In search of a product mix for semiconductor wafer fabrication facilities by a combined simulation/neural network approach

Abstract

The product mix decision has a direct impact on the process tools utilization. While the commonly seen methodologies may be effective in solving the product mix decision problems from many manufacturing settings, they are not adequate in solving the problems from wafer fabrication facilities due to both the complex nature of the wafer fabrication process and the high tool inventory costs. This paper proposes a combined simulation/neural network based approach to solve the product mix problem for a given tool utilization/WIP level. Solution quality of the proposed methodology is justified through a real world case study. Empirical results illustrate its efficiency and effectiveness in solving a product mix problem for semiconductor wafer fabrication facilities. Significance: Product mix decision for wafer fabrication facilities is intrinsically complicated. The realization of a feasible product mix for a given tool utilization/WIP level will facilitate the planning and control of a manufacturing system. It will also have a significant impact to the tool inventory decision.