This Small Business Innovation Research Phase I project has the potential to substantially reduce the mean cycle-time as well as the variance of cycle-times in semiconductor integrated circuit manufacturing plants. Semiconductor manufacturing plants possess a special type of re-entrant process flow that is unlike that found in traditional manufacturing operations such as flow shops and job shops. This project proposes the development of new scheduling policies based on what we call the First Law of Queueing Systems, that reducing fluctuations in flows reduces delay. In preliminary studies these policies have shown the ability to reduce mean waiting time by about 20% and standard deviation cycle-time by 40% on single process flow plants. The scheduling algorithms will address issues related to batch operations and set-up times in addition to the diverse random disruptions occurring in a production facility. The performance of this new class of scheduling policies will be evaluated through the development of a stand-alone simulation environment that will incorporate realistic models of semiconductor production facilities.
