Simulation benchmark reveals variable accuracy of wake prediction tools in complex atmospheric wind flows, highlighting the essential role of inflow characterization.
Key Points
To assess the predictive capability of diverse wind farm simulation tools against observational data across complex atmospheric flow conditions.
Engaged 16 international research groups to evaluate simulation tools ranging from engineering wake models to large-eddy simulations using diurnal data from the AWAKEN campaign.
Implemented a three-phase benchmark assessing blind predictions first, followed by progressive model adjustments as observational data availability increased.
Initial blind predictions revealed that high-fidelity simulations did not uniformly outperform simpler tools, with models exhibiting spatial bias near low-level jets and terrain-induced flow acceleration.
Model ensemble mean absolute error decreased with increased data availability, yielding up to a 40% error reduction in engineering wake models through targeted calibration.