Summary: | 碩士 === 中原大學 === 工業工程學系 === 88 === Semiconductor manufacturing is growing very fast. In addition to the development in products and manufacturing technologies, operation management such as releasing and dispatching is one way for the IC fab to become competitive in the market.
In this research, a dynamic state-dependent dispatching (DSDD) algorithm is proposed for a foundry fab. Fab equipment is dynamically classified in to bottleneck machines when its average utilization of previous day is higher than a threshold value. For a dynamic bottleneck machine, different dispatching rules are used based on its queue length. When the queue is long, a TB+ rule is used: select the lot with the largest WIP deviation and surplus deviation. When the queue is short, an STNV rule is used: select the lot with the shortest expected processing time until its next visit to the bottleneck machine. For a dynamic non-bottleneck machine, a FGCA+ rule is used: select the lot with the least number of steps toward a bottleneck machine. Besides, three different lot release rules are studied in this research.
AutoSched simulation models are built to evaluate the performance of the proposed DSDD algorithm. A foundry fab with five major product types each consisting of 300 steps is studied. Simulation results show that DSDD can lead to better performance measure such as cycle time and wafer in process than the other dispatching rules.
|