Timetable Coordination For Last Trains On Urban Rail Networks With Maximal Accessibility

In recent years,China has entered a rapid development period of urban rail transit.Urban rail networks have been built and subsequently expanded in many cities in China.Coordination of train timetable always plays a vital role in railway operation and service provision.With the increase of passenger demand at late night,the last train timetable coordination on urban rail networks has been attracted great attentions of rail operators.This dissertation studies on the last-train timetable coordination from the view of network accessibility.The network accessibility is defined with the consideration of both transfer process and the whole process between origins and destination(OD).A computation method is designed to evaluate the network accessibility.And then,the last train timetable is synchronized at given departure time with the objective of maximal transfer connection and OD accessibility,respectively.Two strategies "dwell time extension" and "OD accessibility-oriented optimization" are proposed to improve the network accessibility.Further,the service closure times of different lines are coordinated based on the departure time optimization of last trains by taking account of both network accessibility and operational cost.The main contents and conclusions of the dissertation are shown as follows:(1)The network accessibility is consisted of two terms transfer accessibility and OD accessibility.Transfer accessibility indicates whether passengers on one line can transfer to the trains on another lines.The transfer accessibility is measured by transfer redundant time.It can be concluded that the mutual directions between two last trains can be both connected if and only if the dwell times of the two last trains are both longer than the total of arrival time difference and transfer walking time of passengers.OD accessibility indicates whether passengers can arrive in their destinations from their origins.A label setting algorithm is designed to calculate the OD accessibility based on the notion of the classic Dijkstra's algorithm.The method is applied in a real-world metro network in Shenzhen.The results show that the method is a powerful tool in solving the OD accessibility problem.It has the ability to allow passengers to plan an optimal itinerary and to allow operators to analyze the network accessibility.Comparison analysis indicates that the label setting algorithm can provide exact solutions in much shorter time,compared with a path enumeration method.Extensive tests on a set of random networks indicate that the method is efficient enough in practical applications.The execution time for an O-D pair on a personal computer with 2.8GHZ CPU and 4GB of RAM is only 1.2 seconds for urban rail transit networks with 100 transfer stations.(2)A service-oriented last train timetable synchronization model is proposed to improve the successful transfer connections.The model takes the inter-station running times and dwell times at station of last trains as decision variables.The model does not set the upper bound of dwell times at transfer station but minimizes the maximal dwell time and the total travel time to reduce the dwell times.Moreover,the difference of transfer walking times is considered and characterized by the logarithm normal distribution.The model is formulated into a mixed integer linear structure.And it is solved by the commercial optimization solver CPLEX.Lastly,the model is verified by the Shenzhen metro network.The results indicate that the model is effective to enhance the successful transfer connections between last trains.The number of successful connections is improved to 49 from 40.And 4 pairs of mutual directions are both connected.Passenger travel times will not be increased after optimization though the dwell times of last trains at some transfer station are extended.The extended dwell times at transfer stations is effective to enhance the transfer accessibility.By taking transfer walking time distribution into account,the model can further increase passengers with successful transfer and thus the network accessibility.(3)A demand-oriented last train timetable synchronization model is developed to improve the OD accessibility.The objective of the model is to maximize the number of accessible OD pairs instead of the successful transfer connections.The model is formulated by a mixed integer linear model.To solve for large-scale networks,a genetic algorithm combining with the label setting algorithm is designed.A numerical case is conducted to make a deep comparison between the OD accessibility model and transfer accessibility model.The Shenzhen metro network is applied to evaluate the proposed model and solution methodology in practice.The results indicate the proposed model enhances the OD accessibility significantly.The number of accessible OD pair is increased by 12.9%and the number of accessible passenger is increased by 16.6%.The performance of the proposed GA is satisfactory in both solution quality and computation time.Comparison analysis shows that the proposed model significantly outperforms the transfer model in OD accessibility optimization if the dwell times are not extended.And if the dwell times at transfer stations can be extended,the model still has an advantage in the optimization of OD accessibility though the model difference is becoming lower.Therefore,the OD accessibility optimization method is more effective to enhance the network accessibility.(4)This dissertation lastly focuses on the service closure times of different rail lines.A bi-level bi-objective model is constructed with the trade of network accessibility and operation cost based on OD demand.For the upper model,given the earliest and latest departure times of last trains as time horizon,it takes the number of train services in the time horizon as decision variables with the fixed service headway.The objectives of the upper model are the maximal number of passenger reaching their destination stations and the minimal total running distance of trains on all lines.For the lower model,it takes as the running times and dwell times of trains in the time horizon as decision variables.The objective of the lower model is the transfer accessibility of the trains.The model is solved by ?-constraint method to obtain the pareto solutions.The bi-level model is then verified by the Shenzhen metro network.The results indicate the proposed model is effective to balance the network accessibility and the operational cost.From the pareto solutions,the running distance of trains can be reduced at the given number of passengers with successful trip.The number of passengers reaching their destination stations can also be increased at the given train running distance.From the pareto front,a better solution is obtained with lower running distance and more passengers with successful trip,compared with the original solution.Based on the pareto set,the optimal order of last trains' departure can be obtained.The departure order need to be determined by both passenger demand and transfer coordination.