Small-Scale Modeling of Road Deterioration in an Active Transportation System

In this article the authors build upon a basic model of road deterioration and congestion and improve it to consider agent behaviour. The basic model is based on two assumptions, linking road deterioration to the edge betweenness centrality, and the traffic flow (congestion) to the Wiener index. To improve and enhance the initial model, agents are launched within the graph. For this purpose the graph of the city is clustered. Two popular algorithms for clustering are used: Fluid Communities and K-Means Clustering. They are compared using two metrics: the David-Bouldin index and the silhouette coefficient. The clustering in this work is done in a way that assures that the dynamics of the number of road repairs are roughly split into two halves when the intracluster edges and the intercluster edges are considered separately. Then, a new graph consisting of clusters is created and agents are launched between the new nodes. After that the shortest paths are recalculated and the road deterioration dynamics in the original graph are adjusted by recalculating the edge betweenness centrality considering the updated shortest paths.