Smart Sampling of Representative Hourly Power Generation Scenario With High Renewable Penetration

The increasing penetration of renewable energy sources and energy storage in the power grid has led to an exponentially large number of possible power generation scenarios. This makes it difficult to analyze, plan, and operate the power system, and necessitates a scenario reduction method. This paper presents a smart sampling approach for identifying a subset of generation scenarios from one year of production cost modeling data. The approach takes into account the seasonal and diurnal variability of renewable energy sources, utilizing a hierarchical design followed by sliced Latin-hypercube sampling method to select a statistically representative subset of hours. A series of validation tests provides strong evidence that the sampled subset exhibits the same statistical characteristics as the original one-year data.