WakeNet Dataset

Dataset used to train WakeNet, a framework for neural networks to predict wind turbine wakes. Details of the model can be found in Asmuth and Korb (2022, doi:10.1088/1742-6596/2265/2/022066). The model itself is available at https://source.coderefinery.org/wind_energy_uu/wakenet. Please cite doi:10.1088/1742-6596/2265/2/022066 when using this dataset. The data consists of 900 simulations of a single NREL 5MW turbine in different turbulent inflow conditions. Turbine height, inflow velocity, turbulence intensity, shear and tip-speed ratio is varied. Turbulence is imposed by Mann turbulence. Data consists of cross stream planes of mean velocity and Reynolds stresses. The first plane is 1 diameter upstream of the turbine and then there is a plane every D/4 up to 10D downstream of the turbine. Data is provided as vkt image files (.vti), so further parameters of the cases are provided in the header of the files. Each plane is 6Dx6D. Information about turbine location etc. can be found in the file actuator.dict