Context-Independent Centrality Measures Underestimate the Vulnerability of Power Grids

Abstract: Power grids vulnerability is a key issue in society. A component failure may trigger cascades of failures across the grid and lead to a large blackout. Complex network approaches have shown a direction to study some of the problems faced by power grids. Within Complex Network Analysis structural vulnerabilities of power grids have been studied mostly using purely topological approaches, which assumes that flow of power is dictated by shortest paths. However, this fails to capture the real flow characteristics of power grids. We have proposed a flow redistribution mechanism that closely mimics the flow in power grids using the PTDF. With this mechanism we enhance existing cascading failure models to study the vulnerability of power grids. We apply the model to the European high-voltage grid to carry out a comparative study for a number of centrality measures. `Centrality' gives an indication of the criticality of network components. Our model offers a way to find those centrality measures that give the best indication of node vulnerability in the context of power grids, by considering not only the network topology but also the power flowing through the network. In addition, we use the model to determine the spare capacity that is needed to make the grid robust to targeted attacks. We also show a brief comparison of the end results with other power grid systems to generalise the result.