Wind Farm parametrization in the mesoscale model WRF

Abstract

The project’s objective is to investigate and develop methods for prediction of mesoscale climate, wake effects and atmospheric feedbacks, for scenarios where large portions of the sea are covered with wind farms. The atmospheric flow is simulated with the WRF mesoscale model, since it has significantly lower computational costs compared to high resolution models. Due to the fact that its typical horizontal grid spacing is on the order of 2km, the energy extracted by the turbine, as well as the wake development inside the turbine- containing grid cells, are not described explicitly, but are parametrized as another sub-grid scale process. In order to appropriately capture the wind farm wake recovery and its direction, two properties are important, among others, the total energy extracted by the wind farm and its velocity deficit distribution. In the considered parametrization the individual turbines produce a thrust dependent on the background velocity. For the sub-grid scale velocity deficit, the entrainment from the free atmospheric flow into the wake region, which is responsible for the expansion, is taken into account. Furthermore, since the model horizontal distance is several times larger then the turbine diameter, it has been assumed that the generated turbulence and dissipation are balanced. From version 3.2.1 onwards, the WRF (Weather Research and Forecast) model includes a wind farm parametrization option (Fitch Scheme). In contrary to the above described parametrization where the wind turbines are positioned explicitly, the wind farms in the default scheme are treated as a density distribution, which limits the description of the internal wind farm velocity deficit development and its related efficiency. In the Fitch Scheme the wind turbines act as drag devices, where the extracted force is proportional to the turbine area interfacing a grid cell. The sub-grid scale wake expansion is achieved by adding turbulence kinetic energy (proportional to the extracted power) to the flow. The validity of both wind farm parametrizations has been verified against observational data. We use Synthetic Aperture Radar (SAR) satellite data, as well as mast measurements from meteorological masts and power measurements from wind turbines, at Horns Rev and Nysted. From the SAR satellite data the wake extension can be derived. The wind farm measurements have been used to compare the total thrust produced by both types of parametrization. In case studies the wake deficit has been estimated by the deflection of the wake due to the slowing down of the wind speed. The results of the wind farm parametrization will be used to investigate eventual climate impacts of large wind farms. Furthermore it will develop techniques for up-scaling the effects simulated by wind farm wake models into mesoscale atmospheric planetary boundary layer (PBL) parameterisations and perform simulations using these parameterisations to understand the feedbacks between the wind farms and the regional wind climate. The work will extend the current knowledge about wake effects from observations and small-scale models to potential feedbacks in the PBL atmosphere.