Dynamic Traffic Assignment with Rolling Horizon Implementation

Abstract

This paper discusses an analytical dynamic traffic assignment (DTA) model with the extended capability of performing rolling horizon implementation. This model is formulated as a link-based variational inequality and can be solved efficiently to convergence by a relaxation (diagonalization) algorithm. The rolling horizon mechanism makes this proposed approach a flexible tool in evaluating simulation-based DTA models. According to different assumptions of travelers’ route choice behavior, three different classes of travelers are categorized: fixed (predetermined) route class, stochastic dynamic user-optimal class, and dynamic user-optimal class. Model formulation, steps of the algorithm, and rolling horizon implementation are discussed in detail. Computational examples and results are presented and analyzed.