晶圓代工廠屬於接單式生產型態,製造現場之在製品管制以交期達成為目標,本研究提出一以交期為導向之爐管派工方法,以期能作為晶圓代工廠之爐管機台派工依據。爐管機台具有批次生產 (Batch Process) 特性,當材料進入機台加工前,需先在等候區集結,再一起進行加工,由於爐管機台的加工時間長以及加工過程中不可中斷,因此,需透過控制策略 (Control Strategy) ,在機台開始加工之前決定合適的開工時機,Control Strategy為爐管派工法則設計的核心。 文獻上針對Batch Process之派工法則研究,以降低工件等候時間為主,然而晶圓代工產業的特性,交期成為現場排程重要的考慮因子,不同績效指標之間的取捨,而影響派工決策的訂定,本研究提出以達交率為改善目標之爐管派工法則,稱為最晚可延遲法則 (Wait Until Delay, WUD),在交期允許的時間範圍內,將爐管開工時間延後到最晚,並結合Look Ahead機制,以標準週期時間預估在製品來到爐管的時間,並決定是否等待這些在製品來到後再一起加工,預期能在滿足交期的前提下,提升爐管機台加工批量。 籍由eM-Plant模擬軟體建構模擬模式,系統由八部爐管機台與七種產品組成,各產品之加工時間、加工最大批量與可加工機台不同。實驗設計在三種WIP水準與三種交期鬆弛度的環境因子設定下,探討不同的派工法則對於各項績效指標(交期達成率、週期時間、平均加工批次大小與機台使用率)之表現差異。由模擬結果得知,本研究所提出之最晚可延遲法則,可有效的提升達交率與加工批次。
Foundry is based on build-to-order manufacturing plan, the purpose of inline WIP control is toward delivery on time. This research proposed a due-date-oriented dispatching rule of furnace station which has Batch Process characteristic. In this circumstance, WIP gather in the queuing area first, and then process together. Since long processing time and non-preemptive are furnace manufacturing features, through Control Strategy to determine suitable time to begins to process in advance. Control Strategy effective to the result of system performance most on Batch Process dispatching decision. According to the literatures on Batch Process dispatching, most studies put focus on reducing waiting time. However, in most foundry companies, to fulfill due-date is at the top of production control. The diversity on the way of goal setting from literatures leads to different dispatching rule design. In this paper, we provide a new furnace dispatching rule which intends to improve delivery performance, being called “Waiting Until Due (WUD)” principle. WUD considers to due-date to decide the time span for postponing furnace operation. Furthermore, WUD combines Look Ahead mechanism which estimates future arrival events according to standard cycle time in advance. As a result, WUD is expected to be satisfied to due-date in primary objective and second is raising batch size. The experiment is built and constructed the simulation model by eM-Plant simulation software, made up of eight furnace machines and seven kinds of products on which all products have different processing time, batch size and individual furnace machines. In experimental design, three kinds of WIP level and three level of slack setup on due, probed into different dispatching rules affected to every performance indices, including CLIP, cycle-time, batch size and utilization. From the result, WUD performed on CLIP and batch size in very effective improvement. Keywords: BTO, Batch Process, Control Strategy, WUD, Look Ahead