Efficient setup/dispatching policies in a semiconductor manufacturing facility

Abstract

Semiconductor manufacturing requires repetitive use of several similar processing operations. Owing to economic considerations, these expensive machines are always shared among many products or a product at different stages of its manufacturing process. The machines require setups between the production of different products or different entries of the same product. Since a product may enter a machine several times for processing, the manufacturing system is often called a re-entrant shop. We consider the problem of reducing mean and variance of the cycle time in semiconductor manufacturing environments. Manufacturing systems are usually controlled in two major ways. In the first, a release policy specifies when new lots are to be released into the system. Whereas in the second, a scheduling policy controls the lots already in the system. It decides which lot is to be processed next at each machine, when it becomes available for processing. A bulk of previous research on the control of manufacturing systems focuses on the control of bottleneck machines. Performance measures of a manufacturing system depends on the ways in which the conflicts are resolved between products competing for the machines. Lu et al. (1994) develop a class of scheduling policies for systems where there is no machine setup required when switching products. We deal with a more general situation in which the machines require setups. We introduce scheduling policies incorporating setups, since the policies developed in the literature for systems requiring no setups are found to perform poorly for systems requiring considerable setups.