The nature and origin of chaos in manufacturing systems

This paper presents initial efforts to demonstrate chaotic behavior in manufacturing systems, and to explore its origins. We characterize chaotic behavior operationally as small changes bringing about large effects. The manufacturing system studied here by deterministic simulation has only four processing steps and four pieces of equipment. Small changes in this system fall into two categories. We explore changes to policies for the release of raw materials and for the withdrawal of jobs from queues of work in progress, and we investigate changes to the contents and ordering of initial and dynamic queues. Large effects in this system are tracked through changes to throughput time distributions and changes in temporal patterns of finished jobs flowing out of the system. Having shown that small policy or queue changes induce large performance changes, we alter the structure of the manufacturing system to find the origins of the chaotic behavior. This includes the mapping of processing steps onto production machines, the volume of work being added to the system relative to its capacity, and the size of batches being placed into batchable machines. It is shown that each of these factors contributes to the complexity required for the onset of chaotic behavior