Large Eddy Simulation of Turbulent Flows in Wind Energy

This research is devoted to the Large Eddy Simulation (LES), and to lesser extent, wind tunnel measurements of turbulent flows in wind energy. It starts with an introduction to the LES technique associated with the solution of the incompressible Navier-Stokes equations, discretized using a finite volume method. The study is followed by a detailed investigation of the Sub-Grid Scale (SGS) modeling. New SGS models are implemented into the computing code, and the effect of SGS models are examined for different applications. Fully developed boundary layer flows are investigated at low and high Reynolds numbers, and thereafter, the fully-developed infinite wind farm boundary later simulations are performed. Sources of inaccuracy in the simulations are investigated and it is found that high Reynolds number flows are more sensitive to the choice of the SGS model than their low Reynolds number counterparts. Wind tunnel measurements of an airfoil at Reynolds numbers ranging from 40,000 to 400,000 are carried out. The measurements include detailed surface pressure as well as force balance measurements for obtaining the lift, drag and pressure distribution over the airfoil. Measurements are performed in the upstroke and downstroke pitching for angles of attack between −10◦ and +25◦ and the static stall hysteresis phenomenon is investigated experimentally. Following the wind tunnel measurements, LES of the airfoil is performed using a numerical wind tunnel for Re=40,000 and Re=100,000 at a range of angles of attack. Laminar-turbulent transition, generation of laminar boundary layer separation, and formation of stall cells are investigated. The simulated airfoil characteristics are validated against measurements. It is concluded that the LES computations and wind tunnel measurements are in good agreement, should the mesh resolution, numerical discretization scheme, time averaging period, and domain size be chosen wisely. A thorough investigation of the wind turbine wake interactions is also conducted and the simulations are validated against available experimental data from external sources. The effect of several parameters on the wake structures and blade loadings are investigated. In particular, the role of SGS modeling on the flow structures and wind turbine loadings is quantified in great detail. It is found that, for the studied cases (using body-force to represent wind turbines), when a fine mesh is used to capture the tip vortices somewhat accurately, the particular choice of the SGS model is not a determining factor in simulation accuracy. To increase the role of SGS models therefore, one needs to coarsen the computational mesh, which, in return, results in poor wake predictions.