Heuristics and Bounds for an Inventory System with an All-or-Nothing Yield Pattern and Lead-times

We study a single-item, periodic-review inventory system with stochastic demand, arbitary delivery lead times and all-or-nothing production yields: whenever a shipment (of an order) arrives, the firm conducts a sample test to confirm the quality; if the sample test fails, the entire shipment is returned or destroyed. Although the problem is impossible to solve exactly except in special cases, we utilize properties of the problem to develop a heuristic that, in numerical tests, performs well in comparison to a popular heuristic and in comparison to the optimal policy.