半導體晶圓代工業具有接單生產的特性,產能規劃需要考慮的因素包含客戶需求限制、客戶偏好廠區限制,工廠產能與製程限制、產品驗證廠區限制,與單一光罩於單一廠區的使用限制等。本研究提出以購買附屬工具(光罩)為增加產能彈性的方法,提供給決策者面臨產能淡旺季時調節產能的建議,並建構數學模型做為產能規劃的機制,有效分配客戶需求預測。 本研究考慮顧客需求為不確定性參數,發展以情境為基礎的兩階段隨機規劃(Two-Stage Stochastic Programming)數學模式,其目的是為了因應需求波動劇烈的情況發生,縱使未來的需求具有不確定性,產能分配的結果仍然是較穩健(robust)的決策。最後利用晶圓代工廠實例驗證本研究的可行性,並建議購買附屬工具和增加產品認證廠區以增加產能分配的彈性,可有效提升產能利用率與客戶需求滿足率。
This study develops an extended stochastic programming model for the deterministic approach proposed by Chen et al., (2013). Capacity planning which plays a role in the production planning is a challenging problem for semiconductor manufacturing industry. There are three original characteristics of semiconductor industry that caused capacity planning become critical, including (1) fast progress of technology in manufacturing process, (2) high manufacturing cost, and (3) great variability in demand. Therefore, this study dealt with uncertain demands of customer for a foundry industry in Taiwan. Through three scenarios, the stochastic programming model is exploited for solving the optimal medium-term capacity planning problem. To demonstrate the value of our stochastic programming model, the data collected from a real wafer foundry in Taiwan, was treated as a fundamental scenario. The forecast demand variations of other scenarios were complied with a normal distribution, the impact of different probability distribution of demand scenarios on capacity allocation was considered. From the series of capacity planning tests with varying initial customer’s demand, changes in customer order fulfillment rate, or capacity utilization rate were determined. Based on different demand scenarios, the numerical results in this study reveal the concordance for the deterministic model and the proposed stochastic programming model. Furthermore, this stochastic model also expressed the flexible consideration of auxiliary tools and rising number of certified fabrication to capacity planning problem in wafer foundry.