A Simulation study of dispatching rules and rework strategies in semiconductor manufacturing

The semiconductor industry is fast paced and on the cutting edge of technology, resulting in very short life spans of semiconductor products. In order to stay competitive, manufacturers must be able to quickly adapt to produce new products, and they must achieve a high level of productivity. Two major operational components of semiconductor fabrication plants (fabs) that effect productivity are dispatching rules and rework strategies. Although prior research has been conducted independently on these two issues, the hypothesis is that the interrelationship between the dispatching rules and rework strategies has a significant effect on the productivity of the fab. Moreover, the goal is to determine which combination of widely-used dispatching rules and new and existing rework strategies results in the highest level of fab productivity. To test this hypothesis, the significance of rework is evalutated, and a four-factor experiment is conducted to determine the effect of dispatching rules, rework strategies, fab types, and rework levels on key fab performance measures. Five dispatching rules are combined with three previously studied rework strategies and the first bottleneck strategy which is developed in this study. The treatment combinations are compared based on fab performance measures including cycle time, percentage on time, work-in-process, and the XTheoretical value. Simulation models based on actual fab data are constructed to carry out the experiments. The detailed results of the experiment show that combinations of dispatching rules and rework strategies have a significant impact on fab performance measures at each rework level in both fab types. In general, two dispatching rules, rework priority and first-in-first-out, in combination with the first bottleneck rework strategy perform the best. Further analysis concludes that the rework priority dispatching rule and the first bottleneck rework strategy result in the highest level of fab performance and are most robust over alterative fab configurations