A Simulation Optimization Approach for Reactive ConWIP Systems

The dynamic adaptation of the number of cards in ConWIP systems has been addressed in a few articles. The aim is to reduce the work in process and to maintain a good customer satisfaction. A common feature of most existing approach is that they use rules, which are based on thresholds, to add or remove cards. These thresholds are known to influence the system performance, so that they should be set in the best way. We present an approach based on simulation optimization to set these parameters in the design of reactive pull system. Our approach also includes a specific module that avoids changing the number of cards too often. The simulation optimization is based on ARENA and OPTQUEST. The suggested approach is illustrated for a ConWIP problem already addressed in the literature. The simulation optimization experiments illustrate the potential of the approach and highlight the fact that the frequency of changes in the number of cards is an important issue. Future research works suggested from these first experiments are pointed out.