A queuing network model for wafer fabrication using decomposition without aggregation

Deterministic capacity models and simulation models cannot fully comprehend the modern wafer fabrication facility´s need for cycle time (CT) forecasts that are both accurate and rapid. Queuing networks, being stochastic yet analytical models, have the potential of providing sufficiently accurate `on the spot´ CT approximations. In this paper, we characterize a queuing network suited for fab modeling and then formulate a queuing network model for a mini-fab, using decomposition without aggregation method to yield CT approximations. Numerical results demonstrate that this proposed approach generates fairly accurate approximations with an average error of 7% compared to simulation results, better than other existing methods to date.