Detecting critical links of urban networks using cluster detection methods

Clusters of a network are sets of nodes that are strongly connected to each other but weakly connected to the rest of the network. A network link is considered critical if loss of it significantly diminishes the integrity or functionality of the network. Therefore, networks are most vulnerable to losing their critical links. Integrity of the network is measured by the relative size of the giant component. The functionality of the network is measured by the temporal network efficiency. Temporal network efficiency is the sum of reciprocal of the time it takes to traverse between node pairs of the network and is more suitable in transportation networks than the well-known network efficiency which is based on the distance. It is shown in this paper that links connecting neighboring clusters are the most critical links of the network in comparison to links with highest congestion, flows, or betweennesses. Second most important metric is found to be betweenness of links. Flow, and congestion (ratio of link flow and its capacity) are third and fourth, respectively. It was also found that the links located on the borders of communities are not those with highest values of flows, congestion, or betweenness. Infomap was found to be the most suitable cluster detection method for the urban network under study.