École Nationale Supérieure de
Techniques Avancées

Résumé

According to the latest report "Futurs énergétiques 2050" from RTE [RTE 2021], the installed capacity of offshore wind power should reach 50 GW by 2050. One of the major challenges in installing wind farms is predicting how much they can produce. Indeed, mechanisms such as interactions between wind turbines with wake effects (Fig. 0.1) can significantly reduce, by 10% to 20% [R. Barthelmie F. 2006], the production of these farms. It is therefore crucial to develop efficient tools for atmospheric flow modeling in order to study atmospheric flows and evaluate wind energy potential. The goal of these tools is to simulate the micro-meteorology, i.e. the local wind conditions (speeds, turbulence, etc.), in the presence of a wind farm to estimate the expected power output under various global meteorological situations. To fulfill this purpose, the CFD tool code_saturne [EDF 2024], equipped with an atmospheric flow module, is developed by the ’Mécanique des Fluides, Energies et Environnement (MFEE)’ departement at EDF R&D. With appropriate data input (e.g., localized thrust forces representing the presence of turbines, wind and temperature profiles), code_saturne is capable of modeling a wind farm by finely representing the wind field. Different closures to model the turbulence are possible, although this study focuses on the Reynolds Averaged Navier Stokes (RANS) k − ϵ closure. However, these simulations come with a computational cost that may be unsuitable for operational situations. That is the reason why engineering models (e.g., PyWake developed by DTU [Pywake 2024a]), with their simple formulations for wake deficit, blockage effect, etc., running in Python, can then be used. It would therefore be interesting to compare engineering models and CFD models in situations of interest (farm configurations, atmospheric conditions) to assess their agreements and divergences.