良好的設施規劃可提昇生產力、減少物料搬運距離、增加機器設備使用率、充份的利用空間並提昇工廠的競爭力。背脊式設施佈置設計是目前半導體製造廠常見的佈置方式,此種佈置是將工作站沿著中央走道設置,並於中央走道中設置一中央天車物料搬運系統。其目的是為了使晶圓搬運距離最低、物流最便捷並減少搬運之碰撞與震動。 本文提出並探討五種晶圓製造廠設施規劃啟發式改善法︰兩階段改善法、陡降程序法、修正陡降程序法、門檻接受法與混合改善法,求解晶圓製造廠工作區佈置方案,以縮短晶圓搬運距離。混合改善法結合修正陡降程序法並門檻接受法架構,並採用空間填滿構想,取代共同邊或面積相等之傳統交換限制。混合改善法先進行修正陡降程序搜尋區域解,而後運用應用門檻接受法跳脫區域解,反覆搜尋一較佳之佈置組合。 我們將所發展之改善型演算法以案例加以測試,測試案例為8至20個工作區之晶圓製造廠。以晶圓製造廠中三種平均200個製程之主要產品作為測試依據。測試結果修正陡降程序法、門檻接受法與混合改善法於求解時間上表現良好,大幅縮短求解時間,可快速求得近似最佳解之搬運成本,其求解品質績效接近96 %以上。其中混合改善法具有較低之求解變異。
Good facility planning of wafer fabs lead to a higher productivity, shorter wafer move distance, higher machine utilization, better utilization of plant space, and better competition. A spine bay configuration arranges bays along a central spine, and each bay is served by a central automated material handling system. This is a typical design for wafer fabs. This design results in low transportation cost and an efficient flow to reduce vibration during the transportation. This research proposes and evaluates five layout improvement heuristics for spine bay fab design: Two-Phase, Steepest Descent, Steepest Descent Plus (SDP), Threshold Accepting (TA) and Hybrid. Hybrid heuristics combines TA and SDP and adopts space-filling concept to manage traditional place constraints. This heuristics first uses SDP to search local solutions and then uses TA to jump out from local optimal and continue to search better solutions. The layout improvement heuristics are evaluated on the basis of fab layouts with 8 to 20 bays. Three major products with an average of 200 steps are taken into account. SDP, TA, and Hybrid heuristics lead to better solutions then the other. The quality of solution is more then 96%. Among these three heuristics, Hybrid has the lowest solution variation.