以瓶頸機台負荷評估晶圓廠投料決策

Abstract

本論文以半導體晶圓製造廠為對象，針對同時存在訂單式生產與庫存式生產模式的日投料決策進行研究，此決策目的主要在使瓶頸機台的每日在製品需求變異最小化，以降低產品生產週期時間的變異。本研究發展出四種投料方法(M-1、M-2、M-3、M-4)後，以Flexsim模擬軟體建立模型進行實驗，並利用ANOVA對模擬結果進行分析。其分析結果指出M-1投料方法明顯優於其他三種投料方法。

This research studies the daily job release policy for a semiconductor wafer fab, operating in a scenario which involves both MTO (make-to-order) and MTS (make-to-stock) orders. The objective of the job release policy is to minimize the work-in-process (WIP) deviation of the bottleneck workstation, in order to reduce the deviation of cycle time. Four heuristic methods (M-1, M-2, M-3, and M-4) are developed in this study and justified by a simulation program coded in Flexsim (proprietary discrete event simulation software). Simulation results indicate that method M-1, by the ANOVA test, significantly outperforms the other three ones.