Creating a U.S. Fleet-wide Historical (70+ years) Hourly Wind Generation Potential Dataset

Accurate modeling of wind energy resources is crucial for potential wind farm siting, grid integration studies, and analyses of existing wind farm operations. Moreover, the variable nature of wind resources necessitates long time series of realistic asset-level hourly generation to characterize critical risks of persistent wind lulls or adverse correlations to regional demand. Unfortunately, the recency of the wind energy buildout precludes the possibility of such an analysis on actual historical generation data alone.

In this talk, we present a time series dataset of theoretical hourly wind generation for the entire US utility-scale wind energy fleet using high-resolution historical weather data from 1950 to the present. After downscaling ERA5 data to 3km spatial-resolution and bias-correcting via the Global Wind Atlas, we apply turbine-specific wind power curves in conjunction with efficient three-dimensional wake models to calculate hourly wind generation potential over 70+ years of data. In other words, for every operational wind farm in the US we calculate the amount of wind energy that farm could have produced for any hour from 1950-present, regardless of when the farm was actually built.

To validate this dataset of 1950-present hourly generation potential for all US wind farms, we compare against wind farm-level historical hourly generation actuals for 100+ wind farms as reported in Security Constrained Economic Dispatch (SCED) data from the Texas energy grid (ERCOT). After cleaning the data to normalize for curtailment and maintenance, we find good agreement between theoretical and non-curtailed historical values, with an average of 0.84 correlation between hourly generation values and an average annual mean capacity factor bias of 0.045.

Finally, we discuss applications of this 70+ year wind generation potential data to analyze wind-fleet generation potential, calculate farm-to-farm generation correlations, and support energy grid planning and resource adequacy exercises.