震災物流系統之決策模式

Abstract

自921集集大地震發生後，各界漸漸發現救援物資配送是救災工作中之重要一環，快速且具效率地將外界大量援助物資輸送入災區各需求點有助於提昇救災效率。震災物流系統主要包括物流據點配置、以及運輸路線規劃等二個部分，必須於災前妥善規劃震災物流系統與運作機制，同時在災後更需依照實際災損情形，馬上反應檢討調整區位與路線。為使規劃者能夠系統化且快速地進行物流系統規劃，非常需要一套輔助的決策分析工具。 本研究經由對震災物流系統運作機制之探討與設計，以數學規劃方法構建決策模式。其中據點區位選擇包括縣市政府負責運作之區域性大型轉運中心與各鄉鎮公所負責運作之地區性物資配送中心。在目前救災體系下，區位選擇由兩級地方政府各自決定，決策為互相影響之二階規劃，但理想的機制應為專責機構統一決策之一階規劃，本研究對這二種規劃模式及其效果進行探討與比較。另外據點配置與運輸路線規劃同時決策之問題規模龐大，求解效率不適合災後迅速應用，本研究將二部分分開依序進行決策分析，以期在合理的時間內獲得不錯的配置規劃結果。 本研究應用地理資訊系統與Lingo軟體建立資料庫並配合電腦演算，對921集集大地震災區進行實例分析，驗證決策模式之可行性與功能，獲得三項重要結論：一是決策模式之規劃結果在總運輸時間上，確實較當時實際作業方式為佳；二是若能統一決策權限，同時對區域性轉運中心與地區性配送中心之區位進行整體規劃，其總運輸時間會較當時分由縣市政府與鄉鎮公所各自決定區位之方式為佳；三是由於物資需求量大，大部分運輸車次為單點往返式配送，故可將物流據點區位選擇與車輛配送路線分開決策，以提高電腦演算效率。

It was found that logistics system of rescue materials is one of the important relieving tasks in the Chi-Chi great earthquake disaster area. To deliver the rescue materials from outside area into the demand points of disaster area efficiently can improve the efficiency of whole rescue tasks. There are two major elements in the relief distribution system（RDS）：the locations of collecting and distribution stations（CDS）, and the vehicles routing. The RDS and its operational mechanism should be drawn up before the disaster. Furthermore, based on the actual damage situations, the RDS must be adjusted as soon as possible. Therefore, a decision support system to design the RDS systematically and efficiently is very important. This research explored the operational mechanism of RDS in earthquake disaster area, and developed the decision model by mathematical programming. Two levels of CDS were considered in this study：the regional CDS operated by county government and the local CDS operated by township government. Two types of decision models were discussed in this research. One is a bi-level programming, in which two levels of CDS are distributed by decision makers individually, for the practical operational mechanism. Another is a single level programming, in which all of CDS are distributed by single decision maker simultaneously, for the ideal operational mechanism. On the other hand, the problem considering the locations of CDS and vehicles routing simultaneously is too large to solve efficiently. The decision model developed in this research analyzes the locations of CDS and vehicles routing separately by which the approximate optimum results can be recommended in a reasonable calculating time. In order to verify the operational feasibility and functions of decision model, a case study in the disaster area of Chi-Chi great earthquake was proceeded by the use of GIS and LINGO package. There were three major conclusions in this study. First, the distribution of CDS designed by developed model is better than the actual distribution of Chi-Chi case in total travel time. Second, single decision maker to distribute regional CDS and local CDS simultaneously is better than that of two decision makers in total travel time. Third, since the needs on rescue materials were large and most of vehicles shuttled between specified CDS and demand area, the locations of CDS and vehicles routing can be analyzed separately to improve the efficiency of model resolution.