Exact Quantification of System Reliability Indexes in Probabilistic Contingency Analysis

Abstract

Probabilistic contingency analyses (PCAs) can provide complementary information to existing deterministic analyses for tackling the increasing uncertainties introduced by deployment of renewable generation. The PCA built into existing software calculates probabilistic indices of system problems based on a rare event (RE) approximation. When the probabilities of outage events are relatively large, the RE approximation is no longer applicable, and the calculated probabilities may deviate significantly from the real values, or even be larger than 1.0. To overcome this issue, this paper introduces two quantification methods, namely, a minimal cutset upper bound (MCUB) and a binary decision diagram (BDD). These two methods are capable of better approximating and calculating the exact system problem probabilities, respectively. Additional investigation includes the calculation of system problem frequencies and durations using both the MCUB and BDD methods. A Python implementation was integrated into an enhanced PCA (ePCA) tool. A comparison study was performed to demonstrate the feasibility and performance of the proposed methods.