Lagrangian Dynamic Large-eddy Simulation Of Wind Turbine Wakes Combined With An Actuator Line Method

The effects of wind turbine wake(WTW) have an important impact on the production efficiency of wind turbines, rotor fatigue life and stability of electricity grid systems.Therefore, WTW has received extensive attention of the researchers in the wind energy community. However, WTW is a complicated unsteady turbulent flow behind wind turbines located in wind farms, and it depends on many factors such as wind turbine geometry, operating states, wind farm layouts, complex terrain and atmospheric wind condition.The mechanism of WTW is still not fully understood, therefore, most of the current design of wind turbines and farms have to resort to low-accuracy engineering empirical models of WTW. Given the current trend of large- or Mega-scale wind turbine generators,there will be a more significant challenge for the aerodynamic and aero-elastic loading of wind turbines, hence further research on high-accuracy models of WTW and its flow mechanism is imperative to enhance the annual power output and reduce the rotor loads.A new large eddy simulation(LES) numerical model is proposed in the dissertation,which combines the Lagrangian dynamic subgrid-scale(SGS) model with the actuator line method(ALM) technique. With the help of the numerical model proposed, the aerodynamics in the near and far wakes of the MEXICO wind turbine rotor is studied in detail.The major points of this dissertation can be summarized as follows:(1) Developing a Lagrangian dynamic LES model for WTW combined with ALM.WTW is generally an inhomogeneous and anisotropic turbulent flow due to the impacts from the rotor and tower of wind turbines, topography, atmospheric boundary layer, etc.In order to address numerous issues in the complicated anisotropy turbulence, an Lagrangian dynamic subgrid-scale model is introduced into LES. In addition, if the boundary layer of full rotors is computed, it is disadvantageous to the engineering application due to the excessive requirement of computing resources. Hence the ALM model, based on the blade element momentum(BEM), is introduced to replace the effects of airfoil viscous boundary layer on WTW.(2) Introducing the Prandtl-Glauert tip/hub-loss correction model and the Snel threedimensional correction model. Firstly, the ALM model, based on BEM, generally has a low accuracy at the tip and hub of the blades with a finite length, hence, the Prandtl tip/hub-loss factor need to be introduced to improve the ALM. Secondly, the ALM model generally depends on the tabulated airfoil lift, and the drag aerodynamic data. However,most of the existing airfoil aerodynamic data is obtained in two-dimensional airfoil measurement condition, while further investigations have found that there is a considerable difference between the airfoil data from practical three-dimensional rotor and that from two-dimensional airfoil measurement due to the effects of the centrifugal and Coriolis forces. Therefore, the Snel et al. three-dimensional correction model for two-dimensional airfoil aerodynamic data is introduced into ALM.(3) Introducing the Shives et al. selection approach of the length-scale factor of the Gaussian smoothing function. According to the traditional selection approach, the lengthscale factor of the Gaussian smoothing function in ALM model generally relies on the grid size of rotor domain. However, it is generally difficult to estimate the grid size in the simulation process. In the present model, according to Shives et al., the length scale of the Gaussian distribution is related to both the physically meaningful local airfoil chord and the grid size of rotor domain to greatly facilitate its own choice and meshing.(4) The characteristics of wind turbine near wakes are studied in detail using the proposed model. This part consists of two topics. One is to verify the WTW model proposed in this dissertation, and the other is to study the flow mechanism of near wakes using the proposed model. A detailed simulation of the near wakes of wind turbine is performed, in which the computational domain is based on the EU project “model rotor experiments in controlled conditions(MEXICO)”. In this work, three representative operational cases,i.e. turbulent wake state, design conditions, and stalled conditions are simulated. Firstly,the simulation results of the power coefficients, the spanwise and streamwise distributions of time-averaged axial velocities for the three cases are obtained. And then simulation results are compared to the experimental data from the MEXICO to verify the proposed model. Finally, the characteristics of the vortex system and turbulence intensity in wind turbine near wakes are analyzed and discussed in detail.(5) The characteristics of wind turbine far wakes are studied in detail using the proposed model. This part attempts to further reveal the flow mechanism of WTW as an extension of the above part. To be specific, the aerodynamic characteristics of far wakes are investigated at the same operational cases as for near wake, the characteristics include the spanwise distributions of time-averaged axial velocities and turbulence intensity, the coherent structures of far wake vortices and Reynolds stresses(six components). Simulation results show a noticeable difference between near and far wakes. In far wakes, there exist various unique phenomena such as velocity deficit recovery, turbulence intensity stability, and far wake expansion, and the phenomena are related to tip speed ratio(TSR).In addition, it is found that helical wake vortex instability is one of the important factors affecting the Reynolds stress distribution, which will affect the distribution of turbulence intensity.