Embed Size (px)
Citation preview
Loughborough UniversityInstitutional Repository
Improving the new productdevelopment process
This item was submitted to Loughborough University's Institutional Repositoryby the/an author.
Additional Information:
• A Doctoral Thesis. Submitted in partial ful�llment of the requirementsfor the award of Doctor of Philosophy of Loughborough University.
Metadata Record: https://dspace.lboro.ac.uk/2134/6941
Publisher: c© D.J. Stockton
Please cite the published version.
This item is held in Loughborough University’s Institutional Repository (https://dspace.lboro.ac.uk/) and was harvested from the British Library’s EThOS service (http://www.ethos.bl.uk/). It is made available under the
following Creative Commons Licence conditions.
For the full text of this licence, please go to: http://creativecommons.org/licenses/by-nc-nd/2.5/
IMPROVING THE NEW PRODUCT
i
DEVELOPMENT PROCESS
by
D. J. STOCKTON
(B. Sc., M. Sc., M. I. M., C. Eng. )
A Doctoral Thesis submitted in partial fulfilment of the requirements for the award of
Doctor of Philosophy of the Loughborough University of Technology ( February 1983 )
O by D. J. Stockton (1983)
1
Acknowledgements
I would particularly like to thank Mr. J. Middle for the
help, advice and encouragement given throughout the period of the
research York.
In addition my thanks are due to the Design and Production
Engineering staff at Herbert Morris Ltd. for their valuable
assistance, to the Science Research Council for financial
support and to Mrs. J. Morley for the efficient and speedy
way in which this thesis was typed.
Finally I would like to thank Herbert Morris Ltd. for
allowing me to carry out the research at their company and for
the financial assistance given.
11
SUMMARY
A system has been developed and is being used at H. M. -Ltd. for
estimating the labour and overhead costs of components manufactured
by a wide variety of production processes. The system uses multiple
linear regression analysis to develop estimating equations that
quantitatively measure the relationship between the production time
of a component and the factors that influence this time. Production
times can then be converted to cost using appropriate labour and
overhead cost rates.
The system uses design features only for predictor variables in
the estimating equations. Hence designers with little concept of
manufacturing methods can use the system to cost designs as they
evolve. This feature therefore provides designers with a powerful
cost optimization tool.
The manufacturing time data used to develop estimating equations
represents current operating conditions at Herbert Morris Ltd. Hence
the estimated times can be used directly as standard times for the
planning and control of manufacturing. In this way manufacturing costs
will be directly linked to the design features of a product.
Software has been developed to allow a computer to retrieve
appropriate equations and compute the production times and costs of
components. This software could form the basis for a larger system
that also generates producibility data for designers.
111
A method of allowing designers to estimate the development times
for individual components and assemblies has been developed. This
facility enables the design process to be scheduled such that the
overall new product development time could be minimized. An important
element of this scheduling method is the ability to allocate resources
between components to be designed on the basis of relative cost and
importance to. the overall success of the project.
Models have been developed that allow :
(i) The total cost of a product to be estimated within a
reasonable degree of accuracy from a knowledge of the
product's ten highest cost components.
(ii) The various types of cost factors involved in alternative
design solutions to be established, hence enabling minimum
cost design solutions to be identified.
Finally research has been carried out into:
(i) The factors to consider when developing a technique for 0
forecasting the need for product development.
(ii) The factors to consider when developing a prescriptive
guide frr ensuring that a product when launched into the
market is a success.
iv
(iii) Alternative methods of solving the "assortment" problem
(i. e. determining the optimum number and specification of
standard sizes to manufacture).
(iv) The integration of Value Engineering into the new product
development process.
(v) The use of cost/design specification relationships in the
generation of minimum cost designs.
The contribution of existing Computer-Aided Draughting and
Design systems to the new product development and original
design processes.
V
CONTENTS
Page
Acknowledgements
Summary i
List of Contents
List of Figures
List of Tables
Nomenclature included in Text
Abbreviations
Introduction
Chapter 1 Product Planning
1.0 Introduction
ý ýý
f
1.1 The New Product Development Process
1.1.1 Product Specification
1.1.2 Product Development
1.2 NPD Success/Failure Criteria
1.3 Forecasting for NPD
f Chapter 2 Aspects of the Design Process
2.0 Introduction
2.1 The Make or-Buy Decision
2.2 Value Engineering
ý °4 "ýý
0-
1
11
V
xi
xiv
xvii
xix
1
8
9
16
19
24
25
33
58
59
64
76
vi
2.3 The Assortment Problem
2.3.1 Solution Techniques
2.3.2 The Minaddition Technique
2.4 Design Planning and Control
2.5 Design Scheduling
2.6 Alterations to Product Designs
Page
81
86
91
99
109
110
Chapter 3 Costing Product Designs 115
f 3.0 Introduction 116
3.1 Total Manufacturing Costs 116
3.1.1 Material Costs 117
3.1.2 Scrap Costs 119
3.1.3 Overhead Costs 125
3.1.4 Direct Labour Costs 128
3.2 Analysis of Cost Estimating at H. M. Ltd. 132
3.2.1 Function 133
3.2.2 Inputs and Outputs 136
3.2.3 Human Agents 138
3.2.4 Sequence 140
3.2.5 Environment and Physical Catalysts 142
3.3 Work Measurement Techniques 144
3.3.1 Time Study 154
3.3.2 Pre-determined Motion Time Standards 155
3.3.2.1 MTM 156
3.3.3 Category Estimating 161
3.3.4 Synthetic Time Standards 166
vii
Page
3.3.5 Work Measurement using Computers 167
3.3.6 Work Measurement Algorithms 172
3.3.6.1 Synthetic Techniques 173
3.3.6.2 Empirically Derived Equations 180
3.3.6.3 Operational Research 186
3.3.6.4 Multiple Regression 188
3.4 Selection of Work Measurement Method for Design Costing at H. M. Ltd. 198,
Chapter 4 Cost Estimating System Development 202
4.0 Introduction 203
4.1 Classification of Estimating Models 204
4.1.1 Manufacturing Processes and Operations 206
4.1.2 Size and Power of Machines 207
4.1.3 Manual Tasks 210
4.1.4 Component Tolerances and Surface Finish 213
4.1.5 Material Processed 213
4.2 Data Collection 214
4.2.1 Selection of Predictor Variables 217
4.2.2 Data Source 219
4.3 The Multiple Regression Procedure 226
4.4 Equation Validation 231
4.4.1 Accuracy of the Existing Estimating System at H. M. Ltd. 240
4.5 Comparison with Time Study Data 240
4.6 Comparison with Existing Time Standards (STT's) 257
4.7 System Computerisation 257
viii
I
Page
4.7.1 System Size ' 267
4.7.2 Variation of Machine Types 269
4.7.3 The Effective use of Computer Time 270
4.7.4 System Expansion 272
Chapter 5 Design Cost Forecasting 273
5.0 Introduction 274
5.1 Cost Relationships 275 -
5.2 Costing using the Pareto Distribution 285
Chapter 6 The Development of ä Design Scheduling Method 298
6.0 Introduction 299
6.1 Specifying the Objective Criterion 299
6.2 Sequencing Component Design 300
6.2.1 Estimating Component Development Times 301
6.2.2 Interactions between Development Tasks 307
6.2.3 Types of Activity Interactions 308
6.2.4 Accuracy of Estimated Development Times 313
6.3 Determining the Relative Importance of Components 317
6.4 Preventing Design Changes 320
Chapter 7 Computer Aided NPD 321
7.0 Introduction 322
7.1 Function Analysis 327
7.2 Data Acquisition and Transfer 329
ix
7.3 Design Creation and Improvement
7.4 Cost, Marketing and Manufacturing Analysis
7.5 Draughting
7.6 Integration with Overall NPD Process
Of
Page
331
335
337
338
Chapter 8 Discussion 341
8.0 Introduction 342
8.1 Comparison of Work Measurement Techniques 351
8.2 Cost Estimating System Development 358
8.3 The Make or Buy Decision 373
8.4 Cost Forecasting 376
8.5 Design Scheduling 380
8.6 CAD 385
Chapter 9 Conclusions
Chapter 10 Recommendations for Further Work
References
Bibliography
Appendix I Development of the New EHB Range
Appendix II Estimating System: Manual Method Description
Appendix III Design Cost Estimating Program (DESCOSTIMATE): Operation Notes and Program Listing
395
400
404
424
429
441
453
A. 3.0 Introduction 454
A. 3.1 Running Procedure 456
A. 3.2 Program Listing 469
x
Appendix IV Descriptions of Variables
A. 4.0 Introduction
A. 4.1 Variable Description Notes for Individual Cost
Centres
A. 4.2 Common Variables
A. 4.3 Variables Applicable to Single Cost Centres
A. 4.4, Illustrations of Welding Positions
Page
482
483
484
506
509
510
Appendix V Data File for Design Cost Estimating Program 512
Appendix VI Modifying the Regression Procedure 521
X1
LIST OF FIGURES
1- Strategic Planning Framework
Page
13
2 NPD Decision Flow Chart 17
3 The Product Life Cycle and Dynamic Competitive Strategy 34
4 Profitability Gap during Product Replacement 36
5 Number of Attributes Examined per Buyer during the Buying Decision 55
6 Optimum Number of Standard Hoist Sizes to Develop 97
7 70%, 80% and 90% Learning. Curves 122
8 Comparison of Scrap Levels Estimated from 90%, 80% and 70% Learning Curves 124
9 Relative Accuracy against Application Time for the MTM Systems 158
10 Manufacturing Time Bands 162
11 Outline Procedure for Applying the Category Estimating
Technique 165
12 Machining Time Parameter Relationships 171
13 Method of Least Squares 190
14 Comparison of Average and Regression Lines 190
15 Effect of Increasing the Number of Regression Models for a Specific Data Range 194
16 Comparison of*R with Average Proportional Error 234
17 Comparison of F with Average Proportional Error 234
18 Comparison of Coefficient of Variability with Average
Proportional Error 235
xii
Page
19 A Comparison of Actual and Estimated Labour Costs at H. M. Ltd. 241
20 The Estimation of Negative Machining Times using Regression Models 244
21 Frequency ency Distribution of 'E' Values 248
22 Frequency Distribution of 'S' Values 248
23 Actual versus Estimated Costs - Milling Fixtures 253
24 Actual versus Estimated Costs - Drill Jigs (Box Type) 253
25 Actual versus Estimated Costs - Plate Gauges 254
26 Actual versus Estimated Costs - Plug Gauges 254
27 Actual versus Estimated Costs - Copy Templates 255
28 Actual versus Estimated Costs - Drill Jigs (Normal) 255
29 STT versus Estimated Times - Bar Lathes 258
30 STT versus Estimated Times - Chuck Lathes 258
31 STT versus Estimated Times - Bar Capstans 259
32 STT versus Estimated Times - Chucking Capstans 259
33 STT versus Estimated Times - Centre Lathes 260
34 STT versus Estimated Times - Horizontal Borers 260
35 STT versus Estimated Times - Vertical Borers 261
36 STT versus Estimated Times - Copy Lathes 261
37 STT versus Estimated Times - Radial Drills 262
38 STT versus Estimated Times - Cylindrical Grinding 262
39 STT versus Estimated Times - Milling Machines 263
40 STT versus Estimated Times - Saws 263
41 STT versus Estimated Times - Surface Grinders 264
42 STT versus Estimated Times - Gear Cutters 264
Xlii
Page
43 STT versus Estimated Times - NC Chuck Lathe 265
44 STT versus Estimated Times - NC Bar Lathe 265
45 STT versus Estimated Times - Spline Cutters 266
46 STT versus Estimated Times - Shears 266
47 Descostimate Standard Input Data Form 271
48 Assembly Cost Inter-relationships for the New EHB 281
49 Total Costs versus Ld: D Ratio for the 3.2 Tonne EHB 284
50 The Pareto Curve of Cumulative Number of Stock Items
versus Cumulative Stock Cost 286
51 Log-Log Graph of the Pareto Distribution 290
52 Log-Log Graph of Cumulative Costs for H. M. Ltd. 's Products 291
53 Actual versus Estimated Product Costs'(-<£300) 297
54 Actual versus Estimated Product Costs (>E300) 297
55 Cumulative Average Task Times versus Cumulative Number
56
57
58
59
of NPD Tasks 305
Component NPD Network Diagram 309
Determination of Development Times = PARTS A and B'*, 315
Determination of Development Times - PARTS C and D 316
Actual versus Estimated Development Times 318
60 Pareto Curve of the 0.8 Tonne EHB Component Development Times
61 Integrated Information System
318
340
62 Schematic Diagram of the New EHB Concept 438
63 Load Spectrum of the New EHB Range 439
64 First Idler MT Spurwheel - 443
65 Flow Chart for Descostimate 457
66 The Modified Regression Objective 523
xiv
LIST OF TABLES
Page
1 Product Planning Strategies 11
2 The-Contribution of Product Planning to Marketing and Financial Based Strategies 12
3 Product Development Matrix for H. M. Ltd. . 22
4 The Design Functions Influence on New Product Success/ Failure Criteria 30
5 Variability in Attribute Ranking between Product Types 43
6 Supplier Attributes Affecting the Purchasing Decision 47
7 Relative Importance of New EHB Attributes 51
8. Stages in the Design Process 61
9 Make or Buy Decision Criteria 65
10 Types of Batches Present in a Manufacturing Organisation 70
11 Manufacturing Costs of Alternative Rope Guide Assembly
Designs (3.2 Tonne New EHB) 73
12 Cost Reduction Methods 79
13 Factors Affecting the Assortment Problem 82
14 Data Required for the Minaddition Technique 92
15 Optimum Product Range Data 94
16 Methods of Improving Product Profitability 105
17 Major Causes of Design Changes 112
18 Factors Essential to Work Place Standardization 145
19 Assessment of a Work Measurement Department Check List
and Score Sheet 149
20 Operator Performance Factors 151
xv
Page
21 Accuracy and Application Time for the MTM Systems 157
22 Manufacturing Time for the New EHB Range 160
23 50 Interval Estimating Chart 163
24 Criteria for Establishing Nd 176
25 A Comparison of Estimated and Observed Machine Times 178
26 Equations for Estimating Metal Removal Times 181
27 Speeds and Feeds for Drilling Operations 184
28 Classification of Costing Equations 208
29 Manipulation Task Times for a Devlieg Spiramatic
. Jigmil 212
30 Material Factor (Machinability) Groups 215
31 Time Standard Estimation for NC Flame Cutting'at H. M. Ltd. 221
32 Design Factors Affecting Welding Duty Cycles 224
33 SPSS Output File . 228
34 Examples of Altering Input Variables to Improve
Estimating Accuracy 243
35 Minimum Operation Times 245
36 Causes of Variability in Time Study Data 246
37 Component Classification for Determining Assembly Times 250
38 Cost/Hoist Capacity Relationships 276
39 Cost/Hoist Capacity Relationships Developed using the
0.8 Tonne and 3.2 Tonne Units 278
40 Total Costs at Various Drum Tube Lengths 283
41 Cost Estimates using Equation 47 295
42 NPD Activity Times 302
xvi
43 Factors Affecting Project Activity Times
44 Objectives of using CAD
45 Proportion of Time Spent on Design Tasks
46 Process Cost Centres
47 Rules for Determining General Machining Cost Centre
48 Variable Description Notes for Cost Centre 202
49 Predictor Variable Coefficients for Cost Centre 202
50 Labour and Overhead Cost Rates (f's/Hr)
51 Data File for Determining General Machining Input Codes
52 Data File for Predictor Variable Coefficients
53 Data File for Labour and Overhead Cost Rates
54 Comparison of the Accuracy of Conventional and Modified Regression Procedures
Page
306
324
327
444
445
447
448
451
513
513
520
527
XV11
Nomenclature Included in Text
AI = Association Index
B2. = Machine Tooling Capacity
C= Cumulative Product Costs (i's)
Ca = Additional Costs incurred from adding one extra Standard Size (£'s)
Ce = Prime Costs (£'s)
Cm = Total Material Costs per Unit (£'s)
Cp = Total Variable In-House Labour Costs per Unit (£'s)
Csv = Cost Saving obtained from adding one extra Standard Size (£`s)
D= Rope Drum Tube Outside Diameter (mm)
Dm = Mean Diameter (inches)
e= Average Error Ratio
E= Average Proportional Error
f, g = Constants the values of which depends on the form
of the Pareto distribution
Ia = Ideal point on the ath attribute
Ki = Machining Factor
Lt = Total Machined Length (inches)
Ld = Rope Drum Tube Length (mm)
Lm = Labour Cost Rate at the"mth. Manufacturing Cost Centre
(f's/Hr)
Nac = Cumulative Number of Adopters in Time c
Nd = Component Complexity Factor
Nm = Number of Machines Controlled by Operator
Qb = Purchased Batch Size
xviii
Qp
r
Manufacturing Batch Size
Recovery Period for Capital Outlay (Years)
R
S
Correlation Coefficient
= Standard Deviation of the Error Proportions
Sa = Percentage average scrap per batch for a specific "learning situation"
Sp = Percentage Scrap
Sr2 = Standard Error Ratio
St = Machine Setting Time (mins)
Tm = Production Time at the mth Manufacturing Cost Centre
w= The Number of Standard Frame Sizes in the Interval
Xi
Xö
xT, x2... xn
y
yi
y
Corresponding estimated value of the dependent
variable
A Constant equivalent to the intercept on the y-axi. s
of a straight line graph
Coefficients of the independent variables zl, z2... zn
Mean Value of the yi's
Actual Value of the dependent variable
The Value of the dependent variable
Time at which 50% market potential is achieved
a= Standard Deviation of time
xix
ABBREVIATIONS
BM's - Basic Minutes
CAD - Computer-Aided Design
CPM - Critical Path Method
CRT - Compensating Rest Allowance
DCL -" Davy Computing Ltd.
DIAD - Distributed Interactive Automated Draughting
DP - Dynamic Programming
EHB - Electric Hoist Block
EOQ - Economic Order Quantity
EUCLID - Easily Used Computer Language for Illustrations
and Drawings
FEM - Federation Europeene de la Manutention
FTF - Floor-to-Floor Time
H. M. Ltd. - Herbert Morris Ltd.
LOCAM - Logan Computer Aided Manufacture
LP - Linear Programming
MBO - Management by Objectives
MLR - Multiple Linear Regression
MOST - Maynard Operation Sequence Technique
MRP - Materials Requirement Planning
MTM - Methods Time Measurement
NC - Numerically Controlled
NPD - New Product Development
OEM - Original Equipment Manufacturers
PBC - Period Batch Control
xx
PERT - Performance Evaluation and Review Technique
PLC - Product Life Cycle
PMTS - Pre-determined Motion Time Standards
PPBS - Planning, Programming, Budgeting and Systems Analysis
p/t - Prototype
REAP - Rapid Estimating and Planning
SAPPHO '- Scientific Activity Predictor from Patterns with Heuristic Origins,
SPSS - Statistical Packages for the Social Sciences
STT - Sales Target Time
VE - Value Engineering
1
INTRODUCTION
Since the company's foundation in 1884 to market pull' blocks(t),
Herbert Morris Ltd. (H. M. Ltd. ) has become a leading UK manufacturer of
a wide range of mechanical handling equipment including Overhead Cranes,
Goliath Cranes, Container-Handling Cranes and Electric Hoist Blocks.
In 1978 the company became part of Davy Corporation Limited the
international engineering and construction group.
In recent years the competitive action of other European crane
manufacturers has led to many of the company's products becoming
uncompetitive in the market place. Although the product designs of
H. M. Ltd. are still technically acceptable they are labour intensive
in manufacture and are therefore more expensive than many of the leading
competitors products. In addition modular design of competitive products
enables both manufacturers and their distributors to achieve larger
reductions in stock holding costs compared to the current range of the
products of H. M. Ltd.
Since the products of H. M. Ltd. are financially uncompetitive the
present recession in world markets has resulted in a considerable decline
in sales levels and hence profitability. In consequence the company's
workforce has been reduced and product lines sold off.
In order to recover from this situation H. M. Ltd. are currently
undertaking an extensive programme of development of several major
product lines, which includes the wire-rope and chain hoist product
ranges.
2
The need to simultaneously develop several product lines places
a heavy burden on the development resources of H. M. Ltd. and leaves
little scope for further product development should the replacement
products fail to achieve their expected profit objectives.
The research reported here arises from the development of an
improved range of electric wire-rope hoists (New EHB) which is necessary
due to the continual decline in sales of the existing Ropemaster range
(Appendix I). The EHB is also a major sub-assembly in H. M. Ltd. 's range
of Universal Cranes. The constraints surrounding the New EHB development
project are:
(i) Limited development time available since an improved
product range needs to be launched quickly to prevent
further serious decline in the market presence of the
company and to restore profitability levels.
(ii) Limited resources available for research and development
since several development projects (electric wire-rope
hoist, electric chain hoist and universal crane range)
are currently being undertaken in parallel.
(iii) Limited investment resources arising from the company's
current reduced profitability levels.
(iv) An inherent manufacturing cost disadvantage since the
lack of capital investment funds-limits the annual
quantity that can be manufactured. Competitors therefore
have the advantage of cost reductions that arise through
the manufacture of larger volumes.
3
(v) The need to develop a product design that increases the
company's in-house manufacturing work load and hence
reduces the manufacturing overhead burden on other
products manufactured by the company.
The present work attempts to reduce the effect of these constraints
on the development of the New EHB range with the further objective of
providing a contribution to the improvement of the new product
development (NPD) process in general. Initially Chapter 1 examines
the structure of the NPD process and its contribution to the strategic
planning process. Product development is an essential element of
product planning, a strategic planning process appropriate to the low
volume, multi-product manufacturing environment of H. M. Ltd.
Chapter 1 also reviews the literature relating to the factors
that determine the eventual success or failure of new product developments
with the objective of developing a prescriptive guide for the New EHB
and future projects, i. e. the product must enable the company to regain
an adequate market presence and prevent the complete erosion of its
hoist block market by competitors.
Chapter 1 finally reviews existing sales forecasting techniques
in order to develop a technique for forecasting the date at which to
begin the NPD process with the objective of allowing sufficient time
for the development process to be carried out, and hence minimize the
loss of revenue that normally accompanies product change overs.
4
Various aspects of the development process associated with the
overall New'EHB development constraints have been examined in Chapter 2:
(i) A model has been developed and demonstrated for comparing
the costs of alternative design solutions and allowing
the minimum cost solution to be identified. The model 1.0
considers only those costs affected by the choice
and enables the effect of variations in purchasing and
manufacturing batch sizes and recovery period for capital
investment costs to be examined.
(ii) The integration of Value Engineering work into the NPD
process is considered and a method of improving the rate
at which cost reduction tasks can be carried out is
proposed.
(iii) Alternative methods of solving the "assortment" problem
(i. e. determining the optimum number and specification of
standard sizes to manufacture) are examined and the
Minaddition technique identified as appropriate. This
technique is then used to determine the optimum number and
the capacities of New EHB standard units.
(iv) Existing methods of planning and controlling projects are
examined in relation to their suitability for scheduling
the development process of a new product range such that
the overall time required to develop a product is minimized.
5
Chapter 2 concludes with an examination of the causes of design
changes and estimates the additional design resources necessary to
carry them out.
Chapters 3 and 4 exämine: the existing cost estimating system at
H. M. Ltd. and a system is developed that can be used by designers. The
major objective of the present research is to develop a responsive
estimating system that allows cost data to be generated quickly
therefore increasing the use of cost data in the planning and control
of the NPD process and the identification and generation of minimum
cost design solutions. IL
When examined from a systems viewpoint the faults with existing
estimating systems are clearly identified. The prime causes for
estimating-systems being unresponsive are the work measurement techniques
used to estimate component manufacturing times since they require
personnel with detailed knowledge of the production processes and
methods used within a company in order to develop accurate cost
estimates. Hence the design function must request cost data from a
separate estimating or manufacture planning department. Since these
departments often have other responsibilities, requests for cost data
must often queue until service time becomes available and so delaying
the provision of cost data to the design function.
Allowing designers to cost estimate designs would undoubtedly
improve the responsiveness of an estimating system. However, a method
of estimating component manufacturing times needs to be developed that
6
does not require users to possess an in depth knowledge of manufacturing
processes. In this respect the technique of Multiple Regression Analysis
has been found useful in developing cost estimating equations by allowing
the causal relationship between manufacturing times and suitable predictor
variables to be established.
The development of a system for estimating component manufacturing
times which does not rely on a knowledge of machine process variables
(this type of data is not normally available, to designers) required the
generation of a large number of estimating equations. This is because
of the wide variety of manufacturing processes used at H. M. Ltd. Hence
a method of classifying estimating models has been developed to allow
appropriate equations to be quickly retrieved.
The estimating system developed allows cost data to be generated
if general arrangement drawings for designs are available. However there
are instances when it is an advantage-to know total product costs without
the necessity of preparing general arrangement drawings, e. g. choosing
between alternative product design concepts. Also it may not be possible
to prepare detailed drawings'for determining the optimum number and
specification of standard units, therefore, Chapter 5 examines the
development and use of cost/design specification models for estimating
total product costs and for generating minimum cost designs. In
addition since cost/design specification models have limitations
a model has_been developed that allows total product costs to be
alternatively estimated from a knowledge of a product's ten highest
cost components.
7
The examination of existing project planning techniques (e. g. CPM
and PERT) indicated that they were unsuitable for scheduling the NPD
process. Chapter 6 therefore describes the development of a method for
estimating the development time for individual components and identifying
the type of task interactions that exist.
The designer can use the method developed to quickly identify those
components that would lie on the critical path as design proceeds and
hence constantly reschedule the NPD process to avoid delay.
An important element in a scheduling system when limited development
time is available is the effective allocation of this time. It is
proposed therefore (Chapter 6) that both development time and resources
are allocated between components to be designed on the basis of relative
cost and importance to the overall success of the project.
Since H. M. Ltd. are currently installing a Computer Aided
Draughting system, and are planning to extend this into a design and
manufacturing aid, the opportunity was taken of assessing the contribution
of CAD systems to a NPD project in. which original design worksformed an
important element of the total design activities. The individual tasks
involved in the original or creative design process are therefore examined
(Chapter 7) and the present contribution of Computer-Aided Draughting
and Design systems assessed. In addition the inclusion into a CAD system
of the cost estimating system developed in the present work is examined.
8
CHAPTER 1
PRODUCT PLANNING
9
1.0 Introduction
Of the established methods of strategic planning (2'3) the
product planning technique (4, S) seeks to maintain an acceptable
profitability level by directing long term plans towards developing
and maintaining an effective product mix for a manufacturing company.
The technique therefore involves selecting an appropriate long term
planning strategy for each product type manufactured within an
organisation..
Ansoff(3) examining the manner in which strategic planning had
developed since 1950 established that the evolution of competitive
strategies and corresponding management responses had not been one
of replacement but one of proliferation, which in conjunction with
the decreasing predictability of the future had now resulted in
management being confronted with strategic planning problems of
complex structure and containing many inter-related factors.
Product planning techniques help to reduce the complexity of
the strategic planning process (6) by isolating the influence of such
inter-related factors as:
(i) Manufacturing resources, i. e. the compatibility of
product design with manufacturing facilities and labour
skills can be established.
(ii) Changing market requirements.
(iii) Competitive action.
10
(iv) Product profitability.
(v) Product obsolescence.
(vi) Product worth, i. e. the total benefit of a product to
a company. For example, does the product allow total
product range to be increased or improved. . -P
(vii) Investment risk.
(viii) Market environmental changes, e. g. inflation, recession,
government legislation and resource scarcity.
(ix) Product interactions,. i. e. a decline in sales of one
product type effectively increases the required overhead
recovery rate for all products in order to retrieve
consistent overhead costs.
(x) Price/quality relationships.
The need to adopt product planning strategies is dependent on the
number and variety of product types manufactured within a company, since
these two factors influence the type and level of inter-relationships
that exist between strategic planning variables and hence affect the
complexity of the long term planning process. Accordingly product
planning strategies would be particularly applicable to H. M. Ltd.
since the company's product range is diverse, i. e. includes heavy
custom designed cranes, standard crane ranges, electric wire-rope
hoists, electric chain hoists, manual lifting equipment and hydraulic
lifting equipment.
11
The basic types of product plans identified by Lawrence (7) are
listed in Table 1.
TABLE 1 PRODUCT PLANNING STRATEGIES
(i) Add new products.
(ii) Delete obsolete products.
(iii) Modify existing products.
Altering the volume ratio of product types within the overall
product mix could be considered an alternative product planning strategy.
However this option is usually restricted to enabling organisations
to achieve such short term objectives as the maximization of annual
profit levels.
Alternative strategic planning methods (i. e. market and financially
orientated actions) are considered by Kotler(8) to be equally effective
as product planning techniques in achieving an organisation's long term"
objectives. However the author of the work reported here considers
that these alternative strategies can benefit from an element of
product planning as illustrated in Table 2.
12
TABLE 2 THE CONTRIBUTION OF PRODUCT PLANNING TO MARKETING AND
FINANCIAL BASED STRATEGIES
Strategy Product Planning Contribution
1. Find new users of the product ) Improve product specifications to
increase the number of possible 2. Find more 'uses of the product)
users and/or uses.
3. Provide more effective sales Modify product to improve sales
promotion appeal, e. g. include optional extras,
improve product design style.
4. Provide superior services Improve product quality, reliability
(Sales, Technical, Maintenance) and maintainability to reduce need
for service facilities.
5. Provide price discounts Reduce product manufacturing costs
to offset reduction in product
profitability.
6. Sell to specific customers Modify product design to customer's
particular requirements.
Significant literature on corporate and product planning procedures
is available with useful surveys being published by Gilmore and
Bandenburg(9), Cunningham and Hammouda(l0) and Berry(ll). The
detailed framework for preparing strategic plans provided by Gilmore
and Bandenburg, Figure l,. can be used to illustrate the product planning
process, i. e. the activities constituting the "Formulation of Competitive
Strategy" and "Specification of Programme of Action" need to be under-
taken for each product type manufactured within a company.
13
FIGURE 1 STRATEGIC PLANNING FRAMEWORK(after Gilmore and Bandenburg (9) )
14
I Vf a) 4) N U "r C C ý! f "r N ýr "r
"r V) i 0 -0 G) rt3
4- ' t0 Q1 G1. S.. i].. C
C. U -r - u
N
,Q N
. "ý i tt) CL E 0 U
4)
co 4-3 C eý >
'C7 ý
N >
.ý +3 "r 4-) a) CL E 0
U CL Q:
> . r- 4-3 V R$ -0
1/1, N G) Q) "r
tii 1a V) C 4""3 Vf ý C G) CN >N +ý a) C 0
. r. 4J U
'O 0 i
>c a) a)
> rn ý "r- C) +-) U C) C
Y (0 S. C
(0 "r G. M: LL.
b C eo
O "r ý V
4- O
ý O S_ CD) O S.,
a
N CM .4 td) . 1.0 .9 CO ... CM . d- . Lnýý .. 00 01 . O. r. N. 0, O rN NNNNNNNN rM MMM C'M M C'M MMM d' st "C: r
,a
0 U
r
W
C, '3
U.
Y '-I
. ti
E 0 U
ý ý "r r "r ß "r H ro W
LL.
N aý
.ý 4J . r. c
0 0. CL 0
v C cc >C
0 Ný "r ý "r ýN CU 4-3
r G) i "r .a
G1.4) E oE C. U vº "ý-
.-CCo ooC "ý o +3 iý d-ý U ouu ai (L) t0 (3 +> a. o. 3 oEE a) a º-" º-+ z
N C)
.ý ý >
>, rn
"r 4-3 V (0
S- a) (1) -U c
>2 >., .Yý NNN N (CS N
rn (1) +j i
N
fN aý
"r 4-3
4)
C 0
"r +' > Si Ui +) O.
co 4- 0
++ 4-)
m aý EW co Lc rn 0 ca
O r- -r- 0 rt) r- >> > "r "r {-, 1 (O 4.. ) 0 -r- 4) C (0 r
M C) L1. "r Cý ttS ý G) OU tý
a fl. 4- G) tu O +> >U
C "r V1 rt3 +ý V1
-p i rt3 -
a) týý "r CO "r Lt_ N-4 LL
a) e ý tC L rn 0 5- 0
a E 0 U
ý +ý "r r" . r.. .a "r N to (1)
LL.
c 0
.r N N
"r E
V
E 0
0 U
W
.. t. f) . l0 .9 CO Ct1 . Cý. .. l, C) l0 n. 00 011 p r. r-- N Mý NM rrrrrrrrrNN
i "r- O. 4--
ý . '- r- 4-
0 4-" m i. oýa
0 H NC
'C O H
aj "r-
"r- S- 4- 0
15
Figure 1 is also useful in illustrating important aspects of the
product planning process such as:
(i) The need for co-operation and co-ordination between
functional departments, i. e. both assessing the business
and manufacturing capabilities of a company (activities
13-19) and undertaking product development tasks
(activities 34-39) require a high level of departmental
integration and synergy.
(ii) Individual product plans must be developed and integrated
such that the overall objectives of an organisation, i. e.
the economic mission, can be achieved. Wilson, et. al. (12)
indicate that in addition to considering the total business
environment of an organisation, other characteristics
essential to the development of effective product plans
are:
(a) they must be future orientated,
(b) they must be pro-active, i. e. plans need to be
actioned now with future benefits in mind, and
(c) they must be flexible, 'i. e. easily adapted to such
changing circumstances in the business environment
as unexpected competitive action.
This need for product plans to be future orientated, pro-
active and flexible indicates the importance of forecasting
to the strategic planning process since a knowledge of
future activities is essential to the development of
effective product plans.
16
(iii) The product development process (activities 34-38) for
. modifying existing or developing new products is
essential to the basic product planning strategies
listed in Table 1 since even abandoning an obsolete
product normally necessitates the development of a
replacement product. i
(iv) During the product development process the two basic.
elements of risk involved in the product planning process (13)
i. e. information and performance uncertainty, interact.
Information uncertainty is associated with the uncertainty
surrounding the accuracy of various types of data used
(e. g. sales forecasts and cost estimates) and is often
the prime cause of poor product plans being developed.
Performance risk arises due to the uncertainty surrounding
the ability of personnel involved in the development
process, i. e. the relative success of development activities
is unknown and can be responsible for effective product
plans proving unsuccessful.
1.1 The New Product Development Process
The new product development process (NPD) has been well researched
with Uman(14) providing a detailed account of the stages involved, and
decision flow charts being developed by Capon and Hulbert(15) and
Cardozo and Ross (16). The flow chart developed by Capon and Hulbert,
Figure 2, is used in the present work to illustrate the variability
between manufacturing organisations that exists in the NPD process
and hence the difficulties involved in standardising development
procedures.
17
FIGURE 2 NPD DECISION FLOW CHART
11
MAtttTIML SttvlCn ºttºAttS at TAILED AXIKET
DATA iwtvtT IP MlCt3YtT)
It. POTENTIAL WR[[T ýaouect "vý14U
Im . P" ºOOOUCT CMýdTT[t ! cllv[
t0t T117M1 t IDtA?
sm4e Mf
/YTUOt Yis
1. ºtOFlTA4Lln L CAPITAL
REQUIRED
oon º! OllR
r&roM ºtlSar
CO.. Y1. f.. tJC, U, I?
.
tow. cu surIc" ot. ncº nrt
SCMMIJLJ. CAPITAL ttwttrtwii
O(iY120 COSTS
M[t //Oa, cT cu. +mc
"nunu ºtGRCT,
D1sCUSSIi & wu. n uroaT
I. ºwawr.. u., : ý......
. ea. ýen
Does !I INTiva corN rT! [
LººpvB ºfO1lRt
non aSCUm.! cawrru
tWý ous
.ý Rlsws.
1. ºfORT. ruTT ,. c.. I". º
ý
I DOES I COW AMT
SOARD YºtOr!
ºfOJlCTT
5lGM ' rROOUCTIOM
TRIAL
UND IMITLL
REPORT To eaecunre cONrTTee
(Ref. 's
1. SIZE OF POTENTIAL
IURRIT 2 COMPANY KNOWLIDGI
OF THIS MARKET 3. COM/LG1INTARIT7 TTN
COMPANY STRUCTURE
n IDEA
WORTH PussumC?
ºlO1IR Gl. ººOIMTIO, IYOGIT
AMC ºlOOUCT
ºolrAUI1o,
ºIIOONCTIOq T111A13 COW IKI
ºlO1lCT DISCVSBD IT PIODUCT
OIYIISIPICGTIOM COrwTTII TO
DETERMINE WPLL CATbMS FOR WOOL
CDMPUNT
1W
LS P OOVCT -
TIIT $WJIA! TO
ISIST7MC naoucTSl
18
The flow chart indicates that the NPD process is controlled by
a New Product Committee and a Product Manager. However these formal
management systems are but two of several alternative methods of
controlling the NPD process. Other methods are Research and
Development Departments, Commercial and Technical Project Managers,
Commercial and Technical Line Management and Commercial and Technical
One-Man Efforts. The type of management system appropriate to a NPD
project is according to Souder(17) dependent on:
(i) The nature of the technologies being managed.
(ii) The markets being served.
(iii) The relative cost of the development project.
(iv) The capability of the system to handle innovative
conditions.
At H. M. Ltd. the NPD process is controlled by a Progress Committee
consisting of members of each of the main functional departments involved,
i. e. Design, Commercial, Quality Assurance, Production Engineering and
Manufacturing.
Considering the level of technology involved in develooing the
New EHB range a multi-functional NPD committee would according to
Souder(17) prove effective. However experience gained during the present
work indicates that the effectiveness of using an NPD Committee is limitel
if the frequency with which committees meet and attendance at committee
meetings is adversely affected by assigning a low priority to NPD work.
19
Under these conditions it is difficult to optimize development resources
expended with benefits (optimality of the product design) obtained.
The type and level of activities involved in Product Formalization
are dependent on the type of product being developed, e. g.;..
(i)ce High sales volumes indicate that it is important to
optimize a product design in terms of manufacturing
cost, ease of manufacture and range of market
specifications covered.
(ii) Innovative and high technology products indicate that
extensive resources may require allocating to the Product
Formal. ization stage. In addition a large proportion of
the activities carried out could involve basic research.
(iii) The need to develop a reliable product could influence
the resources allocated to prototype testing activities.
To launch a new product the two basic activities necessary are
product specification and product development. The individual tasks
required to specify and develop a new product are obviously dependent
on the type, of product being developed and would therefore vary
considerably between manufacturing organisations.
1.1.1 Product Specification
The tasks involved in specifying a new product include:
(i) Finding new product ideas.
20
(ii) Selecting suitable new product ideas.
(iii) Obtaining detailed design, cost and market
specifications for the new product.
For each stage of the NPD process Fogg (18) estimated the
average number of new product ideas needed to achieve one success-
ful product -as:
NPD Stage Average Number of New Product Ideas
Idea 40
Technical Search 10
Technical Feasibility 6
Prototype Testing 3
Pilot Production 2.5
Launch 2
Success 1
The high rate of project failures indicates that organisations
should endeavour to maintain an adequate flow of new product
ideas in order to ensure an acceptable number of product
successes. However the available development resources with-
in an organisation effectively limits the number of new -
product ideas that can be considered. Under these circumstances
to ensure successful products are developed, organisations must
adopt alternative approaches to the problem of finding new
products, e. g. H. M. Ltd. formerly reduced the cost of developing
successful products by concentrating on improving existing
21
products and hence reducing the technical, manufacturing and
marketing effort required. According to Cooper (19) by remain-
ing with existing products and markets the probability of
marketing a successful product is improved since the company
can acquire an in-depth knowledge of product technology and
market characteristics. However the concept of market evolution
described by Kotler( 8) indicates that since the development of
a market is accompanied by a marked increase in competitive
action the chances of launching successful-products can be
reduced.
The obvious strategy to adopt therefore is to balance the
two approaches (i. e. modify and improve existing products and
develop, new products), a strategy which H. M. Ltd. is presently
following. Table 3 indicates the present status of development
work at H. M. Ltd. and shows that both improved (New Electric
Chain Hoist, New Electric Wire-Rope Hoist, Centrelift and
Universal Crane) and new products (Hydraulic Crane and Piledriver)
are being actively developed.
Obtaining a detailed design specification for a product is
normally the responsibility of the marketing department since
this function has direct access to the market place and hence
customer requirements. Several techniques are available (20)
to aid the methodical collection of product design data such as
questionnaires and surveys. However the effectiveness of the
data collection stage is dependent on the type of data collected
22
ý i O
4- ý W
m
. r. ý C)
'Y
rO
. r..
t0 C)
U
n
TABLE 3 PRODUCT DEVELOPMENT MATRIX FOR H. M. LTD.
Increasing Technological Effort ý
No Product Product New
Change Improvement Products
011 Universal Crane
Hydraulic Jack Range
+) a) a)rn L (0 Centre
Lift z Crane Range
New Electric
Chain Hoist N
110W x V) +-J W
0Y W i-3 i
M 4-)
LO C
4-- º-+ W
New EHB Hydraulic Range Piledriver
Z Hydraulic N Oä Crane
,- -14 m s. >b
Adapted from Berry (11
23
as well as the method used to collect it, i. e. it is essential
that all functional departments involved in the NPD process can
influence the type of data collected since their information
requirements are often not compatible with those of the marketing
department. Consequently problems can arise during the NPD
process. For example, several problems arising during the
development of the New EHB range were:
(i) The maximum length of hoist block that the market
would accept was not obtained leading to problems
in minimizing drum tube and related component
costs. It was necessary to determine the drum
tube length to outside diameter ratio that minimized
overall design costs.
(ii) The maximum speed of lift acceptable to the market
was not defined. This data would have allowed gear
reduction ratios to be minimized and hence gear box
material costs.
(iii) The relationship between height of lift (and hence
rope length and rope drum size) and sales volumes
was difficult to ascertain from the type of data
gathered causing uncertainty during the determination
of the optimum number of standard drum tube lengths
to adopt for each standard of New EHB unit.
24
I.,
(iv) The market survey was primarily designed to obtain
data for EHB units sold for "runway duty", i. e. , Class I cranes
(21). However the objectives of the
design department required developing hoist blocks
for "crane duty", i. e. Class II cranes. For the
marketing department the variation between hoist
blocks for Class I and Clase II cranes is negligible,
however to the design department the variation
represents significant design differences. Hence
data obtained from the market survey was not always
relevant to the design problem in-hand.
1.1.2 Product Development
Product development is an iterative process(22) consisting
of design, manufacture and test-activities carried out in the
following sequence:
(i) Design
(ii) Prototype build
(iii) Prototype test
(iv) Jig and fixture design
(v) Jig and fixture manufacture
(vi) Pre-production manufacture
(vii) Pre-launch market trials
25
The iterative nature of the NPD process often requires the
entire or parts of the above sequence to be repeated.
Since the risk involved in the product development process
arises from the uncertainty with which development tasks could be
performed and the uncertainty surrounding the accuracy of data
used during the NPD process, the present work considers the
overall risk involved (hence the effectiveness of the product
development process) and relates this to the number and type
of components that form the product. These factors determine
the number of activities, type of activities and input data
involved in the development process, e. g. the cost estimates
required, type of design decisions, manufacture of prototype
parts, reliability tests required and the jigs and fixtures
necessary.
In addition since planning the NPD process requires that
development activities are co-ordinated such that development
costs are minimized whilst at the same time ensuring that
maximum benefits are obtained from the product design, the
complexity of the development process is also influenced by
the number and type of components involved.
p) 1.2 NPD Success/Failure Criteria
Identifying the major factors that differentiate between success-
ful and unsuccessful new industrial products is an initial step to
developing prescriptive guides for the new product development process
26
since many of the variables which might separate the "winners" from
the "losers" are within the control of the firm. Cooper(23) considers
that, additionally, determining the relative importance of these
factors is essential to the effective allocation of development
resources.
Initially investigations such as SAPPHO (Scientific Activity
Predictor from Patterns with Heuristic Origins) (24), and that
carried out by Gerstenfeld (25) centred on identifying the problems
facing product developers by comparing successful and unsuccessful
new products and establishing the differences between them. For
example, the SAPPHO project employed a "paired comparison" technique
in which the criterion for the formulation of a pair was commercial,
i. e. the successful and unsuccessful product competed for the same
market.
The effectiveness of the New EHB project can be-examined by
comparison with the five main areas of difference between success
and failure that emerged from the SAPPHO study:
(i) Successful innovators were observed to have a much better
understanding of user needs. In this respect the detailed
market research study(26) carried out by H. M. Ltd. into
the wire-rope hoist market has provided much useful
information of user needs.
Companies established within a market must continually
monitor user needs since these can alter with changes in
market environment. For example, at H. M. Ltd. the Electric
27
Chain Hoist range has recently been re-designed on a
modular concept to enable distributors to maintain an
adequate product range with low stock holding costs,
i. e. current interest rates on borrowed capital are
high hence stock holding costs have assumed greater
importance.
(ii) Successful innovators pay more attention to marketing and
publicity. The level of marketing and advertising resources
to employ during the NPD process is related to the method
of distribution. At H. M. Ltd. the majority of hoist
blocks are sold through a small number of major distrib-
utors, hence the level of marketing and publicity required
is low.
Publicity is of course generated by the reputation in the
market place of the company developing the new product. In
the case of H. M. Ltd. market reputation is high.
(iii) Successful innovators make more use of outside technology
and scientific advice, not necessarily in general but in
the specific area concerned. In this respect the development
policy of H. M. Ltd. is to make full use of outside institutions
such as Leicester University, Loughborough University and
the National Engineering Laboratory.
28
(iv) The responsible individuals in the successful attempts
were usually more senior and had greater authority than
their counterparts in unsuccessful projects. The
individual responsible for the successful development
of a new product must have the authority to integrate
the various functional departments involved. The
objectives of these individual functional departments
are normally at variance during product development(27)
hence the responsible individual must have the authority
to reconcile those objectives. In recognition of these
facts responsibility for NPD at H. M. Ltd. is that of the
Technical Director, who is a member of the main board of
directors.
(v) Successful innovators perform their development work more
efficiently, but not necessarily more quickly than in the
case of failed initiatives. It is in this area that
determining the relative importance of product development
criteria would allow effective allocation'of development
resources.
A variety of techniques (e. g. rating scales and checklist
models) were initially used to determine the relative
importance of development criteria to the success or failure
of a new product. Wind(28) developed a conceptual test
model to screen new product ideas for market acceptability
by estimating subjectively the relative importance of
development criteria. Cooper (29)
compared the relative
29
importance of factors using an arbitrary weighted average,
i. e. each factor was assigned a score of 1.0 if a primary
reason and 0.5 if a contributory cause of failure. The
weighted rating score was then established for each factor
from 101 examples of new product failures.
In later work Cooper (23) overcame the limitations of
previous work (i. e. basing the relative importance of
development criteria on subjective estimates) and developed
empirically based weightings for success/failure criteria.
The application of Cooper's work to the present New EHB
project is limited since the SAPPHO(24) project observed
that the individual factors found to be significant for
success or failure varied between industries. In addition
Evans (30) observed that the relative importance of such
success/failure criteria as product price, reliability,
quality, service and competitive product prices altered
with changes in the market environment (e. g. from energy
abundance to scarcity).
At H. M. Ltd. therefore it is not possible to develop
empirically based models for predicting the potential
success or failure of a new product since the appropriate
data on past new product developments and corresponding
market conditions does not exist.
30
oe
However Cooper(23) developed a comprehensive list of
success/failure criteria from past research projects
which could be used as a guide for the planning and
control of the present New EHB and future development
projects. The criteria listed by Cooper indicated the
importance of the design function to the success of a
new product since many of the factors involved are the
responsibility of the design department. This is
particularly true at H. M. Ltd. since the design department
has the added responsibility of planning and controlling
the NPD process. Table 4 has been adapted from Cooper (23)
to define more clearly the influence of the design function
on new product success/failure criteria.
TABLE 4 THE DESIGN FUNCTIONS INFLUENCE ON NEW PRODUCT SUCCESS/FAILURE CRITERIA
1.
Adapted from Cooper(23)
Technical and Production Synergy and Proficiency
(i) Ensure compatibility of engineering skills and
production resources with product design.
(ii) Ensure the design department carry out preliminary
technical assessment and product development well.
(iii) Ensure that product design does not prevent
prototype testing and pilot production start-up from
being efficiently carried out.
31
TABLE 4 (Cont'd)
(iv) Ensure design staff fully understand product
technology, product design and manufacturing
technology used within the market and company.
2. Marketing Knowledge and Proficiency
(i) Ensure that preliminary market assessment, market
study, test marketing and market launch phases
provide the design function with relevant data on
which to base product design.
(ii) Ensure design department fully understands customer's
needs and wants, buyer price sensitivity, competitive
situation, buyer behaviour and size of potential market.
3. Newness to the Firm
(i) When potential customers, product use, production
processes required, distribution method and product
technology are new to the firm the design function's
responsibility is to ensure that product design is
compatible with these factors.
(ii) When new competitors enter the market the design
department must have the capability of analysing
competitive product designs. a
32
TABLE 4 (Cont`d)
4. Product Uniqueness and Superiority
(i) The design department must ensure that the product
design, if necessary, has unique features for the
customer, is superior to competing products in meeting
customer's needs, allows customers to reduce costs,
carries out unique tasks for customers and is of a
higher quality than competitors products.
5. Market Competitiveness and Customer Satisfaction
(i) Product design must allow company to sell, if
necessary, in a highly competitive market where
product price is important to the buying decision
and potential customers are presently satisfied with
competitors' products.
6. Marketing and Managerial Synergy
(i) The design department must ensure that product design
is compatible with financial resources available, the
company's managerial skills and aids the sales force
when advertising and marketing the product.
7, Market Dynamism
(i) The design department must be able to handle
satisfactorily situations in which new products are
frequently introduced to the market and users' needs
change rapidly.
33
1.3 Forecasting for NPD
The ability to forecast the need for NPD provides many advantages
including:
(i) The control of R&D costs could be improved since product
development is carried out only when necessary as opposed
to the alternative of maintaining planned product
obsolescence strategies.
(ii) Adequate time can be allowed to carry out the NPD process
hence reducing the effect on development costs of schedule
slippages.
(iii) Adequate time can be allowed to ensure that the new
product design satisfies both market and manufacturing
requirements, i. e. in terms of product cost, specification
and function.
Bentley (31)
and Wasson (32) have demonstrated that effective product
planning strategies can be developed using the sales or profit behaviour
of a product during its market life as a planning tool. Hence the
problem of forecasting for NPD involves predicting the shape of a
products' market life cycle curve (PLC).
Figure 3 illustrates the shape of a typical PLC curve and
indicates:
BEST COPY
AVAILABLE
Variable print quality
34
FIGURE 3 THE PRODUCT LIFE CYCLE AND DYNAMIC COMPETITIVE STRATEGY
(Ref. 32)
> )Dt, ioiS1 R
leaf Tiny product)
N L
CL
A-Low Learning Products
1126,
TIME
ST tATEG7 Injsalss Iesrdng requirements. To establish a strong brand market To maintain and To deferd brand position against cote- To n: t n it. oC., t g DISUECTIVE loco, and r.. nedy offering de- and discrlbution i. h. as quickly as strengttao the mar- Pilling brands and product category a- dry of ail passable
facts puLkly. develop . Idespread possible kilt niche achieved Blast other potential prm. acts. through profit vnr. oeos of benefits. and gals through dealer and eosstant att, taba to prod. wt Isnprrn. - vial by nrfy dopt. rs eooz-, for LLY meet opportunüles and fresh promo-
tional and distribution approaches
OUTLOOK FnIt f: ooe is 13ualy to be attracted is Early . nuance of numerous aggr.. - Price and dlstrlbu- Cowpetitbs stabslLad. with fe. or w Slmatar oon. Peeltioo COMPETTTIOY the arty. sprofitablit stages also emulators tlon squeeses on the n- exrases am mariet shares taut deeis-C nod drops.
industry. shakl, g subject to substautlsl change Is the ab- ping our becaaase or out the . esker en- searo of a substa. tal ;. created tea- decrease La Ce. tra" is protement is some brand eemar arrest
iR(. Dt. CT Llmasd her of models with t! odular design to fac111tat. flaalbl. Ine. naufid . tt-ti.. A co. naus alert for _. artet py_sme. t ug Cur-stet '-tog .1 7 LS; G:: hyslcel , uodact sd offering s. Shclbn of variants to appeal to so- to product improve- opPoetsait1es through tithe, hold cost- line to rti. niart. OIy ECTTVC . slgss tsoth fcoossed on m1o1- cry new segment and new tea. -spa went, tlght., ung up ad price-penvtra! un of ae. wartets say r . us rot en-
miaing learoing requirements. teed as fast as discovered ei line to elimnca:. or major product . ha-ges laraiucttoei tura. ag a dl resw aalgtu tx a_ and us. -englneerd uoneceasary special of Danker products Coootase nUeotlos prof..
it. appeal to east rer. t'tlve set- ties with little mu- to possibilities for poxes tieveveemest
. srL t tr. o« attention to quality kilt appeal iid cost cutting- Re oe. vioatluo of snontroi and gaick . iimination of neceuLry of deaigs co-premises
mart. t-rwreald defects in design
PRI21716 To imp_ the wlnlnmm of value A price lie. for every taste, from Increased attontlolt Defeufto pricing to preserve product Mal-. re of Pre-
ONECTrVC prcepcbs Ie rntng and to match to. -. td to premium m. d. l. to marke(-broaden- category franchise. Search for Incr. - ft level prlc Log . hl the gal. r.! erer. Perception of Customary trade discounts Ing and promotional mental pricing opport naties. tochling tom, le".. disregard the most rev sp(IvS segm trans Aggressive promot 1-1 pricing, pricing oppe. rtunl- private Label eowncts, to boo. . alum. of an.. -!!. a es
" High red. discounts and sampil, tf f with prices cal as fast as coati ilea end µd, 1 on -prrirnee d. snsage market slur. tuft l "his decline due to accumulated pro-
duct los experience lntena lficatlun of sampling
(yT; O AI. al Create widespread awarenen Create and strengthen brad prof- Maintain consumer S inst. consumer and trade lot slay. Phase out. keepteg
GU'DEI. UE. 5 and onto stasdieq of offering erenre mmnng trade and final users franchise and . itta atroe emphases on dealers and din- Lust esauth I---Ie- Cen .e aliens benefits Stimulate general trial strengthen dealer trihutora. Promor. on of greater tee tat- profit. blit dts- Oo"ectý"es b)Go:. trial by surly adopters ties frequetsey tntaicw
bloss valuable To order of slum --__- Masa media Mass media %Iasa md). Cut d-. sD wedis so
1-du mu puytviry Personal sites Dealer pronotions Dealer . nested peomolbos she tax-+so ne Pursusi sales Sales pnn. oliona, including same- Pnrsoul selllrg; n0 We. pr-on. -u.,, a of
Mass e-oenewaodeuions ling dealer. any k. ad Publicity Sales promotions
Public dy
JUTHIDt ii i E-lulte or selectl. e. with Intensive and enensiv.. with dealer Intensive and exen- )Moot,. and crania., with . rang I`tus. -. a eaten, as POLICY dlstrltsaoe margins "rh enuuZ, to margins )teat high enough to keep sir., ad a strong emphasis on keet'mg dealer .. it auiplied. they ; even, saar-
justify honey promotional . Lending them interested. Close attention to emphasis on keeping but . mlaimum sen. nitory co. to him g, nal rapid resupply of dlstrllaaor stocks deader . ell supplied and heavy inventories u oil level, but . Ito minimum
Inventory cost to hi
- To Ideetuy e-ual de"elcying uso- Detallcd attention to hand poshlo,. Close attention to Intensifev attertiw to pmsablo product Infornatbo tslniq iT E LLIGENCL
systetss and to uncover .. y prod- to gnpn In med-l and market con- product In. pro" em<n ýImprvnemew a. Sharp alert for potential to de-ally the p. h. t i OCI: 3
uN . eake., s. s . rage. M to gilortu miles for 1 tads. to market- nor Inner-pedua c. "m pea lib. lid for at . roes LL Prodset
market segmtntatlnn broadening ch., Ccii. . 1gns of teglnnisg product decline shu. d Se passed o. a . - sryf to L'oasihie fresh
pron.. Inn them..
35
(i) The rate at which a new product's sales (or profit)
increases.
(ii) The maximum sales (or profit) level reached.
(iii) The period maximum sales (or profits) are sustained
before they begin to decline.
(iv) The rate at which sales (or profits) decline.
Hence forecasting the time and rate at which sales decline for
existing products, and forecasting the rate at which sales increase
and the maximum sales level reached for new products could allow the
NPD process to begin such as to minimize the reduction in profit levels.
This profit level reduction normally occurs during the period between
the sales decline of an existing product and the increase in profit of
a replacement product as illustrated in Figure 4.
The PLC has been formulated as an explicit, verifiable model of
a product's sales behaviour and tested against actual data in 140
categories of non-durable products by Polli and Cook(33). However
attempts by Doyle (34) and Polli and Cook to develop a PLC curve that
expressed the sales behaviour of a range of product types failed since
although most products have an identifiable life cycle, such cycles are
too irregular to permit a common PLC curve to be used as a forecasting
tool. Presently the PLC concept is used to aid the formulation of an
organisation's short-term objectives. For example, Fox and Rink(35,36)
describe how the purchasing department can control buying policies based
on the stage of the PLC occupied by a product and Figure 3 indicates the
36
FIGURE 4 PROFITABILITY GAP DURING PRODUCT
REPLACEMENT
4- 0
U +4 >U
ý------- , (U co 4-3ýýý ý .-atE 4- U I.
s.. (1) (1) ( ýý acs..
.e
I
i i
G) N>
r0 G) 1 >a) Q) En
ý ý
ý ý
ý ý
ý 'S
f �S
ýc\
++ . `ýc ý
.ývý y.. ý o +) iU aý
-v co
"r i a.
C) c cý
"r C P-" "r Ui (L) = o10
i i i i I I I I i L
I
ý a) C.
"C_ (L) ov
C) C_ +J C Ec)o o= tc) C)T)r
0 Vf L +) I/f C. C) 0ý
pi
C. +3 C. N
o"r- 3 a. Xa) ZWZ
I
N L fC3 Q)
aj E
. '.. F-
Profit (E's)
37
use of the PLC curve in aiding a company to set product design and
pricing objectives, promotional guidelines and distribution policy.
Established forecasting techniques such as Time Series Analysis,
Moving Averages and Trend Projection (20,37) do not have the ability to
forecast the PLC curve since these techniques merely project past sales 'Of demand forward into the short term future in order to allow effective
production planning and scheduling to be. carried out. Hence these
techniques cannot forecast PLC "turning points" (i. e. points at which
maximum sales are reached and the sales level begins to decline) since
these points are not related to past sales levels but to such factors
as:
(i) Purchasing motives.
(ii) Product acceptability in the market place.
(iii) Present and future competitive action.
(iv) Strength of competitors in the market place.
The present New EHB development provides a typical example of
the problems that could arise when using only short term forecasting
techniques. The market share of the existing Ropemaster EHB range had
been declining for approximately 5 years before the decline was
identified as a trend, i. e. the sales of the Ropemaster range increased
during this period due to an expansion of the total market size.
However H. M. Ltd. 's market share was gradually being eroded (Appendix I).
38
Short term forecasting methods gave no warning that development work
was necessary since these techniques analysed sales trends only and
not underlying causes such as, poor product design, uncompetitive
product prices and poor marketing strategies.
By the time the need for extensive improvement of the wire-rope 'Oe hoist design was identified an additional product range (Electric Chain
Hoist) had also become unprofitable. Insufficient development funds
were available for beginning the development of both product ranges
simultaneously. The Electric Chain Hoist (ECH) range was given
priority since the profit potential for this range was greater than
for the EHB range. The NPD process for the EHB range was therefore
delayed still further.
Diffusion theory used to model the growth of a new product in
the market (i. e. the initial stages of the PLC curve) suggests that
there is a time lag in the adoption of products by different companies
within a market place. Rogers (38) proposed that a product is first
adopted by a select few "innovators" who in turn influence others to
adopt the product. Thus it is the interaction between adopters and
non-adopters that is assumed to account for the rapid growth stage in
the diffusion process as seen in the PLC curve.
Initial attempts to model the diffusion process(39) employed a
deterministic interpretation of the time dependent aspects of diffusion
(assuming a constant market size) resulting in an S-shape diffusion
39
curve for the cumulative number of adoptions, Nac (Equation 1).
r" t=C
Nac =
oe
1
exp. 1 2a 2 �'L"
t=o
Where:
exp. = Exponential Constant.
Nac = Cumulative number of adopters in time-c.
a= Standard deviation of time.
t= Time period.
11 = Time at which fifty per cent market potential is
achieved.
(1)
Mahajan, et. al. (40,41)
considered that assuming a constant market
potential for a product was unrealistic since it was equivalent to the
statement that "the fate of a product is pre-determined at the time of
introduction". A model was presented by Mahajan and co-workers for such
consumer products as, washing machines and gas water heater, in
which the ceiling on the cumulative. adoption or the market potential
for a product changed with time.
Since both deterministic and dynamic diffusion models contain
variables (a u) that are unknown functions of total marketing activity
within the market they specify, diffusion models can only be used for
dt
L2a2J
40
interpreting historical data and not for predicting future performance.
That is diffusion models cannot forecast the initial stages of the PLC
curve.
Chambers, et. al. (37)
compared eighteen forecasting techniques
in terms of ability to predict both long and short term events, PLC
"turning points", requirements for developing the forecasting method
and operational cost. Although qualitative methods such as Delphi
and Market Research were identified as adequate predictors of PLC
"turning points", these techniques rely heavily on subjective
evaluations by "experts" and hence are prone to errors.
Chambers, et. al. (37)
suggest that causal forecasting techniques
(e. g. Regression Analysis and Econometric Models) overcome the need for
subjective evaluations since these methods employ equations that relate
sales trends to their underlying causes. In addition causal forecasting
methods are applicable to product planning strategies since to be
effective models must be developed for individual product types. -
This reduces the level of inter-relationships between forecasting
variables.
Hence to develop techniques for forecasting the shape of the PLC
curve it is necessary to determine the criteria that affect product
sales. In this respect the research into"buyer behaviour"is relevant
since this is an area of market research that examines, from the buyers
viewpoint, the socio-economic factors behind the purchase decision.
41
Current research by Newa11(42) and GrOnhaug(43) is based on the theory
that since the buyer seldom has full product information or knows the
full consequences in choosing a particular supplier the purchase
decision is perceived as a risk problem. The determinants of
perceived risk in industrial buying identified by Newall are:
(i)ce Those describing the purchasing problem:
(a) the size of the product expenditure,
(b) the type of purchase or buying task,
(c) the degree of product essentiality, and
(d) the factors provoking the purchase.
(ii) Those describing the industrial buyer:
(a) his level of general and specific self-confidence,
(b) his level of decision experience,
(c) his purchase history, and
(d) his degree of technical and professional affiliation.
(iii) Those describing the buying environment:
(a) the size of the buyer company,
(b) the financial standing of the buyer company,
(c) the degree of decision centralisation, and
(d) the degree of decision routinisation.
42
Many of the determinants of perceived risk identified by Newall
are beyond the control of the supplier company, i. e. those describing
the industrial buyer and the buying environment. However the factors
describing the purchasing problem need to be considered in relation
to their effect on sales levels. In this respect the work by Lehmann
and O'Shaughnessy(44) has shown that the perceived risk (i. e. the type
loo of problems: likely to arise in adopting a particular product) influences
the relative importance of the criteria affecting sales levels. Table 5
illustrates the variation in ranking between four product types tested
by Lehmann and O'Shaughnessy, each presenting different types of risk
to the potential buyer:
Type I Routine Order Products - products that are frequently
ordered and there is no problem in learning how to use such
products.. In addition there is no doubt that the product will
carry out its function.
Type II Procedural Problem Products - the buyer is confident
that the product will carry out its function, however a problem
exists in training personnel to use the product.
Type III Performance Problem Products - there is doubt as to
whether the product will perform satisfactorily in the applic-
ation for which it is being considered.
43
TABLE 5 VARIABILITY IN ATTRIBUTE RANKING BETWEEN PRODUCT TYPES Ref. (44)
I
Attribute Ranking
I II III IV
Reputation 4752
Financing 9 16 16 13
Flexibility 3525
Past Experience 6 13 9 10
Technical Service 12 137
Confidence in Salesmen 14 15 15 16
Convenience in Ordering 15 17 17 17
Reliability data 11 11 43
Price 2881
Technical Specifications 5966
Ease of Use 10 278
Preference of User 13 14 13 14
Training Offered 16 3 12 11
Training Required 17 12 14 15
Reliability of Delivery 1414
Maintenance 8 10 11 12
Sales Service 76 10 9
44
Type IV Political Problem Products - when products necessitate
large capital outlays problems can arise since there are always
rival projects for the allocation of funds. More frequently
political problems arise when the product is an input to
several departments whose requirements may not be congruent.
The variation in relative importance of factors affecting the
purchasing decision with perceived risk involved in adopting the
product (Table 5) indicates that a supplier must relate his product
to the prospective application and attempt to deduce from this the
problems the customer is likely to encounter in adopting the product.
In this way the criteria affecting a product's sales levels can be
deduced.
In terms of forecasting the decline of a product's sales level
Kotler(8) identifies two basic causes:
(i) Changing market needs - environmental effects-often lead
to fundamental changes in market requirements, e. g.
increasing oil prices is leading to a decline in demand
for large automobiles . In addition the relative.
importance of factors affecting sales levels can be
influenced, e. g. high interest rates on borrowed capital
tend to increase the importance of product price .
45
(ii) Competitive action - the introduction of an improved
product onto the market by the competition can lead to
.. e
an existing products sales decline. This agrees with
the behaviour of buyers observed by Cunningham and
White (45) , i. e. the comparison of product offerin. gs by
potential buyers has a considerable influence on the
purchase decision.
A detailed account of the method of comparing alternative product
offerings by potential buyers is provided by Webster(46) which shows that
industrial buyers carry out two tasks which require the collection and
analysis of alternative product data:
(i) Criteria must be established against which potential
supplier's offerings must be evaluated. These include
product specification, acceptable price band, delivery
needed and number of products to be purchased. Alternative
supplier's offerings must be identified hence a search is
carried out for suppliers who may be able to provide the
type of product needed and an examination is made of what
exactly they have to offer.
(ii) Buyers initially select those product offerings that
possess the specifications demanded. Those products that
cannot meet the specifications are eliminated. The buyer
then compares the remaining product offerings and seeks to
determine the product that minimizes the perceived risk
involved in the purchase decision. Cunningham and White (47)
46
!
in examining the purchase of four types of standard
machine tools sold into the UK market support this
observation. They found that many of the supplier and
product attributes thought to affect the purchase
decision did so to the extent that buyers required
an absolute level of achievement on the part of
suppliers before offers could be considered for
comparison with competing products.
In order to develop a method of forecasting PLC "turning points"
therefore it is necessary to:
(i) Identify the relevant criteria affecting product
sales levels.
(ii) Determine the relative importance of these factors.
(iii) Determine the relationship between these factors and
sales levels.
(i) Identifying the Relevant Criteria
Kiser, et. al. (48)
using attributes to which researchers
have imputed some importance in the evaluation of suppliers,
developed a definitive list of supplier attributes (Table 6),
which can be used in the form of a check list. Since the
relative importance of attributes varies with such factors
as product type and background of the purchasing executive
all factors listed in Table 6 could not be appropriate to
any specific buying situation.
47
TABLE 6 SUPPLIER ATTRIBUTES AFFECTING THE PURCHASING DECISION
(Ref. 48)
Convenience-Related Attributes: "'1. Deliveries without constant follow-up
2. Can deliver quickly 3. Answers all communications promptly 4. Advises us of potential trouble
'S. Handles rejections promptly and efficiently 6. Adapts to specific needs
'7. Allows credit for scrap and salvage 8. Offers broad product line
'9. Accepts small order quantities '10. Is located in close proximity 11. Anticipates our requirements 12. Is "direct" source of supply
Economic-Financial Attributes: 01. Has competitive prices '2. Offers volume discounts 43. Guarantees price protection '4. Has favourable financial position 'S. Sells at a lower price '6. Offers higher cash discounts '7. Offers extended payment terms
Calibre- Capability Attributes: 1. Regularly meets quality specifications 2. Maintains up-to-date stock
'3. Has good packaging. including packaging slips
'4. Has knowledgeable salesmen 'S. Has high calibre management '6. Has technical ability and knowled, ̂, e '7. Has potential to expand capacity 68. Helpful in providing special handling
equipment 9. Has research and development facilities
Laagt-Depc'. ' R!! ity Atrrtuees: '1. Reliable in quality '2. Reliable in delivery '3. Is fair and honest in dealings 04. Keeps promises S. Has favourable reputation
'6. Has favourable attitude '7. Exhibits desire for business 8. Maintains favourable fabcur management
relations '9. Is a progressive firm
'10. Offers well-known brands and/or products
'11. Utilises effective selling methods '12. Is a well known firm '13. Is a large firm
"1. 2.
"3. 4.
"S. "6. 7.
1. "2 "3. "4. "S. 6.
Inter-corporate Relations Attributes: Is a current supplier Source has been used before Is accepted by our other departments Is recommended by our other departments Is known to our firm Is affiliated with our firm Is a customer or our Service-Related Attributes: Helps in emergency situations Provides needed service information Invoices correctly - Makes salesmen available as needed Offers frequent delivery service Helpful in overcoming our occasional
errors '7. Maintains consignment stock at vendor
plant '3. Supplies parts lists and operating manuals 9. Offers better warranties
'10. Maintains technical service in the field '1I. Maintains repair service
12. Makes available test or demonstration models
"13. Supplies special reports '14. Helpful in providing special handling
equipment IS. Maintains frequent sales calls
'16. Provides information through advertising '17. Provides information through
promotional activities
48
A more appropriate method of identifying relevant criteria
is by market research since this identifies only those
factors relevant to a particular buying situation. From
the market survey carried out by H. M. Ltd. (26) the major
factors affecting user choice of wire-rope hoists were
found to be: I
(a) Specification, i. e. correct capacity,, height of
lift and duty rating.
(b) Price.
(c) Component and product standardisation through
range.
(d) Delivery.
(e) Life of product.
(f) Reliability.
(g) Ease with which hoist can be serviced.
(h) Technical and maintenance service levels required.
In general hoists are a -l ova level buying decision in an
organisation and the Works Engineer usually has the greatest
influence in determining the make of hoist to be purchased.
Because of this ease of maintenance and reliability are
important criteria affecting the purchase decision.
49
It is obvious that reliability and serviceability would
be important factors affecting the make of product
purchased since hoists are often located in difficult
access positions (i. e. overhead) and when out of service
frequently prevent or seriously interfere with the
purchasers manufacturing operations. The author has
practical experience of the effect inoperative hoists
have on manufacturing processes. The overhead crane
unloading aluminium ingots from the pre-heat furnaces
to the hot rolling mill at Alcan Sheet Ltd., Newport,
Gwent, when inoperative for several weeks prevented the
majority of the company's manufacturing processes (e. g.
cold rolling, annealing, slitting, embossing and packing)
from operating, since no hot rolled stock could be produced
to supply these processes. At H. M. Ltd. itself much inter-
operation handling is dependent on overhead travelling
cranes, failure of which could seriously affect production.
i
Distributors and Original Equipment Manufacturers (OEM's)
form a significant part of the hoist block market and are
concerned primarily with quick deliveries, price,
competitive discounts and a product range with a large
variety of options (e. g. dual lifting speeds, secondary
load brake and overload warning) and mounting positions.
50
(ii) Determining Relative Importance
ýe
Several methods are available for determining the relative
importance of product and company attributes, i. e. Regression
Analysis (49), the use of the Pareto Distribution (50) and
Market Research (26).
The regression technique involves determining values for
design attributes of competitive products and using
multiple regression analysis to statistically determine
the relationship between attribute values and corresponding
sales levels , hence obtaining the relative importance of
attributes as coefficients in the regression model.
Regression analysis is examined in detail in Section
3.3.6.4 of this thesis. The use of this method is limited
since extensive data on competitive products (e. g. delivery
time, product reliability and level of technical back-up
provided) is often difficult to obtain.
Starkef(50) determined the relative importance of product
design attributes by subjectively ranking each attribute
in order of importance and using the Pareto Distribution
to assess their relative importance. However, two basic
problems exist when using this method:
51
(a) The initial ranking of the attributes is a subjective
decision and therefore prone to error.
(b) Assuming that the Pareto Distribution (Figure 50)
can be applied to attribute ranking is not always
correct. Table 7 compares the relative importance
of attributes determined by market research and the
use of the Pareto Distribution and indicates that
discrepancies can result. Market research is normally
the most effective method of determining the relative
importance of product attributes since the reasons for
purchasing a product can be elicited directly from
purchasers. The relative importance of attributes
applicable to the development of the New EHB is
listed in Table 7.
TABLE 7 RELATIVE IMPORTANCE OF NEW EHB ATTRIBUTES
Attribute Market Research Pareto Distribution Ref. 50)
Reliability 0.39 0.49
Serviceability 0.24 0.35
Price 0.14 0.07
Life 0.10 0.04
Delivery 0.09 0.03
Service Levels 0.04 0.02
1.00 1.00
52
When using marketing analysis to obtain data it is
necessary to consider the frequency with which market
studies are conducted since the relative importance of
product and company attributes alter with changes in the
market environment.
(iii, ) Determining the Relationship between Product Attributes
and Sales Levels
When the relative importance of product attributes is known
it is then necessary to use this information to determine
the customer preference for a particular products'sales
levels.
Models that view the choice process as a multi-attribute
decision making problem have been established to predict
buyer preference for product brands (i. e. a buyer has to
choose between alternative brands characterised by attributes
such as price, quality and reliability).
Generally marketing studies on multi-attribute decision
making have concentrated on linear models of attribute
combination (51). Although complex multiplicative models
have been-developed, Huber (52) reports that these models
have only similar accuracy levels to simple linear models.
Effective linear attitude models of the form shown in
Equation 2 have been developed by Beckwith and Lehmann (49)
for predicting the relative preferences of individuals for
53
various television programmes and Pekelman and Sen(51)
for predicting consumer preferences for alternative
cereal brands and automobiles.
Pfb = /
a=d
a=1
Where:
Wa (Uba - a) (2)
Pfb = Estimated overall preference for product b.
Wa = Relative importance of the ath attribute.
Uba = Perception of product b on the ath attribute.
d= Total number of attributes.
Ia = Ideal point on the ath attribute.
The research by Wildt and Bruno (53) indicates that linear
attitude models are more applicable to industrial products
than consumer products. This is possibly due to greater
care being exercised during the industrial buying situation
than when purchasing consumer goods.
Since buyers compare alternative products in order to lower
the perceived risk inherent in the purchase decision,
Equation 2 must be re-specified for the industrial buying
situation. That is 'a would now represent the attribute
ý
54
value for a competitive product. The problem begins to
become complex when several alternative products are being
considered since a value for Ia is difficult to assess.
Pekelman and Sen(51) observed that most buyers used very
few attributes when comparing alternative products : (Figure 5).
This is possibly due to the buyers poor information handling
ability and limited time in which to make the purchase
decision. Both factors have been proposed by Gr$nhaug(43)
to explain the search behaviour of buyers, i. e. Equation 3.
Search is a function of (PR, TP, AH) (3)
where:
PR = Perceived performance risk.
TP = Time pressure.
AH = Ability in information handling.
Figure 5 therefore indicates that the complexity of the
multi-attribute decision making problem could be significantly
reduced by using only the small number of the critical
attributes in Equation 2.
The use of multi-attribute decision models in predicting
buyer preference and hence product sales levels is limited
in two ways:
55
FIGURE 5 NUMBER OF ATTRIBUTES EXAMINED PER
BUYER DURING THE BUYING DECISION
ý
C) ý
4- 0
C 0
4-3 S- o
0 S-
136
of
0.8
0.6
0.4
0.2
1234
Ref. 51
Number of Attributes Examined per Buyer
56
(i) The method requires extensive analysis of competitors.
According to Rothschild(54) a disciplined approach is
necessary to assess each major competitor (and
interactions between competitors) which should aim
at providing answers to basic questions such as:
(a) Who is the competition now and who will it be
in the future?
(b) What are the key competitors' strategies,
objectives and goals?
(c) How important is a specific market to the
competitors and are they committed enough to
continue to invest?
(d) What unique strengths do the competitors have?
(e) Do they have any weaknesses that make them
vulnerable?
(f) What changes are likely in the competitors
future strategies?
(g) What are the implications of competitors
strategies on the market, the industry and
ones own company?
57
(ii) Multi-attribute decision models do not take into
consideration the observation of Cunningham and
White(47) that a strong determinant of a buyer's
patronage decision is his past experience and
natural reluctance to change supplier. Hence the
extreme difficulty in recapturing lost markets. le
58
CHAPTER 2
ASPECTS OF THE DESIGN PROCESS
59
2.0 Introduction
Product design can be considered essentially as adecision-making
process since the evaluation of alternative design solutions requires
decisions to be made. That is, a single strategy must be chosen from
a number of alternatives. It is to be expected therefore that the
stages involved in the design process (55,56)
represent those of (55 decision-making models,
57) /
The recognition of design problems presents a serious challenge
to the designer since the rapid changes occurring in the manufacturing
environment both create new problems and alter the relative importance
of existing ones, e. g.
(a) The need to design products on a modular basis,
i. e. the use of standard components and sub-assemblies
between products is becoming increasingly important in
allowing distributors to maintain an adequate range of
product types and sizes with minimum stock costs.
(b) Shaftel(58) also considers that modular design is
necessary if there are economies of scale-in manufacturing
sub-assembly modules. That is, it may be worthwhile to
standardize on a few module designs that will be used in
all products or applications even though a few more parts
than needed are installed. The cost of delivering extra
parts by using few standard modules must be balanced
against the fixed costs of producing more types of
standard modules.
60
(c) There is a growing interest in the concept of
Terotechnology(59'60) in which the objectives of the
design function are to minimize the total costs
incurred during the working life of a product as
compared with the current objectives of minimizing
supplier manufacturing costs.
(d) The design function may be required to develop products
that can be manufactured in existing Group Technology
cells (61)
or are applicable to modern production control
concepts such as Period Batch Control(62) and Materials
Requirement Planning(63).
(e) The growing use of flexible automation(64) can be expected
to influence design criteria since it improves the flexi-
bility of use of machine tools and hence allows greater
product design variability to be considered.
The use of a formal decision-making model (Table 8) could enable
effective solutions to be obtained for each design problem that occurs
during the NPD process. However this is normally prevented due to
insufficient design resources since it would entail considerable design
effort. Hence it is necessary to select those problems that would
yield the greatest benefits from use of a formal decision-making process.
61
TABLE 8 STAGES IN THE DESIGN PROCESS Ref. (55)
(i) Recognition of the design problem objectives.
(ii) Specification of the objective constraints.
(iii) Synthesis of alternative design problem solutions.
i (iv) Evaluation of alternative design problem solutions.
- (v) Deciding the solution to adopt.
The obvious method of selecting design decisions is to determine
their relative importance to the overall design objectives. Starkey (65)
and Sanders (66) identify three basic groups (Fundamental, Intermediate
and Minor), in which the contribution to the total manufacturing cost
involved in a design decision or the cost and ease of changing that
decision is used to determine the relative importance of a problem.
It is suggested that the relative numbers of Fundamental, Intermediate
and Minor decisions follow the Pareto distribution in that a small
proportion of the total number of decisions are Fundamental and
Intermediate. Hence it would appear that these groups of decisions
could be solved using formal decision models. However a Fundamental
or Intermediate decision is often the sum of numerous minor decisions.
For example, deciding the fundamental design concept of the New EHB hoist
(Hydraulic or Electro-Mechanical) involved the detailed design (i. e.
minor design decisions), of each concept in order that comparisons
could be carried out.
62
During the synthesis stage of the decision-making process the
engineering designer is called on to use his creative ability and
experience in proposing design solutions. Ostrofsky(56) suggests
that the higher the number of alternative design solutions proposed
the greater the possibility of choosing the optimum solution. During
the design of the New EHB range it was observed that designers
encountered little difficulty in proposing alternative design
solutions but it was the evaluation and comparison of design solutions
that restricted the number that could be considered since:
(i) The amount of data required to evaluate designs
(e. g. costs, producibility, reliability and safety)
is large.
(ii) Several functional departments are often required to
evaluate designs hence co-ordination of input data is
difficult.
(iii) The resources available within the various functional
departments are limited, i. e. departmental resources
are normally sufficient to cope with the day-to-day
company data requirements and cannot cope with the
additional amounts required during the NPD process.
The criteria used to evaluate alternative designs can be grouped
into the following categories:
63
c>> Specifications, i. e. design characteristics that determine
the range of market specifications covered. Hence a
market specification is defined as that characteristic
of a product required by a potential buyer before the
product can be considered for comparison with competing
products. i
(ii) Attributes, i. e. those product characteristics such as
reliability, quality and style that are used by a buyer
to compare competing products.
(iii) Producibility, i. e. the ease with which a product can be
manufactured and raw materials purchased.
(iv) Cost - of prime importance in appraising designs since it
is used to evaluate other design criteria, e. g. the range
of market specifications to design for, the level of
reliability, quality and safety to design for and the
ease of producibility in terms of labour costs. In
addition the manufacturing cost of a product is normally
a major determinant in the purchasing decision.
Several areas of H. M. Ltd. 's new product design process have been
identified as limiting the number of alternative design solutions that
can be economically compared and/or leading to wasted design resources
and/or preventing an optimum product range being developed.
A major factor that causes problems is the inability of the current
estimating system at H. M. Ltd. to quickly generate the large amounts of
64
cost data required during the design process. In consequence Chapter 3
examines in detail the estimation of product costs at H. M. Ltd. The
need for an estimating system that can be used by engineering designers is proposed and such a system is developed in Chapter 4.
Other areas of the design process called into question are:
(i) the "make" or "buy" decision,
(ii) value engineering,
(iii) the "assortment" problem,
(iv) planning and control of the design process,
(v) scheduling of the design process, and
(vi) the changes that occur in product designs.
These areas of the design process are now examined in greater
detail.
2.1 The Make or Buy Decision
The decision to manufacture a component in-house or purchase
from an outside supplier is a frequent problem facing the designer.
In theory each component of a product design must be assessed in this
manner. However the decision to make or buy is often obvious. For example
65
either the company does not possess the equipment to manufacture
in-house, or the component is a standard, mass produced part, or the
development objectives are to design for either in-house manufacture
or purchasing.
For components in which the decision to make or buy is not
obvious the factors to be taken into consideration have been identified
by Claret(67) and Haas and Wotruba(68). These factors are listed in
Table 9.
TABLE 9 MAKE OR BUY DECISION CRITERIA .
(i) The need to maintain economic utilization of capital
equipment.
(ii) The need to absorb high overhead costs.
(iii) Assurance of supply.
(iv) If the component design is a trade secret.
(v) Transportation costs and ease of transportation.
(vi) Company reputation gained through manufacturing parts
rather than merely factoring them.
(vii) The need to diversify operations.
(viii) The need to smooth seasonal variations in order patterns.
(ix) Lack of sufficient capital to finance required investment.
(x) The familiarity or experience with the technology involved.
66
Haas and Wotruba examined the make or buy decision in relation
to marketing products which the potential customer could also make,
i. e. the prospective customer is at the same time a potential competitor.
They concluded that the decision to make or buy was dependent on the
difference between the cost to the prospective customer should they
manufacture the required product in-house and the prices of that 11
product from possible suppliers.
The manufacturing costs highlighted by Haas and Wotruba included
both the direct and indirect costs of manufacturing components (i. e.
direct material costs, factory supplies costs, direct labour costs,
indirect labour costs and factory overhead costs). However Claret(67)
considers that only those costs directly related to the volume of
components manufactured (i. e. variable manufacturing costs) should be
used in the make or buy decision.
The approach adopted by Claret is more realistic than that of
Haas and Wotruba since normally only variable costs are affected'by
the decision to make or buy. Fixed costs such as factory overheads
are usually incurred irrespective of the type of decision.
Claret considers that in certain circumstances (where labour
resources cannot be used elsewhere) that direct labour costs can be
considered fixed costs and should not be included in the make or buy
decision. However during the development of a new product range
the labour resources required to produce a component should not be
considered fixed since an objective of the design department is to
67
minimize a component's manufacturing costs which include a labour cost
element. Labour costs must therefore be included in the make or buy
decision during new product development.
Levy and Sarnet(69) considered that the make or buy decision should
be based on the differential risk involved in the various components
of the cash flows (e. g. investment outlay for capital equipment, forecast
sales quantities and the purchase and production costs per component)
engendered by the alternatives. However this approach need only be
considered if one alternative necessitates a higher level of capital
investment, in which case the basic Uncertainty surrounding forecast
quantities and estimated costs makes it apparent that the alternative
that requires the more expensive equipment constitutes the greatest
risk.
Normally component designs contain elements of both make and buy,
i. e. the raw material is purchased from which a component is manufactured
in-house. Frequently therefore a designer must decide in what form to
buy raw material or components (e. g. bar, casting and part moulded
plastic components) and which manufacturing operations to carry out
in-house. The present approach to the make or buy decision allows
designers to make such choices by using classical Economic Order
Quantity theory (EOQ) (70) to develop a model for calculating the
total variable costs per unit of a component when purchased,
manufactured in-house or a combination of both. The use of stock
control models also allow the variability of component cost with
purchasing and manufacturing batch sizes to be taken into consideration.
68
The total material cost of a component is calculated(71) using
Equation 4.
Cm =k+k. Ic + Cb. Qb-'
Where:
Cm = Total material costs per unit.
k= Purchase cost per unit.
Ic = Stock holding interest rate per unit per year.
Cb = Purchase cost per batch.
Qb = Purchased batch size.
(4)
The variable costs of in-house manufacturing operations is given
by the following expression(71) :
Cp =C ý+ Cs. Qp-1 (5)
Where:
Cp = Total variable in-house labour costs per unit.
Cy, = Total labour cost per unit.
Cs = Total setting up cost per batch.
Qp = Production batch size.
69
Capital investment costs such as special tooling costs, although
fixed and non-recurring over the short term, are recovered over a
specific time period the length of which is normally a management
decision (at H. M. Ltd. this period is normally three years) and is
therefore influenced by managements attitude to the risk involved in
the development project and financial constraints on borrowed capital.
Equation 6 for recovering capital investment costs is given by
Ostwald and Toole (72):
Ct = CT. 2.718-r. ¢
p=r Ncp
p=1
Where:
Ct = Tooling or investment cost per component.
CT = Total tooling or investment costs.
Ncp = The number of components manufactured for the
first p years of the product life.
r= Recovery period for capital outlay (Years).
= Company's discount'rate for capital investment.
(6)
70
The expression 2"718-r. is used to take into consideration that
tooling or investment costs are incurred at the beginning of the payback
period and must therefore be discounted to give a more balanced
perspective to the investment costs.
The total variable cost involved in buying and/or making a
component is therefore the sum of the individual costs obtained from
Equations 4,5 and 6, i. e.
Cv = Cm + Cp + Ct (7)
Where:
Cv = Total variable component costs.
According to Burbidge(73) Qb and Qp are not always identical
since there are different ways in which parts are batched together
for convenience in production, Table 10.
TABLE 10 TYPES OF BATCHES PRESENT IN A MANUFACTURING
ORGANISATION Ref. (73)
(i) Order Quantity (OQ) - quantity of a component authorised
for production or purchase by a manufacturing order or
purchase order.
71
TABLE 10 (Contd)
Run Quantity (RQ) - quantity of a component run-off at
a machine or other work centre before changing to some
other component.
(iii) Transfer Quantity (TQ) - quantity delivered at one time
by a supplier to a customer, or transferred in a factory
as a batch between two machines for succeeding operations.
(iv) Set-Up Quantity (SQ) - quantity of components (not
necessarily all the same) which are produced on a
machine before changing the tooling and machine set-up.
Qb and Qp must therefore be determined separately. Classical
stock control theory (71) can be used to calculate the value of Qb
that minimizes Cm. It must be assumed that the values of Qp for each
operation required to manufacture a component are identical since this
represents efficient control of a company's manufacturing operations.
In addition a value for Qp that minimizes Cp can then be
established by the classical stock control approach, i. e. differentiating
Cp with respect to Qp, with the minimum value of Cp occurring when:
dC=0 d (Qp)
72
Equation 7 is therefore a useful tool for determining the costs
involved in the make or buy situation and can be used by engineering
designers for comparing design alternatives. For example, various
design alternatives were proposed for the New EHB Rope Guide, the
assembly which prevents adjacent rope strands from rubbing together
whilst the rope is being wound around the drum tube and hence prevents 110,
rope wear. The alternative designs considered were:
IA purchased phosphor bronze casting (currently used on
existing hoist block range) with tooling costs included
in the component purchase cost.
II A purchased, fully machined plastic component (machined
by the supplier).
III A purchased plastic moulding with separate tooling
costs and no subsequent machining required.
IV A purchased plastic moulding with separate tooling
costs that required finish machining operations at
H. M. Ltd.
VA purchased plastic moulding with separate tooling costs
and finished machined by the supplier.
The comparative costs (determined using Equation 7) of the
alternative rope guide designs are shown in Table 11 for r=3 years
(the current H. M. Ltd. payback period) and r= 10 years (the nominal
life of the moulding tools).
73
TABLE 11 MANUFACTURING COSTS OF ALTERNATIVE ROPE GUIDE
ASSEMBLY DESIGNS (3.2 TONNE NEW EHB)
Manufacturing Costs/Unit (£'s)
EOQ Method
Design r=3 years r= 10 years Alternatives
I 69.24 69.24
11 71.25 71.25
III 31.48 10.61
IV 32.31 18.33
V 30.74 16.70
PBC System
r=3 years r= 10 years
69.24 69.24
71.25 71.25
31.95 11.08
39.55 25.51
37.80 23.76
The influence of payback period on the comparative costs of design
alternatives is seen, i. e. when r=3 years alternative V is the least
expensive. However using the Terotechnology approach(59) and recovering
tooling costs over their nominal life (r = 10 years) indicates that
alternative III is the least expensive. Of course, a compromise
solution could be obtained by using alternative III but recovering
tooling costs over 3 years. However this approach tends to inflate the
cost of a product at a period in the PLC when competitive prices are an
advantage to increasing the growth rate of a product's sales in the
market place. Hence choosing the appropriate design alternative could
become a marketing decision.
Burbidge(62) suggests that the use of Period Batch Control (PBC)
reduces manufacturing costs since this production control technique
organises and plans manufacturing operations by controlling the flow
74
of stock. That is, the batch quantities ordered and their due dates
are determined from production programmes based on sales orders
received or short term sales forecasts as opposed to the EOQ method
which controls the level of stocks.
The PBC production control technique(62) releases orders in
product sets (on a series of common due dates). Hence all batch
quantities are identical, 'i. e. OQ = RQ = TQ = SQ (Table 10). The
actual batch quantity selected is dependent on the manufacturing
"cycle time". The year is divided into a number of approximately
equal time periods termed "cycles". The production volume per
product assigned to each"cycle"must be in balance with the market
demand and is allowed to fluctuate with demand. The "cycle time"
cannot be less than the throughput time for any of the components'
included in the PBC product sets and must not increase the proportion
of machine setting time so that capacity is reduced below the level
required to meet demand.
Assuming a PBC system was operating Equation 7 could be used to
compare the alternative rope guide assemblies, Table 11, i. e.
Qb = Qp = Ns. Te. Nw 1 ($)
Where:
Ns = Sales demand.
TI = Longest individual throughput time (weeks).
Nw = Number of working weeks in the year (weeks).
75
For the 3.2 tonne New EHB unit Ns = 380/year, TZ =6 weeks and
Nw = 48 weeks and Qp = 47.5.
From Table 11 it can be seen that the effect of lowering batch
sizes through the introduction of a PBC system tends to increase the
cost effectiveness of components requiring tooling or investment
costs to the disadvantage of machined components. In addition
several difficulties are encountered by the design function when
developing products for a PBC system:
(i) It is necessary to determine the "cycle time" before make
or buy decisions and comparisons of alternative design
concepts can be made. This requires a knowledge of the
longest individual component throughput time within a
product.
(ii) A PBC system requires detailed planning of manufacturing
schedules in order to reduce set-up and operation times.
Hence scheduling flexibility is reduced and new design
constraints arise, since the design function must now
consider the overall manufacturing sequence of a product.
(iii) PBC systems use Group Technology (62) manufacturing cells
to reduce manufacturing cycle times and stock levels.
Hence the design function could need to consider designing
components to fit into existing cells in order to avoid
76
the cost of redesigning the cell structure. This is
particularly applicable in multi-product organisations
if a cell structure is in existence for other products
within the company's product mix.
(iv) Engineering design changes introduced to improve materials
or manufacturing methods could form causes of inaccuracies
in computerised inventory control systems such as PBC (63)
and Materials Requirement Planning (MRP)_
2.2 Value Engineering
During the New EHB development period a Value Engineering (VE)
Committee was formed to examine methods of reducing manufacturing costs.
In accordance with value engineering methodology(74) the members of
the committee were chosen from each of the main functional departments
involved in the new product development process (i. e. design, production
engineering, purchasing and estimating). Although VE is a recognised
method of reducing product manufacturing costs the committee at
H. M. Ltd. achieved limited success since:
(i) VE meetings were not held regularly therefore the number
of components that could be examined was limited. The
frequency with which a VE committee should meet in order
to adequately examine a product range can be represented
as:
Vf = Ta. ý9)
Tp. Nw. TA
77
Where:
Vf = Frequency of VE committee meetings (per week).
Ta = Average VE time per component (hours).
Ne = Number of components examined.
Tp = Product development period (years).
Nw = Number of working weeks per year.
TA = Average length of VE committee meetings (hours).
i
From examination of the current value e. ngineering work
Ta =2 hours,. Ne = 400, Tp = 2.5 years, Nw = 46 and
TA =3 hours. Therefore from Equation 9 the frequency
with which VE committee meetings must be held is:
Vf =2x 400
46 x 2.5 x3 2.3 per week
Clearly holding over two 3 hour VE meetings per week would
be difficult since committee members could not have
sufficient time available to attend. Therefore methods
must be found to either reduce the number of components
examined or improve the rate at which VE work can be carried
out. Since VE is a creative cost reduction tool (rather
than an analytical technique) improving the rate at which.
VE work could be carried out would be inherently difficult
78
whereas obviously concentrating VE time on high cost
components achieves the initial option.
However to increase the rate at which VE work can be carried
out a method employed by the author during VE exercises (75)
was found to produce advantages. The technique involved
applying an individual cost reduction method to all the
components within a product design. Each of the cost
reduction methods listed in Table 12 were applied to a
product design in this manner. In this way the Value
Engineer takes advantage of a "learning curve" effect.
That is, as the number of components examined increases,
the speed with which succeeding components can be examined
and potential cost savings identified also increases.
To reduce VE time further it is also possible to employ
this method in conjunction with the technique of examining'
high cost components, i. e. the order in which components
are examined is determined by their relative cost. If
sufficient time is not available low cost components could
be ignored during the NPD process and value analysed at a
later stage.
79
TABLE 12 COST REDUCTION METHODS
(a) Reducing the volume of material used per component.
(b) Reducing the number of bought out componentsy
i. e. improving overhead recovery.
(c) Reducing-the labour cost per component.
. (d) Converting material costs to labour costs.
(e) Examining the possibility of using cheaper materials.
(f) Examine components for possibility of using plastics.
(9) Examine components for possibility of using standard
replacement parts.
(h) Reducing the number of components required.
(i) Reducing. the number of components required by examining
the possibility of components carrying out the functions
of others.
(j) Standardization of chamfers, radii, bolts, nuts, bearings,
screws and threads.
(k) Examining the possibility of reducing dimensional tolerances
of components.
80
(ii) During meetings formalised VE procedures were not
followed. To achieve results a carefully planned
programme must be followed consisting of seven basic
stages (Information gathering, Speculation phase,
Evaluation phase, Programme planning stage, Programme
execution phase, Recommendation stage and Implementation
stage). Again a reduction in the number of components
examined would effectively allow additional time for
the formal VE procedure to be adopted.
(iii) When VE is carried out during the development process it
is necessary to recommend design changes prior to prototype
build and test in order to prevent further development
work being necessary. However because of the inherently
slow speed at which VE work can proceed (compounded by
the lack of regular VE meetings) the prototype build and
test stages of many assemblies took place before the VE
committee could examine the designs involved. To prevent
the need for design changes occurring after the prototype
test stage value engineering of a component design should
be completed prior to the release of the design for
prototype manufacture. VE should therefore be considered
an essential preliminary step prior to the prototype build
and test phases.
81
2.3 The Assortment Problem
When developing a product to cover a range of market specifications
(e. g. size, capacity and power), fundamental decisions to be made are,
the number of standard sizes to manufacture and the exact specification
for each standard size. With respect to the New EHB development project
these problems centred around deciding:
(i) the number of basic frame capacities to develop as
standards, and
(ii) the Safe Working Load capacities of each standard size.
Decisions of this type fall into an important class of problems
defined as "the assortment problem" which according to Swanson (76)
are composed of two basic sub-classes:
(i) The "set-up and inventory"assortment problem.
(ii) The "job-shop assortment"problem.
In general the assortment problem looks at the determination
of what, if any, system of standard sizes might be manufactured such
that economies of scale obtained through manufacturing a large number
of a few standard sizes more than offset the lost extra capacity
(material and manufacturing time) in supplying products whose size
is greater than the customer requires. With the "set-up and inventory"
assortment problem the method of manufacture is assumed to have a
82
certain set-up cost and optimal production run length associated with
it. Therefore when different size products are combined to form a
standard the inventory, set-up and manufacturing costs must all be
considered in deciding whether or not to combine the various sizes
into a standard. During the "job-shop" assortment problem manufacturing
is assumed to be performed exactly to the specification of the customer. i
Hence there is relatively little inventory and the main consideration
is the differing costs for various methods of manufacture.
The assortment problem involves many subjective criteria which
influence the optimality of the solution and hence assortment problems
are often difficult to formulate objectively. The solution to the
problem must also reconcile the competing objectives of the functional
departments involved in new product development.
Important criteria to consider are listed in Table 13.
TABLE 13 FACTORS AFFECTING THE ASSORTMENT PROBLEM
A, MARKETING FACTORS
(i) Customer acceptance of sizes other than those historically
available.
(ii) Determining whether customers would accept sizes larger
than their requirements.
(iii) Competitive product sizes which may prove to provide poor
price comparisons if standards were chosen that were larger
than the 'competitors.
83
TABLE 13 (Contd)
(iv) The actual demand pattern, i. e. the preferential
sizes favoured by the market.
(v) The range of sizes provided for the market.
(vi) The profitability pattern, i. e. product size versus
profit.
(vii) Manufacturing lead times required.
(viii) The need to market a full range of product sizes,
e. g. at H. M. Ltd. product sales are strongly
dependent on customers standardizing hoist block
requirements in order to simplify stock holding and
maintenance procedures. Therefore customers tend to
buy from manufacturers who can supply their specific
requirements.
B. MANUFACTURING FACTORS
c> > Production flexibility, i. e. can the manufacturing
facilities handle the level of product and component
design variation associated with increasing the number
of standard sizes.
(ii) Stock policy, i. e. stock costs can be expected to rise
as the number of standard sizes manufactured increases.
84
TABLE 13 (Contd)
(iii) Production control methodology, e. g. batch scheduling,
PBC, MRP
(iv) Machine utilisation.
(v) Machine set-up and operation costs.
Machine type profile, i. e. the number and type of
automated machines.
C.
(vii) Type of manufacturing system, i. e. job shop, batch, high
volume production, group technology layout or automated
production lines.
(viii) Cost of standardization.
DESIGN FACTORS
(i) The development resources available influence the number
of standard sizes that can be designed.
C 1 i) "Bought out" to "made in" component levels in the
product design.
(iii) The degree of standardization between product sizes.
(iv) The suitability of the product to modularisation.
(v) Level of material, labour and overhead costs.
85
TABLE 13 (Contd)
D. CORPORATE FACTORS
(i) New product range development time scale.
(ii) Corporate policy. . le
(iii) Long term objectives for the product range.
Labour policy, i. e. degree of flexibility in labour
allocation.
(v) Working capital constraints tend to limit stock and
work-in-progress levels therefore reducing the number
of standard sizes that could be manufactured.
Since many of these factors are difficult to quantitatively define
the assortment problem often centres around the estimation of the
quantifiable benefits in order to weight these against the subjective
criteria. The problem definition then becomes either:
(i) determine the minimum number of standard sizes such
that the loss from manufacturing product sizes larger
than customer requirements is acceptable, or
(ii) determine the number of standard sizes such that the
disadvantages of manufacturing one extra standard size
are greater than the benefits incurred.
86
Each of the above constraints(Table 13) must be examined for its rela- tive importance to the assortment problem. For example development time
scale is an important constraint in the assortment problem for the New
EHB range since it is necessary to limit the development period. in
order that a new product range can be launched before a serious
loss in sales levels of the existing Ropemaster range.
2.3.1 Solution Techniques
There are several established techniques for solving the
assortment problem; complete enumeration, random enumeration,
dynamic programming and the minaddition technique. Each method
attempts to determine the number of standard units and their
respective sizes that minimizes manufacturing costs.
Complete enumeration for example involves the analysis of
all-possible combinations of standard sizes to determine the
relative savings obtained from each, with the solution yielding
the maximum benefits being adopted. However the number of possible
combinations to consider increases exponentially with a rise in
the number of possible sizes and the number of standard sizes.
Therefore complete enumeration cannot be used to solve H. M. Ltd. 's
assortment problem since there are eighteeen sizes to select from
(and hence 6.4 x 1015 possible combinations) and the problem
solution would require an excessively long time to evaluate.
87
Random enumeration of a limited number of combinations is
possible but it is difficult to determine how close to optimal
any trial solution is. Selecting combinations based on
intuition and experience is also liable to the same uncertainty.
Several investigators (76'77'78) have used Dynamic
Programming (DP) algorithms to solve the assortment problem,
i. e. the problem is represented as a sequential decision problem
which is solvable in stages. The first stage is to select
standard size one (the largest size). This decision is
normally a marketing or manufacturing constraint, e. g. the
largest size'requiredto cover the intended market segment or
the maximum size that can be manufactured due to component
size limitations on production equipment.
The subsequent stages involve selecting each individual
standard in order of decreasing size. The underlying philosophic
argument of the DP techniqe is known as the "principle of
optimality", which states that an optimal policy has the
property that whatever the initial state (number of sizes
available) and initial decision, the remaining decisions must
constitute an optimal policy with regard to the state resulting
from the first decision.
88
Jackson and Zerbe (77) initially solved the assortment
problem using the DP technique but according to Swanson(76) they
specified the problem illogically since:
i
(i) The solution required the pre-determination of the
number of standard sizes, i. e. this is a decision
variable and cannot therefore be pre-specified.
(ii) The solution determined which particular sizes
should become standards whilst leaving the other
sizes as custom designed. However, the overall
objective of determining the optimum number of
standards is basically to reduce the number of
sizes produced by manufacturing only a limited
number of standard sizes.
Taking these factors into account Swanson(76) found it
necessary at the initial decision stage to examine all feasible
single standards beginning with the largest item and working
down to the smallest. At the second decision stage all
feasible double standards were examined in the same manner.
It can be seen that extensive amounts of computation are
required when considering even small numbers of standard
sizes, i. e. 5,6,7 ...
89
The DP algorithm can be solved for simple problems using
a graphical technique developed by Swanson but for more complex
problems the sequential decision process method is required.
As the complexity of the problems increase the DP technique
rapidly becomes cumbersome and a computer is necessary to solve
the problem in a reasonable length of time. For a problem the
lo, size of H. M. Ltd. 's approximately 11K core storage is necessary
and the computing time is approximately 20 minutes.
A major disadvantage of using the DP technique to solve
assortment problems is the'difficulty of including non-quantifiable
criteria in the decision making process. This problem has been
highlighted by Martin and piff(78) when attempting to determine
the optimum number of fork lift trucks to manufacture. The DP
solution to this problem considered factors such as manufacturing
quantities, range of sizes possible, manufacturing costs and
sales prices and recommended that 22 standard fork lift truck
sizes should be manufactured. The subjective estimate of
management indicated that only 5 standard sizes should be
produced. The large variation in results indicated the effect
that the subjective criteria, listed in Table 13, have on the
solution to the assortment problem.
Considering the effect that subjective criteria have on the
optimum solution to the assortment problem the Minaddition
technique(79) has certain advantages over the use of DP
algorithms, since an assortment problem of H. M. Ltd. 's magnitude
can be solved in a reasonable amount of time without the aid of
90
a computer. Hence the cost of developing computer software
when employing DP algorithms is avoided.
The Minaddition technique can be used to solve the
assortment problem assuming that any required size not a
standard can be provided from the next standard size up.
This constraint is applicable to the present New EHB
development and, results in excess material and labour costs
being required. The Minaddition technique is then used to
select the particular set of standard sizes that minimizes
excess costs. When profit margins vary with product size
then the problem can be respecified to minimize total profit
loss.
Using the Minaddition technique the assortment problem
resolves into two distinct sections:
(i) Use the Minaddition technique to determine the
cost or profit loss from manufacturing various
numbers of standard sizes.
(ii) Determine the extra costs incurred from increasing
the number of standard sizes.
The optimum number of standard sizes to manufacture therefore
is that number where the savings obtained from adding one extra
standard is considered unjustified when compared with the extra
costs and manufacturing problems involved, e. g. reduction in
manufacturing efficiency, higher work-in-progress levels.
91
2.3.2 The Minaddition Technique
Wolfson (79) published a detailed description of. the
Minaddition method and included a flow diagram to aid computer-
isation when complex problems need to be solved. The technique
can be used during the New EHB development for determining the
optiwum number and lifting capacities of standard frame sizes
to manufacture.
Initially the data to be obtained includes the range of
frame sizes that are being considered, forecast sales quantities
per frame size and the "cost" of each frame size, Table 14.
The type of "cost" determined depends on the specific circumstances
of each problem and may be profit per unit, manufacturing costs
per unit, direct costs per unit, labour cost per unit or
material cost per unit. During the present analysis the prime
costs (i. e. material and labour costs) are used since profit
margins are similar throughout the product range and would not
therefore affect the outcome of the Minaddition solution.
The maximum and minimum frame sizes considered in Table 14
were determined from a knowledge of the size range used within (26)
the total market.
The Minaddition technique considers the range of possible
standard sizes in sections and temporarily regards the sizes
that bound each section (termed "interval") as standards. The
92
TABLE 14 DATA REQUIRED FOR THE MINADDITION TECHNIQUE
Ref. (26)
Hoist Unit Sales Forecast Prime Costs Frame Size (tonnes) (Units per Year) ("s)
Of
0.16
0.20
0.25
0.32
0.40
0.50
0.63
0.80
1.00
1.25
1.60
2.00
2.50
3.20
4.00
5.00
6.30
8.00
21
10
89
10
24
383
74
170
184
142
192
94
289
79
131
74
24
8
241
252
266
284
306
333
367
449
466
533
627
734
868
1016
1270
1515
1886
2368
93
total prime costs for each interval are then calculated in
stages, i. e. the cost that results when all other sizes in
the interval are provided from the larger of the two bounding
sizes. The prime costs in any interval would therefore remain
the same no matter which sizes greater or smaller than the
interval bounding frame sizes were varied. Costs are denoted
by:
Ce = AW (min' max
Where:
Ce, A = Prime costs.
Sm. = The lower bounding frame size.
Smax = The upper bounding frame size.
w= The number of standard frame sizes in the
interval.
(10)
The objective of the Minaddition technique is therefore
to minimize Ce for successive values of w.
Minimum prime costs have been determined for successive
values of w and are listed (with the corresponding optimum
standard frame sizes to manufacture) in Table 15.
94
TABLE 15 OPTIMUM PRODUCT RANGE DATA
v aý
. r..
.0 0 rn
ý ý f0
N
+) N O
U
Vf ý N O
C.. )
C) E
ý
d
VI C) N
N
C) E
LL.
ýI e
.,.. 4-) 0
,a b c
N
4- O
m E
4J x aýI aý
cm c
E 0
4-
Cl 0 0 r
'x ý
. -ý N G . )I
C O
ý
(L) N
. r.. N
i t0
C du
ý N
dý co ONN CM 01 M Ol to q; r ct -d" N N
ct O O1 r-- r- C11 1, -ý M r- r- 00 0T1 N 00
O LO N ý-- fM
C) 06
> N U v
C) 0 CD, r-- X
W v
r-
C) Cö
Cý C; CD
OÖ Lf) C9.0w
cö . w LA v 9ww
O LA LI) LAw
N" N. Cp Lj) ww C
C; CO "N. C9. Ll) w e-
w Lj) Ow Ow
CQ C. LC) N
CÖ C4 ... w CV r- r- p-.
LA L j) ww NN CD LA LAw L1')
.. r- r- r- 06C.
N CV) Ct LO l0 n
I
95
Since the Minaddition technique determines the prime
costs from manufacturing a fixed number of standard sizes the
cost saving resulting from adding one extra standard size can
be determined using Equation 11.
loe
Csv = Cep - Cep+l (1 1)
Where:
Csv = Cost saving obtained from adding one
extra standard size.
Cep = Prime cost for p standard sizes.
Cep+1 = Prime cost for p+1 standard sizes.
Having determined the cost savings it is necessary to
compare them with the increases in manufacturing costs that
are incurred through manufacturing additional standard sizes.
These cost increases arise primarily since:
(i) Increasing the number of standard sizes within a
product range effectively lowers the average
manufacturing and purchase batch sizes, hence
incurring additional purchase and setting-up costs.
(ii) The increased number of batches that require
processing effectively lowers the machine
utilization level.
96
(iii) Increased costs are incurred for special tooling
(e. g. rope guide moulding tools) and jigs and
fixtures.
(iv) The efficiency of production planning and control
systems could be lowered hence labour costs could
be increased, i. e. extra idle time is incurred.
(v) Work-in-progress and stock levels increase.
(vi) The reduction in labour costs and scrap levels
through the "learning curve" effect (see Section
3.1.2) is decreased.
In order to determine the optimum number of standard sizes
to manufacture it is necessary to compare the cost savings with
the added costs involved in increasing the number of standard
sizes. The optimum number occurs when Csv < Ca where Ca represents
the additional costs incurred.
It is an advantage to use the Minaddition technique in
determining the optimum number of standard sizes since accurate
estimation of Ca is not required. Figure 6 illustrates the
rapid rate at which cost savings are incurred when adding extra
standard sizes and indicates therefore that only rough estimates
of the additional costs incurred are needed to determine the
optimum number of standard sizes. H. M. Ltd. estimated that the
extra costs incurred in adding an additional standard size lay
97
5
FIGURE 6 OPTIMUM NUMBER OF STANDARD HOIST SIZES TO DEVELOP
Cost Savings (Csv)
ý Ln
0
X
ý
V1 C'e C
"r > ý
N
y. i N O
U
4
3
2
1
0.5
Si Vli
L- ----------------
Estimate of -r \ increase in costs from_adding onp-____ extra standard
II
I I i ýº
Optimum range of standard sizes to manufacture
345678
Number of Standard Sizes
98
between 150,000 and 1100,000. Hence Figure 6 indicates that
the optimum number of standard sizes to develop would be five
or six. The actual number chosen would depend on a company's
long term objectives. That is, choosing to manufacture 6
standard sizes would aid the company in lowering material and
labour costs but would require the organisation to improve or A,
maintain a high level of manufacturing efficiency. H. M. Ltd. 's
actual policy, to develop five standard sizes, was adopted
because the limited development time available was the overriding
constraint on the number of standards that could be designed.
The basic frame sizes chosen by H. M. Ltd. are 0.8 tonnes,
1.6 tonnes, 3.2 tonnes, 5.0 tonnes and 8.0 tonnes. The prime
costs arising from this policy is equal to £238,600. The
optimum policy for five standard sizes (0.5 tonnes, 1.25 tonnes,
2.5 tonnes, 5.0 tonnes and 8.0 tonnes) gives rise to prime costs
of £191,000 (Table 15). Hence an additional saving of
£238,600 - £191,000 = £47,400 is possible. by changing. to the
optimum policy. However the range of standard sizes adopted
by H. M. Ltd. was chosen primarily because these lifting capa-
cities are traditionally those required by the market for crane
duty. Hence these sizes are necessary since the EHB is a major
sub-assembly in H. M. Ltd. 's Universal crane range. However the
Minaddition analysis has shown the additional cost to H. M. Ltd.
of adopting a non-optimum policy from which management can
judge the value of that policy.
99
2.4 Design Planning and Control
Since the design of a new product range is essentially a conceptual
process, the design time must be allocated using subjective judgement,
i. e. the design phase is planned by apportioning available design time
amongst the major sub-assemblies that makeup the product. It is
then the responsibility of the engineering designer to ensure that
each sub-assembly is designed within the time allowed. Normally the
design time allocated in this manner is not subsequently distributed
between the individual components that form each sub-assembly. Hence
there is no effective control over the allocation of time to design a
component since a target design time for each component will not have
been allocated.
In addition since target times for components are not allocated
the possibility exists of engineering designers mis-apportioning
available design time between components. Under these conditions
the relative design time allowed for components may not always reflect
the relative importance of the components to the overall success of
the product design. For example, during the design of the New EHB
range, the time expended on minor cost items such as bearing housings,
was not proportional to the items relative contribution to the overall
cost, quality, reliability and producibility of the new product.
To enable design resources to be allocated effectively requires
a project planning technique that takes into consideration the relative
importance of each component within a design.
100
The conventional methods of planning and controlling development
projects are Gantt Charts, Critical Path Method (CPM), Performance
Evaluation and Review Technique (PERT), Management by Objectives (MBO)
and Planning-Programming, Budgeting and Systems Analysis (PPBS).
Gantt Charts, CPM and PERT networks (80)
graphically illustrate
the chronological order in which development tasks should be performed
and are therefore practical tools for monitoring the progress of
development projects. CPM or PERT networks are superior to Gantt
Charts since these methods illustrate the relationships between
project tasks and can therefore determine the "critical path" for
a project, i. e. those activities which cannot be delayed without
delaying the project completion date.
PERT networks are useful for planning and controlling the new
product development process since the uncertainty surrounding development
task times can be taken into consideration. Task times are determined (81)
using Equation 12.
T= act
T, + 4T2 + T3
6
Where:
Tact = Activity time for a task.
T1
T2
= Optimistic time for a task.
Most likely time for a task.
Pessimistic time for a task.
(12)
101
Although the PERT technique was originally designed for
controlling development projects, the use of the technique for
controlling the development of the New EHB range had several
disadvantages:
c>> The cost of implementing PERT is high, hence the project
being planned must be sufficiently large to justify
this expense. Smith and Mahler(82) in comparing
computer CPM and PERT programs indicate the relationship
between number of activities being planned and program
cost. Using this relationship the development of each
EHB component if planned in detail (i. e. approximately
3400 events) would represent a cost of approximately
117,000 for a PERT software package. Purchasing a
PERT system would not therefore be cost effective when
compared with the equal opportunity of using this cash
to increase the size of the H. M. Ltd. design team.
(ii) To construct a PERT network diagram projects must be
comprised of a series of identifiable tasks. However,
since the desigq activity is essentially a creative
process, the tasks resulting from the design phase cannot
always be identified in advance of design being carried
out, e. g. the type and length of reliability trials
depends on the assembly design concept adopted, hence
a PERT network diagram cannot be constructed accurately.
102
(iii) PERT de-emphasises cost (both development and manufacturing
costs) as a specific factor to be considered during a
ol
project since resources are directed at maintaining the
scheduled development time. Although PERT/COST network
diagrams are designed to take financial objectives into
consideration, Kahalas(83) indicates that the technique
is difficult to implement since the allocation of costs
to individual tasks within a project is not easy. That
is, the projects suitable for PERT/COST are generally
"one offs", hence the development of standard costs for
any particular task is extremely difficult and often not
cost effective.
(iv) Network analysis techniques such as PERT are concerned
with controlling development time and the allocation of
project resources to the cost effective minimization of
development time. As such these techniques do not consider
the relative contribution of each project task to the
overall success of the development process. This is necessary
when allocating project resources.
(v) Equation 12 is concerned with determining the time required
to carry out a particular project task. However when
development time is limited the primary aim of a planning
technique should be to allocate development time between
priority tasks such that it is used efficiently.
103
MBO(83) is a planning technique designed to motivate management
at every level of an organisation by involving them in the planning
process (i. e. goal setting) and evaluating them on their effectiveness
in achieving the goals which they participate in setting. Using the
MBO philosophy to plan and control the development of a new product
would force management to establish a consistent and comprehensive
set of development objectives for the organisation and could produce
a higher degree of commitment to the development goals.
However the MBO technique is normally limited by the degree to
which the organisational and the personal goals of individual managers
can be made compatible. This is a major problem during new product
development since managers are conscious that decisions are effective
for the life of the product. The tendency, therefore, with using MBO
is to focus on short term goals that lead to departmental functions
concentrating on individual objectives rather than on organisational
aims. For example, the manufacturing function could tend to choose
materials to minimize labour costs rather than total manufacturing
costs.
PPBS(83'84) was developed for use by the USA Department of Defence
to assist decision-makers in choosing, planning and controlling defence
projects with varying levels of costs and resulting benefits. When
numerous projects are competing for limited development resources a
cost-benefit analysis is carried out to compare the cost of achieving
benefits from one project with the cost of receiving equivalent benefits
from an alternative project.
104
The philosophy of the PPBS technique is applicable to the new
product development process since each component within a design can
be considered a candidate for design time. The design time must
then be allocated to each component based on the benefits to be gained,
i. e. the relative contribution to the overall development objectives.
The need to carry out a resource-benefit analysis requires a
common basis for comparing benefits to be established. However since
the outcome of alternative projects can vary considerably alternative
benefits are often compared qualitatively rather than quantitatively.
In addition projects can have multiple objectives and benefits, hence
the formation of a common basis often depends on the ingenuity of the
analyst in using the information at hand to develop benefit and cost
measures which provide meaningful and comprehensive outcome descriptions
and in addition provide a basis for rational choice amongst alternatives.
The basic benefit to be gained from increasing the level of
design resources allocated to a component is of course an increase in
the overall profitability of a product design, i. e. the total profit
obtained during the total market life of the product. However overall
profitability can be derived in numerous ways, examples of which are
listed in Table 16.
105
TABLE 16 METHODS OF IMPROVING PRODUCT PROFITABILITY
1. Lowering product selling price and therefore increasing
sales levels.
2. Designing a more attractive or desirable product.
3. Improving the delivery time of the product.
-4. Enabling the product launch to coincide with an
increase in sales demand.
5. Widening the range of specifications covered to improve
the potential market.
In examining the means by which overall profitability can be
improved the present work considers that the four basic activities
involved are:
(i) Improving product design specification.
(ii) Improving product characteristics such as style,
reliability, quality and serviceability.
(iii) Lowering the product manufacturing costs.
(iv) Improving the producibility (ease of manufacture) of
a product design.
106
Hence the allocation of design resources must determine the
affect of resource allocation on design characteristics (i. e.
specification, attributes, producibility and cost) and determine
their effect on overall product profitability.
When comparing alternative designs manufacturing cost is the Ile
design characteristic that can be related to product profitability
with accuracy and consistency. Product specifications, attributes
and producibility are difficult to accurately relate to product
profitability since the effect of these characteristics depends on
such subjective criteria as the purchasing motives of potential
buyers. Hence the relative cost of components within a product
design is the most logical criterion to use for allocating development
and design resources.
Although it is feasible to allocate design resources on the
basis of a component's relative cost within a product design, several
factors that nevertheless must be considered are:
(i) The necessity to determine the influence of other
design criteria (specifications, attributes and
producibility) on the level of resources allocated.
(ii) The allocation of resources can only be used for design
improvement, i. e. it is necessary to develop an initial
design solution, estimate the manufacturing costs of
this solution and use these costs to estimate the level
of resources to use for design improvement.
107
(iii) The necessity to develop cost targets for designs in
order to ensure the efficient use of design resources.
Here. target costs could be defined as those costs
beyond which additional cost reductions would not
justify the cost of the extra design resources
required to obtain them.
In practice such cost targets are difficult to assess, hence more
practical cost limits could be obtained using competitors product
prices, expected profit margins required and quality-levels necessary.
Brooks (85) used the concept of target costing to develop a
programme for the control of costs during NPD. In order to provide
management with a usable cost control tool a "cost trend chart" was
constructed that indicated the anticipated growth in product costs as
the design proceeded. An anticipated cost trend allows a greater degree
of freedom in the sequencing of design tasks. In addition cost
variances (the difference between actual and target costs) when plotted
on the chart provide a visual indication of whether total product costs
are within the pre-defined limits. The extent of the variance between
actual and target costs could be used to indicate the type of cost
reduction action required. That is, small differences would tend
to indicate that cost reduction could be achieved by the engineering
designer whereas large variances would indicate that a more radical
cost reduction technique was necessary, e. g. value engineering or
value analysis.
108
If negative cost variances were obtained (i. e. actual costs were
lower than target costs) this could indicate that the amount of time
the designer spends in attempting to improve manufacturing costs could
be reduced, hence the overall design time could be lowered.
It is important during the development of the New EHB to
effectively' plan and control the limited design resources available.
Since existing project planning techniques are not appropriate an
alternative method needs to be developed. Such a planning method
would need to contain:
(i) A method of allowing designers to obtain component cost
data quickly. This data would then be used to allocate
design time.
(ii) A relationship between design time and the relative
cost of components within a product.
(iii) A method of determining component cost targets. A
suitable method would be to base cost targets on
competitor's product costs and expected profit margins.
(iv) A method of monitoring and controlling design time
against product costs. In this case a suitable method
would be the use of the "cost trend charts" proposed (85)
by Brooks
109
2.5 Design Scheduling
Undertaking development tasks in the correct sequence is essential
in minimizing total development time by ensuring that tasks that form
the critical path are not delayed. However during the design of a new
product the critical path may be constantly changing since the design
process may create development tasks that lie on the critical path.
For example, a component may be designed requiring a lengthy delivery
period for raw materials or special tooling hence affecting the
existing critical path, or a component may need to be long term tested
therefore automatically forming a new critical path.
The scheduling of work through the design phase has important
effects on the other departmental functions within the NPD process.
Normally design work is released from the design department in the
form of finished product or major sub-assembly drawings which result
in high work loads being placed on other departmental functions.
For example, the level of resources within a Jig and Tool department
is normally sufficient to cope with the relatively low level. (as
compared with NPD requirements) of short term tooling requirements.
Limited spare capacity is normally available for NPD requirements.
In addition short term tooling tasks are needed to allow day-to-day
manufacturing operations to continue, hence they take priority over
NPD tooling requirements. It is therefore necessary to avoid releasing
high levels of tooling requirements during the NPD process since this
normally overloads the tooling department. Hence to ensure the
minimum delay to the development schedule-work must be sub-contracted
leading to increased development costs.
110
Scheduling design work such as to minimize total development
time is not always the major objective since it is often necessary
to plan work such that total product costs can be estimated. This
is in order to financially justify the development work with the
minimum loss in development resources. In addition the early
estimation of product costs allows control procedures to be established
that would allow costs to be reduced to an acceptable level.
However since development time is limited during the NPD project
the ability to sequence design work could be of significant benefit.
Again existing project planning methods are not appropriate hence an
alternative method needs to be developed. Such a technique would
need to contain a method of allowing designers to determine the total
time required for components and assemblies to be developed and
development task interactions (in order to identify "critical path"
components).
2.6 Alterations to Product Designs
Altering component designs after their release to the prototype
build and test stages of the NPD process, was observed to be a frequent
activity during the New EHB development process. A sample of components
chosen at random were used to determine the total number of design
changes that had occurred in this manner, i. e.
Nt = Na. Nr. Nq (13)
111
Where: I
Nt = Total number of design changes incurred.
Na = Average number of design changes per component.
N -r = Total number of components within the product range.
NA = Average number of design changes per component that
result from design interactions between components,
i. e. caused by Na type changes.
For the New EHB process Na = 4.5, Nr. = 400 and NA = 2.5
hence substituting in Equation 13:
Nt = 4.5 x 400 x 2.5 = 4500
From a sample of experienced designers it was established that
the average time to carry out each design change was approximately
0.50 hours (includes draughting and functional calculations).
Therefore total design time needed to alter existing designs
15:
4500 x 0.5 = 2250 hours
2250 hours represents 17.6% of the total design resources
available for the New EHB project. Hence it would be an advantage
to reduce the number of design changes that take place during the NPD
112
process. This would also reduce the effect of design changes on other
stages in the development process, e. g. prototype testing and jig and fixture build. From an examination of the design changes that occurred
during the development of the New EHB range the major causes of design
changes were identified, Table 17.
i TABLE 17 MAJOR CAUSES OF DESIGN CHANGES
Cause Proportion of Proportion of Total Changes Design Resources
Cost Reductions 0.78
Functional improvements 0.10
Attribute improvements 0.05
Producibility improvements 0.04
Improving specification range 0.03
0.137
0.018
0.009
0.007
0.005
Table 17 indicates that the majority of design alterations were
carried out to reduce product manufacturing costs arising from the delay
in obtaining cost data from the present cost estimating system. ' During
the NPD process designs often had to be released to the prototype build
and test stage to prevent serious delay to the project completion date.
Subsequently when designs had been costed, cost reduction exercises
found to be necessary normally resulted in changes to component designs.
Table 17 also indicates the proportion of the available design resources
required to carry out design alterations.
113
Design changes arising from cost reduction tasks can be reduced
in two basic ways:
(i) Develop a more responsive cost estimating system to
provide cost data quickly in order to allow cost
reduction exercises to be performed prior to the loe
release date of designs to the prototype design,
build and test stages.
(ii) The decision to alter a component design should be
related to the potential benefits that could be obtained.
Only minor savings would be obtained from reducing the
cost of low cost components, hence these savings should
be compared with such disadvantages of design changes as
delay in project completion date, alterations to jig and
fixture design and resources required to improve component
design.
Functional design improvements arise due to prototype testing
revealing design deficiencies. Design alterations to improve the
functioning of a product are therefore a natural part of the develop-
ment process. The small proportion of changes that arise because of
design defects indicates the power of mechanical engineering models; 86'87)
such as bearing life equations, and the experience of the designer in
developing working designs.
114
The proportion of design changes arising from the need to alter
product specifications and attributes could be reduced by ensuring
that market data is collected in a suitable format for the design
department. However a certain proportion of specification and
attribute design improvements could be necessary to counteract
competitive action during the NPD process.
Requests by the manufacturing function to improve the producibility
of designs should take place prior to the release of designs for
prototype build and test. Hence the manufacturing function must be
allowed to examine designs early in the design process.
115
CHAPTER 3
COSTING PRODUCT DESIGNS
116
3.0 Introduction
Costing is essential to the development of effective new product
designs. To determine the manufacturing costs of a product design it
is necessary to identify the relevant cost elements involved and
decide how each element is to be estimated.
Z
Equation 7 is useful to designers when comparing alternative
product designs but since this model contains only the variable costs
of manufacturing it cannot indicate the true cost of a specific
product design to a company. In order to achieve this objective
the total manufacturing costs (which includes fixed costs) must be
obtained.
3.1 Total Manufacturing Costs
At H. M. Ltd. total manufacturing costs are determined using
Equation 14. This model is similar to Equation 7 but in addition
contains an allowance for defective components manufactured and
elements for recovering both fixed and variable overhead costs.
CM =
.. rI --- 1 -1
("im.
im) k(1+Sk) +1+ Om + Ct
m=1
m=n
1+ Oa (14)
117
Where:
Tm
m=n
= Total manufacturing costs.
Sk- = Allowance for defective components.
Production time at the mth manufacturing cost centre.
Labour cost rate at the mth manufacturing cost centre.
Tm. Lm = Cp (See Equation 5).
m=1
Om .= . Manufacturing overhead rate for the m"" operation.
= Administration overhead rate for the company.
3.1.1 Material Costs
The material cost data used within a company is initially
procured by the purchasing department since they are normally
the principal link with material suppliers. Up-dating material
costs can involve a considerable amount of effort on the part of
the purchasing function when product ranges contain many
thousands of different types of raw materials and purchased
components as in the case at H. M. Ltd. However, the effort
involved in up-dating material costs is normally avoided by
assigning each type of component and raw material a standard
cost. Lists bf standard costs for commonly used parts and raw
118
materials are maintained and issued to departments requiring
material cost data. These lists are re-standardised at monthly
intervals using average percent price increases based on such
economic factors as current inflation rates. At intervals
(the length of which is determined by the cost of the raw
material) each raw material is up-dated using actual cost
data obtained from material suppliers.
The use of standard cost data for costing new product
designs could represent a possible source of error since
standard costs do not always reflect the true price of the
materials. In addition standard costs may be based on discount
structures for current order quantities. The manufacture of a
new product could therefore alter these order quantities and
hence the discount rate that was applicable.
During the development of"a new product engineering
designers frequently obtain material cost data directly from
-potential. suppliers. However, the final responsibility for
obtaining cost data must belong to the purchasing department
since such factors as supplier reliability, material availability,
multiple sourcing and vendor rating(88) need also to be considered
when choosing the ultimate material supplier. This task can be
more efficiently carried out by purchasing personnel who possess
the relevant background experience.
119
3.1.2 Scrap Costs
The basic types of scrap costs that must be considered
when costing designs are:
(i) Waste material costs such as machining swarf,
' flash produced during forging and waste material
incurred during flame cutting.
Defective component costs, i. e. components that
are manufactured outside the accepted quality
requirements and must therefore be scrapped. It
is assumed during the present discussion that
defective bought out components are replaced free
of charge.
At H. M., Ltd. the excess cost that arises from material loss
and the scrapping of defective components is compensated for by
increasing standard material costs by a fixed cost percentage,
the value of which depends on the raw material type (e. g. bar,
sheet, casting or tube) and is based on past experience of scrap
levels produced.
However, waste material costs should be determined from the
difference between the initial raw material form and final component
shape since this could provide essential cost data for value
engineering tasks. In addition if the waste material produced
can be used for, other purposes or has a significant scrap value
then this can be taken into consideration when determining the
final material cost of components.
120
Basing defective component costs on material type over-
simplifies the problem since other factors that need to be
considered are:
(i) Component design.
(ii) Batch size.
(iii) Quality requirements.
(iv) Dimensional tolerance requirements.
(v) Operator experience.
(vi) Complexity of manufacturing process used.
(vii) Reliability of production equipment used.
A simple model (Equation 15) for predicting the number of
defective components produced based on the manufacturing batch
size is given by Sims (89):
Ndef. QP
Where:
Ndef.
Qp
Number of defective components.
= Manufacturing batch size.
(15)
121
Sims then uses the "learning curve" concept to take into
consideration the reduction in both labour cost and scrap loss
arising from the "experience effect". The "learning curve" is
a graphical representation of the improved performance (reduction
in the percentage of defective components manufactured) arising
from increasing quantities, skills and experience. The "learning
curve" is exponential in character and is therefore transformed
into a straight line,, Figure 7, if plotted on log-log graph
paper.
According to Sims the reduction in scrap loss arising from
the "experience effect" is taken into consideration by using
Equation 16 to estimate the number of defective components.
Ndef.
Where:
= Sa jQp
(16)
Percentage average scrap per batch for a
specific "learning situation".
The- value of Sa depends on the particular "learning
situation" 'which can be defined as the relative decrease in the
rate (i. e. -learning curve performance) at which defective
components are manufactured. The learning curve performance
is dependent on such factors as the component quality and
122
FIGURE 7 70%, 80% AND 90% LEARNING CURVES
Adapted from Ref. (89)
ae 0 rn
C) kýo
C C C
0 c C
0 c
c CD N
C) N
c C
C' C
aý ý ý U N
O J
ae CC)
aR 0
a N "r
C tn
ý
V
c
C".
0 0 ý
CD CJ
ý (PS) (ale: )s 60l) dea: )s a6eaanya6e; u30J3d
CV
v ý (D . L..
123
dimensional tolerances required and varies between manufacturing
operations. For example machine shop work has a learning curve
performance of approximately 90% and complex assembly work one
of 65% to 85%.
-A comparison, Figure 8, of the scrap proportions (Sp)
produced when calculated using Equation 17 for 90%, 80% and 70%
learning curve performances indicates that only small differences
exist.
. SP = Ndef.
x 100% Qp
(1 7)
In addition these differences decrease with increasing batch size.
Hence a common learning curve slope could be used for a variety
of manufacturing tasks without incurring large errors (relative
to accepted estimating errors) in the scrap proportions evaluated.
Since Sims indicates that complex machine type work usually
develop learning curves of approximately 90% and assembly work
approximately 70% a suitable curve for use at H. M. Ltd. (where
manufacturing operations involve both machining and assembly
tasks) would be the average of these two values, i. e. 80%.
Manufacturing batch sizes for the New EHB product average
approximately 100 components, hence from Figure 8 the errors
in-the scrap levels produced by using an 80% learning curve
124
FIGURE 8 COMPARISON OF SCRAP LEVELS ESTIMATED FROM 90%, 80% AND 70% LEARNING CURVES
i.
ýI ýI^
I II1 -1 LO
I
ct MN r-
(dS) de. A: )S %
125
(as opposed to individual learning curves for specific processes)
would be +2.4% to -1.4%. These error levels are within the
accuracy limits of ±10% specified by management and the actual
accuracy (see Section 4.4.1) of the current estimating system
at H. M. Ltd.
3.1. J' Overhead Costs
Overhead costs form a significant part of the operating
expenses of a company and are recovered using either of two
financial accounting methods, i. e. absorption costing and
marginal costing(90). Absorption costing is the method adopted
by H. M. Ltd. and is carried out by apportioning total overhead
costs to each component manufactured within a company using
suitable overhead recovery rates (Om and Oa, Equation 14).
However, when costing new product designs using the absorption
costing technique problems that exist are: -
(i) A large proportion of overhead costs are fixed
costs and hence do not vary with the volume of
components manufactured. Therefore if a fixed
overhead cost is to be recovered and fixed profit
margins maintained overhead recovery rates must be
increased when the number of components manufactured
decreases. Hence if new product development takes
place when a company's sales volumes are low the
higher overhead recovery rates in operation could
126
effectively "overcost" the new product design. This
effect occurred during the development of the New
EHB range when H. M. Ltd. increased overhead rates
by an average of 135%. However, since the new
product is intended to improve the sales demand a
more realistic product design cost could be obtained
by using forecast sales volumes to determine overhead
recovery rates. This would appear a logical method
since both material and labour costs are also based
on forecast sales demand.
When costing new product designs it is essential that
only relevant overhead expenditures are absorbed by
a product to prevent the true cost of a product to
a company becoming distorted. At H. M. Ltd. this
is particularly important since both standard and
custom-designed products are manufactured, i. e. the
custom-designed products require additional overhead
services (e. g. drawing office, tendering and costing)
which are not necessary for the production of standard
product ranges. Hence using a common set of overhead.
rates for both product types would tend to inflate
the manufacturing cost of the standard products. The
benefit to a company in manufacturing standard
products would not then be genuinely reflected.
127
(iii) If the prime manufacturing costs (labour + material
costs) of a product are reduced through design
improvements the overhead cost recovered could also
be lowered since a lower cost product contributes
less to the overhead recovery of a company. Hence
the design could therefore be considered less
desirable.
Marginal costing is an appropriate method of costing new
product designs. Using this method fixed overhead costs-are
separated from the variable elements of total costs and the
contribution of products to the total overhead costs is
determined. However, when using the marginal costing technique
it is necessary to determine the proportion of total company
overheads that are incurred by the product range being examined
in order to adequately reflect the true financial contribution
of the product. 1 1.
Since overhead costs form a large proportion of H. M. Ltd. 's
annual manufacturing costs, the absorption costing method presently
used by the company enables greater control to be exercised over
the recovery of overhead costs than would the marginal costing
. technique.
128
3.1.4 Direct Labour Costs
The determination of labour costs is dependent on the
accurate estimation of the time required to carry out each of
the operations needed to manufacture a component. Hence estimat-
ing labour costs is synonymous with that of estimating'the time
to carry out manufacturing tasks.
The time required Tm (Equation 14) at the mth manufacturing
cost centre is termed the "floor-to-floor" time (FTF) and is
the time required to complete all machinery operations at a
cost centre. When costing production operations the FTF time
is broken down into elements, each of which have distinct
factors that affect their costs. Blore(91) considers FTF
times to be made up of preparation tasks, process tasks,
process manipulation tasks and component manipulation tasks
whilst Knott(92) considers gauge and/or inspect to be an
additional activity. These tasks are now considered in detail:
(i) Set-up or Preparation Time
The set-up time is that time required to position
the various tools and fixtures necessary to carry
out the manufacturing operations and involves such
tasks as adjusting machine settings and changing
the type of machine jaws.
129
I i
Normally each type of operation that is necessary
to set-up a manufacturing process is given a standard
time. However Owen(93) points out that determining
the process equipment settings, tools, fixtures and
adjustments required to machine a component is not
always simple since the problems of "complete set-up",
in which the set-up is started from a stripped machine
and "adjust set-up", in which the correct tools are
already on the production equipment and only need
adjusting, must be considered.
In addition production scheduling rules, such as the
maximization of machine utilization also affect set-
up times since components are sequenced to minimize
set-up tasks.
Component Manipulation Time
Component manipulation tasks involve loading and
unloading components onto and off the production
equipment and if necessary re-positioning the
component when loaded. The time taken to manipulate
a component is dependent on:
(a) component size, e. g. large components often
need to be loaded and unloaded using mechanical
handling devices which result in manipulation
times being dependent on their availability.
130
(b) design shape, i. e. holding fixtures may be
required to position components on machines,
. 41'
(c) type of operation, i. e. the accuracy and hence
positioning time with which components are
positioned varies depending on the type of
operation to be performed.
(iii) Process Time
The process time represents the time required, for the
production equipment to carry out the processing
tasks, e. g. metal removal, welding, forging and
pressing. The factors that influence process time
are dependent on the process being carried out. For
example, welding time is dependent on welding speed
and welding process; forging is dependent on power
availability; metal removal times are dependent on
speeds, feeds, machine power rating and condition
of the machine and tools used.
(iv) Process Manipulation Time
Process manipulation tasks are those required to
ensure that the processing conditions are satisfactory,
e. g. altering feeds and speeds, positioning tools,
selecting welding conditions and tools.
131
In the present research this element contains the
time required to carry out the quality control tasks
referred to by Knott(92) as separate tasks, e. g.
checking component dimensions and surface finishes.
Quality checks (although often separate tasks) are
an extension of machine manipulation tasks since
they are necessary to verify the effectiveness of
process control activities. In many instances
quality checks also form part of a manipulation
task, for example, when positioning a horizontal
boring tool.
The production time to carry out all activities necessary
to manufacture the component is converted to cost using an
appropriate labour cost rate (variable Lm, Equation 14). Labour
cost rates used at H. M. Ltd. are listed in Table 50, Appendix II.
A factor to consider when calculating production cost is the
number of operatives involved in each operation. That is, when
operators control more than one machine hourly cost rates must
be divided by the number of machines controlled as *shown in
Equation 18:
Llm (1s)
Nm
132
Where: -
Lim = Labour cost rate for the mth cost centre.
= Number of machines controlled by operator.
McGrath and Conlon(94) have examined the "interference
effect" when an operator controls two machines in order to develop
-incentive schemes that ensure the operator has the motivation
to keep both pieces of equipment operating at maximum possible
capacity. The "interference effect" occurs when the operator
through setting-up or operating one machine must allow the other
machine to be idle. It can be seen that this effect would tend
to increase the FTF time of a component. Situations in which
one-operator controls a number of machines are becoming more
common in industry hence the "interference effect" will become
of, increasing importance and subsequently may need to be included
in the model (Equation 14) for estimating total manufacturing.
costs. Since this type of situation is not common at H. M. Ltd.
the "interference effect" on FTF times has been ignored during
the present work.
3.2 Analysis of Cost Estimating at H. M. Ltd.
The 'estimation of manufacturing costs may be considered a "works (95)
system" as defined by Nadler i. e. "a works system is the operational
combination of the human, financial and physical resources which
processes its input material, information or humans to achieve the
purpose(s) of a desired state of a product or service".
133
Recognition of this fact therefore enables the cost estimating
function within a company to be described using Nadler's seven system
characteristics (Inputs, Outputs, Environment, Physical Catalysts,
Function, Sequence and Human Agents). By examining these factors
individually it is possible to determine the role each system
characteristic plays in achieving the overall system objectives.
In addition. relationships or interactions between system character-
istics can be more easily identified. Hence incompatibilities between
system characteristics and system objectives, and among system features,
can be highlighted and if necessary improved.
3.2.1, Function
The function of a cost estimating system is to provide
management with the type, quantity and quality of cost data:
needed to'efficiently plan and control a manufacturing
organisation.
Anthony and Lahiff(96) point out that the quantity of
cost data required by management can be expected to increase due
to the increasing use of sophisticated management science
techniques to plan and control complex operations. A cost
estimating system must therefore, have the ability to respond
to future increases in demand for cost data. This would prove
a particular advantage during new product development which
normally requires an estimating system to quickly generate
large quantities of cost data.
134
An aspect of cost estimating not referred to in published
literature is that of system responsiveness which is the rate
at which the estimating function can provide cost data. A poor
response from an estimating system could lead to delays in new
product development programmes or the inability to react to
quickly changing market forces such as competitors price
reductions. During the design of the New EHB difficulty in
quickly obtaining cost data often resulted in designers making
subjective, and therefore possibly sub-optimum, design decisions
in order to prevent lengthy delays to the. NPD programme.
Consequently a large number of design changes were incurred
during the NPD process.
The accuracy of cost estimates is an essential consideration
when designing an estimating system since this factor represents
an important element of risk in management decision making. The
level of risk acceptable and hence estimate accuracy required is
dependent on the volume of product manufactured within an
organisation. In large volume manufacturing systems changes
in component design or work methods, although leading to small
changes in a components manufacturing cost, could when multiplied
by the large number of components manufactured result in
substantial cost variations. Hence the accuracy of cost
estimates under these circumstances would be a fundamental
consideration and could therefore take precedence over other
cost estimating system characteristics, such as system
responsiveness and estimate preparation costs.
135
The cost of preparing estimates is also an important
consideration since the estimating function is normally an
overhead expenditure that must be held at an acceptable level.
The amount of overhead expenditure that can be justified for
the estimating function is related to the type of manufacturing
system. In large volume manufacturing systems the estimating r
cost per component would be relatively small because of the
large numbers of components made. On the other hand, in small
batch or job shop manufacturing systems, particularly if products
are not of a standard design (i. e. designed to-customers specific
requirements) the ratio of cost estimates required to components
manufactured could be relatively high. Hence the number of
estimates required to control the manufacturing activities could
represent a significant overhead expenditure.
The estimating system at H. M. Ltd. has been designed to
generate the relatively low level of manufacturing time standards
and cost data that enables short term planning and control
activities to be performed. Consequently the estimating system
is not sufficiently responsive to the needs of the NPD process
which requires large amounts of cost data quickly.
The accuracy of cost and time standards data at H. M. Ltd.
is allowed to vary depending on the annual quantity of components
manufactured. Hence the overhead burden of the estimating
function is maintained within acceptable limits.
136
3.2.2 Inputs and Outputs
Input data required to prepare cost estimates is provided
by several functional departments, i. e.
(i) Marketing provides information on annual sales
quantities and long range sales forecasts.
(ii) Design provides general arrangement drawings,
detailed product designs, parts lists and quality
requirements.
(iii) Purchasing provides material/component costs and
quantity discount data.
(iv) Manufacturing provides the production data with which
to estimate component manufacturing times, e. g.
machining feeds and speeds, process set-up times,
and jig and fixture costs.
(v) Finance provides labour and overhead cost rates.
The amount and detail of input data is dependent on the
accuracy of the estimate required. For example, estimating
manufacturing costs within a high volume manufacturing system
would require a detailed breakdown of the operations necessary
to produce a component, whereas for less accurate cost estimates,
which may be appropriate for jobbing and small batch situations,
individual operations may be aggregated hence reducing the
detail of the input data.
137
In addition since the input data used to develop a cost
estimate reflects the manufacturing conditions for a component
it is essential that these conditions are controllable. Any
permanent deviation from the manufacturing conditions used
to prepare a cost estimate would reduce the validity of that
estimate. i
Output data must be provided in many forms to satisfy
the needs of different functions. For example, financial
planning and control uses cost data on a product and
manufacturing cost centre basis, whereas the production,
planning and control department use production time standards,
i. e. the manufacturing times of single or batches of. components.
The designer requires cost data to enable design concepts
to be evaluated. This could necessitate costing in detail
individual design features such as machined bores, chamfers.
and threads.
It is essential therefore that an estimating system is
able to provide cost information in a variety of forms and
detail. That is, on a design feature, process, component,
sub-assembly, product basis or cost centre basis.
138
3.2.3 Human Agents
The human agents (estimators) are an important character-
istic of an estimating system since they must be familiar with
the estimating methods used and types of input data required
to develop cost estimates. Since conventional estimating
methods require a detailed breakdown of the method of
manufacture of components it is usually essential that
, estimators are fully familiar with products, processes and work
methods used within a company. It is normally the degree of
background experience of estimators that determines the speed
and accuracy with which estimates can be developed.
Ostwald and Toole (97)
surveyed the type of background
training given to estimators and concluded that the majority
have no formal training or estimating education. This situation
reflects the current attitude towards the estimating function,
in that estimating is considered "more of an art than a
science"(98). Therefore estimators need to develop
estimating skills through obtaining a considerable grounding
in the manufacturing processes of a company.
Familiarity with the processes and methods being costed
enables the cost of preparing estimates to be reduced since
estimators have the ability to make subjective estimates of
manufacturing costs and hence avoid the time required to
gather detailed input data.
139
Anthony and Lahiff(96) found that additional barriers
existed to the development of reliable estimates. For example,
undue pressures on estimators such as high work loads and
insufficient time allowed to prepare estimates, were found to
lead to inconsistent and inaccurate estimates being developed.
It is necessary to consider these factors when costing new
product designs since the large amounts of cost data required
during the development process often overloads the estimating
department.
Companies such as H. M. Ltd. are unable to cope with the
cost data requirements of the new product development process
" when the estimating function has been designed to provide the
relatively low levels of cost information required to control
day-to-day manufacturing activities. The low levels of estimates
required to support short term planning activities of a company
are insufficient to make an individual estimating function cost
effective particularly (as in the case of H. M. Ltd. ) if a wide
range of manufacturing methods and processes require the
background experience of numerous estimating engineers.
Therefore estimators must perform other duties. At H. M. Ltd.
estimators are also responsible for maintaining short term
manufacturing operations. These tasks normally take precedence
over cost estimating for new product development.
140
3.2.4 Sequence
The number of stages involved in developing cost estimates
can influence the responsiveness of an estimating system. The
time to develop a cost estimate can be considered, using
Queueing Theor3 70), to be the sum of a queueing time (the time
a request for a cost estimate must wait before the actual
task of estimating begins) and a service time (the time taken
to carry out the estimating task). The responsiveness of an
estimating system is the sum of the queueing and service times
and is therefore dependent on the number of stages involved in
developing a cost estimate.
Queueing times are dependent on:
(i) The priority given to requests for cost estimates.
(ii) Human factors such as motivation. 4
(iii) The current work load and capacity of the
estimating function.
Service times are dependent on:
. ,t
(i) The cost estimating method used since this factor
determines the level of detail of input and
output data.
141
(ii) The accuracy of estimates required since this
factor determines the estimating method used.
(iii) The experience and ability of the estimators.
(iv) Capacity of the estimating function.
) biased to Anthony and Lahiff(96ased cost estimates
could be developed if the action that resulted from the estimate
effected the person or department preparing the estimate in
an adverse or favourable manner. Similarly if the action
resulting from not preparing an estimate does not have an
adverse or favourable effect on the estimating function then
this tends to reduce the responsiveness of an estimating system.
For example, since there is a tendency during the design of
the New EHB range for designers to make decisions without the
benefit of cost data, estimators have little incentive in
preparing cost data quickly.
The number of stages involved in developing cost estimates
can be a function of a company's internal management structure.
At H. M. Ltd. the production engineering department determine
manufacturing times for components and the commerical department
convert this data into manufacturing costs. Hence two distinct
queues are formed therefore reducing the effectiveness of the
estimating system.
142
For estimating the costs of non-standard products
H. M. Ltd. have recently altered their estimating procedure to
improve its responsiveness by using the commercial department
to estimate both initial manufacturing times and costs, i. e.
removing the production engineering departmental link.
Although the accuracy of estimates can be adversely
effected (since the actual estimating is no longer carried
out by the personnel most intimately connected with the
manufacturing processes) allowing production engineering
to prepare the total cost estimates was not possible since
the extra work load on the department could not be adequately
handled.
3.2.5 Environment and Physical Catalysts
The environment (working conditions of a company) and
physical catalysts (estimating methods used) of a cost estimating
. system are closely related, since it is the working conditions
of a company (e. g. variety of products, variety in manufacturing
processes, level of planning carried out and level of control
over operations) that determines to a large extent the appropriate
estimating methods used.
Cost estimating methods vary in the manner in which the
time taken to perform a manufacturing task is determined. They
are derived from work measurement techniques which themselves
differ according to the level of detail into which manufacturing
tasks are broken down for estimating purposes.
143
In general the higher the level of detail used to
estimate manufacturing times:
(i) The greater the accuracy of estimates.
(ii) The greater the time required to prepare an
ý estimate.
(iii) The more familiar an estimator must be with the
working methods and processes in use.
(iv) The less responsive the estimating system.
(v) The higher the cost to prepare an estimate.
Since the method of work measurement has a significant
effect on the accuracy and responsiveness of an estimating
system, it is this system characteristic that determines the
-appropriateness of an estimating system for costing product
designs during the new product development process. The
estimating method influences:
(i) The personnel who use the system.
(ii) Estimate accuracy. .
(iii) Time to prepare the estimates.
(iv) System operational costs.
(v) System development costs.
144
System responsiveness.
(vii) Degree of subjective estimating and input data
used.
(viii) Range of manufacturing tasks costed.
loe (ix) Level of cost detail obtained.
(x) Ease with which computer software can be
developed.
3.3 Work Measurement Techniques
It is necessary to measure the work content involved in
manufacturing a component in order to develop production time standards
which can be expected to accurately represent the time an operator
would be expected to take to carry out manufacturing tasks. Time
standards allow operators to regulate their work performance to a
pre-set rate (99)
and therefore facilitates work planning, scheduling
and incentive scheme operation. Since production time standards help
to regulate manufacturing time variability they can be used to
determine the labour cost of component manufacture.
Work measurement techniques are the methods by which standard
production times can be established. Since many factors affect the
time taken to manufacture a component it is necessary to state the
conditions, under which time standards have been developed and
subsequently adhere to these conditions during actual manufacture of
the component.
145
The development of standard production times is therefore
synonymous with the development of standard working conditions.
Defining, standardising and maintaining working conditions can be
difficult, 'particularly in large, multi-product organisations such
as H. M. Ltd., since these tasks are dependent on the level of
management control exercised over the manufacturing operations. Ile
That is, it is necessary to optimize the level of management control
in multi-product companies with its cost. The criteria that affect
working conditions and production times are basically of two types,
i. e. `workplace orientated factors and operator performance factors
(e. g. the pace at which operators work).
Workplace Factors
cooling(100) in discussing management's responsibility in
maintaining the working conditions under which time standards were
developed indicates the types of factors that are important (Table 18).
TAB 18 FACTORS ESSENTIAL TO WORK PLACE STANDARDISATION
Ref. (100)
(i) Material arriving at the work place on time.
(ii) Consistent material quality.
(iii) Consistent quality specifications.
(iv) Well maintained tools and equipment.
146
TABLE 18 (Contd)
(v) The avoidance of incorrect or faulty tooling causing inconsistent work quality.
(vi) Correct positioning of tools, equipment and material
at the workplace.
(vii) Consistent training of operators.
. (viii) Reduction in the variety of methods used, i. e. greater
job specialisation.
(ix) Effective communications.
However Cooling assumes that manufacturing organisations have
sufficient management resources to plan and control operations in
detail and hence avoid problem situations arising. However, in
practice the level of management control is restrained by financial
considerations and work place problems arise as a'matter of course.
The affect of work place environmental conditions, such as,
work place temperature, amount of light available, absence of glare,
noise levels, machine design, layout of control panels on operator
performance and hence production times are the specific province of
Ergonomics and have been adequately examined(101).
147
Operator Performance Factors
Fein(102), argues that the rate at which work is carried out is
not rigidly quantifiable in a measurement sense, since in order to
make measurements of any kind there must be a means of measuring work
rate that is exact, clearly definable and reproducible and a set of
reference standards (normal work rates) against which these measure-
ments can be gauged. Fein argues that work measurement literature
does not sufficiently provide definitive guides for the establishment
of normal (standard) work rates.
Techniques used to. fulfill this reference standard requirement
all have one characteristic in common, i. e. all are based on the
subjective judgement of the analyst. For example, the technique
used in time study synthesis is called performance rating in which
the analyst mentally compares the operator at work against his own
concept of normal and establishes a numeric performance rating factor
which indicates the relationship of the observed work to his mental
concept for that work. In calculating a time standard, the observed
time values are adjusted by the performance rating factors. However,
performance rating is often criticised as being inaccurate. For
example, Owen (103)
examined the effectiveness of present work
measurement techniques in producing realistic time standards. He
found that the inability of work study practioners to adequately rate
work performance was a major cause of loss of management control over
working conditions, leading to the introduction of rates negotiated
between trade unions and management.
148
Mansoor(104) found that the accuracy of time standards
estimations was dependent on the ability of the time study engineer
carrying out the rating. This dependence led Faherty et. al. (105)
to develop a check list, Table 19, of factors essential to successful
estimating which could enable large groups of independent companies
to reduce. the variations in time standards developed by individual i
time study departments.
The difficulty in estimating performance ratings has been largely
overcome by applying statistical techniques to the analysis of time
study data(106,107). An extension of this approach is the development
of Pre-determined Time Standards such as, MTM, in which work tasks
have been broken down into a set of basic motions each of which have
been given a basic time. However, although the application of Pre-
determined Time Standards data eliminates the need for performance
rating, rated performances were used in their development. They are
also only applicable to large volume manufacturing systems in which
work tasks are repetitive.
The individual factors, Table 20, that affect operator performance
have been adequately examined by Archer and Siemens(108) and fall into
four main groups.
149
TABLE 19 ASSESSMENT OF A WORK MEASUREMENT DEPARTMENT
CHECK LIST AND SCORE SHEET Ref. (105)
A. MANAGEMENT X. Does the department occupy a suitable place in the company organisation (i. e. as a
central service not tied to a particular department)?
Wiz. Is the status of the head of the department commensurate with his responsibilities (Le. at least equal to that of the heads of the departments served)?
3. Are the staff deployed to match the production organisation? Ile ¢ Is there a central cell responsible for policy, research, etc?
s. Are staff interchanged from time to time? 6. Is ' there a policy manual? 7. Is there a manual of standard procedures?
U. ` SELECTION 8. Are staff recruited from both internal and external sources? 9. Is selection by single interview, several interviews (1), board (2) or examination (3)?
Io. Is a formal qualification required? (ONC, C&G, AMIWSP, etc. )
G TRALNING II. Are untrained recruits given a formal course of at least 4 weeks? 12. - Is the conduct and syllabus of the course satisfactory? (e. g. to IWSP standards. )
13. Is there a probationary period for trainees of at least 6 months? 14. Are staff sent on continuation or refresher courses-every 2 years at least?
IS. ' Are departmental seminars held from time to time as a regular practice? 16. Is there a departmental conference at least once every 2 years?
t D. PERSONNEL
17. is promotion by seniority, recommendation (t), board (2), or examination (3)?
1g. Are the salary levels reasonable for the type of work involved?
19. Is the staff turnover between ao;; and 5% per annum?
$, WORK NJASUREMEN'T PROCEDURES
20. Are operations method-studied before measurement?
21. Are methods specified in detail when standards are issued?
22. Is maximum use made of PMTS?
23. Is maximum use made of local synthetics, job comparisons, tables etc.?
z. }. Is the BSI rating scale used?
25. Are all elements rated in stop-watch time study?
26. P. re stop-%ratch time studies checked for consistency (1), adequate sample size (z), time elapsed against time recorded (1)?
27. Is activistsa pli snore some ether
systematic means used to provide data regarding contingencies q
28. Are contingency allowances properly defined?
29. Are guidance tables used for fatigue allowances?
30. Are personal allowances defined and standardised?
31. Are the various allowances reviewed periodically?
32. Are standard times issued for setting-up machines?
33" Is due allowance made for the `learning curve' when long runs are expected?
34. Are short run allowances or temporary values used when small quantities are processed?
35" Is analytical or comparative estimating used for non-repetitive work?
Possible Points
2
2
I
2
I
24
I}
2 3 z
3 3 2 2 z 2
2
2
2
4 2 3 4 t ý
3
2
I
2
I
I
2
2 I
3
Score
150
TABLE 19 (Contd)
F. QUALITY CONTROL 36. Are regular rating check sessions held, at least once every 2 months? 37. Where possible, are time studies repeated before a standard is issued:
(a) by another Time Study Man (TSM)? (b) with another operator?
23. Are standards approved by section (section leaders) before issue?
-39. Are studies and/or computations scrutinised by seniors as a regular routine? 40- Do TSM review levels of operator performance in their areas at regular intervals?
G. WORK LOAD 4t. Is the ratio of TSM to operators adequate? 42. Is the number of permanent standards set per week reasonable? (Bearing in mind cycle time, batch sizes and duration of studies. )
FL APPRECIATION 43. Arc appreciation courses given as a regular routine to:
(a) New employees (b) Shop stewards and union representatives (c) Supervisors (d) Managers
SL'ALMARY Management Selection Training Personnel Work Measurement Procedures Quality Control Work J; oad Appreciation General Impression (at Assessor's Discretion)
Possible Points Scor,
5
2 s i I 4
2}
2}
I I I I
II 7
13 7
33 IS 5 4
S. TOTAL I zoo
TABLE 11 SSONTTORL`iG OF A WORK MEASL'RF-NiENT DEPARTMENT
Items to be checked
Number of operators on PBR
Operator Performance (by depts.! product centres)
Number of permanent standards set
Dumber of permanent standards revised
Number of Time Study Men
GraFh of results of rating check sessions
Turnover of Time Study Men
Changes in : Management
Changes in policy
Changes in standard procedures
Total hours recorded on Measured Work
Total hours recorded on Unmeasured Work
Total hours Allowed Time
Y
Frequency I Monthly ( 6-. North
X
X
X
X
X
as received
x
x
x x
x
x
151
TABLE 20 OPERATOR PERFORMANCE FACTORS
Ref. (108)
(i) Job Aptitude - Rote Factor, Comprehension Factor,
Dexterity Factor, Stamina.
(iiY Operator Versatility - Style, Materials.
(iii) Job Responsibility - Repair, Reject.
(iv) Job Conditions - Equipment.
(i) Job Aptitude
(a) The Rote Factor described by Archer and Siemens is
equivalent to the Incipient Learning Stage of the
"learning curve"(109), in which the operator is
becoming acquainted with the set-up, the tooling,
instructions, blueprints, the workplace arrangements
and the conditions of the process. The time required
to memorize the essential details of a job task
depends on the experience and training of the
operator. Carlson and Rowe (109) considers this
stage is also influenced by the "forgetting or
interruption" portion of the learning curve. That
is, when a job is interrupted the amount of "forgetting"
decreases as the number of units completed before an
interruption occurs is increased.
152
It is at the final stage of the learning curve
(the Maturity Stage) where the limit to job
improvement has been reached that the remaining
factors examined by Archer and Siemens begin to
influence operator performance rate.
r (b) Comprehension - How well the job details have been
learned.
(c) Dexterity - The ease with which a trainee is able
to achieve and maintain the standard production pace
for a specified time.
(d) Stamina - Refers to the time required to build up
the stamina necessary to perform the job at the
standard pace for the daily working period.
Operator Versatility
Reflects the time required by an operator to adjust to
the introduction of minor variations in job description
(style) and the effects of changes in materials on
machine settings.
(iii) Job Responsibility
The time required to either reject or repair mistakes in
work operation.
153
Job Conditions
The time needed to repair or adjust equipment. _
The operator performance factors listed in Table 20 indicate that
maintaining a standard work rate over the daily working period could
be difficult. This is supported by Dudley (110) who established that
operator performance is cyclical in character, since performance rate
is low'during the initial work period, rises to a peak and falls off
towards mid-day, with the cycle repeating in the afternoon work period.
The cyclical nature of operator performance indicates that the
use of a single time standard for a specific manufacturing task is
suspect. Again however, the accuracy of time standards needs to be
. compared with the cost of preparing those standards and the consequences
of using an inaccurate estimate.
The problems inherent in estimating reliable time standards, such
as maintaining standard working conditions and operator performance,
can be expected to increase with growing use of manufacturing
concepts, such as Job Enrichment("') and Job Rotationt112), since
the level of work planning carried out is reduced. Job Enrichment
for example, involves manufacturing tasks being designed or redesigned
such that workers are able to use more of their skills and determine
their own work pace and methods. The result is a reduction in
management control over working conditions which could increase the
difficulty in developing standard production times.
154
Within manufacturing industry a wide variation exists in
the level of control exercised by management over working conditions
, and operator performance rates. Tn , consequence a number of work
measurement techniques have evolved with each technique being
suitable for specific manufacturing conditions. Existing methods
of: work measurement have therefore been examined in order to
-identify the appropriate system for use at H. M. Ltd.
3.3.1 Time Study
Time study(113) develops standards by direct observation of
work whilst it is being performed. The method cannot therefore be
used to estimate manufacturing times and hence is unsuitable for
estimating design costs during new product development.
However during the pre-production manufacturing stage of the
NPD cycle, time study analysis could be useful in highlighting
design features that reduce the producibility of a product.
However, the cost effectiveness of using the method needs to be
considered, since time study is a relatively slow and expensive
technique and therefore most suited to high volume repetitive work.
Cost effectiveness in low volume manufacturing systems, as
in th'e case with H. M. Ltd., would be dependent on the possible
savings that could result from the use of time study analysis.
In order to be cost effective:
155
' c-Fs <N (CI-C2)
Where:
.e
Cry = Cost of time study analysis per component.
N= Number of components manufactured.
Cl = Component manufacturing cost prior to time
study.
C2 = Component manufacturing cost after time
study.
(19)
It would appear logical therefore to focus available time
study resources on high cost components since larger Cl-C2
values could be possible.
3.3.2 Pre-Determined Motion Time Standards
Pre-determined Motion Time Standards work measurement
systems (PMTS) are based on the concept that all manual work
tasks can be broken down into a limited number of basic human
motions which when allocated basic times enable total job
times to be built-up from a knowledge of the basic motions
involved. To avoid the need to subjectively rate operator
performance the basic times for each basic motion are
developed statistically.
156
Methods-Time Measurement (MTM)(113), Work Factor(114)
and Maynard Operation Sequence Technique (MOST)(115,116)
are examples of PMTS systems, each varying in the type of
basic motions into which work tasks are broken. Since these
systems are similar in terms of method of application and
-application cost, one such system can be used to determine
the r suitability of PMTS for costing product designs during
the NPD process. MTM has been chosen since this system is
already extensively used throughout the UK.
3.3.2.1 MTM
There are various PMTS systems in the MTM family
which form an integrated work measurement system. The
individual systems I, II, III and V have a specific
relationship to each other in terms of characteristic
features, such as speed of analysis, capacity for
methods description and accuracy of time standards.
Willey(117) comparing the speed and accuracy of the
MTM family by analysing a simple work cycle, Table 21,
indicates that if the measurement system is simplified,
by aggregating basic motions into larger groups, the
speed of application increases and the accuracy of the
time standards developed is reduced, Figure 9.
157
TABLE 21 ACCURACY AND APPLICATION TIME FOR THE
MTM SYSTEMS
Adapted from Ref. (117)
MTM I MTM II MTM III MTM V
% Relative Error (as
compared with MTM I) 0 5.5 12.3 16.3
Application Time (mins) 93 43 15 5
Estimated Job Time (mins) 0.268 0.283 - 0.301 0.312
Accordingly, in deciding the MTM system to adopt
management must balance the economics of greater estimat-
ing accuracy against application time and cost. Considering
H. M. Ltd. 's manufacturing conditions (small batch
production, high level of custom designed products and
a low level of manufacturing planning), MTM V offers the
greatest potential for cost effectiveness since this
system has the lowest application time and therefore
cost.
The comparison of application times for the MTM
family, Table 21, indicates the relationship between
application time and the estimated manufacturing time.
Using the relationship for MTM V the total time required
to estimate the New EHB product design costs can be
determined as:
158
FIGURE 9 RELATIVE ACCURACY AGAINST APPLICATION TIME FOR THE MTM SYSTEMS
Adapted from Ref. (> »)
i
O 0i
0 Co
0
N
ta "E
0) E
F-
oc LC) o
"r 4J
-ra u . '...
CD %: r
C) cY)
0 N
CD ,-
CD N ý .-
N ý
00 v
I W1W 44ýM paaeowoo se AoJa3 anPelaa %
159
Test - Tap'TM
Where:
Test = Total time required to prepare
/ estimates.
Tap = MTM V application time per hour of
estimated time.
(20)
TM = Total manufacturing time for the New
EHB range.
For MTM V, Tap = 16.03 hrs/hr.
The manufacturing costs estimated using H. M. Ltd. 's
existing synthetic work measurement methods have been
converted to labour hours, Table 22. The total number of
labour hours for the complete range is 431.5 hours.
Hence using Equation 20 the total time to estimate the
New EHB manufacturing times is:
16.03 x 431.5 = 6917 hours
Number of man years = Total Application Time (Hours) No. of Working Weeks/Yr x Hrs worked/Wk
6917 4.06 man years 46x37
160
It is considered therefore that the use of MTM V
is not feasible for costing low volume product designs
such as the New EHB. Knott(92) examining the use of
MTM V within a machine shop supports this conclusion.
He states that "MTM V like the other systems in the MTM
family are more suitable to mass production operations
than to low volume or jobbing shop manufacturing
conditions".
TABLE 22 MANUFACTURING TIME FOR THE NEW EHB RANGE
Labour Costs (i's)
General Multi- ac fining Machining. Assembly Fabrication
0.8 Tonne Hoist Block 60.1 7.7 40.3 2.5
1.6 11 if 99 71.5 10.2 44.7 3.0
3.2 80.7 11.5 48.5 5.7
5.0 95.2 15.7 57.9 9.5
8.0 138.7 20.3 69.8 12.1
3.2 C. L. Trolley 30.7 - 6.2 10.7 2.9
6.3 35.7 7.1 12.1 3.9
TOTAL 512.6 78.7 284.0 39.6
Labour Cost Rate (Vs/hr) 2.60 0.85 2.25 2.55
Labour Hours 197.2 92.6 126.2 15.5
Total Labour Hours = 197.2 + 92.6 + 126.2 + 15.5 = 431.5 Hours
161
3.3.3 Category Estimating
Category estimating divides the time range distribution of
manufacturing tasks into a sequence of non-overlapping intervals,
as illustrated in Figure 10. Manufacturing jobs are then
allocated to an interval and assigned a single time standard
associated with that interval. Hence it is not necessary to
. estimate an, individual time standard-for a manufacturing task.
A simple approach, Table 23, developed by Boehringer(118)
uses an interval chart representing 50 equal time spaces. Since
estimating time standards merely requires an estimator to use
personnel judgement to decide which interval the job time falls
within, the ease with which work measurement can be carried
out makes category estimating suitable for low volume manufacturing
situations.
More sophisticated category estimating systems have been
developed using statistical analyses of job time distributions.
Secor and Kogovsek(119) and Steele(120; 121) assume that the
range of job times within a company's manufacturing processes
, are normally distributed. However job times examined at
H. M. Ltd. for welding (122), turning, milling, gear cutting,
assembly and fitting, horizontal boring and flame cutting
exhibit a log-normal distribution..
162
FIGURE 10 MANUFACTURING TIME BANDS
tKtuV=Nl-r
U
C)
() 5-
LL-
(1)
-, ý F--
-0 O 7
17
SLOT 3
SLOT 4
Ref. (g1)
SLOT 5
SLOT 6
Range of Job Times within a Company
163
. TABLE 23 50 INTERVAL ESTIMATING CHART
Ref. (118)
�
Std item/min 5.87-6.14
5.60-5.86
5.35-5.59
5.11-5.34
4.88-5.10
4.66-4.87
4.45-4.6S'
4.25-4.44 AG
IS i
*. 064.24
3.88-4.05
3.70-3.87
3.54-3.69
3.38-3.53
3.22-3.37
3.08-3.21
2.94-3.07
2.81-2.93
2.68-2.80
2.56-2.67
2.45-2.55
2.34-2.44
2.23-2.33
2.13-2.22
2.03-2.12
1.94-2.02
Pcs/hr 10.0
10.5
11.0
11.5
12.0
12.6
13.2
13.8
14.5
15.1
15.8
16.6
17.4
18.2
19.1
20.0
20.9
21.9
22.9
24.0
25.1
26.3
27.5
28.8
30.2
Std item/min Pcs/hr 1.86-1.93 31.6
1.77-1.85 33.1
1.69-1.76 34.7
1.62-1.68 36.3
1.54-1.61 38.0
1.47-1.53 39.8
1.41-1.46 41.7
1.34-1.40 43.7
1.28-1.33 45.7
1.23-1.27 47.9
1.17-1.22 50.1
1.12-1.16 52.5
1.07-1.11 55.0
1.02-1.06 57.5
. 974-1.01 60.3
. 931-. 973 63.1
. 889-. 930 66.1
. 849-. 888 69.2
. 811-. 848 72.4
. 774-. 810 75.9
. 739-. 773 79.4
. 706-. 738 83.2
. 674-. 705 87.1
. 644-. 673 91.2
. 615-. 643 95.5
. 587-. 614 100.0
164
Recent developments in category estimating have recognised
the log-normal distribution of job times. Kirby(123) outlined
a procedure for applying the category estimating technique to
log-normal. distributions in setting time standards for non-
repetitive jobbing work, Figure 11. Harris and Melven(124)
have used this technique for estimating time standards for
tool repair tasks.. The procedure for developing a category
estimating system outlined in Figure 11 requires management
to choose the number of categories in which the job time
distribution should be divided, Figure 10, since the width of
the time bands determines the individual accuracy of the time
standards developed. According to Fairhurst(125) it is
management's responsibility to optimize the widths of the
time bands and estimating cost since when system accuracy is
increased the cost in terms of planning time also rises.
Although category estimating is suitable for developing
time standards in small batch manufacturing systems it is generally
inappropriate for costing new product designs because:
c>> Its use requires a detailed knowledge of manu-
facturing processes and methods. Hence estimators
need to be fully familiar with the manufacturing
conditions used within a company. Therefore it
is suitable for use by foremen and chargehands
but not by engineering designers.
165
FIGURE 11 OUTLINE PROCEDURE FOR APPLYING THE
CATEGORY ESTIMATING TECHNIQUE
Ref. (123)
t
oe
5a) Take further movie
6aI TA* fun. lw movie
6W Repeat 6
_ 6c) Check cortgatabdlity
of samples rjý iý2 Szý
ýSý ` Ef, Efý
6d) Add sump4es ar+d repeat 5&6
tie) Repeat 6abcd until sample is big enough
a
Group job durations into clash which
tuocessirely increase in range by a Common factor (e.. a 21.
Convert boundaries to togs and calculate x Sic. 5.7 t
Use Xa ten to c hm* goodness of fR
to 'tog normal'
Check accuracy /Avrnoc S <E4vTaos
i
of sample size t4
10 Calculate mid-values for each category by working through probabilities at msd-boundanes and de. iatgns.
It. Calculate standard time for Nwage
job in each category by use of data from 7 and 8 and relaxation allowance.
12. Set up work standards on master heel
11. Supervision ssumatt job eategar. es before they are don..
14. Obtaat sample of ategwtsed cards
IS. Venfy eslomating in ample by Comparing with work standadi
14 Calculate performance by canoarn. q total standard fours (from ess. mated Categories) with total clocked Wur%
17. Prrpue incent. ve Scheme based on lxrformance.
166
(ii) The method is not designed to provide accurate cost
estimates for individual jobs but an accurate
average time over large numbers of parts.
(iii) Detailed manufacturing data is not used to establish
time standards therefore detailed cost breakdowns of
product designs could not be developed.
(iv) A new product introduction could significantly alter
the existing job time distribution of a company.
Hence basing component costs on an established job
time frequency distribution could give rise to
inaccurate cost estimates.
3.3.4 Synthetic Time Standards
Synthetic time standards (113) estimating is an established
method of cost estimating within low volume manufacturing environ-
ments. The method is similar to the use of PMT standards in
that'manufacturing tasks are broken down into basic elements
and the elements assigned basic times. However synthetic
time standard elements are specifically related to a company's
products and manufacturing activities and therefore unlike PMT-
standards are normally only suitable for use in the company for
which they are developed.
The benefit of using synthetic standards arises since
elements chosen as standards can be any combination of individual
work tasks that are common to a variety of manufacturing processes
167
and products. In order to develop a synthetic standards
estimating system it is necessary to identify, define and
measure a large number of work elements. Definition of the
type and extent of work tasks included in the work element
is essential in ensuring estimating accuracy.
. ol
Synthetic time standards data is presently being used
., at H. M. Ltd. to cost product designs. However,, the application
time for estimate development using synthetic time standards is
similar to that of MTM V. Consequently estimating is a slow
process and cannot cope with the large amounts of cost data
required during the NPD process.
In addition estimators must'be familiar with the manufacturing
processes: and methods, thus it is more difficult to improve the
responsiveness of H. M. Ltd. 's estimating system.
3.3.5 Work Measurement using Computers
The initial development of work measurement software
programs was directed at using the fast processing speed and
high accuracy of computers to achieve the advantages highlighted
by'Ke11er(126):
(i) To increase the speed at which estimates can be
provided.
168
(ii) To relieve cost estimators of the routine part
of work measurement.
(iii) To eliminate human errors that occur when large
amounts of data are handled manually.
Initially, therefore, software packages were developed
around conventional work measurement systems. For example,
. Kopp(127) describes the structure of a computer system for
improving time study synthesis. The computer is used to collect,
store and process time study data, in the rapid development of
time standards. Since the basic time study technique remains
unchanged the computer system examines only existing work
methods and like manual time study synthesis cannot estimate
time standards for the new product development process.
Since the estimation of time standards using conventional
work measurement techniques (e. g. PMTS) requires methods planning,
early attempts to develop software for these techniques required
extensive amounts of input data to define the work tasks being
timed. Software was limited in the type of work'tasks that
could be estimated. For example, Mason and Towne(128) developed
ARMAN, a computer work measurement system using MTM I time
standards, to measure the work content of complex electronic
maintenance tasks.
169
The American "WOCOM" production planning system(129ýl30)
includes both MTM I and Work Factor work measurement modules.
To reduce the amount of input data required to define complete
work, tasks the modules operate on two levels which are the
"motion" level using input symbols that correspond to Work-
Factor or MTM I motions and the "task" level in which motion
patterns and work place layouts for common operations can be
retrieved from the computer memory. - The work measurement modules
are applicable to a wide section of industry and can be adapted
to different computers including time sharing devices.
Although the speed of estimating is increased, through the
use of computers, work measurement modules based on PMTS standards (131)
cannot estimate labour time standards within job shop or small
batch manufacturing environments since the detail of work
methods description is normally too imprecise. That is, jobs
are not planned in sufficient detail to enable MTM-I work
measurement to be performed.
Steele (120) recognising the imprecise nature of methods
planning within low volume manufacturing systems, developed work
measurement software (based around MTM-2 data) for measuring
fitting and assembly work. Again however it is unlikely that
this system would be suitable for use at H. M. Ltd. since to
estimate time standards for non-repetitive work tasks Steele
resorts to a category estimating technique.
170
Computer work measurement systems have gradually evolved
since the late 1960's to become an integral part of much wider
production planning systems. The range of manufacturing tasks
covered by such systems is diverse and includes manufacture of
cardboard cartons (132)
, storage tanks (133) and machining
(134,91)
The systems differ according to their accuracy, ability to lo,
optimize production plans and the level of subjective judgement
, required by the estimator in order to develop time standards.
In terms of ability to optimize, Halevi and Stout(135)
developed a software planning procedure for specifying machines,
machining operations and metal cutting criteria in order to
optimize the machining of turned components. Although the
software was limited to optimizing the "turning" process the
number of variables, Figure 12, affecting the optimization
process resulted in a large amount (80K bytes) of core storage
being required.
A recent software system developed by Logan (136) succeeds
in reducing the level of operator interaction required to
develop work plans and hence time standards. The system,
LOCAM, consists of a data base containing all the relevant
information about a company's manufacturing facilities and a
manufacturing logic program which is designed to extract data I
171
FIGURE 12 MACHINING TIME
PARAMETER RELATIONSHIPS
94VE OF PLAT
POWER
I \NN. :/
`'`, ` . ýýý
I
XIOL TYPE ý
SURa6CE FlN1SH
DE p 1). 4 CUT-17N. 7. CF Ni FOPCE
E14T ý-"ý i "".
CVrr . wG
dýr. s'E 4 'CRSAOw
FC; K--E ý
ý ýý
CW7(! +G K'Simq /\\ / CVTTwG
ý; K--E
F=ED OE7i+-- -- ------- GEP'M /
R<iF OF CUT OF CU'
Ref. (135)
/I IY. Or
// Cra. Cýc: vG
CwU; ER ///
LOCA'b+ Of Ga1Cxl, {
LE OF CUt
T9lFR WCES
C7a°Or. ýT Mn[ul\ýN ý
AE`. iU, aFý, cti'S
172
from the data base, manipulate it and perform calculations
with it. In essence LOCAM simulates the logic process which
the planning engineer would follow through in his mind. LOCAM
has been designed for use by planning engineers and is therefore
unsuitable for use by engineering designers.
loe
3.3.6 Work Measurement Algorithms
The estimation of time standards can be performed by
developing algorithms that express the causal relationship
between manufacturing tasks and the factors that effect these
tasks. Although the development of estimating models can be
difficult and may require computers, their use provides the
following benefits:
(i) The cost of estimating is reduced since estimates
can be developed quickly.
(ii) The level of subjective judgement required to
develop an estimate is reduced.
(iii) The need to hold extensive manufacturing time data
files is avoided.
Many factors affect manufacturing task times and the
difficulty involved in constructing estimating algorithms arises
due to the difficulty in establishing the effect of individual
variables on the manufacturing times. There are four basic
173
methods of constructing estimating algorithms (i. e. synthetics,
empirically based studies, use of operational research
techniques and the use of statistical techniques) each of
which is capable of developing models with acceptable accuracy
levels.
3.3.6.1 Synthetic Techniques
The development of estimating models synthetically
requires the use of human judgement to alter both variables
and the method of combining variables until an equation
is arrived at that produces time standards of acceptable
accuracy. The method therefore relies on the skill and
judgement of the personnel constructing the model.
A critical examination of the model synthetically
derived by Mahmoud(137,138) illustrates the problems
inherent in using subjective judgement to develop
estimating equations. Mahmoud developed a "universal"
model (Equation 21) for estimating the costs of turned
components.
1 C=
Bi . Nd Kc. Ll . Dm +
St +k (21)
mnf 60 4.77 , BL. Qp
J
174
Where:
Cmnf = Manufacturing cost of the component being
turned.
Bi = Hourly labour rate (including Factory
Overheads). /
Nd = Component complexity factor.
Kc = K1. K2. Kmat.
K1 = Machining factor.
KZ = Machine type factor.
Kmat = Material factor.
LI = Total machined length (inches).
Dm = Mean diameter (inches).
St = Machine setting time (mins).
B2. = Machine tooling capacity.
Qp = Batch size.
k= Material cost.
Equation 21 is similar to the model used to estimate
manufacturing costs at H. M. Ltd. (Equation 14). The
expression Nd (Kc.
Li. Dm +
St represents the 4.77 B2. Qp
floor-to-floor time of a component.
175
The component complexity factor, Nd, is calculated
by determining the number of discontinuities on the machine
surface and is, according to Mahmoud, a measure of the
component design complexity. Mahmoud states that Nd
represents "... the probable number of passes of the: tool
over the machined surface. Therefore the more irregular
and interrupted the surface, the greater the number of
tool passes required to produce the component, i. e. it
is therefore indicative of the amount of work done on
the component". However the method of calculating Nd
(i. e. the types of discontinuities considered, Table 24)
varies depending on the value of (Li. Dm)and the type of
stock material (e. g. bar, casting, fabrication) the
component is manufactured from.
!
By varying the method of determining Nd, the equation
may be designed to enable estimated component costs to fit
the observed costs rather than to express the exact causal
relationship between component costs and their influencing
factors. It is possible therefore that variables which
are not causally related to the manufacturing costs can
be included in the estimating model, hence poor estimating
accuracy can result.
This argument is supported by examination of other
variables within Equation 21. For example, St is assumed
to be related to Nd. That is, the number of surface
I 176
TABLE 24 CRITERIA FOR ESTABLISHING Nd
Ref.. (138)
Machining Feature Discontinuity Rating
1. Number of set ups 1 for each
2. Unmachined surface 0
3. Machined surface 1
4. " Threaded parts - if unmachined before threading 1
- if machined before threading, e. g. turned or drilled 2
- if machined before threading, e. g. Drilled and bored s' 3
5. Shoulders or faces - if ` 1- Y6- 0
- if > 1" 1
. 16
6. Chamfers - if
- if> 1" . _ý 00, ts
7. Radii - if ý1" 8
8. Drilled holes
.x
0 1
0
1
1
-if bored 2
- if bored and reamed 3
9. Bored holes - previously cored or cast 1
10. Reamed holes - previously drilled or bored 1
11. Recesses - if < 1" wide
- if> 1" 4
1
3
177
features is related to the number of tools used to form
those features. Although this could apply there are
instances when a single tool is often used to machine a
number of design features, particularly during the turning
operation, hence St could be independent of Nd.
According to Mahmoud(138) St must be corrected by
multiplying by Nd . Again however the methods of calculating B2
the actual value of Nd vary: Bs
(i) When Nd > Bz then Nd =i
Using this rule to B2
determine the setting time correction factor
does not take into consideration cases in which
single tools machine numerous surface features
and therefore Nd must be less than one to B2.
reduce the setting time to the correct value.
(ii) When Nd. < B2- then St is reduced by a factor
Nd . This rule does not consider components Bz that require a different tool for each surface
feature.
The variable Ki is dependent on the dimensional
tolerances required. If dimensional tolerances are
tightened Ki increases to reflect the increase
178
in machining time experienced due to the need for
smaller depths of cut, increased number of machining
passes and use of slower machining speeds. However
the use of a single variable to account for dimensional
tolerances required is unsuitable since KI for components
with various dimensional tolerances, in which each design
feature is machined to different dimension tolerances,
is difficult to determine.
Of
The problem involved with using variables that are
not truly causally related to component manufacturing
times is clearly illustrated by Table 25 which compares
machining times estimated using Equation 21 and the
actual observed times. Poor estimate accuracy is
indicated.
TABLE 25 A COMPARISON OF ESTIMATED AND OBSERVED MACHINE TIMES
Estimated Times Observed Times % Equation (H. M. Ltd. Time Study Data) Error
297.2 290.7 + 2.2
35.4 39.0 - 9.2
55.0 81.7 - 32.7
69.0 48.0 + 43.8
45.7 53.8 - 15.1
261.9 61.4 +326.5
52.1 52.6 - 0.9
179
TABLE 25'(Contd)
Estimated Times Observed Times % (Equation (H. M. Ltd. Time Study Data) Errors
115.8 78.0 + 48.5
15.0 66.2 - 77.3
,, 147.0 182.9 - 19.6
35.2 31.1 + 13.2
25.7 33.0 - 22.1
9.5 18.0 - 47.2
102.8 90.0 + 14.2
22.3 37.5 - 40.5
65.1 34.3 + 89.8
14.8 8.0 + 85.0
1.2 1.5, - 20.0
2.5 12.0 - 79.2
4.7 14.0 - 66.4
2.3 4.2 - 45.2
19.7 13.5 + 45.9
Estimating Accuracy: E=-O. C8
S=0.55
Equation 21 could not be used to estimate New EHB
manufacturing times since the variability in the accuracy of
estimates developed is not within H. M.. Ltd. 's current accuracy
limits of ,S=.
'O. 314 (See Section 4.4.1).
180
3.3.6.2 Empirically Derived Equations
Empirically derived estimating equations of the form
shown in Table 26 are developed under "laboratory"
conditions. That is, the variables affecting manufacturing
times are individually controlled therefore allowing the
r exact influence of each variable on the total manufacturing
times to be determined.
The majority of empirically derived equations estimate
machine process times (e. g. metal removal times for turning,
boring, milling and drilling) since these operations are
applicable to a wide range of industries and the factors
that influence machine process times can be easily varied
with respect to each other.
However empirical models are valid only under the
exact conditions for which they were derived. For example,
although Equation 22 is used to estimate the time required
to carry out turning operations, its use with Numerically
Controlled (NC) machine tools is, according to Moore(139)
inapplicable since NC machine tools are provided with a
constant cutting speed facility enabling optimum cutting
conditions to be maintained. Hence a linear relationship
between the spindle speed component (used to calculate
cutting speed) and Dl(Equation 23) does not apply.
181
The cost effectiveness of using empirical models
for estimating is questionable since the variables used
to determine machine process times in Equations 22 to 26
are themselves dependent on a wide range of factors.
i TABLE 26 EQUATIONS FOR ESTIMATING METAL
REMOVAL TIMES
Turning (Manual Operation) Ref. (140)
tm = n. Dt. Lz. Vc-1. Fl-1
Where:
tm = Metal removal time (mins).
DI = Diameter of work surface (mm).
(22)
Lz= Length of surface to be machined (mm).
Vc = Cutting speed (m/min).
F1; = Feed rate (mm/rev).
Turning (Numerically Controlled Machines) Ref. (139)
Lm = -ff. (D1-dt). (F1. Vc)-1. (D1. dt). (D1+d1)-1
Where:
di = The turned component diameter (mm).
(23)
182
TABLE 26 (Contd)
Drilling Ref. (141)
tm = L3. VS-1. Fj -1
Where:
Ls = Length of hole (m).
VS = Spindle speed (rpm).
(24)
Face Milling (Reciprocating Cutting Operations Ref. (142)
tm = 63. Lz. F2- -1 . (V1 + V2-1 ) (25)
Where:
B3 = Width of surface to be machined (mm).
Fi
vi
= Feed per Stroke (mm/min).
= Average forward cutting speed (m/min) .
V2 = Average return cutting speed (m/min),
Gear Cutting (Roughing) Ref. (143)
tm =nt. Mod. (Fw + 20). 600-1. F3. Fß. Fv. Fu. Fm (26)
Where:
RE = Number of teeth.
M= Module (mm).
183
Fw = Face width (mm).
F3 = Tooth umber coefficient.
Fß = Helix angle coefficient.
Fa = Coefficient of the strength of the material.
Fu = Machinability coefficient.
Fm = Coefficient for influence of machine size.
r
Figure 12 illustrates the process parameters
influencing material machinability. Since the relation-
ships-between the variables shown in_Fiqure 12 are
difficult to derive empirically the estimator must resort
to the use of reference tables such as Table 27, with a
consequent reduction in estimating speed and therefore
increase in estimating costs.
The accuracy of empirical estimating models is
dependent on the degree of control exercised over machining
conditions, material specifications and quality.
Bhattacharyya and Coates (145) found, during an examination
of machining conditions used within industry, a wide
variation in both the machining characteristics of
materials (both between and within batches) and machining
parameters used (feed rates and cutting speeds). In
addition Bhattacharyya and Coates established that in
124
TABLE 27 SPEEDS AND FEEDS FOR DRILLING
OPERATIONS Ref. (144)
cn) ö _ ° [o I '
-
'H I I' I ý
to Q r-
I ä
N O ý`
cJ ö fý ý'ý
°l O ýý
rý SI rl $ ý
4
oý OI
_I 1`)ý
of ln
ový
In N I N ap
- I ý
( LSv - -, --
Q
-- ý .p ° "ý oý
Cl o äý
, CIC g to
.o a
.° ? o( p N Y ° !
"J O ý pý O In ° tn
f a O A Ca
° 1ý ao
, nýýt , tn
0 ÖI O ° a ,, '0
ý ý
t
.ý
ý
=,. J
' O1 D
ö O oý ° oý
4 oi o)
O o!
r1 0ý ý
0
n1 p a0 -4 0
1 !
ö ; ý ö
ö fi O
l ln . aT $
O :.:. 'I
O Ö
O) i
O ? ̀ ' F p Q; D ý( - --ý---
l _
° L' in
; .0 ö
"0 c
I s , 1, °
c_J ý .0 l °{ cc O'
10 lI to I .a 1n; ý I ý s
_' Ji -. 0 ý I ä In
ý
ý o . . + ; o ý- od (An
ý caI öý _
si ý cr1 C-4
O ýN OI ,,, O Q ; r ý
ýýý T LIL L DI ý
d \
: o .ýýs ý o
ý
tn Oc o 9 ( O
ö`ý'i 0 ýC
- 4 oI cº~ iO ýº aI of
` Iv ý ( I r
--
, \ n
-5 G .o
Ir o of
o sl .o
r
ýi o o
kö 1O I r:
W, ýn O o, ýi ol ý oi ýgl tc
cx ý ! ä'
( V? ý u; l
°co' '
04 O N' "ý'
P-. GI °-
; . ;, . , . r
. ; - . - `` cc , OI oc ai ^- ý: i 00
00 . ný i of ý a; gl °;
ý n^ ! "
o! o! o -0 ip c: I `a 4! o ý'' N ý
ý ro o J ý . °! ° _ ý, , z _ , o !
, . , ;,. ý ':
IA\; pi °i O ^'
p1 ö O! GC
T ° ,Z
ö CI
ä ö1 ö N
!o IO °
f -i .0 L`1
°ý ýi
ö C
', ý' I J:
01 1-; a! N r? t0! Q ! ln) 1OI V1 OI " I 2i `D O V' ^ý " ý? °
l! 1 ýO r O
ý-l I lI
r -D
1 Öt äý ö_ý ö O o. °i ^?
N ^ _
ý\
0±N OO
ýa ° O;
' r"4' '01
0: rl Oi -$ ý!
ý;
; "D ' ,. r f e: Vl ; 0! r) Oý Oý C1: :( O( Q c , ý; ä"ý oi ý? i
Or
o} lrl
ä
0 :Oiý
ý 1Iö + ý ... , _ 1 ý,
ý a;,, 1 ö
0 of 1 Öý -1) 1 rJ i0 °ý a, o
0'
Ll c c
_' Ci
, 4: C röo oj ä: 8i nt
( ý
1 öý
01 $;
S .;
-+ o O
:0N f60
fi o ,o "r r ý fr ti : ý ,
J !1"
ý. o ý o ý ýn ý
o 1
a l n
1
1 46
ý t !
M 40 i
L I
185
60% of the samples studied the selection of machining
parameters was controlled by the operator. These results
indicate that in the majority of companies the use of
empirical modus to estimate manufacturing times could
be suspect since insufficient control is exercised over
the machining conditions used on the shop floor and the i
specification of raw materials. However the increasing
use of NC machine tools would effectively improve the
value of empirical models for estimating since:
(i) Machine process parameters can be closely
controlled.
(ii) Equations for estimating the manual content
of machine processes (i. e. loading, unloading
and tool manipulation tasks) are not required.
In addition specific equations can be developed for
optimizing machining operations, e. g.
(i) Cost optimization of the turning operation(146).
(ii) Optimizing machining conditions with sales
demand (147)
(iii) Selection of the optimum machine tools (148).
186
3.3.6.3 Operational Research
The operational research techniques that are suitable
for developing estimating equations are:
(i) Simulation. - Manual simulation is idhntical to
, or the synthetic development of estimating
equations in that both variables and method
of combining variables are altered until an
equation is obtained that produces accurate
time standard estimates.
The use of a computer to carry out the simulation
technique greatly improves the accuracy of the
models developed since a larger number of
variables, variable combinations and actual
time standard observations (used to construct
the estimating model) can be considered.
(ii) Queueing Theory - Queueing Theory has been used
by the author (122) to examine the waiting times
of welders using overhead cranes to manipulate
heavy fabrications. The data obtained allowed
the indirect costs of the handling content of
welding to be quantified.
187
11,
(iii) Linear Programming - Observations of
manufacturing tasks can be formulated into
a series of simultaneous equations (constraints)
which when solved using linear programming (LP)
yield the relationship between manufacturing
variables and manufacturing times.
Ballard and Hichkad(149) and Oliver (150)
formulated LP models of the form shown in
Equation 27:
=n Aq. Xqj + Bi .- E2 = Yi
Where: ý
Aq
Xqj
(27)
= Estimated time coefficient to. perform
one unit of the qth work element.
Frequency count of the qth work
element in the jth operation.
Eij; Ezj = Error elements which represent the
amount of time per unit of the jth
operation requiring an adjustment in
the total time for that operation,
i. e. these elements are the times
either unaccounted for or in excess
and result from the chosen set. of
work elements.
Yý = Total standard time per unit of the
nth operation.
188
To solve Aq > 0, Xqi > 0, Ei J>0, Ezi > 0, Yi >0
The objective of the LP program involved the
minimization of the sum of the error elements
E1j and E2j. The result is then a set of time
coefficients, Aq's, for each work task. i
The benefits of using LP to construct estimating
equations are:
(i) Management control data can be introduced
into the model in the form of extra constraints
that influence the time coefficients.
(ii) Negative coefficients for Aq's are avoided.
3.3.6.4 Multiple Regression
Multiple linear regression (151) is essentially an
extension of the "line of leas squares method", i. e.
simple linear regression, a statistical technique used to
establish the relationship between an independent variable
and a dependent variable. Multiple linear regression (MLR),
however, determines the effect of more than one independent
variable on a dependent variable using a first order equation
of the following type as a model, i. e.:
Y= X0 + x1 Zl + X2Z2 + ... . XnZn (28)
189
Where:
Y= The value of the dependent variable.
X0'=A constant equivalent to the intercept
on the y-axis of a straight line
graph. or
Xi, X2... Xn = Coefficients of the independent
variables Zl, Z2... Zn respectively
and represent the individual effect
that each independent variable has
on the dependent variable.
When used to develop estimating equations, the
regression technique minimizes the sum of the squares'of
the differences between estimated and actual values of
the dependent variable, Figure 13. Estimates can
therefore be derived from the equation of the "least
squares" line through actual observations. The regression
technique guarantees that no other line through the
actual observations of the dependent variable would
have a smaller sum of squares as illustrated in Figure 14.
It therefore yields an unbiased and efficient estimate of
the coefficients of the independent variables with which
to predict the value of the dependent variable, given
the values of the independent variables.
190
FIGURE 13 METHOD OF LEAST SQUARES
Ref. (114)
Z
Dependent variable
berm
Constant
Vzrahla
m si -minimum
1
i-1 --m
FIGURE 14 COMPARISON OF AVERAGE AND REGRESSION LINES
Ref. (114)
14 ý
12
10
8 'ý= 1 ..,; 6
RANGE OF DATA
i
ix . - ý
0 ý? 'ý i ý:, _. :
2I ýCaE'"i'
0
x
x
e 1ö 4yý :]E! 11ý 1$ ý
I-c. Y. S x Ma
191
Although a linear model is produced (with or
without a constant term Xo) it can be used to predict
values for a dependent variable either when it expresses
the exact functional relationship between a dependent
variable and independent variables or when it is an
acceptably accurate approximation of a more complex Ile
relationship, e. g. a non-linear relationship.
Using the multiple linear regression technique
enables data to be used from an existing system (while
it operates) to develop the regression model. This therefore
removes the need to generate data from designed experiments
in which independent variables have to be controlled -'
individually to determine the influence of each on the
dependent variable; When developing regression models
for estimating production time standards, data can be
obtained from direct observation of manufacturing tasks
or previous estimates if sufficiently accurate.
However the effectiveness of the regression technique
depends on the assumption of "cause and effect", i. e. that
the independent variables chosen actually influence the
dependent variable being examined. The regression model
cannot distinguish between a natural relationship or one
occurring by chance.
192
r
Crocker (152) observed, "the ease with which non-
existent relationships can be found is surpassed only by
the ease of missing important ones". It is, therefore,
essential to choose independent variables with care.
The use of personnel familiar with the work being studied
can aid in the identification of independent variables.
MLR in Cost Estimating
The regression technique has been known to be applicable
in the field of production time standards estimation for over
20 years, although the applications for which the method has
been used, though diverse, have been limited in number.
Nevertheless an examination of available literature is
useful in highlighting critical factors that influence
the development of estimating models.
Canning (153) used the regression technique to establish
models for estimating the man-hour requirements for a jobbing
shop assembling transformer cores. Regression models that
gave an acceptable degree of accuracy were developed for
individual sections of the job shop and for each product
made within a section. The work undertaken by Canning
illustrates how the accuracy of estimating models can be
improved. By establishing models for each section of
the job shop and for each product type made within a
section both the range of variablity of assembly times
193
and the number of possible variables affecting assembly
times were reduced. Reducing the assembly time
variability effectively, lowers the maximum possible
error obtained from the estimating model and therefore
improves estimate accuracy.
i In addition if the actual causal relationship between
dependent and independent variables is non-linear, a number
of estimating models may be used to improve estimate
accuracy, as illustrated in Figure 15.
Canning effectively reduced the number of variables
that influenced assembly times by removing the effects of
factors such as:
(i) Variation in production methods between
stations.
(ii) Variation in production methods between
products.
(iii) Variation in productivity levels between
stations.
(iv) Variation in productivity levels between
products.
194
FIGURE 15 EFFECT OF INCREASING THE NUMBER OF REGRESSION MODELS FOR A SPECIFIC DATA RANGE
Ref. (154)
/
ý
... C)
-0 t0
S.. tC
ý
C) . O, C' C)!
C)
4- O
C)
10.
Independent Variable Values, (xý, x2, x3, n)
A
Bi , B2, B3
the regression model for complete data range, max. error = e2
the regression models for sections of the data range,
max. error = el
el < e2
195
Cochran (155) states the limit to reducing data
variability in this manner is the point where the errors
obtained from using the average value of the dependent
variable are acceptable and the use of a regression model
would not specifically improve this accuracy. In practical
terms, however, the number of models developed would be a
management decision since the greater the number of models
developed the higher the application cost. Hence application
cost would need to be optimized against estimate-accuracy
in determining the number of models-to develop.
i
Salem(156) determined the indirect labour time
required to load pallets with packed material using the
predictor variables; number of cases packed per pallet,
number of orders per pallet load, weight of packed material
per pallet load and the volume of cases per pallet load.
It was found that the number of cases packed and number of
orders per pallet load were the best measures of job time,
whilst other- predictor variables did little to enhance
the model. The criterion for choosing predictors was that
residual standard deviation resulting from the regression
model should be minimized., A high inter-correlation
existed between the two remaining variables (i. e. 'number
of cases and number of orders per pallet load) which
bought into question the need for both variables within
the model. Therefore Salem used Student t-tests to
determine which of the two variables should be dropped
196
from the equation. These tests indicated that the
coefficient of the number of cases packed per pallet
had the best statistical precision and in addition gave
an adequate estimation of packing times per pallet.
Doney and Gelb(157) used the multiple linear
regression technique to estimate the cycle times of
turned and bored components on an automatic lathe.
Component design features selected as having the greatest
effect on cycle times were stock outer diameter, material
machinability factor, overall length of the component,
finished turned outer diameter, hub outer diameter and
bore diameter. Several variables when transformed however
improved the regression model:
, or
Variable 1= Stock outer diameter xn (29) Material Factor
Variable 2= Stock OD - Finished Outer diameter , (30)
2x Material Factor
Although the model developed gave low percentage
estimating errors (+2.3% to -2.6%) the variability in cycle
times used to construct the model was narrow (1.2 minutes
to 2.4 minutes). Regression equations can only be assumed
to be applicable within the data range from which they were
constructed. Hence-the regression model developed by
197
Doney and Gelb would be limited to a narrow data range.
It is essential therefore when developing regression models
to provide actual data points to limits that would include
all work likely to be encountered.
An important advantage of using MLR to develop
estimating models is the wide range of manufacturing
tasks for which the technique is applicable. For example,
MLR has been. used to develop models for estimating time
standards for cost control and scheduling of Flexowriters
in a sales order office (158),
setting incentive payments
for vehicle deliverers (159), estimating tooling costs
(160,161)
developing gross time standards (162)
and for estimating
assembly and fitting times for heavy, custom-designed (154)
cranes .
r
Logan (136) recognised the cost-effectiveness of using
regression based estimating systems. Regression analysis
was used to reduce thousands of individual time standards
to simple formulae hence improving the estimating applic-
ation time, removing the high level of skill and background'
experience required to develop time standards and therefore
easing the problem of training new estimators.
Software for carrying out the MLR solution technique
now forms part of several work measurement computer systems,
e. g. WOCOM(129), LOCAM(136) and REAP(91) where the
198
technique's ability to reduce individual time standards
to estimating models ensures that these systems are
commercially viable in jobbing shop and low volume
production systems.
3.4 Selection of Work Measurement Method for Design Costing at H. M. Ltd.
When selecting the most appropriate work measurement system for
developing an estimating system capable of costing designs during the
new product development process, several methods can be ignored:
(i) Time Study since this method cannot predict time
standards.
(ii) PMTS - since these time standards have been
shown to be suitable in high volume
manufacturing systems and are not
appropriate to the small-batch manu-
facturing environment of H. M. Ltd.
(iii) Comparative - since this method cannot estimate
Estimating accurately the individual manufacturing
costs of components.
Synthetic - since these are difficult to develop and
Algorithms prone to high inaccuracies.
(v) Operational - normally operational research techniques
Research are used for optimization problems, hence
Algorithms their use to develop estimating equations
199
(v) Contd ... - could prove costly, i. e. the computing
cost to determine optimization solutions
is in excess of that required to find the
relationships between dependent and
independent variables.
Estimating systems used in low volume manufacturing systems
normally employ both synthetic time standards (for estimating manual
task times) and empirically derived algorithms (for estimating machine
process times). Therefore, the problem of selecting a suitable work
measurement method for costing design involves determining the
suitability of using either multiple regression models or synthetic
time standards and empirically derived algorithms. Synthetic time
standards and empirical estimating models are the basis of the present
estimating system at H. M. Ltd. and cannot handle efficiently the
costing requirements of the New EHB development project. Since it
would not be cost effective to increase estimating resources over
the present level, these work measurement methods must also be
discounted.
When examined with respect to the estimating system requirements of
the new product design process, the development of equations using
regression analysis appears possible, i. e.:
(i) Accuracy - The level of accuracy obtained using
regression models can be improved by reducing data
variability, i. e. increasing the number of equations
developed. Accuracy therefore is a management decision
since the number of equations developed must be optimized
against system operating cost.
200
The present work develops estimating equations within
the accuracy limits of H. M. Ltd. 's existing cost
estimating system (Section 4.4.1).
(ii) Setting-up Cost- The cost of developing regression
equations can be expected to be low since existing
time study data can be used, i. e. it is unnecessary
to design and carry out a costly time study programme.
(iii) Operating Costs - The cost of using such a system would
be low since time standards retrieval from extensive
data files would be unnecessary enabling estimates to
be prepared quickly.
(iv) Responsiveness - The responsiveness of the system would
be dependent on the personnel who carry out the
estimating. The present work establishes an estimating
system capable of use by design engineers, therefore
reducing the effect of queueing time on system
responsiveness.
(v) Type and Detail of Cost Data - Previous literature has
shown the diversity of manufacturing tasks which have
been estimated using regression models. Therefore it
can be expected that regression equations could be
developed for the wide range of processes used at
H. M. Ltd. and at various levels of cost detail.
201
(vi) Subjective Judgement - Since design engineers are to use
the estimating system, the level of subjective judgement
i
and consequently degree of familiarity with manufacturing
methods and processes required to develop time standards,
must be minimized. This factor is dependent on the
types of predictor variables chosen and methods of
quantitatively evaluating these variables.
202
CHAPTER 4
COST ESTIMATING SYSTEM DEVELOPMENT
203
4.0 Introduction
The development of a cost estimating system that uses. regression
analysis to develop estimating equations necessitates defining the
manufacturing tasks to which each model relates. This allows estimators
to identify relevant equations and the conditions for which specific
equations jare valid. Factors to consider are:
. (i) The ease with which estimating equations can be accessed
It is essential that equations can be located quickly since
both the wide range of manufacturing processes carried out
at H. M. Ltd. and the need to achieve estimating accuracy
levels within the current levels operating at the company
(i. e. E=0.096, S=0.314, see Section 4.4.1) requires
developing a large number of equations (>200). The system
used to code or classify estimating models must allow the
retrieval of the small number of equations required to cost
each design to be carried out quickly.
(ii) The number of predictor variables within each estimating
equation
This number should be minimized where possible to reduce the
effect of interactions between variables and therefore help
to develop estimating models of greater accuracy. However,
reducing the number of variables per estimating equation
effectively increases the number of equations within the
total system. Hence it is necessary to balance estimating
accuracy with system size.
204
(iii) The type of factors chosen as predictor variables
Since the estimating system is being developed for use by
engineering designers it is necessary to minimize the effect
on manufacturing cost estimation of those factors for which
data is inaccessible to designers. Such factors would
include machine process variables (e. g. machining feeds,
speeds and depths of cut) since the generation of this
type of data lies outside the normal experience and
training of engineering designers.
(iv) The level of. detail of cost data that can be provided by the estimating system
The system needs to estimate the cost of machining individual
design features in order to allow designers to carry out
detailed cost comparisons of alternative designs.
4.1 Classification of Estimating Models
Initially an examination was carried out to determine the
feasibility of using existing component coding systems to classify
cost centres. Coding systems. such as VUOS0(163), Opitz(164) and
Brisch(165) enable design features to be coded and therefore
could possibly enable estimating equations to be quickly
referenced.
However existing systems are primarily concerned with either
the rapid retrieval of drawings or selecting groups of similar
components with respect to production requirements (166,167). As
such they have not been specifically designed for cost estimating
205
purposes. Since GombinskV(165) points out that the use of
systems of classification for purposes other than those
intended is not always possible it appears unlikely that
existing systems would be suitable for costing purposes.
This assumption is confirmed by work carried out by Remmerswaal (1
and Schilperoort68) and Bloreý91).
i
Remmerswaal and Schilperoort evaluated component coding
systems in terms of their ability to provide the type of
information required to enable various departments to function.
They found that although classification systems provided
information for planning and scheduling, the manufacturing
time standards used were assumed to be known. Blore(91)
abandoned the use of existing coding systems during the
development of the REAP system (Rapid Estimating and Planning).
During the present work the classification of estimating
models required the exclusion of variables considered inaccessible
to designers since this would then-overcome the need to request
data from other functional departments and therefore improve the
response of the estimating system. Accessible data is considered
to be that which is involved in the design decision making
process, e. g. component dimensions, design features, tolerances,
materials and the manufacturing operations required. Inaccessible
data is associated with the manufacturing input to the estimating
task, e. g. machining speeds, feeds and manual manipulation tasks.
206-
Examination of relevant machining literature (169) indicates
that process variability is caused by:
(i) The process being carried out, e. g. machining,
welding and flame cutting.
(ii) The operation being performed, e. g. turning, boring
and milling.
(iii) The size and power of the machine used.
(iv) The manual work content involved.
(v) The tolerances and surface finish required.
(vi) The material being processed.
4.1.1 Manufacturing Processes and Operations
The manufacturing processes carried out at H. M. Ltd.
have been identified and classified for ease of access into
four main groups:
(i) Pre-machining - processes used to prepare material
for subsequent machining operations. Examples
are sawing, shearing, flame cutting and
fabrication.
(ii) General Machining - those processes, such as
turning, boring, drilling and thread cutting,
that can be carried out on a wide range of
machine types, e. g. capstan lathes, centre
207
lathes, bar lathes, horizontal borers,
numerically controlled lathes and vertical
borers.
(iii) Special Machining - those processes that are
carried out on specially designed machine
tools. Examples are gear cutting, milling,
radial drilling, spline cutting, cylindrical
grinding and surface grinding.
/
(iv) Post Machining - processes such as painting
and assembly.
Estimating equations were developed for each operation
carried out within a process group. The process and operations
for which estimating equations were developed are listed in
Table 28. This table also includes other manufacturing
cost items for which estimating equations were developed.
4.1.2 Size and Power of Machines
At H. M. Ltd. "General Machining" and gear generating
operations represent approximately 70% of the machine shop
capacity and are carried out on a range of machine sizes.
The variation of power input of machines results in a broad
range of process variables being used.
In order to effectively remove the influence of machine
process variables on manufacturing costs it is necessary to
divide the range of general machining and gear cutting
208
TABLE 28 CLASSIFICATION OF COSTING EQUATIONS
PRE-MACHINING
Flame Cut
Saw
Shear
Weld
GENERAL MACHINING
Turn
Face
Drill
Bore
Ream
Chamfer
Radius
Groove
Circlip Groove
Thread
Tap
Rope Groove
Recess
MANUAL TASKS
Load/Unload Machine
Machine Manipulation
Machine Set-up
SPECIAL MACHINING
Gear Cut
Mill
Cylindrical Grind
Surface Grind
Spline
Radial Drill
POST MACHINING
Assembly
OTHER COSTS
Milling Fixtures
Drill Jigs (Normal)
Drill Jigs (Box Type)
Copy Templates
Plate Gauges
Plug Gauges
MATERIAL COSTS
Rigid Steel Joists
Rigid Steel Girders
EN8 Round Bar
EN8 Plate
Hollow Bored Bar
209
machines into groups. In this way the machine process
variability within a group does not have a significant
effect on manufacturing times.
Machines that performed general machining operations
were therefore grouped (according to machine size and power
availability) by machine type, i. e. i
Bar Lathes
Bar Capstans
Chuck Lathes
Chuck Capstans
Centre Lathes
Horizontal Borers
Vertical Borers
Copy Lathes
NC Bar Lathes
NC Chuck Lathes
Within each group of machines there is insufficient
variation in machine size and power to significantly influence
the accuracy of cost estimates. These factors can therefore,
be ignored.
Although gear cutting equipment varied by both size
and type an examination revealed that the particular
equipment on which a component is machined is dependent
on the number and module of gear teeth.
210
Estimating equations could therefore be developed
for specific size ranges of spur and pinion gears
(Appendices IV and V) enabling the variability in machine
process factors to be effectively removed. It was thus
not necessary to classify the gear cutting equipment used
at H. M. Ltd. r
In addition the material type from which the gear or
pinion was manufactured was found to be related to the
module so that a Material Factor equation (see Section 4.1.5)
was not required.
4.1.3 Manual Tasks
The model developed by Mahmoud(138) does not estimate
separately the manual content involved in machine processing.
The usefulness of this model is therefore limited. For
example, the manual work content removed by process
automation cannot be determined. The present work therefore
was designed to enable manual task times to be estimated
independently of machine process times.
Halevi and Stout('35) determined manual task times
independently of machine processing time but as a proportion
of machine process time. The present author considers this
approach unsuitable since manual tasks times could be
dependent on factors other than the machine process time.
211
Tool manipulation times for example, are often dependent
on the distance moved between machining operations rather
than on the time spent machining.
When classifying manual tasks a detailed breakdown
of each operation, such as illustrated in Table 29, was
° considered inappropriate since a knowledge of methods
planning would then be required for their use. The method
adopted by Blore(91) of grouping manual tasks according to
their main objectives (i. e. machine set-up tasks, machine
manipulation tasks and machine loading and unloading tasks)
has been used in the present work since sufficient detail
could be provided for design costing purposes.
Although Knott(92) regards"inter-process measurement"
of components as a further manual task objective, for the
present work this form of measurement is regarded as an
integral part of machine manipulation` tasks. The low
level of quality control checks required during the
machining of components at H. I. I. Ltd. also allowed these
measurement tasks to be included in the machine manipulation
tasks without loss of estimating accuracy.
The manual content involved in gear cutting operations
represented less than 5% of the total operation time and
was therefore included in the machine process task time
since a significant loss of accuracy would not be incurred.
212
TABLE 29 MANIPULATION TASK TIMES FOR A DEVLIEG SPIRAMATIC JIGI4IL
S.
Ref. (144-)
List of Elements B. Ma.
Load and Unload Load/Unload Component to Table Up to 50 lb to Over ". 50 lb 10 cwt 10 cwt
- . 35 2.43 Position on Table or Fixture (No True) . 40 Position on Table (True by Clock). 2.33 Position on Table (True by Scribe Line). 1.62 Tighten/Release Gland or Clamp (1'Nut or. Screw) . 58
Machine Manirulation Fit/Remove Tool. to Flash Change
... . 25 Fit/Remove Tool to Quick Change* . 20 Position/Retract to Cut . 28 Index. Table . 13 Change Feed or Speed . 16 Position to Axis (Tape or Manual).
. 52 Apply Tapping Grease . 14 Deburr by Hand
"13
Retract Spindle Move In Cut. . 59
Clear Workplace of Cuttings . 26
Blow Cuttings Out of Bore . 14
Set Depth of Cut' . 29
Set 'X' Axis to Zero 9.49 Set 'Y' Axis to Zero 3.49
Gau Micrometer - Internal . 32
- External . 31
Rule . 20 Depth Vernier . 45 Caliper Vernier . 41 Plug Gauge . 30 Caliper - Internal . 32
- External . 35 Screw Gauge - Plug . 46 Air Gauge - Plug
B. Ms. = Basic Minutes
213
However, the number of estimating equations within the
system would be reduced hence lowering operating costs
and increasing the speed at which estimates could be
prepared.
4.1.4 Component Tolerances and Surface Finish
Separate estimating equations were developed for
those machining operations with dimensional tolerances of
<+ 0. lmm and those operations with tolerances of >+ 0.1mm.
This took account of the variation in machining speeds,
feeds and depths of cut caused by differences in dimensional
tolerance requirements.
Components requiring a surface finish of <2. Oum center
line average and dimensional tolerances of <+ 0.05mm did
not influence the variability in General Machining process
factors since these tolerances were achieved by a subsequent
grinding operation. In this case a grinding allowance of
+0.20mm with tolerances on the allowance of s+ 0. lmm had
to be left on the work during the General Machining operations.
4.1.5 Material Processed
Obviously machining times are influenced by the ease
with which a material can be machined. This property of a
material is termed its "machinability". Various criteria
can be used to quantitatively measure a materials
machinability.
'Of
Of the common criteria (170,144)
used to define
machinability the metal removal rate per horsepower is
suitable for use during the present work. This criterion
indicates the relative metal removal rates for different
materials and can therefore be related to machining times
and hence manufacturing costs. The relative rates of
metal removal are determined under constant power
conditions. These conditions exist during the present
work due to the method'of developing estimating equations
for specific machine sizes.
Using metal removal rate per horsepower data obtained
from the P-E Consulting Group(144) materials were classified
according to their relative machinability (Table 30). Since
this classification is not influenced by the specific
manufacturing conditions of machine shops it can be used
during the present work. To take into consideration the
effect of a materials machinability on machine process
times, the machining times estimated using the regression
equations must be factored using the Material Factor
equations listed in Appendices IV and V.
4.2 Data Collection
The collection of data for developing the estimating equations
initially required predictor (independent) variables to be chosen
and a suitable data source identified.
215
TABLE 30 MATERIAL FACTOR (MACHINABILITY) GROUPS
Ref. (144)
Material Factor Group
MF1
i
EN9 EN18 EN36C EN43B EN24T
EN15 EN19 EN36T EN43C EN5B
EN16 EN19C EN39A EN45
EN16J EN23 EN39B EN56A
EN16S EN24 EN42 EN56B
EN16T EN36A EN43 EN58
EN16U EN36B EN43A EN58A
High carbon, Alloy and Stainless Steels
MF2 EN6 EN8A EN14A EN20B
EN6A EN8B EN16B EN22
EN7 EN8D EN16H EN33
EN8 ENGE EN19A EN34
SAE 8620H
Medium Carbon Steels
MF3 Cast Iron
MF4
Cast Steel
ENIA EN3A EN32A
EN2 EN3B EN32B
E7112A EN3C EN352
EN2C EN8t4 EN355
EN2D EN32
216
TABLE 30 (Cont'd)
Material Factor Group
MF4 Free Cutting Low Carbon Steels Cont'd
Low Carbon Steels loe Nodular Iron
MF5 PB1 LMG LM9
PB2 LM4M LM9M
PB3 LM4WP
Phosphor Bronze, Leaded Bronze
Gun Metal, Hard Brass, Soft Brass
Free Cutting Brass
Aluminium Alloy Castings
Plastic Materials
217
4.2.1 Selection of Predictor Variables
Selecting predictor variables is a critical part of
the regression procedure since the variables chosen
(Appendix IV) are assumed to have an actual causal effect
on the-dependent variable. Although this assumption can
often be proven by empirical evidence obtained from
laboratory studies of manufacturing processes (e. g.
Equations 22 to 26), many situations exist where this
is not possible. Examples are, welding, machine
manipulation tasks and tooling costs. Hence care must
be taken when choosing independent variables. During
the present work predictor variables were identified with
reference to personnel, such as foremen and production
engineers, who were familiar with the manufacturing tasks
being examined.
Estimating accuracy and system complexity were
improved as follows:
(i) Various types and combinations of predictor
variables were examined to determine the most
accurate estimating model for each dependent
variable. For example, it was found that during
the development of estimating equations for
boring operations that the use of length of bore
and diameter of bore as separate predictor
variables gave more accurate results than using
the product of these two dimensions.
218
. -Ir
(ii) When estimating models gave unacceptable error
levels both the independent variables and the
data source used to develop the models were
re-examined for suitability in estimating the
dependent variable. For example, the use of
H. M. Ltd. 's standard welding time data was
found unsuitable and other more reliable
sources of data had to be found (see Section
4.2.2).
(iii) When small differences existed in accuracy
levels between alternative equations, consider-
ation was given to the standardization of
predictor variables between equations and
hence the simplification of the system. For
example, a number of estimating equations were
developed for estimating the times to turn
components on a range of machine types, i. e.
bar lathes, bar capstans, chuck lathes, chuck
capstans, NC lathes and centre lathes. For a
small proportion of machine types using the
product of the dimensions "turned length" and
"depth of cut" as a predictor variable gave
the greatest estimating accuracy. However,
"turned length" and "depth of cut" were used
as separate predictor variables since this did
not seriously impair estimating accuracy but
allowed the predictor variables for turning
times to be standardized between all machine types.
219
4.2.2 Data Source
A library of measured time study data was. held by the
production engineering department. This data was collected
by carrying out observations on operators with rating
performances of 80 on the 0/100 rating scale (113). This
pace is slower than the average performance on the 0/100
scale, i. e. the average rate at which a large group of
qualified and motivated workers could be expected to
achieve. However it represents the standard rate at which
operators can be expected to work under -the manufacturing
conditions at H. M. Ltd. These conditions (low volume
manufacturing) result in operators frequently having to
become familiar with the production requirements of new
jobs. Hence overall work performance is lowered.
In order to represent current operating conditions,
and hence allow this data to be suitable for the present
study, a compensating rest allowance (CRT) (113) was added
to all manual task times. This allowance represents the
non-working time required to take account of such factors
as operator fatigue and personal needs. The standard
CRT allowance used at H. M. Ltd. was 17.5%.
Manufacturing tasks for which time study data had not
been collected by the production engineering department
were dealt with in the following manner:
220
(i) Sawing, Spline Cutting and Shearing Operations
The standard production times issued to the
of
shop floor for control purposes at H. M. Ltd. are
termed Sales Target Times (STT's) and have been
estimated using synthetic work measurement
techniques. These times were used to develop
estimating equations for sawing, spline cutting
and shearing operations since:
(a) Lack of effective control of job time
recording by shop floor personnel resulted'
in recorded job times having little
comparison with the actual times taken
to perform tasks.
(b) The STT values used were initially
examined by production engineering
personnel and considered appropriate.
(ii) Numerically Controlled Profile Flame Cutting
An existing procedure employed by the estimating
department (Table 31) for developing flame cutting
times was adapted (Appendix IV) for use within tie
estimating system. The adaption involved no loss
of estimating accuracy over the existing method
and still allowed the individual tasks involved
in flame cutting to be costed separately.
221
TABLE 31 TIME STANDARD ESTIMATION FOR NC FLAME CUTTING AT H. M. LTD.
OPDE2. No. TA? E No
Nd. ý ..
SETuP.
A* No aF TAPES X- S. 00 shtis =
No CP NOZZLE . SET? tNGS
_x 3.00 sds =
PýoEtLý
C. LOAD FZ. ATE C ST= -1 S. OO sýts -
. L' uNtvý-. -: ALJ x Is. Go D FLATS TutCKNESS C mm]
=E No OF PIZEHEATS x CDI__, x*0.10 F &U CLCIM. OF PP. rFtLE1 CMl x CV, ) VS=
G UN1L0i1D COWONEN TS Y. 2. oo srns=
&tt? NtNG VALUES. M VJofZK VALUES.
Tl-iK. ZNtMý LMn-t/nuN
SM VALUE SET UP
6 C55 1 . 850 A+B ID 560 2.080 12 510 2.313 VALUE ISSUED = 15 405 2 "Eeo I9
. 305 3.855 RZOFILE
2-2. 305 3. $S5
25 305 3"B55 C+E+F+G= 32 255 4. co2b 35 255- 4.626 VAL. 'UE tSSuED
50 2o-3 5.7ý2 75. i50 7.7t0
SM's = Standard Minutes
222
When flame cutting is used to prepare edges for
welding then the time estimated for initial
flame cutting of component shape from the
stock material must be doubled to obtain the
total time required.
i (iii) Welding
The method of estimating labour time standards
(STT's) for welding tasks at HJI. Ltd. involved
calculating the total joint length to be welded
and applying a single rate factor. Since the
influence of factors such as, type of joint,
welding position, electrode wire diameter,
handling time and plate thickness was not
taken into consideration, the use of STT times
for developing regression equations was considered
inappropriate. That is, STT times would be
inconsistent with the actual work content
involved. Examination of welding data confirmed
that similar fabrications had widely varying
STT's issued (up to 200% difference).
The approach adopted by the present author
initially involved developing models to determine
the relationship between the "basic arc time
standards" issued by the Welding Institute (171)
and weld design factors (see Variable Description
Notes, Appendix IV). Since the "basic arc time"
223
O
for a weld is that amount of time the arc is
actually burning, it is necessary to add on a
time allowance for such tasks performed by the
welder as weld joint preparation, manipulating
the fabrication to gain access to the weld joint
and adjustment of protective clothing. In
addition since the standards issued by the
Welding Institute are "basic times" they must
be converted to "Standard Times" (113) by adding
appropriate relaxation allowances.
The'basic arc time" is related to the total
time required to weld a fabrication as shown in
Equation 31(172,173)
Or = Tarc
Twe1d (31)
Where:
OF = The operating factor or duty cycle.
Tare = Basic arc time (mins).
Tweld = Total fabrication time (mins).
From Equation 31 it*can be seen that the duty
cycle is the proportion of the welders time in
which he is actually welding or depositing weld
metal. Duty cycles are influenced by the working
conditions in the welding shop, the welding
224
process used, the type of work involved and the
help given to the welder in the form of unskilled
labour. The actual influence of these factors
on the duty cycle can only be measured by time
and motion studies. A previous study(122) of
the welding of standard fabrications at H. M. Ltd.
indicated that the average duty cycle was
approximately 0.35 for a combination of 80%
Gas Metal Arc welding and 20% Manual Metal Arc.
However, an average duty cycle for a range of
fabrications is unsuitable for the present. work
since duty cycles can be expected to vary
between fabrication types and welding processes.
/
Duty cycles must thus be related to specific
design features. The principal features that
influence duty cycles are listed in Table 32.
TABLE 32 DESIGN FACTORS AFFECTING WELDING DUTY CYCLES
(i) The number of planes that welding is carried out on.
This factor determines the number of times the
fabrication must be re-positioned to enable the
welder to gain access to the weld joint.
(ii) The number of individual weld lengths. This factor
determines the number of weld start locations and
ancillary operations.
225
TABLE 32 (Cont'd)
(iii) Location of welds and their ease of access.
(iv) Shape, size and weight factors that influence the
time for manual or mechanised manipulation of the
fabrication.
The primary process used to weld standard
fabrications at H. M. Ltd. is semi-automatic
Gas Metal Arc welding and is therefore the only
welding process considered in the present work.
The range of the duty cycle for the Gas Metal
Arc welding process is considered to be 0.25 to
0.60(172). Although Myers(173) considers that
for small amounts of welding the duty cycle could
approach 1.0, a figure of 0.6 is considered during
the present work to be a practical upper limit
since the maximum relaxation allowance that
could be added is. 35%. _
To this figure must
be added allowances for assembly and tack welding
of the fabrication and equipment handling.
Although actual time studies are required to
accurately estimate the influence of the factors
listed in Table 32 on the welding duty cycle
subjective estimates (using experienced personnel)
are considered during the present work to be
sufficiently accurate for estimating purposes.
226
(iv) Chamfers, Circlip Grooves, Radii and Tapping
/
Little variability was observed in the machining
times for these features. Cochran(' 55) points
out that in such circumstances, adequate
accuracy levels can be obtained by using the
average of the operation times.. Hence regression
equations were not developed for these operations.
4.3 The Multiple Regression Procedure
Ezekiel and Fox (174) and Hawkins (175)
provide methods of
carrying out the regression procedure (i. e. developing estimating
models) using a desk top calculator. However, manual techniques
rapidly lose their effectiveness as the number of independent
variables within a regression model increases. In addition
when large numbers of estimating models need to be developed
the use of a desk top calculator becomes both tedious and time
consuming.
However since multiple linear regression is a standard
statistical procedure used in a wide range of business and
research applications computer software packages are available
for most micro-, mini- and main frame computers. The regression
software packages vary in the number of independent variables
that can be dealt with and the type and range of supporting
statistical data produced.
227
The regression package at the Loughborough University
Computer Centre is a module of the Statistical Package for the (176) Social Sciences (SPSS) and has been used during the present
work since the package:
/ (i)
(ii) provides a wide range of supporting statistical data
(see Table 33).
(iii) combines standard and stepwise multiple linear
regression procedures which enable considerable control
to be exercised over the inclusion of independent
variables in the final model, and
(iv) has the facility to test independent variables for
inter-correlation and exclude variables from the
can deal with 50 independent variables and hence
the maximum number of independent variables (i. e. 20)
expected to be required to develop estimating
equations,
model-on this basis. This overcomes the need to test
independent variables for inter-relationships prior
to running the regression procedure.
The stepwise regression procedure is suited to the present
work since this procedure allows those independent variables to
be chosen which provide the best prediction model possible with
the fewest independent variables. That is, the method recursively
constructs a prediction equation one independent variable at a time.
BEST COPY
AVAILABLE
Variable print quality
I
228
TABLE 33 SPSS OUTPUT FILE
I
M n
10,
LL
6
r N
M M
ý
C K
(. ý
ý
I-
C c z ý
_ G
b- ý ö ý x- GC Z4 t\
N
Wý ýL
P- W 11 Z
W ýt
: Z.
uC
2ý Ca
Hv
.
.
. s
"
.
.
z c
H
H
W
C
W
ý J
d J
.. h C
r I' s
J
F6G- C ýYX
<< 2IS
> W G"
*C I-
ýý
"W ý " I.
.'_ º. n "n
": s .LAN
"z .'c
""C ý
""'JY C 1V
""<F
""QY. ýL
"fý2
C_ «"ra
2N Ir
"f ýL 1L J
CSC
""2< W2
f6f2
QWK p=< Y<6
NC 22 CW
< yý
Iz -- .u< WýH"G2ý
N V
ý
N
WNN a M1 p <C M1
^ CNN IA "
C
zr < ý I
vNC Wp .r aC ýI <MQ
r ä V. _r
ýý
T
Y.
~ ` C. A
.. rcc ý n, r
L6 <J
C6
< CS
V. V <C'
tr Vý >WG 1V JT <vV. r
<ssü
4CJ CC .C <hNý1 N r" J "V CS4P
4C G
s. V V^3
CZ 2Ný
Sýfý
.. ý_ 3LW> <f ;ý ýMNG
rR<Y.
III f
Q W
. 0-
Z
W f. i 2
Q ý J O H
.. C
ý Vf W
6 f
Q ý 7
ý
ý
ý ý ý ý ý ý ý ý ý ý : ý H
< ý ý ý
J f
F K f 6
° J 6 f
a t ý
` I- ý ý I I I
ý
ý Y Y
I > º ý ' ~ C Yi
I I
I
ý ý _ ` �3 r
Z
V.
V ý.
OC
;1
1 iI
I, "I II $1 Is
ý,
ý
-r t
1 hý t
ýýa
I ý 9 I
oI
Cl 0 C 0 c
C
C C C C C
ý
_ G
ý IC ý _
rw W-. arn CANCA aC aC"G Nr. r. ý, ̀
CCNGry N"ý "A " 1G Gr
ý rC 1r'"0. ýAC pCNpýGAý
rr .r .r" rcrJ "C MNr
rJ J"ýA
Q 1ý
Nh fe vY �ýJY
CCh; r" Kr J ýf
ý: ý ý U. ýý I.. "ý". ^j _-.
U. C
-r CI is .ý..
.. ý
ý ý u.
C-
ý I
fn GCý %
T! :.
ý ý P
`
P
U.
J
O H
~ 7. y
ý ý ý
VI 2
n. HW
N
iV W LN
S t
ýs r 2V
J r z> ý I- ýý
ls 7=
I
ý F
ý ý
ý z I-
2
r
ý V
,
229
TABLE 33 (Cont'd)
/ -0 N
am tv
J N
M M
4 ý
G <
ý
ý
ý I.: C
G 2
z C
' C J
C
7
a " "
."
"
" "
e
LL
ý. ̂.
R
Wt
J
J
2
CC zsw
'" r W\f 2S
ýr r
WN z"W
W
-rr- cä <ý <nfa
zrz ý0
- c º- " .c. 2W
C Cc "" tJ .L
Hý. "J 6 {: ""
Ix WK"" O2tt rf¢""
Nc"" 0-- z z2 O Y. ""W
J I- """2 <W 1V W -a"G O- W ýº N. O
Y/
ý
f0
<
r
ý a
t
vr
I Wr
0 JM IC N
> Ifý IC
0
O[ A aC A
'O M a0
.ýCp CD omý ýO
N
W 4i CC NO y V' C "ý 2OO
w cc CA O2 1ý OO HV"""
a
W4nM
<d <A ý4 eC Y+ OAAt Yi "..
C9 Q
ý J Q.
r
2
ACR KAC CC{
't m cc 4O
{Y I-' z t aa
W
z v
v.
ý U.
Cr . - CdG ý a0 C
c
ce r-n :. M
on cc r' J
ýr
W
G LL
C LL LL = .a 62 aC
ýLG Ot ý. C < LL GOZi ý "+ WN
236GCC CL
. ºý ZxX N2ý<<
LL 2ZZ
n i r. N
I
230
N 1ý:
W ý t G
.;
CC
0
r
. CO
ý
º-
ý a c z i z
H ý J
0.
z ý x^ Cý zý
N �. ý Wý
ýý
ýý WM 2
SH yt
c ý.
Z Y. O
=W Ca
46 a.:
I
.
.
r
.
.
.
z c
V1
H
tL'
a
C
ý
2
ý J
CL
r
J
.
.
.
.. ..
O[ WK "}
o s< Veý"f
i yG"" _=
1L: ""
0-
J"""
tW ýº
ýa W" WN"
TABLE 33 (Cont'd)
I
O
O -------------------------- -
II
G Y. N a
C CCCCCCCCý
Y. WWY. IV Y. 1L Wý
J
F C r
JýCCPd. C K"; PNN It zýM0MACCdCCW. NNNdM. cc AO; J aC ýMM N"^ NCOdwW. aC d C. NNdrX 4' ANMNdCN. NM dNOPN"ýPý NNM* NdGP O. OP N. p NP "ýA .O apN MNd H º.
pMM aC CNCCAd aC 0C 1' -ACM eý M CC r" dzNJAO E( 7C N1A N ACfbCNMA0ddýCC-A A- CCAO. aCr0C. AGAAM6ýº
NrANC r' hNdaC Cr :; AA N^ AOlC fý MCý. dddCIVKACCC yý ' NrOMMNN"O MMdC OM MM "dN M"ýNA ý. p MN It """"""_____""""" ý" "" ý" ""-""MH
IlttttItttIIIIItIIVZF.
ý < cs C W=G Z LL t:
r t: s
W H t
~ Y V
>
ý' =C C r.. +
tL - J9pPJ JN JMC O NNNKO pJ MNC NK p NN W4J JA ýO Jp W MNO NK CK ý GIMC NX C NNM7A t ti C N. O A OýA M Ný C1 A 1. ' CCNvGNNJ V' riKhA0C K<ý"' K V` ýhJApJKMJ ýº ýCaCCP: 0 0 0v JeCC C ýJH CrrP. r. CV C ýr. A CL fVý Ahr. FW . ý"Ar4C - ANC. r"K ." CA AL C. C C ýO AKOCJC JO OK Ar! t aý ýN JJ i` C. M'ý.: C ý A^ M CCV YN JAV JJM AM Arti. ý
. "KJ ýJM'r'. C CA e" rýCs
ýý " "ýý """" "ý " orrr
: r. r. aa
r. +rar. ýr.
r. v
uC -
ýCGCCCCGCCCCGGCGCCCCC_GGCCCCCCCCCC ýCGCCCCCCGCCGGGCýCCCGGCGCCr'CCCCCCC JOCG Gý
GrCCCOOCOOGCCC070rGGCCOOCCCC t0 /1C C0 > �. 000CCCCCCCCCC CGCGCCC CG GGC C RrJ 0[C CCC CC
hý V` Cýýýý.: e= "= C r, ý .ý ý_ ý .rý. J C""' Jf<SCý. Jt hý 7" ý . 1ý. t N JN. f MNA aCh'"'h
ý .-. -... "..... ". Y V. } Vhrrhf M` M r. Mýý .-
z C ý :ýNMJ Vý O f. eC P C. ý NKJ NG Iý eC 0O r` hMJ N'G 1ý 4' PGýn. M
C
U. H
-= « i rr
Nr ýC
C. VI {º N
r s., < º" "i r` a- 2 C. r=
"a v
a
2
P Yi G
ff W
C
a -z
ý .. rý ý h s N W v
Ot y Cý
º- .. r w
Cý
v. C u. . Y V".
I- Ok
WrY. C-
'Z Si `uG
7=
231
The first step is to choose the single variable which is
the best predictor. The second independent variable to be added
to the regression equation is that which provides the best
prediction in conjunction with the first variable. The program
then proceeds in this recursive fashion adding variables step-
by-step until the desired number of independent variables are in
the equation or until no other variable will make a significant
contribution (normally a user decision) to the prediction
equation. At each step the optimum variable is selected given
the other variables in the equation.
The SPSS regression package allows the data for each'machine
tool to be held in separate sub-files. The regression procedure
can then be carried out on individual or combinations of these
sub-files. Hence equations can be developed for single or ranges
of machine types enabling the total number of estimating models
within a system to be minimized.
4.4 Equation Validation
Since the multiple linear regression technique is to be
used when the exact relationship between dependent and independent
variables is unknown, it is essential that the accuracy of the
estimating models is confirmed to be within the current limits
of H. M. Ltd. (E = 0.096 and S=0.314, Section 4.4.1). To indicate
formula accuracy the most common method adopted(174) is the
Correlation Coefficient (Equation 32):
232
i=N
R=1 iY; - Y)(X; - X) (32) N. 0'y. 0"x
i=1
Where:
/
II
R= Correlation Coefficient.
öy = Standard deviation of actual values.
(Tx = Standard deviation of estimated values.
yi = Actual value of the dependent variable.
xi = Corresponding estimated value of the dependent variable.
y, x = Mean values of y and x respectively.
N= Number of data points.
The Correlation Coefficient is said to measure. fornula
accuracy by indicating the proportion of the original variation in
y that has been accounted for. The value of R lies between 0
and 1 and the closer to unity the greater the estimating accuracy
of the model. However, Cochran(155) points out that since R
does not directly measure the difference between actual and
estimated values (termed "residuals") it cannot truly measure
a model's estimating accuracy. The present work supports this
observation since little correlation is observed (Figure 16)
between the value of R and the Average Proportional Error
(Equation 36) which is a measure of formula accuracy based
233
on the size of residuals. Hence the use of R to measure
formula accuracy could be misleading since when R is close
to unity high residuals could still be obtained and the model
would have poor predictive powers.
Other commonly used methods of measuring the estimating lo,
accuracy of regression models that is, the F-ratio (174) and
Coefficient of Variability (176), also possess little correlation
with the size of residuals as illustrated in Figures 17 and 18.
The quantity 1-e (where e is the average error ratio and
is calculated using Equation 33) proposed by Cochran makes direct
use of residuals to measure formula accuracy and is, therefore,
an improvement on the use of the Correlation Coefficient.
However, Cochran points out that the estimates obtained from a
formula could appear to be accurate, simply because the actual
dependent variable data has a narrow range. In these circumstances
the use of the quantity 1-e could be a poor measure of accuracy
since it fails to consider that a narrow base of experience could
be an unreliable foundation for projecting future results.
However, the present author considers that the use of an
adequate data range is a pre-requisite to regression model
building hence removing this objection to the use of 1-e.
xi) (33) Ny
234
FIGURE 16 COMPARISON OF R WITH AVERAGE PROPORTIONAL ERROR
1.0 01x
° 0.8 x ýcx x..
iX
ý 0.6 xx öXXx
-P 0.4 x aý . 'v
xXX . r- 44-
14 0.2 0
C-) xX
X
-1.0 -0.5 0 +0.5 +1.0
Average Proportional Error (E)
FIGURE 17 COMPARISON OF F WITH AVERAGE PROPORTIONAL ERROR
70
. -. ý v
0 .ý
(Ki
U
t0 . r. i (0
60
50
40
30
20
10
ý
V
)c
x
x
xX x
x it
IL x XYX yl X
19 x x
X
it
xx
-1.0 -0.5 0 +0.5 +1.0
Average Proportional Error (E)
235
FIGURE 18 COMPARISON OF COEFFICIENT OF VARIABILITY WITH
AVERAGE PROPORTIONAL ERROR �
X
41 "r r "r -0
"r L tC
4- O
d-)
aJ "r U
.r 4- 4-
O
70
60
50
40
30
20
10
ý
ý
-1.0
X
X
X ix
xx
X
x X
xx
x I
x
Xx xx
x
x ý
X
1t1ý11111
-0.5 0 +0.5 +1.0
Average Proportional Error (E)
236
To take into consideration the range of data used to
construct an estimating model, Cochran suggested using the
Association Index (AI, Equation 34) which again increases in
direct proportion to the degree to which the formula accounts
for variability in the cost data.
ý
AI =1-T TY
(34)
Where:
e2 =1 ýI iY; - xi)
21
zI Qy =N2
Q'e2 = Standard Deviation of the Residuals. 2 0y = Standard Deviation of the Actual Values.
However, although the Association Index enables the variability
in actual data to be considered, by doing so it effectively results
in a factor (y) other than the difference between actual and
estimating values influencing the calculated accuracy of a model.
AI is not, therefore, a true measure of formula accuracy. In this
respect the use of 1-e is a disadvantage since this measure of
formula accuracy also includes the term y in its determination.
� .
237
An index of formula accuracy, the Standard Error Ratio
(Sr2, Equation 35), is provided by Ezekiel and Fox(174). This
index is suitable for measuring the estimating accuracy of costing
models since it uses only the size of residuals in its determination.
-ý ý ---- ý Sr2 =N,
I(Yi - xi)` (35)
Sr2 is a measurement of the range of estimating errors
obtained using an equation and therefore indicates the closeness
with which any specific estimate may approach the unknown actual
value.
The present author considers that an indication of the
overall estimating accuracy can be obtained using the Average
Proportional Error (Equation 36). The use of this equation
takes into consideration circumstances where residuals, although
large, may however only be a relatively small proportion of
corresponding y values.
E=1 N yi
Where:
(36)
E= Average Proportional Error
238
The Average Proportional Error is a measurement of the
average error as a proportion of actual values and is appropriate
for product costing equations since the accuracy of cost estimates
is normally determined as a proportion of actual costs.
For the present work the value E will be used to indicate e
the accuracy of estimates despite the apparent conflict between
this method of computing estimating accuracy and the objectives
of the regression procedure. Minimizing the values (yi - xi) 2
using regression analysis does not necessarily-result in the
error proportions being minimized.
Yi
However the regression procedure effectively improves the
estimating accuracy of high cost items which have been shown
(Section 5.2) to form the majority of the total costs of the
products manufactured at H. M. Ltd. This advantage together
with the ability to improve the estimating accuracy of equations
at the low end of the data range using minimum operations times
(see Section 4.5) make the normal regression technique the most
suitable method of developing costing models at H. M. Ltd.
It is possible to modify the regression procedure to %1)
minimize yi - xi ` values and to take into consideration
y; / the relative numbers of components manufactured. This modified
procedure (Appendix VI) would be more applicable to high volume
239
manufacturing systems in which the large batch quantities
involved could magnify small costing errors into large cost
variances.
To indicate the variability in individual errors and
hence the effective accuracy of regression models, the present Ile author proposes the use of the quantity S (Equation 37). S
represents the Standard Deviation of the proportional errors and
hence indicates the variability in individual errors.
s=
Where:
(Pi - E)2I
N
i (37)
S= Standard Deviation of the error proportions.
Pi = Proportional error yýY xý )
for the ith
observation.
E and S values can be used as descriptive measures of
accuracy, location and variation irrespective of the type of
frequency distribution of the labour times. They cannot be used
to determine confidence intervals since this requires an assumption
that the random variation of labour times is normally distributed.
N
240
4.4.1 Accuracy of the Existing Estimating System
at H. M. Ltd.
Using the quantities E and S the accuracy of the
estimating system at H. M. Ltd. was determined. The
results (Figure 19) indicated that the Average Proportional
f Error is 0.096. This is within the +10% accuracy limits
that management specify. However, there is a high
variability in the individual estimating errors since
S=0.314 (i. e. 31.40).
4.5 Comparison with Time Study Data
The accuracy of each estimating model developed (Appendix V)
was tested against time study data with the acceptance criteria
set at H. M. Ltd. 's current accuracy limits of E=0.096 and
S=0.314. Time study data used to develop the equations was
not subsequently employed to test the accuracy of these equations.
In this way a more valid indication of estimating accuracy would
be obtained.
Those equations that failed to meet the acceptance criteria
were examined for possible methods of imoroving estimating accuracy.
The causes of poor estimating accuracy were found to be:
(i) Incorrect combination of predictor variables.
Frequently either changing the combination of
predictor variables used and/or adding new
variables and/or removing variables improved the
241
FIGURE 19 A COMPARISON OF ACTUAL AND ESTIMATED-
LABOUR COSTS AT H. M. LTD.
E=0.096 S=0.314
N,
d-ý
O
O . L] tC
J
r t0
ý V
3.0
2.0
1.0
le
1.0 2.0 3.0
Estimated Labour Costs (VS)
242
estimating accuracy of models. Examples of
typical changes carried out are shown in Table 34.
i
(ii) The use of negative coefficients within the
estimating models. A disadvantage of using the
regression procedure is the possibility of
developing equations which contain independent
variables with negative coefficients. Hence the
estimation of negative manufacturing times is
possible.
If the negative sign is contrary to the suggested
causal effect of the independent variables Thelwell(151)
suggests that the variables can be excluded from the
model without serious loss of estimating accuracy.
This proved correct with many equations developed
during the present work. The regression procedure
had to be re-run without variables producing negative
coefficients included.
For certain equations exclusion of the predictor
variables with negative coefficients produced
equations that failed to meet the stated acceptance
criteria for accuracy levels. Inclusion of these
variables however improved the overall accuracy of
equations but resulted in negative operation times
being estimated at low values of predictor variables
(see Figure 20). In order to overcome this difficulty
243
TABLE 34 EXAMPLES OF ALTERING INPUT VARIABLES TO IMPROVE
ESTIMATING-ACCURACY.
N W C'S Z
S
i
O rr F-
P-4 co M: O U
J
º-ý 1- r-+
ý
Ec 4- Wi a) "r eo O. C S- E F- +J (3) S-
m (A TJ Cl O Q) CO Q1 O 4-
. X O. C C s. "r fö"r i-) "r4- a _C g (n aJ i-) oa C) CE cZ co >ý
s- ý aýi c
o "r (1) c. 1 .nO 'O +J r- E "r = 4-3 Q C:: U¢ÜC co
+
C) 01 w Ný Z º-+ W f- M: rý C7 tn ýZd
ý--ý O QZ
r--4 J 2Q
XU F- QO ýýH
+++ zzz o00 ý V) N V) ZZZ WWW
ýiý 1--n 1--4 º-1
000 Q 0ý U (I)
a O
>G XXi ddd= ýýýZ
00 äö c w F-- O "ý r--
NUQ.
=w Rs "r- U 0- ZG iO
i CU C CU .C C71 O
+ý U tC "ý CU +) i +) E "ý Q) (l) rt3
O¢ +) ý O> Oý"C7 E
NO -C. 7 C3O C1 O to U Oi Cl) G)
c3: ý CD > .C> S- 0 4-) O CU
v-v E aO .N -a C CU Niý ¢ ro Cr cz c7
+ - r*l + N
. . -. r= 1 C)
I- .-
¢ a. F-ý
0W
WO dO Ow CY- C. 7 `i v
t 4-3
ZZO
+ ý z v
v) G) > 0 O S.
C. 7
4- O
5- a)
.a E ý
Z
+ C) >
+ .. c O 4J O O.
U (1) C7 G
If.
Z l0 N
au C. (1) rn
> (c (1) 0 S- a. o a) 0 S.. > ý cý cc
4J CL
aý
t
. r. 4-3 ý c
.ý
.n E +ý OO OU O G)
F-- >i O G)
'o G) Q
-C 0
.C iý CL
aý 0 t 41 0 0 ý---
:N~
1--1 .: ý ci CV 3
+
r
T7 O
to O
C7
+ . -ý
ý aý m
a- 4-. 0
a .0
+
LL
4-)
"r
aý V to
LL.
aý vº l0 ýý O1
> OaO r- td Nu00NU GJ
ý d' i C7 M: co
I
I--,
C e0
ý-- m
4- O
a) 4-) a) E
a) -0 ., _ N
4J
O
} ý--ý
., _ -0
G) CL C1.
ý N C)
4-) O
O "r. ý 0.
U ý C)
C)
.0
.,...
tö
C) C) N
H C)
.0 t0 "r..
f0 >
O ý U
. '.. ,a C) i 0
ýF- O
O "r" ý 0.
. r... i U N a
i O
LL
äi +j 0
244
FIGURE 20 THE ESTIMATION OF NEGATIVE MACHINING TIMES USING REGRESSION MODELS
. le
4
Regression Line
(n C
. r. ý
ý ý
"r I-
C, C
"r C
I
UI ý
ý
+ve 0
-ve ii ý
. -I
Negative Machining Times
oe
Independent Variable Values (xl, x2, x3 ... xn)
245
oe
the dependent variables concerned were allocated the
minimum operation times shown in Table 35 below
which estimated operation times could not fall.
Since the allocated times are based on an
examination of actual minimum operation times
this procedure effectively improves the accuracy
of estimates from equations with negative
coefficients.
TABLE 35 MINIMUM OPERATION TIMES
Cost Centre Equation Number Minimum
(see Section 4.7.1) Operation Times
(rains) 201 5
203 1,2,17,18
204 1,2,3,4
214 8
216 8
219 2
216 1
202 1,2,3,4
204 18
209 5
210 17
236 1,2
214 1
0.5
1.0
2.0
246
TABLE 35 (Cont'd)
Cost Centre Equation Number Minimum (see Section 4.7.1) Operation Times
(mi ns ) 202 9,13,18 )
3.0 219 1)
i
210 1-54.0
206 5) ) 5.0
209 18
The time study data used to construct the estimating
equations was also checked for accuracy. The causes of variability
in this data source were identified and are listed in Table 36.
TABLE 36 CAUSES OF VARIABILITY IN TIME. STUDY DATA
1. Variation in machine process data (e. g. machining speeds
and feeds) between identical machining operations arising
from the absence of detailed methods planning and control.
2. Variation in the dimensions and quality of materials used,
e. g. the dimensional tolerances of cast materials and the
presence of casting defects.
3. The rating of the work performance of operators is a
subjective decision and would therefore tend to vary
between time study engineers.
247
Models were not developed for estimating material costs
since preliminary attempts indicated that a major predictor
variable was the volume of a particular component or material
manufactured by suppliers. This type of data is normally
not available to H. M. Ltd.
The frequency distributions (Figures 21 and 22) for the E
and S values of the estimating equations developed indicate that
the accuracy of the majority of the equations is within the
current accuracy levels of H. M. Ltd. A number of estimating
equations had accuracy levels that lay outside the current
limits of H. M. Ltd.
These were equations for estimating:
(i) assembly task times,
(ii) tooling costs,
(iii) machine manipulation times for horizontal and
vertical borers,
(iv) load/unload times for horizontal borers,
(v) boring times (tols. ! O. lmm) for chuck lathes,
vertical borers and horizontal borers,
(vi) turning times (tols. >0. lmm) for chuck capstans,
vertical borers and copy lathes, and
(vii) turning times (tols. <0.1mm) for centre lathes
and vertical borers.
248
FIGURE 21 FREQUENCY DISTRIBUTION OF 'E' VALUES
100
N 4) N
O
i d) 50:
0.25 0.20 0.15 0.10 0.05 0 0.05 0.10 0.15 0.20 0.25
Average Proportional Error (E)
FIGURE 22 FREQUENCY DISTRIBUTION OF 'S' VALUES
100
N
IN
C.. )
4- O
CU
.a E
50 / /
// // /I
F I
-I- I 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Standard Deviation of Proportional Errors (S)
249
(i) Assembly Task Times
Since time studies of assembly tasks had-not been
carried out at H. M., Ltd. data was not available from
which to develop estimating models. The use of MTM
time standards are more applicable to highly
repetitive work tasks which are normally not found 'Oe
in-low volume manufacturing systems such as H. M. Ltd.
At
An attempt was made to use STT times for various
sub-assemblies and determine the relationship between
STT times and the types of components that formed
the sub-assembly (Table 37). The regression procedure
produced models (e. g. Equation 38) the coefficients
of which clearly did not reflect the actual causal
effect of the predictor variables. Hence the
equations could not be included in the estimating
system.
4.48a1 + 93.87a2 + 48.01a3 - 101.09a4
- 80.50a5 - 44.86a6 + 130.73a7 + 71.89a8 (38)
+ 76.61a9 - 38.75a10 - 9.49a11 - 514.66
Where:
At = Total assembly time (mins).
al, a2, a3... all = The number of component types
(Table 37) within the assembly.
250
TABLE 37 COMPONENT CLASSIFICATION FOR DETERMINING ASSEMBLY TIMES
VARIABLE COMPONENT DESCRIPTION EQUATION 38 NAME CODE
DISCA , Disc shaped with outside diameter <150mm a4
DISCB Disc shaped with outside diameter $, 150mm
SHAFTA Shafts with length <250mm
SHAFTB Shafts with length p250mm
BRINGS Bearings, keys
SCREWS Screws, plugs, nipples, nuts, bolts
CCLIPS Circlips, dowel pins, dust caps,
a10
a6
a5
a2
all
retaining rings, springs, washers a1
PLATE Plate sections
SMALLFAB Castings, Pressings and Fabrications
with a maximum dimension of <150mm
a9
a8
LARGEFAB Castings, Pressings and Fabrications
with a maximum dimension of >150mm a7
SEAL Seals, gaskets a3
251
/
Equation 38 illustrates a disadvantage of using
regression analysis to determine causal relationships.
That is, as the number of predictor variables used
to construct an estimating equation increases it
appears that the possibility of the regression
technique finding non-existent relationships also
increases.
The technique adopted for developing assembly time
regression models was that used previously by the
author when establishing models for estimating times
for the. assembly of heavy custom designed cranes(154).
The technique involved using personnel familiar with
the assembly of existing product ranges within
H. M. Ltd. to choose predictor variables and estimate
assembly times for various levels and combinations
of variables.
From this data, estimating models were developed that
expressed the relationship between assembly time and
the dimensions of various component types, such as
disc shaped components, pins, shafts, castings,
plate and fabrications. In addition numerous
components (Appendix IV) were allocated the standard
times shown in Appendix V.
252
i
The accuracy of the estimates developed by
these equations could not be measured against current
STT assembly times because of inconsistency in
the STT times. This inconsistency is particularly
noticeable when comparing STT times for assembling
similar products. In these cases maximum differences
of 200% have been observed.
(ii) Tooling Costs
Attempts to develop equations for estimating the costs
of various special tooling items were only partially
successful as illustrated in Figures 23 to 28.
Whilst examining the data used. to construct the
estimating models factors were highlighted that
suggested the use of regression models for
estimating special tooling costs was inappropriate.
The primary difficulty arises since the design and
manufacture of jigs, fixtures and special tooling is
sub-contracted. Hence the influence of variations
in sub-contractors profit levels, overhead rates,
price increases, labour rates and material costs
are difficult to isolate. In addition since
numerous sub-contractors could be employed these
factors would also vary between sub-contracting
firms.
253
800
t/f
W 700
4-3 N O
U
ý ý ý U 4
600 ý
. 500
400
700
Vf
W ....
+1 N 0
U
ý V
600
400 500 600 700
Estimated Cost (i's)
600 Estimated Cost (£'s)
800
700
254
. --4 (N
W
ý " t/1
0 I
r
ý ý ý U Q
100
80
60 r
- 40
20
20 40 60 80 100 Estimated Cost (S's)
Vf
14
12
10 W v
4-) N 0
U
ý U
8
6
4
2
2468 10 Estimated Cost (I's)
12 14
255
1 07 W ý
4-, 1/f O
U
r ý
ý i-, V
.¢
ý N1
W v
4-3 N O
U
4-) V
C
700
600
500
400
300
300
. 150 200
Estimated Cost (£'s)
400 500 600 Estimated Cost Ws)
700
250
256 '
Therefore it was decided to remove the jig and
fixture costing equations from the estimating
/
system. However in the event that H. M. Ltd.
decide to manufacture jigs and fixtures in-house
the procedure for calculating tooling costs has
been retained in the estimating software program
(Appendix III). At the present moment the lack
of estimating equations in the data file will,
initiate a default message informing the user
that equations for estimating tooling costs are
not yet available.
(iii)-(vii) Manual Task Times and Machine Process Times
The equations for estimating machine manipulation
times for horizontal and vertical borers; load/unload
times for horizontal borers; boring times (tols.
n. lmm) for chuck lathes, vertical borers and
horizontal borers; turning times (tols. <O. lmm)
for copy lathes and vertical borers represent a
small percentage (0.3ö) of the total number of
equations developed. The causes of inconsistency
between estimates and time study data could not be
found. It is therefore assumed that this inconsistency
arises from the factors listed in Table 36.
257
4.6 Comparison with existing Time Standards (STT's)
Although the estimating accuracies of the majority of the
equations are within the stated limits it is necessary to
determine whether the equations can be used to build-up
accurate manufacturing times for complex designs which require
more, than one manufacturing operation. This will ensure that
estimating errors are not wholly additive. In order-to achieve
this objective, estimates using the regression equations have
been compared with STT's since these times represent the total
machining time for a component processed on individual machine
tools and are therefore the sum of the individual operation
times.
The accuracy with which the regression equations can
estimate STT's is illustrated by Figures 29 to 46 in which
ideally all points should lie on the 45° lines.
4.7 System Computerisation
Although the system can be used manually as described in
Appendix II, software has been developed('77) to allow a computer
to carry out the tedious calculations involved and hence reduce
computational errors. Operation notes for the computer program
are contained in Appendices III to V:
ý
I- N
"r
ý F- t- cn ý
a) E
"r H
Q" i.. cc
F-N
G)
N
ý N c
"r ý ý..
.ý I- F- N ý
Ei E
H
ý aý
(ii I-
r m
N
20
15
�
10
5
50
40
30
20
10
10
5
258
10 15 Estimated Time (Mins)
20 30 40 Estimated Time (Mins)
20
50
259
N
.ý
ý F- N
N E
"r", F-I
+-) N cn LI
NI 4)
r0 tJ')
... N
ý F-
d E
F-
4-3 N 0) L to
H
N C1
5
4
3 Of
2
1
25
20
15
10
5
1
S
234
Estimated Time (Mins)
10 15 20 Estimated Time (Mins)
5
25
260
ý
tN C
.ý ý
F-- F- N
E "r ý
4-3 ý
F--
N 4)
r
NI
200
160
120,
80
. 40
40 80 120 160 Estimated Time (Minn)
ý Ni CI
.r ý ý
f--
N E
"r F 4-J (I)
S.. t0
f"
V1 G)
r t0
V)
100
80
60
, 40
20
20 40 60 80
200
100
Estimated Time (Mins)
N C
. r. ý ý
ý F- I-- N v
a E
"r r h-
41 ai
1
rn
cC H Vf G)
r ý
N
. - t/1 ý
.ý ý ... i
ý H H N ý
C, E
"r F-
i-ý G) C, S-. rö
F-
vf C)
r r6
Cl)
50
40
30 ý
20
10
25
20
15
10 ý
5
10
5
261
20 30 40 50
Estimated Time (Mins)
10 15 20 Estimated TimeMins)
25
262
N
. r. Z:
ý I- H N ý
Q) E
." "r I- 4-, aý rn S. --
" U, N
r fý', tN
25
20
15,
10
5
5 10 15 20
Estimated Time (Mins)
0- N
"r
F- F- cN
() E
. r- F-
4-3 G) O) t. r0
F-
N G)
r IO
N
25
20
15
L 10
5
5 10 15 20 Estimated Time (Minn)
25
25
263
ý Vf C
"r ý v
, -, I . - F- N .. -I
C) E
H ti +J
ý rn ý F- ý
t/)
. -ý tN C
"�-. i ý..
1ý I- I-c/)
"r. I-ý
4-2 N
L ý
1-`
cn a)
r b
ý/)i
15
10 e
5
10
5
5 10
Estimated Time (Mins)
FIGURE 40 STT VERSUS ESTIMATED TIMES
COST CENTRE 236
SAWS
I
15
5 10 Estimated Time (Mins)
264
, C)
ýý
rn S-
N ý
R3ý N
"r
I
30 N i
"r M: . -, . -ý
! 2ý r-
20 H 4J C)
10
200
160
120
%80
40
40
10 20
Estimated Time (Mips)
80 120 Estimated Time (Mips)
160
30
200
265
t/1 C
. r. ý v
0- F- ý N v
a E
If-
4-3 a S- to F- N
. G)
r- ý tN
N C
"r .ý
f-
N
dJ E
.,. H
4J d)
al
RV
(If C)
r ý
V)
25
20
15 , or
10
5
25
20
15
10
5 10 15
Estimated Time (Mins) 20
FIGURE 44 STT VERSUS ESTIMATED TIMES-
COST CENTRE 258
NC BAR LATHE (BATCHMATIC-) xx
X
x
Xx
x x Xx Y.
I1
x
25
I
ý
5 10 15 20 25 Estimated Time (Mins)
266
ý t/1, C'
ýI ý
ý-.
F-- N ý...
W E
"r F-
4-) N Q L
N a)
r
i
FIGURE 45 STT VERSUS ESTIMATED TIMES
COST CENTRE 270
SPLINE CUTTERS
X
I
xx
1 Estimated Time (Mins)
xx
x
'ix
1
2
N c
. �- ý v
. -, I- I- �I) ý
a) P "r ý
4-3 (1) r. P S- (0
F-CA
a)
rrd N
5
4
3
2
1
1 234 Estimated Time (Mins)
5
267
(i) Appendix III contains the operation notes for using
the program and a full program listing. Included is
a list of the input codes for the manufacturing
processes that can be estimated using the system
(Flow Chart Code 3).
(ii) Appendix IV contains definitions of predictor variables,
equation codes and the order in which values for
variables should be entered into the computer program.
(iii) Appendix V is the program data file and contains the
relevant data required to determine the machine type
when general machining processes are being costed
(Table 51), a list of the predictor variable estimat-
ing coefficients for each operation that can be
costed (Table 52) and the labour cost rates,
manufacturing overhead rates and administration
overhead rates for each cost centre (Table 53).
4.7.1 System Size
The system consists of a large number of estimating
models. A method of numerical coding was required to allow
the appropriate equations to be quickly retrieved. Each
equation was initially coded according to the applicable
machine tool group (using H. P. Ltd. 's management accounting
cost centre classification) and secondly allocated an
268
operation reference number in order to allow each individual
equation to be located in the data file, e. g.:
AA
L i
Equation number for a rough turning operation
Cost centre coding for bar lathes
Specific equations can therefore be isolated using
the cost centre classification and equation number.
Exceptions are cases in which different machine types
have identical cost centre codings. At H. M. Ltd. these
are:
Cost Centre Coding
Cylindrical and Surface Grinding 216
NC Bar Lathe (Batchmatic)
NC Chuck Lathe (Churchill) 258
It has been necessary therefore to add an extra
numerical code to the cost centre to form input codes (see
Flow Chart Code 3, Appendix III) for the computer program.
The use of H. M. Ltd. 's coding system enables the
software to be simplified for the conversion of production
times to costs since both labour and overhead cost rates
269
vary with machine type. Hence an additional classification
number is not required to identify each equation for
costing purposes.
4.7.2 Variation of Machine Types
The range of machine groups capable of carrying out
general machining operations is large. It was necessary
to develop decision rules to enable engineering designers
to choose the appropriate machine group, i. e. normally
production engineers choose machines for components.
This type of data is therefore inaccessible to designers.
0
Examination of the component limitations of machine tools
revealed that the main factors affecting the choice of
machine type are:
(a) Whether cylindrical or prismatic operations
are required.
(b) Whether a boring operation is required.
(c) The component diameter.
(d) The component length.
(e) The annual quantity of components.
270
In the present study prismatic operations are defined
as, "those operations not carried out about the axis of
rotation of the component" and are associated at H. M. Ltd.
only with horizontal borers (input code 0,209). Hence
it is not necessary to include this group of machines in
the decision rules for determining the relevant input
code. ý
The length, diameter and annual quantity of components
have been used to develop decision rules-(see Table 51,
Appendix V). Two groups of rules are necessary since the
maximum component length per machine type is dependent
on whether or not a boring operation is to be performed.
The annual quantity of components machined determines
the cost effectiveness of using numerically controlled
machine tools.
4.7.3 The Effective use of Computer Time
To obtain effective use of computer time it was
necessary to enable estimators to collect the input data
for the estimating system. This involved allowing
estimators to determine the relevant input code using
Table 47, Appendix III and to identify the independent
variables using the appropriate Variable Description Notes
contained in Appendix IV. A standard form (Figure 47)
271
FIGURE 47 DESCOSTIMATE STANDARDINPUT DATA FORM
MANUFACTURING TINE/COST ESTIMATIONS
PRCDUCT:
PAGE: bF
DATE:
D.: tING No.
F: /C CENT. '2Lý
OPERATION CODE A B C D Il? /C
TIME
F. T. _
TIi: 7--, / COST
i
272
has been designed in order to aid the collection and
input of data.
Although the decision rules for determining the
appropriate General Machining input code have
been included in the system software (for use when
computer time is not limited) the inclusion of the
Variable Description Notes could only be justified if
the estimating software formed part of a Computer Aided
Design system. Cost estimating could then take place
without operator intervention.
Estimators with a detailed knowledge of manufacturing
techniques (e. g. production engineers) would know the
relevant machine code for estimating equations, the
module for machine code determination can therefore be by-
passed hence reducing the system application time.
4.7.4 System Expansion
Since in the course of time new manufacturing processes
are likely to be introduced into H. M. Ltd. the system must
be easily expanded. Hence data file space has been left
vacant for the introduction of two extra machine groups.
In addition each machine group contains vacant data lines
to enable additional equations to be included. This is
beneficial since the flexibility of the regression procedure
enables estimating equations for specific operations to be
developed. For example, the machining of grooves for rope
drum tubes (Equation Code 20609).
273
CHAPTER 5
DESIGN COST FORECASTING
274'
5.0 Introduction
The regression based estimating system developed allows designers
to quickly estimate product costs if general arrangement drawings exist
from which to obtain the values of predictor variables. This avoids,
having to carry out more detailed design work which is necessary for
costing purposes when using conventional costing techniques. The
regression based estimating system permits economic consideration of
a greater number of alternative designs and therefore has considerable
advantages over conventional costing techniques.
The ability to add to the number of design alternatives that
could be economically considered at the concept stage would improve
still further the likelihood of emerging with the ideal or best
possible system to meet defined needs. This could only be economically
accomplished by reducing the amount of resources required to cost
concept designs. Rather than being an estimating task using
established drawings and costing techniques the costing of designs
would become a forecasting task. Cost forecasting of designs in the
. present work is therefore considered to be, "the estimation of total
product costs without the necessity to undertake high levels of
design and cost estimating". The objective of a cost forecasting
technique would be to enable total product costs to be determined as
early in the design process as possible. The benefits to be gained
in achieving this objective are:
(i) limited design resources could be allocated more
efficiently,
(ii) the level of design resources employed on cost
ineffective design concepts could be reduced,
27 5
(iii) specification for the optimum standard sizes could
be identified early in the NPD process, and
(iv) cost data on competitors products for sales
forecasting techniques could be more easily
provided.
5.1 Cost Relationships
An obvious method of cost forecasting when designing a new
product range is the use of cost/design specification relationships.
Pedder-Smith (178) has shown that relationships can exist between
design characteristics such as product size and costs. Hence, with
reference to the New EHB development, using cost relationships
developed from selected hoist capacities the costs of remaining
hoist sizes can be predicted.
Since regression analysis is an effective method of determining
causal relationships the technique has been used to quantify the
relationships between manufacturing costs of components and hoist
capacity. Examples of costing equations are shown in Table 38,
i. e. only the coefficients of the costing model (Equation 28) are
listed. The predictor variable in all cases is "basic hoist
capacity (Tonnes)".
276
TABLE 38 COST/HOIST CAPACITY RELATIONSHIPS
Component Type of Equation 28 Cost Coefficients
Estimating Accuracy
Xo Xý ES
(£'s ) i
Brake Material 9.3 29.1 -0.002 0.077
Gearbox Material 13.2 58.6 0.003 0.030
Gearbox Assembly Labour 20.6 5.6 0.002 0.056
Hoist Block Material + 204.3 278.0 -0.008 0.042 Labour
Dual Wound Motor Material 207.9 115.3 0.007 0.040
Squirrel Cage Motor Material 56.2 36.4 0.020 0.090
Rope Band Material 0.451 0.234 0.021 0.097
Moulding Tool for
Rope Guide - Tooling 5393 504.6 0.000 0.023
It can be seen that the accuracies of the cost models listed in
Table 38 are well within the present accuracy limits of E=0.096,
S=. 0.314. However these cost models have been developed using the
costs of the total product range. In order to reduce the amount of
cost estimating required cost models must be constructed using the
minimum number of hoist sizes. In order to construct a linear cost
model the minimum number of hoist sizes is two. It is therefore
'necessary to develop cost models using only two hoist sizes and
determine the overall accuracy of these models.
277
When using regression analysis to develop cost models it is
necessary to employ the full range of cost data expected to be
encountered. Hence in terms of the New EHB development cost estimating
equations must be constructed using the smallest (0.8 tonne) and largest
hoist capacities (8.0 tonne) that form the product range. Therefore,
the initial general arrangement drawings prepared must be for these
units. However this is not always possible since the order in which
units are designed must often consider marketing factors. For example,
the schedule for the New EHB range required the initial development
of the 0.8 tonne, 1.6 tonne and 3.2 tonne units-since these units
represented 92.0% of the total forecast sales and hence the largest
potential profit. In addition the development and launch of a limited
product range would allow early profits to be obtained.
Developing cost models from a limited data range and their
subsequent use to predict costs outside this range is not recommended
since the cost models may not represent the causal effect of the total
data range. This is illustrated in Table 39 which lists the accuracy
levels of cost models developed using the 0.8 tonne and 3.2 tonne
hoist units.
278
TABLE 39 COST/HOIST CAPACITY RELATIONSHIPS DEVELOPED USING THE 0.8 TONNE AND 3.2 TONNE UNITS
Component Type of Equation 28 Estimating Cost Coefficients AccuraSy
x0 xi Es
/ (i's)
Brake Assembly Material 13.7 28.8 0.049 0.077
Gearbox Assembly Material 18.0 54.0 -0.036 0.038
is Labour 22.3 4.3 -0.056 0.060
Hoist Block Material + Labour 269.2 245.1 -0.024 0.040
Dual Wound Motor Material 196.5 114.8 -0.018 0.036
Squirrel Cage Motor Material 33.4 51.5 0.121 0.123
Rope Band Material 0.333 0.333 -0.014 0.272
Moulding Tool for
Rope Band Tooling 5400 500 -0.001 0.022
Comparing Tables 38 and 39 indicates that the accuracy of the cost
models is reduced. Although the accuracy levels shown in Table 39 are
still within the present estimating accuracy levels of H. M. Ltd. this
could not always be assumed. Hence the use of cost models is restricted
to development programmes in which costing requirements do not conflict
with those of the marketing function.
Cost models are also useful in the cost optimization of product
designs. For example, the wire-rope drum tube is used to wind and store
the wire-rope during operation of the hoist block. The outside diameter
279
of the mild steel drum tube is machine grooved to prevent adjacent
rope strands rubbing together, and hence causing premature rope wear.
The main dimensions to be determined during the design process
are:
loe
(i) wall thickness (t)
-(ii) length (Ld)
and (iii) outside diameter (D)
These will now be considered in detail.
Wall Thickness (t)
The wall thickness required is related to the hoist capacity
since it should be sufficient to prevent the drum tube buckling
under the maximum hoist load. The required wall thickness can
be determined from the appropriate crane design standards(179).
Since the present example looks at the design of the 3.2 tonne
drum tube the wall thickness can be considered constant. Hence
it is not a design variable.
Length and Outside Diameter
Since the drum tube stores the wire-rope during hoist
operation it must be capable of containing the full length of
rope. The relationship between length (ld) and diameter (D)
for a hollow tube is given by Equation 39.
280
Where:
Ld = H. N. 1.125G
ir. D
H= Height of Lift (Metres).
N= Number of Rope Falls.
1.125G = Rope Groove Pitch (where G is the rope.
diameter in millimetres).
J Drum tube length allowance to allow the
rope guide to locate the rope on the
outer rope grooves of the tube, (mm's).
(39)
It will be obvious that as the tube diameter increases,
the tube length required to hold unit length of rope decreases.
However, as D increases the cost per unit length of drum tube
also increases. In addition, as Figure 48 indicates, drum tube
dimensions influence the costs of other components viz:
(a) The length of the drum tube influences the length
of the drive shaft, tiebars, rope anchorage and
support tubes.
(b) The diameter of the drum tube influences the
diameters of the rope guide assembly and drum
tube end plates, and the gearbox reduction ratio
required to provide the specified hoisting speed.
28.1
FIGURE 48 ASSEMBLY COST INTERRELATIONSHIPS FOR THE NEW EHB
ai Y (10 i m
*
X O
t:
Q) CJ
*
a:
.n F- E L
0
aV n 0 ý
a) v'
co
CL 0 w
U 0
r--
E 0
4-) 4-) 0 co
4-3 V-
r
> "r
0
*
N S. -
C) . f- I-
rn c
If. +1 (1) Cý = (0 Or-
M: d
a)
rts O
a)ý CL U OC
Q: Q
Motor
Brake
*
* *
* *
* * * *
*
Gearbox
Drum Tube
Rope
Rope Guide
Bottom, Block
* *
*
*
*
*
Drive Shaft
Tiebars
Support Tubes
Mounting Plate
*- Cost Relationships
282
Since both D and Ld influence the total cost of these
components the problem of cost optimization faced by the
designer is to determine the Ld: D ratio that minimizes total
Costs.
Using regression analy le
were determined:
Cost Components Cost
Identification . Relationship
Drive Shaft + Tiebars + Support Tubes + Rope
Anchorage
Ct2 Drum Tube End Plates + Rope Guide Assembly +
Gearbox Casing
Ct, = 56. OLd (40)
Ct2 = 46.0 + 143.94D (41)
Ct3 Drum Tube Ct3 = 6.17 + 0.00233 (TT(D2_(D_2t)2)Ld)(42)
4
Using regression analysis the relevant cost relationships
The total costs (Table 40) were determined over a range
of drum tube lengths.
283
Table 40 TOTAL COSTS AT VARIOUS. DRUM TUBE LENGTHS
Drum Tube Drum Tube Total
Length (Ld Diameter (D) Component Costs (£'s) Costs (M) (mm)
. Ct1 Ct2 Ct3 Ct1 + Ct2 + Ct3
0.70
0.75
0.80
0.85
0.90
0.95
1.00
1.05
1.10
oo,
0.458 39.20 111.82 42.41 193.43
0.423 42.00 106.89 41.92 190.81
0.393 44.80 102.57 41.49 188.86
0.367 47.60 98.83 41.11 187.54
0.343 50.40 95.37 40.64 186.41
0.324 53.20 92.64 40.44 186.28
0.306 56.00 90.05 40.14 186.19
0.290 58.80 87.75 39.87 186.42
0.275 61.60 85.58 39.54 186.72
Using the total costs listed in Table 40 the minimum'
cost curve (Figure 49) has been constructed. The curve indicates
that "minimum" costs are incurred over a range of Ld: D ratios
(i. e. 2.6 - 3.3). Factors to consider when deciding the
actual Ld: D ratio to adopt. are:
(i) The minimum drum tube to rope diameter ratio
permissible under the relevant hoist design (180)
standards. Using these standards the
minimum drum tube diameter is 320mm.
284
FIGURE 49 TOTAL COSTS VERSUS Ld: D RATIO FOR THE-3.2 TONNE EHB
aý ý 4-)
Q) E (0
ýr
ý ~OM
.-O i
E -0 :3 +ý E J "r
L "ý ý
"ý ý cn LO 'O - O" C) " O
O M tu
JCi CD
ö I^
LO Q1
F--
I
IIIII1III
0 rn ý
N ýn E -P N
.ý Vf ý ipW
U ý- s-I oI w
C. )
Ln
Co r
er ý" O
LU, ý
M
M
ý
ko N
LLC) ^
0 . r. 4-) ro
výo J cm
J
285
(ii) The maximum length of hoist block acceptable in the
market (an increase in the length of the hoist
block reduces the working area of the hoist).
Since the market survey(26) did not determine the
maximum hoist length acceptable in the market it is
necessary to choose a drum tube length at the low
end of the minimum cost area.
(iii) As the drum tube diameter is reduced the lifting
speed for constant drum tube revolutions per minute
decreases. Thus to maintain a constant lifting
speed the gearbox reduction ratio must be lowered.
In order to lower the reduction ratio the number of
teeth and outside diameter of the spur sears must be
reduced. However reducing the outside diameter of the
spur gears forces the gear centre lines closer
together. Therefore, the limit to the minimum
drum tube diameter which is feasible occurs at the
point where the gear centres cannot be moved any
nearer to each other.
5.2 Costing using the Pareto Distribution
If the cumulative percentage of an element is plotted against
its cumulative percentage effect then the Pareto distribution predicts
that the resulting curve will be of the form illustrated in Figure 50.
The Pareto distribution, although applicable to many areas of industry,
286
FIGURE 50 THE PARETO CURVE OF CUMULATIVE
, or
.. ý v
G) ý
r ý
ý
G) >
.r ý fl7
r ý ý ý c. )
80
60
40
NUMBER OF STOCK ITEMS VERSUS CUMULATIVE STOCK COST
20
-1 11i
20 40 60 80
Cumulative Item Count (%)
287
is principally used to design selective stock control systems which
control only the minority of stocked parts that account for the majority (181ý
of stock costs
The equation for the curve shown in Figure 50 is of the general
type: i
Cstk - f'Nstk 9
where for stock control purposes:
Cstk = cumulative value of stock costs.
Nstk = cumulative. number of stock items.
f, g = constants the values of which depend on the
form of the curve.
(43)
In terms of costing products Equation 43 finds a wide application
-in the electronic and aerospace industries where the evolution of the
design of a product to a stable configuration takes place during the
manufacturing stage. In these circumstances Equation 43 also
represents the effect of learning(182) and can therefore be used
, to predict the reduction in manufacturing costs that arises due to
an increase in the number of products manufactured. When the
cumulative cost curve is assumed to be log-linear the resulting
unit cost can be determined according to Hill(182) using Equation 44.
ý
288
Where:
Nuh
C=f (Nu 1+9
- Nu 1+g ) hh h-1 J
Ch = The cost per unit for the hth unit.
i The cumulative number of units = h.
Nuh-1 = h-1 units.
(44)
Examination of the component costs of a wide range of products
manufactured at H. M. Ltd. revealed that component costs followed a
typical Pareto distribution. It could be expected therefore that
Equation 43 could be used to determine the total costs of a product
if the coefficient f and indice g could be determined. Since the
equation is assumed to be log-linear f and g would represent the
C-intercept and Slope respectively of the log-log straight line curve
of Equation 45.
Log10C = Loglof + gLogloNu
Where:
C
Nu
= Cumulative product costs (S's) .
= Cumulative number of components .
(45)
Therefore using only a small proportion of the high cost
components of a product both the C-intercept and the slope of the
straight line curve may be determined. Hence if the total number
289
of components within a product are known the total cost of a product
may then be estimated. The obvious advantage of this product costing
method is the reduction in the total number of components that need
to be accurately costed.
However this method when used to cost various products manufactured
at H. M. Ltd. resulted in large inaccuracies being obtained. That is,
estimated costs were found to be up to three times greater than actual
product costs.
A model of the type expressed by Equation 43 was found by Swan
and Worrall(183) to be useful in solving various industrial and
organisational problems. For example, a small number of shoe models
were found to account for the majority of annual sales hence forecasting
sales could be improved. The "over costing" effect experienced when
using Equation 43 can be explained since Swan and Worrell indicate
that the slope of the straight line curve obtained from Equation 44
decreases at a distance along the Nu-axis (Figure 51). Examination
of H. M. Ltd. 's products indicated a similar effect (Figure 52).
Determining the coefficients f and g using various percentages
(10-50%) of Nu did not significantly improve the estimating accuracy
of the models developed. In addition developing cost models using
greater than fifty percent of a product's components would not
seriously reduce the amount of detailed cost estimating required.
290
-FIGURE 51 LOG-LOG GRAPH OF THE PARETO DISTRIBUTION
Z
"- C-axi s %rber at ? Wotiows ""
C Tatirr (i. e., t"l, : "1s2.3"1": s3. etc. ).
soo
I
ISO
too /;, i ;J
ed
10
I.
J6
Ref. (183)
I,
dc dNu
t_34S6 7$ 9 10 IS "
:o4; rwy rri., rity
Nu-axis " Organization problems. Croup priorities.
291
FIGURE 52 LOG-LOG GRAPH OF CUMULATIVE COSTS
FOR H. M. LTD'S PRODUCTS
N
4--l ý 0 ý
rn ý 4-J
U
a) ý
Q) >
"r 4--3 r6
U
100
50
10
1 1
ý -ý.
1 i
t.
- ---r ... .
. 't
5 10
Cumulative Percentage Components
A-1.6 Tonne New EHB
B-3.2 Tonne New EHB
C-0.5 Tonne Electric Chain Hoist
f
50 100
292
Values for coefficients f and g that produce an acceptable
estimating accuracy are difficult to determine since the point
dc (Figure 5) at which the slope of the straight line curve Uffu
changes cannot be predicted. However, it is reasonable to assume
that the relative values of f, g and Nu affect the position of point
dc . Visual examination of previous error levels and values of f, g
BNu
and Nu indicated a relationship may exist. In order to test this
assumption and hence quantitatively determine the relationship
involved, the following actions were taken:
c>> From existing H. M. Ltd. products the values of f and g
were determined using the 10 highest cost components.
(ii) Using the regression technique the relationship
between total product cost (the dependent variable)
and f, g and Nu (the predictor variables) was examined.
To ensure that the variables are linearly related the
logarithmic form of each variable was entered into the
regression procedure.
(iii) Various combinations of independent variables were tested
in order to determine the model with the highest
estimating accuracy. The combinations tested were:
Log10 C= Log10 f+g Log10Nu
Log10 C= Log10 f+g+ Log10Nu ............... (46)
Log10 C= Log10 f+ Log10g + Log10Nu
293
Using the method of determining equation accuracy developed
previously (S and E) an expression of the form shown in Equation 46
was found to yield the highest correlation between dependent and
independent variables. However the model developed was not considered
sufficiently accurate for product costing since 401o of the product
costs estimated had errors greater than +15%. When the products
producing unacceptable estimating accuracies were examined for
possible causes the following reasons were found:
(i) The highest cost items used to determine the regression
coefficients were found to be inappropriate since the
costs used were those of sub-assemblies and not of
individual components. Hence cost items need to be of
a common form.
(ii) When determining Nu items such as washers, screws,
circlips, dowel pins, locknuts, precision nuts and
split pins had been included. However, such items
are normally considered consumables and are therefore
included in H. M. Ltd. 's indirect overhead costs. Hence
the inclusion of such items when calculating Nu can lead
to inaccurate estimates being obtained particularly in
products that require large numbers of such items.
294
(iii) The coefficient f represents the cost of the most
expensive component within a product. Although the
regression procedure estimates the value of f it was
found that using the actual value of the highest cost
component improved estimating accuracy.
Upon correcting the input data the costing accuracy of
the subsequent regression model (Equation 47) was
improved.
C=0.5916 f0.9948 101.209g Nu0.162 (47)
Comparisons of actual and estimated costs using
Equation 47 are shown in Table 41 and Figures 53 and
54.
Table 41 shows that:
65.2% of estimated product costs are within +10% of actual values
91.3% 11 11 11 11 11 it +15% " it 11
8.7% 11 If 11 11 11 ý +15% is 11 to
Although the small Proportion of products with high error
levels were re-examined no obvious reasons could be found
for these levels. It is therefore concluded that although
Equation 47 has improved the estimating accuracy it cannot
precisely reflect the point of deflection of the slope. The
overall accuracy is however within H. M. Ltd. 's current limits,
since E= -0.015 and S=0.096.
295
TABLE 41 COST ESTIMATES USING EQUATION 47
Nu fa Actual Estimated 0/
Cost Cost Difference
57 27.9 0.441 99.1 106.6 + 7.6
174 64.6 0.529 354.7 375.8 + 5.9
58 7.6 0.682 60.1 57.4 - 4.5
25 ,1 .50.594 7.7 7.7 0.0
175 107.1 0.471 522.8 529.7 + 1.3
29 1.8 0.639 10.2 10.5 + 2.9
178 198.1 0.453 907.3 931.1 + 2.6
33 1.9 0.642 11.5 11.9 + 3.5
40 20.0 0.519 80.2 89.7 +11.8 67 152.5 0.814 1637.2 1659.6 + 1.4
37 80.4 0.798 732.3 769.1 + 5.1
27 24.3 0.810 226.4 229.1 + 1.2
19 36.5 0.445 117.4 117.5 + 0.1
19 48.7 0.408 142.7 141.3 - 1.0
25 7.4 0.247 16.4 14.5 -11.6 202 63.0 0.397 279.0 280.5 + 0.5
40 74.0 0.653 434.6 478.0 +10.0 26 46.5 0.440 169.6 154.9 - 8.7
29 4.6 0.509 19.3 19.3 0.0
79 36.5 0.835 434.6 436.5 + 0.4
122 240.10 0.522 1381.0 1288.2 - 6.7
12 23.2 0.369 57.9 56.2 - 2.9
127 339.2 0.580 2220.4 2138.0 + 3.7
12 29.0 0.360 69.8 69.2 - 0.9
291 78.7 0.521 648.8 486.4 -25.0 45 5.6 0.535 29.7 26.9 - 9.4
60 8.4 0.666 70.8 60.8 -14.1 81 62.0 0.311 207.8 173.8 -16.4 31 1.0 0.732 7.7 8.1 + 5.2
34 15.4 0.223 30.0 29.5 - 1.7
63 9.2 0.67,1 80.0 68.4 -14.5
296
TABLE 41 (Cont'd)
Nu fg Actual Estimated % -' - Cost Cost Difference
39 140.6 0.664 840.8
73 3.3 0.526 16.2
60 "9.5 '0.727 95.0
20 2.9 0.676 15.7
29 0.7 0.858 6.8
22 1.1 0.671 5.9
70 0.6 0.799 6.8
77 9.7 0.719 106.9
91 15.5 0.578 83.8
566 45.8 0.546 390.8
63 11.2 0.785 138.7
31 1.0 0.617 6.9
202 69.0 0.383 244.9
70 27.5 0.590 166.4
52 67.4 0.509 315.6
931.1 16.6 81.3 18.2
7.4 6.6 6.4
84.7
93.3
338.8 113.8
6.0 274.2 164.1 305.6
+10.7 + 2.5
-14.0 +15.9
+ 8.8
+11.9
- 5.9
-20.8 +11.3
-13.3 -18.0 -13.0 +12.0
- 1.4
- 3.2
297
FIGURE-53 ACTUALVERSUS ESTIMATED PRODUCT COSTS
250
ý i/f ý W v
(A 4-) t/f 0
V
4-) U
0
C3-
r tö 7
4-, U
Q
200
150
100
50
<£300)
50 100 150 200 Estimated Product Costs (i's)
FIGURE 54 ACTUAL VERSUS ESTIMATED COSTS (>E300)
N
N d-3 Vf 0
U
4--3 U
-0 O i
CS.
+ýI U
200
150
100
50
250
500 1000 1500 2000
Estimated Product Costs (Its)
298
CHAPTER 6
THE DEVELOPMENT OF A DESIGN SCHEDULING METHOD
299
6.0 Introduction
The scheduling of components through the design process is similar
to the sequencing of operations within a machine shop where the
theoretical approach to developing effective scheduling rules
described by King (184) centres upon:
(i) Specifying an objective criterion by which alternative
schedules can be judged.
-(ii) Developing a method of ranking components in the order
with which their design should be completed.
(iii) Developing a method of determining the relative
importance of each component to the overall success
of a product thus enabling available design resources
to be effectively allocated.
In addition, scheduling the design process requires a method of
determining or reducing the probability of design changes occurring
since these changes could affect any design schedules developed. This
is analogous to the problem of re-working defective components during
manufacturing.
6.1 Specifying the Objective Criterion
The conflicting aims of the various departmental functions involved
in the new product development process can make the task of specifying
an objective criterion difficult. For example, a major concern of the
financial department is the minimization of product cost whereas that of
the manufacturing function could be the minimization of the manufacturing
time of a product.
300
However, the time available for development of the New EHB range
was limited by marketing policy. The objective criterion chosen was
therefore the minimization of development time.
The choice of this obective criterion reflects the growing
importance of development time to the ultimate success of a new product 11 e
in the market place (23 However the success of a product also depends
on its profitability and-design optimization. Therefore the total
cost effectiveness of any reduction in development time must be related
to the effect on design optimization and profitability.
6.2 Sequencing Component Design
The sequence with which components are designed is important to
the minimization of development time and forms the basis of conventional
project planning techniques such as PERT and CPM(80'83). Accordingly
correct sequencing of component design requires a knowledge of the total
time required to develop each component, i. e. the sum of the individual
times required at each of the development stages, and the inter-
relationships that exist between various development tasks.
- Since large numbers of components are normally involved in the new
product development process it is necessary to remove the dependence of
designers on the need to request component development time data from
other functional departments.
301
6.2.1 Estimating Component Development Times
Since the tasks required to develop each component within
a product are a direct result of the design process it is
reasonable to assume that the development time per task can be
related to specific design factors. In order to develop a method
for estimating component development times it is therefore
necessary to identify the design criteria associated with each
development task. Hence ascertaining the development time per
task would allow designers to directly estimate component
development times.
From discussions with relevant functional departments
(particularly the Purchasing Department), and examination of the
present and past NPD projects (i. e. Electric Chain Hoist), the
minimum, maximum and average times (Table 42) were estimated for
the various tasks involved in the NPD process. Using Table 42 it
is possible for an engineering designer to examine a component
design and identify the appropriate development tasks required
and hence determine the total development time per component.
The tables can be used in the form of a checklist to ensure that
all relevant development tasks are considered. Initially however
it is necessary to identify the interactions between activities
in order to determine those tasks that can be carried out
simultaneously and that would thus reduce the total development
time of a component. In this way a knowledge of such interactions
would allow "critical path" components to be identified.
302
TABLE 42 NPD ACTIVITY TIMES
X
Q rN CO LC) N 00
rr rN
r NO rd
O LD
Lt) LC) r
(Z LC)
O 00
N Lt1 LC) tC)
Ön . zr CO rd O tD to O O c+') O O O ýv ¢ N re) M LC)
F_ W. vz. 4 r H
. r Z
CD Mr L1ý rr ON NO O Lo Ln O r ý Z r r r M
11 U O
ý N
ýc
CL öö
CCCC "r L "r S.. t Q) t Q)
4-) a-ý "r XrX 3 a) 3 a)
r- ý- tio -0 vv
N r- 0 O
O)
N 4)
ý
ý
cr u
Q
.ýý ý tu
.ýv
tu +1 de -v c
"r N d)
-0
"r S.
"r
cr u
Q
N rn
.ý i
-0
(L)
4-3 O
O L
G)
h N
ý--ý
(A 43 tý
ý O
CL
a) >
. r. 4J . r. 4-3 (L) Q E O V
L 4-) .ý
N C 0 N
"r S- co a" E 0 V
C ý
"ý N a)
a) Y
ý
a) .ý
¢ä¢ä
4-3 0 C
aý VN +ý o a)
N E ,G
+)
U O
. }ý Vf
ý O
Uf C 'Q Y C1 L) (A vf 0 C)
+1 0 Kn -v 0 EN -p
a) Y i
"ý- O C) a) +ý "r- "ý YNý ý-- VIC. C. O s= C- C_
+ý OC NCNN
r0 a r0 .GU vvvýv
aý 10 (0
r s- s-
4--) C) C) ý "r "r C) r- r S- C- CL S- C_ C- CCC t) NN
. -. ý . -. . -. e6 CUü vvvv
CL d EE 4) O +) 0 "ý U0U "p a) a) v
NtNtC r' a-1 r-" +-ý R) fO fCT
"r C "r CN L "r Si "r 'fl 4) .C (U L
r+36 4-3 4J ýO
ý3ý "3
E
C C) C a) N fC N (0 tN r"
0 0
ý 4J >1 4-J e-- C r-- +) () +) CCC t0 G! 0 a) "r
Cý L O. L. S. EL 4)
C) Ü
C) VCD. 7
r-- NM mmm
¢ C]
303
TABLE 42 (Cont'd)
X r O O O LP) O O N N O Q r d' N r N
N
M}-4 U) Ö U) O O O U) O r r LC)
rr Q N 00 r r r f- LLJ U= cc P-4
F" r LA LO
r-ý O N O O U) O LO O O N
Y U 0
4-3 t/N
-v ý V1
ý >� ý ý
0 NO
Y UN 0O ýO N .0
LL NU
ý "ý "r Nrr ýaa s. aa UNN
tu -0 U Z7
ý a) E
"r
ý aJ
ý
"r ý
a) F..
R: r m
c co 0 ýr N
C) ý (A Co v) Lin C) N c. ý Oý i lp 01 C1 MM
N 04 ý ýww a1 r 01 Crr ONN d ýww
o er ý l. i O
NN Vf
aý ww LM C)
ýO NN
aJ Vww
N 01 4-3 OO V1 NN
0 ww U
a, CD CNN ., _
rd t0 -0 ý ýý v
ý U N V
ý . -. ýý vv
ý c b ý ýa .ý L ý b ý 3 rt
Q. '
r- c N G
Up
304
Maximum and minimum times for each development task have
been included in Table 42 to enable designers to gain. a realistic
outline of the limits of the possible development time required.
Although Table 42 provides a quick estimation of development task
times the actual development times should be obtained when
searching for potential material suppliers.
If the cumulative percentage activity time of the various
tasks is plotted against number of tasks a Pareto distribution
(Figure 55) is obtained indicating the possibility (as with
Stock Control policies) of ranking components in order of
length of development time required. This procedure would
identify those components that could form the "critical path".
Hence it would be possible to isolate the critical components
for which accurate development times must be estimated.
305
FIGURE 55 CUMULATIVE AVERAGE TASK TIMES VERSUS CUMULATIVE NUMBER OF NPD TASKS
/
CD Co C)
t. 0 C) cr
CD N
(%) sawýl Jsel anýTpLnwno
0
Lr) M
M
LO NN
Y t/1
iJ
G d Z
O N 4- O
i. a)
.0
LC)
a) ý
. r..
O r- U
LO
306 -
In addition to the development tasks listed in Table 42
there are other NPD activities (Table 43), the task times of
which are difficult to obtain from past NPD projects since the
factors that affect these tasks vary considerably between
projects.
. le
TABLE 43 FACTORS AFFECTING PROJECT ACTIVITY TIMES
Activity Influencing Factors
1. Prototype (a) parts marshalling Number of standard sizes, (b) assembly complexity of product design
and type of products
2. Pre-production (a) parts marshalling Launch quantities required, (b) manufacture sales forecasts, product type, (c) assembly design complexity
3. Launch (a) parts marshalling Sales forecasts, product type,
(b) assembly design complexity
(c) stock build up
4. Test area organisation Type of product, type of tests carried out, length of testing period
5. Plant layout/rearrangement Level of changes required, type
of plant layout adopted
Removal of obsolete stock Present stock levels, rate of sales
7. Design and Manufacture of Sales quantity required, cost
Production/Assembly aids feasibility of manufacturing aids
307
6.2.2 Interactions between Development Tasks
The two types of interactions involved in project
scheduling(55) are:
(i) Precedence Interactions
O
_ý Task A
Iýi Task B
F--* Task C
I-a
Here task B is preceded by task A (which must be
carried out first) and leads to task C (that must be
carried out next). The project scheduling objectives
are therefore to determine the order in which tasks
should be carried out and the total development time
required is the sum of the individual task times
involved.
(ii) Simultaneous Interactions
Tas k A
Task B
Task -01 c
Here tasks A, B and C can be carried out simultaneously
(i. e. in parallel) and it is the task that requires the
longest time to perform that is used to determine the
total development time for a component.
308
It was possible to classify the tasks involved in the
New EHB development process using the two types of activity.
interactions and hence construct a network diagram (Figure 56)
for the development of individual components. Task interactions
are described in Section 6.2.3.
' An element not shown in Figure 56 is the need for an
information feedback system from each of the NPD tasks to the
design stage since the design function should be involved if
corrective action is required.
6.2.3 Types of Activity Interactions
(1) Design Tasks of individual sub-assemblies can normally
be carried out simultaneously (except when combining
sub-assemblies) if sufficient design resources are
available. In general however lack of sufficient
resources requires the design of sub-assemblies to
be sequential. In addition the design of each
component within a sub-assembly is a precedence
interaction since this is the basis of creative
design.
(ii) Procurement Tasks can be carried out simultaneously
since the time involved in placing procurement orders
is minimal when compared with the delivery period of
the supplies ordered. That is, procurement task times
are basically "waiting periods" which can occur
simultaneously since they require small amounts of
development resources.
»-
309
FIGURE 56 COMPONENT NPD NETWORK DIAGRAM
4" O
0 i Cl.
(L) CL >,
}ý N ý--
ý 41
rn c . "- ý V1 N F-
aý S. -
4)
a v
"r 4-J
ý .r 4J C. I fC3
u` (0 -1b. 4-
rn .ý ý N ei ý--'
(A N
F"
CCC O00
41
OýOO
ýO -0 M
"ý O. "r- ""- U iÖLi
"OII td U rti mO UUUi
r- N Q. N t0 NýN
- U). - (41 4-)
t6 "- RS -0 td t3l : -) i +1 E +i r-
"r' N CL
Otz CGN nG
Ö Z"r" ZbZU
. ý,
310
O
(iii) Quality Checks (Incoming Supplies) can be regarded
as simultaneous activities since the time required
to test materials, components and tools is small
when compared with the overall development period.
Components that determine the critical path through
this stage are those that either require long term
reliability tests or must be tested outside the
company.
(iv) Prototype Manufacture of Sub-Assemblies are subject
to precedence constraints. Sub-assemblies must be
sequenced in order of the length of the prototype
test period required. For example, the prototype
hoist assembly of the New EHB had to be developed
at the beginning since long term reliability trials
had to be carried out whereas the suspension frame
and bottom block (hook) sub-assemblies required only
short term function testing.
(v) Prototype Manufacture of Components within each
Sub-Assembly. The total time required to manufacture
a prototype component consists of a "waiting period"
(the time a component must wait before a manufacturing
operation begins) and a "service-time" (the time to
carry out an operation). Since "waiting periods" are
in excess of "operation times", i. e. measured in days
as compared with hours for "operation times" the
latter can be ignored when estimating total
manufacturing times.
311
lo,
(vi )
Manufacturing activities between components can
occur simultaneously (even though more than one
component may require the use of the same equipment,
since in these situations all "development components"
are normally batched for manufacturing purposes)
unless components need to be processed together
(e. g. precision machining and welding fabrications)
when precedence constraints are imposed.
The "waiting periods" for each operation carried out
on an individual component must however be added
since precedence interactions exist, i. e. the
component incurs a "waiting period" between each
manufacturing operation required. Hence total
waiting time is the sum of the individual "waiting
periods".
Post Manufacturing Quality Checks can occur after
each operation required to manufacture a component
and are therefore subject to precedence constraints.
This is particularly important to consider when
quality checks are sub-contracted. However between
components, quality checks can be regarded as
simultaneous for similar reasons to the testing
of incoming supplies, i. e. testing time is minimal
when compared with overall development time except
when reliability trials are required.
312
(vii) Prototype Testing is an important precedence task
since it determines the order in which sub-assemblies
are designed and manufactured. The order in which
sub-assemblies are designed is dependent on the
length of the prototype test period. Those sub-
assemblies not directly involved in long term
reliability trials can therefore be placed at the
end of the design schedule.
i
The actual prototype testing of sub-assemblies can
of course be carried out simultaneously.
(viii) Jig and Fixture Design are precedence tasks when
limited tooling departmental resources are available.
The actual manufacture of jig and fixture components
are subject to identical task interactions as the
manufacture of prototype components.
When jig and fixture design and manufacturing tasks
are sub-contracted then they could be carried out
simultaneously.
(ix) Procurement of Pre-production Materials are tasks
similar to the procuring of prototype materials and
are therefore subject to identical task interactions.
313
(x) Pre-production Manufacturing tasks are similar to
prototype manufacturing tasks. That is, the tasks
are normally simultaneous between components but
operations carried out on individual components are
subject to precedence constraints.
(xi) Product Launch is a major precedence task since it
could determine the order in which the standard sizes
within a product range are developed. In the case of
the New EHB project it is desirable to develop a
limited product range (e. g. 0.8,1.6 and 3.2 tonne
units) in order to reduce the total development
period that elapses before revenue can be obtained
from the new products.
6.2.4 Accuracy of Estimated Development Times
It is important to recognise that Figure 56 represents the
network diagram of a component that must undergo each task
involved in the development process. Not all components within
a product design will require every development activity. For
example, bearings are purchased components and hence require no
manufacturing and jig and fixture building tasks.
Since the order in which product sizes and sub-assemblies
are developed is determined by major precedence constraints
(e. g. marketing constraints and length of prototype testing
period) the designer need only consider the order in which
314
components within sub-assemblies are designed. In addition
the order in which these components are designed is determined
by the amount of development time required to prepare components
for prototype assembly. Hence the use of Figure 56 in conjunction
with development activity times (Table 42) allows designers to
estimate the total development time of each component within a
product design. Examples of development time estimation are
illustrated in Figures 57 and 58.
Upon completion of the prototype testing stage engineering
drawings are the tools that allow development tasks between
components to be carried out simultaneously (although for
individual components each development task must be carried
out in the correct order) except when components need to be
manufactured together.
The validity of estimating development times (using
Table 42 and Figure 56) was tested using the actual development
times for the 0.8 tonne and 3.2 tonne New EHB components. The
results shown in Figure 59 indicates that close agreement is
obtained between estimated and actual development times.
However a proportion of the points indicate that differences
exist between estimated and actual development times. These
large variances arise primarily since in actual practice many
tasks that could have been carried out simultaneously were not,
i. e. development activities were not scheduled such as to
minimize total development time.
315
FIGURE 5.7 DETERMINATION OF DEVELOPMENT TIMES - PARTS A AND B
PART A
OUTPUT GEAR
Procure Metric B3c Hobber 30
DESIGN
i. __ _
Procure B2a ýaterial 6 Procure Metric B3c Cutter 30
4r
z
Examine Dla 1 JTurn, Face
Bore(202) Cla 10
1Cut Gear Clb 15 Teeth 217
Examine D2 5
Surface C2 10 Harden
Examine D2 5
. +º ý 4r-- Turn Faro
Procure Bla Material 0.5
nle 1Iý. ý,,. �. MV\. ri.. 9A
Bore 2fl4 Ala iu
Cut Splines
PART B
BEARING HOUSING
217) C1b15
Examine Dlb 1 f .. A 1
/ 1/
ý
Task and Task Code
ý t I t L_ Task
Tine
Total = 76 days PROTOTYPE Total = 56.0 days ASSEMBLY AND
TEST
316
FIGURE 58 DETERMINATION OF DEVELOPMENT TIMES -- PARTS C AND D
PART -C',
GEARCASE
Examine Dla
Total = 122 days
DESIGN
PROTOTYPE ASSEMBLY AND
TEST
PART p
ENDPLATES
Total = 47.5 days
317
The cumulative distribution of development times
estimated for each standard unit were found to form a typical
Pareto distribution (Figure 60) indicating the possibility of
reducing the level of management control resources by operating
a "management by exception" policy. This would involve
directing management resources to the control of those i
components with the longest development times.
6.3 Determining the Relative Importance of Components
According to Ostrofsky(56) engineering design requires the
development of alternative design solutions and the comparison of
these solutions in order to choose the most appropriate. However,
this design methodology is difficult to adopt in projects with limited
development and design resources since there is frequently only
sufficient design time to develop an initial solution. Hence to
ensure that available design resources are used effectively it is
necessary to adopt an alternative design methodology. The stages in
such a methodology would involve:
(i) developing the initial design solution.
(ii) determining the benefits to be gained by allocating
extra design resources to improving the design,
(iii) determining the relative importance of these extra
benefits to the overall success of the product,
and (iv) use of this relative importance to allocate available
design resources.
318
FIGURE 59 ACTUAL. VERSUS ESTIMATED DEVELOPMENT TIMES
100
N ý
E
"r
4-3 cl UJ
O
N
4-) U
Q
80
60
40
20
20
FIGURE 60
40 60 80 100 Estimated Development Time (Days)
PARETO CURVE OF THE 0.8 TONNE EHB COMPONENT DEVELOPME. 1T TIMES
ae
a) E
t- +J c a) E 0 a) a)
a) >
"r r0
r
80
60
40
20
20 40 60 80 Cumulative Number of Components (ö)
319
Determining the relative importance of design solutions by
necessity requires a common basis of comparison with which to relate
designs that could perform quite different functions. Examination of
the PPBS project planning technique (Section 2.5) has shown that the
design process can produce many types of benefit (e. g. improve
producibility, reduce manufacturing cost, improve quality and widen
market specifications). However, manufacturing cost is the most
convenient for comparison purposes particularly since the regression
based cost estimating system previously developed allows designers
to quickly assess manufacturing costs.
To allocate the available design time the Pareto curve of
cumulative product costs versus number of components is useful since
the cumulative cost axis can be converted directly to available design
time. The amount of design time to allocate to each component can
thus be obtained as a direct measure of the relative cost of a
component.
The Pareto distribution could only indicate the design time to
allocate to each component since factors other than cost need to be
considered. However it is an advantage to use such a chart since the
designer must then weigh the additional benefits to be gained with the
cost (in terms of design resources) required to achieve them.
320
6.4 Preventing Design Changes
Design changes have been estimated (Section 2.6) to consume a
considerable amount of design resources (2250 hours). Since a high
proportion (78%) of design changes resulted from the need to improve
design cost effectiveness it is reasonable to assume that an improved
cost estimating ability would enable the number of design changes that
occur to be reduced. That is, the cost effectiveness of designs can
be ascertained and improved prior to their release to other departmental
functions. Hence the regression based estimating system could be of
benefit since cost estimates can be quickly prepared and help to
reduce the-pressure on the design department to prematurely release
design drawings in order that prototype manufacture can begin.
In addition the possibility of having to change designs is a
prime reason for not releasing design work in a smooth flow. However,
releasing design work in a smooth flow of "stabilized" component designs
would ensure that high workloads do not arise in subsequent development
areas such as jic and fixture manufacture. The estimating system
developed allows design costs to be established (and therefore designs
stabilized) earlier in the design process. Hence the ability to
release designs in a smooth flow for subsequent development activities
would be improved.
321
CHAPTER 7
COMPUTER AIDED NPD
322
7.0 Introduction
The growing use of computers to aid various manufacturing,
planning and control tasks (185,186) will in the future undoubtedly
include the NPD process. The immediate challenge according to
Hancock (187) is to devise systems and procedures that provide the
links to integrated manufacturing systems since progress with sub-
systems has reached the stage where the majority of the tasks involved i
in manufacturing industry can be automated.
With respect to the development process it is the design phase
that has received most attention with the development of a wide variety
of Computer-Aided Design and Draughting (CAD) systems. Systems have
been developed by both academic institutions and commercial enterprises
who have analysed the needs of a drawing office, studied the manner in
which a design evolves and produced CAD packages (e. g. Distributed
Interactive Automated Draughting (DIAD)(188), Applicon(189), Polysurf(190,191)
IV(192) and Ferranti (193) )Comdraw , to meet the majority of the require-
ments possible in keeping with a reasonably priced total system.
H. M. Ltd. intend to install an Autotrol AD/380 system(194) which
although not the lowest cost CAD system available (i. e. £150,000 compared
with the Ferranti system cost of £llO, 000), has the advantage of being
marketed by Davy Computing Ltd. (DCL) who in common with H. M. Ltd. are
part of the Davy Group of Companies. The primary advantage of purchasing
the Autotrol system is the ability to use DCL's own CAD system, EUCLID(195)
for training purposes at a relatively low cost. H. M. Ltd. use the
EUCLID system on DCL's Univac mainframe computer through an existing
323
computer link. Since the actual plotting of drawings is carried out
at DCL, H. M. Ltd. have only to purchase and install low cost peripheral
equipment such as an interactive graphics unit.
The Autotrol AD/380 system however runs on a dedicated mini-
computer which is part of the package purchased by the user company.
The system supports peripheral equipment such as workstations, plotters,
digitisers and teletypes and since it has magnetic tape and disc
storage/reader unit facilities, the system can compile in-house
libraries of data.
The drawing language. of the Autotrol system is similar to that of
the EUCLID system and uses "every day" words. Hence the changeover from
EUCLID can be accomplished quickly. In addition data files generated
using the EUCLID system can be transferred to the Autotrol system.
Features that make the Autotrol system preferable to the EUCLID
system are:
(i) The ability to zoom into or out from a particular area
of a drawing.
(ii) The ability to draw in refresh, i. e. if the refresh
command is invoked a particular feature can be moved
around the screen and positioned as required.
3 24
(iii) The ability to invoke a series of commands by the touch
of a single key. The AD380 workstation has a 240
position keyboard which can be overlayed by a menu card
and then touching one of the 240 keys invokes the set of
commands which have been assigned to that key. Since
there is no limit to the number of different menu cards
which can be set up, the 240 position keyboard can be
considered as multi-layer rather than single layer.
Although the advantages obtained through using CAD systems have
been well documented(196,197) the level of benefits achieved are
dependent on the specific design application. Hence Laxon (198)
considers it necessary to have a clear idea of the objectives
(Table 44) to be achieved before embarking on the selection and
implementation of a CAD system.
TABLE 44 OBJECTIVES OF USING CAD Ref. (198)
1.
2.
A better product through improved design. CAD may result in a
reduction in design errors and in improved performance at lower
costs through a better optimization of the design parameters.
Solving design problems which cannot be easily tackled without
a computer. Examples include finite-element analysis, computer
representation of surfaces and computerised design of integrated
circuits.
325
TABLE 44 (Cont'd)
3. Reduced design costs through a reduction in manhours. The
replacement of engineers and draughtsmen by a computer system
can lead to substantial savings in direct costs. These savings
will increase in time as salaries rise and computers become
cheaper.
4. A greater design capacity. Many companies cannot reach their
full productive potential for a product because of a bottleneck
in the design office. If it is impossible to recruit designers
and draughtsmen, CAD may offer a solution.
5. Reduced design time. The use of CAD at the conceptual design
stage can reduce the lead time needed to get a new product onto
the market. For detailed design CAD can result in a faster
response to customer's queries and faster delivery.
Determining the benefits to be achieved is not straight forward
when using a CAD system as an aid during the NPD process, since the
relative proportion of resources expended on the individual activities
involved in the design process varies considerably between NPD projects.
This is influenced by such factors as:
(i) The innovativenessof the product design. The more
original the product design the greater the proportion
of time expended on design creation activities.
326
(ii) The type and number of functions the product carries out.
Products such as automobiles and heavy cranes require
large amounts of stress, function and performance analysis.
(iii) The complexity of the design. This factor could increase
the amount of detailed draughting work required.
Hence for a particular NPD project it is necessary to determine
the relative amounts of resources expended on the various activities
involved in the design process. The approach adopted during the New
EHB development process for achieving this objective involved the use
of the Activity Sampling work measurement technique(113). The
activities required to prepare various assembly designs from their
conception to the preparation of production drawings were examined
at random intervals in order to determine the proportion of time
(Table 45) a designer spends on each type of task.
When examining the advantages of CAD systems to the development
of the New EHB range, Table 45 allows the relative importance of each
CAD facility to be recognised.
327
V
TABLE 45 PROPORTION OF TIME SPENT ON DESIGN TASKS
Function Analysis
Data Acquisition and Transfer e
Design Creation and Improvement
Cost, Marketing and Manufacturing Analysis
Draughting
Integrating with overall NPD process
7.1 Function Analysis
Proportion of Design Time
0.46
0.23
0.12
0.08
0.06
0.05
In their basic form CAD systems are merely automated draughting
and data storage systems and hence do not have the capability of
performing the complex mathematical calculations required for
functional and performance analysis of designs. However software
modules are available and can be included in the total CAD package
to allow advanced analytical design to be performed.
Unfortunately the development of software in the mechanical
engineering area has tended to be piecemeal in nature. This results
from the general diversity of the topic and the difficulty in
analytically predicting the overall performance of a mechanical
system in terms of the performance of its individual components.
In certain fields such as finite element analysis(199) and
bearing design (200) "off-the-shelf" software is available. However,
during the development of a new product there is often a need to
develop "customised" software for specific analytical requirements.
For example, the design of trim cages for controlling the velocity
of flow within gas valves (201)
and predicting stability criteria
for four wheeled vehicles in the Railway Industry(202)
With respect to the New EHB development software would need to
be designed for:
(i) Modelling rope twist during lowering of hoist load.
(ii) Estimating drum tube thicknesses.
(iii) Determining motor and brake sizes.
(iv) Modelling hoist deflection under load.
(v) Determining rope diameters required.
However since such software would be useful only during new
product development and therefore infrequently it is doubtful that
its use would be economic when compared with the present manual
methods used.
329
7.2 Data Acquisition and Transfer
Since a major function of CAD systems is to provide data storage
facilities the acquisition and transfer of data during the NPD process
can be expected to be greatly improved, i. e. CAD systems allow
extensive data files to be maintained and information retrieved
rapidly.
-A wide variety of data can be stored including:
(i) existing product data,
(ii) competitive product data,
(iii) parts lists,
(iv) material specifications,
(v) component details (e. g. material, dimensions and
tolerances),
(vi) manufacturing data,
(vii) cost data,
and (viii). component, assembly and product drawings.
The ability to store and quickly retrieve data aids in the
development of an effective product design. Since engineering drawings
can be electronically stored a complete definition of a component can
be filed. This facility is of benefit since the capturing by the
330
computer of the definition of a component at the design stage provides
instant data transfer facilities for associated manufacturing and
business operations. For example, the Autotrol AD380 basic software
package has the ability to create a full parts list of the components
used within a particular assembly. This information when placed on
magnetic tape or disc can be transferred to the main frame computer Oe
handling such processes as stock control and product structure files.
The need to write out a parts list-remote from the drawing board and/or
transpose the parts lists into punchcards is therefore unnecessary.
Hence the interface between the design function (where information is
established) and the data processing department (where information
is processed) could be improved.
The amount of data stored on an engineering drawing can be
significantly increased producing benefits to the engineering department.
Using manual methods, everything written or displayed on a drawing
occurs on a single layer namely the sheet of paper. Hence a great
deal of information which could be usefully stored on the drawing has,
because of lack of space, to be stored using other means.
However the Autotrol AD380 system produces a total drawing by
storing different layers of information. Hence if layers 1 to 4 were
utilised by the designer, layer 5 could be made available to the Jig
and Tool designers. Layer 6 could then be allocated to the production
engineers to store data on production methods. A centralized library
of information could then be compiled.
331
To transfer data it would be necessary to provide access to the
-CAD system for the various functional departments involved and to ensure that appropriate security systems allow design changes to be made by
design personnel only.
The manufacturing functions will benefit from a CAD system since
it could be possible to produce drawings reflecting the operations to
*take place for individual types of processes such as, welding, casting
and machining. Presently drawings attempt to convey information to
both a founder and a machinist or a welder and a machinist, hence the
machine operator or welder must search drawings and identify only
relevant information. However using the layering technique, and an
appropriate structured drawing number system, 'it is possible to
produce separate fabrication and machining drawings or separate
casting and machining drawings at no extra effort to the designer.
The growing use of numerically controlled machine tools within
H. M. Ltd. also benefits from CAD systems in that component geometry
can be recalled for use in the preparation of NC tapes.
7,3 Design Creation and Improvement
Product design is essentially a creative process although the
level of creativity involved is dependent on the innovativeness of the
design. The amount of creative design involved in the development of
the New EHB did not represent a major proportion of design time since
332
the level of product innovativeness was limited, i. e. the New EHB
design is similar to that of several competitors products. This was
a deliberate policy to enable development time to be reduced.
Design improvement also takes place primarily by creative
synthesis although various optimization techniques such as Linear
Programming(55) and Dynamic Programming(78) can also be used.
Indeed it is through the use of optimization techniques that the
difficulty in transferring the creative element of design from a
manual task to a computer task can be illustrated. 'For example,
the solution to the "assortment problem" (Section 2.3) indicated
that there are many subjective factors involved in the design process
which cannot be taken into consideration when using optimization
techniques.
According to Masters(203) human creativity and judgement will
always be necessary in design projects where:
(i) the design work is not a mathematically exact process,
(ii) there are many variables and influencing factors
involved,
(iii) it is difficult to define when a design is "acceptable",
(iv) the number of decisions to make are immense hence human
intuition or judgement is necessary to solve a problem,
333
(v) where the work being performed cannot logically be
defined,
(vi) where the problem in hand cannot be specified to a
computer,
and (vii) when the volume of input data for problem specification
is large, i. e. input specification may be as great. a
task as producing the design manually.
Since the large proportion of original design problems contain
one or more of the above elements it can be seen that CAD systems have
little to contribute to the initial creation of design solutions.
Once a design solution is generated however computers are ideal tools
for design optimization providing the optimization procedure is
correct (204)
and can be logically defined.
In addition to CAD systems being of limited use to the creative
design process, Cooley (205) suggests that "the crude introduction of
computers into the design activity ... may result in a deterioration
of the design quality". This argument is based on the known effects
of introducing high-capital equipment with an increasing obsolescence
rate into the manufacturing environment. Essentially employers seek
to alter the working conditions of their workers in order to exploit
high-capital equipment 24 hours per day. In consequence companies
could seek to impose on the design staff (as occurred at Rolls Royce,
Bristol (205)) the following conditions:
334
(i) the acceptance of shift work in order to exploit high
capital equipment,
(ii) the acceptance of work-measurement techniques,
and (iii) the division of work into basic elements and the setting
of times for these elements, such times to be compared
with actual performance.
Overall there would be a tendency for designers to become paced
by the machine and since CAD systems can produce quantitative data at
a high rate the stress on the designer could be enormous.
Clearly the conditions described by Cooley(205) would not provide
a working environment in which factors that improve the creativity of
a designer (206)
could be generated. These factors are:
(i) A designer's level of self confidence.
(ii) Adoption of a positive attitude.
(iii) Adopting an open mind.
(iv) Adopting an attitude of constructive discontent.
(v) A designer having the courage of his own convictions.
lle
335
The use of sophisticated CAD systems capable of carrying out
detailed functional analysis of designs could be expected to reduce
the design resources required to generate a design solution. As a
result two strategies are possible. Either the amount of design
resources required during a NPD project can be reduced or the number
of alternative design solutions generated could be increased. Given
the adverse effect of the introduction of CAD systems on the creative
process the latter strategy would appear difficult to achieve.
7.4 Cost, Marketing and Manufacturing Analysis
Cost, marketing and manufacturing analysis of design work is
essentially a data collection process with the relevant data being
generated at present by the appropriate departmental function. Hence
the actual examination of this data by the design function represents
a small proportion of the total design time. In addition the limited
development time available restricts the number of design alternatives
that can be generated hence reducing the amount of cost, marketing
and manufacturing analysis that needs to be undertaken.
Since it is the rate at which data can be generated that influences
the effectiveness of the design process, the ability of a computer system
to manipulate large amounts of data could allow designers to generate
their own cost, marketing and manufacturing data in order to avoid
delay to the design process.
336
The cost estimating system developed helps achieve this aim by
providing designers with a powerful tool for generating cost data for
their designs as they evolve and hence enabling cost effective designs
to be more readily achieved.
The generation of manufacturing data is difficult'since the
factors involved are complex and varied. The solution to the problem
according to Colding(207) and Coates, et. al. (208) is the establishment
of manufacturing data bases or manuals. The effectiveness of data
bases would increase with the growing use of NC equipment since they
allow optimum manufacturing conditions to be selected.
" The estimating system developed by the author could form the basis
of a manufacturing data base. For instance, the costing equations for
difficult production operations that the manufacturing function wish to
avoid could be removed. Format statements could then be inserted
asking the designer to consult the manufacturing department before
stabilizing the product design.
In addition the use of a particular costing equation could cause
appropriate manufacturing requirements and cost reduction suggestions
to be displayed. For instance the use of an equation to estimate the
times required to bore a hole with tolerances of <0. lmm could initiate
the display of such statements as:
337
(i) Avoid blind holes.
(ii) Can these tolerances be relaxed?
The generation of manufacturing data bases and their use by
designers is possibly a long term solution to the problem of analysing
designs in'terms of producibility. More immediate solutions would be
the construction of lists providing "producibility tips" for the
designer(209) and the use of a "data file layer" for a Value
Engineering committee. The advantage of using a VE committee is
the ability to carry out the multi-disciplined analysis of designs
during a co-ordinated design improvement effort.
7.5 Draughting
CAD systems are mechanical draughting systems in the sense that
they relieve the draughtsman of such tasks as dimensioning, scaling,
creating lead lines and arrowheads, calculating dimensions, cross
hatching and creating manufacturing symbols. In addition the user
friendly interactive graphics system enables drawings to be quickly
and accurately generated. However during the design of the New EHB
the task of draughting represented a small proportion of a designer's
overall time. Hence CAD systems could be expected to have a minor
effect in terms of reducing required design time.
338
CAD systems are suitable for designing small numbers of complex,
custom-designed products (e. g. heavy cranes) where design is
accomplished by using various arrangements of standard components
and assemblies. In these circumstances the ability of CAD systems
to hold large amounts of component drawings in file and quickly
retrieve them enables design time to be significantly reduced. In a
addition storing cost and manufacturing data will allow other
departments to function. For example, materials can be purchased
and tenders and manufacturing schedules prepared.
In companies that manufacture both special and standard products
the use of a CAD system during the design of a standard product range
could lead to the generation of ineffective product designs. This
may occur if files initially created for say jobbing quantities were
used in the development of designs for higher volume component
manufacture.
7.6 Integration with Overall NPD Process
The responsibility for co-ordinating NPD tasks at H. M. Ltd. is
that of the designer and a multi-disciplined NPD committee. The
present NPD management system allows effective control of the technical
development process to be achieved but the level of planning carried
out does not enable development time to be minimized.
339
A CAD system could improve this situation by allowing more
effective communication between departments. Also the use of a project
planning technique such as the design scheduling method developed in
the present work would allow overall development time to be minimized.
The future direction of CAD/Computer-Aided Manufacture integration
should, according to Knox(209), be directed at developing a "coherent,
cohesive information system" for combining a series of business tools.
Knox suggests that a forecast for a company's products and the
corresponding bills of material should be used to create part
requirements (Figure 61).
A Pareto cost analysis and segregation both by a family classi-
fication system and manufacturing polycodes should then be used to
determine both the type of manufacturing required and the correct
machine tools to perform the work.
r
340
FIGURE 61 INTEGRATED-INFORMATION SYSTEM
Ref. (209)
PRODUCT BtU. OF MATERIAL
341
CHAPTER 8
DISCUSSION
342
8.0 Introduction
In recent years the increased action of competitors and the rapid
development of manufacturing technology have increased the rate at which
companies must undertake product development. The new product development
-(NPD) process is gradually becoming a "continuous" process and demands
increasingly larger proportions of the technical, financial and
management resources of a company. Therefore, the management and
implementation of product development must change to reflect the need
to actively minimize development costs and improve the planning and
control of the NPD process.
,, An important effect of the increase in the rate at which NPD
projects must be undertaken is the need to consider a greater number of
alternative product design concepts. The work carried out at H. M. Ltd.
has_shown that the major obstacle to the economical comparison of concept
designs is not the synthesis of such designs but the generation of data
with which alternative concepts can be compared. In particular a major
limiting factor is the inability of conventional estimating systems to
generate quickly and economically the large amounts of cost data
required during the NPD process.
Two approaches are available for increasing the economic number
of alternative design concepts that can be costed during NPD. Both
approaches have been considered by the author and they are:
343
(i) reduce the cost of generating cost data, and
reduce the amount of cost data required to compare
designs.
In order to reduce the cost of generating cost data the existing
estimating, -system at H. M. Ltd. was examined using Nadler's(95) seven
system descriptors. The examination revealed that the work measurement
technique used to estimate manufacturing times of components controlled
the overall effectiveness and estimating accuracy of an estimating system.
In particular, the work measurement technique employed determines
the type of personnel who could estimate manufacturing times. Conventional
work measurement techniques such as the type used at H. M. Ltd. (i. e.
synthetic time standards and empirically derived algorithms) require
personnel with extensive background knowledge of the methods and processes
used to manufacture components. Cost estimating for the NPD process
normally forms only part of an estimators duties. Hence the responsiveness
of an estimating system is dependent on the priority given to the design
department's requests for cost data. The large amounts of cost data
required during NPD and the extensive detail required to estimate
manufacturing costs using conventional techniques exacerbates this
situation.
It has been explained how the responsiveness of an estimating
system can be greatly improved when cost data can be generated as the
design evolves. In order to achieve this objective the present work has
examined existing work measurement systems to identify those that can be
adapted for use by designers. The examination of existing work
measurement techniques is discussed more fully in Section 8.1.
344
An estimating system has been developed that reduces the time and
cost of generating cost data. The system has used regression analysis to
develop costing equations capable of estimating the manufacturing times
and hence labour and overhead costs of components for a wide variety
of production processes. The system which can be used by designers is
discussed more fully in Section 8.2. It is capable of estimating
manufacturing costs quickly and economically as the design evolves.
High cost items can therefore be quickly identified and cost reduction
exercises initiated. In addition the estimating system enables many
types of cost data to be generated, i. e. labour costs. of products,
assemblies and components, process and operation costs, the manual
and machine work content for manufacturing processes and costs arising
in individual cost centres. Hence the cost data requirements for functions
such as management accounting, budgeting, production control, tendering,
production engineering and marketing can be satisfied.
Because of the inability of H. M. Ltd. 's current estimating system
to quickly generate cost data component designs were often released for
prototype manufacture before they had been costed. This procedure was
necessary to prevent delay to the development process. The ability of
designers using the regression based estimating system to cost designs
as they evolve can, therefore, be expected to reduce the number of
design changes that arise in this manner. These design alterations
accounted for 13.7% of the resources available for designing the New
EHB range. It cannot be argued that reducing the number of design
changes that take place during NPD would necessarily reduce the total
amount of resources required to design the New EHB range. This is
I
345
because the resources required to achieve acceptable manufacturing costs
for component designs cannot be accurately determined since they are
dependent on such subjective factors as the experience and ability
of the engineering designers carrying out the design work.
However the possibility of design changes occurring can effect the
scheduling of development tasks adversely since production drawings
for designs are prevented from being draughted. Tasks such as jig and
fixture design cannot therefore be effectively scheduled, and high work
-loads on such departments frequently occur due to the inability of the
design' department to release component designs in a smooth flow of
finished drawings. Hence the reduction in the level of design changes
that would take place as a result of using the estimating system
developed by the author can be expected to significantly improve the
scheduling of development tasks with an overall reduction in lead time
to manufäcture.
During the concept and detail design phase of the NPD process cost
data is of paramount importance not only to ensure that manufacturing
costs are within pre-defined targets but also to allow the economic use
of decision tools such as:
(i) make or buy analysis,
value engineering, and
(iii) determining the optimum number of standard units and their
respective sizes to manufacture in a product range.
346
At H. M. Ltd. the extensive use of such design decision tools is
limited at the present moment due to the inability of the existing
estimating system to quickly and economically generate cost data.
The implementation of the estimating system developed by the author
can be expected to significantly improve this situation. Design
decision tools can then be used more extensively to enable cost
effective designs to be developed.
In particular, increased use can be made of the model (Equation 7)
proposed by the author for allowing designers to carry out "make or buy"
analyses. The model which enables both the purchase and in-house
manufacturing costs of designs to be quickly estimated by designers
is discussed more fully in Section 8.3.
Value engineering is by its nature a relatively slow process.
Normally large numbers of components need to be examined during the NPD
process, if a complex product such as a crane hoist is being developed.
To avoid wastage of design resources it is necessary to enable
design changes arising from value engineering to be implemented prior
to the prototype build phase of the NPD process. A responsive estimating
system such as that developed by the author, is therefore essential in
achieving this objective.
Value engineering must be integrated into the overall NPD process
to enable cost effective designs to be developed. This aim requires
effective control of the VE function. The model developed by the author
(Equation 9) can be expected to assist in this area by allowing calculation
347
of the number of components that can be value analysed taking into account
the time and resources available for the NPD process. Candidates for
VE could then be selected on the basis, of their relative cost or
importance to the overall success of the product design. Again a
responsive estimating system is essential to the selection of
candidates for value engineering. /
During the present work it was observed that the rate at which
cost reduction exercises could be accomplished was increased by taking
advantage of the "experience effect". Each cost reduction method listed
in Table 12 was selected in turn and then each component within the
total product design analysed with respect to a specific method. Hence
the'experience gained through examining a previous component enables
subsequent components to be more rapidly analysed.
The alternative approach to the problem of increasing the economic
number of design concepts that can be considered during the NPD process
involves reducing the amount of cost data necessary to compare product
designs. The research carried out by the author in this area involved
the development of models for forecasting the total costs of product
designs. The basic methods examined for developing costing models were:
(i) the determination of relationships between cost and design
specifications such as hoist capacity and the horse power
of electric motors, and
(ii) the use of a small number of the high cost components of
a product design.
348
The models forecast the total costs of a product design are
discussed in Section 8.4. Using these models detailed cost estimating
is not required, hence the time and resources necessary for costing
purposes can be expected to be significantly reduced. In addition,
since cost is a major factor effecting price and sales demand, a
knowledge of competitive product costs through the ability to quickly
estimate these costs would enable causal forecasting techniques to be
Used. This is of benefit to a company since causal techniques have
been shown to be suitable for forecasting the decline stage of the
market life cycle (PLC) of a product. The ability to forecast the
"turning points" of the PLC of a product enables the optimum time,
to be determined for developing a replacement or improved design
fora product whose sales demand is declining.
The model developed using a modified Pareto distribution estimates
the manufacturing cost of competitive product designs that would result
if the product was manufactured at H. M. Ltd. Hence only differences
in manufacturing costs between H. M. Ltd. 's and their competitors
products would be used to initiate the NPD process. The market price
of competitors products would not enter into the decision to carry out
product development work. This is beneficial since the price of a
product does not always reflect the manufacturing costs of the product
design. 'For example, marketing policy may require product prices to
be reduced artifically in order to capture a larger share of the
market.
349
The inability-of current forecasting techniques to accurately
predict the start of the decline stage of the market life cycle of
a product can lead to a situation in which new products must be
developed in a limited period of time. If this occurs to achieve an
optimum product design within the NPD period additional development
resources may be required. However to minimize the required design
resources a'greater degree of planning and control must be exercised
over the NPD process. Existing project planning techniques have been
examined during the present work and found to be inappropriate to the
development of a new product range. The disadvantages of such project
planning techniques as PERT, MBO, PPBS and CPM are:
(i) High cost of implementation.
(ii) Inability to integrate the important factors of the NPD
process, i. e. optimizing product design.
(iii) The need to predict the times required to carry out
development tasks prior to the design phase.
To allow a greater degree of planning and control to take place
during NPD a method has been developed by the author to allow designers
to estimate the development times of individual components. Hence the
sequencing of components through the design process with the objective
of minimizing total development time required can be accomplished. In
addition development resources can be allocated on the basis of the
relative cost or importance of components to the overall success of the
design. The ability to sequence components through design and allocate
development resources enables limited development resources to be used
effectively.
350
'_The ability to schedule design allows use of cost targets to make
a greater contribution to the reduction of design time'and the develop-
ment of a cost effective design. For example, if a cost target was
reached for a component before the allocated resources had been expended
then the unused resources could be employed more effectively elsewhere.
If cost estimates for a new component design fail to meet pre-defined
, targets then the size of the cost variance could be used to determine
the appropriate action to be taken. High cost variances for example
would indicate that a value engineering exercise was necessary whilst
the recovery of small cost variances would be the responsibility of
the designer. The scheduling method developed by the author is
, discussed more fully in Section 8.5.
In order to develop an optimum product design the tasks performed
by-the various functional departments involved in the NPD process must
be properly integrated. That is, all activities carried out must be
correctly sequenced with compatible objectives. A fundamental require-
ment for the integration of the activities of the departmental functions
is the removal of the barriers (interfaces) to information flow between
departmental functions. An alternative solution which the cost
estimating and scheduling systems developed by the author achieves
is to remove the need for information to flow across departmental
interfaces. The total integration of the NPD process could be
accomplished using a CAD system. Since H. M. Ltd. are installing a CAD
system, the opportunity was taken of examining the contribution CAD
makes to the NPD process. A valuable part of this contribution is the
extent to which CAD aids in the integration of the departmental functions
during NPD. The work carried out in this area is discussed more fully
in Section 8.6.
351
8.1 Comparison of Work Measurement Techniques
Existing work measurement methods are concerned with the development
of standard times for production tasks with which to control manufacturing
operations. Since production times are affected by the conditions under
which work is performed, the estimation of standard manufacturing times
must also consider the development and maintenance of standard working 11
conditions.
It is the level of precision in the development and maintenance of
standard working conditions that is the limiting factor to the degree
of estimating accuracy that can be obtained. The high level of management
resources required for precise planning and control of a working environ-
ment is normally uneconomic in low volume, multi-product organisations
such as H. M. Ltd. Hence variability in the working conditions listed
in Tables 18 and 20 result in variability in the times required to carry
out specific tasks.
The variation in the degree of planning of manufacturing methods
carried out within industry has led to the development of a variety of
work measurement systems. These vary in the degree of methods planning
required to estimate time standards.
Of the methods used to estimate time standards, time study
synthesis can be ignored during the present work since this technique
cannot predict manufacturing times and hence is unsuitable for costing
products during the design phase.
352
PMTS systems such as MTM are accurate methods of estimating time
standards, but require detailed methods planning, including the detailed
design of the working area. These systems are therefore relatively
slow and costly, and of little practical use for cost estimating at
H. M. Ltd. For example, using MTM V, which is the fastest of the MTM
family of PMTS systems, the time required to estimate the manufacturing
costs of the New EHB range would be approximately 4 man years. This
level of resource expenditure on the cost estimating function is clearly
unsuitable. MTM standards also form the basis of computerised work
measurement systems (120,129). However, although the application time
is greatly reduced through using a computer detailed methods planning
is still required. Hence these systems would be unsuitable for use by
designers.
, Estimating using synthetic time standards is the method currently
used at H. M. Ltd. This technique uses pre-specified time standards and
is therefore similar in application to the use of PMTS systems. However
the work tasks for which synthetic standard times are measured are
specifically related to the manufacturing activities carried out within
a company. In addition synthetic time standards are normally larger
aggregates of basic work motions than PMT-standards. The level of
methods development required to estimate manufacturing times is
therefore reduced and the estimating application time lowered.
However, the rate at which cost estimates can be prepared using
synthetic time standards is not sufficient to cope with the current
NPD cost data requirements. In addition their use requires extensive
353
knowledge of manufacturing to identify the relevant work tasks involved
in the production of a component. This method cannot thus be used
directly by designers and the responsiveness of the estimating system
is dependent on the priority given by the estimating function to
requests for cost data.
1 -Category estimating, although a work measurement technique
appropriate to low volume manufacturing systems, cannot be used to
estimate the labour costs of product designs since the method is
statistically designed to provide acceptable estimating accuracy over
a large number of components. The time standards for individual
components could therefore be poor. This method requires estimators
to determine the appropriate time band (Figure 10) for a component.
It is reasonable to assume that narrowing the time bands would improve
the accuracy of individual estimates since individual components are
allocated the specific manufacturing time allotted to each time band.
However Fairhurst(125) points out that narrowing the time band increases
the estimating application time and reduces system responsiveness.
Again category estimating requires personnel (foreman, chargehands)
familiar with the manufacturing methods and processes used within a
company and cannot therefore be used by designers.
A suitable method of estimating time standards for a low volume
manufacturing system is the use of work measurement algorithms. These
allow time standards to be developed quickly and accurately by expressing
the causal relationship between manufacturing task times and the factors
354
that affect those times. In addition, work measurement algorithms
could reduce both the cost of developing estimates and the level of
subjective judgement involved, and overcome the need to hold extensive
data files of time standards.
To achieve these benefits the variables used within algorithms e
must be easily accessible to designers. That is, if the system is to
be used by designers inaccessible predictor variables such as machining
feeds, speeds and depths of cut must be avoided.
Four basic methods can be used to develop work measurement
algorithms:
c>>
developed by Mahmoud(138) for estimating the costs of turning
Synthetic Development - Human judgement is used to select
both the predictor variables and the method of combining
variables. The obvious disadvantage with this method is the
possibility of developing an algorithm to fit the data used
in its development rather than a model that expresses the
actual causal effect between manufacturing times and
predictor variables. This disadvantage of the synthetic
development technique is illustrated by the model (Equation 21)
i
components. The model employs a discontinuity factor (Nd)
the actual value of which varies with component volume and
the form of the raw material used (e. g. casting, solid bar
and fabrication). Nd is intended to represent the probable
355
number of passes of the tool over. the machined surface and
therefore the design complexity. However it is obvious
that design complexity is not always related to component
volume or the form of the raw material used. Hence Nd
is not causally related to manufacturing time and a variety
of methods of calculating Nd are required to allow estimated 'Oe turning times to fit the observed data. Therefore poor
estimating accuracy can result as illustrated in Table 25.
(ii) Empirically Derived Equations - are developed under laboratory
conditions hence the predictor variables examined can be
individually controlled allowing the effect of each variable
to be assessed. Empirical techniques are normally used to
develop machining time algorithms such as Equations 22 to 26,
which could provide rapid estimating facilities. Usually
however the many parameters (Figure 12) used to estimate
machining times cannot be causally related, hence it is
necessary to construct standard parameter tables of the form
shown in Table 27. The use of such tables reduces the speed
with which estimates can be prepared and prevents designers
using the algorithms since familiarization with machining
methods is needed. In addition empirically derived models
are useful only under the conditions from which they were
constructed. Hence the variability in working conditions
associated with low volume manufacturing systems could lead
to variances between the estimated and actual manufacturing
times for components.
356
(iii) Operational Research Techniques - are capable of developing
work measurement algorithms:
(a) Simulation. - This method is basically a synthetic
development procedure although the use of a computer
allows the examination of a large number of potential
models. However since models are synthetically
developed it is possible to obtain a model contain-
ing predictor variables that are not causally related
to the independent variable being estimated.
(b) Queueing Theory -This method has been used to estimate
the waiting time involved in fabricating crane sub-
assemblies(122) and is useful when interference
occurs between manufacturing operations.
(c) Linear Programming - This method can be used to develop
linear models that optimize specific objectives. The
method develops equations similar to those developed
by multiple linear regression analysis. However,
in terms of computing cost because LP is an optimization
tool it is an expensive alternative. This factor is
important during the present work since a large number
of equations needed to be developed.
(iv) Regression Analysis - is a statistical method of determining
relationships between dependent and predictor variables.
The method employs partial differentiation to determine
the coefficients of predictor variables (Equation 28)
357
that minimizes the sum of the squares of the differences
between estimated and observed values of the dependent
variable. The technique therefore constructs a model
that expresses the relationship between a dependent
variable and any number of predictor variables. Hence
a large number of individual time standards can be
transformed into simple formulae.
Using regression analysis it is possible to use data
from an existing system without the need to isolate the
effects of each predictor variable involved under laboratory
conditions.
" Although the regression technique cannot distinguish between
an actual causal effect between variables and one that occurs
by chance, investigators (158,159,162,160) have shown that
if predictor variables are chosen with care (i. e. which
naturally requires the use of personnel familiar with the
manufacturing tasks being examined) effective estimating
models can be developed. In addition experienced
personnel are not required when using estimating models.
They are however essential in maintaining a model's
applicability should manufacturing conditions alter.
An examination of previous research(153,156,157) highlighted
several important factors to consider when developing regression
based estimating models:
358
(a) Although the estimating accuracy of models can be
improved by reducing the data variability range, as
illustrated in Figure 15, additional models are
/
necessary to enable the total data range to be
covered. Hence the estimating system becomes more
complex and less responsive. Accuracy must therefore
be balanced with estimating application time and cost.
(b) It is necessary to define the manufacturing tasks and
data range appropriate to an estimating model since
regression equations are applicable only within the
data range used in their development.
(c) The complexity of equations can be reduced by removing
predictor variables that are inter-correlated with
other predictor variables in a model.
8.2-' Cost Estimating System Development
The development of a cost estimating system that employs regression
models requires the definition of the manufacturing tasks (termed cost
centres) to which the estimating models apply since this allows estimators
to identify the relevant equations and the conditions under which these
equations are valid. In addition, a classification system must allow
estimating equations to be accessed rapidly particularly when large
numbers of equations are used within a system.
359
The number of equations required for a system is dependent on
the range of manufacturing tasks that need to be costed and must often
be increased in order to improve the estimating accuracy of individual
equations, i. e. decreasing the number of factors considered variables
within an equation (whilst improving estimating accuracy) 'naturally
requires additional models to be developed. The objective during
the present work is to balance estimating accuracy with system size
and application costs.
The development of an estimating system must also consider the
types of predictor variables used and ensure they are accessible to
. designers, hence maximizing system responsiveness. In this respect
factors such as machining speeds and feeds, depths of cut and number
of cuts must be excluded from the estimating equations.
In addition, a system must enable cost data to be estimated on a
level that allows designers to carry out detailed comparisons of
alternative design solutions, hence cost data is required for individual
design features such as splines and gear teeth.
An examination of existing component coding and classification
systems such: as VUOSO(163), Opitz(164) and Brisch(165) indicated these
systems were not applicable to the present work since they are designed
for purposes other than design costing. It is pointed out by Gombinski (165)
that their adaption for costing purposes may not be possible. Blore(91)
supports this conclusion since he abandoned the use of existing coding
. systems during the development of a computerised estimating and planning
system.
360
Data inaccessible to the designer is associated with the manufacturing
input to the estimating task and variability of this type of data was
found to be caused by:
(i) The process being carried out, e. g. machining, welding
and assembly.
(ii) The operation being performed, e. g. turning, boring and
drilling.
(iii) The size and power of the machine used.
The manual work content involved.
(v) The tolerances and surface finish required.
(vi) The material being processed.
The removal of manufacturing variables initially required the
identification of processes and operations (Table 28) carried out at
H. M. Ltd. with the objective of developing estimating models for each
process/operation combination.
Since the variability of machine process criteria arises due to
the size and power of the machines used it was necessary to develop
estimating equations for each basic type of machine used at H. M. Ltd.
In the case of General Machining operations a component could be processed
on a wide range of machine types. Hence it was also necessary to develop
a method (Table 47, Appendix II) of enabling designers to select the
appropriate machine for components.
361
Allowing the manual work content involved in machining operations
to be calculated separately from the machine content could both improve
estimating accuracy and allow the cost estimating system to be useful
in areas other than design costing. For example, cost comparisons
could be made between manual and automated machining systems since
the manual work content saved by automating a. process could be e
specifically determined.
The method of estimating manual task times adopted by Halevi and
Stout (135) involved estimating manual task times as a direct proportion
of the machine process time. However this method was considered
unsuitable since manual task times are dependent on factors other
than the machine process time, e. g. tool manipulation times depend
on the distance a tool is moved between machining operations.
,: To reduce the number of equations involved in estimating manual
task times and in addition, overcome the need for detailed methods
planning being required, individual manual tasks were grouped according
to Blore(91), i. e. load/unload machine, machine manipulation and machine
set-UP tasks. During the present work an additional manual task group
(gauging and/or inspection) recognised by Knott (92) is considered an
integral part of machine manipulation tasks since the two types are
often difficult to separate. For example during horizontal boring,
manipulating a tool to the required position and measuring tool
location are often carried out simultaneously.
362
Although the type of dimensional tolerances required from a
machining operation affects machine process variables such as feeds,
speeds and depths of cut, this variation is normally a step function.
Hence estimating equations had only to be developed for two tolerance
bands, i. e. <+0.01mm and ! ýO. Olmm). For dimensional tolerances=<±0.05mm
an additional grinding operation was necessary for which a grinding
allowance (+0.1) had to be left on the component.
Since machining times are influenced by the ease with which
materials can be machined it was necessary to include a "machinability"
factor in the system. An examination of the common criteria used to
define or measure machinability identified the index of "metal removal
rate per horsepower" as a suitable measure. This index, measures the
effect of a material's relative machinability on machining time since
it measures the relative rate at which metal can be removed when the
available machine horsepower is constant. Materials were therefore
classified (Table 30) according to their machinability using data
obtained from the P-E Consulting Group (14 4) and Material Factor
equations included in the estimating system.
The collection of data from which to develop estimating equations
involved two basic tasks:
(i) Choosing the predictor variables, i. e. these by necessity
had to be primarily design criteria to allow designers
to use the system.
363
(ii) Collection of the predictor and corresponding dependent
variable data from a suitable source.
With respect to choosing predictor variables it was important to
bear in mind that the regression procedure could not distinguish
between an actual causal relationship between variables and a Of
relationship that occurs by chance. - Hence the need to choose
predictor variables with care required the examination of relevant
literature and the use of personnel familiar with the manufacturing
tasks under examination.
In addition various types and combinations of predictor variables
were examined in order to obtain effective estimating models for each
dependent variable. When small differences (E-<0.05) existed in the
accuracy levels of alternative equations consideration was given to the
standardization of predictor variables between equations in order to
simplify the overall estimating system.
Several data sources (MTM V standards, synthetic time standards
and a planned time study programme) were examined and considered
unsuitable when compared with the use of an existing extensive library
of time study data held at H. M. Ltd. With respect to sawing, spline
cutting'and shearing operations, time study data was not available.
For these operations it was considered that the standard production
times (STT's) issued to the shop floor could represent a suitable data
source, particularly since the times associated with these tasks are
relatively low.
364
An existing procedure (Table 31) for developing flame cutting
times' for steel plate was adapted without loss of estimating accuracy.
Since no loss of estimating accuracy over the existing method was
incurred, the equations did not require validation.
The times taken to machine features such as chamfers, circlip e
grooves and radii exhibited little variability within a machine type.
The arithmetic mean of each operation was therefore used since
adequate accuracy levels could be obtained(155).
Since STT values for welding tasks were inaccurate the approach
adopted for developing equations for estimating welding times involved
(i)developing equations to estimate "basic arc times" using standard
data issued by the Welding Institute(171)9 and (ii) developing equations
to convert "basic arc time" to total welding time using experienced
personnel to estimate the affects of variables on the welding "operating
factor" (Equation 31).
Since a large number of estimating equations needed to be developed
a computer was used to perform the actual regression analysis. Hence
much tedious effort was avoided and therefore possible errors eliminated.
The regression package(176) available at the Loughborough University
Computer Centre was used during the present work since it provided a
wide range of supporting statistical data (Table 33) with which to
determine the quality of the estimating equations developed. In
addition, the regression package could recursively construct prediction
365
equations using stepwise regression procedures. This enabled the
maximum estimating accuracy to be obtained using the minimum number
of predictor variables.
The package also has the facility to test predictor variables
for inter-correlation and when necessary remove correlated variables
from the final estimating model.
s The data handling facilities of the SPSS system (of which the
regression procedure is part) enabled data to be held in sub-files
such that equations could be developed for individual or combinations
of sub-files. This facility enabled estimating equations to be developed
in a single regression run for individual or combinations of machine
sizes. 'Hence the number of equations within the system could be
reduced when an equation for a group of machines gave acceptable
accuracy results.
Since the regression technique has been used when the exact
relationship between dependent and predictor variables is unknown it
is essential that the accuracy of the estimating models proved to be
within the current accuracy limits required by H. M. Ltd.
. The commonplace methods (R, R2, F-test, t-test and the Coefficient
of Variability) of indicating formula accuracy have been shown to be
inappropriate to the measurement of a model's estimating accuracy,
since they are not a direct measure of the difference between actual
366
and estimated values (i. e. residuals). The present work supports this
opinion since Figures 16 to 18 show that R, F-value and the Coefficient
of Variability are unrelated to Average Proportional Error (Equation 36)
which is determined directly from residual values.
Cochran(155) proposed the use of the quantity 1-e where e is
determined using Equation 33 but states a limitation when narrow ranges
of dependent variable data is involved. In these circumstances the
quantity 1-e is a poor measure of estimating accuracy since it fails
to consider that a narrow base of experience could be an unreliable
foundation for projecting future results. However a basic limitation
when'using regression models is the fact that models are only applicable
over the data range from which they were derived. Hence Cochran's
limitation to the use of 1-e is not valid.
The present objection to the use of l-e and a further measure of
an equations estimating accuracy proposed by Cochran the Association
Index (Equation 34) is the inclusion of the term y since this results
in the indices not being "completely" directly related to residual values.
A suitable index of the estimating accuracy of models provided
by Ezekiel and Fox (174) is the Standard Error Ratio (Equation 35). It
requires only the size of residuals for its determination and therefore
indicates the closeness with which estimated values may be expected
to approximate to the true but unknown actual values. It is therefore
a measure of the range of the accuracy of individual estimates.
367
A measure of the estimating accuracy of a model needs to consider
the proportional errors yi-xi since the accuracy of cost estimates yi
is presently measured in this manner. By using the index E (Equation 36)
an improved measure of the estimating accuracy of models is obtained
since this value is a measure of the average proportional error that
arises through using an estimating equation.
Unfortunately the conventional regression procedure of minimizing
the sum of the squares of the residuals does not necessarily minimize
the value of E, as illustrated in Table. 54, so that the most
effective estimating equation is not always developed. However the
conventional regression procedure is suitable for use in the present
work since:
. (i)
/
The low volume manufacturing environment at H. M. Ltd.
prevents proportional errors being magnified into large
cost variances.
Analysis of product manufacturing costs indicated that
a Pareto distribution exists (See Section 5.2). A small
number of high cost components accounted for the majority
of a product's total manufacturing costs. In this way the
use of the conventional regression procedure is suitable
because the technique effectively improves the estimate
accuracy of high cost components.
368
For high volume manufacturing systems in which the reduction in
average proportional error is important, the conventional regression
procedure can be modified (Appendix VI) to minimize proportional
estimating errors. Appendix VI also illustrates how the relative
manufacturing volumes of components can be allowed to influence
estimating accuracy. /
The value E indicates the average error proportion arising from
the use of an estimating equation and does not therefore indicate the
range of individual errors that could arise. The range of individual
estimating errors that could arise from an equation is indicated by S
which is the Standard Deviation of proportional errors (Equation 37).
The values of E and S (and R2, F-value, Student t-test, 1-e and AI)
are normally descriptive measures of a model's estimating accuracy and
therefore can only be used to provide confidence limits(213) when the
manufacturing times examined are normally distributed. However labour
times usually form a negative exponential distribution (122,123). The
calculation of confidence intervals cannot therefore, be achieved using
a simple statistical package such as SPSS. They could however be obtained
using a more sophisticated package such as GLIM(214) and specifying a
skew error distribution such as Log Normal. However this approach is
not practicable since the proposed estimating method is intended for
routine use in industry.
Since the values of E and S were obtained using new data (i. e. time
study data not used to construct the estimating model) it is acceptable
to use these values as norms by which to judge the estimating accuracy
of equations.
369
Examination of the equations that failed to meet the accuracy
'acceptance criteria (i. e. the acceptance levels were set at H. M. Ltd. 's
current estimating accuracy limits of E=0.096 and S=0.314) indicated
that the prime causes of poor estimating were:
(i) Variation in machine process data arising from lack of
control over work methods used.
Variation in operator experience.
(iii) Variation in the quality and dimensional accuracy of the
raw materials used.
However these conditions are characteristic of low volume manu-
facturing systems since economics require a low level of management
planning and control to be exercised over working conditions. Hence
although equations appear to have poor estimating accuracy levels they
would minimize the affects of variations in working conditions on the
inherent variability of task times.
Additional causes of poor estimating accuracy (i. e. incorrect type
and combination of predictor variables and the use of negative predictor
variable coefficients) were corrected and estimating errors therefore
reduced. In the latter case the inclusion of negative coefficients
within an equation resulted in negative labour costs being generated
for low values of predictor variables. To overcome this problem
operation times (Table 35) were allocated to the manufacturing tasks in
question below which estimated operation times could not fall. Hence
the estimating accuracy at the low end of the data range of a dependent
variable could be improved whilst not influencing the accuracy levels
outside this area.
370
The frequency distributions for the E and S values of the equations developed (Figures 21 and 22) indicate that the majority of the equations
are acceptable. However since more than one equation is usually required
to estimate the total labour costs of a component, as illustrated in
Appendix II, it was necessary to determine whether the individual
estimating errors from each equation were "additive" and would
therefore create large estimating errors. Hence the estimating
equations were used to determine the manufacturing times for existing
H. M. Ltd. components and these total times compared with the corresponding
Sales Target Times (STT's). The comparisons (Figures 29 to 46) indicated
that good estimating accuracy could be achieved since a large proportion
of the points lie on or near the 450 lines which indicate perfect accuracy.
If equations are used to develop manufacturing time standards and the
manufacturing function consider* these standards over estimated, then
the equation coefficients can be altered to compensate for this. In
this manner a direct link exists between design criteria and the manu-
facturing time standards used to control working conditions.
Initial attempts to develop equations for estimating assembly
task times involved the use of a relatively large number of predictor
variables, i. e. 11. The resultant model (Equation 38) indicated that
the probability of the regression procedure finding non-existent
relationships appears to increase with increasing number of predictor
variables. Hence it was beneficial to classify manufacturing operations
into-groups which reduced the number of variables influencing manufacturing
tasks.
371
A reduction in estimating time of approximately 70% was observed
compared with H. M. Ltd. 's current estimating methods, by using the
regression based estimating system. The response of the estimating
system can be expected to be much greater since little "queuing"
time is involved (i. e. cost estimating for the design fynction is not
reliant on the work loads of other departments) when designers generate le their own cost data. In addition the lack of a subjective element in
the estimation of design costs reduces the risk of "user bias" entering
into the process.
Although the estimating system can be used manually (Appendix II)
software has been developed (177) to allow a computer to carry out the
tedious calculations involved. This improves the rate at which estimates
can be developed and reduces the'occurrence of computational errors.
Since the system is composed of a large number of equations, a
method of retrieving the appropriate equations from the data file was
required. The obvious method was to group equations according to the
appropriate machine type since this corresponded with the existing cost
centre classification at H. M. Ltd. Manufacturing times were easily
converted to costs by application of appropriate overhead and labour
cost rates for the particular cost centres.
Equations within a cost centre were then allocated an identification
number which, in conjunction with the cost centre classification, allows
each equation to be individually retrieved. With cylindrical and surface
grinding (i. e. cost centre 216) and the NC bar and chuck lathes (i. e. cost
centre 258) processes had identical cost centre classifications.
372
Hence instructions were included in the system operating procedure
(Appendix III) to enable users to retrieve the appropriate group of
equations.
Since the computer program had to be interactive (i. e: predictor
variable data needs to be input) it was necessary to determine the tasks i
carried out by the users and those by the computer. In this respect it
was necessary to make effective use of the computer since the computer
facilities of the design department were obtained on a "time sharing"
basis. Hence it was necessary to allow users to identify the relevant
predictor variables involved in order that the appropriate input data
could be obtained from a component's engineering drawing prior to using
the'computer program. Using Appendix IV, therefore, to identify
predictor variables avoids delay arising whilst using the program.
An advantage of using design criteria for estimating manufacturing
costs is the possibility of cost data being freely available to all
departments within an organisation.
To fully computerise the estimating system would require the direct
transfer of predictor variable data from component drawings to the cost
estimating module. In this respect the ability in CAD systems to retrieve
dimensional data from stored component drawings could enable the estimating
procedure to be partially computerised. To complete the computerisation
of the estimating system would require additional software developing
to enable the relevant estimating equations to be identified and
organize predictor variable data in the correct form and order for
these equations.
373
8.3 The Make or Buy Decision
The decision to manufacture "in-house" or to purchase-from an
outside supplier (the make or buy decision) is fundamental to the
development of a product since all components need to be examined in
this manner.
A model (Equation 7) has been developed during the present work
that allows designers to cost designs and therefore to compare the costs
of alternative design solutions. Since the equation includes both the
make and buy elements involved in a component's manufacture (i. e.
component designs often include an element of both make and buy),
practical decisions involving the economical amount of "in-house" manu-
facturing can be undertaken. In addition, a tooling cost element is
included since tool costs (which include capital equipment costs) have
a significant influence on the economics of alternative design
solutions.
The model has been adapted from the classical stock control models
used for determining the variable costs involved in stock holding and
manufacturing of alternative batch sizes. The affect of adopting
economic batch sizes (under various manufacturing situations such
as, when shortage costs occur or, when simultaneous manufacture and
usage is applicable)on the economics of alternative design solutions
can be examined. In this respect, care must be taken when determining
economic batch sizes since purchase and manufacturing batch sizes can
differ.
374
The use of Equation 7 to compare the New EHB rope guide assembly
costs highlighted the influence of the payback period for tooling costs
on the economics of alternative design solutions (Table 11). As can be
expected varying the tool cost recovery period influences the relative
economics of alternative rope guide designs. The choice of which
design solution to adopt could then become a marketing decision.
Using the alternative rope guide assemblies as examples, if the cheapest
design over the nominal life of the tools is chosen (Alternative III,
Table 11) but tooling costs recovered over the company's normal payback
period, for H. M. Ltd. this period is 3 years, then the initial cost
of a product could be higher than if the cheapest design solution over
the short term were chosen (Alternative V, Table 11). The marketing
department therefore need to decide if the higher costs are acceptable
or if product sales (at a period when the product is introduced to
the market and rapid sales growth is required) will be reduced.
If an increase in price is not advisable an alternative solution
would be to accept a reduced profit margin during the investment cost
recovery period. In this manner. production technology requiring high
capital investment costs can be more easily introduced into a company.
In addition the purchasing policy of a company towards production
equipment with a long introduction period until its maximum utilization
was achieved could also benefit from this type of investment policy.
This type of equipment would include industrial robots where a previous
study by the author (122) indicated that a considerable amount of redesign
of existing fabrications was necessary before the maximum benefits could
be achieved from a continuous arc welding robot.
375
Equation 7 can also be used to determine the effect of alternative
production control systems on the economics of design solutions. For
example, the basis of Period Batch Control systems (62) is the
determination of a manufacturing "cycle" time for a product which can
be a'function of both the time required to manufacture the component
with the longest production time and the forecast sales demand. The
purchase and manufacturing batch sizes (which are equal under PBC systems)
can be calculated from these two factors and as illustrated by
the analysis of the alternative rope guide assemblies can be reduced
in size. Reducing batch sizes increases the labour and purchasing costs
per component whilst tooling and investment costs per component remain
constant, i. e. these cost items are not dependent on batch sizes but
on annual quantities manufactured. The outcome of this change in costs
would be, therefore, an improvement in the chances of adopting design
solutions that require tooling and investment costs.
Accurate costing of a product design could not take place until
manufacturing and purchasing batch sizes are known. Under a PBC system,
therefore, product designs could not be costed until sufficient detail
design and manufacturing planning had taken place to enable cost
estimators to identify the component with the longest individual
throughput time. This would present difficulties in costing general
arrangement drawings. In addition the sequence of detail design
required to identify the component with the longest individual through-
put time may not be compatible with the overall development plan.
376
-Normally when costing product designs, the effects of varying
order quantity and manufacturing batch sizes on manufacturing costs are
not considered. Using Equation 7, however, requires designers to use
these quantities to determine total manufacturing costs. Hence
alternative design solutions could be compared at their respective
minimum costs which occur at the optimum values of order quantity
and manufacturing batch size.
8.4. Cost Forecasting
The regression based estimating system developed in the present
work-allows designers to estimate component labour costs from general
arrangement drawings. However during the concept design stage of the
NPD process (particularly if several alternative design concepts are
being compared) the need to develop general arrangement drawings in
order to determine manufacturing costs is undesirable. The ability to
determine product manufacturing costs without the need for such design
drawings would:
(, ) enable unattractive design alternatives to be abandoned
early in the design process and therefore allow design
resources to be used effectively,
(ii) increase the number of design alternatives that could be
analysed,
(iii) allow fundamental design decisions (e. g. the "assortment"
problem) to be taken, and
-(iv) allow a greater number of competitive product designs
to be costed.
377
An obvious method of cost "forecasting" when developing a new
product range is to identify the cost/design specification relationships
that exist, hence knowing the design specification (e. g. capacity, size)
the appropriate cost could be found. In the present work since regression
analysis has been shown to be effective in determining causal relationships
the technique has been used to construct cost/design specification models. i
To ensure model estimating accuracy the total range of product
costs must be used to construct cost/design specification models. This
means that the smallest and largest units in a product range must be
costed. In this way the ability to develop effective costing models
depends on the order in which product sizes are developed. Development
schedules however are not always suitable for costing purposes. For
example, during the New EHB project the 0.8 tonne, 1.6 tonne and 3.2 tonne
units were developed in preference to the 5.0 tonne and 8.0 tonne units
to enable early launch of a limited product range for which the greatest
market share was forecast.
Although during the present work cost relationships have been useful
in the optimization of designs (e. g. drum tube assembly) their basic
disadvantage when used to estimate total product design costs is the
need to carry out detailed design and costing of those product sizes
used to develop the cost models. This need limits the number of products
that can be costed and hence the estimating accuracy of models. With
respect to the minimization of individual product design costs it is
an advantage for a designer to initially identify the existence of cost
relationships between the various assemblies of a product. Representing
378
these cost relationships in a table of the form shown in Figure 48
provides a useful design aid. For example when the drum tube assembly
of the New EHB was being designed, Figure 48 provided a quick reference
to the assembly cost relationships that needed to be determined in
order to enable total product manufacturing costs to be minimized.
As might be expected examination of a wide range of products
manufactured at H. H. Ltd. indicated that when cumulative costs were
compared with the cumulative number of components a Pareto distribution
is formed. Although the form of the curves varied between products it
could be expected that for each product the curve is log-linear.
Converting both cumulative cost and number of components to their
logarithmic form (Equation 45) would produce a straight line curve.
If a small number of the highest cost components (which are easily
identifiable since they form the basis of the conceptual design of a
product) are then used to extrapolate the whole straight line curve,
the constants f and g (Equation 43) required to determine total product
costs could be found so that only a small number of components would
need to be costed in detail. The costing of products also requires a
knowledge of the total number of components within a product although
the accuracy with which this figure is determined need not be high since
the relative contribution to the cumulative product costs decreases
rapidly with increasing number of components.
However, when models of the form shown in Equation 43 were used
to cost H. M. Ltd. 's products these models grossly over estimated the
actual product costs. The over costing effect can be explained with
379
. reference to the work carried out by Swan and Worrall(183). Log-log
curves established by these researchers indicated that the slope of the
curve decreases at a distance along the Nu-axis. This explanation was
confirmed by constructing log-log cost curves (Figure 52) for several
of H. M. Ltd. 's products.
0
Increasing the number of components used to determine the values
of f and g (up to 50% of the total number'of components was examined)
did not improve the estimating accuracy of cost models to an acceptable
level, i. e. to the accuracy of H. M. Ltd. 's existing estimating system.
Using proportions greater than 50% was not considered by the author to
be an effective method of improving estimating accuracy since this
procedure would not significantly reduce the amount of detailed cost
estimating required to determine total product costs.
From visual observation of the relative values of the variables
that form Equation 43 it was reasonable to assume that they were related
to the point at which the slope changes. Hence regression analysis was
used to test this theory since conversion of the variable values to
their logarithmic form ensured that they were, if at all, linearly related.
Although significantly improving the accuracy of the estimates
developed the initial models developed did not provide sufficiently
accurate estimates. Poor estimating accuracy was found to be caused by:
(i) Not using a common component structure level, i. e. using
both component and assembly costs.
380
(ii) Using items classed as consumables (e. g. washers) in the
determination of the total number of components within
a product.
(iii) Not using the actual value of f to determine product costs.
Upon, correcting the input data the new model developed, Equation 47,
was found to give an acceptable estimating accuracy, i. e. E= -0.015 and
S=0.096.
8.5 Design Scheduling
When the time available to develop a new product is restricted the
scheduling of development tasks in order to minimize overall development
time is important. In addition the efficient allocation of design
resources needs to be considered, i. e. resources need to be allocated
in relation to the relative importance of tasks to the overall success
of a product design.
Existing network analysis project planning techniques have been
found to be unsuitable for controlling the development of a new product
range since:
(i) They are expensive to implement (PERT implementation
cost is £17,000 for a project of similar size to that
of the New EHB development) and are an inferior choice
for the expenditure of development funds when compared
with the equal cost opportunity of increasing the design
staff.
381
(ii) To construct a network diagram requires a series of
identifiable tasks. Due to the creative nature of the
new product design process the remaining tasks required
to develop a component cannot be accurately identified
until the design process is complete.
le In addition existing project planning techniques (PERT, CPM) are
concerned with determining the time required to perform a development
task whereas when limited design time is available a planning technique
should concentrate on effective allocation of the available resources.
Since component designs are initially created at the design stage
of the NPD process it is here that the scheduling of development work
should begin. That is, components need to be sequenced through the
design stage such that overall development time is minimized. The
major constraints influencing the sequence with which components are
scheduled through the design process are:
(i) The order in which products are developed which can
be dependent on marketing or financial considerations.
For example, H. M. Ltd. need to market a "mini-product
range" to enable revenue to be obtained quickly.
(ii) The length of the prototype testing period which is often
considerably longer than the sum of the remaining project
task times particularly if long term reliability testing
needs to be undertaken.
382
Although these constraints determined the order in which sub-
assemblies are to be developed it is necessary to sequence the detail
design of each component within each sub-assembly. In this'respect
scheduling components through the design phase is considered in the
present research analogous to the scheduling of operations within a
batch manufacturing system. For the design process the main steps are: /
(; ) Specifying the objective criterion by which alternative
schedules can be judged. Although the conflicting aims of
the various departmental functions involved in the new
product development process could make the specification
of a single objective criterion difficult, the importance
of minimizing development time during the present project
and its increasing importance to the ultimate success of
a new product (29)
makes this the obvious choice.
(ii) Sequencing the design of components. In this respect it
is necessary to determine the time required to carry out
the individual development tasks for each component and
the interactions that exist between development tasks.
Since a*large number of components are involved in the New EHB
project it is essential that designers are able themselves to estimate
development task times to avoid delay in the acquisition of data from
other functional departments. In order to achieve this situation the
present research related development activities and activity times to
design criteria and developed a check list of activity times (Table 42).
333
Using the two types of task interactions that arise during product
development (precedence and simultaneous) the interactions between tasks
required to develop individual components were identified (Section 6.2.3).
This table used in conjunction with the NPD network diagram (Figure 56)
for a component allows a designer to determine the interactions between
tasks and hence calculate the overall development time for each
component within a sub-assembly.
The validity of the method established'for estimating component
development times when tested against actual components from the New EHB
project (Figure 59) indicated that close agreement existed between
actual and estimated development times.
The cumulative distribution of development times estimated for
each standard unit were found to form a typical Pareto curve (Figure 60)
indicating the possibility of reducing the level of management control
resources by concentrating on the small proportion of components with
long development times. In addition, maintaining a list of components
in order of the "length of development time required" and constantly
updating this list (as design proceeds) enables designers to quickly
isolate those components that could affect the overall development
time. The construction of such a list could remove the need to develop
and update a network diagram.
When limited design resources are available the conventional design
methodology of creating and comparing alternative design solutions cannot
always be undertaken, particularly if development time is limited. Under
384
these circumstances an appropriate design methodology to adopt would be
to create an initial solution and allocate design resources for the
improvement of this solution. Allocation of resources could be based
on the relative contribution to be gained from the overall success of
the product design. If the initial solution cannot be improved
sufficiently by the designer then it must be the value engineering
teams responsibility to create alternative design solutions.
Pleasuring the relative importance of alternative design solutions
requires a common basis of measurement. The difficulties with introducing
PPBS project planning techniques illustrates the problems involved in
finding a common basis of measuring the benefits obtained from the
use of additional design resources. This difficulty also applies to
the NPD process since although the overall benefit to be gained from
additional resources is an increase in profitability there are a variety
of routes by which this can. be achieved. The effect on profitability
of each route is not always easily quantified.
The effect of manufacturing cost changes can be related to a
product's profitability. This factor could therefore be used as a
common basis of measurement. Design resources would therefore be
allocated on the relative cost of a component or assembly. In this
respect the relative cost of components would-be used to allocate
resources on the reasonable assumption that the greater the initial
cost the greater the potential for cost reduction. Manufacturing cost
is also convenient for the allocation of design resources since the
regression based estimating system developed in the present work allows
385
designers to generate cost data rapidly and therefore enables design
resources to be allocated as the design process proceeds. In addition
as new designs are created cost comparisons can be made with existing
components and re-allocation of design resources carried out when
necessary.
oof To allocate design resources therefore a simple method would involve
the construction of a Pareto distribution curve for component costs and
convert the cumulative cost axis to cumulative design improvement time,
hence allowing design time to be obtained as a direct measure of
component costs.
Since cost is but one of several important product characteristics
the allocation of design improvement time based on relative cost needs to
be weighed against the need to improve other characteristics which
increase the overall worth of a product.
8.6 CAD
Through the installation of the Autotrol system(194) H. M. Ltd. will
have the use of a sophisticated Computer-Aided-Draughting system for
designing custom-specified products assembled from existing standard
parts, and extension into the Computer-Aided Planning and Computer-Aided
Engineering areas (e. g. NC part programming and stock control).
The initial introduction of DCL's EUCLID system(195) can be
accomplished at low cost through a time sharing facility which overcomes
the need to purchase expensive hardware such as a digital plotter. Hence
386
both the experience gained and the initial preparation of data files
using this system will enable the payback period for the more expensive
Autotrol system to be reduced.
Since Autotrol is essentially an automated draughting system,
the system in its basic form has numerous limitations when used to
aid the design phase of the NPD process. In order to assess the
relative significance of these limitations an examination of the
tasks involved in the original design of the New EHB was undertaken.
This examination identified design tasks and determined the
proportion of time a designer devoted to the accomplishment of each task
during the preparation of a product design. It must be recognised that
the results of the examination (Table 45) particularly the proportion
of time devoted to each task applies only to the design of the New EHB
product and would vary between product types. For example, an innovative
product design could require a higher proportion of creative design
effort when compared with the design of the New EHB range since
H. M. Ltd. 's design is based on those of competitive products.
In determining the applicability of the Autotrol system in aiding
the development of the New EHB the contribution of CAD to each activity
involved in the design process has been examined.
387
(i) Function Analysis
In their basic form CAD systems are merely automated draughting
and data storage systems and therefore require additional
software modules in order to carry out the functional
analysis of product designs. Because of the diverse
nature of product types within the mechanical engineering
industry it is possible that a wide range of functional
analysis software packages for such applications as gearbox
design and stress analysis would be applicable to the
design of wire-rope hoist blocks, and would be available
"off-the-shelf".
In those cases where "off-the-shelf" software is not avail-
able the economics of developing software depends on the
complexity of the analysis required and the frequency of
its use. Since new product ranges are developed at H. M. Ltd.
at infrequent intervals it is unlikely that functional
analysis software would be an economic alternative to the
present manual methods that involve using desk top
calculators and scaled models. However use of software
that could be applied to heavy, custom-designed cranes
(which represent the major portion of the design work carried
out at H. M. Ltd. ) could in economic terms be more easily
justified. It would be of benefit, therefore, to examine
the functional analysis requirements for the complete range
of products manufactured at H. M. Ltd. in order to identify
software packages that would be economically viable.
388
(ii) Data Acquisition and Transfer
The ability of CAD systems to hold large quantities of
design data (including a complete component definition)
enables the acquisition and transfer of data by the design
department to be greatly improved. The advantages of a CAD
system would include:
(a) An increase in the amount of data stored on an
engineering drawing.
(b) Instant data transfer facilities.
(c) The possibility of preparing NC part programmes
directly from the CAD system.
(d) The ability to produce drawings that hold information
for individual manufacturing operations.
Since the design process relies on the acquisition of data
from other functional departments involved in the NPD process,
CAD systems can be expected to be useful in the development
of optimum designs. However although CAD systems can store
large quantities of data it is the generation of this data
by the various functional departments that could be expected
to limit the usefulness of CAD systems to the design process.
For example, the present research has shown that the generation
of cost data using H. M. Ltd. 's present estimating system is
389
slow and cannot provide the amounts of cost data required
by the design process. Hence the development of the
estimating system in the present work could be considered
helpful to the economics of CAD systems when used to aid
the NPD design process, in that the time delay involved
in data acquisition could be significantly reduced. e Linking the estimating system to a CAD system such that
operator interaction was not required would then completely
eliminate the time delay between data acquisition and use.
(iii) Design Creation and Improvement
There are reasons to suppose (205) that CAD systems could
have an adverse effect on the creation of design solutions
because of the changes that would take place in the
conditions under which designers work. These changes,
such as the introduction of shift work and measured work
rates, could arise from the need of a company to quickly
recover the high investment cost of purchasing CAD systems.
The present author considers it doubtful that effective
original design would take place at a CAD terminal since
the creation of designs is an unpredictable process and
incompatible with the types of tasks such as, draughting
normally carried out at visual display units. In addition
it has been shown that a relaxed atmosphere aids the creation
of design solutions which is an environment normally not
possible at man/machine interfaces.
390
Conventional techniques for creating alternative design
solutions are also incompatible with the use of CAD.
Brainstorming(206) for example, requires the creation of
as many design solutions as possible prior to-their analysis.
To carry out the technique effectively it is necessary to
quickly sketch design ideas then concentrate on generating
'Of more ideas, i. e. to create a flow of design ideas. Hence
the need to record using a keyboard(which requires translat-
ing design ideas to the input language of a CAD system)
would constantly interrupt the creative flow.
Although having no direct input to the creation of design
solutions the use of CAD could be expected to increase the
proportion of time a designer allocates to the creative
tasks. That is, the increase in speed with which designs
could be analysed could allow more design solutions to be
considered.
The use of CAD systems"to actually create designs is limited
by the subjective nature of design problems and the ability
to logically specify requirements. In addition the number
of decisions to be made can be sufficiently large that
design solutions could be generated manually in the same
time it takes to develop software to specify the problem
to the computer.
391
(iv) Cost, Marketing and Manufacturing Analysis
At the present moment the major difficulty in carrying
out the cost, marketing and manufacturing analysis of a
product design is the need to obtain relevant data from
the other departments in the NPD process. The departments
involved are usually not designed to generate the large
amounts of data required during product development.
This is a major reason for the development of the
regression based cost estimating system. In addition
the "day-to-day" tasks, such as production scheduling
and purchasing carried out by these departments, often
have priority over requests for data by the design
department.
The contribution of CAD systems to this area of the design
process will arise when systems are developed to allow
designers to quickly generate cost, marketing and manu-
facturing data. The development of the regression based
estimating system is a significant contribution. Linking
this estimating system to the CAD system will be possible
with the CAD software extracting data from the stored
drawing of a component and transferring data to the estimating
module.
The eventual generation of manufacturing data by the designer
centres around the construction of lists of producibility
tips (209)
and data bases (207,208) that are developed for
use by designers. An alternative would be the allocation
392
of a data file layer to a VE team, who could then analyse
the manufacturing elements of a design and propose changes
that could quickly be implemented to the original design.
(v) Draughting
Since CAD systems can include mechanised draughting systems f they could be expected to have a significant effect on the
design process. However, the time a designer spends
committing ideas to paper is not a significant proportion
of the total design activity. For the New EHB project CAD
systems could not be justified solely in terms of their
impact on the draughting activities, since these tasks
represented only 60110 (763 hours) of the total design time
expended. Hence the maximum saving possible through use
of CAD is approximately £8,000.
The feasibility of using CAD to reduce the time required to
prepare engineering drawings is dependent on the design
complexity of the product since this affects the amount of
detailed design work necessary to prepare drawings for the
manufacturing functions. This makes the use of CAD to
design customer specified heavy cranes economically feasible.
(vi) Integration with the overall NPD Process
The integration of design with the remaining NPD tasks is
primarily a problem of data generation and transfer between
departments. The contribution of CAD systems in this area
would be limited if departments are unable to quickly
393
generate cost, manufacturing and marketing data with which
to progress the development of a new product. An effective
contribution to the integration of the overall NPD process
could be made, therefore, if CAD systems were developed
to aid the generation of product design data. The transfer
of data between departments can already be effectively
I-e accomplished using CAD systems.
In addition to aiding the generation of NPD data it is also
considered important by the present author, that CAD systems
must be expanded to enable:
(a) A coherent, cohesive information system to be
developed that allows a series of design optimization
tools (e. g. Linear Programming, Dynamic Programming,
Make or Buy models) to be combined. The basis of
this information system would be the analysis of the
common data requirements and the development of software
to transfer this data to specific decision-making and
optimization tools.
(b) The planning and control (including the allocation of
design resources) of the design process using an
appropriate project planning technique. The basis of
such a system would be the necessity to allow designers
to generate their own planning and control data. Hence
the importance of the scheduling technique developed
in the present work.
394
ý
(c) A system that ensures successful product designs will
be developed. At H. M. Ltd. this is likely to be
achieved by an expansion of the existing Quality
Assurance scheme (215)
since product quality is but
one important design characteristic of a product.
The draft guide to design management procedures for
the assurance of quality issued by the British Standards
Institute (216) lists with respect to quality assurance
the important factors to be considered by the design
department. These procedures could be expanded to
include other essential design characteristics of
a product.
395
CHAPTER 9
CONCLUSIONS
396
1. An estimating system has been developed and is currently being
used at H. M. Ltd. that enables designers to quickly cost product
designs as they are created, using data readily available to the
designer. The system therefore allows estimates (the accuracy of
which are within the current accuracy levels of H. M. Ltd. ) to be
generated during the design process, hence cost data can be used
to greater effect for:
(i) developing minimum cost designs,
(ii) reducing the number of design changes that occur after
the prototype build phase has begun and so reducing the
design resources needed to develop a product,
(iii) allocating limited design resources based on the relative
cost of designs,
(iv) allowing the designer to maximize product worth, and
(v) allowing the greater use of design optimization techniques.
Methods have been proposed for modifying the conventional regression
procedure in order to:
(i) Minimize the sum of the squares of "residual" values when
measured as proportions of their corresponding "observed"
values. This modification (which enables the method of
calculating the accuracy of estimates and the objectives
of the regression procedure to be made compatible) is more
suited to high volume manufacturing systems than to the
small batch environment of H. M. Ltd.
397
(ii) Enable the volumes of individual components to influence
the accuracy of their cost estimates. This procedure
is again suitable for high volume manufacturing systems
where it can help avoid errors in cost estimates being
magnified into large cost variances due to the volume of
components involved. i
2. A model, which allows the total cost of a product to be estimated
from the product's ten highest cost components, has been developed
by modifying the equation for the Pareto distribution. Product
designs can be accurately costed at the concept design stage
without the need to prepare general arrangement drawings.
3.
4.
A check list of development task times has been constructed from
which the time required to develop individual components can be
estimated. The check list when used in conjunction with the data
provided on task interactions allows designers to identify those
components that would lie on the critical path without the need
to construct the whole task network for the NPD project. Planning
schedules can, therefore, be constantly updated as design proceeds.
A model which allows designers to compare the variable manufacturing
costs of alternative design solutions has. been constructed using
classical EOQ theory.
The model can be modified to apply to a variety of manufacturing
situations, e. g. simultaneous purchasing and usage, when shortage
398
costs apply and using the Period Batch Control (PBC) production
control technique. In addition the amount of in-house manu-
facturing that minimizes total manufacturing costs can be
determined for a product design.
5. A method of improving the rate at which the manufacturing costs
of a'product design can be reduced has been proposed. The
technique takes advantage of the "learning curve" effect by
requiring cost reduction methods to be applied individually to
all components within a product design.
6. The tasks required to allow H. M. Ltd. to develop a method of
forecasting the need for new product development have been described.
The development of such a forecasting technique at H. M. Ltd. is
prevented at the present moment by the lack of data on past
development projects.
7. The cost effectiveness of using CAD systems to aid new product
development at H. M. Ltd. has been found to be dependent on:
(i) the amount of draughting involved,
(ii) the availability of software for performing functional
analysis of designs, and
(iii) the innovativeness of the new product design. .
399
In addition it is considered that the use of a CAD system could
adversely effect the creative design process but would benefit
the NPD process by enabling effective data transfer between
the various departments involved. This would improve the level
of integration, planning and co-ordination achieved between
product development tasks and hence improve the chances of
marketing a successful product.
400
CHAPTER 10
RECOMMENDATIONS FOR FURTHER WORK
401
In the short term further work should include:
1. Extension of the range of processes and operations that can be
costed. In this area work is presently being carried out at
Leicester Polytechnic to develop regression based models for
estimating the cost of cast components. i
2. Aggregation of individual estimating equations. The objectives
in this area would be to reduce the number of design variables
required to estimate the manufacturing costs of designs. Hence
the time required to estimate costs would be reduced and an
estimating system could be developed for use by sales personnel
" for such functions as sales quotes and tendering.
3. Expansion of the software developed for the estimating system into
a comprehensive data base for the design process. Research is
necessary to identify the type of manufacturing and product
attribute (reliability, quality and serviceability) data that
needs to be considered during the design process. This data must
then be related to such factors as type of design feature, process
being carried out, operation being performed, type and size of
equipment in use, tolerances and surface finishes required and
type of material being processed. The use of a particular
costing equation from the estimating system developed could
then be used to display a list of relevant design data that
needs to be considered. In particular this concept could be
402
used to provide designers with relevant manufacturing data.
Additionally coding of design features would enable data files
containing software for the analysis and solution of functional
design problems to be referenced.
In the medium term further work would include: le
1. The development of a comprehensive cost and design optimization
system for the design process. This would involve writing
software for the available cost optimization, resource allocation
and design scheduling techniques and the development of a common
data base. The objectives would be to allow designers to
prepare and input relevant data into the system and call-up
appropriate optimization tools. The cost forecasting and design
scheduling techniques developed in the present work would
undoubtedly be included in such a system whilst the estimating
system developed could both generate cost data and transfer
this data to a suitable data file.
2. The development of software to link the cost estimating system
developed to a CAD system. The objective would be to allow the
CAD system to retrieve relevant design variable data from an
engineering drawing and transfer this data directly to the
estimating system. Hence removing the interactive element
presently required to use the estimating system and reducing
the time required to develop cost estimates.
403
In the long term further work would need to consider the integration
of all functions involved in the NPD process. The objectives would be
to remove all current barriers to the development and transference of
data between departmental functions. A computer-aided engineering
system would play an important role in this area by allowing large
amounts of information to be stored and quickly retrieved by all
departments. In addition, a scheme would need to be developed for
co-ordinating the activities of departmental functions with the
objective of ensuring that the success of a new product design is
considered at all stages during the NPD process. Such a scheme would
undoubtedly be an extension of existing Quality Assurance methodology.
404
REFERENCES
1. WAINWRIGHT, D. Cranes and Craftsmen: The Story of Herbert
Morris Ltd., 1st Ed., Hutchinson Benham Ltd., 1974.
2. NEWMAN, W. H. and LOGAN, J. P. Strategy, Policy and Management, 7th Ed., South-Western Publishing Co., 1976.
Ile
3. ANSOFF, H. I. "The Changing Shape of the Stategic Problem",
J. of General Management, 4,4, Summer 1979, pp. 42-58.
4. GUTMANN, P. M. "Strategies for Growth", IBM Product Strategy
and Management Seminar Manual, University of Bradford Management
Centre, Oct. 16-17,1967, pp. 10-11.
5,
6.
7.
8.
9.
BROWNE, W. G. and KEMP, P. S. "A Three Stage Product Review Process",
Industrial Marketing Management, 5,1976, pp. 333-342.
CRAWFORD, C. M. "Strategies for New Product Development", Business
Horizons, Dec. 1972, pp. 49-58.
LAWRENCE, R. J. "Planning the Product Strategy", Marketing,
Dec. 1970, pp. 34-37.
KOTLER, P. Marketing Management: Analysis, Planning and Control,
4th Ed., Prentice-Hall, 1980.
GILMORE, F. F. and' BANDENBURG, R. G. "Anatomy of Corporate Planning",
IBM Product Strategy and Management Seminar, University of Bradford
Management Centre, Oct. 16-17,1967, pp. 15-25.
10. CUNNINGHAM, M. T. and HAMMOUDA, M. A. A. "Product Strategy for
Industrial Goods", The J. of Management Studies, May 1969, pp.
223-242.
11. BERRY, D. "Strategic Planning - Stability and Growth in the Face
of Adversity", Industrial Marketing Management 5,1976, pp. 209-211.
405
12. WILSON, I. H. et. al. "Strategic Planning for Marketing",
Business Horizons, Dec. 1978, pp. 65-73.
13. BELL, P. "P_ow to Cope with Uncertainty", Management Today, April 1978, pp. 67-70.
14. OMAN, D. B. "New Product Programs"; 1st Ed., American Management
Association Inc., 1969. f
15. CAPON, N. and HULBERT, J. "Decision Systems Analysis in Industrial
-Marketing", Industrial Marketing Management, 4,1975, pp. 143-160.
16. CARDOZO, R. N. and ROSS, I. "New Product Decisions by Marketing
Executives", J. of Marketing, Jan. 1972, pp. 10-16.
17. SOUDER, W. E. "Effectiveness of Product Development Methods",
" Industrial Marketing Management, 7,1978, pp. 299-307.
18. FOGG, C. D. "New Business Planning: The Internal Development
Process", Industrial Marketing Management, 5,1976, pp. 179-195.
19. COOPER, R. G. "Why New Industrial Products FaiZ", Industrial
Marketing Management, 4,1975, pp. 315-326.
20. RAWNSLEY, A. "Manual of Industrial Marketing Research", 1st Ed.,
J. Wiley and Sons, 1978.
21. B. S. I. Electric Overhead Travelling Cranes for Cenral Use in
Factories, Workshops and Warehouses, British Standards, BS466,1960.
22. LOVE, S. F. Planning and Creating SuccessfuZ Engineering Designs,
1st Ed., Van Nostrand Reinhold Company, 1980.
23 COOPER, R. G. "The Di.:. ens^., ons of Industrial New Product Success
and Failure", J. of Marketing, 43, Summer 1979, pp. 93-103.
406
24. ROTHWELL, R. et. al. "SAPPHO Updated - Project SAPPHO Phase II", Research Policy, 3,1974, pp. 258-291.
25. GERSTENFELD, A. "A Study of Successful Projects, Unsuccessful
Projects, and Projects in Process in West Germany", IEEE
Transactions on Engineering Management, Aug. 1976, pp. 116-123.
26. ANON,, "EZectric Wire Rope Hoist Market Survey and Recorrnnendations for a New EHB Range", H. M. Ltd. Internal Report, Aug. 1977.
27. -REA, W. H. "Designing for the Engineering Industries" The Engineering Designer, Oct. 1969, pp. 16-18.
28. WIND, Y. "A New Procedure for Concept Evaluation". J. of Marketing, 37, Oct. 1973, pp. 2-11.
29. COOPER, R. G. "Introducing Successful New Industrial Products", European Journal of Marketing, 10,1976, pp. 298-329.
30. EVANS, R. H. "Assessing Introduction Factors for a New Industrial
Product", Industrial Marketing Management, 7,1978, pp. 128-132. I
31. BENTLEY, T. "ldeasuring Product Profitability'; Management Accounting, July/Aug. 1979, pp. 17-18.
32. WASSON, C. R. "The Importance of the Product Life Cycle to the
Industrial Marketer", Industrial Marketing Management, 5,1979,
pp. 299-308.
33. POLLI, R. and COOK, V. "Vatidity of the Product Life CycZe", The J. of Business, Oct. 1969, pp. 385-400.
34. DOYLE, P. "The Realities of the Product Life Cycle", Quarterly
Review of Marketing, Summer 1978, pp. 1-6.
407
35. RINK, D. R. "The Product Life Cycle in Formulating Purchasing
Strategy", Industrial Marketing Management, 5,1976, pp. 231-242.
36. FOX, H. W. and RINK D. R. "Purchasing's RoZe Across the Product
Life CycZe", Industrial Marketing Management, 7,1978, pp. 186-192.
37. CHAMBERS, J. C. et. al. "Row to Choose the Right Forecasting
Technique", Harvard Business Review, July/August 1971, pp. 45-74.
38. ROGERS, E. M. "New Product Adoption and Diffusion", J. of Consumer
Research, 2, March 1976, pp. 290-301.
39. LANCASTER, G. A. and WHITE, M. 'Industrial Diffusion, Adoption
and Communication", European J. of Marketing, 10,5,1976,
pp. 280-298.
40-. . MAHAJAN, V. et. al. "A New Product Growth Model with Dynxnic
Market Potential", Long Range Planning, 12,1979, pp. 51-58.
41. MAHAJAN, V. and MULLER, E. "Innovation Diffusion and New Product
Growth Models in Marketing", J. of Marketing, 43, Fall 1979, pp. 55-68.
42. NEWALL, J. "Industrial Buyer Behaviour", European J. of Marketing,
11,1977, pp. 166-211.
43. GRONHAUG, K. "Search Behaviour in Organisational Buying",
Industrial Marketing Management, 4,1975, pp. 15-23.
44. - LEHMANN, D. R. and O'SHAUGHNESSY J. "Difference in Attribute
Importance for Different Industrial Products", J. of Marketing,
38, April 1974, pp. 36-42.
45. CUNNINGHAM, M. T. and WHITE, J. G. "The Behaviour of Industrial
Buyers in Their Search for Suppliers of Machine Tools'; The J.
of Management Studies, May 1974, pp. 115-128.
408
46. WEBSTER, F. E. "Modeling the Industrial Buying Process", J. of Marketing Research, 2,3, Nov. 1965, pp. 370-376.
47. CUNNINGHAM, M. T. and WHITE, J. G. "The Determinants of Choice of Supplier", European J. of Marketing, 7,3,1973, pp. 189-202.
48. KISER, G. E. et al. "Clues to the Design of a Marketing European J. of Marketing, 8,2,1974, pp. 168-179.
49. BECKWITH, N. E. and LEHMANN, D. R. "The Importance of Differential Weights in Multiple Attribute Models of Consumer Attitude",
J. of Marketing Research, 10, May 1973, pp. 141-5.
50. STARKEY, C. V. "Profit by Design", the Engineering Designer,
June 1969, pp. 12-17.
5T. - PEKELMAN, D, and SEN, S. K. "Mathematical Programming Models for
the Determination of Attribute Weights", Management Science, 20, 8, April 1974, pp. 1217-1229.
52. HUBER, G. P. "Multi Attribute Utility Models: A Review of Field
and Field-Like Studies", Management Science, 20,10, June 1974,
pp. 1393-1402.
53. WILDT, A. R. and BRUNO, A. V. "The Prediction of Preference for
"Capital Equipment using Linear Attitude Models", J. of Marketing
Research, 11, May 1974, pp. 203-5.
54. - ROTHSCHILD, W. E. "Competitor Analysis: The Missing Link in
strategy", The McKinsey Quarterly, Autumn 1979, pp. 42-53.
S STARR, M. K. Product Design and Decision Theory, 1st Ed.,
Prentice-Hall, 1963.
56. OSTROFSKY, B. Design, Planning and Development Methodology,
ist Ed., Prentice-Hall, 1977.
409
57. ARCHER, E. R. "A Decision Making Model", Industrial Engineering
April 1975, pp. 27-29.
58. SHAFTEL, T. L. "How Modular Design Reduces Production Costs",
Arizona Review, 21,6-7 June-July 1972, pp. 1-5.
59. HARVEY, G. "Life CycZe Costing: A Review of the Technique",
Management Accounting, Oct. 1976, pp. 343-347.
60. DEPARTMENT OF INDUSTRY, "Terotechnology: An Introduction to the
Management of Physical Resources", HMSO, 1975.
61. TEDFORD, J. D. and FERRIS, T. J. "The Application of Group
Technology at the Production Planning Phase of a New Product",
The Production Engineer, Nov. 1976, pp. 643-646.
62. BURBIDGE, J. L. The Introduction of Group Technology, 1st Ed.,
Heinemann, 1975.
63. DEVARAJAN, D. A. "Care and Feeding of MRP", Industrial Engineering
Aug. 1978, pp. 22-27.
64. BELIANIN, P. N. "Industrial Robots in Production Automation",
The Industrial Robot, Sept. 1978, pp. 141-143.
65. STARKEY, C. V. "Profit By Design: Part II Design Decisions"..
The Engineering Designer, May 1969, pp. 12-16.
66.. SANDERS, G. S. "The Long Range Planning of Research", British
Institute of Management, Long Range Planning for Profits and
Growth, Product Market Development Planning, July 1968.
67. CLARET, J. "To Make or Buy - It's More than a Question of Economy",
Accountants Weekly, 28th April, 1978, pp. 20-21.
410
68. HAAS, R. W. and WOTRUBA, T. R. "Marketing Strategy in a Make or Buy Situation", Industrial Marketing Management, 5,1976,
pp. 65-76.
69. LEVY, H. and SARNET, M. "The Make or Buy Decision", J. of General Management, 4,1, Autumn 1976, pp. 46-50.
70. " TAHA, H. A. Operations Research: An Introduction, 2nd Ed.,
Collier MacMillan, 1976.
71. WILD, R. Essentials of Production and operations Management,
1st Ed., Holt, Rinehart and Winston Ltd., 1980.
72. OSTWALD, P. F. and TOOLE, P. J. "How Do You Handle Tooling Costs", Industrial Engineering, Nov. 1975, pp. 19-23.
73. BURBIDGE, J. L. "The Pseudo Economic Batch Quantity", Institute
of Production Control, Jan/Feb. 1981, pp. 3-8.
74. CRUM, L. W. Value Engineering: The Organised Search for Value,
1st Ed., Longman Group Ltd., 1971.
75. STOCKTON, D. J. "New EHB Range: Value Engineering, Parts I-III",
H. M. Ltd. Internal Reports, June 1980, pp. 1-23.
76. SWANSON, E. "Determination of Standard Sizes to be Manufactured
Using Dynamic Programming - An Extension", AIIE Transactions,
Sept. 1970, pp. 191-202.
77. JACKSON, B. M. and ZERBE, D. R. "Determination of Standard Sizes to
be Manufactured Using Dynamic Progrru77fling", The J. of Industrial
Engineering, 19,8, Aug. 1968, pp. 401-406.
78. MARTIN, M. J. C. and PIFF, C. I. "Case Study: Determining the Range
of Standard Sizes of Capital Goods to be Manufactured", International
J. of Production Research, 9,4,1971, pp.. 443-456.
411
79. WOLFSON.. M. -L,
"Selecting the Best Lengths to Stock", Operations Research, 13,1965, pp. 570-585.
80. REUTER, V. G. "Using Graphic Management Tools", J. of Systems
Management, 30, April 1979, pp. 6-17.
81. BUFFA, E. S. Modern Production Management, 5th Ed., John Wiley
and Sons, 1977, pp. 466-467. /1
82. SMITH, L. A. and MAHLER, P. "Comparing CommerciaZZy Available
. CPM/PERT Computer Programs", Industrial Engineering, April 1978,
pp. 37-39.
83. KAHALAS, H. "A Look at Major Planning Methods: Development,
Implementation, Strengths and Limitations", Long Range Planning,
11,4,1978, pp. 84-90.
84. CLELAND, D. I. and KING, W. R. Systems Analysis and Project
Management, Ist Ed., McGraw-Hill, 1968.
85. BROOKS, D. T. "Designing for ProducibiZity: Design Influenced
Production Cost Progrcun", Society of Automotive Engineers,
National Aeronautic and Space Engineering and Manufacturing
Meeting, Los Angeles, Calif., Sept. 28-30,1971, SAE Report
No. 710747, pp. 1-6.
86. JOHNSON, R. C. Optimum Design of Mechanical Elements, 1st Ed.,
John Wiley and Sons, Inc., 1961.
87. SIEGEL, M. J. et. al. Mechanical Design of Machines, 4th Ed.,
International Textbook Company, 1965.
88. HERZOG, R. E. "Engineering and Purchasing: Partners in Design",
Machine Design, Aug. 23rd 1973, pp. 46-50.
412
89. SIMS, E. R. "Precision Costing of Manufacturing Operations",
Society of Automotive Engineers, Report No. 680575, pp. 1-16.
90. SIZER, J. An Insight into Management Accounting, 1st Ed.,
Penquin Books, 1977.
91. BLORE, D. "Taking a New Look at Work Measurement", The
Production Engineer, May 1981, pp. 34-36.
92. KNOTT, K. "MTM-V Application", The J. of Methods-Time
. Measurement, 4,3, pp. 22-27.
93. OWEN, E. J. "Estimating in Light Engineering", Work Study,
Oct. 1978, pp. 30-35.
94. MCGRATH, M. and CONLON, J. "Use Probability to Estimate Machine
Interference AZtouaances", Industrial Engineering, July 1970,
pp. 35-36.
95. NADLER, G. Work Systems Design: The Ideals Concept, Rev. Ed.,
Richard D. Irwin, Inc., 1970.
96. ANTHONY, T. F. and LAHIFF, J. M. "Estimate Acquisition: The
AchiZZes' Heel of Decision Making", Managerial Planning, July/
Aug. 1976, pp. 25-29.
97. OSTWALD, P. F. and TOOLE, P. J. "IE's and Cost Estimating",
Industrial Engineering, Feb. 1978, pp. 40-43.
98. VERNON, I. R. Realistic Cost Estimating for Manufacturing,
ist Ed., Society of Manufacturing Engineers, Michigan, 1968.
99. HITCHINGS, G. G. "Reliability of Production Norms in Manual
Repetitive Tasks", Work Study, Dec. 1978, pp. 30-37.
100. COOLING, W. C. "Understanding Production Standards", Work Study,
Dec. 1978, pp. 22-25.
413
101. SHACKEL, B. Applied Ergonomics Handbook, 1st Ed., IPC Science
and Technology Press Ltd., 1974.
102. FEIN, M. "A Rational Basis for Normal in Work Measurement",
. The J. of Industrial Engineering, 18,6,1967, pp. 341-346.
103. OWEN, E. J. "Realism in Work Measurement"3 Work Study, April 1981,
pp. 25-27.
104. MANSOOR, E. M. "An Investigation into Certain Aspects of Rating
Practice", The J. of Industrial Engineering, Feb. 1967, pp. 184-190.
105. FAHERTY, B. T. et. al. "The Quality of Work Measurement", Work
Study and Management Services, June 1970, pp. 467-470.
106. SMITH, S. "Better Work Standards Through Statistical Time
Analysis", The J. of Industrial Engineering, July/Aug. 1959,
pp. 261-265.
107. KLEIN, M. "Double ScunpZe Estimation in Work Measurement", The J.
of Industrial Engineering, May/June 1959, pp. 193-196.
. 108. ARCHER, E. R. and SIEMENS, N. "Quantitative Evaluation for
Repetitive Production Jobs", Industrial Engineering, March 1971,
pp. 10-15.
109. CARLSON, J. G. and ROWE, A. J. "How Much Does Forgetting Cost',
Industrial Engineering, Sept. 1976, pp. 40-47.
110. DUDLEY, N. A. Work Measurement: Some Research Studies, Ist Ed.,
MacMillan and Co. Ltd., 1968.
111. HITCHINGS, G. G. "Job Enrich'n nt", Work Study, Aug. 1978, pp. 28-35.
414
112. CARNALL, C. A. and BUTTERISS, M. "Job Design in the Process Industries", Chartered Mechanical Engineer, Sept. 1977, pp. 57-59.
113. CURRIE, R. M. Work Study, 4th Ed., Pitman Publishing, 1978.
114. HEWARD, J. "Computers and Work Study in the USA"., Work Study
and Management Services, Feb. 1970, pp. 102-114.
i 115. THRELFALL, J. "Making the Most of Work Study", Management
Accounting, April 1979, pp. 19.
116. MAYNARD, R. B. "Maynard Operation Sequence Technique", Work
Study, Nov. 1979, pp. 10-14.
117. WILLEY, D. E. "MTM--V: A User's Report", The J. of Methods-Time Measurement, 5,2, pp. 19-31.
118. BOEHRINGER, J. R. "50-Interval Chart Helps in Calculating
1.1 Incentive Rates", Industrial Engineering, June 1978, pp. 31.
119. SECOR, H. W. and KOGOVSEK, E. P. "PDS Work Measurement", The J.
of Industrial Engineering, April 1967, pp. 254-258.
120. STEELE, P. M. "Computerised Fitting and Assembly Estimating",
Work Study, Feb. 1981, pp. 10-16.
121. STEELE, P. M. "Case Study of Computerised Fitting and Assembly
Estimating", Work Study, Feb. 1981, pp. 34-39.
122. STOCKTON, D. J. Industrial Robots and Work Handling for Welding,
MSc. Thesis, Loughborough University of Technology, 1979. (Unpublished)
123. KIRBY, P. W. "Category Estimating". Work Study and Management
Services, April 1970, pp. 302-306.
415
124. HARRIS, D. G. and MELVEN, D. G. "Tootroom Work Measurement the Easy Way", The Production Engineer, Jan. 1974, pp. 26-30.
125. FAIRHURST, J. H. "The Accuracy of Comparative Estimating",
International J. of Production Research, 12,3,1974, pp. 399-407.
126. KELLER, L. W. "The Use of a Computer in Cost Estimating", ASTME Technical Paper No. MM67-80, pp. 1-13.
127. KOPP, K. K. "A Computer Program for Time Study Analysis",
The J. of Industrial Engineering, 18,2,1967, pp. 147-152.
128. MASON, A. K. and TOWNE, D. M. "Toward Synthetic Methods Analysis", The J. of Industrial Engineering, 18,1,1967, pp. 52-56.
129. WEAVER, R. F. et. al. "Developing Standards by Computer", Industrial Engineering, Jan. 1978, pp. 26-31.
130. HOLLINGUM, J. "Techniques for Measuring Work Yrainned in Computer
System", The Engineer, 236,12th April, 1973, pp. 54-57.
131. O'NEAL, M. H. and MOORE, C. "Mutti-Plant Computer System for
Standards Provides Tool for Overall Manufacturing Control",
Industrial Engineering, Aug. 1980, pp. 54-62.
132. ANON. "A Packaging Package", Computer Management, Nov. 1976,
pp. 27-36.
133. HILDEBRANDT, H. and WOOD, H. "Mamvmoth Task just a Mini Job",
Computing Europe, 7th July, 1977, pp. 14.
134. ANON. "Estimating System for Vickers", Computer-Aided Design,
10,2,1978, pp. 149-150.
416
135. HALEVI, G. and STOUT, K. J. "A Computerised Planning Procedure for Machined Components", The Production Engineer, April 1977, pp. 37-42.
136. LOGAN, F. "Let LOCAM Take the Paperwork out of Planning", The Production Engineer, June 1981, pp. 15-18.
137. MAHMMUD, M. A. M. and PUGH, S. "Turned Component Costs", Engineering, Jan. 1981, pp. 35-37.
138. 'MAHMOUD, M. A. M. The Costing of Turned Components at the Design
Stage, Ph. D. Thesis, Loughborough University of Technology, 1979. l(Unpublished)
139. MOORE, S. J. "Calculation of Machining Times", Work Study, Oct. 1978, pp. 12-20.
140. CHISHOLM, A. J. et. al. The Action of Cutting Tools, Ist Ed., Machinery Publishing Co. Ltd., 1959, pp. 57.
141. KEMPSTER, M. H. A. Introduction to Workshop Technology, 3rd Ed., The English Universities Press Ltd., 1972, pp. 372.
142. HASLEHURST, M. Manufacturing Technology, 1st Ed., The English Universities Press Ltd., 1969, pp. 32.
143. ANON. MAAG Gear Book, Ist Ed., MAAG GearWheel Co. Ltd., Zurich, 1963, pp. 215-216.
144. ANON. Machine Shop Synthetics, PE Consulting Group Ltd., 1979.
145. BHATTACHARYYA, S. K. and COATES, J. B. "The Assessment of Manufacturing
Technology", International J. of Production Research, 12,2,1974,
pp. 159-177.
417
146. RAMASWAMY, K. V. and LAMBERT, B. K. "Cost Optimization for a Single Pass Turning Operation Including Inventory and Penalty
Costs", International J. of Production Research, 12,3,1974,
pp. 331-344.
147. RAVIGNANI, G. L. Machining for a Fixed Demand - An Approach to
Optimization", International J. of Production Research, 12,3,1974,
pp. 361-376.
148. BARTALUCCI, B. "On the Optimum Selection of Machine Tools in
Turning Operations", Annals of the CIRP, 17,1969, pp. 445-452.
149. BALLARD, C. and HICHKAD, R. "An Equation to Estimate Assembly
Time", Industrial Engineering, Aug. 1969, pp. 20-22.
150. OLIVER, R. 'An OR Approach to Work Center Standards", Industrial
Engineering, Dec. 1969, pp. 21-24.
151. THELWELL, R. R. "On Evaluation of Linear Programming and Multiple
Regression for Estimating Manpower Requirements", The J. of
-Industrial Engineering, 18,3,1967, pp. 227-236.
152. CROCKER, D. C. "IntercorreZation and the Utility of Multiple
Regression in Industrial Engineering", The J. of Industrial
Engineering, 18,1,1967, pp. 79-85.
4
153. CANNING, F. L. "Estimating Load Requirements in a Job Shop",
The J. of Industrial Engineering, 10,, 6,1959, pp. 447-449.
154. STOCKTON, D. J. "An Approach to Improving Cost Estimating",
International J. of Production Research, 20,6,1982, pp. 741-751.
155. COCHRAN, E. B. "Using Regression Techniques in Cost Analysis:
part I", International J. of Production Research, 14,4,1976,
pp. 465-468. "Using Regression Techniques in Cost Analysis:
Part II", International J. of Production Research, 14,4,1976,
pp. 489-511.
418
156. SALEM, M. D. "Multiple Linear Regression Analysis for Work Measurement of Indirect Labour", The J. of Industrial
Engineering, 18,5,1967, pp. 314-319.
157. DONEY, L. and GELB, T. "Regression Short-Cuts CycZe-Time
Estimating", Industrial Engineering, Feb. 1971, pp. 22-23.
158. RICHARDSON, W. J. "Measurement of Indirect Work using Multiple
Regression", International J. of Production Research, 9,4, 1971, pp. 481-486.
159. MELLOR, R. 'Mlultiple Regression'; Work Study and Management Services, July 1970,, pp. 572-574.
160. KLEIN, R. S. and TAIT, H. J. "Faster Better Tooling Estimates", Industrial Engineering, Dec. 1971, pp. 12-17.
161. COATES, J. B. "Tool Costs and Tool Estimating", The Production
Engineer, April 1976, pp. 201-208.
162. MARTIN, D. D. "Instant Time Standards", Industrial Engineering, Nov. 1971, pp. 24-27.
163. KOLOC, J. "The Use of Workpiece Statistics to Develop Automatic
Progrcvrnning for N. C. Machine Tools", International J. of Machine Tool Design and Research, 9,1969, pp. 65-80.
164. OPITZ, H. et. al. "Workpiece Classification and its Industrial
Application", International J. of Machine Tool Design and Research,
9,1969, pp. 39-50.
165. GOMBINSKI, J. "Fundamental Aspects of Component Classification", Annals of the CIRP, 17,1969, pp. 367-375.
419
166. FOGG, B. et. al. "Component Classification as an Aid to the Design
of Press Tools", Annals of the CIRP, 19,1971, pp. 141-151.
167. SWIFT, K. G. and REDFORD, A. H. "Classification for Automatic
Assembly of SmaZZ Products", Annals of the CIRP, 27,1978,
pp. 435-440.
, 168. REMMERSWAAL, J. L. and SCHILPEROORT, B. A. "An Evaluation of
Existing Classification Systems in Production Engineering",
Annals of the CIRP, 20,1,1971, pp. 91-92.
169. CRAWFORD, S. Basic Engineering Processes, 2nd Ed., Hodder and
Stoughton, 1972.
170. BOSTON, O. W. Metal Processing, 2nd Ed., John Wiley and Sons
Inc., 1958.
171. ANON. Standard Data for Arc welding, The Welding Institute
Publication, 1978.
172. ANON. "Welding Economics: Some Basic Practical Aspects",
Welding and Metal Fabrication, March 1977, pp. 85-91.
173. MYERS, L. W. "How to Calculate and Control Welding Costsºº,
Welding Design and Fabrication, Nov. 1978, pp. 78-86.
174. EZEKIEL, M. and FOX, K. A. Methods of Correlation and Regression
Analysis, 3rd Ed., John Wiley and Sons, Inc., 1959.
175. HAWKINS, J. "Applying Matrices to Regression", Management
Accounting, April 1979, pp. 37.
176. NIE, N. H. et. al. SPSS: Statistical Package for the Social
Sciences, 2nd Ed., McGraw Hill Book Co., 1975.
420
177. MISTRY, B. and STOCKTON, D. J. "DESCOSTIMATE", H. M. Ltd.,
Internal Report, 1981.
178. PEDDER-SMITH, D. W. "Price, Cost and Product Size", Management Accounting, April 1976, pp. 136-137.
179., AUSTRALIAN STANDARDS. Drwns and Sheaves, AS1418 Part 1,1977.
i 180. FEM. Rules for the Design of Serial Lifting Equipment,
Federation Europeenne de la Manutention, Section 9.
181. THOMAS, A. B. Stock Control in Manufacturing Industries, 1st Ed.,
Gower Press Ltd., 1968.
182. HILL, L. S. "Towards a Co-ordinated Approach to the Structuring
of Cost and Design Models: A Case Example", Operations Research,
13,1965, pp. 619-631.
183. SWAN, A. W. and WORRALL, B. M. "A New Quantitative Technique for
Identification of Industrial and Organisational Problems
International J. of Production Research, 12,2,1974, pp. 313-330.
184. KING, J. R. "The Theory-Practice Gap in Job-Shop Scheduling",
The Production Engineer, March 1976, pp. 137-143.
185. NUTTMAN, L. et. al. "Computers - The Effective Way to Manage
Manufacturing", The Production Engineer, July/Aug. 1979, pp. 21-26.
186. HOLLINGUM, J. "Keeping Tabs on Shop Floor Activity is a job for
the microprocessor", The Engineer, 17th Jan. 1980, pp. 43,45.
187. HANCOCK, C. "Integrated Manufacturing Systems -a ChaZZenge",
The Production Engineer, Dec. 1979, pp. 49-50.
188. MORTIMER, J. "Computer Draughting a TooZ - Not a Takeover",
Eureka, Feb. 1981, pp. 47-48.
421
189. ANON. "CAD to be used in the Development of Nuclear Reactors", Engineering Today, 3rd August, 1981, pp. 8.
190. SIMKINS, P. "The Computer as a Production TooZ'; Chartered Mechanical Engineer, Jan. 1980, pp. 53-55.
191. NUSSEY, I. D. "Manufacturing and Computers: PROLAMAT '76", The Production Engineer, April 1977, pp. 51-52.
192. ANON. "Easy Automation for the Drawing Office", Engineering Today, 22nd June, 1981, pp. 11.
193. WALKER, B. Designing by Computer", Engineering, Oct. 1980,
pp. 1104-1108.
194. HEWITT, E. C. "Computer Assisted Design and Draughting at Davy-Loewy Limited", Managing Computer Aided Design, I. Mech. E. Conf. Publications, 1980, pp. 9-20.
195. BOND, A. "Starting CAD at a Lower Entry Fee", Computer Systems,
June 1981, pp. 78-81.
196. HOLLINGUM, J. "Computer Graphics Playing a More Vital RoZe"3 The Engineer Survey, 24th July, 1980, pp. 28-29.
197. WATTS, G. A. and VEENMAN, P. "Computer-aided Design in the Non- Ferrous Metals Industry", Computer-Aided Design, 8,1,1976,
pp. 13-16.
198. LAXON, W. R. "Selecting and Evaluating CAD Systems", Computer-Aided
Design, 9,4, Oct. 1977, pp. 233-237.
199. - SHOPPEE, G. J. V. "Current Structural analysis systems and development tends", CAD 76, IPC Science and Technology Press,
Guildford, 1976, pp. 206-212.
422
200. HEATH, H. H. "MELBA Program Solves Bearing Problems", Design Engineering, May 1977, pp. 53-55.
201. NEWTON, F. J. et. al. "Design of Low-Noise Gas Valves", Computer Aided Design, 9,2, April 1977, pp. 135-140.
202. COATES, P. J. "Design: Ends and Means", Computer-Aided Design, 1,1, Autumn 1968, pp. 27-32.
203. MASTERS, E. H. "The Engineering Design Functions which the Computer
. can most Effectively Aid" Computer-Aided Design, Autumn 1968,
pp. 33-38.
204. PITTS, G. "Computer Aided Design: Use and Misuse", Computer-Aided
Design, Summer 1970, pp. 41-43.
205. COOLEY, M. J. E. "Impact of CAD on the Designer and the Design
Function", Computer-Aided Design, 9,4, Oct. 1977, pp. 238-242.
206. ALGER, J. R. M. and HAYS, C. V. Creative Synthesis in Design, Ist Ed., Prentice-Hall, 1964.
207. COLDING, B. "Machining Economics and Industrial Data Manuals",
Annals of the CIRP, 17,1969, pp. 279-288.
208. COATES, J. B. et. al. "Improve Productivity by Establishing a Technological Data Base", Management Accounting, Dec. 1976,
pp. 437-440.
209. KNOX, C. "CAD/CAM and Group Technology: The Answer to Systems
Integration", Industrial Engineering, Nov. 1980, pp. 66-73.
210. ANON. 'International Report: Sverre Munck A/S'; Dun and Bradstreet, Inc., 31st Jan., 1980.
423
211. BRITISH STANDARDS INSTITUTION. Specification for Permissible
Stresses in Cranes and Design Rules, BS2573, Part 2,1980.
212. SIMPSON, L. "How to Make Better Decisions", Management Accounting, Jan. 1976, pp. 21-22.
213. ROMANO, A. Applied Statistics for Science and Industry, 1st Ed.,
Allyn and Bacon Inc., 1977.
214. BAKER, R. J. and NELDER, J. A. Generalised Linear Interactive
Modelling: The GLT. 4 System, 1st Ed., Royal Statistical Soceity,
1978.
215. BSI. "Guide to the Operation and Evaluation of Quality Assurance
Systems", BS5179, Part 3, December 1974.
21.6. BSI. "British Standard Guide to Design Management Procedures
for the Assurance of Quality", Draft for Public Comment,
79/62358 DC, May 1979.
424
BIBLIOGRAPHY
LILIEN, G. L. and LITTLE, J. D. C. "The ADVISOR Project: A Study of Industrial Marketing Budgets", Sloan Management Review, 17, Spring 1976,
pp. 17-31.
SAUNDERS, A. and SMITH, N. J. "Hydraulic Cranes from Armstrongs
EZswick Works Z846-2936: A Product Life Cycle in Capital Goods",
IMRA Journal, May 1970, pp. 55-70.
WILLSMER, R. "Long-range Corporate Planning and the Marketing Men",
Management Decision, Autumn 1967, pp. 12-15.
REISMAN, A. "Forecasting Short-Term Demand", Industrial Engineering,
May 1976, pp. 38-45.
PARKER, G. G. C. and SEGURA, E. L. "How to get. a better forecast",
Harvard Business Review, £4arch/April 1971, pp. 99-109.
KELLY, J. P. and COAKER, J. W. ' The Importance of Price as a Choice
Criterion for Industrial Purchasing Decisions", Industrial Marketing
Management, 5,1976, pp. 281-293.
CUNNINGHAM, M. T. and ROBERTS, D. A. "The RoZe of Customer Service in
Industrial Marketing", European J. of Marketing, 8,1,1974, pp. 15-41.
BANTING, P. M. "Customer Service in Industrial Marketing: A Comparative
Study", European J. of Marketing, 10,3,1976, pp. 136-145.
GOTTWALD, C. "Decision Making in Preliminary Design", Society of
Automotive Engineers, Report No. 730889,1973, pp. 1-8.
HEGINBOTHAM, W. B. and TEWARI, N. K. 'A Mechanised Assembly System:
Its Influence on Component Design and Tolerances", International J. of
Production Research, 16,1,1978, pp. 63-75.
425
DOGRAMACI, A. "Design of Common Components Considering Implications of Inventory Costs and Forecasting", AIIE Transactions, June 1979,
pp. 129-135.
BURBIDGE, J. "What is Wrong with Materials Requirement Planning", The Production Engineer, Oct. 1980, pp. 57-59.
PENDLEBURY i A. J. "Trends in Production and Inventory Management", Management Accounting, Jan. 1979, pp. 18-22.
SATO, N. et. al. "Group Technology and Materials Requirements Planning: An Integrated Methodology for Production Control, ", Annals of the CIRP, 27,1978, pp. 471-473.
SERPA, J. R. "Manufacturing and Design Engineering: Partnership in
Cost Reduction", IEEE Transactions on Manufacturing Technology, Dec. 1972,
pp. 13-15.
BIRCHALL, J. "Marketing and Design: Part One", Engineering Designer, Sept/Oct. 1975, pp. 13-16.
HERZOG, R. E. "Designing for Foreign Markets", Machine Design, Aug. 22,
1974, pp. 73-77.
MILITZER, R. W. "Needed: Manufacturing Engineering Input to Product
Design", Manufacturing Engineering, 82,2nd Feb., 1979, pp. 2.
HODSKINSON, M. G. "On VaZue Added by Design", Engineering Designer,
Nov/Dec. 1978, pp. 15-16.
BOOTHROYD, G. "Design for Economic Manufacture", Annals of the CIRP,
28,1,1979, pp. 345-350.
GORB, P. 'Design and Its Use to Managers", J. of the Royal Society of Arts, Feb. 1980, pp. 144-158.
426
ARCHER, B. "Design as an Element in a New Approach to Britain's Industrial Future", Quarterly Review of Marketing, Winter 1976,
pp. 1-7.
RAO, V. R. and SOUTAR, G. N. "Subjective Evaluations for Product Design Decisions", Decision Sciences, 6, Jan. 1975, pp. 120-134.
OXFORD, T. R. "Designing for Production"3 Engineering Designer, Feb. 1976, pp. 1'1-14.
PHILBIN, L. and RENEGAR, B. "Federal Express Corp. Develops Performance Measurement Index for Indirect Labour Sector", Industrial Engineering,
Aug. 1980, pp. 48-53.
KANON, D. "Application of Nonparametric Statistics to Industrial
Engineering", The J. of Industrial Engineering, 18,4,1967, pp. 272-276.
ALLAN, W. G. "Decision Systems and Work Study", Work Study and Management
Services, Nov. 1970, pp. 882-889.
ANON. "MTM-3 and Integrated Work Measurement", Work Study and Management
Services, Oct. 1970, pp. 833.
LASHER, W. R. and WECHSLER, D. B. "Estimating Electronics Assembly Costs", Industrial Engineering, Sept. 1972, pp. 17-21.
BLANCHFIELD, D. E. et. al. "A Simplified Approach for Estimating
Manufacturing Cost at the Part Level", Society of Automotive Engineers,
Report No. 730908, pp. 1-9.
MURPHY, R. "A Computerised Standard Data System", Industrial Engineering,
Oct. 1970, pp. 10-17.
BRISLEY, C. L. and DOSSETT, R. J. "Computer Use and Non-Direct Labour
Measurement will transform Profession in the new decade", Industrial
Engineering, Aug. 1980, pp. 34-43.
427
PARK, W. R. "How to Make Quick Manufacturing Cost Estimates", Consulting Engineer, Jan. 1972, pp. 44,46 and 48.
KIDDNEY, R. P. "A SmaZZ Company's Guide to Computerised Shop Standards", Industrial Engineering, Sept. 1978, pp. 54-55.
LIPTON, P. R. and DUFFY, D. J. "An Application of Statistical Control to a Work Measurement Problem", The J. of Industrial Engineering, 18,9,
01 1967, pp. 526-529.
DAVIES, E. C. "Time Study on TriaZ", Work Study and Management Services, June 1970, pp. 462-465.
ALLEMANG, R. D. "New Technique CouZd RepZace Learning Curves", Industrial Engineering, Aug. 1977, pp. 22-25.
KINACK, R. J. "Activity Evaluation Technique -A Quick and Easy Procedure for Developing Time standards"; The J. of Industrial Engineering, 18,5, 1967, pp. 13-16.
ALLEN, D. S. "Comparing Productive Activities at Different Times" Industrial Engineering, March 1978, pp. 40-43.
BECK, C. "Work SannpZing Sets Direct Labour Standards", Industrial
Engineering, Feb. 1969, pp. 22-25.
LIPTON, P. R. "An Application of Factorial Experimentation to the Work Measurement Process", The J. of Industrial Engineering, 18,8,1967,
pp. 449-455.
NANDA, R. "Developing Variability Measures for Predetermined Time
Systems", The J. of Industrial Engineering, 18,1,1967, pp. 120-122.
CHRIST, C. "MRT: An Analytical Way to Measure Indirect Work",
Industrial Engineering, June 1969, pp. 36-38.
428
NADKARNI, R. P. "Activity Sequence Analysis", Work Study, June 1978,
pp. 35-38.
SMERILSON, H. H. "Standards for Engineers", Industrial Engineering, Oct. 1975, pp. 1217.
MODER, J. J. "Activity Sampling with Applications to Time Standard
Estimation", The J. of Industrial Engineering, 18,1,1967, pp. 24-29.
ANON. "Employing Synthetics in the Workshop", Work Study, 24, Feb. 1975,
pp. 32-36.
ANON. "Obtaining Synthetic Values for use in Grinding Work", Work
Study, 24, Nov. 1975, pp. 32-42.
EILON, S. "Job Shop Scheduling"; The Production Engineer, Jan/Feb. 1977,
pp. 47-48.
FREEMAN, J. M. "Factors Affecting Cost Behaviour", Management Accounting,
June 1976, pp. 226-227.
HARLOW, I. P. "Cost Models HeZp to Answer the Questions"3 Management
Accounting, July/Aug. 1976, pp: 266-268.
GUTMAN, N. "How to Keep Product Costs in Line", Machine Design,
22nd Jan., 1981, pp. 50-53.
RIST, L. "From CAD to CAM via Expanded Factory Applications of Computers",
American Machinist, March 1980, pp. 134-137.
429
APPENDIX I
DEVELOPMENT OF THE NEW EHB RANGE
430
Initially the existing Ropemaster hoist concept was designed as
an electric hoist block for runway duty. However in the early 1950's
the hoist unit then in production was adapted for use on cranes in an
endeavour to match competitors strategy.
During this period the UK market was supplied from a large number
of local manufacturers with H. M. Ltd. holding the predominant market
share. In the late 1950's/early 1960's European competitors began to
enter the market seriously with units designed to FEM standards(180)
originally as suppliers of hoist blocks to end users but gradually
entering the crane field through minor manufacturers and increasing
their penetration because of the competitiveness of the hoist block
type crane. The Ropemaster during this period was modified to increase
the range from a maximum of 5.5 tonnes to a maximum of 10 tonnes lifting
capacity.
The success of H. M. Ltd. 's Universal Crane range launched in 1962,
the acceptance of this type of crane within the market and the gradual
increase in crane capacities and spans resulted in further extensions
of the Ropemaster range so that in 1969, a standard 4-Fall 10-tonne unit
was introduced which was subsequently stretched to 20 tonnes capacity.
The period from 1960-1974 showed an increase in the market
requirement for electric hoists so that whilst the market share of
H. M. Ltd. was being eroded by foreign imports, high volumes were
being sustained with order levels of 1300 in 1969,1200 in 1970,
431
900 in 1971,900 in 1972 and 1250 in 1973. Extended delivery times
due. to production constraints during this period resulted in regular
customers turning to other suppliers to meet their requirements.
Since 1974 Ropemaster sales have declined (particularly for units
sold for runway applications) with current sales levels being approximately
700 per annum. The ratio of "hoist blocks sold for runway duty" to
"blocks sold as part of a Universal crane" is approximately 1.5: 1.
The delivery constraints created market acceptability for competitors
which due to the concept of their design H. M. Ltd. have subsequently
been unable to overcome.
The delivery constraints arose due to:
(i) Complex production methods being used.
(ii) Poor component supply.
(iii) Lack of priority given to the manufacture of the
Ropemaster.
Thus, whilst in the early 1970's 40% of the machine shop capacity
was occupied with Ropemaster production, this is currently running at
22%.
1 432
Traditional markets, predominently home orientated, have shrunk
and due to the Ropemasters design philosophy H. M. Ltd. have been unable
to replace this by export sales. The design is labour intensive and
coupled with the low volume is uncompetitive in the market place as
a straight hoist. When used on a crane, the cost of the hoist unit
to the customer is not separately identified but, whilst still
technically acceptable in the home market the price content of the
hoist within the crane necessitates selling at a lower profit margin.
The uncompetitiveness of the Ropemaster as a straight hoist has
led to a change in the hoist concept within the organisation. The
Ropemaster unit is now an essential part of the total crane philosophy
and not a secondary EHB product range. The development of hoist block
cranes through market acceptance is again emphasising the need to
extend the capacity of the current range of cranes. However, the
Ropemaster cannot be stretched to meet these objectives.
It is not possible to increase hoist sales volumes appreciably
since the present Ropemaster design is no longer acceptable to other
crane distributors and assemblers who form the major portion of home
and export markets. The design philosophy means either a high stock
commitment by clients or alternatively long delivery times since each
unit must effectively be built to contract. Current design parameters
do not allow a ready change of suspension method, lifting capacity by
re-reeving or any optional extras such as speed variations. These can.
be readily accommodated by competitors products. It is therefore an
obsolete design, expensive to produce and inadaptable to meet current
market demands.
433
Since the hoist block is an integral part of the Universal crane
it was important that H. M. Ltd. continue to market such a product.
This would also enable the company to retain its market image (built
up over several decades) as materials handling experts.
Four options existed, each of which could allow H. M. Ltd. to
continue to market a hoist block.
Option 1
To factor complete units from another manufacturer. This could
have been achieved by buying low cost hoist units from the distributor
Kone (UK) Ltd. This would have allowed H. M. Ltd. to maintain their
current selling prices but with reduced profit margins. However, the
price structure of Kone (UK) Ltd. represented a "dumping exercise" to
enable them to penetrate the market. This eventually could result in
the setting up of their own manufacturing facilities which could lead
to a gradual decrease in H. M. Ltd. 's crane sales. This option also
limited H. M. Ltd. 's export potential.
The decision not to manufacture a hoist block range would also
produce the following problems:
(a) Since the machine tool investment programme involving
1200,000 over the last 6 years has been primarily aimed
at improving Ropemaster manufacturing productivity the
majority of this capital expenditure would be wasted.
1 434
(b) Approximately 37% of the machine shop operatives would
need to be made redundant. The redundancy would lower
the flexibility of the manufacturing facilities across
the company's whole product range and prevent a quick
response to unusual market demands.
(c) ' Redundancy on such a scale could have a repercussive
effect on the workforce in other areas with a consequent
fall-off in performance generally.
(d) The other product groups would need to absorb the
remaining manufacturing expenses and would become less
profitable as a result.
(e) There are many components common to the Ropemaster and
other product ranges. Inevitably therefore the cost of
materials would increase due to the reduction in volumes.
(f) The profitability of the spares division would be
lowered.
(g) Increased stock holding costs would be incurred.
(h) Market reaction could be unfavourable since many customers
place continuing business simply because it is an H. M. Ltd.
product. These customers may be reluctant to purchase a
foreign block through H. M. Ltd. and may prefer to purchase
direct. This attitude could also extend into the Universal
Crane business with consequent reduction in order input.
1 435
Option 2
To obtain a licensing agreement from a suitable foreign
manufacturer. This is not feasible since all current manufacturers
are suffering due to a down turn in the world market and are therefore
seeking to increase their in-house manufacturing levels and hence are
not interested at the present time in this type of arrangement.
Option 3
Collaborative manufacture of a composite range in conjunction
with another manufacturer. An examination of possible manufacturers
with whom to collaborate revealed that many were strong in the market
place with respect to hoist block manufacture and design expertise.
Consequently any deals would give them minimum advantages and a strong
hand in any negotiations.
However one company Swerre Munck A/S (Munck) of Norway showed
greater potential in this area, since they themselves appeared to be in
a difficult position in terms of a rapidly reduced turnover and recent
financial difficulties (210). Numerous discussions were held with
Munck management to establish the extent to which mutual advantages
existed. However, although there were many advantages to be gained
(i. e. reduction in the new product range introduction time due to
Munck already having part developed a new product range, reduction
in development cost and a possible extension of the product range
through joint manufacturing of an 8.0 tonne unit) agreement could
not be reached of the division of the in-house work content and the
amount of royalty paid by H. M. Ltd. for the previous development
work carried out by Munck.
436
Option 4
The fourth option was therefore the only feasible choice to
follow, i. e. to develop and manufacture a new range of wire rope
hoists. This option would then give maximum utilization of existing
plant, maximum overhead recovery, conform to the company's image as
in-house manufacturers of the total range of crane products, protect
interests in export markets and future growth and ensure that control
is retained of own manufacturing costs and selling prices.
However the disadvantages are long development time, current
low sales volumes of company's products negating capital investment,
investment of the technical resource at the expense of other areas,
limited experience in hoist block design philosophy and a possible
compromise on a limited range.
From the detailed market survey carried out in 1977(26) the
basic design specifications for the New EHB range were considered to
be:
(i) The hoist block should be cylindrical in shape with a
standard flange mounted squirrel cage motor at one end,
driving through a central drive shaft to a triple
reduction gearbox at the other end. The gearbox is a
separate module making assembly and replacement easy.
(ii) A standard failsafe brake is required, mounted at the
back-end of the gearbox and therefore easily accessible
by removal of a weatherproof cover.
437
(iii) The construction of the frame must be open for ease of
assembly of the rope and rope guide and to lend itself to
/
varying drum lengths with the minimum of components. The
design of theframe must allow units to be easily converted
to either normal or low headroom monorail units, crab
versions or fixed suspension units, as illustrated in
Figure 62.
-(iv) The unit must be designed to be easily convertable from
single to two or four rope fall versions.
(v) The units must conform to FEM Ib(180) and BS 2573 Part 2
M3(211) standards.
(vi) The range of capacities, heights of lift and lifting
speeds required are as follows:
Safe Working Load (SWL) Height of Lift Speed of Lift
(tonnes) 1 Fall of Rope (Metres) (Metres/min)
0.8 10 20
1.6 11 20
3.2 12 20
5.0 12 20
8.0 12 20
This is to allow the range to be extended as illustrated in
Figure 63 to other FEM groups at lower capacities.
438
FIGURE 62 SCHEMATIC DIAGRAM OF THE NEW EHB CONCEPT
Ref. (26)
ý h:
1, /ý'ý 1,
439
FIGURE 63 LOAD SPECTRUM OF THE NEW EHB RANGE
\. ý\ \\`11
1I1 211
ý- --- .
640 ý 1250
Hoisf pa it y kg k9 kg
160 320 640
200 40 0j__ 800
250 500 1000
320
1000 1 2000
800 ö00 __
IT
1000 2000 1250 2500
1250
1600
2 500
3200
411
3 2OO
4000
50 vO
6400
8000 2000 1 4000
L500 3200
4000
5000
6300
8000
10000
5000 1 10000
6400
8000 10000
12500
16000
20000
2500
"5000
20000
25000
32000
4-' j 'J
Lcad Spectrum
Light
Medium
Heavy
F E. M roup
<1
< 0.5
I6
8-16
4-8
-4
II Hill
-j i
-L. * 71 c
Ref. (26)
Average operating time p? r wcrkiny dai in hours
2-4 I4-8
1-2 ; 2-4 ý
0-5-1
a
ý
i
xc IN
!nj'. - --- ' ---- -ý ý .. ý
-1- ----___---- ; ý: ý '
I
vý ^\
W iý
K, itc
vý :, ý
i ý
%941 1
Z;
i ist
ýý 1
- >16 i
8-16 f >1t
>9 4-B 1
Ivv
i
ý ýý
ý
I ---ý
.ý ný I
! ý .: i:
S
.. O;
Y. iti
Ji /ý ý
440
(vii) For ease of service (which is intended to be a particular
design feature) standard proprietary motors and brakes
must be used as well as metric bearings, oilseals and
threaded components to meet export requirements.
(viii) From H. M. Ltd. 's marketing department a forecast of the
e future sales of the hoist are:
Sales Forecast
SWL Market 1st 2nd 3rd (tonnes) Demand Year Year Year
0.8 37 % 259 444 740
1.6 34 % 238 408 680
3.2 19 % 133 228 380
5.0 8% 56 96 160
8.0 2% 14 24 40
(ix) The unit must be designed for economic manufacture and
assembly by designing on a modular principle. The
modules are to be motor, gearbox, wire-rope drum tube,
frame, suspension unit, brake, rope guide and bottom
block (hook arrangement).
I 441
i
APPENDIX II
ESTIMATING SYSTEM: MANUAL METHOD DESCRIPTION
442
The sequence of tasks involved in estimating labour costs is
illustrated using the example of the 30 tonne Goliath Crane idler
spur gear (Drawing Number 4/14162/2) which is machined from an
EN24T forging:
(ý) From the component drawing (Figure 64) identify the
machining operations required to manufacture the part.
These are:
(a) Turn outside diameter (To1s.,: ý; O. lmm)
(b) Face both flange faces (Tols. )O. lmm)
(c) Face both boss faces (Tols. > O. lmm)
(d) Bore centre hole (Tols. )O. lmm)
(e) Grind centre hole
(f) Grind outside diameter
(g) Cut gear teeth
(2) Determine the appropriate cost centres using Table 46.
443
FIGURE 64 FIRST IDLER MT SPURWHEEL
A
139.70
I ý i, iý ý; i, i; ýýý
a
i'
il
1ý
i 1 ý i
i ý ý ý
i 1
nfl
I I
I I ý
I I
ý
92.10
A
O Ql
O co co O CO O
CC) L(7 M 00 M Q1
aM co
Tooth Depth = 14.29mm Number of Teeth = 45
Module = 6.35 Machining Limits = +0.40mm
All dimensions in mm's (unless otherwise stated)
Material EN24T Annual Quantity = 50
444
TABLE 46 PROCESS COST CENTRES
Pre-, Special and Post Machining General Machining
Processes Cost Machine Type Cost Centre Centre
/
Flame Cut 248 Bar Lathes 201
Saw 236 Bar Capstans
Shear 238
Weld 239
Gear Gutting 217
Milling 214
Grinding (Cylindrical) 216
Grinding (Surface) 216
Spline Cutting 270
Radial Drilling 211
Assembly 227
Chuck Lathes
202
203
Chuck Capstans 204
Centre Lathes 206
Vertical Borers 210
Horizontal Borers (Prismatic Operations) 209
NC Bar Lathe 258
NC Chuck Lathe 258
Copy Lathes 219
(3) If general machining operations are carried out and the
appropriate machine on which the component is processed is
not known, determine the relevant cost centre using Table 47,
which is extracted from Appendix V and shown here for
convenience.
The user initially chooses the correct section of the table
depending on whether or not a boring operation is required.
It is then necessary to move down the columns until a cost
centre is found where all conditions (i. e. component diameter,
445
TABLE 47 RULES FOR DETERMINING GENERAL MACHINING
COST CENTRE
Component Variables
Diameter (mm) Length (mm) Annual Quantity Input Code
.. e Boring Operation Required Cost Centre
es e 70 0,203 < 51 152
92 305
< 76 152
92 305
394 178
686 305
610 229
686 305
<1778 610
< 914 <5639
2 70 0,201 NC Bar
-e1000 1,258Lathe
ie 1000 0,201
70 0,204 Z
70 0,202 *
z1000 0,258 NC Chuc Lathe
, ei 000 0,202
191000 0,210
191000 0,206
No Boring Operation Required
51 178 70
92 610 70
76 e 254 21000
92 610 21000
394 178 70
686 305 70
610 279 21000
686 305 1000
21829 2 762 1000
22134 t5639 21000
0,203
0,201
258 NC Bar Lathe
0,201
0,204
0,202
0,258 NC Chuc Lathe
0,202
0,210
0,206
K
, 1c
* General machining cost centre for the spur gear
446
length and annual quantity) are satisfied. Hence with
reference to the spur gear the first cost centre that can
machine a component 396.0mm in diameter, 92.1cm in length
and with an annual quantity of 50 is 202.
Cost centres for machining the spur gear are:
Operations Cost Centre le
a-d 202
e-f 216
g 217
(4) For operations carried out on machine group 202 identify
the relevant predictor variables using Appendix IV. An
abstract from this appendix for cost centre 202 is shown
in Table 48.
(5) From the component drawing determine the values of predictor
variables:
(i) Turn (Tols. ! ýo. lmm)Trurned lengths = 92.1mm
Depths of cut = 5. omm
No. Turned Diameters =1
(ii)- (iii) Face (Tols. sO. lmm)
Flange Flange Boss Boss Side 1 Side 2 Side 1 Side 2 Total
FTool Travel Lengths 28.6 28.6 23.6 28.6 114.4
No. of Facing Ops. 11114
(iv) Bore (Tols. ! ýO. lmm). EBored Lengths = 92.1 nn . 57Bored Diameters = 88.9 mm 9o. of Bored Holes =1
N O N
W
I-
W
I- N O U
m: O LL..
t/) W I- O
O r-ý
9-+
Ln W C
W J co Q
0-4 ý ¢ ý
CO qe" W J m
F--
i
I
N
N
W
f- 1-ý
OW UU
N_
C
F-- Z
_C C U
Ci Z
Y U
U
z
r U Q
CJ
CC W
W C9
N N LJ <) C7 R
C. o o c C c-,
C. C, c: (: o
I
ý C. o
V9 N L. L1
O O ü C ? O C C C) O CJ U U ' C
ý.. _
t--) C
N 'L
i)
! -
I .
" Ell J
S _ O L- = _ ý
C O
Z-ý W N
p W
LLJ Q U W O O J O \ O O G O N
J W W C [t - - r? "
C: ) = O W
} ý, W
Vf
- O C O W L W O
C L J C) C) C7 O C
i. O
L p M ý ' _ . . W ý f_
N W J C . a(
}
N "ý" N (il .1 _
N S Z C)
W n
(1 co G ` .:: O p O C; r G ^ .ý ""
ý lV O w LAý
.. -
ý. ý C ) :
C) (. ) ý
r_- Ö
_
n . .. r co
W = C) _
- If - W - - W W x ý N .i ! - N
C7 W
1 N F-- U V)
N . N W p . ..
-J F" n J 7
W W W . 1. '-
W
J U
4 W } J
W } I. . ý -
ý LL. J C- laJ
4J N . t.. i
J J Ca \ - C7 - C
N ý _
-' J J M U C_) I-- W -
O 0 Q LA- Ö ta. Vj _ G 1-- -
W W W ° W ° O L ý( Z
- W
M S -
\ ý E
- m
° = = - Cý VI A /ý v //ý ý/
/ý
. 7- W T L ö O
_ Y J (_ ! -- U F- W
N -J J
Z- O K g
_1
G CC J a O U u . - '7 º-- v
f -- J . "! v v v v v _ " > J N W W !v O () !ý -"
PC - -
f: ' ý . 'J L.. )
O CC
Cr i W
-Z O
" 's F, W º- - Lr (
ý I - N M Q tf\ %D h N LT O ` :V h'1 C U1 %D r" t ` "- ý ý ý ý ý
( : n ý `) ý
W
I
ý ý
448
(6) Insert the coefficients of the independent variables (Table 49, Appendix V) in the relevant equations. An
extract from this appendix for cost centre 202 is shown
here for convenience.
Equation Code
i 20201 (0.113 x 92.1) + (0.041 x 5.0) -(2.24 x 1) = 8.37
20203 (0.058 x 114.4) - (0.18 x 4) = 5.92
20206 (0.049 x 92.1) + (0.037 x 88.9)-(2.09 x 1) = 5.71
Total = 20.00 mins.
TABLE 49 PREDICTOR VARIABLE COEFFICIENTS FOR COST CENTRE 202
111:
112: COST CENTRE 202 - CHUCKING COMBINATIONS
113:
114: 20201 0.113 0.041 -2.24 0.00 0.00 0.0
115: 20202 0.147 0.041 -5.77 0.013 0.0 0.0
116: 20203 0.058 -0.18 0.00 0.00 0.00 0.0
117: 20204 0.081 -0.68 0.00 0.00 0.00 0.0
118: 20205 0.032 0.065 -0.92 0.0 0.0 0.0
119: 20206 0.049 0.037 -2.09 0.0 0.0 0.0
120: 20207 0.095 0.032 -1.81 0.0 0.0 0.0
121: 20208 0.50 0.00 0.00 0.0 0.0 0.0
122: 20209 0.147 0.041 0.013 0.0 0.0 0.0
123: 20210 1.00 0.00 0.00 0.0 0.0 0.0
449
TABLE 49 (Cont'd)
124: 20211 0.00 0.0 0.0 0.0 0.0 0.0
125: 20212 0.50 0.00 0.00 0.00 0.00 0.00
126: 20213 0.095 0.037 -2.09 0.0 0.0 0.0
127: 20214 0.00 0.00 0.00 0.0 0.0 0.00
128: 20215 0.0 0.0 0.0 0.0 0.0 0.0
129: 20216 1.10 1.00 1.00 1.00 0.36 0.0
130: 20217 0.003 0.016 -0.10 0.0 0.0 1.2
131: 20218 0.002 -0.009 0.95 0.14 0.0 -0.88
132: 20219 14.08 0.00 0.00 0.0 0.0 0.49
(7) Multiply machining time by the relevant Material Factor
(Equation Code 20216, Table 49) to take into consideration
the machinability of the material. Use Table 30 to
determine appropriate Material Factor group.
The total machining time for the spur gear
= 20.00 x 1.1 = 22.00 mins.
(8) To estimate the "floor to floor" time (FTF) time, the time
required for machine set-up, loading/unloading components
and machine manipulation tasks must be calculated using
the relevant equations and predictor variable coefficients
from Tables 48 and 49.
450
(i) Load/Unload Machine
Equation Code 20217 Max. A Dimension = 396.0mm
Max. C Dimension 92.1mm
Total Ops =6
/
Load/Unload time. =
(0.003 x 396.0)+(0.016 x 92.1)-(0. l x 6)
+ 1.2 = 3.26 mins
(ii) Machine Manipulation
Equation Code 20218 Max. A Dimension = 396.0mm
Max. C Dimension = 92.1mm
Total Ops. =6
Machine Time = 22.00 mins
Machine Manipulation Time =
(0.002 x 396.0)-(0.009 x 92.1)+(0.95 x 6)
+ (0.14 'x 22.0) - 0.88 = 7.86 mins
(iii) Machine Set-up
Equation Code 20219 No. of main ops. =3
Batch Size = 20
Set-up time =
(14.08 x 3) + 0.49 =
20
Total Manual Task Time =
2.14 mins
13.26 mins
(9) The total FTF time for the general machining of the spur
gear is:
Machining Time = 22.00
Manual Task Time = 13.26
Total Machining Time = 35.26 mins
451
(10) To estimate the labour and overhead costs the total FTF
time must be multiplied by the appropriate labour and
overhead rates for cost centre 202. Labour and overhead
cost rates for all cost centres are listed in Table 50,
which is extracted from Appendix V and shown here for
convenience. e
(i) Labour Cost = 35.26 x 2.60 =£1.53 60
(ii) Manufacturing Overheads = 1.53 x 8.25 = £12.61
(iii) Administrative Overheads =
(1.54 + 12.61) x 0.2
Total Labour + Overhead Costs
=12.83
£16.97
TABLE 50 LABOUR AND OVERHEAD COST RATES (f's/Hr)
Cost Labour Manufacturing Administration
Centre Rate Overhead Rate Overhead Rate
248 2.55 6.70 0.20
236 1.30 8.25 0.20
238 2.60 7.00 0.20
239 2.60 8.25 0.20
201 2.60 8.25 0.20
202 2.60 8.25 0.20
203 2.60 8.25 0.20
204 2.60 8.25 0.20
452
TABLE 50 (Cont'd)
Cost Labour Manufacturing Administration
Centre Rate Overhead Rate Overhead Rate
206 2.60 8.25 0.20
209 2.60 8.25 0.20
210` 2.60 8.25 0.20
219 2.60 8.25 0.20
258 2.60 8.25 0.20
214 2.60 8.25 0.20
216 2.60 8.25 0.20
215 2.60 8.25 0.20
270 2.60 8.25 0.20
211 2.60 8.25 0.20
217 0.85 7.00 0.20
227 2.25 4.90 0.20
(11) Stages 1- 10 are then repeated to determine the labour
and overhead costs for the grinding and gear cutting
operations.
453
APPENDIX III
DESIGN COST ESTIMATING PROGRAM (DESCOSTIMATE):
OPERATION NOTES AND PROGRAM LISTING
454
A. 3.0 Introduction
The function of the program is to allow both the time. and cost of
manufacturing a component to be estimated. It carries out this function
by calculating the time and cost for each individual machining operation
performed on a component and building these into total floor-to-floor
times and costs. Machining costs are then determined using the
appropriate labour and overhead cost rates.
The manufacturing operations to be carried out on a component must
be analysed in the following order:
(i) Initially the total machining time per cost centre must
be calculated. Machining operations can be analysed in
any order.
(ii) The total machining time calculated must then be input as
a variable into the "Material Factor" equation to allow a
correction to be made for the type of material being
machined.
(iii) If Load/Unload and Machine Manipulation times/costs are
required these must be carried out after the total
machining times have been adjusted using the Material
Factor equations.
To enable the estimating equations to be quickly accessed they
are filed under their appropriate works cost centre number and process
type, i. e. Pre-Machining, General Machining, Special Machining and
455
Post Machining. The cost centre that the component is to be machined
under must be known to determine the input code that allows access to
the relevant estimating equations. However, General Machining operations
can be carried out on a range of cost centres. A sub-program is there-
fore available to select the correct input code if not already known.
Operation times and costs can be calculated for each operation
used to manufacture the component and for the total number of operations
used.
The information required to run the program is:
(i) The cost centre or machine type if known. The cost centre
identification forms part of the input code for retrieving
relevant estimating equations from the data file.
(ii) If the cost centre is unknown, i. e. for General Machining
only, the component diameter, length and number of components
machined per year.
(iii) The operations and costs required to be estimated.
(iv) The design variables for the relevant estimating equations
and the order in which values for these variables must be
entered into the program (See Appendix IV).
Many design variables are common to a range of estimating equations.
However to simplify variable identification a variable description sheet
is available in Appendix IV for each cost centre. To gain the full
456
benefit of the system (and reduce computer costs) a standard form has
been prepared (Figure 47) to allow the values for the relevant
variables to be recorded prior to using the program.
Unless indicated to the contrary all component dimensions are
measured in mm's. r
The equations for each machining cost centre can be used to calculate
total floor-to-floor times and costs and machine set-up times and costs.
Where the Load/Unload or Machine Manipulation times represent only a
small proportion of the total floor-to-floor time of the component,
separate equations have not been included, i. e. these times form part
of the equation for calculating machining times.
A. 3.1 Running Procedure
The system flow chart (Figure 65) and the following procedure
'form the operating instructions for the estimating of costs using the
'computer program "Descostimate". The example used to illustrate the
procedure is that of the component used to illustrate the manual
estimating method in Appendix II.
457
FIGURE 65 FLOW CHART FOR DESCOSTIMATE START
1
2
i
3
5 4
61 "I 9
1 7
22
8
21 15
23 16
END 17
ZO
11
T
12
13
18
19
20
1 I
14
BEST COPY
AVAILABLE
Variable print quality
458
ý ý O v
/
ý
.... ý G.
ý
.0 N
an (L)
. L]
E rO i ý O i. G.
a)
H
ý :ý I_`.:
- (r F- ý! F... Ea
F-_, L`: t 1. I. L.
ý. ..
1i1_"
.:,
r- ý ý 0 ý
ý", ýý, >, r- H C-
,. _ .-a :"H
i-" ,"r
'-. "'
r La cu f%
F-- F" v1 1ý C
Fýi J. ý
;^`.. ý
I _4 � r,..
.ý c 0
. I- 4-) CL 0
> N U
., _ 3
r
C)
0 a
O U
0
,a CJ
. r.. E b x C)
N C O
41
S.. C) C1 0
ý
L O
F-
41
Q. r-
0-4
N
459
WI
O
i
U
O 4-
N a)
-0 O U
N . '..
cý
0 U
a
. t-
+ý C)
"r H-"r
ý O N1
ýH v1 to
"r r -0
C C) co
C) P r G) G -P ý "r Üý
N t0 O
U N ý bO
Q� N N V1
"r C) ýU
O O C) a.
ý cn c
ý "r f0 L S.. ý Cn ý pU S- rt L7.4--
C) C s (o F- E
c 3 0 c
Y
I
Q-1
(.. -A 1ý! f 't
ý CL
-y 1: '11 17-11 _r { Sj 1 ;; {T"{ IXI cc t 17"1 lyt 1=t 1-t i=l --r ý_: uj
Ij 1 IJ"1 . -1 >_+ 1ýj 1: 'ýj 1: 'ýj 1: j r'ýj 1: 'ýj 1'ýj 1, -. j 1: 'ýj t: "ýj 1ýt 1ý1 1_1 1: ý 1ý1 1y1 1ý; 1ý1 1; { 1ý1 ,r týl ý*'ý 1r{
"y I{[ 11) f11 /ý"{ iý'{ I11 iT 07, CC w (i C. `j"- F- tij I- LLj P=W LLI 13 = i2_ =1! J=IJjLcký WI-1= 1-- L I- L 1= 1_1 1=1 = 1= H I=C; LL-4 ! 1-j L'eF-Jitj J 1_t _1
i_1 1-. 1= 1= Ckf 4G
1_' E': 11 L L: :L1. L. G 1Z 12 == 1'J W Cl '=' 1_t IT 1J 1-- +1i wl 1_J 1.. J {_1 _: _ 0')
xi i tD ('"- '? " ý. l} k, L'i im U 7' c '-f C" J ý, -" i +", q . --, ---r ý (""ý i=i i' l t: '"i i: '"1 i; '*i i'+! i: ""1 i: '1 ': '"1 ý; '"i t^i i.., } ID 171 i"yi Fri 0 iri IM U-5 i: J
rI LL
ý
U, tz HA c4 otHir: Jy
ij: ?W JIi J F-i F-/ Jt==LL'L; - L: 1: = LL CO tiä--:. CO ýCtoJ
ch
ý-' ý'- -+
'=1 {yj0
L_ I--
F ..
J. ýi
º_ <L
�4
I.. iý t: C. -t.. ý
L:
ý.. _ L.
J rFý--ý ý ý=
ýý- ýiý J
ý=t ^_ _I W
º-a'-''='
IX 1--IL
W
'-4 .... 1 U- 1"1 I ;
460
ý
0 4-)
E rts S-
S- CL
C)
a. a
N C) N
b U
'L7 C)
a ý ý N
-0
C 0
r i-3 N C)
Q'
C) L
O d-)
O
Cý
L C)
N C
Q
+-)
(A
a) i
C 0
N "r U C)
C "r C
"r to
C O U
1ý a,
C O
"r 4-) U C)
N ý
() C
"r 4--)
0 S..
.0 N
f0
O 4-)
C. E
"r)
aý
4-3
0 aý
0 4-) C) v 0 CJ
a .ý cý c
"r i
"r t V A
ý
r-
tu L aý 0 a, co aý
.ý ý 0 ý a a
aý
aý ý rö c aý
ý ý
a ý
S b L rn 0 i a
ý
ý N
. r.
V1 G) >1
v
>-
(4 . 1-
ý. a I -. r. . I--
1.: ýý
L: L.
Lý" L:
.
L. }-" LY: l. 1
Lt
a O u ý CL
.ý C) 4J
"- N 4.3 C OC CO O "r- a 4-A EU O () tJ V)
to 0 4-)
ON 4- C.
E
C) "r7 4-A (n C O () U t
4--) C)
ýE rC
Oi i-) Cl
0 () i i O.
C) (A .C C +) 0
",.. ý 4. ) C t0 rt)
C) p dý
0o C) V1 S- "e- 0 Eo
a 0 s. ca
ý 4- c ý aý
Lin to rl% 4: r
W
U .o U
E
0 4-)
CL E =3
. In 0 4-)
C .ý C . r. L U ý
i Es i S- cm (L)
O. C7
C) s-
a. i 4-
ýn C) C) 'fl v, o
rtý U a, "., D
II .0ý oý U
Gý 4J 41
ý "LýJ
0
cu t1,
+ý ¢ý O
(1) A tl C 41 tý ý
CI Cý
1
CC "r"I "r" OC', L "r 'r. a) Vi
C G) W 4-) N
(L) 0 S- o
4-
E ro L ý
co 1l N
I
461
ý C.
E b
O S.. C_
F--
S
_}- ! - F-I
I-
Ii
-) ýJ
J
=t
II
u p.
1-
L1J
-j
I rt: W ý L'i
H LJ
li..
ýJ
LLJ
}--
LL
C1
m "r
w C) V
C)
O- C)
E L O
4-
4-) V C) L i O tý
C)
i-ý
.,...
.ý
.0 N
E
RS i-ý
ý
4-3
fb ý
r
L
d
F-
CD r-
462
d
>,. ý ý
CD, r- ý
4
<- LD
w r
N 01
E
4J M C (U
d-3 C (U C O
ý O
U
ý-.
C) 4 C) E
4-) c (L) c 0 G E 0
U
M
N ý
ý
N .ý
4-1
E
1. Q1 O i a aý
}- W 5 LL
ýJ
u -
I-
1=4
u
L: S CL.
Li!
C.
tý tý r`ti
ui
tý cC tJ
a-ý C: ýt-
tý ý
a 1.11 F-f rr ". ý.
E
vn 0 L a aý
4-1
C3.
"r
Vf "r
0 "ý 4-) V1 a) ý
V1 .ý
4-J
O
a) V1
O CL V1
a) S"" a) ý
"r i C1. 0 i C1 C1. to
a) ý
a) 3
4J` r- N G O r- r-
N
fo r- CL ý
. '.. ý
ftif
ýr.
ý--ý
1=4 L:: ý-- r_": w -j Lu ij;
LL t=.
` ý i ý.
Ls ý=
M ý--
463
W
eo
OW s.. ýý CL CL
e
++ W A-3
C '-
ö !W C].. i -ý-- ý : F- 6
+ý LL
ý ¢ _ v
C J
W
O c. ý
ý .0 1=1
LO 4- W
cc `"' " o "r- o +' 4- W U
ß J=
W N
1: 4 1-1
LU i- tJ1 CWJ CL CL
= tý. J H
"n ci s- T LLi Oý tz ý
. OC
O iJ
ýV
Eý Weý.
tn 1-r C, W
t Loý h- W _tQý
ý... t-+ "p t-. i t! t, [
WW 2J F- L:
_
ýr ý
oý
O
a.. º U
N
O
ý
i
E
O i O.
F-
N
N O
U
t0
N
E . r.
O .ý 4J b
aý
O
Q C
. '.. 4J t0
r--
V r-t0
U
O V-
E (0 S- tr) O L
d
ý N
LL..
464
_I- w
H Ll. l
ýJ
. _.. ý
J
LL-
I'd lal CA
N ý ý
r
N "r
E ý i c7) O L a
ý
I---
W
I-
W
I- I-I., Iýi III
i_I '. ' CI
Hý f=
I1i (l
yW Cl CL ý Lt. W Wi. ý
LL W W Cl
t: t: W W= wi-
Li ý ID
W sii =W f- A
-r -r- .ý Fýt ý--
Lc) r
w 0 i a)
. [] E.
a rp a) ý°" sý +' L C
C r0 C) OýL
.ý 4-+ '. 4- (o N 1. N (A
OO C) +J C) L1 r .r CO r0
N 4-3
Oi i"C pý
ý 4- 0 -t()
4-) 10.. fö 4-) Nd
NN E C) OE C) iý CJ t O0 'a +j
4- V C) C) .0L. 01 COH 4-3
C) 0O +ý "r C) "r d.. º i +ý C)
O 00 OCNS () r Q' 0 C) {. t
W "r a0 C) +1 0
.C "" ýC u- 4J
r C) Cd C C. O rý
"r wOýN
+J t0 4- rO O
OrOOU CL
i cr C)
"r t0
w C) . Li
.0 4- O OEO . ýº
wýLN Cy) OC
NýO
E
(a S- bwL
Cl. ýn F-- O
t0 r-
O
N ý fö
r
N . '.. "a
c
a 4-J
E (ts S aI 0 S- CL
465
cu U
c aý
a) N Q)
.aC ýON C TJ "r" a)
Ll; (0 J3
ýE0 -r ;°
o a) s.
, i, "" c> LL Ü
+j (1) o
l t'_ i r- a) i3 L; S- OU HON3 "r" t: tf -, UNa"
(1) C) Qj L.
{: i{ .C (1) L. d F- _1 4-º -0 a)
RD 0ý lt')
WHC . r. a) C i:: _C
rt7 s :V "v 0 (A
iJ f-r i 4-) Oý i-ý ý a) rC iN C, LL ýO a) tA W l"'d
_ ýýý _ a) a) 0I a) S_ CN ýý O "r C ?W t1.., C "r 1-+ 1. - !- +ý "r fC
ID ): L V) " rt3 4-1 (i ) Cj )-ti a) 5C W Lt_ E
... "r ýÜ --I LL 'J Vi W CJ III dw I= Ld
L. ý V ý
Cc 0ý __- ý""- ý_
ý" ý W
CL
W CL WW
- LL if) rr 1: J '"1
ý , --
iQ ý >C
O C) "r-
0 w
0
J 4J
+ý V aý 4- N
O N "eý
aJ r- 3 ro O > .ý N C)
I- t0
Co r-
466
t- 92-1
c. ý
0 J LL
Wý C O V w
O
Vf T r0
ä N
.Q
C)
E r0 i ý O i a C) ý---
t! i
.ý ý
r"_ , iü
Iil i
...:, w
M-+
I-
ý=t f--r ! --
7-Li CL 17J
C1
r--
,; c, . -ý
ti t-
ý_}
4={
+_ +11 _ : G:
W ; =,
ý RQ
ct
ý.. J
_I Cl IW4
J
i
ý t t
ý
CL N
. r.. 'L7
E
i
O i O_
a)
ý
"a (1)
4-1 N 0 c. ý
Cn
"r a)
.0
(L) S-
44 X
4-
O
N ý
"r
w
ý O
ý
ýi- P-4
II Ih- rr}
r=r ýJ J CI rý F-
c b c 0
. 1" ý rq c
EE u m S- 41 aý V
4-3 N
0 u
0 .ý. +. ) m
ýa " i "ý d dw ov Qý
m >, , -
CL o "ý 4- tt N cu
4-J J3 N
ov- r- x aj "rý
om "*- cn ý ro rC N
Vf cr (Li W
Oi C
e y-. L. º., a
H
G
W C4
LL!
º-ý J
H i= N
F-
J
W -j 6-4 LL
ý F-
{-=i
. -, A J IJ LL
Lti W
F-
1N 1M al IN
Lt! i:: l IJ tý
467
rw
r-ý 1 eý d
S- a 'C7 Ol
N C) E
"r )N
C O
"r 4-) f0 L C) d 0
C)
to
. r. ý
n
O . r..
U C)
N
0
N
i..
4-J C) S_
C)
E
S-
O S_
C)
Fý
ý (t3
r--
V) -ý . II
a)
ý
E
O i Q.
a)
F-
v a1 +-A E
4-) an G1
ý ß
a1 ý fo
t0 r
O
4-J V a1
V1
.ý
C1 i-ý N
N .. C 4J 4-1 (A OO
-0 'fl U C=
V1 cp 'C C) al
'0 S. C) O :3
Ua C) C rt3 >
º-+ ,. J 0
r"- ý
t4
ý
i17
I. i. l
1 '! u. l Iý i
H w
iiý H
iý F- ý-ý W ic.. ý
V Li L
W 2 '-' ~ fit
; i. 1-:
LL. ILL,
I lI '
H
ý 2 fJ It
.ý
.ý
,_. it: w 7*
w Cl
w 2 t-
E--
J
W 7
W J-D
LW 2 F-
N C) N N C. 1
O
LL
C)
41 0
ý
C) -0
O
"r
M
O . r. i-3 V
N
0 41
N
ý Cl S-
c N
ý
E
ý O
Q.
N
F-
f
0 CV
4-) tic
a
ý
ý
"r
0 w 0
C{- H
a ý ý 0 u Cý
-0 c eb cý
ý
Cý
0
E 0 cý
m
ý C1 b
4-
ý E
0 . bi
.0 C) S..
0- CJ
N
a CL N
. r.
E
s. Cf O L a. aj
4-J
O . r.. U
OCC 4- " tC "r-
C) eC) " r- "r
+ý"ý At b "r C)
,a "` U #O vl C) ý
(A C) Q C). V) Nr
10 CA A C) wS. :.: i-) (1)
UONC C S- O C)
"-y ddS?
I I. -
r- -- "rC) "' T F- ` 1-4 -ý
.ý '_l 1. il - 1=L ! d--
.. L- . . 7. ý F=.
''t " 'r' +f I %1 0-) W 'Tf ! _- r r_n .. º-+ º-+
W
ý º-ý F-- ýJ ýT
CL.
W ýw
.ir .. J
W Cj .ý_W 'J W-+
W
LL i- LL. ý_r) º- ý ýJ
W F- iýi
U ;:; 1_ L++
J
F-
F-
W
F-
r- N
i .., .: i:: lJ
LU týa , -ý ýJ
I. j LL
ý w 1-4 t '± =ý r
r
ý. L. ý ýJ
_ _0- , "' "
ý :. "1
r.. , r. ý,
ý H
468
t- ý ¢ U 3 O J Li.
O LL! )
O C. ý
f
r-
O "r 4-) U
N
O 4-3
V1
S_
4-3 Q) i
s
i Cn O S- G
Q) ý ý
4-)
C.
"r
N . r.
. -" N GJ T
v
i' 4- ý
N N
-0
N "r
E co 5-
0 i L]..
ý
4-)
CL . r.. N
"r
rý O C
v
Z
. 4- ý--ý
M N
G)
tn N G)
S- o S- S- CL)
ýrt
N t_t b tZ
N Li i....
.a EW
p'f LLi O S- CL Cl.
C) ý t i-J
4-) 0-
'r' iý t/1 w .ý a CL aý
4J
ý 0
rts " w
4- " �--. "�_
469
A. 3.2 Program Listing
1T, IT'1 1. ýýý
f:. t CID
. -! r", J
ý1J
! il H J
W t-
, ll-
111
1 ý1
! jl t1J a
.. e
H crý 17-1 ýJ
ýJ J
H
ý ý H i
ýi I
f_i
i-f tJ. - F- cl GJ P--- ý
Iý h+! i_i C-: ýýt t j0 CID fiý
Cl f' 1ü ID
""1 -1 i1 Ls.!
f_: r- : -if ý. 1 -: - 'A ( F-
.. , _( _( . i::
{F- "" ii. F- r_, ! yý - ý-=" ý :ý F- ý: 'd JJJ iij i. Ll. W Ll. Wý ; it
I-A CC, ii iW0_ý ;ý. ý 0i
r I_ FýWý ýý W 4.4t
ýý ý '- D-- i_I IyI H ".
ýa I_t I= ti Iyt I-- ý_i aý 1 IJ t ý" ti" Iý '""
-. L'L LL' `=' '" "I t"'
H
r_l rJ r".
N ý-", ý li1
1ý IT, f- tTl
t_J 1 . 11 F-- --. '0
. 1-- rl ief I_t1J 0
ýij (". +ý.
ý. .., ý., _ . . c;, WW " JJ ý_ %
: ",
. -, 44
Ui 44
ý 4 ý
.ý
ý 174 Lt1 ý. L LL 1-j-, I
-j t- -6 -". ". ^. ti r". 0.. ý..
L'J W CL """ i- W J1= i i't JW. iyi I-- ýi, -º- :, ý ý-. W Wcj c; I- F- -º- W CO i CC' W L't I= - C. 0 '-- Ltl i'"_ "". . T". Li-. 1 u L1J cJ 1.3 WW ýy f= I.. _ : ýM^ JJ J LL L. t lJ CL ý5 ý F- .! ýt.; 1 '"ý L31 i"t ""- iý c, "". ""
1=4 C-1 F-. tt; L) ý"-+ WW I- J "y 1=4 J 'T wý ^" i= LL' lt: ý lL .
ýý r-a : i'-' LLJ LL
_:. U.
F- -'
WW ; Z4 _
1_1 I_ . Ltý CL - fl:: , 4" F- F"- F- F- 1- C". J 17,
i_i i_ý ýj W i_t '" '" -" " 'ý- iy "'-' ýL -' Ilý I^ I- F- I"- I- W r=tl ýti %. IIII11iL`IIWUW(A (-ii n
I-i (-.
-1
LL ý_, ýsln_. 1 r_1 I-- r I-= _I r F= ý F= º=a v) ýjC, LL W. =' i_} i ý} ý0 i- III i º"-r JiýiHJ i±'1 s" it %- " _. +-. "v " . rf.. .. r 0. L'1 it: W CL c, c. 0 Lt:, ! 1+ Cl I: r: ýy 11
_L'L' W LL LL LL LL LL r^ r 11
J 4=1 CL i. J I. i"I li.. iZ i_J I: J 1: ý4_. Lr r-+ I--. H n"' r-" ": r'i
, --I --ý ý". i
't ý: J LO
" "W
üü
L; t: ý" LJ ::: ý: rJ'1 IJ 1 F- .. 3 r_ýr AC4- F-F-ýwtiF-
I= It 0 (0 ".,:, 0.0 .0
. "ýJ1, ý ý. yltil
LLJ: " 4%j ... ý N Cl CA CA r--j
(") i_-, ý71 -+
_l i-J t-!
Iyý fri i-i '=i o_I Ci ýý 1ý1 C. ý"ý, IM 1=1 t=l ýLI i l, " i "3" L_j
( "",
_Lý ( . -
I. -. 1 T, ý^i .rf .ý , ',! 'ý IJ ý ý. lJ ý'
. ':. 1 ýT'ý ý :. i -"ý i" i, ý', 1'". 1 I'" C"i 1 0i ý., i .I,.. -.. _. _.., _.. , -ý ra . -+ IiI! 1iýýý ý'"5 I'ý 1I "'I ý '1 i' i Yý ýIý1Y : 1'
470
.
ý'ýý . _#- " CID
LL - Lt1
a, "
ý, " " 111 ý^
i Iý1 F-i 1_f iý _
F- I-- ls T , "' yW
W IJ I--
LL
1ý1 111 IJ Lý ::
--- ýiý ý- W º-r i
--j +. CY: 1.1) ' -' + F- l'L"' 'J
t= t_t
-- t- + LtJ
ý tý' ., + , ".
ýý =4 `' =
LL: + it. '.. 4-- ýii : -". F- , +. iJ W ='i Cl 1= . y- CL ýýý . -" ,
L1: U., Cl F- + -r- - tu Ld -0 r- Va -s- jy ý- .
IT- T F- 4. t*t .
*. ,- LL '- t% 11. i r- _ '_' HC;
'` 'ý' ý: " iJ ý. ý_ý Q) t. 0 ll. l 14 t_i 1_: J = LL J "
-` '" -"J' }--
"J `-' _ 'ý ý"ý lL' --
ýýý : ý- y
tµ+ "ý1
F- L ". Ld _J
#ý+ P-- # ý: _: '_# 1ý #_J -" iý F' .. ý Il
LLJ Lt: 0.. "i #=I 1ýj f
_} 1". 1. # 1... 1 f_. # 'ý r. _.
tt: ýý, " - 'J +-I LL i #_# t i. +_+ F- r t'- 1--, ý. IZ ... . 'ý
1:.. - 1-, " =0 r" LL '-" F,.. -. LLJ l1J -" -+1,1 1 1-+
ý. IrI _" týl IT"# v 1ý1 '"ý Ir'I z F-" r", J Cu ýý ý f, "wr.,
1 -" t-. '-" 1=1 l. Lý J: K 1'T" !
_i" i. iý IT1 FM iJ ! --4 in _-q ". +wx".
."
#L #ý # ly Iyl I-C 1y+ ýtI ý# 'Z +ýI F-- ID 1y', 1 ",. e -º =F.. {. '. hÜ " yl =-r "': # _- -r 1ý: _- '7' 1_ #-# 1. ýI
_. Il ! 't I: Li + F- -+ ""e-#-. +-+
i-I 1ý) -. ^# *7" ( ""#
'. ri i '1 -. i j1 F- F- . "-". . ý" .., I`, .r1ý-#. '! I" i '. #11, I, ", 1, -.
ý ,ý
I 1y I. 'r1+. '# r1 11 #ý --#
W ++- I .-.,.. 1 ."". : ._ý. _ LLI _. ., ý " "".: "ýý". : L{ .... ý '-' i. ý'""" Li, J
t! ý ýlu CD Lu 1' LL LJ # -1 1-' . 171 1,1 W"J LLJ ^ 'ý.
_r '_
rý F- i- -7 r- -- _I
F- F-- 1-"" iý 1"1 I- F- F- 1 t't ý" ýf... ý. _
F-- Jº"J #--" F-- F- 1
L: #t L! '.. ; _. "'
ý~I I .L Lýý C; i'I iýI '"i. CL C, L. I. E.
1_
F- ý" ! _ý1
+' 1ý1 1ý# I
_I , ý1
. .T ! ý! _.
! +. ý. #_.
_.. 1ý #-.. I
... 1 ýI [- ý F-r ,. "J .!. 1I I'1 + LL 11 '' 'y
"-" "ý ii ýi .
17i 1=: ID # ; "'1 1^, "+ý'. 17J -4
T
#ý # ý" IT, 1,!::,
i1I -5 .J ''"., ht # 's L: ' ý! ý 'ii 'ý- I.. I I.. 1 ý"
ý"! . I, "
J i. '! ""!
'"ý :II *-" ! i, J " r1' ý. ., '" ý# "
I; _I _" I _I
I ý". #ý,: 17D 1 -1 1". J 17: . ý# . +, 1'ýI #ý! ý1 11;: 1ý ý" 1ýr. 1 7_ ý ý% I. il , ̂ +'.... J Lý, , ý! #ý IýI "ý, lyl 1ý) 1"", 1.. I ý-ý
k 1,.! .+.:. I #7, #Ä --a '. { #`- 'T .Jt., r_"! '! "t "t+.! tj #1 ý iTy 1""", 1 ý 1- '. -.
.r ý#" .t-, - -t t: h#:! ,
it tl'1 i" . It '#" 1 I.. # 1 ! 11 .:
', '. 1'1': '. I, I 1"". ("". T"
471
rsj .
, -,
CI I= {A iIJ
-S oi_i )
r
ý Lý
1ý" ý
't
. ý'.. LL T4
t- W [ýJ t1 CL
.. wC, ýr, = w
1=1 F- _J-J =F-W1J
. F-- ' "..
J C-. 1 - ý_ ý
w
1=1 I: -, i LL
" ý-ry
.ýýJi {"ýJ
f_f-t t= t: "ý1 I- ? 7'F- ý, - ý1 r f! tý ýý
LL t. "ºt t= tý
W Lt = lrJ
JWt=ý"- fJ JH
W r: fý! iJ. -
t= rý i fý t= J t: 'J
f"'ý
li , tr, r. 4 ". .
ý, iý! _ C. ' N
ü 1 .1 tr/
ý. t " rrs
AH f=tiif'"-W Cr- 0 i"'. j i. t_
t
- N. r. ". 1 ý_ý W
, -ý iJ ýýý '""_ C Li1
ý ý, jiiý", J= tý . ýý W F--
I'D im t". j {A Z: .4-. F- W ý', jý-,. -+'"". _iýi Im ü_ Co '_
C, n Cl F-f ý, -"ý} ".
J r"" = J _J U. CE: 1""M =r o-"". '_ . -. ý., VLLW
Cl F--
". "-+ LL
Li 1.1 LSi ý. iý J '"
k1; CA j""_ Ci ;n ti , ý: ti1 ü1
to -" rJ W -. C'J 0ý. =
; ý; ý.. -. ý.. F- ý ijiý. CýhT-+J W=1 J -"+ =-"fjJüi'J-º- F- CL
F=H W0 1=1
º_, H
E--ifiCrJ
WW Ui J
1: 74 CA t ý" ý'"ý "1 \
IZ"i 1,171 LLJ 1 ý J N
i_i= -"
ti ti +.
":;. o .
j"ý "JJ ".!
ýý1":
{S "% : %; {rJ "11
+
+ +
_. L_ ; #, + F- .
f. l" 1-.. # CL T '- ._+0 F- . A. # I 0-1, *- # 1°- jiy
i_l i. -r) CA +J t: t: , .: * 1J
of co r, ;, P, - 0 0 U-LL -0- F-
_ý {-- +. " W
=I ii}CO # lL
'+ i-º-+i-f-# ý "ýf 'tt i ý- -*
[t: F-
rý
W A
WW "WW=i- 0 4-4 . '_ ý'" = 4" (M F.. tiJ5J0 #
ý I. J CL CL
ý#- =iý. rJ.... ýJýJý CL . º-t1_§ = LU LJ s-
ý#- io lk: - LL ý-+ Cl ==
-I. - ID =- :Z.... F- i- -4- I. F-1 #
LX, + H La! Lý+IJIJt
L2: 1. L = LL LL + ý. t L11
ýý=sra? 21=Lt1F-+ " ý 111 1ý 0 ý` z rJl 4- I.. L'. F- �W IJ 1--+ V+ 1= ý`". 1= C. L' 174 1-- IJ += °ti Jýý =1: - I: i ff Wr_rr=r=. __tJ". -e rr, -2 rJ º r= S# tC
F-1--" W F- }- + {_
I-re JrJZ2-+i-I-+r, W '_ -'-
I- : +: tz (-f F-}- WW irWZrr0 ('i22+ rI 1=4 W F-- i=+ = F- r_r +. ti +2 i_r j r_r ff2ff# º-"r f v", -"vº-ýtýv+I- '`"". F-F-F--F-2i-I-+=J fJ r= r= r= t_t r= r= +O CL CL T2r'r= r_r2LLl. t: L2. ýlt', l"i -+ 1-4 ý-- r~ tJ LJ t_r Cl rJ +J iiýf LL LL LL '" LL LL#"U1 . }; 1 ": ý%: =ic
rý W rýr r' ýj rýr + U'i r. lr f'- rrr r iý + rrA ci --r r^r 8±
"S' 7 '7 iý
u ýý . f-.. L ýý
+{'ýJ A
fJJ CL ýZ 0
LLý
T
rrJ ý CIO rr
". ý ýJ W
W ýý
º=1 W
" F-'"Ia Iý ý= rr w 11 1=i
1: 1.
=-- 1- ü J;; .J F- ý ". "= . -'.. LO ..: - r-
.ý IJ O_ 10 Q w -6 ;i,, iCL W'_ 2ii`%. : :-W
Hý ý
ü: V CJ LD $U O %, r.
CU C+J f_I F- "" F- UO ý"ý ýº tUUi_1 ctWW + CJýJ ý"" L"'rSýýý'ý WW W Or . ti1. Q.
ýý
w ý rn
G ': fj
F- _. 3 .= CZ 'º- -.. üIzWWF-1- ý=ýti.: 11 ctº-+ 1- W^... ý=f_ýýývvv
00 : º-LLLLLL 1J.. Jf, J4-a. +Hb-4
ý ýJ
11
ýi
G
tt: ü
ý ,., ý ýý "-ý Crt
:L, + rý", s+'. rJJ [t: l:.
lJ! tL f- '- i+J 0ü ý--'lnnü
+ . ̀. ý. 11
tr7ck: WvF-lr1 F-Ar, C+J
tt: WNC
= tt: :L
in G 11') c ý-i t1 ýi ý ý, ij" ý, .r . "-.. -ý .ý
4
CI Cl 0 IM l. r/ 1--I as IM IM tn fJ l=
lý Irl Iý IýI 1.. 'al Im in in I-_I 1ý 1ý1 IýI Irl lyl tJ1 I:: 1 Irl Cl trj 1.. ^. I Im l=- 4=1 Ia -. 4 1"iJ 1"'1 ý (JýI l"L (`ý i-. l lil 1y1 ý.. y Iý"1 lyl . -ý I-'. j 1, ' Iý(iJ 1. i1 (""ý I. tJ IT', Iýt +-/ lý, j 1? 1 In t ý1 IýI I, rJ Iy1 1ý1 111 1-1 Ir1 1;. 1 . -. ý. ("` 111
I I"I 1 IJ co 1 Ij cc. .-1I1 c- I: i ýiIý1.1 1? j Cr, "IT, T. ITý ITr ITl lit IiI IT. 1 . ý+ ... ý.. -.. ý. ... ý. ý. . ý. ,., / ý� ý. l". _ t"". f""- -
472
1: 11 i: '. l
iJ F-
IJ i_}
a
I
-,
ý
r. --,
! 'J
º-, ., 0W A 11 F-. = 0 _. J
F- U_r: tiýýLL
W 7- '
"-, ".. H týt
r_t ia CL CA t=t ytJ iit
" F"- t; ", 1 4: -4
ý.. - "
ýä tit t_t ý ts. t t_t
"0 iy W -4
ýi ""1 . "" tý
ý" ww -j "
ý--' c. º. 1 CO
I
-"1=1 JLY. --. r r10 = --#. ILýi +. LL. %"" ilý _=' " 1- ýJiri I= -'WiJ º-ti 0=-J -- . _I
1-1
Cri
t=titjt=t t7 0 : 1- " i- IF- :. ý;
N0ýCl 80 yy i_i Cl f0 I= tT,. r. . r. . -%
.. CO ýýj ýJ ljj . -ý . -r am am rr rti 17,1 CC'
"" "--1 U-) i w t--;; i "J i'"1
P- 17'1 -4 (-ýi)
F- . (I'
r_, CZ
UJ CL
Ui-º- ,., 1-
r
WI-- U'i
º-4O ý-u Cl.. CL
1'". j Oj iJ 4J
.v
ýW ."". " . IJIýiIýiIJ i ". i: WWU. i W 's }'. "" V \-.. l". . -s
" h-; I'". f_! W15a F-1 ~- ~'
jW 11 -- ý _l CL 1- ý 11 0 `- _r -. t-. '. il , t= _t- 11 O1 U i_t'1S W
ý, "-, 'ýJ V
: 11 LLl CL LLý_ Ni=; H
ss ., =ýý W'ý ifssJ s_c s=i s_s Cl s .ý
WF-C'lF-ýJýJF-= . '-. _. _ - *-. 4 IL J -". 11 J rý_ýCt: : W'= ty, 0 r-_: LL LL LL L.
`ý`ý ýJ t=+ ý Cý ýJ r+ H rý r. i 1. -. r-+ º-+
Uý 1""t ý: J CA
r"1ý.; -rUiý. ºýr"rrý: Týýý--+, _", ýý"ý-r --. 1"'-
ýr.,. r. ýr. ý. ý_. ý-"... -ý-r
, . -4 '. / J. ºJ
LL. LL. b- r-.
ý '-ý ' ý?. "J ', "-ýt", ýý I I""ýý: ýý, r�.,. I{'}I, ': I II 'ý
-4-
++ W F- W =ýJ! +
F- f_t " º-+ýJ IL
-. r" lG
"W W ý"
ý .v
CL. C=- -_ W o 1J r '' I. L! LL r.
il 11 ol 11 11 I'} F-- W t- W 1- 11
1- rý i-- M- r, _ ýr"__ý. ý,, . ý, f..
171 Oo .J r_t J U. I. 1, W r_s r_j I.; J rJ I:. j º- r_) º-+ - L. t
4717, "1 C7f 0 9ý"1 {'ý1 '. ''ýl ý. +1
+
-: -ý. +
LLi
. }. _
y. ý... * ci: t ýJ
LL
Ll. )
tý + fý
W
LK -+§J
t
LLJ
+
, S. U-
C4 -\ �-4
C) W P-1 .. S. /1 F- ý LL ow "-+ º-+ l.: rý1 i. e 0
F- - F- .ý ýý
(j) = Wºr W Wi-j F- 4-i
I"L U-11 r. D C.
= CL 1=0 F- L CL ýJiW
IJLL ýJ ýJ U
ýý tY ' W º-ý W LL
.ý 1-- W WL iai ýJ -, Lý
ýt 1
F-}º-ý1ý)
-J- -0- O. --e lu; dw rn i -- Wý w+-a i-- +w i= + I_ F- -.: ºr tJýZa. [�
+ f_ F- + :; - ýL ý+W
iJ + lLý
+ F-
+ H+
+ý- + ýt
+ (, L
+1ý
ý. jl'_if i
w H
r, J
rýýº..
º.. W lX.
I- U) 0 rJ
,= U_' ltl wT CL ý
LL r_I
...
.:.
1ý1
1.1)
tz u W 1-. + I: i1 ýJ .^ ýJ
-W
WH r. ýy
h ý. \" i1 F- Lj. I 4ºi 2
1-- ý" .` F-- t = : - LL ý.: ý
'_i h- - 1=ý IZJ W {-1 Y
%J r{'ý
~ ý_ W CL LL L--, 1J H 1. J
,1 �J h- _J ýr J =SJ L LL
t ti VLL "5i "7: i "%ý
V! ý1: ý ý:! ýý"!
ýýý -- utua . 4r
tJ (ýt
": '. t1 i+t --
'ý t. Lt f' . "1 -T -" T -t -ý_. ý... ý...
O, iý , ý; ) in ýý In- Ir/ 1M 17'1
r" . a; . -. C1 . _- T U-.. ty,. f".
--" . -d w. +_. _I -Q -4 _1 - -. 4
C I;, I'D t: J # 6r/ rr
-r 't _.. ... r-
473
e
w Fý
ý F- rý) Co
u t. L.. 1J
ý F-
-j v Li 1-
iý F-
w -.. ý (J
ý�_ ~J
T1
LLI i_ U i_ I=
F-
-. i'1 -ý , �v
. d- =-
CL ýZ
t-" F- CL 1_ Ci
W -- H
, i-- - -: y
AL W L. t. - '_ c: r ij ý tJ tC F- F-- - it: l_ c' i_ W '= 4=1 F- -- ti. -, -.
! 1? -"W LiJ LL
ý-+ '- 'L" := F- F-- LL I. L J
iýCL. i Q. ir: CL ==N µ+WCc EE rii
"ý:
Ci t. 3 1.71) ti 1_ j. Fµi 1=)
.. _. CA -, r, o-y *t =y ."
LLt ..., 1
CL ý ýJ ý=. J
.0
L=i
li
ttil V ý
rý _} .., ^
"tü a"
"ý W 11 J
1 -4-
r. ý---
v -. j i--' . e= A -o, -
ce- r. -" + º-" /1 V I: D ý .. ºý 2ýJ CA tr J",
vT". p7 -"" _r p -r ffj
ý_ :. ý . ý. " -. ýýn. ý . -"`ý -- iyý ý= U1 ý' W" fJ 1=0 O :L U')
'. .` ý. LH r"-- W= =) . rty : l7 ýJ I1ý H '-'l .ý "^ ==. -=W ý-. In =' l.
i
fJ = U7 r". U"! w ryl -0 º-q J-Jr U-) 1: 3 1--; p U7 F-- -. * -6 ii: --* --ý U7 C+l ý: "i J 11 J" CO) W -N Co C. ü . Ct', -. ý- '"ri +'"J: º'-' "" al 11 i*. l % .4 It 11 =. 1= V O'. JD C. l 11 1"0
WJ'^-eUä '"ý'_ -"' fJl --4 "- C 1'71 CA .=. CA 17'1 """ L. 2: .0=. %-* CC v (D W "ý W 11 v 11 º-+ 'J 11 '1 11 F-" , t- as '.. e 2 '-+ Wý i- LU WJ1-E=1F-JWt=tWJF-W1-1WJýUiA I. - I_1-. I=ý-, JiL'= = 1- ": '. = F-.:. 2 =W LL` W !: L 2V º-y º--' '-0 O 'J º-. - (j'1 0W LL tý a CL` O Ui o: º-+iC= iY- -2' r) wLL J_i1'wLL J... 1JAi. C1-+Q ü : f. 'j : A:.?. lJ)
. ý.
ý tý ý F- W ý
t'-rl trj Irº at {tJ U -j 0 Irf . 1: '"1 1. ''J {''1 C''1 U) IJ 1 -' r"' ý. ý U
1ý^, 1 1rJ 1ý1 Iri 1ý1 lyl Irl lyl IrJ 1r1 1J 1ri 1r1 trt IrJ 1m 0, Ir1 IrI IrJ 0 Irl 1M 1ý 1ý 1ý iý 1ý1 1ý 1^J IrJ Irt trl "C' 07-1 IM ý.. t. '"! 1-11 'r U-. 1 I, lj r.. 'L' 1T'1 1=1 "'t i'"J lrl -T U1 k11 i'- $'I) '-To Ir< .. + 1'", 1
1Iýt uI o; U-I U- 1 o-I UI U-1 1. L1 1. t1 1. L k i, ti1 1i1 1i1 i1 1 I. tº 1 ý;
I: . I'7' Ui U r1 If1IJýt; r
ru U- tt r"_ rý rý rýy r". r". r". r. r.. - ýý ýý r+ ýý+N ý. -, rýý ýr _.. ý/ --". -. ýý ý ýý ýýý ý ýý ý ýý ý. ýý ý ýýý
LL, r-1-4
! yý ft..
Iýiý 1 fjj! 'ti
L, i 11yt in = I'"J
I
. ý.
ff) 1' -
º_º ", 1 W LL -" i_º W CL : Im = ºi
474
* * ±
* * *
* ý- -.
*11:
'd! ' ýý
* : ¢. :. _, I i11 Ch: F- -" " -ý -. LL " 1JJ J --
iW LtJ ý F- w
w
. c_ " 1=1 " z7.
Hrl W-- ." ý.: a H i_i t
; ; r= t-ý WW #iL º-:. wZg. ý , -" F- LL'
1 .0 All. H "ti F-' F- }-a ý_- +Mr C! ý CI I_t "`
, r: F- ufý iJ !!! : -: i: " WAr I W!
Z+W 1= 's IW-4. -
W cr.: it L) I_' J) -. : b: " F- W0I. lJ '-. CI ý r=A. it. LWQ: W F-*=}1 =i 4"=4 #"
4 6. cI WCl -- ; ý"
F-- ý =I IZ -. #
'r ýW '-' L. L. F- LL1 I-- '.. CI -9k"
. -_ -ý- Z CL _1
-1 r-i ý fl. "? I: ' F- 1D rl CL CL Tý
"! }; " 1_J 4- Fj-1 '-' i i. I . -Cl
L I-- _
t--1 ", 1_'4 _. . #. ilk.
-d! - u_ '"rL-WW'. 1: ' iis_T_
: +F- "-" " I--= F- F- Ct F- '" "i: ý b º- LLJ `' =dý"
ý: ., F- .
L1 ". F- .
L'' =ý: " . ý:. ti :H ="' ý ""J W ., ý. ., ý 'ý. " ý I-- I- I-'_' F-- ý
F- F- ý -'y°r" Cc � rT L r= t' I T' :_". ý'f. - : ýI'LI ICFýI. _=ýi" 34 Lt: L.:: 1_! L 1. = I. _ t2- h "'p" hI ' ý"; 1ý1 Iýt 1=1 !. L CzI 1_I ý ll. LL. '" Ll.. " LL. U. ýf1: -
#: .r ii "ii ý: '
"ý#. laý, l '
ii~i iý1 }ý .'
.. ý" '. fl r. "_ IT, IT"I r". 1
HjF" 1 ±, '1 i :l'. +rl C1 Cl
.. r. I) .1
1J~1 1.. 1'ý"' fj
t- IJ
_I-
L. ý ý
t_t
fl: L: J
J
F--
H
IJ CL LI
t_I
J I_I
I_I i-
IýI ~ýJ
W 1-# I`I lýt Iýi : '".
- I'y!
F- A W +_+ i- F- +_i ýi fi+CL
ý" ""-
L f=i i7 t) J'"",
+". Ltl_i Q +1+ F-t t-. "i.. º-"i i iý, j ! '". '-5CL1- tJ' '
-. i= += QW *-+ W
_S -. F=; '"1 += F- ýW +J F- -_-.
WJ ý. JJ'= ý-º-''_+
W i-e r". +_+ += iJ i} i LO U C). '" JW', r" L"G .. -,, ý_'ý 7-- CY: LL) L0 iJ t- r-" Cr- LL '"". *-+
ww ý= i- Oil: ýiH^'ý_týffjrJA º-+ '-ý ý f^f ý f-- ü F- }- f L: '-" lL f-f
f=f1J1-ýr_t .
f C. ý fýf ; 'ý fif t= -_ý _ý i_f f=. f =t 0W
J lt:. i= W
ý+ W ý--
ýý
i"M .- . -". . -. 11 *-+ ý' J *-+ E-- .
r_"ý ý F- i_ týt
i-- L4
IPS
aa CIi L: J WW cl T i_l . r". .. co I. j; l r. LL W r"- W W-?, "-" 1't ... 0 I::. LL 03 ry LL Hi". i i! V', . ̂ I ITy I=I
i_! W I. - IT", 1.. _ 11 CT , I"". i 17"f :_ii il , -I "".: ºýI º-ý ýý_, F- .=Z: 1=' IyJ _I 1--1 a h'-'- tl., t.!
ý F.. , _I I1. LL
1 .I ý-., li ºLI r-+ F"-. LL .:
tJý_II. I_ f ,: l
al -4
I -j +_ i ý-l
!JjI:, iri iJ ý 1äI I^I Iit yi 0 IM 1 yi , yI 1M, 1_I , yi Iý , yi 1ý1 C z; 17J t =l On. ; 7:. I-, j 17,, i 1=1 I'D I'. i 1' ý Lj 1. ýý1 ýT'ý , "ý". 1 .rY1.1 iý 1 -ý" ý7/. j1 f'"- i ýýI ýy 1ý ýr 1_
:ý1 i'1 1 I. I. Ii1
+ -fi iJ ~_ r" , t. ... rr ý' J: Pt
1111 1~I 1 : _! I I"1 1"I .II "{ i 1,1 1 III? iJ I�hI
IT., IS, 171 r"1 -1 Iý1 :;; 1 la 1 1ü LiJ , ýll fýI
ýý . ý, ... ý ... f ýI J -.. 1 «I ý, «1 . «. 1 ýý J ýI ýý ... 1 -. i ýJ «. 1 ý"1 .. ý. 11ýi1 1i I ". i I'". i I"1II
W
4
W
ýý ". ý_ý W LtJ =S F- F-
I t- ü
Lt: " E-ý+ LL " Lti " ý=+ '_
W }- ý_ý Ct = cc . ý
(L -H1, '"1 W ti = --4
LI IJ I. J
HI -I-
iyI
,r c", J . -4
T
1-1
_1 a, 1=i C, cl
t=i . 1J i ;, i _t
W
}--'_ 1- _ _º- W T -, - p.. ý- . ̀- --
.., -1
IM '"' i, y1 i:: 1
fy, i ýI .. d W i" i"f iý 1"1 i
475
d
'it i It H
F-i
C: -
LtJ
Lý
J ý I.,
ýýI
I. ýJI. "+1
'DC) F- !-
, _l rr CI F-
i_II_1 Iýi 1ý1 1ý{
1Vý "1-
... s1 WW i iý ý ýL e. ir'"
.. U'1 1'11 17. ') . ti) 1J W ý'"- co Ci'ý'1 LJ -. + Hli Uýt f, '! C-t
ý i_t ' ! '- ;ft: 1 '" i iJ _ tT"t IT',
t tý !) t-i +_t ý"
t_t i't ""f t_t NL'J_lA
. LL ; -ý, ý ,Ti; ;-" ; _. (ý. H ý_;
-+º-+ LL J Lt +_; LL 1-: 1 +-" LL t_;
p ýJ
U-) f^i
: -". T., b-1 LO ý. L C. ) '"t 0 '.. U"a 11 (' 1 Cr., '. Y" ý
11 CI W F--
T
F1Hý1ýI
LL J LU iJ
fJ ýý ýt
lti ýä '. t-j +-1 1; ri ;r
1- 1ý1 1ý 1ý1 ýJ ýJ f_1
"'ý". . ̂ . , r. 1ý1 1ý1 I+j'1 1ji I. i) 1ý1 1IJ 1., L1 1}ýi ! 7' Iýrl 1; "j 1, 'ýý ' y. ý t. "ýý
. WW ý s.., . r:
. /'. " . ý+ L t: U 'j 0 Lt: 1ý1 ro 1. J i: t: "'J --ý H 'T I+I i- ý) ti -i Irl I- I. p C. - i_I ý"_ I1I, .j i_ý .o I1 j13 '+ Iý i! I
'"ý ý 17) tjý ij F"' i-J iýrý"il ý-+ ý=1 iJý '_II ý-+ . ý4 WJAWJI-- AWJ ., 1 %. H -.. -ý-
+:. +. L=1 ''-"'. ý. ý=ý V.. - -- 11 LL " M-A = I^I LL Ir ýý
º-i JW i_, º-+ Lt_ JW i_I I-. ;. 1. J'slj
U
, -, ! jä U"o IT h ,;.: l
u
jl i"'1 1=1 1=1 1=1 r=: 1=1 I=I 1=1 IM Iy1 Iy1 i=l Ijl ly, '1 1�"ý', `I 1ý1 Iý Irj Iý1 1ý1 I ýI 1; y1 Iyi 1ýý, 1 1ý 1r1 +, '. I rý IýI -r LO I". Ü r'"- 1_r ri, r-- -! -. j ý'''+ "1" fJ'I I"JJ ý'".. 1ýI"1 ITy {^I , --r I'". J I''! 't' U: I
. -ý r.. l --+ ---r r"I . -, }"- ;I f+: ''J I, "f r'1( 1; 4 j 1; '" (1jI; ': j I"^. j L' :I 1"ý'r 1: ' "1 1 'I 1'ý 1 I'ý'I i; ý! I. ''. f i: ''i I. i. -r ý" . ý. ''II""Ji. (I-, (i"ýý 1i ii,;: -fi. 1i''11''ýII'JI`''ii''fI'''II''"Ji''(I'ýii. fi". I1''. (i". ji". fý'': (I"': fI'eIýýII'"I';
476
/
lrj ci ', ý ý_ý.. ý r'"- 'jj 'Ty . -, .. -1
'.. ýý 171 'J
r, F- i-- I-- E--
, _, I_,
- '_, ''- "+-". ; -. 'rj ID Cl cl , y, 0 IM, i.: ' '=, , f, ,;, i'J l'", a ,. i, 'f"'
,: j tiý ýý trJ ': j ':. J
. 1, '": J .....
wwwwwww ,yz
Wý ^"" Jý "-" iý '' .
Yý ýý "
.. IT, . '1" {", 1 j""_ 1r) " i=j Co (r§ , -4 . kLi 't iyt ýýÜ 'ý- ýýý Lý ýjs i"; J ti+ý i: +; ý iýi ll= ý: '"ý li 1 ýýý I`ý ý"1! f'- iiý ý CtJ ýý ýýý lk: iýi ý: '"1 '-ý
*-+ i i" _ . --" i^i . - ý- ý {- { "_ (it 1-- CI ý 1- +. + -T, C Fr IM .4;: r I--- , -"e 1. '71 1'", J F- ý. ,j ý"11 i ,J I. ', J i_i , -"4 111. ": j i_i". .tIIt1j! i ('J 1ý-) '% I"L ü C, 1 (J i, Jit". fri II ý: ,J 1- IJy ý t--" ". ry , -+ ý F- CT., -" ý F- 17'1 - --: I"- 17"1 *"^ .ý I- 17"1 '. ", 1
ýýi i iý"1 IIýýiiiij11 +-4 i=t i} j'".., 11ý 1i io-a Hci ij., if F-+ ii1.0', r" 11 +-+ i_t o) "'" 11 F. "ý ý't ý--yAW_! h"ý-_:: LU_.. 11-AWJI-GttlJ1-::: 1_"aWJF-`". 'NWJI-: 4 1=4 WJ1-
.. i.. -. T '"1 +... H' =' .. ý +ý.. ý` ."+. ý3 -i ^"": ' +.. i. ý '"_' .. ý r:. i--ý ''ý "+_ .ýT i_i LL º-" ý ! _i LL L. i- LL rý º-+ ý i=i LL ý+ "-ý ,, _ #_I LL
:J '-+ I-, JW i_, - -1 LU '_J º-+ LL JW i_I ý-+ LL J L11 1:: I 1""-+ LL "J WI-J F-+ L1. J LU CiH LL JW
Ui ,...
ýý ý'; Iji kiJ ýý
C
r"_ i i: ý 'Ti -
IM IM In 17-I lyi lyl ý, ýý Iýi Irl In I=I I=I 1=1 i'm I=1 I:: t i=i !m : -j !, 'Z ':. -i I. ýjl ! rJ I= I: j Iwt I=, Iý1 IM Il i ýI I, "ýýI Iý ! ýJ !:. I I1 -f' U-1 I. LI r- ii1ý I'y 1ý1 .. r I: J I: 'J 'r" U. I --. 3 ('". I i; l IT"1 I,.. / ' ... r I. J ", 1'ý'1 .I (".
- IT', +-+ i". J i'. I ?U:! I"LI ; ""ý 'Lýý1 11 'j- .t -t 'j" 'r i' 'ý Ij} IJ :I Ur' U-1 U""1 U-1 Lil U 1J"'I 6-. 1 ! "jj i"ýI ti) 1.1--l 1" LI '" LI '. il
. fl !. G LLj ("ý... j_ ('"- ('". j ". I'"- (' (.
i, Ji""II. I1"'": Ji", I{", Ii, j1`. Ii, Ii", ji""j1'". j! "11! '"Ii. li". I1, '"")1'j yI'"111', Ii. If. 1i", I! '", I1'. II`"ji. Ii, )1'"")! ')i. {
477
I '1 -1
cl r-- i_[
W
. ? ", t:: N F-- ý. '" 4 i_i "". F- Cr C') ""_ ...: 1=4
ýý ý !! _
Ir, l
-i
r4
!
ýyý ý. ýý ifý iýýi . -, l; i , -+
i--
rý rýt
Cr. ' rý
o! -i +-i ý_' J C'd
Ci Ci F- F- F-
1_t +=1 t_t i_t t, l +_t
Iyl 1ý1 Iýt Cl Cl lil r"_ ý
Li) l'ýý
, WW .... r.
t: '-J
WW ý: ý-. p. _ . 1, ' "1 1ý . U't 1: ý 1. , -f . 1.11 *. j-
*: - 1ý 1 Ltý U-1 r'"_ al cký r., IT. 1 111 ý 1-, -1 lyi Lt i: ': t
IT"1 i; ". ) 1^I F- I. ij (', 1 '", I Cl I- 1; A U-1 i', 1 1D H 1-. , j) 1D 1- i; ý i'". LI C9 1.: 1
,ii t1 IIt ,1 i'( "". iiII t 1, " i't +. r"- I CA I i"1 t: '"1 1_i , U- !i1: '"1 tI11 ;! 1- 1? '+ 1; ,Iº1- IT-, 1: ", 1 r-- = f- Ir"1 (. _t .: -. i1 t--t 1_t ii; t" if t-+ 1ý1 ist"(I ºý +1ii! a11 1"-4 ht CQ1.. 1 il º-ý týt
i-j {- "-" i-1 Lt1 z W_1 F- J! - ý t=ý t11 J
_{ _" ýý--. ý_I -ý _''.:.
iT .ý .+ ý-ý4 1'-H t-+ 1
^( LL- º. 'ý i 1ý LL I-r ý r"- tý(
F-f .ýi:: t LL 1-"+ t--1 ! C' L. L.
in F-1 U. J _L -. 1 LL",
_1 _! '. -
I-"ý -4 ri i ., i
Crl -4
1_ il G-1 ciJ '* I! I
Ci F-- t-- ý_ý }--
, y, r.
! '"ý1 ý''d v .. WW
,ý-. ýý ..
. Ch r""_ . t'". J 1=1 .
t+t i""- t. ý! tTt t; j tý"t F-- t: ! t: ' J t; ý i iry i_t '" 't IIý: ý. i i 5. ý, Lt IIý, J ý_; i 1-- CT *1 t'+ ý t't ". " 11 0-4 1-'1 '. ""1 fI º-, i-a , ý!
TWJF-; Aý. ý74 W
LL ý-+ ýýý= lL F-. º-+ -= t=, U. F°+ Lt- .JWi:; t .. J Li ii-, r-.
1: i'1 ý, ''ý j i, '" i ýTý
r yl iy: t`. Iýr hI im C' =, 1M Iyl I, rl f^y1 r; v 171 1,71 lýt 1ýi 1ý iý; lyi '=' C! hi 1M , y. In 1i r'"'1 I=1 1^I I=: 1=1
4: 0 IT, Ir rI 1, 'Iý 1ý' 1 ý" U-1 LJ_. r"ý 1D --I ýt rt V. I. LI r".
_ !, 1; a11. ý"" *""g i", j i. ' t jJ i I. LI r"_ I 11 IT,
lji 1T"! .
, T"; _ I.;. r!. - ý. _
ý" ,; iiiiiiit: ý 1: 7! i '! iIt 1_i"1 ". ! _'I ýTi: 1_I1i; 1". i Iý 1
1 ; i"ri :i ! I. {i i ti. 1i, ": ;i. j ii,!! ':: i"1'. !ý i': 1tý: 11,
ha-ýi"; }.; i'ii"I ý! "
478
t
ý
iJ
1-i H iy
ti
e-a
J
I
ki:, ý_ý ý- Cl t. J
t. ý ý., I-L)
L) i--
CI I-
cl C,
I_J fJ
I^y1 ijl 1ý
ly, i lyl
. ry i, " i ý'"ý
ý CA Cl "CA CA I. A
}'j y I, IJ 11
I1'ý
ýf
LL-
I SS
wu - -.. y . -.. ý"iý ý"
ý ý_'ý __ý:
:ý . r. _'
`iý `t +? y ým I: k: 1=, 1 Ct
::. {+} i+f h- "r -+ "i Hij i_j ". "ýi l
C' i= F-" iT} -± ý j "_+ ýý ýý_ iI }t
Jhý: AW -'A".. .
:Z= Z= i_iý - H -
; Ww iý N--iL _JU
It
L; i
1: J L-1 L'! 'T LY: l'J LI J ýrJ F- UJ 1' ýjý F- "i' ý"L7 't i" j i_ý ýýýý, -'=1 ý_ý ýý-ý iIf
.- "ý ý-- ý? 'i 'ýý
_ 'iJ It ý-ý ýý iii "_ 1ý ý-r !:. f WJHW ". := ý--4 ". _ "=., ý T
o LL ~ 4-E LL H 1-1 =
t: r-+ J LºJ i*3 -+ LL -W
! T'1 -i
F- Iml
T. 1 in
Li LC'
T
1ý1
ý
ly
t
iý "!
1-"" l i, l
I. -:
i
T V I; -.. 1
ý ý tL
, . ý. ý ii-:;
1
" II
J \
J
'1- ""ý ! --
Tt i
1`1
ý
1-)
1-1
�+ i-I, --. ý :- !- -" -. t "_ . =+ 'i «1 1 _j ül IM
-4 1= .1
-. Hr-.
P-} W P4
Cu- Wo ýiy
T Wý
IJ ""ti fJ b.. . _,. p.. +
'ý" ""-r 'ý ý_ý ýý f-- I- I-1- iýa
W 1_1 ýJ ý_ý =u (r,
+= W Z: C, º-+ : _". : =.
ý=t ".: J -+ U7 0 4'ý 1=1 ' cý °-' "'1 '"ýr" " : 'y. "" 0 In '. ý ^4 clý i""- w in F- I-- WWTi "La .ý '"t t ý-i
" .: " = Ci to
W (". J §: 'J V. i ' il 1-) Wt''J i: 1 WW _1 _1 J"W-. -. , ýý , -, ., W- CL , 1ý1 -, = f! (i , ý= T k. 0 k. u - ti J =t il _I. - yt p ,yH º-. :A 7- ITl F"- ... - : . -- ,J =" =W I- %. 1. F- F- ý: = , yF-. W Wu l-Wto U- CL =1 11-1A1-1! 'T-F"-h1--f"-'. ijF-ý""+ýhi-ý=ýýý _
Ir-'-'. ýHH ýJý! '-".: F.. ý ". ý. "0W LL V lti` IX 1-1 G i_i I. D [:. LL L1: ! _, LL Ll-
--J 1--1 LL: F"-+ 01 -3t, ýt 1=-ý c- ýý I=i r-e ý 1� jº-4 a-.
11
1=1
iJ
1=1
I -, j UJ
f--- r r,
J
ýn
ii", im
r. -
1 ýI Iýi týl 1ý1 "ý D IM trJ I'D 1,1 0: 1 I wl Isl }:: I I7j lyl I, j 17D 1-i1 1=1 IM I y1 1=1 1=1 J'ý1 I-- :ý 1 týt 1'"1 . tJ 1=
.. J } 1"CI (- t iý "_ . 1ý1 -ý .j . -r 1 ýf- Ij'e Lj; t (".
" i t"I 1''11ý: _". + 1 rl' .I 'T il I IJJ ('"- 1ýi rTy I; J .r"! 1. ' 1ýY ý! "I 1 t. 1 (ý
"" I"1 -T J'ý .-{. 1 1; -1 I "! ( '"J 1, _1 j '. 3 I' itr: i rI ;.: t" ý I'ý ý 1? i 1, ýi1, ri I" "I 1" "'1 -r ,- -7" .ý t_t. ... .. 4 , ra -4
iý'ý . -i ('"1 1'"! ; '" Ii 'r 11"I I'' t : -" Iý i t' i I+ý !! ' : 1ý 1 Iý 1 1"'ýt i '? I' I I'"1 t' i f'i t"ý i 1"''1 1"ý"t 11'I i"' o r"" 1 "" r" "' 1" 1{ "I 1ý: I'"1 1'"1
479
i
t -;
H.
(! -j in W Vs C-s it- I-! U-) IJýs ý: JH
s_s t_s ._f C, 4- t-- ILL, in h, t_1 1- ! -
(LI 1ý1 1ý1 .,.. 1 V'ý1
rl IT, 1 'ý1 .r . -r eý"1 iw l -r 1 ý'1 :y'. *'
"" i'"ý1 ""1, 'ý1
WLi1 " "' W! t1 "0 _1 1_, 1 .J t=S " _1 _J t _' !-
.. . tom W"" Ct!
0 ! --, i. t:
J 111
-" - -' ''"' ., _
;= : -ý Cl 1_i t_, W i_t t_, t i-j !: =4 LL. Li.. CL i=l týt t_J } -; ""
.. . Li
LL
*-, ý-ý : ". ýý ý-ý ý-: F-; ý ý-.
ý-i 1- i '": t+a 1_1 W '"f
I_i 11 H , -! 0 al CL CL 1_1 47"1
, ý i-1 iý i=ii i i_ ýy
T
.Lf
tZtrtsiit .+ .. 1yt t1"t tit
tit it tit --r 1yt t. Lt ID s-1 ft --+ *i' "--r 1: ' ý t: '! -1- i "ýj i "ýt 1ý1 i'y: 1j1 . 1ý1 . 1: ''J 1: V 1'"J t_;
""" tU LLJ t ="ý i'S t ý; . t_t Lti Ltj I .' ý- ý"'- i... " 1ý'"! iil 1 :1
I -j 17', it --+ LJ 1- r F-- I=t "-^ ttj FH ti! iii ý1ý11-1 t, '"j ý-^ý 1. ý1 IýI
_ "":. " r, '4 �' ? f' ýi "1`'i
" 1.. l: W 1=1 HH 1=1 " 1...
- y. ., i
" : ý.
1= ti
I. ýI t-,
1ý1 "" *"ý i"'l
(. J +yi 1r1 '"'ý '. -r Iyl Lt. l
I.: ri l''! 1'i+ +. ''J 1--1 " iyt C: 'I "0
t --. J C, 14
" "W " "I: L
i_I {11 '. L3 . ý. IZI . "F-" . 1ý
W CL I: X: u-,, p"- " .ý I-- F- ''r ++ 1. lr 1::. 1 11 F- F- 1- 1= i -1 11ý LL. en t tý fl Cl C)
.: z :t" r-+
ü H ...:
0 Li1
ý: ý ý ýt ý ý
" F- it iýi º'(_
. , r. " 1, ýJ 1ý117ý1 -1 U -j 1-: 1 -4 I-Vj 1ý". j "r.. IT, I j! ! "r" IJI lyl 1 "ýý " Iý'. J
" 1. -1 t-L. t
! -_j1=; t. U Ia. J " Cl +..
-. I, _I º-, -. ý-.. -". ý-. ý"--ý
~y . ýý "ti: I"-"
Iýl I-a LU IT', Cl L1J 4=M r+ i-t ý:: týl 1; '. 1 1�L 1=1
..
týUJ "Lt= IýI "1 -i IýI
" ". y"
I- V-I cl ~, " CI {'"j " ." IýI I. y C. i, L! C. C. O1 C-i 1r1 f'. j
I Cl r1 . 1 'i yj "" jjJ
W"" ttJ
-". I=I tyj :, I ;".. I--:. l7I I'i 0: ' Lt;
{ý1 ""J _-"I F- , _: Lt1 1' "J 1ý. ý W F-"i 1=1 F-- t:: i CI CI ý""j Q) I7-I
I'- .,
ffý tl. ! lL LL; LL ý_ +LL. iJL tL I. J.. Ll. IJ.! +-L C, LL IU LL ILL u i4 ý! 'ý
I_ý, I t--.
If' C: C) y- ýr
uu -4 F. + " [_I
"-i i
: -; ý, ""ý Lt. l
1 71 -"- 1--+
-I
rf, "r
Jý
W .J ý-It
ýI
1ý1 IýI
Iy1 J
1: 1 IýI 1:: 1 Iýl it U-1 " U71 1: J
--" 11 "-ý II ý-+ il I" ̀I F- LU WW '"_+ Iý1 .
ý: -.. 0 :ý 1=1 i. WI -'
I_' 1'L 1=1 il 1=1 CL 'Z l. L. I-" C-1 1:
_, 1 'J 1: I i--. = L. r
-r -* ýý 1...
U) U'1 i-ii=1
: ý: týi ýý' iýi 1 ý: sýs t; ys týs : T;,: : ', 4: " s ýt s""'_,, ý tý's t ýý tsý : y, tý: tyt sr"i s�ýI t ýs ý :ý týs sys 'ýý sýý s"ý�t tit ýý., týt _-, s',! ''': ' ý#' Is 1 `"I . (" siiiit ýs ý',! ,,,... I! ', '. [: f"..
. 'i itf 'a . .tr!:, _j_ , j..
IJ. t(I}DaIJ. t11}1! 1UJ[ts. ý, t.. s! i, ýL. "ý t. Lt. ý s, ýti C f_i .
f". t , '! t'. ", }: t t', "s t', "t t-tt"11 "i t,. "t t ^: ,ý11i "t t. ý , ": i: , '. i' :. "ý 1'" ; t'ý ý i"ý ýf . ", i""'i ' 6", 1:,: "'. ; ,. t :ý
480
/
111 "'- ýf
+ ": ý
* L_ 'ý,
-d- CL ý. 'r- °.
t- -ý- r.
.4 rt 7'
-ý: ý ý
-ý- 71: +
ý: ý rt. ý ý
-q". ý 'tb'
'ý" }-: " ": ý'
ý 1--i
t- '"ý
iir"ý !; ý- +
!- -ý -.
. t. 'r 4- !ýa..
t L' "ý- i: ' -" F- .: 4: ,` -ic- 01 CA
rr :.:: t-ý !! ý__ ý" l_
f::, Li i
1=ý ý_. ý " 4ýT , ý:
4i "_. 1ý ý-LIT:. '-ý Li_ 4" ,, i--:
=--ýi =i: IT .rJ. " 4:. Cl i_. 't
+ + ý : ý. ý :t
_ý ti_
1; -'t.. T= i- -. ip-J. _-_4 w º-+ j- Cl 1D . '"
:r t_. t t_: 1 Cý,. 17, -l .. :r t� 1_I yr" "--.
1--- 1_ 1e W4
t'; t 12: ý: " Lt:
-+i:.! LL 1_I LL LJ 1_, -" ..
l_ ! Z: T
I. il 1 =t !. L ý. r"' : -". 1; _t LL ý= ! i! "
ja i --+ W ".: ý.. ": ". _. ".. ý ý; "j-F- ... .
- ly t, __ L' +s :: IT' "_
Li:. 1--: I i. L.
ly 1. ' 77 LL L L2 LL LJ t, t -L L:: ll_ L"# ti[ 1.511. F-i
F-- F- i-- !- f-- = -+'. -' }y= ;- _L. 4 :
-
i_c 1i i_7 ! _} iý :ü Lt. Lj- 4_ L
'tä "ýý
ýMý ý_ý i _". f^ ý. ý . {..
. -tý"iýýi 1: ý
- (j Cý 1ýý Ij: fýt ii1 aý
!. 1 I. _
111 ii : 1' ý 1. ".
I-r, ý- .
r.. ý: "ýI r, ý
_ý 17-
LL -"
'"ý'ý '- f=' 2
-- ý-, '
lyi
tIt '_'
ý i-L!
IZ F- Li_ t. t +-. ý
4- 7't
ý- ý
ý ý- ý ý :ý :ý ý
-ý- -ý
, s; .-
= y- . ý..
L* --ý. L: '6"
- _ -C-ý __.
-ý
ý ý ý -ý *
ý Ct:: Fw
ý- - E- iý. ý
ý !y
ýý ý ý ý
I.
Let
=` - LL L. .. -U . LL
* :,,:
t . ý.. ý F--i
IL
ly
ý F-
w fýl
ii
w ! ý
i:; I_i
ý 1=
ý ý. ý ý= t- hh , <- ý. r� `ý ýý , _, 0
ý : ý- ý ý t ý .: ý. ý >
ý ý ý :ý
ý Q-
*
ý
ý ý ý ý ý + * ý ý :..
ý ýº- ý- ý ý ý ý
ý .4 -fi ý y
.4
T 1ý1 ^ý f
-a- *
--r -+ -r. ý- --
( "_ t_ý. tv
ý"iýýf': 1I ý7 ii , ._-'. ý ýýý, ii ýý f'f
i /
481
ýi + ++ ++ ý- + + ="- *
I- CI
CL
+_s
+_ 1
ý i-
i'1
"_i týt
1ý1 Iýt IýI [i
t_i
ý: t. " ý'- , r. " 1ý1 iyi rý 1oi 111'ý+
". º_{
J--.: T
F- t! - iii11I= u LL ! _! LuF-ý "ri 5º-si: S
Lr f- ii. Ci -1 IZP W if, i- - iJ LL F- W wiijF- `, c-5 o-= i; i ýi _ý i_t º...
'1,1
I^1 y-=
" r- 1? I I-. _t . -. 'f' Cl -t JII.; iý " r- u n. =t F- IT, -f= Pi IT1
ý H ýý . 1-N
lt: W
Li. i F-- 1=
[i 1-
i
ýt W#
`-- v) u ºZZý. f--C N C+_
=Z -4- " '. ' -9- l. U -lb- 0.11 -. -4-
.. ý 1ý F- . f. r.. f ". jý
-" ii -ý Wrii i- t= lL ý-
ýý w
L:. 1
.-º., . -» - J ýjl ºr ý ;i
LsJ -. 04
Ci ý_ý Lý ti L =J ýJ
cl (_I t- 0 LL W II F- ,. +
iL 8 w --: - E- r-; . r.. lIJ w" rýCt "-ýi". jF-ü-t - I-- W -. Of CL CL . 11 -- =I '"fj -- I=. 11 %A) ti LWr.. 131-....
tJ: ". r {' ºý+ i !i ýý+ f- ! _ý r-ý L' { ! ýý :Jý: : Lij
AWJ=.... 1: 4- -- r-4 1_
LL 0-4 º-r .T ý--t , _l L' J ý_, L. LJWJý.. iic:
= -' F- h- F- = 0_9 1... LL1 = º-+ ý-- u ... U Cl: 0 Cl- = Lt º-+ º-+ ý_ 1 =. 1J li t; =t _7
C. '
IT 11yý. -ýI;. f -i'U'II. j. 1(
"'! ýf . `... r .t rt. týf... r
1::,
IT,
# i- ý ef t *
. r. .9- CA -+
.+ tfl LL
". -+- ; +; U, i
. -, tl -+ý
("-v, Lt-ýJ+ ý..
"'h
k go {_"2
-ý- CC) W %= "+. ý- l rt Lr -+-
i- º-+ LJ ýi- . ý:.
tti+ ~. 't=t ý- !-
4ý lsj . ... F- ++ º .. (.. ý[ -i- "-. -Gý...
a':.. f': --t(. -4.;. Lt: +iI1 ::. 1, -T- 1'-_i +" 1-+ "M-
G, +I, -JF--r + LL lri + + -. I_1_1
D 1: L W -e- X. !- b_. + "W ý^ " r-r ý-
-" N rý' tilt V lt: + P'! " '""., WW"',.
ZO; 0-4 It º. " i ". " ý-+ r'd + .. _1 _J -ý"
ý ý ý
+
r. "r 0 off
. +- " ý- -. -M U'i ýf -r" : ý.. ",, r "v. `-' . 1,.
F- +L tL !ý -a Lt1 &'i t. t: + LL LL Lt.
, -4- l. -W r. i ±
4: 1 -:. ý , -, , =-, 1-1, . rl+ , 'ý, ... 4 +1 ..
. -. -ý- ý.. .ý "- :ý
r+. I, wl t"y'r rý' IýI ir r±r r. a: r :"r är rý r ý" rý ý :" rýý rý rý" r- r. ý r": .. r V"r L} Iirr rT " rýr "- {. ý r, , .r tr ý. r ý. . ý. -ý +.,.. ý ý. r. ý. ... t "t 1
.; "". t r" J! '} r! '? .t .t .t .r! "f +? +t ýt ý .ýr .t
482
APPENDIX IV
DESCRIPTIONS OF VARIABLES
483
A. 4.0 Introduction
For description purposes design variables have been placed into
four main sections:
A. 4.1 Variable Description Notes for Individual Cost Centres
I'll This section contains for all cost centres: '
(i) an individual list of predictor variables,
(ii) the order in which values for predictor variables
should be entered into the computer program and,
(iii) the equation codes for each estimating equation.
In addition for specific cost centres descriptions of
variables can also be found in this section. Those variable
descriptions not present can be found in Sections A. 4.2 - A. 4.4
A. 4.2 Common Variables
Variables used to estimate manufacturing costs in more
than one cost centre. In this section variables are listed in
alphabetical order.
A. 4.3 Variables Applicable to Single Cost Centres
Variables that are specifically related to individual
cost centres. Variables are listed in their particular cost
centre number.
A. 4.4 Illustrations of Welding Positions
484
lf
A. 4.1 Variable Description Notes for Individual
Cost Centres
485
ö N
W
cn oW UU
Cl) ý
H ý
G4 ý RA
W
U 4
L lV
ý
ý
ý
>
W L
N W f. J O cr- C{..
W C C> O C) C C C.: C G O C O O C C tat G c CJ
O C) C 0 0 C) L. 0 0 0 O O C) O O O C. C]
- Vf W
J L7 J .. O W O Z S G
C O O N C C O O W
O J r_S N N J W l. 1
ý -2 _ U M S C W ly C L S :[
la. t. ý. lL CL O U
-D t7 C O O cc
O
N 4J 1- N
rJ a N G ° n li.. C t a O O
ö r ä C Cý ,
Q CV 1- L J W tj lJ Cý W
W ft: G J W
~L-
O lt C r r" -- C. CL C.. i
- N
. U ý tJ
b7 W . -. ýJ V1 V) ý
1- =
J
w
F-- r' G
r"1 ý" O
CLr O
W J
S
W 1-- N Z V i. w 9 J t; ^ S !1 W
Q J . W J :* 1= C W t .r lý J (A C I. - W ý ; M 4 N Ü ý
-yW l.. tý 3 ýý
Ö L) 1- C. ) W a
Lý LL. p J
W f ^I '
ýA) `
O ý/ýý 1, ý'
' II II ºI / `I
--
ý ý/ Y
O 0 ; lc
ý C
/n V /n V /ý J K C .ý
ý. ý - C) %Aj L-
I: ' L I i
C. ) -c _ r- c\ ". ý ý
u -- W ~
[p tV M ýt t. 1 v^ O ý tV N\ !T tt1 ýO tn C)% ý ...
W
4b-E
ý
N C N
W K
F- f-- Ný O t+. F UU
cn
U
LJ ý
U 4
.z J "S er_ W
W c. 7
N L. l C) CJ CT. O-
c' C) o o a o C', u c: c: o o c: ý, r c n
N L, f
C
O O W C Q O C C C) C U U C LI C
_ .i N
W ý .
tn t't laJ ._ N J ý ý
C) I C Z -o
C uf w O W
U w O O LLJ J O
< O C G O N C J W lu 1. J
rý. . "- W ". ) N ý (L' Ö C " y t - o L C 7
_ O
L. C) C) M C O ý1 la
. G . . ý
... * ".
cn W
<(
C- -. (
.1 . N "= t") O 9 F- C3 . -^1
'Li co G .... ` O C
O O t" O ^ rý -
t.! 41 = tT U . 41 " W r"ý N t J 7 ý- t7 L . 1"- 6.. -
J l ý! "
=ý
. O O
W Ga . ;
C > t r7 O 17.1 r-) .'
W 'w' ýAI
f I `I I, tý
N S C'f F- N W
7 . N C) N
N W =
u Li º- J º- _ ''" -ý S3 (ýý.
" IJiJ W 0. W .L
<. "J lJ ? J ýJ
4 yJ :: - 4 J : 3
W ua.
!) W N
n ~ _ J O
J J .M J C) CJ 1-- 4 t:., ..
O Q LL. :3
L. La. Y C It 1=3 O 46. ) U =I
/ 'f -
ºi =
Y' - - O
",
o
ý
_ o _ . v i n .. /A V /A V /A W - W ý� º". J º-. W tz
C- J J ^ rS C. 1 tr Cy
v u o c s "7 ý- '2 tý ` .. 1"ý :J ýl. ' "
v v v v v W N `- ;
' } ln _ "
2 lyJ t tJ
l ; - : -- :o :, ;; .. t . F- a t- C: t ºý Lf \
in
ý F- "]
L H
PT Q Ul ýO fý N Qý 0 ^J N1 4 V1 %O p" ct .
.
W
487
K1 G N
W
ý- ý- cn C) w UU
N ý ý ý G L U
cý cm
W Z
2 U ý
Ca
v a
s a
I.. i ý c^
r/1 w
c:. ý
W O G O C C O C O O C O O .:: C l.. G C.
C C) O (=; Q
.
C3 G C O C O O 0 O O G 3
O C. 0
laý W
C)
= rý O C O C C O O U F-- W
O U J G O N .J V.
C O
4, U.
O O
ý Z L Z V
Ell ý N
ý... r [3.. - -S N 0 - C) LAJ
co O O O ý O ý O O p - N
Z Cý = W ý - s ?
. Q J
W v
W ;,,. ý U. 7
0. 4- fC is : - ^ý. '1 W -r O O W r, 3
w o - W o 2 r
S
L . _ :3
r- N in rn
N L f J N J F- p i
/- C7 rn .., z J 1-- ý. tD ý - J C7 Q j.; -. j W W F- N tJ
Q LLJ J
Q L
J CD W ý O J >
OC C W W Li J W fn O J J ` C. ) ". ý
W L s
. _ J J W W
K T t`
%. ) "y
Q
V O -
C O U F- F- O C. 1 Co LL- J th- C= W ,,, - ( L t
W L O
C7 L Q
- I\ I\ Iý ""
"
° ý ö - - - ; C J in
V in v /n
°
J
``' =
Ö `
2
°- %S
J v º_ ý_. .. CJ
i
"C N _C c-
d
I., CS .r
CJ L--) F-
-- - C) "'
rl CJ q -. 4/i 2
=) -
W
- Ö
J "C ` lº. f: ß: a
-- W C
C) - N M t; 1, ̂1 W H u) C, a -. tV vý vý P co tr Jrý ý ý ý
488
v c N
W f-º- N pw UC. )
N
H
7
U ---ý 17- C-)
W Z
U C
co
C1: w.. i
W L
N N W fJ
C1.
W C C O O O O C C O O CD O CD C C ý
C C O
D C O C. O O Cr C C C O O O O O C a
C C O
N W
W O
D C O O O O C U ~
C C J LAJ
O O
''' ý J N
D
Li. U. C.
O - O B Hv Z C
: Z L, ~
CN N W
ý W F- C N
4 c3 CL
C Q
C <L . D J O O O ý C
CO O p u LAJ
F- : 4 J W : -J W
.. t O W U O W O Li. f=. ̀ ý. _ C: "2 ý. i
W - = W O C C
lJ Z N F- ý
V) J N N J ý- L p W
UO L .
J G - -1 4 C
i l7 W
_ J C7 ý ac ts. W
C J
D W \ L-i U. J J W W
K 4 . N
F- C3 CJ C Ö J p Q G. K
W ý W
: W z Z C, w
:s :; i m m
Zi '2 ö
O
G
ý -
o h
-- = .d U ~ C .J
! N C I-
ý O..
ß
v v v v J C :4
z w L"i = "ý _
rJC J = - º
[ ea - i- tL " - =ý :J ) -ý
c3 ý- W QÖ
_ V M t %Q r" .n ^ O '- LV )n U 10 14 t? ý O r. ý ý ý ... ý
laJ
4by
%M c N
w c=
�- I-- Ný G !ý UU
-. j
u
W
w z_ s v 4
cý ý
s J 4
s W
ý ý ý
Cý; N
t. )
W G C O O C O O C) C; C G O Q V Cs C Cl
w
C C O O C O C C O "C
C: 0 C) 0 C k C
t )1
ý
f N W
f
d -
JO C N
L) W O J O
W
O _ N
ý G O
N LLJ
N '
L` O
CL J
ca
lu Cý
°
La. C_ O
C ý"j W 4 N
_
n - '- Ö
0 o :.
C t, r- n x
^ - N
C: S >.
1- N C7 "t Lý
Ü Ö
CO L. - C9 _ ' _
`: O 6'
- O O
W -
tr1
u s N J ry ý - N
W - W J w
rý c ' '
W W ä
t. 3 ^
li.
0 "r n
G J - - G". C: L
W = o ! = W - - .r 1-' w W ý= .t s N
I
N ý
L ý
N $
N LJ S ^ Cýi
U W P- S4
C7 W W
Cý u t.. l J
J ýO
Q _WJ - J 4 U.
W Q > K-
N . L. (> W
\
Q tý G. `ý
LJ N 1-.
W .ý
i ü
1 J
J J
O JJ 1:. 1 U
O U H W yr, [�^
=3 Ö Ö L.
0 G. 0
L. 0
4 W
C6. i
W - - w
- - O
v O A
.a m 2 t
-c C. O
0 - - ý
0
/jj\ ,
V /n `
A /ri
\
M J
ý W 1 _
O
W /, r J J °ý W L)
ý 1 `C O r
fn
C7 .
1- C
= LV ýL
v j t. 1
tY .ý , ý. J ls:
al
, L. W I)
W w "
W Cl- Cf. :. s
W t
V j
W C
ý f"- l+- ý .ý r) ý.. CL !ý J .
ý. ý '
t7
r.. ýV M ý7 tn %, o t- C17 v O IN M 1V 111 %p h . Sý n ; ._ ý
L. ý
490
C, N
W
r_ N Q UU
C/)
Cl ý
J
v
W
_ v
U
j
f, r^
N ýn ý
CA-
W C C: O O O C C C O co C O O C C Lºt C C O
O O O CG C O C C C C C O O O O C: .. ý
O C. O
L/) L. i J O
LA.; La. ý C] J
O u O =
ý w ý
C C O N
C U
C O J C Ca N J W
Ü L: W M C/I 12
C) L a. C " G C3 [il LL Q C
LL CL L J
C 0 O cc C p Z Z
N W L-- ¢
_ 4
« R
co LL _
Ö
a ä 0 0 0 °
C U CD C W } L. J =
S U « J N N J
G.. ta. K_ L. J r O W CD O W r') LL _ C C' C. ̂. C: LL V
_ O _ w
w P,
L
N _ ý -- N N LD L LJ
N W :. 7 .: J l/1 l/f J 1-- G O W =
! =
C . 41 J
W
O W
J l/1 O
cc 1-
O LaJ
W J
W Li +C
M 4 N
" -' L
J J
W W
Lý� C"U (Y CL C-) :. ) I- U Ci ß
º- C)
Ö C)
LC') Ö "ý i -K O J
W - W
L W z 7 1 = 11
:: f ý.
C ý S S J - ý ý "
O V
- " _[ " 0 :n C
in v in v in ö W ö N W ý .
J `C_' N N ýC K : 1. .1 .
ýr
L. J p _ _ ' c
Ö Ö zC tj
v ºý
W W
Ü H v v v v
1- ý
W \ J :2
W L+J W ý _ W "ý
K c
ý) C
'r O
r L. J .
-- " º - ."
LL. CO tt: > ýt L
W C p - LNJ M ý Lt1 V7 fý W ^ O = 1V M of 1 ui fý N pº rI - ý" ý ý
lJ
491
.
C cv
W
t- ý- N O UU
O
ý
.
x ý
sý ý
ý ýJ1
"t ý 4-; W 4i
n
LAJ ö
ý
t. i C C O C O C C: C. ̀i Co G C O 0 C: C ýi,
G C V
C7 O O C: i O C3 C: 0 0 0 O O O O O C, ̀ C. O
N : -t J ý
N W J -
O tý M-
r_- 0 0 O O %0 C-) W
O C Cl O J O 7 J M tJ
O tý. C O
Ö _ - 4 L - 1
O f1C O = % Z - L V .. r
W
W O Z tA
cc O O O O ýt = ä v Ci i` Q W
O ý
a - :i Z Q
J W ý. J
iv 4 4 _i _. 9
tL LL- Z 0 _ C. )
W N_
º- Vf N N p".
} Lj
ý, W -
''ý tA N
0- J 1- ..
cr W -j
x
ý - W W 1- J
(A 92 0
r 4
41 J
T Z
J W :C ý
Q i. ý J ttýýff
v CC O
ý J ý *ýi " tJ O I-- W .. + <
-C f-) .i vs
ý rý" i L. J La :L #
^ ý r- ý- C C t.. tA- eýG e. "c " C O
W - W - W ý A %q W
ý ý " v
.ý .. .ý . -. ..
f f f " L 4
/A V /A V /A J
lý =
1 J v W
!I
. ̀ý N ', N =L .ý . a. +
W J J I L - V i () º- t': � y t/1 V
. i. K
v v v " W -j «
f
W ý
tJ W
º- tL , f w ý \
c`. ý lJ ýÖ
- V r+1 %r ti1 ý. 1 1% ) !, º 0 tV Mt "1 UN ttY P CO Cj%
tJ
492
N
W
r_ º- N 1_
OW UU
N J J
2
lil
x U ý
ca
J
cc 0
f. 1 (il
W v ý
W C O O C O C
C ...
N
Q
O C C O G C C
C G N "
ý
Cn W J C
W U L, O C C C/5 C J J U' O
_ ý C"r J ft, C KL .[ H. } C
F--
N ln Z Q W
, C CD tn
J" z o C w W C s
W J ý C
C W U C] C C
W o ý c
V7 V)
H ý
` <L w
W N N
J "d W 1
C t--" /- t,,, r W J W 4. C Cn J "= 0
Q CL O c ý ä
W W _ c
G t _ _ O
ý tL c r- n C: " 'L t. i - ý -
Q CL Z
U L..; "S C -t a
t= - º- J
tJ - 41
'l U N e. ", � ttN .D 1- . rg
41
v N
W
NZ O taJ (, ý U
W Z
._ 4
CD ý
J J
(N tN W U O ý G.
493
W O O O
N CL O
O C-0 C) O
J " J
O
N_
W CL. L'i
U p O O
co
H O J
N O - ý N v
e Z F co W 0
Fl: N
/-- O = O s, j O Z
XX
W ý
v
J
N Z ý 2 O _ F- ý
Q r^ N Z = J W 4J J J
F- O Ü
U Z
W ý v
Z N O
4
ý W Cl- r'
ý
V O fl. J = J O 1
Z F- 4 W S N
Li tO
N M -q ll1 v7 n r0 Oý
s* W
N
F-
Z O
1- " QC J 1.
Z QC
1
W c7 ZZ-"
U =ü UmOL
Uý O<< z 4.
ae Ü" W ý7 J
W z>
< U !-C < d' f= " 2 L. S
CaJ F- C C7 0
zzCz c_ º- u
ca j a {. 7 U. OztO º- J
z0
Z, cC0 I C7 1- H
rU 2z O
Q`
77
ýCO O r`. tC
JL 1-_ z< C-
H t) ºý OR Mý O
ý IL . tº O
XO
d. ý
11v F"
_2G zu GOJ
NN11 2 u i.. S ýýrx
Cý CJ ; RX
X Fý ºý <<1
=sÜU
.....
-NM sT l! 1
.. 4. I
I. - c ý
0
494
N
W
F-ý N ow UU
ý W
J
U
Ö
J
r U
W Z
2 Sc-,
6
cý
<ý
(/7 tN l. J U U C: G.
w C C O O
Cl C C O
N
C C/) C
U C o M
G C7
C
ä C O
t, ", Zn
C Z C ý.. w C
c cý c ý v U
cr- C. m
. x ^ i- - ö
ý J J 4
Z L
C) _
Q 4
O x r^
W _ - ....
O rý W
4 O V7 Cr J w W Cl. - Ü
J -C -t
- -ý
tO - N M V lLl "O r- co C%
a
495
.O
N
W
r.: ý--
N UU
ý ý W ý
ý C7
W U 4' U. (i
Cn
W
2
ý
ýý ý ý
67 (/) tJ !) C-) (:: C...
W C C C C C
O C C C C C
N W U C C C G G J U
:Z
Cl G C c C 4
C`
'Z LA n' 7 N - W
_ /ý
4: Z ýZ: Z
/n V `ý C'7. tu -_ y
U I C C ý
ý
r"-ý
41
ý
z 0 ý
r ý 0 ý N
N
X Q
1-
Z C
N S; W
1" W CL
" .. t
ý
ýC C
C7 t)
4J
ü<
. ý v` 7ä ý)
-Z Q '<
III
ý" 'J ý
`: .rr
III
Z.:
.ý ýi ýýý ýýW
S
r!.. *
ýi I-
496
_ý N
W C O C O C C C O C.
C C O C C O C O C C]
W U O C C O C O C O C
ý
ý
i1 C.
1- W
C] O O C O C O C v
H J
O
W 1Y
ý
W W W
li. Vf C
M C'
J V `V / `
r
\
\ý =
M 'cr M ý yý ~ ý V
in A r n W
cc M
O N "J C_ = l! 1 - U1 _ -
C F- ý ý !
V - - n - - ý- M
ý : _ý '
-' N M � V1 n f. Z ý
Sf ý
ý_
C' N G-t, 47
_m w cý F- F---
N< O l. J Cý U
W
S 4
ý ý r- ý v ý taJ : 'ý
(! 1 1, -i
ý
x -, 4 - L- -" Ci F- ý)
u"1. J _ý
t+. -ö c ý- - W`^ .3.? Cv
.r C
w 7-1 . .J
r= c.? t3 W
UV
u 6. ý-- F-
CC
Ku
_ .., 1
ý ºC"
º- Cý
NC .ý ý^ QW "ýý
NJ C~
=Ö
º=i G
x <-
ý- a W G' ý .: ý ...
< b- Z. - - CV L'
Hýý `
X p.
r2Cýý sG M' C»
h" _;:
a
.im 41 1 ýW 1ý7 C
<r !ý G 4. C r" F" 1ý
... O -m- r I.
ýý o , ý"ý cr -'. ý _ ý+ W\O
rL f" r
L. =
I- I- =_ It ..
X. 4
ý .:. ý >- - c. º= 0 4 14
V7 W
-a
497
C, C"i w CC
w U
N
N
w ý a J
>- C-
C-,
U ýS
lý
ý ... f.
ý ;_ ý ý
17 V) W
O
UN w o C C o 0 0 o o c n
C O O C O C G C C C
M U O C C O C C ý . O C C
r +ý
N W J
¢ Ct G C C O O O ý, O C G
N N _
Z ! N c7 .
N W W W W
4 W J L} ? - ý J ý S Q ý
Q "`
- - cý c^ s s
, - o . _ º..
W o z _ . W ! a s
. -. .ý
v1 cn N !
p O C .. ý W
V LA. v ...
A v C- .., _ U ... v Y W - N - .r
- f- 4. U fý 7 i
ü ý- W
, ýD v In .; -n 1-0 r- iz CN C . ̂J rn ý' ý,
---r - - ... .ý -
Z
ý
s3
ý
Q ý
N r ri e a ., ý
ý .ý J
ý :z
0
v1 w r ý,;
498
K,
c"I N
4J
a: r- i-- oW v c. ý
r .J W y N N Z
C/7 V7 W
pr- : ý..
W C
TI
C C C O O C C C
G C C O O G al 1 C C C C
U F
Z . u N
, ý O - 7 Vf Q N
W t_ F
_. n J 4 C N Y C >- ß. < !)
W " Y Z C7
Z .. _ ý'= Z Z
N W J co
ý . rJ N
c7 CL
_ N J N
_ W J -J
L- ' = -- - - -J ' i
-
.ý V) N
Z F- O -ý o z
¢ ý
W Z N _ cc
- Z N_ ý Cl
Cl, _
W l.. "^ -C ý
W Q _ ý H Ü V) Y"
7zi _ z -C W Q X t N r
Z Z ::
- N M Q
rn
Li ý G 7 O
1-- C^ t J ý .
U Z 'ý J
ý ý 4. F... f.. "; r J . "ý S ý "f C. '7
_ "C J -t = rn v - 1,,.
-s
W V M Q ý ý r CC C7ý C N M Q V1 ý'J n St C7ý
ý
ý
u
.:
ý
ß
VI H ý °l ü ý
tý
cý.
ý w ý
ý y
W
H
V
499
.O K1 N
La. J
ý ý- º- cn z oW c -)c-)
W
ä
N (/1 lJ
4.
W C C 0
o O C 0
W U C O 0
co c c o
N
N v
W ` N
O 4
ý c
[ F"
C. -. 'v W d
U i H-
t -
W Cn cN
lJ
N rý1 ý; lC% 19 1- co C% J ý1
W
W
O N
J
z a 0
x W
ý "
W
N Li a
w`
J
z O_
U ý
N N O
U
V cr.
. 1. - O
J J Q"
O
N cr. W W W C:
2 0
N z u ý
O
J
O
N
N O a U
?C S
O_
N '1. W M'
C3
Sc x
00 N
Wý Fý c: j
0
bUU
Co rn ý
lJ
. º- ý- Ný C) W UU
Ljj
C3
ý ýi ý.; ý
f/1 cn W
v ýý-
w c c O
O O O O
W U O O O
ZZ
d C
O
i
N
n + N aýJ Z
H Ü
T
ý
ý Ö ý:.. . -. W
F"" ` L
W v ý
ý-- N N
L. J
äö - w rn c ul -ýo n m rn _ý.
a
Co
L.. 1
C F- F'- Ný OW UU
C= 0
C) U
W
C..? 6
tn cn W rý U
G.
Li C G O O C C C
O O C C C C C C
W U ý
Z } } C G C G
Z x
v x
v
C] Y 3 _ _ C c C
F-
Q J
F N
W (1)
J 4.
lt,. t. 7
-- ' == ý ý p..
"1 L^ tI', Oý L'1 ý J
V M Iý w ý
N C -C
v - 'M r. J
o ,, ) :s w
" r_ ) 1-. w ~ C. ]
_ - 1 º- Z C
a : " . n
" ' " J I+)
ý LJ
. tO -0 f- (: ) Oý ý tJ
w
Vf
CL 1-W
Fu
W J
W J
4. 0
ti ... (n bf ý ý X C>
I-
u 1- < J
n
F-
i
N
LW OO O
F+ 'O 0 cii L
.. r U F+ c7 F+ W ý uý t, o p, CZ
...
ub N n7
"r1 C! l+
ea u ca
... ... U
ri d w ý. N CJ
d CO" . "C N 41
.r w r+ r co
....
-ý N rn v
Öý
Co LC1 N
w C--
N OW UU
ý ý a ý
U z
H R
n
W L
U
ý
J
ý W
W c. 7
(N C/) lJ C, )
CL. CL
U, W C C O O C 0 O ü C. 7 r; C O
U
ý
O O C O O O C C O
I
c L: O O 6 O C C
v
N
O
O
LLJ ý J
4
. . t+.. O C
_ W v _ ý
t) w O O J
C C C O N C C W
ý " "' U M N N L ^ ý O t . F-
. ý ` ý
W . W 0- -j J Z O O
G
O O C. C
O O C? " : =ý
. ý ý L I- I- v Cl) LJ J G KL
=L Z
I- N N I- C ý1. . N - ý U CD v' C - _ i"º
O _ LL . O N v
[C Cl < 0 0 0 ^ Cl. N
N U .
C) W .
41 N
G . `'
_ ^ \ ý 1
_. + ti
J W +.. .ý W W C
O U W (T C Ö .
ý Z UU z ' :.
W TA
ý O "ý c .. 'ý
ý y t7
'.. ~9
. >
N N 4J 'L N O N S ý, _, N
VI W T CJ W F- C'7 a
'
J 1- ý _
^' C7
J C7 -. T " C: J
.J N W J Ö
d W
.ý
W ? < _
W .J d
C t,. J Ii
. ]..
- J W
J
= - ý
v
'7 F-- ý W w c-
C IAJ
n ='J ' cW
7 J W N C-) ~
ý= O
_ J -i U Lý H W * r
ý O . O O ý C Ö ý W L. G < t . ý O .
W w O O
Z O O w
O
CT3
x ä 2 - ý v
II
I
- "
_ S 2 1 ý G
C `iI Y
° 8 ^
2E ý=
. ?
` ' A IA K ý ý, o .: c `y //
\\ V J Z: : T- ý v v ý
. ý lA
" "
L J ýy_ j
. _ ý lý a C O..
A. W
G W ta. J ý
N
W -L W .' ' n
-ý :
Y < `r - -C
4i _w
= a O
ti U
) \
cý \
4"- W ý I- . .J J .
Ci
:
h : :
4
: :
-W ý- :V M v 1r V
- - - " - _: c. ý - I
whý 503
Co LN
W
Cl- V) z o UU
ý
ý
c7 z H
x U ý
U
U
C N 7
J J
U
C. J
w z
U Q
J
ui c. 7
Vf N ki
O ý L-
w c c
C3 a o 0 o U U rý c o 0 o c U-1 a c. ý o
U N
-
O O O
"
C C) C C C) O C: L: i O C Li O O
i .
v U i
O
C
oe W Q
W W N J O
G` ''_]' W W v
CU Ow
C) Li L'i J O
ý C) O G O N O C)
Z J W W j U `L C--
C C rr O
r_ C) W ^ L+"
1- . La co :1 C ü O
ý Z
C) o c = ö
0 o c M U.
F"" o Z = Z _ _ 7. . r
N W J Cr
F- vN d N y c .. a. n N - '= o Z V G ý
_ O O G O -
Cß <L G
S C') <
'O 0 0 0 O r CJ, Vf
-" ,, ., J i L! _
Li _
ä N
G ý W w c < 6
N 1-- J
M- LL. J W
u 1. _ cc C) C)
co C _ C] C] : _1
W O W = _ _ _ 1ý ̀ / Z G7 ý v
N N }
CO S CC W C =
N 2 U W 1--" "
' 1--
J ý L ý L7 ý
i , '. L. J Z
C. 7 C .J " C/) - W J
:C W W CL' 1 J - v J
- J <i" lwi. J
W fm
ä ý W
ý C3 LJ _ CL. LAJ W _ yz C LJ
1" - W -.
ý y F...
W - O
1 ý J J
W U - Lý ! - W ri C
O G ý = =
O CC O 4.
Q L. U.
1) W LWa. G Q _
O Z
F- ! - o c: C C) C: CD O C" LaJ =
w w ^ _ o Z o 7 W w o
'2 x a _ o
. -. i-+
ý ä ý "
O Ö
7 1
ö O -
ýý ý ; 2 ýn v ýn v W `ý--^ J J 'ýý
F- Li. "
O :) L A
-: c J
Ci. J K Ü
. ý "".
7 lb Ö =
U o c : _ ý º- ý@ cý ; - º- v W ý ý ` .. v v v v 'ý y ý ý� -< V)
G W w ýy ¢ Lr
W Q: d
` iJ
rL . U Cr . L º- c cn C "I cc F-
L: )
1- '. J - hl nl V tlý ýO r aJ aý O N M R: I. U-i %0 h (X) Lh "t ý) - - - - - - - - - ýc. ý
-t %0 N
ý H x ý
cD z H A
Cl) ý U
a
504
ý 0 0 \ O 0 0
" ý a a a a r+ ,a
. 1ý C'1 ý7 M
n ý ýo O N ý7 N N 00 I-
yp 0 0 0 0 O 0 N ý -- ý ý ý
a a a a a a ý co -Ir ý (31, n zr co u, Ln in
Q ýt tn -4 CV .7 C
O O O H H O O ý a a ý a a a
r-i O ý ö r, r-4 ý
n N O N " r-1 0
O O O C O C
u ä a a a º-
cn w 'a C H
A n 4 ä a a a ä A 9 H H
C'
U O O O O O ý O O v
n a a a a a ý N H P 4j aj aj = .G O 0 C: 1 I
ý
ý O O O O O O O PQ C
u ä a a a a wl H v
Q 41 JJ ai .c .c ý G C
0 00
d Ö O O O O 0 0 .
11 a a a a a a H .
H .
A +ý . +ý ý+ +ý .ý
. .C z z
Z U C, ) 0 ~ ý ý
H E + t n v i
Aý 1 > DG ý
a > a A =t
Ö [ ý 6 O O
a 0 tý P 4
H H H W H ý-+ v v
ý 9 C W
... ý
H H
1-4 ýöO H
0 Hw ýp
r-1 N c'1 -7 tn .o rý 00 0N DU 0"
cr v
4
. -. o +1
vU C)
aý S .r 3 . 4- 0 N ý
ý ý
C . r.
ý
C ro s +ý N N CJ
r- Ul "r C
vp "1
C . 4-, ýb ýV
"I-
d 4-) to
Oi U -0
. r.. i
(a 4-
4- 0
. -. O v1 iC
d
a. ca a "ý- E
v
E
0 U
41 U O S. - 5- 0 U
.ý
ý m ý to >
F-
0
aý +J ý C!
a, vº +ý ta"c +. + E i. a
"r U1 L +ý CO a) "r- d-)
-v rn E CCvC
__. _ OU +J
ý l0 Y r- C) ý 3
r- >9 b co0
N aý "
", - aý :3
v r. ý 113 ei
¢ý+L-º OEO OOr-
.ýu '- ý
QJý C cn
,p . r.
C 0
au
rn c
. "- oýa
vý G1 u
+l . bf r t MI tL LL
(L) NN RS C) f0 7C1&-
04) ý. NUL". -
L () C) L #. A w3N
41 Vf 41 () ". - U Eb+. ) CCý vz ý". - c c, o"r-
r-o LacE "r" O º-ý "c tn. ". - . b. + Nf (1)
C)ý-~+ i=^Q1S.. Crý-" ýY
(1) 41
S-CD aciDCL
3WLýn
C. `" r- rt7 ý-. H 41 GNIb
C) C Q1 LA Q1 C) () tJCN C) .CLC i""ý (1) PW C) C) ++ C) to NE
3 r- C d-º r- 471 "r- 4- Y "C L () O. ]L +J O4--0 Ur- C) EO
O r- "r- C) iJ rC) 4- Ln U L C)L 3C". -0 VL O (ß. ä-4-) W"C ... to
4J C) "r- .-O C) i Cl WU b4.
Jý br-S Eºi-. "N
. r" C) O r- "r" O= H" to O -0 11 C. V- ýCCm
11 L II II 11 " II J 11 O ý. 1 W
GLL. J+1 . Ce z3
........ C
r- N c" ) ct tn tp m% Co r-
t! f 01
O
z
+ý " 0 .0 F- +ý Eý
C) 'O L (U du 4-j cý ýc
u " .. - a -v "c . r. . r. w yý
N R7 s 41 rn
ý M aý 3
r
.,.
v.. 0 L v
ý E 5 C
U
C
ý P-
^ r- b c . r.. t/)
O L
N
10 C (0
C. ý r N
ui
NZ
pw UU
T
U Q
_
w
N
V7 (i7 lJ U U ý ý
505
LA-
ui O _ O
p L. O
W U O IMi O n 1
ý
aý
Co c 4. rG
O
W
J
O
Z, -
Q S v W
0 lr. W d -1 Z N J
N v
C7 ý Z O Z- O Z m-
<c Li N
C S
W
21 W
J Z
N
ü _w LO
N LA � tI1 O i- Co 014
W
WI h- Q
a
,0i dp %_ ,/i
506
A. 4.2 Common`Variables
"variable" If a particular type of operation is carried out
on a component more than once (e. g. turning several
diameters, boring more than one diameter hole) then
the >
sign indicates that the sum of the
variable values for the operation must be input.
e. g. Where i= number of times a type of operation is I =n
>(Variable)
i-1
carried out, i. e. i=1,2,3,4 ... n.
DEPTHS OF CUT The difference between the initial and machined radii
of a component.
MACHINING TIME The time calculated for the machining operations
after inserting in the MATERIAL FACTOR equation.
It should not include UNLOAD/LOA& M/C MANIPULATION
or SET-UP times.
MATERIAL FACTOR See Table 30 for a full list of material groups.
NO. OF TURNED DIAMETERS
NO. OF FACING OPS
NO. OF DRILLED HOLES
NO. OF BORED HOLES.
NO. OF RECESSED HOLES
NO. OF THREADING OPS
The total number of each type of
operation carried out on a component.
507
MAX. A DIMENSION
MAX. B DIMENSION
MAX. C DIMENSION
Component rotating, tool stationary
r
fi B
Tool Axis
C A
Tool Axis
fi C
A B
pq
For round components the MAX. A DIMENSION and MAX. B DIMENSION
represent the diameter of the component.
Component stationary, tool rotating
For non-circular components the MAX. B DIMENSION is the maximum
dimension of the component at 900 from the MAX. A DIMENSION.
508
No. MAIN OPS Total number of major operations (i. e. Bore,
Face, Turn, Recess, Thread, Ream, Groove) to
be performed on the component.
(0) No "Operation" If the type of "operation" is not performed
on the component input 0.
(1) "Operation" If the type of "operation" is performed on
the component one or more times input 1.
"Operation" DIAMETER The machined diameter for the particular
"Operation".
"Operation" LENGTH The machined length for the particular
"operation".
TOOL TRAVEL LENGTH The difference between the maximum and minimum
radii of a facing operation.
TOTAL OPS The total number of operations carried out on
a component. This variable includes the
number of chamfers, circlip grooves and
tapping operations.
509
A. 4.3 Variables Applicable to Single Cost Centres
202 GROOVE
14-- 4J --ý
(a) Max. groove width =w
(b) Groove depth =d
(c) Groove Diameter =D
206 ROPE GROOVE
r ý.
This is a specific equation for grooving of wire rope drum tubes.
(Rope Diameter - 1.27mm) Diameter of the wire rope diameter
less 1.27mm
Groove Pitch =PP
N (Variable) This indicates that the variable
enclosed in the brackets should firstly
be multiplied by the number of grooves
machined (N).
510
A. 4.4 ILLUSTRATIONS OF WELDING POSITIONS
WELDING POSITIONS - BUTT JOINTS (Ref. (171))
Flat
9 / 8= 60; f= ý3211i; g= ý32 in,
9=5 0"; f= ; ýWa; g4? 16in.
Horizontal vertical
-1 990
9f ý fir4Tth _-r
- -- -e V/ I --
g=0 9= 1132 in.
ris in.
Vertical
--j-r- I32 In. ---"1 t'- ýI16 in.
L9j'J 9=60" 8_50"
Overhead
-'1 i'-9
g= O g= I/tsin.
E
511
WELDING POSITIONS - FILLET JOINTS
Ref. (171)
1
Horizontal vertical
Flat
A \VOOI*
E
B
,I 1
C
, 2ý 2-
0
12'ý N ýý ýý II
h ý-- --* LEG LENGTH
512
APPENDIX V
DATA FILE FOR THE
DESIGN COST ESTIMATING PROGRAM
513
ý, 1+. ý FILE FL Ei GýýItlr CCSi LSTII A'iIi"C fI"CC%Af"
1: ýEE'SIi. I. CAic. % 19-2-81 2; DATA FCF- CLSICI CGE"T ESTIt'ATIfýG FF<rrýr Ai"i. 3: TABLE DAJA FILE FOR DETERMINING GENERAL F"ACNIMItiG INPUT u" º, Iýý. tr ES- iV LE'iýý"t
It'IFI; T
7; DIAI. ETEk(Mý) ltf"GTH(týt, ) IvO-. I EA 1EAr, ý CODES
9: ECF E=1 10: 51 152 70 09203 1-; ý 92 305 70 0,201
76 152 1U. ̂. 1,258 13: 92 :, 05 J, 201 144 394 173 70 0,204 15: 680 305 7ü (j9202 16: 610 229 10r. " 0,258 17: ö36 305 10"' : 0,20"^_
18: 1778 610 10 l"i :0ý 210 19; 914 5639 100 0,206 20: Ei-FE=2
178 70 0,203 92 0 70 0. ý01 2..; b1
23; 76' 254 1OJ -" 10 25ö 92 201
25: 3014 178 70 09204
26: 68ý3 305 70 0,202
17: 610 279 10ý: 0,258
^o " 686 z05 10. ' 00 ý02 Gv ý
9 1329 762 10 0,210 30: 2134 5639 10ý: 0r206 31 32: TABLE 52 DATA FILE FOR PREDICTOR VARIABLE COEFFICIENTS
3t}: M/C CGh, S i Ah. 7ý F vý {-F f_-I" ACHTi .iG ý'ýI: f=ATIýraS----.... --.... ---.... - 35: CCCc AECCEA,:, 36: CC.: " T CEI. T I. c 248-Esl. f t"tIý li 37: 24301 1.07 0.1+. 0. il 0.;; 0";; C. 13 38 2+80' 3.86 0.10 0.0: 0.: 0.1 0.;;,; 392l+803 4. t]3 ()04 i. 0. J" = 00i «% O.: . 40 : 248U4 1.43 0.10 1.93 0. . 0.0
11: 24305 0.0 0. ü 0..: 42: 24306 0.0 0.0 0.1 0. ') 0. J 0. ýº u. 3: 24807 7.00 0.0"_ 3.::. C...; 4. J: U.
4-: 24aoýa 17. O;: 0.13_ }. 1; e. 3 ', 5: 24809 °. Ot? 0.;, ". 0. ü*_ 0.;. ü. ýJ O. JI
, 7: 3; Cýt. J? CEI""7Lc 23G-.: AS
`_. ," E0: 23c01 G J. 8 :.! - 51 - 52:? 3603 O.: J G. 0 U.: 0.: ý.:
J3: 23b04 O. ul 0. ý, 00; 0. 'J ") 0G U.. 34: 23605 0. t; 0. "3 J. l. ,. u 0.: 0..
: 23cüv ü. J U. V
. 0.. J.. r}.; J .: t]: Zýc07 (l.. , (C.
J J" J". ."! 'ý" ý 7:
LrJ8 ýi. 'ý l:. J: "- II ", I"fu".
ý 'ý.... at.. Uc' J. 'ý'. 'ý" ._
ti ". f. . , f.
"f ý" f
CODES
514
59: 60: 61: 62: 63: 23801 64: 23802 65: 23803 66: 23804
" ä7: 23E05 68: 23806 601: 23807 70; 23803 71: 23809 72: 73: 74: 75: - 76: 23901 77: 23902 78: 23903 79: 23904 80: 23905 81: 23906 82: 23907 83: 23908 84: 23909 85: 86: 87: 88: h /C 89: CCCE 90: 91: 201ý1
" 92: 20102 93: 20103 94: 20104 95: 20105 90: 20106 97: 20107 98: 20108 99: 21,101u9
100: 2C110 1p1: 201'_:. 1Ö2: 20112 1U3: 201.3 1Ö4: 2C1=4 1Ö5: 20115 106: 201.6 107: 20117 1ý8: ý0113 109: 2C119
.
112: 1'3: 114: 20201 1! 5: 20202 1, ä: 20293 1; "7.20`"'"4 1 L8 ;2 02 :, 5
CC57 CEt. 7E. E 238-SHEAF'
0.046 0. G 0.03 O. QU 0.01.0
0.311 0. U. " O. OJ 0.00 0.0: t:. 32 O. U O. C ýJ. O Q"0 0.0 0. ý
0.0 0"U 0.0 Q. 0 0.0 .: ri .00.
Q 0.0 O. U 0.0 Q. t; 0.0 Q. o 0.0 ö. o 0.0 Q. 0
0.0 0.3 0.0 0.0 0.0 0.0 0.0 0.0 04 0.0 0.0 u. 0
Ö. 0 0.0 0.0 0.0 0.0 0.0 e
CCST CEI-7F. t, 239-t"ELD
1.37 1.10 3.4tr 2.38 -4.07 0.:
1.74 1.77 3.57 3. G5 -3.54 0. ti 3.2c 2.5f`. 4.94 5.68 U. J 2.54 3.83 4.4+ 4. Lä -10.14 0. ") 0.63 0.58 0.9ri 1.05 -ý. 77 U"' 1.19 1: U6 1.01 3.14 -2620 0": I 0 . L_5 O. G: 3 U. G: 2 1.23 1.4 u' .2 Q. C2 1.23 1.4 '")"0
t:. v O. t,
CCF. ýTA!, T ý: FLt- GEt Et AL t'ACi-4ItIt"G UFEi. A T ILt ý-------------
A6. C0c: A. 1
CGST Ct! ". iVc 201-8AF CObßlt. l. TICt. S
0. ^13 0.042 -0.17 0.0 0.0 0.0
0.026 0.042 1.00 0.0 0.0 06C 0.0:. 3 0.28 0.00 0.0 0.0 0.0
0.03 -0.22 " 0.00 0.0 0.0 0.4 0,04 0.01. -0.71 0.0 0.7 0.0
O. 0 25 0.011 -0.25 0.0 0.0 0.0
0.072 0.0=]6 0.03 0.0 0.0 0.0
0.15- 0.0:: 0.011 0.0 0.0 0 . öZ.
0.0': O. UC 0.00 0.0 tJ: O 0.0
0.50 0.00 0.0J O. G 0.3 002-
0.0 0.0 0.0 0.0 0.0 "0.0 0.50 0.00 0.00 O. J 0.0 O. J.:
0,072 0.006 0.03 0.0 0.0 U. 0
O. Or_3 0.08 0. OC 0.0 000 U. ý
0.0 0.0.0.0 0.0 U. a c . 1.11 1.20 1.00 0.70 0.85 0.36 0. Vl:
0.0J, 1 O. UG1 0.0 . 1,; 0.0 0.0
0.43 0.0c, U. 00 0.0 0.0.0.4: 11.65 0. ü: 0"00 0.0 0.0 5. t3:
Cc.!, T CEr. Tc. E 202-ChI. cF IN" rLr. p1r; ATICr4-S
0.1ý3 U.. r. 41 -2. ý4 O. Ui) 0.147 O. Cý1 -5.77 0.013
-0.18 0.: )81 -0 . G8 U. J 0.0;; ý.; `ý2 L. C6S -ýJ. 92 O. J
0. "J. U. i 0.0 0": li . C: U.: 0. `i ; . '"? ü.. 0.:
515
`119: 202C6 0,049 120: 20207 0.095 12120208 0.50 122: 20209 0.1,47 123: 20210 1.00 124: 20211 0.0_ 125: 20212 0.50
'126: 20213 0.095 127: 20214 0.07 128: 20215 0.0_
'129: 20216 1.10 130: 20217 0.003 131: 20218 _O.
0C2 132: 20219 ý 14.08
'133:
0.037 -2.09 0.032 -1.81 0400 0.00 0.041 0.013 0.00 0.00 0.0 0.0 0.00 0.00 U. 037 -2.09 0.00 0.00 0.0 0.0 1.00 1.00 U. 016 -0.10
-0.609 0.95 0.00 0.00
0.0 0.0 0.0 0.0 0.0 0.0 0.00 0.0 0.0 0.0 1.00 Q. 0 0.14 0.0
134: '135: Ci: ST CEI: TE E 203-2Ar' CAFSTAI 5 136:.
_ 137: 20301 0. Q05 133: 20302 0.025 139: 20303 0.003
-'140: 20304 0.006
141: 20305 0.04. 142: 20306 Q. 037 143: 20307 0.06 144: 203a8 0.15 145. : 20309 0.00 14b: 20310 0:. 15
147: 20311 0.0 148: 20312 0.15
-149: 20313 0.06 150: 20314 Q. Ot; s
151: 20315 0.0_
' 152: 2Q31ö 1.20 _
153: 20317 0.021 154: 20318
_0.043 155: 20319 12.50
156 : 157: 158: i59:
0.004 0.12 0.00 0"004 -Q. 1E 0.00 0.28 O. OfI . 0.00 0.56 0.. 00 0.3; j 0.01 -0.76 0.001 1.20 O. OC 0. JG 1.47 0.01) 0.0 0.0t2 0 . Ot1 0. C0 0.00 0.03 0.01 0.00 0.013 0. t: : 0.0
o. Z)o 0. 1.47 0.01 0. Cr_ 0.08 0. Cü 0.; 1
0.0 G. 0 0.0 1.. 00 0.70 0.85 0.001 0.. 00 0.00 0.001 0.19 0.00 0.001 0 . 0.3 0.00
CCý7 CENTRE 204-CHI. CE ING CAFSTAt. S
160: 20401 0.043
161: 20402 0.056
162: 20403 0.001
163: 20404 0.028
164: 20405 0.032
165: 2046 0.037
166: 20407 0.06
-167: 20408 0.50
"1623: 20409 0.00
169: 20410 0.20
170: 20411 0.0.
171: 20412 0.50
172: 20413 0.06
173: 2041+ 0.00 '174: 20415 0.0
175: 20416 1.29.
176: 20417 0.001
177: 20`18 10.00ý. 178: 20L1°
0. Ots1 -0.02 0.00 (3.001 0.0 25 0. Qw 0.37 0.00 0.02 0.32 O. 00 0.0) Q. 065 -0.92 0.00 1.20 0.00 0. C' 1.47 0.0J 0.00 0.00 0.00 0.03 0.00 0.00 O. CO 0. UC 0.20 0.0 Q. ý 0. t) 0.0 0. O G0 . 0') 0.0 . '', 1.47 0000 0. CO 0.0 0. C 0.0 0.0 1.18 0.97 1.00 C'. lt36 C. 3i) 0. J::
-0.318 C. 29 0.1; t). L . U. l.. ý U. J 'ý
ö: 0 a'. ti 060 000 0.0 0.0 0.0 cý. 0 0.0 0.0 0.0 0.0 0.00.0i o. u 0.0 0.0 0. t`0 0.. o 0.0 0.36 040 Q. 0 1.2 0.0 -O. ES 0.0 0.49
o.. o Q. o 0.0 0.0 0.0 000 0.0 U"0: 0.0 Q. C3 0.0 U. ý 0.0 0. C 0.0 0.36 0.0 0.0 ö. u
010 Q. U WC 0.0 Q. 0 0.0 0.0 0.0" 0.0 O. OL 0.0 Q. o 010 0.0 0.0 a. 0y 0.13
-0.36 5. Ö
Q. 0 0.0 0"a 0.0 0.03 0. U 0. u 0.3 o. c O. U O. v 0. () J. C 0.0 0.0 0.56
0.: 0.. ß. )
0. u 0.0 0.0 0. L` 0.0 0.0 0.0 0.0 U. 0 0.0" 0.0 u. u: O. l, ý 0. J: 0.0 0. g 1.06
-U. 42 1;;. U
516
179: 180:
_182: - 183: 20601 184: 20002 185: 20603
, 136: 20604 187: 20605 188: 20606 189: 20607 190: 20608 191: 20609 192: 20610 193: 20611 194: 20612 195: 20613 196: 20614 197: 20615 198: 20616 199: 20617 200: 20618 201: 20619 202: 203: 204: 205: _ 206: 20901 207: 20902 208: 20903 209: 2090.4 210: 20905 211: 20906
'212: 20907 213: 20908 214: 209Q9 215: 20910 216: 20911 217: 20912 218: 20913 219: 20914 220: 20915 221: 20916 222: 20917 223: 20918 224: 20919
. 225" ý" 2^6"
227: 2L8i ----
229: 21001 230: 21002 231: 21003 232: 210J4 23:,: 21005 234: 210U6 235: 21007 236: 210t; 8 237: 210ý9 238: 21010
-239: 21011,
i
CCST CEh, 7 fiE 206-CEti TfE LATF: EE
0.03 0.486 -4.00 0.00 0.039 0.486 -4.42 0.00 0.009 1.88 0.00 0.00 0.028 3.94 0.00 0.00 0.031 0.087 -2.48 0.00 0.026 -0.005 1.14 0.00 0.06 0.041 -0.23 0.00 1.00 0.00 Q. 0C Q. OJ 0.32 -0.055 0.26 Q. 00 3.00 0.00 0.03 0.0
0.00 0.00 0.00 0.00 0.50 0.00
- 0.00 0.00
0.06 0.041 -0.23 0.00 0.13 0.05 -0.87 0.00 0.0 0.0,0.0 000 1.00 1.00_ 1.00 1.00 0.028 0.001 0.10 0.00 0.008 -0.003 1.64 0.15
10.00 0.00 0.00 0.00
CGST CEt. FE 209-F, CF'12(. NTAL ECF-E1ýS
0.125 0.025 -1.06 0.. 0 0.159 0.038 -4.59 0.0 0. Q17 2.93 0.0: 0.03 0.075 3.94 0.0ý Q. 00 0.130 Q. 142 -5.26 0.0 0.132 -0.007 -0.073 0.0 0.206 0.004 -0.025 0.0 1.00.0.00 0.09.. 0.00 0.000 0.000 0.000 0.0 5.00 0.00 0.00 0.00 Q. 0- 0.0
. 0.0 Q. O.
0.50 O. Uil_ O. G9 0.00 0.132 -0.007 -0.073 0.0 0.0 0.0 0.0 0.0 Q6 0_ 0.0 0.0 0.0 1.30 1.00 0.90 1.00 0.011 0.004 0001 Q. 00 0.030 -0.007 1.680 0.0
1gýa0 0.00 0.00 0.01
COST CENTRE 210-% ERTICAL E! 01'-. EFS
0.104 6.08 0000 O. OJ 0.186 5.66 0.00 Q. QU
0.047 -1.36 0.00 0.00
0.075 -2.46 0.00 0.0: )
0 . 027 Q. 044 0.23 Q. 0
0.065 0.025 -1.27 0.0 0.093 0.028 0.00 0.0 1.0u C. OJ J. Oü 0.0:
0.0 0.!. J 0.0 0.0 2.50 O. ýJ 0. ý:! U. V.: 0.0 0.0 0.0 0.0
0.0 0.0 0.0 o. c 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Q. 0 0.0 Q. o 0.0i 0.0 0.0 0.00 Q. o^ Q'. o 0.0 Q. 0 0.0 Q. o Q. o 0.06 0.0 Q. 0 -0.13 u. Q
_1.32 0.0 15.0
Q. U 0.0 0.0 0. c 0.00 0.0 0.0C0 0.0 0.0 0.0 0.00. U. 0 0.0 0.0
0.0.. 0.00 0"0. U"0. Q. CO 0.04 Qý-0. Q. 0
. 0.03 O. On
0.0 0.0 o. 0 0.0 0.0.0. a 0.50 0. v Q. 0 _4.23 0.0 -18. E2 0. OU 1.15
c. 3u .ýV
O. OJ 0.0;; 0.0 O. u 0.0
ý 0. � O. u G. 13 w O. ...
0.0 0.0 0.0 O. J 0.0 0.0 0.0 O. Ca 0.0 00L
o. e
517
240: 21012 0.50 0.00 0.00 0.00 0.00 0.00 241: 21013 0.186 -Q. 473 0.00 0.9) 0.0 0.0
242: 21014 0.0 0.00 0.0 0.0 0.0 0 .0 243: 21015 0.0_ 0.0 0.0 0.0 0.0 0.0
244: 21016 1.70 1.70 0.78 1.. 00 0.36 0.0
245: 21017 0.026 0.021 -0.21 0.0 " 0.0 -7.41 246: 21018 14 _0.260 _1.51
0.0 0.0 -1.28 247: 21019 30.000 15.000 15.00 0.0 0.0 0.00
248: `49: 250: .
CCST CENTRE 219-CCFY LATHES
, 251: 252: 21901 0.076 0, "00 0.. 00 0.00 0.. 0 -12.35 253: 21902 0.017 0.00 0.00 0.0 0.0 -2,. 38
254: 21903 0.002 0.00 0.. 00 0.0 0.0 1.29
255: 21904 0.0 0.0 0.0 0.0 Q. 0 0.0
256: 21905 0.0 0.0 0.0 0.. 0 0.0 0.0
257: 21906 0.0 0.0 0.0 0.0 0.0 0.0
258: 21907 O. C -
0.0 0.0 0.0 0.0 O. Q
259: 219Q8 0.10 0. Q0 0.00 0.0 0.0 0.0
260: 21909 0.00 04100 0.00 0.0 0.0 0.6
261: 21910 0.10 O. 0 t? 0.0 0.0 0.0
262: 21911 0.10 0.0+l 0.00 0.0 0.0 0.0
263: 21912' 0.0 0.0 0.0 0.0 0.0 0.0
264: 21913 0.0 0.0 0.0 0.0 0.0 0.0
265: 21914 0.0 0"0 0.0 0.0 0.0 0.0
. 266: 21915 0.0 0.0 0.0 0.0 0.0 0.0
267: 21916 1.15 0.50 0.50 0.50 0.40 0.0
268: 21917 0.005 . 0.00 0.00 0.00 0.00 Q. 48
269: 21918 0.002 Ö. Oi3 0.00 0.00 0.00 0.33
270: 21919 9.36 0.00 0. CO 0.00 0.00 5893
271: 272. '273 % CC-ST CENTF: E 258-CHI. FCHILL 420 KC LATF-E
274: -- --- . 275: 25801 0.012 0.007 -1.. 87 0.00 0.. 00 0.0
276: 25802 0.018 . 0.000 -0.90 0.00 0000 O. U
277: 25803 0.012 -0.20 0.00 Q. 00 0.. 00 0.0
278: 25804 Q. 014 0.05 Q. 0 0 Q. 00 Q. 00 0.0
274: 25805 0.005 0.015 . 0.18 0.00 0.00 0.0
230: 25806 0.001 0.017 -0.42 0.00 0.00 0.0
281: 25807 0.004 0.022 -0.20 0.00 0. OC 0.0.
282: 25808 0.30.0.00 0.00 0.00 0.00 0. Q0
283: 25809 0.200 0.00 0.00 Q. 00 0.00 0.0
284: 25810 0.50 0.00 0.00 0.00 0.00 0.00
285: 25811 0.0_ O. U 0.0 00 0.0 0.0
286: 25812 0.20 0.00^ 0.00 0.. 00 0, -00 Q. Oü
287: 25313 0.004 O. U2ý -Q. ýO U. 00 0 00
288: 25814 0.008 0.08 0.00 0.00 0.0
289: 25815 Q. U_ 0.0 Q. Q .
0.0,0.0 0.0
290: 25816 1.50 1.00 1.0 0 1.00 0.6 0.0
291: 25817 042 -0.021 046 0.00 0.02 -0.79
292: 25818 -_0.0f)9 0.00
_0.00 0. a; 0.20
293: 25819 15.00 15. OfJ 15.013 1Ö. 0'ý 0.0:; lO. tl
294: 295 : 296:
COST CEt. TF-E 258-BATCHMATIC tic MACF+It. E
297' O. C12 0.03 O. UO O. Or 0.40 29E: 25301 0"0ýý1
299: 25802 0 . 0; J, 2 0.02 0.10 0.00 0.4J 0.79 0. G+.? 3 U. J ` 0.33
30(): 25803
518
301: 25804 Q. 0Q6 302: 25805 0.019 303: 25806 O. 0C6 304: 25807 0.017 305: 25808 0,615 306: 25809 1.00 307: 25810 0440 308: 25811 O. OQ 309: 25812 Q.. 2Q 310: 25813 Q. 017 311: 25814 0.008 312: 25815 1.00_, 313: 25816 0,. 011 314: 25817 r Q. 0U5 315: 258113
. 0001
316: 25819 20.00 317: 318: 319: 320: 321: 322:. ____
0.00 0.00 0.00 O. UO 6.49 0.034 0.. 29 0.00 9000 Q. Q 0.007 0.53 0.00 Q. 00 Q. U 0.012 1.30 0.01 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.00 O. UO Q. 00 . Q. CU Q. Q 0.0 0.0 0.0 0.0 (ý. 0 0ý. 0 0.0 0.0 0.0 0.0 0.0.
- Q. 0_ 0.0- 0.0 0. -0 0.012 1.30 0.00 0.00 0.0 0.08 0.00 0.00 0.00 Q. 0 1.00.1.00 1.00' 0.6 0.0. 0.002 0.28 Q. QO 0.0. -U. 57 0.002 0.00 0.00 O. QU Q,. 10 =0. Q06 0.15
_Q. 00 0.0 -Q. 49 15.00 15.00 10.00 0.00 10.0
CGP. STAI TS FOP SFECIAL h'ACF'IPlING GFFkATIGt`'.
323: 21701 0.101 324: 21702 0.05 325: 21703 0.606 326: 21704 0.211 327: '21705 0.394 328: 21706 0,. 345 329: 21707 0.0 330: 21708 0.15 . 331: 21709 0.47 332: 333: ' 334:. 335:. _. _ 336: 21401 0.019 337: 21402 0.0 338: 21403 0.0 339: 21404 0.0 340: 21405 Q. 0 341: 21406 0.0 342: 21407 0.0_.. 343: 21408 O. C14 344: 21709 0.30 345:. 346: 347:. 348: 349: 21501 350: 21502 351: 21503 352: 21504
-, 353: 21505 354: 21506 355: 21507 356: 21508 357: 21509 353: 359:.... -=60:, ' .:,
0.0 O. G 0.0 0.0 0.0 0.0 0.0 0.0 0.0
COST CEh1kE 217-GEAr
0.00 Q. 00 O.. OC 0.0G O. OJ 0.00 0.00 Q"ýü t). 0() O. OQ 0.0C 0.00 0. t) 0.0 0.20 0.400 3.07 0.00
CCST CENTRE 214-MILLING
0.029 IN -1.03 0.. 0 0.0 0.0
0.0 NO 0.0 0.0 0.0 0,. 0 0.. 0 O. Q Q, 0 O. 0 0.0 Q. 0 0.0 000.0.0
-0.003 0.00 Q. 00 0.15 0.00 0.00
CCST CEFaTAE -
0.. 0 O. a 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Q. 0 0.0 0.0 0.0
Cl: 7iII"G
Q. QU 0.40 14'. 5 Q. co 0.00 12.1 0.0i- Q,. OJ -36.0 0.00 O. 00 13.5 Ö. oc 0.00 2.38 0.00 0*. 60
-19.9 0.0 Q. 0 0.0 Q. oJ 0.00
, 0.0 0.00 0.00 114.55
Q0 Q. Ou 0.0 0.0 Qý0 Q. 0 0"0 0.0 0. Q u. 0 Q. 0 0.0 000- 0.0 0.00 -1.24 0.0v 0.0
ö: ö 0.0 U. 0 0.0 0.0 O. 0 U. 0 0.0 0.0 O. J 0.0 U 0.0 0.0 0.0 0.0 0.0 0.0
COSY CENTRE 216C4 - Ir. CIr :G tMUCf ICAI, )
519
361: _ 362: 21601 363: 21602 364: 21603
-365: 21604
366: 21605 367: 21606
. 36 6: 21607 369: 21608 370: 21609 371: 372: 373: 374: -- 37521601 37621602 377: 21603 378: 21604 379: 21605 380: 21606 381: 21607 332: 21608 383: 21609 354: 385: 386: 387: 383: 27001 389: 27002 390: 27003 391: 27004 392: 270C5 393: 27006 -394: 27007 395: 27008 396: 27009 397: 398: 399: 400:: 40121101 40221102- 403 21103 40421104 40521105 4 0621106 407 21107 408 21108 409: 21109 410= 413' 412ý 413:: 41'4- 415s22301 416: 22301 417: 2303
j+1g: 2n304 419.2305 420: 2_^_306
j=21: 22307
/
0,034 0.013 0.0 0.0 0.0 0.0 0.0 0.0 Oro 0.. 0 0.0
. 0.0
0.024 0.002 0.. 060 0.. 00,2
20.00 0.00
-0.62 0.0 0.0 0.0 0.0 0.0 0.00 0.00 0.00
O. CO O. CO 0.0 0.0 0.0 0.0 0.0 0. v 0. u 0.0 0.0 0.0 0.0 " 0.0 C. 0 0.0 0.0 0. Q 0.0C 0.00 -0.13 0.00 0.00 -1.90 0.00 0.00 O. C
C(; ST CENTRE 216-GF: IPýDIP. G (SLRFACE)
0,. 032 0.00 0.005 a. 00 Q. 0 0.0 Q. 0 0.. 0 0.0 0. C 0.0_ 0"0.. 0.016 0.0u 0.001 0100
20.00 10.00
CCST CEVTRE
0.047 0.00 0.0 0. "0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0_ 0.00 0.00 1.20 1.. 00
25.00 0.00
CG5T CEt: TE"E
0.008 U. 0ü8 0.025 1.2ý 0.20 0.00 0.00 0.00 0.0 0.0 0.. 0
._0.0 -O. OC1 0.007
O.. QC4 U. 0C2 25.00 0.00
o. oo öý. ö0 0.0 0.0 ö. ö 0. ci^ 0. öJ ö. o º Ö. 0C
270-SPLINE CI, T i II-jG
0.. 00 O. OJ 6.65 -0"16 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0. C 0.0 0.0 0.0 0.0 0.00 0.00 0.0v Q. 0 1.00 1. Ot) 0.5 0.0 0.00 0.00 0.0 0.0
211-PADIA. L Cr-ILLS
(ý". OÜ0. GQ -2'. 4 0.00 Q. 00 0.32 Q. 0 U"0 0.0 o. Q Q. 0 0.0 0.0 0.0 0.0 O. o
. 0.0_ 04
0.00 0.09 -1.2 4.00 0.00 0.0a 0 . 0'0 0.00 0.0
0.33 0.00 0.00 0.0 0.. 00 0.00 0.00 0.0 0.00 0.00 0.00 0.0 0.00 0.00 0.00 0.0 0.0 0.0 0.0 0.0 0.0.. ü. ü 0.0 0.0 0.00 0. *00 o. OU 1.11 0.07 0.00 0"J0 0.16 O. OC 0.00 0. CO 0.0
------- CCheSTAh: TS FQF% F CST, MAChININC CFEF"A T ICN:: ---------
CCS'4 CELTFE 0.0 0.0 Q. 0 Q. Q
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0u 0.0 0.0
ö. 0 Q. 0 OA ý. ý o. 0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0. J 0. ýJ 0.0 O. U 0.0 U. 0 0. 'ý O. p U. J 0.0 0.0 0. la 3 . r)
422: 22306 423: 22304 424: 22310 425: 22311 426: 22312 427: 22313 428: 22314
ý429: 2ý315 430: 22316 431: 22317 432: 22318
. 433: 22319 434: 435:
_ -436: 22701 437: 22702 438: 22703 439: 227Q4 440: 22705 441922706 442: 22707 443: 22708 444: 2709 445: 22710 446: 22-711 447: 22712
'448: 22713 449: 2714 450: 22715
. -451922716 452: 22717 453: 22718 454: 22719 466: 467: 468: 469: 470: 471: 472: 473: 474:
. 475: 476: 477:
-478: 479: 480: 431: 482: 483: 484:
. 48. ý. 486: 487: 488: 489: 490: 491: 492: 493:
ýe
520 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0. .0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
_, _ Cu;; T CENTF: E
0.012 0.045 0 . 104 0.006, O. 0J7 0.043 0.052 0.0C4 0.018 0.005 0.005 0.024 2.50 5.00 1.50 1.50 0.50 1.00 1.00 1.50 0.0 0.. 0 0.0 0.0 0.0 0.0 0.0 O. J 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0,. 0 0.0 0.0 0.0 0.0 o. o a. 0 O. U 0.0 0.0 0.0 0.. 0 0.0 0.0 0.0 0.0 0.0
227-_ASI. EN"BlY 1.16 0.00 0.52 O. Q0 1.42 Q. 00 1. OS ' 0.00 5.03 Q. 00 5.42 0.00 3.00 0.00 1.50 Q. 00 1.00 0.00 2.. ý0 Q. 00 0.0 0.0 0.0 0.0 0,. il 0.0 0.0 0.00 0.0 0.0 0.0 0.0 060 0.0 0.0 0.0 0.0 0.0
TABLE 53 DATA FILE FOR LABOUR AND OVERHEAD COST RATES
CEItiTcF', 248 236 238 239 2Q1 202 203 204 206 209 210 219 258 214 216 215 270 211 217 227
ýwI tl'-L- t, AItZ: (GACFE) 1-5-81) LA6ül, F. (. ý/i-fý ) ü%EF HEADS ADMIt±.
2.55 7.. 70 1.2Q 1.30 9.25 1.. 20 2.60 8.00 1.2Q 2.60 9.25 1.20 2.60 9.25 1.20 2.60 9.25 1.20 2.60 9.25 1.20 2.60 9.25 J. 2Q 2.60 9.25 1.20 ý. 60 9.25 1.20 2.60 9.25 1.20 2.60 9.25 1.20 2.60 9.25 1.20 2.60 9.25 1.20 2.60 9.25 1.20 2.60 9.25 1.20 2.60 9.25 1.20 2.60 9.25 1.20 0.85 8.00 1.20 2.25 5.90 1.20
u. c 6.6 0.0 000 0.01 0.0 0.0 000 000 0.. 0 000 0.0 0.0 0.0 000 0.0 0.0 0.0 0.0 0.0 0.0 040 000 0.0
O. OU 000 0.00 ý. p 0.00 0.0 Q"QJ Q. 0 Q. 00 000 O. OC 0.0 0.00 q"0 0.. 00 Q. 0 0.00 0.0 0.00 0. Ü 0.0 0.. 0 U. 0 0.0 0.0 ö. q 0.0 0.0 0.0 Q. 0 0.0 0.0 0.0 0.0 0.. 0 0.0 0.0 0.0
494: ///, ' y 495: º -' ''
521
APPENDIX VI
MODIFYING THE REGRESSION PROCEDURE
522
To reduce proportional estimating errors the conventional
regression procedure can be modified to minimize the sum of the
squares of the residual proportions (Figure 66). The solution to
the multiple linear regression procedure for two independent
predictor variables then becomes:
i (i) The equation that minimizes the sum of the squares
of the residual proportions is obtained by determining
the values of A, B and C that minimize Equation 48.
i=N
(Pi)2 =
i=ý
Where:
Pi
Yi
1=ci
i=1
(Yi -A- BX1i- CX21 )2 (48)
Yi
= The proportional difference between the ith
estimated (i. e. using the expression
A+ BXl + CX2) and observed values of the
dependent variable Y.
= ith observed value of the dependent variable Y.
N= Number of observations.
Xli, X2i = ith values of the predictor variables Xl and X2.
A, B, C= The coefficients of the estimating equation that
minimizes (Pi)?
523
FIGURE 66 THE MODIFIED REGRESSION OBJECTIVE
aý. ý .0 ý (0
ý
v
ý aýI 0
Predictor Variable (X)
Objective of the Modified Regression Technique
i=m
Minimize ý Pi
Yi i=1 (')
i=1,2,3, ... m
524
(ii) Transforming Equation 48 and dropping the subscripts to
simplify the notation the expression becomes:.
yp2 = (1 - AY-1 - BX1Y-1 - CX2Y-1)2 (49)
(iii) The minimization of P2 is accomplished by the partial
differentiation of Equation 49 with respect to A, B and
C and allowing the partial derivatives to equal zero,
i. e. Equations 50 to 52.
aFP 2=
-2 Y-1 (1 - AY-1 - BX1Y-1-CX2Y-1) =0 (50) aA
82P 2= -2 X1Y-1 (1 - AY-1 - BX1Y-1 - CX2Y-1) =0 (51)
aB
ayP2 = -2 X2Y-1 (1 - AY-1 - BX1Y-1 - CXZY-1) =0 (52) ac
(iv) Dividing Equations 50 to 52 by -2, applying the summation
operator (2: ) to the terms inside the brackets and replacing
the subscripts a set of simultaneous equations are formed
the solution of which yields the values of A, B and C that
minimize the sum of the squares of the residual proportions,
i. e. Equations 53 to 55.
525
i=N i=N i=N i=N
Yi-1 =A Tyi-2
+B EXliyi-2
+C X2iYi_2 (53) i=1 i=1 i=1 i=1 ý'
i=N i=N i=N i=N T-Xl
iY i-1 = ALXl iY i-2 +B Xl i 2Yi-2
+C XliX2iY 1-2 (54)
i=1 ,e i=1 i=1 i=1
i=N i=N i-N i=N
X2iYi-1 =A ZX2iYi-2
+B2: X1iX2iYi-2 +C2: X21 2Y1-2 (55)
i=1 i=1 i=1 i=1
To test the effectiveness of modifying the regression
procedure in the above manner the estimating accuracy
obtained using the modified regression equations was
compared with that of. equations developed using the
conventional regression procedure.
The results of this comparison (Table 54) indicated that:
(i) In general the modified regression procedure improves
the average estimating accuracy (when measured using
the Average Proportional Error) although small
reductions in average estimating accuracy can
occur (e. g. equations for estimating gear cutting
times for T>65, M<3; T>65, M>3<4 and T<65, M<3).
526
(ii) The modified regression procedure significantly
improves the variation in proportional estimating
errors.
(iii) The minimization of absolute residual values by the
conventional regression technique effectively results
in high proportional errors arising at the low end
of the dependent variable range. Hence the use
of the modified regression technique effectively
improves the estimating accuracy of estimating
equations in this range.
(iii) By improving the estimating accuracy for low'values
of the dependent variable the modified regression
procedure helps to avoid negative time standards
being estimated in this area when negative predictor
variable coefficients are present in the estimating
equations.
t i
527
TABLE 54 COMPARISON OF THE ACCURACY OF CONVENTIONAL AND MODIFIED REGRESSION PROCEDURES
CONVENTIONAL MODIFIED
DEPENDENT VARIABLE REGRESSION PROCEDURE REGRESSION PROCEDURE
ESES loe
Gear Cutting Times (T>65, P1<3) -0.050 0.255 0.066 0.190
It ' 11 It (T>65, P9>3<4) -0.004 0.133 -0.018 0.129
(T>65, M>4) 0.066 0.108 -0.012 0.098
'(T<65, M>3) 0.030 0.202' -0.036 0.188 11 11 11
(T<65, M>3) 0.019 0.143 0.012 0.133
Sawing Times (Bar) 0.158 0.416 -0.090 0.285
Facing Times (Bar Lathes) 0.718 1.348 -0.272 0.446
11 11 (Vertical Borers) -0.112 1.251 -0.177 0.462
T= number of gear teeth
M= module size
Consideration of Manufacturing Volume
The individual time standard observations used to develop estimating
equations using the regression technique are normally associated with
various types of components. Since the volume of individual components
manufactured within a company can vary considerably it would be useful
528
to weight the accuracy of an estimating equation with respect to
component volumes, i. e. the higher the component volumes manufactured
the greater the accuracy of the time standard estimates obtained from
a regression model.
In order to achieve this objective it is again possible to modify
the conventional regression procedure to take account of the relative
volumes of the components used to develop the estimating models.
Equation 48 would need to contain a term for the volume of components
manufactured as shown in Equation 56.
(P; )2
Where:
_2
v1 Yi -A- BX, i - CX2i
Yi (ý 56
Vi = The volume manufactured of the component that
, represents the ith observation of Y.
The solution to Equation 56 is carried out in an identical manner
to that of Equation 48, i. e. a set of simultaneous equations (Equations
57 to 59) are obtained the solution to which yields the required values
of predictor variable coefficients.
529
ý1Y1-1 = A\ v1Y1-2 +Bý V1X11Y1-2 +C ýv1X2'Y1-2 (57)
ýiX1iYi-1 - AT ýiX1iYi-2 +B LviXli2Yi-2 + CL ViX1iX21Y1-2 (58)
ý ViX2iYi_1 - A7 ViX2iYi_2 + B) viXliX2iYi_2 + CT ViX2i2Yi-2 (59)
r
Although modifying the regression procedure in this manner would
enable the estimating accuracy of high volume components to be improved,
estimating equations would need to be redeveloped if changes in the
relative volumes of various components 'occurred.
![53rd NCAA Wrestling Tournament 1983 3/10/1983 to … 1983.pdf · 53rd NCAA Wrestling Tournament 1983 3/10/1983 to 3/12/1983 at Oklahoma City ... Jan Michaels [US] ... Barry Davis](https://img.pdfslide.net/doc/110x75/5acd84bf7f8b9a93268db5de/53rd-ncaa-wrestling-tournament-1983-3101983-to-1983pdf53rd-ncaa-wrestling.jpg)
