Planning order releases for an assembly system with random operation times

A multi-stage assembly network is considered. A number of end items should be delivered at a certain time. Otherwise a delay cost is incurred. End items and components that are delivered before they are needed will cause holding costs. All operation times are independent stochastic variables. The objective is to choose starting times for different operations in order to minimize the total expected costs. We suggest an approximate decomposition technique that is based on repeated application of the solution of a simpler single-stage problem. The performance of our approximate technique is compared to exact results in a numerical study.