Heuristic policies for inventory ordering problems with long and randomly varying lead times

Abstract

In most inventory models, lead time is an insignificant factor in inventory modeling and implementation. There are situations, however, in private industry and governmental procurement in which this is not the case. For example, orders for castings and forgings with a low order size may have lead times of from 12 to 36 months. The actual lead time is random and depends on economic factors that determine the load at the foundry. A model developed by Liberatore in 1977 [1] was the first approach to incorporate random lead time and yet had a feasible solution approach for a multi-period problem. However, this approach requires significant development time for implementation on a digital computer by a technically skilled analyst. Our sensitivity analysis of this model indicates that a solution of the model for every specific case may not be necessary. In fact, some heuristic policies are revealed that offer a manager a finite set of alternatives to consider with some assurance that the policy selected will be close to optimum. Our conclusions are: 1) Consider 1,2 or 3 orders per year with the order sizes roughly equal. 2) The time of the first order should be approximately E (lead time) + 1.25σ LT prior to the first demand point. 3) Large increases in σ LT will not affect these policies.