Optimal pacing in an assembly-based multi-stage production system

We examine periodic pacing in a multi-stage non-synchronous production system where production stages are distributed over various locations. The processing rates and inspection failure rates at each stage are random. The optimal policy for pacing (e.g., buffering) the flow of in-process inventory between stages is pursued to minimize operational costs over a finite horizon. It is known that the optimal control is extremely complex in general. However, separability across the stages is shown under weighted penalties which results in a decentralized heuristic pacing. Numerical comparisons of heuristic and optimal solutions are reported. We obtained sufficient conditions under which the heuristic pacing is finite. As a result, an efficient algorithm is developed for the design of inter-stage buffers.