晶圓代工廠限制驅導式生產管理系統之應用

Abstract

目前在台灣的半導體業有應用限制驅導式排程與管理技術(Drum-Buffer-Rope， DBR)於生產管理的公司包括有台積電、聯電、華邦電、旺宏、茂矽…等多家公司，但是這些公司在限制驅導式排程與管理技術的實際推展上，對於限制驅導式管理的精神是否吻合，以及是否存在一些差距無法真正的落實，對於此眾說紛紜的推展成果，必須加以釐清。 本研究期望藉由個案公司的檢討，針對號稱為限制驅導式排程與管理技術(Drum-Buffer-Rope， DBR)的A公司，實際生產運作情況是否有符合DBR的精神，對於實施不夠完善的部分，說明原因以及提出改善方案，期望能夠讓個案公司藉由檢視生產系統的過程，探討其在運用此技術時，對於DBR似乎僅僅是”觀念”的運用，而在實際的生產線上所面臨的困難以及不足的部分加以強化，並將在DBR推展上無法真正落實以及尚有差距的部分加以釐清，在檢視過程中不斷的提出改善生產系統方法，例如：以實體的晶圓代工廠依據DBR所發展出適用於A公司且能達到合理且有彈性之關鍵層別比值法 (Critical Ratio by Critical Layer，CRCL)，來建立整廠的的派工模式，期望藉由持續的改善過程使A公司的生產系統能夠更趨近於真正的DBR管理的公司。 由於半導體景氣的循環，景氣由谷底翻揚到現在的活絡，出貨量由5000片/月攀升到25000片/月，在出貨量增加5倍的情況下， A公司在兩年半的期間不斷的檢視生產系統且持續的改善，才有生產週期時間縮短9%以及產品達交率增加27%的成果，但是我們還是要持續不斷的努力讓系統更完善，期望此研究能夠給予更多推展DBR瓶頸管理的公司有一些啟發，能夠正視問題的癥結點，不要誤以為僅只有DBR的觀念應用於生產系統就是導入DBR的公司了。

In the semiconductor industry in Taiwan, the technology of Drum-Buffer -Rope (DBR) management is widely adopted in many companies such as tsmc, UMC, Winbond, MXIC and Mosel etc. However, in the real application of DBR management, there still exist unclear results to be clarified regarding whether the gist of the theory is really carried out and whether some gaps exist in its execution.

A case study, Company A that announced adoption of DBR in its production, is examined in this research. At the beginning, company A is considered to satisfy the DBR spirit, but only concepts are employed after a series of examination. This thesis tries to deal with the difficulties that Company A faces in the production line and suggest further improvements to strengthen the operational system in the hope that Company A becomes a real company applying DBR.

Buffer Management is used to examine the system of company A. Continuous improvements are proposed during examining procedure. For example, according to DBR management, a much more reasonable and flexible auto-dispatching model via the CRCL (Critical Ratio by Critical Layer) method is developed and established in the whole factory of company A. In two and a half years, due to the hot season coming, the shipping quantity is five times in production, the performance of cycle time is shortened to 9%, and the on-time delivery is increased to 27%. Continuous improvements are still performed to make the system more perfect. This research is expected to inspire more companies to employ DBR management to face the real problem. One is not a DBR company if it just introduces the concept to the production system.