Modeling of supply chain decision logic using PID controllers

Most manufacturing takes place in the context of a supply chain. Each station in the supply chain must not only manufacture a product but also decide how much to produce. This decision is influenced by the supply of materials/components from the next station down in the supply chain and the demand from the next station up. With the advent of increased customization, inventory management is increasingly becoming a critical issue in the manufacturing process. In this paper we model the decision logic at each stage of a supply chain system through the use of system identification and PID controllers. The goal is to investigate and manage the costs of manufacturing a product in the context of a supply chain. It is assumed that the supply chain has well-understood interactions between individual positions, allowing for a focus on the ordering decision logic. A review of ordering strategies is presented, and a discussion of the difficulties in determining PID gains for human decision makers is included. The results show a range of correlation between the PID simulation and measured supply chain inventories. This stems from a number of factors, which are discussed. Additionally, ordering strategies to optimize the supply chain are investigated.