太陽能廠晶錠－晶圓製程主生產排程規劃系統之構建

Abstract

太陽能廠晶錠－晶圓之製造流程中，存在著雙瓶頸工作站，為長晶站及切晶站，分別屬於存貨式生產及訂單式生產。其中長晶站需很長之加工及整備時間，而切晶站有多種機型，各機型皆以批量機台方式進行生產，各機型各機台有不同之訂單加工時間，加工一次所產出之數量亦有所不同。為成為有效之生產系統並提升競爭力，太陽能廠晶錠－晶圓製程需妥善規劃各機台於各週期之生產計畫，以滿足客戶之訂單需求。 本文針對太陽能廠晶錠－晶圓製程之生產環境及機台加工特性，建構一「太陽能廠晶錠－晶圓製程主生產排程規劃系統」，此規劃系統包含兩個模組，分別為「產能估算模組」及「切晶工作站排程模組」。 「產能估算模組」旨在規劃幅度內，估算雙瓶頸工作站之產能是否足以生產客戶訂單需求。「切晶工作站排程模組」承接產能估算模組之結果，針對屬於訂單式生產之切晶工作站，考量機台產能負荷、整備時間、訂單需求、訂單到臨日及交期等限制，以訂單延遲達交數量最小化為目標，設計一整數規劃模式，求解各機台於各規劃週期加工各訂單之最佳生產數量，期望能使所有訂單皆能如期達交，以滿足客戶需求。 根據案例分析之結果顯示，本文所提出之排程模組可達到整體最佳解。而同時包含多個週期之規劃方式，可節省跨期之整備次數及時間。此外，逐期規劃追求前期剩餘產能盡量生產使得後期有多餘產能以因應客戶之緊急插單。整體而言，本文設計之模組可在有效時間內求得各機台於各規劃週期生產各訂單之最佳數量，可作為現場規劃及生產人員之參考依據。

The manufacturing of ingot－wafer process in solar energy factory includes two bottlenecks, named crystal-growing workstation and wafer-slicing workstation, which belongs to make-to-stock and make-to-order. The process time and setup time on crystal-growing workstation are very long. The wafer-slicing workstation owns multiple-type machines which are in batch processing. For each macine in each machine type, the process time and throughput rate for an order is different. To be an effective manufacturing system and to improve competitiveness, the solar energy factory needs to make production plan properly for each period appropriately so as to satisfy the demand of orders. Thus, this thesis develops a master production scheduling (MPS) system for the solar energy factory for the manufacturing environment. The MPS system includes rough cut capacity planning module and bottleneck scheduling module. The rough cut capacity planning module is used to calculate the capacity supply of each bottleneck workstation and the time needed by all orders. Based on the result of rough cut capacity planning, the bottleneck scheduling module constructs the integer programming model for the wafer-slicing workstation, which considers the machines load, setup time, quantity of each order, release time and due date, to generate the production schedule on each machine and the production amount for each order, with minimizing number of tardy orders as the objective. According to the result of cases analysis, this system can attain optimal solution. Furthermore, by planning multi-planning periods in a horizon simultaneously, it can reduce the number of setup times and the setup time between periods. In addition, the thesis also proposes another scheduling system which pursue earlier period can produce as far as possible enables later period to have surplus produces energy for urgent orders of customers. Hence, the production scheduling system proposed in this thesis can allocate the best production quantity of each order on each machine in each planning period in effectively time.