This paper proposes using mixed inventory policies to determine ordering quantities for all organizations in a supply chain. Agents of organizations choose inventory policies from a set of candidate policies, and determine their ordering quantities for each period according to the chosen policies. A genetic algorithm is formulated to search for the optimal combination of inventory policies to minimize the total costs of the whole supply chain over a time horizon. To illustrate and evaluate the performance of our approach, many different setups of experiments are conducted by computer simulation of an MIT Beer game.
International Journal of Modelling and Simulation 26(2), pp.129-136