Optimization Research On Feeder-bus Service In Response To The Urban Rail Transit Service Disruptions

As one of the main urban public transport modes in China, urban rail transit (URT)supports the urban mobility. However, due to some irresistible factors, URT service disruption occurs occasionally, which may break the URT normal operation,lead to serious passenger delay, and increase safety risk. When service disruption occurs, it could cause negative impacts such as a large number of passenger delay and social order chaos if no proper service disruption measures are applied timely. Routine bus system has the features of intensive distribution, flexible dispatch and large transport capacity,and thus it is an important way to respond to the URT service disruption by evacuating passengers quickly and guaranteeing consecutive transportation. Based on theories of transportation planning, operational research and intelligent optimization algorithm, the study investigates feeder-bus service design in an URT service disruption scenario. The study contains a systematic research of response mechanism, service demand, service plan and service system designing, aiming at providing theoretical and practical application basis for URT emergency organization and management in order to reduce risk, decrease passenger delay and save emergency costing. The main contents of the study are listed as the following four aspects:(1) The study investigates the feeder-bus evacuation mechanism in response to the URT service disruptions. Evacuation mode, evacuation classification, service scope of feeder-bus, and evacuation demand are investigated to provide theory frame foundation for the formulation of specific service plan. According to the location and influence scope, the study divides the service disruption into two types: single-line service disruption and network service disruption. Combined with the feature that URT can operate between sections, the study presents an emergency evacuation mode of URT part route supplemented by routine bus connection to make full use of the transport capacity of uninfluenced URT and reduce the influence of service disruption, which means the URT in uninfluenced section will operation normally and make sure the normal trips of passenger by connecting bus system. For different service disruption types, the feeder-bus functions are refined: as in single-line service disruption, the influence scope is small and the passenger flow is simple, the feeder-bus can replace the influenced URT to maintain the transportation function; for the network service disruption, due to the large scope of influence and complex passenger flow, the feeder-bus should supplement the connection capacity and enlarge the evacuation region to the whole URT network, instead of replacing the connection function only. The evacuation demand is classified into static demand and dynamic demand based on the occurrence time of service disruption,and different static passenger flow situations due to different locations of train when the service disruption occurs are deeply investigated;contents and problems to be solved that should be contained in the feeder-bus service plan are listed for further mathematic modeling.(2) The optimization of the feeder-bus dispatch plan in response to single-line service disruption in URT. In single-line service disruption, the feeder-bus will replace the connection role only and its line direction as well as stop stations will be arranged as the URT line. Therefore, the feeder-bus dispatch plan need to solve the problems of bus dispatch and operating routes, including the parking station that the feeder-buses are dispatched, the number of buses dispatched from each parking station to each connection station, the number of round trips of each feeder-bus dispatched, and the connection stations that the bus will return to its original parking station.The study builds a feeder-bus dispatch plan optimization model in response to single line URT service disruption. A method based on equivalent parking station principle is designed to solve the model, and the original nonlinear programming model is transformed to an easy-solved transportation problem model. By using Nanjing subway as an example, the feasibility of the model and algorithm is verified and a sensitivity analysis is conducted on the model parameters. Results show that the total evacuation cost and the number of feeder-buses in need decrease with the increase of specified rescue time. Besides, no significant correlation is observed between the number of dispatched buses and the number of parking stations, but the demand distribution of both ends of the part route obviously influences the feeder-bus dispatch plan. The findings can provide guidance for the resource allocation of feeder-bus and improve the utilization efficiency of public transportation resource in emergency evacuation.(3) The optimization of the feeder-bus route design in response to URT network service disruption. In network service disruption, the service scope of feeder-bus expands to the entire URT network. Thus, both replacement and supplement feeder-bus routes are needed. The replacement feeder-bus route can be determined by the influence scope of the service disruption, but the supplement feeder-bus route need to be further confinned according to the demand. Therefore, the number of feeder-bus routes which are needed for supplement, the direction of each supplement feeder-bus route as well as the choosing of parking stations become principal problems to be solved in making network evacuation plan.From the viewpoint of ensuring the travel time of influenced passengers, the study confirms the supplement feeder-bus route set gradually by the way of "station"determining "line". Determining "station": by calculating and comparing the shortest path travel time of each URT station pair in normal URT network, URT network under service disruption, URT network under service disruption with supplement feeder-bus routes as well as the real road network, the station pair can be classified into completely interrupted station pair, partially influenced station pair and uninfluenced station pair;when the shortest path travel time of partially influenced and uninfluenced station pairs in real road network reaches the minimum, the station pair set of feeder-bus route in need for supplement will be obtained, i.e. the supplement feeder-bus station set.Determining "line": the study uses the minimum supplement feeder-bus route network length as the objective and builds the supplement feeder-bus route line optimization model. Based on the Shortest Path theory and the correlation between required station and shortest path, an optimization algorithm for candidate route combination and recombination is designed, and the feasibility of the model and algorithm is verified by example analysis.(4) The optimization of the feeder-bus dispatch plan in response to URT network service disruption. The feeder-bus tasks can be divided into suspended passenger evacuation task and consecutive transportation guarantee task. For static passenger flow, due to the large amount and long delay time, the feeder-bus should complete the evacuation as soon as possible. For the dynamic passenger flow, which is regular and stable, the feeder-bus can operate according to time table or certain departure interval. Thus, in order to satisfy different connection task demand, the study presents a two-stage dynamic emergency dispatch mode for feeder-bus.The first stage "to complete the passenger evacuation": the feeder-bus dispatch mode in this stage is to keep feeder-bus continuous operation; using the minimum passenger delay as objective, the study builds the first stage dynamic dispatch optimization model for feeder-bus and designs genetic algorithm based on multilayer coding to solve the model. The second stage "to guarantee consecutive transportation":the feeder-bus dispatch mode in this stage is to operate according to the given time table;using the minimum total connection time of feeder-bus as objective, the study builds the second stage feeder-bus dynamic dispatch optimization model and designs an improved simulated annealing algorithm to solve the model. Besides, by setting state end point in the first stage model and setting virtual parking station in the second stage model, the connection and interaction of the two stage models are realized to obtain the end time of the first stage and the returning number of feeder-bus. Finally, the feasibility of the two stages models and algorithms is verified by example analysis.(5) The system development for feeder-bus service plan design in response to URT service disruption. In order to realize a quick emergency response, the study designs a feeder-bus emergency service planning system in response to URT service disruption based on the above-mentioned research. The combination of service plan self-establishment with man-machine-aided generalization improves the emergency planning efficiency greatly.