A Comparative Study of Centralized and Decentralized Decision Making on Flexibility Investment

This paper studies the flexibility investment decision making problem in a manufacturing system. Two different scenarios of decision making, i.e., centralized and decentralized, are investigated. In the centralized case, we formulate this problem as a two-stage stochastic program. In the decentralized case, the multi-agent simulation technology, bidding mechanism and contract net protocol are jointly employed to simulate the subcontracting activities among autonomous plants in the manufacturing system. Simulation results show that by taking the changeover cost into account, shortfall in the centralized case using the ldquochain effectrdquo principle is lower than that in decentralized case. However, if the demand variability faced by different plants differs greatly, the ldquochain effectrdquo principle for flexibility investment would not be effective anymore. To solve this problem, we propose a novel investment principle based on variability matching, which is proved to be effective by experimental results.