Agent-based model for optimising supply-chain configurations

With the increasing importance of supply-chain operations on manufacturing successes, an interest for OEM manufacturers is to optimally configure their supply chains to meet customer demand with minimum cost. While a completely dynamic supply chain, where best combination of resources is found and used for every customer order, may be applicable in certain situations, other situations demand a relatively stable supply chain, which evolves over time with respect to changes. This paper proposes a methodology for optimising supply-chain configurations to cope with customer demand over a period of time. With this methodology, a multi-agent system is used to model resource options available in a supply chain as well as dynamic changes taking place at the resources and their operational environment. Demand is modelled by a time-dependent sequence of customer orders, which are processed by the supply chain one after another. Agents within the supply chain interact with tasks in each customer order, under the coordination of an iterative bidding mechanism, to identify the optimum resource combination to satisfy each order. The resulting resource combinations for individual orders are then clustered to identify frequently used resource groups, which are refined further based on qualitative criteria, for the identification of a future chain structure. The method is tested on a simple example and its feasibility confirmed by initial results.