisture loss and the degreeof overall restraint provided by the central core,which will have a higher moisture content thanthe surface region. The rate and amount ofshrinkage and the tendency for the surface zonesto crack are therefore affected.
In particular, longer moisture diffusion pathslead to lower shrinkage rates. For example, amember with a large surface area to volume ratio,e.g. a T-beam, will dry and therefore shrink morerapidly than, say, a beam with a square cross-section of the same area. In all cases, however,the shrinkage process is protracted. In a studylasting 20 years, Troxell et al. (1958) found thatin tests on 300�150mm diameter cylindersmade from a wide range of concrete mixes and
stored at relative humidities of 50 and 70 percent, an average of only 25 per cent of the 20-year shrinkage occurred in the first two weeks, 60per cent in three months, and 80 per cent in oneyear.
The non-uniform drying and shrinkage in astructural member will result in differentialstrains and hence shrinkage-induced stresses –tensile near the surface and compressive in thecentre. The tensile stresses may be sufficient tocause cracking, which is the most serious con-sequence for structural behaviour and integrity.However, as discussed above, the effects in prac-tice occur over a protracted timescale, and thestresses are relieved by creep before crackingoccurs. The structural behaviour is thereforecomplex and difficult to analyse with any degreeof rigour.
19.1.4 Prediction of shrinkage
It is clear from the above discussion that,although much is known about shrinkage and the
Drying shrinkage
149
700
800
900
1000
1100
1200
300 400 500 600 700 800
Cement content (kg/m3)
Shr
inka
ge (m
icro
stra
in)
Water/cement ratio
Water content(kg/m3)
0.5 0.450.4
0.35
0.3
230
0.25
210
190
170150
FIGURE 19.8 Typical effects of cement content, water content and water/cement ratio on shrinkage of con-crete – moist curing for 28 days followed by drying for 450 days (Shoya, 1979).
factors that influence its magnitude, it is difficultto estimate its value in a structural situation withany degree of certainty. It has been shown that itis possible to obtain reasonable estimates of long-term shrinkage from short-term tests (Neville etal., 1983), but designers often require estimateslong before results from even short-term tests canbe obtained. There are a number of methods ofvarying degrees of complexity for this, oftenincluded in design codes, e.g. BSI (1997), ACI(2000a), all of which are based on the analysisand interpretation of extensive experimental data.
19.2 Autogenous shrinkageContinued hydration with an adequate supply ofwater leads to slight swelling of cement paste, asshown in Figure 19.2. Conversely, with no mois-ture movement to or from the cement paste, self-desiccation leads to removal of water from thecapillary pores (as described in Section 13.4) andautogenous shrinkage. It may be the only form ofshrinkage occurring in the centre of large mass ofconcrete, but its magnitude is normally at least anorder of magnitude less than that of dryingshrinkage. It can, however, be larger and moresignificant in concrete with very low water/cement ratios.
19.3 Carbonation shrinkageCarbonation shrinkage differs from dryingshrinkage in that its cause is chemical, and it doesnot result from loss of water from the hcp or con-crete. Carbon dioxide, when combined withwater as carbonic acid, attacks most of the com-ponents of the hcp, and even the very dilute car-bonic acid resulting from the low concentrationsof carbon dioxide in the atmosphere can havesignificant effects. The most important reaction isthat with the calcium hydroxide:
CO2 �Ca(OH)2 →CaCO3 �H2O (19.3)
Thus water is released and there is an increase inweight of the paste. There is an accompanyingshrinkage, and the paste also increases in strengthand decreases in permeability. The most likely
mechanism to explain this behaviour is that thecalcium hydroxide is dissolved from more highlystressed regions, resulting in the shrinkage, andthe calcium carbonate crystallises out in thepores, thus reducing the permeability and increas-ing the strength.
The rate and amount of carbonation depend inpart on the relative humidity of the surroundingair, and within the concrete. If the pores are satu-rated, then the carbonic acid will not penetratethe concrete, and no carbonation will occur; ifthe concrete is dry, then no carbonic acid is avail-able. Maximum carbonation shrinkage occurs athumidities of about 50 per cent and it can be ofthe same order of magnitude as drying shrinkage(Figure 19.9). The porosity of the concrete is alsoan important controlling factor. With averagestrength concrete, provided it is well compactedand cured, the carbonation front will only pene-trate a few centimetres in many years, and withhigh-strength concrete even less. However, muchgreater penetration can occur with poor qualityconcrete or in regions of poor compaction, andthis can lead to substantial problems if the con-crete is reinforced, as we shall see in Chapter 23.
19.4 Thermal expansionIn common with most other materials, cementpaste and concrete expand on heating. Knowledgeof the coefficient of thermal expansion is needed intwo main situations; first to calculate stresses dueto thermal gradients arising from heat of hydrationeffects or continuously varying diurnal tempera-tures, and second to calculate overall dimensionalchanges in structures such as bridge decks due toambient temperature variations.
The measurement of thermal expansions onlaboratory specimens is relatively straight-forward, provided sufficient time is allowed forthermal equilibrium to be reached (at most a fewhours). However, the in situ behaviour is compli-cated by the observation that, unlike most othermaterials, the thermal movement is time-dependent, and, as with shrinkage, it is difficultto estimate movement in structural elements fromthose on laboratory specimens.
Deformation of concrete
150
19.4.1 Thermal expansion of hardenedcement paste
The coefficient of thermal expansion of hcp variesbetween about 10 and 20�10�6 per °C, depend-ing mainly on the moisture content. Figure 19.10shows typical behaviour, with the coefficientreaching a maximum at about 70 per cent rh. Thevalue at 100 per cent rh, i.e. about 10�10�6 per°C, probably represents the ‘true’ inherent valuefor the paste itself. The time-dependence of thebehaviour takes the form of an initial expansionon an increase in temperature showing somereduction over a few hours if the temperature isheld constant.
Explanations for this behaviour have allinvolved the role of water, and relate to the dis-turbance of the equilibrium between the watervapour, the free water, the freely adsorbed water,the water in areas of hindered adsorption and theforces between the layers of gel solids (seeChapter 13, Section 13.6). Any disturbances willhave a greater effect at intermediate humidities,when there is a substantial amount of waterpresent with space in which to move. On anincrease in temperature, the surface tension of the
capillary water will decrease, and hence itsinternal tension and the corresponding compres-sion in the solid phases will decrease, causingextra swelling, as observed. However, changes ininternal energy with increased or decreased tem-perature will stimulate internal flow of water,causing the time-dependent volume change in theopposite sense to the initial thermal movementmentioned above.
19.4.2 Thermal expansion of concrete
The thermal expansion coefficients of the mostcommon rock types used for concrete aggregatesvary between about 6 and 10�10�6 per °C, i.e.lower than either the ‘true’ or ‘apparent’ valuesfor cement paste. The thermal expansion coeffi-cient for the concrete is therefore lower than thatfor cement paste, as shown in Figure 19.10. Fur-thermore, since the aggregate occupies 70 to 80per cent of the total concrete volume, there is aconsiderable reduction of the effects of humiditythat are observed in the paste alone, to the extentthat a constant coefficient of thermal expansionover all humidities is a reasonable approximation.
Thermal expansion
151
�0.2
�0.16
�0.12
�0.08
�0.04
0
0 20 40 60 80 100
Relative humidity (%)
Carbonationshrinkage
Dryingshrinkage
TotalshrinkageS
hrin
kage
str
ain
(%)
FIGURE 19.9 The effect of surrounding relative humidity on drying and carbonation shrinkage of mortar(Verbeck, 1958).
The value depends on the concrete mix propor-tions, chiefly the cement paste content, and theaggregate type; for normal mixes the latter tendsto dominate. The curves for quartz and limestoneaggregate concrete shown in Figure 19.10 repre-sent the two extremes of values for most normalaggregate concrete. Such values apply over a tem-perature range of about 0 to 60°C. At higher tem-peratures, the differential stresses set up by thedifferent thermal expansion coefficients of thepaste and aggregate can lead to internal microc-racking and hence non-linear behaviour. We shalldiscuss this further when considering fire damagein Chapter 24.
19.5 Stress–strain behaviour
19.5.1 Elasticity of the hardened cementpaste
Hardened cement paste has a near linear com-pressive stress–strain relationship for most of its
range and therefore a modulus of elasticity canreadily be determined from stress–strain data.Water-saturated pastes generally have a slightlyhigher modulus than dried pastes, indicating thatsome of the load is carried by the water in thepores. Nevertheless, the skeletal lattice of thepaste carries most of the load, and the elasticresponse is governed by the lattice properties. Asmight therefore be expected, the elastic modulus(Ep) is highly dependent on the capillary porosity(pc); the relationship has been found to be of theform
Ep �Eg(1�pc)3 (19.4)
where Eg is the modulus when pc �0, i.e. it repre-sents the modulus of elasticity of the gel itself. Thisis a similar expression to equation (13.9) for thestrength of the paste, and therefore it is to beexpected that the same factors will influence bothstrength and modulus. This is indeed the case; forexample, Figure 19.11 shows that decreasingwater/cement ratio and increasing age both
Deformation of concrete
152
Concrete:quartz aggregatelimestone aggregateTh
erm
al e
xpan
sion
coe
ffic
ient
(�
10
�6/°
C)
0
5
10
15
20
25
0 20 40 60 80 100
Relative humidity (%)
Very dry Partially dry Saturated
Cementpaste
FIGURE 19.10 The effect of dryness on the thermal expansion coefficient of hardened cement paste and con-crete (Myers, 1950).
increase the elastic modulus, a directly comparableeffect to that on strength shown in Figure 13.9.
19.5.2 Models for concrete behaviour
Concrete is, of course, a composite multiphasematerial, and its elastic behaviour will depend onthe elastic properties of the individual phases –unhydrated cement, cement gel, water, coarse andfine aggregate and their relative proportions andgeometrical arrangements. The real material istoo complex for rigorous analysis, but if it is con-sidered as a two-phase material consisting of hcpand aggregate, then analysis becomes possible,and instructive.
The model for the concrete behaviour requiresthe following:
1. The property values for the phases; in thissimple analysis, three are sufficient:(a) the elastic modulus of the aggregate, Ea;(b) the elastic modulus of the hcp, Ep; and(c) the volume concentration of the aggre-
gate, g.2. A suitable geometrical arrangement of the
phases; three possibilities are shown in Figure19.12. All the models consist of unit cubes.Models A and B have the phases arranged as
Stress–strain behaviour
153
FIGURE 19.11 Effect of water/cement ratio and ageon the elastic modulus of hardened cement paste(Hirsch, 1962).
FIGURE 19.12 Simple two-phase models for con-crete (Hansen, 1960; Counto, 1964).
adjacent layers, the difference being that in Athe two phases are in parallel, and thereforeundergo the same strain, whereas in B thephases are in series and are therefore sub-jected to the same stress. Model C has theaggregate set within the paste such that itsheight and base area are both equal to �g,thus complying with the volume requirements.This intuitively is more satisfactory in that itbears a greater resemblance to concrete.
Analysis of the models is not intended to give anydetail of the actual distribution of stresses andstrains within the concrete, but to predict averageor overall behaviour. Three further assumptionsare necessary:
1. the applied stress remains uniaxial and com-pressive throughout the model;
2 the effects of lateral continuity between thelayers can be ignored;
3. any local bond failure or crushing does notcontribute to the deformation.
Model A – phases in parallel
Strain compatibilityThe strain in the concrete, �c is equal to the strainin the aggregate, �a and the paste, �p, i.e.
�c � �a � �p (19.5)
EquilibriumThe total force is the sum of the forces on each ofthe phases. Expressed in terms of stresses andareas this gives
�c.1� �a.g� �p.(1�g) (19.6)
Constitutive relationsBoth of the phases and the concrete are elastic,hence
�c � �c.Ec �a � �a.Ea and �p � �p.Ep (19.7)
where Ec, Ea and Ep are the elastic moduli of theconcrete, aggregate and paste respectively.
Substituting into equation (19.6) from equation(19.7) gives
�c.Ec ��a.Ea.g� �p.Ep.(1�g)
and hence, from equation (19.5)
Ec �Ea.g�Ep.(1�g) (19.8)
Model B – phases in series
EquilibriumThe forces and hence the stresses (since the forcesact on equal areas) in both phases and the com-posite are equal, i.e.
�c � �a ��p (19.9)
StrainsThe total displacement is the sum of the displace-ments in each of the phases; expressed in terms ofstrain this gives
�c ��a.g� �p.(1�g) (19.10)
Substituting from equations (19.7) and (19.9)into (19.10) and rearranging gives
� � (19.11)
Model C – combined
This is a combination of two layers of hcp alonein series with a third layer of hcp and aggregatein parallel, as in model A. Repetition of the abovetwo analyses with substitution of the appropriategeometry and combination gives:
� � (19.12)�g
��Ea.g�Ep.(1��g)
(1��g)�
Ep
1�Ec
(1�g)�
Ep
g�Ea
1�Ec
Deformation of concrete
154
Figure 19.13 shows the predicted results of equa-tions 19.8, 19.11 and 19.12 from the threemodels for varying aggregate concentrations,with Ep �Ea as is normally the case. Models Aand B give upper and lower bounds respectivelyto the concrete modulus, with model C, not sur-prisingly, giving intermediate values. The effect ofaggregate stiffness is shown in non-dimensionalform in Figure 19.14, on which some typicalexperimental results are also plotted. It is clearthat, for concrete in which Ea/Ep is near 1, e.g.with low modulus lightweight aggregates, allthree models give a reasonable fit, but for normalaggregates which are stiffer than the paste, i.e.Ea/Ep �1, model C is preferable.
19.5.3 Measured stress–strain behaviourof concrete
The stress–strain behaviour of both the hcp andaggregate is substantially linear over most of therange up to a maximum. However, the compositeconcrete, although showing intermediate stiffnessas predicted from the above analysis, is markedlynon-linear over much of its length, as shown inFigure 19.15(a). Furthermore, successive unload-ing/loading cycles to stress levels below ultimateshow substantial, but diminishing, hysteresisloops, and residual strains at zero load, as inFigure 19.15(b).
The explanation for this behaviour lies in thecontribution of microcracking to the overallconcrete strains, i.e. assumption (3) in the aboveanalysis is invalid. As we will see in Chapter 20,the transition zone between the aggregate andthe hcp or mortar is a region of relative weak-ness, and in fact some microcracks will bepresent in this zone even before loading. Thenumber and width of these will depend on suchfactors as bleeding characteristics and theamount of drying or thermal shrinkage. As thestress level increases, these cracks will increasein length, width and number, thereby making aprogressively increasing contribution to theoverall strain, resulting in the non-linear behavi-our. The cracking eventually leads to completebreakdown and failure, and therefore we will
Stress–strain behaviour
155
FIGURE 19.13 The effect of volume concentrationof aggregate on elastic modulus of concrete calculatedfrom simple two-phase models.
FIGURE 19.14 Prediction of elastic modulus ofconcrete (Ec) from the modulus of the cement paste(Ep) and the aggregate (Ea) for 50 per cent volume con-centration of aggregate.
CB
A
Aggregate
Concrete
(b) Behaviour of concreteunder successive loading
cycles
Hardenedcementpaste
Str
ess
Strain Strain
(a) Typical behaviour of hcp,aggregate and concrete
First load cycle
Secondload cycle
Str
ess
FIGURE 19.15 Stress–strain behaviour of cement paste, aggregate and concrete.
postpone more detailed discussion of crackinguntil the next chapter.
Subsequent cycles of loading will not tend toproduce or propagate as many cracks as theinitial loading, provided the stress levels of thefirst or previous cycles are not exceeded. Thisexplains the diminishing size of the hysteresisloops shown in Figure 19.15(b).
19.5.4 Elastic modulus of concrete
The non-linear stress–strain curve for concretemeans that a number of different elastic modulusvalues can be defined. These include the slope ofthe tangent to the curve at any point (giving thetangent modulus, A or B in Figure 19.15 (b)) orthe slope of the line between the origin and apoint on the curve (giving the secant modulus, Cin Figure 19.15(b)).
A typical test involves loading to a workingstress, say 40 per cent of ultimate, and measuringthe corresponding strain. Cylindrical or prismspecimens are usually used, loaded longitudinally,and with a length at least twice the lateral dimen-sion. Strain measurements are usually taken overthe central section of the specimen to avoid endeffects. To minimise hysteresis effects, the speci-mens are normally subjected to a few cycles ofload before the strain readings are taken over aload cycle lasting about five minutes. It is usual tocalculate the secant modulus from these readings.This test is often called the static test, and theresulting modulus the static modulus, to distin-guish it from the dynamic modulus test whichwill be described in Chapter 22.
The elastic modulus increases with age anddecreasing water/cement ratio of the concrete, forthe reasons outlined above, and as with paste, thesetwo factors combine into an increase of moduluswith compressive strength, but with progressivelysmaller increases at higher strength. However, thereis no simple relationship between strength andmodulus since, as we have seen, the aggregatemodulus and its volumetric concentration, whichcan vary at constant concrete strength, also have aneffect. The modulus should therefore be determinedexperimentally if its value is required with any
certainty. This is not always possible, and estimatesare often needed, e.g. early in the structural designprocess, and a number of empirical relationshipshave been suggested, including
Ec �3.32(fcyl)0.5 �6.9 for strengths between 21 and 83MPa (ACI, 2000b)
(19.13)
where Ec �elastic modulus (in GPa) andfcyl �cylinder compressive strength (in MPa). (Wewill discuss the difference between cylinder andcube compressive strength in the next chapter.)
Alternatively:
Ec �9.5 (fck �8)0.33 (BSI, 1992) (19.14)
where fck �characteristic cylinder compressivestrength (in MPa).
Both of these expressions apply to concrete ofnormal density, i.e. made with normal densityaggregates. Lightweight concrete will have alower modulus for the same strength, and ACI(2000a) gives an equation incorporating density:
Ec �4.3�10�5 � �1.5 � fcyl0.5 (19.15)
where � �concrete density (in kg/m3).Although at first glance these expressions (and
others that have been suggested) appear conflict-ing, when plotted they all give largely similarresults. The important thing to remember is thatthey only give not very good estimates of theelastic modulus.
19.5.5 Poisson’s ratio
The Poisson’s ratio of water-saturated cement pastevaries between 0.25 and 0.3; on drying it reducesto about 0.2. It seems to be largely independent ofwater/cement ratio, age and strength. For concrete,the addition of aggregate again modifies the behavi-our, with lower values obtained with increasingaggregate content. For most concrete, values liewithin the range 0.17–0.2.
19.6 CreepThe general nature of the creep behaviour of con-crete was illustrated in Figure 19.1. The magni-
Deformation of concrete
156
tude of the creep strains can be several timeshigher than the elastic strains on loading, and theytherefore often have a highly significant influenceon structural behaviour. Also, the creep does notappear to tend to a limit, as shown in Figure19.16 for tests of 30 years’ duration. This figurealso shows that the creep is substantially increasedwhen the concrete is simultaneously drying, i.e.creep and shrinkage are interdependent. This leadsto the definitions of creep strains shown in Figure19.17. Free shrinkage (�sh) is defined as the shrink-age of the unloaded concrete in the drying con-dition, and basic creep (�bc) as the creep of asimilar specimen under load but not drying, i.e.sealed so that there is no moisture movement to orfrom the surrounding environment. The totalstrain (�tot) is that measured on the concrete whilstsimultaneously shrinking and creeping, and, asshown in Figure 19.16, it is found that
�tot � �sh ��bc (19.16)
The difference, i.e. �tot � (�sh ��bc) is called thedrying creep (�dc). It follows that the total creepstrain (�cr) is given by
�cr � �dc ��bc (19.17)
Creep
157
FIGURE 19.16 Creep of concrete moist-cured for28 days, then loaded and stored at different relativehumidities (Troxell et al., 1958).
Time
Str
ain
(b) Basic creep (stress, no loss of moisture)
(a) Free shrinkage (no stress)
(c) Total creep (stress and drying)
Totalstrain
�sh
�bc
�bc
�sh
�cr
�dc � Drying creep
FIGURE 19.17 Definitions of strains due to shrink-age, creep and combined shrinkage and creep of hard-ened cement paste and concrete.
19.6.1 Factors influencing creep
Apart from the increase in creep with simultane-ous shrinkage just described, the followingfactors have a significant effect on creep.
• A reduced moisture content before loading,which reduces creep. In fact, completely driedconcrete has very small, perhaps zero, creep.
• The level of applied stress; for any given con-crete and loading conditions, the creep isfound to increase approximately linearly withthe applied stress up to stress/strength ratiosof about 0.4 to 0.6 (different studies haveindicated different limits). It is therefore oftenuseful to define the specific creep as the creepstrain per unit stress in this region. At higherstress levels increased creep is observed, whichcan ultimately result in failure, as will be dis-cussed in the next chapter.
• Increasing concrete strength, which decreasesthe creep.
• Increasing temperature, which increases the
creep significantly for temperatures up toabout 70°C. Above this, moisture migrationeffects lead to lower creep.
• The aggregate volume concentration, illus-trated in Figure 19.18, which shows that theaggregate is inert as regards creep, and hencethe creep of concrete is less than that ofcement paste. This is, therefore, directly com-parable to the shrinkage behaviour shown inFigure 19.7.
Neville (1964) suggested a relationship betweenthe creep of the concrete (Cc) and that of neatcement paste (Cp) of the form
Cc/Cp � (1�g�u)n (19.18)
where g and u are the volume fractions of aggreg-ate and unhydrated cement respectively, and n isa constant which depends on the modulus of elas-ticity and Poisson’s ratio of the aggregate and theconcrete. This therefore shows that:
• the properties of the aggregate are important,and that they can have a substantial effect ofthe magnitude of the creep;
• the effect of the water/cement ratio and age ofconcrete need not be considered separately,since they both effect the elastic modulus;
• the effect of other materials which also effectthe rate strength gain, such as admixtures andcement replacement materials, can be treatedsimilarly.
19.6.2 Mechanisms of creep
Since the creep process is occurring within thecement paste, and the moisture content andmovement have a significant effect on its magni-tude, it is not surprising that the mechanisms pro-posed for creep have similarities with thoseproposed for shrinkage and discussed in Section19.1.2. As with shrinkage, it is likely that a com-bination of the mechanisms now outlined isresponsible.
Moisture diffusion
The applied stress causes changes in the internal
Deformation of concrete
158
Range fornormal
concrete
Spe
cific
cre
ep (m
icro
stra
in/M
Pa)
0
100
200
300
400
500
0 20 40 60 80 100
Aggregate content (% by volume)
FIGURE 19.18 The effect of aggregate content oncreep of concrete (Concrete Society, 1973).
stresses and strain energy within the hcp, result-ing in an upset to the thermodynamic equilib-rium; moisture then moves down the induced freeenergy gradient, implying a movement fromsmaller to larger pores, which can occur atseveral levels:
• in capillary water as a rapid and reversiblepressure drop;
• in adsorbed water moving more graduallyfrom zones of hindered adsorption – thismovement should be reversible;
• in interlayer water in diffusing very slowly outof the gel pores. Some extra bonding maythen develop between the solid layers, and sothis process may not be completely recover-able.
In sealed concrete, there are always sufficientvoids to allow the moisture movement, hencebasic creep can occur with this mechanism. Withsimultaneous drying, all of the processes aremuch enhanced, hence explaining drying creep.
Structural adjustment
Stress concentrations arise throughout the hcpstructure because of its heterogeneous nature, and
consolidation to a more stable state without lossof strength occurs at these points by either:
• viscous flow, with adjacent particles slidingpast each other; or
• local bond breakage, closely followed byreconnection nearby after some movement.
Concurrent moisture movement is assumed todisturb the molecular pattern, hence encouraginga greater structural adjustment. The mechanismsare essentially irreversible.
Microcracking
We have seen that hcp and concrete containdefects and cracks before loading, and propaga-tion of these and the formation of new cracks willcontribute to the creep strains, particularly athigher levels of stress. This is the most likelyexplanation of the non-linearity of creep strainwith stress at high stress levels. In a drying con-crete, the stress gradient arising from the mois-ture gradient is likely to enhance the cracking.
Delayed elastic strain
The ‘active’ creeping component of the hcp orconcrete, i.e. mainly the water in its variousforms in the capillary or gel pores, will be actingin parallel with inert material that will undergoelastic response only. In hcp this will be solid gelparticles, the unhydrated cement particles, andthe calcium hydroxide crystals, augmented inconcrete with the aggregate particles. The stressin the creeping material will decline as the load istransferred to the inert material, which thendeforms elastically as its stress graduallyincreases. The process acts in reverse on removalof the load, so that the material finally returns toits unstressed state; thus the delayed elastic strainwould be fully recoverable in this model.
19.6.3 Prediction of creep
As with shrinkage, it is often necessary to estim-ate the likely magnitude of the creep of a struc-tural element at the design stage, but, again,
because of the number of factors involved, pre-diction of creep with a degree of certainty isproblematic. Brooks and Neville (1978) suggestthat a satisfactory method is to carry out short-term (28 day) tests, and then estimate creep at alater age by extrapolation using the expressions:
basic creep ct �c28 �0.5t0.21 (19.19)total creep ct�c28�(�6.19�2.15loget)0.38 (19.20)
where t�age at which creep is required (days �28)c28 �measured specific creep at 28 daysct � specific creep at t days in microstrain
per MPa.
If short-term tests are not feasible, there are, asfor shrinkage, a number of empirical methods ofvarying degrees of complexity for estimatingcreep, often included in design codes, e.g. BSI(1997), ACI (2000a).
19.7 ReferencesACI (2000a) ACI 209R-92 Prediction of creep, shrink-
age and temperature effects in concrete structures.ACI Manual of Concrete Practice, Part 1 Materialsand General Properties of Concrete, American Con-crete Institute, Michigan, USA.
ACI (2000b) ACI 363R-23 State of the art report onhigh strength concrete. ACI Manual of ConcretePractice, Part 1 Materials and General Properties ofConcrete, American Concrete Institute, Michigan,USA.
Bazant, Z.P. (1972) Thermodynamics of hinderedadsorption and its implications for hardened cementpaste and concrete. Cement and Concrete Research2, 1, January, 1–16.
Brooks, J.J. and Neville, A.M. (1978) Predicting thelong-term creep and shrinkage from short-term tests.Mag. of Concrete Research 30, 103, 51–61.
BSI (1992) DD ENV 1–1: 1992 Eurocode 2: Design ofConcrete Structures, Part 1 General Rules and Rulesfor Buildings, British Standards Institution, London.
BSI (1997) BS 8110–1: 1997 Structural Use of Con-crete. Code of Practice for Design and Construction,British Standards Institution, London.
Concrete Society (1973) The Creep of Structural Con-crete, Technical Report No. 101, London.
Counto, U.J. (1964) The effect of the elastic modulusof aggregate on the elastic modulus, creep and creeprecovery of concrete. Magazine of ConcreteResearch 16, 48, Sept., 129–38.
References
159
Feldman, R.F. and Sereda, P.J. (1970) A new model forhydrated Portland cement and its practical implica-tions. Engng J. 53, 8/9, 53–9.
Hansen, T.C. (1960) Creep and stress relaxation ofconcrete. Proceedings of Swedish Cement and Con-crete Research Institute 31.
Helmuth, R.A. and Turk, D.M. (1967) The reversibleand irreversible drying shrinkage of hardened Port-land cement and tricalcium silicate pastes. Journal ofthe Portland Cement Association, Research andDevelopment Laboratories 9, 2, 8–21.
Hirsch, T.J. (1962) Modulus of elasticity of concrete asaffected by elastic moduli of cement paste matrixand aggregate. Proc. Amer. Conc. Inst. 59, March,427.
Ishai, O. (1965) The time-dependent deformationalbehaviour of cement paste, mortar and concrete.Proceedings of the Conference on Structure of Con-crete and its Behaviour Under Load, Cement andConcrete Association, London, Sept., 345–64.
Meyers, S.L. (1950) Thermal expansion characteristicsof hardened cement paste and concrete. Proceedingsof the Highway Research Board 30, 193.
Neville, A.M. (1964) Creep of concrete as a function ofits cement paste content. Magazine of ConcreteResearch 16, 46, March, 21–30.
Neville, A.M., Dilger W.H. and Brooks, J.J. (1983)Creep of Plain and Structural Concrete, Construc-tion Press, London, 361 pp.
Pickett, G. (1956) Effect of aggregate on shrinkage ofconcrete and hypothesis concerning shrinkage. J.Amer. Conc. Inst. 52, January, 581–90.
Powers, T.C. (1965) Mechanisms of shrinkage andreversible creep of hardened cement paste. Proceed-ings of the Conference on Structure of Concrete andits Behaviour Under, Load, Cement and ConcreteAssociation, London, Sept., 319–44.
Shoya, M. (1979) Drying shrinkage and moisture lossof superplasticiser admixed concrete of lowwater/cement ratio. Transactions of the Japan Con-crete Institute II – 5, 103–10.
Soroka, I. (1979) Portland Cement Paste and Concrete,Macmillan, London.
Troxell, G.E., Raphael, J.M. and Davis, R.E. (1958)Long-time creep and shrinkage tests of plain andreinforced concrete. Proc. Amer. Soc. Test andMater. 58, 1101–20.
Verbeck, G.J. (1958) Carbonation of hydrated Port-land cement. ASTM Special Publication 205, Ameri-can Society for Testing and Materials, 17–36.
Verbeck, G.J. and Helmuth, R.A. (1968) Structure andphysical properties of cement paste. Proceedings ofthe Symposium on the Chemistry of Cement, Tokyo,3, 1–37.
Wittman, F.H. (1968) Surface tension, shrinkage andstrength of hardened cement paste. Mater. Struct. 1,6, 547–52.
Deformation of concrete
160
Chapter twenty
Strength andfailure ofconcrete
20.1 Strength tests20.2 Factors influencing strength of Portland cement concrete20.3 Strength of concrete containing CRMs20.4 Cracking and fracture in concrete20.5 Strength under multiaxial loading20.6 References
Strength is probably the most important singleproperty of concrete, since the first considerationin structural design is that the structural elementsmust be capable of carrying the imposed loads.The maximum value of stress in a loading test isusually taken as the strength, even though undercompressive loading the test piece is still whole(but with substantial internal cracking) at thisstress, and complete breakdown subsequentlyoccurs at higher strains and lower stresses.Strength is also important because it is related toseveral other important properties which aremore difficult to measure directly, and a simplestrength test can give an indication of these pro-perties. For example, we have already seen therelation of strength to elastic modulus; we shalldiscuss durability in the next chapter, but inmany cases a low permeability, low porosity con-crete is the most durable, and, as discussed whenwe considered the strength of cement paste inChapter 13, this also means it has high strength.
In structural situations concrete will be subjectto one of a variety of types of loading, resultingin different modes of failure. Knowledge of the
relevant strength is therefore important; forexample, in columns or reinforced concretebeams, the compressive strength is required; forcracking of a concrete slab the tensile strength isimportant. Other situations may require torsionalstrength, fatigue or impact strength or strengthunder multiaxial loading. As we shall see, moststrength testing involves the use of a few, relat-ively simple tests, generally not related to aparticular structural situation. Proceduresenabling data from the tests described in thischapter to be used in design have been obtainedfrom empirical test programmes at an engineeringscale on large specimens. The reader is referred totexts on structural design for a description ofthese design procedures.
In this chapter we shall describe the mostcommon test methods used to assess concretestrength, discuss the factors influencing the resultsobtained from them, and then return to a moredetailed consideration of the cracking and frac-ture processes taking place within the concrete.Finally, we shall briefly discuss strength undermultiaxial loading situations.
20.1 Strength tests
20.1.1 Compressive strength
The simplest compressive strength test uses a con-crete cube, and this is the standard test in the
161
UK.1 The cube must be sufficiently large to ensurethat an individual aggregate particle does notunduly influence the result; 100mm is recom-mended for maximum aggregate sizes of 20mmor less, 150mm for maximum sizes up to 40mm.The cubes are usually cast in lubricated steelmoulds, accurately machined to ensure thatopposite faces are smooth and parallel. The con-crete is fully compacted by external vibration orhand tamping, and the top surface trowelledsmooth. After demoulding when set, the cube isnormally cured under water at constant tempera-ture until testing.
The cube-testing machine has two heavyplatens through which the load is applied to theconcrete. The bottom one is fixed and the upperone has a ball seating which allows rotation tomatch the top face of the cube at the start ofloading. This then locks in this position duringthe test. The load is applied to a pair of faceswhich were cast against the mould, i.e. with thetrowelled face to one side. This ensures that thereare no local stress concentrations which wouldresult in a falsely low average failure stress. Avery fast rate of loading gives overhigh strengths,and a rate to reach the ultimate stress in a fewminutes is recommended. It is vital that the cubeis properly made and stored; only then will thetest give a true indication of the properties of theconcrete, unaffected by such factors as poor com-paction, drying shrinkage cracking, etc.
The cracking pattern within the cube (Figure20.1 (a)) produces a double pyramid shape afterfailure. From this it is immediately apparent thatthe stress within the cube is far from uniaxial.The compressive load induces lateral tensilestrains in both the steel platens and the concretedue to the Poisson effect. The mismatch betweenthe elastic modulus of the steel and the concrete,and the friction between the two, result in lateralrestraint forces in the concrete near the platen.This concrete is therefore in a triaxial stress state,with consequent higher failure stress than the
Strength and failure of concrete
162
1 A list of relevant standards is included in ‘Further reading’on page 222.
true, unrestrained strength. This is the majorobjection to the cube test. The test is, however,relatively simple, and capable of comparing dif-ferent concretes. (We shall consider triaxial stressstates in more detail later in the chapter.)
An alternative test, which at least partly over-comes the restraint problem, uses cylinders; this ispopular in North America, most of Europe and inmany other parts of the world. Cylinders with aheight/diameter ratio of 2, most commonly300mm high and 150mm diameter, are testedvertically; the effects of end restraint are muchreduced over the central section of the cylinder,which fails with near uniaxial cracking (Figure20.1 (b)), indicating that the failure stress is muchcloser to the unconfined compressive strength. Asa rule of thumb, it is often assumed that the cubestrength is about 25 per cent higher than thecylinder strength, but this ratio has been found todepend on several factors, and in particular,reduces with increasing strength. For example,CEB-FIP (1990) gives
fcube �1.25fcyl for fcyl �40MPaand fcube �1.25fcyl for 40MPa� fcyl �80MPa
A general relationship between the height/diameter ratio (h/d) and the strength of cylindersfor low- and medium-strength concrete is shownin Figure 20.2. This is useful in, for example,
Cracking atapprox 45°to axis near
ends
Crackingparallel toloads awayfrom ends
(a) Cube (b) Cylinder with a height/diameter � 2
FIGURE 20.1 Cracking patterns during testing ofconcrete specimens in compression.
interpreting the results from testing cores cutfrom a structure, where h/d often cannot be con-trolled. It is preferable to avoid an h/d ratio ofless than 1, where sharp increases in strength areobtained, and high values, although giving closerestimates of the uniaxial strength, result in exces-sively long specimens which can fail due to slen-derness ratio effects.
Testing cylinders has one major disadvantage;the top surface is finished by a trowel and is notplane and smooth enough for testing, and ittherefore requires further preparation. It can beground, but this is very time consuming, and thenormal procedure is to cap it with a thin(2–3mm) layer of high-strength gypsum plaster,molten sulfur or high early strength cement paste,applied a day or two in advance of the test. Alter-natively, the end of the cylinder can be set in asteel cap with a bearing pad of an elastomericmaterial or fine dry sand between the cap and theconcrete surface. Apart from the inconvenience ofhaving to carry this out, the failure load is sensi-tive to the capping method, particularly in high-strength concrete.
20.1.2 Tensile testing
Direct testing of concrete in uniaxial tension, asshown in Figure 20.3 (a), is more difficult thanfor, say, steel or timber. Relatively large cross-
Strength tests
163
0.8
1
1.2
1.4
1.6
1.8
0 1 2 3 4
Height/width ratio
Rel
ativ
e st
reng
th
Height
Width
FIGURE 20.2 The relationship between height/width (or diameter) ratio and strength of concrete incompression.
sections are required to be representative of theconcrete, and, because the concrete is brittle, it isdifficult to grip and align. Eccentric loading andfailure at or in the grips is then difficult to avoid.A number of gripping systems have beendeveloped, but these are somewhat complex, andtheir use is confined to research laboratories. Formore routine purposes, one of the following twoindirect tests is preferred.
Splitting test
A concrete cylinder, of the type used for compres-sion testing, is placed on its side in a compressiontesting machine and loaded across its vertical dia-meter (Figure 20.3 (b)). The size of cylinder usedis normally either 300 or 200mm long (l) by 150or 100mm diameter (d). The theoretical distribu-tion of horizontal stress on the plane of the verti-cal diameter, also shown in Figure 20.3 (b), is anear uniform tension (fs), with local high com-pression stresses at the extremities. Hardboard orplywood strips are inserted between the cylinderand both top and bottom platens to reduce theeffect of these and ensure even loading over thefull length.
Failure occurs by a split or crack along the ver-tical plane, the specimen falling into two neathalves. The cylinder splitting strength is definedas the magnitude of the near-uniform tensilestress on this plane, which is given by
fs �2P/ld (20.1)
where P is the failure load.The state of stress in the cylinder is biaxial
rather than uniaxial, and this, together with thelocal zones of compressive stress at the extremes,results in the value of fs being higher than the uni-axial tensile strength. However, the test is veryeasy to perform with standard equipment usedfor compressive strength testing, and gives consis-tent results; it is therefore very useful.
Flexural test
A rectangular prism, of cross-section b�d(usually 100 or 150mm square) is simply
supported over a span L (usually 400 or600mm). The load is applied at the third points(Figure 20.3 (c)), and since the tensile strength ofthe concrete is much less than the compressivestrength, failure occurs when a flexural tensilecrack at the bottom of the beam, normally withinthe constant bending moment zone between theloading points, propagates upwards through thebeam. If the total load at failure is P, then analy-sis based on simple beam-bending theory andlinear elastic stress–strain behaviour up to failure
gives the stress distribution shown in Figure20.3 (c) with a maximum tensile stress in the con-crete, fb, as
fb �PL/bd 2 (20.2)
fb is known (somewhat confusingly) as themodulus of rupture.
However, as we have seen in the precedingchapter, concrete is a non-linear material and theassumption of linear stress distribution is notvalid. The stress calculated from equation (20.2)
Strength and failure of concrete
164
P
P
Area � A
fd
fd � P/A
(a) Direct tension
Hardboardpacking
(Length � l)
P
Cylinder splitting tensile strength � fs � 2P/ld(P � failure load)
Tension
d
P
fs
Comp
(b) Indirect – cylinder splitting
Compression
L/3L/3 L/3
Cross-sectionb � d
Tension
Longitudinal bendingstress distribution at
mid-point
fb
Modulus of rupture � fb � PL/bd2
(P � failure load)
(c) Indirect – modulus of rupture
P
FIGURE 20.3 Tensile testing methods for concrete.
is therefore higher than that actually developed inthe concrete. The strain gradient in the specimenmay also inhibit the crack growth. For both ofthese reasons the modulus of rupture is alsogreater than the direct tensile strength.
20.1.3 Relationship between strengthmeasurements
We have already discussed the relationshipbetween cube and cylinder compressive strengthmeasurements.
The tensile strength, however measured, isroughly one order of magnitude lower than thecompressive strength. The relationship betweenthe two is non-linear, with a good fit being anexpression of the form
ft �a(fc)b (20.3)
where ft � tensile strength, fc �compressivestrength and a and b are constants. The Eurocodefor Design of Concrete Structures (BSI 1996)gives a�0.30 and b�0.67 when fc is the charac-teristic cylinder strength and ft is the designtensile strength. This relationship, converted toactual cube compressive strengths and tensile
strengths, is plotted in Figure 20.4 together withequivalent data from cylinder splitting andmodulus of rupture tests obtained by UCL under-graduate students in laboratory classes.
It is clear from this figure that, as we have alreadysaid, both the modulus of rupture and the cylinder-splitting tests give higher values than the directtensile test. The modulus of rupture is the highervalue, varying between about 10 and 17 per cent ofthe cube strength (the higher value applies to lowerstrengths). The cylinder-splitting strength is betweenabout 8 and 11 per cent of the cube strength, andthe direct tensile strength between about 6 and 8per cent of the cube strength. Figure 20.4 alsoshows that, as with all such relationships, there is aconsiderable scatter of individual data points aboutthe best-fit line, although in this case some of thismay be due to the inexperience of the testers.
20.2 Factors influencing strength ofPortland cement concreteIn this section we will consider the strength ofconcrete with Portland cement as the sole binder.The effect of cement replacement materials willbe discussed in the next section.
Factors influencing strength of Portland cement concrete
165
0
2
4
6
8
0 20 40 60 80
Compressive (cube) strength (MPa)
Tens
ile s
tren
gth
(MPa
)
Modulus of rupture Cylindersplitting
Direct tension(EC2)
FIGURE 20.4 The relationship between direct and indirect tensile strength measurements and compressivestrength (EC2, 1996, and UCL data).
20.2.1 Transition/interface zone
Before looking at the relationships between thestrength of concrete and the many factors whichinfluence it, we need to introduce an extremelyimportant aspect of the concrete’s structure. InChapter 13 we described the microstructure ofthe hardened cement paste that is formed duringhydration. Concrete is, of course, a mixture ofthe paste and aggregate, and it is the interfacebetween these that is of great interest. The pasteclose to the aggregate surface is substantially dif-ferent to that of the bulk paste, and crucially thistransition or interface zone is significantly weakerthan the rest of the paste. As the load on the con-crete increases, cracking will start in this zone,and subsequently propagate into the hcp untilcrack paths are formed through the concrete, asshown in Figure 20.5, which when sufficientlyextensive will result in complete breakdown, i.e.failure.
The formation, structure and consequences ofthe transition zone have been the focus of muchresearch since the mid-1980s, even to the extentof two major international conferences beingdedicated to the subject (Maso, 1992; Katz et al.,1998). Although there are some differences ofopinion, there is a general consensus that thezone is between 30 and 50 microns wide, andthat its structure in a much simplified form is asshown in Figure 20.6. This shows two main fea-tures:
• a very thin surface layer of calcium silicatehydrate on the aggregate, also containingsome small calcium hydroxide (calcite) crys-tals;
• most of the zone consists mainly of largercalcite crystals and fine needles of calciumsulfoaluminate (ettringite – see page 99).
Suggested mechanisms for the zone’s formationinclude an increased water/cement ratio at thepaste aggregate/interface due to:
• the ‘wall effect’, whereby the cement grainscannot pack as efficiently next to the aggreg-ate surface as they can in the bulk paste;
• mix water separation at the interface due tothe relative movement of the aggregate parti-cles and cement paste during mixing, com-pounded by localised bleeding (see page 136).
We should therefore think of concrete as a three-phase material – hardened cement paste, aggreg-ate and the transition zone, and it will be usefulto bear this model in mind during the discussionof the more important factors that effect concretestrength that now follows. We will discuss somefurther aspects of the cracking and failure processlater in the chapter.
20.2.2 Water/cement ratio
In Chapter 13 we saw that the strength of cementpaste is governed by its porosity, which in turn
Strength and failure of concrete
166
Aggregate
Hardenedcementpaste
Crack path anddirection ofpropagation
Transitionzone
FIGURE 20.5 Cracking patternin normal strength concrete.
depends on the water/cement ratio and degree ofhydration. The overall dependence of strength ofconcrete on the amount of cement, water and airvoids within it was recognised by Feret in 1896,who suggested a rule of the form:
fc �K(c/{c�w�a})2 (20. 4)
where fc � strength, c, w and a are the absolutevolumetric proportions of cement, water and airrespectively, and K is a constant.
Working independently, Abrams, in 1918,demonstrated an inverse relationship with con-crete strength of the form:
fc �k1/(k2w/c) (20.5)
This has become known as Abrams’ law,although strictly, as it based on empirical obser-vations, it is a rule.
The constants K, k1 and k2 are empirical anddepend on age, curing regime, type of cement,amount of air entrainment, test method and, to alimited extent, aggregate type and size.
Feret’s rule and Abrams’ law both give aninverse relationship between strength andwater/cement ratio for a fully compacted concrete
Factors influencing strength of Portland cement concrete
167
Aggregate
Thin surface layerof C–S–H fibres
Transition zone withlarge Ca(OH)2 crystalsand ettringite needles
Bulk cementpaste
30–50 microns
FIGURE 20.6 Features of the transition zone at thepaste–aggregate interface (adapted from De Rooij etal., 1998).
of the form shown in Figure 20.7. It is importantto recognise the limitations of such a relationship.First, at low water/cement ratios, the concretebecomes less workable and increasingly more dif-ficult to compact. Feret’s rule recognises thatincreasing air content will reduce the strength,and in general the strength will decrease by 6 percent for each 1 per cent of included air byvolume. This leads to the steep reduction instrength shown by the dashed lines in Figure20.7. The point of the divergence from the fullycompacted line can be moved further up and tothe left by the use of more efficient compactionand/or by improvements in workability withoutincreasing the water/cement ratio, for example byusing plasticisers or superplasticisers, which werediscussed in Chapter 14. Without such admix-tures, it is difficult to achieve adequate workabil-ity for most normal compaction methods atwater/cement ratios much below 0.4; with admix-tures this limit can be reduced to 0.25 or evenless.
At the other end of the scale, Abrams himselfshowed that his rule was valid for water/cementratios of up to 2 or more. However, at these highvalues the paste itself is extremely fluid, and it isvery difficult to achieve a homogeneous, cohesiveconcrete without significant segregation. In prac-tice, water/cement ratios in excess of 1 are rarelyused.
Str
engt
h
Reduction due toincompletecompaction
Fully compacted concrete
Water/cement ratio
Improved compactionmethods or increasedworkability
FIGURE 20.7 The general relationship betweenstrength and water/cement ratio of concrete (adaptedfrom Neville, 1995 by permission, Pearson Education).
Figure 20.8 shows a recent set of resultsobtained with class 42.5N Portland cement,where good compaction was achieved in labora-tory conditions at water/cement ratios down to0.33. This gives a good idea of typical concreteperformance, but we must add our normalproviso that the use of other constituent materials(cement source, aggregate type, etc.) will give dif-ferent strength levels.
We will be discussing examples of achievingstrengths significantly higher than those in Figure20.7 in Chapter 24 when we consider highperformance concrete.
20.2.3 Effect of age
The degree of hydration increases with age,leading to the effect of age on strength apparentfrom Figure 20.8. The strength at 28 days is oftenused to characterise the concrete for specificationand compliance purposes, probably because itwas originally thought to be a reasonable indica-tion of the long-term strength without having towait too long for test results. The rate of gain ofstrength varies with water/cement ratio, and, as
discussed in Chapter 13, the rate of hydrationand therefore the rate gain of strength will varywith cement composition and fineness, and thus itis impossible and futile to generalise.
It is important to remember that, as also dis-cussed in Chapter 13, the hydration reactions arenever complete, and, in the presence of moisture,concrete will continue to gain strength for manyyears, although, of course, the rate of increaseafter such times will be very small.
20.2.4 Temperature
As we discussed in Chapter 18, a higher tempera-ture maintained throughout the life of a concretewill result in higher short-term strengths butlower long-term strengths. Also, an early ageheating/cooling cycle from heat of hydrationeffects can lead to lower long-term strength, butthe effect can be reduced or even eliminated bythe incorporation of pulverised fuel ash orground granulated blast furnace slag. We shalldiscuss the effect of transient high temperatureswhen considering the durability of concrete in firein Chapter 23.
Strength and failure of concrete
168
0
10
20
30
40
50
60
70
80
90
0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1
Water/cement ratio
Com
pres
sive
str
engt
h (M
Pa)
Age56 days28 days7 days3 days
Curing in water at 20°Ccrushed gravel aggregate
FIGURE 20.8 Compressive strength versus water/cement for concrete made with class 42.5N Portland cement(Balmer, 2000).
20.2.5 Humidity
The necessity of a humid environment for ade-quate curing has already been discussed; for thisreason concrete stored in water will achieve ahigher strength than if cured in air for some or allof its life, as shown in Figure 20.9. Also, speci-mens cured in water will show a significantincrease in strength (5 per cent or more) ifallowed to dry out for a few hours before testing.
20.2.6 Aggregate properties, size andvolume concentration
As discussed above, for normal aggregate it is thestrength of the paste/aggregate bond or transitionzone that has a dominant effect on concretestrength; the aggregate strength itself is generallysignificant only in very-high strength concrete orwith the relatively weaker lightweight aggregates.Struble et al. (1980) have concluded that withsome carbonate and siliceous aggregates there isevidence that the structure and chemistry of thetransition zone are influenced by both the aggreg-ate mineralogy and surface. Also, increasedsurface roughness can improve the bond, prob-ably due to mechanical interlocking. For thisreason, and also the increased mechanical inter-
Factors influencing strength of Portland cement concrete
169
FIGURE 20.9 The influence of curing conditions onconcrete strength (Portland Cement Association, 1968).
locking of the aggregate particles themselves, con-cretes made with crushed rocks are typically some15 per cent stronger than those made withuncrushed gravels, provided all other mix propor-tions are the same.
The use of a larger maximum aggregate sizereduces the concrete strength, again provided allother mix proportions are the same. The reduc-tion is greater at lower water/cement ratios andlarger aggregate sizes (Figure 20.10). The largeraggregates have a lower overall surface area witha weaker transition zone, and this has a morecritical effect on the concrete strength at thelower water/cement ratios. In fresh concrete, thedecreased surface area with the increased aggreg-ate size leads to increased workability for thesame mix proportions, and therefore for mixdesign at constant workability the water contentcan be reduced and a compensating increase instrength obtained.
Increasing the volumetric proportion of aggreg-ate in the mix will, at constant water/cementratio, produce a relatively small increase in con-crete strength (Figure 20.11). This has been attri-buted, at least in part, to the increase in aggregateconcentration producing a greater number of
FIGURE 20.10 The effect of aggregate size andwater/cement ratio on concrete strength (Cordon andGillespie, 1963 by permission, American ConcreteInstitute).
secondary cracks prior to failure, which requiregreater energy, i.e. higher stress, to reach fracture.This effect is only valid if the paste contentremains high enough to at least fill the voids inthe coarse/fine aggregate system, thereby allowingcomplete consolidation of the concrete. Thistherefore imposes a maximum limit to the aggreg-ate content for practical concretes.
20.3 Strength of concrete containingCRMsWe discussed the nature, composition andbehaviour of four cement replacement materials –pfa, ggbs, microsilica and metakaolin – inChapter 15, and in particular we described thesupplementary or secondary pozzolanic reactionthat leads to the formation of further calcium sili-cate hydrates. When each CRM is used within itsoverall replacement level limitation (page 119),the general effect is an increase in long-termstrength compared to the equivalent Portlandcement mix (i.e. the two mixes being compareddiffering only in the binder composition). This isdue to a combination of:
• better packing of the particles in the freshstate, leading to an overall reduced porosity ofthe hardened cement paste after hydration;
• preferential enhancement of the transitionzone, which, as we have seen, is of higherporosity and is rich in calcite and is thereforea prime target for the secondary reactions.
Not surprisingly, the strength of mixes withCRMs does take some time to reach and overtakethat of the equivalent Portland cement mix.Figure 20.12 shows a schematic of the strengthdevelopment behaviour when each CRM is usedwithin its normal replacement limitation. The‘cross-over’ point for the microsilica mixes is veryearly, sometimes within a day; metakaolin mixestake a few days, ggbs mixes days or weeks, andpfa mixes weeks or months. The reasons for thedifferent rates of strength gain with the four dif-ferent materials lie in their descriptions given inChapter 15, which can be summarised as:
• the extremely fine particles, which act asnucleation sites for hydrate deposition, andvery high active silica content of the micro-silica results in the strength quickly overtak-
Strength and failure of concrete
170
FIGURE 20.11 The effect of aggregate concentra-tion and water/cement ratio on concrete strength(Erntroy and Shacklock, 1954).
Age
Com
pres
sive
str
engt
hDays Weeks Months
microsilica 5 – 20%metakaolin 10 – 40%ggbs 30 – 80%pfa 20 – 40%
FIGURE 20.12 Schematic of average strength gaincharacteristics of concrete containing cement replace-ment characteristics.
ing that of the equivalent Portland cementmix;
• the metakaolin is somewhat coarser so it is alittle slower to react, but it contains a highactive silica content and is therefore not farbehind the microsilica mixes;
• the ggbs and pfa have similar particle sizes tothe Portland cement, but the ggbs also con-tains its own calcium oxide which contributesto the secondary reactions.
If the slower rate of strength gain is a problem,mixes can, of course, be modified accordingly,e.g. by using plasticisers to maintain workabilityat a reduced water/cement ratio.
The contribution of the CRMs to strength isoften expressed in terms of an activity coefficientor cementing efficiency factor (k) which is ameasure of the relative contribution of the CRMto strength compared to an equivalent weight ofPortland cement. This means that if the amountof CRM is xkg/m3, then this is equivalent tokxkg/m3 of cement, and the concrete strength isthat that would be achieved with a cementcontent of c�kx, where c is the amount ofcement.
If k is greater than 1, then the CRM is moreactive than the cement, and if less than 1, it is lessactive. Its value will clearly vary with the age ofthe concrete, and also with the amount of CRMand other mix proportions. For 28-day-old con-crete and proportions of the CRM within theoverall limits of Figure 20.12, values of 3 formicrosilica, 1 for ggbs and 0.4 for pfa have beensuggested (Sedran et al., 1996). These can be usedin mix design as will be discussed in Chapter 21.
It is, however, difficult to do more than gener-alise on the time-scales and magnitude of thestrength characteristics, for two reasons:
• the vast amount of published information onthe properties of concrete containing CRMsshows that, with each CRM, there is a widerange of performance (not just of strength,but also of all other properties), due mainly tothe differences in physical and chemical com-position of both the CRM and the Portlandcement in the various test programmes;
• each set of tests will have been designed for adifferent purpose and therefore will have adifferent set of variables, such as changing thewater content to obtain equal workability orequal 28 days’ strength, and therefore it isoften difficult to compare like with like(indeed, this is a problem facing students innearly all areas of concrete technology).
20.4 Cracking and fracture inconcrete
20.4.1 Development of microcracking
As we discussed in Chapter 19, the non-linearstress–strain behaviour of concrete in compres-sion is largely due to the increasing contributionof microcracking to the strain with increasingload. Four stages of cracking behaviour havebeen identified, illustrated in Figure 20.13(a).Below 30 per cent of ultimate load, the transitionzone cracks remain stable, and the stress–straincurve remains approximately linear (Stage 1). Asthe stress increases beyond 30 per cent of ulti-mate, the cracks begin to increase in length,width and number, causing non-linearity, but arestill stable and confined to the transition zone(Stage 2). At loads above 50 per cent ultimate(Stage 3), the cracks start to spread into thematrix and become unstable at loads approaching75 per cent ultimate, resulting in further devia-tion from linearity. Above 75 per cent ultimate(Stage 4), spontaneous and unstable crack growthbecomes increasingly frequent, leading to veryhigh strains. Also at this stage the excessivecracking results in transverse strains increasing ata faster rate than the longitudinal strains, result-ing in an overall increase in volume (Figure20.13(b)).
Complete breakdown, however, does not occuruntil strains significantly higher than those atmaximum load are reached. Figure 20.14 showsstress–strain curves from strain controlled testson paste, mortar and concrete. The hcp has asmall descending branch after maximum stress;with the mortar it is more distinct, but with theconcrete it is very lengthy. During the descending
Cracking and fracture in concrete
171
region, excess cracking and slip at the paste–aggregate interface are occurring before thecracking through the hcp is sufficiently welldeveloped to cause complete failure.
20.4.2 Creep rupture
We discussed in Chapter 19 the contribution of
Strength and failure of concrete
172
FIGURE 20.13 Stress–strain behaviour of concrete under compressive loading: (a) from Glucklich (1965); (b)from Newman (1966).
FIGURE 20.14 Typical stress–strain characteristicsof aggregate, hardened cement paste, mortar and con-crete compressive loading (Swamy and KameswaraRao, 1973).
microcracking to creep. This increases with stresslevel to the extent that if a stress sufficiently closeto the short-term ultimate is maintained thenfailure will eventually occur, a process known ascreep rupture. There is often an acceleration increep rate shortly before rupture. The behaviourcan be shown by stress–strain relationshipsplotted at successive times after loading, giving anultimate strain envelope, shown for compressiveand tensile loading in Figures 20.15(a) and (b)respectively. The limiting stress below whichcreep rupture will not occur is about 70 per centof the short-term maximum for both compressionand tension.
20.4.3 The fracture mechanics approach
Griffith’s theory for the fracture of materials andits consequent development into fracture mechan-ics was described in general terms in Chapter 6.Not surprisingly, there have been a number ofstudies attempting to apply linear fracturemechanics to concrete, with variable results.Three main reasons for the difficulties encoun-tered have been suggested (American ConcreteInstitute, 1980):
1. Failure in compression, and to a lesser extentin tension, is controlled by the interaction of
many cracks, rather than by the propagationof a single crack.
2. Cracks in cement paste or concrete do notpropagate in straight lines, but follow tortu-ous paths around cement grains, aggregateparticles, etc., which both distort and bluntthe cracks (Figure 20.5).
3. Concrete is a composite made up of cementpaste, the transition zone and the aggregate,and each has its own fracture toughness (Kc),in themselves difficult to measure. There isalso disagreement on the size of concrete speci-men necessary to determine an overall frac-ture toughness.
Despite these difficulties, Kc values for cementpaste have been estimated as lying in the range0.1 to 0.5 MN/m3/2, and for concrete betweenabout 0.45 and 1.40MN/m3/2 (Mindess andYoung, 1981). Kc for the transition zone seems tobe smaller, about 0.1MN/m3/2, confirming thecritical nature of this zone. Comparison of thesevalues with those for other materials given inTable 6.1 shows the brittle nature of concrete.
20.5 Strength under multiaxialloadingSo far in this chapter our discussions on compres-sive strength have been concerned with the effectsof uniaxial loading, i.e. where �1 (or �x) is finite,and the orthogonal stresses �2 (or �y) and �3 (or�z) are both zero. In many, perhaps most, struc-tural situations concrete will be subject to amultiaxial stress state (i.e. �2 and/or �3 as well as�1 are finite). This can result in considerable modi-fications to the failure stresses, primarily byinfluencing the cracking pattern.
A typical failure envelope under biaxial stress(i.e. �3 �0) is shown in Figure 20.16, in whichthe applied stresses, �1 and �2 are plotted non-dimensionally as proportions of the uniaxial com-pressive strength, �c. First, it can be seen thatconcretes of different strengths behave very simi-larly when plotted on this basis. Not surprisingly,the lowest strengths in each case are obtained inthe tension–tension quadrant. The effect of com-bined tension and compression is to reduce con-siderably the compressive stress needed for failureeven if the tensile stress is significantly less thanthe uniaxial tensile strength. The cracking patternover most of this region (Type 1 in Figure 20.16)
Strength under multiaxial loading
173
FIGURE 20.15 The effect of sustained compressive and tensile loading on the stress–strain relationship forconcrete: (a) compressive loading; (b) tensile loading (Rusch, 1960; Domone, 1974).
is a single tensile crack, indicating that the failurecriterion is one of maximum tensile strain, withthe tensile stress enhancing the lateral tensilestrain from the compressive stress. In the regionof near uniaxial compressive stress, i.e. close tothe compressive stress axes, the cracking pattern(Type 2) is essentially the same as that in thecentral region of the cylinder shown in Figure20.1; i.e. the cracks form all around the specimenapproximately parallel to the compressive load.In the compression–compression quadrant, thecracking pattern (Type 3) becomes more regular,with the cracks forming in the plane of theapplied loads, splitting the specimen into slabs.Under equal biaxial compressive stresses, thefailure stress is somewhat larger than the uniaxialstrength. Both Type 2 and Type 3 crack patternsalso indicate a limiting tensile strain failure crite-rion, in the direction perpendicular to the com-pressive stress(es).
With triaxial stresses, if all three stresses arecompressive then the lateral stresses (�2 and �3)
act in opposition to the lateral tensile strain pro-duced by �1. This in effect confines the specimen,and results in increased values of �1 beingrequired for failure, as illustrated in Figure 20.17for the case of uniform confining stress (i.e.�2 � �3); the axial strength (�1ult) can be related tothe lateral stress by the expression
�1ult � �c �K�2 (or �3) (20.6)
where K has been found to vary between about 2and 4.5.
In describing strength tests in Section 20.1.1,we said that when a compressive stress is appliedto a specimen by the steel platen of a testmachine, the lateral (Poisson effect) strains inducerestraint forces in the concrete near the platendue to the mismatch in elastic modulus betweenthe concrete and the steel. This is, therefore, aparticular case of triaxial stress, and the cause ofthe higher strength of cubes compared to longerspecimens such as cylinders.
Strength and failure of concrete
174
FIGURE 20.16 Failure envelopesand typical fracture patterns for con-crete under biaxial stresses �1 and�2, relative to uniaxial stress �c
(Kupfer et al., 1969; Vile, 1965).
20.6 ReferencesBalmer, T. (2000) Investigation into the effects on the
main concrete relationship using class 42.5N Port-land cement at varying compliance levels. Associ-ation of Concrete Technology Diploma Report,BCA, Crowthorne.
British Standards Institution (1996) DD ENV:1992–1–2 Eurocode 2: Design of Concrete Struc-tures.
CEB-FIP (1990) Model Code for the Design of Con-crete Structures, Thomas Telford, London.
Cordon, W.A. and Gillespie, H.A. (1963) Variables inconcrete aggregates and Portland cement pastewhich influences the strength of concrete. J. Amer.Conc. Inst. 60, 8, August 1029–50.
De Rooij, M.R. and Bijen, J.M.J.M. and Frens, G.(1998) Introduction of syneresis in cement paste.Proc. of Int. RILEM Conf. on the Interfacial Trans-ition Zone in Cementitious Composites (eds A. Katz,A. Bentur, M. Alexander and G. Arliguie), Israel,March, E & FN Spon, London, pp. 59–67.
Domone, P.L. (1974) Uniaxial tensile creep and failureof concrete. Magazine of Concrete Research 26, 88,Sept., 144–52.
Erntroy, H.C. and Shacklock, B.W. (1954) Design ofhigh strength concrete mixes. Proceedings of theSymposium on Mix Design and Quality Control of
References
175
FIGURE 20.17 The effect of lateral confining stress(�2, �3) on the axial compressive strength (�1) of con-cretes of two different strengths (FIP/CEB, 1990).
Concrete, Cement and Concrete Association,London, pp. 55–73.
FIP/CEB (1990) State-of-the-Art Report on HighStrength Concrete, Thomas Telford, London.
Glucklich, J. (1965) The effect of microcracking ontime-dependent deformations and the long-termstrength of concrete. Proceedings of the Inter-national Conference on Structure of Concrete and itsBehaviour Under Load, Cement and ConcreteAssociation, London, Sept., pp. 176–89.
Katz, A., Bentur, A., Alexander, M. and Arliguie, G.(eds) (1998) Proc. of Int. RILEM Conf. on the Inter-facial Transition Zone in Cementitious Composites,Israel, March, E & FN Spon, London, p. 342.
Kupfer, H., Hilsdorf, H.K. and Rusch, H. (1969)Behaviour of concrete under biaxial stress. Proc.Amer. Conc. Inst. 66, 660.
Maso, J.C. (ed.) (1992) Proc. of Int. RILEM Conf. onInterfaces in Cementitious Composites, Toulouse,October, E & FN Spon, London, p. 315.
Mindess, S. and Young, J.F. (1981) Concrete, Prentice-Hall, New Jersey.
Neville, A.M. (1995) Properties of Concrete, 4th edn,Pearson Education, Harlow, p. 844.
Newman, K. (1966) Concrete systems. In CompositeMaterials (ed. L. Hollaway), Elsevier, London.
Portland Cement Association (1968) Design andControl of Concrete Mixes, 11th edn, Stokie, Illi-nois, USA.
Rusch, H. (1960) Researches towards a general flexuraltheory for structural concrete. J. Amer. Conc. Inst.57, 1, July, 1–29.
Sedran, T., de Larrard, F., Hourst, F. and Contamines,C. (1996) Mix design of self-compacting concrete(SCC). Proceedings of Int. RILEM Conf. on Produc-tion Methods and Workability of Concrete (edsP.J.M. Bartos, D.L. Marrs and D.J. Cleland) E &FN Spon, London, pp. 439–50.
Struble, L., Skalny, J. and Mindess, S. (1980) A reviewof the cement-aggregate bond. Cement and ConcreteResearch, 10, 2, March, 277–86.
Swamy, R.N. and Kameswara Rao, C.B.S. (1973)Fracture mechanism in concrete systems under uni-axial loading. Cement and Concrete Research 3, 4,July, 413–28.
Vile, G.W.D. (1965) The strength of concrete undershort-term biaxial stress. Proceedings of the Inter-national Conference on Structure of Concrete and itsBehaviour Under Load, Cement and ConcreteAssociation, London, Sept., pp. 275–88.
176
Chapter twenty-one
Concrete mixdesign
21.1 The mix design process21.2 UK method of ‘Design of normal concrete mixes’ (BRE,
1997)21.3 Mix design with cement replacement materials (CRMs)21.4 Design of mixes containing admixtures21.5 References
Mix design is the process of selecting the propor-tions of cement, water, fine and coarse aggregatesand, if to be used, cement replacement materialsand admixtures to produce an economic concretemix with the required fresh and hardened proper-ties. The cement and other binder constituents areusually the most expensive component(s), and‘economic’ usually means keeping its/theircontent as low as possible, without, of course,compromising the resulting properties. There mayalso be other advantages, such as reduced heat ofhydration (Chapter 17) or drying shrinkage(Chapter 19)
21.1 The mix design processFigure 21.1 shows the stages in the complete mixdesign process; we will discuss each of these inturn.
21.1.1 Specified concrete properties
The required hardened properties of the concreteresult from the structural design process, and aretherefore provided to the mix designer. Strengthis normally specified in terms of a characteristicstrength (or grade) at a given age (see Preface).
Durability requirements, to be discussed inChapter 23, may impose a limit on some mixproportions, e.g. a minimum cement content ormaximum water/cement ratio, or demand the useof air-entraining agent or a particular aggregatetype.
The choice of workability will depend on themethods selected for transporting, handling andplacing the concrete (e.g. pump, skip), the size ofthe section to be filled and the density of rein-forcement. The workability must clearly be suffi-cient at the point of placing, which in the case ofready-mixed concrete transported by road to site,may be some time after mixing.
21.1.2 Constituent material properties
As a minimum, the fine and coarse aggregate size,type and grading, and the cement type must beknown. The relative density of the aggregates, thecement composition, and details of any cementreplacement materials and admixtures that are tobe used or considered may also be needed.
21.1.3 Initial estimate of mix proportions
An initial best estimate of the mix proportionsthat will give concrete with the required proper-ties is then made. In doing this, as much use aspossible is made of previous results from concretemade with the same or similar constituent mater-ials. In some cases, for example in producing anew mix from an established site plant, thebehaviour of the materials will be well known. In
other circumstances, there will be no such know-ledge, and typical behaviour such as that given inthe preceding few chapters has to be used.
There are a considerable number of step-by-stepmethods of varying complexity that can be used toproduce this ‘best estimate’. Many countries havetheir own preferred method or methods, and, asan example, we will describe a current UK methodbelow. Whichever method is used, it is importantto recognise that the result is only a best estimate,perhaps even only a good guess. Because the con-stituent materials will not be exactly as assumedand their interaction cannot be predicted with anygreat certainty, the concrete is unlikely to meet therequirements precisely.
21.1.4 Laboratory trial mixes
It follows from the above that a trial mix todetermine the resulting properties of the ‘bestguess’ mix is essential. This is normally firstcarried out at a relatively small scale in the labo-ratory. Some adjustment to the mix portions willprobably be necessary when the test results areobtained, e.g. a decrease in the water/cementratio if the strength is too low. A second trial mixwith the revised mix proportions is then carriedout, and the process repeated until a satisfactorymix in all respects is obtained.
The mix design processs
177
Specified concrete properties,e.g strength, workability, durability
Constituent material propertiese.g aggregate size and grading,cement type, admixtures
Initial estimate ofmix proportions
Laboratory trial mix
Compare measured andspecified properties
Adjust mixproportions
End
Satisfactory Not satisfactory
Full-scale trial mix
Satisfactory Not satisfactory
Adjust mixproportions
FIGURE 21.1 The mix design process.
21.1.5 Full-scale trial mixes
The laboratory trials do not provide the completeanswer. The full-scale batching and mixing proce-dures will not be exactly the same as those in thelaboratory, and may cause differences in the con-crete properties. Complete confidence in the mixcan therefore only be obtained with further trialsat full scale, again with adjustments to the mixproportions and re-testing if necessary.
21.2 UK method of ‘Design ofnormal concrete mixes’ (BRE, 1997)This method of mix design provides a goodexample of the process of making an initial estim-ate of the mix proportions. It has the advantageof being relatively straightforward and producingreasonable results with the materials most com-monly available in the UK. It should be emphas-ised that it is not necessarily the ‘best’ methodavailable world-wide, and that it may not givesuch good results with other materials.
The main part of the method is concerned withthe design of mixes incorporating Portlandcement, water and normal density coarse and fineaggregates only, and with characteristic strengthsof up to about 70MPa. It encompasses bothcrushed and uncrushed coarse aggregate. Thesteps involved can be summarised as follows.
Concrete mix design
178
0
2
4
6
8
10
0 10 20 30 40 50 60 70
Characteristic strength (MPa)
Sta
ndar
d de
viat
ion,
s (
MPa
) s for fewerthan 20results
Min s for 20or moreresults
FIGURE 21.2 Standard deviation versus character-istic strength of concrete (BRE, 1997).
21.2.1 Target mean strength
As described in the Preface, the specified character-istic strength is a lower limit of strength to be usedin structural design. As with all materials, concretehas an inherent variability of strength, and anaverage cube compressive strength (or target meanstrength) somewhat above the characteristicstrength is therefore required. The differencebetween the characteristic and target mean strengthis called the margin; a 5 per cent failure rate is nor-mally chosen for concrete, and the margin shouldtherefore be 1.64 times the standard deviation ofthe strength test results (Figure 0.4).
This means that a knowledge of the standarddeviation is required. For an existing concreteproduction facility this will be known from previ-ous tests. Where limited or no data are available,the upper values given in Figure 21.2,1 which hasbeen derived from analysis of the data from manyproduction facilities, can be used. When produc-tion is under way, this can be reduced if justifiedby test results, but not below the lower values inFigure 21.2. The advantage of reducing the vari-ability by good practice is clear.
21.2.2 Free water/cement ratio
For a particular cement and aggregate type, theconcrete strength at a given age is assumed to begoverned by the free water/cement ratio only. Thefirst step is to obtain a value of strength atwater/cement ratio of 0.5 from Figure 21.3 forthe relevant age/aggregate type/cement type com-bination (this figure has been drawn from data ina table in the method document). This value isthen plotted on the vertical line in Figure 21.4 togive a starting point for a line which is con-structed parallel to the curves as shown. Thepoint of intersection of this line with the horizon-tal line of the required target mean strength thengives the required free water/cement ratio. Theranges of the axes in Figure 21.4 indicates thelimits of validity of the method.
1Figures 21.2 to 21.7 inclusive have been taken from BRE(1997), and Figures 21.2, 21.4, 21.6 and 21.7 are ©BRE andhave been reproduced with the permission of BRE.
21.2.3 Free water content
It is now assumed that, for a given coarse aggreg-ate type and maximum size, the concrete work-ability is governed by the free water content only.The workability can be specified in terms ofeither slump or Vebe time (see Chapter 17),although slump is by far the most commonlyused. Figure 21.5 is a graph of data for slump,again drawn from data given as a table in the
method document, from which the free watercontent for the appropriate aggregate can beobtained.
21.2.4 Cement content
This is a simple calculation from the values offree water/cement ratio and free water contentjust calculated.
UK method of ‘Design of normal concrete mixes’ (BRE, 1997)
179
20
30
40
50
60
70
0 10 20 30 40 50 60 70 80 90 100
Age (days)
Com
pres
sive
str
engt
h (M
Pa)
42.5N cement52.5R cement
Uncrushed aggregate
Crushed aggregate42.5N cement52.5R cement
0
10
20
30
40
50
60
70
80
90
0.3 0.4 0.5 0.6 0.7 0.8 0.9Free water/cement ratio
Com
pres
sive
str
engt
h (M
Pa)
Starting line withdata from Figure 21.3
FIGURE 21.3 Compressivestrength versus age for concretewith a water/cement ratio of 0.5(BRE, 1997).
FIGURE 21.4 Compressivestrength versus water/cementratio of concrete (BRE, 1997).
21.2.5 Total aggregate content
An estimate of the density of the concrete is nowrequired. This is obtained from Figure 21.6, usingknown or assumed values of the relative densityof the aggregates. A weighted mean value is usedif relative densities of the coarse and fine aggreg-ate are different. Subtraction of the free watercontent and cement content from this densitygives the total aggregate content.
21.2.6 Fine and coarse aggregate content
The estimated value of the proportion of fineaggregate in the total aggregate depends on themaximum size of aggregate, the concrete work-ability, the grading of the fine aggregate (specifi-cally the amount passing a 600 micron sieve) andthe free water/cement ratio. Figure 21.7 shows therelevant graphs for obtaining this proportion fora maximum aggregate size of 20mm. Sufficient
Concrete mix design
180
0
20
40
60
80
100
120
140
100 150 200 250 300
Free water content (kg/m3)
Slu
mp
(mm
)
Crushed aggregate:
10 mm20 mm40 mm
Uncrushedaggregate:
10 mm20 mm40 mm
FIGURE 21.5 Slump versusfree water content of concrete(BRE, 1997).
2100
2200
2300
2400
2500
2600
2700
2800
100 120 140 160 180 200 220 240 260
Free water content (kg/m3)
Wet
den
sity
(kg
/m3)
Relative density of aggregate(SSD basis)
2.9
2.4
2.5
2.6
2.7
2.8
Assume for:crushed aggregatesuncrushed aggregates
FIGURE 21.6 Wet densityof fully compacted concreteversus free water content (BRE,1997).
fine aggregate must be incorporated to produce acohesive mix that is not prone to segregation, andFigure 2.17 shows that increasing quantities arerequired with increasing water/cement ratio,increasing slump and if the aggregate itself iscoarser. The mix design document also givesequivalent graphs for 10 and 40mm coarseaggregate; with the former between 5 and 15 percent more fine aggregate is required, and with thelatter between 5 and 10 per cent less.
The fine and coarse aggregate content is nowcalculated by simple arithmetic, and the amounts(in kg/m3) of free water, cement, coarse and fineaggregates for the laboratory trial mix have nowall been obtained.
It is important to note the simplifying assump-tions used in the various stages. These make themethod somewhat simpler than many otheralternatives, but highlight the importance of trialmixes and subsequent refinements.
21.3 Mix design with cementreplacement materials (CRMs)As we have seen, CRMs effect both the fresh andhardened properties of concrete, and it is oftendifficult to predict their interaction with the Port-land cement with any confidence. The mix designprocess for concretes including CRMs is therefore
more complex, and again, trial mixes are essen-tial.
The mix design method described above (BRE,1997) includes modifications for mixes contain-ing good quality low-lime pfa or ggbs. With pfa:
• the amount, expressed as a proportion of thetotal binder, first needs to be selected, forexample for heat output, durability or eco-nomic reasons, subject to a maximum of 40per cent;
• the increase in workability is such that thewater content obtained from Figure 21.5 canbe reduced by 3 per cent for each 10 per centof pfa;
• the effect upon the strength is allowed for bythe use of a cementing efficiency factor, k,which we discussed in Chapter 20. This con-verts the amount of PFA to an equivalentamount of cement. The total equivalentcement content is then C�kF, whereC�Portland cement content and F�pfacontent. The value of k varies with the type ofpfa and Portland cement and with the age ofthe concrete, but a value of 0.30 is taken for28-day strength with a class 42.5 Portlandcement. Thus, if w�water content, a value ofthe equivalent water/cement ratio w/(C�kF)is obtained from Figure 21.4;
Mix design with cement replacement materials (CRMs)
181
% of fineaggregatepassing600 µmsieve
10
20
30
40
50
60
70
80
0.2 0.4 0.6 0.8
10080
60
40
15
60–180 mm
10
20
30
40
50
60
70
80
0.2 0.4 0.6 0.8
10080
60
40
15
30–60 mm
10
20
30
40
50
60
70
80
0.2 0.4 0.6 0.8
10080
60
40
15
10–30 mm
10
20
30
40
50
60
70
80
0.2 0.4 0.6 0.8
10080
60
40
15
Slump: 0–10 mm
Prop
ortio
n of
fin
e ag
greg
ate
(%)
Free water/cement ratio
FIGURE 21.7 Proportionsof fine aggregate according topercentage passing 600�msieve (for 20mm max. coarseaggregate size) (BRE, 1997).
• subsequent calculations follow using C�Fwhen the total binder content is required.
With ggbs:
• again the amount as a proportion of thebinder is first chosen, with values of up to 90per cent being suitable for some purposes;
• the improvements in workability are such thatthe water contents derived from Figure 21.5can be reduced by about 5kg/m3;
• the cementing efficiency factor approach usedfor pfa is more difficult to apply as the valueof k is dependent on more factors, includingthe water/equivalent cement ratio, and for 28-day strengths varies from about 0.4 to over1.0. It is assumed that for ggbs contents of upto 40 per cent there is no change to thestrength, i.e. k�1, but for higher proportionsinformation should be obtained from thecement manufacturer or the ggbs supplier.
21.4 Design of mixes containingadmixtures
21.4.1 Mixes with plasticisers
As we have seen in Chapter 14, plasticisersincrease the fluidity or workability of the con-crete. This leads to three methods of use:
1. to provide an increase in workability, bydirect addition of the plasticiser with no otherchanges to the mix proportions;
2. to give an increase in strength at the sameworkability, by allowing the water content tobe reduced, with consequent reduction in thewater/cement ratio;
3. to give a reduction in cement content for thesame strength and workability, by couplingthe reduction in water content with a corre-sponding reduction in cement content tomaintain the water/cement ratio.
(1) and (2) change the properties of the concrete,(3) will normally result in a cost saving as theadmixture costs much less than amount ofcement saved.
Table 21.1 gives typical figures for these effectsfor an average strength concrete mix and a ligno-sulfonate-based plasticiser. The figures have beenobtained from data provided by the admixturesupplier. The admixture amount is a ‘standard’dose. The important changes are, in each case:
1. an increase in slump from 75 to 135mm;2. an increase in strength from 39 to 45MPa;3. a reduction in cement content of 30kg/m3.
Plasticisers can have some effect on the settingtimes, but mechanical properties and durability atlater ages appear largely unaffected and aresimilar to those expected for a plain concrete ofthe same water/cement ratio, with two relativelyminor exceptions:
1. there is some evidence of a slight increase in28-day strength, attributed to the dispersionof the particles causing an increased surface
Concrete mix design
182
TABLE 21.1 Methods of using a plasticiser in average quality concrete (typical data from admixture supplier’sinformation)
Mix Cement Water w/c Plasticiser dose Slump 28 day strength(kg/m3) (% by wt of (mm) (MPa)
cement)
control 325 179 0.55 0 75 391 325 179 0.55 0.31 135 39.52 325 163 0.5 0.3 75 453 295 163 0.55 0.3 75 39
Note:1. A ‘standard’ dose.
area of cement being exposed to the mixwater (Hewlett, 1988);
2. some plasticisers entrain about 1–2 per centair because they lower the surface tension ofthe mix water. This will reduce the densityand strength of the concrete.
21.4.2 Mixes with superplasticisers
For the reasons explained in Chapter 14, it is verydifficult to generalise about the effects and uses ofsuperplasticisers other than to say that they canproduce greater increases in workability and/orstrength and/or cement reduction than plasticis-ers. They are more expensive than plasticisers,and therefore the economic advantages of cementreduction may not be as great. The suppliers willprovide information for each specific product orformulation, but a mix designer must ensurecompatibility with the proposed binder. This canoften be judged by tests on paste or mortar inadvance of trial mixes on concrete (Aitcin, 1994).
Superplasticisers enable a much greater rangeof concrete types to be produced than with plasti-cisers, e.g. high workability flowing concrete, self-compacting mixes and high-strength mixes lowwater/cement ratios. These will be discussed inChapter 24.
21.4.3 Mixes with air entraining agents
As discussed in Chapters 14 and 23, air entrain-ment is used to increase the resistance of concreteto freeze–thaw damage, but the entrained airincreases the workability and reduces the sub-sequent strength. The method of mix designdescribed above (BRE, 1997) includes the follow-ing modifications to allow for these effects if thespecified air content is within the normal range of3 to 7 per cent by volume:
• it is assumed that the strength is reduced by5.5 per cent for each 1 per cent of air; thetarget mean strength is therefore increased bythe appropriate amount;
• the slump is reduced by a factor of about twofor the selection of water content from Figure21.5;
• the concrete density obtained from Figure21.6 is reduced by the appropriate amount,
21.5 ReferencesAitcin, P.C., Jolicoeur, C. and MacGregor, J.G. (1994)
Superplasticizers: how they work and why theyoccasionally don’t. Concrete International 16, 5,45–52.
BRE (1997) Design of Normal Concrete Mixes, 2ndedn, Building Research Establishment, Watford.
Hewlett, P.C. (ed.) (1988) Cement Admixtures: Useand Applications, 2nd edn, Longman, Essex.
References
183
184
Chapter twenty-two
Non-destructivetesting ofhardenedconcrete
22.1 Surface hardness – rebound (or Schmidt) hammer test22.2 Resonant frequency test22.3 Ultrasonic pulse velocity test (upv)22.4 Near-to-surface tests22.5 References
There are a wide variety of available methods andtechniques available for the non-destructivetesting of structural concrete, which can bebroadly divided into those which assess the con-crete itself, and those which are concerned withlocating and determining the condition of thesteel embedded in it. We are going to describethree well-established tests for concrete which arestrictly non-destructive, and more briefly discussothers which involve some minor damage to theconcrete – the so-called near-to-surface tests. Wedo not have space to consider tests to assess thelocation and condition of reinforcing and pre-stressing, important though these are. Some refer-ences to useful selected publications on these andother tests on concrete are included in the‘Further reading’ on page 222.
Non-destructive testing of concrete is used fortwo main purposes:
• in laboratory studies, where it is particularlyuseful for repeated testing of the same speci-
men to determine the change of propertiestime, for example to provide information ondegradation in different environments;
• to assess the properties of concrete in a struc-ture, for example for compliance with speci-fications, after damage due to fire or overloador when a change of use is proposed.
Two of the tests that we will describe, therebound hammer and ultrasonic pulse velocity,can be used for both these purposes; the third, theresonant frequency test, can only be used on pre-pared specimens in the laboratory.
An estimation of the strength of the concrete isoften required, and therefore the degree of corre-lation of the non-destructive test measurement(s)with strength is important, and will be discussedin each case. It will be apparent that a single non-destructive test rarely gives a single definitiveanswer, and engineering judgement is required ininterpreting the results. Nevertheless, the useful-ness of such tests will become apparent.
22.1 Surface hardness – rebound(or Schmidt) hammer testThis is perhaps the simplest of the commonlyavailable tests, and can be used on laboratory
specimens or on in-situ concrete. The apparatus iscontained in a hand-held cylindrical tube, andconsists of a spring-loaded mass which is firedwith a constant energy against a plunger heldagainst the concrete surface (Figure 22.1). Theamount of rebound of the mass expressed as thepercentage of the initial extension of the spring isshown by the position of a rider on a graduatedscale, and recorded as the rebound number. Lessenergy is absorbed by a harder surface, and so therebound number is higher. A smooth concretesurface is required, but even then there is consid-erable local variation due to the presence ofcoarse aggregate particles (giving an abnormallyhigh rebound number) or a void (giving a lownumber) just below the surface, and therefore anumber of readings must be taken and averaged.Typical recommendations are for ten readingsover an area of 150mm diameter, with no tworeadings being taken within 25mm of each other.1
Also, the concrete being tested must be part of anunyielding mass; laboratory specimens such ascubes should therefore be held under a stress ofabout 7MPa in a compression testing machine.
The test clearly only measures the properties ofthe surface zone of the concrete, to a depth ofabout 25–30mm. The correlation betweenrebound number and concrete strength dependson:
• the aggregate type, since the hardness is afunction of both strength and elastic modulusof the concrete;
• the moisture condition of the surface;• the angle of the hammer with the vertical,
which will vary since the test must be carriedout with the plunger normal to the concretesurface.
There is therefore no single universal correlation.Figure 22.2 shows the relationship betweenrebound number and strength obtained by stu-dents at UCL in laboratory classes over severalyears. The degree of scatter is somewhat higherthan that reported by other workers, the most
Resonant frequency test
185
FIGURE 22.1 Rebound hammer: (1) plunger; (2)concrete; (3) tubular housing; (4) rider; (5) scale; (6)mass; (7) release button; (8) spring; (9) spring; (10)catch (Neville, 1995 by permission, Pearson Educa-tion).
1A list of relevant standards is included in ‘Further reading’ onpage 222.
likely explanation being the inexperience of theoperatives. Even with more skilful operatives,strength cannot be predicted with great certainty,but the test is very simple and convenient, and sois often used as a first step in an investigation ofin situ concrete, for example to assess uniformityor to compare areas of known good quality andsuspect concrete.
22.2 Resonant frequency testThis is a laboratory test on prepared specimens,and can be used to assess progressive changes inthe specimen due, for example, to freeze/thawdamage or chemical attack. It is therefore particu-larly useful for generating data in durabilitytesting.
The specimen is in the form of a beam, typically500�100�100mm; the test normally consists ofmeasuring the beam’s fundamental longitudinalresonant frequency when it is supported at itsmid-point. A value of elastic modulus called thedynamic elastic modulus can be obtained fromthis frequency (n), the length of the beam (l) andits density (�) using the relationship:
Ed �4.n2.l 2.� (22.1)
The resonant frequency is measured with the testsystem shown in Figure 22.3. The vibration isproduced by a small oscillating driver in contactwith one end of the beam, and the response of thebeam is picked up by a similar device at the otherend (Figure 22.3 (a)). The amplitude of vibration
varies along the beam as in Figure 22.3 (b). Thefrequency of the driver is altered until amaximum amplitude of vibration is detected bythe pick-up, indicating resonance (Figure22.3 (c)). The frequency is normally displayeddigitally and manually recorded.
The test involves very small strains, but, as wehave seen in Chapter 19, concrete is a non-linearmaterial. The dynamic modulus, Ed, is therefore
in effect the tangent modulus at the origin of thestress–strain curve, i.e. the slope of line B inFigure 19.15(a), and it is higher than the static orsecant modulus (Es) measured in a conventionalstress–strain test, i.e. the slope of line C in Figure19.15(b). The ratio of Es to Ed depends on severalfactors, including the compressive strength, but isnormally between 0.8 and 0.85.
As with the static modulus, for a particular set
Non-destructive testing of hardened concrete
186
0
10
20
30
40
50
60
70
80
10 20 30 40 50 60 70Rebound number
Com
pres
sive
cub
e st
reng
th (
MPa
)
Tests on cubesMoist surfaceSiliceous gravel aggregateEach result average of 5 readings
FIGURE 22.2 Relation between strength of concrete and rebound test results (UCL data).
Driver frequency
Pick
-up
ampl
itude
(c) Frequency response curve(b) Amplitude of vibration
BeamDriver Pick-up
(a) Test system
Support
Resonance
FIGURE 22.3 Measurement of the longitudinal resonant frequency of a concrete beam.
of constituent materials the dynamic modulusand strength can be related; Figure 22.4 showsdata obtained by students at UCL. The amount ofscatter is less than that for rebound hammerversus strength (Figure 22.2), mainly because thedynamic modulus gives an average picture of theconcrete throughout the beam, not just atlocalised points. The relationship is clearly non-linear, as with those for static modulus andstrength given in equations (19.13) to (19.15).The applicability of relationships such as those inFigure 22.4 to only a restricted range of para-meters (aggregate type, curing conditions, etc.)must be emphasised.
22.3 Ultrasonic pulse velocity test(upv)This is an extremely versatile and popular test forboth in-situ and laboratory use. It involves mea-suring the time taken for an ultrasonic pulse totravel through a known distance in concrete,from which the velocity is calculated. The ultra-sonic signal is generated by a piezo-electric crystal
housed in a transducer, which transforms an elec-tric pulse into a mechanical wave. The pulse isdetected by a second similar transducer, whichconverts it back to an electrical impulse, and thetime taken to travel between the two transducersis measured and displayed by the instrumenta-tion. Various test arrangements, illustrated inFigure 22.5, are possible. Efficient acoustic cou-pling between the transducers and the concrete isessential, and is usually obtained by a thin layerof grease. The pulse velocity is independent of thepulse frequency, but for concrete fairly low fre-quencies in the range 20–150kHz (most com-monly 54kHz) are used to give a strong signalwhich is capable of passing through severalmetres of concrete. Transducers which producelongitudinal waves are normally used, althoughshear wave transducers are available.
The velocity (V) of the longitudinal ultrasonicpulse depends on the material’s dynamic elasticmodulus (Ed), Poisson’s ratio (�) and density (�):
V���� (22.2)Ed(1� �)
���(1��)(1�2�)
Ultrasonic pulse velocity test (upv)
187
0
10
20
30
40
50
60
70
80
30 35 40 45 50 55
Dynamic elastic modulus (GPa)
Com
pres
sive
cub
e st
reng
th (
MPa
)
Siliceous gravel aggregateWater curing
FIGURE 22.4 Relation between strength and dynamic elastic modulus of concrete (UCL data).
Hence upv is related to the elastic properties, and,as with Ed, can be correlated empirically tostrength, but with the similar limitations ofdependence on constituent materials and as withthe rebound hammer, moisture conditions. Figure22.6 shows UCL students’ data. The relation isclearly non-linear, which is to be expected sinceupv is related to dynamic modulus, but shows agreater degree of scatter than the strength/Ed rela-tionship in Figure 22.4. Two factors contribute tothis. First, the upv test requires a little more skillthan the resonant frequency test, e.g. in ensuringgood acoustic coupling between the transducerand the concrete. Second, the results wereobtained on 100mm cubes, and therefore asmaller and inherently more variable volume ofconcrete was being tested. It is extremely import-ant to bear both of these factors in mind wheninterpreting any non-destructive test data.
The ultrasonic pulse is travelling through both
the hardened cement paste and the aggregate, andhence the pulse velocity will depend on the veloc-ity through each and their relative proportions.The velocity through normal density aggregate ishigher than that through the paste, which leads tothe broad relationships between upv and strengthfor paste, mortar and concrete shown in Figure22.7.
The upv test has the great advantage of beingable to assess concrete throughout the signalpath, i.e. in the interior of concrete element.Direct transmission is preferred, but for in situmeasurements, semi-direct or indirect operationcan be used if access to opposite faces is limited(Figure 22.5). With in situ testing, it is also veryimportant to ensure that measurements are takenwhere they are not influenced by the presence ofreinforcing steel, through which the pulse travelsfaster (upv�5.9km/sec), and which can thereforeresult in a falsely low transit time.
Non-destructive testing of hardened concrete
188
(b) Semi-direct transmission (c) Indirect transmission
Transmittingtransducer
Receivingtransducer
Path length l
Concrete
(a) Direct transmission
Time t
Pulse velocity � t/l
FIGURE 22.5 Measurement of ultrasonic pulse velocity in concrete.
Ultrasonic pulse velocity test (upv)
189
0
10
20
30
40
50
60
70
80
4 4.2 4.4 4.6 4.8 5
Ultrasonic pulse velocity (km/sec)
Com
pres
sive
cub
e st
reng
th (
MPa
)
Siliceous gravel aggregateWater curing
FIGURE 22.6 Relation between strength and ultrasonic pulse velocity (UCL data).
0
10
20
30
40
50
60
70
80
2 2.5 3 3.5 4 4.5 5 5.5
Com
pres
sive
str
engt
h (M
Pa)
Ultrasonic pulse velocity (km/sec)
Hardenedcement paste
Mortar
Concrete
FIGURE 22.7 Envelope of strength and ultrasonic pulse velocity for hardened cement paste, mortar and con-crete (based on Sturrup et al., 1984, and UCL data).
22.4 Near-to-surface tests
The specific need to assess the strength of in situconcrete has led to the development of a range oftests in which the surface zone is penetrated orfractured. The limited amount of damageincurred does not significantly affect the struc-
tural performance of the concrete elements ormembers, but it does normally require makinggood after the test for aesthetic or durabilityrequirements.
The five main types of test are shown in Figure22.8; there are a number of commercial versionsof each test. The penetration resistance test, in
Non-destructive testing of hardened concrete
190
Load
Reactionringforce
Wedgeanchor bolt
in drilled hole
Fractureplane
(a) Internal fracture
LoadReactionring force
Embeddedinsert
Fractureplane
(b) Pull-out
Penetrationdepth
Fracture surface
Load
Coredconcrete
30 mm
Scale
(c) Pull-off
(d) Penetration resistance
(e) Break-off
Load
Reactionring force
Circular disk, resin bondedto concrete surface
Failure surface
FIGURE 22.8 The main types of near-to-surface tests for concrete (all to same approx. scale).
which the depth of penetration of a bolt fired intothe concrete with a standard explosive charge ismeasured, is somewhat different from the otherswhich involve the measurement of the forcerequired to cause a fracture of some form. Theforce is applied via a device either inserted into apre-drilled hole, as in the internal fracture andbreak-off tests, or glued to the concrete surface,as in the pull-off tests, or cast into the concrete,as in the pull-out test. This last type involves pre-planning before or during construction, althougha version which fits into an under-reamed drilledhole is available.
To give an estimate of compressive strength,prior collaboration in the laboratory is necessaryfor all tests, and, as with the truly non-destructivetests already described, considerable scatter isobtained. Table 22.1 gives recommendations forthe number of test repetitions required to give asatisfactory mean value, and the resulting confi-dence of the strength estimation. It can be seen
that this is of a similar order of magnitude to thatin Figures 22.2 and 22.6. All of the correlationsare also dependent on a number of factors,particularly the aggregate type.
The non-destructive and near-to-surface testsare often used in advance of taking larger coresamples, which after preparation can be tested togive a direct measurement of strength, but whichinvolved a lot more work, and result in moredamage to the structure.
22.5 ReferencesBungey, J. (1992) Near-to-surface strength testing.
Concrete, Sept–Oct, 34–6.Neville, A.M. (1995) Properties of Concrete, 4th edn,
Longman, Essex, p. 625.Sturrup, V.R., Vecchio, F.J. and Caratin, H. (1984)
Pulse velocity as a measure of concrete compressivestrength. In In-situ/Non-destructive Testing of Con-crete (ed. V. Malhotra), ACI SP-82, American Con-crete Institute, Detroit, USA, pp. 201–7.
References
191
TABLE 22.1 Required repetitions and accuracy of strength estimation for near-to-surface tests (Bungey, 1992by permission, Concrete Society)
Test method Recommended number of Likely 95% confidence limittests for a mean value at for compressive strength
each location estimation
Internal fracture 6 !30%Pull-out 4 !10–20%Pull-off 6 !15%Penetration resistance 3 !20%Break-off 5 !20%
192
Chapter twenty-three
Durability ofconcrete
23.1 Transport mechanisms through concrete23.2 Measurements of flow constants for cement paste and
concrete23.3 Degradation of concrete23.4 Durability of steel in concrete23.5 Recommendations for durable concrete construction23.6 References
Durability can be defined as the ability of amaterial to remain serviceable for at least therequired lifetime of the structure of which itforms a part. The specified design life can typi-cally be 50 or 100 years, but for many structuresthis is not well defined, and then the durabilityshould be such that the structure remains service-able more or less indefinitely, given reasonablemaintenance. For many years, concrete wasregarded as having an inherently high durability,but experiences in recent decades have shownthat this is not necessarily the case. Degradationcan result from either the environment to whichthe concrete is exposed, for example frostdamage, or from internal causes within the con-crete, as in alkali–aggregate reaction. It is alsonecessary to distinguish between degradation ofthe concrete itself and loss of protection and sub-sequent corrosion of the reinforcing or prestress-ing steel contained within it.
The rate of most of the degradation processesis controlled by the rate at which moisture, air orother aggressive agents can penetrate the con-crete. This penetrability is a unifying theme whenconsidering durability, and for this reason weshall first consider the various transport mechan-
isms through concrete – pressure-induced flow,diffusion and absorption – their measurementand the factors which influence their rate. Weshall then discuss the main degradation processes,firstly of concrete – chemical attack by sulfates,sea water, acids and alkali–silica reaction, andphysical attack by frost and fire – and then thecorrosion of embedded steel. This will show howpotential problems in all of these areas can beeliminated, or at least minimised, by dueconsideration to durability criteria in the designand specification of new structures. By way ofillustration, some typical recommendations areincluded. Ignorance of, or lack of attention to,such criteria in the past has led to a thriving andexpanding repair industry in recent years; it is tobe hoped that today’s practitioners will be able tolearn from these lessons and reduce the need forsuch activities in the future. It is beyond the scopeof this book to discuss inspection and repairmethods; publications on this subject are nowextensive, and a useful review is listed in the‘Further reading’ on page 222.
23.1 Transport mechanisms throughconcreteAs we have seen in Chapter 13, hardened cementpaste and concrete contain pores of varying typesand sizes, and therefore the transport of materialsthrough concrete can be considered as a particu-lar case of the more general phenomenon of flowthrough a porous medium. The rate of flow will
not depend simply on the porosity, but on thedegree of continuity of the pores and their size –flow will not take place in pores with a diameterof less than about 150nm. The term permeabilityis often loosely used to describe this general pro-perty (although we shall see that it also has amore specific meaning); Figure 23.1 illustrates thedifference between permeability and porosity.
Flow can occur by one of three distinctprocesses:
• movement of a fluid under a pressure differen-tial – i.e. permeation;
• movement of ions, atoms or molecules undera concentration gradient, i.e. diffusion;
• capillary attraction of a liquid into empty orpartially empty pores, i.e. sorption.
Each of these has an associated ‘flow constant’,defined as follows:
1. In the flow or movement of a fluid under apressure differential, the flow passagesthrough concrete and the flow rates are suffi-ciently small for the flow of either a liquid orgas to be laminar, and hence it can bedescribed by Darcy’s law:
ux � �K∂h/∂x (23.1)
where, for flow in the x-direction, ux �meanflow velocity, ∂h/∂x� rate of increase in pres-sure head in the x-direction, and K is a con-stant called the coefficient of permeability, thedimensions of which are (length)/(time), e.g.m/sec.
Transport mechanisms through concrete
193
FIGURE 23.1 Illustration of the difference betweenpermeability and porosity (Concrete Society, 1988 bypermission).
The value of K depends on both the porestructure within the concrete and the proper-ties of the permeating fluid. The latter can, intheory, be eliminated by using the intrinsicpermeability (k) given by
k�K /� (23.2)
where �coefficient of viscosity of the fluidand � �unit weight of the fluid. k has dimen-sions of (length)2, and should be a property ofthe porous medium alone and thereforeapplicable to all permeating fluids. However,for liquids, it depends on the viscosity beingindependent of the pore structure, and for hcpwith its very narrow flow channels in which asignificant amount of the water will be subjectto surface forces this may not be the case. Fur-thermore, comparison of k values from gasand liquid permeability tests has shown theformer to be between 5 and 60 times higherthan the latter, a difference attributed to theflow pattern of a gas in a narrow channel dif-fering from that of a liquid (Bamforth, 1987).It is therefore preferable to consider perme-ability in terms of K rather than k, and acceptthe limitation that the values given apply toone permeating fluid only, normally water.
2. The movement of ions, atoms or moleculesunder a concentration gradient is a process ofdiffusion, and is governed by Fick’s law:
J� �D∂C/∂x (23.3)
where, for the x-direction, J� transfer rate ofthe substance per unit area normal to the x-direction, ∂C/∂x�concentration gradient andD is a constant called the diffusivity, whichhas the dimensions of (length)2/(time), e.g.m2/sec.
Defining the diffusivity in this way treatsthe porous solid as a continuum, but thecomplex and confining pore structure withinconcrete means that D is an effective, ratherthan a true, diffusion coefficient. We are alsointerested in more than one type of diffusionprocess, for example moisture movementduring drying shrinkage, or de-icing salt dif-fusion through saturated concrete road decks.
Furthermore, in the case of moisture diffusion(in, say, drying shrinkage) the moisturecontent within the pores will be changingthroughout the diffusion process. There is,however, sufficient justification to consider Das a constant for any one particular diffusionprocess, but it should be remembered that, aswith the permeability coefficient K, it isdependent on both the pore structure of theconcrete and the properties of the diffusingsubstance.
3. Adsorption and absorption of a liquid intoempty or partially empty pores occurs by cap-illary attraction. Experimental observationsshow that the relationship between the depthof penetration (x) and the square root of thetime (t) is bi- or tri-linear (Figure 23.2), with aperiod of rapid absorption in which the largerpores are filled, being followed by moregradual absorption (Buenfeld and Okundi,1998). A constant called the sorptivity (S) canbe defined as the slope of the relationship(normally over the initial period), i.e.:
x�S.t0.5 (23.4)
As before, S relates to a specific liquid,often water. It has the dimensions of(length)/(time)0.5, e.g. mm/sec0.5.
The applicability of these mechanisms in practiceis illustrated by the case of an offshore concretestructure, which is exposed to a range of differentenvironments. Figure 23.3 shows the predomi-
Durability of concrete
194
Pene
trat
ion
dept
h (x
)
Square root of time (t)
Sorptivity (S) = x/t 0.5
FIGURE 23.2 Typical form of results from sorptiv-ity measurements.
FIGURE 23.3 Primary transport mechanisms in thevarious exposure zones of a concrete offshore structure(Concrete Society, 1988 by permission).
nant mechanisms in the respective exposurezones.
23.2 Measurements of flowconstants for cement paste andconcrete
23.2.1 Permeability
Permeability is commonly measured by subjectingthe fluid on one side of a concrete specimen to apressure head, and measuring the steady-stateflow rate that eventually occurs through the speci-men, as illustrated in Figure 23.4. The specimenis normally a circular disc, the sides of which aresealed to ensure uniaxial flow. If the fluid isincompressible, i.e. it is a liquid such as water,the pressure head gradient through the specimenis linear, and Darcy’s equation reduces to:
�Q/A� �K.�P/l (23.5)
where �Q�volumetric flow rate, A� total cross-sectional area of flow perpendicular to the z-direction, �P�pressure head and l�flow pathlength.
Much of the fundamental work on the perme-ability of cement paste to water was carried out
by Powers (1954, 1958). As the cement hydrates,the hydration products infill the skeletal struc-tures, blocking the flow channels and hence redu-cing the permeability. As might be expected fromour earlier description of cement hydration inChapter 13, the reduction of permeability is highat early ages, when hydration is proceedingrapidly. In fact, as shown in Figure 23.5, itreduces by several orders of magnitude in the first2–3 weeks after casting.
Although, as we discussed above, permeabilityand porosity are not necessarily related (Figure23.1) there is a general non-linear correlationbetween the two for cement paste, as shown inFigure 23.6. The greatest reduction in permeabil-ity occurs for porosities reducing from about 40to 25 per cent, where increased hydrationproduct reduces both the pore sizes and the flowchannels between them. Further hydrationproduct, although still reducing porosity signific-antly, results in much lower changes in the per-meability. This explains the general form ofFigure 23.5, and also accounts for the effect ofwater/cement ratio on permeability shown inFigure 23.7 for a constant degree of hydration. Atwater/cement ratios above about 0.55 the capil-lary pores form an increasingly continuoussystem, with consequent large increases in perme-ability.
It is apparent from the above arguments and
Measurements of flow constants for cement paste and concrete
195
FIGURE 23.4 Typical test system for measurementof concrete permeability under steady-state flow.
FIGURE 23.5 The effect of hydration on the perme-ability of cement paste (waste/cement�0.7) (Powers etal., 1954).
FIGURE 23.6 The relationship between permeabil-ity and capillary porosity of hardened cement paste(Powers, 1958).
from those in Chapter 13 that high strength andlow permeability both result from low porosity,and in particular a reduction in the volume of thelarger capillary pores. In general, higher strength
Durability of concrete
196
FIGURE 23.7 The relationship between the perme-ability and water/cement ratio of mature cement paste(93 per cent hydrated) (Powers et al., 1954).
FIGURE 23.8 Comparison between test results of permeability of concrete to water (Lawrence, 1985 by per-mission, Concrete Society).
implies lower permeability, although the relation-ship is not linear, and may be different for differ-ent curing histories and cement types.
The permeability of the concrete will also be
influenced by the permeability of the aggregate.Many of the rock types used for natural aggre-gates have permeabilities of the same order asthat of cement paste (Table 23.1), despite havingrelatively low porosities. Lightweight aggregates,which are highly porous, can have much higherpermeabilities. However, in practice the perme-ability of the composite concrete is often found tobe substantially higher than that of either theaggregate or the paste, as can be seen by compar-ing the data for concrete shown in Figure 23.8with those for aggregate and cement paste inTable 23.1 and Figure 23.7 respectively. This isprimarily due to the presence of defects or cracks,particularly in the weaker transition zone at thecement/aggregate interface. Larger aggregates,with larger transition zones, exaggerate the effect.
As with cement paste, similar factors controlboth the permeability and strength of the con-crete, and it is therefore possible to produce lowpermeability by attention to the same factorsrequired to produce high strength. These includeusing a low water/cement ratio (Figure 23.8) and
an adequate cement content, and ensuring propercompaction and adequate curing; in addition, theproperties of the transition zone may beimproved by the use of cement replacementmaterials (see Chapter 20), although longercuring periods are necessary to ensure continu-ance of the pozzolanic reaction. The avoidance ofmicrocracking from thermal or drying shrinkagestrains and premature or excessive loading is alsoimportant.
23.2.2 Diffusivity
The principle of diffusivity testing is relativelysimple. A high concentration of the diffusant isplaced on one side of a suitable specimen (nor-mally a disc) of hcp, mortar or concrete, and thediffusion rate calculated from the increase of con-centration on the other side. In the case of gasdiffusion, the high concentration side may be anatmosphere of the pure gas; in the case of salts, ahigh concentration aqueous solution would beused. The test is therefore similar to the perme-ability test without the complication of high pres-sure. It is generally found that, after an initialperiod for the diffusant to penetrate through thespecimen, the concentration on the ‘downstream’side increases linearly with time. The diffusivitywill change if the moisture content of the con-crete changes during the test, and so the speci-mens must be carefully conditioned beforetesting.
Control of test conditions is important, and
Measurements of flow constants for cement paste and concrete
197
TABLE 23.1 Comparison between permeabilities ofrocks and cement paste (Powers, 1958)
Type of rock Permeability Water/cement ratio(m/sec) of cement paste
of same permeability
Dense trap 2.47�10�14 0.38Quartz diorite 8.24�10�14 0.42Marble 2.39�10�13 0.48Marble 5.77�10�12 0.66Granite 5.35�10�11 0.70Sandstone 1.23�10�10 0.71Granite 1.56�10�10 0.71
TABLE 23.2 Chloride ion diffusivities of paste and concrete
Binder w/c Diffusivity (m2/sec)
Paste (Page et al., 1981)100%pc 0.4 2.6�10�12
100%pc 0.5 4.4�10�12
100%pc 0.6 12.4�10�12
70%pc�30%pfa 0.5 1.47�10�12
30%pc�70%ggbs 0.5 0.41�10�12
Concrete (Buenfeld et al., 1998)100%pc 0.4 18�10�12
100%pc 0.5 60�10�12
60%pc�40%pfa 0.4 2�10�12
25%pc�75%ggbs 0.4 2�10�12
diffusivity measurements from different test pro-grammes are not entirely consistent. Table 23.2shows values of chloride ion diffusivity that havebeen obtained on mature saturated pastes andconcrete. The beneficial effects from lowerwater/cement ratio and the use of CRMs areclear.
23.2.3 Sorptivity
Sorptivity can be calculated from measurementsof penetration depth, and tests are carried out onsamples in which penetration is restricted to onedirection only, such as cylinders with the curvedsurface sealed with a suitable bitumen or resincoating. The penetration depth at a particulartime can be measured by splitting a sample open,but this requires a considerable number ofsamples to obtain a significant number of results.Alternatively, it can be estimated from weightgain, providing the concrete’s porosity is known;this can be conveniently found by drying thespecimen after the test. Such tests can only becarried out in the laboratory, either on specimenscast for this purpose, or on cores cut from struc-tural concrete.
Values of sorptivity at various distances for thesurface of a concrete slab are shown in Figure23.9. These were obtained on slices of cores cutfrom concrete slabs 28 days old which had beenmoist cured for 4 days and then air cured for 24
days. The sorptivity decreases with depth, attri-buted to the air drying causing imperfect curingof the surface zone. However, although thesimilar strength mixes containing CRMs hadsimilar sorptivities in the 15mm thick surfacezone, they generally had lower values than theplain Portland cement concrete at greater depth,again indicating the advantages to be gained fromthese materials with sufficient curing.
A number of tests have been developed tomeasure the absorption and permeabilitycharacteristics of in situ concrete whilst still in thestructure, i.e. avoiding the need to cut cores.These all measure the penetration rate of a fluid(normally air or water) into the concrete, eitherthrough the concrete surface with the fluid con-tained in a chamber fixed to the concrete (Figure23.10(a)) or outwards from a hole drilled intothe concrete, with the fluid delivered via a smalltube or needle inserted in the hole (Figure
Durability of concrete
198
FIGURE 23.9 Variation of sorptivity with distance from cast surface of concrete made with Portland cementand cement replacement materials (Bamforth and Pocock, 1990).
FIGURE 23.10 Techniques for surface permeabilityand absorption measurements on in situ concrete.
23.10(b)). Depending on the fluid pressure,which varies from test to test, estimates are madeof either the absorption characteristics of thesurface combination of the two.
A commonly used test of this type is the Initial
Surface Absorption Test (ISAT), shown in Figure23.11. A cap is clamped to the concrete surfaceand a reservoir of water is set up with a constanthead of 200mm. The reservoir is connectedthrough the cap to a capillary tube set level withthe water surface. At the start of the test, water isallowed to run through the cap (thus coming intocontact with the concrete surface) and to fill thecapillary tube. The rate of absorption is thenfound by closing off the reservoir and observingthe rate of movement of the meniscus in the capil-lary tube. Readings are taken at standard timesafter the start of the test (typically 10 mins, 20mins, 30 mins, 1 hour and 2 hours), andexpressed as flow rate per surface area of the con-crete, e.g. in units of ml/m2/sec. The rate drops offwith time, and in general increases with the sorp-tivity of the concrete.
Typical results showing the effect ofwater/cement ratio of the concrete and the dura-tion of the initial water curing period on the 10minute ISAT value for tests carried out on con-crete 28 days old are shown in Figure 23.12. Notsurprisingly, decreasing water/cement ratio andincreased curing time both decrease the ISATvalues; the results clearly reinforce the import-ance of curing.
In common with the other tests of this type, theISAT has two main disadvantages. First, the
Degradation of concrete
199
FIGURE 23.11 Initial Surface Absorption Testsystem (Concrete Society, 1988 by permission).
results depend on the moisture state of the con-crete at the start of the test, which is particularlydifficult to control if the test is carried out in situ.Second, the flowpath of the fluid through the con-crete is not unidirectional but diverges; a funda-mental property of the concrete is therefore notmeasured, and it is difficult to compare resultsfrom different test systems.
However, the tests all measure some propertyof the surface layers of the concrete and, as weshall see, this is all important in ensuring gooddurability.
23.3 Degradation of concreteThe degradation agencies that affect concrete canbe divided into two broad groups:
1. those whose action is initially chemical, beforesubsequently leading to loss of physicalintegrity, including sulfates, sea water, acidsand alkali–silica reactions;
2. those which directly lead to physical effects,such as frost and fire.
Each of these is now discussed.
FIGURE 23.12 Effect of water/cement ratio andinitial curing on surface absorption measured by theISAT test (Dhir et al., 1987 by permission, ThomasTelford Ltd).
23.3.1 Attack by sulfates
We have seen in Chapter 13 that a controlledamount of calcium sulfate, in the form ofgypsum, is added to Portland cement during itsmanufacture to control the setting process.Further sulfates can arise from contaminatedaggregates, a particular problem in some MiddleEastern countries, or from ground-water contain-ing sulfates from clay soils, fertilisers or industrialeffluent coming into contact with concrete infoundations, retaining walls, etc. Sodium, potas-sium, magnesium and calcium sulfates are allcommon; when these are in solution they canattack the hardened cement paste. We brieflydescribed the nature of the problem when dis-cussing sulfate-resisting Portland cement inChapter 13; specifically, the sulfates and thehydrated aluminate phases in the hardenedcement paste react to form ettringite, accordingto equation (13.12), which in its full chemicalnotation is::
3CaO.Al2O3.12H2O�3CaSO4 �20H2O→3CaO.Al2O3.3CaSO4.32H2O (23.6)
This is an expansive reaction, with the solidphases more than trebling in volume, causing dis-ruption.
With sulfates other than calcium sulfate, reac-tions can also occur with the calcium hydroxidein the hcp, forming gypsum, again with anincrease in volume, and a loss of stiffness andstrength of the paste, thereby increasing thedegradation. For example, the reaction withsodium sulfate is:
Ca(OH)2 �Na2SO4.10H2O→CaSO4.2H2O�2NaOH�8H2O (23.7)
With magnesium sulfate, a similar reaction takesplace, but the magnesium hydroxide formed isrelatively insoluble and poorly alkaline; thisreduces the stability of the calcium silicatehydrate which is also attacked:
3CaO.2SiO2.3H2O�3(MgSO4.7H2O)→3(CaSO4.2H2O)�3(Mg(OH)2)�2SiO2.aq (23.8)
Thus the severity of attack depends on the type of
sulfate; magnesium sulfate is more damaging thansodium sulfate, which, in turn, is more damagingthan calcium sulfate. In each case, attack occursonly when the amount of sulfates present exceedsa certain threshold; the rate of attack thenincreases with increasing sulfate, but at a redu-cing rate of increase above about 1 per cent con-centration. Also, the rate of attack will be faster ifthe sulfates are replenished, for example if theconcrete is exposed to flowing groundwater.
Concrete which has been attacked has awhitish appearance; damage usually starts atedges and corners, followed by progressive crack-ing and spalling, eventually leading to completebreakdown. Although this stage can be reached ina few months in laboratory tests, it normallytakes several years in the field.
For any given concentration and type ofsulfate, the rate and amount of the deteriorationdecreases with:
• the C3A content of the cement, hence the lowC3A content of sulfate-resisting Portland cement;
• higher cement content and lower water/cement ratio of the concrete; higher qualityconcrete is less vulnerable due to its lowerpermeability. This is a more significant factorthan the C3A content, as shown in Figure23.13;
• the incorporation of cement replacementmaterials, which can decrease the permeabil-ity, reduce the amount of free lime in the hcp,and effectively ‘dilute’ the C3A.
Recommendations for suitable concrete for use insulfate-containing environments follow from adetailed knowledge of the above factors. As anexample, Table 23.3 is an extract of the require-ments in the UK (BRE, 1996). Low sulfate levelsdo not require any special considerations overand above those for all concrete construction.With increasing sulfate levels a number of combi-nations of sulfate resistance of the binder andoverall quality of the concrete are possible, withfewer options with high sulfate levels. In the mostsevere class 5, the concrete itself cannot be madesufficiently durable, and some form of surfaceprotection or coating is also required.
Durability of concrete
200
Degradation of concrete
201
FIGURE 23.13 The effect of C3A content of cementand cement content of concrete on deterioration in asoil containing 10 per cent Na2SO4 (Verbeck, 1968).
TABLE 23.3 Extract of the requirements for well-compacted concrete exposed to a permeable sulfate soil orfill (BRE, 1996, ©BRE, by permission)
Concentration inMinimum Maximumgroundwater
cement content water/cementClass SO4 (%) Mg (%) Cement type (kg/m3) ratio
1 �0.4 – any 275 0.652 0.4–1.4 – Portland or Portland blended with moderate 330 0.50
amounts of ggbs or pfaSulfate-resisting 280 0.55Portland blended with 74–90% ggbs or 300 0.5525–40% pfa
3 1.4–3.0 – Sulfate-resisting 320 0.50Portland blended with 74–90% ggbs or 340 0.5025–40% pfa
4 3.0–6.0 �1 Sulfate-resisting 360 0.45Portland blended with 74–90% ggbs or 380 0.4525–40% pfa
�1 Sulfate-resisting 360 0.455 �6.0 �1 as for class 4 plus surface protection
�1
23.3.2 The thaumasite form of sulfateattack
This form of attack, known as TSA for short,also involves sulfates, but has distinct differences
from the sulfate attack described above, and canhave more serious consequences.
Thaumasite is a rare mineral that occursnaturally in some basic rocks and limestones. Itis a compound of calcium silicate, carbonateand sulfate with the formula CaSiO3.CaCO3.CaSO4.15H2O. To be formed in concrete andmortar it requires:
• a source of calcium silicate, clearly availablefrom the hydrated or unhydrated Portlandcement;
• a source of sulfate ions, for example from soilor groundwater;
• a source of carbonate, usually in the aggregateor inert fillers, or less commonly from carbondioxide dissolved in the surrounding groundwater; and
• a very wet, cold (below 15°C) environment.
Clearly all of these requirements do not oftenoccur together, the most common case where theydo being in concrete made with limestone aggreg-ate used for foundations of structures in sulfate-bearing soils in temperate or cold climates.Because the attack involves the calcium silicate
hydrates, it can lead to complete disintegration ofthe cement paste, which turns into a soft, white,mushy mass.
Incidents of attack are not widespread, themost notable involving buried concrete in houseand bridge foundations in the west of England,column building supports in Canada, tunnellinings, sewage pipes and road sub-bases. Theincidents in England came to light in the early1990s and led to a major investigation. TSA hasalso been found in brickwork where the Portlandcement render or mortar has been in contact withsulfate-bearing clay bricks in prolonged wet, coldconditions; the source of the carbonate wasthought to be atmospheric carbon dioxide dis-solved in surface run-off water (ThaumasiteExpert Group, 1999)
Research into the mechanisms, rates and avoid-ance of TSA is incomplete, but the followingfactors are apparent:
• As with other forms of attack involving agentsexternal to the concrete, it generally starts atthe surface and works inwards at a ratedepending on the composition and quality ofthe concrete and the curing history, all ofwhich, as we have seen, effect the permeabil-ity and porosity.
• Although the ultimate outcome of TSA iscomplete loss of strength, the deteriorationwill be relatively slow and, even though thismay be in buried foundations, visual signs ofdistress to a structure should become apparentlong before its integrity is significantlyimpaired.
• The expansive disruption associated withother forms of sulfate attack may or may notaccompany the formation of thaumasite.
• Thaumasite can form in pre-existing voidsand cracks in the concrete along with ettrin-gite and gypsum, or on its own, but this is notnecessarily disruptive. Disruption only occurswhen the thaumasite replaces some or all ofthe hardened cement matrix.
• TSA has occurred in concrete which has beendesigned to resist the other forms of sulfateattack described above. The specifications for
this are therefore not sufficient for TSA avoid-ance.
• In assessing the risk from groundwater andsoils, account should be taken of the sulfide aswell as the sulfate content. Sulfides oftenoccur as pyrite (FeS2), which when exposed toatmospheric conditions, when the soil is dis-turbed during excavation and backfill, willoxidise to form more sulfates, thus increasingthe risk of attack.
• An acceptable threshold level of calcium car-bonate within either the coarse or fine aggreg-ate has not been rigorously defined, but thelikelihood of TSA is reduced when the CaCO3
content of the total aggregate is reduced to alevel between 10 per cent and 30 per cent,depending on whether this is in the fine orcoarse aggregate fraction respectively.
• A safe limit for the amount of carbonatematerial within a cement such as Portlandlimestone cement is difficult to define, but it isconsidered that the 5 per cent addition offinely divided limestone allowed in Portlandcement will not be detrimental.
• Some concrete made with sulfate-resistingPortland cement is more resistant to TSA thanothers, but C3A is not involved in the reactionand the reason for this is not clear. In labora-tory tests, concrete made with a blendedcement of 70 per cent ggbs and 30 per centPortland cement has given promising results.
The result of a survey of an extreme case of TSAwhich illustrates the distinctive features of thedegradation are shown in Figure 23.14. This wason a high-quality concrete (cement content370kg/m3, w/c about 0.5, estimated in situ cubestrength greater than 60MPa) in a 29-year-oldburied highway bridge column, made with amixture of dolomitic and oolitic limestone coarseaggregate and a quartz sand, and surroundedwith a reworked clay with mobile ground water.As can be seen, there are four zones of increas-ingly severe attack, with a total depth of 25mm,and the sulfate concentration in the surface zonewas about twenty times that in the undamagedconcrete. However, the TSA was not in itself the
Durability of concrete
202
greatest threat to the structural durability. Chlo-rides from the highway drainage system were ableto penetrate the reduced cover to the reinforcingsteel and initiate steel corrosion. This subject isdiscussed in detail later in this chapter.
Recommendations for minimising the risk ofTSA are limited by the lack of complete informa-tion in some areas, as mentioned above. Thoseresulting from the UK investigation into TSA(Thaumasite Expert Group, 1999) are based on
those for the other forms of sulfate attack dis-cussed above, and include:
• classification of the soil and groundwater con-ditions in a similar way, but taking particularaccount of:– the whole ground profile to the full depth of
the concrete penetration,– the amount of sulfides as well as sulfates,– the presence and mobility of groundwater;
Degradation of concrete
203
Effectivedepth of cover
to steel
Zone 1needles of
thaumasite andoccasional
ettringite withinvoids
Zone 2thin cracksinfilled withthaumasite
Zone 3wide cracks filled
with thaumasite andhaloes of thaumasite
around aggregateparticles
Zone 4thaumasite and
aggregateparticles only –mushy mass
Reinforcingsteel
Unaffectedconcrete
Sulfates inground water
from oxidation ofpyrite in clay
Concretesurface
5 mm 7 mm 8 mm 5 mm
Sul
fate
con
tent
(% b
y w
t ce
men
t) 40
30
20
10
FIGURE 23.14 Typical effects of severe thaumasite attack in high-quality concrete after 29 years’ exposure(Thaumasite Expert Group, 1999, HMSO. Crown copyright reproduced with the permission of the Controller ofHer Majesty’s Stationery Office).
• when carbonate-containing aggregates with apotential for contributing to TSA are used insulfate classes 3 to 5 (see above), they areranked according to their carbonate content,and enhanced concrete quality is required forthe higher rankings. The ranking system takesaccount of the greater contribution of carbon-ates in the fine aggregate fractions.
23.3.3 Sea water attack
Concrete in sea water is exposed to a number ofpossible degradation processes simultaneously,including the chemical action of the sea salts,wetting and drying in the tidal zones and justabove, abrasion from waves and water-borne sedi-ment and, in some climates, freezing andthawing.
The total soluble salt content of sea water inthe oceans is typically about 3.5 per cent byweight, the principal ionic contributors and theirtypical amounts being 2.0 per cent Cl�, 1.1 percent Na�, 0.27 per cent SO4
�, 0.12 per cent Mg��
and 0.05 per cent Ca��. The action of the sulfatesis similar to that of pure sulfate solutionsdescribed above, with the addition of some inter-active effects. Importantly, the severity of theattack is not as great as for a similar concentra-tion of sulfate acting alone and there is littleaccompanying expansion. This is due to the pres-ence of chlorides; gypsum and ettringite are moresoluble in a chloride solution than in pure water,and therefore tend to be leached out of the con-crete by the sea water, and their formation doesnot, therefore, result in expansive disruption. Themagnesium ions also participate in the reactionsas in equation (23.8), and a feature of sea waterdamaged concrete is the presence of whitedeposits of Mg(OH)2, often called brucite. Inexperiments on concrete permanently saturatedwith sea water, a form of calcium carbonatecalled aragonite has also been found, arising fromthe reaction of dissolved carbon dioxide withcalcium hydroxide. The brucite and aragonite canhave a pore-blocking effect, effectively reducingthe permeability of the concrete near the surface(Buenfeld and Newman, 1984).
As illustrated in Figure 23.3, the transport ofsalts into concrete in a marine structure, or theirleaching from it, may be a permeability, diffusionor absorption controlled process. In the areassubject to wetting and drying cycles, salts willcrystallise as the water evaporates, which canlead to high salt concentrations and to disrup-tion from the pressure exerted by the crystals asthey rehydrate and grow during subsequentwetting/drying cycles – a process known as saltweathering. This can be compounded by damagefrom freeze–thaw cycles or wave action, depend-ing on the environment. These areas thereforetend to be more vulnerable.
The key to elimination, or at least reduction, ofall of these problems is, not surprisingly, the useof a low permeability concrete, perhaps combinedwith some limits on the C3A content of thecement, or the use of cement replacement mater-ials. However, for the reasons given above, thedegradation processes in many climates do notcause rapid deterioration, which explains whyconcrete of even relatively modest quality has along and distinguished history of use in marinestructures, both coastal and offshore.
The salts in sea water can contribute to twoother, potentially much more critical, degradationprocesses, namely alkali–aggregate reaction andcorrosion of embedded steel. Both are discussedlater.
23.3.4 Acid attack
We have seen that the hardened cement pastebinder in concrete is alkaline, and therefore noPortland cement concrete can be considered acidresistant. However, it is possible to produce aconcrete which is adequately durable for manycommon circumstances by giving attention to lowpermeability and good curing. In these circum-stances, attack is only considered significant if thepH of the aggressive medium is less than about 6.
Examples of acids that commonly come intocontact with concrete are dilute solutions of CO2
and SO2 in rain water in industrial regions, andCO2 and H2S-bearing groundwater from moor-lands. The acids attack the calcium hydroxide
Durability of concrete
204
within the cement paste, converting it, in the caseof CO2, into calcium carbonate and bicarbonate.The latter is relatively soluble, and leaches out ofthe concrete, destabilising it. The process is thusdiffusion controlled, and progresses at a rateapproximately proportional to the square root oftime. The C–S–H may also be attacked, as cancalcareous aggregates such as limestone. The rateof attack increases with reducing pH.
As mentioned above, the quality of the con-crete is the most important factor in achievingacid resistance, but well-cured concretes contain-ing cement replacement materials also havegreater resistance due to the lower calciumhydroxide content as a result of the pozzolanicreaction. In cases where some extra acid resis-tance is required, such as in floors of chemicalfactories, the surface can be treated with dilutedwater glass (sodium silicate), which reacts withthe calcium hydroxide forming calcium silicates,blocking the pores. In more aggressive conditions,the only option is to separate the acid and theconcrete by, for example, applying a coating ofepoxy resins or other suitable paint systems to theconcrete.
23.3.5 Alkali–aggregate and alkali–silicareaction
We described the general nature and compositionof natural aggregates in Chapter 16. Amongmany other constituents, they may contain silica,silicates and carbonates, which in certain mineralforms can react with the alkaline hydroxides inthe pore water derived from the sodium andpotassium oxides in the cement. The mostcommon and important reaction involves activesilica, and is known as alkali–silica reaction(ASR). The product is a gel which can destroy thebond between the aggregate and the hardenedcement paste, and which absorbs water andswells to a sufficient extent to cause cracking anddisruption of the concrete.
For the reaction to occur, both active silica andalkalis must be present. In its reactive form, silicaoccurs as the minerals opal, chalcedony, crysto-balite and tridymite and as volcanic glasses.
These can be found in some flints, limestones,cherts and tuffs. The sources of such aggregatesinclude parts of the USA, Canada, South Africa,Scandinavia, Iceland, Australia, New Zealandand the midlands and south west of England.Only a small proportion of reactive material inthe aggregate (as low as 0.5 per cent) may benecessary to cause disruption to the concrete.
In unhydrated cement, sodium and potassiumoxides (Na2O and K2O) are present in small butsignificant quantities (see Table 13.1), either assoluble sulphates (Na2SO4 and K2SO4) or a mixedsalt (Na, K)2SO4. There is also a small amount offree CaO, which is subsequently supplemented byCa(OH)2 from the hydration reactions of C3S andC2S. During hydration, these sulfates take part ina reaction with the aluminate phases in a similarway to gypsum (see Section 13.1.3), the productagain being ettringite with sodium, potassiumand hydroxyl ions going into solution:
(Na, K)2SO4 � 3CaO.Al2O3 �Ca(OH)2 →(C3A)
3CaO.Al2O3 .3CaSO4.32H2O�(ettringite)
Na�, K�, OH� (23.9)(in solution)
The resulting pH of the pore water is 13–14,higher than that of saturated calcium hydroxidesolution alone. Alkalis may also be contributedby some admixtures, by pfa and ggbs and byexternal sources such as aggregate impurities, seawater or road de-icing salts.
The reactions between the reactive silica andthe alkalis which forms the alkali–silicate geloccur first at the aggregate/cement paste interface.They require the presence of water, and, it isbelieved, Ca�� ions. The nature of the gel iscomplex, but it is clear that it is a mixture ofsodium, potassium and calcium silicates. It is soft,but imbibes a large quantity of water by osmosisand swells considerably. The hydraulic pressurethat is developed leads to overall expansion of theconcrete and can be sufficient to cause crackingof the aggregate particles, the hcp and the transi-tion zone between the two.
Degradation of concrete
205
Continued availability of water causes enlarge-ment and extension of the cracks which eventu-ally reach the outer surface of the concrete,forming either pop-outs if the affected aggregateis close to the surface, or more extensive crazing,or map cracking, on the concrete surface, as illus-trated in Figure 23.15. These surface cracks areoften highlighted by staining from the soft geloozing out of the cracks. In general, the crackingadversely affects the appearance and serviceabil-ity of structure before reducing its load-carryingcapacity.
Durability of concrete
206
FIGURE 23.15 Typical cracking patterns resultingfrom alkali–silica reaction.
FIGURE 23.16 Effect of acid-soluble alkali content ofconcrete on expansion and cracking after 200 days fromalkali–silica reactions (tests at critical silica/alkali ratio)(Hobbs, 1986 by permission, Thomas Telford Ltd).
The whole process is often very slow, and thecracking can take years to develop in structuralconcrete. A description was first published in theUSA in 1940 (Stanton) since when numerousexamples have been reported in many countries.Over 100 cases were identified in the UK between1976 and 1987, triggering much research aimedat understanding and quantifying the mechanismsinvolved, determining its effect on structuralperformance and providing guidance for minimis-ing the risk in new concrete. The latter can beconsidered successful as there have been no newconfirmed cases of ASR in the UK since 1987.
Even though laboratory tests have sometimesnot satisfactorily explained all field observations,the most important factors influencing theamount and rate of reaction can be summarisedas follows:
• The amount of alkalis available, which is nor-mally expressed as the total weight of sodiumoxide equivalent�Na2O�0.65 K2O. Thereaction rate and amount increase withincreasing alkali level, but test data such asthose shown in Figure 23.16 indicate thatthere is a threshold level below which no dis-ruption will occur, due to the lowering of thepH of the pore water, even with reactiveaggregates. This level is typically about3.5–4kg/m3 of concrete, which corresponds toa lower limit of about 0.6 per cent by weightof cement. Cements with a sodium oxideequivalent of less than this are called low-alkali cements.
• The presence of cement replacement materials,which reduce the rate of reaction through twomain mechanisms:– They remove Ca(OH)2 (see Chapter 15),
and hence reduce the amount of Ca�� ionsin the pore water which are necessary forthe reaction to take place.
– They reduce the permeability of the con-crete and therefore reduce the mobility ofthe aggressive agents, i.e. the same generaleffect as on other degradation processes.
• The amount of reactive silica. Typical effectsfrom tests on mortars are shown in Figure
23.17. The expansion increases with activesilica content, but only up to a certaincontent, beyond which the expansion reduces.There is thus a pessimum content of silica formaximum expansion, in this case about 1 percent, but which is higher for lowerwater/cement ratios and higher cement con-tents. The following explanation for thebehaviour in the four regions shown in Figure23.17 has been proposed (Hobbs, 1988):– Region A: the reactive silica content is low
and gel growth after the concrete has hard-ened is insufficient to cause cracking.
– Region B: the reaction after the concretehas hardened is sufficient to cause cracking.There is an excess of alkalis, and the reac-tion continues until all the active silica hasbeen used up.
– Region C: there is an excess of silica overalkalis, and the reaction continues until allthe alkalis have been used up or their con-centration falls below the threshold level.The uptake of water by the reactionproduct decreases with decreasing alkali/reactive silica ratio.
– Region D: the reactive silica content is sohigh and the reaction is so rapid that by thetime the concrete has hardened the rate ofgel growth is too slow to induce cracking.
Degradation of concrete
207
FIGURE 23.17 Effect of active silica content onexpansion of mortars due to alkali–silica reaction(opaline silica, w/c�0.53, A/c�3.75, Na2O�4.4kg/m3) (Hobbs, 1988).
At the pessimum point the amount of reactivesilica is just sufficient to react with all thealkalis present; the reactive silica/alkali ratiousually lies in the range 3.5 to 5.5.
• The aggregate particle size, which affects theamount of reactive silica exposed to the alkali;fine particles (20 to 30�m) can lead to expan-sion within a few weeks, larger ones only aftermany years.
• The availability of moisture. The gel swellingwill cease if the relative humidity within theconcrete, which depends on the environmentand the concrete permeability, falls belowabout 85 per cent. Alternative wetting anddrying may be the most harmful, possiblybecause it can lead to local high concentra-tions of the reacting materials.
• The ambient temperature. Higher tempera-tures accelerate the reaction, at least up to40°C.
Once started, the only effective way of stoppingASR is by eliminating water, which is clearlyimpractical in many structural situations. Itfollows that it is important to reduce or eliminatethe risk of ASR occurring by careful materialsselection and concrete mix design. Considerationof the factors influencing the occurrence and rateof ASR described above leads to the followingpossibilities:
• Avoiding the use of reactive aggregates. Thisis more difficult than it sounds, particularlywith mixed mineral aggregates. There is nouniversally agreed test for aggregate reactivity,and recent guidance notes (Concrete Society,1999) suggest classifying the aggregate aseither low, normal or high reactivity depend-ing on its composition.
• Limiting the amount of alkalis in the cement,for example by using a low-alkali cement, i.e.with alkali content of less than 0.6 per cent byweight, as discussed above.
• Combining the Portland cement with acement replacement material, which can be ofbenefit in three ways:– although pfa and ggbs themselves contain
alkalis, the extent to which these contribute
to the total alkalinity of the pore waterappears to be small when they are com-bined with a high alkalinity Portlandcement, and only a proportion of theiralkali content needs be taken into account(figures of 17% pfa and 50% for ggbs havebeen suggested) and therefore the alkalicontent of the binder may be reduced;
– they reduce the permeability of the con-crete, thus reducing the rates or reaction;
– they contain divided reactive silica, andtherefore, if added in adequate quantitieswill increase the total reactive content towell above the pessimum value, e.g. intoregion D of Figure 23.17, thereby reducingor eliminating expansion. However, suffi-cient quantities must be added to ensurethat the pessimum level is exceeded by awide margin, and minimum replacementlevels of 30–40 per cent pfa, 20 per centmicrosilica and 50–65 per cent ggbs havebeen suggested.
However, the exact mechanisms and quantita-tive nature of the role of cement replacementmaterials are complex and unclear, and arethe subject of continuing research.
• Limiting the total alkali content of the con-crete. Alkalis from all sources – cement,CRMs (but taking account of the reductionfactors discussed above), de-icing salts, etc.should total less than 3.0kg/m3 of concrete.
• Ensuring that the concrete remains drythroughout its life – obviously difficult orimpossible in many structures.
Much more detailed guidance has recently beenpublished (Concrete Society, 1999).
23.3.6 Frost attack – freeze–thaw damage
In cold climates, frost attack is a major cause ofdamage to concrete unless adequate precautionsare taken. We discussed this briefly when consid-ering air-entraining agents in Chapter 14. Whenfree water in the larger pores freezes, it expandsby about 9 per cent and, if there is insufficientspace within the concrete to accommodate this,
then potentially disruptive internal pressures willresult. Successive cycles of freezing and thawingcan cause progressive and cumulative damage,which takes the form of cracking and spalling,initially of the concrete surface.
It is the water in the larger capillary poresand entrapped air voids that has the criticaleffect; the water in the much smaller gel pores(see Chapter 13) is adsorbed on to the C–S–Hsurfaces, and does not freeze until the tempera-ture falls to about �78°C. However, after thecapillary water has frozen it has a lower ther-modynamic energy than the still-liquid gelwater, which therefore tends to migrate to sup-plement the capillary water, thus increasing thedisruption. The disruptive pressure is alsoenhanced by osmotic pressure. The water in thepores is not pure, but is a solution of calciumhydroxide and other alkalis, and perhaps chlo-rides from road de-icing salts or sea water; purewater separates out on freezing, leading to saltconcentration gradients and osmotic pressureswhich increase the diffusion of water to thefreezing front.
The magnitude of the disruptive pressuredepends on the capillary porosity, the degree ofsaturation of the concrete (dry concrete willclearly be unaffected) and the pressure relief pro-vided by a nearby free surface or escape bound-ary. The extent of this pressure relief will dependon:
1. the permeability of the material;2. the rate at which ice is formed; and3. the distance from the point of ice formation to
the escape boundary. In saturated cementpaste, the disruptive pressures will only berelieved if the point of ice formation is withinabout 0.1mm of an escape boundary. A con-venient way of achieving this is by use of anair-entraining agent (see Chapter 14), whichentrains air in the form of small discretebubbles and an average spacing of about0.2mm is required.
As we saw in Chapter 13, the capillary porosityof a cement paste or concrete, and hence its sus-ceptibility to frost attack, can be reduced by low-
Durability of concrete
208
ering the water/cement ratio and ensuring that byproper curing the hydration is as complete as pos-sible. Bleeding, which results in local high poros-ity zones, should also be minimised. Thecombined effects of air entrainment andwater/cement ratio are illustrated in Figure 23.18.
Certain aggregates are themselves susceptibleto frost action, and their use must be avoided if adurable concrete is to be achieved. The first signof damage caused by aggregate disruption is nor-mally pop-outs on the concrete surface. Vulner-able aggregates include some limestones andporous sandstones; these generally have highwater absorption, but other rocks with highabsorption are not vulnerable. Similar argumentsof pore size and distribution for cement pasteapply to the aggregates; for example, it has beenfound that pores of about 4 to 5�m are critical,since these are large enough to permit water toenter but not large enough to allow dissipation ofdisruptive pressure. Aggregate size is also afactor, with smaller particles causing less disrup-tion, presumably because the average distance toan escape boundary on the aggregate surface isless. The only satisfactory way of assessing an
Degradation of concrete
209
FIGURE 23.18 The effect of air entrainment andwater/cement ratio on the frost resistance of concretemoist-cured for 28 days (US Bureau of Reclamation,1955).
FIGURE 23.19 The effect of temperature andaggregate type on the compressive strength of concreteheated and tested hot (average initial strength�28MPa) (Abrams, 1971).
aggregate is by its performance when incorpo-rated in concrete, using field experience or labo-ratory testing.
23.3.7 Fire resistance
Concrete is incombustible and does not emit anytoxic fumes when exposed to high temperatures.It is thus a favoured material, both in its ownright and as protection for steelwork, when struc-tural safety is being considered. However,although it can retain some strength for a reason-able time at high temperatures, it will eventuallydegrade. The rate and amount of degradationdepends on the maximum temperature, theperiod of exposure, the induced temperature gra-dients, the concrete constituents and moisturecontent and the size of the element, and willtherefore vary considerably.
Figure 23.19 shows typical results of testingsmall elements by holding them at elevated tem-peratures for a reasonable period of time. Fortemperatures up to about 500°C the strengthreduction is relatively gradual, but thereafter thedecline is more rapid, giving almost total lossapproaching 1000°C. There are three main con-tributions to the degradation:
1. Evaporation of water within the concrete,which starts at 100°C, and continues withprogressively more tightly held water beingdriven off. If the concrete is initially saturatedand also of low permeability, then the steamcannot disperse quickly, and build-up of pres-sure can lead to cracking and spalling. This istherefore a particular problem with high-strength, low-porosity concrete. Even thoughthe total volume of water in the concrete islow, the induced pressures are very high, andprogressive explosive spalling of the surfacelayers can occur within a few minutes ofexposure to the fire. The inclusion of poly-propylene fibres in the mix is one way of over-coming this effect; these rapidly melt andprovide pressure relief channels.
2. Differential expansion between the hcp andaggregate, resulting in thermal stresses andcracking, initiated in the transition zone. Thisis mainly responsible for the more rapid lossof strength above about 500°C, and alsoexplains the superior performance of the lime-stone and lightweight aggregate concrete; theformer has a coefficient of thermal expansioncloser to that of the hcp (see Section 15.3) andthe latter is less stiff and hence the thermalstresses are lower. Lightweight aggregateshave the additional advantage of decreasingthe thermal conductivity of the concrete, thusdelaying the temperature rise in the interior ofa structural member.
3. Breakdown of the hydrates in the hcp, whichis not complete until the temperatureapproaches 1000°C, but results in a total lossof strength at this point.
23.4 Durability of steel in concreteNearly all structural concrete contains steel,either in the form of reinforcement to compensateprimarily for the low tensile and shear strength ofthe concrete, or as stressed pretensioned tendonswhich induce stresses in the concrete to opposethose due to the subsequent loading. Sound con-crete provides an excellent protective medium forthe steel, but this protection can be broken down
in some circumstances, leaving the steel vulner-able to corrosion. Crucially, the corrosion pro-ducts – rust in various forms – occupy aconsiderably greater volume than the originalsteel. Rusting in the concrete can therefore causecracking and spalling of the concrete covering thesteel, leading to more rapid corrosion of theexposed steel and eventual loss of structuralintegrity.
Although the processes involved are lesscomplex than those of the various degradationmechanisms of the concrete itself, describedabove, they are much more difficult to avoid andcontrol. Corrosion of steel in concrete is thegreatest threat to the durability and integrity ofconcrete structures in many countries. In the lastfew decades the concrete repair industry hasbenefited considerably and is thriving.
In this section we shall first describe the generalnature of the phenomenon, and then consider thefactors that control its onset and subsequent rate.
23.4.1 General principles of the corrosionof the steel in concrete
The electrochemical nature of the corrosion ofsteel was described in Chapter 11, but it is worthsummarising the main reactions here:
at the anode: Fe→Fe�� �2e (23.10)
electron conductance:2e (anode)→2e (cathode) (23.11)
at the cathode:4e�O2 �2H2O→4OH� (23.12)
near the surface:Fe�� �2OH� →Fe(OH)2
ferrous hydroxide, black rust (23.13)
followed by:4Fe(OH)2 �O2 →2Fe2O3.H2O�H2O
ferric hydroxide, red rust (23.14)
For iron or steel rusting in oxygenated water ormoist air, the water on or near the metal surfaceacts as the electrolyte of the corrosion cell, andthe anode and cathode are close together, e.g.
Durability of concrete
210
across a single crystal or grain. The oxide isformed and deposited near but not directly on themetal surface, as illustrated in Figure 23.20,allowing the corrosion to be continuous. In con-crete, different conditions prevail. The electrolyteis the pore water in contact with the steel, and, aswe have seen, this is normally highly alkaline(pH�12.5–13) due to the Ca(OH)2 from thecement hydration and the small amounts of Na2Oand K2O in the cement. In such a solution, theprimary anodic product is not Fe�� as in reaction(23.10) but is Fe3O4, which is deposited at themetal surface as a tightly adherent thin film, andstifles any further corrosion. The steel is said tobe passive, and thus sound concrete provides anexcellent protective medium. However, the pas-sivity can be destroyed by either:
1. a loss of alkalinity by carbonation of the con-crete, in which the calcium hydroxide is neu-tralised by carbon dioxide from the air,producing calcium carbonate; or
2. chloride ions, e.g. from road de-icing salts orsea water.
Either of these can therefore create conditions forthe corrosion reactions (23.10) to (23.13) or(23.14) to occur. The corrosion can be localised,for example in load-induced cracks in the con-crete, leading to pitting, or the corrosion cells canbe quite large, for example if anodic areas havebeen created by penetration of chlorides into alocally poorly compacted area of concrete.However, it is important to remember that
Durability of steel in concrete
211
FIGURE 23.20 Spacial arrangement of corrosionreactions of iron in moist air or oxygenated water.
oxygen and water must still be available at thecathode to ensure reaction (23.12) continues.
As mentioned above, the corrosion products(ferric and ferrous hydroxide) have a much largervolume than the original steel, by about two tothree times, and hence lead to bursting pressuresin the concrete and, eventually, cracking. Thisdamage can take various forms, as illustrated in Figure 23.21. The steel is then completelyexposed, and the corrosion can be very fast anddestructive.
Since the carbon dioxide or chlorides will nor-mally have to penetrate the concrete cover beforethe corrosion can be initiated, the total time toconcrete cracking will consist of two stages, illus-trated in Figure 23.22:
FIGURE 23.21 Different forms of damage fromsteel corrosion (Browne, 1985).
FIGURE 23.22 The two stages of corrosiondamage: t0 � time to initiation of corrosion, t1 � timefor sufficient corrosion to crack the concrete cover(Browne, 1983).
1. the time (t0) for the depassivating agents (thecarbon dioxide or chlorides) to reach the steeland initiate the corrosion;
2. the time (t1) for the corrosion to then reachcritical levels, i.e. sufficient to crack the con-crete, which depends on the subsequent corro-sion rate.
The processes of carbonation-induced corrosionand chloride-induced corrosion are now con-sidered separately.
23.4.2 Carbonation-induced corrosion
We discussed carbonation and its associatedshrinkage in Chapter 19. Atmospheric carbondioxide, when dissolved in the pore water in con-crete, reacts rapidly with the calcium hydroxideproduced during the cement hydration, formingcalcium carbonate, and reducing the pH from 12or more to about 8. There are also some reactionsbetween the carbon dioxide and the otherhydrates, but these are not significant in thiscontext.
The carbonation reaction occurs first at thesurface of the concrete and then progressesinwards, further supplies of carbon dioxide dif-fusing through the carbonated layer. Extensiveanalysis by Richardson (1988) showed that thecarbonation depth (x) and time (t) are related bythe simple expression
x�k.t0.5 (23.15)
where k is a constant closely related to the dif-fusion characteristics of the concrete. The form ofthis equation is the same as that of equation(23.4), which indicates that carbonation may beconsidered as a sorption process. The value of kdepends on several factors, chiefly:
1. The degree of saturation of the concrete. It isnecessary for the carbon dioxide to be dis-solved in the pore water, and so concretewhich has been dried by storing at low rela-tive humidities will not carbonate. At theother extreme, the diffusion will be slow inconcrete completely saturated with water, andso the fastest advance of the carbonation front
Durability of concrete
212
FIGURE 23.23 The relationship between carbona-tion depth and concrete strength (Nagataki et al.,1986).
occurs in partially saturated concrete at rela-tive humidities of between 50 and 70 per cent.Thus concrete surfaces which are shelteredwill carbonate faster than those exposed todirect rainfall.
2. The pore structure of the concrete. Parrott(1987) suggested that relating carbonationdepth to concrete strength, as in Figure 23.23,is a useful way of combining the effects ofwater/cement ratio, cement content and incor-poration of cement replacement materials.Adequate curing at early ages is also animportant factor. Although cement replace-ment materials can result in lower overallporosity with full curing, the pozzolanic reac-tion can also reduce the calcium hydroxidecontent before carbonation, and so they donot necessarily have the same benefits as theydo with other degradation processes.
3. The carbon dioxide content of the environ-ment.
Observed rates of carbonation, such as thoseshown in Figure 23.23, are such that with highquality, well-cured concrete the carbonatedregion, even after many years’ exposure tonormal atmospheric conditions, is restricted towithin 20 to 30mm of the concrete surface. It is
difficult to estimate or predict the rate of corro-sion once the steel has been depassivated, andtherefore design recommendations are aimed atensuring that the depth and quality of concretecover are sufficient to achieve a sufficiently longinitiation period, t0. It should also be noted thatcarbonation is not entirely detrimental. Thecalcium carbonate formed occupies a greatervolume that the calcium hydroxide, and so theporosity of the carbonated zone is reduced,increasing the surface hardness and strength, andreducing the surface permeability.
23.4.3 Chloride-induced corrosion
There are four common sources of the chlorides:
1. calcium chloride, a cheap and effective accel-erator (see Chapter 14);
2. contamination in aggregates;3. sea water, for coastal or marine structures;
and4. road de-icing salts, a particular problem on
bridge decks.
Calcium chloride, or any chloride-containingadmixture, is normally no longer permitted inconcrete containing steel, but it is generallyaccepted that small amounts of chlorides fromany of the other sources can be tolerated and thata threshold level is required to depassivate thesteel allowing the corrosion to start. There are,however, differences of opinion about the magni-tudes involved, including the following (Glassand Buenfeld, 1997):
• some of the chlorides react with the aluminatephases in the cement forming chloroalumi-nates, and therefore are no longer free chlo-ride ions. It was thought that these boundchlorides are not harmful, but this is nolonger believed to be true;
• corrosion has occurred at a very wide range oftotal chloride contents, and it may be better tothink of the chloride content as giving a riskof corrosion, rather than there being anabsolute threshold value below which no cor-rosion can ever occur.
Despite this, many recommendations include athreshold limit, with a value of 0.4. per cent chlo-ride ion by weight of cement being typical.
If the chlorides are included in the concrete onmixing, then the steel may never be passivated,and the initiation period, t0, will be zero. Chloridesfrom the external sources, sea water or de-icingsalts have to penetrate the concrete cover in suffi-cient quantities, however defined, to depassivatethe steel before the corrosion is initiated: t0 is finitein these circumstances. The transport mechanismsmay be governed by the permeability in the caseof, say, concrete permanently submerged in seawater; diffusivity, where salts are deposited ontosaturated concrete; or sorptivity, where salts aredeposited onto partially saturated concrete. Thecorrosion risk in situations in which the salts arewater-borne and deposited onto the surface byevaporation, such as in the splash zone of marinestructures or on run-off from bridge decks, isparticularly high as the reservoir of salts is con-stantly replenished. An absorption mechanism maydominate in the early stages of such contamina-tion, with diffusion being more important at laterstages (Bamforth and Pocock, 1990).
These processes result in chloride profiles such asthose shown in Figure 23.24(a). A large number ofsuch profiles showing the effect of a large number ofvariables have been generated both experimentallyand analytically, and these have been used to for-mulate design recommendations to ensure that theconcrete properties and cover to the reinforcing issufficient to keep the chloride content at the steel toacceptable levels for the required design life. Theirform and magnitude depend on time of exposure,the exposure conditions and the concrete properties;clearly the properties in the cover zone are critical.Lower sorptivities and diffusivities, and henceincreased values of t0, are associated with lowerwater/cement ratios, higher cement contents andefficient curing. The use of CRMs can also reducethe chloride penetration significantly, as shown inFigure 23.24(b) for concrete containing pfa.
Although many recommendations for concretecover and quality are aimed at extending theperiod t0 as far as possible, there are circum-stances in which it is impossible to prevent
Durability of steel in concrete
213
corrosion being initiated. Much research hastherefore been carried out to determine thefactors which control the rate of corrosion. Thesehave been found to include the following.
1. The spacing and relative size of the anode andcathode in the corrosion cell. Relativelyporous areas of a concrete member, such as apoorly compacted underside of a beam, willallow rapid penetration of chlorides, depassi-vating a small area of steel to form the anode.The reinforcement throughout the structure isnormally electrically continuous, and so theremainder forms a large area cathode, result-ing in a concentration of corrosion current,and hence a high corrosion rate, at the anode.
2. The availability of oxygen and moisture,particularly to sustain the cathodic reaction. Ifthe supply of either is reduced, then the corro-sion rate is reduced. Hence little corrosionoccurs in completely dry concrete, and onlyvery low rates in completely and permanentlysaturated concrete, through which diffusion ofoxygen is difficult.
3. The electrical resistivity of the electrolyte of
the corrosion cell, i.e. the concrete. High resis-tivities reduce the corrosion current and hencethe rate of corrosion, but increasing moisturecontent, chloride content and porosity allreduce the resistivity.
Much analysis of the extensive and increasingamount of data on this subject is aimed at pro-ducing guidelines to ensure adequate durability(see below). There are, however, circumstances inwhich protection against corrosion cannot beguaranteed by selection of the materials and pro-portions of the concrete, depth of cover andattention to sound construction practice. Theseinclude, for example, marine exposure in extremeclimatic conditions, and regions in which aggre-gates containing excess chlorides must be used.One or more of the following extra protectivemeasures may then be taken:
1. the addition of a corrosion inhibiting admixturesuch as calcium nitrite to the fresh concrete;
2. the use of corrosion-resistant stainless steelreinforcement bars, or epoxy-coated conven-tional bars;
Durability of concrete
214
FIGURE 23.24 Chloride concentration profiles in concrete after marine exposure in tidal/splash zone: (a) theeffect of exposure period (Portland cement concrete, moist-cured for 3 days, exposed from 28 days) (Bamforth andPocock, 1990); (b) the effect of pfa (moist-cured for 1 day, exposed from 28 days for 2 years) (Thomas et al., 1990).
3. applying a protective coating to the concrete,to reduce chloride and/or oxygen ingress;
4. cathodic protection of the reinforcement, i.e.applying a voltage from an external sourcesufficient to ensure that all of the steel remainspermanently cathodic (see Section 11.4.3).
23.5 Recommendations for durableconcrete constructionWe have already discussed some of the measuresthat can be taken to avoid degradation from the
specific problems of sulfates, acids, alkali–silicareaction and fire. A further example is theapproach adopted in the UK for the action ofother environmental conditions. This combinesthem into a single set of exposure conditions withcorresponding requirements for concrete qualityand cover to embedded steel, which is useful andconvenient (BS 8110: 1997). The exposure con-ditions are given in Table 23.4, and the combina-tion of concrete quality and cover in Figure23.25. The engineer has some flexibility of choiceof combination in each exposure condition. As
Recommendations for durable concrete construction
215
TABLE 23.4 Exposure conditions for concrete (BS 8110: 1997)
Exposure condition Environment of concrete surface
Mild Protected against weather or aggressive conditions
Moderate Sheltered from severe rain or freezing whilst wetSubject to condensationContinuously under waterIn contact with non-aggressive soils
Severe Exposed to severe rain, alternate wetting and drying or occasional freezing or severecondensation
Very severe Occasionally exposed to sea-water spray, de-icing salts, corrosive fumes or severe freezingwhilst wet
Most severe Frequently exposed to sea-water spray or de-icing salts in sea-water tidal zone down to 1mbelow lowest low water
Abrasive Exposed to abrasive action, e.g. machinery, metal-tyred vehicles or water carrying solids
Min
imum
cov
er t
o re
info
rcem
ent
(mm
)
50
40
30
20
300.65275
350.6300
400.55325
450.5350
500.45400
Minimum concrete characteristic strength (MPa)Maximum water/cementMinimum cement content (kg/m3)
Mild
Moderate
Severe
Very severe
Most severe
If these concretes areexposed to freezing when wet,air entrainment is also required
For abrasive exposure, use theother relevant value and addallowance for abrasion loss.
FIGURE 23.25 Requirements for concrete quality and cover to steel for the exposure conditions in Table 23.4(BS 8110: 1997).
we mentioned in Chapter 21, the concrete qualityrequirements may form a limiting value for theconcrete mix design.
As in other areas of construction technologyand practice, there has been extensive recentwork to produce combined European recommen-dations for durable concrete. These need interpre-tation for the requirements of individualcountries, and a look at Hobbs (1998) will givean idea of the resulting extent and complexities ofthe subject.
23.6 ReferencesAbrams, M.S. (1971) Temperature and concrete. Amer.
Conc. Inst. Special Publication 25, pp. 33–58.Bamforth, P.B. (1987) The relationship between perme-
ability coefficients for concrete obtained using liquidand gas. Magazine of Concrete Research, 39, 138,March, 3–11.
Bamforth, P.B. and Pocock, D.C. (1990) Proceedingsof Third International Symposium on Corrosion ofReinforcement in Concrete Construction, ElsevierApplied Science, pp. 119–31.
Browne, R.D. (1983) Proceedings of the Symposium onDurable Concrete, Institute of Concrete Technology,London.
Browne, R.D. (1985) Proceedings of Seminar onImprovements in Concrete Durability, Institute ofCivil Engineers, London, pp. 97–130.
Buenfeld, N. and Okundi, E. (1998) Effect of cementcontent of transport in concrete. Mag. of Conc. Res.50, 4, Dec. 339–51.
Buenfeld, N. and Newman, J.B. (1984) The permeabil-ity of concrete in a marine environment. Magazineof Concrete Research 36, 127, June, 67–80.
Building Research Establishment (1996) Sulfate andacid resistance of concrete in the ground. Digest363, BRE, Garston, UK.
Concrete Society (1999) Alkali–silica reaction: mini-mising the risk of damage to concrete. TechnicalReport No. 30, 3rd edn, Slough, p. 72.
Concrete Society (1988) Permeability testing of siteconcrete. Technical Report No. 31, London.
Dhir, R.K., Hewlett, P.C. and Chan, Y.N. (1987)Near-surface characteristics of concrete: assessmentand development of in-site test methods. Magazineof Concrete Research 39, No. 141, Dec., 183–95.
Glass, G.K. and Buenfeld, N.R. (1997) The presenta-tion of the chloride threshold level for corrosion ofsteel in concrete. Corrosion Science 39, 5, May1001–17
Hobbs, D.W. (1986) Deleterious expansion of concretedue to alkali–silica reaction: influence of PFA andslag. Magazine of Concrete Research 38, 137, Dec.191–205.
Hobbs, D.W. (1988) Alkali–Silica Reaction in Con-crete, Thomas Telford, London, p. 183.
Lawrence, C.D. (1985) Concrete Society MaterialsResearch Seminar on Serviceability of ConcreteSlough, July, published in Concrete Society (1988)above.
Nagataki, S., Ohga, H. and Kim, E.K. (1986) Proceed-ings of the 2nd International Conference on Fly Ash,Silica Fume, Slag and Natural Pozzolans in Con-crete. Amer. Conc. Inst. Special Publication SP-91,521–40.
Page, C.L., Short, N.R. and El Tarras, A. (1981)Cement and Concrete Research 11, 3, 395–406.
Parrott, L.J. (1987) A Review of Carbonation in Rein-forced Concrete, Cement and Concrete Association,Slough.
Powers, T.C. (1958) Structure and physical propertiesof Portland cement paste. Journ. Amer. Ceramic Soc.41, Jan., 1–6.
Powers, T.C., Copeland, L.E., Hayes, J.C. and Mann,H.M. (1954) Permeability of Portland cement paste.Journ. Amer. Conc. Inst. 51, November, 285–98.
Richardson, M.G. (1988) Carbonation of ReinforcedConcrete: Its Causes and Management, CITIS Ltd,Dublin.
Rose, D.A. (1965) RILEM Bulletin, No. 29.Stanton, T.E. (1940) The expansion of concrete
through reaction between cement and aggregate.Proc. Am. Soc. Civ. Engrs. 66, 1781–811.
Thomas, M.D.A., Matthews, J.D. and Haynes, C.A.(1990) Proceedings of Third International Sympo-sium on Corrosion of Reinforcement in ConcreteConstruction, Elsevier Applied Science, pp. 198–212.
Thaumasite Expert Group (1999) The ThaumasiteForm of Sulfate Attack: Risks, Diagnosis, RemedialWorks and Guidance on New Construction, Dept.of Environment, Transport and Regions, London, p. 180.
US Bureau of Reclamation (1955) Concrete LaboratoryReport No. C-810, Denver, Colorado.
Verbeck, G.J. (1968) in Performance of Concrete (ed.E.G. Swenson), University of Toronto Press.
Durability of concrete
216
Chapter twenty-four
Highperformanceconcrete
24.1 High strength concrete24.2 Self-compacting concrete24.3 References
‘High strength concrete’ became a familiar phrasein concrete technology in the late 1980s, and afew years later was widened to ‘high performanceconcrete’. It has been the subject of many papers,books and conferences, but it may have beenpreferable to use the term higher performanceconcrete, since it is essentially an extension andenhancement of existing conventional concretepractice.
High performance concrete is not a single typeof concrete, but includes any concrete whoseproperties significantly exceed or extend therange of concrete currently being used. There isalso another important condition – the concretemust be capable of being mixed, handled andplaced with the same equipment and proceduresas conventional concrete, albeit with higherstandards of quality control and more carefulselection of materials and mix design. The supe-rior properties can thus be readily exploited inconcrete practice. In nearly all cases the enhance-ment of properties is through the use of cementreplacement materials and superplasticisers, bothof which we have discussed in previous chapters.
We should add here that there are a number ofspecialist concretes that, although demonstrating
the great versatility of the material, are not nor-mally referred to as high performance concrete.Examples are underwater concrete, foamed con-crete, autoclaved aerated concrete and roller com-pacted concrete; discussion of these is beyond thescope of this book.
There are three broad divisions of highperformance concrete – high strength concrete,high durability concrete and self-compacting con-crete. We will consider the first and last of these –in which the high performance involves the hard-ened and fresh properties respectively. As we dis-cussed in the last chapter, the achievement ofhigh durability often results in high strength, butthis is not always the case, e.g. freeze–thaw resis-tance is obtained by air entrainment whichreduces the strength.
24.1 High strength concreteAn accepted definition of a high strength concreteis a strength significantly higher than that used in normal practice. As practice is continuallyadvancing, this means that the quantitative defini-tion needs regular adjustment, and can vary fromcountry to country. An accepted current value isa characteristic strength in excess of about80MPa.
It will be clear from Chapter 20 that the use ofa low water/cement or water/binder ratio is a
217
prime requirement, but this in itself is not suffi-cient. The main considerations arising from themany research studies and development pro-grammes can be summarised as follows:
• Microsilica at levels of up to about 10 percent of the binder is important, particularlyfor achieving very high strengths (Figure24.1). The main benefit is improvement of thepaste/aggregate transition zone.
• All materials must be very carefully selectedfor optimum properties:– high quality aggregates are required, with
crushed rocks normally preferred, and somelimestones giving particularly good perform-ance. Limiting the maximum aggregate sizeto 10mm is normally suggested.
– Superplasticisers are essential for producingadequate workability, but the binder(particularly the cement) and superplasti-ciser must be compatible to avoid problemssuch as rapid loss of workability.
• Even at slumps in excess of 175mm, themixes can be cohesive and difficult to handle,i.e. they have a high plastic viscosity. Atten-tion to aggregate gradings and particle sizecan reduce this problem; the lubrication pro-vided by the very fine spherical microsilicaparticles is beneficial.
• Problems that are of minor significance fornormal strength concrete, such as loss of
workability and heat of hydration effects, willbe exaggerated in high strength mixes, andtherefore may become critical without dueconsideration. The use of ternary cementblends, i.e. Portland cement plus microsilica(csf) and either pfa or ggbs can be helpful inmany cases
• Mix design is considerably more complexthan for normal strength concrete because ofthe larger size of variables involved and theirinteractive effects. A more extensive set oftrial mixes at both laboratory and full scaleare required.
• All production and quality control issues needmuch greater attention than for normalstrength concrete, since the consequences ofvariations and fluctuations will be much moreserious.
• The elastic modulus continues to increasenon-linearly with strength, with a typical rela-tionship being (Kakizaki et al., 1992):
Ec �3.65(fcyl)0.5
for strengths between 80 and 140MPa (24.1)
where Ec �elastic modulus (in GPa) andfcyl �cylinder compressive strength (in MPa).
However, the stress–strain behaviourbecomes distinctly more brittle and the con-crete fails at increasingly lower strains as
High performance concrete
218
60
80
100
120
140
160
0.2 0.25 0.3 0.35 0.4Water/binder ratio
18
0 d
ay c
ompr
essi
vest
reng
th (
MPa
) csf(% of binder)
1510
5
0
10 mm granite aggregateall mixes with superplasticiserto 150 mm slump
FIGURE 24.1 Strength versus water/binder ratio for high strength concrete (Domone and Soutsos, 1995).
strength increases. This has consequences forreinforcement design.
All of these factors mean that high strength con-crete is relatively expensive, but its use may leadto overall structural economies, and there aregrowing numbers of successful examples of itsuse. Reduced section sizes mean much lower self-weights – a major factor in long-span bridges –and in high-rise buildings, another significantarea of use, the reduced column cross-sectionsgiving higher usable space, particularly in thelower storeys.
The upper limits of strength levels shown inFigure 24.1 are approaching those that can beachieved with ‘conventional’ concrete materialsand practice, but do not represent an overallceiling or a limit. There have been several develop-ments of materials with water/binder ratios con-siderably less than 0.2. One example is themacro-defect free (MDF) cement already discussedin Chapter 13, p. 105. Another example is DSPcement (densified with small particles) which usesthe action of a superplasticiser and microsilica toproduce low porosity, which when combined witha strong aggregate of 4mm maximum size pro-duces compressive strengths of up to 260MPa(Bache, 1994). RPC (reactive powder concrete)takes this a stage further by using a maximumaggregate size of, at most, 600 microns, increasingthe density by optimising the proportions of all ofthe particles, and curing under pressure followedby heat treatment. Unconfined compressivestrengths of more than 200MPa are obtained, andductility can be provided by including short steelfibres (Bonneau et al., 1996).
Although such materials may only find use inspecialist applications, they do demonstrate that,by understanding and applying the principles ofmaterials science to cement composites, a wideand continuous spectrum of performance can beachieved.
24.2 Self-compacting concreteSelf-compacting concrete (SCC) can achieve fulland uniform compaction without the need for
any help from vibration. It is distinctly differentfrom other types of high workability concrete inthat it is able to flow through and around heavilycongested reinforcement whilst retaining itsintegrity and homogeneity. It was developed inJapan in the late 1980s in response to a lack ofskilled construction workers, and it was quicklyadopted into Japanese construction practice, andits use is now rapidly spreading to other countries(RILEM, 2001). A major advantage is environ-mental – construction sites and precast works aremuch less noisy, and the health risks associatedwith hand-held vibrators are eliminated.
The high performance relates to the fresh pro-perties only. It is possible to produce the com-plete range of strengths and other properties ofconcrete described hitherto in this book, apartfrom very low strengths. SCC requires a combi-nation of:
• high fluidity and stability, achieved by a com-bination of low water/binder ratios and super-plasticisers, often supplemented by viscosity-enhancing agents or ‘thickeners’. These twoproperties are described as filling ability andsegregation resistance respectively, and in rheo-logical terms, require a very low yield stressand a moderate to high plastic viscosity (butnot so high that flow times are excessive);
• the avoidance of aggregate particles bridgingbetween reinforcing bars and blocking theflow, achieved by an increase in the volume ofpaste or mortar, and a consequent reductionin the coarse aggregate volumes (Figure 24.2).This property is called passing ability, and thevolume of coarse aggregate in SCC is typicallyin the range 30–34 per cent of the concretevolume, compared to 40–55 per cent innormal concrete.
This combination of properties can only beachieved with a reasonably high binder contentand low water/binder ratios, typically in theranges 450–600kg/m3 and 0.3 to 0.5 respectively.Control of strength and heat of hydration effectsis obtained through the use of significant quanti-ties of CRMs and/or inert fillers such as limestonepowder. Apart from the slump flow test, where a
Self-compacting concrete
219
value of at least 600mm is required, a consensusof appropriate tests and limiting values tomeasure and specify the properties of SCC hasnot yet emerged. However, it is clear that SCCwill have an increasing impact on concrete prac-tice in the next few years.
24.3 ReferencesBache, H.H. (1994) Design for ductility. In Concrete
Technology: New Trends, Industrial Applications(ed. R. Gettu, A. Aguado and A. Shah), E & FNSpon, London, pp. 113–25.
Bonneau, O., Poulin, C., Dugat, J., Richard, P. and
Aitcin, P.-C. (1996) Reactive powder concrete fromtheory to practice. Concrete International 18, 4,April, 47–9.
Domone, P.L. and Soutsos, M.N. (1995) Properties ofhigh strength concrete mixes containing pfa andggbs. Mag. of Con. Res. 437, 173, Dec. 355–67.
Kakizaki, M. (1992) Effect of mixing method onmechanical properties and pore structure of ultra-high strength, concrete. Proc. of 4th CANMET/ACIInt. Conf. on Fly Ash, Silica Fume, Slag and NaturalPozzalans in Concrete, Istanbul, Turkey, ACI SP-132, American Concrete Institute, Michigan, USA,997–1015.
RILEM Technical Committee 174-SCC (2001) Self-compacting Concrete: State-of-the-art Report (ed. A.Skarendahl and O. Petersson), RILEM, Paris.
High performance concrete
220
Normal concrete – blocking dueto aggregate particle bridging
Successful SCC with goodpassing ability – no blocking
Aggregate
Reinforcingbars
FIGURE 24.2 The passing ability of self-compacting concrete.
Further reading
GeneralNeville, A.M. (1995) Properties of Concrete, 4th edn,
Pearson Education, Harlow, p. 844.Since its first edition in 1963, this has been the defini-
tive reference book of all aspects of concrete techno-logy. This fourth edition encompasses the manyrecent developments, and it is the first source for allwith an interest in concrete. It is not a teaching text,but a book to be consulted regularly.
Hewlett, P.C. (ed.) (1998) Lea’s Chemistry of Cementand Concrete, Arnold, London, p. 1052.
An update of a book first published in 1935. Theauthoritative text on the subject, with much detail atan advanced level. Not for the faint-hearted, butworth consulting for project work, etc.
Murdock, L.J., Brook, K.M. and Dewar, J.D. (1992)Concrete Materials and Practice, 6th edn, EdwardArnold, London.
Aimed at professionals in the concrete industry, quitedetailed, but useful when you want to know moreabout practice.
Clarke, J. (ed.) (1993) Structural Lightweight AggregateConcrete, Blackie Academic and Professional, London.
A useful source for all aspects of this subject.
History of concreteBritish Cement Association (1999) Concrete Through
the Ages, BCA, Crowthorne, p. 37.Fairly short, but interesting and well illustrated.
Cements/CRMsBye, G.C. (1999) Portland Cement: Composition, Pro-
duction and Properties, 2nd edn, Thomas Telford,London, p. 225.
Detailed and comprehensive – a good reference source.
Concrete Society (1991) The use of ggbs and pfa in
concrete. Technical Report No. 40, Wexham, p. 142, and
Concrete Society (1993) Microsilica in concrete. Tech-nical Report No. 41, Wexham, p. 54.
Both summaries produced for professionals in the con-crete industry, but also useful for students.
AdmixturesRamachandran, V.S., Malhotra, V.M., Jolicoeur, C.
and Spiratos, N. (1998) Superplasticisers: Propertiesand Applications in Concrete, CANMET, Ottawa,p. 404.
Specialised and not very easy reading, but comprehen-sive.
Rixom, R. and Mailvaganam, N. (1999) ChemicalAdmixtures for Concrete, 3rd edn, E & FN Spon,London, p. 456.
A welcome new edition – thorough.
Mix designACI 211.1–91 (2000) Standard practice for selecting
proportions for normal heavyweight and mass con-crete: ACI Manual of Concrete Practice. Part 1:Materials and General Properties of Concrete,American Concrete Institute, Farmington Mills,Michigan, USA.
The US procedure.
BRE (1997) Design of Normal Concrete Mixes, 2ndedn, Building Research Establishment, Watford.
A fairly straightforward UK method discussed inChapter 21.
Day, K.W. (1995) Concrete Mix Design, QualityControl and Specification, E & FN Spon, London, p. 350.
deLarrard, F. (1999) Concrete Mixture Proportioning:a Scientific Approach, E & FN Spon, London, p. 350.
221
Dewar, J.D. (1999) Computer Modelling of ConcreteMixtures, E & FN Spon, London, p. 272.
All these three are for those with an interest in consid-ering mix design in great detail. They contain avarying mixture of science, empiricism and practicaladvice.
Fresh concreteMasterston, G.G.T. and Wilson, R.A. (1997) The plan-
ning and design of concrete mixes for transporting,placing and finishing. CIRIA Report 165, Construc-tion Industry Research and Information Association,London.
Useful to those concerned with practical issues of con-crete construction.
Tattersall, G.H. and Banfill, P.F.G. (1983) The Rheol-ogy of Fresh Concrete, Pitman, London.
Contains all the relevant theory, background anddetails of application of rheology to fresh cementand concrete, and summarises all the pioneeringstudies of the 1960s and 1970s. An excellent refer-ence text.
Tattersall, G.H. (1991) Workability and QualityControl of Concrete, E & FN Spon, London.
Discusses the nature of workability and workabilitytesting, summarises the background to rheologicaltesting, and considers quality control issues in somedetail. The source for much of the information inthis chapter.
Non-destructive testingBungey, J. (1989) The Testing of Concrete in
Structures, 2nd edn, Surrey University Press,London.
Fairly easy to read, and covers most aspects of NDT ofconcrete.
Malhotra, N.J. and Carino, V.M. (1991) CRC Hand-book on Non-destructive Testing of Concrete, CRCPress, Boca Raton, Florida.
More detailed, a good reference source.
DurabilityMays, G.C. (1991) Durability of Concrete Structures:
Investigation, Repair, Protection, E & FN Spon,London.
Considers the whole subject in detail, from fundamen-tal mechanisms through to repair methods.
High performance concreteAitcin, P.-C. (1998) High-performance Concrete, E &
FN Spon, London, p. 591.A comprehensive text on this fast developing area of
concrete technology; takes the subject from the back-ground science to many case studies of applications.
Concrete Society (1997) Technical Report TR47Design with High-strength Concrete, ConcreteSociety, Crowthorne, p. 150.
Provides design guidance for structural applications ofhigh-strength concrete not covered by current codesof practice.
Nawy, E.G. (1996) Fundamentals of High-strength Con-crete Performance Concrete, Longman, Harlow p.340.
Considers the whole subject of high-strength concretefrom selection of materials, properties, uses and appli-cations, generally from an American perspective.Aimed at students and practising engineers.
Price, W.F. (2001) The Use of High-performance Con-crete, E & FN Spon, London.
Considers the uses and practical applications of thevarious types of high performance concrete, includinghigh strength, controlled density, high durability, highworkability and self-compacting concrete.
Standards referred to in the text
UK Standards (published by the BritishStandards Institution)BS 8110: 1997 The structural use of concrete. Part 1:
Code of practice for design and construction.
Cements
BS 12: 1996 Specification for Portland cement.BS 4027: 1991 Specification for sulfate-resisting Port-
land cement.BS 6699: 1992 Specification for ground granulated
blast furnace slag for use with Portland cement.BS 3892: Part 1 1993 Specification for pulverised fuel
ash for use with Portland cement.
Admixtures
BS 5075: Concrete Admixtures.Part 1: 1982 Specification for accelerating admix-tures, retarding admixtures and water reducingadmixtures.
Further reading
222
Part 2: 1982 Specification for air-entraining admix-tures.Part 3: 1985 Specification for superplasticizingadmixtures.
Aggregates
BS 812 Testing aggregates (24 parts).BS 882: 1992 Specification for aggregates from natural
sources for concrete.
Test methods
BS 1881: Testing concrete.Part 102: 1983 Method for determination of slump.Part 102: 1983 Method for determination of flow.Part 103: 1983 Method for determination of com-pacting factor.Part 104: 1983 Method for determination of Vebetime.Part 116: 1983 Method for determination of com-pressive strength of concrete cubes.Part 201: 1986 Guide to the use of non-destructivemethods of test for hardened concrete.Part 202: 1986 Recommendations for surface hard-ness testing by rebound hammer.Part 203: 1986 Recommendations for measurementof velocity of ultrasonic pulses in concrete.Part 207: 1986 Recommendations for the assessmentof concrete strength by near-to-surface tests.Part 208: 1986 Recommendations for the determina-tion of the initial surface absorption of concrete.Part 209: 1986 Recommendations for the measure-ment of dynamic modulus of elasticity.
BS 5168: 1975 Glossary of rheological terms.
European Standards (published by theBritish Standards Institution)EN 197–1: 2000 Cement – Part 1: Composition, speci-
fications and conformity criteria for commoncements.
EN 12350–2: 2000 Testing fresh concrete. Slump test.EN 12350–3: 2000 Testing of fresh concrete. Vebe test.EN 12350–4: 2000 Testing of fresh concrete. Degree of
compactability.EN 12350–5: 2000 Testing of fresh concrete. Flow
table test.DD ENV 1992–1–2: 1996 Eurocode 2: Design of con-
crete structures.DD ENV 206: 1992 Concrete: Performance, produc-
tion placing and compliance criteria.
US Standards (published by the AmericanSociety for Testing and Materials)ASTM C150–94 Specification for Portland cement.ASTM C494–92 Specification for chemical admixtures
for concrete.ASTM C618–94a Specification for coal fly ash and raw
or calcined natural pozzola for use as a mineraladmixture in hydraulic-cement concrete.
ASTM C989–93 Specification for ground granulatedblast furnace slag for use in concrete and mortars.
ASTM C1240–93a Specification for silica fume for useas a mineral admixture in hydraulic-cement con-crete, mortar and grout.
ASTM Specification C 143–90a Test for slump ofhydraulic cement concrete.
ASTM Specification C 232–92 Test for bleeding ofconcrete.
ASTM Specification C 125–93 Standard definitionsand terms relating to concrete and concrete aggre-gates.
Standards referred to in the text
223
Part Four
Bituminousmaterials
D.G. Bonner
227
Introduction
The term ‘bituminous materials’ is generallytaken to include all materials consisting ofaggregate bound with either bitumen or tar.Materials of this kind are used almost exclusivelyin road construction. However, bitumen and taron their own have other uses in construction. Forexample bitumen is used in roofing materials andas a protective/waterproof coating. This part willconcentrate on the use of bituminous materials inroad construction.
The use of tar in road building materials beganto grow significantly in the UK just after the turnof the century following the advent of the pneu-matic tyre and the motor vehicle. Up to that time,roads were constructed following the principlesdeveloped by Macadam using waterbound,graded aggregate. Under the action of pneumatictyres and the higher speeds of motor vehicles, agreat deal of dust was generated on macadamroads which led to the use of tar as a dressing tobind the surface. Tar was eminently suitable forthis purpose since it could be made sufficientlyfluid by the use of heat to be sprayed, but stiff-ened on cooling. Furthermore, it protected theroad from the detrimental effects of water. Thebenefits of using tar were quickly realized and arange of ‘coated stone’ materials, or ‘tarmac-adams’, were developed.
References to natural sources of bitumen dateback to biblical times. However, the first use ofnatural rock asphalt for paving roads was in themiddle of the nineteenth century. The first refin-ery bitumens to be used in this country camefrom the Mexican oilfields around 1913. But itwas the opening of the Shell Haven refinery in1920 which gave rise to the rapid development ofbitumen for road construction. Bitumen was
found to be less temperature-susceptible than tar.Thus it is less soft than an equivalent grade of tarat high temperatures, making it more resistant todeformation, and less stiff than tar at low temper-atures, making it less brittle and more resistant tocracking. As the performance required of bitumi-nous materials increased, due to the increase inquantity and weight of traffic, bitumen becamemore widely used than tar. Although tar-boundmaterials predominated during the Second WorldWar due to difficulties in importing crude oil, theintroduction of North Sea gas in the late 1960sdramatically reduced the production of crudecoal tar. Now the use of tar in road constructionis limited to tar/bitumen blends for surface dress-ing and dense tar surfacing used in areas such ascar parks where the greater resistance of tar todiesel and oil spillage is an advantage. Thereforethe following chapters will deal only withbitumen.
A very wide range of bituminous mixtures haveevolved to suit the wide variety of circumstancesin which they are used. They vary according totheir bitumen content and grade as well as theiraggregate grading and size. However, they can allbe classified into two groups, namely asphaltsand macadams. (Here it should be noted that theterm ‘asphalt’ is used with its European meaning,i.e. a particular type of mixture of bitumen andaggregate. In North America, ‘asphalt’ meansbitumen.) Figure IV.1 illustrates the fundament-ally different characteristics of asphalts andmacadams. Asphalts rely on their dense, stiffmortar for strength and stiffness, whereasmacadams rely on the stability of the aggregatethrough its grading. Thus the role of bitumen isquite different in each case, and the properties of
asphalts are more strongly dependent on thenature of the bitumen than the properties ofmacadams. However, although Figure IV.1 por-trays two very different types of material, there is,in fact, almost a continuous spectrum of mater-ials between these two extremes. Thus there areasphalts which have a larger coarse aggregatecontent than suggested in Figure IV.1 so that theoverall aggregate grading is more continuous andthe materials resemble macadams in that respect.Similarly there are macadam mixes which aredense and contain more bitumen, and tendtowards the model for asphalts. This will be dis-cussed in more detail in Chapter 29.
Bituminous materials are used in so-called
Introduction
228
Up to 80 Msa
Wearing course
Basecourse
Roadbase
Sub-base
Capping
Subgrade
Lower roadbase
Upper roadbase
Greater than 80 Msa
Bituminousmaterials areused for theselayers
Wearing course
Sub-base
Capping
FIGURE IV.1 The essential features of asphalts andmacadams: (a) asphalt; (b) macadam.
FIGURE IV.2 The structure of a flexible roadaccording to the traffic loading in millions of standardaxles (Msa).
‘flexible’ construction. The alternative is ‘rigid’construction where the road consists essentiallyof a concrete slab. In flexible construction thereare a number of layers to the road structure eachhaving a specific function. Figure IV.2 illustratesthose layers and where bituminous materials maybe used. The nature of the materials will varyaccording to their position and function in thestructure. Thus both the wearing course andbasecourse may be asphalts but the propertiesrequired of the wearing course at the surface aredifferent from those required just below thesurface in the basecourse. Therefore a wearingcourse asphalt differs from a basecourse asphalt.Also, particular types of material are selectedaccording to their suitability. Thus when trafficloads are very high, an asphalt roadbase may notprovide sufficient resistance to deformation, butwould give the necessary resistance to fatiguecracking. Therefore asphalt is used at the bottomof the roadbase where tensile stresses inducingcracks are greatest, and macadam is used for theupper part to provide improved resistance todeformation. This will be discussed more fully inChapter 27.
Chapter twenty-five
Structure ofbituminousmaterials
25.1 Constituents of bituminous materials25.2 Bitumen25.3 Types of bitumen25.4 Aggregates25.5 Reference
25.1 Constituents of bituminousmaterialsBituminous materials consist of a graded aggreg-ate bound together with bitumen. Thus they aretwo-phase materials and their properties dependupon the properties of the individual phases aswell as the mix proportions. The two phases arequite different in nature. Whilst the aggregate isstiff and hard, the bitumen is flexible and soft andis particularly susceptible to temperature change.Therefore the proportion of bitumen in the mixhas a great influence on the mix properties and iscrucial in determining the performance of thematerial.
Bitumen may be supplied in a number of formseither to facilitate the mixing and laying processor to provide a particular performance. Aggre-gates may come from a wide range of rock typesor from artificial sources such as slag. Thegrading of the aggregate is important and rangesfrom continuous grading for mix types known asmacadams through to gap grading for mixesknown as asphalts. The very fine component ofthe aggregate (passing 75 microns) is called filler.
Although the graded aggregate will normallycontain some material of this size, it is usuallynecessary to provide additional filler in the formof limestone dust, pulverized fuel ash or ordinaryPortland cement.
25.2 Bitumen
25.2.1 Sources
There are two sources of bitumen: naturaldeposits and refinery bitumen.
Natural asphalts
Bitumen occurs naturally, formed from petroleumby geological forces, and always in intimateassociation with mineral aggregate. Types ofdeposit range from almost pure bitumen tobitumen-impregnated rocks and bituminoussands with only a few per cent bitumen.
Rock asphalt consists of porous limestone orsandstone impregnated with bitumen with atypical bitumen content of 10 per cent. Notabledeposits are in the Val de Travers region ofSwitzerland and the ‘tar sands’ of North America.
Lake asphalt consists of a bitumen ‘lake’ withfinely divided mineral matter dispersed through-out the bitumen. The most important deposit ofthis type, and the only one used as a source ofroad bitumen in the UK, is the Trinidad Lake.
229
The lake consists of an area of some 35ha andextends to a depth of 100m. Asphalt is dug fromthe lake, partially refined by heating to 160°C inopen stills to drive off water, then filtered, bar-relled and shipped. The material consists of 55per cent bitumen, 35 per cent mineral matter and10 per cent organic matter. It is too hard in thisform to use directly on roads and is usuallyblended with refinery bitumen.
Refinery bitumen
This is the major source of bitumen in the UK. Inessence, bitumen is the residual material left afterthe fractional distillation of crude oil. Crudesvary in their bitumen content. The lighter paraf-finic crudes, such as those from the Middle Eastand North Sea, have a low bitumen contentwhich must be obtained by further processes afterdistillation. Heavier crudes, known as asphalticcrudes, such as those from the United States,contain more bitumen which is more easilyextracted.
25.2.2 Manufacture
The process of refining crude oil yields a range ofproducts, as shown in Figure 25.1. These prod-ucts are released at different temperatures withthe volatility decreasing and viscosity increasingas the temperature rises. Bitumen is the residual
material but its nature will depend on the distilla-tion process and, in particular, on the extent towhich the heavier oils have been removed. If theresidual material contains significant amounts ofheavy oils, it will be softer than if the heavy oilshad been more thoroughly extracted. Modernrefinery plant is capable of very precise controlwhich enables bitumen to be produced consis-tently at a required viscosity.
25.2.3 Chemistry and molecular structure
Bitumen is a complex colloidal system of hydro-carbons and their derivatives which is soluble intrichloroethylene. The usual approach to determi-nation of the constituents of a bitumen is throughuse of solvents. It may be subdivided into thefollowing main fractions:
1. carbenes – fraction insoluble in carbon tetra-chloride;
2. asphaltenes – fraction insoluble in lightaliphatic hydrocarbon solvent, e.g. heptane;
3. maltenes – fraction soluble in heptane.
The last two fractions predominate and themaltenes may be further subdivided into resins(that part adsorbed on an active powder such asFuller’s earth) and oils. The asphaltenes have thehigher molecular weight but their exact nature isdependent on the type of solvent and the volumeratio of solvent to bitumen. If small amounts of
Structure of bituminous materials
230
FIGURE 25.1 Preparation of refinery bitumens.
solvent are used, resins, which form part of themaltene fraction, may be adsorbed onto theasphaltene surfaces, yielding a higher percentageof asphaltene. Although they may vary accordingto method of extraction, the appearance ofasphaltenes is always of a dark brown to blacksolid which is brittle at room temperature. Theyhave a complex chemical composition but consistchiefly of condensed aromatic hydrocarbons andinclude complexes with nitrogen, oxygen, sulphurand metals such as nickel and vanadium. Thestructure of asphaltenes is not known with cer-tainty. One suggestion is of two-dimensional con-densed aromatic rings, short aliphatic chains andnaphthenic rings combined in a three-dimensionalnetwork (Dickie and Yen, 1967). Another sugges-tion is that there are two different moleculartypes, one being a simple condensed aromaticunit and the other consisting of collections ofthese simple units (Speight and Moschopedis,1979). It is likely that all of these may exist inbitumens from different sources since the natureof the molecules present in a crude oil will varyaccording to the organic material from which thecrude was formed, and to the type of surroundinggeology.
Maltenes contain lower molecular weight ver-sions of asphaltenes called resins and a range ofhydrocarbon groups known as ‘oils’ includingolefins, naphthenes and paraffins. The aromaticoils are oily and dark brown in appearance andinclude naphtheno-aromatic type rings. The satu-rated oils are made up mainly of long, straightsaturated chains and appear as highly viscouswhitish oil.
Bitumen is normally described as a colloidalsystem where the asphaltenes are solid particles inthe form of a cluster of molecules or micelles in acontinuum of maltenes (Girdler, 1965). Dependingon the degree of dispersion of the micelles in thecontinuous phase, the bitumen may be either a sol,where there is complete dispersal, or a gel, wherethe micelles are flocculated into flakes. Bitumenswith more saturated oils of low molecular weighthave a predominantly gel character. Those withmore aromatic oils, which are more likeasphaltenes, have a predominantly sol character.
Types of bitumen
231
In terms of their influence on the properties ofbitumen, asphaltenes constitute the body of thematerial, the resins provide the adhesive andductile properties, and the oils determine the vis-cosity and rheology.
Although bitumens are largely complex mix-tures of hydrocarbons, there are other elementspresent. The high molecular weight fraction con-tains significant amounts of sulphur, the amountof which influences the stiffness of the material.Oxygen is also present, and some complexes withoxygen determine the acidity of the bitumen. Thisis important in determining the ability of thebitumen to adhere either to aggregate particles orto road surfaces.
25.3 Types of bitumen
25.3.1 Penetration grades
Refinery bitumens are produced with a range ofviscosities and are known as penetration gradebitumens. The term derives from the test which isused to characterize them according to hardness.The range of penetration grades for road bitu-mens is from 15 to 450, although the most com-monly used are in the range 50 to 200. The rangeis produced partly through careful control of thedistillation process and partly by fluxing ‘resid-ual’ bitumen with oils to the required degree ofhardness. The specification for penetration gradebitumens is contained in BS 3690 Part 1(4).1
Table 25.1 indicates a range of tests with whichpenetration grade bitumens for road purposesmust comply. These bitumens are specified bytheir penetration value (BS 2000: Part 49: 1993)and softening point (BS 2000: Part 58: 1993)which indicate hardness and viscosity respec-tively. However, they are designated only by theirpenetration, e.g. 50 pen bitumen has a penetra-tion of 50!10. In addition there are limits forloss on heating (BS 2000: Part 45: 1993) whichensures that there are no volatile componentspresent whose loss during preparation and laying
1A list of all standards referred to in the text is included in‘Further reading’ on page 266.
would cause hardening of the bitumen, and solu-bility (BS 2000: Part 47: 1993), which ensuresthat there are only negligible amounts ofimpurity. The permittivity test (BS 2000: Part357: 1983) measures the dielectric constant forthe material which has been shown to be relatedto the weathering characteristics.
25.3.2 Oxidized bitumens
Refinery bitumen may be further processed by airblowing. This consists of introducing air underpressure into a soft bitumen under controlledtemperature conditions. The oxygen in the airreacts with certain compounds in the bitumenresulting in the formation of compounds ofhigher molecular weight. Thus the asphaltenecontent increases at the expense of the maltenecontent, resulting in harder bitumens which arealso less ductile and less temperature-susceptible.Although these bitumens are mostly used forindustrial applications such as roofing and pipecoatings, there is a road bitumen produced bythis process known as heavy duty bitumen. It isregarded as a penetration grade bitumen and isincluded in Table 25.1 (40 pen HD).
25.3.3 Cutbacks
Penetration grade bitumen is thermoplastic andthus its viscosity varies with temperature. At
ambient temperature it can be more or less solidand to enable it to be used for road constructionit must be temporarily changed into a fluid state.This may simply be achieved by raising the tem-perature. However, for surface dressing and sometypes of bituminous mixture it is necessary tohave a fluid binder that can be applied and mixedat relatively low temperatures, but have an ade-quate hardness after laying. Cutback bitumensare penetration grade bitumens which have theirviscosity temporarily reduced by dilution in avolatile oil. After application the volatile oil evap-orates and the bitumen reverts to its former vis-cosity.
The curing time and viscosity of cutbacks canbe varied according to the volatility of the dilut-ing oil and the amount of dilutent used. In theUK, cutbacks are manufactured using 100 or 200pen bitumen diluted with kerosene. Three gradesare produced to comply with a viscosity specifica-tion based on the standard tar viscometer (STV).Table 25.2 from BS 3690 shows that cutbacksmust also comply with solubility, distillation andrecovered penetration requirements. The last twoare to ensure that the dilutent will evaporate atthe required rate and that the residual bitumenwill have an appropriate hardness for theperformance requirements.
Structure of bituminous materials
232
TABLE 25.1 Specification for penetration grade bitumens (The Shell Bitumen Handbook, 1990)
Property Test method Grade of bitumen
40 pen HD 50 pen 100 pen
Penetration at 25°C dmm1 BS 2000: Part 49 40!10 50!10 100!20Softening point, °C min BS 2000: Part 58 58 47 41
max 68 58 51Loss on heating for 5h at 163°C BS 2000: Part 45
(a) loss by mass, % max 0.2 0.2 0.5(b) drop in penetration, % max 20 20 20
Solubility in trichloroethylene, % by mass min BS 2000: Part 47 99.5 99.5 99.5Permittivity at 25°C and 1592Hz min BS 2000: Part 357 2.630 2.630 2.630
Note:1. 1dmm�0.1mm.
25.3.4 Emulsions
An emulsion is a two-phase system consisting oftwo immiscible liquids, one being dispersed asfine globules within the other. A bitumen emul-sion consists of discrete globules of bitumen dis-persed within a continuous phase of water, and isa means of enabling penetration grade bitumensto be mixed and laid.
Dispersal of the bitumen globules must bemaintained electrochemically using an emulsifierwhich consists of a long hydrocarbon chain ter-minating with either a cationic or an anionicfunctional group. The hydrocarbon chain has anaffinity for bitumen, whereas the ionic part hasan affinity for water. Thus the emulsifier mole-cules are attracted to the bitumen globules withthe hydrocarbon chain binding strongly to thebitumen leaving the ionic part on the surface ofthe globule as shown in Figure 25.2. Each dropletthen carries a like-surface charge depending onthe charge of the ionic part of the emulsifier.Cationic emulsions are positively charged andanionic emulsions are negatively charged. Theglobules therefore repel each other, making theemulsion stable. Cationic emulsions are preferredbecause they also aid adhesion, the positivelycharged bitumen globules being strongly attractedto the negatively charged aggregate surface.
Emulsions must satisfy two conflicting require-ments in that they need stability for storage but
Types of bitumen
233
TABLE 25.2 Specification for cutback bitumens (BS 3690: Part 1: 1989) Reproduced with the permission ofthe British Standards Institution under licence number 2001/SK0230.
Property Test method Grade of cutback
50 sec 100 sec 200 sec
Viscosity (STV) at 40°C, 10mm cup, secs BS 2000: Part 72 50!10 100!20 200!40Distillation BS 2000: Part 27
(a) distillate to 225°C, % by volume max 1 1 1360°C, % by volume 10!3 9!3 7!3
(b) penetration at 25°C of residue from BS 2000: Part 49 100 to 350 100 to 350 100 to 350distillation to 360°C, dmm1
Solubility in trichloroethylene, % by mass min BS 2000: Part 47 99.5 99.5 99.5
Note:1. 1dmm�0.1mm.
FIGURE 25.2 Schematic diagram of charges onbitumen droplets (The Shell Bitumen Handbook,1990).
also may need to break quickly when put intouse. The stability of an emulsion depends on anumber of factors as follows:
1. The quantity and type of emulsifer present.Anionic emulsions require substantial waterloss before they break, whereas cationic emul-sions break by physicochemical action beforemuch evaporation. The more free emulsifierions there are in the continuous phase, theeasier it is for the negatively charged aggreg-ate surface to be satisfied without attractingthe bitumen globules.
2. Rate of water loss by evaporation. This inturn depends on ambient temperature, humid-ity and wind speed as well as rate and methodof application.
3. The quantity of bitumen. Increasing thebitumen content will increase the breakingrate.
4. Size of bitumen globules. The smaller theirsize, the slower will be the breaking rate.
5. Mechanical forces. The greater the mixingfriction or, in the case of surface dressing, therolling and traffic action, the quicker theemulsion will break.
The viscosity of emulsions is important because alarge proportion of emulsions are applied byspray. The viscosity increases with bitumencontent and is very sensitive for values greaterthan about 60 per cent. The chemistry of theaqueous phase is also important with viscositybeing increased by decreasing the acid content orincreasing the emulsifier content. The viscosityfor road emulsions is specified in BS 434: Part 1.
25.4 AggregatesAggregates make up the bulk of bituminousmaterials; the percentage by weight ranges fromabout 92 per cent for a wearing course asphalt toabout 96 per cent for an open-textured macadam.The aggregate has important effects on thestrength and stiffness of bituminous mixtures. Inmore open-textured types of mix, the strengthand resistance to deformation are largely deter-mined by the aggregate grading with the bitumen
acting principally as an adhesive. Here thegrading is continuous and provides a densepacking of particles leading to a stable aggregatestructure. In denser mixes, aggregate grading isagain important, but the properties are largelydetermined by the matrix of fines and bitumen.
The majority of aggregates used in bituminousmixes are obtained from natural sources, eithersands and gravels or crushed rock. The main non-natural aggregate source is slag, and blast furnaceslag is the most commonly used. As with con-crete, aggregates in bituminous mixes areregarded as inert fillers. However, whereas inconcrete both the aggregate and the hardenedcement paste are relatively stiff, in a bituminousmix, the bitumen is very soft compared to theaggregate. Therefore the role of the aggregate indetermining mix stiffness and strength is moreimportant in bituminous mixtures.
Three size ranges are recognized in aggregatesfor bituminous mixes. These are coarse, fine andfiller. Coarse material is that retained on a2.36mm sieve, fine material passes 2.36mm butis retained on the 75�m sieve, and the filler is thematerial passing 75�m.
25.4.1 Properties
The importance of grading has already been men-tioned. In addition, aggregates suitable for use inbituminous mixes must have sufficient strength toresist the action of rolling during construction.For surfacing materials, they must also be resis-tant to abrasion and polishing in order to providea skid-resistant surface. Here the shape andsurface texture of aggregate particles are import-ant.
Figure 25.3 gives typical grading curves for adense macadam and an asphalt wearing course.The curve for the macadam clearly shows thecontinuous nature of the grading, whereas thatfor the asphalt shows a gap grading with a lackof material in the range 600�m to 10mm. This istypical of an asphalt where the ‘mortar’ of finesand filler bound with bitumen characterizes thematerial, the coarse aggregate providing the bulk.
The strength of aggregate is assessed in two
Structure of bituminous materials
234
ways. For resistance to crushing, the aggregatecrushing value test is used. This test (BS 812: Part110: 1990) determines the extent to which anaggregate crushes when subjected to a steadilyincreasing load up to 40 tonnes. The test sampleconsists of single-sized particles, 10–14mm, andthe amount of fines produced in the test isexpressed as a percentage by weight of the ori-ginal sample. A variation of this test for weakeraggregates is the 10 per cent fines test. Here themaximum crushing load which will produce 10per cent fines from the original single-sizedsample is determined. The importance of this testis to assess the extent of crushing which mayoccur during compaction.
Resistance to impact loads is also required forroad aggregates. The impact value test (BS 812:Part 112: 1990) determines the response of
Aggregates
235
FIGURE 25.3 Aggregate grading curves.
Aggregate particles Fine sand
Texture depth
FIGURE 25.4 Measurement of macrotexture using the sand patch test.
aggregate to blows from a heavy hammer. Onceagain the outcome of the test is the percentage offines produced from the original single-sizedsample.
The skid resistance of a road surface is pro-vided largely by the aggregate exposed at thesurface. There are two components which arereferred to as macrotexture and microtexture.The macrotexture is the overall road surfaceroughness which arises from the extent to whichthere are spaces between aggregate particles. Thisis a function of mix proportions. For example, anasphalt provides an extremely low macrotexturebecause the coarse aggregate content is low andcoarse particles are immersed in the fines/filler/bitumen mortar. Consequently, a layer ofsingle-sized aggregate particles is rolled into thesurface to provide the required macrotexture.These are precoated with a film of bitumen toimprove adhesion to the asphalt surface and areknown as coated chippings. Macadams on theother hand have a lower proportion of fines andfiller and provide a rough surface.
Macrotexture is measured in terms of texturedepth using the sand patch test (BS 598: Part 105:1990). The test involves spreading a knownvolume of sand in a circular patch over the roadsurface until it can be spread no further. The sandfills the spaces between aggregate particles asshown in Figure 25.4. The diameter of the patchis measured and, knowing the volume of sand,the average depth can be calculated.
Microtexture refers to the surface texture ofindividual particles and varies according to thetype of aggregate. Here it is important not only touse an aggregate which has a rough surfacetexture, but also which will retain that texture
under the action of traffic. This is assessed usingthe polished stone value test (BS 812: Part 114:1989). Here samples of aggregate are subjected toa simulated wear test, where a pneumatic tyreruns continuously over the aggregate particlesunder the abrasive action of emery powder. Theskidding resistance of the samples is determinedafter the test using the pendulum skid tester.
25.5 ReferenceDickie, T.P. and Yen, T.F. (1967) Macrostructure of
asphaltic fractions by various instrumental methods.Analytical Chemistry 39, pp. 13–16.
Girdler, R.B. (1965) Constitution of asphaltenes andrelated studies. Proceedings of the Association ofAsphalt Paving Technologists 34, p. 45.
Shell Bitumen UK (1990) The Shell BitumenHandbook, Shell Bitumen UK.
Speight, J.G. and Moschopedis, S.E. (1979) Someobservations on the molecular ‘nature’ of petroleumasphaltenes. American Chemical Society, Division ofPetroleum Chemistry 24, pp. 22–5.
Structure of bituminous materials
236
Chapter twenty-six
Viscosity anddeformation ofbituminousmaterials
26.1 Viscosity and rheology of binders26.2 Measurement of viscosity26.3 Influence of temperature on viscosity26.4 Resistance of bitumens to deformation26.5 Determination of permanent deformation26.6 Factors affecting permanent deformation26.7 References
26.1 Viscosity and rheology ofbindersThe viscosity of a liquid is the property thatretards flow so that, when a force is applied tothe liquid, the higher the viscosity, the slower willbe the movement of the liquid. The viscosity ofbitumen is dependent upon both its chemicalmake-up and its structure. In sol-type bitumens,the asphaltene micelles are well dispersed withinthe maltene continuum. The viscosity depends onthe relative amounts of the asphaltenes andmaltenes, decreasing as the asphaltene contentreduces. In gel-type bitumens, where the asphal-tene micelles have aggregated, the viscosity ishigher and dependent upon the extent of theaggregation. The degree of dispersion of theasphaltenes is controlled by the relative amountsof resins, aromatics and saturated oils. If thereare sufficient aromatics they form a stabilizinglayer around the asphaltene micelles, promoting
the dispersion. However, if they are not presentin sufficient quantity the micelles will tend to jointogether. A schematic representation of the twostates is shown in Figure 26.1 (The Shell BitumenHandbook, 1990). In practice most bitumens aresomewhere between these two states. Themaltene continuum is influenced by the saturatedoils which have low molecular weight and, con-sequently, a low viscosity. These saturates havelittle solvent power in relation to the asphaltenes,so that as the saturate fraction increases, there isa greater tendency for the asphaltenes to aggreg-ate to form a gel structure. Thus a high propor-tion of saturates on the one hand tends to reduceviscosity because of their low molecular weight,but on the other hand encourages aggregation ofthe asphaltene micelles which increases viscosity.The relative importance of these two opposingeffects depends on the stabilizing influence on theasphaltenes of the aromatics.
The asphaltenes exert a strong influence onviscosity in three ways. First, the viscosityincreases as the asphaltene content increases.Second, the shape of the asphaltene particlesgoverns the extent of the change in viscosity. Theasphaltene particles are thought to be formedfrom stacks of plate-like sheets of aromatic/naphthenic ring structures. These sheets are held
237
together by hydrogen bonds. However, theasphaltenes can also form into extended sheetsand combine with aromatics and resins so thatparticle shape varies. Third, the asphaltenes maytend to aggregate, and the greater the degree ofaggregation the higher is the viscosity.
26.2 Measurement of viscosityAn absolute measure of viscosity can be deter-mined using the sliding plate viscometer. If a thinfilm of bitumen is held between two parallelplane surfaces, and one surface is moved parallelto the other, the movement is resisted by thebitumen, according to its viscosity. The force of
Viscosity and deformation of bituminous materials
238
FIGURE 26.1 The structure of bitumen: (a)schematic diagram of a sol-type bitumen; (b) schematicdiagram of a gel-type bitumen (The Shell BitumenHandbook, 1990).
resistance, F, depends on the area of the surfaces,A, the distance between them, d, and the speed ofmovement of one plate relative to the other, V,such that:
F� (26.1)
The factor is the coefficient of viscosity, and isgiven by:
� "
� (26.2)
This is known as the dynamic viscosity and hasunits of Pa sec. Viscosity may also be measured inunits of mm2/sec. This unit refers to kinematicviscosity and is related to dynamic viscosity bythe expression:
Kinematic viscosity� (26.3)
However, it is not necessary to know absoluteviscosity, and a number of tests have beendeveloped which are empirical and provide arbi-trary relative measures of viscosity. The two mostcommon measures of viscosity are the softeningpoint test and the standard tar viscometer.
The softening point (BS 2000: Part 58: 1983) isthe temperature at which a bitumen reaches aspecified level of viscosity. This viscosity isdefined by the ring and ball test apparatus as theconsistency at which a thin disc of bitumen flowsunder the weight of a 10mm diameter steel ballby a distance of 25mm. Figure 26.2 shows a dia-grammatic representation of the test. The moreviscous the bitumen, the higher the temperatureat which this level of viscosity is reached.
The standard tar viscometer, as the nameimplies, is normally used to assess the viscosity oftars. Tars have a greater range of viscosity thanbitumens, extending to lower values, and the testallows for this range by permitting a variation intest temperature. The test involves the measure-ment of the time taken for 50ml of the tar to runout of a cup through a standard orifice (BS 2000:
Dynamic viscosity���
Mass density
Shear stress��Rate of strain
V�d
F�A
AV�
d
Part 72: 1993). Results obtained at different testtemperatures selected to ensure that the effluxtime lies between 10 and 140sec are related to acommon equiviscous temperature scale. The vis-cosity defined for this scale is that at which 50mlflows out in 50sec. The test is not used for pene-tration grade bitumens since these are too viscousto reach the defined viscosity except at very hightemperatures which cannot confidently be con-verted to the empirical equiviscous temperaturescale. However, the test is used for cutback bitu-mens. The efflux time depends on kinematic vis-cosity, and the test results may be converted toabsolute viscosity in Pasec according to:
�flow time�density�constant. (26.4)
For cutback bitumens, using a cup with a 10mmorifice, the constant is 0.40.
Another test which is commonly applied tobitumens, and is the basis for their characteriza-tion, is the penetration test. The test measureshardness, but this is related to viscosity. The testconsists of measuring the depth to which a needlepenetrates a sample of bitumen under a load of
Influence of temperature on viscosity
239
FIGURE 26.2 Apparatus for ring and ball test.
100g over a period of 5 sec at a temperature of25°C. Thus the test differs from the previous twoin that, rather than determining an equiviscoustemperature, the viscosity is determined at aparticular temperature.
However, because bitumen is viscoelastic, thepenetration will depend on the elastic deforma-tion as well as the viscosity. Therefore, since vis-cosity changes with temperature, differentbitumens may have the same hardness at 25°Cbut different hardnesses at other temperatures. Itis the varying elasticity of bitumens which pre-vents correlation between these empirical tests.
26.3 Influence of temperature onviscosityBitumens are thermoplastic materials so that theysoften as the temperature rises but become hardagain when the temperature falls. The extent ofthe change in viscosity with temperature variesbetween different bitumens. It is clearly import-ant, in terms of the performance of a bitumen inservice, to know the extent of the change in vis-cosity with temperature. This is referred to astemperature susceptibility and, for bitumens, isdetermined from the penetration value, P, andsoftening point temperature, T. These are relatedempirically by the expression:
logP�AT�k (26.5)
where A is the temperature susceptibility of thelogarithm of penetration and k is a constant.From this relationship, an expression has beendeveloped (Pfeiffer and Van Doormaal, 1936)which relates A to an index, known as the pene-tration index (PI), such that for road bitumensthe value of PI is about zero.
A� � · (26.6)
It has been determined that, for most bitumens,the penetration at their softening point (SP) tem-perature is about 800. Thus if the penetration at25°C and the softening point temperature areknown, the PI can be evaluated from:
1�50
20�PI�10�PI
d(logP)�
dT
� (26.7)
� ·
For example, for a 50 pen bitumen with a soften-ing point of 48°C:
� (26.8)
� �0.0523
Therefore:
0.0523� ·
giving:
PI�1.7
Pfeiffer and Van Doormaal produced a nomo-graph (Figure 26.3) to evaluate the above expres-sion, and it can be seen that for the aboveexample a similar result is obtained.
Bitumens for road use normally have a PI inthe range �2 to �2. If the PI is low, bitumens aremore Newtonian in their behaviour and becomevery brittle at low temperatures. High PI bitu-mens have marked time-dependent elastic proper-ties and give improved resistance to permanentdeformation.
The influence of chemical composition on tem-perature susceptibility is illustrated in Figure26.4. In general the PI increases as theasphaltenes increase at the expense of the aromat-ics. This change can be achieved by controlled airblowing. Heavy duty bitumen (40 pen) is pro-duced by this technique and has a PI of about2.5. As a consequence, mixes incorporating thisbitumen have a better resistance to deformation.
26.4 Resistance of bitumens todeformationSince bitumen is a viscoelastic material, theresponse to an applied load depends on the sizeof the load, the temperature, and the duration of
1�50
20�PI�10�PI
1.204�
23
log800� log50��
48�25
d(logP)�
dT
1�50
20�PI�10�PI
log800� logP��
SP�25
d(logP)�
dT
Viscosity and deformation of bituminous materials
240
FIGURE 26.3 Nomograph to evaluate penetrationindex from softening point and penetration (Pfeifferand Doormaal, 1936).
100
90
80
70
60
50
40
30
20
10
0�2 �1 0 1 2 3 4 5 6 7 8 9
Saturates
Aromatics
Resins
Asphaltenes
Penetration index
Perc
enta
ge
FIGURE 26.4 Relationship between chemical com-position and penetration index (Lubbers, 1985).
its application. In other words there is no simplerelationship between stress and strain. Thereforeit is difficult to predict the elastic modulus ofbitumen. However, this modulus is of crucialimportance in determining the resistance of abituminous mixture to permanent deformation.Therefore if the performance of a mix is to beassessed, it is necessary to obtain a value forelastic modulus. Van der Poel (1954) has intro-duced the concept of stiffness modulus to takeaccount of the viscoelastic nature of bitumen.This modulus is dependent on both temperatureand time of loading, and is given by:
St,T � (26.9)
where � is the tensile stress and �t,T is the resul-tant strain after loading time t at temperature T.Figures 26.5 and 26.6 illustrate the effect ofloading time and temperature for bitumens of dif-ferent PI. For low PI bitumens (Figure 26.5) thestiffness is constant for very short loading timesand virtually independent of temperature. Thisrepresents elastic behaviour. For longer loadingtimes the curves have a consistent slope of 45°and have a significant variation with temperatureindicating viscous behaviour. The effect ofincreasing PI can be seen by comparing Figures26.5 and 26.6. High PI bitumens are much stiffer
���t,T
Determination of permanent deformation
241
FIGURE 26.5 The effect of temperature andloading time on stiffness of a low PI bitumen (The ShellBitumen Handbook, 1990).
FIGURE 26.6 The effect of temperature andloading time on stiffness of a high PI bitumen (TheShell Bitumen Handbook, 1990).
at high temperatures and longer loading times.Thus under conditions which are more likely togive rise to deformation, namely slow-moving orstationary traffic and high temperatures, a high PIbitumen offers greater resistance to deformationby virtue of its higher stiffness.
When considering a bituminous mix, consistingof a graded aggregate bound with bitumen, thestiffness of the mix is dependent on the stiffnessof the bitumen and the quantity and packing ofaggregate in the mix. The quantity and packingof aggregate particles depends on grading, parti-cle shape and texture, and method of com-paction.
26.5 Determination of permanentdeformationRutting of bituminous pavements is the mostcommon type of failure in the UK. It is thereforeimportant to be able to predict the deformationfor a bituminous mix and this depends on the lowstiffness response, that is the stiffness at long loadtimes. Two tests which have been commonly usedto determine fundamental values of stiffness arethe creep test and the repeated load triaxial test.
In the creep test, a uniaxial load of 0.1MPa isapplied to a cylindrical specimen for 1 hour at
40°C. During the test, deformation is measuredas a function of time. The stiffness, Smix, of themix may then be determined from:
Smix � (26.10)
Although simple, and giving good correlationwith rutting measurements, the test does notemploy a confining stress. In situ materials willclearly be confined and the effect of the confiningstress on the vertical strain may be important.
The repeated load triaxial test overcomes thisdisadvantage and simulates more closely theactual conditions of stress in a bituminous pave-ment layer. This test is similar to that used in soilmechanics. A cylindrical specimen with aheight/diameter ratio of at least 2.5:1 is subjectedto a static confining pressure whilst a cyclic verti-cal stress is applied. The resulting vertical andhorizontal strains can be measured and stiffnessdeduced. However, the good correlation betweencreep and rutting tests has led to the creep testbeing more commonly used.
A model has been developed (Hills et al., 1974)linking creep results and rutting as follows:
R� (26.11)
where R� rut depth, Cm �correlation factorvarying between 1 and 2, H� layer thickness,�av �average stress in the pavement related towheel load and stress distribution, and Smix � stiff-ness of the mix as determined from the creep test.
26.6 Factors affecting permanentdeformation
26.6.1 Bitumen viscosity
When a stress is applied to a bituminous material,both the aggregate particles and the bitumen willbe subjected to the stress. But the aggregate parti-cles, being hard and stiff, will undergo negligiblestrain, whereas the bitumen, being soft, willundergo considerable strain. Thus the deforma-tion is associated with movement in the bitumenand the extent of the movement will depend on
Cm �H� �av��
Smix
Applied stress��
Axial strain
Viscosity and deformation of bituminous materials
242
Softening point (ring and ball) (°C)W
heel
-trac
king
tes
t: r
ate
of t
rack
ing
(mm
/h a
t 45
°C)
10
5
3
2
1
0.5
0.3
0.2
20
30
30 40 50 60 70 80 90
Rolled asphalt30% coarse aggregate
200 pen
100 pen
50 pen
35 pen
HD Bitumen (40 pen)Bitumen � EVA
Trinidadlake epuré
(soluble part)
FIGURE 26.7 Effect of softening point of binder onresistance to deformation (Szatkowski, 1980).
its viscosity. Figure 26.7 shows how the rate ofdeformation, as measured by the wheel-trackingtest, varies with the softening point of thebitumen. It can be seen that a reduction in soften-ing point of 5°C will approximately double thedeformability.
26.6.2 Aggregate
Bituminous mixtures which utilize a continuouslygraded aggregate – macadams – rely mainly onaggregate particle interlock for their resistance todeformation. Thus the grading and particle shapeof the aggregate are major factors governingdeformation. The characteristics of the fineaggregate are particularly important in gap-graded materials which rely on a dense bitumenand fine mortar for their strength. These are theasphalt mixes. Sand particles can vary consider-ably from spherical glassy grains in dune sands,to angular and relatively rough grains from somepits. Mixes made with a range of sands all at thesame bitumen content have been shown to givedeformations, when tested in the laboratorywheel-tracking test, that varied by a factor of 4from the best to the worst sand (Knight et al.,1979).
26.6.3 Temperature
Figure 26.8 shows permanent strain againstnumber of test cycles in a repeated load triaxialtest. It can be seen that permanent strainincreases with temperature. This is due to thereduction in viscosity of bitumen and consequentreduction in bitumen stiffness. The figure alsoindicates the effect of the aggregate grading. Atlow temperatures, the permanent strain in theasphalt and macadam are the same. Here the highdegree of aggregate particle interlock in themacadam and the high viscosity bitumen in theasphalt provide a similar resistance to deforma-tion. However, at higher temperatures, theasphalt deforms more because its stiffness isreduced by the reduced bitumen viscosity, whichis not compensated by the aggregate interlockeffect. In the macadam, although the bitumen willalso be less stiff, the aggregate grading continuesto provide a compensating resistance to deforma-tion.
26.7 ReferencesBrown, S.F. (1978) Material characteristics for analyti-
cal pavement design. In Developments in HighwayPavement Engineering – I, Applied Science Publish-ers, London.
Hills, J.F., Brien, D. and Van de Loo, P.J. (1974) TheCorrelation of Rutting and Creep Tests on AsphaltMixes, Institute of Petroleum, IP-74–001.
Knight, V.A., Dowdeswell, D.A. and Brien, D. (1979)Designing rolled asphalt wearing courses to resistdeformation. In Rolled Asphalt Road Surfacings,ICE, London.
Lubbers, H.E. (1985) Bitumen in de weg – en water-bouw. Nederlands Adviesbureau voor bitumen-topassingen.
Pfeiffer, J.Ph. and Van Doormaal, P.M. (1936) Therheological properties of asphaltic bitumens. Journalof Institute of Petroleum 22.
Shell Bitumen UK (1990) The Shell BitumenHandbook.
Szatkowski, W.S. (1980) Rolled asphalt wearingcourses with high resistance to deformation. InRolled Asphalt Road Surfacings, ICE, London.
Van der Poel, C. (1954) A general system describingthe viscoelastic properties of bitumen and its relationto routine test data. Journal of Applied Chemistry 4,224–36.
References
243
FIGURE 26.8 Comparison of permanent strain fordense macadam and hot rolled asphalt base mixes(Brown, 1978).
244
Chapter twenty-seven
Strength andfailure ofbituminousmaterials
27.1 The road structure27.2 Modes of failure in a bituminous structure27.3 Fatigue characteristics27.4 References
27.1 The road structureA flexible road structure consists of a number oflayers of different materials, as illustrated inFigure IV.2 of the Introduction. In structuralterms, the purpose of the road is to distribute theapplied load from the traffic to a level which theunderlying subgrade can bear. The stressesinduced by the loads are high at the surface butreduce with depth. Thus, the surfacing materialmust be of high quality, but at greater depthsbelow the surface, economies can be achieved byusing materials of lower strength.
27.2 Modes of failure in abituminous structureRoads deteriorate in a number of ways, butbroadly there are two forms of failure. First, theroad surface may deteriorate. This may bethrough breakdown of the surface material eithergenerally, for example through fretting or stoneloss, or locally when a pot hole develops as a
result of a local weakness. Alternatively, thesurface texture of the wearing course may bereduced, through polish or abrasion, so that theskidding resistance drops below an acceptablelevel.
Second, the road structure deteriorates. It isthis structural deterioration which will be dis-cussed here. The key feature of such deteriorationis that it is gradual, and develops with the con-tinued application of wheel loads. In the earlystages, the rate of deterioration is very small andthe structural changes are not perceptible and aredifficult to measure. But with continued service,signs of structural change become clearer and therate of deterioration accelerates. There are twomodes of breakdown which are illustrated inFigure 27.1. First, permanent deformation occursin the wheel tracks. This ‘rutting’ is associatedwith deformation of all the pavement layers andis linked to a loss of support from the underlyingsubgrade. Deformation within the bituminouslayers is an accumulation of the small irrecover-able part of the deformation caused by eachwheel load. It is a function of the viscoelasticnature of the bitumen together with the mechani-cal support offered by the grading of the aggreg-ate. The second mode of failure is cracking,which appears along the wheel tracks. The crack-
ing is caused by the tensile strain developed in thebound layers as each wheel load passes. It istherefore a function of both the size of tensilestrain, and the repetitive nature of the loading;that is a fatigue failure. It is important to note, asFigure 27.1 shows, that the cracking initiates atthe base of the bound layer. This is where thetensile stresses are highest, as shown in Figure27.2. It follows that, by the time the cracking isvisible at the surface, the damage has beenpresent for some time.
In both modes of failure, the breakdown iscaused by (a) the repetitive nature of the loading,and (b) by the development of excessive strains inthe structure. This leads to the notion that, iffailure can be defined, the life of a road can bedetermined provided that the loading can beassessed and the performance of the materials eval-uated. Alternatively, the structural design of theroad together with the make-up of the materials togive required properties may be determined, inorder to provide a given life. In either case, it isnecessary to define failure. In most civil engi-neering structures, structural failure renders thestructure unusable and is often associated with col-lapse. However, whilst roads may become lesscomfortable to drive on and less safe, they do not,except in very extreme cases, become unusable.Therefore failure for roads must be identified interms of serviceability and/or repairability; that is,
Modes of failure in a bituminous structure
245
FIGURE 27.1 Modes of failure and critical strainsin a flexible pavement (Brown, 1980).
FIGURE 27.2 Vari-ation of vertical andradial stresses belowthe centreline of a40kN wheel loadacting over a circulararea of radius 160mm,with a contact pressureof 0.5MN/m2 (Peattie,1978).
the extent of cracking and deformation must bedefined which is just acceptable to drivers and/orwhich represents a condition which may beeconomically restored by repair. A definition isprovided, in terms of both modes of failure, whichidentifies three conditions: sound, critical, andfailed. Table 27.1 shows that if any cracking isvisible at the surface, then the road is regarded asbeing at critical condition or as having failed. Herethe term ‘critical’ means that failure is imminentbut the road still has sufficient structural capabilityto support strengthening and to provide anextended life from the strengthened road. If thereare no signs of cracking, then the condition isdefined in terms of the extent of permanent defor-mation. Thus, if the rut depth reaches 20mm, theroad is regarded as having failed.
27.3 Fatigue characteristicsThe development of permanent deformation hasalready been discussed. Here the fatiguecharacteristics of bituminous mixes, leading tocracking, will be examined. Fatigue crackingarises from the fact that, under repeated applica-tions of tensile stress–strain, a bituminous mater-ial will eventually fracture. The higher the level ofstress and strain applied, the smaller the numberof load applications before cracking occurs. For aparticular level of stress and strain, the mix pro-portions and nature of the bitumen dictate thenumber of cycles before cracking occurs.
A number of laboratory tests have beendeveloped to assess the fatigue characteristics of
Strength and failure of bituminous materials
246
TABLE 27.1 Criteria for determining pavement condition (Department of Transport, 1983, Advice NoteHA/25/83. Reproduced with permission of the Controller of HMSO. © Crown copyright).
Wheel-track rutting under a 2m straight edge
Wheel-track Less than From 5mm From 10mm 20mm orcracking 5mm to �10mm to �20mm greater
None Sound Sound Critical Failed� half width or Critical Critical Critical Failed
single crack� half width Failed Failed Failed Failed
FIGURE 27.3 Methods for fatigue testing of bitu-minous materials (Brown, 1978).
bituminous materials. The tests, illustrated inFigure 27.3, are flexure tests and simulate therepeated bending action in the stiff bound layerof a pavement caused by the passage of eachwheel load. The number of load cycles which aparticular specimen can endure before failuredepends on a number of factors discussed below.
27.3.1 Stress and strain conditions
Fatigue tests may be conducted in two ways.They may be constant stress tests, where eachload application is to the same stress level regard-less of the amount of strain developed. Alterna-tively they may be constant strain tests, whereeach load application is to the same strain levelregardless of the amount of stress required.
These two alternatives produce quite differentresults. Figure 27.4 (a) shows the general patternof results from constant stress tests. Each line rep-resents a different test temperature, i.e. a differentstiffness, and it can be seen that mixes withhigher stiffness have longer lives. Figure 27.4 (c)shows the general pattern of results from con-stant strain tests. Again each line represents a dif-ferent temperature or stiffness and it can be seenthat the outcome is reversed, with the mixes ofhigher stiffness having the shortest lives. Thiscontrast may be explained in terms of the failuremechanism. Cracks initiate at points of stress
concentration and propagate through the mater-ial until fracture occurs. If the stress level is keptconstant, the stress level at the tip of the crackcontinues to be high so that propagation is rapid.However, in a constant strain test, the develop-ment of a crack causes a steady reduction in theapplied stress level because the cracks contributemore and more to the strain as they propagate.Thus the stress at the crack tips reduces and rateof propagation is slow. Thus it is important toestablish which test condition is most relevant toactual pavement behaviour. It has been shown(Monismith and Deacon, 1969) that straincontrol is appropriate to thin layers (for examplesurfacing layers), whereas stress control is appro-priate to thicker structural layers. This is becausepavements are subject to a stress controlledloading system, so that the main (and normallythick) structural layers are stress controlled.However, the thin surface layer must move withthe lower structural layers and so is effectivelysubject to strain control. Nevertheless, under low
Fatigue characteristics
247
FIGURE 27.4 Fatigue lines representing number of cycles to failure, Nf, under different test conditions(Brown, 1980).
temperature conditions giving high stiffness,crack propagation is relatively quick even understrain controlled loading so that the differencebetween the two loading conditions is small.
27.3.2 The strain criteria
If the results of a controlled stress test areexpressed in terms of an equivalent strain thenthe log–log plot of strain against number of loadcycles produces a single linear relationship for alltest conditions for a particular mix, as shown inFigure 27.4 (b). In other words the relationship isindependent of mix stiffness. This suggests thatstrain is the principal criterion governing fatiguefailure, and it has been demonstrated (Cooperand Pell, 1974) that flexure tests on a wide rangeof mixes produce unique fatigue lines for eachmix. The general relationship defining the fatigueline is:
Nf �C�1
���
m
(27.1)
where Nf is the number of load cycles to initiate afatigue crack, � is the maximum applied tensilestrain, and C and m are constants depending onthe composition and properties of the mix.
The fatigue lines for a range of mixes areshown in Figure 27.5 together with details of themix compositions in Table 27.2.
27.3.3 Effect of mix variables
A large number of variables associated with themix affect the fatigue line. However, it has beenshown (Cooper and Pell, 1974) that two variablesare of prime importance. These are:
1. the volume of bitumen in the mix;2. the viscosity of the bitumen as measured by
the softening point.
As the volume of bitumen increases up to 15 percent, the fatigue life increases, and as the bitumenbecomes more viscous, with softening pointincreasing up to 60°C, the fatigue life alsoincreases.
Other factors are important insofar as theyaffect the two main variables. The void content ofthe mix has an effect on the volume of bitumen.The total void content is, in turn, affected by theparticle shape and grading of the aggregate, thecompactive effort, and the temperature. In otherwords there is a link between workability, com-pactive effort and void content which is con-
Strength and failure of bituminous materials
248
FIGURE 27.5 Fatigue lines under controlled stress conditions for a range of bituminous mixes (Brown, 1980).
Fatigue characteristics
249
TABLE 27.2 Details of mixes represented in Figure 27.5 (Brown, 1980)
Description Coarse Fine Filler Binder Mean void Fatigue lineof mix aggregate aggregate (% by mass) (% by mass) content (%) constants
(% by mass) (% by mass)C m
Mastic asphalt 42 23 20 15:70/30 0 1.1�10�15 5.5wearing crushed rock limestone limestone TLA1/20course pen bit.Rolled asphalt 30 53.2 8.9 7.9 2.9 1.3�10�14 5.1wearing crushed rock sand limestone 45 pencourseRolled asphalt 65 29.3 – 5.7 4.0 3.2�10�8 3.2basecourse crushed rock sand 45 penDense bitumen 62 28.6 4.7 4.7 6.8 2.0�10�11 3.8macadam crushed rock crushed rock crushed rock 100 penbasecourseDense bitumen 62.3 28.7 4.7 4.3 6.9 2.5�10�12 4.0macadam crushed rock crushed rock crushed rock 200 penbasecourseDense tar 61.7 28.4 4.7 5.2 7.5 1.0�10�7 2.7macadam crushed rock crushed rock crushed rock B54basecourse
Notes:1. Trinidad Lake asphalt.
FIGURE 27.6 The influence of voids on fatigue life.
trolled by the bitumen content. However, whilst ahigher bitumen content improves fatigue life, italso reduces stiffness which leads to increasedstrain. Figure 27.6 illustrates the double influenceof void content. Figure 27.6 (a) shows the effectthat an increased bitumen content has on stiff-ness. The stiffness is reduced which increases the
strain under constant stress conditions causing ashift to the left along the fatigue line. Thus thefatigue life is reduced. Figure 27.6 (b) shows theinfluence that void content has on the fatigue lineso that, for the same strain, the fatigue life isreduced as void content increases. The change inposition of the fatigue line corresponds to a
change in material type, as was seen in Figure27.5, whereas the shift along a fatigue line due toa stiffness change corresponds to a change indegree of compaction. However, in practice, botheffects occur if the change in void content is asso-ciated with a change in bitumen content.
27.4 ReferencesBrown, S.F. (1978) Material characteristics for analyti-
cal pavement design. In Developments in HighwayPavement Engineering – 1, Applied Science Publish-ers, London.
Brown, S.F. (1980) An Introduction to the AnalyticalDesign of Bituminous Pavements, Department ofCivil Engineering, University of Nottingham.
Cooper, K.E. and Pell, P.S. (1974) The Effect of MixVariables on the Fatigue Strength of BituminousMaterials. Transport and Road Research Labora-tory, Laboratory Report 633, Department of Trans-port (1983) Departmental Advice Note HA/25/83.
Monismith, C.L. and Deacon, J.A. (1969) Fatigue ofasphalt paving mixtures. Journal of TransportEngineering Division, ASCE 95, 154–61.
Peattie, K.R. (1978) Flexible pavement design. InDevelopments in Highway Pavement Engineering,Applied Science Publishers, London.
Strength and failure of bituminous materials
250
Chapter twenty-eight
Durability ofbituminousmixtures
28.1 Introduction28.2 Ageing of bitumen28.3 Permeability28.4 Adhesion28.5 References
28.1 IntroductionDurability is the ability to survive and continue togive an acceptable performance. In the case ofroads, it is necessary that the structure shouldsurvive for the specified design life, although it isaccepted that not all aspects of performance canbe sustained for this duration without somerestorative maintenance. The design guide for UKroads suggests a design life of 40 years (Depart-ment of Transport, 1987), although this can onlybe achieved in stages, as shown in Figure 28.1.The durability of a flexible road structuredepends on the durability of the materials fromwhich it is constructed, in particular the bitumen-bound materials. Bituminous materials may dete-riorate in a number of ways. The bitumen itselfwill harden with exposure to oxygen and temper-ature effects, the aggregate may not be of suffi-cient quality so that some individual particlesmay break down, or there may be loss of adhe-sion between the bitumen and aggregate particles.These forms of deterioration are caused byweathering and the action of traffic. These agentsact at the road surface which is particularly
vulnerable. However, deterioration can alsooccur in the body of the material and this is con-trolled by its permeability.
28.2 Ageing of bitumenThe ageing or hardening of bitumen is aninevitable result of exposure of bitumen to theatmosphere. The rate of hardening will dependon the conditions and the nature of the bitumen.There are two main processes which occur: oxi-dation and loss of volatiles.
28.2.1 Oxidation
In the oxidation process, oxygen molecules fromthe air combine with the resins to form
251
FIGURE 28.1 The life of a flexible road.
asphaltenes. Thus there is an increase in thepolar, high molecular weight fraction at theexpense of the lower molecular weight resins. Theresult is an increase in viscosity of the bitumen.Also, the bitumen becomes unstable due to thediscontinuity which develops between the satu-rates and the rest of the components. This insta-bility causes a lack of cohesion within thebitumen which will lead to cracking. The rate ofoxidation is highly dependent on temperature,and is rapid at the high temperatures used formixing and laying bituminous materials.
28.2.2 Loss of volatiles
Loss of volatiles will occur if there is a substantialproportion of low molecular weight componentsin the bitumen, and if the bitumen is subjected tohigh temperatures. However, for penetrationgrade bitumens, the loss of volatiles once thematerial has been laid is relatively small.
28.2.3 Ageing index
The hardening of bitumen results in a lowering ofpenetration, an increase in softening point and anincrease in penetration index. Therefore an exces-sive amount of hardening will cause the materialto be brittle at low temperatures and vulnerableto cracking. A convenient means of evaluating theability of a bitumen to resist hardening is to usethe microfilm durability test (Griffin et al., 1955).The test consists of ageing films of bitumen 5 microns thick on glass plates in an oven at225°F for two hours. The hardening whichoccurs is determined by measuring the viscosityof the material before and after this exposurewith the sliding plate microviscometer. The ratioof the viscosity of the aged asphalt to that of theoriginal asphalt is the ageing index.
Ageing index is strongly influenced by theinitial molecular weight of the asphalt. Figure28.2 shows that ageing index can be large inasphalts with low molecular weight. A satisfac-tory durability (ageing index less than 10) can beobtained if components with a molecular weightless than 400 are eliminated.
Durability of bituminous mixtures
252
FIGURE 28.2 Influence of initial molecular weighton durability (Griffen et al., 1959).
In practice, the ageing of bitumen is mostmarked during the mixing process because hightemperatures are involved. For example, the pene-tration value of a 50 pen bitumen will fall tobetween 30 and 40 depending on the duration ofthe mixing and the temperature used; subsequenthigh temperature storage will cause further ageing.Thus the penetration value could be reduced by asmuch as half. Ageing of bitumen on the road isgenerally a much slower process. This is becausethe temperatures are much lower and the availabil-ity of oxygen is restricted by the permeability ofthe mix. In more open-textured mixes with a largevolume of interconnected voids, air can readilypermeate the material allowing oxidation to occur.Macadams generally fall into this category.However, in dense mixes such as asphalts, the per-meability is low and there will be very little move-ment of air through the material. In both cases,ageing will be more rapid at the surface than in thebulk of the material because there is a continualavailability of oxygen and the surface will reachhigher temperatures.
28.3 PermeabilityPermeability is an important parameter of a bitu-minous mixture because it controls the extent towhich both air and water can migrate into thematerial. The significance of exposure to air was
described in the previous section. Water may alsobring about deterioration by causing the bitumento strip from the aggregate particles, or causingbreakdown of the aggregate particles themselves.Furthermore, permeability controls the extent offrost damage. Thus permeability may be regardedas a measure of durability.
28.3.1 Measurement and voids analysis
The measurement of permeability is, in essence, asimple task, achieved by applying a fluid underpressure to one side of a specimen of a bituminousmixture and measuring the resulting flow of fluidat the opposite side. Both air and water have beenused as the permeating fluid. A permeability celldeveloped for the measurement of permeability ofconcrete, shown in Figure 28.3, has been used suc-cessfully to measure the permeability of hot rolledasphalt to oxygen (Robinson, 1991). Here a con-stant fluid pressure is maintained. By contrast, afalling head permeameter has been used (Khalid,1990) with water as the permeating fluid. Morerecently an air permeability test (Cabrera andHassan, 1994) has been described which can beused both in the laboratory and in the field.
28.3.2 Factors affecting permeability
The permeability of a bituminous mixturedepends on a large number of factors. Of particu-
lar importance is the quantity of voids, the distri-bution of void size and the continuity of thevoids. Figure 28.4 shows how permeability varieswith total voids in the mix for an open-texturedbituminous macadam. It can be seen that the per-meability is more sensitive to void content at highbinder contents. This is likely to be due to the sizeand continuity of pores. At low binder contents,the pores are large and although a reduction inthe volume of voids will tend also to reduce thepore size, the size will remain sufficiently large topermit high permeability. At high binder con-tents, however, the pores are relatively small anda reduction in the volume of voids will cause asignificant change in pore size with a muchgreater effect on permeability. A change in con-tinuity of the voids is also more likely at higherbinder contents when pores will be plugged withbitumen as the volume of voids reduces.
The voids are also affected by the nature of theaggregate. The shape, texture and grading of theparticles will govern the packing and hence voidcontent at a particular bitumen content. Theamount of compactive effort employed is alsoimportant.
28.4 AdhesionThe quality of the adhesion of a bitumen to anaggregate is dependent on a complex assemblageof variables. Table 28.1 identifies a number of
Adhesion
253
FIGURE 28.3 Schematic diagram of permeability test cell.
factors which have an influence on the adhesionperformance of bituminous mixes. Althoughsome of these relate to the ambient conditionsand aspects of the mix as a whole, the principalfactors are the nature of the aggregate and, to alesser extent, the bitumen.
28.4.1 The nature of the aggregate
The mineralogical and physical nature of theaggregate particles has an important bearing onadhesion, with the adhesive capacity being afunction of chemical composition, shape andstructure, residual valency, and the surface areapresented to the bitumen.
Generalizations about the effect of mineralogyare difficult because the effects of grain size,shape and texture are also important. However,there seem to be a preponderance of reports offailures with the more siliceous aggregates such asgranites, rhyolites, quartzites, siliceous gravel andcherts. The fact that good performance with thesematerials has also been reported, and that failuresin supposedly good rock types such as limestonesand basic igneous rocks have occurred, emphas-izes the complexity of the various material inter-actions. Therefore caution should be exercisedwhen attempting to make generalizations on the
Durability of bituminous mixtures
254
FIGURE 28.4 Relationship between the coefficientof permeability, k, and voids content of an open-textured bituminous macadam mix (Khalid, 1990).
TABLE 28.1 Material properties and external influences which can act singularly or together to affect the adhe-sion and stripping resistance of a bituminous mix
Aggregate properties Bitumen properties Interactive mix properties External influences
Mineralogy Composition and source Compaction Annual precipitation*Surface texture Durability and weathering Grading Relative humidity*Porosity Viscosity Permeability pH of water*Surface coatings and dust Curing time Binder content Presence of salts*Mechanical durability Oxidation Cohesion Temperature*Surface area Electrical polarity Film thickness Temperature cycling*Absorption Use of additives Filler type Light, heat and radiation*Moisture content Type of mix TrafficAbrasion pH Method of production Construction practiceWeathering grade Use of additive DesignExposure history WorkmanshipShape DrainageAdditives
* Factors considered uncontrollable.
adhesion performance of aggregates of differentor even similar mineralogy.
The surface character of each individual aggreg-ate type is important particularly in relation to thepresence of a residual valency or surface charges.Aggregates with unbalanced surface chargespossess a surface energy which can be attributed to a number of factors, including brokencoordination bonds in the crystal lattice, the polarnature of minerals, and the presence of adsorbedions. Such surface energy will enhance the adhesivebond if the aggregate surface is coated with aliquid of opposite polarity.
Absorption of bitumen into the aggregatedepends on several factors, including the totalvolume of permeable pore space, and the size ofthe pore openings. The presence of a finemicrostructure of pores, voids and microcrackscan bring about an enormous increase in theabsorptive surface available to the bitumen. Thisdepends on the petrographic characteristics of theaggregate as well as its quality and state of weath-ering.
It is generally accepted that rougher surfacesexhibit a greater degree of adhesion. A balance is,however, required between the attainment of goodwettability of the aggregate (smooth surfaces beingmore easily wetted), and a rougher surface whichholds the binder more securely once wetting isachieved. The presence of a rough surface texturecan mask the effects of mineralogy.
28.4.2 The nature of the bitumen
The important characteristics of the bitumenaffecting adhesion to aggregate are its viscosityand surface tension, and its polarity.
The viscosity and surface tension will governthe extent to which bitumen is absorbed into thepores at the surface of the aggregate particles.Both these properties are altered with tempera-ture, and mixing of aggregate and bitumen isalways done at high temperature – up to 180°Cfor 50 pen bitumen – in order that the bitumencoats the aggregate surface readily.
Bitumen will also chemically adsorb ontoaggregate surfaces. Strongly adsorbed bitumen
Adhesion
255
FIGURE 28.5 Adsorption of bitumen molecules tothe aggregate surface (Ensley, 1975).
fractions have been identified at the bitumenaggregate interface, forming a band of the orderof 180Å thick. Ketones, dicarboxylic anhydrides,carboxylic acids, sulphoxides and nitrogen-bearing components have been found in this layer(Ensley, 1975). The strongly adsorbed com-ponents have been found to have sites capable ofhydrogen bonding to the aggregate, though in thepresence of water the available bonds prefer themore active water. A migration of some bitumencomponents to the interface is inferred and there-fore a dependence on binder composition, mixingtemperature and viscosity. Figure 28.5 illustratesthe process, with molecules of bitumen at thesurface aligned in the direction of polarity of thesubstrate (aggregate), usually a negative surface.The zone of orientation of bitumen moleculesextends for a thickness of several thousand mol-ecules.
28.4.3 Mechanisms for loss of adhesion
Breakdown of the bond between bitumen andaggregate, known as stripping, may occur for a
number of reasons. However, the principal agen-cies are the action of traffic, weathering andmoisture, and these often act in combination.
Weathering may affect the aggregate itself butis a strong influence on the bitumen causingageing through exposure to ultra-violet light. Theconsequent loss of ductility of the bitumen willrender it more vulnerable to brittle fracture andthe coating of aggregate particles may be broken.
The effect of moisture is more significant sinceit causes loss of adhesion in a number of ways(Hargreaves, 1987). A number of mechanismshave been postulated for adhesion loss and mostinvolve the action of water. These are describedbelow and each may occur depending on the cir-cumstances.
Displacement
This occurs when the bitumen retracts from itsinitial equilibrium position as a result of contactwith moisture. Figure 28.6 illustrates the processin terms of an aggregate particle embedded in abituminous film (Majidzadeh and Brovold, 1968).Point A represents the equilibrium contact posi-tion when the system is dry. The presence ofmoisture will cause the equilibrium point to shiftto B leaving the aggregate particle effectively dis-placed to the surface of the bitumen. The posi-tions of points A and B will depend on the typeand viscosity of bitumen.
Durability of bituminous mixtures
256
FIGURE 28.6 Retraction of the binder–water inter-face over the aggregate surface in the presence of water(Majidzadeh and Brovold, 1968).
FIGURE 28.7 Thinning of the bitumen film on anaggregate with rough surface texture (b, c). Smoothaggregates (a) retain an unstressed and even film.
Detachment
This occurs when the bitumen and aggregate areseparated by a thin film of water or dust thoughno obvious break in the bitumen film may beapparent. Although the bituminous film coats theaggregate particle, no adhesive bond exists andthe bitumen can be clearly peeled from thesurface.
Film rupture
This occurs when the bitumen fully coats theaggregate but where the bitumen film thins,usually at the sharp edges of the aggregate parti-cle (Figure 28.7).
Blistering and pitting
If the temperature of bitumen at the surface of aroad rises, its viscosity reduces. This reduced vis-cosity allows the bitumen to creep up the surfaceof any water droplets which fall on the surface,and may eventually form a blister (Figure 28.8).With further heating, the blister can expand andcause the bitumen film to rupture, leaving a pit.
Spontaneous emulsification
Water and bitumen have the capacity to form anemulsion with water at the continuous phase. The
emulsion formed has the same negative charge asthe aggregate surface and is thus repelled. Theformation of the emulsion depends on the type ofbitumen, and is assisted by the presence of finelydivided particulate material such as clay mater-ials, and the action of traffic.
Hydraulic scouring
This is due principally to the action of vehicletyres on a wet road surface. Water can bepressed into small cavities in the bitumen film infront of the tyre and, on passing, the action ofthe tyre sucks up this water. Thus a compres-sion–tension cycle is invoked which can causedisbonding.
Pore pressure
This mechanism is most important in open orpoorly compacted mixes. Water can becometrapped in these mixes as densification takes placeby trafficking. Subsequent trafficking acts on thistrapped water and high pore pressures can result.This generates channels at the interface between
bitumen and aggregate (Figure 28.9) and eventu-ally leads to disbonding.
Chemical disbonding
Diffusion of moisture through the bituminouscoatings can lead to the layers of water buildingup in the interfacial region.
28.5 ReferencesCabrera, J.G. and Hassan, T.Q.M. (1994) Quality
control during construction of bituminous mixturesusing a simple air permeability test. In: Performanceand Durability of Bituminous Materials, E & FNSpon, London.
Ensley, E.K. (1975) Multilayer absorption with molecu-lar orientation of asphalt on mineral aggregate andother substrates. Journal of Applied Chemistry andBiotechnology 25, 671–82.
Griffin, R.L., Miles, T.K. and Penther, C.J. (1955) Micro-film durability test for asphalt. Proceedings of theAssociation of Asphalt Paving Technologists 24, 31.
Griffin, R.L., Simpson, W.C. and Miles, T.K. (1959)Influence of composition of paving asphalt on vis-cosity, viscosity–temperature susceptibility, anddurability. Journal of Chemical and EngineeringData 4, 89–93.
References
257
FIGURE 28.8 Formation of blisters and pits in abituminous coating (Thelan, 1983).
FIGURE 28.9 Pore pressure disbonding mechanism(McGennis et al., 1984).
Hargreaves, A. (1987) An investigation of the prema-ture failure of bituminous macadam wearing courseswith particular reference to the effects of moistureand aggregate quality. PhD thesis, Hatfield Polytech-nic.
Khalid, H. (1990) Permeability: implications in asphaltmix design. Highways Asphalt 90 Supplement,Highways, Faversham House Group.
Majidzadeh, K. and Brovold, F.N. (1968) State of theArt: Effect of Water on Bitumen–Aggregate Mix-tures, Highway Research Board, Publication 1456,Special Report 98, 77.
McGennis, R.B., Kennedy, T.W. and Machemehl, R.B.(1984) Stripping and Moisture Damage in AsphaltMixtures, Research Report 253.1, Project39–79–253, Centre for Transportation Research,University of Texas, USA.
Robinson, D. (1991) An Investigation into WearingCourse Failure on the A414 Breakespeare Way,Hemel Hempstead. Unpublished project report, Hat-field Polytechnic.
Thelan, E. (1958) Surface energy and adhesion proper-ties in asphalt–aggregate systems. Highway ResearchBoard Bulletin 192, 63–74.
Durability of bituminous mixtures
258
Chapter twenty-nine
Practice andprocessing ofbituminousmaterials
29.1 Bituminous mixtures29.2 Recipe and designed mixes29.3 Methods of production29.4 References
29.1 Bituminous mixturesThere are a very large number of bituminousmixtures which vary according to density,bitumen content, bitumen grade, and aggregatesize and grading. However, most can be classifiedinto two groups, namely asphalts and macadams.
29.1.1 Asphalts
Asphalts are dense materials and are character-ized by their high bitumen content and highfiller/fines content. They derive their strength andstiffness from a dense, stiff mortar of bitumen,filler and fines. The coarse aggregate content isrelatively low so that the overall particle size dis-tribution is gap-graded. Figure IV.1 (a) of theIntroduction illustrates these features and it canbe seen that the material transmits load throughthe mortar continuum. This mortar, being rich inbitumen, is expensive and the coarse aggregateserves to increase the volume of the mortar with arelatively cheap material, thereby reducing the
overall cost. The binder used for asphalt will nor-mally be 50 or 70 pen and may either be penetra-tion grade bitumen or a blend of bitumen withpitch or lake asphalt. Because of the hardness ofthe binder and the higher filler/fines content, theworkability is low, and high temperatures have tobe used to mix, lay and compact the material,which adds to the cost. Freshly laid wearingcourse asphalt (known as hot rolled asphalt) pre-sents a smooth surface with coarse aggregate par-ticles submerged with the mortar. In order toprovide a skid-resistant surface, coated chippingsare rolled into the surface. This also adds to thecost. Asphalts have a very low permeability, andare capable of transmitting high stresses whilstproviding some ductility. They are therefore verydurable, and normally used where traffic loadsare high or durability is important.
There are three groups of asphalt mixtures (BS594: 1992): Group 1 mixtures are for roadbases,base courses and regulating courses; Group 2 mix-tures are designed wearing course mixtures; Group3 mixtures are recipe wearing course mixtures.Wearing course mixtures may be either type Fincorporating fine sand, or type C incorporatingcrushed rock or slag fines which are more coarselygraded. The coarse aggregate is either 14mm or20mm nominal size and varies in content from
259
30–55 per cent. Higher proportions are associatedwith larger size. In the designed wearing coursemixtures, the binder content is determined from adesign process which will be described in thefollowing section. Table 29.1 shows the specifica-tion for Group 1 mixtures as an example. It can beseen that each mixture is designated according tothe coarse aggregate content and its nominal size.Thus a 50/20 mixture has 50 per cent coarseaggregate with a nominal size of 20mm.
29.1.2 Macadams
Macadams range from dense mixtures to open-textured mixtures. They are characterized by arelatively low binder content and a continuouslygraded aggregate.
Macadams rely on the packing and interlock ofthe aggregate particles for their strength and stiff-ness. The binder coats the aggregate, and acts asa lubricant when hot and an adhesive and water-proofer when cold. The grade of binder used issofter than for asphalts being 100 or 200 pen.Figure IV.1 (b) of the Introduction illustratesthese features and it can be seen that the materialtransmits load through the aggregate structure.Because of their lower binder content, macadamsare cheaper than asphalts. In general macadamshave a higher void content than asphalts so theyare more permeable and less durable. They lackductility and are generally used on less heavilytrafficked roads.
Macadams are classified according to thenominal size of the aggregate, the grading of the
Practice and processing of bituminous materials
260
TABLE 29.1 Composition of roadbase, basecourse and regulating course hot rolled asphalt mixtures (BS 594:1992) Reproduced with the permission of the British Standards Institution under licence number 2001/SK0230.
Column number 1 2 3
Designation* 50/14† 50/20† 60/20
Nominal thickness of layer (mm) 35 to 65 45 to 80 45 to 80
Percentage by mass of total aggregrate passing BS test sieve50 mm – – –37.5 mm – – –28 mm – 100 10020 mm 100 90 to 100 90 to 10014 mm 90 to 100 65 to 100 30 to 6510 mm 65 to 100 35 to 100 –6.3 mm – – –2.63 mm 35 to 55 35 to 55 30 to 44600 µm 15 to 55 15 to 55 10 to 44212 µm 5 to 30 5 to 30 3 to 2575 µm 2 to 9 2 to 9 2 to 8
Binder content, % by mass of total mixture for:Crushed rock or steel slag 6.5 6.5 5.7Gravel 6.3 6.3 5.5Blast furnace slag: bulk density (kg/m3)
1440 6.6 6.6 5.71360 6.7 6.7 5.91280 6.8 6.8 6.01200 6.9 6.9 6.11120 7.1 7.1 6.3
* The mixture designation numbers (e.g. 50/14 in column 1) refer to the nominal coarse aggregate content of themixture/nominal size of the aggregate in the mixture respectively.† Suitable for regulating course.
aggregate and the intended use of the material(BS 4987: 1993). The main mixtures availableare:
1. roadbase – 40mm dense roadbase2. basecourse – 40 and 20mm open-
textured basecourse– 40mm single course– 40, 28 and 20mm dense
basecourse3. wearing course – 14 and 10mm open-
textured wearing course– 14 and 10mm dense
wearing course– 10mm medium-texture
wearing course– 6mm fine-textured wearing
course.
(The last two mixes are referred to as coarse coldasphalt and fine cold asphalt respectively eventhough they are macadams.)
For the open-textured mixes, the fines contentis low (less than 15 per cent), whereas for themedium- and fine-textured materials, the finescontent is higher (40–60 per cent and more than75 per cent respectively). The dense materialshave a closely specified aggregate grading and alow voids content (about 4 per cent). Theirdensity and consequent strength are more in linewith some types of asphalt.
Although asphalts and macadams have distinc-tive features, they in fact represent opposite endsof a spectrum of materials. This range covers avariation in the proportions of the constituentsand the void content, and is illustrated in Figure29.1. It can be seen that in the middle of therange the roadbase asphalts and dense macadamsare not so different from one another.
29.1.3 Other mixtures
There are other bituminous mixtures which donot fit the general description of asphalts ormacadams described in the previous sections.Two that are beginning to be used widely areporous asphalt and stone mastic asphalt.
Bituminous mixtures
261
FIGURE 29.1 Constituent proportions of variousbituminous materials.
Porous asphalt
In spite of its name, porous asphalt is essentially agraded aggregate bound with bitumen, and isincluded in the British Standard for macadammixtures (BS 4987: 1993). It is designed toprovide a large volume (at least 20 per cent) ofinterconnected air voids so that water can drainthrough the material and run off within the thick-ness of the layer. This requires the underlyingbasecourse to be impermeable. The aggregategrading consists predominantly of coarse aggreg-ate: about 75 per cent passing 2.36mm. The fineaggregate fractions are added to enhance thecohesion and stiffness of the mixture but in suffi-ciently small quantity so as not to interfere withthe interlock of the coarse particles, and to leaveenough voids to provide a pervious structure.Because of its porous nature, the material isvulnerable to ageing through oxidation of thebitumen. To counteract this, the bitumen contentmust be sufficient to provide a thick coating onthe coarse aggregate particles. Typically 5 per
cent of 100 pen bitumen is used. The advantagesof porous asphalt are that it minimizes spray inwet weather, reduces surface noise, improvesskidding resistance, and offers lower rolling resis-tance than dense mixtures. However, it is lessstrong than denser mixtures and UK pavementdesign practice assumes that a 60mm layer ofporous asphalt is structurally equivalent to20mm of hot rolled asphalt wearing course.
Stone mastic asphalt
Stone mastic asphalt (SMA) has a coarse aggreg-ate skeleton but the voids are filled with a mortarof fines, filler and bitumen. It thus resembles hotrolled asphalt, particularly the high stone contentmixtures, but it may best be considered as havinga coarse aggregate structure similar to porousasphalt but with the voids filled. SMA differsfrom hot rolled asphalt in that the quantity ofmortar is just sufficient to fill the voids in thecoarse aggregate structure. It therefore provideshigh stiffness due to the interlock of the coarseaggregate particles, and good durability becauseof a low void content. Since SMAs do not have amortar continuum, they are less well able to with-stand tensile strain than hot rolled asphalt. Inorder to improve their resistance to tensile strain,it is usual to include cellulose fibres in themixture.
29.2 Recipe and designed mixesThe majority of bituminous mixtures are recipemixes. In other words, the mixtures are puttogether according to prescribed proportions laiddown in the appropriate British Standard. Thesemix proportions have been derived throughexperience in use and, provided the separateingredients meet their specifications, the mix willprovide the required performance in most situ-ations. This approach is consistent with theempirical method for the structural design ofroads which has predominated until relativelyrecently. Thus an empirical chart to determine thethickness of roadbase was used which dependedon the roadbase material having the same mix
proportions as the materials in the roads fromwhich the design chart was originally established.
It is important to note that tests for complianceof recipe mixes can only be compositional testsand tests of the nature of individual ingredients.Typically, these materials are broken down todetermine binder content and the aggregategrading. It is axiomatic that there are no tests ofthe properties of the material produced or of itsperformance.
Recipe mixes provide a satisfactory perform-ance in many cases and there is some advantagein the simplified approach which recipe mixesoffer. However, there are limitations to the use ofrecipe mixes which match the limitations ofempirical structural design of roads. These are asfollows:
1. Non-specified materials cannot be used. Forexample, a locally available sand may notmeet the grading requirements of the specifi-cation but may produce a satisfactory mix.Recipe mixes preclude any assessment of theproperties of a mix containing that sand.
2. Modified binders cannot be used. The use ofchemical additives in binders can give usefulenhancements to their properties. The absenceof any end-test of recipe mixes prevents theevaluation of these changes.
3. No procedure is available to assess causes offailure.
4. No procedure is available to optimize the mixproportions. This is particularly important asfar as the bitumen is concerned because this isthe most expensive ingredient and has astrong bearing on the performance of mixes,especially those which are more dense.
These drawbacks have led to the development ofa procedure for the design of bituminous mixes,which has occurred in parallel with the develop-ment of analytical procedures for the structuraldesign of roads. An analytical approach to roaddesign enables the determination of the thicknessof the road structure through an analysis of itsbehaviour under the applied load. This clearlyrequires a knowledge of certain properties of thematerials and it follows that materials will have
Practice and processing of bituminous materials
262
to be produced with particular characteristics.However, the design system, with material pro-perties feeding into the analysis, has not yet beenfully developed and what are now called designedmixes are, in fact, only partially designed.
The procedure for mix design (BS 598: Part107: 1990) is based upon the Marshall test whichwas originally developed in the USA for designingmixes for use on airfield runways. The objectiveof the procedure is to determine an optimumbinder content from a consideration of:
1. mix strength (stability);2. mix density;3. mix deformability (flow).
Test samples of binder/aggregate mixtures areprepared using the materials to be used in thefield. The aggregate grading is kept constant andsamples with a range of binder contents are pro-duced. The samples are prepared and compactedin a standard way into moulds 101.6mm in dia-meter and 70mm high. The state of compactionachieved is determined by measuring the bulkdensity and calculating the compacted aggregatedensity. At low binder contents the mix will lackworkability and the densities will be correspond-ingly low. At high binder contents, aggregate willeffectively be displaced by bitumen, and again thedensities will be low. Each of these measures ofdensity will thus produce an optimum bindercontent as shown in Figures 29.2 (a) and (b).
To test the strength and resistance to deforma-tion of the material, the specimens are heated to60°C and subjected to a compression test usingspecial curved jaws which match the curved sidesof the specimens as shown in Figure 29.3. Thusthe load is applied radially. The jaws of themachine are driven together at a constant rate of50mm per minute until the maximum load isobtained which is termed the ‘stability’. Thedeformation of the sample at this maximum loadis also recorded and termed the ‘flow’. Typicalplots of stability and flow against binder contentare shown in Figures 29.2 (c) and (d). Thestability plot gives a third optimum bindercontent, and the design binder content is obtainedfrom the average of this and the optima from the
Recipe and designed mixes
263
FIGURE 29.2 Analysis of mix design data from theMarshall test.
FIGURE 29.3 Testing arrangement for a Marshallasphalt design.
two density plots. The flow at this design bindercontent can then be read off. Minimum values ofstability and flow are specified according to theamount of traffic which the road will carry.
In evaluating mixes it is helpful to consider theMarshall quotient, Qm. This is derived from thestability and flow:
Qm � stability/flow (29.1)
Thus Qm bears some resemblance to a modulus(ratio of stress to strain) and may be taken as ameasure of mix stiffness. It has been found tocorrelate well with wheel-tracking tests to assessdeformation resistance. Since stability is used todetermine the optimum binder content, the designprocedure may be regarded as determining stiff-ness at optimum binder content, and the largerthe value of Qm, the more resistant the mix willbe to deformation.
More recently new approaches to bituminousmix design have been proposed. A completedesign process has been suggested (Cabrera,1994) which is applicable to all hot bituminousmixtures, i.e. both hot rolled asphalts andmacadams. The procedure, known as the LeedsDesign Method, aims to design mixtures to meetboth structural requirements and in-serviceperformance requirements. It involves measure-ment of a range of parameters: workability, airpermeability, Marshall stability and flow, staticstiffness and vapour diffusion.
Another development is the use of a repeatedload compression test for assessing resistance todeformation as part of a design process (Gibband Brown, 1994). The test employs the Notting-ham Asphalt Tester which enables a specimen tobe subjected to repeated load in a controlledenvironment. In order to represent vehicleloading effects, stress pulses of 100kPa areapplied for a duration of one second with aninterval of one second between pulses. Resultscorrelate well with wheel tracking tests.
29.3 Methods of productionThe process of manufacture of bituminous mater-ials involves three stages. First, the aggregatesmust be proportioned to give the requiredgrading, second the aggregates must be dried andheated, and third, the correct amount of bindermust be added to the aggregate and mixed tothoroughly coat the aggregate particles andproduce a homogeneous material.
The most common type of plant in the UK isthe indirectly heated batch mixing plant. Aschematic diagram of this type of plant is shownin Figure 29.4. The aggregate is blended fromcold bins and passed through a rotary
Practice and processing of bituminous materials
264
Fourteen major parts1 Cold bins2 Cold feed gate3 Cold elevator4 Dryer/heater5 Dust collector6 Exhaust stack7 Hot elevator
8 Screening unit9 Hot bins
10 Weigh box11 Mixing bowl or pugmill12 Mineral filler storage13 Hot bitumen storage14 Bitumen weigh bucket
1
2
3 4 5
11 14
6 7
8
9
10
12 13
FIGURE 29.4 Schematic diagram of an indirectly heated batch mixing plant (The Shell Bitumen Handbook,1990).
drier/heater. Here the moisture is driven off andthe aggregate temperature raised to the prescribedmixing temperature for the type of material beingproduced. The aggregate is then transported byhot elevator to hot storage bins where it is sepa-rated into fractions of specified size. Aggregatesare released into the weigh box in the desiredproportions and then released into the mixer.Bitumen heated to the prescribed temperature isalso introduced to the mixer, the quantity beingdetermined using a weigh bucket or volumetri-cally using a flow meter. The mixing time variesup to 60 seconds but should be as short as pos-sible in order to limit oxidation of the binder.After mixing, the material is discharged directlyinto a wagon.
This type of plant is very versatile, beingcapable of producing both asphalt and macadammixes, and being able to easily adjust to a widerange of mix specifications.
A variation on this type of plant is to dry andheat the aggregate in batches before beingcharged into the mixer. This eliminates the needfor hot aggregate storage but the proportioningof the cold aggregate needs to be very carefullycontrolled.
An alternative type of mixer is the drum mixerwhich gives continuous production rather thanbatches. Here the cold aggregates are propor-tioned and conveyed directly into a drum mixer.The drum has two zones. The first zone is wheredrying and heating occur, and in the second zonethe bitumen is introduced and the mixing takesplace. The advantages of this type of plant arethat the amount of dust emission is reduced, theprocess is simpler and, above all, the rate of pro-duction can be very high – up to 500 tonnes perhour. This is advantageous where large quantitiesof the same type of material are required, but it isdifficult to change production to a different mix.
29.4 ReferencesCabrera, J.G. (1994) Hot bituminous mixtures – design
for performance. In Performance and Durability ofBituminous Materials, E & FN Spon, London.
Gibb, J.M. and Brown, S.F. (1994) A repeated loadcompression test for assessing the resistance of bitu-minous mixes to permanent deformation. In Perfo-rmance and Durability of Bituminous Materials, E & FN Spon, London.
Shell Bitumen UK (1990) The Shell BitumenHandbook.
References
265
266
Further reading
Hunter, R. (2000) (ed.) Asphalts in Road Construction,Thomas Telford, August.
This book gives an excellent coverage of asphaltsincluding recent developments in asphalt technologysuch as stone mastic asphalt, thin surfacings, andhigh modulus bases. As well as covering the mater-ials themselves, it deals with the design and mainte-nance of pavements, laying and compaction, surfacedressing and other surface treatments, and failure ofsurfacings. The book is an updated version of Bitu-minous Materials in Road Construction. This too isworth reading since it deals with all types of bitumi-nous material.
O’Flaherty, C.A. (2000) Highways: The Location,Design, Construction and Maintenance of RoadPavements, Butterworth-Heinemann, June.
A comprehensive textbook on all aspects of roadengineering from the planning stages through to thedesign, construction and maintenance of road pave-ments.
Whiteoak, D. (1990) The Shell Bitumen Handbook,Shell.
Essential reading if you want to get to grips withthe fundamentals of bitumen. It deals with the struc-ture and constitution of bitumen and goes on to linkthis to the engineering properties. The book alsocovers the design and testing of bituminous mixes aswell as pavement design.
StandardsBritish Standard 434: Part 1: 1984: Bitumen road
emulsions (anionic and cationic). Part 1. Specifica-tion for bitumen road emulsions.
British Standard 594: 1992: Hot rolled asphalt forroads and other paved areas. Part 1: Specificationfor constituent materials and asphalt mixtures.
British Standard 598: Part 105: 1990: Methods of testfor the determination of texture depth.
British Standard 598: Part 107: 1990: Method of testfor the determination of the composition of designwearing course asphalt.
British Standard 812: Part 110: 1990: Methods fordetermination of aggregate crushing value.
British Standard 812: Part 112: 1990: Methods fordetermination of aggregate impact value.
British Standard 812: Part 114: 1989: Method fordetermination of polished stone value.
British Standard 2000: Part 27: 1993: Distillation ofcutback asphaltic (bituminous) products.
British Standard 2000: Part 45: 1993: Loss on heatingof bitumen and flux oil.
British Standard 2000: Part 47: 1983: Solubility ofbituminous binders.
British Standard 2000: Part 49: 1983: Penetration ofbituminous materials.
British Standard 2000: Part 58: 1983: Softening pointof bitumen (ring and ball).
British Standard 2000: Part 72: 1993: Viscosity ofcutback bitumen and road oil.
British Standard 2000: Part 357: 1983: Permittivity ofbitumen.
British Standard 3690: Part 1: 1989: Bitumens forbuilding and civil engineering. Part 1: Specificationfor bitumens for road purposes.
British Standard 4987: 1993: Coated macadam forroads and other paved areas. Part 1: Specificationfor constituent materials and for mixtures.
Department of Transport (1987) Departmental Stan-dard HD 14/87.
Part Five
Brickwork andBlockwork
R.C. de Vekey
Introduction
This introduction outlines the history of masonryand explains the terminology used. The sub-sequent chapters cover: materials, compositionand manufacturing processes for the componentsof masonry; structural forms, architecture anddetailing; structural behaviour and response toactions such as wind and movement; and durabil-ity and other important properties in relation toheat, noise, fire and rain, and weather.
The title ‘Brickwork and blockwork’ has beenchosen to emphasize that these materials are themost common of this type used in construction atthe present time. The term ‘masonry’
• has recently been widened from its traditionalmeaning of structures built of natural stone toencompass all structures produced by stack-ing, piling or bonding together discretechunks of rock, fired clay, concrete, etc., toform the whole
• ‘masonry’, in this wider sense, is what thesechapters are about.
Second to wood, masonry is probably the oldestbuilding material used by man, and it certainlydates from the ancient civilizations of the MiddleEast and was used widely by the Greeks andRomans. Early cultures used mud building bricksand very little of their work has survived, butstone structures such as the Egyptian pyramids,Greek temples and many structures made fromfired clay bricks have survived for thousands ofyears. The Romans used both fired clay bricksand hydraulic (lime/pozzolana) mortar andspread this technology over most of Europe.
The basic principle of masonry is of buildingstable bonded (interlocked) stacks of handleablepieces. The pieces are usually chosen or manufac-
tured to be of a size and weight that one personcan place by hand but, where additional power isavailable, larger pieces may be used which givepotentially more stable and durable structures.
This greater stability and durability is con-ferred by the larger weight and inertia whichincreases the energy required to remove one pieceand makes it more resistant to natural forces suchas winds and water as well as human agency.There are four main techniques for achievingstable masonry:
1. Irregularly shaped and sized but generallylaminar pieces are chosen and placed by handin an interlocking mass (e.g. dry stone walls).
2. Medium to large blocks are made very pre-cisely to a few sizes and assembled to a basicgrid pattern either without mortar or withvery thin joints (e.g. ashlar).
3. Small to medium units are made to normalprecision in a few sizes and assembled to abasic grid pattern and the inaccuracies aretaken up by use of a packing material such asmortar (e.g. normal brickwork).
4. Irregularly shaped and sized pieces are bothpacked apart and bonded together withadherent mortar (e.g. random rubble walls).
Only type (4) structures depend largely on themortar for their stability; all the other types relylargely on the mechanical interlocking of thepieces. Figure V.1 shows typical examples.
These descriptions are given to emphasize thatmost traditional masonry owes much of itsstrength and stability to interlocking action,weight and inertia while the mortar, whenpresent, is not acting as a glue but as somethingto fill in the gaps resulting from the imperfect
269
fitting together of the pieces. Most contemporarymasonry is type (3) and modern mortars do havean adhesive role but much of the strength stillderives from the mass and interlocking of shapesand it is important to remember this in design.
It is also important to remember that, althoughthe wall is the most useful and effective masonrystructure, many other structural forms are usedsuch as columns, piers, arches, tunnels, floors androads. Normal plain masonry must be designedsuch that the predominant forces put it intocompression since it cannot be relied on to resisttensile forces. If, however, tension structures suchas cantilevers, earth-retaining walls and beams
are required, masonry may be reinforced or post-tensioned in the same way as concrete.
Terminology
Components
Units – pieces of stone, brick, concrete or calciumsilicate which may be assembled to makemasonry. Usually, but not invariably, they are inthe form of rectangular parallelepipeds.Mortar – a material that is plastic (flows) whenfresh but sets hard over a period of hours to days.Its purpose is to fill the gaps caused by variations
Introduction
270
(a)
(c)
(b)
(d)
FIGURE V.1 The main types of masonry: (a) dry stone walls, (b) ashlar, (c) jointed brick and blockwork and(d) rubble masonry.
in the size and shape of units such that themasonry is stable and resists the flow of air andwater. Mortar is compounded from a binder (e.g.cement) and a filler/aggregate (usually sand). Thebinder is a finely ground material which, whenmixed with water, reacts chemically and then setshard and binds aggregates into solid masses.
Size and dimensions (Figure V.2)
Work size – the size of a masonry unit specifiedfor its manufacture, to which its actual sizeshould conform within specified permissible devi-ations. As a rough guide for the following sec-tions, bricks are considered to be units with facedimensions of up to 337.5mm long by 112.5mmhigh and with a maximum width of 225mm,while blocks are larger units up to face dimen-sions of 1500mm by 500mm. UK units aresmaller than these limits. A standard UK brick is215mm long by 65mm high and with a depth of102.5mm. There is no standard block size but thecommonest size is 440mm long by 215mm high
and with a width of 100mm, with a limiting sizeof 650mm for any dimension for standard con-crete units.Co-ordinating size – the size of a co-ordinatingspace allocated to a masonry unit, includingallowances for joints and tolerances. Thiscoordination grid into which they fit is generally10mm larger for each dimension.
Units
Units may be produced in a number of forms,illustrated in Figures 30.6 and 31.7.
• Solid – having no designed voids (holes,depressions or perforations) other than thoseinherent in the material.
• Frogged – with a depression (or frog) in oneor both of the bed faces where the totalvolume of the frog(s) does not exceed 20 percent of the gross volume.
• Cellular – having one or more deep holes ordepressions in one bed face with an aggregatevolume exceeding 20 per cent of the gross
Terminology
271
FIGURE V.2 Masonry dimensions.
volume and which do not penetrate throughto the other side.
• Perforated – having one or more holes passingfrom one face to the opposite face. UK prod-ucts have vertical perforations, i.e. the holespass from one bed face to the other and willbe vertical when laid normally in a stretcherbonded wall. Horizontally perforated clayblocks are, however, not uncommon incontinental Europe.
• Hollow – having one or more formed holes orcavities which pass through the units.
• Specials (special shapes) – a range of standard
bricks for curves, non-right-angled corners,plinths, cappings, etc., are available. They aredescribed in BS 4729: 1990.
General
Fair-faced – masonry, within the variability ofindividual units, precisely flat on the visible face.This is normally only possible on one side of solidwalls.
Other relevant definitions are contained in BS6100: Part 3: 1984.
Introduction
272
Chapter thirty
Materials andcomponents forbrickwork andblockwork
30.1 Materials used for manufacture of units and mortars30.2 Other constituents and additives30.3 Mortar30.4 Fired clay bricks and blocks30.5 Calcium silicate units30.6 Concrete units30.7 Natural stone30.8 Ancillary devices – ties and other fixings/connectors30.9 References
30.1 Materials used for manufactureof units and mortars
30.1.1 Sands and fillers
Sand
Sand is used widely as a constituent of masonryin mortar, in concrete units and sandlime units, ingrouts and renders. It is a mixture of rock parti-cles of different sizes from about 10mm diameterdown to 75�m diameter.
Sand is usually extracted from recent naturallyoccurring alluvial deposits such as river beds andsea beaches, or from older deposits from alluvialor glacial action. In some areas it may be derivedfrom dunes or by crushing quarried rocks. Thechemical and geological composition will reflectthe area from which it is derived. The commonest
sands are those based on silica (SiO2), partlybecause of its wide distribution in rocks such assandstones and the flint in limestones, and partlybecause silica is hard and chemically resistant.Other likely constituents are clay, derived fromthe decomposition of feldspars, calcium carbon-ate (CaCO3), in the form of chalk or limestonefrom shells in some marine sands, and micas insands from weathered granites. Crushed rockssuch as crushed basalts and granites will reflecttheir origins.
Sands should be mostly free of particles of clay(with a size of between 75–30 µm) which causeunsatisfactory shrinkage characteristics andchemical interactions with binders. Most of theconstituents of sand are relatively chemically inertto environmental agents but chalk or limestoneparticles will be dissolved slowly by mild acidsand clays may react in time with acids, or alkalis.Most sand constituents are also fairly hard andare resistant, in themselves, to mechanical abra-sion and erosion by dust in wind and water.
Mortar and rendering sands
Mortar sand must contain no particles with adiameter greater than about half the thinnestjoint thickness, i.e. around 5mm. It should also
273
have a good range of particle sizes from thelargest to the smallest (a good grading) since thisleads to good packing of the particles to give adense, strong mass resistant to erosion, perme-ation and chemical attack. Many naturally occur-ring alluvial deposits fall naturally into therequired grading and may be used as dug or justwith a few coarser particles screened off. Theseare usually termed pit sands. Sands that areoutside the normal range must be sieved toremove coarse fractions and washed to removeexcess clay particles.
The shape of the particles is also important formortar sands. Very flaky materials such as slatesand micas are not very suitable as it is difficult tomake them workable. Very absorbent materialsare also unsatisfactory for dry-mixed mortarssince they cause rapid falls in workability byabsorbing the mixing water. They may be suit-able for mixes based on wet premixed lime-sand‘coarse stuff’. Sand may be sieved into fractionsand regraded but this is rarely done for a mortarsand. Figure 30.1 shows the grading curves forthe sands allowed under BS 1200: 1976.1 Thisgives two allowed grades, S for structural use and
Materials and components for brickwork and blockwork
274
FIGURE 30.1 Grading curves for S and G sands.
1 A list of all standards referred to in the text is included in‘Further reading’ on page 329.
G, with slightly wider limits, for general pur-poses. Rendering mixes require sands withbroadly similar characteristics to mortars but agood grading is even more important to avoidshrinkage, cracking and spalling and to give goodbond to the substrate.
Concreting sands
These have been discussed in Chapter 16.
Ground sand
Finely ground silica sand is used particularly inthe manufacture of autoclaved aerated cement(AAC) materials.
Pulverized fuel ash (pfa)
Pfa is the main by-product of modern coal-firedelectricity generating stations which burn finelyground coal. Typical composition ranges aregiven in Table 30.1 and fuller information isavailable in Central Electricity Generating Board(1972). It has also been discussed in Chapter 15.
Chalk (calcium carbonate, CaCO3)
In a finely ground state, chalk is used as a fillerand plasticity aid in masonry cement and somegrouts.
30.1.2 Clays
Clay is a very widely distributed material which isproduced by weathering and decomposition ofacid alumino-silicate rocks such as the feldspars,granites and gneisses. Typical broad types are thekaolin group of which kaolinite has a composi-tion Al2O3.2SiO2.2H2O, the montmorillonitegroup of which montmorillonite itself has thecomposition Al2O3.4SiO2.nH2O, and the claymicas which typically have a compositionK2O.MgO.4Al2O3.7SiO2.2H2O.
They will frequently contain iron, which cansubstitute for the aluminium, and other transitionmetals. The clays used for clay brick manufactureare normally only partly actual clay minerals,which impart the plasticity when wetted, thebalance being made up of other minerals. Brickearths, shales, marls, etc., mostly contain finely
divided silica, lime and other materials associatedwith the particular deposit, e.g. carbon in coalmeasure shales. Most brick clays contain ironcompounds which give the red, yellow and bluecolours to fired bricks. Table 30.2 gives the com-positions of some typical clays in terms of theircontent of oxides and organic matter (coal, oil,etc.).
The properties of clays are a result of theirlayer structure which comprises SiO4 tetrahedrabonded via oxygen to aluminium atoms whichare also bonded to hydroxyl groups to balancethe charge. The layers form loosely bound flatsheet-like structures which are easily parted andcan adsorb and bond lightly to varying amountsof water between the sheets. As more water isabsorbed the clay swells and the intersheet bondbecomes weaker, i.e. the clay becomes moreplastic and allows various shaping techniques tobe used.
30.1.3 Aggregates
Natural aggregates, sintered pfa nodules, expandedclay and foamed slag have been described in
Materials used for manufacture of units and mortars
275
TABLE 30.1 Typical chemical composition ranges of pulverized fuel ash (weight %)
Range SiO2 Al2O3 Fe3O4 CaO MgO K2O Na2O TiO2 SO3 Cl
Minimum 45 24 7 1.1 1.5 2.8 0.9 0.8 0.3 0.05Maximum 51 32 11 5.4 4.4 4.5 1.7 1.1 1.3 0.15
TABLE 30.2 Chemical compositions of some representative clays (weight % of oxides) (X � alkali metal ion)
Broad type SiO2 Al2O3 Fe3O4 CaO MgO X2O CO2 H2O Organic matter
London brick 49.5 34.3 7.7 1.4 5.2 _ _ _ 1.9Blue clay 46.5 38 1 1.2 _ _ _ _ 13.3Loam 66.2 27 1.3 0.5 _ _ _ _ 5Fletton clay 50 16 7 10 1 3 _ 6 6Marl 33 10 3 26 3.5 _ 20.5 4 _Burnham clay 42.9 20.9 5 10.8 0.1 0.3 8.1 6.9 5Red brick clay 49 24 8 7 _ 1 11 _ _Gault clay 44 15 6 17 _ _ _ 18 _Washed china clay 46 40 _ 1 _ _ _ 13 _Stourbridge fireclay 65 22 2 1 _ _ _ 10 _
Chapter 16. Other aggregates used particularly forunit manufacture are the following:
1. Furnace clinker, a partially fused ash from thebottom of solid-fuelled industrial furnaces.
2. Furnace bottom ash. Most large, modern fur-naces, especially those used to raise steam inpower stations, burn finely ground coal dustas a dust/air mixture. A proportion of the ashsinters together in the gas stream, then falls tothe base of the furnace as particles too largeto stay suspended in the gas stream. Thisclinker-like material is termed ‘furnacebottom ash’.
3. Perlite. Volcanic ash is deposited as a fineglassy dust and can be converted to a light-weight aggregate by hot sintering.
4. Pumice, a light foamed rock formed when vol-canic lava cools. It is normally imported fromvolcanic regions such as Italy.
30.1.4 Binders
The binder is the component which bindstogether mixtures of sands, aggregates, fillers,plasticizers, pigments, etc., used to make mortars,concrete units, sandlime units and grouts. Allthose used widely are based on: hydrauliccements which react chemically with water atnormal factory/site temperatures; lime–silica mix-tures which react only in the presence of high-pressure steam; or lime which sets slowly in airby carbonation. Because they must be finelydivided to be able to penetrate the spaces betweensand grains and to react in some way to give thechange from a formable plastic material to a hardadhesive, they must inherently be more chemi-cally reactive than the other components. Theirchemical reactivity is their weakness in that theyoften react with chemicals in the environmentwith resultant deterioration.
Portland cement
Currently, the most popular binder for generalpurposes is Portland cement, often still known asopc (ordinary Portland cement). Related products
are rapid hardening Portland cement and sulfateresisting Portland cement. The chemistry andmanufacturing methods of hydraulic cementshave been discussed in Chapter 13.
Masonry cement
A factory prepared mixture of opc with a fineinert filler/plasticizer (around 20 per cent) and anair-entraining agent to give additional plasticity issupplied as masonry cement. It is intended solelyfor mixing with sand and water to make beddingmortars. The fine powder is normally groundchalk but a wide range of inert and semi-inertmaterials can be substituted. In the future, Port-land cements blended with pulverized fuel ash(pfa) may well be used for mortars either asmasonry cements or as blended cements.
Lime and hydraulic lime
Lime (CaO) is widely used as an ingredient inmortars, plasters and masonry units. The pureoxide form, called quicklime, was used widely inthe past for mortars for stonework. It is preparedby heating pure limestone to a high temperatureand then ‘slaking’ with water to produce hydratedlime, Ca(OH)2. Since it does not have any settingaction in the short term, it may be kept wet fordays or weeks provided it is covered and pre-vented from drying out. The wet mix with sand istermed ‘coarse stuff’. Contemporary lime mortarmay be made from pre-hydrated lime but is other-wise similar. The initial setting action of thismortar depends only on dewatering by contactwith the units so it is not suitable for constructionof slender structures which require rapid develop-ment of flexural strength. Over periods of monthsor years the lime in this mortar carbonates andhardens to form calcium carbonate as in equation(30.1), but it is never as hard or durable as prop-erly specified hydraulic cement mortars.
Impure lime was widely used in the past and isgaining popularity for repairing historic buildingsto match the existing mortar. It is basically aquicklime – calcium oxide – produced by heatingimpure limestone to a high temperature. The
Materials and components for brickwork and blockwork
276
impurities, usually siliceous or clay, lead to theformation of a proportion of hydraulically activecompounds such as calcium silicates or alumi-nates. The mortar is made as normal by gauging(mixing in prescribed proportions) with sand andwater, but there will be some emission of heatwhile the lime is slaking. The classic systematiza-tion work was done by Vicat (1837) and laterCowper (1927) and more recent background isgiven by Ashurst (1983).
Ca(OH)2 �CO2 →CaCO3 �H2O (30.1)
Most of the hydraulic cements may be blendedwith pure hydrated lime in various proportions tomake hybrid binders which give mortars with alower strength and rigidity but still maintain theplasticity of the 1:3 binder:sand ratio. This leadsto mortars which are more tolerant of movementand more economical.
Sandlime
The binder used for sandlime bricks and auto-claved aerated concrete (AAC) blocks is lime(calcium hydroxide, Ca(OH)2) which reacts withsilica during autoclaving to produce calcium sili-cate hydrates. The reaction, in a simplified form,is:
Ca(OH)2 �SiO2 →CaSiO3 �H2O (30.2)
The lime is usually added directly as hydrated cal-cined limestone or may be derived in part fromopc which is also incorporated in small quantitiesto give a green strength to the unit.
30.2 Other constituents andadditives
30.2.1 Organic plasticizers
Many organic compounds improve plasticity, orworkability, of mortars, rendering mortars, infill-ing grouts and concrete used for manufacture ofunits. All the classic mortar plasticizers operateby causing air to be entrained as small bubbles.These bubbles fill the spaces between the sandgrains and induce plasticity. Typical materials are
based on Vinsol resin, a by-product of cellulosepulp manufacture, or other naturally available orsynthetic detergents. They are surfactants andalter surface tension and other properties. Superplasticizers, used only for concrete and groutmixes, plasticize by a different mechanism whichdoes not cause air entrainment. Admixtures arediscussed in more detail in Chapter 14.
30.2.2 Latex additives
A number of synthetic copolymer plastics may beproduced in the form of a ‘latex’, a finely divideddispersion of the plastic in water usually stabi-lized by a surfactant such as a synthetic detergent.Generally the solids content is around 50 per centof the dispersion. At a temperature known as thefilm-forming temperature, they dehydrate to forma continuous polymer solid. When combined withhydraulic cement mixes these materials have anumber of beneficial effects: they increase adhe-sion of mortar to all substrates; increase thetensile strength and durability; reduce the stiff-ness and the permeability. Because of these effectsthey are widely used in flooring screeds andrenders but are also used to formulate high-bondmortars and waterproof mortars. The better poly-mers are based on copolymerized mixtures ofbutadiene, styrene and acrylics. Polyvinylidenedichloride (PVDC) has also been marketed forthis application but it can lose chlorine which canattack buried metals. Polyvinyl acetate (PVA) isnot suitable as it is unstable in moist conditions.Polyvinyl propionate has been found to give lesssatisfactory flow properties than the acryliccopolymers. These materials should never be usedwith sands containing more than 2 per cent ofclay or silt particles. Dosage is usually in therange of 5–20 per cent of the cement weight.Table 30.3 gives some properties of commontypes from data in de Vekey (1975) and de Vekeyand Majumdar (1977).
30.2.3 Pigments
Through-coloured units and mortars of particularcolours may be manufactured either by selecting
Other constituents and additives
277
suitably coloured natural sands and binders or byadding pigments. Units may also be coloured byapplying surface layers but this is more commonfor fired clay than for concrete or calcium silicateunits. Pigments are in the form of inert colouredpowders of a similar fineness to the binder, sothey thus tend to dilute the mix and reducestrength. Most pigments should be limited to amaximum of 10 per cent by weight of the binderin mortars and carbon black to 3 per cent. Sometypical pigments, from information in ASTM taskgroup C09.03.08.05 (1980) are synthetic red ironoxide, Fe2O3; yellow iron oxide; black iron oxide,FeO (or Fe3O4) and brown iron oxide, Fe2O3,xH2O; natural brown iron oxide, Fe2O3, xH2O;chromium oxide green, Cr2O3; carbon black (con-crete grade); cobalt blue; ultramarine blue;copper phthalocyanine; and dalamar (hansa)yellow. Only pigments resistant to alkali attackand wettable under test mix conditions areincluded. All but the last two are not faded bylight.
30.2.4 Retarders
Retarders are used to delay the initial set ofhydraulic cement mortars. They are generally
polyhydroxycarbon compounds. Typical examplesare sugar, lignosulphonates and hydroxy-carboxylic acids. These have been more fully discussed in Chapter 14.
30.2.5 Accelerators
Accelerators have been marketed, usually basedon calcium chloride (CaCl2), which is used insmall amounts in concrete block manufacture.Alternatives such as calcium formate(Ca(CHO2)2) may be satisfactory. Acceleratorsare not effective when building with mortar infrosty weather and are no substitute for properprotection of the work.
Again, they have been discussed in Chapter 14.
30.3 MortarMortar has to cope with a wide range of, some-times conflicting, requirements. To obtainoptimum performance the composition must betailored to the application. The broad principlesare as follows:
1. Mortars with a high content of hydrauliccements are stronger, denser, more impervious
Materials and components for brickwork and blockwork
278
TABLE 30.3 Latex polymer additives
Chemical name: Reaction product with Properties of polymer/cement1 as proportion of neatPolymer-of or copolymer-of cement slurry cement paste
Elastic Modulus Flexural strengthAir2 Water2 Air2 Water2
Vinyl acetate Acetate ions 0.623 0.593 1.603 0.663
Vinyl proprionate None 0.623 0.74 1.243 1.223
Butadiene and styrene None 0.51 0.71 1.003 1.633
Vinylidene dichloride Chloride ions 0.664 0.844 1.524 1.364
Acrylic acid and styrene None 0.59 0.69 1.56 1.95Acrylic acid None 0.4 0.65 1.29 1.91Acrylic acid and methacrylic acid None 0.32 0.73 1.44 1.68
Notes:1. Properties are for an opc cement paste with a w/c of 0.3 and a polymer solids/cement ratio of 0.1 after storage for two years.2. Storage conditions are air at 65 per cent relative humidity or water at 20°C.3. These figures are the mean of two products.4. These figures are the mean of four products.
and more durable, bond better to units undernormal circumstances and harden rapidly atnormal temperatures. They also lead to a highdrying shrinkage and rigidity of the masonry.They are likely to cause shrinkage cracks ifused with shrinkable, low-strength units,particularly for long, lightly loaded walls suchas parapets and spandrels.
2. Mortars with decreased or no content ofhydraulic cements are weaker and moreductile and thus more tolerant of movement.They are matched better to low-strength unitsbut at the cost of a reduction in strength,durability. and bond. There is a correspond-ing reduction in shrinkage and hardening rate.
3. Mortars made with sharp, well-graded sandscan have very high compressive strength, lowpermeability and generally good bond butpoor workability, while fine loamy sands givehigh workability but generally with reducedcompressive strength and sometimes reducedbond.
4. Lime addition confers plasticity and, particu-larly for the wet stored mixes, water retentiv-ity – the ability of the mortar to retain itswater in contact with highly absorbent bricks– which facilitates the laying process andmakes sure that the cement can hydrate. Limemortars perform poorly if subjected to freez-ing while in the green (unhardened) state but,when hardened, are very durable. Lime iswhite and thus tends to lighten the colour ofthe mortar. In some circumstances it can beleached out and may cause staining.
5. Air entrainment improves the frost resistanceof green mortar and allows lower water/cement ratios to be used, but such plasticizedmixes may be less durable and water-retentivethan equivalent lime mixes. Air-entrainedmixes also need careful manufacture andcontrol of use since over-mixing gives veryhigh air contents, and retempering (addingmore water and reworking the mix) can leadto very poor performance due to the highporosity of the set dry mortar.
6. Pigment addition weakens mortar and thecontent should never exceed 10 per cent of the
Mortar
279
2 A list of BRE Digests on brickwork and blockwork isincluded in ‘Further reading’ on page 330.
weight of the cement in the mortar. Carbonblack is a special case and should be limited to3 per cent.
7. Polymer latex additives can markedly improvesome properties, such as bond, flexuralstrength and resistance to permeation bywater and air but, they are costly and shouldonly be used where there is a particularrequirement.
8. Retarders are widely used in the manufactureof ready-mixed mortars, delivered to site inthe same way as ready-mixed concrete. Theretarder is dosed as required by the supplier togive a ‘pot life’ of between 1 and 3 days. Careis needed in the use of retarded mortars, espe-cially in hot weather, because if they dry outtoo rapidly, the curing process never takesplace and the mortar never hardens.
9. Mortar has a relatively high thermal conduc-tivity and thus causes heat loss in walls oflow-density units. To achieve better thermalinsulation, insulating mortars are now avail-able which use a low-density replacement forthe coarse sand particles. Another option is tojust use a thinner layer. Special high-bond‘Thin-bed mortars’ (see BRE Digest (1998)2)designed for use in thicknesses from 1–5mmare now available but require units with closetolerances on their size.
30.3.1 Properties of unset mortar
Test methods for fresh mortar are given in BS4551 (1980), RILEM (1978), ISO (1991) and theEN 1015 series (1998–9). The important proper-ties of unset mortar are the workability, i.e. howeasy it is to handle and place on to the masonry;the pot life, i.e. how long it may be used for aftermixing (see EN 1015–9 (1999)); the water reten-tivity, i.e. how good it is at retaining wateragainst the suction of the units; the setting time;and the hardening rate. Associated parametersare the cement content, water content (often
expressed as the water/cement ratio w/c) and theair content (see EN 1015–7 (1999)).
The workability is measured in terms of theslump, consistence or the flow. The slump test isseldom now used for mortar. Flow is measuredon a standard 254mm diameter (ASTM) flowtable in EN 1015–3 (1999). Consistence is meas-ured by the dropping ball test or by the plungerpenetration test in EN 1015–4 (1999).
Water retentivity is measured by weighing theamount of water extracted from a dropping ballmould full of mortar by weighted layers of filterpaper through two layers of cotton gauze (see EN1015–8 (1999)). The consistence retentivity ismeasured by repeating the consistence test on thedewatered mortar.
Air content may be measured by weighing aknown volume (0.5 litres) of mortar and then cal-culating, using data on the relative densities andproportions of the constituents, from the follow-ing formula:
A�100(1�K�) (30.3)
where A is the air content, � is the relative densityof the mortar and K is derived from:
K�
where M1, M2, . . . , etc., are the relative masses ofthe constituents of the mortar of relative densities(specific gravities) d1, d2, etc., and Mw is the rela-tive mass of water. The sum of all the values ofM1, M2, . . . and Mw will be equal to 1. For precisemeasurements it is necessary to measure the spe-cific gravity of all the constituents separately byuse of the density bottle method. If this cannot bedone, the following default values are suggested:opc 3.12, masonry cement 3.05, silica sand 2.65,white hydrated building lime 2.26, grey hydratedlime 2.45. Alternatively the pressure method canbe used. All the methods are detailed in BS 4551(1980), which will be replaced by the BS EN1015-series (1999). The water content can bedetermined independently on fresh mortar byrapid oven drying of a weighed quantity. A
simple site test to independently measure cementcontent of fresh and unhardened mortars hasbeen described (Southern, 1989). A quantity ofmortar is weighed and dissolved in a predilutedand measured volume of acid held in an insulatedcontainer. The cement content can be calculatedfrom the temperature rise measured with an elec-tronic thermometer.
30.3.2 Properties of hardened mortar
The important properties of hardened mortar arethe density, permeability, Young’s modulus, com-pressive strength, flexural strength, bond strengthto units and drying shrinkage. Most of theseproperties may be measured by tests given instandards BS 4551 (1980), RILEM (1978), ISO(1991) and the BS EN 1015 series (1998–9). Thedurability is influenced by the combination ofother properties and may be determined using themethods proposed by Harrison et al. (1981,1986, 1990) and Bowler et al. (1995) now stan-dardized by RILEM (1998) and the draft Euro-pean method – prEN 1015–14 (1999).
The compressive strength is measured by acube-crushing test and the flexural strength ismeasured by the three-point bend (or modulus ofrupture) test (see BS 4551 (1980)) the draft of EN1015–11 (1999). The modulus of rupture (MOR)is based on the formula:
MOR�3PL/2bd 2 (30.4)
where P is the maximum load applied, L is thespan of the support rollers, b is the width of theprism and d is the depth of the prism.
The bond strength to typical units is conven-tionally measured by the parallel wallette test. Inthis test, a small wall is built and tested in thevertical attitude by a four-point bend test usingarticulated loading arms and supports, to preventthe application of any twisting moments, andresilient bolsters to prevent uneven loading.
The bond wrench has been introduced andspecified in some codes and standards such asASTM C1072–86 (1986) and Standards Associ-ation of Australia (1988) as a simpler way tomeasure bond, although it may give slightly
��M
d1
1� � �
M
d2
2� � . . .�Mw�
���M1 �M2 � . . .�Mw
Materials and components for brickwork and blockwork
280
higher values than the wallette test. This devicemeasures the moment required to detach a singleunit from the top of a wall or stack-bonded prismusing a lever clamped to the unit as in Figure30.2. The load may be applied in a variety ofways, most simply by filling a container with leadshot. An electronically gauged version isdescribed in BRE Digest 360 (1991) where theload is applied manually via a load cell trans-ducer. Its main advantage is that it can be usedon site for quality control and diagnosis of prob-lems and failures as well as a laboratory tool.
Table 30.5 gives the common formulations ofmortars used today and also lime mortars forrestoration works. The table gives some rangesfor the performance of the mortars in terms ofthe range of compressive, flexural and bondstrength. A ‘designation’ is a term used for a
Mortar
281
FIGURE 30.2 Bond wrench A before and B aftertest.
group of prescribed mortars giving approximatelysimilar performance.
It is clear from the table that a very wide rangeof strengths is possible for any nominal mix ratioand that parameters other than just the bindertype and content influence the strength, includingwater/cement ratio, sand grading and air content.A further factor which only affects the propertiesof the mortar in the bed (and not, of course,mortar specimens cast in impervious moulds) isthe amount of dewatering and compaction by theunits. Where dewatering occurs, it generallyincreases the intrinsic strength and density of themortar but may reduce the bond by also causingshrinkage and local microcracking.
Other key parameters are: density, porosity(indicative of permeability), Young’s modulusand drying shrinkage (see Chapter 19). Density(�) is measured by dividing the weight of a prismor cube by its volume. The volume may beobtained by measuring all the dimensions or byweighing the prism/cube submerged in water andcalculating:
� �m/bdL or � �m/(m�ms) (30.5)
where m�mass derived from a weighing, b is thewidth of the prism/cube, d is the depth of theprism/cube, L is the length of the prism/cube andms is the submerged mass. The density will varywith moisture content and W is measured as satu-rated or oven dry. Typical densities for mortarsare usually in the range 1500–2000kg/m3. Thewater porosity is obtained by measuring thewater absorption by evacuating the weighed dryspecimen then immersing it in water at atmos-pheric pressure and weighing. The gain in weightof water can be converted to volume and thendivided by the volume of the specimen to give thepercentage of porosity. The Young’s modulusmay be derived from the flexural strength testprovided deflection is measured. Drying shrink-age is measured by attaching precision referencepoints to the ends of a damp or saturated prismand then measuring the length with a micrometerscrew gauge or other precision length measuringdevice. After a drying regime the length is againmeasured and the change is expressed as a
percentage of the overall length. Various dryingregimes have been used (BRE Digest 35 (1963),RILEM (1975) and CEN (1991)). Some data onthese properties are given in Table 32.4.
30.3.3 Thin bed and lightweight mortars
Thin bed mortars (see BRE Digest 432 (1998),Fudge and Barnard (1998) and Phillipson et al.(1998)) are proprietary formulas but normallycontain some fine sand and cement plus bond-improving additives. They are normally suppliedas a dry premix which just needs thoroughmixing with water before use. The applicationtechnique is similar to that used for bedding tilesby use of serrated spreaders to produce thinribbed layers of the order of 2–5mm thick.Accurate levelling is crucial for good qualitywork. Lightweight mortars (see Stupart et al.(1998)) are thermally more efficient replacementsfor normal mortars and contain cement, option-ally lime and low-density fine aggregates such aspumice, pfa and perlite.
30.4 Fired clay bricks and blocksFired clay units are made by forming the unitfrom moist clay by pressing, extrusion or casting,followed by drying and firing (burning) to a tem-perature usually in the range 850–1300°C.During the firing process there are complex chemi-cal changes and the clay and other particles thatgo to make up the brick are bonded together bysintering (transfer of ions between particles atpoints where they touch) or by partial melting toa glass. During the drying and the firing processthe units generally shrink by several per cent fromtheir first-made size and this has to be allowed forin the process. Some clays contain organic com-pounds, particularly the coal measure shales andthe Oxford clay used to make Fletton bricks.Some clays are deliberately compounded withwaste or by-product organic compounds sincetheir oxidation during firing contributes to theheating process and thus saves fuel. The burningout of the organic material leaves a more open,lower density structure. The ultimate example ofthis is ‘Poroton’ which is made by incorporating
Materials and components for brickwork and blockwork
282
TABLE 30.4 Strength ranges for mortar
Mortar Proportions by volume of ingredients Strength properties (ranges)designation
Cement: Masonry: Cement: sand Observed strength ranges at 28 days forlime:sand cement:sand �plasticizer cement mortars and 90 days for lime
mortars
Compressive Flexural Bond(MPa) (MPa) (MPa)
(i) 1:0–��:3 – – 8–30 2.8–6.6 0.6–1.6(ii) 1: ��:4 �� 1:2.5–3.5 1:3–4 5–18 1.8–4.5 0.3–1.0(iii) 1:1:5–6 1:4–5 1:5–6 2–12 0.7–3.7 0.2–1.1(iv) 1:2:8–9 1:5.5–6.5 1:7–8 0.8–5.5 (0.7–1.7) (.36–.5)(v) 1:3:10–12 1:6.5–7 1:8 0.5–1.0 (0.7–0.9) no data(vi)1 0:1:2–3 – – 0.5–1.0 no data no data(vii)2 0:1:2–3 – – 0.5–1.0 no data no dataThin bed – – 3 5–10 – 0.14–0.66Lightweight 1:1:53 - 1:53 1.7–3.6 0.7–1.4 0.05
Notes:1. Hydraulic lime mortars.2. Pure lime mortars (referred to as ‘air lime motars’ in Europe).3. The exact composition of thin bed and lightweight mortars is a commercial secret.
fine polystyrene beads in the clay. Wood and coaldust can be used to achieve a similar effect insome products.
30.4.1 Forming and firing
Soft mud process
The clay is dug, crushed and ground then blendedwith water using mixers to make a relativelysloppy mud. A water content of 25–30 per cent isrequired for this process. In some plants otheradditives may be incorporated such as a propor-tion of already fired clay from crushed rejectbricks (grog), lime, pfa, crushed furnace clinkerand organic matter to act as fuel. In the well-known yellow or London Stock brick, groundchalk and ground refuse are added. The mud isformed into lumps of the size of one brick andthe lump is dipped in sand to reduce the sticki-ness of the surface. In the traditional techniquethe lump is thrown by hand into a mould and theexcess is cut off with a wire. This gives rise to thecharacteristic ‘folded’ appearance of the faces ofthe brick caused by the dragging of the clayagainst the mould sides as it flows. Nowadaysmost production is by machine which mimics thehandmaking process. These bricks usually have asmall frog (depression) formed by a raised centralarea on the bottom face of the mould. Because ofthe high drying shrinkage of such wet mixes andthe plasticity of the unfired (green) brick, the sizeand shape of such units are fairly variable. Thisvariability adds to their ‘character’ but meansthat precision brickwork with thin mortar beds isnot feasible. The finished brick is also fairlyporous which improves its insulation properties,and, paradoxically, its effectiveness as a rainscreen, but limits the strength.
Stiff plastic process
The clay is dug, crushed and ground then blendedwith water using mixers to make a very stiff butplastic compound with a water content of 10–15per cent. This is then extruded from the mixerand cut into roughly brick-shaped pieces and
allowed to dry for a short period before beingpressed in a die. The clay is very stiff so, whenejected straight from the mould, it retains veryprecisely the shape of the die. The low moisturecontent means that the shrinkage is low andtherefore the size is easier to control and thedrying time is relatively short. Another advantageis that the unfired brick is strong enough to bestacked in the kiln or on kiln cars without furtherdrying. This type of unit will usually have at leastone shallow frog and may have frogs in both bedfaces. The process is used to produce engineeringbricks, facing bricks, bricks with very accuratedimensions and pavers.
Wire cut process
Clay of softer consistency than the stiff plasticprocess is used with a moisture content of 20–25per cent and the clay is extruded from a rectangu-lar die with the dimensions of the length andwidth of the finished unit. The ribbon of clay, the‘column’, is then cut into bricks by wires set apartby the height of the unit plus the allowance forprocess shrinkage. The cutting machines areusually arranged such that the group of cuttingwires can travel along at the same speed as thecolumn while the cut is made. This means thatthe process is fully continuous and the cut is per-pendicular to the face and ends of the unit. Aplain die produces a solid column with just thecharacteristic wire cut finish, and these bricks willhave no depressions in their bed faces. In thisprocess it is easy, however, to include holes orperforations along the length of the column byplacing hole-shaped blockages in the die face.This has the following advantages:
l. Reduction in the weight of clay required perunit, so transport costs at every stage of theproduction and use of the units and all claypreparation costs, i.e. for shredding, grinding,mixing, etc., are also reduced.
2. A reduction in the environmental impact byreducing the rate of use of clay deposits andtherefore the frequency of opening up newdeposits.
Fired clay bricks and blocks
283
3. Reduction in the mass and opening up of thestructure of the units, thus speeding up dryingand firing, cutting the fuel cost for theseprocesses and reducing the capital cost of theplant per unit produced.
4. The oxidation of organic matter in the clay isfacilitated by increasing the surface area tovolume ratio and reducing the chance ofblackhearting.
5. The thermal insulation is improved. This hasa modest effect for UK-Standard size bricksbut the improvement can be substantial forlarge clay blocks.
6. The units are less tiring to lay because of thelower weight.
Because of these factors and the very large pro-portion of the production cost spent on fuel, mostclay units are perforated at least to the extent of10–25 per cent by volume. It should be stressedthat there is a penalty in that the clay must bevery well ground and uniform in consistence forsuccessful production of perforated units. Anylumps or extraneous air pockets can ruin the
column. To improve consistence, the mixers arecommonly heated and the front of the extruder isde-aired (evacuated) to prevent air bubbles.Figure 30.3 illustrates the production of a typicalthree-hole perforated brick by stiff plastic extru-sion and wirecutting.
Semi-dry pressing
This is one of the simplest processes of formingbricks. In the UK only Lower Oxford clay (orshale) is used, which comes from the Vale ofAylesbury and runs in a band towards the eastcoast. This clay contains about 7 per cent naturalshale oil which reduces the cost of firing but givesrise to some pollution problems. It is dug andthen milled and ground to go through a 2.4mmor 1.2mm sieve without altering the watercontent markedly from that as-dug (8 to 15 percent). The coarser size is used for common bricksand the finer for facings. The powdered clay isthen fed into very powerful automated presseswhich form a deep frogged, standard size brick
Materials and components for brickwork and blockwork
284
Extruder barrel
Die
Clay column
Movement ofcutting wires
Perforationsin column end
FIGURE 30.3 Stiff plastic extrusion of perforated clay bricks.
known as a Fletton (named after one of the earlymanufacturers). Unmodified Flettons are limitedto a single barred pink/cream colour. All facingFlettons are either mechanically deformed to givea patterned surface (rusticated) or an appliedsurface layer, such as coloured sand, is fired on.
Drying and firing in Hoffman kilns
The Hoffman kiln is a multichamber kiln inwhich the bricks remain stationary and the firemoves. It is mainly used for manufacture of Flet-tons. In the classic form it consists of a row ofchambers built of firebricks in the form of shorttunnels or arches. In the most efficient form thetunnels form a circle or oval shape and are con-nected together and to a large central chimney bya complex arrangement of ducts. A single ‘fire’runs round the circle and at any one time onechamber will be being loaded or ‘set’ ahead of thefire and one will be being unloaded behind thefire. The bricks are stacked in the kiln in groupsof pillars termed ‘blades’, which leaves lots of
space between units to enable the gases to circu-late freely. Chambers immediately in front of thefire will be heating up using the exhaust gasesfrom the hottest chamber and those further aheadwill be being dried or warmed by gases fromchambers behind the fire which are cooling down.
Figure 30.4 illustrates the broad principles ofthe system showing only the ducts actually in usefor the fire in one position. In practice the ductsare positioned to ensure a flow through from theinlets to the outlets. The whole process is veryefficient, particularly as a large proportion of thefuel is provided by the oil in the bricks them-selves. Because of the organic content of thebricks, the firing has to be done under oxidizingconditions during the last phases in order to burnout the oil. If this is not done the bricks have adark, unreduced central volume, known as ablackheart, which can give rise to deleterioussoluble salts. During this phase some fuel isadded to keep the temperature up. This is essen-tially a batch process and the average propertiesof the contents will vary a little from chamber to
Fired clay bricks and blocks
285
FIGURE 30.4 Principles of the Hoffman kiln.
chamber. Additionally, the temperature and oxi-dizing condition will vary with position in thechamber and thus some selection is necessary tomaintain the consistency of the product.
Drying and firing in tunnel kilns
Tunnel kilns, as in Figure 30.5, are the comple-ment of Hoffman kilns in that the fire is station-ary and the bricks move through the kiln asstacks on a continuous train of cars. In practice along insulated tunnel is heated in such a way thattemperature rises along its length, reaches amaximum in the centre and falls off again on theother side. To maximize efficiency, only the firingzone is fuelled and hot gases are recycled fromthe cooling bricks and used to heat the dryingand heating-up zones of the kiln. Most extrudedwirecuts and stiff plastic bricks are now fired insuch kilns which are continuous in operation.Stocks and other mud bricks may also be firedthis way after a predrying phase to make themstrong enough to withstand the stacking forces.
Clamps
Clamps are the traditional batch kilns comprisinga simple insulated refractory beehive-shapedspace with air inlets at the base and a chimneyfrom the top.
Intermittent kilns
These are the modern version of the clamp wherethe units are fired in batch settings using oil orgas as a fuel. They are now only used for produc-tion of small runs of specially shaped brick ‘spe-cials’.
30.4.2 Properties
Clay bricks probably have the widest range ofstrengths of any of the manufactured masonrymaterials with the compressive strength rangingfrom 10MPa for an underfired soft mud brick toas much as 200MPa for a solid engineering brick.The common shapes and terminology are shownin Figure 30.6. The compressive strength is meas-
Materials and components for brickwork and blockwork
286
Flow of hot gases
Flow of hot gasesFuel/air
Stacking Drying Heating Firing Initial cooling Final cooling Unloading
1000
500
20
Tem
pera
ture
°C
Position along the kiln
Kiln car movement
FIGURE 30.5 Tunnel kiln – principles.
ured by a crushing test on whole units with thestress applied in the same direction as the unitwould be loaded in a wall. Solid and perforatedunits are tested as supplied, but frogs are nor-mally filled with mortar as they would be in awall. In BS 3921 (1985), thin plywood sheetpacking is used to try to reduce the effect of highspots and unevenness of the faces. The new Euro-pean test method EN 772–1 (2000) uses eithermortar capping or face grinding to achieve evenloading. The quoted strength is the average of sixto ten determinations of stress based on the loaddivided by the area of the bed face. The flexuralstrength and modulus of elasticity is not normallya designated test parameter in unit standards butis important to obtain data for Finite elementmodels. A standard three-point bending methodis available in RILEM LUM A.2 (1994). Units
Fired clay bricks and blocks
287
FIGURE 30.6 Shape types and terminology ofmasonry units. Based on Figure 1 in BRE Digest–441published by CRC Ltd.
having a span to depth ratio in the test geometryexceeding 4 may be tested without any furtherpreparation provided they are sufficiently flat andtrue to allow even loading. Units which are notflat or true may need to be made true by applyinga layer of rapid-setting mortar or plaster to thebearing area. Units having a span/depth ratio ofless than 4 have too low an aspect ratio to givetrue bending. The span/depth ratio may beincreased by gluing three units together in a lineand testing the central unit.
The flexural strength (S) is given by:
S�M/Z (30.6)
whereM� the bending moment at failure, in Newton
millimetres�Pl/4,Z �section modulus of test specimen in cubic
millimetres�bd 2/6,P �maximum load applied to the prism, in
Newtons,l �distance between the axes of the support
rollers, mm,b � the mean width of prism at the line of frac-
ture, mm� (b1 �b2)/2,d � the mean height of prism at the line of frac-
ture, mm� (dl �d2)/2.
Units having a span/depth ratio of less than 4have too low an aspect ratio to give true bending.The span/depth ratio may be increased by gluingthree units together in a line and testing thecentral unit.
Other important properties are the dimensions,water absorption (and porosity), suction rate,density and soluble salts content. In BS 3921 thedimensions are measured by laying twenty-fourunits in a line and dividing the overall length bytwenty-four, i.e. it is an averaging process andgives no data on individual variability. Theprocess is repeated for each dimension with theunits in the appropriate attitude. The equivalentCEN standard has individual tolerances for eachunit. Water absorption and density are measuredin the same way as for mortar (see Section30.3.2), except that the preferred saturation tech-nique is to boil the units in water for 5 hours.
The initial rate of absorption (IRA, suction rate)is measured to give some idea of the effect of theunit on the mortar. Units with high suction ratesneed very plastic, high water/cement ratiomortars, while units with low suction rates needstiff mortars. The parameter is measured bystanding the unit in 3mm depth of water andmeasuring the uptake of water in 60 seconds.
The IRA is calculated using:
wi � (m2 �m1)/Lb (30.7)
wherewi � initial rate of absorption,m1 � initial mass of the unit/specimen,m2 �mass after 60 seconds of water absorption,L� length of the bed (mortar) face in service toan accuracy of 0.5 per cent,b�width of the bed (mortar) face in service to anaccuracy of 0.5 per cent.
The result is normally given in units ofkg/m2/min.
The content of soluble salts is measured by stan-dard wet chemical analysis techniques or bymodern instrumental techniques such as flamephotometry. The elements and compounds ofconcern are sulfates, sodium, potassium, calciumand magnesium. Table 30.5 gives typicalvalues/ranges for some of the key properties ofclay bricks.
In most brickwork, bricks are loaded upontheir normal bed face but often they are loadedon edge or on end. Typical examples are headers
and soldiers (see Figure 31.3) in normal walls,stretchers in arches and reinforced beams andheaders in reinforced beams. While solid bricksshow a small variation in strength for loading indifferent directions, due to the change in aspectratio (height/thickness), perforated, hollow orfrogged units may show marked differences asillustrated in Table 30.6 from data in Lenczner(1977), Davies and Hodgkinson (1988) and Sinhaand de Vekey (1990).
Taking the simplest geometry as an example, itcan be seen from Figure 30.7 that the minimumcross-sectional area of the 5-slot unit resisting theload will be 80 per cent on bed but 71 per centon edge and 25 per cent on end. The ratios of thestrengths in Table 30.6 follow approximately theratios of the areas. Other factors such as the slen-derness of the load-bearing sections and the effectof high local stresses at rectangular slot endscomplicate the behaviour and may explain thevariations between different types. It can also beshown that porosity, in the form of vertical perfo-rations result in a smaller reduction of thestrength of a material than does generally distrib-uted porosity and is the more efficient way ofreducing the weight. This is illustrated in Figure30.8. More detailed information on clay brickproperties is contained in BRE Digests 441 Part 1and 2 (1999).
Materials and components for brickwork and blockwork
288
TABLE 30.5 Properties (typical ranges) for UK brick types
Brick type Compressive Water Water Suction rate Flexuralstrength absorption porosity (IRA) Bulk density strength(MPa) (Weight %) (Volume %) (kg/m2/min) (kg/m3) (MPa)
Handmade facing 10–60 9–28 19–42 – – –London Stock 5–20 22–37 36–50 – 1390 1.6Gault wirecut 15–20 22–28 38–44 – 1720 –Keuper marl wirecut 30–45 12–21 24–37 – 2030 –Coalmeasure shale 35–100 1–16 2–30 – 2070 –Fletton 15–30 17–25 30–40 1.0–2.0 1630 2.8Perforated wirecut 35–100 4,5–17 – 0.2–1.9 1470–2060 7Solid wirecut 20–110 4–21 10–35 0.25–2.00 1700–2400 6.5
Fired clay bricks and blocks
289
FIGURE 30.7 Area of 5-slot brick resisting load inthe three orientations.
FIGURE 30.8 Compressive strength of ceramicbodies as a function of: (a) general porosity; (b) verti-cally aligned perforations of constant diameter.
TABLE 30.6 Properties of some UK brick types in various orientations
Type Ref. Perforation CompressiveProportion of the on-bed strength
Water Suctionstrength on bed
as a percentageabsorption rate (IRA)
(Vol. %) (MPa) On bed On edge On end (Weight %) (kg/m2/min)
23 hole (a) ? 65.5 100 29 11 6.9 0.614-hole (b) 21.3 74.3 100 35 14 3.9 –14 hole (c) ? 44.8 100 29 11 4.7 1.810 hole (d) 23.1 70.2 100 42 31 5.4 –3 hole (e) 12.2 82 100 65 49 4.2 –3 hole (f) ? 57.8 100 36 25 8.5 1.55 cross slots (g) 20 64.1 100 81 22 3.4 –16 hole (A) 20.1 64.7 100 31 13 5.5 0.35Frogged (B) 15.1 25.4 100 41 36 21.7 2.86Frogged (C) 6.2 33.7 100 49 51 14.4 1.06Frogged (D) 8.6 31.7 100 92 50 11.9 0.54Solid (E) 0 43.5 100 67 65 22.8 3.31
Note: The reference given links the strength data in this table to strength data for masonry specimens in Table 32.1.
30.5 Calcium silicate unitsThese are manufactured from mixtures of lime,silica sand and water. Aggregates such as crushedrocks or flints may be incorporated to alter theperformance and appearance, and pigments maybe used to vary the colour. Common colours arewhites, blacks, buffs and grey-blues. Reds areproduced but they seldom have the richness offired clay units. There is only one basic process,in which the mixture is pressed to high pressuresin a die in a static press, ejected, set on cars andthen placed in autoclaves and cured in high pres-sure steam for several hours. The mix is invari-ably fully compacted and makes a very preciselyshaped, low tolerance unit with sharp, well-defined corners (arrises) and a fairly smoothfinish. The properties commonly measured are
given in Table 30.7. Additionally a drying shrink-age test may be carried out.
30.6 Concrete unitsConcrete units have been made since the 1920sand were widely used, in the form of the ‘breeze’block, to build partitions in houses in the build-ing boom of the 1930s. In the past 40 years,however, the range of products has expandedenormously to cover facing bricks and blocks,high-strength units, simulated stone units, ther-mally insulating blocks and pavers. Chapters 13to 24 describe the performance and behaviour ofconcrete. Most aggregate concrete units are pro-duced by pressing specially designed mixes. Someof the processes used and the resulting products,
Materials and components for brickwork and blockwork
290
TABLE 30.7 Property ranges of calcium silicate bricks (West et al., 1979)
Brick type Compressive Water absorption Initial rate of Bulk density Frog volumestrength suction(MPa) (weight %) (kg/m2/min) (kg/m3) (%)
Solid 20–50 8–22 0.25–2.0 1750–2000 0Frogged 20–55 13–20 0.5–1.2 1650–1950 4–7
TABLE 30.8 Properties of some typical aggregate concrete blocks
Unit type Size Void Bulk Concrete Compressive Flexural Young’s Water3
L�H�T density density strength1 strength2 modulus absorption(mm) (%) (kg/m3) (kg/m3) (MPa) (MPa) (GPa) (weight %)
Brick 215�65�103 0 2160 2160 32.5 – – 6.3DA Block 438�213�98 0 2140 2140 15.5 2.59 – 10.9DA Block 390�190�140 41.6 1350 2320 31.6 – 42.3 –LWA Block 390�190�140 22.1 1630 2090 21.5 – 32.8 –LWA Block 390�190�140 0 2170 2170 29.9 – 23.6 –LWA Block 390�190�90 0 2060 2060 21.6 – 17.5 –LWA Block 390�190�90 19.9 1100 1380 8.1 – 9 –LWA Block 439�215�98 0 2190 2190 6.6 0.5 – 35
Notes:1. Compressive strength tested wet and mortar capped in accordance with BS 6073 Appendix B. Similar values are obtained by
testing between fibreboard at ex-factory water content.2. Flexural strength (MOR) measured in accordance with BS 6073 Appendix C.3. Measured by vacuum absorption.
e.g. autoclaved aerated concrete (AAC), arewholly different from normal concrete; thus a fulldescription is given.
30.6.1 Production processes
Casting concrete
Concrete blocks can be manufactured by pouringor vibrating a concrete mix into a mould anddemoulding when set. While this method is used,particularly for some types of reconstituted stoneor faced blocks, it is not favoured because of itsslowness and labour demands.
Pressing of concrete
This is a widely used method for producing solidand hollow bricks and blocks either in dense con-crete or as a porous open structure by using gap-graded aggregates and not compacting fully. Themachine is basically a static mould (or die), whichis filled automatically from a mixer and hoppersystem, and a dynamic presshead which compactsthe concrete into the die. After each productioncycle the green block is ejected onto a conveyorsystem and taken away to cure either in air oroften in steam. The presshead may have multipledies. A variation of the method is the ‘egg layer’.This performs the same basic function as a staticpress but ejects the product straight onto thesurface on which it is standing and then movesitself to a new position for the next productioncycle.
Curing
All aggregate concrete products may either becured at ambient or elevated temperature. Theelevated temperatures are usually achieved by theuse of steam chambers and allow the manu-facturer to decrease the curing period or increasethe strength or both. Products cured externallyshould not be made when the temperature is nearor below 0°C since they will be damaged byfreezing while in the green state.
Autoclaved aerated concrete (AAC)
AAC is made by a process, developed originallyin Scandinavia, to produce solid microcellularunits which are light and have good insulatingproperties. The method involves mixing a slurrycontaining a fine siliceous base material, a binder,some lime and the raising agent, aluminiumpowder, which reacts with the alkalis (mainlycalcium hydroxide) to produce fine bubbles ofhydrogen gas as per equation (30.8). Thismixture is poured into a mould maintained inwarm surroundings and the hydrogen gas makesthe slurry rise like baker’s dough and set to aweak ‘cake’. The cake is then cured for severalhours at elevated temperature, demoulded,trimmed to a set height and cut with two orthog-onal sets of oscillating parallel wires to the unitsize required using automatic machinery. The cutunits are then usually set, as cut, onto cars whichare run on rails into large autoclaves. Thecalcium silicate binder forms by reaction underthe influence of high-pressure steam. Additionalcuring after autoclaving is not necessary as all theunits can be incorporated in work as soon as theyhave cooled down.
The binder reaction is conventional, as given inSection 30.1. The cellular structure gives theproduct good thermal properties and a highstrength/density ratio. The light weight allowslarger units to be handled comfortably so somedouble-size units are now produced for use withthin-bed mortar.
Ca(OH)2 �2Al�2H2O→CaAl2O4 �3H2 (30.8)
This produces a structure of small, closed cellssurrounded by cell walls composed of a finesiliceous aggregate bound together by calciumsilicate hydrates. The nature of the principalsiliceous material is identifiable from the colour:ground sand produces a white material and pul-verized fuel ash a grey material.
Because of the light weight (low density), theproduct can be made into large-sized blocks whileremaining handleable. The largest units currentlyavailable are 447mm by 447mm face size whichare designed for building thin-joint masonry (see
Concrete units
291
BRE Digest 432 (1998)). Still the most commonsize for normal work is 440�215�100 (orthicker).
30.6.3 Products and their properties
Dense aggregate concrete blocks andconcrete bricks
These are generally made from well-gradednatural aggregates, sands, pigments and ordinaryor white Portland cement by static pressing to awell-compacted state. Figure 30.9 (a) illustratesthe principle of such materials where the voidsbetween large particles are filled with smaller par-ticles. They are strong, dense products and areoften made with a good surface finish suitable forexternal facing masonry. They are also suitablefor engineering applications. Bricks are producedmainly as the standard size (215�102�65mm)in the UK but in a wide range of sizes in conti-nental Europe. Blocks are produced either solidor hollow by varying the quantity of mix and theshape of the press platen. In order to facilitatedemoulding the hollows will always have a slighttaper. The hollows in UK products are alldesigned to run vertically in the finished masonryas this gives the optimum strength to weightratio. The face size of UK units is generally440mm long by 215mm high but the thicknessmay vary from 50mm to 300mm. Some of theimportant properties are summarized in Table30.9.
Reconstructed stone masonry units
These have essentially the same specification asdense-aggregate concrete blocks except that themain aggregate will be a crushed natural rocksuch as limestone or basalt and the other mater-ials will be chosen such that the finished unitmimics the colour and texture of the naturalstone.
Materials and components for brickwork and blockwork
292
Lightweight aggregate concrete blocks
These are generally produced as load-bearingbuilding blocks for housing, small industrialbuildings, in-fill for frames and partition walling.High strength and attractive appearance arerarely the prime consideration but handlingweight, thermal properties and economy areimportant. Inherently low-density aggregates areused and are often deliberately gap-graded, asillustrated by Figure 30.9 (b), and only partlycompacted to keep the density down. They willfrequently be made hollow as well to reduce theweight still further. The aggregates used are sin-tered pfa nodules, expanded clay, furnace clinker,furnace bottom ash, pumice or foamed slagtogether with sand and binder. Breeze is a tradi-tional term for a lightweight block made fromfurnace clinker. Often low-density fillers or aggre-gates such as sawdust, ground bark or poly-styrene beads are incorporated to further reducethe density. They are produced either by staticpressing or in egglayer plants. Some of theimportant properties are summarized in Table30.8. The properties commonly measured includethe compressive strength by the method used forclay bricks for brick-sized units but capped withmortar to achieve flat parallel test faces forblocks. The flexural strength has also been usedto evaluate partition blocks which bear lateralloads but only self-weight compressive loads. It isa simple, three-point bend test of the typedescribed in Section 30.4.3 for clay units with anaspect ratio greater than 4. Other propertiesmeasured include dimensions, water absorptionby the method of vacuum absorption, density anddrying shrinkage.
Autoclaved aerated concrete (AAC)
Fine sand or pulverized fuel ash or mixturesthereof is used as the main ingredient. The binderis a mixture of ordinary Portland cement (opc), togive the initial set to allow cutting, and limewhich reacts with the silica during the auto-claving to produce calcium silicate hydrates andgives the block sufficient strength for normal
building purposes. Some of the key properties aresummarized in Table 30.9.
30.7 Natural stoneThe UK possesses a great variety of geologicaltypes of natural stone and has a long tradition ofbuilding masonry using locally available stones.Used locally, it is environmentally friendly sincethe only energy input is for cutting and trans-portation. The main types used in the UK are‘sedimentary rocks’ formed from compressed sedi-
ments from the bottom of ancient seas – the lime-stones and sandstones; ‘metamorphic rocks’formed by the action of pressure and high tem-perature on sedimentary deposits – the marblesand slates; and ‘igneous rocks’ formed by meltingof rock during volcanic activity – the granites andbasalts. Both strength and durability are veryvariable and will depend on the porosity and the distribution of pores. Generally, strengthincreases and porosity decreases from sedimen-tary through metamorphic to igneous. Igneousrocks are fairly homogeneous, i.e. have similar
Natural stone
293
(a) (b)
FIGURE 30.9 Types of aggregate (a) well graded; (b) gap graded (schematic).
TABLE 30.9 Properties of AAC
TypeDry
Compressive strengthFlexural Tensile Young’s Thermal
densitytested to BS 6073: 1981
strength strength modulus2 conductance(Typical (Nominal (Typical1 (Typical (Typical (GPa) @ 3% moisturekg/m3) MPa) MPa) MPa) MPa) (W/m°K)
Low density 450 2.8 3.2 0.65 0.41 1.6 0.12525 3.5 4 0.753 0.52 2.003 0.14
Standard 600 4 4.5 0.85 0.64 2.4 0.16675 5.8 6.3 1.003 0.76 2.55 0.18
High density 750 7 7.5 1.25 0.88 2.7 0.2
Notes:1. Nominal values are manufacturer’s declared values while the alternatives are values typical of modern production plants.2. Linear extrapolation of a limited range of splitting tests (Grimer and Brewer, 1965).3. Interpolated value.
Materials and components for brickwork and blockwork
294
TABLE 30.10 Properties of some types and examples of stone
Type of stone Density Water Porosity Compressive Young’sabsorption strength modulus
(kg/m3) (% w/w) (% v/v) (MPa) (GPa)
Typical rangesLimestone 1800 to 2700 0.1 to 17 0.3 to 30 20 to 240 1 to 8Sandstone 2000 to 2600 0.4 to 15 1 to 30 20 to 250 0.3 to 8Marble 2400 to 2800 0.4 to 2 0.4 to 5 40 to 190 –Slate 2600 to 2900 0.04 to 2 0.1 to 5 50 to 310 –Granite 2500 to 2700 0.04 to 2 0.1 to 4 80 to 330 2 to 6Basalt 2700 to 3100 0.03 to 2 0.1 to 5 50 to 290 6 to 10
Individual examplesYork sandstone 2560 2.62 7.6 72.6 –Portland limestone 2209 5.33 11.8 32 –
properties in all directions, but the other typeshave a layer structure and will be significantlystronger normal to the bed plane than in theother two directions (inhomogeneous). To ensurethe optimum performance the layered rocks areusually cut to maintain the bed plane perpendicu-lar to the compressive stress field in the building,e.g. horizontal in normal load-bearing masonry.This also gives the optimum durability. BREDigest 420 (1997) is a good introduction andgives a useful set of references, and the GeologicalSociety SP16 (1999) has up-to-date and compre-hensive coverage. Some performance character-istics of well-known types are given in Table30.10 and movement data is given in Table 32.4.
30.8 Ancillary devices – ties andother fixings/connectorsIn order to ensure their stability, masonry ele-ments need to be connected to either othermasonry elements to form stable box structuresor to other elements such as frames, floors, roofs,beams and partitions. A huge range of devicesexists, summarized in Table 30.10 together withinformation on the current and future EuropeanStandards and guidance literature. Most of thesedevices are made from metal, predominantly gal-vanized mild steel, austenitic stainless steel and
bronzes. A few light-duty tie products are madefrom plastic.
30.9 ReferencesAshurst, J. (1983) Mortars, plasters and renders. In
Conservation, Ecclesiastical Architects and Survey-ors Association, London.
BRE Digest–441, Clay bricks and clay brickwork (parts1 and 2) (1999) CRC Ltd, Watford.
BS EN 772–1 (2000) Methods of test for masonryunits. Part 1. Determination of compressive strength(Note other BS EN standards are listed in the furtherreading section).
Central Electricity Generating Board (1972) PFA Uti-lization.
Cowper, A.D. (1927) Lime and Lime Mortars(reprinted 1998, BRE Ltd).
Davies, S. and Hodgkinson, H.R. (1988) TheStress–Strain Relationships of Brickwork WhenStressed in Directions Other Than Normal to theBed Face: Part 2, RP755, British Ceramic ResearchEstablishment, Stoke on Trent.
Fudge, C.A. and Barnard, M. (1998) Development ofAAC masonry units with thin joint mortar forhousebuilding in the UK, Proc. 5th InternationalMas. Conf. London 1998 (BMS Proc. 8), 384–7.
Grimer, F.J. and Brewer, R.S. (1965) The within cakevariation of autoclaved aerated concrete. Proceed-ings of Symposium on Autoclaved Calcium SilicateBuilding Products, Society of Chemical Industry,May, 163–70.
Harrison, W.H. (1981) Conditions for sulphate attackon brickwork. Chemistry and Industry.
295
TA
BL
E 3
0.1
1A
ncill
ary
devi
ces
Con
nect
ing
devi
ceP
urpo
seSp
ecifi
cati
on d
ocum
ent1
Gui
danc
e do
cum
ents
1
Cav
ity
wal
l tie
sT
o lin
k th
e tw
o le
aves
of
cavi
ty w
alls
and
sha
reB
S 12
43,
BR
E I
P16/
(198
8),
tens
ion,
com
pres
sion
and
she
ar lo
ads
BS
DD
140,
BS
5628
(19
93),
3
EN
845
–12
BR
E I
P11
(200
0)
Supp
ort
ties
To
link
wal
ls t
o fr
ames
BS
DD
140,
BR
E I
P17/
(198
8),
EN
845
–12
BS
5628
:p1
(199
3)B
RE
IP1
1 (2
000)
Shea
r ti
esT
o gi
ve s
hear
con
nect
ions
bet
wee
n el
emen
tsE
N 8
45–1
2B
S 56
28:p
1 (1
993)
4
BR
E I
P11
(200
0)
Slip
tie
sT
o gi
ve s
hear
con
nect
ions
bet
wee
n el
emen
ts w
hile
EN
845
–12
BS
5628
:p3
(198
5)al
low
ing
in-p
lane
mov
emen
tB
RE
IP1
1 (2
000)
Stra
psT
o ti
e w
alls
to
roof
s an
d flo
ors
BS
8103
,B
S 81
03 (
1995
), B
RE
E
N 8
45–1
2G
BG
29
(199
7)
Jois
t ha
nger
sT
o su
ppor
t flo
ors
on u
nper
fora
ted
wal
ls v
ia t
he jo
ists
BS
6178
: Par
t 1,
BS
8103
(19
95),
EN
845
–12
BS
5628
:p1
(199
3)
Bra
cket
sT
o su
ppor
t th
e lo
ad o
f w
alls
on
fram
esE
N 8
45–1
21
Ang
les
To
supp
ort
wal
ls o
n fr
ames
11
Anc
hors
and
Fixi
ng d
evic
es s
uch
as s
crew
s, n
ails
, scr
ew-p
lug
1B
RE
Dig
est
329,
(20
00 e
diti
on)
fixin
gssy
stem
s, r
esin
-anc
hors
, exp
andi
ng a
ncho
rs, f
orC
IRIA
TN
137
: 199
1 at
tach
ing
fitti
ngs
to w
alls
W.S
. Pat
erso
n
Not
es:
1.M
ost
man
ufac
ture
rs c
an s
uppl
y da
ta o
n sp
ecifi
cati
on a
nd in
stal
lati
on in
stru
ctio
ns.
2.T
he C
EN
sta
ndar
d E
N 8
45–1
cov
ers
mos
t of
the
se d
evic
es b
ut is
unl
ikel
y to
be
issu
ed in
its
final
BS
EN
for
m t
ill 2
001.
3.Il
lust
rate
d in
Fig
ure
31.4
.4.
Illu
stra
ted
in F
igur
e 31
.5.
Harrison, W.H. (1986) Durability tests on buildingmortars – effect of sand grading. Magazine of Con-crete Research 38, 135, 95–107.
Harrison, W.H. and Bowler, G.K. (1990) Aspects ofmortar durability. Br. Ceram. Trans. J. 89, 3,93–101.
Lenczner, D. (1977) Strength of Bricks and BrickworkPrisms in Three Directions, Report No. 1, Universityof Wales Institute of Science and Technology,Cardiff.
Phillipson, M.C., Fudge, C.A., Garvin, S.L. andStupert, A.W. (1998) Construction with thin jointmortar systems and AAC blockwork, Proc. 5thInternational Mas. Conf. London (BMS Proc. 8),388–90.
RILEM (1975) Recommendations for testing methodsof aerated concrete. Materials and Structures 8, 4S,211.
RILEM (1978) Recommendations for the testing ofmortars and renderings. Materials and Structures 11,63, 207.
RILEM Technical recommendations for the testing anduse of construction materials (1994), E & FN Spon,London, LUM A.2 Flexural strength of masonryunits, 459–61.
Sinha, B.P. and de Vekey, R.C. (1990) A study of thecompressive strength in three orthogonal directionsof brickwork prisms built with perforated bricks.Masonry International 3, 3, 105–10.
Smith, M.R. (ed.) (1999) Building Stone, Rock Fill and
Armourstone in Construction, Special PublicationNo. 16, Geological Society, London.
Southern, J.R. (1989) BREMORTEST: A RapidMethod of Testing Fresh Mortars for CementContent, BRE Information Paper IP8/89, BuildingResearch Establishment, Watford.
Stupert, A., Skandamoorthy, J.S. and Emerson, F.(1998) Thermal and strength performance of twolightweight mortar products, Proc. 5th InternationalMas. Conf. London (BMS Proc. 8), 103–6.
Task group of ASTM Subcommittee sectionC09.03.08.05 (1980) Pigments for integrallycoloured concrete. Cement, Concrete and Aggregates,CCAGDP, 2, 2, Winter 1980, 74–7.
de Vekey, R.C. (1975) The properties of polymer modi-fied cement pastes. Proceedings of the First Inter-national Polymer Congress, Congress on PolymerConcrete, London, VI, 94–104.
de Vekey, R.C. and Majumdar, A.J. (1977) Durabilityof cement pastes modified by polymer dispersions.Materials and Structures 8, 46, 315–21.
Vicat, L.J. (1837) Treatise on Calcareous Mortars andCements, translated by Smith, J.T., reprintedDonhead Publishing, Shaftesbury, 1997.
West, H.W.H., Hodgkinson, H.R., Goodwin, J.F. andHaseltine, B.A. (1979) The Resistance to LateralLoads of Walls Built of Calcium Silicate Bricks,BCRL, Technical Note 288.
WRC Information and Guidance Note IGN 4–10–01(1986) Bricks and Mortar, Water Research Council,Oct.
Materials and components for brickwork and blockwork
296
Chapter thirty-one
Masonryconstructionand forms
31.1 Introduction31.2 Mortar31.3 Walls and other masonry forms31.4 Bond patterns31.5 Specials31.6 Joint-style31.7 Workmanship and accuracy31.8 Buildability, site efficiency and productivity31.9 Appearance31.10 References
31.1 IntroductionThis chapter is concerned with how masonry isbuilt and the architectural forms used. Appear-ance is a very important aspect; the basic struc-tural form of many types of masonry is expressedon the surface of the buildings and other struc-tures and can be a very attractive and reassuringaspect of these structures. Appearance is a synthe-sis of the size, shape and colour of the units, thebond pattern, the mortar colour and finish, themasonry elements – walls, piers, columns,corbels, arches, etc. – and the scale and propor-tion of the whole structure. Other key aspects arethe workmanship, the accuracy, the detailing inrelation to other features and the use of speciallyshaped units.
The basic method of construction has hardlychanged for several thousand years: units are laidone on top of another in such a way that theyform an interlocking mass in at least the two
horizontal dimensions. It is not practical toachieve interlocking in a third dimension withnormal rectangular prismatic units but a degreeof such interlocking is sometimes used in ashlarstonework. Most practical masonry employs amortar interlayer to allow for small inaccuraciesof size between units and to make walls water-tight, airtight and soundproof.
31.2 MortarMortar must be mixed thoroughly and its pro-portions kept constant. It should not be over-mixed as this will introduce too much air andreduce the durability. Portland cement mortarshave a pot life of about one hour and they shouldbe discarded after the time limit or if they becomeunworkable. Traditional lime mortars andretarded opc mortars have a much longer pot lifewhich makes them easier to use. For optimumbond and ease of laying, the mortar shouldalways be matched to the units being laid. As ageneral rule dense low absorption units need afairly firm, dry mortar (e.g. with a dropping ballconsistence of around 10 or less), while highabsorption units need a sloppy wet mortar with adropping ball consistence of 13–15. Alternativetechniques for improving usability are the use ofagents in the mortar which hold water in (waterretentivity aids) or wetting the units to reducetheir suction.
297
31.3 Walls and other masonry forms
Walls are built by laying out a plan at foundationlevel and bringing the masonry up layer by layer.To maximize the strength and attractiveness it isimportant to make sure that all the foundationlevels are horizontal, are accurate to the plan andallow multiples of whole units to fit most runsbetween returns, openings, etc. It is also essentialto maintain the verticality (plumb), the level ofbed joints and the straightness of the masonrywithin reasonable limits. The thickness of themortar joints must be kept constant within asmall range, otherwise the masonry will lookuntidy. The standard technique used is to gener-ate reference points by building the corners(quoins) accurately using a plumb bob and line, a
builder’s level, a straight edge and a rule. Anyopenings are then filled with either a temporaryor permanent frame placed accurately in the planposition. Lines are stretched between the refer-ence points and the intervening runs of masonryare built up to the same levels.
Columns, piers and chimneys are built in thesame way but need plumbing in two dimensionsand more care because of their small dimensions.
Arches and tunnels must follow a curved shapedefined by the architect or engineer and aretraditionally built on timber formwork.Adjustable reusable metal formwork systems arealso available. Some arches use special taperedunits called voussoirs but large radius or shallowarches may be built with standard units andtapered joints. Figure 31.1 illustrates the main
Masonry construction and forms
298
FIGURE 31.1 Structural elements and terminology of arches.
elements of arches and the special vocabulary forthem.
Reinforced and post-tensioned masonry areused to a limited extent in the UK mainly for civilengineering structures, high single-storey halls,retaining walls and lintels within walls. Masonrylintels may sometimes be constructed by layingspecial bed joint reinforcement in the mortar.This acts as a tension reinforcement for amasonry beam. Most other reinforced masonry isformed by building masonry boxes in the form ofhollow piers, walls with cavities or walls withslots in them, and then locking the reinforcingelements into the voids using a concrete grout.
Post-tensioned masonry may be built in the sameway but the reinforcement is then passed throughthe cavities and stressed against end plates whichremoves any necessity to fill with grout. Figure31.2 shows some typical reinforced masonryforms.
31.4 Bond patternsMost modern masonry is the thickness of a singleunit breadth and is built by overlapping half thelength with the next unit. This is known asstretcher bond or half bond and is shown inFigure 31.3. Variations of stretcher bond may be
Bond patterns
299
FIGURE 31.2 Reinforced and post-stressed masonry forms: (a) bed joint reinforced; (b) reinforced pocketwalls; (c) grouted cavity; (d) quetta bond.
achieved by using third or quarter bond (alsoshown). Soldier courses, where all the units standon their ends, may be incorporated as a decora-tive feature but reduce the strength and robust-ness of the masonry. Much of this stretcherbonded work is used as cladding to frame struc-tures where the strength is less important becauseof the presence of the supporting structure. Inoccupied structures it is widely used in the formof the cavity wall, as illustrated in Figure 31.4,which comprises two such walls joined with flex-ible metal ties across a space which serves princip-ally to keep out rain and keep the inner wall dry.Blockwork is almost universally built with thisbond and broader units are used to achievethicker walls. Stretcher bonded walls may be builtthicker by linking two or more layers with strongmetal ties and filling the vertical ‘collar’ joint withmortar. The collar jointed wall, Figure 31.5, maybe built with a smooth true face (‘fair face’) onboth visible sides by taking up any thickness vari-ations by varying the thickness of the collar joint.
Masonry construction and forms
300
FIGURE 31.3 Half-brick bonds.
FIGURE 31.4 Typical block/brick cavity wall.
FIGURE 31.5 Collar-jointed brick wall.
In thicker walls, built in multiples of a singleunit breadth, there are a large number of possibletwo-dimensional bonding patterns availableknown by the traditional names. A few of thewidely practised bonds are shown in Figure 31.6and more are given in BS 5628: Part 3 and Handi-syde and Haseltine (1980).
31.5 SpecialsIt has always been possible to make structuresmore interesting by using specially shaped units
to vary angles from 90°, to generate tapers,plinths, curves, etc. In recent years such featureshave, if anything, become more popular. A verylarge range of shapes is available on a regularbasis and are known as ‘standard specials’. Addi-tionally, it is possible to get almost any shapemanufactured to order although it is inevitablyquite expensive for small quantities. As an altern-ative, some specials can be made by gluing cutpieces of standard bricks with high performanceadhesives. Some of the typical varieties are illus-trated in Figure 31.7.
Specials
301
FIGURE 31.6 Bonded wall types: (a) English bonded wall; (b) Flemish garden wall; (c) Heading bond; (d) Rat-trap bond.
31.6 Joint-styleIt is often not realized how much the joint colourand shape influence the appearance and perform-ance of masonry. Obviously the colour contrastbetween the mortar and the units must have aprofound effect on the appearance but so doesthe shape of the finished joint. The common jointstyles are shown in Figure 31.8. Recessed andweathered joints cast shadows and increase thecontrast between mortars and light-colouredbricks in most lighting conditions.
Masonry construction and forms
302
FIGURE 31.7 Standard special bricks (a) Coping(b) Queen-closer (c) Single bull-nose (d) External angle(e) Plinth.
Struck flush or wiped Bucket handle or ironed Weathered Recessed
FIGURE 31.8 Joint styles.
31.7 Workmanship and accuracyStandards of good workmanship in terms of howto lay out work and avoid weather problems byprotection of new work against rain, wind andfrost are covered in BS 5628 (1985). Realistic tol-erances for position on plan, straightness, level,height and plumb are given in BS 8000: Part 3(1985) based on BS 5606 (1999); principles andtighter tolerances are specified in SP56 (1980) forstructural quality masonry. The draft of EN1996–2 contains similar information.
31.8 Buildability, site efficiency andproductivityThe process of constructing masonry hastraditionally been regarded as difficult to mecha-nize and ‘bring fully up to date’. This is partlybecause it is a skilled craft and often has to beadapted to compensate for inaccuracies in othercomponents. There is also a tendency to a highwastage rate, because of the use of mortar whichhas a finite life and because of poor handling andstorage conditions on site and losses betweenstores and the workpoint. Many of these prob-lems have been improved by innovations such asshrink wrapping of materials, crane delivery ofpacks direct to the work points and the use ofretarded ready-to-use mortar with a long shelflife.
Despite many attempts it has not yet been pos-sible to economically automate the site construc-tion of masonry. Bricklaying machines have been
developed but only suitable for building the sim-plest of walls. For a brief period in the 1960/70sprefabricated brickwork cladding panels becamepopular in the USA and some degree of mecha-nized construction was feasible for relativelysimple factory-built panels. Recent continentalinnovations are the use of very large masonryunits, with cranes to lift them if necessary, andthe use of thin joint mortar to bond them. Thelarger size and simple technique necessary canimprove productivity and speed to similar levelsto that of timber frame for building the structuralcore of small–medium buildings. They can thenbe clad with conventional brickwork or otherfinishes. Since the cladding is off the critical path,it is possible to simultaneously finish the outsideand the inside in the same way as framed build-ings.
31.9 AppearanceThis is very much a matter of taste and expecta-tion but there are some general rules to follow.Precisely shaped bricks with sharp arrises demandaccurate layout with perpends lined up verticallyand evenly sized mortar joints throughout, other-wise they tend to look untidy. Less accurate,uneven bricks will tolerate some variation in jointsize and position without looking ugly. Exceptwith very accurate bricks, walls can only be madefair faced on one side while the other side has tosuffer from any variability in thickness of theunits. If a solid, 220mm thick wall is required to
have two fair faces it should be built as a collarjointed wall. If an internal half-brick wall isexposed on both faces it is probably best to use ajoint which is tolerant of some inaccuracy, suchas a recessed joint.
Exposed external walls should be protected asmuch as possible from run-off of rain. Any detailwhich causes large amounts of rain water tocourse down the wall in one spot or leach outthrough mortar joints at damp-proof membraneswill eventually lead to discoloration due to stain-ing by lichen, lime or efflorescence.
31.10 ReferencesBS 5628: Part 3: (1985) British Standard Code of Prac-
tice for use of masonry: Part 3: Materials, com-ponents design and workmanship. BSI, London.(New issue due in 2001).
BS 8000: Part 3: (1989) British Standard Code of Prac-tice for Workmanship on Building Sites: Part 3:Masonry. BSI, London (New issue due in 2001).
BS 4729: (1990) Specification for dimensions of bricksof special shapes and sizes. BSI, London (Newversion due 2001).
BS 5606: (1990) British Standard Guide to accuracy inbuilding. BSI, London.
Draft EN 1996–2 Eurocode 6 Design of MasonryStructures: Part 2 Design, selection of materials andexecution.
Handisyde, C.C. and Haseltine, B.A. (1980) Bricks andBrickwork, Brick Development Association.Windsor.
SP 56 (1980) Model Specification for Clay and CalciumSilicate Structural Brickwork, British CeramicResearch Establishment, Stoke on Trent.
References
303
304
Chapter thirty-two
Structuralbehaviour andmovement ofmasonry
32.1 Introduction32.2 Compressive loading32.3 Shear load in the bed plane32.4 Flexure (bending)32.5 Tension32.6 Elastic modulus32.7 Movement and creep of masonry materials32.8 References
32.1 IntroductionLike any structural material, masonry must resistloads or forces due to a variety of external influ-ences (actions) and in various planes. Figure 32.1from BRE Digest 246 (1981) illustrates thevarious forces that can arise and the likelyactions. Like plain concrete, unreinforcedmasonry is good at resisting compression forces,moderate to bad at resisting shear but very poorwhen subjected to direct tension. Masonry struc-tures which are required to resist significanttensile forces should be reinforced by adding steelor other tension components. Unlike concrete,however, masonry is highly anisotropic becauseof its layer structure and this must always beborne in mind in design.
Masonry is quite effective at resisting bendingforces when spanning horizontally between verti-cal supports, but it is somewhat less effective at
resisting bending forces when spanning verticallyor cantilevering from a support (Figure 32.2)because the resistance of a lightly loaded wall inthat direction is dependent solely on the mass andthe adhesion of the units to the mortar. Much ofthe resistance to bending and collapse, especiallyof simple cantilever masonry structures, is simplydue to self-weight. Masonry is a heavy material,
Tensile
Lateral(Flexural)
In-planeshear
Out-of-planeshear
Tensile
Compressive
FIGURE 32.1 Forces on walls.
usually with a density in the range 500–2500kg/m3, i.e. between 0.5 and 2.5 tonnes percubic metre. In relatively squat structures such assome chimneys, parapets or low or thick bound-ary walls, the force needed to rotate the wall to apoint of instability is sufficient to resist normalwind forces and minor impacts. Any masonryunder compressive stress also resists bendingsince the compressive prestress in the wall mustbe overcome before any tensile strain can occur.
There is much empirical knowledge about howmasonry works and many small structures arestill designed using experience-based rules. TheLimit State Code of Practice in the UK (BS 5628:Part 1: 1992) gives a calculated design procedurebut much of this code is based on empirical datasuch as given by Davey (1952), Simms (1965),West et al. (1977, 1986), de Vekey et al. (1986,1990). Broadly, it predicts the characteristicstrength of masonry elements, such as walls, fromdata on the characteristic strength or othercharacteristics of the materials using various
engineering models for the different loading con-ditions. A check is then made that the predictedstrength is greater than the expected worseloading derived from data about wind, deadloads, snow loads and occupancy loads. To allowfor statistical uncertainty in loading data, a safetyfactor f is applied, and to allow for uncertaintyabout the strength of the masonry a further factorm is used. The combination of these partialfactors of safety (FOS) gives an overall safetyfactor against failure of the structure which isusually in the range 3.5–5. This relatively highFOS is used because of the high variability ofmasonry properties and the brittle failure modewhich gives very little warning of failure.
32.2 Compressive loadingMasonry is most effective when resisting axialcompressive loads normal to the bedding plane(Figure 32.1). This is, clearly, the way in whichmost load-bearing walls function but also the way
Compressive loading
305
FIGURE 32.2 Mechanisms for resisting bending forces in masonry cantilevers.
that arches and tunnels resist load since an inwardforce on the outside surface of a curved plate struc-ture or tube will tend to put the material intocompression, as in Figure 32.3. If a load or force isput on a wall at a point, it would logically spreadoutward from the point of application in astretcher bonded wall since each unit is supportedby the two units below it. This mechanism, showndiagrammatically in Figure 32.4 by representing themagnitude of the force by the width of the arrows,leads to some stress being spread at 45° in a half-bond wall but the stress still remains higher nearthe axis of the load for 2m or more. Such a com-pressive force causes elastic shortening (strain) ofthe masonry. As a result of Poisson’s ratio effects, atension strain and hence a stress is generatednormal to the applied stress. In bonded masonrythe overlapping of the units restrains the propaga-tion of cracks which are generated in the verticaljoints by the tension until the stress exceeds thetensile resistance of the units.
The compressive strength of masonry is meas-
Structural behaviour and movement of masonry
306
FIGURE 32.3 Forces on masonry cylinders, e.g.arches, tunnels and sewers.
FIGURE 32.4 Load spreading in stretcher bondedwalls.
ured by subjecting small walls or prisms or largerwalls of storey height (2–3m) to a force in acompression test machine. Loading is usuallyaxial but may be made eccentric by offsetting thewall and loading only part of the thickness.
Masonry is not so good at resisting compres-sion forces parallel to the bedding plane becausethere is no overlapping and the bed joints faileasily under the resultant tensile forces. Addition-ally, most bricks with either frogs, perforations orslots are weaker when loaded on end than on bedbecause the area of material resisting the load isreduced and the stress distribution is distorted bythe perforations. Data on bricks are given inTable 30.6. The equivalent data in Table 32.1 forprisms show that the strength does vary withloading direction, although not to the sameextent as for units because the aspect ratios of themasonry specimens are all similar.
The axial strength of masonry might beexpected to depend on the strength of the unitsand of the mortar and, to a first approximation,the contribution of each to the overall strengthshould be related in some way to the volume pro-portion of each. This gives a reasonable modelfor behaviour of squat structures. A complicationis that most masonry comprises units that aremuch stronger than the mortar, and the three-
dimensional confining restraint increases theeffective strength of the thin mortar beds. Figure32.5, based on data for wallettes tested incompression in de Vekey et al. (1990) and Edgellet al. (1990), shows the minor effect of mortarstrength (i) and (iii) on the compressive strengthof masonry made with a range of strengths ofbricks but the considerable influence of the brick
strength. The diagram shows a linear relationshipbetween brick strength and brickwork strengthfor two strengths of mortar, one 4� the strengthof the other. The third line shows the differentbehaviour of blocks tested to BS 6073. Platenrestraint in the standard test method, whichinhibits tension failure in the zones shown shadedin Figure 32.6, enhances the measured strength of
Compressive loading
307
70
00
Unit strength (MPa)
Cor
resp
ondi
ng m
ason
ry s
tren
gth
MPa 60
50
40
30
20
10
20 40 60 80 100 120 140 160
Bricks laid in Designation (i) mortar
Bricks laid in Designation (iii) mortar
Blocks laid in a range of mortars
Mortar strengthcirca 12 mpa
Mortar strengthcirca 3 mpa
FIGURE 32.5 Influence of mortarand units on masonry strength.
TABLE 32.1 Masonry prism strengths with bricks loaded in various directions
Type Ref.Prism compressive
Proportion of the on-Mortar mix
strength on bedbed strength (%)
(C:L:S) (MPa) (MPa) On bed On edge On end
23 hole (a) 1: ��:3 23.1 22.4 100 71 4814 hole (b) 1: ��:3 26.6 28.9 100 29 5114 hole (c) 1: ��:3 23.1 19.9 100 58 5710 hole (d) 1: ��:4 �� 8.3 15.2 100 104 11110 hole (d) 1: ��:3 26.6 22 100 68 913 hole (e) 1: ��:3 26.6 37.6 100 81 583 hole (e) 1: ��:4 �� 7.8 20.3 100 96 833 hole (f) 1: ��:3 23.1 23.2 100 61 295 cross slots (g) 1: ��:3 26.6 34.1 100 85 415 cross slots (g) 1: ��:4 �� 7.7 18 100 132 5416 hole (A) 1: ��:3 – 26 100 20 29Frog (B) 1: ��:3 – 9.7 100 55 55Frog (C) 1: ��:3 – 10.8 100 133 122Frog (D) 1: ��:3 – 19.2 100 93 86Solid (E) 1: ��:3 – 16 100 73 64
Note: The reference relates the data in this table to those in Table 30.6.
squat units (units wider than they are high) asopposed to slender units. This anomaly is takenaccount of in BS 5628 by using different tablesfor differently shaped units. In BS DD ENV1996–1–1 the measured strength of units is cor-rected to a common value for a unit with asquare cross-section using a conversion table. Theprocess, called normalization, also corrects to astandard air-dry state whatever the test conditionis.
To calculate masonry strength, BS DD ENV1996–1–1 uses an equation of the form:
fk �K.fba.fm
b (32.1)
Where fk, fb and fm are the strength of themasonry, the normalized strength of the units and
Structural behaviour and movement of masonry
308
the strength of the mortar respectively, K is afactor which may vary to take into account theshape or type of the units, a is a fractional powerof the order of 0.7–0.8, and b is a fractionalpower of the order 0.2–0.3.
32.2.1 Stability: slender structures andeccentricity
If a structure in the form of a wall or column issquat, so that the ratio of height to thickness(slenderness ratio) is small, then the strength willdepend largely on the strength of the constituentmaterials. In real structures the material will bestiffer on one side than the other, the load willnot be central and other out-of-plane forces mayoccur. This means that if the slenderness ratio isincreased, at some point the failure mechanismwill become one of instability and the structurewill buckle and not crush.
Loads on walls, typically from floors and roofs,are commonly from one side and thus stress thewall eccentrically. Figure 32.7 illustrates, in anexaggerated form, the effect of an eccentric loadin reducing the effective cross-section bearing theload and putting part of the wall into tension.
This is recognized in practice and usually aformula is used to reduce the design load capacityof structures in which the reduction is a functionof the slenderness ratio and the net eccentricity ofall the applied loads. BS 5628: 1992 gives a stan-dard formula, where � is the reduction factor, em
is the design eccentricity and t is the thickness ofthe element as follows:
� �1.1(1� (2em /t)) (32.2)
32.2.2 Concentrated load
Many loads are fairly uniform in nature, beingderived from the weight of the structure above ormore locally to floors. There will always be,however, some point loads, termed ‘concentratedloads’, in structures at the ends of beams, lintels,arches, etc. In general the masonry must becapable of withstanding the local stresses result-ing from the concentrated loads but the designer
Test machine platen
Zones of the specimeninfluenced by platenrestraint
Slender unit
Squat unit
Test machine platen
Test machine platen
Test machine platen
FIGURE 32.6 Platen restraint (shown as shadedareas) which enhances the measured strength of squatunits more than slender units.
may assume that the load will spread out in themanner of Figure 32.4 so that it is only critical inthe first metre or so below the load. Additionally,because the area loaded is restrained by adjacentunloaded areas, some local enhancement can beassumed.
Figure 32.8 (a) shows the condition for a smallisolated load applied via a pad where there isrestraint from four sides, and Figures 32.8 (b), (c)and (d) show further conditions with reducingrestraint. The local load capacity of the masonryin the patches compared to the average loadcapacity (the enhancement factor) can vary from0.8 for case (d) to as much as 4 for case (a) (Aliand Page (1986), Arora (1988) and Malek andHendry (1988)). The enhancement factordecreases as the ratio of the area of the load to
Compressive loading
309
Joints maygo into tension
Section of wallbearing load
Load
FIGURE 32.7 Effect of eccentric load on masonry(exaggerated).
FIGURE 32.8 Concentrated loads on masonry(restraint indicated by arrows).
the area of the wall increases, as the load movesnearer the end of a wall and as the load becomesmore eccentric. Formulae describing this behavi-our are proposed in many of the references inSection 32.8.
32.2.3 Cavity walls in compression
If a compression load is shared equally by thetwo leaves of a cavity wall then the combinedresistance is the sum of the two resistances, pro-vided their elastic moduli are approximately thesame. If one leaf is very much less stiff than theother, the stress is all likely to predominate in thestiffer wall and be applied eccentrically, and thenthe combined wall may have less capacity thanthe single stiff wall loaded axially. It is commonpractice to put all the load on the inner leaf anduse the outer leaf as a rain screen. In this loadcondition more stress is allowed on the loadedwall because it is propped by the outer leaf. Thisis achieved mathematically by increasing theeffective thickness of the loaded wall used forslenderness calculations.
32.3 Shear load in the bed planeIf a wall is loaded by out-of-plane forces, e.g. bywind, impact, or seismic (earthquake) actions, theforce will act to try to slide the wall sideways(like a piston in a tube). In practice the action canbe at any angle in the 360° plane although it ismost common parallel or normal to the wall face.This is a very complex loading condition and theresult is rarely a pure shearing failure. For smalltest pieces measured in an idealized test, the shearstrength tv can be shown to follow a friction lawwith a static threshold ‘adhesion’ fv0 and adynamic friction term K dependent on the forcenormal to the shearing plane sa. The formula issimply:
fv � fv0 � sa.K (32.3)
Measurement of pure shear is very difficultbecause of the tendency to induce rotation in vir-tually any physical test arrangement. The simpledouble shear test of the type shown in Figure32.9 is suitable for measuring shear resistance ofdamp-proof course (dpc) materials as per EN1052–4 (1999) but is unsatisfactory for mortar
joints where a much shorter specimen is preferredof the type described in EN 1052–3 (1999). Table32.2 gives some typical shear data derived fromHodgkinson and West (1981).
In the example sketched in Figure 32.10 theends are supported so the wall tends to adopt acurved shape. In this case it is shown as failing byshear at the line of the dpc. If a wall is loaded bylateral forces acting on the end, as in Figure32.11 (which can arise from wind load on a flankwall at right angles), the force will initially tendto distort the wall to a parallelogram shape as inFigure 32.11(a). If the wall fails it may be bycrushing at the load point or at the ‘toe’ of thewall as in Figure 32.11(a), by sliding shear failurealong the dpc, by ‘shear’ failure, i.e. actual diago-nal tensile failure, or by rotation as in Figure32.11(b).
A further important shear condition occursbetween masonry elements bonded together, i.e.where piers or buttressing walls join other walls,as in Figure 32.11(a), or in box structures at thecorners. If the composite is loaded to generateflexure there is a vertical shear force at the inter-face which is resisted by the punching shear resis-
Structural behaviour and movement of masonry
310
TABLE 32.2 Shear data for two mortars and some damp proof course (dpc) materials (Hodgkinson and West,1981)
Brick dpc 1: �� :3 Mortar 1:1:6 Mortar
sa K sa K
16 hole w/c Blue brick1 0.72 0.82 0.44 1.14Bituminous 0.4 0.8 0.31 0.91Permagrip2 0.8 0.58 0.43 0.61Hyload3 0.21 0.8 0.17 0.58Vulcathene4 0.06 0.47 0.04 0.54
Frogged semi-dry pressed Blue brick1 0.73 0.95 0.36 1.57Bituminous 0.45 0.84 0.36 0.72Permagrip2 0.55 0.93 0.41 0.72Hyload3 0.17 0.82 0.18 0.79Vulcathene4 0.09 0.52 0.06 0.49
Notes:1. Where blue engineering bricks are used for the dpc the shear strength is limited effectively by the mortar.2. Permagrip is a trade name for a reinforced bitumen product with a coarse sand surface.3. Hyload is probably a pitch/polymer system.4. Vulcathene is a trade name for a polyethylene-based dpc.
tance of the units crossing the junction. In stan-dard bond, units cross the shear plane for 50 percent of the height of the structure and the shearstrength will be, at best, the sum of these resis-tances plus a small contribution from the mortar.
Flexure (bending)
311
FIGURE 32.9 Specimen for measuring shear resis-tance of dpc materials, etc.
FIGURE 32.10 Shear failure on dpc line of laterallyloaded wall.
FIGURE 32.11 Stresses and strains resulting froman in-plane shear load. The wall may fail by crushingin the zones of higher compressive stress or by shear atthe junction of the two walls at right angles, by hori-zontal shear at the dpc line, by tension splitting or byrotation.
32.4 Flexure (bending)Traditionally, masonry was made massive ormade into shapes which resisted compressionforces. Such structures do not depend to anygreat extent on the bond of mortar to units.Much of the masonry built in the last few decadeshas, however, been in the form of thin walls forwhich the critical load condition can result fromlateral forces, e.g. wind loads. This phenomenonwas largely made possible by the use of opcmortars which give a positive bond to most unitsand allow the masonry to behave as an elasticplate. There are two distinct principal modes offlexure about the two main orthogonal planes: (1)the vertically spanning direction shown in Figure32.2 which is commonly termed the parallel (orp) direction because the stress is applied aboutthe plane parallel to the bed joints; and (2) thehorizontally spanning direction, shown in Figure32.12, which is commonly termed the normal (or
n) direction because the stress is applied aboutthe plane normal to the bed joints. Clearly thestrength is likely to be highly anisotropic since thestress in the parallel direction is only resisted by
Structural behaviour and movement of masonry
312
Tensile fracture throughperpend and unit
Position of loadingand reaction rodsin a wallette test
Shear of mortarjoints
FIGURE 32.12 Modes of failure for laterally loadedmasonry in the strong (n) direction.
TABLE 32.3 Flexural strength ranges (MPa) for common masonry units bedded in designation (iii) mortar
Material Normal (strong) direction Parallel (weak) direction
Clay brick (0–7% water absorption) 1.8 to 4.7 0.35 to 1.1Clay brick (7–12% water absorption) 1.9 to 3.2 0.3 to 1.3Clay brick (�12% water absorption) 1.0 to 2.1 0.3 to 0.8Concrete brick (25–40MPa strength) 1.9 to 2.4 0.5 to 0.9Calcil brick (25–40MPa strength) 0.7 to 1.5 0.05 to 0.4AAC block (100mm thick) 0.3 to 0.7 0.3 to 0.6LWAC block (100mm thick) 0.7 to 1.3 0.3 to 0.5DAC block (100mm thick) 0.7 to 1.7 0.2 to 0.7
Note: The ranges given are for designation (iii) mortars (1:1:6 or equivalent). There is some dependence of flexural strength onmortar strength and thicker blocks lead to reduced strengths.LWAC� lightweight aggregate concrete and DAC�dense aggregate concrete.
the adhesion of the units to the mortar while thestress in the normal direction is resisted (a) by theshear resistance of the mortar beds, (b) by theelastic resistance of the head joints to the rotationof the units, (c) by the adhesion of the head jointsand (d) by the flexural resistance of the unitsthemselves. Generally the limiting flexural resis-tance will be the lesser of (a)� (b) or (c)� (d)giving two main modes of horizontal spanningfailure – shearing, shown in the lower part ofFigure 32.12 and snapping, shown in the upperpart of Figure 32.12.
Using small walls (wallettes), either as shownin Figure 32.12 for measuring horizontalbending, or ten courses high by two bricks widefor vertical bending, and tested in four-pointbending mode, the flexural strength of a largerange of combinations of UK units and mortarshas been measured for the two orthogonal direc-tions and a typical range is given in Table 32.3;further data are given in West et al. (1986), deVekey et al. (1986) and de Vekey et al. (1990).
The ratio of the strength in the two directions,expressed as p-direction divided by n-direction, istermed the orthogonal ratio and given the symbol�. In cases where only the bond strength (p-direction) is required, a simpler and cheapertest is the bond wrench discussed in Section23.3.2.
The flexural strengths given in Table 32.3 arefor simply supported pieces of masonry spanningin one direction. If the fixity of the masonry at the
supports (the resistance to rotation) is increased,the load resistance will increase in accordance withnormal structural principles (Figure 32.13). Again,if one area of masonry spans in two directions theresistances in the two directions are additive.Haseltine et al. (1977), Sinha et al. (1978, 1997),Seward (1982) and Lovegrove (1988) cover someaspects of the resistance of panels.
32.5 TensionMasonry made with conventional mortars has avery limited resistance to pure tension forces and,
for the purposes of design, the tensile strength isusually taken to be zero. In practice it does havesome resistance in the horizontal direction andsomewhat less in the vertical direction. In anattempt to make a viable prefabricated masonrypanel product for use as a cladding material,polymer latex additives can be used to improvethe tensile strength. Panels of storey height and ametre or more wide have been manufactured andcould be lifted and transported without failure.
Horizontal tensile strength of masonry wasmeasured but little other data exists, and no stan-dard test or any significant data are available.
Tension
313
Applied force (e.g. wind load)
Reaction
Wall with simple edge support
Reaction
Reaction
Wall with fixed edge support
Reaction
Fixed support encastré
FIGURE 32.13 Effect of edge support conditions on the flexural behaviour of masonry.
Tensile bond strength is often measured, the mostcommonly reported test being the ASTM crossbrick test (ASTM C952–76), but this has nowbeen discontinued and simple two-brick prismsillustrated in Figure 32.14 are now favoured.Data from such tests indicate that the directtensile strength across the bond is between one-third and two-thirds of the parallel flexuralstrength (see Table 32.3). Other tests have beendeveloped along similar lines including one inwhich one unit is held in a centrifuge and thebond to another unit is stressed by centrifuging.A useful review is given by Jukes and Riddington(1998).
Structural behaviour and movement of masonry
314
Articulatedplate
Adhesive
Brick
Mortar joint
Brick
Adhesive
Articulatedplate
FIGURE 32.14 Stack-bond couplet tensile test.
32.6 Elastic modulusThe stiffness or elastic modulus of masonry is animportant parameter required for calculations ofstresses resulting from strains arising from loads,concentrated loads, constrained movement andalso for calculating the required area of reinforc-ing and post-stressing bars.
The most commonly measured value is theYoung’s modulus (E) but the Poisson’s ratio (v) isalso required for theoretical calculations usingtechniques such as finite element analysis. Ifrequired the bulk (K) and shear (G) moduli maybe derived from the other parameters. Young’smodulus is normally measured in a compressiontest by simultaneously measuring strain (�p) paral-lel to the applied stress (�) whereupon:
E� �/�p (32.4)
If the strain (�N) perpendicular to the appliedstress is also measured, Poisson’s ratio (�) may bederived:
� � �N/�p (32.5)
Masonry is not an ideally elastic material becauseit is full of minor imperfections such as micro-cracks in the bond layers and because the differ-ences between the unit and mortar stiffnesses andPoisson’s ratios produce high local strains at theinterface which result in non-linear behaviour.This means that the stress–strain curve is typicallyof a broadly parabolic form with an early elasticregion. The instantaneous Young’s modulus isderived from the tangent to the curve at any pointbut for some calculations, such as creep loss ofpost-stressed masonry, the effective Young’smodulus required is derived from the secantvalue. Figure 32.15 illustrates this behaviour.Data on elastic properties in compression aregiven in Davies and Hodgkinson (1988). Elasticmodulus is also important in estimating thedeflections of walls out of plane due to lateralloads such as wind. In this case the modulus canbe measured by using load deflection data forsmall walls tested in four-point bending. In thiscase E is derived from:
E�8Wa(3L2 �4a2)/384I� (32.6)
where W is the applied force, L is the span of thesupports, a is the distance from the supports tothe loading points, I is the moment of inertia and� is the deflection.
In compression tests the value of E has gener-ally been found to be in the range 500–1000times the compressive strength. For typical mater-ials this is likely to be around 2 to 30GPa
In flexure the early tangent modulus has beenfound to be in the range 2–4GPa for tests in thestrong (normal) direction and 1–2GPa for
Movement and creep of masonry materials
315
FIGURE 32.15 Stress–strain behaviour of masonry.
TABLE 32.4 Movements and elastic modulus of masonry materials
Masonry component Thermal expansion Reversible moisture Irreversible moisture Modulus ofmaterial coefficient a movement movement elasticity E
(per °C�10�6) (± %) (%) (GPa)
Granite 8 to 10 – – 20 to 60Limestone 3 to 4 0.01 – 10 to 80Marble 4 to 6 – – 35Sandstone 7 to 12 0.07 – 3 to 80Slate 9 to 11 – – 10 to 35Mortar 10 to 13 0.02 to 0.06 �0.04 to �0.1 20 to 35Dense concrete brick/blockwork 6 to 12 0.02 to 0.04 �0.02 to �0.06 10 to 25LWAC blockwork 8 to 12 0.03 to 0.06 �0.02 to �0.06 4 to 16AAC blockwork 8 0.02 to 0.03 �0.05 to �0.09 3 to 8Calcil brickwork 8 to 14 0.01 to 0.05 �0.01 to �0.04 14 to 18Clay brickwork 5 to 8 0.02 �0.02 to �0.10 4 to 26
equivalent tests in the weak (parallel) direction.Some more data are given in Table 32.4.
32.7 Movement and creep ofmasonry materialsUnrestrained masonry is subject to cyclic move-ment due to moisture and temperature changes,permanent creep due to dead loads or post stressloads, permanent shrinkage due to drying/carbon-ation of mortar, concrete and calcium silicatematerials and expansion due to adsorption ofmoisture by fired clay materials. Table 32.4 con-tains some typical ranges for common masonrycomponents derived mainly from Digest 228(1979).
In simple load-bearing masonry elements, verti-cal movements are accommodated by the struc-ture going up and down as required and are noproblem. Problems can arise, however, wherematerials with different movement characteristicsare joined or where thick elements have differen-tial temperature/moisture gradients throughthem. Elements with a vertical prestress muchlarger than the differential stresses will toleratethe movement, but lightly stressed and unre-strained elements will need a slip plane or softjoint between the elements to avoid problems. Aclassic case is (expanding) clay masonry on
(shrinking) concrete frames where soft joints areused to stop stress transfer to the masonry.
Where restraint is present, horizontal shrink-ages will be converted into tensile forces andexpansions into compressive forces. Since wallsare probably two orders of magnitude weaker intension than in compression the result is thatrestrained walls in tension usually crack whilethose in compression usually just build up stress.Where walls are unrestrained the reverse isusually the case: the shrinking wall simply con-tracts but the expanding wall interacts with flankwalls (those at right angles) and causes crackingby generating rotation at the corner.
Because there is nearly always differentialrestraint the pure tensile or compressive forceswill usually give rise to some associated shearforces.
32.8 ReferencesAli, S. and Page, A.W. (1986) An elastic analysis of
concentrated loads on brickwork. InternationalJournal of Masonry Construction, 6, Edinburgh,9–21.
Arora, S.K. (1988) Review of BRE research intoperformance of masonry walls under concentratedload. Proceedings of the 8th InternationalBrick/Block Masonry Conference, Dublin, 446.
Davey, N. (1952) Modern research on loadbearingbrickwork. The Brick Bulletin 1–16.
Davies, S. and Hodgkinson, H.R. (1988) TheStress–Strain Relationships of Brickwork, Part 2,BCRL research paper 7S5.
Haseltine, B.A., West, H.W.H. and Tutt, J.N. (1977)Design of walls to resist lateral loads. The StructuralEngineer 55, 10, 422–30.
Hodgkinson, H.R. and West, H.W.H. (1981) TheShear Resistance of Some Damp-proof CourseMaterials, Technical Note 326, British CeramicResearch Ltd, Stoke on Trent.
Jukes, P. and Riddington, J.R. (1998) A review ofmasonry tensile bond strength test methods.Masonry International 12, 2, 51–7.
Lovegrove, R. (1988) The effect of thickness and bondpattern upon the lateral strength of brickwork. Proc.Brit. Masonry Soc. 2, 95–7.
Malek, M.H. and Hendry, A.W. (1988) Compressivestrength of brickwork masonry under concentratedloading. Proc. Brit. Masonry Soc., 2, 56–60.
Seward, D.W. (1982) A developed elastic analysis oflightly loaded brickwork walls with lateral loading.International Journal of Masonry Construction 2, 2,129–34.
Simms, L.G. (1965) The strength of walls built in thelaboratory with some types of clay bricks andblocks. Proc. Brit. Ceram. Soc. July, 81–92.
Sinha, B.P. (1978) A simplified ultimate load analysisof laterally loaded model orthotropic brickworkpanels of low tensile strength. The Structural Engi-neer 56B, 4, 81–4.
Sinha, B.P., Ng, C.L. and Pedreschi, R.F. (1997)Failure criteria and behaviour of brickwork inbiaxial bending. J. of Materials in Civil Engg, 9, 2,61–6.
de Vekey, R.C., Bright, N.J., Luckin, K.R. and Arora,S.K. (1986) Research results on autoclaved aeratedconcrete blockwork. The Structural Engineer 64A,11, 332–41.
de Vekey, R.C., Edgell, G.J. and Dukes, R. (1990) Theeffect of sand grading on the performance and pro-perties of masonry. Proc. Brit. Masonry Soc. 4,152–9.
West, H.W.H., Hodgkinson, H.R. and Davenport,S.T.E. (1968) The Performance of Walls Built ofWirecut Bricks With and Without Perforations,Special Publication No. 60, British CeramicResearch Ltd, Stoke on Trent.
West, H.W.H., Hodgkinson, H.R. and Haseltine, B.A.(1977) The resistance of brickwork to lateralloading. The Structural Engineer 55, 10, 411–21.
West, H.W.H., Hodgkinson, H.R., Haseltine, B.A. andde Vekey, R.C. (1986) Research results on brick-work and aggregate blockwork since 1977. TheStructural Engineer 64A, 11, 320–31.
Structural behaviour and movement of masonry
316
Chapter thirty-three
Durability andnon-structuralproperties ofmasonry
33.1 Introduction33.2 Durability33.3 Chemical attack33.4 Erosion33.5 Staining33.6 Thermal conductivity33.7 Rain resistance33.8 Sound transmission33.9 Fire resistance33.10 Sustainability issues33.11 References
33.1 IntroductionThe main non-structural function of masonry ele-ments is as a cladding to buildings and the keyfunction of such elements is to maintain habitableconditions within the building. It is thus para-mount that the masonry be durable with respect tothe external actions it is subjected to. It is alsoimportant to know how effective masonry is incontrolling the temperature, preventing ingress ofwind and rain, reducing noise transmission andlimiting the spread of fire should it break out.Another key concern that is becoming increasinglyimportant is the effect of a product on the environ-ment – the sustainability or ‘green’ credentials of aproduct. Increasingly, materials will be chosen notjust for their inherent performance, appearance or
economy but their energy cost and other effects onthe global environment such as emission of green-house gases will be taken into account.
33.2 DurabilityThere are a large number of mechanisms bywhich masonry structures can deteriorate whichcan be categorized into:
1. chemical/biological attack on either themortar or the units or both, due to water andwaterborne acids, sulfates, pollution andchemicals released by growing plants;
2. corrosion of embedded metal (usually steel)components, particularly ties, straps, reinforc-ing rods, hangers, etc., which is a special caseof chemical attack;
3. erosion of units or mortar by particles inflowing water and wind, by frost attack andby salt crystallization;
4. stress-related effects due to movement offoundations, movement/consolidation/washoutof in-fill, vibration, overloading, moisturemovement of bricks and blocks, thermalmovement, growth of plants;
5. staining due to efflorescence, lime, iron, vana-dium and biological growth.
317
Mortar is generally a less durable material thanfired clay or dense concrete, because it containsreactive and finely divided binders such as opcand it usually has a relatively high porosity and,comparatively, a lower hardness and abrasionresistance. The durability of natural stone is veryvariable, ranging from the highest given by denseimpermeable granites and marbles to quite poorperformance of porous limestones and sand-stones. Erosion processes such as wind and waterscour attack both units and mortar but erode thesofter of the two at a faster rate. Frost and saltcrystallization are complex processes where thesusceptibility is dependent as much on pore distri-bution as hardness and overall porosity. Stresseffects normally cause cracking of varying typesbut the effect is on the masonry composite andnot on the individual components. There aresome problems related to faults in manufacture ofthe units, particularly under- or over-firing of claybricks and the inclusion of foreign particles inbricks or other types of unit. A good range ofcoloured illustrations of problems is given in BREDigest 361 (1991) and in BRE Digest 441 (1999).
33.3 Chemical attack
33.3.1 Water and acid rain
Water percolating into masonry is always apotential source of damage and, where possible,the structure should be designed to channel wateraway or at least to allow it to escape via weep-holes. Absolutely pure water will have no directchemical effect but some of the constituents ofmortar are very slightly soluble and will leachaway very slowly. Rain water contains dissolvedcarbon dioxide which forms a very mild acid thatdissolves calcium carbonate by formation of thesoluble bicarbonate via the reaction:
CO2�H2O→H2CO3�CaCO3 →Ca(HCO3)2 (33.1)
This means that lime mortars, weak opc : limemortars, porous limestones, porous limebondedsandstones and porous concrete blocks madewith limestone aggregate will eventually bedestroyed by percolating rain water because
calcium carbonate is a key constituent. Strongopc mortars with well-graded sand and most con-crete and sandlime units are less susceptible,partly because the calcium silicate binder is lesssoluble but mainly because they are less perme-able and so prevent free percolation. Typicalvisible effects of water leaching on mortar areloose sandy or friable joints, loss of mortar in theoutside of the joints giving a raked joint appear-ance, and in serious cases the loss of bricks fromthe outer layer of masonry particularly fromtunnel/arch heads. The process will sometimes beaccompanied by staining due to reprecipitation ofthe dissolved materials. Stones lose their surfacefinish and may develop pits or rounded arrises.
Sulphur dioxide reacts with water to form ini-tially sulphurous acid but can oxidize further inair to sulphuric acid:
2SO2 �2H2O→2H2SO3 �O2 →2H2SO4 (33.2)
There is no systematic evidence that rain acidifiedby sulphur dioxide from flue gases at the normallevels has a significant effect on mortar, but inspecial cases such as where there is a high level ofgaseous pollution near industrial sites there maybe a very significant increase in the deteriorationrate of all forms of structure including mortarand limestones. Unlined chimneys are a particularcase and can suffer severe sulfate attack in theexposed parts where rain saturation or condensa-tion of fumes containing sulphur occurs. Onemechanism of failure is the conversion of calciumcarbonate to gypsum:
CaCO3 �H2SO4 →CaSO4 �H2CO3 (33.3)
33.3.2 Carbonation
Gaseous carbon dioxide (CO2) at humiditiesbetween about 30 per cent and 70 per cent neu-tralizes any alkalis present. This process occursfor all lime and opc binders with the conversionof compounds such as sodium hydroxide(NaOH) and, most commonly, calcium hydrox-ide (Ca(OH)2) to their respective carbonates. Inlime mortars this process probably increases thestrength and durability. In opc-based materials
Durability and non-structural properties of masonry
318
the key effect of the process is to reduce the pHfrom around 12–13 down to below 7, i.e. con-verting the material from highly alkaline toslightly acid. This can have a profound effect onthe durability of buried steel components (seeSection 17.4.2). There is also some evidence thatthere is a slight associated shrinkage which mayreduce the strength of very lightweight concreteunits. Very dense concrete and calcium silicateunits will carbonate slowly and may take 50–100years or more to carbonate through.
33.3.3 Sulfate attack
Sulfate attack is the next most common problemwhich is due to the reaction between sulfate ionsin water solution and the components of set Port-land cement to form Ettringite. A detailed cover-age of the physical and chemical mechanisms wasgiven in Chapter 23.
The resulting expansion, which can be of theorder of several per cent, causes both local dis-ruption of mortar beds and stresses in the brick-work but only in wet or saturated conditions andwhere there is a source of water-soluble sulfatecompound. It will never occur in dry or slightlydamp masonry. The common sulfates found inmasonry are the freely soluble sodium, potassiumand magnesium salts and calcium sulfate which isless soluble but will leach in persistently wet con-ditions. The sulfates may be present in ground-waters and can affect masonry below the dpc,and masonry in contact with the ground such asretaining walls, bridges and tunnels. In this situ-ation porous concrete units are also affected – seeBRE Digest 363 (1996). Sulfates are also presentin some types of clay brick and will be trans-ported to the mortar in wet conditions. Old typesof solid brick with unoxidized centres (‘black-hearts’) often have large amounts of soluble sul-fates, as do some Scottish composition bricks,while semi-dry pressed bricks made from Oxfordclay (Flettons) have high levels of calcium sulfate.Sulfates may attack lime mortars by conversionof the lime to gypsum in a similar reaction toequation (33.3). Sulfate-resisting Portland cementis deliberately formulated to have a low C3A
content but may be attacked in very extreme con-ditions. Sulfate attack is more likely in porousmortars while rich, dense, impervious mortars areaffected less despite their higher cement content.Recent research reported by the Thaumasiteexperts group (1999) has indicated that thauma-site may also form in brickwork mortar as theproduct of the reaction between dicalcium andtricalcium silicate, sulfate, carbonate and water.This process can turn mortar to a mush.Although no known failures have occurred wheresulfate-resisting cement has been used, thiscement may not protect mortar against thauma-site formation.
Visible effects of such attack on mortar areexpansion of the masonry where it is unrestrainedand stress increases where it is restrained. Typ-ically the mortar is affected more within the bodyof the wall than on the surface, so small horizon-tal cracks are sometimes visible in the centre ofeach bed joint as in Figure 33.1. In thickmasonry, vertical cracks may appear on the exter-nal elevations due to the greater expansion in thecentre of the wall which remains wetter forlonger than the outside which can dry out byevaporation. Horizontal cracking may also occur,but it is likely to be less obvious where there is ahigh vertical deadload stress. Rendered masonryexhibits a network of cracks termed ‘map-cracking’. If walls also have a face to an internalspace which is kept dry, then efflorescence(growth of crystals) may occur due to transportof the sulfates to the surface. These faults aremost common where the water is leaking into thestructure from faulty details, leaking roofs orfrom ground contact, especially with sulfate-bearing groundwaters. The susceptibility ofmortar to sulfate attack (and frost attack or acombination of the two) can be tested using theHarrison technique (see references in Section30.3.2) which has been standardized as RILEM(1999) and CEN 1015–14 (2000).
Sulfates do not usually have a significant chemi-cal effect on properly fired clay units, althoughthey are commonly the source of such com-pounds. Sulfate attack on concrete blocks andbricks can occur but is not a common problem.
Chemical attack
319
Certain precautions are advisable when buildingin ground containing sulfates (see BRE Digest363 (1996)) or constructing flumes and tunnels tocarry contaminated water (WRC, 1986).
A special type of ‘engineering quality’ concretebrick is now available which is designed to bestable in effluents containing sulfates and otherharmful compounds. These units are manufac-tured to a high strength and low permeabilitywith srpc as the binder. A limited test has beencarried out on samples of three different ‘engi-neering quality’ concrete bricks using simulatedindustrial and commercial effluent solutions(Concrete, 1986).
Mundic concrete blocks made using tin-minetailings as aggregate in south-west England havesuffered attack from indigenous sulfates (seeSection 17.3.1, Bomley and Pettifer (1997) andRICS (1997)).
33.3.4 Acids
The effects of acids, e.g. rain run-off from peatmoors, or industrial or agricultural pollution, oncement-based products is covered in Section17.3.1. Clay products are normally resistant.
33.3.5 Chlorides
Chlorides can have a weakening effect on calcium
Durability and non-structural properties of masonry
320
FIGURE 33.1 Sulfate attack.
FIGURE 33.2 Corroded timber frame masonry walltie showing both white rust (ZnO) and red rust (FE2O3
H2O).
silicate units but have little effect on mortars, clayunits or concrete masonry units.
33.3.6 Corrosion of buried metals
Iron and steel expand on rusting, and a typicalcrack pattern from wall tie corrosion is shown inFigure 33.2. A detailed discussion of the corro-sion of steel in cement and concrete was given inChapter 24, and specific coverage of wall tie cor-rosion is given in de Vekey (1990a, 1990b and2000).
33.4 Erosion
33.4.1 Frost
Frost is the principal eroding agent of masonryexposed to normal exterior conditions. Quiteclearly it will not affect masonry buried more thana few feet and so will not affect foundations, theinsides of tunnels or buried culverts. It may affectany masonry exposed on the exterior of structuresbut is more likely to affect exposed componentsthat become saturated. Typical problem areas arecivil engineering structures such as bridges, earth-retaining walls and exposed culverts and parts ofbuildings such as parapets, chimneys, freestandingwalls and masonry between the ground and thedamp-proof course.
Frost attack is due to the stresses created by theexpansion of water on freezing to ice in the poresystems of units and mortars and thus onlyoccurs in water-saturated or near-saturatedmasonry. Some clay bricks and natural stones areparticularly susceptible, whereas most goodquality clay engineering bricks, stock bricks andconcrete and calcium silicate units are very resis-tant. Mortar is also subject to frost damage,particularly in combination with any of the otherforms of chemical deterioration.
Typical effects are the spalling of small piecesof either the unit or the mortar or both forming alayer of detritus at the foot of the wall. Figure33.3 is a case where the brick has a lower resis-tance than the mortar. Clay bricks, particularlylightly fired examples of types made from someshales and marls, are especially susceptible andtend to delaminate. Semi-dry pressed bricks madefrom the Oxford clay tend to break down intograins the same size as they were made from (e.g.around 7–14 mesh). Old types of solid clay brickwith unoxidized centres tend to spall from theboundary between the heart and the outside ofthe brick. Some natural stones with a large pro-portion of fine pores are also susceptible. Modernperforated clay bricks are generally more frost-resistant because the more open structure allowsmore even drying and firing with less chance ofshrinkage/firing cracks forming.
Erosion
321
FIGURE 33.3 Erosion due to frost attack.
The susceptibility of building materials to frostattack can be very difficult to understand. Clearlymaterials such as glass, plastic and metals whichare totally non-porous to water are not affected.Materials such as well-fired ‘glassy’ clay bricks,well-compacted concretes, low-porosity granites,marbles and slates are also affected very little byfrost, since the tiny amount of water that canpenetrate will cause only a trivial level of stresson freezing. Materials with water absorptionsranging from about 4 per cent to 50 per cent tendto suffer frost damage but not invariably. As ageneral rule, materials with a wide range of poresizes from very large to fine tend to be resistant,while those having a limited range of pore sizes,usually of the finer sizes, tend to fail. This isprobably because it is difficult to saturate fullythe mixed pore system, the water filling the finebut not the larger pores. Providing around 10 percent of the pore system is still air filled, there issufficient space for the ice crystals to expand intowithout damaging the structure.
This has given rise to tests of pore distributionbeing used to predict frost durability. For stone atest is used in which the stone is saturated withwater then pressurized to a level at which thelarger pores will be emptied. This gives a measureof the ratio between the volume of pores aboveand below a chosen critical size and some idea ofdurability.
Most of the older data on susceptibility to frostattack are based on experience but this is a veryslow and inefficient way of evaluating new prod-ucts. To try to speed up the process, acceleratedfreezing tests have been developed. A typicalexample is the panel test (West et al., 1984) whichhas latterly been published both by RILEM (1998)and as a draft CEN standard (1999). Work hasalso been done by Beardmore et al. (1987 and1990) and Stupart (1996) to evaluate the numberof days of frost cycling combined with driving rainwhich is what causes the damage.
The method is designed to test small panels ofbrickwork. The panel is first totally immersed inwater for seven days, then one face is exposed torepeated cycles of freezing and thawing usuallyup to a limit of 100 cycles. The other face and thetop and the sides of the panel are enclosed in aclose-fitting jacket of 25mm thick expanded poly-styrene. Bricks showing no signs of failure after100 cycles should be frost-resistant, those failingbetween 10 and 100 cycles would be expected tobe durable under most conditions of exposure,but some failure could occur if they were used ina situation where repeated freeze–thaw cyclingoccurred when the bricks were saturated withwater. Bricks which fail in less than 10 cycles areconsidered to be suitable for internal use only.
33.4.2 Crypto-efflorescence damage
This is basically the same process as efflorescence(see below) but at certain temperature/humidityconditions it occurs just below the surface of themasonry unit. The hydrated crystals of speciessuch as calcium and sodium sulfates growing inthe pore structure result in a tensile force on thesurface layers in a way which is analogous tofrost attack. In UK conditions it is more likely toaffect natural stones than other units. It normallyoccurs in warm conditions where there is rapiddrying of water from the face causing the salts tocrystallize out. Additionally there needs to be asource of water or water containing salts.
Typical appearance is similar to that of frostdamage but it will usually be associated withefflorescence crystals.
33.4.3 Abrasion
Abrasion by particles in wind and water probablyacts more in concert with other processes than inisolation. Likely areas for such erosion are bridgecolumns founded in river beds and buildings nearroad surfaces where splash-up can occur fromvehicles. All types of marine/hydraulic structuressuch as dams, culverts, lock walls, flumes, etc.,where high-velocity flows can occur may sufferfrom localized abrasion damage known as ‘scour’.
33.5 Staining
33.5.1 Efflorescence
This is simply a staining process caused by disso-lution of soluble salts such as sodium, potassium,magnesium and calcium sulfates by rain or con-struction water from within brickwork whichthen crystallize on the surface as a white powderor encrustation. The surface may be any surfacefrom which drying occurs. It is commonly theexternal façade but may be the interior of solidwalls particularly those in contact with earth, e.g.cellar walls and ground-floor walls in older struc-tures without damp courses. The salts commonlyderive from clay bricks, but may also come fromgroundwaters.
Lime staining, shown in Figure 33.4, is caused bycalcium hydroxide leaching from mortar and beingcarbonated at the surface to form a white deposit ofcalcium carbonate crystals. It may also result fromthe dissolution of calcium carbonate in carbonatedrainwater to form calcium bicarbonate whichreverts to the carbonate at the surface (stalagtitesgrow by this mechanism). It is most commonly seenas white ‘runs’ from mortar joints on earth-retain-ing walls but can occur on walls of buildings.
33.5.2 Iron staining
Iron compounds can be present in either bricks ormortars or both. They give little problem if welldistributed as in many red-brown sands but willgive ugly brown run-off stains if present as largerdiscreet particles. Iron may be present in the clay
Durability and non-structural properties of masonry
322
deposit in the form of pyrites and is sometimes aconstituent of additives to clay.
33.5.3 Biological staining
Coloured deposits, usually white/green tobrown/black, can build up due to the growth ofalgae fungi, lichens, mosses, etc. Such depositsonly form if the masonry is wet for significantperiods and is frequently the result of blockeddownpipes or leaking gutters. Differential deposi-tion can result from water streaming unequally offfeatures such as windows and mullions and thiscan become noticeable due to the colour change.
33.6 Thermal conductivityThe thermal conductivity determines the rate ofheat flow through a given material. Generallymetals have the higher and very porous materials(containing a lot of air or other gas) at the lowerconductivities. Masonry materials fall in a bandbetween the two extremes. The property is import-ant in that it affects the amount of heat loss froma building through the walls and thus the energyefficiency of the structure. Surprisingly, although itis often of lower density and quite porous, normalbedding mortar is frequently a poorer insulatorthan many bricks and blocks, as illustrated byFigure 33.5. This has led to the development ofinsulating mortars containing low-density aggreg-ate particles and thin-joint mortar. The latter type
Thermal conductivity
323
FIGURE 33.4 Lime staining.
reduces heat loss because of its much smaller areaproportion of the wall face. Because porosity is akey parameter, the thermal conductivity is boundto be partially related to material density andgeneral equations for dry solid porous buildingmaterials tend to be a function of density and givea regression expression of the form:
k�0.0536�0.000213� �0.0000002�2 (33.4)
The presence of moisture increases the conductiv-ity of porous materials because evaporation/condensation heat transfer mechanisms becomepossible.
Hollow and perforated products give someimprovement over plain, solid products, althoughthe potential gain from the trapped air pockets iscompromised by the convection of the air withinthem. If the air is prevented from convecting byfilling the hollows with foamed plastic materialssuch as urea–formaldehyde or polyurethane thereis a further substantial improvement. Clay bricksor blocks with many small perforations alsoperform better than solid units because thesmaller size of the holes reduces convection andthe smaller solid cross-section reduces conduc-tion. There is also an improvement if the holesare staggered such that the direct conduction paththrough the solid material is as long as possible.Figure 33.6 illustrates the effects of different per-foration patterns on thermal resistance.
FIGURE 33.5 Infra-red photograph showing heatloss through joints.
The properties of walls used as thermal barriersare normally expressed in the form of the ‘Uvalue’, the overall heat transfer coefficient whichis a synthesis of the k value of the actual mater-ials and the heat transfer coefficients at the hotand cold sides. More detail on thermal insulationis given in Diamant (1986), BRE Digest 108(1975) and BRE Digest 273 (1983).
33.7 Rain resistanceFrom the earliest use of built housing it has beenone of the primary requirements that the wallswill keep the occupants dry, and thus mostmasonry forming the perimeter walls of housesand other buildings is called upon to resist theingress of rain. All masonry component materialsare porous, however, and there are always someimperfections in the bond and seal of the mortar
joints (de Vekey et al., 1998) and the workman-ship (Newman et al., 1981 and 1982) which willadmit some water, so no solid masonry wall islikely to be absolutely watertight.
Paradoxically, it is normally easier under UKconditions to make a rain-resistant wall fromporous bricks. This is because some leakagealways occurs at the joints which is mopped upby high absorption units but allowed free passageby low absorption units. Provided the rain doesnot persist until the units are saturated, they candry out in the intervening dry spell and do notactually leak significant amounts of moisture tothe inner face. Under similar conditions, somemodern, low-absorption facings may leak quiteseriously during a moderate storm. As would beexpected, resistance is greater for thicker walls orif a coating is applied to the wall which reducesthe surface rate of absorption. Typical coatings
Durability and non-structural properties of masonry
324
FIGURE 33.6 Effect of perforations on thermal resistance.
are renders and paints or overcladding systemssuch as vertical tiling.
The commonest technique for avoiding rainpenetration in the UK is the cavity wall. This is atwin layer wall with an enclosed air spacebetween the two leaves. Some leakage throughthe outer layer (leaf) is anticipated and any suchwater is directed back out through weepholesbefore it reaches the inner leaf by use of damp-proof membranes and cavity trays. This is some-times thought of as a very recent wall form but itwas probably used in ancient Greece and hasbeen in use in the damper parts of the UK fornearly 200 years. It has given remarkably goodservice and is quite tolerant of workmanship vari-ations and the main problems have been leakageof rainwater, which affects a small percentage ofcases, and the corrosion of the steel ties used toensure shared structural action of the two leaves.Figure 33.7 illustrates some routes for moistureand dampness to penetrate firstly the outer leaf(or a solid masonry wall) and then to reach theinner leaf usually because of bad design or work-manship in the construction of the cavity.
33.8 Sound transmissionSound transmission is another parameter which isvery dependent on density because it is the massof the wall which is critical. Generally, the greaterthe mass, m, of a wall the more effective it is atattenuating (absorbing) the sound passingthrough it. A typical equation for the resistance,R, in decibels, dB, is given by BRE Digest 273(1983).
R�14.3 logm�1.4dB (33.5)
Any holes will short circuit the wall so that wetplastered walls where any minor perforations orimperfections are repaired by the plaster layer aremore effective than dry-lined walls (walls coveredwith plasterboard). There are additional tech-niques to try to cut out sound such as air gapsand cavities with fireproof blankets hung in themwhich will damp out some frequencies.
More details on principles and basic values aregiven in Diamant (1986), BRE Digest 337 (1988),
Fire resistance
325
FIGURE 33.7 Mechanisms for rain water leakagethrough the external leaf and for tracking across cavities.
BRE Digest 338 (1988) and BRE Digest 333(1988).
33.9 Fire resistanceFire resistance of masonry is an important char-acteristic since it has long been recognized thatmasonry is a very effective material for resistingand preventing the spread of fire. This is nowenshrined in various building regulations datingfrom the Great Fire of London. Its effectiveness inthis role is due to the following characteristics:
1. a relatively low thermal conductivity whichprevents the fire spreading by inhibiting therise in temperature of the side of a wall awayfrom the fire;
2. a relatively high heat capacity which alsoinhibits the rise in temperature of the side of awall away from the fire (this is especially trueof concrete-based products which contain alot of loosely bound water which absorbs heatwhile being boiled away);
3. zero flammability and surface spread of flame;4. refractory properties which mean that it
retains its strength and integrity up to veryhigh temperatures, approaching 1000°C insome cases.
These properties mean that it does not catch fireitself, it inhibits spread of fire by conduction andradiation and it is not easily breached by the fire.Insulating finishes such as vermiculite plasterimprove performance still further. It has beenshown to resist fire for between half-an-hour andsix hours, depending on material, thickness andfinishes. The classic data on masonry walls arecontained in Davey and Ashton (1953). RelevantCodes of Practice are BS 5628: Part 3: 1985 andBS DD EN 1996–1–2 (1998).
33.10 Sustainability issuesThere is a social trend in existence to make build-ings and structures more ‘sustainable’ which isinterpreted as both minimizing the inputs ofresources and the impact on the environment atall stages of use from original construction toeventual removal and replacement. This broadintention is interpreted as the minimization ofenergy and raw materials input and output ofpollutants at the manufacturing stage, over thelife of the building or component and ultimatelyits recycling at the end of life of the building.Materials are ideally fully sustainable, i.e. theycome from renewable sources like timber fromtrees (which also absorb CO2 as they grow)whereas materials which are mined or quarriedand/or require fossil fuel energy to manufactureare not fully sustainable and cause a net gain of
CO2 and other pollutants. It should be remem-bered, however, that most materials have to betransported and that will consume some fossilfuels. The BRE have published a broad greenguide to building materials, Howard et al. (1998),which tries to assess them on a broad spectrum ofgreen issues and give a combined rating on a scaleof three. In this chapter the simpler principle hasbeen adopted that, excepting unusual toxic/carcinogenic materials, most of the harmfuleffects of materials can be linked to the energycost to produce and deliver them (their embodiedenergy).
For example, clay bricks are inherently fairlyhigh-energy input materials and therefore there isa considerable emission of greenhouse gasesduring manufacture and transport, together withother pollutants such as sulphur dioxide and par-ticulates. Additionally the quarrying operationsto win the clay and mining operations to obtainthe energy input (i.e. coal, oil or gas) will createaesthetic problems. Thus clay bricks cannot betreated as fully renewable; however, they have along life, on average, and can be recycled in manysituations where the building they form has cometo the end of its useful life. The industry hassought to minimize the environmental impact bymaking plant more efficient and optimizing therecycling of heat during the process and bymoving to more highly perforated bricks whichminimize resource usage. Data from the BCRLsurveys of energy usage (Beardmore, 1988, 1993,1998) in Figure 33.8, indicates that the overallenergy input to clay brick manufacture and trans-port has fallen by about 20 per cent over theperiod 1974 to 1999.
The other common materials, concrete andcalcium silicate, have lower embodied energythan clay bricks and also have a long life.
Masonry, if correctly specified and constructed,has the advantage of a very long-life with verylow maintenance costs, and this can reasonablybe set against its high initial environmental costprovided it is used for long-life structures such ashousing and public buildings. For shorter lifestructures it would be reasonable to use it if itwere readily recyclable. There is a difficulty here
Durability and non-structural properties of masonry
326
because modern opc-based mortars are tooadherent and make recycling of whole bricksuneconomic. The current UK code of practicedoes not provide a basis for design of slenderwalls constructed with air lime mortar whichwould facilitate recycling. However, this is not asignificant problem because the replacement rateof housing in the UK is only of the order 0.25 percent per annum and the overall construction rateis only just over 1 per cent. The current replace-ment rate implies that the UK expects itsdwellings to last several hundred years which isperhaps the reason why long-lived materials arepopular.
The average life, to date, for all the standingstock of masonry dwellings, is around 58 yearsbut a significant proportion is over 125 years old.Figure 33.9 illustrates the annual cost perdwelling of the process energy for contemporarybrick or block manufacture as a function of thelife of the dwelling (or walls). The upper curve isbased on a typical clay brick outer leaf with con-crete block inner walls, while the lower line isbased on an all-concrete unit wall system.
If the reasonable assumption is made of aminimum average of 100-year life for all theexisting stock, the energy consumption perannum of the clay brick option is around��GJ/year and under half this if only concrete unitsare used, assuming that no energy-consumptivemaintenance is necessary. To put this intocontext, the annual average energy cost forheating in the UK, according to Shorrock et al.(1993) is about 50 GJ per dwelling, thus the long-term energy cost of the brick masonry is between�� per cent and 1 per cent of the heating cost orless.
Thus the typical contemporary masonry wallsystem usually gives a long, low maintenance,low overall energy cost lifetime with a futureoption to recycle. Improvements in the sustain-ability can be effected by use of more timbercomponents, by incorporating recycled materials,e.g. aggregates PFA, etc., in the materials, byimproving process efficiency and by engineeringto expedite eventual recycling, e.g. by usingmortars which separate easily from bricks.
Sustainability issues
327
FIGURE 33.8 Gross manufacturing energy inputper clay brick unit (including fuel embodied in the clayand transport). Reproduction of Figure 9 from BREDigest-441, CRC Ltd (1999) Watford.
FIGURE 33.9 Lifetime embodied energy cost ofmasonry component of dwellings as a function of thelife of the building compared to current heating cost.Based on Figure 10 from BRE Digest-441, CRC Ltd.Watford.
33.11 ReferencesBeardmore, C. (1988) Fuel usage in the manufacture of
clay building bricks (6). Ceram. Research RP761,.Beardmore, C. (1990) Winter weather records relating
to potential frost failure of brickwork (2). ResearchPaper 781, British Ceramic Research Ltd.
Beardmore, C. (1993) Fuel usage in the manufacture ofclay building bricks (11). Ceram. Building Techno-logy Research Paper 804, BCRL, Stoke on Trent.
Beardmore, C. (1998) Fuel usage in the manufacture ofclay building bricks (16). Ceram. Research ResearchPaper 812, BCRL, Stoke on Trent.
Beardmore, C. and Ford, R.W. (1987) Winter weatherrecords relating to potential frost failure of brick-work (1). Trans. Brit. Ceram. Soc. J. 86, 7.
Bomley, A.V. and Pettifer, K. (1997) Sulfide-relateddegradation of concrete in Southwest England.
Concrete (1986) The concrete engineering qualitybrick. Concrete April.
Davey, N. and Ashton, L.A. (1953) Investigation ofBuilding Fires. Part V. Fire Tests on Structural Ele-ments, National Building Studies No. 12, HMSO,London.
Diamant, R.M.E. (1986) Thermal and Acoustic Insula-tion, Butterworth, London.
Howard, N., Shiers, D. and Sinclair, M. (1998) TheGreen Guide to Specification and EnvironmentalProfiling System for Building Materials and Com-ponents, BRE, Watford.
Newman, A.J. (1988) Rain Penetration ThroughMasonry Walls: Diagnosis and Remedial Measures.Building Research Occasional Paper 117. Garston,Construction Research Communications Ltd.
Newman, A.J. and Whiteside, D. (1981) Water and airpenetration through brick walls – a theoretical andexperimental study. Trans. Brit. Ceram. Soc. 80, 27.
Newman, A.J., Whiteside, D. and Kloss, P.B. (1982)Full-scale water penetration tests on twelve cavityfills – Part II. Three built-in fills. Building andEnvironment 17, 3, 193–207.
Newman, A.J., Whiteside, D., Kloss, P.B. and Willis,W. (1982) Full-scale water penetration tests ontwelve cavity fills – Part I. Nine retrofit fills. Buildingand Environment 17, 3, 175–91.
Royal Institute of Chartered Surveyors (1997) The‘Mundic’ Problem – A Guidance Note. 2nd edn.RICS, London.
Shorrock, L.D., Henderson, G. and Bown, J.H.F.(1992 and 1993) Domestic Energy Fact File. BR 220and BR 251 (update report). Garston, ConstructionResearch Communications Ltd.
Stupart, A.W. (1996) Possible extensions to developinga frost index. Masonry International 7, 1, 4–9.
Thaumasite Experts Group, Department of theEnvironment, Transport and the Regions (1999) TheThaumasite Form of Sulfate Attack: Risks, Diagno-sis, Remedial Works and Guidance on New Con-struction, DETR, London.
de Vekey, R.C. (1990a) Corrosion of Steel Wall Ties:Background, History of Occurrence and Treatment,BRE Information Paper IP12/90, Building ResearchEstablishment, Watford.
de Vekey, R.C. (1990b) Corrosion of Steel Wall Ties:Recognition and Assessment, BRE InformationPaper IP13/90, Building Research Establishment,Watford.
de Vekey, R.C. (1993) Cavity walls – Still a good solu-tion. BMS Proc. M(5), 35.
de Vekey, R.C. (1999) BRE Digest 441: Clay Bricksand Clay Brick Masonry, Part 1 and Part 2. CRCLtd. Watford.
de Vekey, R.C., Russell, A.D., Skandamoorthy, J. andFerguson, A. (1997) Bond and water resistance ofmasonry walls. Proceedings of 11th InternationalBrick/Block Masonry Conference, Shanghai 2,836–44.
West, H.W.H., Ford, R.W. and Peake, F.A. (1984) Apanel freezing test for brickwork. Proc. Brit. Ceram.Soc. 83, 112.
Durability and non-structural properties of masonry
328
Further reading
British StandardsBS 187: 1978 Calcium silicate bricks.BS 1200: 1976 (amended in 1984) Sands from natural
sources: sands for mortar for plain and reinforcedbrickwork, blockwork and stone masonry.
BS 3921: 1985 Specification for clay bricks.BS 4551: 1980 Methods of testing mortars, screeds and
plasters.BS 4729: 1990 Specifications for dimensions of bricks
of special shapes and sizes.BS 5606: 1999 Code of practice for accuracy in build-
ing.BS 5628: Part 1: 1993 Code of practice for the use of
masonry. Part 1: Structural use of unreinforcedmasonry.
BS 5628: Part 2: 1985 Structural use of reinforced andprestressed masonry.
BS 5628: Part 3: 1985 Materials and components,design and workmanship.
BS 6073: Part 1: 1981 Precast concrete masonry units.Part 1: Specification for precast concrete masonryunit; Part 2: Method for specifying precast concretemasonry units.
BS 6100: Part 5.3: 1984 Glossary of building and civilengineering terms. Part 5: Masonry.
BS 6677: Part 1: 1985 Specification for clay andcalcium silicate pavers.
BS 6717: Part 1: 1986 Specification for precast con-crete paving blocks.
BS 8000: Part 3: 1985 Workmanship in building. Part3: Masonry.
Codes of Practice and guidesBS S606: 1978 Code of practice for accuracy in building.BS 5628: Part 1: 1978 covers normal unreinforced
structural masonry such as walls, arches, tunnels,columns, etc., subject to compressive, lateral andshear loads.
BS 5628: Part 2: 1985 is the section for design of rein-forced or prestressed brickwork, e.g. for earth-retaining walls, chamber covers, etc., and has auseful treatment of the use of corrosion-resistant
reinforcement for service in harsh conditions such asfoul drains.
BS 5628: Part 3: 1985 covers all non-structural aspectsof brickwork design, particularly the specification ofunits and mortars for durability over a wide range ofapplications and also workmanship, detailing,bonding patterns, fire resistance and resistance toweather conditions.
BS 6100: Part 5.3: 1986 defines terms relevant tobricks and brickwork.
British Masonry Society. Eurocode for masonry, ENV1996–1–1: Guidance and worked examples. SpecialPublication No. 1, 1997. Stoke-on-Trent, BritishMasonry Society, 1997.
Materials standardsDRAFT EN 771–1 Clay units.DRAFT EN 771–2 Calcium silicate units.DRAFT EN 771–3 Aggregate concrete units.DRAFT EN 771–4 Autoclaved aerated concrete units.DRAFT EN 771–5 Reconstituted stone units.DRAFT EN 771–6 Natural stone units.DRAFT EN 845–1 Specification for ancillary com-
ponents for masonary: Part 1: Ties, straps, hangersand brackets.
DRAFT EN 845–2 Specification for ancillary com-ponents for masonry: Part 2: Lintels.
DRAFT EN 845–3 Specification for ancillary com-ponents for masonry: Part 3: Bed joint reinforce-ment.
DRAFT EN 998–1 Specification for mortar formasonry: Part 1: Rendering and plastering mortar.
DRAFT EN 998–2 Specification for mortar formasonry: Part 2: Masonry mortar.
Methods of test for masonry unitsBS EN 772–1: 2000 Determination of compressive
strength.BS EN 772–2: 1998 Determination of percentage area
of voids in aggregate concrete masonry units (bypaper indentation).
329
BS EN 772–3: 1998 Determination of net volume andpercentage area of voids of clay masonry units byhydrostatic weighing.
BS EN 772–4: 1998 Determination of real and bulkdensity and of total and open porosity for naturalstone masonry units.
prEN 772–5: Determination of active soluble saltscontent of clay masonry units.
prEN 772–6: Methods of test for masonry units –Determination of bending tensile strength of conretemasonry units.
BS EN 772–7: 1998 Determination of water absorptionof clay masonry damp course units by boiling inwater.
BS EN 772–9: 1998 Determination of volume and per-centage area of voids and net volume of calcium sili-cate masonry units by sand filling.
BS EN 772–10: 1999 Determination of moisturecontent of calcium silicate and autoclaved aeratedconcrete units.
BS EN 772–11: 2000 Methods of test for masonryunits – Determination of water absorption of clay,aggregate concrete, autoclaved aerated concrete,manufactured stone and natural stone masonry unitsdue to capillary action.
prEN 772–12: Determination of length change duringmoisture movement in autoclaved aerated concreteunits.
BS EN 772–13: 2000 Determination of net and grossdry density of masonry units (except for naturalstone).
prEN 772–14: Methods of test for masonry units –Determination of moisture movement of aggregateconcrete masonry units.
BS EN 772–15: 2000 Determination of water vapourpermeability of autoclaved aerated concrete masonryunits.
BS EN 772–16: 2000 Determination of dimensions.BS EN 772–18: 2000 Determination of freeze–thaw
resistance of calcium silicate masonry units.BS EN 772–19: 2000 Determination of moisture
expansion of large horizontally-perforated claymasonry units.
BS EN 772–20: 2000 Methods of test for masonryunits – Determination of flatness of faces aggregateconcrete manufactured stone and natural stonemasonry units.
prEN 772–22: Determination of freeze–thaw resistanceof clay masonry units.
Methods of test for mortar for masonryEN 1015–1999: Determination of particle size distribu-
tion (by sieve analysis).
BS EN 1015–2: 1999 Bulk sampling of mortars andpreparation of test mortars.
BS EN 1015–3: 1999 Determinatoin of consistence offresh mortar (by flow table).
BS EN 1015–4: 1999 Determination of consistence offresh mortar (by plunger penetration).
BS EN 1015–6: 1999 Determination of bulk density offresh mortar.
BS EN 1015–7: 1999 Determination of air content offresh mortar.
prEN 1015–8: Determination of water retentivity offresh mortar.
BS EN 1015–9: 1999 Determination of workable lifeand correction time of fresh mortar.
BS EN 1015–10: 1999 Determination of dry bulkdensity of hardened mortar.
BS EN 1015–11: 1999 Determination of flexural andcompressive strength of hardened mortar.
BS EN 1015–12: 1999 Determination of adhesivestrength of rendering and plastering mortars on sub-strates.
BS EN 1015–14: 1999 Determination of durability ofhardened mortar.
BS EN 1015–17: 2000 Determination of water-solublechloride content of fresh mortars.
prEN 1015–18 Determination of water absorptioncoefficient due to capillary action of hardened ren-dering mortar.
BS EN 1015–19: 1999 Determination of water vapourpermeability of hardened rendering and plasteringmortars.
prEN 1015–21 Determination of the compatibility ofone-coat rendering mortars with backgroundsthrough assessment of adhesive strength and waterpermeability after conditioning.
Methods of test for masonryBS EN 1052–1: 1999 Determination of compressive
strength.BS EN 1052–2: 1999 Methods of test for masonry –
Determination of flexural strength.prEN 1052–3: Methods of test for masonry – Determi-
nation of initial shear strength.BS EN 1052–4: 2000 Methods of test for masonry –
Determination of shear strength including dampproof course.
prEN 1052–5 Methods of test for masonry – Determi-nation of bond strength.
EurocodesBS ENV 1996–1–1: Eurocode 6 Design of masonry
structures – Part 1–1: Structural design of masonry.BS ENV 1996–1–2: Eurocode 6 Design of masonry struc-
tures – Part 1–2: General rules – Structural Fire design.
Further reading
330
ENV 1996–1–3: Eurocode 6 Design of masonry struc-tures – Part 1–3: General rules for Buildings –Detailed rules on lateral loading.
ENV 1996–2: Eurocode 6 Design of masonry struc-tures – Design, selection of materials and executionof masonry (Together with UK National ApplicationDocument – NAD).
ENV 1996–3: Eurocode 6 Design of masonry struc-tures – Part 3: Simplified calculation methods andsimple rules for masonry structures.
EN 1745: Masonry and masonry products – Methodsfor determining design thermal values.
Methods of test for ancillary componentsfor masonryBS EN 846–2: 2000 Determination of bond strength of
prefabricated bed joint reinforcement in mortarjoints.
BS EN 846–3: 2000 Determination of shear strength ofwelds in bed joint reinforcement.
PR EN 846–4 Determination of tensile and compres-sive load capacity and stiffness of wall ties (walltest).
EN 846–5: 2000 Determination of tensile and com-pressive load capacity and stiffness of wall ties(couplet test).
EN 846–6: 2000 Determination of tensile and com-pressive load capacity and stiffness of wall ties(single end test).
BS EN 846–7: 2000 Determination of shear loadcapacity and load-displacement characteristics ofshear ties and slip ties (couplet test for mortar jointconnections).
EN 846–8: 2000 Determination of strength and stiff-ness of joist hangers.
EN 846–9: 2000 Determination of flexural resistance,shear load resistance and stiffness of lintels.
EN 846–10: 2000 Determination of flexural resistanceand stiffness of shelf angles.
EN 846–11: 2000 Determination of dimensions andstraightness or bow of lintels.
PR EN 846–12 Determination of strength and stiffnessof straps.
PR EN 846–13 Determination of resistance to impact,abrasion and corrosion of organic coatings.
Other standardsASTM C9S2–76 (1976) Standard test method for bond
strength of mortar to masonry units.ASTM C1072–86 Standard method for measurement
of masonry flexural bond.
Building Research EstablishmentpublicationsThe BRE, Watford, have an extensive series of publica-
tions, particularly Digests, relating to brickwork andblockwork. The following are those of the most rele-vance to the content of this book.
Digest 35 (1963) Shrinkage of natural aggregates inconcrete (revised 1968).
Digest 108 (1975) Standard U-values.Digest 157, Calcium silicate (sandlime, flintlime) brick-
work.Digest 164, Clay brickwork: 1.Digest 165, Clay brickwork: 2.Digest 228 (1979) Estimation of thermal and moisture
movements and stresses.Digest 240, Low rise buildings on shrinkable clay soils:
Part 1.Digest 246 (1981) Strength of brickwork and block-
work walls: design for vertical load.Digest 250, Concrete in sulphate bearing soils and
groundwaters.Digest 273 (1983) Perforated clay bricks.Digest 298, The influence of trees on house founda-
tions in clay soils.Digest 329, Installing wall ties in existing construction.Digest 333 (1988) Sound insulation of separating walls
and floors. Part 1: Walls.Digest 337 (1988) Sound insulation: basic principles.Digest 338 (1988) Insulation against external noise.Digest 359, Repairing brickwork.Digest 360 (1991) Testing bond strength of masonry.Digest 361 (1991) Why do buildings crack?.Digest 362, Building mortar.Digest 363 (1991) Sulfate and acid resistance of con-
crete in the ground.Digest 420 (1997) Selecting natural building stones.Digest 432 (1998) Aircrete: thin joint mortar systems.
Performance specifications for wall ties – BRE report.BRE CP24/70, (ATO) Some results of exposure tests on
durability of calcium silicate bricks.BRE CP23/77 (1977) Chemical resistance of concrete.
Concrete, 11, 5, 35–7.Harrison, W.H. (1987) Durability of concrete in acidic
soils and groundwaters. Concrete, 21, 2.Harrison, H.W. and de Vekey, R.C. (1998) Walls,
Windows and Doors. In BRE Building Elementsseries, CRC, Watford.
Brick Development AssociationpublicationsBrick diaphragm walls in tall single storey buildings
(and earth retaining walls).
Miscellaneous publications
331
BDA Design note 3, Brickwork Dimensions Tables.BDA Design note 7, Brickwork Durability.
British Ceramic Research LimitedpublicationsTechnical Note 368, The performance of calcium sili-
cate brickwork in high sulphate environments.SP56: 1980: Model specification for clay and calcium
silicate structural brickwork (in process of updat-ing).
Supplement No. 1 to SP56, Glossary of terms relating tothe interaction of bricks and brickwork with water.
SP108, Design guide for reinforced clay brickworkpocket-type retaining walls.
SP109, Achieving the functional requirements ofmortar.
British Cement Association
Miscellaneous publicationsTechnical report TRA/145, The effects of sulphates on
Portland cement concretes and other products.
Concrete Brick ManufacturersAssociation publicationsCBMA Information Sheet 2, Concrete bricks – product
information.
Miscellaneous publicationsHendry, A.W., Sinha, B.B. and Davies, S.R. An lntro-
duction to Load Bearing Brickwork Design, EllisHorwood, Chichester.
Further reading
332
Part Six
Polymers
L. Hollaway
Introduction
Pioneers in the development of plastics includeAlexander Parkes in the UK in the 1860s and theUS chemist Leo Baekeland, who developed‘Parkesine’ and ‘Bakelite’ respectively. Much ofthe development and exploitation of polymersover the past 100 years has stemmed from thegrowth of the oil industry. Since the 1930s oil hasbeen our main source of organic chemicals, fromwhich synthetic plastics, fibres, rubbers and adhe-sives are made. The by-products of the distillationof petroleum are called basic chemicals and theyprovide the building blocks from which manychemicals and products, including plastics, can bemade.
A large variety of polymers, with a wide rangeof properties, have been developed commerciallysince 1955. For example, phenol formaldehyde
(PF) is a hard thermosetting material, polystyreneis a hard, brittle thermoplastic; polythene andplasticised polyvinyl chloride (PVC) are soft,tough thermoplastic materials. Plastics can alsoexist in various physical forms: bulk solids, rigidor flexible foams, sheet or film.
Many of these have found use in the construc-tion industry, and this section describes the pro-cessing, properties and applications of those thathave been used as sealants, adhesives, elastomersand geosynthetics.
Although many of the materials are relativelystrong, their stiffness is too low for most struc-tural applications. As we will see in the nextsection they can be combined with fibres of highstiffness and strength to form composites withimproved structural properties.
335
Chapter thirty-four
Polymers
Types, propertiesand applications
34.1 Polymeric materials34.2 Processing of thermoplastic polymers34.3 Polymer properties34.4 Applications and uses of polymers34.5 References
34.1 Polymeric materialsPolymers are produced by combining a largenumber of small molecular units (monomers) by thechemical process known as polymerisation to formlong-chain molecules. There are two main types.Thermoplastics consist of a series of long-chainpolymerised molecules. All the chains of the mole-cules are separate and can slide over one another. Inthermosetting polymers the chains become cross-linked so that a solid material is produced whichcannot be softened and which will not flow.
Polymers are usually made in one of twopolymerisation processes. In condensation-polymerisation the linking of molecules creates by-products, usually water, nitrogen or hydrogen gas.In addition-polymerisation no by-products arecreated. Both thermosetting and thermoplasticpolymers can be manufactured by these processes.
34.1.1 Thermoplastic polymers
In a thermoplastic polymer the long-chain mole-cules are held together by relatively weak Van der
Waals forces, but the chemical bond along thechain is extremely strong (Figure 34.1 (a)). Whenthe material is heated, the intermolecular forcesare weakened and the polymer becomes soft andflexible; at high temperatures it becomes a viscousmelt. When it is allowed to cool again it solidifies.The cycle of softening by heating and hardeningby cooling can be repeated almost indefinitely,but with each cycle the material tends to becomemore brittle.
Thermoplastic materials can have either a semi-crystalline ordered structure or an amorphousrandom structure. Polypropylene, Nylon 66 andpolycarbonate are examples of amorphous ther-moplastic polymers. Recent developments in thefield of engineering polymers include the intro-duction of high-performance polymers, such as
337
FIGURE 34.1 (a) Schematic representation of long-chain molecules of a thermoplastic polymer; (b)schematic representation of long-chain molecules of athermosetting polymer illustrating the cross-linking.
polyethersulphone (PES), which is amorphous,and polyetheretherketone (PEEK), which is semi-crystalline. These offer properties far superior tothose of the normal thermoplastic polymers.
34.1.2 Thermosetting polymers
The principal thermosetting polymers which areused in composites in construction are polyesters,epoxies and phenolics. Thermosetting polymersare formed in a two-stage chemical reaction.First, a substance consisting of a series of long-chain polymerised molecules, similar to those inthermoplastics, is produced; then the chainsbecome cross-linked. This reaction can take placeeither at room temperature or under the applica-tion of heat and pressure. As the cross-linking isby strong chemical bonds, thermosetting poly-mers are rigid materials and their mechanicalproperties are affected by heat. Figure 34.1 (b)shows a schematic representation.
34.1.3 Foamed polymers
A rigid foam is a two-phase system of gas dis-persed in solid polymer, and is produced byadding a blowing agent to molten resin. In theexothermic polymerisation reaction the gas isreleased and causes the polymer to expand,increasing its original volume many times by theformation of small gas cells. Like solid polymers,rigid foam polymers can be either thermoplasticor thermosetting and generally any polymer canbe foamed.
34.2 Processing of thermoplasticpolymersThermoplastic polymers may readily be processedinto sheets or rods or complex shapes in oneoperation, which is often automated. Stages suchas heating, shaping and cooling will ideally be asingle event or a repeated cycle. The principalprocessing methods are extrusion, injectionmoulding, thermoforming and calendering.
The first is the most important method fromthe civil engineering viewpoint, and this is there-
Polymers
338
fore outlined below. Powder or granules of ther-moplastic polymer are fed from a hopper to arotating screw inside a heated barrel; the screwdepth is reduced along the barrel so that thematerial is compacted. At the end of the barrelthe melt passes through a die to produce thedesired finished article. Changing the die allows awide range of products to be made, such as:
1. profile products;2. film-blown plastic sheet;3. blow-moulded hollow plastic articles;4. co-extruded items;5. highly orientated grid sheets.
34.2.1 Profile products
With different extrusion dies, many profiles canbe manufactured, such as edging strips, pipes,window-frames, etc. However, success dependsupon the correct design of the die.
34.2.2 Film-blown plastic sheet
Molten plastic from the extruder passes throughan annular die to form a thin tube; a supply of airinside the tube prevents collapse and when thefilm is cooled it passes through collapsing guidesand nip rolls and is stored on drums. Biaxial ori-entation of the molecules of the polymer can beachieved by varying the air pressure in thepolymer tube, which in turn controls the circum-ferential orientation. Longitudinal orientation canbe achieved by varying the relative speeds of thenip roll and the linear velocity of the bubble; thisis known as draw-down.
34.2.3 Blow-moulded hollow plasticarticles
A molten polymer tube, the Parison, is extrudedthrough an annular die. A mould closes round theParison and internal pressure forces the polymeragainst the sides of the mould. This method isused to form such articles as bottles and coldwater storage tanks. The materials commonlyused are polypropylene, polyethylene and PET.
34.2.4 Co-extruded items
A multilayered plastic composite is sometimesneeded to withstand the end use requirements.Two or more polymers are combined in a singleprocess by film blowing with an adhesive filmbetween them. Reactive bonding processes tochemically cross-link the polymers are underdevelopment.
34.2.5 Highly orientated grid sheets
Polymer grids are used in civil engineering as thereinforcement to soil in reinforced earth. Contin-uous sheets of thermoplastic polymers, generallypolypropylene or polyethylene, are extruded tovery fine tolerances and with a controlled struc-ture. A pattern of holes is stamped out in thesheet and the stampings are saved for re-use. Theperforated sheet is stretched in the longitudinaland then in the transverse direction to give ahighly orientated polymer in the two directionswith a tensile strength similar to that of mildsteel. The low original stiffness of the materialcan be increased ten-fold. The stiffness of unori-entated high-density polyethylene (HDPE), forinstance, is initially only 1GPa and after forminginto an orientated molecular structure it increasesto 10GPa. The use of these sheets is discussedlater in the chapter.
In injection moulding, softened thermoplasticpolymer is forced through a nozzle into aclamped cold mould. When the plastic becomescold, the mould is opened and the article isejected; the operation is then repeated.
34.3 Polymer properties
34.3.1 Mechanical properties
The physical and mechanical properties of ther-mosetting resins, specifically polyesters andepoxies, can vary greatly. As mentioned above,thermosetting polymers are cross-linked and forma tightly bound three-dimensional network ofpolymer chains; the mechanical properties arehighly dependent upon the network of molecular
Polymer properties
339
units and upon the lengths of cross-link chains.The characteristics of the network units are afunction of the chemicals used and the length ofthe cross-linked chains is determined by thecuring process. It is usual to cure composites byheating to achieve optimum cross-linking andhence to enable the mechanical properties torealise their potential. Shrinkage during curingdoes occur, particularly with polyesters, and con-traction on cooling to ambient temperature canlead to stress build-up. This latter problem is dueto the differences between the thermal expansioncoefficients of the matrix and fibre, and it canhave a major effect on the internal micro-stresseswhich are sometimes sufficient to produce micro-packing, even in the absence of external loads.
Thermoplastic polymers which are not cross-linked derive their strength and stiffness from theproperties of the monomer units and the highmolecular weight. Consequently, in crystallinethermoplastic polymers there is a high degree ofmolecular order and alignment, and during anyheating the crystalline phase will tend to melt andform an amorphous viscous liquid. In amorphousthermoplastic polymers, there is a high degree ofmolecular entanglement so that they act like across-linked material. On heating, the moleculesbecome disentangled and the polymer changesfrom a rigid solid to a viscous liquid.
Table 34.1 gives the most important mechani-cal properties of thermosetting and thermoplasticpolymers.
34.3.2 Time-dependent characteristics
Polymer materials exhibit a time-dependent strainresponse to a constant applied stress; this behavi-our is called creep. Conversely, if the stress on apolymer is removed it exhibits a strain recovery.Figure 34.2 illustrates the total creep curve for apolymer under a given uni-axial tensile stress atconstant temperature, which is divided into fiveregions. When a polymer is subjected to a con-stant tensile stress, its strain increases until thematerial fails at the end of tertiary creep region.This process is called creep rupture, and thefailure point depends on the level of stress. The
designer must be aware of this failure mode andrealise that polymers, which are tough undershort static loads, may become embrittled underlong-term loading.
A further important consequence of the time-dependent behaviour of polymers is that whenthey are subjected to a particular strain, the stressnecessary to maintain this strain decreases withtime.
Polymer materials have mechanical propertieswhich lie somewhere between the ideal Hookeanmaterial, where stress is proportional to strain,and the Newtonian material, where stress is pro-portional to rate of strain. They are termed‘visco-elastic’ materials and their stress is a func-
Polymers
340
tion of strain and time, and can be described byan equation of the form:
� � f(�, t) (34.1)
This non-linear visco-elastic behaviour can besimplified for design purposes into the form:
���f(t) (34.2)
This linear visco-elastic response indicates that,under sustained tensile stress, after a particulartime interval the stress is directly proportional tostrain. Figure 34.3 describes the various types ofresponse discussed above.
Short-term tensile tests are commonly used tocharacterise metals. However, they must betreated with caution when testing polymers. It ispossible to obtain quite different results bychanging, for instance, the rate of extension ofthe specimen under a tensile force. At high ratesof strain an un-plasticised PVC will show brittlecharacteristics and relatively high modulus andstrength, whereas at low rates of strain the mater-ial will be ductile and have a lower modulus andstrength. Consequently, the test conditions mustbe consistent with the service condition for whichthe material is being designed.
Creep tests are often used to measure the defor-mational behaviour of polymeric materials. Thevariation of strain with time, when the specimenis under constant load, is usually recorded on a
TABLE 34.1 Mechanical properties of common thermosetting and thermoplastic polymers
Material properties Specific weight Ultimate tensile Modulus of Coefficient ofstrength elasticity in tension linear expansion(MPa) (GPa) (10�6/°C)
ThermosettingPolyester 1.28 45–90 2.5–4.0 100–110Epoxy 1.30 90–110 3.0–7.0 45–65Phenolic (with filler) 1.35–1.75 45–59 5.5–8.3 30–45
ThermoplasticsPolyvinyl chloride (PVC) 1.37 58.0 2.4–2.8 50Acrylonitrile butadiene
styrene (ABS) 1.05 17–62 0.69–2.82 60–130Nylon 1.13–1.15 48–83 1.03–2.76 80–150Polyethylene (high-density) 0.96 30–35 1.10–1.30 120
FIGURE 34.2 Total creep curve for a polymerunder a given uniaxial stress state.
Applications and uses of polymers
341
FIGURE 34.3 Stress–strain behaviour of elastic andviscoelastic materials at two values of time.
FIGURE 34.4 Typical relationship for a linear viscoelastic material.
logarithmic time-scale so that the time depen-dence for long periods can be extrapolated ifnecessary. Figure 34.4 represents a typical rela-tionship between strain and time; a family ofcurves may be obtained for different stress levels.Figure 34.5 shows the relationship drawn on alogarithmic time-scale.
Figure 34.6 (a) shows a family of strain–timegraphs for a typical non-linear polymer. Fromthis, it is clear that the higher the stress applied tothe polymer, the higher is the creep rate. Replot-ted as a stress–time relationship, Figure 34.6 (b)represents the relaxation of stress in the polymer
at constant strain; this is termed an isometricgraph. If the stresses at a particular time t on thecreep curve are divided by the respective strainvalues, then the modulus of elasticity–time curveis shown in Figure 34.6 (c). Finally, if at time t thestress values are plotted against the respectivestrain values, an isochronous graph, Figure34.6 (d), results. Isochronous data are oftenrepresented on log–log scales; the graph willgenerally be a straight line and its slope will indi-cate the degree of non-linearity of the polymermaterial; a slope of 45° represents a linear mater-ial, whereas a non-linear material will have aslope less than 45°.
34.4 Applications and uses ofpolymers
34.4.1 Sealants
Sealants are elastomeric materials which can beused for sealing joints against wind and water inconstruction. Thin curtain wall constructionemploys highly effective materials to provide theheat installation, but generally there is no cavityfor the dispersion of water that may leak throughthe joints on the outside. In addition, in the eventof air blowing directly to the inside of a jointthere must be an effective material to provideheat insulation; consequently, a baffling system toprovide this must be installed. Therefore, adhe-sive and elastic sealants are required to enablethis type of construction to be used efficiently.
The largest variety of sealants fall into the clas-sification of solvent release and are composed ofthree component parts:
1. the basic non-volatile vehicle (the liquidportion of the compound);
2. the pigment component;3. a solvent or thinner to make application
easier.
The non-volatile vehicle can vary from a veget-able oil (e.g. linseed) to a synthetic elastomer.Opacity or colour will be introduced into thematerial by the pigment component. To enablethe sealant to be applied easily and to ensure the
Polymers
342
FIGURE 34.6 Typical creep curves for a non-linear viscoelastic material for varying stress values. Isometricand isochronous stress–strain curves.
FIGURE 34.5 Typical relationship for a linear vis-coelastic material drawn on a logarithmic time-scale.
correct thickness is achieved a solvent is intro-duced. The sealant is cured and its required vis-cosity is reached by the evaporation of thesolvent. The butyl rubber solution and the acryliccopolymer solution fall into this category.
Another group of sealants are those which arechemically cured. The polysulphide compoundsand the silicone-based compounds are the mainsealants under this heading. The latter, which is atwo-part sealant, is highly dependent upon theenvironmental conditions for its rate of cure;thus, if the temperature and humidity are low, thecuring period could be very long. The chemicallycured compounds require adhesion additives inorder to develop a bond to a surface as they donot generally contain much solvent.
The desired properties of a sealant are:
• a good adhesion with the joint;• low rate of hardening;• low rate of shrinkage;• permanent elasticity.
The choice of sealants is a compromise as no oneproduct has all the above-mentioned attributes.
34.4.2 Adhesives
Within the context of the construction industrythe term adhesive embraces not only those mater-ials which are used to bond together two com-ponents of a structure, but also those materialswhich provide a specific function in themselves(e.g. protection, decoration) and are at the sametime self-adhesive to the substrate whose surfacethey modify. Thus, a mortar which may be usedto bond bricks together, may also be applied as aself-adhesive protection and often decorative ren-dering over the finished blockwork.
Adhesive bonded connections are the primaryconcern. The physical nature of the fibre/matrixcomposites does introduce problems that are notencountered in metals. The fibre type andarrangement, as well as the resin type and fibrevolume fraction will influence the behaviour ofthe joint. In addition, composites are not gener-ally homogeneous throughout their thickness asmany thermosetting polymers have gel coats;resin-rich surface layers are brittle and whenoverloaded are liable to display a brittle fracture.An appropriate resin should therefore be chosen,such as a compliant one, which will distribute theapplied load over a large area, thus reducing thestress taken by the friable surface of the compos-ite. There are two particular problems associatedwith adhesive bonding of FRP materials:
• the attachment to the surface of a layeredmaterial;
• the surface may be contaminated with mouldrelease agents remaining on it from the manu-facturing procedures.
As the matrix material in a polymer compositeis also an organic adhesive, the polymers which
Applications and uses of polymers
343
are then used to join composite materials togetherare likely to be similar in terms of chemical com-position and mechanical properties. Currently theepoxy and the acrylic based toughened adhesivesare used for general application and have provedover the years to be very versatile and easy to use.Durable, robust and relatively free from toxichazards, the toughened adhesives exhibit highpull strengths.
The basic requirements for the production of asatisfactory joint are:
• selection of a suitable adhesive (the commongeneric type of adhesive used in civil engin-eering is an epoxy, which can be formulatedin a wide range of forms to provide a broadrange of application characteristics);
• adequate preparation of the adherend surfaces;• appropriate design of the joint;• controlled fabrication of the joint itself.
Adhesive formulations are, in general,complex. To the base resin is added one of arange of different types of hardener and additives,such as fillers, toughening agents, plasticisers sur-factants, anti-oxidants and any other requiredmaterials. The hardeners are chosen dependingupon whether the cure of the resin is to be atambient or at elevated temperatures; the rate ofchemical reaction is approximately doubled forevery 8°C rise in temperature. It will be clear thatthe properties of the adhesive will be altered withthe large variety of additives that can be incorpo-rated into the base resin.
If the adhesive is required to join two dissimilarmaterials, such as polymer composite and con-crete or steel, the mechanical and thermal proper-ties should be considered in relation to these twomaterials. The effects of environmental and otherservice conditions on the adhesive material andon the behaviour of bonded joints must be con-sidered carefully.
With some bonding surfaces, such as steel, itwill generally be necessary to apply an adhesive-compatible primer coat to generate a reliable andreproducible surface. With concrete surfaces itmight be advisable to use a primer to enable suit-able conditions on which to apply a relatively
viscous adhesive. It will not generally be neces-sary to prime a composite surface.
As with all resins it is necessary to keep a closecheck on the following items to ensure that theadhesive, when used in the field, is in pristinecondition:
• the shelf life is within the manufacturer’s recommended time limit;
• the viscosity and wetting ability is satisfactory;• the curing rate is correct;• the ambient temperature does not fall below
the specification value;• post cure is complete before load is applied to
the joint.
34.4.3 Elastomers
The elastomer is another member of the polymerfamily. The material consists of long chain mole-cules which are coiled and twisted in a randommanner and the molecular flexibility is such thatthe material is able to undergo very large defor-mations. The material is cross-linked by a processknown as vulcanisation, which prevents the mole-cules of the elastomer under load, moving irre-versibly relative to each other. After a curingprocess, the molecules are cross-linked like a ther-mosetting polymer. As the vulcanisation processdoes not change the form of the coiled molecules,but merely prevents them from sliding, the elas-tomeric material will completely recover its ori-ginal shape after the removal of an external force.
34.4.4 Geosynthetics
One area of the major advances made by poly-mers in the last 25 years has been the burgeoninguse of these materials in the geotechnical engin-eering industry. The most commonly knownmaterial is the geotextile; a simple definition ofthis is a textile material used in a soil environ-ment.
In the early 1970s, these materials werereferred to as civil engineering fabric or filterfabric where their primary use was for filterapplications. In the latter part of the 1970s, they
Polymers
344
were referred to as geotechnical fabrics as theywere primarily used in the geotechnical soilengineering applications. It was in the early1980s that the term geotextile was suggested assuitable. At the same time impermeable polymericmembranes were also being increasingly used,and these became known as geomembranes. Thusin the mid 1980s many types of polymeric-basedmaterials were being used in the geotechnicalengineering industry and many of these could notbe classed as either a geotextile or a geomem-brane. To encompass all these polymeric mater-ials the new name ‘geosynthetic’ was derived,which is defined as a synthetic (polymeric) mater-ial used in a soil environment.
Geosynthetics, which are all thermoplasticpolymers, can be divided into five broadly basedcategories:
• Geotextiles: polymeric textile materials usedin geotechnical engineering applications.These materials are essentially permeable tothe passage of water.
• Geogrids: open, mesh-like polymeric structures.• Geomembranes: polymeric materials in sheet
form which are essentially impermeable to thepassage of water.
• Geo-linear elements: long, slender, polymericmaterials normally used as reinforcingtendons in soils and rocks.
• Geocomposites: covers all polymeric materialsused in a soil environment not covered by theabove four categories.
Each of these is now discussed.
Geotextiles
Geotextiles are usually classified by their methodof manufacture and are made in two stages: themanufacture of the linear elements, such as fibres,tapes, etc., and the fabrication of those linear ele-ments into geotextiles. The fibres are the basicload-bearing elements in the material and theforming technique determines the structure andhence the physical and mechanical characteristicsof the system. The main fibres used in geotextilesare the synthetic ones such as polyethylene,
polypropylene, polyester and polyamide; thesewill be described in more detail in the nextsection.
Geomembranes
Geomembranes are synthetic materials manufac-tured in impermeable sheet form from thermo-plastic polymers or bituminous materials. Bothmaterials can be reinforced or unreinforced; the former is manufactured on a production lineand the latter can be produced on a productionline or in situ. The matrix can be reinforced bytextiles.
Geo-linear elements
Geo-linear elements are long, slender strips orbars consisting of a unidirectional filament fibrecore which is made from a polyester, aramid orglass fibre in a polymer sheath of a low-densitypolyethylene or a resin. The components of thesystem form a composite, in that the fibre pro-vides the strength and extension characteristicsand the matrix protects the fibre from internalinfluences and provides the bonding character-istics with the soil.
Geogrids
Geogrids are often grid-like structures of thermo-plastic polymer material, and in conjunction with
References
345
the soil form a quasi-composite system, where thegrid structure is the fibre and the soil is assumedto be the ‘matrix’ and forms an efficient bondwith the fibre. Geogrids are of two forms:
1. cross-laid strips; and2. punched thermoplastic polymer sheets.
The manufacturing techniques will be discussedin Chapter 35.
Geocomposites
Geocomposites consist of two or more differenttypes of thermoplastic polymer systems combinedinto a hybrid material. Their main function is toform a drainage passage along the side of thewater course, with a polymer core as the drainagechannel and the geotextile skin as the filter.
As is apparent, many of the materials in eachof the above groups are fibre composites of aparent polymer reinforced with polymer fibres,and so will be discussed in more detail in the nextsection.
34.5 ReferencesBrandon, D. and Kaplan, W.D. (1997) Joining
Processes, John Wiley and Sons, Chichester.Hollaway, L. (1993) Polymer Composites for Civil and
Structural Engineering, Blackie Academic and Pro-fessional, London.
Rodriguez, F. (1996) Principles of Polymer Systems, E & FN Spon, London.
Part Seven
Fibre Composites
Section One:PolymercompositesL. Hollaway
Section Two:Fibre-reinforcedcements andconcreteD.J. Hannant
349
Introduction
This part is divided into two sections to reflectthe difference in the mechanics of the reinforcingprocess between fibre-reinforced polymers andcementitious materials. The first, on fibre-rein-forced polymers, is mostly concerned with pre-cracking behaviour, and the second, onfibre-reinforced cement and concrete, is mainlyconcerned with post-cracking performance.
The history of fibre-reinforced composites asconstruction materials is more than 3000 yearsold. Well-known examples are the use of straw inclay bricks, mentioned in Exodus, and horsehair inplaster. Other natural fibres have been used overthe ages to reinforce mud walls and give addedtoughness to rather brittle building materials.
In the last section we commented that poly-mers, although relatively strong, have low stiff-ness. They can, however, be combined with fibresof high stiffness and strength to form a range ofcomposites with improved structural performancewhich have an increasing number of applications.These include ‘hi-tech’ uses of carbon fibres inresin systems for aircraft components and sportsequipment and glass fibre reinforced systems(GFRP) for car bodies and ships’ hulls. In civilengineering there has been a steadily increasinguse of the latter group for in-fill panels supportedby main structural elements of steel or reinforcedconcrete, and for modest span lightweight bridgesconsisting entirely of the fibre composite. Anincreasing use of both types of system is forpanels fitted to existing structures for strengthen-ing after damage or for upgrading. The composi-tion, manufacture, analysis, properties and usesof these systems are described in Section 1.
Fibre-reinforced composites can also be basedon inorganic cements and binders. These aredescribed in Section 2. Most fibre-reinforced,
cement-based composites differ from fibre-rein-forced polymers in that most of the reinforcingeffects of the fibres occur after the brittle matrixhas cracked either at the microscopic level orwith visible cracks through the composite. This isthe result of the relatively low strain to failure ofthe cement matrix (0.01–0.05 per cent) com-pared with the high elongation of the fibres(1–5 per cent). The fibres in cement-basedsystems often have a lower modulus of elasticitythan the cement matrix and hence little or noincrease in cracking stress is expected from thefibre reinforcement.
The most notable exception to this division isasbestos cement which has been the most com-mercially successful fibre-reinforced compositethis century, in terms of both tonnage andturnover, the latter having been in excess of £1billion per annum world-wide for many years.Asbestos cement was invented in about 1900 byHatschek and its high tensile cracking stress andfailure strain (in excess of 0.1 per cent) results inpart from the suppression of cracks propagatingfrom flaws. Usage of asbestos fibres has beenreduced since about 1980 due to their carcino-genic effects and this has stimulated the searchfor alternative fibre systems, chiefly glass andpolypropylene, which have enabled the tradi-tional products to be safely marketed even if withconsiderably changed properties.
Starting in the 1960s, there has been growinginterest in modifying the properties of fresh andhardened concrete by the addition of fibres. Theseinclude both steel and polypropylene, and, to alimited extent, glass fibres. Their use is nowwidespread in the construction industry, and therange of materials, their properties and mostcommon applications are described.
351
Section 1
Polymercomposites
IntroductionComposites which are used to form engineeringmaterials and which consist of strong, stiff fibresin a polymer resin require scientific understandingfrom which design procedures may be developed.The mechanical and physical properties of thecomposite are clearly controlled by their con-stituent properties and by the micro-structuralconfigurations. It is, therefore, necessary to beable to predict properties when parameter varia-tions take place.
The next important aspect of composite mater-ial design is the property of anisotropy; it isnecessary to give special attention to the methodsof controlling this property and its effect on ana-lytical and design procedures.
Research work has demonstrated that with the
correct process control and a soundly basedmaterial design approach, it is possible toproduce composites that can satisfy stringentstructural requirements. However, in the con-struction industry, because it is difficult to estab-lish the durability of the composite over a periodof, say, 50 years, great care has to be exercised inpredicting stress and deformations over this timespan. It will, however, be apparent from thissection that there are many examples of success-ful structural applications of polymer composites.Some of the examples discussed below are outsidethe construction field, but they have beenincluded because they are particularly note-worthy, demonstrating how fibre-reinforcedpolymer composites have offered unique solu-tions to a range of product demands
Chapter thirty-five
Fibres forpolymercomposites
35.1 Fibre manufacture35.2 Fibre properties
When a load is applied to a fibre-reinforced com-posite consisting of a low-modulus matrix rein-forced with high-strength, high-modulus fibres,the visco-elastic flow of the matrix under stresstransfers the load to the fibre; this results in ahigh-strength, high-modulus material whichdetermines the stiffness and strength of the com-posite and is in the form of particles of highaspect ratio (i.e. fibres), is well dispersed andbonded by a weak secondary phase (i.e. matrix).
Many amorphous and crystalline fibres can beused, including glass, carbon and boron and fibresproduced from synthetic polymers. Making a fibreinvolves aligning the molecules of the material, andthe high tensile strength is associated withimproved intermolecular attraction resulting fromthis alignment. Polymeric fibres are made fromthose polymers whose chemical composition andgeometry are basically crystalline and whose inter-molecular forces are strong. As the extensibility ofthe material has already been utilised in the processof manufacture, fibres have a low elongation.
The following sections will discuss the manufac-ture and make-up of fibres which can be used toupgrade polymers, cements, mortars and concretes.The last three will be covered in greater detailin Chapters 42 and 43. Glass, carbon and Kevlar
fibres are used in conjunction with thermosettingpolymers such as polyesters and epoxies.
35.1 Fibre manufacture
35.1.1 Glass fibres
Glass fibres are manufactured by drawing moltenglass from an electric furnace through platinumbushings at high speed; see Figure 35.1. The fila-ments cool from the liquid state at about 1200°Cto room temperature in 10�5 seconds. On emerg-ing from the bushings the filaments are treatedwith a lubricant or size and 204 filaments arebundled together to form a strand. The lubricant
1. facilitates the manufacturing of the strandsand moulding of the composite;
2. reduces damage to the fibres during mechani-cal handling;
3. reduces the abrasive effect of the filamentsagainst one another.
The following four types of glass are used forfibres.
1. E-glass of low alkali content is the commonestglass on the market and is the major one usedin composites in the construction industry. Itwas first used in 1942 with polyester resin andis now widely used with polyester and epoxyresin.
352
353
Fibre manufacture
2. A-glass fibre of high alkali content was for-merly used in the aircraft industry, but it isnow little used.
3. Z-glass (zirconia glass) was developed forreinforcing cements, mortars and concretesbecause of its high resistance to alkali attack(see Chapter 42).
4. S2-glass fibre is used in extra-high-strengthand high-modulus applications in aerospace.
Strands of glass fibre are combined to formthicker parallel bundles called rovings which,when twisted, can form several different types ofyarn; rovings or yarns can be used individually orin the form of woven fabric.
Glass strands for reinforcing thermosettingpolymers may be used in a number of differentforms:
1. woven rovings, to provide high strength andstiffness characteristics in the direction of thefibres;
2. chopped fibres, to provide a randomly orien-tated dispersion of fibres in the composite;
3. chopped strand mat, to provide a quasi-isotropic reinforcement to the composite;
FIGURE 35.1 Schematic representation of themanufacture of glass fibre.
4. surface tissues, to provide a thin glass-fibremat when a resin-rich surface of composite isrequired.
35.1.2 Carbon fibres
Carbon fibres are produced commercially as syn-thetic fibres similar to those used for making tex-tiles and from pitch, which is obtained by thedestructive distillation of coal.
There are three different precursor materialsused at present to produce carbon fibres.
Rayon precursors – the earliest precursors usedto make carbon fibres were derived from cellu-losic materials. The significant disadvantage ofthis precursor is that only a small part (typically25 per cent) of the initial fibre mass remains aftercarbonisation which makes it an expensive fibre.
Polyacrylonitrile (PAN) precursors – the major-ity of carbon fibres commercially available arederived from this precursor. The carbon fibre con-version yield is 50–55 per cent. Carbon fibres basedupon the PAN precursor have a higher tensilestrength than fibres based upon other precursors.This is due to their having less surface imperfec-tions than other manufactured carbon fibres.
Pitch precursors – pitch is a relatively low-costand a high-yield material. However, non-unifor-mity from batch to batch is a serious problem.
The conversion process for carbon fibreincludes stabilisation at temperatures up to400°C, carbonisation at temperatures from800°C to 1200°C and graphitisation in excess of2000°C. Surface treatments of the fibres are thenundertaken which includes sizing and spooling.
With the increased use of carbon fibres in theengineering field and particularly civil engin-eering, new processes are being developed. Allcommercial production of PAN precursor carbonfibres is by spinning and the cross-section isround. New manufactured carbon fibres are nowcommercially available to produce different pre-cursor shapes including ‘I’ and ‘cross’ types. Theadvantage of non-round shapes is that closer fibrepacking in composites is possible, permittinghigher load-carrying capacity than that forround-section fibres.
There are three grades of carbon fibres:
• Type I is the stiffest and has the highestmodulus of elasticity of the three. The heattreatment temperature is in excess of 2000°C.
• Type II is the strongest and is heat treated atabout 1500°C.
• Type III is the least stiff of the three and has aheat treatment at about 1350°C.
35.1.3 Aramid fibres
The aramid fibre (aromatic polyamide) was intro-duced in 1972 by Du Pont under the trade nameof Kevlar. Recently several other manufacturershave produced related aramid fibres. The struc-ture of the fibre is anisotropic, resulting in higherstrength and stiffness in the fibre longitudinaldirection.
Aramid fibres are resistant to fatigue, bothstatic and dynamic. They are elastic in tensionbut exhibit non-linear characteristics undercompression and care must be taken when highstrain compressive or flexural loads are involved.Aramid fibres exhibit good toughness and generaldamage tolerance characteristics.
There are two forms of Kevlar fibre, the Kevlar29 with a high strength and intermediate stiffnessand Kevlar 49 with a high modulus and equival-ent strength to that of the Kevlar 29. Table 35.2gives the mechanical properties of the two fibres.The Kevlar 49 is used for high-performance com-posite materials.
The fibre is produced by an extrusion and spin-ning process. A solution of the polymer in a suit-able solvent at a temperature between �50°C and�80°C is extruded into a hot-walled cylinder at200°C; this causes the solvent to evaporate andthe resulting fibre is wound onto a bobbin. Thefibre then undergoes a stretching and drawingprocess to increase its strength and stiffness.
35.1.4 Linear organic fibres
By arranging the molecular structure of simplepolymers during their manufacture, to be orien-tated into one direction, a high-strength and high-
modulus organic fibre can be produced. Thisfibre, in future, could be one of the major rein-forcements for civil engineering structures. With aspecific weight of 0.97, high modulus polyethyl-ene fibres, produced in the USA and The Nether-lands, have mechanical properties of the sameorder as those of aramid fibres with modulus ofelasticity and strength values of 117GPa and2.9GPa respectively. These values were deter-mined at ambient temperature but will decreaserapidly with increasing temperature. Further-more, with non-cross-linked thermoplasticpolymer fibres, creep will be significant.However, recently it has been claimed that theseproblems can be overcome by cross-linking usingradiation.
35.1.5 Other synthetic fibres
The most important fibres for upgrading cementsand mortars or for use in reinforced earth situ-ations are polypropylene, polyethylene, polyesterand polyamide. The first two are utilised in themanufacture of cement/mortar composites andare discussed further in Chapter 42, but all fourare used in geosynthetics, especially to form geo-textiles and geogrids as described in Chapter 39.Synthetic fibres are the only ones which can beengineered chemically, physically and mechani-cally to suit particular geotechnical engineeringapplications. Natural fibres (e.g. cotton, jute) andthe majority of regenerated fibres (e.g. cellulose,rayon) are seldom used to make geotextilesbecause they are biodegradable; however, geotex-tiles made from natural fibres and even paper(another fibrous product) may serve temporaryfunctions where biodegradation is desirable (e.g.temporary erosion control).
Manufacture begins with the transformation ofthe raw polymer from solid to liquid either bydissolving or melting. Synthetic polymers such asacrylic, modacrylic, aramid and vinyladpolymersare dissolved into solution, whereas the polyolefinand polyester polymers are transformed intomolten liquid; chlorofibre polymers can be trans-formed into a liquid by either means. A spinneretconsisting of many holes is used to extend the
Fibres for polymer composites
354
Fibre properties
355
liquid polymer which is then solidified into con-tinuous filaments.
The filaments undergo further extension intheir longitudinal axes, thus further increasing theorientation of the molecular chain within the fila-ment structure, with a consequent improvementin the stress–strain characteristics. Different typesof synthetic fibre or yarn may be produced,including monofilament fibres, heterofilamentfibres, multifilament yarns, staple fibres, stapleyarns, split-film tapes and fibrillated yarns.
35.2 Fibre propertiesThe advantage of composite materials over con-ventional ones is that they have high specificstrength and high specific stiffness, achieved bythe use of low-density fibres with high strengthand modulus values. The strength and stiffnessvalues of carbon, glass and Kevlar fibres are givenin Table 35.1.
It has been stated that there are three types ofcarbon fibre: the high-modulus, the high strengthand the intermediate-modulus fibre. The degreeof alignment of the small crystalline units in thefibres varies considerably with the manufacturingtechnique, which thus affects the stiffness of thefibre. The arrangement of the layer planes in thecross-section of the fibre is also important,because it affects the transverse and shear proper-ties. Table 35.1 gives the principal mechanicalproperties of these fibres and illustrates this point.
The strength and modulus of elasticity of glassfibres are determined by the three-dimensionalstructure of the constituent oxides which can beof calcium, boron, sodium, iron or aluminium.The structure of the network and the strength ofthe individual bonds can be varied by the addi-tion of other metal oxides and so it is possible toproduce glass fibres with different chemical andphysical properties. Unlike carbon fibres theproperties of glass fibres are isotropic and themodulus of elasticity is the same along and acrossthe fibre.
The main factors which determine the ultimatestrength of glass fibres are the processing con-dition and the damage sustained during handlingand processing. The mechanical properties of twotypes of glass fibre are given in Table 35.1.
The manufacturing processes for Kevlar fibresalign the stiff polymer molecules parallel to thefibre axes, and the high modulus achieved indic-ates that a high degree of alignment is possible.Typical properties of Kevlar fibres are given inTable 35.1. When the fibres have been incorpo-rated into a matrix material, composite actiontakes place and as discussed in the next chapter, aknowledge of the fibre alignment, fibre volumefraction and method of manufacture is necessaryto obtain the mechanical characteristic of thematerial. To illustrate this, Table 35.2 gives thetensile characteristics of the different systems ofGFRP composites; the various methods of manu-facture of the composites are given in Chapter 37.
TABLE 35.1 Typical mechanical properties for the three fibres used in construction
Material properties Relative density Ultimate tensile Modulus of elasticitystrength in tension(GPa) (GPa)
Carbon fibreType I 1.92 2.00 345Type II 1.75 2.41 241Type III 1.70 2.21 200
E-glass fibre 2.55 2.4 72.4S2-glass fibre 2.47 4.6 88.0Kevlar fibres
29 1.44 2.65 6449 1.45 2.65 127
Fibres for polymer composites
356
TABLE 35.2 Typical mechanical properties for glass-reinforced polymer composites
Material Glass content Relative Tensile Tensile(weight %) density modulus strength
(GPa) (MPa)
Unidirectional rovings 50–80 1.6–2.0 20–50 400–1250(filament windingor pultrusion)
Hand lay-up with 25–45 1.4–1.6 6–11 60–180chopped strandmat
Matched dye 25–50 1.4–1.6 6–12 60–200moulding withpreform
Hand lay-up with 45–62 1.5–1.8 12–24 200–350woven rovings
DMC polyester 15–20 1.7–2.0 6–8 40–60(filled)
SMC 20–25 1.75–1.95 9–13 60–100
35.3 ReferencesHull, D. (1992) An Introduction to Composite
Materials, Cambridge University Press, Cambridge.Holloway, L. (1993) Polymer Composites for Civil and
Structural Engineering, Blackie Academic and Pro-fessional, Glasgow.
Kim, D.-H. (1995) Composite Structures for Civil andArchitectural Engineering, E & F Spon, London.
Chapter thirty-six
Analysis of thebehaviour ofpolymercomposites
36.1 Characterisation and definition of composite materials36.2 Elastic properties of continuous unidirectional laminae36.3 Elastic properties of in-plane random long-fibre laminae36.4 Macro-analysis of stress distribution in a fibre/matrix
composite36.5 Elastic properties of short-fibre composite materials36.6 Laminate theory36.7 Isotropic lamina36.8 Orthotropic lamina36.9 The strength characteristics and failure criteria of
composite laminae36.10 References
36.1 Characterisation and definitionof composite materialsThe mechanical properties of polymers can begreatly enhanced by incorporating fillers and/orfibres into the resin formulations. Therefore, forstructural application, such composite materialsshould:
1. consist of two or more phases, each with theirown physical and mechanical characteristics;
2. be manufactured by combining the separatephases such that the dispersion of one mater-ial in the other achieves optimum propertiesof the resulting material;
3. have enhanced properties compared withthose of the individual components.
In fibre-reinforced polymer materials, theprimary phase (the fibre) uses the plastic flow ofthe secondary phase (the polymer) to transfer theload to the fibre; this results in a high-strength,high-modulus composite. Fibres generally haveboth high strength and high modulus but theseproperties are only associated with very fine fibreswith diameters of the order of 7–15mm; theytend to be brittle. Polymers, however, may beeither ductile or brittle and will generally havelow strength and stiffness. By combining the twocomponents a bulk material is produced with astrength and stiffness dependent upon the fibrevolume fraction and the fibre orientation.
The properties of fibre/matrix compositematerials are highly dependent upon the microstructural parameters such as:
1. fibre diameter;2. fibre length;3. fibre volume fraction of the composite;4. fibre orientation and packing arrangement.
It is important to characterise these parameterswhen considering the processing of the compositematerial and the efficient design and manufactureof the composite made from these materials.
The interface between the fibre and the matrixplays a major role in the physical and mechanical
357
Analysis of the behaviour of polymer composites
358
properties of the composite material. The transferof stresses between fibre and fibre takes placethrough the interface and the matrix, and in theanalysis of composite materials a certain numberof assumptions are made to enable solutions tomathematical models to be obtained:
1. the matrix and the fibre behave as elasticmaterials;
2. the bond between the fibre and the matrix isperfect and consequently there will be nostrain discontinuity across the interface;
3. the material adjacent to the fibre has the sameproperties as the material in bulk form;
4. the fibres are arranged in a regular or repeat-ing array.
The properties of the interface region are veryimportant in understanding the stressed compos-ite. The region is a dominant factor in the frac-ture toughness of the material and in itsresistance to aqueous and corrosive environ-ments. Composite materials which have weakinterfaces have low strength and stiffness but highresistance to fracture, and those with strong inter-faces have high strength and stiffness but are verybrittle. This effect is a function of the ease ofdebonding and pull-out of the fibres from thematrix material during crack propagation.
Using the above assumptions, it is possible tocalculate the distribution of stress and strain in acomposite material in terms of the geometry ofthe component materials.
36.2 Elastic properties of continuousunidirectional laminae
36.2.1 Longitudinal stiffness
A basic laminate is shown in Figure 36.1 and it isassumed that the orthotropic layer has threemutually perpendicular planes of property sym-metry; it is characterised elastically by fourindependent elastic constants (refer to Section36.6 and to Section 36.8, Figure 36.5). They are:
E11 �modulus of elasticity along fibre directionE22 �modulus of elasticity in the transverse direc-
tion FIGURE 36.1 Basic laminate.
�12 �Poisson’s ratio, i.e. strains produced indirection 2 when specimen is loaded indirection 1
G12� longitudinal shear modulus�21 �Poisson’s ratio, i.e. obtained from the equa-
tion E11�21 �E22�12
If the line of action of a tensile or compressiveforce is applied parallel to the fibres of a unidirec-tional lamina, the strain �m in the matrix will beequal to the strain �f in the fibre, provided thebond between the two components is perfect. Asboth fibre and matrix behave elastically then:
�f �Ef �t and �m �Em�m where �t � �m
As Ef �Em the stress in the fibre must begreater than the stress in the matrix and willtherefore bear the major part of the applied load.
The composite load Pc �Pm �Pf or
�cAc � �mAm � �fAf
�c � �mVm � �fVf (36.1)
where A� the area of the phase, V�volume frac-tion of the phase with Vc �1, Vc � the volume ofcomposite.
As the bond is perfect
�c � �m ��f
from equation (36.1)
Ec�c �Em�cVm �Ef �cVf (36.2)Ec �EmVm �EfVf
Ec �E11 �Em(1�Vf)�EfVf. (36.3)
Macro-analysis of stress distribution in a fibre/matrix composite
359
This equation is often referred to as a law of mix-tures equation.
36.2.2 Transverse stiffness
The same approach can be used to obtain thetransverse modulus of a unidirectional laminaE22.
The applied load transverse to the fibres actsequally on the fibre and matrix and therefore
�f � �m
�f � �22/Ef and �m � �22/Em (36.4)�22�Vf �f �Vm�m (36.5)
Substituting equation (36.4) into equation (36.5)
�22 �Vf �22/Ef �Vm�22/Em (36.6)
Substituting
�22 �E22�22 into equation (36.6)E22 �EfEm/[Ef(1�Vf)�EmVf] (36.7)
Equation (36.7) predicts E22 with reasonableagreement when compared with experimentalresults. The equation (36.8) has been proposedand takes account of Poisson contraction effects.
E22 �E#mEf /[Ef(1�Vf)�VfE#m] (36.8)
FIGURE 36.2 Orientation dependence of themodulus of elasticity of a fibre/matrix composite.
where
E#m �Em/(1�v2m).
36.3 Elastic properties of in-planerandom long-fibre laminateLaminae manufactured from long randomly ori-entated fibres in a polymer matrix are, on amicroscopic scale, isotropic in the plane of thelamina. The general expression (the proof is givenin Hollaway (1989)) for the elastic modulus oflaminae consisting of long fibres is:
1/E� � (1/E11)(cos4�)� (1/E22)(sin4�)� [(1/G12)� (2v12/E11)] cos2� sin2� (36.9)
where � �angle defining the direction of requiredstiffness. Figure 36.2 shows the relationship of Eo
when angle � varies between 0° and 90°.It can be seen then that laminae can be made
with predetermined fibre orientation distributionso that elastic and other mechanical propertiescan be designed to meet specific needs.
36.4 Macro-analysis of stressdistribution in a fibre/matrixcompositeThe behaviour of composites reinforced withfibres of finite length l cannot be described by theabove equations. As the aspect ratio, which isdefined by the fibre length divided by the fibrediameter (l/d), decreases, the effect of fibre lengthbecomes more significant.
When a composite containing uniaxiallyaligned discontinuous fibres is stressed in tensionparallel to the fibre direction there is a portion atthe end of each finite fibre length, and in thesurrounding matrix, where the stress and strainfields are modified by the discontinuity. The effi-ciency of the fibre to stiffen and to reinforce thematrix decreases as the fibre length decreases.The critical transfer length over which the fibrestress is decreased from the maximum value,under a given lamina load, to zero at the end ofthe fibre is referred to as half the critical length
Analysis of the behaviour of polymer composites
360
of the fibre. To achieve the maximum fibre stress,the fibre length must be equal to or greater than the critical value lc.
Figure 36.3 shows diagrammatically the defor-mation field around a discontinuous fibre embed-ded in a matrix and subjected to a tensile forcewhere the line of action of the force is parallel tothe fibre. Figure 36.4 shows a schematicrepresentation of a discontinuous fibre/matrixlamina subjected to an axial stress; the stress dis-tributions of the tensile and shear components areshown.
36.5 Elastic properties of short-fibrecomposite materialsAs discussed above the reinforcing efficiency ofshort fibres is less than that for long fibres. Inaddition the orientation of short fibres in alamina is random and therefore the lamina can beassumed to be isotropic on a macro scale.
The rule of mixtures as given in equation(36.3) can be modified by the inclusion of a fibreorientation distribution factor , thus the com-posite modulus of elasticity is given by
Ec �E11 � EfVf �EmVm. (36.10)
Values of have been calculated by Krenchel(1964) for different fibre orientations:
�0.375 for a randomly orientated fibre array�1.0 for unidirectional laminae when tested
parallel to the fibre�0 for unidirectional laminae when tested per-
pendicularly to the fibre�0.5 for a bidirectional fibre array.
The following sections are important for a com-plete understanding of composite theory and youmay find that you have to go back to study themin more detail after the first reading.
36.6 Laminate theorySections 36.2, 36.3, 36.4 and 36.5 discussed theindividual lamina properties; this section concen-trates upon laminates which are formed whentwo or more laminae are combined to produce acomposite. It describes the methods used to calcu-late the elastic properties of the laminates, andbriefly introduces the elasticity theory.
The stresses at a point in a body are generallyrepresented by stress components acting on thesurface of a cube; Figure 36.5 shows the threenormal and the three shear stresses. The notation
FIGURE 36.3 Diagrammatic representation of the deformation field around a discontinuous fibre embedded ina matrix.
Laminate theory
361
FIGURE 36.4 Schematic representation of an aligned, discontinuous fibre composite subjected to an axialstress; stress distribution at failure is shown.
FIGURE 36.5 Components of stress acting on an elemental unit cube.
Analysis of the behaviour of polymer composites
362
employed here is such that the first subscriptrefers to the plane upon which the stress acts andthe second subscript is the coordinate direction inwhich the stress acts; the equivalent strains havethe same notation. As laminae are assumed to besufficiently thin the through thickness stresses arezero. Thus �33 ��31 � �13 �0 and plane stressconditions hold.
36.7 Isotropic laminaFor homogeneous isotropic lamina the stress–strain relationship for a lamina and laminates is:
�11 � (E/(1�v2))(�11 �v�22)�22 � (E/(1�v2))(�22 �v�11)�12 � (E/2(1�v))(�12) (36.11a)
or in matrix form
� � �� � � �[�] � [Q][�] (36.11b)
where
Q11 �E/(1�v2)�Q22
Q12 �vE/(1�v2)�Q21
Q33 �E/2(1�v)�G.
There are two independent constants in theseequations; these are E and � and this indicatesisotropic material properties.
The corresponding set of equations to thosein equation (36.11b), which relate strains tostresses, are:
� � �� �� � (36.12a)
[�] � [S][�] (36.12b)
where
S11�1/E�S22
S33�1/GS12� �v/E.
36.8 Orthotropic laminaThe orthotropic lamina can be assumed to beisotropic in plane 1, as shown in Figure 36.5 (i.e.the plane normal to the axis direction 1), as theproperties are independent of direction in thatplane. The stress–strain relationship for anorthotropic lamina is
�11 � [E11/(1�v12v21)][�11 �v21�22]�22 � [E22/(1�v12v21)][�22 �v12�11]�12 �G12�12 (36.13a)
or in matrix form
� � �� �� �[�] � [Q][�] (36.13b)
where
Q11 �E11/(1�v12v21); Q22 �E22/(1�v12v21)Q12 �v21E11/(1�v12v21); Q21 �v12E22/(1�v12v21)Q33 �G12.
As the Q matrix is symmetric we havev21E11 �v12E22. The Poisson’s ratio v12 refers tothe strains produced in direction 2 when thelamina is loaded in direction 1.
There are four independent constants in theseequations; these are E11, E22, �12 and �21 and thisindicates orthotropic material properties.
From the above equation it can be seen that theshear stress �12 is independent of the elastic prop-erties E11, E22, �12 and �21, and therefore no cou-pling between tensile and shear strains takes place.
The corresponding set of equations to those inequation (36.13b) which relate strains to stressesare:
� � �� �� � (36.14a)
[�] � [S][�] (36.14b)
where
S11 �1/E11; S33 �1/G12
�11
�22
�12
00
S33
S12
S22
0
S11
S21
0
�11
�22
�12
�11
�22
�12
00
Q33
Q12
Q22
0
Q11
Q21
0
�11
�22
�12
�11
�22
�12
00
S33
S12
S22
0
S11
S21
0
�11
�22
�12
�11
�22
�12
00
Q33
Q12
Q22
0
Q11
Q21
0
�11
�22
�12
Orthotropic lamina
363
S22 �1/E22; S12 � �v21/E22 � �v12/E11.
If the line of application of the load is alongsome axis other than the principal one, then thelamina principal axes do not coincide with thereference axes x, y of the load and the formeraxes must be transformed to the reference axes.Figure 36.6 illustrates the orientation of theorthotropic lamina about the reference axis.
Hollaway (1989) shows that the stress–strainrelationship in the (x, y) coordinate system atan angle � to the principal material directionbecomes
� � �� �� � (36.15a)
or
� � � [Q� ]� � (36.15b)
where
Q� 11 � Q11m4 �Q22n4 �2(Q12
�2Q33)n2m2
Q� 12 � Q� 21 � (Q11 �Q22 �4Q33)n2m2
�Q12(n4 �m4)
Q� 13 � Q� 31 � (Q11 �Q12 �2Q33)nm3
� (Q12 �Q22 �2Q33)n3mQ� 22 � Q11n4 �Q22m4 �2(Q12
�2Q33)n2m2
Q� 23 � Q� 32 � (Q11 �Q12 �2Q33)n3m� (Q12 �Q22 �2Q33)nm3
Q� 33 � (Q11 �Q22 �2Q12
�2Q33)n2m2 �Q33(n4 �m4)
where Q11, Q22, Q12, Q21 and Q33 have beendefined in equation (36.13b) and m�cos�,n� sin� and the equivalent expression for straincomponents in the reference axis x, y in terms ofthe stress components in that axis become:
� � � [S�]� � (36.16)
where [S�] is a 3�3 compliance matrix where thecomponents are:
S�11 � S11m4 �S22n4 � (2S12 �S33)n2m2
S�12 � S�21 �S12(n4 �m4)� (S11 �S22
�S33)n2m2
S�13 � S�31 �2(S11 �2S12 �S33)nm3 � (2S22
�S12 �S33)n3mS�23 � S32 � (2S11 �2S12 �S33)n3m� (2S22
�2S12 �S33)nm3
�xx
�yy
�xy
�xx
�yy
�xy
�xx
�yy
�xy
�xx
�yy
�xy
�xx
�yy
�xy
Q� 13
Q� 23
Q� 33
Q� 12
Q� 22
Q� 32
Q� 11
Q� 21
QQ� 31
�xx
�yy
�xy
FIGURE 36.6 Orientation of orthotropic lamina about reference axis.
Analysis of the behaviour of polymer composites
364
S�22 � S11n4 �S22m4 � (2S12 �S33)n2m2
S�33 � 2(2S11 �2S22 �4S12 �S33)n2m2
�S33(n4 �m4)
where S11, S22, S12, S21 and S33 have been definedin equation (36.14b).
36.9 The strength characteristicsand failure criteria of compositelaminaeIn the two preceding sections, the stiffness relation-ships in terms of stress and strain were presentedfor isotropic and orthotropic materials. It is nownecessary to have an understanding of the ultimatestrengths of the laminae to enable a complete char-acterisation of the composite material to be made.The stress–strain relationship stated in the previoussections described the actual stresses occurring atany point in a lamina, and the strength character-istics may be considered as describing the allow-able stress at any point.
When the formulation of the stiffnesscharacteristics of the lamina was developed,properties in both tension and compression wereassumed. However, the ultimate strength behavi-our of composite systems may be different intension and compression and the characteristicsof the failure mode will be highly dependent uponthe component materials. Therefore, a systematicdevelopment of the strengths of these materials isnot possible; consequently a series of failure cri-teria for composite materials will be given.
36.9.1 Strength theories for isotropiclaminae
In isotropic materials both normal and shearfailure can occur, but it is usual to equate the com-bined stress situation to the experimentally deter-mined uniaxial tension or compression value.When a tensile load is applied to a specimen in auniaxial test it is possible for failure in the speci-men to be initiated by either an ultimate tensilestress or a shear stress, because a tensile stress of s(the maximum principal stress in this type of test)on the specimen produces a maximum shear value
of s/2. Consequently the failure theories are relatedto the applied tensile or compressive stress thatcauses failure, irrespective of whether it was anormal or a shear stress failure.
Many theories and hypotheses have beendeveloped to predict the failure surface forcomposite materials under tensile loads and prob-ably the best known theories which have beenused to predict failure and which are discussed inHolmes and Just (1983) are:
Maximum principal stress theory
�xx � �t* (36.17)
where
�xx �maximum principal stress�t* � failure stress in a uniaxial tensile test
or ��zz � �c*
�zz �minimum principal stress�c* � failure stress in a uniaxial compressive test.
Figure 36.7 shows the principal stresses actingon an element of material.
Maximum principal strain theory
�xx � �t* (36.18)
where
�xx�maximum principal tensile strain�t* � tensile strain at failure
in terms of stress
(�xx �v�yy �v�zz)/E� �t*/E
or
(�xx �v�yy �v�zz)� �t*
v�Poisson’s ratio.
Similarly
�zz ��c*�zz �minimum principal strain�c* �compression strain at failure
or �zz �v(�xx � �yy)��c*.
The strength characteristics and failure criteria of composite laminae
365
Both the above theories assume failure to bedue to normal stresses and ignore any shear stresspresent. Consequently the theories are relevant tothe failure of brittle materials under tension.
The total strain energy theory
The above theories express the failure criterion aseither limiting stress or limiting strain; the totalstrain energy theory attempts to combine thesetwo theories. The development of the theory,which is based upon strain energy principles, hasbeen discussed in Hollaway (1989) and only thefinal solution will be given here.
The lamina theory gives the solution as:
�*2 ��2xx � �2
yy �2�xx�yyv. (36.19)
The theory applies particularly to brittle mater-ials where the ultimate tensile stress is less thanthe ultimate shear stress.
Deviation strain energy theory
This theory is known as the von Mises criterionand in it the principal stresses �xx, �yy and �zz canbe expressed as the sum of two components,namely the hydrostatic stresses �xx, �yy and �zz
which causes only a change in volume and thedeviation stress which causes distortion of thebody. The system is shown in Figure 36.8.
The hydrostatic stress components produceequal strains in magnitude and are consistent inthe three directions and therefore produce equalstrain in these directions. The system, therefore,undergoes change in volume but not change inshape. The stress deviation system will cause thebody to undergo changes in shape but not involume.
Again the theory has been developed in Holla-way (1989) and will not be repeated here; thelamina theory gives the solution as:
�2xx ��2
yy � �xx�yy � �*2 (36.20)
FIGURE 36.7 Isotropic and orthotropic materials under normal stress: (a) element under three principalstresses �xx ��yy ��zz; (b) orthotropic material under normal stress.
Analysis of the behaviour of polymer composites
366
The above failure criterion is most relevant toductile materials. It is not obvious which of theabove failure criteria is most relevant to compo-sites as the fibre volume fraction and orientationof the fibres in the polymer will influence theirstrength and ductility properties. However, thelast theory has been applied to quasi-isotropiccomposites with some success.
Strength theories of orthotropic laminae
The theories based upon the strength character-istics of orthotropic materials are considerablymore complicated than those for the isotropicones. As with these latter materials the strengthhypothesis is based upon simple fundamentaltests, but because orthotropic materials havedifferent strengths in different directions a moreintensive set of data is required than for isotropicmaterials. Three uniaxial tests are required, onein each of the principal axis directions to deter-mine the three moduli of elasticity, Poisson’s ratioand the strength characteristics; tests in thesedirections will eliminate any coupling effects ofshearing and normal strains which would occur ifthe laminate were tested in any other direction.The shearing strengths with respect to the prin-cipal directions must be determined fromindependent experiments.
Figure 36.9 shows schematically the criticalstress values in the principal material axes. Fororthotropic materials the stress condition at apoint is resolved into its normal and shearingcomponents relative to the principal material axisat the point. Consequently the failure criteria inthese materials become functions of the basicnormal and shearing strengths described forisotropic materials.
Maximum stress theory
The maximum stress theory of failure assumesthat failure occurs when the stresses in the prin-cipal material axes reach a critical value. Thethree possible modes of failure are:
�11 � �*11 the ultimate tensile orcompressive stress indirection 1.
�22 � �*22 the ultimate tensile orcompressive stress in (36.21)direction 2.
�12 � �*12 the ultimate shear stressacting in plane 1 indirection 2.
If the load were applied to the lamina at anangle � to the principal axis direction shown inFigure 36.6 then by transformation
FIGURE 36.8 Volume and deviation stress system: (a) principal stress; (b) volume stress system; (c) stress devi-ation (or distortion) system.
The strength characteristics and failure criteria of composite laminae
367
�11 � �xx cos2���� cos2��22 � �xx sin2���� sin2� (36.22)�12 � ��xx sin�cos����� sin�cos�
The failure strength produced by the maximumstress theory would depend upon the relativevalues of s11, s22 and s12 and would therefore bethe smallest value of the following:
�� ��*11/cos2��� ��*22/sin2� (36.23)�� ��*12/sin�cos�
Maximum strain theory
The maximum strain theory of failure assumesthat failure occurs when the strains in the prin-cipal material axes reach a critical value. Hereagain there are three possible modes of failure:
�11 � �*11
�22 � �*22 (36.24)�12 � �*12
Where �*11 is the maximum tensile or compres-sive strain in direction 1, �*22 is the maximumtensile or compressive strain in direction 2 and�*12 is the maximum shear strain on plane 1 indirection 2.
Tsai-Hill energy theory
The Tsai-Hill criterion is based upon the vonMises failure criterion which was originally
applied to homogeneous isotropic bodies. lt wasthen modified by Hill to suit anisotropic bodies,and finally applied to composite materials byTsai.
Hollaway (1989) has discussed the derivationof the equation which describes the failure enve-lope and this may be expressed as:
� � � �1. (36.25)
For most composite materials �*11 $ �22; con-sequently, the second term of equation (36.25) isnegligible and this equation becomes:
� � �1. (36.26)
Equations (36.25) and (36.26) only apply toorthotropic laminae under in-plane stress con-ditions.
To enable a prediction to be made of thefailure strength in direction � to the principalaxes (Figure 36.6) on unidirectional laminae,equations (36.25) and (36.23) can be combinedto give:
�xx � �� � � � � �sin2�cos2�
� ��1/2(36.27)
Hull (1992) has stated that when equation(36.27) has been fitted to the results of experimen-tal tests on carbon fibre-epoxy resin laminates the
sin4���*22
2
1��*12
1
1��*12
2
cos4��
�*121
�212
��*12
2
�222
��*22
2
�211
��*12
1
�212
��*12
2
�222
��*22
2
�11�22�
�*121
�211
��*12
1
FIGURE 36.9 Critical stress values in the principal material axes.
Analysis of the behaviour of polymer composites
368
predicted values are much better than for themaximum stress theory, equation (36.21).
Finally, Figure 36.10 shows a laminate madefrom three lamina. Providing lamina 1 and 3 havethe same thickness, the laminate would bedescribed as symmetric; lamina 2 could have anyvalue of thickness. If, however, the thickness oflamina 1 and 3 were different, the laminatewould be described as non-symmetric. Under athermal and mechanical load, coupling forces areintroduced into a non-symmetric laminatebecause of the different mechanical properties ofthe individual lamina. For this reason it iscommon practice in many applications to usesymmetric laminates which are not subjected tothis type of coupling.
36.10 ReferencesHollaway, L. (1989) Design of composites. In Design
with Advanced Composite Materials (ed. L. Phillips),The Design Council, London.
Holmes, M. and Just, D.J. (1983) GRP in StructuralEngineering, Applied Science Publishers, Londonand New York.
Hull, D. (1992) An Introduction to Composite Mater-ials, Cambridge University Press, Cambridge.
Krenchel, H. (1964) Fibre Reinforcement, AkademiskForlag, Copenhagen.
FIGURE 36.10 Laminate arrangements.
369
Chapter thirty-seven
Manufacturingtechniques forpolymercomposites
37.1 Manufacture of fibre-reinforced thermosetting composites37.2 Manufacture of fibre-reinforced thermoplastic composites37.3 References
The two parts of this chapter concentrate uponthe manufacturing techniques for fibre-reinforcedthermosetting and thermoplastic polymer com-posites respectively.
37.1 Manufacture of fibre-reinforcedthermosetting compositesFibre-reinforced thermosets are manufactured inthree ways and examples of each are given here:
1. manually – hand lay-up, spray-up, pressurebag and autoclave moulding;
2. semi-automatically – cold pressing, compres-sion moulding and resin injection;
3. automatically – pultrusion, filament windingand injection moulding.
Processes which use only one mould are knownas open-mould techniques; the surface of theproduct in contact with the mould is able toreproduce the surface of the mould completely.Processes in which the product is formed within aclosed space by two moulds are known as closed-
mould techniques. Figure 37.1 shows a schematicdiagram of the various fabrication processes.
37.1.1 Manual processes
Manual processes are all open-mould processes,involving contact moulding, resin transfer mould-ing and resin infusion processes.
The hand lay-up technique
This is the simplest and most common techniquefor producing fibre-reinforced polymer com-ponents (Figure 37.2 (a)). It is ideally suited to the
FIGURE 37.1 Schematic diagram of the fabricationprocesses for thermosetting polymer composites.
Manufacturing techniques for polymer composites
370
production of a small number of similar com-ponents such as fibre-reinforced polymer infillpanels which could be used in civil engineering.Most materials are suitable for mould making,but the most common are glass fibre reinforcedpolymers (GFRP). A master pattern is preparedfrom which GFRP moulds may be made. Arelease agent is used to prevent bonding to themoulds during manufacture. It is necessary toprotect fibres from exposure to the atmosphere toprevent fracture and this protection is achievedby applying a resin-rich area, known as a gelcoat, on the exposed surface of the composite.
The function of the gel coat is:
1. to protect fibres from external influences,especially moisture penetration to the inter-face of the fibre and matrix with consequentbreakdown of the interface bond;
2. to provide a smooth finish and to reproduceprecisely the surface texture of the mould.
The thickness of the gel coat is generally about0.35mm. Sometimes a surface tissue mat is usedto reinforce the gel coat.
After the gel coat has become tacky but firm, aliberal coat of resin is brushed over it and the firstlayer of glass reinforcement is placed in positionand consolidated with brush and roller. The glassfibre may be in the form of chopped strand mat
FIGURE 37.2 (a) Hand lay-up moulding method;(b) spray-up moulding technique.
or woven fabric, precut to the correct size. Layersof resin and reinforcement are then applied untilthe required thickness of composite is reached.
Spray-up technique
The spray-up technique, shown in Figure 37.2 (b),is less labour intensive than the hand lay-upmethod. It involves the simultaneous depositionof chopped glass fibre roving and polymer ontothe mould with a spray gun. The roving is fedthrough a chopping unit and projected into theresin stream. The glass resin mixture is thenrolled by a split washer roller to remove any air.
The technique requires considerable operatorskill to control the thickness of the composite andto maintain a consistent glass/polymer ratio. Afterthe initial polymerisation of the composite, andwhen it has been demoulded, the unit must becured usually by heating for eight hours at 60°C.
As the tooling costs for the above two processesare low, the designer has considerable versatilityfrom the point of view of shape and form.
Pressure bag techniques
Pressure is exerted on the open face of the mould-ing to enable:
1. a greater compaction of the composite (thevoids are reduced to zero);
2. a higher fibre volume ratio;3. a higher quality of surface finish on this face.
These techniques are ideal for the manufactureof low-cost open-mould products such as struc-tural panels without sophisticated mechanisation.
Vacuum bag
A contact moulding is produced by the hand lay-up method, but before curing commences a rubbermembrane is placed over the composite compo-nent and all joints between the membrane andmould are sealed and a vacuum is applied betweenthe bag and composite (Figure 37.3(a)). The pres-sure of about one atmosphere applied to thesurface of the moulding forces out air and excess
Manufacture of fibre-reinforced thermosetting composites
371
polymer; glass/polymer weight ratios of up to 55per cent are possible. A protective sheet of cello-phane is used between the rubber and composite.
Pressure bag
This technique is similar to the vacuum bag exceptthat a pressure is applied to the open surface(Figure 37.3(b)). The contact moulding is pro-duced and before curing commences a rubber bagwith the cellophane sheet is sealed to the mouldwith a plate and pressurised up to about threeatmospheres. Because of the higher pressure com-pared with that for the vacuum bag, a greaterfibre/matrix weight ratio is obtained. This valuecan be increased to about 65 per cent. Superiorcomposites can be manufactured by this techniquebut more equipment is required than for thevacuum bag, with a consequent increase in cost.
Autoclave
The autoclave is a modification of the pressurebag method; pressures of up to 6 atmospheres aredeveloped within the autoclave and the systemproduces a high-quality composite with afibre/matrix weight ratio of up to 70 per cent.The cost of production also increases.
In addition to manufacturing structures, madecompletely from polymer composites, by thehand lay-up or spray-up techniques, commer-cially available procedures to retrofit compositematerials to existing structures, to improve theirtensile and shear strengths, do exist.
Hand lay-up process for retrofitting
A commercially available method of retrofitting isthe REPLARK process (REPLARK is a tradename used by Sumotomo Corporation, EuropePlc) which is utilised, in industry, to strengthenstructures in flexure and shear by retrofitting thematerial to their tensile and shear faces. (In thesecases the surface of the structure forms the mouldfor the composite; the technique is more fullydescribed in Chapter 39.) The fabrication tech-nique to form this material is basically a hand
FIGURE 37.3 Pressure bag techniques.
lay-up method in which the mould is usually thestructural unit onto which is retrofitted thepolymer composite. In addition, planar and non-planar composites can be manufactured indepen-dently and used as structural units.
A second method for retrofitting procedures isknown as the Dupont method which uses Kevlarfibres; it is marketed as a repair system for con-crete structures.
37.1.2 Semi-automated process forretrofitting
These methods are used for the manufacture of thecomposite material and to retrofit them, in situ,onto concrete, steel and cast-iron structures. Thesemi-automated resin infusion under flexibletooling (RIFT) process has been developed byDML, Devonport, Plymouth, to allow quality com-posites to be formed. In this process dry fibres arepreformed in a mould in the fabrication shop andthe required materials are attached to the preformbefore packaging. The preform is taken to site andis attached to the structure; a resin supply is thenchannelled to the prepreg. The prepreg and resinsupply is then enveloped in a vacuum baggingsystem. As the resin flows into the dry fibre
Manufacturing techniques for polymer composites
372
preform it develops both the composite materialand the adhesive bond between the CFRP and thestructure. The process provides high fibre volumefraction composites of the order of 55 per cent;these have high strength and stiffness values.
XXsys Technologies, Inc., San Diego, Californiahas developed a wrapping system for seismic retro-fitting to columns. The technology associated withthe technique is based upon the filament winding ofprepreg carbon fibre toes around the structural unitthus forming a carbon fibre jacket; currently, thestructural unit to be upgraded would be a column.The polymer is then cured by a controlled tempera-ture oven and can, if desired, be coated to matchthe existing structure.
37.1.3 Automated processes
Filament winding
Filament winding is a highly mechanised andsophisticated technique for the manufacture ofpressure vessels, pipes and rocket casings whenexceptionally high strengths are required. It is anopen mould-process in which continuous rein-forcement, usually rovings, is fed through a traversing bath of activated resin and is thenwound onto a rotating mandrel. If resin preim-pregnated reinforcement is used, it is passed over ahot roller until tacky and is then wound onto therotating mandrel. Figure 37.4 illustratesthe process and it is evident that the angle of the
helix is determined by the relative speeds of the traversing bath and the mandrel. After comple-tion of the initial polymerisation, the composite isremoved from the mandrel and cured; the compos-
FIGURE 37.4 Filament winding technique.
ite unit is then placed into an enclosure at 60°C foreight hours.
In the construction industry, filament windinghas been used to form high-pressure pipes andpressure vessels. Sewerage pipes have also beenmanufactured in this way.
The rest of the automated processes that aredescribed below are closed-mould processes, inwhich the product is formed within a closed-mould system, utilising pressure and sometimesheat rapidly to produce high-quality units. Becauseof the high capital equipment outlay, particularlyfor the manufacture of the metal moulds, it isessential that large production runs are performed.Because of the mechanisation of the system only asmall, skilled workforce is required. There are twotypes of process – matched die and pultrusion andmodified pull-winding.
Matched die
In cold press moulding, pressure is applied to twounheated matched metal moulds to disperse resinthroughout a prepared fibre fabric stack placed inthe mould; a release agent and a gel coat wouldbe applied to the mould surface before the fibrestack is placed in position. The activated polymeris poured onto the top of the mat and the mouldis then closed; the polymer spreads throughoutthe fabric under the mould pressure. Pressuresapplied may be as low as 100kPa and the heatgenerated during the exothermic polymerisationprocess warms the tools; this helps in the curingprocess but additional curing after demoulding isessential.
In hot press moulding, glass fibre reinforcementand a controlled quantity of hot curing catalysedresin are confined between heated, matched, pol-ished metal dies brought together under pressure(Figure 37.5). The pressures vary depending uponthe process but will normally be between 0.5 and15MPa and the mould temperatures will bebetween 120°C and 150°C. The heat ensuresrapid curing and so no subsequent curing isrequired. Both preform moulding and premixmoulding of the fibre and the polymer can beused in this process.
Manufacture of fibre-reinforced thermosetting composites
373
In preform moulding chopped rovings are pro-jected onto a rotating fine metal mesh screenshaped to the required dimensions; the fibres areheld on the screen by suction. The strands arebonded together by spraying the preform with aresinous binder in the form of a powder or anemulsion and the whole is transferred to an ovenat 150°C for two to three minutes, after whichtime the preform is ready for the press.
Sheet moulding compound (or prepreg) is apolyester-resin-based moulding material whichconsists of a mixture of chopped strand mat orchopped fibre resin, fillers, catalyst and pigment.It is produced and supplied in the form of a con-tinuous sheet wound into a roll and protected onboth sides by sheets of polythene film which areremoved before loading into the press. Anadvantage of this method is that no liquid resin isinvolved and the prepreg sheets can be preparedto the design size by cutting.
Dough moulding compound (or bulk mouldingcompound) contains a mixture of choppedstrands (20 per cent by weight) with resin, cata-lyst and pigment. As the compound flows readilyit may be formed into shape by compressiontransfer or injection and the pressure required toproduce a component is relatively low so thatlarge mouldings can readily be produced. Curingtakes about 2 minutes for moulding temperaturesin the region of 120–160°C although this willdepend upon the section thickness.
FIGURE 37.5 Hot press moulding technique.
Resin injection
Resin injection is a cold mould process using lowpressures of about 450kPa. The surfaces of themould are prepared with release agents and gelcoat before the glass fibre reinforcement is placedin position in the bottom mould and allowed toextend beyond the sides of the moulds. The uppermould is clamped in position to stops and theactivated resin is injected under pressure into themould cavity. Figure 37.6 shows the arrange-ment. It is possible to obtain a fibre/matrix ratioby weight of 65 per cent.
Pultrusion and modified pull-winding
The pultrusion technique consists of impregnat-ing continuous strands of a reinforcing materialwith a polymer and drawing them through a dieas shown in Figure 37.7. Thermosetting polymersare used in this process although research is beingundertaken currently to pultrude thermoplasticmaterials. Curing of the composite component isundertaken when the die is heated to about135°C. A glass content of between 60 and 80 percent by weight can be achieved. Compositesmanufactured by this method tend to be rein-forced mainly in the longitudinal direction withonly a relatively small percentage of fibres in thetransverse direction. A technique has beendeveloped (Shaw-Stewart, 1988) to ‘wind’ fibresin the transverse direction simultaneously withthe pultrusion operation. The process is known aspull-winding and gives the designer greater flexi-
FIGURE 37.6 The resin injection process.
Manufacturing techniques for polymer composites
374
bility in the production of composites, particu-larly those of circular cross-section.
The pultrusion technique is the process usedextensively in the reinforced plastics industry andis becoming an important technique in civilengineering to form structural units. The finishedpultrusion sections are generally straight and diescan be manufactured to give most geometricalshapes; the most common of these are I, L, Tee,and circular sections
37.2 Manufacture of fibre-reinforcedthermoplastic compositesReinforced thermoplastic composites can be man-ufactured by most of the thermoplastic processingtechniques such as extrusion, blow moulding andthermoforming of short-fibre reinforced thermo-plastics. However, the most important techniquefor industrial use is injection moulding. It is asimilar technique to the manufacture of unrein-forced thermoplastics but the melt viscosity ishigher in the reinforced polymer process and con-sequently the injection pressures are higher. Withall the techniques, production difficulties canoccur because the reinforced composite is stifferthan the unreinforced one. The cycle time is lessbut the increased stiffness can affect the ejectionfrom the mould so the mould design has to bemodified from that of the unreinforced polymermould.
One of the problems of thermoplastic compos-ites is that they use short fibres (typically0.2–0.4mm long) and consequently their fullstrength is not developed. Continuous fibre tapesand mats in the form of prepregs can help toovercome this. The best known examples of thesesystems are the aromatic polymer composites
FIGURE 37.7 The pultrusion technique.
(APC) and the glass-mat-reinforced thermoplasticcomposites (GMT). The systems use unidirec-tional carbon fibre in a matrix of polyethersul-phone (PES) or polyetheretherketone (PEEK). Thematerial for the APC comes in prepreg form ofunidirectional or 0°/90° fibre and for GMT in atape prepreg form. The composite is manufac-tured by the film stacking process with theprepregs arranged in the desired directions.
37.2.1 The film stacking process
This is a relatively new method in which theprepreg is made from cloth reinforcement and hasa polyethersulphone polymer content of about 15per cent by weight. The final volume fractions offibre and resin are obtained by adding matrix inthe form of film. The film stacking process, there-fore, consists of alternating layers of fibre impreg-nated with insufficient matrix, with polymer filmsof complementary mass to bring the overall lami-nate to the correct fibre volume ratios. Therequired stacked sequence is placed into one partof a split mould; the two half moulds are broughttogether and heat and pressure are applied.
This technique is used mainly for high techno-logy composites in the aerospace and space indus-tries.
37.3 ReferencesHollaway, L.C. and Head, P.R. (2001) Advanced
Polymer Composites and Polymers in the CivilInfrastructure, Elsevier Science, Oxford.
Shaw-Stewart, D. (1988) Pullwinding. Proceedings ofthe Second International Conference on AutomobileComposites 88, Noordwijkerhout, The Netherlands.
375
Chapter thirty-eight
Durability anddesign ofpolymercomposites
38.1 Temperature38.2 Fire38.3 Moisture38.4 Solution and solvent action38.5 Weather38.6 Design with composites38.7 References
Polymer composites change with time and themost significant influencing factors are:
1. elevated temperatures;2. fire;3. moisture, particularly when immersed for
long periods;4. adverse chemical environments;5. natural weathering particularly when exposed
to the sun’s ultra-violet radiation.
38.1 TemperatureAs we have seen in the preceding chapter, temper-ature plays an important part in the manufactureof GFRP composites which are normally cured(and post-cured) with heat to a state of chemicaland physical stability. A certain amount ofshrinkage takes place during the curing process,due to the polymer chains being drawn together.
On occasions post-curing of composites at ele-vated temperatures presents practical difficultiesfor large structural composites and a gradualchange in physical and dimensional propertiesmay be expected for a period if a laminate isimperfectly cured. However, this effect is not nor-mally significant unless there is excessive under-cure. This situation could occur if, for instance,immediately after manufacture the units wereplaced in the fabricator’s yard because there wasno room in the fabrication shop. In this situationonly partial curing would have occurred beforethe unit was exposed to environmental con-ditions.
Constantly fluctuating temperatures have agreater deleterious effect on GFRP. At a microscale, the difference in the coefficients of thermalexpansion of the glass and the resin may con-tribute to progressive debonding and weakeningof the materials, although the extensibility of theresin system will usually accommodate differen-tial movement.
When GFRP composites are exposed to hightemperatures a discoloration of the resin mayoccur; this is noticed by the composite becomingyellow. Both polyester and epoxy show this effectand the problem will be aggravated if flame retar-dants are added to the resin during manufacture
Durability and design of polymer composites
376
of the composite. In addition, as a result of theexposure to high temperatures, the composite willbecome brittle.
38.2 FireA composite material must meet the appropriatestandards of fire performance. It is usually pos-sible to select a resin system which will achievethe standard laid down in British Standard Speci-fication BS 476. The Building Regulations requirethat, depending upon their use, building com-ponents or structures should conform to givenstandards of fire safety. The fire tests by whichthese are measured fall into two categories:
1. reaction to fire – tests on materials;2. fire resistance – tests on structures.
The tests under these two headings are laiddown in BS 476: Parts 4–7, and in BS 476: Parts3–8 respectively.
Resins will probably contain fillers andcolourants which may affect fire properties. Somemineral fillers, such as specially treated calciumcarbonate, can improve the mechanical propertiesof composites made from polyester and epoxies.No more than 25 per cent of the filler should beused, although as an exposed aggregate in a deco-rative surface a greater proportion of the coarseparticles may be used. To enable flame-retardantproperties of the plastic to be improved, alu-minium trihydrate and antimony trioxide may beused as fillers for both lamination and gel-coatedresins, but the use of flame retardants can affectthe colour retention of the polymer; the pigmentwould have been added to produce a particularcolour in a structural component. Het-acid-basedresins can be used where flame-retardantcharacteristics are required.
38.3 MoisturePolymers which are cross-linked are not easilyhydrolysed but they are sufficiently hydrophilicfor water to be absorbed. The effect of thisuptake of water is to reduce the modulus of elas-ticity and strength. Wet strength can be 25 per
cent lower than that of the dry specimen. The wetstrength values are sometimes used in design toreflect the possible effects of water and weather.The long-term effects on these two properties seemto be greater for epoxy resins than for polyesters.The absorption of water by both the polyestersand epoxies leads to swelling of the laminate withan increase in thickness; the wet laminates can alsowarp if the structure is unbalanced.
Water can have a deleterious effect on glassfibre. The high strength of fibres is largely due tosurface imperfections but a small amount of solu-tion on the surface of the fibres can producesurface flaws which will reduce the overallstrength of the laminate. The alkali-resistant glassfibre has a better resistance to water than the E-glass fibre which is generally used for the manu-facture of GFRP.
On a short-term basis water absorption is asurface effect and, if the composite has a gel coat(see later in the section), this effect is of little con-sequence. On a long-term basis, however, immer-sion of the composite in water will allowpenetration through the thickness of the coatingand water-filled voids will occur at thefibre/polymer interface and may weaken thebond. High-quality materials and good control ofthe composite fabricating process can reduce theabove effects to a minimum. The glass fibres, asmentioned earlier, are treated with a size toprotect them after manufacture and when the in-service requirements demand it, a water-resistantsize should be used to minimise blistering.
The ends of fibres will be exposed if GFRP iscut during shaping of the component, and thefibres may suffer from a wicking action, butexperience shows that only a narrow strip at thecut edge will be affected by water penetration. If,however, the composite fibre volume fraction ishigh the effect of water may be significant andshould be investigated.
Work by the Building Research Establishment(1963) showed that wetting and drying of GFRPcomposites, from natural exposure to environ-mental conditions, is much less severe than con-tinuous exposure to water in producing changesat the glass/resin interface.
Weather
377
38.4 Solution and solvent actionA fully cured polyester resin exhibits good resis-tance to acidic and alkaline attack if selected anddesigned properly; resin manufacturers should beconsulted when choosing a resin to be utilised fora specific corrosive environment.
38.5 WeatherNatural weathering does cause some deteriora-tion of GFRP composites; sunlight degrades bothpolyester and epoxy resins. The first sign ofdegradation is a discoloration of the materialwhich develops into a breakdown of the surfaceof the composite. In addition there is a loss oflight transmission in translucent sheeting. Theultraviolet component of sunlight is largelyresponsible for this degradation; the short wave-length band at 330nm has most effect on poly-esters but longer wavelengths are also significant.To reduce this weathering problem with polyesterresin, ultraviolet absorbers and stabilisers areadded to the resin formulations during fabrica-tion of the composite. There is, however, littleimprovement in the stability of epoxy resins whenultraviolet stabilisers are added.
When pigments are placed in polyester resininfra-red and visible radiation can accelerate therate of degradation of the polymer by raising itstemperature; a rise of 10°C approximatelydoubles chemical reaction and hence degradation.Consequently, resin degradation proceeds morerapidly in hot than in more temperate climates.The rate of degradation of epoxy resins is similarto that of polyesters.
Weathering can affect the mechanical proper-ties of GFRP composites; surface debonding offibres as a result of degradation of the resin willreduce the load-bearing capacity. The deteriora-tion could be caused by solar effects or by wateron the surface of the composite. Because weather-ing is largely a surface effect, the thickness of thecomposite has a significant influence upon themechanical properties. It has been shown (Scholz,1978) that a 3mm thick GFRP laminate has areduction in flexural strength of between 12 per
cent and 20 per cent after 15 years’ exposure tonatural weathering, whereas it has been estimatedthat for a 10mm laminate a reduction of no morethan 3 per cent would take place after 50 years’exposure.
The above discussion has assumed a generalpurpose resin without any surface protection. If apolyester resin has been specifically designed bythe manufacturer to resist weathering, degrada-tion would not be so severe.
The three principal types of polyester used as alaminating resin are orthophthalic, isophthalicand het acid resins. The orthophthalic type is ageneral-purpose resin, the isophthalic one hassuperior weathering and chemical resistanceproperties and the het acid resin is used for flame-retardant purposes. Fillers and pigments may beused in resins, the former principally to improvemechanical properties and the latter for appear-ance and protective action. Fillers such as alu-minium trihydrate may be used to improveflame-retardant characteristics, but it is importantthat the correct amount of the fillers should beincorporated; if too high a proportion isemployed, adverse effects on weathering proper-ties may result. This is only applicable for com-posites exposed to the weather and internalapplications; where water or chemicals are not incontact with the polymer, the composites are notaffected. Ultraviolet stabilisers can be incorpo-rated into the resin at the time of fabrication anda gel coat surface coating can be applied to thecomposite for increased weather protection.
It is clear that the durability of a GFRP lami-nate depends upon the quality of the exposedsurface and a good in-service performance is mostcommonly achieved by the use of a surface coator gel coat of pigmented resin. This protects theglass fibre reinforcement from the action of mois-ture and the laminating resin from harmful radia-tion and from moisture. The polymer that is usedfor this purpose has been formulated to providegood weathering and chemical resistance proper-ties. The gel coat is generally applied in a uniformthickness of 0.5mm to the mould surface and canbe pigmented to give the required colour. Asurface tissue of glass fibre is often incorporated
Durability and design of polymer composites
378
into the gel coat to reinforce it and to enhance itsappearance, whilst maintaining the resin-richsurface.
It is important that, for the satisfactory designof composites, the resin system, the reinforcementand the structure of the laminate should beassessed properly. This may involve an evaluationprogramme to assess the design and manufactureof the system to ensure that the product gives andmaintains satisfactory performance throughoutits service life. As an aid, the British Plastics Fed-eration has produced guidance notes on the con-struction of glass-reinforced plastics claddingpanels, dealing mainly with the in-service require-ments.
38.6 Design with compositesDesigning with composites is an interactiveprocess between the designer and the productionengineer responsible for the manufacturing tech-nique. It is essential that a design methodology isselected and rigorously used, because many dif-ferent materials are on the market and they canbe affected by quality, the environment and themanufacturing process. It is also important thatthe designer recognises the product cost, becausethe constituents of composite materials (the fibreand the matrix) can vary significantly in price and the manufacturing process can range fromsimple compact moulded units cured at roomtemperature to sophisticated high-temperatureand pressure-cured composites.
The design process can be divided into fivemain phases:
1. the design brief and an estimation of cost;2. the structural, mechanical and in-service
details;3. the manufacturing processes and cost details;4. the material testing and specification informa-
tion;5. the quality control and structural testing
information.
The selection of design factors of safety is an
important aspect of the work; these factors arelikely to be covered in the relevant code of prac-tice. However, if the design is unique, it may benecessary for the designer/analyst to select them,bearing in mind the exactness of the calculations,the manufacturing processes, the in-serviceenvironment, the life of the product and theloading. The selection follows the pattern forother materials but, with the variation in proper-ties, due to the anisotropic nature and the differ-ent manufacturing techniques of the composites,a more involved calculation and a greater relianceupon the design factors will result.
Polymer composite structures can be manufac-tured from thin plate and shell laminated plies toform continuum systems or from pultruded or fila-ment wound tubes to form skeletal structures.The properties of these thin laminated plate ortube elements may be calculated in terms of lami-nate structures and ply thicknesses and propertiesby using laminate theory or by commerciallyavailable microcomputer programs such asEngineering Science Data Unit (1987a,b) andThink Composites Software (1987).
38.7 ReferencesBS 476: British Standards Institution, London.Building Research Establishment (1963) Internal
records, Building Research Establishment, Garston,UK.
Engineering Science Data Unit (1987a) Stiffnesses andProperties of Laminated Plates, ESDU 20–22, ESDUInternational, London.
Engineering Science Data Unit (1987b) Failure of Com-posite Laminates, ESDU 20–33, ESDU International,London.
Hollaway, L. (1993) Polymer Composites for Civil andStructural Engineering, Blackie Academic and Pro-fessional, London.
Hollaway, L.C. and Head, P.R. (2001) AdvancedPolymer Composites and Polymers in the CivilInfrastructure, Elsevier Science, Oxford.
Kim, D.-H. (1995) Composite Structures for Civil andArchitectural Engineering, E & FN Spon, London.
Scholz, D. (1978) Kunststoffe, 68, 556.Think Composites Software (1987) MIC-MAC, Think
Composites Software, Dayton.
379
Chapter thirty-nine
Uses of polymercomposites
39.1 Marine applications39.2 Applications in truck and automobile systems39.3 Aircraft, space and civil applications39.4 Pipes and tanks for chemicals39.5 Development of uses in civil engineering structures39.6 Composite bridges39.7 Retrofitting bonded composite plates to concrete beams39.8 Composite rebars39.9 References
The use of polymers and polymer composites for‘all composite’ construction in the civil engin-eering industry falls into three categories. Theseare:
1. non-load-bearing;2. semi-load-bearing;3. load-bearing.
As we have seen in Chapter 34, unreinforcedpolymers are used in non-load-bearing and semi-load-bearing applications, whilst fibre-reinforcedpolymers are used in load-bearing applications.Thermoplastic materials can be drawn into fibreswhich are then used in geotextiles.
Polymers and fibre-reinforced polymers offermany advantages over other materials, but themost appropriate resin must be selected for theparticular end use since every material does notpossess all the following characteristics:
1. high light transmission for glazing and light-ing fixtures;
2. infinite texture possibilities;3. minimum maintenance requirements;4. infinite design possibilities;
5. resistance to water and corrosion;6. high specific strength;7. high impact resistance.
The disadvantages that have impeded the utili-sation of polymers in construction are:
1. The cost of the materials is relatively high,although when their low density is consideredand the structure as a whole has beendesigned as a composite system (e.g. thereduced foundation size compared with whatwould be required for conventional material),then polymers are competitive.
2. The stiffness and strength of the polymer orthe composite is less than that of competitormaterials and therefore they must be used inconjunction with the latter.
3. The scratch resistance is poor.4. Ultraviolet light can attack the material unless
stabilisers are incorporated into the resin for-mulations.
5. Organic materials will burn, but fire-retardantadditives can be used with the polymers toretard or eliminate burning.
In this chapter we will discuss some typicalexamples of successful structural applications ofpolymer composites; some other applicationsoutside construction are first briefly discussedbecause they are particularly noteworthy indemonstrating how such composites have offeredunique solutions to a range of product demands.
Uses of polymer composites
380
39.1 Marine applicationsPolymer composites are now the dominant mater-ials for pleasure craft. These materials have beenparticularly effective in replacing wood becauseof the following factors:
1. Design: the moulds to obtain the hydro-dynamic and aesthetic forms of the craft canbe readily manufactured.
2. Manufacture: the operations of cutting,fitting, assembly and finishing are reduced oreliminated in composite constructions com-pared with those required for a wood con-struction.
3. Structure: continuous stiff and strong shellforms can be produced in a composite con-struction.
The Sandown Class Single Role Minehunterhas been built for the Royal Navy by VosperThorneycroft and features an advanced glass-reinforced plastic hull structure.
39.2 Applications in truck andautomobile systemsPolymer composites are used in the manufactureof certain sports car bodies and truck cabs.Strength, stiffness, toughness, corrosion resistanceand the high-quality finishes are the physical andmechanical properties that must be satisfied, buteconomics is the crucial consideration governingthe choice of composites over conventionalmaterials.
39.3 Aircraft, space and civilapplicationsThe high technology composites are increasinglybeing used for components in the aircraft indus-try. The fin of the European Airbus is a compo-nent made from a sandwich construction withcarbon fibre/epoxy resin face material. The West-land helicopter rotor blades are made fromcarbon fibre composite material. In space, thesolid-fuel rocket-motor cases house the propellantfor many missiles. These cases are fabricated in
the form of a cylinder, having dome ends madeby the filament winding process.
39.4 Pipes and tanks for chemicalsThe chemical industry uses polymer compositepipes and tanks for storage. Whilst the criticalconsideration is the corrosive resistance underextreme environmental conditions, the ease offabrication of complex tank shapes and the sim-plicity of connections makes the compositematerial cost-effective.
39.5 Development of uses in civilengineering structuresTwo sophisticated GFRP structures have played amajor role in the development of polymer com-posite materials for construction; these are thedome structure erected in 1968 in Benghazi,Libya, and the roof structure at Dubai Airport,built in 1972. The composite units for the latterwere designed and manufactured in the UnitedKingdom and shipped to Dubai.
During the 1970s and early 1980s, prestigiousbuildings were erected in this country, notablyMorpeth School, Mondial House (the GPOHeadquarters in London), Covent Garden FlowerMarket and the American Express Building inBrighton. Figure 39.1 shows a photograph ofMondial House situated on the bank of theThames. Because of the relatively low modulus ofelasticity of the material, these buildings wereerected as a composite system, with either steel orreinforced concrete units as the main structuralelements and the GFRP composite as the load-bearing infill panels.
In the mid 1970s, Lancashire County Councilmanufactured a classroom system, using onlyGFRP, by folding flat plates into a folded platesystem so that the structural shape provided thestiffness to the building. In the late 1980s, moreambitious structural elements were produced.Two examples are the Manchester City FootballClub grandstand, in which the construction is abarrel vault system using glass-reinforced poly-ester composite, and the dome at Sharjah Airport,
Retrofitting bonded composite plates to concrete beams
381
where the dome is manufactured in the form ofcurved panels which in the erected position drainto composite channels concealed under thepanels.
Hollaway (1993) has given further examples ofthe use of polymers and polymer composites inconstruction.
39.6 Composite bridgesThe development of bridges constructed entirelyout of polymer/fibre composites commenced withthe prototype footbridges in Europe and NorthAmerica in the late 1970s. The first GFRPhighway bridge is believed to be the 10 metresspan bridge constructed in Bulgaria in 1982/1983using the hand lay-up technique. The bridge wasbuilt rather like Ironbridge 200 years earlier, toshow what could be achieved with the material.The second all GFRP bridge to be built is theMiyun Bridge in Beijing, China.
There have been some polymer compositebridges built in this country just recently; the bestknown one is the Aberfeldy cable stayed foot-bridge over the river Tay in Scotland. The deck ofthe bridge was manufactured using a modularsystem known as the Advanced Composite Con-struction System (ACCS) manufactured by thepultrusion technique (described in Chapter 37).
FIGURE 39.1 Mondial House International tele-phone exchange on the River Thames.
The ACCS module consists of a number of inter-locking fibre-reinforced polymer composite unitswhich can be assembled into a large range of highperformance structures for use in construction.The tower of the bridge, which has a main spanof 63 metres and an overall length of 113 metres,was also fabricated from the ACCS plank. Thecable stays are made from aramid fibres. Figure39.2 (a) shows a section of the Maunsell plankfrom which the deck was constructed; the wholesystem was designed by Maunsell Structural Plas-tics. Figure 39.2 (b) shows a cross-section of tenACCS planks fabricated into a box beam fromwhich the single bascule lift bridge at Bonds Millin Gloucester was manufactured and is shown inFigure 39.3. There have been other all-compositebridges built in Europe and North Americaduring the latter part of the 1990s.
39.7 Retrofitting bonded compositeplates to concrete beamsIn recent years various research groups through-out the world have focused, not only upon theutilisation of polymer composites as civil engin-eering structural units or complete compositestructures, but also upon retrofitting the materialto concrete and steel structures to improve theirstrength. The rehabilitation topics consideredhave included the confinement of concretecolumns with composite sheets or plates bondedto them, the use of polymer composite plates forthe flexural and shear strengthening of beams andslabs, wrapping of reinforced concrete structuralelements for corrosion protection and the dura-bility of concrete repairs. Figure 39.4 shows theretrofitting of a pultruded plate to a concretebridge to upgrade it to take greater load capacity.
39.7.1 Flexural and shear upgrading
Reinforced concrete structures may for a varietyof reasons (e.g. marginal design/design errorscaused by inadequate factors of safety, use ofinferior materials, etc.) be found to be unsatisfac-tory. In service, increased safety requirements, achange in use or modification causing redistribu-
Uses of polymer composites
382
80
80
80 760
Dimension shown in mm (b)
(a)
2125
603
FIGURE 39.2 Cross section of the ACCS Structural Unit (The Maunsell Plank).
FIGURE 39.3 Bonds Mill Lift Bridge, Gloucester-shire (photograph by kind permission of MaunsellStructural Plastics, Beckenham, Kent).
tion of stresses, an increase in magnitude orintensity of the applied loads required to be sup-ported, or an upgrade of design standards mayrender all or part of a structure inadequate. Onall highway structures, corrosion of the internal
reinforcement is exacerbated by the applicationof de-icing salts.
These inadequacies may manifest themselves bypoor performance under service loading in theform of excessive deflections and material failure.When maintenance or local repair can restore thestructure to the required standard, there are twoalternatives, complete or partial demolition andrebuild or a programme of strengthening.
Strengthening can be carried out by severaltechniques to achieve the desired improvement.These include increasing the size of the defectivemember, replacing non-structural toppings, exter-nal post tensioning, etc. The development ofstructural adhesive has lead to the evolution of afurther method of structural repair in which ini-tially steel plates were externally bonded to thestructure but latterly carbon fibre compositeplates are being used due to several disadvantagesinherent in the use of steel plates.
The directional reinforcing fibres are positionedlongitudinally relative to the beam and the com-posite would have a high fibre volume fraction to
Retrofitting bonded composite plates to concrete beams
383
obtain maximum efficiency, thus allowing thecomposite to be tailor made to suit the requiredshape and specification. The resulting materialsare easy to handle and to transport, require littlefalsework on site and can be used in areas oflimited access and do not add significant deadloads to the structure after installation. Themethod of manufacture of the composite plateswould be either pultrusion, hand-lay REPLARKor semi-automated RIFT methods (described inChapter 37)
As with all bonding operations the adherendsmust be free of all dust, dirt and surface grease.Consequently, the concrete, or steel surface onto
FIGURE 39.4 Plate bonding on the soffit of abridge (photograph by kind permission of Sika Ltd,Welwyn Garden City, Herts, UK).
which the composite is to be bonded, would begrit blasted to roughen and clean the surface. Itwould then be air blasted to remove any looseparticles and wiped with acetone or equivalent toremove any grease before the bonding operation.The thickness of the carbon fibre composite platewould generally not be greater than about1.2mm. The total length, as delivered to site,would be of the order of 18 metres. It is possibleto roll the material into a cylinder of about 1.5metre diameter for transportation and forbonding the plate onto the beam in one opera-tion. Both flexural and shear upgrading can beundertaken using carbon fibre composites.
39.7.2 Wrapping technique to upgradestructural columns
It is already known that confinement of concreteenhances its durability and strength. Currently, toenhance reinforced concrete columns, additionallongitudinal steel bars and concrete are addedaround existing columns. Another method con-sists of placing a steel jacket around a column.The two methods are difficult to apply.
Confinement of concrete columns with polymercomposite strands or sheets of composite prepregshows many advantages in compression overother confinement methods. These include theadvantageous properties of the compositemethods such as high specific strength and stiff-ness and the relative ease of applying the compos-ite materials to construction site situations.
Composite wrapping systems have been usedthroughout the world on a number of bridges,mainly in seismic loading, predominately in Japanand the USA. The available composite systemsinclude epoxy with glass fibre, aramid fibre orcarbon fibre fabric materials. A column consistingof a hybrid of fibre/polymer composite and con-crete can deform much more under an extremestress state than a conventional material systembefore failure. Confinement of the concrete later-ally provides an order of magnitude in theimprovement of the ultimate compressive strain.The REPLARK and the XXsys (described inChapter 37) are the two main systems available
Uses of polymer composites
384
for site work. Figure 39.5 illustrates theREPLARK system applied to a column
39.8 Composite rebarsConcrete structures have traditionally been rein-forced with steel rebars which are durable in thealkaline environment of the concrete. However,concrete beams crack in the tensile region underload, thus allowing aggressive environments tocome into contact with the steel. The corrosion ofthe steel reinforcement in concrete is a majorconcern in some civil engineering constructionenvironments such as coastal, marine and chemi-cal plants. In order to overcome this problem,polymer composite rebars may be used in place ofthe steel rebars.
The most suitable fibres currently used for thereinforcement of concrete are glass, carbon and
FIGURE 39.5 The use of REPLARK to wrap acolumn with carbon fibre polymer composite (photo-graph by kind permission of Sumitomo Corporation(UK) London).
aramid. The reinforced polymer matrix rebarswould be manufactured by the pultrusion processusing thermosetting polymers such as epoxy,vinylester and polyester. The final choice willdepend upon their durability and cost. There areno limits on the size and shapes of pultruded barswhich could be manufactured. The bars and rodswould be manufactured in straight lengthsbut unlike steel they cannot be bent (thermo-plastics/fibre composites can be bent by heatingbut these composites have not been developedsufficiently yet to be used as rebars). Othermethods of developing the bond length of com-posite rebars have to be developed, such as lapsplicing onto it; a 90° hook of the same cross-section as the main reinforcement could be used.
The most suitable fibres for use as prestressedtendons for concrete are the glass, aramid andcarbon fibres. These materials are possiblereplacements for the high tensile strength steels.The first two fibre types have been utilised as pre-stressing tendons; the latter fibre is available butno actual applications are known. The three com-mercial systems which are currently available, allutilise parallel filaments but they differ in con-struction techniques. ‘Polystal’ is a pultrusion ofglass fibre in a resin matrix; ‘Parafil’ is a rope thatderives its strength from aramid yarns; ‘Arapree’is a pultrusion of aramid fibres.
39.9 ReferencesHollaway, L. (1993) Polymer and Polymer Composites
for Civil and Structural Engineering, Blackie Acade-mic and Professional, Glasgow.
Hollaway, L.C. and Head, P.R. (2001) AdvancedPolymer Composites and Polymers in the CivilInfrastructure, Elsevier Science, Oxford.
Section 2
Fibre-reinforcedcements andconcrete
386
Chapter forty
Properties offibres andmatrices
40.1 Physical properties40.2 Structure of the fibre–matrix interface
40.1 Physical propertiesThe performance of the composite is controlledmainly by the volume of the fibres, the physicalproperties of the fibres and the matrix, and thebond between the two. Values for bond strengthrarely exceed 4MPa and may be much less forsome polymer fibres. The bond strengths will alsochange with time and with storage conditionswhich may permit densification of the interfaceregion due to continuing hydration. Typicalranges for other physical properties of fibres andmatrices are shown in Table 40.1.
It is apparent from this table that the elonga-tions at break of all the fibres are two or threeorders of magnitude greater than the strain at failure of the matrix and hence the matrix will usually crack long before the fibre strength is approached. This fact is the reason for theemphasis on post-cracking performance in thetheoretical treatment. On the other hand, the modulus of elasticity of the fibre is generallyless than five times that of the matrix and this,combined with the low fibre volume fraction,means that the modulus of the composite is notgreatly different from that of the matrix.
The low modulus organic fibres are generally
subject to relatively high creep which means thatif they are used to support permanent highstresses in a cracked composite, considerableelongations or deflections may occur over aperiod of time. They are therefore more likely tobe used in situations where the matrix is expectedto be uncracked, but where transitory overloadssuch as handling stresses, impacts or wind loadsare significant.
Another problem with the low modulus fibresof circular cross-section is that they generallyhave large values of Poisson’s ratio and this, com-bined with their low moduli, means that ifstretched along their axis, they contract sidewaysmuch more than the other fibres. This leads to ahigh lateral tensile stress at the fibre–matrix inter-face which for smooth circular section fibres islikely to cause a short aligned fibre to debond andpull out. Devices such as woven meshes, fibril-lated fibres with loose hairs on the surface or fib-rillated networks may therefore be necessary togive efficient composites.
Even the high modulus short fibres may requiremechanical bonding to avoid pull out unless thespecific surface area is very large. Thus steel fibresare commonly produced with varying cross-sections or bent ends to provide anchorage andglass fibre bundles may be penetrated withcement hydration products to give a more effect-ive mechanical bond after a period of time.
The mortar and concrete matrices in Table
40.1 are differentiated mainly by particle size andstrain to failure. The maximum particle size ofthe matrix is important because it affects the fibredistribution and the quantity of fibres which canbe included in the composite. The average parti-cle size of cement paste before hydration isbetween 10 and 30 microns whereas mortar isconsidered to contain aggregate particles up to5mm maximum size. Concrete which is intendedto be used in conjunction with fibres should nothave particles greater than 20mm and preferably
not greater than 10mm otherwise uniform fibredistribution becomes difficult to achieve.
In order to avoid shrinkage and surface crazingproblems in finished products it is advisable to useat least 50 per cent by volume of inert mineralfiller, which may be aggregate or could includepulverised fuel ash, or limestone dust. However, ifthe inert filler consists of a large volume of coarseaggregate the volume of fibres which can beincluded will be limited which will, in turn, limitthe tensile strength and ductility of the composite.
Physical properties
387
TABLE 40.1 Typical properties of cement-based matrices and fibres
Material or Relative Diameter or Length Elastic Tensile Failure Volume infibre density thickness (mm) modulus strength strain (%) composite
(microns) (GPa) (MPa) (%)
Mortar matrix 1.8–2.0 300–5000 – 10–30 1–10 0.01–0.05 85–97
Concretematrix 1.8–2.4 10000–20000 – 20–40 1–4 0.01–0.02 97–99.9
Aromaticpolyamides(aramids) 1.45 10–15 5-continuous 70–130 2900 2–4 1–5
Asbestos 2.55 0.02–30 5–40 164 200–1800 2–3 5–15
Carbon 1.16–1.95 7–18 3-continuous 30–390 600–2700 0.5–2.4 3–5
Cellulose 1.5 20–120 0.5–5.0 10–50 300–1000 20 5–15
Glass 2.7 12.5 10–50 70 600–2500 3.6 3–7
Polyacrylonitride(PAN) 1.16 13–104 6 17–20 900–1000 8–11 2–10
Polyethylene pulp 0.91 1–20 1 – – – 3–7HDPE filament 0.96 900 3–5 5 200 – 2–4High modulus 0.96 20–50 continuous 10–30 �400 �4 5–10
Polypropylenemonofilament 0.91 20–100 5–20 4 – – 0.1–0.2Chopped film 0.91 20–100 5–50 5 300–500 10 0.1–1.0Continuous nets 0.91–0.93 20–100 continuous 5–15 300–500 10 5–10
Polyvinyl alcohol(PVA, PVOH) 1–3 3–8 2–6 12–40 700–1500 – 2–3
Steel 7.86 100–600 10–60 200 700–2000 3–5 0.3–2.0
Strength of the matrix is mainly affected by thefree water/cement ratio and this parameter alsohas a lesser effect on the modulus so that theproperties of the matrices can vary widely.
40.2 Structure of the fibre–matrixinterfaceThe strength, deformation and failure character-istics of the whole range of fibre-reinforcedcementitious materials is critically dependent onthe bond between fibre and cement, which itselfdepends on the microstructure of the interface.The interface consists of an initially water-filledtransition zone which does not develop the densemicrostructure typical of the bulk matrix andcontains a large volume of calcium hydroxidecrystals which deposit in large cavities. Threelayers are commonly observed in this zone, a thin(less than one micron) calcium hydroxide-richrather discontinuous layer directly in contact withthe fibre, a massive calcium hydroxide layer and aporous zone up to 40 microns from the surfaceconsisting of calcium silicate hydrate and someettringite. The porosity of these layers is affectedby W/C ratio, age and whether or not microsilica,metakaolin or other additives are used in themixture.
When a force is applied to the composite,particularly local to a crack, high shear stresses
exist at the fibre cement interface, but failure maybe initiated in the porous layer rather than at theinterface itself. It is partly the changing naturewith continuing hydration of this rather weakinterfacial zone which is used to explain theembrittlement of some fibre cements with time.
When considering the microstructure in theinterfacial zone, a distinction should be madebetween discrete monofilaments such as steel andelements of polypropylene nets and bundled fila-ments such as glass or asbestos. With monofila-ments the entire surface of the fibre can be indirect contact with the matrix and the perimeter(Pf) carrying the shear forces is known. Withbundled filaments only the external fibres initiallyhave direct contact with the matrix and theperimeter transferring shear stress can only beestimated. As time goes on the vacant spacesbetween filaments in a strand may slowly becomefilled with hydration products. An intermediatestage is when some central fibres in a bundle arefree to slide while the outside fibres are rigidlylocked into the matrix. The final stage is completepenetration of the bundle which may then showbrittle fracture.
Thus the accurate calculation of bond stress(�), which relies on a knowledge of the perimeterin contact with cement, is not possible with fibrebundles because this area is unknown andchanges with time.
Properties of fibres and matrices
388
Chapter forty-one
Structure andpost-crackingcompositetheory
41.1 Theoretical stress–strain curves in uniaxial tension41.2 Uniaxial tension – fracture mechanics approach41.3 Principles of fibre reinforcement in flexure41.4 References
As mentioned in the introduction to fibre com-posites, the main benefits of the inclusion of fibresin hardened cement-based products relate to thepost-cracking state, where the fibres bridging thecracks contribute to the increase in strength,failure strain and toughness of the composite.This type of reinforcing mechanism requires a dif-ferent theoretical approach from that described inChapter 36 although the principles covered inthat chapter adequately describe the pre-crackingbehaviour.
Another significant difference from fibre-rein-forced polymers is that the fibre volume in fine-grained cement-based products rarely exceeds 15per cent and is more commonly less than 10 percent. In concrete the fibre volume is usually lessthan 2 per cent, often with low modulus fibresbeing included, and it is for these reasons,together with the relatively stiff matrix, that littleemphasis is placed on increasing the elasticmodulus of the composite by the inclusion offibres.
41.1 Theoretical stress–strain curvesin uniaxial tension
41.1.1 Characteristic shapes ofstress–strain curves
Fibre-reinforced cements and concretes are gener-ally considered to be most useful when carryingbending or impact loads but unfortunately atheoretical analysis of the mechanics of reinforce-ment in these systems is very complex. A morefundamental and a more easily understood stresssystem is that of direct tension and a soundknowledge of the behaviour of fibres in such asystem provides a good background by which anengineer can judge the potential merits of a fibrecement composite for a given end use.
The three basic types of tensile stress–straincurve available to the engineer are shown inFigure 41.1. Curves B and C are based on theassumption that the stress in the composite isincreased at a constant rate.
For all three curves, the portion OX defines theelastic modulus of the uncracked composite (Ec).Curves A and C are for composites in which thereare insufficient fibres to carry the load in the com-posite after continuous cracks in the matrix propa-gate completely across the component. In curveA, after the cracks have formed at X the fibres
389
slowly pull out and absorb energy leading to atough but rather weak composite typified by steelfibre concrete or short random fibre polypropy-lene concrete. In curve C, which is representativeof asbestos-cement and of some cellulose fibre-based composites when dry, the relatively highcrack propagation stress leads to a sudden largerelease of energy and to an almost instantaneousfibre fracture or fibre pull-out at C. However,stable micro-cracks may exist well before point Cis reached. The mechanism of the reinforcementof cement by asbestos fibre is very complex andleads to a special case of a high tensile strength,rather brittle, composite. A full understanding ofthe fracture process has not yet been achieved butit appears to require a combination of the rein-forcing theory described for curve B with aknowledge of fracture mechanics as outlined inSection 41.2.
Curve B is typical of a composite in whichthere are sufficient fibres to maintain the load onthe composite when the matrix cracks. The hori-zontal portion X–Y is a result of multiple crack-ing at approximately constant stress and Y–Brepresents extension or pull-out of fibres up to
Structure and post-cracking composite theory
390
FIGURE 41.1 Theoretical tensile stress–straincurves for different fibre composites.
separation of the component into two parts atpoint B. Curve OXYB is typified by fine-grainedmaterials such as glass-reinforced cement or somecontinuous fibre composites such as polypropy-lene network reinforced cement.
As will be seen from the equations in thischapter, the specific value of the bond strength (�)does not affect the shape of the stress–strain curveOXYB in Figure 41.1, the length of the multiplecrack region (��c), nor the failure strain (�cu). Bondstrength does, however, affect crack spacing andcrack width which are significant parametersaffecting the performance of the composites inpractical situations and, thus, time and weatheringwill affect crack spacing and crack width.
Apart from the special case of curve OXC it isunusual for fibres to significantly increase thecracking stress (�c) of the cement-based compos-ite and therefore it is of value to consider theload-carrying ability of the fibres after matrixcracking has occurred. An important conceptwhich it is necessary to understand in this respectis the critical volume fraction of fibres.
41.1.2 Critical volume fraction (Vfcrit) inuniaxial tension
The critical fibre volume is the volume of fibreswhich after matrix cracking will carry the loadwhich the composite maintained before cracking.
This definition needs to be used with a littlecare because material which has less than thecritical volume of fibre in tension (curve OXA)may have considerably greater than the criticalfibre volume required for flexural strengthening.However, it is common practice to assume thatthe above definition of critical fibre volume refersonly to uniaxial tensile stresses.
For the simplest case of aligned fibres withfractional bond, let:
�mu � matrix cracking strain�c � composite cracking stressVfcrit � critical volume of fibres�fu � fibre strength or average pull-out stress of
fibre depending on whether fibres break orpull out at a crack.
Just before cracking
�c �Ec�mu (41.1)
After cracking, the whole stress is carried bythe fibres. Assume that there are just sufficientfibres to support this stress, i.e. fibrevolume�Vfcrit
�c � �fuVfcrit (41.2)
From (41.1) and (41.2)
Vfcrit � (41.3)
From (41.2)
Vfcrit � (41.4)
There are important points to note about equa-tions (41.3) and (41.4).
1. Ec, �mu and hence �c may vary with time andEc and �c will generally increase if the cementcontinues to hydrate. This implies that a com-posite which has just sufficient fibre volume toexceed Vfcrit at early ages may, after someyears, have less fibre than Vfcrit at that age andhence may suffer brittle fracture.
2. Vfcrit can be decreased by decreasing �c.3. Poor bond may reduce �fu by allowing fibre
pull-out at a fraction of the fibre strength.Additionally the orientation of the fibres willhave a large effect on Vfcrit because randomfibre orientation will reduce the number offibres across a crack surface compared withthe aligned case.
In the common practical case of short, ran-domly orientated fibres where the fibres pull outat a crack rather than break, the stress in the fibrecausing pull-out may be substantially less than �fu
and hence Vfcrit for such composites may be up toten times that required for continuous alignedfibres. In the equations in Section 41.1.3 it isassumed that the orientation and stress efficiencyfactors are 1.0 and hence Vf and �fu for the widevariety of composites in current use must bemodified by the use of efficiency factors to allowfor the effective volume of fibres in the direction
of stress. The effect of short random fibres onVfcrit may be calculated using the equations inSection 41.1.6.
41.1.3 Stress–strain curve, multiplecracking and ultimate strength
If the critical fibre volume for strengthening hasbeen reached then it is possible to achieve multiplecracking of the matrix. This is a desirable situationbecause it changes a basically brittle material witha single fracture surface and low energy require-ment to fracture, into a pseudo-ductile materialwhich can absorb transient minor overloads andshocks with little visible damage. The aim of thematerials engineer is often therefore to produce alarge number of cracks at as close a spacing aspossible so that the crack widths are very small(say �0.1mm). These cracks are almost invisibleto the naked eye in a rough concrete surface andthe small width reduces the rate at which aggres-sive materials can penetrate the matrix whencompared with commonly allowable widths inreinforced concrete of up to 0.3mm.
High bond strength helps to give a close crackspacing but it is also essential that the fibres de-bond sufficiently local to the crack to give ductil-ity which will absorb impacts.
The principles behind the calculation of thecomplete stress–strain curve, the crack spacingand the crack width for long aligned fibres for thesimplified case where the bond between the fibresand matrix is purely frictional and the matrix hasa well-defined single value breaking stress hasbeen given by Aveston et al. (1971) and Avestonet al. (1974). The following has been simplifiedfrom these publications.
Long fibres with frictional bond
The idealised stress–strain curve for a fibre-rein-forced brittle matrix composite is OXYB inFigure 41.1.
If the fibre diameter is not too small, the matrixwill fail at its normal failure strain (�mu) and thesubsequent behaviour will depend on whether the
�c��fu
Ec�mu�
�fu
Theoretical stress–strain curves in uniaxial tension
391
fibres can withstand the additional load withoutbreaking, i.e. whether �fuVf � �c.
If they can take this additional load, it will betransferred back into the matrix over a transferlength %# (Figure 41.2) and the matrix will even-tually be broken down into a series of blocks oflength between %# and 2%# with an averagespacing (C) of 1.364%#.
We can calculate %# from a simple balance ofthe load �muVm need to break a unit area of thematrix and the load carried by N fibres of indi-vidual cross-sectional area Af and perimeter Pf
across the same area after cracking. This load istransferred over a distance %# by the limitingmaximum shear stress �. i.e.
N�Vf /Af (41.5)
PfN�%#� �muVm (41.6)
or
%# � �V
Vm�f
· ��
�
mu� �
A
Pf
f� (41.7)
This stress distribution in the fibres and matrix(crack spacing 2%#) will then be as shown inFigure 41.2.
The additional stress (��f) on the fibres due tocracking of the matrix varies between �muVm/Vf atthe crack and zero at distance %# from the crackso that the average additional strain in the fibres,which is equal to the extension per unit length ofcomposite at constant stress �c, is given by:
��c � �1
2��mu · �
V
Vm
f
� · �E
1
f
� (41.8)
i.e.
��c � � (41.9)
where
� �EmVm/EfVf (41.10)
and the crack width ( ), bearing in mind that thematrix strain relaxes from �mu to �mu/2 will begiven by
�2%#���
2mu�� �
�
2mu�� (41.11)
��mu�
2
�muEmVm�
2EfVf
Structure and post-cracking composite theory
392
FIGURE 41.2 Strain distribution after cracking ofan aligned brittle matrix composite (Aveston et al.,1971). (Crown copyright reserved.)
or
� �mu(1��)%# (41.12)
For practical composites, the factor � mayrange from about 2 for asbestos-cement to over100 for some polymer-fibre composites, thusmaking a very wide range of properties availableto the engineer.
At the completion of cracking the blocks ofmatrix will all be less than the length (2%#) requiredto transfer their breaking load (�muVm) and sofurther increase in load on the composite results inthe fibres sliding relative to the matrix, and thetangent modulus becomes (EfVf) (see Figure 41.1).
In this condition, the load is supported entirelyby the fibres and the ultimate strength (�cu) isgiven by
�cu � �fuVf (41.13)
If the average crack spacing C�1.364 %# isused in the calculations instead of the 2%# crackspacing then parameters associated with curveOXYB in Figure 41.1 are:
Average crack spacing,
C�1.364�V
Vm
f
� ��
�
mu� �
A
Pf
f� (41.14)
Length X–Y,
��c �0.659��mu (41.15)
Average crack width between X and Y�W
W��mu ·0.9 (1� �)%# (41.16)
Composite failure strain
�cu � �fu �0.341��mu (41.17)
The relationship between the average crackwidth W at the end of multiple cracking and fibrevolume is shown in Figure 41.3 for the case inwhich � is much greater than unity. It is apparentthat, even using the most optimistic assumption,large variations in crack width are likely to occurfor small variations in fibre volumes at less than 2per cent by volume of fibre and volumes above 5per cent are desirable if a uniformly and invisiblycracked composite is to be achieved. Hence thefabrication techniques must be adjusted toproduce a uniformly dispersed fibre and prefer-ably a high (5–10 per cent) fibre volume if invis-ible cracking is required in the product.
41.1.4 Efficiency of fibre reinforcement
The efficiency of fibre reinforcement depends onfibre length and orientation. Different efficiencyfactors are required for pre-cracking perform-ance, as described in the chapter on polymercomposites, and post-cracking performance.Allowance has also to be made for staticfrictional resistance prior to sliding and dynamicfrictional resistance during sliding. A detailedtreatment of efficiency factors is beyond the scopeof this section but may be found in Laws (1971),Aveston (1974), Hannant (1978) and Bentur andMindess (1990). A simplified approach for spe-cific cases is given in paragraph 41.1.6.
41.1.5 Application of theory to realcomposites
Typical stress–strain curves for real compositesmay be obtained using the theory in the previoussections and four such curves for different fibretypes are shown in Figure 41.4, in which V#f is the
Theoretical stress–strain curves in uniaxial tension
393
FIGURE 41.3 Relationship between crack width(W) at the end of multiple cracking and fibre volume(Vf) for the case in which � is much greater than unity.
effective volume of fibre in the direction of stresscalculated from the total Vf using the appropriateefficiency factors.
The values for � shown in Figure 41.4 havebeen calculated from equation (41.10), and, whensubstituted in equation (41.15), they give an indi-cation of the possible range for real composites ofthe horizontal part ��c of Figure 41.1.
Figure 41.4 (b) shows that it is possible forglass-reinforced cement to have a typical exten-sion due to multiple cracking alone of abouttwelve times the matrix cracking strain, whereasasbestos cement (Figure 41.4 (a)) can only extendby about one times the matrix cracking strainbefore the fibres take over completely. Likewise, asimilar effective volume of continuous polypropy-lene to that of asbestos (Figure 41.4 (c)) couldincrease the strain due to multiple cracking aloneby fifty-one times the matrix cracking strain evenunder fairly rapid loading. This would probablybe exceeded for extended loading periods becausethe modulus of polypropylene is time dependent.
The lack of a horizontal portion in Figure
41.4 (a) is a serious deficiency in that it results inthe material having limited capability to absorbshock and accidental over-strains. It is possiblethat the asbestos fibres have such a small dia-meter (0.02–20�m) that matrix cracking is sup-pressed until successively higher strains arereached as described in Section 41.2 and finalfailure is by single fracture with some pull-out.
The strain to failure for the polypropylene con-tinuous fibre composite in Figure 41.4(c) is anorder of magnitude greater than for the other com-posites and hence the energy-absorbing capabilityunder overload conditions, which depends on thearea under the stress–strain curve, is likely to behigher. However, the ultimate strain may not bereached in practice due to excessive deformation.
Figure 41.4 (d) is typical of concretes contain-ing less than the critical volume of short choppedfibres, generally steel or polypropylene in randomthree-dimensional orientation. The fibres in thesecircumstances are generally sufficiently short topull-out, rather than break, when cracks occur inthe matrix and the ultimate load on the compos-ite is then controlled by the numbers of fibresacross a crack, the length/diameter ratio of thefibres and the bond strength.
The stress in the fibre at pull-out is often lessthan half the strength of the fibre and hencehigher bond strength, rather than high fibrestrength may be the most important requirementfor this type of composite.
Structure and post-cracking composite theory
394
FIGURE 41.4 Typical stress–strain curves for real composites: (a) asbestos fibres in cement paste; (b) two-dimensional random glass fibres in cement paste; (c) continuous aligned polypropylene fibres in cement paste; (d)short, random, chopped steel or polypropylene fibres in concrete.
41.1.6 Short random fibres which pull outrather than break
Factors affecting a realistic estimate of Vf(crit) andpost cracking strength are:
1. number of fibres across a crack;2. bond strength and fibre pull out load.
Number of fibres across a crack
The situation in a cracked composite may berepresented by Figure 41.5. For short randomfibres, such as steel or chopped fibrillatedpolypropylene, which are generally shorter thanthe critical length for fibre breakage, the fibresmostly pull-out across a crack. A realistic estimateof the load carried after cracking can therefore beobtained by multiplying the number of fibrescrossing a unit area of crack by the average pull-out force per fibre. For fibres which break beforepulling out, the situation is more complicated.
The appropriate number of fibres, N per unitarea, can be calculated as follows:
Af �cross-sectional area of a single fibre.
For fibres aligned in one direction:
N� (41.18)
For fibres random in two dimensions:
N� �
2� �
V
Af
f
� (41.19)
Vf�Af
Theoretical stress–strain curves in uniaxial tension
395
FIGURE 41.5 Change in fibre orientation at acrack. (Reprinted by permission of John Wiley & SonsLtd, from D.J. Hannant (1978).)
For fibres random in three dimensions:
N� �1
2� �
V
Af
f
� (41.20)
Bond strength and fibre pull-out force
If the composite failure is by fibre pull-out, it hasbeen shown that the mean fibre pull-out length is(l/4) (see Figure 41.6).
Provided that the average sliding friction bondstrength (�) is known and assuming that this doesnot vary with the angle of the fibre to the crackthen the average pull-out force per fibre (F) isgiven by:
F� �Pf l/4 (41.21)
where Pf � the individual fibre perimeter incontact with the cement.
The ultimate stress (�cu) sustained by a unitarea of composite after cracking is therefore givenby N�F, i.e.
�cu � (41.22)
and the average fibre stress at pull-out (�f) is
�f � ��
A
P
f
f� �
4
l� (41.23)
which for round fibres reduces to
�f � � · �d
l�
Substituting for N from equations (41.18),
N�Pfl�
4
FIGURE 41.6 Average pull-out force per fibre.(Reprinted by permission of John Wiley & Sons Ltd,from D.J. Hannant (1978).)
(41.19) and (41.20) in equation (41.22) gives thefollowing:
Aligned fibres in one direction
�cu �Vf� �4
l� · �
A
Pf
f
� (41.24)
Random two-dimensional
�cu � �
2�Vf� �
4
l� · �
A
Pf
f
� (41.25)
Random three-dimensional
�cu � �1
2�Vf� �
4
l� �
A
Pf
f
� (41.26)
Thus, the 3-D random fibre concrete shouldhave about half the post crack strength of thealigned composite.
The critical fibre volume for short randomfibres can be obtained by re-arranging equations(41.24), (41.25) and (41.26) to give Vf on the left-hand side of the equation.
In contrast to the use of the above equationsfor short chopped polypropylene fibre or steelfibre concrete, the glass-reinforced cement indus-try has sometimes used empirical efficiencyfactors. Thus, for a commonly used, two-dimensional random chopped glass fibre–cementcomposite the effectiveness of the fibre at the endof the multiple cracking zone has been shown tobe between 0.16 and 0.27 depending on thedirection of stress to the spray-up direction.
41.1.7 Toughness in uniaxial tension
One of the major attributes of fibre–cement com-posites is their increased toughness in comparisonwith plain concrete. However, the theoretical pre-diction of the increased toughness and its mea-surement in practice are both fraught withproblems. Commercial tests tend to rely onempirical comparisons between composites withand without fibres, tested with drop weights to‘no rebound’ or with beams deflected to specificvalues. These tests do not determine basic mater-ial parameters and are often highly dependent onsupport conditions or on specimen and machinedimensions and stiffnesses.
Structure and post-cracking composite theory
396
A more fundamental approach is to calculatethe very large increase in area of the tensilestress–strain curve resulting from the presence ofthe fibres. This increase, which may be severalorders of magnitude, is indicative of greatlyenhanced toughness and ductility, propertiesoften loosely associated with impact strength orimpact resistance. The theory outlined in Section41.1.3 can therefore be used as the basis oftoughness predictions.
In fibre cements and concretes there are essen-tially two modes of failure, one mainly involvingfibre pull-out after single fracture and the otherinvolving multiple fracture of the matrix followedby fibre pull-out or fracture. In the former case,which is typical of steel fibre concrete, the energyU to pull-out N fibres per unit area of compositecross-section is proportional to the frictional fibrematrix bond strength, �, and to the square of thefibre length (l) as in equation (41.27) (see Hibbertand Hannant, 1982).
U� (41.27)
In the case where failure occurs by multiplecracking followed by fibre fracture rather thanfibre pull-out, the energy is absorbed throughoutthe volume of material and may be calculatedfrom the area under the tensile stress–strain curveOXYB in Figure 41.1. The resulting energy isthat absorbed per unit volume of composite:
i.e. U�0.5�fu�fuVf �0.159�Ec�2mu (41.28)
where the first term represents the fibre strainenergy and the second term the contribution frommultiple cracking of the matrix.
For practical situations, we really require toknow the energy which can be absorbed by theproduct before it breaks into two pieces or beforeit is rendered unserviceable. The amount ofenergy which can be absorbed before reaching anacceptable amount of damage in terms of visiblecracking depends very much on whether fractureoccurs mainly by a single crack opening or bymultiple cracking. Thus a careful decision has tobe made as to whether the energy should be pre-dicted from a material parameter related to unit
N·Pf ·�l2�
24
area of fracture surface (J/m2) for the case of asingle crack or from a parameter for energyabsorbed per unit volume of material (J/m3) formultiple cracks.
An additional complexity is that because thecomposite modulus (Ec) and the matrix failurestrain (�mu) may increase with time in externalenvironments, then Vf(crit) calculated from equa-tion (41.3) may increase above the original valuerequired to ensure multiple cracking. Thus a com-posite which is tough and ductile at early agesmay suffer single fracture after some years withan energy to failure defined by the area under thecurve OX in Figure 41.1 (equation 41.29).
U�0.5Ec�2mu (41.29)
Although Ec and �mu may have increased withageing, the energy calculated from equation(41.29) is very much less than the energy given byequation (41.28).
Some typical toughness values for real compos-ites calculated from the area under the tensilestress–strain curves are shown in Table 41.1.
41.2 Uniaxial tension – fracturemechanics approachIt has been observed that asbestos cement has astrain to failure or to first visible crack often inexcess of 1000�10�6 and sometimes up to
Uniaxial tension – fracture mechanics approach
397
TABLE 41.1 Typical toughness values for real com-posites
Material Energy to failure(kJ/m3)
Asbestos cement 5.5
Glass-reinforced cementdry 120wet 3
Continuous polypropylenenetworks in cement 1000
Kevlar in cement 150
2000�10�6. This implies that the cracking strainof the matrix has been increased by the presenceof fibres since the cracking strain of the cementpaste from which the composite is made isusually less than 400�10�6. The compositematerials approach outlined in Section 41.1 is notadequate to explain this observation.
The implication is that microcrack propagationfrom pre-existing flaws in the matrix has beensuppressed by the very high volume (12 percent) of very small diameter, high modulus ofelasticity fibres and the concepts of fracturemechanics are required to quantify the possibleimprovements in performance. It has been shownthat typical pre-existing flaws in the matrix may be3mm long with a critical opening of about 1micron. For asbestos cement and some other fibrecomposites the flaws will be crossed by many fibresbut to demonstrate the principle of the argument,Figure 41.7 shows a flaw traversed by a single fibre(Hannant et al., 1983)
If the strain in the specimen is increased fromzero with the crack initially closed, the fibre willbe stressed by the crack opening to width B andthis stress will be transferred back into the matrixas shown in the strain diagram in Figure 41.7.The matrix will slide back over the fibres by adistance l, producing a difference in strainbetween fibre and matrix at the crack face of �f.
� Ef�f � (41.30)
Also, the crack opening B is obtained from theintegrated difference in strain between reinforcingfibres and matrix, i.e.
l� (41.31)
So, by eliminating l, the stress in the fibre �f isgiven by
�f � �EfB�1/2(41.32)
The fibres therefore exert a closing force on thecrack and reduce its opening. If there are N fibrescrossing a crack, the total closing force F redu-cing the crack opening is
Pf�Af
B��f
Pf�l�Af
FN�fmaxAf (41.33)
Using typical values for bond, modulus andspecific surface area for glass fibres indicatesstresses in the fibres of 300MPa before the pre-existing cracks reach sufficient width to propa-gate. The flaws will propagate catastrophically assoon as the rate of energy release becomes infini-tesimally greater than the rate of energy absorp-tion. The calculation is made using a computer-iterative process which shows, in accord withexperiment, that the ultimate failure strain ofasbestos fibres in asbestos cement at Vf �5% and10% is reached before the crack becomes un-stable. We therefore could expect breaking of thespecimen and first cracking of the matrix to occursimultaneously at a strain of 860�10�6 forVf �5% and at 1200�10�6 for Vf �10%.
Although similar effects should exist in all fibrecements where the fibre spacing is less than the
flaw size, it is only where composites contain veryhigh volumes, of high-modulus small-diameterfibres with high surface area and good bondingcharacteristics, that the effects of increasedmatrix cracking strain become sufficientlysignificant to be observable in practice.
41.3 Principles of fibrereinforcement in flexure
41.3.1 Necessity for theory
In many of their major applications, cement-bound fibre composites are likely to be subjectedto flexural stresses in addition to direct stresses,and hence an understanding of the mechanism ofstrengthening in flexure may be of equal import-ance to the analysis of the direct stress.
The need for a special theoretical treatment forflexure arises because of the large differences
Structure and post-cracking composite theory
398
FIGURE 41.7 Possible zone of relaxation of strain around a crack under stress that is traversed by a singlefibre. The left side of the figure illustrates the variation in strain within the fibre and matrix along the line A#A##(Hannant et al., 1983).
which are observed experimentally between theflexural strength and the direct tensile strengths,both in glass-reinforced cement and in steel-fibreconcrete. In both of these materials the so-calledflexural strength can be up to three times thedirect tensile strength even though, according toelastic theory, they are nominally a measure ofthe same value. The same situation occurs to alesser degree with plain concrete.
The main reason for the discrepancy infibre–cement composites is that the post-crackingstress–strain curve X–A and XYB in Figure 41.1on the tensile side of a fibre cement or fibre con-crete beam are very different from those incompression and, as a result, conventional beamtheory is inadequate. The flexural strengtheningmechanism is mainly due to this quasi-plasticbehaviour of fibre composites in tension as aresult of fibre pull-out or elastic extension of thefibres after matrix cracking.
Figures 41.8 (a) and (b) show a cracked fibrereinforced beam with a linear strain distributionand the neutral axis moved towards the compres-sion surface. At a crack the fibres effectivelyprovide point forces holding the section together(Figure 41.8 (c)). However, the exact stresses inthe fibres are generally ignored in flexural calcula-tions and an equivalent composite stress blocksuch as in Figure 41.8 (d) is assumed. The shapeof this stress block depends on fibre volume,bond strength, orientation and length efficiencyfactors. An accurate analysis of such a systempresents formidable problems but a simplifiedtreatment which is satisfactory for many practicalsituations is given below.
41.3.2 Analysis using a rectangular stressblock in the tensile zone of a beam
The analysis which follows is based on a simpli-fied assumption regarding the shape of the stressblock in the tensile zone after cracking.
The stress block for an elastic material inbending is shown in Figure 41.9 (a) and this isusually used to calculate the flexural strength (�fl)even though it is known to be grossly inaccuratefor quasi-ductile fibre composites.
Figure 41.9 (b) shows a simplified stress blockin bending for the type of tensile stress–straincurve OXY in Figure 41.1. This is typical of afibre–concrete composite after cracking, wherethe fibres are extending or are pulling out at con-stant load across a crack throughout the tensilesection. The ultimate post-cracking tensilestrength of the composite is �cu and �comp is thecompressive stress on the outer face of the beam.Figure 41.9 (b) approximates to the stresses insteel-fibre concrete where the crack widths aresmall (�0.3mm) compared with the fibre lengthand possibly to glass-reinforced cement at earlyages when the fibres are poorly bonded andextend before fracture or pull-out after fracture atroughly constant load. It also simulates compos-ites with �Vfcrit of continuous polypropylene nets.
A conservative estimate for the distance of theneutral axis from the compressive surface is D/4and using this assumption the moments of resis-tance of the two stress blocks can be compared:
moment ofresistance
� �flD2 for Figure 41.9 (a) (41.34)
moment ofresistance
� �cuD2 for Figure 41.9(b) (41.35)
In order that the two beams represented inFigure 41.9 can carry the same load, theirmoments of resistance should be equal, i.e.
�1
6��flD2 � �
1
3
3
2��cuD2 (41.36)
Therefore
�fl �2.44�cu (41.37)
Equation 41.37 implies that if the critical fibrevolume in uniaxial tension has just been achievedas shown in Figure 41.10(b) then the flexuraltensile strength will appear to be about 2.44times the composite cracking stress. Conversely, amaterial with less than half the critical fibrevolume in tension which has the uniaxial tensilestress–strain curve shown in Figure 41.10 (a) willnot exhibit a decrease in flexural load capacity
13�32
1�6
Principles of fibre reinforcement in flexure
399
Structure and post-cracking composite theory
400
FIGURE 41.8 Strain and stress distributions in a cracked fibre-reinforced concrete beam. (Note: scales of (c)and (d) are different.)
FIGURE 41.9 Stress blocks in flexure. (a) Elastic material; moment of resistance � (1/6)�flD2. (b) Elastic incompression, plastic in tension; moment of resistance�13/32�cuD2.
after cracking, implying that the critical fibrevolume in flexure has been achieved.
The limiting condition in Figure 41.9 (b) iswhen the neutral axis reaches the compressive
surface of the beam while maintaining themaximum tensile strength (�cu) throughout thesection. In this case:
�1
2��cuD2 � �
1
6��flD2, i.e. �fl �3�cu (41.39)
This type of simplified analysis explains whythe flexural strength for fibre cements and fibreconcretes is often quoted to be between 2 and 3times the tensile strength. Because the flexuralstrengths calculated using the normal ‘elastictheory’ approach often imply unrealistically hightensile strengths, it is unwise to use such strengthsin the design of fibre-reinforced cement or con-crete components. For the same reason it ispreferable to avoid flexural tests wherever pos-sible when the tensile strengths of fibre-reinforcedcement-based composites are required.
41.3.3 Effect of loss in ductility in tensionon the flexural strength
The importance of the post-cracking tensile straincapacity in relation to the area of the tensile stressblock in bending has already been demonstrated.
Principles of fibre reinforcement in flexure
401
FIGURE 41.10 Stress–strain curves in uniaxialtension: (a) no decrease in flexural load capacity aftercracking; (b) load capacity after cracking�2.44 timesthe cracking load for compressive strength/tensilestrength &6.
A further result of this major factor is thatchanges in strain to failure in the composite canresult in changes in the flexural strength evenwhen the tensile strength remains constant. Thisis particularly relevant to glass-reinforced cementwhere the tensile strain capacity can reduce by anorder of magnitude (1 to 0.1 per cent) over aperiod of years of natural weathering or watercuring. The movement of the neutral axistowards the compressive surface depends on ahigh post-cracking strain in tension and if thistensile strain decreases sufficiently, the compositewill have a reduced moment of resistance (Lawsand Ali, 1977). This effect is shown in Figure41.11. For instance, OAC is a tensile strain curve
FIGURE 41.11 Apparent bending flexural strengthcurves predicted for assumed direct tensile curves(Laws and Ali, 1977). (By permission of the Institutionof Civil Engineers.)
and OAC# is the associated bending curve. If thetensile strain reduces from C to A at constantstress, the bending strength will reduce from C#to A with an increasing rate of reduction as A isapproached and the material becomes essentiallyelastic.
41.4 ReferencesAveston, J., Cooper, G.A. and Kelly, A. (1971) Single
and multiple fracture. Paper 2 in The Properties ofFibre Composites, Conference Proceedings of theNational Physical Laboratory, IPC Science andTechnology Press, Guildford, UK
Aveston, J., Mercer, T.A. and Sillwood, J.M. (1974)Fibre-reinforced cement – scientific foundations forspecifications. Composites Standards Testing andDesign, National Physical Laboratory ConferenceProceedings, Guildford, UK
Structure and post-cracking composite theory
402
Bentur, A. and Mindess, S. (1990) Fibre ReinforcedCementitious Composites, Elsevier Applied Science,Guildford, UK
Hannant, D.J. (1978) Fibre Cements and Fibre Con-cretes, Wiley, Chichester, UK
Hannant, D.J., Hughes, D.C. and Kelly, A. (1983)Toughening of cements and other brittle solids withfibres. Phil. Trans. R. Soc. Lond. A, 310, 175–90.
Hibbert, A.P. and Hannant, D.J. (1982) Toughness offibre–cement composites. Composites, 105–11.
Laws, V. (1971) The efficiency of fibrous reinforcementof brittle matrices. J. Physics D: Applied Physics, 4,1737–46.
Laws, V. and Ali, M.A. (1977) The tensile stress–straincurve of brittle matrices reinforced with glass fibre.Fibre Reinforced Materials Design and EngineeringApplications Conference, Institution of Civil Engi-neers, London, pp. 101–9.
Chapter forty-two
Fibre-reinforcedcements
42.1 Asbestos cement42.2 Glass-reinforced cement (GRC)42.3 Natural fibres in cement42.4 Polymer fibre-reinforced cement42.5 References
42.1 Asbestos cementAsbestos cement is familiar to all engineers as theubiquitous roofing and cladding material whichhas had a very low cost and excellent durabilityduring the past 100 years. A reason for its successhas been the great durability of asbestos fibreswhich are shown in Figure 42.1. The fibresshown in this micrograph have been exposed toweathering for more than 10 years but the sub-micron fibres within the fibre bundle show nosign of deterioration. Other studies have shownthat the strengths of the fibre bundles varybetween 400MPa and 1400MPa irrespective ofexposure up to seven years.
However, due to the well-publicised healthhazards associated with asbestos fibres, which areknown to be carcinogens, there has been a rapiddecrease since 1980 in the UK in sales of asbestoscement sheeting products. Stringent safety pre-cautions are essential not only during the manu-facturing process but also to prevent inhalation ofdust during cutting and drilling such sheets onsite.
Another significant problem is that the materialis brittle and the impact strength is notoriouslylow so that there are a number of deaths in theUK every year as a result of people falling
through roofs when not using the required crawl-ing boards.
The net result of these adverse factors has beento virtually eliminate sales of asbestos cementproducts in Europe and the USA but neverthelessthere is still more asbestos cement on structuresthroughout the world than all other types of fibrecement combined and therefore a detaileddescription of its properties is warranted herein.
42.1.1 Mix design and manufacture
The proportion by weight of asbestos fibre is nor-mally between 9 to 12 per cent for flat or corru-gated sheet, 11 to 14 per cent for pressure pipesand 20 to 30 per cent for fire-resistant boards,and the binder is normally a Portland cement.Fillers such as finely ground silica at about 40 percent by weight may also be included in auto-claved processes where the temperature mayreach 180°C. Fibre volume, stress direction andproduct density all have an effect on propertiesand hence the properties depend to a certainextent on the manufacturer.
The most widely used method of manufactureof asbestos cement was developed from paper-making principles in about 1900 and is known asthe Hatschek process. A slurry or suspension ofasbestos fibre and cement in water at about 6 percent by weight of solids is continuously agitatedand allowed to filter out on a fine screen cylinder.The filtration rate is critical and coarser cementthan normal (typically with a specific surface areaof 280m2/kg compared with the normal value of
403
320m2/kg) is used to minimise filtration losses.Referring to Figure 42.2, in very much simplifiedterms, the Hatschek machine operates as follows.A dilute slurry pours into the vat and drainsthrough a porous sieve cylinder depositing thesolid contents as a layer on the surface of thesieve. The water passes through the sieve surfaceand returns to the vat via the backwater circuit.The sieve cylinder rotates and the layer rises outof the vat. A continuous felt runs in a loop fromthe sieve cylinder to an accumulation roll. Surfacetension forces the layer to transfer from the topof the sieve cylinder to the underside of the felt.The movement of the felt transfers the layer fromthe sieve to the accumulation roll and on the wayit is vacuum dewatered.
Typical outputs are 1 tonne per hour per metrewidth of vat. Felt speeds range from 40 to 70metres per minute. A typical three-vat machinewill make a 6mm thick sheet in six to nine revo-lutions and will make one sheet every 20 seconds.
Manufacturing costs are therefore very low.Due to the process, the fibres are essentially dis-persed in two directions with the predominantalignment of the fibres in the direction of rotationof the sieve drum. The stacks of products mayreach 60°C and autoclaving is sometimes used toreduce the shrinkage of the products.
Fibre-reinforced cements
404
FIGURE 42.1 Asbestos fibre bundle in cement pasteafter natural weathering for more than 10 years.
FIGURE 42.2 Hatschek manufacturing process forasbestos cement sheeting.
0
Strain (� 10�6)
Str
ess
(MPa
)
26
24
20
16
12
8
4
600 1200 1800 2200
FIGURE 42.3 Tensile stress–strain curve forasbestos cement.
42.1.2 Properties
Asbestos cement is the only fibre composite forwhich there are International Standards require-ments for certain properties. These are generallyexpressed in terms of minimum bending strength,density, impermeability and frost resistance. Forinstance, the minimum bending strength generallyvaries between 15MPa and 23MPa when testedunder defined conditions and depending onwhether the sheet is semi- or fully compressed.
Glass-reinforced cement (GRC)
405
Also, various loading requirements are definedfor corrugated sheets such as snow loads up to1.5kPa, and point loads to simulate men workingon a roof. Water absorption should not exceed20 to 30 per cent of the dry weight depending onthe type of product.
A typical tensile stress–strain curve for a com-mercial product is shown in Figure 42.3 wherethe failure strain is about 2000�10�6. No crackswere visible before failure which may be con-trolled by crack suppression as described inSection 41.2. A modulus of elasticity of about20GPa in tension and compression and amodulus of rupture well in excess of 30MPa havecombined to provide probably the most success-ful example of all time of a fibre-reinforced com-posite both in terms of tonnage and profitability.
42.1.3 Durability
Asbestos cement is known to be very durableunder natural weathering conditions and littledeterioration in flexural properties takes placedue to weathering, although the material becomesprogressively more brittle.
However, it has been shown that asbestosfibres in cement sheets at ages of 2, 16 and 58years do suffer a certain amount of corrosionwhich is compensated for, in terms of compositestrength, by an increase in bond between the fibreand the cement. The corrosion of the fibre is pro-moted by the penetration of airborne carbondioxide which causes carbonation at the surfaceof the fibre. Also, certain magnesium hydroxidesand carbonates may be formed as reaction prod-ucts. The natural variability in fibre strengthtends to mask any measured changes in fibrestrength due solely to weathering.
42.1.4 Uses
Corrugated roofing and cladding for agriculturaland industrial buildings has formed by far thelargest application, and the ability to be mouldedinto complex shapes has enabled a wide range ofaccessories to be produced for roofing applica-tions. Flat sheet has been used for diagonal tiles
to replace natural slate. Asbestos cement pressurepipes have been used for many years for convey-ing mains water, sewage, sea water, slurries andindustrial liquors. Diameters range from 50mmto 900mm with working pressures from0.75MPa to 1.25MPa. It is the extraordinaryability of the asbestos fibres to suppress crackingin the matrix which has made these watertightapplications possible and no other fibre cementhas been able to emulate its performance. Also,non-pressure fluid containers such as rainwatergoods, conduits, troughs, tanks and flue pipeshave accounted for a large proportion of theminor applications of asbestos cement.
42.2 Glass-reinforced cement (GRC)Glass-reinforced cement is normally made withalkali-resistant glass fibre bundles combined witha matrix consisting of Ordinary Portland cementplus inorganic fillers. E-Glass fibres have beenused with a polymer modified cement matrix toprotect the glass against attack by the alkalis inthe cement. The material described in this sectionrelates to zirconia-based alkali-resistant fibreswhich are normally produced in the form ofstrands consisting of 204 filaments each ofbetween 14 and 20 microns in diameter. Up to 64strands may be wound together as a roving whichis cut during the making of GRC into strands12mm to 38mm long. Pre-chopped strands maybe supplied in lengths as short as 3mm. The indi-vidual filaments are bonded together in the strandby an organic ‘size’ which determines the physicalnature of the end product. A photograph of astrand embedded in cement is shown in Figure42.4.
The presence of zirconia (ZrO2) in the glassimparts resistance to the alkalis in the cementbecause the ZrBO bonds, in contrast to theSiBO bonds, are only slightly attacked by theOH� ions thus improving the stability of the glassnetwork.
42.2.1 Manufacture
Spraying and premixing are the main productionprocesses for glass-reinforced cement.
In spray processes the cement/sand mortarslurry is fed to the spray gun and is broken intodroplets by compressed air. Glass fibre roving isfed to a chopper on the spray head which chopsthe fibre into 25mm to 40mm lengths and injectsthem into the mortar spray to be deposited on themould. The slurry mortar is typically 1:1sand:cement ratio with a W/C ratio of 0.33.Admixtures are used to increase the workabilityand the resulting composite contains typically 5per cent by weight of glass fibre.
The composite may be sprayed manually or bymachine and some processes use dewatering byvacuum to result in an end product with a freeW/C ratio of about 0.28 which has greaterdensity and strength that the non-dewateredproduct.
Premixing processes, as the name implies,involve the blending together of cement, sand,water, admixtures and chopped strands in amixer before placing in the mould. The fibres areadded at the end of the mixing process at a lowerspeed to avoid damage. Typical mixes contain asand:cement ratio of 2:3, a water:cement ratio of
Fibre-reinforced cements
406
FIGURE 42.4 Glass fibre strand in cement paste.
about 0.35, workability aids and, typically, 3 percent by weight of 12mm long fibres. Having beenmixed, the composite may be cast by pumpingwith or without vibration and also by spraying.Pressing combined with dewatering may also beused with premixes.
Prebagged formulations containing between0.5 per cent and 2.5 per cent by weight of glassfibre have also been developed for renderingbrickwork or blockwork.
As with all cement-based products, it is import-ant to moist cure glass-reinforced cement prod-ucts for as long as possible to ensure that amaximum amount of hydration and hencestrength gain has occurred before loading.
42.2.2 Properties
The tensile stress–strain curves for glass-rein-forced cement will typically follow either ofcurves of type A or B in Figure 41.1. Curve typeA is representative of the lower fibre volumesused in the premix composite where a singlecrack and no increase in the post-cracking stressis expected. Curve B is representative of the earlyage sprayed composite with more than the criticalfibre volume at that age. This results in multiplecracking and a high failure strain as shown inFigure 42.5.
Fibre length, volume and orientation also affectthe performance of the composite at 28 days inuniaxial tension (Ali et al., 1975). The shape ofthe curves is approximated by the theoreticalapproach described in Section 41.1 and it can beseen that strength and strain to failure are bothincreased by increases in fibre length and volume.Nominal flexural tensile strengths may varybetween 15MPa and 50MPa and follow the rela-tionship between tension and flexure shown inFigure 41.11.
However, although the mechanical propertiesare good at early ages, the strength and toughnessof some GRC formulations may change with timeand hence design stresses are conservative.Typical design stresses quoted from trade liter-ature are shown in Table 42.1. In this table, thedifference in performance between sprayed and
premixed material due to fibre volume and orien-tation is demonstrated.
42.2.3 Durability
In the 1970s, the durability of GRC was found tobe strongly influenced by the environment towhich it was exposed (Majumdar et al., 1991)For instance, in dry air, there was little change inflexural or tensile strength during a 10-yearperiod, whereas under water or in natural weath-ering there was a decrease in tensile strength ofthe composite by more than 50 per cent. Thedecrease in impact strength was even more severe,reducing by an order of magnitude. The changeswere due to a combination of factors including a
Glass-reinforced cement (GRC)
407
FIGURE 42.5 Tensile stress–strain curves of spraydewatered glass-reinforced cement (Oakley andProctor, 1975).
loss in fibre strength and failure strain due toalkali attack, and an increase in matrix crackingstress due to continuing cement hydration whichincreases the critical fibre volume above theincluded fibre volume. Another contributingfactor was the filling of the voids in the fibrebundles with lime crystals and calcium silicatehydrates which increased the bond with indi-vidual glass filaments so that matrix cracks easilypropagated through the fibre bundle. The netresult of these changes was to reduce the compos-ite strain to failure, in some cases to matrixfailure strain, thus reducing the once ductilematerial to a brittle material.
An example of the way in which the increase inmatrix strength and reduction in composite strainto failure can affect the critical fibre volume andtoughness of the composite is shown below usingFigure 42.6.
For glass-reinforced cement, the followingvalues have been published (Building ResearchEstablishment, 1979):
28 days, Ec �22.5GPa with a bend-overpoint at 9.5MPa, which gives �mu �422*10�6,�fu �1000MPa;10 years, Ec �28.5GPa, no accurate value forthe bend-over point is given. If we assume nochange in �mu and take �fu equal to 600MPaafter 10 years, equation (41.3) gives an increasein Vfcrit from 1 per cent at 28 days to 2 per centat 10 years for aligned fibres.
Most glass-reinforced cement is sprayed withshort fibres in a random (2-D) orientation. To
TABLE 42.1 Typical Cem-FIL GRC design stresses for different manufacturing methods (from trade literature)
Design value Loading example Hand or premix Machine spray
Compression compressive 12MPa 12MPaTension cylinder hoop stress 3MPa 2MPa
bending of sandwich panelsTension/bending bending of box sections or channels 4MPa 2.5MPaBending bending of solid beams or plates 6MPa 4MPaShear shear loading 1MPa 1MPa
Note: These design values may be varied in certain product areas, e.g. formwork. Limit state methods are also used.
take account of the non-alignment, we multiplythe value of �fu by 0.27 (the efficiency factor forstress) and find that a total fibre volume fractionof more than 3.5 per cent at 28 days, and morethan 7.4 per cent at 10 years is required to main-tain ductility. Typical total fibre volumes areabout 4 per cent and it is known that the energyabsorbed to failure, found from the area underthe measured stress–strain curve, reduces fromabout 120kJ/m3 at 28 days to less than 5kJ/m3 at5 years. This change may therefore be explainedby the increase in critical fibre volume with time.
An understanding of these potential problemsresulted in research to eliminate calcium hydrox-ide growth by including a metakaolin syntheticpozzolana (described in Chapter 15) to react withthe free lime in a controlled way to prevent limebuild up within the fibre bundle. The metakaolinparticles also appear not to migrate into the fibrebundle. Results from these modified matricesusing accelerated tests have indicated that greatlyimproved durability and long-term toughness ofGRC composites should now be possible usingabout 25 per cent cement replacement bymetakaolin. Thus, long-term stable properties formetakaolin-based matrices approximate to curveA in Figure 42.6 rather than the embrittled curvetype B. In contrast, silica fume with a particle sizeof less than 0.1 micron, although reacting withthe lime, may penetrate to individual filamentsand result in the formation of hard calcium sili-cate hydrates which are as damaging to the fibresas lime deposition.
However, a note of caution should be maderegarding the long-term predictive ability of high-temperature accelerated tests because, as withmany durability problems with cement-basedmaterials, total confidence can only be gained by‘real time’ trials in natural weathering conditionsover many years.
42.2.4 Uses
Cladding panels are a major field of applicationfor glass-reinforced cement. Due to crackingcaused by restrained warping in early applica-tions particular attention should be paid to
Fibre-reinforced cements
408
FIGURE 42.6 Effect of natural weathering on thetoughness of glass-reinforced cement.
thermal and moisture movements and to fixingdetails in large, double-skinned sandwich panelsof this type of construction. Light colouring ofthe panels is preferred because this helps toreduce thermal stresses particularly where there isan insulating core. There is increasing emphasison the use of single-skin cladding panels attachedto prefabricated steel frames by flexible anchors.This is known as a GRC Stud Frame system. Thefixings are designed to allow unrestrained thermalor moisture movements of the glass-reinforcedcement skin which may be 6m long by storeyheight.
An important use where the early age strengthand toughness are beneficial is in permanentformwork for bridge decks, the advantage beingthat no temporary support works are requiredand a dense, high-quality cover is provided to thereinforcement.
The greater efficiency ensured by using contin-uous glass fibre rovings in the main stress direc-tion has been utilised in a process for producingcorrugated sheeting. The corrugated sheets startas a continuous flat sheet made from two layers,each 3.25mm thick. Short strands as cross rein-forcement are immersed in the matrix on theupper/underside and this leads to an ideal sand-
wich structure. The lengthwise reinforcement ismostly made up of one-directional inlaid glassfibre rovings, which are sandwiched between thetwo layers that make up the sheet. The flat sheetsare then corrugated on formers before curing.
There are many other uses such as cable ducts,agricultural products, sewer linings, culverts,sound barriers, drainage channels, fire-resistantboards, septic tanks, roofing slates and mortarrenders for dry block wall construction. Sometrials have been made using glass fibre tendonsprotected with polyester resin as prestressingtendons in prestressed concrete structures.
42.3 Natural fibres in cementThe use of natural cellulose or vegetable fibres incement or mortar products is common in both‘developed’ and ‘developing’ countries and thesubject has been reviewed in detail by Swamy(1988) and Bentur (1990).
42.3.1 Wood fibre products
Manufacture
In developed countries the bulk usage is for woodcellulose fibres from trees. The wood is mechani-cally and chemically pulped to separate the indi-vidual fibres which may be between 1mm and3mm long and up to 45 microns in width. Hard-woods and softwoods are used and the elasticmodulus of individual fibres may vary between18GPa and 80GPa with strengths between350MPa and 1000MPa depending on the angleof cellulose chains in the cell wall. Cellulose fibresproduced from timber have several advantageswhen used in thin cement or autoclaved calciumsilicate sheets. The fibres are low cost comparedwith most man-made fibres, they are a renewableresource, there is considerable experience in theuse of such fibres in existing plant for asbestoscement, and they have an adequate tensilestrength for cement reinforcement. However, cel-lulose is sensitive to humidity changes and theelastic modulus of the fibres reduces when wet so
that the properties of the composite may varyconsiderably from dry to wet. The fibres are usedin volumes up to 10 per cent in conjunction withpolymer fibres in asbestos-free products in whichmanufacturing procedures are very similar to theHatschek process already described for asbestoscement.
Properties
When suitably pre-treated by refining, woodfibres used in conjunction with polyvinyl alcoholfibres in a matrix of Portland cement and fillerscan provide a tough and durable fibre cement.This composite is suitable for the commercialproduction of corrugated sheeting and pressedtiles on traditional slurry dewatered systems suchas the Hatschek machine (Figure 42.2). The vari-ation in composite properties wet to dry andpressed to unpressed is shown in typical tensilestress–strain curves for commercial composites inFigure 42.7.
Uses
In Australia, cellulose fibres have completelyreplaced asbestos fibres in flat sheeting productsmade from an autoclaved calcium silicate. Auto-claved systems are said to have the advantageover air-cured hydrated cement binders in thatthere is greater dimensional stability in relation tomoisture and temperature movements. Alsobecause of the absence of free alkalinity theboards can be more easily decorated. In the UK,flat sheet for internal and external applicationshas been available for a number of years, pro-duced from cellulose fibres in an autoclavedcalcium silicate matrix. A wide range of proper-ties is available in cellulose fibre boards, typicalvalues being an elastic modulus of 12GPa withthe tensile strength varying between 6MPa and20MPa and the modulus of rupture between15MPa and 30MPa depending on whether thecomposite is wet or dry and on the fibre volume.Typical tensile stress–strain curves in the dry andwet states for these materials are shown in Figure42.8.
Natural fibres in cement
409
Wood cellulose fibres in concrete
Wood cellulose fibres are not suitable for use inbulk concrete applications because of difficultiesin mixing and compaction and their use is there-fore limited to automated factory processes. Insome of these processes wood chips or flakes atup to 20 per cent are mixed with cement and finesand to make a variety of compressed wood chipboards or particle boards. These are generally
used internally and have low flexural tensilestrength, often below 1MPa. They are not strictlyfibre-reinforced cements.
42.3.2 Vegetable fibre products
The use of vegetable stem fibres in developingcountries is generally aimed at producing cheapbut labour-intensive, locally constructed cement-
Fibre-reinforced cements
410
FIGURE 42.7 Effect of moisture condition on the tensile properties of a cement composite containing cellulosefibres and artificial fibres – pressed, unpressed.
FIGURE 42.8 Effect of moisture condition on tensile properties of autoclaved calcium silicate containing cellu-lose fibres.
based roof sheeting often of corrugated or foldedplate design. Long fibres which are indigenous tothe locality are used, such as akwara, banana,bamboo, coir, elephant grass, flax, henequen,jute, malva, musamba, palm, plantan, pineappleleaf, sisal, sugar cane and water reed. Lengths offibres may be up to 1m or more and are handplaced in a matrix of sand and cement. Corru-gated sheets of up to 2m by 1m in size of6–10mm thickness and tiles may be producedwith fibres in preferential directions.
Manufacture and applications
Composites made from vegetable stem fibres aremanufactured by simple hand lay-up processeswhich are potentially suitable for low-costhousing applications.
In these applications the fibre content is usuallyless than 5 per cent when applying mixing tech-nologies, but it may be greater when using thetechnologies of hand lay-up of long fibre rovings.Hand laying involves the application of a thinmortar layer on a mould, followed by alternatelayers of fibres and mortar matrix. The fibres canbe rolled into the matrix or worked into it manu-ally, and the process may involve some vibration.In the mixing technique, there is a limit to thecontent and length of fibres that can be incorpo-rated, since as with any other fibres workability isreduced. However, many of the natural fibrecomposites are intended for the production ofthin components such as corrugated sheets andshingles, and for these applications there is arequirement for both plasticity and fresh strengththat will permit the shaping of the product. Forsuch purposes, reduced flow properties are not asdetrimental as in the case of conventional con-crete. In these components, which have a typicalthickness of about 10mm, the matrix is a cementmortar, and the mix with fibres, or with hand-laid fibres, is spread on a mould surface and thenshaped. Corrugation can be achieved by pressingbetween two corrugated sheets.
Properties
The cracking stress and strength of the compos-ites are not greatly increased compared with theunreinforced matrix (i.e. about 1MPa to 3MPa)but the fibres enable the sheets to be formed inthe fresh state and handled and transported in thehardened state. Considerable toughness isachieved in the short term.
Bamboo, when split into strips and woven intomeshes, has been used as reinforcement for avariety of applications from roads and structuresto water tanks. Tensile strengths of the fibre arecommonly in excess of 100MPa with elasticmoduli between 10 and 25GPa. Toughening andpost-cracking performance are the most import-ant characteristics and optimum fibre volumesbetween 1.5 per cent and 3 per cent have beenquoted.
Durability
The high alkalinity of the pore water preventsmicrobiological decay in the fibres but, to setagainst this, the calcium hydroxide penetrates thefibre to mineralise or petrify it. The high alkalin-ity can cause severe reduction in fibre strengthbut where carbonation has penetrated, this rateof reduction of strength is reduced. However,natural stem fibres are not expected to give thecomposite a long lifetime, although short cellu-lose fibres as used in alternatives to asbestoscement products have been shown to be moredurable than natural stem fibres.
42.4 Polymer fibre-reinforcedcementThe inclusion of polymer fibres into cement-basedproducts is potentially a very large world-widemarket. For instance, about 90 countries haveproduced asbestos cement for cladding, roofingor pipes and about 3.5 million tonnes of asbestosfibre have been used annually in the asbestoscement and building products industries, givingabout 28 million tonnes of products. This marketwill either be lost to other products or man-made
Polymer fibre-reinforced cement
411
fibres will be substituted for the asbestos fibres. Itis clear therefore that there is a potential marketand considerable inroads have been made sincethe early 1980s by polymer fibres into this indus-try.
Although a wide variety of polymers has beenused on a trial basis in cement-based materials,only a few have been commercially successful.Polypropylene and polyvinyl alcohol have beenthe most used although polyethylene pulp is alsoused in some thin sheet products.
42.4.1 Chopped polypropylene films
Chopped polypropylene films have been used atfibre volumes of 3 per cent to 5 per cent toproduce alternative products to asbestos cementwith some modifications being required to thetraditional machinery.
The polypropylene in this case was speciallystretched and heat treated to give elastic moduliof 9 to 18GPa with tensile strengths from 500 to700MPa and ultimate strain of 5 to 8 per cent.Various surface treatments to improve wetting ofthe films and increase their bond were carried outbefore splitting the tape and chopping intolengths between 6 and 24mm to give fibres with abasically rectangular cross-section but withfrayed edges.
42.4.2 Continuous opened polypropylenenetworks
Layers of networks of continuous polypropylenefilms, as shown in Figure 42.9, in combinationwith continuous glass fibre rovings have beenused commercially in fine grained cement-basedmaterials to produce alternative products toasbestos cement. The advantage of this system isthat the full fibre strength of both fibres is usedbecause there is no pull-out and excellentmechanical bonding in the polypropylene isachieved by virtue of the uneven micro- andmacro-slits in the films and the many fine hairsproduced in the production process. Also it hasbeen found that there is a synergistic interactionbetween the fibre types so the performance of
Fibre-reinforced cements
412
FIGURE 42.9 Polypropylene networks.
each fibre is enhanced by the presence of theother fibre.
Manufacture
The continuous networks are laid up in twodirections in twelve-layer packs. Three or fourpacks are fed simultaneously to a machine whichimpregnates them with cement slurry in sequence,producing a wet, flat sheet with good two-dimensional strength. This is then corrugated in avacuum corrugator. Both chopped and continu-ous glass fibres can be included in the process.
Properties
All the requirements for roofing sheet can be metand additionally, toughness values of 1000kJ/m3
are possible. The failure strain remains in excessof 5 per cent after weathering, provided that thecritical fibre volume is exceeded at the appropri-ate age.
Durability
Polypropylene fibres show excellent resistance toalkalis and in a cement matrix have been shownto have good resistance to a variety of weathering
conditions up to 18 years of real time exposure(Figure 42.10). Equally important is that thebond strength has been shown to be time stableup to 18 years (Hannant, 1999), giving confi-dence in the long-term toughness of polypropy-lene fibre composites.
42.4.3 Polyvinyl alcohol fibres (PVA)
High-strength and stiffness PVA fibres are usedwidely as an asbestos replacement in asbestoscement products. However, taken by themselvesin a cement slurry, they offer little retention ofthe cement grains and hence must be used in con-junction with cellulose pulp to keep the cement inthe system as water is sucked out by vacuum. Thefibres are treated on the surface to enhance theircompatibility with the matrix, the quantity offibres being typically 3 per cent by volume. Flex-ural strengths of the sheeting are adequate tomeet the requirements of the appropriate Euro-pean standards. Alkali resistance has been statedto be excellent and the fibres can survive expo-sure to temperatures of 150°C without loss instrength.
42.4.4 Polyethylene pulp
Polyethylene pulp made from short fibres hasmainly been used as a cement retention anddrainage aid as a substitute for asbestos fibres inHatschek type process for the manufacture ofthin sheet products. Up to 12 per cent by volumehas been used and at this level improvements inflexural strength and ductility have also beenobserved. Because the fibres do not swell in thepresence of water, the durability of the productsis said to be improved in comparison with similarsystems using cellulose fibres.
42.4.5 Continuous networks of high-modulus polyethylene fibres
Highly orientated polyethylene fibres may be pro-duced by gel spinning or high draw ratios andfibres have been produced with the elasticmodulus of glass and the strength of steel. Com-mercial fibrillated tapes with initial elastic moduliof about 30GPa have been used in thin cementsheets in a similar fashion to polypropylene nets.
Durability in alkalis is expected to be good butthe films which have been available have suffered
Polymer fibre reinforced cement
413
0
Time (years)
Fibr
e st
reng
th (
MPa
)
300
5
250
200
150
100
50
010 15 20
Natural weathering
Laboratory air
Under water
95% confidence limits
FIGURE 42.10 Durability of polypropylene networks in cement.
from high creep strain in comparison withpolypropylene.
42.5 ReferencesAli, M.A., Majumdar, A.J. and Singh, B. (1975) Pro-
perties of glass fibre cement – the effect of fibrelength and content. J. Materials Science, 10,1732–40.
Bentur, A. and Mindess, S. (1990) Fibre ReinforcedCementitious Composites, Elsevier Applied Science,London and New York.
Hannant, D.J. (1999) The effects of age up to 18 yearsunder various exposure conditions on the tensileproperties of polypropylene fibre reinforced cementcomposites. Materials and Structures, RILEM, 32,March, 83–8.
Oakley, D.R. and Proctor, B.A. (1975) Tensilestress–strain behaviour of glass fibre reinforcedcement composites. In Fibre Reinforced Cement andConcrete, Construction Press Ltd, Lancaster, UK,pp. 347–59.
Swamy, R.N. (ed.) (1988) Natural Fibre ReinforcedCement and Concrete, Vol. 5, Concrete Technologyand Design, Blackie, Glasgow.
Fibre-reinforced cements
414
Chapter forty-three
Fibre-reinforcedconcrete
43.1 Steel fibre concrete43.2 Polypropylene fibre-reinforced concrete43.3 Glass fibre-reinforced concrete43.4 Reference
43.1 Steel fibre concreteConcrete reinforced with chopped steel fibres involumes generally less than 1 per cent has atensile stress–strain curve of the type shown asOXA in Figure 41.1. The reason for this is that itis physically very difficult to include sufficientfibres in the mix to exceed the critical fibrevolume which, for short random 3-D orientedfibres, is twice the value given by equation (41.4).
This critical fibre volume for short steel fibresgenerally exceeds 2 per cent. Apart from the eco-nomics, it is physically difficult to include thesefibre volumes because concrete contains about 70per cent by volume of aggregate particles whichobviously cannot be penetrated by fibres. Also,the fibres tend to end up in a three-dimensionalrandom distribution when mixed in a rotarymixer which, together with their short length,makes them very inefficient at reinforcement inany given direction of tensile stress. Nevertheless,useful properties in the composite have beenachieved by many practical, if rather specialist,systems.
A great variety of fibre shapes and lengths areavailable depending on the manufacturingprocess, a few of the more common types beingshown in Figure 43.1. Cross-sectional shapesinclude: circular (from drawn fibres); rectangular
(from slit sheet or milled from ingots); sickleshaped (from the melt extract process); andmechanically deformed in various ways toimprove the bond strength. Fibre lengths rangefrom 10 to 60mm with equivalent diametersbetween 0.1 and 1.0mm. Mild steel, high-tensilesteel and stainless steel fibres are available.
It should be realised that the average fibre pull-out length is 1/4 which, for the longest 60mmfibres, is only 15mm. This length is insufficient toallow efficient use to be made of the high tensilestrength of drawn wire unless devices such asbends or crimps are used to improve anchorageefficiency.
43.1.1 Mix design and compositemanufacture
In the early development of steel fibre concrete,mixing problems often occurred with balling upof fibres in unsuitable mixes. These problemshave now largely been resolved by an appropriatechoice of fibre type and volume although inap-propriate mix designs and poor introduction ofthe fibres to the mix can still cause balling prob-lems. Advice on mix proportions for specific fibretypes and volumes is readily available from fibremanufacturers.
Mix designs for rotary mixers generally recom-mend the use of fibres with an equivalentlength/diameter ratio of 70 or less, and typicalweights of fibre are between 25kg/m3 and60kg/m3. These weights are equivalent to fibrevolumes between 0.3 per cent and 0.8 per cent.
415
It is important to use a relatively high propor-tion of fines, for example a typical mix maycontain 800kg/m3 of river sand and 300 to350kg/m3 of cement, often with the further addi-tion of pulverised fuel ash. Aggregates larger than20mm should be avoided, and it is preferable tolimit the aggregate fraction larger than 14mm to15 per cent to 20 per cent. Free water cementratios of less than 0.55 are preferable and work-ability is commonly improved by the addition ofplasticisers or superplasticisers to give slumps ofmore than 100mm.
It is unfortunate that workability is reduced byan increase in fibre volume and equivalent
Fibre-reinforced concrete
416
FIGURE 43.1 Steel fibres showing a variety of typesof mechanical deformation.
length/diameter ratio (l/d ratio) because the post-cracking tensile strength of the composite (equa-tions (41.24), (41.25), (41.26)) is improved by anincrease in both these parameters.
Therefore the requirements of reinforcementand workability act against each other and acompromise must be reached usually at l/d ratiosbetween 40 and 70 with fibre lengths between20mm and 60mm.
Generally the fibres are added last to the freshconcrete, care being taken to ensure that noclumps are added and the fibres are rapidlyremoved from the entry point to the mixer. Alter-natively they may be added onto the aggregate onthe conveyor belt. Proprietary systems exist forintroducing fibres into a ready-mix truck byblowing the individual fibres at high velocity intothe back of the truck. Collated fibres with a watersoluble glue also considerably assist the batchingand mixing actions with bags of fibres beingadded to the mixer drum, the paper bags them-selves also being water soluble.
Guniting or sprayed steel fibre concrete isanother manufacturing process which is widelyused for tunnel linings and rock slope stabilisa-tion. Fibre lengths typically range from 25mm to40mm. As with normal concrete, the greater thefibre aspect ratio and fibre volume, the better isthe performance of the sprayed concrete but themore difficult it is to mix, convey and spray.Mixes generally have less than 10mm maximumsized aggregate with cementitious contents inexcess of 400kg/m3 and fibre contents between30kg/m3 and 80kg/m3. Cement replacements andadmixtures are widely used.
In a third manufacturing process known asSIFCON fibre volumes up to 20 per cent are pre-placed into a mould before mixing and are theninfiltrated with a fine-grained cement-basedslurry. This gives very high strength and tough-ness in localised regions such as beam/columnintersections. Flexural strengths up to 60MPaand tensile strengths up to 16MPa are said to bepossible with this system.
43.1.2 Properties
Steel fibres provide virtually no increase in thecompressive or uniaxial tensile strength of con-crete. The main benefits in uniaxial tension resultfrom the control of crack widths due to shrinkageor thermal effects in slabs and tunnel linings andthis is not an easily quantifiable parameter butrelates to post-cracking fibre pull-out forces. Post-cracking uniaxial tensile strengths of 0.4 to1.0MPa are possible at commonly used fibrevolumes, at crack openings up to about 2mm.
The relationship between tensile strength andflexural strength for various fibre volumes isshown in Figure 43.2 from which it can be seenthat at the fibre volumes commonly used in prac-tice of less than 0.8 per cent, the increase in post-cracking flexural strength is rather limited.Nevertheless, this post-cracking flexural strength,which results from the increased area of thetensile part of the stress block, as shown inFigures 41.8 and 41.9, is a most important partof the commercial uses of steel fibre concrete,enabling reductions of thickness to be made in
Steel fibre concrete
417
Str
engt
h (M
Pa)
10
15
5
0 1 2 3Volume of fibres (%)
Flexural strength
Tensile strength
FIGURE 43.2 Typical direct tensile and flexuralstrengths of steel fibre-reinforced concrete and mortar(Edgington, 1973).
sections subject to flexure or point loads. Typicalstress/deflection curves for beams are shown inFigure 43.3 and demonstrate the post-crackingcharacteristics in bending which result from theductile characteristics of the tensile stress blockeven though the fibre volume is less than the criti-cal fibre volume in tension. Impact strength andtoughness are greatly increased, the increasedtoughness resulting from the increased area underthe load deflection curve in tension and flexure. Avariety of toughness indices have been proposedin the literature depending on the flexural deflec-tion which is chosen to represent a typical ser-viceability limit. Improved fatigue resistance isoften claimed but this is a complex parameterwhich depends so heavily on fibre type andvolume that no generalised improvements can bestated.
43.1.3 Durability
Steel fibres are generally well protected inuncracked concrete where the high alkalinity pro-vides a passive layer on the fibre surface. Evenwhen the fibres are near the surface in a carbon-ated zone, serious corrosion takes many years tooccur and surface spalling is rare. It may be thatthe short fibre length results in a more or lessuniform potential adjacent to the fibre which
Flex
ural
str
ess
MPa
6
1
1.0 1.5Deflection mm
2
3
4
5
0.50
FIGURE 43.3 Typical stress–deflection curve for40kg/m3 of steel fibres in a 100mm square beam at400mm span.
limits the setting up of corrosion cells. Any fibresprotruding from a concrete surface will readilycorrode or wear away and have not been foundto give significant problems.
The main durability problem is likely to occurwhere load-bearing carbon steel fibres areexposed across cracked sections in the presence ofchlorides, where they will readily corrode and itwould be wise in such conditions to use stainlesssteel fibres. If these precautions are not taken, themode of failure in cracked concrete subject towater and chlorides will eventually change fromductile with fibre pull-out, to brittle with fibrefracture and this sudden change will not be pre-dictable by the extrapolation of real time datafrom durability experiments.
43.1.4 Applications
A major use of steel fibres in the developed coun-tries is to use them as a replacement for conven-tional steel mesh in industrial ground floor slabs.Fibre weights of between 15kg/m3 and 60kg/m3
are commonly used in floor thicknesses between120mm and 200mm. The fibres are particularlybeneficial in the laser screed process (Figure 43.4)for large area pours (�1000m2/day) because theyavoid interruptions to the construction processcaused by placing formed joints and mesh rein-forcement. Joints are normally still necessary andinduced joints should be cut to about 1/4 of theslab depth with a diamond saw within the first 24hours after casting at between 5m and 10mcentres. Alternatively, where joints have to beformed, steel-edged permanent shutters withlateral movement capability are beneficial in pre-venting edge breakdown. When sawn joints haveformed due to restrained contraction, the fibresrestrict joint opening and maintain aggregateinterlock at the joint surface. Also when subjectedto flexural stresses, the load at which cracksbecome visible on the top surface can beincreased compared to plain concrete slabs.
The main advantages claimed by specialistusers of the laser screed process are: eliminationof mesh which makes faster, more simplified con-struction, reduction of manpower costs, saving of
Fibre-reinforced concrete
418
dowel bars, tie wire, etc., better performance ofjoint arises induced by diamond blade cutting,and more uniform joint opening than when nomesh or fibres are used.
The typical stress maintained across a jointafter cracking may be calculated from equation(41.26), i.e. for 30kg/m3 of steel fibres, 60mmlong by 1mm in diameter with a bond strength��4MPa, the post-cracking tensile strength �cu
will be:
�cu � ��Vf .�.�4
l�.�
A
Pf
f
� (41.26)
Vf �30/7860�0.0038
��cu �0.5�0.0038�4� �6
4
0� �
�0.46MPa
This theoretical post-cracking strength is sup-ported by experimental studies of stresses sus-tained across induced cracks. This apparentcomposite post-cracking strength is equivalent toa force of about 75kN/m (7.5T/m) width of a200mm thick ground floor slab sawn to a depthof 50mm to induce cracks. This force can becompared with a force of 65kN/m (6.5T/m)width supplied by a typical mesh reinforcementconsisting of one layer of fabric. Hence thereplacement of fabric with steel fibres for thecontrol of induced crack widths appears to be apractical proposition.
�1��12
FIGURE 43.4 Laser-screed machine placing a steelfibre-reinforced concrete floor (by permission ofBekaert Building Products).
There are a few specialist systems in which theneed for saw cut joints is eliminated completelyand this prevents any problems with curling orspalling of the slabs.
Other uses of mild steel fibres with conven-tional mixing and compaction techniques haveincluded hydraulic structures such as spillwaysand sluices, airfield pavements and precast com-ponents such as segments for tunnel linings.When overlaying concrete roads, an isolatinglayer of 30mm of asphalt has been found to beimportant and bonding the new and old worktogether is a critical part of the operation. Jointsbetween 10m and 15m are essential and reflectedcracks must be expected. However, performancehas been shown to be very good when properlydesigned and constructed.
Another major application of steel fibre con-crete is in situ tunnel lining and rock slope stabili-sation using the gunite technique and the use ofsteel mesh in these situations is being steadilyreplaced.
One of the most successful uses of stainless steelfibres has been in castable refractory concretes foruse at temperatures up to 1500°C. In these prod-ucts, aluminous cement is used and initial cost isnot the prime consideration provided that productlife can be increased, typically by 100 per cent.
43.2 Polypropylene fibre-reinforcedconcrete
Extensive use has been made in the constructionindustry of small quantities (0.1 per cent byvolume) of short (�25mm long) fibrillated ormonofilament polypropylene fibres as shown inFigure 42.15(b) to alter the properties of thefresh concrete. Fresh concrete means in the firstfour or five hours after mixing, either before orjust after the concrete has stiffened but could notbe described as hardened. ‘Reinforcement’ isprobably not a correct term to use to describe theeffects of the fibres and there is no theoreticaltreatment which adequately predicts their effects.Also, the type of polymer from which the fibre isformed is probably not very important because
Polypropylene fibre-reinforced concrete
419
FIGURE 43.5 Types of chopped polypropylenefibres: (a) twisted twine; (b) untwisted.
little use is made of the fibre stiffness or strengthduring this critical initial period and alkalis havehad no time to cause damage. The description oftheir mode of action is therefore mainly qualitative.
Claims made for the benefits endowed by thesesmall volumes of fibres include:
• reduction in sedimentation and plastic shrink-age;
• elimination of the need for crack control steelmesh;
• increases in strength, durability, impact andabrasion resistance.
The other type of polypropylene used for manyyears is chopped twisted fibrillated twine (Figure43.5 (a)) typically in volumes of 0.44 per cent in
40mm lengths. This fibre has only been used forprecast products subject to impact loading suchas pile shells.
43.2.1 Mix design and manufacture
In the most used mixes the fibre volume is so low(typically 0.1 per cent) that mixing techniquesrequire little or no modification from normalpractice. Usually, the fibre comes prepackaged in0.9kg water-soluble bags which are placed in themixer for each cubic metre of concrete mixed.The fibres are released and dispersed during thenormal mixing cycle. For the higher fibre volumescontaining chopped twine, about 30 seconds ofmixing time is recommended to prevent the fibrebundles from breaking down into individual ele-ments.
Water-soluble bags containing a combinationof steel and polypropylene fibres are also mar-keted; these give some of the benefits of each fibretype.
43.2.2 Properties of fresh concrete
As discussed in Chapter 18, badly made andcured concrete may suffer from sedimentation ofthe aggregate particles with consequential bleed-ing of water to the surface of the cast layer. Withexcessive sun or wind, this water may rapidlyevaporate with an increase in suction and shrink-age on a horizontal surface which can cause sub-stantial cracks to form, known as plasticshrinkage cracks. Some tests have shown thatsmall quantities of fibres in the mix increase thecohesion and prevent sedimentation due to theirinterlocking network characteristics. The resultmay be that in some, but not all cases, the quan-tity of bleed water may be reduced. This will notnecessarily have a beneficial effect on plasticshrinkage cracking but may help to limit plasticsettlement cracking. The small reduction in bleed-ing where this is validated is unlikely to be due tothe additional specific surface area of the fibresbecause a similar surface area increase could alsobe gained by an additional 1kg/m3 of cement.
43.2.3 Plastic shrinkage cracking
This is a topic which is so complex that there isno agreed mechanism for plain concrete let alonean understanding of the effects of fibres. How-ever, it is an area where the earlier theory canindicate some possibilities.
For plastic shrinkage cracks to occur, it is pre-sumed that the tensile stress exceeds the concretetensile strength, which is very low during the first4 hours. It has been found that typical concretesmay have a tensile strength of about 0.02N/mm2
at 4 hours. In order that the cracking stress of0.02N/mm2 should be sustained by 0.1 per centof random fibres, the fibre stress can be calculatedfrom:
�f �2·�c /Vf �2�0.02/0.001�40MPa (43.1)
Sufficient bond stress with the polypropylenecan be developed during the first 4 to 5 hoursafter placing for fibre stresses of this order to bedeveloped so that the width of plastic shrinkagecracks can be limited by fibres. The fibres alsoendow the concrete with some post-cracking duc-tility and increased strain capacity at these veryearly ages which could influence plastic shrinkagecracking.
43.2.4 Properties of hardened concrete
For a system containing 0.1 per cent by volumeof polypropylene fibres it can easily be showntheoretically that the fibres will have little mea-surable affect on the tensile or flexural strength ofhardened concrete and that they cannot be con-sidered as a primary reinforcement. This isbecause the most optimistic estimate of criticalfibre volume from a rearranged equation (43.1) isfor the unlikely case of all fibres breaking (ratherthan pulling out) at a stress of 400MPa. Thus fora typical concrete tensile strength of 3MPa
Vfcrit �2�3/400�1.5%
More realistically �f 200MPa and hence Vfcrit
3%.The actual post-cracking stress for 0.1% Vf will
depend on the fibre–cement bond strength.
Fibre-reinforced concrete
420
Assuming that an average fibre stress of 200MPacan be sustained then
�cu � �V
2f
� ·�f �0.0
2
01� ·2000.1MPa
thus giving minimal post-cracking strength in thehardened state.
This estimate is supported by experimentalevidence where the maximum pull-out forceacross a pre-formed crack is about 0.15MPa. In aground floor slab of the same dimensions asdescribed for steel fibres this would result in arestraining force of only 22.5kN/m (2.25T/m)width, which is much less than for the steel fibrecase.
Systems using 0.44 per cent by volume ofpolypropylene twine, as in shell piles and marinedefence units, contain sufficient volume of fibre toincrease the impact strength of the concrete byholding cracked sections together to enable themto withstand further impacts. The tensile andbending strengths are virtually unchanged butgreater ductility and toughness in the post-cracking zone compared with plain concrete canbe achieved.
43.2.5 Durability
Polypropylene is extremely resistant to the alkalisin concrete and the concrete matrix protects thefibres from ultra-violet light which otherwisewould cause chain scission and degradation.Little change in fibre strength has been observedup to 18 years in a variety of exposure environ-ments (Figure 42.10) and accelerated tests havepredicted a lifetime considerably in excess of 30years. However, for the early age properties, thedurability of the fibres is not critical and it is onlysecond order effects on hardened propertieswhich are affected by fibre durability. An import-ant consideration for the long-term toughness offibre concrete in general is the potential increasein bond strength with time which could lead tofibre fracture rather than fibre pull-out, thusreducing the toughness. However, for fibrillated
films, the bond strength has been shown to betime stable, thus giving increased confidence inlong-term composite performance.
43.2.6 Applications
Regardless of the theoretical technical merits ofusing 0.1 per cent by volume of polypropylenefibres, there is an increasing use in the industrialground floor industry for similar reasons as forsteel fibre concrete. Also, an indication of itsgrowing use in general construction is that nearly10 per cent of ready-mixed concrete in the USAcontains polypropylene fibres.
Concrete containing 0.44 per cent by volumeof chopped twine has been extensively used since1970 in piling shells, and in two tonne blocks assea defences. This type of application whereimpact loads form a substantial proportion of theapplied stress is likely to remain a useful andcontinuing specific use for chopped polypropy-lene twine.
43.3 Glass fibre-reinforced concreteThere are few uses for glass fibre reinforcement inbulk concrete because the cost of including a suf-ficient quantity of random glass fibre wouldrarely be cost effective. Also, significant mixingdifficulties would occur for the normal range ofconcretes at fibre volumes in excess of 1 per cent.
However, fibre quantities as low as 0.03 percent by volume (0.8kg/m3) of 12mm long fibrescan offer improved resistance to plastic shrinkagecracking for normal concretes in a similar way topolypropylene fibres. These mixes will inhibitbleeding and plastic shrinkage cracking and willhave between 2 million and 200 million fibres perm3 of concrete depending on whether the fila-ments are allowed to disperse from the strands.
43.4 ReferenceEdgington, J. (1973) Steel fibre reinforced concrete.
Ph.D. Thesis, University of Surrey.
Reference
421
422
Further reading
Ashbee, K. (1989) Fibre Reinforced Composites, Tech-nomic Publishing Co. Ltd, Lancaster, Pennsylvania:Basel.
Balaguru, P.N. and Shah, S.P. (1992) Fibre ReinforcedCement Composites, Mcgraw-Hill, New York.
Bentur, A. and Mindess, A. (1990) Fibre ReinforcedCementitious Composites, Elsevier Applied Science,London.
Clarke, J.L. (1993) Alternative Materials for the Rein-forcement and Prestressing of Concrete, E & FNSpon, London.
Fordyce, M.W. and Wodehouse, R.G. (1983) GRCand Buildings, Butterworth, London.
Hannant, D.J. (1978) Fibre Cements and Fibre Con-cretes, Wiley, Chichester, UK.
Hollaway, L. (1986) Pultrusion, Chapter 1 of Develop-ments in Plastic Technology – 3 (eds A. Whelan andJ.L. Craft), Elsevier Applied Science, London.
Hollaway, L. (ed.) (1990) Polymers and Polymer Com-posites in Construction, Thomas Telford, London.
Hollaway, L. (1993) Polymer Composites for Civil andStructural Engineering, Blackie Academic and Pro-fessional, London.
Hollaway, L. (ed.) (1994) BPF Handbook of PolymerComposites for Engineers, Woodhead PublishingLtd, Cambridge.
Hollaway, L.C. and Head, P.R. (2001) AdvancedPolymer Composites and Polymers in the CivilInfrastructure, Elsevier Science, Oxford.
Hollaway, L.C. and Leeming, M.B. (eds) (1999)Strengthening of Reinforced Concrete Structures –
Using Externally Bonded FRP Composites in Struc-tural and Civil Engineering, Woodhead PublishingLtd, Cambridge.
Hull, D. (1992) An Introduction to Composite Mater-ials, Cambridge University Press, Cambridge.
Majumdar, A.J. and Laws, V. (1991) Glass Fibre Rein-forced Cement, Blackwell Scientific Publications,London.
Meier, U. and Kaiser, H.P. (1991) Strengthening ofstructures with CFRP laminates. Proc. AdvancedComposite Materials in Civil Engineering Structures,Mats Div ASCE, Las Vegas, Jan. 1991, 224–32.
Phillips, L.N. (1989) Introduction. In Design withAdvanced Composite Materials (ed. L.N. Phillips),The Design Council, London.
Priestley, M.J.N., Seible, F. and Calvi, M. (1996),Seismic Design and Retrofit of Bridges, John Wiley& Sons, Inc., New York.
Richmond, B.R. and Head, P.R. (1988) Alternativematerials in long span bridge structures. KerenskyMemorial Conference, London, June.
Swamy, R.N. (ed.) (1988) Natural Fibre ReinforcedCement and Concrete, Blackie, London.
Triantafillou, T.C. and Plevris, N. (1995) Reliabilityanalysis of reinforced concrete beams strengthenedwith CFRP laminates. In Non-metallic (FRP) Rein-forcement for Concrete Structures (ed. L. Taerwe), E& FN Spon, London, pp. 576–83.
Weatherhead, R.C. (1980) FRP Technology: FibreReinforced Resin Systems, Applied Science Publish-ers, London.
Part Eight
Timber
J.M. Dinwoodie
Introduction
In the industrial era of the nineteenth centurytimber was used widely for the construction notonly of roofs but also of furniture, water-wheels,gearwheels, rails of early pit railways, sleepers,signal poles, bobbins and boats. The twentiethcentury saw an extension of its use in certainareas and a decline in others, due to its replace-ment by newer materials. Despite competitionfrom the lightweight metals and plastics, whetherfoamed or reinforced, timber continues to be usedon a massive scale.
World production of timber in 1993 (the lastyear for which complete data is available) was3400�106 m3. As much as 55 per cent of thisvolume (1880�106 m3) is used as fuelwood, butthere is still 1520�106 m3 used for industrial andconstructional purposes.
In 1997, the UK consumed 48.3�106 m3 oftimber, panels, paper and pulp based on anunderbark wood raw material equivalent basis(Anon, 1998); this was equivalent to a per capitaannual consumption of 0.82m3. Consumption oftimber and wood-based panels on a wood rawmaterial basis was 27.07�106 m3 comprising17.48�106 m3 of softwood, 1.86�106 m3 ofhardwood, and 7.73�106 m3 of wood-basedpanels.
About 80 per cent of consumption is met byimports; the cost of these for timber and wood-based panels (after deduction of a small volumeof re-exports) was £4843�106, a far frominsignificant import bill and one that is more than15 per cent of the total UK annual trade deficit.
The remaining 20 per cent of consumption ismet from home production, equivalent to5.51�106 m3 with a value of about £900�106.Although this contribution will increase steadily
over the next two decades, it is then expected topeak at a value corresponding to only 25 per centof expected consumption.
In the UK, timber and timber products are con-sumed by a large range of industries, but the bulkof the material continues to be used in construc-tion, either structurally, such as roof trusses orfloor joists (about 43 per cent of total consump-tion), or non-structurally, e.g. doors, windowframes, skirting boards and external cladding(about 9 per cent of total consumption). The con-struction industry, therefore, consumed in 1997timber and wood-based panels to a value ofabout £3000�106. On a volume basis, annualconsumption continues to increase slightly andthere is no reason to doubt that this trend will bemaintained in the future, especially with thedemand for more houses, the increasing price ofplastics, the favourable strength–weight andstrength–cost factors of timber and panel prod-ucts, and the increased emphasis on environ-mental performance and sustainability in whichtimber is the only renewable construction mater-ial.
Timber is cut and machined from trees, them-selves the product of nature and time. The struc-ture of the timber of trees has evolved throughmillions of years to provide a most efficientsystem which will support the crown, conductmineral solutions and store food material. Sincethere are approximately 30000 different speciesof tree, it is not surprising to find that timber isan extremely variable material. A quick mentalcomparison of the colour, texture and density ofa piece of balsa and a piece of lignum vitae, for-merly used to make playing bowls, will illustratethe wide range that occurs. Nevertheless, modern
425
society has found timber to be a cheap and effect-ive material and, as we have seen, continues touse it in vast quantities. However, we must neverforget that the methods by which we utilise thisproduct are quite different from the purpose thatnature intended and many of the criticisms lev-elled at timber as a material are a consequence ofour use or misuse of nature’s product. Unlike somany other materials, especially those used in theconstruction industry, timber cannot be manufac-tured to a particular specification. Instead thebest use has to be made of the material alreadyproduced, though it is possible from the widerange available to select timbers with the mostdesirable range of properties. Timber as a mater-ial can be defined as a low-density, cellular, polymeric composite, and as such does not conve-niently fall into any one class of material, rathertending to overlap a number of classes. In termsof its high strength performance and low cost,timber remains the world’s most successful fibrecomposite.
Four orders of structural variation can berecognised – macroscopic, microscopic, ultra-structural and molecular – and in subsequentchapters the various physical and mechanicalproperties of timber will be related to these fourlevels of structure. In seeking correlationsbetween performance and structure, it is temptingto describe the latter in terms of smaller andsmaller structural units. Whilst this desire forrefinement is to be encouraged, a cautionary notemust be recorded, for it is all too easy to overlookthe significance of the gross features. This isparticularly so where large sections of timber arebeing used under practical conditions; in thesesituations gross features such as knots and grainangle are highly significant factors in reducingperformance.
ReferenceAnon (1998) A reference for the forestry industry.
1997 Yearbook of the Forestry Industry Council ofGreat Britain, Stirling, 72 pp.
Introduction
426
Chapter forty-four
Structure oftimber and thepresence ofmoisture
44.1 Structure at the macroscopic level44.2 Structure at the microscopic level44.3 Molecular structure and ultrastructure44.4 Variability in structure44.5 Appearance of timber in relation to its structure44.6 Mass–volume relationships44.7 Moisture in timber44.8 Flow in timber44.9 References
44.1 Structure at the macroscopiclevelThe trunk of a tree has three physical functions toperform; first, it must support the crown, a regionresponsible for the production not only of food,but also of seed; second, it must conduct themineral solutions absorbed by the roots upwardsto the crown; and third, it must store manufac-tured food (carbohydrates) until required. As willbe described in detail later, these tasks are per-formed by different types of cell.
Whereas the entire cross-section of the trunkfulfils the function of support, and increasingcrown diameter is matched with increasing dia-meter of the trunk, conduction and storage arerestricted to the outer region of the trunk. Thiszone is known as sapwood, while the region inwhich the cells no longer fulfil these tasks is termed
the heartwood. The width of sapwood varieswidely with species, rate of growth and age of thetree. Thus, with the exception of very young trees,the sapwood can represent from 10 to 60 per centof the total radius, though values from 20 to 50per cent are more common (Figures 44.1 and44.2); in very young trees, the sapwood willextend across the whole radius. The advancementof the heartwood to include former sapwood cellsresults in a number of changes, primarily chemicalin nature. The acidity of the wood increasesslightly, though certain timbers have heartwood ofvery high acidity. Substances, collectively calledextractives, are formed in small quantities andthese impart not only coloration to the heartwood,but also resistance to both fungal and insectattack. Different substances are found in differentspecies of wood and some timbers are devoid ofthem altogether: this explains the very wide rangein the natural durability of wood about whichmore will be said in Section 47.3.1. Many timbersdevelop gums and resins in the heartwood whilethe moisture content of the heartwood of mosttimbers is appreciably lower than that of thesapwood in the freshly felled state. However, inexceptional cases high moisture contents can occurin certain parts of the heartwood. Known aswetwood, these zones are frequently of a darker
427
colour than the remainder of the heartwood andare thought to be due to the presence of micro-organisms which produce aliphatic acids and gases(Ward and Zeikus, 1980: Hillis, 1987).
With increasing radial growth of the trunk bydivision of the cambial cells, commensurateincreases in crown size occur, resulting in theenlargement of existing branches and the produc-tion of new ones; crown development is not onlyoutwards but upwards. Radial growth of thetrunk must accommodate the existing branchesand this is achieved by the structure that weknow as the knot. If the cambium of the branchis still alive at the point where it fuses with thecambium of the trunk, continuity in growth willarise even though there will be a change in orien-tation of the cells. The structure so formed istermed a green or live knot (Figure 44.3). If,however, the cambium of the branch is dead, andthis frequently happens to the lower branches,there will be an absence of continuity, and the
Structure of timber and the presence of moisture
428
FIGURE 44.1 Diagrammatic illustration of a wedge-shaped segment cut from a five-year-old hardwood tree,showing the principal structural features (© Building Research Establishment).
FIGURE 44.2 Cross-section through the trunk of aDouglas fir tree showing the annual growth rings, thedarker heartwood, the lighter sapwood and the bark(© Building Research Establishment).
Structure at the microscopic level
429
FIGURE 44.3 Green or live knot showing con-tinuity in structure between the branch and tree trunk(© Building Research Establishment).
44.2 Structure at the microscopiclevelThe cellular structure of wood is illustrated inFigures 44.5 and 44.6. These three-dimensionalblocks are produced from micrographs of samplesof wood 0.8�0.5�0.5mm in size removed from aconiferous tree, known technically as a softwood(Figure 44.5), and a broad-leaved tree, known as a
trunk will grow round the dead branch, oftencomplete with its bark. Such a knot is termed ablack or dead knot (Figure 44.4), and will fre-quently drop out of planks on sawing. The graindirection in the vicinity of knots is frequently dis-torted and in a later section the loss of strengthdue to different types of knots will be discussed.
FIGURE 44.4 Black or dead knot surrounded bythe bark of the branch and hence showing discontinu-ity between branch and tree trunk (© BuildingResearch Establishment).
FIGURE 44.5 Cellular arrangement in a softwood(Pinus sylvestris – Scots pine, redwood) (© BuildingResearch Establishment).
FIGURE 44.6 Cellular arrangement in a ring-porous hardwood (Quercus robur – European oak) (©Building Research Establishment).
hardwood (Figure 44.6). In the softwoods about 90per cent of the cells are aligned in the vertical axis,while in the hardwoods there is a much widerrange in the percentage of cells that are vertical(80–95 per cent); the remaining percentage ispresent in bands, known as rays, aligned in one ofthe two horizontal planes known as the radialplane or quarter-sawn plane (Figure 44.1). Thismeans that there is a different distribution of cellson the three principal axes and this is one of thetwo principal reasons for the high degree ofanisotropy present in timber.
It is popularly believed that the cells of wood areliving cells: this is certainly not the case. Wood cellsare produced by division of the cambium, a zone ofliving cells which lies between the bark and thewoody part of the trunk and branches (Figure 44.1).In the winter the cambial cells are dormant andgenerally consist of a single circumferential layer.With the onset of growth in the spring, the cells inthis single layer subdivide radially to form a cambialzone some ten cells in width. This is achieved by theformation within each dividing cell of a new verticalwall called the primary wall. During the growingseason these cells undergo further radial subdivisionto produce what are known as daughter cells andsome of these will remain as cambial cells whileothers, to the outside of the zone, will develop intobark or, if on the inside, will change into wood.There is thus a constant state of flux within thecambial zone with the production of new cells andthe relegation of existing cambial cells to bark orwood. Towards the end of the growing season theemphasis is on relegation and a single layer ofcambial cells is left for the winter period.
To accommodate the increasing diameter of thetree, the cambial zone must increase circumferen-tially and this is achieved by the periodic tangen-tial division of the cambial cells. In this case, thenew wall is sloping and subsequent elongation ofeach half of the cell results in cell overlap, oftenfrequently at shallow angles to the vertical axis,giving rise to the formation of spiral grain in thetimber. The rate at which the cambium dividestangentially has a significant effect on the averagecell length of the timber produced.
The daughter cells produced radially from the
Structure of timber and the presence of moisture
430
cambium undergo a series of changes extendingover a period of about three weeks; this process isknown as differentiation. Changes in cell shape areparalleled with the formation of the secondarywall, the final stages of which are associated withthe death of the cell; the degenerated cell contentsare frequently to be found lining the cell cavity. It isduring the process of differentiation that the stan-dard daughter cell is transformed into one of fourbasic cell types (Table 44.1).
Chemical dissolution of the lignin–pectincomplex cementing the cells together will result intheir separation and this is a useful technique forseparating and examining individual cells. In thesoftwood (Figure 44.7) two types of cell can beobserved. Those present in greater number areknown as tracheids, some 2–4mm in length withan aspect ratio (L/D) of about 100:1. These cells,which lie vertically in the tree trunk, are respons-ible for both the supporting and conducting roles.The small block-like cells, some 200�30�m insize, known as parenchyma, are mostly located inthe rays and are responsible for the storage offood material.
In contrast, in the hardwoods (Figure 44.8), fourtypes of cell are present, albeit that one, thetracheid, is present in small amounts. The role of storage is again primarily taken by theparenchyma, which can be present horizontally inthe form of a ray, or vertically, either scattered orin distinct zones. Support is effected by long, thincells with very tapered ends, known as fibres; theseare usually about 1–2mm in length with an aspectratio of about 100:1. Conduction is carried out incells whose end walls have been dissolved awayeither completely or in part. These cells, known asvessels or pores, are usually short (0.2–1.2mm) andrelatively wide (up to 0.5mm) and, when situatedabove one another, form an efficient conductingtube. It can be seen, therefore, that while in thesoftwoods the three functions are performed bytwo types of cell, in the hardwoods each function isperformed by a single cell type (Table 44.1).
Although all cell types develop a secondarywall, this varies in thickness, being related to thefunction that the cell will perform. Thus, the wallthickness of fibres is several times that of the
Structure at the microscopic level
431
TABLE 44.1 The functions and wall thicknesses of the various types of cell found in softwoods and hard-woods
Cells Softwood Hardwood Function Wall thickness
Parenchyma ✓ ✓ Storage
Tracheids ✓ ✓ Support conduction
Fibres ✓ Support
Vessels (pores) ✓ Conduction
FIGURE 44.7 Individualsoftwood cells (�20) (©Building Research Establish-ment).
vessel (Table 44.1). Consequently, the density ofthe wood, and hence many of the strength prop-erties, will be related to the relative proportionsof the various types of cell. Density, of course,will also be related to the absolute wall thicknessof any one type of cell, for it is possible to obtainfibres of one species of wood with a cell wallseveral times thicker than those of another. Therange in density of timber is from about 120 to1200kg/m3 corresponding to pore volumes from92 per cent to 18 per cent (see Section 44.6.1).
Growth may be continuous throughout theyear in certain parts of the world and the woodformed tends to be uniform in structure. In thetemperate and subarctic regions and in parts ofthe tropics growth is seasonal, resulting in theformation of growth rings; where there is a singlegrowth period each year these rings are referredto as annual rings (Figure 44.1).
When seasonal growth commences, the domin-ant function appears to be conduction, while inthe latter part of the year the dominant factor issupport. This change in emphasis manifests itself
Structure of timber and the presence of moisture
432
in the softwoods with the presence of thin-walledtracheids (about 2�m) in the early part of theseason (the wood being known as earlywood)and thick-walled (up to 10�m) and slightlylonger (10 per cent) in the latter part of theseason (the latewood) (Figure 44.5).
In some of the hardwoods, but certainly not allof them, the early-wood is characterised by thepresence of large-diameter vessels surrounded pri-marily by parenchyma and tracheids; only a fewfibres are present. In the latewood, the vessel dia-meter is considerably smaller (about 20 per cent)and the bulk of the tissue comprises fibres. It is notsurprising to find, therefore, that the technicalproperties of the earlywood and latewood arequite different from one another. Timbers withthis characteristic two-phase system are referred toas having a ring-porous structure (Figure 44.6).
The majority of hardwoods, whether of temper-ate or tropical origin, show little differentiation intoearlywood and latewood. Uniformity across thegrowth ring occurs not only in cell size, but also inthe distribution of the different types of cells (Figure44.9): these timbers are said to be diffuse-porous.
FIGURE 44.9 Cellular arrangement in a diffuse-porous hardwood (Fagus sylvatica – beech) (© BuildingResearch Establishment).
FIGURE 44.8 Individual cells from a ring-poroushardwood (�50) (© Building Research Establishment).
Structure at the microscopic level
433
FIGURE 44.10 Electronmicrograph of the softwoodbordered pit showing themargo strands supporting thediaphragm (torus), whichoverlaps the aperture (�3600)(© Building Research Estab-lishment).
In addition to determining many of the tech-nical properties of wood, the distribution of celltypes and their sizes is used as a means of timberidentification.
Interconnection by means of pits occursbetween cells to permit the passage of mineralsolutions and food in both longitudinal and hori-zontal planes. Three basic types of pit occur.Simple pits, generally small in diameter and takingthe form of straight-sided holes with a transversemembrane, occur between parenchyma andparenchyma, and also between fibre and fibre.Between tracheids, a complex structure known asthe bordered pit occurs (Figure 44.10; see alsoFigure 44.27(a) for sectional view). The entrance tothe pit is domed and the internal chamber is char-acterised by the presence of a diaphragm (the torus)which is suspended by thin strands (the margostrands). Differential pressure between adjacent tra-cheids will cause the torus to move against the pitaperture, effectively stopping flow. As will be dis-cussed later, these pits have a profound influenceon the degree of artificial preservation of thetimber. Similar structures are to be found intercon-
necting vessels in a horizontal plane. Betweenparenchyma cells and tracheids or vessels, semi-bordered pits occur and are often referred to as raypits. These are characterised by the presence of adome on the tracheid or vessel wall and the absenceof such on the parenchyma wall: a pit membrane ispresent, but the torus is absent. Differences in theshape and size of these pits is an important diag-nostic feature in the softwoods.
The general arrangement of the verticallyaligned cells is referred to as grain. While it isoften convenient when describing timber at ageneral level to regard these cells as lying trulyvertical, this is not really true in the majority ofcases; these cells generally deviate from the verti-cal axis in a number of different patterns.
In many timbers, and certainly in most of thesoftwoods, the direction of the deviation from thevertical axis is consistent and the cells assume adistinct spiral mode which may be either left- orright-handed. In young trees the helix is usuallyleft-handed and the maximum angle which is nearto the core is frequently of the order of 4°, thoughconsiderable variability occurs both within a
species and also between different species oftimber. As the trees grow, so the helix angle in theouter rings decreases to zero and quite frequentlyin very large trees the angle in the outer rings sub-sequently increases, but the spiral has changedhand. Spiral grain has very significant technicalimplications; strength is lowered, while the degreeof twisting on drying and the amount of pick-upon machining increase as the degree of spirality ofthe grain increases (Brazier, 1965).
In other timbers the grain can deviate from thevertical axis in a number of more complex ways,of which interlocked and wavy are perhaps thetwo most common and best known. Since each ofthese types of grain deviation give rise to a char-acteristic decorative figure, further discussion ongrain is reserved until Section 44.5.
44.3 Molecular structure andultrastructure
44.3.1 Chemical constituents
Chemical analysis reveals the existence of fourconstituents and provides data on their relativeproportions. This information may be summar-ised as in Table 44.2: proportions are for timberin general and slight variations in these can occurbetween timber of different species, or in differentparts of a single tree.
Cellulose
Cellulose (C6H10O5)n occurs in the form of long,slender filaments or chains, these having been builtup within the cell wall from the glucose monomer(C6H12O6). Whilst the number of units per cellu-lose molecule (the degree of polymerisation) canvary considerably even within one cell wall, it isthought that a value of 8000–10000 is a realisticaverage for the secondary cell wall while theprimary cell wall has a degree of polymerisation ofonly 2000–4000 (Simson and Timell, 1978). Theanhydroglucose unit C6Hl0O5 which is not quiteflat, is in the form of a six-sided ring consisting offive carbon atoms and one oxygen atom (Figure44.11); the side groups play an important part inintra- and intermolecular bonding as will be notedlater. Successive glucose units are covalently linkedin the 1,4 positions giving rise to a potentiallystraight and extended chain; i.e. moving in a clock-wise direction around the ring, it is the first andfourth carbon atoms after the oxygen atom thatcombine with adjacent glucose units to form thelong-chain molecule. The anhydroglucose unitscomprising the molecule are not flat, as notedabove; rather they assume a chair configurationwith the hydroxyl groups (one primary and twosecondary) in the equatorial positions and thehydrogen atoms in the axial positions (Figure44.11).
Structure of timber and the presence of moisture
434
TABLE 44.2 Chemical composition of timber
Component Per cent mass Polymeric state Molecular Function
Softwood Hardwoodderivatives
Cellulose 42!2 45!2 Crystalline Glucose ‘fibre’highly orientedlarge linear molecule
Hemicelluloses 27!2 30!5 Semi-crystalline Galactosesmaller molecule Mannose
Xylose‘matrix’
Lignin 28!3 20!4 Amorphous Phenyl-propanelarge 3-D molecule
Extractives 3!2 5!4 Generally compounds Terpenes extraneoussoluble in organic Polyphenolssolvents Stilbenoids
Glucose, however, can be present in one of twoforms dependent on the position of the BOHgroup attached to carbon 1. When this group liesabove the ring, i.e. on the same side as that oncarbon 4, the unit is called �-glucose, and whenthis combines with an adjacent unit with theremoval of HBOBH (known as a condensationreaction) the resulting molecule is called starch, aproduct which is manufactured in the crown andstored in the parenchyma cells.
When the BOH group lies below the ring, theunit is known as �-glucose and on combiningwith adjacent units, again by a condensationreaction, a molecule of cellulose is produced inwhich alternate anhydroglucose units are rotatedthrough 180°: it is this product which is the prin-cipal wall-building constituent of timber.
Cellulose chains may crystallise in many ways,but one form, namely cellulose I, is characteristicof natural cellulosic materials. Over the yearsthere have been various attempts to model thestructure of cellulose I. One of the more recent,and one which has gained wide acceptance, isthat proposed by Gardner and Blackwell (1974).Using X-ray diffraction methods on the celluloseof Valonia, these authors proposed an eight-chainunit cell with all the chains running in the samedirection. Forty-one reflections were observed intheir X-ray diffractions and these were indexedusing a monoclinic unit cell having dimensionsa�1.634nm, b�1.572nm and c�1.038nm (thecell axis) with � �97°; the unit cell thereforecomprises a number of whole chains or parts ofchains totalling eight in number.
All but three of the reflections can be indexedby a two-chain unit cell almost identical to theearlier model by Meyer and Misch (1937),though this model had adjacent chains aligned inopposite directions. These three reflections arereported as being very weak, which means thatthe differences between the four Meyer andMisch unit cells making up the eight-chain cellmust be small. Gardner and Blackwell thereforetake a two-chain unit cell (a�0.817nm,b�0.786nm and c�1.038nm) as an adequateapproximation to the real structure. Their pro-posed model for cellulose I is shown in Figure44.12, which shows the chains lying in a parallelconfiguration, the centre chain staggered by0.266�c (�0.276nm).
Cellulose which has regenerated from a solutiondisplays a different crystalline structure and isknown as cellulose II: in this case there is completeagreement that the unit cell possesses an anti-parallel arrangement of the cellulose molecule.
Within the structure of cellulose I, bothprimary and secondary bonding are representedand many of the technical properties of wood canbe related to the variety of bonding present.Covalent bonding both within the glucose ringsand linking together the rings to form themolecular chain contributes to the high axialtensile strength of timber. There is no evidence ofprimary bonding laterally between the chains:rather this seems to be a complex mixture of thefairly strong hydrogen bonds and the weak vander Waals forces. The same OH groups that giverise to this hydrogen bonding are highly attractive
Molecular structure and ultrastructure
435
FIGURE 44.11 Structural formula for the cellulose molecule in its chair configuration (© Building ResearchEstablishment).
to water molecules and explain the affinity of cel-lulose for water. Whereas some earlier workers,though placing the intermolecular hydrogenbonds in the ac plane, recorded that theintramolecular hydrogen bonds were on a diago-nal plane thereby linking different layers,Gardner and Blackwell (1974) identify the exist-ence of both intermolecular and intramolecularhydrogen bonds, all of which, however, are inter-preted as lying only on the ac plane (Figure44.12); they consider the structure of cellulose asan array of hydrogen-bonded sheets held togetherby van der Waals forces across the cb plane.
The degree of crystallinity of the cellulose isusually assessed by X-ray and electron diffractiontechniques, though other methods have beenemployed. Generally, a value of about 60 per centis obtained, though values as high as 90 per centare recorded in the literature. This wide range invalues is due in part to the different techniquesemployed in the determination of crystallinityand in part to the fact that wood is comprised notjust of the crystalline and non-crystalline con-stituents, but rather a series of substances ofvarying crystallinity. Regions of complete crys-tallinity and regions with a total absence of crys-talline structure (amorphous zones) can berecognised, but the transition from one state tothe other is gradual.
The length of the cellulose molecule is about5000nm (0.005mm) whereas the average size ofeach crystalline region determined by X-rayanalysis is only 60nm in length, 5nm in widthand 3nm in thickness. This means that any cellu-lose molecule will pass through several regions ofhigh crystallinity – known as crystallites ormicelles – with intermediate non-crystalline orlow-crystalline zones in which the cellulose chainsare in only loose association with each other(Figure 44.12). Thus, the majority of chainsemerging from one crystallite will pass to thenext, creating a high degree of longitudinalcoordination; this collective unit is termed amicrofibril and has infinite length. It is clothedwith chains of cellulose mixed with chains ofsugar units other than glucose (see below) whichlie parallel, but are not regularly spaced. This
Structure of timber and the presence of moisture
436
brings the microfibril in timber to about 10nm inbreadth, and in some algae, such as Valonia, to30nm. The degree of crystallinity will thereforevary along its length and it has been proposedthat this could be periodic.
Readers desirous of a more comprehensiveaccount of the structure of cellulose are referredto Chapter 1 of Dinwoodie (2000).
Hemicelluloses and lignin
In Table 44.2 reference was made to the otherconstituents of wood additional to cellulose. Twoof these, the hemicelluloses and lignin, areregarded as cementing materials contributing tothe structural integrity of wood and also to itshigh stiffness. The hemicelluloses, like celluloseitself, are carbohydrates built up of sugar units,but unlike cellulose in the type of units they com-prise; these units differ between softwoods andhardwoods and generally, the total percentage ofthe hemicelluloses present in timber is greater inhardwoods compared with softwoods (Table44.2). Both the degree of crystallisation and thedegree of polymerisation of the hemicelluloses aregenerally low, the molecule containing less than200 units; in these respects, and also in their lackof resistance to alkali solutions, the hemicellu-loses are quite different from true cellulose (Siau,1984).
Lignin, present in about equal proportions tothe hemicelluloses, is chemically dissimilar tothese and to cellulose. Lignin is a complex, three-dimensional, aromatic molecule composed ofphenyl groups with a molecular weight of about11000. It is non-crystalline and the structurevaries between wood from a conifer and from abroad-leaved tree. About 25 per cent of the totallignin in timber is to be found in the middlelamella, an intercellular layer composed of ligninand pectin together with the primary cell wall.Since this compound middle lamella is very thin,the concentration of lignin is correspondinglyhigh (about 70 per cent). Deposition of the ligninin this layer is rapid.
The bulk of the lignin (about 75 per cent) ispresent within the secondary cell wall, having
437
FIGURE 44.12 Relationship between the structure of timber at different levels of magnitude. The lower twodiagrams are projections of the Gardner and Blackwell two-chain cell used as an approximation to the eight-chainunit cell of the real structure. On the left is the projection viewed perpendicular to the ac plane; on the right, theprojection viewed perpendicular to the ab plane (i.e. along the cell axis). Planes are characterised according toNorth American rather than European terminology. Note that the central chain (in black) has the same orienta-tion as the other chains and is staggered vertically with respect to them by an amount equal to c/4. (The lowertwo diagrams from K.H. Gardner and J. Blackwell (1974), by permission of John Wiley and Sons, Inc; the upperdiagrams adapted from J.F. Siau (1971), reproduced by permission of Syracuse University Press.)
Structure of timber and the presence of moisture
438
been deposited following completion of the cellu-losic framework. Initiation of lignification of thesecondary wall commences when the compoundmiddle lamella is about half completed andextends gradually inwards across the secondarywall (Saka and Thomas, 1982). Termination ofthe lignification process towards the end of theperiod of differentiation coincides with the deathof the cell. Most cellulosic plants do not containlignin and it is the inclusion of this substancewithin the framework of timber that is largelyresponsible for the stiffness of timber, especiallyin the dried condition.
Extractives
Before leaving the chemical composition of wood,mention must be made of the presence of extrac-tives (Table 44.2). This is a collective name for aseries of highly complex organic compoundswhich are present in certain timbers in relativelysmall amounts. Some, like waxes, fats and sugars,have little economic significance, but others, forexample rubber and resin (from which turpentineis distilled), are of considerable economic import-ance. The heartwood of timber, as described pre-viously, generally contains extractives which, inaddition to imparting coloration to the wood,bestow on it its natural durability, since most ofthese compounds are toxic to both fungi andinsects. Readers desirous of more information onextractives are referred to the comprehensive textby Hillis (1987).
Minerals
Elements such as calcium, sodium, potassium,phosphorous and magnesium are all componentsof new growth tissue, but the actual mass of theseinorganic materials is small and constitutes on thebasis of the oven-dry mass of the timber less than1 per cent for temperate woods and less than 5per cent for tropical timbers.
Certain timbers show a propensity to conductsuspensions of minerals which are subsequentlydeposited within the timber. The presence ofsilica in the rays of certain tropical timbers, and
calcium carbonate in the cell cavities of iroko aretwo examples where large concentrations of min-erals cause severe problems in log conversion andsubsequent machining.
Acidity
Wood is generally acidic in nature, the level ofacidity being considerably higher in the heart-wood compared to the sapwood of the same tree.The pH of the heartwood varies in differentspecies of timber, but is generally about 4.5 to5.5; however, in some timbers such as eucalypt,oak and western red cedar, the pH of theheartwood can be as low as 3.0. Sapwood gener-ally has a pH at least 1.0 higher than the corre-sponding heartwood; i.e. the acidity is at least tentimes lower than the corresponding heartwood.
Acidity in wood is due primarily to the genera-tion of acetic acid by hydrolysis of the acetylgroups of the hemicelluloses in the presence ofmoisture; this acidity in wood can cause severecorrosion of certain metals and care has to beexercised in the selection of metallic fixings espe-cially at higher relative humidities.
44.3.2 The cell wall as a fibre composite
In the introductory remarks, wood was defined asa natural composite and the most successfulmodel used to interpret the ultrastructure ofwood from the various chemical and X-ray analy-ses ascribes the role of ‘fibre’ to the cellulosicmicrofibrils while the lignin and hemicellulosesare considered as separate components of the‘matrix’. The cellulosic microfibril, therefore, isinterpreted as conferring high tensile strength tothe composite owing to the presence of covalentbonding both within and between the anhydro-glucose units. Experimentally it has been shown that reduction in chain length followinggamma irradiation markedly reduces the tensilestrength of timber; the significance of chainlength in determining strength has been con-firmed in studies of wood with inherently lowdegrees of polymerisation. While slippage
between the cellulose chains was previously con-sidered to be an important contributor to thedevelopment of ultimate tensile strength, this isnow thought to be unlikely due to the forcesinvolved in fracturing large numbers of hydrogenbonds.
Preston (1964) has shown that the hemicellu-loses are usually intimately associated with thecellulose, effectively binding the microfibrilstogether. Bundles of cellulose chains are thereforeseen as having a polycrystalline sheath of hemi-cellulose material and consequently the resultinghigh degree of hydrogen bonding would makechain slippage unlikely: rather it would appearthat stressing results in fracture of the CBOBClinkage.
The deposition of lignin is variable in differentparts of the cell wall, but it is obvious that itsprime function is to protect the hydrophilic(water-seeking) non-crystalline cellulose and thehemicelluloses which are mechanically weakwhen wet. Experimentally, it has been
demonstrated that removal of the lignin markedlyreduces the strength of wood in the wet state,though its reduction results in an increase in itsstrength in the dry state calculated on a net cellwall area basis. Consequently, the lignin isregarded as lying to the outside of the microfibrilforming a protective sheath.
Since the lignin is located only on the exteriorit must be responsible for cementing together thefibrils and in imparting shear resistance in thetransference of stress throughout the composite.The role of lignin in contributing towards stiff-ness of timber has already been mentioned.
There has been great debate over the years asto juxtaposition of the cellulose, hemicelluloseand lignin in the composition of a microfibril,and to the size of the basic unit. Two of the manymodels proposed are illustrated in Figure 44.13.The model on the left (a) depicts cellulosic sub-units some 3nm in diameter. These units, com-prising some 40 cellulose chains, are known aselementary fibrils or protofibrils. Gaps (1nm)
Molecular structure and ultrastructure
439
FIGURE 44.13 Models of the cross-section of a microfibril. In (a) the crystalline core has been subdivided intoelementary fibrils, while in (b) the core is regarded as being homogeneous ((a) adapted from D. Fengel (1970)reproduced by permission of TAPPI; (b) adapted from R.D. Preston (1974) reproduced by permission ofChapman and Hall).
between these units are filled with hemicellulosewhile more hemicellulose and lignin form thesheath. In passing, it is interesting to note thatsub-units as small as 1nm (sub-elementary fibril)have been claimed by some researchers.
In the second model (Figure 44.13(b)), thecrystalline core is considered to be about5nm�3nm containing about 48 chains in either4- or 8-chain unit cells; this latter configurationhas received wider acceptance. Both these models,however, are in some agreement in that passingoutwards from the core of the microfibril thehighly crystalline cellulose gives way first to thepartly crystalline layer containing mainly hemicel-lulose and non-crystalline cellulose, and then tothe amorphous lignin: this gradual transition ofcrystallinity from fibre to matrix results in highinterlaminar shear strength which contributesconsiderably to the high tensile strength andtoughness of wood.
44.3.3 Cell wall layers
When a cambial cell divides to form two daugh-ter cells a new wall is formed comprising themiddle lamella and two primary cell walls, one toeach daughter cell. These new cells undergochanges within about three days of their forma-tion and one of these developments will be theformation of a secondary wall. The thickness ofthis wall will depend on the function the cell willperform, as described earlier, but its basic con-struction will be similar in all cells.
Early studies on the anatomy of the cell wallused polarisation microscopy, which revealed thedirection of orientation of the crystalline regions.These studies indicated that the secondary wallcould be subdivided into three layers and measure-ments of the extinction position was indicative ofthe angle at which the microfibrils were orientated.Subsequent studies with transmission electronmicroscopy confirmed these findings and providedsome additional information with particular refer-ence to wall texture and variability of angle.However, much of our knowledge on microfibril-lar orientation has been derived using X-ray dif-fraction analysis. Most of these techniques yield
Structure of timber and the presence of moisture
440
TABLE 44.3 Microfibrillar orientation and percent-age thickness of the cell wall layers in spruce timber(Picea abies)
Wall layer Approximate Angle tothickness (%) longitudinal axis
P 3 RandomS1 10 50–70°S2 85 10–30°S3 2 60–90°
only mean angles for any one layer of the cell wall,but recent analysis has indicated that it may bepossible to determine the complete microfibrillarangle distribution of the cell wall (Cave, 1997).
The relative thickness and mean microfibrillarangle of the layers in a sample of spruce timberare illustrated in Table 44.3.
The middle lamella, a lignin–pectin complex, isdevoid of cellulosic microfibrils while in theprimary wall (P) the microfibrils are loose packedand interweave at random (Figure 44.14); nolamellation is present. In the secondary wall layersthe microfibrils are closely packed and parallel toeach other. The outer layer of the secondary wall,the S1, is again thin and is characterised by havingfrom four to six concentric lamellae, the microfib-rils of each alternating between a left- and right-hand spiral (S and Z helix) both with a pitch to thelongitudinal axis of from 50° to 70° depending onthe species of timber.
The middle layer of the secondary wall (S2) isthick and is composed of 30–150 lamellae, theclosely packed microfibrils of which all exhibit asimilar orientation in a right-hand spiral (Z helix)with a pitch of 10–30° to the longitudinal axis asillustrated in Figures 44.14 and 44.15. Since overthree-quarters of the cell wall is composed of theS2 layer, it follows that the ultrastructure of thislayer will have a very marked influence on thebehaviour of the timber. In later sections,anisotropic behaviour, shrinkage, tensile strengthand failure morphology will all be related to themicrofibrillar angle in the S2 layer.
Kerr and Goring (1975) were among the firstworkers to question the extent of these concentric
The S3 layer, which may be absent in certaintimbers, is very thin with only a few concentriclamellae; it is characterised, as in the Sl layer, byalternate lamellae possessing microfibrils orien-tated in opposite spirals with a pitch of 60–90°though the presence of the right-handed spiral (Zhelix) is disputed by some workers. Generally, theS3 has a looser texture than the S1 and S2 layersand is frequently encrusted with extraneous mater-ial. The S3, like the S1, has a higher concentration
Molecular structure and ultrastructure
441
FIGURE 44.14 Simplified structure of the cell wallshowing mean orientation of microfibrils in each of themajor wall layers (© Building Research Establishment).
FIGURE 44.15 Electron micrograph of the cell wallin Norway spruce timber (Picea abies) showing theparallel and almost vertical microfibrils of an exposedportion of the S2 layer (© Building Research Establish-ment).
lamellae in the S2 layer; these workers found thatthough there was a preferred orientation of ligninand carbohydrates in the S2 layer, the lamellae werecertainly not continuous. Thus, the interruptedlamellae model proposed by them embraced ligninand carbohydrate entities which were greater in thetangential than in the radial direction. Cellulosemicrofibrils were envisaged as being embedded in amatrix of hemicelluloses and lignin.
The existence within the S2 layer of concentriclamellae has been questioned again in the last fewyears. Evidence has been presented (Sell and Zin-nermann, 1993) from both electron and lightmicroscopy which indicates radial, or near radial,orientations of the transverse structure of the S2
layer. The transverse thickness of these agglomer-ations of microfibrils is 0.1–1.0nm and they fre-quently extend the entire width of the S2 layer. Amodified model of the cell wall of softwoods hasbeen proposed (Sell and Zinnermann, 1993).
of lignin than in the S2 (Saka and Thomas, 1982).Electron microscopy has also revealed the pres-ence of a thin, warty layer overlaying the S3 layerin certain timbers.
Investigations have indicated that the values ofmicrofibrillar angle quoted in Table 44.3 are onlyaverage for the layers and that systematic variationin angle occurs within each layer. Thus, Abe et al.(1991) have shown in Abies sachalinensis that themicrofibrillar angle of the secondary wall, as seenfrom the lumen, changed in a clockwise directionfrom the outermost S1 to the middle of the S2 andthen in a counter-clockwise direction to the inner-most S3. This resulted in the boundaries betweenthe three principal layers being very indistinct con-firming reports by previous workers on otherspecies and suggesting that the wall structure canbe viewed as a systematically varying continuum.
Microfibrillar angle appears to vary systematic-ally along the length of the cell as well as acrossthe wall thickness. Thus, the angle of the S2 layerhas been shown to decrease towards the ends ofthe cells, while the average S2 angle appears to berelated to the length of the cell, itself a functionof rate of growth of the tree. Systematical differ-ences in microfibrillar angle have been foundbetween radial and tangential walls and this hasbeen related to differences in degree of lignifica-tion between these walls. Openings occur in thewalls of cells and many of these pit openings arecharacterised by localised deformations of themicrofibrillar structure.
Further information on the variability ofmicrofibrillar angle is to be found in Dinwoodie(2000).
44.4 Variability in structureVariability in performance of wood is one of itsinherent deficiencies as a material. It will bediscussed later how differences in mechanicalproperties occur between timbers of differentspecies and how these are manifestations of dif-ferences in wall thickness and distribution of celltypes. However, superimposed on this geneticalsource of variation is both a systematic and anenvironmental one.
Structure of timber and the presence of moisture
442
There are distinct patterns of variation in manyfeatures within a single tree. Length of the cells,thickness of the cell wall and hence density, angleat which the cells are lying with respect to thevertical axis (spiral grain), angle at which themicrofibrils of the S2 layer of the cell wall arelocated with respect to the vertical axis, all showsystematic trends outwards from the centre of thetree to the bark and upwards from the base to thetop of the tree. This pattern results in the forma-tion of a core of wood in the tree with manyundesirable properties including low strength andhigh shrinkage. This zone, usually regarded assome ten to twenty growth rings in width, isknown as core wood or juvenile wood asopposed to the mature wood occurring outsidethis area. The boundary between juvenile andmature wood is usually defined in terms of thechange in slope of the variation in magnitude ofone anatomical feature (e.g. cell length, density)when plotted against ring number from the pith.
Environmental factors have considerable influ-ence on the structure of the wood and anyenvironmental influence, including forest manage-ment, which changes the rate of growth of thetree will affect the technical properties of thewood. However, the relationship is a complexone; in softwoods, increasing growth rate gener-ally results in an increase in the width of early-wood with a resulting decrease in density andmechanical properties. In diffuse-porous hard-woods increasing growth rate, provided it is notexcessive, has little effect on density, while inring-porous hardwoods, increasing rate ofgrowth, again provided it is not excessive, resultsin an increase in the width of latewood and con-sequently in density and strength.
There is a whole series of factors which maycause defects in the structure of wood and conse-quent lowering of its strength. Perhaps the mostimportant defect with regard to its utilisation isthe formation of reaction wood. When trees areinclined to the vertical axis, usually as a result ofwind action or of growing on sloping ground, thedistribution of growth-promoting hormones isdisturbed, resulting in the formation of an abnor-mal type of tissue. In the softwoods, this reaction
tissue grows on the compression side of the trunkand is characterised by having a higher thannormal lignin content, a higher microfibrillarangle in the S2 layer resulting in increased longitu-dinal shrinkage, and a generally darker appear-ance (Figure 44.16): this abnormal timber,known as compression wood, is also considerablymore brittle than normal wood. In the hard-woods, reaction wood forms on the tension sideof trunks and large branches and is thereforecalled tension wood. It is characterised by thepresence of a gelatinous cellulosic layer (the Glayer) to the inside of the cell wall; this results ina higher than normal cellulose content to thetotal cell wall which imparts a rubbery character-istic to the fibres resulting in difficulties in sawingand machining.
One other defect of considerable technicalsignificance is brittleheart, which is found inmany low-density tropical hardwoods and is onemanifestation of the presence of longitudinalgrowth stresses in large diameter trees. Yield ofthe cell wall occurs under longitudinal compres-sion with the formation of shear lines through thecell wall and throughout the core wood; compres-sion failure will be discussed in greater detail inSection 46.7.1.
More information on the variability in struc-ture and its influence on the technical perform-ance of timber is to be found in Chapters 5 and12 of Desch and Dinwoodie (1996).
44.5 Appearance of timber inrelation to its structureMost readers will agree that many timbers areaesthetically pleasing and the various andcontinuing attempts to simulate the appearanceof timber in the surface of synthetic materialsbears testament to the very attractive appearanceof many timbers. Although a very large propor-tion of the timber consumed in the UK is usedwithin the construction industry, where thenatural appearance of timber is of little con-sequence, excepting the use of hardwoods forflush doors, internal panelling and wood-blockfloors, a considerable quantity of timber is still
Appearance of timber in relation to its structure
443
FIGURE 44.16 A band of compression wood(centre left) in a Norway spruce plank, illustrating thedarker appearance and higher longitudinal shrinkageof the reaction wood compared with the adjacentnormal wood (© Building Research Establishment).
utilised purely on account of its attractive appear-ance particularly for furniture and various sportsgoods. The decorative appearance of manytimbers is due to the texture, or to the figure, orto the colour of the material and, in manyinstances, to combinations of these.
44.5.1 Texture
The texture of timber depends on the size of thecells and on their arrangement. A timber such asboxwood in which the cells have a very smalldiameter is said to be fine-textured, while acoarse-textured timber such as keruing has a con-siderable percentage of large-diameter cells.Where the distribution of the cell-types or sizesacross the growth ring is uniform, as in beech, orwhere the thickness of the cell wall remains fairlyconstant across the ring, as in some of the soft-woods, e.g. yellow pine, the timber is described asbeing even-textured: conversely, where variationoccurs across the growth ring, either in distribu-tion of cells as in teak or in thickness of the cellwalls as in larch or Douglas fir, the timber is saidto have an uneven texture.
44.5.2 Figure
Figure is defined as the ‘ornamental markingsseen on the cut surface of timber, formed by thestructural features of the wood’, but the term isalso frequently applied to the effect of markedvariations in colour. The four most importantstructural features inducing figure are grain,growth rings, rays and knots.
Grain
Mention was made in Section 44.2 that the cellsof wood, though often described as verticallyorientated, frequently deviate from this conve-nient concept. In the majority of cases this devia-tion takes the form of a spiral, the magnitude ofthe angle varying with distance from the pith.Although of considerable technical importancebecause of loss in strength and induced machin-ing problems, the common form of spiral grain
Structure of timber and the presence of moisture
444
has no effect on the figure presented on the fin-ished timber. However, two other forms of graindeviation do have a very marked influence on theresulting figure of the wood. Thus, in certainhardwood timbers, and the mahoganies areperhaps the best example, the direction of thespiral in the longitudinal–tangential plane alter-nates from left to right hand at very frequentintervals along the radial direction; grain of thistype is said to be interlocked. Tangential faces ofmachined timber will be normal, but the radialface will be characterised by the presence of alter-nating light and dark longitudinal bandsproduced by the reflection of light from thetapered cuts of fibres inclined in different direc-tions (Figure 44.17). This type of figure isreferred to as ribbon or stripe and is desirous intimber for furniture manufacture.
If instead of the grain direction alternatingfrom left to right within successive layers alongthe radial direction as above, the grain directionalternates at right angles to this pattern, i.e. in thelongitudinal–radial plane, a wavy type of grain isproduced. This is very conspicuous in machinedtangential faces where it shows up clearly asalternating light and dark horizontal bands(Figure 44.18); this type of figure is described asfiddleback, since timber with this distinctive typeof figure has been used traditionally for themanufacture of the backs of violins: it is to befound also on the panels and sides of expensivewardrobes and bookcases.
Growth rings
Where variability occurs across the growth ring,either in the distribution of the various cell typesor in the thickness of the cell walls, distinct pat-terns will appear on the machined faces of thetimber. Such patterns, however, will not beregular like many of the man-made imitations,but will vary according to changes in width of thegrowth ring and in the relative proportions ofearly and latewood.
On the radial face the growth rings will be ver-tical and parallel to one another, but on the tan-gential face a most pleasing series of concentric
arcs is produced as successive growth layers areintersected. In the centre part of the plank oftimber illustrated in Figure 44.19, the growthrings are cut tangentially forming these attractivearcs, while the edge of the board with paralleland vertical growth rings reflects timber cut radi-ally. In the case of ring-porous timbers, it is thepresence of the large earlywood vessels whichmakes the growth ring so conspicuous, while in
Appearance of timber in relation to its structure
445
FIGURE 44.17 Illustration of ribbon or stripefigure on the cut longitudinal–radial plane. The fibresin successive radial zones are inclined in opposite direc-tions (© Building Research Establishment).
FIGURE 44.18 ‘Fiddleback’ figure due to wavygrain (© Building Research Establishment).
timbers like Douglas fir or pitch pine, the strikingeffect of the growth ring can be ascribed to thevery thick walls of the latewood cells.
Rays
Another structural feature which may add to theattractive appearance of timber is the ray, espe-cially where, as in the case of oak, the rays areboth deep and wide. When the surface of theplank coincides with the longitudinal–radialplane, these rays can be seen as sinuous light-coloured ribbons running across the grain.
Structure of timber and the presence of moisture
446
FIGURE 44.19 The effect of growth rings on figure(© Building Research Establishment).
Knots
Knots, though troublesome from the mechanicalaspects of timber utilisation, can be regarded as adecorative feature; the fashion of knotty-pine fur-niture and wall panelling in the early 1970s is avery good example of how knots can be a decora-tive feature. However, as a decorative feature,knots do not possess the subtlety of variation ingrain and colour that arises from the other struc-tural features described above.
Exceptionally, trees produce a cluster of smallshoots at some point on the trunk and the timbersubsequently formed in this region contains amultitude of small knots. Timber from theseburrs is highly prized for decorative work, espe-cially if walnut or yew.
44.5.3 Colour
In the absence of extractives, timber tends to be arather pale straw colour which is characteristic ofthe sapwood of almost all timbers. The onset ofheartwood formation in many timbers is associ-ated with the deposition of extractives, most ofwhich are coloured, thereby imparting colorationto the heartwood zone. In passing, it should berecalled that although a physiological heartwoodis always formed in older trees, extractives arenot always produced; thus, the heartwood oftimbers such as ash and spruce is colourless.
Where coloration of the heartwood occurs, awhole spectrum of colour exists among the differ-ent species. The heartwood may be yellow, e.g.boxwood; orange, e.g. opepe; red, e.g. mahogany;purple, e.g. purpleheart; brown, e.g. Africanwalnut; green, e.g. greenheart; or black, e.g. ebony.In some timbers the colour is fairly evenly distrib-uted throughout the heartwood, while in otherspecies considerable variation in the intensity ofthe colour occurs. In zebrano, distinct dark brownand white stripes occur, while in olive woodpatches of yellow merge into zones of brown.Dark gum-veins, as present in African walnut, con-tribute to the pleasing alternations in colour. Vari-ations in colour such as these are regarded ascontributing to the ‘figure’ of the timber.
Mass–volume relationships
447
It is interesting to note in passing that the non-coloured sapwood is frequently coloured artifi-cially to match the heartwood, thereby adding tothe amount of timber converted from the log. In afew rare cases, the presence of certain fungi intimber in the growing tree can result in theformation of very dark coloured heartwood: theactivity of the fungus is terminated when thetimber is dried. Both brown oak and green oak,produced by different fungi, have always beenprized for decorative work.
44.6 Mass–volume relationships
44.6.1 Density
The density of a piece of timber is a function notonly of the amount of wood substance present,but also the presence of both extractives andmoisture. In a few timbers extractives are com-pletely absent, while in many they are present,but only in small amounts and usually less than 3per cent of the dry mass of the timber. In someexceptional cases, the extractive content may beas high as 10 per cent and in these cases it isnecessary to remove the extractives prior to thedetermination of density.
The presence of moisture in timber not onlyincreases the mass of the timber, but also resultsin the swelling of the timber, and hence bothmass and volume are affected. Thus, in the deter-mination of density where
� � �m
v� (44.1)
both the mass (m) and volume (v) must be deter-mined at the same moisture content. Generally,these two parameters are determined at zeromoisture content. However, as density is fre-quently quoted at a moisture content of 12 percent since this level is frequently experienced intimber in use, the value of density at zero mois-ture content is corrected for 12 per cent if volu-metric expansion figures are known, or else thedensity determination is carried out on timber at12 per cent moisture content.
Thus, if
mx �m0(1�0.01�) (44.2)
where mx is the mass of timber at moisturecontent x, m0 is the mass of timber at zero mois-ture content, and � is the moisture contentpercentage, and
vx �v0 (1�0.01 sv) (44.3)
where vx is the volume of timber at moisturecontent x, v0 is the volume of timber at zeromoisture content, and sv is the volumetric shrink-age/expansion percentage, it is possible to obtainthe density of timber at any moisture content interms of the density at zero moisture contentthus:
�x��m
vx
x�� ��0 � (44.4)
As a very approximate rule of thumb, the densityof timber increases by approximately 0.5 per centfor each 1.0 per cent increase in moisture contentup to 30 per cent. Density therefore will increase,slightly and curvilinearly, up to moisture contentsof about 30 per cent as both total mass andvolume increase; however, at moisture contentsabove 30 per cent, density will increase rapidlyand curvilinearly with increasing moisturecontent, since, as will be explained later in thischapter, the volume remains constant above thisvalue, whilst the mass increases.
The determination of density by measurementof mass and volume takes a considerable period oftime and over the years a number of quicker tech-niques have been developed for use where largenumbers of density determinations are required.These methods range from the assessment of theopacity of a photographic image which has beenproduced by either light or �-irradiation passingthrough a thin section of wood, to the use of amechanical device (the Pilodyn) which fires aspring-loaded bolt into the timber, after whichthe depth of penetration is measured. In all thesetechniques, however, the method or instrumenthas to be calibrated against density valuesobtained by the standard mass/volume technique.
1�0.01���1�0.01 sv
m0(1�0.01�)��v0(1�0.01 sv)
In Section 44.2, timber was shown to possessdifferent types of cell which could be characterisedby different values of the ratio of cell-wall thick-ness to total cell diameter. Since this ratio can beregarded as an index of density, it follows thatdensity of the timber will be related to the relativeproportions of the various types of cells. Density,however, will also reflect the absolute wall thick-ness of any one type of cell, since it is possible toobtain fibres of one species of timber, the cell wallthickness of which can be several times greaterthan that of fibres of another species.
Density, like many other properties of timber,is extremely variable; it can vary by a factor often, ranging from an average value at 12 per centm.c. of 176kg/m3 for balsa, to about 1230kg/m3
for lignum vitae (Figure 44.20). Balsa, therefore,has a density similar to that of cork, while lignumvitae has a density slightly less than half that ofconcrete or aluminium. The values of densityquoted for different timbers, however, are merelyaverage values: each timber will have a range ofdensities reflecting differences between early andlatewood, between the pith and outer rings, andbetween trees on the same site. Thus, forexample, the density of balsa can vary from 40 to320kg/m3.
In certain publications, reference is made to theweight of timber, a term widely used in com-merce; it should be appreciated that the quotedvalues are really densities.
44.6.2 Specific gravity
The traditional definition of specific gravity (G)(also known as relative density) can be expressed as:
G� ��
�
w
t� (44.5)
where �t is the density of timber, and �w is thedensity of water at 4°C�1.0000 g/ml. G willtherefore vary with moisture content and con-sequently the specific gravity of timber is usuallybased on the oven-dry mass, and volume at somespecified moisture content. This is frequentlytaken as zero though, for convenience, green orother moisture conditions are sometimes used
when the terms basic specific gravity and nominalspecific gravity are applied respectively. Hence:
G� ��V
m
��
0
w
� (44.6)
where m0 is the oven-dry mass of timber, V� isthe volume of timber at moisture content �, �w isthe density of water, and Gµ is the specific gravityat moisture content �.
At low moisture contents, specific gravitydecreases slightly with increasing moisturecontent up to 30 per cent, thereafter remainingconstant. In research activities specific gravity isdefined usually in terms of oven-dry mass andvolume. However, for engineering applicationsspecific gravity is frequently presented as the ratioof oven-dry mass to volume of timber at 12 percent moisture content; this can be derived fromthe oven-dry specific gravity, thus:
G12 � (44.7)
where G12 is the specific gravity of timber at 12per cent moisture content, G0 is the specificgravity of timber at zero moisture content, � isthe moisture content percentage, and Gs12 is thespecific gravity of bound water at 12 per centmoisture content.
The relationship between density and specificgravity can be expressed as
� �G(1�0.01�)�w (44.8)
where � is the density at moisture content �, G isthe specific gravity at moisture content �, and �w
is the density of water. Equation (44.8) is validfor all moisture contents. When � �0 the equa-tion reduces to
� �G0 (44.9)
i.e. density and specific gravity are numericallyequal.
44.6.3 Density of the dry cell wall
Although the density of timber may vary consid-erably among different timbers, the density of theactual cell wall material remains constant for
G0��1�0.01�G0/Gs12
Structure of timber and the presence of moisture
448
449
FIGURE 44.20 Mean densityvalues at 12 per cent moisture contentfor some common hardwoods andsoftwoods (© Building ResearchEstablishment).
all timbers with a value of approximately1500kg/m3 (1.5g/ml) when measured by volume-displacement methods.
The exact value for cell-wall density dependson the liquid used for measuring the volume:thus, densities of 1.525 and 1.451g/ml have beenrecorded for the same material using water andtoluene respectively.
44.6.4 Porosity
In Section 44. 2 the cellular nature of timber wasdescribed in terms of a parallel arrangement ofhollow tubes. The porosity (p) of timber isdefined as the fractional void volume and isexpressed mathematically as
p�1�Vf (44.10)
where Vf is the volume fraction of cell wall sub-stance.
The calculation of porosity is set out inChapter 3 of Dinwoodie (2000).
44.7 Moisture in timber
44.7.1 Equilibrium moisture content
Timber is hygroscopic, that is it will absorb mois-ture from the atmosphere if it is dry and, corre-spondingly, yield moisture to the atmospherewhen wet, thereby attaining a moisture contentwhich is in equilibrium with the water vapourpressure of the surrounding atmosphere. Thus,for any combination of vapour pressure and tem-perature of the atmosphere there is a correspond-ing moisture content of the timber such that therewill be no inward or outward diffusion of watervapour: this moisture content is referred to as theequilibrium moisture content (emc). Generally, itis more convenient to use relative humidity ratherthan vapour pressure. Relative humidity isdefined as the ratio of the partial vapour pressurein the air to the saturated vapour pressure,expressed as a percentage.
The fundamental relationships between mois-ture content of timber and atmospheric con-
ditions have been determined experimentally andthe average equilibrium moisture content valuesare shown graphically in Figure 44.21. A timberin an atmosphere of 20°C and 22 per cent relativehumidity will have a moisture content of 6 percent (see below), while the same timber if movedto an atmosphere of 40°C and 64 per cent rela-tive humidity will double its moisture content. Itshould be emphasised that the curves in Figure44.21 are average values for moisture in relationto relative humidity and temperature, and thatslight variations in the equilibrium moisturecontent will occur due to differences betweentimbers and to the previous history of the timberwith respect to moisture.
44.7.2 Determination of moisture content
It is customary to express the moisture content oftimber in terms of its oven-dry mass using theequation
� � �100 (44.11)
where minit � initial mass of timber sample (g),mod �mass of timber sample after oven-drying at105°C (g) and � �moisture content of timbersample (per cent).
The expression of moisture content of timberon a dry mass basis is in contrast to the pro-cedure adopted for other materials where mois-ture content is expressed in terms of the wet massof the material.
Determination of moisture content in timber isusually carried out using the basic gravimetrictechnique above, though it should be noted thatat least a dozen different methods have beenrecorded in the literature. Suffice it here tomention only two of these alternatives. First,where the timber contains volatile extractiveswhich would normally be lost during ovendrying, thereby resulting in erroneous moisturecontent values, it is customary to use a distillationprocess, heating the timber in the presence of awater-immiscible liquid such as toluene, and col-lecting the condensed water vapour in a
minit �mod��
mod
Structure of timber and the presence of moisture
450
Moisture in timber
451
FIGURE 44.21 Chart showing the relationship between the moisture content of timber and the temperatureand relative humidity of the surrounding air; approximate curves based on values obtained during drying fromgreen condition (Building Research Establishment).
calibrated trap. Second, where ease and speed ofoperation are preferred to extreme accuracy,moisture contents are assessed using electricmoisture meters. The type most commonly usedis known as the resistance meter, though thisbattery-powered hand-held instrument actuallymeasures the conductance or flow (the reciprocalof resistance) of an electric current between twoprobes. Below the fibre saturation point (about27 per cent moisture content, see later) anapproximately linear relationship exists betweenthe logarithm of conductance and the logarithmof moisture content. However, this relationship,which forms the basis for this type of meter,changes with species of timber, temperature andgrain angle. Thus, a resistance-type meter isequipped with a number of alternative scales,each of which relates to a different group oftimber species; it should be used at temperatures
close to 20°C with the pair of probes insertedparallel to the grain direction.
Although the measurement of moisture contentis quick with such a meter, there are, however,two drawbacks to its use. First, moisture contentis measured only to the depth of penetration ofthe two probes, a measurement which may not berepresentative of the moisture content of theentire depth of the timber member; the use oflonger probes can be beneficial, though these aredifficult to insert and withdraw. Second, theworking range of the instrument is only from 7 to27 per cent moisture content.
44.7.3 The moisture content of greentimber
In the living tree, water is to be found not only in the cell cavity, but also within the cell wall.
Consequently the moisture content of green wood(newly felled) is high, usually varying from about60 per cent to nearly 200 per cent depending onthe location of the timber in the tree and theseason of the year. However, seasonal variation isslight compared to the differences that occurwithin a tree between the sapwood and heart-wood regions. The degree of variation is illus-trated for a number of softwoods and hardwoodsin Table 44.4: within the former group thesapwood may contain twice the percentage ofmoisture to be found in the corresponding heart-wood, while in the hardwoods this difference isappreciably smaller or even absent. However,pockets of ‘wet’ wood can be found in the heart-wood as described in Section 44.1.
Green timber will yield moisture to the atmos-phere with consequent changes in its dimensions:at moisture contents above 20 per cent manytimbers, especially their sapwood, are susceptibleto attack by fungi: the strength and stiffness ofgreen wood is considerably lower than for thesame timber when dry. For all these reasons it isnecessary to dry or season timber followingfelling of the tree and prior to its use in service.
44.7.4 Removal of moisture from timber
Drying or seasoning of timber can be carried outin the open, preferably with a top cover.However, it will be appreciated from the previousdiscussion on equilibrium moisture contents that
the minimum moisture content that can beachieved is determined by the lowest relativehumidity of the summer period. In this country itis seldom possible to achieve moisture contentsless than 16 per cent by air seasoning. The planksof timber are separated in rows by stickers(usually 25–30mm across) which permit air cur-rents to pass through the pile; nevertheless it maytake from two to ten years to air-season timberdepending on the species of timber and the thick-ness of the timber members.
The process of seasoning may be accelerated arti-ficially by placing the stacked timber in a dryingkiln, basically a large chamber in which the temper-ature and humidity can be controlled and alteredthroughout the drying process: the control may becarried out manually or programmed automatically.Humidification is sometimes required in order tokeep the humidity of the surrounding air at adesired level when insufficient moisture is comingout of the timber; it is frequently required towardsthe end of the drying run and is achieved either bythe admission of small quantities of live steam or bythe use of water atomisers or low-pressure steamevaporators. Various designs of kilns are used andthese are reviewed in detail by Pratt (1974).
Drying of softwood timber in a kiln can beaccomplished in from four to seven days, theoptimum rate of drying varying widely from onetimber to the next; hardwood timber usually takesabout three times longer than softwood of the samedimension. Following many years of experimenta-
Structure of timber and the presence of moisture
452
TABLE 44.4 Average green moisture contents of the sapwood and heartwood
Moisture content (%)
Botanical name Commercial name Heartwood Sapwood
HardwoodsBetula lutea Yellow birch 64 68Fagus grandifolia American beech 58 79Ulmus americana American elm 92 84
SoftwoodsPseudotsuga menziesii Douglas fir 40 116Tsuga heterophylla Western hemlock 93 167Picea sitchensis Sitka spruce 50 131
tion, kiln schedules have been published for differ-ent groups of timbers; these schedules provide wet-and dry-bulb temperatures (maximum of 70°C) fordifferent stages in the drying process and their useshould result in the minimum amount of degrade interms of twist, bow, spring, collapse and checks(Pratt, 1974). Most timber is now seasoned bykilning: little air drying is carried out. Dry stress-graded timber in the UK must be kiln-dried to amean value of 20 per cent moisture content with nosingle piece greater than 24 per cent. However, UKand some Swedish mills are now targeting 12 percent (‘superdried’), as this level is much closer tothe moisture content in service.
Recently, solar kilns have become commerciallyavailable and are particularly suitable for use inthe developing countries to season many of the dif-ficult slow-drying tropical timbers. These smallkilns are very much cheaper to construct than con-ventional kilns and are also much cheaper to run.They are capable of drying green timber to about 7per cent moisture content in the dry season andabout 11 per cent in the rainy season.
44.7.5 Influence of structure
As previously mentioned in Section 44.7.3, waterin green or freshly felled timber is present both inthe cell cavity and within the cell wall. During theseasoning process, irrespective of whether this is byair or within a kiln, water is first removed fromwithin the cell cavity: this holds true down tomoisture contents of about 27–30 per cent. Sincethe water in the cell cavities is free, not beingchemically bonded to any part of the timber, it canreadily be appreciated that its removal will have noeffect on the strength or dimensions of the timber.The lack of variation of the former parameterwhen moisture content is reduced from 110 to 27per cent is illustrated in Figure 44.22.
However, at moisture contents below 27 percent water is no longer present in the cell cavity,but is restricted to the cell wall where it is chemi-cally bonded (hydrogen bonding) to the matrixconstituents, to the hydroxyl groups of the cellu-lose molecules in the non-crystalline regions and tothe surface of the crystallites; as such, this water is
Moisture in timber
453
FIGURE 44.22 Relationship between longitudinalcompressive strength and moisture content (© BuildingResearch Establishment).
referred to as bound water. The uptake of waterby the lignin component is considerably lowerthan that by either the hemicellulose or the amor-phous cellulose; water may be present as amonomolecular layer though frequently up to sixlayers can be present. Water cannot penetrate thecrystalline cellulose since the hygroscopic hydroxylgroups are mutually satisfied by the formation ofboth intra- and intermolecular bonds within thecrystalline region as described in Section 44.3.1.This view is confirmed from X-ray analyses whichindicate no change of state of the crystalline coreas the timber gains or loses moisture.
However, the percentage of non-crystallinematerial in the cell wall varies between 8 and 33per cent and the influence of this fraction of cellwall material as it changes moisture content on thebehaviour of the total cell wall is very significant.The removal of water from these areas within thecell wall results first in increased strength andsecond in marked shrinkage. Both changes can beaccounted for in terms of drying out of the water-reactive matrix, thereby causing the microfibrils to
come into closer proximity, with a commensurateincrease in interfibrillar bonding and decrease inoverall dimensions. Such changes are reversible, oralmost completely reversible.
44.7.6 Fibre saturation point
The increase in strength on drying is clearly indi-cated in Figure 44.22, from which it will be notedthat there is a threefold increase in strength as themoisture content of the timber is reduced fromabout 27 per cent to zero. The moisture contentcorresponding to the inflexion in the graph istermed the fibre saturation point (fsp), where intheory there is no free water in the cell cavitieswhile the walls are holding the maximum amountof bound water. In practice this rarely exists; a littlefree water may still exist while some bound wateris removed from the cell wall. Consequently, thefibre saturation ‘point’, while a convenient concept,should really be regarded as a ‘range’ in moisturecontents over which the transition occurs.
The fibre saturation point therefore corres-ponds in theory to the moisture content of thetimber when placed in a relative humidity of 100per cent; in practice, however, this is not so sincesuch an equilibrium would result in total satura-tion of the timber (Stamm, 1964). Values of emcabove 98 per cent are unreliable. It is generallyfound that the moisture content of hardwoods atthis level are from 1 to 2 per cent higher than forsoftwoods. At least nine different methods ofdetermining the fibre saturation point are recordedin the literature; the value of the fibre saturationpoint is dependent on the method used.
44.7.7 Sorption
Timber, as already noted in Section 44.7.1,assumes with the passage of time a moisturecontent which is in equilibrium with the relativevapour pressure of the atmosphere. This process ofwater sorption is typical of solids with a complexcapillary structure and this phenomenon has alsobeen observed in concrete. The similarity inbehaviour between timber and concrete withregard to moisture relationships is further illus-
trated by the presence of S-shaped isotherms whenmoisture content is plotted against relative vapourpressure. Both materials have isotherms whichdiffer according to whether the moisture content isreducing (desorption) or increasing (adsorption),thereby producing a hysteresis loop (Figure 44.23).
Readers desirous of more information on sorp-tion and diffusion, especially on the differenttheories of sorption, are referred to the compre-hensive text by Skaar (1988).
44.8 Flow in timberThe term flow is synonymous with the passage ofliquids through a porous medium such as timber,but the term is also applicable to the passage ofgases, thermal energy and electrical energy; it isthis wider interpretation of the term that isapplied in this chapter, albeit that the bulk of thechapter is devoted to the passage of both liquidsand gases (i.e. fluids).
The passage of fluids through timber can occurin one of two ways, either as bulk flow throughthe interconnected cell lumens or other voids, orby diffusion. The latter embraces both the trans-fer of water vapour through air in the lumens andthe movement of bound water within the cell wall(Figure 44.24). The magnitude of the bulk flow ofa fluid through timber is determined by its perme-ability.
Looking at the phenomenon of flow of moisturein wood from the point of view of the type ofmoisture, rather than the physical processesinvolved as described above, it is possible toidentify the involvement of three types of moisture:
(a) free water in the cell cavities giving rise tobulk flow above the fibre saturation point (seeSection 44.7.5);
(b) bound water within the cell walls whichmoves by diffusion below the fibre saturationpoint (see Section 44.7.5);
(c) water vapour which moves by diffusion in thelumens both above and below the fibre satura-tion point.
It is convenient when discussing flow of any typeto think of it in terms of being either constant or
Structure of timber and the presence of moisture
454
Flow in timber
455
FIGURE 44.23 Hysteresis loop resulting from the average adsorption and desorption isotherms for six speciesof timber at 40°C (© Building Research Establishment).
FLOW IN TIMBER
Liquids andgases (fluids)
Bulk flow(permeability)
Thermal energy
Diffusion(diffusion coefficient)
Electricalenergy
– viscous– turbulent– non-linear– molecular slip
(Knudsendiffusion)
Intergas diffusion– transfer of water
vapour throughair in the lumens
Bound waterdiffusion –within thecell wall
FIGURE 44.24 The different aspects of flow intimber which are covered in this chapter.
variable with respect to either time or locationwithin the specimen; flow under the former con-ditions is referred to as steady-state flow, whereaswhen flow is time and space dependent it is referredto as unsteady-state flow. Because of the complex-ity of the latter, only the former is covered in thistext; readers desirous of information on unsteady-state flow are referred to Siau (1984).
One of the most interesting features of steady-state flow in timber in common with many othermaterials is that the same basic relationshipholds, irrespective of whether one is concernedwith liquid or gas flow, diffusion of moisture, orthermal and electrical conductivity. The basicrelationship is that the flux or rate of flow is pro-portional to the pressure gradient:
�k (44.12)flux
�gradient
where flux is the rate of flow per unit cross-sectional area, gradient is the pressure differenceper unit length causing flow, and k is a constant,dependent on form of flow, e.g. permeability, dif-fusion or conductivity.
44.8.1 Bulk flow and permeability
Permeability is simply the quantitative expressionof the bulk flow of fluids through a porous mater-ial. Flow in the steady-state condition is bestdescribed in terms of Darcy’s law. Thus,
permeability� (44.13)
and for the flow of liquids, this becomes
k� (44.14)
where k is the permeability (cm2/atms), Q is thevolume rate of flow (cm3/s), �P is the pressuredifferential (atm), A is the cross-sectional area ofthe specimen (cm2), and L is the length of thespecimen in the direction of flow (cm).
Because of the change of pressure of a gas andhence its volumetric flow rate as it moves througha porous medium, Darcy’s law for the flow ofgases has to be modified as follows:
kg � (44.15)
where kg is the superficial gas permeability andQ, L, A and �P are as in equation (44.14), P isthe pressure at which Q is measured, and P� is themean gas pressure in the sample (Siau, 1984).
Of all the numerous physical and mechanicalproperties of timber, permeability is by far themost variable; when differences between timbersand differences between the principal directionswithin a timber are taken into consideration, therange is of the order of 107. Not only is perme-ability important in the impregnation of timberwith artificial preservatives, fire retardants andstabilising chemicals, but it is also significant inthe chemical removal of lignin in the manufactureof wood pulp and in the removal of free waterduring drying.
QLP�A�PP�
QL�A�P
flux�gradient
Structure of timber and the presence of moisture
456
Flow of fluids
The bulk of flow occurs as viscous (or laminar)flow in capillaries where the rate of flow is relat-ively low and when the viscous forces of the fluidare overcome in shear, thereby producing an evenand smooth flow pattern. In viscous flow Darcy’slaw is directly applicable, but a more specific rela-tionship for flow in capillaries is given by thePoiseuille equation, which for liquids is:
Q� (44.16)
where N is the number of uniform circularcapillaries in parallel, Q is the volume rate offlow, r is the capillary radius, �P is the pressuredrop across the capillary, L is the capillary lengthand is the viscosity. For gas flow, the aboveequation has to be modified slightly to take intoaccount the expansion of the gas along the pres-sure gradient. The amended equation is:
Q� (44.17)
where P� is the mean gas pressure within the capil-lary and P is the pressure of gas where Q wasmeasured. In both cases
Q' ��
L
P� (44.18)
or flow is proportional to the pressure gradient,which conforms with the basic relationship forflow.
Other types of flow can occur, e.g. turbulent,non-linear and molecular diffusion, the last men-tioned being of relevance only to gases. Informa-tion on all these types is to be found in Siau(1984).
Flow paths in timber
SoftwoodsBecause of their simpler structure and theirgreater economic significance, much more atten-tion has been paid to flow in softwood timbersthan in hardwood timbers. It will be recalledfrom Section 44.2 that both tracheids and
Nr4�PP���
8 LP
Nr4�P�
8 L
parenchyma cells have closed ends and thatmovement of liquids and gases must be by way ofthe pits in the cell wall. Three types of pit arepresent. The first is the bordered pit (Figure44.10) which is almost entirely restricted to theradial walls of the tracheids, tending to be locatedtowards the ends of the cells. The second type ofpit is the ray or semi-bordered pit which intercon-nects the vertical tracheid with the horizontal rayparenchyma cell, while the third type is thesimple pit between adjacent parenchyma cells.
For very many years it was firmly believed thatsince the diameter of the pit opening or of theopenings between the margo strands was verymuch less than the diameter of the cell cavity, andsince permeability is proportional to a powerfunction of the capillary radius, the bordered pitswould be the limiting factor controlling longitudi-nal flow. However, it has been demonstrated thatthis concept is fallacious and that at least 40 percent of the total resistance to longitudinal flow inAbies grandis sapwood that had been speciallydried to ensure that the torus remained in itsnatural position could be accounted for by theresistance of the cell cavity (Petty and Puritch,1970).
Both longitudinal and tangential flowpaths insoftwoods are predominantly by way of the bor-dered pits as illustrated in Figure 44.25, while thehorizontally aligned ray cells constitute the prin-cipal pathway for radial flow, though it has beensuggested that very fine capillaries within the cellwall may contribute slightly to radial flow. Therates of radial flow are found to vary very widelybetween species.
It is not surprising to find that the differentpathways to flow in the three principal axes resultin anisotropy in permeability. Permeability valuesquoted in the literature illustrate that for mosttimbers longitudinal permeability is about 104
times the transverse permeability; mathematicalmodelling of longitudinal and tangential flowsupports a degree of anisotropy of this order.Since both longitudinal and tangential flow insoftwoods are associated with bordered pits, agood correlation is to be expected between them;radial permeability is only poorly correlated with
Flow in timber
457
FIGURE 44.25 On the left, a representation of thecellular structure of a softwood in a longitudinal–tangential plane illustrating the significance of the bor-dered pits in both longitudinal and tangential flow; onthe right, softwood timber in the longitudinal radialplane, indicating the role of the ray cells in defining theprincipal pathway for radial flow (© Building ResearchEstablishment).
that in either of the other two directions and isfrequently found to be greater than tangentialpermeability.
Permeability is not only directionally depend-ent, but is also found to vary with moisturecontent, between earlywood and latewood,between sapwood and heartwood (Figure 44.26)and between species. In the sapwood of greentimber the torus of the bordered pit is usuallylocated in a central position and flow can be at amaximum (Figure 44.27(a)). Since the earlywoodcells possess larger and more frequent borderedpits, the flow through the earlywood is consider-ably greater than that through the latewood.However, on drying, the torus of the earlywoodcells becomes aspirated (Figures 44.10 and
44.27(b)), owing, it is thought, to the tensionstresses set up by the retreating water meniscus(Hart and Thomas, 1967). In this process themargo strands obviously undergo very consider-able extension and the torus is rigidly held in adisplaced position by strong hydrogen bonding
This displacement of the torus effectively sealsthe pit and markedly reduces the level of perme-ability of dry earlywood. In the latewood, thedegree of pit aspiration is very much lower ondrying than in the earlywood, a phenomenon thatis related to the smaller diameter and thicker cellwall of the latewood pit. Thus, in dry timber in
Structure of timber and the presence of moisture
458
FIGURE 44.27 Cross-section of a bordered pit inthe sapwood of a softwood timber: (a) in timber in thegreen condition with the torus in the ‘normal’ position;and (b) in timber in the dried state with the torus in anaspirated position (© Building Research Establishment).
marked contrast to green timber, the permeabilityof the latewood is at least as good as that of theearlywood and may even exceed it (Figure 44.26).Rewetting of the timber causes only a partialreduction in the number of aspirated pits and itappears that aspiration is mainly irreversible.
Quite apart from the fact that many earlywoodpits are aspirated in the heartwood of softwoods,the permeability of the heartwood is usuallyappreciably lower than that of the sapwood dueto the deposition of encrusting materials over thetorus and margo strands and also within the raycells (Figure 44.26).
Permeability varies widely among differentspecies of softwoods. Thus, Comstock (1970)found that the ratio of longitudinal-to-tangentialpermeability varied between 500:1 to 80000:1.Generally, the pines are much more permeablethan the spruces, firs or Douglas fir. This can beattributed primarily, though not exclusively, tothe markedly different type of semi-bordered pitpresent between the vertical tracheids and the rayparenchyma in the pines (fenestrate or pinoid
(a) (b)
FIGURE 44.26 The variation in rate of longitudinalflow through samples of green and dry earlywood andlatewood of Scots pine sapwood and heartwood (fromBanks (1968), © Building Research Establishment).
type) compared with the spruces, firs or Douglasfir (piceoid type).
HardwoodsThe longitudinal permeability is usually high inthe sapwood of hardwoods. This is because thesetimbers possess vessel elements, the ends of whichhave been either completely or partially dissolvedaway. Radial flow is again by way of the rays,while tangential flow is more complicated, relyingon the presence of pits interconnecting adjacentvessels, fibres and vertical parenchyma; however,intervascular pits in sycamore have been shownto provide considerable resistance to flow (Petty,1981). Transverse flow rates are usually muchlower than in the softwoods, but somewhat sur-prisingly, a good correlation exists between tan-gential and radial permeability; this is due in partto the very low permeability of the rays in hard-woods.
Since the effects of bordered pit aspiration, sodominant in controlling the permeability of soft-woods, are absent in hardwoods, the influence ofdrying on the level of permeability in hardwoodsis very much less than is the case with softwoods.
Permeability is highest in the outer sapwood,decreasing inwards and reducing markedly withthe onset of heartwood formation as the cellsbecome blocked either by the deposition of gumsor resins or, as happens in certain timbers, by theingrowth into the vessels of cell wall material ofneighbouring cells, a process known as theformation of tyloses.
Permeability varies widely among differentspecies of hardwoods. This variability is due inlarge measure to the wide variation in vesseldiameter that occurs among the hardwoodspecies. Thus, the ring-porous hardwoods whichare characterised with having earlywood vesselsthat are of large diameter, generally have muchhigher permeabilities than the diffuse-poroustimbers which have vessels of considerably lowerdiameter; however, in those ring-porous timbersthat develop tyloses (e.g. the white oaks) theirheartwood permeability may be lower than thatin the heartwood of diffuse-porous timbers. Inter-specific variability in permeability also reflects the
different types of pitting on the end walls of thevessel elements.
Timber and the laws of flow
The application of Darcy’s law to the permeabil-ity of timber is based on a number of assump-tions, not all of which are upheld in practice.Among the more important are that timber is ahomogeneous porous material and that flow isalways viscous and linear; neither of theseassumptions is strictly valid, but the Darcy lawremains a useful tool with which to describe flowin timber.
By de-aeration and filtration of their liquid,many workers have been able to achieve steady-state flow, the rate of which is inversely related tothe viscosity of the liquid, and to find that, invery general terms, Darcy’s law was upheld intimber (see e.g. Comstock, 1967).
Gas, because of its lower viscosity and the easewith which steady flow rates can be obtained, is amost attractive fluid for permeability studies.However, at low mean gas pressures, due to thepresence of slip flow, deviations from Darcy’s lawhave been observed by a number of investigators.At higher mean gas pressures, however, anapproximately linear relationship between con-ductivity and mean pressure is expected and this,too, has been observed experimentally. However,at even higher mean gas pressures, flow rate issometimes less than proportional to the appliedpressure differential due, it is thought, to theonset of non-linear flow. Darcy’s law may thusappear to be valid only in the middle range ofmean gas pressures.
44.8.2 Moisture diffusion
Flow of water below the fibre saturation pointembraces both the diffusion of water vapourthrough the void structure comprising the cellcavities and pit membrane pores and the diffusionof bound water through the cell walls (Figure44.24). In passing, it should be noted thatbecause of the capillary structure of timber,vapour pressures are set up and vapour can pass
Flow in timber
459
through the timber both above and below thefibre saturation point; however, the flow ofvapour is usually regarded as being of secondaryimportance to that of both bound and free water.
Moisture diffusion is another manifestation offlow, conforming with the general relationshipbetween flux and pressure. Thus, it is possible toexpress diffusion of moisture in timber at a fixedtemperature in terms of Fick’s first law, whichstates that the flux of moisture diffusion isdirectly proportional to the gradient of moistureconcentration; as such, it is analogous to theDarcy law on flow of fluids through porousmedia.
The total flux F of moisture diffusion through aplane surface under isothermal conditions is givenby
F� � �D. (44.19)
where dm/dt is the flux (rate of mass transfer perunit area), dc/dx is the gradient of moisture con-centration (mass per unit volume) in the x direc-tion, and D is the moisture diffusion coefficientwhich is expressed in m2/s (Siau, 1984; Skaar,1988).
Under steady-state conditions the diffusioncoefficient is given by
D� (44.20)
where m is the mass of water transported in timet, A is the cross-sectional area, L is the length ofthe wood sample, and �M is the moisture contentdifference driving the diffusion.
The vapour component of the total flux isusually much less than that for the bound water.The rate of diffusion of water vapour throughtimber at moisture contents below the fibre satu-ration point has been shown to yield coefficientssimilar to those for the diffusion of carbondioxide, provided corrections are made for differ-ences in molecular weight between the gases. Thismeans that water vapour must follow the samepathway through timber as does carbon dioxideand implies that diffusion of water vapour
through the cell walls is negligible in comparisonto that through the cell cavities and pits (Tarkowand Stamm, 1960).
Bound water diffusion occurs when water mole-cules bound to their sorption sites by hydrogenbonding receive energy in excess of the bondingenergy, thereby allowing them to move to newsites. At any one time the number of moleculeswith excess energy is proportional to the vapourpressure of the water in the timber at that moisturecontent and temperature. The rate of diffusion isproportional to the concentration gradient of themigrating molecules, which in turn is proportionalto the vapour pressure gradient.
The most important factors affecting the dif-fusion coefficient of water in timber are tempera-ture, moisture content and density of the timber.Thus, Stamm (1959) has shown that the boundwater diffusion coefficient of the cell wall sub-stance increases with temperature approximatelyin proportion to the increase in the saturatedvapour pressure of water, and increases exponen-tially with increasing moisture content at constanttemperature. The diffusion coefficient has alsobeen shown to decrease with increasing densityand to differ according to the method of determi-nation at high moisture contents. It is also depend-ent on grain direction; the ratio of longitudinal totransverse coefficients is approximately 2.5.
Various alternative ways of expressing thepotential which drives moisture through woodhave been proposed. These include percentagemoisture content, relative vapour pressure,osmotic pressure, chemical potential, capillarypressure and spreading pressure, the last men-tioned being a surface phenomenon derivablefrom the surface sorption theory of Dent whichin turn is a modification of the Brunauer-Emmet-Teller (BET) sorption theory (Skaar andBabiak, 1982). Although all this work has led tomuch debate on the correct flow potential, it hasno effect in the calculation of flow; moistureflow is the same irrespective of the potentialused, provided the mathematical conversionsbetween transport coefficients, potentials andcapacity factors are carried out correctly (Skaar,1988).
100mL�tA��M
dc�dx
dm�dt
Structure of timber and the presence of moisture
460
As with the use of the Darcy equation for per-meability, so with the application of Fick’s lawfor diffusion, there appears to be a number ofcases in which the law is not upheld and themodel fails to describe the experimental data.Claesson (1997) in describing some of the failuresof Fickian models claims that this is due to acomplicated, but transient sorption in the cellwall; it certainly cannot be explained by highresistance to flow of surface moisture.
The diffusion of moisture through wood hasconsiderable practical significance since it relatesto the drying of wood below the fibre saturationpoint, the day-to-day movement of wood throughdiurnal and seasonal changes in climate, and inthe quantification of the rate of vapour transferthrough a thin sheet such as the sheathing used intimber frame construction.
What is popularly called ‘vapour permeability’of a thin sheet, but is really vapour diffusion, isdetermined using the wet-cup test and its recipro-cal is now quantified in terms of the water vapourresistance factor (�). Values of � for timber rangefrom 30–50, while for wood-based panels, �ranges from 15 for particleboard to 130 for OSB(see also Dinwoodie, 2000).
44.8.3 Thermal conductivity
The basic law for flow of thermal energy isascribed to Fourier and when described mathe-matically is:
Kh � (44.21)
where Kh is the thermal conductivity for steady-state flow of heat through a slab of material, H isthe quantity of heat, t is time, A is the cross-sectional area, L is the length, and �T is thetemperature differential. This equation is analo-gous to that of Darcy for fluid flow.
Compared with permeability, where the Darcyequation was shown to be only partially valid fortimber, thermal flow is explained adequately bythe Fourier equation, provided the boundary con-ditions are defined clearly.
Thermal conductivity will increase slightly with
HL�tA�T
References
461
increased moisture content, especially when cal-culated on a volume-fraction-of-cell-wall basis;however, it appears that conductivity of the cellwall substance is independent of moisture content(Siau, 1984); at 12 per cent mc, the averagethermal conductivity of softwood timber parallelto the grain is of the order of 0.38W/mK. Con-ductivity is influenced considerably by the densityof the timber, i.e. by the volume-fraction-of-cell-wall substance, and various empirical and linearrelations between conductivity and density havebeen established. Conductivity will also vary withtimber orientation due to its anisotropic struc-ture: the longitudinal thermal conductivity isabout 2.5 times the transverse conductivity.Values of thermal conductivity are given in Siau(1984) and Dinwoodie (2000).
Compared with metals, the thermal conductiv-ity of timber is extremely low, though it is gener-ally up to eight times higher than that ofinsulating materials. The average transverse valuefor softwood timber (0.15W/mK) is about one-seventh that for brick, thereby explaining thelower heating requirements of timber housescompared with the traditional brick house.
Thermal insulation materials in the UK areusually rated by their U-value, where U is theconductance or the reciprocal of the thermalresistance. Thus
U-value�Kh/L (44.22)
where Kh is the thermal conductivity and L is thethickness of the material.
44.9 ReferencesAbe, H., Ohtani, J. and Fukazawa, K. (1991) FE-SEM
observations on the microfibrillar orientation in thesecondary wall of tracheids. IAWA Bull. new series12, 4, 431–8.
Banks, W.B. (1968) A technique for measuring thelateral permeability of wood. J. Inst. Wood Sci. 4, 2,35–41.
Brazier, J. (1965) An assessment of the incidence andsignificance of spiral grain in young conifer trees.For. Prod. J. 15, 8, 308–12.
Cave, I.D. (1997) Theory of x-ray measurement ofmicrofibril angle in wood. Part 2 The diffraction
diagram, x-ray diffraction by materials with fibretype symmetry. Wood Sci. Technol. 31, 225–34
Claesson, J. (1997) Mathematical modelling of mois-ture transport. Proc. International conference onwood-water relations (ed. P. Hoffmeyer), Copen-hagen, and published by the management committeeof EC COST Action E8, 61–8.
Comstock, G.L. (1967) Longitudinal permeability ofwood to gases and nonswelling liquids. For. Prod. J.17, 10, 41–6.
Desch, H.E. and Dinwoodie, J.M. (1996) Timber –Structure, Properties, Conversion and Use, 7th edn,Macmillan, Basingstoke, 306 pp.
Dinwoodie, J.M. (2000) Timber – Its Nature andBehaviour, 2nd edn, E & FN Spon, London, 257 pp.
Fengel, D. (1970) The ultrastructural behaviour of cellwall polysaccharides. In The Physics and Chemistryof Wood Pulp Fibres, TAPPI STAP 8, 74–96.
Gardner, K.H. and Blackwell, J. (1974) The structureof native cellulose. Biopolymers 13, 1975–2001.
Hart, C.A. and Thomas, R.J. (1967) Mechanism ofbordered pit aspiration as caused by capillarity. For.Prod. J. 17, 11, 61–8.
Hillis, W.E. (1987) Heartwood and Tree Exudates,Springer-Verlag, Berlin, 268 pp.
Kerr, A.J. and Goring, D.A.I. (1975) Ultrastructuralarrangement of the wood cell wall. Cellulose Chem.and Technol. 9, 6, 563–73.
Meyer, K.H. and Misch, L. (1937) Position des atomesdans le nouveau modele spatial de la cellulose. Hel-vetica Chimica Acta 20, 232–44.
Petty, J.A. (1981) Fluid flow through the vessels andintervascular pits of sycamore woods. Holz-forschung 35, 213–16.
Petty, J.A. and Puritch, G.S. (1970) The effects of dryingon the structure and permeability of the wood ofAbies grandis. Wood Sci. and Techol. 4, 2, 140–54.
Pratt, G.H. (1974) Timber Drying Manual, HMSO,London, 152 pp.
Preston, R.D. (1964) Structural and mechanical aspectsof plant cell walls. In The Formation of Wood inForest Trees (ed. H.M. Zimmermann), AcademicPress, New York, 169–88.
Preston, R.D. (1974) The Physical Biology of PlantCell Walls, Chapman & Hall, London, 491 pp.
Saka, S. and Thomas, R.J. (1982) A study of lignifica-tion in Loblolly pine tracheids by the SEM-EDXAtechnique. Wood Sci. Technol. 12, 51–62.
Sell, J. and Zinnermann, T. (1993) Radial fibrilagglomeration of the S2 on transverse fracture sur-faces of tracheids of tension-loaded spruce and whitefir. Holz, Roh-Werkstoff 51, 384.
Siau, J.F. (1971) Flow in Wood, Syracuse UniversityPress, Syracuse, New York, 99 pp.
Siau, J. (1984) Transport Processes in Wood, Springer-Verlag, Berlin, 245 pp.
Simson, B.W. and Timell, T.E. (1978) Polysaccharidesin cambial tissues of Populus tremuloides and Tiliaamericana. V. Cellulose. Cellul. Chem. Technol. 12,51–62.
Skaar, C. (1988) Wood–Water Relations, Springer-Verlag, Berlin, 283 pp.
Skaar, C. and Babiak, M. (1982) A model for bound-water transport in wood. Wood Sci. Technol. 16,123–38
Stamm, A.J. (1959) Bound water diffusion into woodin the fiber direction. For. Prod. J. 9, 1, 27–32.
Stamm, A.J. (1964) Wood and Cellulose Science,Ronald, New York, 549 pp.
Tarkow, H. and Stamm, A.J. (1960) Diffusion throughthe air filled capillaries of softwoods: I, Carbondioxide; II, Water vapour. For. Prod. J. 10, 247–50and 323–4.
Ward, J.C. and Zeikus, J.G. (1980) Bacteriological,chemical and physical properties of wetwood inliving trees. In Natural Variations of Wood Proper-ties (ed. J. Bauch). Mitt Bundesforschungsanst Forst-Holzwirtsch, 131: 133–66.
Structure of timber and the presence of moisture
462
Chapter forty-five
Deformation intimber
45.1 Introduction45.2 Dimensional change due to moisture45.3 Thermal movement45.4 Deformation under load45.5 References
45.1 IntroductionTimber may undergo dimensional changes solelyon account of variations in climatic factors; onthe other hand, deformation may be due solely tothe effects of applied stress. Frequently stress andclimate interact to produce enhanced levels ofdeformation.
This chapter commences by examining thedimensional changes that occur in timber follow-ing variations in its moisture content and/or tem-perature. The magnitude and, consequently, thesignificance of such changes in the dimensions oftimber are much greater in the case of alterationsin moisture content compared with temperature.Consequently, the greater emphasis in this firstsection is placed on the influence of changingmoisture content. Later in the chapter the effect ofstress on deformation will be examined in detail.
45.2 Dimensional change due tomoistureIn timber it is customary to distinguish betweenthose changes that occur when green timber isdried to very low moisture contents (e.g. 12 percent), and those that arise in timber of low mois-ture content due to seasonal or daily changes in
the relative humidity of the surrounding atmo-sphere. The former changes are called shrinkagewhile the latter are known as movement.
45.2.1 Shrinkage
As explained in Chapter 44, removal of waterfrom below the fibre saturation point occurswithin the amorphous region of the cell wall andmanifests itself by reductions in strength and stiff-ness, as well as inducing dimensional shrinkage ofthe material.
Anisotropy in shrinkage
The reduction in dimensions of the timber, tech-nically known as shrinkage, can be considerablebut, owing to the complex structure of the mater-ial, the degree of shrinkage is different on thethree principal axes; in other words, timber isanisotropic in its water relationships. The vari-ation in degree of shrinkage that occurs betweendifferent timbers, and, more important, the vari-ation among the three different axes within anyone timber is illustrated in Table 45.1. It shouldbe noted that the values quoted in the table repre-sent shrinkage on drying from the green state (i.e.�27 per cent) to 12 per cent moisture content, alevel which is of considerable practical signific-ance. At 12 per cent moisture content, timber isin equilibrium with an atmosphere having a rela-tive humidity of 60 per cent and a temperature of20°C; these conditions would be found in build-ings having regular, but intermittent heating.
463
From Table 45.1 it will be observed thatshrinkage ranges from 0.1 per cent to 10 per cent,i.e. a 100-fold range. Longitudinal shrinkage, itwill be noted, is always an order of magnitudeless than transverse, while in the transverse planeradial shrinkage is usually some 60–70 per centof the corresponding tangential figure.
The anisotropy between longitudinal and trans-verse shrinkage amounting to approximately 40:1is due in part to the vertical arrangement of cellsin timber and in part to the particular orientationof the microfibrils in the middle layer of the sec-ondary cell wall (S2). Thus, since the microfibrilsof the S2 layer of the cell wall are inclined at anangle of about 15° to the vertical, the removal ofwater from the matrix and the consequent move-ment closer together of the microfibrils will resultin a horizontal component of the movement con-siderably greater than the corresponding longitu-dinal component (see Table 45.1).
Various theories have been developed over theyears to account for shrinkage in terms ofmicrofibrillar angle. The early theories werebased on models which generally consider the cellwall to consist of an amorphous hygroscopicmatrix in which are embedded parallel crystallinemicrofibrils which restrain swelling or shrinkingof the matrix. One of the first models consideredpart of the wall as a flat sheet consisting only ofan S2 layer in which microfibrillar angle had aconstant value (Barber and Meylan, 1964). Thismodel treated the cells as square in cross-section
and there was no tendency for the cells to twist asthey began to swell. An improved model (Barber,1968) was to treat the cells as circular in cross-section and embraced a thin constraining sheathoutside the main cylinder which acted to reducetransverse swelling; experimental confirmation ofthis model was carried out by Meylan (1968).Later models have treated the cell wall as twolayers of equal thickness having microfibrillarangles of equal and opposite sense, and thesetwo-ply models have been developed extensivelyover the years to take into account the layeredstructure of the cell wall, differences in structurebetween radial and tangential walls, and varia-tions in wall thickness. The principal researcherusing this later type of model was Cave whosemodels are based on an array of parallel cellulosemicrofibrils embedded in a hemicellulose matrix,with different arrays for each wall layer; thesearrays of basic wall elements were bondedtogether by lignin microlayers. Earlier versions ofthe model included consideration of the variationin the stiffness of the matrix with changing mois-ture content (1972, 1975). The model was latermodified to take account of the amount of highenergy water absorbed rather than the totalamount of water (1978a,b). Comparison withpreviously obtained experimental data was excel-lent at low moisture contents, but poorer at mois-ture contents between 15 and 25 per cent. Allthese theories are extensively presented and dis-cussed by Skaar (1988).
Deformation in timber
464
TABLE 45.1 Shrinkage (%) on drying from green to 12 per cent moisture content
Transverse
Botanical name Commercial name Tangential Radial Longitudinal
Chlorophora excelsa Iroko 2.0 1.5 �0.1Tectona grandis Teak 2.5 1.5 �0.1Pinus strobus Yellow pine 3.5 1.5 �0.1Picea abies Whitewood 4.0 2.0 �0.1Pinus sylvestris Redwood 4.5 3.0 �0.1Tsuga heterophylla Western hemlock 5.0 3.0 �0.1Quercus robur European oak 7.5 4.0 �0.1Fagus sylvatica European beech 9.5 4.5 �0.1
The influence of microfibrillar angle on degreeof longitudinal and transverse shrinkagedescribed for normal wood is supported by evid-ence derived from experimental work oncompression wood, one of the forms of reactionwood described in Section 44.4. Compressionwood is characterised by possessing a middlelayer to the cell wall, the microfibrillar angle ofwhich can be as high as 45° though 20–30° ismore usual. The longitudinal shrinkage is muchhigher and the transverse shrinkage correspond-ingly lower than in normal wood, and it has beendemonstrated that the values for compressionwood can be accommodated on the shrinkage–microfibrillar angle curve for normal wood.
Differences in the degree of transverse shrink-age between tangential and radial planes (Table45.1), are usually explained in terms of, first, therestricting effect of the rays on the radial plane;second, the increased thickness of the middlelamella on the tangential plane compared withthe radial; third, the difference in degree of lignifi-cation between the radial and tangential cellwalls; fourth, the small difference in microfibrillarangle between the two walls; and fifth, the alter-nation of earlywood and latewood in the radialplane, which, because of the greater shrinkage oflatewood, induces the weaker earlywood toshrink more tangentially than it would if isolated.Considerable controversy reigns as to whether allfive factors are actually involved and their relativesignificance. Comprehensive reviews of the evid-ence supporting these five possible explanationsof differential shrinkage in the radial and tangen-tial planes is to be found in Boyd (1974) andSkaar (1988).
Recently Kifetew (1997) has demonstrated thatthe gross transverse shrinkage anisotropy of Scotspine timber, with a value approximately equal totwo, can be explained primarily in terms of theearlywood–latewood interaction theory by usinga set of mathematical equations proposed by him;the gross radial and transverse shrinkage valueswere determined from the isolated early and late-wood shrinkage values taken from the literature.
Volumetric shrinkage, sv, is slightly less thanthe sums of the three directional components.
Practical significance.
In order to avoid shrinkage of timber after fabri-cation, it is essential that it is dried down to amoisture content which is in equilibrium with therelative humidity of the atmosphere in which thearticle is to be located. A certain latitude can betolerated in the case of timber frames and rooftrusses, but in the production of furniture,window frames, flooring and sports goods it isessential that the timber is seasoned to theexpected equilibrium conditions, namely 12 percent for regular intermittent heating and 10 percent in buildings with central heating, otherwiseshrinkage in service will occur with loosening ofjoints, crazing of paint films, and buckling anddelamination of laminates. An indication of themoisture content of timber used in different envi-ronments is presented in Figure 45.1.
45.2.2 Movement
So far only those dimensional changes associatedwith the initial reduction in moisture contenthave been considered. However, dimensionalchanges, albeit smaller in extent, can also occurin seasoned or dried wood due to changes in therelative humidity of the atmosphere. Suchchanges certainly occur on a seasonal basis andfrequently also on a daily basis. Since thesechanges in humidity are usually fairly small,inducing only slight changes in the moisturecontent of the timber, and since a considerabledelay occurs in the diffusion of water vapour intoor out of the centre of a piece of timber, itfollows that these dimensional changes in sea-soned timber are small, considerably smaller thanthose for shrinkage.
To quantify such movements for differenttimbers, dimensional changes are recorded overan arbitrary range of relative humidities. In theUK, the standard procedure is to condition thetimber in a chamber at 90 per cent relativehumidity and 25°C, then to measure its dimen-sions and to transfer it to a chamber at 60 per cent relative humidity, allowing it to come to equilibrium before remeasuring it; the
Dimensional change due to moisture
465
corresponding average change in moisturecontent is from 21 per cent to 12 per cent. Move-ment values in the tangential and radial planesfor those timbers listed in Table 45.1 are pre-sented in Table 45.2. The timbers are recorded inthe same order, thus illustrating that although abroad relationship holds between values ofshrinkage and movement, individual timbers canbehave differently over the reduced range ofmoisture contents associated with movement.Since movement in the longitudinal plane is sovery small, it is generally ignored. Anisotropywithin the transverse plane can be accounted forby the same set of variables that influence shrink-age.
Where timber is subjected to wide fluctuationsin relative humidity, care must be exercised toselect a species that has low movement values.
Moisture in timber has a very pronouncedeffect not only on its strength (Figure 44.22), butalso on its stiffness, toughness and fracture mor-phology; stiffness is discussed later in thischapter, while the other two parameters are dis-cussed in Chapter 46.
Aware of the technological significance of theinstability of wood under changing moisturecontent, many attempts have been made over theyears to find a solution to the problem; these aresummarised in Section 48.3.1 on dimensionalstabilisation.
45.3 Thermal movementTimber, like other materials, undergoes dimen-sional changes commensurate with increasingtemperature. This is attributed to the increasingdistances between the molecules as they increasethe magnitude of their oscillations with increasingtemperature. Such movement is usually quantifiedfor practical purposes as the coefficient of linearthermal expansion and values for certain timbersare listed in Table 45.3. Although differencesoccur between species these appear to be smallerthan those occurring for shrinkage and move-ment. The coefficient for transverse expansion isan order of magnitude greater than that in thelongitudinal direction. This degree of anisotropy(10:1) can be related to the ratio of length to
Deformation in timber
466
FIGURE 45.1 Equilibrium moisture content of timber in various environments. The figures for differentspecies vary, and the chart shows only average values (© Building Research Establishment).
breadth dimensions of the crystalline regionswithin the cell wall. Transverse thermal expan-sion appears to be correlated with specificgravity, but somewhat surprisingly this relation-ship is not sustained in the case of longitudinalthermal expansion where the values for differenttimbers are roughly constant (Weatherwax andStamm, 1946).
The expansion of timber with increasing tem-perature appears to be linear over a wide temper-ature range: the slight differences in expansionwhich occur between the radial and tangentialplanes are usually ignored and the coefficients areaveraged to give a transverse value as recorded in
Table 45.3. For comparative purposes, the coeffi-cients of linear thermal expansion for glass- andcarbon-reinforced plastic, two metals and twoplastics are also listed. Even the transverse expan-sion of timber is considerably less than that forthe plastics.
The dimensional changes of timber caused bydifferences in temperature are small when com-pared to changes in dimensions resulting from theuptake or loss of moisture. Thus for timber witha moisture content greater than about 3 per cent,the shrinkage due to moisture loss on heating willbe greater than the thermal expansion, with theeffect that the net dimensional change on heating
Thermal movement
467
TABLE 45.2 Movement (%) on transferring timber from 90 per cent relative humidity to 60 per cent at 25°C
Transverse
Botanical name Commercial name Tangential Radial
Chlorophora excelsa Iroko 1.0 0.5Tectona grandis Teak 1.2 0.7Pinus strobus Yellow pine 1.8 0.9Picea abies Whitewood 1.5 0.7Pinus sylvestris Redwood 2.2 1.0Tsuga heterophylla Western hemlock 1.9 0.9Quercus robur European oak 2.5 1.5Fagus sylvatica European beech 3.2 1.7
TABLE 45.3 Coefficient of linear thermal expansion of various woods and other materials per degree centi-grade
Timber Coefficient of thermal expansion�10�6
Longitudinal Transverse
Picea abies Whitewood 5.41 34.1Pinus strobus Yellow pine 4.00 72.7Quercus robur European oak 4.92 54.4GRP, 60/40, unidirectional 10.0 10.0CFRP, 60/40, unidirectional 10.0 �1.00
Mild steel 12.6Duralumin (aluminium alloy) 22.5Nylon 66 125.0Polypropylene 110.0
will be negative. For most practical purposesthermal expansion or contraction can be safelyignored over the range of temperatures in whichtimber is generally employed.
45.4 Deformation under loadThis section is concerned with the type and mag-nitude of the deformation that results from theapplication of external load. As in the case ofboth concrete and polymers, the load–deformation relationship in timber is exceedinglycomplex, resulting from the fact that:
1. timber does not behave in a truly elasticmode, rather its behaviour is time-dependent;and,
2. the magnitude of the strain is influenced by awide range of factors; some of these are pro-perty dependent, such as density of thetimber, angle of the grain relative to directionof load application, and angle of the microfib-
rils within the cell wall; others are environ-mentally dependent, such as temperature andrelative humidity.
Under service conditions timber often has towithstand an imposed load for many years,perhaps even centuries; this is particularly rele-vant in construction applications. When loaded,timber will deform and a generalised interpreta-tion of the variation of deformation with timetogether with the various components of thisdeformation is illustrated in Figure 45.2. On theapplication of a load at time zero an instanta-neous (and reversible) deformation occurs whichrepresents true elastic behaviour. On maintainingthe load to time t1 the deformation increases,though the rate of increase is continually decreas-ing; this increase in deformation with time istermed creep. On removal of the load at time tl
an instantaneous reduction in deformation occurswhich is approximately equal in magnitude to theinitial elastic deformation. With time, the remain-
Deformation in timber
468
FIGURE 45.2 The various elastic and plastic components of the deformation of timber under constant load(© Building Research Establishment).
ing deformation will decrease at an ever-decreasing rate until at time t2 no further reduc-tion occurs. The creep that has occurred duringloading can be conveniently subdivided into areversible component, which disappears withtime and which can be regarded as delayed elasticbehaviour, and an irreversible component whichresults from plastic or viscous flow. Therefore,timber on loading possesses three forms of defor-mation behaviour – elastic, delayed elastic andviscous. Like so many other materials, timber canbe treated neither as a truly elastic materialwhere, by Hooke’s law, stress (see Section 45.4.1)is proportional to strain but independent of therate of strain, nor as a truly viscous liquid where,according to Newton’s law, stress is proportionalto rate of strain, but independent of strain itself(see Section 45.4.2). Where combinations ofbehaviour are encountered the material is said tobe viscoelastic and timber, like many high poly-mers, is a viscoelastic material.
Having defined timber as such, the reader willno doubt be surprised to find that half of thischapter is devoted to the elastic behaviour oftimber. It has already been discussed how part ofthe deformation can be described as elastic andthe section below will indicate how, at low levelsof stressing and short periods of time, there isconsiderable justification for treating the materialas such. Perhaps the greatest incentive for thisviewpoint is the fact that classical elasticitytheory is well established and, when applied totimber, has been shown to work very well. Thequestion of time in any stress analysis can beaccommodated by the use of safety factors indesign calculations.
Consequently, this section will deal first withelastic deformation as representing a very goodapproximation of what happens in practice, whilethe following section will deal with viscoelasticdeformation, which embraces both delayed elasticand irreversible deformation. Although techni-cally more applicable to timber, viscoelasticity iscertainly less well understood and developed inits application than is the case with elasticitytheory.
Deformation under load
469
45.4.1 Elastic deformation
When a sample of timber is loaded in tension,compression or bending, the instantaneous defor-mations obtained with increasing load areapproximately proportional to the values of theapplied load. Figure 45.3 illustrates that thisapproximation is certainly truer of the experi-mental evidence in longitudinal tensile loadingthan in the case of longitudinal compression. Inboth modes of loading, the approximationappears to become a reality at the lower levels ofloading. Thus, it has become convenient to recog-nise a point of inflection on the load-deflection
FIGURE 45.3 Load-deflection graphs for timberstressed in tension and compression parallel to thegrain. The assumed limit of proportionality for eachgraph is indicated (© Building Research Establish-ment).
curve known as the limit of proportionality,below which the relationship between load anddeformation is linear, and above which non-linearity occurs. Generally, the limit of propor-tionality in longitudinal tension is found to occurat about 60 per cent of the ultimate load tofailure while in longitudinal compression the limitis considerably lower, varying from 30 per cent to50 per cent of the failure value.
At the lower levels of loading, therefore, wherethe straight-line relationship appears to be valid,the material is said to be linearly elastic. Hence:
deformation 'applied load
i.e. �a constant
The applied load must be quantified in terms ofthe cross-sectional area carrying that load, whilethe deflection or extension must be related to theoriginal dimension of the test piece prior to loadapplication. Hence:
� stress (N/mm2), denoted by �
and � strain (unitless),denoted by �
Therefore, �a constant�E (N/mm2) (45.1)
where � is the strain (change in dimension/original dimension), � is the stress (load/cross-sectional area), and E is a constant, known as themodulus of elasticity. The modulus of elasticity(MOE), is also referred to in the literature as theelastic modulus, Young’s modulus, or simply andfrequently, though incorrectly, as stiffness.
The apparent linearity at the lower levels ofloading is really an artefact introduced by the rateof testing. At fast rates of loading, a very goodapproximation to a straight line occurs, but asthe rate of loading decreases, the load-deflectionline assumes a curvilinear shape (Figure 45.4).Such curves can be treated as linear by introduc-ing a straight line approximation, which can take
stress (�)��strain (�)
deformation (mm)���original length (mm)
load (N)��csa (mm2)
applied load��deformation
Deformation in timber
470
FIGURE 45.4 The approximation of a curvilinearload-deflection curve for timber stressed at low loadingrates, by linear tangents or secants (© BuildingResearch Establishment).
the form of either a tangent or secant. Tradition-ally for timber and wood fibre composites,tangent lines have been used as linear approxima-tions of load-deflection curves.
Thus, while in theory it should be possible toobtain a true elastic response, in practice this israrely the case, though the degree of divergence isfrequently very low. It should be appreciated inpassing that a curvilinear load-deflection curvemust not be interpreted as an absence of trueelastic behaviour. The material may still behaveelastically, though not linearly elastically: theprime criterion for elastic behaviour is that theload-deflection curve is truly reversible, i.e. nopermanent deformation has occurred on releaseof the load.
The modulus of elasticity or elastic modulus inthe longitudinal direction is one of the principalelastic constants of our material. Readersdesirous of information on the different methodsused to determine the value of the modulus underboth static and dynamic loading are referred toDinwoodie (2000).
The elastic behaviour of a material can be char-acterised by three elastic constants, the first ofwhich, the modulus of elasticity, E, has beendescribed above. The second constant is themodulus of rigidity, G. Within the elastic range ofthe material, shearing stress is proportional toshearing strain. Thus
G� (45.2)
where � is the shearing stress and is the shearingstrain.
The third constant is Poisson’s ratio, �. Gener-ally, when a body is subjected to a stress in onedirection, the body will undergo a change indimensions at right angles to the direction ofstressing. The ratio of the contraction or exten-sion to the applied strain is known as Poisson’sratio and for isotropic bodies is given as:
� � � (45.3)
where �x, �y are strains in the x, y directionsresulting from an applied stress in the x direction.(The minus sign indicates that, when �x is atensile positive strain, �y is a compressive negativestrain.) In timber, because of its anisotropicbehaviour and its treatment as a rhombic system,six Poisson’s ratios occur.
Orthotropic elasticity and timber
In applying the elements of orthotropic elasticityto timber, the assumption is made that the threeprincipal elasticity directions coincide with thelongitudinal, radial and tangential directions inthe tree. The assumption implies that the tangen-tial faces are straight and not curved, and that theradial faces are parallel and not diverging.However, by dealing with small pieces of timber
removed at some distance from the centre of thetree, the approximation of rhombic symmetry fora system possessing circular symmetry becomesmore and more acceptable.
The nine independent constants required tospecify the elastic behaviour of timber are thethree moduli of elasticity, one in each of the lon-gitudinal (L), radial (R) and tangential (T) direc-tions; the three moduli of rigidity, one in each ofthe principal planes longitudinal–tangential (LT),longitudinal–radial (LR) and tangential–radial(TR); and three Poisson’s ratios, namely �RT, �LR,�TL. These constants, together with the threedependent Poisson’s ratios �RL, �TR, �LT are pre-sented in Table 45.4 for a selection of hardwoodsand softwoods. The table illustrates the highdegree of anisotropy present in timber. Compari-son of EL with either ER or ET, and GTR with GLT
or GRL will indicate a degree of anisotropy whichcan be as high as 60:1; usually the ratio of EL toEHORIZ is of the order of 40:1. Note should betaken that the values of �TR are frequently greaterthan 0.5.
Values of EL for some additional timbers are tobe found in Table 46.1.
Factors influencing the elastic modulus
The stiffness of timber is influenced by manyfactors, some of which are properties of thematerial while others are components of theenvironment.
Grain angleFigure 45.5, in addition to illustrating the markedinfluence of grain angle on stiffness, shows thedegree of fit between experimentally derivedvalues and the line obtained using transformationequations to calculate theoretical values (seeequation (46.7) in Section 46.6).
DensityStiffness is related to density of the timber, a rela-tionship which was apparent in Table 45.4 andwhich is confirmed by the plot of over 200species of timber contained in Bulletin 50 of theformer Forest Products Research Laboratory
�y��x
��
Deformation under load
471
TA
BL
E 4
5.4
Val
ues
of t
he e
last
ic c
onst
ants
for
five
har
dwoo
ds a
nd f
our
soft
woo
ds d
eter
min
ed o
n sm
all c
lear
spe
cim
ens
Spec
ies
Den
sity
Moi
stur
eE
LE
RE
Tv T
Rv L
Rv R
Tv L
Tv R
Lv T
LG
LT
GL
RG
TR
(kg/
m3 )
cont
ent
(%)
Har
dwoo
dsB
alsa
200
963
0030
010
60.
660.
018
0.24
0.00
90.
230.
4920
331
233
Kha
ya44
011
1020
011
3051
00.
600.
033
0.26
0.03
20.
300.
6460
090
021
0W
alnu
t59
011
1120
011
9063
00.
720.
052
0.37
0.03
60.
490.
6370
096
023
0B
irch
620
916
300
1110
620
0.78
0.03
40.
380.
018
0.49
0.43
910
1180
190
Ash
670
915
800
1510
800
0.71
0.05
10.
360.
030
0.46
0.51
890
1340
270
Bee
ch75
011
1370
022
4011
400.
750.
073
0.36
0.04
40.
450.
5110
6016
1046
0
Soft
woo
dsN
orw
ay s
pruc
e39
012
1070
071
043
00.
510.
030
0.31
0.02
50.
380.
5162
050
023
Sitk
a sp
ruce
390
1211
600
900
500
0.43
0.02
90.
250.
020
0.37
0.47
720
750
39Sc
ots
pine
550
1016
300
1100
570
0.68
0.03
80.
310.
015
0.42
0.51
680
1160
66D
ougl
as fi
r*59
09
1640
013
0090
00.
630.
028
0.40
0.02
40.
430.
3791
011
8079
*L
iste
d in
ori
gina
l as
Ore
gon
pine
.
Eis
the
mod
ulus
of
elas
tici
ty in
a d
irec
tion
indi
cate
d by
the
sub
scri
pt (
N/m
m2 )
.G
is t
he m
odul
us o
f ri
gidi
ty in
a p
lane
indi
cate
d by
the
sub
scri
pt (
N/m
m2 )
.v i
jis
the
Poi
sson
’s r
atio
s fo
r an
ext
ensi
onal
str
ess
in j
dire
ctio
n,
�
From
Hea
rmon
(19
48),
but
wit
h di
ffer
ent
nota
tion
for
the
Poi
sson
’s r
atio
s.
com
pres
sive
str
ain
in i
dire
ctio
n�
��
�ex
tens
iona
l str
ain
in j
dire
ctio
n
(Figure 45.6); the correlation coefficient was 0.88for timber at 12 per cent moisture content and0.81 for green timber and the relation is curvi-linear. A high correlation is to be expected, sincedensity is a function of the ratio of cell wall thick-ness to cell diameter; consequently, increasingdensity will result in increasing stiffness of thecell.
Owing to the variability in structure that existsbetween different timbers, the relation betweendensity and stiffness will be higher where only asingle species is under investigation. Because ofthe reduced range in density, a linear regression isusually fitted.
Similar relations with density have beenrecorded for the modulus of rigidity in certainspecies: in others, however, for example spruce,both the longitudinal-tangential and longitudi-
Deformation under load
473
FIGURE 45.5 Effect of grain angle on the modulusof elasticity (© Building Research Establishment).
nal–radial shear moduli have been found to beindependent of density. Most investigators agree,however, that the Poisson’s ratios are independ-ent of density.
KnotsSince timber is anisotropic in behaviour, andsince knots are characterised by the occurrence ofdistorted grain, it is not surprising to find that thepresence of knots in timber results in a reductionin the stiffness. The relationship is difficult toquantify since the effect of the knots will dependnot only on their number and size, but also ontheir distribution both along the length of thesample and across the faces. Dead knots, espe-cially where the knot has fallen out, will result inlarger reductions in stiffness than will green knots(see Section 44.1).
UltrastructureTwo components of the fine or chemical structurehave a profound influence on both the elastic andrigidity moduli. The first relates to the existenceof a matrix material with particular emphasis onthe presence of lignin. In those plants devoid oflignin, e.g. the grasses, or in wood fibres whichhave been delignified, the stiffness of the cells islow and it would appear that lignin, apart fromits hydrophilic protective role for the cellulosiccrystallites, is responsible to a considerable extentfor the high stiffness found in timber.
The significance of lignin in determining stiff-ness is not to imply that the cellulose fractionplays no part: on the contrary, it has been shownthat the angle at which the microfibrils are lyingin the middle layer of the secondary cell wall, S2,also plays a significant role in controlling stiffness(Figure 45.7).
A considerable number of mathematical modelshave been devised over the years to relate stiffnessto microfibrillar angle. The early models were two-dimensional in approach, treating the cell wall as aplanar slab of material, but later the modelsbecame much more sophisticated, taking intoaccount the existence of cell wall layers other thanthe S2, the variation in microfibrillar angle betweenthe radial and tangential walls and consequently
the probability that they undergo different strains,and, lastly, the possibility of complete shearrestraint within the cell wall. These three-dimensional models are frequently analysed usingfinite element techniques. These early models werereviewed by Dinwoodie (1975).
Later modelling of timber behaviour in termsof its structure has been reviewed by Astley et al.(1998); it makes reference to the work of Cave(1975) who used the concept of an elastic fibrecomposite consisting of an inert fibre phaseembedded in a water-reactive matrix. The consti-tutive equation is related to the overall stiffness ofthe composite, the volume fraction, and the stiff-ness and sorption characteristics of the matrix.Unlike previous models, the equation can beapplied not only to elasticity, but also to shrink-age and even moisture-induced creep (Cave,1975).
Recently, Harrington et al. (1998) havedeveloped a model of the wood cell wall based on
the homogenisation first, of an amorphous lignin-polyose matrix, and then of a representativevolume element comprising a cellulose microfib-ril, its polyose-cellulose sheath and the surround-ing matrix. The model predicts orthotropic elasticconstants which are in good agreement withrecorded values (Astley et al., 1998). The pre-dicted variation of axial stiffness with the S2
microfibrillar angle is consistent with observedbehaviour and aligned with results from other cellwall models.
Stiffness of a material is very dependent on thetype and degree of chemical bonding within itsstructure; the abundance of covalent bonding inthe longitudinal plane and hydrogen bonding inone of the transverse planes contributes consider-ably to the moderately high levels of stiffnesscharacteristic of timber.
Moisture contentThe influence of moisture content on stiffness is
Deformation in timber
474
FIGURE 45.6 Effect of specific gravity on the longitudinal modulus of elasticity for over 200 species of timbertested in the green and dry states (© Building Research Establishment).
similar to, though not quite so sensitive as, thatfor strength which was illustrated in Figure44.22. Early experiments in which stiffness wasmeasured on a specimen of Sitka spruce, as ittook up moisture from the dry state, clearly indi-cated a linear loss in stiffness as the moisturecontent increased to about 27 per cent, corre-sponding to the fibre saturation point as dis-cussed in Chapter 44; further increase in moisturecontent has no influence on stiffness. Theseresults for the variation in longitudinal moduliwith moisture content under static loading havebeen confirmed using simple dynamic methods.Measurement of the frequency of vibration wascarried out at regular intervals as samples of Sitkaspruce were dried from 70 per cent to zero mois-ture content (Figure 45.8).
Deformation under load
475
FIGURE 45.7 Effect of the mean microfibrillarangle of the cell wall on the longitudinal modulus ofelasticity of the wall material in Pinus radiata. Calcu-lated values from a mathematical model are alsoincluded (from Cave (1968) by permission of Springer-Verlag).
FIGURE 45.8 Effect of moisture content on the lon-gitudinal modulus of elasticity and the modulus ofrigidity in the L–R plane in Sitka spruce. Both moduliwere determined dynamically (© Building ResearchEstablishment).
Confirmation of the reduction in modulus ofelasticity with increasing moisture content isforthcoming first, from Figure 45.6, in which theregression lines of elasticity against density forover 200 species of timber at 12 per cent mois-ture content and in the green state are presented,and second, from the review of more recentresults by Gerhards (1982). The effect of mois-ture increase has a far greater effect on the
modulus perpendicular to the grain than on themodulus along the grain (Gerhards, 1982).
TemperatureIn timber, like most other materials, increasingtemperature results in greater oscillatory move-ment of the molecules and an enlargement of thecrystal lattice. These in turn affect the mechanicalproperties and the stiffness and strength of thematerial decreases.
Although the relationship between stiffness andtemperature has been shown experimentally to becurvilinear, the degree of curvature is usuallyslight at low moisture contents and the relation isfrequently treated as linear, thus:
ET �Et[1�a(T� t)] (45.4)
where E is the elastic modulus, T is a higher tem-perature, t is a lower temperature, and a is thetemperature coefficient. The value a for longitudi-nal modulus has been shown to lie between 0.001and 0.007 for low moisture contents.
Deformation in timber
476
115
110
105
100
95
90
85
80
75
70�30 �20 �10 0 10 20 30 40 50 60
Temperature–°C
Mod
ulus
of e
last
icity
–% Moisturecontent
0%
8%
12%
20%
FIGURE 45.9 The interaction of temperature andmoisture content on the modulus of elasticity. Resultsare averaged for six species of timber and the modulusat 20°C and 0 per cent moisture content is taken asunity (© Building Research Establishment).
The effect of a temperature increase is greateron the perpendicular modulus than on the longi-tudinal modulus (Gerhards, 1982).
At higher moisture contents the relationship ofstiffness and temperature is markedly curvilinearand the interaction of moisture content and tem-perature in influencing stiffness is clearly shownin Figure 45.9. At zero moisture content thereduction in stiffness between �20°C and �60°Cis only 6 per cent: at 20 per cent moisture contentthe corresponding reduction is 40 per cent. Thisincrease in the significance of temperature withincreasing moisture content has been confirmedby Gerhards (1982).
Long-term exposure to elevated temperatureresults in a marked reduction in stiffness as wellas strength and toughness, the effect usually beinggreater in hardwoods than in softwoods. Evenexposure to cyclic changes in temperature overlong periods will result in a loss of stiffness(Moore, 1984).
45.4.2 Viscoelastic deformation
In the introduction to Section 45.4, timber wasdescribed as being neither truly elastic in itsbehaviour nor truly viscous, but rather a combi-nation of both states; such behaviour is usuallydescribed as viscoelastic and, in addition totimber, materials such as concrete, bitumen andthe thermoplastics are also viscoelastic in theirresponse to stress.
Viscoelasticity infers that the behaviour of thematerial is time dependent; at any instant in timeunder load its performance wilI be a function of itspast history. Now if the time factor under load isreduced to zero, a state which we can picture inconcept, but never attain in practice, the materialwill behave truly elastically, and we have seen inSection 45.4.1 how timber can be treated as anelastic material and how the principles oforthotropic elasticity can be applied. However,where stresses are applied for a period of time, vis-coelastic behaviour will be experienced and, whileit is possible to apply elasticity theory with a factorcovering the increase in deformation with time, thisprocedure is at best only a first approximation.
In a material such as timber, time-dependentbehaviour manifests itself in a number of ways, ofwhich the more common are creep, relaxation,damping capacity and the dependence of strengthon duration of load. When the load on a sampleof timber is held constant for a period of time,the increase in deformation over the initial instan-taneous elastic deformation is called creep andFigure 45.2 illustrates not only the increase increep with time, but also the subdivision of creepinto a reversible and an irreversible component ofwhich more will be said in a later section.
Most timber structures carry a considerabledead load and the component members of thesewill undergo creep; the dip towards the centre ofthe ridge of the roof of very old buildings bearstestament to the fact that timber does creep.However, compared with thermoplastics andbitumen, the amount of creep in timber is appre-ciably lower.
Viscoelastic behaviour is also apparent in theform of relaxation where the load necessary tomaintain a constant deformation decreases withtime; in timber utilisation this has limited prac-tical significance and the area has attracted verylittle research. Damping capacity is a measure ofthe fractional energy converted to heat comparedwith that stored per cycle under the influence ofmechanical vibrations; this ratio is time depend-ent. A further manifestation of viscoelasticbehaviour is the apparent loss in strength oftimber with increasing duration of load; thisfeature is discussed in detail in Section 46.6.12and illustrated in Figure 46.7.
Creep
Creep parametersIt is possible to quantify creep by a number oftime-dependent parameters of which the twomost common are creep compliance (known alsoas specific creep) and relative creep (also knownas the creep coefficient); both parameters are afunction of temperature.
Creep compliance (cc) is the ratio of increasingstrain with time to the applied constant stress,i.e.:
Deformation under load
477
FIGURE 45.10 The increase in deformation withtime of urea-formaldehyde bonded chipboard (particle-board): the regression line has been fitted to the experi-mental values using equation (45.10) (© BuildingResearch Establishment).
cc(t, T) � (45.5)
while relative creep (cr) is defined as either thedeflection, or more usually, the increase in deflec-tion, expressed in terms of the initial elasticdeflection, i.e.:
cr(t, T) � or (45.6)
where �t is the deflection at time t, and �0 is theinitial deflection.
Relative creep has also been defined as thechange in compliance during the test expressed interms of the original compliance.
Creep relationshipsIn both timber and timber products such asplywood or chipboard (particleboard), the rate ofdeflection or creep slows down progressively withtime (Figure 45.10); the creep is frequentlyplotted against log time and the graph assumes anexponential shape. Results of creep tests can alsobe plotted as relative creep against log time or ascreep compliance against stress as a percentage ofthe ultimate short-time stress.
�t � �0�
�0
�t��0
strain varying���applied constant stress
In Section 45.4.1 it was shown that the degreeof elasticity varied considerably between thehorizontal and longitudinal planes. Creep, as oneparticular manifestation of viscoelastic behaviour,is also directionally dependent. In tensile stressingof longitudinal sections produced with the grainrunning at different angles, it was found that rela-tive creep was greater in the direction perpendicu-lar to the grain than it was parallel to the grain.
Timber- and wood-based panels, therefore, areviscoelastic materials, the time-dependent proper-ties of which are directionally dependent. Thenext important criterion is whether they arelinear viscoelastic in behaviour. For viscoelasticbehaviour to be defined as linear, the instanta-neous, recoverable and non-recoverable com-ponents of the deformation must vary directlywith the applied stress. An alternative definitionis that the creep compliance or relative creepmust be independent of stress and not a functionof it.
Timber- and wood-based panels exhibit linearviscoelastic behaviour at the lower levels ofstressing, but at the higher stress levels thisbehaviour reverts to being non-linear. Examplesof this transition in behaviour are illustrated inFigures 45.11(a) and 45.11(b), where for bothredwood timber and UF bonded particleboardrespectively, the change from linear to non-linearbehaviour occurs between the 45 and 60 per centstress level.
The linear limit for the relation between creepand applied stress varies with mode of testing,with species of timber or type of panel, and withboth temperature and moisture content. Intension parallel to the grain at constant tempera-ture and moisture content, timber has been foundto behave as a linear viscoelastic material up toabout 75 per cent of the ultimate tensile strength,though some workers have found considerablevariability and have indicated a range from 36 to84 per cent. In compression parallel to the grain,the onset of non-linearity appears to occur atabout 70 per cent, though the level of actualstress will be much lower than in the case oftensile strength since the ultimate compressionstrength is only one-third that of the tensile
Deformation in timber
478
0.6
0.5
0.4
0.3
0.2
0.1
00 30 40 60 67.5 75
Stress level (%)
Rel
ativ
e cr
eep
Time � 1000 minConditions � 20˚C/65% rh
FIGURE 45.11 (a) The relation of relative creep tostress level at fixed time periods illustrating the transi-tion from linear to non-linear viscoelastic behaviour inredwood timber (Pinus sylvestris) (© Building ResearchEstablishment).
strength. In bending, non-linearity seems todevelop very much earlier at about 50–60 percent (Figure 45.11(a) and (b)); the actual stresslevels will be very similar to those for compres-sion.
In both compression and bending, the diver-gence from linearity is usually greater than in the
case of tensile stressing; much of the increaseddeformation occurs in the non-recoverablecomponent of creep and is associated with pro-gressive structural changes including the develop-ment of incipient failure.
Increase not only in stress level, but also intemperature to a limited extent, and moisturecontent to a considerable degree, result in anearlier onset of non-linearity and a more markeddeparture from linearity. For most practical situ-ations, however, working stresses are only a smallpercentage of the ultimate, rarely approachingeven 50 per cent, and it can be safely assumedthat timber, like concrete, will behave as a linearviscoelastic material under normal service con-ditions.
Deformation under load
479
0 30 40 60 67.5 75Stress level
(as percentage of ultimate strength)
Rel
ativ
e cr
eep
50 mm UF particle board0.18
0.16
0.14
0.12
0.10
0.08
0.06
0.04
0.01
0
90 min.
10 min.
FIGURE 45.11 (b) The relation of relative creep tostress level at fixed time periods illustrating the transi-tion from linear to non-linear viscoelastic behaviour in50mm UF particleboard (© Building Research Estab-lishment).
Principle of superpositionSince timber behaves as a linear viscoelasticmaterial under conditions of normal temperatureand humidity and at low to moderate levels ofstressing, it is possible to apply the Boltzmann’sprinciple of superposition to predict the responseof timber to complex or extended loadingsequences. This principle states that the creepoccurring under a sequence of stress increments istaken as the superposed sum of the responses tothe individual increments. This can be expressedmathematically in a number of forms, one ofwhich for linear materials is:
�c(t)��n
1
��icci (45.7)
where n is the number of load increments, ��i isthe stress increment, cci is the creep compliancefor the individual stress increments applied fordiffering times, t� �1, t� �2, . . . , t� �n and �c(t) isthe total creep at time t; or in integrated form
�c(t)��t
�1
cc(t��) (�)d� (45.8)
In experiments on timber, it has been foundthat in the comparison of deflections in beamsloaded either continuously or repeatedly for twoor seven days in every fourteen, for six months atfour levels of stress, the applicability of the Boltz-mann’s principle of superposition was confirmedfor stress levels up to 50 per cent (Nakai andGrossman, 1983).
The superposition principle has been found tobe applicable even at high stresses in both shearand tension of dry samples. However, at highmoisture contents, the limits of linear behaviourin shear and tension appear to be considerablylower, thereby confirming views expressed earlieron the non-linear behaviour of timber subjectedto high levels of stressing and/or high moisturecontent.
Mathematical modelling of steady-statecreepThe relationship between creep and time has beenexpressed mathematically using a wide range ofequations. It should be appreciated that such
d��d�
expressions are purely empirical, none of thempossessing any sound theoretical basis. Their rela-tive merits depend on how easily their constantscan be determined and how well they fit theexperimental results.
One of the most successful mathematicaldescriptions for creep in timber under constantrelative humidity and temperature appears to bethe power law, of general form
�(t)� �0 �atm (45.9)
where �(t) is the time-dependent strain, �0 is the initial deformation, a and m are material-specific parameters to be determined experimen-tally (m�0.33 for timber), and t is the elapsedtime.
The prime advantages of using a power func-tion to describe creep is its representation as astraight line on a log/log plot, thereby makingonward prediction on a time basis that mucheasier than using other models.
Alternatively, creep behaviour in timber, likethat of many other high polymers, can be inter-preted with the aid of mechanical (rheological)models comprising different combinations ofsprings and dashpots (piston in a cylinder con-taining a viscous fluid). The springs act as amechanical analogue of the elastic component ofdeformation, while the dashpot simulates theviscous or flow component. When more than asingle member of each type is used, these com-ponents can be combined in a wide variety ofways, though only one or two will be able todescribe adequately the creep and relaxationbehaviour of the material.
The simplest linear model which successfullydescribes the time-dependent behaviour of timberunder constant humidity and temperature forshort periods of time is the four-element modelillustrated in Figure 45.12; the central part of themodel will be recognised as a Kelvin element. Tothis unit has been added in series a second springand dashpot. The strain at any time t under aconstant load is given by the equation
Y� � �1�exp ��� (45.10)�t� 3
�tE2�
2
��E2
��E1
Deformation in timber
480
FIGURE 45.12 Mechanical analogue of the com-ponents of creep: the springs simulate elastic deforma-tion and the dashpots viscous flow. The modelcorresponds to equation (45.10) (© Building ResearchEstablishment).
where Y is the strain at time t, E1 is the elasticityof spring 1, E2 is the elasticity of spring 2, � is thestress applied, 2 is the viscosity of dashpot 2, 3
is the viscosity of dashpot 3.The first term on the right-hand side of equa-
tion (45.10) represents the instantaneous defor-mation, while the second term describes thedelayed elasticity and the third term the plasticflow component. Thus, the first term describesthe elastic behaviour while the combination ofthe second and third terms accounts for the vis-coelastic or creep behaviour. The response of thisparticular model will be linear and it will obeythe Boltzmann superposition principle.
The degree of fit between the behaviourdescribed by the model and experimentallyderived values can be exceedingly good: anexample is illustrated in Figure 45.10, where thedegree of correlation between the fitted line andexperimental results for creep in bending of urea-formaldehyde chipboard (particleboard) beamswas as high as 0.941.
A much more demanding test of any model isthe prediction of long-term performance fromshort-term data. For timber and the variousboard materials, it has been found necessary to
make the viscous term non-linear in these modelswhere accurate predictions of creep (!10 percent) are required for long periods of time (�10years) from short-term data (6–9 months) (Din-woodie et al., 1990a). The deformation of thisnon-linear mathematical model is given by theequation
Y��1 ��2[1�exp(��3t)]� �4t�5 (45.11)
where �5 is the viscous modification factor with avalue 0�b�1.
An example of the successful application of thismodel to predict the deflection of a sample ofcement-bonded particleboard after ten years fromthe first nine months of data is given in Din-woodie et al. (1990a).
Various non-linear viscoelastic models havebeen developed and tested over the years rangingfrom the fairly simple early approach by Ylinen(1965) in which a spring and a dashpot in hisrheological model are replaced by non-linear ele-ments, to the much more sophisticated model byTong and Ödeen (1989a) in which the linear vis-coelastic equation is modified by the introductionof a non-linear function either in the form of asimple power function, or by using the sum of anexponential series corresponding to ten Kelvinelements in series with a single spring.
All this modelling is further confirmation of thenon-linear viscoelastic behaviour of timber andwood-based panels when subjected to high levels ofstressing, or to lower levels at high moisture levels.
The development of models for unsteady statemoisture content are described later.
Reversible and irreversible componentsof creep
In timber and many of the high polymers, creepunder load can be subdivided into reversible andirreversible components: passing reference to thiswas made in Section 45.4 and the generalisedrelationship with time was depicted in Figure45.2. The relative proportions of these two com-ponents of total creep appears to be related tostress level and to prevailing conditions of tem-perature and moisture content.
Deformation under load
481
The influence of level of stress is clearly illus-trated in Figure 45.13, where the total complianceat 70 per cent and 80 per cent of the ultimatestress for hoop pine in compression is subdividedinto the separate components: at 70 per cent, theirreversible creep compliance accounts for about45 per cent of the total creep compliance, while at80 per cent of the ultimate, the irreversible creepcompliance has increased appreciably to 70 percent of the total creep compliance at the longerperiods of time, though not at the shorter dura-tions. Increased moisture content and increasedtemperature will also result in an enlargement ofthe irreversible component of total creep.
Reversible creep is frequently referred to in theliterature as delayed elastic or primary creep, andin the early days was ascribed to either polymeric
FIGURE 45.13 The relative proportions of therecoverable and irrecoverable creep compliance insamples of hoop pine (Araucaria cunninghamii)stressed in bending (Kingston and Budgen (1972) bypermission of Springer-Verlag).
uncoiling or the existence of a creeping matrix.Owing to the close longitudinal association of themolecules of the various components in the amor-phous regions, it appears unlikely that uncoilingof the polymers under stress can account formuch of the reversible component of creep.
The second explanation of reversible creeputilises the concept of time-dependent, two-stagemolecular motions of the cellulose, hemicelluloseand the lignin constituents. The pattern of molec-ular motion for each component is dependent onthat of the other constituents and it has beenshown that the difference in directional move-ment of the lignin and non-lignin moleculesresults in considerable molecular interferencesuch that stresses set up in loading can be trans-ferred from one component (a creeping matrix) toanother component (an attached, but non-creeping structure). It is postulated that the ligninnetwork could act as an energy sink, maintainingand controlling the energy set up by stressing(Chow, 1973).
Irreversible creep, also referred to as viscous,plastic or secondary creep has been related to eithertime-dependent changes in the active number ofhydrogen bonds, or to the loosening and sub-sequent remaking of hydrogen bonds as moisturediffuses through timber with the passage of time(Gibson, 1965). Such diffusion can result directlyfrom stressing; thus early work indicated that whentimber was stressed in tension it gained in moisturecontent, and conversely when stressed in compres-sion its moisture content was lowered. It is argued,though certainly not proven, that the movement ofmoisture by diffusion occurs in a series of stepsfrom one adsorption site to the next, necessitatingthe rupture and subsequent reformation of hydro-gen bonds. The process is viewed as resulting inloss of stiffness and/or strength, possibly throughslippage at the molecular level. Recently, however,it has been demonstrated that moisture movement,while affecting creep, can account for only part ofthe total creep obtained, and this explanation ofcreep at the molecular level warrants more investi-gation; certainly not all the observed phenomenasupport the hypothesis that creep is due to thebreaking and remaking of hydrogen bonds under a
Deformation in timber
482
stress bias. At moderate to high levels of stressing,particularly in bending and compression parallel tothe grain, the amount of irreversible creep is closelyassociated with the development of kinks in the cellwall (Hoffmeyer and Davidson, 1989).
Boyd (1982) in a lengthy paper demonstrateshow creep under both constant and variable rela-tive humidity can be explained quite simply interms of stress-induced physical interactionsbetween the crystalline and non-crystalline com-ponents of the cell wall. Justification of his view-point relies heavily on the concept that the basicstructural units develop a lenticular trellis formatcontaining a water-sensitive gel which changesshape during moisture changes and load applica-tions thereby explaining creep strains.
Attempts have been made to describe creep interms of the fine structure of timber and it has beendemonstrated that creep in the short term is highlycorrelated with the angle of the microfibrils in theS2 layer of the cell wall, and inversely with thedegree of crystallinity. However, such correlationsdo not necessarily prove any causal relation and itis possible to explain these correlations in terms ofthe presence or absence of moisture which wouldbe closely associated with these particular variables.
Environmental effects on rate of creep
Temperature: steady-stateIn common with many other materials, especiallythe high polymers, the effect of increasing tem-perature on timber under stress is to increaseboth the rate and the total amount of creep.Figure 45.14 illustrates a two-and-a-half foldincrease in the amount of creep as the tempera-ture is raised from 20°C to 54°C; there is amarked increase in the irreversible component ofcreep at the higher temperatures.
Various workers have examined the applicabil-ity to wood of the time/temperature superpositionprinciple; results have been inconclusive and vari-able and it would appear that caution must beexercised in the use of this principle.
Temperature: unsteady-stateCycling between low and high temperatures will
induce in stressed timber and panel products ahigher creep response than would occur if thetemperature was held constant at the higher level;however, this effect is most likely due to changingmoisture content as temperature is changed,rather than to the effect of temperature itself.
Moisture content: steady-stateThe rate and amount of creep in timber of highmoisture content is appreciably higher than thatof dry timber; an increase in moisture contentfrom 6 to 12 per cent increases the deflection intimber at any given stress level by about 20 percent. It is interesting to note the occurrence of asimilar increase in creep in nylon when in the wetcondition.
Hunt (1999) recorded that the effect of humid-ity can be treated by the use of an empiricalhumidity-shift factor curve to be used with anempirical master creep curve. At high moisture
Deformation under load
483
FIGURE 45.14 The effect of temperature on rela-tive creep of samples of hoop pine (Araucaria cunning-hamii) loaded in compression for 20, 40 and 60 hours(Kingston and B. Budgen (1972) by permission ofSpringer-Verlag).
contents this logarithmic shift factor increasesrapidly; creep at 22 per cent moisture contentcompared with that at 10 per cent was found tobe 101.5 or 32 times as fast.
Moisture content: unsteady-stateIf the moisture content of small timber beamsunder load is cycled from dry to wet and back todry again, the deformation will also follow acyclic pattern. However, the recovery in each cycleis only partial and over a number of cycles thetotal amount of creep is very large: the greater themoisture differential in each cycle, the higher theamount of creep (Armstrong and Kingston, 1960;Hearmon and Paton, 1964). Figure 45.15 illus-trates the deflection that occurs with time inmatched test pieces loaded to 3/8 ultimate short-term load where one test piece is maintained in anatmosphere of 93 per cent relative humidity, whilethe other is cycled between 0 and 93 per cent rela-tive humidity. After 14 complete cycles the lattertest piece had broken after increasing its initialdeflection by 25 times; the former test piece wasstill intact having increased in deflection by onlytwice its initial deflection. Failure of the first test
FIGURE 45.15 The effect of cyclic variations inmoisture content on relative creep of samples of beechloaded to 1/8 and 3/8 of ultimate load (© BuildingResearch Establishment).
piece occurred, therefore, after only a short periodof time and at a stress equivalent to only 3/8 of itsultimate (Hearmon and Paton, 1964).
It should be appreciated that creep increasedduring the drying cycle and decreased during thewetting cycle with the exception of the initialwetting when creep increased. It was not possibleto explain the negative deflection observed duringabsorption, though the energy for the change isprobably provided by the heat of absorption. Thenet change at the end of a complete cycle ofwetting and drying was considered to be a redis-tribution of hydrogen bonds which manifestsitself as an increase in deformation of the stressedsample (Gibson, 1965).
More early work was to show that the rate ofmoisture change affects the rate of creep, but notthe amount of creep; this appears to be propor-tional to the total change in moisture content(Armstrong and Kingston, 1962).
This complex behaviour of creep in timberwhen loaded under either cyclic or variablechanges in relative humidity has been confirmedby a large number of research workers (e.g.Schniewind, 1968; Ranta-Mannus, 1973; Hunt,1982; Mohager, 1987). However, in boardmaterials the cyclic effect appears to be somewhatreduced (Dinwoodie, 1990b).
Later test work, covering longitudinal compres-sion and tension stressing as well as bending, indi-cated that the relationship between creepbehaviour and moisture change was more complexthan first thought. The results of this work (e.g.Hunt, 1982) was to indicate that there are threeseparate components to this form of creep; theseare illustrated in Figure 45.16 and are:
1. an increase in creep follows a decrease inmoisture content of the sample, as describedpreviously;
2. an increase in creep follows any increase inmoisture content above the previous highestlevel reached after loading; three examplescan be seen in the middle of the graph inFigure 45.16;
3. a decrease in creep follows an increase inmoisture content below the previous highest
Deformation in timber
484
FIGURE 45.16 Creep deflection under changingmoisture content levels for small samples of beechstressed in tension. Note the different responses toincreasing moisture content (from Hunt (1982) by per-mission of the Institute of Wood Science).
level reached after loading as described previ-ously.
It follows from points 1 and 2 above that therewill always be an initial increase in creep duringthe initial change in moisture content, irrespectiveof whether adsorption or desorption is takingplace.
Further experimentation has established thatthe amount of creep that occurs depends on thesize and rate of the moisture change, and is littleaffected by its duration or by whether suchchange is brought about by one or more steps(Armstrong and Kingston, 1962). These findingswere to cast doubt on the previous interpretationthat such behaviour constituted true creep.
Re-inforcement of that doubt occurred withthe publication of results of Arima andGrossman (1978). Small beams of Pinus radiata(680�15�15mm) were cut from green timberand stressed to about 25 per cent of their short-term bending strength. While held in their
deformed condition, the beams were allowed todry for 15 days, after which the retaining clampswere removed and the initial recovery measured.The unstressed beams were then subjected tochanges in relative humidity and Figure 45.17shows the changes in recovery with changinghumidity. Most important is the fact that totalrecovery was almost achieved; what was thoughtto have been viscous deformation in the post-drying and clamping stage turned out to bereversible.
These two phenomena – that creep is related tothe magnitude of the moisture change and not totime, and that deformation is reversible undermoisture change – cast very serious doubts onwhether the deformation under changes in mois-ture content constituted true creep.
It was considered necessary to separate thesetwo very different types of deformation, namelythe true viscoelastic creep that occurs under con-stant moisture content and is directly a functionof time, from that deformation which is directlyrelated to the interaction of change in moisturecontent and mechanical stressing which is a func-tion of the history of moisture change and isrelatively uninfluenced by time. A term of conve-
nience was derived to describe this latter type ofdeformation, namely mechano-sorptive behaviour(Grossman, 1976).
Changing levels of moisture content, however,will result in changes in the dimensions of timberand an allowance for this must be taken intoaccount in the calculation of mechano-sorptivedeformation. Thus,
�m � �vc � �ms � �s (45.11)
where �m � total measured strain, �vc is the normaltime (constant moisture content) viscoelasticcreep, �ms is the mechano-sorptive strain underchanging moisture content and �s is the swellingor shrinkage strain of a matched, zero-loadedcontrol test piece. The swelling/shrinkage strain(�s) (sometimes referred to as pseudo-creep bysome workers), which is manifest during moisturecycling by an increase in deflection during desorp-tion and a decrease during adsorption, has beenascribed to differences in the normal longitudinalswelling and shrinkage of wood: a tensile strainresulting in a smaller shrinkage coefficient and acompression strain resulting in a larger one (Huntand Shelton, 1988).
Mechano-sorptive behaviour is linear only at
Deformation under load
485
FIGURE 45.17 The amount of recovery of both viscoelastic and mechano-sorptive deflection that occurredwhen dried bent beams were subjected to a sequence of humidity changes (adapted from Arima and Grossman(1978) by permission of the Institute of Wood Science).
low levels of stress; in the case of compressionand bending the upper limit of linear behaviour isof the order of 10 per cent, while in tension it isslightly higher (Hunt, 1980).
Susceptibility to mechano-sorptive behaviour ispositively correlated with elastic compliance,microfibrillar angle of the cell wall and dimen-sional change rates (Hunt, 1994). Thus, bothjuvenile wood and compression wood have beenshown to creep much more (up to five timesgreater) than adult wood.
Modelling of deformation undervariable moisture content
The requirement for a model of mechano-sorptivebehaviour was originally set out by Schniewind(1966) and later developed by Grossman (1976)and Mårtensson (1992). More recently, the list ofrequirements that must be satisfied for such amodel has been reviewed and considerablyextended by Hunt (1994); aspects of sorption arenow included on the list.
Many attempts have been made over the lasttwo decades to develop a model for mechano-sorptive behaviour. A few of these models havebeen explanatory or descriptive in nature-seekingrelationships at the molecular, ultrastructural ormicroscopical levels. Most of them, however,have been either purely mathematical, with theaim of producing a generalised constitutive equa-tion, or partly mathematical, where the derivedequation is linked to some physical phenomenon,or change in structure of the timber under stress.
It will be recalled that it is 40 years since theconcept of mechano-sorptive deformation wasfirst established (Armstrong and Kingston, 1960).Since then, the concept of mechano-sorptivedeformation, which is primarily a function of theamount of moisture change, has been clearly sep-arated from that of the viscoelastic creep which isprimarily a function of time. Indeed, over the lasttwo decades the concepts have become morepolarised and considerable success has been madein recent times in the derivation of effective con-stitutive equations for mechano-sorptive defor-mation.
It now appears that the thinking on creep hasgone full circle and is now back to the stagewhere it was once thought that viscoelastic creepand mechano-sorptive behaviour were but twodifferent manifestations of one basic relationship.The published findings by Gril (1988), Hanhijärvi(1995) and Hunt and Gril (1996) indicating thatmechano-sorptive deformation may be accountedfor by a coupling between creep and hygroexpan-sion, together with the recent findings by Hunt(1999) that time-dependent creep and mechano-sorptive behaviour are different means of reach-ing the same creep result, are certainly contraryto views presented over the last 20 years and givecause for a complete re-evaluation of existingconcepts.
Hunt’s results led him to a new way of charac-terising wood creep by plotting data in the formof strain rate against strain: solution of this dif-ferential equation led to the more normal relationof strain against time. It was then found that nor-malisation of both the ordinate and abscissaresulted in a single master creep curve for bothjuvenile and mature wood from a single sampleand, more important, approximately also for alltest humidities. The effects of humidity changesrequire the additional measurement of theincreased activity associated with the moleculardestabilisation, and its relaxation-time constraint,associated with the physical-ageing phenomenon(thermodynamic equilibrium), the application ofwhich suggests that the speed of moisture changemight be important in mechano-sorptive defor-mation.
More information on the parameters thatshould be included in these mathematical modelsas well as the types of models which have beendeveloped over the years are to be found in Tongand Ödeen (1989b), Hunt (1994), Morlier andPalka (1994), Hanhijärvi (1995) and Dinwoodie(2000).
45.5 ReferencesArima, T. and Grossman, P.U.A. (1978) Recovery of
wood after mechano-sorptive deformation. J. Inst.Wood Sci. 8, 2, 47–52.
Deformation in timber
486
Armstrong, L.D. and Kingston, R.S.T. (1960) Effect ofmoisture changes on creep in wood. Nature 185,4716, 862–3.
Armstrong, L.D. and Kingston, R.S.T. (1962) Theeffect of moisture content changes on the deforma-tion of wood under stress. Aust. J. App. Sci. 13, 4,257–76.
Astley, R.J., Stol, K.A. and Harrington, J.J. (1998)Modelling the elastic constraints of softwood. PartII: The cellular microstructure. Holz als Roh-undWerkstoff 56, 43–50.
Barber, N.F. (1968) A theoretical model of shrinkingwood. Holzforschung 22, 97–103.
Barber, N.F. and Meylan, B.A. (1964) The anisotropicshrinkage of wood. A theoretical model. Holz-forschung 18, 146–56.
Boyd, J.D. (1974) Anisotropic shrinkage of wood.Identification of the dominant determinants.Mokuzai Gakkaishi 20, 473–82.
Boyd, J.D. (1982) An anatomical explanation for vis-coelastic and mechano-sorptive creep in wood, andeffects of loading rate on strength. In New Perspec-tives in Wood Anatomy (ed. P. Bass), MartinusNijhoff/Dr W. Junk, The Hague, 171–222.
Cave, I.D. (1968) The anisotropic elasticity of the plantcell-wall. Wood Sci. Technol. 2, 268–78.
Cave, I.D. (1972) A theory of shrinkage of wood.Wood Sci. Technol. 6, 284–92
Cave, I.D. (1975) Wood substance as a water-reactivefibre-reinforced composite, J. of Microscopy, 104, 1,47–52.
Cave, I.D. (1978a) Modelling moisture-related mechani-cal properties of wood. I. Properties of the wood con-stituents. Wood Sci. and Technol. 12, 75–86.
Cave, I.D. (1978b) Modelling moisture-relatedmechanical properties of wood. II. Computation ofproperties of a model of wood and comparison withexperimental data. Wood Sci. Technol. 12, 127–39.
Chow, S. (1973) Molecular rheology of coniferouswood tissues. Transactions of the Society of Rheol-ogy 17, 109–28.
Dinwoodie, J.M. (1975) Timber – a review of thestructure–mechanical property relationship. J. ofMicroscopy 104, 1, 3–32.
Dinwoodie, J.M. (2000) Timber: Its Nature andBehaviour, 2nd edn, E & FN Spon, London, 257 pp.
Dinwoodie, J.M., Higgins, J.A., Paxton, B.H. andRobson, D.J. (1990a) Creep in chipboard. Part 7:Testing the efficacy of models on 7–10 years dataand evaluating optimum period of prediction. WoodSci. Technol. 24, 181–9.
Dinwoodie, J.M., Higgins, J.A., Paxton, B.H. andRobson, D.J. (1990b) Creep research on particleboard – 15 years’ work at UK BRE. Holz als Roh-und Werkstoff 48, 5–10.
Gerhards, C.C. (1982) Effect of moisture content and
temperature on the mechanical properties of wood.An analysis of immediate effects. Wood & Fiber 14,1, 4–36.
Gibson, E. (1965) Creep of wood: role of water andeffect of a changing moisture content. Nature (Lond)206, 213–15.
Gril, J. (1988) Une modélisation du compartementhygro-rhéologique du bois à partir de sa microstruc-ture. Doctoral thesis, University of Paris.
Grossman, P.U.A. (1976) Requirements for a modelthat exhibits mechano-sorptive behaviour. WoodSci. Technol. 10, 163–8.
Hanhijärvi, A. (1995) Modelling of creep deformationmechanisms in wood. Tech. Res. Centre of FinlandPublication 231, 143 pp.
Harington, J.J., Booker, R. and Astley, R.J. (1998)Modelling the elastic constraints of softwood. Part1: The cell-wall lamellae. Holz als Roh-und Werk-stoff 56, 37–41.
Hearmon, R.F.S. and Paton, J.M. (1964) Moisturecontent changes and creep in wood. For. Prod. J. 14,357–9.
Hoffmeyer, P. and Davidson, R. (1989) Mechano-sorp-tive creep mechanism of wood in compression andbending. Wood Sci. Technol. 23, 215–27.
Hunt, D.G. (1980) A preliminary study of tensile creepof beech with concurrent moisture changes. Proceed-ings of the Third International Conference onMechanical Behaviour of Materials, Cambridge, 1979(eds K.J. Miller and R.F. Smith), Vol. 3, 299–308.
Hunt, D.G. (1982) Limited mechano-sorptive creep ofbeech wood. J. Inst. Wood Sci. 9 , 3, 136–8.
Hunt, D.G. (1994) Present knowledge of mechano-sorptive creep of wood. In Creep in Timber Struc-tures (ed. P. Morlier), RILEM Report 8, E & FNSpon, London, 73–97.
Hunt, D.G. (1999) A unified approach to creep inwood. Proc. Roy. Soc. London, A, 455, 4077–95.
Hunt, D.G. and Gril, J. (1996) Evidence of a physicalageing phenomenon in wood. J. Mat. Sci. Letters 15,80–92.
Hunt, D.G. and Shelton, C.F. (1988) Longitudinalmoisture–shrinkage coefficients of softwood at themechano-sorptive creep limit. Wood Sci. Technol.22, 199–210.
Kifetew, G. (1997) Application of the early–latewoodinteraction theory of the shrinkage anisotropy ofScots pine. Proc. International Conference onWood–Water Relations, Copenhagen (ed. P.Hoffmeyer), and published by the management com-mittee of EC COST Action E8, 165–71.
Kingston, R.S.T. and Budgen, B. (1972) Some aspectsof the rheological behaviour of wood, Part IV: Non-linear behaviour at high stresses in bending andcompression. Wood Sci. Technol. 6, 230–8.
Mårtensson, A. (1992) Mechanical behaviour of wood
References
487
exposed to humidity variations. Report TVBK-1066,Lund Institute of Technology, Sweden, 189 pp.
Meylan, B.A. (1968) Cause of high longitudinal shrink-age in wood. For. Prod. J. 18, 4, 75–8.
Mohager, S. (1987) Studier av krypning hos trä(studies of creep in wood). Report 1987–1 of theDept. of Building Materials, The Royal Institute ofTechnology, Stockholm, 140 pp.
Moore, G.L. (1984) The effect of long term tempera-ture cycling on the strength of wood. J. Inst. WoodSci. 9 , 6, 264–7.
Morlier, P. and Palka, L.C. (1994) Basic knowledge. InCreep in Timber Structures (ed. P. Morlier), RilemReport 8, E & FN Spon, London, 9–42.
Nakai, T. and Grossman, P.U.A. (1983) Deflection ofwood under intermittent loading. Part 1: Fortnightlycycles. Wood Sci. Technol. 17, 55–67.
Ranta-Maunus, A. (1973) A theory for the creep ofwood with application to birch and spruce plywood.Technical Research Centre of Finland, BuildingTechnology and Community Development, Publica-tion 4, 35 pp.
Schniewind, A.P. (1966) Uber den Einfluss vonFeuchtigkeitsanderungen auf das Kriechen vonBuchenholz quer zur Faser unter Berucksichtigung
von Temperatur und temperaturanderungen. Holzals Roh-und Werkstoff 24, 87–98.
Schniewind, A.P. (1968) Recent progress in the studyof rheology of wood. Wood Sci. Technol. 2,189–205
Skaar, C. (1988) Wood–Water Relations, Springer-Verlag, Berlin, 283 pp.
Tong, L. and Ödeen, K. (1989a) A non-linear vis-coelastic equation for the deformation of wood andwood structure. Report No. 4 of the Department ofBuilding Materials, Royal Institute of Technology,Stockholm, 6 pp.
Tong, L. and Ödeen, K. (1989b) Rheological behavi-our of wood structures. Report No. 3 of the Depart-ment of Building Materials, Royal Institute ofTechnology, Stockholm, 145 pp.
Weatherwax, R.G. and Stamm, A.J. (1946) The coeffi-cients of thermal expansion of wood and woodproducts. U.S. For. Prod. Lab., Madison, Report no.1487.
Ylinen, A. (1965) Prediction of the time-dependentelastic and strength properties of wood by the aid ofa general non-linear viscoelastic rheological model.Holz als Roh und Werkstoff 5, 193–6.
Deformation in timber
488
Chapter forty-six
Strength andfailure in timber
46.1 Introduction46.2 Determination of strength46.3 Strength values46.4 Variability in strength values46.5 Inter-relationships among the strength properties46.6 Factors affecting strength46.7 Strength, toughness, failure and fracture morphology46.8 Design stresses for timber46.9 References
46.1 IntroductionWhile it is easy to appreciate the concept of defor-mation primarily because it is something that canbe observed, it is much more difficult to define insimple terms what is meant by the strength of amaterial. Perhaps one of the simpler definitions ofstrength is that it is a measure of the resistance tofailure, providing of course that we are clear in ourminds what is meant by failure.
Let us start therefore by defining failure. Inthose modes of stressing where a distinct breakoccurs with the formation of two fracture sur-faces, failure is synonymous with rupture of thespecimen. However, in certain modes of stressing,fracture does not occur and failure must bedefined in some arbitrary way such as themaximum stress that the sample will endure or,in exceptional circumstances such as compressionstrength perpendicular to the grain, the stress atthe limit of proportionality.
Having defined our end point, it is now easierto appreciate our definition of strength as thenatural resistance of a material to failure. Buthow do we quantify this resistance? This may be
done by calculating either the stress necessary toproduce failure or the amount of energy con-sumed in producing failure. Under certain modesof testing it is more convenient to use the formermethod of quantification as the latter tends to bemore limited in application.
46.2 Determination of strength
46.2.1 Test piece size and selection
Although in theory it should be possible to deter-mine the strength properties of timber independ-ent of size, in practice this is found not to be thecase. A definite though small size effect has beenestablished and in order to compare the strengthof a timber sample with recorded data it is advis-able to adopt the standard sizes set out in theliterature.
The size of the test piece to be used will bedetermined by the type of information required.When this has been decided, it will determine thetest procedure; standardised test proceduresshould always be adopted and a choice is avail-able between National, European or Internationalmethods of test (see Figure 46.1).
Use of small clear test pieces
This size of test piece was originally used for thederivation of working stresses for timber, but inthe mid-1970s it was superseded by actual struc-tural size timber. However, the small clear testpiece still remains valid for characterising new
489
timbers and for the strict academic comparison ofwood from different trees or different species,since the use of small, knot-free, straight-grained,perfect test pieces represent the maximum qualityof wood that can be obtained. As such, these testpieces are not representative of structural-sizetimber with all its imperfections; several arbit-rarily defined reduction factors have to be used inorder to obtain a measure of the working stressesof the timber, when small, clear test pieces areused (see Section 46.8.1).
Two standard procedures for testing smallclear test pieces have been used internationally;the original was introduced in the USA as early as1891 using a test sample 2�2 inches in cross-section; the second, European in origin, employsa test specimen 20�20mm in cross-section. Priorto 1949 the former size was adopted in the UK,but after this date this larger sample was super-seded by the smaller, thereby making it possibleto obtain an adequate number of test specimensfrom smaller trees. Because of the differences insize, the results obtained from the two standardprocedures are not strictly comparable and aseries of conversion values has been determined(Lavers, 1969).
The early work in the UK on species characteri-sation employed a sampling procedure in which
Strength and failure in timber
490
Small clears:cross-section20 � 20 mme.g. BS 373
or 50 � 50 mme.g. ASTM D143–52
UsingNational
standardse.g. BS 5820
Using newEuropeanstandards
e.g. EN 480
Test piece size
Structural size
FIGURE 46.1 Alternative sizes of test piece to beused in the determination of the strength of timber.
the test samples were removed from the log inaccordance with a cruciform pattern. However,this was subsequently abandoned and a methoddevised applicable to the centre plank removedfrom a log; 20�20mm sticks, from which theindividual test pieces are obtained, are selected atrandom in such a manner that the probability ofobtaining a stick at any distance from the centreof a cross-section of a log is proportional to thearea of timber at that distance. Test samples arecut from each stick eliminating knots, defects andsloping grain; this technique is described fully byLavers (1969).
Use of structural-size test pieces
The use of these larger test pieces reproducesactual service loading conditions and they are ofparticular value because they allow directly fordefects such as knots, splits and distorted grainrather than by applying a series of reductionfactors as is necessary with small clear test pieces.However, use of the large pieces is probably morecostly.
46.2.2 Standardised test procedures
Europe at the present time is in a transitionperiod in which National test procedures arebeing replaced by European procedures. Withinthe first decade of this century all National stand-ards relating to the use of timber and panel prod-ucts in construction will be withdrawn; it isuncertain at this stage whether the testing ofsmall, clear test pieces will be retained as aNational Standard provided it is not used for thederivation of characteristic values (workingstresses). It is interesting to note that many of theEuropean standards (ENs) have now beenadopted as International Standards (ISOs).
46.2.3 Methods of test
Methods of test are given in Desch and Din-woodie (1996) and Dinwoodie (2000).
46.3 Strength values
46.3.1 Derived using small clear testpieces
For a range of strength properties determined onsmall, clear test pieces, with the exception oftensile strength parallel to the grain, the meanvalues and standard deviations (see below) arepresented in Table 46.1 for a selection of timberscovering the range in densities to be found inhardwoods and softwoods. Many of the timberswhose elastic constants were presented in Table45.1 are included. All values relate to a moisturecontent which is in equilibrium with a relativehumidity of 65 per cent at 20°C; these are of theorder of 12 per cent and the timber is referred toas ‘dry’. Modulus of elasticity has also beenincluded in Table 46.1.
Table 46.1 is compiled from data presented inBulletin 50 of the Forest Products Research Labo-ratory (Lavers, 1969) which lists data for boththe dry and green states for 200 species of timber.The upper line for each species provides the esti-mated average value while the lower line containsthe standard deviation.
In Table 46.2 tensile strength parallel to thegrain is listed for certain timbers and it is in thismode that timber is at its strongest.
Comparison of these values with those for compression strength parallel to the grain in Table 46.1 will indicate that, unlike many othermaterials, the compression strength is only aboutone-third that of tensile strength along the grain.
46.3.2 Derived using structural size testpieces
Structural strength values may still be recordedand used as mean values with a standard devia-tion where National standards are still being usedin testing (e.g. BS 5820) and design (e.g. BS 5268Pt2 1996).
An example of mean strength values derivedfrom testing dry structural size timber to BS5820 is given in Table 46.3 for each of themajor strength modes. Not only are these mean
values considerably lower than the mean valuesderived from small, clear test pieces (Tables 46.1and 46.2), but the tensile strengths are nowlower than the compression strengths. This isdirectly related to the presence of knots andassociated distorted grain in the structural sizetest pieces.
By the year 2010, however, all structural testwork and design within Europe will have to becarried out according to the new European stand-ards by testing to EN 384 and EN 408 anddesigning using Eurocode 5 (ENV 1995–1–1).Within the European system, the characteristicvalue for the strength properties is taken as the 5-percentile value; for modulus of elasticity thereare two characteristic values, one the 5-percentile,the other the mean or 50-percentile value.
The sample 5-percentile value is determined foreach sample by the equation
f05 � fr (46.1)
where f05 is the sample 5-percentile value, and fr isobtained by ranking all the test values for asample in ascending order. The 5-percentile valueis the test value for which 5 per cent of the valuesare lower. If this is not an actual test value (i.e.the number of test values is not divisible by 20)then interpolation between the two adjacentvalues is permitted.
The characteristic value of strength (fk) is calcu-lated from the equation
fk � f–
05kskv (46.2)
where f–
05 is the mean (in N/mm2) of the adjusted5-percentile values (f05) for each sample weightedaccording to the number of pieces in each sample,ks is a factor to adjust for the number of samplesand their size, and kv is a factor to allow for thelower variability of f05 values from machinegrades in comparison with visual grades; forvisual grades kv �1.0, and for machine gradeskv �1.12
46.4 Variability in strength valuesIn Chapter 44 attention was drawn to the factthat timber is a very variable material and that
Variability in strength values
491
TA
BL
E4
6.1
Ave
rage
and
sta
ndar
d de
viat
ion
of v
ario
us m
echa
nica
l pr
oper
ties
of
sele
cted
tim
bers
at
12 p
er c
ent
moi
stur
e co
nten
t fr
omsm
all,
clea
r te
st p
iece
s
Den
sity
Stat
ic b
endi
ng*
Impa
ctC
ompr
essi
onH
ardn
ess
Shea
rC
leav
age
Dry
Mod
ulus
of
Mod
ulus
of
Ene
rgy
toE
nerg
y to
Dro
p of
Par
alle
l to
On
side
Par
alle
l to
Rad
ial
Tan
gent
ial
(kg/
m3 )
rupt
ure
elas
tici
tym
ax. l
oad
fact
ure
ham
mer
grai
ngr
ain
grai
npl
ane
plan
e(N
/mm
2 )(N
/mm
2 )(m
mN
/mm
3 )(m
mN
/mm
3 )(m
)(N
/mm
2 )(N
)(N
/mm
2 )(N
/mm
(N/m
mw
idth
)w
idth
)
Har
dwoo
dsB
alsa
176
2332
000.
018
0.03
5–
15.5
–2.
4–
–7.
310
600.
007
0.01
74.
430.
62O
bech
e36
854
5500
0.05
80.
095
0.48
28.2
1910
7.7
9.3
8.4
6.5
620
0.01
00.
015
0.07
23.
0026
80.
671.
821.
58M
ahog
any
497
7890
000.
070
0.12
80.
5846
.436
9011
.810
.014
.0(K
haya
ivor
ensi
s)15
.015
200.
026
0.04
40.
149
8.45
816
2.56
2.08
2.90
Syca
mor
e56
199
9400
0.12
10.
163
0.84
48.2
4850
17.1
16.8
27.3
11.0
1160
0.02
80.
049
0.13
64.
8363
92.
322.
953.
91A
sh68
911
611
900
0.18
20.
281
1.07
53.3
6140
16.6
16.6
2170
0.04
50.
097
0.21
67.
7311
582.
52O
ak68
997
1010
00.
093
0.16
70.
8451
.654
7013
.714
.520
.116
.819
600.
026
0.05
10.
209
7.98
911
2.38
2.86
2.08
Afz
elia
817
125
1310
00.
100
0.20
30.
7979
.278
7016
.610
.513
.326
.617
600.
043
0.08
70.
215
12.0
291
42.
282.
002.
49G
reen
hear
t97
718
121
000
0.21
30.
395
1.35
89.9
1045
020
.517
.522
.220
.919
900.
047
0.08
80.
207
8.49
1531
3.06
4.79
4.97
Soft
woo
dsN
orw
ay s
pruc
e41
772
1020
00.
086
0.11
60.
5836
.521
409.
88.
49.
1(E
urop
ean)
10.2
2010
0.02
20.
040
0.11
65.
2635
31.
441.
071.
20Y
ello
w p
ine
433
8083
000.
089
0.09
70.
5642
.120
509.
38.
211
.6(C
anad
a)10
.914
400.
015
0.01
90.
100
6.14
473
1.61
1.57
1.77
Dou
glas
fir
497
9110
500
0.09
70.
172
0.69
48.3
3420
11.6
9.5
11.4
(UK
)16
.921
600.
038
0.08
10.
200
8.03
865
2.29
1.90
2.17
Scot
s pi
ne51
389
1000
00.
103
0.13
40.
7147
.429
8012
.710
.313
.0(U
K)
16.9
2130
0.03
20.
053
0.16
79.
2569
72.
451.
822.
47C
arib
bean
Pit
ch p
ine
769
107
1260
00.
126
0.25
30.
9156
.149
8014
.312
.113
.314
.518
000.
042
0.06
00.
196
7.76
1324
2.81
1.23
1.58
*In
thre
e-po
int
load
ing.
for many of its parameters, e.g. density, celllength and microfibrillar angle of the S2 layer, dis-tinct patterns of variation could be establishedwithin a growth ring, outwards from the pithtowards the bark, upwards in the tree, and fromtree to tree. The effects of this variation in struc-ture are all too apparent when mechanical testsare performed.
Test data for small, clear test pieces are usuallyfound to follow a normal distribution and, asdescribed earlier an efficient estimator of the vari-ability which occurs in any one property is thesample standard deviation, denoted by s. It is thesquare root of the variance and is derived fromthe formula:
s��� (46.3)
where x stands for every item successively and nis the number of items in the sample of eithersmall, clear or structural size test pieces.
In a normal distribution, approximately 68 percent of the results should lie, in theory, within�1 s and �1 s of the mean, and 99.87 per centshould fall within !3 s of the mean.
As a general rule, a normal distribution curvefits the data from small, clear test pieces betterthan does the data from structural size test pieces,where as noted earlier, the 5-percentile character-istic value is determined simply by ranking theresults.
The standard deviation provides a measure ofthe variability, but in itself gives little impressionof the magnitude unless related to the mean. Thisratio is known as the coefficient of variation, i.e.
CV��me
s
an�% (46.4)
�x2 ��(�x
n
)2
�
��n�1
Variability in strength values
493
TABLE 46.2 Tensile strength parallel to the grain of certain timbers using small, clear test pieces
Timber Moisture Tensile content (%) strength (N/mm2)
HardwoodsAsh (Home grown) 13 136Beech (Home grown) 13 180Yellow poplar (Imported) 15 114
SoftwoodsScots pine (Home grown) 16 92Scots pine (Imported) 15 110Sitka spruce (Imported) 15 139Western hemlock (Imported) 15 137
TABLE 46.3 Mean values for dry strength derived on structural size test pieces (approx. 97 �47mm) using BS5820 (m.c.�15–18 per cent)
Timber Mean values (N/mm2)
Bending Tension Compression
Sitka spruce (UK) 32.8 19.7 29.5Douglas fir (UK) 35.7 21.4 32.1Spruce/pine/fir (Canada) 43.9 26.3 39.5Norway spruce (Baltic) 50.9 30.5 45.8
The coefficient of variation varies considerably,but is frequently under 15 per cent for many bio-logical applications. However, reference to Table46.1 will indicate that this value is frequentlyexceeded. For design purposes the two mostimportant properties are the moduli of elasticityand rupture which have coefficients of variationtypically in the range of 10–30 per cent.
46.5 Inter-relationships among thestrength properties
46.5.1 Modulus of rupture and modulusof elasticity
A high correlation exists between the moduli ofrupture and elasticity for a particular species: butit is doubtful whether this represents any causalrelationship; rather it is more probable that thecorrelation arises as a result of the strong correla-tion that exists between density and eachmodulus. Whether it is a causal relationship ornot, it is nevertheless put to good advantage for itforms the basis of the stress grading of timber bymachine (see Section 46.8). The stiffness of apiece of timber is measured as it is deflectedbetween rollers and this is used to predict itsstrength.
46.5.2 Impact bending and total work
Good correlations have been established betweenthe height of drop in impact bending test andboth work to maximum load and total work:generally the correlation is higher with the latterproperty.
46.5.3 Hardness and compressionperpendicular to the grain
Correlation coefficients of 0.902 and 0.907 havebeen established between hardness and compres-sion strength perpendicular to the grain of timberat 12 per cent moisture content and timber in thegreen state respectively. It is general practice to
predict the compression strength from the hard-ness result using the following equations:
Y12 �0.00147x12 �1.103 (46.5)
Yg �0.00137xg �0.207 (46.6)
where Yg and Y12 are compression strength per-pendicular to the grain in N/mm2 for greentimber and timber at 12 per cent moisturecontent respectively, and xg and x12 are hardnessin N.
46.6 Factors affecting strengthMany of the variables noted in the previouschapter as influencing stiffness also influence thevarious strength properties of timber. Once again,these can be regarded as being either materialdependent or manifestations of the environment.
46.6.1 Anisotropy and grain angle
The marked difference between the longitudinaland transverse planes in both shrinkage and stiff-ness has been discussed in previous chapters.Strength likewise is directionally dependent andthe degree of anisotropy present in both tensionand compression is presented in Table 46.4 forsmall clear test pieces of Douglas fir. Irrespectiveof moisture content, the highest degree ofanisotropy is in tension (48:1): this reflects thefact that the highest strength of clear, straight-grained timber is in tension along the grain whilethe lowest is in tension perpendicularly. A similardegree of anisotropy is present in the tensilestressing of both glass-reinforced plastics andcarbon-fibre-reinforced plastics when the fibre islaid up in parallel strands.
Table 46.4 also demonstrates that the degree ofanisotropy in compression is an order of magni-tude less than in tension. Whilst the compressionstrengths are markedly affected by moisturecontent, tensile strength appears to be relativelyinsensitive, reflecting the exclusion of moisturefrom the crystalline core of the microfibril; it isthis crystalline core that imparts to timber its veryhigh longitudinal tensile strength. The compari-
Strength and failure in timber
494
son of tension and compression strengths alongthe grain in Table 46.4 reveals that clear,straight-grained timber, unlike most other mater-ials, has a tensile strength considerably greaterthan the compression strength. In structuraltimber containing knots and distorted grain, theopposite is the norm (see Table 46.3)
Anisotropy in strength is due in part to the cel-lular nature of timber and in part to the structureand orientation of the microfibrils in the walllayers. Bonding along the direction of themicrofibrils is covalent whilst bonding betweenmicrofibrils is by hydrogen bonds. Consequently,since the majority of the microfibrils are alignedat only a small angle to the longitudinal axis, itwill be easier to rupture the cell wall if the load isapplied perpendicular than if applied parallel tothe fibre axis.
Since timber is an anisotropic material, itfollows that the angle at which stress is appliedrelative to the longitudinal axis of the cells willdetermine the ultimate strength of the timber.Figure 46.2 illustrates that over the range 0–45°tensile strength is much more sensitive to grainangle than is compression strength. However, atangles as high as 60° to the longitudinal axisboth tension and compression strengths havefallen to only about 10 per cent of their value in straight-grained timber. The sensitivity ofstrength to grain angle in clear straight-grainedtimber is identical with that for fibre orientationin both glass-fibre- and carbon-fibre-reinforcedplastics.
It is possible to obtain an approximate value ofstrength at any angle to the grain from know-ledge of the corresponding values for both paral-
Factors affecting strength
495
TABLE 46.4 Anisotropy in strength in small, clear test pieces
Timber Moisture Tension Compressioncontent
� � �:� � � �:�(%)(N/mm2) (N/mm2) (N/mm2) (N/mm2)
Douglas fir �27 131 2.69 48.7 24.1 4.14 5.82
Douglas fir 12 138 2.90 47.6 49.6 6.90 7.19
FIGURE 46.2 Effect of grain angle on the tensile,bending and compression strength of timber.
lel and perpendicular to the grain using thefollowing formula, which, in its original form,was credited to Hankinson:
f� � (46.7)
where f� is the strength property at angle � fromthe fibre direction, fL is the strength parallel to thegrain, fT is the strength perpendicular to grain,and n is an empirically determined constant; in
fL � fT��fL sinn � � fT cosn �
tension, n�1.5�2 while in compression,n�2�2.5. The equation has also been used forstiffness where a value of 2 for n has beenadopted.
46.6.2 Knots
Knots are associated with distortion of the grainand since even slight deviations in grain anglereduce the strength of the timber appreciably, itfollows that knots will have a marked influenceon strength. The significance of knots, however,will depend on their size and distribution bothalong the length of a piece of timber and acrossits section. Knots in clusters are more importantthan knots of a similar size which are evenly dis-tributed, while knots on the top or bottom edgeof a beam are more significant than those in thecentre; large knots are much more critical thansmall knots.
It is very difficult to quantify the influence ofknots; one of the parameters that has been suc-cessfully used is the knot area ratio, which relatesthe sum of the cross-sectional area of the knots ata cross-section to the cross-sectional area of thepiece. The loss in bending strength that occurredwith increasing knot area ratio in 200 home-grown Douglas fir boards is illustrated in Figure46.3.
The very marked reduction in tensile strengthof structural size timber compared with smallclear test pieces (Tables 46.2 and 46.3) is due pri-marily to the presence and influence of knots inthe former.
46.6.3 Density
In Section 44.2 density was shown to be a func-tion of cell wall thickness and therefore depend-ent on the relative proportions of the various cellcomponents and also on the level of cell walldevelopment of any one component. However,variation in density is not restricted to differentspecies, but can occur to a considerable extentwithin any one species and even within a singletree. Some measure of the interspecific variationthat occurs can be obtained from both Figure
Strength and failure in timber
496
FIGURE 46.3 Effect of knot area ratio on thestrength of timber (© Building Research Establish-ment).
44.20 and the limited amount of data in Tables45.1 and 46.1. It will be observed from the latterthat as density increases, so stiffness and thevarious strength properties increase. Density con-tinues to be the best predictor of timber strengthsince high correlations between strength anddensity are a common feature in timber studies.
Most of the relations that have been estab-lished throughout the world between the variousstrength properties and timber density take theform of
f�kgn (46.8)
where f is any strength property, g is the specificgravity, k is a proportionality constant differingfor each strength property, and n is an exponentthat defines the shape of the curve. An example ofthe use of this expression on the results of over200 species tested in compression parallel to thegrain is presented in Figure 46.4: the correlationcoefficient between compression strength anddensity of the timber at 12 per cent moisturecontent was 0.902.
Similar relationships have been found to holdfor other strength properties, though in some thedegree of correlation is considerably lower. Thisis the case in tension parallel to the grain wherethe ultrastructure probably plays a more signific-ant role.
Over the range of density of most of thetimbers used commercially, the relationshipbetween density and strength can safely beassumed to be linear with the possible exceptionof shear and cleavage; similarly, within a singlespecies, the range is low and the relationship canagain be treated as linear.
46.6.4 Ring width
Since density is influenced by the rate of growthof the tree, it follows that variations in ring widthwill change the density of the timber and hencethe strength. However, the relationship is consid-erably more complex than it first appears. In thering-porous timbers such as oak and ash (seeChapter 44), increasing rate of growth (ring width)results in an increase in the percentage of the late-wood which contains most of the thick-walledfibres; consequently, density will increase and sowill strength. However, there is an upper limit toring width beyond which density begins to fall
Factors affecting strength
497
FIGURE 46.4 The relation of maximum compression strength to specific gravity for 200 species tested in thegreen and dry states (© Building Research Establishment).
owing to the inability of the tree to produce therequisite thickness of wall in every cell.
In the diffuse-porous timbers such as beech,birch and khaya, where there is uniformity instructure across the growth ring, increasing rateof growth (ring width) has no effect on densityunless, as before, the rate of growth is excessive.
In the softwoods, however, increasing rate ofgrowth results in an increased percentage of thelow-density earlywood and consequently bothdensity and strength decrease as ring widthincreases. Exceptionally, it is found that verynarrow rings can also have very low density: thisis characteristic of softwoods from the verynorthern latitudes where latewood developmentis restricted by the short summer period. Hencering width of itself does not affect the strength ofthe timber: nevertheless, it has a most importantindirect effect working through density.
46.6.5 Ratio of latewood to earlywood
Since the latewood comprises cells with thickerwalls, it follows that increasing the percentage oflatewood will increase the density and therefore thestrength of the timber. Differences in strength of150–300 per cent between the late and earlywoodare generally explained in terms of the thicker cellwalls of the former: however, some workers main-tain that when the strengths are expressed in termsof the cross-sectional area of the cell wall, the late-wood cell is still stronger than the earlywood.Various theories have been advanced to account forthe higher strength of the latewood wall material;the more acceptable are couched in terms of thedifferences in micro-fibrillar angle in the middlelayer of the secondary wall, differences in degree ofcrystallinity and lastly, differences in the proportionof the chemical constituents.
46.6.6 Cell length
Since the cells overlap one another, it follows thatthere must be a minimum cell length below whichthere is insufficient overlap to permit the transferof stress without failure in shear occurring. Someinvestigators have gone further and have argued
Strength and failure in timber
498
FIGURE 46.5 Effect of microfibrillar angle on thetensile strength of Pinus radiata blocks (from Cave(1969) by permission of Springer-Verlag).
that there must be a high degree of correlationbetween the length of the cell and the strength ofcell wall material, since a fibre with high strengthper unit of cross-sectional area would require alarger area of overlap in order to keep constantthe overall efficiency of the system.
46.6.7 Microfibrillar angle
The angle of the microfibrils in the S2 layer has amost significant effect in determining the strengthof wood. Figure 46.5 illustrates the markedreduction in tensile strength that occurs withincreasing angle of the microfibrils: the effect onstrength closely parallels that which occurs withchanging grain angle.
46.6.8 Chemical composition
In Chapter 44 the structure of the cellulose mole-cule was described and emphasis was placed on
the existence in the longitudinal plane of covalentbonds both within the glucose units and alsolinking them together to form filaments contain-ing from 5000 to 10000 units. There is littledoubt that the high tensile strength of timberowes much to the existence of this covalentbonding. Certainly, experiments in which manyof the � 1–4 linkages have been ruptured bygamma irradiation, resulting in a decrease in thenumber of glucose units in the molecule fromover 5000 to about 200, have resulted in a mostmarked reduction in tensile strength; it has alsobeen shown that timber with inherently low mole-cular lengths, e.g. compression wood, has a lowerthan normal tensile strength.
Until the 1970s, it had been assumed that thehemicelluloses which constitute about half of thematrix material played little or no part in deter-mining the strength of timber. However, it hasnow been demonstrated that some of the hemi-celluloses are orientated within the cell wall andit is now thought that these will be load-bearing.
It is known that lignin is less hydrophilic thaneither the cellulose or hemicelluloses and, as indi-cated earlier, at least part of its function is toprotect the more hydrophilic substances from theingress of water and consequent reduction instrength. Apart from this indirect effect onstrength, lignin is thought to make a not tooinsignificant direct contribution. Much of thelignin in the cell wall is located in the primarywall and in the middle lamella. Since the tensilestrength of a composite with fibres of a definitelength will depend on the efficiency of the trans-fer of stress by shear from one fibre to the next, itwill be appreciated that in timber the lignin playsa most important role. Compression strengthalong the grain has been shown to be affected bythe degree of lignification not between the cells,but rather within the cell wall, when all the othervariables have been held constant.
It would appear, therefore, that both the fibreand the matrix components of the timber com-posite are contributing to its strength as, in fact,they do in most composites: the relative signific-ance of the fibre and matrix roles will vary withthe mode of stressing.
46.6.9 Reaction wood
Compression wood
The chemical and anatomical properties of thisabnormal wood, which is found only in the soft-woods, were described in Section 44.4. Whenstressed, it is found that the tensile strength andtoughness is lower and the compressive strengthhigher than that of normal timber. Such differ-ences can be explained in terms of the changes infine structure and chemical composition.
Tension wood
This second form of abnormal wood, which isfound only in the hardwoods, has tensilestrengths higher and compression strengths lowerthan normal wood: again this can be related tochanges in fine structure and chemical composi-tion (see Section 44.4).
46.6.10 Moisture content
The marked increase in strength on drying fromthe fibre saturation point to oven-dry conditionswas described in detail in Section 44.7.5 andillustrated in Figure 44.22; experimentation hasindicated the probability that at moisture con-tents of less than 2 per cent the strength of timbermay show a slight decrease rather than the previ-ously accepted continuation of the upward trend.
Confirmatory evidence of the significance ofmoisture content on strength is forthcoming fromFigure 46.4 in which the regression line for over200 species of compression strength of greentimber against density is lower than that fortimber at 12 per cent moisture content; strengthdata for timber are generally presented for thesetwo levels of moisture content (Lavers, 1969).
However, reference to Table 46.4 will indicatethat the level of moisture has almost no effect onthe tensile strength parallel to the grain. Thisstrength property is determined by the strength ofthe covalent bonding along the molecule andsince, the crystalline core is unaffected by mois-ture (Chapter 44), retention of tensile strength
Factors affecting strength
499
parallel to the grain with increasing moisturecontent is to be expected.
Within certain limits and excluding tensilestrength parallel, the regression of strength,expressed on a logarithmic basis, and moisturecontent can be plotted as a straight line. The rela-tionship can be expressed mathematically as:
log10 f� log10 fs �k(�s � �) (46.9)
where f is the strength at moisture content µ, fs isthe strength at the fibre saturation point, �s is themoisture content at the fibre saturation point,and k is a constant. It is possible, therefore, tocalculate the strength at any moisture contentbelow the fibre saturation point, assuming fs to bethe strength of the green timber and �s to be 27per cent. This formula can also be used to deter-mine the strength changes that occur for a 1 percent increase in moisture content over certainranges (see Table 46.5); the table illustrates forsmall clear test pieces how the change in strengthper unit change in moisture content is non-linear.
This relation between moisture content andstrength may not always apply when the timbercontains defects as is the case with structural sizetimber. Thus, it has been shown that the effect ofmoisture content on strength diminishes as thesize of knots increase.
The relation between moisture content andstrength presented above, even for knot-freetimber, does not always hold for the impact resis-tance of timber. In some timbers, though cer-tainly not all, impact resistance or toughness ofgreen timber is considerably higher than it is in
the dry state; the impact resistances of green ash,cricket bat willow and teak are approximately 10per cent, 30 per cent and 50 per cent higherrespectively than the values at 12 per cent mois-ture content.
In the case of structural timber, several types ofmodel have been proposed to represent moisture-property relationships; these models reflect thefinding that increases in strength with drying aregreater for high strength structural timber thanfor low strength material.
46.6.11 Temperature
At temperatures within the range �200°C to�200°C and at constant moisture content,strength properties are linearly (or almost lin-early) related to temperature, decreasing withincreasing temperature. However, a distinctionmust be made between short- and long-termeffects.
When timber is exposed for short periods oftime to temperatures below 95°C the changes instrength with temperature are reversible. Thesereversible effects can be explained in terms of theincreased molecular motions and greater latticespacing at higher temperatures. For all of thestrength properties, with the possible exception oftensile parallel to the grain (Gerhards, 1982), agood rule of thumb is that an increase in temper-ature of 1°C produces a 1 per cent reduction intheir ultimate values.
At temperatures above 95°C, or at tempera-tures above 65°C for very long periods of time,
Strength and failure in timber
500
TABLE 46.5 Percentage change in strength and stiffness of Scots pine timber per 1 per cent change in moisturecontent (Lavers, 1983)
Property Moisture range (%)
6–10 12–16 20–24
Modulus of elasticity (MOE: stiffness) 0.21 0.18 0.15Modulus of rupture (MOR: bending strength) 4.2 3.3 2.4Compression, perpendicular to the grain 2.7 2.0 1.40Hardness 0.058 0.053 0.045Shear, parallel to the grain 0.70 0.53 0.36
there is an irreversible effect of temperature dueto thermal degradation of the wood substance,generally taking the form of a marked shorteningof the length of the cellulose molecules and chemi-cal changes within the hemicelluloses (see Section47.2.3). All strength properties show a markedreduction with temperature, but toughness isparticularly sensitive to thermal degrade.Repeated exposure to elevated temperature has acumulative effect and usually the reduction isgreater in the hardwoods than in the softwoods.Even exposure to cyclic changes in temperatureover long periods of time has been shown toresult in thermal degradation and loss in strengthand especially toughness.
The effect of temperature is very dependent onmoisture content, sensitivity of strength to tem-perature increasing appreciably as moisturecontent increases (Figure 46.6), as occurs alsowith stiffness (Figure 45.9); these early resultshave been confirmed by Gerhards (1982). Therelationship between strength, moisture contentand temperature appear to be slightly curvilinearover the range 8–20 per cent m.c. and �20° to60°C. However, in the case of toughness, while atlow moisture content it is found that toughnessdecreases with increasing temperature, at high
Factors affecting strength
501
FIGURE 46.6 The effect of temperature on thebending strength of Pinus radiata timber at differentmoisture contents.
moisture contents toughness actually increaseswith increasing temperature.
46.6.12 Time
In Chapter 45 timber was described as a visco-elastic material and as such its mechanicalbehaviour will be time dependent. Such depen-dence will be apparent in terms of its sensitivityto both rate of loading and duration of loading.
Rate of loading
Increase in the rate of load application results inincreased strength values, the increase in ‘green’timber being some 50 per cent greater than thatof timber at 12 per cent moisture content; strainto failure, however, actually decreases. A varietyof explanations have been presented to accountfor this phenomenon, most of which are based onthe theory that timber fails when a critical strainhas been reached and consequently, at lower ratesof loading, viscous flow or creep is able to occurresulting in failure at lower loads.
The various standard testing proceduresadopted throughout the world set tight limits onthe speed of loading in the various tests: unfortu-nately, such recommended speeds vary through-out the world, thereby introducing errors in thecomparison of results from different laboratories;the introduction of European standards (ENs)and the wider use of International standards(ISOs), should give rise to greater uniformity inthe future.
Duration of load (DOL)
In terms of the practical use of timber, the dura-tion of time over which the load is applied isperhaps the single most important variable. Manyinvestigators have worked in this field and eachhas recorded a direct relationship between thelength of time over which a load can be sup-ported at constant temperature and moisturecontent and the magnitude of the load. This rela-tion appears to hold true for all loading modes,but is especially important for bending strength.
The modulus of rupture (maximum bendingstrength) will decrease in proportion, or nearly inproportion, to the logarithm of the time overwhich the load is applied; failure in this particulartime-dependent mode is termed creep rupture orstatic fatigue. Wood (1951) indicated that therelation was slightly different for ramp and con-stant loading, was slightly curvilinear and thatthere was a distinct levelling off at loadsapproaching 20 per cent of the ultimate short-term strength such that a critical load or stresslevel occurs, below which failure is unlikely tooccur; the hyperbolic curve that fitted Wood’sdata best for both ramp and sustained loading,and which became known as the Madison curve,is illustrated in Figure 46.7.
Other workers have reported a linear relation,though a tendency to non-linear behaviour atvery high stress levels has been recorded by some.Pearson (1972), in reviewing previous work inthe field of duration of load and bendingstrength, plotted on a single graph the resultsobtained over a 30-year period and found that,
Strength and failure in timber
502
FIGURE 46.7 The effect of duration of load on thebending strength of timber (after Wood (1951) by per-mission of the Forest Products Laboratory, Madison,and Pearson (1972) by permission of Walter deGruyter & Co., Berlin, New York).
despite differences in method of loading (ramp orconstant), species, specimen size, moisturecontent, or whether the timber was solid or lami-nated, the results showed remarkably little scatterfrom a straight line described by the regression:
f�91.5�7 log10t (46.10)
where f is the stress level (percentage), and t is theeffective duration of maximum load in hours.This regression is also plotted in Figure 46.7 forcomparison with Wood’s curvilinear line.Pearson’s findings certainly threw doubt on theexistence of a critical stress level below whichcreep rupture does not occur. These regressionsindicate that timber beams which have to with-stand a dead load for 50 years can be stressed toonly 50 per cent of their ultimate short-termstrength.
Although this type of log-linear relationship isstill employed for the derivation of duration ofload factors for wood-based panel products, thisis certainly not the case with solid timber.
By the early 1970s, there was abundant evid-ence to indicate that the creep rupture response ofstructural timber beams differed considerablyfrom the classic case for small clear test piecesdescribed above.
Between 1970 and 1985 an extensive amountof research was carried out in America, Canadaand Europe; readers desirous of following thehistorical development of the new concepts inDOL are referred to the comprehensive reviewsby Tang (1994) and Barrett (1996). This researchconfirmed that the DOL effect in structuraltimber was different to that in small clear testpieces and was also less severe than the Madisoncurve predicted for loading periods up to oneyear: it also confirmed that high strength timberpossessed a larger DOL effect than low strengthtimber.
The above test work clearly indicated that theDOL factors then in current use were conserva-tive and in order to obtain a more realistic predic-tion of time to failure, attention moved to thepossible application of reliability-based designprinciples for the assessment of the reliability oftimber members under in-service loading con-
ditions. In particular, this approach led to theadoption of the concept of damage accumulation.
It should be appreciated that in the applicationof this concept there does not exist any methodfor quantifying the actual damage; the develop-ment of damage is simply deduced from the time-to-failure data from long-term loadingexperiments under a given loading history. Thesemodels generally use the stress-level history as themain variable and are thus independent of mater-ial strength. In order to calculate time to failurefor a given stress history under constant tempera-ture and humidity, the damage rate is integratedfrom an assumed initial value of 0 to the failurevalue of 1 (Morlier et al., 1994). Several types of damage accumulation models have beenrecorded, of which the most important are thoselisted by Morlier et al. (1994) and Tang (1994).
The dependence of these damage models on thestress ratio results in a logistical problem, sinceboth the short-term and the long-term strengthhas to be known for the same structural testpiece, but the test piece can be tested only once.This problem is usually resolved by using twoside-matched test pieces and assuming equalstrengths! A comparison among four of the above
damage accumulation models, together with onemodel based on strain energy (Fridley, 1992a), ispresented in Figure 46.8, from which it will benoted how large is the variability among them.
The level of both moisture content and temper-ature has a marked effect on time to failure.Thus, increasing levels of relative humidity resultin reduced times to failure when stressed at thesame stress ratios (Fridley et al., 1991), whilevarying levels of humidity have an even greatereffect in reducing times to failure (Hoffmeyer,1990; Fridley et al., 1992b).
46.7 Strength, toughness, failureand fracture morphologyThere are two fundamentally differentapproaches to the concept of strength and failure.The first is the classical strength of materialsapproach, attempting to understand strength andfailure of timber in terms of the strength andarrangement of the molecules, the fibrils, and thecells by thinking in terms of a theoretical strengthand attempting to identify the reasons why thetheory is never satisfied.
The second and more recent approach is much
Strength, toughness, failure and fracture morphology
503
FIGURE 46.8 Comparison of duration of load predictions among four damage accumulation models and onemodel based on strain energy (Fridley, Tang and Soltis (1992a) with permission of the American Society of Engi-neers).
more practical in concept since it considerstimber in its present state, ignoring its theoreticalstrength and its microstructure and stating thatits performance will be determined solely by thepresence of some defect, however small, whichwill initiate on stressing a small crack; the ulti-mate strength of the material will depend on thepropagation of this crack. Many of the theorieshave required considerable modification for theirapplication to the different fracture modes in ananisotropic material such as timber.
Both approaches are discussed below for themore important modes of stressing.
46.7.1 Classical approach
Tensile strength parallel to the grain
Over the years a number of models have beenemployed in an attempt to quantify the theo-retical tensile strength of timber. In these modelsit is assumed that the lignin and hemicellulosesmake no contribution to the strength of thetimber; in the light of recent investigations,however, this may be no longer valid for some ofthe hemicelluloses. One of the earliest attemptsmodelled timber as comprising a series of endlesschain molecules, and strengths of the order of8000N/mm2 were obtained. More recent model-ling has taken into account the finite length of thecellulose molecules and the presence of amor-phous regions. Calculations have shown that thestress to cause chain slippage is generally consid-erably greater than that to cause chain scissionirrespective of whether the latter is calculated onthe basis of potential energy function or bondenergies between the links in the chain; preferen-tial breakage of the cellulose chain is thought tooccur at the CBOBC linkage. These importantfindings have led to the derivation of minimumtensile stresses of the order of 1000–7000N/mm2
(Mark, 1967).The ultimate tensile strength of timber is of the
order of 100N/mm2, though this varies consider-ably between species. This figure corresponds to avalue between 0.1 and 0.015 of the theoreticalstrength of the cellulose fraction. Since this
accounts for only half the mass of the timber(Table 44.2) and since it is assumed, perhapsincorrectly, that the matrix does not contribute tothe strength, it can be said that the actualstrength of timber lies between 0.2 and 0.03 of itstheoretical strength.
In attempting to integrate these views of molec-ular strength with the overall concept of failure, itis necessary to examine strength at the next orderof magnitude, namely the individual cells. It ispossible to separate these by dissolution of thelignin–pectin complex cementing them together(Section 44.2 and Figures 44.7 and 44.8). Usingspecially developed techniques of mounting andstressing, it is possible to determine their tensilestrengths: much of this work has been done onsoftwood tracheids, and mean strengths of theorder of 500N/mm2 have been recorded by anumber of investigators. The strengths of the late-wood cells can be up to three times that of thecorresponding earlywood cells.
Individual tracheid strength is thereforeapproximately five times greater than that forsolid timber. Softwood timber also containsparenchyma cells which are found principally inthe rays, and lining the resin canals, and whichare inherently weak; many of the tracheids tendto be imperfectly aligned and there are numerousdiscontinuities along the cell; consequently it is tobe expected that the strength of timber is lowerthan that of the individual tracheids. Neverthe-less, the difference is certainly substantial and itseems doubtful if the features listed above canaccount for the total loss in strength, especiallywhen it is realised that the cells rupture on stress-ing and do not slip past one another.
When timber is stressed in tension along thegrain, failure occurs catastrophically with little orno plastic deformation (Figure 45.3) at strains ofabout 1 per cent. Visual examination of thesample usually reveals an interlocking type offracture which can be confirmed by opticalmicroscopy. However, as illustrated in Figure46.9, the degree of interlocking is considerablygreater in the latewood than in the earlywood.Whereas in the former, the fracture plane isessentially vertical, in the latter the fracture plane
Strength and failure in timber
504
follows a series of shallow zig-zags in a generaltransverse plane; it is now thought that thesethin-walled cells contribute very little to thetensile strength of the timber. Thus, failure in thestronger latewood region is by shear, while in the earlywood, though there is some evidence ofshear failure, most of the rupture appears to betranswall or brittle.
Examination of the fracture surfaces of thelatewood cells by electron microscopy revealsthat the plane of fracture occurs either within theS1 layer or, as is more common, between the S1
Strength, toughness, failure and fracture morphology
505
FIGURE 46.9 Tensile failure in spruce (Picea abies)showing mainly transverse cross-wall failure of the earlywood (left) and longitudinal intra-wall shearfailure of the latewood cells (right) (�110, polarisedlight) (© Building Research Establishment).
and S2 layers. Since shear strengths are lower thantensile strengths these observations are in accordwith comments made previously on the relativesuperiority of the tensile strengths of individualfibres compared with the tensile strength oftimber. By failing in shear this implies that theshear strength of the wall layers is lower than theshear strength of the lignin–pectin materialcementing together the individual cells.
Confirmation of these views is forthcomingfrom the work of Mark (1967) who has calcu-lated the theoretical strengths of the various cellwall layers and has shown that the direction andlevel of shear stress in the various wall layers wassuch as to initiate failure between the S1 and S2
layers. Mark’s treatise received a certain amountof criticism on the grounds that he treated onecell in isolation, opening it up longitudinally inhis model to treat it as a two-dimensional struc-ture; nevertheless, the work marked the beginningof a new phase of investigation into elasticity andfracture and the approach has been modified andsubsequently developed. The extension of thework has explained the initiation of failure at theS1–S2 boundary, or within the S1 layer, in terms ofeither buckling instability of the microfibrils, orthe formation of ruptures in the matrix or frame-work giving rise to a redistribution of stress.
Thus, both the microscopic observations andthe developed theories appear to agree thatfailure of timber under longitudinal tensile stress-ing is basically by shear unless density is lowwhen transwall failure occurs. However, undercertain conditions the pattern of tensile failuremay be abnormal. First, at temperatures in excessof 100°C, the lignin component is softened andits shear strength is reduced. Consequently, onstressing, failure will occur within the cementingmaterial rather than within the cell wall.
Second, transwall failure has been recorded inweathering studies where the mode of failurechanged from shear to brittle as degradation pro-gressed; this was interpreted as being caused by abreakdown of the lignin and degradation of thecellulose, both of which processes would bereflected in a marked reduction in density(Turkulin and Sell, 1997).
Finally, in timber that has been highly stressedin compression before being pulled in tension, itwill be found that tensile rupture will occur alongthe line of compression damage which, as will be explained below, runs transversely. Con-sequently, failure in tension is horizontal, givingrise to a brittle type of fracture (see Figure 47.1).
In the literature a wide range of tensile failurecriteria is recorded, the most commonly appliedbeing some critical strain parameter, an approachwhich is supported by a considerable volume ofevidence, though its lack of universal applicationhas been pointed out by several workers.
Compression strength parallel to thegrain
Few attempts have been made to derive a mathe-matical model for the compressive strengths oftimber. One of the few, and one of the most suc-cessful is that by Easterling et al. (1982); in modelling the axial and transverse compressivestrength of balsa, the authors found that theirtheory which related the axial strength linearly tothe ratio of the density of the wood to the densityof the dry cell wall material, and the transversestrength to the square of this ratio, was well sup-ported by experimental evidence. It also appearsthat their simple theory for balsa may be applic-able to timber of higher density.
Compression failure is a slow yielding processin which there is a progressive development ofstructural change. The initial stage of thissequence appears to occur at a stress of about 25per cent of the ultimate failing stress (Dinwoodie,1968), though Keith (1971) considers that theseearly stages do not develop until about 60 percent of the ultimate. There is certainly a verymarked increase in the amount of structuralchange above 60 per cent which is reflected bythe marked departure from linearity of thestress–strain diagram illustrated in Figure 45.3.The former author maintains that linearity here isan artefact resulting from insensitive testingequipment and that some plastic flow hasoccurred at levels well below 60 per cent of theultimate stress.
Strength and failure in timber
506
FIGURE 46.10 Formation of kinks, in the cell wallsof spruce timber (Picea abies) during longitudinalcompression stressing. The angle � lying between theplane of shear and the middle lamella varies systematic-ally between timbers and is influenced by temperature(�1600, polarised light) (© Building Research Estab-lishment).
Compression deformation assumes the form ofa small kink in the microfibrillar structure, andbecause of the presence of crystalline regions inthe cell wall, it is possible to observe this featureusing polarisation microscopy (Figure 46.10).The sequence of irreversible anatomical changesleading to failure originates in the tracheid orfibre wall at that point where the longitudinal cellis displaced vertically to accommodate the hori-zontally running ray. As stress and strain increasethese kinks become more prominent and increasenumerically, generally in a preferred lateral direc-tion, horizontally on the longitudinal–radialplane (Figure 46.11) and at an angle to the vertical axis of from 45° to 60° on the longitudi-nal–tangential plane. These lines of deformation,generally called a crease and comprising numer-ous kinks, continue to develop in width andlength; at failure, defined in terms of maximumstress, these creases can be observed by eye on theface of the block of timber (Dinwoodie, 1968). Atthis stage there is considerable buckling of the cellwall and delamination within it, usually betweenthe S1 and S2 layers. Models have been produced
to simulate buckling behaviour and calculatedcrease angles for instability agree well withobserved angles (Grossman and Wold, 1971).
At a lower order of magnification, Dinwoodie(1974) has shown that the angle at which thekink traverses the cell wall (Figure 46.10) variessystematically between earlywood and latewood,between different species, and with temperature.Almost 72 per cent of the variation in the kinkangle could be accounted for by a combination ofthe angle of the microfibrils in the S2 layer andthe ratio of cell wall stiffness in longitudinal andhorizontal planes.
Attempts have been made to relate the size andnumber of kinks to the amount of elastic strain or
Strength, toughness, failure and fracture morphology
507
the degree of viscous deformation; under con-ditions of prolonged loading, total strain and theratio of creep strain to elastic strain (relativecreep), appear to provide the most sensitive guideto the formation of cell wall deformation; thegross creases appear to be associated with strainsof 0.33 per cent (Keith, 1972).
The number and distribution of kinks aredependent on temperature and moisture content:increasing moisture content, though resulting in alower strain to failure, results in the productionof more kinks, although each is smaller in sizethan its ‘dry’ counterpart; these are to be found ina more even distribution than they are in drytimber. Increasing temperature results in a similarwider distribution of the kinks.
Static bending
In the bending mode timber is subjected tocompression stresses on the upper part of thebeam and tensile on the lower part. Since thestrength of clear timber in compression is onlyabout one-third that in tension, failure will occuron the compression side of the beam long beforeit will do so on the tension side. In knotty timber,however, the compressive strength is often equalto and can actually exceed the tensile strength. Asrecorded in the previous section, failure incompression is progressive and starts at low levelsof stressing: consequently, the first stages offailure in bending in clear straight-grained timberwill frequently be associated with compressionfailure and, as both the bending stress and con-sequently the degree of compression failureincrease, so the neutral axis will move progres-sively downwards from its original central posi-tion in the beam (assuming uniform cross-section)thereby allowing the increased compression loadto be carried over a greater cross-section. Frac-ture occurs when the stress on the tensile surfacereaches the ultimate strength in bending.
Toughness
Timber is a tough material, and in possessingmoderate to high stiffness and strength in
FIGURE 46.11 Failure under longitudinal compres-sion at the macroscopic level. On the longitudinalradial plane the crease (shear line) runs horizontally,while on the longitudinal tangential plane the crease isinclined at 65° to the vertical axis (© Building ResearchEstablishment).
addition to its toughness, it is favoured with aunique combination of mechanical propertiesemulated only by bone, which, like timber, is anatural composite.
Toughness is generally defined as the resistanceof a material to the propagation of cracks. In thecomparison of materials it is usual to expresstoughness in terms of work of fracture, which is ameasure of the energy necessary to propagate acrack, thereby producing new surfaces.
In timber the work of fracture, a measurementof the energy involved in the production of cracksat right angles to the grain, is about 104 J/m2; thisvalue is an order of magnitude less than that forductile metals, but is comparable with that forthe man-made composites. Now the energyrequired to break all the chemical bonds in aplane cross-section is of the order of 1–2J/m2;that is, four orders of magnitude lower than theexperimental values. Since pull-out of themicrofibrils does not appear to happen to anygreat extent, it is not possible to account for thehigh work of fracture in this way (Gordon andJeronimidis, 1974; Jeronimidis, 1980).
One of the earlier theories to account for thehigh toughness in timber was based on the workof Cook and Gordon (1964) who demonstratedthat toughness in fibre-reinforced materials isassociated with the arrest of cracks made possibleby the presence of numerous weak interfaces. Asthese interfaces open, so secondary cracks are ini-tiated at right angles to the primary, thereby dis-sipating the energy of the original crack. Thistheory is applicable to timber as Figure 46.12illustrates, but it is doubtful whether the total dis-crepancy in energy between experiment andtheory can be explained in this way.
Subsequent investigations have contributed to abetter understanding of toughness in timber(Gordon and Jeronimidis, 1974; Jeronimidis,1980). Prior to fracture it would appear that thecells separate in the fracture area; on furtherstressing these individual and unrestrained cellsbuckle inwards generally assuming a triangularshape. In this form they are capable of extendingup to 20 per cent before final rupture therebyabsorbing a large quantity of energy. Inward
Strength and failure in timber
508
FIGURE 46.12 Crack-stopping in a fractured rotorblade. The orientation of the secondary cracks corres-ponds to the microfibrillar orientation of the middlelayer of the secondary cell wall (�990, polarised light)(© Building Research Establishment).
buckling of helically wound cells under tensilestresses is possible only because the microfibrilsof the S2 layer are wound in a single direction.Observations and calculations on timber havebeen supported by glass-fibre models and it isconsidered that the high work of fracture can beaccounted for by this unusual mode of failure. Itappears that increased toughness is possiblyachieved at the expense of some stiffness, sinceincreased stiffness would have resulted from con-trawinding of the microfibrils in the S2.
So far, we have discussed toughness in terms of
only clear timber. Should knots or defects bepresent, timber will no longer be tough and thecomments made earlier as to viewpoint areparticularly relevant here. The material scientistsees timber as a tough material: the structuralengineer will view it as a brittle material becauseof its inherent defects and this theme is developedin Section 46.7.2.
Loss in toughness, however, can arise not onlyon account of the presence of defects and knots,but also through the effects of acid, prolongedelevated temperatures, fungal attack, or the pres-ence of compression damage with its associateddevelopment of cell wall deformations; theseresult from overstressing within the living tree, orin the handling or utilisation of timber after con-version (Section 47.2.4 and Figure 47.1) (Din-woodie, 1971; Wilkins and Ghali, 1987). Underthese abnormal conditions the timber is said to bebrash and failure occurs in a brittle mode.
Fatigue
Fatigue is usually defined as the progressivedamage and failure that occurs when a material issubjected to repeated loads of a magnitudesmaller than the static load to failure; it isperhaps the repetition of the loads that is thesignificant and distinguishing feature of fatigue.
In fatigue testing the load is applied generallyin the form of a sinusoidal or a square wave;minimum and maximum stress levels are usuallyheld constant throughout the test, though otherwave forms, and block or variable stress levelsmay be applied. In any fatigue test the three mostimportant criteria in determining the character ofthe wave form are:
1. the stress range, ��, where �� � �max � �min;2. the R-ratio, where R� �min/�max, which is the
position of minimum stress (�min) andmaximum stress (�max) relative to zero stress.This will determine whether or not reversedloading will occur; this is quantified in termsof the R-ratio, e.g. a wave form lying symmet-rically about zero load will result in reversedloading and have a R-ratio� �1;
Strength, toughness, failure and fracture morphology
509
3. the frequency of loading.
The usual method of presenting fatigue data isby way of the S–N curve where logN (the numberof cycles to failure) is plotted against the meanstress S; a linear regression is usually fitted.
Using test pieces of Sitka spruce, laminatedKhaya and compressed beech, Tsai and Ansell(1990) carried out fatigue tests under load controlin four-point bending. The tests were conducted inrepeated and reversed loading over a range of fiveR-ratios at three moisture contents (Figure 46.13).Fatigue life was found to be largely speciesindependent when normalised by static strength,but was reduced with increasing moisture contentand under reversed loading. The accumulation offatigue damage was followed microscopically intest pieces fatigued at R�0.1 and was found to beassociated with the formation of kinks in the cellwalls and compression creases in the wood.
In related work, Bonfield and Ansell (1991)investigated the axial fatigue in constant ampli-tude tests in tension, compression and shear inboth Khaya and Douglas fir using a wide range ofR-ratios and confirmed that reversed loading isthe most severe loading regime. Fatigue lives
FIGURE 46.13 The effect of moisture content onsliced Khaya laminates fatigued at R�0. Themaximum peak stresses are expressed as a percentageof static flexural (bending) strength (from Tsai andAnsell (1990) by permission of Kluwer Academic Pub-lishers).
measured in all-tensile tests (R�0.1) were con-siderably longer than those in all-compressiontests (R�10), a result which they related to thelower static strength in compression relative totension. S–N data at different R-ratios yielded aset of constant lifelines when alternating stresswas plotted against mean stress; these lifelinespossessed a point of inflection when loadingbecame all compressive.
More information on fatigue in wood is to befound in Dinwoodie (2000).
46.7.2 Engineering approach to strengthand fracture
The second approach to the concept of strengthand failure is a more practical one and is basedon the premise that all materials contain flaws orminute cracks, and that performance is deter-mined solely by the propagation of cracks arisingfrom these defects. The largest flaw will becomeself-propagating when the rate of release of strainenergy exceeds the rate of increase of surfaceenergy of the propagating crack.
The development of this concept has becomeknown as fracture mechanics: its application totimber did not take place until as late as 1961.Part of the reason for this is due to the modellingof wood as an orthotropic material, and con-sequently there are six values of ‘fracture tough-ness’ (a material parameter, also known as thecritical stress intensity factor, Kc) for each of thethree principal modes of crack propagation. Intimber, however, macroscopic crack extensionalmost always occurs parallel to the grain eventhough it is initiated in a different plane, therebygiving rise to a mixed mode type of failure.
The value of Kc (fracture toughness) is depend-ent not only on orientation as inferred above, butalso on the opening mode, orientation, wooddensity, moisture content, specimen thickness andcrack speed (see, e.g. Dinwoodie, 2000). Thus,the value of KIc (Kc in mode I) in the four weakparallel-to-the-grain systems is about one-tenththat in the two tough across-the-grain systems;KIc increases with increasing density and withincreasing specimen thickness.
Strength and failure in timber
510
Fracture mechanics has been applied to variousaspects of wood behaviour and failure, e.g. theeffect of knots, splits and joints; good agreementhas been found between predicted values usingfracture mechanics and actual strength values.Examples are to be found in Dinwoodie (2000).
46.8 Design stresses for timberTimber, like many other materials, is gradedaccording to its anticipated performance inservice; because of its inherent variability, distinctgrades of material must be recognised.
46.8.1 Grade stresses in UK prior to 1973
These were derived from the testing of small cleartest pieces (Section 46.2.1) as set out in Table46.6 and described fully in Dinwoodie (1981), aswell as in Desch and Dinwoodie (1996).
46.8.2 Grade stresses in UK from 1973 to1995
The realisation in the 1970s that the duration ofload values derived from the testing of small cleartest pieces were not appropriate for structuraltimber led to the derivation of grade stressesdirectly from actual structural-size timber.
This approach necessitated the introduction ofgrading of the timber by either visual or mechani-cal means about which more is said below. Inorder to define these grade classes in terms ofactual strength, tests on structural timber had tobe performed which led to the derivation of actualgrade stresses either in the form of strengthclasses, or as grade stresses for individual speciesand grades (see Table 46.6 and Desch and Din-woodie, 1996). The advantage of the strengthclass system over the listing of stresses for indi-vidual species and grades is that it allows suppliersto meet a structural timber specification by sup-plying any combination of species and grade listedin BS 5268: Part 2 as satisfying the specified class.
Structural design using these stresses is by wayof permissible stress design according to BS 5268:Part 2.
TA
BL
E4
6.6
Cha
nges
in t
he d
eriv
atio
n of
des
ign
stre
sses
ove
r th
e pe
riod
bef
ore
1973
to
abou
t 20
05
UK
EU
RO
PE
→19
7319
73–1
995
1995
–abo
ut 2
005
1996
→
Dat
a fr
om t
esti
ng s
mal
l cl
ear
test
pie
ces
Stru
ctur
al t
imbe
rSt
ruct
ural
tim
ber
Stru
ctur
al t
imbe
r↓
↓↓
↓M
ean
valu
e le
ss 2
.33s
Gra
ding
Tes
ts°
Gra
ding
Tes
ts°
Gra
ding
Tes
ts°
↓↓
↓B
S 58
20↓
↓E
N 4
08↓
↓E
N 4
08D
ivid
ed b
y sa
fety
V
isua
lM
achi
ne↓
Vis
ual
Mac
hine
EN
384
Vis
ual
Mac
hine
EN
384
fact
or*
to g
ive
BS
4978
BS
4978
↓B
S 49
78E
N 5
19↓
EN
518
EN
519
↓B
asic
str
ess
(198
8)(1
988)
↓(1
996)
↓↓
↓↓
↓B
S 57
56↓
BS
5756
↓↓
(198
0)↓
(199
7)↓
↓↓
↓↓
↓↓
↓↓
↓↓
↓M
ulti
plie
d by
G
rade
or S
tren
gth
clas
ses
↓↓
→
S
tren
gth
clas
ses
and
Stre
ngth
cla
sses
stre
ngth
rat
ioB
S 49
78 (
1988
)↓
↓C
hara
cter
isti
c va
lues
and
↓↓
↓↓
ΕΝ
338
Cha
ract
eris
tic
valu
esD
eriv
ed G
rade
str
esse
sSt
ress
es in
sta
ndar
d↓
↓Ε
Ν 3
38, Ε
Ν 1
912
in B
S 52
68 P
art
2B
S 52
68 P
art
2↓
fact
ored
↓↓
↓→
Gra
de s
tres
ses
↓↓
↓↓
Des
ign
code
Des
ign
code
Des
ign
code
Des
ign
code
BS
5268
Par
t 2
BS
5268
Par
t 2
BS
5268
Par
t 2
EN
V 1
995–
1–1
s�st
anda
rd d
evia
tion
.*
�sa
fety
fac
tor
(1.4
for
com
pres
sion
�to
the
gra
in; 2
.25
for
all o
ther
mod
es)
to c
over
eff
ects
of
spec
imen
siz
e an
d sh
ape,
rat
e of
load
ing
and
dura
tion
of
load
.°�
test
ing
used
onl
y to
der
ive
valu
es f
or in
clus
ion
in t
he s
tand
ards
.
46.8.3 Characteristic values for structuraltimber in Europe (including UK) from 1996
The formation of the EC as a free trade area andthe production of the Construction ProductsDirective (CPD) led automatically to the intro-duction and implementation of new Europeanstandards and the withdrawal of conflictingNational standards. Such an approach has muchmerit, but as far as the UK is concerned, it has ledto changes in both the derivation and use ofworking stresses. First, test results are nowexpressed in terms of a characteristic valueexpressed in terms of the lower 5th percentile, incontrast to the former use in the UK of a meanvalue and its standard deviation. Second, in thedesign of timber structures the new Eurocode 5(see below) is written in terms of limit statedesign in contrast to the former UK use ofpermissible stress design. Third, the number ofstrength classes in the new European system isgreater than in the UK system thereby giving riseto mismatching of certain timbers.
However, apart from these three majorchanges, the grading of timber into strengthclasses, and the original characterisation of theseby testing, adopts a similar procedure to thatused in the UK after 1973 (see Table 46.6)
Characteristic values for structural timber in arange of strength classes is to be found in EN338.
As noted above, the design of timber structuresmust be in accordance with Eurocode 5. Thecharacteristic values for both timber and wood-based panel products must be reduced accordingto the period of loading and service class (definedin terms of level of relative humidity). Values ofKmod (duration of load and service class) and Kdef
(creep and service class) are set out in Eurocode 5for each of the three service classes.
46.8.4 The UK position from 1995 toabout 2005
This is the transition period for all Europeancountries in which a choice is available betweentheir existing National grading and design stand-
ards and the new European system. In theoryboth systems exist side by side, but in practicemany countries, including the UK, have wiselytaken advantage of this period to modify theirexisting standards in order to bring them nearerto the European system and reduce the magni-tude of the change at the termination of thisperiod.
The UK has already adopted machine gradingof softwoods to EN 519 and has redrafted BS4978 (1996) so that it now relates only to thevisual grading of softwoods. Machine gradedtimber is assigned to strength classes contained inEN 338. From this standard, the characteristicvalue for a particular strength class and materialproperty are read off and in turn converted (fac-tored) to a grade stress for use in the continuingpermissible stress design system set out in BS5268: Part 2 (Anon, 1996). Visual graded timberis still assigned to grade stresses directly (seeTable 46.6).
46.8.5 Visual grading
Visual grading, as the title implies, is a visualassessment of the quality of a piece of structuraltimber: this is carried out against the permissibledefects limits given in standards BS 4978 (1996)and BS 5756 (1997) which conform to EN 518.However, visual grading is a laborious processsince all four faces of the timber should be exam-ined. Furthermore, it does not separate naturallyweak from naturally strong timber and hence ithas to be assumed that pieces of the same sizeand species containing identical defects have thesame strength: such an assumption is invalid andleads to a most conservative estimate of strength.
46.8.6 Machine grading
Many of the disadvantages of visual grading can beremoved by machine grading, a process which wasintroduced commercially in the early 1970s. Theprinciple underlying the process is the high correla-tion that has been found to exist between themoduli of elasticity and rupture; this was describedin Section 46.5.1. Grading machines based on this
Strength and failure in timber
512
relationship usually provide higher yields of thehigher grades than are achieved by visual grading.Since the above relationship varies among the dif-ferent species, it is necessary to set the gradingmachine for each species or species group.
The equation of the regression line is only oneof a number of inputs to the mathematical modelused to determine the machine settings for aparticular timber. These settings also depend on:
1. bandwidth (separation of the grade bound-aries on the x-axis) which, in turn, is relatedto the grade combination being graded;
2. the overall mean and standard deviation ofthe modulus of elasticity of the species;
3. the interaction of bandwidth and the MOEparameters; and
4. the cross-sectional size, and whether the timberpiece is sawn or planed all round (PAR).
The mathematical model can be designed toselect a number of grades or strength classes.
Having established the basic relationship foreach species, the grading machine can then be setup to grade the timber automatically. Dependingon the type of machines, as each length of timberpasses through, it is either placed under a con-stant load and deflection is measured, or sub-jected to a constant deflection and load ismeasured. A small computer quickly assesses theappropriate stiffness-indicating parameter every100 to 150mm along the length of the timber.The lowest reading of load, or the highest readingof deflection, is compared with the pre-set valuesstored in the computer and this determines theoverall grade of the piece of timber. Gradestamps are then printed on one face of the timbertowards one end of the piece.
Under the new European system this grademark will include the specification number usedin the grading (EN519):
• the species or species group;• the timber condition (green or dry);• the strength class;• the grader or company name;• the company registration number;• the certification body (logo).
46.9 References
46.9.1 Standards and specificationsASTM Standard D143–52 (1972) Standard methods of
testing small clear specimens of timber. AmericanSociety for Testing Materials.
BS 373 (1986) Methods of testing small clear speci-mens of timber. BSI, London
BS 4978 (1988) Softwood grades for structural use,BSI, London.
BS 4978 (1996) Visual grading of softwood. BSI, London.BS 5268 (1996) Structural use of timber. Part 2: Code
of practice for permissible stress design, materialsand workmanship. BSI, London.
BS 5756 (1996) Visual strength grading of hardwoods.BSI, London.
BS 5820 (1979) Methods of test for determination ofcertain physical and mechanical properties of timberin structural sizes. BSI, London.
EN 338 (1995) Structural timber – Strength classes.EN 384 (1995) Structural timber – Determination of
characteristic values on mechanical properties anddensity.
EN 408 (1995) Timber structures – Structural timberand glued laminated timber – Determination of somephysical and mechanical properties.
EN 518 (1995) Structural timber – Grading – Require-ments for visual strength grading standards.
EN 519 (1995) Structural timber – Grading – Require-ments for machine strength graded timber andgrading machines.
EN 1912 (1998) Structural timber – Strength classes –Assignment of visual grades and species.
ENV 1995–1–1 (1994) Eurocode 5: Design of timberstructures. Part 1.1. General rules and rules forbuildings.
46.9.2 LiteratureAnon (1996) Specifying structural timber, Digest 416.
Building Research Establishment, Watford, 8 pp.Barrett J.D. (1996) Duration of load – the past, present
and future. In Proc. of International Conference onWood Mechanics, Stuttgart, Germany (ed. S.Aicher), and published by the management commit-tee of EC COST 508 Action, 121–37.
Bonfield, P.W. and Ansell, M.P. (1991) Fatigue proper-ties of wood in tension, compression and shear. J.Mat. Sci. 26, 4765–73.
Cave, I.D. (1969) The longitudinal Young’s modulus ofPinus radiata. Wood Sci. Technol. 3, 40–8.
Cook, J. and Gordon, J.E. (1964) A mechanism for thecontrol of crack propagation in all brittle systems.Proc. Roy. Soc. London, A 282, 508.
References
513
Desch, H.E. and Dinwoodie, J.M. (1996) Timber –Structure, Properties, Conversion and Use, 7th edn,Macmillan, Basingstoke, 306 pp.
Dinwoodie, J.M. (1968) Failure in timber, Part I:Microscopic changes in cell wall structure associatedwith compression failure. J. Inst. Wood Sci. 21,37–53.
Dinwoodie, J.M. (1971) Brashness in timber and itssignificance. J. Inst. Wood Sci. 28, 3–11.
Dinwoodie, J.M. (1974) Failure in timber, Part II: Theangle of shear through the cell wall during longitudinalcompression stressing. Wood Sci. Technol. 8, 56–67.
Dinwoodie, J.M. (1981) Timber: its Nature andBehaviour, van Nostrand Reinhold, Wokingham,190 pp.
Dinwoodie, J.M. (2000) Timber: its Nature andBehaviour, E & FN Spon, London, 256 pp.
Easterling, K.E., Harrysson, R., Gibson, L.J. andAshby, M.F. (1982) The structure and mechanics ofbalsa wood. Proc. Roy. Soc, London 383, 31–41.
Fridley, K.J, Tang, R.C. and Soltis, L.A. (1991) Mois-ture effects on the load-duration behaviour oflumber. Part I: Effect of constant relative humidity.Wood and Fiber Sci. 23, 1, 114–27.
Fridley, K.C., Tang, R.C. and Soltis, L.A. (1992a)Load-duration effects in structural lumber: strainenergy approach. J. Struct. Eng., Structural Div.ASCE. 118, 9, 2351–69.
Fridley, K.J, Tang, R.C. and Soltis, L.A. (1992b) Mois-ture effects on the load-duration behaviour oflumber. Part II: Effect of cyclic relative humidity.Wood and Fiber Sci. 24, 1, 89–98.
Gerhards, C.C. (1982) Effect of moisture content andtemperature on the mechanical properties of wood:an analysis of immediate effects. Wood and Fiber14, 1, 4–26.
Gordon, J.E. and Jeronimidis, G. (1974) Work of frac-ture of natural cellulose. Nature (London) 252, 116.
Grossman, P.U.A. and Wold, M.B. (1971) Compres-sion fracture of wood parallel to the grain. WoodSci. Technol. 5, 147–56.
Hoffmeyer, P. (1990) Failure of Wood as Influenced byMoisture and Duration of Load. State University ofNew York, Syracuse, New York, USA.
Jeronimidis, G. (1980) The fracture behaviour of woodand the relations between toughness and morphol-ogy. Proc. Roy. Soc., London B208, 447–60.
Keith, C.T. (1971) The anatomy of compression failurein relation to creep-inducing stress. Wood Sci. 4, 2,71–82.
Keith, C.T. (1972) The mechanical behaviour of woodin longitudinal compression. Wood Sci. 4, 4,234–44.
Lavers, G.M. (1969) The strength properties of timber.Bulletin 50, For. Prod. Res. Lab. (2nd edn) HMSO(3rd edn revised by G. Moore, 1983).
Mark, R.E. (1967) Cell Wall Mechanics of Tracheids,Yale University Press, New Haven.
Morlier, P., Valentin, G. and Toratti, T. (1994) Reviewof the theories on long-term strength and time tofailure. Proc. Workshop on Service Life Assessmentof Wooden Structures, Espoo, Finland (ed. S.S.Gowda), the management committee of EC COST508 Action, 3–27.
Pearson, R.G. (1972) The effect of duration of load onthe bending strength of wood. Holzforschung 26, 4,153–8.
Tang, R.C. (1994) Overview of duration-of-loadresearch on lumber and wood composite panels inNorth America. Proc. Workshop on Service LifeAssessment of Wooden Structures, Espoo, Finland(ed. S.S. Gowda), published by the managementcommittee of EC COST 508 Action, 171–205.
Tsai, K.T. and Ansell, M.P. (1990) The fatigue proper-ties of wood in flexure. J. Mat. Sci. 25, 865–78.
Turkulin, J. and Sell, J. (1997) Structural and fracto-graphic study on weathered wood. Forschungs-undArbeitsbericht 115/36 Abteilung Holz, EMPA,Switzerland, 4 pp.
Wilkins, A.P. and Ghali, M. (1987) Relationshipbetween toughness, cell wall deformations anddensity in Eucaluptus pilularis. Wood Sci. Technol.21, 219–26.
Wood, L.W. (1951) Relation of strength of wood toduration of load. For. Prod. Laboratory (Madison)Report No. 1916.
Strength and failure in timber
514
515
Chapter forty-seven
Durability oftimber
47.1 Introduction47.2 Chemical, physical and mechanical agencies affecting
durability and causing degradation47.3 Natural durability and attack by fungi and insects47.4 Performance of timber in fire47.5 References
47.1 IntroductionDurability is a term which has different conceptsfor many people: it is defined here in the broadestpossible sense to embrace the resistance of timberto attack from a whole series of agencies whetherphysical, chemical or biological in origin.
By far the most important are the biologicalagencies, the fungi and the insects, both of whichcan cause tremendous havoc given the right con-ditions. In the absence of fire, fungal or insectattack, timber is really remarkably resistant andtimber structures will survive, indeed have sur-vived, incredibly long periods of time, especiallywhen it is appreciated that it is a natural organicmaterial. Examples of well-preserved timber itemsnow over 2000 years old are to be seen in theEgyptian tombs. Both fungal and insect attack aredescribed in Section 47.3 together with the import-ant aspect of the natural durability of the timber.
Another important aspect of durability oftimber is its reaction to fire and this is discussedin Section 47.4.
The effect of photochemical, chemical, thermaland mechanical actions are usually of secondaryimportance in determining durability; these willbe briefly considered first in Section 47.2.
47.2 Chemical, physical andmechanical agencies affectingdurability and causing degradation
47.2.1 Photochemical degradation
On exposure to sunlight the colouration of theheartwood of most timbers will lighten (e.g.mahogany, afrormosia, oak), though a fewtimbers will actually darken (e.g. Rhodesianteak). Indoors, the action of sunlight will be slowand the process will take several years: however,outdoors the change in colour is very rapid,taking place in a matter of months and is gener-ally regarded as an initial and very transient stagein the whole process of weathering.
In weathering the action not only of lightenergy (photochemical degradation), but also ofrain and wind results in a complex degradingmechanism which renders the timber silvery-greyin appearance: more important is the loss ofsurface integrity, a process which has been quan-tified in terms of the residual tensile strength ofthin strips of wood (Derbyshire and Miller, 1981;Derbyshire et al., 1995). The loss in integrityembraces the degradation of both the lignin, pri-marily by the action of ultraviolet light, and thecellulose by shortening of the chain length,mainly by the action of energy from the visiblepart of the spectrum. Degradation results inerosion of the cell wall and in particular the pitaperture and torus. Fractography using scanningelectron microscopy has revealed that the pro-gression of degradation involves initially the
Durability of timber
516
development of brittleness and the reduction instress transfer capabilities through lignin degrada-tion, followed by reductions in microfibrilstrength resulting from cellulose degradation(Derbyshire et al., 1995).
However, the same cell walls that are attackedact as an efficient filter for those cells below andthe rate of erosion from the combined effects ofUV, light and rain is very slow indeed; in theabsence of fungi and insects the rate of removalof the surface by weathering is of the order ofonly 1mm in every 20 years. Nevertheless,because of the continual threat of biologicalattack, it is unwise to leave most timbers com-pletely unprotected from the weather; it shouldbe appreciated that during weathering, theintegrity of the surface layers is markedlyreduced, thereby adversely affecting the perform-ance of an applied surface coating. In order toeffect good adhesion the weathered layers mustfirst be removed (see Section 48.4).
47.2.2 Chemical degradation
As a general rule, timber is highly resistant to alarge number of chemicals and its continued usefor various types of tanks and containers, even inthe face of competition from stainless steel, indic-ates that its resistance, certainly in terms of cost, ismost attractive. Timber is far superior to cast ironand ordinary steel in its resistance to mild acidsand for very many years timber was used as sepa-rators in lead-acid batteries. However, in its resis-tance to alkalis timber is inferior to iron and steel:dissolution of both the lignin and the hemicellu-loses occurs under the action of even mild alkalis.
Iron salts are frequently very acidic and in thepresence of moisture result in hydrolytic degrada-tion of the timber; the softening and darkish-bluediscolouration of timber in the vicinity of ironnails and bolts is due to this effect.
Timber used in boats is often subjected to theeffects of chemical decay associated with the cor-rosion of metallic fastenings, a condition fre-quently referred to as nail sickness. This isbasically an electrochemical effect, the rate ofactivity being controlled by oxygen availability.
47.2.3 Thermal degradation
Prolonged exposure of timber to elevated temper-atures results in a reduction in strength and avery marked loss in toughness (impact resistance).Thus, timber heated at 120°C for one monthloses 10 per cent of its strength; at 140°C, thesame loss in strength occurs after only one week(Shafizadeh and Chin, 1977). Tests on three soft-wood timbers subjected to daily cycles of 20°C to90°C for a period of three years resulted in areduction in toughness to only 44 per cent of itsvalue of samples exposed for only one day(Moore, 1984). It has been suggested thatdegrade can occur at temperatures even as low as65°C when exposed for many years.
Thermal degradation results in a characteristicbrowning of the timber with associated caramel-like odour, indicative of burnt sugar. Initially thisis the result of degrade of the hemicelluloses, butwith time the cellulose is also affected, with areduction in chain length through scission of the�1–4 linkage; commensurate degrade occurs inthe lignin, but usually at a slower rate (seeSection 47.4).
47.2.4 Mechanical degradation
The most common type of mechanical degrada-tion is that which occurs in timber when stressedunder load for long periods of time. The conceptsof duration of load (DOL) and creep areexplained fully in Sections 46.6.12 and 45.4.2respectively. Thus, it was illustrated in Figures46.7 and 46.8 how there is a loss in strength withtime under load, such that after being loaded for50 years the strength of timber is approximatelyonly 50 per cent. Similarly, there is a markedreduction in stiffness that manifests itself as anincrease in extension or deformation with timeunder load, as is illustrated in Figures 45.10 and45.13. The structural engineer, in designing histimber structure, has to take into account the losswith time of both strength and stiffness by apply-ing two time-modification factors.
A second and less common form of mechanicaldegradation is the induction of compression
Natural durability and attack by fungi and insects
517
failure within the cell walls of timber, which canarise either in the standing tree first, in the formof a natural compression failure due to highlocalised compressive stress, or second, as brittle-heart due to the occurrence of high growthstresses in the centre of the trunk, or underservice conditions where the timber is over-stressed in longitudinal compression with the pro-duction of kinks in the cell wall as described inSection 46.7.1. Loss in tensile strength due to theinduction of compression damage is about 10–15per cent, but the loss in toughness can be as highas 50 per cent. An example of failure in a scaffoldboard in bending resulting from the prior induc-tion of compression damage due to malpracticeon site is illustrated in Figure 47.1
47.3 Natural durability and attack byfungi and insects
47.3.1 Natural durability
Generally when durability of timber is discussed,reference is being made explicitly to the resistanceof the timber to both fungal and insect attack:this resistance is termed natural durability.
FIGURE 47.1 Kink bands and compression creasesin the tension face of a scaffold board which failed inbending on site. Note how the crack pathway has fol-lowed the line of the top compression crease which hadbeen induced some time previously (magnification �150) (© Building Research Establishment).
In the UK, timbers have been classified into fivedurability groups which are defined in terms ofthe performance of the heartwood when buried inthe ground. Examples of the more commontimbers are presented in Table 47.1. Such an arbi-trary classification is informative only in relativeterms, though these results on 50�50mmground stakes can be projected linearly forincreased thicknesses. Timber used externally, butnot in contact with the ground, will generallyhave a much longer life, though quantification ofthis is impossible.
Recalling that timber is an organic product, itis surprising at first to find that it can withstandattack from fungi and insects for very longperiods of time, certainly much greater than itsherbaceous counterparts. This resistance can beexplained in part by the basic constituents of thecell wall, and in part by the deposition of extrac-tives (Sections 44.1 and 44.3.1; Table 44.2).
The presence of lignin which surrounds andprotects the crystalline cellulose appears to offer aslight degree of resistance to fungal attack: cer-tainly the resistance of sapwood is higher thanthat of herbaceous plants. Fungal attack can com-mence only in the presence of moisture and thethreshold value of 20 per cent for timber is abouttwice as high as the corresponding value for non-lignified plants.
Timber has a low nitrogen content being of theorder of 0.03–0.1 per cent by mass and, since thiselement is a prerequisite for fungal growth, itspresence in only such a small quantity contributesto the natural resistance of timber.
The principal factor conferring resistance to bio-logical attack is undoubtedly the presence ofextractives in the heartwood. The far higher dura-bility of the heartwood of certain species com-pared with the sapwood is attributable primarilyto the presence in the former of toxic substances,many of which are phenolic in origin. Otherfactors such as a decreased moisture content,reduced rate of diffusion, density and deposition ofgums and resins also play a role in determining thehigher durability of the heartwood.
Considerable variation in durability can occurwithin the heartwood zone: in a number of
Durability of timber
518
TABLE 47.1 Durability classification (resistance of heartwood to fungi in ground contact) for the morecommon timbers*
Durability class Approximate life in contact Hardwoods Softwoodswith ground (years)
Perishable �5 AlderAsh, EuropeanBalsaBeech, EuropeanBirch, EuropeanHorse chestnutPoplar, blackSycamoreWillow
Nondurable 5–10 Afara Hemlock, westernElm, English Parana pineOak, American red Pine, Scots (redwood)Obeche Pine, yellowPoplar, grey PodoSeraya, white Spruce, European (whitewood)
Spruce, Sitka
Moderately durable 10–15 Avodire Douglas firKeruing LarchMahogany, African Pine, maritimeOak, TurkeySapeleSeraya, dark redWalnut, EuropeanWalnut, African
Durable 15–25 Agba Pine, pitchChestnut, sweet Western red cedarIdigbo YewMahogany, AmericanOak, EuropeanUtile
Very durable �25 AfrormosiaAfzeliaEkkiGreenheartIrokoJarrahKapurMakoreOpepePurpleheartTeak
*Note that the sapwood of all timber is perishable.
Natural durability and attack by fungi and insects
519
timbers the outer band of the heartwood has ahigher resistance than the inner region, owing, itis thought, to the progressive degradation of toxicsubstances by enzymatic or microbial action.
Durability of the heartwood varies consider-ably among the different species, being related tothe type and quantity of the extractives present:the heartwood of timbers devoid of extractiveshas a very low durability. Sapwood of all timbersis susceptible to attack owing not only to theabsence of extractives, but also the presence inthe ray cells of stored starch which constitutes aready source of food for the invading fungus.
The sapwood and heartwood of many speciesof timber can have its natural durability increasedby impregnation with chemicals; the preservativetreatment of timber is considered in Chapter 48.
47.3.2 Nature of fungal decay
In timber, some fungi, e.g. the moulds, arepresent only on the surface and, although theymay cause staining, they have no effect on the strength properties. A second group of fungi,the sapstain fungi, live on the sugars present in the ray cells and the presence of their hyphaein the sapwood imparts a distinctive colourationto that region of the timber when it is often referred to as ‘blue-stain’. One of the bestexamples of sapstain is that found in recentlyfelled Scots pine logs. In temperate countries thepresence of this type of fungus results in onlyinappreciable losses in bending strength, thoughseveral staining fungi in the tropical countriescause considerable reductions in strength.
By far the most important group of fungi arethose that cause decay of the timber by chemicaldecomposition; this is achieved by the digestingaction of enzymes secreted by the fungal hyphae.Two main groups of timber-rotting fungi can bedistinguished:
1. The brown rots, which consume the celluloseand hemicelluloses, but attack the lignin onlyslightly. During attack the wood usuallydarkens and in an advanced stage of attacktends to break up into cubes and crumbles
under pressure. One of the best known fungiof this group is Serpula lacrymans whichcauses dry rot. Contrary to what its name sug-gests, the fungus requires an adequate supplyof moisture for development.
2. The white rots, which attack all the chemicalconstituents of the cell wall. Although thetimber may darken initially, it becomes verymuch lighter than normal at advanced stagesof attack. Unlike attack from the brown rots,timber with white rot does not crumble underpressure, but separates into a fibrous mass.
In very general terms, the brown rots are usuallyto be found in constructional timbers, whereasthe white rots are frequently responsible for thedecay of exterior joinery.
Decay, of course, results in a loss of strength,but it is important to note that considerablestrength reductions may arise in the very earlystages of attack; toughness is particularly sensi-tive to the presence of fungal attack. Loss in massof the timber is also characteristic of attack anddecayed timber can lose up to 80 per cent of itsair-dry mass.
The principal types of fungal attack of wood inthe standing tree, of timber in felled logs, or oftimber in service are set out in Table 47.2. Moreinformation on fungal attack of timber is to befound in Desch and Dinwoodie (1996).
47.3.3 Nature of insect attack
Although all timbers are susceptible to attack byat least one species of insect, in practice only asmall proportion of the timber in service actuallybecomes infested. Some timbers are more suscep-tible to attack than others and generally theheartwood is much more resistant than thesapwood: nevertheless, the heartwood can beattacked by certain species, particularly whendecay is also present.
In certain insects the timber is consumed by theadult form, and the best known example of thismode of attack are the termites where the adult,but sexually immature, workers cause mostdamage. Few timbers are immune to attack by
Durability of timber
520
these voracious eaters and it is indeed fortunatethat these insects generally cannot survive thecooler weather of this country. They are to befound principally in the tropics, but certainspecies are present in the Mediterranean regionincluding southern France.
In this country insect attack is mainly by thegrub or larval stage of certain beetles. The adultbeetle lays its eggs on the surface of the timber,frequently in surface cracks, or in the cut ends ofcells; these eggs hatch to produce grubs whichtunnel their way into the timber, remaining therefor periods of up to three years or longer. Thesize and shape of the tunnel, the type of detritusleft behind in the tunnel (frass) and the exit holesmade by the emerging adults are all characteristicof the species of beetle.
The principal types of insect attack of wood inthe standing tree, of timber in the form of felledlogs, or of timber in service which may be freefrom decay or partially decayed are set out inTable 47.3. More information on insect attack oftimber is to be found in Desch and Dinwoodie(1996).
47.3.4 Marine borers
Timber used in salt water is subjected to attackby marine-boring animals such as the shipworm(Teredo sp.) and the gribble (Limnoria sp.).Marine borers are particularly active in tropicalwaters; nevertheless, around the coast of GreatBritain Limnoria is fairly active and Teredo,though spasmodic, has still to be considered apotential hazard. The degree of hazard will varyconsiderably with local conditions and there arerelatively few timbers which are recognised ashaving heartwood resistant under all conditions:the list of resistant timbers includes ekki, green-heart, okan, opepe and pyinkado.
47.4 Performance of timber in fireThe performance of materials in fire is an aspectof durability which has always attracted muchattention, not so much from the research scient-ist, but rather from the material user who has toconform with the legislation on safety and who isinfluenced by the weight of public opinion on the
TABLE 47.2 The principal types of fungal attack
Fungus Location of attack Effect on the timber
Tree Logs Timber Gross Micro
Brown rot ✓ 1 ✓ ✓ Darkening of timber with Attacks cellulose andcuboidal cracking. hemicellulose.
White rot ✓ 1 ✓ ✓ Bleaching of timber which Attacks cellulose,turns fibrous. hemicellulose and lignin.
Soft rot – – ✓ Superficial; small cross- Attacks cellulose of S2 layer.cracking; mostly occurringin ground contact.
Sapstain – ✓ ✓ Stains the sapwood of Stain due to colour of(Blue-stain) timber in depth. hyphae.Moulds – ✓ ✓ Superficial staining due to Live on cell contents; may
spores. increase permeability oftimber.
Bacteria ✓ ✓ ✓ Subtle changes in texture Increases permeability; getand colour. significant decay in ground
contact.
1Commonly as pocket rots.
Performance of timber in fire
521
use of only ‘safe’ materials. While various testshave been devised to assess the performance ofmaterials in fire, there is a fair degree of agree-ment in the unsatisfactory nature of many ofthese tests, and an awareness that certain mater-ials can perform better in practice than is indi-cated by these tests.
Thus, while no-one would doubt that timber isa combustible material showing up rather poorlyin both the ‘spread of flame’ and ‘heat release’tests, nevertheless in at least one aspect ofperformance, namely the maintenance of strengthwith increasing temperature and time, wood per-forms better than steel.
TABLE 47.3 The principal types of insect attack of timber
Type of insect Location of attack Comments
Tree Logs Timber in service
Sound Decayed
Pin-hole borers ✓ ✓ – – Produce galleries 1–2mm diameter which(Ambrosia beetle) are devoid of bore dust and are usually
darkly stained; attack is frequently present intropical hardwoods.
Forest longhorn ✓ (✓ ) – – Galleries oval in cross-section; no bore dustbut galleries may be plugged with coarsefibres; exit holes oval, 6–10mm diameter.
Wood wasp (✓ ) ✓ – – Galleries circular in cross-section andpacked with bore dust; attacks softwoods;exit holes circular, 4–7mm diameter.
Bark borer beetle (✓ ) ✓ (✓ ) – Requires presence of bark on timber;(Ernobius mollis) galleries empty and mainly in bark, but will
also penetrate sapwood.; exit holes circular,1–2mm diameter.
Powder-post – ✓ ✓ – Attack confined to sapwood of hardwoodsbeetle (Lyctus) having large-diameter vessels; bore dust
fine, talc-like; exit holes circular, 1–2mmdiameter.
Common – – ✓ – Attacks mainly the sapwood of bothfurniture beetle softwoods and European hardwoods; bore(Anobium) dust like lemon-shaped pellets; exit holes
circular, 1–2mm diameter.House longhorn – – ✓ – Attacks sapwood of softwoods mainly in thebeetle (Hylotropes) roof space of houses in certain parts of
Surrey; bore dust sausage-shaped; exit holesfew, oval, often ragged, 6–10mm diameter.
Death-watch – – (✓ ) ✓ Attacks both sap and heartwood of partially-beetle (Xestobium) decayed hardwoods, principally oak; bore
dust bun-shaped; exit holes circular, 3mmdiameter.
Weevils – – – ✓ Attacks decayed softwoods and hardwoods in(e.g. Euophryum) damp conditions; exit holes small, ragged
about 1mm diameter.Wharf borer – – – ✓ Attacks partially decayed timber to produce(Narcerdes) large galleries with 6mm diameter oval exit
hole.
✓ – main site of attack (✓ ) – much reduced incidence of attack.
Durability of timber
522
There is a critical surface temperature belowwhich timber will not ignite. As the surface tem-perature increases above 100°C, volatile gasesbegin to be emitted as thermal degradation slowlycommences; however, it is not until the tempera-ture is in excess of 250°C that there is a sufficientbuild-up in these gases to cause ignition of thetimber in the presence of a pilot flame. Where thisis absent, the surface temperature can rise toabout 500°C before the gases become self-igniting. Ignition, however, is related not only tothe absolute temperature, but also to the time ofexposure at that temperature, since ignition is pri-marily a function of heat flux.
Generally chemical bonds begin to break downat about 175°C and it is recognised that the firstconstituent of the timber to degrade is the lignin,and this continues slowly up to 500°C. The hemi-celluloses degrade much more quickly between200°C and 260°C as does the cellulose within thetemperature range 260°C–350°C. Degradation ofthe cellulose results in the production of the flam-mable volatile gases and its marked reduction in
degree of polymerisation (chain length) (Le Vanand Winandy, 1990).
The performance of timber at temperaturesabove ignition level is very similar to that ofcertain reinforced thermosetting resins whichhave been used as sacrificial protective coatingson space-return capsules. Both timber and theseablative polymers undergo thermal decomposi-tion with subsequent removal of mass, leavingbehind enough material to preserve structuralintegrity.
The onset of pyrolysis in timber is marked by adarkening of the timber and the commencementof emission of volatile gases; the reactionbecomes exothermic and the timber reverts to acarbonised char popularly known as charcoal(Figure 47.2). The volatiles, in moving to thesurface, cool the char and are subsequentlyejected into the boundary layer where they blockthe incoming convective heat: this most import-ant phenomenon is known as transpirationalcooling. High surface temperatures are reachedand some heat is rejected by thermal radiation:
FIGURE 47.2 Diagrammatic representation of the thermal decomposition of timber (© Building ResearchEstablishment).
References
523
the heat balance is indicated in Figure 47.2. Thesurface layers crack badly both along and acrossthe grain and surface material is continually, butslowly, being lost.
A quasi-steady state is reached, therefore, witha balance between the rate of loss of surface andthe rate of recession of the undamaged wood. Formost softwoods and medium-density hardwoodsthe rate at which the front recedes is about0.64mm/min: for high density hardwoods thevalue is about 0.5mm/min (Hall and Jackman,1975).
The formation of the char, therefore, protectsthe unburnt timber which may be only a few millimetres from the surface. Failure of the beamor strut will occur only when the cross-sectionalarea of the unburnt core becomes too small tosupport the load. By increasing the dimensions ofthe timber above those required for structuralconsideration, it is possible to guarantee struc-tural integrity in a fire for a given period of time.This is a much more desirable situation than thatpresented by steel, where total collapse of thebeam or strut occurs at some critical temperature.
47.4.1 Methods of assessing performance
At the present time in the UK (1999), four stan-dard tests are applicable to the evaluation of theperformance of timber in fire, though this situ-ation will change in the first few years of the newmillennium with the introduction of the newEuropean test methods to assess reaction to fire.These four tests assess non-combustibility,ignitability, fire propagation and spread of flame.
In three out of these four standard fire testscurrently carried out, timber and board productsdo not fair at all well. None of the four testsdemonstrates the predictability of the perform-ance of timber in fire, nor do they indicate theguaranteed structural integrity of the material fora calculable period of time. The performance oftimber in the widest sense is certainly superior tothat indicated by the present set of standard tests.
Early in the first decade of the new millennium,National standards on reaction to fire will bereplaced by new European standards which
generally measure different aspects of the behavi-our of building materials and products in firethan do the current British Standards. All con-struction products will be classified into one ofsix Euroclasses (A–F) according to their reaction-to-fire performance in fire tests. Two of thesetests will be used to classify the least combustiblematerials (Euroclasses A1 and A2); these two newtests are a furnace test for non-combustibility(prEN ISO 1182) and an oxygen bomb calorime-ter test (prEN ISO 1716).
At the lower end of the range of Euroclasses(classes E and F), construction products of appre-ciable combustibility will be assessed using asimple ignitability test (prEN ISO 11925–2).Products that fall into Classes A2, B1 C and D(and D will probably contain timber and wood-based panels) will be tested using the singleburning item test (SBI) except where the productsare used as floor coverings when the critical flux(radiant panel) test (prEN ISO 9239–1) will beused to determine performance in Euro classesB–E. For both floor and non-floor applications,generally only two of the above tests will berequired to characterise performance of any oneproduct. More information on the methods ofassessing reaction to fire is given in Dinwoodie(2000).
The above reaction-to-fire tests relate to theproduct. When that product is incorporated into abuilding element, the fire resistance of that elementwill be determined by a whole series of other tests;most of these are still at the drafting stage, but it isanticipated that there will be at least sixty Euro-pean standards (or part-standards) relating to fireresistance of parts of a building.
47.5 References
47.5.1 Standards and specificationsprEN ISO 1182. Reaction to fire tests for building
products – Non-combustibility test.prEN ISO 1716. Reaction to fire tests for building
products – Determination of the gross caloric value.prEN ISO 9239–1. Reaction to fire tests for floor cov-
erings – Part 1: determination of the burning behavi-our using a radiant heat source.
Durability of timber
524
prENISO 11925–2. Reaction to fire tests for buildingproducts – Part 2: ignitability when subjected todirect impingement of flame.
prENISO 13823: 2000 Reaction to fire tests for build-ing products – Building products excluding floorings– exposed to thermal attack by a single burning itemtest (SBI).
47.5.2 LiteratureDerbyshire, H. and Miller, E.R. (1981) The photo-
degradation of wood during solar radiation. Holzals Roh-und Werkstoff 39, 341–50.
Derbyshire, H., Miller, E.R., Sell, J. and Turkulin, H.(1995) Assessment of wood photodegradation bymicrotensile testing. Drvna Ind. 46, 3, 123–32.
Desch, H.E. and Dinwoodie, J.M. (1996). Timber-
Structure, Properties, Conversion and Use, 7th edn,Macmillan, Basingstoke, 306 pp.
Dinwoodie, J.M. (2000) Timber: its Nature andBehaviour, 2nd edn, E & FN Spon, London, 257 pp.
Hall, G.S. and Jackman, P.E. (1975) Performance oftimber in fire. Timber Trades J. 15 Nov. 1975,38–40.
Moore, G.L. (1984). The effect of long-term tempera-ture cycling on the strength of wood. J. Inst. WoodSci. 9, 6, 264–7.
Shafizadeh, F. and Chin, P.P.S. (1977) Thermal degra-dation of wood. In Wood Technology: ChemicalAspects (ed. I.S. Goldstein), ACS Symposium Series4, American Chemical Society, Washington D.C.,pp. 57–81.
Le Van, S.L. and Winandy, J.E. (1990) Effects of fireretardant treatments on wood strength: a review.Wood and Fiber Sci. 22, 1, 113–31.
525
Chapter forty-eight
Processing oftimber
48.1 Introduction48.2 Mechanical processing48.3 Chemical processing48.4 Finishes48.5 References
48.1 IntroductionAfter felling, the tree has to be processed in orderto render the timber suitable for man’s use. Suchprocessing may be basically mechanical or chemi-cal in nature or even a combination of both. Onthe one hand timber may be sawn or chipped,while on the other it can be treated with chemi-cals which markedly affect its structure and itsproperties. In some of these processing operationsthe timber has to be dried and this technique hasalready been discussed in Section 44.7.4 and willnot be referred to again in this chapter.
The many diverse mechanical and chemicalprocesses for timber have been described in greatdetail in previous publications and it is certainlynot the intention to repeat such description here:readers desirous of such information are referredto the excellent and authoritative texts listedunder Further Reading. In looking at processingin this chapter, the emphasis is placed on theproperties of the timber as they influence orrestrict the type of processing. For convenience,the processes are subdivided below into mechani-cal and chemical, but frequently their boundariesoverlap.
48.2 Mechanical processing
48.2.1 Solid timber
Sawing and planing
The basic requirement of these processes is quitesimply to produce as efficiently as possible timberof the required dimensions, having a quality ofsurface commensurate with the intended use.Such a requirement depends not only on the basicproperties of the timber, but also on the designand condition of the cutting tool; many of thevariables are inter-related and it is frequentlynecessary to compromise in the selection of pro-cessing variables.
In Section 44.5 the density of timber wasshown to vary by a factor of ten from about 120to 1200kg/m3. As density increases, so the timetaken for the cutting edge to become bluntdecreases: whereas it is possible to cut over10000 feet of Scots pine before it is necessary toresharpen, only one or two thousand feet of adense hardwood such as jarrah can be cut.Density will also have a marked effect on theamount of power consumed in processing. Whenall the other factors affecting power consumptionare held constant, this variable is highly corre-lated with timber density, as illustrated in Figure48.1.
Timber of high moisture content nevermachines as well as that at lower moisture levels.There is a tendency for the thin-walled cells to bedeformed rather than cut because of their
Processing of timber
526
increased elasticity in the wet condition. After thecutters have passed over, these deformed areasslowly assume their previous shape, resulting in anirregular appearance to the surface which is verynoticeable when the timber is dried and painted;this induced defect is known as raised grain.
The cost of timber processing is determinedprimarily by the cost of tool maintenance, whichin turn is related not only to properties of thetimber, but also to the type and design of the sawor planer blade. In addition to the effect of timberdensity on tool life, the presence in certaintimbers of gums and resins has an adverse effectbecause of the tendency for the gum to adhere tothe tool, thereby causing overheating; in sawblades this in turn leads to loss in tension result-ing in saw instability and a reduction in sawingaccuracy.
A certain number of tropical hardwood timberscontain mineral inclusions which develop duringthe growth of the tree. The most common is silicawhich is present usually in the form of smallgrains within the ray cells. The abrasive action ofthese inclusions is considerable and the life of theedge of the cutting tool is frequently reduced toalmost one-hundredth of that obtained when
FIGURE 48.1 Effect of timber density and feed-speed on the consumption of power using a circularsaw to cut along the grain (rip-sawing) (© BuildingResearch Establishment).
cutting timber of the same density, but free ofsilica. Timbers containing silica are frequentlyavoided unless they possess special features whichmore than offset the difficulties which result fromits presence.
Moisture content of the timber also plays asignificant role in determining the life of cuttingtools. As moisture content decreases, so there is amarked reduction in the time interval betweenresharpening both saw and planer blades. Thefibrous nature of tension wood (Section 44.4) willalso increase tool wear.
Service life will also depend on the type anddesign of the tool. Although considerably moreexpensive than steel, the use of tungsten-carbide-tipped saws and planer blades extends the life ofthe cutting edge especially where timbers areeither dense or abrasive. Increasing the number ofteeth on the saw or the number of planer bladeson the rotating stock will increase the quality ofthe surface, provided that the feedspeed issufficient to provide a minimum bite per revolu-tion; this ensures a cutting rather than a rubbingaction which would accelerate blunting of thetool edge.
One of the most important tool design vari-ables is the angle between the edge and the timbersurface. As discussed in Section 44.2, timber isseldom straight-grained, tending in most cases tobe in the form of a spiral of low pitch; occasion-ally, the grain is interlocked or wavy. Under thesecircumstances, there is a strong tendency forthose cells which are inclined towards the direc-tion of the rotating cutter to be pulled out ratherthan cut cleanly, a phenomenon known as pick-up or tearing. The occurrence of this defect canbe removed almost completely by reducing thecutting angle (rake angle) of the rotating blades,though this will result in increased power con-sumption.
The cost of processing, though determined pri-marily by tool life, will be influenced also by theamount of power consumed. In addition to theeffect of density of the timber previously dis-cussed, the amount of energy required willdepend on the feedspeed (Figure 48.1), tooldesign and, above all, on tool sharpness.
Mechanical processing
527
Steam bending
Steam bending of certain timbers is a long-established process which was used extensivelywhen it was fashionable to have furniture withrounded lines. The backs of chairs and woodenhat stands are two common examples from thepast, but the process is still employed at thepresent time, albeit on a much reduced volume.The handles for certain garden implements,walking sticks and a few sports goods are all pro-duced by steam bending.
The mechanics of bending involves a presteam-ing operation to soften the lignin, swell thetimber, and render the timber less stiff. With theends restrained, the timber is usually bent rounda former, and after bending the timber must beheld in the restrained mode until it dries out andthe bend is set. In broad terms the deformation isirreversible, but over a long period of time, espe-cially with marked alternations in humidity of theatmosphere, a certain degree of recovery willarise, especially where the curve is unrestrainedby some fixing. Although most timbers can bebent slightly, only certain species, principally thehardwood timbers of the temperate region, canbe bent to sharp radii without cracking. Whenthe timber is bent over a supporting, but remov-able strap, the limiting radius of curvature isreduced appreciably. Thus, it is possible to bend25mm thick ash to a radius of 64mm and walnutto a radius of only 25mm.
48.2.2 Board materials
The area of board materials is recognised as beingthe fastest growing area within the timber indus-try over the last two decades. Not only does thisrepresent a greater volume of construction(particularly in the domestic area) and of con-sumer goods (e.g. furniture), but it also reflects alarge degree of substitution of board materials forsolid timber.
Production of wood-based panels in Europe in1997 (the last year for which complete data are available) was 40.1�106m3, of which 72.5 percent was particleboard, 14 per cent MDF, 7.5
per cent plywood, 4.5 per cent fibreboard and 1.5per cent OSB. Consumption would be in excessof production by about 6�106 m3 (estimate),most of which can be attributed to large importsof plywood.
In the UK, the consumption of board materialin 1997 (provisional data) was 5.8�106 m3 ofwhich 54 per cent was chipboard and OSB com-bined, 31 per cent was plywood, 11 per cent wasMDF, 4 per cent was other fibreboards, and 0.3per cent was CBPB. About 70 per cent of the UKconsumption of chipboard and OSB combinedwas home produced.
As a material, timber has a number of deficien-cies:
• it possesses a high degree of variability;• it is strongly anisotropic in both strength and
moisture movement;• it is dimensionally unstable in the presence of
changing humidity;• it is available in only limited widths.
Such material deficiencies can be lowered appre-ciably by reducing the timber to small units andsubsequently reconstituting it, usually in the formof large, flat sheets, though moulded items arealso produced, e.g. trays, bowls, coffins, chairbacks. The degree to which these boards assumea higher dimensional stability and a lower level ofanisotropy than is the case with solid timber isdependent on the size and orientation of thecomponent pieces of timber and the method bywhich they are bonded together. There are aninfinite variety of board types though there areonly four principal ones – plywood, chipboard(particleboard), OSB and fibreboard.
In comparison with timber, board materialspossess a lower degree of variability, loweranisotropy, and higher dimensional stability: theyare also available in very large sizes. The reductionin variability is due quite simply to the randomrepositioning of variable components, the degreeof reduction increasing as the size of the com-ponents decrease. The individual board materialsare discussed below in separate sections.
Processing of timber
528
Plywood
Over one million cubic metres of plywood areconsumed annually in the UK. Most of this ismade from softwood and is imported from theUnited States, Canada and Finland. Less than 1per cent of plywood is home produced.
Plywoods made from temperate hardwoods areimported mainly from Germany (beech) andFinland (birch – or birch/spruce combination)while plywoods from tropical hardwoods comepredominately from South East Asia (mainlyIndonesia and Malaysia) and to a lesser, butincreasing, extent from South America and Africa.
Logs, the denser of which are softened byboiling in water, may be sliced into thin veneerfor surface decoration by repeated horizontal orvertical cuts, or, for plywood, peeled by rotationagainst a slowly advancing knife to give a contin-uous strip. After drying, sheets of veneer forplywood manufacture are coated with adhesiveand are laid up and then pressed with the graindirection at right angles in alternate layers.Plywood frequently contains an unequal numberof plies so that the system is balanced around thecentral veneer; some plywoods, however, containan even number of plies, but with the two centralplies having the same orientation thereby ensur-ing that the plywood is balanced on each side ofthe central glue line.
As the number of plies increase, so the degreeof anisotropy in both strength and movementdrops quickly from the value of 40:1 for timberin the solid state. With three-ply construction andusing veneers of equal thickness, the degree ofanisotropy is reduced to 5:1, while for nine-plythis drops to 1.5:1. However, cost increasesmarkedly with number of plies and for mostapplications a three-ply construction is regardedas a good compromise between isotropy and cost.
The common multilayered plywood is techni-cally known as a veneer plywood in contrast tothe range of core plywoods where the surfaceveneers overlay a core of blocks or strips ofwoods.
Plywood (veneer type) for use in constructionin Europe must comply with the requirements of
the new European specifications, the mostimportant requirement of these being that ofbond performance.
The mechanical and physical properties of theplywood, therefore, will depend not only on thetype of adhesive used, but also on the species oftimber selected. Both softwoods and hardwoodswithin a density range of 400–700kg/m3 are nor-mally utilised. Plywood for internal use is pro-duced from the non-durable species andurea–formaldehyde adhesive (UF), while plywoodfor external use is generally manufactured usingphenol–formaldehyde (PF) resins; however, withthe exception of marine grade plywood in theUK, durable timbers, or permeable non-durabletimbers which have been preservative treated, areseldom used.
It is not possible to talk about strength proper-ties of plywood in general terms since not onlyare there different strength properties in differentgrain directions, but that these are also affectedby configuration of the plywood in terms ofnumber, thickness, orientation and quality of theveneers and by the type of adhesive used. Thefactors which affect the strength of plywood arethe same as those set out in Chapter 3 for thestrength of timber, though the effects are notnecessarily the same. Thus the intrinsic factors,such as knots and density, play a less significantpart than they do in the case of timber, but theeffect of the extrinsic variables such as moisturecontent, temperature and time is very similar tothat for timber.
Plywood is the oldest of the timber sheetmaterials and for very many years has enjoyed ahigh reputation as a structural sheet material. Itsuse in the Mosquito aircraft and gliders in the1940s, and its subsequent performance for smallboat construction, for sheathing in timber-framehousing, and in the construction of web andhollow-box beams all bear testament to its suit-ability as a structural material.
When materials are compared in terms of theirspecific stiffness (stiffness per unit mass),plywood is stiffer than many other materials,including mild steel sheet; generally, plywoodalso has high specific strength. Another important
Mechanical processing
529
property of plywood is its resistance to splitting,which permits nailing and screwing relativelyclose to the edges of the boards; this is a reflec-tion of the removal of a line of cleavage along thegrain which is a drawback of solid timber. Impactresistance (toughness) of plywood is very highand tests have shown that to initiate failure aforce greater than the tensile strength of thetimber species is required.
Plywoods tend to fall into three distinct groups.The first comprises those which are capable ofbeing used structurally. Large quantities of soft-wood structural plywood are imported into theUK from North America, supplemented bysmaller volumes from Sweden and Finland; thelatter country also produces a birch/spruce struc-tural plywood.
The use of this group of structural plywoods isnow controlled in that they must first complywith the new European specification and second,the characteristic values for use in design musthave been derived from semi-sized test piecesaccording to the new European test methods.
The second group of plywoods comprises thosewhich are used for decorative purposes, while thethird group comprises those for general-purposeuse. The latter are usually of very varied perform-ance in terms of both bond quality and strengthand are frequently used indoors for infill panelsand certain types of furniture.
Chipboard (particleboard)
In the UK the boards made from wood chips andresin are known as chipboards; however in mostother countries the product is referred to as particleboard. The chipboard industry dates fromthe mid-1940s and originated with the purpose ofutilising waste timber. After a long, slow start,when the quality of the board left much to bedesired, the industry has grown tremendouslyover the last two decades, far exceeding the sup-plies of waste timber available and now relying toa very large measure on the use of small trees forits raw material. Such a marked expansion is duein no small part to the much tighter control inprocessing and the ability to produce boards with
a known and reproducible performance, fre-quently tailor-made for a specific end use. Over60 per cent of UK consumption is home-produced.
In the manufacture of chipboard the timber,which is principally softwood, is cut by a series ofrotating knives to produce thin chips which aredried and then sprayed with adhesive. Usually thechips are blown onto flat platens in such a waythat the smaller chips end up on the surfaces ofthe board and the coarse chips in the centre. Themat is usually first cut to length before passinginto a single or multi-daylight press where it isheld for 0.10–0.20 min per mm of board thick-ness at temperatures up to 200°C. The density ofboards produced range from 450 to 750kg/m3,depending on end-use classification, while theresin content varies from about 9–11 per cent onthe outer layers to 5–7 per cent in the centrelayer, averaging out for the board at about 7–8per cent on a dry mass basis. Over the last decademany of the new chipboard plants have installedlarge continuous presses; as the name implies, themat is fed in at one end to reappear at the otherdistant end as a fully-cured board. This type ofpress has the advantage of being quick to respondto production changes in board thickness, adhe-sive type or board density.
Instead of using the very long continuous press,chipboard can also be made continuously usingeither the Mendé or an extrusion process. Theformer is applicable only in the manufacture ofthin chipboard, i.e. 6mm or less, and the processis analogous to that of paper manufacture in thatthe board is sufficiently flexible to pass betweenand over large heated rollers. In the extrusionprocess, the chipboard mat is forced out througha heated die, but this results in the orientation ofthe chips at right angles to the plane of the boardwhich reduces both the strength and stiffness ofthe material; it is used primarily as a core in themanufacture of doors and composite panels.
The performance of chipboard, like that ofplywood, is very dependent on the type of adhe-sive used. Much of the chipboard (particleboard)produced in Europe is made using urea–formaldehyde (UF) which, because of its sensitivity
Processing of timber
530
to moisture, renders this type of chipboardunsuitable for use where there is a risk of thematerial becoming wet, or even being subjectedto marked alternations in relative humidity over along period of time. More expensive boardspossessing some resistance to the presence ofmoisture are manufactured using eithermelamine-fortified urea–formaldehyde (MUF), orphenol–formaldehyde (PF), or isocyanate (IS)adhesives: however, a true external-grade boardhas not yet been produced commercially.
Chipboard, like timber, is a viscoelastic mater-ial and an example of the deformation over anextended period of time has already been pre-sented (Figure 45.10). However, the rate of creepin particleboard is considerably higher than thatin timber though it is possible to reduce it byincreasing the amount of adhesive or by modify-ing the chemical composition of the adhesive.
Within the new framework of European speci-fications, six grades of particleboard are specified,of which four are rated as load bearing (i.e. theycan be used in structural design) and two arenon-load bearing. The characteristic values forthe load-bearing grades for use in structuraldesign are given in EN 12369–1.
Particleboards are also produced from a widevariety of plant material and synthetic resin ofwhich flaxboard and bagasse board are the bestknown examples.
Wet-process fibreboard
Although much smaller quantities of fibreboardare used in the UK than either chipboard orplywood, it is nevertheless a most importantpanel product, used extensively in the UK forinsulation and the linings of doors and backs offurniture, and in Scandinavia as a cladding androofing material.
The process of manufacture is quite differentfrom that of the other board materials in that thetimber is first reduced to chips which are thensteamed under very high pressure in order tosoften the lignin which is thermoplastic in behavi-our. The softened chips then pass to a defibratorwhich separates them into individual fibres, or
fibre bundles, without inducing too muchdamage.
The fibrous mass is usually mixed with hotwater and formed into a mat on a wire mesh; themat is then cut into lengths and, like particle-board, pressed in a multi-platen hot press at atemperature of from 180°C to 210°C; the boardproduced is smooth on only one side, the under-side bearing the imprint of the wire mesh. Bymodifying the pressure applied in the final press-ing, boards of a wide range of density are pro-duced ranging from softboard with a density lessthan 400kg/m3, to mediumboard with a densityrange of 400–900kg/m3, to hardboard with adensity exceeding 900kg/m3. Fibreboard, like theother board products, is moisture sensitive, but inthe case of hardboard, a certain degree of resis-tance can be obtained by the passage of thematerial through a hot oil bath thereby impartinga high degree of water repellency: this material isreferred to as tempered hardboard.
MDF (dry process fibreboard)
There has been a phenomenal increase in the pro-duction of MDF world-wide over the last twodecades. In the period from 1986 to 1997 Euro-pean production rose from 0.58�106 m3 to5.5�106 m3. Consumption of MDF in the UK in1997 was 0.6�106 m3 of which 0.4�106 m3 washome produced.
MDF is a dry process fibreboard. The fibrebundles are first dried to a low moisture contentprior to being sprayed with an adhesive andformed into a mat which is hot-pressed toproduce a board with two smooth faces similar tothe production of particleboard; both multi-day-light and continuous presses are employed.
Various adhesive systems are employed; wherethe board will be used in dry conditions, aurea–formaldehyde (UF) resin is employed, whilea board with improved resistance to moisture foruse in humid conditions is usually manufacturedusing a melamine-fortified urea–formaldehyderesin (MUF), though phenol–formaldehyde (PF)or isocyanate (IS) resins are sometimes used.
The European specification for MDF includes
Mechanical processing
531
both load-bearing and non-load-bearing gradesfor both dry and humid end uses. Characteristicvalues for structural use of the former are given inEN 12369–1; however, it should be noted that theuse of the load-bearing panel under humid con-ditions is restricted to short periods of loading.
A very large part of MDF production is takenup in the manufacture of furniture where non-load-bearing grades for dry use are appropriate.
OSB (oriented strand board)
Like MDF, OSB manufacture is fairly new and iscertainly a growth area with over 50 mills world-wide with a capacity in 1997 in excess of16�106 m3; European capacity in 1997 wasabout 1�106 m3; UK production from one mill inthe same year was about 240000m3.
Strands up to 75mm in length with amaximum width of half its length are generallysprayed with an adhesive at a rate correspondingto about 2–3 per cent of the dry mass of thestrands. It is possible to work with much lowerresin concentrations than with chipboard manu-facture due to the removal of dust and ‘fines’from the OSB line prior to resin application: in afew mills powdered resins are used, though mostmanufacturers use a liquid resin. In the majorityof mills a phenol–formaldehyde (PF) resin is used,but in one or two mills, a melamine-fortifiedurea–formaldehyde (MUF) or isocyanate (IS)resin is employed.
In the formation of the mat the strands arealigned either in each of three layers, or only inthe outer two layers of the board. The extent oforientation varies among manufacturers, withproperty level ratios in the machine to crossdirection of 1.25/1 to 2.5/1, thereby emulatingplywood. Indeed, the success of OSB has been asa cheaper replacement for plywood, but it mustbe appreciated that its strength and stiffness areconsiderably lower than those of high qualitystructural grade plywood, though only marginallylower than those of many of the current struc-tural softwood plywoods. It is widely used forsuspended flooring, sheathing in timber-frameconstruction and flat roof decking.
The European specification for OSB sets outthe requirements for four grades, three of whichare load bearing, covering both dry and humidapplications. Characteristic values for structuraluse of the load-bearing grades are given in EN12369–1.
CBPB (cement bonded particleboard)
This is very much a special end-use product manu-factured in relatively small quantities. It com-prises by mass 70–75 per cent of Portland cementand 25–30 per cent of wood chips similar tothose used in particleboard manufacture. Theboard is heavy with a density of about1200kg/m3, but it is very durable (due to its highpH of 11), is more dimensionally stable underchanging relative humidity (due to the highcement content), has very good performance inreaction to fire tests (again because of the highcement content) and has poor sound transmission(due to the high density). The board is thereforeused in high hazard situations with respect tomoisture, fire or sound.
Comparative performance of the wood-based boards
With such a diverse range of board types, eachmanufactured in several grades, it is exceedinglydifficult to select examples in order to make someform of comparative assessment.
In general terms, the strength properties ofgood quality structural softwood plywood arenot only considerably higher than all the otherboard materials, but they are usually similar to orslightly higher than that of softwood timber.Next to a good quality structural plywood instrength are the hardboards, followed by MDFand OSB. Chipboard (particleboard) is of lowerstrength, but still stronger than the mediumboards and CBPB. Table 48.1 provides the 5thpercentile strength values included in the ENproduct specifications, with the exception ofplywood where actual test data for Douglas firplywood have been used. In passing, it is interest-ing to note the reduction in anisotropy in bending
Processing of timber
532
strength from 4.5 for 3-ply construction to 1.8for 7-ply layup. Other structural softwood ply-woods can have strength values lower than thosefor Douglas fir, being similar to or only slightlyabove those of OSB of high quality. Actualstrength values of individual manufacturer’sproducts of non-plywood panels may be higherthan these minimum specification values. Itshould be realised that these specification valuesare only for the purpose of quality control andmust never be used in design calculations.
Comparison of the behaviour of these productsto the effect of 24 hours’ cold water soaking isalso included in Table 48.1. CBPB is far superiorto all other boards. Even higher swell values thanthose recorded in the table are to be found in15mm OSB/1 (general purpose board) of 25 percent and in 3.2mm HB.LA hardboard (load-bearing, dry) of 35 per cent. For those boardslisted in the specifications for use under humidconditions that are included in Table 48.1, thetable provides information on their moistureresistance in terms of their retention of internalbond strength following either the cyclic exposuretest (EN 321), the boil test (EN 1087–1) or both.
48.2.3 Laminated timber
The process of cutting timber up into strips andgluing them together again has three main attrac-tions. Defects in the original piece of timber suchas knots, splits, reaction wood, or sloping grainare redistributed randomly throughout the com-posite member, making it more uniform inquality in comparison to the original piece oftimber where the defects often result in stressraisers when load is applied. Consequently, thestrength and stiffness of the laminated productwill usually be higher than that of the timberfrom which it was made. The second attraction isthe ability to create curved beams or complexshapes, while the third is the ability to use shorterlengths of timber which can be end-jointed.
Glulam, the popular term for laminated timber,has been around for many years and is to be foundin the form of large curved beams in public build-ings and sports halls. In manufacture, strips of
timber about 20–30mm in depth are coated withadhesive on their faces and laid up parallel to oneanother in a jig, the whole assembly being clampeduntil the adhesive has set. Generally, cold settingadhesives are used because of the size of thesebeams; for dry end use a urea–formaldehyde (UF)resin is employed, while for humid conditions aresorcinol–formaldehyde (RF) resin is employed.The individual laminae are end jointed using eithera scarf (sloping) or finger (interlocked) joint. Struc-tural characteristic values for glulam are deter-mined by the strength class of the timber(s) fromwhich it is made, factored for the number andtype(s) of laminates used.
48.2.4 Mechanical pulping
The pulping industry is the single largest con-sumer of wood. In the UK the consumption ofpaper and board in 1997 was 12.2�106 tonnes,of which approximately half was home producedfrom home-grown softwood and hardwood, soft-wood residues, recycled fibre and importedwoodpulp.
Pulp may be produced by either mechanical orchemical processes and it is the intention to post-pone discussion on the latter until later in thischapter. In the original process for producingmechanical pulp, logs with a high moisturecontent are fed against a grinding wheel which iscontinuously sprayed with water in order to keepit cool and free it of the fibrous mass produced.The pulp so formed, known as stone ground-wood, is coarse in texture, comprising bundles ofcells rather than individual cells, and is mainlyused as newsprint. To avoid the necessity toadopt a costly bleaching process, only light-coloured timbers are accepted. Furthermore,because the power consumed on grinding is alinear function of the timber density, only thelow-density timbers with no or only small quanti-ties of resin are used.
Much of the mechanical pulp now used is pro-duced by disc-refining. Wood chips, softened inhot water, by steaming, or by chemical pretreat-ment, are fed into the centre of two high-speed,counter-rotating, ridged, metal plates; on passing
533
TA
BL
E4
8.1
Five
per
cent
ile s
tren
gth
and
stif
fnes
s an
d 95
per
cent
ile s
wel
l val
ues
for
tim
ber
and
boar
d m
ater
ials
: qua
lity
cont
rol v
alue
s in
the
EN
pro
duct
spe
cific
atio
ns
EN
310
EN
319
EN
317
Moi
stur
e re
sist
ance
Thi
ckne
ssD
ensi
tyB
endi
ng s
tren
gth
EIn
tern
al24
hr
Inte
rnal
bon
d(m
m)
(kg/
m3 )
(N/m
m2 )
(N/m
m2 )
bond
swel
ling
(N/m
m2 )
aft
er
Par
alle
l Per
pend
icul
arP
aral
lel P
erpe
ndic
ular
(N/m
m2 )
thic
knes
sC
yclic
Boi
l(%
)E
N 3
21E
N 1
087–
1
Solid
tim
ber
– D
ougl
as fi
rSm
all c
lear
tes
t pi
eces
2059
080
°2.
2°16
400°
1100
°–
––
–St
ruct
ural
tim
ber
100
580
22–
8110
––
––
–Pl
ywoo
d (E
N 6
36–3
)D
ougl
as fi
r (t
hree
-ply
)4.
852
0*51
*11
*84
62*
624*
––
––
Dou
glas
fir
(sev
en-p
ly)
1960
0*42
*23
*75
24*
2496
*–
––
–O
SB/3
(E
N 3
00)
1867
0*20
1035
0014
000.
3215
0.15
0.13
OSB
/418
670*
2815
4800
1900
0.45
120.
170.
15Pa
rtic
lebo
ard
(EN
312
)T
ype
P4 (
Loa
d-be
arin
g dr
y)15
720*
1521
500.
3515
––
Typ
e P7
(H
eavy
load
-bea
ring
1574
0*20
3100
0.70
80.
360.
23hu
mid
)C
BPB
(E
N 6
34)
1810
009
4000
0.50
1.5
0.3
–Fi
breb
oard
(E
N 6
22)
Part
2 h
ardb
oard
(lo
ad-
3.2
900*
3838
000.
8015
–0.
5be
arin
g hu
mid
) H
B.H
LA
1Pa
rt 3
med
ium
boar
d (l
oad-
1050
0*18
1800
0.10
15–
–be
arin
g dr
y) M
BH
.LA
1Pa
rt 5
MD
F (l
oad-
1279
0*32
2800
0.80
100.
250.
15be
arin
g hu
mid
) M
DF.
HL
S
*N
ot in
the
spe
cific
atio
ns.
°M
ean
valu
es.
Processing of timber
534
from the centre of the plates to the periphery, thechips are reduced to fine bundles of cells or evenindividual cells. This process is capable of accept-ing a wider range of timbers than the traditionalstone groundwood method.
48.3 Chemical processing
48.3.1 Treatability
The ease with which a timber can be impregnatedwith liquids, especially wood preservatives, isgenerally referred to as its treatability. Treatabil-ity is related directly to the permeability of timberwhich was discussed in some detail in Section44.8 where the pathways of flow were described;it will be recalled that permeability was shownto be a function not only of moisture contentand temperature, but also of grain direction,sapwood/heartwood, earlywood/latewood andspecies.
Longitudinal permeability is usually about104 � transverse permeability owing principally tothe orientation of the cells in the longitudinaldirection. Heartwood, owing to the deposition ofboth gums and encrusting materials, is generallymuch less permeable than the sapwood, whileearlywood of the sapwood in the dry conditionhas a much lower permeability than the sametissue in the green state due to aspiration of thebordered pits in the dry state.
Perhaps the greatest variability in ease ofimpregnation occurs between species. Within thesoftwoods this can be related to the number anddistribution of the bordered pits and to the effi-ciency of the residual flow paths which utiliseboth the latewood bordered pits and the semi-bordercd ray pits. Within the hardwoods variabil-ity in impregnation is related to the size anddistribution of the vessels and to the degree ofdissolution of the end walls of the vesselmembers.
Four arbitrary classes of treatability are recog-nised, different timbers being assigned to theseaccording to the depth and pattern of penetrationwhen treated with wood preservatives (see EN350–2). These classes are permeable, moderately
resistant, resistant and extremely resistant. Thisclassification is equally applicable to impregna-tion by flame retardants, or dimensional stabilis-ers for, although differences in viscosity willinfluence degree of penetration, the treatability ofthe timber species will remain in the same relativeorder.
Artificial preservatives
Except where the heartwood of a naturallydurable timber is being used, timber shouldalways be treated with a wood preservative ifthere is any significant risk that its moisturecontent will rise above 20 per cent during itsservice life. At and above this moisture content,wood-destroying fungi can attack. The relation-ship between service environment and risk ofattack by wood-destroying organisms is definedin EN 335–1 using the hazard classification ofbiological attack. Clearly, those timbers ofgreater permeability will take up preservativesmore easily and are to be preferred over thosethat are more resistant. It is normally not neces-sary to protect internal woodwork which shouldremain dry. However, where the risk of waterspillage, leakage from pipes or from the roof isseen as likely or significant, application of woodpreservatives may be deemed a sensible precau-tion.
A variety of methods for the application ofwood preservatives are available. Short-termdipping and surface treatments by brush or sprayare the least effective ways of applying a preserv-ative because of the small loading and poor pene-tration achieved. In these treatments only thesurface layers are penetrated and there is a risk ofsplits occurring during service which will exposeuntreated timber to the risk of attack by wood-destroying organisms. Such treatments are usuallyconfined to do-it-yourself treatments, or treat-ments carried out during remediation or mainte-nance of existing woodwork.
The most effective methods of timber impreg-nation are industrial methods in which changes inapplied pressure ensure controlled, more uniformpenetration and retention of preservative. The
Chemical processing
535
magnitude of the pressure difference depends onthe type of preservative being used. Essentially,the timber to be treated is sealed in a pressurevessel and a vacuum drawn. While undervacuum, the vessel is filled with the preservativeand then returned to atmospheric pressure duringwhich some preservative enters the wood. At thispoint, an over-pressure of between zero and 13bar is applied, depending mainly on the preserva-tive being used, but also on the treatability of thetimber. This can be held for between severalminutes and many hours after which the vessel isdrained of preservative. A final vacuum is oftenapplied to recover some of the preservative and toensure the treated timber is free of excess fluid.Details of the choice, use and application ofwood preservatives can be found in BS 1282.
There are three main types of preservative ingeneral use. The first group are the tar oils ofwhich coal tar creosote is the most important. Itsefficacy as a preservative lies not only in itsnatural toxicity, but also in its water repellencyproperties. It has a very distinctive and heavyodour and treated timber cannot be paintedunless first coated with a metallic primer.
The second group are the water-borne preserv-atives which are suitable for both indoor andoutdoor uses. The most common formulationsare those containing copper, chromium andarsenic compounds, but combinations ofcopper/chromium and copper/chromium/boronare also used. All of these preservatives areusually applied by a vacuum high-pressure treat-ment; the chemicals react once in the wood andbecome fixed, i.e. they are not leached out inservice.
Inorganic boron compounds are also used aswater-borne preservatives, but they do notbecome fixed within the wood and therefore canbe leached from the wood during service. Theiruse is therefore confined to environments whereleaching cannot take place.
The third group are the solvent-type preserva-tives, which tend to be more expensive than thoseof the first two groups, but they have the advant-age that machined timber can be treated withoutthe grain being raised, as would be the case with
aqueous solutions. The formulations of thesolvent type are based on a variety of compoundsincluding pentachlorophenol, tri-n-butyltin oxide,and copper and zinc naphthanates. Morerecently, copper and zinc versatate, zinc ocoate,and the preparations known as acypetacs zincand acypetacs copper have been introduced.Some organic solvent preservatives include insec-ticides and water repellents. BS 5707 lists thestandardised formulations. These preservativesfind uses in the DIY and industrial sectors. Indus-trial treatment processes include double vacuumand immersion techniques, while DIY and on-sitetreatment includes dipping and brushing.
In looking at the application of these three dif-ferent types of preservatives, creosote and CCA(copper-chrome-arsenic) are able to protecttimber in high hazard situations, i.e. groundcontact, while organic solvent preservatives areused in timber out of ground contact and prefer-ably protected with a paint film.
Although it is not an impregnation process asdefined above, it is convenient to examine herethe diffusion process of preservation. The timbermust be in the green state and the preservativemust be water soluble. Timber is immersed for ashort period in a concentrated (sometimes hot)solution of a boron compound, usually disodiumoctoborate tetrahydrate, and then close-stackedundercover for several weeks to allow the preser-vative to diffuse into the timber.
Gaseous diffusion has been examined as amethod of preservative treating wood-basedpanels as well as timber. Treatment by trimethylborate vapour in a sealed, evacuated chamber ofmaterial at a low moisture content results in thereaction of the gaseous trimethyl borane with thewater in the wood to produce the biocide boricacid which is then deposited where the reactionoccurs (Turner and Murphy, 1998).
For many years there was some difference ofopinion as to whether these preservatives merelylined the walls of the cell cavity or actuallyentered the cell wall. However, it has been welldemonstrated by electron microanalysis thatwhereas creosote only coats the cell walls, thewater-borne preservatives do impregnate the cell
Processing of timber
536
wall. Although there is some evidence to showthat PCP (pentachlorophenol) can penetrate thecell wall, it is doubtful whether solvent-typepreservatives in general penetrate the wall due totheir large molecular size and by being carried ina non-polar solvent.
There is considerable variation in preservativedistribution in treated dry timber; in the soft-woods, only the latewood tends to be treated dueto aspiration of the earlywood bordered pits asdescribed in Section 44.8.1. In the hardwoods,treatment is usually restricted to the vessels andtissue in close proximity to the vessels, again asdescribed in Section 44.8.1.
In those timbers which can be impregnated, itis likely that the durability of the sapwood afterpressure impregnation will be greater than thenatural durability of the heartwood, and it is notunknown to find the heartwood of telegraph andtransmission poles to be decayed while thetreated sapwood is perfectly sound.
Mention has been made already of the diffi-culty of painting timber which has been treatedwith creosote. This disadvantage is not commonto the other preservatives and not only is it pos-sible to paint the treated timber, but it is also pos-sible to glue together treated components.
Flame retardants
Flame-retardant chemicals may be applied assurface coatings or by vacuum-pressure impreg-nation, thereby rendering the timber less easilyignitable and reducing the rate of flame spread.Intumescent coatings will be discussed later andthis section is devoted to the application of fireretardants by impregnation.
The salts most commonly employed in the UKfor the vacuum pressure impregnation process aremonoammonium phosphate, diammonium phos-phate, ammonium sulphate, boric acid andborax. These chemicals vary considerably in solu-bility, hygroscopicity and effectiveness againstfire. Most proprietary flame retardants are mix-tures of such chemicals formulated to give thebest performance at reasonable cost. Since thesechemicals are applied in an aqueous solution, it
means that a combined water-borne preservativeand fire-retardant solution can be used which hasdistinct economic considerations. Quite fre-quently, corrosion inhibitors are incorporatedwhere the timber is to be joined by metal connec-tors. Treatment of the timber involves high pres-sure/vacuum processes; as aqueous solutions areinvolved, redrying of the timber is required.
Considerable caution has to be exercised indetermining the level of heating to be used indrying the timber following impregnation. Theammonium phosphates and sulphate tend tobreak down on heating, giving off ammonia andleaving an acidic residue which can result indegradation of the wood substance. Thus, it hasbeen found that drying at 65°C following impreg-nation by solutions of these salts results in a lossof bending strength of from 10 per cent to 30 percent. Drying at 90°C, which is adopted in certainkiln schedules, results in a loss of 50 per cent ofthe strength and even higher losses are recordedfor the impact resistance or toughness of thetimber. It is essential, therefore, to dry the timberat as low a temperature as possible and also toensure that the timber in service is never sub-jected to elevated temperatures which would initi-ate or continue the process of acidic degradation.Most certainly, timber which has to withstandsuddenly applied loads should not be treated withthis type of fire retardant, and care must also beexercised in the selection of glues for construc-tion. The best overall performance from timbertreated with these flame retardants is obtainedwhen the component is installed and maintainedunder cool and dry conditions.
Conscious of the limitations of flame retardantsbased on ammonium salts, a number of companieshave developed effective retardants of very differentchemical composition; these have much reduceddegradation of the timber, or do not degrade thetimber, and are usually non-corrosive to metalfixings. However, they are considerably moreexpensive than those based on ammonium salts.
Dimensional stabilisers
In Section 45.2, timber, because of its hygro-
Chemical processing
537
scopic nature, was shown to change in dimen-sions as its moisture content varied in order tocome into equilibrium with the vapour pressureof the atmosphere. Because of the compositenature of timber, such movement will differ inextent in the three principal axes.
Movement is the result of water adsorption ordesorption by the hydroxyl groups present in allthe matrix constituents. Thus it should be possibleto reduce movement (i.e. increase the dimensionalstability) by eliminating or at least reducing theaccessibility of these groups to water. This can beachieved by either chemical changes or by theintroduction of physical bulking agents.
Dimensional stability is imparted to wood byswelling of the substrate due to chemical modifi-cation, since the bonded groups occupy volumewithin the cell wall. At high levels of modifica-tion, wood is swollen to near its green volume,and anti-shrink efficiencies close to 100 per centare achieved. After extended reaction, swelling inexcess of the green volume can occur, which isaccompanied by cell wall splitting.
Various attempts have been made to substitutethe hydroxyl groups chemically by less polargroups and the most successful has been by acety-lation (Rowell, 1984). In this process acetic anhy-dride is used as a source of acetyl groups. A verymarked improvement in dimensional stability isachieved with only a marginal loss in strength.Using carboxylic acid anhydrides of varying chainlength, Hill and Jones (1996) obtained gooddimensional stabilisation which they attributedsolely to the bulking effect. Other reagents withpotential for chemical modification include iso-cyanates and epoxides. Readers desirous of moreinformation on chemical modification arereferred to the review by Rowell (1984) and textedited by Hon (1996).
Good stabilisation can also be achieved byreacting the wood with formaldehyde which thenforms methylene bridges between adjacenthydroxyl groups. However, the acid catalystnecessary for the process causes acidic degrada-tion of the timber.
In contrast to the above means of chemicalmodification, a variety of chemicals have been used
to physically stabilise the cell wall; these impreg-nants act as bulking agents and hold the timber in aswollen condition even after water is removed, thusminimising dimensional movement.
Starting in the mid-1940s and continuing on amodest scale to the present time, some solidtimber, but more usually wood veneers, areimpregnated with solutions of phenol–formalde-hyde. The veneers are stacked, heated and com-pressed to form a high-density material with gooddimensional stability which still finds wide usageas a heavy-duty insulant in the electrical distribu-tion industry.
Considerable success has also been achievedusing polyethylene–glycol (PEG), a wax-like solidwhich is soluble in water. Under controlled con-ditions, it is possible to replace all the water intimber with PEG by a diffusion process, therebymaintaining it in a swollen condition. The tech-nique has found application, among other things,in the preservation of waterlogged objects ofarchaeological interest, the best example of whichis the Swedish wooden warship Wasa, which wasraised from the depths of Stockholm harbour in1961 having foundered in 1628. From 1961 thetimber was sprayed continuously for over adecade with an aqueous solution of PEG whichdiffused into the wet timber gradually replacingthe bound water in the cell wall without causingany dimensional changes.
PEG may also be applied to dry timber by stan-dard vacuum impregnation using solutionstrengths of from 5 to 30 per cent. Frequently,preservative and/or fire-retardant chemicals arealso incorporated in the impregnating solution. Itwill be noted from Figure 48.2 that followingimpregnation, the amount of swelling has beenreduced to one-third that of the untreated timber.
Developments in the production and use ofwater-repellent preservatives based on resins dis-solved in low-viscosity organic solvents haveresulted in the ability to confer on timber a low,but nonetheless important level of dimensionalstability. Their application is of considerableproven practical significance in the protection ofjoinery out-of-doors and these are discussedfurther in Section 48.4.
Processing of timber
538
48.3.2 Chemical pulping
The magnitude of the pulping industry hasalready been discussed, as has the production ofmechanical pulp. Where paper of a higher qualitythan newsprint or corrugated paper is required, apulp must be produced consisting of individualcells rather than fibre bundles. To obtain this typeof pulp the middle lamella has to be removed andthis can be achieved only by chemical means.
There are a number of chemical processeswhich are described in detail in the literature. Allare concerned with the removal of lignin, whichis the principal constituent of the middle lamella.However, during the pulping process lignin willalso be removed from within the cell wall as wellas from between the cells: this is both acceptableand desirous since lignin imparts a greyishcolouration to the pulp which is unacceptable forthe production of white paper.
However, it is not possible to remove all thelignin without also dissolving most of the hemi-celluloses which not only add to the mass of pulpproduced, but also impart a measure of adhesionbetween the fibres. Thus, a compromise has to bereached in determining how far to progress withthe chemical reaction and the decision is depend-
ent on the requirements of the end product. Fre-quently, though not always, the initial pulpingprocess is terminated when a quarter to a half ofthe lignin still remains and this is then removed ina subsequent chemical operation known asbleaching, which, though expensive, has relativelylittle effect on the hemicelluloses. The yield ofchemical pulp will vary considerably dependingon the conditions employed, but it will usually bewithin the range of 40–50 per cent of the drymass of the original timber.
The yield of pulp can be increased to 55–80 percent by semi-chemical pulping. Only part of thelignin is removed in an initial chemical treatmentwhich is designed to soften the wood chips: sub-sequent mechanical treatment separates the fibreswithout undue damage. These high-yield pulpsusually find their way into card and board-linerwhich are extensively used for packaging whereultimate whiteness is not a prerequisite.
48.3.3 Other chemical processes
Brief mention must be made of the destructivedistillation of timber, a process which is carriedout either for the production of charcoal alone orfor the additional recovery of the volatile by-
FIGURE 48.2 The comparative rates of swelling in water of untreated pine timber and timber impregnatedwith a 50 per cent (by mass) solution of PEG (polyethylene–glycol): this is equivalent to 22 per cent loading on adry wood basis (adapted from R.E. Morén (1964) by permission).
Finishes
539
products such as methanol, acetic acid, acetoneand wood-tar. The timber is heated initially to250°C, after which the process is exothermic: dis-tillation must be carried out either in the com-plete absence of air, or with controlled, smallamounts of air.
Timber can be softened in the presence ofammonia vapour as a result of plasticisation ofthe lignin. The timber can then be bent ormoulded using this process, but, because of theharmful effects of the vapour, the process hasnever been adopted commercially.
48.4 FinishesFinishes have a combined decorative and protec-tive function. Indoors they are employed primar-ily for aesthetic reasons, though their role inresisting soiling and abrasion is also important;outdoors, however, their protective function isvital. In Section 47.2.1, the natural weatheringprocess of timber was described in terms of theattack of the cell wall constituents by ultravioletlight and the subsequent removal of breakdownproducts by rain; the application of finishes is toslow down this weathering process to an accept-able level, the degree of success varying consider-ably among the wide range of finishescommercially available.
In Sections 44.1–44.3 the complex chemicaland morphological structure of timber wasdescribed, while in Section 44.7 the hygroscopicnature of this fibre composite and its significancein determining the movement of timber was dis-cussed. The combined effects of structure andmoisture movement have a most profound effecton the performance of coatings. For example, inthe softwoods the presence of distinct bands ofearly and latewood with their differential degreeof permeability results not only in a difference insheen or reflectance of the coating between thesezones, but also in marked differences in adhesion;in Douglas fir, where the latewood is most con-spicuous, flaking of paint from the latewood is acommon occurrence. In addition, the radialmovement of the latewood has been shown to beas high as six times that of the earlywood and
consequently the ingress of water to the surfacelayers results in differential movement and con-siderable stressing of the coatings. In those hard-woods characterised by the presence of largevessels, the coating tends to sag across the vesseland it is therefore essential to apply a paste fillerto the surface prior to painting; even with this,the life of a paint film on a timber such as oak(see Figure 44.6) is very short. For this reason,the use of exterior wood stains is common, as thistype of finish tends not to exhibit the same degreeof flaking. The presence of extractives in certaintimbers (see Section 44.3.1 and Table 44.2)results in the inhibition in drying of most finishes;with iroko and Rhodesian teak, many types offinish may never dry.
Contrary to general belief, deep penetration ofthe timber is not necessary for good adhesion, butit is absolutely essential that the weathered cellson the surface are removed prior to repainting.Good adhesion appears to be achieved by molecu-lar attraction rather than by mechanical keyinginto the cell structure.
Although aesthetically most pleasing, fullyexposed varnish, irrespective of chemical compo-sition, has a life of only a very few years, princip-ally because of the tendency of most types tobecome brittle on exposure, thereby cracking anddisintegrating because of the stresses imposed bythe movement of the timber under changes inmoisture content. Ultraviolet light can readilypass through the majority of varnish films,degrading the timber at the interface and causingadhesion failure of the coating.
A second type of natural finish which over-comes some of the drawbacks of clear varnish isthe water-repellent preservative stain or exteriorwood stain. There are many types available, butall consist of resin solutions of low viscosity andlow solids content: these solutions are readilyabsorbed into the surface layers of the timber.Their protective action is due in part to the effec-tiveness of water-repellent resins in preventingwater ingress, and in part from the presence offinely dispersed pigments which protect againstphotochemical attack. The higher the concentra-tion of pigments the greater the protection, but
Processing of timber
540
this is achieved at the expense of loss in trans-parency of the finish. Easy to apply and maintainthese thin films, however, offer little resistance tothe transmission of water vapour into and out ofthe timber. Compared with a paint or varnish,the water-repellent finish will allow timber to wetup and dry out at a much faster rate, therebyeliminating problems of water accumulationwhich can occur behind impermeable paintsystems; the presence of a preservative constituentreduces the possibility of fungal developmentduring periods of high moisture uptake. The filmsdo require, however, more frequent maintenance,but nevertheless have become well established forthe treatment of cladding and hardwood joinery.
By far the most widely-used finish, especiallyfor external softwood joinery, is the traditionalopaque alkyd gloss or flat paint system embracingappropriate undercoats and primers; a three- orfour-coat system is usually recommended. Mul-tiple coats of oil-based paint are effective barriersto the movement of liquid and vapour water;however, breaks in the continuity of the film afterrelatively short exposure constitute a ready meansof entry of moisture after which the surrounding,intact film will act as a barrier to moisture escape,thereby increasing the likelihood of fungal attack.The effectiveness of the paint system is deter-mined to a considerable extent by the quality ofthe primer. Quite frequently window and doorjoinery with only a priming coat is left exposedon building sites for long periods of time. Mostprimers are permeable to water, are low in elas-ticity and rapidly disintegrate owing to stressesset up in the wet timber; it is therefore essentialthat only a high quality of primer is used. Emul-sion-based primer/undercoats applied in two con-secutive coats are more flexible and potentiallymore durable than the traditional resin-basedprimers and undercoats.
A new range of exterior quality paints has beenproduced in the last decade; these are eithersolvent-borne or water-borne formulations. Someof the formulations have a higher level of moisturepermeability than conventional paint systems andhave been described as microporous. These areclaimed to resist the passage of liquid water, but
to allow the passage of water vapour, therebyallowing the timber to dry out; however, thereappears to be no conclusive proof for such claims.
The solvent-borne exterior paints come inmany forms; for example, as a three-layer systembased on flexible alkyd resins which produce agloss finish, or a one-can system which is appliedin two coats which produces a low-sheen finish.
The water-borne exterior paints are based onacrylic or alkyd-acrylic emulsions applied ineither two- or three-coat systems. The water-borne system has a higher level of permeabilitycompared with the solvent-borne system. Evenmore important is the high level of film extensi-bility of the water-borne system which is retainedin ageing (Miller and Boxall, 1994), and whichcontributes to its better performance on-site thanthe solvent-borne exterior paints.
Test work has indicated that the pretreatmentof surfaces to be coated with a water-repellentpreservative solution has a most beneficial effectin extending the life of the complete system, firstby increasing the stability of the wood surfacethereby reducing the stresses set up on exposure,and second by increasing adhesion between thetimber surface and the coating. This concept ofan integrated system of protection employingpreservation and coating, though new for timber,has long been established in certain other mater-ials; thus, it is common practice prior to thecoating of metal to degrease the surface toimprove adhesion.
One specialised group of finishes for timberand timber products is that of the flame-retardantcoatings. These coatings, designed only to reducethe spread of flame, must be applied fairly thicklyand must neither be damaged in subsequentinstallation and usage of the material, nor theireffect negated by the application of unsuitablecoverings. Nearly all the flame retardants on theUK market intumesce on heating and the result-ing foam forms a protective layer of resistantchar.
541
48.5 References
48.5.1 Standards and specificationsBS 1282 (1975) Guide to the choice, use and applica-
tion of wood preservatives. BSI, London.BS 5707. Specification for solutions of wood preserva-
tives in organic solvents.Part 1: (1979) Specification for solutions for generalpurpose applications, including timber that is to bepainted. BSI, London.Part 2: 1976 (1986) Specification for pen-tachlorophenol wood preservative solution for useon timber that is not required to be painted. BSI,London.
EN 335–1 (1992) Hazard classes of wood and wood-based products against biological attack – Part 1:Classification of hazard classes.
EN 350–2 (1994) Durability of wood and wood-basedproducts – Natural durability of solid wood – Part2: Guide to natural durability and treatability ofselected wood species of importance in Europe.
EN 12369–1 (2001) Wood-based panels – Characteris-tic values for use in structural design. Part 1: Particleboards, OSB and fibreboards.
48.5.2 LiteratureHill, C.A.S. and Jones, D. (1996) The dimensional sta-
bilisation of Corsican pine sapwood by reactionwith carboxylic acid anhydrides. Holzforschung 50,5, 457–62.
Hon, D.N.S. (ed.) (1996) Chemical Modification ofLignocellulosic Materials, Marcel Dekker Inc., NewYork.
Miller, E.R. and Boxall, J. (1994) Water-borne coat-ings for exterior wood. Building Research Establish-ment Information Paper IP 4/94.
Morén, R.E. (1964) Some practical applications ofpolyethylene glycol for the stabilisation and preser-vation of wood. Paper presented to the British WoodPreserving Association annual convention.
Rowell, R.M. (1984) Chemical modification of wood.For. Prod. Abstracts 6, 75–8.
Turner, P. and Murphy, R.J. (1998) Treatments oftimber products with gaseous borate esters. Part 2:Process improvement. Wood Sci. Technol. 32,25–31.
References
542
Further reading
Barrett, J.D., Foschi, R.O., Vokey, H.P. and Varoglu,E. (eds) (1986) Proc. International Workshop onDuration of Load in Lumber and Wood Products.Held at Richmond, B.C., Canada, 1985, spec. pubno. SP–27, Forintek Canada Corp., 115 pp.
Barrett, J.D. and Foschi, R.O. (eds) (1979) Proceedingsof the First International Conference on Wood Frac-ture, Banff, Alberta, 1978, Forintek Canada Corp.,304 pp.
Bonfield, P.W., Dinwoodie, J.M. and Mundy, J.S. (eds)(1995) Workshop on Mechanical Properties of PanelProducts. Proc. of workshop held at Watford, 1995,and published by the management committee of ECCOST Action 508, 317 pp.
Bodig, J. and Jayne, B.A. (1982) Mechanics of Woodand Wood Composites, Van Nostrand Reinhold,New York, 712 pp.
Bravery, A.F., Berry, R.W., Carey, J.K. and Cooper,D.E. (1987) Recognising Wood Rot and InsectDamage in Buildings, Building Research Establish-ment Report, BRE, Watford, England, 120 pp.
Desch, H.E. and Dinwoodie, J.M. (1996) Timber –Structure, Properties, Conversion and Use, 7th edn,Macmillan, Basingstoke, England, 306 pp.
Dinwoodie, J.M. (2000) Timber – Its Nature andBehaviour, E & FN Spon, London, 256 pp.
Gordon, J.E. (1976) The New Science of Strong Mater-ials, 2nd edn, Penguin, 269 pp.
Hearmon, R.F.S. (1961) An Introduction to AppliedAnisotropic Elasticity, Oxford University Press, 136pp.
Hillis, W.E. (1987) Heartwood and Tree Exudates,Springer-Verlag, Berlin, 268 pp.
Hoffmeyer, P. (ed.) (1997) International Conference onWood–Water Relations. Proc. of conference held inCopenhagen in 1997, published by the managementcommittee of EC COST Action E8, 469 pp.
Jane, F.W. (1970) The Structure of Wood, 2nd edn,Adam & Charles Black, London, 478 pp.
Kollmann, F.F.P. and Côté, W.A. (1968) Principles ofWood Science and Technology, I, Solid Wood,Springer-Verlag, Berlin, 592 pp.
Kollmann, F.F.P., Kuenzi, E.W. and Stamm, A.J.(1975) Principles of Wood Science and Technology.II, Wood-Based Materials, Springer-Verlag, Berlin,703 pp.
Morlier, P. (ed.) (1994) Creep in Timber Structures,RILEM report 8, E & FN Spon, London, 149 pp.
Morlier, P., Valentin, G. and Seoane, I. (eds) (1992)Workshop on Fracture Mechanics in WoodenMaterials. Proc. of workshop held at Bordeaux,France, April 1992, and published by the manage-ment committee of EC COST Action 508, 203 pp.
Pizzi, A. (1983) Wood Adhesives; Chemistry andTechnology, Marcel Dekker, New York, 364 pp.
Preston, R.D. (1974) The Physical Biology of PlantCell Walls, Chapman and Hall, London, 491 pp.
Siau, J.F. (1988) Transport Processes in Wood,Springer-Verlag, Berlin, 245 pp.
Skaar, C. (1988) Wood–Water Relations, Springer-Verlag, Berlin, 283 pp.
Wise, L.E. and Jahn, E.C. (1952) Wood Chemistry,2nd edn, Rheinhold Publishing Corp., New York,688 pp.
Index
AAC see autoclaved aerated concrete Abrams law 167absorbed water 158absorption
aggregates 126clay bricks and blocks 287–8concrete 194, 212
accelerators see admixturesacid attack
concrete 204–5masonry 320
acid rain 318acidity timber 427, 438acrylic copolymers 342acrylics 343activation energy 12, 15active silica 118, 119, 205, 206activity coefficient, cement replacement
materials 171adhesion bitumen 253–4adhesives 41–2, 343–4admixtures 93, 110–16, 148, 182–3
accelerators 112–13, 278calcium chloride 112chloride free 113
air entraining agents (AEAs)114–15, 182–3, 208
classification 115corrosion inhibiting 214plasticisers (water reducers,
workability aids)110, 182–3,277
retarders 110, 112, 113–14, 278,279
superplasticisers (high range waterreducers) 111–12, 182–3, 217,218, 219
adsorption 14, 42–3, 194, 212energy 43
AEA see admixtures, air entrainingagents
ageing index 126aggregate content concrete 148, 158,
169, 180aggregate crushing value 235aggregates
absorption 126bulk density 124
carbonate 202classification 123–5
shape 123–5size 123–5
coarse 93, 123–5, 180–1crushed rocks 121–2, 234elastic properties 126fine (sand) 93, 123–5, 180–1fineness modulus 123for bituminous materials 227–8,
234–6, 242, 254–5, 260for concrete 121–6for masonry 274gap graded 259, 293graded 123grading 123–4, 227–8, 229, 234,
274gravels 121–2, 234heavyweight 122–3impact resistance 235 lightweight 92, 122, 148, 210
expanded clay 122expanded shale 122foamed slag 122sintered pfa 122
limestone 202, 205, 209, 218normal density 121–2, 148permeability 196porosity 126sieve analysis 123–4single size 123skid resistance 235 stiffness 148strength 126surface characteristics 127, 234
air content mortar 280air entraining agents see admixturesair entrainment
concrete 209mortars 277, 279
alite 97, 99alkali resistant glass fibres 405alkali-aggregate reaction 205alkalis in cement 206, 207, 211alkali-silica reaction (ASR) 205–8allotropes 60allotropy–iron 24alloys 29, 60–1
alumina 9, 97aluminium and alloys xxiii, xxv, 9,
49, 55, 59, 68, 73, 77, 85–7aluminium–copper equilibrium diagram
23aluminium-silicon equilibrium diagram
21–2ammonia 40amorphous solids 12anisotropy, timber 494–5annealing 67, 70annual rings, timber 432anodic reaction 74appearance, timber 443–7aragonite 204aramid fibres (Kevlar) 352, 354 argon 26Armco iron 57asbestos cement 393, 403–5asbestos fibres 387ashlar stonework 269, 297asphalt 227–8, 229–30, 252, 259–60
basecourse 228porous 261–2stone mastic 262wearing course 228
asphaltenes 230–1, 237, 252ASR see alkali–silica reaction atomic
bonding 25–30spacing 32structure 25–30
austenite 24austenitic steel 84autoclaved aerated concrete (AAC)
277, 291–2, 292–3autoclaved polymer composites 371autogenous shrinkage 150
bakelite 12basecourse 228basic creep 157belite 97, 100Bessemmer converter 55binders for masonry 274Bingham constants concrete 129,
133–4Bingham equation 38
543
Bingham model, cement and concrete129
bitumens 68, 227, 229–34adhesion 253–4ageing 251–2blistering 256deformation 240–3dispersion hardening 68loss of volatiles 252manufacture 230modulus 241oxidation 251–2rheology 238–40sources 229–30stiffness 241types 231–4viscosity 231, 238–40, 242, 243,
248, 255weathering 256
bituminous materials 227–66, 345bitumen content 249cracking 244–6durability 251–7failure 244–50
modes 244–6fatigue 245–50
characteristics 246factors effecting 248–50tests 247–8
permeability 252–3practice and processing 259–65production methods 264–5strength 244–50structure 229–36viscosity 237void content 248, 253
bituminous mixtures, mix design262–4
Blaine method 96bleeding, concrete 136–8blended Portland cements 117blistering, bitumens 256Bogue formulae 97bolting of metals 72, 75, 86Boltzmann’s
superposition principle 37, 479–80constant 16equation 34
bondfibre/matrix 358patterns, brickwork 299–301
bond strength fibre/matrix 386, 395–6mortar 280
bonding covalent 27–8ionic 26–7metallic 28
bound water, timber 453, 454brasses 87brazing 71
brazing copper 87
break-off test see concrete near-to-surface tests
bricks and blocks bricks and blocks calcium silicate
290bricks and blocks clay 282–9bricks and blocks concrete 290–3
brickwork and blockwork 267–331 see also masonry
brickwork bond patterns 299–301brittle fracture 44, 45–6, 48brittleheart 443bronzes 87brucite 204bulk density aggregates 124bulk modulus 9, 10, 12butyl rubber 342
C2S see dicalcium silicateC3A see tricalcium aluminateC3S see tricalcium silicateC4AF see tetracalcium aluminoferritecadmium 59, 77calcined clay 118calcining 92, 95, 96calcite 114, 166, 170
see also calcium hydroxidecalcium
aluminates 96carbonate 95, 273, 275chloride 112, 213, 278formate 113, 278hydroxide (calcite) 101, 104, 118,
150, 204, 208, 291, 319, 322,388, 408
monosulfoaluminate 99nitrite 214silicate 14, 96, 201, 205silicate bricks and blocks 290silicate hydrate (C-S-H) 100, 101,
113, 139, 145, 208sulfoaluminate see ettringite
cambial cells 428cambium 430capillary
pores, Portland cement paste 102,103, 104, 195, 208
tension 146water 147, 158
carbenes 230–1carbon
fibre composites xxiii, 371–2,382–4
fibres 352, 353–4, 387carbonate aggregates 202carbonation
concrete 211, 212–13 mortar 315, 318–19shrinkage 150steel-fibre concrete 417–18
carboxylic acid 111, 114, 115cast aluminium 85
cast iron xxiii, 23, 78–9casting of metals 69–70cathodic protection 77
steel in concrete 215cathodic reaction 74cavity walls 300, 309CBPB see cement bound particleboardcell walls, timber 440–2cellulose 434–5, 464, 499, 504, 517
fibres 387cement 13, 30, 92, 95–108
see also Portland cementbound particleboard (CBPB) 531content 178extenders see cement replacement
materialspaste 129, 137
content 158diffusivity 197permeability 195–6porosity 195
cement replacement materials (CRMs,cement extenders, latenthydraulic materials, mineraladmixtures, supplementarycementitious materials) 93,117–20, 197, 206, 212, 213
see also microsilica, ggbs,metakaolin, pfa
activity coefficient 171calcined clay 118cementing efficiency factor 171,
181composition 119dosage levels 119effect on concrete strength 170–1metakaolin 118particle sizes 119physical properties 119rice husk ash 118specification 120supply 120
cementing efficiency factor 171, 181characteristic strength xxv–xxvii
concrete 176, 178high strength concrete 217timber 491, 512
Charpy test 81chemical
attack concrete 185disbonding bitumens 257processing timber 534–9pulping timber 538
chemisorption 43chipboard 529–30chlorides 320
in concrete 211, 213–15chlorine 26, 27chloroaluminates 213chromium plating 75chromium 59, 73, 75, 77, 84citric acid 114
Index
544
classification admixtures 115aggregates 123–5cement 108–9, 115
Clausius–Clapeyron equation 19clay 14, 38, 39, 95clay bricks and blocks 282–9
drying and firing 285–6forming 283–5properties absorption 287–8properties 286–9
clinker 95coarse aggregate see aggregates, coarsecobalt 59coefficient of diffusion 6coefficient of thermal conductivity
timber 461typical values 49
coefficient of thermal expansion concrete 141, 151–2hardened cement paste 151masonry 315timber 466–8
coefficient of variation xxv, 493cohesiveness fresh concrete 128cold
press moulding polymer composites372–3
rolled steel 82rolling 70working 64, 70
compactability fresh concrete 128compacting factor test fresh concrete
130composition Portland cement 97, 98composition–temperature diagrams
19–24compound phases
concrete 97 metals 61
compression wood 443compressive strength 33
brittle materials 46concrete see concrete compressive
strength masonry 306–8mortar 280timber 491, 494, 506
concentration cell 75concentric cylinder viscometer 129concrete xxiii, xxv, 9, 49, 91–223
see also admixturesadmixtures see admixturesadsorption and absorption 194,
212aggregate content 180aggregate see aggregateautogenous shrinkage 150behaviour after placing 136–8bleeding 136–8bricks and blocks see bricks and
blocks concrete
carbonation 150, 211, 212–13characteristic strength 176, 178chemical attack 185coefficient of thermal expansion
141, 151–2compressive strength 161–3,
165–71 cracking 162, 163, 171–3creep 143, 156–8
basic 157factors affecting 157–8drying 157mechanisms 158–9prediction 158–9rupture 172total 157
curing 138–9deformation 91, 143–59degradation 199–210diffusivity 193, 197, 213drying shrinkage 143–50
factors effecting 148–50prediction 148–50
durability 91, 176, 192–216acid attack 204–5alkali–aggregate reaction 205alkali–silica reaction (ASR)
205–8exposure conditions 215fire resistance 209freeze–thaw resistance 114–15frost attack 185, 208–9recommendations 200, 201,
203, 208, 215salt weathering 204sea water attack 204, 213steel in concrete see corrosion,
steel in concrete sulfate attack 200–1thamausite damage 201–4
early age properties 136–41elastic modulus 156elastic modulus dynamic 156,
185–7, 188elastic modulus static 156fracture 171–3
mechanics 172–3toughness 173
fresh 38, 91, 128–35 see also freshconcrete
grade 176heat of hydration 140high performance 111, 217–20high strength 217–19high workability 219history 91–2maturity 139microcracking 152, 154, 159,
171–3mix design 176–83
UK method 178–82with admixtures 182–3
with air entraining agents 182–3with cement replacement materials
181–2with plasticisers 182–3with superplasticisers 182–3
mix proportions 93, 176modulus of rupture 164near-to-surface tests 184, 190–1non-destructive testing 184–91
correlation with strength 185,187, 188, 191
rebound hammer 184–5resonant frequency 184, 185–7ultrasonic pulse velocity 184,
187–9 permeability 193, 194–7, 206, 208plastic settlement 137–8
cracking 138plastic shrinkage 138
cracking 138Poisson’s ratio 156properties for fibre composites 387secant modulus 156segregation 136–8self-compacting 219setting 136settlement 137shrinkage 143–50, 157slump 130, 132, 180slump-flow 130, 132, 219sorptivity 194, 197–8, 212, 213specific creep 157steel corrosion 113strength 161–74
biaxial loading 174effect of
age 168aggregate properties 169aggregate size 169aggregate volume 169cement replacement materials
170–1ggbs 170–1humidity 169metakaolin 170–1microsilica 170–1pfa 170–1transition zone 166water/cement ratio 166–7
factors effecting 165–71gain 139–42gain effect of temperature
139–41multiaxial loading 173–4tensile vs compressive 165tests 161–5
compressive 161–3cubes 162cylinder splitting 163cylinders 162flexural 163–4splitting 163
Index
545
concrete strength tests (cont’d)tensile 163–5
triaxial loading 174stress–strain behaviour 153–6
models 153–4tangent modulus 156target mean strength 178thermal expansion 151–2trial mixes 177–8workability 115
condensed silica fume see microsilicaCondon–Morse curves 32, 43, 45conductivity
electrical 29, 49–50thermal 29, 49–50
configurational entropy 34copper and alloys 59, 49, 87copper–nickel equilibrium diagram 20core wood 443corrosion
anodic and cathodic reaction 74cells 211, 214concentration cell 75control 76dry oxidation 73electrochemical series 74–5electrode potentials 74galvanic series 75of metals in masonry 317, 320of metals 73, 77protection 76sacrificial anodes 77steel in concrete 113, 210–15water line 75, 76wet corrosion 73
corundum 22costs of materials, typical values xxvcovalent bonding 27–8cover to steel in concrete 215cracking
bituminous materials 244–6concrete 162, 163, 171–3, 202,
205–6, 207, 211fibre composites 386fibre reinforced cements and concrete
389, 391–2glass reinforced cement 407in fatigue 47in fracture mechanics 46masonry 319polypropylene fibre reinforced
concrete 420roads 244–6
creep 34–5concrete 143, 156–8masonry 314–15polymers 339, 340rupture 44
concrete 172polymers 339timber 502
specific 157
test bitumens 241–2timber 468, 477–86
creosote 535cristobalite 19critical volume fraction fibre composites
390–1, 394, 407, 420CRMs see cement replacement materialscrushed rock 121–2, 234crystal structure 58–60
body centred cubic 59, 67face centred cubic 20, 58, 67hexagonal close packed 58
crystalline solids 12crystallization, water of 40crystals
close packed 29dislocations 46, 65, 66, 67, 70growth 67internal energy 59line defects 65metallic 29slip 64, 66slip planes 66unit cell 58
csf see microsilicaC-S-H see calcium silicate hydrate cube test concrete 162curing concrete 138–9, 291cutback bitumen 232, 239cylinder test concrete 162
Daniel cell 75Darcy’s law 193–4, 456deformation
bitumens 240–3concrete 91, 143–59timber 463–88
degradation concrete 199–210timber 515–20
degree of hydration, Portland cement100
de-icing salts 213delayed elastic strain 159dendrites 57dendritic growth 58density xxii
mortars 281timber measurement 447timber 432, 447–8, 473, 496–7
design for minimum cost xxiifor minimum weight xxii, 3life 76
roads 251destructive distillation timber 538–9detachment, bitumens 256diamond xxiii, xxv, 9dicalcium silicate (C2S) 97, 99, 100,
107, 113differentiation, timber 430diffuse-porous structure, timber 432,
442diffusion 5–6
in concrete 193moisture 149,158in timber 454–5, 459–61
diffusivity cement paste and concrete193, 197, 213
dilatancy 14dimensional stabilisers, timber 536–7disjoining pressure 147dislocation energy 66dislocations 46, 65, 66, 67, 70dispersion hardening 67displacement, bitumens 256drift velocity 50dry oxidation 73dry stone walls 269drying
creep, concrete 157kilns, timber 452shrinkage 143–50, 315shrinkage concrete 148–50shrinkage hardened cement paste
105–6, 143–7shrinkage mortars 279, 281, 315
dry-process fibreboard (MDF) 530–1DSP cement 219ductile fracture 44, 46–7, 48ductile iron 79ductile/brittle transition 47, 81ductility in metals 64, 66durability
aluminium alloys 86bituminous materials 251–7concrete 91, 114–15, 192–216, 176fibre reinforced cements
asbestos cement 405glass reinforced cement 407
masonry 317–22 steel in concrete see corrosion, steel
in concrete steel-fibre concrete 417–18timber 515–24
duralumin 86dynamic modulus concrete 156
earlywood 432, 458, 498efflorescence in masonry 318, 322Einstein’s equation 38elastic constants 9–12
relation between 11–12elastic deformation timber 468,
469–76elastic modulus (modulus of elasticity,
stiffness, Young’s modulus)aggregates 126asbestos cement 405bitumens 241concrete 156, 185–7, 188 definition 9fibres 355, 386–7glass reinforced cement 406–7
Index
546
high strength concrete 218masonry 314–15mortars 281polymers properties 339–40timber 470–6, 491, 494typical values 9
elastic resilience 10elastic limit definition 62elasticity 31–7
hardened cement paste 152elastomers 344electrical conductivity 29, 49–50electrochemical series 74electrode potentials 74electromotive series 74, 75electrons 25, 49electroplating 74emulsification bitumens 256–7emulsions and emulsifiers 233energy
activation 12, 15and equilibrium 15–24dislocation 66free 17internal 15of adsorption 43surface 39–40
entropy 16, 34epoxies, epoxy resins xxiii, xxv, 9,
338, 340, 343, 375, 376, 377epoxy coated reinforcement for concrete
214equilibrium
diagrams 17–24, 60, 61aluminium/copper 23aluminium/silicon 21–2copper/nickel 20–1iron/carbon 23–4iron-carbon 80silica/alumina 22single component 17–19two component 19–24water 17
thermodynamic 17ettringite (calcium sulfoaluminate) 99,
107, 200, 204, 319eutectic
composition 22point 22systems 21–4temperature 22
eutectoid 24expanded clay 122expanded shale 122extraction metallury 55extraction of iron 77extractives, timber 427, 438extrusion of metals 70
face-centred cubic crystal 20failure
bituminous materials 244–50
concrete 171–3rate xxviiroads 245–6strain, fibres 386–7timber 503–10
fair-faced masonry 272fast fracture 48fatigue 44, 48
bituminous materials 245–50cracks 47timber 509–10
Ferets’s rule 167ferric hydroxide 210, 211ferrite 24, 80ferritic steel 84ferrous hydroxide 210, 211fibre fibre bond 358, 386fibre composites 346–422
fibre reinforced cement and concretes349, 385–422
polymer composites 346–84timber as 438–40
fibre reinforced cements 403–14asbestos cement 403–5glass reinforced cement 405–9polymer fibre reinforced cement
411–14with natural fibres 409–11with vegetable fibres 410–11with wood fibres 409–10
fibre reinforced cements and concrete385–422
bond strength 396cracking 389, 391–2critical volume fraction 390–1, 394,
407fibre and matrix properties
386–8fibre pull-out 395, 396fibre–matrix interface 388flexure 398–402fracture mechanics 390, 397–8stress–strain curves 389–90, 391,
393–4structure and post-cracking
composite theory 389–402toughness 396–7uniaxial tension 389–98
fibre reinforced concrete 415–21fibre saturation point, timber 454,
475, 500fibre volume, fibre reinforced cements
and concrete 387fibre/matrix interface 357, 376fibreboard 530–1fibres
aramid 352, 354asbestos 387carbon 352, 353–4, 387cellulose 387glass 352, 387
in timber 430polyamide 354polyester 354polyethelyne 354, 387polypropylene 210, 354, 387polyvinyl alcohol 387steel 219, 387
fibrils, timber 439Fick’s law 6, 193, 460filament winding technique, polymer
composites 372filling ability, self-compacting concrete
219film rupture, bitumens 256film stacking process, polymer
composites 374fine aggregate see aggregate finefineness modulus 123finishes, timber 539–40fire resistance
concrete 209masonry 325–6timber 520–3
fitness-for-purpose xxiiflame retardants, timber 536, 540flexural strength
concrete 163–5masonry 311–12steel-fibre concrete 416
flow curve, fresh concrete 129flow in timber 454–61flow table test, fresh concrete 131flowing concrete 133fluidity, fresh concrete 128fluids 5–7
Bingham 38Newtonian 38yield stress 38
fluorine 30foamed polymers 338foamed slag 122forging of metals 67, 70forming of metals 69–72fracture 44–8
brittle 44, 45–6, 48ductile 44, 46–7, 48energy 47
fracture mechanics 47–8concrete 172–3cracks 46ductile materials 47ductile/brittle transition 47fast fracture 48fibre reinforced cements and concrete
390, 397–8fracture toughness 47, 48Griffith’s theory 45strain energy 46stress intensity factor 47, 48surface energy 45timber 510toughness 47
Index
547
fracture toughness 47, 48concrete 173timber 510
free energy 17free lime 97, 98free water content 126, 179free water, timber 453, 454free water/cement ratio 126, 178free-electron theory 28freeze–thaw damage
damage concrete 114–15, 185damage masonry 321–2
fresh concrete 91, 128–35Bingham behaviour 38cohesiveness 128compactability 128compacting factor test 130flow table test 131fluidity 128plastic viscosity 129, 132–3rheology 128, 129single point tests 130–2slump test 130, 132slump-flow test 130, 132two point workability test 129Vebe test 130workability 128–9
factors effecting 132–4loss 134–5measurement 129–32
yield stress 129, 132–3stability 128
fresh properties concrete 128–35 seealso fresh concrete
frost attack concrete 208–9 see alsofreeze thaw damage
frost damage masonry 321–2frost resistance mortars 279
galvanic cell 75galvanic series 75galvanizing 75gap graded aggregate 259, 293gas law 6gases, inert 26, 30gases, viscosity 5gel bitumen 231, 237gel pores, Portland cement 101gels 12–13geosynthetics 344geogrids 345geo-linear elements 345geomembranes 345geotextiles 344ggbs (ground granulated blast furnace
slag) 118, 141, 148, 182, 201,207, 218
composition 119effect on concrete strength 170–1
Gibbs free energy 17Gibbs phase rule 18glass and glasses v, 12, 49
glass fibre 387polymer composites xxiii, 355–6,
370, 375, 377reinforced cement 393, 405–9reinforced concrete 421
glass fibres 352–3alkali resistant 376, 405
gluing of metals 72glulam see laminated timbergold 59grade, concrete 176grading
curves 234–5curves aggregates 123–4requirements aggregates 123–4timber 510–13
grain angle, timber 496grain boundaries, metals 66graphite 23
spheroidal 79gravel 121–2
aggregate 234GRC see glass reinforced cementgrey cast iron 79GRFP see glass fibre reinforced
polymersGriffith theory 45, 172ground granulated blast furnace slag see
ggbsgrout 93growth rings, timber 432, 444–6GRP see glass fibre reinforced
polymers guniting steel-fibre concrete 416gunmetal 87gypsum 96, 97, 98, 112, 113, 204
half, third and quarter bond 299–300hand-lay up technique, polymer
composites 369hardboard 530hardened cement paste (hcp) 93, 121,
166, 205absorbed water 105capillary water 105chemically combined water 105drying shrinkage mechanisms
146–7drying shrinkage reversible and
irreversible 145drying shrinkage 105–6, 143–7elasticity 152evaporable/non-evaporable water
105inner product 101interlayer water 105microstructure 100outer product 101porosity 143, 152, 208strength 104–5stress-strain behaviour 152structure 104–5
thermal expansion 151ultrasonic pulse velocity 188water in 105–6
hardness bitumen 231timber 494
hardwood 429–30, 449, 459cell structure 430, 432
Hatschek process 403hcp see hardened cement pasteHDPE see high density polyethyleneheartwood 427, 458, 517heat affected zone 71heat of hydration 140heat of transformation 19heat treated steels 83heavyweight aggregates 122–3helium 25Helmohotz free energy 17, 31hemicellulose 436–8, 464, 499high density polyethylene (HDPE) 338high early strength cement 97high performance concrete 111,
217–20high strength concrete 217–19
characteristic strength 217elastic modulus 218plastic viscosity 218slump 218stress–strain behaviour 218
high workability concrete 219high-lime pfa 119high-range water reducers see
admixtures, superplasticisersHoffman kiln 285–6Hooke’s law 9, 32, 469hot press moulding polymer composites
372–3hot working 70HRM see metakaolinhydrates 40hydration 14
Portland cement 93, 97, 98, 104,118, 134, 195
hydraulic cement 92hydraulic lime 276hydraulic scouring, bitumens 257hydrogen 25, 30
bond 30hydrostatic pressure 7hydroxycarboxylic acids 110
impact loading 44impact resistance aggregate 235 impact strength
fibre reinforced cements and concrete396
timber 500inert gases 26, 30initial surface absorption test (ISAT)
198initiation period 211–12, 213
Index
548
interaction energy 31interfacial zone see transition zoneinterlayer water 147, 158intermediate components 60internal energy 15
crystals 59internal fracture test see concrete near-
to-surface testsinterstitial solid solutions 60intrinsic permeability 193iodine 30ionic bonding 26–7iron 9, 55, 59, 78–9
atomic structure 25carbide 61cast 78–9ductile 79extraction 77hydroxide 74malleable 79nodular 79oxide 97pig iron 78, 79sulphide 79
iron-carbon equilibrium diagram23–4, 80
irreversible creep timber 481–2ISAT see initial surface absorption test
joining of metals 70–1joints, masonry 302juvenile wood 442
Kelvin’s equation 146Kevlar see aramid fibres kinematic viscosity 238knot area ratio 496knots, timber 428, 446, 473, 496krypton 26
laminated timber (glulam) 532latent hydraulic materials see cement
replacement materialslatewood 432, 458, 498latex additives 277, 279lead 49, 59Lever rule 20, 22lightweight
aggregate concrete blocks 292aggregates see aggregates lightweightmortars 282
lignin 436–8, 473, 482, 499, 517lignin-pectin complex 430, 440lignosulfonates 111, 114lignosulfonic acids 110lime 22, 92, 97, 276lime mortars 281, 297lime staining, masonry 322limestone 254, 273
aggregates 202, 205, 209, 218dust 96powder 118, 219
limit of proportionality 62line defects in crystals 65liquids 5liquidus line 20lithium 25load deflection behaviour, timber
469–70loss of volatiles, bitumens 252low heat Portland cement 107low-lime pfa 119
macadam 227, 242, 252, 260–1macro-defect-free cement (MDF) 105magnesium 26, 55, 59
chloride 26oxide 26
malleable iron 79maltenes 230–1, 237manganese 79, 80margin xxvii, 178Marshall quotient (Qm) 264Marshall test, bituminous materials
263martensite 68, 83martensitic steel 84masonry 269, 297–332
see also brickwork and blockworkcement 276clay bricks and blocks 282–9components 270–2construction and forms 297–303
appearance 303arches 298–9, 306bond patterns 299–301buildability, site efficiency and
accuracy 302–3joints 302tunnels 306walls 298workmanship and accuracy 302
durability 317–29abrasion 322biological attack 317chemical attack 317, 318–21corrosion of embedded metals
317, 329efflorescence 318, 322erosion 317, 321frost attack 321–2and non-structural properties
317–28staining 318, 322–3
fair-faced 272fire resistance 325–6forms 269mortar 269, 270, 273–82, 297
component materials 273–8mixing and forming 297properties 278–2
hardened mortar 280–2unset mortar 279–80
thin-bed mortars 279
natural stone 293–4rain resistance 324–5reinforced and post-tensioned 299sound resistance 325strength see masonry structural
behaviour and movementstructural behaviour and movement
304–16bending 304, 311–13compression 304, 305–9concentrated loads 308creep 314–16elastic modulus 314–15shear 304, 310–11stability 308tension 313–14
sustainability issues 326thermal conductivity 323–4ties and fixings 294units 270–2
size 271types 271
matched die technique, polymercomposites 372
maturity of concrete 139MDF cement see macro defect free
cementMDF timber see dry-process fibreboardmechanical metallurgy 56mechanical pulping, timber 532–3mechano-sorptive behaviour, timber
485mediumboard 530melamine formaldehyde 111metakaolin (HRM) 118, 388, 408
composition 119effect on concrete strength 170–1
metallic bonding 28metallic crystals 29metallography 57metals and alloys 53–88
annealing 67, 70bolting 72, 75brazing 71casting 69–70chromium plating 75cold working and rolling 64, 70–1compound phases 61corrosion 73crystal structure 58–60dispersion hardening 67ductility 64, 66elasticity 62extrusion 70forging 67, 70forming 69–72galvanizing 75gluing 72grain boundaries 66grain structure 57–8hot working 70joining 70–1
Index
549
metals and alloys continuedmechanical properties 62–8oxidation 73pinning 72plasticity 63, 64–6recovery 67recrystallization 67riveting 72, 75rolling 67soldering 71solutions and compounds 60–1strain hardening 67strengthening 66–7stress-strain behaviour 62tensile strength 63–4toughness 66ultimate tensile strength 64welding 70work hardening 67
micelles 238microcracking, concrete 152, 154,
159, 171–3microfibrils, timber 438–42, 464, 473,
498microfilm durability test 252microsilica (csf, condensed silica fume)
118, 119, 148, 170–1, 207, 218,388, 408
middle lamella 440mild steel 80mill scale 70mineral additives see cement
replacement materialsmineral admixtures see cement
replacement materialsminerals in timber 438mix design
bituminous mixtures 262–4concrete 176–83
UK method 178–82with admixtures 182–3with air entraining agents 182–3with cement replacement materials
181–2with plasticisers 182–3with superplasticisers 182–3
polypropylene fibre reinforcedconcrete 420
steel-fibre concrete 415–16mix proportions by weight and volume
93concrete 93, 176
mixing 16modulus of elasticity see elastic modulusmodulus of rigidity see shear modulus modulus of rupture
concrete 164timber 494
moisture content, timber 427, 447,449–52, 474, 491, 499–500,517, 525
moisture diffusion 149, 158
timber 459–61moisture in timber 427–62, 450–4,
463–6molybdenum 59, 84montmorillonite clay 39mortar 92, 93
for masonry 269, 270, 273–82spread test 111ultrasonic pulse velocity 188matrix properties for fibre composites
387mortars lightweight 282movement masonry see masonry
structural behaviour andmovement
movement timber 465–6mullite 22
naphthalene formaldehyde 111natural asphalt 229–30necking 47, 63, 64neon 26neutrons 26Newtonian behaviour, fluids and liquids
5, 38, 129, 340, 469nickel 59, 73, 77, 84niobium 59nodular iron 79non-destructive testing of hardened
concrete see concrete non-destructive testing
normal brickwork 269normal density aggregates 121–2, 148nylon xxiii, xxv, 9, 340
Ohm’s law 50oriented strand board (OSB) 531OSB see oriented strand boardoxidation
bitumens 251–2metals 73
oxidised bitumen 232
paints 77, 540parenchyma 430, 432, 504partial cement replacement materials see
cement replacement materialsparticleboard 529–30passing ability, self-compacting concrete
219pearlite 24, 60, 80penetration grade bitumen 231penetration index (PI) 239–40penetration resistance test 190 periodic table 26permeability
aggregate 196asphalts 259bituminous mixtures 252–3cement paste 195–6concrete 193, 194–7, 206, 208macadams 260
timber 454–6pessimum silica content 207PET 338pfa (pulverized fuel ash, fly ash) 118,
141, 148, 181, 207, 218, 273,416
composition 119effect on concrete strength 170–1high-lime 119low-lime 119
phase rule 18, 20phosphorus 81physical metallurgy 56, 57–61PI see penetration indexpig iron 55, 78, 79pigments 277pinning of metals 72pitch 77pitting blistering 256plastic deformation 46plastic shrinkage
concrete 138polypropylene fibre reinforced
concrete 420plastic viscosity
fresh concrete 129, 132–3high strength concrete 218
plasticisers see admixturesplasticity 37–8, 63, 64–6plywood 528–9Poiseuille equation 456Poisson effect, concrete 162, 174Poisson’s ratio 11
concrete 156fibres 386–7limits 12masonry 314timber 471–2typical values 9
polyamide 345fibres 354
polycarboxylates 111polyester fibres 354 polyesters, polyester resins 338, 339,
340, 345, 375, 376, 377polyethylene 12, 33, 49, 338, 339,
340, 345fibres 354, 387
polyethylene fibres in cementcomposites 413–14
polyethylene pulp in cement composites413
polymer composites 346–84analysis of behaviour 357–68
elastic properties 358–60laminate theory 360–4strength and failure 364–8stress-distribution 359
design 378durability 375–8
effect of fire 376effect of moisture 376
Index
550
effect of solvents 377effect of temperature 375–6effect of ultraviolet light 377,
379weathering 377–8
fibres for 346–56manufacturing techniques 369–74
thermoplastic composites 374thermosetting composites
369–74uses 379–84
aircraft and space applications380
civil engineering applications380–4
column wrapping 383–4composite bridges 381marine applications 380pipes and tanks 380reinforcing bars 384retrofitting of bonded plates
381–3trucks and automobiles 380
polymer fibre composites see polymercomposites
polymer fibre reinforced cement411–14
polymers 12, 28, 33, 335–45applications and uses 341–4
adhesives 343–4elastomers 344geosynthetics 344sealants 341–2
blow moulded 338co-extruded 339dislocation 46oriented grid sheets 339profile film-blown 338profile products 338properties 339–40
elastic modulus 339–40strength 339stress–strain behaviour 340–2
thermoplastic 12, 335, 337–8, 339,345
thermosetting 12, 335, 378, 339 types, applications, uses 335–45
polypropylene 338, 345fibre reinforced concrete 419–21
applications 421durability 421fresh properties 420hardened properties 420–1mix design and manufacture 420plastic shrinkage cracking 420
fibres 210, 354, 387, 394, 419in cement composites 412–13in concrete 394, 396
polysulfides 342polyvinyl alcohol (PVA) fibres 387,
413polyvinyl chloride (PVC) 340
porosityaggregates 126hardened cement paste 143, 152,
195timber 430, 450
porous asphalt 261–2Portland blast furnace cement 120Portland cement 22, 92, 95–108, 403
ASTM types 108capillary pores 102, 103, 104CEM types 108composition 97, 98, 112compounds 97, 98degree of hydration 100dormant period 98, 101final set 98free lime 97gel pores 101heat output 98, 106high early strength 97hydration 14, 93, 97, 98–104, 118,
195initial set 98low heat 107manufacture 95modifications 106–8oxide composition 97, 98phase composition 97, 98porosity 101–4rapid hardening 107relative density 93, 96self-desiccation 102setting 106specific surface area (SSA) 96standards and nomenclature 108strength 104
gain 106requirements 108
sulfate resisting 97, 107, 200, 202white 107
Portland composite cements 117Portland pozzolanic cement 120potassium silicate 205pozzolanic behaviour 117–18pozzolans 92, 118preform moulding, polymer composites
373preservatives timber 534–6pressure bag techniques, polymer
composites 370–1pressure-temperature diagram, water
17processing of timber 525–41proof stress 62protons 25pull-off test 190–1pull-out test 190–1pull-winding technique, polymer
composites 373–4pultrusion technique, polymer
composites 373–4pulverised fuel ash see pfa
PVA see polyvinyl alcoholPVC see polyvinyl chloridepyrite 202
Qm see Marshall quotientquartz 19quenching steel 68
random rubble walls 269rapid hardening Portland cement 107rays, timber 446reaction wood 442reactive powder concrete (RPC) 219rebound hammer see concrete non-
destructive testingrebound number 185reconstructed stone 292recovery, metals 67recrystallization in metals 67refinery bitumen 230reinforced and post-tensioned masonry
299reinforcement composite bars 384relaxation spectrum 36relaxation time 36REPLARK process 371resilience 10resin injection polymer composites
373resonant frequency test see concrete
non-destructive testingretarders see admixturesretrofitting, polymer composites
371–2, 381–4reversible creep, timber 481–2rheological models 480
see also viscoelastic modelsrheology
bitumens 238–40fresh concrete 128, 129
rice husk ash 118RIFT process 371ring-porous structure timber 432riveting aluminium 86riveting of metals 72, 75roads
cracking 244–6design life 251failure 245–6rutting 241–2, 244skidding resistance 244
Roman cement 92rotary kiln 95RPC see reactive powder concreterubber xxiiirusting of steel 77
sacrificial anodes 77salt weathering concrete 204sand for mortars 273–4sand see aggregates, finesandlime 276
Index
551
sapwood 427, 457, 517saturation point, superplasticisers 112saturation vapour pressure 7Schmidt hammer see concrete, non-
destructive testing, reboundhammer
sea water attack concrete 204, 213sealants 341–2seasoning timber 452–3, 465secant modulus 9
concrete 156masonry 315timber 470
segregation resistance self-compactingconcrete 219
segregation, concrete 136–8selenium 27self-compacting concrete 219self-desiccation 102, 145, 150setting, concrete 136settlement, concrete 137shale 95shear modulus (modulus of rigidity)
10, 66timber 471–2typical values 9
shear strain definition 9shear strength masonry 310–11shear stress 9shrinkage
autogenous 150carbonation 150concrete 143–50, 157drying 143–50, 315plastic 138timber 442, 463–5
sieve analysis 123–4SIFCON steel fibre concrete 416silica 12, 19, 22, 95, 97, 118
active 205, 206gel 14, 43glass xxiii, xxv, 9, 19
silica-alumina equilibrium diagram 22silicon 79silicone putty 14silver 49, 59single point tests, fresh concrete
130–2sintered pfa 122skid resistance 244, 259slip in crystals 64, 66slump
high strength concrete 218loss 134test 130, 132
slump-flow in self-compacting concrete
219–20test 130, 132
sodium 26chloride 26oxide equivalent 206, 212
silicate 205softboard 530softening point (SP) 339softwood 429, 449, 456–7
cell structure 429–30soldering 71
copper 87solid solutions 60solids 7–12
amorphous 12crystalline 12yield strength 47
solidus line 20solvus line 23sorption, timber 453sorptivity, concrete 194, 197–8, 212,
213SP see softening pointspalling of concrete 211specific creep
concrete 157timber 477
specific heat 50spheroidal graphite 79spherulites 12spraying, steel-fibre concrete 416spray-up technique, polymer composites
370spreading force 41SSA see Portland Cement, specific
surface area.stability fresh concrete 128stainless steel 73, 80, 83–4
stainless steel reinforcement forconcrete 214
standard deviation xxv, 178, 493static modulus concrete 156steel 55, 49, 77, 79–84
austenitic 84carbon 47cathodic protection in concrete 215cold rolled 82corrosion in concrete see corrosion,
steel in concreteductile-brittle transition 81ferritic 84fibre concrete 396, 415–18fibres 219, 387heat treated 83high strength xxiiimartensite 68, 84mild 80quenching 68rusting 77stainless 73, 80, 83–4stainless reinforcement for concrete
214strength grades 81stress corrosion cracking 84structural xxiii, xxv, 80–3tempering 68weld decay 84
welding 82yield strength 81
steel-fibre concrete applications 418–19durability 417–18mix design and manufacture
415–16properties 417
steric hindrance 110, 111stiffness aluminium 85stiffness xxii–xxiii, 33 see also elastic
modulusstone mastic asphalt 262strain energy 46
density 10strain hardening 62, 67strain 9strength xxii
aggregates 126bituminous materials 244–50concrete 161–74
gain 139–42tests 161–5
grade steel 81timber 489–514
effect of moisture content 453typical values xxiii
strengthening of metals 66stress 9
corrosion cracking 44, 84intensity factor 47, 48relaxation 34–5
stress–strain behaviour asbestos cement 405concrete 153–6fibre reinforced cements and concrete
389–90, 391, 393–4glass reinforced cement 406hardened cement paste 152in tension 44metals 62
stretcher bond 299–300structural behaviour masonry see
masonry structural behaviourand movement
structural steels xxiii, xxv, 80–3structure, sensitive and insensitive
properties 62substitutional solid solutions 60sucrose 114sulfate attack
concrete 200–1mortar 319
sulfate resisting Portland cement 97,107, 200, 202
sulfonic acid 114sulphur 27, 81superplasticisers see admixturessuperposition theory 37surface
adhesives 41–2characteristics, aggregates 127
Index
552
energy 39–40, 45, 146hardness concrete see concrete non-
destructive testing reboundhammer
laitance 138tension 39–40, 42, 146wetting 40–1
surfaces 39–44
tangent modulus 9concrete 156timber 470
tar 77, 227tar viscometer 232target mean strength concrete 178tarmacadam 227tellurium 27temperature rise in concrete 140tempering of steel 68tensile strength 33
fibres 386–7masonry 313metals 63–4polypropylene fibre reinforced
concrete 420–1steel-fibre concrete 416timber 491
tension wood 443tetracalcium alumina ferrite (C4AF)
97, 99, 107, 113thamausite damage concrete 201–4,
319thermal conductivity 29, 49–50
masonry 323–4mortars 279timber 460–1Wiederman–Franz ratio 50
thermal expansion aluminium 85concrete 151–2hardened cement paste 151timber 466–8
thermodynamics 3, 15, 31entropy 16equilibrium 17first law 3, 16, 17, 31second law 3, 17, 31
thermoplastic polymers 12, 335,337–8, 339, 345
thermosetting polymers 12, 335,338–9
thin-bed mortars 279, 282thixotropy 14timber xxiii, 423–542
anisotropy 463appearance 443–7
figure 444texture 444
brown rot 519cell structure 505–6characteristic strength 491, 512chemical constituents 434–8
chemical degradation 516chemical processing 534–9consumption 425creep rupture 502deformation 463–88
moisture movement 463–6thermal movement 466–8under load 468–86
creep 468, 477–86elastic modulus 470–6elastic 468, 469–76limit of proportionality 470load-deflection behaviour
469–70Poisson’s ratio 471–2
viscoelastic 476–82, 501degradation processes 515–20density 432, 447–8, 473design stresses 510destructive distillation 538–9diffusion 454, 459–61dimensional changes 463–6dimensional stabilisers 536–7drying 452–3durability 515–24effect of sunlight 516 fatigue strength 509–10 fibre saturation point 454, 475, 500finishes 539–40flame retardants 536fracture toughness 510fungal decay 519grading 510–13
machine 512visual 512
grain 433, 444hardness 494impact strength 500impregnation 534insect attack 519–20marine borers 520mass–volume relationships 446–50mechanical degradation 516mechanical processing 525–34mechano-sorptive behaviour 485modulus of rupture 494moisture content 427, 474, 491,
517, 525movement 465–6natural durability 517–20performance in fire 520–3, 536permeability 454–6photodegradation 515–16preservatives 534–6processing 525–41
board materials 527–32mechanical 525–34sawing and planing 525–6
pulping 532–3, 538raised grain 526seasoning 452–3, 465shrinkage 442, 463–5
strength 489–513compressive 491, 494, 506determination 489–90effect of
anisotropy 494–6cell length 498chemical composition 498–9density 496–7fibre angle 498grain angle 494–6knots 496latewood/earlywood ratio 498moisture content 499–500rate of loading 501–2reaction wood 499ring width 497–8temperature 500–1time 501–2
factors effecting 494–503tensile 491values from small clear test pieces
491values from structural size test
pieces 491variability 491–4
structure 427–62macroscopic 427–9microscopic 429–34molecular and ultrastructure
433–42variability 442–3
thermal degradation 516thermal expansion 466–8toughness 501, 507–9treatability 534uses 425weathering 516 wet-process fibreboard 530white rot 519
titanium xxiii, xxv, 9, 59, 84total creep concrete 157toughness xxii, 44, 47
fibre reinforced cements and concrete396–7
glass reinforced cement 406, 407metals 66steel-fibre concrete 417timber 501, 507–9typical values xxiii
tracheids 430, 540transformations 19transition
temperature 48zone 121, 137, 166, 170, 171, 173,
196, 205, 388ductile/brittle 47
transport mechanisms through concrete192–9
trial mixes concrete 177–8triaxial test, bitumens 241–2tricalcium aluminate (C3A) 97, 99,
107, 108, 112, 113, 319
Index
553
tricalcium silicate (C3S) 97, 99, 107,113
tridymite 19Trinidad Lake asphalt 229true stress 64tungsten carbide 9tunnel kiln 286two point workability test 129
ultimate tensile strength 64ultrasonic pulse velocity see concrete
non-destructive testing
vacuum bag polymer composites370–1
valency 26Van der Waals bonds 29–30, 43, 147Van der Waals equation 6vanadium 59vapour pressure 7vapour–liquid transition 6–7variability
of materials xxv–xxviitimber 442–3, 491–4
Vebe test fresh concrete 130vegetable fibres in cement composites
410–11vessels timber 430viscoelastic
creep 34deformation timber 476–82, 501models 36relaxation spectrum 36relaxation time 36
viscoelasticity 34–7polymer composites 352bitumens 240polymers 340
viscosity 38bitumens 131, 231, 237, 238–40,
242, 243, 248, 255emulsions 234gases 5liquids 5
wall effect 166water
bound 453, 454content free 126, 178in hardened cement paste 105–6of crystallization 40retentivity mortar 280structure 30
water/binder ratio 117, 218, 219water/cement ratio 93, 104, 112, 145,
166–7, 196, 197, 218free 126, 178
waterline corrosion 75, 76water-reducers see admixtures,
plasticiserswearing course
asphalt 228mixtures 259
weathering asbestos cement 405bitumen 256polymer composites 377–8timber 516
weld decay steel 84welded joints 48welding
aluminium 86cast iron 79copper 87heat affected zone 71metals 70
steel 82wet corrosion 73wetting surface 40–1wetwood 427white cast iron 78white cement 107Wiederman–Franz ratio 50wood fibres in cement composites
409–10work hardening 62work to fracture xxii
fracture timber 508workability
aids see admixtures, plasticisersconcrete 115, 176 see also fresh concretemortar 280
wrought aluminium 86wrought iron 55
xenon 26
yield point 62yield strength
strength metals 64, 66, 67strength solids 47strength steel 81
yield stressstress fluids 38stress fresh concrete 129, 132–3stress metals 62
Young’s modulus 9 see also elastic modulus
zinc 59, 75, 77zirconium carbide 26
Index
554
