考量多通路需求公平性與延遲交貨之主規劃排程演算法

Abstract

在全球化的影響之下，彼此息息相關的全球供應鏈也變得更加複雜並且充滿各種不確定性。面對供應鏈不確定性所造成的需求波動，製造商需要發展更能適應環境變動的規劃方式。傳統的規劃方式主要是以最小化成本為主要目標，儘管此種規劃方式能夠為製造商帶來較大的利潤，但也使得製造商在面對需求的波動時處於極高的風險之中。此外，脆弱的供應鏈所造成的需求波動也讓製造商面臨該怎麼在產能不足的情況下妥善分配產能的問題。在過去的生產規劃研究之中，許多研究都假設製造商在規劃時擁有足夠的產能，或是預設製造商會以保守的態度來接受訂單。但在實際規劃與應用時，產能不足使得製造商需要延遲交貨的情形確實是製造商需要面對的重要課題。本研究提出一考量多通路需求公平性與延遲交貨之主規劃排程演算法，解決上述製造商在生產規劃時所遇到的議題。 本研究首先量化公平性和延遲交貨兩個概念，並且將同時考量公平性與延遲交貨的生產規劃問題建立一非線性整數數學規劃模型。由於此模式過於複雜無法找出解答，因此本研究以建立一啟發式演算法的方式予以實作解決，並且以公平性和延遲懲罰成本為指標，與數學模型的求解結果相互驗證。在求解結果表現部分，啟發式演算法與數學模型求解之結果在延遲懲罰成本上平均僅落差0.61%。另外，本研究亦將提出之啟發式演算法應用於一間位於臺灣的精密機械公司，並針對此公司的產品線進行情境設計與實驗。求解規模部分，演算法可將最佳化求解一小時仍找不到任何可行解的案例縮短在1秒之內完成。 綜合以上，本研究所提出之演算法可以迅速並且有效地產出一考量公平性與延遲交貨的主生產規劃排程。對於重視需求公平性以及延遲交貨的製造商來說，此演算法可供其使用；而對於成本導向為主的製造商來說，此演算法的計算與設計亦可做為其調整生產規劃工具的參考。

Due to globalization, global supply chains become more complicated and uncertain. Traditionally, minimizing the total cost is the main objective of master planning. Though minimizing the total cost can bring more profits to manufacturers, this planning method may risk manufacturers overproduction or underproduction when demands fluctuating. Besides, how to properly allocate their capacities when the capacities are in shortage is another issue to manufacturers. Therefore, to solve the abovementioned problems, this study proposes a master planning algorithm considering fairness and backorder. This study first defines and quantifies the concept of fairness and backorder and then formulates a mixed-integer nonlinear programming (MINLP) model to solve the master planning production problem. Due to the complexity of the MINLP model, a heuristic fairness and backorder master planning algorithm (FBMPA) is proposed to generate a master plan which considers fairness and backorder simultaneously. As for the performance of the algorithm, the average difference in total delay penalty between the results obtained by FBMPA and the MINLP model is 0.61%. In addition, FBMPA is applied to the real cases of a machining manufacturing company located in Taiwan to verify the applicability. Compared to the MINLP model which cannot find any feasible solution within one hour, the experimental outcome shows that FBMPA only spends less than one second solving the problems. In conclusion, this study provides an algorithm to help manufacturers generate a fair master plan under the situation of capacity shortage. To manufacturers which prioritizing demand fairness, the proposed algorithm is an effective and efficient planning tool; as for manufacturers whose main planning objective is cost efficiency, the result of this study still can help them improve their planning methods.