THROUGHPUT TIME REDUCTION THROUGH ON-LINE SIMULATION AND INTELLIGENT SCHEDULING FOR SEMICONDUCTOR FACTORY AUTOMATION

In today's capital intensive (~$2B per facility) semiconductor fabrication facilities, maximizing product output and minimizing throughput-time are two critical metrics to measure manufacturing performance. A method of enhancing fab performance may be attained by implementing an integrated environment where one can analyze factory performance, create and test new work flow management rules via on-line simulation model, and then apply new rules via an intelligent scheduling/dispatching system where rule compliance is ensured.
Traditionally, scheduling has most frequently been accomplished through fab managers' experience and intuition. However in recent years, rule based prioritization has been known to improve fab performance. Nonetheless, this method of prioritization is manual, dependent on individual's interpretation, and does not quantitatively incorporate factory-wide analysis. Conversely, discrete process simulation is a proven method of analyzing factory operations. By integrating process simulation with a dynamic rule-based scheduling system, a factory can be optimized for maximum delivery performance.
With the increasing amount of automation in fabs, accurate models have been developed by extracting information from the manufacturing execution systems (MES). In addition, best known method (BKM) rules are drafted and prioritized for lot dispatch purposes. A snapshot of the current state of the factory can be used to simulate forward to test the BKM rules. Once a set of rules have been optimized and applied, a dispatch list can be called up in real time from each operating station in the factory. Feedback from each transaction is used to update the factory states, which in turn are used to calculate the new dispatch list.