Experimental Investigation of Wind Turbine Wakes and Their Interaction

Abstract

In a wind farm, the interaction between downstream turbines and the wake of the machines positioned upstream is a complex problem, which influences the overall energy production of the turbine cluster and the lifetime of the machines. Despite the intensive investigation performed in the latest years, the physics of wake-rotor interaction is, to some extent, not yet understood. In addition, the accuracy of wind turbine simulation computational models was proven by many studies to be not satisfactory. The thesis aims to improve the knowledge on the rotor wake dynamics and on rotorwake interaction issues, by means of wind tunnel investigation of the performance of model turbines in single and tandem arrangement and of porous discs. Moreover, the collected experimental data on wake-rotor interaction represent a solid test case for simulation benchmarking purpose. The experiments were run on two three-bladed model wind turbines. Despite having a slightly different support structure, both turbines had the same blade set, based on the S826 airfoil from NREL, and rotor diameter,D = 0.9m. The turbines were operated in a low turbulence uniformflow, with a free stream velocity of U∞ = 10 m/s, both in single and tandem arrangement at two different downstream separations, namely 3 and 5 rotor diameters. The tip chord local Reynolds number for the upstream turbine, which was running at the optimum tip speed ratio of λ = 6throughout all the experiments, was approximately Retip c ≈ 105, which was demonstrated to guarantee an acceptable rotor scaling. The non-symmetries observed in the structure of wind turbine wakes were investigated via a dedicated experiment where one of the model turbines was equipped with two symmetrical support towers. The results highlighted that the wake asymmetries are a consequence of an asymmetry in cross-stream momentum transport, induced by the tower wake, which forces the rotor wake to sink downward. The analysis of the performance and of the wake of a turbine tandem in a low turbulence flow showed that the power production of the downstream turbine is almost insensitive to the downstream distance between the turbines, while it strongly depends on the rotational speed of the downstream rotor. When the downstream turbine is operated at higher-than-optimum tip speed ratios, the wake of the turbine tandem recovers faster. The efficiency of the downstream turbine, on the other hand, is only marginally lower than the efficiency of the upstream turbine, meaning that most of the power loss is to be ascribed to the reduction in mean velocity rather than to inefficiencies of the rotor. An assessment of the most common models for turbine simulation was performed via a "Blind Test" challenge, where experimenters were invited to submit computer simulations reproducing the behavior of the two in-line turbines by the only knowledge of the boundary conditions at which the test was run. The comparison of different models to the experimental results highlighted a big scatter among the thrust and power predictions of the different computations. The tandem mean wake was generally not accurately reproduced. For thrust and power prediction purposes, a fully resolved approach simulating the whole blade and turbine surface gave the best results. An actuator line rotor model coupled with a LES solver outputted the best mean and turbulent wake predictions. The experimental-numerical comparison campaign represents an important database which can become a reference for modellers who need to choose the appropriate numerical model to solve a wake interaction problem. The near wake of static wind turbine wake simulators was also analyzed, as a function of their geometry. Experiments on a porous disc with a diameter of D = 0.8 m, run for a diameter based Reynolds number of ReD = 5· 105, showed that the modification of the planar arrangement of the bars composing the mesh can strongly affect the mean wake structure and induce a non-axisymmetric near wake.