Modeling the Dynamics of Supply Chains

It is estimated that, on average, 60-70% of the costs of manufactured goods come from raw materials and purchased components [Har92]. As industry moves toward an increasingly more global market economy in which companies focus more on the production of core highvalue-adding components, this gure can only increase. This in turn will further increase the interdependence between manufacturers and their suppliers when it comes to improving due date performance, reducing costs or increasing quality and will put a premium on the ability of managers to grasp the full complexity of the supply chain environment in which their companies operate. Current supply chain analysis techniques and tools often prove inadequate in this regard, due to modeling limitations (e.g. xed leadtimes, inability to account for nite capacities, steady state assumptions, omission of important costs), an inability to take advantage of opportunities provided by Electronic Data Interchange (EDI) technology and a lack of support for operationalizing recent manufacturing philosophies (e.g. \Lean Production”,TQM and JIT). This paper reviews important research issues in supply chain management and presents initial work towards the development of decision support tools for analysis of supply chain dynamics. Our approach relies on the development of an extensible multi-agent simulation testbed in which a wide range of supply chain problems can be quickly and accurately modeled, and alternative solutions to these problems can be compared via simulation. We summarize initial results obtained with an early prototype that indicate some performance e ects of di erent information sharing protocols. 1 Objectives and Motivations To remain competitive, industrial organizations are continually faced with challenges to reduce product development time, improve product quality, and reduce production costs and leadtimes. Increasingly, these challenges cannot be e ectively met by isolated change to speci c organizational units, but instead depend critically on the relationships and interdependencies among di erent organizations (or organizational units). With the movement toward a global market economy, companies are increasingly inclined toward speci c, high-value-adding manufacturing niches. This, in turn, increasingly transforms the above challenges into problems of establishing and maintaining e cient material ows along product supply chains. The ongoing competitiveness of an organization is tied to the dynamics of the supply chain(s) in which it participates, This research was supported by the Advanced Research Projects Agency under contracts F30602-91-F-0016 and F30602-90-C-0119. and recognition of this fact is leading to considerable change in the way organizations interact with their supply chain partners. Broadly speaking, supply-chain management can be subdivided into three inter-related topics: 1. Supply Chain Con guration: Decisions here relate to determination of an optimal number of suppliers as well as the selection of speci c suppliers (internal or external) based on considerations such as quality, leadtimes, costs, reliability, expected learning curves, locations, capacities, earlier experiences, etc. 2. Buyer-Supplier Relations: These decisions have to do with assessing the merits of alternative contracts and agreements between buyers and suppliers. They include understanding tradeo s involved in setting up cost-sharing agreements, determining the length of contracts, agreeing to share di erent types of information (e.g. open-book audits of suppliers), or committing to buying a percentage of the supplier’s capacity. 3. Buyer-Supplier Coordination: Here buyers and suppliers are concerned about identifying e cient coordination policies to maintain a smooth ow of materials and products through the supply chain, avoiding stockouts while keeping inventories to as low a level as possible. Decisions of interest at this level include the selection of proper inventory policies and associated reordering policies (how much to reorder and when) as well as evaluating the impact of di erent information exchange protocols. In this paper, we summarize ongoing work aimed at the development of an extensible modeling and simulation framework for analyzing supply chain management problems. Our objectives are two-fold: 1. First, we are interested in providing new insights into the nature of tradeo s in currently ill-understood aspects of supply chain coordination such as buyer-supplier information exchange, buyer-supplier contractual agreements and buyer-supplier decisions under dynamically changing supply chain relationships. Analyses conducted to date in each of these areas has either su ered in their relevance to practical industrial situations, due to the limiting assumptions that are necessary to construct tractable analytical models, or have been retroactive and limited in applicability, relying on post hoc trend analysis of speci c organizational entities. Our work, in contrast, seeks to construct and analyze models that capture the assumptions and dynamics of these decision tradeo s in actual organizational decision-making contexts, and to provide relevant, prescriptive advice in di erent decision-making circumstances. 2. More generally, we are interested in increasing the relevance of analysis results to practical decision-making contexts, and in providing practical decision-support tools to supply chain management decision-makers. In developing our modeling and simulation testbed, our goal is a modular framework for specifying models of arbitrary delity to a given application context; allowing analysis of decision tradeo s under assumptions that match the actual circumstances facing supply chain managers and their decision-making requirements. We expect, as a by product of investigating the above mentioned tradeo s, to produce an extensible library of model building blocks (e.g., supplier/buyer agents, reordering policies, contractual agreements, information exchange protocols) for subsequent adaptation and re-use. In the longer term, we envision this research leading to the development of practical decision support tools, directly accessible to supply chain managers and integrated with the EDI capabilities of industrial organizations. The balance of this paper is organized as follows. Section 2 brie y introduces concepts and issues in supply chain management and reviews existing research and approaches in this area. Section 3 summarizes the multi-agent modeling and simulation testbed we are developing. In Section 4, we present initial results obtained with an early prototype.