Nowadays wind turbines are clustered in wind farms and the wake development plays an important role in energy production and blade fatigue load of tubines. The actuator line method is an effective modeling approach that gives improtant wake flow characterstics of a wind turbine. In the last few years, numerous studies have been conducted based on this method using Ellipsys3D, a computational fluid dynamics (CDF) flow solver based on finitie volume approach. However, due to the limited order of accuracy of this solver, it is not capable of a linear stability analysis with small amplitude of perturbation. Therefore, the present work investigates implementing the actuator line ethos into a  high order method, Nek5000, a flow solver based on the spectral element approach. The main goal of the present work is to validate the code implementation by comparing the simulations results with the previous Ellipsys3D data. Both 2-D and 3-D Gaussian distribution functions are discussed for the actuator line force distribution. Parametric study is carried out regarding the smoothing parameter ε and the partitioning of the actuator line.

Understanding the flow around wind turbines is a highly relevant research question due to the increased interest in harvesting energy from renewable sources. This thesis approaches the topic by means of numerical simulations using the actuator line method and the incompressible Navier–Stokes equations in the spectral element code Nek5000. The aim is to gain enhanced understanding of the wind turbine wake structure and wind turbine wake interaction. A verification study of the method and implementation is performed against the finite volume solver EllipSys3D using two types of turbines, an idealized constant circulation turbine and the Tjæreborg turbine. It is shown that Nek5000 requires significantly lower resolution to accurately compute the wake development, however, at the cost of a smaller time step.The constant circulation turbine is investigated further with the goal of establishing guidelines for the use of the actuator line method in spectral element codes, where the mesh is inherently non-equidistant and currently used guidelines of force distribution based on Gaussian kernels are difficult to apply. It is shown that Nek5000 requires a larger kernel width in the fixed frame of reference to remove numerical instabilities. Further, the impact of different Gaussian widths on the wake development is investigated in the rotating frame of reference, showing that the convection velocity and the breakdown of the spiral tip and root vortices are dependent on the Gaussian width. In the second part, the flow around single and multiple wind-turbine setups at different operating conditions is investigated and compared with experimental results. The focus is placed on comparing the power and thrust coefficients and the wake development based on the time-averaged streamwise velocity and turbulent stresses. Further the influence of the tower model is investigated both upstream and downstream of the turbine. The results show that the wake is captured accurately in most cases. The loading exhibits a significant dependence on the Reynolds number at which the airfoil data is extracted. When the helical tip vortices are stable the turbulent stresses at the tip vortices are underestimated in the numerical simulations. This is due to the finite resolution and the projection of the actuator line forces in the numerical domain using a prescribed Gaussian width, which leads to lower induced velocities in the helical vortices. / <p>QC 20170523</p>

In the present thesis the wake flow behind wind turbines is analyzed numerically using large-eddy simulations. The wind turbine rotors are modeled by using either the actuator disc method or the actuator line method in which the blades are represented by body forces computed with airfoil data. Using these models, the boundary layers of the turbine blades are not resolved and most of the computational power is preserved to simulate the wake flow. The actuator disc method is used for the wake interaction studies of the Lillgrund wind farm. In this study the power production is simulated for two different wind directions and compared to measurements. A grid sensitivity study and a turbulence intensity study are also performed. As a last step the front row turbines are derated in an attempt to increase the total production of the farm. The results show that it is important to impose atmospheric conditions at the inlet in the simulations, otherwise production will be unrealistically low for some turbines in the farm. The agreement between the simulated and measured power is very good. The study based on derating the front row turbines does not show any positive increase on the farm production. The actuator line method is used for near wake analysis of the MEXICO rotor. In this study the near wake is simulated for five different flow cases and compared with particle image velocimetry (PIV) measurements. The analysis is performed by comparing size and circulation of the tip vortices, the radial and streamwise velocity distributions, the spatial expansion of the wake and the axial induction factor. The simulations and measurements generally are in agreement. In some cases, however, the measurements are affected by tunnel effects which are not captured in the simulations. In connection to the actuator disc method a power control strategy for operating conditions below rated power is implemented and tested. The strategy is first validated using an in-house developed blade element momentum code and then is implemented in the actuator disc method used in the EllipSys3D code. The initial tests show that the strategy responds as expected when changing the moment of inertia of the rotor and when varying the inlet conditions. Results from the implementation of the strategy in the actuator disc method in EllipSys3D show that the turbine adapts to the conditions it is operating in by changing its rotational velocity and power output when the inlet conditions are varied. / <p>QC 20130111</p>