Wind Turbine Model Validation Is Improved by High‐Resolution, Measurement‐Derived Inflows

ABSTRACT There is an increased need for accurate validation of turbine models used by original equipment manufacturers to fine‐tune prototypes and predict power performance and maintenance needs while avoiding costly warranty payouts. Perhaps the most substantial source of uncertainty in typical wind turbine model validation procedures is the use of stochastically generated turbulent inflows using only 10‐min mean data from field measurements as inputs. As part of the Rotor Aerodynamics, Aeroelastics, and Wake (RAAW) campaign, we hub‐mounted a SpinnerLidar on a 2.8‐MW turbine for unobstructed and high‐fidelity measurement of the turbine's inflow. These data allowed us to create real‐time, measurement‐derived inflows that are compatible with OpenFAST. Herein, we compare the results of using these SpinnerLidar‐derived inflows to a standard approach using TurbSim‐generated inflows that use only 10‐min mean data from meteorological (met) tower anemometers as inputs. Results from multiple quantities of interest across 1645 10‐min bins of data are compared. Both inflow methods perform similarly on control related statistics, though results from Spinner inflows demonstrate higher correlation coefficients to the real turbine. Because Spinner inflows include veer and improved spatial matching of yaw misalignment, we see the largest differences in loads affected by these, such as the tower side–side moments. The Spinner inflows also allow us to identify that the tower side–side and tower top torque loads likely have model errors. Overall, we demonstrate that this method, or similar methods of turbine model validation through measurement‐derived inflows, shows considerable promise for identifying sources of error within the turbine model.