An approach to statistical estimation of cascading failure propagation in blackouts

Load power is progressively shed as large, cascading blackouts of electric power transmission systems evolve. We propose a statistical estimator to measure the extent to which the load shedding is propagated. The estimator uses data from a series of simulated blackouts. The estimator is derived from a continuous state branching process that is a high level probabilistic model of the cascading process. The estimator is tested on failure data generated by a power system model of cascading line outages. The estimates for propagation of load shed are consistent with estimates for the propagation of line outages. Estimating the initial load shed and the propagation of load shed leads to estimates of the probability distribution of blackout size. This work opens up possibilities of monitoring infrastructure failures to quantify the vulnerability to cascading and the overall risk of large cascading failures.