Title: | 利用DDMRP之緩衝管理及歷史需求與市場預測資訊改善晶圓製造廠之存貨管理 Integrating Historical Demand and Market Forecasts in DDMRP Buffer Management to Improve Inventory Management in Wafer Foundry |
Authors: | 王雅群 Wang, Ya-Chun 張永佳 Chang, Yung-Chia 工業工程與管理系所 |
Keywords: | 限制理論;需求拉動補貨策略;緩衝管理;DDMRP;Theory of constraints;Demand-pull replenishment policy;Buffer management;Demand driven material requirement planning |
Issue Date: | 2015 |
Abstract: | 限制理論(Theory of Constraints, TOC)所提出的需求拉動補貨(demand-pull replenishment)結合緩衝管理管理(buffer management) (以下簡稱為DPBM)的方法較傳統的存貨理論中的補貨策略簡單且容易應用,在產業界也有許多成功應用的實績。目前已有研究提出改善傳統的DPBM方法使其能適用於需求變異大、生產前置時間長且生命週期短產品的存貨管理,但無論是傳統的DPBM方法或是現有對DPBM方法的改良,都還是將目標庫存(即緩衝)等分為三區,依照庫存水位與目標庫存的相對關係作為調整目標庫存的依據,並未將需求的變化情況納入考量。有別於傳統對緩衝的分區方式,本研究應用在需求驅動之物料需求規劃(demand driven material requirement planning,DDMRP)方法中對緩衝分區的概念,利用產品的歷史需求資訊及需求之變異程度為依據以決定緩衝分區大小,使緩衝的分區方式更貼近實際需求狀況。本研究並利用歷史的需求資訊與顧客所提供之對市場的滾動式預測,設計一套補貨與緩衝調整機制。為驗證本研究所提出之方法之可行性與有效性,本研究利用國內某晶圓製造廠所提供之實際產品資料與本研究以模擬方式所產生出之四種不同型態的需求資料對本研究所提出之方法進行測試,研究結果顯示,並與文獻中所提之方法比較,本研究所提出之方法確實能維持高服務水準但卻降低平均庫存,讓DPBM方法應用於需求變化大、前置時間長與生命週期短產品的存貨管理,有更好的效果。 The demand-pull replenishment policy and buffer management (shorted as DPBM) method proposed by the theory of constraints (TOC) is simple and easy to implement. Man successful applications have been reported in practice. There are some studies developed different methods to improve traditional DPBM so that it can apply to management the inventory of the product with large demand variation, long production lead times and short product life cycle. All the existing DPBM-based method evenly divide the target buffer into three zones and adjusting the size of target buffer based on the inventory status. This study, applied the concept presented in the buffer management buffer of the demand driven material requirement planning (DDMRP) method to determine the zone assignment of the target buffer. In this method, the size of each zone is determined by the variable of demand. This study uses real demand data provided by a wafer foundry in Taiwan as well as simulated demand data to demonstrate the efficiency and effectiveness of the proposed method. The numerical experiments showed that the proposed method outperformed the existing ones in terms of average inventory and service level. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#GT070253316 http://hdl.handle.net/11536/126405 |
Appears in Collections: | Thesis |