An analysis of the "K-step ahead" minimum inventory variability policy(R) using SEMATECH semiconductor manufacturing data in a discrete-event simulation model

Abstract

This paper addresses one alternative to optimizing resource scheduling, the Minimum Inventory Variability Policy(R) (MIVP(R)). The second level of K-step ahead MIVP(R), will be compared to the first level 1-step ahead MIVP(R) and in turn each to first-in-first-out (FIFO) using a SEMATECH factory dataset (including random machine failures and repairs) to reduce cycle time. The heuristic explanation and theory behind MIVP(R) including Little's law (1961) and Kingman's formula are described. Justification to use simulation modeling is discussed. Cycle time (CT) comparison results are presented. A secondary observation of comparing one simulator to another using the same factory dataset and identical scheduling rules is presented.