MULTIITEM LOTSIZING IN CAPACITATED MULTISTAGE SERIAL SYSTEMS

This research deals with the capacitated, multi-item, multi-stage lotsizing problem for serial systems by examining the performance of heuristics found effective for the capacitated multiple-product, single-stage problem. The items have proportional processing requirements across stages; that is, if item 1 requires twice the processing time of item 2 at one stage, it requires twice the processing requirements at all other stages. The single-stage heuristics included in the study are: Dixon/Silver, Lambrecht/Vanderveken, the Dogramaci, Panyiotopoulos and Adam (both single and multi-pass), and different versions of the ABC heuristic of Maes and Van Wassenhove. These heuristics have been altered in two ways. First, they allow for the inclusion of the cost modification procedures developed by Blackburn and Millen. Second, the feasibility routines have been modified to work in multi-stage environments. Both modifications attempt to coordinate decisions made across stages concerning lotsizes. Experiments were conducted to determine the relative performance of these methods. The experimental design factors include: the TBOs (or time between orders), the coefficient of variation of demand, the capacity available at each stage, the number of products, and the number of stages.