Study case in Rio de Janeiro
Study case in LA
Using a multi-city framework, we link the spatial congestion propagation patterns and the macroscopic vehicle throughput. Using the fundamental diagram of each link in a traffic network, we determine the number and the size of congested links, which are spatially connected (i.e. connected components of congested links in a network). The features of these congested clusters exhibit an interesting evolution over time and a surprising similarity to the evolution of the average network flow as given by the MFD. As a result, we show that an urban traffic network reaches its maximum vehicle throughput when the maximum number of congested clusters is found. This key moment serves as a precursor for the subsequent process, where the clusters merge into a single giant component. This phenomenon is called percolation and characterized by a very rapid growth of the giant component, which in traffic is accompanied by a drastic deterioration of the average network speeds. Thus, we show that the traffic collapse in a city results from a percolation process. Therefore, we offer novel methods to model traffic in urban road networks, e.g. relating the maximum vehicle throughput and the network’s structure.
Publication:
Ambühl, L., Olmos, L. E., Menendez, M., & Gonzalez, M. C. (2020). The collapse of urban traffic networks (in preparation)