考慮不確定性因素下回收供應鏈主規劃排程之研究

Abstract

近年來由於環境保護意識的覺醒以及考量回收所帶來的可觀的經濟效益，回收廢棄物再製成新產品已是一種不可以抵擋的新趨勢，也使得回收供應鏈管理議題備受重視。然而在回收供應鏈中納入了商品拆分解的流程，拆解與一般生產製造流程是不對稱的，再生原物料的需求也無法被獨立規劃，故回收供應鏈的主規劃排程較一般製造生產規劃困難。針對回收流程的規劃問題，過往已有研究提出了一些模型與方法，然而這些模型較為簡化，或附帶假設限制，而無法反映出真實的回收環境。因此，本研究提出了完整的回收供應鏈結構，同時考量供給量與需求量的不確定性因素，致力解決回收供應鏈規劃問題。 本研究考量多種需求商品與供給廢棄物、多層級產品結構、規劃多期、有產能限制下，以間斷時間模式進行回收供應鏈中各期的生產、運輸、存貨、整備活動之規劃。因為考量供給與需求量之不確定性，本研究採用存貨政策來控制各個成員在各期間的生產以及運輸的數量，目標為求長期最大利潤下，制定各成員合適的存貨政策。對此規劃問題，本研究先建構對應的隨機模型，並提出一啟發式演算法，使得在有效率的時間下，獲得一趨近最佳解的可行方案。 本研究考量長鞭效應會負面影響供應鏈的利潤，因而認定供應鏈中每個成員都須採用相同的基本存貨政策。本啟發式演算法致力縮減存貨政策的搜尋範圍，並透過模擬來搜尋出最合適的存貨政策。其流程為：先找出網路中最重要的節點，接著針對該節點的成本結構與整體網路資訊決定出存貨政策的搜尋上下界，再使用三層的搜尋方法獲得一最佳存貨政策。 最後，本研究實際建立出此規劃系統，並進行情境分析與實例討論，用以驗證本研究之啟發式演算法可行且具高效率性。

Because of the environmental awareness and economical reasons, the manufacturers are pushed to recover the used products and thus the recycling supply chain is recently receiving a lot of attentions. However, disassembly and assembly processes are asymmetric so that the planning problem for the recycling process is different from the one for the regular production process. Although some studies have focused on solving such problems, their models are simplified with unrealistic assumptions. In this study, we focus on solving the master planning problems for the recycling supply chain with uncertain supply and demand. The recycling supply chain network includes members such as collectors, disassemblers, re-manufacturers and re-distributors with the recycling processes from collecting the returned goods to distributing these recovery products to market. The objective of this study is to maximize the total profit of the entire recycling supply chain. Considering the stochastic property, this study institutes the stocking and processing policies for each member of the recycling supply chain to better respond to the unknown upcoming demand. To solve the master planning problems for the recycling supply chain with supply and demand uncertainties, we propose a stochastic model. To improve the effectiveness and efficiency of finding a solution, a heuristic algorithm, Heuristic Stochastic Recycling Process Planning Algorithm (SRPPA) is proposed. The idea of SRPPA is to narrow down the search space and compares the long-term profit result of some valuable sets by simulation to find the best one. The main process of SRPPA consists of three phases. In Leaders Finding Algorithm, SRPPA determines the most important node to be the leader of the recycling supply chain. In Candidate Policies Sets Generating, SRPPA defines the searching range of the inventory policy for the leader and forms the candidate policies sets based on the characteristics of the leader. In Step Best Policies Set Selecting, SRPPA constructs the simulation process for each inventory policy candidate to find the best one. To show the effectiveness and efficiency of SRPPA, a scenario analysis is conducted.