Enhancing Power Flow Studies Through Representative Scenario Selection

This paper presents an approach to scenario selection with the goal of improving the accuracy of power flow simulations, particularly with vast datasets involving load and weather variables. With large power systems and large amounts of available data, it is computationally expensive to choose important scenarios with a higher impact on the operation, considering load and weather for renewable generation output. Using the K-Means method for clustering, representative points are strategically chosen to simulate various solar, wind, and load conditions. The two selected representative points include an average and an outlier. Choosing these two points allows for baseline data analysis as well as anomalies, which can cause stress in the grid. The method is then demonstrated in this paper to show its functionality and how it captures the diversity of a dataset. The resulting clusters help finding interesting scenarios by addressing the variability that is inherent in power systems. This leads to improving grid reliability by preparing for a range of scenarios.