Quantifying Cascading Failure Propagation in Blackouts of Electrical Transmission Networks

Large blackouts involve complicated series of tens or hundreds of events, but usually involve the cascading outage of transmission lines. Small cascades of line outages occur routinely, but usually do not lead to load shed because the power system is designed to be robust to a few outages. I want to count the line outages that are recorded in order to describe how much on average the line outages propagate as they cascade. This leads to a branching process probabilistic model of the cascading line outages. Then, given assumed or estimated initial line outages, we can calculate the probability distribution of the extent of the cascading outage from about one year of observed utility data. This gives a way to extend risk analysis to include the effect of cascading and monitor how well the power system inhibits cascading. I can also observe how cascades spread in utility networks as evidence that outages do not primarily propagate along the electrical network. I conclude that mathematical models parsimoniously describing cascading at a high-level can be validated, and their parameters summarizing the cascading can be practically estimated from observed or simulated data.