THROUGHPUT ANALYSIS IN MANUFACTURING NETWORKS

The throughput of a plant is a measure of major importance when assessing its ability to compete successfully in the market place. Managers often rely on changes in capacity (production rate) and process improvements as two major factors that impact throughput. The optimal allocation of resources to these two factors is difficult to determine without the support of appropriate mathematical models. In this paper we attempt to quantify the tradeoffs between capacity and process improvements, through variance reductions, and throughput. We consider multiproduct manufacturing systems modeled by open networks of queues and formulate the throughput characterization (TC) and variability reduction (VR) problems as nonlinear programs. These formulations are based on the decomposition approach for estimating the work-in-progress in open queueing networks. We show, by demonstrating the applicability of greedy-type heuristics for the TC and VR problems, that the overall impact of a wide variety of process improvement practices on WIP and throughput can be evaluated very efficiently.