Wake downstream of the Lillgrund wind farm - A Comparison between LES using the actuator disc method and a Wind farm Parametrization in WRF

Abstract

Simulations of the Lillgrund wind farm (located between Malmo and Copenhagen) are performed using both Large Eddy Simulation (LES) and mesoscale simulations in WRF. The aim is to obtain a better understanding of wakes generated by entire wind farms in order to improve the understanding of farm to farm interactions. The study compares the results from the two used models for the energy production and the wake characteristics downstream of the wind farm. A comparison is also performed with regards to the production data from the Lillgrund wind farm which has been filtered to be comparable to the case used in the simulations.The studied case, based on a prerun in WRF without any wind farm, has an inflow angle of 222 ± 2.5 deg, a wind speed at hub height of 9.8 m/s and a near neutral atmosphere. A logarithmic wind shear is used in LES and the turbulence intensity is 5.9%.The WRF simulations use a parameterization for wind farms. The wind farm is treated by the model as a sink of the resolved atmospheric momentum. The total energy extraction and the electrical power are respectively proportional to specified thrust and power coefficients. The generated turbulent kinetic energy arethe difference between the total and the electrical power.The LES are performed using the EllipSys3D code applying the actuator disc methodology for representing the presence of the rotors. Synthetic atmospheric turbulence is generated with the Mann model. Both the atmospheric turbulence and the wind shear are introduced using body forces.The production was found to be better estimated in LES. WRF show a slightly higher recovery behind the farm. The internal boundary layer is for the compared simulation setups higher in LES while the wake expansion is about the same in both models. The results from the WRF parameterization could potentially be improved by increasing the grid resolution. For farm to farm interaction a combination of the two methods is found to be of interest.