晶圓製造廠考量機台當機變異下產能擴充規劃模式之構建

Abstract

晶圓製造為一資本密集與技術密集兼具的產業環境，投資資金相當的龐大。以一座12吋晶圓廠為例，其所需投資金額約為新台幣2100~2400億元，其中，機台設備成本即佔整體成本65%~75%。因此，晶圓製造業者在提供一高水準之製造環境的同時，必須搭配良好之產能規劃。此外，對於現有廠而言，當產能無法滿足未來需求時，晶圓業者必須在預期之未來最有可能之需求組態下進行產能擴充規劃，並能達到預期之績效。然而，由於廠內之現有機台隨著工作站當機頻率之增加，使其機台所需之維修時間拉長，對生產物流順暢產生干擾，進而影響系統整體產出以及降低生產週期時間之穩定性。另一方面，在進行機台購置時，對於同一功能之機台可能會有多家供應商提供相關之機型以及其相關之機台資料，包括可用率，平均機台當機時間等相關資訊，其對於系統亦有不同之績效表現。因此，當進行產能擴充規劃時，決策者必須兼顧廠內現有機台之產能表現以及機型之選購以求滿足預定之績效。 有鑑於此，本文首先於給定之未來需求組態下，根據每個工作站之產能負荷情況決定出需要進行產能擴充之工作站。接著，分別針對這些工作站評估在給定之需求組態下，目前之機台可用產能之量(平均值)以及質(變異數)對於生產績效的影響，並將此結果作一排序以作為產能擴充規劃之順序依據。本文之產能擴充規劃將根據預定之生產週期時間、需求組態以及有限的機台購置成本等資訊，根據工作站產能擴充排序之結果逐一進行機台購置，並確保在規劃過程中，其生產績效可以達到預定之水準，同時搭配投資報酬率之指標用以找出最佳之機型選購目標。 驗證結果顯示，本文之產能擴充規劃模式，相較於一般只透過利用率之排序或是只購置穩定性較高之機台卻未顧及廠內之各機種之產能表現，其規劃結果不但可以達到預定之生產績效，亦可在最低成本下達到最佳之投資報酬率。整體而言，本文所發展之模式，具有即時性與有效性，可作為產能擴充規劃決策之參考依據。

Semiconductor wafer fabrication is a capital-intensive and technology-intensive industry. Take the 12-inch secmiconductor wafer fab as an example, it’s average capital investment can be as high as about NT$2100- NT$2400 billion, of which, 65%-75% will be spent on equipments. Due to high amount of investment, many semiconductor manufacturing companies have to pursue good capacity planning to match with high level manufacturing environment. Usually, semiconductor manufacturing companies allocate budges every year for new types of equipments when the given demand portfolio of each wafer type can’t be satisfied with the capacity currently in the company. However, equipment within a fab is subject to many kinds of variability, such as machine failures, which will disturb the manufacturing flows, will reduce the throughput and the delivery performance. On the other hand, different suppliers provide different equipment models with the same function. These models have different availibility, mean time to repair (MTTR), mean time to failure (MTTF), prices, and etc, will have different performance for a manufacturing system. Thus, planner must consider both the performance of machines currently within the fab and choose the right type of new equipments so as to achieve the planned systment performance. In this study, we estimate the loading of each workstation under a given demand portfolio to determine the workstation whose capacity is needed to expansed. Second, we consider the mean and variance of each workstation’s capacity to rank the purchase sequence for those workstations that need capacity expansion. The capacity expansion plan will be developed based on information about capacity availability and cost. We purchase the tool type according to the rank to keep the performance level of production system and the maximization of return on investment simultaneously. Simulation experiment reveals that the developed capacity expansion plan procedure can make the production performance truly satisfy the requirements of the system. Compared to the other planning method by ranking the utilization rates in order or by choosing the higher performance equipment type, the proposed model can achieve the minimization of total purchase cost and maxization of the return on investment. Thus, the capacity expansion plan model proposed in this thesis is effective and efficient, and can be used as a feasible reference of capacity planning.