Promoteur : Prof. Laurent BRICTEUX

Résumé :

Multiscale modeling of the atmospheric boundary layer: Mesoscale/microscale coupling
Numerical modeling of atmospheric flows is one of the main topics in the field of environmental fluid dynamics due to its wide range of applications. These applications include for instance wind energy problems, solid particle or heavy gas dispersion modeling in the atmosphere, emergency response studies to pollutant release, estimation of meteorological conditions for space missions… For wind energy applications, the principal modeling approach consists in using computational fluid dynamics (CFD micro-scale models with domain lengths smaller than 1km) using logarithmic inflow conditions, the reduced form of Monin-Obukhov similarity theory, which is only valid within the first 10% of the atmospheric boundary layer and based on the horizontal homogeneity assumption. However, under extreme atmospheric stability conditions and for heterogeneous terrain studies, the applicability of these idealized inflow conditions is open to question. The other approach that is usually considered is to use numerical weather prediction (NWP mesoscale codes with domain lengths comparable to the size of a region ~10km). These NWP models are used for both operational and research purposes, for instance, for weather forecasting, or for the prediction of regional climate and hydrology. Due to the wide extends of the NWP computational domains, the spatial resolution reached with mesoscale simulations is limited. Even if the micro-scale codes allows for higher spatial resolution, their inflow conditions depends on very large length and times scales that cannot be covered with CFD simulations. Hence, there is an increasing interest in the use of NWP codes, to provide boundary and initial conditions to micro-scale CFD simulations. However, the turbulence modeling approaches for the mesoscale and micro-scale models are quite different. Therefore, the goal of this thesis is to bridge the gap between mesoscale and micro-scale models and provide a proper coupling methodology for micro-scale models to use the inflow conditions of mesoscale models.
In order to achieve that, the objectives pursued in this thesis are:
 Development of two-equation eddy viscosity closures for micro-scale simulations
 Reformulation of the developed closures for coupled CFD/NWP simulations under near-neutral atmospheric stability conditions
 Extension of the proposed models to the non-neutral conditions
 Application to an urban boundary layer case
In this framework, first, a theoretical analysis has been carried out to determine the coefficients of the eddy viscosity closures and the developed models are applied to an homogeneous atmospheric boundary layer case. Then, the applicability of the proposed models above the surface layer is investigated and coupled simulations are performed for a near-neutral complex terrain configuration: the Askervein Hill. The models are also extended to include the non-neutral conditions and coupled simulations are performed for a flat terrain case and then a sea-breeze event. Finally, the coupling methodology is applied to an urban boundary layer case: the Oklahoma City case. The results have shown the inability of standard turbulence closures and the classical logarithmic inflow conditions for the modeling of the atmospheric boundary layer with respect to coupled simulations. For the homogeneous and complex terrain cases, coupled simulations are able to provide reasonable agreement with the full-scale experiments. However, the same level of accuracy is not achieved for the Oklahoma City case.