THESIS
2012
xiii, 124 p. : ill. ; 30 cm
Abstract
In face of the serious urban traffic congestion problem and consistent public rejection to congestion pricing, this dissertation proposes and studies a novel idea for traffic regulation – tradable credit scheme. With such a scheme, the social planner initially distributes a certain number of credits to all eligible travelers, charges a link-specific number of credits from travelers using that link, and allows for free trading of the credits among travelers. The performance of such tradable credit schemes in traffic regulation is investigated in general networks with both homogeneous users and heterogeneous users....[
Read more ]
In face of the serious urban traffic congestion problem and consistent public rejection to congestion pricing, this dissertation proposes and studies a novel idea for traffic regulation – tradable credit scheme. With such a scheme, the social planner initially distributes a certain number of credits to all eligible travelers, charges a link-specific number of credits from travelers using that link, and allows for free trading of the credits among travelers. The performance of such tradable credit schemes in traffic regulation is investigated in general networks with both homogeneous users and heterogeneous users.
In the homogeneous user case, we demonstrate that, given a credit charging and distribution scheme, a unique equilibrium link flow pattern with either fixed or elastic demand can be obtained by solving a standard traffic equilibrium model subject to a total credit consumption constraint, and the credit price at equilibrium of the trading market is unique under very mild conditions. As a result, through appropriate distribution of credits among travelers and selection of link-specific credit charging rates, the tradable credit scheme can obtain and sustain most desirable network flow patterns, such as the social optimum and Pareto-improving flow pattern generated by the various congestion pricing or congestion pricing cum revenue-refunding schemes proposed in the literature, in a revenue-neutral manner.
However, in the presence of user heterogeneity in terms of different VOTs, no equivalent convex program is identified to yield the user equilibrium (UE) and market equilibrium (ME) conditions from its KKT conditions. Therefore, the combined UE ad ME conditions are instead formulated into a VI problem with existence of solutions guaranteed. Sufficient conditions for unique aggregate UE link flows and credit price are provided. We also demonstrate how appropriate credit schemes can be set up to decentralize system optimal and Pareto-improving flow patterns.
Similar as setting an optimal congestion price scheme, to set an appropriate credit scheme for system optimal flow patterns, we need complete information of the demand function, the speed-flow relationship, and the generalized cost (or value of time). To overcome the difficulty of demand function calibration in reality, trial-and-error methods for implementation of tradable credit scheme on both a single road and general networks are proposed. For the single road case, a tradable credit scheme is gradually adjusted by a modified bisection method based on observed credit price, whereas for the general network case, the scheme is updated by successive average method based on observed UE link flows and credit price. Last but not least, this dissertation firstly points out the non-convergence problem of the bisection-based trial-and-error method for congestion pricing proposed by Li (2002), and suggests a modified version to address this problem.
Post a Comment