服務性零組件最後訂購情況下最適服務水準之研究 －以汽車產業為例

The Optimal Service-Levels of Final Order for Auto Service Parts

滕喆安(Chi-An Teng)

指導教授 ： 蔣明晃

共同指導教授 ： 郭瑞祥(Ruey-Shan Guo)

國立臺灣大學/工學院/工業工程學研究所/碩士(2009年)

https://doi.org/10.6342/NTU.2009.00463

摘要

對許多產品公司而言，包括汽車製造商，在車輛產出後，由於無法直接在世界各地提供製造後服務，因此通常會透過當地代理商利用其原先所建立的通路來取得售後服務的先機。由於用以維修及更換的服務性零組件服務期間往往比車輛生產期間還長，在車輛停止生產後，零組件所需持續服務的期間通常還需持續一段時間。在此時零組件生命週期多數已步入衰退期，零組件供應商為考量經濟生產批量的成本因素下，並不會持續生產至服務期間結束。因此，這時代理商面臨上游供應商即將停產某些服務性零組件的情況，得做最後一次訂購以滿足剩餘服務年限內所有的需求。代理商必須在成本控管與服務水準之間取得平衡，決定最適的最後訂購量。
然而，若代理商的最後一次訂購太多將產生滯銷存貨，訂購太少則會有顧客抱怨與銷貨損失。現行汽車產業為方便起見，通常以歷史資料的平均需求做為預測起始參考依據，假定服務期間結束時需求為零，剩餘年限需求以一定數量隨時間遞減，如此不僅過於直覺且可能造成多餘存貨。
因此，本研究希望利用卜瓦松迴歸模型，建立一套更有效的衰退期零組件需求預測模式，但是零組件的種類太多，因此接著根據不同零組件中需求資料不確定的特性以FCM法進行分群，制定各群的服務水準優先順序，以提出有效的存貨管理機制。最後在報童模型極小化呆料與缺貨成本的情況下，滿足整體服務水準，求得各群最適訂購量，各群內之零組件皆以該群最適訂購量向上游廠商訂購，如此可減低訂單複雜度。並以實際需求資料比較個案T公司和本研究提出最後訂購方法之優劣，使在面臨最後訂購問題的廠商，不僅能以更具成本效益的訂貨方式進行訂貨決策，而且可以有效的滿足顧客對各種類別所有零組件的需求。

關鍵字

最後訂購 ； 服務水準 ； 卜瓦松迴歸 ； Fuzzy C-Means法 ； 報童模型

並列摘要

For many car manufacturers, to sell products in the global market usually need local agents to provide after-sales service through their channels. Because the periods of maintenance and replacement of the service parts have much longer than vehicle’s production periods, the service parts requirement will last for a certain period of time even the vehicle manufacturer stops producing the car. The period after the vehicle stops production is called the end-of-life service period.
At that period, the majority of parts are within the declining phase of part life cycle. The supplier of service parts may no longer manufactures the parts after the car stop producing for considering economic production scale. Before the suppliers stop supplying service parts, the agents must place a final order to meet future demands at a certain service level to lower the customers’ complaints at the minimum cost. If the final order size is larger than actual demand, the salvage parts become obsolete inventory. On the other hand, if the final order size is smaller than actual demand, it will cause customer complaints and results in lost sales.
Therefore, this study firstly establishes a Poisson regression model for service parts in the period of declining demands. Secondly, a FCM method is applied to cluster service parts into different groups based on their costs and essentialness such that service level priorities can be determined in order to provide effective inventory management. Next, the newsboy model considering dead-stock cost and shortage cost under the overall service level constraint is developed to determine the optimal order quantities for each group. Parts within each group are ordered the same optimal quantity to reduce complexity. Finally, we use the real data from company T to compare our model with current practice of company T. The result shows our model reduces cost effectively and meets customer needs for various categories of parts.

並列關鍵字

Final Order ； Service Level ； Poisson Regression ； Fuzzy C-Means ； Newsboy model

參考文獻

1.林姿依，建立適合顧客關係管理之模糊分群模型 - 以汽車維修服務為例，國

2.陳育珊，考慮產品生命週期末端之服務性零組件最後訂購模型–以汽車產業

1. Anderberg, M. R., “Cluster Analysis for Applications,” Academic Press, 8, 1, (1973), 2-7

3. Bezdek, J. C., “Pattern Recognition with Fuzy Objective Function Algorithms”,

5. Cameron, A. C. and Trivedi, P. K., “Econometric Models Based on Count Data: Comparisons and Application of Some Estimators and Tests”, Journal of Applied Econometrics, 1, (1986), 29-53