Optimizing Lot Priority Rules in Semiconductor Production Systems

Abstract

Production processes in the semiconductor industry (e.g. wafer production) are very complex. They are characterized by a high product variety, low-volume production, high cycle time variability and diversity in lot priorities. Low prices and low profit margins lead to immense competitive pressure and semiconductor manufacturers have to bring their products to the market with short lead times according to their contractual agreements. To meet customer requirements in terms of quantity, cost and lead time, high priority lots are pushed through the factory as quickly as possible. This prioritization of certain lots affects the other lots in the line. This results in longer lead times for the non-priority lots and may require additional setup operations. We build a queueing network model to model the manufacturing system under study and develop a hybrid decomposition method to analyse it. We optimize diverse configurations of the system under different contractual agreements regarding lead times/penalty costs.