基於可變空間規畫之半導體封測產能重新配置

Abstract

過去幾年來，由於各種新興資通訊產品(如第五代行動通訊、高效能運算等技術)穩步發展，半導體測試之需求不斷將會再提昇。廠商於追求營收與獲利成長的同時，必須考量，如何求公司內部生產設施與外包廠商產能之最適規劃，以兼顧測試品質，並且可避免產能擴充過於快速，於景氣衰退時面臨利用率過低的問題。過去，多有學者導入新型規劃法(De novo Programming)，求廠內與外包廠商產能配置之最佳化，唯新型規劃法所求解與渴望水準(Aspired Level) 仍有所差距，並非實際可達之最佳化。因此，本研究導入可變空間規劃法(Changeable Space Programming)，發展測試廠渴望產能之規劃模型。本研究將以某封測廠產能規劃之案例，實證分析模型之可行性。依據實證研究，可變空間規劃確實較新型與線性規劃法所得之結果為佳。發展完善之架構，將可用於將任何工廠之設備配置最適化。In recent years, the demand for semiconductor testing has continued to expand as a result of the continuing development of new information and communication products (such as fifth-generation mobile communication, high-performance computing, and other technologies). While pursuing revenue and profit growth, manufacturers must consider how to find the optimal planning of the company's internal production facilities and the capacity of outsourced manufacturers, taking test quality into account and avoiding the rapid expansion of production capacity and overutilization during recessions. low problem. In the past, a number of academics have presented a novel planning method (De novo Programming) to optimize the capacity allocation of in-house and outsourced manufacturers. Despite this, there is still a gap between the new planning approach and the aspired level, which is unattainable. The optimal answer Consequently, this study introduces the approach of changeable space programming and builds the planning model for the test factory's production capacity aspirations. This study will verify the model's feasibility using the empirical case modified from a real capacity planning problem of a semiconductor assembly and packaging plant. Based on the results of the empirical study, changeable space programming indeed can derive better results than the ones derived by De Novo and linear programming techniques. In the future, capacity plan of any factory can be optimized by utilizing a well-developed framework.