Simulating Wind Turbine Wakes from Multiple Interacting Wind Farms in the Western US Using the WRF Wind Farm Parameterization

Characterizing wind turbine wakes and their effects on downwind turbines is important for integrating renewably-generated electricity into power grids by correctly predicting available wind power. In this study, we advance validation efforts of the Weather Research and Forecasting model’s Wind Farm Parameterization (WRF-WFP). We employ meteorological observations, wind turbine power production data, and wind turbine nacelle observations from multiple wind farms located in the western United States at elevations above one kilometer. Data from four neighboring wind farms, including 348 turbines spread along a longitudinal distance of about 63 km, are available. Wind data from a 60-m meteorological tower upwind of the wind farm are also available. We simulate cases where we anticipate the largest wake effects, with upwind wind speed ranges from 7–11 m s^-1 and when most, if not all, of the turbines are operating. Simulations with a number of WRF permutations are compared: a) WRF power prediction with no WFP, b) WRF power prediction with the current version of the WFP, and c) WRF power prediction with a number of modifications to the WFP. Modifications to the WFP include density corrections to the turbine power curve, consideration of rotor-equivalent wind speed, and adjustments to the turbulence generated by the turbines. We find that the inclusion of the WFP greatly improves wind power predictions as compared to WRF simulations with no wake impacts. Further, the density corrections contribute a small improvement, and the inclusion of turbine-generated turbulence is necessary in this location where turbines are located between 5D and 15D downwind of each other. Simulations and observations of wind speed and power will be analyzed and discussed.