Optimal Design of Robust Feeder Bus Network to Rail Lines

Abstract

There is a growing concern with respect to public transport (PT) use and its inability to encourage people to switch from solo to shared means of travel. Designing an operational and economically public bus network is widely recognized as an efficient way of reducing mobility-related problems such as traffic congestion, parking overload, air and noise pollution, and road accidents. One of the major elements of PT network is the way buses can feed railway stations with the goal to approach seamless bus-rail transfers. By optimizing the routes of corresponding feeder-bus network services, better connections with urban rail transit are attained. The design problem for a set of feeder bus lines is analyzed by considering one high capacity railway station linked with a road network. This thesis is three-fold. Firstly, it documents and illustrates the survey results of feeder-bus service demand and consumer requirements, and exceptions for the feeder bus at Auckland, New Zealand, for selected rail stations. Secondly, using the survey results, an ant colony optimization algorithm is developed for an efficient design of feeder-bus network within an acceptable running time. A case study of a major rail station in Auckland demonstrates the effectiveness and performance measures of the proposed algorithm. Finally, a work shows how to efficiently design feeder bus lines to and from railway stations, in consideration of uses of online smartphone data. This design aims to generate better service than that provided by private cars or fixed-route bus service. A bi-level optimization model has been developed for the purposes of designing this integrated, smart PT feeder/shuttle service. The upper level problem is to minimize user, operator, and community costs, and the lower level problem is to maximize the number of rail passengers who will use the feeder bus service given travel behavior of railway users. A genetic algorithm is used to solve the model in an acceptable running time. A numerical small example and a real life case study in Chengdu City, China, are used to illustrate the properties of the bi-level problem and the performance of the proposed algorithm.