Simulation-based optimization of facility locations in supply chain network design and redesign

Abstract

Determining facilities’ locations is a strategic decision in the design and redesign of supply chain networks. It affects the network’s performance over a long time horizon and sets limitations for further decisions. Various decision support and solution approaches exist, typically, utilizing simplified models of the network configuration. However, a multitude of factors, which may not be considered in the model development, influence the real network’s performance. Therefore, all solutions need to be scrutinized before implementing them in the real network to anticipate the extent of the performance’s variation. Simulation modeling facilitates a straightforward evaluation of various factors’ impact on a network’s performance. This approach can be used for the determination of locations as well as generating results for the succeeding analyses. However, the partially considerable computation effort, inherent in this approach, limits its applicability.

An agent-based simulation model for a three-tier retail supply chain network is developed. Various warehouse location alternatives are generated and the locations minimizing the costs and order fulfillment times are selected utilizing the analytical hierarchy process. The selected alternative is subjected to parameter variations of initially constant factors to determine their influence on the model’s performance. These responses are examined utilizing various quantitative analyses. It is found that factors connected to products or facilities with considerably differing specifications have a notable influence on particular responses. However, this effect is restricted to responses on which the adjacent tiers’ behavior has a significant influence. Simulation modeling offers the opportunity to include these factors in the location decision to lessen the difference in the modeled and real network’s performance and mitigate the risk of flawed decisions.