標題: 新製程導入階段新舊製程產能分配模式之構建
Capacity allocation model during production ramp up stage
作者: 洪正哲
Hung, Cheng-Che
許錫美
Hsu, Hsi-Mei
工業工程與管理學系
關鍵字: 半導體產業;學習良率;產能分配;Semiconductor;Yield learning;Capacity allocation
公開日期: 2010
摘要: 半導體產業為了提升市場的競爭優勢,會不斷地改善製程來降低產品成本,然而發展一新製程需要龐大的研發費用,另一方面IC產品的產品生命週期太短,因此大部分公司都希望能夠快速地將產能轉換到新製程上生產,將所有產能投入到新製程前的這段期間,我們稱為新製程的ramp up。記憶體IC具有新舊製程產品共用的特性,因此在新製程剛導入的階段就能夠上線生產,新舊製程的產能分配成為公司面臨的重要議題。 本研究探討ramp up期間內新舊製程產能規劃的問題。藉由修改的良率模式,將累積投料批量與製程導入時間這兩個因子納入考量,預測ramp up 期間內製程良率的變化,最後透過動態規劃手法,以收益最大化為目標,在瓶頸機台產能限制下,求解每期最佳的新舊製程產能分配。
To increase competitive power, semiconductor manufactures should continuously improve their production process for cutting cost. Shorter product lifecycle and rapid decline of sales price have driven firms to introduce new production process into production line when it is ill understood. The period between completion of development and full capacity utilization is called as production ramp up stage. During that period the yield of new production process is low and unstable. On the contrary the yield of current production process is high and stable. Therefore, current and new production processes are frequently existed simultaneously in production line at production ramp-up stage for meeting customer demands and for cutting future production cost. How to allocate capacity during ramp up stage to current and new production processes is an important issue. The yield learning model is defined as a function of cumulative releasing lots of new production process and learning time. Considered capacity constraints and yield learning a dynamic programming model is established for maximizing total profit to determine the optimal capacity allocation.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT079833540
http://hdl.handle.net/11536/47887
顯示於類別:畢業論文