Optimal transit fare structure and frequency design with seat and standing capacity

Abstract

Public transit systems are of great significance to societies in terms of their social, economic, and environmental benefits. A well-designed public transit system can generate profits to the transit operators. Meanwhile, it has beneficial effects on traffic mobility, energy consumption, and environmental problems, especially for cities with high population density such as Hong Kong. The design and planning of public transit systems are formally considered in the Transit Network Design (TND) problem. Various TND models have been proposed with a diverse collection of behavioral assumptions, problem characteristics, and solution methods. Yet, effects of design of transit vehicle space allocation for seat capacity, consideration of seat availability, and passengers’ seat preferences are seldom addressed together in existing TND models. Moreover, limited effort has been put to investigate and evaluate different fare structures in the TND problem literature. This thesis develops a bi-level TND model for the design of most profitable fare, frequency, and seat space allocation of transit routes using a three-stage approach. At the first stage, an approach-based stochastic user equilibrium with elastic demand (SUEED) transit assignment problem is formulated and solved by the c-SRAM algorithm. At the second stage, a bi-level fare and frequency optimization model is proposed based on the approach-based SUEED assignment problem formulated at the first stage. A sensitivity analysis-based ascent search method is proposed to solve this model. At the third stage, the bi-level TND model is extended to further capture seat availability and multi-class passenger demand in the lower level transit assignment problem and to design the optimal seat space allocation in transit vehicles in the upper level problem. The solution methods proposed at the previous stages are adapted for solving the extended multi-class TND model. At each stage, model properties and effectiveness of solution methods are examined and illustrated by numerical examples. Based on the numerical results, three widely adopted fare structures (i.e., flat fare, distance-based fare, and sectional fare) are compared in terms of profitability. Insights of maximum feasible percentage of seat space in transit vehicles are also provided.