台灣高鐵列車組車隊需求分析

Abstract

列車組為鐵路營運機構主要核心系統及生產設備之一，投資成本非常龐大，其採購費用動輒數十億元，除了具有沉沒成本（Sunk Cost）的特性之外，回收年期亦相當長；另外，列車採購從下訂到交車的過程需要很長的前置時間（Lead Time），因此，如何根據營運需求、擬訂列車服務計畫、時刻表，進而推估列車組的需求規模以作為採購決策的依據，為鐵路營運機構所面臨的重要課題。 臺灣高鐵自2007年通車營運以來，從營運初期每週133班次，歷經多次增班，目前（2014年）每週營運班次數已高達954班次。隨著營運班次的成長，列車調度捉襟見肘。展望未來將新增南港、苗栗、彰化、雲林等四站，旅次量及列車往返時間亦將逐漸增加，故須及早進行列車需求評估供採購決策，除了避免惡化旅客服務水準之外，亦可對於整體之財務調度預作規劃。 因此，本研究以臺灣高鐵為實證研究對象，從上游之旅運需求預測、列車服務計畫、列車排班，以及維修計畫，並運用「臺灣高鐵營運決策輔助系統」，針對各種發展情境，進行車隊規模的推估。研究成果可供臺灣高鐵公司採購列車的參考，在兼顧服務品質以及財務負擔之下，作出最佳的決策。 本研究對未來類似列車組車隊研究提出兩項建議，(1)除列車組需求評估外，尚包括財務分析、車型選擇、車隊（含維修備料）的汰換更新…等因素均應綜合考量。(2)由於未來的不確定性甚高，因此長年期的營運規劃應在各分析程序裡考慮敏感度分析。

The trainset is one of the most expensive components in railway operations . The investment is substantial and the cost is usually over 10 billion NTdollars per trainset. It also has the characteristics of a sunk cost and long payback period. Additionally, trainset procurement usually has a long lead time from ordering to delivery. Therefore, while making decisions on trainset procurement it is an important issue for railway operators to estimate their fleet size based on operational needs, and future train service plans. Taiwan High Speed Rail (THSR) has inaugurated to commercial operations in 2007. At the beginning the train service was 133 trains per week and has increased to 954 trains per week at present (2014). As operation services grew, train dispatching has become increasingly difficult. Moreover, four new stations coming online, Nangang, Miaoli, Changhua and Yunlin, will gradually increase the ridership as well as the round trip time. As a result, it is necessary to forecast the future fleet size in advance for a trainset procurement decision to prevent service quality to deteriorate and to arrange a long term financial plan. This research has proposed two suggestions for future study. (1) Besides trainset requirement assessment, financial analysis, trainset modeling, equipment replacement and so on should be considered together. (2) Due to future uncertainty, long term operation planning should consider sensitive analysis.