Product Cost Estimation Theoretical Development and Industrial Validation PDF

Product and Manufacturing Cost Estimation: Theoretical Development and Industrial Validation
By
Adnan Niazi
This thesis is submitted to the Department of Mechanical Engineering, School of Physical Sciences and Engineering, Kings College London, University of London, for the Degree of Doctor of Philosophy.
To
My Mum and Dad My Wife

and
My Children
Abstract
Product Cost Estimation (PCE) deals with predicting the cost of a product before it is manufactured. Due to market competitiveness, there is a need to predict product costs early and accurately. However, the available methods of cost estimation compromise accuracy in an attempt to deliver early results. Conversely, the accuracy can only be fully achieved once the design and process planning details are available by which time cost estimation will be too late. The main aim of the work is to develop a methodology for early and accurate estimation of a products cost without relying on design and process planning details.
The main contributions of the thesis are as follows. First, following a comprehensive literature review of the available techniques, an extensive hierarchical classification system is developed. The classification is based on categorizing the techniques into qualitative and quantitative with further subdivisions down to four levels. The developed system identifies that qualitative techniques deliver early results and quantitative techniques are known for accuracy. The review also identifies that overall accuracy greatly depends on accurate estimation of overheads. Secondly, based on the feedback from the review and the proposed classification system, a new method of overhead estimation based on separating time- and material-dependent overheads is developed with improved accuracy. Thirdly, a comprehensive mathematical model based on combining the attributes of early and accurate estimation from the qualitative and quantitative techniques is developed and called a Hybrid Model. The developed model is optimized through the introduction of the cost deviation indices. Fourthly, the deviation indices are modelled considering past product cost details, inflation and other deviations in order to predict future costs early and accurately without requiring product design and process planning details.
Abstract
The developed models are validated by industrial trials on two distinct global locations adding further towards understanding the implications of geographical locations on aspects of cost control.
Table of Contents
Abstract ........................................................................................................................ 3 Table of Contents.......................................................................................................... 5 List of Figures............................................................................................................. 11 List of Tables.............................................................................................................. 14 List of Notations ......................................................................................................... 16 List of Acronyms ........................................................................................................ 22 Acknowledgements..................................................................................................... 24 C hapt e r 1 1.1 1.2 1.3 1.4 1.5 Introduction ......................................................................................... 27
Overview..................................................................................................... 27 Problems identification................................................................................ 29 Aims and objectives .................................................................................... 33 Thesis structure ........................................................................................... 35 Conclusions................................................................................................. 38 Background and Related Work ............................................................ 39
C hapt e r 2 2.1 2.2 2.3 2.4 2.5
Introduction................................................................................................. 39 Manufacturing and cost control ................................................................... 41 Cost estimation............................................................................................ 45 Cost estimation in the early design stages .................................................... 47 Cost estimation for specific applications...................................................... 48 Cost estimation for a specific segment in a production cycle................ 49 Cost estimation for specific machining and manufacturing processes... 50
2.5.1 2.5.2
Table of Contents
2.5.3 2.5.4 2.6
Cost forecasting for specific parts and products ................................... 50 Cost estimation for generic systems ..................................................... 51
Conclusions................................................................................................. 53 PCE Technique Classification System ................................................. 55
C hapt e r 3 3.1 3.2 3.3
Introduction................................................................................................. 55 Development of hierarchical classification system (HCS)............................ 57 Intuitive cost estimation techniques ............................................................. 60 Case-based methodology ..................................................................... 60 Decision support systems (DSS) .......................................................... 63
3.3.1 3.3.2
Rule-based systems ......................................................................................... 65 Fuzzy logic approach....................................................................................... 68 Expert systems ................................................................................................ 69 3.4 Analogical cost estimation techniques ......................................................... 70 Regression analysis models.................................................................. 70 Back-propagation neural network (BPNN) models............................... 70
3.4.1 3.4.2 3.5 3.6
Parametric cost estimation techniques.......................................................... 71 Analytical cost estimation techniques .......................................................... 73 Operation based approach .................................................................... 73 Breakdown approach ........................................................................... 75 Tolerance-based cost models ............................................................... 76 Feature-based cost estimation .............................................................. 78 Activity-based costing (ABC) system .................................................. 79
3.6.1 3.6.2 3.6.3 3.6.4 3.6.5 3.7
Conclusions................................................................................................. 81 MRO and TRO Estimation Methods .................................................... 87
C hapt e r 4 4.1 4.2
Introduction................................................................................................. 87 Cost estimation methodology at the selected company................................. 91 6
Table of Contents
4.2.1 4.2.2 4.2.3 4.3
Material cost estimation ....................................................................... 92 Direct labour costs ............................................................................... 93 Overheads estimation........................................................................... 93
Proposed methodology for overheads estimation ......................................... 95 MRO estimation model........................................................................ 95 TRO estimation model......................................................................... 98
4.3.1 4.3.2 4.4 4.5
Model implementation and validation........................................................ 101 Conclusions............................................................................................... 112 PCE Hybrid Model ............................................................................ 114
C hapt e r 5 5.1 5.2 5.3
Introduction............................................................................................... 114 Product cost and modelling approach......................................................... 116 Direct cost elements .................................................................................. 122 Direct material costs .......................................................................... 123 Direct labour...................................................................................... 126
5.3.1 5.3.2
Labour units .................................................................................................. 128 Labour rate.................................................................................................... 130 5.4 Indirect cost elements ................................................................................ 132 Processing cost .................................................................................. 132
5.4.1
Processing units............................................................................................. 133 Processing rate .............................................................................................. 134 5.4.2 5.4.3 Material dependent cost ..................................................................... 135 Tooling cost....................................................................................... 136
Machine tool rate........................................................................................... 137 Labour tool rate............................................................................................. 138 5.4.4 Building space cost ............................................................................ 139
Building space rate........................................................................................ 140 7
Table of Contents
Manufacturing space ..................................................................................... 141 5.5 5.6 Production overheads ................................................................................ 142 Conclusions............................................................................................... 143 Industrial Implementation and Analysis of the PCE Hybrid Model .... 145
C hapt e r 6 6.1 6.2 6.3
Introduction............................................................................................... 145 HMI algorithm and the implementation methodology................................ 148 PCE at the company .................................................................................. 157 Information and details ...................................................................... 157 Material cost estimation ..................................................................... 160 Labour cost estimation....................................................................... 163 Overhead estimation .......................................................................... 165
6.3.1 6.3.2 6.3.3 6.3.4 6.4
Implementation of the PCE Hybrid Model................................................. 168 Material cost estimation ..................................................................... 169 Labour cost estimation....................................................................... 171 Processing cost estimation ................................................................. 173 MDC estimation ................................................................................ 175 Production overheads estimation........................................................ 177
6.4.1 6.4.2 6.4.3 6.4.4 6.4.5 6.5
Conclusions............................................................................................... 179 Comparisons and Validation Analysis................................................ 181
C hapt e r 7 7.1 7.2
Introduction............................................................................................... 181 Preparations for comparisons..................................................................... 183 Labour and machine cost estimation .................................................. 185 Material cost and factory expenses..................................................... 187
7.2.1 7.2.2 7.3 7.4 7.5
Comparison analysis for product cost ........................................................ 189 Comparison analysis for cumulative costs ................................................. 202 Comparison analysis for product and production overheads....................... 206 8
Table of Contents
7.6 7.7
Cost breakdown analysis ........................................................................... 212 Conclusions............................................................................................... 223 Conclusions ....................................................................................... 225
C hapt e r 8 8.1 8.2
Summary................................................................................................... 225 Contributions............................................................................................. 231 Development of a technique classification system.............................. 231 Development of a decision support model (DSM).............................. 231 Development of time- and material-based overhead estimation methodology...................................................................................... 232
8.2.1 8.2.2 8.2.3
8.2.4 8.2.5 8.2.6 8.2.7 8.2.8 8.3 8.4
Development of a PCE methodology for batch production................. 233 Development of cost deviation indices............................................... 234 Development of HMI algorithm and industrial implementation.......... 234 Comparison and validation ................................................................ 235 By products..................................................................................... 235
Current Trends and Future Work ............................................................... 236 Concluding remarks .................................................................................. 239
Publications Arising from the PhD Study.................................................................. 241 References ................................................................................................................ 243 Appendix A A.1 A.2 Bill of Material (BOM)...................................................................... 261
Introduction............................................................................................... 261 BOM for Hammer Drill ............................................................................. 262 Deviation Indices............................................................................... 274
Appendix B B.1 B.2 B.3 B.4
Material cost deviation index..................................................................... 274 Labour cost deviation index....................................................................... 276 Processing cost deviation index ................................................................. 277 MDC deviation index ................................................................................ 277 9
Table of Contents
B.5 B.6 B.7
Tool cost deviation index........................................................................... 278 Building cost deviation index .................................................................... 279 PO deviation index .................................................................................... 279
10
List of Figures
Figure 1-1: Main area of research................................................................................ 29 Figure 2-1: Relationship between productivity and flexibility ..................................... 41 Figure 2-2: Flexibility and productivity of different manufacturing systems................ 44 Figure 3-1: PCE Preliminary Technique Classification................................................ 59 Figure 3-2: Flow Diagram of the Case-Based Approach for Cost Estimation............... 61 Figure 3-3: Decision support system approach to cost estimation ................................ 64 Figure 3-4 Cost Estimation Process Model Based On User Constraints ....................... 66 Figure 3-5: Classification of the PCE Techniques ....................................................... 84 Figure 3-6: Decision Support Model for cost estimation methodology selection.......... 86 Figure 4-1: Break down of Selling Price and Manufacturing costs .............................. 89 Figure 4-2: Cost estimation results for 25, 100, 200, 500 and 1000kVA transformers............................................................................................ 104 Figure 4-3: Cost element values in DT and PT .......................................................... 106 Figure 4-4: TRO and MRO values breakdown for DT and PT................................... 107 Figure 4-5: Cost trends in (a) DT and (b) PT............................................................. 109 Figure 4-6: Cost trends comparison in DT and PT..................................................... 110 Figure 5-1: Pictorial representation of the mathematical model for product cost estimation ............................................................................................... 119 Figure 5-2: Development of the Hybrid Model within the framework of the technique classification system ............................................................... 125 Figure 5-3: Types of work centres............................................................................. 126 Figure 6-1: PCE Hybrid Model implementation phase .............................................. 150
11
List of Figures
Figure 6-2: Material cost estimation implementation process .................................... 151 Figure 6-3: Labour cost estimation implementation process ...................................... 153 Figure 6-4: Processing cost estimation implementation process................................. 154 Figure 6-5: MDC estimation implementation process................................................ 155 Figure 6-6: Manufacturing cost estimation implementation process .......................... 155 Figure 6-7: PO estimation implementation process ................................................... 156 Figure 7-1: The comparison of the actual costs against the estimated costs................ 190 Figure 7-2: Estimation error trends for the three methods (2003 2005) ................... 192 Figure 7-3: Percentage cost estimation variations from actual product costs. ............. 194 Figure 7-4: Estimation error trend across the product range....................................... 196 Figure 7-5: Error linearization for the results given by the companys method .......... 198 Figure 7-6: Error linearization for the results given by the Jungs method ................. 199 Figure 7-7: Error linearization for the results given by the Hybrid Model.................. 200 Figure 7-8: (a) Cumulative actual costs against total estimated values, (b) estimation errors for cumulative costs..................................................... 203 Figure 7-9: Optimization for the estimation accuracy achieved by the Hybrid Model against (a) the companys method (b) the Jungs model ............... 205 Figure 7-10: Estimation error trends for overheads using the three methods (2003 2005) ......................................................................................... 208 Figure 7-11: Overheads estimation analysis for the cumulative values ...................... 209 Figure 7-12: Optimization achieved for overhead estimation based on (a) the Companys method (b) the Jungs model ................................................ 211 Figure 7-13: Production cost (actual) break down analysis (2002 2005) presented in values and percentage ......................................................... 213 Figure 7-14: Production cost elements trends ............................................................ 214 Figure 7-15: (a) Production and manufacturing costs (b) elemental costs effect on production and manufacturing costs ................................................... 216
12
List of Figures
Figure 7-16: Overheads break down analysis (2002 2005) presented in values and percentage........................................................................................ 217 Figure 7-17 Overhead elements trends (2002 2005) ............................................... 218 Figure 7-18: Production overheads breakdown (2002 2005)................................... 219 Figure 7-19: Production overhead elements trends analysis....................................... 220 Figure 7-20: Processing cost elements trends ............................................................ 221 Figure 7-21: MDC elements trends ........................................................................... 222 Figure A-1 Hammer drill .......................................................................................... 262 Figure A-2 Product structure (Hammer Drill)............................................................ 264 Figure A-3 Dismantled drill with assemblies and sub-assemblies.............................. 265 Figure A-4 Parts and components in product structure .............................................. 266 Figure A-5 Winding (Stator and rotor) and drive assembly ....................................... 266 Figure A-6 Driven assembly (with gear and shock absorber) and Drill/Hammer switch ..................................................................................................... 267 Figure A-7 Trigger assembly .................................................................................... 267
13
List of Tables
Table 2-1: Summary of the published literature references for various applications of PCE .................................................................................. 53 Table 3-1: The PCE techniques; key advantages, limitations and list of discussed references.................................................................................. 85 Table 4-1: MRO for power and distribution transformers (2003-2004)........................ 97 Table 4-2: Budgeted time related overhead rate calculation (2003-2004)..................... 99 Table 4-3: Summary of TRO rates, MRO percentage fractions and overhead rates for 4 years ...................................................................................... 100 Table 4-4: Cost estimation for 25kVA transformer.................................................... 102 Table 6-1: Industrial output values for the CMD (2002 2005) ............................... 158 Table 6-2: Product range at the SPSD ....................................................................... 159 Table 6-3: Yearly production quantities for the product range in the SPSD (2001 2005) and Actual unit product cost for the given product range (2002 2005)................................................................................ 160 Table 6-4: Actual material cost (Cumulative and per unit costs)................................ 161 Table 6-5: Estimated unit material costs for the given product range (2003 2005) ...................................................................................................... 162 Table 6-6: Aggregate labour rate calculation............................................................. 164 Table 6-7: Total man-hours and estimated labour costs (2003 2005) for the given product range ................................................................................ 165 Table 6-8: Overhead costs for individual elements (2001 2005) ............................. 167 Table 6-9: Estimated product cost values (2003 2005)............................................ 168
14
List of Tables
Table 6-10: Material cost deviation index and estimated unit material cost values (2003 2005) .............................................................................. 170 Table 6-11: Labour units calculation ......................................................................... 172 Table 6-12: Labour cost deviation index and estimated labour costs.......................... 173 Table 6-13: Estimation of processing units, deviation indices, rates and costs ........... 174 Table 6-14: MDC deviation index and estimated MDC per unit values ..................... 176 Table 6-15 PO fractions (actual & estimated) and deviation indices .......................... 177 Table 6-16: Estimated per unit values for manufacturing, PO and product costs........ 178 Table 7-1: Lead times, machine running costs, rates (operator and machine) and labour and machine costs.................................................................. 186 Table 7-2: The total factory expenses, the expenses rate, the estimated per unit values (factory expenses, product cost) (2003-2005)............................... 188 Table A-1 Hammer Drill (product level) 1.6 Kg........................................................ 268 Table A-2 Cumulative material quantities at the lowest level for the hammer drill......................................................................................................... 273
15
List of Notations
The following is a list of the main symbols used for modelling in the thesis, together with their brief descriptions.
a
b d h
i
Total no. of broken or worn out labour tools in the nth year Total no. of broken or worn out machine tools in the nth year No. of time dependent cost elements No. of building space cost elements No. of machine centres No. of work centres Amount of the kth material in product p amount of nth direct material No. of overhead elements No. of work centres routed by the pth product Work time consumed by the pth product at the ith machine centre in the nth year Work time consumed by the pth product at the jth work centre in the nth year. time spent by a skilled labour in the jth work centre working on product p time spent by a semi-skilled labour in the jth work centre working on product p time spent by a non-skilled labour in the jth work centre working on product p No. of material dependent cost elements
j
mk mn
q
r
n t ip
tn jp
tx
ty
tz
16
List of Notations
x y z
n Cd
No. of skilled labour working on product p in the jth work centre No. of semi-skilled labour working on product p in the jth work centre No. of non-skilled labour working on product p in the jth work centre Total cost in the individual time dependent cost elements (such as utility cost, maintenance cost, repair cost, depreciation, insurance, tax etc.) for the nth year overall direct material costs in the (n-1)th year unit cost of the nth direct material overall freight & transportation costs in the (n-1)th year Total cost in the individual building space cost elements (such as plant depreciation, building insurance, maintenance, repair, tax, utilities etc.) for the nth year overall inspection costs in the (n-1)th year overall indirect material costs in the (n-1)th year Unit cost of the kth material in product p in the nth year Total material dependent costs in the nth year Estimated material dependent cost for the pth product in the (n+1)th year Material cost for pth product in the nth year Material cost for pth product in the (n+1)th year Cumulative material cost for p products in the nth year. Cumulative material cost for p products in the (n+1)th year. Estimated cost for the pth product in the (n+1)th year overall purchase department costs in the (n-1)th year overall stores & inventory costs Total time dependent costs in the nth year
n 1 Cdmt
Cdn
1 Cn ft
n Ch
Cin 1
n 1 Cim
C kn
n C md n +1 C mdp n C mp n +1 C mp n C mt n +1 C mt n +1 Cp n 1 CP n 1 C si n C td
17
List of Notations
n +1 Ctdp n 1 Ctotal n Cw
Estimated processing cost for the pth product in the (n+1)th year total capacity in the (n-1)th year Total cost in the individual material dependent cost elements (such a indirect material, purchasing cost, stores & inventory cost, freight & transportation cost, material inspection cost, packaging cost, quality cost, etc.) for the nth year Estimated building cost for the pth product in the (n+1)th year Total building space cost in the nth year Estimated engineering cost for the pth product in the (n+1)th year Estimated manufacturing cost for the pth product in the (n+1)th year Total manufacturing cost in the nth year Direct labour cost for the pth product in the nth year Estimated direct labour cost for the pth product in the (n+1)th year Total labour tool cost for the nth year Total labour cost in the nth year Total estimated direct labour for the (n+1)th year Total machine tool cost for the nth year Estimated tooling cost for the pth product in the (n+1)th year Total tooling cost in the nth year Total depreciation of the broken labour tool Total depreciation of the broken machine tool Total depreciation of the Lth tool in the nth year Total depreciation of the Mth tool in the nth year Total wages for jth work centre in the nth year,
n +1 C Bp n C Bt n +1 C Ep n +1 CGp n C Gt n C Lp n +1 C Lp n C LT n C Lt n +1 C Lt n C MT n +1 CTp n CTt
Da Db
n DL n DM n Gtj
18
List of Notations
n Hq
Total cost in the individual overhead elements (such as computer software cost, general administration cost, financing expenses, selling expenses etc.) for the nth year Total no. of useable labour tools in the nth year Labour units consumed by the pth product at the jth work centre in the nth year. Total no. of useable machine tools in the nth year Processing units consumed by the pth product at the ith machine centre in the nth year manufacturing lead time No. of units of pth product produced in the nth year. No. of units of pth product produced in the (n+1)th year. Occupied space by the jth work centre in the nth year Estimated overheads for the pth product in the (n+1)th year Space occupied by the pth product at the rth work centre in the (n+1)th year Total overheads for the nth year Material-related overhead for a new product in the nth year time-related overhead for a new product in the nth year Initial purchase price of the broken labour tool Initial purchase price of the broken machine tool Building space rate for the nth year Estimated building space rate for the (n+1)th year Actual labour rate for the nth year; Labour tool rate for the nth year Estimated labour tool rate for the (n+1)th year
Lnjp
M
n M ip
MLT
n Np n +1 Np
On j
n +1 Op n +1 Orp
Otn
OM
OT Pa Pb
n RB n +1 RB n R LA n RLT n +1 RLT
19
List of Notations
n +1 RLE n RMA n +1 RME n RMT n +1 RMT
Estimated labour rate for the (n+1)th year Actual processing rate for the nth year Estimated processing rate for the (n+1)th year Machine tool rate for the nth year Estimated machine tool rate for the (n+1)th year Total scrap value in the nth year Total space occupied by the pth product in the (n+1)th year Total space occupied by all work centres in the nth year Total unoccupied space on the manufacturing floor in the nth year Total manufacturing space for the pth product in the (n+1)th year Total manufacturing space in the nth year total time-related overhead in the (n-1)th year Labour units for pth product in the nth year Labour units for the pth product in the (n+1)th year Total labour units consumed in the nth year. Total labour units consumed in the (n+1)th year Processing units consumed for pth product in the nth year Processing units for the pth product in the (n+1)th year Total processing units consumed in the nth year skill index for semi-skilled labour (0.4 0.8) skill index for non-skilled labour (0.25 0.4) Building space cost deviation index in the (n+1)th year
Sn
n +1 S op n S ot n S ut n +1 S Gp n S Gt n 1 TROtotal n U Lp n +1 U Lp n U Lt n +1 U Lt n U Mp n +1 U Mp n U Mt
n+1
20
List of Notations
n +1
Labour cost deviation index in the (n+1)th year Material cost deviation index in the (n+1)th year Machine tool cost deviation index in the (n+1)th year Machine index (1.25 2.0) Processing cost deviation index in the (n+1)th year MDC deviation index in the (n+1)th year Labour tool cost deviation index in the (n+1)th year PO deviation index in the (n+1)th year
n+1 n +1 i n +1 n+1 n +1 n +1
21
List of Acronyms
The following is a list of acronyms used in the thesis, together with their brief descriptions.
ABC ASSD BCDI BOM BPNN CBR CCS CIM CMD CMR CUI DFC DSM DSS DT HCS HMI ICSD
Activity based costing assembly & services sub-division building cost deviation index bill of materials back-propagation neural network case-based reasoning cost control system computer integrated manufacturing crane manufacturing division cumulative material requirements cost uncertainty index Design for cost decision support model decision support system distribution transformers hierarchical classification system Hybrid Model Implementation installation & commissioning sub-division
22
List of Acronyms
LCDI JIT LTCDI MCDI MDC MDCDI MLT MRO MRP MTCDI MTO PCE PCDI PO PODI PT QFD SED SPSD TRO
labour cost deviation index Just-in-time labour tool cost deviation index material cost deviation index material-dependent costs material-dependent costs deviation index manufacturing lead time material related overheads material requirement planning machine tool cost deviation index make-to-order product cost estimation Processing cost deviation index production overheads production overheads deviation index power transformers Quality Function Deployment ship engineering division spares & parts sub-division time related overheads
23
Acknowledgements
Alhamdulillah, with the thesis finally in my hand, a long and hard struggle seems to have concluded with nothing less than a sweet reward. My efforts alone in treading this difficult journey from its very start would take me no where, had it not always been coupled with the explicit or implicit support of various people all along.
I would like to heartily acknowledge the never-ending support of Prof Jian Dai as not only my PhD Supervisor but as a kind and compassionate mentor. His valuable guidance in matters of both personal and professional grooming played a pivotal role. I would like to extend my special thankfulness to Dr Stavroula Balabani whose continued assistance in issues of academic and non-academic relevance has been instrumental to the overall success of the project. I would also like to acknowledge the support of Prof Lakmal Seneviratne as Head of the Division of Engineering, Kings College London. The active support and dedication of all these academics made the publication of two quality journal papers possible. The submission of another three papers in leading journals was also made easier with their continued support.
It is difficult to find words to pay gratitude to Dr Salim Habibs generosity through Pakistan Scholarship without which my PhD simply would not have been achieved. His unceasing personal interest in my academic progress remained a revitalizing element
24
Acknowledgements
throughout my PhD studies. Kings Colleges support through Principals Discretionary Fund during crucial stages of my PhD proved vital. Caroline Usher played an important part not only in liaising with Dr Habib but in guiding me towards the discretionary fund.
I would like to remember the support of Mr Muhammad Arif Hasan and Mr Saeed Iqbal in making the industrial visit to an Electrical Engineering Company in South Asia possible. I would also like to name Mr Peter Faccenda from Manufacturing Advisory Service UK for making the industrial visit in the UK possible by liaising with the Crane and Ship Engineering Company. I would have liked to name the companies involved that were central to the success of the project. I would acknowledge the support provided by both the companies that agreed to disclose very sensitive data crucial to their business competitiveness and equally vital to advance my research studies. Due to the confidentiality accord and understanding the need for their anonymity, I would like to thank all those namelessly who provided key information necessary for the research progress and its eventual completion. I would also like to acknowledge the positive feedback and interest from Professor Frank J Fabozzi of Yale School of Management (USA) for my first journal publication.
This uphill task simply would not have been possible without the support and care from a loving family like mine. My parents visited the UK twice from Pakistan during the course of my PhD to reassure their heartfelt love at times when I really needed it most. I would not be where I am now without the infinite love and guidance from my parents. My wife Uzma stood by all my judgements with an unwavering belief in my abilities at times when any uncertainty on her part would not be unreasonable. My daughters, Daanya and Iqra remained the inner strength for me. Last but not least, the support from 25
Acknowledgements
my brother, Kamran and my sister, Farhea peaked when they visited me from Pakistan during some of the very difficult times during my studies.
26
Chapter 1
Introduction
This chapter outlines the scope of the thesis by giving an overview of the research area under study. Main problems are identified in the field and aims and objectives of the study are outlined. The chapter also outlines the structure of the thesis and ends with the conclusions.
1.1
Overview
Chapter 1 is an introductory chapter and is aimed at providing an overview of the research field; identifying main problems and setting the relevant aims and objectives for the research study.
The advent of information technology has brought with it an unprecedented era of globalization. The ever-increasing pressure on firms to stay competitive in such an environment is constantly forcing them to remain innovative in all aspects of their business. No wonder, not only products and services are increasingly customized to suit the end users needs but the required business tools are ever more novel. One of the key business tools is a pricing strategy that combined with market awareness not only allows an enterprise to remain competitive but thrive in the market.
27
Chapter 1: Introduction
A good pricing system rests on the application of cost engineering. The fundamentals of the former and the principles of the latter are not mutually exclusive. Cost estimating, cost control and profitability are the elements of cost engineering [1]. The American Association of Cost Engineers (AACE) defines cost engineering as that area of
engineering practice where engineering judgement and experience are utilized in the application of scientific principles and techniques to the problems of cost estimation, cost control, and profitability.
Whereas market awareness can deal with the issues of profitability to some extent, maximizing profits comes down to better cost control. An effective cost control system (CCS) aims to reduce the gap between the budgeted and the actual costs. Setting the realistic budgets in turn is one of the functions of cost estimation. The ultimate responsibility of maximizing profits is, therefore, often closely linked with the provisions of accurate and timely cost estimates in order to facilitate key managerial decisions.
Cost estimating unlike cost accountancy requires sound engineering knowledge in order to deal with problems involving scientific principles and techniques. Cost estimation deals with predicting the cost of a product, project or a service. Product cost estimation (PCE) methodologies aim to predict product costs early and accurately before the actual production takes place and sometimes even before the design cycle. The scope of the thesis is PCE with specific focus on manufacturing costs. Figure 1-1 highlights the area of research followed in the thesis.
28
Area of research
Project Cost Estimation Service Cost Estimation Product Cost Estimation
Applied Cost Engineering Cost Estimation Cost Control Profitability
Figure 1-1: Main area of research
1.2
Problems identification
There are a number of issues in the area of PCE that need investigation. Following is the brief outline of some of the problems that form the basis of the research study:
A vast number of estimation techniques are available but no classification system available
Difficulty in selecting an estimation methodology for a given condition Early estimation and accuracy are counter to each other Available methodologies mostly estimate only part or component costs not the entire product cost
Most of the methodologies that can predict the entire product cost, fall short of predicting accurately the costs breakdown 29
One of the breakdown elements is overhead (sometime referred to as indirect costs) that is difficult to predict accurately for individual products
A vast number of methods and techniques have been developed over a period of time to facilitate estimators and designers to predict a products cost. However, they differ in applications in terms of compatibility to the needs of a system, delivering optimized results in given conditions and the level of resource consumption. A great number of methods, on the other hand, share common grounds. Estimators often find it difficult to select a methodology to suit the needs of a system under consideration. This is normally because an in depth analysis of a specific methodology to check its compatibility with an organizational framework is often time-consuming let alone analyzing more than one techniques. The selection of a particular methodology is normally based on the availability of data, level of accuracy and the stage of estimates required. A better exploitation of the differences and similarities of the techniques could result in helping estimators to select a specific methodology to suit the needs of a system. The nonavailability of a classification system is a barrier to the notion. Such a system could also help in developing a decision support tool in order to help estimators of product costs to make decisions to select an estimation methodology to satisfy a systems needs and make the best use of available resources.
The aim of estimating a products cost is not just predicting it accurately but as early as possible in order to facilitate key business policy decisions. Often a methodology selected for PCE in the early stages of a design cycle, does not furnish accurate enough results due to the non-availability of the design details. Historical data can overcome the problems to some extent but the requirement of extensive past results limits the use of
30
relevant techniques. Such techniques also fall short of predicting accurate results for new designs. Study of a products features could be helpful in such circumstances but not only require skilled estimators but mostly fall short of delivering early results due to the detailed designs required. The reliance of the existing methods on product design and process planning details could result in accurate estimation but often leads to unnecessary delays. On the other hand, the uses of past experience, data or knowledge do have the potential to deliver early estimates but compromise accuracy. Such discussion is elaborated with necessary references in detail in chapter 3. Due to the significance of an early and accurate estimation process, it is important that a correct technique is employed within a specific set of conditions. However, most of the available techniques make a compromise between early estimation and accurate results.
Most of the available techniques predict part or component costs instead of an entire products cost using certain design parameters (such as design features, dimensions, tolerances etc.) and manufacturing operation times. The individual components costs add up to an entire products cost along with considering any process planning details for assembly operations based on standard techniques, historical data or time and motion studies etc. The process is time and resource consuming. A methodology aiming to predict an entire products cost by not relying on details like product designs or process planning details could be the answer. However, an alternative to such details would have to be found.
A more traditional approach can predict the entire product cost based on allocating the factory-wide resources to individual products. Such a method divides the entire product costs into direct and indirect costs. Direct costs refer to direct material and labour 31
consumed for the benefit of the product whereas, indirect costs, also referred to as overheads, are incurred for the shared benefit of all the products. The method obtains material and labour costs either from past data for the existing products or uses design and process planning details to calculate for the new products. However, using such details for a new product design could not allow the method to predict costs early. The method allows the allocation of the third cost element, overheads, to individual products by considering their lead times and an aggregate overhead rate. However, again not only the similar problem arises for any new products but the use of the aggregate rate results in less accurate estimates.
Another problem with using the traditional method is its breakdown of the entire product cost into only three elements (i.e. material, labour, overheads) resulting in overlooking possible areas of optimization for cost control. In a production environment where a product is the result of a mix of activities, the degree of consumption of these activities is reflected in a products cost. However, in such an environment, finding the costs incurred on individual breakdown elements and sub-elements is difficult, especially for overheads. The early and accurate estimation of this element and its subelements is crucial to the overall accuracy of a products cost. Predicting such costs beforehand requires much more than just accountancy laws. The early and accurate estimation of the individual elements and sub-elements costs could provide effective cost control opportunities.
32
1.3
Aims and objectives
The set of identified problems in the area of PCE is helpful in establishing an outlook within which the aims and objectives for the overall research work can be set. The individual problems identified would help to establish the objectives for the research work in order to find or attempt to find pragmatic solutions. The overall aim of the work, however, is to make a genuine and significant contribution to the area of cost engineering in general and to PCE in particular.
Following is the list of the objectives set for the research study: 1. 2. Development of a technique classification system Development of a decision support model (DSM) to allow estimators and designers to select an estimation methodology in given conditions 3. 4. Development of an overhead estimation methodology Development of a PCE methodology with cost deviation indices for early and accurate estimation of a products cost with its elements costs 5. 6. Industrial implementation of the developed cost estimation technique Comparison analysis for the validation of the developed model
A comprehensive review of the available cost estimation techniques could help to identify the similarities and the differences between them. A careful analysis of their advantages and limitations could be useful in grouping them. This in turn would allow the development of a technique classification system. Such a system would also lead the way to establish conditions under which designers and estimators make decisions to select a specific methodology. A DSM can, therefore, be developed too. 33
The development of a PCE methodology for early and accurate estimation of a products cost would require a thorough analysis of the existing methods. One of the problems of the existing methods is their reliance on product designs and process planning details. The developed method should aim at eliminating or reducing the need of such details. This can be achieved by making use of the past production details in order to predict the estimates for the future production. However, if a precedent is not available, the use of such technique can be limited. Such a limitation can be overcome by making use of past design cases resulting in a case-based framework. One of the objectives of the study would, therefore, be to develop a case-based framework in order to facilitate effective utilization of past details for early and accurate estimation of product costs.
The determination of an entire products cost instead of only part and component costs would require evaluating any existing methodologies that can predict such costs. The areas of improvement in those methods will be identified. One of the areas, for example, is overhead estimation that leads to inaccurate estimation of an entire products cost. Another objective of the study, therefore, would be to identify problems with the existing method of overhead estimation and develop an improved methodology. The effect of the proposed methodology on the overall estimation of a products cost will then be studied. An indication of improved overhead estimation results would pave a way for modelling an entire products cost with its breakdown elements.
Finding pragmatic solutions would require industrial implementation, comparisons and validation to back up any theoretical advancement in the field. One of the objectives of the research would, therefore, be to validate the proposed mathematical models from 34
industrial applications. The aim of the study is, therefore, to make theoretical advancement in the area of PCE and present industrial validation analysis.
1.4
Thesis structure
The overall research work is detailed in the thesis with the aim of disseminating not just the research findings but knowledge in the area of PCE. A logical sequence is maintained in presenting the work throughout the thesis starting from general concepts and literature review to methodology development, implementation, application and validation. The work comprises eight chapters in all.
Chapter two uncovers the background and related work in the area of cost estimation. The concepts of costs and costing are outlined. Price and pricing are briefly described. Cost estimation is defined and discussed with a special focus on manufacturing and cost control. Manufacturing systems are discussed with a view of selecting a suitable system to develop a methodology for. The functions of cost estimation are outlined and its significance in the early stages of a design cycle is highlighted. A comprehensive literature review in the area of cost estimation with special emphasis on its specific applications is also presented in this chapter. The chapter identifies a batch manufacturing environment as a potential area to develop a cost estimation methodology for.
Chapter three establishes comprehensive theories in the area of PCE. Existing technique classification systems are briefly discussed before developing a comprehensive
35
classification system for the available techniques in the area. The literature for each category is comprehensively reviewed along the course with the identification of strengths and weaknesses for the individual methods. The underlying principles for the categories are mentioned and the conditions for selecting a method from each one of them are set out. Establishing such conditions helped to develop a DSM for methodology selection. The chapter also presents a framework for the case-based methodology. A useful summary of the reviewed work and the key advantages and limitations for the proposed categories is then presented in the end. The chapter identifies that a hybrid approach combining the elements of qualitative and quantitative techniques could result in early and accurate estimation of a products cost.
Chapter four forms part of the methodology development for PCE. It considers a representative case of the existing method for overhead estimation. Problem identification leads to the development of a new overhead estimation methodology based on material and time dependent overheads. The effects of the proposed methodology on the estimation results for the overall product costs are analysed retrospectively for a four year period. The industrial validation analysis results reveal the superiority of the proposed methodology. However, the room for further improvement in the proposed methodology are also identified. The chapter also reveals cost elements breakdown statistics typical of the South Asian region and help to understand the implications of geographical locations on the effectiveness of a CCS.
Chapter five is a step forward in the development of a comprehensive methodology for PCE in a batch type manufacturing environment. The concept of overhead estimation proposed in the preceding chapter is carried forward to develop comprehensive 36
mathematical models for estimating an entire products cost in a batch type manufacturing environment. The developed model is hybrid in nature based on a cost breakdown structure and the concepts of modified activity-based costing (ABC) system. The proposed methodology is based on an effective utilization of past data to predict future costs facilitated by incorporation of cost deviation indices.
Chapter six is based on the industrial implementation and application of the proposed Hybrid Model in a batch type manufacturing environment in the UK. The chapter is a step towards the overall validation of the developed Hybrid Model. In order to facilitate the industrial implementation of the proposed model, an implementation algorithm is developed and the already proposed indices from the preceding chapter are modelled. The developed model is then implemented retrospectively in a crane and ship engineering company based in the UK. The implementation process generates the cost estimates for a given product range. The companys own method of cost estimation is also described and is used to generate cost results for further comparison analysis.
Chapter seven presents the comparisons and validation analysis of the Hybrid Model. The developed model is compared against a published model and the companys own method as the representative cases from the two domains. A carefully selected published model is used to furnish cost estimates. Comparison analysis includes comparing the estimated costs obtained from the three methods against the actual costs. The developed model is validated based on generating more accurate and more consistent results as opposed to both the companys own method and the published model. The chapter also reveals cost elements breakdown statistics typical for the UK.
37
Chapter eight concludes the research work and the thesis. The overall contents in general and the research work in particular are summarized. The main achievements of the work are highlighted and the success of the research study is evaluated. This involves a comparison of the achievements against the planned aims and objectives. The work is also evaluated for the originality and its contribution. The areas for possible improvements in the research work are pointed out with a focus on the existing limitations. The chapter also discusses the current and future research trends in the field. Finally, the possible avenues for future research directions in the established research work are highlighted with a view of providing a platform for any further research in the area.
1.5
Conclusions
This chapter presented the scope of the research work. An overview of the research area focussed on the importance of cost estimation for cost control. Some of the problems were outlined and discussed briefly in order to establish the likely aims and objectives for the study. Aims and objectives were set with the view of furnishing solutions to the outlined problems and establishing a significant theoretical development in the area. Finally, the overall structure of the thesis was outlined with the aim of translating the objectives in logical order in different chapters.
38
Chapter 2
Background and Related Work
The aim of Chapter 2 is to develop background information in the area of cost estimation. The concepts of cost and costing are established. Manufacturing systems with a focus on CCS are discussed. Cost estimation as a key component of CCS is then discussed. Various application areas for cost estimation techniques are discussed with a comprehensive literature review in the area. The rationale for developing a cost estimation methodology in a batch type manufacturing environment is established. The chapter ends with conclusions.
2.1
Introduction
Chapter 2 deals with background information in the chosen area of study and provides a comprehensive literature review in the area with a focus on cost estimation techniques applications.
Our day to day lives are full of instances of buying and selling goods and services. The amount paid to buy a commodity reflects the incurred cost from a buyers perspective. The same amount is perceived a selling price from the sellers view point. The commodity may be resold with a higher amount making a profit. Cost can therefore, be defined as: 39
Chapter 2: Background and Related Work
The amount of money paid or required as a payment to buy a goods or service
The original commodity may often be altered or modified before being resold and is said to have been added with a value. The value addition process thus incurs further costs. The incurred costs could be the result of tangible and/or intangible values. The process of determination of the final cost of the modified commodity before being resold is referred to as costing.
Costing determines selling price. Setting prices in turn is called pricing. As costing normally refers to considering all the costs already incurred for a commodity, any unexpected costs at the beginning of the value addition process are discovered and accounted for. For the same reason, however, its role is normally limited to pricing and price adjustments. Cost control may, therefore, not be its domain.
Since selling price is one of the major factors of gaining competitive advantage in the consumer market, cost control can not be ignored. Cost accounting also known as management accounting could provide solutions to cost control by recording cost values and providing opportunities to minimize them in future.
Operational matters requiring aspects of cost control within the framework of a project execution, service delivery or manufacturing set up may be different. The aspects relating to manufacturing environment come under the scope of the current study.
40
2.2
Manufacturing and cost control
Manufacturing involves converting materials from one form to another by adding value to them. This often requires a series of specific treatments or processes to be carried out on different materials. These processes accomplish the desired final product form and shape specified by a design engineer prior to actual manufacturing. Selection of a particular process among various alternatives is closely linked with many factors including the associated costs. Incorporating certain features (from available alternatives into a product design) that require costly processes is often a result of poor product design. Therefore, the final product cost reflects the design quality and also determines selling price. The latter is also governed by consumer demand and is crucial in gaining competitive advantage. Maximizing profits (while remaining competitive in the market) requires an effective CCS alongside keeping high quality standards. An effective CCS monitors various aspects of the final product cost.
P Productivity (P)
Increasing production volume Increasing product variety
Flexibility (F)
Figure 2-1: Relationship between productivity and flexibility
41
In a manufacturing environment, where high volumes of products are produced without making many changeovers in production setup, the aspects of cost control are a routine task. The effectiveness of a CCS, however, is more strongly required within a manufacturing setup of routine changeovers in production setup. The ability to
changeover to a new production setup economically and quickly to produce new products in response to market or engineering changes is called Product Flexibility. For example, the automobile industries are subject to tough competitions and in order to remain competitive, develop different car models and introduce them at shorter intervals. On the other hand, Productivity is a measure of the extent to which the resources of an organization are consumed effectively in transforming inputs to outputs. The overall productivity of a firm is greatly influenced by flexibility. The productivity of a manufacturing system remains in conflict with flexibility where increasing productivity compromises the systems flexibility. The relationship is elaborated in Figure 2-1 between flexibility and productivity. An increased flexibility of a manufacturing system demands higher changeover time thereby reducing its productivity. Increased productivity also refers to higher production volumes whereas; an increased flexibility is a reflection of an increased product variety.
Business goals for a manufacturing enterprise are set within the framework of the organizational corporate planning structure. Strategic planning framework in return ensures that a manufacturing system be in place in order to achieve those objectives. For example, the level of flexibility and productivity required to achieve business objectives would need a specific set up for a manufacturing system. High production volumes and low flexibility are the attributes for mass production and can be achieved by setting up transfer lines. Such systems make use of automated assembly lines using industrial 42
robots etc. The investment needed for such a manufacturing set up is high. However, both labour skills and labour costs are relatively low. Small production quantities with a high flexibility, on the other hand, can be achieved by conventional job shops. However, the level of skill required to operate general-purpose machines in such a set up is high. Since the production rate is low, the unit cost in job shops is usually high. Conventional flow line, flexible manufacturing system, manufacturing cells, numeric control systems etc. can be effective in achieving a desired combination of flexibility and productivity and normally result in batch production. As a result multi-skilled labour (low, medium, high) is generally employed to handle tasks for varied levels of flexibility and productivity. Small to medium size batches can be produced on machines ranging from general-purpose with computer controls to equipment with specifically designed fixtures and tools. Figure 2-2 elaborates various levels of productivity and flexibility achievable through different manufacturing systems. The overlap between the systems is due to the various levels of technical attributes (automation, computer control etc.) achievable in each system. Due to the impact of the changeovers on product costs, the need for an effective CCS in a batch manufacturing set up is wider in its scope. Job shop, although, highly flexible can be partly covered by batch production cost control strategies. Both mass production and a job shop system are, therefore, left with little scope for comprehensive CCS.
43
Increasing Productivity
Transfer Line
Mass Production
Batch Flow Line
Flexible Manufacturing System Manufacturing Cell
Batch Production
NC Production
Job Shop System
Job Shop Production
Increasing Flexibility
Figure 2-2: Flexibility and productivity of different manufacturing systems
Whether a manufacturing system is based on mass, batch or job shop production, an effective CCS makes sure at every stage of the execution process that the planned budget is adhered to. Accountancy laws are easier to apply once the plans are executed and actual costs are available. The aim is to minimize the difference between the planned budget and the actual costs. In a highly competitive business environment, any non-compliance to planned budgets could seriously jeopardize the accomplishment of an enterprises business objectives. Setting realistic budgets in turn is, therefore, also an important element of a CCS and requires accurate predictive tools. A carefully devised cost estimation methodology based on sound engineering principles serves as a reliable tool for predicting accurate costs likely to be incurred during an execution phase.
44
2.3
Cost estimation
Cost estimation is the process of determining a combination of tangible & intangible values associated with the expected and unexpected cost occurrences linked with a set of activities prior to their execution. Accurate cost estimation is essential whether the end product is a project, manufactured goods, or a service. The cost estimated close to the accounting cost should serve as a valuable decision aid tool at the early stages of an execution plan.
Different cost estimation approaches have been developed to best suit the needs of end users or customers. One approach may require the evaluation of uncertainty as an intangible value associated with the loss of goodwill owing to delayed job completion, whereas another may necessitate the estimation of expected cost tied with the development or manufacture of a product. The use of risk assessment and uncertainty evaluation techniques is quite common in project proposals and bidding processes. PCE techniques, on the other hand, generally tend to consider the effect of expected cost occurrences.
Cost estimation affects certain organizational functions. The following is a list of functions or processes in a manufacturing company where cost estimation is involved. The list is not exhaustive but provides a good idea of the impact of cost estimation in the operational and corporate planning structure of an organization. The functions include:
deciding whether to produce in-house, purchase, or out-source; 45
selecting material, manufacturing processes, and routings; evaluating different design alternatives; setting economic lot sizing for batch production; assessing suppliers quotations as a trade-off between customer needs (quality) and available budget (cost);
quoting prices to customers before actual manufacturing starts; allocating enterprise-wide resources (man, machinery, tooling, etc.) and budgeting to different technical attributes of a design in proportion to their desirability (extent of customer requirements);
manufacturing cost control and hence overall control of production cost.
Good cost estimation has a direct bearing on the performance and effectiveness of a business enterprise as overestimation can result in loss of business and goodwill in the market whereas underestimation may lead towards financial losses to the enterprise. Due to this sensitive and crucial role in an organization, cost estimation has been a focal point for design and operational strategies and a key agenda for managerial policies and business decisions. As a result, a substantial research effort has been expanded in exploring design implications, new techniques and methods for producing accurate and consistent cost estimates not only to generate optimum design solutions but also to achieve the maximum customer satisfaction in terms of low cost, high quality and intime product delivery.
46
2.4
Cost estimation in the early design stages
The key to thrive for a manufacturing enterprise in the 21st century is based on product quality, competitive cost, fast delivery, and flexibility. On the other hand, factors like globalisation and mass customisation put an extra pressure to a business enterprise to survive and remain profitable at the same time. Whereas, an innovative approach and a new product development process may attempt to deal with issues such as flexibility and product quality, they may still be time consuming and less cost effective. In addition, the prospective end user of a would-be product often demands a price quote as soon as possible, sometimes even unconcerned and oblivious of factors such as the extent of the customisation, the nature of the data required and the design complexity. To make matters worse, often a manufacturer ignores the significant factors like design module availability, manufacturability and the level of accuracy required for processing time estimation. The overall situation, therefore, could either lead to an underestimation resulting in a profit loss and a blow to operational targets or a more profound strategic damage caused by overestimation leading towards the loss of customer goodwill and market share. All the above highlights the ever-increasing importance of devising methods to forecast the cost for a new product in the early design and development phases with accuracy.
Since most of the product costs sustained during later in the production life cycle are determined during the conceptual design phase [2], the cost estimation in the early phase of the design cycle is crucial. Many researchers have emphasized the importance of cost estimation at the early design stages when 70 to 80 percent of a total product cost is determined [3-6]. Some researchers have developed methodologies with a special
47
emphasis on early cost estimation [7, 8]. A framework for developing a cost database was suggested by Sheldon et al. [9] and aimed to serve different groups of design for cost (DFC) system users to determine appropriate cost structures by analysing the information provided by a cost accounting system. Knowledge representation in such a way facilitated the generation of cost estimates at an early design stage. A framework to integrate design costs into Quality Function Deployment (QFD) was used by Bode and Fung [10]. The approach adopted is a helpful tool for designers at the early stages of product design for making trade-off decisions between quality and cost prioritising the attainment of technical attributes based on customer requirements. All the above mentioned highlights the significance of not only estimating a product cost accurately but as early as possible.
2.5
Cost estimation for specific applications
A number of cost estimation methods and techniques were developed with reference to particular applications. The techniques only suit the conditions for which they were tailored and can only be effective for their application areas. These techniques range from the evaluation of certain manufacturing and machining processes to the dedicated techniques designed to suit specific manufacturing systems, from composite material costing to the cost analysis of parts and assemblies and from dealing with specific segments in a production cost cycle to covering a product life cycle cost. The section is aimed at providing a comprehensive review of the available techniques for cost estimation with an emphasis on their application areas. This not only leads to develop better understanding of the application areas but could also be helpful in providing a
48
platform for any future exploration within a specific application area of the cost estimation.
2.5.1
Cost estimation for a specific segment in a production cycle
Different costs are associated with various stages in a production cycle starting from the ones incurred during the early phases of a design cycle to those linked with the manufacture of an actual product on the shop floor. Many researchers devised methods to evaluate the costs associated with a specific segment in a production cycle. For example, if a methodology is applicable at the QFD stage [10] the other is developed for the costs associated with the design and development phase of a product [11]. Methods for process planning cost evaluation and optimization can also be found in [12-14]. Aldrich [15] estimated the cost associated with the bill of materials (BOM) using MRPII software. Cost calculation in manufacturing and machining can be found in [16]. Costs associated with conventional manufacturing processes [6], non-conventional manufacturing processes [17] and the machining accuracy [18] can also be found.
Some researchers developed methodologies that could be used for cost estimating in several stages of a product design cycle. For example, Weustink et al. [19] developed a framework for product cost control by estimating the various cost elements and storing the data in a generic way. The methodology allows cost estimation on different aggregation levels, e.g. feature level, component level, assembly level etc. Koonce et al. [20] developed a system capable of generating cost estimates in all phases of the design stage. The system, which was prototyped in JAVA, used simple parametric cost relations in the early stages of a design, when detailed design was not available to 49
produce cost estimates for a product. When the design was developed, estimates could be produced using design features, manufacturing features or a process plan.
2.5.2
Cost estimation for specific machining and manufacturing processes
The selection of suitable manufacturing processes is governed by an accurate estimation of the costs associated with them among other factors. Methods have been devised to predict the costs of specific machining and manufacturing processes. These include the assembly costing techniques [21, 22] and the cost models for die-casting [23]. Further models were developed for a hole-making process [24], welding [25] and milling and drilling [26].
2.5.3
Cost forecasting for specific parts and products
Many researchers focussed on the application of the cost estimation techniques to specific products ranging from standard parts and components to a particular product group. Schreve et al. [27] presented a cost model using mild steel fabricated parts, whereas the one for machined components can be found in [28]. Hicks et al. [8] developed four cost modelling approaches for various classes of engineering components. These components were defined as standard selected, standard designed and bespoke designed. French [29] used a function cost modelling methodology to estimate the costs of mechanical components whereas Ulrich and Eppinger [30] used previous orders and procurement records to estimate the cost of similar components. Kendall et al. [31] presented cost information for automotive components using a 50
simulation technique. Gutowski et al. [32] presented a process-oriented cost model to estimate manufacturing cost of advanced composite aerospace parts. Cost models for injection-moulded components can also be found in [33-35]. On the other hand, cost models for specific product groups include those for PCB manufacturing [36-38], developmental equipment [39], packaging products [40, 41], injection moulding tool [42] and gear drive manufacturing [6].
Another type of products covered by cost estimation techniques is based on composite material. For example, cost and consolidation model for commingled yarn based composites was presented in [43]. Process flow simulation techniques to evaluate manufacturing costs for composite products were discussed in [44]. Cost analysis of thermoplastic composites using different techniques can also be found in [45, 46], whereas Walls and Crawford [47] used historical data to produce cost information for continuous fibre-reinforced thermoplastic products.
2.5.4
Cost estimation for generic systems
Cost estimation models to suit the needs of a generic system were also developed by many researchers. For example, cost models for job shop manufacturing systems can be found in [48, 49]. On the other hand, a mathematical and simulation model to estimate the manufacturing and product cost in material requirement planning (MRP) and just-intime (JIT) systems was proposed in [50]. Similarly, a simulation model to estimate the cost in a flexible manufacturing environment was proposed in [51], whereas the one for cellular manufacturing configuration was proposed in [52].
51
Cost estimation techniques were further developed for specific industry sectors [53]. For instance, work has been carried out for airplane manufacturing industries [54], electronics industries [55], automotive industries [56], aerospace and defence industries [57], telecommunications [58], and ship building industries [59].
Table 2-1 summarizes the discussed literature references for various applications of PCE. It is clear that the techniques have been designed to suit the conditions for a given application area and may not be suitable for any other applications. For example, a methodology designed to predict machining costs may not be applicable for other areas of manufacturing due to the differences in the requirements of the parameters for the two conditions. There is, thus, a need to develop a method that can cover a number of application areas without compromising accuracy yet furnishing estimates in the early stages of a design cycle or even before. Such a system could eliminate or minimize the need for a methodology selection under varying conditions. The techniques designed to work for a specific generic system usually do not have the limitations for other application areas. For example, a methodology developed for a job shop system, although, may not be suitable to work in a batch manufacturing environment but can provide estimates for manufacturing parts and be applicable in the different phases of a design cycle. Maximizing the application areas for a methodology could, therefore, be achieved by originating it from the applications for generic systems. In other words, if an estimation methodology is designed to suit the needs of a generic system, it could work well for the other application areas.
52
Table 2-1: Summary of the published literature references for various applications of PCE
S. No. Application area 1. 2. Cost estimation for a specific segment in a production cycle Cost estimation for specific machining and manufacturing processes Parts and components 3. Cost estimation for specific parts and products Special products Composite material products Specific manufacturing system 4. Cost estimation for generic systems Specific industrial sector
References [6], [10-20] [21-26], [75-79] [8], [27-35] [6], [36-42] [43-47] [48-52] [53-59]
It was already established in Section 2.2 that the scope of a batch manufacturing system is wider in terms of not just encompassing a range of manufacturing set ups but for a realistic need of a CCS. The effectiveness of such a CCS was then linked with the effectiveness of a cost estimation methodology. Now, that the need to develop a cost estimation methodology for a generic system is evident from the view point of maximizing the application areas already mentioned, the only manufacturing system left for exploring the possibilities for a methodology development is a batch type environment.
2.6
Conclusions
This chapter presented a comprehensive survey of literature in the area of cost estimation in order to establish a background in the research area and to develop an 53
understanding of the problems. The techniques used for cost estimation in various application areas were investigated.
The chapter started with defining basic concepts of cost and costing. Manufacturing systems were discussed with cost control implications. Flexibility and productivity were discussed and their relationship was detailed. It was noted that various combinations of production volumes and product varieties resulted in three main categories of manufacturing environment: mass, batch and job shop production. It was found that batch production environment demanded an effective CCS and was suggested as a desirable area for developing an estimation technique. Cost estimation, its functions and the significance of early estimation were discussed.
A comprehensive literature review with a focus on four main application areas was presented. Techniques from each application area were thoroughly analysed. They included techniques for specific segment in a production cycle, techniques for specific machining and manufacturing processes, techniques for specific parts and products and finally the techniques for generic systems. It was noted that techniques developed for generic systems did have the potential to fulfil the demands of the other application areas. In order to maximize the scope of an estimation methodology, it was deemed necessary to develop a technique for a generic system. Batch manufacturing environment was considered as a viable generic system for which a cost estimation technique could be developed. Finally, the published literature references mentioned for various applications of cost estimation were summarized for any useful research in future.
54
Chapter 3
PCE Technique Classification System
Chapter 3 provides a technique classification system based on the state of the art on product cost estimation. PCE techniques are classified to form an extensive hierarchical classification system. The developed classification system is further used to form a decision support system for methodology selection. A case-based model is also developed for effective utilization of past product details. Key advantages and limitations of the techniques are discussed along the course. The chapter concludes with the suggestion of developing a methodology for cost estimation based on combining concepts from both qualitative and quantitative techniques for early and accurate estimation of a products cost.
3.1
Introduction
Product cost estimation refers to predicting all the costs associated with manufacturing a product from raw material to a finished product. Accurate product cost estimates are important in establishing a good CCS and help in setting competitive price plans. The role of PCE is, therefore, to predict the overall product cost likely to be incurred throughout the
55
Chapter 3: PCE Technique Classification System
entire production phase. The accuracy of the predicted costs will determine the effectiveness of the technique employed and, combined with the information access, can help the entire production enterprise achieve its corporate goals. It, therefore, not only serves operational matters (such as cost control, resource allocation, price setting etc.) but is also deemed as a key component for strategic production planning and control system.
The published literature on PCE covers a wide variety of issues ranging from manufacturing cost estimation of standard mechanical components to cost analysis of highly customized assembled products, from process cost optimisation techniques to specific methods for overhead costing, from unique approaches for estimation at the conceptual design stage to general costing rules designed for use at a later stage in the design cycle and also from classical costing methods to highly novel cost estimation techniques. Several textbooks [60-62] can be found on some of the subjects.
Due to the significance of the cost estimation process in the organizational structure of an enterprise, the influence of an effective estimation methodology can not be underestimated. On the contrary, the availability of an extensive range of estimation techniques makes the selection of an appropriate methodology to suit the prevailing conditions a daunting task. The classification of these techniques based on certain criteria is a step towards overcoming the issue. A number of researchers have attempted to categorize the PCE techniques using certain criteria. Zhang et al. [40] categorized some techniques into traditional detailedbreakdown, simplified breakdown, group-technology based, regression-based and activitybased cost estimation techniques. Ben-Arieh and Qian [11] classified cost estimation 56
methods into intuitive, analogical, parametric and analytical methods. Shehab and Abdalla [63] mentioned intuitive, parametric, variant-based and generative cost estimating approaches without defining them clearly. The same authors [64] later classified cost modelling approaches at the design stage into knowledge based, feature based, function based, and operations based approaches. Cavalieri et al. [2] identified three approaches for cost estimation: analogy-based, parametric and engineering approaches. However, a comprehensive hierarchical classification of the estimation techniques has not been exploited.
3.2
Development of hierarchical classification system (HCS)
The availability of a wide range of PCE techniques demands an extensive classification system in order to facilitate the selection of a suitable methodology for a given condition. The existing classifications fall short of furnishing a complete framework for methodology selection. These classifications group together the techniques with similarities in separate categories but do not consider the differences within the respective groups. Filtering a methodology is, therefore, constrained. The development of HCS is aimed at providing a framework for not only classifying the PCE methodologies but facilitating their selection to suit different problems. The developed system comprising an extensive hierarchical classification is based on grouping the techniques with similar features into various categories. The methodologies discussed in different categories are distinct and reflect the nature of that category. An effort is also made to elaborate each group or category with
57
reference to published work. Meanwhile, mathematical models are presented on occasions with particular references in order to better understand the nature of a given category. The hierarchical classification expands due to the identification of the variations within a group. The PCE techniques are broadly classified into qualitative and quantitative techniques followed by an extensive sub-classification.
Qualitative cost estimation techniques are primarily based upon a comparison analysis of a new product with the products that have been manufactured previously in order to identify the similarities in the new one. The identified similarities help to incorporate the past data into the new product so that the needs to obtain the cost estimate from scratch are greatly reduced. In that sense the past design and manufacturing data or previous experience of an estimator can provide a useful help to generate reliable cost estimates for a new product which is similar to a past design case. Sometimes, this can be achieved by making use of the past design and manufacturing knowledge encapsulated in a system based on rules or decision trees etc. Historical design and manufacturing data for products with known costs may also be used systematically to obtain cost estimates for new products. For example, regression analysis models and neural network approaches could provide an efficient way to predict costs for new products by using historical cost data. In general, qualitative techniques help obtain rough estimates during the design conceptualisation. These techniques can further be categorised into Intuitive and Analogical Techniques, which are discussed in detail in Sections 3.3 and 3.4 respectively.
58
Quantitative techniques, on the other hand, are based on a detailed analysis of a product design, its features and corresponding manufacturing processes instead of simply relying on the past data or knowledge of an estimator. Costs are, therefore, either calculated using an analytical function of certain variables representing different product parameters or as the sum of elementary units representing different resources consumed during a whole production cycle of a given product. Although, these techniques are known to provide more accurate results, their use is normally restricted to the final phases in the design cycle due to the requirement of a detailed product design. Quantitative techniques can be further categorised into Parametric and Analytical Techniques, which are discussed in detail in Sections 3.5 and 3.6 respectively.
These techniques categorised as qualitative and quantitative can be illustrated in a tree diagram in Figure 3-1 showing the preliminary classification of PCE techniques.
Product Cost Estimation Techniques
Qualitative Techniques
Quantitative Techniques
Intuitive Techniques
Parametric Techniques
Analogical Techniques
Analytical Techniques
Figure 3-1: PCE Preliminary Technique Classification
59
3.3
Intuitive cost estimation techniques
The intuitive cost estimation techniques are based on using the past experience. A domain experts knowledge is systematically used to generate cost estimates for parts and assemblies. The knowledge may be stored in the form of rules, decision trees and judgments etc. at a specific location, e.g. a database, to help the end user to improve the decision making process and prepare cost estimates for new products based on certain input information. The present study identified three sub-categories under intuitive techniques.
3.3.1
Case-based methodology
This approach also known as case-based reasoning (CBR) attempts to make use of the information contained in previous design cases by adapting a past design from a database that closely matches the attributes of a new design. This often requires making necessary changes to parts and assemblies of previous design cases and incorporating missing details to it. Figure 3-2 shows a complete framework for the case-based approach with the dotted lines representing the cost interfaces to the system. The process starts by outlining a new products design specifications followed by retrieving a closest design match from a design database. The system then attempts to find the changed assemblies and subsequently changed parts in the assemblies. Changes are incorporated in the design either by retrieving similar parts and/or similar assemblies from the design database or by designing the new ones altogether. All the necessary changes are incorporated in a similar way until the new design conforms to the outlined design specifications.
60
Figure 3-2: Flow Diagram of the Case-Based Approach for Cost Estimation
61
The new design is later stored in the design database. This technique allows the cost estimation for a new product by combining the past results with those for the newly designed components and assemblies, thereby greatly reducing the need to design from scratch. The approach is, therefore, helpful in making good estimates at the conceptual design stage, since the use of the past cost data to generate new estimates greatly minimizes the estimation time. However, the methodology is applicable only when similar past designs are available to incorporate the relevant cost data during cost estimation for new products.
A typical example can be seen in the methodology presented by Rehman and Guenov [3] in which an attempt is made to predict design features from incomplete design descriptions based on past designs and production knowledge. In their work, a system allows the retrieval of past design cases that match the new problem description. The cost modeller detects necessary modifications in the retrieved designs and the cost data is updated accordingly using the adaptation rules stored within the design models. The method, thus, allows the cost estimation and evaluation for innovative designs. The method applied by LiHua and Yun-Feng [65] evaluated costs for new products by implementing the functions of CBR. These functions included organizing case bank, indexing case, initializing case, seeking and searching case and adapting case. The method was useful for rapid costing to satisfy customers demands on pricing. Ficko et al. [66] conceived a CBR system for predicting total cost of the tool manufacture. The system is based on extracting geometrical features from CAD-models stored in a database and calculating the similarities with the problem description of a new products features.
62
Although, the developed system is only limited to tools for manufacture of sheet metal products by stamping, it provides good-quality predictions based on enough similar cases. Balarman and Vattam [67] analyzed other applications of CBR in the domains of helpdesks, diagnosis, cost estimation and design based on its functions such as representations, indexing, matching, adaptation and process of problem solving. Their work is an effort towards building a general-purpose case-based problem solver.
3.3.2
Decision support systems (DSS)
These systems are helpful in evaluating design alternatives. The main purpose of these systems is to assist estimators in making better judgments and decisions at different levels of the estimation process by making use of the stored knowledge of experts in the field. This is illustrated in Figure 3-3. One such system incorporating expert rules has been developed by Kingsman and De Souza [68] for cost estimation and price setting in versatile manufacturing companies dealing with make-to-order (MTO) systems. The developed system not only focused onto different factors that influence the decision making process in handling a customer enquiry but discussed the rules that cost estimators apply when making decisions about these factors.
To incorporate experts experience, the artificial intelligence (AI) philosophy is used to represent and utilize a domain experts knowledge in a way that is oriented towards problem solving and serves as a decision aid tool. In the particular context of the PCE, for example, it may constitute a segment of the system containing information about machining
63
processes, manufacturability analysis and constraints, product characteristics with design functions and relationships with each other set out in logical statements. It may also incorporate rules about the actions to be taken or more conventional mathematical formulae. It can point outside to external programs and databases that can be associated with it including some that can cope with uncertain or conflicting judgments.
Figure 3-3: Decision support system approach to cost estimation
64
Shehab and Abdalla [69] developed knowledge-based cost models for the PCE in early design stages whereas Luong and Spedding [24] developed a knowledge-based system by integrating process planning into cost estimation. Another approach adopted by Gayretli and Abdalla [4] focussed on developing a prototype system for manufacturing process optimization. The system assisted designers to create real-time cost estimates and feasible process plans by retrieving manufacturing form features and parameters from the feature database.
One of the most common ways to represent DSS is based on storing design, manufacturing or other constraints as a set of rules. Since many practical situations deal with uncertainty and non-availability of heuristic data, fuzzy logic techniques are used to some extent to overcome such problems. Another non-conventional approach makes use of Expert systems (ES) or Expert Support Systems in the domain of DSS.
Rule-based systems These systems are based on process time and cost calculation of feasible processes from a set of available ones for the manufacture of a part based on design and/or manufacturing constraints. Such a system reflects these constraints in a respective rule class with the information encapsulated in it by an expert in the area. A rule-based algorithm is an example of one such approach, which helps to establish design and manufacturing constraints. This approach is shown diagrammatically in Figure 3-4. Based on a set of user constraints, manufacturing processes are selected, which are then used to calculate the
65
product cost. The set of constraints may need to be changed to obtain a different set of manufacturing processes to obtain an acceptable product cost estimate. This methodology is helpful for cost optimization based on process evaluation criteria. However, obtaining the optimized results can be very time-consuming especially, when there are a large number of processes to be evaluated.
Figure 3-4 Cost Estimation Process Model Based On User Constraints
Gayretli and Abdalla [70] developed a rule-based algorithm for the selection and optimization of feasible processes to estimate process time and cost based on parts features.
66
A detailed description of part features with possible processes and constraints was given. Process times were calculated using a standard formula as follows:
Pr ocessTime =
FormFeatureVolume Material Re movalRate
(3-1)
The process time is then used to calculate Lot Time, which is based on a form feature quantity. The total process cost is subsequently calculated as follows:
Total Process Cost = Lot Time PHC
(3-2)
Where, PHC is the Productive Hour Cost given by a cost estimation database [71]. The total cost is then calculated as follows:
Total Cost = Material Cost + [(Lot Time PHC) + Tool Cost + Setup Cost]
(3-3)
The proposed system allowed the selection of a combination of feasible processes from the possible ones and hence the calculation of process time and cost based on the user input constraints, e.g. maximum allowable cost and process time for a particular feature. A criterion of feasibility was judged against the level of satisfaction for input constraints. The process allowed flexibility based on user constraints. Another example of this category can be found in [3] where manufacturing and assembly rules were used to update cost data in
67
the proposed system, whereas, an object-oriented and rule-based system can also be found in [64] for product cost modelling and estimation.
Fuzzy logic approach

This approach to cost estimation is particularly helpful in handling uncertainty. Fuzzy rules such as those for design and production are applied to such problems to get more reliable estimates. However, estimating the costs of objects with complex features using this approach is quite tedious and requires further research in the area. Shehab and Abdalla [63] used fuzzy rules with linguistic expressions and assigned truth-values to them. They used several steps to develop a fuzzy logic model. These steps were fuzzification of input variables followed by fuzzy inference based on a set of rules and finally defuzzification of the inferred fuzzy values. A fuzzy technique consisting of a decision table providing a means for system rules and indicating the relationships between the input and output variables of the fuzzy logic system, is used to handle the uncertain knowledge on cost estimation. The construction of a set of rules from the decision table enables the estimation of the machining time (Ti) for a given feature, which is multiplied by the unit time cost (Ri) to get the machining cost (Cm) for that feature, i.e.:
Cm = Ri Ti
(3-4)
The developed fuzzy logic based system was capable of estimating the total product cost apart from enabling the material selection and estimating the assembly cost. The same
68
investigators [64] carried out a similar but more comprehensive study by considering other essential costs such as non-productive and set up costs.
Expert systems
This approach is based on storing the knowledge in a database and manipulating it on demand to infer quicker, more consistent and more accurate results based on an attempt to mimic the human expert thought process with the help of an automated logical reasoning approach, normally achieved by rule based programming. Within the specific context of cost estimation, the expert system approach refers to a model and associated procedure exhibiting a degree of expertise comparable to that of a human expert in generating or to help in generating reliable cost estimates. Expert systems applied to the PCE have mainly focussed on formalising the theoretical techniques largely from textbooks etc. rather than encapsulating the practical knowledge (e.g. the expert conceptual estimator developed by Musgrove [72]). Further research in the area considering the human contemplating process of an estimator has a better potential to exploit the typical characteristic. Cost estimation methods for various applications using expert systems or expert support systems can be found in [73, 74].
69
3.4
Analogical cost estimation techniques
These techniques employ similarity criteria based on historical cost data for products with known cost such as regression analysis models or back propagation methods. The following sub-sections describe these methods in detail.
3.4.1
Regression analysis models
These models make use of the historical cost data to establish a linear relationship between the product costs for the past design cases and the values of certain selected variables so that the relationship can be used to forecast the cost of a new product. The regression analysis approach based on the similarity principle was adopted by Hundal [75] and Poli et al. [42] to use a basic cost value and consider the effects of variable cost factors by assuming linear relationships between the final product cost and the cost factors. Lewis [76] further used existing designs to provide cost estimates for similar new designs whereas Pahl and Beitz [5] provided more general costing approaches based on similarity.
3.4.2
Back-propagation neural network (BPNN) models
These models use a neural network (NN) that can be trained to store knowledge to infer the answers to questions that may even not have been seen by them before. This means that such models are particularly useful in uncertain conditions and are adaptable to deal with non-linearity issues as well. The back-propagation neural network (BPNN) is the most
70
common of all network types and also suits better the nature of the PCE. The application of neural network in cost engineering is discussed in [77].
Shtub and Zimerman [78] compared the cost results obtained with the regression model and the back-propagation neural network model and observed the superiority of the latter in many ways. In another study [40], a featured-based methodology was proposed using BPNN to estimate the cost of packaging products. Cost-related features of packaging products were used in conjunction with historical cost data to obtain a relationship between cost and cost-related features based on BPNN. The proposed method overcame the limitations of regression analysis models such as the assumption of non-linear relationships between product cost and its variables as well as those of traditional breakdown approaches, e.g. the requirement of detailed cost information like process planning cost. Zhang and Fuh [41] proposed a similar approach for early cost estimation, whereas Chen M-Y and Chen D-F [79] proposed a BPNN model for strip-steel coiler. Further, a backpropagation algorithm [2] was used with momentum and a flat spot elimination term for a Multilayer Perceptron (MLP) Neural Network, in which neurons are organized in several layers including an input layer, a number of hidden layers and an output layer.
3.5
Parametric cost estimation techniques
Parametric models are derived by applying the statistical methodologies and by expressing cost as a function of its constituent variables. These techniques could be effective in those
71
situations where the parameters, sometimes known as cost drivers, could be easily identified. Parametric models are generally used to quantify the unit cost of a given product. Cavalieri et al. [2] developed a parametric model for the estimation of unit manufacturing costs of a new type of brake disk using the weight of the raw disk, unit cost of raw material and the number of cores as parameters in their model, which is expressed as follows:
C TF C = FC + C co N co + rm W 1 SC
(3-5)
where, C = Unit cost of disk brake, FC = Fixed Cost Factor (Coefficient), Cco = Core Cost per Kg of Cast iron (Coefficient), Nco = Number of cores, Crm = Unit cost of raw material, SC = Scrap rate (Coefficient), TF = cast iron / steel conversion factor (Coefficient), W = weight
A simple linear regression model using one of the cost drivers would not be effective because of variances between the data. However, the developed model overcame this problem by using more parameters. Validation analysis of the model by comparing the estimated costs with the actual ones of the brake disks demonstrated the superiority of the proposed parametric model over the linear regression model.
A wide range of parametric models can be found in the literature. For example, Hajare [80] modelled parametric costing of components using the product specifications. Roberts and 72
Hermosillo [81] used approximate tool paths and process parameters from available factory resources to estimate time and cost for small surface units. Boothroyd and Reynolds [82] adopted a parametric costing approach using the volume of typical turned parts as a parameter to estimate the cost in the early design stages. Unlike the detailed-breakdown approach, the method adopted by them could be used in the early design stage without the need of a process plan. Similar work can be found in [83].
3.6
Analytical cost estimation techniques
This approach requires decomposing a product into elementary units, operations and activities that represent different resources consumed during the production cycle and express the cost as a summation of all these components. These techniques can be further classified into different categories, which are discussed in detail below.
3.6.1
Operation based approach
This approach is generally used in the final design stages due to the type of information required and is one of the earliest attempts to estimate manufacturing costs. The approach allows the estimation of manufacturing cost as a summation of the costs associated with the time of performing manufacturing operations, non-productive time, and set-up times. Several techniques have been developed to select the alternative manufacturing operations that optimise the machining cost.
73
The cost model proposed by Jung [84] estimated the manufacturing cost by considering three different times including set up time, operation time and non-operation time. Formulation was provided. The total cost was given by:
Mfg cost = (Ro + Rm)[( Tsu/Q)Tot + Tno] + material cost + factory expenses.
(3-6)
where, Ro = operators rate, Rm = machine rate, Tsu = set-up time, Q = batch size, Tot = operation time, Tno = non- operation time.
The model could not be used to evaluate design alternatives due to its availability only in the final stages of design cycle. Feng et al. [12] presented a digraph based mathematical model that uses the geometric features including cylinder, rectangular block, chamfer, flat surfaces and hole, and developed an algorithm to estimate the minimum cost using an operation-based approach. The process plans of alternative design solutions with explicit modelling of the machining time of various features represented the criteria of estimating manufacturing costs. Gupta et al. [13] developed a similar methodology using manufacturing features for the evaluation of alternative process plans to estimate the manufacturing cost of the part. Wei and Egbelu [85] used geometric design data and developed a method based on a tree representation of alternate processes to estimate the product manufacturing cost. Although the approach focussed on obtaining the optimum results, it did not consider direct labour cost. Further, Kiritsis et al. [14] proposed a method for the cost estimation of the machining of parts based on the description of given features and associated alternative manufacturing operations. The proposed methodology was based 74
on Petri nets to determine overall costs including machining cost, moving cost, setup cost and tool change costs. However, getting the optimised results using the proposed methodology was time consuming.
3.6.2
Breakdown approach
This method estimates the total product cost by summing all the costs incurred during the production cycle of a product, including material costs and overheads. The method requires detailed information about the resources consumed to manufacture a product including purchasing, processing and maintenance details.
The cost model developed by Son [86] included labour costs, machining cost, tool cost, set up cost, space occupied cost, computer software cost and material cost. The model also separated the raw material cost and the labour cost into different categories. The proposed model included insurance, utility, maintenance, repair and property costs. The machining cost (Cm,), is hence represented in the following equation as
Cm= (utility cost) + (maintenance cost) + (repair cost) + (insurance cost) + (property cost)
= (CuTm + CmtTmt + CrTr + aFk + bFk)
(3-7)
75
where, Cu = utility cost per unit time, Tm = machining time, Cmt = maintenance cost per unit time, Tmt = total maintenance time, Cr = repair cost per unit time, Tr = total repair time, a = insurance premium, Fk = initial investment, and b = property tax.
Further, equations for other cost elements including labour costs, tool cost, set up cost, space occupied cost, computer software cost and material cost were also provided. The requirement of such detailed information restricted the use of the model in the final design stage. Further, manufacturing costs were considered by Bernet et al. [43] as the sum of material, labour and overhead costs and Ostwald [60] estimated product cost as the summation of material cost, manufacturing cost, labour cost and overhead expenses based on hourly usage of machinery or direct labour. Such traditional cost estimation and cost accounting techniques were also discussed in detail in [61].
3.6.3
Tolerance-based cost models
The objective of such models is to estimate product cost considering design tolerances of a product as a function of the product cost.
Singh [87] presented a framework for the concurrent design of product and processes considering the criteria of minimum cost, maximum quality and minimum manufacturing lead time. Three models were presented to jointly design the products and processes. They are unit cost of production model, the quality model and the lead time model. The unit cost model was expressed as follows:
76
X0 (d, j) = Ki (d, j) [Xi + f(j)] Ks (d, j)Xs
(3-8)
where, Ki and Ks are technology coefficients that can be found from the following equations:
Ki = 1/ [1- SC (d, j)]
(3-9)
Ks = SC (d, j)/ [1-SC (d, j)]
(3-10)
SC is the scrap rate given by the following equation:
SC (d, j) = [-d / (j)] + 1 [d / (j)]
(3-11)
where, j = jth manufacturing process selected for producing a product, Xo (d, j) = the unit cost with tolerance d, Xi = the unit raw material cost, f(j) = the unit processing cost for jth process, Xs = the unit salvage value, Ki = technology coefficient (input), Ks = technology coefficient (scrap), SC (d, j) = scrap rate, (j) = standard deviation, (x) = Cumulative distribution function of probability distribution with the mean equal to 0 and the standard deviation equal to 1.
The modelling methodology was based on obtaining the optimal tolerances and hence setting up the acceptance regions for the design variables meeting certain criteria. The
77
objective of the cost model was to select the process and design variables that minimize the cost function. However, the modelling methodology eliminated the needs for design changes because it considered various design and manufacturing factors at the early stage of the design. The cost-tolerance relationships and relevant models can also be found in [88, 89].
3.6.4
Feature-based cost estimation
The feature-based cost estimation methodology deals with the identification of a products cost related features and the determination of the associated costs. A considerable research has been carried in order to extract and quantify representative product features that contribute to the total cost. These features can either be design related such as the type of material used for a specific product, geometric details, etc. or process oriented, i.e. a particular process required for manufacturing the product, e.g. machining, casting, injection-moulding etc. The methodology allows the selection of a particular design or manufacturing form feature for DFC system users. However, the approach can have limitations for complex or very small geometric features especially if machining processes are used to produce these features.
Zhang et al. [40] proposed a feature-based cost estimation system for packaging products extracting 32 cost related features in both the design and manufacturing domain. These features were then quantified based on a relative cost influence among the various possible states of a particular feature. However, no attempt was made to quantify these features
78
objectively. Ou-Yang and Lin [6] looked into the feature-based costing by focussing on the machining type features and developed a manufacturing cost estimation model based on feature shapes and precision. With the process planning information and geometrical data, the machining time of a feature was estimated. One limitation of their proposed framework was that it only considered conventional machining processes. Further research work in the area of feature-based cost estimation can be found in [90-92].
3.6.5
Activity-based costing (ABC) system
The ABC system focuses on calculating the costs incurred on performing the activities to manufacture a product. The method was first discussed by Cooper and Kaplan [93]. They presented the ABC system as a useful means to distribute the overhead costs in proportion to the activities performed on a product to manufacture it. Hundal [94] presented similar methods. The ABC system proved a good alternative to traditional estimation techniques since it provided more accurate product manufacturing cost estimates [95].
Various information sources for the implementation of the ABC system within a specific context can be found in [96-101]. The effectiveness of the ABC system was discussed by Kaplan [102] in providing helpful cost information to product designers for developing economic designs. The capabilities of the ABC were investigated by Tornberg et al. [103] with a particular emphasis on providing useful cost information to product designers. The study focused on modelling product design, purchasing and manufacturing processes with graphic flow charts in the form of activity chains. With the help of the process models and
79
the activity-based cost calculations, product designers were able to estimate the effects of different design options on product costs. The work of Tseng and Jiang [104] attempted to combine the feature-based costing methodology with the ABC approach. Their ABC analysis model could evaluate different manufacturing costs for multiple feature-based machining methods. Yang et al. [105] used process planning, scheduling and cost accounting information to estimate manufacturing and machining cost through an activitybased method. Other examples of manufacturing and machining cost estimation using the ABC approach can be found in [106, 107].
Some other researchers used the ABC approach to model the manufacturing costs in a specific manufacturing set-up. For example, Koltai et al. [108] estimated costs for flexiblemanufacturing systems based on the ABC analysis, whereas Aderoba [48] developed an ABC model for job shops. The latter was based on the classification of all the activities into machine-based production, labour-intensive production, technical services and
administrative services. Cost rates for all such activities were provided which were then used to estimate the cost of a new order. For example, the cost of a machine-based production activity (C) was given as follows
C = [(M+m)tm + (L+l)tl + btb + utu]
(3-12)
where, M = cost rate for machines in the activity per hour, m = periodic rate for complimentary tools in the activity, L = labour rate for direct worker on machine activity, l = labour rates for ancillary worker on machine activity, b = building space rate for machine 80
Chapter 3: PCE Technique Classification System activity, u = utility rate for machine activity, tm = machining time, tl = labour working time, tb = time spent on building space occupied by activity, tu = time of utility being used.
Expressions were also provided for the cost rates for machines and tools, the labour rates for direct and ancillary worker, the building space rate, and the utility rate, which are not described here for brevity of presentation. The proposed method proved useful in highlighting high cost elements; however, its accuracy depended on how reliable the activity time estimates for a new product were.
3.7
Conclusions
This chapter extensively reviewed the pertinent literature on manufacturing and product cost estimation along with a critical evaluation of some of the techniques developed in the area. An extensive classification scheme is developed. A pictorial representation of the classification system adopted in the present study is given in Figure 3-5. The horizontal dotted lines are used to show the different levels in the hierarchical tree diagram proposed.
The techniques were classified into two main groups as qualitative and quantitative, which were then subdivided into two categories each. The chapter examined all the categories in detail with references to the published literature. Mathematical models were presented on some occasions to illustrate certain techniques. In addition, the PCE techniques discussed
81
in the chapter are tabulated together with the key advantages, limitations and corresponding published literature in Table 3-1.
The study of individual techniques also revealed the key conditions under which they can be applied. The conditions can be grouped together to form a decision support model (DSM) for cost estimation methodology selection and is presented in Figure 3-6. The developed model is a helpful tool for estimators in making decisions about selecting a suitable estimation methodology. It can be observed that a particular technique linked with a specific class is more applicable in certain situations. During the early phases of the design cycle, when limited data is available, qualitative cost estimation techniques are more appropriate and provide a helpful starting point for a detailed analysis at a later stage. For example, the proposed case-based methodology systematically makes use of available past data to generate estimates for a similar new product. One problem linked with such techniques is the limited availability of past data, which is overcome to some extent by making use of the past experience or knowledge of the estimator generally encapsulated in the form of decision rules. Qualitative techniques, therefore, are helpful either in furnishing rough cost estimates or serve as a decision aid tool for designers or estimators especially during the early phases of design process. However, when the detailed design becomes available, quantitative techniques provide more accurate estimates, which are necessary for factors like design rationalization and determination of profit margins etc. The data requirements restrict the use of such techniques in the final phases of design and development process. Techniques such as the ABC systems overcome the problem to some extent by making use of the pre-determined activity rates to calculate the total amount of
82
activities consumed to manufacture a product rather than requiring any detailed design and manufacturing information. This, however, requires lead times for individual products in the early design stages, which may be obtained using methodologies such as the case-based approach. Therefore, a combination of the two approaches, the qualitative and the quantitative techniques, could play an important role in developing a cost evaluation system capable of providing useful cost information on various stages of design and development phases.
The proposed classification system and the developed decision support model are aimed at supporting the decision making process of the estimators and designers. The two elements are formulated to provide guidelines to the users for the selection of an effective estimation methodology. A rightful selection is eventually an aspect of a CCS.
83
Level 1
Level 2
Level 3
Regression Analysis Model Decision Support Techniques
Back-Propagation Neural Network Model OperationBased Approach Breakdown Approach
Case-Based Technique
ToleranceBased Cost Models
Activity-Based Cost Estimation
Feature-Based Cost Estimation
Level 4
Rule-Based System
Fuzzy Logic System
Expert System
Figure 3-5: Classification of the PCE Techniques
84
Table 3-1: The PCE techniques; key advantages, limitations and list of discussed references
Product Cost Estimation Techniques Intuitive Cost Estimation Techniques Case-Based Systems
Key Advantages Innovative design approach Can provide optimized results Handles uncertainty, reliable estimates Quicker, more consistent and more accurate results Simpler method
Limitations Dependence on past cases
References [3], [65-67] [3], [64], [70] [63, 64]
Qualitative Cost Estimation Techniques
Decision Support Systems
[4], [24], [68-69]
Rule-based Systems Fuzzy logic systems Expert Systems
Time-consuming
Estimating complex features costs is tedious Complex programming required
[72-74]
Analogical Cost Estimation Techniques
Regression Analysis Model Back Propagation neural network model
Limited to resolve linearity issues Completely datadependant, Higher establishment cost Ineffective when cost drivers can not be identified Time-consuming, require detailed design and process planning data
[5], [42], [75, 76] [2], [41], [77-79],
Deal with uncertain and non-linear problems Utilize cost drivers effectively
Parametric Cost Estimation Techniques Quantitative Cost Estimation Techniques Operation-based cost models Analytical Cost Estimation Techniques
[2], [80-83]
Alternative process plans can be evaluated to get optimized results Easier method
[12-14], [84, 85]
Breakdown cost models
Detailed cost information required about the resources consumed Require detailed design information
[43], [60, 61], [86]
Cost tolerance models
Cost effective design tolerances can be identified. Features with higher costs can be identified Easy and effective method using unit activity costs
[87-89]
Feature-based cost models
Difficult to identify costs for small and complex features Require lead times in the early design stages
[6], [40], [90-92]
Activity-based cost models
[48], [93108],
85
Start
Input for selected methodology Selected methodology Decision Node / Available alternatives
NO
Apply quantitative techniques
Estimates at conceptual design stage?
YES
Apply qualitative techniques
Back propagation algorithm

YES
NO
Parametric relation available?

NO
YES
Past experience available?
Past product data available?
Apply analogical techniques
YES
NO
Apply analytical techniques
Apply parametric model
Apply intuitive techniques

NO
NO
Resolving linearity?
Apply BPNN models
Process planning details / operation times available Cost elements breakdown available Cost as a function of design tolerances Individual features cost known Unit activity costs known
Operation-based models
Breakdown models Tolerance-based models Feature-based models ABC models
Product parameters. Shape, Size, Weight etc. Past experience stored in AI technology? Rule-based algorithm, Fuzzy logicbased system, Expert system
Past product cases available?
YES
Apply regression models

NO YES
Apply casebased technology Matching algorithm
YES
Linear relationship using historical cost data
Apply DSS Stop
Figure 3-6: Decision Support Model for cost estimation methodology selection
86
Chapter 4
MRO and TRO Estimation Methods
Overhead is a major element of the manufacturing cost. The chapter identifies the need to implement an improved overhead estimation methodology based on dividing manufacturing overheads into time- and material-related costs. New overhead estimation methods called material-related overhead (MRO) and time-related overhead (TRO) are, therefore, introduced, developed and implemented in an electrical engineering company for a four-year period. The results based on a four-year retrospective validation analysis confirm the superiority of the proposed methodology over the existing one, as overheads are more accurately estimated.
4.1
Introduction
The aim of Chapter 4 is to present material-related overhead (MRO) and time-related overhead (TRO) estimation methods. This is essential to lay the foundation for developing a comprehensive methodology for early and accurate estimation of a product cost. It has already been considered in previous analysis that early and accurate estimation would require adopting a hybrid approach combining the elements from qualitative and quantitative techniques. It has also been shown that developing a methodology for estimating a products cost in a batch type manufacturing environment 87
Chapter 4: MRO and TRO Estimation Methods
is in conjunction with a greater need for the system than for either mass production or job shop system. Further, the development of a methodology to predict an entire products cost rather than part or component costs demand a careful analysis of the developed classification system. Breakdown approach presents the entire product cost by summing up all the costs incurred during the production cycle of a product and could, therefore, hold the key to start the development process. In a broader sense, the development of a hybrid system for product cost estimation in a batch type manufacturing environment is, therefore, congruent to the research area.
Developing such a system, however, requires identifying the elements from the qualitative and quantitative techniques for hybrid approach. One of the identified elements is from the quantitative techniques and is based on predicting an entire products cost. Breakdown approach requires defining the total cost as a summation of the constituent elements. Before these elements and the total cost are defined; there is a need to understand any existing breakdown models and to identify problem areas.
Product cost consists of general administrative costs, engineering costs (costs incurred during design and development phases including those linked with customer requirements, designing of the part, its process planning, and prototyping), and manufacturing costs. The latter is the largest element of the overall product cost and is mostly determined during the product design phase. Better understanding and control of this cost element is a major step towards achieving an effective CCS. Figure 4-1 shows a breakdown of the selling price and manufacturing costs of a typical product [109]. It can be seen that the manufacturing cost accounts for 40 percent of the selling price. Half of this cost is incurred on parts and material, whereas direct labour accounts for 12 88
percent. Manufacturing overheads (indirect labour, indirect materials, plant and machinery depreciation, energy costs, etc.) account for 38 percent of the manufacturing costs, which is a substantial figure.
Figure 4-1: Break down of Selling Price and Manufacturing costs
Manufacturing overhead is a significant contributor to the overall product cost and refers to all the production support costs incurred on or off the shop floor and generally for the shared benefit of several products within a manufacturing facility. A traditional approach to estimating such costs relies on summing up all the expenses, other than those included in direct material and labour, and ascertaining a cost rate. The rate is then used to estimate the overheads for a new product by multiplying it with the expected manufacturing lead time (MLT) of the new product. A major disadvantage of this methodology is a disproportionate allocation of the total manufacturing overheads to different products with respect to the actual resources consumed by individual products. Furthermore, since the overheads recovery is based only on MLT and ignores material quantities, costs are underestimated for products with higher material requirements and overestimated for those with lower material quantities.
89
Since, overhead is a major element of the overall product cost and estimating it accurately has a direct bearing on the overall estimated cost value, the current chapter will focus on developing a new overhead estimation methodology. The proposed methodology will then form the basis for the development of a model for the estimation of the overall product cost. The methodology developed in this chapter divides overheads into time- and material-dependent cost elements and overcomes the above mentioned limitations by providing a framework for apportioning the total manufacturing overheads to individual products based on the products degree of consumption of manufacturing activities and resources.
Product cost depends greatly on the geographical location of an enterprise owing to factors (such as governmental policies, taxation, political situation, weather, labour cost, and material availability) that could significantly alter the manufacturing cost of a product on different locations around the globe. The contributions of individual subelements (material costs, direct labour, and overheads) towards the overall manufacturing cost also vary within this context. It is not surprising that many western manufactures have moved their operations to Asian countries.
With this in mind, the present study analysed manufacturing cost and its various subelements in an electrical engineering company in South Asia. It is also important to note that any validation through industrial trials will not only make an attempt to endorse the model but help to understand the implications of geographical locations. The developed mathematical model for overhead estimation is implemented retrospectively in the company for a comparative analysis covering a four-year period. The study sets
precedence for similar studies on other geographical locations with the aim of providing 90
a good foundation for comparative analysis in order to better understand manufacturing costs and develop an effective CCS.
4.2
Cost estimation methodology at the selected company
A well-established ISO9001 certified electrical engineering company was selected for the study as a representative case of the industrial sector in South Asia because it follows the standardized industrial code of practises acknowledged internationally. Because of the sensitivities around the confidentiality issues related to competitive cost data provided by the company, the companys name is not mentioned here. The company designs, develops, manufactures, and markets a wide range of technologically advanced electrical and electronic products and provides after sales services. The transformers manufacturing unit of the Power Transmission & Distribution (PTD) division of the company was chosen for the study, as it is the largest manufacturing unit of the company based on a batch type manufacturing set up. The unit deals with the manufacturing of distribution transformers (DT) and power transformers (PT). One of the major reasons for selecting the company in general and the transformer unit in particular is the units existing method of cost estimation that estimates the entire products cost based on a cost breakdown approach. Three major cost elements make-up the total manufacturing cost of transformers at the company: material costs, direct labour costs, and overheads. These are discussed in more detail below.
91
4.2.1
Material cost estimation
The material costs are estimated using the material quantities obtained directly from BOM (see example in appendix A) which is developed based on the engineering design details. Scrap margin is accounted for based on past trends. In most of the cases standardized transformers are manufactured and thus engineering designs are already established and so are BOMs. Even when a non-standard transformer is designed and developed, the quantities of different materials required are similar to a closest standard match even though the technical design of the product may be significantly different. Thus, the match provides a good starting point for the non-standard transformer to estimate material cost. As a result, material cost estimation is possible in the early stages of the design and development stage. In such circumstances, material cost calculation, based on past trends and certain commercial indicators, becomes standard. The following commercial indicators were used based on past trends in order to calculate material cost for the year 2003-2004:
15 percent increase in purchase price of locally sourced material; 10 percent increase in price of materials bought from foreign suppliers; Exchange rate = US $61.4 and 71 Euro (i.e. US $1.0 = Rupees (Rs) 61.4 and Euro 1.0 = Rs 71.0).
The cumulative material requirements (CMR) for transformer manufacturing for a given period of time change in accordance with planned orders. Thus, the factory-wide changes in material consumption over different time intervals can be used to reflect changes in production levels.
92
4.2.2
Direct labour costs
Work centres in both DT and PT comprise operators and machines. Direct Labour here accounts for the cost incurred on wages for total man-hours required for production. The total manufacturing lead time (MLT) for a transformer is calculated using standard routings and process times. These process times are established either by using standard formulas or by deploying time and motion studies. The lead time for a non-standard transformer is determined from the results for the closest standard match. In this way, the total number of man-hours required for manufacturing a transformer can be determined in the early stages of the design process and are subsequently multiplied by the wage rate to give the labour cost incurred. Setting the wage rate for a particular period of time is an organizational policy matter which is influenced by various factors (such as minimum wage rate set by the government, geo-political conditions, economic growth, etc.).
4.2.3
Overheads estimation
Overheads here refer to manufacturing overheads and account for costs other than direct material and direct labour. Since such costs represent more than a third of total manufacturing costs (see figure 4-1), estimating it accurately can significantly enhance the accuracy of the overall PCE process. The new methodology proposed in the study is implemented retrospectively for a four-year period starting from 1999-2000. The methodology employed by the company is discussed below in order to facilitate comparisons between the two methods and provide insight into the optimisation achieved through the implementation of the proposed methodology.
93
The existing method is based on summing up all the expenses other than those incurred on direct labour and material. For example, the costs incurred on handling, transportation, inspection, storage, inventory control, and purchasing of different materials are classified as overheads. Energy and other utility costs necessary to run different workstations are also included in this category. Several small components such as nuts, bolts, washers, etc. are classed as indirect materials and the costs incurred on these items form part of the overheads as well. By summing all the above-mentioned costs and dividing by the capacity hours, the overhead rate for the following year is calculated. Capacity hours in this context are the hours available for production. The rate is then used to estimate the overheads for a new transformer by multiplying by the MLT of the transformer. The total transformer manufacturing cost is finally estimated by adding the three cost elements: material cost, direct labour cost, and overheads.
The major drawback of this methodology is that the allocation of overheads to an individual transformer in most cases is not in proportion to the material and energy consumed for its production. For example, a transformer with a comparatively lower lead time is allocated lower overheads even though the quantity of material required for its manufacture is higher. Thus, the amount of material required for the manufacture of a transformer is ignored with respect to overheads allocation. This results in underestimation for transformers with higher material requirements and overestimation for those with lower material requirements.
94
4.3
Proposed methodology for overheads estimation
A careful analysis of the overheads at the selected company revealed that they were largely driven by either material quantities or lead times. In some cases, however, material quantities and lead times may interchangeably influence the same overhead cost element. For example, the cost incurred on tools and equipment is greatly affected by lead times which in turn may be linked with the amount of material to be processed. However, based on consistency, a more direct and stronger link can be established only between tooling costs and lead times as in some cases lead times may be independent of material quantities. The proposed methodology for overheads allocation and estimation is, therefore, based on dividing the overheads into MRO and TRO. This allows more realistic allocation of overheads to individual products by taking care of not only their lead times but their respective material quantities. Mathematical models are developed for the estimation of these overheads in the present study and are validated against transformer data obtained from the selected company.
4.3.1
MRO estimation model
MRO include the costs incurred on material handling and transportation, material
inspection, material storage and inventory control, material purchasing etc. It is important to note that activity times (such as time consumed for material handling, inspection etc.) are not same as MLT that form the basis for TRO. These activity times are effected by variations in material quantities and hence any costs associated with them are considered MRO. The costs incurred on indirect materials are also included in MRO. As the material consumption increases, MRO also go up. Total MRO consumed
95
during a certain period of time, therefore, can be expressed as a percentage of total direct material cost. A fraction can be set to express the same at the end of the (n-1)th year based on total MRO and the cumulative direct material cost obtainable at the end of the (n-1)th year from the cost accountancy data. The resulting MRO fraction (which is a reflection of the MRO as a percentage of the direct material cost) is then multiplied by the total material cost of an individual product to calculate the MRO for a new product
(OM) in the nth year. Total material cost for the product is based on the estimated
material quantities for that product obtained from the BOM and their respective unit costs, i.e.
Om = (C d 1 m1 + C d 2 m2 + .... + C dn mn )
n 1 1 n 1 n 1 (C im + Cn + C in 1 + C si S) ft + C P n 1 C dmt
(4-1)
Where,
Cd1, Cd2, ., Cdn = Unit costs of direct material 1, direct material 2, ., direct material
n respectively used in the manufacture of the new product; m1, m2, ., mn = amounts of
n 1 material 1, material 2, ., material n consumed; Cim = overall indirect material costs in
th 1 the (n-1)th year, C n year, ft = overall freight & transportation costs in the (n-1)
n 1 = overall purchase department costs in the (n-1)th year, Cin1 = overall inspection CP n 1 costs in the (n-1)th year, Csi = overall stores & inventory costs, S = Sale proceed of n 1 scrap, Cdmt = overall direct material costs in the (n-1)th year.
96
It can be observed that the model can easily accommodate any other MRO to adapt to the system where it is implemented. For example, if a systems quality control costs are material dependent, equation (4-1) can easily accommodate such costs to adapt to the new system. The percentage fractions of MRO calculated based on cost data (PT & DT) provided by the company for the year 2003-2004 are shown in Table 4-1. All the values given in the table are in (000) Rs. and the scrap value is given in brackets to denote subtraction in the model. The MRO for a new transformer can then be calculated using the percentages obtained and the amount of material used for its manufacture.
Table 4-1: MRO for power and distribution transformers (2003-2004)
Material Related Overheads for Transformers (2003-2004) Power Transformers (PT) Indirect Materials Sale Proceeds of Scrap Freight & Transportation Purchase Department Costs Incoming Inspection Stores & Inventory Control Total MRO Total Material Consumption MRO (2003-2004) 6,500 (1,000) 500 1,478 790 1,350 9,618 203,000 5% Distribution Transformers (DT) 19,500 (3,000) 1,500 4,433 2,370 4,050 28,853 445,700 6% 26,000 (4,000) 2,000 5,910 3,160 5,400 38,470 648,700 Total Transformers
97
4.3.2
TRO estimation model
TRO refers to the overheads other than the MRO and are proportional to MLT. For
example, energy and other utility costs rise when the lead times increase. Different workstations are identified and all the costs incurred on running these stations are gathered together. For example, energy and other utility costs, repair and maintenance costs, costs incurred on special tools and spares, financing expenses, etc. are grouped under TRO. The total TRO can easily be obtained from cost accountancy data at the end of the (n-1)th year and are divided by the total capacity hours to yield the budgeted TRO rate for the nth year. The TRO for a new product (OT) can then be estimated by multiplying the TRO rate with the MLT of the new product. i.e.
OT = MLT
n 1 TROtotal n 1 C total
(4-2)
Where,
n 1 n 1 TROtotal = total TRO in the (n-1)th year; C total = total capacity in the (n-1)th year
TRO rates calculated based on cost data (PT & DT) provided by the company for the year 2003-2004 are shown in Table 4-2. All the cost figures given in the table are in (000) Rs. Since, the workstations running costs obtained from accounts include wages they are excluded from the rates calculation. The TRO for a new transformer can be calculated using the rates in Table 4-2 and the MLT for the transformer.
98
Table 4-2: Budgeted time related overhead rate calculation (2003-2004)
Cost Centre No. 8200 8210 8220 8230 8240 8250 8260 8290 Total Less; Wages
Power Transformers 9,989 31,110
Distribution Transformers 29,966 13,333 126,631
Total Transformers 39,954 44,443 126,631 8,129 1,307 3,211 4,580 3,126 231,381 -16,300 215,081 2,346 6,658 224,086 382,633
2,032 915 2,248
6,097 392 963 4,580 3,126
46,293 -4,400 41,893
185,088 -11,900 173,188 1,760 4,994 179,941 291,017 619
Add; Financing Expenses General Administration Total (Rs. in 000) Total Normal Capacity Hours Budgeted TRO rate
587 1,665 44,145 91,616 482
The estimated MRO and TRO for the new product can be added together to give the overall overheads for the new product. i.e.
O = OM + OT
(4-3)
The total transformer cost is finally estimated by adding the three cost elements discussed: i.e. material cost, direct labour and overheads (MRO and TRO).
99
Table 4-3: Summary of TRO rates, MRO percentage fractions and overhead rates for 4 years
Material Capacity Year consumed Hours (Rs.)
Proposed Methodology TRO TRO Rate 506 500 604 619 338 369 451 482
Existing Methodology
MRO Total Overhead MRO Fraction Overheads Rate (%) 13,114 30,000 31,371 28,853 6,629 5,986 7,359 9,618 3 7 8 6 3 3 7 5 117,514 133,000 207,171 208,853 34,979 36,986 48,659 53,763 570 645 712 718 416 440 531 587
99-00 206,200 DT 00-01 206,200 02-03 291,017 03-04 291,017 99-00 84,000 PT 00-01 84,000 02-03 91,616 03-04 91,616
511,591 440,000 407,392 445,700 237,337 182,000 112,608 203,000
104,400 103,000 175,800 180,000 28,350 31,000 41,300 44,145
In a similar way, TRO rates and MRO percentage fractions can be calculated for the remaining three years of the validation period by using the data provided by the company. Table 4-3 summarizes the TRO rates and the MRO percentage fractions for the four-year period for both DT and PT obtained by the application of the proposed methodology. The table also determines the overhead rates for the same duration using the existing methodology for overheads estimation. The cost figures in the table are given in (000) Rs. except the final rates. The values can be effectively used to calculate the TRO, MRO and the overheads (companys method) for a given product by using MLT and material cost of the product as mentioned in Table 4-4 for 25kVA transformer.
100
4.4
Model implementation and validation
Data were collected for a period of 4 years to test the proposed methodology for overhead estimation. Results were systematically analysed using the cost elements breakdown discussed earlier. Costs were estimated for a selected range of products using both the existing and the proposed method for overheads estimation. The results were compared with actual manufacturing costs obtained from the cost accountancy data, and superiority of the proposed methodology was demonstrated.
As opposed to the two methods for cost estimation mentioned earlier, the actual manufacturing costs are based on cost calculation after the actual production and at the end of the financial year. Although, the precise methodology for the actual cost calculation is unknown due to confidentiality issues involved, the cost results and a brief guideline for the methodology involved was provided by the company. The actual costs of the individual transformers are the results of the redistribution of the plant-wide total costs incurred during the financial year in which they were manufactured. The actual costs are based on the actual MLT and materials (including the actual scrap material) consumed for the manufacture of a product. Total actual overheads are allocated to individual products based on a combination of factors including order related costs, tools and equipment utilization for specific products, traceable utility units to individual products, exclusive packaging, etc. Measurable activities (such as testing, inspection, quality control etc.) are accounted for by setting activity rates and determining activity units. A single cost rate is determined for all the other costs based on total production time. This cost is then allocated to individual products based on their MLT.
101
Table 4-4: Cost estimation for 25kVA transformer
Direct Labour Costs (Rs.)
Material Costs (Rs.)
Old method Overheads (Rs.) Total estimated costs (Rs.)
New Method Total estimated costs (Rs.)
Actual manufacturing costs (Rs.)
% Variation from actual cost New method Old method
Year
99-00 00-01 02-03 03-04
250 250 257 257
19400 3962 23612 3520 582 23752 24290 2.79 2.22 21825 4484 26559 3472 1528 27075 28121 5.55 3.72 24662 4949 29868 4199 1973 31092 31682 5.74 1.88 27622 4989 32868 4300 1657 33836 34877 5.78 3.01
Table 4-4 presents cost estimates for the 25kVA transformer. The costs are based on a MLT of 417 minutes and a wage rate of 36 Rs./hr for 1999-00 and 2000-01 and 37 Rs./hr for 2002-03 and 2003-04. A pictorial representation of the cost estimates produced using the two methods against the actual costs is shown in Figure 4-2 (a and b). It is apparent that the estimates obtained by applying the proposed methodology for overhead estimation are closer to the actual manufacturing costs than those obtained by the previously employed methodology. For example, the difference between the actual and estimated costs based on the previously employed methodology is on average 5% whereas the one resulting from the proposed methodology is on average 2.5%. This is attributed to the fact that the previous method ignores material quantities in the calculation of overheads and thus underestimates overheads and subsequently manufacturing cost.
MRO (Rs.)
TRO (Rs.)
102
Similar results were obtained for 100kVA, 200kVA, 500kVA and 1000kVA transformers as shown in Figure 4-2 (c to j).
25 kVA Transformers
Cost Value (Rs in 000)
36 34 32 30 28 26 24 22 20 1999-00 2000-01 2002-03
(a)
7.00
25 kVA Transformers
(b)
% Variation from actual cost
6.00 5.00 4.00 3.00 2.00 1.00 0.00 1999-00

2.79 2.22
5.55
5.74
5.78
3.72 3.01 1 .88
2003-04
Year
estimated cost (old method) actual cost estimated cost (new method)
2000-01
2002-03
2003-04
Year
old method new method
100 kVA Transformers

55
(c)
6.00
(d)
50 45 40 35 30 1999-00 2000-01 2002-03 2003-04
5.00 4.00 3.00 2.00 1.00 0.00 1999-00

1 .61 1 .03
5.60 4.44
5.48
2.58 1 .73
2.70
Year
2000-01
2002-03
2003-04
Year

60
(e)
7.00
(f)
5.96
55 50 45 40 35 1999-00 2000-01 2002-03 2003-04
6.00 5.00 4.00 3.00 2.00 1.00 0.00 1999-00 2000-01 2002-03
2.66 2.09 1 .51 2.1 1 4.52
5.1 0
2.31
Year
2003-04
Year
Figure 4-2 Contd.
103

120
(g)
7.00
(h)
110 100 90 80 70 1999-00 2000-01 2002-03 2003-04
6.00 5.00 4.00 3.00 2.00 1.00 0.00 1999-00 2000-01

2.40 1 .82 4.89
5.89 5.33
3.04 2.54 2.04
Year
2002-03
2003-04
Year

140 130 120 110 100 90 80 1999-00 2000-01 2002-03 2003-04
(i)
7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00
(j)
6.1 1 5.1 9 5.41
3.34 2.01 1 .43 2.26 2.62
Year
1999-00
2000-01
2002-03
2003-04
Year
Figure 4-2: Cost estimation results for 25, 100, 200, 500 and 1000kVA transformers
These results clearly demonstrate the superiority of the proposed methodology in the estimation of overheads and subsequently manufacturing costs. It can thus become an important part of an effective CCS. Indeed, following the analysis presented in this study, the company has adopted the proposed methodology since 2005-06.
The methodology can be further optimized and fine-tuned by considering the following points. The gradual yearly increase in the overall estimated and actual costs shown in Figure 4-2 is due to inflation. Since, the overheads for a particular year are used to determine the rates for the following year; the rates are slightly underestimated due to the effects of inflation which are not accounted for resulting in the underestimation of
104
the overall manufacturing costs. This explains why the estimated costs shown above are lower than the actual costs. In addition, capacity hours are used to determine the overhead rates instead of the actual manufacturing hours. This means, if the production system remains under capacity for a particular year, the overhead rates are under estimated for the following year and the same is true for the overall manufacturing costs. Conversely, if the production works over capacity, overestimation may be observed. Since the production for the observed period of four years remained under capacity, underestimation is observed in all the cases presented in the study. To refine the proposed method, the use of normal capacity hours during TRO rate estimation can be replaced by the actual manufacturing hours followed by an adjustment to account for inflation in the following year. If the actual material cost for a product is higher than the estimated one, the actual MRO for the product would also be higher than the estimated ones resulting again in an underestimation of the product cost. This is especially true when the actual wastage on the shop floor exceeds the estimated one. The case-based approach developed by Niazi et al. [110] could prove a useful tool for more accurate material cost estimation.
The contribution of each of the three cost elements to the total manufacturing costs company wide were also analysed in order to ascertain the manufacturing cost breakdown statistics needed to better understand an effective CCS. Figure 4-3 shows the breakdown of the overall manufacturing cost for DT and PT in a diagrammatic form and illustrates the variation in the values of each cost element over the four-year period. The peak values represent the total manufacturing costs for each year which can also reflect the variation in production output over the four-year period examined. For example, the lowest peak value observed in years 2003-04 for PT reflects the lowest level of 105
production during the four-year period. It can be seen clearly that direct labour cost contributes the least to the total manufacturing cost (owing to the geographical location of the company) whereas material cost is the most prominent cost element. The overheads values shown in Figure 4-3 include both TRO and MRO.
Figure 4-3: Cost element values in DT and PT
The values can be expressed in percentages. For example, the material cost accounts for about 65-85 percent of the total manufacturing cost whereas, labour cost is an insignificant, almost negligible cost contributor (1-2 percent of the total cost). Indirect
106
labour and indirect material costs are included in overheads. Overheads that combined both TRO and MRO varied between 13 and 33 percent.
The overheads breakdown showing MRO and TRO is plotted in Figure 4-4 for both DT and PT over the four-year period investigated. The TRO turns out to be the dominant component out of the two. One of the reasons is that the material handling costs that are part of MRO are significantly lower than the workstation running costs that come under TRO.
Figure 4-4: TRO and MRO values breakdown for DT and PT
Since the labour cost is insignificant, a comparison is only made between material costs and overheads in the form of cost trends over the four-year period. Figure 4-5 (a) illustrates the cost trends for DT. It can be seen that the total overheads kept on increasing despite a decline in material consumption until 2002-2003. However, this increase reflects an increase in TRO rather than MRO. The decline in material
107
consumption indicates the decline in productivity. The lower productivity, thus, resulted in increased overhead rates due to the allocation of company wide overheads expenditure towards the under capacity manufacture of transformers. Another factor that resulted in increased overheads was an inefficient cost control during the low productivity period.
It can be seen that MRO also increased but their effect on the total overheads was minimal. This was attributed to the fact that energy and other utility costs which are elements of TRO jumped up during that period. Figure 4-5 (b) shows a similar trend for PT. However, the TRO does not exhibit a steep increase as in DT despite a much sharper decline in material consumption compared to DT. This is because the lead times for the PT transformers are much higher than those for DT and so are the TRO. The variation in production levels, therefore, has more profound effect on TRO in PT than in DT. As a result, the decline in production in PT led to low figures for TRO.
108
Cost trend over time (DT)

600 Cost Figure (Rupees in million) 500 400 300 200 100 0 1999-00 2000-01 Year 2002-03
(a)
2003-04
time related overheads Total Overheads
material related overheads Parts and material cost
Cost trend over time (PT)

300
Cost Figures (Rupees in million)
(b)
200
100
0 1999-00 2000-01
Year
2002-03
2003-04
time related overheads Total overheads
material related overheads Parts and material cost
Figure 4-5: Cost trends in (a) DT and (b) PT
109
Figure 4-6 compares the cost trends for both DT and PT by considering material costs against overheads. Because of the reasons discussed earlier, more irregularities are found in DT than in PT.
Material Cost against Overheads

600
Cost Value (Rupees in million)
500 400 300 200 100 0 1999-00 2000-01 Year Material Cost in DT Material Cost in PT Overheads in DT Overheads in PT 2002-03 2003-04
Figure 4-6: Cost
trends comparison in DT and PT
The cost element breakdown analysis and the discussion presented above highlight some key points that are useful for establishing an effective CCS. This is particularly relevant where manufacturing cost is calculated using material cost, direct labour costs and overheads. These points are listed below:
Estimating direct material cost of a product accurately is vital as this is the most dominant cost element with a significant impact on the overall manufacturing cost 110
Material cost of a product can be estimated precisely using an effective estimation methodology such as the case-based approach [110]
Material cost is directly linked with the product design; however, minimising wastage on the shop floor is an important aspect of an effective CCS because it not only reduces direct material costs but also minimizes overheads
Direct labour cost is affected by factors like MLT, wage rate, geographical locations, etc.
Unlike material costs, overheads are not an integral part of product design and hence play an important role in controlling the overall product cost
Controlling overheads during low productivity periods is especially crucial to minimizing product costs
Identifying overheads correctly and adopting an effective estimation methodology can significantly alter product cost estimates
Determining overhead rates based on material and time consumed to manufacture a product and using the rates to estimate overheads for a new product is an easy, effective and accurate method to produce product cost estimates and is applicable in the early stages of the design and development process.
111
4.5
Conclusions
This chapter laid foundation for developing a PCE modelling methodology by establishing an overhead estimation method. In order to predict the entire product cost instead of only part or a component cost, breakdown approach was considered. The chapter considered the existing breakdown of a product cost from a published model in order to portray the potential areas of importance. This led to the identification of overheads as an area worthy of exploration. Problems with the existing methods of overhead estimation were then identified. Some of the existing methods that partially resolved the problems were also discussed. However, this led to the identification of possible solutions for estimating overheads more accurately. It was decided to validate the developed model for overhead estimation through industrial trial. A batch manufacturing set up with the estimation methodology based on a breakdown approach was chosen for the implementation and validation purposes.
The manufacturing costs in an electrical engineering company in South Asia were systematically analysed with emphasis on the estimation of overheads which contribute about a third of total manufacturing costs. The analysis identified problems with the previous practise of estimating overheads and a new methodology was proposed. The new methodology was retrospectively implemented for a four-year period in the company and a cost analysis was carried out. The estimates obtained with the previously employed and the proposed methodologies were compared against the actual costs. It was found that the proposed methodology leads to more accurate manufacturing cost estimation results.
112
The study identified irregularities in the cost trends of the selected products over the period of observations, especially during low productivity periods. Recommendations were made to improve the proposed methodology further such as the use of actual manufacturing hours instead of normal capacity hours during TRO rate estimation followed by an adjustment for inflation in the following year. Job shops and batch manufacturing set up based on flexible manufacturing and MRP based systems are some of the examples of potential candidates for the proposed methodology.
The study also revealed that material cost accounts for 75 percent, labour costs for only 2 percent and overheads for almost 23 percent of the overall manufacturing costs at the company. This cost breakdown, however, is representative of that region characterized by lower wage rates. Thus, extensive testing of the proposed methodology in other geographical locations is ongoing in order to further fine-tune it and make it an integral part of an efficient CCS.
113
Chapter 5
PCE Hybrid Model
The aim of the Chapter 5 is to develop mathematical models for accurate product cost estimation with a focus on batch manufacturing systems. The modelling methodology presented is a hybrid approach consisting of a breakdown technique and an activity-based costing system and focuses on an effective utilization of cost data obtained at the end of a given period to predict a products cost for the following phase. The modelling framework proposed is based on time- and material-dependent cost elements.
5.1
Introduction
Chapter 5 is aimed at modelling the overall product cost from the foundation already provided in the previous chapter. It was established that the development of a hybrid system for PCE in a batch type manufacturing environment is a viable research option in the area. Breakdown approach was then selected as one of the hybrid element. The other elements from the classification system for the hybrid approach will now be identified in this chapter. However, since, the breakdown approach was selected and a complete overhead analysis was already carried out in the previous chapter, the focus is first given here to establish the breakdown elements for the proposed model. This would require an indepth analysis of what constitutes a products cost.
114
Chapter 5: PCE Hybrid Model
Manufacturing is a transformation process that consumes material and resources. The degree of consumption of these resources is reflected in a products cost. The origin of the consumed resources and those of the factors influencing a products cost may not only be traced to the shop floor but outside it. The emerging picture of a products cost is, therefore, a complex blend of distinct and indistinct contributions made to the product. The contributions or the cost elements are often easier to be allocated to individual products once they are produced. However, predicting these cost contributions even beforehand requires efficient and accurate predictive tools.
The ever-increasing pressure on estimators to predict the cost early (sometimes even before the conceptual phase) and accurately is also accentuating the need to develop innovative techniques combining knowledge, experience, resources and historical data. Often the techniques are customized to suit the needs of a system and/or a product. Whether, the methodology is customized or generic, the final product cost determines selling price. An effective cost estimation methodology, therefore, bears a potential of setting competitive price without compromising profits yet securing commercial advantage to the enterprise.
In a batch manufacturing environment, where the available facilities are effectively utilized to carry out a wide range of operations, estimating manufacturing costs accurately requires an in-depth analysis of the shop floor resources and the way they are consumed. For this reason, determining manufacturing cost in addition to the overall product cost does not often receive enough attention. The estimation techniques available, although predict the overall product cost, fail to yield accurate estimates of the manufacturing cost component. 115
An opportunity to make any optimization into the same or other components may, thus, be lost. The chapter presents a comprehensive model for the estimation of manufacturing cost and its sub-elements in addition to the overall product cost. However, before developing the mathematical models for PCE, an analysis of the product cost and modelling methodology is drawn.
5.2
Product cost and modelling approach
The previous chapter considered manufacturing cost as a combination of direct material, direct labour and overheads (also termed as indirect manufacturing costs). Manufacturing cost, on the other hand, is a sub-element of the overall product cost which is a combination of engineering costs (costs incurred during design and development phases starting with customer requirements, design of the part, its process planning and prototyping), manufacturing costs and production overheads. Manufacturing cost is the largest element of the overall product cost and is largely determined during the product design phase. This element accounts for all the costs associated with the resources that can be traced to the manufacturing floor where the actual production takes place.
Overall product cost can also be divided into direct and indirect costs incurred on a product during its transformation from raw material to finished entity. Direct costs refer to the cost elements that can be directly traced to a product and are generally incurred for the benefit of the individual product. For example, direct material costs incurred on a product refer to
116
the costs incurred on material that is integral to the product design. Similarly, direct labour costs refer to the costs incurred on labour spent purely on manufacturing the product and is easily determined using lead times obtainable from process planning details. Direct material and labour are the elements of manufacturing cost. Indirect costs refer to all the production support costs incurred on or off the shop floor and generally for the shared benefit of several products within a manufacturing facility. These refer to all the costs other than those included in direct costs. These include manufacturing overheads, production overheads and engineering costs.
From the analysis presented in the previous chapter, it was found that manufacturing overheads (part of indirect costs) can be divided into time and materialdependent cost elements. However, after the implementation and validation analysis, areas for further optimization were discovered. It was found out that a further two classes of manufacturing overheads can be identified. Some of the indirect costs depend on both processing time and material quantities to be produced. Cost spent on tooling, for example, depends on both processing time and the amount of material to be processed and can not be grouped with either time or materialdependent cost groups. Some of the other indirect costs, on the other hand, are spent on keeping the manufacturing shop floor in a running condition. A realistic approach to allocate such costs on different products is required that is based on the products level of consumption of resources from the manufacturing shop floor. If a product stays for a longer period of time and occupies a larger space on the manufacturing floor, it is highly likely to consume more resources than a product that stays for a shorter duration and occupies little space. Such indirect costs can be termed as building space cost
117
and are, thus, based on lead times and occupied spaces for individual products. Yet another set of indirect costs are those that can not be directly traced to a manufacturing shop floor or only partially. For example, costs incurred on building security, general administration costs and financing expenses etc. can be grouped together. These costs need to be allocated to individual products using a pragmatic approach. It is highly likely that a product with higher manufacturing costs should incur these costs in greater proportion than those with lower manufacturing costs. It is because a product with a higher manufacturing cost consumes more resources on the manufacturing shop floor and is thus likely to incur higher proportions of other costs also. Such indirect costs can, therefore, be allocated to individual products in proportion to their manufacturing costs. These indirect costs refer to the only set of overheads not directly traceable to the manufacturing shop floor and can be termed as production overheads. The other four indirect costs can be traced to the manufacturing shop floor and can be termed indirect manufacturing costs. These include: timedependent cost (also referred to as processing cost), materialdependent cost, tooling cost and building space cost. Engineering cost is also a form of indirect cost and refers to the costs incurred during design and development phases starting with customer requirements, designing of the part, its process planning and prototyping.
Based on the breakdown approach, product cost can now be expressed as a summation of the individual cost elements (manufacturing cost, engineering cost and production overheads). A pictorial representation of the detailed product cost breakdown followed in the present study is shown in Figure 5-1 together with the cost distribution for a typical
n +1 , can thus be expressed as the product [109]. The estimated cost for a new product, C p
118
n +1 sum of the three cost elements: estimated manufacturing cost , CGp , estimated engineering n +1 n +1 , and estimated overheads, O p : cost, C Ep
n +1 n +1 n +1 n +1 Cp = C Gp + C Ep + Op
(5-1)
Selling Price Selling Price

20% 25%
Product Cost
Profit
40% 15%
Manufacturing Cost
Engineering Cost
Production Overheads
Manufacturing Cost
38%
Direct Material Cost
Direct Labour
Indirect Manufacturing Cost

12%
50%
Processing Cost
Material Dependant Cost
Tooling Cost
Building Cost
Figure 5-1: Pictorial representation of the mathematical model for product cost estimation
Manufacturing costs refer to the costs associated with the activities taking place on the manufacturing shop floor and the cost elements that can be traced to it. The manufacturing
119
n +1 cost, CGp , for an individual product, p, can be estimated by summing the direct cost n +1 n +1 , and labour cost, C Lp ) and the indirect manufacturing cost elements (material cost, Cmp n +1 n +1 n +1 elements (processing cost, Ctdp , material-dependent cost, Cmdp , tooling cost, CTp , and n +1 building cost, C Bp ) for the product, i.e.:
n +1 n +1 n +1 n +1 n +1 n +1 n +1 C Gp = C mp + C Lp + C tdp + C mdp + CTp + C Bp
(5-2)
Determining indirect costs accurately and early for an individual product in a manufacturing environment where resources are shared between several different products is challenging and requires sound engineering knowledge of product design, planning, production and costing. A traditional approach to estimating such costs relies on summing up all the expenses other than those included in the direct costs and ascertaining a cost rate. The rate is then used to estimate the indirect costs for a new product by multiplying it with the expected manufacturing lead time (MLT) of the new product. However, a major drawback of this methodology is a disproportionate allocation of indirect costs to different products. Furthermore, since the indirect costs recovery considers only MLT, the amount of material required for the manufacture of a product is ignored. This often results in cost underestimation for products with higher material requirements and overestimation for those with lower material quantities.
The ABC system was presented [93] as a useful means to distribute the indirect costs in proportion to the activities performed on a product during manufacturing. The system 120
proved a good alternative to the traditional estimation techniques since it provided more accurate manufacturing cost estimates [95]. However, the effectiveness of the methodology depends on the way the activity rates are determined and the accuracy of the estimated activity units. Moreover, time-based activity units are generally employed ignoring material related costs during PCE using the ABC system. For example, the ABC model developed by Aderoba [48] only considered activity times ignoring material quantities for the cost estimation of a new product. Material dependent costs such as transportation, packaging, storage and inventory, etc. may, therefore, not be accurately estimated for the new product.
The model presented in the current chapter overcomes the limitations of the existing methodologies, as mentioned above, by providing a framework for PCE based on time and materialdependent cost elements. The modelling methodology is a hybrid approach combining a breakdown model (including manufacturing costs, engineering costs and production overheads) with a modified ABC system and skilfully exploits the already validated models in Chapter 4. The above mentioned limitations associated with determining activity rates and activity units using the conventional ABC system are overcome based on an effective utilization of the cost data obtained at the end of a given year n to predict a products cost in the beginning of the following year (n+1) by determining the rates for cost elements. It means that the developed method incorporates the attributes of a case-based system into the ABC system by making use of past data. The modified ABC system, therefore, forms second element of the Hybrid Model. Figure 5-2 is a pictorial representation of the overall modelling framework within the context of the developed technique classification system. It is evident that the developed model is also
121
hybrid at the top level of the classification system by making use of the attributes of the qualitative and the quantitative techniques. The breakdown approach helps to estimate the overall product cost. The ABC takes care of the accuracy of the results and the use of the case-based approach helps to produce early results. Although, at the lower levels of the classification system, three elements appear to contribute towards the proposed model, the modified ABC system and the breakdown elements are the two elements of the Hybrid Model. The modified ABC component is itself hybrid in nature. The developed model is a comprehensive and an integrated costing tool that not only estimates the overall product cost but predicts the essential individual cost elements. A low to medium volume batch type production environment where various indirect costs are involved is a highly suitable environment for the application of the proposed methodology.
5.3
Direct cost elements
Direct costs refer to the costs associated with the elements directly identifiable to and linked with a product and are generally incurred for its exclusive benefit. Costs associated with material and labour are largely considered direct manufacturing costs. These are examined in more detail below.
122
5.3.1
Direct material costs
The material costs can be estimated using the material quantities obtained directly from the BOM developed on the basis of the product structure and engineering design details. A scrap margin is usually accounted for in BOMs based on past trends. In a manufacturing environment where products are standardized, their engineering designs and BOMs are also established. Even when a non-standard product is designed and developed, the required quantities of different raw materials are similar to those in a closest standard match even though the technical designs of the two may be different. Thus, the closest match provides a good starting point to estimate material cost for the non-standard product. As a result, material cost estimation is possible in the early stages of the design and development stage.
n Under such circumstances, the material cost calculation, based on past product cost, Cmp ,
(already stored in a database) and certain commercial indicators termed as material cost deviation index (MCDI), n +1 , (such as inflation rate, foreign exchange rate, price increase etc.) for the following year, becomes a standard practice. i.e.,
n +1 n C mp = (1 + n +1 ) C mp
(5-3)
Past product costs can be established using the unit cost, Ckn , of the kth material in the product and its corresponding quantity, mk , i.e.
123
n C mp = (C kn )(mk ) k
(5-4)
n The total cost based on the cumulative material requirement, Cmt , for manufacturing, p,
products for a given period of time changes in accordance with the number of units
n produced, N p , for a given product. Thus, the factory-wide changes in material
consumption over different time intervals can be used to reflect changes in production levels and can be obtained by the following equation.
n n n C mt = (C mp )( N p ) p
(5-5)
n +1 , based on cumulative material consumption can be obtained The total planned cost, Cmt n +1 from the planned orders, N p .i.e.
n +1 n +1 n +1 C mt = (C mp )( N p ) p
(5-6)
124
Regression Analysis Model
Back-Propagation Neural Network Model
Decision Support Techniques
Operation-Based Approach
Breakdown Approach Feature-Based Cost Estimation
Case-Based Technique Expert System Rule-Based System Fuzzy Logic System ToleranceBased Cost Models
Activity-Based Cost Estimation
Hybrid Model for PCE

Figure 5-2: Development of the Hybrid Model within the framework of the technique classification system
125
5.3.2
Direct labour
The manufacturing shop floor layout consists of work centres which can be either labourintensive or machine-intensive centres or both. Fabrication, manual assembly and painting are examples of labour-intensive centres sometimes referred to as labour centres also. CNC machines, automatic assembly lines, industrial robots are examples of machine-intensive centres. A combination of the two is also common and can be termed as hybrid centres; examples include lathes, mills, semiautomatic assembly lines, etc. More commonly the term machine centre is also used to refer to either machine-intensive or hybrid centres i.e. any work centre that is not a labour centre can also be termed a machine centre. Figure 5-3 illustrates the three different work centres.
Work Centre
Labour Centre/ Labour-intensive centre Fabrication, manual assembly, painting etc.
Machine Centre
Machine-intensive centre CNC machines, automatic assembly lines, industrial robots etc.
Hybrid centre Lathes, mills, semiautomatic assembly lines etc.
Figure 5-3: Types of work centres
126
The direct labour cost incurred on a product accounts for the fraction of the wages of the workers in the work stations through which the product is routed. The time spent by the workers to manufacture individual products and their respective wage rates (generally set on the basis of their skill levels) can be used to determine this cost. However, in a batch manufacturing environment where multi-skilled workers are placed on work centres intermittently and handle multiple tasks, a more pragmatic approach is required to ascertain these costs.
The direct labour in the proposed methodology is based on the shop floor wide aggregate labour rate and the labour units spent to manufacture a product. Thus, the direct labour cost
n already incurred on an individual product, C Lp , can be obtained as a product of the actual
n n , and the labour units consumed, U Lp , to manufacture it, i.e. labour rate, RLA
n n n C Lp = ( RLA ) (U Lp )
(5-7)
n +1 The estimated labour cost, C Lp , can be given as a product of the estimated labour rate, n +1 n +1 RLE for a given period and the estimated labour units, U Lp , for a given product.
n +1 n +1 n +1 CLp = ( RLE ) (U Lp )
(5-8)
127
n +1 n +1 Total planned labour units, U Lt , can be used to estimate the total labour cost, C Lt , for
planned orders, i.e.
n +1 n +1 n +1 C Lt = ( R LE ) (U Lt )
(5-9)
Labour units
The MLT of a product is the summation of the individual lead times in the relevant work centres. The lead time for a job on an individual work centre (sometimes referred to as work centre cycle time WCCT) is the time spent by the job on the work centre before moving on to the next work centre. The lead time is different from the man-units consumed which are referred to as the total time spent by all the workers on the job in that work centre. The man-units on an individual work centre can be converted into labour units by considering the output levels (skilled, semi-skilled and non-skilled levels) of the individual workers. The total time spent by skilled labour, tx, semi-skilled labour, ty and non-skilled labour, tz, can be expressed in a unified form of labour units, Lnjp , by applying skill indices,
, for semi-skilled labour with ranging from 0.4 to 0.8 and, , for non-skilled labour with
ranging from 0.25 to 0.4.
Lnjp = t x + t y + t z
x y z
(5-10)
128
The values for skill indices can be set based on feedback form shop floor supervisors. For example, the value of = 0.6 can be set for a semi-skilled labour whose skill or output level is almost 60 percent of a standard skilled-labour. Wages of a worker on the shop floor may also be a criterion to set the indices as variations in wages of the workers can be in accordance with their skill or output levels.
This approach is particularly helpful in a computer integrated manufacturing (CIM) environment where time cards are filled at individual work centres with job codes, employee numbers, etc. Based on the relative skill levels of semi- and non-skilled labour compared to those of the skilled ones, line supervisors are generally in a good position to ascertain skill indices for individual workers. If an individual lead time, t n jp , is taken as the time spent by all the workers working on a job in a work station, then the number of skilled, x, semi-skilled, y, and non-skilled, z, labour can be used to obtain labour units on the work centre, using the formula:
Lnjp = t n jp ( x + y + z )
(5-11)
The total labour units consumed for the product can then be given as:
n U Lp = Lnjp j
(5-12)
129
n This in turn allows the estimation of the shop floor wide labour units consumed, U Lt , for a
given period of time based on the number of different products produced during that period:
n n n U Lt = (U Lp )( N p ) p
(5-13)
The determination of labour units for a product, therefore, requires lead times at individual work centres and number of different workers with their skill levels working on the product. The lead times can be obtained from standard routings and are established either by using standard formulas or by deploying time and motion studies. For example, a range of time estimation models and time standards can be found in [111]. In a CIM environment, a feedback mechanism ensures that the actual lead times are standardised for future use. The lead time for a non-standard or a new product can be established from the results for the closest standard match and incorporating the changes in consultation with the shop floor supervisors and planning engineers. The labour units required to manufacture the product can then be estimated based on the number of operators required on individual work centres.
Labour rate
n The shop floor-wide direct labour cost incurred during a given period, C Lt , reflects the
wages paid to the shop floor workers during that period and is the summation of all the total
n wages in the individual work centres, Gtj , i.e.:
130

n n C Lt = Gtj j
(5-14)
The total wages and the total labour units consumed in a given period can be combined to calculate the actual labour rate, i.e.:
G
n RLA = j n U Lt
n tj
(5-15)
The labour rate calculated in this way differs from the conventional wage rate for a given worker. The wage rate for a particular period of time is a matter of organizational policy and is influenced by various factors (such as minimum wage rate set by the government, geo-political conditions, economic growth, and an individual workers skill level etc.). The actual labour rate during a period n can be used to ascertain an estimated labour rate for period n+1 based on the expected variance called labour cost deviation index (LCDI), n +1 , (such as the effects of inflation, forecasted wage differential, etc.):
n +1 n RLE = (1 + n +1 ) RLA
(5-16)
131
5.4
Indirect cost elements
These are costs associated with contributions that can be traced to the manufacturing shop floor but not to a specific product or only partially. The costs incurred on indirect material (material shared for the benefit of many different products), utilities, machine repair and maintenance, quality control, tooling and equipment, building space, etc. comprise indirect costs. The indirect cost elements for an individual product can be determined by adopting a suitable methodology for apportioning the total indirect costs based on the products degree of consumption of manufacturing activities and resources. The study identifies four kinds of indirect costs that can be linked with manufacturing shop floor and presents comprehensive models to estimate them for a specific product.
5.4.1
Processing cost
n The processing costs incurred during a given period on the manufacturing shop floor, Ctd ,
refer to all the processing-time-dependent costs (excluding labour costs) that contribute to the smooth running of the machine centres on the shop floor. Thus, this cost can be
n expressed as a summation of all the individual time-dependent cost elements, Cd , (such as
utility cost, maintenance cost, repair cost, machine depreciation, machine insurance etc.) for a given period:
n n Ctd = Cd d
(5-17)
132
n +1 The processing cost for an individual product, Ctdp , can be determined from the estimated n +1 n +1 , and processing units for the product, U Mp : processing rate, RME
n +1 n +1 n +1 C tdp = ( RME ) (U Mp )
(5-18)
Processing units
n The job processing time, tip , on the, ith, machine centre (time spent by a job between its
arrival and departure from a machine centre while the machine is running) can be used to
n , on the centre by applying a machine determine processing units consumed by the job, M ip
index, i .
n n M ip = i t ip
(5-19)
The machine index accounts for repair, maintenance, utilities etc. and takes values ranging from 1.25 to 2.0 [9]. Older machines requiring more maintenance, repair, etc. will have a higher index value. The job processing time is generally the same as lead time and can be obtained directly from time cards in a CIM environment. By summing up all the processing
n units at individual machine centres, the total processing units for the product, U Mp , can be
obtained:
133
n n U Mp = M ip i
(5-20)
n The shop floor-wide processing units consumed, U Mt , for a given period of time can be
calculated based on the number of different products produced during that period:
n n n U Mt = (U Mp )( N p ) p
(5-21)
The job processing times for a new product can be estimated using closest standard matches allowing the estimation of the processing units for the new product.
Processing rate
n A shop floor-wide aggregate processing rate, RMA , is proposed from the processing cost
data collected during a given period and the corresponding processing units consumed:
n MA
n C td = n U Mt
(5-22)
However, the same methodology can be used to establish separate rates for individual machine centres if their corresponding processing costs can be traced and cumulative processing units obtained easily. The estimated rate for the following period can be set
134
based on a processing cost deviation index (PCDI), n +1 , that reflects the effects of inflation, utility costs variations, maintenance and insurance forecasts, etc.:
n +1 n RME = (1 + n+1 ) RMA
(5-23)
5.4.2
Material dependent cost
Several indirect costs incurred in a manufacturing enterprise can be attributed to material quantities used to manufacture different products, called material-dependent costs (MDC). Examples of such costs include: indirect material, purchasing, stores & inventory, freight & transportation, material inspection, packaging, and quality costs, etc. These costs do not depend on the lead times or processing times but on the quantity of materials consumed. Thus when the direct material quantities go up, MDC also go up. As a result, MDC is also a significant contributor to the manufacturing costs. The total material-dependent costs for a
n given period, Cmd , can be expressed as:
n n C md = Cw Sn w
(5-24)
n Where, C w refers to the total cost incurred on an individual MDC element (such as
packaging) and S n is the salvage value for scrap material. Since the MDC increase with the rise in direct material quantities, the fraction based on the total MDC and the cumulative
135
direct material cost for a given period can be used to estimate the MDC for a specific
n +1 n +1 product, Cmdp , using its estimated direct material cost, Cmp and MDC deviation index
(MDCDI), n+1 (the effect of inflation, variations in material, freight and transportation prices etc.):
n C md n +1 C mdp = Cn mt
n +1 n +1 (1 + ) C mp
(5-25)
5.4.3
Tooling cost
Tools range from manual equipments to jigs, fixtures, moulds, dies etc. The total tooling
n , comprises the cost cost incurred on the manufacturing floor for a given period, CTt
n incurred by tools utilization and replenishment in machine centres, CMT , and labour n centres, CLT .
n n n CTt = CMT + CLT
(5-26)
Tooling cost depends not only on the amount of material processed on the manufacturing floor but also on the job processing times on machine and/or labour centres. The proposed methodology, therefore, allocates the total tooling cost incurred on the shop floor to individual products on the basis of their material costs and corresponding processing and/or labour units. Tooling cost rates in labour and machine centres for a given period n are
136
determined for this purpose. Estimated tooling cost rates for the following period n+1
n +1 n +1 based on machine centres, RMT , and labour centres, RLT , can then be effectively utilized to n +1 predict the tooling cost for an individual product, CTp , i.e.:
n +1 n +1 n +1 n +1 n +1 n +1 CTp = C mp RMT U Mp + RLT U Lp
) [(
)(
) (
)(
)]
(5-27)
Determining the tooling cost for an individual product based on an aggregate shop floorwide tooling cost rate often results in disproportionate allocation. This is due to the cost variances between the tools at machine centres and labour centres and the different time spent by the product on the respective centres. The proposed model overcomes this problem by providing separate rates for tools used at machine and labour centres respectively.
Machine tool rate

n The machine tool rate, RMT , is associated with the total tooling cost incurred in machine n centres, C MT , during a given period. It is given by the formula:
n RMT =
n (CMT ) n n (U Mt ) (Cmt )
(5-28)
The total machine tool cost for a given period can be expressed as the sum of total
n depreciation, DM , of M number of machine tools during that period. The costs associated
137
with number of b machine tools that are broken down can be incorporated by considering their initial purchase prices, Pb , and respective depreciated values, Db .
n n CMT = DM + ( Pb Db ) M b
(5-29)
Standard techniques can be applied to ascertain the depreciation values for individual tools. The estimated machine tool rate for the following period can then be set based on a machine tool cost deviation index (MTCDI), n +1 , to account for the effects of inflation, expected variation in tool utilization, life expectancy factor of the current tools, etc.:
n +1 n RMT = (1 + n +1 ) RMT
(5-30)
Labour tool rate

n n The labour tool rate, RLT , considers the total tooling costs incurred in labour centres, C LT ,
for a given period. It is determined using the formula:
n R LT =
n (C LT ) n n (U Lt ) (C mt )
(5-31)
The total labour tool cost for a given period can be expressed as the sum of total
n depreciation, D L , of L number of labour tools during that period. Similarly to the machine
138
tools, the costs linked with number of a broken labour tools with their initial purchase prices, Pa , and respective depreciated values, Da , can be incorporated as follows:
n n CLT = DL + ( Pa Da ) L a
(5-32)
A labour tool cost deviation index (LTCDI), n +1 , is then employed to account for inflation, expected variation in tool utilization, life expectancy factor of the current tools, etc. and thus the labour tool rate for the following period will be:
n +1 n RLT = (1 + n +1 ) RLT
(5-33)
5.4.4
Building space cost
The building space cost includes all the essential costs incurred to keep the manufacturing shop floor and the overall plant in a usable condition. Plant depreciation, building insurance, maintenance, repair, and utilities (excluding those supplied to the shop floor) are some of the examples of building space cost elements. The number of h building space cost
n elements, Ch for a given period can be added together to determine the total building space
n cost, C Bt , for that period.
n n C Bt = Ch h
(5-34)
139
The proposed methodology allocates the total building space cost to individual products based on the areas they occupy on the manufacturing shop floor and their lead times. Products requiring bigger spaces and higher lead times incur larger proportion of building space costs. Building space rate for a given period is determined for this purpose. An
n +1 estimated building space rate, RB , can then be set for the following period. This allows
the estimation of the building space cost for an individual product by multiplying the
n +1 estimated rate with the manufacturing space, SGp , and units (processing and labour) for the
product, i.e.:
n +1 n +1 n +1 n +1 n +1 C Bp = RB S Gp U Mp + U Lp
) (
) (
(5-35)
Building space rate

n Building space rate for a given period, RB , can be set by dividing the total building space n , and cost incurred during that period from the total manufacturing units (processing, U Mt
n n labour, U Lt ) and the total area of the manufacturing shop floor, S Gt .
n RB =
n (U Mt
n C Bt n n + U Lt ) S Gt
(5-36)
140
Again a building cost deviation index (BCDI), n +1 , can be defined to account for any variations due to inflation, forecast variations for building maintenance, repair, utilities etc. and the final adjusted rate will be:
n +1 n RB = (1 + n +1 ) RB
(5-37)
Manufacturing space
Manufacturing space here refers to the overall area of the manufacturing shop floor and is a
n combination of the space occupied by all work centres, S ot , and the total unoccupied space
n on the floor, Sut .
n n n SGt = S ot + Sut
(5-38)
Shop floor cell layout records can be used to find out occupied spaces by individual work centres, O n j , that can be summed together to give the total occupied space.
n S ot = On j j
(5-39)
The ratio of the total manufacturing space to the total space occupied can be effectively
n +1 used to allocate the manufacturing space for an individual product, SGp .
141
n n +1 n +1 S Gt S Gp = S op Sn ot
(5-40)
n +1 , refers to the total occupied space for the product and can easily be obtained Where, S op
by summing up the number of r different work centre spaces through which the product is
n +1 routed, Orp .
n +1 n +1 Sop = Orp r
(5-41)
5.5
Production overheads
Overheads here refer to the production overheads (PO) and are different from indirect manufacturing costs. These include costs incurred on elements like security services, computer software, general administration, financing, sales and marketing, etc. They are not normally traced to manufacturing shop floor. The total PO, Otn , can be found by adding
n the number of q different PO cost elements, H q , i.e.:
n Otn = H q q
(5-42)
142
Since, the POs are incurred to maintain manufacturing activities; they can be effectively allocated to individual products as fractions of their manufacturing costs. The fraction
n calculated based on the total PO and the total manufacturing cost, CGt , incurred during a
given period can, therefore, be used to estimate the POs for an individual product for the
n +1 n +1 following period, O p , using its estimated manufacturing cost, C Gp , and PO deviation
index (PODI), n+1 (the effect of inflation, variations in selling expenses, general and administration costs etc.):
Otn n +1 n +1 n +1 Op = Cn (1 + ) C Gp Gt
(5-43)
The total manufacturing cost incurred for a given period is a summation of all the individual total manufacturing cost elements, i.e. material, labour, time-dependent, material-dependent, and building space costs.
n n n n n n n CGt = Cmt + CLt + Ctd + Cmd + CTt + C Bt
(5-44)
5.6
Conclusions
This chapter presented a comprehensive modelling methodology for PCE in a batch type manufacturing environment. The overhead estimation method developed in the previous chapter formed the basis for extending the modelling framework. The limitations associated 143
with the existing methods of overheads estimation resulted in defining sets of new indirect costs which were modelled later on. The only element that could not be modelled was engineering cost due to the insufficient experience, data and knowledge obtained from the industrial domain. However, the nature and the extent of the theories developed for the other elements and the clarity of the developed models contain the intrinsic characteristics that can serve as guidelines to model the engineering cost.
The adopted modelling methodology is a hybrid approach combining the attributes of a breakdown technique with the ABC system in order to estimate the total product cost. In this respect, the model was presented as a summation of individual cost elements (manufacturing cost, engineering cost and production overheads) which were broken down further to their sub-elements. Activity rates and units were defined and modelled for individual elements based on an effective utilization of historical cost data in order to predict product costs early and accurately. Thus comprehensive models for estimating manufacturing cost and indirect cost elements for individual products were given.
The proposed modelling framework is based on time- and material-dependent cost elements and overcomes the current limitations in cost estimation. The model has been tested and validated using data from a crane assembling unit and the results are presented and discussed in the next chapter.
144
Chapter 6
Industrial
Implementation and Analysis of the PCE Hybrid Model

The aim of Chapter 6 is to implement the developed PCE Hybrid Model in a batch manufacturing environment in the UK. The process will focus on developing an implementation algorithm, modelling cost deviation indices and obtaining the estimated cost values from the models implementation for a given product range. The estimated results will also be obtained from the companys own method.
6.1
Introduction
Chapter 6 establishes basis for a thorough validation analysis for the developed PCE Hybrid Model. This is achieved by the industrial implementation and application of the Hybrid Model developed and presented in the previous chapter. When considering the validation process, it is important to determine the criteria for validation and that if the adopted criteria are widely acceptable and recognizable standards within the framework of a specialism. Although, mathematical models can be simulated to validate the results, those designed for industrial applications need to be backed up by industrial trials to best serve
145
Chapter 6: Industrial Implementation and Analysis of the PCE Hybrid Model
the purpose. Industrial trials can be time consuming requiring attention to details, the results based on such attempts can only strengthen the confidence in the entire validation process itself.
The developed model as part of the current research study was not only conceptualized with the aim of theoretical development in the field but to serve the wider industrial needs also. The criterion for its validation, therefore, is fittingly set to proceed with the industrial implementation and application. The process involves implementing the model and validating it by comparing the results with those generated by any existing system in place.
Setting the objectives for the validation process and defining how they will be achieved are fundamental to the success of the entire process. Before any of the objectives can be set, the active conditions have to be considered. Within the context, getting the right balance between the duration of the trials and the consistency of the results generated during that time is important. The trial period, therefore, is set for three years (2003 2005) retrospective analysis requiring field data from 2001 to 2005 based on the requirement of the developed model. The validation period of three years is a reasonable time to allow the process to generate reliable and consistent results. Since the model is developed to mainly fulfil a batch manufacturing systems requirements, the selection of such an industrial environment for the implementation and validation analysis is yet another active condition. A careful consideration was given to the conditions to initiate the validation process.
146
The main objective of a PCE process is to predict the likely product cost early and accurately. Accurate estimation refers to minimizing the difference between the actual costs and the predicted values. The developed model should aim to predict product costs more accurately than any of the existing systems in use at the selected company. Although cost estimation has no direct influence on cost control, the estimated results should help the decision-making process to devise cost control strategies. Cost estimation methods that can predict not just the entire product cost but the elemental values can highlight the potential areas for better cost control. The developed model should, therefore aim to present better structured and more elaborate elemental results than those generated by the existing system.
Section 6.2 details the procedure and methodology for the industrial implementation of the PCE Hybrid Model. A comprehensive algorithm is developed to facilitate the implementation process. Section 6.3 details the PCE process at the company. The necessary business information is presented followed by the cost estimation for the given product range using the companys own method of cost calculation. Section 6.4 deals with the implementation of the PCE Hybrid Model in the selected company with the help of the modelled indices and based on the developed algorithm in section 6.2 in order to ascertain the estimated costs for the given product range for the duration of the trial period. Section 6.5 concludes the chapter.
147
6.2
HMI algorithm and the implementation methodology
The Hybrid Model Implementation (HMI) algorithm is developed and presented in the section to facilitate the industrial implementation of the model. The overall aim of the implementation of the Hybrid Model in return is to facilitate the entire validation process. With this in mind, it is essential to compare the results generated by the implementation of the Hybrid Model against those generated by the companys own method. The companys method of cost estimation with the relevant results would, therefore be discussed to support the implementation of the Hybrid Model. But first the necessary steps and the overall procedure of the Hybrid Models implementation will be discussed.
The Hybrid Model is implemented in the selected manufacturing company for a retrospective analysis of three years (2003 2005). The developed model makes use of the available data at the end of a year (n) and the preceding year (n-1) to predict the costs for the following year (n+1). Data obtained from 2001 to 2004 could therefore be used to predict costs from 2003 to 2005. Since, the comparison of the estimated costs predicted at the beginning of a year is made with the cost data obtained at the end of that year; cost data for 2005 were also required in order to make comparison with the estimated costs for that year. Therefore, the data obtained from 2001 to 2005 were used for the implementation and validation analysis for (2003 2005). The estimated results will then be compared against the actual product costs in the next chapter as part of the validation analysis. The current chapter details the entire implementation process. Figure 6-1 to Figure 6-7 represents the Hybrid Model Implementation (HMI) Algorithm.
148
The stepwise preliminary implementation of the Hybrid Model is presented in Figure 6-1. It is clear that the entire PCE process for a product is outlined in 13 different steps starting from material cost estimation to the estimation of a products overall cost. The figure cross references relevant sections, figures and equations. The logical sequence of necessary steps is shown with arrows. The dotted arrows follow sub-systems detailed in separate cross referenced figures with step numbers.
149
Start Step 1 Step 2 (refer to Figure 6-2) Material Cost Estimation (Section 5.3.1) Step 3 1A Step 4 (refer to Figure 6-3) Labour Cost Estimation (Section 5.3.2) 1
Step 5
2A Step 6 (refer to Figure 6-4)
Processing Cost Estimation (Section 5.4.1) 3A Step 8 (refer to Figure 6-5) MDC Estimation (Section 5.4.2) 4A
Step 7
Step 9
Manufacturing Cost Estimation Equations (5-2), (5-44)
Step 10 (refer to Figure 6-6) 5
Step 11
5A Step 12 (refer to Figure 6-7)
Production Overhead Estimation (Section 5.5)
Step 13
6A
Product Cost Estimation (Section 5.2) Figure 5-1, Equation 5-1
Stop
Figure 6-1: PCE Hybrid Model implementation phase
150
Detailed implementation steps are facilitated with necessary links in the preliminary implementation phase. For example material cost estimation process is presented with links 1 and 1A and is elaborated in Figure 6-2. The logical steps presented in rectangular boxes are connected with arrows. Inputs to the system are presented in parallelograms together with necessary references to the equations developed as part of the Hybrid Model. For example, equation (5-4) can be used to calculate material cost for the pth product in the nth and the (n-1)th years using the input provided in the same parallelogram. The input in this case refers to material cost data provided by the company from its financial accounts.
1
Material cost data for the nth and (n-1)th years
Material cost calculation for pth product in the nth and (n-1)th years
Eq (5-4)
Cumulative material costs in the nth and (n-1)th years
Number of units produced in the nth and (n-1)th years Eq (5-5)
MCDI calculation for the (n+1)th year
Eq (B-5)
Estimated material cost for the pth product in the (n+1)th year
Eq (5-3)
1A
Refer to Figure 6-1 (step 3)
Figure 6-2: Material cost estimation implementation process
151
The inputs to the system are either primary or secondary. Primary inputs are the details provided by the company. Secondary inputs to the system are the results obtained during the implementation phase and are fed to the system to generate further results. For
example, during the calculation for PCDI mentioned in Figure 6-4, the inputs to equation (B-9) are aggregate processing rates (secondary input) and inflation rates (primary input). Aggregate processing rates are termed secondary inputs because they were calculated in the preceding step using equation (5-22). It is, therefore, fitting to sequence the execution steps in the HMI algorithm to maintain the logical flow and to effectively utilize the secondary data. The algorithm is designed to facilitate the Hybrid Model implementation at the selected company and as such adapts to the system. However, minor changes can make the algorithm more generic. Some of the customizations are considered in light of the requirements mentioned in section 6.4.
The HMI algorithm facilitates the generation of the results by effectively utilizing the primary and secondary inputs and the PCE Hybrid Model. The results include not only the estimated product and manufacturing costs but the elemental values including the estimated material, labour, processing and material-dependent costs. In addition, the cumulative costs and the deviation indices are effectively calculated. All the results generated through the implementation of the HMI algorithm are discussed in detail in section 6.4.
152
2
Labour units (LU) calculation for pth product in the nth year LU calculation at jth work centre (WC)
Lead times at jth WC, number of skilled, semi-skilled and non-skilled labour, labour indices Eq (5-10) or (5-11)
No
LU calculated at all j WC ?
Yes
Total LU calculation for pth product
Eq (5-12)
No
Total LU for all p products calculated?

Yes
Number of units produced in the nth year for all p products Eq (5-13)
Shop floor-wide LU calculation in the nth year
Total wages paid in the nth year to all workers Eq (5-14)
Direct labour cost calculation in the nth year
Actual labour rate (LR) calculation for the nth year
Eq (5-15)
LCDI calculation for skilled, semi-skilled and non- skilled labour and then average LCDI for the (n+1)th year
Average per month wages of skilled, semi-skilled and non-skilled labour in the nth and (n-1)th years, inflation in the nth and (n+1)th years Eqs (B-6), (B-7) and (B-8)
Estimated LR calculation for the (n+1)th year Estimated labour cost for pth product in the (n+1)th year
Eq (5-16)
Eq (5-8)
2A
Figure 6-3: Labour cost estimation implementation process
153
3 Processing units (PU) calculation for pth product in the nth year PU calculation at ith machine centre (MC)
Processing time at the ith MC, machine index Eq (5-19)
No
PU calculated at all i MC?

Yes
Eq (5-20)
Total PU calculation for pth product

Number of units produced in the nth year for all p products
No
Total PU for all p products calculated?

Yes
Eq (5-21)
Shop floor-wide PU calculation in the nth year

th
Financing expenses and insurance, maintenance, repair and energy costs incurred in the nth year Eq (5-17)
Processing cost calculation in the n year

Eq (5-22)
Aggregate processing rate (PR) calculation for the nth year

Aggregate processing rates for the nth and (n-1)th years, inflation in the nth and (n+1)th years Eq (B-9)
PCDI calculation for the (n+1)th year
Estimated PR calculation for the (n+1)th year

Eq (5-23)
Estimated processing cost for pth product in the (n+1)th year
Eq (5-18)
3A
Figure 6-4: Processing cost estimation implementation process
154
Total MDC calculation for the nth and the (n-1)th years
Indirect material, stores & inventory, freight & transportation, material inspection, purchasing, packaging and material handling costs and scrap for the nth and (n-1)th years Eq (5-24)
MDCDI calculation for (n+1)th year
Total MDC and cumulative material costs for the nth and (n-1)th years, inflation in the nth and (n+1)th years Eq (B-10)
Estimated MDC for the pth product in the (n+1)th year

Eq (5-25)
4A
Figure 6-5: MDC estimation implementation process
Manufacturing cost estimation (MCE) for pth product in the (n+1)th year
Estimated material, labour, processing and materialdependent costs for the pth product in the (n+1)th year Eq (5-2)
Cumulative manufacturing costs in the nth and the (n-1)th years
Cumulative material, labour, processing and material dependent costs for the nth and (n-1)th years Eq (5-44)
5A
Figure 6-6: Manufacturing cost estimation implementation process
155
Total PO in the nth and the (n-1)th years
Computer related and general admin costs and selling expenses in the nth and the (n-1)th years Eq (5-42)
PODI calculation for the (n+1)th year
Total PO and cumulative manufacturing costs in the nth and (n-1)th years, inflation in the nth and (n+1)th years Eq (B-14)
Estimated PO for the pth product in the (n+1)th year

Eq (5-43)
6A
Figure 6-7: PO estimation implementation process
In order to fully appreciate and understand the PCE Hybrid Model, the HMI algorithm and the generated results, it is essential to understand the companys method of cost estimation and the estimated results. The next section, therefore, details the companys estimation method and the generated results before elaborating the Hybrid Models results.
156
6.3
PCE at the company
A well established crane and ship engineering company based in the UK (identity is shielded due to confidentiality issues) was selected for the study as a representative case in the country. The company designs, develops, manufactures, and markets a wide range of cranes and parts for cranes and ships. It also provides after sales, installation and commissioning services. The organizational structure of the company is divided into crane manufacturing division (CMD) and ship engineering division (SED). The structural verticalization within the organization meant that the divisions and sub-divisions carry out their businesses independently. The total industrial output in the year 2005 was reported in excess of 100m with 25m and 75m divided between the CMD and the SED respectively. The CMD is divided into three sub-divisions namely spares & parts (SPSD), assembly & services (ASSD) and installation & commissioning (ICSD) with the respective industrial output in excess of 3m, 12m and 10m in the year 2005.
6.3.1
Information and details
The SPSD of the CMD was selected for the study. The main reason for the selection of this sub-division in particular and the company in general was that the developed models scope in terms of its parameters was likely to be covered there. The company was keen on initiating the validation process from a relatively smaller unit. Table 6-1 highlights the industrial output values in (s) from 2002-2005 for the CMD.
157
Design of parts and products at the company takes place at the divisional level. New product designs are normally generated for the ASSD where crane assemblies and subassemblies are manufactured from the outsourced parts and/or procured from the SPSD. The designs for products manufactured at the SPSD are already made, product range being the standard. However, any minor changes are customized to the actual product designs. Table 6-2 outlines the product range for the SPSD for the duration of the study.
The SPSD comprise fabrication and assembly units. There are a total of 20 different machines in the sub-division. During its manufacture a product is first routed on the machines within the fabrication unit before being routed through the assembly unit machines until its final completion. The products route through a set of predetermined machines within each unit based on its processing requirements. On its completion, a product is handed over to the CMD sales from where it is either shipped to the customers or sent to the ASSD.
Table 6-1: Industrial output values for the CMD (2002 2005)
Industrial Output Values

SPSD ASSD ICSD Total (CMD)
2002
2,192,384 10,983,782 9,901,843 23,078,009
2003
2,534,222 11,625,460 9,955,987 24,115,669
2004
2,895,843 12,098,570 9,980,350 24,974,763
2005
3,200,353 12,595,000 10,048,380 25,843,733
158
Table 6-2: Product range at the SPSD
S. No. Product Number

1. 2. 3. 4. 5. 6. 7. 8. MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024
Product Description
Tripod crane Tripod crane Overhead mounted bridge crane gripping manipulator Overhead mounted gripping manipulator Overhead trolley mounted gripping manipulator Overhead mounted linear track gripping manipulator Column mounted double wire gripping manipulator Column mounted gripping manipulator
Products within a unit are manufactured either in small batches or independently. Product quantities are planned based on quarterly sales forecast released a quarter in advance of the actual production. For example, the quantities for production in the third quarter are released at the end of the first or the beginning of the second quarter in order to make necessary arrangements for material procurement and production planning and control. Product cost evaluation and review is carried out yearly. Table 6-3 gives out the details of the product quantities produced in the SPSD from 2001 to 2005.
Product cost at the company is made up of material cost, labour cost and overheads. Actual unit product costs are obtained at the end of a financial year from the cost accountancy data
159
based on the cumulative production cost of a given product and its respective number of units produced in a given year. The values are given in the Table 6-3 and were provided by the company.
Table 6-3: Yearly production quantities for the product range in the SPSD (2001 2005) and Actual unit product cost for the given product range (2002 2005)
Product 2001
MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024 430 265 145 175 280 432 265 318
No of units produced 2002

450 250 150 150 325 375 245 325
Actual Product Cost per unit ()
2003
400 230 165 135 360 450 298 350
2004
485 200 190 165 345 319 335 245
2005
324 265 175 185 378 316 285 329
2002
1068 955 1948 1267 1492 487 807 386
2003
1197 1082 2155 1388 1659 540 911 435
2004
1344 1226 2370 1551 1851 628 1005 477
2005
1520 1428 2665 1740 2100 729 1174 555
6.3.2
Material cost here refers to the direct material costs. Material quantities are ordered based on the planned orders. The actual material quantities take into account the actual scrap and 160
wastage. The actual material costs are, therefore, only available after the production and can be obtained from the financial data at the end of the production year. Table 6-4 presents the actual material costs in (s) for the product range during the period of study. The cumulative material costs were obtained at the end of a given financial year and used to furnish the unit material costs for the respective products for that year.
Table 6-4: Actual material cost (Cumulative and per unit costs) Product 2001 MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024 Total 214140 110240 176610 123550 225400 101520 84535 52470 1,088,465 Total material cost (Actual) () 2002 228100 105120 185210 112125 280520 95240 87620 54200 1,148,135 2003 211200 101430 215325 104625 327600 117900 110856 59500 1,248,436 2004 265780 92000 259540 135300 331200 92510 123950 40425 1,340,705 2005 181440 129850 252350 158175 383670 98592 112575 56588 1,373,240 Unit material cost (Actual) () 2001 498 416 1218 706 805 235 319 165 2002 507 420 1235 748 863 254 358 167 2003 528 441 1305 775 910 262 372 170 2004 548 460 1366 820 960 290 370 165 2005 560 490 1442 855 1015 312 395 172
The material cost calculation for a specific product in a given year is based on the amounts of different materials in that product and their respective unit costs for that year. Material quantities are obtainable from BOMs (see example in appendix A). Since the products are
161
standardized, BOMs are established. Actual material costs incurred for a product can vary based on a number of factors and are available after its production and generally at the end of a financial year.
Table 6-5: Estimated unit material costs for the given product range (2003 2005)
Product
2003 MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024 530 439 1290 781 902 265 374 174
Estimated unit material cost ()

2004 554 463 1370 814 956 275 391 179 2005 574 482 1430 859 1005 304 387 173
Material cost estimation at the SPSD is based on considering the actual product costs at the end of a given year to predict for the following year. Average inflation values are used to estimate the material costs for the following year. A cost uncertainty index (CUI) value is also used on top of the inflation to account for any uncertainties. The average inflation values of 2.4, 2.5, 3.0, 3.5 and 3.2 percent were used respectively for 2001 2005 along 162
with a fixed CUI value of 1.5 percent. For example, the actual material cost value of 507 for MQ4033 obtained at the end of year 2002 was used to estimate for the year 2003 to be 530. Table 6-5 gives out the estimated material cost values for the given product range predicted by the company for 2003 2005.
6.3.3
Labour cost estimation
The total labour cost in a given year for the SPSD is the summation of the wages paid to all the workers in both the fabrication and the assembly units for that year. The yearly capacity is given in man-hours and is calculated by multiplying the number of working days in a year, the average number of shifts per day, the number of hours per shift and the average number of workers working per shift. The total labour cost obtained at the end of a given year is divided by the capacity to provide an aggregate shop floor-wide labour rate. Table 6-6 provides the aggregate labour rate values for the respective years.
163
Table 6-6: Aggregate labour rate calculation
2001
Shifts per day No. of working days per year No. of hours per shift No. of workers per shift Capacity (man hours) Total Wages () Aggregate Labour Rate (/hr) 1 250 8 21 42,000 240,120 5.72
2002
1 250 8 26 52,000 336,600 6.47
2003
1 250 8 26 52,000 387,000 7.44
2004
1 250 8 26 52,000 460,080 8.85
2005
1 250 8 26 52,000 507,600 9.76
Based on the lead times and number of workers working on different products in each of the shop floor units, the total man-hours for individual products can be calculated as shown in Table 6-7. The labour rate available at the end of a given year is used for the following year to estimate the labour costs for individual products by multiplying the rate with the products corresponding man-hour values. For example, the labour rate of 6.47 /hr obtained at the end of 2002 was set as an estimated rate for the following year giving an estimated labour cost value of 337 in the year 2003 for MQ4033. Table 6-7 tabulates the estimated labour costs also.
164
Table 6-7: Total man-hours and estimated labour costs (2003 2005) for the given product range Estimated Labour Cost ()
Product
Fabrication Unit
Assembly Unit
Total
Lead time (min) MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024 240 330 180 150 220 120 220 120
Lead time (hr) 4.0 5.5 3.0 2.5 3.7 2.0 3.7 2.0
No. of workers
manhours
Lead time (min) 240 200 220 180 180 100 230 180
Lead time (hr) 4.0 3.3 3.7 3.0 3.0 1.7 3.8 3.0
No. of workers
manhours
manhours
2003
2004
2005
5 5 6 6 7 6 6 3
20.0 27.5 18.0 15.0 25.7 12.0 22.0 6.0
8 5 4 6 6 4 5 3
32.0 16.7 14.7 18.0 18.0 6.7 19.2 9.0
52.0 44.2 32.7 33.0 43.7 18.7 41.2 15.0
337 286 211 214 283 121 266 97
387 329 243 246 325 139 306 112
460 391 289 292 386 165 364 133
6.3.4
Overhead estimation
Overhead costs at the company refer to all the costs other than material and labour costs. Overheads incurred in a given year are known at the end of a financial year. The overheads incurred in that year are summed up and divided by the man-hour capacity to give the overhead rate. The rate is used for the following year to estimate the overheads likely to be incurred on individual products by multiplying it with the corresponding man-hour values for the products. Table 6-8 details the yearly overheads in s obtained at the end of each 165
financial year. Scrap values are presented in negative to show the salvage value. The individual overhead cost elements represent categories with like costs. Indirect material here not only represents the cost on material used for shared benefits of products but the tools utilized. Since the tooling is not heavy with comparatively lower costs, the company has not assigned a separate cost pool for the tooling cost. Insurance costs here refer to the costs of insuring machinery, equipment, building and all other insurable items. However, the costs of insuring the equipment and machinery are higher than any other insurance costs. Any costs not incurred for the direct benefit of the SPSD but for the shared benefit of the company are charged proportionately.
Total actual production cost for a year can be calculated by summing the total material, wages and overheads in that year and are also given in Table 6-8. The estimated values for product costs using the companys method can now be given by summing the three estimated elements: material, labour and overheads as mentioned in Table 6-9. The table also gives the cumulative values for the estimated production costs (2003 2005).
166
Table 6-8: Overhead costs for individual elements (2001 2005) 2001 Indirect material Stores & Inventory control Freight & Transportation Material Inspection Purchasing Packaging Material handling Insurance cost Financing Expenses Maintenance Cost Repair Cost Energy cost Computer related cost General admin cost Selling Expenses (Scrap) Total Overheads (actual) Overhead rate (/hr) Total production cost (actual) 55050 45600 35200 24500 17800 26050 16890 67300 24500 20850 18853 57200 44280 139660 95255 -34350 654,638 15.59 1,983,223 2002 56500 48900 39900 25600 18970 27700 17290 69500 28000 25424 21220 65000 45550 145650 109245 -36800 707,649 13.61 2,192,384 2003 64200 54200 58200 26500 19250 32540 19260 72000 36000 38586 38106 98000 64870 182560 132364 -37850 898,786 17.28 2,534,222 2004 75200 56240 62400 27000 21130 45230 19265 88000 42000 51384 25814 130000 85925 210220 194500 -39250 1,095,058 21.06 2,895,843 2005 75800 62450 85240 27560 25360 51230 21450 98400 45000 62598 41852 160000 105425 238954 256450 -38256 1,319,513 25.38 3,200,353
167
Table 6-9: Estimated product cost values (2003 2005) Product Estimated product cost per unit 2003 MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024 Total 1574 1326 1946 1444 1779 640 1200 475 2004 1840 1555 2178 1630 2035 737 1409 549 2005 2129 1802 2407 1845 2311 862 1619 621 Estimated production cost (Cumulative) 2003 629579 305059 321139 194918 640397 288117 357726 166425 2,903,361 2004 892490 311029 413818 268905 702155 234996 471852 134603 3,429,848 2005 689762 477660 421251 341409 873572 272356 461284 204425 3,741,718
6.4
Implementation of the PCE Hybrid Model
The developed model presents the estimated product cost as a summation of manufacturing cost, engineering cost and production overheads. Since the engineering designs for the SPSD are already established, the engineering costs in terms of design and development expenditures are negligible. Any minor costs within the context are incurred at the ASSD and subsequently charged to the SPSD within general administration overheads. These costs were not separately provided by the company. Manufacturing cost and production overheads would, therefore, add up to the estimated product cost for an individual product. 168
Manufacturing cost is presented as material, labour and indirect manufacturing costs. Indirect manufacturing costs comprise processing, material-dependent, tooling and building space costs. Any cost incurred on tooling at the company is not separately recorded but combined with the indirect material cost and would form part of material-dependent cost (MDC) estimation. Plant-wide building maintenance, repair and energy costs are allocated in proportion to the industrial output and charged to the division and subsequently combined with equipment & tool maintenance, repair and energy costs respectively. These costs would, therefore, be dealt with part of processing cost estimation. Manufacturing cost estimation in this case would be the sum of material, labour, processing and MDC.
The application of the HMI algorithm presented in section 6.2 can be followed from the following sub-sections in order to facilitate the implementation of the Hybrid Model and generate estimated cost values.
6.4.1
Material cost estimation using the presented Hybrid Model required calculation of material cost deviation index, n +1 . Material cost deviation trends from the past years could be used to predict the index values for the following years based on the model derived (see
appendix B) as follows:
n +1 p = I n +1 + (1 + I n +1 )
n C mp 1 I n n 1 C mp
(6-1)
169
n +1 Where, p is the deviation index for the pth product estimated for the following year. I n n and and I n+1 are the interest rates in the current and the following year respectively. C mp n 1 C mp are the material costs for the pth product in the current and the preceding year
respectively.
Table 6-10: Material cost deviation index and estimated unit material cost values (2003 2005) Material cost deviation index (%)
n +1 p
Estimated unit material cost () Product (Hybrid Model)
2003 MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024 518 427 1257 796 931 276 404 169
2004 553 465 1387 808 965 272 389 174 Average index (
n +1
2005 567 479 1426 866 1010 321 366 159 )
2003 2.26 1.53 1.84 6.48 7.86 8.74 12.90 1.53 5.39
2004 4.71 5.45 6.29 4.20 6.01 3.67 4.55 2.40 4.66
2005 3.50 4.03 4.41 5.58 5.26 10.62 -0.97 -3.45 3.62
By using all the known values, the model can predict the indices for the product range for the following year. The values can then be used to predict the estimated material cost 170
values. An average index value can also be used for the purpose but it may compromise the accuracy of the individual results. However, the purpose of the average value is best served for a new product with no known past data. The material cost values available at a given time and the average index value can then be used to predict the product cost for the following year. Table 6-10 gives out the indices and the estimated material cost values.
6.4.2
Labour cost estimation
Labour cost estimation using the hybrid method requires the estimation of shop floor-wide aggregate labour rate and the labour units consumed for each product. The lead times for the individual products in the two work centres and the corresponding data obtained from the company for the number of skilled (x), semi-skilled (y) and non-skilled (z) workers required to manufacture the respective products can be used to determine labour units for each product. Skill indices of 0.75 and 0.35 were set for semi-skilled and non-skilled workers respectively based on an average skill level of around 75 and 35 percent with respect to their skilled counterparts. Table 6-11 tabulates the labour units in hours required for the manufacture of the individual products. The table also calculates the total number of labour units consumed in each year using the data for the number of units produced in corresponding years.
The total labour units and the total wages paid in a year can be used to determine the labour rate for that year. The rate can then be used to determine the estimated labour rate for the following year by considering the LCDI, n +1 for the following year. The LCDI values can
171
be found by using the equations (B-6), (B-7) and (B-8). Average monthly wages of the skilled, semi-skilled and non-skilled workers obtained from the company were used for that purpose. The average index values as obtained can be used to determine the estimated labour rates. The estimated rates can be used to generate estimated labour costs for individual products based on their respective labour units as shown in Table 6-12.
Table 6-11: Labour units calculation Product Fabrication x MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024 1 2 1 2 2 1 2 1 y 2 2 2 1 3 2 1 1 z 2 1 3 3 2 3 3 1 Assembly x 2 2 2 2 1 1 2 1 y 3 1 1 2 2 1 1 1 z 3 2 1 2 3 2 2 1 Labour Units (hr) 34.00 32.68 22.02 22.10 28.80 11.18 27.16 10.50 Total 2001 14620 8659 3192 3868 8064 4831 7197 3339 53770 2002 15300 8169 3303 3315 9360 4194 6654 3413 53706 2003 13600 7515 3633 2984 10368 5033 8093 3675 54900 2004 16490 6535 4183 3647 9936 3567 9098 2573 56029 2005 11016 8659 3853 4089 10886 3534 7740 3455 53231 Total labour units consumed
0.75
0.75
0.35
0.35
172
Table 6-12: Labour cost deviation index and estimated labour costs Average wages per month (single employee) () 2001 Skilled Semi Nonskilled 980 810 2002 1100 950 2003 1290 1080 2004 1580 1280 2005 1700 1400 Labour cost deviation index (%) 2003 13.04 18.23 2004 18.27 14.56 2005 22.79 18.70 MQ4033 MQ4030 Estimated unit labour cost () 2003 248 248 2004 273 272 2005 326 326
Product
650
800
910
1050
1200
24.19
14.63
15.46
MQ4024
165
181
217
Average Index ( Labour Rate (Hybrid Model) (/hr) Actual Estimated 4.47 6.27 7.05 7.43
n +1
18.49
15.82
18.98
MQ3522 MQ2538
167 210 82 208 79
183 231 90 228 86
219 277 107 273 103
8.21 8.16
9.54 9.77
MQ1033 GQ4026 GQ4024
6.4.3
Processing cost estimation
The total processing costs spent on the shop floor refer to the necessary costs to keep the floor facilities in the running condition. Financing expenses, insurance, maintenance, repair and energy costs obtained at the end of the financial year add up to the total processing costs for the year. Processing costs for a product can be estimated by multiplying the estimated rate with the processing units of the product. Processing units refer to the total time the product spends on the work centres while being processed. The combined lead time, therefore, for a given product spent on the fabrication and the assembly units would
173
refer to the processing units for the product. The total processing costs for a given year and the total processing units consumed in that year can be effectively utilized to produce the processing rate for the year.
Table 6-13: Estimation of processing units, deviation indices, rates and costs Estimated processing cost per unit () 2005 2592 2341 1167 1018 2520 1159 2138 1645 14578 407850 27.98 20.15 26.96 2003 128 142 107 88 107 59 120 80 2004 198 219 165 136 165 91 186 124 2005 216 238 180 148 180 99 202 135
Product
Processing Units (hr) 2001
Processing units (Total)
2002 3600 2208 1000 825 2167 1375 1838 1625 14638 209144 14.29
n +1
2003 3200 2032 1100 743 2400 1650 2235 1750 15109 282692 18.71 12.37 16.06
2004 3880 1767 1267 908 2300 1170 2513 1225 15028 337198 22.44 32.42 24.78
MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024 Total
8.00 8.83 6.67 5.50 6.67 3.67 7.50 5.00
3440 2341 967 963 1867 1584 1988 1590 14738 188703 12.80
Processing costs (Total) Processing rate (actual)
Processing cost deviation index ( Processing rate (estimated) (/hr)
) (%)
174
However, predicting the rate for the following year would require considering any processing costs deviations. The method for calculating such deviations mentioned earlier can be used to predict the PCDI values, n +1 , for the following year by using the equation (B-9). The PCDI values can then be used to generate the estimated processing rates.
Estimated processing costs can then be determined for the range of products. Table 6-13 covers the estimation of processing units per unit and the total processing units required in a given year. It also calculates the rates and the deviation index values and then gives out the estimated processing cost per unit for the given products.
6.4.4
MDC estimation
The total MDC for the SPSD for a given year is obtained by summing up indirect material, stores and inventory, freight and transportation, material inspection, purchasing, packaging and material handling costs and subtracting the scrap value for that year. MDC per unit material consumed can then be used to determine the MDC deviation index ( n+1 ) values using the method described earlier and given by equation (B-10). This allows the estimation of the MDC per unit material cost for the following year resulting in the estimation of the MDC per unit of the corresponding products for that year. Table 6-14 outlines all these values.
175
Table 6-14: MDC deviation index and estimated MDC per unit values 2001 Total MDC ( C md ) () Cumulative material cost ( C mt ) () MDC fraction ( C md / C mt ) MDC deviation index ( Estimated MDC fraction ( C md / C mt )x( 1 +
n n n +1 n n n n
2002 198,060
2003 236,300
2004 267,215
2005 310,834
186,740
1,088,465
1,148,135
1,248,436
1,340,705
1,373,240
0.172
0.173
0.189
0.199
0.226
) (%)
0.99
10.46
5.06
n+1 )
0.174
0.209
0.209
Product MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024
Estimated MDC per unit () 90 74 219 139 162 48 70 29 116 97 290 169 202 57 81 36 119 100 299 181 212 67 77 33
176
6.4.5
Production overheads estimation
The total PO for the SPSD in a given year can be obtained by adding selling expenses, computer related costs and general administration costs. The total actual manufacturing costs for the SPSD in a year refers to the summation of the cumulative material costs, total wages, total MDC and the total processing costs incurred in that year.
Table 6-15 PO fractions (actual & estimated) and deviation indices 2001 Total manufacturing cost (actual) ( CGt ) () Production overheads ( Ot ) () PO fraction ( Ot / CGt ) PO deviation index (
n +1 n n n n
2002
2003
2004
2005
1,704,028
1,891,939
2,154,428
2,405,198
2,599,524
279,195
300,445
379,794
490,645
600,829
0.164 ) (%)
n n n +1
0.159
0.176 -2.74
0.204 11.79
0.231 15.81
Estimated PO fraction ( Ot / CGt )x( 1 +
0.154
0.197
0.236
The total PO and the total manufacturing costs can be used to establish the PO fractions and subsequently the estimated values of PO fractions. The estimated values of the fractions require the calculation of the PO deviation indices, n+1 using the method mentioned earlier for the calculation of the indices and given by equation (B-14). Table 6-15 explains the calculation of the fractions and the indices. By adding the already estimated manufacturing cost elements, the estimated values for manufacturing costs for individual products in a
177
given year can be obtained. These values and the corresponding estimated PO fractions can be used to estimate the production overheads for the individual products. Estimated product cost can then be determined by combining the manufacturing and the PO cost elements estimated for the product. Table 6-16 tabulates the estimated manufacturing, PO and product costs per unit.
Table 6-16: Estimated per unit values for manufacturing, PO and product costs Estimated product manufacturing cost per unit () 2003 MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024 990 886 1747 1187 1414 466 796 357 2004 1144 1048 2022 1293 1567 511 878 420 2005 1234 1136 2120 1411 1683 596 911 430
Estimated production overheads per unit ()
Estimated product cost per unit ()
2003 153 137 270 183 218 72 123 55
2004 225 207 398 255 309 101 173 83
2005 291 268 501 333 398 141 215 102
2003 1142 1023 2017 1370 1633 538 919 412
2004 1370 1254 2420 1548 1875 611 1051 503
2005 1525 1405 2621 1745 2081 737 1126 532
178
6.5
Conclusions
This chapter discussed the implementation of the Hybrid Model in a batch manufacturing environment in the UK as part of the validation process for the model. The implementation phase involved the retrospective application of the PCE Hybrid Model along with the companys method of cost estimation. The implementation of the Hybrid Model was facilitated by the HMI algorithm. Cost data at the end of a given year were effectively used with the help of the proposed deviation indices, resulting in furnishing the estimated costs for the given product range well before the actual manufacturing started.
It was noted that the Hybrid Model with the help of the HMI algorithm generated detailed cost breakdown values for every product. The companys method was not able to predict such breakdown values. The provision of breakdown values is helpful in identifying high cost elements and subsequently helps in making cost control decisions. The cumulative cost values were also obtained that can be helpful in making statistical and financial analysis of the cost data at the organizational level. Such analyses support key operational and strategic decisions for middle and senior management.
The Hybrid Model attempted to overcome the limitations associated with TRO and MRO models. For example, inflation and other cost deviations were incorporated through the introduction of deviation indices that also optimized the use of past data. The use of normal capacity hours during labour cost and overheads (indirect costs) calculation was replaced by a new method based on calculating labour and processing units etc.
179
Deviation indices are crucial to the overall accuracy of the estimated results and as such demand further investigations. For example, there is a need to investigate the effects of more than two years past cost trends. The HMI algorithm can also be made more generic by adding components for tooling and building space costs calculation in line with the Hybrid Model. Care should be taken while categorizing the indirect elements to different subcategories. For example, material handling costs are categorized under material-dependent costs based on its relation with the increasing material quantity. Increased material quantities result in increased handling and thus in increased costs. However, an increased handling time should not be confused with processing time and the resulting costs should not be categorized under processing costs. Because processing costs are based on job processing times at work or machine centres. On the other hand, an increased material handling time is accounted for in increased material quantities. Similar care should be taken for other overheads elements.
In short, previous limitations found in TRO and MRO estimation methods were removed. The obtained results are compared in the next chapter for the validation analysis.
180
Chapter 7
Comparisons and Validation Analysis
The aim of Chapter 7 is to validate the developed PCE Hybrid Model by comparing the estimated costs obtained in the previous chapter and in the current chapter against the actual costs for a range of products. The process will involve analysing the costs at different levels including the entire product costs, cumulative production costs, overheads for an individual product and overheads at cumulative level. The chapter will also present a breakdown analysis.
7.1
Introduction
The route to successful validation is inscribed in achieving the objectives. This in turn requires outlining the validation method. In Chapter 4, the MRO and TRO estimation methods were presented, implemented and validated by comparing the results obtained from the application of the model with those generated by the companys method. The model presented in Chapter 5 is more comprehensive and requires more than just scrutinizing it against an in-house methodology. A decision, therefore, was made to implement a published model and obtain the estimated cost results from its application alongside the already obtained results in Chapter 6 from the application of the PCE Hybrid
181
Chapter 7: Comparisons and Validation Analysis
Model and the companys method. The three results will then be compared against the actual product costs. Comparing the estimated product costs from the three methods against the actual costs can only strengthen the argument to justify the superiority of the developed methodology. The newly developed Hybrid Model was therefore, implemented alongside a published model in the selected manufacturing company for a retrospective analysis of three years (2003 2005).
Chapter 7 presents the comparisons and validation analysis. This is based on comparing the estimated cost values (2003 2005) obtained using the three methods (the companys method, the selected published model and the Hybrid Model) against the actual cost values provided by the company and given in Table 6-3, Table 6-4, Table 6-6 and Table 6-8. An analysis of the cost variations from the actual costs is presented by not just considering the cumulative production costs but the individual product costs and the cost elements, where possible. Almost every level of the analysis demonstrates the superiority of the developed model. An analysis for cost elements breakdown based on the actual production costs is also drawn. The breakdown results obtained highlight the typical tendencies of the region through the analysis of a representative case. The analysis when compared with the results obtained for the South Asian region presented in Chapter 4 provides the basis for understanding and developing an effective CCS.
The actual product cost values were provided by the company and calculated based on the actual cost figures obtained from the cost accountancy data at the end of each year. These considered the actual material and labour costs. Material costs considered the actual 182
material quantities and wastage. Labour costs were based on the actual time spent on a product or a batch of products taken from the time cards on the shop floor. Any other costs were allocated to the products based on either activity rates and units or their traceability to the products. For example, material handling, stores & inventory, inspection and quality control costs were allocated based on activity rates and units. Freight & transportation, packaging, financing expenses etc. were in most cases traceable to the products. However, any exact method for their calculation was not given by the company.
Before the validation analysis, the results from a published model would be obtained in order to compare the three estimated results against the actual costs. Section 7.2 details the cost estimation results for the given product range using the selected published model as a representative case from the domain. The section, therefore serves the preparation before comparison analysis. Section 7.3 presents the product cost analysis based on comparing the estimated product costs from the three methods against the actual costs. Section 7.4 compares the estimated and the actual cumulative costs. Section 7.5 analyses estimated and actual overheads at product and production level. Section 7.6 presents a statistical analysis of cost breakdown results. Section 7.7 concludes the chapter.
7.2
Preparations for comparisons
The purpose of the implementation of a published model is to validate the developed model by not just comparing it against the industrially established tool but to present its
183
superiority against an already established and recognized representative case from the published research domain. Therefore, the selected model should already have been tested. In order to draw comparison between the developed and the published models, it is essential that the published model should have the necessary parameters to be applicable within the framework of the manufacturing enterprise. The available field data should form the basis for the selection of the model and to further evaluate costs.
The model presented by Jung [84] was selected for the implementation and validation purposes as it not only complied with the above criteria but is established in the field. Jung described major cost elements as material, labour, engineering and burden cost. He did not model engineering costs but manufacturing costs were presented as follows:
Mfg cost = (Ro + Rm)[( Tsu/Q)Tot + Tno] + material cost + factory expenses
(7-1)
where, Ro = operators rate, Rm = machine rate, Tsu = set-up time, Q = batch size, Tot = operation time, Tno = non- operation time.
Three different times are considered but refer to the manufacturing lead time (MLT) time for a product. Labour and machine related costs were described using the MLT. Burden cost or factory expenses refer to the costs other than material, labour and machine related costs and can be considered as the overheads spent for the combined benefit of all the products. In an industrial environment where engineering costs are relatively low, negligible or accumulated with the overheads, Jungs model refers to the overall product 184
cost. In the given conditions where engineering costs, if any, are part of the general administration costs, the model can predict product costs. The major cost elements, therefore, comprise labour and machine costs, material cost and factory expenses.
7.2.1
Labour and machine cost estimation
Labour and machine cost estimation using Jungs model requires the operators and machine rates calculation. The operators rate (Ro) is based on the shop floors direct labour cost. The total wages can, therefore, be divided by the shop floor-wide cumulative MLT to obtain the rate. The floor-wide MLT in a given year can be obtained by lead times of the individual products and the respective number of units produced in that year. Similarly, machine running costs (maintenance, repair and energy costs) can be divided by the total MLT to determine the machine rates (Rm).
The obtained rates at the end of a given year can be used for the following years labour and machine cost estimation process by considering the effect of inflation in that year. Since, Jungs model does not describe the ways to consider any cost variances, only the known variation (i.e. average inflation) is considered for the estimation purposes. The summation of the estimated rates can be multiplied by the individual lead times to estimate the labour and machine costs for individual products. Table 7-1 tabulates the lead times (individual and cumulative), machine running costs (the summation of maintenance, repair and energy costs), the rates (operator and machine) and the estimated labour and machine costs per unit.
185
Table 7-1: Lead times, machine running costs, rates (operator and machine) and labour and machine costs Manufacturing lead time (MLT) (total) Product Lead time per unit (hr) 2001 MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024 Total (hr) Operator's rate (Ro ) (/hr) Machine running cost () Machine rate (Rm) (/hr) 6.57 7.63 11.56 13.79 18.14 96,903 111,644 174,692 207,198 264,450 16.29 23.00 25.61 30.61 34.82 8.00 8.83 6.67 5.50 6.67 3.67 7.50 5.00 3440 2341 967 963 1867 1584 1988 1590 14738 2002 3600 2208 1000 825 2167 1375 1838 1625 14638 2003 3200 2032 1100 743 2400 1650 2235 1750 15109 2004 3880 1767 1267 908 2300 1170 2513 1225 15028 2005 2592 2341 1167 1018 2520 1159 2138 1645 14578 2003 252 279 210 173 210 116 237 158 2004 308 340 257 212 257 141 289 192 2005 367 405 305 252 305 168 344 229 (hr) Labour and machine cost per unit ()
186
7.2.2
Material cost and factory expenses
The amounts of all the material used for the manufacture of a product and their unit prices can be used to ascertain the total material cost of the product. The only known cost deviation (i.e. the average inflation) can then be accounted for in order to predict the total material cost for the product for the following year. For example, the material cost of 522 can be set for MQ4033 for 2003 based on the known value of 507 from 2002 and the inflation rate of 3 percent for 2003.
Ascertaining the factory expenses for an individual product requires the calculation of an expenses rate. The rate can be calculated by dividing the total factory expenses from the cumulative MLT for a given year. The rates after taking into account any deviations effect (inflation in this case) for the following year can be multiplied by the total lead times of the products to furnish their corresponding estimated factory expenses values in that year. Whereas, the total factory expenses for a given year can be obtained by summing all the costs other than material and labour costs. Since, the maintenance, repair and energy costs were considered in the calculation for machine related costs; they would also be excluded from the factory expenses. The estimated product costs can then be given by adding the corresponding factory expenses, material, labour and machine cost components. Table 7-2 calculates the factory expenses rate, the estimated factory expenses per unit and the estimated product costs per unit.
187
Table 7-2: The total factory expenses, the expenses rate, the estimated per unit values (factory expenses, product cost) (2003-2005) 2001 Factory expenses (Total) () Factory expenses rate (/hr) Product 2002 2003 2004 2005
557,735
596,005
724,094
887,860
1,055,063 Estimated product cost per
37.84
40.72
47.92
59.08
72.37
unit ()
Estimated factory expenses per unit () 2003 488 539 406 335 406 224 457 305 1110 1082 1762 1174 1379 531 919 539 2004 1251 1234 1938 1287 1529 594 1046 616 2005 1420 1418 2122 1434 1703 691 1183 704
MQ4033 MQ4030 MQ4024 MQ3522 MQ2538 MQ1033 GQ4026 GQ4024
336 370 280 231 280 154 315 210
397 438 331 273 331 182 372 248
188
7.3
Comparison analysis for product cost
The estimated product costs using the Hybrid Model (new method), the companys method and the Jungs model mentioned in Table 6-16, Table 6-9 and Table 7-2 respectively can be plotted against the actual product costs given in Table 6-3.
Figure 7-1 gives the cost comparison for the entire product range for the three year period (2003 2005). The green vertical columns represent the actual costs for respective products. The other three dotted lines represent the estimated product costs. The analysis clearly shows the superiority of the Hybrid Model as the dotted line representing the results from the Hybrid Model remains closer to the green vertical bars (the actual costs). For example, the estimated costs of 1110, 1574, 1142 were given by the Jungs model, the companys method and the Hybrid Model respectively for the year 2003 against the actual product cost of 1197 for the same year.
189
Figure 7-1: The comparison of the actual costs against the estimated costs
190
Any deviation of the estimated cost values from the actual cost represent the estimation error linked with a given estimation method. In addition to the cost analysis presented earlier, there is a need to analyse the estimation errors using the three methods. The estimation error trends using the three methods for the investigation period are given in Figure 7-2. Error estimations are based on deviations from the actual costs and presented in percentages. The actual cost is represented by the x-axis with zero error. It is clear that on almost every occasion the estimated values predicted by the Hybrid Model are closer to the actual values than those predicted by the other two methods. Although on an isolated case (MQ4030), the Jungs model predicted slightly better values, the Hybrid Models prediction was not only within the acceptable limit for that case but the overall accuracy is far more consistent. It can be seen that a cautious approach by the companys adopted method resulted in not just overestimation on almost every occasion but resulted in as much as 40 percent error. This was mainly due to an effort to avoid under estimating and setting a price more than the actual cost value. However, the loss of goodwill and sales as a result of overestimation could far outweigh any apparent profits as a result of overestimation resulting in net loss of potential profits. A yearly cost rise in the actual values and the estimated values is also visible due to an increase in almost all the cost element values. Jungs model underestimated on many occasions partly because of no provisions in the model for considering cost variances that could be useful in keeping up with the cost trends. Only the known variances were accounted for. Both under- and overestimations were observed within acceptable limits using the Hybrid Model. It is also clear that the use of deviation indices has eliminated the problem of underestimation associated with the TRO and MRO methods presented in chapter 4.
191
Figure 7-2: Estimation error trends for the three methods (2003 2005)
192
It is not uncommon in an industrial environment to focus on net error estimation analysis. While some of the products during a given period may be highly overestimated, the others remain underestimated. The net estimation error may be within acceptable limits. However, the approach can be dangerous as some of the products may be priced very low while the others could be highly priced. This could lead to losses on individual products due to underestimation. On the other hand, those overestimated and thus over-priced products could also result in loss of consumer confidence and loss of market. Therefore, instead of relying on net error estimation, there is a need to estimate individual products cost accurately.
An analysis based on a yearly consolidated representation for the investigated period considering the entire product range is given in Figure 7-3. Error estimation values are also given. Big error distributions across the product range are seen in a given year using both the companys and the Jungs method. The companys method overestimated for almost the entire product range throughout the investigation period. Jungs method mostly under estimated for the given product range throughout the investigation period. The values estimated by the Hybrid Model were either underestimated or overestimated within very close range of the actual costs. In any given year, the values were evenly spread across the entire product range.
193
Figure 7-3: Percentage cost estimation variations from actual product costs.
194
In order to visualise better the error distribution trends across the product range for a given year using the three methods, results are presented in Figure 7-4. The error trends for a given year demonstrate the effectiveness of the Hybrid Model when compared with the other two methods. Huge fluctuations are seen for the estimation error values using either the companys method or the Jungs model across the entire product range for the three year period. The Hybrid Models estimation trends, on the other hand, are clearly even and stay close to the actual values. In terms of not just the accuracy, but the consistency of the results, the Hybrid Model can be seen as a superior methodology. Estimation errors displayed on various occasions are almost negligible unlike those displayed by the other two methods. The effective use of the past data along with accurate prediction of the various indices is behind the accurate prediction of the overall product costs among other factors. Such precision is mainly down to eliminating the already observed shortcomings in the methodology presented in Chapter 4.
195
Figure 7-4: Estimation error trend across the product range
196
Due to huge fluctuations of estimation error across the entire product range using the two methods, some of the products may be highly over estimated while the others highly underestimated resulting in unrealistic price settings. Such a situation could have serious implications for the business objectives of an enterprise. Many companies tend to even out the errors by linearizing the error distribution. Even any error linearization could do little to help the situation. For example, the linearization of the errors for the companys method could help to even out the error distribution across the product range, but the overall error still stays unreasonably high (around 20 percent) as shown in Figure 7-5. Most importantly, such a linearization has little practical implications as such a comparison requires the actual costs which are not discovered until usually the products are produced with price quotes already furnished. However, for statistical analysis and accuracy optimization (based on the rectification of the consistently inaccurate trends) purposes, such results can be helpful. Past error distribution trends, however, may have to be used and relied upon at increased risk of compromising the accuracy for future results.
Similarly, the linearization carried out for both the Jungs model and the Hybrid Model are shown in Figure 7-6 and Figure 7-7 respectively. The linearization of the errors for the Jungs model finds the underestimation of up to around 20 percent and an overestimation of around 10 percent. The values for the Hybrid Model, on the other hand, do not exceed 5 percent for both under and overestimation.
197
Figure 7-5: Error linearization for the results given by the companys method
198
Figure 7-6: Error linearization for the results given by the Jungs method
199
Figure 7-7: Error linearization for the results given by the Hybrid Model
200
Another point noticeable from Figure 7-7 is a changing response for error estimation trend on yearly basis as opposed to the ones presented in Figure 7-5 and Figure 7-6. This is due to the fact that the Hybrid Model takes into account the changing factors to estimate costs. Factors like, inflation rate, expected cost deviations etc. are accounted for very well and by adjusting the response, the estimated values are always kept close to the actual costs. The incorporation of the case-based technique into the modelling framework is also responsible for sensitizing the model by keeping it aligned with the changing trends. While the model is sensitized to the trends, its robustness remains unquestionable. The overall validation analysis spread over a three-year period covering a range of products under varying conditions and huge fluctuations of cost deviations still produced consistent results within acceptable limits of error estimations. The estimated values are, therefore, reliable and can serve the operational purposes (such as price setting, cost control, lot sizing, etc.) well. More strategic decisions can also be relied upon (such as make or buy decisions, business resizing, etc.). The results produced by the other two methods are hugely desensitized to the changing cost trends and consistently produced results with huge estimation errors and unchanged trends. Although, the statistical data analysis can help to exploit the error consistencies to manually adjust the estimated values, the process would not be risk free and the essence of using an estimation method will be lost.
201
7.4
Comparison analysis for cumulative costs
In addition to the analysis for individual products, it is essential to carry out the validation for the cumulative cost values. This is also necessary to quell any notion of a possible selfadjustment of individual under and overestimation on the cumulative level. This means an attempt will be made to determine if any under or overestimation at product level even out each other at cumulative level. In other words, the current analysis is aimed at finding out if a methodology less accurate at estimating individual product costs can be more accurate at cumulative level. This would be done by comparing the total actual costs incurred in the SPSD during a given period against the cumulative values for the estimated results obtained by using the three methods.
Figure 7-8 (a) compares the three estimated cumulative costs with the actual costs at the cumulative level. The values are given in million . The black line representing the actual costs stays closer to the green vertical bars representing the estimated values given by the Hybrid Model (new method). Figure 7-8 (b) quantifies the estimation error. The estimation error values represent the percentage deviations from the actual costs. The estimated values predicted by the Hybrid Model at the cumulative level fluctuated between -3 and 2 percent for the overall period of the investigation. The companys method overestimated on the cumulative level by as much as around 19 percent despite fluctuating under and overestimation values at individual levels. The Jungs model, on the other hand, underestimated costs by around 9 percent at the cumulative level despite fluctuating values at the individual products. This demonstrate that even after taking into account any
202
adjustments of fluctuating values for individual products at the cumulative level, high degree of under and overestimations were observed using the two methods.
Actual Cost Vs Estimated Cost

Millions
(a)
4.0 3.5 3.0 2.5 2.0 2003 2004

Production Year
Cost Values (s)
2005
Total Estimated Production Cost (Jung's) Total Estimated Production Cost (New )
Total Estimated Production Cost (Comapny) Total Production Cost (Actual)
Estimation errors from cumulative actual costs (%)
(b)
20 14.57 Estimation variation (%) 10 1.79 0 -3.01 2003 -10 -7.66 2004 -8.40 Production Year -20 Estimation error using Jung's model (%) Estimation error using hybrid model (%) Estimation error using company's method (%) -1.17 2005 -9.29 18.44 16.92
Figure 7-8: (a) Cumulative actual costs against total estimated values, (b) estimation errors for cumulative costs
203
The Hybrid Model remained close to the actual costs at the cumulative level of prediction. Even if the cumulative results for the two methods were closer to the actual costs after considering any adjustments, the impact of huge under and overestimation on individual products could not be ignored. However, more accurate results for the Hybrid Model even at the cumulative level demonstrate the ultimate superiority of the model over the two methods considered. In addition, the highly accurate values predicted by the Hybrid Model both at the individual and the cumulative levels with consistency would be very difficult for any other method (not considered in the current validation analysis) to rival.
The Hybrid Model gives out optimized values at the cumulative level (i.e. more accurate than those predicted by the other two methods). The quantification of the optimized values against the two methods is shown in Figure 7-9. The results are based on considering the estimation error values for the cumulative costs. Only the error values in the respective years are considered ignoring the symbolic negative signs, if any (meant for showing underestimation), thus allowing the quantification of the deviations from the actual costs. The differences between the estimation error values for a given year using the Hybrid Model and the companys method resulting in the optimized values are mentioned in the first part of the figure. Those based on the Jungs model are presented in the second part. A maximum of over 16 and 8 percent optimizations were obtained against the companys and the Jungs methods respectively. Linearized trends are also included in both cases showing similar results even after adjusting the errors over the duration of the investigation. The results show significant improvements achieved by the Hybrid Model against the two methods.
204
Optimization achieved 2003-2005 (%)

18.00 16.65
(a)
15.75 15.00 y = 2.09x + 10.47 Optimization (%) 12.00 11.56 9.00
6.00
3.00 2003 0.00 2004 Production Year Optimization from company's method (%) Linear optimization trend from company's method (%) 2005
Optimization achieved 2003-2005 (%)

10.00 8.12 8.00 6.61 Optimization (%) 6.00 4.65 4.00 y = 1.73x + 2.99
(b)
2.00 2003 0.00 Optimization from Jung's model (%) Linear optimization trend from Jung's model (%) 2004 Production Year 2005
Figure 7-9: Optimization for the estimation accuracy achieved by the Hybrid Model against (a) the companys method (b) the Jungs model
205
7.5
Comparison analysis for product and production overheads
An important aspect of the validation analysis is to investigate the effects of the three methods on the accuracy of the breakdown elements. The accurate estimation of the breakdown elements would eventually lead to the accurate prediction of the overall product costs. Identifying the least accurate elements could help to better understand the problem areas in the existing methods and reason why the developed model is more efficient.
Direct and indirect elements are estimated in different ways using the three methods. Direct elements are easier to analyse because they are mainly estimated using the standard procedures. Since, the indirect costs were estimated with different approaches and comprise a larger portion of the overall product cost, it warrants investigation. The companys method termed indirect costs as overheads. Factory expenses were considered using the Jungs model and are part of the indirect costs. However, the machining costs have to be considered also for the total indirect costs. Processing cost, material dependent cost and production overheads comprise the total indirect cost using the Hybrid Model. Since, the three methods divide the indirect costs into different elements; the best way to make an analysis is to consider the overall indirect cost or total overheads instead of considering any sub-levels.
In order to compare the estimated overheads using the three methods, the actual values of overheads incurred by different products are required. The actual values were obtained 206
from the accountancy data at the end of a financial year and are allocated to individual products based on the ABC method. Activity rates are set and the activity times are based on the recorded values during a manufacturing phase. Although, the exact method was not disclosed by the company, the allocated overhead values for individual products were supplied.
The comparison of the estimated values for overheads using the three methods against the actual values is presented in Figure 7-10. The comparison is based on analysing the estimation error trends for the given product range. The error trends represent the percentage deviations from the actual costs. The solid lines represent the estimation error trends for the Hybrid Model. On almost every occasion, the estimation error produced by the Hybrid Model is not only lower than those presented by the other two methods but remains in acceptable limits. The companys method largely overestimated overheads for individual products based on a cautious approach, whereas, huge fluctuations can be observed in the estimation errors for the product range using the Jungs model. It is also noticeable that the estimation errors for overheads are greater than those for the overall product costs using all the methods. This refers to the complexities in predicting overheads accurately. The other elements must also be more accurately predicted such that the overall results stay more accurate than those for just overheads. Despite that the overheads predicted by the Hybrid Model stay within reasonable limits unlike those predicted by the other two methods.
207
Figure 7-10: Estimation error trends for overheads using the three methods (2003 2005)
208
Due to the inconsistent results shown by the two methods for the individual products, the cumulative overhead values are also analysed in Figure 7-11. The results are based on comparing the estimated and actual overhead values at the cumulative level using the three methods. The effects of adjustment result in the overall underestimation of overheads predicted by the Jungs model by around 15 percent and those for the companys method result in an overestimation of up to 32 percent. The values for the Hybrid Model varied between -12 and 4 percent. The results clearly demonstrate the superiority of the Hybrid Model over the other two methods for overheads estimation at both individual and cumulative levels of estimation.
Estimation errors for cumulative overhead (%)
40 30 20 10 0 2003 -10 -20 Production Year Estimation error using company's method (%) Estimation error using Jung's model (%) Estimation error using hybrid model (%) 2004 2005
Estimation variation (%)
Figure 7-11: Overheads estimation analysis for the cumulative values
209
The above analysis leads to the quantification of the optimization achieved in the estimation of overheads as a result of using the Hybrid Model in comparison to the other two methods. The results are presented in Figure 7-12 (a) & (b) along with linearization. The optimization is based on considering the deviation values from the actual costs without considering the negative signs (meant for representing underestimation). The differences in the values of error estimation of the two methods against those shown by the Hybrid Model presented in the respective years result in the optimization values. For example, the overhead estimation values predicted by the Hybrid Model in 2004 were almost 29 percent more accurate as compared to those estimated by the companys method. The optimization of around 14 percent was also achieved against the Jungs model estimated values for overheads. For the purpose of statistical analysis, the linearized results are also incorporated and demonstrate linear trends of the optimization for the investigation period.
Every aspect of the validation analysis demonstrated the superiority of the developed Hybrid Model for PCE. Individual products estimated costs and the cumulative costs were analysed and checked for not just accuracy but consistency of the results. The analysis for overhead estimation at products level and cumulative level also demonstrated the developed methods superiority.
210
Overhead estim ation optim ization based on Com apny's m ethod 28.79
(a)
30 24 % Optimization
23.97 y = 5.58x + 10.70
18 12 6 0 2003 2004 Production Year Estimation Optimization (%) Linear estimation optimization trend 2005 12.82
Overhead estim ation optim ization based on Jung's m odel
(b)
15 14.10 11.41 % Optimization 10 y = 4.91x + 0.11
4.28
0 2003 2004 Production Year Estimation Optimization (%) Linear estimation optimization trend 2005
Figure 7-12: Optimization achieved for overhead estimation based on (a) the Companys method (b) the Jungs model
211
7.6
Cost breakdown analysis
One of the advantages of incorporating the breakdown approach in the Hybrid Model is its ability to present a cost breakdown analysis that could lead to the identification of the cost elements requiring better cost control. A cost breakdown analysis is also presented in this study based on the actual costs incurred. All the three methods discussed here consider different breakdown structures. However, since the Hybrid Model is already validated for its accuracy and consistency and found to be superior to the other two models, it will be applied to the actual costs to get the breakdown statistics and for further analysis.
Figure 7-13 shows the production cost (actual) breakdown analysis from 2002 to 2005. Overheads refer to the total indirect costs and are made up of processing costs, materialdependent costs and production overheads. The values are based on the cumulative expenditure in respective years for a given element and are represented in percentages. The breakdown can be considered a typical representation of the UK manufacturing sector based on the analysis of a representative case. Material cost is found to be the highest share (47 percent on the average) of the overall product cost followed by overheads and then labour costs with 37 and 16 percent average values respectively. It can be noted that labour costs in UK are significantly higher than those obtained from the South Asian region. Overhead is a big share of the overall product cost and its accurate estimation is important to the overall accuracy of an entire products cost.
212
Figure 7-13: Production cost (actual) break down analysis (2002 2005) presented in values and percentage
213
An analysis of the three elements based on yearly trends is presented in Figure 7-14. The values represent the cumulative costs incurred on material, labour, and overheads. A gradual yearly rise for the direct elements (material and labour) is not surprising and is due to inflation and other variants. A steeper rise in overheads needs further investigation. However, one explanation could be a less-effective control on overheads resulting in a steeper rise. This is due to the companys adopted cautious approach of overheads estimation resulting in overestimation and thus setting cost control strategies accordingly.
Production Cost Element Trends

1600
Thousands Values (s)
1400
1200
1000
800
600
400
200
0 2002 2003 2004 2005
Production Year
Total Overheads (actual) Actual Labour Cost (Total) Total material cost (Actual)
Figure 7-14: Production cost elements trends
214
Before analysing in detail why overheads rose more steeply, it is necessary to see their effect on the overall production and manufacturing cost. Figure 7-15 (a) gives out the yearly production and manufacturing costs in million . Part (b) considers the effect of elemental costs on these values. It is clear that production and manufacturing costs kept rising relative to a steeper rise in overheads. Although material and labour costs are proportionately higher in the overall product cost, their effect on the overall manufacturing and production costs remained unaffected throughout the duration of the analysis. However, a direct effect of a steep rise in overhead values on production and manufacturing cost necessitates its further analysis.
Material cost is directly associated with design attributes and although cost control aspects consider it, the discussion is beyond the scope of the thesis. Similarly, labour cost control require effective process planning and operations management strategies, the scope of the study (cost estimation) does not cover such discussions. However, an analysis of the overhead values could lead to the identification of the elements with higher cost values in order to keep in place an effective CCS. Figure 7-16 is a breakdown of the actual overheads based on the Hybrid Model application. Production overheads make the largest portion with an average value of 44 percent from 2002 to 2005. Processing cost and MDC make up 31 and 25 percent respectively. The figure also describes the values for the four-year period.
215
Production and Manufacturing Costs (2002 - 2005)

4
(a)
Millions
3.20
3
2.53 2.41
2.90 2.60
Values (s)
2.1 9
2.1 5
1 .89
1 2002 2003 2004 2005
Production Year
Total Manufacturing Cost (Actual)
Total production cost (actual)
Cost elements effect on production & manufacturing costs

3500
(b)
Thousands Values (s)
3000
2500
2000
1500
1000
500
0 2002 2003 2004 2005
Production Year
Total Overheads (actual) Actual Labour Cost (Total) Total production cost (actual) Total material cost (Actual) Total Manufacturing Cost (Actual)
Figure 7-15: (a) Production and manufacturing costs (b) elemental costs effect on production and manufacturing costs
216
Figure 7-16: Overheads break down analysis (2002 2005) presented in values and percentage
217
The analysis of the three sub-elements presented in Figure 7-17 shows a steeper rise in the values for production overheads after 2003. While the processing costs displayed a steady yearly rise for the four year period, MDC were more controlled from 2003 2005.
Overhead Elements Trends

700 600 500 400 300 200 100 0 2002 2003 2004 2005 Production Year Material Depandent Cost Processing Cost Production Overheads
Values (s)
Thousands
Figure 7-17 Overhead elements trends (2002 2005)
A further breakdown of the largest overhead element (production overhead), into its constituent elements is presented in Figure 7-18. General administration costs ranged from 40 to 49 percent of the overall production overheads and mainly remained the largest constituent element. Selling expenses with 35 to 42 percent and computer related costs with 15 to 18 percent of the production overheads were the other two constituents.
218
Figure 7-18: Production overheads breakdown (2002 2005)
An analysis of the cost trends for the constituents presented in Figure 7-19 clearly shows that selling expenses increased rapidly and became the largest constituent in 2005 leaving behind general administration costs. General administration costs remained the largest value constituent until 2004 but its rise remained gradual throughout four-year period. Similar results were noted for computer-related costs. Although, the analysis revealed the
219
selling expenses as a possible area for cost control, a steeper rise in its values in the later half of the trend was partly a result of the business policy decisions to boost sales.
Production Overhead Elements Trends

300 250 200 150 100 50 0 2002 2003 2004 2005 Production Year Computer related cost General administration cost Selling Expenses
Values (s)
Thousands
Figure 7-19: Production overhead elements trends analysis
A similar analysis can be carried out for the constituent elements for processing costs and MDC. For example, the one carried out for processing costs in Figure 7-20 reveals a sharp rise in energy costs. Repair costs were kept under control and went down between 2003 and 2004. Maintenance costs showed a steadier trend. Financing expenses can also be seen to be kept under control. It is clear that energy costs were a major contributor to the processing cots.
220
Thousands
180
Processing Cost Elements Trends

160 140 120
Values (s)
100 80 60 40 20 0 2002 2003 2004 2005
Production Year
Insurance cost Repair Cost Financing Expenses Energy cost Maintenance Cost
Figure 7-20: Processing cost elements trends
An analysis for the MDC constituent elements presented in Figure 7-21 shows that freight and transportation costs could have been better controlled especially between 2002 and 2003 and then 2004 and 2005. Packaging costs between 2003 and 2004 could also be controlled more efficiently. Costs associated with material handling, inspection and scrap were largely kept under control.
221
MDC Elements Trends Thousands Values (s)

90 80 70 60 50 40 30 20 10 0 2002 2003 2004 2005
Production Year
Indirect material Material Inspection Material handling Stores & Inventory control Purchasing Scrap Freight & Tranportation Packaging
Figure 7-21: MDC elements trends
The cost element analysis presented is a result of the Hybrid Models application and can be effectively used to identify high cost elements providing opportunities for better cost control. In that sense, the developed model can not only furnish early cost estimates for an entire product with accuracy and consistency but provide cost control opportunities by identifying high cost elements.
222
7.7
Conclusions
This chapter presented the validation analysis that comprised the comparisons of the actual costs of the given product range against the estimated costs for the same range given by the developed Hybrid Model, the companys method and a published model. The Jungs model was selected as a representative case from the published domain being a proven case that could be implemented within the framework of the selected manufacturing set up. Every level of the analysis performed demonstrated the superiority of the developed model.
The cost estimates for individual products given by the Hybrid Model were found to be more accurate and more consistent. The companys method adopted a cautious approach and as a result overestimated on almost every occasion with as much as 40 percent estimation error on isolated cases. Although the Jungs model predicted slightly better values for an isolated case, the Hybrid Models prediction was not only within the acceptable limit for that case but the overall accuracy was found to be more consistent. The incorporation of the case-based technique into the modelling framework sensitized the model without compromising its robustness. The overall validation analysis spread over a three-year period covering a range of products under varying conditions and huge fluctuations of cost deviations still produced consistent results within acceptable limits of error estimations.
Cumulative cost analysis showed similar results. The companys method overestimated by as much as around 19 percent and the Jungs model underestimated by around 9 percent at
223
the cumulative level. The Hybrid Model remained close to the actual costs even at the cumulative level of prediction. The quantification of the optimization against the two methods showed that a maximum of over 16 and 8 percent optimizations were obtained against the companys and the Jungs methods respectively.
Another aspect of the validation analysis considered analysing the overhead estimation accuracy at product and cumulative levels. On almost every occasion, the estimation error (overheads for individual product) produced by the Hybrid Model was found to be not only lower than those presented by the other two methods but remained in acceptable limits. The cumulative overhead values were also analysed and resulted in the overall underestimation of overheads predicted by the Jungs model by around 15 percent and those for the companys method in an overestimation of up to 32 percent. The values for the Hybrid Model varied between -12 and 4 percent.
A cost breakdown analysis was also presented for a four-year period (2002 2005) based on the actual costs incurred. Material cost was found to be 47 percent; labour costs 37 percent and overheads 16 percent of the overall production cost. A further analysis of the overheads breakdown values was carried out. The cost element analysis presented was a result of the Hybrid Models application and helps to identify high cost elements providing opportunities for better cost control.
224
Chapter 8
Conclusions
Chapter 8 concludes the thesis by summarizing the overall work and contributions. The methodology and approaches developed in this thesis are evaluated by comparing the aims and objectives of the work set at the beginning of the study with the outcome of the research. The chapter describes current and future trends in the field of cost estimation and describes how the developed model conforms to those trends. The chapter also identifies future research avenues within the specific context of the established research.
8.1
Summary
The main aim of the study was to develop a comprehensive methodology for product cost estimation in a batch type manufacturing environment. In order to achieve that, a genuine theoretical development in the area was maintained followed by industrial trials and validation. The overall process of the theoretical development was based on identifying problems that the cost estimators face with the existing methods of PCE. This led to an extensive review of the existing methods for PCE and identification of possible solutions. The objectives were set in light of the possible solutions and were followed up with a careful plan throughout the study. New theories, methods, techniques and mathematical models were developed in line with the objectives. The developed models were validated
225
Chapter 8: Conclusions
through industrial trials due to the practical nature of the overall research area. The overall research work was carefully structured in the thesis.
Chapter 1 described the scope of the research study. To do that, an overview of the research area was provided. It was highlighted that a good pricing system requires the application of cost engineering. The significance of cost estimation on cost control was discussed. The role of cost estimation in the set up of a business enterprise with reference to maximizing its profits was then noted. The chapter outlined and briefly discussed the problems in the area of PCE. The aims and objectives of the research study were set and briefly discussed. Finally, the chapter described the overall structure of the thesis.
Chapter 2 established background information in the area of cost estimation. The concepts of cost and costing were introduced and the implications of cost control on manufacturing systems were discussed. Three kinds of manufacturing systems were discussed: mass, batch and job shop production. The rationale for developing a cost estimation methodology for a batch type manufacturing environment was then presented. The chapter also presented a comprehensive literature review with a focus on the techniques for different applications of cost estimation. One of the application area noted was for generic systems. It was found out that developing a methodology for generic systems could widen the scope of the developed methodology. It was, therefore, decided to develop the methodology for generic systems applications. Since, a batch type manufacturing environment is also a form of generic system, and was already considered a viable option to develop a methodology for; a
226
decision was made to develop a methodology to suit the needs of a batch type environment to satisfy the needs of a generic system also.
Chapter 3 extensively reviewed the literature on manufacturing and product cost estimation and provided a detailed survey along with a critical evaluation of some of the techniques developed in the area. Based on the similarities and the differences of the existing methods, a comprehensive hierarchical classification system was suggested. The developed system along with the key conditions for the application of a methodology from a given category helped in developing a decision support model. The support model was developed with the aim of assisting designers and estimators in selecting a methodology. During the course of the classification system development, a framework for case-based methodology was proposed in order to make an effective use of past product details. The proposed classification system was broadly categorized in qualitative and quantitative techniques. Qualitative techniques were found to be useful in providing early estimates and quantitative techniques were noted for their accuracy. The chapter concluded with the suggestion of developing a methodology for cost estimation based on combining concepts from both qualitative and quantitative techniques for early and accurate estimation of a products cost.
Chapter 4 formed the basis for developing a PCE modelling methodology by establishing an overhead estimation method. In order to fulfil the objectives of developing a method for predicting an entire products cost, a cost breakdown approach was considered. An analysis of the existing breakdown techniques identified overhead as a key element for the overall accuracy of a product cost. Problem identification with the existing method of overhead 227
estimation led to the development of a new overhead estimation methodology based on material and timedependent costs. The proposed methodology was validated in a batch type manufacturing environment in South Asia. The validation analysis was based on comparing the estimated product costs obtained with the application of the proposed methodology and the companys method against the actual product costs. The proposed model for overhead estimation was found to be superior to the companys method of overhead estimation as the former allowed more accurate estimation of product costs. The chapter also revealed manufacturing cost breakdown statistics at the company. The cost breakdown can be considered a representative case for that region characterized by lower wage rates. The chapter finally discussed room for further optimization in the proposed method.
Chapter 5 provided a complete framework for estimating a product cost early and accurately in a batch type manufacturing environment. The overhead method developed in Chapter 4 formed the basis for extending the work. The limitations identified in the already developed method were overcome. In order to make use of the attributes of the qualitative and the quantitative techniques, a hybrid modelling approach was considered. Case-based approach from the qualitative techniques was selected to facilitate an effective use of past product details to allow an early estimation of new products costs. This was achieved by making use of the past product costs obtained at the end of a given year to predict the costs in the following year. Two methods from the quantitative techniques were selected. The use of breakdown approach allowed the estimation of an entire products cost instead of only part or component costs. The activity-based costing method was incorporated to get
228
accurate results. Although, the activity rates and units were modelled, the modelling framework was established on time and materialdependent cost elements giving the Hybrid Model even better accuracy than the conventional ABC systems.
Chapter 6 was based on the industrial implementation and application of the Hybrid Model in a batch manufacturing environment in the UK selected as a representative case of the manufacturing sector in the country. The implementation presented the procedure of using the Hybrid Model. This was achieved by developing a comprehensive Hybrid Model Implementation (HMI) algorithm. The algorithm provided all necessary steps in logical sequence to implement the model. The provisions of primary and secondary data input to the system were accommodated in the algorithm along with inclusions of the Hybrid Models comprehensive set of equations. Equations for cost deviation indices were introduced in order to effectively utilize past data. The algorithm successfully implemented the model and the cost estimate results were obtained. Key benefits of using the algorithm and the Hybrid Model were demonstrated. The cost estimates from the companys method were also obtained. The implementation phase facilitated the generation of cost estimates and is part of the entire validation process. The estimated results were obtained retrospectively from 2003 to 2005 by using the field data from 2001 to 2004.
Chapter 7 was based on a follow up of the implementation phase and compared the results obtained in the preceding chapter in order to validate the Hybrid Model. The Jungs model was also selected as a representative case from the published domain based on its already established validity within its domain and its compatibility with the field data and the 229
selected manufacturing set up. Estimated cost values using the Jungs model were obtained. All the estimated cost values for the given product range were compared against the actual costs for the same range. The estimated costs for that purpose were obtained by the application of the Hybrid Model, the companys own method and the selected published model. Every level of the validation analysis demonstrated the superiority of the developed model. For example, the costs for individual products were more accurately and more consistently estimated using the Hybrid Model. Cumulative production costs were also estimated more accurately and more consistently. The optimization of over 16 and 8 percent were obtained against the companys and the Jungs methods respectively by using the Hybrid Model. The developed model was also checked for its overhead estimation accuracy at product and cumulative production levels. The optimization achieved for overhead estimation was found to be up to 29 percent against the companys method and 14 percent against the Jungs model.
A cost breakdown analysis was also presented for a four-year period (2002 2005) based on the actual costs incurred. The breakdown values revealed can be considered a typical representation of the UK manufacturing. A further analysis of the overhead breakdown values was also carried out based on the Hybrid Models application. The analysis proved helpful in identifying the high cost elements thus providing opportunities for better cost control. The developed model, therefore, forms part of a cost control system development also.
230
8.2
Contributions
In order to evaluate the outcomes from this thesis, the contributions from the overall research work are outlined below.
8.2.1
Development of a technique classification system
Following a comprehensive literature review in the area of PCE, methodologies were categorised into qualitative and quantitative. An extensive analysis of the categorized techniques led to further sub-division of up to four levels and thus resulting in a comprehensive hierarchical classification system. It was found that qualitative techniques mainly focussed on early estimates and quantitative ones delivered accuracy by making use of product design and process planning details. The developed system was found helpful in identifying that an accurate overhead estimation methodology is essential for accurate estimation of a products cost. The proposed system is also helpful in visualizing the nature of an estimation technique and comparing it with the other from the same category or a different one.
8.2.2
Development of a decision support model (DSM)
Following a comprehensive literature review and the developed classification system, key conditions for the implementation of a technique from a given category were identified. This led to the development of a decision support model. The aim of the system is to assist
231
designers and estimators in selecting an estimation methodology to suit the needs of a system and given conditions. This is particularly necessary when several cost estimation techniques may be applied in a given condition. However, due to the availability of a large number of estimation techniques, estimators either find the selection process timeconsuming or erroneous. The developed model overcomes the limitations by providing consistency in the selection process.
8.2.3
Development of time- and material-based overhead estimation methodology
In order to develop a methodology that could predict an entire products cost, breakdown approach was considered. Following the review and the proposed classification system, overhead was identified as an important breakdown element. The significance of accurate estimation of this element led to the identification of any existing methods for its prediction. The ABC system was found to overcome some of the problems but still left few areas for improvement. In light of the identified problems with the existing method of overhead estimation, an improved overhead estimation methodology was proposed based on time and materialdependant cost elements. The industrial validation based on a fouryear retrospective analysis in a batch manufacturing environment in South Asia revealed the superiority of the developed methodology against a representative method from the manufacturing domain.
232
8.2.4
Development of a PCE methodology for batch production
A main objective of the study was to develop a methodology that could predict an entire products cost early and accurately. This was achieved by developing the Hybrid Model for PCE in a batch type environment. The model was based on combining attributes from qualitative and quantitative cost estimation techniques from the proposed classification system. Case-based approach from qualitative techniques provided means of utilizing past cost details in order to furnish estimates for similar new products in future. It allowed the early estimation by eliminating the need for product design details necessary to furnish estimates from scratch. The already developed methods for overhead estimation (TRO and MRO) provided basis for further improvement by incorporating the concepts of improved ABC system features. Overheads were sub-divided and calculated based on improved concepts of activity rates and activity units. In this way, the accuracy attributed to quantitative techniques was achieved by making use of the improved and modified concepts of the ABC system. The use of breakdown approach not only allowed the estimation of the entire product cost, but the representation of the cost into elements and sub-elements.
The developed model can also predict manufacturing cost in addition to the overall product cost. The representation of a product cost into its elements and sub-elements allows the identification of any high cost areas. The identification in return facilitates cost control opportunities. The developed model in that sense goes beyond just furnishing early and
233
accurate cost estimates for a product. It constitutes a key component of a cost control system.
8.2.5
Development of cost deviation indices
The introduction of the case-based framework in the Hybrid Model was facilitated by cost deviation indices. The indices were based on effectively utilizing past two years data to predict future costs. Inflation and other cost deviations were also incorporated in the indices for optimized use of past data and accurate prediction. This also resulted in overcoming the limitations of TRO and MRO methods. Deviation indices are crucial to the overall accuracy of the estimated results from the developed Hybrid Model.
8.2.6
Development of HMI algorithm and industrial implementation
Development of the PCE Hybrid Model could serve no real purpose if it could not be successfully implemented in industries. One of the objectives was to implement the developed model in an industrial environment. This thesis devised a comprehensive procedure for the industrial implementation of the developed PCE Hybrid Model. This resulted in the development of HMI algorithm to facilitate the implementation procedure. The Hybrid Model was subsequently implemented in the selected UK batch manufacturing industry. The Hybrid Model with the help of the HMI algorithm and deviation indices generated the estimation results for the given product range. The HMI algorithm after minor changes can be made more generic for a wider scope of industrial implementation.
234
8.2.7
Comparison and validation
The validity of the developed model could only be established after the comparisons of its generated results against the ones from the already recognized and validated methods. Therefore, the model was validated after a three-year retrospective industrial implementation trial. The cost estimation results that were obtained after a successful industrial implementation were not only compared against an industrially recognized and representative method but against an already established published model. The results from the Hybrid Model were found to be more accurate as the furnished estimated costs were found to be closer to the actual product costs than those predicted by the other two methods. However, the validation analysis went even further ahead by comparing not just the product costs but the cumulative costs and the elemental costs. Every aspect of the comparison analysis validated the Hybrid Model as the costs estimated were not only more accurate but more consistent than those predicted by the other two methods.
8.2.8
By products
The overall study resulted in not just achieving the set objectives but in delivering valuable by-products.
A comprehensive literature review carried out in the field not only presented a critical analysis of the developed techniques with the key advantages and limitations but provided useful work with reference to the applications in the area. Similarly, the developed case-
235
based model serves as a valuable component for not just the Hybrid Model but provides a framework for developing a data retrieval system for other applications of production planning and control system. The developed model makes use of past product details in order to furnish ones for similar new designs. Only the changes in the new designs can be incorporated thereby eliminating the need to furnish details from scratch. Another valuable outcome of the study was the revelation of statistical cost breakdown figures on two different geographical locations that were typical of the regions and reflected contemporary trends. The figures reflect the geographical implications on cost occurrences and are helpful in creating a better understanding for developing an effective CCS.
The overall contributions resulted in theoretical development to the field of PCE. This development was fully backed up by industrial trials and validation analyses. All in all the work presented in the thesis resulted in some pioneering contributions to the field of cost engineering in general and PCE in particular.
8.3
Current Trends and Future Work
Before exploring the possibilities for further research avenues in the already established research domain, it is befitting to analyse how it conforms to the current and future trends in the area.
236
With the advent of computers and the advancement in technology, non-conventional approaches such as knowledge-based techniques and neural network models have been applied effectively utilizing past knowledge to predict the future costs in the early design phases. Current trends in cost estimation exploit the feature technology and a simpler trend is based on estimating the costs by calculating the amount of activities performed to manufacture a product. However, more recent works in the field focus on getting quicker and more accurate results by developing integrated systems combining two or more approaches. For example, a mix of neural network approach and feature technology is an emerging trend. Applying neural networks in CBR [112], cost-tolerance analysis using neural networks [113] and fuzzy activity-based costing [114] are also among some of the new concepts. Yet another area of ongoing research activities combines rule-based, fuzzy logic based, and feature-based methodologies together. An approach that blends some of these techniques could provide more promising results. For example, there is a need to combine the feature technology with the ABC method to study the effects in detail. An approach dealing with the ABC systems and neural networks at the same time may yet be another research area.
In line with the recent and ongoing research trends in the area of PCE, the developed methodology exploits the benefits of the three individual techniques from two branches (qualitative and quantitative) to develop a Hybrid Model that conforms to the current and future trends. The established research, therefore, opens new avenues for exploring further in the field. Following are some of the suggested projects.
237
Validation and optimization analysis of the Hybrid Model for PCE through simulation
The aim is to analyze and investigate thoroughly the behaviour of different parameters of the models under varying conditions and compare the results. This should lead to further optimization of the Hybrid Model.
Development of a data retrieval system based on the case-based methodology
The aim of the proposed project is to develop a matching algorithm to facilitate the adaptation of a past product design with closest similarities to a new design. The developed algorithm will form part of the overall data retrieval system by interfacing it into the already proposed case-based system.
The development of a comprehensive prototype computer-based cost estimation system based on the developed Hybrid Model
The already established research when combined with the elements from the above mentioned proposed projects could lead to the development of a comprehensive prototype computer-based cost estimation system. The optimization analysis through simulation and the development of a data retrieval system based on the proposed case-based approach would form integral elements for the proposed computer-based system. The project should aim at interfacing the system with product design data-base. The aim of such an integrated 238
system would be to generate product cost estimates for new product designs at design conceptualization stages. Such a system would facilitate designers, estimators, planners and managers to make necessary decisions from design changes and price quotes to devising production planning and control strategies. More strategic decisions like make or buy and business resizing etc. would also be possible.
8.4
Concluding remarks
The validation of the Hybrid Model demonstrated through various aspects of the analyses can be attributed to a number of factors. A comprehensive study of the available techniques through a well-structured literature review helped to identify the problem areas. This led to the identification of ways to develop solutions based on an effective utilization of strengths of some of the best known methods and eliminating their shortcomings. The initial methods proposed as part of this thesis that were already found to be more accurate than some of the existing techniques, were further analysed. As a result, a framework for a more comprehensive methodology based on further optimizations was presented. The selfassessment procedure fine-tuned the developed methodology.
The work presented in the thesis resulted in contributions to the field of PCE and forms part of the pioneering work at Kings College London. The encouraging results from the industrial validation analysis open avenues for further exploration in the field. The established work is also an attempt to provide a platform for researchers and practitioners
239
alike to explore further in the field. For example, Professor Frank J Fabozzi of Yale School of Management (USA) and editor of the Journal of Portfolio Management not only cited the published work as excellent but incorporated the article as a basic chapter of his recently published book [115]. Finally, the overall work, therefore, provides contributions to the field of applied engineering and opens up new channels for further explorations.
240
Publications Arising from the PhD Study
The research carried out as part of the current PhD study and covered in the thesis has resulted in the following leading journal publications:
Niazi A, Dai JS, Balabani S, and Seneviratne L. Product cost estimation: Technique classification and methodology review. Transactions of the ASME. Journal of Manufacturing Science and Engineering, May-2006, 128(2), 56375, Publisher: ASME, USA, 2006.
Niazi, A., Dai, J. S., Balabani, S., and Seneviratne, L. D. A new overhead estimation methodology: A case-study in an electrical engineering company. Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture, Jan-2007, 221: 699-710, 2007.
Niazi, A., Dai, J. S., and Balabani, S. PCE Hybrid Model for cost estimation in a batch type manufacturing environment. Transactions of the ASME. Journal of Manufacturing Science and Engineering, 2007, (submitted).
Niazi, A., and Dai, J. S., HMI Algorithm and industrial implementation framework for PCE Hybrid Model. Transactions of the ASME. Journal of Manufacturing Science and Engineering, 2007, (submitted).
241
Publications Arising from the PhD Study
Niazi, A., and Dai, J. S., Methodology comparisons and validation analysis for PCE Hybrid Model. Transactions of the ASME. Journal of Manufacturing Science and Engineering, 2007, (submitted).
242
References
[1]. Clark FD and Lorenzoni AB. Applied Cost Engineering, New York, M. Dekker; 1985.
[2]. Cavalieri S, Maccarrone P and Pinto R. Parametric vs. neural network models for the estimation of production costs: A case study in the automotive industry. International Journal of Production Economics, 91(2): 165-177, 2004.
[3]. Rehman S and Guenov MD. A methodology for modelling manufacturing costs at conceptual design. Computers & Industrial Engineering, 35(3-4): 623-626, 1998.
[4]. Gayretli A and Abdalla HS. A feature-based prototype system for the evaluation and optimisation of manufacturing processes. Computers & Industrial Engineering, 37(12): 481-484, 1999.
[5]. Pahl G and Beitz W. Engineering Design: A Systematic Approach, (2nd edition), Springer; 1999.
[6]. OuYang C and Lin TS. Developing an integrated framework for feature-based early manufacturing cost estimation. International Journal of Advanced Manufacturing Technology, 13(9): 618-629, 1997.
243
References
[7]. Sheldon DF, Perks R, Jackson M, Miles BL, and Holland J. Designing for whole life costs at the concept stage. Journal of Engineering Design, 1(2): 131-45, 1990.
[8]. Hicks BJ, Culley SJ, and Mullineux G. Cost estimation for standard components and systems in the early phases of the design process. Journal of Engineering Design, 13(4): 271-292, 2002.
[9]. Sheldon D, Huang G, and Perks R. Specification and development of cost-estimating database for engineering design. Design for Manufacturability, 52: 91-96, 1993.
[10]. Bode J and Fung RYK. Cost engineering with quality function deployment. Computers & Industrial Engineering, 35(3-4): 587-90, 1998.
[11]. BenArieh D and Li Q. Activity-based cost management for design and development stage. International Journal of Production Economics, 83(2): 169-183, 2003.
[12]. Feng C, Kusiak A, and Huang C. Cost evaluation in design with form features. Computer-Aided Design, 28(11): 879-885, 1996.
[13]. Gupta S, Nau DS, Regli WC, and Zhang G. A methodology for systematic generation and evaluation of alternative operation plans. In: Shah JJ, Mantyla M, and Nau DS, editors. Advances in Feature Based Manufacturing, Amsterdam, Elsevier Publishing Company; 161-184, 1994. 244
References
[14]. Kiritsis D, Neuendorf KP, and Xirouchakis P. Petri net techniques for process planning cost estimation. Advances in Engineering Software, 30(6): 375-387, 1999.
[15]. Aldrich RL. Costing a bill of material/route of thermoformed part using integrated MRPII software. 1995.
[16]. Roztocki N and Needy KL. Integrating activity-based costing and economic value added in manufacturing. Engineering Management Journal, 11(2): 17-22, 1999.
[17]. Park CS and Son YK. Computer-assisted estimating of non-conventional manufacturing costs. Computers in Mechanical Engineering, 6(1): 16-25, 1987.
[18]. Diplaris SC and Sfantsikopoulos MM. Costtolerance function: A new approach for cost optimum machining accuracy. International Journal of Advanced Manufacturing Technology, 16(1): 32-38, 2000.
[19]. Weustink IF, Ten Brinke E, Streppel AH, and Kals HJJ. A generic framework for cost estimation and cost control in product design. Journal of Materials Processing Technology, 103(1): 141-148, 2000.
[20]. Koonce D, Judd R, Sormaz D, and Masel DT. A hierarchical cost estimation tool. Computers in Industry, 50(3): 293-302, 2003/4.
245
References
[21]. Boothroyd G, Dewhurst P, and Knight W. Product Design for Manufacture and Assembly, New York, Marcel Dekker, 1994.
[22]. Daabub AM and Abdalla HS. A computer-based intelligent system for design for assembly. Computers & Industrial Engineering, 37(1-2): 111-115, 1999.
[23]. Dewhurst P and Blum C. Supporting analyses for the economic assessment of diecasting in product design. CIRP Annals, 38(1): 161-164, 1989.
[24]. Luong, L. H. S. and Spedding TA. An integrated system for process planning and cost estimation in hole making. International Journal of Advanced Manufacturing Technology, 10(6): 411-415, 1995.
[25]. Ramirez JC and Touran A. An integrated computer system for estimating welding cost. Cost Engineering, 33(8): 7-14, 1991.
[26]. Taiber JG and Ishii K, editors. Development of an optimization method for determination of process sequences considering prismatic work pieces. Proceedings of International Computers in Engineering Conference (1994) USA. ASME, Vol. 1, pp (271-280), 1994.
246
References
[27]. Schreve K, Schuster HR, and Basson AH. Manufacturing cost estimation during design of fabricated parts. Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture, 213(7): 731-735, 1999.
[28]. Boothroyd G and Radovanovic P. Estimating the cost of machined components during the conceptual design of a product. CIRP Annals, 38(1): 157-160, 1989.
[29]. French MJ. Function costing: A potential aid to designers. Journal of Engineering Design, 1(1): 47-53, 1990.
[30]. Ulrich KT and Eppinger SD. Product Design and Development, (3rd edition), McGraw Hill, 2003.
[31]. Kendall K, Mangin C, and Ortiz E. Discrete event simulation and cost analysis for manufacturing optimisation of an automotive LCM component. Composites Part A: Applied Science and Manufacturing, 29(7): 711-720, 1998.
[32]. Gutowski TG, Henderson R, and Shipp C. Manufacturing costs for advanced composite aerospace parts. SAMPE Journal, 27(3): 37-43, 1991.
[33].Mileham AR, Currie GC, Miles AW, and Bradford DT. A parametric approach to cost estimating at the conceptual stage of design. Journal of Engineering Design, 4(2): 117-125, 1993. 247
References
[34]. Shing ON. Design for manufacture of a cost-based system for moulded parts. Advances in Polymer Technology, 18(1): 33-42, 1999.
[35].Chen Y and Liu J. Cost-effective design for injection moulding. Robotics and Computer-Integrated Manufacturing, 15(1): 1-21, 1999.
[36]. Lee P-and Sullivan WG. Integrating concurrent engineering and activity-based costing in a knowledge-based support system. International Journal of Flexible Automation and Integrated Manufacturing, 6(1-2): 1-28, 1998.
[37]. Ong NS. Manufacturing cost estimation for PCB assembly: An activity-based approach. International Journal of Production Economics, 38(2-3): 159-172, 1995.
[38]. Ong NS and Lim LEN. Activity-based cost-modelling procedures for PCB assembly. International Journal of Advanced Manufacturing Technology, 8(6): 396-406, 1993.
[39]. Arenz DR. Using AI to estimate developmental equipment costs. AACE Transactions, F.2.1-F.2.5, 1991.
[40]. Zhang YF, Fuh JYH, and Chan WT. Feature-based cost estimation for packaging products using neural networks. Computers in Industry, 32(1): 95-113, 1996.
248
References
[41]. Zhang YF and Fuh JYH. A neural network approach for early cost estimation of packaging products. Computers & Industrial Engineering, 34(2): 433-450, 1998.
[42]. Poli C, Escudero J, and Fermandez R. How part design affects injection-moulding tool costs. Machine Design, (101-104), November, 24, 1988.
[43]. Bernet N, Wakeman MD, Bourban PE, and Manson, J. A. E. An integrated cost and consolidation model for commingled yarn based composites. Composites Part A: Applied Science and Manufacturing, 33(4): 495-506, 2002.
[44]. Olofsson K and Edlund A. Manufacturing parameter influences on production cost. International Conference on Advance Composites. SICOMP Technical Report TR 98009, 1998.
[45]. Mayer C, Hartmann A and Neitzel M. Cost-conscious manufacturing of tailored thermoplastic composite intermediates using a double belt press. 18th International SAMPE Europe Conference, Proceedings SAMPE Europe, (339-351), 1997.
[46]. Foley M and Bernardon E. Thermoplastic composite manufacturing cost analysis for the design of cost effective automated system. SAMPE J, 6(4): 67-74, 1990.
249
References
[47]. Walls KO and Crawford RJ. The design for manufacture of continuous fibrereinforced thermoplastic products in primary aircraft structures. Composites Manufacturing, 6(3-4): 245-254, 1995.
[48]. Aderoba A. A generalised cost-estimation model for job shops. International Journal of Production Economics, 53(3): 257-263, 1997.
[49]. Lizhang Z and Burns G. Activity-based costing in non-standard route manufacturing. International Journal of Operations & Production Management, 12(3): 38-60, 1992.
[50]. Ozbayrak M, Akgun M, and Turker AK. Activity-based cost estimation in a push/pull advanced manufacturing system. International Journal of Production Economics, 87(1): 49-65, 2004.
[51]. Takakuwa S. The use of simulation in activity-based costing for flexible manufacturing systems. Proceedings of the 1997 Winter Simulation Conference (1997) USA. (793-800), 1997.
[52]. LaScola Needy K, Billo RE, and Warner RC. A cost model for the evaluation of alternative cellular manufacturing configurations. Computers & Industrial Engineering, 34(1): 119-134, 1998.
250
References
[53]. Wen Hsien T. Activity-based costing model for joint products. Computers & Industrial Engineering, 31(3-4): 725-729, 1996.
[54]. Haedicke J and Feil D. In a DOD environment: Hughes aircraft sets the standard for ABC. Management accounting, 72(8): 29-33, 1991.
[55]. Merz CM and Hardy A. ABC puts accountants on design team at HP. Management accounting, 75(3): 22-27, 1993.
[56]. Miller SH. The view from inside: GMs general auditor looks back. Journal of Accountancy, 177(3): 44-46, 1994.
[57]. Soloway L, J. Government contracting: Using activity-based cost management systems in aerospace and defence companies. Journal of Cost Management, 6(4): 5666, 1993.
[58]. Hobdy T, Thomson J, and Sharman P. Activity-based management at AT&T. Management Accounting, 75(10): 35-39, 1994.
[59]. Porter T, J. and Kehoe J, G. Using activity-based costing and value analysis to take the pain out of downsizing at a naval shipyard. National Productivity Review, 13(1): 115-125, 1994.
251
References
[60]. Ostwald PF. Engineering Cost Estimating (3rd Edition), Prentice Hall, 1991.
[61]. Clark F and Lorenzoni AB. Applied Cost Engineering, (3rd edition), CRC, 1996.
[62]. Brimson JA and National Association of Accountants. Activity Accounting: An Activity-Based Costing Approach, New York, John Wiley and Sons, 1991.
[63]. Shehab EM and Abdalla HS. Manufacturing cost modelling for concurrent product development. Robotics and Computer-Integrated Manufacturing, 17(4): 341-353, 2001.
[64]. Shehab EM and Abdalla HS. A design to cost system for innovative product development. Proceedings of the Institution of Mechanical Engineers, Part B (Journal of Engineering Manufacture), 216(B7): 999-1019, 2002.
[65]. Lihua X and Yunfeng W. Research on rapid cost evaluation based on case-based reasoning. Computer Integrated Manufacturing Systems, 10(12): 1605-1609, 2004.
[66]. Ficko M, Drstvensek I, Brezocnik M, Balic J, and Vaupotic B. Prediction of total manufacturing costs for stamping tool on the basis of CAD-model of finished product. Journal of Materials Processing Technology, 164-165: (1327-1335), 2005.
252
References
[67]. Balarman V and Vattam SS; Sasikumar M, Rao DD, et al. (editors). Finding common ground in case-based systems. Knowledge Based Computer Systems. Proceedings of the International Conference: KBCS'98, (1998) India, (25-37), 1998.
[68]. Kingsman BG and De Souza AA. A knowledge-based decision support system for cost estimation and pricing decisions in versatile manufacturing companies. International Journal of Production Economics, 53(2): 119-139, 1997.
[69]. Shehab E and Abdalla H. An intelligent knowledge-based system for product cost modelling. International Journal of Advanced Manufacturing Technology, 19(1): 4965, 2002.
[70]. Gayretli A and Abdalla HS. An object-oriented constraints-based system for concurrent product development. Robotics and Computer-Integrated Manufacturing, 15(2): 133-144, 1999.
[71]. Ostwald PF. AM Cost Estimator: Cost Estimating Database, (4th edition), Ohio, Penton Publishing, 1998.
[72]. Musgrove JG. Expert conceptual estimator. A multi-media expert system cost estimating tool. New Generation Knowledge Engineering. IAKE '92 Proceedings of Third Annual Symposium of the International Association of Knowledge Engineers (1992) USA, 480-489, 1992. 253
References
[73]. Venkatachalam AR, Mellichamp CM, and Miller DM. A knowledge-based approach to design for manufacturability. Journal of Intelligent Manufacturing, 4(5): 355-366, 1993.
[74]. Waring CW. Product costing automation: The impact of the learning curve. Computers & Industrial Engineering, 21: 313-317, 1991.
[75]. Hundal MS. Design to cost. In: Parsaei HR, Sullivan WG, editors. Concurrent Engineering: Contemporary Issues and Modern Design Tools, (1st edition), pp (330351), Springer, 1993.
[76]. Lewis J. Metrics mapping cuts estimating time. Design News, 55(18): 107-110, 2000.
[77]. McKim RA. Neural network applications to cost engineering. Cost Engineering, 35(7): 31-35, 1993.
[78]. Shtub A and Zimerman Y. A neural-network-based approach for estimating the cost of assembly systems. International Journal of Production Economics, 32(2): 189-207, 1993.
[79]. ManYi C and DingFang C. Early cost estimation of strip-steel coiler using BP neural network. Proceedings of 2002 International Conference on Machine Learning and Cybernetics (2002) USA, 1326-1331, IEEE, 2002. 254
References
[80]. Hajare AD. Parametric costingsteel wire mill. Proceedings of the Annual Convention of the Wire Association International, (172-178), 1998.
[81]. Roberts CA and Hermosillo EP. An automated machining cost estimator. Journal of Engineering Valuation and Cost Analysis, 3(1): 27-42, 2000.
[82]. Boothroyd G and Reynolds C. Approximate cost estimates for typical turned parts. Journal of Manufacturing Systems, 8(3): 185-93, 1989.
[83]. Dewhurst P and Boothroyd G. Early cost estimating in product design. Journal of Manufacturing Systems, 7(3): 183-91, 1988.
[84]. Jong Yun J. Manufacturing cost estimation for machined parts based on manufacturing features. Journal of Intelligent Manufacturing, 13(4): 227-238, 2002.
[85]. Wei Y and Egbelu PJ. A framework for estimating manufacturing cost from geometric design data. International Journal of Computer Integrated Manufacturing, 13(1): 50-63, 2000.
[86]. Son YK. A cost estimation model for advanced manufacturing systems. International Journal of Production Research, 29(3): 441-452, 1991.
255
References
[87]. Singh N. Integrated product and process design: A multi-objective modelling framework. Robotics and Computer-Integrated Manufacturing, 18(2): 157-168, 2002.
[88]. Yeo SH, Ngoi, BKA, Poh LS, and and Hang C. Cost-tolerance relationships for nontraditional machining processes. The International Journal of Advanced Manufacturing Technology, 13(1): 35-41, 1997.
[89]. Sfantsikopoulos MM, Diplaris SC, and Papazoglou PN. Concurrent dimensioning for accuracy and cost. International Journal of Advanced Manufacturing Technology, 10(4): 263-268, 1995.
[90]. Leibl P, Hundal M, and Hoehne G. Cost calculation with a feature-based CAD system using modules for calculation, comparison and forecast. Journal of Engineering Design, 10(1): 93-102, 1999.
[91]. Brimson JA. Feature costing: Beyond ABC. Journal of Cost Management, January/February 1998: (6-12), 1998.
[92]. Wierda LS. Linking design, process planning and cost information by feature-based modelling. Journal of Engineering Design, 2(1): 3-19, 1991.
[93]. Cooper Rand Kaplan RS. How cost accounting distorts product costs. Management Accounting, 69(10): 20-27, 1988. 256
References
[94]. Hundal. Product costing: A comparison of conventional and activity-based costing methods. Journal of Engineering Design, 8(1): 91-103, 1997.
[95]. Andrade MC, Filho RCP, Espozel AM, Maia LOA, and Qassim RY. Activity-based costing for production learning. International Journal of Production Economics, 62(3): 175-180, 1999.
[96]. Beaujon GJ and Singhal VR. Understanding the activity costs in an activity-based cost system. Journal of Cost Management, 4(1): 51-72, 1990.
[97]. Noreen E. Conditions under which activity-based cost systems provide relevant costs. Journal of Management Accounting Research, 3: 159-168, 1991.
[98]. Borjesson S. What kind of activity-based information does your purpose require?: Two case studies. International Journal of Operations & Production Management, 14(12): 79-99, 1994.
[99]. Malik SA and Sullivan WG. Impact of ABC information on product mix and costing decisions. IEEE Transactions on Engineering Management, 42(2): 171-176, 1995.
[100]. Kee R and Schmidt C. A comparative analysis of utilizing activity-based costing and the theory of constraints for making product-mix decisions. International Journal of Production Economics, 63(1): 1-17, 2000. 257
References
[101]. Spedding TA and Sun GQ. Application of discrete event simulation to the activitybased costing of manufacturing systems. International Journal of Production Economics, 58(3): 289-301, 1999.
[102]. Kaplan RS. In defence of activity-based cost management. Management Accounting, 74(5): 58-63, 1992.
[103]. Tornberg K, Jamsen M, and Paranko J. Activity-based costing and process modelling for cost-conscious product design: A case study in a manufacturing company. International Journal of Production Economics, 79(1): 75-82, 2002.
[104]. Tseng YJ and Jiang BC. Evaluating multiple feature-based machining methods using an activity-based cost analysis model. International Journal of Advanced Manufacturing Technology, 16(9): 617-623, 2000.
[105]. Yang YN, Parsaei HR, Hamid R, Leep HR, and Wong JP. A manufacturing cost estimating system using activity-based costing. International Journal of Flexible Automation and Integrated Manufacturing, 6(3-4): 223-243, 1998.
[106]. Ong NS. Activity-based cost tables to support wire harness design. International Journal of Production Economics, 29(3), 271-289, 1993.
258
References
[107]. Karbhari VM and Jones SK. Activity-based costing and management in the composites product realization process. International Journal of Materials and Product Technology, 7(3): 232-244, 1992.
[108]. Koltai T, Lozano S, Guerrero F, and Onieva L. A flexible costing system for flexible manufacturing systems using activity-based costing. International Journal of Production Research, 38(7): 1615-1630, 2000.
[109]. DeGarmo EP, Black JT, and Kohser RA. Materials and Processes in Manufacturing, (8th edition), New York, John Wiley & Sons, 1999.
[110]. Niazi A, Dai JS, Balabani S, and Seneviratne L. Product cost estimation: Technique classification and methodology review. Transactions of the ASME, Journal of Manufacturing Science and Engineering, 128(2), 563-75, 2006.
[111]. Boothroyd G, Dewhurst P, and Knight W. Product Design for Manufacture and Assembly, (2nd edition), New York, Marcel Dekker, 2002.
[112]. Lotfy EA and Mohamed AS. Applying neural networks in case-based reasoning adaptation for cost assessment of steel buildings. International Journal of Computers & Applications, 24(1): 28-38, 2002.
259
References
[113]. ZoneChing L and DarYuan C. Cost-tolerance analysis model based on a neural networks method. International Journal of Production Research, 40(6): 1429-1452, 2002.
[114]. Nachtmann Hand Lascola Needy K. Fuzzy activity-based costing: A methodology for handling uncertainty in activity-based costing systems. Engineering Economist, 46(4): 245-273, 2001.
[115]. Frank J. Fabozzi, Pamela Peterson Drake and Ralph S. Polimeni, The Complete CFO Handbook: From Accounting to Accountability, New Jersey, John Willey & Sons, 2007.
260
Appendix A
Bill of Material (BOM)
The purpose of appendix A is to elaborate BOM. This is carried out by taking the example from a hammer drill. The drill is gradually dismantled and assemblies, sub-assemblies and parts & components are identified. Material and material quantities for different parts are identified. The information along the process is recorded and helps to establish BOM for the drill.
A.1 Introduction
Bill of material (BOM) is a representation of different material with their respective quantities used in a product. Such details are extracted from product design documents. Details contained in design documents help in establishing a product structure based on creating sub-assemblies, assemblies and eventually product levels.
In order to better understand BOM and product structure a hammer drill presented in Figure A-1 is selected and dismantled gradually. The process of dismantling the product identifies assemblies and sub-assemblies and hence results in the overall product structure as presented in Figure A-2. Figure A-3 shows the dismantled drill with the assemblies and sub-assemblies. The other assemblies and sub-assemblies mentioned in the product structure are shown from Figure A-4 to Figure A-7. 261
Appendix A: Bill of Material (BOM)
Figure A-1 Hammer drill
A.2 BOM for Hammer Drill

Product structure is a result of design specification and is a helpful tool in identifying assemblies, sub-assemblies and parts & components. The information when presented in tabular form with parts and assemblies names and their respective material quantities constitute BOM. BOM presents product design details by exploding assemblies and sub-
262
assemblies to part and component level and identifying the required materials with their respective quantities as shown below for the selected hammer drill in Table A-1.
263
Figure A-2 Product structure (Hammer Drill)
264
Figure A-3 Dismantled drill with assemblies and sub-assemblies
265
Figure A-4 Parts and components in product structure
Figure A-5 Winding (Stator and rotor) and drive assembly
266
Figure A-6 Driven assembly (with gear and shock absorber) and Drill/Hammer switch
Figure A-7 Trigger assembly
267
The BOM for the drill at product level comprise parts & components and the drill assembly. The drill assembly can be considered a phantom assembly. Phantom assembly represents a group of components or parts that cannot be assembled together unless some other parts and/or assemblies are added to it. This type of imaginary assembly adds to no operation/processing cost. Respective material and material quantities are also shown in Table A-1 for each component used. BOM explodes the drill assembly into electrical and mechanical assemblies with further explosions later on. Material quantities at the part and component levels add up to sub-assemblies and assemblies level and eventually the overall product level.
Table A-1 Hammer Drill (product level) 1.6 Kg
Component
No of items
Material
Quantity per item (total quantity) gm 10 (10)

40 (40)
Depth gauge Chuck key Chuck Shell Front support handle Chuck screws Body screws Drill Assembly
1 1 1 2 1 2 6 1
Plastic Steel Steel Plastic Thermoplastic Steel Steel -
120 (120) 50 (100) 150 (150) 0.5 (1) 0.5 (3) 1176 (1176)
Drill Assembly (phantom assembly) 1176gm
268
Electrical Assembly Mechanical Assembly
1 1
629 (629) 547 (547)
Electrical Assembly (phantom assembly) 629gm Winding Connections 1 1 325 (325) 304 (304)
Winding (assembly level) 325gm Stator Rotor 1 1 207 (207) 118 (118)
Stator (sub-assembly level) 207gm Shell Stator winding Insulators Pegs Holding strip Cables 1 1 8 4 2 4 Electrical steel Copper Plastic Thermoplastic Steel Copper 87 (87) 52 (52) 3 (24) 2 (8) 12 (24) 3 (12)
Rotor (sub-assembly level) 118gm Shell Core Copper plate Rotor winding 1 1 1 1 Electrical Steel Plastic Copper Copper 28 (28) 12 (12) 15 (15) 63 (63)
Connections (phantom assembly) 304gm 269
Rotor Connection Fuse Trigger Assembly Mains
1 1 1 1
28 (28) 5 (5) 31 (31) 240 (240)
Rotor Connection (phantom assembly) 28gm Brush Assembly Plastic cover Plastic end Wire holder 1 2 2 2 Plastic Plastic Copper 14 (14) 2 (4) 2 (4) 3 (6)
Brush Assembly (sub-assembly level) 14gm Spring Copper Wire Copper plate Brush 2 2 2 2 Steel Copper Copper Coal 2 (4) 1 (2) 1 (2) 3 (6)
The Trigger Assembly (assembly level) 31gm Spring Screw Clamp Connector type I Connector type II 1 4 2 2 2 Steel Steel Steel Copper Copper 1 (1) 0.5 (2) 1.5 (3) 1.5 (3) 1.5 (3)
270
Plastic Mould Trigger Lock assembly
1 1 1
Plastic -
5 (5) 10.5 (10.5) 3.5 (3.5)
Trigger (sub-assembly level) 10.5gm Spring Plastic shell Clamp Trigger frame 2 1 1 1 Steel Plastic Plastic Plastic 1.5 (3) 3 (3) 2 (2) 2.5 (2.5)
Lock assembly (sub-assembly level) 3.5gm Spring Plastic Press 1 1 Steel Plastic Mains 240gm Cable Plug 1 1 Insulated Cu Wire 155 (155) 85 (85) 1.5 (1.5) 2 (2)
Mechanical Assembly (phantom assembly) 547gm Drive Assembly Driven Assembly Drill/Hammer Switch 1 1 1 281 (281) 167 (167) 99 (99)
Driven Assembly (assembly level) 167gm Driven shaft 1 Carbon Steel 55 (55)
271
Driven gear Fastening spring Shock absorber
1 1 1
Carbon steel Steel -
23 (23) 12 (12) 77 (77)
Shock Absorber (sub-assembly level) 77gm Spring Shell 1 1 Steel Alloy steel 15 (15) 62 (62)
Drive Assembly (assembly level) 281gm Driving shaft Fan Journal bearing Ball bearing Gear holder assembly 1 1 1 1 1 Steel Plastic Stainless steel Stainless steel 86 (86) 35 (35) 32 (32) 26 (26) 102 (102)
Gear Holder Assembly (sub-assembly level) 102gm Static gear Gear ring Holding shell 1 1 1 Steel Brass Alloy steel 24 (24) 24 (24) 54 (54)
Drill-Hammer Switch (sub-assembly level) 99gm Steel strip Sphere Plastic Mould 1 1 2 Steel Steel Plastic 35 (35) 38 (38) 13 (26)
272
From MRP point of view, BOMs are maintained at the cumulative level for a product and represent combined quantities of similar materials at the lowest level as shown in Table A-2. These quantities are helpful in procuring the material based on the product manufacturing quantities.
Table A-2 Cumulative material quantities at the lowest level for the hammer drill
S. No. 1 2 3 4 5 6 7 8 9 10 11 12
Material Plastic Thermoplastic Steel Electrical steel Carbon steel Alloy steel Stainless steel Copper Brass Insulated copper wire Coal Screws
Quantity (gm) 229.5 158 406.5 115 78 116 58 158 24 155 6 12 Pcs
273
Appendix B
Deviation Indices
The appendix B is aimed at deriving the equations for cost deviation indices. The indices form part of the PCE Hybrid Model.
B.1 Material cost deviation index

Material cost deviation index value refers to the amount of deviation in the cost of a material from its original cost. For example, material cost deviation index for the pth
n product in the nth year represented by p can be given in terms of material cost for that n n 1 product in the nth year ( Cmp ) and the (n1)th year ( C mp ) as follows:
n p =
n n 1 C mp C mp n 1 C mp
n C mp n 1 C mp
(B-1)
Any deviation in the costs is a result of not just inflation but other factors. If I n is the
n deviation index due to inflation in the nth year and J p represents the index due to other n factors for the pth product in the same year, then p can also be given as follows:
274
Appendix B: Deviation Indices
n n n n p = In + Jp Jp = p In
(B-2)
The above equation (B-2) can be converted in a similar way for the (n+1)th year.
n +1 n +1 p = I n +1 + J p
(B-3)
n +1 Where, p is the material cost deviation index for the pth product in the (n+1)th year, I n+1 n +1 is the deviation index due to inflation in the (n+1)th year and J p represents the index due n +1 to factors other than inflation for the pth product in the (n+1)th year. The value of J p is n based on J p and can be obtained by taking into account I n+1 i.e.
n +1 n Jp = Jp (1 + I n +1 )
(B-4)
n +1 Therefore, the value of J p can be replaced in the equation (B-3) as follows:
n n +1 n p = I n +1 + J p (1 + I n +1 ) ; Now replacing the value of J p from equation (B-2):
n n +1 n p = I n+1 + ( p I n )(1 + I n +1 ) ; and now replacing the value of p from equation (B-1):
275
n +1 p = I n +1 +
n C mp
n 1 mp
1 I n (1 + I n+1 )
(B-5)
Equation (B-5) can be used to determine material cost indices for number of p products in the (n+1)th year using their respective material costs in the nth and the (n1)th years and the inflation values in the nth and the (n+1)th years. Therefore, the known cost data can be used to predict the indices for the product range.
B.2 Labour cost deviation index

Let average per month wages of nonskilled, semiskilled and skilled labour in the nth year be W0n , W1n and W2n respectively and for the (n1)th be W0n 1 , W1n 1 and W2n 1
respectively. If labour cost deviation index for nonskilled, semiskilled and skilled labour
n +1 in the (n+1)th year are represented by 0n +1 , 1n +1 and 2 respectively, equation (B-5) can
be used in a similar way to give these indices as follows:
0n +1 = I n +1 +
W0n n +1 1 I n n 1 (1 + I ) W 0
(B-6)
1n +1 = I n +1 +
W1n 1 I n (1 + I n +1 ) n 1 W1
(B-7)
276
2n +1 = I n +1 +
W2n 1 I n (1 + I n +1 ) n 1 W2
(B-8)
In order to determine the estimated labour rate value for the (n+1)th year, an average labour cost deviation index value ( n +1 ) can be determined.
B.3 Processing cost deviation index

The aggregate shop floor wide processing rate in the nth and the (n1)th years represented as
n n 1 RMA and R MA respectively can be used in a similar way to predict the processing cost
deviation index value for the (n+1)th year ( n +1 ) as follows:
n RMA n n +1 n +1 = I n +1 + R n 1 1 I (1 + I ) MA
(B-9)
B.4 MDC deviation index

n If MDC fraction values ( C md / C mt ) in the nth and the (n1)th years are represented as RMDC n 1 and R MDC respectively, the above method can be used in a similar way to predict the MDC
deviation index value for the (n+1)th year ( n +1 ) as follows:
277
n R MDC n n +1 n +1 = I n +1 + R n 1 1 I (1 + I ) MDC
(B-10)
B.5 Tool cost deviation index

Machine tool cost deviation index value for the (n+1)th year ( n +1 ) can be presented using
n n 1 actual machine tool rates in the nth and the (n1)th years (i.e. RMT and RMT respectively) as
follows:
n RMT n n +1 n +1 = I n +1 + R n 1 1 I (1 + I ) MT
(B-11)
Similarly, labour tool cost deviation index value for the (n+1)th year ( n +1 ) can be given by
n n 1 using the actual labour tool rates in the nth and the (n1)th years (i.e. R LT and R LT
respectively) as follows:
n R LT n n +1 n +1 = I n +1 + R n 1 1 I (1 + I ) LT
(B-12)
278
B.6 Building cost deviation index

Again an index value for building cost deviation in the (n+1)th year ( n +1 ) can be obtained
n n 1 by using the actual building space rates in the nth and the (n1)th years (i.e. RB and R B
respectively) in the following way:
n RB n n +1 n +1 = I n +1 + R n 1 1 I (1 + I ) B
(B-13)
B.7 PO deviation index

If production overhead fraction values ( Ot / C Gt ) in the nth and the (n1)th years are
n n 1 represented as R PO and R PO respectively, the PO deviation index value for the (n+1)th
year ( n+1 ) can be given as follows:
n R PO n n +1 n +1 = I n +1 + R n 1 1 I (1 + I ) PO
(B-14)
279

Product Cost Estimation Theoretical Development and Industrial Validation PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Product Cost Estimation Theoretical Development and Industrial Validation PDF

Uploaded by

Copyright:

Available Formats

Product and Manufacturing Cost Estimation: Theoretical Development and Industrial Validation

My Mum and Dad My Wife

C hapt e r 2 2.1 2.2 2.3 2.4 2.5

2.5.3 2.5.4 2.6

Conclusions................................................................................................. 53 PCE Technique Classification System ................................................. 55

C hapt e r 3 3.1 3.2 3.3

3.4.1 3.4.2 3.5 3.6

3.6.1 3.6.2 3.6.3 3.6.4 3.6.5 3.7

Conclusions................................................................................................. 81 MRO and TRO Estimation Methods .................................................... 87

C hapt e r 4 4.1 4.2

Introduction................................................................................................. 87 Cost estimation methodology at the selected company................................. 91 6

4.2.1 4.2.2 4.2.3 4.3

4.3.1 4.3.2 4.4 4.5

C hapt e r 5 5.1 5.2 5.3

Building space rate........................................................................................ 140 7

C hapt e r 6 6.1 6.2 6.3

6.3.1 6.3.2 6.3.3 6.3.4 6.4

6.4.1 6.4.2 6.4.3 6.4.4 6.4.5 6.5

Conclusions............................................................................................... 179 Comparisons and Validation Analysis................................................ 181

C hapt e r 7 7.1 7.2

7.2.1 7.2.2 7.3 7.4 7.5

C hapt e r 8 8.1 8.2

8.2.1 8.2.2 8.2.3

8.2.4 8.2.5 8.2.6 8.2.7 8.2.8 8.3 8.4

Appendix B B.1 B.2 B.3 B.4

B.5 B.6 B.7

n +1 C Bp n C Bt n +1 C Ep n +1 CGp n C Gt n C Lp n +1 C Lp n C LT n C Lt n +1 C Lt n C MT n +1 CTp n CTt

n +1 RLE n RMA n +1 RME n RMT n +1 RMT

Project Cost Estimation Service Cost Estimation Product Cost Estimation

Applied Cost Engineering Cost Estimation Cost Control Profitability

Figure 1-1: Main area of research

Aims and objectives

Background and Related Work

Chapter 2: Background and Related Work

The amount of money paid or required as a payment to buy a goods or service

Chapter 2: Background and Related Work

Manufacturing and cost control

Figure 2-1: Relationship between productivity and flexibility

Chapter 2: Background and Related Work

Chapter 2: Background and Related Work

Chapter 2: Background and Related Work

Batch Flow Line

Flexible Manufacturing System Manufacturing Cell

Job Shop System

Job Shop Production

Chapter 2: Background and Related Work

deciding whether to produce in-house, purchase, or out-source; 45

Chapter 2: Background and Related Work

manufacturing cost control and hence overall control of production cost.

Chapter 2: Background and Related Work

Cost estimation in the early design stages

Chapter 2: Background and Related Work

Cost estimation for specific applications

Chapter 2: Background and Related Work

Cost estimation for a specific segment in a production cycle

Chapter 2: Background and Related Work

Cost estimation for specific machining and manufacturing processes

Cost forecasting for specific parts and products

Chapter 2: Background and Related Work

Cost estimation for generic systems

Chapter 2: Background and Related Work

Chapter 2: Background and Related Work

Chapter 2: Background and Related Work