Risk Analysis of Robust System Design

Robustness has become a popular issue in engineering. For products, Taguchi suggests to adjust the design so that product performance is insensitive to the effects of uncontrolled environmental variations. For systems (to be distinguished from products), we propose to adjust the design so that the risk of getting poor performance is minimized. This risk is evaluated by simulating the system over a sample of environmental scenarios: this procedure yields an estimate of the probability distribution of the outcome. We illustrate this risk/uncertainty approach to robustness through the comparison of four pull productioncontrol systems, including Kanban and Conwip. We conclude that the traditional approach consisting of optimizing the design of a system for a specific environment (base scenario) may be risky.

• Type: Conference
• Year: 1998
• Where: WSC
• Authors: E. G. A. Gaury, Jack P. C. Kleijnen
• Keywords: Modeling And Simulation, Popular Issue, Robustness, Uncontrolled Environmental Variations, WSC 1998