This Small Business Innovation Research (SBIR) Phase I project will evaluate the use of innovative multiclass fluid models for analyzing factory capacity and finding optimal scheduling policies for semiconductor fabs. Historically, queuing theory models and other analytical tools have not modeled wafer fabs well because of their massive size and highly reentrant flows. Recently, researchers in modern queuing theory have made considerable progress in developing a new type of model known as a multiclass fluid network, which addresses many of the shortfalls of earlier analytic methods. The objective of this project is improved analytic and computational techniques and tools for design, optimization, and simulation of fab operation systems using queuing theory and stochastic processes (as models of complex, dynamic fab systems and processes) for production planning and control in scheduling wafer processing and integrated material logistics (including automated material handling and human delivery to and operation of equipment). The project will show the feasibility of several algorithms for solving fluid models when applied to realistic fab data in a robust, usable, portable, and scalable computing environment. Anticipated results include the development of new software tools to improve the performance of manufacturing production systems. The commercial benefits will be to semiconductor manufacturing companies and other firms that operate large factories with reentrant flows (such as flat panel and disk drive producers). The worldwide semiconductor revenue was over $203 billion in 2000 (with 914 fabs in current operation), and is projected to be nearly $283 billion by 2004 (with an additional 38 fabs on line).