台灣半導體設備業零件存貨模式之研究

Abstract

近年來，半導體市場的競爭愈來愈激烈，使得各廠紛紛在成本壓力下尋求解決方案，所以零件庫存問題的重要性已逐漸浮現出來。由於半導體產品製程複雜，常需要至少200到500個反覆步驟，且更新快速，又機台精密度高，零件種類多，大約一部機台就有上百、千種不同的零件；此外，採購前置時間的不易控制及差異大的零件單位成本等因素，都使得零件的存貨管理問題更趨複雜。 論文中係以零件供應商的觀點，構建零件存貨系統，並依半導體機台零件的特性，將存貨模型分為兩類，一是針對使用次數非常少，價格非常昂貴的零件，希望藉由零件的特性、所屬機台的生命週期情況，提供一個簡單的模型，以便求得安全存貨水準與服務水準之間的關係，並且根據不同零件特性訂定不同的服務水準，正確地儲存零件，使得總存貨成本降低。二是針對需求量較大的零件，分別以服務水準及成本結構為考量，建構兩個模型，其一與前者相近，討論存貨水準與服務水準的關係，其二則依其所屬機台的重要性區分為關鍵性零件與非關鍵性零件，採用 (Q, r)存貨政策，將再訂購點以下的存貨量留給關鍵性零件，並於總成本最小的目標下，求出最適訂購量及再訂購點。

Efficient customer service is essential to the supplier as it affects the sales directly. As to the spare parts supply chain in the semiconductor equipment industry, inventory control of spare parts is one of the most important activities for the supplier in Taiwan. Due to the following complex nature of semiconductor manufacturing: machines are upgraded rapidly; they have high precision and many kinds of spare part requirements; one machine can contain hundreds or thousands of different parts; large differences in unit cost and varying demand rate of spare parts in semiconductor equipment; the design and evaluation of spare parts inventory system for local supplier becomes an important issue. This paper aims to establish inventory models for spare pats of the semiconductor equipment. In model I, we focus on the case of spare parts with high cost and low usage. Using varying lead times and machine life cycles, a simple method to evaluate the relationship between inventory level and service level is developed. In model II, we focus on the spare parts in high usage of semiconductor equipment. The same as model I, we evaluate the stock level for given service level. In model III, we discuss the cost structure of inventory cost for the spare parts in high usage. Demand for spare parts can be classified into critical and non-critical demand, depending on the criticality of the equipment for production process. To effectively handle this situation in spare parts inventory control, we propose a stocking policy based on the stochastic continuously reviewed fixed reorder quantity policy (Q, r), where some of the stock is reserved for critical demand.