標題: IC設計公司產能分派模式之構建
The Construction of Capacity Allocation Model
作者: 方淑儒
Shu-Ju Fang
彭文理
李慶恩
Dr. W.L. Pearn
Dr. Ching-En Lee
工業工程與管理學系
關鍵字: IC設計公司;線性規劃;產能分派;IC Design House;Linear Programming;Capacity Allocation
公開日期: 2000
摘要: 台灣的半導體產業,由於分工體系的形成,使IC設計公司更能專注於新產品的研發,而將生產相關的晶圓製造、晶圓針測、IC封裝及最後測試委由專業代工廠製造。這種全數委外加工的生產方式,使得如何掌握下游廠家的生產資訊,及有效運用外部的生產資源,以配合內部的生產計劃,成為IC設計公司的規劃重點。 在日益競爭的產業環境下,為了確保訂單的達交,IC設計公司除須掌握各代工廠的產能狀況,下單前還需慎選合適的代工廠。故本研究主要是以專業IC設計公司為對象,以降低生產成本及延誤率為目標值,顧客訂單為輸入值,在下游代工廠的產能狀況、生產成本、生產時間及機台限制等相關因素的考量下,利用線性規劃建構產能分派模式,並以線性規劃軟體-lingo求得一個最佳產能指派途徑。最後並設計一個求解複雜度的實驗設計,探討本線性規劃模式求解速度的影響因子及模式可承受的複雜度。結果顯示,求解時間的長短取決於限制式及變數的多寡,而模式中影響兩者的主要因子為規劃天數及訂單張數,此外,在下游網路結構較複雜的狀況下,產能利用率亦對求解速度有顯著影響。但在影響因子複雜的條件下,本模式仍有不錯的求解速度,故不失為IC設計公司進行產能指派時的參考。
Due to the trend of deverticalization in semiconductor industry, IC design houses focus on design of new product while outsourcing their product to different subcontractors ( e.g. fabs, packaging houses, testing houses). How to effectively obtain subcontractors’ manufacturing related information and utilize external resources become the major planning issues for IC design houses. IC design houses must not only monitor the status of all subcontractors’ capacity utilization, but also choose suitable subcontractors carefully in order to ensure delivery products to the market on time in the competitive environment. This study develops a capacity allocation model using linear programming for IC design houses. The objective function is to minimize both production cost and unfilled rate. Customer orders are the major inputs. Subcontractor’s characteristics including capacity status, production cost, production time and special machine constraints are taking into consideration. Capacity allocation routes are then obtained by solving the proposed model. In addition, an experiment regarding which factors influence the problem solving speed of linear programming model is designed. The experimental results indicate that the problem solving speed depends on the number of constraints and variables while planning period and the number of orders are two principal influential factors. Furthermore, the utilization of capacity is also an influential factor as the network structure becomes more complex. The proposed model in this research performs well even in fairly complex situations.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT890031023
http://hdl.handle.net/11536/66503
Appears in Collections:Thesis