On cascading failure models and robustness metrics in power networks

Abstract: Network research tries to give solutions within several areas, beginning from social interconnections, logistic problems, virus spreading, supply networks...etc. Some metrics have been developed in order to predict blackouts and improve power grid robustness. Metrics use part of the grid information seeking weakness without simulating the blackout process. On the other hand, simulation models unfold blackouts. In this report, we study such metrics and models, seeking that model that could give us the most relevant information. The models studied in this report represent blackout as a branch breaking process, a singular line cut or a combination of line cuts might cause subsequent disconnections and unfold a blackout. The IEEE-14 and IEEE-96 RTS bus systems are used in order to simulate blackouts. Metrics suggestions are compared with the simulations in these grids in order to know which metric better captures the critical branches. None of the studied metrics turns out to indicate always the worst blackout result as the first option. In addition, the results suggest that the effective resistance metric did not predict the critical transmission lines of the grid. Electrical betweenness and net-ability metrics obtain the critical transmission lines within their first suggestions. However, the results might be better if we also use other relevant information on the grid as the nominal power flow in each branch or the probability of having a line cut. Nevertheless, lines that carry a high amount of power flow are usually more protected and more difficult to being cut and develop a blackout. The probability of having a single line cut or a combination of line cuts will be useful in future assessments in order to seek not only the most critical branch cuts, but the most probable ones.