AICTE-SPONSORED QIP SHORT-TERM COURSE

ON

SUPPLY CHAIN MANAGEMENT

(03.10.07 – 07.10.07)

LECTURE NOTES

Prepared By

Prof. S. P. Sarmah Prof. M. Jenamani

Principal Coordinator Co-coordiantor

Department of Industrial Engineering and Management

Indian Institute of Technology, Kharagpur
West Bengal- 721 302
 Contents

Part A: Introductory Topics

I Introduction to Supply Chain Management 1

II Materials Management 11

III Sourcing Decisions 17

IV Bullwhip Effect and Supply Chain Management 27

V Distribution Management Overview 39

Part B: Advanced Topics

VI Supply Chain Management and Multi-echelon Inventories 55

VII Supply Chain Contract and Coordination 65

VIII A Method for Supply Base Rationalization Considering Supply Risk 79

IX E-procurement 103

X Economic Theory of Auctions 113

XI Technologies for Supply Chain Integration 127

XII Security and Payment Issues in Integrated Supply Chain 141

XIII Automatic Data Capture using RFID and its Implications 155

XIV Dynamic Vehicle Routing with GPS and GIS 175

XV Optimization of Supply Chain Network: Simulation, Taguchi, and Psychoclonal 189

Algorithm Embedded Approach
 PART A

INTRODUCTORY TOPICS
 CHAPTER I
Introduction to Supply Chain Management
-An Overview

1.1 Introduction
A Supply Chain encompasses all activities in fulfilling customer demands and requests. These
activities are associated with the flow and transformation of goods from the raw materials stage,
through to the end user, as well as the associated information and funds flows. There are four stages in
a supply chain: the supply network, the internal supply chain (which are manufacturing plants),
distribution systems, and the end users. Moving up and down the stages are the four flows: material
flow, service flow, information flow and funds flow.
Different entities of the suply chain may be owned by one individual/organization or by several
individuals/organizations. Most supply chains of today belong to the later category. In such supply
chains, the owner of each entity attempts to maximize its benefit. Focus on individual links of the
supply chain invariably leads to inefficient and high cost product/service delivery system. In the
process, such a supply chain looses to supply chain that is customer focussed where the individual
links orient their business processes and decisions to ensure least cost delivery of products/services to
the ultimate customer.

1.2 Increasing role of supply chain management

In the last two decades, both academicians as well as practitioners have shown keen interest on the
subject supply chain management (SCM). Globalization of market, increased competition, reducing
gap between products in terms of quality and performance are compelling the academicians and
industry to rethink about how to manage business operations more efficiently and effectively. Since,
scope for improvement within the organization is decreasing, the academicians and captains of
industry are looking for newer alternatives of integrating the business activities beyond the
organization’s boundary. More specifically, they are trying to align and coordinate the business
processes and activities of the channel members to improve the overall performance and effectiveness
of supply chain. As a result, producer, vendor and buyer have started aligning their operations to make
the business more focussed. The alignment and integration lead to deliver more value and satisfaction
to the customer for the same price. This makes the supply chain more competitive. In the process, the
channel partners increase their market share and profit.
In principle, all the steps from procurement of raw materials to final delivery of products can be
included into a supply chain, connecting raw materials suppliers, manufacturers, distributors and
finally customers. Thus, a supply chain can be viewed as a group of entities interacting to transform
raw material into finished product and then final delivery of the product to the customer. Each
member of the chain provides some activity necessary for the transformation (value addition) and
interactions among the members can take place in the form of information, material and money flow.
Narasimhan and Carter (1998) in their work have mentioned that a well-integrated supply chain
involves coordinating the flows of materials and information between suppliers, manufacturers, and
customers. Effective supply chain management requires planning and coordination among the various
channel members including manufacturers, retailers and intermediaries if any (see Thomas and
Griffin, 1996). Due to severe competition in the market, to-day companies are more focusing on their
core competencies and therefore, increased cooperation between the members of the channel and
coordination of decisions are important. For coordination between various parties to be effective, faith
and effective communication between the members are essential.
Bowersox and Closs (1996) put forward argument that to be fully effective in today’s competitive
business environment, firms must expand their integrated behavior to incorporate customer and
supplier and refer this extension of integrated behaviors through external integration as supply chain
management. Further, no organization has enough resources so that it can single handedly manage the
entire supply chain. Therefore, to compete in the market in the present day, the obvious choice is to
work in coordination with the other members of the channel.

1.3 Evolution and growth of SCM

Supply chain management passed through various stages of evolution in last two decades to reach the
present position. The present supply chain management thought could be traced back to early sixties
when Mallen (1963) in his theory developed within the framework of marketing, advocated for
extension of the organization to include all other members of the distribution channel. Croomi et al.
(2000) in their review paper have mentioned that initial development of SCM is along the lines of
managing physical distribution and transport systems. It used techniques such as industrial dynamics
and minimum total cost approach to distribution and logistics. However, when the focus on
opportunities for competitive advantage started shifting from inside the manufacturing plant to
develop relationship with supplier and then finally with customers, work on SCM research started
picking up and since 1980’s, the subject is receiving attentions of both academicians and practitioners.
Maloni et al. (1997) have mentioned that evolution of intra-firm functional integration has occurred
for most firms over the last few decades. SCM extends the concept of intra-firm functional integration
to inter-firm integration of all the firms in the supply chain. Stevens (1989) has shown that the
process of integration of supply chain undergoes following four phases.
(i) Initially, every department of business organization functions independently. In such an
environment, there is no coordination to achieve an overall customer service objective.
(ii) In the second phase, focus shifts to cost reduction without any consideration of
performance achievement. The business organization goes for functional integration
within the organization through materials management, manufacturing management and
distribution for smooth flow of goods.
(iii) Recognizing the importance of customer’s in the business, the third phase aims to
integrate those aspects which are directly under the control of company including the
outward goods management and thereby integrating supply and demand along the
company’s own chain.
(iv) Phase four is the extension of the integration to external activities and in the process;
company becomes customer oriented by linking customer’s purchasing activities with
company’s own procurement, manufacturing and subsequently marketing activities.
From the above discussion, it is clear that supply chain management has evolved around a customer
focused business vision. This focus has led firms to change their internal and external linkages and
capture synergy of inter-functional and inter-organizational integration and coordination.
When we talk about supply chain of two or more parties linked by flows of goods, information and
funds then some interpret that SCM research is essentially the same as that of multi-echelon inventory
research addressed way back in sixties by Clark and Scarf (1960). But, multi-echelon inventory theory
is mainly concerned with controlling the timing and quantity of material flows whereas SCM research
encompasses a much broader set of issues. As for example, the issue of transportation delay in multi
echelon inventory theory is studied via lead-time whereas SCM studies the alternative modes of
logistics (Tsay et al., 1999).

1.4 Definition of Supply Chain

In supply chain management literature, there are numerous definitions available indicating the
fact that still there is not one universally acceptable definition for supply chain. This is probably
because of the fact that authors in this area use the term supply chain rather loosely to cover a wide

2
range of subject areas. This has however, helped supply chain management research to grow at much
faster rate.
The definition of supply chain is given in the eighth edition of APICS dictionary (1995) as “The
processes from initial raw materials to the ultimate consumption of finished product linking across
supplier-user companies.”
Houlihan (1985) is credited with coining the term Supply Chain and he mentioned some
characteristics unique to SCM. Jones and Riley (1985) defined supply chain as an integrative
approach to dealing with the planning and control of the materials flow from supplier to end-users.
Oliver and Webber (1992) state that a supply chain should be viewed as a single entity that is guided
by strategic decision-making. Stevens (1989) defines it as a connected series of activities from
supplier to customer. Villa (2001) has commented that in principle, all the activities from raw material
supplies to the final delivery of product to the customers can be included within the purview of the
supply chain.
From the large number of definitions available in the literature of supply chain, it is observed that
though they are apparently different from each other, yet, they are carrying more or less the same
meaning representing a system of supplier, manufacture, distributor, retailer and customer, where
materials flow downstream from supplier to customer whereas information and financial flows are bi-
directional.
Further, another important observation is that each definition has recognized that supply chain works
beyond the boundary of one organization and as a result a clear cut demarcation of boundary in a
supply chain is difficult since, where the boundary of one organization finishes, the boundary of the
other member of the supply chain starts.
Though above definitions of supply chain are mostly related with manufacturing organizations, yet,
there exists supply chain related with service organizations also. Some authors (e.g. Anderson et al.,
2000a, 2000b) have focused their study relating to supply chain of service organization.

Supply chain management

Supply chain management presupposes that there exists a supply chain, which needs to be managed
efficiently. As mentioned above, each entity of the supply chain performs a specific activity and
therefore, in a supply chain, there will be number of activities. Thus management of the supply chain
means coordination of activities among the entities as one system. Thomas and Griffin (1996) have
mentioned that supply chain management strategy is to coordinate the various organizations’
objectives in order to increase the efficiency of the entire supply chain.
Therefore, one major challenge of supply chain management is the development of effective
coordination mechanisms between the various entities of supply chain that ultimately lead to
reduction of lead time, cost and uncertainty. Time, cost and uncertainty are important (though in
varying degrees) to all supply chains. For improving the profit of the firm, SCM must focus in
effective and efficient management of materials, information and financial flows in the supply chain.
Ellram and Cooper (1990) have mentioned that since SCM extends the concepts of functional
integration beyond a firm to all firms in the supply chain, therefore, each member of supply chain
should help each other to improve the competitiveness of the chain. According to Christopher (1992),
the real competition is not between a company and another company, but rather between a supply
chain and another supply chain. A supply chain as a whole may have its own identity and function
like an independent firm.
The council of logistics management defines supply chain management as
“The systematic strategic coordination of the traditional business functions and the tactics across these
business functions within a particular company and across businesses within the supply chain for the

3
purposes of improving the long term performance of the individual companies and the supply chain as
whole”
Lee and Billington (1992) mentioned that supply chain management focuses in the coordination of the
manufacturing, logistics, and materials management functions within an organization. They have
defined supply chain management
“The integration activities taking place among a network of facilities that procure raw materials,
transform them into immediate goods and then final products, and deliver products to customers
through distribution system.”
Mentzer (2001) has mentioned SCM as the systematic, strategic coordination of the traditional
business functions within a particular company and across business within the supply chain for the
purpose of improving the long-term performance of the individual companies and the supply chain as
a whole.
New and Pyne (1995) have described supply chain management as the chain linking each element of
the manufacturing and supply process from raw materials through to the end user, encompassing
several organizational boundaries.
Further, La Londe (1997) elaborating the activities of SCM states that it is the process of managing
relationships, information, and materials flow across enterprise borders to deliver enhanced customer
service and economic value through synchronized management of the flow of physical goods, money
and associated information from sourcing to consumption.
Table 1.1 shows the important components of a supply chain management function. These
components can be grouped into two main business processes. They are: Materials Management and
Physical Distribution Management.
As a whole, supply chain management is a set of approaches used to integrate various activities of
suppliers, manufacturers, warehouses efficiently, so that merchandise is produced and distributed at
the right locations and the right time to minimize system-wide costs while satisfying customers’
service level requirements. Successful integration of various processes depends on the accurate and
timely sharing of information by all members of the supply chain. Due to various activities involved
in a supply chain, there may be multiple stakeholders (e.g. suppliers, manufacturers, distributors,
retailers and customers) in a supply chain.
From the above-mentioned views of various authors, it is clear that the scope of SCM is not only
confined to functional activities but also organizational. The functional scope includes a broad range
of traditional business functions, whereas the organizational scope is concerned with relationship
issues important to the participating firms. There is a need to develop a partnership relation between
the participating firms to reap the full benefit of SCM. Therefore, firms must take steps to break down
both intra and inter firm barriers to smoothen uncertainty and to improve control over distribution
channels. From intra-firm functional integration to external integration is the demand of the situation
and supply chain partnership can bridge the gap between the buyer and the supplier.

4
Table 1.1 Components of Total Supply Chain Management (Source: Monackza et al., 2002;
Min and Zhou, 2002)

Inbound Production Outbound

Materials management Work in Finished Physical distribution

function process goods function
inventory inventory
• Sourcing activities • Order receipt and
processing
• Inbound transportation
• Pricing
• Receiving and
warehousing • Promotional support
• Production planning • Outbound transportation
and scheduling
• Return product handling
• Materials control
• Field warehousing
• Intra and inter plant
• Customer service and
movement
material availability
• Quality control
• Finished goods delivery
• Inventory control

Materials Management Physical distribution

Supply chain Management

1.5 Characteristics of supply chain

There are some characteristics unique to supply chain management that differentiates it from earlier
researches on integrated logistics. Houlihan (1985) has mentioned that inclusion of strategic decision-
making aspect is one of the key characteristics that differentiate it from earlier research of integrated
logistics. Ganeshan et al. (1999) while referring to characteristics of SCM have mentioned that supply
chain management reaches out beyond the boundaries of cost containment and links operating
decisions to strategic considerations within and beyond the company and in channel wide supply
chain. Thus, SCM as a management philosophy has the following characteristics
• It is a system approach that considers the channel as a whole to manage the flow of goods
from the supplier to the ultimate customer.
• A strategic orientation towards cooperative efforts to synchronize and converge intra-firm
and inter-firm operational and strategic capabilities into a united whole.
• A customer focus to create unique and individualized sources of customer value, leading
to customer satisfaction.

5
In adopting a supply chain management philosophy, firms must establish management practices that
permit them to act or behave consistently with the philosophy. Some of the activities necessary for
implementing SCM philosophy can be mentioned as follows
• Integration of processes
• Mutually aggreable goal with same focus of customer service
• Mutual sharing of information
• Mutual sharing of channel risks and benefits
• Development of partnership to maintain long term relationship and cooperation.

1.6 Centralized vs. decentralized supply chain

Generally, one encounters supply chains with two different structures. In one structure of supply
chain, all upstream and down stream members of the chain have same owner. In this type of supply
chain, there will be perfect information regarding cost structure and demand pattern of the members.
In such chains, a central planner has the power to impose a globally optimal solution that is
implemented by all memebers.
On the other hand, in the other type of supply chain structure, different entities involved in the supply
chain have different ownership. In such a structure, there are many independent decision makers. This
supply chain is fundamentally different from the first supply chain in two aspects: First, there is
information asymmetry as a party in the chain may not be willing to share information regarding
his/her cost structure and demand. Secondly, the objective of a member of the chain may be different
from another member.
The second kind of supply chain as mentioned above is mostly prevalent in to-days business
environment. This class of supply chains is receiving attention of researcher and practitioner. Planning
and coordination issues of such supply chains have attracted attention of researchers as well as
practitioners (Ertogral et al., 2001). Such supply chains may be termed as decentralized chains.
Differences in focus of individual channel members make such supply chain more costly and less
effective. Decentralized supply chains are less efficient compared to centralized (single owner) supply
chains. Much of the current researches focus on how this gap of inefficiency can be reduced by
implementing novel coordination mechanisms (e.g. contracts). Tsay, Nahmias and Agarwal (1999)
refer to this as system wide performance improvement objective. The most commonly used term in
the literature for system wide performance improvement is the channel coordination.

1.7 Important elements of supply chain management

Supply chain management can broadly be divided into the following four elements
(i) Purchasing elements: Purchasing is an extremely important element in supply chain management
since incoming material quality, delivery time and purchase price are dependent on buyer supplier
relationship and capabilities of the supplier. Problems with suppliers will ultimately affect the end
customers. One of the crucial issues in purchasing is supplier management. This involves assessing
supplier’s current capabilities and how can it be improved. One of the key activities in supplier
management is supplier evaluation. This occurs both when potential suppliers are being evaluated for
a future purchase and when existing suppliers are periodically evaluated for performance purpose.
Over time, careful and effective supplier management efforts allow firms to selectively screen out
poor performing suppliers and build successful, trusting relationships with the remaining top
performing supplier. These suppliers can provide immense benefits to the buying firm and the entire
supply chain.

6
(ii) Operations Elements
Once materials, components and other purchased products are delivered to the buying organization, a
number of internal operations elements become important in assembling and or processing the items
into finished products, ensuring that the right amount of product is produced and that finished product
meets specific quality, cost and customer service requirements. When actual demand does not
materialize with forecasted demand then the firm is left with either too much inventory or not enough.
In both the situations, firm is incurring cost. To minimize these costs, firms rely on demand
management strategies. The objective of the system is to match demand with the available capacity,
either by improving production scheduling, curtailing demand, using a back order system, or
increasing capacity. Further controlling or managing inventory is one of the most important aspects of
operations and certainly valuable to the firm. Firms can and typically do have some sort of material
requirement planning software system for managing their inventory. These systems can be linked
through out the organization and its supply chain partner using enterprise resource planning systems
providing real time sales data, inventory, and production information to supply chain partners.
(iii) Distribution elements
The finished good are delivered to customer a number of different nodes of transportation. Delivering
products to customers at right time, right quality, and right time and in right volume require a high
level of planning and cooperation between the firm, its customers, and the various distribution
elements. Transportation management decisions involve a tradeoff between cost and delivery timing
and customer service. In order to provide the desired level of customer service, firms must identify
customer requirements and then provide the right combination of transportation, storage, packaging,
and information services to successfully satisfy those requirements. Further, designing and building a
distribution network is one method of ensuring successful product delivery. Again there is a trade off
between the cost of the distribution system’s design and customer service.
(iv) Integration Elements
Activities in supply chain are said to be coordinated when members of the supply chain work together
while making delivery, inventory, production, and purchasing decisions that impact the profit of the
supply chain. Successful supply chain integration occurs when the participants realize that supply
chain management must become part of all of the firms’ strategic planning processes, in which
objectives and policies are jointly determined on the basis of final customers’ needs and what the
supply chain as a whole does well. Finally, firms act together to maximize total supply chain profits
by determining optimal purchase quantities, product availabilities, service levels, lead times and
production quantities.

1.8 Partnership formation and vendor managed inventory

Focus on external integration as means to minimize rise of cost naturally led to the development of
partnership relation amongst the player of a supply chain. This trend did not escape the attention of
the supply chain management researchers. Parlar and Weng (1997) have mentioned that long-term
relationship makes both parties of the supply chain better off. However, Boddy et al. (1998) have
mentioned that a change from traditional unfavorable relationship between the customer and the
supplier to one of closer partnering requires careful consideration and attention of management.
Kotler (1997) observed that as firm globalize, they realize that no matter how large they are, they lack
the total resources for success of a complete supply chain that produces value. They recognize the
necessity of partnering with other organizations in the supply chain. In a comprehensive review paper
on supply chain partnership, Maloni and Benton (1997) have mentioned that a supply chain
partnership is a relationship formed between two independent entities in the supply channel to achieve
specific objectives and benefits. From this definition, one can assume that partnership relationship
between the members increases level of information sharing. This ultimately helps in improving the
overall performance of the supply chain through reduction in inventories and reduction in total cost of
managing the supply chain. Besides, one looks ahead to form partnership with the supplier to have

7
better customer service, better capacity utilization, technological innovation, and to new products. A
number of authors (e.g. Ellram, 1991) are quite optimistic about the success of supply chain
partnership. Ellram (1991) has mentioned that the central theme of these partnership relationships is
the establishment of, and commitment to, an interactive exchange where both parties benefit from
sharing of risks and resources.
Further, supply chain management to be effective, it requires mutually sharing of channel risks and
rewards among partners (Ellram and Cooper, 1990). This assures competitive advantage in the long
run. In a complex relationship in which performance is difficult to measure, profit or income sharing
based on incentive scheme is an important cooperation mechanism. Again, how to share the extra
benefit due to cooperation is an important issue. Probably a win-win approach to sharing of benefit
between parties is essential for cooperation. For sharing of benefits, many a times, a negotiation
process may be needed where both the parties are free to exchange views.
According to Rubin and Carter (1990), negotiation is the process of reviewing, planning and
analyzing used by two parties to reach acceptable agreement or compromise. It is a process where
both parties adjust their expectations during the resolution of conflict, as one party does not have
absolute power over the other party. In partnership relationship, it is important to note that each party
must be rational. Negotiation building is integral to successful long-term business relationship
(Sharland, 2001).
In the negotiation process, it is also important what to say and how to say since; it may have
significant impact on the outcome of the process. When one party inappropriately presents the things
to the other party, then the later party may perceive that long-term relationship may not be feasible
with the former party. Kelle et al. (2003) have recently studied buyer supplier partnership in JIT
environment and developed models that can be used as quantitative tools for contract negotiation
between the two parties. This could be done through either price correction or through price premium.
Another important facet of cooperation is information sharing, an essential enabler to minimize
inventory in the supply chain. Information systems must be able to track and communicate production
and customer requirements at different levels in the supply chain (Cooper, Lambert and Pagh, 1997).
Despite the perceived benefits of cooperation, it is seen that there is hardly perfect cooperation as
dominant player tend to be opportunistic. Munson et al. (1999) have nicely elaborated about misuse of
power by channel leader. Sometimes, it is noticed that a powerful manufacturer in the channel
controls dependent suppliers, subcontractors and retailers.
Another new dimension to the supplier-buyer partnership is the vendor managed inventory concept. In
vendor managed inventory (VMI) strategy, supplier manages inventory at customer’s premises and
assumes responsibility to replenish the inventory to meet the needs of the buyer who withdraws items
as per his/her requirements. In this strategy, supplier takes decision on inventory replenishment
without waiting for the customer to order the product. Recently, few articles as mentioned below have
cited the benefits of vendor-managed inventory or supplier owned inventory (SOI). VMI helps in
supply chain coordination leading to improvement in the performance of a supply chain.
Dong and Xu (2002) in their work have stated that VMI is an effective strategy of realizing most of a
fully coordinated supply chain. Piplani et al. (2003) have used the term supplier owned inventory
(SOI) instead of VMI. In their study on the effect of SOI strategy on the cost of a supply chain
concluded that total cost in SOI is never be more than the non-coordinated supply chain. Some other
authors such as Cetinkaya and Lee (2000), Waller et al. (1999), Hung et al. (1995) have also studied
the VMI strategy and its benefits. From the study of VMI literature, it is seen that by implementation
of VMI strategy, a buyer reduces total inventory related cost. However, whether VMI reduces the
supplier cost is still an open question.

1.9 Future trend in supply chain management

The practice of supply chain management is recent phenomenon, as many organizations are just now
beginning to realize the benefits and problems that accompany an integrated supply chain. Supply

8
chain management is an incredibly complex undertaking involving cultural change among most or all
of the participants, investment and training in software and communication system, and realignment
of the competitive strategies employed among the participating firm. In the competitive business
environment products, technology and customers change and subsequently the priorities for the
supply chain must also change, requiring supply chains to be ever more flexible to respond quickly to
these changes. The future issues for the supply chains that need to be addressed include increasing
supply chain responsiveness, creating an environment friendly supply chain, and reducing total supply
chain cost.
(i) Supply chain expansion
The supply chain dynamic to day is changing and companies are now working with firms located all
over the globe to coordinate purchasing, manufacturing, shipping and distribution activities. While
this global expansion of the supply chain is occurring, firms are also trying to expand their control of
the supply chain to include second and third-tier suppliers and customers. Thus supply chain
expansion is occurring in two fronts : increasing breadth of the supply chain to include foreign
manufacturing, office and retail sites, along with foreign suppliers and customers; and increasing the
depth of the supply chain to include second and third tier suppliers and customers. As the firm
becomes more comfortable and experienced with their supply chain relationships with immediate
suppliers and customers, there is a tendency to expand the depth of the supply chain by creating
relationships with second and third tier suppliers and customers. This span expansion phenomenon is
just now taking place in most industries and will continue to increase as the practice of supply chain
management matures.
(ii) Increasing supply chain responsiveness
Agile manufacturing, JIT, mass customization, efficient consumer response and quick response are all
terms referring to concepts that are intended to make the make the firm more flexible and
responsiveness to customer requirements and changes. In to-days intense competitive market
environment firms and their supply chains are looking today at ways to become more responsive to
their customers. To achieve greater levels of customer responsiveness, supply chains must identify the
end customers’ needs and position the supply chain’s products and services to successfully compete,
and then consider the impact of these requirements on the supply chain participants and the
intermediate products and services they provide. Once these issues have been adequately addressed
among the firms in the supply chain, additional improvement in responsiveness comes from designing
more effective and faster product and service delivery systems as the products are passed through the
supply chain and by continuously monitoring changes occurring the market place and using this
information to reposition the supply chain to stay competitive.
To improve customer responsiveness, firms require to reevaluate their supply chain relationships, to
utilize business process reengineering, to reposition warehouses, design new products and services,
reduce new product design cycles, standardize processes and products, empower and train workers on
multiple skills, build customer feedback into daily operations, and, finally, link together all of the
supply chain participants’ information and communication systems. To day web based systems are
proving to be ideal for connecting supply chain members efficiently. One such tool is Formation
systems’ Optiva 4.0. a web based product life cycle management platform that provides business
intelligence and collaboration from product concept through introduction to improvement. It can be
integrated within a supply chain to help product gets to market faster.
(iii) Greening of supply chains
Producing, packaging, moving, storing, repackaging, and delivering products to their final
destinations can pose a significant threat to the environment in terms of discarded packaging
materials, carbon monoxide emission, noise, traffic congestion, and other forms of industrial
pollution. As the practice of supply chain management becomes more widespread, firms and their
supply chain partners will be working harder to reduce these environmental problem.
***********

9
References
[1] Bowersox, D.J. and Closs, D.J., 1996, Logistical management: The integrated supply chain
process, New York, NY: Mc Graw-Hill
[2] Christopher, M., 1992, Logistics and Supply Chain Management: Strategies for reducing costs
and improving services, Publisher: FT Pitman Publishing.
[3] Croomi, Simon, Roamno, P. and Giannakis, M., 2000, Supply chain management: an analytical
framework for critical literature review, European Journal of Purchasing and Supply
Management, Vol. 6, No. 1, pp. 67-83.
[4] Cetinkaya, S. and Lee, C.Y., 2000, Stock replenishment and shipment scheduling for vendor
managed inventory systems, Management Science, Vol. 46, No. 2, pp. 217-232.
[5] Dong, Y. and Xu, K., 2002, A supply chain model of vendor managed inventory, Transportation
Research Part E, Vol. 38, No. 2, pp. 75-95.
[6] Ertogral, K., and Wu, D. S., 2001, A bargaining game of supply chain contracting, Source internat
website www.lehigh.edu/sdw1/ertogral3.pdf
[7] Houlihan, J.B., 1985, International supply chain management, International Journal Of Physical
Distribution and Material Management, Vol. 15, No. 1, pp. 22-38.
[8] Kelle, P., Khateeb, F. and Miller. A.P., 2003, Partnership and negotiation support by joint optimal
ordering/set up policies for JIT, International Journal of Production Economics, Vol. 81-82, pp.
433-443.
[9] Lee, H.L. and Billington, C., 1993, Material management in decentralized supply chains,
Operations Research, Vol. 41, No. 5, pp. 835-847.
[10] Maloni, J.M. and Benton, C.W., 1997, Supply chain partnership: Opportunities for operations
research, European Journal of Operational Research, Vol. 101, No. 3, pp. 419-429.
[11] Mentzer, T.J., 2001, Supply Chain Management, Sage Publisher.
[12] Min, Hockey., Zhaou, G., 2002, Supply chain modelling: past,present and future, Computers
and Industrial Engineeering, Vol. 43, No. 1-2, pp. 231-249
[13] Monczka, R., Trent, R. and Handfield, R., 2002, Purchasing and Supply Chain Management,
Second Edition: publisher: Thomson Asia Pte Ltd. Singapore
[14] Narasimhan, R., Carter,J.R., 1998, Linking business unit and material sourcing strategies.
Journal of Business Logistics. Vol. 19, No. 2, pp. 155-171
[15] Piplani, R. and Viswanathan, S., 2003, A model for evaluating supplier owned inventory
strategy, International Journal of Production Economics, Vol. 81-82, pp. 565-571.
[16] Thomas, D. J. and Grifin, P. J., 1996, Coordinated supply chain management, European Journal
of Operational Research, Vol. 94, No. 1, pp. 1-15.
[17] TSay, A., Nahmias,S., & Agarwal,N., 1999, Modeling supply chain contracts: A review, In: S.
Tayur, M. Magazine, R. Ganeshan, (Eds.), Quantitative models for supply chain management,
Published by Kluwer academic publishers, 1999, pp. 301-336
[18] Waller, M., Johnson, M.E., Davis, T., 1999. Vendor-managed inventory in the retail supply
chain. Journal of Business Logistic, Vol. 20, No. 1, pp. 183-203.

10
 CHAPTER II
Materials Management

2.1 Introduction
Putting in the simplest terms materials management is about moving the materials within an
organization. What do “materials” mean? Materials can basically be defined as those objects or things
that are to be moved in order to produce goods. Material is one of the 5M’s that a manager has at his
command, the other being Men, Machine, Methods and Money. Materials could be in the form of raw
materials, paperwork, messages or information etc. So materials can be both tangible and intangible.
You see the newspaper boy delivering the newspaper to your doorstep everyday or the mi1kman
delivering the milk packets to you. These are tangible materials. There is also some material moved
when you watch a movie on your television or when you receive a phone call. These are the intangible
materials that are moved. So materials management is an important function of every business. The
better is the materials management in a company the better is the health of that company.

2.2 Materials management and its functions

Materials can be put in three categories. First category is purchased materials like the raw materials,
components, spare parts and items that are used and do not appear in the end product. The second
category is of in-process materials or the materials in the semi-finished stages and lastly the finished
goods that are ready for customers. One has to manage these materials. The aim of this management is
to obtain the materials at the minimum possible price while maintaining quality also and to maintain
the inventories in such a way that minimum cost is incurred while maintaining adequate materials for
the production process.
Let us see what materials management actually means. It is defined as a function that integrates
purchasing, storage, inventory control, materials handling and standardization etc in an organization
to achieve its objective of reducing the costs. Every organization wishes to maximize its profit by
maximizing its production and minimizing the cost of production. The average material cost in a
manufacturing setup is around 50-70% of the total expenditure, which further goes up if one takes into
account the inventory costs, storage, waste and other factors etc. It is therefore imperative for an
organization to have a sound materials management with an objective to reduce material costs, control
inventories, ensure uniform flow of materials and maintain good relations with suppliers. Materials
Management has to do activities related to planning, accusation and utilization of materials.
Materials Management as a subject started picking up from early sixties and has gained importance
thereafter. Since the amount of money incurred on materials is higher than the cumulative amount for
machines, men and methods, one has to give high importance to the materials. It is the most feasible
area that can offer opportunities for reduction of costs and improvement of profits. Materials add
value to the product, as the product quality is directly dependent on the materials used. Materials
Management thus can be seen as a system that assures the availability of products to the customers at
minimum cost. In a nutshell, we can say that materials management is about making available the
right materials in right quantity at a right price on the right time.
The functions of the materials management are materials planning and control, purchasing, inventory
control, store keeping, material handling, warehousing, standardization & simplification and
organization & appraisal of materials. Let us discuss them briefly.
1) Materials planning and control: Material requirement lies at the core of successful material
management. This function is at the core of all the material requirements in any manufacturing
process.
2) Purchasing: This function identifies the sources of supply, does market research, call tenders
and select suppliers, negotiate with them and thus make available the raw materials.
3) Inventory control: This function is responsible for the location and storage of materials so that
they remain available at the minimum cost and quickest time.
4) Store keeping: This function is responsible for the receipt and issue of the materials. The
materials are stored in such a way that minimum handling is required and wastage is minimal.
5) Material handling: This function aims at minimizing handling and provision of equipments for
handling materials. This function is crucial for minimizing space requirements, effective
distribution and for providing better working space.
6) Warehousing: This function is responsible for the storage facilities for the materials, weighing
facilities, materials handling equipments, material distribution facilities, fire fighting
instruments etc.
7) Standardization and simplification: This function selects items of great demand and sets the
standards for quality, raw material, sizes and performance of any product.
8) Organization & appraisal of materials: This function helps in effective functioning by
proving smooth flow. It provides coordination and avoid delays and wastages
Management of materials embodies various costs. Since the ultimate aim of materials management is
to reduce the costs of materials and hence the final product, it is worth seeing what these costs are. Let
us take a glimpse of what these costs are:

Table 1.1: Costs involved in the Management of Materials

Sl.
Costs Description
No.
1 Cost of materials The basic cost of materials that has to be paid to suppliers
The cost incurred in purchases e.g. cost on staff, tendering,
2 Purchasing cost
stationary, postage, processing supplies, receiving, inspection
The cost incurred on storage including buildings, costs on staff,
3 Inventory carrying costs
interest on capital locked/ borrowed, obsolescence
Costs incurred on paper, plastic, metal foils, metal and wood
4 Packaging cost
containers etc.
Costs incurred on moving the goods to different desired
5 Transportation cost
locations from time to time
Costs incurred on handling equipments like cranes and
6 Material handling cost
conveyors
Wastage during Costs incurred on holding scrap, obsolete stock and their
7
production disposal

Source: Shah N.M. (1996), An Integrated Concept of Materials Management

An integrated materials management system helps in taking judicious decisions that in turn leads to
lower cost for materials. Similarly if an organization has low inventory carrying costs, less stock outs
etc., it is bound to do well.

2.3 Management of flow of materials

In any organization, the responsibility for maintaining the quality of the product and incurring less
cost on its production is the responsibility of the production/ operation, deciding the price of the
product and finding the customers that will buy it comes under marketing. The question arises that if

12
it is so, what do the materials management function does? The answer is that from the time the
materials enter the warehouse of the organization from the suppliers, the role of materials
management starts and gets going till the final product is obtained. The interrelated activities that are
carried out to achieve this are sequenced after each other in a systematic manner. Management of this
flow of materials is called materials management. This flow of materials is met through a set of
activities presented in Table 1.2 given below
Table 1.2: Set of activities for flow of materials.

Sl. No. Activity Function

Setting the goals, indicating the sources of
1 Planning
finance

Requirements specification, quantum and

2 Scheduling
delivery schedules

3 Purchasing and Procurement Vendor selection, vendor contracts

4 Inspection and Quality Control Conforming quality

Determining inventories, maintenance and

5 Stores and Inventory Control
upkeep

Materials handling and

6 Controlling flows, distribution, shipments
distribution logistics

Source: Dutta A.K (1998), Materials Management: Procedures, Text and Cases

The table above highlights the importance of integrated systems and dependence of function models
for decision-making. The organizations have now become multidimensional in nature. Total materials
management concept evolved to address this dimension and avoid conflicting objectives. Total
material management helps in establishing accountability so that response to a problem is quick and
appropriate. The material functions are accomplished in more coordinated ways with the help of this
integrated approach. When this happens there is increased communication for the need of materials
and hence one gets lower costs, better inventory turnover, reduce stock outs and other significant
benefits. Data processing systems are designed on the basis of the integrated material function.

2.4 Materials logistics management

Materials logistics management program (MLM) started in eighties for an American university
management students to meet the industry requirements. Bowersox et al. (1984) presented an
overview of this philosophy.
Figure 1.1 shows the value-added activities included in MLM. There are two flows that are depicted
here. One is a requirements information flow from customers to suppliers and other is a value-added
materials flow from the suppliers to the customers. The whole process is directed by an integrated
database. MLM covers three essential areas required for moving materials i.e. purchasing,
manufacturing and physical distribution and asserts an integrated logic to ensure smooth flow of
materials. As can be seen in the figure the MLM seeks the achievement of objective like controlled
customer service performance, inventory reduction, minimum variance in planned operations,
minimum total cost of operations and procurement and product quality control.

13
 Value-added materials flow

Industrial enterprise

Physical
Customers distribution Manufacturing Purchasing Suppliers

Requirements information flow

* Demand management * Master schedule management * Supply management

* Scheduled distribution * JIT Scheduling * Scheduled requirements
* Postponement * Flexibility * Responsiveness
Target objectives
* Controlled customer service performance * Minimum Variance * Min total cost of
operations & procurement
* Inventory reduction * Product quality control
Figure 1.1: Materials Logistics Management Process
Source: Bowersox et al. (1984)
Bowersox et al. (1984) further described the three interfaces that MLM covers. These are the physical
distribution interface, the manufacturing interface and the purchasing interface. Let us see how these
interfaces contribute to the MLM productivity. The Table 1.3 summarizes this.
Table 1.3: The MLM productivity and various interfaces

Interface Interfaces with Perception 1 Perception 2 Perception 3

The Physical Customers and Demand Scheduled Postponement
Distribution manufacturing Management distribution
Interface
The Physical Master schedule JIT scheduling Flexibility
Manufacturing distribution and management
Interface purchasing
The Purchasing Manufacturing Supply Schedule Responsiveness
Interface and external management requirements
supplier network
Source: Based on Bowersox et al. (1984)
There are some perceptions attached with each interface.
From the point of view of the physical distribution interface,
• Demand management coordinates and modifies how customers order products in an effort to
reduce uncertainty and simplify transactions,
• Scheduled distribution aims to fulfill customer order in a short span of time and
• Postponement carries a planned delay of an activity as long as possible until a profitable
preposition is achieved.
From the point of view of the manufacturing interface,
• Master schedule management resolves the conflicts between manufacturing and marketing as it
is the point where overall requirements determined by forecasts, customers orders, back orders

14
 and physical distribution are collated.
• Just in time (JIT) scheduling or Kanban means bringing inventories to zero level. To reduce the
inventory, methods like reducing lot sizes, load leveling, quality control and preventive
maintenance can be used.
• Flexibility should be achieved by using “pull” systems, computer-based planning and control
systems. Achieving flexibility is important to manufacturing as it will reduce manufacturing
activities unless or otherwise specifically asked for.
From the point of view of the purchasing interface,
• Supply management identifies the manufacturing trends and initiates effective purchasing for
long-term competitive advantage.
• Schedule requirements expedite purchasing. They must be specified so that suppliers provide
exact lead-time information and purchasers provide exact requirement information to the supply
network. This can be achieved by employing a suitable integrated data-processing system.
• Responsiveness of the supply network identifies frequent changes in customer requirements and
product life cycles.

2.5 Interfaces of materials management

According to Dutta (1998), “When we say that materials management contains an integrated process
of materials flow, in, through and out of an organization, we give some indication that materials
management has interfaces of two kinds, internal and external”.
The interfaces are shown in Table 1.4
Materials management actually does not start with purchase of materials and end with production of
materials. The total financial management concept forces it to do much more. Dutta (1998) further
stresses the materials management functions as follows:
1) Materials forecasting, budgeting, planning and programming
2) Scheduling, purchasing and procurement
3) Receiving and receiving inspection as to quantity and quality
4) Inventory control, storage and warehousing
5) Materials handling, movement control and traffic etc.
6) Dispatch, shipping and disposal
In addition to this materials management also needs to put attention to coordinate all the above
activities and keep liaison between manufacturing, finance and marketing etc.

2.6 Materials and information flow

The aim of any organization is to manage its 5 M’s as effectively as it can. These 5 M’s as discussed
earlier are Men, Machines, Money, Methods and Materials. The purpose of this coordination is
production of superior goods at minimal costs. In this discussion, you have focused on the materials.
If you have to exercise proper control over materials then you have to take care of your material flow
as well as information flow.
There is a definitive flow of materials from vendor/supplier to the warehousing/customer and the
organization is abounded in information flow. This is very important as materials and information
both are extremely important and both should be readily available at a time when needed.

15
 Table 1.4: Various Interfaces of Materials Management
S.No. Interface Description
1 Market forecasting Forecast demands to determine production
on the basis of existing/ expanded facilities,
equipment, processes, manpower and
materials.
2 Production Materials flow begins before the production
cycle, runs throughout and continues even
after production, ensuring uninterrupted
flow of materials to feed the production
process. Internal
3 Finance Materials budget is affected by non-
availability of finance. A major chunk of
Internal

finance is invested in materials and

inventors.
4 Inventory control Materials management has a critical role
here as availability of materials and access
to physical supply has to be assured.

5 Inspection and quality Close liaison with materials management is

control required as it also runs throughout like
materials.
6 Materials Handling and Materials management ensures that
physical distribution materials are physically distributed at the
logistics right time with a minimum of handling.
External
7 Consumers/Customers Materials management interfaces with
customers rarely but it happens sometimes.
External

8 Suppliers/Other Materials Management is responsible for

Companies keeping a close liaison with outside vendors
and other companies seeking trade
relationships. Evaluation of supplier
performances should also be done.

Source: Based on Dutta A.K (1998), Materials Management: Procedures, Text and Cases

The materials flow is starting from the vendor/supplier from which material is to be purchased. Once
the material is purchased, it is received and inspected. After inspection stores accepts it. Production/
manufacturing and its subsystems ca1l for the materials as and when it is required and logistics take
control after that. Later on warehousing and customer comes.
Information flow embodies much more than the materials flow. Be it production planning and control
or sales and marketing, inventory control or purchasing and procurement activities. The effectiveness
of the materials flow is thus dependent on decision-information. If an organization can control these
two flows easily and effectively then it will definitely render goods products at a low cost and also
would be able to offer good service.

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

16
 CHAPTER III
Sourcing Decisions

3.1 Sourcing Decisions: The Make-or-Buy Decision

While the term outsourcing popularly refers to buying materials and components from suppliers
instead of making them in-house, it also refers to buying materials or components that were
previously made in-house. In recent years, the trend has been moving toward outsourcing combined
with the creation of supply chain relationships, although traditionally firms preferred the make option
by using backward integration and forward integration. Backward integration refers to acquiring
sources of supply, whereas forward integration refers to acquiring customers’ operations. For
example, an end-product manufacturer acquiring a supplier’s operations that supplied component
parts is an example of backward integration. Acquiring a distributor or other outbound logistics
providers would be an example of forward integration.
Whether to make or buy materials or components is a strategic decision that can impact an
organization’s competitive position. It is obvious that most organizations buy their MRO
(Maintenance, repair and operational) items and office supplies rather than make the items
themselves. Similarly, seafood restaurants usually buy their fresh seafood from fish markets.
However, the decision on how to acquire highly complex engineering parts that impact the firm’s
competitive position is a complicated one.
Traditionally, cost has been the major driver when making sourcing decisions. However,
organizations today focus more on the strategic impact of the sourcing decision on the firm’s
competitive advantage. For example, Honda would not outsource the making of its engines because it
considers engines to be a vital part of its automobiles’ performance and reputation. However, Honda
may outsource the production of brake drums to a high-quality, low-cost supplier that specializes in
brake drums. Generally, organizations outsource non-core activities while focusing on core
competencies. Finally, the make-or-buy decision is not an exclusive either-or option. Firms can
always choose to make some components or services in-house and buy the rest form suppliers.

3.2 Reasons for Outsourcing

Organizations buy or outsource materials, components, and /or services from suppliers for many
reasons. Let us review these now:
(i) Cost advantage: For many firms, cost is an important reason for buying or outsourcing,
especially for supplies and components that are non vital to the organization’s operations
and competitive advantage. This is usually true for standardized or generic supplies and
materials for which suppliers may have the advantage of economy of scale because they
supply the same item to multiple users. In most outsourcing cases, the quantity needed is
so small that it does not justify the investment in capital equipment to make the item.
Some foreign suppliers may also offer a cost advantage because of lower labor and/or
materials costs.
(ii) Insufficient capacity: A firm may be running at or near capacity, making it unable to
produce the components in-house. This can happen when demand grows faster than
anticipated or when expansion strategies fail to meet demand. The firm buys parts or
components to free up capacity in the short term to focus on vital operations. Firms may
even subcontract vital components and /or operations under very strict terms and
conditions in order to meet demand. When managed properly, subcontracting is an
effective means to expand short-term capacity.
(iii) Lack of expertise: The firm may not have the necessary technology and expertise to
manufacture the item. Maintaining long-term technological and economical viability for
 non core activities may be affecting the firm’s ability to focus on core competencies.
Suppliers may hold the patent to the process or product in question, thus precluding the
make option, or the firm may not be able to meet environmental and safety standards to
manufacture the item.
(iv) Quality: Purchased components may be superior in quality because suppliers have better
technology, process, skilled labor, and the advantage of economy of scale. Suppliers may
be investing more in research and development. Suppliers’ superior quality may help
firms stay on top of product and process technology, especially in high-technology
industries with rapid innovation.

3.3 Reasons for Making

An organization also makes its own materials, components, services, and/or equipment in-house for
many reasons. Some of the reasons are as follows:
(i) Protect proprietary technology: A major reason for the make option is to protect
proprietary technology. A firm may have developed an equipment, product, or process
that needs to be protected for the sake of competitive advantage. Firms may choose not to
reveal the technology by asking suppliers to make it, even if is patented. An advantage of
not revealing the technology is to be able to surprise competitors and bring new products
to market ahead of competition, allowing the firm to charge a price premium. For
example, Intel or Advanced Micro Devices are not likely to ask suppliers to manufacture
their new central processing units.
(ii) No competent supplier: If the component does not exist, or suppliers do not have the
technology or capability to produce it, the firm may have no choice but to make an item
in-house, at least for the short term. The firm may use supplier development strategies to
work with a new or existing supplier to produce the component in the future as a long-
term strategy.
(iii) Better quality control: If the firm is capable, the make option allows for the most direct
control over the design, manufacturing process, labor, and other inputs to ensure that
high-quality components are built. The firm may be so experienced and efficient in
manufacturing the component that suppliers are unable to meet its exact specifications
and requirements. On the other hand, suppliers may have better technology and processes
to produce better-quality components. Thus, the sourcing option ensuring a higher
quality level is a debatable question and must be investigated thoroughly.
(iv) Use existing idle capacity: A short-term solution for a firm with excess idle capacity is to
use the excess capacity to make some of its components. This strategy is valuable for
firms that produce seasonal products. It avoids laying off skilled workers and, when
business picks up, the capacity is readily available to meet demand.
(v) Control of lead-time transportation, and warehousing cost: The make option provides
easier control of lead-time and logistical costs since management controls all phases of
the design, manufacturing, and delivery processes. Although raw materials may have to
be transported, finished goods can be produced near the point of use, for instance, to
minimize holding cost.
(vi) Lower cost: If technology, capacity, and managerial and labor skills are available, the
make option may be more economical if large quantities of the component are needed on
a continuing basis. Although the make option has a higher fixed cost due to capital
investment, it has a lower variable cost because it precludes suppliers’ profits.

18
3.4 Roles of Supply Base
The supply base or supplier base refers to the list of supplies that a firm uses to acquire its materials,
services, supplies, and equipment. Firms engaging in supply chain management emphasize long-term
strategic supplier alliances by reducing the variety of purchased items and consolidating volume into
one or fewer suppliers, resulting in a smaller supply base. An effective supply base that complements
and contributes to a firm’s competitive advantage is critical to its success. Savvy purchasing managers
develop a sound supply base to support the firm’s overall business and supply chain strategies, based
on an expanded role for the supplier. It is thus vital to understand the strategic role of suppliers.
Besides supplying the obvious purchased items, preferred or top-performing suppliers also supply
(i) Product and process technology and knowledge to support the buyer’s operations, particularly
in product design – termed early supplier involvement;
(ii) Information on the latest trends in materials, processes, or designs;
(iii) Information on the supply market, such as shortages, price increases, or political situations
that may threaten supplies of vital materials;
(iv) Capacity for meeting unexpected demand; and
(v) Cost efficiency due to economies of scale, since the supplier is likely to produce the same
item for multiple buyers.
When developing and managing the supply chain, high-performance suppliers are found or developed
to provide these services and play a very important role in the success of the supply chain.

3.5 Supplier Selection

The decision of which supplier to use for office supplies or other non-critical materials is likely to be
an easy one. However, the process of selecting a group of competent suppliers for important
materials, which can potentially impact the firm’s competitive advantage, is a complex one and
should be based on multiple criteria. In addition to cost and delivery performance, firms should also
consider how suppliers can contribute to product and process technology. Factors that firms should
consider while selecting suppliers include:
(i) Product and process technologies: Suppliers should have up-to-date and capable
products, as well as process technologies to produce the material needed.
(ii) Willingness to share technologies and information: With the current trend that favors
outsourcing to exploit suppliers’ capabilities and to focus on core competencies, it is vital
that firms seek suppliers that are willing to share their technologies and information.
Suppliers can assist in new product design and development through early supplier
involvement to ensure cost-effective design choices, develop alternative conceptual
solutions, select the best components and technologies, and help in design assessment. By
increasing the involvement of the supplier in the design assessment. By increasing the
involvement of the supplier in the design process, the buyer is free to focus more attention
on core competencies.
(iii) Quality: Quality levels of the purchased item should be a very important factor in
supplier selection. Product quality should be high and consistent since it can directly
affect the quality of the finished goods.
(iv) Cost: While unit price of the material is not typically the sole criterion in supplier
selection, total cost of ownership is an important factor. Total cost of ownership includes
the unit price of the material, payment terms, cash discount, ordering cost, carrying cost,
logistical costs, maintenance costs, and other more qualitative costs that may not be easy

19
 to assess. Total cost analysis demonstrates how other costs beside unit price can affect
purchase decision.
(v) Reliability: Besides reliable quality levels, reliability refers to other supplier
characteristics. For example, is the supplier financially stable? Otherwise, it may not be
able to invest in research and development or stay in business. Is the supplier’s delivery
lead-time reliable? Otherwise, production may have to be interrupted due to shortage of
material.
(vi) Order system and cycle time: How easy to use is a supplier’s ordering system, and what
is the normal order cycle time? Placing orders with a supplier should be easy, quick, and
effective. Delivery lead-time should be short, so that small lot sizes can be ordered on a
more frequent basis to reduce inventory-holding costs.
(vii) Capacity: the firm should also consider whether the supplier has the capacity to fill orders
to meet requirements and the ability to fill large orders if needed.
(viii) Communication capability: Suppliers should also possess a communication capability
that facilitates communication between the parties.
(ix) Location: Geographical location is another important factor in supplier selection, as it
impacts delivery lead-time, transportation, and logistical costs. Some organizations
require their suppliers to be located within a certain distance from their facilities.
(x) Service: Suppliers must be able to back up their products by providing good services when
needed. For example, when product information or warranty service is needed, suppliers
must respond on a timely basis.
There are numerous other factors, some strategic while others tactical, that a firm must consider when
choosing suppliers. The days of using competitive bidding to identify the cheapest supplier for
strategic items are long gone. The ability to select competent strategic suppliers directly impacts a
firm’s competitive success. Strategic suppliers are trusted partners and become an integral part of the
firm’s design and production efforts.

3.6 Development of successful partnership

In the last two decades, we have learnt from the Japanese that good supplier relations can provide
many benefits such as flexibility in terms of delivery, better quality, better information, and better
material flows between buyers and suppliers.
True partnerships are not easily created and much has to be done to get the most out of any
partnership. Several key ingredients for developing successful partnerships follow.
(i) Building Trust
Trust is critical for any partnership or alliance to work. Trust enables organizations to share valuable
information, devote time and resources to understand each other’s business, and achieve results
beyond what could have been done individually. Jordan Lewis, in his book Trusted Partners, points
out that “Trust does not imply easy harmony. Obviously, business is too complex to expect ready
agreement on all issues. However, in a rusting relationship, conflicts motivate you to probe for deeper
understandings and search for constructive solutions. Trust creates goodwill, which sustains the
relationship when one firm does something the other dislikes.” With trust, partners are more willing to
work together, find compromise solutions to problems, work toward achieving long-term benefits for
both parties, and, in short, go the extra mile.
(ii) Shared Vision and Objectives
All partnerships should state the expectations of the buyer and supplier, reasons and objectives of the
partnership, and plans for the dissolution of the relationship. Both partners must share the same vision
and have objectives that are not only clear but mutually agreeable. Many alliances and partnerships
have failed because objectives were not well aligned or were overly optimistic. The focus must move

20
beyond tactical issues and toward a more strategic path to corporate success. When partners have
equal decision-making control, the partnership has a higher chance of success.
(iii) Personal Relationships
Interpersonal relationships in buyer-supplier partnerships are important since it is people who
communicate and make things happen.
(iv) Mutual Benefits and Needs
Partnering should result in a win-win situation, which can only be achieved if both companies have
compatible needs. Mutual needs create not only an environment conducive for collaboration but
opportunities for increased innovation. When both parties share in the benefits of the partnership, the
relationship will be productive and long lasting. An alliance is much like a marriage, and if only one
party is happy, then the marriage is not likely to last.
(v) Commitment and Top Management Support
First, it takes a lot of time and hard work to find the right partner. Having done so, both parties must
dedicate their time, best people, and resources to make the partnership succeed. Commitment must
start at the highest management level. Partnerships tend to be successful when top executives are
actively supporting the partnership. The level of cooperation and involvement shown by the
organization’s top leaders is likely to set the tone for joint problem solving further down the line.
Successful partners are committed to continuously looking for opportunities to grow their businesses
together. Management must create the right kind of internal attitude needed for alliances to flourish.
Since partnerships are likely to encounter bumps along the way, it is critical that management adopt a
collaborative approach to conflict resolution instead of assigning blame.
(vi) Change Management
With change comes stress, which can lead to a loss of focus. As such, companies must avoid
distractions from their core businesses as a result of the changes brought about by the partnerships.
(vii) Information Sharing and Lines of Communication
Both formal and informal lines of communication should be set up to facilitate free flows of
information. When there is a high degree of trust, information systems can be customized to serve
each other more effectively. Confidentiality of sensitive financial, product, and process information
must be maintained. Any conflict that occurs can be resolved if the channels of communication are
open. For instance, early communication to supplies of specification changes and new product
introductions are contributing factors to the success of purchasing partnerships. Buyers and sellers
should meet regularly to discuss any change of plans, evaluate results, and address issues critical to
the success of the partnerships. Since there is free exchange of information, nondisclosure agreements
are often used to protect proprietary information and other sensitive data from leaking out. It is not the
quantity but rather the quality and accuracy of the information exchanged that indicates the success of
information sharing.
(viii) Capabilities
Organizations that have a long history of using cross-functional teams to solve problems and who
have shown that their employees can collaborate successfully internally have the skills to do so
externally. We all know that things do not always turn out as planned. Thus, companies must be
willing to accept responsibility and have the capability to correct errors effectively when they are
detected. Key suppliers must have the right technology and capabilities to meet cost, quality, and
delivery requirements. In addition, suppliers must have the flexibility to respond quickly to changing
customer requirements. Before entering into any partnership, an organization must conduct a thorough
investigation of the supplier’s capabilities and core competencies. Organizations prefer working with
suppliers who have the technology and technical expertise to assist in the development of new
products or services that would lead to a competitive advantage in the marketplace.

21
3.7 Supplier Performance Evaluation
Performance Metrics
Measures related to quality, cost, delivery, and flexibility have traditionally been used to evaluate how
well supplies are doing. Information provided by supplier performance will be used to improve
efficiency in the entire supply chain. Thus, the goal of any good performance evaluation system is to
provide metrics that are understandable, easy to measure, and focused on real value-added results for
both the buyer and supplier.
By evaluating supplier performance, organizations hope to identify suppliers with exceptional
performance or developmental needs, improve supplier communication, reduce risk, and manage the
partnership based on an analysis of reported data. After all, it is not unusual that the best customers
want to work with the best suppliers. Additionally, the best suppliers are commonly rewarded and
recognized for their achievements.
In a survey of buyers carried out by Purchasing Magazine, although price/cost was rated the most
important factor when selecting suppliers, other criteria such as technical expertise, lead times,
environmental awareness, and market knowledge were also rated highly by the respondents. An
earlier study on the electronics industry by Dr. Pearson and Dr. Ellram showed that quality was the
most important criterion for selection, followed by cost, current technology, and design capabilities. It
would appear that in the electronics industry, which pioneered the six-sigma revolution, quality is the
prime selection criteria due to its strategic importance. Thus it is seen that a multi-criteria approach is
needed to measure performance. Examples of broad performance metrics are shown in Table 3.1.
Over the past several years, total cost of ownership (TCO), a broad-based performance metric, has
been widely discussed in the supply chain literature. TCO is defined as “all costs associated with the
acquisition, use, and maintenance of a good or service” and is comprised of pre-transaction,
transaction, and post-transaction costs. Explanations of these three major cost categories follow:
• Pre-transaction costs: These costs are incurred prior to order and receipt of the purchased
goods. Examples are cost of certifying and training suppliers,
• Transaction costs: These costs include the cost of the goods/services and cost associated with
placing and receiving the order. Examples are purchase price, preparation of orders, and
delivery costs.
• Post-transaction costs: These costs are incurred after the goods are in the possession of the
company, agents, or customers. Examples are field failures, company’s goodwill/reputation,
maintenance costs, and warranty costs.
TCO provides a proactive approach for understanding costs and supplier performance leading to
reduced costs. However, the challenge is to effectively identify the key cost drivers needed to
determine the total cost of ownership. A recent exploratory study of total cost of ownership models
indicates that leading-edge companies actually use such models.

Table 3.1 Examples of Performance Metrics

(i) Cost / Price
• Competitive price
• Availability of cost breakdowns
• Productivity improvement/cost-reduction programs
• Willingness to negotiate price
• Inventory cost
• Information cost
• Transportation cost
• Actual cost compared to: historical (standard) cost, target cost, cost-reduction goal,
benchmark cost
• Extent of cooperation leading to improved cost

22
(ii) Quality
• Zero defects
• Statistical process controls
• Continuous process improvement
• Fit for use
• Corrective action program
• Documented quality program such as ISO 9000
• Warranty
• Actual quality compared to : historical quality, specification quality, target quality
• Quality improvement compared to : historical quality, quality-improvement goal
• Extent of cooperation leading to improved quality
(iii) Delivery
• Fast
• Reliable/on time
• Defect-free deliveries
• Actual delivery compared to : promised delivery, window (i.e., two days early to zero
days late)
• Extent of cooperation leading to improved delivery
(iv) Responsiveness and Flexibility
• Responsiveness to customers
• Accuracy of record keeping
• Ability to work effectively with teams
• Responsiveness to changing situations
• Participation/success of supplier certification program
• Short-cycle changes in demand/flexible capacity
• Changes in delivery schedules
• Participation in new product development
• Solving problems
• Willingness of supplier to seek inputs regarding product/service changes
• Advance notification given by supplier as a result of product/service changes
(v) Environment
• Environmentally responsible
• Environmental management system such as ISO 14000
• Extent of cooperation leading to improved environmental issues
(vi) Technology
• Proactive improvement using proven manufacturing/service technology
• Superior product/service design
• Extent of cooperation leading to improved technology
(vii) Business Metrics
• Reputation of supplier/leadership in the field
• Long-term relationship
• Quality of information sharing
• Financial strength such as Dun & Bradstreet’s credit rating
• Total Cash flow
• Rate of return on investment
• Extent of cooperation leading to improved business processes and performance
(viii) Total Cost of Ownership
• Purchased products shipped cost-effectively
• Cost of special handling
• Additional supplier costs as the result of the buyer’s scheduling and shipment needs
• Cost of defects, rework, and problem solving associated with purchases

23
3.8 Supplier Evaluation and Certification
Only the best suppliers are targeted as partners. Companies want to develop partnerships with the best
suppliers to leverage suppliers’ expertise and technologies to create a competitive advantage.
Learning more about how an organization’s key suppliers are performing can lead to greater visibility,
which can provide opportunities for further collaborative involvement in value-added activities. A
supplier evaluation and certification process must be in place so that organizations can identify their
best and most reliable suppliers. In addition, sourcing decisions are made based on facts and not
merely on perception of a supplier’s capabilities. Providing frequent feedback on supplier
performance can help organizations avoid major surprises and maintain good relationships. For
example, Honeywell has a Web-based monthly reporting system for evaluating supplier performance.
Suppliers can access their ratings on-line and see how they are performing with respect to the other
suppliers. While it is important to evaluate the suppliers, it is equally important that suppliers be
allowed to provide constructive feedback to the customer to enhance long-term partnerships.
One of the goals of evaluating suppliers is to determine if the supplier is performing according to the
buyer’s requirements. An extension of supplier evaluation is supplier certification, defined by the
Institute for Supply Management as “an organization’s process for evaluating the quality systems of
key supplies in an effort to eliminate incoming inspections. The certification process implies a
willingness on the part of customers and suppliers to share goals, commitments, and risks to improve
their relationship. A supplier certification program also indicates long-term mutual commitment. For
example, a certification program might provide incentives for suppliers to deliver parts directly to the
point of use in the buyer firm, thus reducing costs associated with incoming inspection and storage of
inventory.
Implementing an effective supplier certification is critical to reducing the supplier base, building long-
term relationships, reducing time spent on incoming inspections, improving delivery and
responsiveness, recognizing excellence, developing a commitment to continuous improvement, and
improving overall performance. Supplier certification allows organizations to identify the suppliers
who are most committed to creating and maintaining a partnership and who have the best capabilities.
Table 3.2 presents criteria generally found in many certification programs.

Table 3.2 Criteria Used in Certification Programs

• No incoming product lot rejections (e.g., less than 0..5 percent defective) for a specified
time period
• No incoming non-product rejections (e.g., late delivery) for a specified time period
• No significant supplier production-related negative incidents for a specified time period
• ISO 9000/Q9000 certified or successfully passing a recent, on-site quality system
evaluation
• Mutually agreed-upon set of clearly specified quality performance measures
• Fully documented process and quality system with cost controls and continuous
improvement capabilities
• Supplier’s processes stable and in control

3.9 The Weighted-Criteria Evaluation System for supplier

One approach of evaluating and certifying suppliers is to use the following weighted-criteria
evaluation system:
(i) Select the key dimensions of performance mutually acceptable to both customer and supplier.
(ii) Monitor and collect performance data.

24
 (iii) Assign weights to each of the dimensions of performance based on their relative importance
to the company’s objectives. The weights for all dimensions must sum to 1.
(iv) Evaluate each of the performance measures on a rating between zero (fails to meet any
intended purpose or performance) and 100 (exceptional in meeting intended purpose or
performance).
(v) Multiply the dimension rating by the importance weight and sum to get an overall score.
(vi) Classify vendors based on their overall score:
• Unacceptable (less than 50) : Supplier is dropped from further business.
• Conditional (between 50 and 70) : Supplier needs development work to improve
performance but may be dropped if performance continues to lag.
• Certified (between 70 and 90) : Supplier meets intended purpose or performance.
• Preferred (greater than 90): Supplier will be considered for involvement in new
product development and opportunities for more business.
(vii) Audit and perform ongoing certification review.
An example of the preceding evaluation and certification process is shown in Table 3.3.
Table 3.3
Supplier Scorecard Used for the XYZ Company
Performance Rating × Weight = Final Value
Measure
Technology 80 0.10 8.00
Quality 90 0.25 22.50
Responsiveness 95 0.15 14.25
Delivery 90 0.15 13.50
Cost 80 0.15 12.00
Environmental 90 0.05 4.50
Business 90 0.15 13.50
Total Score 1.00 88.25
Note: Based on the total score of 88.25, the XYZ company is considered a certified
supplier

3.10 Supplier Relationship Management Software (SRM)

Many organizations are investing in SRM software modules due to the wealth of information that can
be derived from these systems. SRM software can organize supplier information and provide answers
to questions such as:
• Who are our vendors? Are they the right set of suppliers?
• Who are our best suppliers and what are their competitive rankings?
• What is our suppliers’ performance with respect to on-time delivery, quality, and costs?
• Can we consolidate our buying to achieve greater scale economies?
• Do we have consistency in suppliers and performance across different locations and facilities?

25
 • What products/services do we purchase?
• What parts can be reused in new designs?
In general, SRM software varies by vendors in terms of capabilities offered. AMR Research has
identified five key tenets of an SRM system:
• Automation of transactional processes between an organization and its suppliers.
• Integration that provides a view of the supply chain that spans multiple departments,
processes, and software applications for internal users and external partners.
• Visibility of information and process flows in and between organizations. Views are
customized by role and aggregated via a single portal.
• Collaboration through information sharing and suppliers’ ability to input information directly
into an organization’s supply chain information system.
• Optimization of processes and decision making through enhanced analytical tools such as
data warehouse and Online Analytical Processing (OLAP) tools with the migration toward
more dynamic optimization tools in the future.
The key benefits of SRM include the following: (i) Better internal and external communications
providing visibility into various cost components; (ii) Automated creation, negotiation, execution and
compliance leading to more strategic, long-term relationships; (iii) Common and consistent
measurements that help focus resources, identify performance glitches, and develop strategies for
supply chain improvements; and (iv) The elimination of time-intensive, costly processes of
performing paper-based business transactions.

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

26
 CHAPTER IV
Bullwhip Effect and Supply Chain Management

4.1 Introduction
In recent years, many suppliers and retailers have observed that while customer demand for specific
products does not vary much, inventory and back-order levels fluctuate considerably across their
supply chain. For instance, in examining the demand for pampers disposable diapers, executives at
Procter & Gamble noticed an interesting phenomenon. An expected, retail sales of the product were
fairly uniform; there is no particular day or month in which the demand is significantly higher or
lower than any other. However, the executives noticed that distributors’ orders placed to the factory
fluctuated much more than retail sales. In addition, Procter & Gamble’s orders to its suppliers
fluctuated even more. This increase in variability as we travel up in the supply chain is referred to as
the bullwhip effect.
To understand the impact of the increase in variability on the supply chain, consider the second
stage in our example, the wholesaler. The wholesaler receives orders from the retailer and places
orders to its supplier, the distributor. To determine these order quantities, the wholesaler must
forecast the retailer’s demand. If the wholesaler does not have access to the customer’s demand data,
it must use orders placed by the retailer to perform the forecasting. Since variability in orders placed
by the retailer is significantly higher than variability in customer demand, the wholesaler is forced to
carry more safety stock than the retailer or else to maintain higher capacity than the retailer in order
to meet the same service level as the retailer.
This analysis can be carried over to the distributor as well as the factory, resulting in even higher
inventory levels and therefore higher costs at these facilities. Consider, for example, a simple widget
supply chain. A single factory, Widget Makers, Inc., supplies a single retailer, the Widget Store.
Average annual widget demand at the Widget Store is 5200 units, and shipments are made from
Widget Makers to the store each week.

External Demand

Retailer
Delivery lead-time

Order lead-time
Wholesaler
Delivery lead-time

Order lead-time Distributor

Delivery lead-time

Order lead-time
Factory
Production lead time
If the variability in orders placed by the Widget Store is low, such that the shipment every week is
about 100 units, Widget Makers’ production capacity and weekly shipping capacity need by only
about 100 units. If weekly variability is very high, such that during certain weeks Widget Makers
must make and ship 400 units and some weeks no units at all, it is easy to see that production and
shipping capacity must be much higher and that some weeks this capacity will be idle. Alternatively,
Widget Makers could build up inventory during weeks with low demand and supply these items
during weeks with high demand, thus increasing inventory-holding costs.

4.2 Causes of bullwhip effect

It is therefore, important to identify techniques and tools that will allow us to control the bullwhip
effect, i.e., to control the increase in variability in the supply chain. For this purpose, we need to
first understand the main factors contributing to the increase invariability in the supply chain.
(i) Demand forecasting. Traditional inventory management techniques practiced at each
level of the supply chain lead to the bullwhip effect. To explain the connection between
forecasting and the bullwhip effect, consider an approach that is used frequently to
manage inventory, the min-max inventory management policy. Here, whenever the
inventory at a facility is less than a given number, referred to as the reorder point, the
facility orders a quantity that will increase its inventory to a given target level. This
target level is set based on average demand and the variability of that demand.
Typically, managers use standard forecast smoothing techniques to estimate average
demand and demand variability. An important characteristic of all forecasting
techniques is that as more data are observed, the more we modify the estimates of the
average demand and demand variability. Since the order target level strongly depends
on these estimates, the user is forced to change order quantities, thus increasing
variability.
(ii) Lead-time. It is easy to see that the increase in variability is magnified with increasing
lead-time. Indeed, the reorder level consists of two quantities; the first is the average
demand during lead-time, and the second is the safety stock, which depends on lead-
time, demand variability, and service level. Thus, with longer lead times, a small change
in the estimate of demand variability implies a significant change in reorder level,
leading to a significant change in order quantities. This, of course, leads to increase
invariability.
(iii) Batch ordering. The impact of batch ordering is quite simple to understand. If the
retailer orders in batches, then the wholesaler will observe a large order, followed by
several periods of no orders, followed by another large order, and so on. Thus the
wholesaler sees a distorted and highly variable pattern of orders.
Recall that firms use batch ordering for a number of reasons. As pointed a firm that is
faced with fixed ordering costs needs to minimize these costs, which leads to batch
ordering. Second, as transportation costs become more significant, retailers may order
quantities that allow them to take advantage of transportation discounts (e.g., full-
truckload quantities). This may lead to some weeks with large orders and some with no
orders at all. Finally, the quarterly or yearly sales quotas or incentives observed in many
businesses also can result in unusually large orders observed on a periodic basis.
(iv) Price fluctuation. Price fluctuation also can lead to the bullwhip effect. If prices
fluctuate, retailers often attempt to stock up when prices are lower. This is accentuated
by the prevailing practice in many industries of offering promotions and discounts at
certain times or for certain quantities.
(v) Inflated orders. Inflated orders placed by retailers during shortage periods tend to
magnify the bullwhip effect. These are common when retailers and distributors suspect
that a product will be in short supply and therefore anticipate receiving supply

28
 proportional to the amount ordered. When the period of shortage is over, the retailer
goes back to its standard orders, leading to all kinds of distortions and variations in
demand estimates.
(vi) Lack of centralized information. One of the most frequent suggestions for reducing the
bullwhip effect is to centralize demand information within a supply chain, i.e., to
provide each sage of the supply chain with complete information on the actual customer
demand. To understand why centralized demand information can reduce the bullwhip
effect, note that if demand information is centralized, each stage of the supply chain can
use the actual customer demand data to create more accurate forecasts rather than
relying on the orders received from the previous stage, which can vary significantly
more than the actual customer demand.
Now consider two types of supply chains: one with centralized demand information and a second
with decentralized demand information. In the first type of supply chain, the centralized supply
chain, the retailer, or the first stage in the supply chain, observes customer demand, forecasts the
average demand, determines its target inventory level, and places an order to the wholesaler. The
wholesaler, or the second stage of the supply chain, receives the order along with the retailer’s
forecast average demand, uses this forecast to determine its target inventory level, and places an
order to the distributor.
Since the wholesaler has full information on the retailer inventory levels and customer demand, the
wholesaler can predict an incoming order from the retailer and hence be ready for this order, thus
reducing lead-time. This lead-time reduction leads to reduction in the increase in variability.
Similarly, the distributors, or the third sage of the supply chain, has information about the
wholesaler and the retailer inventory levels as well as customer demand and hence can significantly
reduce lead time and as a result reduce the bullwhip effect.
The second type of supply chain that we consider is the decentralized supply chain. In this case the
retailer does not make its forecast average demand available to the remainder of the supply chain.
Instead, each sage of the supply chain must estimate mean demand based on the orders received
from its customer, without knowledge of the retailer’s forecast.
What can we conclude about the bullwhip effect in these two types of supply chains? For either
type of supply chain, centralized or decentralized, the variability of the order quantities becomes
larger as we move up the supply chain so that the orders placed by the wholesaler are more variable
than the orders placed by the retailer, and so on. The difference in the two types of supply chains is
in terms of the ability to respond to orders from down stream facilities. Centralized information
allows to reduce lead-time and hence variability in the supply chain.
Indeed, the variability of orders increases dramatically more in the decentralized system. In other
words, a decentralized supply chain, in which only the retailer knows the customer demand, can lead
to significantly higher variability than a centralized supply chain, in which customer demand
information is available at each stage of the supply chain, particularly when lead times are large. We
therefore conclude that centralizing demand information can reduce the bullwhip effect significantly.

4.3 Methods for Coping with the Bullwhip Effect

Ability to identify the causes of the bullwhip effect leads to a number of suggestions for reducing
the bullwhip effect or for eliminating its impact. These include reducing uncertainty, reducing the
variability of the customer demand process, reducing lead times, and engaging in strategic
partnerships.
(i) Reducing uncertainty. One of the most frequent suggestions for decreasing or eliminating
the bullwhip effect is to reduce uncertainty throughout the supply chain by centralizing
demand information, i.e., by providing each stage of the supply chain with complete
information on actual customer demand.

29
 Note, however, that even if each stage uses the same demand data, each may still employ
different forecasting methods and different buying practices, both of which may contribute
to the bullwhip effect. In addition, even when each stage uses the same demand data, the
same forecasting method, and the same ordering policy, the bullwhip effect will continue to
exist, albeit at a significantly reduced level. Thus centralized demand information reduces
the bullwhip effect but does not eliminate it.
(ii) Reducing variability. The bullwhip effect can be diminished by reducing the variability
inherent in the customer demand process. For example, if we can reduce the variability of
customer demand seen by the retailer, then even if the bullwhip effect occurs, the variability
of demand seen by the wholesaler also will be reduced.
We can reduce the variability of customer demand through, for example, the use of an
everyday low pricing (EDLP) strategy. When a retailer used EDLP, it offers a product at a
single consistent price rather than offering a regular price with periodic price promotions.
By eliminating price promotions, a retailer can eliminate many of the dramatic shifts in
demand that occur along with these promotions. Therefore, everyday low pricing strategies
can lead to much more stable – i.e., less variable – customer demand patterns.
Of course, variability of customer demand depends not only on the retailer pricing strategy
but also on its competitors’ strategies. Thus, while EDLP is an important tool used to
reduce demand variability, its impact can be limited.
(iii) Lead-time reduction: As we observed earlier, the longer the lead-time, the larger is the
increase in variability. Therefore, lead-time reduction can reduce the bullwhip effect
significantly throughout a supply chain.
Observe that lead times typically include two components: order lead times (i.e., the time it
takes to produce and ship the item) and information lead times (i.e., the time it takes to
process an order). This distinction is important because order lead times can be reduced
through the use of cross-docking, whereas information lead times can be reduced through
the use of electronic data interchange (EDI).
(iv) Strategic partnership. The bullwhip effect can be eliminated by engaging in any of a
number of strategic partnerships. These strategic partnerships change the way information is
shared and inventory is managed within a supply chain, possibly eliminating the impact of
the bullwhip effect. For example, in vendor-managed inventory (VMI;), the manufacturer
manages the inventory of its product at the retailer outlet and therefore determines for itself
how much inventory to keep on hand and how much to ship to the retailer in every period.
Therefore, in VMI, the manufacturer does not rely on the orders placed by a retailer, thus
avoiding the bullwhip effect entirely.
Other types of partnerships are also applied to reduce the bullwhip effect. As we discussed earlier,
for example, centralizing demand information can dramatically reduce the variability seen by the
upstream stages in a supply chain. Therefore, it is clear that these upstream stages would benefit
from a strategic partnership that provides an incentive for the retailer to make customer demand data
available to the rest of the supply chain.

4.4 Matching supply and demand

The business environment today is such that we have consumers who are more demanding and
discriminating. The market has evolved into a “pull” environment with customers dictating to the
supplier what products they desire and when they need them delivered. If a retailer cannot get the
product it wants at the right quantity, price, and time from one supplier, the retailer will look for
another company that can meet its demands. Any temporary stock-out has a tremendous downside
on sales, profitability, and customer relationships. Managing demand is challenging because of the
difficulty in always forecasting future consumer requirements accurately.

30
In order for supply chain integration to be successful, suppliers must be able to accurately forecast
demand so that they can produce and deliver the right quantities demanded by their customers in a
timely and cost-effective fashion. There are several ways to closely match supply and demand. One
way is for a supplier to hold plenty of stock available for delivery at any time. While this approach
maximizes sales revenues, it is also expensive because of the cost of carrying inventory and the
possibility of write-downs at the end of the selling season. Use of flexible pricing is another
approach. During heavy demand periods, prices can be raised to reduce peak demand. Price
discounts can then be used to increase sales during periods with excess inventory or slow demand.
This strategy can still result in lost sales, though, as well as stock-outs and thus cannot be considered
an ideal or partnership-friendly approach to satisfying demand. In the short term, companies can also
use overtime, subcontracting, or temporary workers to increase capacity to meet demand for their
products and services. In the interim, firms will lose sales as they train workers and quality may also
tend to suffer.
Thus, it is imperative that suppliers along the supply chain find ways to better match supply and
demand to achieve optimal levels of cost, quality, and customer service to enable them to compete
with other supply chains. Any problems that adversely affect the timely delivery of products
demanded by consumers will have ramifications throughout the entire chain.

4.5 Collaborative Planning, Forecasting, and Replenishment

The American Production and Inventory Control Society (APICS) defines collaborative planning,
forecasting, and replenishment (CPFR) as “a collaboration process whereby supply chain trading
partners can jointly plan key supply chain activities from production and delivery of raw materials to
production and delivery of final products to end customers. Collaboration encompasses business
planning, sales forecasting, and all operations required to replenish raw materials and finished
goods”. The objective of CPFR is to optimize the supply chain by improving demand forecast
accuracy, delivering the right product at the right time to the right location, reducing inventories
across the supply chain, avoiding stock-outs, and improving customer service. Basically, this can be
achieved only if the trading partners are working closely together and willing to share information
and risk through a common set of processes.
The real value of CPFR comes from an exchange of forecasting information rather than from more
sophisticated forecasting algorithms to improve forecasting accuracy. The fact is that forecasts
developed solely by the firm tend to be inaccurate. When both the buyer and seller collaborate to
develop a single forecast, incorporating knowledge of base sales, promotions, store openings or
closings, and new product introductions, it is possible to synchronize buyer needs with supplier
production plans, thus ensuring efficient replenishment. The jointly managed forecasts can be
adjusted in the event that demand or promotions have changed, thus avoiding costly corrections after
the fact.
On the surface, when decisions are made with incomplete information, it may appear that companies
have “optimized” their internal processes when, in reality, inventory has merely shifted along the
supply chain. Without trading partners in the supply chain collaborating and exchanging
information, the supply chain will always be suboptimal, resulting in less-than-maximum supply
chain profits.
CPFR is an approach that addresses the requirements for good demand management. The benefits of
CPFR include the following:
• Strengthens partner relationships
• Provides analysis of sales and order forecasts upstream and downstream
• Uses point-of-sale data, seasonal activity, promotions, new product introductions, and store
openings or closings to improve forecast accuracy

31
 • Manages the demand chain by exception and proactively eliminates problems before they
appear
• Allows collaboration on future requirements and plans
• Uses joint planning and management of promotions
• Integrates planning, forecasting, and logistics activities
• Provides efficient category management and understanding of consumer purchasing habits
• Provides analysis of key performance metrics (e.g., forecast accuracy, forecast exceptions,
product lead times, inventory turnover, percentage stock-outs) to reduce supply chain
inefficiencies, improve customer service, and increase sales and profitability
The Global Commerce Initiative (GCI) created the GCI Recommended Standard for Globalizing
CPFR. GIC is a voluntary body created in USA in 1999 to “improve the performance of the
international supply chain for consumer goods through the collaborative development and
endorsement of recommended standards and key business processes.” A description of the CPFR
process model used by GCI follows:
• Step 1: Develop Collaboration Arrangement
The buyer and seller must agree on the objective of the collaboration, ground rules for
resolving disagreements, confidentiality of information to be shared, sales forecast
exception criteria, review cycle, time frame, frozen time period with acceptable
tolerances, resource commitments, financial incentives, and success metrics. Some
examples of objectives are to improve customer service levels, reduce stock-outs,
reduce inventories, increase sales, reduce costs, improve forecast service levels, reduce
stock-outs, reduce inventories, increase sales, reduce costs, improve forecast accuracy,
and synchronize production with the forecast.
• Step 2: Create Joint Business Plan
A joint business plan is developed by sharing the companies’ business strategies and
plans. The plan typically involves developing a joint product category and promotional
plan in which the appropriate category strategies, inventory policies, promotional
activities, and pricing policies are specified. A product category is a manageable group
of products perceived by consumers to be similar that can be substituted in meeting their
needs. For each item in the product category, an item management profile is developed
that includes the minimum order quantity, lead-time, and time between orders, frozen
time period, and safety stock guidelines. The trading partner should be informed of
changes such as store openings or closings or changes of items in each product category.
It is important that trading partners be able to understand the impact of new product
introductions, promotions, and marketing campaigns have on demand and ultimately on
the effective management of the supply chain.
• Step 3: Create Sales Forecast
The trading partners use Web-based technologies to share data such as retailer point-of-
sale information, distribution center withdrawals, manufacturing consumption, planned
events including store openings or closings, and new product introductions. Using
multiple inputs into the forecasting process including information about the future, as
well as the past, results in the creation of a shared forecast that reflects the most accurate
and real-time information available. Either partner or both partners may generate the
sales forecast. The forecasting techniques used can be qualitative or quantitative. When
both partners each generate a forecast, middleware is used to highlight the differences,
based on predetermined exception criteria previously agreed upon the partners.

32
 • Step 4: Identity Exceptions for Sales Forecast
Irrespective of how the initial forecast is generated, all exceptions are identified in Step
4. Examples of sales forecast exception criteria are: retail in-stock is less than 95 per
cent, sales forecast error is greater than 20 percent, the difference in sales forecast from
the same period of the previous year is greater than 10 percent, or any changes that have
occurred in timing of promotional activities or number of active stores. The real-time
joint decision-making reduces the risk and increases the confidence in the single
forecast.
• Step 5: Resolve/collaborate on Exception Items
In Step 5, all exceptions are resolved through a collaborative process to create a single
consensus forecast.
• Step 6: create Order forecast
Data are analyzed – such as point-of-sale (POS) data; historical demand; shipment data;
current capacity limits; minimum order quantities; lead times; time between orders;
frozen time periods; safety stock rules, impact events such as new product introductions,
store openings, and store closings; and current inventory positions (on hand, on order, in
transit) – to generate the order forecast consistent with the sales forecast and joint
business plan developed earlier. The order forecast represents detailed time-phased
ordering needs with inventory objectives by product and receiving location. The order
forecast enables the manufacturer to effectively schedule production capacity based on
the demand and to minimize safety stock. For the retailer, there is greater confidence
that orders will be met. In effect, the real-time collaborative effort minimizes the
uncertainty between trading partners, leading to reduced supply chain inventories, along
with improved customer service levels.
• Step 7: Identity Exceptions for Order Forecast
The items that fall outside the order forecast exception criteria such as customer service
measures, order fill rate, or forecast error measures, established jointly by the buyer and
seller in the collaboration agreement, are identified as exception items. Examples of
order forecast exception criteria are retail in-stock less than 95 percent, order forecast
errors greater than 20 percent, annual inventory turnover less than agreed-upon goal,
addition of new event than affects inventory/orders, or requested emergency orders
greater than 5 percent of weekly forecast.
• Step 8: Resolve/Collaborate on Exception Items
Any order forecast exceptions are investigated by examining the shared data, e-mail,
telephone conversations, meetings, and other supporting information. If the analysis
justifies a change in the forecast, a revised forecast is submitted.
• Step 9: Order Generation
This last step involves converting the order forecast into a committed order. The actual
order is expected to consume the forecast. The committed order is generated based on
the product demand in the frozen time period of the order forecast.
Common performance metrics, such as gross margin percent, return on investment, and sales
growth, are developed to measure the success of the relationship. Other metrics include in-stock
percent a point of sale, inventory turnover, inventory level, sales forecast accuracy, potential sales
lost due to stock-out, manufacturing cycle time, order cycle time, shipping cycle time, problem
resolution time, rate of emergency or cancelled orders, and percent shipped or delivered on time.

33
4.6 Forecasting Techniques
Considering the importance of forecast, in this section, a brief discussion about various techniques
of forecasting is presented. Both quantitative and qualitative forecasts can be improved by seeking
inputs from trading partners. Qualitative forecasting methods are based on opinions and intuition,
whereas quantitative forecasting methods use mathematical models and relevant historical data to
generate forecasts. The quantitative methods can be divided into two groups: time series and
associative models.

4.6.1 Qualitative Methods

Qualitative forecasting methods are approaches to forecasting based on intuition or judgmental
evaluation and are generally used when data are limited, unavailable, or not currently relevant.
While this approach can be very low cost, the effectiveness depends to a large extent on the skill and
experience of the forecasters(s) and the amount of relevant information available. The qualitative
techniques are often used to develop long-range projections when current data is no longer very
useful, and for new product introductions when current data does not exist. Discussions of four
common qualitative forecasting models follow:
• Jury of executive opinion: A group of senior management executives who are
knowledgeable about the market, competitors, and the business environment collectively
develop the forecast. This technique has the advantage of several individuals with
considerable experience working together, but if one member’s views dominate the
discussions, then the value and reliability of the outcome can be diminished. This technique
is applicable for long-range planning and new product introductions.
• Delphi method: A group of internal and external experts are surveyed during several
rounds in terms of future events and long-term forecasts of demand. Group members do not
physically meet and thus avoid the scenario where one or a few experts could dominate a
discussion. The answers from the experts are accumulated after each round of the survey
and summarized. The summary of responses is then sent out to all the experts in the next
round, wherein individual experts can modify their responses based on the group’s response
summary. The iterative process goes on until a consensus is reached. The process can be
both time-consuming and very expensive. This approach is applicable for high-risk
technology forecasting; large, expensive project; or major, new product introductions. The
quality of the forecast depends largely on the knowledge of the experts.
• Sales force composite: The sales force represents a good source of market information.
This type of forecast is generated based on the sales force’s knowledge of the market and
estimates of customer needs. Due to the proximity of the sales personnel to the consumers,
the forecast tends to be reliable but individual biases could negatively impact the
effectiveness of this forecast there is a tendency for the sales force to under forecast.
• Consumer survey: A questionnaire is developed that seeks input from customers on
important issues such as future buying habits, new product ideas, and opinions about
existing products. The survey is administered through telephone, mail, Internet, or personal
interviews. The challenge is to identify a sample of respondents who are representative of
the larger population and to get an acceptable response rate.

4.6.2 Quantitative Methods

Quantitative forecasting models use mathematical techniques that are based on historical data and
can include causal variables to forecast demand. Time series forecasting is based on the
assumption that the future is an extension of the past, thus, historical data can be used to predict
future demand. Associative forecasting assumes that one or more factors (independent variables)
are related to demand and, therefore, can be used to predict future demand. Since these forecasts rely

34
solely on past demand data, all quantitative methods become less accurate as the forecast’s time
horizon increases. Thus, for long time horizon forecasts, it is generally recommended to utilize a
combination of both quantitative and qualitative techniques.
Components of Time Series Data
Time series data typically have four components: trend, cyclical, seasonal, and random variations:
• Trend variations: Trends represent either increasing or decreasing movements over many
years and are due to factors such as population growth, population shifts, cultural changes,
and income shifts. Common trend lines are linear, S-curve, exponential, or asymptotic.
• Cyclical variations: Cyclical variations are wavelike movements that are longer than a year
and influenced by macroeconomic and political factors.
• Seasonal variations: Seasonal variations show peaks and valleys that repeat over a
consistent interval such as hours, days, weeks, months, years, or seasons. Due to seasonality,
many companies do well in certain months and not so well in other months.
• Random variations: Random variations are due to unexpected or unpredictable events such
as natural disasters (hurricanes, tornadoes, fire), strikes, and wars.

Time Series Forecasting Models

As discussed earlier, time series forecasts are dependent on the availability of historical data.
Forecasts are estimated by extrapolating the past data into the future. Time series forecasting is one
of the most widely used techniques. A survey of purchasing professionals indicates that the top three
quantitative forecasting techniques used are simple moving average, weighted moving average, and
exponential smoothing. Some of the more common time series approaches such as simple moving
average, weighted moving average, exponential smoothing, and trend-adjusted exponential
smoothing are discussed next.
(i) Simple Moving Average Forecasting Model: The simple moving average forecasting method
uses historical data to generate a forecast and works well when the demand is fairly stable over time.
The n-period moving average forecast is
tt

∑A
i = t − n +1
i
Ft +1 = where
n
Ft+1 = forecast for Period t + 1,
n = number of periods used to calculate moving average, and
Ai = actual demand in Period i.

The average tends to be more responsive if fewer data points are used to compute the average.
However, random events can also impact the average adversely. Thus the decision maker must
balance the cost of responding slowly to changes versus the cost of responding to random variations.
The advantage of this technique is that it is simple to use and easy to understand.
(ii) Weighted Moving Average Forecasting Model: Weighted moving average forecasting, which
is based on an n-period weighted moving average, follows:
tt
Ft +1 = ∑w A
i = t − n +1
i i

35
where
Ft+1 = forecast for Period t + 1,
n = number of periods used to determining the moving average,
Ai = actual demand in Period i, and
Wi = weight assigned to Period i (with ∑W i = 1)
The weighted moving average allows greater emphasis to be placed on more recent data to reflect
changes in demand patterns. Weights used also tend to be based on experience of the forecaster.
Although the forecast is more responsive to underlying changes in demand, the forecast still lags
demand because of the averaging effect. As such, the weighted moving average method does not do
a good job of tracking trend changes in the data.
(iii) Exponential Smoothing Forecasting Model: Exponential smoothing forecasting is a
sophisticated weighted moving average forecasting in which the forecast for the next period’s
demand is the current period’s forecast adjusted by a fraction of the difference between the current
period’s actual demand and its forecast. This approach requires less data to be kept than the
weighted moving average method because only two data points are needed. Due to its simplicity and
minimal data requirement, exponential smoothing forecasting is one of the more popular techniques.
This model, like the other time series models, is suitable for data that show little trend or seasonal
patterns. The exponential smoothing formula is
Ft +1 = Ft + α ( At − Ft ) or Ft +1 = α At + (1 − α ) Ft
where
Ft+1 = forecast for Period t + 1,
Ft = forecast for Period t,
At = actual demand in Period t, and
α = a smoothing constant (0 ≤ α ≤ 1).
With an α value closer to 1, there is a greater emphasis on recent data, making the model more
responsive to changes in the recent demand. When α has a low value, more weight is placed on past
demand (which is contained in the previous period’s forecast value) and the model responds more
slowly changes in demand. The impact of using a small or large value of α is similar to the effect of
using a large or small number of observations in calculating the moving average. In general, the
forecast will lag any trend in the actual data because only partial adjustment to the most recent
forecast error can be made. The initial forecast value could be estimated using one of the qualitative
methods, such as the Delphi forecast, or by simply setting the initial forecast equal to the demand for
that period.
(iv) Trend-Adjusted Exponential Smoothing Forecasting Model. The exponential smoothing
method can be modified to include a trend component when the time series show a systematic
upward or downward trend in the data over time. This method requires two smoothing constants,
one for the smoothed forecast (α ) and the other for the trend (β ). The equations for this model are
Ft = α At + (1 − α ) ( Ft −1 + Tt −1 ),
Tt = β ( Ft − Ft −1 ) + (1 − β ) Tt −1 ,
and the trend-adjusted forecast,
TAFt +m = Ft + mTt
where

36
 Ft = exponentially smoothed average in Period t,
At = actual demand in Period t,
Tt = exponentially smoothed trend in Period t,
α = smoothing constant ( 0 ≤ α ≤1), and
β = smoothing constant for trend ( 0 ≤ β ≤ 1).
A higher value of β indicates greater emphasis on recent trend changes, while a small β places less
weight on recent changes and has the effect of smoothing out the current trend. The smoothing
constants, α and β, are estimated using a trial and error approach, matching actual historical demand
data to the forecasted demand in search of the smoothing constants that minimize the forecast errors.
Linear Trend Forecasting Model: The trend can be estimated using simple linear regression to fit
a line to time series of historical data. The linear trend method minimizes the sum of squared
deviations to determine the characteristics of the linear equation:
Here,
∧
Y = b0 + b1 X
where
∧
Y = forecast or dependent variable
X = time variable,
b0 = intercept of the line, and
b1 = slope of the line.
The coefficients b0 and b1 are calculated as follows:
n∑ ( xy) − ∑ x∑ y
b1 = and
n∑ x 2 − (∑ x )
2

b0 =
∑ y −b ∑ x1

n
where b1 = slope of the line,
x = independent variable values,
y = dependent variable values,

x = average of the x values

y = average of the y values, and
n = number of observations.
Associative Forecasting Models: Associative forecasting generally uses regression analysis to
estimate future demand. One or several external variables are identified that are related to demand,
which are hopefully easier to determine than demand. Once the relationship between the external
variable and demand is determined, it can be used as a forecasting tool. The detail discussion of this
type of model is avoided here.
*****************

37
 CHAPTER V
Distribution Management Overview

5.1 Introduction
Transportation happens to be the most fundamental part of strategic logistic management and
transport costs include all costs associated with movement of products from one location to another.
The average transport costs ranges from 5 to 6% of the recommended retail price of the product.
Transportation is the movement of products, materials and services from one area to another, both
inbound and outbound. It can also be said as movement from one node of the supply chain to the
other.
Indian Army is a typical example of ideal transportation mixes in our country. It uses the aerial, land,
sea and rail routes to maintain its forces strewn all over the country and aboard. The logistics is
enormous and the various modes of transport are aircraft, train, trucks, animals, and human beings. It
transports supplies, ration, fuel, oil lubricants, arms, ammunition, clothing and personal loads over
vast distances and over varied terrain and climatic conditions.

5.2 Transportation System

Transportation system has a strategic bearing on a company’s operation efficiency. Therefore, failure
to identify the best transportation mode can directly affect the growth of a company. Since, higher
transport costs will raise prices; it will directly affect the customer satisfaction in a negative way. The
three factors as mentioned by Gattoma & Walters required to consider are
• Customer
• Environment
• Product & company
Organization, which involves physical movement of goods require transport services that varies from
mode to mode. The best suitable mode is required to be identified depending upon the nature of
product that has to be moved. Like if coal or carbon has to be moved use the railways from the source
to the production unit directly, so as to minimize losses, time & cost factor. Therefore, in order to
identify the right transport system the following points to be considered
• Impact of the transport system on the supply chain
• Factors that determine the choice of transport mode
• Feedback and reporting both from within and the environment on the choice of transport, and
rectify in case you went wrong the first time.
• Your foresight, flexibility & integration of available resources in planning stage will be one of
the crucial factors that will dictate the choice of transport.
• Customer communications: In order to obviate delays in transportation and handling of
logistics both the supplies and distributors are relying more and more on electronic transfer
systems, IT & the Internet. This will help in considerable reduction in time delays and ensure
better cooperation between the chains.
• Market coverage: Transportation costs influence the size of markets covered in a big way.
The characteristics are: costs, flexibility, reliability and availability. The product per se will
influence the economics of the decision. A low volume and high value product will be above
to support higher costs, which means extended delivery distances and increase in delivery
frequency.
 • Sourcing decisions: The geographical dimension of the source markets can be influenced by
low cost transportation system. Companies therefore have to consider a trade off between
price and quality and the costs involved in delivering to the processing point, i.e. volume and
cost of transportation.
• Manufacturing operations: Cost of transporting has a direct bearing on the location of the
manufacturing market center. That is hwy, extraction based units are close to the source of
raw materials and the products related to customer satisfaction are closer home, i.e. near to the
customer hub center.
• Pricing decision: Transportation happens to be the important component of product costs.
Therefore, selection of the appropriate transportation mode will have a direct bearing on the
product costs per se, with more relevance to exports. Increase in transportation costs increases
the product pricing.
• Customer service decisions: Both customer service policy and transportation decisions go
hand in hand and hence one cannot be considered in isolation of the considerably. Therefore,
it is pertinent to overrule the cost factor while servicing the medical customers, since speed is
more important than cost in selecting the transport mode.

5.3 Transport in supply chain

Transport is vital to the overall gambit of SCM operation and therefore cannot be considered in
isolation. The entire transportation process is to be monitored, in order to gauge the exact location and
stage of the materials being transported. Transport, is the process, which transports materials between
two or more stations, and therefore, the form of transport to be used should not only be responsive and
compatible to the terminal stations, but also with the operating environment through which the
product moves. In order to achieve the best, it is therefore mandatory that sufficient information be
made available to enable the movement to be monitored by the producer, consumer, agencies,
financial institutions and relevant groups.
Transport profile
Operating characteristics dictate the transport requirements of an organization. The transport
requirement depends on the different and versatile nature of tasks that are to be performed. Therefore,
to generalize, an organization, which doesn’t have versatility and varieties in operating its transport
for varied tasks, will operate much below the optimum level of efficiency.
Operational factors
Operational factors that determine the transport mode are:
• Environmental factors
• Characteristics of alternate transport modes
• Combination approach
The various characteristics of alternative transportation mode are:
• Useful load: physical capability and maximum load as a percentage of gross weight.
• Density: cargo density, i.e., weight per cubic unit.
• Overheads: fixed costs as a percentage of total cost.
• Productivity: calculated in tonne-miles per direct man-hour.
It is very important to establish and determine the accurate operating characteristics of each available
transport mode, so that suitability of matching these to the operating factors can be established. Each
type of transport offers different characteristics and as a supply chain manager you have to understand
the efficacy of these aspects:

40
 OPERATIONAL FACTORS

CUSTOMER ENVIRONMENTAL PRODUCT COMPANY

CHARACTERISTICS

CHARACTERISTICS OF TRANSPORT ™ ROAD

MODES ™ RAIL
™ AIR
™ WATER

CHOICE OF TRANSPORT MODE ™ CUSTOMER SERVICE

LEVEL
™ COSTS

Figure 5.1: Operational Factors for Deciding the Transport Mode (Deshmukh and
Mohanty), 2004
Transportation costs
Transport cost vary from less than 1 per cent (for machinery) to over 30 per cent (for food) of the
recommended selling price of products, depending upon the nature of the product range and its
market. However, the average transport costs is between 5 to 6 per cent of the recommended retail
price of a product. With inflation, transport costs also rise because the major components are the
workforce (labor), fuel & maintenance, spares, driver’s cost. Similarly, transport represents a direct
cost added to the price of the product and any reduction in transport costs would lead to an increase in
profit, with price remaining constant. Transportation rates are almost linear with distances and not
with volume, be it road, rail, water or air. Transportation costs for company owned fleet is simple and
is evolved by annual cots per truck, annual mileage, amount delivered and trucks effective capacity.
All this information could be effectively utilized to calculate cost per mile per SKU.

5.4 Method of transport selection

The selection procedure for the transport mode could vary from the simple decision either to identify
one feasible method of distribution or to follow the competitor’s procedures, to the complex decision
that calculates the cost incurred and produces an optimum solution. The three potential methods are
• Judgment: Identification of the important factors affecting the transport problem, and the
transport mode from a list of alternatives available, so that the important features of the
transport requirements are met. The shortcomings are tremendous in such a process, since
transport is considered as a service rather than a distribution system.
• Cost trade-off: It is where the impact of transport is calculated in relation to immediate
terminal objectives and activities, and the total cost of distribution system is optimized. This
particular approach acknowledges the existence of trade-off within the numerous alternative
approaches in an attempt to assess the situation to minimize total costs.

41
 • Distribution models: This identifies and explains the interrelationships between the
components of the distribution system at various levels of daily, weekly or monthly demands.
These models could be built to examine the impact of alternative transport modes and
methods, as either the demand changes or the components in the system change.
Therefore, in order to carry out a systematic selection of the transporter a framework consisting of the
following stages is recommended

SYSTEMATIC SELECTION

STAGE 1 STAGE 2 STAGE 3 STAGE 4 STAGE 5

IDENTIFICATION CATEGORIZING DETERMINING MEASURE & MATRIX
OF FACTORS & DISTRIBUTION MONITOR COSTS ANALYSIS TO
IDENTIFICATION NETWORKS SELECTION
OF POTENTIAL

Figure 5.2: The Selection System

• Stage 1: Identifying the factors affecting the choice of transport selection

• Stage 2: Categorizing the significant factors and identifying the potential risks
• Stage 3: Determining the distribution network in terms of number and size of depots
• Stage 4: Applying the matrix analysis to select the most appropriate transport method
• Stage 5: Measuring and monitoring cost factors.

5.4.1 A transportation decision

Determining an organization’s transport requirement will be based on the following underlying
considerations, (Figure 5.2)
• The available depots, their sizes including movement requirements of raw materials to
manufacturing units and finished products to the warehouses and on to the consumers.
• The best choice of mode available depending on the distance involved.
• Product characteristics that will further dictate the type of transport for each requirement.
• The financial option that could be employed in terms of individual type of equipment.
• The operation needs in terms of usage of the equipment for maximum utilization and
minimum operational costs.
We have to understand that transportation operations cannot be seen in isolation, and hence
warehousing and depot locations are equally important to understand the choice of transport selection
process. Warehouse location decision is discussed in next section.
The total number and sizes of depots and warehouses will also have a direct bearing on transport
operations of all companies across the board. Let us se these in more details.

42
 DECISION FRAMEWORK

NUMBE & SIZE CHOICE OF THE CHOICE OF CHOICE OF CHOICE OF

OF DEPOTS TO TRANSPORT EQUIPMENT FINANCIAL OPERATIONAL
INCLUDE MODE SPECIFICATION OPTIONS NEEDS AS
MOVEMEN OF COINCIDENTAL S ON TYPE OF AVAILABLE REGARDS
RAW TO ECH TRANSPORT EQUIPMENTS TO
MATERIALS/FINI POTENTIAL DEFINED BY MAX
SHED PRODUCTS MOVEMENT IN PRODUCT UTILISATION &
DISTANCE CHARACTERIS MINIMISE COSTS
TRAVELED TICS

Figure 5.3: Decision Framework

Decision matrix for appropriate transport option

A decision matrix approach helps in identifying the most appropriate transport option from the
substantial range available. This approach uses the following steps:
• Selection of initial decisions required based upon known alternatives; like, choice of transport
mode, choice of equipment specification, choice of financial options and operational needs.
• To select two options (factors) so that a matrix can be formulated suing one in vertical axis
and the other on horizontal.
• Selection of basic alternatives, which adequately cover the conditions, imposed by the vertical
and horizontal axes.
• Determination of organization needs by analysis of the important factors generated to produce
the matrix and use of the matrix to select the options required.
• Selection of the resources required by considering the results of the matrix analysis plus other
factors of importance.
• The combination of the matrix solutions to provide an effective and efficient profile, which
identifies the transport tasks and appropriate resources for the tasks.
This approach will require imagination to develop the selection of the initial decisions, to determine
the important factors to use for the vertical and horizontal axes on the matrix, and to construct the
matrix. Yet, the majority of the question could be answered by a combination of brainstorming,
analysis and categorization of important factors, which affect the choice of transport selection.

5.5 Fleet sizing and configuration

Fleet sizing objective is to employ through ownership, hire, lease and or rental the fewest possible
trucks to manage the company’s load profile/shipping requirements. This decision is akin to the
decision of how much inventory is to be made available to the consumers / customers. In fleet sizing,
increased availability yields fewer lost sales, shorter customer cycle times, improved customer
services but higher fleet costs. Fleet sizing projections should be developed a few times during the
year and at any time when a major shift in demand pattern occurs. In certain cases, the cost of vehicle

43
shortages can be estimated and a cost of shortages versus cost of ownership analysis can be made to
determine the optimal fleet size. Fleet size can be regulated and minimized by:
• Utilizing standard size pallets and transport containers.
• Vigorously monitoring fleet utilization levels annually.
• Maintaining total fleet visibility, including loading times, unloading, transit times and
maintenance times.
• Choosing low-use periods to conduct routine maintenance.
• Monitoring and charging for demiurges for fleet detention by suppliers, customers and
carriers.
• Utilizing alternative coverage means during super peak periods to avoid carrying the burden
of an oversized fleet.
Therefore, it can be seen that whatever be the fleet size, the company has to use it judiciously and
constantly monitor its progress for optimum utilization of the available resources and at the same time
cut down cost of maintenance e and tile lag.
Fleet maintenance is one means of reducing the ownership cost of the fleet by delaying potential
replacements and improving customer service through improved reliability.

5.6 Routing and scheduling

India is one of the best examples of routing and re-scheduling, wherein such activities are optioned in
the shortest possible time. Delay in delivery due to routing problems increase costs of goods manifold.
Therefore, to tide over this the company has to plan these activities well in advance with detailed
coordination and judicious and realistic planning. India is a versatile country with equally versatile
terrain and climate condition. Companies have to gear itself to such changing scenarios and terrain
since the very inception. Efficient versus inefficient routing can save tremendous amount of money in
fuel, labor, and capital expenditures and significantly enhance customer satisfaction. The objective
should be to minimize:
• Total route costs
• Number of routes
• Distance traveled
• Route time
The constraints are:
• Customer requirements and time available
• Balancing of the route for the driver, to minimize overtaxing
• Maximum route time
• Vehicle capacity
• Start & stop points en-route
• Infrastructure constraints.
Routing problems are some of the most difficult criticalities encountered, and cannot actually be
solved optimally.

44
5.7 Future directions in transportation
One salient aspect that we all have to understand that with e-services, our lead time to delivery has
reduced considerably, but somehow the movement of the product and raw materials performance
cannot move through e-services and have to restrict movement to roads, rail, air and waterways. Yes,
the order can be placed through e-services in a faster mode and so can payment be but the products
cannot be physically moved through the net. A truck, rail wagon or a ship or the cargo aircraft has to
move it from the place of origin to the consumer’s destination.
Transportation too, has improved considerably with the advent of technology and mechanical
developments within a short span. Certain programs and organizations help in coordinating
transportation in a better way and as time passes they are bound to improve transportation in a big
way. They can be clubbed under as follows
• Carrier relationship management: Through carrier relationship management programs we
can bind transporters and those working with it under one roof/ enterprise. These programs
are designed to formalize communication, partnering, negotiating, and performance
monitoring aspects of carrier management. At the heart of most carrier relationship
management programs is a set of guidelines for selecting core carriers, the minority of carriers
who carry a majority of the enterprise’s weight, cube and shipments.
• Corporate traffic councils: These help in bringing together all personnel working in the area
of transportation within an enterprise. The traffic council sets corporate transportation policy
and explores opportunities for leveraging transportation spending across the corporation.
• Training and certification: Corporations should aim at making and maintain transportation as
a value added activity. For this everyone should be in one plane and therefore, such training
activities are carried out to get all under one platform.
• Driver quality: Improvements in drivers with better working environment and better wages
will help in a big way to improve the driver’s capability and capacity in the long run.
• Joint Procurement: Significant cost reduction can be carried out if the purchase and
negotiation of transportation services is consolidated across both inbound/outbound
transportation activities within a business unit, across units and even with no-competitors.
• Logistics compliance & security officer: Forming the chief logistics security officer will
enable a company to cope with global logistics law and to anticipate security lapses within the
logistics network.

5.8 Facility location

Facilities and their locations are major issues in an organization’s logistics system efficiency and its
ability to successfully implement its competitive advantage. Facility location decisions are of major
importance to a company’s ability to compete in the market. Determining appropriate locations for
facilities such as plants, warehouses, retail stores, hospitals etc. represents an important strategic
decision. Location decisions come under the category of long-term decisions. They involve long-term
commitments. A plant location decision cannot be reviewed until after quite sometime as they involve
huge investments. Location decisions also have effect on the operating costs/revenues. For example, a
bad location decision may call for excessive transportation costs, shortage of skilled workforce, loss
of competitive advantage, inadequate supply of raw materials etc. Organizations are involved in
location decisions for a variety of reasons such as the followings
• Expansion of existing facilities
• Addition of new facilities
• Closing down the plant at one location and moving to another

45
5.8.1 Plant location
Choice of location for a plant is one of the earliest problems facing management. But location,
perhaps, is one of the most neglected aspects of business, although the manufacturing and distribution
costs may vary by over 10 percent simply by virtue of choice of location.
There are two types of factors (or criteria) on which location decisions are based: quantitative (or
objective factors and qualitative or subjective) factors. The objective factors involve cost of land,
transportation costs, utilities rates etc. The subjective factors include labour availability, climate,
community environment, quality of life, local politics etc. The presence of objective and subjective
factors results in greater degree of complexity in the structure of the plant location problem as well as
its solution. A decision made on these factors is difficult as they are consistent over all locations. For
example, a plant may be located far from work but have lower utility bills related to the area closer to
work. Some factors may be more dominant than others. For example, on mineral production plants,
raw materials dominate the situations due to which processing is located near mines. On the other
hand, output oriented activities, such as service organizations tend to be located near consumers.
Table 5.1 presents a list of some of the important location factors.

Transportation Utilities Labor Factors Climate, States and

Factors Factors Community, Local
Environment Political
etc. Factors
• Proximity to raw • Power • Labor supply • Climate and • Taxation
material sources living policies
conditions

• Closeness to • Water • Labor • Education • Tax

markets management structure
relations

• Modes of • Fuel • Availability • Community

transportation of skilled attitude
labor

• Transportation • Waste • Labor costs • Religions

costs disposal factors

Table 5.1: Location Factors Determining Plant Location

Steps for Location Planning
Location planning involves the following steps:
(i) Determine the criteria to evaluate location alternatives (for example: minimize costs)
(ii) Identify relevant location factors
(iii) Develop location alternatives
(a) Identify the general region
(b) Identify a small number of community site alternatives
(iv) Evaluate the alternatives and make a selection

46
5.9 Evaluation of Location Alternatives
As stated earlier, the plant location problem involves both qualitative and quantitative factors. Finding
the best location alternative considering all the above factors is not an easy one. Attempts have been
made to combine the qualitative and quantitative factors and score the alternatives. One of the scoring
(or rating) models is outlined below (Table5.2).
The procedure starts by listing the various factors and assigning weight to each factor to represent the
relative importance of various factors. The score for each alternative is found by multiplying each
factor’s score by its weight and summing the results. Table 5.2 gives the details of this rating
approach for two locations A and B.
Table 5.2: Rating Approach

Location factor Weight Score (out of 100) Weighted Score

A B A B
1. Labor Costs 0.40 70 90 70 × 0.40 = 28 90 × 0.40 = 36
2. Water supply 0.10 80 90 8 9
3. Climate 0.15 80 90 12 13.5
4. Proximity to Raw 0.20 40 70 8 14
Materials
0.10 80 90 8 9
5. Transportation
0.05 80 80 4 4
Costs
Total Score 68 85.5

From Table 5.2, we see that location B that has a higher score is preferred. However one has to be
careful in the use of the rating approach because of the assessment of scores, which might have
involved some amount of subjectivity. For example, if the total score for location B were 70, which is
very close to that of A, one need to go for further analysis before arriving at the final decision.

5.10 Distribution problem

The distribution problem is concerned with the allocation of goods flow to minimize overall
distribution cost. Most distribution systems are three-tired structures in which goods start from the
plant; flow to warehouse and ultimately to outlets. Warehousing plays a crucial role in the total
distribution system. Consider for example, a large chain, which manufactures many products,
maintains regional distribution centers and owns a large number of retail store. The firm has control
over the location of all intermediate members of the logistics system.
In the absence of any warehouse, shipments of finished goods would have to be made directly from
the plant to the retail stores. If the plant is located far from the raw materials sources, inbound
transportation costs would be high and delivery times would be high, thereby increasing the chances
of material shortage for production. If the plant is located far from the group of retail stores, then also
transportation costs (i.e. outbound transportation costs) would be high and its takes longer to deliver
orders to retail stores. This may result in out-of-stock situations thereby reducing the level of
customer service. Warehouses placed close to the market can provide quick and efficient delivery to
retail stores, while still permitting the plants to be placed near raw material sources. Warehouses and
distribution centers play an important intermediary between plants and retail stores. They allow a
company to store finished goods for efficient distribution to points of use. The role of warehouses is
illustrated in Figure 5.4

47
 Plant 1 Plant 2 Plant 1 Plant 2

Warehouse

1 2 3 4 5 1 2 3 4 5

Retail Stores Retail Stores

(a) Without Warehouse (a) With Warehouse

Figure 5.4: Role of Warehouse

It can be seen from the above figure that the provision of intermediaries like warehouse reduces
considerably the number of interactions. There are other benefits associated with the provision of
warehouses since they support both manufacturing as well as retail stores. For example, warehouse
facilitates consolidation of orders.
A number of decisions should be made with regard to warehousing. Among the most important
warehousing decisions are the determination of number and location of warehouses. The other
decisions include the following:
• Which products should be stored in each warehouse?
• Should public or private warehouses be used ?
• What type of material-handling equipment should be used ?
• Which customers should be assigned to each warehouse ?
We shall discuss the warehouse location problem in the following section.

5.11 Warehouse location

We present here method(s) for determining the location(s) for warehouse(s). It is assumed that there
are a number of existing facilities in place and we wish to find the optimum location of a new facility
(or new facilities). The existing facilities could be plants, retail stores. The new facilities could be
warehouse. The approaches take into account the locations of plants and markets, volume of goods
moved and transportation costs. All these models focus on minimizing transportation costs.
Measures of Distance
Two measures of distance for movement of items are commonly used: Rectilinear distance and
Euclidean (straight line) distance.
Let the existing facility be located at the point (a, b) and let (x, y) be the location of the new facility.
The rectilinear distance between (a, b) and (x, y) is x − a + + y − b , whereas

The Euclidean distance is ( x − a )2 + ( y − b ) 2

48
Rectilinear distance is appropriate for many location problems such as in metropolitan areas. In many
manufacturing situations, material is transported along aisles arranged in rectilinear pattern.
Fortunately, rectilinear distance problem is easier to solve than Euclidean distance problem.
The problem of locating a simple new facility with respect to a number of existing problems is known
as the single facility location problem whereas the problem of locating multiple new facilities is
known as the multi-facility location problem.

5.12 Single Facility Location Problem

Here, we will discuss about Single Facility Rectilinear Distance Location Problem, Squared Euclidean
Distance Problem (known as the Gravity Problem) and the Straight-Line Distance Problem.
5.12.1 Single Facility Rectilinear Distance Location Problem
Let there be “n” existing facilities located at points (a1 , b1 ), (a 2 , b2 )......, (a n , bn ).

The objective is to locate the new facility to minimize a weighted sum of the rectilinear distance from
the new facility to existing facilities. The goal is to find the values of x and y such that

minimize f ( x, y ) =
n

∑w
i =1
i ( x − a + y − b ), where
i i

wi is the flow of material / goods between the new facility and ith existing facility. The optimum
values of x and y can be determined separately.
f ( x, y ) = g1 ( x) + g 2 ( y ), where
n n
g1 ( x ) = ∑ wi x − ai and g 2 ( y ) = ∑ wi y − bi
i =1 i =1

An example of the single facility location problem could be location of a new storage warehouse for a
company with an existing network of production and distribution centers.

5.12.2 Euclidean Distance Problems

Although the rectilinear distance measure is applicable in many location problems, there are situations
in which the appropriate measure is the Euclidean or Straight-line distance. Location of power
generating facilities so as to minimize the total length of electric cable that must be laid out to connect
the plant and customer is an example where the Euclidean distance measure is appropriate.
We shall discuss here the squared Euclidean Distance Problem (known as the Gravity Problem) and
the Euclidean Distance Location Problem.

5.12.3 The Gravity Problem

The Gravity problem corresponds to the case where the distance measure is square of the Euclidean
distance. This measure is appropriate for location of emergency facilities. The objective is to find the
value of (x, y) to minimize
[
f ( x, y) = ∑ wi ( x − ai ) 2 + ( y − bi ) 2 ]
The solution to this problem is straightforward and is often used as an approximation to the more
common straight-line distance problem.
To find the optimum value of x and y, the partial derivatives of the objective function with respect to x
and y are found and equated to zero.

49
 ∂f ( x, y ) n
We get , = 2∑ wi ( x − a i )
∂x i =1

∂f ( x, y ) n
We get , = 2∑ wi ( y − ai )
∂y i =1

Setting these partial derivatives equal to zero and solving for x and y, we get
n

+
∑w a i i
X = i =1
n

∑w i =1
i

+
∑w a i i
Y = i =1
n

∑w
i =1
i

Thus X+ and Y+ are the weighted averages of x and y coordinates and hence the name Gravity
problem.

5.12.4 The Straight-Line Distance Problem

The straight-line distance measure arises much more frequently than the Gravity problem. The
objective is to find (x, y) to minimize
n
f ( x, y ) = ∑ wi ( x − a i ) 2 + ( y − bi ) 2
i =1

Unfortunately, it is not easy to find the optimum solution mathematically. The partial derivatives
become undefined when the location of the new facility coincides with that of an existing facility.
There are no known simple algebraic solutions; all existing methods require an iterative procedure.
The Gravity solution is usually selected as the starting solution for this iterative process.

5.13 Multi-facility Location Problem

The problem of locating multiple new facilities with respect to existing facilities is known as the
Multi-facility Location Problem. For example, a countrywide consumer goods manufacturer might be
considering where to locate four new regional warehouses. There is interaction among new facilities
as well as between new and existing facilities. In some special situations, multi-facility location
problem can be solved a s a sequence of single facility location problems.
Linear programming can be used to solve the multi-facility rectilinear distance location problem.
Assume that there are “n: existing facilities located at points (a1, b1), (a2, b2), … (an, bn). Suppose the
new facilities are to be located at (x1, y1), (x2, y2), … (xm, ym).
The objective function to be minimized is written as minimize f1 ( x) + f 2 ( y )
m n
Where f1 ( x) = ∑
i≤ j ≤k ≤m
V jk x j − x k + ∑
j =1
∑
i =1
w ji x j − ai

50
 m n
and f 21 ( y ) = ∑
i≤ j ≤k ≤m
V jk y j − y k + ∑
j =1
∑
i =1
w ji y j − ai

Vjk represents the interaction between new facilities j and k and wji represents the interaction between
new facility j and existing facility i. The optimum x and y values can be determined independently as
in the case of single facility location problem.
Multi-facility gravity problems require the solution of a system of linear equations, so that gravity
problems involving large number of facilities are easily solved. Multi-facility Euclidean distance
location problems are solved by using multi dimensional version of the interaction solution procedure
described in the previous section.

5.14 Retail facility location

The major criterion used for retail facility location is the volume of demand and hence estimates of
demand must be known for potential locations. For locating facilities that are oriented toward sales,
the principal factors are market related and the important data are demographic in nature. Other
intangible factors, which affect retail location, are competition, zoning laws, traffic patterns and
accessibility etc. Like in plant location, scoring models may be used to rank potential sites.

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

51
 PART B

ADVANCED TOPICS
 CHAPTER VI
Supply Chain Management and Multi-echelon Inventories

6.1 Introduction
In this section, we consider a multi facility supply chain that belongs to a single firm. The objective of
the firm is to manage inventory so as to reduce system wide cost. Therefore it is important to consider
the interaction of the various facilities and the impact of this interaction on the inventory policy that
should be employed by each facility.

6.2 Multi-echelon inventory system

Multi-echelon models examine the entire system; they might recommend holding just a few parts at
the warehouse level, and none at the retailers, system-wide savings could be enormous. With faster
moving parts of course it is likely that more units would be held at the retail level. Multi-echelon
methods help to determine how many to hold at the retailers and how many to hold at the warehouse
level.

Fig. 6.1: A Three-Echelon System

The first echelons, retail outlets, are replenished from branch warehouses (the second echelon), which
are supplied from a central warehouse (the third echelon). Finally, it is supposed that central
warehouse is replenished from outside sources.
Inventory management in this system is complex because demand at the central warehouse is
dependent on the demand (and stocking decisions) at the branches. And demand at the branches is
dependent on the demand (and stocking decisions) at the retail outlets. More generally, we refer to
this as dependent demand situation (as compared to other inventory control module where demand for
different stock keeping unit was considered to be independent). Multistage manufacturing situation
are conceptually very similar to multi-echelon inventory systems.
6.2.1 Deterministic demand environment
Here external demand rates are known with certainty. The model will reveal the basic interactions
among replenishment quantities at the different situations. Here stocking points are serially connected

Fig. 6.2: A Typical Supply-Chain

We consider here two stages denoted by a warehouse (W) and a retailer (R).

Fig. 6.3: A Two-stage Supply Chain

Here we will discuss the optimum quantity to be ordered for a two-stage supply chain
problem under deterministic environment
Notation used:
D = Deterministic demand at retailer’s end (unit/unit time)
AW = Fixed (set up) cost associated with a replenishment at the warehouse
AR = Fixed (set up) cost associated with a replenishment at the retailer
VW = Unit variable cost or value of the item at the warehouse in Rs/unit
VR = Unit variable cost or value of the item at the retailer in Rs/unit
r = Inventory carrying charge in Rs/Rs/unit time
QW = Replenishment quantity at the warehouse in units
QR =Replenishment quantity at the retailer in units
Here, the two controllable variables are the replenishment sizes QW and QR. For the deterministic case
warehouse QW would be an integer multiple of QR. Therefore, QW = nQR where n=1,2,3, is a positive
integer.
Now, the two decision variables here will be QR and n.

56
 Fig. 6.4: Warehouse Inventory

Fig. 6.5: Retailer Inventory Level

From the above two figures, one can say that
ƒ Inventory at the warehouse does not follow the usual saw-tooth pattern, even though
the end usage is deterministic and constant with time.
ƒ The withdrawals from the warehouse inventory are of size QR.
ƒ Determination of average inventory levels becomes complicated.
To determine the value of inventory, here a concept of echelon stock is introduced. Below, we give
the definition of two types of stock:
Installation Stock: It is simply the stock at the given location without regard for the downstream
stock.
Echelon Stock: To understand multi-echelon inventory, it is necessary to understand the concept of
echelon inventory. As mentioned above, in the distribution system, each stage or level is i.e. the
warehouse or the retailer is referred to as an echelon. The echelon inventory at any stage or level of
the system is equal to the inventory on hand at the echelon, plus all downstream inventories.
For example, the echelon inventory at the warehouse is equal to the inventory at the warehouse, plus
the inventory in transit to the retailer and inventory in stocks at the retailers. Similarly, the echelon
inventory position at the warehouse is the echelon inventory at the warehouse, plus those items
ordered by the warehouse that have not yet arrived minus all items that are backordered. When
backorders are permitted the echelon stock can be negative. Each echelon stock has a saw-tooth
pattern with time and therefore, it is simple to calculate the average value of an echelon stock.

57
However, we cannot simply multiply each average echelon stock by the standard holding cost term
and sum to obtain total inventory carrying costs.
The reason is that the same physical units of stock can appear in more than one echelon inventory.
ƒ When the decision being made is whether to store inventory at an upstream location or
at a downstream location that it supplies, the relevant holding cost is the incremental cost of
moving the product to the retailer.
ƒ The warehouse echelon inventory is valued at,
′ ′
Vw = Vw while the retailer echelon inventory is valued at only VR = VR − Vw .

ƒ In a production assembly context, the echelon valuation V ′ at a particular stage is

given by,
′
Vi = Vi − ∑ V j where the summation is all immediate predecessor J.
The total relevant cost per unit time for the two stages are given by,
AW D A D
TRC (Qw , QR ) = + I W VW′ r + R + I RVR′ r
QW QR

where, IW = Average value of the warehouse echelon inventories in units

I R = Average value of the retailer echelon inventories in units

Using the concept of echelon stock,
′
VW = Vw and VR′ = V R − Vw . Echelon stock follows saw-tooth pattern.

⎡A D Q V ′ r A D Q V ′r ⎤
TRC (n, QR ) = ⎢ W + n R W + R + R R ⎥
⎣ nQR 2 QR 2 ⎦

⎡D ⎛ A ⎞ Q r ⎤
=⎢ ⎜ AR + W ⎟ + R (nVW′ + VR′ )⎥ (1)
⎣ QR ⎝ n ⎠ 2 ⎦
∂TRC D ⎛ A ⎞ r
Now, = − 2 ⎜ AR + W ⎟ + (nVW′ + V R′ ) = 0
∂Q R QR ⎝ n ⎠ 2

⎛ A ⎞
2⎜ AR + W ⎟ D
Or, QR (n ) =
* ⎝ n ⎠
(2)
(nVW′ + VR′ )r
Substituting the value of QR (n ) in eqn. (2), we get
*

⎛ A ⎞
TRC * (n ) = 2 Dr ⎜ AR + W ⎟(nVW′ + VR′ ) (3)
⎝ n ⎠

We have to determine integer value of n that minimizes TRC * (n ) . The value of n that minimizes the
expression is,

58
 ⎛ A ⎞
F (n ) = ⎜ AR + W ⎟(nVW′ + V R′ ) (4)
⎝ n ⎠
For minimization,
∂F (n )
= 0,
∂n
∂TRC ⎡ AW ⎤ ⎛ A ⎞
⇒ = ⎢− 2 ⎥ (nVW′ + VR′ ) + ⎜ AR + W ⎟VW′ = 0,
∂QR ⎣ n ⎦ ⎝ n ⎠

∂TRC ⎛ AW ⎞
⇒ = ⎜ − 2 ⎟VR′ + ARVW′ = 0,
∂QR ⎝ n ⎠
AW VR′
⇒ n* = (5)
ARVW′
which in general, will not be an integer.
If n * < 1 , assume n = 1 and calculate the respective value. Otherwise, ascertain two-integer values
n1 and n 2 that surround n * .

⎛ A ⎞
F (n1 ) = ⎜⎜ AR + W ⎟⎟(n1VW′ + VR′ )
⎝ n1 ⎠

⎛ AW ⎞
and, F (n2 ) = ⎜⎜ AR + ⎟⎟(n2VW′ + VR′ ) .
⎝ n2 ⎠
If F (n1 ) ≤ F (n2 ) use n = n1 ,

If F (n1 ) > F (n2 ) use n = n2 .

Then evaluate,

⎛ A ⎞
2⎜ AR + W ⎟ D
⎝ n ⎠
QR = , then calculate, QW = nQ R . (6)
(nVW′ + VR′ )r

6.2.2 Multi-echelon stocking points with time varying demand

Here, end item demand is known but varies from period to period. D( J ) in period J . Carrying costs
are incurred only on period ending inventories. Now we have a multi-stage assembly or distribution
system and we will discuss the situation for an assembly structure.
One approach would be simply to use sequential decision echelon by echelon. As for example, Silver-
Meal heuristic could be utilized to schedule replenishments for the retailer. This would imply a
pattern of requirements for the warehouse (a primary processing stage in production), which would
then be used as input to the Silver-Meal heuristic to plan the replenishments of the warehouse.
Though this approach is simple, it ignores the cost interdependency of the two echelons. Specifically,
in choosing the replenishment strategy at the retailer, the method does not take account of the cost
implications at the warehouse. Here, a procedure is developed that can be applied sequentially to keep
implementation effort at the reasonable level.

59
It captures the essence of cost interdependencies. An examination of level demand case provides
considerable insight.

⎡D ⎛ A ⎞ Q r ⎤
TRC (QW , QR ) = ⎢ ⎜ AR + W ⎟ + R (nVW′ + VR′ )⎥ where, QW = nQR .
⎣ QR ⎝ n ⎠ 2 ⎦
This problem is analogous to a single echelon problem (i.e. the selection of QR ) if the adjusted fixed
replenishment cost is represented as,
A
Aˆ R = AR + W ,
n
and adjusted unit variable cost of the item is, VˆR = nVW′ + VR′

AW
The term reflects that there is a warehouse setup only every nth retailer setup.
n

⎛ A ⎞
2⎜ AR + W ⎟ D
We can select QR from, QR = ⎝ n ⎠
.
(nVW′ + VR′ )r
Properly taking account of the cost impact at the warehouse if we have a good pre-estimate of n.
Further, we can use the steps followed in the earlier problem.
Step 1

AW VR′
n* = (7)
ARVW′
Step 2
Ascertain the two integer values n1 and n 2 that surround n * .
Step 3

⎛ AW ⎞
Evaluate, F (n1 ) = ⎜⎜ AR + ⎟⎟(n1VW′ + VR′ )
⎝ n1 ⎠
⎛ AW ⎞
and, F (n2 ) = ⎜⎜ AR + ⎟⎟(n2VW′ + VR′ ) .
⎝ n2 ⎠
If F (n1 ) ≤ F (n2 ) use n = n1 ,

If F (n1 ) > F (n2 ) use n = n2 .

AW VR′
But, Blackburn and Miller have found that simply using n =
*
and ensuring that n is ‘at
ARVW′
least’ unity works well, particularly for more complex assembly structure.
⎡ AW V R′ ⎤
Thus, n = ⎢ , 1⎥ (8)
⎢⎣ ARVW′ ⎥⎦
For time-varying demand, compute n and then employ n value to obtain adjusted setup and unit
variable costs for the retailer according to following equation,

60
 A
Aˆ R = AR + W ,
n
VˆR = nVW′ + VR′

The Silver-Meal heuristic can then be applied to the retailer using ÂR and VˆR . The resulting
replenishment again implies a replenishment pattern for the warehouse. Subsequently, the Silver-Meal
or another lot-sizing procedure is used at the warehouse with Aw and Vw .

6.3 Coordinated Ordering

In the earlier inventory control problem studied in the literature, it is assumed that different items in
an inventory could be controlled independently.
In this part, we now relax this assumption and consider situations where there is a need to coordinate
orders for different items. When coordinating the replenishments for different items, it is common to
use cyclic schedules and especially so-called ‘power of two policies’. One reason for coordinated
replenishments is that, we want to trigger orders for a group of items at the same time. This can be
advantageous in many situations.
• It may be possible to get a discount if the total order from the same vendor is greater
than a certain break point.
• It may also be possible to reduce the transportation costs, e.g., by filling a truckload.
• Sometimes, set up costs can also be lowered substantially if a group of items are
produced together in a machine.

6.3.1 Power of two ordering policies

It is very common to use power of two policies in connection with coordinated replenishments. Here
cycle times are restricted to be powers of two times of a certain basic period.
Suppose we order two items with cycle times 5 weeks and 7 weeks respectively. Now suppose both
the items are ordered in week 1. Then item with cycle time 5 week is ordered in week 1, 6, 11, 16, 21,
26, 31, 36 etc and the other item in weeks 1, 8, 15, 22, 29, 36 etc.
Thus week 36 is the first time after week 1 when both the items are ordered together.
ƒ We will show that a restriction of power of two policy will give a solution, which is very
close to the optimal solution. If we could somehow synchronize the order interval so that
order for different products often arrives on same day, we could greatly reduce our
coordination costs.
ƒ Roundy (1985) devised an elegant and simple method called ‘power of two ordering
policies’ to ensure that order for multiple products are well synchronized.
ƒ Roundy proved that using the power of two policy, to round the order interval to a
neighboring power of 2 will increase the sum of ordering and holding cost at most 6%. The
result of power of two policy is that different products will be ordered frequently at the same
time. In many cases, this will greatly reduce coordination cost. In most circumstances this
policy will reduce coordination cost by more than maximum possible 6% increase in total
cost.

6.3.2 Proof of Roundy’s result

For an arbitrary order quantity Q, the total inventory related cost is given by,

61
 TC (Q ) =
AD hQ
+
Q 2

⇒ TC (T ) =
A hDT A hD
+ = + gT , where, g = .
T 2 T 2
( ) 2 ADh ,
Total cost at EOQ is given as TC Q * =

Or, Optimal cost per unit time TC (T ) = 2 Ag .

*

It is natural to consider policies where the reorder interval T is restricted to values that can be easily
implementable.
Now, we put power of two restrictions and T is restricted to be a power of multiple of some fixed base
planning period TB. B

T = TB 2 k , where, k ∈ {0, 1, 2, 3,.....} (9)

Such policy is called power of two policy. The base planning period TB may represent a day, week or
month, etc and is usually fixed beforehand. It represents minimum possible reorder interval.
Now, the basic question is that how does one find the best power of two policy that minimize the cost
over all possible power of two policy?
Secondly, how much this best power of the policy deviates from the optimal policy?

A
T* = is the optimal reorder interval under unrestricted condition and let T be the optimal power
g
of two reorder interval.
Since total cost is a convex function, the optimal k in (9) is the smallest integer k satisfying the
condition.
TC (TB 2 k ) ≤ TC (TB 2 k +1 )

⎛ A k ⎞ ⎛ A k +1 ⎞
⇒ ⎜⎜ k
+ gTB 2 ⎟ ≤ ⎜
⎟ ⎜ T 2 k +1 + gTB 2 ⎟⎟
⎝ B
T 2 ⎠ ⎝ B ⎠
⎡ 1⎤
⎢1 − 2 ⎥ ≤ gTB 2 (2 − 1)
A
⇒ k

TB 2 k ⎣ ⎦

≤ (TB 2 k )
A 2
⇒
2g

⇒
1 A
g
(
≤ TB 2 k )
2
1
⇒ T* ≤T (10)
2
By the definition of optimal k, it must also satisfy the following condition since the cost curve is
convex in nature that is,
TC (TB 2 k ) ≤ TC (TB 2 k −1 )

62
 ⎛ A ⎞ ⎛ A ⎞
⇒ ⎜⎜ k
+ gTB 2 k ⎟⎟ ≤ ⎜⎜ k −1
+ gTB 2 k −1 ⎟⎟
⎝ TB 2 ⎠ ⎝ TB 2 ⎠
A ⎡ 1 ⎤ k⎛ 1⎞
⇒ ⎢1 − 2 −1 ⎥ ≤ gTB 2 ⎜ − 2 ⎟
TB 2 k ⎣ ⎦ ⎝ ⎠

⇒
2A
k
(
≥ gTB 2 k )
TB 2

≥ (TB 2 k )
2A 2
⇒
g

A
⇒ TB 2 k ≤ 2
g

⇒ T ≤ 2T * (11)
1
From equation (10) and (11), T * ≤ T ≤ 2T * (12)
2
Hence the optimal power of two policy for a given base planning period TB must be in the interval
B

⎡ 1 * ⎤
⎢ T , 2T * ⎥ .
⎣ 2 ⎦
The maximum discrepancy between the total cost for the power of two ordering policy and the total
1
cost for T * will occur if power of two reorder interval equals either 2T * or, T*.
2
We have already derived that,

TC (Q ) ⎛ AD hQ ⎞ 1 1 ⎡ Q Q* ⎤
=⎜ + ⎟ = ⎢ +
( )
TC Q * ⎜⎝ Q
⎥,
2 ⎟⎠ ADh 2 ⎣ Q * Q ⎦

Q* Q
Since, T * = and T =
D D
TC (T ) 1 ⎡ T T * ⎤
= ⎢ *+ ⎥
( )
TC T * 2 ⎣T T ⎦

1
Now, upper bound of T = 2T * and lower bound of T = T*.
2
⎛ 1 ⎞
TC ⎜ 2T * , or , T*⎟
⎝ 2 ⎠ 1 ⎡ 2T T* ⎤ 1 ⎡ 1 ⎤
*
Then, = ⎢ * + = ⎢ 2+ ⎥ = 1.06.
TC T * ( ) 2⎣ T *⎥
2T ⎦ 2 ⎣ 2⎦

Thus, average inventory purchasing and carrying cost of the best power of two policy is guaranteed to
be within 6% of the average cost of overall minimum policy.

Let us consider the following example to understand the power of two ordering policy
Three products are with reorder intervals of 3.5 days, 5.6 days and 9.2 days respectively

63
 (
T * = 2 2 k TB , )
Let, TB =1 period.
Starting with power of two k =0,

⇒ 3.5 ≤ 2 2 0 . It is found that condition is not satisfied. We now increase the power k to 1 and
check whether it satisfies the condition. It is found that

3.5 ≤ 2 21
⇒ 3.5 ≤ 1.44 x2 ≤ 2.88 and therefore not satisfied the condition.
Again increasing the power k to two, it is found that

3.5 ≤ 2 2 2
3.5 ≤ 2 x 4
⇒ 3.5 ≤ 1.44 x4 ≤ 5.76 , it satisfies the condition i.e. power of 2 2 =4 weeks.
Similarly for second case, 5.6 ≤ 2 2 2 ≤ 5.76 . This will satisfy the condition, so follow order
interval 2 2 =4 weeks.
Similarly for the third case, 9.2 ≤ 1.44 x 2 3
9.2 ≤ 1.44 x 2 3
⇒ 9.2 ≤ 1.44 x8 , it satisfies the condition. So, follow the order interval 2 3 =8 weeks.
So, according to Roundy’s policy, quantity should be ordered at 2 2 , 2 2 , and 2 3 order interval time.
Thus orders for two items will be given at the end of 4 weeks whereas order for the third item should
be given at the end of 8 weeks.

Reference
Silver,E A., Pyke. D.F, Peterson, R., and; Inventory Management and Production Planning and
Scheduling; Publisher: John Wiley and Sons

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

64
 CHAPTER VII
Supply Chain Contract and Coordination

7.1 Introduction
Managing the flows in the supply chain network implies the presence of many decision makers within
the supply chain where each one operates a part of it. These decision makers could be either distinct
firms or managers of different departments within a firm. Every individual decision maker will
attempt to maximize his own profit keeping in mind that others will also do the same thing. This
individual competetive behavior of the members of the supply chain adversely affect the overall
performance goal of the supply chain.
Thus to avoid this undesirable situation, one of the most important issue arises in the management of
supply chain is how to have perfect coordination amongsts the members of a supply chain to have
cost savings and increase in channel efficiency. Philosophy of supply chain contracting is to develop
coordiantion policies through pair wise interaction of an upstream (supplier) and downstream
(retilaer) agent at a period of time. Decision-making at different levels in the organization should be
so coordinated that operating policies are optimal for the organization as a whole. Without
coordination, improvement at one level may be lost due to inefficiencies at another level. As for
example, reducing inventory at one level may not be beneficial if it is accumulated in any succeeding
stage. To avoid this undesirable situation, academicians have studied the issue and developed various
mathematical models. Porter (1985) has mentioned that cooperative relationship between buyer and
supplier is not a zero sum game in which one gains only at the expense of the other, but a relationship
in which both gain. Coordination is actually in the form of cooperative decisions, that is, the
individual entities make decisions, which are in the best interests of the entire supply chain.
As discussed above, it is clear that in a supply chain (SC), there are multiple firms owned and
operated by different parties, and each of these firms takes decisions, which are in line with their own
goals and objectives. As in all decentralized systems, the actions chosen by SC participants might not
always lead to the “optimal” outcome if one considers the supply chain as one entity. That is, since
each player acts out of self-interest, we usually see inefficiencies in the system, i.e., the results look
different than if the system was managed “optimally” by a single decision-maker who could decide on
behalf of these players and enforce the type of behavior dictated by this globally (or centrally) optimal
solution. In this section, we will take a look at the nature of inefficiencies that might result from the
decentralized decision-making in supply chains, and if and how one can design contracts such that
even though each player acts out of self interest, the decentralized solution might approach the
centralized optimal solution. For excellent reviews on SC contracts and coordination, we refer the
reader to Tsay, Nahmias and Agrawal (1999) and Cachon (2001).
Mechanisms of channel coordination and vertical control in production/distribution channels have
received attention from several disciplines such as Economics, Marketing apart from Operations
Management. The models in Economics literature generally assume a deterministic demand function.
Marketing literature has focused on channel coordination to maximize joint profits. Though
Marketing and Economics literature provide different motivations for coordination yet, ultimately the
motivations share the common objective of maximizing system welfare.

7.2 Supply chain Contract

To better formalize the coordination ideas, we will look at a simple stylized two-stage supply chain
with one supplier and one retailer, where the retailer buys goods from the supplier and sells them in
the end market:
For simplicity, assume that the supplier is uncapacitated, and has unit cost of production c. The
retailer faces a market where the price is inversely related to the quantity sold. For simplicity, let us
assume a linear demand curve P = a - bq where P is the market price and q is the quantity sold by the
retailer6. Assume that all of this information is common knowledge.
First let us consider the simple “wholesale price” contract where the supplier charges the retailer

DSC CSC

Supplier’s wholesale price (w) w = (a + c)/2 w

Retailer’s quantity (q) q = (a - c)/4b Q* = (a - c)/2b

Market Price (P) P = (3a + c)/4 P* = (a + c)/2

Supplier’s profit (Πs) ΠS = (a - c)2/8b Π*S = (w - c)q

Retailer’s profit (ΠR) ΠR = (a - c)2/16b Π*R = (P - w)q

Total SC profits (Π) Π = 3(a - c)2/16b Π= (a - c)2/4b

Table 1: Wholesale price contract versus CSC.

w per unit. The supplier’s and the retailer’s profits are ΠS = (w - c)q and ΠR = (a - bq - w) q,
respectively. The supply chain’s profits are Π = ΠS + ΠR = (a - bq - c)q. Note that the choice of w
only indirectly affects the total SC profits, since the choice of w impacts the choice of q.
Decentralized Supply Chain (DSC)
As in most real-world supply chains, suppose that the supplier and the retailer are two independently
owned and managed firms, where each party is trying to maximize his/her own profits. The supplier
chooses the unit wholesale price w and after observing w, the retailer chooses the order quantity q.
Note that this is a dynamic game of complete information with two players, supplier and retailer,
where the supplier moves first and the retailer moves second. Hence, we can solve this game using
backwards induction. Given a w, first we need to find the retailer’s best response q(w). The retailer
will choose q to maximize ΠR = (a - bq - w)q. This is a concave function of q, and hence from FOC
we get
∂π R
= a − 2bq − w = 0
∂q
a−w
⇒ q ( w) =
2b

66
Next, given the retailer’s best response q(w) = (a-w)/2b, the supplier maximizes ΠS = (w-c)q = (w -
c)(a - w)/2b. This is a concave function of w and from FOC we get
∂π s a+c
= a−w−w+c = 0 ⇒ w =
∂w 2
The equilibrium solution for this decentralized supply chain is given in the second column of Table 1.
In this contractual setting, the supplier gets two-thirds of the SC profits, the retailer gets only one-
third. This is partly due to the first-mover advantage of the supplier.
Now, let us consider a centralized (integrated) supply chain (CSC) where both the retailer and the
supplier are part of the same organization and managed by the same entity. Centralized Supply Chain
(CSC): In this case there is a single decision-maker who is concerned with maximizing the entire
chain’s profits Π= (a-bq -c)q. This is a concave function of q and from first order condition (FOC),
we get
∂π a−c
= a − 2bq − c = 0 ⇒ q* =
∂q 2b
The solution for the CSC is given in the third column of Table 1. From Table 1, we see that the
quantity sold as well as the profits are higher and the price is lower in the CSC than in the DCS.
Hence, both the supply chain and the consumers are better off in the CSC. What about the retailer and
the supplier? Are they both better off, or is one of them worse off in the CSC? What is the wholesale
price? How does the choice of w affect the market price, quantity, and the supply chain profits? A
closer look would reveal that w has no impact on these quantities. Any positive w would result in the
same outcome for the CSC because the firm would be paying the wholesale price to itself ! However,
the choice of w in the CSC is still very important as it determines how the profits will be allocated
between the supplier and the retailer. We can interpret w as a form of transfer payment from the
retailer to the supplier. What is the minimum w that is reasonable? For positive supplier profits, we
need w ≥ c. If we set w = c, the supplier’s profits are zero whereas the retailer captures the entire
supply chain’s profits. What is the w that splits the SC profits equally between the retailer and the
supplier? If we set w = (a+3c)/4, w -c = P -w = (a-c)/4 and each party’s profits are (a-c)2 /8b. Note that
this is the same as the supplier’s profits in the DSC. Hence, if the supplier and the retailer split the
profits equally in the CSC, the supplier is at least as good, and the retailer is strictly better off than in
the DCS. In the DSC, the outcomes are worse for all the parties involved (supplier, retailer, supply
chain, and consumer) compared to the CSC, because in the DSC both the retailer and the supplier
independently try to maximize their own profits, i.e., they each try to get a margin, P - w and w - c,
respectively. This effect is called “double marginalization” (DM).
In a serial supply chain with multiple firms there is coordination failure because each firm charges a
margin and neither firm considers the entire supply chain’s margin when making a decision.
In this stylized model, the profit loss in the DSC due to DM is 25% (also referred to as the DM loss).
It is clearly in the firms’ interest to eliminate or reduce double marginalization, especially if this can
be done while allocating the additional profits to the firms such that both firms benefit. This simple
model suggests that vertical integration could be one possible way of eliminating double
marginalization. However, for reasons we discussed at the beginning of this chapter, vertical
integration is usually not desirable, or not practical. Then the question is, can we change the terms of
the trade so that independently managed companies act as if they are vertically integrated? This is the
concept known as “supply chain coordination.” In this stylized model, the retailer should choose q* =
(a - c)/2b in any coordinating contract. One can easily think of some very simple alternative contracts
to eliminate double marginalization:

67
7.2.1 Take-it-or-leave-it-contract
The supplier offers the following contract to the retailer: Buy q* at the wholesale price w = (a + c)/2,
or nothing. In this case the supplier’s profit is Π, i.e., the supplier captures 100% of the CSC profits.
The supplier offers the contract and grabs whole of the CSC profit as the first-mover advantage. This
contract requires a very powerful supplier.

7.2.2 Marginal pricing

The supplier sets w = c. In this case, the retailer’s profit is Π*, i.e., the retailer captures 100% of the
CSC profits. The supplier sets the wholesale price equal to its marginal price, so that the retailer grabs
whole of the CSC profit. This contract requires a very powerful retailer.
Note that the take-it-or-leave-it contract would require a very powerful supplier whereas the marginal
pricing contract would require a very powerful retailer. In practice, neither the supplier nor the retailer
is so powerful in general to dictate such contract terms. Hence, we need to consider alternative
contracts that coordinate the supply chain.
Above cases are the extreme cases of supply chain contracting and are not easily implementable when
both supplier and retailer have similar bargaining powers. The goal of negotiations of supply chain
contracting is to design “channel coordinated” contracts where the players’ Nash equilibrium
coincides with the supply chain optimum1 and at the same time satisfies the conditions of individual
rationality (IR) and incentive compatibility (IC) (Wu, 2004). Before getting into the types of supply
chains that fall within these extreme cases we discuss a few characteristics of supply chain contracts
in the following.
Profitability : Achieve profits close to optimum
Fairness and flexibility Allow the flexible division of profits
Implementability : Should be easy and low-cost to administer

7.2.3 Revenue Sharing Contract

In a revenue sharing contract the retailer shares a fraction α < 1 of his revenues with the supplier.
Cost of each unit sold c = c S + c R , where, c S and c R are the unit costs of the supplier and the
retailer.
R(q ) = retailer’s revenue as a function of quantity q .
∂R(q )
In the CSC, the profits are given by Π = R(q ) − cq and from FOC we have, = c . This means
∂q
that, the marginal revenue is equal to the marginal cost at the optimum.
In the DSC, the retailer’s profits are Π R = R(q ) − (w + c R )q where the wholesale price is w . From
∂R(q )
FOC we have, = w + c R . Note that q < q * if w > c − c R , i.e., unless the supplier sells at
∂q
marginal cost, the retailer orders less than the optimal CSC quantity resulting in double
marginalization.
The retailer’s profits in the revenue sharing contract are Π R (q ) = αR(q ) − (w + c R )q

∂R (q ) (w + c R ) ∂R(q )
From FOC we have, = . We had = c in the integrated chain. Hence, by
∂q α ∂q
setting the two right hand sides equal to each other, we achieve the same quantities CSC and DSC.

68
That is w + c R = αc , i.e., w = αc − c R , we have marginal revenue equal to marginal cost in the DSC
as well, and q = q* . In this case, the retailer’s profit is
Π R = αR(q ) − (αc − c R )q − c R q = α (R (q ) − cq ) = αΠ , i.e., the retailer captures α fraction of
* * *

the optimal SC profit.

Note that in the revenue sharing contract, the retailer’s objective function becomes an affine
transformation of the supply chain’s objective function, i.e., the retailer’s and the supply chain’s
objectives are perfectly aligned. Since the total SC profits are higher compared to the traditional
wholesale price contract, partners can choose α such that both parties benefit. The choice of α
depends on several factors, including the bargaining power of the supplier and the retailer.
Drawbacks:
1. The additional cost and administrative burden it creates compared to the straightforward
wholesale price contract. This takes an organizational effort to set up the deal and follow its
progress. It is worthwhile to go into such contracts only if the increase in profits is relatively
large compared to the additional cost and the administrative effort.
2. In case of multiple retailers, coordination is not guaranteed unless the supplier has the
flexibility to offer different contracts to different retailers. Unfortunately, such discriminations
may not be possible always due to legal considerations.
3. This contract looses its appeal if the revenues depend on retailer’s sales effort. For a retailer
who is taking only a fraction of the revenues he generates, the incentive to improve sales goes
down.

Example of revenue sharing contracts implemented in practice-

Blockbuster is a retailer, which purchases movies from the studios (suppliers) and rents them to
customers. The supplier’s wholesale price impacts how many videos Blockbuster orders and hence,
how many units are eventually rented by customers. Before 1998, the price of purchasing a tape from
the studio was very high, around $65. Given that rental fees are in the order of $3-4, Blockbuster
could purchase only a limited number of videos and this resulted in lost demand; especially during the
initial release period, where the demand was high (peak demand usually lasts less than 10 weeks),
20% of customers could not find what they were looking for on the shelf. Hence, the studio’s high
wholesale price impacted the quantity purchased by Blockbuster, and in turn, the revenues and the
profitability of both firms. Seeing this problem, Blockbuster and the studios went into a revenue
sharing agreement. According to this, Blockbuster pays only $8 per tape initially, but then gives a
portion (somewhere around 30 to 45%) of the revenues generated from that tape back to the supplier.
Since this agreement reduces Blockbuster’s initial investment in movies, it orders more tapes from the
studio, hence, is able to meet more demand, generate more revenues, and give back a portion of those
revenues back to the supplier. Blockbuster increased its overall market share from 25% to 31% and its
cash flow by 61% using this agreement. This is clearly a win-win situation. The supplier might be
better of even if he sells each unit below its productioncost. A similar agreement is used between
studios and theaters as well. (Cnet News.com, October 18, 2000) [7]

7.2.4 Buyback Contract

One of the reasons of ordering less than the optimal quantity by the retailer is due to risk of excess
inventory. Buyback contract allocate the risk between the supplier and the retailer. Retailer can return
the goods to the supplier and get some money back. In this contract the supplier purchases the leftover
units at the end of the selling season for b per unit, where b < w .
The retailer’s revenue = R(q ) = pS (q ) , where p is the unit selling price and S (q ) is the total
amount of sales made by the retailer, where S (q ) < q .

69
In the DSC, the supplier’s profit is
Π R = revenue + returns – cost – purchase cost
⎛ R(q ) ⎞ ⎛ b⎞
= R(q ) + b⎜⎜ q − ⎟⎟ − (c R + w)q = R(q )⎜⎜1 − ⎟⎟ − (c R + w − b )q
⎝ p ⎠ ⎝ p⎠

( ( ) )
In revenue sharing contracts we have, Π R = αΠ * = α R q * − cq * , i.e., the retailer’s profit is an
affine transformation of the centralized supply chain’s profits. Hence the retailer’s optimal quantity is
the same as the centralized supply chain’s optimal quantity. For a similar situation to hold buyback
contracts, the retailer’s revenue should be equal to αΠ * , i.e.,

⎛ b⎞ ⎛ b⎞
R(q )⎜⎜1 − ⎟⎟ − (c R + w − b )q = αΠ * and we need to have (c R + w − b ) = αc and ⎜⎜1 − ⎟⎟ = α .
⎝ p⎠ ⎝ p⎠

⎛ b⎞
⎜⎜1 − ⎟⎟ = α ; ⇒ b = p(1 − α )
⎝ p⎠
and, (c R + w − b ) = αc; ⇒ wb = αc + b − c R = p (1 − α ) + αc − c R

Hence, if the supplier chooses w and b in this fashion, the retailer will get a fraction of the CSC’s
profits.
In the revenue sharing contract we had, w = αc − c R . In buyback contracts, we have,
wb = p(1 − α ) + αc − c R and, b = p(1 − α ) . That is, wb = b + w . Hence, in the buyback contract
the supplier charges a little more for the wholesale price and in return guarantees to give back b for
any unit that is not sold.
In revenue sharing contracts the wholesale price did not depend on the selling price. But in a buyback
contract, the wholesale price and the return price depend on the selling price. Hence, in a buyback
contract the supplier should set w and b as functions of the selling price p. Is the buyback contract
flexible in terms of how the profits are shared between the supplier and the retailer? The answer is
yes. Actually, for every revenue sharing contract, there is an equivalent buyback contract and vice
versa.

7.2.4 Quantity flexibility contract

In quantity flexibility contract, the supplier provides full refund for returned (unsold) items as long as
the number of return is no larger than a certain quantity. Thus this contract gives full refund for a
portion of the returned items, whereas a buyback contract provides partial refund for all returned
items.

7.2.5 Two-Part Tariff

In a two-part tariff, the supplier charges a fixed fee F and a wholesale price w per unit.
∂R(q )
In the CSC, the optimality condition is = c = c R + c S , i.e., marginal revenue is equal to
∂q
marginal cost.
In the DSC, under two-tariff the retailer’s profit is Π R = R(q ) − (w + c R )q − F . From the FOC

70
∂Π R ∂R(q ) ∂R(q )
= − w − cR = 0 ⇒ = w + cR
∂q ∂q ∂q
In order to have the CSC solution, we need w = c s , i.e, supplier must sell at cost. In this case the
( ) ( )
retailer gets Π R = R q * − (w + c R )q * − F = R q * − (c S + c R )q * − F = Π * − F . Supplier’s profit
is Π S = F + (w − c S )q = F .
*

Notice that F determines the allocation of profits between the supplier and the retailer. This is a
flexible contract as any allocation is possible.

7.2.6 Quantity Discount Contract

In quantity discount contract ifs, if the retailer purchases more units then the supplier lowers the unit
price, i.e, the supplier charges w(q ) where w is a decreasing function of q . In general w(q ) is a step
function. But for simplicity we assume it is continuous and differentiable.
The retailer’s profit in DSC is Π R = R(q ) − (w(q ) + c R )q . We observed earlier that the DSC has the
same optimal quantity as the CSC, if Π R is an affine transformation of Π . Hence we need,
⎛ R(q * ) ⎞
Π R = R (q * ) − (w(q * ) + c R )q * = α (R (q * ) − c )q * ⇒ w(q * ) = (1 − α )⎜⎜ * ⎟⎟ − c R + αc .
⎝ q ⎠
⎛ R(q ) ⎞
If the supplier sets the wholesale price as w(q ) = (1 − α )⎜⎜ ⎟⎟ − c R + αc , there is a tradeoff for the
⎝ q ⎠
retailer. If he chooses q < q * , he will pay too much per unit, increasing his marginal cost. If he
chooses q > q * , then the unit price will decrease but the marginal revenue will decrease more, making
the extra unit unprofitable. Hence the optimal quantity choice for the retailer is q * . The supplier’s
profit in this case is
⎛ (1 − α )R(q * ) ⎞
Π s = (w − c s )q = ⎜⎜
*
*
− c R + αc − c S ⎟⎟q * = (1 − α )R (q * ) − (1 − α )q * c = (1 − α )Π
⎝ q ⎠
In both revenue sharing and quantity discount contract the retailer’s (and the supplier’s) revenue is
proportional to the centralized supply chain’s profit. If there is demand uncertainty, in a quantity
discount contract the retailer bears the risk.

7.3 Present trend in the study of supply chain coordination from Operations Management
perspective
In the last couple of years, interest in the field of supply chain coordination from Operations
Management perspective has grown considerably. One line of research employs quantity discount and
quantity commitment as coordination mechanism (Aderohunmu et al. 1995; Lariviere, 1999). Further,
authors like Ertogral et al. (2001) have put emphasis on the need to incorporate negotiation process in
supply chain coordination. Negotiation process focusses on dynamic sharing of surplus between the
two parties where both can take part in the decision. Here, the negotiation ends with the win-win
feeling for both. This is considered superior to a pre-determined static division of surplus through side
payment strategy for a decentralized supply chain.
Many authors (e.g. Kohli and Park, 1989) have assumed that both buyer and the supplier have full
information about each other. But in practice, such a comfortable situation hardly exists. Realizing
this fact, recently some authors (e.g. Corbett et al., 2000; Ha, 2001) have incorporated the information
asymmetry factor in their models of supply chain coordination problem.

71
Another area of supply chain coordination that has drawn the attention of researchers is on
development of suitable mechanism to coordinate the logistic processes that are controlled by various
companies. Swenseth et al. (2002) have reported that often about 50% of total annual logistics cost of
a product can be attributed to transportation cost. Therefore, for the overall performance improvement
of the supply chain, there is a need to develop coordination mechanism to coordinate the logistics
processes between the various parties of a supply chain. Particularly, in the multi buyers case where
buyers are located in different geographical regions, individual shipments to the buyers by the vendor
increases the total system cost. In such a situation, coordinated shipment from the vendors to multiple
buyers helps to reduce channel cost.

7.4 Supply chain coordiantion through Quantity discount

Our intention here is to cover only quantity discount supply chain coordination mechanism literature.
Since quantity discount is considered to be one of the most popular mechanisms of coordination
between the business entities, this section primarily investigates supply chain coordination models
that have used quantity discount as coordination tool under deterministic environment. Here, the word
vendor, supplier and manufacturer is used alternatively to represent the same upstream member in the
supply chain who sells the item to the buyer unless specifically mentioned.
In many practical situations, it is found that manufacturer has a high production setup cost or high
shipping cost. In such situations, the manufacturer prefers to produce or ship the product in large
quantities. The manufacturer encourages buyers to order in large quantities by offering some
incentives. Similarly, manufacturer also by ordering larger quantity to its raw material supplier may
ask for some incentives from his supplier. Larger orders will lead to fewer productions setup. Larger
orders may allow the vendor to take the advantage of economy of scale in transportation. Thus, it
helps in improving the overall efficiency of the channel. The joint economic lot sizing literature has
examined the case where the supplier wishes to induce the buyer to choose a higher lot size than she
would of her own accord. Thus, in any supply chain, where buyer, manufacture and supplier are
involved, there is also a scope to reduce channel cost and inefficiency. This is possible if channel
partners jointly optimize order quantities/batch sizes and agree to the resultant benefits. Recently,
some authors (e.g. Munson et al., 2001; Yang et al., 2001) have extended the two-stage supply chain
model to a three-stage supply chain model where the raw material supplier to the manufacturer is the
third partner.
Quantity discount and credit by supplier to the buying firms are used as instruments for developing
coordination mechanism between the two parties. Gurnani (2001) in his recent work has mentioned
that industrial supplier-buyer relations have undergone significant changes with increasing emphasis
on coordination and information sharing. This is made with an attempt to reduce transaction related
costs and to capture a larger market share.

Notation:
Subscript 1 and 2 represent vendor and buyer respectively,
D = the buyer’s annual demand for the product,
Si = Setup and ordering cost for the firm i, i = {1,2}
ri = Annual inventory holding cost expressed as a percentage of the value of the item for the
firm i, i = {1,2}
Q = the buyer’s order quantity
M2 = the vendor/manufacturer’s gross profit on sales expressed as a percentage
dk = discount per unit offered by the manufacturer
R2 = the manufacturer’s production rate in unit per year

72
P0 = the buyer’s base purchase price without quantity discount
C2 = the manufacturer’e manufacturing cost per unit excluding order processing, setup, and
inventory holding costs per unit

7.4.1 Background
Study of integrated inventory models can be viewed as one of the origin of supply chain coordination
study from Operations Management perspective. These models mainly examine the benefits accrued
in the system due to coordination in order quantities between the two parties. Earlier, Goyal and
Gupta (1989) have reviewed the literature of buyer vendor coordination models. Benton and Park
(1996) and Munson and Rosenblatt(1998) have also reviewed some of the papers discussed here
under different context. We have mainly considered here the literature of channel coordination/
supply chain coordination models that have operations approach. Operations approach mainly
concentrates on the operating cost of the channel. Operating cost is considered as a function of
retailer’s / buyer’s order quantity where a fixed retail price is assumed and this leads to a fixed final
demand.
The traditional inventory model assumes that a rational buyer would prefer to purchase his optimal
order quantity (EOQ) as any deviation from this quantity would increase his total cost. The buyer’s
annual total cost for order quantity Q can be expressed as

⎛D⎞ ⎛Q⎞
TC (Q ) = P0 D + ⎜⎜ ⎟⎟ S1 + ⎜ ⎟r1P0 (1)
⎝Q⎠ ⎝2⎠
When quantity discount is not allowed, the buyer’s optimal order size is given as

. Thus, the total annual cost is given as TC (Q* ) = P0 D + 2 DS1r1 P0

2 DS1
Q* = (2)
r1P0
Corresponding to buyer’s order quantity Q, the vendor’s/ manufacturer’s yearly net profit considering
only order processing and setup costs for lot for lot policy can be written as

⎛ DS ⎞
YNP2 = DM 2 P0 − ⎜⎜ 2 ⎟⎟ (3)
⎝ Q ⎠

The total channel cost is the sum of the individual cost component of buyer and vendor respectively

and can be writen as TCC (Q) = P0 D +

D
(S1 + S2 ) + ⎛⎜ Q ⎞⎟r1P0 (4)
Q ⎝2⎠
Vendor’s/ manufacturer’s order processing and setup cost per order are considered to be larger than
the buyer’s order processing cost per order. If a buyer adopts his EOQ as the order quantity for
minimizing his total annual costs, the vendor/ manufacturer incurs a significant cost penalty.
Therefore, vendor /manufacturer induces the buyer through quantity discounts to order larger
quantity to maximize his profits. Manufacturer can maximize his profit when the lot size is as high as
infinity ! When buyer purchases in larger order quantity Q > EOQ then there is an increase in profit
for the vendor because of potential savings in order processing cost, manufacturing setup costs and
transportation costs. By selling fewer but larger orders, the vendor generates lower sales cost. Also,
vendor may save by seeking quantity discounts on raw materials he receives from his supplier. The
increase size of order quantity or lot size ultimately helps in improving the channel profits. Increase in
profits should be shared in some equitable fashion so that coordination in real sense is useful and
parties in the channel shows interest to coordinate.
Here, we have categorised the various coordination models as follows

73
 (i) One can maximize the supplier’s yearly net profit as shown by equation (3) in our general
model by adopting different lot size by giving incentive to the buyer. The authors who
have attempted the coordination problem from this perspective are classified here as
vendor’s /manufacture’s perspective coordination models.
(ii) Similarly, one can minimize the total system cost with respect to coordinated lot size or
the order quantity as shown by equation (4) and thereby improves the system savings.
We have classified here those models as joint buyer and seller / manufacturer perspective
coordination models.
(iii) On the otherhand, some authors have studied the buyer vendor coordination through
quantity discount as a non cooperative and cooperative game. In a non cooperative
game, each member will try to maximize his profit or minimize his cost. Thus the
objective will be here to maximize equation (3) and minimize equation (1) of the general
model. However, in a cooperative game, the objective will be to maximize system profit
subject to the constraint that no player looses or incurs more from their non cooperative
solution. We have categorized these models as a buyer and a seller/ manufacturer
coordination models under game theoretic frame work.
In this stream of literature, most of the models have assumed that seller /manufacturer knows or can
estimate the buyer’s setup and holding costs. Further, EOQ assumptions are considered for the buyer.
The buyer is assumed to act optimally and order the quantity leading to his lowest total cost.

7.5 (a) Manufacturer’s / Seller’s perspective coordination models

Monahan (1984) in his model suggested that a vendor could encourage his customer to increase the
order quantities from EOQ by offering a price discount. With the quantity discount, the buyer will be
motivated to increase the order size up to KQ* where K is a factor by which the vendor entices the
buyer’s order size The amount of discount offered by the vendor compensates buyer’s increased in
inventory costs. For the increased order size, total cost of the buyer is given as

⎛ (K − 1)2 ⎞
( )
TC KQ = P1D + 2 DS1r1P0 ⎜⎜1 +
*
⎟
⎟ (5)
⎝ 2 K ⎠
The increase in cost resulting from larger order size is the difference between the costs at the
EOQ and costs at the order size KO* as given by equation (5). The vendor offers a price discount per
2 S1r1 P0 ⎧ (K − 1) ⎫
2
unit equal to the increase in cost at buyer’s side, which is given as dk = ⎨ ⎬
D ⎩ 2K ⎭
(6)
Supplier’s yearly net profit after giving discount amount is given as follows

YNP2 = D(M 2 P0 − d k ) −
D
S2 (7)
KQ *
Substituting the value of dk in equation (7), maximize the supplier’s profit equation YNP2 with respect
S2
to K. The optimal value of K is obtained as K = +!
*
(8)
S1

From the expression of K * in equation (8), one can easily say that when the value of S2 is large, the
supplier can entice the buyer to order in larger quantity and the value of K* is independent of the
amount of discount offered by the supplier. One important issue here is that when buyer is exactly
compensated for increase in cost due to larger order size, buyer will be indifferent towards increasing

74
his order quantity. Monahan developed the model considering lot-for-lot policy, an all unit quantity
discount schedule with single price break
(b) Joint buyer and seller / manufacturer coordination models
Some authors have used quantity discount as a coordination mechanism to maximize the joint
profit of the buyer and the vendor. The objective function here in all likelihood is to minimize the
total channel cost as shown by equation (4). The models here provide some explicit mechanism for
division of surplus generated in the channel due to coordination. Like the seller’s perspective model,
here also it is assumed that seller’s have full information about buyer’s cost structure.
The idea of joint optimization for buyer and vendor was initiated by Goyal (1976) and later reinforced
by Banerjee (1986a). The objective of Goyal’s model was to minimize total relevant cost for both the
vendor and the buyer for the order quantity Q. He assumed that manufacturer does not produce the
item and in fact purchases it from another supplier. Moreover, he assumed that inventory holding
costs are independent of the price of the item
Banerjee (1986a) formulated a joint economic lot size (JELS) model for a buyer and a vendor system
where the vendor has a finite production rate. He determines the JELS Q* by differentiating the total
system cost equation with respect to Q.

⎛ ⎞
TC (Q ) = (S1 + S2 ) + Q r ⎜⎜ P0 + D C2 ⎟⎟
D
(11)
Q 2 ⎝ R2 ⎠

2 D(S1 + S 2 )
Q* = (12)
⎛ D ⎞
r ⎜⎜ P0 + C2 ⎟⎟
⎝ R2 ⎠
The assumption they consider is that a production setup is incurred every time when an order is
placed. He finds that without quantity discount, the buyer incurs loss, but the supplier gets benefit if
JELS is adopted rather than buyer’s EOQ. He developed the two bounds of discounts that allow the
joint benefit to both the parties if the buyer increases the order quantity from EOQ to the JELS
quantity. When discount amount is fixed at lower bound, all the benefits go to the supplier and the
buyer is indifferent where as when amount of discount is set at maximum level, all benefits shift to the
buyer and the supplier is indifferent. While suggesting equal distribution of the gains from Joint
Economic Ordering, Banerjee (1986a) mentioned that question of pricing and lot-sizing decisions are
settled through negotiations between the buyer and the seller. Later on, we will see how some authors
have incorporated in their model the bargaining power of the channel members in fixing the order
quantity and amount of discounts.
Viswanathan (1998) in his paper has compared two supply policies for an integrated vendor buyer
inventory model. In first policy, the vendor produces a batch and supply to the buyer in number of
equal shipment size at constant interval. The second policy is to supply the production batch to the
buyer in increasing shipment size. He identified problem parameters under which the equal shipment
size policy and increasing shipment size policy is optimal. The author has observed that neither of the
two policies dominates the other for all problem parameters. The second policy attempts to shift
inventory to the buyer as quickly as possible. This type of strategy works better if the holding cost for
the buyer is not much higher than that for the vendor.
Three level coordination models
Munson and Rosenblatt (2001) have extended the two level supply chain to a three level supply chain
by considering a supplier (who is supplying raw materials to manufacturer), a manufacturer and a
retailer and they explored the benefit of using quantity discount on both ends of the supply chain to
decrease cost. Like the earlier scenario, manufacturer’s production lot size is an integer multiple of the
buyer’s order quantity and the manufacturer orders an integer multiple of his production lot size to the

75
raw materials supplier. They have shown that by quantity discount mechanism; company can
coordinate its purchasing and production functions. This creates an integrated plan that dictates order
and production quantities throughout a three firm channel. They have considered manufacture as the
dominant member in the channel who takes the lead role in coordinating the channel.
Yang and Wee (2001) in their paper have also considered integration of producer, distributors, and
retailers a three-stage supply chain. They have developed an economic ordering policy under constant
demand for the arborescent (i.e. a tree like) inventory model structure. They have shown that the
integrated approach results in a significant cost reduction compared to that of independent decision
making by each individual entity of the supply chain. The model however, has not considered how the
increase in cost at retailer level is to be compensated due to implementation of the integrated policy.
Khouja (2003a) has also considered three stage supply chain of tree like inventory model structure.
He has considered three coordination mechanisms between the members of the supply chain and has
shown that some of the coordination mechanisms can lead to significant reduction in total cost. The
author however, has not considered the distribution of savings between the different members of the
supply chain.
Khouja (2003b) also studied coordination of the entire supply chain from raw materials to customer
considering single and multiple components. They consider components scheduling decisions at each
stage in which manufacturing occurs and its impact on the holding cost. They have shown that
complete synchronization in the chain leads to loss of some members of the supply chain. They
provide an algorithm for optimal synchronization of supply chain and incentive alignment along the
supply chain.

(c ) A buyer and a seller/ manufacturer coordination models under game theoretic framework
Some authors have viewed the buyer vendor coordination problem through quantity discount
mechanism as a two-person game. They can be formulated as non-zero sum game having elements of
both conflict and cooperation. In a non-cooperative game playing independently, the intention of the
players (vendor and the buyer) is to maximize their individual gain. The objective function for this
game from the general model can be written as
DS1 Q
Minimize TC = P0 D + + r1P0
Q 2
DS 2
Maximize YNP2 = DM 2 P0 −
Q
Generally, the solution to the non-cooperative game can be obtained by using established equilibrium
concept. Different types of game models have different solution concept. In the Stackelberg game,
the player who holds more powerful position is called the leader and enforces his strategy on the other
and the other player who reacts to the leader decision is called the follower. The solution obtained to
this game is the Stackelberg equilibrium solution.
On the other hand in a cooperative game both buyer and seller would consider maximizing system
profit subject to buyer’s total annual cost at cooperation should be less than or at most equal to those
at non-cooperation. Similarly, seller’s total annual profit at cooperation should be greater than or at
least equal to those at non-cooperation. The objective function for this game from the general model
can be written as
Max λYNP2 − (1 − λ )TC
Subject to TC ≤ TC*
YNP2 ≥ YNP2*

76
Where TC* and YNP2* represents the cost and profit of buyer and seller before cooperation.
Depending upon the bargaining power of the seller and the buyer, the value of λ varies between 0
and 1. In the cooperative game a group of strategies is called a pareto efficient point when at least one
player will be better off and no player will be worse off from the initial condition. In the decentralized
supply chain where the members belong to two different firms, the method of bargaining and
negotiated solution which is dynamic in nature may result better coordination in the supply chain as
compared to static coordinated solution in a centralized supply chain.

7.6 Future directions of research

From the study of the above models, it is seen that this stream of literature describes the supply chain
in a highly aggregated level and often considers only two decision makers. The important insight
provided by the above literature is that there is an increase in profit for the manufacturer when the
buyer purchases more than his EOQ. It is based primarily on the facts that (i) Manufacturer’s setup
cost is much higher compared to the buyer’s ordering cost, and (ii) Manufacturer may use a
production cycle which is an integer multiple of ordering cycle of the buyer.
• Excepting a few, majority of the models are developed considering deterministic demand,
zero lead-time, no stockouts. Holding cost of buyer is considered to be independent of
purchase price.
• With a few exception, rest of the models are developed where supplier offers all unit
quantity discount with a single price break point. Further, the manufacturer is assumed to
have two ways of acquiring the item either by outside purchasing or manufacturing the item
subject to specific production capacity.
• Most of the models simplify the purchasing / production system to one product and one
machine.
• Many of the models fail to specify how the incremental savings to the manufacturer can
be passed onto the buyer. Some authors have mentioned about equal splitting of the surplus,
whereas some have suggested splitting the surplus according to their investment. Most of the
models are silent about conflict resolution between the supply chain partners e.g. division of
surplus between buyers and suppliers. Such new problem may call for the use of game
theoretic negotiation model.
• Most of the models assume that a supply chain partner has complete information
(including cost, demand, lead time etc.) about the other partner. This is considered to be
major limitations of these models. In a decentralized supply chain, hardly will be the situation
where complete information will be available with the parties. Coordination under limited
information sharing is an important issue of concern to be studied for the decentralized supply
chain.
• Single vendor multiple buyers’ literature is still in its infancy state. Particularly, how a
supplier should develop a quantity discount schedule when dealing with many buyers with
different demand and cost structure is not known. Thus, mechanism for additional profit
sharing between vendor and multiple heterogeneous buyers is an important issue that needs
investigation.
• In single vendor multiple buyers’ literature very little work is available considering
vendor as a manufacturer producing the items to supply multiple heterogeneous buyers.
Under such situation, how to tackle the discrete vendor inventory depletion into the model is
an area that requires further study.
In conclusion, in this section mainly we have tried to give an exposure on the various research issues
on supply chain coordination in future.

77
References
[1] Bannerjee, A., (Summer) 1986a, A joint economic lot size model for purchaser and vendor,
Decision Science, Vol. 17, No. 3, pp. 292-311.
[2] Cachon, P.G. and Lariviere, A. M., 2001, Contracting to assure supply: How to share demand
forecasts in a supply chain, Management Science, Vol. 47, No. 5, pp. 629-646.
[3] Corbett, C. J., and Tang, C. S., 1999, Designing supply contracts: contract type and information
asymmetry, In: S. Tayur, M. Magazine, R. Ganeshan, (Eds.), Quantitative models for supply
chain management, Published by Kluwer Academic Publishers, 1999, pp. 269-297.
[4] Goyal, S. K., 1976, An integrated inventory model for a single supplier single customer problem,
International Journal of Production Research, Vol. 15, No. 1, pp. 107-111.
[5] Ha, A., 2001, Supplier buyer contracting: Asymmetric cost information and cut off level policy
for buyer participation, Naval Research Logistics, Vol. 48, No. 1, pp. 41-64.
[6] Khouja, Moutaz., 2003, Optimizing inventory decisions in a multistage multi customer supply
chain, Transportation Research part E, Vol. 39, No. 3, pp. 193-208.
[7] Munson, L.C. and Rosenblatt, J.M., 2001, Coordinating a three level supply chain with quantity
discounts, IIE Transactions, Vol. 33, No. 4, pp. 371-384.
[8] Sarmah, S P., Acharya, D., Goyal, S.K., 2006. Buyer vendor cooridnation models in supply chain
management: An invited review, Europoen Journal of Operational Research, Vol.175, pp. 1-15
[9] TSay, A., Nahmias,S., & Agarwal,N., 1999, Modeling supply chain contracts: A review, In: S.
Tayur, M. Magazine, R. Ganeshan, (Eds.), Quantitative models for supply chain management,
Published by Kluwer academic publishers, 1999, pp. 301-336
[10] Viswanathan, S., 2001, Coordinating supply chain inventories through common
replenishment epoch, European Journal of Operational Research, Vol. 129, No. 2, pp. 277-286.
[11] Viswanathan, S., 1998, Optimal strategy for the integrated vendor buyer inventory model,
European Journal of Operational Research, Vol. 105, No. 1, pp. 38-42.
[12] Yang, C.P. and Wee, M.H., 2001, An arborescent inventory model in a supply chain system,
Production Planning and Control, Vol. 12, No. 8, pp. 728-735.

*********

78
 CHAPTER VIII
A Method for Supply Base Rationalization Considering Supply Risk *

8.1 Introduction
Collaborative sourcing (alternatively named as “partnership sourcing”) has been widely proposed in
the literature [1,2,3] to foster a long-term collaboration between a buyer and its suppliers based on
trust and cooperation, with the buyer relying on a single or a small number of preferred suppliers for
sourcing a product. It has merits over adversarial competition because of its lower operational costs
[1] arising out of fewer dedicated suppliers and because risks and rewards are shared between them. A
prerequisite for developing a strong buyer-supplier relationship is to have a small and rational supplier
base. However, it is a very tedious task for a practicing manager to take its supplier base to a rational
level. The word ‘rationalization’ is most commonly associated with the task. A rational supplier base
leads to: (1) reduced supplier development costs, (2) close and workable supplier relationships, and
(3) business rewards to its suppliers [4]. There has been much confusion with regard to the concepts
underlying the supplier base reduction and supplier base rationalization. Cousins [5] and Dubois [6]
have used the term ‘supplier rationalization’ to principally mean supplier base reduction. Supplier
base reduction presupposes the existence of a large supplier base and is concerned with retaining only
the top performers so as to limit the downsized supplier base to a predetermined size. Supplier base
rationalization, on the other hand, consists of two phases: (1) Determination of the optimum size of
the supplier base and (2) Identification of the suppliers who should constitute this base.

Supplier base rationalization may result in an expanded or contracted supplier base depending on the
number of existing suppliers vis-à-vis the optimal size of the supplier base. Industrial firms in many
developing countries still follow the traditional purchase management practices and have large
supplier bases, especially for MRO items. For these organizations, the problem of identification of the
constituents of the supplier base reduces to a problem of reducing the supplier base to a rational level.
Supplier base rationalization may be viewed as a one-time selection of a small group of suppliers so
as to reduce transactional costs and purchasing complexity and build long-term buyer-supplier
relationships. However, no distinguishing approach for supplier base rationalization has been
forwarded in the purchasing literature. The process of supplier base rationalization is strategic in
nature, with suppliers being evaluated based on factors that represent their short- and long-term
characteristics. Many factors, generally considered in the literature for supplier evaluation, are
qualitative in nature. These factors can be measured, at best, subjectively, and are therefore tend to be
imprecise. Thus, the process of supplier base rationalization can be thought of as a process that
consists of the following three steps:
i. Determination of optimal size of supplier base,
ii. Selection of a method to be used for evaluation of suppliers, and
iii. Identification of the constituents of the supplier base.
Whereas there are large volumes of literature with regard to supplier evaluation methods, the literature
on the first and the last issues are rare. Therefore, in this paper we have addressed the first and third
issues only and they have been presented one after the other in the following sections.

*
Contributed by Ashutosh Sarkar
Lecturer, Department of Mechanical Engineering
Institute of Technology, Banaras Hindu University, Varanasi 221005
8.2. Determination of optimal size of supplier base
Majority of literature evaluates the effect of the size of the supplier base on the inventory and the
replenishment lead time [7, 8, 9]. Further, these studies take the number of suppliers as input to study
its effect and do not advance any method for finding the optimal supplier base. Studies that
specifically deal with the problem of determining the optimal size of the supplier base are due to
Agrawal and Nahmias [10], Weber et al. [11], Berger et al. [12], and Kauffman and Leszczyc [13].
Agrawal and Nahmias [10] formulated a profit maximization problem to determine the optimal lot
size and optimal number of suppliers. The model assumes that having more number of suppliers
reduces the yield uncertainty but increases the fixed cost associated with operating multiple suppliers.
The model trades-off the cost of yield with the fixed cost to determine the optimal size of supplier
base. Kauffman and Leszczyc [13] used the concept of buyer utility and decision-related costs to
derive the optimal choice set size for one-time- and repeat-purchase situations. They have also used
the data on actual bid prices and cost data from the industrial steel pipe market to empirically arrive at
the optimal size of the choice set. Weber et al. [11] used both multi-objective programming and data
envelopment analysis techniques to solve the problem. Considering that the biggest motivation for
having multiple suppliers is to prevent complete disruption of supplies due to an unforeseen natural
disaster (like earthquake, cyclone, tsunami, and flood) and/or man-made disaster (like power grid
failure, strike, and communal violence), the above two modeling approaches are inadequate to
determine the optimal size of the supplier base.
Berger et al. [12] argued that supply disruption or interruption of the inbound supply network can
obstruct the functionality of the whole chain. In order to determine the optimal size of the supplier
base, they considered supply risks, the risks posed by the occurrence of catastrophic events that lead
to complete disruption of inbound supply network. They classified these events as (1) ‘super-events’
that can affect all suppliers simultaneously and disrupt supplies from all the suppliers, exhibiting total
effect, (2) ‘semi-super-events’ that affect only a subset of suppliers, exhibiting regional effect, and (3)
‘unique-events’ that affect a particular supplier uniquely, exhibiting local effect. The purchasing
environment determines the classification of an event as a super-, semi-super-, or unique-event. For
example, a cyclone in a coastal region may be termed as a super-event if all suppliers of an item are
located in this region and supplies from these suppliers fail. It may be labeled as a semi-super-event if
a few but not all the suppliers fail. Berger et al. [12] considered the probabilities of occurrence of
super- and unique-events and used the decision tree approach to find the financial loss and operating
cost of working with multiple suppliers. Although Berger et al. [12] defined three types of
catastrophic events, they considered only two types of events while developing their model. In this
chapter, we have considered all three types of events to determine the optimal size of suppliers.

8.2.1 Supply Risk

The incidences, like 2001 World Trade Tower event, sudden deluge in Mumbai, tsunami in the Indian
Ocean, and many more, have forced ‘risks to be factored into all business functions and processes’
[14], meaning that risks should be considered while modeling business processes. As inbound supply
affects the supply chain performance, any study on suppliers must consider the various types of risks
associated with it. The occurrence of unforeseen events, mentioned above, may disrupt the inbound
supply of a supply chain and these are required to be considered for any study on inbound supply.
Zsidisin et al. [15] recognized two distinct concepts - probability and impact - with the definition of
inbound supply risk. The first relates to the ‘measure of how often a detrimental event that results in a
loss occurs’ and the second relates to the ‘significance of that loss to the organization’. Any model,
meant for determination of optimal size of supplier base, has to consider supply risks and its impact
on the profitability of the purchasing firm. Considered from this point of view, Kraljic’s [16]
purchasing portfolio approach is of great relevance. He classified all products, which are procured,
into four portfolios, based on supply risk and profit impact. The classification gives sufficient
‘insights relevant to supply costs and risks’ [14]. In this chapter, we define supply risk as ‘the

80
probability that supplies of an item will be affected because of problems at the supplier’s end’ and the
cost of supply disruption as its impact.
We assume a semi-super-event to be location specific. Occurrence of such an event disrupts all
suppliers in a geographical location while it does not affect suppliers in other locations. We also
assume that all suppliers in a location will have more or less similar risks due to the occurrence of a
unique-event, and, so, individual variation of supply risks for the suppliers can be neglected for that
location. Furthermore, a decision based on small variations in the probability of occurrence of a
unique-event for each individual supplier is inappropriate, considering that precise information on all
potential suppliers (both the existing suppliers as well as those who are not) may not be available and
the existing information base needs to be updated over time for use during the actual process of
supply base rationalization. Individual variations are needed be considered during the actual reduction
of the supplier base. Thus, for our case, the supply risk due to a unique-event represents more the
character of the supply market rather than the individual supplier, and it is a function of the number of
suppliers engaged.

8.2.2 Relevant costs

In spite of the fact that multiple sourcing increases the fixed cost, the biggest motivation for having
such a strategy is to avoid any emergency situation that may lead to sudden, complete disruption of
supplies. Mohebbi and Posner [9] argued that multiple sourcing reduces the shortage costs but
increases the ordering costs. An increase in the number of suppliers increases the expenditures for
inviting quotations and loses the advantage of availing price discounts (for bulk buying). Further, the
increase makes the decision more complex and so the time spent by the managers for taking a
decision also goes up. Reducing the number of suppliers, on the other hand, increases the supply risk,
causing shortages. Shortage of an item may affect the production in terms of stoppages, delay,
inefficiency, and quality problems. Thus, two opposing types of cost influence the decision in optimal
size determination of the supplier base.
Kraljic [16] has shown that both supply risk and corresponding impact on profit of the organization
vary with the nature of item being purchased. With the same level of supply risks, the costs or impacts
on the organization will vary from item to item. We recognize that it is the criticality of an item that
defines the resulting impact of a supply disruption on the profitability of the business. The word
‘criticality’ that we refer here has a different meaning from that used by Kraljic. Kraljic used it to
refer to those groups of items whose risks and profit impacts are high. However, we use the word
‘criticality’ in a more general sense to mean a characteristic of the item. The less critical an item, the
less is its profit impact in the case of supply disruption. We refer to the cost associated with the
criticality of the item as cost of criticality and denote it with CS .

8.2.3 The model

We assume that the item is sourced from suppliers at L different locations. The objective is to
determine the number of suppliers at each location, and therefore the total number of suppliers, who
should constitute the supplier base, so that the monetary loss owing to supply disruption due to
occurrence of unforeseen events is minimized.

Let, P* = Probability of occurrence of a super-event causing all suppliers to fail.

Pl** = Probability of a localized semi-super-event causing all suppliers at location l,
(l )
= 1, ..., L , to fail.
ρ jl = Probability of a unique-event causing supplier j, (j )
= 1, ..., J l at location l to fail.

81
 J l = Number of suppliers available at location l .

il = Number of suppliers chosen from location l .

y = Total number of suppliers engaged by a company.

Fig. 1 is a decision tree-like representation named here as the probability tree of the alternative ways
in which supply disruptions can take place. Chance nodes ( Ο ) indicate the occurrence of disjoint and
collectively exhaustive events. A branch, emanating from such a chance node, indicates the
occurrence of an event and is labeled by its probability of occurrence. The node, denoted as a square
( ), indicates the alternative branches to location-specific and supplier-specific events. The node,
denoted with symbol ⊗ , indicates that the events are multiplicative in nature. The symbol is used in
two places in the diagram to show (1) the joint probability of suppliers at all locations failing due to
the simultaneous occurrence of semi-super-events and (2) the joint probability of all suppliers failing
due to the simultaneous occurrence of unique-events. From Fig. 1,
Super event
P*
All suppliers fail
because

P*

K
Location
. Pl** ∏P l
**

(1 − P ) ∏ P
L
k =1
⊗
* **
. l
Semi-super- All suppliers fail because l =1
.
. of Semi-super-events
Suppliers
Supplier
Location l
.
1 − Pl** . Jk
.
. ρ jl ∏ρ jl
1− P * . j =1

. No Semi-super .
Unique ⊗ All suppliers fail because
. event
event of unique events
. j .
. .
. . No Unique event No supplier fails
. .
.
No super
.
. Supplier Jl

Location L

Fig. 1 Probability tree for suppliers

82
various probabilities can be easily derived. We refer P ( . ) to represent the probability of
occurrence of an event.
P (All suppliers at location l fail because of a semi-super-event)

(
= 1 − P* Pl** )
P (All suppliers at location l fail because of either a super-event or a semi-super-event)

[ ] [(
= P* + 1 − P* Pl** ) ]
P (All suppliers at all locations fail because of either a super-event or a semi-super-event)

[ ] [(
= P* + 1 − P* P1** P2** ... Pl** ) ]
Similarly we get, P (All suppliers at all locations fail because of either a super-event or a semi-super-
event or a unique event)

⎡ ⎤ ⎡ ⎧ ⎫⎪⎤ Assuming, ρ1l = ρ 2 l = ....... = ρ J l = ρ l , we can

[ ] ( ) ∏P ( ) ∏ ⎪⎨(1 − P ) ∏ ρ
L L Jl
= P* + ⎢ 1 − P* **
⎥ + ⎢ 1− P
* **
⎬⎥ l
(1)
⎦⎥ ⎢⎣ ⎪⎭⎥
l l jl
⎣⎢ l =1 l =1 ⎪⎩ j =1
⎦
rewrite the equation (1) as

⎡ ⎤ ⎡ ⎤
[ ] ( ) ∏P ( ) ∏ {(1 − P )ρ }⎥
L L (2)
= P* + ⎢ 1 − P* l
**
⎥ + ⎢1− P
*
l
**
l
Jl

⎣⎢ l =1 ⎦⎥ ⎣⎢ l =1 ⎦⎥
The number of units short during the period of supplier failure can be estimated from the information
on demand and the duration of time required for recovering (T ′ − T ) from a catastrophic event failing
the suppliers to supply the products. Number of units short during the period (T ′ − T ) is D (T ′ − T )

(3)
Total shortage cost during the period of supply failure, CT = CS (T ′ − T ) D(T ′ − T )

We refer to CT as the cost of supply disruption. Thus, the total cost of engaging y suppliers, f ( y ) ,
is the sum of cost of operating y suppliers and costs due to shortages when all suppliers fail to make
their supplies.
⎡ ⎤
( )∏P ( ) ∏ {(1 − P ) ρ }⎥
L L
f (y ) = C (y ) + CT ⎢P * + 1 − P * l
**
+ 1 − P* l
**
l
il
(4)
⎢ ⎥
⎣ l =1 l =1
⎦

Where, C ( y ) is the cost of operating y suppliers and y = ∑ i . The problem, thus, is to find the
l =1
l

optimum number of suppliers y that minimizes the sum of the cost of operating the suppliers and the
cost of shortage due to supply disruption. Stated more succinctly, the problem is to
“Find y that minimizes (4).”

83
This problem is hereafter called Supplier Size Problem where y is the decision variable and y > 0 .
If the choice of y suppliers has to be economically justified compared to the choice of y − 1
suppliers, the following condition must be satisfied:

C ( y ) − C (1)
CT
( )(
< 1 − P * 1 − Pl ** ρ ly −1 (1 − ρ l ) ) (5)

We assume that the operating cost, C ( y ) , has two components: (1) a fixed component, a and (2) a
variable component, b , that varies with the size of the supplier base, y . Therefore,
C ( y ) = a + b ( y ) and inequality (5) can be rewritten as

b
CT
( )( )
< 1 − P * 1 − Pl** ρ ly −1 (1 − ρ l )
(6)

From inequality (6) it can be concluded that the existing supplier base could be reduced as long as the
following condition holds:
y > 1 + ln b { [C T (1 − P )(1 − P )(1 − ρ )] } (ln ρ )
*
l
**
l
(7)
Inequality (6) can also be extended to multiple locations and can be rewritten as for the case when
choosing y + 1 suppliers is superior to choosing y suppliers

( ) ∏ (1 − P )ρ (1 − ρ )
u
b (8)
< 1 − P* l
**
l
yl
u
CT l =1

⎛ u ⎞
where, yl ⎜ ∑ yl = y ⎟ represent the number of suppliers chosen from location l ,
⎜ l =1 ⎟
⎝ ⎠
( )
l = 1, 2, ..., u and the ( y + 1) th supplier is chosen from location u .

8.2.4 Solution approach to the supplier size problem

The decision tree approach, when used in the formulation of the problem with semi-super event,
results in unmanageable number of trees. For example, Fig. 2 shows the decision tree for a problem of
three locations with only two suppliers in each location. The decision maker has three choices for
each location, namely (1) no supplier chosen from a location (0S), (2) only one supplier is chosen
(1S), and (3) two suppliers are chosen from that location (2S). Considering that at least one supplier
from a location is to be selected, the total number of decision alternatives that are to be evaluated is 26
(=33-1). The number of decision alternatives grows geometrically as the number of suppliers in a
location increases and as the number of locations increases. We propose an alternative method to
solve the supplier size problem – a method that is both elegant and simple. The proposed solution
method for the supplier size problem avoids evaluation of a large number of non-optimal solutions.
Thus the proposed method is a partially enumerative one. The avoidance of alternatives from
evaluation is justified using the theorems presented in the next section.

84
 0 suppliers
(not permitted)
Location 3 1S 1 supplier
2S
0S 2 suppliers
0S 1 supplier
1S 1S 2 suppliers
Location 2 Location 3
2S
3 suppliers
2S 0S 2 suppliers
1S
0S Location 3 3 suppliers
2S 4 suppliers
0S 1 supplier
1S 2 suppliers
Location 3
2S
0S 3 suppliers
0S 2 suppliers
1S 1S 1S 3 suppliers
Location 3
2S 4 suppliers
2S 0S 3 suppliers
1S 4 suppliers
Location 3
2S 5 suppliers

0S
2S 2 suppliers
1S 3 suppliers
Location 3
2S
0S 4 suppliers
0S 3 suppliers
1S 1S
Location 2 Location 3 4 suppliers
2S 5 suppliers
2S 0S 4 suppliers
1S
Location 3 5 suppliers
2S 6 suppliers
Fig. 2 Decision tree

8.2.4.1 Theorems to Reduce the Feasible Solution Space

It can be observed in Fig. 2 that the supplier size problem is combinatorial in nature and finding an
optimal solution to the problem is computationally cumbersome. The proposed solution method is
based on the following theorems which help to reduce the solution space considerably and make the
solution computationally very simple.

Theorem 1: If selecting y suppliers from a set of u locations is advantageous compared to selecting

y + 1 suppliers from the same set of locations then the former will be more economic compared to
selecting y + 2 suppliers chosen from the same locations.

85
Proof: If the selection of y suppliers from a set of u locations is economically more advantageous
compared to choosing y + 1 suppliers then from inequality (8), we get

( ) ∏ (1 − P )ρ (1 − ρ )
u
b (9)
> 1 − P* l
**
l
yl
u
CT l =1

assuming that y + 1 th supplier is chosen from location u .

Similarly, assuming that both y + 1 th and y + 2 th supplier are chosen from location u , the selection
of y + 2 suppliers is economically more advantageous compared to selecting y + 1 suppliers from
the same set of locations if the following condition holds:

( )∏ (1 − P )ρ (1 − ρ )ρ (10)
u
b
< 1 − P* l
**
l
yl
u u
CT l =1

Since ρ u < 1 , inequality (10) contradicts (9) and thus

( )∏ (1 − P )ρ (1 − ρ )ρ
u
b
</ 1 − P * l
**
l
yl
u u
CT l =1

Thus, if for same the set of locations, selecting y suppliers is more economical than selecting y + 1
suppliers, then it will be also more economical than selecting y + 2 suppliers from the same set of
locations.
Theorem 2: If y suppliers are to be chosen from a set of L locations, with at least one from each
location, then it is always economically advantageous to choose as many suppliers as possible from
the location that has the minimum of unique-event probabilities.

Proof: Let us consider that y suppliers are to be chosen from u locations. Then the total cost
function f ( y ) is given as

f ( y ) = C ( y ) + C T [{ P * + P1** .... Pu** + (1 − P * )(1 − P1** )...(1 − Pu** ) ρ 1y1 .... ρ uyu }]

where, y1 , y2 , …, yu represent the number of suppliers chosen from the location 1, 2, ... , u ,
respectively.

Instead of selecting a supplier from location 1, if we select a supplier from location u , then the total
cost
f ′( y ) is given as
[{ ( )( )( )
f ′( y ) = c ( y ) + CT P * + P1** .... Pu** + 1 − P * 1 − P1** ... 1 − Pu** ρ1y1 −1 .... ρ uy u +1 }]
If the latter alternative has to be more advantageous than the former, then the following condition has
to be satisfied.
f ( y ) − f ′( y ) > 0

86
 ⎧ ρ ⎫
or, (1 − P )(1 − P )...(1 − P )ρ
* ** ** y1
....ρ uyu ⎨1 − u ⎬ > 0
⎩ ρ1 ⎭
1 u 1

ρu
or, <1
ρ1
The above condition will be violated only if ρ u > ρ 1 .

While searching for solutions to the supplier size problem two questions has to be answered. First,
what should be the total number of suppliers to be engaged and second, how these chosen suppliers
will be distributed among the various locations. The above two theorem helps in searching the
solution to the supplier size problem efficiently. Theorem 1 restricts the engagement of more number
of suppliers from a set of locations once the total cost f ( y ) starts increasing, thus, limits evaluations
for higher values of y . The number of ways a given number of suppliers y may be chosen from a
set a locations is large. Theorem 2 allows us to evaluate only the combination that selects as much
suppliers as possible from the location for which ρ l is least. Thus theorem 1 and 2 helps in reducing
the solution space for the supplier size problem considerably.

8.2.4.2 A Tabular Method for Solving the Supplier Size Problem

We follow a tabular method of finding an optimal solution to the above-mentioned problem. The
method is much like the tabular method often used when a dynamic programming solution approach
is adopted. The problem of selecting y suppliers from L locations is combinatorial:
• All suppliers may be chosen from a single location. The number of ways the location can be
selected is LC1 .
• All suppliers may be chosen from any two locations. The number of ways two locations can
be selected is L C 2 .
And so on.
Thus, if we have to choose y suppliers from L locations, then we have q′ different combinations of
locations from where these y suppliers can be chosen where,

q′ = L C1 + LC2 + .......+ LC L
Thus, the number of combinations of v′ suppliers chosen from L locations is L
C v . When v′ is
fixed, there are a number of ways in which suppliers can be selected from these v′ locations. The
proposed tabular method evaluates each alternative for each combination of locations separately and
helps find the optimal number of suppliers to be chosen from each location that minimizes the total
cost, f ( y ) . For example, given five locations if we have to evaluate for various alternative values of
y , there will be five separate tables. The first table evaluates the various alternative values of y for
each location, the second table for all possible combinations of two locations, the third for three
locations, and so on. The minimum possible value of y in a table will be equal to v′ , the number of
locations considered in that table, i.e., at least one supplier will be chosen from each location for a
combination. For a given value of y , all possible alternatives for the number of suppliers that can be
chosen from each location are considered for evaluation. Table 1 shows an example of such tables
when v ′ = 4 . The first column in the table represents the various alternative values of y , the

87
variables y1, y2 , y3 , y4 represent the number of suppliers chosen from the first, second, third and the
fourth location, respectively. Location 1:2:3:5 signifies that suppliers are to be chosen from location 1,
2, 3 and 5 only for that column. Entries in these columns are to be the total cost.
Table 8.1: Decision alternatives for a 5C4 problem
y y y y y Location Location Location Location Location
1 2 3 4
1:2:3:4 1:2:3:5 1:2:4:5 1:3:4:5 2:3:4:5
4 1 1 1 1
5 2 1 1 1
5 1 2 1 1
5 1 1 2 1
5 1 1 1 2
6 3 1 1 1
6 1 3 1 1
6 1 1 3 1
6 1 1 1 3
6 2 2 1 1
6 2 1 2 1
6 2 1 1 2
6 1 2 2 1
6 1 2 1 2
6 1 1 2 2

Based on Theorem 1 it can be argued that total costs for a column in the table has to be calculated to a
value of y such that f ( y ) > f ( y − 1) for that column. Thus, for a column, there is no need for
calculating the total cost beyond a value of y for which the condition f ( y ) > f ( y − 1) is satisfied.
For a given value of y , the number of ways in which y can be distributed among the locations is
large. However, if we arrange the locations sequentially in increasing order of their unique-event
probabilities, then Theorem 2 allows us to retain only those rows that choose maximum possible
number of suppliers from the first location while discarding all other alternatives for the given value
of y . Thus, if we assume that the unique-event probabilities are related as ρ1 < ρ2 < ... < ρ5 ,
then the table in Table. 1 can be reduced. Table 2 is the reduced table with various non-optimal rows
deleted. It can be observed that the reduced table has twelve rows less than the table shown in Table 1.
This leads to significant saving in computational time. The optimum solution is the minimum of all
best solutions for the individual table.

88
Table 8.2: Reduced table for decision alternatives for a 5C4 problem
y y y y y Location Location Location Location Location
1 2 3 4
1:2:3:4 1:2:3:5 1:2:4:5 1:3:4:5 2:3:4:5
4 1 1 1 1
5 2 1 1 1
6 3 1 1 1

8.2.5 Illustrative example

We take the following values of various parameters for illustrating the application of the method:
D = 2 stock keeping units per day T = 7 days T ′ = 30 days
a = 50 (Rupees) b = 20 (Rupees/supplier)
P = 0.01
*
Super-event probability:

= P2 = P3 = P4 = 0.02
** ** ** **
Semi-super-event probabilities: P 1

Unique-event probabilities: ρ 11
= ρ 21
= ρ 31
= ....... = ρi 1
1
= 0.031

ρ 12
= ρ 22
= ρ 32
= ....... = ρi 2
2
= 0.030

ρ 13
= ρ 23
= ρ 33
= ....... = ρi 3
3
= 0.033

ρ 14
= ρ 24
= ρ 34
= ....... = ρi 4
4
= 0.032

Number of locations: L=4

Cost of criticality, C s
= 33.00 (Rupees/unit time/unit of the item)

Therefore,
Cost of supply disruption (vide Eq. 2.6), C T
= 33.00× 23 × 2 × 23 = 35000 (Rupees)

In the example, there are four locations ( L = 4 ). We arrange these locations in order of their unique-
event probabilities and label them as location A, B, C and D respectively. Table 3 gives the assumed
occurrence probability values for the semi-super and the unique-events. Later, these probability values
have been changed to study their effects.

Table 8.3 Location labels and assumed probabilities

Location A Location B Location C Location D
Semi Super Event 0.02 0.02 0.02 0.02
Unique Event 0.03 0.031 0.032 0.033

89
As there are four locations, the first table will have 4 ( = 4C1 ) columns and all suppliers are to be
chosen from a single location only; the second table will have 6 ( = 4C2 ) columns with all suppliers to
be chosen from any two locations; the third table will have 4 ( = 4C 3 ) columns with all suppliers to be
chosen from any three locations; and the fourth table will have one ( = 4C4 ) column with all suppliers
to be chosen from any four locations. We have also assumed a maximum of three suppliers can be
chosen from each location. The variables y1 , y2 , y3 , and y4 represent the number of suppliers
chosen from the first, second, third and the fourth location, respectively. When all suppliers are
selected from one location (Table 4), the maximum value of y can be 3, because we have assumed
every location to have three suppliers. When two locations are selected (Table 5), a maximum of six
suppliers can be chosen. It is not necessary to compute total costs for each case, however. A good
approach for achieving computational efficiency is to compute the total cost for each location starting
from the lowest value of y . We can stop whenever a higher value of the total cost is obtained.
Thereafter we proceed to the next location and adopt the same procedure. Table 6 and 7 are the
reduced tables for the number of locations considered for selecting suppliers v ′ = 1 and 2,
respectively. The entries in each cell of the tables are the total cost of engaging y suppliers selected
from the corresponding set of locations. The minimum costs for individual Tables are highlighted.
Although there is no reduction in the size of Table 6, the number of rows in Table 7, have reduced to
3 from 15 in Table 5. It may be noted that Table 7 could have been directly obtained without
considering the last 12 rows of Table 5. Table 8 and Table 9 show the total costs of various
alternatives for v ′ = 3 and v ′ = 4 , respectively. The minimum of all the best values from each
table is the solution for the supplier size problem. The minimum occurs in Table 8. The optimal
solution for the above example, therefore, is obtained as y = 3 and with one supplier chosen from
each of the A, B, and C.
Table 8.4: Complete tabular evaluation of all alternatives for v ′ = 1
y Location A Location B Location C Location D
1 2131.710000 2165.667000 2199.624000 2233.581000
2 1163.561300 1165.632677 1167.771968 1169.979173
3 1153.916839 1154.011613 1154.112703 1154.220313

Table 8.5: Complete tabular evaluation of all alternatives for v ′ = 2

y y y2 Location Location Location Location Location Location
1

A:B A:C A:D B:C B:D C:D

2 1 1 484.808409 485.806745 486.805081 486.871637 487.903250 489.001420
3 2 8
1 474.788452 6
474.818402 4
474.848352 1
474.883360 8
474.915340 2
474.984525
3 1 3
2 474.819400 4
474.882295 4
474.947187 8
474.916372 8
474.983427 4
475.019666
4 3 7
1 493.887853 9
493.888752 7
493.889650 4
493.891724 3
493.892715 9
493.895984
4 1 6
3 493.889741 1
493.892713 6
493.895877 2
493.893803 6
493.897073 8
493.898269
4 2 4
2 493.888782 5
493.890668 2
493.892615 9
493.892747 1
493.894826 493.897109
5 4 0
1 513.860835 9
513.860862 6
513.860889 5
513.860983 2
513.861014 3
513.861151
5 1 6
4 513.860922 6
513.861046 5
513.861183 4
513.861081 2
513.861223 5
513.861262
5 3 0
2 513.860863 8
513.860920 9
513.860978 7
513.861015 4
513.861079 9
513.861187
5 2 5
3 513.860892 1
513.860981 5
513.861076 2
513.861047 6
513.861149 5
513.861224
6 5 2
1 533.860025 4
533.860025 3
533.860026 9
533.860030 3
533.860031 6
533.860036
1 9 7 5 4 8
90
6 1 5 533.860028 533.860033 533.860039 533.860034 533.860040 533.860041
6 4 2 6
533.860025 5
533.860027 1
533.860029 6
533.860031 4
533.860033 7
533.860038
6 2 4 9
533.860027 6
533.860031 4
533.860035 5
533.860033 5
533.860037 533.860040
6 3 3 7
533.860026 4
533.860029 5
533.860032 5
533.860032 9
533.860035 4
533.860039
8 4 3 5 6 2

Table 8.6: Reduced tabular evaluation of all alternatives for v ′ = 1

y Location A Location B Location C Location D
1 2131.710000 2165.667000 2199.624000 2233.581000
2 1163.561300 1165.632677 1167.771968 1169.979173
3 1153.916839 1154.011613 1154.112703 1154.220313

Table 8.7: Reduced tabular evaluation of all alternatives for v ′ = 2

y y y2 Location Location Location Location Location B:D Location
1

A:B A:C A:D B:C C:D

2 1 1 484.8084098 485.8067456 486.8050814 486.8716371 487.9032508 489.0014202
3 2 1 474.7884523 474.8184024 474.8483524 474.8833608 474.9153408 474.9845254
4 3 1 493.8878536 493.888752 493.8896506 493.8917242 493.8927156 493.8959848

Table 8.8: Reduced tabular evaluation of all alternatives for v ′ = 3

y y1 y2 y3 Location Location Location Location
A:B:C A:B:D A:C:D B:C:D
3 1 1 1 461.2477421 461.2780716 461.3103578 461.3447963
4 2 1 1 480.3063163 480.3072261 480.3081947 480.3102955

Table 8.9: Complete tabular evaluation of all alternatives for v ′ = 4

y y1 y2 y3 y 4 Location
A:B:C:D
4 1 1 1 1 480.0369313
5 2 1 1 1 500.0064856

8.3. The supply base rationalization problem

8.3.1 Nature of purchase and supplier relationship
Effective and efficient supplier relationship management greatly contributes to the building up of
competitive advantage of an organization. Identification of key buyer-supplier relationships helps an
organization to allocate its resources for building and developing such relationships [17]. Raw
materials and MRO items are usually sourced from a large number of suppliers. Portfolio analysis has
been continuously used to get useful insights into the management of the supplier relationships and
development of possible action plans. Nellore and Soderquist [18] observed that purchasing portfolio
models have three steps in common and they are: (1) Analysis of the products and classification, (2)
Analysis of the supplier relationships required, and (3) Action plan to match requirements. Based on
profit impact and supply risk Kraljic [16] classified purchases into routine, bottleneck, leverage and
strategic purchase. Extending this work, Olsen and Ellram [19] suggested a classification method
based on (1) the strategic importance of the purchase and (2) the difficulty in managing the purchase.
Strategic importance of the purchase can be measured on the basis of internal factors such as its

91
contribution to develop core competencies and boost the buyer’s image among customers and
suppliers. Difficulty of managing a purchase situation can be assessed on the basis of external factors
such as the nature of the product, the characteristics of the supplier market, and the risk and
uncertainty associated with a supply. The nature of the purchase influences many purchasing
decisions like the size of supplier base, the extent of resources to commit to supplier development and
other long-term involvement with suppliers. Fig. 3 lists various categories of purchases and their
features and possible action plans. The buyer-supplier relationships for each category of purchase will
be different from the others. As a strategy, long-term relationships are preferred for sourcing of
bottleneck items, and a medium-to-long-term relationship with one or a smaller group of suppliers is
prescribed for strategic items. For other types of items, the relationship is of very short duration for
which procurements can be done in a more traditional manner. For bottleneck items, there are very
few suppliers available in the market, and so a supplier reduction strategy may not be desirable in this
case. However, a common approach can be adopted for identifying the constituents of the choice set.
Supplier base rationalization process should be done through a careful evaluation of suppliers
considering both their short-term performances and long-term capabilities.

Purchase Category Features

Routine Items ¾ More number of suppliers available
¾ Very short term supplier relationship
¾ Supplier Monitoring
¾ Simplification and automation of purchasing procedure
¾ Delegation of decision making power to lower level of management
Bottleneck Items ¾ Monopolistic supplier market
¾ Long term supplier relationship
¾ Security of inventories
¾ Internally develop alternatives
¾ Contingency planning
¾ Delegation of decision making power to higher level of management
Leverage Items ¾ More number of suppliers available
¾ Short term supplier relationship
¾ Exploitation of full purchasing power
¾ Delegation of decision making power to medium level of management
Strategic Items ¾ Few suppliers are available
¾ Medium/ long term supplier relationship
¾ Detailed evaluation of suppliers
¾ Supplier development efforts
¾ Delegation of decision making power to top level of management
Fig. 3 Characteristics of purchases

92
8.3.2 Performance versus capability of a supplier
Sourcing decisions should be normally based on the consideration of a large number of factors [20].
However, a majority of practitioners focus only on such factors as cost, quality and service, while
neglecting other important factors like technological and financial capabilities, quality systems, etc.
This has not helped in distinguishing suppliers with strong long-term capabilities from those who
excel when measured against short-term criteria. In their study on evaluation and rationalization of
suppliers, Narasimhan et al. [21] have used organization’s capability factors as input resources and
performance factors as output variable in their DEA study. We suggest that the supplier evaluation
criteria can be classified based on their influence on the short-term and the long-term goals of the
supply chain. We define two important dimensions of a supplier’s abilities: performance and
capability. Performance is defined as the demonstrated ability of a supplier to meet a buyer’s short-
term requirements in terms of cost, quality, service, and other short-term criteria. Capability is defined
as the supplier’s potential that can be leveraged to the buyer’s advantages in the long term. Various
criteria identified for supplier selection by Katsikaes et al. [22], Choi and Hartley [23], Swift [24], and
Weber et al. [20] are classified as performance and capability factors and are presented in Table 10.

Table 8.10: Performance and capability factors

Capability Factors Reference Performance Factors Reference
Quality Systems in operation at the 23 Price 20,22,23,24
Supplier’s place/ Quality Philosophy

Financial Capability of the supplier 20,23,24 Quality/Reliability 20, 23,24

of the product
Technological Capability/ R&D Capability 20, 22, 23 Ability to meet delivery 20, 22, 23,
promise/ Delivery lead 24
time/ Consistent Delivery

Reputation for Integrity/ 20, 22, 23, Management sensitivity to 20

24 buyer’s
Believability and Honesty/ Vendor’s Image
requirements/Attitude

Existence of IT standards/ Communication 20, 22 After-Sales Support/ 22, 23, 24

System Technical support
available
Performance Awards/ Performance History 20, 23 Positive attitudes towards 22
complaints

Bidding Procedural Compliance 20

Profitability of Suppliers 23
Breadth of product line/Ability of a supplier 24
to supply a number of items
Supplier’s Proximity/ Geographic Location 20, 24
Management and Organization 20
Contribution to Productivity 24
Conflict Resolution 23

93
Production Facilities and Capacity 20
Communication Openness 23
Labour problems at the supplier’s place 20
Business volume/Amount of Past business 20

It can be seen in Table 10 that while most of the performance factors are quantitative and can be
measured relatively easily, most of the capability factors are qualitative and present measurement
problems.

8.3.3 Supplier evaluation methods

A consensus approach to supplier evaluation is generally followed in practice in view of the
multidimensional and subjective nature of the problem of supplier evaluation. Based on an exhaustive
search of the literature, De Boer et al. [25] have divided the various supplier selection methods into
five categories: (1) Linear Weighting models [26, 27, 28, 29]), (2) Total Cost of Ownership models
[26, 30, 31]), (3) Mathematical Programming models [32, 33, 34, 35, 36, 37], (4) Statistical models
[38, 39], and (5) Artificial Intelligence (AI) based models [40]. The problem of supplier base
rationalization belongs to the pre-qualification phase, rather than to the choice phase. De Boer et al.
have identified a number of models that are particularly suitable for pre-qualification of suppliers,
notable among them being case-based reasoning [41, 42], cluster analysis [43, 44], and data
envelopment analysis [21]. Sarkar and Mohapatra [45] highlighted the limitations of the above
methods and recommended the use of fuzzy set theoretic analysis for evaluation of suppliers.

8.3.4 The proposed method

A methodology for supplier base rationalization has been proposed (Fig. 4) following the framework
of supplier selection proposed by De Boer et al. [25]. The process starts with an analysis of the nature
of purchase to identify the type of relationship that is desired. An analysis of the supplier market is
also necessary to perceive the risk and set a target size of the supplier base. The type of relationship
that is desired also determines the supplier characteristics that are to be considered for evaluation in
order to find a best-fit supplier. As has been discussed extensively in Section 3.2, the supplier base
rationalization process should group these supplier characteristics (factors) into long- and short term
factors. Vokurka et al. [40] and Hong et al. [44] argued that the supplier selection problem should
also include suppliers that are not in the existing supplier base. Identifying potential suppliers requires
a supplier market survey to be conducted. Whereas the organization may have rich experience and
strong information database on known, existing suppliers, it has to obtain information about unknown
suppliers from the Internet [41] and by peer feedback and onsite visit to the suppliers’ facilities [46].
Sometimes the potential suppliers list may be very long. In such cases, screening of the suppliers is
done to reduce the list of potential suppliers. The supplier’s willingness to associate is for sustaining a
relationship and it can be assessed from the amount of incentives the buyer provides to the supplier.
Suppliers will be more committed when the business volume is more. However, for routine items, the
supply risk and the profit impact are very low and the buyer’s interest is only to have a working
relationship. So, we define supplier incentive to do business with the buyer as the total value spent by
the buyer on purchase of items that are/(can be) supplied by the supplier. An initial screening of
suppliers can be done based on the supplier’s incentives to do business with the buyer.
Fig. 5 depicts a matrix where supplier incentive is put against the length of relationship desired by the
buyer. The ‘tick’ and ‘cross’ marks are used to show whether, the incentive will be sufficient to attract
supplier’s willingness. The matrix gives an easy guideline for initial screening of the suppliers.
94
95
As many of the criteria used for supplier evaluation are qualitative in nature, we propose the use of
experts’ opinion-based methods for ranking the suppliers. Experts may be drawn from the users of the
product and from the purchase personnel. Senior management can also be involved if the purchase has
strategic implications. The experts give their subjective assessment of the relative importance of the
factors and evaluate the suppliers against each factor in terms of scores in a predefined scale. Factor
ratings with capability rankings along the horizontal direction and Performance Ratings in the vertical
direction are plotted in a matrix (Fig. 6). We call this the ‘capability-performance matrix’. The
position of a supplier in the matrix denotes the rank the supplier has considering the two factor
categories. It can be used to arrive at an ordered list of suppliers with decreasing preference. Finally,
this order of preference is used to retain the desired numbers of suppliers.

Scores secured by a supplier against the capability and performance ratings define the position of the
supplier in Fig. 6. A diagonal line drawn from the top-left corner to the right-bottom corner divides
the suppliers into three classes: (1) Balanced suppliers, (2) Motivated suppliers, and (3) De-motivated
suppliers. All suppliers on the diagonal line are balanced suppliers (supplier H). They are deemed to
have a performance level which is commensurate with their level of capability. The suppliers lying
below the diagonal line (suppliers E, D, and B) fail to match their performance with their capability.
They are the de-motivated suppliers. They are not sufficiently motivated to leverage on their
96
capability. Either such a supplier is unable to use his capability efficiently or he does not know how to
operate within the framework of the relationship. Suppliers who are above the diagonal line (suppliers
A, C, F, and G) have performed better than their capability. They are the motivated suppliers. They
either have high stake in doing business with the buyer or are committed and are able to efficiently
capitalize on their capability. However, a cross-evaluation of the reasons for over-performance for this
category of suppliers is very necessary. An evaluation of the sustainability of this performance in the
long run should also be carried out. The cost of capability enhancement for them for consistent
performance over a long period should be traded-off against the cost of motivating and improving the
performance of a de-motivated supplier in case of a tie.
In case two suppliers have the same location on the diagonal and the same length of the perpendicular,
ranking is made arbitrarily. In this case the buyer will have to evaluate the cost of supplier
motivational efforts against the cost of capability enhancement. The first option is relatively easier to
implement whereas the latter will demand a lot of resources for both the suppliers and the buyer.
Supplier location on the matrix diagonal as a reference to determine the order of preference is logical
when we give equal importance to both performance and capability ranking. For a supplier the sum of
its rank in the performance factors and the capability factors can also be used in place of using the
intersecting point. In such a case the supplier with a greater sum will be placed ahead of the others in
the order of preference.

8.3.5 Illustrative example

We take a hypothetical case for illustrating the proposed method of supplier base rationalization.
There are ten suppliers and the objective is to reduce the present number of suppliers to two. We
consider four performance factors and ten capability factors given in Table 11 for the evaluation
purpose.

Table 8.11: Performance and capability factors considered for the illustrative example
Performance Factors Capability Factors
Price Quality systems in operation at the supplier’s place
Quality Financial capability of the supplier
Delivery lead time Production facilities and capacity
Attitude Management and organization
Technological capability
Breadth of product line
Supplier’s proximity
Existence of IT standards
Labour problems at the supplier’s place
Reputation
The ranks of the suppliers are given in Table 12.

97
Table 8.12: Rankings of suppliers
Supplier Based on Performance Based on Capability
Rank Factors Factors
01. A2 A10
02. A7 A4
03. A4 A6
04. A10 A9
05. A6 A3
06. A3 A1
07. A8 A7
08. A9 A2
09. A1 A8
10. A5 A5

The individual ranks of suppliers for capability and performance factors are plotted in the capability-
performance matrix and are shown in Fig. 8. The capability-performance matrix shows that supplier
A5 lies on the diagonal and is a balanced supplier; however, it performs the least both on the
capability and the performance factors. Suppliers A7, A2, and A8 are located above the diagonal and
suppliers A1, A3, A4, A6, A9 and A10 are located below the diagonal. The suppliers A7, A2, and A8
are the motivated suppliers and the suppliers A1, A3, A4, A6, A9, and A10 are the de-motivated
suppliers. The locations of suppliers A4 and A10 on the diagonal are same, as those of A2 and A7. On
the basis of the length of the perpendiculars, A4 is ranked ahead of A10 and A7 is ranked ahead of
A2.

98
In order to rank the suppliers based on the preferences for developing long-term relationships we
move from the top-left corner of the capability-performance matrix and sequence the supplier based
on their locations on the diagonal. The order of preference is shown in Table 13.

Table 8.13: Final order of preference for suppliers

Order of preference 01 02 03 04 05 06 07 08 09 10
Supplier A4 A10 A6 A7 A2 A3 A9 A1 A8 A5

As we have to retain only two suppliers for the case, we retain suppliers A4 and A10 and all other
suppliers are removed from the registered supplier list.

8. 4. Conclusions
Supplier base reduction is a prerequisite for developing long-term relationships with suppliers. In this
chapter, we propose a structured method of rationalizing the supplier base. A major contribution of
our work is that the problem has been addressed considering, in addition to the above two types of
supply risks, the possibility of occurrence of a semi-super-events. The decision tree approach, when
all the three types of supply risks are considered, results in an unmanageable number of decision
alternatives. In order to avoid it, we develop a simple, but elegant, method to determine the optimal
size of supplier base. The method uses tables to evaluate all possible decision alternatives. From the
practitioners point of view this tool is very simple and spreadsheets can be very easily used to find the
optimal solution.
One of the important points of the proposed method is the use of the capability-performance matrix,
which increases the visibility about each supplier’s strengths and weaknesses and facilitates a more
rational judgment. The increased visibility through the use of capability-performance matrix helps
classifying the suppliers into ‘motivated’ and ‘de-motivated’ ones. Tracing the causes for a supplier
being ‘motivated’ or ‘de-motivated’ can reveal important information with respect to ‘consistency’ in
the supplier performance. The ‘capability-performance matrix’ also helps easy ranking of the
suppliers with whom a sustainable long-term relationship can be established.
An issue that needs further development is how to develop a mechanism for continuously evaluating
supplier performance and maintaining of a knowledge base of suppliers. While the knowledge about a
new supplier is usually continually updated over time, the issue is how to include such new
information on suppliers in the method. After the supplier base is rationalized, issues related to the
subsequent management of the supplier base have to be addressed. How to develop and build a
sustainable relationship with this reduced supplier base is also an area that needs further development.
Development of such long-term relationship requires efforts and resources on the part of the supplier
as well as the buyer. A future area of research can be to include and adapt the dynamics of supplier-
development potential as well as related costs into the proposed method.

References
[1] Parker, D., and Hartley, K., (1997) The Economics of Partnership Sourcing versus Adversarial
Competition: A Critique. European Journal of Purchasing & Purchasing Management, 3(2), 115-125.
[2] Mudambi, R. and Schrunder, C.P., (1996) Progress towards Buyer-Supplier Partnerships:
Evidence from Small and Medium-sized Manufacturing Firms. European Journal of Purchasing and
Supply Management, 2(2/3), 119-127.

99
[3] Bechtel, C., & Patterson, J.L., (1997). MRO Partnerships: A Case Study. International Journal of
Purchasing and Materials Management, 33(3), 18-23.
[4] Dowlatshahi, S. (2000). Designer-Buyer-Supplier Interface: Theory versus Practice. International
Journal of Production Economics, 63, 111-130.
[5] Cousins, P.D. (1999) Supply Base Rationalization: Myth or Reality?. European Journal of
Purchasing and Supply Management, 5, 143-155.
[6] Dubois, A., (2003). Strategic Cost Management across Boundaries of Firms. Industrial Marketing
Management, 32, 365-374.
[7] Ramasesh, R.V., Ord, J.K., Hayya, J.C. and Pan, A.C., (1991) Sole versus Dual Sourcing in
Stochastic Lead Time (s, Q) Inventory Models. Management Science, 37(4), 428-443.
[8] Chiang, C. and Benton, W.C., (1994) Sole sourcing versus dual sourcing under stochastic
demands and lead times. Naval Research Logistics, 41, 609-624.
[9] Mohebbi, E. and Posner, M.J.M., (1998) Sole versus Dual Sourcing in a Continuous Review
Inventory System with Lost Sales. Computers and Industrial Engineering, 34(2), 321-336.
[10] Agrawal, N. and Nahmias, S., (1997) Rationalization of the supplier base in the presence of yield
uncertainty. Production and Operations Management, 6(3), 291-308.
[11] Weber, C.A., Current, J. and Desai, A., (2000) An Optimization Approach to Determining the
number of Vendors to Employ. Supply Chain Management: An International Journal, 5(2), 90-98.
[12] Berger, P. D., Gerstenfeld, A., & Zeng A.Z., (2004) How Many Suppliers are best? A Decision-
Analysis Approach. Omega, 32, 9-15.
[13] Kauffman, R.G., and Leszczyc P.T.L. P., (2005). An Optimization Approach to Business Buyer
Choice Sets: How Many Suppliers Should be Included?. Industrial Marketing Management, 34, 3-12.
[14] Cavinato, J.L. (2004) Supply chain logistics risks: From the back of the room to the board room.
International Journal of Physical Distribution & Logistics Management, 34(5), 383-387.
[15] Zsidisin, G.A., Ellram, L.M., Carter, J.R., and Cavinato, J.L., (2004) An Analysis of Supply Risk
Assessment Techniques. International Journal of Physical Distribution & Logistics Management,
34(5), 397-413.
[16] Kraljic, P., (1983) Purchasing must become supply management. Harvard Business Review,
61(5), 109-117.
[17] Zolkiewski, J., and Turnbull, P., (2002) Do Relationship Portfolios and Networks Provide the
Key to Successful Relationship Management?. Journal of Business & Industrial Marketing, 17(7),
575-597.
[18] Nellore, R., & Soderquist, K., (2000) Portfolio Approaches to Procurement: Analysing the
Missing Link to Specifications. Long Range Planning, 33, 245-267.
[19] Olsen, R.F., & Ellram, L.M., (1997) A Portfolio Approach to Supplier Relationships. Industrial
Marketing Management, 26, 101-113.
[20] Weber, C. A., Current, J. R. and Benton, W. C. (1991) Vendor Selection Criteria and Methods,
European Journal of Operational Research, 50, 2-18.
[21] Narasimhan, R., Srinivas, T. and Mendez, D., (2001 summer) Supplier Evaluation and
Rationalization via Data Envelopment Analysis: An Empirical Examination. The Journal of Supply
Chain Management: A Global Review of Purchasing and Supply, 37(3), 28-37.
[22] Katsikeas, C.S., Paparoidamis, N.G., and Katsikea, E., (2004) Supplier Source Selection Criteria:
The Impact of Supplier Performance on Distributor Performance. Industrial Marketing Management,
33, 755-764.
100
[23] Choi, T.Y., & Hartley, J.L., (1996) An Exploration of Supplier Selection Practices across Supply
Chain. Journal of Operations Management, 14, 333-343.
[24] Swift, C.O., (1995) Preferences for Single Sourcing and Supplier Selection Criteria. Journal of
Business Research, 32, 105-111.
[25] De Boer, L., Labro, E. and Morlacchi, P., (2001) A review of Methods Supporting Supplier
Evaluation, European Journal of Purchasing and Supply Management, 7, 75-89.
[26] Timmerman, E., (1986) An Approach to Vendor Performance Evaluation. Journal of Purchasing
and Supply Management, 1, 27-32.
[27] De Boer, L, Van der Wegen, L. and Telgen, J., (1998) Outranking Methods in Support of
Supplier Evaluation. European Journal of Purchasing and Supply Management, 4(2/3), 109-118.
[28] Narasimhan, R., (1983) An Analytic Approach to Supplier Selection. Journal of Purchasing and
Supply Management, 1, 27-32.
[29] Petroni, A. and Braglia, M., (2000) Vendor Selection Using Principal Component Analysis. The
Journal of Supply Chain Management: A Global Review of Purchasing and Supply, 36(2), 63-69.
[30] Monczka, R. M. and Trecha, S. J., (1988) Cost-based Supplier Performance Evaluation. Journal
Purchasing and Materials Management, 24 (Spring), 2-7.
[31] Smytka, D. L. and Clemens, M. W., (1993) Total Cost of Supplier Selection Model: A Case
Study. International Journal of Purchasing and Materials Management, 29, 42-49.
[32] Weber, C. A. and Desai, A., (1996) Determination of paths to vendor market efficiency using
parallel co-ordinates representation: a negotiation tool for buyers. European Journal of Operational
Research, 90, 142-155.
[33] Weber, C. A., Current, J. R., and Desai, A. (1998) Non Co-operative Negotiation Strategies for
Vendor Evaluation. European Journal of Operational Research, 108, 208-223.
[34] Karpak, B., Kumcu, E., and Kasuganti, R., (1999) An Application of Visual Interactive Goal
Programming: A Case in Vendor Decisions. Journal of Multi Criteria Decision Analysis, 8, 93-105.
[35] Degraeve, Z. and Roodhooft, F., (1998) Determining Sourcing Strategies: A Decision Model
Based on Activity and Cost Driver Information, Journal of Operational Research Society, 49(8), 781-
789.
[36] Degraeve, Z. and Roodhooft, F., (1999) Improving the Efficiency of the Purchasing Process
Using Total Cost of Ownership Information: The Case of Heating Electrodes at Cockerill Sambre
S.A. European Journal of Operational Research, 112(1), 42-53.
[37] Degraeve, Z. and Roodhooft, F., (2000) A Mathematical Programming Approach for
Procurement Using Activity Based Costing. Journal of Business Finance and Accounting, 27 (1-2),
69-98.
[38] Ronen, B. and Trietsch, D., (1988) A Decision Support System for Purchasing Management of
Large Projects. Operations Research, 36 (6), 882-890.
[39] Soukup, W. R., (1987) Supplier Selection Strategies. Journal of Purchasing and Materials
Management, 23(3), 7-12.
[40] Vokurka, R. J., Choobineh, J. and Vadi, L. (1996) A Prototype Expert System for Evaluation and
Selection of Potential Suppliers. International Journal of Operations & Production Management, 16
(12), 106-127.
[41] Choy, K.L., Lee, W.B., Lau, H.C.W. and Choy, L.C., (2005) A Knowledge-Based Supplier
Intelligence Retrieval System for Outsourcing Manufacturing. Knowledge-Based Systems, 18, 1-17.

101
[42] Luu, D.T., Ng, S.T., Chen, S.E. & Jefferies, M., (2006). A strategy for evaluating a fuzzy case-
based construction procurement selection system. Advances in Engineering Software, 37, 159-171.
[43] Holt, G.D., (1998) Which contractor selection methodology?. International Journal of Project
Management, 16(3), 153-164.
[44] Hong, G.H., Park, S.C., Jang, D.S. & Rho, H.M., (2005) An Effective Supplier Selection Method
for Constructing a Competitive Supply-Relationship, Expert Systems with Applications, 28, 629-639.
[45] Sarkar, A. and Mohapatra, P.K.J., (2006) Evaluation of Supplier Capability and Performance: A
Method for Supply Base Reduction. Journal of Purchasing and Supply Management, 12, 148-163.
[46] Avery, S., (1999) MRO Report: Supplier Alliances Help Power Wisconsin Electric. Purchasing,
June 3, 62-64.

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

102
 CHAPTER IX
E-procurement

9.1. Introduction

Online procurement (e-procurement) is a technology solution to facilitate corporate buying using the
Internet and other Information and Communication Technologies (ICT). It has been identified as the
most important element of e-business [1]. As specified in the literature and experienced by top
corporate houses the benefits of e-procurement are manifold. These advantages include reducing
administrative costs, shortening the order fulfillment cycle time, lowering inventory levels and the
price paid for goods, and preparing organizations for increased technological collaboration and
planning with business partners[1].
The potential is so great that e-procurement has turned the formerly looked-down-upon traditional
purchasing function into a competitive weapon. One such example is General Electric’s (GE’s)
Trading Process Network (TPN) [2]. Here the buyer posts a request for proposal on the Internet for
access by pre-qualified suppliers. The suppliers download the request and submit bids electronically.
The buyer evaluates the bids, negotiates on line, and places the order with the lowest bidder. The
system also facilitates transaction processing by, for example, automatically reconciling purchase
orders with invoices as part of the payment process. A solution like the TPN impacts both the supplier
selection and contract agreement components of the purchasing process.
Some of the benefits that GE has realized from its e-procurement operations in the initial days are
listed below [2]
• Reductions in labor costs in the purchasing process are one of the reasons that transaction costs
fall so precipitously with e-procurement. For example, in traditional labor-intensive, paper-based
purchasing process in GE the transaction cost. was ranging from $70 to $300 per purchase order.
GE saw those costs drop 30%.
• Material cost reductions in the range of 5% to 20% were realized because GE’s e-procurement
solution helped the firm reach a wider supplier base and identify heretofore unidentified and
qualified sources of supply.
• The system also allows the Company’s purchasing departments around the world to share
information about their best suppliers. GE’s purchasing departments gained 6 to 8 days per month
to work on more strategic initiatives.
• e-procurement systems enable firms to more efficiently and accurately capture and aggregate how
much they are spending corporate-wide in various purchased product areas, allowing the firm to
bring what may be significant buying power leverage to market. This benefit contributes to the
5% to 20% material cost reductions that GE has experienced.
Adoption of technology solution requires reengineering of the traditional purchasing process. We
describe below the traditional purchasing process and compare with a typical reengineered process.
9.2. Traditional purchasing process
Fig. 1 provides an overview of a typical purchasing process. It begins with the need to define buying
requirements based on the demands of the firm’s final customer. At this stage, specifications are
developed. The step involves early purchasing involvement (EPI) and early supplier involvement
(ESI), as well as inputs of a cross-functional buying team that may include, in addition to supply and
engineering, representatives from operations and marketing. Once the specifications have been
developed, a buying team led by the supply manager will pre-qualify suppliers, generate requests for
proposals, evaluate the proposals, and select a supplier based on established selection criteria.
Contract negotiations result in the terms and conditions of a formal contract. Ordering routines and
transaction-processing guidelines are established for all purchases that take place under the umbrella
of the negotiated contract. Closing the loop is a supplier evaluation system that assesses supplier
performance that provides information to be used as the basis for rating the supplier (e.g., excellent,
good, fair, unacceptable).
.

Define
Requirement Select Contract Supplier
Supplier Agreement Evaluation

• Specification • Pre-qualification • Negotiate • Assessing supplier

Development • RFP • Formalize Contract performance
• EPI/ESI • Select Supplier • Establish Ordering • Supplier Rating
• Cross- Routines/ Transaction
Functional Processing
Teams
Fig. 1 The traditional purchasing process.
(Adapted from A.J. Van Weele, Purchasing Management 1994)

9.3. Purchasing process reengineered using ICT

The basic activities in the purchasing process are not altered in an organization using ICT tools and
techniques. Rather they are reengineered for the convenience of automation leading to value creation.
Figure 2 (adopted from [3]) provides an overview of the main activities in sourcing and procurement
in a typical e-procurement setup.

Spend
Analysis Sourcing Procurement Settlement

• Data • Catalog buying / • Requisitioning • Purchase ordering

aggregation Contract negotiation • Approval workflow • Invoicing
• Sourcing • RFx, • Supplier enablement • Payables
strategy • Supplier selection • Catalog Management • Receivables
Vendor Management

Fig. 2. Overview of a Purchasing Department Activities Reengineered using ICT

i. Spend Analysis: The focus of this activity is to develop an aggregate view of the procurement
spend across the organization using the transaction data. The aggregate spend by commodity,
supplier, plant etc provides a basis for identifying cost saving strategies. A typical example is to find
commodity classes or plants where reducing suppliers and increasing volume to a small number of
(preferred) suppliers might allow for better price negotiations. Another piece of analysis is to track the
performance of each supplier based on past behavior. This is a strategic activity.

104
ii. Sourcing: One of the fundamental aspects of sourcing is supplier selection (for a commodity class
identified by spend analysis) using one of many negotiation techniques (such as RFx, auctions etc).
Once the suppliers are selected the relationship with the selected suppliers is then managed through
the negotiated contracts. This step operationalizes the strategy developed by spend analysis.
iii. Procurement: This is a tactical activity where purchasing is (ideally) performed within the
umbrella of existing contracts. Typical purchasing within an enterprise starts with a requisition that is
approved and purchased from within catalogs of selected suppliers. An additional activity that is
supported at this level is the enablement of (new) supplier catalogs and the management of these
catalogs.
iv. Settlement: This is the follow through activity where the purchase is ordered, invoiced etc. This is
the routine of bookkeeping of each purchase and will not be elaborately discussed.
Comparison of Fig 1 and Fig. 2 reveals the following facts:
• The basic activities for developing product specification is an unstructured decision making activity
involving human intelligence, awareness of the recent trends and technologies and other external
factors. Thus, they are not the candidates for automation. Therefore, absence of these activities in
Fig. 2 is not surprising.
• Integration of the information systems across geographically dispersed units of an organization,
developments in data warehousing and mining technologies has enabled the managements to take
strategic decisions on spend management. Therefore, these activities do not appear in Fig. 1.
• Spending in purchasing is directly affected by the suppliers’ performance. Therefore, this strategic
activity is integrated with spend management (Fig. 2).
• Other strategic decision making activities which are automated are clubbed together under the
umbrella of sourcing in Fig 2. These activities were otherwise spread under supplier selection and
contract negotiation steps of the traditional process.
• All the activities involving ICT in tactical decision making, such as organization-wide workflow,
connecting to suppliers’ catalog systems are under the umbrella of procurement (Fig 2.). This step is
not present in the traditional purchasing process.
• The transaction activities which do not involve any decision making tools are clubbed together as
settlement activity (Fig 2). In Fig. 1, these activities were considered as a part of contract
negotiation.
We describe below the ICT tools and techniques used for different steps of the reengineered
purchasing process [3].
9.4. Spend Analysis
This term is a general umbrella term used to capture various strategic activities that are important for
designing a sourcing strategy for the corporation. Following steps are involved in spend analysis.
9.4.1 Data Warehouse for Spend Analysis
As evident from GE’s case e-procurement systems enable firms to more efficiently and accurately
capture and aggregate how much they are spending corporate-wide in various purchased product
areas, allowing the firm to bring what may be significant buying power leverage to market. This
requires the creation of a homogeneous data warehouse from disparate (heterogeneous) databases
(from various departments, locations). Some of the subtasks to creating a data warehouse are:
a) Supplier Normalization: It is likely that the same supplier (e.g. HP) might have been referred
differently in different systems (e.g. HP India, H.P., etc.). These aliases should be mapped to a unique
supplier name before creation of the data warehouse. Moreover, the parent child relationships often
need to be resolved – this is particularly difficult since mergers and acquisitions often lead to parent-
child relations within companies that are completely different. This entails the creation of a list of
distinct suppliers so that transactions to the same supplier can be grouped together.

105
b) Commodity Mapping: This requires each transaction to be mapped to an appropriate commodity
code, such as the UNSPSC code or a company proprietary code. This is due to the fact that the
transaction records at the invoice level often provide only a part level description of the commodity
and maybe associated supplier codes.
c) Data Visualization: Once the data is scrubbed and cleansed, a set of visualization and rendering
tools are required to view the different cross sections of the data so as to get an enterprise-wide view
of procurement spend.
9.4.2 Sourcing Strategy
Once a data warehouse is available for analyzing the procurement spend, the next step is to evaluate
the different sourcing options for each commodity class or other dimension and identify the potential
cost savings. This then provides a basis for a list of actionable sourcing initiatives. The subtasks to
creating such a strategy report as follows:
a) Demand Aggregation: The data warehouse provides a means to examine spend by each category,
supplier, plant etc. An important first step is to establish the number of suppliers being used for each
commodity class across all plants. Often such an exercise might reveal that the number of suppliers
that are being used for each commodity is very large and it presents an opportunity to allocate the
demand to a few suppliers and leverage the aggregate demand volume to negotiate better prices.
b) Supplier Scorecarding: While consolidating the supplier set for any commodity class it is important
to analyze the supplier performance against a set of company’s strategic metrics. The score carding
function helps identify the top suppliers to whom future allocation awards would likely go (despite
potentially having higher prices) as well as the bottom suppliers who would need to be more
aggressively managed as part of the “supplier relationship” activities.
c) E-Procurement model Selection: Different kinds of product require different e-procurement models.
Mostly four types of models are discussed in the literature: Use of E-Procurement Software, Internet
market exchange, B-to-B auctions, and Internet purchasing consortia (Table 1). These models can be
broadly classified under two technology options (Table 1): Catalog Buying and Contract Negotiation.
These topics will be elaborated in the next section.
d) Report Generation: Finally a report needs to be generated that outlines the sourcing strategy based
on the spend analysis.
Table 1 E-Procurement Models and Technology Options[1]
E-Procurement model Description

E-Procurement software Any Internet-based software application that enables employees to

purchase goods from approved electronic catalogues in accordance
Technology Option with company buying rules, while capturing necessary purchasing data
• Catalog Buying in the process.
Internet market exchanges Web sites that bring multiple buyers and sellers together in one central
Technology Option virtual market space and enable them to buy and sell from each other at
• Catalog Buying a dynamic price that is determined in accordance with the rules of the
• Contract Negotiation exchanges.
Internet B2B auctions Internet B2B auctions are events in which multiple buyers place bids to
Technology Option acquire goods or services at an Internet site. There are a variety of e-
• Contract Negotiation auction formats. Auctions enable organizational buyers to identify the
best offer from an expanded base of potential suppliers from around
the world. Sellers benefit by obtaining access to bid for business on a
level playing field rather than attempting to obtain business based on
networks of personal relationships.
Internet purchasing Internet service that gathers the purchasing power of many buyers to
consortia negotiate more aggressive discounts. Some organizations aggregate
Technology Option buying power for manufacturing inputs, while others perform similar
• Catalog Buying functions for indirect goods.
• Contract Negotiation

106
9.5 Sourcing
Activities carried out under this step depend on the type of the e-procurement model and the
corresponding technology selected at the strategic level.
9.5.1 Catalog buying
Catalog buying is carried out using Internet-based software application that enables employees to
purchase goods from approved electronic catalogues in accordance with company buying rules, while
capturing necessary purchasing data in the process. The employee’s selection of a good for purchase
from a supplier catalogue is automatically routed through the necessary approval processes and
protocols. E-Procurement software investment may take several forms, including purchase of a
software package from a third party technology provider (e.g., Ariba, CommerceOne), use of an e-
procurement system embedded in an Internet market exchange, subscription to e-procurement
software hosted and supported by an application service provider (ASP), or development of a
proprietary in-house system.
9.5.2 Contract Pricing
Another aspect of a sourcing strategy is to decide on a negotiation technique for contract pricing. The
ability to provide good estimates depends on how well the cost types of the suppliers can be
characterized. In addition, it is important to model and analyze the risks associated with the
uncertainty in the demand and choose contracted volumes optimally.
9.5.3 Core Ingredients of Contract Pricing
1. RFx: The RFI (Request for Information), RFQ (Request for Quotation), and RFP (Request for
Proposal) – collectively referred to as RFx – each represents a document and a means for a buyer to
specify the requirements of a purchase along multiple dimensions from multiple suppliers.
2. Protocol for price discovery: It can be done in a single round or in multiple rounds. The single
round systems are equivalent to sealed bids in traditional purchasing environment. Multiple round
auctions are like traditional English auctions and called reverse auctions for obvious reason.
3. Contracts: The negotiations lead to a contract which is then executed with one or more suppliers.

9.6 RFx
In B2B settings, the specification of purchases can get quite complex and require sophisticated
capabilities that allow the specification of complex items or services. Complex RFQs also need to
allow for a variety of bid structures that exploit complementarities and economies of scale in cost
structures of suppliers. An RFx is a document with an associated process initiated by a buyer in order
to solicit information, competitive quotes, or proposals from multiple suppliers. A sourcing platform
should make the RFx process as easy and straightforward as possible for all of the parties involved. It
should also be versatile enough to be used for both goods and services, and for both direct and indirect
spend categories. It should also support a wide range of RFx types and sizes, from simple RFIs to
complex RFPs.
Most RFx applications support a common set of capabilities such as the creation and editing of an
RFx document that mimics its paper-based counterpart. For example, this includes being able to add
any number of questions with response fields of the attribute types expected by the buying
organization (e.g., numeric, date, text, units of measurement, etc.) for each line item. Table 2
summarizes some of the major requirements in terms of bids (from the seller side) that are supported
in RFx systems.

107
 Table 2. Description of Complex Bid Types
Bid Types Description
Simple multi-line bids A bid includes multiple items, and specifies the unit price for each item.
Multi-attribute bids A bid includes multiple items, and specifies various relevant attributes for
each item, including unit price.

Bundled bids A bid includes multiple items, specifies the quantity of each item, and
provides a total bid price for all the items.

Volume discount bids A bid includes multiple items, and specifies the price curve of each item.
Configurable bids A bid includes multiple items, and specifies various relevant sets of values
for each attribute for each item. This provides a compact representation for
a large number of configurations (e.g. PCs) and needs to support mark-up
based pricing.

9.7 Protocols

9.7.1. Single round price discovery

RFQs are often used in a single round process that is similar to a one shot sealed bid auction where
the winners are selected (based on the recommendations of the bid evaluation engine) once all the
bids are in. After receiving such bids the buyer needs to identify the set of bids that minimizes total
procurement cost subject to business rules such as:
• The number of winning suppliers should be greater than a certain number (to avoid depending
too heavily on just a few suppliers), but smaller than a certain number (to avoid too much
administrative overhead);
• The maximum amount purchased from each supplier is bounded to a certain limit;
• At least one supplier(s) from a target group (e.g., minority) needs to be chosen; and
• If there are multiple winning bid sets, then one needs to pick the set that arrived first.
Decision support capabilities are essential to facilitate the creation and evaluation of such complex
RFQs and bids. Identifying the cost minimizing bid set subject to these business rules is a hard
optimization problem and difficult to do by hand (as is a common practice today). In addition, the
buyer is required to specify a scoring function that specifies the tradeoff of the non-price attributes
against price. This is difficult to do in a consistent manner without a rational process to elicit the
tradeoffs.
However, in a price negotiation context, it is often desirable to have a multiround process where after
each round the suppliers are allowed to reformulate their bids based on information about the winning
bids (more like based on feedback from the auctioneer).
9.7.2. Multiple round price discovery
A typical flow for negotiation is to get a bid response to the RFQ from the suppliers and choose the
appropriate bid/bids that satisfy the requirements of the purchase at minimum cost. With the advent of
the Internet, online reverse auctions represent a new tool in the purchasing department’s toolbox to

108
potentially increase competition through open, real-time, competitive bidding, which requires an
iterative bidding protocol.

Fig. 3 Process Flow for Iterative Auctions

Such a multi-round process is illustrated in Fig 3. The bid evaluation engine provides the decision
support for all the three functions required for multi-round negotiations and iterative auctions. Winner
determination identifies the winning bids from a given set of bids to minimize the total procurement
cost, the pricing module prescribes the payment to be made by each winner (this could be in general
different from the bid price to promote efficiency in the market), and signaling provides a “market
clearing” price for bid reformulation. This iterative process continues until there are no new bids or
closing time.

Most reverse auction formats allow for live, real-time, open, competitive bidding where bidders must
outbid the current winning bid in order to win the business. There are a variety of auction formats and
settings. The most basic reverse auction is a price-only auction for a single item. Most auction
providers (and there are many) provide a wide range of formats and settings including multiple
quantities of a line item, multiple line items, time extensions, start and reserve prices, partial quantity
bids and award allocations, and bundled bids – to name only a few. However, there are three advanced
auction formats/settings of note:
• Combinatorial Auction – allows suppliers to mix bundled bids along with un-bundled bids
• Volume Discount Auction – allows suppliers to establish price discounts at certain quantities
• Multi-attribute Reverse Auction – the winner(s) is determined by a score (rather than just
price) calculated using the buyer’s weights and preferences for price, quantity, and any
number of other attributes.

9.8 Contract Management

One of the main goals of a sourcing project is to execute one or more purchasing contracts with one or
more suppliers. The prices and terms of the line items covered by a contract were previously
negotiated in RFx and auction rounds. The contracts themselves, however, also go through a different
form of negotiation at a more legally precise level. Once executed, these contracts are meant to be
used to procure the contracted line items (perhaps via a procurement system) using the negotiated

109
prices and terms. A sourcing platform should provide a means to generate a contract based on its
preceding RFx and auction negotiations, support contract negotiations, and monitor compliance to the
contracts’ business commitments over time.
Contract monitoring capabilities incorporated into contract management software include:
• An alert notification is sent when a contract is soon to expire.
• The buyer’s purchase volume commitments can be monitored with alert notifications sent if there is
danger of buying under the minimum quantity within the designated time period.
• Notifications can be sent alerting the buyer and/or supplier of a supplier’s violation of a delivery
commitment.
In all of the contract management solutions today, there is a specific and important shortcoming.
Namely, there are two key parts enabling automated contract monitoring which currently must be
performed manually.
• First, the contract commitments to be monitored must be manually selected out of the contract’s
negotiated legalese into a structure easier to analysis.
• Second, the business process data and raw transaction data needed to assess whether commitments
are being fulfilled or violated is also captured mannually.
Business process integration and management (BPIM) and business activity monitoring (BAM)
systems are beginning to address the challenges associated with contract monitoring systems.
9.9 Risks Associated with e-Procurement Technologies
In a survey conducted by Davila et al. [1], respondents perceive certain risks linked to the adoption of
e-procurement technologies that need to be addressed before these technologies are widely accepted.
These risks include:
Internal business risks: companies are uncertain about whether they have the appropriate resources to
successfully implement an e-procurement solution. The experimentation of the companies following a
‘wait and see’ strategy may help to develop the required absorptive capacities. Implementing an e-
procurement solution requires not only that the system itself successfully performs the purchasing
process, but most important, that it integrates with the existing information infrastructure. This
internal information infrastructure includes systems such as accounting, human resources, asset
management, inventory management, accounts payable, production planning, and cash management
systems. Most organizations adopting or looking to adopt e-procurement software already have
significant investments in these other systems; integrating these new technologies with existing
platforms should happen as smoothly as possible. Failure to integrate creates duplicative work steps
and jeopardizes the reliability of organizational information.
External business risks: e-procurement solutions need not only ‘talk’ with internal information
systems, but also need to cooperate with external constituencies — mainly customers and suppliers.
External constituencies need to develop internal systems that facilitate the communication through
electronic means — an issue that demands technology investments as well as incentives for these
constituencies. For e-procurement technologies to succeed, suppliers must be accessible via the
Internet and must provide sufficient catalogue choices to satisfy the requirements of their customers.
Ideally, suppliers will provide e-catalogues in the formats required by customers, reflecting custom
pricing and/or special contractual agreements, and will send updates on a regular basis. However,
suppliers, especially in low margin industries, may be hesitant or even unable to meet such demands
without guarantees of future revenue streams. Lack of a critical mass of suppliers accessible through
the organization’s e-procurement system would limit the network effects that underlie these
technologies, further hindering the acceptance and adoption of the technology. Cooperation with
external parties also requires new suppliers and customers to meet the business criteria that
organizations have set to accept them in their networks. Since some of the business models associated
with e-procurement technologies (e.g. auctions, consortia, and exchanges) clearly envision the use of
suppliers with whom the buyer has not previously transacted business, companies need to develop

110
mechanisms that provide the buyer with assurances that the supplier meets or exceeds recognizable
and industry enforced standards relating to supplier quality, service, and delivery capabilities.
Technology risks: companies fear the lack of a widely accepted standard and a clear understanding of
which e-procurement technologies best suit the needs of each company. The lack of a widely accepted
solution blocks the integration of different e-procurement software across the supply chain. The
significance of this risk factor seems to suggest the need for clear and open standards that would
facilitate inter-organization e-procurement technologies. Without widely accepted standards for
coding, technical, and process specifications, e-procurement technology adoption will be slow and
will fail to deliver many of the benefits expected.
E-procurement process risks: another set of risks has to do with the security and control of the e-
procurement process itself. Organizations must be confident, for example, that unauthorized actions
will not disrupt production or other supply chain activities when committing to e-procurement
technologies. Thus, the challenge for the e- procurement technology adoption is to provide evidence
to non-users that these technologies (1) do not undermine control, security, or privacy requirements;
(2) they are not so technically complex that organizations without a sufficient technology skill set
cannot use them, and (3) the new business model provides the right incentives to supply chain
constituencies to effectively use these technologies.
Table 3 identifies the changes in the buyer – supplier relationship as a major barrier to e-procurement
technology use. While technology is perceived as a barrier, reflected in the ‘lack of common
standards’ concerns for e-procurement software, most barriers point to the need for redesigning these
relationships. If, for example, the use of e-procurement undermines amicable trading relationships,
buyers are concerned about how they will obtain needed goods when supplies get tight. Buyers are
also concerned that these technologies will push prices down to the point where suppliers cannot
invest in new technology or product development, upgrade facilities, or add additional productive
capacity. Additional price pressures can even push suppliers with a poor understanding of their cost
structure out of business. Finally, integration with existing mechanisms is seen as another barrier.

Table 3: Three Most Frequently Identified Barriers to E-Procurement Technologies Utilization

E Procurement Software
• Problems integrating with existing system
• Lack of common standard for e-commerce software development
• Lack of suppliers’ accessibility to the organization’s e-procurement system and/or lack of
supplier investment in catalog development
Internet Exchanges
• Not enough suppliers to create a liquid marketplace
• Suppliers reluctance to participate in selling environments where preeminent focus is on price
• Supplier’s reluctance to participate because control is lost over the presentation of brand name
and product features.
E-Auctions
• Organizational discomfort with auctions, as opposed to honoring commitment to supplier
partnering and consolidation.
• Downward price pressure on vendors resulting in diminished customer service or quality.
• Inability to identify potential items for auction.
Purchasing consortia
• Pricing is not significantly better than available without consortia.
• Getting sufficient number of vendors into the process.
• Ensuring conformance to state laws and regulations

111
Reference

[1] Antonio Davila, Mahendra Gupta and Richard Palmer Moving Procurement Systems to the
Internet:: the Adoption and Use of E-Procurement Technology Models European Management
Journal, Volume 21, Issue 1, February 2003, Pages 11-23
[2] William D. Presutti Supply management and e-procurement: creating value added in the supply
chain Industrial Marketing Management, Volume 32, Issue 3, April 2003, Pages 219-226
[3] Robert Guttman, Jayant Kalagnanam, Rakesh Mohan and Moninder Singh, Strategic Sourcing
and Procurement, In Supply Chain Management on Demand Strategies, Technologies,
Application, Chae An and Hansjörg Fromm Eds., Springer Berlin Heidelberg, 2005, pages 117 –
142

***********

112
 CHAPTER X
Economic Theory of Auctions

10.1. Introduction
Auctions have been widely adopted as tool for buying and selling goods and services. Auctions can be
used to sell (allocate) almost all kinds of goods. The governments use them to sell public resources
such as radio spectrum licenses and oil drilling rights; the firms and individuals use them to sell
houses, flowers, antiques, etc. They also find applications in fields of computer science, such as
allocating bandwidth in the communication networks. The online auction business model is the one in
which participants bid for products and services over the Internet. The functionality of buying and
selling in an auction format is made possible through auction software which regulates the various
processes involved. eBay, the world's largest online auction site, is one of the better known examples
of buying auctions. Similarly, a reverse auction is a tool used in industrial business-to-business
procurement. It is a type of auction in which the role of the buyer and seller are reversed, with the
primary objective to drive purchase prices downward. One example of a reverse auction site could be
Metal Junction – a consortium of Tata Steel and SAIL. A less know online auction business is
adopted by Dell and GM. They use auction to sell second hand products (used) which they buyback
from their clients. A well designed auction mechanism can add to company’s profitability. Similarly,
a well designed bidding strategy can save a bidder falling into a trap like that of winner’s curse.
Studying auction theory can help understanding basic design issues. Auction theory itself is an
important part of economic theory and it helps to understand properties of the markets, such as the
price formulation and information structures.

10.2. Basic Auction Types

The exposition of auction theory typically starts with the introduction of four basic auction formats:
two open auctions and two based on sealed bids:

Open bids
• Ascending bid auction (also called English Auction) – In this auction the price is successively
raised until one bidder remains. This bidder wins the object at the final price. The auction can
run by the auctioneer calling price, the bidder submitting prices, or electronic bids with the
highest bid posted continuously. Once somebody quits the process they are not allowed back
in. This auction format is common in art, livestock, and some Internet-based procurement
auctions.
• Descending bid auction (also called Dutch auction) – In this auction the price starts from a
high level and called down. The first bidder who accepts the current price wins. This is how
the Dutch flower auctions are managed, but there are not many other examples of Dutch
auction.

Sealed Bid Auctions

• First price sealed bid auction (1SB) – Each bidder submits a single bid (independently) and
the item is sold to the highest bidder who pays the winning bid. This is the most common
form in procurement auctions and in many government contract auctions.
• Second price sealed bid auction (2SB) – Each bidders submit a single bid (independently) and
the item is sold to the highest bidder. Unlike the first price sealed bid auction, however, the
 price that the winner pays is the second highest price (“second price”). This type of auction
was first suggested by Vickrey. Although it is rarely practiced, its value is the insight it offers
to other auction mechanisms.

10.3. A framework for defining and categorizing auction

Kalagnanam and Parkes[1] have developed a framework for classifying auctions based on the
requirements that need to be considered to set up an auction. These core components are:
Resources The first step is to identify the set of resources over which the negotiation is to be
conducted. The resource could be a single item or multiple items, with a single or multiple units of
each item. An additional consideration common in real settings is the type of the item, i.e. is this
standard commodity or multi-attribute commodity. In the case of multi-attribute items, the agents
might need to specify the non-price attributes and some utility/scoring function to tradeoff across
these attributes.
Market Structure An auction provides a mechanism for negotiation between buyers and sellers. In
forward auctions a single seller is selling resources to multiple buyers. Alternately, in reverse
auctions, a single buyer is sourcing resources from multiple suppliers, as is common in procurement.
Auctions with multiple buyers and sellers are called double auctions or exchanges, and these are
commonly used for trading securities and financial instruments and increasingly within the supply
chain.
Preference Structure The preference structure of agents in an auction is important and impacts some
of the other factors. The preferences define an agent's utility for different outcomes. For example,
when negotiating over multiple units agents might indicate a decreasing marginal utility for additional
units. An agent's preference structure is important when negotiation over attributes for an item, for
designing scoring rules used to signal information.
Bid Structure The structure of the bids allowed within the auction defines the flexibility with which
agents can express their resource requirements. For a simple single unit, single item commodity, the
bids required, are simple statements of willingness to pay/accept. However, for multiunit identical
items setting bids need to specify price and quantity. Already this introduces the possibility for
allowing volume discounts, where a bid defines the price as a function of the quantity. With multiple
items, bids may specify all-or-nothing bids with a price on a basket of items. In addition, agents might
wish to provide several alternative bids but restrict the choice of bids.
Matching Supply to Demand A key aspect of auctions is matching supply to demand also referred to
as market clearing, or winner determination. The main choice here is whether to use single-sourcing,
in which pairs of buyers and sellers are matched, or multi-sourcing, in which multiple suppliers can be
matched with a single buyer, or vice-versa. The form of matching influences the complexity of winner
determination, and problems range the entire spectrum from simple sorting problems to NP-hard
optimization problems.
Information Feedback Another important aspect of an auction is whether the protocol is a direct
mechanism or an indirect mechanism. In a direct mechanism, such as the first price sealed bid
auction, agents submit bids without receiving feedback, such as price signals, from the auction. In an
indirect mechanism, such as an ascending-price auction, agents can adjust bids in response to
information feedback from the auction. Feedback about the state of the auction is usually
characterized by a price signal and a provisional allocation, and provides sufficient information about
the bids of other agents to enable an agent to refine its bids. In complex settings, such as multi-item
auctions with bundled bids, a direct mechanism can require an exponential number of bids to specify
an agent's preference structure. In comparison, indirect mechanisms allow incremental revelation of
preference information, on a “as required basis”. The focus in the design of indirect mechanisms is to
identify how much preference information is sufficient to achieve desired economic properties and
how to implement informationally-efficient mechanisms. A related strand of research is to provide
compact bidding languages for direct mechanisms.

114
Each of the six dimensions provides a vector of choices that are available to set up the auction. Putting
all of these together generates a matrix of auction types. The choices made for each of these
dimensions will have a major impact on the complexity of the analysis required to characterize the
market structure that emerges, on the complexity on agents and the intermediary to implement the
mechanism, and ultimately on our ability to design mechanisms that satisfy desirable economic and
computational properties.

10. 4. Evaluating Auctions

When deciding between various auction mechanisms, the auctioneer has a very large number of
auction designs to choose from. The most important criteria in the choice of auction format are the
following [2]:
• Revenue – auctioneers are looking for the auction that will yield the maximum revenue for
the item sold. While this is an important tenet of auction theory and will be assumed to be the
case in the bulk of this section, other considerations are also important not only to
governments but to many corporations.
• Efficiency – an auction is successful if the bidder that values the item most ex post - actually
gets it. In some contexts, such as government auctions this is important, especially when the
government is selling public assets. When the sale involves future delivery of services, as in
many procurement auctions, efficiency means that the contract is more likely to be carried out
and the service provided at a high level.
• Time and Effort – many B2B auctions involves the trading of many (sometimes tens of
thousands) of items while soliciting bids from many (sometimes hundreds) of suppliers.
Furthermore such auctions are conducted periodically, every year (for many MRO services)
or every product model (for direct material). Auctioning organizations have to devote a great
deal of time and effort to such auctions and thus mechanisms that minimize the time and
effort involved are the ones that will be used.
• Simplicity – One of the objectives of auctioneers in most auctions is to get as many
participants as possible. Keeping the rules simple, especially knowing that many suppliers
have to respond to hundreds of auctions every month, helps participation in many situations.
While the literature on auction theory focuses on the first two criteria, the last two are very important
in procurement auctions.

10.5. Basic Approaches for Auction Design:

The basic economic methodology used in the design of electronic intermediaries first models the
preferences, behavior, and information available to agents, and then designs a mechanism in which
agent strategies result in outcomes with desirable properties. Kalagnanam and Parkes[1] consider two
approaches to modeling agent behavior:
Game-theoretic/mechanism design The first model of agent behavior is game theoretic and relates
to mechanism design theory. In this model the equilibrium state is defined by the condition that agents
play a best-response strategy to each other and cannot benefit from a unilateral deviation to an
alternative strategy.
Price-taking/competitive equilibrium The second model of agent behavior is price-taking, or
myopic best-response, and relates to competitive equilibrium theory. In this model the equilibrium
state is defined by the condition that an agent plays a best-response to the current price and allocation
in the market, without modeling either the strategies of other agents or the effect of its own actions on
the future state of the market.

115
10.5.1 Game-theoretic/mechanism design approach for auction design
An auction is a game with partial information where a player's valuation of an object is hidden from
other players. It is served as a popular way in resources (goods) allocation by specifying a set of rules
to determine the winner(s) and the related payments. A typical setting of the auction is that a seller
attempts to sell one or more items to a set of bidders. The involved players (seller and bidders) do not
have complete information about the value of the items on sale in the sense that they do not know
other’s value but know their own value, which may or may not be affected by others. All players are
assumed to be selfish and payoff-maximizing. The auction theory studies the behavior of the players
in this non-cooperative environment.

10.5.1.1 Basic Auction Models

In any auction process the seller does not know the value that bidders place on the item auctioned
off and bidders do not know with certainty how other bidders value the item. The information
available to bidders and the corresponding type of auctions can be classified as follows:
• Private value (PV) auctions – where each bidder knows only the value of the item to
himself. Even if he would know what other bidders are willing to pay this would not affect his
own valuation. Such a model is appropriate when the value of an item to a bidder is derived
from its consumption alone and not from later resale.
• Interdependent value auctions – where the value of the items sold is not known to the
bidders. Each bidder has only an estimate (signal) regarding the value (this may be an expert
opinion or a test result). If a given bidder would have known the signals of other bidders, his
own estimate of the true value may change.
• Common value (CV) auctions – a special case of interdependent values in which the value
of the item ex post is the same for all bidders. For example, when oil leases are auctioned off,
bidders have only their own test results regarding the actual amount of oil in the tract being
leased. After the auction, however, the winner will find out exactly the amount of oil in the
ground and this oil has a certain market value.
The simplest and most thoroughly investigated auction model is the symmetric independent private
values (SIPV) model with risk-neutral bidders, in which (1) a single indivisible object is for sale
(single-object auction); (2) each bidder knows his own valuation about the object but no one else dose
(private value). The unknown valuations are independent and identically distributed (independence,
symmetry); (3) all bidders are ex ante identical (symmetry); (4) Bidders are risk-neutral. This model
serves as a prototype in the research of auction theory. All results in the literature either can be
derived directly from this model or come from relaxing some assumptions or with some other features
and information in different situations.

10.5.1.2 Mechanism Design Problem

The mechanism design approach to solving distributed allocation problems with self-interested agents
formulates the design problem as an optimization problem. Mechanism design addresses the problem
of implementing solutions to distributed problems despite the fact that agents have private information
about the quality of different solutions and that agents are self-interested and happy to misreport their
private information if that can improve the solution in their favor. A mechanism takes information
from agents and makes a decision about the outcome and payments that are implemented. It is useful
to imagine the role of a mechanism designer as that of a game designer, able to determine the rules of
the game but not the strategies that agents will follow.
A mechanism defines a set of feasible strategies, which restrict the kinds of messages that agents can
send to the mechanism, and makes a commitment to use a particular allocation rule and a particular
payment rule to select an outcome and determine agent payments, as a function of their strategies.

116
Game theoretic methods are used to analyze the properties of a mechanism, under the assumption that
agents are rational and will follow expected-utility maximizing strategies in equilibrium.

10.5.1.3 Direct Revelation Mechanism

The space of possible mechanisms is huge, allowing for example for multiple rounds of interaction
between agents and the mechanism, and for arbitrarily complex allocation and payment rules. Given
this, the problem of determining the best mechanism from the space of all possible mechanisms can
appear impossibly difficult. The revelation principle allows an important simplification. The
revelation principle states that it is sufficient to restrict attention to incentive compatible direct-
revelation mechanisms. In a direct-revelation mechanism (DRM) each agent is simultaneously asked
to report its type. In an incentive-compatible (IC) mechanism each agent finds it in their own best
interest to report its type truthfully. The mechanism design problem reduces to defining functions that
map types to outcomes, subject to constraints that ensure that the mechanism is incentive-compatible.
The IC constraints require that when other agents truthfully report their types an agent's best response
is to truthfully report its own type, for all possible types. In technical terms, this ensures that truth-
revelation is a Bayesian-Nash equilibrium, and we say that the mechanism is Bayesian-Nash incentive
compatible. In a Bayesian-Nash equilibrium every agent is plays a strategy that is an expected utility
maximizing response to its beliefs over the distribution over the strategies of other agents. An agent
need not play a best-response to the actual strategy of another agent, given its actual type. This
equilibrium is strengthened in a dominant strategy equilibrium, in which truth-revelation is the best-
response for an agent whatever the strategies and preferences of other agents.

10.5.1.4 Efficiency Vis-à-vis Optimality in Auction

There are two natural design goals in the application of mechanism design to auctions and markets.
One goal is allocative efficiency, in which the mechanism implements a solution that maximizes the
total valuation across all agents. This is the efficient mechanism design problem. Another goal is
payoff maximization, in which the mechanism implements a solution that maximizes the payoff to a
particular agent.
Optimal auctions are designed to maximize the expected revenue of the seller by a set of tools
including posing a reserve price or charging an entry fee, whereas the objective of efficient auctions is
to maximize the social welfare, the sum of the players' surplus. Roughly speaking, the efficient design
aims to maximize the system welfare, whereas the optimal design aims to maximize the seller's
individual revenue. Since optimality and efficiency usually can not be achieved simultaneously, the
auction designers have to make the choice before starting address the rules. A financial self-interested
agent may prefer the optimal auctions, while a public agent like the government may prefer the
efficient auctions to gain more social welfare. Nevertheless, all agents need to balance optimality and
sufficiency to make the auctions practical.

10.5.1.5 Efficient Auction Mechanism

The efficiency problem in the single-object auction where buyers have private values is theoretically
solved in Vickrey's pioneering work. The winner is the buyer whose valuation of the good is highest
and will pay the second highest value. Truth-telling is a weakly dominant strategy for any buyer. The
mechanism is known as Vickrey's auction, which also applies in the case of multiple identical objects.
This format is significantly extended to a mechanism called Vickrey-Clarke-Groves (VCG)
mechanism4. The VCG mechanism works for homogeneous goods as well as heterogeneous ones in
private value environment.

117
10.5.1.5.1 Vickrey’s Mechanism: an incentive compatible direct revelation mechanism
In the sealed second price auction the dominant strategy is to bid the actual valuation, regardless of
the other players’ strategy. To illustrate the validity of this strategy, we demonstrate using a
hypothetical situation
Consider bidding (v − Δ ) , if the highest bid other than this is bmax , then the following cases arise:

a. If v < bmax , the bid is not won (same as bidding v )

b. If (v − Δ ) > bmax , the bid is won and payment of bmax (same as bidding v )

c. If v > bmax > (v − Δ ) , bidding (v − Δ ) means the auction is lost. However bidding v would
have helped win the auction with a surplus of (v − Δ )

Similarly, now consider bidding (v + Δ ) . If the highest bid other than this is bmax , then the following
cases may occur:
a. If (v + Δ ) < bmax , the bid is not won (same as bidding v )

b. If v > bmax , the bid is won and payment of bmax (same as bidding v )

c. If v < bmax < (v + Δ ) , bidding (v + Δ ) means the auction is won. However, while bidding
v would have meant losing the auction the final payment is greater than the valuation by an
amount of with a surplus of bmax .

Therefore, as demonstrated, bidding more than or less than the actual valuation never helps and may
cause losses. Bidding the true value is the dominant strategy.
VA is concerned with auctioning off a single good. Clarke and Groves mechanism extends the
concept of strategyproofness to the combinatorial auction setting.

10.5.1.5.2 Optimal Auctions

The problem of designing an auction that maximizes the seller's revenue (optimal auction) is usually
more challenging, especially in the case of multiple heterogeneous objects. However, Myerson's
revelation principle allows one, without loss of generality, to restrict attention to a direct-revelation
mechanism that is used to prove a surprising result called “revenue equivalence theorem". His paper
provides the framework that has become the paradigm for the research of optimal mechanism design
for selling one or more homogeneous objects.
In this section, we will review the results on the revenue equivalence theorem under different
circumstances. In the case that this theorem does not hold, we compare different auction formats
based on the revenue-generating ability. Since the reserve price is one of the main tools to maximize
the seller's revenue, we also present the main results in this aspect. We will discuss the tradeoff
between optimality and efficiency in the auctions design in the next section.

10.5.1.5.3 Strategic Equivalence of Auctions

Open auctions require the bidders to be all present (physically or digitally) at the same time and place,
offering bidders the opportunity to observe the behavior of other participants. Sealed bids, on the
other hand, are mailed in or submitted electronically with no opportunity for information feed-back.
Interestingly, some of these differences do not matter to rational decision makers.
Note first that submitting the winning bid in a 1SB auction is equivalent to buying the item at that
price through Dutch auction, provided the item is still available. This is true despite the fact that a

118
(first price or otherwise) sealed bid auction offers no opportunity to observe other bidders behavior
while the Dutch auction is “open.” Naturally, the reason is that in a Dutch auction the first time that
somebody enters, he wins and the auction terminates. Thus, in 1SB and Dutch auction bidders have to
decide a-priori how much to bid and the auctioneer will get what the highest bidder submitted. In
economic terms, 1SB auctions are strategically equivalent to Dutch auctions in the sense that for
every realization of bidders’ estimates the two auctions induce identical equilibrium outcomes (i.e.,
winners and prices).
In English auctions it makes sense for a bidder to stay in the auction until his value has been reached
(in fact, it is his dominant strategy, as discussed below). He should not drop beforehand and certainly
should not stay longer. Thus the winning bidder will stay until the next-to-last bidder drops out,
meaning that he will pay the price in which the last bidder dropped out, or the second highest bid.
Thus English auctions are equivalent to 2SB auctions. This equivalence is not as strong as the one
between 1SB and Dutch auctions since with English auctions bidders do get information throughout
the process when they see the prices in which other bidders drop out and can therefore adjust their
estimate of the value of the item being sold and their strategy. This is not a consideration with private
value auctions but important in common and interdependent value auctions. These equivalences are
depicted in the following Figure

PV
Dutch 1st Price
CV

PV
English 2nd Price
Equivalence of Open and Sealed Bid Auctions

The most common types of procurement auctions (where the winner is the participant who offered
the lowest bid) are the 1SB, where the winner gets to supply the item to the buyer at the lowest bid
price and English auction, where the winner supplies the item or service at the second lowest price.
English auctions became popular with procurement managers only during the 1990-s with the
advent of Internet-based auctions.
10.5.1.5.4 Revenue Equivalence
In this section we obtain the revenue’s generated from the First Price and Second Price auction
models.
Assume that there are n bidders and their values {V1 , V2 ,..., V N } are drawn from independent identical
distributions, F (v) with density function f (v). Also the order statistics are {V1′, V2′,..., V N′ }and the
density function of the kth lowest value is given by

119
 ′ n
f ⎛⎜ v k ⎞⎟ = ⋅ f (v ) ⋅ [F (v )] ⋅ [1 − F (v )]
k −1 n−k

⎝ ⎠ (k − 1)!⋅(n − k )!
Considering that the valuations are drawn from a uniform distribution U [0, 1], the distribution of the
kth order statistics is given by:

′ n
f ⎛⎜ v k ⎞⎟ = ⋅ v k −1 ⋅ [1 − v ]
n−k

⎝ ⎠ (k − 1)!⋅(n − k )!
This is a beta distribution with parameters k and n-k+1, the mean of the distribution is
′ k
E ⎡v k ⎤ =
⎢⎣ ⎥⎦ n + 1

First Price Revenue

In the First Price Auctions, the bidders shave their bids by a certain amount. The winner is the bidder
with the highest value, in other words the highest order statistic, which from the calculations above is
n
E [v n ] =
n +1
Further, from the previous section, the best bidding strategy is
n −1
b∗ = ⋅ E [v n ]
n
Therefore the expected winning bid is
n −1
b∗ =
n +1
This is also the auctioneer’s payoff.

Second Price Revenue

In the Second Price Auction system the winner pays the second highest bid. Therefore, continuing the
earlier illustrations, the expected payoff for the auctioneer will be the expected value of the second
highest bid in the auction. From Section 4, this is
n −1
E [v n −1 ] =
n +1
Therefore, we observe that the First Price and the Second Price Auction formats lead to the same
expected payoff for the auctioneer. This is the illustration of one of the most fundamental results in
auction theory, namely the Revenue Equivalence Theorem. This theorem basically states that under
certain conditions, any type of auction mechanism lead to the same amount of payoff for the
auctioneer.
Revenue Equivalence Theorem
The Revenue Equivalence Theorem is the most fundamental result of Auction Theory. The statement
of the theorem is:
Assume each of a given number of risk neutral potential buyers of an object has a privately
known signal independently drawn from a common, strictly increasing distribution. Then any
auction mechanism in which (i) the object always goes to the buyer with the highest signal,

120
 and (ii) any bidder with the lowest feasible signal expects zero surplus, yields the same
expected revenue and results in each bidder making the same expected payment as a function
of their signal.

This result applies to both private value models as well as the more general common value model;
however the signals have to independent. Thus all the common auction formats, ascending bid,
descending bid, first price sealed bid, second price sealed bid and even some non – standard auction
formats always lead to the same expected revenue for the auctioneer under the stated conditions.

Proof of the Revenue Equivalence Theorem

We consider the ‘independent private values’ model, where n bidders compete for a single unit.
Bidder i, values the unit at vi drawn from the same continuous distribution F (v ) on v, v (so [ ]
()
that F (v ) = 0, F v = 1 ) with density function f (v ) . All bidders are risk neutral.

For any given mechanism, for a given bidder i, let S i (v ) be the expected surplus, as a function of the
type. Let Pi (v ) be the probability of receiving the object. The following equation is the key

S i (v ) ≥ S i (v~ ) + (v − v~ ) ⋅ P(v~ )

The right hand side is the surplus that player i would obtain if she had type v but deviated from
equilibrium behavior and followed the strategy of type v~ . In equilibrium, v must not prefer to deviate
from it, so the left hand side must (weakly) exceed the right hand side. Therefore we have,
S i (v ) ≥ S i (v + dv ) + (− dv ) ⋅ Pi (v + dv )

And since v + dv must not want to imitate type v we have

S i (v + dv ) ≥ S i (v ) + (dv ) ⋅ Pi (v )

Combining the previous two equations,

S i (v + dv ) − S i (v )
Pi (v + dv ) ≥ ≥ Pi (v )
dv

Taking the limit as dv → 0 , we obtain

dS i
= Pi (v )
dv

Integrating,

121
 v
S i (v ) = S i (v ) + ∫ P (x )dx
i
x =v

At any type v̂ the slope of the surplus function is Pi (vˆ ) , so if we know S i (v ) , we have the complete
picture. Now, considering any two mechanisms with the same S i (v ) and the same Pi (v ) functions for
all v and for every player i. They have the same S i (v ) function. So any given type, v, of player i,
makes the same expected payment in each of the two mechanisms. Therefore the expected payment
averaged across the different possible types is also the same. Since this is true for all bidders, these
mechanisms yield the same revenue for the auctioneer.

10.5.1.5.5 Auctions with Interdependent values

In auctions with interdependent values it is assumed that bidders have only partial information about
the value of the item being auctioned off, in the form of a signal, which is a random variable. The
extreme case of interdependent value auctions are common value auctions in which all bidders have
the same ex post value for the item being auctioned off. Before the auction, however, each bidder has
only a random signal regarding the value.
Naturally, each bidder’s estimate of the true underlying value of the item will improve if he gets
information about the other bidders’ signals – for example, when they drop from the process during an
English auction.
Interdependent value auctions are important to consider since it is difficult to imagine a situation in
which there is not at least some portion of the value which is common. Even if bidders intend to
purchase an item only for private consumption (such as a house or a painting), they may keep an eye
towards selling it in the future, meaning that they would like also to consider the “future market
value” of the item purchased, which is common, or at least depends on others’ valuations.
Revenue Comparison
When analyzing interdependent and common value auctions, many economists also drop the
assumption that the signals are independent and assume that bidders’ signals are affiliated. Affiliation
is a strong form of positive correlation. Such treatment makes sense because in most cases the
technology used to get the signals is similar. Since the revenue equivalence does not hold any more
with affiliated signals, it is natural to question which auction format yields more revenue to the
auctioneer.
With affiliated signals and common values the English auction may yield higher revenue for the
auctioneer than a first price auction. The reason is that the transmission of information allows bidders
to continuously upgrade the quality of their estimate of the true value of the item being sold, and thus
they need to hedge less, bidding higher to the benefit of the auctioneer.
Efficiency of Auctions
In pure common value auctions, there is no issue of efficiency since ex post all bidders value the item
identically. In auction with interdependent values, however, the item sold will be awarded to the
bidder with the highest signal, who may or may not be the bidder with the highest value.26 Efficiency
is assured only if the signals are such that the ex post values of all bidders can be ordered in the same
way that their signals are ordered. Under this condition, all the standard auctions will be efficient,
provided only that the signals are symmetric.
Winner’s Curse
An interesting phenomenon which takes place when bidders fail to account for it and in the presence
of interdependent values is the winner’s curse. The winner’s curse takes place when winners pay too
much, due to their failure to anticipate and correct, in their bidding strategy, a bias in their estimate of
the value of the item being auctioned off. The phenomenon arises since the winner of an auction with

122
interdependent values is the bidder who submitted the highest bid, which is the bidder who had the
highest signal. When notified of the win, the bidder knows one thing for sure – his signal was higher
than the signal of all other (n -1) participants in the auction. This means that the winner is the bidder
who most overestimated the value of the item sold.

10.5.1.5.6 Basic Design Considerations for Auctions

There are several considerations to be kept in mind while designing auction mechanisms. Even in the
case of established auction mechanisms, there are several aspects that have to be catered to in order
for the auctioneer to gain the most benefit. This section illustrates some such aspects.
Number of Bidders: It is observed that the auctioneer increases the returns from the auction process
with an increase in the number of bidders. Therefore the auctioneer should invest in getting more
participants in the bidding process, especially if the number of bidders in the fray is low.
Reservation Price: Auctioneers should set a reservation price, i.e. a minimum asking price below
which they are not willing to sell the item. The setting of this reservation price should be such so as to
maximize the returns to the auctioneer. As an illustration supporting the case for setting a reservation
price, consider the following scenario:
Assume that the auctioneer’s value for the item is w = 0 and he sets a positive (low) reservation price
of r. For small r, the probability that a bid is less than r is low, that is F(r) << 1. The range of
valuations is U [0, 1], the following cases are possible:
a. If both bids are above r, setting r did not hurt the auctioneer
b. If both bids are below r, then there is no sale. However the probability of this event occurring
is F [(r )] and the maximum loss is r
2

c. If one bid is below it and one bid is above then the winner pays r instead of the second
highest price and the auctioneer gains. The probability of this occurring is 2F(r) [1=F(r)].
Thus for small r, the probability of gaining is proportional to F(r) and the probability of losing is
proportional to F [(r )] . Hence it is clearly advantageous to set a reservation price.
2

For determining the optimal reservation price we use the following argument:
Assume that the auctioneers value is 0, there are n bidders and the reserve price is set to r (where r>0)
and the auctioneer considers raising it by a small amount δ.
If the highest bidder bids above (r+ δ), then there is a gain of δ and the probability of the occurrence
of this event is n ⋅ F [(r )] ⋅ [1 − F (r + δ )] . This is a bad move if the highest bidder bids between r
n −1

and (r+ δ). The auctioneer loses r and the probability of this happening is
n ⋅ F [(r )] ⋅ [F (r + δ ) − F (r )]
n −1

Therefore the net gain per increment of δ in r is:

n ⋅ F (r ) ⋅ [1 − F (r + δ )] F (r + δ ) − F (r )
n −1
G (δ ) = ⋅ δ − n ⋅ F (r )
n −1
⋅ ⋅r
δ δ
Simplifying and taking the limit as the increment of the reservation price goes to zero:

G (δ ) = n ⋅ F (r ) ⋅ {[1 − F (r )] − f (r ) ⋅ r}
n −1
lim
δ →0

Setting G (δ) = 0 for optimum, we have

1 − F ( r′)
r′ =
f ( r ′)

123
The optimal reservation price r ′ is the solution of the above equation; this is independent of the
number of bidders.

10.5.2 Price-taking/competitive equilibrium approach for auction design

Competitive equilibrium theory is built around a model of agent price-taking behavior. At its heart is
nothing more than linear-programming duality theory. One formulates a primal problem to represent
an efficient allocation problem, and a dual problem to represent a pricing problem. A competitive
equilibrium condition precisely characterizes complementary-slackness conditions between an
allocation and a set of prices and implies that the allocation is optimal and therefore efficient.
Competitive equilibrium conditions are useful because they can be evaluated based on myopic best-
response bid information from agents, and without requiring complete information about agent
valuations. This is the sense in which prices can decentralize decision-making in resource allocation
problems.
The modeling assumption of price-taking behavior states that agents will take prices as given and
demand items that maximize payoff given their valuations and the current prices. This is commonly
described as price-taking or myopic best-response behavior. In the language of mechanism design,
this can be considered a form of myopic, or bounded, incentive-compatibility.
To illustrate competitive equilibrium (CE) prices we will impose some structure on choice set K and
define the combinatorial allocation problem (CAP). Let G define a set of items, and S ∈ K a subset,
or bundle, of items. A choice k ∈ K defines a feasible allocation of bundles to agents. Introduce
variables, xi ( S ) ∉ {0,1} , to indicate that agent i receives bundle S in a particular allocation. Agent
i has value vi ( S ) for bundle S . Assume for the purpose of exposition that we have knowledge of
agent valuations. The CAP can be formulated as the following integer programming problem.

max ∑ ∑ x i ( S )vi
xi ( S )
S ⊆ G i∈I

s.t.
∑ x (S ) ≤ 1,
S
i ∀i ∉ I

∑∑ x (S ) ≤ 1,
S∋ j i∈I
i ∀j ∉ G

xi ( S ) ∈ {0,1}
where, S ∋ j indicates that bundle S contains item < j. To apply linear-programming duality theory
we must relax this IP formulation, and construct an integral LP formulation. Consider [LP1] in which
the above equation is relaxed to xi ( S ) ≥ 0 . Then, the dual is simply written as:

min
πi , p( j )
∑ π +∑ p( j )
i
i
j

s.t.
π i + ∑ p( j ) ≥ vi ( S ), ∀i ∈ I , ∀S ⊆ G
j∈S

π i , p( j ) ≥ 0, ∀i, j

124
< The dual introduces variables for items p ( j ) ≥ 0 which we can interpret as prices on items. Given
prices, p ( j ) , the optimal dual solution sets π i = max S {vi ( S ) − ∑ p( j ), 0} . This is the maximal
j∈S
payoff to agent i, given the prices. The dual problem computes prices on items to minimize the sum of
the payoffs across all agents. These are precisely a set of CE prices when the primal solution is
integral. The complementary-slackness (CS) conditions on a feasible primal, x i ( S ) , and feasible
dual p ( j ) , solution define conditions for competitive equilibrium:

π i > 0 ⇒ ∑ x i ( S ) = 1 ∀i
S

p ( j ) > 0 ⇒ ∑∑ x i ( S ) = 1, ∀j
S∋ j i∉I

x i ( S ) > 0 ⇒ π i + ∑ p( j ) ≥ vi ( S ) ∀i, ∀S
j∈S

These conditions have a natural economic interpretation. First and third conditions state that the
allocation must maximize the payoff for every agent at the prices. The second condition states that the
seller must sell every item with a positive price, and maximize the payoff to the seller at the prices.
The prices are said to support the efficient allocation. A seller can announce an efficient allocation
and CE prices, and let every agent verify that the allocation maximizes its own payoff at the prices. In
practice we will need an auction to provide incentives for agents to reveal the information about their
valuations, and to converge towards a set of CE prices.

10.6. Formulating the Winner Determination Problem

In the context of auctions, the problem of computing an allocation is often referred to as the winner-
determination problem. In simple designs (such as the English, Vickrey etc.) where only a single
winner is permitted in the allocation, the optimization problem can be solved in a straightforward
fashion. However, in settings where the allocation rule permits multiple winners, the optimization
problem that needs to be solved can become quite computationally complex depending on the market
and bid structures. In this section we outline the different settings and the associated complexity of the
winner determination problem.
Example: A Combinatorial Reverse Auction
Auctions can be hosted for multiple heterogeneous items, where multiple items can be auctioned
together. This is known as combinatorial auction problem. Combinatorial auctions are also proposed
for procurement problems in markets with one buyer and multiple sellers. The reverse combinatorial
auction is formulated as a set covering problem rather than a set packing problem. An interesting (and
complicating) issue that arises in this setting is that there are various business rules that are used to
constrain the choice of winners. These business rules appear as side constraints in the winner
determination problem.
Formulation: Let G = (1,.., N ) denote the set of items for sale. The bidders are allowed to specify
bundles S ⊆ G with a single price on the entire bundle and submit bids for multiple bundles. We
formulate this problem by introducing a decision variable x i ( S ) for each bundle S offered by
bidder i . Each bidder provides a bid set B i ⊆ 2G Let p i ( S ) denote the price offered by agent i for
bundle S . The winner determination problem with no side constraints can be formulated as follows:

125
 m in
xi ( S )
∑∑
S ∈ Bi i
x i (S ) p i (S )

s .t.
∑
S ∈ Bi
x i ( S ) ≤ 1, ∀i

∑∑
S ∈ Bi S ∋ j
x i ( S ) ≥ 1, ∀j

x i ( S ) ∈ {0 ,1} , ∀ i, S

This is posed as a cost minimization problem with a demand covering constraint. In this formulation
the problem is to procure a single unit of each good, but this can be generalized by increasing the
RHS of the first set of constraints.
Introducing side constraints: In a real world setting there are several considerations beside cost
minimization. These considerations often arise from business practice and/or operational
considerations and are specified as a set of constraints that need to be satisfied general, the specific
form of these side constraints depends on the market structure. Some of these constraints are:
Number of Winning Suppliers: An important consideration while choosing winning bids is to make
sure that the entire supply is not sourced from too few suppliers, since this creates a high exposure if
some of them are not able to deliver on their promise. On the other hand, having too many suppliers
creates a high overhead cost in terms of managing a large number of supplier relationships. These
considerations introduce constraints on the minimum, LS , and maximum, U S , number of winning
suppliers in the solution to the winner determination problem.
yi ≤ ∑
S ∈ Bi
x i ( S ) ≤ Ky i , ∀ i ∈ N

LS ≤ ∑y i
i ≤US

Budget Limits on Trades: A common constraint that is often placed is an upper limit on the total
volume of the transaction with a particular supplier. These limits could either be on the total spend or
on the total quantity that is sourced to a supplier. These types of constraints are largely motivated (in a
procurement setting) by considerations that the dependency on any particular supplier is managed.
Similarly, often constraints are placed on the minimum amount or minimum spend on any transaction,
i.e. if a supplier is picked for sourcing then the transaction should be of a minimum size. Such
constraints reduce the overhead of managing a large number of very small contracts.
Marketshare Constraints: Another common consideration, especially in situations where the
relationships are longterm, is to restrict the market share that any supplier is awarded. The motivations
are similar to the previous case.
Reservation Prices: A reservation price allows the buyer to place an additional constraint on the most
she will pay for some items. This can arise, for example, due to a fall-back option such as an external
commodity market.
References
[1] Jayant Kalagnanam and David C. Parkes, Auctions, Bidding and Exchange Design,
In Handbook of Quantitative Supply Chain Analysis: Modeling in the E-Business Era, David
Simchi-Levi, S. David Wu, and Max Shen (eds.), Chapter 5, Kluwer, 2004.
http://www.eecs.harvard.edu/econcs/pubs/ehandbook.pdf
[2] MIT Open Course Ware: Auction Theory, http://ocw.mit.edu/NR/rdonlyres/ Engineering-
Systems-Division/ESD-260JFall2003/2CECCCEB-0165-42A3-B86A-
B4BBA5A6930B/0/l18ch22auctheory.pdf
[3] Roger L. Zhan and Zuo-Jun Max Shen, Optimality and Efficiency in Auctions Design: A Survey
(With Z. Shen, to appear in Pareto Optimality, Game Theory and Equilibria, A. Migdalas, P.M.
Pardalos, and L. Pitsoulis, eds, Springer (2006), http://plaza.ufl.edu/zhan/paper/zhanShenV2.pdf

126
 CHAPTER XI
Technologies for Supply-Chain Integration

11.1 Introduction
Information and communication technology (ICT) has emerged as the key enabler of supply chain
integration. Specifically, the businesses can use the Internet to gain global visibility across their
extended network of trading partners and help them respond quickly to a range of variables, from
customer demand to resource shortages. This chapter shall discuss the existing and emerging
technologies for supply chain integration.

11.2. B2B integration

B2B integration or B2Bi provides a technology framework for supply chain integration. It promises to
dramatically transform the way business is conducted between partners, suppliers and customers. All
companies (i.e., large, medium, small and new) can experience increased growth and success through
tightly integrated partnerships. Companies from across a variety of industries from consumer
packaged goods, high technology products, logistics, pharmaceuticals to Chemical, manufacturing
and financial services, are embracing B2Bi.
They are realizing the enormous competitive advantage of B2Bi, through faster time to market,
reduced cycle times and increased customer service. Through integration of business and technical
processes, companies are able to strengthen relationships with service partners and customers, achieve
seamless integration inside and outside the enterprise, gain real-time views of customer accounts,
increase operational efficiencies and reduce costs.

B2B integration is a challenging task because of the following factors:

• Information formats are becoming more diverse.
• The information space is large and dynamic.
• Semantic integration of data is far from resolved.
• Most systems are autonomous (i.e. their architecture is not centralized).
• The integration needs to be simple, fast, secure and adaptable

Business-to-business integration and enterprise application integration (EAI) are the top priorities of
companies these days. B2Bi requires exchange of data and sharing of business processes across
multiple trading partners, such as buyers, suppliers and distributors. EAI, on the other hand, requires
internal applications, such as CRM, ERP and legacy systems, to interact with each other
seamlessly.B2Bi and EAI are accomplished by data, application and business process integration. The
integration challenges in both B2Bi and EAI have a lot in common and can be overcome through a
single, integrated solution.
 Fig 1: B2B Integration at a glance

Fig 2. B2Bi and EAI integrate all internal and external applications

11.3. Middleware technologies for B2B integration

Integration can be conducted at the various layers of an e-commerce system, using various
Middleware technologies. Middleware is defined as a set of common business-unaware services that
enable applications processes (i.e. components) and end users to interact and interoperate with each
other across a network. In essence, middleware is the software that resides above the network and
below the business-aware application software. The ideal middleware should mask out the differences
in network protocols, operating systems, platforms and programming languages in addition to the

128
quality attributes such as performance, reliability, safety, security and real-time issues. In other terms,
the perfect middleware would make the process and the content connectivity among dispersed
autonomous systems as if they are components of the same systems.

Fig 3: B2B Interoperability layers

The B2B Integration Framework provides integration at the following levels:

i. Communication Layer (Transport) is concerned with the exchange of messages among partners.
Multiple protocols and frameworks have been developed that varies from network level of
communication to the distributed objects frameworks. This layer’s interoperability objective is to
provide independence from such protocols and frameworks by translating and converting messages
between heterogeneous protocols
ii. Content Layer (Data) resolves semantic and structural heterogeneity issues. For example, it
determines if a document represents a purchase order or a request for a quote or a product description
etc. Structural differences arise from the use of diverse information formats. Semantic differences
come from different interpretations of the same concept. For example, a data item called “price” can
mean a price that includes or excludes tax. Therefore, this layer’s interoperability objective is to
provide independence from data models, formats and languages. Solutions are based on information
translation and integration, Examples include wrappers and mediators.
iii. Business Process Layer (Process Flows) deals with the semantics of interactions that correspond
to joint business processes. For example, the following steps constitute a joint business process: send
order, process order, deliver product, and make payment. This layer resolves issues such as what is
the meaning of a message, what actions are allowed, what responses are expected, etc. Therefore, the
layer’s interoperability objective is to allow transparent peer-to-peer interactions with any partners.
This is a very difficult problem. Amongst potential solutions are: Application Programming Interface
(API), Document-based solutions and Workflow-based solutions.

Each of these three layers is discussed in detail below:

129
11. 3.1 Communication (transport) level middleware
The term low-level middleware refers to the integration process that is based on the lowest levels of
interconnectivity. Two concepts are included at this level: the Network protocols that are used to
transfer raw data and the remote procedure call (RPC) as first form of distributed computing
framework.

11.3.1.1 Network-protocol level

Early distributed systems were built using few available options to achieve interconnectivity. Those
options are using relatively low-level programming and more related to the underlying network
protocols. Distributed systems that are built directly over the TCP/IP protocol-suite, have considered
these protocols as an underlying middleware. TCP/IP enables the communication of different systems
across many operating systems, platforms and programming languages as it is defined for a diverse
set of operating systems and platforms. APIs are also defined for many programming languages. Yet
this approach needs tremendous programming skills and knowledge of low-level issues of the
network. Therefore, it is time consuming, cumbersome and can be error-prone as the programmer has
to reinvent the wheel in every program code.

11.3.1.2 Object and component-based middleware

Object Oriented and component-based middleware have become the leading methodologies in
developing distributed systems. The different types of middleware presented in this section are best
suited for the centralized architecture for B2B distributed systems because they support tightly
coupled connectivity. The nature of tightly coupled connectivity of this middleware mandates that the
interconnected components be aware of each other object/component interface, therefore, it is not a
good choice for inter-enterprise connectivity.
Object Oriented (OO) development approach has emerged steadily as a paradigm that focuses on
granularity, productivity and low maintenance. Unlike the procedural paradigm, which is still
endorsed by some developers, the OO has proven itself to be an effective development method. This
is because it represents real-time entities as objects, therefore, reducing the gap between real-world
and software concepts. Nevertheless, OO is not the silver bullet for software development because not
everything can be modeled as an object. There should be a more granularity level that allows a
number of objects to cooperate to perform an atomic task and to be used as a plug-and- play
component in software. The ambition of the component development paradigm is to make the
construction and software as simple as possible.
A component is an independent delivery piece of functionality presented as a black-box that provides
access to its services through a defined interface. Objects and components have been in use in the
distributed computing in such a way that an object or a component can invoke another remote object
or component functionality through their interfaces. This requires a middleware environment that
implements the communication issues of objects/components interaction. Two variations of such
middleware are: Java-based (RMI and EJB) and OMG CORBA.
The platform-independent Java language has now become the dominant language for developing
networked systems. The early maturity and stability of Java has led to the development of Java-based
frameworks for facilitating the B2B distributed systems development. There are two major Java
frameworks: RMI which is developed on top of the TCP/IP and the EJB which is developed on top of
the RMI. EJB provides a framework for “plugging in” server components into a larger server
application, thereby extending that application’s functionality.
It is intended to hide the low-level system details of managing server resources such as transaction
processing, threads management and databases access. It also helps in clearly identifying the

130
responsibilities of the client, the server and all other individual components. EJB deals with issues
such as scalability, replication, distributed processing, deployment, security, or transactions.

CORBA: One of the most important standards that have emerged from the Object Management
Group (OMG) is the Common Object Request Broker Architecture (CORBA). The CORBA’s
infrastructure provides mechanisms to deal with platform heterogeneity, transparent location and
implementation of objects, interoperability and communication between software components of a
distributed object environment. CORBA uses a language-independent Interface Definition Language
(IDL), and is therefore more interoperable than Java RMI. However, Java RMI and CORBA
distributed objects can be integrated through the Java RMI versions that run on top of the Internet
Inter-ORB Protocol (IIOP). CORBA can also be integrated with the EJB because EJB specification
does not dictate the protocol used to communicate with an EJB.

11.4. Content level middleware

A data transfer format defines how the data should be formatted in order for different components, not
only within the same distributed business system, but also across different systems resides in different
companies and technologies, to exchange messages, data, and even documents that they can interpret
and understand. Both sender and receiver should agree initially on the shared format. The sender
component usually has software (i.e. embedded component) that encodes the contents of the message
into a well-defined format that is sent using the middleware protocols to remote object/component.
This receiving component interprets the message to retrieve the original application data.
Typical distributed systems components agree on specific data format that each component uses to
encode and interpret sent and received data. Having non- standard data formats (incompatible data
format) would not allow different components in different system to interact unless extra adapter
software is provided to convert between the two different data formats. This difficulty becomes very
clear when considering distributed systems integrated with the web. In most situations, the web
browser would act as a client that should be able to use the same data format of the system it is
interacting with. Having a standard data format would allow new business systems not to bother about
how their web clients would interpret exchanged data.
Two major content level middleware are: Electronic Data Interchange (EDI) and a number of XML-
based frameworks.

11.4.1 Electronic Data Interchange (EDI) based integration frameworks

Electronic Data Interchange (EDI) is a set of standards for structuring information to be
electronically exchanged between and within businesses, organizations, government entities and other
groups. The standards describe structures that emulate documents, for example purchase orders to
automate purchasing. The term EDI is also used to refer to the implementation and operation of
systems and processes for creating, transmitting, and receiving EDI documents.
EDI offers the prospect of easy and cheap communication of structured information throughout the
corporate community, and is capable of facilitating much closer integration among hitherto remote
organizations. A more careful definition of EDI is 'the exchange of documents in standardized
electronic form, between organizations, in an automated manner, directly from a computer application
in one organization to an application in another. Despite being relatively unheralded, in this era of
technologies such as XML services, the Internet and the World Wide Web, EDI is still the data format
used by the vast majority of electronic commerce transactions in the world.

131
11.4.1.1 Architecture for EDI
The essential elements of EDI are:
• The use of an electronic transmission medium (originally a value-added network, but
increasingly the open, public Internet) rather than the dispatch of physical storage media such as
magnetic tapes and disks;
• The use of structured, formatted messages based on agreed standards (such that messages can
be translated, interpreted and checked for compliance with an explicit set of rules);
• Relatively fast delivery of electronic documents from sender to receiver (generally implying
receipt within hours, or even minutes); and
• Direct communication between applications (rather than merely between computers).
EDI depends on a moderately sophisticated information technology infrastructure. This must include
data processing, data management and networking capabilities, to enable the efficient capture of data
into electronic form, the processing and retention of data, controlled access to it, and efficient and
reliable data transmission between remote sites.
A common connection point is needed for all participants, together with a set of electronic mailboxes
(so that the organizations' computers are not interrupted by one another), and security and
communications management features. It is entirely feasible for organizations to implement EDI
directly with one another, but it generally proves advantageous to use a third-party network services
provider.

11.4.1.2 Benefits of EDI

EDI's saves unnecessary re-capture and re-keying of data. This leads to faster transfer of data, far
fewer errors, less time wasted on exception-handling, and hence a more stream-lined business
process. EDI also eliminates other paper-handling tasks.
Benefits can be achieved in such areas as inventory management, transport and distribution,
administration and cash management. EDI offers the prospect of easy and cheap communication of
structured information throughout the government community, and between government agencies and
their suppliers and clients.
EDI can be used to automate existing processes. In addition, the opportunity can be taken to
rationalize procedures, and thereby reduce costs, and improve the speed and quality of services.
Because EDI necessarily involves business partners, it can be used as a catalyst for gaining
efficiencies across organizational boundaries by reducing business cycles. This strategic potential
inherent in EDI is expected to be, in the medium term, even more significant that the short-term cost,
speed and quality benefits.
EDI can positively impact customer service factors such as incidence of errors and timeliness. Other
advantages accrue to organizations that utilize EDI as an opportunity to rationalize operations via
business process.

11.4.1.3 Limitations of EDI

EDI is an established technique for B2B integration. When assessing its role in the various
interoperability layers, EDI is more focused on communication interoperability where VANs are used
to handle message delivery and routing. EDI standards also provide a single homogeneous solution
for content interoperability, but the set of supported document types is limited. This means that EDI is
limited to enable a rich set of possible B2B interactions. In addition, EDI standards, as currently
defined, do not support interoperability at the business process level

132
11.4.1.4 Translating the benefits of EDI into a positive ROI
The benefits of implementing EDI can be both strategic and operational. As an example of the
strategic benefits of EDI, the Gartner Group reports that gross sales for a supplier firm to a major
retailer increased 18% after the company implemented an EDI program (Gartner Group, 1996).
• Administrative cost reduction: Studies suggest that it costs an average of Rs.30 to manually
process a customer order, whereas implementing an EDI-enabled order entry program would
reduce that cost to Rs.10 per customer order.
• Personnel Reduction: Studies have shown as much as a 50% reduction in required staff can
result from a fully functional EDI implementation (Price Waterhouse, 1995)
• Cycle time reduction: In a case study involving the Pfaltzgraff Company (a supplier) and Best
Products (a retailer), a 50% reduction in order cycle time was achieved after implementing an EDI
program. (Gartner Group, 1995).
• Inventory Reduction The Gartner Group (1995) reports an average inventory reduction of 10%
at EDI-enabled manufacturing firms.
• Cash flow improvement: Decreased operating expenses and improved accuracy in procurement
areas are directly beneficial to a firm's financial cash flow. Although this area is difficult to
quantify, most accounting departments can attest to the numerous benefits of improved corporate
cash flow

9.4.2 XML-based frameworks

The Extensible Markup Language (XML) is an emergent set of open standards in the production and
consumption of content managed by the World Wide Web Consortium. XML is a data oriented
technology, based on a lightweight subset of the Standard Generalization Markup Language (SGML),
suitable for the definition, storage and retrieval of structured data.
XML is inherently language independent. XML brings with it a huge promise in the same way Java
did for portability of code; XML claims to do for portability of data. Sun has even been touting the
slogan: “Java + XML = Portable Code + Portable Data”. XML is a compromise between the strong
rigid structure of databases and the weak semantics and comparative freedom of the HTML. It has
been devised with the objective to facilitate sharing of data among applications, both across platforms
within the same enterprise and sharing data between enterprises.

11.4.2.1 The Role of Extensible Markup Language (XML) in B2Bi

XML has become the lingua franca of the B2B e-business revolution. It has created a mechanism to
publish, share, and exchange data using open standards over the Internet. There is no doubt that in the
future XML will be used in each and every B2B application.
XML is not, however, an integration solution in itself - it is just a data definition language. Without
global XML standards there can be no seamless business among companies spread out all over the
world. These standards are a common set of industry-specific definitions representing business
processes. For XML messages to be interpreted by all companies participating in B2Bi they need to
agree on a common XML-based B2B standard, which will define the document formats, allowable
information, and process descriptions.
The need for industry-wide B2B e-commerce standards in vertical industries is becoming increasingly
critical and obvious. Several organizations have been working to define these market-segment-
specific definitions. Standards and groups such as RosettaNet, CIDX, and OASIS are making it
possible for companies to share information with one another without having to completely reengineer
their internal applications. These standards will automate the flow of information across all companies
within a given industry, independent of the underlying software or hardware infrastructure supporting
the activities related to these transactions

133
Inspired by the promising future of being the standard format for data transfer and object
communication in the distributed systems, many special purpose frameworks are built around XML
making use of its extensibility features.

The Electronic Commerce(eCO) framework aims to address the problem of interoperability by

providing standard facilities for businesses to discover each other, determine possible trading
relationships, determine capabilities of trading partner systems, and establish trading relationships
regardless of the e-commerce standards and protocols.
Commerce XML (cXML) targets non-strategic transactions which are Maintenance, Repair and
Operations (MRO) of materials, office supplies, laboratory supplies, etc. It is a simplified, XML and
Internet-based version of EDI.
BizTalk’s approach to applications interoperation is based on leveraging several existing standards
and technologies including the SOAP, XML and Multipurpose Internet Mail extensions (MIME).
RosettaNet provides a set of XML-based standard interfaces for supply chain management in
information technology and electronic component industry.
The Electronic-Business-XML (ebXML) vision is to create a single global electronic marketplace
where enterprises of any size and in any geographical location can meet and conduct business with
each other through the exchange of XML based messages.

11. 4.2.2 Advantages of XML

• It is a simultaneously human and machine-readable format;
• It is platform independent
• It supports Unicode, allowing almost any information in any written human language to be
communicated;
• It can represent the most general computer science data structures: records, lists and trees;
• Its self-documenting format describes structure and field names as well as specific values;
• The strict syntax and parsing requirements make the necessary parsing algorithms extremely
simple, efficient, and consistent.
• XML is heavily used as a format for document storage and processing, both online and offline,
• It is based on international standards;
• It allows validation using schema languages such as XSD and Schematron, which makes effective
unit-testing, firewalls, acceptance testing, contractual specification and software construction
easier;
• The hierarchical structure is suitable for most (but not all) types of documents;
• It manifests as plain text files, which are less restrictive than other proprietary document formats;
• Its predecessor, SGML, has been in use since 1986, so there is extensive experience and software
available.

134
11.4.2.3 Disadvantages of XML
• XML syntax is redundant or large relative to binary representations of similar data. The
redundancy may affect application efficiency through higher storage, transmission and processing
costs.
• XML syntax is too verbose relative to other alternative 'text-based' data transmission formats.
• No intrinsic data type support:XML provides no specific notion of "integer", "string", "boolean",
"date", and so on.
• The hierarchical model for representation is limited in comparison to the relational model or an
object oriented graph.
• Expressing overlapping (non-hierarchical) node relationships requires extra effort.
• XML namespaces are problematic to use and namespace support can be difficult to correctly
implement in an XML parser.

11.5. Business process middleware

Business processes are descriptions of the activities required by an organization to fulfill its mission,
such as, a business contract or satisfying a specific customer request. Gaining control of these
processes allows an organization to reengineer and improve each process or adapt them to changing
requirements. A business process is defined as a set of one or more linked procedures or activities
which collectively realize a business objective or policy goal, normally within the context of an
organizational structure defining functional roles and relationships. A few approaches have emerged
for the purpose of facilitating inter-business process interoperability. Among these approaches are:
• Application Programming Interface (API): consists of working out business processes offline,
determining the overall connection and coordination of operations and then defining universally-
agreed abstract interfaces which provide remote operation invocations and connectors for back-end
systems. Middleware and database technologies are then used for mapping these abstract interfaces to
physical implementations. Example of such approaches is CORBA-based solutions.
• Document-based solutions: a set of documents is exchanged according to a protocol. There is no
prior agreement as partners publish their documents independently. Each document is self-describing
and contains enough information about the business process involved. Examples include EDI,
BizTalk, eCO and RosettaNet.
• Traditional workflow systems are based on the promise that the success of an enterprise requires
the management of business processes in their entirety. Indeed, an increasing number of organizations
have already automated their internal process management using workflows and enjoyed substantial
benefits in doing so. Current business processes within an organization are integrated and managed
either using ERP systems such as SAP/R3, Baan, People-Soft or various workflow systems like
IBM’s MQ Series Workflow or integrated manually on demand- basis.
• The Web Services model is emerging steadily as a loosely- coupled, document-based integration
framework for Internet-based applications. Unlike traditional workflows, Web services support inter-
enterprise workflows where business processes across enterprises can interact in a loosely coupled
fashion by exchanging XML document-based messages. The Web Services model idea is to break
down web accessible applications into smaller services.

11.5.1 Web Services

"Web Services provide a simplified mechanism to connect applications regardless of the technology
or devices they use, or their location. They are based on industry standard protocols with universal
vendor support that can leverage the internet for low cost communications, as well as other transport

135
mechanisms. The loosely coupled messaging approach supports multiple connectivity and information
sharing scenarios via services that are self describing and can be automatically discovered."
These services are accessible through electronic means, namely the Internet. They are self-describing
and provide semantically well-defined functionality that allows users to access and perform the
offered tasks. Such services can be distributed and deployed over a number of Internet-connected
machines.

Fig 4: Web services standards

A service provider is able to describe a service, publish the service and allow invocation of the
service by parties wishing to do so. A service requester may request a service location through a
service broker that also support service search. Web services are loosely coupled allowing external
applications to bind to them. Web services are also reusable allowing many different parties to use
and reuse a service provided.
As shown above there are three emerging standards for web services:
1. Simple Object Access Protocol (SOAP) which is an XML-based protocol for exchanging
document-based messages across the Internet.An XML-based, extensible message envelope format,
with "bindings" to underlying protocols. The primary protocols are HTTP and HTTPS, although
bindings for others, including SMTP and XMPP, have been written.
2. Web Services Description Language (WSDL), which is a general purpose XML-based language
for describing the interface, protocol bindings and the deployment details of Web services, Typically
used to generate server and client code, and for configuration.
3. Universal Description, Discovery and Integration (UDDI) which refers to a set of specifications
related to efficiently publishing and discovering information about Web services, to enable
applications to find Web services, either at design time or runtime.

In conclusion, Web Services model is trying to use XML as basis to re-invent or improve the
integration model by using SOAP at the communication layer and inter-enterprise workflow at the
business process layer.

11.5.2 Essential Features of a B2B Integration Solution

1) Firstly, the integration solution should be able to enable any transaction, any time - end-to-end
and partner-to-partner. It should be able to fully automate real-time exchange of data between
disparate applications.
2) Secondly, the solution should be able to conduct all transactions securely, maintain audit logs, etc.

136
3) Thirdly, the solution should support diverse sets of file formats, protocols, and security standards.
4) Fourthly, the solution should be based on open standards that allow a company and its partners to
send transactions using any combination of applications and file formats, telecommunication
pathways, communication protocols and B2B protocols, and XML standards such as RosettaNet,
ebXML, OAG, Biztalk, OBI, etc. The solution should also provide support for Web Services.
5) Lastly, the solution should be scalable, that is, companies should be able to scale it horizontally
and vertically. Further, it should offer robust load balancing features, critical to the success of
large applications.
A few leading B2Bi solutions include: IBM MQSeries Integrator; Extricity; BEA eLink; webMethods
B2B Enterprise; NEON eBusiness Integration Servers; Vitria BusinessWare; and Microsoft BizTalk
Server.

11.5.3 Web Services as the essential traits of B2B applications

11.5.3.1 Distributed Transaction Management:

It is very tough to maintain distributed transaction control even within disparate systems and
applications within an enterprise. B2B transactions may be spread over disparate systems and
applications across different enterprises, making them several times more difficult to maintain and
control. In their current state, Web Services are not transactional in nature and provide basic
"request/response" functionality.

11.5.3.2 Security:
B2Bi requires two levels of security. Firstly, B2Bi necessitates opening up corporate firewalls to
enable cross boundary communication between enterprises. Thus, whatever mode of integration is
used, companies have to secure their internal network against malicious attacks through these open
ports.
Secondly, the data transmitted over dedicated leased lines, such as EDI, Internet, or any other mode,
has to be secured. The data may contain classified information, such as corporate information and
business transaction information, and thus cannot be left unguarded. In their current state, Web
Services lack broad support and facilities for security. Thus, Web Services based B2Bi architecture
may potentially have big security loopholes.

11.5.3.3 Dynamic
For companies to participate in true dynamic business with other companies, integration between the
systems of the two companies has to happen in real-time. Further, this integration is only possible if
B2Bi is done using open standards over the Internet.
Web Services do provide a dynamic approach to integration by offering dynamic interfaces. Web
Services are based on open standards such as UDDI, SOAP, and HTTP, and this is probably the single
most important factor that would lead to the wide adoption of Web Services for B2Bi.

11.5.3.4 Integration Mode

The integration mode or pattern is the most important element of B2B integration. Is the B2Bi data-,
business process-, application-, function-, or portal-oriented? The answer to this question determines a
lot of answers involved in the modalities and technology used for B2Bi. Typically in B2B integration,
companies involved take a joint decision based on the technology available in-house, budgets, and
level of synchronization needed to support business functionalities.

137
In this generation of Web Services, it is possible to achieve only function level integration between
applications The next generation of Web Services, however, will be functionally and technologically
advanced, offering user interface encapsulation and security. They will be able to package an
application and embed it into another application.

11.5.4 Example of Web Services for B2Bi

The following diagram (Fig 5) shows an example of using Web Services in a B2Bi scenario. In this
example, the corporate procurement application running within an application server requests quotes
from multiple vendors. The procurement application of the buyer gets information about Web
Services offered by suppliers using a private UDDI registry and invokes these services over the
Internet to get quotes for a specific item.

Fig 5: B2B Use of Web services for procurement

The sequence of steps is as follows:

1. The Buyer's procurement application, running within an application server, has to generate a
purchase order for a specific item.
2. The procurement application gets information about Web Services of different suppliers for that
specific item by doing a look up in the private UDDI registry.
3. The location of and WSDL binding information for the Web Services is sent to the procurement
application.
4. The application invokes the Web Services published by the suppliers to get quotes for that item.
The communication is based on SOAP over the Internet.
5. The application receives quotes from different suppliers. The communication is based on SOAP
over the Internet.
6. The information is then analyzed, leading to the creation of the purchase order.

11.5.5 EAI and Web Services

As companies move in the direction of collaborative business-to-business e-commerce, they will first
have to look inward to their own internal systems, applications and processes. Several business
processes span across multiple internal applications. These applications must be able to communicate
dynamically in real-time before a company can effectively e-communicate with the outside world.

138
B2Bi requires exchange of data and sharing of business processes across multiple trading partners,
such as buyers, suppliers and distributors. EAI, on the other hand, requires internal applications, such
as CRM, ERP and legacy systems, to interact with each other seamlessly.
Most companies have an environment of disparate legacy systems, applications, processes, and data
sources, which typically interact by a maze of interconnections that are poorly documented and
expensive to maintain. Web Services are not EAI in and of themselves. Rather, Web Services are just
another technology that enables EAI, and it can significantly change the traditional point-to-point
approach.
Using Web Services that loosely integrate applications, a company achieves just a subsection of EAI.
EAI, on the other hand, takes a complete holistic approach of tightly integrating and connecting all
applications and systems that support a company's business. EAI takes years of continued
commitment and effort from different business and technical units within the company, high
investment, and substantial resources.
The current EAI solutions that predominately focus on integrating applications will have to be
changed significantly, as packaged applications in the future will expose their functions as services
using technologies such as XML, SOAP, and UDDI. Thus, the EAI solutions will have to provide a
broad support for service integration rather than application integration.

11.6. Conclusion
B2B integration is the pervasive enabler of most current business strategies such as collaborative e-
commerce, collaborative networks, supply chain management (SCM) and customer relationship
management (CRM) across multiple channels of delivery including wireless devices and the Internet.
B2Bi strategy should be laid out and executed in such a way so as to: have an integrated, real-time
application-to-application, system-to- system interaction with all the existing and new trading
partners; eliminate all manual steps in business processes; conduct secure and real-time commerce
transactions over the Internet; have the flexibility to accommodate the different mode of interactions
of each partner; and, finally, have the ability to adapt to change quickly and easily in this dynamic age
of B2B collaborative e-commerce. This is what B2Bi is all about — the end-to-end automation and
integration of cross-organization business processes, data, applications and systems.
Web Services certainly have the potential of redefining the whole paradigm of B2B integration by
making it truly dynamic, easily implemented in a modular fashion, and in the longer run being
cheaper. The application of Web Services for B2Bi, however, will be limited if services for
authentication, encryption, access control, and data integrity are not available. Web Services
intermediaries that provide services such as UDDI repository hosting, security services, quality
assurance of Web Services, performance checks, etc., will have a big role to play in the B2Bi space.

References
[1] Feras T. Dabous, Fethi A. Rabhi, Pradeep K. Ray and Boualem Benatallah.Middleware
Technologies for B2B Integration. The International Engineering Consortium (IEC) Annual
Reviews of Communications, IEC Press, USA, Vol 56,July, 2003
[2] Christoph Bussler, The Role of B2B Engines in B2B Integration Architectures, , SIGMOID
Record, Vol 31,No. 1, March 2002
[3] Boualem Benatallah, Olivier Perrin, Fethi Rabhi, Claude Godart: Web Service Computing:
Overview and Directions. Book chapter. In Handbook of Innovative Computing. Editor: Albert
Y. Zomaya. Springer, 2005.
[4] Bussler C, B2B Integration technology architecture,.; Fourth IEEE International Workshop on
Advanced Issues of E-Commerce and Web-Based Information Systems, 2002. (WECWIS
2002). Proceedings.

139
[5] Jones R., B2B Integration, IET Manufacturing Engineer Vol 80,Issue 4,Aug 2001, Pages 165-
168
[6] http://en.wikipedia.org/
[7] http://www.webservicesarchitect.com/content/articles/samtani02.asp
[8] http://www.worldscibooks.com/business/etextbook/p263/p263_chap1.pdf
[9] http://www.webservicesarchitect.com/content/articles/samtani01.asp
[10] http://www.msc-inc.net/Documents/EDI_roi.htm
[11] http://www.webservicesarchitect.com/content/articles/samtani07.asp

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

140
 CHAPTER 12
Security and Payment Issues in Integrated Supply Chain

12.1. Introduction
In today’s Internet world, it is relatively easy to create, alter and transmit information. The
advancement in computing capacity and interconnectivity has presented a situation where small
efforts can cause potentially large losses. Both accidental and intentional breaches are easier and more
likely. This is a major challenge to businesses that want to take advantage of the current information
technology. Concern for information security is fairly widespread. Those in banking, health care,
finance, and telecommunications rate information security as the highest business priority, with
retailers a little less concerned. In every sector, security is regarded as a key business driver.

12.2 Need for e-Business security

There is almost an uncountable number of ways that an e-business setup could be attacked by hackers,
crackers and disgruntled insiders. Common threats include hacking, cracking, masquerading,
eavesdropping, spoofing, sniffing, Trojan horses, viruses, bombs, wiretaps, etc. While the list of
actual manifestation is long, conceptually, they break down to a few categories. These are spoofing,
unauthorized disclosure, unauthorized action, and data alteration. From a business perspective Denial
of Service (DoS) attacks appear to be the most serious threat. DoS attacks consist of malicious acts
that prevent access to resources that would otherwise be available. Even though data may not be lost,
the financial losses that could be incurred from not being able to supply a service to customers could
be of much higher value.
In conducting e-business, every organization ought to be able to:
• Positively identify or confirm the identity of the party they are dealing with on the other end of the
transaction;
• Determine that the activities being engaged in by an individual or machine is commensurate with
the level of authorization assigned to the individual or machine;
• Confirm the action taken by the individual or machine and be able to prove to a third party that the
entity (person or machine) did in fact perform the action;
• Protect information from being altered either in storage or in transit;
• Be certain that only authorized entities have access to information;
• Ensure that every component of the e-business infrastructure is available when needed;
• Be capable of generating an audit trail for verification of transactions.

12.3 e-Business Security Categories

Effective information security policy must have the following six objectives: confidentiality; integrity;
availability; legitimate use (identification, authentication, and authorization); auditing or traceability;
and non-repudiation. If these objectives could be achieved, it would alleviate most of the information
security concerns. Each information security objective is discussed below with emphasis on the
specific challenges it poses to Internet mediated businesses.

12.3.1 Confidentiality: Confidentiality deals with protecting the content of messages or data
transmitted over the Internet from the unauthorized people. For example, it is essential for you to
protect your credit card number from the hackers. Besides other e-business setups, concern for
confidentiality is also major concern in healthcare, insurance, and banking industry. To maintain the
confidentiality of Web users’ information, organizations have to find ways to keep the information
from unauthorized view. From an operational point of view, that means information that is stored has
to be secured in a way that it can only by accessed by authorized parties. Similarly, information in
transit has to be kept from the view of unauthorized parties and that it is retrieved only by a legitimate
entity.
12.3.2 Integrity: Integrity is related to preventing data from being modified by an attacker.
Transmitting information over the Internet (or any other network) is similar to sending a package by
mail. The package may travel across numerous trusted and untrusted networks before reaching its
final destination. It is possible for the data to be intercepted and modified while in transit. This
modification could be the work of a hacker, network administrator, disgruntled employee, government
agents or corporate business intelligence gatherer; it could also be unintentional.
12.3.3 Availability: Availability means that systems, data, and other resources are usable when
needed despite subsystem outages and environmental disruptions. Lack of availability is essentially
loss of use. The most commonly known cause of availability problems is Denial of Service (DoS)
attacks even though there are other common causes such as outages, network issues, or host problems.
The goal is to ensure that system components provide continuous service by preventing failures that
could result from accidents or attacks. From a security point of view, availability is enhanced through
measures to prevent malicious denials of service. Closely related to availability and very important to
e-businesses are reliability and responsiveness. Reliability implies that a system performs
functionally as expected. Responsiveness is a measure of how quickly service could be restored after
a system failure. In other words, it is a measure of system survivability.
12.3.4 Legitimate use: Legitimate use has three components: identification, authentication and
authorization. Identification involves a process of a user positively identifying itself (human or
machine) to the host (server) that it wishes to conduct a transaction with. The most common method
for establishing identity is by means of username and password. The response to identification is
authentication. Without authentication, it is possible for the system to be accessed by an impersonator.
Authentication needs to work both ways: for users to authenticate the server they are contacting, and
for servers to identify their clients. Authentication usually requires the entity that presents its identity
to confirm it either with something the client knows (e.g. password or PIN), something the client has
(e.g. a smart card, identity card) or something the client is (biometrics: finger print or retinal scan).
Biometric authentication has been proven to be the most precise way of authenticating a user's
identity. However, biometric processes such as scanning retina or matching fingerprints to one stored
in a database are often considered intrusive, and there always exists some measure of fear that this
information will be misused.
The approach to authentication that is gaining acceptance in the e-business world is by the use of
digital certificates. A digital certificate contains unique information about the user including
encryption key values. These public/private encryption key pairs can be used to create hash codes and
digitally sign data. The authenticity of the digital certificate is attested to by a trusted third party
known as a "Certificate Authority." The entire process constitutes Public Key Infrastructure.
Once an entity is certified as uniquely identified, the next step in establishing legitimate use is to
ensure that the entity’s activities within the system are limited to what it has the right to do. This may
include access to files, manipulation of data, changing system settings, etc. A secured system will
establish very well defined authorization policy together with a means of detecting unauthorized
activity.
12.3.5 Auditing or Traceability: From an accounting perspective, auditing is the process of officially
examining accounts. Similarly, in an e-business security context, auditing is the process of examining
transactions. Trust is enhanced if users can be assured that transactions can be traced from origin to
completion. If there is a discrepancy or dispute, it will be possible to work back through each step in
the process to determine where the problem occurred and, probably, who is responsible. Order
confirmation, receipts, sales slips, etc. are examples of documents that enable traceability. In a well-

142
secured system, it should be possible to trace and recreate transactions, including every
subcomponent, after they are done. An effective auditing system should be able to produce records of
users, activities, applications used, system settings that have been varied, etc., together with time
stamps so that complete transactions can be reconstructed.
12.3.6 Non-repudiation: It is an attribute of secure system that prevents the sender of a message from
denying having sent it.Non-repudiation is the ability of an originator or recipient of a transaction to
prove to a third party that their counterpart did in fact take the action in question. Thus the sender of a
message should be able to prove to a third party that the intended recipient got the message and the
recipient should be able to prove to a third party that the originator did actually send the message.
This requirement proves useful to verify claims by the parties concerned and to apportion
responsibility is cases of liability. Obviously, this is a crucial requirement in any business transaction
when orders are placed and both buyers and sellers need to be confident that not only are they dealing
with the appropriate parties but also that they have proof to support the claims of any action taken in
the process. Non-repudiation protocol is also useful in forensic computing where the goal is to collect,
analyze and present data to a court of law.

12.4 Introduction to Security Technology

12.4.1 Cryptography
Cryptography is a technique by which data, called plaintext, is scrambled or encrypted in such a way
that it becomes extremely difficult, expensive and time consuming for an unauthorized person to
unscramble or decrypt it. The encrypted text is called the ciphertext.

The steps in Cryptography

„ Encryption: Sender of a message Msg uses an encryption algorithm Encrypt and an
encryption key keye to generate the encrypted version of the message EncryptedMsg
EncryptedMsg = Encrypt ( Msg, keye)
„ Decryption: The receiver of EncryptedMsg uses an decryption key keyd to regenerate the
message Msg.
Msg = Decrypt (EncryptedMsg, keyd)

Cryptographic algorithms can be classified into two broad classes: symmetric key cryptography or
asymmetric key cryptography. If keye= keyd in the above encryption and decryption algorithm then the
corresponding the algorithm is called a Symmetric Key Algorithm. Example of such algorithms are
DES (Data Encryption Standard), TDES, IDEA, RC2, RC4, and RC5. These algorithms can be
implemented either in software form or in hardware form. The hardware implementation is typically
100 times faster than the software implementation. The major problem associated with the symmetric
key algorithm is key distribution. The keys are to be securely distributed before starting the actual
secure communication. Symmetric key algorithm cannot be used for authentication or non-repudiation
of the communication process. This is another disadvantage.

If keye is not equal to keyd then the corresponding algorithm is called an asymmetric key cryptography
algorithm. The entities wishing to use the algorithm must posses a pair of keys: a public and a private
key. The public key is known to all the outside entities. RSA (named after its inventors: Ron Rivest,
Adi Shamir and Leonard Adleman) is an example of an asymmetric cryptography algorithm.
Asymmetric key algorithms are much slower than symmetric key cryptographic algorithms. For
example, RSA is 100 times slower than DES. Private Key operation time grows with k3 where as the
private key operation time grows with k2, where k is the length of the key in bits. In real-life situations
RSA is never used for bulk data transfer. Rather it is used for bulk encryption key (symmetric key)
exchange.

12.4.2 Hash Function

A hash function is a reproducible method of turning some kind of data into a (relatively) small
number that may serve as a digital "fingerprint" of the data. The algorithm substitutes or transposes

143
the data to create such fingerprints. The fingerprints are called hash sums, hash values, hash codes or
simply hashes. In cryptography, a cryptographic hash function is a hash function with certain
additional security properties to make it suitable for use as a primitive in various information security
applications, such as authentication and message integrity. A hash function takes a long string (or
'message') of any length as input and produces a fixed length string as output, sometimes termed a
message digest or a digital fingerprint. In various standards and applications, the two most-commonly
used hash functions are MD5 and SHA-1. Properties of the Hash function are:
• It should be easy to compute h(Msg), where, Msg is the message to be sent.
• It should be hard to obtain Msg given h(Msg)
• It should be very hard to find another message Msg/ such that h(Msg) = h(Msg/)

12.4.3 Digital signature

In cryptography, a digital signature or digital signature scheme is a type of asymmetric cryptography
used to simulate the security properties of a signature in digital, rather than written, form. Digital
signature schemes normally give two algorithms, one for signing which involves the user's secret or
private key, and one for verifying signatures which involves the user's public key. The output of the
signature process is called the “digital signature”. Digital signatures, like written signatures, are used
to provide authentication of the associated input, usually called a "message." Messages may be
anything, from electronic mail to a contract, or even a message sent in a more complicated
cryptographic protocol. The process of digital signature generation and verification is shown below.
Message Received
Site A Message Sent to B Site B
from A

Msg Msg Msg Msg

Internet
MD MD
Hash Hash
function function
MD4
MD5 MD
SHA
SHA-1 =?

MD Encryption Decryption MD

A’s Private Key A’s Public Key

Digital Signature Generation Digital Signature Verification
Figure 1: digital signature generation and verification process

Digital signatures are used to create public key infrastructure (PKI) schemes in which a user's public
key (whether for public-key encryption, digital signatures, or any other purpose) is tied to a user by a
digital identity certificate issued by a certificate authority. PKI schemes attempt to unbreakably bind
user information (name, address, phone number, etc.) to a public key, so that public keys can be used
as a form of identification.

12.4.4 TLS / SSL Protocol

Transport Layer Security (TLS) and its predecessor, Secure Sockets Layer (SSL), are cryptographic
protocols which provide secure communications on the Internet for such things as web browsing, e-
mail, Internet faxing, instant messaging and other data transfers. There are slight differences between
SSL 3.0 and TLS 1.0, but the protocol remains substantially the same. The TLS protocol(s) allow

144
applications to communicate across a network in a way designed to prevent eavesdropping,
tampering, and message forgery. TLS provides endpoint authentication and communications privacy
over the Internet using cryptography. Typically, only the server is authenticated (i.e., its identity is
ensured) while the client remains unauthenticated; this means that the end user (whether an individual
or an application, such as a Web browser) can be sure with whom they are communicating. The next
level of security — in which both ends of the "conversation" are sure with whom they are
communicating — is known as mutual authentication. Mutual authentication requires public key
infrastructure (PKI) deployment to clients.

As explained in the figure below, TLS involves two basic protocols with following functions:
TLS Handshake protocol
• Peer negotiation for algorithm support
• Public key encryption -based key exchange and certificate-based authentication
TLS record protocol
• Symmetric cipher -based traffic encryption

HTTP

-N eg otiation of cryptographic
T LS H an d sh ake P ro toco l and co m p ressio n algo rithm s
-E xcha ng e of secrets
throu gh P K
TLS

-G e nera tion o f se cre te key

T LS R eco rd P rotoco l -E n cryption /d ecryption

-M essa ge au the ntica tion
-C om p ressio n/D ecom pre ssion

TCP
Figure 2: TLS protocol in context

12.5 Public Key Infrastructure -Solution to the e-Business security

Public Key cryptography supports security mechanisms such as confidentiality, integrity,
authentication, and non-repudiation. To successfully implement these security mechanisms, an
infrastructure to manage them should be planned. A public key infrastructure(PKI) is a foundation on
which other applications, system, and network security components are built. In this chapter first we
will discuss about PKI and then some secure protocols which are used in e-Business transactions.
12.5.1 PKI Framework
The framework of a PKI consists of security and operational policies, security services, and
interoperability protocols supporting the use of public-key cryptography for the management of keys
and certificates. The generation, distribution, and management of public keys and associated
certificates normally occur through the use of Certification Authorities (CAs), Registration
Authorities (RAs), and directory services, which can be used to establish a hierarchy or chain of trust.
CA, RA, and directory services allow for the implementation of digital certificates that can be used to
identify different entities. The purpose of a PKI framework is to enable and support the secured
exchange of data, credentials, and value (such as monetary instruments) in various environments that
are typically insecure, such as the Internet.
A major benefit of a PKI is the establishment of a trust hierarchy because this scales well in
heterogeneous network environments. Entities that are unknown to one another, each individually
establish a trust relationship with a CA. The CA performs some level of entity authentication,
according to its established rules as noted in its Certificate Practices Statement or CPS, and then

145
issues each individual a digital certificate. That certificate is signed by the CA and thus vouches for
the identity of the individuals. Unknown individuals can now use their certificates to establish trust
between them because they trust the CA to have performed an appropriate entity authentication, and
the CA's signing of the certificates attests to this fact.
A hierarchical trust model represents the most typical implementation of a PKI. In its most simple
instantiation, this trust model allows end entities’ certificates to be signed by a single CA. In this trust
model, the hierarchy consists of a series of CAs that are arranged based on a predetermined set of
rules and conventions For example, in the financial services world, rather than have a single authority
sign all end entities’ certificates, there may be one CA at a national level that signs the certificates of
particular financial institutions. Then each institution would itself be a CA that signs the certificates of
their individual account holders. Within a hierarchical trust model there is a trust point for each
certificate issued. In this case, the trust point for the financial institution's certificate is the national or
root CA. The trust point for an individual account holder is their institution's CA. This approach
allows for an extensible, efficient, and scalable PKI.
12.5.2 Security Services
The principal business objectives and risk management controls that can be implemented by a PKI are
presented in this section:
Confidentiality: Confidentiality means ensuring that the secrecy and privacy of data is provided with
cryptographic encryption mechanisms. Encryption of data is possible by using either public
(asymmetric), or secret (symmetric) cryptography. Since public key cryptography is not as efficient as
secret key cryptography for data encipherment, it is normally used to encipher relatively small data
objects such as secret keys used by symmetric based encryption systems. Symmetric cryptographic
systems are often incorporated into PKIs for bulk data encryption; thus, they are normally the actual
mechanism used to provide confidentiality.
Integrity: Integrity means ensuring that data cannot be corrupted or modified and transactions cannot
be altered. Integrity can be provided within a PKI by the use of either public (asymmetric), or secret
(symmetric) cryptography. An example of secret key cryptography used for integrity is DES in Cipher
Block Chaining mode where a Message Authentication Code (MAC) is generated. In the PKI
environment, using symmetric cryptographic systems for implementing integrity does not scale
particularly well. Public key cryptography is typically used in conjunction with a hashing algorithm
such as SHA-1 or MD5 to provide integrity.
Authentication: Authentication means verifying that the identity of entities is provided by the use of
public key certificates and digital signature envelopes. Authentication in the e-commerce environment
is performed very well by public key cryptographic systems incorporated into PKIs. The primary goal
of authentication in a PKI is to support the remote and unambiguous authentication between entities
unknown to each other, using public key certificates and CA trust hierarchies. Authentication in a PKI
environment relies on the mathematical relationship between the public and private keys. Messages
signed by one entity can be tested by any relying entity. The relying entity can be confident that only
the owner of the private key originated the message, because only the owner has access to the private
key.
Non-Repudiation: Non-repudiation means ensuring that data cannot be renounced or a transaction
denied. This is provided through public key cryptography by digital signing. Non-repudiation is a by-
product of using public key cryptography. When data is cryptographically signed using the private
key of a key pair, anyone who has access to the public key of that pair can determine that only the
owner of the key pair itself could have signed the data in question. For this reason, it is paramount that
end entities secure and protect their private keys used for digitally signing data.
12.5.3 PKI Logical Components
Different logical components comprise a PKI. The following outlines the typical logical components
in a PKI:
• End entities or subscribers

146
 • Certificate authorities
• Certificate policies
• Certificate practices statement
• Hardware security modules
• Public key certificates
• Certificate extensions
• Registration authorities
• Certificate depositories
End Entities or Subscribers: An end entity or subscriber is any user or thing, including inanimate
objects, such as computers that have a need for a digital certificate to identify them for some reason.
The end entity normally must have the capacity to generate a public/private key pair and some means
of securely storing and using the private key. By definition, an end entity is not a CA.
Certificate Authorities: A Certificate Authority plays a critical role in a PKI. According to the IETF,
a CA is “an authority trusted by one or more users to create and assign public key certificates.”
[Internet X.509 Public Key Infrastructure PKIX Roadmap, March 10, 2000]. CA functions as a
trusted third party and provides various key management services A CA’s public keys must be
distributed to all entities that trust the CA’s certificates. If a CA is a Root CA, that is, at the top of the
trust hierarchy and has no superior CA to vouch for it, then the CA must distribute its public keys as
self-signed certificates with an acceptable key certificate format and distribution protocol. The CA
must also make its clear text public keys available, so that relying entities can resolve the self-signed
certificates.
Certificate Policy: A primary tenet of e-commerce security is the CP statement. The CP statement
provides the overall guiding principles that an organization endorses regarding who may do what and
how to systems and data. A CP also specifies how controls are managed. In addition, a CP names a set
of rules that indicates the applicability of a public key certificate to a particular community or class of
applications with common security requirements.
Certificate Practice Statement: The details of a policy statement should be published in a Certificate
Practices Statement or CPS. The CPS is a statement of the practices that a CA employs in issuing
public key certificates. The CPS document enumerates the procedural and operational practices of a
PKI. The CPS should detail all processes within the life cycle of a public key certificate including its
generation, issuance, management, storage, deployment, and revocation. The CPS should also specify
the original entity authentication process that an end entity must be validated through before
participating in a PKI. The objective of the CPS is to instill trust in the PKI such that the user
community at large will have sufficient confidence to participate in it.
Hardware Security Modules: Hardware Security Modules (HSMs) are another primary component
of a CA. A CA must instill trust in not only its client base but also in those who rely upon the
certificates issued to subscribers. Since that trust is predicated upon the security and integrity of the
CA's private keys used to sign the public key certificates of subscribers, it is necessary that those
private keys be secured as best as possible. For this reason, CAs should only store and use their
private keys in specialized computer equipment known as HSMs. HSMs are also known as Tamper
Resistant. Various standards are used to categorize HSMs, for example, FIPS-140-1.
Public Key Certificates: A CA’s primary purpose is to support the generation, management, storage,
deployment, and revocation of public key certificates. A public key certificate demonstrates or attests
to the binding of an end entity’s identity and its public key. The basic constructs of a certificate should
include the name of an entity, identifying information about the entity, expiration period for the
certificate, and the entity’s public key. Other additional and useful information may be included in a
certificate: serial numbers, the CA's name, the CA's public key certificate itself, the type of algorithms
used to generate and verify the keys and certificate, and any other information that the CA generating

147
the certificate considers useful. The most widely used format for digital certificates are those based on
the IETF X.509 standards

Certificate Extensions: Certificate extensions provide additional information within a certificate and
allow them to be tailored for the particular needs of an organization. Certificate extensions can affect
the interoperability of certificates if a relying party does not recognize the structure or content of the
certificate extensions. The types of information that may be found in a certificate extension include:
policy, usage, revocation, and naming data, which provide particular details unique to an
organization’s PKI.
Registration Authorities: A Registration Authority (RA) is an optional but common component of a
PKI. An RA is used to perform some of the administrative tasks that a CA would normally undertake.
Most importantly, an RA is delegated, with the CA’s explicit permission, the authority to perform
tasks on behalf of the CA. The primary purpose of an RA is to verify an end entity’s identity and
determine if an end entity is entitled to have a public key certificate issued. The RA must enforce all
policies and procedures defined in the CA’s CP and CPS.
Certificate Depositories: A certificate depository, sometimes referred to as a certificate directory, is
also an optional but common component of a PKI. A certificate depository may be an efficient
solution for closed systems (e.g., intranet) or those in isolated processing environments (e.g.,
chipcard-based applications) where the Root CA public key is distributed locally or revocation lists
are stored locally.
Certificate distribution can be accomplished by simply publishing certificates in a directory controlled
by a CA or RA. When the directory is controlled by the CA or RA, the certificate distribution process
is greatly simplified. Rather than trying to distribute every certificate to a unique point, the CA simply
updates the directory. A critical factor is that only the CA must have the authority to update or modify
the directory, but the directory must be publicly readable. LDAP is a good example of a simple and
efficient standards based directory format and protocol that can be used for certificate distribution.
12.5.4 PKI Functions
This basic processes which are common to all PKIs are:
• Public key cryptography – Includes the generation, distribution, administration, and control of
cryptographic keys.
• Certificate issuance – Binds a public-key to an individual, organization, or other entity, or to
some other data—for example, an email or purchase order.
• Certificate validation – Verifies that a trust relationship or binding exists and that a certificate
is still valid for specific operations.
• Certificate revocation – Cancels a previously issued certificate and either publishes the
cancellation to a Certificate Revocation List or enables an Online Certificate Status Protocol
process.

12.5.5 Public Key Certificate Binding

To participate in a PKI, an end entity must enroll or register in a PKI. The result of this process is the
generation of a public key certificate. The binding is declared when a trusted CA digitally signs the
public key certificate with its private key. To generate a certificate, the CA performs the following
steps.
1. Acquire a public key from the end entity.
2. Verify the identity of the end entity.
3. Determine the attributes needed for this certificate, if any.

148
 4. Format the certificate.
5. Digitally sign the certificate data.
Acquiring the Public Key: Depending on the value proposition and business risks associated with
the issuance of a public key certificate, the CA may take basic measures when obtaining the required
identification and certificate information. For example, the CA may allow it to be sent electronically
over the Internet, or the CA could use more sophisticated means such as mandating out of band
manual methods (e.g., using bonded couriers). The type and integrity of the credentials requested by
the CA during the enrollment process also depends on the intentions of the CA.
Verify the Identity of the End Entity: Depending on the business model in place and the amount of
reliance on the public key certificates, the CA may take simple measures to authenticate the end
entity. In any case, the most important factor in the binding process is to ensure that an entity’s
identity is verified unambiguously.
Formatting the Certificate: Before the certificate is signed by the CA, all data to be placed into the
certificate is collected and formatted. The specific data content and format of a public key certificate
can vary depending on the needs of the PKI.
Signing the Certificate: Finally, the certificate is digitally signed under the private key of the CA
used for signing certificates. Once signed it can be distributed and/or published using different
vehicles.
Certificate Validation: Any digital certificate issued by a CA will be valid only for a specified
amount of time mentioned in the certificate. After this period the use of certificate is no longer valid.
So before using any certificate its expiry date must be verified. Along with the expiry date validation
the certificate must also be validated whether the certificate has been signed by the appropriate
certifying authority.
Certificate Revocation: Although public key certificates are issued for a fixed period of time before
they become void, situations can arise where they are no longer trustworthy and thus must be
prematurely expired. This is known as certificate revocation. Certificate revocation must be initiated
by the CA or their delegate such as an RA, which originated the end entity's certificate. The
predominant vehicle for certificate revocation is known as a Certificate Revocation List or CRL. A
CRL is a list generated by the CA that contains unique information about the revoked certificates
which enables relying entities to determine if a certificate is valid or not. A CRL must be published in
a publicly available repository or directory.
12.6. Electronic Payment Systems
A Payment System is a mechanism that facilitates transfer of value between a payer and a beneficiary
by which the payer discharges the payment obligations to the beneficiary. Payment system enables
two-way flow of payments in exchange of goods and services in the economy. An electronic payment
system is needed for compensation for information, goods and services provided through the Internet -
such as access to copyrighted materials, database searches or consumption of system resources - or as
a convenient form of payment for external goods and services - such as merchandise and services
provided outside the Internet. It helps to automate sales activities, extends the potential number of
customers and may reduce the amount of paperwork.
12.6.1 Business to Business Payment Systems
Changes in the marketplace for business-to-business (B2B) payments increasingly demand executives'
attention. Growing numbers of e-payments, decreasing paper check volumes, and new legislation are
the main motivators. Businesses are beginning to realize they must add new payment options to a
process still dominated by paper checks, wire transfers, and automated clearinghouse (ACH)
transactions.
B2B payments involve up to five distinct components:
The buyer - purchases goods and services;

149
The buyer's financial institution - provides banking services to the buyer;
The supplier - provides goods and services;
The supplier's financial institution - provides banking services to the supplier; and
The Federal Reserve or other check-clearing institution - facilitates the check-sorting process.
Figure 3 outlines the traditional B2B check-clearing process in which a buyer writes a check that
subsequently flows through the supplier, the supplier's bank, the Federal Reserve (or other check-
clearing institution), the buyer's bank, and finally back to the buyer. This process is supported by
human intervention throughout the supply chain, including manual entry of check amounts and
associated transaction data.

Figure 3: B-to-B Payment traditional Vis-à-vis electronic

150
Shifting payment processes to an electronic format requires an IT infrastructure to route payments and
related data through an integrated network. Using such end-to-end infrastructure poses a significant
adoption challenge because users' technological sophistication with regard to the payment supply
chain is so varied. Creating uniformity in the e-payment infrastructure is further complicated by the
fact that the standards of the Federal Reserve infrastructure apply to Fed products but not to products
outside the Fed's domain. Therefore, building the infrastructure necessary to securely and reliably
transmit e-payments and related transaction information among a large group of diverse users is a
significant obstacle and expense. However, the result of an improved payment network, as in Figure
3, is a streamlined process supporting the continuous exchange of information and funds.
12.6.1.1 B-to-B E-Payment Options
Three notable e-payment technologies promise to simultaneously address key adoption challenges,
accelerate B2B e-payment adoption, and deliver across multiple value dimensions: Automated
clearing house (ACH) -based bank proprietary e-payment platforms; Enhanced purchasing card
technology; and Open network systems.
ACH-Based Bank Platforms: ACH-based transactions (such as for direct deposit of employee
salaries) have existed in one form or another since the early 1970s. Bank proprietary systems are e-
payment solutions built on the ACH network. In the past decade, banks have invested heavily in
developing software packages that have improved the functionality of B2B ACH transactions. They
now combine financial processing and electronic data warehousing capabilities into one
comprehensive service. Buyers and suppliers send complete trade information (including contracts,
pricing, orders, receipts, and invoices) to bank systems; the information is then stored in a document
available to both parties. These platforms seek to provide comprehensive electronic bill payment and
presentment from the beginning to the end of the payment cycle.
From the standpoint of value attribute comparisons, ACH-based payment systems meet many of the
requirements identified by B2B transaction participants. The importance of these attributes is the
foundation of historical ACH popularity in facilitating relatively small recurring financial transactions
between businesses and consumers. One particularly troublesome issue is the relatively high
implementation costs of ACH-based platforms, particularly for small- and medium-size businesses.
This issue further complicates systems-integration problems. Another significant issue plaguing ACH-
based solutions is that they rely on bilateral closed networks. This means that businesses must share
sensitive financial information, in turn increasing the risk of fraud. Global coverage is another
concern with any form of ACH-based payment, as ACH systems are typically built for domestic use.
Due to the foregoing, along with the lack of other desired characteristics, ACH-based e-payment
systems have not gained significant traction in the B2B e-payment market. Despite this, bank
proprietary solutions are well suited for large corporations and government agencies that execute
relatively low- to medium-value recurring payments within the country.
Enhanced Purchasing Cards: Enhanced purchasing cards, or p-cards, (such as MasterCard's e-P3)
are e-payment solutions that offer management services for B2B purchasing, presentment, and
payment. This e-payment strategy is the result of collaboration between outside software vendors and
existing p-card issuers to provide businesses with customized end-to-end e-commerce. P-card
solutions allow buyers and sellers to collaborate within a common environment, streamline payment
processes, and use p-cards for payment. Businesses transmit orders electronically, view the status of
orders and invoices online, control the initiation of payments, and integrate data into existing financial
systems.
Enhanced p-cards share many of the positive characteristics of desirable e-payment networks. They
are typically Web-based and relatively easy to integrate with legacy software and hardware
environments. This ease of integration greatly reduces the cost and complexity of implementation.
Another p-card benefit is reduced potential for fraud and credit loss, due to enhanced settlement
processes. P-card transactions settle through an online "credit gateway" associated with the p-card
issuer that allows institutions to exchange encrypted information for authorization purposes. These
systems provide a secure environment for payments because sensitive account information is not
exchanged directly, and transactions are easily traceable and challenged when disputes arise.

151
One significant disadvantage of enhanced p-cards is that interchange fees typically still apply. This
means that suppliers may be averse to accepting the enhanced p-card for medium- to large-value
transactions, as fees increase as the dollar value of the transaction increases. Given these benefits and
limitations, the enhanced p-card represents a good way for current users of p-cards and prospective e-
payment-orientated businesses to conduct low-value nonrecurring transactions.
Open Network Systems: Open network systems, like Visa's "Visa Commerce" platform, are rules-
based e-payment systems that utilize open, secure, global-settlement networks to process buyer-
initiated payments. Open networks interface directly with front-end procurement and accounts-
payable systems to provide buyers and suppliers alike with a seamless, integrated corporate payment
solution. Open networks also enable buyers to initiate and settle payments based on preestablished
terms with suppliers.
Unlike enhanced p-card and ACH-based platforms, open networks employ a flexible fee schedule,
allowing for the minimization of fees for relatively high-dollar-amount transactions. In addition,
open-network solutions may be easier to integrate with legacy environments. For example, Visa
Commerce was designed to integrate with existing procurement applications, regardless of their
sophistication. This integration strategy means that financial institutions, buyers, and suppliers that
want to use an open network can do so with minimal up-front investment. Finally, open networks
typically offer global coverage, a major consideration for moderate-to-large-size firms. Open network
payments appear to be best suited for relatively large-value, multiple-invoice directed payments, as
they circumvent interchange fees and support robust transaction data.
12.6.1.2 Challenges with B-to-B E-Payment Systems
Four main challenges confront businesses interested in adopting e-payment processes: systems
integration; the absence of remittance standards; security; and uncertainty in the return on
investment; in addition, candidate technologies must be considered in light of their ability to deliver
value in multiple dimensions.
Systems integration: The most daunting barrier to the adoption of e-payments is a lack of integration
across the many systems needed to support the process.
Remittance standards: Complicating these challenges is a lack of even minimum standards
pertaining to e-payment data. Early adopters of e-payment technologies still complain that
inconsistencies in remittance standards have resulted, in some cases, in the corruption of important
remittance data, including invoice numbers, and other information, including payment amounts and
payer identification.
Security: Since B2B transactions involve multiple parties, they flow across diverse technology
architectures. Each party in an electronic transaction is subject to the security procedures of other
members in its financial supply chain. Arguably, paper-based payments are subject to similar security
challenges, though they are well understood and have long been accounted for.
Value proposition uncertainty: E-payment networks (like other networks) are subject to "network
externalities," so the value of participation is contingent on the size of the network itself; that is, the
greater the number of participants, the more valuable the network is to each participant. Given that the
technical aspects of e-payment networks are still evolving, it has been difficult for potential
participants to estimate the value of joining and assess the appropriate level and speed of the related
investment. Investment decisions in this context are influenced by businesses' perceptions of the
likelihood of agreement on the issues related to standards and integration. The value proposition of e-
payment networks is further clouded by the realization that connecting accounts for only a portion of
the total cost.

152
12.6.2 Business to Consumer Payment Systems
Business-to-consumer is a form of electronic commerce in which products or services are directly
sold from a firm or company to a consumer without any intermediation. Internet has been proved to
be an effective medium for direct selling. B 2 C on-line payment occurs when the enterprise (the
seller) and the individual (the buyer) settle their transaction on the E-business web site on the Internet
and the bank provide them with on-line fund settlement service. Traditional B-to-C payment services
can be cash based, cheque based or credit card based payment. Out of these mechanisms, credit card
based payment is the most popular.
Credit Card Based Payment: Ideally Internet based credit card payment must works in a similar
way like that of the traditional credit card based system. However, security becomes the primary
concern in an Internet based system. The credit card details once submitted to the merchant’s Web site
can be reused by the merchant himself. The protocol like SET (Secure Electronic Transaction) helps
preventing this kind of situation. SET is pragmatic approach that paves the way for easy, fast and
secure transaction over the internet. That transaction is initiated with a handshake, with the merchant
authenticating itself to the payer and fixing all payment data. The payer then uses a sophisticated
encryption scheme to generate payment slip. The goal of the encryption scheme is to protect the
sensitive payment information such as credit card number. Next the payment slip is signed by the
payer and is sent to the merchant. The merchant sends the slip to its acquirer gateway, to authorize
and capture payment. The acquirer checks all signatures and the slip, verifies the credibility of the
payer and sends either a positive or a negative signed acknowledgement back to merchant and the
payer. SET however is not widely accepted as it is not computationally efficient and requires heavy
investment. Another alternative is to use the Web sites of the Trusted Third Parties to submit credit
card details (Ex. Paypal).
Cheque-like System: Electronic cheques work in a similar way like that of paper cheques. The
customer need a wallet software to store information on the browser about the checking account.
During cheque payment the customer sends encrypted customer’s payment information with banks
public key and other information for the merchant encrypted with the customer’s private key to the
merchant. The merchant sends the payment information to the appropriate bank and get the fund
transferred to its account. CyberCash is one example of such a chequing system.
Digital Cash: In a digital cash system, user can withdraw e-cash coins from a bank and use them to
pay other users. Digital cash is kept as computer files containing huge random numbers digitally
stamped by the bank, i.e. encrypted with banks private key followed by customer’s pubic key. They
are maintained by the customer’s wallet software. Digital cash can be classified as identifiable or
anonymous depending on whether it requires the identity of the spender. Depending on the direct
participation of the bank it can be classified as online or offline. One major problem associated the
money when it is both anonymous and Offline is the problem of double spending. In case of
identifiable Electronic Money a serial number is provided by the bank for each of the computer file.
Example of this type of money is CyberCash. In case of anonymous electronic money the serial
number is created and blinded by the customer. It is subsequently signed by the bank. DigiCash is one
such example. Online Electronic Money is validated online with active participation of the bank
(CyberCash) whereas offline money is not validated online during the transaction. This scheme is
appropriate for peer to peer transaction where bank need not be an active participant (DigiCash).

12.7. Conclusion
E-business depends on providing customers, partners, and employees with access to information, in a
way that is controlled and secure. Managing e-business security is a multifaceted challenge and
requires the coordination of business policy and practice with appropriate technology. In addition to
deploying standards bases, flexible and interoperable systems, the technology must provide assurance
of the security provided in the products. As technology matures and secure e-business systems are
deployed, companies will be better positioned to manage the risks associated with disintermediation
of data access. Through this process businesses will enhance their competitive edge while also

153
working to protect critical business infrastructures from malefactors like hackers, disgruntled
employees, criminals and corporate spies.
The volume of e-payment activity has risen steadily since 1979 and shows no sign of slowing down.
Motivating this trend has been consumer willingness to submit payments electronically. Increasingly,
businesses have sought to improve their understanding of the value of e-payment systems. Businesses
that identify and adopt effective e-payment strategies are more likely to realize streamlined business
processes and significant bottom-line savings compared to their peer organizations that don't. Those
seeking to implement e-payment technologies must consider not only their emerging technology
options but evolving payment needs as well.

References
[1] Eben Otuteye, A Systematic Approach to E-business Security,
http://ausweb.scu.edu.au/aw03/papers/otuteye/paper.html
[2] An Oracle white paper, Managing e-Business Security Challenges,
http://www.oracle.com/technology/deploy/security/oracle9ir2/pdf/9iR2hisec.PDF
[3] E-business Resource Group Security Guidelines,
http://www.bc.pitt.edu/ebusiness/arEBSecurityGuide.pdf
[4] Muhammad M. Satti, Brain J. Garner, Mahmood M. Nagril, Information Security Standards for e-
Business, http://www.macquarietelecom.com/whitepapers/Info%20Security%20Standards%20e-
biz.pdf

[5] Canadian banker association, Minding your e-Business, Security and Privacy Matter,
http://www.cba.ca/en/content/general/MYEBUS_Final_Report.pdf

[6] Public Key Infrastructure Overview, By Joel Weise – SUN PS Global Security Practice, Sun
Blueprints Online, August 2001, http://www.sun.com/blueprints/0801/publickey.pdf

[7] Cutting Checks: Challenges and Choices for the Adoption of B2B Electronic Payments, Mark J.
Cotteleer, Christopher A. Cotteleer, Andrew W. Prochnow, Communications of the ACM,
Volume 50, Number 6 (2007), Pages 56-61

[8] The State of the Art in Electronic Payment Systems, N. Asokan, Phillipe A. Janson, Michael
Steiner, Michael Waidner, IEEE Computer, Vol 30 , Issue 9 ( 1997) Pages: 28 - 35

*********

154
 CHAPTER XIII
Automatic Data Capture using RFID and its implications

13.1 Introduction
Optimizing inventory level across the coordinating organizations is an important aspect of supply
chain management. Such efforts require real-time inventory data to be shared among the coordinating
organizations. Collection and dissemination of real-time inventory data requires integration of
automatic data capture and transfer technologies. Automated Identification and Data Capture (AIDC)
includes technology to identify objects, and automatically collects data about them and updates the
data into software systems without human intervention. Some examples of AIDC technologies include
bar codes, RFID, smart cards, voice and facial recognition, and so forth.
Modern AIDC heavily relies upon barcodes, for automated data capture. A barcode basically is a
machine-readable visual representation of information printed on the surface of the objects. The
encoded data on the barcodes is read by barcode readers, which update the backend ERP, SCM, or
WMS systems. However there are some inherent issues with using a barcode as shown in Table 1. To
overcome these issues the industry is now looking at the possibility of using new generation AIDC
technology like the RFID. For example, objects tagged with RFID can be sensed in a wide area, and
there is no need to individually scan all the objects in front of an optical scanner. It can also offer
item-level tagging - that is, each item within a product range can be uniquely identified. The data
collected using RFID technology helps associate production events with each inventory item, which
leads to a tighter inventory control approach that relies on such real time data.
Table 1: Bar code vis-à-vis RFID solution for automated data capture
Bar Code Deficiency RFID improved solution
Line of Sight Technology Able to scan and read from different angles and
through certain materials
Unable to withstand harsh conditions (dust, Able to function in much harsher condition
corrosive), must be clean and not deformed
No potential for further technology Technology advancement is possible due to new
advancement chip and packaging technique
Can only identify the items generally and EPC code will enable to identify up to 296 items
not as unique objects uniquely
Poor tracking Technology, labor intensive Potential to track the items in real time as they
and slow move through the supply chain.

13.2 Technology overview

An RFID system is composed of three main elements: an RFID tag (inlay), which contains data that
uniquely identifies an object; an RFID reader, which writes this unique data on the tags and, when
requested, can read this unique identifier; and an RFID middleware, which processes the data
acquired from the reader and then updates it to the backend database or ERP systems.
A typical RFID system is shown in Figure 1. When the RFID tag comes in the range of the RFID
reader, the reader activates the tag to transmit its unique information. This unique information is
propagated to the RFID middleware, which appropriately processes the gathered information and then
updates the backend database.
 Figure 1: A Typical RFID Setup

There are several versions of RFID that operate at different radio frequencies. The choice of
frequency is dependent on the business requirements and read environment – it is not a technology
where ‘one size fits all’ applications. Three primary frequency bands are being used for RFID:
• Low Frequency (125/134KHz) – Most commonly used for access control, animal tracking and
asset tracking. Tags in this range are not affected by metallic surroundings and hence are ideal for
identifying metal objects like vehicles, tools, containers, and metallic equipments. The reading
range varies from a few centimeters to a meter depending upon the size of the antennae and the
reader used. These tags can also penetrate through water and body tissues, and hence often used
for animal identification. Most LF-based systems can only read one tag at a time—that is, they do
not support reading multiple tags simultaneously
• High -Frequency (13.56 MHz) – Used where medium data rate and read ranges up to about 1.5
meters are acceptable. HF tags can penetrate through most materials including water and body
tissues; however they are affected by metal surroundings. HF tags are comparatively cheaper than
the LF tags. The data transfer rate is higher compared to LF tags, for example, 20ms for read
operation. This is because at high frequency the communication is faster. The reader can read
multiple tags simultaneously.
• Ultra High-Frequency (850 MHz to 950 MHz) – offer the longest read ranges of up to
approximately 3 -6 meters and high reading speeds. UHF tags are normally less expensive than
HF. Such tags are commonly used on objects that are moving at a very high speed, and a large
number of tags are scanned per second in the business contexts such as supply chain, warehouse,
and logistics. UHF tags do not work well in liquid and in metal surroundings. Larger read range
limits their use to banking and access control applications, because the access card may be scanned
from a longer distance and some unauthorized person might gain entry in restricted premises.

13.2.1 RFID Tags

An RFID /tag is a microchip attached with an antenna to a product that needs to be tracked. The tag
picks up signals from the reader and reflects back the information to the reader. The tag usually
contains a unique serial number, which may represent information, such as a customer’s name,
address, and so forth.
RFID tags can be classified using two schemes. First, the tags can be classified based on their ability
to perform radio communication -active, semi-active (semi-passive), and passive tags. Second, the
tags can be classified based upon their memory -read-only, read-write or write-once, and read-many.
Active vs. Semi-Active(or Semi-Passive) vs. Passive RFID tags
• Active Tags have a battery that provides necessary energy to the microchip for transmitting a radio
signal to the reader. These tags generate the RF energy and apply it to the antennae and transmit to
the reader instead of reflecting back a signal from the reader. These batteries need to be recharged
or replaced once they are discharged. Some tags have to be disposed off when the batteries run out
of power. These tags have a read range of several 100 meters and are very expensive and hence are

156
 used for tracking expensive items; for example, the U.S. military uses these tags to track supplies
at ports
• Semi-Active Tags (or semi-passive or battery-assisted) also contain a battery, which is used to run
the circuitry on the microchip, however it still relies on the reader’s magnetic field to transmit the
radio signal (i.e., information). These tags have a larger range because all the energy supplied by
the reader can be reflected back to the reader, which means it can work at low-power signal levels
as well. These tags have a read range up to 100 meters and may cost a dollar or more. Some of
these tags often are dormant - that is, they are activated by the presence of a reader’s magnetic
field. Once activated, the battery runs the circuitry and responds back to the reader. This is a
mechanism to save the battery power.
• Passive Tags completely rely on the energy provided by the reader’s magnetic field to transmit the
radio signal to and from the reader. It does not have a battery. As a result, the read range varies
depending upon the reader used. A maximum distance of 15 meters (or 50 feet) can be achieved
with a strong reader antennae and RF-friendly environment.
Read-Write vs. Read-Only RFID Tags
• Read-only tags: The reader can only read data stored on such tags. The data cannot be modified in
any manner. The tag manufacturer programs the data on the tag. Such tags are comparatively very
cheap.
• Write-once read-many (WORM): The owner of the tag can program the data by writing the
content on the tag. Data stored on this tag can be written only once; however it can be read many
times.
• Read-write tags: Data stored on such tags can be easily edited when the tag is within the range of
the reader. Such tags are more expensive and are not often used for commodity tracking. These
tags are reusable; hence they can be reused within an organization.
13.2.2 RFID Readers
RFID readers send radio waves to the RFID tags to enquire about their data contents. The tags then
respond by sending back the requested data. The readers may have some processing and storage
capabilities. The reader is linked via the RFID middleware with the backend database to do any other
computationally intensive data processing. RFID readers can be classified using two different
schemes. First, the readers can be classified based on their location as handheld readers and fixed
readers. Second, the tags can be classified based upon the frequency in which they operate -single
frequency and multi-frequency.
Fixed Readers vs. Handheld Readers
• Fixed RFID Readers are fixed at one location (e.g., choke point). In a supply chain and
warehouse scenario, the preferred location of a reader can be along the conveyor belt, dock
door antennae or portals, depalletization stations, or any other mobile location.
• Portable or Handheld RFID Readers are designed for Mobile Mount Applications, for
example, vehicles in a warehouse or to be carried by inventory personnel, and so forth.
Single Frequency vs. Multi-Frequency
• Single-frequency operation readers operate in one frequency zone, either in LF, HF, or UHF.
Such readers become inconvenient if tags in a warehouse are operating in different frequencies.
• Multi-frequency operation readers can operate in multiple frequencies. Such readers can
conveniently read tags, which operate in different frequencies (i.e., LF, HF, or UHF). Hence
these are more useful from a practical perspective, however such readers come at a premium
price.

157
13.2.3 RFID Middleware
In a general, the RFID middleware manages the readers and extracts Electronic Product Code (EPC –
To be explained latter) data from the readers; performs tag data filtering, aggregating, and counting;
and sends the data to the enterprise WMSs (warehouse management systems), backend database, and
information exchange broker. Figure 1 shows the relationship between tag, reader, RFID middleware,
and backend database. An RFID middleware works within the organization, moving information (i.e.,
EPC data) from the RFID tag to the integration point of high-level supply-chain management systems
through a series of data-related services.
From the architectural perspective, RFID middleware has four layers of functionality: reader API,
data management, security, and integration management. The reader API provides the upper layer of
the interface interacting with the reader. Meanwhile, it supports flexible interaction patterns (e.g.,
asynchronous subscription) and an active “context-ware” strategy to sense the reader. The data
management layer mainly deals with filtering redundant data, aggregating duplicate data, and routing
data to appropriate destination based on the content.
The integration layer provides data connectivity to legacy data source and supporting systems at
different integration levels and thus can be further divided into three sub-layers: application integra-
tion, partner integration, and process integration. The application integration provides varieties of
reliable connection mechanisms (e.g., messaging, adaptor, or the driver) that connect the RFID data
with existing enterprise systems such as ERP or WMS. The partner integration enables the RFID
middleware to share the RFID data with other RFID systems via other system communication
components (e.g., the Data Exchange Broker in Figure 2). The process integration provides capability
to orchestrate the RFID-enabled business process. The security layer obtains input data from the data
management layer, and detects data tampering which might occur either in the tag by a wicked RFID
reader during the transportation or in the backend internal database by malicious attacks. The overall
architecture of RFID middleware and its related information systems in an organization are depicted
in Figure 2.

Figure 2. RFID middleware architecture

The backend DB component stores the complete record of RFID items. It maintains the detailed item
information as well as tag data, which has to be coherent with those read from the RFID. It is worth
noting that the backend database is one of the data-tampering sources where malicious attacks might
occur to change the nature of RFID item data by circumventing the protection of an organization’s

158
firewall. The WMS integrates mechanical and human activities with an information system to
effectively manage warehouse business processes and direct warehouse activities. The WMS
automates receiving, put-away, picking, and shipping in warehouses, and prompts workers to do
inventory cycle counts.
The RFID middleware employs the integration layer to allow real-time data transfer towards the
WMS. The data exchange broker is employed in this architecture to share, query, and update the
public data structure and schema of RFID tag data by exchanging XML documents. Any update of the
data structure will be reflected and propagate to all involved RFID data items stored in the backend
database. From the standardization view, it enables users to exchange RFID-related data with trading
partners through the Internet. From the implementation angle, it might be a virtual Web services
consumer and provider running as peers in the distributed logistics network.

13.3 Applications of RFID

Applications fall into two principal categories: firstly, short range applications where the reader and
tag must be in close proximity (such as in access control) and secondly, medium to long application,
where the distance may be greater (such as reading across a distribution centre dock door). A sample
of applications is shown below:
Access control for people: there are many areas where RFID tags are carried by people to allow them
to gain access to facilities or services:
• Secure access to work place
• Safety access to dangerous/secure equipment
• Access to a computer or vehicle
• Access to travel on trains/buses
• Access to leisure facilities
Access control for vehicles:
• Secure access on site
• Road tolling
• Instant payment for fuel
Manufacturing automation:
• Control of flexible manufacturing processes by recognizing items being built on a production line
(mass customization enabler)
• Labeling key components for later recycling
Logistics and distribution:
• Tracking parcels from shipment to end customer
• Tracking goods from manufacture through to retail
Retail:
• Supply chain management
• Stock taking
• Reducing loss through shrinkage
• Reverse logistics
• Product availability

159
Maintenance:
• Plant & Equipment
• Fixed assets
• Patients
Product security:
• Tamper evidence
• Product authentication
• Anti-counterfeiting

13.4 RFID technology in the supply chain

The efficiency of the supply chain has a direct impact on the profitability of a company. It is no
surprise therefore to find that many large corporate companies have made it a key part of their
strategy, and invested heavily in software systems (ERP, WMS.) and IT infrastructure designed to
control inventory, track products and manage associated finance. RFID will bring a new dimension to
supply chain management by providing a more efficient way of being able to identify and track items
at the various stages throughout the supply chain. It will allow product data to be captured
automatically, and therefore be more quickly available for use by other processes such as stock
management and real time billing.
Even though RFID applications are still at the early stages of deployment, many companies running
pilot systems have been able to demonstrate some of the significant benefits that RFID promises.
There is no doubt that more will be discovered as the industry adopts the technology on a wider scale.
The following are examples of what has been identified so far by the different studies and tests/pilots
recently completed within the supply chain.
Advanced Shipping Notices (ASN): RFID is able to automatically detect when either a pallet or
shipment has left the warehouse or Distribution Center. This will allow to not only generate an
electronic ASN and notify the recipient, but also to bill clients in real time instead of waiting until the
end of the week or month, and doing a batch operation.
Shrinkage: One of the major problems in the supply chain is product loss or shrinkage, which can
account for anything from 2 to 5 % of stock. The causes may vary from misplaced orders, employee
and customer theft or inefficient stock management. RFID with its superior tracking and identification
capability will be able to localize where losses are occurring.
Returned Goods: Full visibility and automation can be potentially achieved on returned goods-
thereby reducing fraud.
Anti –counterfeit: Illegal duplication and manufacture of high value products, is one of the industries
most well known problems. By integrating a tag into items, for example the body of an expensive
ladies handbag, RFID has the potential to authenticate a product, and combat the sale of false goods
on the black markets.
Supply Chain efficiency: RFID will enable the traceability and reduction in the number of
discrepancies between what a supplier invoiced, and what a customer actually received.
Improved stock management: Managing stock is the key priority for many retailers. Studies have
shown that on average, products are not on the store shelves 7% of the time due to inefficiencies in
stock management, which means of course a potential purchase loss. Implementing RFID at the item
level and on shelves will give an automatic way of knowing and managing stock levels. However in
order to achieve this on a large scale, it is recognized that tags will have to come down in price to
around 5 cents or less, and readers to around 100 USD.

160
Reduction in labor costs: At DC ‘s (Distribution Centers) labor accounts for nearly 70% of costs. It is
estimated that RFID could reduce this by nearly 30% by removing the need for manual intervention
and use of barcodes when loading cases or stocking pallets.

13.5 Enabling RFID Adoption Move in the Supply Chain – The Electronic Product Code
At the heart of the current RFID based technology drive to improve supply chain efficiency and
reduce operating costs, is the EPC (Electronic Product Code).

13.5.1 EPC Origins

In October 1999 the Auto-ID center was created in the Department of Mechanical Engineering by a
number of leading figures at MIT. The potential benefits of RFID tags had been identified long
before, what was stopping the adoption of the technology in the supply chain was the cost of the tags.
The AutoID recognized that in order to solve this problem, tags needed to be as simple as possible,
and act instead as pointers to information held on servers in the same way as information is stored on
the internet. This lead to the idea of the EPC (Electronic Product Code) which would provide fast and
detailed information of products anywhere in the supply chain. The goal however, was not to replace
bar codes, but rather to create a migration path for companies to move from bar code to RFID. The
Auto-ID Center officially closed on October 26th, 2003. The final board meeting was held in Tokyo,
Japan. The Center had completed its work and transferred its technology to EPCglobal
(www.epcglobalinc.org), which will administer and develop EPC standards going forward.
13.5.2 EPC layout
The code is similar to the UPC (Universal Product Code) used in bar codes, and ranges from 64 bits to
256 bits with 4 distinct fields described below in fig 3. .What sets the EPC apart from bar codes is its
serial number which allows to distinguish the uniqueness of an item, and track it through the supply
chain.

Fig 3. Layout of an EPC which is 96 bits in length

• Header (0- 7) bits: The Header is 8 bits, and defines the length of the code in this case O1 indicates
an EPC type 1 number which is 96 bits in length. The EPC length ranges from 64 to 256 bits.
• EPC manager (8- 35) bits: Will typically contain the manufacturer of the Product the EPC tag is
attached to
• Object Class (36-59) bits: Refers to the exact type of product in the same way a an SKU (Stock
Keeping Unit)
• Serial Number (60 – 96) bits: Provides a unique identifier for up to 296 products

13.5.3 EPC infrastructure

161
 Fig. 4 The basic steps of EPC infrastructure

EPC infrastructure will allow immediate access to information, which will not only optimize existing
services such ASN, but also have the potential to create new services, for example; a retailer could
automatically lower prices as the expiry date approaches, or a manufacturer could recall a specific
batch of products due to health concerns, and if needed pinpoint source of the problem down to a
unique product.
Middleware or Savant Software The sheer potential volume of data created by billions of EPC tags
would very quickly grind most existing companies’ enterprise software and IT infrastructure to a
standstill within a matter of minutes. The answer to this problem is middleware or Savants. RFID
savants serve as a software buffer which sits almost invisible between the RFID readers, and the
servers storing the product information. It allows companies to process relatively unstructured tag
data taken from many RFID readers, and direct it to the appropriate information systems. Savants are
able to perform many different operations, such as monitor the RFID reader devices, manage false
reads, cache data and finally query an Object Naming Service (ONS).

162
Object Name Service (ONS) ONS matches the EPC code to information about the product via a
querying mechanism similar to the DNS (Domain Naming system) used in the internet, which is
already proven technology capable of handling the volumes data expected in an EPC RFID system.
The ONS server provides the IP address of a PML Server that stores information relevant to the EPC.
Physical Markup Language (PML) Whilst the EPC is able to identify the individual product, the
real useful information is written in a new standard software language called Physical Markup
Language. PML itself is based on the widely used and accepted extensible markup language (XML),
designed as a document format to exchange data across the internet. It is not surprising therefore, with
so much of the infrastructure for EPC being borrowed from the internet (DNS,XML..), that it is often
referred to as ¨ the internet of things ¨.
PML is designed to store any relevant information about a product; for example,
(1) Location information e.g., tag X was detected by reader Y, which is located at loading dock Z;
(2) Telemetry information [Physical properties of an object e.g., its mass; Physical properties of the
environment, in which a group of objects is located, e.g., ambient temperature];
(3) Composition information e.g., the composition of an individual logistical unit made up of a pallet,
cases and items. The information model will also include the history of the various information
elements listed above e.g., a collection of the various single location readings will result in a location
trace.
(4) Manufacturing and expiry dates

Figure 5 explains how EPC will automate the supply chain.

13.6 Adoption strategy

An adoption strategy provides a roadmap to implement the technology in a way that is consistent with
an organization’s strategic vision and goal. A road map comprises four essential fundamental
principles for RFID to be thoroughly adopted in the organization.
Shared Understanding: In an organization, each decision maker or potential stakeholder of the RFID
adoption may have a considerably different understanding from his or her own perspective towards
RFID. The likelihood of successful adoption of RFID in effect hinges on the capability of the
organization to enforce a broader shared understanding.
Goal Setting: The organization has to set a very clear unambiguous goal for RFID adoption. This
goal is to be aligned with the organization’s existing business goal. With the goal, important
stakeholders can thus be explicitly (rather than implicitly or potentially) identified.
Justification: A cost-benefit analysis (e.g., ROI estimation) to suffice the financial concern will
significantly increase the possibility of RFID being accepted by those hesitant stakeholders. Last,
potential challenges and issues during and after deploying RFID have to be clearly identified at the
early stage. Each challenge should be addressed by possible solutions submitted to all involved
stakeholders.
Planning and Approaching: A good project plan includes evaluation of technology option, standard-
based deployment approach, measurements and optimization techniques, and a continuous
improvement roadmap. During the implementation, an incremental approach shall be taken. Thus
pilot tests and milestones can be used for checking the progress of the implementation and ensuring
the desired outcome is achieved through the delta part implemented in the current adoption iteration.
Strategic step based on these aforementioned principles are explored below.

163
 Fig 5. Using EPC

Steps in RFID Adoption:

Core Competency Reaffirmation: The core competency (CC) is defined as one thing that an
organization can do better than its competitors, and is crucial to its success. Before the organization

164
adopts any candidate technologies, it has to assess them against the core competency. For example,
when retailer giant Wal-Mart realizes that the core competence lies in its dominant distribution
channels, which can greatly benefit from RFID technology, it demands all of its top 100 suppliers to
attach RFID tags onto their goods. On the contrary, if a company XYZ’s CC is manufacturing high-
quality automobile engine assemblies, it should be very cautious in adopting RFID unless its primary
retailers at the downstream urge it to do so, because the distribution is not XYZ’s core competencies
and doing so will only distract itself from doing what they are good at. Therefore, in general, an
organization should be wary of taking aggressive steps in implementing RFID solutions before the
core competence has been thoroughly reaffirmed and evaluated.

Feasibility Analysis: If the RFID technology is being considered to be aligned with the CC, it is time
for the organization to perform the feasibility analysis, which deals with four major aspects:
• It is very important to estimate the capability of existing information systems. Since RFID will
dramatically increase the amount of the data captured from instance level tags, the information
systems have to capture, process, and analyze such a huge amount of data efficiently. Any
insufficiency of information systems will definitely hinder the RFID technology to realize its full
potential.
• RFID solutions could be quite costly; hence continuous and dedicated RFID funding support is
essential.
• Personnel preparation is another feasibility issue needing consideration. Once implemented, the
RFID might change the business process as well as fundamental information systems operations,
which incur substantial IT training and adaptation study. The learning capability of involved
working staff also determines the success of RFID adoption.
• Lastly, the organization has to choose the appropriate time to deploy RFID. Early adoption of
RFID has advantages as well as risks. The feasibility of time should study whether the company is
ready for RFID at that particular time, and moreover can bear the risk associated, even if the
prerequisite conditions are all met.

Candidate Scenarios
The next step is to identify candidate scenarios that would benefit from RFID and to measure the
potential benefit in a self-defined scale. For example, for manufacturing units, RFID can be used to
support quality control by querying components and subassemblies as they enter the facility. This is a
typical candidate scenario that can be identified. It is recommended for the organization to enumerate
all the potential RFID-involved candidate scenarios that can impact an organization’s core
competences. A typical tabular description of a sample candidate scenario (based on RosettaNet,
2001) is presented in Table 1.
Scenario Prioritization
In this step, the company need to prioritize this list against a set of criteria such as estimated impacts
to CC, the ROI, the cost, and so forth. Such a prioritizing process also needs to consider the
preference from different RFID stakeholders. This is achieved by:
1. organizing the criteria set into a hierarchical structure
2. performing the pair-wise comparison between any two candidate scenarios against one specific
criterion
3. providing the pair-wise comparison between any two criteria for each RFID stakeholder
4. computing the stakeholders-aggregated preferences for each criteria
5. calculating the overall weight for each scenario allowing for all criteria with different preferences
ranking the scenario list against the weight value generated in #5, with the biggest weight value

165
 being positioned in the first rank. This step eventually generates a Prioritized Scenario List for
further justification review.
Justification Review:
In this step, for each candidate scenario in the Prioritized Scenario List, the organization conducts the
justification review, a static analysis of RFID deployment only related to that particular candidate
scenario. For example, traditional ROI (return on investment) methods can be utilized to examine
whether RFID technology should be deployed in this candidate scenario. For each candidate scenario,
if the justification review produces negative results, it is removed from the prioritized scenario list,
and the next scenario will be considered to perform the justification review. Otherwise, this scenario
will be subsequently chosen for the following pilot test.
Adoption Issues
Before the pilot test, a clear consensus of issues and their solutions is necessary among all the
stakeholders. Hence in this step, the stakeholders need to unanimously identify adoption challenges
specific to this scenario. The organization itemizes all the possible problems that it might encounter
during the pilot test. More importantly, the solutions that address these challenges should be proposed
and planned in this step. This ensures the smooth progress of the pilot test. Moreover, the solutions
can also be validated and modified during the pilot test.
Pilot Test
Once the justification review is confirmed positively, the pilot implementation can be carried out in an
experimental environment. It can be an RFID prototype in a smaller scope or scale of this candidate
scenario. For example, a pilot deployment in one or two locations allows evaluation of RFID vendors,
equipment, and software, and provides the opportunity for different stakeholders to gain experience
with RFID. Furthermore, such a pilot is to produce the impact analysis for this candidate scenario. As
many impacts can be associated with the RFID deployment, some of them are beneficial to some
shareholders, while some are negative to other shareholders. Hence the pilot implementation can
estimate such impacts brought by RFID deployment. If most of the results tend to be negative, the
likelihood that the RFID to be implemented in this scenario appears to be very low. The organization
might need to consider the next candidate scenario along the Prioritized Scenario List. In contrast,
positive pilot test results with the pilot feedback suggest the start of the RFID implementation in this
candidate scenario.
Implementation
Once the pilot test is passed, the formal implementation can be carried out in an incremental manner
through a set of iterations with a couple of milestones, which ensure that the implementation cost and
risk can be controlled and mitigated to the minimal level. Each milestone has different focuses. Table
3 lists possible sequential tasks within one milestone. Particular attention is given for fostering the
transition between milestones.

166
2

167
 3

13.7 RFID Adoption Challenges

Several key adoption issues will be discussed in this section. The main issues that we address in this
section are cost associated with the deployment of RFID system, security and privacy concerns, and
finally more technical issues in deployment of an RFID system.
Cost
A cost-estimation model for a full-fledged deployment of an RFID system in a supply chain environ-
ment should consider the following factors.
RFID Tags: When deploying an RFID system, one should consider the cost of buying RFID tags. It is
a good idea to consider renewable tags (if possible) as a means to reduce cost. Normally the cost of
active tags is more than the passive tags. Apart from the cost of the tags, companies should also
consider the cost of testing the passive tags. Finally there is cost associated with replacing the
defective tags.
RFID Printers: An RFID label has a similar functionality as an RFID tag. However it can be stuck on
like a label. The RFID label can be printed using RFID printers. Hence if you are planning to use
RFID labels, the cost of the RFID printer should be considered. In some applications RFID labels are
much more preferred because of the environment and the products. For example applications like
express parcel delivery, library book/video checkout, sensitive document tracking, ticketing (sports,
concerts, ski lifts, etc.), and pharmaceuticals prefer RFID labels (Zebra, 2006).
RFID Readers: When deploying an RFID system, one should consider the cost of buying RFID
readers. The fixed readers are normally cheaper than the portable readers. Dumb readers are usually
cheaper, as they do not have any computing capability. On the other hand intelligent readers offer
computing capability to filter data, store information, and execute commands. Agile readers can
communicate with tags using a variety of protocols, while multi-frequency readers can read tags using
different frequencies. All these features contribute to the cost of readers, and the organization should
select a proper reader based on its application requirements.
RFID Antennas: Almost all the readers are equipped with one or more antennas. However in some
cases the need for additional high-power antennas cannot be ruled out and hence this additional cost
should be considered before deciding to deploy an RFID system.

168
RFID Middleware: RFID middleware contributes a major portion of RFID investment. Many vendors
supply RFID middleware, and the cost can vary depending upon the capabilities of the middleware.
Usually factors that contribute to cost include complexity of the application and the number of places
the middleware would be installed. Apart from the middleware, the companies should also consider
the cost of edge servers, which are normally deployed in the warehouse, distribution center, or
production facility. The edge servers are simple servers, which are connected to the RFID reader
using a Universal Serial Bus (USB) port.
Training Existing Staff: The organization will need to train its employees, particularly engineering
staff who will manage readers in manufacturing and warehouse facilities, and IT staff who will work
on the systems that manage RFID data.
Hiring Technology Expertise: Most of the companies, as of now, would not have the expertise to
deploy a complete RFID system. This is partly attributed to the fact that RFID is a relatively new
technology. Hence an organization would need to outsource this task to a third party who knows how
to install the readers, decide the appropriate location for fixing the tag on the products, ascertain that
the data gathered by the reader is properly propagated to the middleware in the right format, and so
on. This is quite important because RFID systems can be sometimes difficult to install, as there are
several factors that can affect the optimum performance of such a system. Hence a major portion of
RFID investment has to be targeted to this area.
Other Miscellaneous Costs: The miscellaneous costs might include regular maintenance of the RFID
readers or replacement of damaged tags or antennas.

Cost Estimation for RFID Deployment in Supply Chain Tracking

When deploying an RFID system for inventory management and control in a supply chain, all the
above-mentioned costs should be considered. According to Forrester Research, the estimated cost of
middleware is around $183,000 for a $12 billion manufacturer looking to meet the RFID tagging
requirements of a major retailer. In the same manner the estimated price of $128,000 could be spent
for consulting and integration, $315,000 for the time of the internal project team, and $80,000 for tag
and reader testing. A simple estimate is provided in Table 3.
On the lower end an estimated cost of around $3 million should be invested in an RFID project for
inventory tracking and management if a total of 10 million items are tagged. The number of readers,
printers, and edge servers would vary depending upon the number of distribution centers and
warehouses in use. Here we assumed 100 readers and printers, and 50 edge servers. These numbers
are used just as an example; it can vary depending upon each company.

169
Read-Rate Accuracy
Achieving 100% read-rate accuracy is a major adoption challenge with RFID deployment. Supply
chains and warehouse management solutions based on RFID are highly vulnerable to read-rate
inaccuracy because of the number of RFID-tagged items that need to be scanned every second.
Consider the scenario when a palette containing 1,000 RFID-tagged items is scanned at the warehouse
exit. There is a high probability that the reader would not scan a few tags. It is difficult to list the main
reasons that result in inaccurate readings because there are too many “ifs” and “buts”. Accuracy is
dependent on so many unrelated variables that it is difficult to list the main factors behind the cause.
However we attempt to outline some basic parameters, which results in inaccurate readings. Some of
the main reasons for inaccurate readings include the environment in which RFID system works,
material of the item being tracked, reader configuration, reader and tag placements, tag orientation,
and so forth. To successfully deploy an RFID system, some key parameters should be considered to
achieve accurate readings.

Tagged Material
Maintain some consistency when tracking materials. It is not a good idea to standardize the reader
configuration to track cartons, trolleys, pallets, glass materials, documents, or metal or plastic bins.
This is because different materials behave differently to RF energy; some materials are RF friendly,
while others are RF absorbent or RF opaque. A reader configured to read tags from RF-friendly
material would definitely fail to give 100% read-rate accuracy if used to track RF-absorbent or RF-
opaque items.

Preplanned Object Movement

To assure good read rates, it is advised to move the tagged objects on a predefined route (or pattern).
You cannot expect good rates if the cartons are moved through a forklift, people, metals trolleys,
plastic trolleys, and so forth. There should be just one or two modes of transport well tested for 100%
read-rate accuracy.

Tags from Different Vendors

Read rate is also affected if tags are used from different vendors because the performance of such tags
varies significantly. Another reason is the use of different standard-compliant tags (like EPC Gen1 &
EPC Gen2). It is also difficult to configure the reader power level at an optimum level where it
supports all different tags with 100% read rate. Ideally one should try to use a single standard and
single vendor tag in one ecosystem. Nevertheless this may change as the standards improve.

Tag Orientation
Orientation is one of the big factors for providing good read rates. Even though dual dipole tags
perform much better in all orientations, it is still advised to follow a policy on tag placement and tag
orientation (TPTO). A standard policy on TPTO across an organization would definitely improve the
read-rate accuracy.

13.8 Security in RFID Systems

RFID tags used in the supply chain will contain data ranging from simple ID numbers (EPC), to more
important information about a product. For example in the health industry, it could be the blood type
of a sample. The main goal of any security system designed to protect data stored in mediums such as
tags, computer disk drives, or smart cards is basically to prevent any unauthorized person from being
able to either;

170
a) Obtain access and learn the data contents
b) Obtain access and modify/corrupt/erase the data contents
c) Copy the data contents to a similar storage device (duplicate)
In a complete system, security of data as defined above not only involves the storage medium, but
also how data is created and transferred from a host to the medium (or vice versa). For example, when
an engineer broke the security of a French bank credit card a few years ago, he did it not by
compromising the chip security, but by hacking the reader terminal. The following are scenarios that
could happen in the supply chain.
1) Industrial Sabotage – somebody with a grievance against a company decides to start corrupting
data in tags by using a hand held device, and erasing or modifying the contents.
2) Industrial Espionage – A rather unlawful competitor would like to know how many, and what
type of products are being manufactured, and shipped by your company. He could possibly achieve
this in the following ways
i. Eavesdropping – listening in on longer range communication systems like UHF which
broadcast signals (albeit very weak) up to 100 meters – some protocols have a basic security
which ensures that the ID N° is never transmitted completely in one stream.
ii. Placing bogus well concealed readers linked to a PC somewhere in the proximity of the
tags moving through the production line
iii. Using hand held devices
3) Counterfeiting – Being able to read or intercept data being written into a tag which uniquely
identifies or certifies a product. Once the data is known, similar read/write tags could be purchased
and updated with the authentic data, thus creating the real possibility of counterfeiting products which
are supposed to be protected by a tag. All the above scenarios are potential risks if no security is
implemented in the tag and reader. The importance attached to protecting data in the supply chain will
depend on the application, and the company’s strategy towards security. In some cases legislation will
impose it. Of course bar codes which are used today, can be easily read, decrypted, and even
destroyed, but not on the wide-spread and automatic scale possible with RFID. Even the simplest
security costs silicon area, and therefore will impact on the final tag price. This goes against the
current trend of trying to produce the smallest and cheapest tag possible. Every company is therefore
faced with this tradeoff between cheaper unsecured tags, and the potential security risks they entail.

13.9 Privacy
If RFID is deployed in a full scale, it may result in many privacy concerns because RFID can be used
to track consumer behavior, which can further be used to analyze consumer habits. It can even be
used for hidden surveillance, for example, deploying secret RFIDs for tracking. With the size of
RFIDs reducing day by day, it has now become possible to hide them within products without the
owners’ consent. For example, RFID tags have already been hidden in packaging. A scenario of
hidden RFID testing was discovered in a Wal-Mart store in Broken Arrow, Oklahoma, where secret
RFID readers tracked customer action. Criminals with RFID readers can look for people carrying
valuable items and can launch selective attacks. However most of these issues can be tackled by
privacy enforcement laws, which can be incorporated into the nation’s legal framework. All these
privacy issues have created a lot of fear in the consumer community. In order to address these issues,
several approaches are proposed in the literature.

13.10 RFID Implementation Examples

In an academic study performed at Wal-Mart, RFID reduced Out of Stocks by 30 percent for products
selling between 0.1 and 15 units a day. Second, the RFID technology can prevent or reduce the

171
sources of errors. Benefits of using RFID include the reduction of labour costs, the simplification of
business processes and the reduction of inventory inaccuracies.

Wal-Mart and the United States Department of Defense have published requirements that their
vendors place RFID tags on all shipments to improve supply chain management. Due to the size of
these two organizations, their RFID mandates impact thousands of companies worldwide. The
deadlines have been extended several times because many vendors face significant difficulties
implementing RFID systems. In practice, the successful read rates currently run only 80%, due to
radio wave attenuation caused by the products and packaging. In time it is expected that even small
companies will be able to place RFID tags on their outbound shipments.

Since January, 2005, Wal-Mart has required its top 100 suppliers to apply RFID labels to all
shipments. To meet this requirement, vendors use RFID printer/encoders to label cases and pallets
that require EPC tags for Wal-Mart. These smart labels are produced by embedding RFID inlays
inside the label material, and then printing bar code and other visible information on the surface of the
label.
The Table below lists few more initiatives and success stories.

13.10 Conclusions
The need to organize and make decisions based on the data provided by the RFID tags is prominent.
To date, research that conjoins RFID technology and item- level inventory management on the
shop floor is at a preliminary stage, only inferring benefits upon application. The challenge is to
collect RFID data in a timely manner, to process such voluminous data, and to make timely decisions
that are tied into manufacturing execution systems. If the challenge is overcome, then the benefits
such as waste elimination, inventory reduction, automatic replenishment, stock-out reduction and
overall cost savings can be easily realized. Therefore, there is a need for RFID data-based effective
decision making algorithms that can lead to such benefits. In this study, a forecasting integrated
inventory management model that relies on real-time RFID data is presented. The goal of this
research is threefold: (1) to model and analyze the decisions made to manage inventory levels of
time sensitive materials in a shop floor manufacturing environment; (2) to investigate new
decision making algorithms that substantiate the use of RFID for real- time visibility; and (3) to
explore the impact of fundamental control parameters on performance measures.

References:
[1] V. Potdar, C. Wu, E. Chang, E-Supply Chain Technologies and Management, chapter ‘Automated
Data Capture Technologies – RFID,’ in IDEA Group Reference, Hershey, PA, USA, March 2007
[2] Radio Frequency Identification news and commentary, http://www.rfidgazette.org/walmart/
[3] A BASIC INTRODUCTION TO RFID TECHNOLOGY AND ITS USE IN THE SUPPLY
CHAIN, Steve Lewis, White Paper from Ship2save
ship2save.com/page_images/wp_printronix_rfid_supplychain.pdf
[4] Want, R., An introduction to RFID technology, Intel Res., Santa Clara, CA, USA; IEEE Pervasive
Computing, Jan.-March 2006
[5] Introduction to RFID Technology, http://www.rfidc.com/docs/introductiontorfid_technology.htm
[6] Radio frequency identification, http://en.wikipedia.org/wiki/RFID
[7] Tony Gale, Divakar Rajamani, Chelliah Sriskandarajah, The Impact of RFID on Supply Chain
Performance, Working Paper, School of Management, University of Texas at Dallas
http://som.utdallas.edu/c4isn/documents/c4isn-Impact-RFID-SC-Perform.pdf

172
173
174
 CHAPTER XIV
Dynamic Vehicle Routing with GPS and GIS

14.1 Introduction
Logistics especially distribution management is a vital component of supply chain management.
Physical distribution includes a set of activities executed to obtain the delivery of a product from the
production location to the end customer. Efficient distribution of goods entails, among other things, a
determination of routes and schedules for the fleet of vehicles so that total distribution costs are
minimized, while various requirements (constraints) are met. The constraints concern various facets
of the operation, such as vehicle capacities, time windows on pick up and/or delivery, time
availability of vehicles, etc.
The most central model in distribution management is the vehicle routing problem (VRP). In the
standard capacitated vehicle routing problem (CVRP) a homogeneous fleet of vehicles serves a set of
customers from a single distribution centre such that:
• the fixed capacity of a vehicle cannot be exceeded;
• each customer has known demand that must be satisfied;
• the demand of each customer is satisfied by exactly one visit of a single vehicle;
• each vehicle must leave and return to the distribution centre.
The objective is to generate a sequence of deliveries for each vehicle so that all customers are serviced
and the total distance traveled by the fleet is minimized. However, there exist several important
problems that must be solved in real-time. In what follows, we review the main applications that
motivate the research in the field of the real-time VRPs.
(i) Dynamic fleet management: Several large-scale trucking operations require real-time dispatching
of vehicles for the purpose of collecting or delivering shipments. Important savings can be achieved
by optimizing these operations.
(ii) Vendor-managed distribution systems: In vendor-managed systems, distribution companies
estimate customer inventory level in such a way to replenish them before they run out of stock.
Hence, demands are known beforehand in principle and all customers are static. However, because
demand is uncertain, some customers (usually a small percentage) may run out of stock and have to
be serviced urgently.
(iii) Couriers: Long-distance courier need to collect locally outbound parcels before sending them to a
remote terminal to consolidate loads. Also, loads coming from remote terminals have to be distributed
locally. Most pick-up requests are dynamic and have to be serviced the same day if possible.
(iv) Rescue and repair service companies: There are several companies providing rescue or repair
services (broken car rescue, appliance repair, etc.).
(v) Dial-a-ride systems: Dial-a-ride systems provide transportation services to people between given
origin–destination pairs. Customers can book a trip one day in advance (static customers) or make a
request at short notice (dynamic customers).
(vi) Emergency services: Emergency services comprise police, fire fighting and ambulance services.
By definition, all customers are dynamic. Moreover, the demand rate is usually low so that vehicles
become idle from time to time. In this context, relocating idle vehicles in order to anticipate future
demands or to escape from downtown rush hour traffic jam is a major issue.
(vii) Taxi cab services: In taxi cab services, almost every customer is dynamic. As in emergency
services, relocating temporary idle vehicles is an issue.
Due to recent advances in information and communication technologies, vehicle fleets can now be
managed in real-time. When jointly used, devices like geographic information systems (GIS), global
positioning systems (GPS), traffic flow sensors and cellular telephones are able to provide relevant
real-time data, such as current vehicle locations, new customer requests and periodic estimates of road
travel times. If suitably processed, this large amount of data can be in principle be used to reduce cost
and improve service level. To this end, revised routes have to be timely generated as soon as new
events occur. In recent years, algorithms relevant in a real-time context have been developed to handle
vehicle routing problem. These algorithms are celled real-time vehicle routing algorithm or the
dynamic vehicle algorithm.
14.2 Technical Requirements
In this section we provide a brief introduction to some of the most essential technologies when
dealing with real-life applications of vehicle routing problems within a dynamic environment.
Communication and Positioning Equipment
The communication between the drivers of the vehicles and the dispatching center is essential in order
to feed the most up-to-date information into the routing system. The equipment for determining the
current position of the vehicles and the communication equipment for passing information on between
the dispatching center and the drivers in the vehicles will be introduced below.
• Naturally, positioning equipment like the GPS (Global Positioning System) is essential to a
dynamic vehicle routing system. The GPS is a constellation of 24 satellites orbiting Earth that
constantly send out signals giving their positions and time. Signals from three or four different
satellites at any given time can provide receivers on the ground with enough information to
calculate their precise location within a few meters depending on which version of the GPS system
is used.
• The communication equipment between the vehicle and the dispatching center is essential for the
structure of the routing system. Mobile telephone communication systems are one example of a
technology capable of providing this information. Another technology is a dedicated radio based
communications system. The main difference in these technologies is the differences in initial and
operating costs. A mobile telephone communications system is relatively costly to operate, but has
low initial costs because the basic technology is provided by the telephone companies and the
GSM system today offers almost full coverage in most western industrialized countries. The initial
costs in implementing a radio based communications system are on the other hand very high,
because transmission masts will have to be put up and relatively expensive radio equipment must
be installed in every vehicle. In all, a radio based communication system has very high initial
costs, while the operating costs are almost negligible. Furthermore, the radio based system does
not offer the same flexibility compared to the mobile telephone communications system.
In Figure 1.2 the basic information flows between the vehicle and the dispatching center are shown.
Ideally, the dispatching center will know in which state the vehicle and the driver are at any given
point in time. However, as the above description indicates, this may prove to be infeasible for some
applications due to the operating costs of this method. However, within a real-life setting the
positioning information is transmitted at fixed intervals and an interpolation scheme is employed in
order to estimate the positions of the vehicles. Alternatively, the driver sends a message about his
current status and position to the dispatch center, each time he finishes the service at a customer.
Obviously, this approach does not offer the same level of information for the dispatcher to support her
decision as to which vehicle to dispatch to the next customer to be served. If the new information
provided by the now idle driver/vehicle makes the dispatcher change her mind on the current planned
route, she will have to call the other drivers manually to inform them about the changes in the current
routes. However, the conclusion of this discussion must be that a careful analysis will have to show
which approach to choose when designing the system. Of course, history shows that the prices of
communication decrease rapidly over the years. This could be the motivation to go for
telecommunications based system. We below describe the principles behind GPS – crucial component
of the communication system.

176
Figure 1. Sketch of the information flow in a GPS based vehicle routing system.

Global positioning system (GPS)

Global Positioning System (GPS) is a technology that uses the position of satellites to determine
locations on earth anytime, in any weather, anywhere. GPS is a constellation of U.S. Government
satellites providing the most advanced and accurate positioning and navigation service. Twenty-four
GPS satellites orbit 12,000 miles above the earth, constantly transmitting the precise time and their
position in space. The GPS receivers listen to these satellite signals and use the information to
determine the location of the receiver, as well as how fast and in what direction it is moving. GPS
satellites circle the earth twice a day in a very precise orbit and transmit signal information to earth.
GPS receivers take this information and use the principle of triangulation to calculate the user's exact
location. Satellites are equipped with very precise clocks that keep accurate time to within three
nanoseconds - that’s 0.000000003, or three billionths, of a second. This precision timing is important
because the receiver must determine exactly how long it takes for signals to travel from each GPS
satellite.
GPS has 3 parts: the space segment, the user segment, and the control segment. The space segment
consists of 24 satellites, each in its own orbit 11,000 nautical miles above the Earth. The GPS
satellites each take 12 hours to orbit the Earth. The user segment consists of receivers, which we can
hold in your hand or mount in the vehicle. These receivers detect, decode, and process GPS satellite
signals. The control segment consists of ground stations (five of them, located around the world) that
make sure the satellites are working properly.
Geographic Information Systems (GIS)

Transportation data is usually associated with spatial data, like traffic counts from particular sites, the
traffic volumes along particular roads or links, etc. Geographical Information System (GIS) can be
used as a database for storing transportation data. The primary advantage of using GIS as a database
for transportation data is the fact that GIS can integrate the spatial data and display the attribute data
in a user-chosen format. The chief sources of spatial data are the existing digitized files (e.g.:
Topologically Integrated Geographic Encoding and Referencing (TIGER) files in the US). The Global
Positioning System (GPS) is widely being used as a tool for collecting the spatial data. Systems which

177
chiefly use GPS as a spatial data source for a GIS are called as GPS-GIS integrated systems. Most
western industrialized countries now have almost fully detailed road network databases.

14.3 The Vehicle Routing Problem

Vehicle routing refers to a broad group of problems that could be expressed as following: a finite set
of customers at fixed locations with defined demand, must be supplied with goods by a number of
vehicles having a finite capacity and predefined starting points and terminals. The vehicle routing and
scheduling problem consists of two sub-problems: the customer grouping to routes (clustering) and
the definition of the optimum tour for every route (cluster). Therefore, route is the total number of
deliveries made by a single vehicle and tour is their sequence. The solution of these sub-problems
results to the routes and tours that minimize the total transportation cost.
The VRP is a combinatorial problem whose ground set is the edges of a graph G(V,E). The notation
used for this problem is as follows:

• V = {v0 , v1 ,..., vn } is a vertex set, where:

o Consider a depot to be located at v0 .

o Let V ′ = V \ {v0 } be used as the set of n cities.

• A = {(vi , v j ) | vi , v j ∈ Vi , i ≠ j} is an arc set.

• C is a matrix of non-negative costs or distance cij between customers vi and vj.
• d is a vector of the customer demands.
• Ri is the route for vehicle i.
• m is the number or vehicles (all identical). One route is assigned to each vehicle.

When cij=cji for all (vi , v j ) ∈ A the problem is said to be symmetric and it is then common to replace
A with the edge set E = {(vi , v j ) | vi , v j ∈ Vi , i < j} .

With each vertex vi in V ′ is associated a quantity qi of some goods to be delivered by a vehicle. The
VRP thus consists of determining a set of m vehicle routes of minimal total cost, starting and ending
at a depot, such that every vertex in V ′ is visited exactly once by one vehicle.

We will consider a service time δ i (time needed to unload all goods), required by a vehicle to unload
the quantity qi at vi. It is required that the total duration of any vehicle route (travel plus service times)
may not surpass a given bound D, so, in this context the cost cij is taken to be the travel time between
the cities. The VRP defined above is NP-Hard.

A feasible solution is composed of:

• a partition R1,…, Rm of V;
• a permutation σ i of Ri ∪ 0 specifying the order of the customers on route i..

The cost of a given route ( Ri = {v0 , v1 ,..., vm +1 } ), where vi ∈ V for i = 1..m and v0 = vm +1 = 0 (0
m m
denotes the depot), is given by: C ( Ri ) = ∑ cij +∑ δ ij .
i =0 i =0

178
A route Ri is feasible if the vehicle stop exactly once in each customer and the total duration of the
route does not exceed a prespecified bound D: C ( Ri ) ≤ D .

m
Finally, the cost of the problem solution S is: FVRP ( S ) = ∑ C(R ) .
i =1
i

The characteristics of the attributes of the information forming the input for the vehicle routing
problem are as follows.
• Evolution of information: In static settings the information does not change, nor is the information
updated. In dynamic settings the information will generally be revealed as time goes on.
• Quality of information: Inputs could either; 1) be known with certainty (deterministic), 2) be
known with uncertainty (forecasts) or 3) follow prescribed probability distributions (probabilistic).
Usually, the quality of the information in a dynamic setting is good for near-term events and
poorer for distant events.
• Availability of information: Information could either be local or global. One example of local
information is when the driver learns of the precise amount of oil the current customer needs,
while a globally based information system would be able to inform the dispatcher of the current
status of all the customers' oil tanks. The rapid advances within information technologies increase
the availability of information. This fast growth in the amount of information available raises the
issue of when to reveal/make use of the information. For instance, the dispatcher may choose to
reveal only the information that is needed by the drivers although she might have access to all
information.
• Processing of information: In a centralized system all information is collected and processed by a
central unit. In a decentralized system some of the information could for instance be processed by
the driver of each truck.

A VRP is said to be static if its input data (travel times, demands,. . .) do not depend explicitly on
time, otherwise it is dynamic. Moreover, a VRP is deterministic if all input data are known when
designing vehicle routes, otherwise it is stochastic.
The Static Vehicle Routing Problem can be defined as the one in which:
1. All information relevant to the planning of the routes is assumed to be known by the planner before
the routing process begins.
2. Information relevant to the routing does not change after the routes have been constructed.
A static problem can be either deterministic or stochastic. In deterministic and static VRPs all data
are known in advance and time is not taken into account explicitly. In stochastic and static VRPs
vehicle routes are designed at the beginning of the planning horizon, before uncertain data become
known. Uncertainty may affect which service requests are present, user demands, user service times
or travel times. If input data are uncertain, it is usually impossible to satisfy the constraints for all
realizations of the random variables. If uncertainty affects the constraints but the objective function is
deterministic, it can be required that constraints be satisfied with a given probability (chance
constrained programming, CCP). In a more general approach, a first phase solution is constructed
before uncertain data are available and corrective (or recourse) actions are taken at a second stage
once all the realizations of the random variables become known. The objective to be minimized is the
first stage cost plus the expected recourse cost (stochastic programming with recourse, SPR).

179
14.4 The dynamic vehicle routing problem
It is argued that a relatively high number of the routing problems being modeled as static problems do
in fact include dynamic elements in a real-life situation. Examples of this phenomenon is the fact that
on site service times as well as travel times despite extensive empirical data are often filled with
noise. This implies that the preplanned routes collapse because of the new temporal conditions of the
system. That is what seemed to be an optimal solution (or at least a good quality solution) might turn
out to be a sub-optimal solution.
In a dynamic Vehicle Routing Problem setting
1. Not all information relevant to the planning of the routes is known by the planner when the routing
process begins.
2. Information can change after the initial routes have been constructed.
The dynamic vehicle routing problem calls for online algorithms that work in real-time since the
immediate requests should be served, if possible. As conventional static vehicle routing problems are
NP−hard, it is not always possible to find optimal solutions to problems of realistic sizes in a
reasonable amount of computation time. This implies that the dynamic vehicle routing problem also
belongs to the class of NP−hard problems, since a static VRP should be solved each time a new
immediate request is received.

Figure 2: A dynamic vehicle routing scenario with 8 advance and 2 immediate request customers.

In Figure 2 a simple example of a dynamic vehicle routing situation is shown. In the example, two un-
capacitated vehicles must service both advance and immediate request customers without time
windows. The advance request customers are represented by black nodes, while those that are
immediate requests are depicted by white nodes. The solid lines represent the two routes the
dispatcher has planned prior to the vehicles leaving the depot. The two thick arcs indicate the vehicle
positions at the time the dynamic requests are received. Ideally, the new customers should be inserted
into the already planned routes without the order of the non-visited customers being changed and with
minimal delay. This is the case depicted on the right hand side route. However, in practice, the
insertion of new customers will usually be a much more complicated task and will imply a re-
planning of the non-visited part of the route system. This is illustrated by the left hand side route
where servicing the new customer creates a large detour.

180
Generally, the more restricted and complex the routing problem is, the more complicated the insertion
of new dynamic customers will be. For instance, the insertion of new customers in a time window
constrained routing problem will usually be much more difficult than in a non-time constrained
problem. Note that in an online routing system customers may even be denied service, if it is not
possible to find a feasible spot to insert them. Often this policy of rejecting customers includes an
offer to serve the customers the following day of operation. However, in some systems – as for
instance the pick-up of long-distance courier mail - the service provider (distributor) will have to
forward the customer to a competitor when they are not able to serve them.
A dynamic problem can also be deterministic or stochastic. In deterministic and dynamic problems,
all data are known in advance and some elements of information depend on time. For instance, the
VRP with time windows belongs to this class of problems. Similarly, the traveling salesman problem
(TSP) with time-dependent travel times is deterministic and dynamic. In this problem, a traveling
salesperson has to find the shortest closed tour among several cities passing through all cities exactly
once, and travel times may vary throughout the day. Finally, in stochastic and dynamic problems (also
known as real-time routing and dispatching problems) uncertain data are represented by stochastic
processes. For instance, user requests can behave as a Poisson process. Since uncertain data are
gradually revealed during the operational interval, routes are not constructed beforehand. Instead, user
requests are dispatched to vehicles in an on-going fashion as new data arrive.
The events that lead to a plan modification can be: (i) the arrival of new user requests, (ii) the arrival
of a vehicle at a destination, (iii) the update of travel times. Every event must be processed according
to the policies set by the vehicle fleet operator. As a rule, when a new request is received, one must
decide whether it can be serviced on the same day, or whether it must be delayed or rejected. If the
request is accepted, it is temporarily assigned to a position in a vehicle route. The request is
effectively serviced as planned if no other event occurs in the meantime. Otherwise, it can be assigned
to a different position in the same vehicle route, or even dispatched to a different vehicle. It is worth
noting that at any time each driver just needs to know his next stop. Hence, when a vehicle reaches a
destination it has to be assigned a new destination. Because of the difficulty of estimating the current
position of a moving vehicle, reassignments could not easily made until quite recently. However, due
to advances in communication technologies, route diversions and reassignments are now a feasible
option and should take place if this results in a cost saving or in an improved service level. Finally, if
an improved estimation of vehicle travel times is available, it may be useful to modify the current
routes or even the decision of accepting a request or not. For example, if an unexpected traffic jam
occurs, some user services can be deferred. It is worth noting that when the demand rate is low, it is
useful to relocate idle vehicles in order to anticipate future demands or to escape a forecasted traffic
congestion.
Particular features
A dynamic VRPs possess a number of peculiar features, some of which have just been described. In
the following, the remaining characteristics are outlined.
(i) Quick response: Real-time routing and dispatching algorithms must provide a quick response so
that route modifications can be transmitted timely to the fleet. To this end, two approaches can be
used: simple policies (like the first-come first served (FCFS) policy), or more involved algorithms
running on parallel hardware (like the tabu search (TS) heuristics). As will be explained, the choice
between them depends mainly on the objective, the degree of dynamism and the demand rate.
(ii) Denied or deferred service: In some applications it is valid to deny service to some users, or to
forward them to a competitor, in order to avoid excessive delays or unacceptable costs. For instance,
the requests that cannot be serviced with a given time windows are rejected. When no time windows
are imposed, some user requests can be postponed indefinitely because of their unfavorable location.
This phenomenon can be avoided by imposing dummy time windows, or by adding a nonlinear delay
penalty to the objective function.

181
(iii) Congestion: If the demand rate exceeds a given threshold, the system becomes saturated, i.e., the
expected waiting time of a request grows to infinity.
The degree of dynamism of a problem
Designing a real-time routing algorithm depends to a large extent on how much the problem is
dynamic. Degree of dynamism of a problem quantifies this concept. Without loss of generality, we
assume that the planning horizon is a given interval [0, T ] , possibly divided into a finite number of
smaller intervals. Let ns and nd be the number of static and dynamic requests, respectively. Moreover,
let ti ∈ [0, T ] be the occurrence time of service request i. Static requests are such that ti = 0 while
nd
dynamic ones have ti ∈ [0, T ) Lund et al. [32] define the degree of dynamism d as δ =
ns + nd
which may vary between 0 and 1. Its meaning is straightforward. For instance, if δ is equal to 0.3,
then 3 customers out of 10 are dynamic. In his recent doctoral thesis, Larsen [10] generalizes the
definition proposed by Lund et al. in order to take into account both dynamic request occurrence times
and possible time windows. He observes that, for a given δ value, a problem is more dynamic if
immediate requests occur at the end of the operational interval [0, T ] . As a result he introduces a new
measure of dynamism:

∑
ns + nd
ti / T
δ '
= i =1

ns + nd

It is worth noting that δ ' ranges between 0 and 1. It is equal to 0 if all user requests are known in
advance while it is equal to 1 if all user requests occur at time T . Finally, Larsen extends the
definition of δ ' to take into account possible time windows on user service time. Let ai and bi be the
ready time and deadline of client i ti ≤ ai ≤ bi , respectively. Then,

∑
ns + nd
[T − (bi − ti )] / T
δ ''
= i =1

ns + nd

It can be shown that ti ≤ ai ≤ bi also varies between 0 and 1. Moreover, if no time windows are
imposed (i.e., ai = ti and bi = T ), then δ ' = δ '' . As a rule, vendor based distribution systems (such
as those distributing heating oil) are weakly dynamic. Problems faced by long-distance couriers and
appliance repair service companies are moderately dynamic. Finally, emergency services and taxi cab
services exhibit a strong dynamic behavior.

14.5 An Example of DVRP- The Traveling Repairman

Consider the situation that arises when for instance a bank teller machine breaks down and must be
repaired by a service technician. The route to be followed by the technician may be determined using
a distance based objective or it might take the urgency of the call (is the teller-machine located in a
high intensity area or is it located in a remote area?) into consideration. This problem is often referred
to as the Dynamic Traveling Repairman Problem" (DTRP) and is one of the most well-studied
dynamic vehicle routing problems. A similar example is the repairman from the electric power
company traveling from house to house to repair sudden break-downs in the electric power supply.
The Dynamic Traveling Repairman Problem (DTRP) was introduced by Bertsimas and Van Ryzin in
the paper entitled \A Stochastic and Dynamic Vehicle Routing Problem in the Euclidean Plane".
Bertsimas's and Van Ryzin's work on the DTRP is by far the most extensive and mathematically
concise work on dynamic vehicle routing problems. Bertsimas and Van Ryzin mention that in real

182
distribution systems orders/demands arrive randomly in time and the dispatching of vehicles is a
continuous process of collecting demands, forming tours and dispatching vehicles. In dynamic
settings the waiting time is often more important than the travel cost. Examples of applications where
the waiting time is the important factor include the replenishment of stocks in a manufacturing
context, the management of taxi cabs, the dispatch of emergency services, geographically dispersed
failures to be serviced by a mobile repairman.
Bertsimas and Van Ryzin define the DTRP as follows:
• A repairman (or a vehicle/server) travels at unit velocity in a bounded convex region A of area
A.
• All demands are dynamic and arrive in time according to a Poisson process with the intensity
parameter λ . The locations of the demands are independently and uniformly distributed in A.
• Each demand requires an independently and identically distributed amount of on-site service
time with mean duration s and second moment s 2 . The fraction of time that the server spends
on on-site servicing the demands is denoted by ρ . For stable systems ρ = λ s .

• The system time, Ti , of demand i is defined as the elapsed time between the arrival of demand i
and the time the server completes the service of the demand. The steady-state system time
denoted by T is defined by T = lim E[Ti ] .
i →∞

• The waiting time, Wi , of demand i is defined as the time elapsed from the demand arrived until
the service starts. Thus, Ti = Wi + si . The steady-state waiting time, W, is defined as
W =T −s .
The problem is to design a routing policy that minimizes T . The optimal value of is denoted T * .
Bertsimas and Van Ryzin stress that although the DTRP resembles a traditional queuing system,
queuing theory does not apply due to the fact that the system time T includes the travel times which
cannot not be regarded as independent variables. The approach of the authors is to derive lower
bounds for all policies for the average system time T . After that Bertsimas and Van Ryzin analyze
several policies and compare their performance to the lower bounds. To obtain these results the
authors use techniques from combinatorial optimization, queueing theory, geometrical probability and
simulation.
Bertsimas and Van Ryzin analyze a wide range of routing policies for the DTRP. A brief description
of these policies is given below.
• First Come First Served (FCFS): The demands are served in the order in which they are received
by the dispatcher.
• Stochastic Queue Median (FCFS-SQM): The FCFS-SQM policy is a modification of the FCFS
policy. According to the FCFS-SQM policy the server travels directly from the median of the
service region to the location of the demand. After the service has been completed, the server
returns to the median and waits for the next demand.
• Nearest Neighbor (NN): After completing service at one location the server travels to the nearest
neighboring demand.
• Traveling Salesman Problem (TSP): The demands are batched into sets of size n. Each time a
new set of demands has been collected, a Traveling Salesman Problem is solved. The demands are
served according to the optimal TSP tour. If more than one set exists at the same time, the sets are
served in an FCFS manner.
• Space Filling Curve (SFC): The demands are served, as they are encountered during repeated
clockwise sweeps of a circle C that covers the service region.

183
14.6 The integration of information technology and OR algorithms
Heuristics for the vehicle routing problem
Many authors have suggested a number of solution methods for the vehicle routing problem.
Heuristics have been developed and applied to many routing and scheduling case studies and several
researchers have compared and evaluated these methods. The advantage of the heuristics is their
ability to handle efficiently a large number of constraints and parameters of the routing problem. They
perform a relatively limited exploration of the search space and generally produce good quality
solutions within modest computing time. Therefore, classical heuristics are still widely used in
commercial software packages. Heuristics for the vehicle routing problem can be classified into two
main classes: classical heuristics, developed between 1960 and 1990 and metaheuristics whose growth
has occurred in the last decade . The most well known classical heuristics are the Savings and Sweep
algorithms. The most successful metaheuristic approach is the tabu search heuristics. In terms of the
solution procedure, classical heuristics can be classified in sequential and parallel. Sequential heuristic
algorithms solve the vehicle routing sub-problems (clustering and finding best tour) separately and
consecutively. Parallel heuristics produce routes and tours concurrently and can be classified in
construction and improvement methods. In metaheuristics, the emphasis is on performing a deep
exploration of the most promising regions of the solution space. These methods typically combine
sophisticated neighbourhood search rules, memory structures and recombination of solutions. The
quality of solutions produced by these methods is usually much higher than that obtained by classical
heuristics but the computing time is increased.
Advanced planning and scheduling systems: The OR practice using IT
Modern information technology supports the definition and analysis of the factors influencing the
routing process, a problem difficult to solve using empirical methods. Nowadays, there are software
applications that manage supply chain processes at all three decisional levels of management,
including features of transportation planning and execution:
• High level (strategic)––these applications offer transportation planning and fleet composition
modules;
• Middle level (tactical)––these applications allocate resources to the general transportation plan
derived from the previous level;
• Low level (operational)––at this level the applications are oriented to detailed scheduling of routes
and tours on a daily basis.
The mathematical technique most commonly applied to strategic transportation planning is linear
programming and related algorithms like network optimization and mixed integer programming.
Linear programming based approaches model the current transportation business including revenues
and costs. The models are extended to include decisions on the location of new facilities, acquisition
of transport resources and implementation of transport strategies. Transportation planning applications
at the tactical level include linear programming based models similar to the ones mentioned above.
Their functionality is found in most ERP systems. The operational level applications utilize mostly
heuristic algorithms. These algorithms can find a near-optimum solution for complex and multi-
parameter problems in short time. Today, the advanced computer programming languages and the
powerful hardware supports the use of these algorithms in advanced planning and scheduling (APS)
software systems. The ability to implement APS systems as a decision support tool directed in
transportation, has been greatly enhanced by improvements in telecommunications, better supply
chain management systems, more comprehensive ERP implementations, and more powerful
forecasting tools. The power of OR models and optimization methods can be enhanced incorporating
them within a decision support system, which takes advantage of modern information technology.

184
14.7 A Generic Architecture for Dynamic Real-Time VRS
Dynamic Real-Time Vehicle Management Model
Most solution approaches to the VRP are in practice implemented in a centralized computer resource
(normally at headquarters), producing a daily plan to be provided to the vehicles before the beginning
of the distribution execution. Some of these approaches have been implemented in commercial
systems that are successfully used by numerous transportation, logistics, and manufacturing
companies over the last twenty years. These systems have not, however, been designed to address the
case in which the execution of delivery cannot follow the prescribed plan due to some unforeseen
event. When there is a need for real-time intervention, it may be necessary to re-compute the plan
using new input data. If a typical VRP approach is used for re-planning (i.e. re-planning the whole
schedule from scratch), many vehicle schedules may be affected, thus causing significant performance
inefficiencies (high overhead, nervousness, errors and high costs).
Thus, re-planning based on classical VRP solution methods may not be a realistic option. In the
absence of algorithms capable of ‘isolating’ the part of the VRP affected by the unexpected event in
order to minimize the disturbance to the overall schedule, interventions are typically performed
manually (for example, through voice communication between drivers and the logistics manager), and
the quality of decisions taken is naturally affected. The need to enhance existing methods, or develop
novel approaches, becomes clearer in view of recent advancements in mobile and positioning
technologies. Using such technologies, information about unforeseen events may be transmitted when
they occur directly from the affected truck(s) through a mobile network to headquarters and/or other
parts of the fleet. Given an efficient re-planning algorithm, appropriate and implementable plan
modifications may be transmitted back to the fleet in a timely fashion to respond effectively to the
new system state.
A real-time vehicle management model is schematically depicted in Figure 3, using control system
formalism. The model includes the following:
a) Control & Detection of the system’s state: This concerns the selection of the parameters to be
monitored, such as truck position, truck speed, truck inventory, and so on. These parameters need to
be regularly monitored, as they will trigger intervention if needed. It is noted that interventions may
lead to system “nervousness”; thus, the cost of intervention should be balanced against expected
benefits.
b) Projection: This concerns the revised distribution plan that will be generated by the system, when it
is observed that a vehicle is out of schedule. Typical parameters that cause significant disturbances to
the original routing schedule are traffic jams, road works and negative environmental conditions.
c) Decision-making and execution: The selection of the problem objectives has significant effects on
the decision-making and execution mechanisms employed. Objectives to be considered may include:
minimize the deviation from the original plan, minimize the cost of non-conformance, minimize risk,
and others.

Fig. 3: Systemic representation of a real-time vehicle management system

185
Important decision-making issues include modeling of the real-time rerouting problem, and
development of appropriate solution methods. In this case, problem complexity and computational
time play a significant role. The reduction of complexity appears to be a necessary condition in
providing timely, implementable solutions. A classic way to reduce complexity is by using a
hierarchical approach, whereby, a complex monolithic problem is decomposed or disaggregated to
multiple, simpler problems that can be solved independently. The solutions of these lower-level
problems are combined to yield the solution of the global, higher-level, problem. By doing so, one
needs to consider the trade-off between optimality and computational efficiency.
System Implementation Issues
The model presented in Figure 3 can be realized through the use of mobile technologies, real-time
decision-making algorithms (along the lines presented in the previous section) and back-office
automated processing. In addition to providing the appropriate directions to the drivers of the fleet, the
customer base may be kept informed in regard to changes in the initial schedule, therefore improving
the company’s service quality and customer relations.
The proposed system architecture is shown in Figure 4. It comprises three major sub-systems. The
back-end system consists of a decision making module to facilitate automated decision making and
ERP connectivity. The Wireless Communication sub-system allows a two-way communication
between the back-end and front-end systems. The Front-end system enables a) a robust user interface,
and b) interaction between the software platform that is installed in the on-board truck computer and
the company’s back-end system.
Back-end sub-system
The back-end system is a decision support system that incorporates algorithms needed for real-time
routing, scheduling and monitoring of the current state of the fleet, as well as a robust database
containing both static (customers, geographical information of the road network, and so on) and
dynamic (orders, quantities, time window information, and so on) data. The back-end system also
provides ERP connectivity, which is especially useful in ex-van sales to provide information, such as
customer sales history, customer credit, and other decision-critical data.
Wireless Communication sub-system
The wireless communication sub-system consists of two parts: a) The mobile access terrestrial
network, which is responsible for the wireless interconnection of the back-end system with the front-
end on-board devices, and b) the positioning system, which is responsible for vehicle tracking.
The Mobile Access Terrestrial Network can be based on any of a number of existing or emerging
mobile technologies. In examining the options available to support an integrated distribution system,
bandwidth is perhaps the most important issue. The bandwidth requirements depend on the
computational model chosen. If vehicle onboard devices support much of the computations, then the
demand for bandwidth is different than in the case in which much of the computation is performed at
the headquarters. In either case, however, the demand for bandwidth is greater when compared to
existing applications, such as fleet tracking, graphical representation of real-time information in
digital maps, and voice communication.
GPRS, TETRA, and UMTS can provide always-on, packet-switched connectivity and high-speed data
rates. GSM is a mature technology. However, it cannot support high-data transmission effectively.
GPRS combines high data rates, always-on connectivity, mature technology, and has also been used
in fleet management systems. As far as TETRA is concerned, it is worth mentioning that it provides
much better security than GPRS, as well as supporting point-to-multipoint voice broadcasting. UMTS
is an emerging standard, and its use cannot be assessed prior to thorough validation testing.
As far as the Positioning System is concerned, positional accuracy of less than 100m is deemed
acceptable for urban distribution (accuracy requirements can, of course, be relaxed in non-urban
settings). An analysis of the technologies that can be used for location identification goes beyond the
scope of this paper.

186
 Fig. 4: Generic architecture for dynamic real-time vehicle management

Front-end system (Access media)

The front–end system generally consists of a mobile device, to which all necessary information is sent
from the headquarters. The selection of the front-end device is important both from a user interface
and from a computational performance perspective. Typical mobile devices that can be used on-board
include mobile phones, personal digital assistants (PDAs), and Tablet PCs. In their present state,
mobile phones do not appear capable of coping with the requirements of the applications under
consideration. By contrast, PDAs and Tablet PCs are already used for specific distribution
applications, such as back-end ERP connectivity. These devices include high-resolution displays,
wireless networking capabilities, and integrated support for peripherals.
14.8 Conclusions
Real-time vehicle management is important in supporting supply chain execution systems, and
minimizing the related logistics risks. It has been demonstrated that a good, near-optimal, distribution
plan is necessary but not sufficient for high performance distribution. This needs to be complemented
by the ability to make and implement sophisticated decisions in real –time, in order to respond
effectively to unforeseen events. The emergence of technologies and information systems allowing for
seamless mobile and wireless connectivity between delivery vehicles and distribution facilities is
paving the way for innovative approaches in addressing this requirement.
In order to develop robust, practical approaches to the real-time vehicle management problem,
research efforts should focus on three fronts: systems design, decision support methods, and system
implementation. In the first area, significant issues to be tackled include the definition of the system’s
objectives (minimize cost, risk and/or deviation from the original plan), the observability of the

187
system’s state, balance of intervention costs vs. expected benefits, the extent of interventions (local
vs. global), and other parameters. System designs cannot be generalized beyond the extent achieved in
this chapter due to their heavy dependence on the characteristics of the problem addressed, and the
algorithmic approach chosen for intervention. Therefore, future research can assess alternative design
specifications against real-life case studies of real-time vehicle routing problems.
In the second area, a review of the vast existing literature on the vehicle routing problem has indicated
that some research is relevant and can be used as a basis for the development of appropriate
enhancements and/or novel decision support approaches in real-time vehicle re-planning. In this case,
problem complexity and computational time play a significant role in system effectiveness.
In the implementation area, it appears that there exist mature technologies to sufficiently address the
requirements of the real-time vehicle management system. In terms of the communication subsystem,
GPRS and TETRA are appropriate mobile access networks, while GPS technologies meet all the
related positioning requirements. For the front-end system, PDAs and Tablet PCs have significant
potential, since both their interface capability and computational power support efficient user
interaction and the local computational system requirements, respectively. All three fronts discussed
above present interesting challenges with significant implications for both VRP-related research and
the technology that will support effective logistics execution.

References

[1] Dimitris Bertsimas and Garrett Van Ryzin. A Stochastic and Dynamic Vehicle Routing Problem
in the Euclidean Plane. Operations Research, 39:601-615, 1991.
[2] Allan Larsen, The Dynamic Vehicle Routing Problem, PhD Thesis, Technical University of
Denmark, 1999.
[3] Vasileios Zeimpekis and George M. Giaglis, A Dynamic Real-Time Vehicle Routing System for
Distribution Operations in Georgios J. Doukidis and Adam P. Vrechopoulos eds., Consumer
Driven Electronic Transformation- applying New Technologies to Enthuse Consumers and
Transform the Supply Chain part-1, Springer Berlin Heidelberg, 2005, pp 23-37
[4] Gianpaolo Ghiani, Francesca Guerriero, Gilbert Laporte and Roberto Musmanno, Real-time
vehicle routing: Solution concepts, algorithms and parallel computing strategies, European
Journal of Operational Research, Volume 151, Issue 1, 16 November 2003, Pages 1-11
[5] Sotiris P. Gayialis and Ilias P. Tatsiopoulos, Design of an IT-driven decision support system for
vehicle routing and scheduling, European Journal of Operational Research, Volume 152, Issue
2, 16 January 2004, Pages 382-398
[6] D. Tarantilis, D. Diakoulaki and C. T. Kiranoudis Combination of geographical information
system and efficient routing algorithms for real life distribution operations European Journal of
Operational Research, Volume 152, Issue 2, 16 January 2004, Pages 437-453
[7] Vehicle Routing Problem (VRP), by Wolfgang Garn
http://osiris.tuwien.ac.at/~wgarn/VehicleRouting/vehicle_routing.html

188
 For Limited Circulation

Chapter 15
Optimization of the Supply Chain Network: Simulation, Taguchi, and
Psychoclonal Algorithm Embedded approach*

In today’s market, increased level of competitiveness and uneven fall of the final demands are pushing
the enterprises to make an effort for optimization of their process management. It involves
collaboration in multiple dimensions like information sharing, capacity planning, and reliability
among players. One of the most important dimensions of the supply chain network is to determine its
optimal operating conditions incurring minimum total costs. However, this is even a tough job due to
the complexities involved in the dynamic interaction among multiple facilities and locations. In order
to resolve these complexities and to identify the optimal operating condition we have proposed a
hybrid approach incorporating simulation, Taguchi method, robust multiple nonlinear regression
analysis and the Psychoclonal algorithm. The Psychoclonal algorithm is an evolutionary algorithm
that inherits its traits from Maslow need hierarchy theory and the artificial immune system. The
results obtained using the proposed hybrid approach is compared with those found out by replacing
Psychoclonal algorithm with the AIS and Response Surface Methodology (RSM) respectively. This
research makes it possible for the firms to understand the intricacies of the dynamics and
interdependency among the various factors involved in the supply chain. It provides guidelines to the
manufacturers for the selection of appropriate plant capacity and also proposes a justified strategy
for delayed differentiation.

Key word: supply chain, simulation, Taguchi orthogonal array, Regression analysis, Psychoclonal
algorithm

1 Introduction
In today’s competitive and global market, the success of an industry is reliant upon the management
of its supply chains. Supply chain is the interlinked network of suppliers, manufacturers, distributors
and customers interconnected by transportation, information sharing, and financial infrastructure
(Chopra and Meindle, 2001). The supply chain aims to provide quality products and services to the
end consumer in most efficient and economical mode (Sahin and Robinson 2002).
Plethora of researches addressing various issues related to different aspects of the supply chains is
available in the literature. Gavirneni et al. (1999) have focused on management of material flow.
While, Strader et al. (1999), and Hewitt (1999) emphasized on the role of information technology in
the supply chain network. Shunk et al. (2006) applied integrated enterprise modeling methodology –
FIDO – to supply chain integration 1 modeling. Spekman (1988) and Tompkins (1998) studied the
interdependencies among various supply chain members like retailers, manufacturers and suppliers.
Kwak et al. (2006) proposed supplier buyer model for the bargaining process, while an agent based
approach for e-manufacturing and supply-chain was given by Zhang et al. (2006). Moreover,
Altiparmak et al. (2006) considered supply chain network as a multiobjective optimization problem
and solved it with the aid of genetic algorithm. In the past, researchers have adopted different
methodologies to cope with the supply chain problems some of them have focused on the
deterministic while others on non deterministic approaches. Cohen and Lee (1988), Arntzen et al.

1
Contributed by Dr. M. K. Tiwari, Department of Industrial Engineering and management, Indian Institute of
Technology Kharagpur, 721302, India. This work is a part of research article which has been accepted for
publication in Computers and Industrial Engineering. The other co-authors of this work are Sanjay Shukla and
Dr. Ravi Shankar.

189
 For Limited Circulation

(1995), and Hariharan et al. (1995) have deployed mathematical programming, a deterministic
approach, to optimize the supply chain network. While Zheng and Zipkin (1990), and Van Houtum et
al. (1996) have concentrated on stochastic process modeling to deal with the supply chain issues.
In order to attain the competitive edge a supply chain should be flexible, quick enough, dependable,
and cost-efficient. These objectives are accomplished by high speed information and material flow
with low overhead costs. Coordination and collaboration in different activities like sale information,
inventory information, promotions, and shipment are essential to the success of the supply chain .One
of the important dimension in the supply chain network is to the minimize its costs by identifying the
optimal operating conditions. However, this is even a tough job due to the complexities involved in
the dynamic interaction among multiple facilities and locations. To resolve these complexities and to
maximize the overall supply chain performance members of the chain have to team up and define
clearly their responsibilities along with the co-operation mechanism.
Many researchers have addressed the problem of a single operating condition such as at what capacity
level will a supply chain achieves its best performance. But, an analytical approach addressing the
issue of depicting an optimal supply chain conditions by considering two or more factors, at a time, is
lacking.
In this research the proposed analysis will clarify the responsibility of each member in implementing
the optimal supply chain strategy. In other words, this research attempts to minimize total supply
chain costs and identify an optimal condition by considering various supply chain parameters
simultaneously viz. delayed differentiation, information sharing, capacity, reorder policy, lead time,
and supplier’s reliability. In order to meet this objective, we are proposing a hybrid approach
encapsulating simulation, Taguchi approach (1986), non-linear regression analysis (Montgomery
2001), and the Psychoclonal algorithm (Tiwari et al. 2004). First, Simulation is used to model a
comprehensive supply chain network, which deals with a range of operational components and
management level. Although useful, simulation only evaluates the effectiveness of pre-specified
conditions and do not provide any means for optimizing the system performance. Due to the
aforementioned shortcomings authors have intended to couple the simulation model with The Taguchi
orthogonal array, robust non-linear regression analysis, and the Psychoclonal algorithm.
In order to respond effectively to the unanticipated events such as changes in demand, order quantity,
or delivery time and to maintain system stability one needs to develop a robust supply chain. Taguchi
approach is the best way to attain such robustness as it is capable of increasing system robustness,
reducing experimental costs, and improving the overall quality by considering factors at a discrete
level. We have explored Taguchi’s orthogonal array to identify the best parameter settings for the
robust supply chain. Montgomery (2001) pointed out that the Taguchi’s orthogonal arrays are not
efficient in exploring the search space completely when process parameters vary on continuous scale.
Therefore, this method is only suitable for optimizing qualitative variables but not for the quantitative.
This limitation has motivated the authors to incorporate regression analysis and the Psychoclonal
algorithm with simulation and Taguchi approach. Regression analysis has been employed to obtain
functional relationship between quantitative process variables and response of the supply chain
network (supply chain costs). Further, Psychoclonal algorithm is utilized as an optimization tool for
depicting optimum setting of quantitative process variables. The Psychoclonal algorithm is an
evolutionary algorithm. It inherits its properties from Maslow’s need hierarchy theory and artificial
immune system. The results achieved by using the proposed hybrid approach are compared with those
obtained by replacing Psychoclonal algorithm with AIS and Response Surface Methodology (RSM)
respectively. These comparisons revealed that Psychoclonal algorithm dominates both the AIS and the
RSM in terms of minimizing the supply chain costs.
The remainder of paper is organized as follows. Section 2 deals with the simulation model of the
supply chain network and parameters affecting its performance. In section 3, the way of conducting
the experiments (using Taguchi method) are explored to obtain costs of supply chain network for
different parameters settings. Then in section 4 mathematical modeling is carried out to obtain
functional relationship between network’s process parameters and supply chain coats. To solve the

190
 For Limited Circulation

mathematical model developed by regression analysis a solution methodology is presented in section

5. Results and discussions are provided in section 6 while section 7 concludes the paper.
2. Simulation Modeling of a Supply Chain Network
Simulation is a convenient source for predicting different interdependencies among constituents of a
supply chain network. It is used to model a comprehensive supply chain network, which deals with
the range of operational components and management level. In past, researchers like Bhaskaran
(1998), Beamon and Chen (2001) have used simulation model to study different dimensions of a
supply chain like performance effects of operational factors, and demand amplification effects. Here,
we are using simulation to model a comprehensive supply chain network with the objective of
identifying optimal factor setting having minimum costs.
We have considered three echelon supply chain that consist of four retailers one manufacturer, and
three suppliers as shown in figure 1.

Retailer 1
Supplier 1

Retailer 2
Supplier 2 Manufacturer

Retailer 3

Supplier 3 Retailer 4

Figure1 A supply chain network

From the figure, it is clear that each player of the supply chain is interdependent and hence output at
one stage acts as input for next stage. We have chosen ARENA (Rockwell software, 2003) as our
working simulation environment. Arena is general purpose simulation software that has full
visualization of model structure and parameters, run control, and animation facilities. The supply
chain under consideration manufactures and markets televisions having short life cycle. The retailers
rely on Bass model (refer to appendix A) for forecasting the demand of the product. They follow the
following strategy to meet their objective:
1. After estimation of the future demand the inventory decision is made depending to its current level.
2. According to inventory level order is placed to the manufacturer
3. Inbound inventories are received from the manufacturer then inventory cost and services are
assessed.
For simplicity we assume that studied problem has two product families based on the size of
television frame. The demand for product is assumed to resemble an S- shaped curve with an average
of 1800 units per month. Since televisions are short life cycle products hence, length of the simulation
run is considered to be small (say, 18 months). The supply chain network variables considered in
present simulation are listed in table 1.
However, each variable may have wide range of possible levels but we are focusing only on three for
each to represent most likely variability corresponding to the first stage of simulation. A brief
description of the factors influencing the supply chain performance (supply chain costs) is described
below:

191
 For Limited Circulation

Table 1. Levels of controllable and uncontrollable factors in simulation

Controllable factors Level

A Delayed differentiation No Partial Complete
B Information sharing No Partial Complete
C Capacity (00s) 50 100 150
D Retailer’s (S, s) Policy, in (00s) 25 50 75
E Replenishment lead time (in 6 10 14
days)
F Supplier reliability 50% 70% 90%

1. Delayed differentiation (postponement): Alderson (1950) was the first who coined the term
Delayed differentiation. This is also termed as postponement. This is the most general method that can
be applied for enhancing the efficiency of a marketing system. It exploits the component commonality
and redesigns the production process so that expensive operation can be delayed. This is an effective
means of minimizing the increased costs, with maintaining the good customer services, arisen due to
product multiplicity (Swaminathan and Tayur 1998).
2. Information sharing ( i ) : Information sharing plays an important role in supply chain network.
Each player makes an ordering decision based on information received from the downstream player.
Lack of information sharing among player results in high amplification of demand (sterman 1989).
This effect is highly undesirable as it exacerbate the supply chain costs.
3. Capacity ( c ) : Capacity indicates the ability to satisfy future demand or it may be defined as the
amount of the products that can be produced per time period. Sharing the capacity information by
each entity in a supply chain is necessary for integrated planning (Gaonkar and Viswanathan 2001).
The manufacturer makes a better production decision, in terms of costs saving, when he knows the
capacity of each supplier. A higher capacity level leads to greater benefit of information sharing
(Gavirneni 1997).
4. Reorder policy ( R ) : Each retailer fallows a (S , s ) policy. When inventory level becomes equal to or
falls below the reorder point, s , then retailer puts an order up to level S. This is also known as min-
max policy.
5. Lead-time (l ) : Lead-time is one of the most important factors influencing the supply chain
performance. It may be defined as the time interval between placing and receiving an order. Longer
lead-time results in smaller benefit of information sharing (Towill et al., 1992). Large lead time
imparts major contribution in enhancing bull Whip effect (Lee et al. 1994).
6. Reliability of the suppliers (r ) : Reliability of the supplier affects the performance of the supply
chain in terms of availability of materials and quality components at the time of production.
Unreliable supply decreases the quality of finished product; and increases demand variability.
The simulation procedure consists of three main phases that are discussed in next three subsections.
2.1 Generation of Demand and Inventory Policy
Each retailer estimates the demand by combining Bass model and demand variance. Forecasted
demand is assumed to be much larger than the error arising in forecasting thus, possibility of
generating negative demand is negligible (Zhao and Xie 2002). Once the demands for all retailers are
generated for 18 month life-cycle, the capacity is designed to match the demand. Let the
manufacturer’s monthly capacity is represented by C. Retailers follow the (S , s ) policy to control the
inventory level. Under min-max policy, every weak, if inventory drops below s then the retailers put

192
 For Limited Circulation

an order to sustain a stock at an explicit level S . This explicit level is determined by calculating the
quantity needed between the time the order is placed and is received along with a quantity of safety
stock to allow variation on demand. Mathematically S can be given as:

S = μ ( T + LT ) + Θ σ T + LT …(1)
Where,
μ is the periodic expected demand which can be estimated by employing Bass model.

T is the time interval between orders which is one week for the present model.
LT is the lead time that includes production time needed when no information sharing and no product
differentiation are considered.
Θ is the standard deviation associated with desired service level, for present simulation model service
level is set to be 97.5%.
σ is the standard deviation of demand per time period.
Due to the S -shaped demand pattern (S , s ) is adjusted according to the bass model. Using the
difference of (S − s ) specified in factor D of table 1, we derive the reorder point, s , for each order
cycle. The parameters of Bass model are derived by previous sales of similar product as shown in
table 2.
Table 2. Estimation of ω , υ and K in Bass model

Month television Cumulative Predicted Squared error

T demand demand D (t) in F (t) dt - Ft
dt in (00s) (00s)
0 0 0 0
1 69 69 63.59 29.26
2 71 140 73.65 7.04
3 82 222 82.97 0.94
4 87 309 92.41 29.31
5 96 405 100.90 23.98
6 110 515 108.42 2.50
7 116 631 114.67 1.78
8 132 763 118.51 181.95
9 115 878 119.46 19.90
10 123 1001 117.32 32.31
11 104 1105 111.96 63.35
12 100 1205 104.96 24.59
13 97 1302 96.09 0.82
14 72 1394 85.49 42.33
15 74 1468 73.62 0.14
16 57 1525 62.79 33.49
17 49 1574 53.66 21.71
18 45 1619 45.27 0.07
19 42 1661 37.12 23.47
20 31 1692 29.14 3.47.
Total 1692 1692.00 542.73
Estimation ω υ K
0.035 0.188 1800.32

193
 For Limited Circulation

These parameters ( ω ,υ , K ) are determined by minimizing the sums of the squared error using excel
solver. Note that for a new product, without history, demand forecast is often estimated based on a
comparable older generation product.
2.2 The Production and Delivery Decision
The manufacturer applies lot sizing method (Chung and Lin 1988) in planning of its production
activities. In order to produce the right demand and to lower its WIP the manufacturer must reduce the
bullwhip effect (Lee et al. 1994). Clearly, the retailers need to share more information, and the
manufacturer needs to make production-planning decisions based on additional information. Under no
information sharing the suppliers only know the orders placed by manufacturer and manufacturer only
receives the order information from retailers. The manufacturer plans his production based on
forecasts obtained from the Bass model. Under partial information sharing the manufacturer plans his
production polices based on forecasts provided by Bass-model but adjusts the production based on the
retailers’ projected future net requirements, which is derived from retailer’s current inventory and
future forecast. When complete information is available retailers provide real-time updated inventory
status. Whenever a demand occurs and manufacturer adjusts the production quantity accordingly.
Therefore it is evident that Bass model is only used for purpose of planning, and that the manufacturer
basically implements a pull production system.
The manufacturer makes shipping decisions from on-hand inventory after finishing the current
period’s production. He will fill each retailer’s order plus back order. If on-hand inventory is
insufficient each retailer will receive a share proportional to its order plus backorder. Shipment
reaches the retailer after the lead time.
2.3 Simulation Run and Performance Measure
The manufactured televisions have short life cycle so we have to study the system behavior over the
product’s entire life cycle. The simulation required to study this sort of system is termed as
terminating simulation. This type of simulation starts at a prescribed initial state and terminates when
system reaches a prescribed terminal state or time. In our case it starts at t = 0, and terminates at t=18
months. In addition to this we have made 1200 replications of the system to attain a good point
estimator.
The objective of the supply chain, considered in this research, is to minimize its costs by identifying
the optimal level of supply chain parameters. The total costs of supply chain consist of transportation
cost, carrying cost, ordering cost, setup cost, and backorder cost. The relation between the explicit
operating situations and supply chain costs are shown in figure 2.1 – 2.6.
3. Parameter Design –Taguchi Method
The aim of the parameter design is to select the optimal level of controllable system parameters. This
leads the high level of performance under diverse range of conditions and makes the system robust
against the noises that cause inconsistency. Studying the design parameters one at a time or by trail is
a widespread approach to optimize the required response. However, this needs either to a very long
time for completing the design or to sudden termination of the design process due to cost of
conducting large number of experiments. By varying the design parameters one at a time; the study of
6-design parameter at 3 levels would necessitate 36 possible experimental evaluations. But the time
and cost required in carrying out such a comprehensive study is not desirable. The Taguchi approach
(1986) provides the convenient and effective means to determine optimal or near optimal design
parameters by using orthogonal array and linear graphs. We choose the L27 (313) design for
controllable factors. By using L27 array, 13 parameters can be studied at 3 levels by running 27
experiments instead of (36). To map our model completely, a linear graph is presented in figure 3.

194
 For Limited Circulation

figure 2.1Total Cost Under Different Level of Figure 2.2 Total Cost Under Different Level of
Delayed Differentiation Information Sharing

155 155

Total cost
Total Cost

145 145
135 Delayed 135 Information
125 Differentiation 125 Sharing
115 115
105 105
No Partial Complete No Partial Complete

Delayed Differentiation Information Sharing

Figure 2.3 Total cost under different level of

capacity Figure 2.4 Total Cost Under Different Level of
Retailer's S-s Values
155
Total cost

145

Total Cost
155
135 Capacity
145
Retailer's S-s
135
125
115
125
115
value
105 105
25 50 75
50 100 150
retailer's S-s Value
Capacity

Figure 2.6 Total cost under Different level of

Supplier Reliability
Figure 2.5 Total Cost Under Different Level of Lead
Time (days)
155
145
155
Total Cost
Total Cost

145 135
135
Supplier
Lead Time (days) 125 Reliability
125
115
115
105
6 10 14 105
0.5 0.7 0.9
Lead Time (in days)
Supplier's Reliability

Figure 2. Total cost verses controllable variables

195
 For Limited Circulation

A 1

A*B
3, 4 A*C
6, 7

D E F

B*C C 5 9 10 11 12 13
8, 11

Figure 3 Linear graph for L27 (313) orthogonal array

In the linear graph, for L27 (313) each circle represents a column of orthogonal array. Column 1, 2 and
5 represents factor A, Band C respectively. Column 3 or 4, 6 or 7 and 8 or 11 can represent
interactions between A and B, A and C , and B and C respectively column 9, 10 and 12 can be allotted
randomly for remaining factors D, E, and F.
3.1 Signal-to-Noise (S/N) ratio
Signal-to-Noise ratio measures the robustness of the response obtained from the Taguchi model.
Signal measures the typical value of the response whereas; noise is a measure of variability and
represents the undesired components. S/N ratios are defined in such a manner that larger is numerical
value, better the performance (Montgomery 2001). In this research the response has the smaller the
better characteristics. Hence S/N ratio for the costs of supply chain can be given as:
S/N ratio = -10log10 So … (2)

⎛1 n ⎞
Where S0 = ⎜ ∑ yi2 ⎟
⎜n ⎟
⎝ i =1 ⎠

yi = i th performanc e charecteri stic

n = number of replicates

We can draw plots between S/N ratio, obtained from Taguchi orthogonal array, and factors level as
shown in figure 4.

figure 4.1 S/N Ratio for Delayed Figure 4.2 S/N Ratio for Information
Differentiation sharing
35 29
28.5
28 S/N ratio
S/N ratio for
S/N
S/N

27.5
30 27 for factor i
d 26.5
26
25 25.5
No Partial Complete No Partial Complete

Delayed Differentiation Information sharing

196
 For Limited Circulation

Figure 4.3 S/N ratio for manufacturing capacity

28

Figure 4.4 S/N ratio for Retailers (s,S) policy

32
31
S/N

26 S/N ratio for

30
capicity 29 S/N ratio for

S/N
28 factor R
27
26
24 25
50 100 150 25 50 75
Capacity
Retailer's S-s Value

Figure 4.5 S/N ratio for Lead T ime

29
Figure 4.6 S/N ratio for Supplier Reliability
28.5 28
28 27
27.5 S/N ratio
S/N

27
S/N 26 S/N ratio for
26.5
for factor l 25 factor r
26
24
25.5
25 23
6 10 14 0.5 0.7 0.9

Lead T ime Supplier Reliability

Figure 4 Plots of S/N ratio from Taguchi method results

From the plots, we find the following optimal level for each factor.
Factor A (Delayed differentiation) = Complete
Factor B (Information sharing) =Complete
Factor C (Monthly capacity) =100
Factor D (Reorder quantity) = 25
Factor E (Lead time) = 6 days
Factor F (Supplier reliability) =90%
In optimization problems use of the Taguchi orthogonal array is limited as it assumes factors at
discreet level. Hence optimal level of factors, varying on continuous scale can not determined by it.
Therefore, in our case only two qualitative factors (delayed differentiation, and information sharing)
can be optimized by using Taguchi approach.
4. Mathematical modeling of a supply chain network using regression analysis
One of our goals is to design a network that is robust when exposed to disturbances. To attain this we
have used the factors level identified by the Taguchi method as the initial point therefore, we start
with:
Capacity=100
Reorder quantity=25

197
 For Limited Circulation

Lead time=6
Suppliers reliability=90%
The Experiments are carried out with suitable step sizes of parameters as listed in table 3.

Table 3. total costs of supply chain corresponding to different settings of factors

S.No. Factors Supply chain
Capacity (x1) Reordered Lead-time Suppliers costs
quantity (x2) reliability
100* 25& 6$ 90#
1. 98.3 25.85 6.43 89.4 203
2. 96.6 26.7 6.86 88.8 176
3. 94.9 27.55 7.29 88.2 162
4. 93.2 28.4 7.72 87.6 148
5. 91.5 29.25 8.15 87 143
6. 89.8 30.1 8.58 86.4 138
7. 88.8 30.95 9.01 85.8 127
8. 86.4 31.8 9.44 85.2 125
9. 84.7 32.65 9.87 84.6 129
10. 83 33.5 10.3 84 130
11. 81.3 34.35 10.73 83.4 134
12. 79.6 35.2 11.16 82.8 136
13. 77.9 36.05 11.59 82.2 143
14. 76.2 36.9 12.02 81.6 150
15. 74.5 37.75 12.45 81 157
16. 72.8 38.6 12.88 80.4 163
17. 71.1 39.45 13.31 79.8 167
18. 69.4 40.3 13.74 79.2 175
19. 67.1 41.15 14.17 78.6 198
20. 66 42 14.6 78 205
* Initial point of capacity
& Initial point of reordered quantity
$ Initial point of lead time
# Initial point of supplier reliability
These step sizes have been selected based on the knowledge of the process or other practical concerns.
For the instance, if the original monthly capacity is 100 units, to understand the supply chain
performance when capacity is changed within range of (65,100) units, a step size of 1.7 units would
be sensible. In terms of lead-time if range of experiment is (6, 15) days, step size of 0.43 days time
would be reasonable. We choose step size for capacity, reorder quantity, lead time, and supplier
reliability 1.7, 0.86, 0.43, and 0.6 respectively.
For simplicity, independent variables are coded in an interval of (-1, 1) by using the fallowing coding
schema:

Actual value ofi th factor − χ i

xi =
γi
(Upper bound )i + (Lower bound )i …(3)
Where, χi =
2
(Upper bound )i − (Lower bound )i
γi =
2

198
 For Limited Circulation

The upper bound and lower bound of parameters are as follows:

Capacity = [65,100]
Reordered quantity = [25, 43]
Lead time = [6, 15]
Supplier reliability = [77, 99]
By using equation 3, network variables can be coded in following manner:

⎛ Capacity − 98.5 ⎞
x1 = ⎜ ⎟
⎝ 8 ⎠
⎛ Reordered quantity - 32.6 ⎞
x2 = ⎜ ⎟
⎝ 6 ⎠ … (4)
⎛ Lead time - 5.3 ⎞
x3 = ⎜ ⎟
⎝ 5 ⎠
⎛ Supplier reliability - 88 ⎞
x4 = ⎜ ⎟
⎝ 10 ⎠
Variables in coded form are listed in table 4

Table 4. coded levels of factor

S.NO. Factors
X1 X2 X3 X4
1. 0.90 -0.45 -0.90 0.91
2. 0.80 -0.41 -0.81 0.75
3. 0.68 -0.36 -0.71 0.72
4. 0.61 -0.31 -0.61 0.63
5. 0.51 -0.26 -0.52 0.54
6. 0.42 -0.21 -0.42 0.44
7. 0.36 -0.17 -0.33 0.35
8. 0.22 -0.12 -0.23 0.26
9. 0.12 -0.07 -0.14 0.17
10. 0.02 -0.027 -0.04 0.08
11. -0.07 0.02 0.05 -0.01
12. -0.16 0.06 0.15 -0.11
13. -0.26 0.11 0.24 -0.20
14. -0.36 0.16 0.33 -0.29
15. -0.46 0.21 0.43 -0.38
16. -0.55 0.25 0.53 -0.48
17. -0.65 0.30 0.62 -0.57
18. -0.75 0.35 0.72 -0.66
19. -0.88 0.39 0.81 -0.75
20. -0.94 0.44 0.91 -0.85

Regression analysis provides the functional relationship between independent variables and response.
Regression equation is given as (Montgomery 2001):

199
 For Limited Circulation

k k
ŷ = βˆ0 + ∑ βˆi xi + ∑ βˆii xi2 + ∑∑ βˆij xi x j
i =1 i =1 i< j … (5)

Where xi is the i th parameter, β i is the regression coefficient and k is the number of process parameters.
Least square estimate method is used to interpret estimated regression coefficient which is given as:
⎛ βˆ ⎞
⎜ 0⎟
⎜ βˆ ⎟
⎜ i⎟
⎜ : ⎟
⎜ : ⎟ −1 … (6)
βˆ =⎜ ˆ ⎟ =⎛⎜ X' * X⎞⎟ * y
β
⎜ k⎟ ⎝ ⎠
⎜ : ⎟
⎜ ⎟
⎜ : ⎟
⎜ˆ ⎟
⎜β ⎟
⎝ kk⎠

⎛1 x 11 x 21 .. .. xk 1 ⎞
⎜ ⎟
⎜1 x 21 x 22 .. .. xk 2 ⎟
⎜: : : : ⎟
Where, X = ⎜ ⎟
⎜: : : : ⎟ … (7)
⎜ ⎟
⎜: : : : ⎟
⎜1 xn1 xn2 .. .. x nn ⎟
⎝ ⎠

y is the response (costs of supply chain), which can be expressed as:

⎛ y1 ⎞
⎜ ⎟
⎜ y2 ⎟ … (8)
⎜ ⎟
y=⎜ : ⎟
⎜ : ⎟
⎜⎜ ⎟⎟
y
⎝ n⎠

By using equations 5, 6, 7, and 8, we get following regression equation (objective function).

Y = 120 .3 + 0.717 x1 − 1.0037 x 2 + 2.111 x 3 + 0.897 x 4 + 0.689 x12

+ 1.109 x 22 + 1.552 x 32 + 1.316 x 42 − 0.328 x1 x 2 + 3.274 x1 x 3 − …(9)
0.078 x1 x 4 + 4.495 x 2 x 3 + 5.727 x 2 x 4 − 0.001 x 3 x 4
Subject to xi ≤ 1

Nonlinear regression equation 9 foresees interdependencies among various supply chain parameters.
By solving this equation with suitable optimization tool, optimum level of these factors can be
predicted.
5. Solution methodology
In past, Shang et al. (2004) have solved this type of model by using Response Surface Methodology
(RSM). Measure drawback allied with RSM is to consider only one parameter as a variable during
partial differentiation and rest as a constant. While in actual practice in dynamic supply chain network
all players are interrelated. Therefore it is not reasonable to assume any variable as a constant at any

200
 For Limited Circulation

time. Ridge effect also limits the use of RSM in resolving the regression equations having four or
more variables. To overcome these drawbacks and for exploring search space completely, we are
adopting Psychoclonal algorithm for the optimization of objective function depicted in equation 9.
5.1 Psychoclonal Algorithm: an Overview
The Psychoclonal algorithm (Tiwari et. al.2004) inherits its attributes from Maslow’s need hierarchy
theory and Artificial Immune System (AIS). The silent features of this algorithm are as follows.
5.1.1 Maslow Need Hierarchy Theory
Abraham Maslow was the investigator who devised an interesting theory about human nature by
amalgamating a large body of research allied to human development (Maier, N.R.F. 1965). Based on
Maslow’s research, hierarchy of human needs can be viewed as a personal evolution. This hierarchy
has a Pyramidal structure as shown in figure 5.

Self-actualization
needs

Growth needs

Social needs

Safety needs

Physiological needs

Figure 5 Maslow’s needs Pyramid

In the hierarchy of human need, deficiency is detected from the Bottom (the starting point) to the top
(the end point). Each detected deficiency is dealt with and removed before progressing to the next
level. Analogous to human needs an evolutionary algorithm can also be considered to have
deficiencies. For efficient evolution of an algorithm all deficiencies should be detected and then dealt
with them. The analogy between Maslow need hierarchy theory and an algorithmic evolution is
shown as follows.
1. Physiological needs: The physiological need lies at the bottom of Maslow Pyramid and refers to
the hunger, thirst, and bodily comfort of people. In optimization, this refers to generation of feasible
solutions based upon the problem environment.
2. Safety needs: This is the second set of needs. It has to do physical and physiological safety from
external threats to our well-being. External threats in engineering milieu are constraints imposed on
the problem. For continued existence, solutions are subjected to these threats or constraints and proper
evaluation.
3. Social needs: Next, social needs become dominant; a person will strive for meaningful relations
with others. In the optimization, this refers to the selection of the candidate solution through
interaction between candidates.
4. Growth needs: Every entity desires to produce entities of its kind through reproduction. Here,
candidate solution diversifies to extend its search space. This is the basic mechanism of every
evolutionary process.
5. Self actualization needs: Self-actualization is the need to maximize one’s potential and to the
fulfillment allied with the realizations of one’s capabilities. In term of Maslow “this need might be
phrased as the desire to become more and more what one is, to become everything that one is capable

201
 For Limited Circulation

of becoming”. This is very true whether it is human or solution to an optimization problem. As per
above description, it has been concluded that more the self-actualization needs are fulfilled stronger
the individual become. This is the reason for carrying out a number of iterations to decide the near-
optimal/optimal solution.
5.1.2 A brief description of Artificial Immune System (AIS)
In last few decades the researchers, involved in designing and optimization of engineering systems,
has paid considerable attention to the evolutionary processes like Artificial Immune System (AIS),
Genetic Algorithms (GA), and Artificial Neural Networks (ANN). AIS provides a way to deal with
the complex computational problems like pattern recognition, elimination, machine learning and
optimization.
The vertebrate immune system is a complex system having large number of functional components.
The main task of the immune system is to survey the organism in search of malfunctioning cells from
their own body and foreign substances that are recognized by system called antigens. The constituents
of the immune system that recognize antigens are called antibodies. There are two major categories of
immune cells: B cells and T cells. B cell recognizes antigens free in solution (in blood stream) on the
other hand T cells require antigens to be presented by accessory cells. After generation of T-cells they
migrate in to thymus where they mature. During maturation, all T-cells that recognize self-antigen are
excluded from population of T-cells this process is termed as negative selection (Nossal 1994). If a B-
cell interacts with a non-self antigen with affinity threshold; it proliferates and differentiates in the
memory and effecter cells, this process is termed as clonal selection (Ada and Nossal 1987). In
contrast, if a B-cell recognizes a self-antigen, it might results in suppression as recommended by the
immune network theory (Jerne 1974).
When a non-self antigen is recognized by a B-cell receptor with threshold affinity, it is selected to
proliferate and produce high volume of antibodies as shown in figure 6.

None-self antigen

Proliferation
and Maturation

Selection

Cells with matured

Receptor

Figure 6. Schematic representation of proliferation and maturation

During reproduction the B-cell progenies (clones) having strong selective pressure participates in
hypermutation process. The whole process of mutation and selection is known as maturation of the
immune response.
Depending upon affinity, a selection is made from matured pool of antibodies. This results a high
quality solution that exhibit prudence. From engineering point of view this is the most alluring
characteristic of the immune system because, the candidate solutions with higher affinity must

202
 For Limited Circulation

somehow be preserved as high quality candidate solutions and will only be replaced by matured
clones. In a T cell dependent immune response, the repertoire of antigen-activated B cells is
diversified by two mechanisms: hypermutation and receptor editing as shown in figure 7.

Affinit

Antigen-binding site

Figure 7 Antigen-binding sites

Random changes are introduced in to the genes responsible for the Ag-Ab interaction, and
occasionally one such change will lead to an increase in affinity of the antibody. The hypermutation
operator works in a similar fashion to mutation. The difference lies in the rate of modification, which
depends upon antigenic affinity. Antibodies with higher affinity are hypermutated at low rate, while
antibodies with a lower affinity are hypermutated at high rate. This phenomenon is called receptor
editing and governs the hypermutation that guides the solutions toward local optima, while receptor
editing helps to avoid local optima.
Based on the Clonal principal and Maslow needs the proposed psycho-clonal methodology is
conceived and it is described in the coming section.
5.2 Proposed Algorithm
On considering the evolution of AIS on the track similar to that of Maslow need hierarchy theory we
come across with the Psychoclonal algorithm. Need level 5 make the real difference between AIS and
the Psychoclonal algorithm. In Psychoclonal algorithm the novel feasible antibodies are constantly
introduced to make a boarder exploration of the search space and prevent the saturation of the
population with similar antibodies. This feature inherits a unique ability of maintaining the diversity
of population. Evolution of the Psychoclonal algorithm is shown in figure 8.

Need level 1. Psychological needs: Defining a problem-specific objective function is pre-requisite. A

randomly generated initial population of antibodies depending upon the problem-environment is also
required.
Need level 2. Safety needs: Here, The initial population is exposed to the threats posed by the
antigens.
2.1 From population Agm randomly choose an antigen Agj and expose it to all
antibodies Ab.
2.2 According to the objective function, determine affinity vector f k that contains the affinity of
Ag j to the entire N, antibodies (Ab’s) in the set Ab.

203
 For Limited Circulation

If
gen < max_gen Y
Output: Ab, fk

Self- Memory
Editing(Abd) actualization set(Abd)
needs

Hypermutation (c*k) Growth needs Hypermutation (c*)

Clone (ck) Social needs Clone (ck)

Affinity (f) Affinity (f*)

Safety needs

Generate (Ab) Physiological needs Generate (Ab)

Input: Ab, gen, n, ρ d, fk,, fk*,

Figure 8. Flowchart illustrating Psycho-Clonal Algorithm

The figure clearly depicts the sequence in which Maslow’s needs are fulfilled. Acronyms
used in algorithm are explained in appendix B.

Need level 3. Social needs: Here, interaction is carried out between antibodies with a view as to
identify the relation with each other.
3.1 From randomly generated set Ab select n highest affinity antibodies and compose a new set
Ab k, n of high affinity antibodies in the relation to Agj.
3.2 Now the new composed set of antibodies is cloned proportional to their antigenic affinities,
generating a repertoire C k of clones. Higher the antigenic affinity, the higher the number of
clones generated for each n selected Ab’s
Need level 4. Growth needs: Set Ck is submitted for hypermutation, inversely proportional to the
antigenic affinity, generating a population C*k of matured clones (the higher the affinity, the smaller
the mutation rate).
After satisfaction of need level 4 it is necessary to check the need level 2 once again for these entities,
as they are new denizens of society. Therefore, they must be exposed to the threats and properly
evaluated as per objective function.

Need level 2. Safety needs: Determine the affinity f k* of the matured clones c*k with relation to an
antigen. Now the whole of n, Ab’s are selected to compose the memory set.
Need level 5. Self-actualization needs: finally replace the d lowest affinity Ab’s from Abd and choose
the best among them to fulfill the self-actualization level. As mentioned above, this level becomes
stronger and stronger after a number of generations. Thus, the process is continuing until N = Ngen .
(maximum number of generation).
5.3. Implementation of the Psychoclonal Algorithm for Optimization of Supply Chain
Network
5.3.1 Need level 1
Physiological needs: Defining search space and then selecting antibodies from search space randomly
satisfy physiological needs of problem. The antibodies consist of string of candidate solution, taken
from search space (-1, 1), shown in figure 9.

204
 For Limited Circulation

0.19 -0.59 0.90 0.85

x1 x2 x3 x4

Supply chain’s cost (affinity) = $121.564 million

Figure 9. Representation of initial antibody

Figure 9 provides randomly chosen initial level of all four factors ( x1 , x2 x3 , x4 ) .

5.3.2. Need level 2

Safety needs: By taking solution from the infeasible region to feasible region satisfies safety needs.
Once the solution moves into feasible region its affinity vectors is calculated using objective function
given in equation (9). This equation guides the solution towards local optima and helps in satisfying
initial safety need of the antibodies.
5.3.3. Need level 3
Social needs: antibodies satisfy their social needs by interacting with other antibodies in the solution
pool.
5.3.3.1 Select n highest affinity antibodies from the set of antibodies filtered from need level 2.
5.3.3.2 Selected antibodies are then cloned independently and proportionally using equation (10).
⎛ ⎞
⎜ ⎟
⎜ fi ⎟
η c ,i = φ⎜ n .η ⎟ … (10)
⎜ f ⎟
⎜∑ i ⎟
⎝ i =1 ⎠
η = total number of antibodies
th
η c , i = number of clones generated for the i antibody

f i = antigenic affinity of ith antibody

n = size of population

φ (.) = function for generating closest integer

5.3.4. Need level 4

Growth need: At this level repertoires of the clones created at need level 4 are subjected to
hypermutation. Hypermutation rate is calculated using equation (11) and is inversely proportional to
the antigenic affinity.
(
Π = exp − ρ × f i ) … (11)

Π = hypermutation rate
ρ = decay factor

f i = antigenic affinity of ith antibody

After satisfying the need level 4, the antibodies formed are again attacked by the antigens, affinity
vector of matured clones are calculated using objective function given in equation (9). The weak
clones are edited at this level using receptor editing
5.3.5 Need level 5

205
 For Limited Circulation

Actualization need: Finally, the best solution is retained in need level 5 as a memory set. The weak
solutions are also edited at this level. Thus, with number of generation, the quality of solution
improves in the memory set.
Process repeats till gen = max_gen.
6. Results and discussions
In this research six factors have been considered for the simulation modeling of a supply chain
network. Among these two factors are qualitative and rests are quantitative, therefore they vary on
continuous and discrete scales respectively. Clearly, optimal level of the former can be identified
using Taguchi orthogonal array while latter can not be dealt with this method. In the past Shang et al.
(2004) have made an attempt to identify optimal setting of quantitative factors by deploying response
surface methodology (RSM) on simulation based regression analysis model. Nevertheless, because of
some draw backs coupled with RSM, we have motivated to use a new evolutionary algorithm named
Psychoclonal algorithm for identifying the optimal setting of qualitative factors incurring minimum
supply chain costs. After successful implementation of the Psychoclonal algorithm on the objective
function given in equation (9), we have obtained the optimal level of the quantitative factors in coded
form (in the range [-1, 1]) as listed in table 5.

Table 5. Optimal level (in coded form ) of factors

obtained from Psychoclonal algorithm

S. No. Factors Optimal level

1. X1 -0.48
2. X2 0.26
3. X3 0.96
4. X4 -0.12
By using equation (3) optimal level of these factors in their natural form is enlisted in table 6.

Table 6. Optimal level (in natural form) of factors

obtained from Psychoclonal algorithm
S.No. Factors Optimal level
1. Capacity 94.66 Unit
2. Reorder quantity 34.16 Unit
3. Lead time 10.10 Days
4. Supplier 88.80%
reliability
In contrast to the above results, Taguchi method provides one value as an optimal out of the pre-
determined level of the factors. For example with Taguchi three level orthogonal array, supplier
capacity can only take up one of the three values: 100, 50, and 150 as an optimal, while Psychoclonal
algorithm explores the search space completely in the range [50, 150] and finds the optimum level of
the supplier’s capacity. The optimal level of supplier’s capacity using Psychoclonal Algorithm is
found out to be 94.66 units. Similarly, by using Psychoclonal algorithm the optimal level obtained for
the reorder quantity and supplier reliability is 34.16 units and 88.8% respectively. Authors have also
solved the model presented in equation (9) by employing Artificial Immune System (AIS) and

206
 For Limited Circulation

Response Surface Methodology (refer to appendix C). The results obtained by employing AIS are
listed in table 7 and 8. While, results found out using RSM is provided in table 9 and 10.

Table 7. Optimal level (in coded form) of factors

obtained from AIS

S. No. Factors Optimal level

1. X1 -0.44
2. X2 0.17
3. X3 -0.32
4. X4 -0.85

Table 8. Optimal level (in natural form) of factors

obtained from AIS
S.No. Factors Optimal Level
1 Capacity 94.98 Unit
2. Reorder quantity 33.62Unit
3. Lead time 3.7Days
4. Supplier 79.5%
reliability

Table 9. Optimal level (coded) of factors obtained from

RSM

S. No. Factors Optimal level

1. X1 0.4272
2. X2 -0.4839
3. X3 -0.4299
4. X4 0.7246

Table 10. Optimal level (in natural forms) of factors

obtained from RSM
S.No. Factors Optimal level
1 Capacity 89.976 Unit
2. Reorder quantity 29.64Unit
3. Lead time 8.065Days
4. Supplier 88.209%
reliability

The programming of the Psychoclonal algorithm and AIS has been done in C++ language on IBM PC
with Pentium IV CPU at 1.9GHz. The minimum total supply chain costs corresponding to the optimal
setting of the factors obtained by Psychoclonal algorithm is $117.01 million. In contrast to this, at
stationary point the minimum total costs obtained by AIS and RSM are $ 118.96 million and $ 120.26
million respectively. This shows the advantage of Psychoclonal algorithm over AIS and RSM. In the
Psychoclonal algorithm convergence of result starts on very few generations, thus the computational
burden has been considerably reduced. This can be attributed to the various levels of needs and

207
 For Limited Circulation

hypermutation, which accelerates the multiple local searches simultaneously. The convergence trend
followed by Psychoclonal algorithm and AIS is shown in figure 10.

Figure 10. Convergence trend of solution

psychoclonal
126 AIS

124

122
Supply Chain Cost

120

118

116

114

112

110
0 25 50 75 100 125 150 175 200 225 250 275 300
Iteration

Figure 10 Convergence trend of solution

This figure shows that the Psychoclonal algorithm converges at the 147th generation to achieve the
minimum possible supply chain costs of $117.01 million, whereas AIS converges at the 166th
generations with the total supply chain costs of $ 118.96 million. Evidently, the Psychoclonal
algorithm converges much earlier and with a lower supply chain costs. The certainty of convergence
of the Psychoclonal is due to its ability to maintain the diversity of population. This is because, the
feasible antibodies are constantly introduced to make a broader exploration of search space and
prevent saturation of the population with similar antibodies. In nutshell, the superior performance of
the Psychoclonal algorithm may be ascribed due to its ability to produce a large volume of antibodies
depending upon the needs level. It is found that the probability of finding optimum solution is
enhanced when the pool considered is large. Further, receptor editing helps in escaping local-optima
and the self-actualization needs level makes it stronger and stronger with each generation.
Parameter tuning is a matter of major concern for any optimization problem as it affects the
performance of the algorithm. Fine tuning of the parameters and scrutinizing their correlation with the
problem under consideration needs random experiments. Psycho-Clonal algorithm is explored mainly
by three parameters; namely: n (number of antibodies to be selected for cloning), Nc (number of
clones generated), and d (the amount of low-affinity antibodies to be replaced). These mainly
influence the convergence speed, and the computational complexity.
7. Conclusion
Implementation of an efficient operational strategy for a supply chain is a tough job due to difficulties
in inferring the outcomes of interactions among various factors. To deal with this difficulty, we have
proposed a hybrid approach encapsulating simulation, Taguchi method, nonlinear robust regression
analysis and the Psychoclonal algorithm. First, simulation is carried out to model a comprehensive
supply chain network. Further Taguchi method is deployed to quickly determine a robust area and to
optimize qualitative factors. Subsequently, regression analysis and Psychoclonal algorithm refines the
result obtained by Taguchi orthogonal array. In addition, authors have also done the same by
replacing Psychoclonal algorithm with AIS and RSM. The comparisons of results revealed that the
Psychoclonal algorithm explores the search space better then AIS and RSM. By adopting the

208
 For Limited Circulation

proposed approach, supply chain players can coordinate their operations to achieve the highest
efficiency for the entire chain. This study makes it possible for the firms to understand the dynamic
relations among various factors, and provide guidelines for management to minimize the impact of
demand uncertainty on the supply chain performance. The results obtained help the manufacturer to
determine the proper plant capacity and right level of delayed differentiation strategy for its products.
Further research may include strategic costs relating to design changes of the supply chain. The
strategic costs may comprise product and process redesign costs for postponement, costs for
establishing information sharing through advanced information technology, and costs for reducing the
lead time.

Appendix A:
2.1 Bass model
Bass model is helpful in evaluation of demand of short life cycle products (Bass 1969). In present
research, authors focus their attention on short life cycle product and use Bass model to estimate the
demand. In past, this model has been used by number of researchers (Lawrence and Lawton 1981 and
Thomas 1985).
−( ω +υ )t
1− e
Dt = K
− ( ω +υ )t
1 + ( ω / υ )e
… (12)
dD
dt = t = ( ω + (υ / K )Dt − 1 )( K − Dt − 1 )
dt
Where,
dt = the demand at period t

D(t ) = the cumulative demand up to period t

K = the potential number of customers in the market

ω = the coefficient of innovation or external influence

υ = the coefficient of information at internal influence

Appendix B:
Ab: available set of antibodies
Abd: set of new antibodies that will replace the amount d of the lower affinity
antibodies from Ab.
Abk,n : antibodies from Ab with highest affinity
Agm : population of m antigens.
Ck : population of Nc clones generated from Abk, n
C*k : population after hypermutation.
fk : vector containing values of objective function Y (.) as the affinity of all
antibodies in the relation to antigen
f*k : vector containing the values of antigenic affinity for matured clones
N : total number of antibodies

209
 For Limited Circulation

Nc : total number of clones generated for each of the antigen.

Ngen : number of generations

Appendix C:
Response surface methodology (RSM)
Response surface methodology (RSM) uses a group of mathematical and statistical techniques to find
the optimal condition and response (s).
The second order model (given in equation 6) can be expressed in matrix format as:
Ŷ = βˆ 0 + X ' b + X ' BX … (13)

Where
⎛ βˆ ⎞ ⎛ βˆ βˆ 12 / 2 .. .. βˆ 1k / 2 ⎞⎟
⎜ 1⎟ ⎜ 11
⎜ βˆ ⎟ ⎜ βˆ / 2 βˆ 22 .. .. βˆ 2k / 2 ⎟
⎜ 2⎟ ⎜ 21 ⎟
b=⎜ : ⎟ , Β =⎜ : : : : : ⎟ … (14)
⎜ : ⎟ ⎜ : : : : : ⎟
⎜ ⎟ ⎜ ⎟
⎜ βˆ ⎟ ⎜β /2
ˆ .. .. .. βˆ kk / 2 ⎟
⎝ k⎠ ⎝ i1 ⎠

The set of X i that optimizes the response surface is called the stationary point, where partial
∂Ŷ
derivative is Zero, i.e. = b + 2 BX = 0 … (15)
∂X
⇒ Optimal input level is X 0 = −0.5 B −1 b . … (16)

Optimal total costs at the stationary point thus becomes

Ŷ0 = βˆ 0 + 0.5 X 0' b … (17)

By using equation 10, 12, 14 and table 4 we obtain stationary point as:
⎡ 0.689 − 0.164 1.637 − 0.039 ⎤ ⎡0.717 ⎤
⎢− 0.164 1.109 2.247 2.863 ⎥⎥ ⎢− 1.0037⎥
X 0 = −0.5 * ⎢ ⎢ ⎥
⎢ 1.637 2.247 1.552 − 0.0005⎥ ⎢2.111 ⎥
⎢ ⎥ ⎢ ⎥
⎣− 0.039 2.863 − 0.005 1.316 ⎦ ⎣0.897 ⎦
⎡0.4227 ⎤
⎢− 0.4839⎥
= ⎢ ⎥
⎢− 0.4299⎥
⎢ ⎥
⎣0.7246 ⎦
Minimum total costs
⎡ 0.717 ⎤
⎢−1.0037⎥
Y0 = 120.23+ 0.5 [0.4557 − 0.4839 − 0.4299 0.7246] ⎢
ˆ ⎥
⎢ 2.111 ⎥
⎢ ⎥
⎣ 0.897 ⎦
= $120.26 million

210
 For Limited Circulation

References

ADA, G. L., and NOSSAL, G. J. V., 1987, The Clonal selection theory. Scientific American,
257(2), 50-57.
ALDERSON, W., 1950, Marketing efficiency and the principal of postponement cost.
Profit outlook, September, 1950.
Altiparmak, F., Gen, M., Lin, L., and Paksoy, T., 2006, A genetic algorithm approach for multi-
objective optimization of supply chain networks. Computers and Industrial Engineering,
51(1) 196-215.
ARNTZEN, B. C., BROWN, G. G., HARRISON and TRAFTON, L. L., 1995, Global supply
chain management at digital equipment corporation interface. 25(1), 69-93.
BASS, F.M., 1969, A new product growth model for consumer durables. Management
science, 15, 215-227.
BEAMON, B. M. and CHEN, V. C. P., 2001, performance analysis of conjoined supply chains.
International Journal of production research, 39(14), 3195-3218.
BHASKARAN, S., 1998, Simulation analysis of manufacturing supply chain. Decision sciences,
29(3), 633-657.
CHOPRA, S. and MEINDL, P., 2001, Supply chain management: strategy, planning and
operation (Englewood cliffs, Prentice-Hall).
COHEN, M. A. and LEE, H. L., 1988, Strategic analysis of integrated production distribution
system: models and methods. Operations research, 36(2), 216- 228.
DE Castro, L.N., and VON ZUBEN, F.J., 2002, Learning and optimization using the
Clonal selection principal. IEEE transaction Evolutionary computation, special issue on the
artificial immune system, 2002, 6(3), 239-251.
GAONKAR, R., and VISWANADHAM, N., 2001, Collaboration and information sharing in
global contract manufacturing networks. IEEE/ASME on mechatronics, 6, 366-376.
GAVIRNENI, S., 1997, Inventories in supply chains under cooperation, PhD Thesis, Carnegie
Mellon University, Pittsburgh, PA, September.
GAVIRNENI, S., KAPUSCINSKI, R. and TAYUR, S., 1999, Value of information in
capacitated supply chains. Management science, 45(1).
HARIHARAN, R., and ZIPKIN, P., 1995, Customer-order information, lead times, and
inventories. Management science, 41(1), 1599-1607.
HEWITT, F., 1999, Information technology mediated business process management – lessons
from the supply chain. International journal of Technology management, Geneva, 17(1-2).
JERNE, N. K., 1974, Towards a network theory of the immune system. Ann. Immunol.
(Inst. Pasteur), 1256, 373-389.
Kwak, T.C., Kim J.S., and Chiung M., 2006, Supplier–buyer models for the bargaining process
over a long-term replenishment contract. Computers and Industrial Engineering, 51(2), 219-
228.
LAWRENCE, K. D. and LAWTON, W.H., 1981, Applications of diffusion models:
some empirical results. In Y.Wind, V.Maha, and R.C. Cardozo, (eds.), New product
forecasting (Lexington, MA: Lexington Books), pp. 529-541
LEE, H., PADMANABHAN, V., and WHANG, S., 1994, Information distortion in a
supply chain: the bullwhip effect. Management sciences, 43(4), 546-558.

211
 For Limited Circulation

MAIER, N. R. F., 1965, Psychology in industry, 3rd edition, 417-419 (Houghton- Mifflin,
Boston, Massachusetts).
METTERS, R., 1997, Quantifying the bullwhip effect in supply chain. Journal of
operational management, 15, 89-100.
MONTGOMERY, D.C., 2001, Design and analysis of experiments (New York: Wiley).
NOSSAL, G.J.V., 1994, Negative selection of Lymphocytes. Cell, 76, 229-239.
SAHIN, F. and ROBINSON, E., 2002, Flow coordination and information sharing in
supply chains: review, implications and directions for future research. Decision sciences, 33,
505-536.
SHANG, J. S., LI, S., and TADIKAMALLA, P., 2004, Operational design of a supply chain
system using the Taguchi method, response surface methodology, simulation, and
optimization. International journal of production research, 42(18), 3823-3849.
Shunk, D.L., Kim, J.I., and Nam, H.Y., 2003, The application of an integrated enterprise
modeling methodology—FIDO—to supply chain integration modeling. Computers and
Industrial Engineering, 45(1) 167-193.
SPEKMAN, R. E., 1988, Strategic supplier selection: understanding long-term buyer relation-
ships. Business Horizons, July-August, pp. 80-81.
STERMAN, J.D., 1989, Modeling managerial behavior: misperceptions of feedback in a
dynamic decision making experiment. Management science, 35, 321-339.
STRADER, T. J., LIN, F. R. and SHAW, M., 1999, The impact of information sharing on order
fulfillment in divergent differentiation supply chains. Journal of Global Information
Management, Harrisburg; Jan-mar, 7(1).
SWAMINATHAN, J. M. and TAYUE, S. R., 1998, managing broader product line
through delayed differentiation using vanilla boxes. Management science, December, 161-
172.
TAGUCHI, G., 1886, Introduction to quality engineering (New York: Asia productivity
organization, UNIPUB, White Plains).
THOMAS, R. J., 1985, Estimating market growth for new products: an analogical
diffusion model approach. Journal of Product Innovation Management, 2(1), 45-55.
TIWARI, M. K., PRAKASH, KUMAR, A. and MILEHAM, A. R., 2004, Determination
of an optimal assembly sequence using the Psychoclonal algorithm. IMechE Journal
of engineering manufacture –part B, 219, 137-149.
TOMPKINS, J. A., 1998, Time to rise above supply chain management. Transportation and
Distribution,39(10).
TOWILL, D.R., NAIM, M.M., and Wikner, J., 1992, Industrial dynamics simulation
models in the design of supply chains. International Journal of Distribution and
Logistics management, 22, 3-13.
VANHOUTUM, G., INDERFURTH, K., and ZIJM, W., 1996, Material coordination in stochastic
multi-echelon systems. European Journal of Operational Research, 95, 1-23.
Zhang, D. Z., Anosike, A.I., Ming K.L., and Akanle O.M., 2006, An agent-based approach for e-
manufacturing and supply chain integration. Computers and Industrial Engineering, 51(2),
343-360.
ZHAO, X. and XIE, J., 2002, Forecasting errors and value of information sharing in a supply
chain. International Journal of Production Research, 40(2), 311-335.
ZHENG, Y., and ZIPKIN, P., 1990, A queuing model to analyze the value of centralized
inventory information. Operations Research, 38(2), 296-307.

212