A Cost Adjustment Heuristic for Dynamic Lot-Sizing with Uncertain Demand Timing

We define a new heuristic, based on the adjustment of set-up costs, for the dynamic lot-sizing problem when demand timing is uncertain. Using a range of simulated problems, we compare the performance of this cost adjustment heuristic with that of a safety lead time heuristic, Zangwill's deterministic algorithm, which ignores uncertainty, and the optimal solution. Results show that the safety lead time heuristic works well for problems with high shortage costs and high probability of early demands, but poorly in other circumstances. The cost adjustment heuristic is quite robust, generally exceeds the performance of the safety lead time approach, and is relatively insensitive to variations in uncertainty.