Simulation-Based Work Load and Job Release Control for Semiconductor Manufacturing

Cycle time and capacity management are two of the most important issues for manufacturing companies, particularly in a dynamic business environment. The goal is to meet customer delivery dates while minimizing manufacturing cost. Semiconductor manufacturing involves a large number of process steps, reentrant process flows and long production cycle times. In addition, technological innovation in the semiconductor industry is rapid, resulting in shorter product life cycles. Thus, semiconductor manufacturing companies must continually invest in order to keep pace with new technology and update their equipment to produce new products. These conditions make cycle time and capacity management more important for semiconductor manufacturing companies. For the purpose of stable control of cycle time and capacity, there have been many studies dealing with semiconductor manufacturing shop floor control. This paper focuses on job release control, which aims to keep the work load stable and meet customer delivery dates in the presence of dynamic demand fluctuations. We introduce a hierarchical control model which consists of two layers: a long-term capacity decision model and a short-term job release control model. Parameters of the proposed linear programming (LP) models are estimated using simulation, which acts as an intermediate layer between the two layers. Numerical experiments on a mini-fab demonstrate the potential gains from using our model, which achieves improved customer delivery and shorter cycle time relative to several conventional job release policies.