賑災物資配送系統之最適規劃

Abstract

1999年9月21日凌晨1點47分，台灣地區發生規模7.3的大地震，造成2246人死亡、8735人受傷。由於地震所引起的災害除了來自其本身直接造成的「一次災害」外，尚可能因人為的處理不當而有「二次災害」、「三次災害」。因此，災後如何迅速採取適當地回應措施以減低大型震災所可能導致的損害，即成為一項重要的課題。「防災」、「救災」以及「重建」是災變防救體系的三大系統，救災的部分係指地震發生以後採取的緊急措施，緊急物資的配送又為救災系統中重要的一環。如何有效地分配與調度物資以避免浪費或堆置無用，將是決定此配送系統優劣的關鍵。 賑災物資配送系統屬於區位－分派問題，但除了追求經濟上的效率外，「公平」亦須納入模式的目標。本研究從最佳化的觀點，以多目標規劃構建賑災物流系統之配送模式，考慮的目標有總成本最小、總旅行時間最短以及最大化最小滿意度，並設定物資轉運站不提供倉儲的功能。本研究設計一研究範圍包含中部四個縣市的簡化個案，求解後發現最佳的物資轉運站應設置於豐原、南投兩處，顯示此次921災變中政府僅設置一大型物資轉運站於南投縣立體育場，並不是最好的選擇，應根據災情的位置分布，選取適當的物資轉運站負責之，不一定只集中在同一地點再進行調度。 最後，本研究認為在賑災物流系統中，如何取得各項參數亦為一項重要的課題，此外並提出三項建議：（1）為提高模式的實用價值，需與決策支援系統做整合；（2）政策上應化被動為主動，事先在各地選取適當的避難位置、物資供給以及轉運站候選點；（3）可引進基因演算法來解決實務上大規模的問題，並透過其平行搜尋的功能，以增加演算效率。

Taiwan had occurred powerful earthquake which measuring 7.3 on the Richter scale at 1:47 a.m. on September 21, 1999. This quake caused the death toll of 2,246 and 8,735 people injured. In addition to the damage caused straightly by earthquake, taking improper action would cause other disaster. Hence, it becomes an important issue that how to react in order to reduce earthquake loss. Disaster prevention, protection, and reconstruction are the major topics of the system for hazard mitigation. Protection means the emergency program after quake, and how to distribute relief efficiency and effect is one of the key points. The model of the relief distribution system is a location-allocation problem. In order to seeking the efficiency in economic, “fairness” also has to be included in it. From the optimization viewpoint, the thesis constructs a relief distribution model using multiobjective programming methodology. There are three objectives in the model – minimize the total cost, total travel time, and maximize the minimal satisfaction during the planning period. Besides, the model also assumes that the transfer center doesn’t offer the service to store any relief. According to the information about 921 earthquake, this thesis designs a case and the solution suggests that we should locate two transfer centers in “Fang-Yuan” and “Nan-Tou”. It means that the government just located one in Nan-Tou at 921 earthquake is not the best choice. Choosing the location of transfer center should accord to distribution of all locations need help, and maybe not dispense the relief after centering at one place. Finally, how to get the data to input the model is also an important issue considered in the thesis, but needs future researches to discuss. Also presenting three suggestions: first, in order to increasing the utility, the model has to integrate with decision support system (DSS); second, government should be proactive in choosing the proper places refuge for victims, gather relief and transfer center candidates; third, it could introduce the genetic algorithm to increase the efficiency in calculation and solve large-scale problems in practical.