Dynamic System Optimum: A Projection-based Framework for Macroscopic Traffic Models

TL;DR

This paper introduces a projection-based framework for achieving dynamic system optimum in regional traffic networks, improving solution quality over classical methods by leveraging macroscopic traffic models and gradient-based optimization.

Contribution

It presents a novel projection gradient-based method for emulating dynamic system optimum in macroscopic traffic models, avoiding marginal travel time approximations.

Findings

The framework effectively improves solutions compared to classical methods.

Application on 8-region networks demonstrates its practical utility.

The approach enhances traffic management strategies in regional networks.

Abstract

This paper proposes the theoretical grounds for emulating the Dynamic System Optimum with desired arrival times on regional networks, using aggregated traffic dynamics based on the Macroscopic Fundamental Diagram. We used a projection gradient-based solution method that avoids the need to compute approximations of marginal travel times. We demonstrate the application of our framework on 8-region networks and show that our approach yields improved solutions compared to classical approximation methods in the literature, such as the Method of Successive Averages and the gap-based approach.