陳靜枝2006-07-262018-06-292006-07-262018-06-292001http://ntur.lib.ntu.edu.tw//handle/246246/18811由於全球經濟體系的成形，企業無論大小都感受到來自全世界的競爭壓力，企業不僅必須加強本身的生產技術與成本控制，更必須在整個供應鏈(supply chain)管理中創造更高的價值及新的市場機會。對於臺灣半導體代工的產業而言，供應鏈管理具價值之處，即為能將所有牽涉產品生產、配銷過程的個體及其相互鏈結與影響均列入考慮；然而也正因影響變數眾多，使得供應鏈管理的研究十分困難。半導體生產的流程可分成四階段：晶圓生產(fabrication)、測試分類(sorting)、封裝(assembly)與檢驗(testing)。每個階段都有不同的廠商可提供加工服務。當晶圓生產廠商接獲訂單，其供應鍊管理者會根據產能、需求量、交貨日、技術水準與成本等考慮因素，決定此訂單應由何晶圓廠區生產、由何測試分類廠做分類、由何封裝廠做封裝與最後由何檢驗廠做檢驗。本研究的主要目的為在各種限制條件下，以最小成本為目標，找出最佳廠商組合。可能的限制包括產能限制、交貨日的滿足、各廠區的技術水準及需求量的大小。本問題可視為產品組合、廠商組合與生產排程的綜合問題，過去常用的解決方法為整數與線性規劃的混合應用，但是由於牽涉的因素太多，常常問題的模式中變數與限制式過多導致無法解決。本研究的第一階段將以資料的收集與模式的建構為主，第二階段則將引用圖學理論與網路流量模式，找出並建構解決問題的演算法，最後並將此演算法與整個半導體產業決策品質促成者結合，整合企業資源規劃系統中的資料，成為其中的一部份。Supply Chain Network Management of the foundry wafer manufacturer considers the demands and supplies triggered by downstream IT related assembly plants or distributors, and upstream wafer material suppliers as well as the production process, transportation process, and the linkage among all these staketakers. A supply chain network of a wafer production involves four stages: Wafer Fabrication, Sorting, Assembly, and Testing. Usually more than one facility could be selected for each stage. Since many facilities could be chosen at each stage, the objective is to find the optimal plant combination that minimize the cost and under certain constraints such as the plant capacities, order quantities, and order due date. The problem could be seen as a plant location problem combined with product mix and due date factors. The usual way of solving this type of problem is the adoption of mix integer and linear programming method that often results in a large formulation and unsolvable situation. The first phase of this study is to review the related literature, collect data from the semiconductor fab, and construct a network configuration and resource allocation model that include all the related variables and parameters. The second phase of this study is to find an easy-to-implement algorithm that provides not only feasible but also near optimal solution to this type of problem. This study will approach the problem from a different angle and solve the problem by adopting graph theory and network flow algorithm. The ultimate goal of this study is to construct an enabler to implement and integrate this algorithm with other modules in a semiconductor SCM total solution in the second phase.application/doc107520 bytesapplication/mswordzh-TW國立臺灣大學資訊管理學系暨研究所供應鍊網路資源分配半導體製造Supply Chain NetworkResource AllocationSemiconductor Manufacturing[SDGs]SDG11otherhttp://ntur.lib.ntu.edu.tw/bitstream/246246/18811/1/892213E002118.doc