Partial characterization of optimal hedging point policies in unreliable two-part-type manufacturing systems
Résumé
The production control problem in unreliable manufacturing systems has generated a great deal of interest recently. The goal is the real-time calculation of production rates to minimize total expected surplus and backlog costs in a production system. A class of feedback control policies, called hedging point policies, has been studied in literature and shown to be optimal for certain single-part-type systems. This paper focuses mainly on the performance of hedging point policies in two-part-type systems. We first generalize known stability results for hedging point policies in single-part-type systems and extend them to two-part-type systems. An average cost analysis of trajectories is then used to partially characterize an optimal policy belonging to this class. A complete characterization is obtained for special parameter values. The results in this paper provide insights for designing simple yet effective heuristics for controlling manufacturing systems producing many part-types.