半導體廠商為了維持其技術及產能競爭優勢，必須持續投資先進製程之生產設備，高度設備資本支出的產業特性，其設備維修在成本結構中佔有相當程度的比重。 為了維持設備的可使用時間(Availability)，減少因保養或故障之產能損失，以及控制維修成本等考量，在保固期過後通常會評估是否與專業廠商簽訂維修合約或是由廠內自行維修保養，以取得較低之設備持有成本(Total Cost of Ownership)。 本研究期望透過資料探勘方法，從龐大的設備保養及生產系統等資料庫中，分析不同廠牌機型之成本特性並建立成本模型，以此為依據並配合公司目標策略以決定合約計價模式，達成與委外廠商簽訂符合雙方利益之合約，以建立長期之合作關係。 透過資料探勘分析，我們求得三種典型的廠牌機型設備之成本模型與其建議之合約計價模式。 1.LAM KIYO2300: 固定成本模型，適用「製程別」之「固定價格」 2.AMAT Enabler: 變動成本模型，適用「固定費率」之稼動率「時間基礎定價」 3.EBARA FRXS: 半變動成本模型，適用「變動費率」之稼動率「時間基礎定價」 ;Semiconductor manufactures continuously invest on production equipments for conversion to advanced process technology and capacity expansion. Therefore, the industry has feature of high equipment capital with maintenance cost. Considering maintenance cost and equipment availability for avoidance of capacity loss due to long period of maintenance or unscheduled down time, manufacturers normally make cost-benefit assessment comparing in-house maintenance and outsourcing contract, purpose to get lower Total Cost of Ownership. The thesis intended to use data mining technique to build up cost model of variety of equipment vendor types, gathering data from big database of maintenance and production system. The resulting models will be basis for making pricing strategy of outsourcing contract, that the service terms can fulfill requirements for both manufacturer and sub-contractor, to build up a win-win situation of long term relationship.
We concluded 3 typical cost models of vendor type equipment and recommended contract pricing method respectively: 1.LAM KIYO2300: Fixed cost model， use “Fixed pricing” by process 2.AMAT Enabler: Variable cost model，use fixed-rate “Hourly pricing” 3.EBARA FRXS: Semi-variable cost model，use non-fixed-rate “Hourly pricing”
