- 學位論文
半導體設備廠在製品拆卸重工之決策探討
A Study on the Decision Making of Disassembling and Reworking Work-In-Process for a Semiconductor Equipment Company
摘要
隨著半導體市場受全球化影響，顧客需求種類越來越多樣化也越來越廣泛，至使企業須提供更多樣化的客製化產品，以滿足不同顧客需求。但因產品種類的激增，使得企業越來越困難準確預測產品的生產數量及規格。 案例公司主要是以生產預估進行當月顧客所需產品數量並進行組裝，當公司遇到生產預估外的顧客訂單而產生預測錯誤時，會考量將現有在製品拆卸重工以滿足額外訂單的需求，但此舉卻可能會造成總成本增加。因此本研究建構以最低總成本為目標的顧客訂單生產成本模型，並以延遲成本、交貨前置時間及顧客訂單分歧點設置點為影響因子。當額外訂單發生時，依照各項參數值並以達成最低總成本為目標求出拆卸現有在製品的數量來滿足額外訂單需求。 本研究運用顧客訂單分歧點成本模型，藉由各項參數得知各個顧客訂單分歧點設置站的最佳拆卸數量，並探討各項因素對於總成本的影響，將所得數據及最佳顧客訂單分歧點設置進行敏感度分析，且由分析結果得知各項因素對於最佳拆卸數值變動差異。因此建議案例公司將顧客訂單分歧點設置為第四站，且發生額外訂單且需拆卸在製品時，拆卸部分在製品即可，再依據額外訂單遲交成本與預測訂單遲交本比例等相關因素，調整需拆卸數量。
並列摘要
Semiconductor market is highly affected by globalization, and equipment products become very diverse to meet various customer requirements. Therefore, the sales forecasting for customized equipment products has become easy to have errors. While forecasting errors happened, such as the number was incorrect or rush order occurred, the manufacturer would consider disassembling work-in-process (WIP) of the similar products in the plant to fulfill the rush order. Although the rush order can be delivered on time in this way, the regular order might become late and the total production cost would increase. In this study, to solve inaccurate forecasting or excess demand problems, a mathematical programming model was developed to minimize total production cost for a semiconductor equipment plant. The assembly process was modeled as a flow line with four consecutive workstations, and components were assembled or disassembled in each station. This model would decide whether and how many to disassemble WIP of similar products to meet the excess demand based on the goal of minimizing the total production cost. The position of customer order decoupling points (CODPs), the penalty cost for tardiness, and delivery lead time were considered as significant factors in this model and their effects were analyzed. The minimum total cost was found with the CODP positioning after the fourth workstation and sensitivity analysis was applied to that model. Based on the results, this study found that while the CODP moved to the customer side, the disassembled number of WIP decreased. The penalty cost of rush order for tardiness has most significant effects on the total cost, and it is suggested that the disassembled number of WIP should be the decision variable. If the rush order arrives, the penalty cost for tardiness and delivery lead time are inputted into the mathematical programming model, and the disassembled number of WIP will be solved to minimize the total cost.