晶圓製造廠產品組合最佳化之設定

Abstract

在競爭市場中，晶圓製造廠必須在追求利潤最大化的同時，充分考慮產能的有效利用及生產的平穩度。而本論文之目的，在於提供一個設定產品組合結構時的有效分析模式，以達到較佳化之生產製造。其主要的三個方法為：資料包絡法 ( DEA )、層級分析法( AHP )及網路分析法( ANP )。 首先是應用資料包絡法( DEA )，在不需專家意見的前提下，亦即不用預先分配權重下，對不同的穩態生產環境進行績效評鑑，並依其多元投入與多元產出項目的績效比值，求解出應如何設定生產組合。接著再利用層級分析法( AHP )，以階層分析投入與產出過程，同時結合Delphi法，考慮專家認為各要素之重要性，以獲得較理想且符合需求的產品組合結構。最後應用網路分析法( ANP )，此法主要以層級分析法為基礎，但增加考慮要素間之交互作用，來決定優先權重。簡易網路分析法可用來快速處理簡單要素間之交互作用，而完整的網路分析法則可處理較複雜要素間交互作用。 本論文可提供晶圓廠設定接單策略之有效分析模式，在生產效率及利潤最大化之考量下，並同時考慮多元投入與多元產出以維持生產之平穩化，以選擇產品組合結構。本論文所提供之方法淺顯易懂，除方便管理者使用外，並可做為有效決策之參考。此分析模式可因應不同的情境與管理需求，提供管理者系統化之解題流程與模式，以選取做出較適宜與較有效的產品組合結構。

Wafer fabrication factory is facing a very competitive market nowadays, and how to acquire a higher profit while maintaining production smoothness and utilizing current capacity effectively become a must. The purpose of this dissertation is to present effective approaches to find a set of product mix for the company to achieve a better performance. Data Envelopment Analysis (DEA) is first used to measure multiple inputs and outputs of product mixes in a semiconductor fabricator under the assumption that experts’ opinions on the importance of factors are not necessary; that is, the pre-assignment of weights to the factors is not required. An efficiency score for producing each product mix relative to other mixes can be obtained. Analytic hierarchy process (AHP) approache is next taken to analyze multiple performance factors, incorporating experts’ opinion on their priority of importance, to obtain suitable product mixes for semiconductor production. Analytic network process (ANP) approach is lastly adopted to consider the interrelationship among the factors. Simplified ANP is used to deal with simple interaction among factors while comprehensive ANP is suggested to solve complex interrelationship among factors in different levels. The results obtained from the various approaches can provide guidance to a fabricator regarding strategies for accepting orders in the attempt of maximizing the manufacturing efficiency and the profit, while simultaneously considering other important input and output factors for maintaining manufacturing smoothness. The models can be easily understood and followed by administrators to determine an efficient product mix for a fab. Depending on the environment and managerial requirement about whether the weights of factors need to be considered and whether the interrelationship of factors should be concerned, different systematic procedures are provided here to let practitioners pick the most suitable model to adopt.