Operational level-based policies in production rate control of unreliable manufacturing systems with set-ups

• Published in: International journal of production research Vol. 44; no. 3; pp. 545 - 567
• Main Authors: Gharbi, A., Kenné, J.-P., Hajji, A.
• Format: Journal Article
• Language: English
• Published: London Taylor & Francis Group 01.02.2006; Taylor & Francis LLC
• Subjects: Control theory; Experimental design; Numerical methods; Operations research; Optimization; Production management; Production rates; Set-up policy; Simulation; Stochastic control theory; Stochastic dynamic programming; Studies
• ISSN: 0020-7543; 1366-588X
• DOI: 10.1080/00207540500270364

This paper deals with the control of the production rates and set-up actions of an unreliable multiple-machine, multiple-product manufacturing system. Each part type can be processed for a specified period on one of the involved machines. When switching the production from one type to another, each machine requires both set-up time and set-up cost. Our objective is to determine the production rates and a sequence of set-ups in order to minimize the total set-up and surplus cost. As an analytical or even a numerical solution of the problem is very difficult to find, a combined approach is presented. The proposed approach is based on stochastic optimal control theory, discrete event simulation, experimental design and response surface methodology. It is proved experimentally that an extended version of the Hedging Corridor Policy is more realistic and guarantees better performance for two study cases. The first consists of the unreliable one-machine case facing exponential failure and repair time distribution. The second, which is more complex and where the optimal control theory may not be easily used to obtain the optimal control policy, consists of five machines facing non-exponential failure and repair time distributions. To illustrate the contribution of the paper and the robustness of the obtained control policy, numerical examples and sensitivity analysis are presented.