Analysis of Critical Scenarios of Congestion Emergence and Propagation in Urban Environments

LICIT laboratory (ENTPE / UGE), Lyon; Lyon; Sur place; EUR 40 000 - 60 000; Plein temps

Résumé du poste

A research laboratory in urban transport is offering a PhD position in Lyon, focusing on the identification of key determinants of congestion in urban environments. The selected candidate will utilize machine learning methods and traffic theory to analyze congestion dynamics and develop predictive models. This position is supported by public funding and aims to enhance urban mobility management.

Qualifications

• Strong understanding of transport systems and urban mobility.
• Experience with machine learning methods is a plus.
• Knowledge of statistical physics and traffic theory is beneficial.

Responsabilités

• Identify key determinants of congestion in urban environments.
• Analyze congestion dynamics at mesoscopic and macroscopic scales.
• Develop models for predicting traffic patterns and system failures.

Formation

Doctorate or equivalent in a relevant field

Description du poste

Topic description

Context

Optimizing transport systems is a key lever for the ecological transition of cities. To reduce their environmental footprint, cities must offer a high-performance multimodal mobility system that ensures safe, rapid, equitable, and low-carbon access to essential services and infrastructures. However, urban congestion represents a major obstacle to these objectives. It generates significant economic, health, and environmental costs, while greatly contributing to air pollution and greenhouse gas emissions.

While the mechanisms of congestion formation and propagation on linear infrastructures (highways, urban corridors) have been extensively studied, the urban context presents specific features that considerably complicate the analysis of this phenomenon. Indeed, the density and heterogeneity of networks, the diversity of demand, and the interdependence with control systems make congestion dynamics particularly difficult to apprehend, even though their effects are critical.

Subject description

In this context, LICIT-ECO7 — a joint research unit of ENTPE and Gustave Eiffel University, specialized in transport and energy systems — is offering a PhD position focusing on the identification of key determinants of congestion in urban environments, as well as a detailed understanding of its propagation mechanisms at mesoscopic and macroscopic scales.

Recent research has highlighted the existence of recurrent spatio-temporal patterns in congestion formation. At the macroscopic scale, the MFD (Macroscopic Fundamental Diagram) constitutes a tool for aggregated representation of traffic dynamics within an urban area. By linking the average traffic density to its average flow, it enables characterization of the overall operational state of the network. At a mesoscopic scale, recent studies have aimed to characterize the stability of tree-like congestion structures (or jam trees), revealing persistent congestion propagation properties.

However, although these contributions explore invariants of congestion structures across different cities, they rely on limited historical datasets, which restricts the temporal scope of their conclusions. The recurrence and predictability of congestion emergence at the mesoscopic scale remain insufficiently studied; and the causal chains linking the phases of emergence, propagation, and dissipation of congestion are still poorly characterized. These gaps hinder the development of robust methods for the automatic identification of critical scenarios, particularly cascading failures, whose understanding could enable operational anticipation of major disruption situations. At the same time, the issue of network capacity reserves — meaning segments capable of absorbing or deferring flows during disruptions — remains largely overlooked, even though it constitutes a strategic lever for dynamic and resilient urban mobility management.

This PhD research will aim to address these issues, standing at the crossroads of several methodological approaches. On the one hand, machine learning methods will be used to automatically identify congestion structures. On the other hand, traffic theory, statistical physics (percolation), and artificial intelligence will be leveraged to characterize the underlying propagation phenomena and identify scenarios likely to lead to major system failures. Finally, the use of microscopic and / or macroscopic simulation models will allow exploration of these scenarios and quantification of their impacts on the transport system.

Starting date

• 10-01

Funding category

Public funding alone (i.e. government, region, European, international organization research grant)

Funding further details

Subject to ongoing validation of funding by the project financier.