Optimization of Re-Entrant Hybrid Flows With Multiple Queue Time Constraints in Batch Processes of Semiconductor Manufacturing

Abstract

This paper proposes optimization methods of re-entrant hybrid flows with multiple queue time constraints in batch processes of semiconductor manufacturing. A smoothing flow concept and a dispatching rule for the re-entrant flows called “re-entrant flow smoothing” [REFS(X)] and a loading rule for the processes with multiple queue time constraints called “synchronized control of kanban and buffer sizes” (SCKB) were developed and evaluated. These optimization methods aimed to both maximize the throughput ratio of quality products and minimize the lead time of work-in-process. A three-view modeling to combine the objective functions was created to respond to an actual problem scale. For a manufacturing process consisting of five flows and four re-entrant flows with seven queue time constraints, SCKB-REFS(X) improved the throughput ratio of the quality products by 20.0% compared with first-come first-served (FCFS)-conditional time in 40 days. The mean lead time was also shortened by 60.1% because SCKB-REFS(X) could reduce the setup rate by 83.1% and improve the overall equipment effectiveness by 41.9%, though the performance of minimum inventory variability scheduler application rules is almost the same as that of basic FCFS. REFS-SCKB(X) computed the problem in 4 s or less.