Performance evaluation of due-date based dispatching rules in dynamic scheduling of diffusion furnace

Abstract

There has been extensive scheduling research relating to use of existing dispatching rules along with/without new dispatching rules and compared their performance behavior in job-shop, flow-shop, open-shop, flexible manufacturing system, and single machine with unit capacity environments using various scheduling objectives. However, it appears that there is no comparative study on analysis of dispatching rules for scheduling bottleneck batch processing machine in discrete parts manufacturing, particularly the diffusion furnace (DF) in semiconductor manufacturing (SM). This study addresses this research issue. For that, this study first, proposes the mathematical models for dynamic scheduling (DS) of DF to optimize the due-date based scheduling objectives: Total weighted tardiness, on-time delivery rate, total earliness/lateness, and maximum lateness. Due to the computational intractability of each the proposed mathematical models for large-scale problem, this study proposes greedy heuristic algorithm (GHA) based on due-date based dispatching rules (DDR). Because, dispatching rules are widely used in the SM industry. Accordingly, in this study twenty variants of GHA-DDR are proposed by considering various due-date based dispatching rules to compare the effects of due-date based dispatching rules in DS of DF. From the series of computational analysis carried out in this study, it is observed empirically that the proposed variants of GHA based on apparent tardiness cost (ATC) and batch ATC (BATC) dispatching rules yield consistently better solution for most of the scheduling objectives considered in this study. This observation is further verified by statistical analysis: Friedman test and Nemenyi multiple comparison test.