Estimation and use of wind inflow for wind turbine performance prediction

This work investigates the dynamic response of a wind turbine to different wind inflows and develops a means of estimating the wind inflow using sparse wind measurements. The simulated different wind inflows used for this study were generated using both a stochastic model (TurbSim) and a physics-based model (LES) for neutral and stable atmospheric conditions. The turbine response to these inflows was assessed using an aero-elastic code FAST for 1.5 MW and 5.0 MW wind turbines. In addition, a POD-LSE complementary approach was used for estimating the resolved wind inflow. Sparse wind data were sampled from the LES wind inflow in a manner similar to the way a Lidar scans the wind inflow.;When used with FAST, the wind inflow from both models (TurbSim and LES) produced different wind turbine response for a given atmospheric stability condition. The largest differences were observed in the higher frequency region and were attributed to the different spectral energy and structure of turbulence present in the two wind inflows. With these results only, it is difficult to conclude which model is more realistic as neither wind inflow was validated using experimental measurements. Using LES simulations to produce wind inflows of varying spatial resolution, highly resolved wind inflows produced more turbine excitations in the higher frequencies. This turbine excitation was due to the smaller turbulent scales captured in the more highly resolved wind inflow. The study also showed that the coupling of mean shear and turbulence amplified the wind turbine response. These results showed the need for capturing smaller turbulence scales in order to allow for better wind turbine design and performance studies. In addition to studying turbine response, the ability to estimate the wind inflow using a scanning Lidar was also studied using the LES simulations. The results revealed that the best estimated wind is obtained for a Lidar scanning time within the integral time scale of the wind inflow. As a result, a better estimation was achieved for smaller grids and higher sampling rates. These findings should help to determine the required ability to scan in a grid and scanning rate of the Lidar for estimating a wind inflow of a given region.