Inventory Management for an Assembly System Subject to Supply Disruptions

Abstract

We consider an assembly system with a single end product and a general assembly structure, where one or more of the component suppliers or (sub)assembly production processes is subject to random supply disruptions. We present a method for reducing the system to an equivalent system with some subsystems replaced by a series structure. This reduction simplifies the computation of optimal ordering policies and can also allow for comparison of disruption impacts across systems with different supply chain structures. We identify conditions under which a state-dependent echelon base-stock policy is optimal. Based on this result, we propose a heuristic policy for solving the assembly system with disruptions and test its performance in numerical trials. Using additional numerical trials, we explore a variety of strategic questions. For example, contrary to what is typically observed in systems without disruptions, we find that choosing a supplier with a longer lead time can sometimes yield lower system costs. We also find that backup supply is more valuable for a supplier with a shorter lead time than one with a longer lead time. In addition, because of component complementarities, we find that choosing suppliers whose disruptions are perfectly correlated yields lower system costs than choosing suppliers whose disruptions are independent, in contrast to the strategy that is typically preferred when choosing backup suppliers for a single product. This paper was accepted by Gérard P. Cachon, operations management.