PLANNED LEAD-TIMES DESIGN IN STOCHASTIC MULTISTAGE ASSEMBLY SYSTEMS

摘要

This paper considers a general multistage assembly system operating on a make to order basis with stochastic manufacturing lead-times. The objective is to design optimal planned lead-times by minimising the expected sum of inventory holding costs and tardiness cost. A perturbation analysis based stochastic approximation (PASA) procedure and a Simulated Annealing (SA) method are developed to solve this problem. Compared with the deterministic backward scheduling, the PASA and SA can reduce the costs significantly. The PASA achieves the similar cost to the SA with substantially less CPU time. Case studies using industrial data are given to demonstrate the results.