A comparison of the exponential and the hyperexponential distributions for modeling move requests in a semiconductor fab

Variability in any manufacturing process negatively impacts performance since it leads to system disruption. Semiconductor manufacturing, with its characteristic reentrant flow, typically experiences extreme variability. The automated material handling system (AMHS) in a semiconductor fab is subject to this variability and yet must still complete deliveries within a specified time limit. When designing the AMHS the variability used in the simulation model will have a direct impact on the equipment set selected. Sizing a system based on the average case scenario creates a system incapable of meeting the extreme conditions often encountered in reality. The challenge for the modeler of a semiconductor fab is to accurately represent this variability. This paper discusses how the hyperexponential distribution more accurately represents the variability in semiconductor fabs than the typically used exponential distribution