Identification of Cascading Failure Scenarios of Infrastructure Systems using Multi-Group Non-Dominant Sorting Algorithm

Abstract

Power transmission networks are critical infrastructure systems of urban communities, but are prone to cascading failures due to their high level of interconnectivity. Therefore, it is of great interest to identify critical components of the network that may trigger cascading failures. However, existing approaches to identify critical cascading failures focus on topological effect for a limited number of initial component failures. Meanwhile, identification based on load flow analysis without a limit on the number of triggering component failures has not been extensively studied. In this study, we simulate the overload-induced cascading failures to find the most critical scenarios of initial failure events in a power grid. The proposed approach uses the multi-group non-dominant sorting algorithm (Choi and Song, 2017) with two objective functions, i.e. network impact measure, and the number of initial component failures. Numerical experiments on a 30-bus network demonstrate that the identified critical cascading scenarios, triggered by single and multiple component failures, may not share common components necessarily. The proposed approach is expected to identify a group of critical components, which may be neglected by existing approaches.