以限制理論需求拉動補貨策略改善擋控片庫存管理-以M半導體工廠為例

Abstract

半導體業發展已經超過40年，其特性為兼具技術與資本的產業。晶圓製造的良率是決定製造成本高低的一項關鍵，而穩定的機台生產條件是維持晶圓製造的良率必備條件。為維持好的機台生產條件，晶圓製造廠便需要擋控片來穩定及監控機台生產狀況。一旦發生缺貨，將會造成生產線機台停擺。所以，為避免產能損失，生產線都會備大量的擋控片庫存因應，但同時大量的擋控片庫存也會造成成本浪費及資金積壓。

本研究以M半導體工廠庫存水準為例，依據限制理論(Theory of Constraints, TOC)以需求拉動補貨與動態緩衝管理方式，透過實際數據的模擬與驗證，來證明比現行使用預測需求補貨的方式，更能夠在避免缺貨的狀況下，提供管理者能用”較低 ”的庫存水準來備庫存，進而提昇工廠競爭能力。

Semiconductor industry has been developed over 40 years. It has been known as a capital and technical intensive industry. The yield of wafer manufacturing process is a key factor to decide the manufacturing cost and a stable equipment condition is a must to sustain the wafer manufacturing yield. Wafer foundry has been using dummy & control wafers to stabilize and monitor its manufacturing condition of the semiconductor equipment for a long time, once there is a shortage of the dummy & control wafers, the manufacturing equipment would be stopped. To avoid manufacturing lost, production line needs to keep large amount of stock for dummy & control wafers. At the mean time, the large amount of stock on dummy & control wafers have created a lot of cost waste and capital burden.

This study is using the Inventory Level of M Semiconductor Foundry as an example. Based on the Theory of Constraints (TOC) and via actual data simulation and verification, this study is applying both the Demand-Pull Replenishment Inventory Control and Dynamic Buffer Management to prove that Demand-Pull Replenishment Inventory Control method is more effective than current Forecast Demand Inventory Control method. Furthermore, under the wafer shortage free condition, management team is able to keep a lower inventory level and enhance the competition of a wafer foundry.