Priority Cycle Time Behavior Modeling for Semiconductor Fabs

Semiconductor wafer fabrication factories (fabs)often offer wafer manufacturing services of multiple priorities in terms of differentiated cycle time-based targets and fab production must be planned accordingly. This paper aims to develop modeling methods and fab behavior models to capture cycle times of differentiated manufacturing services for semiconductor supply chain management. A novel, hybrid decomposition approximation-based priority queueing network model is designed to characterize how cycle times of individual priorities (PCTs) are affected by wafer release rates and fab capacity utilization. The approach integrates queueing network analyzer that approximates production flow behaviors among tool groups, sequential decomposition approximation among priorities, fixed point iteration to exploit the re-entrant flows and an empirical data-based model tuning technique. Model tuning and validation by comparison with simulation results demonstrate that the decomposition approximation based models yield quick and good quality estimations of PCTs and related performance indices. The fab models thus constructed enable efficient what-if analysis for supply chain management of differentiated manufacturing services.