鐵路列車連鎖延滯之模擬模式構建與應用

Abstract

鐵路系統容量不足往往是導致列車延滯之主因，尤其在初始延滯(first delay)發生後，後續所引發的連鎖延滯(knock-on delay)擴散效應對列車正常營運之影響甚大，因此欲確保系統之可靠度及服務品質，快速有效地降低連鎖延滯一直是重要課題。為有效釐清造成連鎖延滯之複雜因素及其交互作用，本研究依據臺鐵系統之運轉特性構建一模擬模式用以推估連鎖延滯，並以臺鐵西部幹線北部路段為案例，分析連鎖延滯之各種相關問題。 本研究經蒐集臺鐵七堵─樹林路段之營運資料驗證模式的合理性後，續針對不同初始延滯程度衍生之連鎖延滯擴散，及不同運轉調度策略(timetable recovery strategy)對連鎖延滯降低進行系統分析，初步發現連鎖延滯有往下游路段擴散，並呈非線性遞增之趨勢，且各種運轉調度策略對連鎖延滯均有明顯降低效果等現象。故本研究進而分別針對列車密度、初始延滯及運轉調度策略對連鎖延滯之影響及其整體交互作用影響進行分析，並以迴歸分析方法構建列車連鎖延滯與該三項因子間之指數關係函數，以作為估算連鎖延滯的簡便工具。另為利了解初始延滯發生區位、持續時間及運轉調度策略對連鎖延滯之影響，本研究亦進行其關聯分析。由研究結果顯示，若初始延滯發生之區位愈靠近上游路段，則連鎖延滯亦將愈大，而運轉調度策略對於降低連鎖延滯之效果也愈佳。 此外，鑑於影響連鎖延滯之部分關鍵因素具有隨機特性，故為確實反映臺鐵列車實際運轉因初始延滯發生後所產生之連鎖延滯特性，及深入探討產生連鎖延滯之班表穩定度問題，本研究更進一步分析站間運轉時間及停站時間之隨機特性，除彙整七堵─新竹路段對號列車及通勤電聯車之站間運轉時間資料，另依尖、離峰時段分開彙整停站時間之隨機資料，並將其納入模擬模式推估連鎖延滯。研究結果顯示，無論就所有車站或二端末車站(七堵及新竹站)而言，其尖峰或離峰時段之連鎖延滯模擬結果皆會接近一定值。本研究所提出之分析架構及內容，除可有效釐清連鎖延滯關鍵影響因素及程度，所構建之模擬模式更可作為相關改善策略之分析工具。研究成果可作為後續營運單位排班規劃、系統可靠度分析及服務品質改善之參考。

Train delays of a railway system are affected by many factors and one of the most important factors is insufficient line capacity. Once a first delay occurs, the delay propagation (i.e., knock-on delay) always interrupts train operation. Thus, how to promptly reduce the knock-on delays becomes an important issue for providing reliable timetable and high quality of service. In order to clarify the impacts of these complicated factors and their interactions on knock-on delay, this research develops a comprehensive simulation model to estimate knock-on delays, and a rail section from northern area of Taiwan railway system is selected for case study. This research first collects real data from the section of Cidu to Shulin of Taiwan railway system to verify and validate the model. The impacts on knock-on delays of different first delays and timetable recovery strategies are then evaluated. The results show that knock-on delay propagates toward downstream sections when a first delay occurs, and the knock-on delay increases nonlinearly toward downstream sections. In addition, timetable recovery strategies are demonstrated to have significant impacts on the reduction of knock-on delays. Based on the simulation results of the case study, regression analyses are employed to calibrate the relationships between knock-on delays and three key factors, which are train density, first delay and timetable recovery strategy. The regression models indicate that the relationship between knock-on delay and these three key factors conforms to an exponential function. This research also explores the effects on knock-on delays of different first delay locations and recovery strategies. The main findings are as follows: (1) the closer that the first delay occurs at upstream section, the greater the knock-on delays at all stations and two end stations are; (2) the effects of timetable recovery strategies are better in recovering to scheduled timetable when the first delay occurs at upstream section. To clarify the stochastic nature of the key factors affecting knock-on delays and to evaluate timetable stability, this research also collects the running time and dwell time data of Cidu-Hsinchu section for express and commuter trains respectively for further analysis. The result shows that the knock-on delays at all stations and two end stations during peak and off-peak hours converge to constant values respectively. In summary, this research proposes a framework and a simulation model which can be applied to analyze the impacts on knock-on delay by all kinds of changes in infrastructures, operational situations and controlling strategies. It is expected that the results can be beneficial to timetable scheduling, system reliability analysis and service quality improvement.