Sociomeeting: Congestion in Traffic Systems - A Statistical Physics Perspective

This talk explores traffic congestion through two complementary perspectives, both grounded in concepts from statistical physics.

The first approach aims to characterize congestion dynamics in the metropolitan area of Bologna. Leveraging GPS trajectory data from TIM—covering approximately 30% of Italian mobile devices—we apply a Fuzzy clustering algorithm to reconstruct the multilayer structure of the urban road network. This functional decomposition allows us to define and compute a congestion index within each identified layer, offering a nuanced view of how different segments of the network respond to traffic demand.

The second approach investigates the emergence of large-scale traffic jams as a first-order phase transition in the number of vehicles trapped within the road network. We explore the connection between this critical behavior and the origin-destination (OD) structure of mobility flows. To this end, we employ a GPU-accelerated microscopic multi-lane traffic simulator, analyzing transitions across various synthetic regional OD scenarios.

Both perspectives lead to mesoscopic models that capture essential features of congestion, highlighting the interplay between individual mobility patterns and collective traffic dynamics.