Afin de bénéficier de vents importants, un nombre croissant d'éoliennes est installé en terrain complexe. Cependant, un terrain complexe accroit la complexité de l'écoulement et donc la prédiction du potentiel éolien. Dans ce travail, le vent en terrain complexe est simulé en soufflerie. L'objectif est d'étudier la capacité de la modélisation en soufflerie.La partie basse de l'atmosphère, appelée couche limite atmosphérique, est le siège d'important gradients de vitesse et de turbulence. Dans la soufflerie, ils sont reproduits grâce à des obstacles placés dans la section d'essai. Leurs tailles varient en fonction du type de terrain à modéliser. Cette approche expérimentale est validée par des données terrain. La reproduction des conditions atmosphériques est le paramètre crucial pour une bonne modélisation.Pour évaluer le vent en terrain complexe, le choix de la zone à reproduire autour du site d'intérêt est une question centrale : elle doit tenir compte de l'effet des reliefs environnants mais doit être assez réduite pour préserver un facteur d'échelle raisonnable dans la soufflerie. Une série d'études sur des collines simplifiées est ainsi réalisé afin de déterminer l'étendue spatiale du sillage en aval d'un relief simplifié afin de rationaliser le choix de la zone d'étude.Deux cas réels sont ensuite traités, l'ile de Bolund au Danemark et la montagne Alaiz en Espagne. Les résultats sont bons pour l'estimation de la vitesse du vent, entre 5 et 10 % mais la modélisation de la turbulence est plus difficile, des écarts jusqu'à 100 % sont enregistrés comparés aux données terrain. / To benefit from strong winds, an increasing number of wind turbines are placed in complex terrains. But complex terrains means complex flows and difficult wind resource assessment. This study proposed to use wind tunnel modelling to evaluate the wind in a complex topography. The goal of this study is to evaluate the possibilities of wind resources assessment by wind tunnel modelling and to quantify the important modelling parameters. The lower part of the atmosphere, the atmospheric boundary layer (ABL) is defined by a velocity and a turbulence gradient. The ABL is reproduced in the wind tunnel by placing obstacles and roughness elements of different size representative to the type of terrain desired. The flow produced in the wind tunnel is validated against field data and a wise choice of the obstacles is discussed to reproduce the desired wind profile. The right reproduction of the inflow conditions is found to be the most important parameter to reproduce. The choice of the area to reproduce around a site in usually difficult to make in order to keep a low scaling factor and to account for the surrounding topography. A series of tests on simplified hills helps the experimentalist in this choice by enlightening the longitudinal and vertical extension of the wake downstream different hills shapes. Finally, two complex topographies are studied in two wind tunnels, the Bolund hill in Denmark and the Alaiz mountain in Spain. The results are giving good results, 5 to 10 %, for predicting the wind speed but more scatter is observed for the modelling of the turbulence, up to 100 %. The laboratory simulation of atmospheric flows proves to be a demanding but reliable tool for the prediction of the mean wind speed in complex terrain.

To benefit from strong winds, an increasing number of wind turbines are placed in complex terrains. But complex terrains means complex flows and difficult wind resource assessment. This study proposed to use wind tunnel modelling to evaluate the wind in a complex topography. The goal of this study is to evaluate the possibilities of wind resources assessment by wind tunnel modelling and to quantify the important modelling parameters. The lower part of the atmosphere, the atmospheric boundary layer (ABL) is defined by a velocity and a turbulence gradient. The ABL is reproduced in the wind tunnel by placing obstacles and roughness elements of different size representative to the type of terrain desired. The flow produced in the wind tunnel is validated against field data and a wise choice of the obstacles is discussed to reproduce the desired wind profile. The right reproduction of the inflow conditions is found to be the most important parameter to reproduce. The choice of the area to reproduce around a site in usually difficult to make in order to keep a low scaling factor and to account for the surrounding topography. A series of tests on simplified hills helps the experimentalist in this choice by enlightening the longitudinal and vertical extension of the wake downstream different hills shapes. Finally, two complex topographies are studied in two wind tunnels, the Bolund hill in Denmark and the Alaiz mountain in Spain. The results are giving good results, 5 to 10 %, for predicting the wind speed but more scatter is observed for the modelling of the turbulence, up to 100 %. The laboratory simulation of atmospheric flows proves to be a demanding but reliable tool for the prediction of the mean wind speed in complex terrain.