薄膜電晶體陣列廠機台配置機制之構建

Abstract

在TFT-LCD產業之三階段製程中，由於第一階段之薄膜電晶體陣列(TFT Array)段其機台設備十分昂貴，因此，薄膜電晶體陣列廠中機台數量之配置規劃為一個相當重要之長期決策。此外，產品生產週期時間為一項重要之生產績效指標，同時亦為機台規劃時不可或缺之考慮因素。再者，由於TFT-Array廠中玻璃基板體積龐大，須以自動化搬運系統進行物料之運送，若系統之搬運車數配置不當，將導致廠內工件之生產週期時間拉長，因此，不可忽略搬運因素對於生產系統之影響。 有鑑於此，本文首先估算系統所需之合適車數，使生產活動受搬運系統影響而延遲之可能性降低，同時避免不必要之成本浪費。吾人並在考量搬運因素下，依據等候理論方法發展一套適用於TFT-Array廠之生產週期時間估算法則，以便快速地求算生產週期時間。本文所設計之機台配置機制，將根據生產週期時間以及機台購置成本等資訊，運用排序方法進行機台數量之配置規劃，使系統之生產績效得以維持在一定水準之上，同時追求機台購置總成本之最小化，以確保整體之競爭力。 驗證結果顯示，本文所推估之搬運車數，在維持瓶頸資源利用率於預定水準之前提下，可使得系統產出量達成預期目標，且不致造成車輛資源的浪費；在產品生產週期時間估算值與模擬結果相比較之下，平均誤差小於3%；而透過本文模式所求得之機台數量配置，確實可滿足系統要求之生產績效水準。整體而言，本文所發展之模式，具有即時性與有效性，可作為機台規劃決策之參考依據。

TFT-Array is the first stage among the three-stage manufacturing process of TFT-LCD, which needs quite expensive equipments. Thus, tool allocation for a TFT-Array factory is a significant long-term decision. Besides, product cycle time is one of the major performance measurements, and also an essential consideration of tool allocation. The volume of glasses in the TFT-Array is quite huge, so that if the number of vehicles in automation transportation system is assigned improperly, the product cycle time in the factory will increase. Therefore, the impact of transportation system on the production system should not be ignored. In this study, we first estimate the ideal number of automatic guided vehicle (AGV) for the system to minimize the probability of production delay resulted from the AGV, and to avoid unnecessary transportation cost. Also, we develop a fast cycle time estimation based on Queueing Theory for a TFT-Array factory considering transportation time. Then, we propose tool allocation mechanism, based on the data of the estimated cycle time and cost, to allocate the number of tools by ranking in order to keep the performance level of production system and the minimization of total purchase cost being pursued simultaneously. Simulation experiment reveals that on the premise of keeping the utilization of bottleneck machine at predetermined level, the number of AGV estimated is enough for supporting the achievement of planned target. Compare the estimated cycle time and the simulation result, the average error is less than 3%. And the machine units derived from the tool allocation mechanism can make the production performance truly satisfy the requirements of the system. Consequently, the mechanism proposed in this research is effective and efficient, and can be used as a feasible reference of tool planning.