607
Decision Sciences
Volume 29 Number 3
Summer 1998
Printed in the U.S.A.
Modeling Supply Chain Dynamics:
A Multiagent Approach
*
Jayashankar M. Swaminathan
Walter A. Haas School of Business, University of California, Berkeley, CA 94720,
email: msj@haas.berkeley.edu
Stephen F. Smith and Norman M. Sadeh
The Robotics Institute, Carnegie Mellon University, Pittsburgh, PA 15213,
email: sfs@ri.cmu.edu and sadeh@ri.cmu.edu
ABSTRACT
A global economy and increase in customer expectations in terms of cost and services
have put a premium on effective supply chain reengineering. It is essential to perform
risk-benefit analysis of reengineering alternatives before making a final decision. Sim-
ulation provides an effective pragmatic approach to detailed analysis and evaluation of
supply chain design and management alternatives. However, the utility of this method-
ology is hampered by the time and effort required to develop models with sufficient
fidelity to the actual supply chain of interest. In this paper, we describe a supply chain
modeling framework designed to overcome this difficulty. Using our approach, supply
chain models are composed from software components that represent types of supply
chain agents (like retailers, manufacturers, transporters), their constituent control elements
(like inventory policy), and their interaction protocols (like message types). The under-
lying library of supply chain modeling components has been derived from analysis of
several different supply chains. It provides a reusable base of domain-specific primitives
that enables rapid development of customized decision support tools.
Subject Areas: Artificial Intelligence, Decision Support System, Simulation, and
Supply Chain Management.
INTRODUCTION
A supply chain can be defined as a network of autonomous or semiautonomous
business entities collectively responsible for procurement, manufacturing and dis-
tribution activities associated with one or more families of related products (see
*The authors wish to thank the associate editor and two anonymous referees whose comments have
greatly improved this paper. The authors also wish to thank Dr. Chae An, Dr. Steve Buckley, and the business
modeling (BPMAT) group at IBM’s T.J. Watson Research Center for introducing the first author to a number
of issues in this domain. The authors also thank Dr. Markus Ettl, Dr. Gerry Feigin, Dr. Grace Lin, and Prof.
David Yao, who primarily developed the tool for asset managers at IBM. This work was funded by an IBM
Graduate Research Fellowship and support from ARPA contracts F30602-91-F-0016 and F30602-90-C-0119.
608 Modeling Supply Chain Dynamics
Figure 1). Different entities in a supply chain operate subject to different sets of
constraints and objectives. However, these entities are highly interdependent when
it comes to improving performance of the supply chain in terms of objectives such
as on-time delivery, quality assurance, and cost minimization. As a result, perfor-
mance of any entity in a supply chain depends on the performance of others, and
their willingness and ability to coordinate activities within the supply chain. A glo-
bal economy and increase in customer expectations regarding cost and service
have influenced manufacturers to strive to improve processes within their supply
chains, often referred to as supply chain reengineering (Swaminathan, 1996). For
example, Hewlett Packard’s Vancouver division reduced inventory costs by
approximately 18% for HP Deskjet printers through delayed product differentia-
tion (Billington, 1994). Similarly, National Semiconductor has managed to reduce
delivery time, increase sales, and reduce distribution cost through effective supply
chain reengineering (Henkoff, 1994).
Supply chain reengineering efforts have the potential to impact performance
in a big way. Often they are undertaken with only a probabilistic view of the future,
and it is essential to perform a detailed risk analysis before adopting a new process.
In addition, many times these reengineering efforts are made under politically and
emotionally charged circumstances. As a result, decision support tools that can
analyze various alternatives can be very useful in impartially quantifying gains and
helping the organization make the right decision (Feigin, An, Connors, & Crawford,
1996). In most organizations, reengineering decisions are generally based on either
qualitative analysis (such as benchmarking) or customized simulation analysis.
This is because complex interactions between different entities and the multitiered
structure of supply chains make it difficult to utilize closed form analytical solu-
tions. Benchmarking solutions provide insights into current trends but are not pre-
scriptive. This leaves simulation as the only viable platform for detailed analysis
for alternative solutions. However, there are two major problems associated with
building customized simulation models: (1) they take a long time to develop and,
(2) they are very specific and have limited reuse. Our aim in this paper is to provide
a flexible and reusable modeling and simulation framework that enables rapid
development of customized decision support tools for supply chain management.
It is essential to understand important issues (decision trade-off) and com-
mon processes in different types of supply chains to develop a generic, modular,
and reusable framework. Our framework is based on supply chain studies con-
ducted in three distinct domains: (1) a vertically integrated supply chain of a global
computer manufacturer (Swaminathan, 1994); (2) a Japanese automotive supply
chain that is less tightly coupled (Sabel, Kern, & Herrigel, 1989); and (3) an inter-
organizational supply chain in the U.S. grocery industry (ECR, 1993). These sup-
ply chains differ in terms of centers of decision making, heterogeneity in the
supply chain, and relationship with suppliers. In the supply chain for the computer
manufacturer we found that the decision-making process was centralized to a great
extent, few suppliers were extremely important whereas others were mainly con-
trolled by the manufacturer, and a major part of the supply chain was owned by the
manufacturer. In the Japanese automotive supply chain, the manufacturer had a
greater control over external suppliers and in some cases partially owned them.
Swaminathan, Smith, and Sadeh 609
However, suppliers made independent decisions many times and the supply chain
involved different companies, though all worked according to the guidelines set by
the manufacturer. In the grocery supply chain, manufacturers and retailers were
equally powerful and sometimes had conflicting interests. The decision making
was decentralized and different organizations (operating under different industrial
environments) were part of the same supply chain.
Despite these differences, we found that there are a number of processes that
are common to these supply chains. We have identified these processes and have
developed a library of software components for modeling them. The library con-
sists of two main categories—structural elements and control elements. Structural
elements (like retailer, distribution center, manufacturer, supplier, and transporta-
tion vehicles) are used to model production and transportation of products. Control
elements are used to specify various control policies (related to information,
demand, supply, and material flow) that govern product flow within the supply
chain. Given this base of primitives, an executable simulation model of a given
supply chain is constructed by instantiating and relating appropriate structural and
control elements. Our framework allows development of models to address issues
related to configuration, coordination, and contracts. Configuration deals with
issues related to the network structure of a supply chain based on factors such as
Figure 1:
Supply chain network.
610 Modeling Supply Chain Dynamics
lead time, transportation cost, and currency fluctuations. Coordination deals with
routine activities in a supply chain such as materials flow, distribution, inventory
control, and information exchange. Contracts control material flow over a longer
horizon based on factors such as supplier reliability, number of suppliers, quantity
discounts, demand forecast mechanisms, and flexibility to change commitments.
Multiagent computational environments are suitable for studying classes of
coordination issues involving multiple autonomous or semiautonomous optimiz-
ing agents where knowledge is distributed and agents communicate through mes-
sages (Bond & Gasser, 1988). Because supply chain management is fundamentally
concerned with coherence among multiple decision makers, a multiagent model-
ing framework based on explicit communication between constituent agents (such
as manufacturers, suppliers, distributors) is a natural choice. We model structural
elements as heterogeneous agents that utilize control elements in order to commu-
nicate and control the flow of products within the supply chain. Our approach
emphasizes models that capture the locality that typically exists with respect to the
purview, operating constraints, and objectives of individual supply chain entities,
and thus promotes simultaneous analysis of supply chain performance from a vari-
ety of organizational perspectives. The modular architecture of our framework
enables one to develop executable models for different situations with limited
additional effort.
A typical supply chain faces uncertainty in terms of supply, demand, and
process. Our framework reduces the effort involved in modeling various alterna-
tives and measuring their performance through simulation under different assump-
tions about uncertainties. This eases the ability of decision makers to
quantitatively assess the risk and benefits associated with various supply chain
reengineering alternatives. In this paper, we describe our framework in its current
state and provide examples to demonstrate how issues relevant to supply chain
management can be analyzed using it. A software application using some of the
concepts from this framework has been developed at IBM.
The rest of this paper is organized as follows. In the next section we review
existing research and approaches. In the following section we describe our multi-
agent framework in greater detail. The section on the supply chain library identifies
key elements required to model supply chain dynamics. We present a cross-dock-
ing prototype from the grocery chain industry in the following section. A full-scale
application developed for IBM asset managers is discussed next and finally, we
provide our conclusions.
LITERATURE OVERVIEW
Benchmarking efforts aimed at identifying new trends and philosophies in supply
chain management based on comparative analysis of current practice in different
countries and different sectors of industry include those reported in Hall (1983),
Helper (1991), and Lyons, Krachenberg, and Henke (1990). Lee and Billington
(1992) provided an insightful survey of common pitfalls in current supply chain
management practices. Some studies indicated that buyer-supplier relationships
are becoming more dependent on factors like quality, delivery performance, flex-
ibility in contract, and commitment to work together, as opposed to traditional
Swaminathan, Smith, and Sadeh 611
relationships based on cost (Helper). Electronic Data Interchange (EDI) and Dis-
tributed Databases have been identified as important technological advancements
that may benefit supply chain performance in a significant manner (Srinivasan,
Kekre, & Mukhopadhyay, 1994). While providing general guidelines and identify-
ing elements of best practice, the benchmarking approach has been of limited help to
managers who are looking for specific quantitative solutions to everyday problems.
On the analytical front, research on multiechelon inventory problems has a
long history (Clark & Scarf, 1958; Svoronos & Zipkin, 1991). A multiechelon sys-
tem is one in which there are multiple tiers in the supply chain. This line of work
typically assumes centralized control of the supply network, thus overlooking the
possibility of decentralized decision making. More recent supply chain models in
this area also include Cohen and Lee (1988), Cohen and Moon (1990), and
Newhart, Scott, and Vasco (1993) in which deterministic scenarios are considered
and a global optimization problem is formulated using mixed integer programs.
Lee and Billington (1993), and Pyke and Cohen (1993, 1994) considered stochas-
tic environments and provided approximations to optimal inventory levels, reorder
intervals, and service levels. Arntzen, Brown, Harrison, and Trafton (1995) devel-
oped an elaborate model for global supply chain management for Digital Equipment
Corporation. The above work has contributed in a significant manner to managerial
decision making. However, these models are limited in handling issues related to
dynamics of supply chains and focus exclusively on global performance measures.
The use of simulation as a vehicle for understanding issues of organizational
decision making has gained considerable attention and momentum in recent years
(Feigin et al., 1996; Kumar, Ow, & Prietula, 1993; Malone, 1987). Towill, Naim,
and Wikner (1992) used simulation techniques to evaluate effects of various sup-
ply chain strategies on demand amplification. Tzafestas and Kapsiotis (1994) uti-
lized a combined analytical/simulation model to analyze supply chains.
Swaminathan, Sadeh, and Smith (1995) utilized simulation to study the effect of
sharing supplier available-to-promise information. Given the utility of this
approach, there is a need for tools that can facilitate rapid development of simula-
tion models. Because simulation models in general have limited reuse, the above-
mentioned tools should provide an environment in which reusable software com-
ponents are essentially combined to construct simulation models for different
problems. Simulation software is more prevalent in the area of business process
reengineering in a broader sense. Swain (1995) provided an extensive survey of
commercial simulation software packages available for process analysis. Among
them, software packages like Extend+BPR, Ithink, SIMPROCESS-III, and Work-
Flow Analyser allow modeling and analysis of business processes. Currently there
is no commercial simulation software that provides domain-specific primitives for
modeling and analyzing supply chain coordination problems. In addition, most of
the above-mentioned software systems are built around simple control mechanisms
for processing events such as first in, first out (FIFO) queues. However, supply
chain interactions typically involve more sophisticated control mechanisms. For
example, when an important order comes in, it may have to be processed first, ahead
of other orders. Also, processing of an item may involve more than just waiting at
the service center for some time. For example, when an order is processed, com-
ponents may have to be assembled and that could, in turn, trigger some events
