Congestion and extreme events in urban street networks

Congestion and extreme events in transportation networks are emergent phenomena with significant socioeconomic implications. In this work, we study congestion and extreme event properties on nearly-planar real urban street networks drawn from four cities and compare it with that on a regular square grid. For dynamics, we employ three variants of random walk with additional realistic transport features. In all the four urban street networks and 2D square grid and with all dynamical models, phase transitions are observed from a free flow to a congested phase as a function of the birth rate of vehicles. These transitions can be modified by traffic-aware routing protocols, but congestion cannot be entirely mitigated. In street networks without any structure, we observe a weakly congested regime with coexistence of both congested and free-flow components. This regime is suppressed in street networks with a grid-type structure (such as in parts of New York city) and is entirely absent in the regular 2D grid lattice. In the free-flow regime, extreme event occurrence probability is larger for small degree nodes than for hubs. Hence, our results indicate that studying congestion and extreme event properties on synthetic lattices are relevant for real street networks.