A practical two-stage online voltage stability margin estimation method for utility-scale systems

This work proposes a practical two-stage architecture for online voltage stability margin estimation using statistical models and classification techniques. The approach models the relationship between reactive power reserves and voltage stability margin using multi-linear regression models. In order to handle uncertainty related to variable loading conditions and network topology, a few regression models are required. A classification tool is then developed in order to discern which regression model should be employed at any given operating condition. Both the regression models and the classification tool are developed offline form a database generated through a comprehensive VSA. The methodology is implemented on a reduced case of the U.S. eastern interconnection, which contains around 21k buses. The studied area represents a large part of the state of Iowa and small portions of neighboring states. NERC category B, C and D contingencies have been considered in the study. Several load increase directions are used in order to account for uncertainty in load variation. Results have shown that the methodology can successfully estimate voltage stability margin in the presence of uncertainty related to variable loading condition and network topology.