In-plane seismic performance of glass facade systems · 2017-02-22 · IN-PLANE SEISMIC PERFORMANCE OF GLASS FAÇADE SYSTEMS SIVANERUPAN SIVAGNANASUNDRAM (SIVA) A thesis submitted

IN-PLANE SEISMIC PERFORMANCE OF

GLASS FAÇADE SYSTEMS

SIVANERUPAN SIVAGNANASUNDRAM (SIVA)

A thesis submitted in total fulfilment of the requirement of the degree of

Doctor of Philosophy

October 2011

Centre for Sustainable Infrastructure and Development, Faculty of Engineering and

Industrial Sciences

Swinburne University of Technology, Melbourne, Australia

ii

Declaration

This is to certify that this thesis comprises:

- No material which has been accepted for the award to the candidate of any

other degree, except where due reference is made in the text;

- Solely of my original work and due acknowledgement has been made

wherever other previously published material and references are used; and

- Less than words in length, exclusive of tables, maps, charts and

bibliographical references.

Sivanerupan Sivagnanasundram (Siva)

October 2011

iii

iv

Abstract

Glass façade systems in buildings are subject to racking action due to inter storey drift

caused by earthquake and wind actions. The performance of façade systems is

dependent on the amount of imposed drift and the interaction of the glass panels with

the façade structural support frames. There are two major concerns related to the glass

façade system performance during and immediately after a seismic event, hazards to

people from falling glass and cost associated with building down time and repair.

Glass façade systems can be classified into two types namely, framed glass façade

system (FGFS) and point fixed glass façade system (PFGFS). It was observed that the

damage to glass façade systems resulting from in-plane racking actions mainly

earthquakes is increasingly common and yet there have been limited number of

published research work available in this field. The research conducted to date mainly

focused on traditional framed glass façade systems. However, the seismic

performance of PFGFS is likely to be quite different from conventional framed

systems. Therefore, the aim of the research presented in this thesis is to assess the in-

plane racking performance of PFGFS which is gaining popularity worldwide.

Two unique full scale in-plane racking laboratory tests (Test #1 and Test #2) on

typical PFGFS with different types of spider arms (brackets to connect the glass and

the structural support frame) were conducted. Detailed 3-D finite element models

were developed and validated against the experimental test results to interpret the

racking behaviour of PFGFS. Specific racking mechanisms were attributed to the drift

capacity in each test. Further detailed FE analyses were conducted to evaluate the

individual drift contributions of each racking mechanism such as rigid body

translation at the built-in standard gaps, spider arm rotation and spider arm

deformation. It was found that most of the drift capacity is attributed to the rigid body

translation at the built-in standard gaps. The FE models were then used to predict the

racking performance of PFGFS with different configurations.

The seismic assessment of glass façade systems requires an estimate of the likely drift

demand from the building. Codified provisions for in-plane drift limits on glass

façade systems can be used as a conservative option. Analysis results presented in

this thesis indicated that the inter-storey drift demand is much less than the 1.5% limit

v

specified in AS 1170.4 (2007) for most buildings in Australia for a 500 year return

period seismic event except for soft storey structures. Standard seismic assessment

procedures can be used to estimate the optimum in-plane seismic drift demands from

the buildings. Based on that some rapid inter storey drift assessment methods were

presented with example calculations.

Conservatively, the in-plane racking capacity of PFGFS resulting from the rigid body

translation of the glass panels at the built-in standard gaps can be used as the design

in-plane drift capacity. If required, the drift capacity can be increased by introducing

special articulation features at the bolted connections. Care should be taken at the

boundary conditions of the perimeter glass panels to achieve the racking capacity of

the PFGFS from the rigid body translation at the built-in gaps. In order to assist

façade engineers, particularly at the conceptual design stage, a quick selection guide is

presented to identify the structural components of PFGFS which can increase the

racking performance of the façade system.

vi

Preface

One refereed journal paper and four refereed conference papers have been published

throughout the course of this research. These publications are listed below:

Chapter 2 has been presented in the following:

• Sivanerupan S, Wilson J.L, Gad E.F & Lam N.T.K 2008. Drift performance of

facade systems. Proceedings of the annual conference of the Australian

Earthquake Engineering Society (AEES). Ballarat, Victoria.


• Sivanerupan S, Wilson J.L & Gad E.F 2011. Structural analysis and design of

glazed curtain wall systems. Australian Journal of Structural Engineering

(AJSE), 12, 57-67.


• Sivanerupan S, Wilson J.L, Gad E.F & Lam N.T.K 2010. In-plane drift

capacity of point fixed glass façade systems. Proceedings of the annual

conference of the Australian Earthquake Engineering Society (AEES). Perth,

Western Australia.


• Sivanerupan S, Wilson J.L, Gad E.F & Lam N.T.K 2010. In-Plane Racking

Performance of Point Fixed Glass Façade Systems. Australasian Conference

on Mechanics of Structures and Materials (ACMSM 21). Melbourne,

Australia.


• Sivanerupan S, Wilson J.L, Gad E.F & Lam N.T.K 2009. Seismic Assessment

of Glazed Façade Systems. Proceedings of the annual conference of the

Australian Earthquake Engineering Society (AEES). Newcastle, NSW,

Australia.

vii

Abbreviations

The following abbreviations are used throughout this thesis.

ADRS - Acceleration displacement response spectrum

CQC - Quadratic combination

CSM - Capacity spectrum method

FE - Finite element

FGFS - Framed glass façade systems

GFS - Glass façade systems

LDP - Linear dynamic procedures

LSP - Linear elastic procedures

MDOF - Multi degree of freedom

NDP - Nonlinear dynamic procedures

NSP - Nonlinear static procedure

SDOF - Single degree of freedom

PFGFS - Point fixed glass façade system

RP - Return period

RSDmax - Maximum displacement demand for site class from the

response spectrum

RSDTe - Displacement demand corresponding to the effective

stiffness on the displacement response spectrum

SRSS - Square root of the sum of the squares

viii

Acknowledgements

I wish to express my profound gratitude to my principal coordinating supervisor

Professor John L. Wilson without whose continuous and valuable guidance my

research would never have concluded. His recognition of the ability in me to conduct

a research gave me the wings to fly to Australia.

The continuous help and encouragement from my coordinating supervisor Professor

Emad F. Gad gave me the impetus to continuously be engaged in my research despite

various difficulties. In particular, his guidance with the finite element modeling and

experimental tests has been invaluable.

I am deeply indebted to my associate supervisor Professor Nelson TK Lam for his

continuous support, guidance and suggestions that enabled me to successfully

complete my research.

The study was carried out in the Faculty of Engineering and Industrial Sciences,

Swinburne University of Technology, Melbourne, Australia. I am very grateful to the

University and the Centre for Sustainable Infrastructure (CSI) for providing me with

the scholarship to carry out my studies. I am very grateful to the Francis Lab and the

University of Melbourne where the laboratory tests were carried out.

My deep gratitude goes to Peter McBean from Walbridge and Gilbert Pty Ltd for his

continuous encouragement and suggestions for the research. I am very grateful to Dr.

Ignatius Calderoneo (Calderoneo and Associates Pty Ltd) for his continuous support

and special help in measuring the fracture strength of glass. I gratefully acknowledge

the support and suggestions from the façade engineering experts Dr. Raghu Pendyala

(Pendyala Consulting Pty Ltd), and Weng Chan (Aurecon).

I owe special thanks to Bill Vun, Jon Yan and Leonard Tan from Australian Glass

Assemblies who provided us with the necessary glass fittings and technical support.

My deep gratitude goes to Lynton Wombwell from Viridian World Glass who

provided us with the glass panels necessary for the laboratory tests. The Melbourne

Testing Services (MTS) provided technical assistance throughout the laboratory tests.

In particular, I thank Rodney Wilkie for his continuous support. I also acknowledge

ix

PhD candidate David heath from the University of Melbourne for his special help

with the Photogrammetry measurements.

I owe special thanks to my valued colleagues Deepti Wagle and Bara Baraneedaran

for their continuous support and help during the lab tests. I am deeply indebted to

fellow PhD candidates Vinoth Jayaratnam, Ari Wibowo, Aatheesan Thurairatnam,

Suthagaran Visvalingam, Tuan Nguyen, Ibrahim Saidi, Charley Lubinbert and other

researchers in the Faculty for their suggestions, support and encouragement. The

academic environment and friendliness among fellow researchers played a major role

in my productivity and eagerness to study. Also I would like to thank my house mates

Kugaruban Chelliah, Prad Pradeepan, Sen Senthilkumar and Tilak Makesan for their

encouragement during my studies.

I am deeply indebted to my parents in Sri Lanka for promoting a learning culture and

continuously encouraging me from thousands of kilometers away to stay engaged in

my research. I extend deep gratitude to my parents who not only encouraged me but

also provided me with emotional support during my stay in Melbourne. Also the

support from my siblings enabled me to carry out my research successfully.

x

Table of Contents

1 INTRODUCTION AND OVERVIEW .................................................................1

1.1 Introduction..................................................................................................1

1.2 Project Aims, Objectives and Methodology ................................................5

1.2.1 Critical Literature Review...................................................................5

1.2.2 Experimental Testing of Façade Systems ...........................................5

1.2.3 Analytical Modelling and Parametric Study.......................................6

1.2.4 Estimation of Inter-Storey Drift Demand in Buildings.......................6

1.2.5 In-plane Seismic Assessment and Design of PFGFS..........................6

1.3 Thesis Overview ..........................................................................................6

2 RESEARCH BACKGROUND .............................................................................9

2.1 Introduction..................................................................................................9

2.2 Glass Types................................................................................................11

2.2.1 Annealed Glass .................................................................................11

2.2.2 Heat-Strengthened Glass...................................................................12

2.2.3 Toughened Glass or Fully Tempered Glass......................................13

2.2.4 Laminated Glass................................................................................14

2.2.5 Insulating Glass.................................................................................15

2.3 Glass Façade Systems ................................................................................17

2.3.1 Framed Glass Façade Systems..........................................................17

2.3.2 Frameless Glass Façade System .......................................................19

2.4 Damage to Glass Façade Systems in Past Earthquakes .............................22

2.5 Previous Research on Framed Glass Façade Systems ...............................25

2.5.1 Experimental Study...........................................................................25

2.5.2 Analytical Study................................................................................30

xi

2.5.3 Standard Provisions for Framed Glass Façades................................35

2.6 Limited Number of Previous Research on PFGFS....................................36

2.6.1 Influence of Bushing Type in Load Bearing Capacity .....................36

2.6.2 Influence of Connection Type in Load Bearing Capacity ................37

2.6.3 In-plane Load Capacity of a PFGFS .................................................38

2.6.4 High Displacement Seismic Glass Systems......................................42

2.7 Codified In-Plane Drift Demands on Façade Systems ..............................46

2.8 Conclusion and Summary ..........................................................................48

3 STRUCTURAL ANALYSIS AND DESIGN OF GFS.......................................50

3.1 Introduction................................................................................................50

3.2 Design of Unitized Framed Glass Façade System.....................................51

3.3 Out-of-Plane Design ..................................................................................52

3.3.1 Structural Design of Glass Panel ......................................................52

3.3.2 Design of Mullion and Transom .......................................................52

3.4 In-Plane Design..........................................................................................52

3.4.1 Thermal Expansion of Mullion and Transom...................................53

3.4.2 Serviceability Limit State Deflection of Spandrel Beam..................53

3.4.3 Building Movement Caused by Wind Loading ................................56

3.4.4 Building Movement Caused by Earthquake Loading .......................57

3.4.5 In-plane Drift Capacity of Unitized Framed glass Façade................57

3.5 Design of Point-Fixed Glass Façade System.............................................58

3.5.1 Out-of-Plane Glass Panel Design......................................................58

3.5.2 In-Plane Glass Panel Design .............................................................60

3.5.3 Bolted Connection Location and Design ..........................................63

3.5.4 Stresses at the Glass Bolted Hole and Bolt Design...........................64

3.6 Conclusion and Summary ..........................................................................65

xii

4 IN-PLANE RACKING TESTS OF POINT FIXED GLASS FAÇADE

SYSTEMS....................................................................................................................66

4.1 Introduction................................................................................................66

4.2 Test #1 – ‘X’-Type Spider Arms and Countersunk Bolt Fitting ...............68

4.2.1 Test #1 – Experimental setup............................................................68

4.2.2 Test #1 - Experimental Results and Discussion................................69

4.3 Test #2 – ‘K’-Type Spider Arms with Button Head Bolt Fitting ..............78

4.3.1 Test #2 - Experimental Setup............................................................78

4.3.2 Test #2 - Experimental Results and Discussion................................78

4.3.3 Test #2 – Ultimate Fracture Strength of Toughened Glass...............79

4.4 Test Summary and Further Studies............................................................88

5 FINITE ELEMENT MODELLING OF THE IN-PLANE RACKING

PERFORMANCE OF PFGFS .....................................................................................89

5.1 Introduction................................................................................................89

5.2 Test #1- Structural Idealisation..................................................................90

5.2.1 Test #1- Racking Mechanism ...........................................................90

5.2.2 Model Assumption............................................................................93

5.2.3 Features of the model ........................................................................94

5.3 Test #1 - Model Description ......................................................................94

5.3.1 Element Description..........................................................................94

5.3.2 Material properties ............................................................................98

5.3.3 Material Model for Silicon Sealant...................................................98

5.3.4 Real Constants for the Elements .....................................................101

5.3.5 Boundary Conditions and Loading .................................................104

5.4 Test #1 - Results Comparison ..................................................................104

5.5 Test #1 – Effect of the Diagonal Strut Loads ..........................................109

xiii

5.6 Test #2 - Structural Idealisation...............................................................111

5.6.1 Test #2 - Racking Mechanism ........................................................111

5.6.2 Model Assumption..........................................................................112

5.6.3 Features of the Model .....................................................................113

5.7 Test #2 - Model Description ....................................................................113

5.7.1 Element Description........................................................................113

5.7.2 Material Properties ..........................................................................115

5.7.3 Real Constants for the Elements .....................................................116

5.7.4 Boundary Conditions and Loading .................................................117

5.8 Test #2 - Results Comparison ..................................................................118

5.9 Test #2 – Effect of the Diagonal Strut Loads ..........................................122

5.10 Summary and Conclusions ......................................................................123

6 PARAMETRIC STUDY ON PFGFS USING FE ANALYSES .......................124

6.1 Introduction..............................................................................................124

6.2 Test #1 - Parametric Study for 2x2 Grid Systems ...................................124

6.2.1 Test #1 – Built-in Standard Gaps at the Structural Support Frame 124

6.2.2 Test #1 - Sealant Types...................................................................127

6.2.3 Test #1 - Sealant Thickness ............................................................129

6.2.4 Test #1 - Glass Geometry................................................................131

6.2.5 Test #1 - Glass Thickness ...............................................................132

6.3 Test #1 - Racking Performance of 2x2 Systems......................................134

6.3.1 Test #1 - Discussion of the Parametric Study for 2x2 Systems ......134

6.4 Test #1 - Parametric Study for Multiple Grid Systems............................141

6.4.1 Test #1 - Grid System .....................................................................141

6.4.2 Test #1- Racking Performance of Grid Systems.............................142

6.5 Test #1 - Performance of Rigidly Connected ‘X’ Type Spider Arms .....151

xiv

6.6 Test #2 - Parametric Study.......................................................................158

6.6.1 Test #2 - Sealant Types...................................................................158

6.6.2 Test #2 - Sealant Thickness ............................................................160

6.6.3 Test #2 - Glass Geometry................................................................161

6.6.4 Test #2 - Glass Thickness ...............................................................163

6.7 Test #2 - Racking Performance of 2x2 Systems......................................164

6.7.1 Test #2 - Discussion of the Parametric Study for 2x2 Systems ......164

6.7.2 Test #2 - Grid system......................................................................170

6.8 Design Formulae Developed for Test #1 and Test #2 .............................172

6.8.1 Test #1, 2x2 Systems ......................................................................172

6.8.2 Test #1, Multiple Grid Systems ......................................................172

6.8.3 Test #1 - Rigidly Connected ‘X’ Type Spider Arms ......................173

6.8.4 Test #2, 2x2 and Multiple Grid Systems.........................................173


7 INTER-STOREY DRIFT CALCULATION AND IN-PLANE SEISMIC

DESIGN OF PFGFS..................................................................................................177

7.1 Introduction..............................................................................................177

7.2 Seismic Analysis Methods.......................................................................179

7.2.1 Linear Static Procedures .................................................................179

7.2.2 Non-linear Static Procedures ..........................................................180

7.2.3 Linear Dynamic Procedure .............................................................184

7.2.4 Non-linear Dynamic Procedures .....................................................186

7.3 Calculation of Inter-Storey Drift in Buildings .........................................187

7.3.1 Code Specified Limits.....................................................................187

7.3.2 LDP - RSDmax from Response Spectrum (AS 1170.4) ...................187

7.3.3 LDP – RSDTe from Response Spectrum (AS 1170.4) ....................190

xv

7.3.4 LDP - Response Spectrum Modal Analysis Method (AS 1170.4) .192

7.4 In-plane seismic design of PFGFS...........................................................193

7.5 Recommended Detailing of PFGFS.........................................................197

7.6 Recommended Selection Guide for Façade Engineers ............................200

7.7 Summary and Conclusion ........................................................................202

8 CONCLUSIONS AND RECOMMENDATIONS ............................................203


8.1.1 Research Background .....................................................................203

8.1.2 Experimental Test and Results........................................................204

8.1.3 FE Analytical Model and Results ...................................................206

8.1.4 Seismic Assessment of Façade Systems .........................................210

8.2 Recommendations for Future Research ...................................................211

REFERENCES: ………………………………………………………………….212

xvi

List of Tables

Table 2.1 Test results on behaviour of glass in a storefront FGFS under dynamic

racking conditions (Saflex Solutia Architectural Glazing, 2007) ......................29

Table 2.2 Geometrical and material properties of the single panel PFGFS tested

(Mocibob, 2008) .................................................................................................39

Table 3.1 Relative vertical and horizontal displacement between panels ..................56

Table 3.2 Earthquake drift demand of a 12 storey building on different soil sites.....57

Table 4.1 Details of the LVDTs used in the Test #1 ...................................................73

Table 5.1 Material properties used in the FE model (Test #1) ...................................98

Table 5.2 Material properties used in the FE model (Test #2) .................................115

Table 6.1 Properties of the sealant types used in the FE analysis.............................128

Table 6.2 In-plane drift capacity for the typical panel sizes due the rigid body

translation only at the built in standard gaps ....................................................140

Table 6.3 Drift capacity for typical panel sizes in multiple grid façade systems

(3x3, 4x4 and 5x5) due the rigid body translation at the bolted connections

only (Test #1) ...................................................................................................146

Table 6.4 In-plane drift capacity of PFGFS with rigidly connected ‘X’ type spider

arms ..................................................................................................................157

Table 6.5 Properties of sealant used in the FE analysis............................................158

Table 6.6 Drift capacity for typical panel sizes due the rigid body translation of the

spider arms at the base connections (Test #2) ..................................................170

Table 7.1 Maximum drift demand on façade systems (regular buildings) ...............189

Table 7.2 Maximum drift demand on façade systems (one directional asymmetric

building) ...........................................................................................................189

Table 7.3 Maximum drift demand on façade systems (two directional asymmetric

buildings)..........................................................................................................189

Table 7.4 Maximum drift demand on façade systems (regular buildings) ...............191

xvii

Table 7.5 Summary of the buildings details .............................................................192

Table 7.6 Maximum drift on buildings (Z = 0.10g)..................................................193

Table 7.7 In-plane racking performance of PFGFS with X-type spider arms for

first mode dominant regular buildings (Drift calculated from the RSDmax

method).............................................................................................................196

xviii

List of Figures

Figure 1.1 (a) Assembling of a unitized glass façade system and (b) PFGFS

system supported by truss system.........................................................................2

Figure 1.2 Movement of glass panel within window frame for a glazed window

(Sucuoglu and Vallabhan, 1997) ..........................................................................4

Figure 2.1 The entrance to Apple’s store in Shanghai (Areddy, 2010) ......................10

Figure 2.2 Crack propagation in annealed glass (G. James, 2010).............................12

Figure 2.3 Crack pattern of heat strengthened glass ...................................................13

Figure 2.4 Toughened glass after breakage ................................................................14

Figure 2.5 The interlayer holds the broken glass fragments in laminated glass

(G. James, 2010).................................................................................................15

Figure 2.6 Double glazed façade system (Diytrade, 2011).........................................16

Figure 2.7 Triple glazed façade system (Nourishingobscurity, 2011)........................16

Figure 2.8 Stick curtain wall (Wall-King, 2011) .......................................................18

Figure 2.9 Assembling of a unitized curtain wall system (Wall-King, 2011) ............19

Figure 2.10 Simple post supported PFGFS at a storefront in Melbourne...................20

Figure 2.11 Truss supported PFGFS covering a 4-storey building ............................21

Figure 2.12 Cable supported PFGFS ..........................................................................21

Figure 2.13 Glass fin supported PFGFS at Swinburne University .............................22

Figure 2.14 Kaiser Permanetate Building, Granada Hills, California, cladding

offset NISEE Steinbruggecollection, photo by Mark Aschheim),

(Shahram and Miranda, 2003)............................................................................24

Figure 2.15 FGFS damage was observed in many residential and commercial

buildings and hospitals throughout central Chile following the 2010

magnitude-8.8 Chile Earthquake (Photo courtesy of Eduardo Miranda,

Stanford University), (FEMA E-74, 2011).........................................................24

xix

Figure 2.16 Broken annealed glass that fell several stories from a multistorey

building in the 1994 Northridge Earthquake (Photo courtesy of Wiss, Jenney,

Elstner Associates) (FEMA E-74, 2011)...........................................................25

Figure 2.17 Dynamic racking test setup (Saflex Solutia Architectural Glazing,

2007)...................................................................................................................29

Figure 2.18 Drift time history for AAMA 501.6 dynamic racking crescendo test

(Behr, 2006)........................................................................................................30

Figure 2.19 General glazing details for curtain wall mock-up test (Memari et al.,

2007)...................................................................................................................31

Figure 2.20 Strain gage locations on architectural glass curtain wall mock-up

(Memari et al., 2007) ..........................................................................................31

Figure 2.21 Load-displacement relationship during static (0.01 cm/sec) racking

test (Memari et al., 2007) ...................................................................................32

Figure 2.22 Link 1element, stress-strain relationship (Shirazi, 2005).......................34

Figure 2.23 Pushover curve comparison of experimental and the calibrated finite

element model results (Shirazi, 2005) ................................................................34

Figure 2.24 Large sealant joints required to accommodate thermal movement and

seismic deformations at the California Academy of Sciences, San Francisco,

California (Photos courtesy of Cynthia Perry, BFP Engineers) .........................36

Figure 2.25 (a) Sketch of a button head bolt connection and (b) Sketch of a

countersunk bolt connection (Maniatis, 2006) ...................................................37

Figure 2.26 Cross section of cylindrical countersunk connection tested

(Bernard and Daudeville, 2009) .........................................................................38

Figure 2.27 Schematic diagram of the test set-up (Bernard and Daudeville, 2009)...38

Figure 2.28 (a) Bolt location and (b) bolt connection detail so the specimen

(Mocibob, 2008) .................................................................................................40

Figure 2.29 (a) Test setup for in-plane racking test and (b) Glass connections from

the structural support frame (Mocibob, 2008)....................................................40

xx

Figure 2.30 Crack pattern observed along the laminated heat strengthened glass

panel (Mocibob, 2008) ......................................................................................41

Figure 2.31 Pushover curve from the experimental test (Mocibob, 2008) ................41

Figure 2.32 Spider arms with horizontally slotted holes (Desai et al., 2005).............42

Figure 2.33 In-plane drift performance of the PFGFS expected (Desai et al., 2005).43

Figure 2.34 Swivel countersunk bolt fitting to connect the glass (Desai et al.,

2005)...................................................................................................................43

Figure 2.35 Schematic diagram of the mock-up frame and test specimen with

frame assembly (Desai et al., 2005) ...................................................................44

Figure 2.36 The structural support frame of the San Jose civic centre dome

(Desai et al., 2005)..............................................................................................46

Figure 2.37 (a) Test specimen for San Jose civic centre dome and (b) In-elastic

sealant joint deformation after testing (Desai et al., 2005)................................46

Figure 3.1 Typical layout of unitized framed glass façade system for façade grid of

9800 mm×3600 mm ...........................................................................................51

Figure 3.2 Relative vertical panels movement due to the deflection of spandrel

beam ...................................................................................................................54

Figure 3.3 Deflection of spandrel beam and glazing units ........................................55

Figure 3.4 Schematic diagram of typical point fixed glass façade at ground floor ....58

Figure 3.5 Detail of the proposed glass panel.............................................................59

Figure 3.6 Schematic representation of glass façade articulation in point fixing.

The slotted and larger holes are in the spider arms supporting the glass panel..61

Figure 3.7 Rigid body rotation of glass panels under in-plane lateral loading ...........61

Figure 3.8 Rotated spider arm and relative movement of adjacent glass panels ........62

Figure 3.9 Typical spider arms with slotted holes and large diameter holes ..............62

Figure 3.10 Swivel button head bolt fittings to connect glass and spider arms..........62

Figure 3.11 Guidelines for the holes in toughened safety glass (Viridian, 2010, G.

James, 2010) .......................................................................................................63

xxi

Figure 3.12 Dead and wind forces on the countersunk bolt fitting.............................64

Figure 4.1 X-type spider arm with countersunk bolt fittings (Test #1) ......................67

Figure 4.2 K-type spider arm with button head bolt fittings (Test #2) .......................67

Figure 4.3 Different types of bolt fittings commonly used in Australia.....................68

Figure 4.4 Schematic diagram of the PFGFS in Test #1 ............................................71

Figure 4.5 Structural support frame (blue frame) assembled into the reaction frame

(yellow frame) (Test #1).....................................................................................72

Figure 4.6 Test specimen - glass panels installed and transparent adhesive film

applied (Test #1).................................................................................................72

Figure 4.7 Locations of the LVDTs and the hydraulic jack and the loading bar

attachment with the structural support frame (Test #1)......................................73

Figure 4.8 The measured racking load versus displacement for the system (Test

#1).......................................................................................................................74

Figure 4.9 The system after failure of a glass panel (Test #1)....................................74

Figure 4.10 The broken glass panel after failure with the adhesive film securing

the glass fragments (Test #1)..............................................................................75

Figure 4.11 Glass panels and spider arms to glass bolted connections labelled and

the compression, tension and the spider arm rotational directions indicated

(Test #1)..............................................................................................................75

Figure 4.12 Translations of the glass panels (Test #1) ...............................................76

Figure 4.13 Displacement of the spider arms (to glass bolted connections) in the

vertical direction due to the rotation of the spider arms (Test #1, +ve

movement upward and -ve movement downward) ............................................76

Figure 4.14 Out-of-plane deformation and distortion of the spider arm PBB4 and

PDB2 after failure of a glass panel (Test #1) .....................................................77

Figure 4.15 Differential out-of-plane movement of the spider arms (Test #1) ..........77

Figure 4.16 Test specimen - glass panels installed, transparent adhesive film

applied and Photogrammetry targets attached (Test #2) ....................................81

xxii

Figure 4.17 Visible built-in standard gap at the spider arm base to cleat connection

(Test #2)..............................................................................................................81

Figure 4.18 (a) Slotted holes in the spider arm base plate and (b) Large bolt holes

at the cleat (Test #2) ...........................................................................................82

Figure 4.19 Racking load versus displacement for the PFGFS (Test #2)...................82

Figure 4.20 The system after failure of a glass panel (Test #2)..................................83

Figure 4.21 Glass panels and spider arms connections labelled and the

compression, tension and the spider arm sliding directions indicated (Test

#2).......................................................................................................................83

Figure 4.22 Translation of the glass panels at the bolt holes (Test #2) ......................84

Figure 4.23 Displacement of the spider arms in the vertical direction (Test #2)........84

Figure 4.24 (a) Initial position of the internal centre spider arm and (b) Relative

vertical sliding of the spider arms after failure (Test #2) ...................................85

Figure 4.25 Differential out-of-plane movement of the spider arms (Test #2) ..........85

Figure 4.26 Out-of-plane deformation of the spider arm PAB3 and PCB1 (Test #2) 86

Figure 4.27 (a) Deformed spider arm PCB3 due to compression and (b) Deformed

spider arm and base plate PDB4 due to tension (Test #2)..................................86

Figure 4.28 Schematic diagram of a typical glass failure or crack origin (Castilone

et al., 2002) .........................................................................................................87

Figure 4.29 Glass fracture originated from the bolted connection PCB4 (Test #2) ...87

Figure 5.1 Diagonal strut mechanism and load transfer through the spider arms

(Test #1)..............................................................................................................91

Figure 5.2 Differential movement of the spider arms in the out-of-plane direction

(Test #1)..............................................................................................................92

Figure 5.3 Glass panels and spider arm connections labelled along with the

racking mechanism (Test #1) .............................................................................92

Figure 5.4 Schematic diagram of countersunk bolt connection (Test #1) ..................93

Figure 5.5 FE modelling of the structural support frame with spider arms (Test #1) 96

xxiii

Figure 5.6 Non-linear springs (green) connecting spider arms (purple) countersunk

bolt fittings (red) whilst non-linear springs (green) connect spider arms to

structural support frame (blue) (Test #1)............................................................96

Figure 5.7 Mesh of bolt heads, glass panel and sealant at the internal centre spider

arm in the FE model (Test #1)............................................................................97

Figure 5.8 ANSYS FE full model (Test #1) ..............................................................97

Figure 5.9 Cross section of the silicon sealant (Test #1) ............................................99

Figure 5.10 Test ANSYS FE model of silicon sealant (Test #1)...............................99

Figure 5.11 Tensile and shear load displacement results for 8mm thick silicone

sealant from ANSYS FE model (Test #1)........................................................100

Figure 5.12 Compression simulation results for 8mm thick silicone sealant from

ANSYS FE model (Test #1).............................................................................100

Figure 5.13 X-type spider arm with built-in standard gaps indicated (Test #1) ......102

Figure 5.14 Real constants (spring properties) used for the rigid body translation

and bearing at the connections (Test #1)..........................................................103

Figure 5.15 Real constants (spring properties) used for the rigid body rotation

about the ‘z’ axis at the spider arm to structural support frame connections

(Test #1)............................................................................................................103

Figure 5.16 Boundary conditions at the central spider arms (Test #1).....................104

Figure 5.17 Translation of the glass panels (Test #1)...............................................106

Figure 5.18 Out-of-plane movement (in mm) of the glass panels (Test #1).............106

Figure 5.19 Maximum principle tensile stress (in MPa) developed - front face

(Test #1)............................................................................................................107

Figure 5.20 Maximum principle tensile stress developed (in MPa) - back face

(Test #1)............................................................................................................107

Figure 5.21 In-plane rotation of the spider arms (in radians) at failure (Test #1) ....108

Figure 5.22 Experimental and analytical pushover curves (Test #1)........................108

Figure 5.23 Maximum tensile stress developed at Bolt PCB4 (Test #1)..................109

xxiv

Figure 5.24 Contact elements assigned to connect bolt head to glass bolt hole

(Test #1)............................................................................................................110

Figure 5.25 Glass panels and spider arms configuration including the sliding

directions of the spider arms bases (Test #2) ...................................................112

Figure 5.26 Schematic diagram of button head bolt connection used in the FE

model (Test #2).................................................................................................113

Figure 5.27 FE modelling of the structural support frame with K-type spider arms

(Test #2)............................................................................................................114

Figure 5.28 ANSYS finite element model (Test #2) ..............................................115

Figure 5.29 Real constants used for the vertical sliding(‘y’ direction) of the spider

arms and bearing at the structural support frame to the spider arm base plate

connections (Test #2) ....................................................................................116

Figure 5.30 Real constants used for the out-of-plane movement (‘z’ direction) of

the spider arms and bearing at the structural support frame to the spider arm

base plate connections (Test #2).......................................................................117

Figure 5.31 Locations leading to gaps in button head bolt fitting (Test #2).............117

Figure 5.32 The deformed model after reaching the failure stress at 4.75% drift

(Test #2)............................................................................................................119

Figure 5.33 Out-of-plane movement (in mm) of the glass panels at 4.75% drift

(Test #2)............................................................................................................119

Figure 5.34 Spider arms deformation and vertical translation (in mm) in the spider

arms at 4.75% drift (Test #2)............................................................................120

Figure 5.35 Maximum principle tensile stress (in MPa) developed at 4.75% drift

front face of the glass panels (Test #2).............................................................120

Figure 5.36 Maximum principle tensile stress (in MPa) at 4.75% drift - back face

of the glass panels (Test #2) .............................................................................121

Figure 5.37 Experimental and analytical pushover curve benchmarked (Test #2)...121

Figure 5.38 Maximum tensile stress developed at the glass hole PCB4 (Test #2)...122

xxv

Figure 6.1 Schematic diagram of the holes provided at the structural support

frame (Test #1) .................................................................................................125

Figure 6.2 Real constants (non-linear spring constant) used for the spring elements

to represent the translation and bearing at the spider arm to structural support

frame connection ..............................................................................................126

Figure 6.3 Analytical pushover curve comparison for the models with circular hole

and slotted hole at the structural support frame (Test #1) ................................126

Figure 6.4 Comparison of the tensile stresses developed at the FE model with

circular hole and slotted hole at the structural support frame (Test #1) ...........127

Figure 6.5 Analytical pushover curve comparison of low, medium and high

modulus silicon sealants (Test #1) ...................................................................128

Figure 6.6 Comparison of the tensile stresses developed in the FE models with

low, medium and high modulus silicon sealants (Test #1) ..............................129

Figure 6.7 Analytical pushover curve comparison with 6mm, 8mm and 10mm

thick silicon weather sealants (Test #1)............................................................130

Figure 6.8 Comparison of the tensile stresses developed in the FE models with

6mm, 8mm and 10mm thick silicon weather sealants (Test #1 ) .....................130

Figure 6.9 Analytical pushover curve comparison of the square, portrait and

landscape panel systems (Test #1) ...................................................................131

Figure 6.10 Comparison of the tensile stress developed at the square, portrait and

landscape panel systems (Test #1) ...................................................................132

Figure 6.11 Analytical pushover curve comparison for 10mm, 12mm and 15mm

thick glass panels (Test #1) ..............................................................................133

Figure 6.12 Comparison of the tensile stresses developed for 10mm, 12mm and

15mm thick glass panels (Test #1) ...................................................................133

Figure 6.13 Analytical pushover curve comparison for the Test #1 with and

without spider rotation restrained.....................................................................135

Figure 6.14 Comparison of the tensile stresses developed for the Test #1 with and

without spider rotation restrained.....................................................................135

xxvi

Figure 6.15 Analytical pushover curve comparison of Test #1 FE and Test #1 FE

rigid body translation only................................................................................138

Figure 6.16 Comparison of the tensile stress developed at the Test #1 FE and Test

#1 FE rigid body translation only.....................................................................138

Figure 6.17 Built-in standard gaps provided collectively at the spider arms to bolt

fitting connections (Test #1 FE Chapter 5) ......................................................139

Figure 6.18 Orientation of the spider arms (Same as Test #1) .................................139

Figure 6.19 Frame in parallelogram action under racking load (Test #1) ................140

Figure 6.20 Analytical pushover curve comparison for 2x2, 3x3 and 4x4 systems

(Test #1)............................................................................................................141

Figure 6.21 Comparison of the tensile stresses developed comparison for 2x2, 3x3

and 4x4 systems (Test #1) ................................................................................142

Figure 6.22 All the spider arms orientated diagonally for a multiple façade grid

system (Test #1) ...............................................................................................143

Figure 6.23 The structural support frame in the modified FE model with the spider

arms diagonally orientated (Test #1) ................................................................144

Figure 6.24 Analytical pushover curve comparison from the Test #1 FE multiple

grid façade system with the rigid body translation only from the built-in

standard gaps ....................................................................................................145

Figure 6.25 Comparison of the tensile stress developed for the Test #1 FE multiple

grid façade system with the rigid body translation only from the built-in

standard gaps ....................................................................................................145

Figure 6.26 Built-in standard gaps provided at the bolted connections (Test #1) ....147

Figure 6.27 Analytical pushover curve comparison of multiple grid façade system

from the rigid body translation at the bolt fittings, Test #1 FE ±3mm rigid

body translation with ±3mm and ±7mm rigid body translation .......................148

Figure 6.28 Comparison of the tensile stress developed for the multiple grid

façade system from the rigid body translation at the bolt fittings, Test #1 FE

±3mm rigid body translation with ±3mm and ±7mm rigid body translation ...148

xxvii

Figure 6.29 Built-in standard gaps provided at the bolt fittings and the structural

support frame (Test #1) ....................................................................................150

Figure 6.30 Analytical pushover curve comparison of multiple grid façade

system, Test #1 FE from the ’rigid body translation at the bolt fittings’ with

‘rigid body translation only at the bolt fittings and structural support frame’ .150

Figure 6.31 Comparison of the tensile stress developed at the multiple grid façade

system, Test #1 FE from the ‘rigid body translation at the bolt fittings’ with

from the ‘rigid body translation at the bolt fittings and structural support

frame’ ...............................................................................................................151

Figure 6.32 Glass panels connected to the structural support frame without spider

arms (a) Geometry of the gaps in the glass panels to transfer loads and (b)

Translation of the glass panel under in-plane loading......................................152

Figure 6.33 Glass panels connected to the structural support frame with

horizontally orientated spider arms (a) Geometry of the holes to transfer

loads with spider arms and (b) Translation of the glass panel under in-plane

loading with spider arms ..................................................................................153


arms (a) Geometry of case study example, (b) Translation of the glass panel

under in-plane loading......................................................................................155

Figure 6.35 Glass panels connected to the structural support frame with

horizontally orientated spider arms (a) Geometry of the case study example,

(b) Translation of the glass panel under in-plane loading ................................155

Figure 6.36 Schematic diagram of typical PFGFS with rigidly connected ‘X’ type

spider arms with articulation holes...................................................................157


modulus silicon sealants (Test #2) ...................................................................159

Figure 6.38 Comparison of the tensile stress developed at the FE models with FE

low, medium and high modulus silicon sealant (Test #2) ................................159

Figure 6.39 Analytical pushover curve compared with 6mm, 8mm and 10mm

thick silicon weather sealant (Test #2) .............................................................160

xxviii

Figure 6.40 Comparison of the tensile stress developed at the FE model with

6mm, 8mm and 10mm thick silicon weather sealant (Test #2)........................161

Figure 6.41 Analytical pushover curve comparison with square, portrait and

landscape glass panels (Test #2).......................................................................162

Figure 6.42 Comparison of the tensile stress developed at the FE models with

square, portrait and landscape glass panels (Test #2).......................................162

Figure 6.43 Analytical pushover curve comparison of 10, 12 and 15mm thick glass

panels (Test #2) ................................................................................................163

Figure 6.44 Comparison of the tensile stresses developed on 10, 12 and 15mm

thick glass panels (Test #2) ..............................................................................164

Figure 6.45 Spider arm vertical translation due to the rigid body translation at the

spider arms base connections (Test #2)............................................................167

Figure 6.46 Rocking mechanism of the glass panels under in-plane racking load

(Test #2)............................................................................................................167


rigid body spider arm vertical translation at the base connections...................168


#2 FE rigid body spider arm vertical translation at the base connections ........168


rigid body translation at the bolt fittings built-in standard gaps.......................169

Figure 6.50 Comparison of the tensile stress developed at Test #2 FE and Test #2

FE rigid body translation at bolt fittings built-in standard gaps.......................169

Figure 6.51 Analytical pushover curve compared for the 2x2, 3x3 and 4x4 systems

(Test #2)............................................................................................................171

Figure 6.52 Comparison of the tensile stress developed at the FE 2x2, 3x3 and 4x4

systems (Test #2)..............................................................................................171

Figure 7.1 Schematic diagram of a building sway under earthquake ground motion

(Su et al., 2008) ................................................................................................177

Figure 7.2 Typical capacity spectrum (Wilson and Lam, 2003)...............................183

xxix

Figure 7.3 Typical 500 years RP acceleration response spectrum for different soil

sites (A to E) in Australia for Z = 0.08, (AS1170.4, 2007) ..............................185

Figure 7.4 Typical 500 years RP displacement response spectrum for different of

soil sites (A to E) in Australia for Z = 0.08 (AS1170.4, 2007) ........................185

Figure 7.5 Displacement for effective stiffness of a building from the ADRS

diagram.............................................................................................................190

Figure 7.6 Front view of a storefront PFGFS with ‘X’ type spider arms in

Melbourne, Australia ........................................................................................198

Figure 7.7 Side view of a storefront PFGFS with ‘X’ type spider arms in

Melbourne, Australia (the perimeter glass panels are free to move)................199

Figure 7.8 Typical PFGFS with the perimeter glass panels sealed to the building

using structural sealant in Melbourne, Australia ..............................................199


using sealant and a two way spider arm used to align the glass panels............200

Figure 7.10 PFGFS recommendations to increase the drift capacity........................201

xxx

Chapter 1 INTRODUCTION AND OVERVIEW

Chapter 1

1 INTRODUCTION AND OVERVIEW

1.1 Introduction

Glass façade systems (GFS) have gained popularity in recent times and are commonly

found in all types of commercial, industrial, institutional and residential buildings.

The GFS have significant impact on the building aesthetics and provide the interface

between the internal and external environments. The design of GFS covers aesthetic

considerations, weather proofing and structural design. The structural design of GFS

normally takes into account in-plane and out-of-plane loading from wind, thermal

movement and deflection from supporting structural elements due to gravity loads and

creep.

Conventionally, the GFS are available in three construction forms namely; stick

systems, semi-unitized systems and unitized systems. The unitized GFS is a more

contemporary framing method which comprises a glass vision panel and spandrel

panel mounted in a prefabricated aluminium frame and illustrated as a complete unit

in Figure 1.1a. Alternatively, a new contemporary frameless glazed façade system is

available which provides transparency and improved aesthetics, known as point fixed

or bolt fixed glass façade systems (PFGFS).

PFGFS are often connected with bolts to steel structural support frames, (which are

exposed architectural elements) to combine structural stability with aesthetic

expression. A typical PFGFS supported by trusses is shown in Figure 1.1b. Therefore,

the GFS can be classified depending on the structural support as either framed glass

façade systems (FGFS) or frameless (point fixed) glass façade systems (PFGFS).

1


(a) (b)

Figure 1.1 (a) Assembling of a unitized glass façade system and (b) PFGFS system

supported by truss system

The GFS may be subject to in-plane racking action due to the relative lateral

movement of the building from extreme events such as earthquakes. The performance

of façade system is dependent on the amount of in-plane drift and the interaction of

the glass panel with the façade structural support frames. There are two major

concerns related to GFS performance during and immediately after a seismic event

(Saflex Solutia Architectural Glazing, 2007):

• Hazards to people from falling glass. This may cause injuries at street level

from broken storefront and elevated glazed panels.

• Building down time and cost to repair. Bringing a building back to operation

can be delayed by a breached building envelope due to glazed façade systems

damage.

GFS damage from earthquakes has historically been reported along with general non-

structural damage. However, due to the significant usage of glass in buildings in

recent decades, increasing emphasis has been placed on glass damage observations in

earthquakes reconnaissance reports (Sucuoglu and Vallabhan, 1997). Sakamoto

(1978) reports that many broken window glass panels (FGFS) were observed in the

1964 Niigata and 1968 Tokochi-oki earthquakes in Japan, especially in flexible

2


structures where nine of the 72 buildings investigated suffered glass damage including

one building with 120 broken glass panels, but no other damage.

From damage observations a strong correlation was observed between the inter-storey

drift and glass damage, indicating that the in-plane deformation response is the

dominant cause of damage to window glass (Evans and Kenett, 1988). Reitherman

and Sabol (1995) discuss that, in the 1994 Northridge earthquake, glazing damage

was extensive and the principal cause of glass door and window failures was the

inadequate edge clearances around the glass to allow the building to deform laterally

without bearing on the glass.

Therefore, it is evident that in past earthquakes, GFS with sufficient clearance

between edges of the glass panel and the supporting structures have performed well.

The performance of fixed windows and storefront glass façade systems has been

tested in laboratories over the past few decades. Researchers have suggested

improvements such as clearance between glass to frame, addition of rounded corners

around each glass panel and adoption of more robust glass types such as heat

strengthened, toughened and laminated glasses (Behr, 2006).

Bouwkamp (1960) observed that the in-plane deformation of window panels (FGFS)

under lateral loading takes place in two phases, as shown schematically in Figure 1.2.

First, the window frame deforms and the glass panel translates within the frame until

contact occurs at two opposite corners of the glass panel (Figure 1.2b). The glass

panel then further rotates until the opposite corners coincide with the adjacent frame

corners and diagonal compressive strut action develops in the glass (Figure 1.2c).

Sucuoglu and Vallabhan (1997) found that the total lateral deformation of the window

panel due to rigid body motion of the glass panel in the window frame can be

expressed in terms of the geometric properties of window panel components as:

∆= 2c 1 + Eq (1.1)

Where Δ is the lateral drift capacity of the glass frame and c, h and b are physical

dimensions as defined in Figure 1.2.

3


Figure 1.2 Movement of glass panel within window frame for a glazed window

(Sucuoglu and Vallabhan, 1997)

ASCE 7-02 (2010) provides a general expression for assessing the FGFS under in-

plane loading as expressed by Equation 1.2. The drift capacity (Δfallout) needs to be

greater than the drift demand which is a function of relative seismic displacement (Dp)

and the occupancy importance factor (I)

Δ ≥1.25ID or13mm whichever is greater Eq (1.2) fallout p

Exceptions are recommended by (ASCE7-10, 2010) for FGFS with sufficient glass-

to-frame clearance such that physical contact between the glass and frame will not

occur at the design drift demonstrated by Equation 1.3 which is an extended version

of Equation 1.1.

⎛ h c ⎞D ≥ 1.25D ; and D ≥ 2c ⎜1+

p 2 ⎟ Eq (1.3)

clear p clear 1 ⎜ ⎟b c ⎝ p 1 ⎠

Where hp = height of rectangular glass; bp = width of rectangular glass, c1 = clearance

(gap) between the vertical glass edges and the frame; and c2 = clearance (gap)

between the horizontal glass edges and the frame.

The Standard for earthquake actions in Australia, AS 1170.4 (2007), limits the inter-

storey drift to 1.5% in buildings and states that, the “attachment of cladding and

façade panels to the seismic-force-resisting system shall have sufficient deformation

and rotational capacity”. However, the seismic drift performance of glass façades is

generally not considered at the design stage by façade engineers. The Australian

4


Standard “Glass in buildings-Selection and installation” AS 1288 (2006), provides

guidance for the strength and serviceability design of glass subject to out-of-plane

wind loading but does not comment on in-plane actions.

Recently Baird et al. (2011) summarised the façade technology in New Zealand and

made some conceptual steps towards the performance based seismic design of façade

systems. The researchers classified the PFGFS (spider glazing) under a separate

façade typology for the seismic performance considering the interaction with the

building frame. The seismic performance of PFGFS is likely to be quite different from

conventional FGFS where previous studies have been undertaken. McBean (2008)

commented that the in-plane drift capacity of PFGFS is at least half of the drift

capacity of FGFS. There is very limited published research available on the

behaviour of PFGFS under in-plane actions due to earthquake loading and a testing

methodology and rational analytical work is required to assess the drift performance

of such systems.

1.2 Project Aims, Objectives and Methodology

The overall aim of this thesis is to assess the performance of PFGFS under in-plane

racking action mainly due to seismic load. A number of objectives were developed

along with appropriate methodology to achieve the overall aim. The objectives and

research methodologies adopted are summarised below.

1.2.1 Critical Literature Review

The objectives of the literature review are to examine studies related to: (1) glass

façade systems and glass types; (2) in-plane racking performance of glass façade

systems; (3) previous research on glass façade systems; (4) standard provisions for

design of GFS against racking actions and inter-storey drift limit in buildings; and (5)

review of the structural design and analysis of GFS including both framed and point

fixed glass façade systems (FGFS and PFGFS). The findings of these reviews provide

a clear foundation for the research to be undertaken in this project.

1.2.2 Experimental Testing of Façade Systems

Experimental tests were vital in this project to assess the in-plane racking

performance of PFGFS under in-plane racking loading including both glass and

5


connections. Therefore, two unique full scale laboratory in-plane racking tests on

typical PFGFS currently in practice were conducted at the Francis Laboratory,

University of Melbourne.

1.2.3 Analytical Modelling and Parametric Study

Detailed three-dimensional non-linear finite element models (FE model) were

developed to replicate the laboratory tests and benchmarked against the experimental

test results. The FE models were then modified and used to estimate the drift

contribution from different mechanisms namely; rigid body translation, spider arm

rotation and spider arm deformation. Further, the validated FE models were utilised to

estimate the in-plane racking capacity of PFGFS with various parameters including,

built in standard gap, sealant type, sealant thickness, glass geometry, glass thickness.

1.2.4 Estimation of Inter-Storey Drift Demand in Buildings

The in-plane racking performance of glass façade systems is dependent on the drift

demand imposed on the façade structural support frame. Therefore, it is necessary to

calculate the drift demand from the buildings imposed on façade systems to design the

GFS. A detailed review of the expected inter-story drift of the buildings in low-to

moderate regions was conducted. Some rapid inter-storey drift assessment methods

are described with example calculations.

1.2.5 In-plane Seismic Assessment and Design of PFGFS

The performance of PFGFS to in-plane racking is discussed by comparing the in-

plane drift demand from the buildings with the in-plane drift capacity in PFGFS.

Detailing approaches to improve the in-plane racking performance of PFGFS are

recommended. Additional built-in gaps and special articulation features are suggested

to improve the racking performance of PFGFS.

1.3 Thesis Overview

The literature review covering the project background is presented in Chapter 2. The

Chapter reviews; glass façade systems and technology, in-plane seismic performance

of glass façade systems, previous research on in-plane seismic performance of glass

façade systems and standard provisions for seismic design of glass façade systems.

6


In Chapter 3, an extensive review of the methodology for the contemporary structural

design of GFS, in Australia is presented along with analytical techniques. The design

and analysis of both FGFS and PFGFS are described with example calculations. The

GFS are designed for in-plane and out-of-plane load and movements. Self-weight,

thermal expansion, spandrel beam deflection and building movements due to wind

and seismic loads are considered for in-plane design whilst wind load on the glass

panel, mullion, transom and structural support frames are considered for out-of-plane

design.

Two unique full scale in-plane racking laboratory tests (Test #1 and Test #2) of

typical PFGFS were conducted. The test setup and the results are described in Chapter

4 along with the discussion of racking mechanism of PFGFS. Chapter 5 describes the

three-dimensional non-linear finite element models (FE models) developed to

replicate the laboratory tests and benchmarked against the test results for both

laboratory tests. The results obtained from the FE models were benchmarked against

the test data including, pushover curve, failure stress and out-of-plane deformation of

glass panels.

A parametric study was conducted by modifying the FE models to calculate the

racking performance of different PFGFS as described in Chapter 6. The FE models

were used to estimate the drift contribution from different mechanisms namely; rigid

body translation, spider arm rotation and spider arm deformation. Further, the

validated FE models were utilised to estimate the in-plane racking capacity of PFGFS

with various parameters including; size of glass panels, type of silicon sealant and

grid number of glass panels. Additional drift capacity can be obtained by introducing

articulation features at the bolted connections at the glass or structural support frame.

Special articulation features are introduced with example calculations for PFGFS with

rigidly connected ‘X’ type spider arms.

In Chapter 7, seismic assessment methods for calculating inter-storey drift demands

are reviewed. Rapid assessment methods are presented for estimating the seismic drift

demand of glass façade systems with increasing levels of sophistication and accuracy.

Applications of these methods are illustrated with example calculations and

conservative factors are adopted for considering the torsional behaviour of buildings.

7


Further, in-plane seismic design of PFGFS is discussed with the examples. Seismic

detailing to improve the racking performance of PFGFS is discussed with existing

projects in Melbourne. In addition, a quick selection guide is presented for façade

engineers to select the structural components of PFGFS which can increase the

racking performance of the system.

Finally, significant conclusions from the research are presented in Chapter 8 together

with recommendations for further research.

8

Chapter 2 RESEARCH BACKGROUND

Chapter 2

2 RESEARCH BACKGROUND

2.1 Introduction

Façade systems play an important role in building construction since they provide the

interface between internal and external environment of the building and are used to

supply sufficient light and air quality to improve the indoor environment. Façades

have had a number of major transitional phases throughout the history. Presently, a

drive towards performance based design can be observed with a strong emphasis on

sustainable development. A number of novel technologies have been suggested such

as double façade systems and the use of solar technology to enable buildings to

produce electrical power. Façade systems will continue to develop and become more

interactive with the external environment whilst key factors regarding aesthetics, cost,

performance, durability, maintainability and sustainability will continue to influence

the design. Lawrence Berkeley National Laboratory (2006) states that the façade

systems should provide the following:

• Enhanced sun protection and cooling while improving thermal comfort and

providing most of the light needed with day lighting;

• Enhanced air quality and reduced cooling loads using natural ventilation

schemes employing the façade as an active air control element;

• Reduced operating costs by minimizing lighting, cooling and heating energy

use by optimizing the day lighting-thermal trade-offs; and

• Improved indoor environments leading to enhanced occupant health, comfort

and performance.

The façade system can be classified into different types based on the material used

namely; metal (aluminium, stainless steel), timber (treated), masonry façade (brick

and terracotta) and glass and natural stone (glass, sandstone and limestone) (Emporis,

2007). However, the popularity of glass façade systems (GFS) is increasing due to a

number of facts including glass technology development, aesthetics, increased natural

9


light and sustainability considerations. Today, a series of developments can be seen

which enhances glass to perform as a changeable building material to fulfil several

functions simultaneously. A few examples mentioned by Knaack et al. (2007) are

listed below:

• Electro-chromatic coating: daylight and radiation transmission can be altered

by applying varying levels of voltage;

• Thin film cells: photovoltatic (PV) cells in the form of screens deposited on

glass generate energy-patterns that can be imprinted by laser with some areas

remaining transparent;

• Phase changeable material (PCM) in glass: can provide thermal storage;

• Holographic coating (films): can provide shading independent of the angle of

the solar radiation (transparent) or focus energy; and

• Heated glass: to balance heat loss and increase surface temperatures (no

comfort reduction through radiation or cold air during winter).

In addition, the development of structural glass with larger dimensions has enabled

architects and façade engineers to design load bearing structural glass façade systems.

An example of such system is the Apple glass temple (Figure 2.1) in Shanghai, China

which consists of 62, ultra-clear, tempered, scratch-, stain- and bubble-free glass

panels, six layers thick, 12.5 m tall and 2.5 m wide (Areddy, 2010).

Figure 2.1 The entrance to Apple’s store in Shanghai (Areddy, 2010)

10


2.2 Glass Types

Annealed glass, heat-strengthened glass, toughened glass or fully tempered glass,

laminated glass and insulating glass are the most common types of glass available.

2.2.1 Annealed Glass

Annealed or float glass is the glass that has been cooled gradually from a high

temperature during manufacture to minimize residual stress, allowing the glass to be

cut by scoring and snapping. It is the most commonly available type of flat glass. This

is one of the weakest glass types and has a significant potential to break when

subjected to excessive loads or when installed incorrectly. On breakage, the glass

tends to form sharp-edged, pointed shards (Figure 2.2). The sharp edges may cause

cutting injuries, while the pointed shards may cause piercing injuries. The post failure

behaviour of the glass will be dominated by its lack of residual strength on breakage.

The glass may not be able to resist loads potentially causing:

• Full or partial collapse of the glass structure;

• Penetration of the glass structure; and

• Glass fragments or shards to fall when used at height.

It is for these reasons that monolithic annealed glass is not used as a highly stressed

structural glass. Annealed glass may be safe if suitably laminated but its low strength

reduces its scope as a structural material. The strength of glass is stochastic and

limited by the presence of random defects. The failure stress of an annealed glass is

also dependent on the duration of load. Annealed glass can however be processed into

other glass types and products, which are stronger, have safer breakage

characteristics, and improved post-failure characteristics. These glasses commonly

include heat-strengthened glass, toughened fully tempered glass, and laminated glass.

11


Figure 2.2 Crack propagation in annealed glass (G. James, 2010)

2.2.2 Heat-Strengthened Glass

Heat-strengthened glass has undergoes a controlled heating and cooling process to

improve the resistance to mechanical and thermal stresses. In this process, a

permanent compressive surface stress and a permanent tensile internal stress are

simultaneously induced in the glass (Haldimann et al., 2008). The compressive

surface stresses gives the glass a bending strength higher than that of annealed glass

and reduce the likelihood of glass failure since the surface prestress has to be

overcome before a bending failure can occur.

Typical crack pattern of heat strengthened glass is shown in Figure 2.3. The failure

stress varies widely in a similar fashion to annealed glass. The residual compressive

surface stress of the heat strengthen glass is in the range 24–69 N/mm2

(AS1288,

2006). The failure stress will depend on the condition of the glass surface, load

duration, environment, and the degree of tempering. Heat-strengthened glass breaks

with failure characteristics similar to annealed glass.

12


Figure 2.3 Crack pattern of heat strengthened glass

2.2.3 Toughened Glass or Fully Tempered Glass

Toughened glass is processed in the same way as heat strengthened glass but is cooled

more rapidly. Float glass is heated to 620-675 ºC in a furnace and then cooled rapidly

(Haldimann et al., 2008). The resulting compressive surface stress is higher in

magnitude than that in heat-strengthened glass. This gives the glass a bending strength

higher than that of heat-strengthened glass. The high strength of the glass means it is

far less likely to fail than either annealed or heat strengthened glass types when

subjected to mechanical or thermal stresses. As with heat strengthened glass the

surface prestress has to be overcome before a bending failure can occur. Again the

failure stress is the sum of the failure stress of the untreated glass and the residual

surface stress. The residual compressive surface stress of toughened glass is over 69

N/mm2

(AS1288, 2006). In the event of breakage, toughened glass generally breaks

into small, relatively blunt, glass fragments which do not have sharp edges and are

unlikely to cause deep cutting injuries (Figure 2.4).

13


Figure 2.4 Toughened glass after breakage

2.2.4 Laminated Glass

Laminated glass is an assembly consisting of glass sheets joined together with one or

more plastic or resin interlayer. The glass assembly can have performance advantages

over monolithic glass, which may make it the most appropriate choice for some

applications. The strength, breakage characteristics, and post failure behaviour of

laminated glass depend upon the glass types, glass thicknesses, and interlayer types

and thicknesses used in construction. Thin flexible sheet materials used as an

interlayer are commonly known as foils. For structural applications, laminates are

normally composed of:

• Toughened glass to provide sufficient strength to resist applied loads; and

• Heat strengthened or annealed glass to govern the post failure behaviour.

Film interlayer in structural laminated glass, are normally polyvinyl butyral (PVB) but

thicker sheet interlayer materials are available including ionoplast, which is stronger

and produces a stiffer laminated glass (Chaszar, 2003). In addition, the interlayer may

need to have sufficient adhesion to hold broken glass fragments on failure and have

sufficient strength to resist tearing should failure occur (Figure 2.5). Further

consideration should be given to the possibility that the plastic interlayer may creep or

deform at elevated temperatures over time if relatively high temperatures are likely to

be encountered during normal building use. In such circumstances the laminate may

14


sag and move out of its frame or fixing. A number of interlayer types are available

with different properties that may make them more appropriate for particular

applications (Norville, 1999).

Figure 2.5 The interlayer holds the broken glass fragments in laminated glass

(G. James, 2010)

Glass laminates may also incorporate plastic sheets, such as polycarbonate and

polymethylmethacrylate, to give spalling resistance in the event of glass breakage.

Such laminates have been used in bullet-resistant glazing and overhead glazing. In

external applications the edge of laminated glass may need to be protected from the

weather or moisture to prevent de-lamination. In addition, any sealants that could

come into contact with the laminate edge should be compatible to prevent

delamination. Small areas of delamination will not normally affect strength.

2.2.5 Insulating Glass

Glass façade systems are assembled using single or multiple glass panels (insulating

glass units). Multiple glass panels can be two units (double glazed, Figure 2.6) or

three panels (triple glazed, Figure 2.7) assembled together. In double glazing two

glass panels are installed into a frame with a gap between them creating a sealed unit.

They are separated by a spacer filled with silica balls which serve to extract moisture

from the air. There are many benefits of double glazing installations versus single

glazing. The double layer of glass prevents much of the heat escaping and reduces

noise levels. Double glazing can be more efficient if the gap between the two panels is

slightly larger, and if argon is used to fill the space instead of air. Argon has a lower

15


convection than oxygen and nitrogen found in the air. Triple glazing introduces a third

piece of glass creating a second sealed unit into the façade system. It provides even

greater energy saving and increased sound insulation, and as such is generally

installed in extreme weather conditions and at highly noisy locations; however, triple

glazing inevitably costs and weighs more than double glazing.

Figure 2.6 Double glazed façade system (Diytrade, 2011)

Figure 2.7 Triple glazed façade system (Nourishingobscurity, 2011)

16


Thermal and acoustic performance of glass can be modified using low emissivity

coating, tinting and coating. Low emissivity glass has a thin coating (often of metal)

on the glass that reflects thermal radiation or inhibits emission, thereby reducing heat

transfer through the glass. A basic low emissivity coating allows solar radiation to

pass through the room; thus, reducing heat loss but allowing the room to be warmed

by direct sunshine. While clear glass is the most common glass component of

insulating glass units, tinted glass may be used to reduce solar heat gain or as an

architectural feature. The heat and sound insulation of glazing may also be improved

by the use of a film or coating applied to the surface. This film is typically made of

polyester or metal, and may give the window a reflective mirrored appearance.

2.3 Glass Façade Systems

The glass façade system (GFS) can be classified depending on the structural support

as either a framed glass façade system or a frameless glass façade system.

2.3.1 Framed Glass Façade Systems

Framed glass façade systems (FGFS) are typically designed with extruded aluminium

members, although the early curtain walls were made of steel. The aluminium frame

is typically in-filled with glass, which provides an architecturally pleasing envelope,

as well as benefits such as natural day lighting. These curtain walls are designed to

span multiple floors, and to accommodate design requirements such as, thermal

expansion and contraction, building sway and movement, water proofing, and thermal

efficiency for cost-effective heating, cooling, and lighting in the building. There are

three common types of aluminium glass façade systems available consisting of stick,

semi-unitized and unitized system.

The stick wall system, shown in Figure 2.8 is the older method of curtain wall

technology and is installed piece by piece. Usually, the mullion members (vertical

members) are installed first, followed by the transom members (horizontal rail

members), and finally the glazing or window units. The stick wall system was used

extensively in the early years of metal curtain wall technology, and is still widely used

in a modified form. The advantages of this component system are low shipping and

handling costs, and ease of dimensional adjustment according to site conditions. On

17


the other hand, the major assembling process is conducted on site and a large amount

of labour support is needed.

Figure 2.8 Stick curtain wall (Wall-King, 2011)

The semi-unitized curtain wall system is a development on the stick wall system,

where mullion members are separately installed and then pre-assembled framing units

are placed between them. These units can be up to a full storey height, or they can be

divided into spandrel units (covering the floor slab and beam) and vision glass units.

This system also needs a significant labour force site assembling work and the

erection time is considerably greater than fully unitized curtain wall.

The unitized curtain wall system is a more contemporary construction system and

consists of sheets of glass and aluminium panels fabricated and installed as a unit as

shown in Figure 2.9. Generally, a unitized glazed curtain wall comprises glass vision

panel and spandrel panel mounted in a prefabricated aluminium frame. The

production of the whole panel is completed at the factory with a high degree of

quality control and production efficiency. The structural aluminium profile around the

panel is fabricated as half sections (female and male) which provide built-in tolerance

to accommodate adjustment during erection and movement during service. The panels

are installed in shingle fashion, starting either from the bottom or top of the building

and moving around the perimeter of each floor until the whole building is complete.

18


Figure 2.9 Assembling of a unitized curtain wall system (Wall-King, 2011)

2.3.2 Frameless Glass Façade System

A growing number of architects are substituting the design of framed glass façade

systems with frameless glass façade systems. These contemporary glass walls can

allow more natural light to enter the buildings. This has a particular advantage in

architectural expression in terms of transparency and having a planar glass surface by

the removal of the mullions and aluminium profiles. Double glazing also can be

utilized in this method to improve energy efficiency by increasing the insulating

capacity.

Modern frameless glass façade systems are often bolt connected using spider arms

steel structural support frames, which are important architectural elements and

combine structure stability with aesthetic expression. The bolted spider arms

connections provide point support to the glass panels. Applications of point fixed

glass façade systems (PFGFS) range from simple structures such as store front shop

windows to more complex multi-storey buildings and large atrium. The bolted fixings

are commonly located towards the corners of the glass panels and additionally at

intermediate points on long edges. Although there are different types of point fixed

glass support systems available such as truss systems, cable systems and steel

supported systems, all consists of four basic components; glass panels, bolted fixings,

glass support attachments (spider arms) and the main structural support frame (Ryan

et al., 1997):

19


• Point fixed glass on base supported steelwork such as simple posts, trusses and

fins (Figure 2.10 and 2.11).

• Point fixed glass on cable systems where the structural support frames is

constructed almost entirely from tension elements, such as rods or wires, and

are therefore very light both physically and visually. Loads have to be

transferred, at both ends of the cables, to boundary structural support frames.

The weight of the vertical glazing is either supported by a tie rod hanger

system or by each panel being suspended from the above panel (Figure 2.12).

• Fin walls comprising one-way spanning glazing supported on glass beams or

fins: The glazing is attached intermittently using bolted connections as shown

in Figure 2.13.

Figure 2.10 Simple post supported PFGFS at a storefront in Melbourne

20


Figure 2.11 Truss supported PFGFS covering a 4-storey building

Figure 2.12 Cable supported PFGFS

21


Figure 2.13 Glass fin supported PFGFS at Swinburne University

2.4 Damage to Glass Façade Systems in Past Earthquakes

Glass damage from earthquakes has historically been reported along with general

non-structural damage. However, due to the significant usage of glass in buildings in

recent decades, increasing emphasis has been placed on glass damage observations in

earthquake damage reports (Sucuoglu and Vallabhan, 1997). Sakamoto (1978) reports

that many broken window glass panels were observed in the 1964-Niigata and 1968

Tokochi-oki earthquakes in Japan, especially in flexible structures where nine of the

72 buildings investigated suffered glass damage including one building with 120

broken glass panels, but no other damage.

In the 1971 San Fernando earthquake, several buildings were reported with glass

damage, with the amount of damage correlated to the degree of resilience of the

glazing sealants (Ayres and Sun, 1973). The off-Miyagi earthquake in February 1978

and the 1983 Mid-Japan Sea earthquake caused considerable glass breakage

(Sakamoto et al., 1984) and the extent of non-structural damage, (mainly glass

damage) reported in these two earthquakes was greater than the Niigata and Tokochi

oki earthquakes. The 1985 Mexico City earthquake was well investigated from the

point of view of glass damage. Seven of the 263 investigated multistorey office

22


buildings sustained minor to moderate structural damage, while over 50% of

investigated buildings received some sort of glass damage and 25% were classified as

having serious glass damage. In this earthquake, broken glass was reported as the

second most serious non-structural damage, but less than the damage to infill walls.

As expected strong correlation is observed between the inter-storey drift and glass

damage, indicating that the in-plane deformation response is the dominant cause of

damage to window glass (Evans and Kenett, 1988). Flexible glass façade systems

(metal curtain walls and mullions) caused aggravated glass damage due to drift

whereas glass panels enclosed by more rigid glass façade systems (precast curtain

walls, or glass panels set in the structure) experienced less damage. Further, it is

reported that glass damage consistently increases with larger window areas and

irregular plan configurations. Glass panels used in the shopfront windows of single

storey or low rise commercial buildings were observed to be extremely vulnerable to

seismic excitations in all recent US earthquakes (Behr et al., 1995).

Reitherman and Sabol (1995) discuss that, in the 1994 Northridge earthquake, glass

façade damage was extensive and the principal cause of glass door and window

failures was the inadequate edge clearances around the glass to allow the building to

deform laterally without bearing on the glass. In some cases glazing damage was so

severe that the supporting metal frames buckled. Low-rise buildings had incorporated

annealed glass (rather than tempered, wired, or laminated glass required for taller

buildings) produced sharp-edged pieces that could have caused serious injuries. In

contrast film-coated windows performed well. An industry survey after the

earthquake revealed that glass façade incorporating silicon sealant performed better

than glass façade with vinyl gaskets (Harter, 1994, Vallabhan, 1994). Systems

equipped with Mylar film to provide seismic protection from sharp glass debris

performed very well in the case of small window glass panels, but proved less

effective for larger window panels, where the entire panel was dislodged and fell as

one big piece (Gates and McGavin, 1998).

More recently the City of Bam, Iran was hit by an earthquake in 2003 destroying 70%

of the buildings in the stricken area, causing extensive non-structural damage to the

buildings that remained structurally intact. The observed cases of non-structural

23


24

damage were mainly architectural façades and damage to false ceilings, glass

finishing such as windows, door glass, parapets and other attachments was also

observed (Hosseinia, 2005). Few examples of damage to FGFS are shown in Figures

2.14, 2.15 & 2.16.

Figure 2.14 Kaiser Permanetate Building, Granada Hills, California, cladding offset

NISEE Steinbruggecollection, photo by Mark Aschheim),

(Shahram and Miranda, 2003)

Figure 2.15 FGFS damage was observed in many residential and commercial buildings

and hospitals throughout central Chile following the 2010 magnitude-8.8 Chile Earthquake

(Photo courtesy of Eduardo Miranda, Stanford University), (FEMA E-74, 2011).


Figure 2.16 Broken annealed glass that fell several stories from a multistorey building in

the 1994 Northridge Earthquake (Photo courtesy of Wiss, Jenney, Elstner Associates)

(FEMA E-74, 2011)

2.5 Previous Research on Framed Glass Façade Systems

2.5.1 Experimental Study

A substantial number of laboratory and analytical studies related to the simulated

seismic performance of architectural glazing within framed glass façade systems have

been performed over the past few decades. The following includes a summary of

previous research findings. Most researchers adopted horizontal, in-plane dynamic

racking tests to assess the behaviour and performance of framed glass façade systems

in buildings during earthquakes (Sakamoto, 1978, Sakamoto et al., 1984, Behr et al.,

1995b).

25


Thurston and King (1992) focused on assessing the behaviour of curtain wall glazing

system when subjected to simulated inter-storey drift as may be expected to occur

during the response of multi-storey buildings to earthquake attack. Both one-

dimensional (planar) and two-dimensional (corner) specimens were tested. Glazing

systems investigated consisted of, neoprene gasket dry-glazed systems, unitized 4

sided structural silicone glazed systems, a two-sided silicone glazed system and

mechanically fixed patch plate systems (with toughened glass). Failure was

considered to occur when more than 5% of the mass of glass supported in any frame

fell from that frame. The test procedure involved cyclically displacing the floor beam

(sliding steel beams) to a designed displacement (steadily increasing towards peak

displacement).

Sucuoglu and Vallabhan (1997) focused on the behaviour of window glass panels

during earthquakes and developed analytical techniques to determine the dynamic

response of window glass and structural glazing systems by using simple mechanical

models. Based on the derived expressions, a simple practical procedure was

developed for the design of glass panels that would sustain the effects of earthquakes.

The dynamic behaviour of window glass panels subjected to earthquake excitation

was investigated by assuming two mechanisms consisting of rigid body translation

and elastic deformation. The rigid body translation depends on the clearance between

glass panel and the window frame, and resilience of the sealant material whilst the

elastic deformation includes diagonal shortening of the glass plate.

Behr (1998) conducted a four year research programme at the Building Envelope

Research Laboratory at the University of Missouri-Rolla (UMR) to investigate the

serviceability and fallout resistance of various types of architectural glass and related

glazing systems under simulated earthquake conditions. In 1996 Behr performed

“crescendo tests” on various types of architectural glass commonly employed in a

popular curtain wall system for mid-rise buildings. The first crescendo tests

performed at UMR on storefront glass included a lower ultimate limit state,

corresponding to the drift required to form a major crack pattern in the glass, and an

upper ultimate limit state, corresponding to the drift required to cause glass fallout.

While performing the subsequent mid-rise glass tests, he modified the definitions of

seismic drift limits for architectural glass to be as follows:

26


• A “serviceability drift limit” corresponding to the drift required to cause

observable glass cracking (a condition that would necessitate glass

replacement, but would not pose an immediate life safety hazard); and

• An “ultimate drift limit” corresponding to the drift required to cause glass

fallout (a condition that would pose a life safety hazard to building occupants

and pedestrians).

“Crescendo tests” (Behr, 2001) revealed distinct and repeatable drift limits related to

glass cracking and glass fallout for various types of architectural glass tested in a

representative storefront wall system and a representative mid-rise curtain wall

system. Demonstrable differences in seismic resistance exist between various types of

architectural glass commonly employed in building design. Wall system stiffness,

glass-to-frame glazing details such as the gasket properties and glass-to-frame edge

clearances are also significant parameters in relation to the seismic performance of

architectural glass. Behr (2001) found notable differences in seismic resistance exist

between architectural glass types commonly used in contemporary building design,

with annealed and heat strengthened laminated glass units showing the highest levels

of resistance to glass fallout.

Memari et al. (2003) carried out in-plane dynamic racking crescendo tests on full-

scale curtain walls dry glazed with six different insulating glass unit (IGU)

configurations and one laminated glass unit configuration. All IGU configurations

tested were manufactured with an annealed monolithic pane and a laminated pane

with an argon fill and an anodized aluminium spacer between the panes. Several

parameters were varied in the laminated pane of each configuration including glass

pane thickness and glass type in the laminated pane (annealed, heat strengthened, and

fully tempered), and PVB interlayer thickness for the laminated pane. Properties of

the annealed inside pane were not varied. The test result showed that IGUs performed

well.

Memari et al. (2004) conducted a pilot study at Pennsylvania State University to

investigate the response of curtain wall mock-ups glazed with 6 mm annealed

monolithic architectural glass panels fitted with applied film under simulated

earthquake conditions. Three common film-to-frame anchoring methods were

27


evaluated. These preliminary tests indicated that anchorage type can noticeably

influence both the serviceability and ultimate limit states of filmed glass panels.

Behr et al. (2003) devised and developed the concept of using architectural glass

panels with modified corner geometries to improve resistance to damage during

earthquakes. The primary aspects of the innovations was the removal of material at

glass panel corners (e.g., by rounding the glass corners) and subsequent finishing of

the glass edges in the modified corner regions to minimize protrusions and edge

surface roughness. The rounded corner glass panels with polished edges showed much

higher resistances to glass cracking and glass fallout as compared to counterparts with

square shaped sharp edges.

Saflex Solutia Architectural Glazing (2007) commissioned studies and participated in

cooperative efforts with universities and the U.S. National Science Foundation to

investigate glazing system performance in seismic events. By using the dynamic

crescendo tests they concluded that, laminated glass tends to remain in openings (the

laminate holds the broken glass) when broken for any reason such as wind, hail, wind-

borne debris, bomb blast, accidental impact, intentional impact and during seismic

activity.

A number of racking tests on similar storefront FGFS with different types of glass

was conducted by Saflex Solutia Architectural Glazing (2007). Glass panels

measuring 1524 mm x 1829 mm x 6mm were configured as single glazing, laminated

glazing (two layers of glass) and double glazing (insulting glass unit) with a 12 mm

air space and tested. Test results on the behaviour of glass in a storefront FGFS under

dynamic racking conditions are compared in Table 2.1. Test results revealed,

however, that fully toughened laminated glass panels were not as advantageous in

seismic applications when dry glazed since the unit tends to fold and fall like a heavy

blanket if both plies are broken during racking motions. The impressive performance

of annealed and heat strengthened laminated glass units in the BERL (Building

Engineering Research Laboratory) tests represents a promising step toward the

development of seismic-resistant glass façade systems similar to the Behr (2001)

“Crescendo tests” results.

28


Table 2.1 Test results on behaviour of glass in a storefront FGFS under dynamic

racking conditions (Saflex Solutia Architectural Glazing, 2007)

Drift at observable crack Drift at glass fallout Glass type

on glass (%) (%)

Annealed (Monolithic Glass) 4.3 5.8

Fully Tempered (Monolithic Glass) 5.4 5.4

Annealed (Laminated Glass, Single) 6.5 8.2

Annealed (Insulating Glass Unit) 6.5 7.5

Fully Tempered (Insulating Glass Unit) 7.1 7.1

Behr (2006) developed a set of seismic design provisions for architectural glazing.

These proposed design provisions were modified through several standards and

finally adopted in a slightly modified format by ASCE 7-02 (ASCE7-10, 2010). The

seismic test details for glazed frames are provided in ASCE 7-02 whilst the dynamic

loading test setup and the loading sequence are shown in Figure 2.17 and 2.16

respectively.

6.4 m Steel Column

Steel Column

Bracing

Tie rods B Sliding steel tube Reaction frame

Pivot arm Fulcrum

B Typical glazed

curtain wall

Typical glazed 4 m curtain wall panel (infill) panel

U-jointA Electrohydraulic

servoactuator ram

A

Connection pin

Figure 2.17 Dynamic racking test setup (Saflex Solutia Architectural Glazing, 2007)

29


Figure 2.18 Drift time history for AAMA 501.6 dynamic racking crescendo test

(Behr, 2006)

2.5.2 Analytical Study

The development of code provisions and test standards addressing a perceived

problem leads to increased attention toward the development of analytical procedures.

However, only limited published literature exists investigating the seismic

performance of framed glass façade systems related to finite element analysis. Shirazi

(2005) and Memari et al. (2007) carried out an analytical study on a FGFS and

compared the results with the experimental results. The glass façade shown in Figure

2.19 considered in the analysis with strain gauges mounted at selected locations on the

glass and the aluminium framing was subject to static racking loading as shown in

Figure 2.20.

30


Figure 2.19 General glazing details for curtain wall mock-up test (Memari et al.,

2007)

Figure 2.20 Strain gage locations on architectural glass curtain wall mock-up

(Memari et al., 2007)

The load-displacement curve obtained from the test is shown in Figure 2.21. The

frame displaced 20mm (1.0% drift) before the glass panel made initial contact with

the frame at the top left corner. The racking displacement then increased to 50mm

(2.5% drift) before the glass to frame contact was made at the bottom right corner,

31


effectively creating a diagonal compression strut mechanism along a diagonal of the

glass panel. The racking force and system stiffness increased significantly, until a

small crack developed at the top left corner at a displacement of 60mm (3.0% drift),

followed by a large crack at 70mm (3.5% drift), when the racking load reduced

dramatically and the test was stopped.

Lo

ad (

kN

)

20

18

16

14

12

10

8

6

4

2

0

G lass-to-fram e contact at top left corner

Start of uniform gasket-to-glass friction

G lass-to-fram e contact at bottom right corner

Sm all crack (bottom right corner)

Sm all crack (top left corner)

Large crack form s

0 1 2 3 4 5 6 7 8

1

2

3

4

5

6

Displacement (cm)

Figure 2.21 Load-displacement relationship during static (0.01 cm/sec) racking test

(Memari et al., 2007)

The sources of the generated stress in a glass panel were the rubber friction between

the gasket and the glass panel edge and the direct contact between the corners of the

glass panel to the frame. The boundary condition of the glass panel changed under

increasing magnitude of applied drift. Initially, a gap exists between the glass panel

and the frame all around the glass panel (except at the locations where the edges

contact the setting blocks and side blocks). Setting blocks were used to cushion the

glass and create the clearance between the glass panel and the frame. Side blocks and

mullions provided the clearance between the vertical edges of the glass panel. By

increasing the applied drift, the gap is overcome at the corners located at both sides of

the shorted diagonal of the deformed frame due an increase in the applied drift.

Finally the glass panel contacts the frame and the glass experiences contact forces as

well as friction.

32


A finite element analysis was then undertaken to replicate the load-deflection

behaviour and the localised strain measurement recorded from the experimental tests.

The main objective of the finite element model was to simulate the strains and the

stresses generated in the glass panel due to a specific applied drift to a framed glass

façade system. In this study, the ANSYS finite element analysis program was used to

develop the model using several element types. Four-node shell elements (SHELL 63)

with six degrees of freedom per node were used to model the glass panel. Aluminium

frame members were modelled using two-node beam elements (BEAM 3) with 2-D

translation degrees of freedom and 1-D rotational degree of freedom per node. The

transom to mullion semi-rigid connection was modelled by the change in the bending

and the shear stiffness of the corresponding connection element. The friction effect in

the gasket will be modelled by the use of elasto-plastic link elements.

In order to model the friction effect, the nodes on the frame elements were connected

to the adjacent nodes on the edges of the glass panel by Link 1 elements. Link 1 is a

2-D uniaxial tension-compression element with two degrees of freedom at each node.

As in a pin jointed structure, no bending of the element is considered. If the link

yields and reaches a constant stress after a small deflection then it can simulate the

rubber friction effect that does not change over the increment of the applied drift to

the curtain wall. Figure 2.22 depicts a general view of the stress-strain of the link

element used. Four node shell elements (SHELL 63) with six degrees of freedom per

node are used to model the setting and side blocks. The material property of the

blocks was assumed to be elasto-plastic. A bilinear stress-strain relationship was

specified for the modules of elasticity of setting blocks.

In order to simulate the contact between the glass panel edge and the frame a Link 10

element was used to connect the boundary nodes at the glass panel edges to the

adjacent nodes on the frame. Link 10 is a 3-D spar element having the unique feature

of a bilinear stiffness matrix resulting in a uniaxial compression-only (or tension-

only) element. Link 10 has three translational degrees of freedom at each node in the

‘x’, ‘y’, and ‘z’ directions. No bending stiffness is included in either the tension-only

(cable) option or the compression-only (gap) option. By assigning a positive initial

strain, a gap can be modelled and the element will not be activated until the gap is

overcome. By the use of a combination of Link 1 and Link 10 (compression only

33


option) elements between the glass panel edge nodes and frame nodes, the rubber

friction effect as well as the contact effect were simulated.

The racking displacement was applied to the top corner of the frame and the drift

increments were applied in small steps. The finite element model was calibrated

against the pushover curve and the ultimate strain associated with the glass failure.

The load-deflection behaviour of the finite element model was in good agreement

with the experimental result as shown in Figure 2.23. It was concluded that the finite

element model developed could be used to model the framed glass facades with

different configurations. However, the finite element model should be calibrated by

the use of mock-up test to obtain reliable predictive results.

σ

ε εο

σο

Figure 2.22 Link 1element, stress-strain relationship (Shirazi, 2005)

Figure 2.23 Pushover curve comparison of experimental and the calibrated finite

element model results (Shirazi, 2005)

34


2.5.3 Standard Provisions for Framed Glass Façades

ASCE 7-10 (2010) provides a general expression for assessing architectural glass

under in-plane loading as expressed by Equation 2.1. The drift capacity (Δfallout) is to

be greater than the drift demand which is a function of relative seismic displacement

(Dp) and the occupancy importance factor (I).

Δ ≥1.25ID or13mm whichever is greater Eq (2.1) fallout p

Exceptions are recommended by (ASCE7-10, 2010) for framed glass façades with

sufficient glass-to-frame clearance such that physical contact between the glass and

frame will not occur at the design drift demonstrated by Equation 2.2.

⎛ h c ⎞D ≥ 1.25D ; and D ≥ 2c ⎜1+

p 2 ⎟ Eq (2.2)

clear p clear 1 ⎜ ⎟b c ⎝ p 1 ⎠

Where hp = height of rectangular glass; bp = width of rectangular glass, c1 = clearance

(gap) between the vertical glass edges and the frame; and c2 = clearance (gap)

between the horizontal glass edges and the frame. The mechanism of contact between

the glass and frame is explained by (Sucuoglu and Vallabhan, 1997), considering rigid

body translation and rotation of the glass panel. In framed glass façade system design,

if the exceptions are not satisfied (drift capacity calculated according to Equation 2.2),

mock-up tests can be carried out to evaluate the Δfallout of the glass panels. American

Architectural Manufacturers Association (AAMA) specifies the laboratory test

methods for both static and dynamic testing to evaluate the in-plane drift capacity of

framed glass façade systems (AAMA, 2001a, 2001b).

“Reducing the Risks of Non-structural Earthquake Damage – A Practical Guide”

recently prepared by Applied Technology Council (FEMA E-74, 2011) explains about

the typical causes of damage to FGFS and with example damages interior and exterior

FGFS from the past earthquakes. However, the practical guide (FEMA E-74, 2011)

does not provide guidance on the in-plane racking performance of the PFGFS other

than an illustration with exceptionally large sealant joints to accommodate the seismic

induced movements as shown in Figure 2.24.

35


Figure 2.24 Large sealant joints required to accommodate thermal movement and seismic

deformations at the California Academy of Sciences, San Francisco, California (Photos

courtesy of Cynthia Perry, BFP Engineers)

2.6 Limited Number of Previous Research on PFGFS

Experimental tests and analytical studies have been performed on structural glass

panels and connections to evaluate the strength of components used in PFGFS and to

asses the shear behaviour of glass panels.

2.6.1 Influence of Bushing Type in Load Bearing Capacity

Thermoplastics and aluminium materials are commonly used as bushing material in

glass point fixed applications and the application of bushings is illustrated in Figure

2.25. Experimental investigations and finite element models were carried out to

examine the influence of different parameters such as different bushing materials,

clearance between bolt and drilled hole, eccentric loading, variation in hole diameter

and variation of distance between hole and panel edge. Maniatis (2006) found that the

stress development around the glass bolt hole is not influenced by the different

material combinations of bushings. However, hole diameter, clearance between bolt

to glass hole and glass and distance between hole to the edge influenced the stress

development. Maniatis (2006) found that increasing the clearance between bolt to

glass hole enhances the stress development.

36


(a)

(b)

Figure 2.25 (a) Sketch of a button head bolt connection and (b) Sketch of a

countersunk bolt connection (Maniatis, 2006)

2.6.2 Influence of Connection Type in Load Bearing Capacity

Bernard and Daudeville (2009) performed a series of tensile tests on countersunk

bolted glass connections to predict the ultimate in-plane load capacity of annealed and

tempered glass panels 350 x 600mm in dimension. Three geometries of countersunk

bolt connections were tested in the range of 24-56mm of internal and external

diameters respectively. One of the tested hole geometries is shown in Figure 2.26 and

the schematic diagram of the test setup is shown in Figure 2.27. The ultimate in-plane

tensile load capacity of the 19mm thick annealed glass was found to be in the range of

12 -24 kN and the capacity of the 19mm thick tempered glass was in the range of 62

118 kN.

37


φ

φ

Figure 2.26 Cross section of cylindrical countersunk connection tested

(Bernard and Daudeville, 2009)

Figure 2.27 Schematic diagram of the test set-up (Bernard and Daudeville, 2009)

2.6.3 In-plane Load Capacity of a PFGFS

In a fully transparent single storey pavilion structure free of any steel or concrete

frame, glass panels are used as unique vertical structural elements (Mocibob, 2008).

In this application, the glass panel is supported by bolted connections on the two sides

(roof and foundation) and subjected to out-of-plane wind load, in-plane shear force

due to lateral wind and in-plane compressive force due to dead load of the roof.

Mocibob (2008) and Mocibob and Belis (2010) also conducted an experimental

investigation and numerical simulation on connection devices in order to understand

the load bearing capacity of different types of glass panels with bolted connections.

38


Tests on full size two layered laminated heat strengthened glass panels measuring

1200 x 3500mm were conducted in order to understand the shear buckling behaviour

of point fixed glass panels. The thickness of each glass panel was 8mm and the

thickness of the PVB interlayer was 1.52mm. The geometry of the glass panel and the

material properties are summarised in Table 2.2.

The experimental setup and details of the supports are shown in Figure 2.28 whilst the

crack pattern for the laminated heat strengthened glass panel under pure in-plane

shear load at ultimate limit state is shown in Figure 2.29. Cracks propagated along the

diagonal compressive loading direction due to splitting tensile stress (Figure 2.30).

The ultimate in-plane drift capacity of the glass panel under 27.5 kN in-plane shear

load was approximately 0.2% (Figure 2.31) and the out-of-plane deflection due to

buckling at the mid span was around 4mm (Mocibob, 2008). However, the connection

details of these tests were not representative of a typical PFGFS in practice.

Table 2.2 Geometrical and material properties of the single panel PFGFS tested

(Mocibob, 2008)

Geometrical properties (mm) Material properties

Glass panel height a 3500 Glass modulus of

elasticity

E

(N/mm2)

70,000

Glass panel width b 1200 Glass Poisson ratio ν 0.23

Glass panel thickness t 8/1.52/8 PVB shear modulus (N/mm2) 0.50

Adhesive width cA 40

Adhesive thickness tA 9.5

Setting block width dA 100

Setting position b/5 240

39


(a) (b)

Figure 2.28 (a) Bolt location and (b) bolt connection detail so the specimen

(Mocibob, 2008)

(a) (b)

Figure 2.29 (a) Test setup for in-plane racking test and (b) Glass connections from

the structural support frame (Mocibob, 2008)

40


Figure 2.30 Crack pattern observed along the laminated heat strengthened glass

panel (Mocibob, 2008)

Figure 2.31 Pushover curve from the experimental test (Mocibob, 2008)

41


2.6.4 High Displacement Seismic Glass Systems

In PFGFS, the bolt connections are snug-tightened to allow movements and the

weather sealant between the glass panels is low in stiffness and has high strain

capacity. Therefore, by introducing special articulations into the bolted connections,

the in-plane drift capacity could be increased to satisfy the seismic drift demands.

This form of articulation technology is very common in the precast concrete façade

industry (PCI, 1989) and has recently been adopted in PFGFS by Desai et al. (2005)

and Gowda and Heydari (2009) in California.

Desai et al. (2005) and Gowda and Heydari (2009) assessed the seismic performance

of PFGFS with different types of structural support frames for use as cladding facades

of buildings in areas of high seismicity. To address the criteria of a drift limit of 2.0%

to 2.5% for cladding systems as per the seismic provisions of the California Building

Code (CBC, 2002) a series of mock-up tests were conducted. The function of the

proposed systems is to isolate the glass from the structural support frame for in-plane

deformations and loads while supporting the system for vertical loads and for out-of

plane loads. In PFGFS, spider arms are used to connect the glass to structural support

frames. Specially designed spider arms with large horizontally slotted holes (Figure

2.32) were used to accommodate the drift by allowing isolated horizontal translation

as shown in Figure 2.33. The sizes of the slotted holes were calculated according to

the height of the glass façade and the drift demand from the building.

Figure 2.32 Spider arms with horizontally slotted holes (Desai et al., 2005)

42


Figure 2.33 In-plane drift performance of the PFGFS expected (Desai et al., 2005)

Swivel countersunk bolt fittings were used to connect the glass to the spider arms to

minimize the bending stress concentrations around the holes. A ball and socket at this

connection point prevents transmission of bending stresses in the glass (Figure 2.34).

Together, the bolts and spiders act like a pin connection for out-of-plane loads, and a

roller connection for in-plane loads. Bushes for the bolts were inserted in the bottom

slots of each spider arm to carry the glass dead load (Figure 2.32). Omitting the

bushings in the top slots eliminated vertical support at that location, allowing the glass

to expand due to thermal movements without inducing additional stresses in the glass.

Figure 2.34 Swivel countersunk bolt fitting to connect the glass (Desai et al., 2005)

43


The point supported glass panels were heat-treated to resist local stresses at the holes,

and low modulus silicone sealant was used to create flexible joints that allowed the

glass panels to translate. For the first test simple tubular steel sections (RHS) were

used as the structural support frame as shown in Figure 2.35. A series of mock-up

tests were conducted on PFGFS with a frame structural support frame measuring

7.6m (25 feet) wide and 6.1m (20 feet) high. The wall was subjected to in-plane

displacement in incremental magnitudes consisting of three cycles of elastic drift

condition (0.4%), one cycle of inelastic drift condition (2.5%) and a third drift

capacity of (2.9%).

Figure 2.35 Schematic diagram of the mock-up frame and test specimen with frame

assembly (Desai et al., 2005)

In general, the system behaved very well and met all code criteria with all of the glass

panels remaining fully intact. The authors concluded that, as anticipated the glass

would translate horizontally and distribute the drift proportionally over the height of

the wall without breaking the glass panels. The authors also concluded that the

silicone sealant would withstand the shear demand imposed without rupturing, and

without failing any other component of the system. Only when the frame was racked

beyond the design range of the spider slots (overload drift), did the glass show

noticeable rotation even though no glass or bolt failures were observed. The results

and findings were applied to large scale mock up tests and conducted with different

types of structural support frame for the following real projects:

44


a) San Jose Civic Centre Rotunda in downtown San Jose, California which is a

27m diameter 12m high dome structure fully enclosed by cable supported

PFGFS.

b) New Belmont police department and city hall in Belmont, California which

consists of series of glass fin supports providing out-of-plane support to the

PFGFS.

A large scale mock up test, representing a portion of the project, was conducted for

the San Jose dome with all of the finalized components of the system. The structural

support frame of the PFGFS was a horizontal cable truss system tensioned between a

vertical steel rib structures with additional vertical cables supporting the dead loads as

shown in Figure 2.36. The spider arm was pin connected to the spreader bars of the

cable truss and the dead load carrying cables were clamped to the front of the spider

arms. The test specimen for the San Jose civic centre dome is shown in Figure 2.37a.

During the lateral drift proof test, the wall was subjected to a 2.0% drift in each

direction for a total of three cycles, returning to the original position after each cycle

without any damage to components and weather seals. The mock-up tests

demonstrated that the system with spider arms connected to the vertical structural

support frame members performed satisfactorily. However, the San Jose project test

applied the lateral displacement using a horizontal structural support frame which is

more favourable than the in-situ system. During lateral translation of the in-situ

system in a simulated seismic event, the spider arms connected to the vertical

structural support frame follow the inclination of the structural support frame

eventually inducing a rotation into the spider arms, which will be transferred through

the bolt fittings into the glass panels. In contrast, the test setup with a horizontal

structural support frame eliminates this rotation, since all components translate

parallel to the horizontal joints and hence reduce the stresses on the glass supports.

The in-elastic joint deformation is shown in Figure 2.37b which shows the horizontal

translation.

45


Figure 2.36 The structural support frame of the San Jose civic centre dome

(Desai et al., 2005)

(a) (b)

Figure 2.37 (a) Test specimen for San Jose civic centre dome and (b) In-elastic

sealant joint deformation after testing (Desai et al., 2005)

2.7 Codified In-Plane Drift Demands on Façade Systems

Drift provisions in Standards are recommended for serviceability and ultimate limit

states. “Structural design actions”, AS/NZS 1170.0 (2002) provides out-of-plane and

in-plane serviceability limit state criteria for building elements. The Standard

recommends an in-plane maximum drift limit of H/600 (0.17%) for the brittle

masonry wall (where H is the height of the wall) but no limits are specified for glazed

façade systems.

46


The Australian Standard “Concrete structures”, AS 3600 (2009) specifies in clause

2.4.3, “unbraced frames and multi-storey buildings subject to lateral loading shall be

designed to limit calculated inter-storey lateral drift to 1/500 of the storey height”.

This is aimed at the serviceability limit state of the building mainly for wind loading.

Whilst the Standard for “Steel structures” AS 4100 (1998) recommends compliance

with AS 1170.4 (2007).

The Council on Tall Buildings, Group SB (1979), examined the serviceability wind

drift criteria from industry and literature and found that drift limits ranging from

0.001H to 0.004H (0.1%-0.4%) were used. However the Council states that buildings

designed in the past have been known to perform satisfactorily when designed for

drift limits from 0.002H to 0.005H (0.2%-0.5%). ASCE Task Committee found that

most of the design for institutional, commercial, and residential building types used

drift ratios in the order of 0.002H to 0.0025H (0.2%-0.25%) for steel framed

buildings.

AS 1170.4 (2007), clauses 5.4.4 and 5.5.4, specify that, “the inter-storey drift at the

ultimate limit state, calculated from the forces determined according to strength and

stability provisions shall not exceed 1.5% of the storey height for any level and “the

attachment of cladding and façade panels to the seismic-force-resisting system shall

have sufficient deformation and rotational capacity”. This requirement is for the

ultimate limit state of the building for seismic performance and results in a 54mm

deflection for a typical 3600 mm storey height.

The New Zealand Standard “Earthquake actions”, NZS 1170.5 (2004) specifies in

clause 7.5 that a maximum inter-storey drift limit of 2.5 % is applicable for the

ultimate limit state of 500 year RP event. In the case of a 2500 year RP near fault

event, this limit is increased to 3.75%. Drift limits of 2.5% and 3.75% create demands

of 90 mm and 135 mm respectively on façade systems, assuming a storey height of

3600 mm.

The 2001 Edition of the California Building Code (CBC, 2002), is the basis for the

design methodology of glass façade systems in California. The seismic provisions of

the CBC require that façade systems be designed to accommodate a maximum

inelastic drift of 2.0% to 2.5% of the building height for Seismic Zone 4.

47


The Australian Standard “Glass in buildings-Selection and Installation” AS 1288

(2006), provides guidance for the strength and serviceability design of glass subject to

out-of-plane wind loading but does not comment on in-plane effects. From

discussions with local industry experts, the glass façades are normally designed for

maximum in-plane inter-storey drift of height/500 for serviceability conditions.

2.8 Conclusion and Summary

Façade systems play an important role in building construction since they provide the

interface between internal and external environment of the building and are used to

supply sufficient light and air quality to improve the indoor environment. The façade

system design involves a sequence of steps including visual consideration, weather

proofing and structural evaluation. Visual assessment covers the overall aesthetics

whilst weather proofing includes air leakage control, vapour diffusion control, heat

loss and gain control and rain water penetration control.

Structurally the GFS are designed for in-plane and out-of-plane load and movements.

Self-weight, thermal expansion, spandrel beam deflection and in-plane building

movements due to wind and seismic loads are considered for in-plane design whilst

the wind load on the glass panel, mullion, transom and structural support frames are

considered for out-of-plane design. From a structural point of view, GFS can be

classified into either framed or point fixed. The conventional method of glazing is

framed; however the current international trend is towards the installation of PFGFS

for increased transparency and improved aesthetics.

From the literature review it has been observed that the damage to GFS resulting from

earthquakes is increasingly common and yet there has been limited number of

laboratory tests and detailed analyses undertaken. The research conducted to date has

focused on traditional FGFS. Researchers have suggested improvements such as

providing a larger gap between glass to frame and adoption of more robust glass types

namely; heat strengthened, toughened and laminated glass. International standard

provisions are available to calculate the in-plane racking capacity and to design

against seismic actions for FGFS whilst static and dynamic testing methods are

available to evaluate the racking performance.

48


Despite its growing popularity, there is very limited published research on the

behaviour of the PFGFS under the in-plane earthquake loading. Seismic performance

of PFGFS is likely to be quite different from conventional FGFS. A very limited

number of experimental tests and analytical studies have been conducted to date to

estimate the in-plane load capacity of standard glass panels with bolted connections.

The available test results showed that typical bolted connections and glass panels have

significant in-plane tensile and compressive load capacity.

However, a series of mock-up racking tests were conducted in California to adopt the

PFGFS in higher seismic regions. The drift capacity of the PFGFS was increased to

2.9% by adopting large horizontally slotted holes in the spider arms and using special

purpose low modulus silicon sealant to allow rigid body translation during the racking

actions. The inter-storey drift demand on the GFS depends on the seismic region and

the building characteristics.

Very limited analytical studies have been conducted to-date to predict the racking

capacity of GFS which is further complicated by the uncertainty in the failure stress of

the glass panel. Therefore, developed analytical models were benchmarked against the

test results only and conservatively the racking capacity of the GFS was limited to the

rigid body translation and rotation only. This was consistent with the racking tests in

California where the drift capacity was estimated from the rigid body translation only

with the test not extended to evaluate the ultimate limit state of the system.

It is clear that the in-plane racking performance of typical PFGFS is not well

established for contemporary systems. In addition, guidelines to define a common

testing protocol have not been available. Similarly, standard analytical techniques for

analysing PFGFS have not been developed or published in the literature. This

research project will develop a standard testing protocol and use the protocol to

undertake racking tests on typical contemporary PFGFS that have been developed for

low to moderate seismic regions such as Australia. The test results will then be used

to calibrate finite element models that can then be extended for other PFGFS systems

and geometric configurations.

49

Chapter 3 STRUCTURAL ANALYSIS AND DESIGN OF GFS

Chapter 3

3 STRUCTURAL ANALYSIS AND DESIGN OF GFS

3.1 Introduction

Before undertaking the experimental investigation into the in-plane racking

performance of PFGFS, some background on the current design and analytical

approach to GFS needs to be investigated to provide a context for the testing. This

chapter develops a design guideline for GFS based on current industry practice and

literature including industry experts and journal articles. Interestingly, a GFS design

guide does not exist and this chapter provides the basis for a guideline which has been

summarised as a journal paper (Sivanerupan et al., 2011) to disseminate to

engineering professionals.

The glass façade systems design involves a sequence of steps including visual

consideration, weather proofing and structural evaluation. Visual assessment covers

the overall aesthetics whilst weather proofing includes air leakage control, vapour

diffusion control, heat loss and gain control and rain water penetration control.

Structurally the curtain wall is designed for in-plane and out-of-plane load and

movements. Self-weight, thermal expansion, spandrel beam deflection and in-plane

building movements due to wind and seismic loads are considered for in-plane design

whilst wind load on the glass panel, mullion, transom and structural support frames

are considered for out-of-plane design.

The size and the profile of the glass façade systems are normally specified by the

architect whilst the structural design is undertaken by façade engineers. Interestingly,

there are very few published guidelines on structural design for engineers especially

for the design of point fixed glass façade systems. The objective of this Chapter is to

present an overview of the methodology for the design of both unitized framed glass

façade systems (Section 3.2) and point fixed glass façade systems (Section 3.3).

50


3.2 Design of Unitized Framed Glass Façade System

A typical example of a unitized framed glass façade system is shown in Figure 3.1 for

a 9.8m x 3.6m façade frame grid with six internal panels 1500mm wide and two end

panels 800mm wide to cover the external columns. Glass façade systems must meet

several out-of-plane and in-plane design criteria to perform satisfactorily. Generally,

the façade is designed for the normal design wind pressure using the general pressure

coefficients and thicker glass panels may be required at the corners to accommodate

the higher pressure associated with vortex shedding. Some engineers conservatively

design all façade panels based on the higher corner wind pressure.

Rollerconnection

Spandrel panelsSpandrelPinbeamconnection

External

columnExternal Transom

columnMullion

Vision panel

84 5 72 61 3

Spandrel

beam

Figure 3.1 Typical layout of unitized framed glass façade system for façade grid of

9800 mm×3600 mm

51


3.3 Out-of-Plane Design

The most important out-of-plane loading is due to wind. Glass panels, mullions and

transoms are designed for the critical wind pressure as determined in accordance with

AS/NZS 1170.2 (2002)

3.3.1 Structural Design of Glass Panel

Glass panels are required to comply with Australian Standard AS 2047 (1999)

Windows in buildings - Selection and installation, which in turn specifies that the

design of glass assemblies must comply with Australian Standard AS 1288 (2006)

Glass in buildings - Selection and installation. Design methods and charts provided in

AS 1288 (2006) are used to determine the thickness of the glass for both strength and

serviceability criteria. AS 1288 (2006) covers the design of annealed, heat

strengthened and toughened glass together with the design of double glazed and

laminated glazed façades.

3.3.2 Design of Mullion and Transom

The mullions and transoms are normally aluminium and fabricated as half sections

(female and male) which provide a built-in tolerance and adjustment for construction

and movement during service life (Figure 3.1). There is a wide variety of sections

available to cater for different levels of structural demands as well as architectural

requirements. The structural analysis of mullions and transoms is based on simple

beam theory. The members are subject to combined actions due to loading in different

directions, for example, the transoms are subject to bending about one axis due to out-

of-plane wind loading and bending about the orthogonal axis due to the weight of the

glass whilst the mullions are subject to bending due to out-of-plane wind loading and

axial tension due to gravity loads. AS/NZS 1664.2 (1997) Aluminium Structures

provides the basis for the structural design of aluminium mullions and transoms.

3.4 In-Plane Design

The glass façade system is designed for the deflection of the spandrel beam due to

gravity floor loading and lateral building movement induced by wind and earthquake

loads. The thermal movement of mullions, transoms and glass panels are also

52


considered to ensure that a sufficient gap exists between adjacent panels to

accommodate the movement.

3.4.1 Thermal Expansion of Mullion and Transom

The thermal expansion of the mullion and transom is critical compared to the glass

panel. Allowance for the thermal differential movement of the mullion, transom and

glass panels has to be provided between the edges of the glass to mullion and glass to

transom connections. The differential movement of the mullions in the horizontal

direction and the transoms in the vertical direction due to thermal expansion of

adjacent panels must also be accommodated between the female and male joints of

the mullions and transoms.

3.4.2 Serviceability Limit State Deflection of Spandrel Beam

Typically each panel is pin connected to the spandrel beam near the top left corner of

each panel as shown in Figure 3.1. The top right corner is connected with a roller

support (sliding arm connection) resting on the adjacent panel’s transom that allows

horizontal movements, except the end panel where the sliding arm rests on the

spandrel beam for the vertical support. The glass panel movements associated with

the deflection of the spandrel beam is shown in Figure 3.2. The boundary condition of

the spandrel beam could be pinned, semi-rigid or rigid and will directly affect the

deflection profile.

The maximum relative displacement between adjacent panels in the vertical and

horizontal directions is calculated using rigid body rotations. Such calculations are

demonstrated below for the case where the spandrel beam has pin connections. It

should be noted that the deflection due to the dead load can be neglected, since the

façade system will be erected following the construction of the superstructure. The

differential deflection of the panels along the spandrel beam has to be accommodated

by both the panel connection to the spandrel beam and the joint detail between the

female and male sections of the mullions and transoms. These differential movements

result in the gap between adjacent panels opening at the bottom and closing at the top

of the panels and both movements need to be considered in the design process.

53


Deflected Beam

84 5 71 2 63

Figure 3.2 Relative vertical panels movement due to the deflection of spandrel beam

A sample calculation is presented for evaluating the differential movement of the

glazing panels due to the deflection of the spandrel beam for a typical layout as shown

in Figures 3.1 and 3.3. It is assumed in the calculation a maximum deflection limit of

L/250, where L is the span of the spandrel beam. This is conservative and the

calculation can be simply scaled to accommodate lower deflections limits due to live

load only.

(a) Maximum displacement in vertical direction:

L 9800 Maximum allowable serviceability deflection of a spandrel beam = = = 39 mm

250 250

where L = 9.8m is the beam span. Since the maximum deflection of a simply supported

5 wl 4 L 5 wL 4 w 384 beam = , it follows that = and therefore =

384 EI 250 384 EI EI 1250L3

54


The deflection 'y' of a simply supported beam at any given distance 'x',

wx(L-x) 2 w 384 y = ⎡L +x(L-x) ⎤ , and substituting = results in the equation, ⎣ ⎦ 324EI EI 1250L

16x(L-x) 2y = ⎡L +x(L-x) ⎤3 ⎣ ⎦1250L

This equation can be used to calculate the vertical deflection between two adjacent

panels as shown in Figure 3.3. Calculated relative vertical and horizontal

displacements between panels are summarised in Table 3.1.

3

Deflected shape

Initial beam position

Column edge 4

Figure 3.3 Deflection of spandrel beam and glazing units

(b) Maximum displacement in horizontal direction:

The rotation of each panel is assumed to be equal to the relative vertical displacement

divided by the panel width. The panel rotation can then be used to estimate the

horizontal displacement at the side of the panel for the simple rigid body rotation as

shown in Figure 3.3.

Horizontal displacement of panel 2 = (17.6/1500)x(3600-625) = 34.8mm

Horizontal displacement of panel 3 = (12/1500)x(3600-625) = 24.2mm

55


Since both panels 2 and 3 move in the same direction, the resultant relative horizontal

displacement (panel opening) equals 10.7mm (= 35.8- 24.2mm).

Displacement between panels 4 and 5 = (4.3/1500)x(3600-625) = 8.6mm

The mid panels 4 and 5 move in the opposite direction (Figure 3.2), and hence the

resultant maximum panel opening equals 17.2mm (= 2x8.6mm). The translations of

each panel and relative movements between panels are summarised in Table 3.1

Table 3.1 Relative vertical and horizontal displacement between panels

Panel

Number

X distance

(mm)

Y value

(mm)

Vertical

differential

movement of

panels (mm)

Horizontal

displacement

of panels

(mm)

Panel

opening in

horizontal

direction (mm)

400 5.1

2 17.6 34.8

1900 22.7 10.7

3 12.2 24.2

3400 34.9 15.6

4 4.3 8.6

4900 39.2 17.2

5 -4.3 -8.6

6400 34.9 15.6

6 -12.2 -24.2

7900 22.7

3.4.3 Building Movement Caused by Wind Loading

The Australian Standard “Concrete structures”, AS 3600 (2009) specifies in clause

2.3.2, that “unbraced frames and multi-storey buildings subject to lateral loading shall

be designed to limit calculated inter-storey lateral drift to 1/500 (0.2%) of the storey

height”. This is aimed for the serviceability limit state of the building mainly for wind

loading. Therefore, for a 3600mm height curtain wall, the drift demand equals 7.2mm

(h/500 = 3600/500) and the in-plane drift capacity of the glazing system is achieved

through rigid body articulation of the panels provided by gaps between the frame to

glass and between the male to female mullion joints.

56


3.4.4 Building Movement Caused by Earthquake Loading

An estimate of the ultimate building movement and drift due to earthquakes can be

obtained using the approximate method proposed by Wilson and Lam (2005) in

conjunction with AS 1170.4 (2007).The maximum drift can be estimated by Equation

3.1:

⎡PDD ⎤Maximum drift = 3 Eq (3.1) ⎢ ⎥

nh ⎣ 1 ⎦

Where, PDD - Peak displacement demand, n- number of stories and h1- Floor height.

Earthquake drift demand on a uniform and regular 12 storey building on different soil

sites is shown in Table 3.2. As expected, the maximum inter-storey drift occurs for

soil site classification “E”. A maximum of 0.63% and 1.15% drift is expected for

importance level 2 (500 years returned period) and 3 (2500 years returned period)

buildings respectively. The Building Code of Australia (BCA, 2011) recommends the

seismic design of glass façade systems be performed in accordance with AS 1170.4

(2007). Previously, it was assumed that façade systems could tolerate earthquake

loading and façade engineers typically did not design explicitly for this action.

Table 3.2 Earthquake drift demand of a 12 storey building on different soil sites

Site

classification

PDD (mm)

500yr

Inter-storey

drift (%)

PDD (mm)

2500yr

Inter-storey

drift (%)

A 20 0.14 40 0.28

B 25 0.17 50 0.35

C 35 0.24 65 0.45

D 60 0.42 105 0.73

E 90 0.63 165 1.15

3.4.5 In-plane Drift Capacity of Unitized Framed glass Façade

In unitized framed glass façade system, the glazing units are hung from the top and

are allowed to slide and rotate between the female and male split joints of the mullion

and transom members. The in-plane movement capacity of the glazing to

accommodate the spandrel beam deflection can be determined by the depth of the

57


split joints. The unique nature of the connection detail in unitized glazing, consisting

of one pin and one roller support provides articulation and makes it effectively an

earthquake isolated curtain wall. This mechanism provides a significant amount of in-

plane drift capacity through sliding and rotational behaviour of the glazing panels and

therefore, in regions of low to moderate seismicity the seismic design of unitized

framed glass façade system is typically not critical.

3.5 Design of Point-Fixed Glass Façade System

A typical example of a point fixed glass façade system is shown in Figure 3.4 for a

9.8m x 4m building façade grid at the ground floor level. The glass façade system

between adjacent columns consists of 10 toughened point fixed panels including four

opaque panels measuring 800mm × 2000mm to cover the columns and eight clear

panels measuring 2250mm × 2000mm between columns as shown in Figure 3.4.

Generally the dimensions and the arrangement of the glass panels including sealant

thickness between panels are specified by the architect and the façade engineer for the

project.

Bolted connections

Figure 3.4 Schematic diagram of typical point fixed glass façade at ground floor

3.5.1 Out-of-Plane Glass Panel Design

A glass panel point supported at four corners and subjected to a uniform out-of-plane

wind load would experience maximum bending stress at the centre of the panel.

However, since the strength capacity of glass is not uniform and tends to be lower

near the edges, the maximum stress along an edge also needs to be considered. For a

58


typical panel as shown in Figure 3.5, the maximum edge stress is at mid-length of the

longer edges of the glass panel. Equations to calculate the stresses and deflection are

provided as follows (Zhou, 2002):

2 wa

The maximum bending stress at the longer edge: σe = 0.916 2

Eq (3.2) t

2 wa

The maximum bending stress at centre: σc = 1.208 2

Eq (3.3) t

4 wa

Maximum deflection at the centre: δ = 0.294 Eq (3.4)c Et

3

Where “a” is the distance between two neighbouring point supports, “t” is the

thickness of the glass and “w” is the design wind pressure. The stresses and the

deflections of point fixed glass panels also can be found using commercial finite

element software packages. If the glass panels are large and require a series of bolted

connections it may be necessary to carry out a finite element analysis to determine the

serviceability deflection and ultimate bending stresses. Design methods are provided

in AS 1288 (2006) to calculate the ultimate tensile strength capacity of the glass

panels and the serviceability deflection.

2250 mm

2000 mm Locations of

maximum stresses

Centre

Edge

Figure 3.5 Detail of the proposed glass panel

59


3.5.2 In-Plane Glass Panel Design

In the point fixed glass façade systems the structural support frame, connection details

and the weather sealant are designed to accommodate the in-plane building movement

due to wind, thermal and earthquake loads. Generally toughened safety glass is used

and each glass panel is supported to allow the panel to articulate and accommodate in-

plane drift of the structural support frame and prevent glass failure. A typical support

configuration is shown in Figure 3.6, with two horizontal slotted holes at the top

corners and bottom fixings with larger diameter holes to allow both relative vertical

and horizontal movements.

The resulting relative movement of the glass panel relative to the building drift is

shown in Figure 3.7 with each glass panel rotating about the slotted holes at the top

corners and relative movement at the other support points. The glass panels will

translate in the direction of applied load and rotate in the opposite direction due to the

differential vertical movement of the spider arm ends. Figure 3.8 shows the enlarged

relative movement between the adjacent glass panels at a typical 4 arm spider fixing

location.

The articulation is made possible through movements between the glass and spider

arms as well as rotation of the spider arms which support the glass panel (Figure 3.9).

In addition, new bolts with swivel heads are now available in the market to connect

the glass to the spider arms as shown in Figure 3.10. These bolts provide additional

capacity for movement between the glass and structural support frames. The in-plane

loadings such as thermal expansion, building movement due to wind load and seismic

movement are accommodated by the slotted and larger diameter holes on the spider

arms. The snug tightened bolts allow the toughened glass to move and rotate and

hence avoid the build-up of stresses in the glass.

However there are no specifications or methods available to façade engineers to

calculate the in-plane drift capacity of the point fixed glass façades with the

articulation described. Weather sealants are normally used to seal adjacent glass edges

to maintain water and air tightness in point fixed glass façade systems. These weather

sealants generally have lower elastic modulus, lower hardness, and less strength

compared to structural sealants, but can accommodate larger movements (Zhou,

2002).

60


Slotted hole

Glass panel Large hole

Figure 3.6 Schematic representation of glass façade articulation in point fixing. The

slotted and larger holes are in the spider arms supporting the glass panel

Figure 3.7 Rigid body rotation of glass panels under in-plane lateral loading

61


Figure 3.8 Rotated spider arm and relative movement of adjacent glass panels

Figure 3.9 Typical spider arms with slotted holes and large diameter holes

Figure 3.10 Swivel button head bolt fittings to connect glass and spider arms

62


3.5.3 Bolted Connection Location and Design

Stress concentrations in the glass can be found in the glass in the locations of

attachments to the spider arms. The magnitude of the stress is dependent on the bolt

location and diameter and such parameters have to be considered for an optimum

design to prevent glass failure. Research has been carried out investigating the

optimal position of the glass panel point support at the University of Karlsruhe,

Germany (Klinkenberg et al., 1998). The edge distances were found to be more

critical for out-of-plane wind loading compared with in-plane dead loads. Special

connections types were considered and the optimal distance for the bolt holes were

specified according to the connection type. Glass manufacturers in Australia (for

example ‘Viridian New World Glass and G. James Glass and Aluminium) provide

detailed guidelines for the glass hole positions in toughened glass as illustrated in

Figure 3.11.

Figure 3.11 Guidelines for the holes in toughened safety glass (Viridian, 2010, G.

James, 2010)

63


3.5.4 Stresses at the Glass Bolted Hole and Bolt Design

The bolts connecting the glass to the spider arms and spider arms to the supporting

structure are selected to have adequate capacity to resist the in-plane and out-of-plane

loads imposed on the panels. The wind load is considered for the out-of-plane design

whilst self-weight of the glass panels, thermal expansion, building movement and

earthquake are considered for in-plane design. The earthquake design involves

ensuring that the façade system has sufficient drift capacity to accommodate the drift

demands imposed on the building structure from the earthquake as discussed in

Section 2.2.4. The bolts to connect the glass to spider arms are normally made of

stainless steel whilst a typical countersunk bolt connection is shown in Figure 3.12.

Figure 3.12 Dead and wind forces on the countersunk bolt fitting

64


3.6 Conclusion and Summary

This chapter presents an overview of the design methodology for both framed and

point fixed glass façade systems in Australia. A glass façade system needs to be

designed to satisfy the ultimate strength limit states and deflections associated with

the serviceability limit state from a combination of loads including dead, live,

thermal, wind and earthquakes. Although, the design methodology for a unitized

framed glass façade system is well established, the design of point fixed glass façade

systems is less established and a recommended methodology has been presented for

unitized framed glass façade for both in-plane and out-of-plane loading. For the point

fixed glass façade system out-of-plane loading, typical techniques for installation and

simple formulas to determine the maximum stresses and deflection were presented.

For in-plane loading, use of slotted holes and swivel bolts to accommodate

movements were suggested. There is little or no standard practice available for in-

plane loading. However, this will be dealt with as part of the research and a design

methodology will be documented.

65

Chapter 4 IN-PLANE RACKING TESTS OF POINT FIXED GLASS FAÇADE SYSTEMS

Chapter 4

4 IN-PLANE RACKING TESTS OF POINT FIXED

GLASS FAÇADE SYSTEMS

4.1 Introduction

Two unique full scale in-plane racking laboratory tests (Test #1 and Test #2) of

typical point fixed glass façade systems (PFGFS) were conducted. The tests utilised

contemporary connections to attach the glass panels to the structural support frame

consist of spider arms and special bolt fittings. As outlined earlier, the in-plane

racking performance of PFGFS is dependent on three main components; the glass

panels, the connection details and the structural support frame. In these tests, a strong

structural support frame was articulated so that the racking performance of the glass

panels and the connection details could be assessed.

In this study, spider arms are configured as X-type (Figure 4.1) or K-type (Figure 4.2)

depending on the type of fixity at the structural support frame. X-type spider arms

were used for Test #1 and K-type spider arms were used for Test #2. The spider arms

were snug tightened to the structural support frame as would normally be done in

practice. The X-type spider arms were connected to the structural support frame using

a single bolt to allow in-plane rotation of the glass panels at the spider arm-to

structural support frame connection. The K-type spider are connected to the structural

support frame using double bolts and which do not allow the glass panels to rotate at

the spider arm-to-structural support frame connection but allow sliding at the base

connection in the vertical direction.

There are different types of bolt fittings available in the market to connect the glass

to-spider arms namely, countersunk, button head and swivel connections as shown in

Figure 4.3. Countersunk and button head bolt fittings are the most common and the

cheapest options whilst swivel connections are used when excessive stress

developments are expected. Test #1 was performed with X-type spider arms and

66


countersunk bolt fittings whilst Test #2 was performed with K-type spider arms and

button head bolt fittings. This Chapter describes the test setup and results.

Figure 4.1 X-type spider arm with countersunk bolt fittings (Test #1)

Figure 4.2 K-type spider arm with button head bolt fittings (Test #2)

67


Countersunk Button head Swivel

Figure 4.3 Different types of bolt fittings commonly used in Australia

4.2 Test #1 – ‘X’-Type Spider Arms and Countersunk Bolt Fitting

4.2.1 Test #1 – Experimental setup

The test was conducted on a typical PFGFS as shown in Figure 4.4, which consisted

of four 1200mm x 1200mm toughened 12mm thick glass panels joined with 8mm

thick silicon weather sealant. A structural support frame (blue frame) was designed to

support the glass panels through the spider arms. It was fabricated using 180PFC

sections and bolted together using M24 bolts which were snug tightened to allow for

free racking mechanisms as shown in Figures 4.4 and 4.5.

The flanges of the vertical PFC were removed at each end to facilitate pin connections

between the vertical webs and the horizontal PFC members. The bottom flange of the

horizontal PFC at the floor level was rigidly connected to the floor using M24 bolts.

The hydraulic jacking system was capable of applying 100kN in-plane lateral load

and more than 150 mm in-plane displacement. The hydraulic jack was supported by a

reaction frame as shown in Figure 4.5.

The structural support frame was prevented from moving in the out-of-plane direction

by four sets of rollers mounted at the top as shown in Figure 4.5. The rollers ensured

that the structural support frame was aligned with the loading direction. Once the

structural support frame was assembled, glaziers fixed the spider arms, glass panels

and applied the weather sealant (Figure 4.6). A special transparent adhesive film was

68


applied to the glass panels to prevent the glass fragments scattering following any

glass fracture.

The hydraulic jacking system was mounted on the reaction frame (yellow) was used

to laterally load the structural support frame (blue frame) and the façade system. As

additional precaution the test area was enveloped with nets to capture any flying glass

fragments following fracture and to ensure safety in the lab.

The racking test procedure and instrumentation setup were as follows:

• A lateral load was applied to the top right hand corner in a step by step manner

with displacement increments of 5mm until failure.

• Two systems of displacement measurement were adopted to achieve good

confidence in the data acquisition:

o Linear Voltage Displacement Transducers (LVDTs)

o Photogrammetry

• Deflections were measured at 11 locations (horizontal, vertical and out-of

plane) with the LVDTs whilst the applied load was measured using a load cell

(Figure 4.7). The details of the LVDTs are summarised in Table 4.1.

• Photogrammetry provided displacement data for the target points that were

tactically positioned and marked with retro-reflective adhesive labels (Figures

4.9 & 4.10). Photographs of the targets were taken before and after a sequence

of loading and the relative movement in their positions were interpreted using

software based on the principle of triangulation. The Photogrammetry

measurements provided movements in all three directions (x, y and z)

4.2.2 Test #1 - Experimental Results and Discussion

The load-displacement curve measured at the top of the structural support frame at a

relative height of 2.72m from the floor (LVDT No. 01 in Figure 4.7) is shown in

Figure 4.8. It indicates that the structure performed linearly until failure. The slightly

jagged nature was reflective of the initial stiffness of the sealant relaxing together

69


with the difference between the dynamic and static frictional movement and rotation

at the connections (structural support frame to spider arm, spider arm to bolt fitting

and bolt fitting to glass connection). A maximum in-plane lateral displacement of 58

mm was measured with a corresponding 16kN racking load before failure.

Surprisingly, this resulted in a maximum of 2.1% in-plane drift capacity for the

system with minor damage to the sealant and yielding of the spider arms, before

catastrophic failure of one of the glass panels. The failure of the system and the glass

panel are shown in Figures 4.9 & 4.10. The adhesive film was very advantageous in

preventing the shattered glass fragments from spreading all around the laboratory.

It was observed during the racking test that the glass panels and the spider arms all

translated as rigid bodies whilst the sealant deformed at the interface followed by

spider arms deformation and yielding (distortion) at one location. The X-type spider

arms used in this experiment had a frictional moment capacity (torsional) after which

rotation would occur. A simple truss analysis was carried out to determine the loading

actions (tension or compression) in the panels as shown in Figure 4.11. The initial

(blue) and the final (red) locations of the panels are shown to scale in Figure 4.12 and

these represent the translations that occurred in the glass panels before failure. Rigid

body translation in both the horizontal and vertical directions was observed at the

built-in standard gaps (gaps) at the connection details (structural support frame to

spider arm, spider arm to bolt fitting and bolt fitting to glass connection).

During the first measurement of loading with approximately 0.1% drift, rotation in the

corner spider arms was observed (Figure 4.13). The measurement was taken at the

spider arm to glass bolted connections. It should be noted that when the spider arm

rotates the bolted connection moves in the vertical direction. Consequently, rotations

were observed in all the perimeter spider arms as shown in Figure 4.13

(positive/upward vertical displacement in the perimeter arms on the left hand side and

negative/downward vertical displacement in the perimeter arms on the right hand side

was measured).

Interestingly the glass panels also displaced in the out-of-plane direction relative to

each other as shown in Figure 4.14. This differential movement was caused by the net

vertical force at the connection deforming the spider arms in the out-of-plane

70


direction. A significant amount of out-of-plane movement was observed between

arms PBB4 (Panel PB spider arm B4) to PDB2 and PAB3 to PCB1 (Figure 4.15) with

a maximum differential movement of approximately 8.5mm. The relative out-of-plane

movement of the spider arms was measured using the photogrammetry targets on the

bolt head and the permanent photogrammetry targets on the floor. This out-of-plane

movement induced combined local bending and tensile stresses in the glass

particularly around the bolt hole (Bolt PBB4) resulting in the initiation of cracking

and catastrophic failure of the bottom right hand glass panel as shown in Figures 4.9,

4.10 & 4. 14.

180

180

94

2720

1200

1200

2596

180PFC

180

PF

C

A

A SECTION A-A

25

5

0

M24 Bolt

1200

180

180

25

Glass panel Glass panel

Glass panel Glass panel

Figure 4.4 Schematic diagram of the PFGFS in Test #1

71


Figure 4.5 Structural support frame (blue frame) assembled into the reaction frame

(yellow frame) (Test #1)

Figure 4.6 Test specimen - glass panels installed and transparent adhesive film

applied (Test #1)

72


GROUND FLOOR

180PFC

18

0P

FC

20

0

3300

90

x90

x10

EA

A A

B

100 x16 Flat Bar

300

550

1

8

2

4

3

56

10

9

7

11

Hydraulic Jack

Load Cell

Glass Panels

Figure 4.7 Locations of the LVDTs and the hydraulic jack and the loading bar

attachment with the structural support frame (Test #1)

Table 4.1 Details of the LVDTs used in the Test #1

Number Description

1 Top displacement

2 Out-of-plane deformations of the glass panel

3 In-plane lateral relative movement of glass panels

4 In-plane vertical relative movement of glass panels

5 Spider arm vertical movement

6 Corner spider arm vertical movement

7 Internal central spider arm movement

8 Lateral movement of the test frame at the bottom

9 Vertical movement of the test frame at the bottom

10 Movement of the reactive frame at the top

11 Out-of-plane deformations of the glass panel

73

14


18

Failure at 16 2.1% drift

12

10

8

6

4

2

0

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

In-plane drift (%)

Figure 4.8 The measured racking load versus displacement for the system (Test #1)

In-p

lan

e l

oa

d (

kN

)

Figure 4.9 The system after failure of a glass panel (Test #1)

74


Figure 4.10 The broken glass panel after failure with the adhesive film securing the

glass fragments (Test #1)

B1

B3 B4

B2

Glass panel (PA)

B1

B3 B4

B2

Glass panel (PB)

B1

B3 B4

B2

Glass panel (PD)

B1

B3 B4

B2

Glass panel (PC)

TENSION COMPRESSION

Figure 4.11 Glass panels and spider arms to glass bolted connections labelled and

the compression, tension and the spider arm rotational directions indicated (Test #1)

75


Translation of the glass panel at the drill holes

2500

2000

He

igh

t (m

m) 1500

1000

500

0 0 500 1000 1500 2000 2500

Width (mm)

Ve

rtic

al d

isp

lac

em

en

t (m

m)

Figure 4.12 Translations of the glass panels (Test #1)

20

15 Left hand side

perimeter spider arms

10

5

0

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8

Central spider arms

2 2.2

-20

-15

-10

-5 PAB1

PDB4

PBB2

PAB2

PAB3

PAB4

PBB4

PDB3

PCB3

Right hand side

perimeter spider arms

-25 In-plane drift (%)

Figure 4.13 Displacement of the spider arms (to glass bolted connections) in the

vertical direction due to the rotation of the spider arms (Test #1, +ve movement

upward and -ve movement downward)

76


Figure 4.14 Out-of-plane deformation and distortion of the spider arm PBB4 and

PDB2 after failure of a glass panel (Test #1)

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

In-plane drift (%)

0

1

2

3

4

5

6

7

8

9

Defo

rmati

on

(m

m)

PDB2 to PBB4

PCA3 to PCB1

Figure 4.15 Differential out-of-plane movement of the spider arms (Test #1)

77


4.3 Test #2 – ‘K’-Type Spider Arms with Button Head Bolt

Fitting

4.3.1 Test #2 - Experimental Setup

The 2x2 glass panel test specimen was constructed with K-type spider arms and

button head bolt fittings is shown in Figure 4.16. The specimen description and the

experimental setup was the same as described in Section 4.2 for Test #1. However,

the structural support frame was modified by welding some ‘T’ sections thus creating

fixed cleats, to which the K-type spider arms were connected with two snug tightened

M10 bolts, to replicate a typical design practice.

Representative built-in standard gaps were provided by drilling 20mm diameter holes

in the cleat to accommodate the M10 bolts. In addition the spider arm base had two

vertical slots measuring 14mmx25mm for the M10 bolts. This provided ±7mm

horizontal (in the out-of-plane direction of the glass panels) and ±17.5mm vertical

gaps for the M10 bolted connections, which is representative of industry practice.

Figures 4.17 and 4.18 show the K-type spider arms to cleat connection details. Button

head bolt fittings were then used to connect the glass panels to the spider arm as

shown in Figure 4.16. The instrumentation and the test procedure was the same as

Test #1 with the lateral load applied at the top right hand corner in a step by step

manner with displacement increments of 5mm up to failure.

4.3.2 Test #2 - Experimental Results and Discussion

The load-displacement curve measured at the top of the structural support frame

(LVDT No 1 in Figure 4.7) for Test #2 is shown in Figure 4.19 and indicates that the

façade system performed almost linearly up to failure. A maximum displacement of

143 mm was measured with a corresponding 38kN racking load at failure.

Surprisingly, this resulted in a maximum 5.25% in-plane drift capacity for the 2.72m

height system. At this high level of drift there was damage to the sealant and yielding

of the spider arms, before catastrophic failure of one of the glass panels. The failure of

the system and the glass panel are shown in Figure 4.20 with the adhesive film again

preventing the shattered glass fragments from falling.

78


It was observed during the racking test that the glass panels translated and the spider

arms moved vertically within the slotted hole connection to the structural support

frame cleat. Damage along the vertical silicon sealant was noticed at a 2.0% drift.

Rigid body translations were observed at the spider arm glass bolted connection. As a

result the glass panels translated up to a displacement of 3.3% drift. Beyond a drift of

3.3% there was no capability for further rigid body translations as the spider arms

were bearing on the edges of the circular and slotted holes and consequently the

spider arms began to deform to accommodate further drift. This resulted in both in-

plane and out-of-plane deformations of the spider arm fixings, and created excessive

bending and tensile stresses in the glass, before catastrophic failure at a large drift

5.25%.

A simple truss analysis was carried out to determine the loading actions (tension or

compression) in the panels as shown in Figure 4.21 (where ‘PA’ is panel ‘A’ and B4

is bolt 4) together with the possible vertical movement direction of the spider arms.

The initial (red) and the final (blue) locations of the panels are shown to scale in

Figure 4.22 and demonstrate the translations and rotations that occurred in the glass

panels before failure. The vertical displacement of the spider arm to glass bolted

connection is plotted in Figure 4.23 whilst the relative vertical measurement of the

internal spider arm is shown in Figure 4.24 after failure of the system. The out-of

plane deformation of the glass panel is plotted in Figure 4.25 and illustrated in Figure

4.26 whilst the damage and distortion of the spider arm fixings at failure is shown in

Figure 4.27.

4.3.3 Test #2 – Ultimate Fracture Strength of Toughened Glass

During glass failure, the crack front propagates through the material, creating fracture

features known as the mirror, mist, and hackle (Figure 4.28). The crack front initially

produces the smooth mirror region. However, as the crack accelerates it becomes

more unstable, creating a dimpled surface known as mist. This instability eventually

causes the crack to branch out, producing the rough hackle region. The hackle region

is characterised by elongated markings that proceed in the direction of crack

propagation. The hackle markings point back to the flaw origin (Frechette, 1990).

79


The ultimate fracture strength of glass can be measured using an empirical expression.

The radius of the mirror is inversely proportional to the square of the stress when the

mirror was formed and it may be used to calculate the stress at the instant of fracture.

From fracture mechanics analysis, the radius of the mirror also relates to the critical

size of the flaw and the time to catastrophic failure under fatigue conditions (Shinkai,

√

1994). The stress that initiates the fracture, fr can be found from the Equation 4.1

(Shand, 1959):

= . Eq (4.1)

In this equation, fr is the tensile stress in MPa and ‘r’ is the mirror radius in metres.

This equation has been found to hold for a variety of sample sizes and surface

conditions. Measurement of the mirror radius parallel to the surface of the sample has

been found to produce the most reliable results (Brungs and Sugeng, 1995). In Test #2

the broken glass panel was investigated and the fracture origin was identified. The

"mirror radius" was measured to be 6 mm at the glass hole where the crack originated

as illustrated in Figure 4.29. Using Equation 4.1 the effective tensile stress at the edge

of the hole where the fracture originated was approximately 28 MPa. Conservatively,

the minimum compressive stress at the edge of the hole should be 69 MPa which is

the minimum toughening stress given in (AS1288, 2006). Therefore, the applied

tensile stress due to the racking loads to overcome the pre-compression and cause

tensile failure could be estimated to be around 97 MPa. In the FE analysis presented

in Section 5 an ultimate tensile failure stress of 94MPa was used for the toughened

glass which is consistent with the requirement of AS1288 (2006).

80


Figure 4.16 Test specimen - glass panels installed, transparent adhesive film applied

and Photogrammetry targets attached (Test #2)

Slots to allow the spider

arm to move vertically

(± 17.5mm)

Figure 4.17 Visible built-in standard gap at the spider arm base to cleat connection

(Test #2)

81


In-p

lan

e lo

ad

(kN

)

14mm x 25mm

slotted holes

20mm diameter holes

(a) (b)

Figure 4.18 (a) Slotted holes in the spider arm base plate and (b) Large bolt holes at

the cleat (Test #2)

40

35

Test #2- Experiment

30 Spider arm base

deformation observed

(3.3 % drift)

25

Damage observed at Failure of a glass panel 20 the sealant joint (2.0% (5.25 % drift) drift)

15

10

5

0

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5

In-plane drift (%)

Figure 4.19 Racking load versus displacement for the PFGFS (Test #2)

82


Figure 4.20 The system after failure of a glass panel (Test #2)

COMPRESSION TENSION

B1

B3 B4

B2

Glass panel (PA)

B1

B3 B4

B2

Glass panel (PB)

B1

B3 B4

B2

Glass panel (PD)

B1

B3 B4

B2

Glass panel (PC)

Figure 4.21 Glass panels and spider arms connections labelled and the compression,

tension and the spider arm sliding directions indicated (Test #2)

83


Translation of the glass panels at the drill holes

0

500

1000

1500

2000

2500

0 500 1000 1500 2000 2500

Width (mm)

He

igh

t (m

m)

Figure 4.22 Translation of the glass panels at the bolt holes (Test #2)

-15

-10

-5

0

5

10

15

20

25

30

35

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

In-plane drift (%)

Vert

ical

mo

vem

en

t (m

m)

PAB1

PAB2

PAB3

PAB4

PBB2

PBB4

PCB3

PCB4

PDB3

PDB4

Figure 4.23 Displacement of the spider arms in the vertical direction (Test #2)

84


(a) (b)

Figure 4.24 (a) Initial position of the internal centre spider arm and (b) Relative

vertical sliding of the spider arms after failure (Test #2)

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

In-plane drift (%)

0

1

2

3

4

5

6

7

8

9

10

11

Defo

rmati

on

(m

m)

PDB2 to PBB4

PAB3 to PCB1

Figure 4.25 Differential out-of-plane movement of the spider arms (Test #2)

85


Figure 4.26 Out-of-plane deformation of the spider arm PAB3 and PCB1 (Test #2)

(a) (b)

Figure 4.27 (a) Deformed spider arm PCB3 due to compression and (b) Deformed

spider arm and base plate PDB4 due to tension (Test #2)

86


Figure 4.28 Schematic diagram of a typical glass failure or crack origin (Castilone

et al., 2002)

Figure 4.29 Glass fracture originated from the bolted connection PCB4 (Test #2)

87


4.4 Test Summary and Further Studies

In Test #1, the maximum drift of 2.1% was much larger than initially anticipated and

demonstrated that the 2x2 system was surprisingly able to sustain high level of in-

plane drift. The glass panel rigid body translation, spider arms rotation and

deformations allowed the system to move laterally and created the increased drift

capacity. A significant amount of out-of-plane movement was observed between arms

PBB4 to PDB2 and PAB3 to PCB1 with a maximum differential movement of

approximately 8.5mm. This out-of-plane movement induced combined local bending

and tensile stresses in the glass particularly around the glass hole (Bolt PBB4)

resulting in the initiation of cracking and catastrophic failure of the bottom right hand

glass panel.

In Test #2, the maximum drift of 5.25% (143mm) was much larger than initially

anticipated and demonstrated that the 2x2 system was surprisingly tolerant to drift.

Damage along the vertical silicon sealant was noticed at 2.0% drift. The spider arms

began to deform at 3.3% drift whilst the base plate of the spider arms also commenced

to yield. The system continued to deform until failure of glass panel ‘C’ at 5.25% drift

due to excessive bending stresses (from out-of-plane displacement of the spider arms)

combined with the in-plane diagonal tensile stresses.

The two tests completed indicated that the PFGFS with X-type and K-type spider

arms had surprisingly large in-plane drift capacities caused by rigid body translation

of the glass panels at the built-in standard gaps, spider arm rotation (Test #1), spider

arm vertical translation (Test #2) and deformation of the spider arm components.

Overall the drift capacities for Test #1 (pin connection) and Test #2 (rigid connection

with slotted holes) were 2.1% and 5.25% respectively.

In order to further understand the racking performance of the systems detailed FE

model were developed to replicate the test results of Test #1 and Test #2 and

discussed in the following Chapter.

88

Chapter 5 FINITE ELEMENT MODELLING OF THE IN-PLANE RACKING PERFORMANCE OF PFGFS

Chapter 5

5 FINITE ELEMENT MODELLING OF THE IN-PLANE

RACKING PERFORMANCE OF PFGFS

5.1 Introduction

Two unique, full scale laboratory experimental racking tests were conducted on two

different types of PFGFS connected to a strong articulated structural support frame. In

practice, PFGFS systems have some variations in terms of glass panel shapes, glass

thickness, sealant thickness, sealant types and facade grid arrangements. Therefore,

based on the laboratory test results, the racking performance of other insitu systems

can be speculated but not quantified without using a predictive analytical model such

as finite element modelling (FE modelling). For example, when a structural sealant

(stiffer than weather sealant) is used instead of a weather sealant, the in-plane drift

capacity of the PFGFS could be reduced. Therefore highly sophisticated FE models

were developed and conservatively benchmarked against the laboratory test results to

demonstrate their suitability.

Three-dimensional non-linear FE models were created using the ANSYS 12.1 FE

program to replicate the laboratory tests (referred to as ‘Test #1’ and ‘Test #2’). The

results obtained from the FE models were benchmarked against the test results

including the pushover curve, failure stress and out-of-plane deformation of the glass

panels. Conservatively, the failure stress of toughened glass panels was calculated in

accordance with AS1288 (AS1288, 2006). In this Chapter the modelling methodology

and non-linear analysis approach undertaken using ANSYS for the Test #1 is

described through Sections 5.2 - 5.6 and Test #2 is described through Sections 5.7

5.10.

89


5.2 Test #1- Structural Idealisation

A 3-D analysis was needed since the racking structural support frame to which the

load was applied was eccentric to the bolted glass panels, which created differential

out-of-plane deformation on the spider arms and glass panels. The differential out-of

plane deformation on the glass panels created the bending stress on the glass panels in

addition to the axial tension and compressive stresses. The non-linearity was

necessary to take account of the presence of built-in gaps which eventually close

during loading, geometric non-linearity associated with out-of-plane differential

movement of glass panels and deformations of sealant and spider arms. However, to

avoid unnecessary complexities the spider arms were modelled as linear elastic

element.

5.2.1 Test #1- Racking Mechanism

It was observed during the experimental test that the glass panels and the spider arms,

all translated and rotated as rigid bodies whilst the sealant deformed at the interface.

A significant amount of rigid body translation followed by bearing action in both the

horizontal and vertical directions was observed at the bolted connections where the

built-in standard gaps allowed. The spider arms used in this experiment had a

frictional moment capacity beyond which rotation occurred on the perimeter spider

arms. The weather sealant offered some resistant against tensile, compressive and

shear actions and resisted some relative movement of the glass panels in both in-plane

and out-of-plane directions. The applied load was eccentric to the bolted glass panels,

which created differential out-of-plane deformations on the spider arms resulting in

the glass panels displacing in the out-of-plane direction relative to each other with a

maximum differential movement of approximately 8.5mm. This induced combined

local bending and tensile stresses particularly around the countersunk bolt holes on

the glass panels.

When a racking load was applied to the structural support frame, the glass panels

developed a diagonal strut action and the diagonal loads were transferred to the glass

panels via the spider arms as illustrated in Figure 5.1. Therefore the glass holes were

under bearing action due to the transfer of load. The net force transmitted from the

glass to the spider arm connection was approximately along the diagonal direction and

eccentric resulting in the arms deforming out-of-plane with one arm moving out from

90


the structural support frame and the other arm moving inwards causing relative out-

of-plane movement of the adjacent glass panels as shown in Figure 5.2. Therefore, the

rigid body translation at the built-in standard gaps, in-plane rotation and deformation

in the spider arms allowed the system to move laterally and created the drift capacity

despite the stiff sealant.

The glass panels and the spider arms are labelled in Figure 5.3 along with the racking

mechanism induced. Using a simple truss analysis, assuming the glass panels as

diagonal struts and ties, the compressive and the tensile diagonals can be identified as

illustrated in Figure 5.1. The diagonal forces which are approximately equivalent in

magnitude cause the in-plane rotation of the spider arms a counter clockwise direction

due to the eccentric resultant forces at the connections which is similar to the

experimental results shown in Figure 4.13.

Figure 5.1 Diagonal strut mechanism and load transfer through the spider arms

(Test #1)

91


Figure 5.2 Differential movement of the spider arms in the out-of-plane direction

(Test #1)

B1

B3 B4

B2

Glass panel (PA)

B1

B3 B4

B2

Glass panel (PB)

B1

B3 B4

B2

Glass panel (PD)

B1

B3 B4

B2

Glass panel (PC)

TENSION COMPRESSION

Figure 5.3 Glass panels and spider arm connections labelled along with the racking

mechanism (Test #1)

92


5.2.2 Model Assumption

It was understood from the racking mechanism that failure occurred in the glass panel

due to combined bending and axial tensile stresses. Bending stresses were developed

due to the differential out-of-plane movement of the spider arms and the tensile

stresses were developed due to the in-plane tensile force along the diagonal of the

glass panel. Therefore, in the FE model both bending and axial effects from the

countersunk bolt were incorporated. Firstly the countersunk tapered bolt head was

conservatively assumed as a uniform cylinder with a 20mm diameter (the smaller

diameter of the tapered section) and 12mm thick (height of the cylinder) as shown in

Figure 5.4. The hole in the glass panel was also assumed to be cylindrical with the

same diameter. This approximation increases the stress due to the in-plane diagonal

strut load on the glass hole because of the bearing area reduction.

Shell elements, 12mm thick were used to model the glass panels and bolt heads. The

bolt heads and the glass panels were meshed together (i.e., glued and there is no gap

between the bolt head to the glass hole) with appropriate material properties. This

distributed the stress due to the in-plane diagonal strut load uniformly around the bolt

hole and hence reducing the stress development around the glass hole. It was assumed

that the stress around the glass hole due to the in-plane diagonal strut load was not

significant compared to the bending stress.

Countersunk bolt

connection Actual hole detail Assumed hole detail

Figure 5.4 Schematic diagram of countersunk bolt connection (Test #1)

93


5.2.3 Features of the model

The model was created with a number of features to represent the racking behaviour,

including;

• The structural support frame, spider arms and M10 bolt to connect the spider

arm to glass panel were modelled using beam elements,

• Allowance for the spider arm to rotate when it overcomes the internal

frictional force at the structural support frame connections. This action was

modelled using non-linear springs with real constants assigned to represent the

frictional torsional moment versus rotation at the spider arm connection,

• Glass panels were modelled using shell elements and finely meshed around the

glass holes to determine the failure stresses,

• Conservatively, the countersunk bolt head was modelled as a 20mm diameter

cylindrical head and modelled using shell elements,

• The translations between (a) bolt fittings and spider arm and (b) spider arm

and support structure were modelled using non-linear springs, and

• Silicon sealant was modelled using a material model specified in the ANSYS

called Blatz and Ko (Bondi and McClelland, 2009) which is a one parameter

model to represent hyper elastic behaviour.

5.3 Test #1 - Model Description

The FE model was developed using different types of elements, material properties

real constants and material models.

5.3.1 Element Description

The structural support frame was modelled using elastic 3D beam elements ‘BEAM4

3-D elastic beam element’ as shown in Figure 5.5 and assigned with the structural

properties of 180PFC’s. BEAM4 element is a uniaxial element with tension,

compression, torsion, and bending capabilities. The element has six degrees of

freedom at each node- translations in the nodal ‘x’, ‘y’, and ‘z’ directions and

94


rotations about the nodal ‘x’, ‘y’, and ‘z’ axes. Large deflection capabilities are

included in the element (ANSYS12.1, 2010).

Each spider arm was modelled using ‘BEAM4 3-D elastic beam element’ as shown in

Figure 5.5 and was assigned the properties of X-type stainless steel spider arms. Each

spider was connected to the structural support frame using three non-linear spring

elements COMBIN39. Two springs were assigned with high stiffness in the in-plane

horizontal (‘x’ axis) and in-plane vertical (‘y’ axis) directions and the third spring was

assigned about the out-of-plane lateral (‘z’ axis) direction to allow in-plane rotation to

simulate the rotation in the spider arms against friction.

The shank of the bolt fittings were modelled using beam elements BEAM3 and

assigned with the properties of M10 stainless steel bolts. Each bolt fitting was

connected to the spider arms using three non-linear springs. Two springs in the ‘x’

and ‘y’ directions were assigned with load displacement curves representing the gaps

(free translation) and bearing (gap closed) between the spider arms to the structural

support frame connections and the spider arm to the countersunk bolt fittings. The

third spring was assigned in the ‘z’ direction with high stiffness to prevent the out-of

plane movement between the bolts to spider arm connections. The springs connecting

the bolt to spider arm and the spider arm to structural support frame are illustrated in

Figure 5.6.

The counter sunk bolt head was modelled as a cylinder 20mm in diameter 12mm thick

using SHELL95 elements as shown in Figure 5.7. 12mm thick glass panels were

modelled using SHELL95 elements and the silicone sealant interface was modelled

using SHELL181 elements. Size controlled manual meshing method was used and a

very fine mesh was created around the glass hole as shown in Figure 5.7. The front

view of the FE model which shows the mesh of glass panels, bolt heads and sealant

interface is shown in Figure 5.8a. Side view of the model is shown in Figure 5.8b

which shows the structural support frame, glass panels and the spider arms.

95


Front view Side view

Figure 5.5 FE modelling of the structural support frame with spider arms (Test #1)

Bolt fitting

connecting spider

arm to glass

Springs connecting spider

arms to bolt (3-springs)

Springs connecting

spider arms to

structural support

frame (3-springs)

Glass panel

Spider arms

Support structure

Figure 5.6 Non-linear springs (green) connecting spider arms (purple) countersunk

bolt fittings (red) whilst non-linear springs (green) connect spider arms to structural

support frame (blue) (Test #1)

96


Figure 5.7 Mesh of bolt heads, glass panel and sealant at the internal centre spider

arm in the FE model (Test #1)

(a) Front view of the model (b) Side view of the

model

Figure 5.8 ANSYS FE full model (Test #1)

97


5.3.2 Material properties

Four different types of material were used namely; mild steel, stainless steel, glass

and silicon sealant. All materials were defined to be linear elastic and isotropic except

the silicon sealant (Refer Table 5.1 for material properties) which is a hyper elastic

material. The FE modelling of silicone presented challenges due to its non-structural

nature. ANSYS has some built-in hyper elastic material models to represent the non

linear behaviour of silicone sealant like materials and the Blatz and Ko (Bondi and

McClelland, 2009) material model was used and is described in the following section.

Table 5.1 Material properties used in the FE model (Test #1)

Material name Elements Young’s

modulus

Poisson’s

ratio

Steel Structural support frame 200,000 0.30

Stainless steel Spider arms, bolt head and bolt

to connect glass and spider arms 190,000 0.30

Glass Glass panels 70,000 0.23

5.3.3 Material Model for Silicon Sealant

A significant amount of deformation was observed in the silicon sealant during the lab

test although the sealant limited the relative movement of the glass panels. The sealant

was deformed due to tensile, compressive and shear loads between the adjacent panels

and therefore it is important to incorporate this hyper elastic behaviour of the material

in the model. Bondi and McClelland (2009) carried out an analytical study on the

Blatz and Ko model for structural silicon sealants application in glass façade systems

and found a good correlation between the analytical and experimental results. The

Silicon sealant was modelled using the same single parameter continuum model

derived from a study by Blatz and Ko with a representative shear modulus of law

(0.05MPa), medium (0.20MPa) and high (0.50MPa) moduli sealants.

98


The model relies on the shear modulus to model the stiffness and uses a fixed

Poisson’s ratio of 0.463 to capture the incompressible nature of the silicone sealant.

The cross section of the silicon sealant used is shown in Figure 5.9 and the effective

cross section was assumed to be a rectangular 8mm x 8mm section. The Blatz and Ko

model was verified using a simple ANSYS FE model as shown in Figure 5.10 using

three different shear moduli representative of law (0.05MPa), medium (0.20MPa) and

high (0.50MPa) moduli sealants with 8mm x 8mm sealant confined between two

12mm thick 100mm long glass panels. The model was fixed at one side and

compressive, tensile and shear loadings were applied. The tensile and shear behaviour

of the sealants are shown in Figure 5.11 whilst the compressive behaviour of the

sealants are shown in Figure 5.12 and the same properties were adopted in the FE

modelling.

12mm

8mm

Glass panel

Figure 5.9 Cross section of the silicon sealant (Test #1)

Figure 5.10 Test ANSYS FE model of silicon sealant (Test #1)

99


Lo

ad

(N

)

250

200

150

100

50

0

Tensile load 0.05MPa



Shear load 0.05MPa

Shear load 0.20MPa

Shear load 0.5MPa

0 2 4 6 8 10

Deformation (mm)

Figure 5.11 Tensile and shear load displacement results for 8mm thick silicone

sealant from ANSYS FE model (Test #1)

Lo

ad

(k

N)

40

35

30

25

20

15

10

5

0

Compressive load 0.20MPa



0 1 2 3 4 5 6 7 8

Deformation (mm)

Figure 5.12 Compression simulation results for 8mm thick silicone sealant from

ANSYS FE model (Test #1)

100

12


5.3.4 Real Constants for the Elements

Linear real constants (sectional properties) were used for the structural support frame,

spider arms, bolt head, bolt between the glass panel and the spider arms, glass panel

and silicone sealant. All real constants were specified with the actual thickness and

properties and for the spider arms an average second moment of area was specified to

incorporate the tapering effect. Non-linear real constants (spring constants) were

specified for the springs to represent the built-in standard gaps and then bearing

effects (gap closed). Non-linear real constants were also used to represent the

frictional moment versus rotational relationship between the spider arms and

structural support frame connection.

Different locations in the bolted connections were identified that had gaps that

allowed rigid body translation followed by bearing action:

1. At the spider arm to structural support frame connection:

a) The bolt holes on the structural support frame (on PFC’s) were 18mm

in diameter whilst 16mm diameter bolts were used to connect the

spider arms. This gave approximately a ±1mm of gap in both the ‘x’

and ‘y’ directions

b) The same size of gap could occur at the base of the spider arm and the

bolt as illustrated in Figure 5.13 where the hole diameter is 18mm and

the bolt diameter is 16mm.

2. The spider arm to bolt fittings (shank) connection as shown in Figure 5.13

could also had a gap of approximately a ±1 mm in both the ‘x’ and ‘y’

directions for circular holes. The slotted holes in the spider arms were

assumed to allow ±5mm in the ‘x’ direction and ±1mm in the ‘y’ direction.

Therefore in total a ±3mm of gap between the structural support structure and glass

bolt connection could be expected for spider arm circular holes in both the ‘x’ and ‘y’

directions. All these gaps were collectively specified in the model at the spider arm to

bolt fitting (shank) connection. Each of these connections was specified with three

different non-linear translational springs; one in the ‘x’ direction, another in the ‘y’

direction and other in the ‘z’ direction. For the slotted holes a different spring constant

was specified in the ‘x’ direction to allow ±7mm movement. The fine tuned non

101


linear real constants (spring constants) used in the model are shown in Figure 5.14

based on the basic friction theory. Initially, the connection overcomes the static

friction, then rigid body translation together with the dynamic friction occurs and

finally a dramatic increase in stiffness takes place due to bearing contact. A very stiff

spring constant was used in the ‘z’ direction to transfer the axial load from the bolt to

the spider arm.

There were another two translational and one rotational springs used at the spider arm

to structural support frame connection to incorporate the rotational behaviour of the

spider arm against the rotational friction moment (torsion). Springs in the ‘x’ and ‘y’

directions were assigned with a very high stiffness to represent the pin connection

whilst a non-linear real constant was specified for the rotational spring about the ‘z’

axis. An experimental study was conducted in the laboratory to calculate the ultimate

moment capacity (torsion) of the snug tightened spider arm to bolt connections and a

non-linear real constant was proposed with moment (torsion) versus rotation as shown

in Figure 5.15.

Figure 5.13 X-type spider arm with built-in standard gaps indicated (Test #1)

102

9000

Circular hole

7000

Bearing

Slotted hole

5000

Gaps

3000

1000

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7-1000

Initial

friction -3000

Bearing

-5000

-7000

40000

20000

0

-1 -0.5 0 0.5 1

-20000

-40000


Fo

rce (N

)

-8 8

-9000

Displacement (mm)

Figure 5.14 Real constants (spring properties) used for the rigid body translation

and bearing at the connections (Test #1)

60000

Mo

men

t (N

m)

-1.5 1.5

-60000

Rotation (rad)

Figure 5.15 Real constants (spring properties) used for the rigid body rotation about

the ‘z’ axis at the spider arm to structural support frame connections (Test #1)

103


5.3.5 Boundary Conditions and Loading

The structural support frame was pin connected at the base by restraining the ‘x’, ‘y’

and ‘z’ directions at the base nodes. The structural support frame and the spider arm

bases were also restrained in the ‘z’ direction to prevent out-of-plane movement. The

spider arms were restrained from rotation about the ‘x’ and ‘y’ axis at the spider arm

bases and the M10 bolts connecting the glass panel to the spider arms were also

restrained from rotation about the ‘x’ and ‘y’ axes (Figure 5.16). A 60mm racking

displacement was applied at the top of the structural support frame and a large

displacement analysis was performed to incorporate the geometrical non-linearity.

The displacement was applied in a step by step manner representing a typical

pushover test.

Base of the spider

arm restrained in

the ‘z’ direction

Bolt restrained

from rotation

about the ‘x’ and

‘z’ axes

Glass panel

Support

structure

y

z

Figure 5.16 Boundary conditions at the central spider arms (Test #1)

5.4 Test #1 - Results Comparison

The nominal tensile strength of 12mm toughened glass at the bolt holes was assumed

to equal 94MPa as per AS1288 (AS1288, 2006). It was estimated that the out-of-plane

bending stress was the major component responsible for the failure of the glass panels

104


which was directly affected by the properties of the spider arms. When a low value of

second moment of area was used for the spider arms, the tensile stress developed in

the glass panels was high due to the excessive out-of-plane deformation. Fine tuning

of basic properties was necessary to bench mark against all the experimental results.

The FE model was calibrated to the failure stress of 94MPa and a pushover curve was

developed.

The translation and rotation of the FE model is shown in Figure 5.17 at an in-plane

racking displacement of 49mm (2.0% drift) at the point of failure of a glass panel

when the glass stress exceeded 94MPa. Interestingly, the in-plane differential

movement of the glass panels relative to each other in the vertical and horizontal

directions was very similar to the test results. Similarly, the deformation of the silicon

sealant at the internal spider arm also correlated well with the test results.

Figure 5.18 illustrates the differential out-of-plane movement of the glass panels due

to the out-of-plane deformation of the spider arms. The blue contours represent the

minimum negative deformation and the red contours represent the maximum positive

deformation. The maximum differential movement occurred between spider arms

PBB4 (Panel PB spider arm B4) to PDB2 which is very similar to the experimental

results. The out-of-plane movement around the internal central spider arm was not

symmetrical due to the combination of circular holes and slotted holes in the spider

arm. A significant amount of out-of-plane movement was observed between arms

PBB4 to PDB2 and PAB3 to PCB1 with a maximum differential movement of

approximately 8mm which is comparable to the maximum measured in the test of

8.5mm.

Rotation of the spider arms are shown in Figure 5.21. Directions of rotation are

similar to the predicted racking mechanism and the maximum rotation is at the corner

spider arm PDB4 which is almost equal to the experimental results. The tensile stress

developed on the glass panels are illustrated in Figure 5.19 (front side) and Figure

5.20 (back side). According to the racking mechanism, the maximum tensile stress

should be developed along the tensile diagonals (B1 to B4). Since all the corner spider

arms rotate, the possible diagonals are PBB1 to PBB4 and PCB1 to PCB4. In the

model, the maximum tensile stress occurred at PCB4 as shown in Figure 5.19. The

105


pushover curve from the analytical result is compared with the experimental result

and illustrated in Figure 5.22 and shows a very good correlation. The failure stress in

the glass panel developed at a lateral displacement of 49mm (2.0% drift) of

displacement compared to 51mm (2.1% drift) in the experiment with a lateral force of

15.5kN as shown in Figure 5.23.

Figure 5.17 Translation of the glass panels (Test #1)

Figure 5.18 Out-of-plane movement (in mm) of the glass panels (Test #1)

106


Figure 5.19 Maximum principle tensile stress (in MPa) developed - front face

(Test #1)

Figure 5.20 Maximum principle tensile stress developed (in MPa) - back face

(Test #1)

107

Chapter 5 FINITE ELEMENT MODELLING OF THE IN-PLANE RACKING PERFORMANCE OF PFGFS In

-pla

ne l

oad

(k

N)

Figure 5.21 In-plane rotation of the spider arms (in radians) at failure (Test #1)

In-plane displacement (mm)

0 7.2 14.4 21.6 28.8 36 43.2 50.4 57.6

20 20

18 18

Test 1- Experiment 16 16

ANSYS FE model 14 14

12 12

10 10

8 8

6 6

4 4

2 2

0 0

0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4

In-plane drift (%)

Figure 5.22 Experimental and analytical pushover curves (Test #1)

108


Te

ns

ile s

tre

ss (

MP

a)

120

100

80

60

40

20

0

Failure of glass panel at 94MPa

0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4

In-plane drift (%)

Figure 5.23 Maximum tensile stress developed at Bolt PCB4 (Test #1)

5.5 Test #1 – Effect of the Diagonal Strut Loads

The glass to countersunk bolt head connection detail was simplified and represented

as a cylindrical hole as described in Section 5.2.2. The simplified model results

showed that the maximum tensile stress was formed at bolt hole PCB4. However, in

the model the in-plane diagonal strut loads were initially uniformly distributed around

the glass bolt holes through the continuous FE mesh and may have under predicted

the maximum tensile stress developed. A more complex model was then developed

using a contact interface between the bolt head and the glass to predict the actual

effect on maximum tensile stress developed in the glass panel. In-plane node-to-node

contact elements were defined using non-linear springs COMBIN39 between the bolt

head and the glass interface and the perimeter nodes were coupled in the out-of-plane

direction to form the bending effect as shown in Figure 5.24. The springs were

assigned with a very high stiffness in compression (spring shortening) and low

stiffness in tension (spring expansion) representing the bearing effect.

109


Figure 5.24 Contact elements assigned to connect bolt head to glass bolt hole

(Test #1)

The remainder of the FE model was the same and the analysis was performed again.

The maximum tensile stresses were compared between the simple and more complex

model at the same glass hole PCB4. In the simplified model, tensile stress was

recorded at a radius of 10mm whilst the stress from the more complex interface model

was noted at 12.5mm radius to avoid the unrealistically high stress concentration at

the spring connected nodes. Very similar results were obtained from both models.

Hence, from this comparison, it is considered reasonable that the simplified model is

capable of predicting the in-plane racking displacement capacity of PFGFS with X-

type spider arms and countersunk bolt fittings.

110


5.6 Test #2 - Structural Idealisation

5.6.1 Test #2 - Racking Mechanism

It was observed during the experimental test that the glass panels translated and the

spider arms moved vertically within the slotted hole connections along the cleats of

the structural support frame whilst damaged to the vertical silicone sealant joint

commenced at 2.0% drift. During the racking action a certain amount of rigid body

translation followed by bearing action in both the horizontal and vertical directions

was observed at the spider arm to glass connection details, where the built-in standard

gaps allowed. The combined rigid body translation was observed up to 3.3% drift

whilst at greater drift the spider arms began to yield and deform. The glass panels also

displaced in the out-of-plane direction relative to each other with a maximum

differential movement of approximately 10mm.

The racking load applied to the structural support frame was transferred to the glass

panel via the spider arms resulting in diagonal tension and compression forces as

shown in Figure 5.25. Beyond 3.3% drift, there was no possibility for further rigid

body translation and consequently the spider arms in tension distorted sideways whilst

the spider arms in compression tended distort in the opposite out-of-plane direction

due to the eccentricity as described in Test #1, Section 5.2.1. This induced combined

local bending and axial tensile stresses in the glass panels, particularly around the bolt

holes, before catastrophic failure at a large drift of 5.25%.

111


COMPRESSION TENSION

B1

B3 B4

B2

Glass panel (PA)

B1

B3 B4

B2

Glass panel (PB)

B1

B3 B4

B2

Glass panel (PD)

B1

B3 B4

B2

Glass panel (PC)

Figure 5.25 Glass panels and spider arms configuration including the sliding

directions of the spider arms bases (Test #2)

5.6.2 Model Assumption

Failure of the glass panel occurred due to the combined bending and axial tensile

stresses. Therefore in the FE model both bending and axial tensile effects from the

button head bolts were incorporated as described in Test #1. In the M10 button head

bolt fittings a 20mm diameter Nylon bush was inserted into the cylindrical glass hole

and the glass panel was clamped by two 55mm diameter, 6.8mm thick stainless steel

disks to transfer the wind load. Therefore the diagonal strut loads were transferred via

the bushes and the bending effects were created by both the bush and disks.

Conservatively, the bolt was assumed to be a 20mm diameter 12mm thick cylinder as

assumed in Test #1, Figure 5.26.

Shell elements, 12mm thick were used to model the glass panels and bolt heads. The

bolt heads and the glass panels were meshed continuously (i.e. glued) with relevant

material properties. Similar to Test #1 bolt shanks were modelled using elastic

‘BEAM 3’ element. However the 20mm diameter bolt head created very high local

bending stress concentration on the glass panels. Therefore the stresses on the glass

112


holes were measured at a radius of 15mm and benchmarked with the experimental

results.

M10 button head bolt connection Hole detail after assumption

Figure 5.26 Schematic diagram of button head bolt connection used in the FE model

(Test #2)

5.6.3 Features of the Model

The vertical PFCs of the structural support frame to the spider arms connections were

modified to allow the vertical sliding and out-of-plane movement of the spider arms

otherwise, all other features were the same as described in Test #1.

5.7 Test #2 - Model Description

Similar modelling techniques and elements to Test #1 were used in the Test #2 model

except the spider arms to structural support frame connections.

5.7.1 Element Description

K-type stainless steel spider arms spider arms were modelled using ‘BEAM4 3-D

elastic beam element’ as shown in Figure 5.27. The base plates of the spider arm were

also modelled using elastic beam elements and were connected to the structural

support frame using three different non-linear spring elements, COMBIN39. Each of

these connections were specified with three different non-linear springs; one each in

the ‘x’ (in-plane horizontal), ‘y’ (in-plane vertical) and ‘z’ (out-of-plane) directions as

shown in Figure 5.27. The spring in the ‘y’ direction represented the vertical sliding

whilst the spring in the ‘z’ direction allowed the base plate to move horizontally in the

out-of-plane direction and the third spring restrained the movement of the base plate

113


in the ‘x’ direction. The front elevation of the FE model, shown in Figure 5.28a shows

the mesh of glass panels, bolt heads and sealant interface whilst the side view, shown

in Figure 5.28b shows the structural support frame, glass panels and the spider arms.

Test - front view Test - side view

FE model - front view FE model - side view

Figure 5.27 FE modelling of the structural support frame with K-type spider arms

(Test #2)

Springs connecting

spider arms to

structural support

Spider arm base plates

114


x

(b) Side view of the

y (a) Front view of the model model

Figure 5.28 ANSYS finite element model (Test #2)

5.7.2 Material Properties

All materials were defined to be linear elastic and isotropic other than the silicon

sealant (Refer Table 5.2 for material properties) as described in Test #1 where the

same Blatz and Ko (Bondi and McClelland, 2009) material model was used. A

representative shear modulus value of 0.10MPa was specified for the sealant material

properties.

Table 5.2 Material properties used in the FE model (Test #2)

Material name Elements Young’s

Modulus (MPa)

Poisson’s

ratio

Steel Structural support frame 200,000 0.30

Stainless steel Spider arms, bolt head and bolt

to connect glass and spider arms 190,000 0.30

Glass Glass panels 70,000 0.23

115

25,000

20,000

15,000

10,000

5,000

0

-15 -10 -5 0 5 10 15

-5,000

-10,000

-15,000

-20,000


5.7.3 Real Constants for the Elements

The spider arm base plate was assigned the actual dimensional properties whilst the

K-type spider arms were assigned average properties given the tapering geometry.

Non-linear real constants (spring stiffness) were proposed for the spring elements

connecting the spider arm base plate to the structural support frame. The spring in the

‘y’ direction was specified to allow ±17.5mm vertical movement followed by high

stiffness to represent bearing contact as shown in Figure 5.29. Similarly the spring in

the ‘z’ direction was specified to allow ±7mm movement before bearing contact on

Figure 5.30. The third spring in the ‘x’ direction was specified with a high stiffness to

restrain the base plate in this direction.

Collectively, ±3mm of gap was expected at the bolt fitting to spider arm and bolt

fitting to glass panel resulting from: (1) bolt to Nylon bush at the spider arm hole; (2)

bolt to Nylon bush at the button head bolt connections; and (3) Nylon bush to glass

hole at the glass to button head bolt connections as shown in Figure 5.31. Therefore in

total (similar to Test #1) ±3mm of rigid body translation was assigned in both the ‘x’

and ‘y’ directions for the circular hole and a ±7mm rigid body translation in the ‘x’

direction and ±3mm in the ‘y’ direction were assigned for the slotted holes as shown

in Figure 5.14.

20

Figure 5.29 Real constants used for the vertical sliding (‘y’ direction) of the spider

arms and bearing at the structural support frame to the spider arm base plate

connections (Test #2)

Lo

ad

(N

)

-20

-25,000

Vertical sliding (mm)

116

20000

10000

0

-8 -6 -4 -2 2 4 6 80

-10000

-20000


30000

Lo

ad

(N

) -10 10

-30000

Out-of-plane movement (mm)

Figure 5.30 Real constants used for the out-of-plane movement (‘z’ direction) of the

spider arms and bearing at the structural support frame to the spider arm base plate

connections (Test #2)

Figure 5.31 Locations leading to gaps in button head bolt fitting (Test #2)

5.7.4 Boundary Conditions and Loading

The same boundary conditions as described in Test #1 were used in Test #2. Racking

displacements up to 125mm (5.2% drift) were applied at the top of the structural

support frame with a number of small load steps and a large displacement analysis

was conducted to incorporate the geometrical non-linearity.

117


5.8 Test #2 - Results Comparison

The nominal tensile strength of 12mm toughened glass at the bolt holes was assumed

to be 94MPa as per AS1288 (AS1288, 2006). The FE model was calibrated to the

failure stress of 94MPa and a pushover curve was developed. The glass panels and the

spider arms configurations and the sliding mechanism of the spider arms are

illustrated in Figure 5.25. The translations and rotations of the FE model is shown in

Figure 5.32 at the failure of a glass panel at a lateral displacement of 114mm (4.75%

drift). Interestingly, the in-plane differential movement of the glass panels relative to

each other in the vertical and horizontal directions were very similar to the

experimental results. Similarly, the deformation of the silicon sealant at the internal

spider arm was also very similar to the experimental test results.

The contour plot in Figure 5.33 illustrates the differential out-of-plane movement of

the glass panels due to the deformation of the spider arms. The blue contours

represent the minimum negative deformation and the red contours represent the

positive maximum deformation. Maximum differential movement occurred between

spider arms PBB4 to PDB2 which is very similar to the experimental results. A

significant amount of out-of-plane movement was observed between arms “PBB4

(Panel PB spider arm B4) and PDB2” and “PAB3 and PCB1” with a maximum

differential movement of approximately 10mm which is equal to the value measured

experimentally. The vertical sliding of the spider arms is shown in Figure 5.34 which

strongly agrees with the experimental results. A similar good correlation between the

analytical and experimental deformations in the base plates of the spider arms also

was observed.

The tensile stress developed in the glass panels are illustrated in Figure 5.35 (front

face of the glass panels) and Figure 5.36 (back face of the glass panels). Interestingly,

as observed in the experimental test, the maximum tensile stress occurred at PCB4, as

indicated in Figure 5.35. The pushover curve from the analytical result is compared

with the experimental result and shows good correlation as illustrated in Figure 5.37.

The failure stress in the glass panel developed at a lateral displacement of 114mm

(4.75% drift) compared to 5.25% drift in the experiment with a lateral force of 35kN

as shown in Figure 5.38.

118


Figure 5.32 The deformed model after reaching the failure stress at 4.75% drift

(Test #2)

Figure 5.33 Out-of-plane movement (in mm) of the glass panels at 4.75% drift

(Test #2)

119


Figure 5.34 Spider arms deformation and vertical translation (in mm) in the spider

arms at 4.75% drift (Test #2)

Figure 5.35 Maximum principle tensile stress (in MPa) developed at 4.75% drift

front face of the glass panels (Test #2)

120


Figure 5.36 Maximum principle tensile stress (in MPa) at 4.75% drift - back face of

the glass panels (Test #2)

In-plane displacement (mm)

0 12 24 36 48 60 72 84 96 108 120 132

0

5

10

15

20

25

30

35

40

0

5

10

15

20

25

30

35

40

In-p

lan

e lo

ad

(kN

)

Test #2- Experiment

Test #2 - FE analysis

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5

In-plane drift (%)

Figure 5.37 Experimental and analytical pushover curve benchmarked (Test #2)

121


Te

ns

ile

str

ess

(M

Pa)

120

100

80

60

40

20

0

Faiure of glass

panel at 4.75%

0 1 2 3 4 5

In-plane drift (%)

Figure 5.38 Maximum tensile stress developed at the glass hole PCB4 (Test #2)

5.9 Test #2 – Effect of the Diagonal Strut Loads

The glass to button head bolt connection detail FE model was simplified as described

in the FE model in Section 5.6.2. The simplified model results showed that the

maximum tensile stress was formed at bolt hole PCB4. However, in the model the in-

plane diagonal strut loads were initially uniformly distributed around the glass bolt

holes through the continuous FE mesh and may have under predicted the maximum

tensile stress developed. A more complex model was then developed using a simple

contact interface between the same bolt head to glass hole to predict the actual effect

on maximum tensile stress developed in the glass panel similar to Test #1, Section

5.5.

In the simplified model tensile stress was recorded at a radius of 15mm whilst the

stress from the more complex interface model was noted at 17.5mm radius to avoid

the unrealistically high stress concentration at the spring connected nodes. Very

similar results were obtained from both models. Hence from this benchmarking

exercise it is considered reasonable that the simplified method is capable of predicting

the in-plane racking displacement capacity of PFGFS with K-type spider arms and

button head bolt fittings.

122

6


5.10 Summary and Conclusions

A sophisticated non-linear finite element models were developed and conservatively

benchmarked against experimental results with excellent correlation. For the Test #1,

the FE model can be confidently applied for different types of PFGFS with X-type

spider arms and countersunk bolt fittings by changing the dimensions and properties.

The same FE model was modified at the spider arm to structural support frame

connection to analyse the racking performance of Test #2 with the K-type spider arms

and button head bolt fittings. The Test #2 FE model also bench marked against the

experimental results with excellent correlation. The FE model can be confidently

applied to different types of PFGFS with K- type spider arms and button head bolt

fittings by changing the dimensions and properties.

In the FE models, load-deflections curves, out-of-plane deformation of the glass

panels, spider arm deformations and the failure modes were obtained very similar to

the experimental results. The models were used for the parametric study and

described in the following chapter.

123

Chapter 6 PARAMETRIC STUDY ON PFGFS USING FE ANALYSES

Chapter 6

6 PARAMETRIC STUDY ON PFGFS USING FE

ANALYSES

6.1 Introduction

The FE models benchmarked against the experimental test results for Tests #1 and

Test#2 in Chapter 5 were used to predict the racking performance of PFGFS with

different configurations. Variations related to built-in standard gap, sealant type,

sealant thickness, glass geometry, and glass thickness were considered in the

parametric study for 2x2 façade grid systems. Further detailed FE analyses were

conducted to evaluate the individual contributions of each racking mechanism to the

total racking displacement for the 2x2 and multiple grid systems.

Moreover, a parametric study on multiple façade grid systems was conducted and

detailed FE analyses were carried out to evaluate individual contributions for each

racking mechanism. Further, the racking performance of PFGFS with rigidly

connected ‘X’ type spider arms (i.e., in-plane rotation prevented) were also discussed

and special articulations were introduced to increase the in-plane raking capacity of

such systems. The results and discussions for racking performance of Test #1 with ‘X’

type spider arm configurations are discussed through Sections 6.2 to 6.6 and repeated

for Test #2 with ‘K’ type spider arm configurations through Sections 6.6 to 6.9.

6.2 Test #1 - Parametric Study for 2x2 Grid Systems

6.2.1 Test #1 – Built-in Standard Gaps at the Structural Support Frame

A standard gap (which is generally a 20 mm x 30 mm slotted hole on the structural

support frame) is provided at the ‘X’ type spider arm to structural support frame

connection in typical PFGFS applications. In Test #1, M16 bolts were used to fix the

‘X’ type spider arms into 18mm diameter holes instead of the standard slotted holes to

check the racking performance of the spider arms, glass panels and the bolted

124


connection system only. The FE model developed and calibrated in Chapter 5 to the

experimental racking results of Test #1 was then modified to incorporate the slotted

holes (20 mm x 30 mm) at the structural support frame as illustrated in Figure 6.1.

Built-in standard gaps in the ‘x’ and the ‘y’ directions were assigned in the model

using non-linear springs which connected the spider arms to the structural support

frame with the stiffness relationship as shown in Figure 6.2.

The FE analysis with the slotted holes at the structural support frame indicated that

the ultimate in-plane drift capacity of the PFGFS increased from 2.1% to 3.1% before

failure at 94MPa tensile stress in the glass panel. The pushover curve and the

maximum tensile stresses developed on the glass panel are compared in Figures 6.3 &

6.4 for the Test #1. The racking performance of the system improved upon the

incorporation of the slotted holes and therefore, it is evident that the provision of

additional articulation at the structural support frame increases the racking

performance of PFGFS with ‘X’ type spider arms.

180PFC 180PFC

Ø18mm

drill hole

20x30mm

slotted hole

Test #1 Parametric studyx

y

Figure 6.1 Schematic diagram of the holes provided at the structural support frame

(Test #1)

125

7000 High stiffness

attributed to

bearing of bolt on

'Y' direction 'X' direction 5000 edge holes

3000

Slack due to large hole size 1000 compared to bolt diameter

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7-1000

Initial stiffness due

-3000 to snug tight bolt

installation

High stiffness

attributed to -5000 bearing of bolt on

edge holes -7000


9000

8

-9000

Displacement (mm)

Figure 6.2 Real constants (non-linear spring constant) used for the spring elements

to represent the translation and bearing at the spider arm to structural support frame

connection

18

Fo

rce

(N

)

-8

In-p

lan

e lo

ad

(kN

)

16

14

12

10

8

6

4

2

0

Test #1FE 18mm diameter hole

Test #1 with slotted hole 20x30mm

Failure of glass panel at 2.1%


0 0.4 0.8 1.2 1.6 2 2.4 2.8

In-plane drift (%)

Figure 6.3 Analytical pushover curve comparison for the models with circular

hole and slotted hole at the structural support frame (Test #1)

126

3.2


Te

ns

ile s

tre

ss (M

Pa

)

120

110

100

90

80

70

60

50

40

30

20

10

0

Test #1 FE 18mm diameter hole

Test #1 with slotted hole 20x30mm



0 0.5 1 1.5 2 2.5 3 3.5

In-plane drift (%)

Figure 6.4 Comparison of the tensile stresses developed at the FE model with

circular hole and slotted hole at the structural support frame (Test #1)

6.2.2 Test #1 - Sealant Types

Three different types of silicon sealants are available in the market namely; structural,

weather and special purpose sealants. Structural sealant has a high shear modulus

compared to the weather sealant and is applied if the sealant joint is required to

transfer stresses whereas weather sealant is used for weather proofing only. Special

purpose sealants have low shear modulus and can resist high displacement or

expansion at the joint. A weather sealant was used in between glass panels in Test #1

with a typical shear modulus of 0.20MPa. A parametric study was conducted using

the FE model to compare the racking performance of the façade system with weather,

structural and special purpose sealants. The typical shear moduli of the sealants

assigned in the FE models are presented in Table 6.1.

The sealant had a significant effect on the racking performance of the PFGFS with the

ultimate in-plane drift capacity of 1.7%, 2.1% and 2.3% estimated from the high

modulus (structural), medium modulus (weather) and low modulus (special purpose)

sealants respectively. In each case, failure was defined when the tensile stresses in the

glass panels exceeded 94MPa. The pushover curves for low, medium and high

modulus sealants are illustrated in Figure 6.5 and the tensile stresses developed

against the applied drift are illustrated in Figure 6.6. The pushover curves showed that

127


a high in-plane lateral load had to be applied when higher modulus sealant was used.

This enhanced the out-of-plane deformation in the spider arms and the tensile stress

development in the glass panel thereby reducing the racking performance of the

system compared to the other sealants.

Table 6.1 Properties of the sealant types used in the FE analysis

Sealant type Shear Modulus

(MPa)

Grid

Arrangement

Glass Panel

Dimension (mm)

Weather (Medium modulus) 0.20 2x2 1200x1200

Structural (High modulus) 0.50 2x2 1200x1200

Special purpose (Low

modulus) 0.05 2x2 1200x1200

In-p

lan

e lo

ad

(kN

)

25

20

15

10

5

0

Test #1 FE Medium Modulus

Test #1 FE Low Modulus

Test #1 FE High Modulus




0 0.4 0.8 1.2 1.6 2 2.4 2.8

In-plane drift (%)


modulus silicon sealants (Test #1)

128


Te

ns

ile s

tress (M

Pa)

120

110

100

90

80

70

60

50

40

30

20

10

0




Failure of glass panel

0 0.4 0.8 1.2 1.6 2 2.4 2.8

In-plane drift (%)

Figure 6.6 Comparison of the tensile stresses developed in the FE models with low,

medium and high modulus silicon sealants (Test #1)

6.2.3 Test #1 - Sealant Thickness

An 8mm thick weather sealant was used in Test #1 with a typical shear modulus of

0.20MPa and a parametric study was then conducted using the FE model to compare

the racking performance with 6mm and 10mm thick weather sealants. The sealant

thickness had a significant effect on the racking performance of the PFGFS with the

ultimate in-plane drift capacity of 1.9%, 2.1% and 2.2% estimated from the 6mm,

8mm and 10mm thick sealants respectively. In each case, failure was defined when

the tensile stresses in the glass panels exceeded 94MPa. The pushover curves for

6mm, 8mm and 10mm thick sealants are illustrated in Figure 6.7 and the tensile

stresses developed against the applied drift are illustrated in Figure 6.8. The pushover

curves showed that a high in-plane lateral load had to be applied for failure of a glass

when a thicker sealant was used. However, when a thicker 10mm sealant was used,

the out-of-plane deformation in the spider arms and the tensile stress development in

the glass panel were reduced, thereby increasing the racking performance of the

system compared to the 6mm and 8mm thick sealants.

129


0

5

10

15

20

25

In-p

lan

e lo

ad

(kN

)

Test #1 FE 6mm Thick Sealant






0 0.4 0.8 1.2 1.6 2

In-plane drift (%)

Figure 6.7 Analytical pushover curve comparison with 6mm, 8mm and

10mm thick silicon weather sealants (Test #1)

Te

nsile s

tre

ss (M

Pa

)

120

110

100

90

80

70

60

50

40

30

20

10

0





0 0.4 0.8 1.2 1.6 2

In-plane drift (%)

Figure 6.8 Comparison of the tensile stresses developed in the FE models with 6mm,

8mm and 10mm thick silicon weather sealants (Test #1 )

130

2.4

2.4


6.2.4 Test #1 - Glass Geometry

The Test #1 FE model was modified and analyses were performed for a 2x2 system

consisting of 1200mm wide and 2400mm high panels (portrait panel system) and a

2x2 system consisting of 2400mm wide and 1200mm high panels (landscape panel

system). The racking performance of the portrait and landscape panel systems was

compared with the square panel system of the Test #1. The pushover curves for

portrait, square and landscape panel systems are illustrated in Figure 6.9 whilst the

maximum tensile stresses developed at the glass holes against the applied drift for the

square, portrait and landscape systems are illustrated in Figure 6.10.

In each case, failure was defined when the tensile stresses in the glass panels

exceeded 94MPa. The FE analyses results showed that, the drift capacity of the

system was dependent on the aspect ratio with drift capacity values of 1.4%

(landscape), 2.1% (square) and 2.3% (portrait) recorded. When the landscape system

was used the in-plane racking load at failure of a glass panel was exceptionally high

(38kN). In practice, a strong structural support structure would need to sustain the

higher load to achieve this drift capacity (1.4%).

In-p

lan

e lo

ad

(kN

)

45

40

35

30

25

20

15

10

5

0 0 0.4 0.8 1.2 1.6 2 2.4 2.8

Test #1 FE 1200(w)x1200(h)

Test #1 FE 1200(w)x2400(h)

Test #1 FE 2400(w)x1200(h)




In-plane drift (%)

Figure 6.9 Analytical pushover curve comparison of the square, portrait and

landscape panel systems (Test #1)

131


Ten

sile

str

ess

(M

Pa)

120

110

100

90

80

70

60

50

40

30

20

10

0

Test #1 FE 1200(w)x1200(h)

Test #1 FE 1200(w)x2400(h)

Test #1 FE 2400(w)x1200(h)

Failure of glass panel at 1.4% Failure of glass

panel at 2.1% Failure of glass panel at 2.3%

0 0.4 0.8 1.2 1.6 2 2.4 2.8

In-plane drift (%)

Figure 6.10 Comparison of the tensile stress developed at the square, portrait and

landscape panel systems (Test #1)

6.2.5 Test #1 - Glass Thickness

The thickness of the glass panels in the Test #1 FE model were modified from 12mm

to 10mm and 15mm (standard glass thicknesses) and the analyses were repeated. The

nominal strength at holes for 12mm thick glass panel is 94MPa in accordance with

AS1288 (2006). Similarly the nominal strengths of the 10mm and 15mm glass panels

are 98MPa and 90MPa respectively and used as the reference for the failure of the

glass panels in the respective analyses. The pushover curves for different thickness of

glass panels (10mm, 12mm and 15mm) are compared in Figure 6.11 and the tensile

stresses developed against the applied drift are illustrated in Figure 6.12. The FE

results indicated that when the glass thickness increased the drift capacity also

increased marginally since the thicker glass panels were stiffer against the out-of

plane deformation of the spider arms and reduced the development of bending

stresses.

132


0

2

4

6

8

10

12

14

16

18

20

In-p

lan

e lo

ad

(k

N)

Test #1 FE 10mm Thick Glass






Te

ns

ile

str

es

s (M

Pa

)

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4

In-plane drift (%)

Figure 6.11 Analytical pushover curve comparison for 10mm, 12mm and

15mm thick glass panels (Test #1)

120

110

100

90

80

70

60

50

40

30

20

10

0




Failure of glass panels

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4

In-plane drift (%)

Figure 6.12 Comparison of the tensile stresses developed for 10mm, 12mm and

15mm thick glass panels (Test #1)

133


6.3 Test #1 - Racking Performance of 2x2 Systems

6.3.1 Test #1 - Discussion of the Parametric Study for 2x2 Systems

The parametric study indicated that in Test #1 the racking performance of the PFGFS

increased when:

1) Slotted holes (standard gaps) were introduced at the structural support frame

2) The stiffness of the silicone sealant was reduced

3) The silicone sealant thickness was increased

4) The height to width ratio of the glass panels was increased

5) The glass thickness was increased.

FE results showed that in all types of 2x2 systems the applied racking displacement

was accommodated by three mechanisms as discussed in Chapters 4 & 5:

1) Rigid body rotation of the spider arms;

2) Rigid body translation at the built-in standard gaps and

3) Deformations of the spider arms.

Therefore, the in-plane drift capacity of a 2x2 system with spider arms arranged

similar to Test #1 can be estimated using the following equation:

DT = DRBR + DRBT + DD Eq (6.1)

Where, DT is the total drift capacity, DRBR is the drift capacity due to the rigid body

rotation of the spider arms, DRBT is the drift capacity due to rigid body translation at

the built-in standard gaps and DD is the drift capacity due to deformation of the spider

arms. By modifying the Test #1 FE model these three components can be quantified.

In the following sections the drift contributions from DRBR , DRBM and DD are

calculated using the FE analyses by modifying the Test #1 FE model appropriately.

a) Drift contribution from the spider arm rotation (DRBR)

The spider arms were restrained against the rotation about the ‘z’ axis at the base

connection of the Test #1 FE model and the analysis was repeated to evaluate the

racking performance of the system due to the rigid body rotation of the spider arms.

The analytical pushover curve comparison is shown in Figure 6.13 whilst the

134


maximum tensile stress developed against the applied drift is shown in Figure 6.14.

When the spider arms were restrained, the ultimate drift capacity reduced from 2.0%

(the model was benchmarked to 2.0% at a failure of glass panel) to 1.6%. Therefore,

the contribution from the provision of the spider arm rotation to Test #1 was 0.4%

drift (2.0% - 1.6%).

20

In-p

lan

e lo

ad

(k

N)

18

16

14

12

10

8

6

4

2

0

Test #1 FE Spider Rotation Allowed

Test #1 FE Spider Rotation Restrained



0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

In-plane drift (%)

Figure 6.13 Analytical pushover curve comparison for the Test #1 with and without

spider rotation restrained

100

90

Te

ns

ile s

tre

ss (M

Pa)

80

70

60

50

40

30

20

10

0

Test #1 FE Spider Rotation Allowed

Test #1 FE Spider Rotation Restrained

Failure of glass panels

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

In-plane drift (%)

Figure 6.14 Comparison of the tensile stresses developed for the Test #1 with and

without spider rotation restrained

135


b) Drift contribution from the built-in standard gaps (DRBT)

The Test #1 FE model was modified to evaluate the racking performance from the

rigid body translations due to the built-in standard gaps only. High stiffness properties

were assigned to the spider arms and the structural support frame elements (to prevent

the deformation) and the spider arms were restrained against the rotation about the ‘z’

axis. The FE model including the glass panels was restrained in the ‘z’ direction to

prevent the out-of-plane deformation of the structure. The analysis was repeated and

the pushover curves compared as shown in Figure 6.15 whilst the maximum tensile

stress developed against the applied drift is shown in Figure 6.16. When only the rigid

body translation at the built-in standard gaps was allowed, bearing occurred at 1.1%

drift which reduced the ultimate drift capacity of the system from 2.0% to 1.1%.

Therefore, the drift contribution from the rigid body translation due to the provision of

built-in standard gap at the bolted connections was 1.1% and the drift contribution

from the spider arms deformation, DD was 0.5% (2.0% - 1.1% - 0.4%).

A simple trigonometric calculation could be used to calculate the drift capacity due to

the rigid body translations at the built-in standard gaps provided at the bolted

connections were symmetrical. In the Test #1 FE model (Chapter 5), the built-in

standard gaps were provided collectively at the spider arms to bolt fitting connections.

The built-in standard gap provided at the Test #1 FE is shown in Figure 6.17. The

bolted connections were assigned ±3mm built-in standard gaps in both ‘x’ and ‘y’

directions except the bolts PAB4 and PBB3 which were assigned ±7mm in the ‘x’

direction and ±3mm in the ‘y’ direction. To introduce symmetry (which will allow for

simple trigonometric calculation), the built-in gaps at the bolts PAB4 and PBB3 were

modified to ±3mm gaps in the ‘x’ direction in the FE model and the FE analysis was

repeated.

Interestingly, similar results to the previous rigid body translation analysis (Figures

6.15 & 6.16) were obtained and the bearing contact occurred at 1.1% drift. When the

2x2 system is racked as shown in Figure 6.18 the system expanded diagonally due to

the rigid body translation and accommodated the racking displacement whilst the

opposite diagonals were in compression. This indicates that the diagonals in tension

expand and diagonals in compression contract due to the rigid body translation.

136


However, the maximum expansion or shortening in the diagonal length is limited by

the built-in gaps provided at the connections. The racking mechanism as illustrated in

Figure 6.19, was simplified using the Pythagoras Theorem as described in the

following equations:

c1 = b2 + h2

c2 = (b + D)2 + h2 and

2 2 2 2 c2 − c1 = (b + D) + h − b + h Eq (6.2)

Where, ‘c1’ is the initial diagonal length between the extreme bolted connections on

the glass panels as shown in Figure 6.19; ‘c2’ is the final diagonal length (after the

racking action) between the extreme bolted connections on the glass panels; ‘b’ is the

horizontal distance between the extreme bolted connections on the glass panels; ‘h’ is

the vertical distance between the extreme bolted connections on the glass panels; and

‘D’ is the possible racking displacement due to rigid body translation only. Maximum

possible translation in the diagonal length as shown in Figure 6.18 due to rigid body

translation at the built-in standard gap provided is 17mm (rigid body translation

allowed diagonally at one bolt is .(3 + 3 ) and therefore, for 4 bolts along the

diagonal 4x .(3 + 3 ) ) when the spider arms PAB1 to PDB4 are aligned to

maximise the diagonal distance.

Therefore, c2 – c1 = 17mm for the example shown in Figure 6.18 and by applying

Equation 6.2 the racking displacement ‘D’ can be calculated:

17 = .(2200 + D ) − .(2200 + 2200 ) , hence D = 24mm

This indicates that the system can accommodate a racking displacement of 24mm

(1.1%) between two vertical extreme bolted connections (h = 2200mm) on the glass

panels due to the rigid body translation only. Drift capacities of 2x2 panel

arrangements with different size of glass panels are presented in Table 6.2. This

shows that the orientation and size of the glass panels influences the drift capacity due

to the rigid body translation at the built-in standard gaps. Moreover, the racking

performance due to only the rigid body translation can be increased further by

137


providing larger gaps for the bolts at the connections between the glass and spider arm

and/or at the connections between the spider arms and support structure as discussed

in Section 6.2.1.

50

45

In-p

lan

e lo

ad

(k

N)

40

35

30

25

20

15

10

5

0

Test #1 FE Rigid Body Movement Only

Test #1 FE

Bearing at the bolt holes occourd at 1.1% drift


0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

In-plane drift (%)


rigid body translation only

Ten

sile

str

ess

(M

Pa

)

120

110

100

90

80

70

60

50

40

30

20

10

0

Test #1 FE Rigid Body Movement Only

Test #1 FE

Bearing at the bolt holes occourd at 1.1% drift

Faiure of glass panels

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2

In-plane drift (%)

Figure 6.16 Comparison of the tensile stress developed at the Test #1 FE and Test #1

FE rigid body translation only

138


B1

B3 B4

B2

Glass panel (PA)

B1

B3 B4

B2

Glass panel (PB)

B1

B3 B4

B2

Glass panel (PD)

B1

B3 B4

B2

Glass panel (PC)

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm- +7mm-

+3mm-

+3mm-

+3mm-

+3mm

+3mm-

+3mm-

+3mm

+3mm-

+3mm-

+3mm-

+3mm-

+3mm

+3mm-

+7mm-

+3mm

+3mm-

+3mm-

+3mm-

+3mm-

+3mm

+3mm-

Figure 6.17 Built-in standard gaps provided collectively at the spider arms to bolt

fitting connections (Test #1 FE Chapter 5)

1200

1200

Glass panel

3111

.3 2°

100

PA PB

PD

PC

B3 B4

B1 B2

B3 B4

B1 B2

B3 B4

B1 B2

B3 B4

B1 B2

b = 2400

h =

22

00

100

Figure 6.18 Orientation of the spider arms (Same as Test #1)

139


b

h

D

c1

c2

Figure 6.19 Frame in parallelogram action under racking load (Test #1)

Table 6.2 In-plane drift capacity for the typical panel sizes due the rigid body

translation only at the built-in standard gaps

Panel

aspect

ratio

Panel

width(mm)

Panel

height

(mm)

b

(mm)

h

(mm)

c1

(mm)

c2

(mm)

c2-c1

(mm) D(mm)

Drift

(%)

1.0 1200 1200 2200 2200 3111 3128 17 24.0 1.1

1.5 1200 1800 2200 3400 4050 4067 17 31.1 0.9

2.0 1200 2400 2200 4600 5099 5116 17 39.1 0.8

0.7 1800 1200 3400 2200 4050 4067 17 20.2 0.8

1.0 1800 1800 3400 3400 4808 4825 17 24.0 0.7

1.3 1800 2400 3400 4600 5720 5737 17 28.5 0.6

0.5 2400 1200 4600 2200 5099 5116 17 18.8 0.8

1.0 2400 2400 4600 4600 6505 6522 17 24.0 0.5

1.5 2400 3600 4600 7000 8376 8393 17 30.9 0.4

0.5 3600 1800 7000 3400 7782 7799 17 18.9 0.5

0.7 3600 2400 7000 4600 8376 8393 17 20.3 0.4

1.0 3600 3600 7000 7000 9899 9916 17 24.0 0.3

• All values are for 2x2 grid system with ‘X’ type spider arms

• All connections with built-in standard gaps

140


6.4 Test #1 - Parametric Study for Multiple Grid Systems

6.4.1 Test #1 - Grid System

The Test #1 FE model was modified to incorporate 3x3 and 4x4 panel systems and

the analyses were repeated. The size of the glass panels remained the same. The

ultimate drift capacities of both systems were reduced to 1.5% compared to the drift

capacity of 2.1% for the 2x2 system. The pushover curves for these three systems are

compared in Figure 6.20 and the tensile stresses developed against the drift applied

are compared in Figure 6.21. The racking performance of the PFGFS reduced with the

increasing number of façade grid system with same size of glass panels.

In-p

lan

e lo

ad

(kN

)

40

35

30

25

20

15

10

5

0

Test #1 FE 2x2 System






0 0.4 0.8 1.2 1.6 2

In-plane drift (%)

Figure 6.20 Analytical pushover curve comparison for 2x2, 3x3 and 4x4

systems (Test #1)

141

2.4


120

110

100

90

80

70

60

50

40

30

20

10

0

In-plane drift (%)

Figure 6.21 Comparison of the tensile stresses developed comparison for 2x2, 3x3

and 4x4 systems (Test #1)

6.4.2 Test #1- Racking Performance of Grid Systems

The parametric study indicated that when the grid number of the Test #1 PFGFS

configuration increased to 3x3 and 4x4 with same size of glass panels the racking

performance of the system reduced and the rotation of the spider arms was reduced

compared to Test #1 FE results. When most of the spider arms (except the corner

spiders) are orientated diagonally as shown in Figure 6.22, there is no possibility of

the diagonal length between bolted connections increasing due to the spider arm

rotation. Therefore, it was assumed, DRBR = 0. Consequently, during the racking

action of such systems the rigid body translation can only occur at the built-in

standard gaps provided. Additionally, the spider arms could deform structurally and

accommodate further racking displacement resulting in the following drift capacity:

DT = DRBT + DD Eq (6.3)

Where, DT is the ultimate drift capacity, DRBT is the drift capacity due to rigid body

translation at the built-in standard gap provided and DD is the drift capacity due to

deformation of the spider arms. If all the spider arms are diagonally oriented as shown

in Figure 6.22, the in-plane drift capacity for multiple grid system (eg. 3x3, 4x4 and

Ten

sile s

tre

ss (M

Pa

)




Failure of glass panel at 1.5% Failure of glass

panel at 2.1%

0 0.4 0.8 1.2 1.6 2

142

2.4


5x5) due to the rigid body translation only will be the same as for a 2x2 system.

Similarly, symmetrical systems will result in the same drift capacity. Therefore, in the

following sections the DT, DRBT and DD were calculated using the FE analyses for a

2x2 system with diagonally orientated spider arms as shown in Figure 6.22 which also

represents multiple grid systems. The DRBM was calculated for three different types of

built-in standard gaps as discussed in the following sections.

a) Total ultimate drift capacity (DT)

Test #1 FE model was modified to evaluate the racking performance of the system

shown in Figure 6.22. The modified structural support frame in the FE model is

shown in Figure 6.23 where the rotation of the spider arms was prevented. The FE

analysis was repeated with the diagonally orientated spider arms to evaluate the

ultimate drift capacity of the system and the pushover curve and the tensile stress

developed against the applied drift are shown in Figures 6.24 & 6.25. The Test #1 FE

with diagonally oriented spider arms had an ultimate drift capacity of 1.45% which is

comparable with the multiple grid systems capacity of 1.5% in Section 6.4.1 as

expected.

1200

1200

Glass panel

Figure 6.22 All the spider arms orientated diagonally for a multiple façade grid

system (Test #1)

143


(a) Front view (b) Side view

Figure 6.23 The structural support frame in the modified FE model with the spider

arms diagonally orientated (Test #1)

b) Drift contribution from the built-in standard gap (DRBT)

Following the calculation of DT, that the structural support frame and spider arms

were assigned very high stiffness and the FE model was restrained in the out-of-plane

direction to evaluate the racking performance of the system due to the rigid body

translation only at the built-in standard gaps. The stiffness of the springs were set to

be the same as the Test #1 FE model (Chapter 5) until the gap closed at which point a

very high stiffness was assigned (i.e. bearing of bolt on spider arm). A rigid body

translation of ±3mm was allowed in both the ‘x’ and ‘y’ directions at the glass panel

spider arm connection to represent built-in standard gaps. The pushover curve shown

in Figure 6.24 indicates that the rigid body translation occurred for 1% drift followed

by bearing contact. A similar racking performance is observed in the maximum

tensile stress developed against the drift applied as shown in Figure 6.25.

144


Figure 6.24 Analytical pushover curve comparison from the Test #1 FE multiple grid

façade system with the rigid body translation only from the built-in standard gaps

Figure 6.25 Comparison of the tensile stress developed for the Test #1 FE multiple

grid façade system with the rigid body translation only from the built-in standard

gaps

145


The results from the FE analysis can be verified using Equation 6.2 introduced in

Section 6.31:

2 2 2 2 c2 − c1 = (b + D) + h − b + h Eq (6.2)

Similar to the Section 6.3.1, c − c = 4 × (√3 + 3 ) = 16.97mm

16.97 =.(2200+D)2+22002-√22002+22002 and D=24mm

This indicates that the system can accommodate a racking displacement of 24mm

between two extreme bolted connections (h = 2400mm) at the structural support

frame due to the rigid body translation from the built-in standard gaps. Therefore the

system has a drift capacity of DRBT = 1.0%. A similar drift capacity (DRBT) can be

expected from the multiple grid PFGFS. However, the drift capacity is dependent on

the geometry of glass panels. The drift capacities due to rigid body translation from

the built-in standard gap in multiple grid systems with different glass panel

configurations are presented in Table 6.3.

Table 6.3 Drift capacity for typical panel sizes in multiple grid façade systems (3x3,

4x4 and 5x5) due the rigid body translation at the bolted connections only (Test #1)

Panel

aspect

ratio

Panel

width (mm)

Panel

height (mm)

b

(mm)

h

(mm)

c1

(mm)

c2

(mm)

c2-c1

(mm)

D

(mm)

Drift

(%)

1.0 1200 1200 2200 2200 3111 3128 17 24.0 1.0

1.5 1200 1800 2200 3400 4050 4067 17 31.1 0.9

2.0 1200 2400 2200 4600 5099 5116 17 39.1 0.8

0.7 1800 1200 3400 2200 4050 4067 17 20.2 0.8

1.0 1800 1800 3400 3400 4808 4825 17 24.0 0.7

1.3 1800 2400 3400 4600 5720 5737 17 28.5 0.6

0.5 2400 1200 4600 2200 5099 5116 17 18.8 0.8

1.0 2400 2400 4600 4600 6505 6522 17 24.0 0.5

1.5 2400 3600 4600 7000 8376 8393 17 30.9 0.4

0.5 3600 1800 7000 3400 7782 7799 17 18.9 0.5

0.7 3600 2400 7000 4600 8376 8393 17 20.3 0.4

1.0 3600 3600 7000 7000 9899 9916 17 24.0 0.3

146

-

-

-

-


c) Drift contribution from deformation of the spider arms (DD)

The ultimate drift capacity of the system is shown in Figures 6.24 & 6.25. This

indicates that the system with diagonally oriented spider arms has an ultimate drift

capacity DT = 1.45% and therefore, the drift from the deformation of the spider arms,

DD = 1.45-1.0 = 0.45%.

d) Drift contribution from the built-in standard gaps at the spider arms in practice

(DRBT)

In real practice, the four way ‘X’ type spider arms are provided with two circular

(bottom) and two slotted (top) holes. Therefore, the built-in standard gap in the ‘x’

direction for slotted holes was modified to ±7mm in the FE model as shown in Figure

6.26 and the analysis was repeated. The pushover curve is shown in Figure 6.27 and

the maximum stress developed against the applied drift is shown in Figure 6.28. The

drift capacity increased from 1% to 1.33% due to the slotted holes.

1200

+3mm+3mm - +3mm- +3mm --

+7mm+7mm +7mm+7mm -- --

+3mm +3mm- +3mm - +3mm--

+3mm+3mm - +3mm- +3mm --

+3mm +3mm- +3mm - +3mm

+7mm+7mm +7mm +7mm- -- -

Glass panel

+3mm +3mm- +3mm - +3mm

1200 +3mm+3mm- +3mm - +3mm- -

Figure 6.26 Built-in standard gaps provided at the bolted connections (Test #1)

147


0

10

20

30

40

50

60

In-p

lan

e l

oa

d (

kN

)

Test #1 FE 3&7mm Rigid Body Movement Only

Test #1 FE 3mm Rigid Body Movement Only

Bearing occured at

1.0% drift

Bearing occured at

1.3% drift

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

In-plane drift (%)

Figure 6.27 Analytical pushover curve comparison of multiple grid façade system

from the rigid body translation at the bolt fittings, Test #1 FE ±3mm rigid body

translation with ±3mm and ±7mm rigid body translation

60

50

Te

nsi

le s

tre

ss (

MP

a)

40

30

20

10

0

Test #1 FE 3&7mm Rigid Body Movement Only

Test #1 FE 3mm Rigid Body Movement Only

Failure of a glass

panel at 1.3%

Bearing occured at

1.0% drift

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

In-plane drift (%)

Figure 6.28 Comparison of the tensile stress developed for the multiple grid façade

system from the rigid body translation at the bolt fittings, Test #1 FE ±3mm rigid

body translation with ±3mm and ±7mm rigid body translation

148


e) Drift contribution from the built-in standard gaps at the structural support

frame in practice (DRBT)

The FE model was modified with the gaps as shown in Figure 6.29. The bolt fittings

were assigned with ±3mm rigid body translation in both the ‘x’ and ‘y’ directions and

additionally, the structural support frame connections were assigned with ±7mm in

the ‘x’ direction and a ±2mm gap in the ‘y’ direction to incorporate the built-in

standard gap (slotted hole Section 6.2.1) at the structural support frame. The FE

analysis was repeated to calculate the drift capacity due to the rigid body translation

associated with these two built-in standard gaps. The pushover curve and the

maximum tensile stresses developed as a function of the applied drift from the

analysis are presented in the Figures 6.30 & 6.31. The drift capacity increased from

1.0% to 1.7% with the introduction of the built-in standard gap at the structural

support frame.

The results from the FE analysis can be verified using Equation 6.2:

2 2 2 2 c2 − c1 = (b + D) + h − b + h Eq (6.2)

Similar to Section 6.3.1, the diagonal expansion due to the rigid body translation at

the bolted fittings = 4 × (√3 + 3 ) = 17mm. The diagonal expansion due to the

rigid body translation at the structural support frame = 2 × (√2 + 2 ) = 5.5mm (Since only the two extreme bolts can move and the middle one cannot move).

Therefore in total, c2-c1 = 17+5.5 = 22.5mm and

22.5 = .(2200 + D) + 2200 − √2200 + 2200 and D = 31.7mm

Further, a remaining possible rigid body translation of 5mm (7mm-2mm) in the ‘x’

direction at all the structural support frame connections will add additional 2x5 =

10mm racking displacement at the top. Therefore, in total the system can

accommodate a racking displacement of 41.7mm (31.7mm+10mm) from the built-in

standard gaps at the structural support frame resulting in a drift capacity =DRBM

1.74%.

149

---


+7mm +7mm +7mm

+2mm-+2mm-+2mm-

+7mm-

+2mm-

+7mm-

+2mm-

+7mm-

+2mm-

+7mm-

+2mm-

+7mm-

+2mm-

+7mm-

+2mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

+3mm-

1200

1200

Glass panel

Figure 6.29 Built-in standard gaps provided at the bolt fittings and the structural

support frame (Test #1)

Figure 6.30 Analytical pushover curve comparison of multiple grid façade system,

Test #1 FE from the ’rigid body translation at the bolt fittings’ with ‘rigid body

translation only at the bolt fittings and structural support frame’

150


Figure 6.31 Comparison of the tensile stress developed at the multiple grid façade

system, Test #1 FE from the ‘rigid body translation at the bolt fittings’ with from the

‘rigid body translation at the bolt fittings and structural support frame’

6.5 Test #1 - Performance of Rigidly Connected ‘X’ Type Spider

Arms

There are occasions where the ‘X’ type spider arms are rigidly connected to the

structural support frame which prevents the in-plane rotation. If the structural support

frame members are horizontally orientated then a higher seismic PFGFS could be

used as discussed in Section 2.6.4 with large horizontally slotted holes on the spider

arms. However, there are occasions where the ‘X’ type spider arms are rigidly

connected to vertically orientated structural support frame members. Racking

performance of such systems could be increased by providing articulation at the

bolted fittings while allowing the glass panels to rotate.

A panel with two horizontally slotted hole connections at the top to transfer the self-

weight and two oversized circular holes at the bottom is shown in Figure 6.32a with a

clearance (gap) “c” assigned between the edge of the bolt and the edge of the hole. In

practice these gaps are provided on the spider arms rather than the oversized holes in

151


the glass. However, for clarity, the holes are shown in Figure 6.32a without the spider

arms. The translation under the in-plane load and the drift capacity “Δ” are shown in

Figure 6.32b considering rigid body translation and rotation of the glass panel and

assuming that the structural support frame is pinned at the base.

Δ

c c

l

c c

b H

θ

(a) (b)


arms (a) Geometry of the gaps in the glass panels to transfer loads and (b)

Translation of the glass panel under in-plane loading

According to Figure 6.32b the drift angle “θ” can be calculated as follows;

2c sinθ = Eq (6.3)

l

Therefore, the drift “Δ” for the height “H” is equal to

2c Δ = H ×Sinθ = H × Eq (6.4)

l

Figure 6.33 shows the geometry of the holes to transfer the loads with the spider arms

horizontally projected to a length of “a/2” (for illustrative purposes, the spider arms

are horizontally projected). Under racking load, it is assumed the glass panel will

translate and rotate as a rigid body with no shear deformation. Hence, only the

152


geometrical positions of the holes influence the drift “Δ” whilst the spider arm

connections will facilitate the movement and rotation of glass panel as shown in

Figure 6.33b.

a/2

b

c c

a/2

H

l

Δ β

θ

α

(a) (b)

Figure 6.33 Glass panels connected to the structural support frame with horizontally

orientated spider arms (a) Geometry of the holes to transfer loads with spider arms

and (b) Translation of the glass panel under in-plane loading with spider arms

In the translated glass panel with the spider arms, the angle “θ” can be taken as the

angle between the structural support frame and the edge of the glass panel (refer

Figure 6.33b). However, the drift angle “α” is equal to “θ-β” as illustrated in Figure

6.33b. The reason for the reduction in the drift angle is due to the rotation of the panel

as the top left spider arm rises up and the top right spider arm lowers down, resulting

in the panel rotating in the clockwise direction. This rotation is indicated by the angle

“β”. Since “θ” and “β” are small angles;

sin α = sin (θ - β) ≈ sin θ - sin β Eq (6.5)

Since the structural support frame drifts through an angle “α”, the rigidly connected

spider arms will also rotate through the same angle “α”. The top left arm rises up by a

153


vertical distance of (a/2) sin α and the top right arm moves down by a vertical

distance of (a/2) sin α. Therefore the angle “β” can be deduced as follows;

a sin αsin β = Eq (6.6)

b

Where “b” is the distance between the holes as shown in Figure 6.33a. Using

Equations 6.3, 6.5 & 6.6 angle “α” can be obtained as follows:

2c a sin αsin α = sin θ - sin β=

l b

2 c ⎛ b ⎞sin α = Eq (6.7) ⎜ ⎟

l ⎝ a + b ⎠

Therefore the drift “Δ” with the spider arm case shown in Figure 6.33 is equal to:

2 c ⎛ b ⎞Δ = H × Eq (6.8) ⎜ ⎟

l ⎝ a + b ⎠

The developed expressions are demonstrated with the following example. Assume a

250 x 250 mm glass panel with 35mm long slotted holes at the top and 35mm

diameter circular hole at the bottom, without spider arms as shown in Figure 6.34.

The connection bolt diameter is assumed to be 15mm and the rigid body translation of

the glass panel for a 500mm structural support frame is shown in Figure 6.34b.

According to Equations 6.3 & 6.4:

-1 20 �θ = sin = 7.66

150

2c 2×10 Δ = H × =500 × = 66.67mm

l 150

Figure 6.35a shows the geometry of the setup with “a”/2 = 50mm long spider arms

whilst the rigid body translation and rotation of glass panel are shown in Figure 6.35b.

154


35 66.67

Ø15 50

60

150 250

Ø35

250

7.66°

250

(a) (b)


arms (a) Geometry of case study example, (b) Translation of the glass panel under in-

plane loading

3.06° 35 40

Ø15 7.65°

250

Ø35

250

4.59°

250

(a) (b)

Figure 6.35 Glass panels connected to the structural support frame with horizontally

orientated spider arms (a) Geometry of the case study example, (b) Translation of the

glass panel under in-plane loading

155


By applying Equations 6.6, 6.7 & 6.8:

2 c ⎛ b ⎞ 2 ×10 ⎛ 150 ⎞ �sinα= = , and α = 4.59 ⎜ ⎟ ⎜ ⎟l ⎝ a+b ⎠ 150 ⎝100+150 ⎠10 × sin 4.59

�sin β = and β = 3.06 150

The sum of the angles “α” and “β” is equal to; α + β = 3.06 + 4.59 = 7.66°, which is

equal to the angle “θ” from the calculation shown for glass panel without spider arms.

The drift capacity “Δ” is therefore equal to:

2 c ⎛ b ⎞ 2 ×10 ⎛ 150 ⎞Δ=H × = 500 × = 40mm ⎜ ⎟ ⎜ ⎟

l ⎝ a+b ⎠ 150 ⎝100 +150 ⎠

Interestingly, the example demonstrates that the drift capacity of the façade system

with spider arms actually reduces from 67mm to 40mm due to the counter rotation of

the glass panels.

Test #1 FE model (Chapter 5) was modified to verify Equation 6.8. Non-linear spring

elements were used to define the gap between the spider arms and the bolt. The

stiffness of the spring was set to be very low until the gap closed and once closed the

stiffness was assigned to be very high (i.e. bearing of bolt on spider arm). Analysis

was carried out and the in-plane drift predicted by Equation 6.8 was confirmed by the

FE analysis results. Equation 6.8 was then applied to a two by two grid of glass panels

shown in Figure 6.36 where the clearance “c” was assumed to be 10mm for “a” =

100mm long arms and the glass wall height “H” was double the height of the glass

panel. Table 6.4 shows the in-plane displacement capacities considering rigid body

translation and rotation for a range of panel sizes and indicates that the drift capacity

reduce as the panel size increases.

156


H

l

c

b

l

b

c

a/2 a

Figure 6.36 Schematic diagram of typical PFGFS with rigidly connected ‘X’ type

spider arms with articulation holes

Table 6.4 In-plane drift capacity of PFGFS with rigidly connected ‘X’ type spider

arms

Dimension

(mm)

Length “b”

(mm)

Length “l”

(mm)

Drift “Δ”

(mm)

Drift “Δ”

(%)

1200x1200 1100 1100 40.0 3.3

1200x1600 1100 1500 39.1 3.2

1600x1600 1500 1500 40.0 2.4

1600x2000 1500 1900 39.4 2.4

2000x2000 1900 1900 40.0 2.0

157


6.6 Test #2 - Parametric Study

6.6.1 Test #2 - Sealant Types

A weather sealant was used in Test #2 with a typical shear modulus of 0.10MPa and a

parametric study was then conducted using the FE model to compare the racking

performance with structural and special purpose sealants. The typical shear modulus

values for the sealants assigned in the FE model are presented in Table 6.5. The

sealant type had a significant effect on the racking performance of the PFGFS with

the ultimate in-plane drift capacity found to be 3.7%, 4.7% and 5.1% for the high

(structural), medium (weather) and low (special purpose) sealants respectively. In

each case, failure was defined when the tensile stresses in the glass panels exceeded

94MPa. The pushover curves for low, medium and high modulus sealants are

illustrated in Figure 6.37 and the tensile stresses developed against the applied drift

are illustrated in Figure 6.38. The pushover curves showed that a high in-plane lateral

load had to be applied when the higher modulus sealant was used. This enhanced the

out-of-plane deformation in the spider arms and the tensile stress development in the

glass panels thereby reducing the racking performance of the system compared to the

other sealants.

Table 6.5 Properties of sealant used in the FE analysis

Sealant type Shear Modulus

(MPa)

Grid

Arrangement

Glass Panel

Dimension

(mm)

Weather (Medium modulus) 0.10 2x2 1200x1200

Structural (High modulus) 0.50 2x2 1200x1200

Special purpose (Low

modulus) 0.05 2x2 1200x1200

158


Te

ns

ile s

tre

ss (M

Pa

) In

-pla

ne

load

(kN

) 50

45

40

35

30

25

20

15

10

5

0 0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2 5.6



Test #2 FE High Modulus Failure of glass panel at 3.7%



In-plane drift (%)


modulus silicon sealants (Test #2)

120

110

100

90

80

70

60

50

40

30

20

10

0

In-plane drift (%)

Figure 6.38 Comparison of the tensile stress developed at the FE models with FE





0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2 5.6

low, medium and high modulus silicon sealant (Test #2)

159


6.6.2 Test #2 - Sealant Thickness

An 8mm thick weather sealant was used in Test #2 with a typical shear modulus of

0.10MPa and a parametric study was then conducted using the FE model to compare

the racking performance of systems with 6mm, 8mm and 10mm thick silicon sealants.

The sealant thickness had an effect on the racking performance of the PFGFS with in-

plane drift capacity of 4.7%, 4.7% and 5.2% estimated for the 6mm, 8mm and 10mm

thick sealants respectively. In each case, failure was defined when the tensile stresses

in the glass panels exceeded 94MPa. The pushover curves for 6mm, 8mm and 10mm

thick sealants are illustrated in Figure 6.39 and the tensile stresses developed against

the applied drift are illustrated in Figure 6.40. The out-of-plane deformation in the

spider arms and tensile stress development on the glass panel were reduced when

thicker sealant was used, thereby increasing the racking performance of the system

compared to the 6mm and 8mm thick sealants.

In-p

lan

e lo

ad

(kN

)

40

35

30

25

20

15

10

5

0




Faiure of glass panel at 4.7%


0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2 5.6

In-plane drift (%)

Figure 6.39 Analytical pushover curve compared with 6mm, 8mm and 10mm

thick silicon weather sealant (Test #2)

160


Te

ns

ile s

tre

ss (M

Pa

)

110

100

90

80

70

60

50

40

30

20

10

0




Faiure of glass panel

0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2 5.6

In-plane drift (%)

Figure 6.40 Comparison of the tensile stress developed at the FE model with 6mm,

8mm and 10mm thick silicon weather sealant (Test #2)

6.6.3 Test #2 - Glass Geometry

Test #2 FE model was modified and analyses were performed for a 2x2 system

consisting of 1200mm wide and 2400mm high panels (portrait panel system) and a

2x2 system consisting of 2400mm wide and 1200mm high panels (landscape panel

system). The racking performance of the portrait and landscape panel systems was

compared with the square panel system of the Test #2 FE results. The pushover

curves for portrait, square and landscape panel systems are illustrated in Figure 6.41

whilst the maximum tensile stresses developed at the glass holes against the applied

drift for the square, portrait and landscape systems are illustrated in Figure 6.42.

In each case, failure was defined when the tensile stress in the glass panels exceeded

94MPa. The FE analyses demonstrated that, the drift capacity of the system was

aspect ratio dependent, with drift capacity values of 3.6% (landscape system), 4.7%

(square system) and 4.4% (portrait system) recorded. Although the racking drifts were

similar, the applied in-plane racking loads were quite different with the landscape

system significantly stiffer than the square and portrait system as shown in Figure

6.41.

161


Te

ns

ile s

tre

ss (M

Pa

) In

-pla

ne

load

(kN

)

80

70

60

50

40

30

20

10

0 0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2

Test #2 FE 1200(w)x1200(h) Glass






In-plane drift (%)

Figure 6.41 Analytical pushover curve comparison with square, portrait and

landscape glass panels (Test #2)

110

100

90

80

70

60

50

40

30

20

10

0





0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2

In-plane drift (%)

Figure 6.42 Comparison of the tensile stress developed at the FE models with square,

portrait and landscape glass panels (Test #2)

162


6.6.4 Test #2 - Glass Thickness

The thicknesses of the glass panels in the Test #2 FE model were modified from

12mm to 10mm and 15mm and the analyses repeated. The nominal strength at holes

for 12mm thick glass panel was 94MPa in accordance with AS1288 (2006) and

similarly the nominal strengths of the 10mm and 15mm glass panels was 98MPa and

90MPa respectively and these values were used as the reference for failure of the

glass panels in the analyses. The pushover curves for different thickness of glass

panels (10mm, 12mm and 15mm) are compared in Figure 6.43 and the tensile stresses

developed against the applied drift are illustrated in Figure 6.44. The FE results

indicated that the drift capacity increased marginally with the thicker glass specimens

since the out-of-plane stiffness of the glass panel increased relative to the stiffness of

the spider arm, reducing the out-of-plane deformation and bending stresses in the

glass.

In-p

lan

e lo

ad

(kN

)

50

45

40

35

30

25

20

15

10

5

0







0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2 5.6

In-plane drift (%)

Figure 6.43 Analytical pushover curve comparison of 10, 12 and 15mm thick

glass panels (Test #2)

163


110

100

90

80

70

60

50

40

30

20

10

0

In-plane drift (%)

Figure 6.44 Comparison of the tensile stresses developed on 10, 12 and 15mm thick

glass panels (Test #2)

6.7 Test #2 - Racking Performance of 2x2 Systems

6.7.1 Test #2 - Discussion of the Parametric Study for 2x2 Systems


was accommodated by three mechanisms as discussed in Chapters 4 & 5.

1) Rigid body translation of the spider arms at the base built in standard gap;

2) Rigid body translation at the built-in standard gaps provided at the bolted

fittings and

3) Deformations of the spider arms.

The parametric study indicated that in Test #2 the racking performance of PFGFS

increased when:


2) The thickness of the sealant was increased

3) The portrait panel arrangement was used

4) The glass thickness was increased

Te

ns

ile s

tre

ss (M

Pa

)





0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2 5.6

164


Therefore, the in-plane drift capacity of a 2x2 system with spider arms arranged

similar to Test #2 can be estimated from the following Equation:

DT = DRSB + DRBM + DD Eq (6.9)

Where, DT is the ultimate drift capacity, DRSB is the drift capacity due to the rigid

body translation at the spider arms base built-in standard gap (relative translation in

the vertical direction), DRBM is the drift capacity due to rigid body translation from the

built-in standard gap at the spider arm bolted fittings and DD is the drift capacity due

to deformation of the spider arms. In the following sections the drift contribution from

DRSB, DRBM and DD are discussed.

(a) Drift capacity due to the rigid body translation at the spider arms base

built-in standard gap (DRSB)

Test #2 FE results showed that the major contribution of the drift capacity was

obtained from the vertical tolerance of the spider arm base connections. This rigid

body translation is equivalent to a rocking mechanism as shown in Figures 6.45 &

6.46. Using trigonometry from Figure 6.51, the displacement ‘∆’ (DRBM in terms of

drift) can be calculated:

∆ s Tanθ = = Eq (6.10) -

Where, “∆” is the in-plane displacement, “h” is the height of the PFGFS, “s” is the

maximum possible vertical tolerance in the spider arm base and ‘b’ is the width of a

glass panel.

s Therefore, DRBM = - x 100 % Eq (6.11)

In Test #2, s = 35mm, b = 1200mm and by applying Eq.6.11;

35 DRBM = x 100 % = 2.92%. Therefore, a 2.92% drift can be attributed from the 00 rigid body translation at the at the spider arm base built-in standard gap.

Test #2 FE model was modified to verify the racking performance due to the base

translation only of the spider arm and the analysis was repeated. The resulting

165


pushover curve and the tensile stress developed against the applied drift are illustrated

in Figures 6.47 & 6.48 respectively. The FE results indicated that bearing occurred at

2.9% drift whilst drift capacities from only the spider arm sliding for the 2x2 panel

arrangements with different sizes of glass panels are presented in Table 6.6 using

Equation 6.11.

The analysis indicated that the orientation and size of the glass panels influences the

drift capacity due to the rigid body translation at the spider arm base connection.

According to Equation 6.11, the only parameter that influences the drift capacity from

the spider arm vertical translation at the base is the vertical tolerance ‘s’ and the glass

panel width ‘b’. This explains the reason for the massive reduction in the ultimate

drift capacity of the portrait system in Section 6.7.3. Therefore, a similar drift capacity

could be obtained in multiple grid PFGFS from the vertical translation of the spider

arms at the base connections (DRSB).

(b) Drift capacity due to rigid body translation at the built-in standard gap at

the bolted fittings (DRBM)

Test #2 FE model was modified to verify the racking performance due to rigid body

translation at the built-in standard gap provided at the bolted fittings only and the

analysis was repeated. The analytical pushover curve and the tensile stress developed

against the applied drift are illustrated in Figures 6.49 & 6.50 respectively. The FE

results indicated that the bearing occurred at 1.3% drift. The rigid body translation at

the built-in standard gaps (DRBM) and the rigid body translation due to the

deformation of the spider arms could also be verified using trigonometry as discussed

in Section 6.3 for Test #1.

(c) Drift capacity due to deformation of the spider arms (DD)

The full system was benchmarked against the experimental results with an ultimate

drift capacity DT = 4.7% and therefore, approximately the drift capacity due to the

deformation of the spider arms could be estimated as DD = DT – (DRSB+ DRBM) = 4.7

(2.9+1.3) = 0.5%. The FE analysis was then extended to investigate multiple grid

systems and the results are discussed in Section 6.8.2.

166


B1

B3 B4

B2

Glass panel (PA)

B1

B3 B4

B2

Glass panel (PB)

B1

B3 B4

B2

Glass panel (PD)

B1

B3 B4

B2

Glass panel (PC)

COMPRESSION TENSION

Figure 6.45 Spider arm vertical translation due to the rigid body translation at the

spider arms base connections (Test #2)

Glass panel (PA) Glass panel (PB)

Glass panel (PD)Glass panel (PC)

h

θ

θ θ

θ

θ

θ

SS

S

Δ

Figure 6.46 Rocking mechanism of the glass panels under in-plane racking load

(Test #2)

167


In-p

lan

e lo

ad

(kN

)

120

100

80

60

40

20

0

Test #2 FE

Test #2 FE Rigid Body Spider Translation Only

Bearing occured at 2.9% drift


0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2

In-plane drift (%)


rigid body spider arm vertical translation at the base connections

Ten

sile

str

es

s (M

Pa

)

160

150

140

130

120

110

100

90

80

70

60

50

40

30

20

10

0

Test #2 FE

Test #2 FE Rigid Body Spider Translation Only


Bearing occured at 2.9%

0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2

In-plane drift (%)


#2 FE rigid body spider arm vertical translation at the base connections

168


In-p

lan

e lo

ad

(kN

)

120

100

80

60

40

20

0

Test #2 FE

Test #2 FE Rigid Body at Bolts Only



0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2

In-plane drift (%)


rigid body translation at the bolt fittings built-in standard gaps

Ten

sile s

tre

ss (M

Pa

)

110

100

90

80

70

60

50

40

30

20

10

0

Test #2 FE

Test #2 FE Rigid Body at Bolts Only



0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2

In-plane drift (%)

Figure 6.50 Comparison of the tensile stress developed at Test #2 FE and Test #2

FE rigid body translation at bolt fittings built-in standard gaps

169


Table 6.6 Drift capacity for typical panel sizes due the rigid body translation of the

spider arms at the base connections (Test #2)

Panel

aspect

ratio

Panel

width (mm)

Panel

height (mm) S (mm) Drift (%)

1.0 1200 1200 35 2.9

1.5 1200 1800 35 2.9

2.0 1200 2400 35 2.9

0.7 1800 1200 35 1.9

1.0 1800 1800 35 1.9

1.3 1800 2400 35 1.9

0.5 2400 1200 35 1.5

1.0 2400 2400 35 1.5

1.5 2400 3600 35 1.5

0.5 3600 1800 35 1.0

0.7 3600 2400 35 1.0

1.0 3600 3600 35 1.0

• Calculation for ‘K’ type spider arms only

• Standard built-in gaps at the spider arm base only

6.7.2 Test #2 - Grid system

Test #2 FE model was modified to model 3x3 and a 4x4 panel systems with glass

panel size remained the same and the analyses were repeated. The pushover curves for

these three systems are compared in Figure 6.51 and the tensile stresses developed

against the drift applied are compared in Figure 6.52. The drift capacity of the 3x3

system was 4.5% and the drift capacity of the 4x4 system was 4.4% compared to the

capacity of 4.7% for a similar 2x2 system. The small deviation in the capacity is due

to the rigid body translation at the spider arm to glass bolt fittings. Therefore, the drift

capacities presented in Table 6.6 due to the spider arm base translation for 2x2 panel

arrangements only with different size of glass panels are also applicable for multiple

grid systems.

170


Te

ns

ile s

tre

ss (M

Pa

) In

-pla

ne lo

ad

(kN

) 60

50

40

30

20

10

0

Test #2 FE 2x2 system






0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2 5.6

In-plane drift (%)

Figure 6.51 Analytical pushover curve compared for the 2x2, 3x3 and 4x4

systems (Test #2)

110

100

90

80

70

60

50

40

30

20

10

0




Faiure of glass panel

0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 4.4 4.8 5.2

In-plane drift (%)

Figure 6.52 Comparison of the tensile stress developed at the FE 2x2, 3x3 and 4x4

systems (Test #2)

171


6.8 Design Formulae Developed for Test #1 and Test #2

6.8.1 Test #1, 2x2 Systems

The in-plane drift capacity of a 2x2 system with spider arms arranged similar to Test

#1 can be estimated using the following equation:

DT = DRBR + DRBT + DD Eq (6.1)

Where, DT is the total drift capacity, DRBR is the drift capacity due to the rigid body

rotation of the spider arms, DRBT is the drift capacity due to rigid body translation at

the built-in standard gaps and DD is the drift capacity due to deformation of the spider

arms.

DRBT is calculated using Equation 6.12:

DRBT = � × 100% Eq (6.12)

Where “D” is calculated using Equation 6.2 and “h” is the vertical distance between

the extreme bolted connections on the glass façade.

2 2 2 2 c2 − c1 = (b + D) + h − b + h Eq (6.2)

Where, ‘c1’ is the initial diagonal length between the extreme bolted connections on

the glass panels as shown in Figure 6.19; ‘c2’ is the final diagonal length (after the

racking action) between the extreme bolted connections on the glass panels; ‘b’ is the

horizontal distance between the extreme bolted connections on the glass panels; ‘h’ is

the vertical distance between the extreme bolted connections on the glass façade; and

“D” is the possible racking displacement due to rigid body translation only.

6.8.2 Test #1, Multiple Grid Systems

DT = DRBT + DD Eq (6.3)

Where, DT is the ultimate drift capacity, DRBT is the drift capacity due to rigid body

translation at the built-in standard gaps calculated using Equation 6.12 and DD is the

drift capacity due to deformation of the spider arms.

172

l


6.8.3 Test #1 - Rigidly Connected ‘X’ Type Spider Arms

The additional drift capacity of “Δ” from the articulation provided in the rigidly

connected spider arms can be calculated using:

2 c ⎛ b ⎞Δ = ×100% Eq (6.8) ⎜ ⎟

⎝ a + b ⎠

Where, “c” is clearance (gap) assigned between the edge of the bolt and the edge of

the hole, “l” is the centre to centre distance between two adjacent bolt holes on the

glass panel and “a” is the horizontally projected spider arm length as shown in Figure

6.33.

6.8.4 Test #2, 2x2 and Multiple Grid Systems

The in-plane drift capacity of a 2x2 and multiple grid system with spider arms

arranged similar to Test #2 can be estimated from the following Equation:

DT = DRSB + DRBM + DD Eq (6.9)

Where, DT is the ultimate drift capacity, DRSB is the drift capacity due to the rigid

body translation at the spider arms base built-in standard gap (relative translation in

the vertical direction), DRBM is the drift capacity due to rigid body translation from the

built-in standard gap at the spider arm bolted fittings calculated using Equation 6.11

and DD is the drift capacity due to deformation of the spider arms.

s DRBM = x 100 % Eq (6.11) -

Where, ”s” is the maximum possible vertical tolerance in the spider arm base and “b”

is the width of a glass panel.

173



The FE models benchmarked against the experimental test results for Tests #1 and

Test#2 were used to predict the racking performance of PFGFS with different

configurations. Variations related to built-in standard gap, sealant type, sealant

thickness, glass geometry, and glass thickness were considered in the parametric

study for 2x2 façade grid systems. Further detailed FE analyses were conducted to

evaluate the individual contributions of each racking mechanism to the total racking

displacement for the 2x2 and multiple grid systems. Moreover, a parametric study on

multiple façade grid systems was conducted and detailed FE analyses were carried out

to evaluate individual contributions for each racking mechanism. Further, the racking

performance of PFGFS with rigidly connected ‘X’ type spider arms (i.e., in-plane

rotation prevented) were also discussed and special articulations were introduced to

increase the in-plane raking capacity of such systems.

The parametric study indicated that in Test #1 the racking performance of the PFGFS

increased when:

1) Slotted holes (built-in standard gap) were introduced at the structural support

frame


3) The silicone sealant thickness was increased

4) The height to width ratio of the glass panels was increased




a) In-plane rigid body rotation of the spider arms – 0.4%.

b) Rigid body translation facilitated by the built-in standard gaps between the

bolts and holes with in the spider arm and structural support frame connections

– 1.1%.

c) Deformations and distortions of the spider arms – 0.5%.

Further detailed FE analyses were conducted to evaluate the racking capacity from

each mechanism. The analyses showed that a significant amount of the drift capacity

174


was attributed to the rigid body translation at the built-in standard gaps provided.

Trigonometric expressions were used to verify the drift capacities from the rigid body

translation. The possible rigid body translations in practice were discussed, FE

analyses undertaken and analytical expressions developed for different configurations.

The in-plane drift capacities of such systems due to rigid body translations were

calculated and presented.

The analysis was further extended for multiple grid PFGFS with ‘X’ type spider arms

and typical drift capacities of multiple grid systems were also presented. The study

was further extended to PFGFS with rigidly connected ‘X’ type spider arms and a

special articulation was introduced to increase the drift capacity of such systems.

The parametric study indicated that in Test #2 the racking performance of PFGFS

increased when:


2) The thickness of the sealant was increased

3) The portrait panel arrangement was used




d) Rigid body vertical translation of the spider arms at the base standard built-in

gap – 2.9%.

e) Rigid body translation facilitated by the built-in standard gaps between the

bolts and holes with in the glass panels and spider arm connections – 1.3%.

f) Deformations and distortions of the spider arms – 0.5%.

Further detail FE analyses were conducted to express the racking capacity from each

mechanism. The analyses showed that a significant amount of the drift capacity was

attributed to the rigid body translation at the spider arm base (vertical sliding)

connections. Trigonometric expressions and FE models were used to verify the drift

capacities from the rigid body translation at the spider arm base connections. The

possible rigid body translations were discussed, with FE analyses undertaken and

analytical expressions developed for different configurations. Drift capacities of such

175


systems due to rigid body translation at the spider arms base built-in standard gaps

were calculated and presented.

The rigid body translation at the built-in standard gaps (DRBT) and the rigid body

translation due to the deformation of the spider arms could be calculated manually as

discussed in Section 6.3 for Test #1. The analysis was further extended for multiple

grid PFGFS with ‘K’ type spider arms and interestingly typical drift capacities of

multiple grid systems were approximately the same as Test #2 2x2 systems.

176

Chapter 7 INTER-STOREY DRIFT CALCULATION AND IN-PLANE SEISMIC DESIGN OF PFGFS

Chapter 7

7 INTER-STOREY DRIFT CALCULATION AND IN

PLANE SEISMIC DESIGN OF PFGFS

7.1 Introduction

Buildings subjected to seismic actions experience reverse cyclic swaying and the

resulting deformations induced in buildings may be quantified for the assessment of

façade systems using the inter-storey displacement (Su et al., 2008). A schematic

diagram of a building sway under earthquake ground motion is shown in Figure 7.1.

The inter-storey drift ratio ‘γi’ at the ‘ith’ floor can be defined as;

γi = (∆i / hi) x100 % Eq (7.1)

where ‘hi’ is the storey height

Figure 7.1 Schematic diagram of a building sway under earthquake ground motion

(Su et al., 2008)

177


The in-plane seismic assessment of glass façade systems requires an estimate of the

likely in-plane drift demand from the earthquake action. AS 1170.4 (2007), clauses

5.4.4 and 5.5.4, specify that, “the inter-storey drift at the ultimate limit state,

calculated from the forces determined according to strength and stability provisions

shall not exceed 1.5% of the storey height for each level” and “the attachment of

cladding and façade panels to the seismic-force-resisting system shall have sufficient

deformation and rotational capacity”. Therefore, for a typical floor height of 3600 mm

the maximum allowable relative storey deflection is 54 mm. Drift provisions in other

international standards and industrial practices were previously discussed in detail in

Chapter 2.

Fardipour et al. (2011) presented the results obtained from a recent study assessing the

drift demand on buildings for a range of projected earthquake scenarios in Australia.

Equations were proposed to conservatively estimate the inter-story drift in buildings.

Example calculations were performed using the peak displacement demand from the

design response spectrum and the maximum inter-story drift calculated for regular

buildings was well below the 1.5% drift limit specified in AS1170.4 (2007).

Similarly, Su et al. (2008) reviewed the seismic engineering research conducted in

Hong Kong and concluded that the maximum inter-storey drift demand at deep soil

sites in conditions of rare earthquake events for a return period of 2500 years is less

than 0.3% for all regular buildings without soft-storey and torsional irregularity

features.

These studies indicate that the likely inter-storey drift is much less than 1.5% for most

buildings in Australia for the 500 year return period (RP) event except for soft storey

structures. However, a simplified approach is required to calculate the likely

maximum in-plane drift demand and to assess the in-plane racking performance of

façade systems.

This chapter reviews the different seismic analysis procedures to calculate the inter-

storey drift in buildings and illustrates some rapid assessment methods to calculate the

inter-storey drift with example calculations. The outcome of this study will be the

development of a simple assessment procedure to assess the likely drift demand to

ensure a minimum level of protection against seismically induced damage in glass

façade systems for both new and existing buildings in regions of low to moderate

178


seismicity such as Australia. Moreover, the a seismic design concept of PFGFS using

the in-plane drift capacities from Chapter 5 & 6 is explained along with recommended

PFGFS boundary conditions.

7.2 Seismic Analysis Methods

Seismic drift demand in buildings can be investigated using different analysis

procedures to ensure compliance with code specified limits. ASCE 41-06 (2007)

provides four analysis options with varying levels of conservatism and specifies the

limitations of each procedure namely; (a) linear elastic procedure (LSP), (b) non

linear static procedure (NSP), (c) linear dynamic procedure (LDP) and (d) non-linear

dynamic procedures (NDP). Each of the different procedures is summarised in the

following sections.

7.2.1 Linear Static Procedures

The Linear Static Procedure (LSP) is the least complicated method to analyse a

structure that is subjected to seismic excitation and is commonly applied for regular

buildings dominated by the first mode of vibration. A set of equivalent static forces in

the direction being considered is distributed simultaneously over the height of the

building and the corresponding internal forces and displacements are determined

using static linear elastic analysis. The base shear force ‘V’ is specified in the

following form in AS1170.4 (2007):

= ( ) /μ Eq (7.2)

Where Kp = Probability factor appropriate for the limit state under

consideration (Kp = 1.0 for a 500 years RP event)

Z = Earthquake hazard factor which is equivalent to an

acceleration coefficient with a 1/500 annual

probability of exceedance in (i.e., a 10% probability

exceedance in 50 years)

Ch(T1) = Value of the normalised spectral shape factor for the

fundamental natural period of the structure

179


Sp = Structural performance factor

µ = Structural ductility factor

Wt = Seismic weight of the structure taken as the sum of Wi

for all levels

The advantage of the LSP is that it is quick, easy, requires minimal inputs, and

inelastic response is implicitly generalised using a ductility factor. The procedure is a

useful tool for conceptual design and produces reasonable results for regular buildings

where the response is dominated by the first mode. Tall buildings and irregular

buildings where higher modes are important are therefore not recommended for such

static design procedures.

7.2.2 Non-linear Static Procedures

The seismic performance of buildings dominated by the first mode can be evaluated

using non-linear static analysis. This involves the estimation of the structural strength

and deformation demands and the comparison with the available capacities at desired

performance levels. Two different NSP are currently used in practice: (a)

displacement modification procedure (co-efficient method) and (b) capacity spectrum

method (CSM). FEMA 356 (2000) utilizes a displacement modification procedure in

which several empirically derived factors are used to modify the response of a SDOF

model of the structure assuming that it remains elastic. The alternative capacity

spectrum method of ATC 40 (1996) is actually a form of equivalent linearization.

This technique uses empirically derived relationships for the effective period and

damping as a function of ductility to estimate the response of an equivalent linear

SDOF oscillator (FEMA 440, 2005).

(a) Displacement co-efficient method

The co-efficient method is used to evaluate the displacement demand with the

displacement capacity at the top of the structure. The displacement coefficient method

described in FEMA 356 (2000) provides a direct numerical calculation of maximum

global displacement demand for structures. A detailed review and evaluation of the

180


method followed by recommendations for improvement are presented in FEMA 440

(2005). The inelastic displacement demand is calculated by modifying the elastic

displacement demand with a series of displacement modification factors. The

displacement capacity is typically estimated from a static pushover analysis, where

monotonically increasing lateral forces are applied to a non-linear mathematical

model of the building until failure results. The target displacement, which represents

the expected maximum displacement of the structure for the earthquake load, is

calculated using Equation 7.3:

2 � = 0 3 �

� Eq (7.3) 4�2 �

Where C0 = Modification factor to relate elastic spectral displacement

of an equivalent SDOF system to the roof displacement

of the corresponding MDOF system

C1 = Modification factor to relate expected maximum inelastic

displacements to displacements calculated for linear

elastic response

C2 = Modification factor to represent the effect of pinched

hysteretic shape, stiffness degradation and strength

deterioration on maximum displacement response

C3 = Modification factor to represent increased displacements

due to dynamic P-∆ effects

Sa = Response spectrum acceleration, at the fundamental

period and damping ratio of the building in the direction

under consideration

g = acceleration of gravity

Te = Effective fundamental period of the building in the

direction under consideration, sec.

181


(b) Capacity spectrum method

The CSM was first introduced in the 1970s as a rapid evaluation procedure in a pilot

project for assessing seismic vulnerability of buildings (Freeman et al., 1975) and is

now the recommended method for the seismic evaluation and retrofit of concrete

buildings (ATC 40, 1996) which are first mode dominant. The procedure compares

the capacity of the structure which is represented in the form of a pushover curve with

the seismic demand on the structure represented in the form of acceleration

displacement response spectra (ADRS). The response of the structure is evaluated

from the graphical intersection of the two curves, which is defined as the

‘performance point’.

This method is suitable for checking the performance of both existing structures and

new structures designed with traditional force based methods. A typical capacity

spectrum diagram is shown in Figure 7.2. The capacity curve is constructed from a

non-linear static pushover analysis for a structure and involves the following steps:

1. Plot base shear force versus roof displacement from a pushover analysis.

2. The base shear force (Vb) is divided by the effective mass (Meff) in the first

mode to obtain the “effective acceleration”. The effective mass is typically in

the range 70%-90% of the total mass of the structure. The maximum

“effective acceleration” is the notional “acceleration capacity”.

3. The roof displacement is divided by the first mode participation factor to

obtain the effective displacement (Deff) of an equivalent SDOF. The maximum

effective displacement is the notional “displacement capacity”.

4. The “Capacity Curve” is constructed by plotting the effective acceleration

against the effective displacement as shown in Figure 7.2.

The performance point as shown in Figure 7.2 represents the maximum displacement

demand experienced by the structure for the design earthquake. This point is defined

as the intersection between the Demand Curve and the Capacity Curve in an iterative

procedure. If the Capacity Curve intersects the 5% damped Demand Curve, then a

conservative estimate of the displacement demand can be established and the structure

is deemed not to fail. However, if an intersection point cannot be obtained, then a

182


more refined procedure involving iterations to modify the Demand Curve for different

damping ratios (reflecting the inelastic energy absorbed by the structure) would have

to be made.

In the example illustrated in Figure 7.2, the performance point is determined initially

at “1” based on intercepting the Capacity Curve with the Demand Curve associated

with 5% damping. The displacement demand obtained from this first iteration

(Deff1) is then used to calculate the ductility demand and the corresponding estimated

effective damping ratio (z) for the 2nd

iteration. The new performance point at “2” is

then determined by intercepting the Capacity Curve with the new Demand Curve

associated with the (updated) effective damping ratio (z). Iteration stops when the

displacement demand determined from subsequent iterations (eg. Deff1 and Deff2)

converges.

RSa or Vb/Meff

RSd or Deff

Demand Curve

Capacity Curve: Force Diaplacement Diagram from

push-over analysis (and divided by Mg)

Deff2 Deff1

5%

10% 2

1

An iterative procedure may be required to determine the true performance point

Vb

Pushover analysis

Figure 7.2 Typical capacity spectrum (Wilson and Lam, 2003)

183


7.2.3 Linear Dynamic Procedure

The linear dynamic procedure (LDP) is used when higher mode effects are important

and consists of either the response spectrum method or time history method.

(a) Response spectrum modal analysis

The simplest form of dynamic analysis is the elastic modal and response spectrum

analysis in which the natural period and shape of deflection of the significant modes

of vibration is calculated assuming linear elastic behaviour of all elements in the

building. When a modal response spectrum is used to define the properties of the

ground shaking, no time history is involved and hence no earthquake record is

required for the analysis.

The dynamic response of a linear elastic MDOF structure can be calculated by

superimposing the responses calculated for each individual natural mode of vibration.

Each model has its own modal frequency and mode shape, and on being subjected to a

seismic action the structure vibrates in all modes simultaneously. For a given design

earthquake a smoothed “response spectrum” can be prepared of response

(acceleration, velocity or displacement) versus frequency or period as shown in

Figures 7.3 & 7.4. The response value for a given frequency then represents the peak

response of a SDOF system of that modal frequency when subjected to the chosen

ground motion.

The elastic response spectrum will vary depending on different levels of damping and

generally 5% damping is assumed for design. The response spectrum used in design

will generally represent the smoothed envelope of a number of different spectra

calculated for different possible earthquake motions at a site (Figures 7.3 & 7.4). By

subjecting the individual structure modes to their appropriate response value from the

response spectrum, the peak response in each mode can be found. These peak modal

responses are then combined to obtain the total response using the square root of the

sum of the squares (SRSS) method or quadratic combination (CQC) method or very

conservatively using the absolute sum.

184


0

0.5

1

1.5

2

2.5

3

3.5

4 S

pe

ctr

al

ord

ina

tes (

mm

/s2

)

A

B

C

D

E

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Period (S)

Figure 7.3 Typical 500 years RP acceleration response spectrum for different soil

sites (A to E) in Australia for Z = 0.08, (AS1170.4, 2007)

120

Dis

pla

cem

en

t (m

m)

100

80

60

40

20

0

A

B

C

D

E

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Period (T)

Figure 7.4 Typical 500 years RP displacement response spectrum for different of soil

sites (A to E) in Australia for Z = 0.08 (AS1170.4, 2007)

185

5

5


(b) Time history analysis

Time history analysis is the most direct method of evaluating seismic effects on a

structure. It is also the most complex, and additionally only applies to a specific

earthquake event and must be repeated many times to find the worst case. This

procedure provides valuable time history information which is not available in the

response spectrum modal analysis procedure. This process involves structural

response and the direct integration of the equations of motion. A step by step

numerical integration is performed to determine forces and displacements for a series

of short time increments from the first application of excitation to any desired time.

The motion of the system is evaluated on the basis of assumed response mechanism

for each time increment. This process is equally applicable to both linear and non

linear analyses. Linear elastic time history analysis is a relatively straight forward

procedure and varies in complexity from an analysis of a single-stick model using an

excel spreadsheet to a detailed 3-D model using commercial computer programmes.

Time history earthquake analyses are usually undertaken in the non-linear range to

explicitly account for ductility and hysteretic effects.

7.2.4 Non-linear Dynamic Procedures

The non-linear dynamic procedure (NDP) involves an inelastic time history dynamic

analysis to directly calculate the seismic response of structures with the stiffness term

replaced by hysteric behaviour of the different elements. Inelastic time history

analysis combine both the seismic demand in terms of the earthquake ground motion

input and the structural capacity (strength and deformation) in terms of inelastic

hysteretic member characterisation. NDP are computationally quite expensive and

generally quite complex requiring an in-depth understanding of both earthquake

ground motions and inelastic structural modelling. For these reasons, NDP are

typically used as a checking tool by experienced designers for important structures,

particularly where higher mode effects are important.

186


7.3 Calculation of Inter-Storey Drift in Buildings

The inter-storey drift limits specified in the code of practice can be used as the first

tier (a conservative option) approach to design the PFGFS. The detailed seismic

assessment methods reviewed in Section 7.2 can be used to calculate more accurate

estimates of the inter-storey drift in buildings. However, a detailed approach is not

considered practical for façade engineers and some simple seismic drift assessment

methods in buildings using LDP which is directly accessible from the codes are

illustrated with example calculations with increasing accuracy and complexity. The

examples are focused on Australian seismicity, but the methods can easily be

extended for other regions of the world.

7.3.1 Code Specified Limits

Seismic design codes usually specify the maximum allowable inter-story drift in

buildings at the design stage. The Australian standard for seismic actions, AS 1170.4

(2007) has outlined a maximum lateral displacement demand of 1.5% of the storey

height for a 500 year RP seismic event. This drift limit is conservative for regular

buildings as the limit includes irregular buildings and soft-storey buildings. A 1.5%

drift for a 3600 mm storey height corresponds to a lateral deflection Δ = 54 mm. If the

capacity of the façade system is higher than this limit, then the façade is considered

compliant and it is not necessary to carry out further assessment.

7.3.2 LDP - RSDmax from Response Spectrum (AS 1170.4)

For Australian seismicity inter-storey drift can be conservatively taken as the

maximum displacement (RSDmax) from the response spectrum for the appropriate soil

type. The displacement response spectrum shown in Figure 7.4 can be idealised as a

bilinear graph which represents the displacement demand linearly increases upto the

1.5 second corner period and continues as a constant plateau. Therefore, regardless of

the building details a maximum possible normalised displacement can be obtained

from the displacement response spectrum for a particular soil type. A simple

expression can be developed to calculate the inter-storey drift which should not

exceed 1.5% considering the generalised dynamic characteristic of the buildings as

follows:

187


RSD M×F ×F max ×F PF TInter-storeydrift, γi = ≤ 1.5% Eq (7.4) n × h ×100

Where,

RSDmax = Maximum displacement demand for site class from

the response spectrum

FPF = Participation factor (1.0 – 1.5)

FM = Mode shape correction multiplier (1.0 - 2.0)

FT = Torsional amplification factor (1.0 - 2.0)

n = Number of storey

h = Storey height

The participation factor can be assumed FPF = 1.5 for buildings regular in elevation

and FPF = 1.0 for single storey buildings and soft-storey buildings. A value of Fm = 1

can be used for buildings less than 5 storeys (Wilson and Lam, 2005) and

conservatively the value of Fm = 2 can be used for buildings between 5-10 storeys to

account for the curved mode shape (Wilson and Lam, 2005). The value of FT = 1 is

recommended for symmetric buildings, whilst values of FT = 1.6 and FT = 2.0 are

conservatively assumed for estimating peak drift demands for buildings that are

asymmetric in one and two directions respectively (Lumantarna et al., 2008). The

RSDmax method is suitable for buildings dominated by the first mode and typically

less than 10 storeys. Tables 7.1, 7.2 & 7.3 summarise the typical drift demands for

different building heights for 500 year RP (Z = 0.10g) and 1500 year RP (Z = 0.15g)

events on different soil types for regular, one and two directional asymmetric

buildings in accordance with AS 1170.4 (2007). The maximum drift demands

calculated for 500yr RP events were caped to 1.5%.

188


Table 7.1 Maximum drift demand on façade systems (regular buildings)

No. of

Storeys

Inter-storey drift (%), (Z=0.10g)

Class B Class C Class D

500yr 1500yr 500yr 1500yr 500yr 1500yr

3 0.47 0.69 0.67 1.00 1.05 1.50

8 0.36 0.53 0.50 0.75 0.78 1.17

10 0.28 0.42 0.39 0.58 0.64 0.94

Table 7.2 Maximum drift demand on façade systems (one directional asymmetric

building)

No. of

Storeys



500yr 1500yr 500yr 1500yr 500yr 1500yr

3 0.75 1.11 1.05 1.58 1.50 2.25

8 0.55 0.83 0.81 1.19 1.25 1.89

10 0.44 0.67 0.64 0.94 1.00 1.50

Table 7.3 Maximum drift demand on façade systems (two directional asymmetric

buildings)

No. of

Storeys



500yr 1500yr 500yr 1500yr 500yr 1500yr

3 0.94 1.39 1.33 1.97 1.50 3.00

8 0.69 1.05 1.00 1.47 1.50 2.25

10 0.55 0.83 0.81 1.19 1.25 1.89

189


7.3.3 LDP – RSDTe from Response Spectrum (AS 1170.4)

When the natural period of a building is known the normalised displacement can be

obtained from the response spectrum. However, the effective moment of inertia (Ie)

should be used to represent the cracked sections rather than the gross sectional

properties (Ig) in seismic analysis. The use of Ig instead of Ie would potentially lead to

a substantial underestimate of the inter-storey drift (McBean, 2008). Paulay and

Priestley (1992) recommends 0.40Ig for beams and 0.60Ig for normal columns as

effective moments of inertia.

In this example an average 0.50Ig has been used for all elements to provide a

reasonable estimate of the building natural period. When the effective stiffness Ke is

equal to 0.50Ig the corresponding Te can be calculated using Equation 7.3 where the

Me is the effective mass and Ke is the effective stiffness. Substituting 0.50Ig in

Equation 7.3 results in the period being extended to Te = 1.4T. The typical

displacement demand corresponding to the effective stiffness (RSDTe) is shown in

Figure 7.5. The RSDTe method also is suitable for buildings dominated by the first

mode and typically less than 10 storeys for which the natural period of the buildings is

required.

M Effective natural period, T e = 2π e Eq (7.5)

K e

T

Te

RSDTe

Acc

eler

atio

n

Displacement

Figure 7.5 Displacement for effective stiffness of a building from the ADRS diagram

190


The inter-storey drift can be estimated from the RSDTe using the Response Spectrum

as follows:

RSD ×F ×F ×F Te PF M TInter-storeydrift, γ i = ≤ 1.5% Eq (7.6) n×h ×100

Where RSDTe = displacement demand corresponding to effective period (Te) and the

remaining factors are the same as explained in Equation 7.4. AS 1170.4 (2007)

specifies that a rigorous structural analysis could be performed to calculate the natural

periods of the structure or the equation specified in the standard could be used to

calculate the fundamental period of the structure. However, for simplicity in this

study, the natural period of the buildings have been crudely estimated using, T = 0.1n

where ‘n’ is the number of storeys.

Table 7.4 compares the typical drift demands for the 500 year RP events with Z =

0.10g on different soil types for regular buildings (Ig and 0.50Ig) in accordance with

AS 1170.4 (2007). Similarly the inter-storey drift can be calculated for irregular

buildings by multiplying the regular drift values by the representative torsional

amplification factors. The inter-storey drift from the RSDTe method in Table 7.4 is

clearly less than the RSDmax method in Table 7.1 for 500 year RP event for buildings

with Te < 1.5 seconds. Therefore, the RSDTe method is more efficient to calculate

inter-storey drift in the first mode dominant buildings compared to the RSDmax

method. For buildings where higher mode effects are important, the response

spectrum modal analysis should be used.

Table 7.4 Maximum drift demand on façade systems (regular buildings)

No. of

Storeys

Period

T (s)

0.50Ig

period

Te (s)



Ig 0.50Ig Ig 0.50Ig Ig 0.50Ig

3 0.3 0.43 0.09 0.13 0.12 0.19 0.12 0.23

8 0.8 1.13 0.19 0.26 0.26 0.37 0.42 0.59

10 1.0 1.41 0.19 0.26 0.26 0.37 0.42 0.59

191


7.3.4 LDP - Response Spectrum Modal Analysis Method (AS 1170.4)

The modal analysis method is suitable for buildings where higher mode effects are

important. The soil class and modal properties including natural period and mode

shapes of the building are required to undertake the analysis. The method has been

applied using the first three modes for three typical buildings with 26, 40, and 66

floors in the low-to-moderate seismic region. Table 7.5 summarises the details and

modal periods of the typical example buildings which natural periods have been

calculated from the gross section properties Ig. The inter-storey drift was calculated

using Equation 7.6 for each mode assuming a regular building configuration whilst

the well-known modal combination rule, the square root of the sum of the squares

method was used to calculate the combined modal inter-storey drifts.

Table 7.6 compares the maximum drift demand for regular buildings (Ig and Ie =

0.5Ig) subject to the 500 year RP event for different soil classes in accordance with AS

1170.4 (2007). Similarly the inter-storey drift in irregular buildings can be calculated

by multiplying the drift values by the representative torsional amplification factors.

The maximum inter-storey drift 0.2% was calculated for the 40 storey building using

the effective stiffness and a Class-D soil with a 500 year RP event. Linear and non

linear time history analyses could be used to more accurately calculate the inter-storey

drift with an increased level of complexity.

Table 7.5 Summary of the buildings details

Building

Reference Description Height Number

of floors

Natural period (sec) Reference

and location (m)

Mode 1 Mode 2 Mode 3

1

Singapore

Frame-tube

Office block 280 66 5.4 1.5 0.7

(Brownjohn

and Pan,

2001)

2

Melbourne

Central core

steel frame

Office block

152 40 3.8 1.0 0.6

(Swaddiwudh

ipong et al.,

2002)

3

Singapore

Concrete

Office block 91 26 1.5 0.4 0.2

(Brownjohn

and Pan,

2001)

192


Table 7.6 Maximum drift on buildings (Z = 0.10g)

No. of

Storeys

Inter-storey drift (%)


Ig 0.50Ig Ig 0.50Ig Ig 0.50Ig

26 0.08 0.08 0.10 0.12 0.16 0.17

40 0.08 0.09 0.10 0.12 0.17 0.20

66 0.05 0.05 0.06 0.08 0.11 0.12

7.4 In-plane seismic design of PFGFS

The racking (in-plane seismic) performance of PFGFS is dependent on the inter-

storey drift demand from the building and the in-plane racking displacement capacity

of the system. The maximum inter-storey drift for the building could be obtained

from; (a) the structural engineer who designs the building; or (b) conducting a seismic

analysis procedure as described in Section 7.2; or (c) using the rapid assessment

procedure described in the Section 7.3. On the other hand, the in-plane racking

displacement capacity of a typical PFGFS could be calculated from the study

presented in the Chapters 5 & 6.

The in-plane racking capacity of the PFGFS in this research study has been obtained

from a static monotonic load application. However, buildings subjected to seismic

actions will experience reversed cyclic swaying. The ultimate drift capacity obtained

from the monotonic laboratory experiment and the analytical model may over predict

the racking performance. So some conservatism is warranted when assessing the

cyclic in-plane racking capacity. There are examples in façade system where the

design in-plane drift capacity is conservatively assumed equal to the drift associated

with the rigid body articulation (translation) only. Three similar practical applications

of rigid body articulations in façade system design are described below:

1. In architectural precast panel design, the in-plane seismic design standards

require that connections and panel joints should allow for the inter-story drift

caused by relative seismic displacements. Connection details, and joint

193


locations and sizes between cladding panels should be designed to

accommodate any shrinkage, story drift, or other expected movement of the

structure, such as sway in tall, slender structures (PCI, 2011).

2. ASCE7-10 (2010) recommends that sufficient glass-to-frame clearance is

provided in framed glass façade systems such that physical contact between

the glass and frame will not occur at the design drift as demonstrated by

Equation 2.2. This Equation calculates the in-plane drift capacity of the

framed glass façade system from the rigid body translation developed at the

gap provided between the glass to the frame.

3. Desai et al. (2005) and Gowda and Heydari (2009) recommended the

introduction of a simple articulation in PFGFS for high seismic regions.

Specially designed spider arms with large horizontally slotted holes (Figure

2.30) were used to accommodate the drift by allowing isolated horizontal

translation as shown in Figure 2.31 to address the criteria of a drift limit of

2.0% to 2.5% for cladding systems as per the seismic provisions of the

California Building Code (CBC, 2002). The sizes of the slotted holes were

calculated according to the height of the glass façade and the drift demand.

The in-plan drift capacity of the system was achieved by the rigid body

translation at the large horizontally slotted holes.

Similarly, the in-plane racking capacity of the PFGFS from the rigid body translation

provided at the built-in standard gap could be used as the design drift capacity against

cyclic racking action. The rigid body translation at the built-in standard gaps of

PFGFS occurred within the application of a small racking load and therefore, the

system is structurally within the elastic state from the experimental and the analytical

test results undertaken in this study. Further, if the drift capacity from the rigid body

translation associated with the built-in standard gap is less than the drift demand from

the building, additional tolerances (articulation) could be introduced.

Most buildings in Australia are structurally designed for the 500 year RP seismic

event and require the façade system to accommodate the associated drift imposed.

The façade in-plane racking assessment procedure developed in this study consists of

194


five steps and provided that the drift demand calculated from any of the steps is less

than the drift capacity, then the façade system is deemed satisfactory:

Step 1. Compare the drift limit specified in the AS1170.4 (2007) standard

provision with the drift capacity of the façade system. If the drift capacity is less

than the drift limit then proceed to Step 2. For example, the PFGFS with X-type

spider arm has a drift capacity of 1% from the rigid body translation at the built-

in standard gaps which is less than the AS1170.4 (2007) limit of 1.5% drift.

Step 2. Compare the drift demand from the RSDmax method and the drift

capacity of the façade system. If the drift capacity is less than the drift demand

then proceed to Step 3. A typical racking assessment procedure for first mode

dominant regular buildings less than 10 storeys consisting of PFGFS with X-

type spider arms (1% drift capacity) is shown in Table 7.7 where the in-plane

drift calculated from the RSDmax method presented in Table 7.1 was used in this

assessment.

Step 3. Similar to Step 2, except the RSDTe method rather than RSDmax

method is used to calculate the drift demand. If the drift demand exceeds the

drift capacity then proceed to Step 4.

Step 4. Carry out a detailed seismic analysis to calculate the structural drift

demand and compare with the drift capacity of the façade system. If the drift

demand exceeds the drift capacity, proceed to Step 5.

Step 5. Introduce additional articulation into the façade system to satisfy the

drift demand and check the façade system performance.

A similar procedure can be applied for irregular buildings by factoring the drift

demands to allow for the torsional or soft storey irregularities. High rise regular

buildings considered in Section 7.3 resulted in less than 0.2% drift for the 500 year

RP seismic event. Since, the PFGFS with X-type and K-type spider arms as tested

have a racking capacity of 1% and 2.9% due the rigid body translation at the built-in

195


standard gaps, a detailed seismic analysis or an experimental test for the PFGFS is not

required.

Table 7.7 In-plane racking performance of PFGFS with X-type spider arms for first

mode dominant regular buildings (Drift calculated from the RSDmax method)

No. of

Storeys Soil type

Return

period

(years)

Inter-

storey

drift (%)

Drift capacity

of PFGFS with

X-type spider

arms (%)

Comments

3

B 500

0.47 1.0 Safe

8 0.36 1.0 Safe

10 0.28 1.0 Safe

3

B 1500

0.69 1.0 Safe

8 0.53 1.0 Safe

10 0.42 1.0 Safe

3

C 500

0.67 1.0 Safe

8 0.5 1.0 Safe

10 0.39 1.0 Safe

3

C 1500

1.0 1.0 Safe

8 0.75 1.0 Safe

10 0.58 1.0 Safe

3

D 500

1.05 1.0 Further study is

required

8 0.78 1.0 Safe

10 0.64 1.0 Safe

3

D 1500

1.5 1.0 Further study is

required

8 1.17 1.0 Further study is

required

10 0.94 1.0 Safe

196


7.5 Recommended Detailing of PFGFS

The racking capacity of the PFGFS without any boundary restraints (free to rack) and

similar to Test #1 and Test #2 can be estimated from the rigid body translation at the

built-in standard gaps. For example, the PFGFS examples shown in Figures 7.6 & 7.7

with X-type spider arms are representative of Test #1. However, there are examples in

practice like the PFGFS shown in Figures 7.8 & 7.9 with ‘K’ type spider arms, where

some glass panels are directly connected to the building frame or wall for structural

purposes such as supporting the dead load or resisting the wind load. This means the

building transfers the racking load directly to the perimeter glass panels which may

cause excessive stresses on the glass since the glass panels are effectively restrained

from rotating locally due to the imposed boundary conditions.

The racking performance of the PFGFS shown in Figure 7.8 could be increased

further by adopting the following measures,

a) Introduce additional façade structural support frames at both perimeter

boundaries (adjacent to the walls) to ensure the racking movement is

not restrained.

b) Structural sealant was used in this façade and this was confirmed by

the glaziers. The internal glass panels should be joined using weather

sealant with low shear modulus instead of structural sealant since the

sealant is used for weather protection only.

c) Introduce special purpose sealant with very low shear modulus which

can resist high displacement at the perimeter glass panel joints with

wall partitions. Alternatively, rebate channels could be used at the

perimeter with gaskets which have a low frictional coefficient against

in-plane movement.

In summary, the glass panels should be connected to the main structure (building

frame) using the façade structural support frame only to allow the façade to rack

freely with rigid body articulation.

197


Rebate channels are the ideal solution at the perimeter glass edges to weather seal the

buildings. The rebate channels with gaskets holding the glass panel edges should have

sufficient clearance to accommodate the differential movement of glass panels in the

in-plane directions and should allow the glass edges to move with minimal frictional

force. If the perimeter glass panels are weather sealed using sealant the sealant should

allow enough movement to accommodate the racking displacement without inducing

excessive stresses on the glass panels. Low modulus or special purpose sealants are

recommended for this application. Structural sealant should not be used at this

location as the sealant could reduce the movement and rotation of the perimeter glass

panels and could create excessive stresses on the system.

Figure 7.6 Front view of a storefront PFGFS with ‘X’ type spider arms in

Melbourne, Australia

198


Figure 7.7 Side view of a storefront PFGFS with ‘X’ type spider arms in Melbourne,

Australia (the perimeter glass panels are free to move)


using structural sealant in Melbourne, Australia

199



using sealant and a two way spider arm used to align the glass panels

7.6 Recommended Selection Guide for Façade Engineers

A quick selection guide recommended for façade engineers is summarised in Figure

7.10 to increase the drift capacity of typical PGFFS due to the rigid body translation at

the standard built-in gaps. In addition, when the PFGFS is designed assuming only the

rigid body translation at built-in gaps:

(a) The racking load and the bending stress on the glass panels and the structural

support frame can be reduced by reducing the spider arm eccentricity in the

out-of-plane direction.

(b) The bending stress in the glass panel can be reduced by using stronger spider

arms that reduce the out-of-plane bending.

(c) The bending stress in the glass panels can be reduced by using larger diameter

bolt heads to connect the glass panels. For example swivel bolt fitting is better

than countersunk bolt fitting.

200


Increasing Drift Capacity of PFGFS:

Glass panel geometry

Landscape Square Portrait

Glass panel thickness

10mm thick 12mm thick 15mm thick

Panel to panel sealant type

Structural sealant (high modulus)

Weather sealant (medium modulus)

Special purpose sealant (low modulus)

Silicon sealant thickness

6mm thick 8mm thick 10mm thick

Glass panel size

2400mmx2400mm 1800mmx1800mm 1200mmx1200mm

Figure 7.10 PFGFS recommendations to increase the drift capacity

201


7.7 Summary and Conclusion

This chapter reviewed the different seismic analysis procedures to calculate the inter-

storey drift in buildings and described rapid assessment methods to calculate the inter-

storey drift with example calculations and conservative factors were presented for

considering the torsional behaviour of buildings. The seismic assessment of glass

façade systems requires an estimate of the likely drift demand from the building.

Codified provisions for in-plane drift limits on glass façade systems can be used as a

conservative option. However, analysis results indicated that the inter-storey drift can

be much smaller than the 1.5% limit in AS 1170.4 (2007) for most buildings in

Australia for the 500 year RP event except for soft storey structures.

The in-plane racking drift capacity of typical PFGFS could be estimated using the

approach described in Chapter 6. Conservatively, the in-plane racking capacity of the

PFGFS from only the rigid body translation at the built-in standard gaps could be used

as the design in-plane drift capacity. The drift capacity could be increased further by

introducing additional articulation tolerances at the bolted connections at the spider

arms or façade structural support frame. Care should be taken in detailing the

boundary condition of the perimeter glass panels to ensure the full racking capacity of

the PFGFS from the rigid body translation at the built-in standard gaps is achieved.

A quick selection guide has been presented for façade engineers to increase the drift

capacity of PFGFS by optimising the detailing and configuration of various

components that make up the façade system.

202

Chapter 8 CONCLUSIONS AND RECOMMENDATIONS

Chapter 8

8 CONCLUSIONS AND RECOMMENDATIONS

Based on the conclusions presented at the end of each chapter, the following major

findings from the research project have been drawn. The conclusions are summarised

into four sections namely, (a) research background; (b) experimental tests and results;

(c) FE analysis and results; and (d) seismic assessment of façade systems.


8.1.1 Research Background

1. The design of glass façade system involves a sequence of steps including

visual consideration, weather proofing and structural evaluation. Visual

assessment covers the overall aesthetics whilst weather proofing includes air

leakage control, vapour diffusion control, heat loss and gain control and rain

water penetration control. Structurally the glass façade system is designed for

in-plane and out-of-plane loads and movements. Self-weight, thermal

expansion, spandrel beam deflection and in-plane building movements due to

wind and seismic loads are considered for in-plane design whilst wind load on

the glass panel, mullion, transom and structural support frames are considered

for out-of-plane design.

2. Despite the growing popularity of point fixed glass façade systems, there is

very limited published research on the in-plane racking behaviour of such

contemporary systems. A very limited number of experimental tests and

analytical studies had been conducted to date to calculate the in-plane load and

drift capacity of standard glass panels with bolted connections and there is

little or no standard practice available for in-plane loading.

203


3. An overview of the design methodology for both framed and point fixed glass

façade systems in Australia was presented. The design methodology for

framed glass façade system is well established. Although, the design

methodology for a unitized framed glass façade system is well established, the

design of point fixed glass façade systems is less established and a

recommended methodology has been presented for unitized framed glass

façade for both in-plane and out-of-plane loading. For the point fixed glass

façade system out-of-plane loading, typical techniques for installation and

simple formulas to determine the maximum stresses and deflection were

presented. For in-plane loading, use of slotted holes and swivel bolts to

accommodate movements were suggested.

8.1.2 Experimental Test and Results

4. A specific test setup was developed and used to undertake racking tests on

contemporary PFGFS commonly used in low-to-moderate seismic regions

such as Australia. Two unique full scale in-plane racking laboratory tests (Test

#1 and Test #2) of typical PFGFS were conducted. Spider arms are configured

as X-type (Figure 4.1) or K-type (Figure 4.2) in this study depending on the

type of fixity at the structural support frame. The spider arms are snug

tightened to the structural support frame in different ways. The X-type spider

arms are connected to the structural support frame using a single bolt to allow

in-plane rotation of the glass panels at the spider arm-to-structural support

frame connection whilst K-type spider are connected to the structural support

frame using double bolts and do not allow the glass panels to rotate at the

spider arm to structural support frame connection but allow sliding at the base

connection in the vertical direction.

There are different types of bolt fittings available in the market to connect the

glass to spider arms namely, countersunk, button head, swivel connections.

Countersunk and button head bolt fittings are the most common and the

cheapest options whilst swivel connections are used when excessive stress

204


developments expected. Test #1 was performed with X-type spider arms and

countersunk bolt fittings whilst Test #2 was performed with K-type spider

arms and button head bolt fittings.

5. For Test #1, a maximum in-plane lateral displacement of 58 mm was

measured with a corresponding 16kN racking load before failure. Surprisingly,

this resulted in a maximum of 2.1% in-plane drift capacity for the system with

minor damage to the sealant and yielding of a perimeter spider arm, before

catastrophic failure of one of the glass panels. A maximum of approximately

8.5mm relative out-of-plane movement was measured at the adjacent glass

bolted connection.

6. The maximum drift of 5.25% (143mm) for Test #2 was much larger than

initially anticipated and demonstrated that the 2x2 system was surprisingly

tolerant to in-plane drift. Damage along the vertical silicon sealant was noticed

at 2.0% drift. The spider arms began to yield and distort at 3.3% drift whilst

the base plate of the spider arms also commenced to yield. The system

continued to have great drift capacity until failure of one of the glass panels at

5.25% drift due to excessive bending stresses (from out-of-plane displacement

of the spider arms) combined with the in-plane diagonal tensile stresses. A

maximum of 32mm of relative vertical translation at the internal central spider

arms and approximately 10mm of relative out-of-plane deformation was

measured between adjacent bolted connections using the photogrammetry

results.

7. It was revealed from the experimental results that in 2x2 systems with X-type

spider arms with countersunk bolt connections (Test #1), the applied racking

displacement was accommodated by three mechanisms:

o In-plane rigid body rotation of the spider arms;

o Rigid body translation facilitated by the built-in standard gaps between

the bolts and holes with in the spider arm and structural support frame

connections; and

o Deformations and distortions of the spider arms.

205


8. Similarly, it was revealed from the experimental results that in 2x2 systems

with K-type spider arms with button head bolt connections (Test #2), the

applied racking displacement was accommodated by three mechanisms:

o Rigid body vertical translation of the spider arms at the base standard

built-in gaps;

o Rigid body translation facilitated by the built-in standard gaps between

the bolts and holes with in the glass panels and spider arm connections;

and

o Deformations and distortions of the spider arms.

8.1.3 FE Analytical Model and Results

9. Detailed three-dimensional non-linear finite element models (FE models) were

developed to replicate the laboratory tests. The results obtained from the FE

models were benchmarked against the test data namely, pushover curve,

failure stress and out-of-plane deformation of glass panels. Using a trial and

error tunning approach by modifying the non-linear spring stiffness to

represent the rigid body translations followed by bearing actions and adopting

a representative non-linear silicon sealant model, an excellent correlation was

achieved between the experimental and analytical results. For Test #1 with X-

type spider arms, the experimental pushover curve with 2.1% drift at the

ultimate limit state was benchmarked with 2.0% drift from the analytical

pushover curve at failure of a glass panel by exceeding the nominal strength of

96MPa in accordance to AS 1288 (2006). A maximum of 8.5mm relative out-

of-plane movement was measured between the adjacent glass bolted

connections in experiment was benchmarked with an 8.0mm of relative out-

of-plane movement from the analytical model at the same set of glass bolted

connections.

10. Further detailed FE analyses were conducted to evaluate the Test #1 racking

capacity from each mechanism. The analyses showed that a significant amount

of the drift capacity was contributed from the rigid body translation facilitated

206


by the built-in standard gaps between the bolts and holes with in the spider

arm and structural support frame connections (1.1% drift). The estimated drift

contributions from each mechanisms using FE models are summarised as

follows:

a) In-plane rigid body rotation of the spider arms – 0.4%.

b) Rigid body translation facilitated by the built-in standard gaps between

the bolts and holes with in the spider arm and structural support frame

connections – 1.1%.


11. Simple expressions were presented to verify the drift capacities from the rigid

body translation facilitated by the built-in standard gaps between the bolts and

holes with in the spider arm and structural support frame connections. The

possible rigid body translations in the contemporary design of PFGFS were

discussed with FE analyses and verified along with analytical expressions for

different configurations. The in-plane drift capacities of PFGFS with different

geometry due to rigid body translations at the standard built-in gaps were

presented. Similarly the analytical study was extended to multiple grid Test #1

façade systems (all properties including the glass panel size were similar to the

2x2 Test #1 with 3x3 and 4x4 systems). Drift contributions from the rigid

body translation at the standard built-in gaps for multiple grid façade systems

were calculated using FE models and confirmed with simple trigonometric

expressions. Typical drift capacities from the rigid body translation of multiple

grid systems with different geometries were also presented.

12. The study was further extended to PFGFS with rigidly connected ‘X’ type

spider arms (in-plane rotation of the spider arm is not allowed) with special

articulations introduced to increase the drift capacity of such systems. Simple

expressions were presented using trigonometry to calculate the drift capacity

of such systems and confirmed with FE models. The in-plane drift capacities

of such systems due to rigid body translations at the articulation features also

were presented.

207


13. The FE models benchmarked against the experimental test results for Tests #1

with X-type spider arms was used to predict the racking performance of

PFGFS with different configurations. Variations related to built-in standard

gap, sealant type, sealant thickness, glass geometry, and glass thickness were

considered in the parametric study for 2x2 façade grid systems. The

parametric study indicated that in Test #1 with X-type spider arms and

countersunk bolt connection the racking performance of the PFGFS increased

when:

a) Slotted holes (built-in standard gap) were introduced at the structural

support frame spider arms bolted connections

b) The stiffness of the silicone sealant was reduced

c) The silicone sealant thickness was increased

d) The height to width ratio of the glass panels was increased

e) The glass thickness was increased

14. Similar to Test #1 with X-type spider arms, detailed three-dimensional non

linear FE models were developed to replicate the laboratory tests for Test #2

with K-type spider arms. An excellent correlation was achieved between the

experimental and analytical results. The experimental pushover curve with

5.25% drift at the ultimate limit state was benchmarked with 4.75% drift from

the analytical pushover curve at failure of a glass panel by exceeding the

nominal strength of 96MPa in accordance to AS 1288. A 35mm of vertical

relative spider arm translation was estimated from the analytical model at the

internal central spider arm compared to the 32mm in experimental results. A

maximum of 10mm relative out-of-plane movement was measured at the glass

bolted connection in experiment was benchmarked with an 8mm of relative

out-of-plane movement from the analytical model at the same set of glass

bolted connections.

15. Detail FE analyses were conducted to express the racking capacity from each

mechanism in Test #2 with K-type spider arms. The analyses showed that a

significant amount of the drift capacity was attributed to the rigid body vertical

translation of the spider arms at the base standard built-in gap (vertical sliding,

208


2.9% drift) connections. The calculated drift contributions from each

mechanisms are summarised as follows:

a) Rigid body vertical translation of the spider arms at the base standard

built-in gap – 2.9%.

b) Rigid body translation facilitated by the built-in standard gaps between

the bolts and holes with in the glass panels and spider arm connections

– 1.3%.


2. Basic trigonometric expressions and FE models were used to verify the drift

capacities from the rigid body vertical translation of the spider arms at the

base standard built-in gaps. The possible rigid body translations in the

contemporary design of PFGFS were discussed with FE analyses and verified

along with analytical expressions for different configurations. The drift

capacities of such systems due rigid body vertical translation at the spider arm

base connections only were calculated and presented. Interestingly, the

analytical expression showed that the drift capacity of multiple grid system

similar to Test #2 (all properties were similar to the 2x2, Test #2 with 3x3 and

4x4 systems) due to the rigid body vertical translation of the spider arms at the

base standard built-in gaps is same as Test #2, 2x2 systems.

16. The parametric study indicated that in Test #2 with K-type spider arms the

racking performance of PFGFS increased when:

a) The stiffness of the silicone sealant was reduced

b) The silicone sealant thickness was increased

c) The height to width ratio of the glass panels was increased

d) The glass thickness was increased

209


8.1.4 Seismic Assessment of Façade Systems

17. The seismic assessment of glass façade systems requires an estimate of the

likely drift demand from the building. Codified provisions for maximum

allowable in-plane drift limits on façade systems can be used as a conservative

option. However, researchers highlighted that the inter-storey drift is much

smaller than the 1.5% limit in AS 1170.4 (2007) for most buildings in

Australia for the design 500 year return period event except for soft storey

structures. Therefore, the standard seismic drift analysis methods could be

used for estimating the optimum in-plane seismic drift demands on glass

façade systems and this will lead to an economical in-plane seismic design of

façade system.

18. Different seismic analysis procedures to calculate the inter-storey drift in

buildings were reviewed. Based on that, rapid assessment methods were

described to calculate the building inter-storey drift with example calculations.

Conservative factors recommended by Lumantarna et al (2008) were adopted

for the torsional behaviour of buildings in inter-storey drift calculations.

19. The ultimate drift capacity of PFGFS is more than the drift capacity due to

rigid body translation at standard built-in gaps. Conservatively, the in-plane

drift capacity of the PFGFS from only the rigid body translation at the built-in

gaps could be used as the design in-plane drift capacity. The drift capacity

could be increased further by introducing additional articulation tolerances at

the bolted connections at the spider arms or façade structural support frame.

Care should be taken in detailing the boundary conditions of the perimeter

glass panels to ensure the full racking capacity of the PFGFS from the rigid

body translation at the built-in gaps can be achieved.

20. A quick selection guide was summarised in Figure 7.10 for façade engineers

to increase the drift capacity of PFGFS systems by optimising the detailing

and configuration of various components that make up the façade system.

210


8.2 Recommendations for Future Research

The research project examined the in-plane racking performance of point fixed glass

façade systems. Although a number of important issues and results have been

described, there remain some areas where further research would be beneficial.

1. The laboratory tests were conducted with strong and articulated structural

support frames. Different types of structural supports are available in practice

for example, cable supported PFGFS. Further research is recommended to

fully assess the racking performance of such support systems, although it is

expected that these more flexible systems will tend to increase the drift

capacity of PFGFS.

2. Modelling the yielding and distortion of the spider arms was beyond the scope

of this study however, it is recommended as a subject of further research.

3. Eccentricity of the spider arms, bolt head geometry and the bolt head diameter

affect the stress development at the glass holes. In addition, different types of

bolt fittings with different geometries are available in the market to connect

the glass to spider arms namely; countersunk, button head and swivel

connections. Further detailed FE analytical works are recommended to

quantify these parameters.

4. The pseudo-static pushover analysis provides a load-deflection curve that can

be directly compared with the FE analysis. Since the strain rate effects

associated with earthquake excitation are quite modest, the pseudo-static test

results are representative of the dynamic behaviour. The non-linear stiffness

parameters used in the static FE analysis could be extended to undertake a

dynamic analysis. Interestingly, the stiffness properties obtained from the

experimental program are directly applicable for a dynamic analysis, which

could form a topic for future research.

211


5. Conducting component tests separately and producing the pushover curve of a

façade system based on the component tests incorporated into FE models

would be an economical evaluation approach. This approach can be practised

using the test results from Test #1 & Test #2 in further studies.

6. Further experimental tests and analytical works are recommended to assess the

racking performance of PFGFS with irregular shapes of glass panels and the

complexities of deformations compatibility at corner glass panels.

212

Chapter 0

REFERENCES:

AAMA 2001a. Recommended static test method for evaluating curtain wall and

storefront systems subjected to seismic and wind induced inter-story drifts.

Publication No. AAMA 501.4-01,, Schaumburg, III.

AAMA 2001b. Recommended dynamic test method for determining the seismic drift

causing glass fallout from a wall system. Publication No. AAMA 501. 6-01,,

Schaumburg, III.

ANSYS12.1 2010. ANSYS 12.1 release, Structural Mechanics Solutions.

Southpointe, 275 Technology Drive, Canonsburg, PA 15317, U.S.A.

AREDDY, J. T. 2010. Apple’s Glass Temple, Made in China. The Wall Street

Journal, http://blogs.wsj.com/chinarealtime/2010/08/30/apples-glass-temple

made-in-china/tab/print/.

AS1170.4 2007. Structural design actions, Part 4: Earthquake Actions in Australia.

Australian Standard, Standards Australia, 1 The Crescent, Homebush, NSW

2140.

AS1288 2006. Glass in buildings - Selection and installation. Australian Standard,

Standards Australia, 1 The Crescent, Homebush, NSW 2140.

AS2047 1999. Windows in buildings - Selection and installation. Australian

Standard, Standards Australia International Ltd, GPO Box 5420, Sydney,

NSW 2001, Australia.

AS3600 2009. Concrete structures. Australian Standard, Standards Australia, 1 The

Crescent, Homebush, NSW 2140.

AS4100 1998. Steel strutures. Australian Standard, Standards Australia, 1 The

Crescent, Homebush, NSW 2140.

AS/NZS1170.0 2002. Structural design actions, Part 0: General principles.

Australian/New Zealand Standard, Published jointly by: Standards Australia,

1 The Crescent, Homebush, NSW 2140 and Standards New Zealand Level 10,

Radio New Zealand House, 155 The Terrace, Wellington 6001, New Zealand.

AS/NZS1170.2 2002. Structural design actions, Part 4: Wind Actions Australian/New

Zealand Standard, Published jointly by: Standards Australia, 1 The Crescent,

Homebush, NSW 2140 and Standards New Zealand Level 10, Radio New

Zealand House, 155 The Terrace, Wellington 6001, New Zealand.

AS/NZS1664.2 1997. Aluminium structures, Part 1: Limit state design.

Australian/New Zealand Standard, Published jointly by: Standards Australia,

1 The Crescent, Homebush, NSW 2140 and Standards New Zealand Level 10,

Radio New Zealand House, 155 The Terrace, Wellington 6001, New Zealand.

ASCE7-10 2010. Minimum design loads for buildings and other structures. 1801

Alexander Bell Drive, Reston, Virginia 20191-4400: The American Society of

Civil Engineers.

213

Chapter 0

ASCE 41-06 2007. Seismic Rehabilitation of Existing Buildings. ASCE/SEI Standard

41-06. Reston, VA.

ATC 40 1996. Seismic Evaluation and Retrofit of the Concrete Buildings. APPLIED

TECHNOLOGY COUNCIL, Redwood City, California 94065: Report No.

SSC 96-01.

AYRES, J. M. & SUN, T. Y. 1973. Non-structural damage, the San Fernando

earthquake of February 9,1971. US Department of Commerce, National

Oceanic and Atmospheric Administration.

BAIRD, A., PLAERMO A, PAMPANIN S, RICCIO P & TASLIGEDIK A S 2011.

Focusing on reducing the earthquake damage to facade systems. Bulletin of the

New Zealand society for earthquake engineering, 44, 2.

BCA 2011. Building Code of Australia. Canberra, A.C.T: Australian Building Codes

Board.

BEHR, R. A. 1998. Seismic performance of architectural glass in mid-rise curtain

wall. Journal of Architectural Engineering 4, 94-98.

BEHR, R. A. 2001. Architectural glass for earthquake-resistant buildings. 8th Glass

Processing Days Conference. Tampere, Finland.

BEHR, R. A. 2006. Design of architectural glazing to resist earthquakes. Journal of

Architectural Engineering, 12, 122-128.

BEHR, R. A., BELARBI, A. & A, B. T. 1995. Seismic Performance of Architectural

Glass. Earthquake Spectra 11.

BEHR, R. A., BELARBI, A. & CULP, J. H. 1995b. Dynamic racking tests of curtain

wall glass elements with in-plane and out-of-plane motions. Earthquake

Engineering and Structural Dynamics, 24, 1-14.

BEHR, R. A., KREMER, P. A. & MEMARI, A. M. 2003. Earthquake damage

resistant architectural glass panels. Proceedings of 2003 -Architectural

Engineering Conference- Building Integration Solution. Austin, TX ASCE

AEI.

BERNARD, F. & DAUDEVILLE, L. 2009. Point fixing in annealed and tempered

glass structures: Modeling and optimization of bolted connections. Eng.

Struct., 31, 946-955.

BONDI, S. & MCCLELLAND, N. 2009. Capturing structural silicone non-linear

behaviour via the finite element method. Glass processing days 2009,

Tampere, Finland, 183–185.

BOUWKAMP, J. G. 1960. Behaviour of window panels under in-plane forces.

Structures Material Research Series, University of California, Berkeley, CA,

100.

BROWNJOHN, J. M. W. & PAN, T. C. 2001. Response of tall building to weak long

distance earthquakes. Earthquake Engineering and Structural Dynamics, 30,

709-729.

BRUNGS, M. P. & SUGENG, X. Y. 1995. Some solutions to the nickel sulphide

problem in toughened glass. Glass Technology 36. No. 4.

214

Chapter 0

CASTILONE, R. J., GLAESEMANN, G. S. & HANSON, T. A. Relationship

Between Mirror Dimensions and Failure Stress for Optical Fibers. Optical

Fiber and Fiber Component Mechanical Reliability and Testing II, M. John

Matthewson, Charles R. Kurkjian, Editors, Proceedings of SPIE Vol. 4639

(2002), 2002.

CBC 2002. Californaia building standards commission and the international

conference of bulding officials. California building code: California code of

regulations, Title 24, Part 2 (Volume 1).

CHASZAR, A. 2003. Hybrid laminations for structural glass. Glass processing days

2003. Tampere, Finland.

COUNCIL ON TALL BUILDINGS GROUP SB 1979. Structural Design of Tall

Steel Building, Volume SB of Monograph on Planning and Design of Tall

Buildings, New York, ASCE.

DESAI, P., GOLMOHAMMADI, A., GARLIPP, R. & GOWDA, B. New point

supported glass seismic system. Proc. of The First. International Conference

on Advances in Experimental Structural Engineering (AESE 2005), 2005

Nagoya, Japan.

DIYTRADE. 2011.

http://www.diytrade.com/china/4/products/2611165/Insulating_glass.html

[Online]. [Accessed 1st of February 2011].

EMPORIS. 2007. Exposed structure, Emporis standards.

http:standards.emporis.com/?nav=realestate&lng=3&esn=41302, viewed 1st

November 2007 [Online].

EVANS, D. & KENETT, E. 1988. Glass damage in the September 19, 1985 Mexico

city earthquake. Steven Winter Associates, Inc, New York, 1988.

FARDIPOUR, M., LUMANTARNA, E., LAM, N. T. K., WILSON, J. L. & GAD, E.

F. 2011. Drift demand predictions in low ot moderated seismicity regions.

Australian Journal of Structural Engineering, 11.

FEMA 356 2000. PRESTANDARD AND COMMENTARY FOR THE SEISMIC

REHABILITATION OF BUILDINGS. Washington, D.C.: Federal Emergency

Management Agency.

FEMA 440 2005. IMPROVEMENT OF NONLINEAR STATIC SEISMIC

ANALYSIS PROCEDURES. Applied Technology Council (ATC-55 Project).

Redwood City, California 94065.

FEMA E-74 2011. Reducing the Risks of Nonstructural Earthquake Damage - A

Practical Guide. Washington, D.C: FEDERAL EMERGENCY

MANAGEMENT AGENCY.

FRECHETTE, V. 1990. Failure Analyses of Brittle Material; Advances in Ceramics.

The American Ceramic Society, Westerville, Ohio, 28.

FREEMAN, S. A., NICOLETTI, J. P. & TYRELL, J. V. Evaluations of Existing

Buildings for Seismic Risk - A Case Study of Puget Sound Naval Shipyard,

Bremerton, Washington. Proceedings of U.S. National Conference on

Earthquake Engineering, Berkeley, U.S.A., pp. 113-122, 1975.

215

Chapter 0

G. JAMES. 2010. G. James Glass and Aluminium, Glass Hand Book, G James is

Glass [Online]. [Accessed 06th May 2010].

GATES, W. E. & MCGAVIN, G. Lessons learned from the 1994 Northridge

earthquake on the vulnerability of non-structural systems. Proceedings of the

Seminar on Seismic Design, Retrofit, and Performance of Non-structural

Components, ATC 29-1, 1998 San Francisco, CA. 93-106.

GOWDA, B. & HEYDARI, N. 2009. High Displacement Glass Seismic Systems.

ASCE Practice Periodical on Structural Design and Construction.

HALDIMANN, M., LUIBLE, A. & OVEREND, M. 2008. Structural Use of Glass,

Zurich, Switzerland, International Association for Bridge and Structural

Engineering (IABSE).

HARTER, D. 1994. Earthquake in Los Angeles, Glazing. California Glass

Association.

HOSSEINIA, M. 2005. Behaviour of non-structural elements in the 2003 Bam, Iran

earthquake. Earthquake Spectra, Earthquake Engineering Research Institute,

21, S439–S453.

KLINKENBERG, A., JAGER B & SAAL H 1998. Untersuchungen Zur statich

optimalen Halterposition bei punktgestutzten Glastafeln. Stahlbau 67,

Germany (in German), 4, 275-280.

KNAACK, U., KLEIN, T., BILOW, M. & AUER, T. 2007. Facades: Principles of

construction.

LAWRENCE BERKELEY NATIONAL LABORATORY 2006. High performance

commercial building facades. Lawrence Berkely National Laboratory,

Berkely, California.

LUMANTARNA, E., LAM, N. T. K., KAFLE, B. & WILSON, J. L. 2008.

Displacement Controlled Behaviour of Asymmetrical Buildings. Procs. of the

Australian Earthquake Engineering Society Annual Conference, Ballarat,

Victoria, November.

MANIATIS, I. 2006. Numerical and Experimental Investigations on the Stress

Distribution of Bolted Glass Connections under In-Plane Loads (PhD Thesis).

Technische Universität München (TUM).

MCBEAN, P. C. 2008. Drift intolerant facade systems and flexible shear walls: Do

we have a problem? Australian Journal of Structural Engineering, 8, 77-84.

MEMARI, A. M., BEHR, R. A. & KREMER, P. A. 2003. Seismic behaviour of

curtain walls containing insulating glass units. Journal of Architectural

Engineering, 9, 70-85.

MEMARI, A. M., BEHR, R. A. & KREMER, P. A. 2004. Dynamic racking

crescendo tests on architectural glass fitted with anchored pet film. Journal of

Architectural Engineering, 10, 5-14.

MEMARI, A. M., SHIRAZI, A. & KREMER, P. A. 2007. Static finite element

analysis of architectural glass curtain walls under in-plane loads and

corresponding full-scale test. Structural Engineering and Mechanics, 25, 365

382.

216

Chapter 0

MOCIBOB, D. 2008. Glass panel under shear loading - Use of glass envelopes in

building stabilization. PhD Thesis submitted to Steel structures laboratory,

The Swiss Federal Institute of Technology in Lausanne (EPFL).

MOCIBOB, D. & BELIS, J. 2010. Coupled experimental and numerical investigation

of structural glass panels with small slenderness subjected to loacally

introduced axial compression. Eng. Struct., 4, 753-761.

NORVILLE, H. S. 1999. Strength factor for laminated glass. Glass processing days

1999. Tampere, Finland.

NOURISHINGOBSCURITY. 2011. http://nourishingobscurity.com [Online].

[Accessed 1st of February 2010].

NZS 1170.5 2004. Structural design actions, Part 5: Earthquake actions. New Zealand

Standard.

PAULAY, T. & PRIESTLEY, M. J. N. 1992. Seismic Design of Reinforced Concrete

and Masonry Building, John Wiley and Sons, New York.

PCI 1989. Architectural Precast Concrete, 2nd Edition, First Printing. PCI

Architectural Precast Concrete Manual Committee, 175 W. Jackson Blvd.,

Chicago, IL, 340 pp.

PCI 2011. Behaviour of Architectural Precast Panels in Response to Drift, Designer's

notebook. Precast/Prestressed Concrete Institute, 200, West Adams Street,

Sutie 2100, Chicago.

REITHERMAN, R. & SABOL, T. 1995. Non-structural damage in Northridge

earthquake of January 17, 1994, Reconnaissance Report. Supplement C to

Earthquake Spectra , Earthquake Engineering Research Institute, 11.

RYAN, P., OTLET M & OGDEN R G 1997. Steel supported glazing systems. UK:

Steel Construction Institute.

SAFLEX SOLUTIA ARCHITECTURAL GLAZING 2007. Glazing systems:

Performance under seismic conditions, Brochure.

SAKAMOTO, I. 1978. Seismic performance of non-structural and secondary

structural elements. University of California, Berkely, California, Rep. No.

EERC-78/10.

SAKAMOTO, I., ITOH, H. & OHASHI, Y. 1984. Proposals for aseismic design

method on non-structural elements. Proceeding of 8th World Conference on

Earthquake Engineering. San Francisco.

SHAHRAM, T. & MIRANDA, E. 2003. Response assessment of non-structural

building elements. PEER Report 2003/05.

SHAND, E. B. 1959. Breaking Stress of Glass Determined from Dimensions of

Fracture Mirrors. Journal of the American Ceramic Society, 42, 474–477.

SHINKAI, N. 1994. The Nature and Fractography of Flat Glass. Fractography of

Glass. Edited by R. Bradt and R. Tressler. Plenum Press, New York, 253–297.

SHIRAZI, A. 2005. Development of a Seismic Vulnerability Evaluation Procedure for

Architerchtural Glass Curtain Wall. Doctor of Philosophy, The University of

Pennsylvania.

217

Chapter 0

SIVANERUPAN, S., WILSON J L & GAD E F 2011. Structural analysis and design

of glazed curtain wall systems. Australian Journal of Structural Engineering

(AJSE), 12, 57-67.

SU, R. K. L., LAM, N. T. K. & TSANG, H. H. 2008. Seismic drift demand and

capacity of non-seismically designed concrete building in Hong Kong.

Electronic Journal of Structural Engineering.

SUCUOGLU, H. & VALLABHAN, C. V. G. 1997. Behaviour of window glass

panels during earthquakes Engineering Structures, 19, 685-694.

SWADDIWUDHIPONG, S., LEE, S. L. & ZHOU, Q. 2002. Effect of axial

deformation on vibration of tall buildings. The Structural Design of Tall

Buildings, 11, 309-328.

THURSTON S J & KING A B. Two directional cyclic racking of corner curtain wall

glazing Building Research Association of New Zealand (BRANZ) 1992.

VALLABHAN, C. V. 1994. Window glass damage during the January 1994 Los

Angeles Earthquake. Texas Tech University.

VIRIDIAN. 2010. Viridian new world glass, Viridian glass processing [Online].

[Accessed 6th May 2010].

WALL-KING. 2011. http://www.wall-king.com/products/unitized-curtain-wall-49

catalog-1.html [Online]. China. [Accessed 27th January 2011].

WILSON, J. L. & LAM, N. T. K. 2003. A recommended earthquake response

spectrum model for Australia. Australian Journal of Structural Engineering, 5.

WILSON, J. L. & LAM, N. T. K. 2005. Earthquake design of buildings in Australia

using velocity and displacement principles. Australian Journal of Structural

Engineering, 6, 103-118.

ZHOU, Y. S. 2002. An Introduction to Design of Curtain Walls, Aluminium Windows,

Glass Walls, Skylights and Canopies, Wilson Curtain Wall Consultant (HK)

Limited.

218

Documents

In-plane seismic performance of glass facade systems · 2017-02-22 · IN-PLANE SEISMIC PERFORMANCE OF GLASS FAÇADE SYSTEMS SIVANERUPAN SIVAGNANASUNDRAM (SIVA) A thesis submitted