Large eddy simulation of wind turbine wakes using adaptative mesh refinement

Abstract The development of turbulent vortical wakes released downstream of wind turbines is a key physical phenomenon as it presents many technological implications for windfarm design and exploitation. The numerical prediction of these wakes constitutes a challenging problem as they involve the shedding of fine vortical structures, their instabilities, and interactions with an ambient turbulent flow. The capture of these complex, three dimensional, unsteady flow phenomena calls for a Large Eddy Simulation (LES) approach. Yet, the computational cost of a scale resolved LES can be huge and the mesh generation process is not obvious when the zones of interest are not known a-priori. Adaptive mesh refinement (AMR) allows generating Eulerian elements only in the regions of interest of the flow, where an action takes place. The AMR strategy proposed here uses the MMG3D library coupled with the YALES2 unstructured finite volume solver. The method is successfully demonstrated on two test cases, the NTNU blind test case for which experimental data exist and the reference NREL 5MW under dynamic yaw conditions.