在製品的變化代表生產現場中生產狀況的變動,在實際生產現場中,生管人員常以在製品變化情況作為是否投料的依據。然而,投料時機是否適當,往往影響產出績效甚鉅。 因此,本研究以半導體晶圓生產中之在製品控管與產出績效為研究標的,使用資料包絡分析法對兩者之間進行效率評估,並且透過本研究所提出之績效評估與改善流程將分析結果回饋至原始設定,透過如此的回饋修改機制,以達到產出效率增加的目的。本研究在假設所有生產條件皆不變的狀況下,調整在製品控制總量、瓶頸設備群組在製品控制總量以及初始瓶頸設備群組在製品控制量等三項變數,期望使當月總產出、瓶頸設備群組使用率、訂單達交率以及瓶頸設備月產出的四項產出指標達到最佳。 之後,再以資料包絡分析法分析所得到的建議改善數據,透過本研究所提出之績效評估與改善流程對原始資料進行修正,透過統計檢定檢視本研究所提出之分析流程是否能夠改善產出績效,以確定該績效評估與改善流程確實能夠對在製品控制數量作出修正並且達到更佳之產出績效。 The change of WIP indicates the real situation of work field, production planner usually decides when to release order by checking the change of WIP. However, a suitable time to release order will make a huge affect of production performance. For this reason, this paper will use Data Envelopment Analysis to evaluate the relation of WIP control quantity and production performance in wafer fabrication. We make the use of analysis information to feedback to the primitive set of WIP control quantity in order to achieve a better performance. Under the assumption that all production condition in unchanged, we adjust total WIP control quantity, bottleneck loading, first pass bottleneck loading and expect to improve the performance of total output wafer, CVP, bottleneck group loading , bottleneck output. Afterwards, a hypothesis test will be practiced to examine whether the process is useful in correction of WIP quantity to make a better production performance