Investigating Travel Time Reliability Measures in Toll Design Problem

Abstract

Congestion pricing has been regarded as an effective method of reducing networkwide travel cost. Previous work on the toll-design problem focused on the deterministic network, assuming that the network operates deterministically without uncertainty. This essentially precludes investigating the recently proposed travel time reliability indices as a system performance measure to evaluate and prioritize the selection of future investment strategies. In this study, we investigate various recently proposed travel time reliability measures in determining the optimal toll strategies to improve travel time reliability in the network. A simulation-based genetic algorithm (SGA) procedure is developed to solve the stochastic bi-level programming problem. Our experimental results using a simple test network show that optimal toll determined by the misery index can improve both travel time reliability for network users and total travel time for the network planner.