Posters

Abstract

Together with on-site measurements, mesoscale models are commonly used to assess the wind resource at locations of planned wind farms, either by directly simulating the site of interest or by utilizing existing wind atlases. Low-fidelity engineering wake models are then used for layout optimization by estimating wake losses. However, with the increasing number of installed turbines, cluster wake effects become more relevant, and the influence of neighboring wind farms can no longer be neglected. Cluster wakes are currently not accurately captured by engineering models, while mesoscale models contain the higher degree of physics needed to simulate cluster wakes, purely using mesoscale models is also undesirable. Due to its coarse resolution, layout effects are not accounted for, which might result in inaccurate yield estimates. Besides this, the higher computational costs would not allow for layout optimization. Here, we present a multi-model approach that simulates external wakes with a mesoscale model and subsequently adds the wind farm of interest with an engineering model. This approach is compared to other pure mesoscale and engineering model simulation strategies. The Weather Research and Forecasting model (WRF) and the engineering model FOXES (Farm Optimization and eXtended yield Evaluation Software) were utilized in this work. Wind turbines are simulated with the default Fitch wind farm parameterization. Besides the wind speed and wind direction, the turbulent kinetic energy (TKE) is extracted from WRF and fed into FOXES. The analytical wake model used in FOXES is the TurbOPark (Turbulence Optimized Park) model, including superposition principle and tuning parameter value. The analysis focuses on a wind farm cluster in the German Bight, namely the Heligoland-cluster (N-4). The simulated turbines match the installed turbines in terms of hub height, rotor diameter and power and thrust coefficient curves. The full climatology of a reference year was simulated, that is representative for the long-term conditions. To demonstrate the proposed modeling tool chain, three simulations strategies are carried out: 1. Pure engineering: WRF without turbines provides the background flow. Full wind farm cluster simulated with FOXES. 2. Pure mesoscale: Mesoscale model simulates all turbines via the wind farm parametrization, FOXES is not used. 3. Coupled engineering-mesoscale: Multi-model tool chain in which WRF simulates the external wakes and FOXES simulates the wind farm of interest. Corresponding to previous results, the wind speed reduction is larger for WRF than for FOXES. Besides, in WRF the wind speed reduction spreads to higher altitudes (up to the boundary layer's capping inversion), which is not the case in FOXES. This clearly illustrates that WRF contains a higher degree of physics than FOXES. The presentation will also discuss computational efforts and potential validation strategies of the presented approach.