Théveninm equivalent estimation for power system voltage stability analysis

Abstract

Previous work has demonstrated that a static equivalent model can be found to accurately support the active and reactive power load and bus voltage variations of a system using only the measurable data seen at any point on the grid.

Furthermore, the algorithm is tuned to circumvent the limitations of two-second-measured unsynchronized SCADA data. An intelligent estimation algorithm containing some rudimentary detection and handling of undesirable phenomenon in the measured data is also developed.

In this thesis, a real-time Théveninm equivalent estimation algorithm is systematically developed, tested and tuned using simulated and real-world-measured power system data. The on-line implementation of the algorithm will allow an operator to receive rapid summary analysis of a robust interconnected system--providing useful information such as the voltage collapse point and stability margin. This will prove most useful when advanced techniques such as state estimation yield uninformative results due to convergence issues; as well as during large unforeseen system variations attributed to line-switching and load changing.

The Bonneville Power Administration has provided simulated PMU data and 24-hour SCADA measurements from a wind hub on the US interconnection. This data is then used to test and validate the methods developed. This data may yield erratic and inaccurate results due to measurement errors and quantization noise. An in-depth analysis of the capabilities and limitations of the Th\'evenin equivalent model in the presence natural and artificial noise will readily lead to a robust and computationally inexpensive real-time estimation solution.

Increasing penetration of renewable power technologies with unpredictable load and generation patterns has engendered the creation of naturally hybrid systems on the US power grid. These systems must be rapidly and accurately analyzed to provide useful information to the system operator or subordinate control process.

Electrical power systems are increasingly in need of advanced modeling, estimation, and control methods. A simple and powerful voltage stability analysis method is to use a static Th\'evenin equivalent model to represent the myriad connections to a power grid or load center. Curve-fitting methods such as recursive least-squares approximation are limited in reliability when there is little variation in measured data. More sophisticated methods are dependent on the computation of multiple derivatives.