Optimizing Last Train Timetable Synchronization For Urban Railway Transits

Since the 1970s, with the increasingly heavy problem of energy crisis, urbanization issued resources shortage, traffic congestion, traffic accidents and environment, urban railway transits have attracted much attention in both developed and developing countries. Due to the rapid expansion of urban railway transit networks, the level of urban traffic supply has been improved; the traffic congestion problem has been released; the urban territory has been expanded; and the social benefit as well as the city competitiveness has also been increased.Urban railway transit is not only the main body of city trips but also an important support of city operations.In recent years, urban railway transit systems in metropolises have become complex networks. Therefore, it is increasingly difficult to manage the operations of urban railway transit under concentrated networking and multi-mode of the system.This thesis focuses on the urban railway transit networks and transfer stations, involving analyses the coordination mechanism of the last trains as well as impact elements of passenger transfer and optimization of the last train timetable synchronization on the basis of the current service quality in the Beijing subway. Concerned with different situations, different models and algorithms were developed to address the problem of last train timetable synchronization, which could support theoretical methods for last train operations. The Beijing subway network was chosen as a case study to illustrate the high efficiency of the proposed models. The main contents of this thesis are listed as follows.First, coordination situations between the last feeder train and the last connecting train were shown. Passengers can transfer successfully when the arrival-time difference between the last connecting train and the last feeder train is greater than or equal to the dwell time at the transfer station. Otherwise, passengers fail to transfer to the last trains. Next, a general last-train timetabling model was proposed, which was further proved to be NP-Hard with NAE-3SAT selected as the reduced problem.Second, after analyzing the operating characteristics of the first train, the last train and ordinary trains comprehensively, an integrated model of minimizing passenger transfer waiting time was established. According to different operating periods, three algorithms to calculate transfer waiting times were developed. Taking the Beijing subway network in year of 2013 as a case study, the results indicate that the first train transfer waiting time was reduced to 21.5% of the original value and five "Just-meet" directions were improved; the last train transfer waiting time was reduced 67.7% (the number of missing directions reduced from 16 to 14) and two "Just-meet" directions were improved; and the ordinary-train transfer waiting time was reduced 63.3% as well as seven "Just-meet" directions were improved.Third, the last-train transfer redundant time was defined, and the relations between the redundant time and waiting time were analyzed. As a result, a last train transfer redundant time model for subway networks was proposed. In addition, a genetic algorithm was developed to solve the model. To verify the effectiveness of the model, a sample network was designed and tested. The optimization results indicate that the model and the algorithm are both applicable and effective. Moreover, a case study for the Beijing subway network was considered. Among the 42 selected key transfer directions, the coordination level is improved by 11.9%, which totally eliminates the influence of just misses.Fourth, a last train rescheduling model for delay in subway networks was proposed, and a genetic-simulated annealing algorithm was developed. We proposed the last train rescheduling principle, i.e., to eliminate transfer influence activated by train delays quickly and economically. Taking the Beijing subway network as a case study, the results indicate that the proposed model and algorithm have high practicability and efficiency. In addition, sectional redundant delay times of the up-train direction of line #1 were calculated, showing that the redundant delay time of section "D1-1 to Gongzhufen station" is 15 min 11 sec. The most sensitive section, which has the minimum redundant delay time, is "Dongdan station to Jianguomen station" and can tolerate a delay of 7 min 46 sec so that transfers will not be affected.