Strategic control of multichannel bulk server diffusion/oxidation processes

Abstract

This research investigates how knowledge of future arrivals can be used to strategically control bulk server diffusion and oxidation processes. The objective is to reduce the average time that lots spend waiting to be processed. A review of past research reveals that several papers have dealt with the control of bulk server queueing systems. However, only one has addressed the use of knowledge of future arrivals and that paper only considered a single product and a single server. This research first examines the single product-single server situation. Next, the single product-multiple servers and multiple products-single server situations are explored. Finally, the multiple products-multiple servers situation is considered. For each situation, a control strategy is devised and then tested through the use of simulation. The performance of each developed control strategy is compared to the performance of the optimal control strategy that does not consider the timing of any future arrivals (i.e. a Minimum Batch size strategy). The experimental results indicate that the control strategies developed perform quite well under most circumstances. In order to see how the multiple products-multiple servers control strategy perform in a realistic setting, a detailed simulation model of an existing wafer fabrication facility was developed. The model was run with several start rates and the results were compared with similar runs that used a Minimum Batch Size strategy. The results indicate that the developed strategy performs well over the wide range of start rates simulated. The developed control strategy has an advantage over a Minimum Batch Size strategy in that it is parameter free. Several areas for future research are identified.