Multiclass Dynamic Traffic Assignment Model: Formulation and Computational Experiences

Abstract

To model the impact of advanced transportation management and information systems, especially route guidance and other traffic information systems, it is often necessary to develop traffic models that acknowledge the existence of different user classes and that acknowledge that each class may respond to traffic information differently. As a minimum the model should differentiate vehicles that receive real-time traffic information from those that do not. Most existing approaches that incorporate multiple user classes are based on simulation. Although there are advantages of doing so, their main disadvantage is lack of solution quality and property. Recognizing the importance of these, an analytic approach in which the solution properties and quality are well defined is developed. The users are divided into three classes: (a) fixed-route travelers, (b) stochastic dynamic user-optimal, and (c) dynamic user-optimal. The approach defines the property of each class and integrates each into one modeling framework through a variational inequality formulation. Subsequently, a solution algorithm for the formulation is provided and the results of the algorithm are verified through six scenarios.