Large-Eddy Simulation of Turbulent Boundary Layer Flow through Wind Turbines and Wind Farms

In this thesis, we develop a new large-eddy simulation (LES) framework to simulate atmospheric boundary layer (ABL) flows through wind turbines and wind farms. A Lagrangian scale-dependent dynamic model is used to compute the Smagorinsky model coefficient dynamically based on the local dynamics of the resolved velocity field. The turbine-generated power outputs and the turbine-induced forces (e.g., thrust, lift, drag) are parameterized using two actuator-disk models: (a) the traditional actuator-disk model without rotation (ADM-NR), which uses 1D momentum theory to relate the power output and the uniform thrust force distribution with an incoming velocity; and (b) the actuator-disk model with rotation (ADM-R), which adopts blade element theory to calculate the lift and drag forces (that produce thrust, rotor shaft torque, and power) based on the local blade and flow characteristics. The performance of the LES framework is first evaluated through comparisons with high-resolution velocity measurements collected in the wind-tunnel experiments with the single wake of a stand-alone miniature wind turbine [Chamorro and Porté-Agel, 2010a] as well as the multiple turbine wakes in a model aligned wind farm [Chamorro and Porté-Agel, 2011]. Emphasis is placed on the structure and characteristics of the simulated turbine wakes using the ADM-NR and the ADM-R. In general, the ADM-R yields improved predictions compared with the ADM-NR in the wake regions, where the ADM-NR tends to underestimate the velocity deficit and the enhancement of turbulence intensity in the wakes. Next, the LES framework is used to model multiple turbine wakes and associated power losses in a large wind farm. Here, we propose a new dynamic procedure coupled with the ADM-R to predict turbine power output based on a turbine-model-specific relationship between the shaft torque and the blade angular speed. The Horns Rev offshore wind farm is chosen as a case study since both power data from eighty Vestas V80 wind turbines and flow data from three nearby meteorological masts were available [Barthelmie et al., 2009]. The torque-speed relationship for the V80 turbine is obtained from a series of stand-alone turbine simulations. In the wind farm simulations, the ADM-R gives the best power prediction. Finally, the validated LES framework is used to study atmospheric turbulence effects on wind-turbine wakes in neutrally-stratified ABL over homogeneous flat surfaces with four different aerodynamic roughness lengths. In this study, the simulation result shows that the different turbulence intensity levels of the incoming flow lead to considerable influence on the spatial distribution of the mean velocity deficit, turbulence intensity, and turbulent shear stress in the wake region. In particular, when the turbulence intensity level of the incoming flow is higher, the turbine-induced wake (velocity deficit) recovers faster, and the locations of the maximum turbulence intensity and turbulent stress are closer to the turbine.