On the reproducibility of spatiotemporal traffic dynamics with microscopic traffic models

TL;DR

This paper evaluates how well microscopic traffic models can replicate spontaneous traffic breakdowns and their spatiotemporal patterns using empirical data, highlighting the influence of boundary conditions on model accuracy.

Contribution

It systematically compares microscopic traffic models against real-world data to assess their ability to reproduce traffic breakdowns and explores the impact of boundary conditions on these phenomena.

Findings

Models can reproduce traffic breakdowns under certain conditions.

Boundary setup significantly influences the emergence of traffic congestion in models.

Model assessment methods help identify key factors affecting traffic dynamics.

Abstract

Traffic flow is a very prominent example of a driven non-equilibrium system. A characteristic phenomenon of traffic dynamics is the spontaneous and abrupt drop of the average velocity on a stretch of road leading to congestion. Such a traffic breakdown corresponds to a boundary-induced phase transition from free flow to congested traffic. In this paper, we study the ability of selected microscopic traffic models to reproduce a traffic breakdown, and we investigate its spatiotemporal dynamics. For our analysis, we use empirical traffic data from stationary loop detectors on a German Autobahn showing a spontaneous breakdown. We then present several methods to assess the results and compare the models with each other. In addition, we will also discuss some important modeling aspects and their impact on the resulting spatiotemporal pattern. The investigation of different downstream boundary conditions, for example, shows that the physical origin of the traffic breakdown may be artificially induced by the setup of the boundaries.