Cost-effective retrofits of power grids based on critical cascading failure scenarios identified by multi-group non-dominated sorting genetic algorithm

Power supply network is one of the most critical infrastructure networks of urban communities, but prone to the risk of cascading failures. In efforts toward disaster risk reduction of urban communities, it is thus important to identify critical cascading failure scenarios of the power supply networks and prepare effective countermeasures based on the identified scenarios. While addressing these issues, previous research efforts focused on cascading failure scenarios induced by a single component although those induced by multiple components may occur under natural or man-made disaster events. A major challenge in this problem is high computational cost required for simulating cascading failure scenarios and solving large-size optimization problems. This paper first presents an effective method to identify multi-component failure combinations entailing critical cascading failures in the network by using the overload cascading model, the multi-group non-dominated sorting genetic algorithm (MG-NSGA), and the concept of critical zone. Based on the identified critical scenarios, cost-effective retrofit combinations against the risk of cascading failures are also identified using the proposed 'elite set updating' method. The proposed disaster risk reduction decision-making methods are demonstrated and tested by numerical examples of two complex power supply networks.