薄膜液晶體陣列廠下雙等級工單之主生產排程設計

Abstract

薄膜電晶體陣列段，因產品具製程規格、垂直鎖定機台與雙等級工單等限制，造成等候時間延長，因而提高產品生產週期時間之變異。規劃主生產排程時不考慮上述限制，將無法精確控制生產週期時間。有鑑於此，針對預定之產品/等級組合及目標產出量，本文擬建立一快速提供主生產排程結果之規劃模組，使產品可準時達交。另外，為得知具長期競爭優勢之最適產品組合/等級比例，本文擬建構一評選機制，以求得良好之產品組合/等級比例。 針對考量雙等級工單、具製程規格與垂直鎖定機台之情境。本論文之「主生產排程模組」，首先建構一混合整數規劃模式，以最小化黃光工作站機台間之產能負荷差異為目標，求得各機台在負荷均衡下之產能分配。再透過Jackson Network演算法得知每一等級產品於各工作站之到達率，最後使用非逐位性等候模式估算各產品於每一工作站之流程時間，並依照產品途程將流程時間加總，以得每一等級產品之生產週期時間。依據估算之各等級產品生產週期時間，估算現場在製品數，並制定對應之投料計劃及派工法則，以確保各機台間產能負荷均勻化，及達成預定生產績效。 驗證結果顯示，本文所提出之主生產排程模組，在考量機台利用率具差異之情況下，所推估之產品生產週期時間，與模擬系統所推估之結果相比，平均誤差在 3%以內；而本文提出之投料規劃，可使生產現場運作能達預定產出目標之99%以上。整體而言，本文所發展之模式，可作為在考量具雙等級工單、製程規格能力與垂直鎖定機台等限制下，規劃主生產排程之參考依據。 本論文之「最適產品組合/等級比例評選機制」，係以上述模組為基礎，運用資料包絡分析法針對產品生產週期時間、訂單達成率與利潤等生產績效指標進行整體評估，決定最具競爭優勢之產品組合/等級比例，以提昇整體生產效率。

In TFT-LCD array environment, the waiting time of products will be extended if products have the characteristics of process window, machine dedication restrictions and priority order. If considering the above characteristics is not considered when planning the master production scheduling (MPS), then it is unable to control cycle time precisely. Hence, MPS mechanism is proposed for such an environment. In this thesis, we develop mix-integer programming to allocate capacity loading between each machine in photo workstation for minimizing the variability of utilization rate at first. Second, in order to calculate the cycle time for every priority order of products, we use Jackson Network algorithm to solve the arrival rate of each product type at each workstation so as to estimate the flow time of products by using non-preemptive priority queuing model. Finally, to ensure the estimated cycle time of products can be achieved. We propose a material release planning module by adopting CONWIP release rule and develop the corresponding dispatching rule. To validate the proposed mechanism, a simulation method is built. The results show that the average of cycle time estimation error is within 3% and 99% of the output target can be achieved. To find the product and priority mix with the most competitive advantage for the factory, we select feasible scenarios as decision making units (DMUs) by applying the proposed MPS mechanism. Then we use data envelopment analysis (DEA) to evaluate and to select sets of the DMUs with considering cycle time, profit, output target, work in process(WIP) and delivery rate.