Towards quantifying cascading blackout risk

Abstract

Blackouts become widespread by initial failures propagating in a diverse and intricate cascade of rare events. We describe this complicated cascade using a bulk probabilistic model in which the initial failures propagate randomly according to a branching process. The branching process parameters can be statistically estimated from observed data or simulations and then used to efficiently predict the probability distribution of blackout size. We review the current testing of these methods on simulations and observed data and discuss the next steps towards achieving verified and practical methods for quantifying cascading failure of electric power systems. The ability to efficiently quantify cascading blackout risk from observed data and simulations could offer new ways to monitor power transmission system reliability and quantify the reliability benefit of proposed improvements.