供應鏈網路中考量回收機制之主規劃排程演算法

Abstract

在競爭激烈的商業環境中，如何整合供應鏈內成員彼此間的政策及行為，以達到整體供應鏈利益的最大化，為目前大多供應鏈管理所探討的問題。而對於供應鏈的定義，隨著全球對環保議題的重視，已經從傳統自供應商至顧客端的範圍，擴大為自供應商至顧客端，再經過回收處理後提供製造商重新利用的過程。本研究即為考量回收機制之主規劃排程研究，以達到整體供應鏈最佳化為目的。 本研究屬於先進規劃排程中之主規劃排程，規劃供應鏈中期的採購、生產與配銷計畫。研究中考量多個最終產品、規劃多期、有多張需求、產品結構具有共同料之特性，再加上回收規費的設定及產品回收流程，以間斷時間模式進行整體供應鏈的規劃。 本研究對於此規劃問題提出一多目標模型為：以最小化總需求延遲成本滿足顧客需求為首要考量；當延遲成本獲得最佳解之後，再以最小化總生產、處理、運輸、存貨成本與回收規費為目標。在本研究中，規費的罰則為影響整體供應鏈成本的重要因素，法令設定的規費高低也會影響製造商等相關供應鏈成員的行為。因此規劃時必須同時考量變動成本與回收規費之間的變化關係，才能使整體供應鏈最佳化。 本研究之規劃問題若以混合整數線性規劃模型求解，在問題規模龐大時，需要花費大量時間求解或完全無法求得解答。因此本研究提出ㄧ啟發性演算法，使得本研究問題能在有效率的時間下，得到一趨近最佳解之可行解決方案。 本研究啟發性演算法之流程為：先進行規劃排程之前置作業，接著進行需求排序，完成後依照排序結果開始為每張需求做規劃。規劃之步驟為設定網路連結成本，尋找需求微調的機會，尋找最小成本之生產路徑，分配產能，增加存貨，以及調整網路結構等。以上規劃步驟重覆進行直到所有需求規劃完成為止。最後，本研究建立一規劃排程系統，並進行情境分析之實例討論，以驗證本演算法實為一可行且高效率之規劃求解選擇。

In a competitive business environment, how to coordinate the policies and behavior of organizations and facilities in supply chain is crucial to the performance of all chain members. And with the increased environmental concerns, the definition of supply chain has been extended and contained the recycle process, so-called green supply chain. Considering multiple final products, multiple discrete planning periods, multiple demands, recycle process, and recycle penalty, this study focuses on “Master Planning” of “Advanced Planning and Scheduling”, which is to determine a production and distribution plan of all demands on each node of the supply chain network. One multiple-goal “Mixed Integer Programming” model is proposed for this planning problem. The objective is to minimize the delay cost first, followed by minimizing the sum of the production(handling) cost, holding cost, transportation cost, and recycle penalty. Owing to the great influence of the recycle penalty on the total cost of the supply chain, recycle penalty and other variable costs should be taken into consideration simultaneously in planning demands to optimize the performance of the overall supply chain. First, MIP is used to solve the planning problem. However, when the problem gets more complex, the MIP model becomes unacceptable in time or unsolved due to the computing resource. Therefore, a heuristic algorithm is proposed to solve the “Mater Planning” problem for green supply chain. The algorithm consists of three steps: (1) Do the preliminary procedures, which include transforming the closed-loop supply chain into an open-loop supply chain prone to planning, transforming all nodes in the network into single-functioned nodes, and searching the sub-networks for each final product. (2) Determine the planning sequence of demands. (3) Plan demands sequentially until all demands are fulfilled. The demand-planning algorithm is named BRVA (Balancing Recycle-penalty and Variable-cost Algorithm), which focuses on two main steps including setting the cost for each link and adjusting the production planning to avoid recycle penalty. In BRVA, both recycle penalty and variable cost are considered to achieve the optimization of the overall supply chain. To show the effectiveness and efficiency of the algorithm, a prototype is constructed and scenario analysis is illustrated.