Studies in Systems, Decision and Control 88

Krishnendu Mukherjee

Supplier
Selection
An MCDA-Based Approach
Studies in Systems, Decision and Control

Volume 88

Series editor
Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland
e-mail: [email protected]
About this Series

The series “Studies in Systems, Decision and Control” (SSDC) covers both new
developments and advances, as well as the state of the art, in the various areas of
broadly perceived systems, decision making and control- quickly, up to date and
with a high quality. The intent is to cover the theory, applications, and perspectives
on the state of the art and future developments relevant to systems, decision
making, control, complex processes and related areas, as embedded in the fields of
engineering, computer science, physics, economics, social and life sciences, as well
as the paradigms and methodologies behind them. The series contains monographs,
textbooks, lecture notes and edited volumes in systems, decision making and
control spanning the areas of Cyber-Physical Systems, Autonomous Systems,
Sensor Networks, Control Systems, Energy Systems, Automotive Systems, Bio-
logical Systems, Vehicular Networking and Connected Vehicles, Aerospace Sys-
tems, Automation, Manufacturing, Smart Grids, Nonlinear Systems, Power
Systems, Robotics, Social Systems, Economic Systems and other. Of particular
value to both the contributors and the readership are the short publication timeframe
and the world-wide distribution and exposure which enable both a wide and rapid
dissemination of research output.

More information about this series at http://www.springer.com/series/13304

Krishnendu Mukherjee

Supplier Selection
An MCDA-Based Approach

123
Krishnendu Mukherjee
Mechanical Engineering
University of Engineering and Management
Jaipur, Rajasthan
India

ISSN 2198-4182 ISSN 2198-4190 (electronic)

Studies in Systems, Decision and Control
ISBN 978-81-322-3698-6 ISBN 978-81-322-3700-6 (eBook)
DOI 10.1007/978-81-322-3700-6
Library of Congress Control Number: 2016963662

© Springer (India) Pvt. Ltd. 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or
dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer (India) Pvt. Ltd.
The registered company address is: 7th Floor, Vijaya Building, 17 Barakhamba Road, New Delhi
110 001, India
I would like to thank my mentors—Prabir
Kumar Bose, Ardhendu Bhattacharayya, and
Bijon Sarkar—who have taught me the value
of decision making and its effect on human
life. I am also grateful to my parents for their
love and encouragement. Specifically, I am
indebted to my mother for her continuous
support to complete this work.
Finally, this entire work is dedicated to Lord
Krishna and to my mother. Without their
blessing it would be difficult for me to give my
best during my odd days.
Preface

Supplier selection is not a mere clerical process. It is a product-specific

multi-criteria-based optimization problem. Judicious selection of suppliers could
minimize procurement cost, enhance quality and reliability of supply, and thereby
enhance profit margin of the company by minimizing upstream supply chain risk.
Today, supply chain surplus of a company depends considerably on proper selec-
tion of its suppliers. Success of Apple Industry is a good indication of that.
A decision maker has to trade off tangible and intangible criteria to select the best
supplier or suppliers for the focal company. Several methods are available for
selecting the best ones. Among these methods, the analytic hierarchy process
(AHP) and the analytic network process (ANP) are the most used for their com-
putational simplicity and consistency. This book encompasses several criteria and
methods for supplier selection in a systematic way based on extensive literature
review from 1998 to 2012 and includes illustrative case studies and examples for
interested researchers.
The purpose of this book is to present a comprehensive review of the latest
research and development trends at the international level for modeling and opti-
mization of the supplier selection process for different industrial sectors. It is tar-
geted to serve two audiences: the MBA and Ph.D. student interested in procurement
and the practitioner who wishes to gain a deeper understanding of procurement
analysis with multi-criteria-based decision tools to avoid upstream risks to get better
supply chain visibility. This book is expected to serve as a ready reference for
supplier selection criteria and various supplier evaluation methods for forward,
reverse, and mass-customized supply chain. This book also encompasses strategic
sourcing in detail for forward, reverse, and sustainable mass-customized supply
chain.
Chapter 1 introduces the basic concept of multi-criteria decision-making process.
It also introduces fundamental idea and steps of decision-making process. In brief,
Chap. 1 gives fundamental idea of AHP, type of scale, prioritization methods,
problem of rank reversal, validation of AHP, fuzzy hierarchical TOPSIS, VIKOR,
uncertainty analysis with multi-criteria decision analysis tools, cascaded fuzzy
inference system, and intuitionistic fuzzy set. Chapter 2 reviews supplier selection

vii
viii Preface

methods for traditional, reverse, and mass-customized supply chain. Chapter 3

mentions steps to develop mathematical model for supplier selection, evaluation,
and order allocation problem. Chapter 4 discusses in detail about strategic sourcing.
It includes several mathematical models with illustrative case studies. Chapter 5
discusses mainly on the limitations of fuzzy analytic hierarchy process and con-
sistency of decision to aware researchers about certain limitations of fuzzy analytic
hierarchy process. All mathematical models are solved with either MATLAB or R
or LINGO. In the appendix, some computer code written in MATLAB, R, and
VB.NET is also included for the interested reader.
I would like to give sincere thanks to Prof. T.L. Saaty, the originator of AHP and
ANP method, and distinguished University Professor of Business Analytics and
Operations Department of Joseph M. Katz Graduate School of Business, University
of Pittsburgh, USA, for sharing his notes, published and unpublished research work
on AHP and ANP. Without his continuous support and inspiration, it was difficult
for me to know AHP and ANP in detail. My special thanks are due to Prof. Rozann
Whitaker Saaty of the Creative Decisions Foundation of Pittsburgh, USA, for
sharing her research work on AHP and ANP. I wish to thank Professor
Timothy J. Ross of Civil Engineering Department of University of New Mexico;
Professor Ahti A. Salo of Mathematics and Systems Analysis Department of Aalto
University School of Science, Finland; and Dr. Matteo Brunelli, postdoctoral
researcher of Systems Analysis Laboratory of Aalto University, Finland, for their
wholehearted support and great encouragement in producing this book. I wish to
thank various researchers and my students for their support to complete my work.
I am also indebted to editorial team at Springer, for their support, assistance, and
guidance on this book.
While every attempt has been made to ensure that no errors (printing or other-
wise) may be present, the possibility of their existence is expected. As the saying
goes, to err is human. I would be grateful to the readers if these errors are pointed
out to me. Suggestions for further improvement of the book would be thankfully
acknowledged. If any fact, data, figure, concept, etc, resemble with any published
and/or unpublished work, then it is unintentional and I would remain grateful to
readers if they kindly acknowledge it.

Jaipur Krishnendu Mukherjee

Contents

1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Characteristics and Classification of Criteria . . . . . . . . . . . . . . . . . 3
1.3 Classification of Decision Problem . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 MCDA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4.1 Analytic Hierarchy Process (AHP) . . . . . . . . . . . . . . . . . 5
1.4.2 Types of Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4.3 Prioritization Methods—EM or LLSM—Which
One Is Better? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4.4 Rank Reversal in AHP . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4.5 Validation of AHP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4.6 Different Forms of AHP . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.4.7 Application of AHP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4.8 Analytic Network Process (ANP) . . . . . . . . . . . . . . . . . . 12
1.4.9 Technique for Order Preference by Similarity
to Ideal Solution (TOPSIS) . . . . . . . . . . . . . . . . . . . . . . . 13
1.4.10 Fuzzy Hierarchical TOPSIS . . . . . . . . . . . . . . . . . . . . . . . 14
1.4.11 Rank Reversal Problem in TOPSIS . . . . . . . . . . . . . . . . . 15
1.4.12 TOPSIS and Other Methods . . . . . . . . . . . . . . . . . . . . . . 15
1.4.13 Application of TOPSIS . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.4.14 VIKOR (VlseKriterijumska Optimizacija I
Kompromisno Resenje; in Serbian) . . . . . . . . . . . . . . . . . 17
1.5 Uncertainty Analysis with MCDA . . . . . . . . . . . . . . . . . . . . . . . . 17
1.5.1 Fuzzy Set—An Introduction . . . . . . . . . . . . . . . . . . . . . . 18
1.5.2 Cascaded Fuzzy Inference System . . . . . . . . . . . . . . . . . . 20
1.5.3 Intuitionistic Fuzzy Set (IFS)—An Introduction . . . . . . . 22
1.5.4 Dealing Uncertainty with AHP . . . . . . . . . . . . . . . . . . . . 24
1.5.5 Dealing Uncertainty with TOPSIS . . . . . . . . . . . . . . . . . . 24

ix
x Contents

1.5.6 Dealing Uncertainty with VIKOR . . . . . . . . . . . . . . . . . . 24

1.5.7 Fuzzy AHP by Hand Calculation . . . . . . . . . . . . . . . . . . 24
1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2 Modeling and Optimization of Traditional Supplier Selection . . .... 31
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 31
2.2 State-of-the-Art Literature Review of Supplier Selection
Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.3 Pareto Analysis of Supplier Selection Criteria . . . . . . . . . . . . . . . 35
2.4 Stages of Procurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.5 Qualities of Good Supplier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.6 How to Prepare Supply Base? . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.7 Supplier Selection for Mass Customized System . . . . . . . . . . . . . 42
2.8 Hybrid Methods for Supplier Selection . . . . . . . . . . . . . . . . . . . . . 43
2.8.1 Modified Extent Fuzzy AHP and GA
(MEFAHP-GA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.8.2 Fuzzy TOPSIS-MOGA . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.8.3 Multi-Objective Model for Supplier Selection . . . . . . . . . 46
2.8.4 Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3 Mass Customization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.2 Constraints of Mass Customization . . . . . . . . . . . . . . . . . . . . . . . . 60
3.3 Postponement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.4 Sourcing Postponement—A New Kind of Postponement
Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.5 Advantages of Postponement Strategy . . . . . . . . . . . . . . . . . . . . . 62
3.6 Drivers of Postponement Strategy . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.7 Customer Order Decoupling Point (CODP) . . . . . . . . . . . . . . . . . 64
3.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4 Modeling and Optimization of Strategic Sustainable Sourcing . .... 67
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 67
4.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 68
4.2.1 Viability of Dickson’s 23 Criteria for Green
Supply Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.3 Economical Aspects of Reverse Supply Chain . . . . . . . . . . . . . . . 73
4.3.1 Disassembly Cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.3.2 Recycling Profit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.3.3 Optimum Level of Disassembly . . . . . . . . . . . . . . . . . . . 75
Contents xi

4.4 Sustainable Mass Customized System . . . . . . . . . . . . . . . . . .... 75

4.4.1 Stages and Enablers of Sustainable Mass
Customized System . . . . . . . . . . . . . . . . . . . . . . . . . .... 76
4.4.2 Sourcing Strategies for Sustainable Mass
Customized System . . . . . . . . . . . . . . . . . . . . . . . . . .... 76
4.5 Mathematical Model for SPS . . . . . . . . . . . . . . . . . . . . . . . . .... 77
4.6 Decision Support System for Strategic Sustainable Sourcing
in Volume Discount Environment. . . . . . . . . . . . . . . . . . . . . .... 81
4.7 Strategic Sourcing of Large Number of Suppliers: An
Illustrative Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.7.1 Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5 A Note on Limitations of FAHP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.2 Other Limitations of Fuzzy AHP Models . . . . . . . . . . . . . . . . . . . 103
5.3 Consistency Index and Optimization Methods for AHP . . . . . . . . 103
5.3.1 Weighted Least Square Method (Chu et al. 1979) . . . . . . 104
5.3.2 Error Minimization Method
(Chen and Triantaphyllou 2001) . . . . . . . . . . . . . . . . . . . 104
5.3.3 Logarithmic Least Square Method . . . . . . . . . . . . . . . . . . 106
5.3.4 Goal Programming Method (Bryson 1995) . . . . . . . . . . . 106
5.4 Alternative Approaches to FAHP . . . . . . . . . . . . . . . . . . . . . . . . . 106
5.4.1 Method of Triantaphyllou and Lin (1996) . . . . . . . . . . . . 107
5.4.2 Least Square Distance Method (Wang and Parkan
2006; with Kind Permission from Elsevier Limited) . . . . 107
5.4.3 Defuzzification-Based Least Square Method
(Wang and Parkan 2006; with Kind Permission
from Elsevier Limited) . . . . . . . . . . . . . . . . . . . . . . . . . . 108
5.4.4 Preference Programming
(Salo and Hämäläinen 1995) . . . . . . . . . . . . . . . . . . . . . . 108
5.4.5 Fuzzy Preference Programming
(Mikhailov and Singh 2003) . . . . . . . . . . . . . . . . . . . . . . 109
5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Appendix: MCDA Tools and Meta-Heuristic Techniques:
Sample Codes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
About the Author

Krishnendu Mukherjee is an operations research scientist with the Operations

Research Machine Learning & Analytics Experts (ORMAE), where he has been
dealing with several projects for multi-national companies in India and abroad. He
did his Bachelor of Engineering in mechanical from Jadavpur University and
Master of Engineering from BITS Pilani. He also worked as invited reviewer of
IJPE, EJOR, JORS, IJAHP, etc. In 2004, he got opportunity to design and develop
innovative carpet backing machine at IICT, Bhadohi, as one of the core team
member. In 2014, he received US copyright for developing computer code to
facilitate selection and evaluation of n-number of suppliers in fuzzy environment. In
2015, he introduced new concept of postponement strategy to prevent upstream
supply chain risk. He has published 15 papers in international journals/conferences.
His areas of research include multi-criteria decision analysis, fuzzy set theory,
mixed-integer programming, linear programming, constrained programming, non-
linear programming, optimization, supply chain management, sustainability, pro-
curement analysis, vehicle routing, airlines scheduling, and tankers scheduling. He
has also completed his Ph.D. work including the preparation of final draft of Ph.D.
thesis at Jadavpur University. This book encompasses extended work of his Ph.D.
thesis. He previously worked with National Institute of Technology Silchar (former
REC); Ministry of Manpower, Oman; Indian Institute of Carpet Technology,
Ministry of Textile, Govt. of India; Heritage Institute of Technology,
Kolkata; MAHE, Dubai, etc.

xiii
Notations

AHP Analytic hierarchy process

ANP Analytic network process
ATO Assembly to order
CA Cluster analysis
CBR Case-based reasoning
CI Consistency index
CODP Customer order decoupling point
DEA Data envelopment analysis
EM Eigenvalue method
EPAs Environmental Protection Agencies
ETO Engineer to order
FAHP Fuzzy analytic hierarchy process
FANP Fuzzy analytic network process
FNIS Fuzzy negative ideal solution
FPIS Fuzzy positive ideal solution
GA Genetic algorithm
GDM Group decision-making method
GHGs Greenhouse gases
GP Goal programming
GPD Green product design
IFS Intuitionistic fuzzy set
LLSM Logarithmic least square method
MADM Multi-attribute decision making
MAUT Multi-attribute utility theory
MCDA Multi-criteria decision analysis
MCDM Multi-criteria decision making
MEFAHP Modified extent fuzzy AHP
MEFAHP-GA Modified extent fuzzy AHP and GA
MIP Mixed-integer programming
MODM Multi-objective decision making

xv
xvi Notations

MOGA Multi-objective genetic algorithm

MTO Make to order
MTS Make to stock
NIS Negative ideal solution
PFA Product family architecture
PIS Positive ideal solution
SCV Supply chain visibility
TFN Triangular fuzzy number
TOPSIS Technique for order preference by similarity to ideal solution
TVP Total value of purchase
TVRP Total value of reliable purchase
VIKOR VlseKriterijumska Optimizacija I Kompromisno Resenje in
Serbian
List of Figures

Figure 1.1 Comparisons of AHP and ANP . . . . . . . . . . . . . . . . ..... 12

Figure 1.2 Distribution of research papers on hybrid TOPSIS
since 2000 onward . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Figure 1.3 Examples of membership functions . . . . . . . . . . . . . . . . . . 19
Figure 1.4 Triangular fuzzy number. . . . . . . . . . . . . . . . . . . . . . . . . . 19
Figure 1.5 Custom membership function . . . . . . . . . . . . . . . . . . . . . . 20
Figure 1.6 2-D representation of IFS . . . . . . . . . . . . . . . . . . . . . . . . . 23
Figure 1.7 Graphical presentation of fuzzy set and intuitionistic
fuzzy set. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Figure 2.1 Supplier selection methods . . . . . . . . . . . . . . . . . . . . . . . . 33
Figure 2.2 Trend of supplier selection methods . . . . . . . . . . . . . . . . . . 35
Figure 2.3 Distribution of research paper for supplier selection . . . . . . . 35
Figure 2.4 Application of supplier selection methods to different
industries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 38
Figure 2.5 Most cited criteria for supplier selection . . . . . . . . . . ..... 41
Figure 2.6 Intersection of two TFNs (from Zhu et al. 1999;
with kind permission from Elsevier Limited) . . . . . . . . . . . . 44
Figure 2.7 Integrated model of f-TOPSIS-MOGA . . . . . . . . . . . . . . . . 47
Figure 2.8 Cost of producing cement . . . . . . . . . . . . . . . . . . . . . . . . . 48
Figure 2.9 Membership function of linguistic values . . . . . . . . . . . . . . 49
Figure 2.10 Objective function for total cost of purchase (TCP) . . . . . . . 53
Figure 2.11 Result of multi-objective GA. . . . . . . . . . . . . . . . . . . . . . . 53
Figure 3.1 Sourcing postponement. . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Figure 4.1 Pareto analysis of Dickson’s 23 criteria . . . . . . . . . . . . . . . 69
Figure 4.2 Six enablers of 3-stage sustainable mass customized
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 76
Figure 4.3 SVD simulation using four rules at stage 1 . . . . . . . . ..... 82
Figure 4.4 Two odd-shaped membership functions for input
at stage 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 82
Figure 4.5 FIS model for sustainable supplier selection . . . . . . . ..... 83
Figure 4.6 Membership function for stage 1 . . . . . . . . . . . . . . . ..... 83

xvii
xviii List of Figures

Figure 4.7 Membership function for stages 2 and 3 . . . . . . . . . . . . . . . 83

Figure 4.8 Output surface of FIS for quality and price . . . . . . . . . . . . . 84
Figure 4.9 Output surface of FIS for economic strategies 1 and 2 . . . . . 85
Figure 4.10 Rules for overall economic strategy . . . . . . . . . . . . . . . . . . 85
Figure 4.11 Simulink model for economic strategy . . . . . . . . . . . . . . . . 86
Figure 4.12 Rule surface for environmental and social strategy . . . . . . . . 86
Figure 4.13 36 rules for environmental and social strategy . . . . . . . . . . . 87
Figure 4.14 Rules for economic, social, and environmental strategies. . . . 88
Figure 4.15 Simulink model for sustainable supplier selection. . . . . . . . . 89
Figure 4.16 Web-based sustainable supplier selection process . . . . . . . . . 89
Figure 4.17 Pareto front of TCP and TVRP . . . . . . . . . . . . . . . . . . . . . 90
Figure 4.18 d index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Figure 4.19 Hubert index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Figure 4.20 Clusters with their centers . . . . . . . . . . . . . . . . . . . . . . . . . 96
Figure 5.1 Feasible region of decision space . . . . . . . . . . . . . . . . . . . . 109
List of Tables

Table 1.1 Integrated AHP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

Table 1.2 Application of AHP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Table 1.3 Types of decision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Table 1.4 Comparison of different defuzzification methods . . . . . . . . . . . . 21
Table 2.1 Single sourcing versus multiple sourcing . . . . . . . . . . . . . . . . . . 32
Table 2.2 Supplier selection technique—a brief review . . . . . . . . . . . . . . . 34
Table 2.3 Various techniques for single-sourcing and multi-sourcing
supplier selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 34
Table 2.4 A partial list of supplier selection criteria . . . . . . . . . . . . . . . .. 36
Table 2.5 Application area wise distribution of research paper
from 2005 to 2012 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 39
Table 2.6 Fuzzy TFN values (from Tolga et al. 2005; with kind
permission from Elsevier Limited) . . . . . . . . . . . . . . . . . . . . . . . 44
Table 2.7 Priority of supplier selection criteria . . . . . . . . . . . . . . . . . . . . . 48
Table 2.8 TFN values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Table 2.9 Fuzzy evaluation matrix of alternatives . . . . . . . . . . . . . . . . . . . 49
Table 2.10 Weighted evaluation for three suppliers . . . . . . . . . . . . . . . . . . . 50
Table 2.11 Fuzzy TOPSIS result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Table 2.12 Approval status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Table 2.13 Supplier performance data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Table 3.1 Common constraints and solutions of mass customization . . . . . 60
Table 3.2 Categories of postponement strategy . . . . . . . . . . . . . . . . . . . . . 64
Table 3.3 Manufacturing strategy and degree of complexity . . . . . . . . . . . 65
Table 4.1 Literature review of green supplier selection . . . . . . . . . . . . . . . 70
Table 4.2 Critical remarks for Humrhreys model for reversed
supply chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Table 4.3 New criteria for green supplier selection . . . . . . . . . . . . . . . . . . 73
Table 4.4 Linguistic terms in stage 1 and stage 2 . . . . . . . . . . . . . . . . . . . 84
Table 4.5 Linguistic terms in stage 2 and stage 3 . . . . . . . . . . . . . . . . . . . 84
Table 4.6 Volume discount schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
Table 4.7 Supplier datasheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

xix
xx List of Tables

Table 4.8 Selection of optimal number of clusters . . . . . . . . . . . . ....... 93

Table 4.9 Master datasheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ....... 95
Table 4.10 Significance of three clusters . . . . . . . . . . . . . . . . . . . . ....... 96
Table 5.1 Limit of principal eigenvalue (RI is from Saaty
and Tran 2007; with kind permission from Elsevier
Limited) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Table 5.2 Method of Triantaphyllou and Lin (1996) . . . . . . . . . . . . . . . . . 107
Table 5.3 Comparison of AHP and different FAHP methods . . . . . . . . . . 110
Chapter 1
Overview

1.1 Introduction

“People make three general types of judgments to express importance, preference,

or likelihood and use them to choose the best among alternatives in the presence of
environmental, social, political, and other influences. They base these judgments on
knowledge in memory or from analyzing benefits, costs, and risks. From past
knowledge, we sometimes can develop standards of excellence and poorness and
use them to rate the alternatives one at a time”.
Thomas L Saaty
Distinguished Professor of Business Analytics and Operations,
Joseph M. Katz Graduate School of Business,
University of Pittsburgh, USA, (INTERFACES, 1994, Vol. 24, pp. 19–43, with
kind permission from Springer Publishing Company)

The quote, above, addresses three types of decision in addressing large, com-
plex, and most complicated decision-making process of human brain. Every
decision-making process encompasses three key characteristics, namely complex-
ity, credibility, and uncertainty. We do take decision even in complete uncer-
tain situation to bring more credibility and thereby to reduce the complexity. Thus,
judicious decision making is an art and the seed of successful life. The word
‘decision’ is originated from the latin word dēcīsiō, means cutting off. It means to
select the best from available alternatives or converging to a conclusion. Every
decision has the potential to change the world. Thus, every people must learn the art
of decision making. Our success is the yield of our decision. Usually, every
decision encompasses spiritual quotient (SQ), emotional quotient (EQ), and intel-
ligent quotient (IQ). Person-to-person priority of each quotient varies. A spiritual
person usually gives more priority to spiritual quotient, an emotional person usually
gives more priority to emotional quotient, and an intelligent person usually gives
more priority to intelligent quotient. It is expected to have different decisions of two

© Springer (India) Pvt. Ltd. 2017 1

K. Mukherjee, Supplier Selection, Studies in Systems, Decision and Control 88,
DOI 10.1007/978-81-322-3700-6_1
2 1 Overview

persons in same environment because of infinite combinations of IQ, EQ, and SQ.
Thus, every decision is the unique identity of each person and because of that we
receive success or failure. Our decision is also influenced by political, social, and
other influences. Sometimes one wrong decision triggers series of disasters. We
learn from our past and use our past experience to take right decision. In Indian
mythology, Lord Krishna asked Arjuna to kill all his enemies including his relatives
and gurus those who opposed to establish dharma. Every word of Lord Krishna
shows that the right decision at right place is always expected to avoid disaster like
the gruesome war of Kurukshetra. The inherent characteristic of every human being
is to move from less favorable state to more favorable state. This eternal journey
demands right decision and consistent effort. Every good decision of human being
depends on insight, evaluation of fact and external stimuli, imagination, ability to
forecast future effect from present and past data, and language (Saaty and Shih
2009). Our every work demands good decision and priority. Thus, the art of
decision making is truly justified to have better life. Usually, decision-making
process consists of the following eight steps:
1. Problem definition: Define problem with root causes, constraints, or limitations.
2. Need identification: Elaborately explain the need to identify feasible solution
space.
3. Define goals: What objective you are looking for to solve the problem?
4. Select alternatives: Alternatives are the paths to achieve predefined objective.
5. Identify suitable criteria: Which criteria are the best to achieve your goal or
objective?
6. Select appropriate decision-making tool: A less complicated multi-criteria tool
which gives error-free result to any problem is always preferred.
7. Evaluate all alternatives against criteria: Which alternatives are the best to
achieve your goal? This assessment could be objective or subjective or com-
bination of two to measure preference of selection of every alternative with
respect to a criterion to achieve predefined objective.
8. Validate result: Always cross-check the result obtained from any
decision-making process prior to its use.
The term ‘problem’ refers to the dissatisfaction or discontent or irritation per-
ceived from some undesired result or effect of ongoing situation. Every human
being and even insects move from less favorable state to more favorable state. Thus,
making decision to survive in real life is a natural phenomenon. In 1980, a prag-
matic method was proposed by Saaty, popularly known as analytic hierarchy
process (AHP), to give hierarchical interpretation of problem with simple diagram.
In 1974, Altshuller proposed TRIZ (theory of solving inventive problems) to
solve problem in uncertain environment. Usually, our cognition power, pattern
matching ability, associative memory, judicious judgment, ability to compare
results, imagination, logical deduction, etc. help to solve complex problems of real
life. Altshuller standardized the method of problem solving as follows:
1.1 Introduction 3

1. Solve problem step-by-step in a systematical way.

2. Try to converge from broad solution space to achieve feasible optimal solution
exact to or near to an ideal solution.
3. Try to produce reliable solution repeatedly.
4. Understand the body of inventive knowledge and use it judiciously.
5. Interact and discuss with inventors and explain the use of aforesaid method.
Saaty and Shih (2009) proposed the following steps to form hierarchy to solve
problem:
1. Define the goal first.
2. Achieve objective of any problem in series. Highest level should be the goal of
the problem.
3. Insert actors into a suitable level to achieve goal or objective.
4. All alternatives should be placed at bottom level.
5. Verify the hierarchic levels in forward and backward calculations.
Both Saaty and Altshuller proposed modular approach to solve complex problem
even in uncertain environment in a systematic way. In decision making, structure is
the first step to prepare the replica of the problem. It is the logical identification and
grouping of similar things together to identify the flow of influence among them
(Saaty and Shih 2009).

1.2 Characteristics and Classification of Criteria

Criteria are used to measure preference associated with each alternative with
respect to goal. Criteria can also be defined w.r.t one or several preference scales
related to the actual problem. The scales can be classified as continuous scale,
discrete scale, ordinal scale, or cardinal scale (Jacquest Eghem Jr et al. 1989).
Criteria should be non-redundant and relevant to the problem, and no. of criteria
should be as small as possible (Baker et al. 2002). Selection of appropriate criteria
is the cornerstone in decision-making process to select right alternative(s).
Decision-maker’s preference can be explained with the four binary relations
introduced by (Jacquest Eghem Jr et al. 1989):
1. Indifference situation: It is denoted by a I b (i.e., a is indifferent to b). For
example, two alternatives are equally important with respect to some criterion to
a decision-maker.
2. Preference situation: It is denoted by a P b (i.e., a is strictly preferred to b). For
example, one alternative is highly preferred with respect to one criterion to a
decision-maker.
3. Weak preference situation: It is denoted by a Q b (i.e., degree of hesitation is
associated with preference and indifference situation). For example, two
4 1 Overview

alternatives cannot be judged with respect to one criterion because of lack of

certainty.
4. Incomparability situation: It is denoted by a R b (i.e., degree of hesitation is
associated with a P b and b P a). For example, two alternatives cannot be
compared with respect to one criterion because of ambiguity.
There may be other preference relations. Usually, MCDA methods are modeled
with above four binary relations.

1.3 Classification of Decision Problem

According to Roy (1985), problem statements can be classified as follows:

1. Choice problem statement: This type of problem assists the selection of one or
several ‘suitable action(s).’
2. Sorting problem statement: This type of problem assists with group’s various
actions.
3. Rank problem statement: This type of problem orders the actions w.r.t the
predefined preference.

1.4 MCDA

Complex and conflicting real-life problems can be solved in simple but pragmatic
way with multi-criteria decision analysis tools (MCDA). MCDA tools can be
broadly categorized as multi-attribute decision making (MADM) and
multi-objective decision making (MODM). MADM is suitable for finite set of
alternatives. MODM, on the other hand, is suitable for infinite number of alterna-
tives. MCDA methodology can be considered as nonlinear recursive process which
consists of four steps: (1) defining the problem, (2) identifying preference or cri-
teria, (3) aggregating preferences, and (4) ordering alternatives (Guitouni and
Martel 1998).
Opricovic and Tzeng (2004) proposed the following steps of MCDM:
1. Define suitable criteria to achieve goal(s).
2. Identify alternatives for achieving desired goal(s).
3. Evaluate each alternative w.r.t each criterion.
4. Use a suitable multi-criteria analysis tool or techniques.
5. Accept suitable alternative(s) to achieve goal(s).
6. If final solution is not feasible or not acceptable, then opt for next iteration until
feasible solutions are not achieved.
1.4 MCDA 5

Every human being does take decision rationally, irrationally, and

non-rationally. In rational decision-making process, we select the best by evalu-
ating all available alternatives. Often, we give priority based on our knowledge and
past experiences. This approach is known as non-rational decision. We also
express our personal like and dislike. This approach is known as irrational deci-
sion (Guitouni and Martel 1998).
Pairwise comparison is inherent characteristics of human being, and the majority
of MCDA tools prefer to use pairwise comparison matrix to come to consensus
through different methods of normalization and aggregation. MCDA is applicable
to finite number of alternatives and it can be classified as follows:
1. The single-synthesizing criterion approach without incomparability (TOPSIS,
MAVT, MAUT, AHP, EVAMIX, and UTA).
2. The outranking synthesizing approach.
3. The interactive local judgments with trial and error approach.

1.4.1 Analytic Hierarchy Process (AHP)

In AHP, palpable and non-palpable criteria are trade off to prepare a framework
based on ratio scales from simple pairwise comparisons. (Schoner and Wedley
1989 as mentioned in Malcom Beynon 2002).
Analytic hierarchy process can easily be employed with the following steps:
Step 1: Define goal or objective.
Step 2: Define criteria and sub-criteria to accomplish goal.
Step 3: Use Saaty’s nine-point preference scale to form pairwise comparison
matrix. Let B is a n
n pairwise comparison matrix.
2 3
b11 b21 . . . b1n
B ¼ 4 b21 b22 . . . b2n 5 where bij ¼ 1 8i ¼ j
bn1 bn2 . . . bnn

Step 4: Normalize the matrix with geometric mean as follows:

hP i1=n
n
j¼1 bij
ui ¼ 8 i; j ¼ 1; 2; 3 . . . ; n
Pn hPn i1=n
i¼1 j¼1 bij

Step 5: If R denotes n-dimensional column vector describing the sum of

R ¼ ½Ri n
1 ¼ BuT ; ¼ i ¼ 1; 2; . . .n
6 1 Overview

2 3
2 3 R1
1 b12 . . . b1n 6 7
where BuT ¼ 4 b21 1 ... b2n 5½u1 u2 . . . un  ¼ 6 7
4 R2 ... 5
bn1 bn2 . . . 1
Rn

Step 6: Saaty proposed the use of maximum eigenvalue kmax to check consis-
tency of judgment. The maximum eigenvalue kmax can be determined as
follows:
Pn
Rui
kmax ¼ i¼1
8i ¼ 1; 2; 3 . . . n
n

Step 7: Calculate consistency index (CI) as follows:

kmaxn
CI ¼
n1

Step 8: Calculate consistency ratio (CR) to check the validity of

decision-making process.

CR ¼ CIRI, where RI denotes average random index. Saaty proposed that CR

value should be less than or equal to 0.1. Some researchers also mentioned that
random index (RI) can be represented as follows: RI = {1.98(m − 2)/m}, where
m is the size of the matrix.

1.4.2 Types of Scale

In multi-criteria decision analysis, every preference is associated with a suitable

scale to evaluate or rank alternatives to achieve goal. For instance, A is two times
better than B. Here, linguistic term ‘two times better than’ needs a scale to measure
degree of preference. A scale is a triplet, encompasses a set of numbers, a set of
objects, and mapping of objects to the number (Saaty 2004). There are different
types of scale such as
1. Nominal scale: A number is assigned to each object. For example, queue in
front of railway ticket reservation counter.
2. Ordinal scale: Numbers are assigned to each object to represent their order,
increasing or decreasing.
3. Interval scale: For example, Y = B – 30, where Y is a dependent variable and
B is an independent variable.
4. Ratio scale: For example, K = BL, where B > 0, and is a proportional constant.
5. Absolute scale: Number is used directly for pairwise comparison. It is com-
monly used in AHP.
1.4 MCDA 7

Apart from above, there are eight different scales identified by Ishizaka and Labib
(2011)—linear scale (Saaty 1977), power scale (Harker and Vargas 1987), geo-
metric scale (Lootsma 1989), logarithmic scale (Ishizaka et al. 2010), root square
scale (Harker and Vargas 1987), asymptotical scale (Dodd and Donegan 1995),
inverse linear scale (Ma and Zheng 1991), and balanced scale (Salo and
Hamalainen 1997). Usually, Saaty’s scale and geometric scale are used. However,
Saaty’s scale is not like geometric scale which is considered as transitive scale (Dong
et al. 2008). In 1994, Saaty mentioned categorically that the determination of
parameter of geometric scale is difficult. Thus, Saaty’s scale is must for AHP or ANP.

1.4.3 Prioritization Methods—EM or LLSM—Which One

Is Better?

There are several prioritization methods available for multi-criteria decision anal-
ysis (Srdjevic 2005; Choo and Wedley 2004). Some of them are as follows:
1. Eigenvalue method (EM)
2. Logarithmic least square method (LLSM)
Selection of the best prioritization method is an open research issue irrespective
of the most preferred eigenvalue method. Dong et al. (2008) proposed two algo-
rithms to evaluate the performance of four scales and prioritization methods. Saaty
(1990) proposed ten best reasons for using eigenvalue method and some of them are
as follows:
(1) EM yields unique solution.
(2) Simplicity of EM.
(3) EM procedure is descriptive.
(4) EM does not consider statistical indices of bias.

1.4.4 Rank Reversal in AHP

In 1984, Belton and Gear identified rank reversal issues as the major concern for the
application of AHP and further proposed to modify existing normalization method
for classical AHP. Saaty and Vargas (1984) counter reply to the normalization
method of Belton and Gear and declared rank reversal problem as natural phe-
nomenon. Belton and Gear (1982) showed that the addition and deletion of alter-
native can change rank order if new alternative is neither strongly dominated by the
least preferred alternative or strongly dominated by the most preferred alternative
for every criterion. Wang and Elhag (2006) proposed that rank order of the alter-
natives cannot be changed if the priority of the new alternative did not influence the
original local priority of every alternative under each criterion. They further
8 1 Overview

proposed that if B ¼ ðbij Þn
n is the pairwise comparison matrix, then after the
addition of new alternative, comparison matrix becomes B1 = ðbij Þðn þ 1Þxðn þ 1Þ and
the priorities are represented by WB ¼ ½W1B ; W2B ; . . . ; WnB T for the old pairwise
comparison matrix and the new priorities are represented by
WB1 ¼ ½W1B ; W2B ; . . .Wðn þ 1ÞB T . The necessary condition to preserve rank of
alternatives after the addition of new alternative can be expressed as follows:

X
n nX
þ1
WiB ¼ WiB1 ¼ 1
i¼1 i¼1

Extensive literature review shows the following modified methods of AHP to

tackle rank reversal problem:
1. B–G modified AHP, proposed by Belton and Gear (1982).
2. Referenced AHP, proposed by Schoner and Wedley (1989).
3. Linking pin AHP, proposed by Schoner et al. (1997).
4. Multiplicative AHP, proposed by Barzilai and Lootsma (1997).

Example The performance of two suppliers A and B is to be measured w.r.t two

criteria price and quality. Equal priority is given to both criteria. Let us consider B is
preferred to A. If another supplier C is introduced, whose performance is not better
than A, then ranking will remain the same but their global or normalized priority
value will change. If the performance of C is much better than B, then rank of B will
change but rank of A remains the same.

A B
0.46 0.54

CBA

A B C
0.37 0.47 0.17

BCA

A B C
0.13 0.36 0.5
1.4 MCDA 9

1.4.5 Validation of AHP

In multi-criteria decision analysis, decision-makers prepare a mathematical model

which is similar to given problem. Result obtained from the theoretical model does
vary from the actual problem due to the assumptions considered to prepare math-
ematical models. The validity of mathematical model in decision-making process,
thus, is highly required. Choice of numerical scale and method of prioritization
highly influence the validity of decision-making process (Dong et al. 2008).
Qureshi et al. (1999) proposed VVS model which consists of verification, valida-
tion, and sensitivity analysis to compare theoretical MCDA model with actual
problem. They further proposed that to compare result, statistical inference
approaches should be used. Interested reader can refer to work of Whitaker (2007)
in this regard. Every real-life problem encompasses uncertainties. Usually,
researchers prefer to use fuzzy AHP instead of classical AHP to tackle uncertainties.
Classical AHP is a stochastic process, and fuzzy AHP, on the other hand, neither
yields better result nor reduces the complexity of analysis. Several researchers
including Saaty showed that fuzzy AHP has certain limitations. In this book, fuzzy
version of different integrated mathematical models for supplier selection are dis-
cussed and the conflict of choice to opt classical AHP or fuzzy AHP is left to the
reader. Rosenbloom (1996) further enhanced the capability of AHP with the
probabilistic interpretation of final rankings. Usually, designers and manufacturers
prefer to use interval range instead of exact dimension to mention size of product to
cope with inevitable inaccuracy of manufacturing process. The concept of tolerance
in metric space can also be used in decision space. Aguarón et al. (2003) adopted
similar approach to propose consistency stability interval (CSI) to associate an
interval range instead of a crisp value with every judgment of pairwise comparison.
Causes for the validation of model can be divided into two categories:
1. Internal cause: They are the inherent part of every MCDA problem and some of
them are as follows:
1. Misinterpretation of actual problem brings ambiguity, vagueness, etc.
2. Wrong selection of appropriate method yields erroneous result.
3. Misleading judgment may accept the worst alternative instead of the best
alternative.
2. External cause: They are external to the system and some of them are as
follows:
1. Wrong data collection method.
2. Wrong interpretation of interdependencies among criteria and/or alternatives.
Therefore, decision-makers must find controllable and uncontrollable causes to
give optimum result.
10 1 Overview

1.4.6 Different Forms of AHP

Extensive literature shows that researchers are more interested in using hybrid
AHP instead of classical AHP to combine the benefits of more than one method. In
Table 1.1, list of hybrid AHP is shown.

Table 1.1 Integrated AHP

Sl. Author/s Journal/conference Remarks
no. name and year
1 Ali Najmi and Ahmad International Journal of They proposed the integrated
Makui Industrial Engineering use of AHP and DEMATEL to
Computations (2010) tackle interdependencies of
criteria
2 David P. Lilly, John Oxford Business and Principal component analysis
Cory, Bill Hissem Economics Conference was developed by Pearson in
Program (2009) 1901 and has been successfully
adapted to many industries to
create summary information.
The principal component
analysis method creates an
overall measure, which is a
refined dashboard type of
performance measurement using
statistical data compression
techniques. The variable
reduction technique creates
artificial variables and reduces
redundancy in the correlations
3 A.A. Zadeh, H.R. International Journal of They proposed an integrated
Izadbaksh Industrial Engineering approach of PCA and AHP and
(2008) further compared the
performance of their proposed
method with the integrated
approach of DEA and AHP. As
per their observation, the
performance of DEA and AHP
is much inferior to their
proposed method in the context
of plant layout
4 Zixue Guo and Yi IEEE conference They used integrated approach
Zhang (2010) of AHP and PCA
5 Weijun Xia and Omega (2007) They used integrated approach
Zhiming Wu of rough sets theory and AHP to
enhance consistency
6 Ehsan Akhlaghi World Academy of Used fuzzy RST
Science, Engineering
and Technology (2011)
7 Chen-Guang Guo, International Journal of Fuzzy AHP and RST
Yong-Xian Liu, Automation and
(continued)
1.4 MCDA 11

Table 1.1 (continued)

Sl. Author/s Journal/conference Remarks
no. name and year
Shou-Ming Hou and Computing. November
Wei Wang (2010)
8 Te-Sheng Li and Expert Systems with They proposed combined
Hsing-Hsin Huang Applications (2009) approach of fuzzy AHP and
TRIZ for product design. In
1946, Altshuller, a mechanical
engineer of Russian Navy,
developed TRIZ, an acronym
for the theory of inventive
problem solving

1.4.7 Application of AHP

From early 1970s, AHP has been extensively used because of its computational
simplicity, flexibility to be integrated with other MCDA tools, and strong mathe-
matical background. In this regard, three review works since 1979 onward are
identified on the application of AHP, as shown in Table 1.2.

Table 1.2 Application of AHP

Authors Period No. of No. of application areas identified
papers
referred
Jung. P. Shim 1979– 141 31 different areas of application
1988
Omkarprasad S. Vaidya Prior to 154 10 different areas are identified, and each
and Sushil Kumar 1990– area is further sub-divided into 9
2003 sub-areas
Nachiappan 1990– 291 Area related to operations management
Subramanian and 2009 are identified as
Ramakrishnan 1. Operations strategy—manufacturing,
Ramanathan technological, socioeconomical, and
environmental strategies
2. Product and process design—product
planning, forecasting, quality
management, measuring and improving
performance, layout of facilities,
managing capacity
3. Planning and scheduling resources—
resource allocation, job design, and work
measurement
4. Project management—project
estimation, project planning, project
control
5. Managing the supply chain—logistics
and supply chain management,
outsourcing, managing stocks
12 1 Overview

1.4.8 Analytic Network Process (ANP)

Usually, every decision is influenced by its surrounding stimuli. On the other hand,
ANP consists of several clusters and they are connected with each other by their
dependence. Cluster basically groups elements with common set of attributes.
According to Saaty (2004), ANP integrates human intuition and judgment with
reason. In Fig. 1.1, a comparison of AHP and ANP is shown. Saaty broadly
classified dependencies as inner dependencies and outer dependencies. Inner
dependencies may exist among elements of cluster and outer dependencies may
exist between two clusters.
2 3
0 0 0 0
Wn ¼ 4 W21 0 0 05 for hierarchical model:
0 W32 0 I
2 3
0 0 0 0
Wn ¼ 4 W21 W22 0 05 for network model:
0 W32 W33 I

In AHP, W22 and W33 are equal to zero. The steps of ANP are briefly mentioned
below:
Step 1: Define goal or objective of the problem.
Step 2: Identify criteria and sub-criteria.
Step 3: Determine the inner dependencies between factors by pairwise com-
parison using 1–9 scale to calculate priority vectors.
Step 4: Determine the inner as well as outer dependencies between sub-factors
by pairwise comparison using 1–9 scale to calculate priority vectors. For
each comparison, inconsistency index should be less than 0.1 to accept
the decision.
Step 5: Form unweighted super matrix by using priority vectors. In ANP, a
stochastic super matrix is formed through the series of matrix operations.

Fig. 1.1 Comparisons of AHP and ANP

1.4 MCDA 13

Step 6: Form weighted super matrix by dividing each element by the sum of the
corresponding column elements. So that the sum of each column will be
equal to one.
Step 7: Form limit matrix by raising the power of weighted super matrix to an
arbitrary high power. Eigenvalues in stochastic matrix are less than one
(Saaty 2004). The sum of each column in limit matrix will be equal and
the sum of limiting priorities will be equal to the number of criteria.
Step 8: Determine the normalized priority value of each sub-criteria by cluster.
Like AHP, rank reversal problem is also expected in ANP. Leung and Cao
(2001) showed that Sinarchy, a particular form of analytic network process (ANP),
could prevent rank reversal.

1.4.9 Technique for Order Preference by Similarity to Ideal

Solution (TOPSIS)

TOPSIS, a three-point approach, is one of the most cited MCDM approaches to find
solutions from the set of finite alternatives by minimizing the distance from an ideal
point and maximizing the distance from a nadir point (Olson 2004; Opricovic and
Tzeng 2004). However, relative importance of these distances is neglected in
TOPSIS (Opricovic and Tzeng 2004). In 1981, Hwang and Yoon developed
TOPSIS by modifying the method of Zeleny (1974). Hwang and Yoon (1981)
proposed the following six steps to use TOPSIS:
1. Prepare of normalized decision matrix.
2. Prepe of weighted normalized decision matrix.
3. Identify positive ideal solution and negative ideal solution.
4. Determine the separation measure.
5. Determine relative closeness coefficient.
6. Rank the alternative.
Brain storming, nominal group technique (NGT), Delphi technique, etc. can
also be integrated with TOPSIS to enhance the quality of decision (Shih et al.
2007). One of the most important advantages of TOPSIS is its ability to identify the
best alternative quickly (Parkan and Wu 1997). Like other MCDA tools, method of
normalization for TOPSIS can be simplified as linear transformation (Saghafian and
Hejazi 2005; Chen 2000). Extensive literature shows the following two different
ways to use TOPSIS in fuzzy environment:
1. Defuzzification of ratings and weights.
2. Generalized fuzzy TOPSIS.
Usually, later method is considered better as it prevents the loss of information
during defuzzification. Wang and Lee (2007) extended TOPSIS in fuzzy environ-
ment with two parameters, Up and Lo. Interval-valued fuzzy set theory, on the
14 1 Overview

other hand, can provide a more accurate modeling. In this regard, researchers (Chen
and Tsao 2008; Ashtiani et al. 2009; Chu and Lin 2009) used the extension of fuzzy
TOPSIS method with interval-valued fuzzy sets. Wang and Lee (2009) further
modified TOPSIS with an innovative weighting mechanism to avoid the subjec-
tivity of DM’s personal bias. Nezhad and Damghani (2009) used preference ratio to
rank alternatives based on closeness coefficient in TOPSIS. They recommended that
consistency rate between two distance measures gradually reduces with the increase
of alternatives. Shih et al. (2007) used internal aggregation for TOPSIS. Taleizadeh
et al. (2009) integrated fuzzy set theory, TOPSIS, and GA to solve inventory
problem. Tsou (2008) used multi-objective particle swarm optimization (MOPSO)
with TOPSIS to solve inventory problems. Lin et al. (2008) combined AHP and
TOPSIS to map customer requirements with design characteristics of product. Shih
et al. (2007) further compared the performance of TOPSIS and AHP in the presence
of large number of criteria and confirmed that AHP can accommodate only 7 or 9
criteria but TOPSIS can accommodate many more. Several normalization pro-
cesses, such as vector normalization, linear normalization and non-monotonic
normalization, can also be used for TOPSIS along with two distance measures
function—Minkowski’s Lp metrics and weighted Lp metrics (Shih et al. 2007).

1.4.10 Fuzzy Hierarchical TOPSIS

In brief, fuzzy hierarchical TOPSIS has the following steps (Wang et al. 2008):
1. Fuzzy analytic hierarchy process uses a top-down structure to determine the
fuzzy weight of each criterion.
2. TOPSIS prepares a normalized fuzzy performance matrix by using characteristic
of each criterion and finally forms a normalized weighted performance matrix.
3. Calculate FPIS and FNIS, and apply the metric distance method to calculate the
dispersion between alternatives under each criterion.
4. Finally, get a best ranking through an appropriate method of aggregation.
Fuzzy hierarchical TOPSIS can be briefly summarized as follows:
Step 1: Define the evaluation criteria and alternatives of the decision-making
problem to frame a top-down structure.
Step 2: Prepare fuzzy pairwise comparison matrix for each criterion to evaluate
alternatives.
Step 3: Use the Lambda-Max method to calculate the fuzzy weight (FAHP) of
each criterion given by the experts.
Step 4: Check the consistency index (C.I.)
Step 5: Take geometric mean to aggregate all expert opinion.
Step 6: Normalize fuzzy performance matrix.
Step 7: Prepare weighted normalized fuzzy performance matrix.
Step 8: Calculate FPIS and FNIS.
1.4 MCDA 15

Step 9: Measure the distance of each point from FPIS and FNIS.
Step 10: Apply the Euclidean distance method to aggregate all of the criteria for
each alternative.
Step 11: Select the best alternative.

1.4.11 Rank Reversal Problem in TOPSIS

Like AHP, TOPSIS does have rank reversal problems. Cascales and Lamata (2012)
identified the following reasons of rank reversal problem in TOPSIS:
1. Selection of norm.
2. Selection of positive ideal solution (PIS) and negative ideal solution (NIS).
Vector normalization, commonly used in classical TOPSIS, is represented as
follows:
xij
rij ¼ Pm 2
8i ¼ 1; 2; 3; . . . ; m and j ¼ 1; 2; 3; . . . ; n
j¼1 ðxij Þ

Chakraborty and Yeh (2009) mentioned that vector normalization is the most
appropriate to maintain rank consistency and weight sensitivity, respectively.
However, Cascales and Lamata (2012) proposed the modification of the vector
x
normalization with new rij for preserving rank, where rij ¼ maxiji ðxij Þ 8i ¼
1; 2; . . . ; m and rij  1. They further mentioned that modification of norm and
selection method for PIS and NIS is must for preserving rank. For example, after
the addition of new alternative, if the normalized matrix A ¼ ½rij m
n , where
maxðrij Þ ¼ 18i ¼ 1; 2; . . . ; m and j ¼ 1; 2; . . . ; n, then positive ideal solution
(PIS) becomes A þ ¼ ½1; 1; . . . ; 1. Here, PIS remains unchanged.
  However, there

are chances that negative ideal solution (NIS) A ¼ ½min rij  may change. With
the change in NIS, closeness coefficient as well as rank of alternatives changes.

1.4.12 TOPSIS and Other Methods

Today, researchers are more keen to use integrated or hybrid TOPSIS (Behzadian
et al. 2012). Govindan et al. (2012) identified the following advantages of TOPSIS:
1. TOPSIS can evaluate large number of alternatives w.r.t large number of criteria.
2. Changes in one attribute can be compensated by other attributes as TOPSIS
allows explicit balancing and interactions among attributes.
16 1 Overview

3. Instead of giving only rank of each alternative like other MADM tools, TOPSIS
gives preferential ranking of alternatives with a numerical value to give better
understanding of differences and similarities between alternatives.
4. According to the simulation comparison from Zanakis et al. (1998), TOPSIS has
the fewest rank reversals among all MADM tools.
To identify the trend of research on TOPSIS, work of Behzadian et al. (2012) is
further extended to show the distribution of research papers on hybrid methods of
TOPSIS since 2000 in Fig. 1.2.
As shown in Fig. 1.2, about 33.33% of total research work has been allotted to
the integrated use of fuzzy set theory and TOPSIS and about 14.86% of the total
research work on hybrid TOPSIS has been allotted to the integrated approach of
AHP and TOPSIS.

1.4.13 Application of TOPSIS

Behzadian et al. (2012) identified the following application areas of TOPSIS:

1. Supply chain management and logistics
2. Design, engineering, and manufacturing systems
3. Business and marketing management
4. Health, safety, and environment management
5. Human resources management
6. Energy management

Application of TOPSIS with other techniques

160 139
Series1
140
120
100
76
80 62
60
40 20 15 14 14 13
12 8 8 7 7 6 5
20 4 4 3
0

Fig. 1.2 Distribution of research papers on hybrid TOPSIS since 2000 onward
1.4 MCDA 17

1.4.14 VIKOR (VlseKriterijumska Optimizacija I

Kompromisno Resenje; in Serbian)

VlseKriterijumska Optimizacija I Kompromisno Resenje or VIKOR means

multi-criteria optimization and compromise solution (Chu et al. 2007). It ranks
alternatives w.r.t ‘closeness’ to the ‘ideal’ solution (Opricovic 1998, as stated in
Oprocovic and Tzeng 2004). VIKOR is helpful in situation where decision-makers
do not know their preferences at the beginning of the system (Opricovic and Tzeng
2004). In VIKOR, ‘compromise’ means a feasible solution space closest to ideal
solution and compromise means an agreement established by mutual consensus
(Opricovic and Tzeng 2007, as stated in Liu and Wang 2011). There are several
methods that have been developed since 1998 such as
1. Fuzzy VIKOR
2. Extended VIKOR
3. OWA (ordered weighted averaging aggregation)—VIKOR
4. Other MCDA tools are integrated with VIKOR.
Opricovic and Tzeng (2004) compared TOPSIS and VIKOR based on four
criteria—procedural basis, method of normalization, aggregation, and solution.
VIKOR is one of the compromise programming which has received much attention
from different fields, namely manufacturing, design, finance, and health care.

1.5 Uncertainty Analysis with MCDA

Uncertainty refers to the absence of precise information. Often, vagueness and

imprecision are used to refer uncertainty in measurement. Vagueness and impre-
cision differ in representing the form of uncertainty. Vagueness usually relates to
non-measureable or qualitative data. For example, the quality of supply is good.
The term ‘good’ is a linguistic variable to refer the level of satisfaction associated
with the product. For example, the delivery date of supply is 10–15 days. For a
planned horizon, if we plot the delivery date of supply, then we can state the
imprecision in probabilistic terms. Uncertainty can be dealt with several theories
such as fuzzy set, vague set, gray set, intuitionistic fuzzy set, probability theory,
possibility theory, and evidence theory. Each of them is used to deal with different
types of uncertainty. Klir and Yuan identified three different types of uncertainty—
fuzziness, discord, and non-specificity. Fuzzy set represents fuzziness, probability
distribution represents only discord, and classical set represents non-specificity
(Jousselme et al. 2005). Classical MCDA tool uses crisp set during pairwise
comparison of criteria, sub-criteria and alternative with respect to each
sub-criteria/criteria. Uncertainty is associated with assigning a crisp value during
pairwise comparison. Moreover, in real-life problem, every data, be cardinal or
ordinal, is associated with different types of uncertainty. Every human being does
18 1 Overview

make decision under certainty, uncertainty, and risk (Taha 2006, p. 503), and in
each situation, different methods are used, as shown in Table 1.3.
In decision making, uncertainty arises when the consequences of an action are
unknown as it depends on future events. Durbach and Stewart (2012) classified
uncertainty into two categories—internal uncertainty and external uncertainty.
Internal uncertainty refers to uncertainty about decision-maker’s preferences,
problem identification, vagueness, imprecise information, etc. External uncertainty
refers to uncontrollable events of surrounding which could affect the outcome of
any decision.

1.5.1 Fuzzy Set—An Introduction

Concept of set is the crux of mathematics. It is difficult to define set as in mathe-

matics it is not well defined yet. Set can be defined as collection of well-defined
objects. A number a is the member of the set A or not is defined by its charac-
teristic function UA(a) which is mentioned below:

1 where a 2 A
U A ð aÞ ¼
0 where a 62 A

Zadeh proposed that there are infinite possible values between the real contin-
uous intervals [0,1] and introduced the concept of ‘degree of membership.’ The sets
on the universe U that encompasses ‘degree of membership’ are called as fuzzy
set. If u 2 U and AðuÞis degree of membership then fuzzy set A ~ ¼ fu; AðuÞg: The
selection of membership functions is highly subjective and sensitive to the appli-
cation area. Let us consider the price of a product is close to Rs. 6 and it varies from
Rs. 5 to Rs. 7. An expert can select either of the two membership functions to
express above statement, as shown in Fig. 1.3.

Table 1.3 Types of decision

Types of decision Types of payoff matrix Method
Decision making All elements of payoff matrix are well defined Linear
under certainty programming
Decision making Payoff associated with each decision alternatives is 1. Decision
under risk described by probability distribution tree analysis
2. Utility
functions
Decision making All elements of payoff matrix are random in nature Laplace
under uncertainty Minimax
Savage
Hurwicz
1.5 Uncertainty Analysis with MCDA 19

Fig. 1.3 Examples of membership functions

Usually, triangular fuzzy shape is considered for its simplicity and stability
during defuzzification process. A positive TFN ñ can be specified by three
parameters (a,b,c), as shown in Fig. 1.4. The membership function µÃ is defined as
8
>
> xa
0;
< xa
; a xb
l A ¼ ba
>
>
cx
; b  xc
: cb
0; cx
 
a ¼ a1; a2; a3 and ~b ¼ ðb1 ; b2 ; b3 Þ are two triangular fuzzy numbers (TFNs),
If ~
then the distance between them calculated by vertex method is as follows:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1h i
~; ~nÞ ¼
d ðo ða1  b1 Þ2 þ ða2  b2 Þ2 þ ða3  b3 Þ2
3

Actually, defuzzified value varies for all fuzzy membership functions for dif-
ferent methods except TFN, as shown in Table 1.4.

Fig. 1.4 Triangular fuzzy number

20 1 Overview

0.9

0.8

0.7
Membership value

0.6

0.5

0.4

0.3

0.2

0.1

0
0 1 2 3 4 5 6 7 8

Fig. 1.5 Custom membership function

1.5.2 Cascaded Fuzzy Inference System

Knowingly or unknowingly, human being prefers to use linguistic variable to

express his shallow or imprecise knowledge about any fact or objects such as about
to, very small, and fairly. With this natural expression, we conclude or make
decision in uncertain environment. In fuzzy inference system (FIS), such expres-
sions are formed with If-then rule as mentioned below:
If premise, then conclusion or if x is A ~ 1 and A
~ 2 Then y is B
~
The above rule can be expressed as if x is A ~ Then y is B
~
where A~¼A f1 \ fA2 ¼ min½l e ð xÞ; l e ð xÞ
A1 A2
The above premise or antecedent is known as multiple conjunctive antecedent.
~ Then y is B,
Similarly, if x is A ~ where A ~¼f A1 [ f
A2 ¼ max½l e ð xÞ; le ð xÞ
A1 A2
The above antecedent is known as multiple disjunctive antecedent. Fuzzy
inference system can be broadly classified as follows: (1) Mamdani systems,
(2) Sugeno models, and (3) Tsukatomoto models. Mamdani models can be further
classified depending upon the method of aggregation, defuzzification, etc.
Depending upon the types of output function, Sugeno models can be classified as
zero-order Sugeno, first-order Sugeno, etc. For several reasons, Mamdani models
are preferred. In this section, a fuzzy inference system is developed with two-input
and one-output Mamdani model. One of the main drawbacks of fuzzy inference
system using intersection rule configuration (IRC) is the combinatorial explosion
of rules (Ross 2007, p. 75). If N is the number of rules, K is the number of linguistic
variables, and n is the number of input variables, then N = Kn. Thus, the explosion
Table 1.4 Comparison of different defuzzification methods
Defuzzification methods
Types of Centroid Bisector Mean of Smallest of Largest of Remark
fuzzy maximum maximum maximum
member (MOM) (SOM) (LOM)
TFN 3 3 3 3 3 For TFN, defuzzified value remains the same for all methods.
1.5 Uncertainty Analysis with MCDA

(1,3,5) Time taken in defuzzification is as follows:

Time in 0.000432 0.003799 0.002213 0.00243 0.00023 LOM < SOM < CENTROID < MOM < BISECTOR
seconds
TRAPMF 2.5 2 2.5 2 3 For trapezoidal membership functions, defuzzified values
(1,2,3,4) varies. Time taken in defuzzification is as follows:
Time in 0.0005 0.000017 0.0000584 0.000052 0.000041 BISECTOR < LOM < SOM < CENTROID < MOM
seconds
Custom 4.9847 6 6.5 6 7 For custom function shown in Fig. 1.5 gives different
functions defuzzified values for different methods. Time taken in
defuzzification is as follows:
LOM < SOM < CENTROID < MOM < BISECTOR
21
22 1 Overview

of rules can be avoided by reducing K and n. In three different ways, combinatorial

rule explosion can be avoided (Davidrajuh 2008):
1. Dividing the fuzzy inference system into multiple numbers of sub-systems.
Instead of preparing a monolithic FIS with large number of rules, it is advisable
to split into several layers. This approach is known as cascaded fuzzy inference
system.
2. Use of limited number of rules.
3. Use of limited number of inputs.
In cascaded fuzzy inference system, number of rules in each layer or stage can be
calculated with the formula mentioned below (Davidrajuh 2008):
     
mN=n1
n1 þ mn1 =n2
n2 þ    þ mnr1 =nr
nr ð1:1Þ

where N is the number of input, ni is the number of sub-system in ith layer, r is the
number of layer of sub-systems, and m is the number of fuzzy membership func-
tions for each input. Still the performance of cascaded fuzzy inference system can
be improved by reducing the number of rules in fuzzy associative memory (FAM)
by using either singular value decomposition (SVD) or Combs method as cas-
caded fuzzy inference system is the collection of several fuzzy inference systems
in hierarchical form. In cascaded fuzzy inference system, the output from one FIS is
fed to other FIS after processing. Thus, the explosion of rule base is quite expected
if it is developed improperly.

1.5.3 Intuitionistic Fuzzy Set (IFS)—An Introduction

In 1986, Atanassov proposed intuitionistic fuzzy set (IFS), the generalized con-
cept of fuzzy set. If U be a universe of discourse, then IFS A can be defined as
A ¼ fðu; lA ðuÞ; mA ðuÞÞju 2 Ug, where µA(u) and mA(u) denote membership and
non-membership functions of A, respectively, and satisfy 0  lA ðuÞ þ mA ðuÞ  1
8u 2 U: For every IFS A in U, degree of hesitation can be defined as pA ðuÞ ¼
1  lA ðuÞ  mA ðuÞ which expresses whether u belongs to A or not.
Any element u that belongs to IFS A should be inside the triangle MNO, as
shown in Fig. 1.6.
If A ¼ fðu; lA ðuÞ; mA ðuÞÞju 2 Ug and B ¼ fðu; lB ðuÞ; mB ðuÞÞju 2 Ug, then nor-
malized hamming distance between A and B can be represented as lðA; BÞ ¼ 2n 1
Pn
i¼1 ½jlA ðuÞ  lB ðuÞj þ jmA ðuÞ  mB ðuÞj þ jpA ðuÞ  pB ðuÞj (Szmidt and Kacprzyk
2000). To rank three IFS, their normalized hamming distance from the ideal solution M
(1,0,0) should be calculated. Lowest distance from M will give better solution. Every
human decision is associated with a degree of hesitation. If degree of hesitation is zero,
then it is certainty. Essence of intuitionistic fuzzy set is to measure this degree of
hesitation associated with human decision. Szmidt and Kacprzyk (2000) mentioned
1.5 Uncertainty Analysis with MCDA 23

νA(u) N(0,1,0)

π=0

μA(u)

u
O M(1,0,0)

Fig. 1.6 2-D representation of IFS

graphical comparison of fuzzy set and intuitionistic fuzzy set, and similar comparison
was prepared with MATLAB, as shown in Fig. 1.7, to explain generalized IFS.
As shown in Fig. 1.7, an intuitionistic fuzzy set U is mapped into the triangle
ABC, shaded in green color, so that each element of U is corresponding to an
element of ABC. Point P refers to all points belonging to intuitionistic fuzzy set, and
point A refers to all points not belonging to intuitionistic fuzzy set. If orthogonal
projection of ABC is taken on plane, triangle A°B°C°, shaded in blue color, is
obtained.

Fig. 1.7 Graphical presentation of fuzzy set and intuitionistic fuzzy set (Color Online)
24 1 Overview

1.5.4 Dealing Uncertainty with AHP

There are several forms of AHP to deal with uncertainties. Some of them are
mentioned below:
1. Extent fuzzy AHP
2. Modified extent fuzzy AHP
3. Intuitionistic fuzzy AHP (IF-AHP)
4. Fuzzification of AHP by Shannon entropy
5. Gray AHP
6. Rough set theory and AHP

1.5.5 Dealing Uncertainty with TOPSIS

Like AHP, different approaches of TOPSIS are available to deal with uncertainty.
Some of them are as follows:
1. Fuzzy hierarchical TOPSIS
2. Extent fuzzy TOPSIS
3. Fuzzy TOPSIS approach with Shannon entropy

1.5.6 Dealing Uncertainty with VIKOR

Like AHP and TOPSIS, the following approaches of VIKOR are available to deal
with uncertainties. Some of them are as follows:
1. Extent fuzzy VIKOR
2. Fuzzy VIKOR approach with Shannon entropy
3. Gray VIKOR

1.5.7 Fuzzy AHP by Hand Calculation

Let us consider a simple three-criteria (C1, C2, and C3) and three-alternative (A1,
A2, and A3) problem. Fuzzy linguistic values used for pairwise comparison are
shown in the table.
Linguistic variables (from Tolga et al. 2005; with kind permission from Elsevier
Limited).
1.5 Uncertainty Analysis with MCDA 25

Linguistic values Triangular fuzzy numbers (l,m,u)

Equal (1,1,1)
Weak (2/3,1,3/2)
Fairly strong (3/2,2,5/2)
Very strong (5/2,3,7/2)
Absolute (7/2,4,9/2)

Fuzzy pairwise comparison matrix for criteria is as follows:

Criteria C1 C2 C3
C1 Equal Very strong Absolute
(1,1,1) (5/2,3,7/2) (7/2,4,9/2)
C2 —— Equal Fairly strong
(1,1,1) (3/2,2,5/2)
C3 —— —— Equal
(1,1,1)

Complete fuzzy pairwise comparison matrix is as follows:

Criteria C1 C2 C3
C1 (1,1,1) (5/2,3,7/2) (7/2,4,9/2)
C2 (2/7,1/3,2/5) (1,1,1) (3/2,2,5/2)
C3 (2/9,1/4,2/7) (2/5,1/2,2/3) (1,1,1)

A triangular fuzzy number (TFN) contains three components—l,m,u. The above

fuzzy pairwise comparison matrix prepared with the following relations is as
follows:
8
< lij ¼ uji
1
>
for every i 6¼ j mij ¼ m1ji
>
:u ¼ 1
ij lji

Fuzzy synthetic extent value

SC1 (7,8,9)ð1=14:85; 1=13:08; 1=11:41Þ (0.471,0.61,0.789)

SC2 (2.79,3.33,3.9)ð1=14:85; 1=13:08; 1=11:41Þ (0.188,0.255,0.342)
SC3 (1.62,1.75,1.95)ð1=14:85; 1=13:08; 1=11:41Þ (0.109,0.134,0.171)
26 1 Overview

Degree of possibility

V(SC1 > SC2) 1 min{1,1} = 1 Normalized priority of three criteria W = [1,0,0]T

V(SC1 > SC3) 1
V(SC2 > SC1) 0 min{0,1} = 0
V(SC2 > SC3) 1
V(SC3 > SC1) 0 min{0,0} = 0
V(SC3 > SC2) 0

Similarly, normalized priority of alternatives can be calculated as shown above.

The final normalized priority for three alternatives is 0.57, 0.16, and 0.27.

1.6 Conclusion

In this chapter, a brief introduction is given to MCDA. Well-known MCDA tools

such as AHP, ANP, TOPSIS, VIKOR, and their fuzzy versions are discussed in
detail. A detail discussion is given on rank reversal problem of TOPSIS, various
applications of AHP, ANP, TOPSIS, and VIKOR, uncertainty analysis with AHP,
TOPSIS, VIKOR, etc.
Recent trend of research related to application and development of MCDA tools
shows that researchers are more interested for the integrated application of MCDA
tools. However, such application raises questions about the legitimacy of integrated
approach, particularly for the combined application of fuzzy set theory and
AHP. Still integrated model of fuzzy AHP-TOPSIS, fuzzy AHP-VIKOR, and
IF-AHP is discussed with illustrative case study, and choice of selection of
appropriate method is left to the interested readers. Limitations of extent fuzzy AHP
are discussed at the end of Chap. 4. A brief discussion is also included in this
chapter.
Few researchers also showed preference to use cascaded fuzzy inference system
to select suppliers in complex decision-making process to avoid subjectivity of
human assessments during pairwise comparison. Such subjectivity of human
assessments during pairwise comparison can be avoided by considering a range of
value instead of a crisp value. In this regard, author would like to say that the
justification of usefulness of any MCDA tool is a priori of decision-making process.
In Sect. 1.5.1, a brief introduction to fuzzy set theory is given with its certain
limitations. Selection of fuzzy membership functions is still subjective. Triangular
fuzzy membership functions are commonly used for the simplicity of calculation
and stability of defuzzification. Fuzzy set theory is well accepted in different
engineering applications but its application in decision sciences is not unanimously
accepted yet. For example, AHP is a stochastic process. Fuzzy AHP brings more
fuzziness to result and thereby computational complexity with erroneous result. In
1.6 Conclusion 27

this book, several methods are discussed and selection of appropriate method is left
to the reader.
The following useful findings are mentioned in this chapter:
• Steps of decision-making process, classification of criteria, etc. are explained
with examples to give brief introduction to the readers.
• Different scales, normalization methods, etc. are mentioned in this chapter.
• Rank reversal problem is discussed in detail for AHP, ANP, and TOPSIS as it is
one of the debated issues of MCDA.
• Recent trend of MCDA is discussed for AHP, ANP, TOPSIS, and VIKOR.
• Three hybrid algorithms are discussed for AHP, TOPSIS, and VIKOR.
• Several comparative analyses are shown in tabular format in this chapter such as
AHP-TOPSIS and TOPSIS-VIKOR.
• Introduction to cascaded fuzzy inference system is included and the same is
discussed in detail Chap. 4 with an illustrative example.
• Finally, a brief discussion is given to deal with uncertainty with MCDA tools
such as AHP, TOPSIS, and VIKOR.

References

Aguarón J, Esocbar MT, Jiménez JMM (2003) Consistency stability intervals for a judgement in
AHP decision support systems. Eur J Oper Res 145:382–393
Akhlaghi E (2011) A rough-set based approach to design an expert system for personnel selection.
World Academy of Science, Engineering and Technology 78
Ashtiani B, Haghighirad F, Makui A, Montazer GA (2009) Extension of fuzzy TOPSIS method
based on interval-valued fuzzy sets. Appl Soft Comput 9:457–461
Baker D, Bridges D, Hunter R, Johnson G, Krupa J, Murphy J, Sorenson K (2002) Guidebook to
decision-making methods. WSRC-IM-2002-00002, Department of Energy, USA. http://emi-
web.inel.gov/Nissmg/Guidebook_2002.pdf
Barzilai J, Lootsma FA (1997) Power relations and group aggregation in the multiplicative AHP
and SMART. J Multi-Criteria Decis Anal 6:155–165
Behzadian M, Otaghsara SK, Yazdani M, Ignatius J (2012) A state-of the-art survey of TOPSIS
applications. Expert Syst Appl 39:13051–13069
Belton V, Gear T (1982) On a shortcoming of Saaty’s method of analytic hierarchies. Omega 11
(3):226–230
Beynon M (2002) An analysis of distributions of priority values from alternative comparison
scales within AHP. Eur J Oper Res 140:104–117
Cascales MSG, Lamata MT (2012) On rank reversal and TOPSIS method. Math Comput Model
56:123–132
Chakraborty S, Yeh CH (2009) A simulation comparison of normalization procedures for TOPSIS.
IEEE. ISSN: 978-1-4244-4136-5/09
Chen CT (2000) Extensions of the TOPSIS for group decision making under fuzzy environment.
Fuzzy Sets Syst 114:1–9
Chen TY, Tsao CY (2008) The interval-valued fuzzy TOPSIS method and experimental analysis.
Fuzzy Sets Syst 159:1410–1428
Choo EU, Wedley WC (2004) A common framework for deriving preference values from pair
wise comparison matrices. Comput Oper Res 31:893–908
28 1 Overview

Chu TC, Lin YC (2009) An interval arithmetic based fuzzy TOPSIS model. Expert Syst Appl
36:10870–10876
Chu MT, Shyu J, Tzeng GH, Khosla R (2007) Comparison among three analytical methods for
knowledge communities group-decision analysis. Expert Syst Appl 33(4):1011–1024
Davidrajuh R (2008) Building a Fuzzy Logic based Tool for E-readiness measurement. Electron
Gov Int J 5(1):120–130
De Luca A, Termini S (1972) A definition of a non-probabilistic entropy in the setting of fuzzy sets
theory. Inf Control 20:201–312
Dodd F, Donegan H (1995) Comparison of prioritization techniques using inter hierarchy
mappings. J Oper Res Soc 46:492–498
Dong Y, Xu Y, Li H, Dai M (2008) A comparative study of the numerical scales and the
prioritization methods in AHP. Eur J Oper Res 186:229–242
Durbach IN, Stewart TJ (2012) Modeling uncertainty in multi-criteria decision analysis. Eur J
Oper Res 223:1–14
Govindan K, Khodaverdi R, Jafarian A (2012) A fuzzy multi criteria approach for measuring
sustainability performance of a supplier based on triple bottom line approach. J Clean Prod.
doi:10.1016/j.jclepro.2012.04.014
Guitouni A, Martel JM (1998) Tentative guidelines to help choosing an appropriate MCDA
method. Eur J Oper Res 109:501–521
Guo Z, Zhang Y (2010) The third-party logistics performance evaluation based on the AHP-PCA
model. IEEE. ISBN:978-1-4244-7161-4/10
Guo CG, Liu YX, Hou SM, Wang W (2010) Innovative product design based on
customer requirement weight calculation model. Int J Autom Comput 7(4):578–583.
doi:10.1007/s11633-010-0543-3
Harker P, Vargas L (1987) The theory of ratio scale estimation: Saaty’s analytic hierarchy process.
Manage Sci 33:1383–1403
Hwang CL, Yoon K (1981) Multiple attribute decision making. Springer-Verlag, Berlin
Ishizaka A, Labib A (2011) Review of the main developments in the analytic hierarchy process.
Expert Syst Appl 38:14336–14345
Ishizaka A, Balkenborg D, Kaplan T (2010) Influence of aggregation and measurement scale on
ranking a compromise alternative in AHP. J Oper Res Soc 62:700–710
May EC, Spottiswoode SJP, James, CL (1994) Shannon entropy: a possible intrinsic target
property. J Parapsychology 58
Jousselme A-L, Liu C, Grnecier D, Bossé Ẻ (2005) Measuring ambiguity in the evidence theory.
IEEE Trans Syst Man Cybern Part A Syst Hum. doi:10.1109/TSMCA.2005.853483
Jr JT, Delhaye C, Kunsch PL (1989) An interactive decision support system (IDSS) for
multicriteria decision aid. Math Comput Model 12(10/11):131l–1320
Leung LC, Cao D (2001) On the efficacy of modeling multi-attribute decision problems using
AHP and Sinarchy. Eur J Oper Res 132:39–49
Li TS, Huang HH (2009) Applying TRIZ and Fuzzy AHP to develop innovative design for
automated manufacturing systems. Expert Syst Appl 36:8302–8312
Lilly DP, Cory J, Hissem B (2009) The use of principal component analysis to integrate blasting
into the mining process. In: Proceedings of 2009 Oxford Business & Economics Conference
Program. ISBN: 978-09742114-1-9
Lin MC, Wang CC, Chen MS, Chang CA (2008) Using AHP and TOPSIS approaches in
customer-driven product design process. Comput Ind 59:17–31
Liu P, Wang M (2011) An extended VIKOR method for multiple attribute group decision making
based on generalized interval-valued trapezoidal fuzzy numbers. Sci Res Essays 6(4):766–776
Lootsma F (1989) Conflict resolution via pair-wise comparison of concessions. Eur J Oper Res
40:109–116
Ma D, Zheng X (1991) 9/9-9/1 Scale method of AHP. In: Proceedings of Second International
Symposium on AHP, Pittsburgh
Najmi A, Makui A (2010) Providing hierarchical approach for measuring supply chain
performance using AHP and DEMATEL methodologies. Int J Ind Eng Comput 1:199–212
References 29

Nezhad SS, Damghani KK (2009) Application of a fuzzy TOPSIS method base on modified
preference ratio and fuzzy distance measurement in assessment of traffic police centers
performance. Appl Soft Comput. doi:10.1016/j.asoc.2009.08.036
Olson DL (2004) Comparison of weights in TOPSIS Models. Math Comput Model 40:721–727
Opricovic S (1998) Multicriteria optimization of civil engineering systems. Faculty of Civil
Engineering, Belgrade
Opricovic S, Tzeng GH (2004) Compromise solution by MCDM methods: a comparative analysis
of VIKOR and TOPSIS. Eur J Oper Res 156:445–455
Opricovic S, Tzeng GH (2007) Extended VIKOR method in comparison with outranking methods.
Eur J Oper Res 178(2):514–529
Parkan C, Wu ML (1997) On the equivalence of operational performance measurement and
multiple attribute decision making. Int J Prod Res 35(11):2963–2988
Qureshi ME, Harrison SR, Wegener MK (1999) Validation of multicriteria analysis models. Agric
Syst 62:105–116
Rosenbloom ES (1996) A probabilistic interpretation of the final rankings in AHP. Eur J Oper Res
96:371–378
Ross TJ (2007) Fuzzy logic with engineering applications. Wiley India Edition
Roy B (1985) Mèthodologie Multicritère D’Aide à la Dècision. Ćollection Gestion—Edition
Economica, Paris
Saaty T (1977) A scaling method for priorities in hierarchical structures’. J Math Psychol 15
(3):234–281
Saaty TL (1990) Eigenvector and logarithmic least squares. Eur J Oper Res 48:156–160
Saaty TL (1994) Highlights and critical points in the theory and application of the analytic
hierarchy process. Eur J Oper Res 74:426–447
Saaty TL (2004) Decision making—the analytic hierarchy and network process (AHP/ANP).
J Syst Sci Syst Eng 13(1):1–34
Saaty TL, Shih HS (2009) Structures in decision making: on the subjective geometry of hierarchies
and networks. Eur J Oper Res 199:867–872
Saaty TL, Vargas LG (1984) The legitimacy of rank reversal. OMEGA Int J Manag Sci 12(5):513–
516
Saghafian S, Hejazi SR (2005) Multi-criteria group decision making using a modified fuzzy
TOPSIS procedure. In: Proceeding 2005 International Conference on Computational
Intelligence for Modelling, Control and Automation, and International Conference on
Intelligent Agents, Web Technologies and Internet Commerce (CIMCA-IAWTIC’05). ISSN
0-7695-2504-0/05
Salo A, Hamalainen R (1997) On the measurement of preference in the analytic hierarchy process.
J Multi-Criteria Decis Anal 6:309–319
Schoner B, Wedley WC (1989) Ambiguous criteria weights in AHP: consequences and solutions.
Decis Sci 20:462–475
Schoner B, Choo EU, Wedley WC (1997) A comment on ‘rank disagreement: a comparison of
multi-criteria methodologies’. J Multi-Criteria Decis Anal 6:197–200
Shih HS, Shyur HJ, Lee ES (2007) An extension of TOPSIS for group decision making. Math
Comput Model 45:801–813
Shim JP (1989) Bibliographical research on the analytic hierarchy process (AHP). Socio-Econ
Plann Sci 23(3):161–167
Srdjevic B (2005) Combining different prioritization methods in the analytic hierarchy process
synthesis. Comput Oper Res 32:1897–1919
Subramanian N, Ramanathan R (2012) A review of applications of analytic hierarchy process in
operations management. Int J Prod Econ 138:215–241
Szmidt E, Kacprzyk J ( 2000) Distance between intuitionistic fuzzy sets. Fuzzy Sets System 114
(3):505–518
Taha HA (2006) Operations research—an introduction. Prentice-Hall of India Pvt, Ltd
30 1 Overview

Taleizadeh AA, Niaki STA, Aryanezhad MB (2009) A hybrid method of Pareto, TOPSIS and
genetic algorithm to optimize multi-product multi-constraint inventory control systems with
random fuzzy replenishments. Math Comput Model 49:1044–1057
Tolga E, Demircan ML, Kahraman C (2005) Operating system selection using fuzzy replacement
analysis and analytic hierarchy process. Int J Prod Econ 97:89–117
Tsou CS (2008) Multi-objective inventory planning using MOPSO and TOPSIS. Expert Syst Appl
35:136–142
Vaidya OS, Kumar S (2006) Analytic hierarchy process: An overview of applications. Eur J Oper
Res 169:1–29
Wang YM, Elhag TMS (2006) An approach to avoiding rank reversal in AHP. Decis Support Syst
42:1474–1480
Wang YJ, Lee HS (2007) Generalizing TOPSIS for fuzzy multiple-criteria group decision-making.
Comput Math Appl 53:1762–1772
Wang TC, Lee HD (2009) Developing a fuzzy TOPSIS approach based on subjective weights and
objective weights. Expert Syst Appl 36:8980–8985
Wang JW, Cheng CH, Cheng HK (2008) Fuzzy hierarchical TOPSIS for supplier selection. Appl
Soft Comput. doi:10.1016/j.asoc.2008.04.014
Whitaker R (2007) Validation examples of the analytic hierarchy process and analytic network
process. Math Comput Model 46:840–859
Xia W, Wu Z (2007) Supplier selection with multiple criteria in volume discount Environments.
Omega 35:494–504
Yager RR (1979) On the measure of fuzziness and negation Part 1: membership in unit interval.
Int J Gen Syst 5:21–229
Zadeh AA, Izadbaksh HR (2008) A multi-variate/ multi-attribute approach for plant layout design.
Int J Ind Eng 15(2):143–154
Zanakis SH, Solomon A, Wishart N, Dublish S (1998) Multi-attribute decision making: a
simulation comparison of selection methods. Eur J Oper Res 107:507–529
Zeleny M (1974) A concept of compromise solutions and the method of the displaced ideal.
Comput Oper Res 1:479–496
Chapter 2
Modeling and Optimization of Traditional
Supplier Selection

2.1 Introduction

Judicious selection of supplier can mitigate upstream supply chain risk by sup-
plying right quantity at right place and time.
Supplier selection is a strategic process as it can mitigate upstream supply chain risk
partially, if not completely. Better supplier–buyer dyadic relationship can enhance
supply chain visibility and capability to cope with high demand volatility. Supplier
selection, thus, is an indispensible part of any business. Any disruption in upstream
supply may cause tremendous disaster in entire supply chain and compel organi-
zation to take risk. Risks in supply chain are broadly classified as internal risk that
appears in normal operation and external risk that come from outside the supply.
Selection of right supplier(s) could minimize external risks. Supplier selection could
be either single sourcing or multiple sourcing. In single sourcing, entire supply
comes from one supplier. In multiple sourcing, on the other hand, entire supply
comes from a group of suppliers. Risk in supply chain could be minimized by
internal integration and external integration of supply chain entities. External
integration strongly encourages single sourcing by strengthening buyer–supplier
relationship. Table 2.1 shows the comparative analysis of single-sourcing and
multiple-sourcing strategies.
Research on supplier selection methods has rich collection, as shown in Fig. 2.1.
Some researchers combined at least two techniques for supplier selection, for
instance, AHP-GP, AHP-LP, DEA-AHP, and DEA-MOP. Supplier selection
problem involves vague and imprecise assessments, which are by nature fuzzy.
Thus, a group of researchers used fuzzy AHP. Various methods have been used to
derive priority vectors from fuzzy pairwise comparison. A partial list is shown in
Table 2.2.
Among all techniques, extent fuzzy AHP is used most frequently because of its
computational simplicity. In Table 2.3, various techniques are classified based on
single-sourcing and multiple-sourcing supplier selection.

© Springer (India) Pvt. Ltd. 2017 31

K. Mukherjee, Supplier Selection, Studies in Systems, Decision and Control 88,
DOI 10.1007/978-81-322-3700-6_2
32 2 Modeling and Optimization of Traditional Supplier Selection

Table 2.1 Single sourcing versus multiple sourcing

Single sourcing Multiple sourcing
Concept of this strategy comes from Multiple sourcing is preferable if reliability of
just-in-time (JIT) philosophy. Uncertainty in one supplier is very poor. It reduces safety
supply is very high as buyer deals with single stock without increasing stock-out problem
supplier (Kelle and silver 1990). It reduces uncertainty
in supply but increases ordering cost
(Agrawal and Nahmias 1997)
No competition exists as only one supplier is Reduction in price is achieved through
involved. It gives quantity discount from competition between suppliers. It gives better
order consolidation and reduces order lead on-time delivery and higher volume
time and logistical lead time (Hahn et al. flexibility (Ramasesh et al. 1991)
1986; Bozarth et al. 1998)
It is applicable where goodwill trust exists Dual sourcing is always effective for low
between buyer and supplier ordering cost and highly variable lead times
(Ramasesh et al. 1991)
Low threat to loss of information Since business data are shared among various
suppliers, proper security measures should be
taken

2.2 State-of-the-Art Literature Review of Supplier

Selection Methods

The abundant work on supplier selection can be broadly classified into eight dif-
ferent categories as follows:
1. selection of supplier for single item or multiple items for deterministic or
stochastic demand and supply;
2. selection of supplier for manufacturing industry;
3. selection of supplier for service industry;
4. selection of supplier with price–order quantity discount;
5. comparative analysis of single-sourcing and multiple-sourcing strategies;
6. decision support system (DSS) for supplier selection;
7. supplier selection for green supply chain; and
8. supplier selection for new product development.
In this chapter, literature review is conducted to find the followings:
1. to identify relevant criteria for supplier selection;
2. to identify different methods for supplier selection; and
3. to identify the trend of supplier selection methods.
Research work related to supplier selection is considerably very high. For
instance, from www.sciencedirect.com, alone 13, 201 articles were found with the
search word ‘supplier selection’ for publication 2009 onwards. About 100 research
manuscripts are selected from peer-review journals from 1998 to 2012. Papers are
selected based on the reputation of journal and citation of papers to find the most
2.2 State-of-the-Art Literature Review of Supplier Selection Methods 33

Fig. 2.1 Supplier selection methods

cited method(s) for supplier selection, recent trend of supplier selection, etc.
Supplier selection methods are broadly classified into two categories—methods for
single model and integrated model, as shown in Fig. 2.1. Single model is further
classified into three subcategories—methods based on mathematics, statistics, and
artificial intelligence. Integrated models usually combined with linear program
(LP) or genetic algorithm (GA) or particle swarm optimization (PSO) to allocate
order among multiple suppliers. Each method has certain limitations. Therefore,
selection of an appropriate method always remains a daunting task for decision
makers. AHP, ANP, and their integrated methods are mostly preferred by various
researchers because of its simplicity and ability to solve complex problem. Figs. 2.2
and 2.3 clearly indicate such trend. Today, majority of the companies prefer to
34 2 Modeling and Optimization of Traditional Supplier Selection

Table 2.2 Supplier selection technique—a brief review

Sl. Techniques Authors
No.
1. Logarithmic least square method (LLSM) for Van Laarhoven and Pedrycz
fuzzy AHP (1983).
2. Modified logarithmic least square method Wang et al. (2006a, b)
(MLLSM)
3. Fuzzy least square priority method Xu (1996).
4. Lambda-Max method Csutora and Buckley (2001)
5. Eigenvector method Wang et al. (2008a, b)
6. Fuzzy preference programming Mikhailov (2003)
7. Extent analysis Chang (1996)

Table 2.3 Various techniques for single-sourcing and multi-sourcing supplier selection
Single sourcing Multi-sourcing
Sl. Methods Remarks Sl. Methods
No. No.
1. Linear Depends heavily on human 1. Mixed integer programming
weighted judgments
point
2. Categorical Depends heavily on human 2. Goal programming
method judgments
3. Cost ratio Very complicated and needs 3. Single- / multi-objective
more financial information programming
4. AHP More accurate than any other 4. Multi-attribute utility theory and
method (Ghodsypour and AHP; AHP-LP; AHP-GA; AHP
O’Brien 1998) and multi-objective possibilistic
linear programming
(AHP-MOPLP) etc

reduce supply base, and because of that, research trend on supplier selection is
gradually moving from multiple supplier selection to single supplier selection.
However, very less number of research papers has been identified on supply base
reduction (SBR). Different criteria used for supplier selection methods are shown
in Table 2.4. Cost, quality, and service are mostly used for traditional supplier
selection process, for example, delivery time, on-time delivery, and delivery reli-
ability. About 78 papers are analyzed thoroughly from 2005 to 2012 to find out the
application of supplier selection methods in different industries, as shown in
Fig. 2.4 (Table 2.5).
Study reveals that major contribution in research related to supplier selection is
obtained from Taiwan, USA, Turkey, Iran, and China. Their cumulative research
work related to supplier selection is about 69% of total research work. Both India
and UK occupies the same position. This major contribution also inspired
researchers to contribute more on supplier selection methods for electronics,
2.2 State-of-the-Art Literature Review of Supplier Selection Methods 35

Fig. 2.2 Trend of supplier selection methods

Fig. 2.3 Distribution of

research paper for supplier
selection

mechanical or manufacturing, and automobile industries, as shown in Fig. 2.4.

Unfortunately, supplier selection for construction industries is yet to be explored.

2.3 Pareto Analysis of Supplier Selection Criteria

Dickson (1966) in his seminal work proposed twenty-three criteria for supplier
selection. Based on the work of Dickson (1966) and Weber (1991), Pareto analysis
is performed to find most cited criterion for supplier selection. Six criteria such as
net price, delivery, quality, production facilities and capacity, geographic location,
and technical capability are identified as the most cited criteria.
Table 2.4 A partial list of supplier selection criteria
36

Year Author/s Cost Quality Service Delivery Reputation Environment Logistical Commercial Production Technology Responsiveness Supplier’s Risk Reliability Lead Flexibility Supplier’s R&D
performance structure profile factor time willingness

1998 Ghodsypour, X X X
O’Brien
2001 Eon-Kyung X X X X
Lee, Sungdo
Ha, and
Sheung-Kown
Kim,
2001 Maggie C. X X
Y. Tam, Rao
Tummala
2002 Robert X
Handfield,
Steven V.
Walton, Robert
Sroufe, and
Steven A.
Melnyk
2005 Ozden Bayazit X X X
and Birsen
Karpak
2006 Huan-Jyh X X X X X
Shyur and
Hsu-Shih Shih
2007 FU Yao and X X X
LIU Hongli
2007 Felix T.S. X X X X X
Chan and Niraj
Kumar
2007 Weijun Xia and X X X
Zhiming Wu
2007 Min Wu X X X X
2007 Sanjay X X X
Jharkharia and
Ravi Shankar
2007 Cevriye Gencer X X
and Didem
Gürpinar
2007 Ezgi Aktar X X X
Demirtas and
Ozden Ustun

(continued)
2 Modeling and Optimization of Traditional Supplier Selection
Table 2.4 (continued)
Year Author/s Cost Quality Service Delivery Reputation Environment Logistical Commercial Production Technology Responsiveness Supplier’s Risk Reliability Lead Flexibility Supplier’s R&D
performance structure profile factor time willingness

2008 Ali Kokangul X X

and Zeynep
Susuz
2008 Jia-Wen Wang, X X X
Ching-Hsue
Cheng, and
Huang
Kun-Cheng
2008 Reuven R. X X
Levary
2008 Jing-Rung Yu X X X X X
and Chao-Chia
Tsai
2008 Ozan Cakir and X
Mustafa S.
Canbolat
2008 Sung Ho Ha X X
and Ramayya
Krishnan
2008 Amy H.I. Lee X X X X X
2.3 Pareto Analysis of Supplier Selection Criteria

2008 Eleonora X
Bottani and
Antonio Rizzi
2008 Rong-Ho Lin X X X X
2008 Semih Önüt, X X X
Selin Soner
Kara, and Elif
Is_ik
2008 Ozeden Ustun X X X
and Ezgi Aktar
Demı̇rtas
2009 Wann-Yih Wu X X
et al.
2009 Chia-Wei Hsu, X X
Allen H. Hu
37
38 2 Modeling and Optimization of Traditional Supplier Selection

Fig. 2.4 Application of supplier selection methods to different industries

2.4 Stages of Procurement

Supplier selection is the process by which suppliers are reviewed, evaluated, and
chosen to become part of the company’s supply chain (Sanayei et al. 2010). The
overall objective of the supplier selection process is as follows (Chena eta al 2006):
1. to reduce the procurement risk;
2. to maximize the overall value of purchase; and
3. to build the closeness and long-term relationships between buyers and suppliers.
Supplier selection is not a mere clerical issue or a mere optimization problem.
Supplier selection is a strategic issue of any business because of the following
reasons:
1. Procurement is considered as value addition process to supply chain.
2. Active supplier involvement can enhance efficiency and effectiveness of supply
chain.
3. Short product life cycle and rapid product innovation give more emphasizes on
integration of material and information flows, both internally and externally.
Supplier selection process consists of four stages—problem definition; formu-
lation of attributes; qualification of potential suppliers; and the final selection of best
suppliers (De Boer et al. 2001). A generalized procurement cycle can be con-
sidered that consists of the following stages:
1. Recognition of need: Identify the demand of product.
2. Specification: Identify part/assembly/raw material specifications.
3. Make or buy decision: It is one of the most crucial stages of procurement cycle
to think over about source materials, goods, price, etc. Usually, a company is
supposed to take, make, or buy decision for the following reasons:
2.4 Stages of Procurement 39

Table 2.5 Application area wise distribution of research paper from 2005 to 2012
Year Author(s) Application areas
2005 Hong et al. Agricultural industry in Korea
2005 Chen et al. Electronic components
2005 Bayazit and Karpak Construction company
2006 Kubat and Yuce General
2006 Mouli et al. General
2006 Sarfaraz and Balu General
2006 Chen et al. High tech manufacturing
2006 Shyura and Shih Local Taiwanese company
2007 Gencer and Gu¨rpinar Electronic company
2007 Demirtas and Ustun Refrigerator producers
2007 Reza Farzipoor Saen General
2007 Che et al. Semiconductor industry
2007 Mehdizadeh and Moghaddam General
2007 Guo et al. General
2007 Yao and Hongli Information & Mgmt Sys outsourcing
2007 Fayez et al. General
2007 Huang and Keskar PC manufacturer
2007 Amid et al. General
2007 Li et al. General
2007 Chan and Kumar Manufacturing company
2007 Xia and Wu General
2007 Min Wu General
2007 Guan et al. General
2008 Kokangul and Susuz Automotive industry
2008 Rong-Ho Lin General
2008 Moghadam et al. General
2008a, b Wang et al. Lithium-ion battery
2008 Che and Wang PDA
2008 Hong and Ha Agricultural industry
2008 Reuven R. Levary Manufacturing company
2008 Yu and Tsai Semiconductor industry
2008 Lin and Chang Manufacturing company
2008 Wan Lung Ng General
2008 Amin and Razmi ISP
2008 Wu et al. TFT-LCD industry
2008 Chou and Chang IT Industry
2008 Ha and Krishnan Automobile industry
2008 Lee et al. TFT-LCD industry
2008 Amy H.I. Lee TFT-LCD industry
2008 Önüt et al. Telecommunication industry
(continued)
40 2 Modeling and Optimization of Traditional Supplier Selection

Table 2.5 (continued)

Year Author(s) Application areas
2008 Zhang et al. General
2008 Bottani and Rizzi Manufacturing company
2008 Rhee et al. Manufacturing company
2008 Wu and Olson General
2008 Yang et al. Electronic manufacturing company
2008 Ustun and Dem˙ırtas Refrigerator manufacturing
2009 Amy H.I. Lee TFT-LCD industry
2009 Wu et al. Notebook manufacturer
2009 Hsu and Hu Electronic manufacturing company
2009 Wang et al. Notebook manufacturer
2010 Wen-Pai Wang Electronic manufacturing company
2010 Wu et al. General
2010 Sanayei et al. Automobile industry
2010 Tadeusz Sawik General
2011 Bilsel and Ravindran General
2011 Tadeusz Sawik General
2011 Amid et al. General
2011 Selin Soner Kara Paper production
2012 Erdem and Göçen White goods manufacturer
2012 Shaw et al. Garment manufacturing
2012 Mukherjee and Kar General
2012 Jin Wang General
2012 Riedl et al. General
2012 Bruno et al. General
2012 Choudhary and Shankar General
2012 Parthiban et al. Automotive industry

1. sudden increase in procurement cost;

2. need for design secrecy;
3. lack of specific technical competency of suppliers in supply base;
4. poor services of existing suppliers; and
5. Unpredictable deterioration of existing supplier’s performance, etc.
4. Source Identification: Prepare a supply base as per requirement.
5. Source selection: Organization has to think about single sourcing or optimal
number of sourcing as per the goal of organization. Usually, it consists of four
stages:
1. Select criteria to consider palpable and non-palpable issues of supplier
selection.
2. Select appropriate method for supplier selection.
2.4 Stages of Procurement 41

Fig. 2.5 Most cited criteria for supplier selection

3. Select supplier(s) as per the unanimous preference of decision makers.

4. Allocate order to selected suppliers as per mathematical model.
For single sourcing, last stage is not considered as entire order is allocated to
single/best supplier.
6. Contracting: Placed order to selected suppliers.
In brief, supplier selection procedure can be classified as preselection, selection,
and post-selection.
Preselection and post-selection are highly subjective and varied from company
to company as per their goal of procurement process.

2.5 Qualities of Good Supplier

Quality of good supplier is highly subjective as it varied from company to com-

pany, product to product, and process to process. It also depends on the type of
sourcing decisions. Sourcing decisions are classified as follows:
1. consumable supplies;
2. production materials and components;
3. capital purchases (e.g., machinery);
4. intellectual property (e.g., software);
5. subcontractors; and
6. services.
Based on the above classifications, a generalized list of qualities of good sup-
pliers can be mentioned as follows:
1. on-time delivery;
2. technical capabilities;
42 2 Modeling and Optimization of Traditional Supplier Selection

3. consistent quality;
4. reasonable low price;
5. good past performance record;
6. ability to maintain volume flexibility to withstand sudden variations in demand;
7. presales and post-sales support;
8. ability to provide his buyers tracking facilities to track the progress of supply.
Such tracking process could enhance the reliability of supply;
9. industrial certifications such as ISO and TUV; and
10. proactive to develop a healthy relation with his buyers.

2.6 How to Prepare Supply Base?

A stable supply base could enhance availability of raw material/parts/assembly,

increase buyers’ bargain power, and increase the possibility to get best supplier(s) to
develop a long-term relationship. Preparation of supply base is not a much discussed
topic in supplier selection literature as major focus is given to selection and evaluation
of supplier(s). Often, it is considered as a preprocess of supplier selection. Author
strongly suggests that due care should be given to prepare supply base as poor supply
base gives low probability of selecting good supplier(s) even if very supplier selection
method is used. Following are some of the criteria for preparing supply base:
1. reputation and industrial certification of suppliers;
2. availability of past performances of supplier with authenticated documents;
3. availability of well-documented product catalog. For raw material, a
well-documented test report which contains chemical and physical properties
should be provided.
4. Ability to provide goods as per delivery due date. Usually, it varies from
company to company.
5. For new product, technical know-how of supplier should be verified. At the
same time, availability of technical equipments of the supplier should be
verified.

2.7 Supplier Selection for Mass Customized System

In Chap. 3, a detailed discussion is given for customized production system,

postponement, and CODP. Mass customization is a strategy to manufacture
customized product from standard product with near mass production efficiency.
Mass customization can be broadly classified as assembly to order (ATO), build to
order (BTO), engineer to order (ETO), and make to stock (MTS) which is
commonly used for standard product to reduce customer’s waiting time. Literature
2.7 Supplier Selection for Mass Customized System 43

review reveals ample work on MTS but very few works on ATO/BTO/ETO. In this
book, main focus is given to supplier selection methods, development of hybrid
model for supplier selection, development of mathematical model for supplier
selection for ATO system, and design of decision support system for sustainable
supplier selection and strategic sourcing.

2.8 Hybrid Methods for Supplier Selection

In this chapter, three hybrid methods are proposed for supplier evaluation, selection, and
order allocation problem. In the first method, modified extent fuzzy AHP is used to
consider tangible and intangible criteria for supplier selection, and for order allocation,
GA is used. In the second method, fuzzy AHP-TOPSIS is used to consider palpable and
non-palpable criteria for supplier selection, and order allocation GA is used.

2.8.1 Modified Extent Fuzzy AHP and GA (MEFAHP-GA)

Chang’s (1996) extent analysis is based on the following steps:

1. If Mgi i are the triangular fuzzy numbers (TFNs) where gi is the goal set
(i = 1, 2, 3…m),
the fuzzy extent value Si with respect to the ith criterion is defined as

X
m n X
X m
Si ¼ Mgi i  ½ Mgi i 1 ð2:1Þ
i¼1 i¼1 i¼1

 
Pm Pm P
m
where Mgi i ¼ i¼1 l i¼1 m u
i¼1
where l is the lower limit value, m is the most promising value, and u is the
upper limit value.
and

X
n X
m  
1 1 1
½ Mgi i 1 ¼ Pn ; Pn ; Pn ð2:2Þ
i¼1 i¼1 i¼1 ui i¼1 mi i¼1 li

2. The degree of possibility of M2  M1 is given by V(M2  M1)

where
8
< 1; if m2  m1
V ðM2  M1 Þ ¼ 0; if l1  u2
: l1 u2
ðm2 u2 Þðm1 l1 Þ ; otherwise
44 2 Modeling and Optimization of Traditional Supplier Selection

If d is the highest intersection point of lM1 and lM2 as shown in Fig. 2.6,

d ðAi Þ ¼ min V ðSi  Sk Þ for k ¼ 1; 2; 3; 4; 5 . . . n; k 6¼ i

The weight vector is W  ¼ ðd ðA1 Þ; d ðA2 Þ; d ðA3 Þ; . . .:; d ðAn ÞÞT

The normalized weight vector is W ¼ PnW

.
d ðAi Þ
I¼1

TFNs are used for fuzzy comparisons, as shown in Table 2.6.

Reason behind the popularity of extent fuzzy AHP is its computational sim-
plicity. However, it is unable to find the true weights from fuzzy comparison
matrix. Wang et al. (Wang et al. 2008a, b) pointed out that Eq. (2.2) should be
modified to find the true fuzzy extent value of ith criteria. This method is known as
modified extent fuzzy AHP (MEFAHP).
" #1
Xn Xm
i
Mgi
i¼1 i¼1
Pn Pn Pn !
lij j¼1 mij j¼1 uij
¼ Pn Pj¼1
n Pn ; n Pn
P ; Pn Pn Pn
j¼1 lij þ k¼1; k6¼i j¼1 ukj k¼1 j¼1 mkj j¼1 uij þ k¼1; k6¼i j¼1 lkj

where i = 1,2,3…n.

Fig. 2.6 Intersection of two TFNs (from Zhu et al. 1999; with kind permission from Elsevier
Limited)
Table 2.6 Fuzzy TFN values (from Tolga et al. 2005; with kind permission from Elsevier
Limited)
Linguistic values Fuzzy numbers
Equal (1,1,1)
Weak (2/3,1,3/2)
Fairly strong (3/2,2,5/2)
Very strong (5/2,3,7/2)
Absolute (7/2,4,9/2)
2.8 Hybrid Methods for Supplier Selection 45

Wang et al. (2008a, b) further proposed that extent analysis method may be
considered as the method for showing how bigger one a decision is than the others
in fuzzy comparison and there is chances of loosing of information when it assigns
irrational zero to some important criteria or subcriteria. Linear program (LP),
integer program (IP), goal program, etc., can be used alone to consider limitations
of supplier(s) in supplier selection problem. However, these methods cannot con-
sider qualitative criteria for supplier selection. Therefore, combination of AHP-GA
or AHP-LP is a better choice. Extent fuzzy AHP alone is more suitable for single
sourcing where best supplier is capable enough to fulfill the entire demand. The
following steps are used to combine extent fuzzy AHP with GA.
Step 1. Define goal or objective of the problem.
Step 2. Select criteria for selecting suppliers.
Step 3. Select fuzzy membership function for fuzzy comparison matrix.
Step 4. Find priority of suppliers by extent fuzzy AHP.
Step 5. Form objective function and constraints.
Step 6. Use genetic algorithm to solve single-objective constrained objective
function.
Order allocation to selected suppliers with GA is discussed in detail in the next
section mentioned below.

2.8.2 Fuzzy TOPSIS-MOGA

The following steps are maintained to rank suppliers from a predefined supply base:
Step 1: TFNs, ãij, are used to find suitability of each alternatives w.r.t criteria.
Step 2: Since TFNs are already normalized, no need for normalization. Calculate
the weighted normalized fuzzy decision matrix, Ũ.

e ¼ ½~uij 
U i ¼ 1; 2; 3. . .n j ¼ 1; 2; . . .j
nj
~uij ¼ ~aij  wi

where wi is the set of weight of each criterion derived by analytic

hierarchy process (AHP).
Step 3: Identify FPIS (fuzzy positive ideal solution) and FNIS (fuzzy negative
ideal solution). Calculate the distance of each alternative from FPIS and
FNIS.
Step 4: Calculate the closeness coefficient of each alternative, CCi.
D
where CCj ¼ D þ þj D
j j

Step 5: Rank suppliers based on higher value of CCi..

46 2 Modeling and Optimization of Traditional Supplier Selection

2.8.3 Multi-Objective Model for Supplier Selection

A multi-objective order allocation model is developed with five objective functions

and three constraints. The following assumptions are considered to develop order
allocation model
Assumptions
1. Selected suppliers will supply only one item.
2. No quantity discount is considered.
3. No shortage of item is allowed for any supplier.
4. Deterministic constant demand is considered.
Ci Procurement cost of per ton of coal from ith supplier
TCi Transportation cost of per ton of coal from ith supplier
CCi Closeness coefficient of ith supplier obtained from fuzzy TOPSIS
ai Reliability of ith supplier
Xi Order quantity to ith supplier
LDi Percentage of late delivery from ith supplier
bi Percentage of coal contains 15–18% of ash in per ton received from ith
supplier
ci Percentage of coal contains 15–16% of moisture in per ton received from ith
supplier
H Handling cost per ton
Order allocation model:
Total cost of purchase (TCP) consists of purchase, transportation, order/setup, and
holding cost. Order/setup cost is neglected in this mathematical model, and material
handling cost is considered as holding cost.
Pn P
n P
n
Minimize total cost of purchase (TCP): Ci Xi þ TCi Xi þ H Xi
i¼1 i¼1 i¼1
Second objective function in our mathematical model is similar to Ghodsypour
and O’Brien (1998). However, their proposed objective function is modified as total
value of reliable purchase (TVRP). Reliability of supply of each supplier is cal-
culated from past performance data of supplier.
Pn
Maximize total value of reliable purchase (TVRP): ai CCi Xi
i¼1
Third objective function is to mitigate supply risk. Fourth and fifth objective
functions are to maintain desired quality level. In any cement company, every lot is
accepted based on two quality parameters—ash content and surface moisture content.
P
n
Minimize number of late deliveries: LDi Xi
i¼1
P
n
Minimize amount of rejected lot based on ash content: ð1  bi Þ Xi
i¼1
P
n
Minimize amount of rejected lot based on moisture content: ð1  ci Þ Xi
i¼1
2.8 Hybrid Methods for Supplier Selection 47

Constraints for supplier selection:

Supplier capacity constraint, minimum order quantity to fulfill demand constraint,
and cost or budgetary constraint are some of the most significant constraints of
order allocation model (Kumar et al. 2004; Ghodsypour and O’Brien 1998). The
following constraints are considered to optimize above five objective functions:
Capacity constraint: Xi Vi for i ¼ 1; 2; 3. . .n
P
n
Demand constraint: Xi ¼ D
i¼1
P
n
Cost constraint: Ci Xi B
i¼1
Non-negativity constraint: Xi  0 for i ¼ 1; 2; 3. . .n
The proposed model is solved by using MATLAB R 2009a and run it on a
personal computer intel(R) Core(TM) 2 Duo 2.00 Ghz. This integrated model
consists of four stages—preprocessing, supplier selection, order quantity

Fig. 2.7 Integrated model of f-TOPSIS-MOGA

48 2 Modeling and Optimization of Traditional Supplier Selection

calculation, and post-processing. In preprocessing, only preparation of supply base

is considered. A supply pool is built initially with predefined supply lead time. In
supplier selection stage, suppliers are selected from supply base by combined fuzzy
TOPSIS and approval status proposed by Chen et al. (2006). Third stage uses
genetic algorithm to optimize multi-objective, and finally, order quantities are
selected from Pareto-optimal solutions. In the last stage, orders are placed to
selected suppliers. Flowchart of the integrated approach is shown in Fig. 2.7.

2.8.4 Case Study

High initial investment, lack of resources, land acquisition problem for expansion,
and long waiting time to get desired return on investment are some of the major
barrier for cement industries. Moreover, full capacity utilization of cement plants is
highly influenced by demand of realty sector. About 67% of the total production of
cement is used in housing sector, 13% is used in commercial construction, 11% is
used in infrastructure project, and only 9% is used in industrial construction. India

Fig. 2.8 Cost of producing cement

Table 2.7 Priority of supplier selection criteria
Criteria Priority Inconsistency
Quality(C1) 0.3132 0.0935
Price(C2) 0.0819
Capacity(C3) 0.0819
Location(C4) 0.5230

Table 2.8 TFN values

Linguistic values Fuzzy numbers
Very low (VL) (0,0,0.2)
Low(L) (0,0.2,0.4)
Medium (M) (0.2,0.4,0.6)
High(H) (0.4,0.6,0.8)
Very High(VH) (0.6,0.8,1)
Excellent (0.8,1,1)
2.8 Hybrid Methods for Supplier Selection 49

is the second largest producer of cement after China. Indian cement industry is
basically oligopolistic in nature with more than 160 companies scattered all over
India. Northern, eastern, southern, western, and central are the five main regions
responsible for cement production in India. Andhra Pradesh, Tamil Nadu, and
Rajasthan are the main contributors to Indian cement industry. Till early 2000,
Ordinary Portland Cement (OPC) was the main variety of cement in India. Since
2005, production of Portland Pozzolana Cement (PPC) was increased at the cost of
production of OPC. Today, about 61% of total production is PPC. Total 20% of
total cost is spent for procuring coal to produce cement, as shown in Fig. 2.8.
An ISO 9001:2000 certified company which is situated in north east is producing
various grades of cement such as Ordinary Portland Cement (OPC) and Portland
Pozzolana Cement (PPC). Limestone and coal are two important raw materials for
cement. Gypsum is essential for OPC, and fly ash is essential for PPC. Company
will select supply of coal if its ash content is 15–18% and surface moisture content
is 15–16%. Moreover, company can wait maximum three days to get supply.
Material handling cost comes to Rs. 350 per ton, and order/setup cost and other
holding cost are negligible. Four criteria, namely quality, price, capacity, and
location of the supplier, have been chosen to select suppliers as per the consensus of
the decision maker’s committee which encompasses senior members from finance,
marketing, purchase, and sales department of the focal company.

Fig. 2.9 Membership function of linguistic values

Table 2.9 Fuzzy evaluation matrix of alternatives

Alternative C1 C2 C3 C4
A1 High Low Medium Medium
A2 Medium Medium High Very high
A3 Very high Low High Excellent
A1 (0.4,0.6,0.8) (0,0.2,0.4) (0.2,0.4,0.6) (0.2,0.4,0.6)
A2 (0.2,0.4,0.6) (0.2,0.4,0.6) (0.4,0.6,0.8) (0,0,0.2)
A3 (0.6,0.8,1) (0,0.2,0.4) (0.4,0.6,0.8) (0.8,1,1)
Weight 0.3132 0.0819 0.0819 0.5230
 50

Table 2.10 Weighted evaluation for three suppliers

Alternative C1 C2 C3 C4
A1 (0.125,0.188,0.251) (0,0.016,0.033) (0.016,0.033,0.049) (0.105,0.209,0.314)
A2 (0.063,0.125,0.188) (0.016,0.033,0.049) (0.033,0.049,0.066) (0,0,0.105)
A3 (0.188,0.251,0.3132) (0,0.016,0.033) (0.033,0.049,0.066) (0.418,0.523,0.523)
A+ (1,1,1) (0,0,0) (1,1,1) (0,0,0)
A− (0,0,0) (1,1,1) (0,0,0) (1,1,1)
2 Modeling and Optimization of Traditional Supplier Selection
2.8 Hybrid Methods for Supplier Selection 51

Pairwise comparison value for each criterion is obtained from each decision
maker. After that, analytic hierarchy process (AHP) is used to calculate priority of
each criterion, as shown in Table 2.6, and linguistic values are shown in Table 2.7
and in Fig. 2.9.
As stated above, detail calculation is shown in Tables 2.8, 2.9, 2.10, and 2.11.
Supplier performance data are shown in Table 2.12.
All three suppliers can be accepted with low risk to supply coal.
Multi-objective functions for supplier selection:
In this problem, a linear total cost function (TC(Q)=a + bQ) is considered for all
three suppliers for simplicity. As shown below, a nonlinear integer function is
developed for total cost of purchase which is to be minimized, as shown in
Fig. 2.10.
Minimize total cost of purchase (TCP): (3099 + 10  1)  1 + (3100 +
10  2)  2 + (3102 + 10  3)  3
Maximize total value of reliable purchase (TVRP): 0.4674  1 + 0.4982 
2 + 0.4133  3
Minimize delay in supply: 0.1  1 + 0.15  2 + 0.2  3
Quality:
1. Minimize defects to maintain permissible ash content in supply:
0.2  1 + 0.25  2 + 0.3  3
2. Minimize defects to maintain permissible moisture content in supply:
0.15  1 + 0.2  2 + 0.2  3
Subject to
Demand constraint:  1 +  2 +  3 = 8000
Production constraint:  1 4000;  2 3000;  3 3000
Budget constraint: (2000 + 10  1)  1 + (2000 + 10 2)  2 + (2000 +
10 3)  3 30000000

Table 2.11 Fuzzy TOPSIS result

Alternatives D+i D−i CCi
A1 2.0282 2.0092 0.4976
A2 1.9228 2.1201 0.5244
A3 2.2134 1.8052 0.4492

Table 2.12 Approval status

Closeness coefficient (CCi) Assessment status
CCi
[0,0.3) Rejected
CCi
[0.3,0.5) Recommended with high risk
CCi
[0.5,0.7) Recommended with low risk
CCi
[0.7,0.9) Approved
CCi
[0.9,1.0] Highly recommended
 52

Table 2.13 Supplier performance data

Sl % failure rate of Reliability Capacity Total cost Transportation Quality % late delivery
No. supply (f) (a = 1 − f) (ton) (Rs/ton) cost (Rs/ton) % of coal contains % of coal contains 15–16%
15–18% ash in per ton moisture in per ton
1. 6 0.94 4000 2000 + 10*Q 749 0.8 0.85 0.1
2. 5 0.95 3000 2000 + 10*Q 750 0.75 0.8 0.15
3. 8 0.92 3000 2000 + 10*Q 752 0.7 0.8 0.2
2 Modeling and Optimization of Traditional Supplier Selection
2.8 Hybrid Methods for Supplier Selection 53

Fig. 2.10 Objective function for total cost of purchase (TCP)

Fig. 2.11 Result of multi-objective GA

The multi-objective problem consists of nonlinear objective function with one

nonlinear constraint. It cannot be solved with MATLAB GA solver. Penalty
function approach is considered to convert constraint optimization problem to
unconstrained optimization problem. Multi-objective GA finds multiple and
54 2 Modeling and Optimization of Traditional Supplier Selection

diverse Pareto-optimal (or near Pareto-optimal) solutions in a single simulation run.

Therefore, it is important to choose a particular solution from a set of nondominated
solutions. A discontinuous Pareto front is obtained, shown in Fig. 2.11.
The number of points on the Pareto front was 50. The average distance mea-
sure of the solutions on the Pareto front was 0.0574885. The spread measure of
the Pareto front was 0.0700832. Finally, ordered quantities to three suppliers are
{2666, 2666, 2668}.

2.9 Conclusion

Supplier selection is not a mere clerical process. It encompasses several palpable

and nonpalpable criteria. It is a multi-criteria-based optimization process. Effective
selection of supplier could reduce uncertainty of availability of raw material, assure
quality throughout the supply chain, reduce upstream supply chain risk, and, finally,
reduce cost of manufacturing of product. About 70% of total cost is usually spent in
procurement. Thus, procurement is most important for any company. In this
chapter, several methods and criteria are mentioned through rigorous literature
survey. Most cited criteria are identified with latest trend of supplier selection. As
per the latest trend, two methods are discussed in detail with a case study. This
chapter discusses in detail supplier pool preparation, supplier selection, evaluation,
and order allocation with the above two methods.
Majority of the supplier selection models proposed by different researchers are
for electronics industries, automobile industries, etc. In this chapter, an attempt has
been made to prepare mathematical model to select suppliers for cement industries.
Cement is a localized product and needs some extra constraints that are required to
select and allocate order to suppliers. Proposed model in this chapter is prepared
accordingly. It is pertinent to mention that application of the proposed models is not
limited to the cement industries alone. It can be used for any industry with simple
modification.
Total value of purchase is commonly used to allocate order to selected suppliers.
In this chapter, total value of reliable purchase (TVRP) is considered instead of total
value of purchase (TVP) to reduce the upstream supply chain risk, if any. TVRP is a
weighted nonlinear objective function which is prepared with the priority obtained
from the proposed MCDA tools. Industries such as cement and R&D usually face
high risk to manufacture their product within due time. In such cases, TVRP should
be used instead of TVP.

References

Agrawal N, Nahmias S (1997) Rationalization of the supplier base in the presence of yield
uncertainty. Prod Oper Manage 6(3):291–308
References 55

Amid A, Ghodsypour SH, O’Brien C (2007) A weighted additive fuzzy multi-objective model for
the supplier selection problem under price breaks in a supply chain. Int J Prod Econ. doi:10.
1016/j.ijpe.2007.02.040
Amid A, Ghodsypour SH, O’Brien C (2011) A weighted max–min model for fuzzy
multi-objective supplier selection in a supply chain. Int J Prod Econ 131:139–145
Amin SH, Razmi J (2008) An integrated fuzzy model for supplier management: a case study of
ISP selection and evaluation. Expert Syst Appl. doi:10.1016/j.eswa.2008.10.012
Bayazita O, Karpak B (2005) An AHP application in vendor selection. In: Proceedings of ISAHP
Honolulu, Hawaii, 2005 July 8–10
Bilsel RU, Ravindran A (2011) A multiobjective chance constrained programming model for
supplier selection under uncertainty. Transp Res Part B 45:1284–1300
Bottani E, Rizzi A (2008) An adapted multi-criteria approach to suppliers and products selection—
an application oriented to lead-time reduction. Int J Prod Econ 111:763–781
Bozarth C, Handfield R, Das A (1998) Stages of global sourcing strategy evolution: An
exploratory study. J Oper Manage 16(2–3):241–255
Bruno G, Esposito E, Genovese A, Passaro R (2012) AHP-based approaches for supplier
evaluation: problems and perspectives. J Purchasing Supply Manage 18:159–172
Cakir O, Canbolat MS (2008) A web-based decision support system for multi-criteria inventory
classification using fuzzy AHP methodology. Expert Syst Appl 35:1367–1378
Chan FTS, Kumar N (2007) Global supplier development considering risk factors using fuzzy
extended AHP-based approach. Omega 35:417–431
Chang DY (1996) Application of The Extent Analysis Method of Fuzzy AHP. Eur J Oper Res
95:649–655
Che ZH, Wang HS (2008) Supplier selection and supply quantity allocation of common and
non-common parts with multiple criteria under multiple products. Comput Ind Eng 55:110–133
Che ZH, Wang HS, Sha DY (2007) A multi-criterion interaction-oriented model with proportional
rule for designing supply chain networks. Expert Syst Appl 33:1042–1053
Chen KL, Chen KS, Li RK (2005) Suppliers capability and price analysis chart. Int J Prod Econ
98:315–327
Chen CT, Lin CT, Huang SF (2006) A fuzzy approach for supplier evaluation and selection in
supply chain management. Int J Prod Econ 102:289–301
Chou S-Y, Chang Y-H (2008) A decision support system for supplier selection based on a
strategy-aligned fuzzy SMART approach. Expert Syst Appl 34:2241–2253
Choudhary D, Shankar R (2012) Joint decision of procurement lot-size, supplier selection, and
carrier selection. J Purchasing Supply Manage. doi:10.1016/j.pursup.2012.08.002
Csutora R, Buckley JJ (2001) Fuzzy hierarchical analysis: the Lamda-Max method. Fuzzy Sets
Syst 120:181–195
Daǧdeviren M, Yavuz S, Kilinç N (2009) Weapon selection using the AHP and TOPSIS methods
under fuzzy environment. Expert Syst Appl 36(4):8143–8151
De Boer L, Labro E, Molrlacchi (2001) A Review of methods supporting supplier Selection. Eur J
Purchasing Supply Manage 7(2):75–89
Deb K (2001) Multi-objective optimization using evolutionary algorithms. John Willey and Sons,
New York
Demirtas EA, Ustun O (2007) Analytic network process and multi-period goal programming
integration in purchasing decisions. Comp Ind Eng. doi:10.1016/j.cie.2006.12.006
Dickson GW (1966) An analysis of vendor selection systems and decisions. J Purchasing 2(1):5–
17
Erdem AS, Göçen E (2012) Development of a decision support system for supplier evaluation and
order allocation. Expert Syst Appl 39:4927–4937
Faez F, Ghodsypour SH, O’Brien C (2007) Vendor selection and order allocation using an
integrated fuzzy case-based reasoning and mathematical programming model. Int J Prod Econ.
doi:10.1016/j.ijpe.2006.11.022
Gaither N (1996) Production and operations management. Duxbury Press, Florence, KY
56 2 Modeling and Optimization of Traditional Supplier Selection

Gencer C, Gürpinar D (2007) Analytic network process in supplier selection: a case study in an
electronic firm. Appl Math Model 31:2475–2486
Ghodsypour SH, O’Brien C (1998) A decision support system for supplier selection using an
integrated analytic hierarchy process and linear programming. Int J Prod Econ 56–57:199–212
Ghodsypour SH, O’Brien C (2001) The total cost of logistics in supplier selection, under
conditions of multiple sourcing, multiple criteria and capacity constraint. Int J Prod Econ
73:15–27
Guan Z, Jin Z, Zou B (2007) A multi-objective mixed-integer stochastic programming model for
the vendor selection problem under multi-product purchases. Inf Manag Sci 18(3):241–252
Guo M, Zhu J, Zhao X (2007) A Bi-level programming model for supplier selection in
constructing logistics service supply chain. In: Proceedings of 2007 IEEE IEEM ISBN:
1-4244-1529-2/07
Ha SH, Krishnan R (2008) A hybrid approach to supplier selection for the maintenance of a
competitive supply chain. Expert Syst Appl 34:1303–1311
Hahn CK, Kim KH, Kim JS (1986) Costs of competition: implications for purchasing strategy.
J Purchase Mater Manag (Fall)
Handfield R, Walton SV, Sroufe R, Melnyk SA (2002) Applying environmental criteria to supplier
assessment: A study in the application of the analytical hierarchy process. Eur J Oper Res
141:70–87
Hong G, Ha SH (2008) Evaluating supply partner’s capability for seasonal products using machine
learning techniques. Comput Ind Eng 54:721–736
Hong GH, Park SC, Jang DS, Rho HM (2005) An effective supplier selection method for
constructing a competitive supply-relationship. Expert Syst Appl 28:629–639
Hsu C-W, Hu AH (2009) Applying hazardous substance management to supplier selection using
analytic network process. J Clean Prod 17:255–264
Huang SH, Keskar H (2007) Comprehensive and configurable metrics for supplier selection. Int J
Prod Econ 105:510–523
Jang J, Sun C, Mizutani E (2004) Neuro-Fuzzy and soft computing: a computational approach to
learning and machine intelligence. Prentice-Hall, India
Jharkharia S, Shankar R (2007) Selection of logistics service provider: an analytic network
process. Omega 35:274–289
Kelle P, Silver EA (1990) Decreasing expected shortages through order splitting. Eng Costs Prod
Econ 19:351–357
Kokangul A, Susuz Z (2008) Integrated analytical hierarch process and mathematical program-
ming to supplier selection problem with quantity discount. Appl Math Model. doi:10.1016/j.
apm.2008.01.021
Kubat C, Yuce B (2006) Supplier selection with genetic algorithm and fuzzy AHP. In: Proceedings
of the 5th international symposium on intelligent manufacturing systems, 29–31 May
Kumar M, Vrat P, Shankar R (2004) A fuzzy goal programming approach for vendor selection
problem in a supply chain. Comput Ind Eng 46:69–85
Lee AHI (2008) A fuzzy supplier selection model with the consideration of benefits, opportunities,
costs and risks. Expert Syst Appl. doi:10.1016/j.eswa.2008.01.045
Lee AHI (2009) A fuzzy supplier selection model with the consideration of benefits, opportunities,
costs and risks. Expert Syst Appl 36:2879–2893
Lee E, Ha S, Kim S (2001) Supplier selection and management system considering relationships in
supply chain management. IEEE Trans Eng Manage 48(3):307–317
Lee AHI, Kang H-Y, Chang C-T (2008) Fuzzy multiple goal programming applied to TFT-LCD
supplier selection by downstream manufacturers. Expert Syst Appl. doi:10.1016/j.eswa.2008.
08.044
Levary RR (2008) Using the analytic hierarchy process to rank foreign suppliers based on supply
risks. Comput Ind Eng 55:535–542
Li GD, Yamaguchi D, Nagai M (2007) A grey-based decision-making approach to the supplier
selection problem. Math Comp Model 46:573–581
References 57

Lin R-H (2008) An integrated FANP–MOLP for supplier evaluation and order allocation. Appl
Math Model. doi:10.1016/j.apm.2008.08.021
Lin H-T, Chang W-L (2008) Order selection and pricing methods using flexible quantity and fuzzy
approach for buyer evaluation. Eur J Oper Res 187:415–428
Mehdizadeh E, Moghaddam RT (2007) A Hybrid Fuzzy Clustering PSO Algorithm for a
Clustering Supplier Problem. In: Proc. 2007 IEEE IEEM ISBN: 1-4244-1529-2/07
Mikhailov L (2003) Deriving priorities from fuzzy pair wise comparison judgements. Fuzzy Sets
Syst 134:365–385
Moghadam MRS, Afsar A, Sohrabi B (2008) Inventory lot-sizing with supplier selection using
hybrid intelligent algorithm. Appl Soft Comput 8:1523–1529
Mouli KVVC, Subbaiah KV, Rao KM, Acharyulu SG (2006) Particle swarm optimization
approach for vendors selection. IE(I) J-PR 87:3–6
Moynihan GP, Saxena P, Fonseca DJ (2006) Development of decision support system for
procurement operations. Int J Logistics Sys Manag 2(1):1–18
Mukherjee S, Kar S (2012) A three phase supplier selection method based on fuzzy preference
degree. J King Saud Univ—Comp Inf Sci http://dx.doi.org/10.1016/j.jksuci.2012.11.001
Ng WL (2008) An efficient and simple model for multiple criteria supplier selection problem. Eur J
Oper Res 186:1059–1067
Önüt S, Kara SS, Işik E (2008) Long term supplier selection using a combined fuzzy MCDM
approach: a case study for a telecommunication company. Expert Syst Appl. doi:10.1016/j.
eswa.2008.02.045
Opricovic S, Tzeng G (2004) Compromise solution by MCDM methods: a comparative analysis of
VIKOR and TOPSIS. Eur J Oper Res 156:445–455
Parthiban P, Zubar HA, Garge CP (2012) A multi criteria making approach for suppliers selection.
Procedia Eng 38:2312–2328
Ramasesh RV, Ord JK, Hayya JC, Pan AC (1991) Sole versus dual sourcing in stochastic lead
time(s, Q) inventory models. Manage Sci 37(4):428–443
Riedla DF, Kaufmann L, Zimmermann C, Perols JL (2012) Reducing uncertainty in supplier
selection decisions: antecedents and outcomes of procedural rationality. J Oper Manag. doi:10.
1016/j.jom.2012.10.003
Saen RF (2007) Suppliers selection in the presence of both cardinal and ordinal data. Eur J Oper
Res 183:741–747
Sanayei A, Mousavi SF, Yazdankhah A (2010) Group decision making process for supplier
selection with VIKOR under fuzzy environment. Expert Syst Appl 37:24–30
Sarfaraz AR, Balu R (2006) An integrated approach for supplier selection. IEEE ISBN:
0-7803-9701-0/06
Sawik T (2010) Single vs. multiple objective supplier selection in a make to order environment.
Omega 38:203–212
Sawik T (2011) Supplier selection in make-to-order environment with risks. Math Comp Model
53:1670–1679
Shaw K, Shankar R, Yadav SS, Thakur LS (2012) Supplier selection using fuzzy AHP and fuzzy
multi-objective linear programming for developing low carbon supply chain. Expert Syst Appl
39:8182–8192
Shyur HJ, Shih HS (2006) A hybrid MCDM model for strategic vendor selection. Math Comput
Model 44:749–761
Soner Kara S (2011) Supplier selection with an integrated methodology in unknown environment.
Expert Syst Appl 38:2133–2139
Srinivas N, Deb K (1994) Mult-iobjective optimization using non-dominated sorting in genetic
algorithms. Evol Comput 2(3):221–248
Tam MCY, Tummala VMR (2001) An application of the AHP in vendor selection of a
telecommunications system. Omega 29(2):171–182
Tolga E, Demircan ML, Kahraman C (2005) Operating system selection using fuzzy replacement
analysis and analytic hierarchy process. Int J Prod Econ 97:89–117
58 2 Modeling and Optimization of Traditional Supplier Selection

Ustun O, Demirtas EA (2008) An integrated multi-objective decision-making process for

multi-period lot-sizing with supplier selection. Omega 36:509–521
Van der Rhee B, Verma R, Plaschka G (2008) Understanding trade-offs in the supplier selection
process: the role of flexibility, delivery, and value-added services/support. Int J Prod Econ.
doi:10.1016/j.ijpe.2008.07.024
Van Laarhoven PJM, Pedrycz W (1983) A fuzzy extension of Saaty’s priority theory. Fuzzy Sets
Syst 11:199–227
Wang W-P (2010) A fuzzy linguistic computing approach to supplier evaluation. Appl Math
Model 34:3130–3141
Wang J (2012) Do firms’ relationships with principal customers/suppliers affect shareholders’
income? J Corp Finan 18:860–878
Wang YM, Parkan C (2006) Two new approaches for assessing the weights of fuzzy opinions in
group decision analysis. Inf Sci 176:3538–3555
Wang YM, Elhag TMS, Hua ZS (2006) A modified fuzzy logarithmic least squares method for
fuzzy analytic hierarchy process. Fuzzy Sets Syst 157(23):3055–3071
Wang YM, Luo Y, Hua ZS (2008a) On the extent analysis method for fuzzy AHP and its
applications. Eur J Oper Res 186(2):735–747
Wang J-W, Cheng C-H, Cheng HK (2008b) Fuzzy hierarchical TOPSIS for supplier selection.
Appl Soft Comput. doi:10.1016/j.asoc.2008.04.014
Wang S-Y, Chang S-L, Wang R-C (2009) Assessment of supplier performance based on
product-development strategy by applying multi-granularity linguistic term sets. Omega
37:215–226
Weber CA, Current JR, Benton WC (1991) Vendor selection criteria and methods. Eur J Oper Res
50:2–18
Wu M (2007) Topsis-AHP simulation model and its application to supply chain management.
World J Model Simul 3(3):196–201
Wu D, Olson DL (2008) Supply chain risk, simulation, and vendor selection. Int J Prod Econ
114:646–655
Wu WY, Shih H-A, Chan H-C(2008)The analytic network process for partner selection criteria in
strategic alliances. Expert Sys Appl. doi:10.1016/j.eswa.2008.06.049
Wu W-Y, Sukoco BM, Li C-Y, Chen SH (2009) An integrated multi-objective decision-making
process for supplier selection with bundling problem. Expert Syst Appl 36:2327–2337
Wu DD, Zhang Y, Wu D, Olson DL (2010) Fuzzy multi-objective programming for supplier
selection and risk modeling: a possibility approach. Eur J Oper Res 200:774–787
Xia W, Wu Z (2007) Supplier selection with multiple criteria in volume discount environments.
Omega 35:494–504
Xu R, Zhai X (1996) Fuzzy logarithmic least squares ranking method in analytic hierarchy
process. Fuzzy Sets Syst 77:175–190
Yang JL, Chiu HN, Tzeng G-H, Yeh RH (2008) Vendor selection by integrated fuzzy MCDM
techniques with independent and interdependent relationships. Inf Sci 178:4166–4183
Yao F, Hongli L (2007) Information systems outsourcing vendor selection based on analytic
hierarchy process. IEEE ISBN: 1-4244-1312-5/07
Yu J-R, Tsai C-C (2008) A decision framework for supplier rating and purchase allocation: A case
in the semiconductor industry. Comput Ind Eng 55:634–646
Zhang D, Zhang J, Lai K-K, Lu Y (2008) An novel approach to supplier selection based on vague
sets group decision. Expert Syst Appl. doi:10.1016/j.eswa.2008.07.053
Chapter 3
Mass Customization

Mass customization combines economy of scale and economy

of scope to enhance supply chain surplus with demand
aggregation.

3.1 Introduction

Heterogeneous customer base, customer demand for configurable product, signifi-

cantly shorter product life cycle, volatile market demand, and demand for customer
responsiveness continuously pressurize companies how to satiate and retain
potential customer. Mass customization strategies can be an obvious panacea for this
issue. However, does mass customization pay? In the late 1960s and the early 1970s,
buyers were tired of the poor quality, long customer lead time, high price, and
monotonous variety of standard product. Increasing demand for product variety and
limited manufacturing facilities compelled academician to think over alternative
operations strategy and competitive priority (Skinner 1969). They realized that
wining strategy of competitive market was moved from price-driven market to cost-,
quality-, delivery-, and flexibility-driven market. In 1950, Alderson introduced the
postponement. In 1965, first study of postponement was started at Buckling. During
this phase, postponement got appreciation in the industry. However, it failed to win
the corporate. In 1980s, Japanese firms first realized the importance of quality and
flexibility within short period of time Japanese firms got monopoly in the world
market. In 1987, Stan Davis introduced mass customization in his seminal
work-Future Perfect. However, people gave scant attention as because they were
preoccupied with quality issues. On the other hand, another concept popularly
known as customer order decoupling point (CODP) was introduced by Bucklin [21]
in 1965, and it got much attention during this period. In 1993, Pine published “Mass
Customization: the New Frontier in Business Competition.” With the advent of
internet technology at the end of 1990s, much awaited process of customization
started to gain its momentum.
We can classify companies as product-based and customer-based. In product-
based company, standard products are manufactured irrespective of customer choice.

© Springer (India) Pvt. Ltd. 2017 59

K. Mukherjee, Supplier Selection, Studies in Systems, Decision and Control 88,
DOI 10.1007/978-81-322-3700-6_3
60 3 Mass Customization

In customer-based company, products are manufactured/assembled as per the need of

individual customer. For example, Adidas develops running shoe equipped with
magnetic sensors and a microprocessor to measure and control the compressive force
for the wearer (Salvador et al. 2009). How to align product attribute as per the
heterogeneous demand of customer? What would be the appropriate strategy to
implement mass customization effectively? According to Salvador et al. (2009), a
company must possess the following three fundamental capabilities to offer cus-
tomized product to customer:
1. Development of the solution space: Identify the need of potential customers
and render the information to develop/modify product characteristics. Create
virtual prototype as per the “virtual concept testing.” “Virtual concept testing”
was implemented by Adidas to minimize redundant variants.
2. Robust process design: Reuse or recombine available resources to develop
customized product. Extensive need of use FMS, process modularity, and
adaptive human capital. Here, adaptive human capital means skilled/trained
people who can take new tasks.
3. Choice navigation: Too many choices for product variants could create cus-
tomer’s confusion. Is properly known as “paradox of choice.” Assortment
matching is one of the effective approaches to map customer’s need with
available product variants.

3.2 Constraints of Mass Customization

The purpose of mass customization is to satisfy individual customer needs with near
mass production efficiency (Tseng et al. 1996).Tseng and Jiao (2004) identified
three aspects of customizability—design customizability, process customizability,
and perceived value of customized product. Based on their work, FDPT constraints
are indentified—financial constraints, design constraints, process constraints, and
time constraints. In Table 3.1, common expectations of customer, constraints, and

Table 3.1 Common constraints and solutions of mass customization

Expectations of customers Constraints Solutions
Cost: lower is better Financial Reduce the cost of procurement, production, and
constraints logistics as much as possible
Design flexibility and Design Identify the need of customer and allow
product variety: higher is constraints customer to work as co-designer. Such approach
better will integrate customer value in right amount
with product
Process flexibility: higher Process Standardize processes and use advanced
is better constraints manufacturing technology such as FMS to
enhance the process flexibility
Time to offer customized Time Reduce total lead time
product: lower is better constraints
3.2 Constraints of Mass Customization 61

their corresponding solutions are shown to implement effective strategies of mass

customization.

3.3 Postponement

Significance of using postponement strategy is to reap benefits of leagile concept

(Van Hoek 2000).The postponement strategy uses demand aggregation to reduce
the difference between high demand and low demand (Cheng, p. 3). Effective
implementation of postponement strategy can reduce process complexity, and
demand and supply uncertainties. There are four basic categories of postponement
strategies, namely pull postponement, logistics postponement, form postponement
(Lee 1998), and price postponement (van Mieghem and Dada 1999). Production
postponement, on the other hand, encompasses first three postponement strategies
stated above (van Mieghem and Dada 1999). Bowersox and Closs (1996) stated
three different postponement strategies—time postponement, place postponement,
and form postponement.

3.4 Sourcing Postponement—A New Kind

of Postponement Strategy

Procure when required

Sourcing postponement is useful for complex product design environment where

complex customized product is prepared from base product by adding auxiliary
components, as shown in Fig. 3.1. Company can purchase all components at a time
to get price discount; however, it could create difficulties in inventory management.
On the other hand, fluctuation of demand at the downstream could create stock out
or over stock situation of components at middle stream or far upstream. In such
situation, one strategic approach could be to purchase raw material/components
from supplier and manufacture base product as per aggregate forecast and procure
auxiliary components at the retailer site after receiving customer order. Such
sharing of inventories at manufacturer and retailer site could reduce the risk of stock
out/over stock, WIP inventory, and customer waiting time.
62 3 Mass Customization

Fig. 3.1 Sourcing postponement

3.5 Advantages of Postponement Strategy

Both the philosophy of postponement and the principle of just in time

(JIT) emphasize procurement strategies to mitigate upstream supply chain risk.
Effective implement of postponement strategy could save time, improve quality,
and reduce cost of product. Efficiency of postponement strategy depends solely on
product family architecture (PFA) as this hierarchical structure of product family
integrates economy of scales and economy of cost. Postponement strategy sim-
plifies forecasting of product as product remains in generic form instead of finished
form. Therefore, internal variability of product is drastically reduced without
spoiling the external variability of product. Modular characteristics of product not
only reduce the cost production but also enhance outsource capability of auxiliary
processes and speed up the process of new product development. Van Hoek (2001)
identified that postponement strategy reduces uncertainties by delaying finalization
of products, increase product variety with product family architecture (PFA), and
reduce complexity of supply chain. In traditional supply chain, lead time increases,
variety brings more obsolescence risks and more uncertainties due to push-type
supply chain (Van Hoek 2001). Integration of push-type and pull-type supply chain
offers an alternative way to tackle such issues with product differentiation or
delayed product finalization. However, product differentiation does not guarantee to
mitigate all expected risks that may develop due product complexity. A large
product variety may create several problems for the company. A well-defined
product family architecture is highly desired in this regard. Modular approach for
manufacturing can be expected as one of the best ways to increase product variety.
3.5 Advantages of Postponement Strategy 63

Responsive, reliability, resilience, and relationship—the 4Rs are the basis of

successful logistic and supply chain management of any business (Christopher
2005, p. 38). Responsiveness gives the ability to a company to respond to cus-
tomers’ demand in shortest possible time. In such fast-changing market place,
agility is the key characteristic of a successful company.
Reliability of product could be improved by appropriate control of process and
quality of material. Reliability of the logistic service could be improved by
increasing the visibility of pipeline. It means that focal company should get more
information about far downstream supply chain.
Resilient supply chain enhances the capability of the company to cope with
uncertain business environment.
Success of business depends upon the effective management of relationship
across the supply chain. Particularly, good relationship between buyer and supplier
could reduce cost, improve the quality of work and product, provide better assur-
ance of delivery of raw material/part/sub-assembly/assembly, and prepare a niche
for innovative work (Table 3.2)
Postponement integrates aforesaid 4Rs to increase responsiveness of the supply
chain. For instance, the use of modular structure of product provides better control
of process and quality of material at sub-assembly/component level.

3.6 Drivers of Postponement Strategy

According to Lee (1998), four enablers are identified as follows:

1. Modularity;
2. Design for postponement;
3. Supply chain collaboration; and
4. Associated costs.
Apart from the above, the following prerequisites for postponement are identi-
fied as follows:
1. Extensive use of information technology to develop healthy relationship
between buyer and supplier;
2. Reduce lead time, particularly after postponement point so that customer waiting
time could be reduced as much possible. This will reduce the chances of
backorder or lost sales (Lee 1996);
3. Simplification of design and process is required to reduce internal complexity of
supply chain; and
4. Postponement is particularly suitable for global supply chain where positioning
of inventory, distribution mode, and production facilities are the critical success
factors for any organization.
64 3 Mass Customization

Table 3.2 Categories of postponement strategy

Sl. Type Definition Application
No.
1 Pull In pull postponement or process 1. ABC Bicycle Company, India
postponement postponement (Brown et al. 2. Benetton, an apparel
2000), decoupling point moves manufacturer, delays its color
upstream in the supply chain to dyeing process until orders are
make forecast easier (Lee 1998) received (Lee and Tang 1998)
2 Logistics Here, customization takes place 1. Hewlett-Packard produces
postponement far downstream (Lee 1998). generic printers at its factory
Packaging postponement and and distributes them to local
labeling postponement (Twede distribution centers, where
et al. 2000) or branding auxiliary components are
postponement (Ackerman 1997) packed
are some of the approaches of 2. All products of IKEA are
logistic postponement assembled at customers site
3 Form Form postponement or product 1. Brown et al. (2000) applied
postponement postponement (Brown et al. form postponement in a
2000) opts to combine semiconductor company
standardized components with (Xilinx) where standard ICs are
auxiliary components to achieve developed and later they are
high customization (Lee 1998 configured to have customized
and Brown et al. 2000) features as per demand
4 Price Van Mieghem and Dada (1999) 1. Bank of China (BOC) Hong
postponement defined price postponement Kong used a price
from economic and marketing postponement strategy in July
perspectives. Here, selling price 2002
of the product is fixed after
receiving customer order
5 Time Customer order initiates the 1. Dell computer: Computers are
postponement forward movement of goods assembled after the orders are
placed by customers
6 Place Goods are distributed after 1. Hindustan Unilever Limited,
postponement receiving the customer order India’s largest consumer
from a central location product company, supplies their
products to modern trade
segment (organized retail
sector), general trade segment
and rural markets by using place
postponement strategy

3.7 Customer Order Decoupling Point (CODP)

Time-based competition, short product life cycle, volatile market, and heteroge-
neous customer demand need optimum balance between limited resources of any
organization and level of satisfaction of customer to yield desired supply chain
surplus. CODP is one of such strategies to balance decision under certainty and
uncertainty concerning customer demand (Rudberg and Wikner 2004). According
3.7 Customer Order Decoupling Point (CODP) 65

Table 3.3 Manufacturing strategy and degree of complexity

Type of Benefit Degree of
Strategy complexity
MTS Shortest customer waiting time
ATO Reduced WIP inventory, product variety is more than
MTS
MTO Long customer waiting time comparing to ATO;
however, more complex product is manufactured
as per the need of customer
ETO Long customer waiting time. Particularly suitable
for very complex product design ex. Ship
manufacturing

to Olhager (2003), who minted this term, CODP is the point after which customized
product is prepared from semi-finished product as per the specification of customer
order. He further mentioned that depending upon the position of CODP along the
supply chain, different manufacturing strategies can be classified as engineer to
order (ETO), make to order (MTO), make to stock (MTS), and assembly to order
(ATO) based on the ratio of production lead time and delivery lead time. According
to Wikner and Rudberg (2005), mass-customization strategies can be classified into
four different categories, as shown in Table 3.3.

3.8 Conclusion

Mass customization is not the panacea for increasing supply chain surplus for any
industry. It needs effective strategy, advanced manufacturing technology, advanced
information and communication technology, and willingness of customer to pay
more for customized product. Customer will pay more for his/her desired product if
standard product fails to give his/her desired utility. Companies usually face
challenges to manufacture customized product as infinite product variety cannot be
developed with their limited resources. Thus, a well-defined product family
architecture (PFA) is required. Four mass customization strategies are discussed in
this chapter, namely MTS, ATO, BTO, and ETO. Companies such as Dell, Nike,
General Motors, and Zara show that with the concept of mass customization they
can reduce WIP inventory without effecting overall lead time.
A brief introduction is also included about customer order decoupling point
(CODP) and drivers of postponement strategy. CODP is used in supply chain to
combine push-type and pull-type strategies. Interested readers are requested to refer
relevant journals in this regard to know in detail about CODP and mass cus-
tomization strategies. Delayed product finalization or postponement is commonly
used to reduce manufacture’s risk, complexity of supply chain, lead time chain and
to enhance the flexibility of supply chain. In this chapter, a new concept of
66 3 Mass Customization

procurement is mentioned in Sect. 3.5, “Sourcing Postponement,” to reduce the risk

between manufacturer and retailers. Such strategy may reduce WIP inventory, risk
for stock out/over stock, and also long waiting time of customer to get his/her
product. Sourcing postponement is based on the concept of JIT and useful for ATO
industries.

References

Ackerman KB (1997) Postponement, practical handbook of warehousing. Chapman and Hall, New
York
Allee V (2000) Reconfiguring the value network. J Bus Strategy 21(4)
Bowersox DJ, Closs DJ (1996) Logistical management: the integrated supply chain process.
McGraw-Hill, New York
Brown AO, Lee HL, Petrakian R (2000) Xilinx improves its semiconductor supply chain using
product and process postponement. Interfaces 30(4):65–80
Christopher M (2005) Logistics and supply chain management. Prentice Hall Publication
Cheng TCE, Li J, Wan CLJ, Wang S Postponement strategies in supply chain management.
Springer 143
Kotha S (1995) Mass customization: implementing the emerging paradigm for competitive
advantage. Strateg Manag J 16(Special Issue):21–42
Lee HL (1996) Effective inventory and service management through product and process redesign.
Oper Res 44(1):151–159
Lee HL (1998) Postponement for mass customization: satisfying customer demands for
tailor-made products. In: Gattorna (ed) Strategic supply chain alignment. Gower, England
Lee HL, Tang CS (1998) Variability reduction through operations reversal. Manage Sci 44
(2):162–172
Olhager J (2003) Strategic positioning of the order penetration point. Int J Prod Econ 85:319–329
Pine II, Joseph B (1993) Mass customization: the new frontier in business competition. Harvard
Business School Press, Boston, Massachusetts
Rudberg M, Wikner J (2004) Mass customization in terms of the customer order decoupling point.
Prod Plann Control 15(4):445–448
Salvador F, Holan PM, Piller F (2009) Cracking the code of mass customization. MIT Solan
Manage Rev 50(3) http://sloanreview.mit.edu/article/cracking-the-code-of-mass-customization/
Tseng MM, Jiao J, Merchant ME (1996) Design for mass customization. CIRP Ann-Manufact
Technol 45(1):153–156
Tseng MM, Jiao J (2004) Customizability analysis in design for mass customization. Comput
Aided Des. doi:10.1016/j.cad.2003.09.012
Twede D, Clarke RH, Tait JA (2000) Packaging postponement: a global packaging strategy.
Packag Technol Sci 13(3):105–115
Van Hoek RI (2000) The thesis of leagility revisited. Int J Agile Manage Syst 2:196–201
Van Hoek RI (2001) The rediscovery of postponement a literature review and directions for
research. J Oper Manage 19:161–184
Van Mieghem JA, Dada M (1999) Price versus production postponement: capacity and
competition. Manage Sci 45(12):1631–1649
Wikner J, Rudberg M (2005) Integrating production and engineering perspective on the customer
order decoupling point. Int J Oper Prod Manage 25(7):623–641
Zipkin P (2001) The limits of mass customization. Solan Manage Rev 42(3):81–87
Chapter 4
Modeling and Optimization of Strategic
Sustainable Sourcing

4.1 Introduction

All organizations—but especially large companies and public sector

organizations—should use procurement as a way of encouraging those in supply
chain to improve environmental performance.
UK Round Table on Sustainable development
Second Annual Report, London, 1997.

The word sustainability is derived from Latin word sustinere. It means to endure
or maintain. The concept of sustainability is introduced in the report ‘Our Common
Future’ of the UN commission Brundtland in 1987. According to the definition of
sustainable development of the commission Brundtland, sustainable development is
the development that meets the needs of the present without spoiling the ability of
next generations to achieve their own needs. With the advent of new technologies,
man enters the world of luxury and comfort by sacrificing the nature. Unfortunately,
ostentatious display of hedonistic society allures people and compels them to forget
about divine reality. Present bears the seed of future. Today, we are facing soaring
temperature, air pollution, water pollution, etc. because of the selfish activities of
mankind. It is crystal clear that creation of wealth should be integrated with
eco-ethical lifestyle of human being. Today, companies are augmenting the practice
of corporate sustainability and co-creation of value. Traditional supply chain deals
with man, money, and material (3M). On the other hand, green supply chain deals
with man, money, material, and environment (3Me). Finally, sustainable supply
chain deals with man, money, material, environment, and society (3MeS).

© Springer (India) Pvt. Ltd. 2017 67

K. Mukherjee, Supplier Selection, Studies in Systems, Decision and Control 88,
DOI 10.1007/978-81-322-3700-6_4
68 4 Modeling and Optimization of Strategic Sustainable Sourcing

4.2 Literature Review

Do procurement policies need to change and if so how? What sort of procurement

policies are required to achieve sustainability or green procurement status? To find
the answers of aforesaid questions, author searched thoroughly different literature.
Literature review shows that before 1980s, purchasing function got scant attention
from corporate sector. It was considered as simple clerical job. Later on, people
understood the very essence of procurement and tried their best to use it as mere
tool to reduce overall cost of procurement. Such myopic view compelled several
academicians to explore hidden benefits of procurements process. At the end of
1980s, clean technologies were introduced for reducing the environmental impact.
At the beginning of 1990s, enterprises changed their operating procedures to
introduce eco-auditing frameworks to rectify products and services (Franke 1995).
Recently, environmental conscious firms, mainly large firms are organizing envi-
ronmental programs to alter their supply chain (Gupta 1995). For instance, Sony’s
procurement policy promotes ‘QCDS + E’, i.e., quality, cost, delivery, service, and
environment (Handfield et al. 2002). Brink et al. (1998) developed a system called
‘ECO-QUEST’ to prepare questionnaire for suppliers to self-audit their environ-
mental performance and gives them some solutions to improve environmental
profile of their product. Noci (1997) suggested a preliminary framework that
identifies measures for assessing environmental performance but gave little atten-
tion on environmental cost data. Because of such cumulative efforts, today’s pur-
chasing managers are working as strategic sourcing decision makers to offer better
products at lower cost, better quality, and with greater flexibility. Strategically
managed long-term relationships with key suppliers have a positive impact on the
firm’s financial performance and intensity of buyer–supplier coordination (Carr and
Pearson 1999; De Toni and Nassimbeni 1999). Managers and policy makers are
gradually realizing that purchasing can redo corporate function/performance along
environmental dimension (Handfield et al. 2002). Procurement of any company can
affect environment directly and indirectly (Handfield et al. 2002). Direct effect of
product is the wastage of product during transportation, storage, processing, use, or
disposal. When a company is purchasing the product, it is indirectly integrating
environmental effect of product with its value chain. Awareness of green pro-
curement is germinated from several new happenings, such as the follows:
1. Introduction of ISO 14000 certification standard which focuses the firm’s
environmental management system (EMS). Procurement is one of the key assets
of ISO 14000 to control disposal as well as procurement of material.
2. Pressure of government agencies like Environmental Protection Agencies
(EPA) to control waste reduction.
Dickson (1966) in his seminal work proposed 23 criteria for supplier selection.
Pareto analysis is further conducted to find most cited criteria, as shown in Fig. 4.1.
Among 23 criteria, some criteria such as performance history, reputation and
position in history, impression and amount of past business are very closely related.
4.2 Literature Review 69

Fig. 4.1 Pareto analysis of Dickson’s 23 criteria

Higher amount of past business shows companies market share as well as its
reputation. On the other hand, amount of past business is a part of performance
history. However, Dickson’s 23 criteria gave a ready-made platform for academi-
cian as well as industrial practitioners. Out of the 23 criteria, only packaging ability,
warranties and claim policies, and technical capabilities are closely related to green
supply chain. Moreover, Dickson gave an indication of sustainability through labor
relations record which is the social responsibility of supplier. Question arises about
the viability of Dickson’s 23 criteria for green supplier selection. Can we use the
same 23 criteria for green procurement process? To find the answer of this question,
author conducted a survey of recent literature, as shown in Table 4.1. Methods such
as AHP, CBR, DELPHI, DEMATEL, gray entropy, etc. are used to trade off
palpable and non-palpable criteria of supplier selection. Majority of the research
works on green supplier selection are based on automotive industry, apparel
industry, electronic industry, or closely related sectors. It is clear that green pro-
curement is not a mere optimization or decision-making problem. It is a strategy to
integrate mission and vision of any company to broaden the way of sustainability.
Humrhreys (2003) proposed a hierarchical model of supplier selection for
reversed supply chain. A brief analysis of his model with critical remarks is shown
in Table 4.2.
70 4 Modeling and Optimization of Strategic Sustainable Sourcing

Table 4.1 Literature review of green supplier selection

Sl. No. Author Year Criteria Method
1 Robert Handfield, 2002 Product attributes, waste management, AHP, Delphi
Steven V. Walton, Robert labeling/certification, packaging/reverse
Sroufe, Steven A. Melnyk logistics, compliance to government
regulations, environmental programs at the
supplier facilities.
2 Paul Humphreys, Ronan 2003 Environmental competencies, environmental CBR and
McIvor, Felix Chan management system, design for multi-attribute
environment, green image, management analysis
competencies, environmental costs
3 P.K. Humphreys, Y.K. 2003 —do— Multi-attribute
Wong, F.T.S. Chan analysis
4 YANG Yuzhong, WU 2007 Quality, price, capability, service, Grey entropy
Liyun environment protection, management
system, and credit standing
5 Satish Nukala and 2007 Quality, on-time delivery, proximity, Taguchi loss
Surendra M. Gupta cultural and strategic issues function, AHP,
fuzzy.
6 C. Y. Chiou, C. W. Hsu, W. 2008 Supply chain base: Cost, quality, delivery, Fuzzy analytic
Y. Hwang service performance, hierarchy process
Green Competencies: green material
selection, cleaner production technologies,
reduced green packaging, reverse logistics,
Environmental management systems:
ISO-14001 certification, Eco-labeling,
supplier environmental evaluation,
environmental management information
system,
Environmental performance: green design,
use of toxic/restricted substances,
Corporate social responsibility:
Public disclosure of environmental record,
green image, relationship with stakeholders,
Contributions to community,
Risk Factor:
Government and local rules/regulations,
Political stability,
Supplier’s credit, Internal management
system
7 YU-ZHONG YANG, 2008 Quality, price, capability, service, and Extensible
LI-YUN WU environment protection, management synthetic
system and credit standing evaluation
8 Amy H.I. Lee, He-Yau 2009 Quality: quality-related certificates, Fuzzy set theory,
Kang, Chang-Fu Hsu, capability of quality management, capability AHP
Hsiao-Chu Hung of handling abnormal quality,
Finance: Past finance performance, stability
of finance,
Price,
Organization: attitude of managers, future
strategy direction, degree of strategic
cooperation,
(continued)
4.2 Literature Review 71

Table 4.1 (continued)

Sl. No. Author Year Criteria Method
Technology capability: capacity, technology
level, capability of R&D, capability of
design, capability of preventing pollution,
Service: credible delivery, capability of
delivery on time, capability of technology
support, flexibility,
Total product life cycle cost: cost of
supplied components,
Green Image: green purchase trend of
customer,
Pollution control:
Use of harmful materials.
Environment management:
Environment-related certificates, internal
control process
9 GE Yan 2009 Environment, quality, operational capacity, AHP and GA
price level, service level,
10 Cao Qingkui, 2009 Product information: delivery time, price, Rough sets and
Ruan Junhu quality, AHP
Geographical position: delivery location,
delivery way,
Green information: resource utilization ratio,
Resource recovery ratio, resource types,
Resource conservation, environmental
pollution, environmental governance,
Cooperative history: cooperation duration,
cooperation condition,
Service: Response time, service level
11 Mehrdad Agha Mohammad — Price, quality, services, and environmental AHP and
Ali Kermani, Amir Malaei, criteria multi-objective
Marzieh Nasiri optimization
(presented
mathematical
model for
sustainability)
12 Alireza Iirajpour, Mehdi 2012 Design management—process/product DEMATEL
Hajimirza, Mahdi Golsefid changeability, ability to design and recycle
Alavi, Sajad Kazemi Technology management—technology
level, ability of R&D, clean technology,
Customer management—controlled use of
hazardous material, energy consumption,
green packaging
Strategic management—green image, green
planning, supervision on and following
environmental laws, social responsibility
72 4 Modeling and Optimization of Strategic Sustainable Sourcing

Table 4.2 Critical remarks for Humrhreys model for reversed supply chain
Critical remarks
Quantitative Pollutant Solid waste Saving energy means saving money. Thus, energy should be
cost costs/effect Chemical waste recovered from used product. However, care should be taken to
reduce exhaust to air, water, and soil as much as possible
Air emission
Water waste disposal
Energy Optimum use of energy should be considered from the
procurement of raw material to shipment to end users
Improvement Buying Select material as per product specification to reduce
cost environmental free environmental cost. In this regard, author suggests using
material SimaPRO or Gabi 4 software to calculate the energy associated
with each raw material
Buying new Earlier trend of manufacturing gave highest priority to low
environmental free cost. Environment consciousness gives highest priority to cost
technology and environmental burden associated with the
manufacturing/assembling of product
Redesign of product It is basically amalgamation of design for assembly, design for
disassembly and environmental conscious manufacturing
process
Staff training It is required to make employee environment conscious
Recycling Every recycling is associated with profit. To maximize the
profit, company should reuse the complete product or partly
disassemble the product as the cost of disassembly initially
decreases and then increases with the degree of disassembly.
Therefore, author strongly suggests that each product should be
associated with optimum disassembly plan
Management Senior management
competencies support
Environment
partners
Training Training gives required competencies to employee. It also
creates an awareness of eco-friendly product
Information Reversed supply chain is very complex in nature. Like forward
exchange supply chain, sharing of information is also required for
reverse supply chain
Green image Customer’s Green market share will increase if customer is willing to
purchasing retention purchase green/eco-friendly product. Government policies also
Green market share play a major role to control green market share of any product

Stakeholders
relationship
Design for Recycle Complexity of reverse supply chain is more than forward
environment Re-use supply chain. However, effective strategies for reverse supply
chain can reduce cost and environmental burden associated
Remanufacturing
with product
Disassembly
Disposal
Environmental Environmental Such factors are considered to measure the capability of
management policies supplier to supply eco-friendly product/parts/sub-assembly/raw
systems Environmental material to the customer.
planning
Implement and
operation
ISO 14001
certification
(continued)
4.2 Literature Review 73

Table 4.2 (continued)

Critical remarks
Environmental Clean technology Environmental management systems can be considered as
competencies availability cause and environmental competencies can be considered as its
Use of environment effect
friendly materials
Pollution reduction
capability
Returns handling
capability

4.2.1 Viability of Dickson’s 23 Criteria for Green Supply

Chain

Dickson’s 23 criteria have been considered as ready-made reference for forward

supply chain. However, author suggests that Dickson’s criteria can be used for
reverse supply chain with little modification. Table 4.3 shows the list of few
unchanged, changed, and new entrant criteria.

4.3 Economical Aspects of Reverse Supply Chain

Does reverse supply chain pay? One of the main advantages of reverse supply chain
is to enhance supply chain surplus by mixing reused parts/products with new
product. It saves time of manufacturing, reduce total cost of procurement but

Table 4.3 New criteria for green supplier selection

Sl. Dickson’s criteria Green supplier selection criteria
No.
Changed criteria
1 Management and Environmental management system, management
organization competencies
2 Training aids Training aids
3 Desire for business Green image
4 Technical capability Ability of R&D, Cleaner production technology
5 Packaging ability Green packaging ability
6 Procedural compliance Compliance to government regulations for eco product
New criteria
1 New entry Corporate social responsibility
2 New entry Design for environment
Unchanged criteria
1 Quality Quality
2 Delivery Delivery
3 Price Price
74 4 Modeling and Optimization of Strategic Sustainable Sourcing

enhances the complexity of supply chain. It also raises questions of quality to reuse
dumped parts/products. Thus, an optimal balance is required.

4.3.1 Disassembly Cost

A disposed product is composed of several parts, each of which may have some
material value or reuse. The purpose of the disassembly operation is to retrieve
valuable parts and then forward those to the appropriate recycling streams (Das and
Yedlarajiah 2002). Output of the disassembly operation could be waste material,
recyclable material, or parts for direct use. In general assembled product is disas-
sembled either manually or through automatic process. For each customized product,
a suitable disassembly plan should be developed to disassemble the product with less
time to reduce the disassemble cost. In this regard, various researchers have developed
various methods to calculate disassembly cost. Special emphasize has been given to
the design of fasteners. Ishii et al. (1994) proposed the disassembly cost DC as

X
l X
m X
n
DC ¼ Ci þ ðfn  FÞj þ ðpn  pÞk
i¼1 j¼0 k¼0

Ci time to remove component

FJ time to remove fastener
fnJ number of fastener associated with one link
pnk number of process points
pk time to remove or undue process
l total number of components in the system
m total number of links with fastener
n total number of links with fastening process.
The calculated disassembly time associated with each product should be multiplied
with labor wages per unit time. Obvious question arises about the importance of
disassembly process in environmental conscious manufacturing. In this regard, Das
and Mathew (1999) found that there are eight common class of material output bins
that are maintained in electronics disassembly facilities; these are ferrous metals
(steel), non-ferrous metals (aluminum, copper), sources of precious metals (gold,
silver, palladium), packaging materials (styrofoam), glass, ceramics, plastics, haz-
ardous parts, and paper. Hence by integrating disassembly process, companies could
reduce their production cost and thereby maximize their profit. Das et al.
(2000) propose a scheme for the estimation of disassembly effort index (DEI) which
is the function of seven factors—(a) time (b) tools (c) fixture (d) access (e) instruct
(f) hazard, and (g) force required to disassemble a product. The DEI scores ranges
from zero upward, with zero indicating no effects. Thus, disassembly is a crucial factor
in implementing environmental conscious manufacturing (Zeid and Gupta 2002).
4.3 Economical Aspects of Reverse Supply Chain 75

4.3.2 Recycling Profit

A product recycling module should be developed in this regard to calculate the

recycling profit of the disposed product. Zhang et al. (2004) calculated the profit
due to recycling as PR

PR ¼ qp  wt  ðdc þ ic þ rc þ tcÞ

where
qp quote price in the market
wt weight of the material recovered
dc disassembly cost
ic inspection cost
rc recycling cost
tc transportation cost.
The product recycling module will interact with LCA module and depending
upon the post consumer use of the product (i.e., reuse, recycling, incineration and
landfill), LCA calculation will be modified.

4.3.3 Optimum Level of Disassembly

Complete disassembly is not profitable as the marginal cost benefit for complete
disassembly becomes less attractive (Das and Naik 2002; Gungor and Gupta 1999).
Total disassembly cost increases with the level of disassembly. The lowest total
disassembly cost determines the termination of the disassembly process (Zhang
et al. 2004). A proper disassembly process planning (DPP) is required to disas-
semble the product up to a certain level to reduce the cost of disassembly (Gungor
and Gupta 1999). Researchers defined the cost of disassembly in several ways. For
instance, disassembly cost is composed of direct labor cost, disassembly effort cost,
and the logistic cost (Das and Naik 2002). According to Zhang et al. (2004), total
cost of disassembly is composed of disassembly cost which includes labor, and
tooling cost; material reprocessing cost etc.

4.4 Sustainable Mass Customized System

Author suggests a new way of mass customization to encompass the concept of

sustainability with the concept of mass customization. Sustainable mass customized
system interacts with planet (the environment) to reduce the emission of green
house gases (GHG), people (the customer as well employee) to maintain a healthy
social life, product with customizable characteristics to give the desired utility to
76 4 Modeling and Optimization of Strategic Sustainable Sourcing

individual customer and prosperity (the economic growth) to the organization.

Therefore, sustainable mass customized system can be considered as a system of
people, planet, product, and their combined prosperity.

4.4.1 Stages and Enablers of Sustainable Mass Customized

System

Green procurement, environmental conscious manufacturing (ECM), green product

design (GPD), reverse logistics, reuse and recycle, etc. are commonly used to make
green supply chain. It is mentioned in the bottom layer of triangle in Fig. 4.2. For
instance, company should look after the welfare of employee’s family. Such
policies enhance the integrity of the employee with his company. It is mentioned in
middle layer of triangle. After achieving the sustainable supply chain, company
should proceed for auditing and bench marking. It is the top layer of sustainable
triangle. Considering the need of sustainable supply chain, the author suggests that
mass customization strategies should incorporate sustainable issues to make prof-
itable, eco-friendly, and eco-ethical ambiance for the organization.

4.4.2 Sourcing Strategies for Sustainable Mass Customized

System

Any complex product can be developed through modular approach. Usually,

companies diversified their portfolio of product to manufacturer different brand to
enhance profit margin through economy of scopes. For instance, each and everyday

Green Product Design

Product Family
Green procurement Architecture (PFA)
Social Issues Modularity &
Commonality
Design for Disassembly
Environmental
(DFD)
Conscious
Manufacturing Auditing
Process

Embed sustainability
Reuse and Recycle into
supply chain Optimization of
logistic process
to reduce GHG
Use core competencies to reduce carbon foot print
emissions

Fig. 4.2 Six enablers of 3-stage sustainable mass customized system

4.4 Sustainable Mass Customized System 77

users are receiving cell phone with different varieties. It is reducing product life
cycle, reducing time to introduce new product in market, and at the same time
triggering tough competition among competitors. To survive in such situation,
companies prefer to develop modular product. A modular product composed of
several standard parts/components. Standard parts/components can be manufac-
tured as per the aggregated demand and auxiliary parts can be developed and
assembled with standard or base parts on receipt of the customer’s request. This
approach can reduce the complexity of inventory, reduce the chances of over stock
or under stock, and enhances the responsiveness of supply chain. It is an integrated
approach of push and pull strategy of supply chain. Therefore, base product can be
made with aggregated demand and auxiliary product can be made with stochastic
demand. This 2-stage procurement problem can be solved with intuitionistic fuzzy
AHP (IF-AHP) and multi-objective genetic algorithm (MOGA).

4.5 Mathematical Model for SPS

China has emerged as a favored low-cost manufacturing spot among electronics

companies because of the presence of competent suppliers and contract electronics
manufacturers, as well as the quality of its infrastructure (roads, electricity, stable
political situation, etc.). Although electronics assembly may take only one to two
days, transporting goods by ship between China and Europe takes three weeks.
Such delay in service can create several problems. One solution to this problem is to
create standard products in the low-cost production center but does final configu-
ration and packaging at a distribution point closer to the customer. Algorithm of this
two-stage sustainable procurement process is mentioned below:
Stage-1: Selection of suppliers for base product with aggregated demand
1. Prepare intuitionistic fuzzy pairwise comparison matrix for each
criterion and alternative.
2. Calculate score (Si) of all intuitionistic fuzzy number with any of
the given formula.
   
SI Xij ¼ lij  mij where SI Xij 2 ½1; 1 ð4:1Þ
   
SII Xij ¼ lij  mij  pij where SII Xij 2 ½1; 1 ð4:2Þ
     
SIII Xij ¼ lij  mij þ pij =2 where SIII Xij 2 ½0:5; 1 ð4:3Þ

SI(Xij) is useful for simple decision-making problem, and SII(Xij)

and SIII(Xij) are useful for complex decision-making problem (Chen
2011).
3. Calculate normalized score matrix with the given formula
78 4 Modeling and Optimization of Strategic Sustainable Sourcing

 
sij  min sij
j
sij ¼     ð4:4Þ
max sij  min sij
j

4. Normalize each row of S with the given formula

 sij
s ¼ Pn 8 i ¼ 1; 2; 3. . .m and j ¼ 1; 2; 3. . .n ð4:5Þ
j¼1 sij

5. Calculate entropy w.r.t ith attribute with the given formula

1 Xn  
Ei ¼  s ij ln s ij ð4:6Þ
ln n j¼1

6. Calculate entropy weight wi with the given formula

1  Ei
wi ¼ Pm ð4:7Þ
i¼1 ð1  Ei Þ

7. Calculate normalized entropy weight to rank criteria or alternative

with the given formula

Wi
wi ¼ Pm ð4:8Þ
i¼1 Wi

8. Prepare weighted order allocation model with the priority obtained

from step-7. The following notations are used to prepare
multi-objective order allocation model mentioned below:

Cij Purchase cost of product j from ith supplier

TCij Transportation cost of product j from ith supplier
CCi Overall performance index of ith supplier
COij Ordering cost of jth product from ith supplier
ai Reliability of ith supplier
Xij Order quantity of product j to ith supplier
LDij Percentage late delivery of product j from ith supplier
Vij Capacity of ith supplier for jth product
Dj Demand for jth product
Hj Handling cost per ton of product j
B Total allocated budget for all products
kj Percentage of jth product disposed at disposal site
nj Level of disassembly of jth product at disassembly site
4.5 Mathematical Model for SPS 79

bi GHG emission factor per weight unit distance due to use of

transportation mode
di Distance of ith supplier from manufacturing/retailing site
alpha Probability value of chance constraint
i 1,2,3…n of suppliers
j 1,2,3…m no of products

Total cost of purchase (TCP) consists of purchase, transportation,

order/setup, and holding costs.
Min TCP:

X
n X
m n X
X m X
m X
n n X
X m
Cij Xij þ TCij Xij þ Hj Xi þ COij Xij
i¼1 j¼1 i¼1 j¼1 j¼1 i¼1 i¼1 j¼1

ð4:9Þ

Maximize TVRP:

X
n X
m
ai CCi Xij ð4:10Þ
i¼1 j¼1

Minimize number of late deliveries:

X
n X
m
LDij Xij ð4:11Þ
i¼1 j¼1

Minimize GHG emission for inbound logistics:

X
n X
m
bi di Xij ð4:12Þ
i¼1 j¼1

Subject to
Capacity constraint:

X
m X
n m X
X n
Xij  Vij for i ¼ 1; 2; 3. . .n and
j¼1 i¼1 j¼1 i¼1 ð4:13Þ
j ¼ 1; 2; 3. . .m

Demand constraint:
m X
X n X
m X
m
Xij ¼ Dj  ð1  kj Þnj Dj ð4:14Þ
j¼1 i¼1 j¼1 j¼1
80 4 Modeling and Optimization of Strategic Sustainable Sourcing

Cost constraint:

X
m X
n
Cij Xij  B ð4:15Þ
j¼1 i¼1

Non-negativity constraint:

Xij  0 for i ¼ 1; 2; 3. . .n and j ¼ 1; 2; 3. . .m: ð4:16Þ

Stage-II: Selection of suppliers for auxiliary product with stochastic demand

9. Repeat steps 1–7 mentioned in stage-1 to find the priority of each
supplier and form the weighted order allocation model mentioned
below:
Minimize TCP:

X
n X
m X
m X
n n X
X m
Cij Xij þ Hj Xi þ COij Xij ð4:17Þ
i¼1 j¼1 j¼1 i¼1 i¼1 j¼1

Maximize TVRP:

X
n X
m
ai CCi Xij ð4:18Þ
i¼1 j¼1

Minimize number of late deliveries:

X
n X
m
LDij Xij ð4:19Þ
i¼1 j¼1

Subject to
10. Demand constraint:
hXn i
pr i¼1
x ij  Dj  alpha ð4:20Þ

Above chance constraint can be converted to a deterministic con-

straint as follows (Guan et al. 2007):
n X
X m
11: Xij  lj þ U1 ðalphaÞrj ð4:21Þ
i¼1 j¼1
12. where
4.5 Mathematical Model for SPS 81

ZX
1 ðXlÞ2
Uð X Þ ¼ pffiffiffiffiffiffi e 2r2 dX ¼ alpha; U1 ðalphaÞ ¼ X ð4:22Þ
2pr
1

13. Capacity constraint:

X
m X
n m X
X n
Xij  Vij ð4:23Þ
j¼1 i¼1 j¼1 i¼1

14. Cost constraint:

X
m X
n
Cij Xij  B ð4:24Þ
j¼1 i¼1

15. Non-negativity constraint:

Xij  0 for i ¼ 1; 2; 3. . .n and j ¼ 1; 2; 3. . .m: ð4:25Þ

4.6 Decision Support System for Strategic Sustainable

Sourcing in Volume Discount Environment

In the late 1970s, decision support tools were developed in DOS and UNIX
environment. In early 1990s, it was developed in Windows environment. With the
development of data processing, microprocessor and networking technology DSS
tools became more flexible, user friendly, and be able to solve complex
decision-making problem with speed and accuracy. The development of wireless
network and cell phones made next big leap of DSS tools. DSS used for supplier
selection commonly uses a database, a knowledge base, a graphical user interface, a
optimization tool, and other analytical tools to make decision-making process more
flexible as well as presentable. Research on sustainable supplier selection is still in
infancy. Only the work of Amindoust et al. (2012) has been identified on fuzzy
inference system for sustainable supplier selection. They used total 132 rules to
select sustainable suppliers. They also claimed robustness and flexibility of their
proposed model. But robustness of their model may not true as defuzzification
value changes for all membership functions except triangular membership func-
tions. Robustness of the proposed model of Amindoust et al. (2012) is true as long
as users are using triangular fuzzy membership function only. Secondly, they did
not tried to reduce the rule base of their proposed system. For instance, at stage 1
rule base could be reduced from 16 to only 4 with SVD method. The complicated
output surface at stage 1 of their proposed model can also be simplified as shown in
82 4 Modeling and Optimization of Strategic Sustainable Sourcing

Fig. 4.3 SVD simulation using four rules at stage 1

Fig. 4.3. Membership function at the input can be modified to two odd-shaped
membership functions as shown in Fig. 4.4.
Finally, Amindoust et al. (2012) could enhance the flexibility of their model by
integrating it with any stand-alone or online software to evaluate a large number of
suppliers. In this section, detailed discussion is provided to design and develop
decision support system with FIS. The proposed FIS is developed with intersection
rule configuration (IRC) model for mamdani fuzzy system for 2-input and
1-output at each stage. Triple bottom line approach is used for proposed model, as
shown in Fig. 4.5.
Price, quality, service, delivery, capacity, and past performance are selected for
economic criteria. Environmental competencies, environmental management system,
green packaging ability, and green product design are selected for environmental
criteria. Health and safety, and education are selected for social criteria. All
sub-criteria are considered to be larger-is-better. Here, term price refers the difference
between the highest quoted price and the price quoted by ith supplier. Thus, higher the
price means lower the price quoted by ith supplier. Triangular membership functions,
shown in Figs. 4.6 and 4.7, are selected for input and output in stage 1. Membership
functions shown in the figure are selected for output in stage 2 and for input and output
in stage 3. Mamdani fuzzy inference system consists of four components—fuzzifier,
fuzzy rule base, defuzzifier, and interface engine. Fuzzifier contains different mem-
bership functions, such as triangular fuzzy membership function, trapezoidal

μ(x) 1
0.891

0.1526

Fig. 4.4 Two odd-shaped membership functions for input at stage 1

4.6 Decision Support System for Strategic Sustainable Sourcing … 83

Economic Strategy Second Stage Third Stage

FIS-11
FIS-123

FIS-12
FIS-21

FIS-14
FIS-145
FIS-15
FIS-31
Environmental Strategy
FIS-11

FIS-13
FIS-22
FIS-12

Social Strategy

th
FIS-11

Fig. 4.5 FIS model for sustainable supplier selection

Fig. 4.6 Membership function for stage 1

Fig. 4.7 Membership function for stages 2 and 3

84 4 Modeling and Optimization of Strategic Sustainable Sourcing

Table 4.4 Linguistic terms in stage 1 and stage 2

Slightly preferred (SP) (0, 0, 0.3)
Preferred (P) (0.1, 0.45, 0.8)
Absolutely preferred (AP) (0.7, 1, 1)

Table 4.5 Linguistic terms in stage 2 and stage 3

Very less preferred (VLP) (0, 0, 0.25)
Less preferred (LP) (0.05, 0.25, 0.45)
Low moderately preferred (LMP) (0.25, 0.45, 0.65)
High moderately preferred (HMP) (0.45, 0.65, 0.85)
Very strongly preferred (VSP) (0.65, 0.775, 0.9)
Absolutely preferred (AP) (0.85, 1, 1)

Fig. 4.8 Output surface of FIS for quality and price

membership function, and Gaussian membership function. Defuzzifier includes

center of area method (COA) and bisector of area method (BOA). The linguistic terms
used for proposed FIS are shown in Tables 4.4 and 4.5.
As shown in Fig. 4.8, economic strategy 1 sharply increases with the increase in
quality and price. It means supplier who will offer low cost and better quality is
expected to score well. As shown in Fig. 4.9, overall economic strategy increases
with the increase in economic strategy 1 and 2. It indicates that a supplier can
expect to score well if he offers low cost, better quality, better service, and delivery.
For instance, if a supplier pays more priority to economic strategy 1&2 and less to
economic strategy 3, then he can expect to score more, as shown in Fig. 4.10. This
is one of the flexible options of the proposed FIS as the rule base can be prepared as
per the long-term goal of the company. A simulink model is prepared with constant
input, as shown in Fig. 4.11, to connect all FIS models. Similarly other FIS models
are prepared for environmental and social strategies. As shown in Fig. 4.12, overall
increase obtained from combined strategy is observed for an increase in environ-
mental and social strategies. As shown in Fig. 4.13, a supplier should pay same
4.6 Decision Support System for Strategic Sustainable Sourcing … 85

Fig. 4.9 Output surface of FIS for economic strategies 1 and 2

Fig. 4.10 Rules for overall economic strategy

priority to both environmental and social criteria to score well in environmental and
social strategy. Thus, the proposed FIS shows that low cost, better quality, or better
service is not the market-winning strategy for any supplier to get order. In sus-
tainable procurement process, supplier has to pay more attention on overall
improvement of his performance for each of social, environmental, and economic
criteria. It is crystal clear that if a supplier pays more attention on economic criteria
and pays less on environmental and social criteria, then he can expect to get poor
score, as shown in Fig. 4.14. In Fig. 4.15, detailed simulink model is shown for
proposed cascaded FIS model.
The proposed FIS model is capable to select large number of suppliers. The
proposed model is integrated with VB.NET and SQL server. A simple website is
86 4 Modeling and Optimization of Strategic Sustainable Sourcing

Fig. 4.11 Simulink model for economic strategy

Fig. 4.12 Rule surface for environmental and social strategy

developed to collect potential suppliers’ details to prepare supply base for future
requirements, as shown in Fig. 4.16. Proposed FIS model remains on a standalone
computer. From website, the details of suppliers are stored in .xls file to feed the
simulink model.
Following notations are used to prepare order allocation model as shown below:
Pi Purchase cost of per ton of raw material from ith supplier
CCi Rank of ith supplier
ai Reliability of ith supplier
Xi Order quantity to ith supplier
Vi Capacity of the ith supplier
Dj Demand of jth product in a planning horizon
ri Volume discount given by ith supplier
4.6 Decision Support System for Strategic Sustainable Sourcing … 87

Fig. 4.13 36 rules for environmental and social strategy

P
Minimize total cost of purchase (TCP): ni¼1 ð1  ri ÞPi Xi
where ri is defined as follows:
8
<0 if 0  Pi Xi \10; 000
ri ¼ 10 if 10000  Pi Xi \20; 000
:
20 if Pi Xi  20000

The above ri is based on the model of (Xia and Wu 2007).P

Maximize the total value of reliable purchase (TVRP): ni¼1 ai CCi Xi
Subject to
Xi  Vi 8 i ¼ 1; 2; 3. . .n
Capacity constraint: P
Demand constraint: ni¼1 Xi ¼ Dj 8i ¼ 1; 2; . . .n and j ¼ 1; 2; . . .m
P
Cost constraint: ni¼1 ð1  ri Þ Pi Xi  3; 000; 000
Non-negativity constraint: Xi  0 8 i ¼ 1; 2; 3. . .n
Each supplier is giving three volume discounts. Three suppliers will give 27 (33)
different combinations of equations (Xia and Wu 2007).
88 4 Modeling and Optimization of Strategic Sustainable Sourcing

Fig. 4.14 Rules for economic, social, and environmental strategies

Example

A company is willing to initiate sustainable procurement process and going to

select suppliers for its product. Volume discount offered by each supplier is shown
in Table 4.6. For sake of simplicity, it is considered that all suppliers are offering
same volume discount. Supplier datasheet is shown in Table 4.7. Demand of the
product in planning horizon is 1200 ton, and total budget is Rs 3 crore. An
appropriate method should be used to select sustainable suppliers.
Proposed cascaded FIS is used to rank suppliers. Only three best suppliers are
selected. Following objective functions and constraints are prepared as per the
supplier datasheet, as shown in Table 4.7.
Minimize total cost of purchase (TCP): ð1  r1 Þ  1500  X1 þ ð1  r2 Þ  1499
X2 þ ð1  r3 Þ  1501  X3
Maximize total value of reliable purchase (TVRP): 0:95  0:5  X1 þ
0:95  0:45  X2 þ 0:96  0:508  X3
Demand constraint: x1 þ x2 þ x3 ¼ 12; 000
Capacity constraint: x1  5000; x2  6000; x3  6000
4.6 Decision Support System for Strategic Sustainable Sourcing … 89

Fig. 4.15 Simulink model for sustainable supplier selection

Fig. 4.16 Web-based sustainable supplier selection process

90 4 Modeling and Optimization of Strategic Sustainable Sourcing

Table 4.6 Volume discount schedule

r Volume discount (in thousand Rs) Percent discount
1 0 to under 10 0
2 10 to under 20 10
3 20 and above 20

Table 4.7 Supplier datasheet

Sl % failure rate of supply Reliability (a = 1 Capacity Total cost
No. (f) −f) (Ton) (Rs/Ton)
1 5 0.95 5000 1500
2 5 0.95 6000 1499
3 4 0.96 6000 1501

Cost constraint: 1500  x1 þ 1499  x2 þ 1501  x3  30; 000; 000

Above multi-objective mixed integer non-linear constrained optimization
(MOMINLP) problem is solved with MATLAB GA solver. Minimum value of
TCP is 1.4396e + 007 and maximum value of TVRP is 5.4341e + 003. Best value
of decision variables are {4.9993e + 03, 5.8943e + 03,1.1064 + e03}. Obtained
pareto front is shown in Fig. 4.17. The average distance measure of the solutions on

Fig. 4.17 Pareto front of TCP and TVRP

4.6 Decision Support System for Strategic Sustainable Sourcing … 91

the pareto front was 0.0105462, and the spread measure of the pareto front was
0.161515.

4.7 Strategic Sourcing of Large Number of Suppliers:

An Illustrative Case

Success begins with right decision. Right decision gives better insight, and better
insight helps to frame effective strategy to mitigate future risk, if any. Strategic
sourcing is a complex decision-making problem, and it demands better insight to
control each parameter at microlevel of a process or system to prevent future risk.
Strategic supplier selection encompasses operational metrics such as cost, quality,
delivery, and other strategic dimensions such as quality management practices,
process capabilities, design and development capabilities, and cost reduction
capabilities into the decision-making process (Talluri and Narasimhan 2004).
Strategic supplier selection demands more effective method to identify the differ-
ence in performance of suppliers across supplier groups to give better insights to
decision makers. Efficiency of cascaded FIS solely depends on various criteria,
namely structure of the membership functions and complexity of the rule base.
Improper selection of membership function may lead to a disaster. Therefore, more
flexible method is required for the strategic selection of large number of suppliers.
Integrated approach of data envelopment analysis (DEA) and non-parametric
test proposed by Talluri and Narasimhan (2004) to cluster supplier for strategic
sourcing is limited to small number of suppliers as their proposed method is time
consuming. In this section, data clustering technique is discussed to cluster sup-
pliers for strategic sourcing. Proposed method is implemented with “R” and
spreadsheet. Cluster analysis (CA) is a statistical technique which is used for the
classification of similar objects into different groups by partitioning of a given data
set into subsets (clusters) to identify common trait of each group or cluster.
Clustering is a method of unsupervised learning and a well-known technique for
statistical data analysis. Clustering algorithms can be classified into various cate-
gories such as hierarchical methods, partitioning techniques, grid-based
methods, density-based methods, and model-based methods. One of the big-
gest challenges of clustering is identifying the optimal number of clusters and the
selection of appropriate clustering method. Thirty indices are identified in this
regard to select optimal value of cluster (Charrad et al. 2014). Out of thirty indices,
five outperform others (Milligan and Cooper 1985). Majority rule was used further
to select optimal number of clusters. The steps of this method are mentioned below
as follows:
1. Define strategic goal of sustainable procurement.
2. Select criteria for sustainable procurement process. Identify cost criteria and
benefit criteria. Benefit criterion is preferred to be larger-is-better except the
92 4 Modeling and Optimization of Strategic Sustainable Sourcing

cost criterion (smaller-is-better). Convert cost criteria as the difference between

highest cost and the cost offered by ith supplier to use it as benefit criteria.
3. Prepare master datasheet from each supplier datasheet with a suitable scale.
Master datasheet is usually prepared with the responses of suppliers. In such
case, it is advisable to conduct the test of hypothesis to confirm that ques-
tionnaire is simple and unambiguous and respondent can answer all without any
difficulty (Talluri and Narasimhan 2004). Conduct data cleaning procedure and
check redundant data. Conducts spend analysis to select major suppliers, if
required.
4. Calculate total score of each supplier w.r.t each criteria. Calculate weighted
total score of each supplier with the priority of each criterion. Rank suppliers
accordingly in descending order to show cross-performance of each supplier
under each criterion.
5. Select optimal number of clusters.
6. Select suitable clustering techniques with cluster validity method.
7. Cluster master data sheet.
8. Interpret each cluster to give better managerial interpretation.
9. Identify best suppliers. Implement appropriate strategy to selected group of
suppliers to achieve strategic goal. Identify key suppliers and prune worst
performer from supply pool. Tier selected suppliers, if required.
10. Select appropriate order allocation model for selected suppliers.

4.7.1 Case Study

To maintain confidentiality of business and secrecy of data, name of the company is

changed and case study is further modified to extend the work of Talluri and
Narasimhan (2004). Company ABC has its manufacturing base in south India and
willing to initiate strategic sustainable procurement process for its new factory
situated in north India. Company ABC has adopted the decentralized procurement
process and highly depends on its local suppliers. Company ABC has operations
plants, R&D facilities, and excellent distribution facilities. The critical objectives of
the Company ABC in procurement and supply management include the following:
• Selection of key suppliers;
• Building long-term relationship with key suppliers;
• Initiating benchmarking program to encourage ineffective suppliers to improve
their performance;
• Identification of worst suppliers to reject them from supply base;
• Initiating strategic sustainable procurement process to fulfill company’s
long-term goal; and
• Initiating strategies to reduce supply base and willing to apply supply base
reduction method.
4.7 Strategic Sourcing of Large Number of Suppliers: An Illustrative Case 93

Company ABC is also interested to implement an effective user friendly method to

select suitable suppliers to fulfill its goal and to achieve it. Company ABC is willing
to start bimonthly meeting with its suppliers to prepare them as per the requirement of
the company. Purchasing department of the company in south India is willing to
adopt supply base reduction method to reduce supply base to increase profit margin
and its new factory in north India is giving more emphasize on the quality of their raw
material for the new product. Therefore, uncertainty of demand is quite expected.
Moreover, traditional spend analysis cannot be used as local suppliers are going to
supply for first time. Overall reduction in supply base can be achieved initially by
rationalizing supply base in each location of the Company ABC. In the next step,
spend analysis as well as integration of common suppliers for both factories could be
done. Such gradual approach of supply base reduction will help new factories of
Company ABC to work effectively with the tempo of local suppliers. Sourcing is a
product-specific process. Commodity strategy is a priori to supplier base reduction
approach. Commodity strategy gives guideline for optimal number of suppliers, type
of suppliers or desired suppliers, desired supplier performance for each component
bought. Thus, Company ABC should develop commodity strategy for its each
product, parts, or components. In the absence of commodity strategy, Company ABC
can use Kraljic matrix which is a 2  2 matrix to compare profit impact with supply
risk. However, Company ABC is willing to start supply base reduction immediately.
Company is also unaware whether supply base reduction approach will bring any
benefit to its stakeholders or not. It is true that reduction in supply base will help
Company ABC to develop long-term relationship with its suppliers and to have more
priority from its suppliers and thereby company can expect to have more purchase
power. But uncertainties associated with the new product demands considerable
number of suppliers to assure availability of raw material for production. Instead of
minimum number of suppliers, company may look for optimal number of suppliers to
have a better result. In order to achieve these goals, company has selected three
metrics and twelve criteria. Instead of sending questionnaire to all suppliers, suppliers
were asked to send detailed supplier datasheet as per the prescribed format of the
company. Company ABC initially screened about 100 suppliers. For the sake of
calculations, only 11 suppliers are considered here. A master datasheet was prepared
with 0-1 scale as shown in Table 4.9. Prior to that, data redundancy check was also
conducted for data cleaning purpose. To select the optimal number of clusters, dif-
ferent indices were used as shown in Table 4.8, Figs. 4.18 and 4.19. Initially, hier-
archical method and partitioning techniques were used with same data, as shown in
Table 4.9, and the best technique was identified as K-means clustering technique for
proposed supplier selection problem by comparing the results.
Table 4.8 Selection of optimal number of clusters
Index Number of clusters
1 Duda 3
2 Calinski and Harabasz 4
3 d index 3
4 Hubert index 3
94 4 Modeling and Optimization of Strategic Sustainable Sourcing

Fig. 4.18 d index

Fig. 4.19 Hubert index

K-means clustering technique categorized 11 suppliers into three different

groups, namely best (B), medium (M), and worst (W) which is to be pruned as
shown in Fig. 4.20.
Only members of the cluster 3 should be selected for order allocation stage.
Performance of cluster 3 members should be benchmarked to encourage
Table 4.9 Master datasheet
Economic strategy Environmental strategy Social
Supp# C1 C2 C3 C4 C5 C6 TS Rank C7 C8 C9 C10 TS Rank C11 C12 TS Rank WS
1 0.45 0.5 0.6 0.6 0.5 0.5 3.15 2 0.5 0.55 0.5 0.55 2.1 2.5 0.7 0.55 1.3 1 2.145
2 0.5 0.5 0.5 0.5 0.6 0.5 3.1 3 0.65 0.5 0.5 0.6 2.25 1 0.6 0.5 1.1 5 2.1285
3 0.7 0.5 0.35 0.5 0.5 0.5 3.05 4.5 0.45 0.5 0.45 0.5 1.9 6.5 0.7 0.5 1.2 1.5 2.0295
4 0.1 0.7 0.5 0.65 0.6 0.5 3.05 4.5 0.5 0.5 0.5 0.5 2 4 0.65 0.5 1.2 4 2.046
5 0.35 0.5 0.5 0.45 0.25 0.5 2.55 11 0.55 0.55 0.55 0.45 2.1 2.5 0.55 0.65 1.2 1.5 1.9305
6 0.5 0.6 0.45 0.55 0.6 0.5 3.2 1 0.45 0.45 0.5 0.5 1.9 6.5 0.55 0.45 1 7.5 2.013
7 0.65 0.35 0.5 0.35 0.35 0.5 2.7 9.5 0.5 0.45 0.5 0.5 1.95 5 0.5 0.5 1 7.5 1.8645
8 0.6 0.35 0.55 0.5 0.5 0.5 3 6 0.4 0.45 0.5 0.5 1.85 8 0.45 0.5 1 9 1.914
9 0.3 0.5 0.45 0.5 0.5 0.5 2.75 8 0.25 0.3 0.5 0.35 1.4 11 0.45 0.6 1.1 6 1.716
10 0.56 0.45 0.5 0.25 0.5 0.5 2.76 7 0.35 0.35 0.35 0.45 1.5 10 0.3 0.5 0.8 11 1.6698
11 0.2 0.65 0.35 0.5 0.5 0.5 2.7 9.5 0.5 0.5 0.3 0.5 1.8 9 0.4 0.45 0.9 10 1.7655
Note C1: Price, C2: Quality, C3: Service, C4: Delivery, C5: Capacity, C6: Past Performance C7: Environmental competencies, C8: Environmental management system,
C9: Green packaging ability, C10: Green product design, C11: Health and safety C12: Education, TS: Total score, WS: Weighted Score
4.7 Strategic Sourcing of Large Number of Suppliers: An Illustrative Case
95
96 4 Modeling and Optimization of Strategic Sustainable Sourcing

Fig. 4.20 Clusters with their centers

Table 4.10 Significance of three clusters

Cluster Supplier Remarks
3 (best) 1, 2, 3, 4 Cluster 3 groups all suppliers who have any combination of the
following traits—low price, better quality, better delivery, better
service, better capacity, good environmental competencies, better
environmental management system, and well-documented policies
and plans for safety, health and education for employees
2 (medium) 5, 6, 7, 8 Cluster 2 groups all suppliers who performed moderately w.r.t 12
sub-criteria and need more improvement
1 (worst) 9, 10, 11 All members of cluster 1 should be rejected

improvement of supplier 5, 6, 7, and 8. As shown in Tables 4.9 and 4.10, supplier 6

has better capacity and also offered better quality and prompt delivery. But he failed
to get selected as he paid less attention to environmental and social metrics. Still
suppliers 5 and 6 could be retained in supplier pool for future requirement as they
performed moderately well. Long-term relationship should be maintained with all
members of cluster 3, and special attention should be paid to suppliers 1 and 2 as
they are the best performers among 11 suppliers. Suppliers 9, 10, and 11 should be
pruned from supply pool. Such differentiation among suppliers performance gives
better insight which is required for strategic sourcing. Tiering of suppliers are not
considered as Company ABC is willing to pay equal priorities to all suppliers of
cluster 3 and want to tier its supplier in near future. Finally, any appropriate
allocation model stated earlier can be considered to allocate order to all members of
cluster 3 as per the requirement of the Company ABC.
4.8 Conclusion 97

4.8 Conclusion

In this chapter, an attempt has been made to discuss strategic sustainable supplier
selection process with three different methods—intuitionistic fuzzy AHP (IF-AHP),
cascaded fuzzy inference system, and K-means data clustering method. Last two
methods are discussed with an example and a case study. Strategic sourcing gives
deeper insight to companies to take better decision to survive in cut throat com-
petition. Intuitionistic fuzzy set is a generalized fuzzy set. It considers the degree of
hesitation of decision makers. It is expected that proposed method could bring more
flexibility to decision-making process for complex strategic sourcing problem.
Proposed cascaded fuzzy inference system demands more resources such as
MATLAB and .NET server. Cascaded FIS brings more computational complexity
and depends highly on rule base. At the same time, structure of membership
function is also subjective. Thus, it demands modification of rule base as well as
modification of membership functions depends upon the need of the problem. The
last method is more or less suitable for large number of suppliers and it gives better
insight for strategic sourcing. This method can easily be implemented with
Microsoft Excel. Companies do prefer simple, easy-to-understand method which
they could implement easily with available resources such as Microsoft Excel. This
is one of the reasons for acceptance of Saaty’s AHP in all major decision-making
processes. In this regard, author strongly suggests interested readers to refer work of
Saaty on dynamic decision-making process. Our decision is not static. Our decision
changes with time and place, and is usually influenced in the presence of external
stimuli. However, judicious decision making is an art as selection of suitable
method can give a better managerial interpretation of the problem and thereby an
extra edge to the company to enhance its supply chain surplus.

References

Amindoust A, Ahmed S, Saghafinia A, Bahreininejad A (2012) Sustainable supplier selection: A

ranking model based on fuzzy inference system. Appl Soft Comput 12:1668–1677
Bai C, Sarkis J (2010) Green supplier development: Analytical evaluation using rough set theory.
J Clean Prod. doi:10.1016/j.jclepro.2010.01.016
Brink S, Diehl JC, Stevels A (1998) ECO-QUEST, an ecodesign self audit tool for suppliers of the
electronics industry. In: Proc. the 1998 international symposium on electronics and the
environment, May 4–6, Oak Brook, Illinois, pp 129–132
Carr AS, Pearson JN (1999) Strategically managed buyer–seller relationships and performance
outcomes. J Oper Manag 17:497–519
Charrad M, Ghazzali N, Boiteau V, Niknafs A (2014) NbClust: An R package for determining the
relevant number of clusters in a data set. J Stat Softw 61(6):1–36
Chen TY (2011) A comparative analysis of score functions for multiple criteria decision making in
intuitionistic fuzzy settings. Inf Sci 181:3652–3676
Council for solid waste solution (1990). www.americanchemistry.com/s_plastics/bin.asp?CID=
1211&DID=4601&DOC=FILE.PDF
98 4 Modeling and Optimization of Strategic Sustainable Sourcing

Das SK, Mathew S (1999) Characterization of material outputs from an electronics demanufac-
turing facility. In: proc. IEEE international symposium on electronics and the environment,
Boston, MA, May 11–13, pp 251–256
Das SK, Naik S (2002) Process planning for product disassembly. Int J Prod Res 40(6):1335–1355
Das S, Yedlarajiah D (2002) An integer programming model for prescribing material recovery
strategies. In: Proc. IEEE international symposium on electronics and the environment, May
6–9, pp 118–122
De Toni A, Nassimbeni G (1999) Buyer–supplier operational practices, sourcing policies and plant
performance: result of an empirical research. Int J Prod Res 37(3):597–619
Franke J (1995) Political evolution of EMAS: perspectives from the EU, National governments
and industrial groups. Bus Strategy Environ 5(3):14–17
Franklin Associates (1991) Product life-cycle assessment: guidelines and principles. EPA Report,
#68-CO-0003
Ghodsypour SH, O’Brien C (1998) A decision support system for supplier selection using an
integrated analytic hierarchy process and linear programming. Int J Prod Econ 56–57:199–212
Guan Z, Jin Z, Zou B (2007) A Multi-Objective Mixed-Integer Stochastic Programming Model for
the Vendor Selection Problem under Multi-Product Purchases. Inform Management Sci 18
(3):241–252
Gungor A, Gupta SM (1999) Issues in environmentally conscious manufacturing and product
recovery: A survey. Comput Ind Eng 36:811–853
Gupta M (1995) Environmental management and its impact on the operations function. Int J Oper
Prod Manage 15(8):34–51
Handfield R, Walton S, Sroufe R, Melnyk S (2002) Applying environmental criteria to supplier
assessment: a study in the application of the analytical hierarchy process. Eur J Oper Res
141:70–87
Hsu C-W, Hu AH (2009) Applying hazardous substance management to supplier selection using
analytic network process. J Clean Prod 17:255–264
Hunt R, Sellers J, Franklin W (1992) Resource and environmental profile analysis: a life cycle
environmental assessment for products and procedures. Environ Impact Assess Rev 12(3):
245–269
Ishii K, Eubanks CF, Marco PD (1994) Design for product retirement and material life-cycle.
Mater Des 15(4):225–233
Kim K, Song I, Kim J, Jeong B (2006) Supply planning model for remanufacturing system in
reverse logistics environment. Comput Ind Eng 51:279–287
Kotha S (1995) Mass customisation: implementing the emerging paradigm for competitive
advantage. Strategic Manage J 16:21–42
Lee HI, Kang HY, Hsu CF, Hung HC (2009) A green supplier selection model for high-tech
industry. Expert Syst Appl 36(4):7917–7927
Levan LS (1998) Life cycle assessment: measuring environmental impact. www.fpl.fs.fed.us/
documnts/pdf1998/levan98b.pdf
Milligan GW, Cooper MC (1985) An examination of procedures for determining the number of
clusters in a data set. Psychometrika 50(2):159–179
Mukherjee K, Sarkar B, Bhattacharya A (2011) Comments on the erratum to “Supply planning
model for remanufacturing system in reverse logistics environment” [Comput. Ind. Eng.
51 (2006) 279–287]. Comput Ind Eng 61:1349–1350
Muralidharan C, Anantharaman N, Deshmukh (2002) A multi-criteria group decision making
model for supplier rating. J Supply Chain Manage Fall 22–33
Noci G (1997) Design “green” vendor rating systems for the assessment of a supplier’s
environmental performance. Eur J Purchasing Supply Manage 3(2):103–114
Pine BJ (1993) Mass customisation. HBS Press, Boston
Sanchez PP, Soyer R (1998) Information concepts and pair-wise comparison matrices. Inf
Processing Lett 68:185–188
Shin H, Collier DA, Wilson DD (2000) Supply management orientation and supplier buyer
performance. J Oper Manage 18:317–333
References 99

Smith KG, Carroll SJ, Ashford SJ (1995) Intra-and inter-organizational cooperation: toward a
research agenda. Acad Manag J 38(1):7–23
Spekman RE (1988) Perceptions of strategic vulnerability among industrial buyers and its effect on
information search and supplier evaluation. J Bus Res 17:313–326
Stilwell JR, Canty PK, Montrone A (1991) Packaging for the environment. American
Management Association, New York
Svoboda S (1995) Note on life cycle analysis www.umich.edu/*nppcpub/resources/compendia/
CORPpdfs/CORPlca.pdf
Szmidt E, Kacprzyk J (2000) Distance between intuitionistic fuzzy sets. Fuzzy Sets, Syst 114
(3):505–518
Tracey M, Tan CL (2001) Empirical analysis of supplier selection and involvement, customer
satisfaction and firm performance. Supply Chain Manage: Int J 6(4):174–188
Van Hoek RI (1999) From reversed logistics to green supply chains. Supply Chain Manage: Int J 4
(3):129–135
Wang M, Perkins JR (2006) Using interval alignment policies for efficient production control of
supply chain systems. Int J Ind Syst Eng 1(1–2):87–108
Wang M, Perkins JR (2011) Time interval alignment (ia) policies, boundary and applications with
multiple stream arrivals. J Syst Sci Syst Eng 20(4):400–415
Wu CH, Kuo TC, Lu YY (2007) Environmental principles applicable to green supplier evaluation
by using multi-objective decision analysis. Int J Prod Res 45(18–19):4317–4331
Xia W, Wu Z (2007) Supplier selection with multiple criteria in volume discount environments.
Omega 35:494–504
Zeid I, Gupta SM (2002) Computational algorithm to evaluate product disassembly cost index. In:
Proc. of the SPIE international conference on environmentally conscious manufacturing II,
February 11, pp 23–31
Zhang HC, Li J, Shrivastava P, Whitely A, Eugene M (2004) A web-based system for reverse
manufacturing and product environmental impact assessment considering end-of-life dispo-
sitions. CIRP Annals-Manuf Technol 53(1):5–8
Chapter 5
A Note on Limitations of FAHP

One of the most difficult tasks in multiple criteria decision

analysis (MCDA) is determining the weights of individual
criteria so that all alternatives can be compared based on the
aggregate performance of all criteria.
Chiang Kao

Applied Mathematical Modelling (2010), Vol. 34,

pp. 1779–1787

5.1 Introduction

In decision-making process, uncertainty can be classified into two categories,

namely internal and external. External uncertainties refer to imprecise data, vague
problem definition, etc. Internal uncertainties refer to subjectivity of the decision
maker. Fuzzy set theory is a well-established subject, and its application is well
accepted in engineering. But application of fuzzy set theory in decision sciences is
not unanimously accepted yet. Fuzziness may work better in metric spaces than in
order spaces (Saaty and Tran 2007). Particularly, fuzzy AHP has certain limitations
(Saaty and Tran 2007, 2010; Zhu 2014, 2012). In 1986, Saaty, the originator of
AHP, proposed four axiomatic foundations of AHP. Some of them are as follows:
1. Reciprocal Axiom: For all judgments aij , its reciprocal aji , must satisfy the
following relation

aij
aji ¼ 1 8i, j ¼ 1,2,. . .,n ð5:1Þ

1
or aij ¼
aji

Fuzzy AHP basically violates the above relation. For example, a triangular fuzzy
number af f
i| ¼ ð2; 3; 4Þ and its reciprocal a |i ¼ ð 4 ; 3 ; 2 Þ
1 1 1

© Springer (India) Pvt. Ltd. 2017 101

K. Mukherjee, Supplier Selection, Studies in Systems, Decision and Control 88,
DOI 10.1007/978-81-322-3700-6_5
102 5 A Note on Limitations of FAHP

af f
i|  a |i ¼ ðl; m; uÞ  ð1=u; 1=m; 1=lÞ ’ ðl=u; m=m; u=lÞ 6¼ ð1; 1; 1Þ ð5:2Þ

where  represents fuzzy multiplication.

The above multiplication approach was used by Chang (1996). Van Laarhoven
and Pedrycz (1983) used following approach:

ðl1; m1; u1Þ  ðl2; m2; u2Þ  ðl1 l2; m1 m2; u1 u2Þ 6¼ ð1; 1; 1Þ ð5:3Þ

The above multiplication is an approximation.

2. Consistency Axiom: In AHP, if all judgments are consistent, then following
relation must satisfy

aik
akj ¼ aij 8 i, j, k ¼ 1; 2; 3; . . .:; n ð5:4Þ

A positive fuzzy reciprocal matrix is consistent if and only if afik 

af f
k| ¼ a i| :
For example, if af f
11 ¼ ð1; 1; 1Þ; a f
12 ¼ ð2; 3; 4Þ and a 21 ¼ ð1=4; 1=3; 1=2Þ, then

af f
12  a f
21 6¼ a 11

3. Continuity Axiom: If A  B  C, then there exist probability p in (0, 1) such

that B  pA þ ð1  pÞC (Zhu 2014). For example, M1(4, 5, 6), M2(2, 3, 4), M3
(1, 1, 1)

M1  M2  M3:

Therefore,

M2  pM1 þ ð1  pÞM3 ð5:5Þ

The above equation can be written as follows:

8
< 2  4p þ ð1  pÞ
3  5p þ ð1  pÞ ð5:6Þ
:
4  6p þ ð1  pÞ
8
< p  1=3
p  1=2 ð5:7Þ
:
p  3=5

None of the above value of p satisfies Eq. (5.6) Thus, no p in (0, 1) satisfies
Eq. (5.7). Hence, fuzzy AHP violates axiom of reciprocal, axiom of consistency,
and axiom of continuity.
Zhu (2014) mentioned that Saaty’s fundamental 1–9 scale itself is fuzzy. Terms
used in fundamental scale such as ‘moderate importance’ and ‘strong
5.1 Introduction 103

importance’ are linguistic variables to express degree of dominance of one

variable to other. Thus, values of fundamental scale are not crisp, and there is
absolute no need of fuzzification of fundamental scale for fuzzy AHP. Saaty and
Tran (2007) mentioned that good judgment produces valid answer in AHP, and
fuzzy AHP simply produces perturbation without producing any better out-
comes. Even inconsistent good judgment yields better result in AHP, but fuzzy
AHP produces worse rather than better outcomes.

5.2 Other Limitations of Fuzzy AHP Models

Some of the limitations of the Van Laarhoven and Pedrycz (1983) method are as
follows:
• Methodology proposed by them to normalize the local fuzzy weights was
problematic (Wang and Parkan 2006).
• Their proposed method for incomplete fuzzy comparison matrices for local
fuzzy weights was not suitable (Wang and Parkan 2006).
Some of the limitations of Buckley (1985) method are as follows:
• If the pairwise comparison matrix is consistent, then their method yields same
priority vectors as the eigenvectors method, which was Saaty’s classical AHP
method. But it fails to produce same result for inconsistent matrix (Csutora and
Buckley 2001).
Some the limitations of Chang (1996) model are as follows:
• Proposed method of Chang could lead to a wrong decision if it assigns zero
weights to some items such as criteria, sub-criteria, or alternatives and excludes
them from the decision analysis (Wang et al. 2008).

5.3 Consistency Index and Optimization Methods

for AHP

Saaty proposed consistency ratio (C.R) which is the ratio of consistency index (C.I)
and random index (R.I) and defined as follows:

kmax  n C:I
C:I ¼ and C:R ¼ \0:1
n1 R:I
104 5 A Note on Limitations of FAHP

Thus C.R can be simplified as follows:

kmax  n
C:R ¼ \0:1 ) kmax \0:1:R ðn  1Þ þ n ð5:8Þ
Rðn  1Þ

With the help of consistency index Saaty imposed an extra constraint on prin-
cipal eigenvalue, shown in Eq. 5.8, to reduce values of all elements of pairwise
comparison matrix except its diagonal elements. For example, values of a12, a13,
and a23 should be reduced through iteration so that a 3  3 pairwise comparison
matrix could satisfy consistency index or maximum limit of principal eigenvalue,
kmax , shown in Table 5.1.
There are several other methods are available to derive priorities from AHP. In
this sec., optimization approach is considered mainly.
Optimization Approaches to find priority for AHP.

5.3.1 Weighted Least Square Method (Chu et al. 1979)

Wi
aij  ð5:9Þ
Wj

X
n X
n
Min S ¼ ðaij  Wi = Wj Þ2 ð5:10Þ
i¼1 j¼1

Subject to

X
n
Wi ¼ 1 ð5:11Þ
i¼1

5.3.2 Error Minimization Method (Chen

and Triantaphyllou 2001)
Wi
aij ¼ ð1 þ eij Þ ð5:12Þ
Wj

Equation 5.10 should be modified with Eq. 5.11.

Table 5.1 Limit of principal eigenvalue (RI is from Saaty and Tran 2007; with kind permission from Elsevier Limited)
Size, n 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Random index 0 0 0.52 0.89 1.12 1.26 1.36 1.41 1.46 1.49 1.52 1.54 1.6 1.58
Principal eigenvalue is 1 2 3.104 4.267 5.448 6.63 7.816 8.987 10.17 11.34 12.52 13.69 14.9 16.05
less than equal to
5.3 Consistency Index and Optimization Methods for AHP
105
106 5 A Note on Limitations of FAHP

5.3.3 Logarithmic Least Square Method

n X
X n
Min ðln aij  ðln Wi  ln Wj ÞÞ2 ð5:13Þ
i¼1 j [ i

Subject to

X
n
Wi ¼ 1 ð5:14Þ
i¼1

5.3.4 Goal Programming Method (Bryson 1995)

!
Wi dijþ
aij ¼ 8i; j ¼ 1; 2; 3; 4; . . .; n and j [ i ð5:15Þ
Wj dij

where dijþ 1 and d

ij 1 are deviation variables and both of them cannot be
greater than 1 simultaneously.

X
n X
n
Min ðln dijþ þ ln d
ij Þ ð5:16Þ
i¼1 j [ i

Subject to
 
ln aij ¼ inWi  inWj þ ln dijþ  ln d
ij ð5:17Þ

Unfortunately, none of the optimization method considered maximum limit of

principal eigenvalue, shown in Table 5.1.

5.4 Alternative Approaches to FAHP

Saaty mentioned that fuzzy AHP brings more fuzziness in the result. To explain
Saaty’s statement, let us consider two triangular fuzzy numbers (TFNs) A e¼
e
f1; 2; 3g and B ¼ f2; 3; 4g are used for pairwise comparison of criteria or alter-
natives. TFNs are considered in this regard, as they are commonly used in fuzzy
AHP. The product of the above two TFNs are equal to {2, 6, 12}. As mentioned in
Chap. 1, triangular fuzzy number A e refers to the linguistic term ‘about to 2’. If we
decrease differences between lower and upper limit of TFN, then we will become
closer to average value. It means certainty increases with the decrease in support of
any TFN or alternatively, we can say multiplication of TFNs brings more fuzziness
5.4 Alternative Approaches to FAHP 107

Table 5.2 Method of Triantaphyllou and Lin (1996)

C1 C2 C3 Calculation of priority
C1 (1, 1, 1) (5/2, 3, (7/2, 4, (35/4, 12, 63/4) = (8.75, 12, 15.75)1/3 = (2.0606,
7/2) 9/2) 2.2894, 2.5066) = (0.187, 0.208, 0.2278)
C2 (2/7, (1, 1, 1) (3/2, 2, (3/7, 2/3, 1) = (0.428, 0.66,1)1/3 = (0.7536,
1/3, 5/2) 0.8707, 1.0000) = (0.0685, 0.079, 0.09)
2/5)
C3 (2/9, (2/5, (1, 1, 1) (4/45, 1/8, 4/21) = (0.088, 0.125,
1/4, 1/2, 0.19)1/3 = (0.4448, 0.5000, 0.5749) = (0.04,
2/7) 2/3) 0.045, 0.052)

to the result as multiplication increases support of the product which is also a TFN.
Some of the methods for deriving priority for FAHP are as follows:

5.4.1 Method of Triantaphyllou and Lin (1996)

Triantaphyllou and Lin (1996) proposed an alternative approach to quantify priority

of alternatives. Their proposed method is explained with an example, previously
solved in Chap. 1 with Extent Fuzzy AHP (Table 5.2).
Obtained priorities show that C1  C2  C3 :
Both Extent Fuzzy AHP (Chang 1995) and method of Triantaphyllou and Lin
(1996) did not consider consistency of fuzzy pairwise comparison matrix and their
result significantly deviates from each other.

5.4.2 Least Square Distance Method (Wang and Parkan

2006; with Kind Permission from Elsevier Limited)

If Rei and f
R| be the two triangular or trapezoidal numbers. The weighted Euclidean
distance between Rei and f R| is defined as follows
m qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
X
dij ¼ ðWi rik  Wj rjk Þ2 ð5:18Þ
k¼1

X
n X
n
Min S ¼ dij2 ð5:19Þ
i¼1 j¼1; j6¼i
108 5 A Note on Limitations of FAHP

Subject to

X
n
Wi ¼ 1 ð5:20Þ
i¼1

5.4.3 Defuzzification-Based Least Square Method (Wang

and Parkan 2006; with Kind Permission from Elsevier
Limited)

If Rei and fR| be the two triangular or trapezoidal numbers then their defuzzified
value Z can be expressed as follows:
For triangular fuzzy number, Z ¼ 13 ðr1 þ r2 þ r3 Þ:
For trapezoidal fuzzy number, Z ¼ 14 ðr1 þ r2 þ r3 þ r4 Þ:
The defuzzified values of the weighted fuzzy numbers should be as close as
possible. Their proposed objective function is as follows:

X
n X
n
Min S ¼ ðWi zi  Wj zj Þ2 ð5:21Þ
i¼1 j¼1; j6¼i

Subject to

X
n
Wi ¼ 1 ð5:22Þ
i¼1

5.4.4 Preference Programming (Salo and Hämäläinen

1995)

Salo and Hämäläinen (1995) extended the work of Arbel (1993) and used series of
optimization to derive priority vectors for criteria or alternatives from the feasible
region of decision space. They divided entire hierarchy into several levels and used
top-down or bottom-up or any feasible combination to derive priority. One of the
major significance of their method is the graphical presentation of feasible region of
decision space and introduction of ambiguity index (AI). Their method is included
in this section as any triangular or trapezoidal fuzzy number can be easily converted
to interval values with fuzzy alpha-cut method. But such generalization of fuzzy
number is always associated with loss of information. Let us consider that a
decision maker has prepared following pairwise comparison matrix for three
alternatives w.r.t a given criterion
5.4 Alternative Approaches to FAHP 109

Fig. 5.1 Feasible region of decision space

2 3 2 W1 W1 W1
3
1 2 3 W W2 W3
6 12 7
S ¼ 4 1=2 1 45 ¼ 4W
W1
W2
W2
W2
W3 5 ð5:23Þ
1=3 1=4 1 W3 W3 W3
W1 W2 W3

By comparing above two matrices, we can derive the following relations:

6 3 2
w1 ¼ 2w2 ; w1 ¼ 3w3 ; w2 ¼ 4 w3 and normalized priority V ¼ ð ; ; Þ
11 11 11
ð5:24Þ

Feasible region, S of the above problem is shown in Fig. 5.1. Salo’s method can
convert the above pairwise comparison matrix is as follows:
2 3
1 2 ½3; 8

S ¼ 4  1=2  1 1  ½3=2; 4

5 and AI ¼ 0:1129 ð5:25Þ

8 ; 1=3 4 ; 2=3
1
1

Ambiguity Index of Salo and Hämäläinen (1995) determines fuzziness associ-

ated in assigning interval limit in fuzzy pairwise comparison matrix. But they did
not consider any consistency index in their calculation. As a result, an interval
matrix with lower value of ambiguity can have higher inconsistency. However,
their proposed method is flexible enough to solve complex multi-level hierarchical
problem.

5.4.5 Fuzzy Preference Programming (Mikhailov and Singh

2003)

Fuzzy Preference Programming (FPP) converts prioritization problems into a fuzzy

linear programming problem to identify non-convex, non-empty, fuzzy feasible
110 5 A Note on Limitations of FAHP

Table 5.3 Comparison of AHP and different FAHP methods

C1 C2 C3 Saaty’s Triantaphyllou and Extent Fuzzy FPP
AHP Lin AHP
C1 1 3 4 0.625 0.2076 1 0.62203
C2 1/3 1 2 0.238 0.0792 0 0.23511
C3 1/4 1/2 1 0.137 0.0457 0 0.14286

region that contains all priorities which satisfies all interval constraints (Mikhailov
and Singh 2003). In FPP, decision makers consider a linear membership function to
express degree of violation of the constraint mentioned below:

~ wi = wj 
lij  ~ uij ð5:26Þ

where  ~ denotes fuzzy less or equal.

The optimization problem prepared with the above constraint can be expressed
as follows:
Maximize k
subject to

ðmij  lij Þ kwj  wi þ lij wj  0 8 i ¼ 1; 2; 3; . . .; n  1; j ¼ 2; 3; 4; . . .; n and j [ i

ð5:27Þ

ðuij  mij Þ kwj þ wi  uij wj  0 ð5:28Þ

X
n
wk 8k ¼ 1; 2; 3; . . .; n ð5:29Þ
k¼1

For example, above problem, previously solved in Chap. 1 with Extent

Fuzzy AHP, is solved with FPP (Table 5.3).
Result obtained from FPP, Extent Fuzzy AHP, and method of Triantaphyllou
and Lin (1996) differs considerably and raises questions to select best method for
the simple problem. Result of FPP is very close to the result of Saaty’s AHP. It
proves that FPP can give more accurate result as it does not consider multiplicative
relation in its entire computation process. Today, more research work is required to
well define consistency index of FAHP. In this regard, interested readers can refer
work of Leung and Cao (2000).

5.5 Conclusion

Selection of appropriate MCDA tools is an open issue. Purpose of this chapter is to

aware researchers about the limitations of some of the fuzzy AHP methods. It is
crystal clear that all three models of fuzzy AHP, namely method of Van Laarhoven
5.5 Conclusion 111

and Pedrycz (1983), Buckley (1985), and Chang (1996) violate axioms of Saaty’s
classical AHP. Among all methods, Fuzzy Preference Programming (Mikhailov
and Singh 2003) gives result closer to the classical AHP. Thus, selection of an
appropriate method is a multi-criteria problem to trade-off conflicting criteria such
as accuracy of result, simplicity of method, and consistency of result.

References

Bryson N (1995) A goal programming method for generating priorities vectors. J Oper Res Soc
641–648
Buckley JJ (1985) Fuzzy hierarchical analysis. Fuzzy Sets Syst 17:233–247
Chang D-Y (1996) Applications of the extent analysis method on fuzzy AHP. Eur J Oper Res 95
(3):649
Chu ATW, Kalaba RE, Spingaran K (1979) A comparison of two methods for determining the
weights of belonging to fuzzy sets. J Optim Theory Appl 27(4):321–538
Csutora R, Buckley JJ (2001) Fuzzy hierarchical analysis: the Lamda-Max method. Fuzzy Sets
Syst 120:181–195
Mikhailov L, Singh MG (2003) Fuzzy Analytic Network Process and its Application to the
Development of Decision Support Systems. IEEE Trans Syst, Man, Cybern Part C 33(1):33–41
Saaty TL, Tran LT (2007) On the invalidity of fuzzifying numerical judgments in the Analytic
Hierarchy Process. Math Comput Model 46:962–975
Saaty TL, Tran LT (2010) Fuzzy Judgments and Fuzzy Sets. Int J Strateg Decis Sci 1(1):23–40
Salo A, Hämäläinen RP (1995) Preference programming through approximate ratio comparisons.
Eur J Oper Res 82:458–475
Van Laarhoven PJM, Pedrycz W (1983) A fuzzy extension of Saaty’s priority theory. Fuzzy Sets
Syst 11:229–241
Wang Y-M, Parkan C (2006) Two new approaches for assessing the weights of fuzzy opinions in
group decision analysis. Inf Sci 176:3538–3555
Wang Y-M, Luo Y, Hua Z (2008) On the extent analysis method for fuzzy AHP and its
applications. Eur J Oper Res 186:735–747
Zhu K (2012) The invalidity of triangular fuzzy AHP—a mathematical justification. http://dx.doi.
org/10.2139/ssrn.2011922
Zhu K (2014) Fuzzy analytic hierarchy process: fallacy of the popular methods. Eur J Oper Res
236(1):209–217
Appendix
MCDA Tools and Meta-Heuristic
Techniques: Sample Codes

Computer Codes
A.1 MATLAB code for Analytic Hierarchy Process (AHP)
******************************************************************
This code is developed on the algorithm proposed by T.L.Saaty for AHP

******************************************************************
tic;
clc;
clear % is must required to clean workspace data %
row = input('Enter no of rows or columns :');
for i = 1:1:row
for j= 1:1:row
if i==j
a(i,j)=1;
elseif (i>j)
a(i,j)= 1/a(j,i) ;
else
a(i,j)=input('Enter pair wise comparison value :');

end;

end
end
disp('Pair Wise Comparison Matrix')
disp(a)

***********************************************************
% Normaliztion of pair wise matrix %

for i=1:1:row
Sum_Col=0.0;
for j=1:1:row
Sum_Col= Sum_Col + a(j,i);
end
Col_Sum(i)= Sum_Col;
end
disp(Col_Sum)
for i=1:1:row
for j=1:1:row
norm_mat(j,i)= a(j,i)/Col_Sum(i);
end
end

© Springer (India) Pvt. Ltd. 2017 113

K. Mukherjee, Supplier Selection, Studies in Systems, Decision and Control 88,
DOI 10.1007/978-81-322-3700-6
114 Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes

disp('Normalized Matrix')
disp(norm_mat)
% Take average of row to calculate priority %
for i=1:1:row
row_sum = 0.0;
for j=1:1:row
row_sum = row_sum + norm_mat(i,j);
end
prio_mat(i)= row_sum/row;
end
disp('Priority Vectors')
disp(prio_mat)
*************************************************
% Calculation of consistency index %

lambda_max=0.0;
for i=1:1:row
lambda_max= lambda_max + Col_Sum(i)* prio_mat(i);
end
CI = (lambda_max - row) /(row -1);
disp('Consistency Index')
disp(CI)
% Random Consistency Index Formula%
RI=1.98*(row-2)/row;
disp('Random Consistency Index')
disp(RI);
CR=CI/RI;
disp('Consistency Ratio')
disp(CR)

************************************************
%Pair Wise Comparison Matrix for Alternatives%

Alt_row =input('Enter No. of Alternatives:');

for k=1:1:row %Comparison with respect to each criterion%
disp(['Comparison with respect to Criteria : ', num2str(k)])
for i=1:1:Alt_row
for j=1:1:Alt_row
if i==j
alt(i,j,k)=1;
elseif (i>j)
alt(i,j,k)= 1/alt(j,i,k) ; % Array of Matrix representation %
else
alt(i,j,k)=input('Enter pair wise comparison value :');

end;
end
end
end
for i=1:1:row
disp(alt(:,:,i))
end

**********************************************************
% Normaliztion of pair wise matrix %
Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes 115

for k=1:1:row
for i=1:1:Alt_row
Sum_Col=0.0;
for j=1:1:Alt_row
Sum_Col= Sum_Col + alt(j,i,k);
end
Col_Sum_alt(k,i)= Sum_Col;
end
end

for i=1:1:row
disp(Col_Sum_alt(i,:))

end

for k=1:1:row
for i=1:1:Alt_row
for j=1:1:Alt_row
norm_mat_alt(j,i,k)= alt(j,i,k)/Col_Sum_alt(k,i);
end
end
end
disp('Normalized Alternative Matrix')
for i=1:1:row
disp(norm_mat_alt(:,:,i))
end

*************************************************
% Take average of row to calculate priority %

for k=1:1:row
for i=1:1:Alt_row
row_sum = 0.0;
for j=1:1:Alt_row
row_sum = row_sum + norm_mat_alt(i,j,k);
end
prio_mat_alt(k,i)= row_sum/Alt_row;
end
end

disp('Local Priority Vectors Of Alternative')

for i=1:1:row
disp(prio_mat_alt(i,:))
end

**************************************************
% Calculation of CR of each pair wise comparison matrix of alternatives %

for i=1:1:row
lambda_max_alt=0.0;
for j=1:1:Alt_row
lambda_max_alt= lambda_max_alt + Col_Sum_alt(i,j)* prio_mat_alt(i,j);
end
lambda_max_new(i)= lambda_max_alt;
end
for i=1:1:row
CI_alt(i) = (lambda_max_new(i) - Alt_row) /(Alt_row -1);
end
116 Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes

disp('Consistency Index of Pairwise Comparion Matrix of Alternatives')

disp(CI_alt)

******************************************
% Random Consistency Index Formula%

RI=1.98*(Alt_row-2)/Alt_row;
disp('Random Consistency Index')
disp(RI);
for i=1:1:row
CR_alt(i)=CI_alt(i)/RI;
end
disp('Consistency Ratio of Pairwise Comparison Matrix of Alternatives')

disp(CR_alt)

************************************************
%disp('Global Priority Vectors Of Alternative')

for i=1:1:Alt_row
Prio_Sum =0.0;
for j=1:1:row
Prio_Sum = Prio_Sum + prio_mat(j)* prio_mat_alt(j,i);
end
global_prio(i)=Prio_Sum;
end
disp('Rank of Alternativies:')
sort_global_prio = sortrows(global_prio);
for i=1:1:Alt_row
for j=1:1:Alt_row
if sort_global_prio(i)==global_prio(j)
rank(j)= i;
end
end
end
for i=1:1:Alt_row
disp(['Alternative : ', num2str(i),' Global Priority: ',num2str(global_prio(i)), ' Rank : ',num2str(rank(i))])
end
toc;

A.2 MATLAB Code for Fuzzy AHP by alpha-cut method

***************************************************************************
This code is developed on fuzzy AHP alpha–cut method

***************************************************************************
tic;
clc;
clear % is must required to clean workspace data %
row = input('Enter no of rows or columns :');
for i = 1:1:row
for j= 1:1:row
if i==j
a(i,j)=1;
elseif (i>j)
a(i,j)= 1/a(j,i) ;
Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes 117

else
a(i,j)= Fuzzy_Alpha_Cut();

end;

end
end
disp('Pair Wise Comparison Matrix')
disp(a)

******************************************
% Normaliztion of pair wise matrix %

for i=1:1:row
Sum_Col=0.0;
for j=1:1:row

Sum_Col= Sum_Col + a(j,i);

end
Col_Sum(i)= Sum_Col;
end
disp(Col_Sum)
for i=1:1:row
for j=1:1:row
norm_mat(j,i)= a(j,i)/Col_Sum(i);
end
end
disp('Normalized Matrix')
disp(norm_mat)

************************************************
% Take average of row to calculate priority %

for i=1:1:row
row_sum = 0.0;
for j=1:1:row
row_sum = row_sum + norm_mat(i,j);
end
prio_mat(i)= row_sum/row;
end
disp('Priority Vectors')
disp(prio_mat)

*************************************************
% Calculation of consistency index %

lambda_max=0.0;
for i=1:1:row
lambda_max= lambda_max + Col_Sum(i)* prio_mat(i);
end
CI = (lambda_max - row) /(row -1);
disp('Consistency Index')
disp(CI)

************************************************
% Random Consistency Index Formula%
RI=1.98*(row-2)/row;
disp('Random Consistency Index')
118 Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes

disp(RI);
CR=CI/RI;
disp('Consistency Ratio')
disp(CR)

*****************************************************
%Pair Wise Comparison Matrix for Alternatives%

Alt_row =input('Enter No. of Alternatives:');

for k=1:1:row %Comparison with respect to each criterion%
disp(['Comparison with respect to Criteria : ', num2str(k)])
for i=1:1:Alt_row
for j=1:1:Alt_row
if i==j
alt(i,j,k)=1;
elseif (i>j)
alt(i,j,k)= 1/alt(j,i,k) ; % Array of Matrix representation %

else
alt(i,j,k)= Fuzzy_Alpha_Cut();

end;
end
end
end
for i=1:1:row
disp(alt(:,:,i))
end

*****************************************************
% Normaliztion of pair wise matrix %

for k=1:1:row
for i=1:1:Alt_row
Sum_Col=0.0;
for j=1:1:Alt_row
Sum_Col= Sum_Col + alt(j,i,k);
end
Col_Sum_alt(k,i)= Sum_Col;
end
end

for i=1:1:row
disp(Col_Sum_alt(i,:))
end

for k=1:1:row
for i=1:1:Alt_row
for j=1:1:Alt_row
norm_mat_alt(j,i,k)= alt(j,i,k)/Col_Sum_alt(k,i);
end
end
end
disp('Normalized Alternative Matrix')
for i=1:1:row
disp(norm_mat_alt(:,:,i))
end
Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes 119

**************************************************
% Take average of row to calculate priority %

for k=1:1:row
for i=1:1:Alt_row
row_sum = 0.0;
for j=1:1:Alt_row
row_sum = row_sum + norm_mat_alt(i,j,k);
end
prio_mat_alt(k,i)= row_sum/Alt_row;
end
end

disp('Local Priority Vectors Of Alternative')

for i=1:1:row
disp(prio_mat_alt(i,:))
end
**********************************************************
% Calculation of CR of each pair wise comparison matrix of alternatives %

for i=1:1:row
lambda_max_alt=0.0;
for j=1:1:Alt_row
lambda_max_alt= lambda_max_alt + Col_Sum_alt(i,j)* prio_mat_alt(i,j);
end
lambda_max_new(i)= lambda_max_alt;
end
for i=1:1:row
CI_alt(i) = (lambda_max_new(i) - Alt_row) /(Alt_row -1);
end
disp('Consistency Index of Pairwise Comparion Matrix of Alternatives')
disp(CI_alt)

*****************************************************
% Random Consistency Index Formula%

RI=1.98*(Alt_row-2)/Alt_row;
disp('Random Consistency Index')
disp(RI);
for i=1:1:row
CR_alt(i)=CI_alt(i)/RI;
end
disp('Consistency Ratio of Pairwise Comparison Matrix of Alternatives')
disp(CR_alt)

********************************************************
%disp('Global Priority Vectors Of Alternative')

for i=1:1:Alt_row
Prio_Sum =0.0;
for j=1:1:row
Prio_Sum = Prio_Sum + prio_mat(j)* prio_mat_alt(j,i);
end
global_prio(i)=Prio_Sum;
end
disp('Rank of Alternativies:')
sort_global_prio = sortrows(global_prio);
120 Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes

for i=1:1:Alt_row
for j=1:1:Alt_row
if sort_global_prio(i)==global_prio(j)
rank(j)= i;
end
end
end
for i=1:1:Alt_row
disp(['Alternative : ', num2str(i),' Global Priority: ',num2str(global_prio(i)), ' Rank : ',num2str(rank(i))])
end
toc;

A.3 MATLAB Code for Extent Fuzzy AHP

***************************************************************************
This code is developed on algorithm proposed by D.Y.Chang for Extent Fuzzy AHP

***************************************************************************
tic;
clc;
clear % is must required to clean workspace data %
row = input('Enter no of rows or columns :');
for i = 1:1:row
for j= 1:1:row

if i==j
a(i,j,1)=1;
a(i,j,2)=1;
a(i,j,3)=1;
elseif (i>j)
for k=1:1:3
n=4-k;
a(i,j,k)= 1/a(j,i,n) ;
end;
else
a1= Extent_Fuzzy_AHP_Linguistic();
a(i,j,1)=a1(1);
a(i,j,2)=a1(2);
a(i,j,3)=a1(3);
end
end
end
sum_u=0.0;
sum_m=0.0;
sum_l=0.0;
for i=1:1:row
for j=1:1:row
sum_u=sum_u+a(i,j,3);
sum_m=sum_m+a(i,j,2);
sum_l=sum_l+a(i,j,1);

end
end
disp('Sum of L M U:')
disp(sum_l)
disp(sum_m)
disp(sum_u)
Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes 121

***************************************************
%Creation of Cell Array %

C=cell(row,3);
for i=1:1:row
sum_u1=0.0;
sum_m1=0.0;
sum_l1=0.0;
for j=1:1:row
sum_u1=sum_u1+ a(i,j,3);
sum_m1=sum_m1+a(i,j,2);
sum_l1=sum_l1+a(i,j,1);

end
sum_l2(i)=sum_l1;
sum_m2(i)=sum_m1;
sum_u2(i)=sum_u1;
C(i,:)={sum_l2(i) sum_m2(i) sum_u2(i)};

end
disp('Extent Fuzzy Synthetic Value')
disp(C)

***************************************************
% Calculation of fuzzy synthetic extent value%

for i=1:1:row
S(i,1)= sum_l2(i)/sum_u;
S(i,2)=sum_m2(i)/sum_m;
S(i,3)=sum_u2(i)/sum_l;
end
disp(S)

****************************************************
% Calculation of Degree Possibility%

for j=1:1:row
for i=1:1:row
if i==j
Val(i)=196;
else
Val(i)= Degree_Possibility(S(j,1),S(j,2),S(j,3),S(i,1),S(i,2),S(i,3));
end
end
min_pos(j) = min(Val);
end
% Calculate Priority Vector%
disp('Local Priority Vector:')
disp(min_pos)

*****************************************************
%Calculate Normalized Priorirty Vector%

norm_sum=0.0;
for i=1:1:row
norm_sum=norm_sum+min_pos(i);
end
for i=1:1:row
122 Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes

norm_priority(i) = min_pos(i)/norm_sum;
end
disp('Normalized Priority')
disp(norm_priority)
toc;

A.4 MATLAB Code for Degree of possibility

function Poss = Degree_Possibility(l2,m2,u2,l1,m1,u1)

%UNTITLED Summary of this function goes here
% Detailed explanation goes here
if m2>=m1
Poss=1;
elseif l1>=u2
Poss=0;
else
deno=(m2-u2)-(m1-l1);
Poss= (l1-u2)/deno;
end

A.5 MATLAB Code for Extent Fuzzy AHP(linguistic variables)

function fuzzy_val = Extent_Fuzzy_AHP_Linguistic()

%UNTITLED2 Summary of this function goes here
% Detailed explanation goes here
% Declaration of TFN
Y=input('Enter Yes or No if you want to flip comparison:','s');
S = input('Enter any of the following linguistic terms : absolute,very strong,fairly strong,weak,equal:','s');
if strcmp(Y,'No')==1
switch lower(S)
case 'absolute'
fuzzy_val = [3.5;4;4.5];
case 'very strong'
fuzzy_val = [2.5;3;3.5];
case 'fairly strong'
fuzzy_val = [1.5;2;2.5];
case 'weak'
fuzzy_val = [.67;1;1.5];
case 'equal'
fuzzy_val = [1;1;1];
otherwise
disp('Unknown Linguistic Term.')
end
else
switch lower(S)
case 'absolute'
fuzzy_val = [0.2222;0.25;0.2857];
case 'very strong'
fuzzy_val = [0.2857;0.3333;0.4];
case 'fairly strong'
fuzzy_val = [0.4;0.5;0.6666];
case 'weak'
fuzzy_val = [0.6666;1;1.4925];

case 'equal'
fuzzy_val = [1;1;1];
otherwise
Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes 123

disp('Unknown Linguistic Term.')

end
end

A.6 MATLAB code for Fuzzy Alpha Cut Method

***************************************************************************
This computer code is developed for fuzzy alpha-cut method

***************************************************************************

function a_alpha_beta = Fuzzy_Alpha_Cut()

%UNTITLED2 Summary of this function goes here
% Detailed explanation goes here
% Declaration of TFN
S = input('Enter any of the following linguistic terms : absolute,very strong,fairly strong,weak,equal:','s');
switch lower(S)
case 'absolute'
fuzzy_val = [3.5;4;4.5];
case 'very strong'
fuzzy_val = [2.5;3;3.5];
case 'fairly strong'
fuzzy_val = [1.5;2;2.5];
case 'weak'
fuzzy_val = [.67;1;1.5];
case 'equal'
fuzzy_val = [1;1;1];
otherwise
disp('Unknown Linguistic Term.')
end
% Defuzzification of fuzzy value by fuzzy alpha cut %
%alpha = input('Enter fuzzy alpha-cut value : ');
%beta=input('Enter degree of satisfaction value : ');
alpha=0.5;
disp(fuzzy_val(2))
a_alpha_lower = (fuzzy_val(2)-fuzzy_val(1))* alpha + fuzzy_val(1);
a_alpha_upper = fuzzy_val(3)-(fuzzy_val(2) -fuzzy_val(3))* alpha;
%a_alpha_beta = (1-beta)* a_alpha_lower + beta * a_alpha_upper;

a_alpha_beta (1) =(a_alpha_lower);

a_alpha_beta (2)=(a_alpha_upper);

A.7 MATLAB Code for Order Allocation to selected suppliers by genetic algorithm-I

***************************************************************************

Allocation of order to selected supplier/s – An example of single objective constrained optimization

***************************************************************************

% GA FITNESS FUNCTION %
124 Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes

function y = simple_fitness_supplier(x)
y=-.30561*x(1)-.38463*x(2)-.30977*x(3);

***************************************************************************

% CONSTRAINT OPTIMIZATION %
function [c,ceq]=constraint_supplier(x)
c=[30*x(1)+60*x(2)+35*x(3)-65000;0.01*x(1)+0.02*x(2)+0.04*x(3)-48];
ceq=x(1)+x(2)+x(3)-1200;

***************************************************************************
clc;
objectiveFunction=@simple_fitness_supplier;
nvars=3;
LB=[0 0 0];
UB=[650 650 550];
constraintFunction=@constraint_supplier;
options=gaoptimset('PopulationSize',20,'CrossoverFraction',0.8,'MutationFcn',{@mutationadaptfeasible,0.05});
options=gaoptimset(options,'PlotFcns',{@gaplotbestf,@gaplotdistance, @gaplotrange,@gaplotbestindiv
},'Display','iter','Generations',60);
[x,fval]=ga(objectiveFunction,nvars,[ ],[ ],[ ],[ ],LB,UB,constraintFunction,options)

A.8 MATLAB Code for Order Allocation to selected suppliers by genetic algorithm-II
**************************************************************************
Order allocation to selected supplier/s – An example of multi-objective genetic algorithm (MOGA)
**************************************************************************
function y = supplier_selection_multiobjective_fitness(x)
y(1)= 3859 * x(1)+ 3850 * x(2)+ 3851 * x(3);
y(2)= -.5654 * x(1)-.5024 * x(2)-.2033 * x(3) ;
y(3)= 0.1 * x(1) + 0.15 * x(2) + 0.2 * x(3);
y(4)= 0.2 * x(1) + 0.25 * x(3) + 0.3 * x(3);
y(5)=0.15 * x(1) + 0.2 * x(2) + 0.2 * x(3);

**************************************************************************
clc;
tic;
FitnessFunction=@supplier_selection_multiobjective_fitness;
numberOfVariables=3;
A=[2760 2750 2749]; b=[28000000];
Aeq=[1 1 1];beq=[9900];
lb=[0 0 0];
ub=[4000 3000 3000];

options=gaoptimset('PlotFcns',{@gaplotpareto});
options=gaoptimset(options,'PopulationSize',80,'HybridFcn',[],'CrossoverFraction',0.85,'CrossoverFcn',@crossoverarithmetic,'M
utationFcn',{@mutationadaptfeasible,0.5});
options = gaoptimset(options,'DistanceMeasureFcn',{@distancecrowding,'genotype'});
options = gaoptimset(options,'ParetoFraction',0.5,'Display','iter');
%options = gaoptimset(options,'PopulationSize',20);
[x,fval,exitFlag,Output,population,scores]=gamultiobj(FitnessFunction,numberOfVariables,A,b,Aeq,beq,lb,ub,options);
display(scores)
display(population)
display(fval)
fprintf('The number of points on the Pareto front was: %d\n', size(x,1));
fprintf('The average distance measure of the solutions on the Pareto front was: %g\n', Output.averagedistance);
fprintf('The spread measure of the Pareto front was: %g\n', Output.spread);
disp(['Elapsed time to solve multi-objective GA is ', num2str(toc)]);
Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes 125

A.9 MATLAB Code for Singular Value Decomposition method for fuzzy rule base reduction method
***************************************************************************
This computer code is developed on SVD method to reduce fuzzy rule base
***************************************************************************
function Z = Single_Value_Decomposition()
%UNTITLED Summary of this function goes here
% Detailed explanation goes here
clc
A=[1.2857 1.2857 4 4;1.2857 4 4 6;4 4 6 6;4 6 6 8.7143] %stage-1 sustainable
%supplier selection
%A=[-10 -7.5 -5 -2.5 0;-7.5 -5 -2.5 0 2.5;-5 -2.5 0 2.5 5;-2.5 0 2.5 5 7.5;0 2.5 5 7.5 10]
%A=a;
%A=[0.8 0.8 2.5 4.5;0.8 2.5 4.5 5.5;2.5 4.5 5.5 7.5;4.5 5.5 7.5 9.2]
rank_A=rank(A);
disp(rank_A)
[U S V]=svd(A);
disp('U matrix')
disp(U)
disp('V matrix')
disp(V)
disp('Diagonal Matrix of Singular Value')
disp(S)
disp('Rank of Diagonal Matrix of Singular Value')
K=rank(S);
disp(K)
if (K>=2)
K=2;
disp('2 input is considered')
end
M=size(U);
N=M(2);
B=size(U);
l=B(1);
UR=U(:,1:K);
UD=U(:,K+1:N);
Cu=blkdiag(sum(UR(:,1)),sum(UR(:,2)),1);%To form the diagonal matrix
disp(Cu)
UR(1,K+1)=0;
U1= UR*Cu;
min_U1=sort(U1);
Col_U1=size(U1);

if (min_U1>= -1)
delta=1;

else
delta= 1/mod(min_U1);

end

for i=1:1:K
for j=1:1:K

if (i==j)
126 Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes

stoch_matrix(i,j)=1+delta;
else
stoch_matrix(i,j)=1;
end
end

end

final_stoch_matrix =(1/(2+delta))*stoch_matrix;
U1_temp=U1(:,1:K);
U2=U1_temp*final_stoch_matrix;
disp('Prototypical Value')
disp(U2)
B=size(U2);
M=B(1);
*************************************************
%Prototypical Membership Value for FIRST Input%

EU=[U2(1,:);U2(M,:)];
%EU(K,1)=0;
EU_inv=inv(EU);
disp(EU_inv)
U3=U2*EU_inv;
disp('Membership Value for first input:')
disp(U3)

***************************************************
%Overlapping Membership Function: Matrix V%

%V=V';
VR=V(:,1:K);
MV=size(V);
NV=MV(2);
VD=V(:,K+1:NV);
Cv=blkdiag(sum(VR(:,1)),sum(VR(:,2)),1);%To form the diagonal matrix
VR(1,3)=0;
V1= VR*Cv;
min_V1=sort(V1);
if (min_V1>= -1)
deltaV=1;

else
deltaV= 1/mod(min_V1);

end

for i=1:1:2
for j=1:1:2

if (i==j)
stoch_matrixV(i,j)=1+deltaV;
else
stoch_matrixV(i,j)=1;
end
end

end
Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes 127

final_stoch_matrixV =(1/(2+delta))*stoch_matrixV;
V1_temp=V1(:,1:2);
V2=V1_temp*final_stoch_matrixV;

*******************************************************
%Prototypical Membership Value: FOR SECOND INPUT%

BV=size(V2);
V=BV(1);
EV=[V2(1,:);V2(V,:)];
EV_inv=inv(EV);
V3=V2*EV_inv;
disp('Membership Value for second input:')
disp(V3)

****************************************************
% Reduced matrix of rule consequent values%

disp('Inverse of Eu')
inv(EU_inv)
disp('Inverse of Du')
inv(final_stoch_matrix)
disp('Inverse of Cu')
Cu1=Cu(1:2,1:2);
inv(Cu1)
S(1:2,1:2)
disp('Inverse of Cv')
Cv1=Cv(1:2,1:2);
inv(Cv1)
disp('Inverse of Dv')
inv(final_stoch_matrixV)
disp('Inverse of Ev')
inv(EV_inv')
Z=inv(EU_inv)*inv(final_stoch_matrix)*inv(Cu1)*S(1:2,1:2)*inv(Cv1)*inv(final_stoch_matrixV)*inv(EV_inv');

end

A.10 VB.NET Code for database connection used to prepare DSS

*****************************************************
A simple VB.NET code to connect SQL SERVER with front end with ADO
*****************************************************
Imports System.Data
Imports System.Data.SqlClient
Imports System.IO
Imports System.Data.Common
Partial Class _Default
Inherits System.Web.UI.Page
Private connect As String
Private price(1000) As Double
Private temp1(100) As Double
Dim _count As Integer

Protected Sub Page_Load(ByVal sender As Object, ByVal e As System.EventArgs) Handles Me.Load

Dim cl As New Class_connect
connect = cl.init_db_con
If Page.IsPostBack = False Then
Button1.Text = "Load Data"
End If
If Page.IsPostBack = True Then
Button1.Text = "Save"
128 Appendix: MCDA Tools and Meta-Heuristic Techniques: Sample Codes

Button1.Enabled = False
Dim con As New SqlConnection(connect)
con.Open()
Dim sql As String
sql = "Select Name,Prod_Id,Prod_Des,Price,Quantity from Supplier where Prod_Id=1002"
'Paramteric SQL connection
Dim myadap As New SqlDataAdapter(sql, con)
Dim myds As New DataSet()
myadap.Fill(myds, "Orders")
con.Close()
'Only the name of the field of database is required
DDgrdVw.DataSource = myds.Tables("Orders")
'DDgrdVw.DataValueField = "Prod_Id"
DDgrdVw.DataBind()

DDgrdVw.SelectedIndex = -1
Dim i As Integer
'Dim j As Integer
Dim _maxValue As Double
Dim sum As Double
_count = myds.Tables("Orders").Rows.Count - 1
Dim temp(_count) As Double
'Dim temp1(j) As Double
ReDim temp1(_count)
ReDim price(_count)
For i = 0 To myds.Tables("Orders").Rows.Count - 1
price(i) = CDbl(myds.Tables("Orders").Rows(i).Item("Price").ToString)
temp(i) = price(i)
Next

temp1 = temp
Array.Sort(temp1)
_maxValue = temp1(_count).ToString
sum = 0.0

For i = 0 To _count
temp1(i) = _maxValue - price(i)
sum = sum + temp1(i)
Next
For i = 0 To _count
temp1(i) = temp1(i) / sum

Next

End If

End Sub

Protected Sub Button1_Click(ByVal sender As Object, ByVal e As System.EventArgs) Handles Button1.Click

Dim st As Stream = File.Open("C:\price.xls", FileMode.Create, FileAccess.Write)
Dim bw As New StreamWriter(st)
Dim i As Integer
For i = 0 To _count
Dim str1 As String
str1 = CDbl(temp1(i).ToString)
bw.Write(str1)
bw.WriteLine()
Next
bw.Close()
End Sub
End Class