修正式簡化型限制驅導式方法於迴流流程工廠生產排程之應用

Abstract

傳統的簡化型限制驅導式(simplified drum-buffer-rope，SDBR)方法假設工件只流經產能受限資源(capacity constrained resource，CCR)或瓶頸一次。當傳統的SDBR應用於迴流流程工廠(reentrant flow shop，RFS)時，其生產績效有以下三點可再改進之處：第一是短生產週期工單永遠較長生產週期工單優先加工；第二是處於不同產品但擁有相同的交期延誤緊急程度比率的情況下，無法區別延遲風險較高的工單；第三是處於同一產品且擁有相同的交期延誤緊急程度比率的情況下，也無法區別延遲風險較高的工單。為了改善SDBR應用於RFS的缺失，本研究提出以緩衝狀態偏差率來決定所有工單在單一產品的RFS環境下之加工優先順序，並將該方法擴充至多產品生產環境下的應用。此外，由於傳統的SDBR並未考慮迴流特性與機台的當機情況，本研究針對此種隨機變動的RFS提出了SDBR的交期承諾機制、投料法則與派工法則的改良。此改良的方法首先依照訂單經過CCR的次數重新排序，並依照非遞增順序將多迴流工單的已規劃負荷累加到CCR上，並以此指派訂單交期與投料日期。隨後利用緩衝狀態偏差率作為改良的派工法則以消除機台隨機當機之影響。本研究將所提出之方法應用於一個模擬的RFS環境中，並將所產出之模擬結果與其他三種方法相比。模擬結果顯示本研究所提出的方法在當產品組合中擁有比重較大的多迴流工單或當CCR的利用率由60%增加至90%時，與六個交期相關的績效衡量指標相比後，可有更佳的表現。

The conventional simplified drum-buffer-rope (SDBR) assumed that each job visits capacity-constrained-resource (CCR) or bottleneck at most once. When the conventional SDBR is applied in a reentrant flow shop (RFS), the production performance could be improved via the following three aspects. Firstly, orders with short production cycles are always prioritized than those with long ones. Secondly, orders with high level of risk of delay cannot be distinguished when different products have the same urgency in delivery. Thirdly, orders with high level of risk of delay cannot be distinguished when the same products have the same urgency in delivery. In order to improve the drawbacks of the conventional SDBR in a RFS, this research proposes the deviation of the buffer status to decide the priority of all orders for RFS in a single-product production environment. Later, this research extends the approach to a multi-product production environment. The conventional SDBR does not consider the application of RFS in a random environment which might involve reentry property and machine breakdowns. In this random RFS, the due-date assignment method, order release rule and dispatching rule of the conventional SDBR were improved. This research determines the sequence of the multi-reentrant orders among the planned load of the CCR in nonincreasing order to assign their due-dates and release dates, and using the deviation of buffer status as a dispatching rule to eliminate the influence of machine breakdowns. This research applied the approaches in a RFS where the simulation results were contrast to the other three methods. The experimental results show that when compared to six performance indexes related to the due-date, the approaches have better performance than the other three methods when the product mix with a large proportion of multi-reentrant orders and when the CCR utilization increases from 60% to 90%.