Train Dispatching Models Towards Reducing Train Delays And Inequity Of Competitors

Providing punctual, efficient and reliable services is a main goal of rail industries in order to maintain and further improve their competitive advantages in the rapidly changing multimodal transportation market. The research on railway transportation management has attracted great attention worldwide. With the development of economy and the construction of high-speed railway, the quality of people’s living is improved and the requirements of train punctuality and reliability are more and more strict.Due to the limited capacity and unavoidable stochastic factors, perturbations may influence scheduled segment running and station dwelling times, thus causing primary delays to normal train operations. Primary delays may further result in snowball effect on other trains with consecutive delays. The main task of train dispatching (conducted by train dispatchers) is to recover impacted schedules from perturbations. Ineffective train dispatching may significantly downgrade the quality, efficiency, punctuality and reliability of train services and the overall system performance.Furthermore, with the rapid development of market economy and the trend to separate government functions from enterprise management, the research considering competitive interaction, related to equity and inequity, is essential.This thesis addresses the problem of dispatching trains with consideration of delay equity among multiple competing operators, in addition to minimize total delay time. According to different purposes, the operators can be viewed as trains, companies or passengers. Note that the company represents Train Operating Companies (TOCs) for foreign (European) countries and Railway Bureaus (RBs) for China. Equity is defined as the degree of homogeneity of the delay cost faced by different trains, companies or passengers. An inequitable situation is some trains (companies or passengers) face much more delay costs than other trains (companies or passengers).The main work in this thesis could be summarized as follows.(1) By analyzing the rail operation and the reforming trend of Chinese railway, and introducing the separation between infrastructure management and transport operation in foreign (European) countries, the necessity and significance of research on train dispatching problem with consideration of equity is presented.(2) In order to give a more comprehensive view on related studies, the literature review section is divided into three subsections:1) general works about online railway traffic control, where delay equity is not explicitly considered;2) studies dealing with fairness in resource allocation in the train line planning stage and3) works which study equitable resource allocation and online traffic control in air transport field.(3) The related terms about train dispatching problem are analyzed, including its effects, goals, basic dispatching measures, and the difference from train line planning problem. The characteristics of perturbations (i.e. causes and classification) are also studied.(4) Based on network flow theory and cumulative flow variable theory, a model to describe infrastructures on railway network is developed. Three optimization models are further presented, which generate dispatching plans for trains with/without consideration of equity, and considering it as soft constraint (in objective function) or hard constraint (in constraints). The model considering equity as a soft constraint is denoted as model (P1). In this way, equity can be represented as the minimization of the maximum deviation of competitors’ delay cost. An additional model (P2) formalizes delay equity through hard constraints where some fixed amount of delays discounted from the deviation which is associated to different competing operators are allowed. In order to provide a benchmark, we further propose a model (P3) in which delay equity is neglected.(5) Computational experiments are first performed on an artificial dataset, in order to test the performance of the proposed model. Furthermore, the assessments based on realistic datasets from China and the Netherlands are made by comparing different models (P1, P2and P3), using two scheduling algorithms (FIFO rule and original timetable order) and setting different parameters (weights, train delay cost per unit time and so on). An extensive number of experiments studies sensitivity of the solutions to a variety of factors, such as:1) differences in generated schedules with/without equitable requirements;2) differences in generated schedules with FIFO rule and original timetable order;3) sensitivity to the delays distributions suffered by trains as primary delays;4) impact of the relative weights between costs of train delays and equity of competitors;5) impact of considering equity for single trains, or considering only aggregated performance per competitor;6) impact of the "equity" of the original timetable, i.e. the degree by which some train paths are systematically more prone to face delays;7) differences in solutions with/without rerouting measure and finally,8) impact of the train delay cost per time unit.