Large-eddy simulations of the interaction between wind farms and mesoscale effects
Anja Stieren is a PhD student in the research group Physics of Fluids. Supervisors are prof.dr. D. Lohse and dr. R.J.A.M. Stevens from the Faculty of Science & Technology.
During my PhD, I studied the interaction of wind turbine wakes within and between wind farms. Wind turbine wakes are regions of decreased velocity and increased turbulence that form behind each turbine. In wind farms, wind turbine wakes reduce the power production of downstream turbines. Consequently, it is essential to study the impact of wake effects on wind farm performance.
Wakes have been observed up to 55km behind wind farms, while the distance between wind farms in regions such as the North Sea is typically much smaller. However, the influence of wind farm wakes on downstream wind farms is not yet well understood. The reason is that there is limited insight into the influence of atmospheric mesoscale effects, such as dynamic wind speed and direction changes, baroclinicity, and atmospheric thermal stability, on the recovery of wind farm wakes. Detailed computer simulations can help to improve our understanding of wake effects on wind farm performance. In contrast to field measurements, the atmospheric conditions in computer simulations are prescribed and known, providing an ideal setting to study the flow physics around wind farms.
In my PhD I performed high-fidelity large eddy simulations (LES) to study the interaction of wind farms with mesoscale effects and neighboring wind farms. LES provide a detailed representation of the turbulent atmospheric flow in and around wind farms. Turbulent flow simulations are very challenging due to the large range of length and time scales that need to be captured. While a typical wind farm has a size of several square kilometers, the smallest turbulent structures range in the scale of millimeters. The large-scale flow features are explicitly resolved in LES, and small-scale turbulence is parameterized using a sub-grid scale (SGS) model. Consequently, the accuracy of LES is highly dependent on the SGS model used to parameterize these processes.
In part I of my thesis we determined the most suitable SGS model for large-scale simulations. Additionally, we developed a new method to include dynamic wind direction changes in the LES of wind farms. Dynamic wind direction changes originate on scales larger than typical wind farm sizes, and are usually not included in fundamental studies of wind farms. We show that these dynamic wind direction changes can positively and negatively affect the power production of wind farms. We also include an additional mesoscale phenomenon, negative geostrophic shear, in the LES. We find that this phenomenon can create an upward flux above the wind farm, which limits the energy entrainment into the farm.
In Part II of my thesis, wind farm wakes, and their impact on downstream positioned wind farms are analyzed using LES. We show that the performance of the leading row and the wake recovery of the downstream farm are highly impacted by the wake of the upstream farm. The results are used to evaluate the wind farm wake recovery predicted by engineering models. All engineering models under consideration overestimate the wind farm wake recovery compared to LES observations. Therefore, we conclude that these engineering models must be updated to include the interaction between wind farms.