Soutenance publique de thèse de Doctorat

Résumé :

As the energy demand has continuously been growing over the last decades, the harvesting of wind energy produced by large wind farms and, more recently, by small wind turbines operating in urban environments is becoming an alternative source of clean and renewable energy. The characterization of the interaction between a wind turbine and the environmental turbulence is crucial for the efficient integration of small wind turbines in the urban environment. In addition, a sound understanding of the interactions between wind turbines and atmospheric turbulence can help to determine the wind load and the quality of the power injected into the grid.The study of small wind turbines involves multidisciplinary knowledge: electrical engineering, mechanical engineering, and aerodynamics and, the fidelity levels in the numerical simulations will depend on the quantity of interest. The objective pursued in this thesis was to achieve advances in CFD techniques that will lead to better flow predictions for small wind turbines operating in urban environments.
An important aspect to consider in the study of small wind turbines is their siting. Wisely choosing the location of the wind turbine allows to take benefit from potential wind acceleration due to the presence of buildings and also to avoid areas where the turbulence levels are too high. This problem is tackled using RANS simulations to assess the impact of a building on the flow field. The mean velocity field and the mean tke field produced by these simulations can help to address wind turbine siting problems and also to generate a realistic turbulent inflow for the LES simulations of a full wind turbine energy conversion chain. The method selected in this study is a standard approach using linear eddy viscosity turbulence models recognized to have some difficulties to handle flows around bluff bodies. An attempt to improve the reliability of this kind of simulation was made by investigating the turbulence model form uncertainty quantification. The methodology considered in this thesis allowed to clearly indicate areas where the flow is highly influenced by the turbulence model considered. This information is essential to interpret the results obtained by the classical RANS and correctly choose the location of the wind turbine.
RANS approaches are not always sufficient to accurately characterize the unsteady aerodynamic behavior of small wind turbines. The unsteady and multiscale character of this kind of flow thus calls for LES approaches which allow to properly capture the most relevant scales of the turbulence, yielding accurate flow diagnostics. The LES approach considered in this thesis properly captures the wake behavior over long distances but cannot handle solid boundaries. Consequently, the atmospheric turbulence is only accounted for at the inlet of the computational domain. In order to take this atmospheric turbulence into account, the turbulent inflow has to be as realistic as possible. In this thesis, a new method has been developed, allowing to generate turbulence inlet with realistic spectral behavior. The next step required to characterize the aerodynamic interactions between small wind turbines and the built environment with a higher level of fidelity, is to consider LES with both wind turbine models and geometrical modeling of the complex environment. Some progress has been made in this direction by setting up a computational framework to perform WMLES of urban flows in the YALES2 flow solver.