Transshipment policy research of multi-location inventory system based on Multi-Agent System

This paper considers a multi-location inventory system with several retailers who share one supplier. Emergent transshipments happen between retailers when in-hand inventory and pipeline stock are not enough to meet the demand. Our model based on Multi-Agent System (MAS) and game theory allows consultation and local decision making for transshipment. Contract Net protocol mechanism is applied to organize the transshipment consultation between retailers. What makes our research different is that Multi-period transshipment cooperation of retailers is considered as an unlimited repeated game and Robert Axelrod’s evolution theory of cooperation is brought into the transshipment cooperation. Many practical factors, such as demand lead-time, transshipment lead-time and replenishment lead-time, which make the model more complex, are also considered. The optimal inventory control policy is obtained by simulation and optimization with software Anylogic.