A Methodology for Bus Network Design

This thesis develops a methodology for solving the bus network design problem (BNDP) in which one seeks to determine the best configuration of bus routes and their frequencies, given information on the bus transit demand, the street network, the available resources and the operational constraints. The proposed methodology follows a heuristic approach and consists of two main components route generation (RG) and vehicle allocation (VA) that are implemented in a single automated design procedure. The proposed methodology addresses the principal shortcomings of previous heuristic approaches to the BNDP and is characterized by its ability to (1) incorporate the fleet size constraint within the route generation process, (2) identify the major trip patterns in the demand matrix for guidance in the route configuration, and (3) solve the BNDP using a general network design or a transit center network design. The RG procedure is based on the concept of route skeletons which consist of three nodes two termini and one intermediate node that are expanded into full routes by means of nodal insertions. RG creates one route at a time and estimates the number of buses required on the network after each route is generated. If the vehicle estimate is less than the available fleet size, an additional route is generated and the network is reevaluated; otherwise, VA is initiated. VA calculates the minimum number of buses required on each route so as to satisfy the loading feasibility and passenger assignment constraints. Surplus buses remaining after the minimum vehicle requirement is determined may be allocated on the network to improve its overall performance, or used by RG to create additional routes. The proposed methodology and automated design procedure are applied to a case study in San Juan, Puerto Rico where a major restructuring of the bus system using a transit center concept has been proposed. Solutions produced by the proposed methodology are compared with the proposed transit center route plan. Thesis Supervisor: Dr. Nigel H. M. Wilson Title: Professor of Civil and Environmental Engineering