Stochastic Cascading Failure Model With Uncertain Generation Using Unscented Transform

Abstract

This paper proposes a novel stochastic cascading failure (CF) model with full ac power flow using the unscented transform (UT) for power girds vulnerability analysis. The quasi-steady state (QSS) model is built upon previous dc power flow model by incorporating several enhancements. Our analysis shows that dc power flow does not provide an accurate estimation of the flow process under highly variable generation thus may underestimate the severity of the cascades. The incorporation of full ac power flow constraints allows us to access voltage profiles dynamics during CF in order to simulate voltage-related failures in the grid. The under-voltage load shedding relays are modeled along with a stochastic time-inverse overload relay to accurately simulate the protective system response. In addition, more realistic assumptions are considered in the modeling of wind power penetration using geographical information of grid topology and wind potential map for a given geographical area. The accuracy of the estimated flow process based on UT method is examined under different operating conditions in a 500-bus synthetic network. The proposed model is benchmarked against the historical blackout data and widely used models in the literature showing similar statistical patterns of blackout size.