Testing and applying the filtered actuator line model for very coarse scales in LES of large wind farms

In the actuator line model (ALM), the wind turbine blade forces are usually applied on the flow field using a Gaussian smearing function. It is known that an optimal kernel length of about 25% chord length is required for accurate ALM-based force and flow predictions. However, for Large Eddy Simulations (LES) of extended wind farms, coarse grid resolutions and much coarser filter kernel sizes are typically required. In this study, the filtered actuator line model proposed by Martinez-Tossas and Meneveau (J. Fluid Mech. 2019, vol 863, pp 269-292) is tested and implemented at very large filtering scales, as large as 25% of rotor radius. We evaluate the accuracy of the ALM implementation by performing simulations for both an idealized blade (constant chord wing) and a wind turbine blade (NREL5MW) in uniform inflow, confirming that filter-scale independent predictions are obtained. We then compare the results with predictions from blade-element momentum (BEM) calculations and prior simulations with the optimal filtering size. We push the simulation grid to a resolution at which ALM-based LES of very large wind farms becomes feasible.