Effective sequencing rules for closed manufacturing networks

This paper develops a method for determining sequencing policies to effectively control a multistation closed network. We assume that several job classes, with different service time distributions, share each server and should be sequenced to maximize the long-run throughput rate of the system. A Brownian control problem that approximates the original queueing network is formulated and used to develop a dynamic sequencing policy that seeks to prevent idleness, unless the system is at a face of the workload imbalance polytope that arises in the Brownian formulation. Several examples are used to demonstrate the benefit of this policy over the static prioritization proposed by Chevalier and Wein, which has previously been shown to dominate traditional scheduling policies in a closed network setting.