Using Measurement-based Approach to Model, Predict and Control Large-scale Power Grids

University Of Tennessee Knoxville, Knoxville TN

Investigators

Abstract

The electric power grid is the backbone of all modern societies. With increasing renewable power generation, it becomes a challenge to operate the already aging U.S. power grids efficiently and reliably. The 2003 U.S. Northeast/Quebec and 2012 India blackouts have demonstrated the catastrophic consequences of a massive blackout. However, blackouts such as these could be prevented if the power system could be monitored and controlled more accurately and timely. The transformative research proposed in this project could potentially make full usage of the high-resolution measurement data available in the power grids and develop a completely new measurement-based approach to steer the power grids away from large blackouts early on. The proposed project is also coupled with a strong educational component to engage students from underrepresented groups and a broad dissemination of research findings. Based on over ten years of observation of the three major North American grids and major grids worldwide via synchrophasor measurement, strong linearity of large-scale power systems has been observed. This observation can also be verified by the interconnection-level dynamic simulations. No longer constricted by the habitual belief that the electric power grid is a nonlinear network that should be always represented by a high-order circuit-based model, this project proposes an entirely new measurement-based method to model, predict and control a large-scale interconnected power grid, especially in regard to small-signal dynamic behaviors. This project will develop measurement-based power system analysis and control applications that take full advantage of this underutilized system linearity. Specifically, the proposed linearity study will characterize the strong linearity of large-scale power grids, which has been understandably neglected by the research community, as the first step. The study will analyze the source of large-scale power grid linearity and re-examine the conventional definition of small signal. Secondly, this project will construct a linear-structured model using measurements to predict a large-scale power grid?s dynamic behavior following a small-signal disturbance. Predicting a power grid's behavior is very important for large-scale interconnected power systems and this predictive capability will provide system operators with true look-ahead capabilities. Thirdly, another novel application of a large-scale power grid?s linearity involves representing the less-interested areas of a large-scale circuit-based model with measurement-based equivalent models. This hybrid circuit and measurement model will easily achieve several orders of magnitudes higher simulation speed while maintaining acceptable accuracy. Finally and most importantly, compared to the circuit-based model that cannot be easily updated frequently, this measurement-derived model could be updated using real-time streaming measurements and keep track of the continuous change of power grids. For example, a measurement based oscillation damping controller could be updated in real time and would be much more accurate and robust, improving the stability of an interconnected power grid. With more high-resolution measurement data available, the proposed research will have a direct and immediate impact on how the U.S. interconnected power system should be modeled, analyzed, and controlled; and this advanced approach will contribute to the energy security and efficiency of the U.S. electric power grid infrastructure.