Setting planned job release times in stochastic assembly systems with resource constraints

This paper considers a stochastic assembly system operating on a make-to-order basis with complex product structure and resource constraints. The problem is to find the optimal planned job release times by minimizing the expected sum of the work-in-progress holding cost, product earliness cost and product tardiness cost. A perturbation analysis algorithm is developed to derive the gradient estimate of the cost function with respect to the job release times. This gradient estimate is shown to be unbiased and may lead to the optimal solution by using a stochastic approximation method. Moreover, a procedure is presented to adjust planned job release times to meet service level constraint for each individual job. Numerical examples, which use manufacturing and assembly data from a capital goods company, are given to demonstrate the results.