球格陣列封裝製程之主生產排程系統設計

Abstract

球格陣列封裝廠具有機台整備時間長、訂單動態來到以及產品會再回流等三大特性。這三大特性使得球格陣列封裝廠難以在考慮最大化產能利用的情況下同時維持高訂單達交率。本論文的目的即在設計一主生產排程系統，用以規劃主生產排程使其滿足高訂單達交率的需求，且能作為新訂單允收評估的參考依據。為了達成此研究目標，準確的生產週期時間估算與良好的機台整備排程具有決定性的影響力，因此本主生產排程系統首先採用區段基礎式生產週期時間估算法來估算生產週期時間，接著再使用混合整數規劃模型來排出最佳的機台整備排程，進而求得主生產排程。當新訂單到臨時，透過前推排程法與既定之主生產排程即可決定該訂單的允收與否。實驗結果顯示，不論在生產週期時間的準確度或是機台整備排程的績效上，本論文提出的方法都具有相當良好之成效。故本主生產排程系統所規劃之主生產排程能有高訂單達交率，亦能作為新訂單允收評估之依據。

Ball Grid Array (BGA) factories are characterized by long machine setup time, dynamical arrival of orders, and re-entrance of products. These three properties make a BGA factory difficult to maximize its capacity utilization while having all orders meeting their due dates. Therefore, this thesis designs a master production scheduling (MPS) system which aims for having high due date achievement as well as doing customer order promising. In order to achieve these goals, obtaining accurate cycle time estimation and good setup schedule is critical. The MPS system firstly adopts block-based cycle time estimation algorithm (BBCT) to estimate the cycle time. Then a mixed integer programming (MIP) model is applied to optimize the setup schedule so that the optimal MPS is planed. When a new order arrives, a forward scheduling algorithm basing on the optimal MPS is used to determine whether the new order can be accepted or not. The experimental results showed that both the accuracy of cycle time estimation and the performance of setup schedules were promised. With estimated cycle times and setup schedules, the master production scheduling system can plan master production schedules which are capable to meet the due date requirement and can handle the task of customer order promising.