Optimal control in a failure prone manufacturing system
Abstract
This paper addresses the optimal control in a failure prone manufacturing system. A discrete-time model is used. A single commodity is produced and there is a constant demand. The optimal control policy minimizes the long run average cost incurred by holding inventory and by failing to meet the demand. We show that the optimal control policy in any finite horizon problem is characterized by a critical number, which we call the ideal inventory level (or hedging point). The system should not produce at all if the resulting inventory level exceeds the ideal inventory level, and it should produce at full capacity if the resulting inventory level is less than the ideal inventory level, it should produce exactly enough to meet the ideal inventory level if the ideal inventory level is attainable. We also show that the ideal inventory level increases and converges to a finite value as the horizon length increases. This implies that the optimal control policy in the infinite horizon problem is also characterized by a critical number.