| 摘要: | 本研究探討半導體封裝廠生產排程問題。個案公司雖為全球封測業前五大公司之一，但是受到客戶要求、工廠製程能力的限制、晶圓凸塊工件迴流特性及機台金屬靶材的限制，造成UBM (Under bump metallurgy)站點經常成為個案公司的瓶頸站點。當瓶頸發生時，個案公司沒有優質的生產排程工具及有效的排程方法控管瓶頸站點的排程，堆貨情況未能有計畫性的改善，導致生產管理相當困難。面對大環境的競爭，產品逐漸走向少量多樣的生產型態，生產管理變得複雜，且滿足客戶交期為工廠績效相當重要的指標。因此，本研究將以個案公司UBM站點之生產排程問題為研究方向，開發新的排程方法及改善排程工具，以提升個案公司生產管理能力。
本研究發展出一套派工演算法，主要以關鍵比值法(Critical Ratio, CR)、優先權(Priority)、最短加工時間法(Shortest Processing Time, SPT)為組合派工法則。透過CR值判斷產品為正常批(生產進度符合預期)，則派工改以SPT法則為排序依據，且優先權高的產品不再有派工的優先權，並使用個案公司生產資料做為研究資料，依據派工演算法流程圖之步驟完成實驗結果，將個案公司現行排程模式做為對照組，產生兩種排程方式的實作結果。最後，選定兩項績效衡量指標分析研究結果，研究結果顯示，本研究所得到的單位時間產量較個案公司現行排程模式高，單位時間產量增加10.9%，本研究方法所得到的平均在製品數量較個案公司現行排程模式低，平均在製品數量減少18%。兩項績效衡量指標皆能說明本研究方法是有效的，主要是本研究方法把CR大於或等於1的產品視為正常批，就算優先權為1的產品也是如此，且正常批的派工原則是以加工時間愈短的產品愈先執行。本研究方法能有效地消化瓶頸站點之在製品數量及改善優先權低的產品落後情況，期望本研究能做為個案公司未來改善生產排程問題的參考方向。
;This study explores the production scheduling issues of semiconductor packaging plants. Although the case company is one of the top five companies in the global packaging and testing industry, it is limited by customer requirements, factory process capabilities, wafer bump workpiece reflow characteristics, and machine metal targets, resulting in UBM (Under bump metallurgy) sites It often becomes the bottleneck site of the case company. When a bottleneck occurs, the individual company does not have high-quality production scheduling tools and effective scheduling methods to control the scheduling of the bottleneck site, and the stockpiling situation cannot be improved in a planned way, making production management very difficult. In the face of competition in the general environment, products are gradually moving towards a small number of diverse production types, production management has become complicated, and meeting customer delivery dates is a very important indicator of factory performance. Therefore, this research will take the production scheduling problem of the case company’s UBM site as the research direction, develop new scheduling methods and improve scheduling tools to enhance the case company’s production
management capabilities.
This research has developed a set of dispatching algorithms, mainly based on the combination of Critical Ratio (CR), Priority, and Shortest Processing Time (SPT). If the product is judged to be a normal batch based on the CR value (the production schedule is in line with expectations), the dispatching is changed to the SPT rule as the ranking basis, and the products with higher priority no longer have priority for dispatching. According to the production data of the individual company and the steps of the dispatch algorithm flow chart, the experimental results are completed. The current scheduling mode of the case company is used as the control group, and the results of the two scheduling methods are generated. Finally, two performance measurement indicators are selected to analyze the research results. The research results show that the output per unit time obtained by this research is higher than the current scheduling model of the case company, the output per unit time increased by 10.9%. The average number of products in process obtained by this research is less than the current scheduling model of the case company, the average work-in-process quantity has been reduced by 18%. Two performance measurement indicators can show that this research method is effective. The main reason is that this research method regards products with CR greater than or equal to 1 as normal batches, even products with priority of 1. The product with the shorter processing time is executed first. Therefore, this research method can effectively digest the number of products in progress at the bottleneck site and improve the lagging situation of low-priority products. It is hoped that this research can be used as a reference direction for the case company to improve production scheduling problems in the
future. |
