Stochastics and Statistics Short-term robustness of production management systems: A case study

Whereas Operations Research concentrates on optimization, practitioners find the robustness of a proposed solution more important. Therefore this paper presents a practical methodology that is a stagewise combination of four proven techniques: (1) simulation, (2) optimization, (3) risk or uncertainty analysis, and (4) bootstrapping. This methodology is illustrated through a production-control study. That illustration defines robustness as the capability to maintain shortterm service, in a variety of environments (scenarios); that is, the probability of the short-term fill-rate remains within a prespecified range. Besides satisfying this probabilistic constraint, the system minimizes expected long-term workin-process. Actually, the example compares four systems––namely, Kanban, Conwip, Hybrid, andGeneric––for the wellknown case of a production line with four stations and a single product. The conclusion is that in this particular example, Hybrid is best when risk is not ignored; otherwise Generic is best; that is, risk considerations do make a difference. 2002 Elsevier Science B.V. All rights reserved.