Dynamic Congestion Pricing Considering Spatial Interactions of Queues

Road pricing has been studied extensively by transportation scientists and economists ever since Arthus Pigou first suggested the idea in 1920's. Vickery's landmark study using the bottleneck model advanced road pricing analysis to a dynamic regime where individuals' departure times are endogenized, hence the optimum tolls charged vary over time. Although the simple, elegant bottleneck model adopted by Vickery enables a wealth of policy/economic insights of the urban congestion problem to be obtained, its treatment of vehicular queues as point queues makes its applicability to general networks questionable. Specifically, the propagation of congestion is ignored in the bottleneck model. As such, congestion is localized within a link and the so-called queue spill-back is overlooked. With the help of a simple diverge, we find that allowing queue spill-back in flow dynamics could produce equilibrium solutions and optimal tolls far different from those predicted by the bottleneck model. We hypothesize that using a realistic dynamic flow model may significantly alter the policy implications and welfare effects of time-varying toll derived from the basic bottleneck model. In the proposed research, we plan to test our hypothesis by exploring dynamic congestion pricing on a wide variety of simple yet general networks with a more realistic traffic flow model. To achieve this objective, we will formulate, solve and analyze the dynamic congestion pricing problem via traffic equilibrium analysis and network optimization, and discuss the policy implications of our research findings.