Mots-clés de l'auteur: anabatic flow ; direct numerical simulation ; dynamic surface roughness model ; energy budget ; immersed boundary method ; katabatic flow ; large-eddy simulation ; Prandtl model ; rough surfaces ; turbulence.

This volume is a collection of lectures given at the two colloquia on atmospheric flows over complex terrain with applications to wind energy and air pollution, organized and sponsored by ICTP in Trieste, Italy. The colloquia were the result of the recognition of the importance of renewable energy sources, an important aspect which grows yearly as the environmental problems become more pronounced and their effects more direct and intense, while at the same time, the wise management of the Earth's evidently limited resources becomes imperative.It is divided into two main parts. The first, which comprises Chaps. 1 to 4, presents the structure of the atmospheric boundary layer with emphasis in the region adjacent to the ground. The second, Chaps. 5 to 10, discusses methods for the numerical computation of the wind field on an arbitrary terrain. The unique feature of this book is that it does not stop at the theoretical exposition of the analytical and numerical techniques but includes a number of codes, in a diskette, where the mechanisms and techniques presented in the main part are implemented and can be run by the reader. Some of the codes are of instructional value while others can be utilized for simple operational work.Some of the lecturers are: D N Asimakopoulos, C I Aspliden, V R Barros, A K Blackadar, G A Dalu, A de Baas, D Etling, G Furlan, D P Lalas, P J Mason, C F Ratto and F B Smith.

Starting with the description of meteorological variables in forest canopies and its parameter variations, a numerical three-dimentional model is developed. Its applicability is demonstrated, first, by wind sheltering effects of hedges and, second, by the effects of deforestation on local climate in complex terrain. Scientists in ecology, agricultural botany and meteorology, but also urban and regional lanners will profit from this study finding the most effective solution for their specific problems.

The present work establishes a complete process to perform the micro-scale simulations of the neutral atmospheric boundary layer (ABL) flow over realistic terrain with multi-fidelity turbulence modeling approaches, and to validate the numerical predictions with the available on-site wind data. The complex terrain is located within the Chokecherry and Sierra Madre wind farm site which encompasses about 500 square kilometers of rugged and diverse terrain in south-central Wyoming. A robust conditional sampling procedure for the meteorological tower (met-tower) data to identify near-neutral ABL condition based on a criterion of the turbulence intensity is developed. The conditionally averaged wind data on fourteen met-towers is used for the model validation. The ABL flow simulations are conducted based on the OpenFOAM-based simulator for on/offshore wind farm applications (SOWFA) with both RANS and LES turbulence modeling approaches. The turbulent inflow is generated through a two-stage iterative approach using a precursor method. Appropriate boundary conditions are developed to adjust the real flow patterns over the complex terrain. In the RANS approach, a new formulation to calculate the production term in the transport equation for the turbulent kinetic energy (TKE) is developed to greatly reduce the commonly observed nonphysical near-surface TKE peak, and to improve the prediction of law-of-the-wall scaling in the near-surface region. In the LES approach, a low-dissipative, scale-selective discretization scheme is applied to the non-linear convection term of the filtered momentum equation. The aim is to reduce the numerical dissipation arising from the typically adopted upwind-biased schemes while maintaining second-order accuracy. The simulation results from both RANS and LES approaches are qualitatively compared with each other, and quantitatively compared to the conditional met-tower data in terms of the mean, standard deviation and direction of the wind at three about ground levels. The instantaneous flow field and turbulent structures are predicted by the LES approach. Overall, the wind statistic obtained by RANS and LES approaches show reasonable agreement compared to the met-tower data, except for some under-predictions at four met-tower located closer to the main ridge of the hill in a region of strong terrain variations.

The present thesis describes the work carried out using the OpenFOAM solver with a Reynolds-Averaged Navier Stokes (RANS) approach to investigate the wind flow at complex sites for wind-energy exploitation. Toward this objective, several physical effects such as buoyancy, forest canopies, Coriolis forces, stratification as well as humidity have been implemented in the model to improve wind-field predictions. First, the wind flow in an urban environment and, more precisely, a university campus is investigated. A stationary logarithmic profile for the wind velocity at the inlet is prescribed. Despite the assumption of a flat terrain, which is a drastic simplification of the real ground, the study shows how a simple canopy model improves the prediction of the flow at the site. The simulation is validated with long term measurements from a network of six stations. Secondly, results from a rural case in the Swabian Alb in Southern Germany, characterized by a forested escarpment, are presented. The model is adapted to atmospheric boundary layer (ABL) flows and a computational domain with a ground conforming to the site orography is built. To get more realistic boundary conditions and to avoid the assumption of logarithmic profiles, the solver is coupled with a numerical weather prediction (NWP) model. The coupling is performed using a one-way approach, i.e the coarse weather model provides input to the OpenFOAM solver through the lateral boundary conditions of the computational domain. Simulations with and without forest are compared. The results with a canopy model clearly show at the lower levels a flow deceleration and an increase in turbulence intensities by a factor of four, when compared to results without forest. The study reveals again the important impact of the forest on the wind-field, especially at turbine-relevant heights. Finally, the transient approach (unsteady RANS) is tested by using time-dependent boundary conditions. The accuracy of the coupling is evaluated by validating the simulation results against measurements from a tall meteorological tower as well as an unmanned aircraft system. Adopting a transient approach leads to an excellent agreement of the model. The thesis shows that an unsteady RANS based solver, which accounts for first-order relevant physics, can be valuable for a wind resource assessment at low computational cost compared to detached-eddy (DES) or large-eddy (LES) simulations.