Towards Estimating the Statistics of Simulated Cascades of Outages With Branching Processes

Branching processes can be applied to simulated cascading data to describe the statistics of the cascades and quickly predict the distribution of blackout sizes. We improve the procedures for discretizing load shed data so that a Galton-Watson branching process may be applied. The branching process parameters such as average propagation are estimated from simulated cascades and the branching process is then used to estimate the distribution of blackout size. We test the estimated distributions with line outage and load shed data generated by the improved OPA and AC OPA cascading simulations on the IEEE 118-bus system and the Northeast Power Grid of China.