Fast fab scheduling rule selection by ordinal comparison-based simulation

摘要

In this paper, an ordinal comparison (OC)-based simulation tool is designed and applied to achieve fast selection of wafer release and lot dispatching rule combination for fab operations. By comparing relative orders of performance among scheduling rules to a specified level of confidence, the OC approach reduces simulation time significantly. The tool consists of (1) a discrete event simulator, (2) a fab model database, (3) a library of scheduling rules, (4) a library of performance indices, (5) an ordinal comparator, and (6) a computation budget allocation. Rule selections from a set of prominent fab scheduling rules under frequently considered fab performance indices such as mean and variance of cycle time and smoothness are studied over various time horizons by using a benchmark fab model. Results demonstrate a potential improvement in efficiency over traditional simulation by two orders of magnitude. In addition to insights about static selection of rules over various objectives and time horizons, our simulation studies also indicate the necessity of dynamic selection.