Cette thèse est consacrée à la simulation du mélange de scalaires passifs à grand nombre de Schmidt (faible diffusion), au moyen d’un modèle de sous-maille structurel pour la Simulation aux Grandes Echelles (LES pour Large Eddy Simulation) reposant sur le caractère multifractal des champs de gradient en turbulence. L’analyse multifractale des champs de dissipation scalaire permet, à l’aide d’une description statistique des singularités, de prendre en compte l’intermittence inhérente à ces champs. Des simulations numériques directes du mélange à différents nombres de Schmidt supérieurs à l’unité sont mises en oeuvre. Une analyse multifractale au moyen de différentes méthodes est menée afin d’obtenir les spectres de singularités de la dissipation scalaire. Une implantation du modèle de sous-maille multifractal pour la vitesse, proposé par Burton et al., est d’abord réalisée dans le code volumes finis YALES2.Une modification du modèle équivalent pour les scalaires, reposant sur une cascade multiplicative pour reconstruire la dissipation scalaire de sous-maille, est proposée afin de prendre en compte le micro-mélange à grand nombre de Schmidt. Ce modèle de sous-maille est alors évalué au moyen de tests a priori. / This thesis is focused on the simulation of turbulent mixing of passive scalars at high Schmidt numbers (low diffusivity). The modeling work is based on a structural subgrid-scale model for Large Eddy Simulation relying on the multifractal nature of gradient fields in turbulence.The multifractal formalism provides a mean to handle the characteristic intermittency of scalar dissipation fields through a statistical description of their singularities. Direct Numerical Simulations of mixing at several Schmidt numbers above unity are run with a dedicated code. Different methods are used to perform a multifractal analysis of scalar dissipation. The multifractal subgrid-scale model of Burton et al. for velocity is implemented in the Finite Volume code YALES2. A modification of the equivalent multifractal model for scalars is proposed to take into account micro-mixing at high Schmidt numbers. The model shows satisfactory results when tested a priori against direct simulations.

The present thesis deals with a number of challenges in the field of large eddy simulation (LES). These include the performance of subgrid-scale (SGS) models at fairly high Reynolds numbers and coarse resolutions, passive scalar and stochastic modeling in LES. The fully-developed turbulent channel flow is used as the test case for these investigations. The advantage of this particular test case is that highly accurate pseudo-spectral methods can be used for the discretization of the governing equations. In the absence of discretization errors, a better understanding of the subgrid-scale model performance can be achieved. Moreover, the turbulent channel flow is a challenging test case for LES, since it shares some of the common important features of all wall-bounded turbulent flows. Most commonly used eddy-viscosity-type models are suitable for moderately to highly-resolved LES cases, where the unresolved scales are approximately isotropic. However, this makes simulations of high Reynolds number wall-bounded flows computationally expensive. In contrast, explicit algebraic (EA) model takes into account the anisotropy of SGS motions and performs well in predicting the flow statistics in coarse-grid LES cases. Therefore, LES of high Reynolds number wall-bounded flows can be performed at much lower number of grid points in comparison with other models. A demonstration of the resolution requirements for the EA model in comparison with the dynamic Smagorinsky and its high-pass filtered version for a fairly high Reynolds number is given in this thesis. One of the shortcomings of the commonly used eddy diffusivity model arises from its assumption of alignment of the SGS scalar flux vector with the resolved scalar gradients. However, better SGS scalar flux models that overcome this issue are very few. Using the same methodology that led to the EA SGS stress model, a new explicit algebraic SGS scalar flux model is developed, which allows the SGS scalar fluxes to be partially independent of the resolved scalar gradient. The model predictions are verified and found to improve the scalar statistics in comparison with the eddy diffusivity model. The intermittent nature of energy transfer between the large and small scales of turbulence is often not fully taken into account in the formulation of SGS models both for velocity and scalar. Using the Langevin stochastic differential equation, the EA models are extended to incorporate random variations in their predictions which lead to a reasonable amount of backscatter of energy from the SGS to the resolved scales. The stochastic EA models improve the predictions of the SGS dissipation by decreasing its length scale and improving the shape of its probability density function. / QC 20110615

A metodologia LES (Large Eddy Simulation) é uma alternativa viável para a solução numérica de escoamentos de interesse prático em virtude da limitação computacional imposta pela resolução direta de todas as escalas presentes em escoamentos turbulentos. Entretanto, a compreensão detalhada do fenômeno da turbulência é ainda uma tarefa desafiadora em consequência do seu comportamento não linear e alta sensibilidade às condições iniciais e de contorno. Dessa forma, o sucesso de simulações LES está associado à utilização de um código computacional eficiente, com modelagem submalha que represente corretamente a dinâmica do escoamento, juntamente com a especificação de condições iniciais turbulentas fisicamente consistentes. Nesse contexto, o presente trabalho tem como objetivo o desenvolvimento de um código LES de alta ordem aliado a um método de geração de perturbações para o estudo de escoamentos turbulentos em camada limite sobre superfície plana. Foi adotada a formulação vorticidadevelocidade. A metodologia numérica baseia-se no método de diferenças finitas em malhas colocalizadas, onde as derivadas nas direções longitudinal e normal ao escoamento são aproximadas usando diferenças compactas de alta ordem. Esse estudo assume periodicidade na direção transversal do escoamento e então um método espectral é adotado nessa direção. A integração temporal é feita através do método Runge-Kutta de 4a ordem e a solução da equação de Poisson se dá por meio de um método multigrid. Para a modelagem submalha é adotado o modelo WALE (Wall-Adapting Local Eddy-viscosity). O método RFG (Random Flow Generation) foi responsável pela geração das flutuações de velocidade. Os resultados obtidos mostraram-se em boa concordância com os dados DNS (Direct Numerical Simulation) e LES presentes na literatura. / LES methodology is a viable alternative for the numerical solution of practical interest flows due to the computational limitations imposed by the direct resolution of all scales presented in turbulent flow. However, the detailed understanding of the turbulence phenomenon is still a challenging task as a result of its non-linear behavior and high sensitivity to initial and boundary conditions. Thus, the success of LES simulations is associated with the use of an efficient computational code, wherein the subgrid scale modeling accurately represents the flow dynamics, together with the specification of realistic inicial boundary conditions. In this context, this study aims to develop a high-order LES code combined with a method for generating velocity fluctuations to compute turbulent boundary layer flows over a flat plate. The vorticity-velocity formulation was adopted. The numerical scheme is based on the finite difference method in collocated grid, where the derivatives in the streamwise and wall-normal are approximated using high order compact finite difference schemes. We also assume periodicity in spanwise direction therefore it is adopted a spectral method in this direction. The method chosen for the temporal evolution is the 4th order Runge-Kutta method and the solution of Poisson equation solution is accessed via a multigrid algorithm. For subgrid modelling it is adopted the Wall-Adapting Local Eddy-viscosity (WALE) model. The RFG (Random Flow Generation) method was responsible for the generation of unsteady turbulent velocity signal. The results obtained were in good agreement with DNS (Direct Numerical Simulation) and LES from the literature.

A metodologia LES (Large Eddy Simulation) é uma alternativa viável para a solução numérica de escoamentos de interesse prático em virtude da limitação computacional imposta pela resolução direta de todas as escalas presentes em escoamentos turbulentos. Entretanto, a compreensão detalhada do fenômeno da turbulência é ainda uma tarefa desafiadora em consequência do seu comportamento não linear e alta sensibilidade às condições iniciais e de contorno. Dessa forma, o sucesso de simulações LES está associado à utilização de um código computacional eficiente, com modelagem submalha que represente corretamente a dinâmica do escoamento, juntamente com a especificação de condições iniciais turbulentas fisicamente consistentes. Nesse contexto, o presente trabalho tem como objetivo o desenvolvimento de um código LES de alta ordem aliado a um método de geração de perturbações para o estudo de escoamentos turbulentos em camada limite sobre superfície plana. Foi adotada a formulação vorticidadevelocidade. A metodologia numérica baseia-se no método de diferenças finitas em malhas colocalizadas, onde as derivadas nas direções longitudinal e normal ao escoamento são aproximadas usando diferenças compactas de alta ordem. Esse estudo assume periodicidade na direção transversal do escoamento e então um método espectral é adotado nessa direção. A integração temporal é feita através do método Runge-Kutta de 4a ordem e a solução da equação de Poisson se dá por meio de um método multigrid. Para a modelagem submalha é adotado o modelo WALE (Wall-Adapting Local Eddy-viscosity). O método RFG (Random Flow Generation) foi responsável pela geração das flutuações de velocidade. Os resultados obtidos mostraram-se em boa concordância com os dados DNS (Direct Numerical Simulation) e LES presentes na literatura. / LES methodology is a viable alternative for the numerical solution of practical interest flows due to the computational limitations imposed by the direct resolution of all scales presented in turbulent flow. However, the detailed understanding of the turbulence phenomenon is still a challenging task as a result of its non-linear behavior and high sensitivity to initial and boundary conditions. Thus, the success of LES simulations is associated with the use of an efficient computational code, wherein the subgrid scale modeling accurately represents the flow dynamics, together with the specification of realistic inicial boundary conditions. In this context, this study aims to develop a high-order LES code combined with a method for generating velocity fluctuations to compute turbulent boundary layer flows over a flat plate. The vorticity-velocity formulation was adopted. The numerical scheme is based on the finite difference method in collocated grid, where the derivatives in the streamwise and wall-normal are approximated using high order compact finite difference schemes. We also assume periodicity in spanwise direction therefore it is adopted a spectral method in this direction. The method chosen for the temporal evolution is the 4th order Runge-Kutta method and the solution of Poisson equation solution is accessed via a multigrid algorithm. For subgrid modelling it is adopted the Wall-Adapting Local Eddy-viscosity (WALE) model. The RFG (Random Flow Generation) method was responsible for the generation of unsteady turbulent velocity signal. The results obtained were in good agreement with DNS (Direct Numerical Simulation) and LES from the literature.

Dans ce travail, des méthodes de diagnostics et des techniques de développement de modèles sous-maille sont proposées pour la simulation des grandes échelles (SGE) du mélange turbulent. Plusieurs modèles sous-maille issus de ces stratégies sont ainsi présentés pour illustrer ces méthodes.Le principe de la SGE est de résoudre les grandes échelles de l'écoulement responsables des transferts principaux et de modéliser l'action des petites échelles de l'écoulement sur les échelles résolues. Au cours de ce travail, nous nous sommes appuyés sur le classement des modèles sous-maille en deux catégories. Les modèles "fonctionnels" qui s'attachent à reproduire les transferts énergétiques entre les échelles résolues et les échelles modélisées et les modèles "structurels" qui cherchent à bien reproduire le terme sous-maille. Le premier enjeu important a été d'évaluer la performance des modèles sous-maille en prenant en compte leur comportement à la fois fonctionnel (capacité à reproduire les transferts d'énergie) et structurel (capacité à reproduire le terme sous-maille exact). Des diagnosctics des modèles sous-maille ont pu être conduits avec l'utilisation de la notion d'estimateur optimal ce qui permet de connaitre le potentiel d'amélioration structurelle des modèles. Ces principes ont dans un premier temps servi au développement d'une première famille de modèles sous-maille algébrique appelée DRGM pour "Dynamic Regularized Gradient Model". Cette famille de modèles s'appuie sur le diagnostic structurel des termes issus de la régularisation des modèles de la famille du gradient. D'après les tests menés, cette nouvelle famille de modèle structurel a de meilleures performances fonctionnelles et structurelles que les modèles de la famille du gradient. L'amélioration des performances fonctionnelles consiste à supprimer la prédiction excessive de transferts inverses d'énergie (backscatter) observés dans les modèles de la famille du gradient. Cela permet ainsi de supprimer le comportement instable classiquement observé pour cette famille de modèles. La suite de ce travail propose ensuite d'utiliser l'estimateur optimal directement comme modèle sous-maille. Comme l'estimateur optimal fournit le modèle ayant la meilleure performance structurelle pour un jeu de variables donné, nous avons recherché le jeu de variable optimisant cette performance. Puisque ce jeu comporte un nombre élevé de variables, nous avons utilisé les fonctions d'approximation de type réseaux de neurones pour estimer cet estimateur optimal. Ce travail a mené au nouveau modèle substitut ANNM pour "Artificial Neural Network Model". Ces fonctions de substitution se construisent à partir de bases de données servant à émuler les termes exacts nécessaire à la détermination de l'estimateur optimal. Les tests de ce modèle ont montré qu'il avait de très bonnes perfomances pour des configurations de simulation peu éloignées de la base de données servant à son apprentissage, mais qu'il pouvait manquer d'universalité. Pour lever ce dernier verrou, nous avons proposé une utilisation hybride des modèles algébriques et des modèles de substitution à base de réseaux de neurones. La base de cette nouvelle famille de modèles ACM pour "Adaptative Coefficient Model" s'appuie sur les décompositions vectorielles et tensorielles des termes sous-maille exacts. Ces décompositions nécessitent le calcul de coefficients dynamiques qui sont modélisés par les réseaux de neurones. Ces réseaux bénéficient d'une méthode d'apprentissage permettant d'optimiser directement les performances structurelles et fonctionnelles des modèles ACM. Ces modèles hybrides allient l'universalité des modèles algébriques avec la performance élevée mais spécialisée des fonctions de substitution. Le résultat conduit à des modèles plus universels que l'ANNM. / This work develops subgrid model techniques and proposes methods of diagnosis for Large Eddy Simulation (LES) of turbulent mixing.Several models from these strategies are thus presented to illustrate these methods.The principle of LES is to solve the largest scales of the turbulent flow responsible for major transfers and to model the action of small scales of flowon the resolved scales. Formally, this operation leads to filter equations describing turbulent mixing. Subgrid terms then appear and must bemodeled to close the equations. In this work, we rely on the classification of subgrid models into two categories. "Functional" models whichreproduces the energy transfers between the resolved scales and modeled scales and "Structural" models that seek to reproduce the exact subgrid termitself. The first major challenge is to evaluate the performance of subgrid models taking into account their functional behavior (ability to reproduce theenergy transfers) and structural behaviour (ability to reproduce the term subgrid exactly). Diagnostics of subgrid models have been enabled with theuse of the optimal estimator theory which allows the potential of structural improvement of the model to be evaluated.These methods were initially involved for the development of a first family of models called algebraic subgrid $DRGM$ for "Dynamic Regularized GradientModel". This family of models is based on the structural diagnostic of terms given by the regularization of the gradient model family.According to the tests performed, this new structural model's family has better functional and structural performance than original model's family of thegradient. The improved functional performance is due to the vanishing of inverse energy transfer (backscatter) observed in models of thegradient family. This allows the removal of the unstable behavior typically observed for this family of models.In this work, we then propose the use of the optimal estimator directly as a subgrid scale model. Since the optimal estimator provides the modelwith the best structural performance for a given set of variables, we looked for the set of variables which optimize that performance. Since this set of variablesis large, we use surrogate functions of artificial neural networks type to estimate the optimal estimator. This leads to the "Artificial Neural Network Model"(ANNM). These alternative functions are built from databases in order to emulate the exact terms needed to determine the optimal estimator. The tests of this modelshow that he it has very good performance for simulation configurations not very far from its database used for learning, so these findings may fail thetest of universality.To overcome this difficulty, we propose a hybrid method using an algebraic model and a surrogate model based on artificial neural networks. Thebasis of this new model family $ACM$ for "Adaptive Coefficient Model" is based on vector and tensor decomposition of the exact subgrid terms. Thesedecompositions require the calculation of dynamic coefficients which are modeled by artificial neural networks. These networks have a learning method designedto directlyoptimize the structural and functional performances of $ACM$. These hybrids models combine the universality of algebraic model with high performance butvery specialized performance of surrogate models. The result give models which are more universal than ANNM.

<p>This thesis is concerned with particle subgrid scale (SGS) modeling in large-eddy simulation (LES) of particle-laden turbulence. Although most particle-laden LES studies have neglected the effect of the subgrid scales on the particles, several particle SGS models have been proposed in the literature. In this research, the approximate deconvolution method (ADM), and the stochastic models of Fukagata et al. (2004), Shotorban and Mashayek (2006) and Berrouk et al. (2007) are analyzed. The particle SGS models are assessed by conducting both a priori and a posteriori tests of a periodic box of decaying, homogeneous and isotropic turbulence with an initial Reynolds number of Re=74. The model results are compared with particle statistics from a direct numerical simulation (DNS). Particles with a large range of Stokes numbers are tested using various filter sizes and stochastic model constant values. Simulations with and without gravity are performed to evaluate the ability of the models to account for the crossing trajectory and continuity effects. The results show that ADM improves results but is only capable of recovering a portion of the SGS turbulent kinetic energy. Conversely, the stochastic models are able to recover sufficient energy, but show a large range of results dependent on Stokes number and filter size. The stochastic models generally perform best at small Stokes numbers. Due to the random component, the stochastic models are unable to predict preferential concentration.</p> / Master of Applied Science (MASc)

Reduced-order models are frequently used in the simulation of complex flows to overcome the high computational cost of direct numerical simulations, especially for three-dimensional nonlinear problems. Proper orthogonal decomposition, as one of the most commonly used tools to generate reduced-order models, has been utilized in many engineering and scientific applications. Its original promise of computationally efficient, yet accurate approximation of coherent structures in high Reynolds number turbulent flows, however, still remains to be fulfilled. To balance the low computational cost required by reduced-order modeling and the complexity of the targeted flows, appropriate closure modeling strategies need to be employed. In this dissertation, we put forth two new closure models for the proper orthogonal decomposition reduced-order modeling of structurally dominated turbulent flows: the dynamic subgrid-scale model and the variational multiscale model. These models, which are considered state-of-the-art in large eddy simulation, are carefully derived and numerically investigated. Since modern closure models for turbulent flows generally have non-polynomial nonlinearities, their efficient numerical discretization within a proper orthogonal decomposition framework is challenging. This dissertation proposes a two-level method for an efficient and accurate numerical discretization of general nonlinear proper orthogonal decomposition closure models. This method computes the nonlinear terms of the reduced-order model on a coarse mesh. Compared with a brute force computational approach in which the nonlinear terms are evaluated on the fine mesh at each time step, the two-level method attains the same level of accuracy while dramatically reducing the computational cost. We numerically illustrate these improvements in the two-level method by using it in three settings: the one-dimensional Burgers equation with a small diffusion parameter, a two-dimensional flow past a cylinder at Reynolds number Re = 200, and a three-dimensional flow past a cylinder at Reynolds number Re = 1000. With the help of the two-level algorithm, the new nonlinear proper orthogonal decomposition closure models (i.e., the dynamic subgrid-scale model and the variational multiscale model), together with the mixing length and the Smagorinsky closure models, are tested in the numerical simulation of a three-dimensional turbulent flow past a cylinder at Re = 1000. Five criteria are used to judge the performance of the proper orthogonal decomposition reduced-order models: the kinetic energy spectrum, the mean velocity, the Reynolds stresses, the root mean square values of the velocity fluctuations, and the time evolution of the proper orthogonal decomposition basis coefficients. All the numerical results are benchmarked against a direct numerical simulation. Based on these numerical results, we conclude that the dynamic subgrid-scale and the variational multiscale models are the most accurate. We present a rigorous numerical analysis for the discretization of the new models. As a first step, we derive an error estimate for the time discretization of the Smagorinsky proper orthogonal decomposition reduced-order model for the Burgers equation with a small diffusion parameter. The theoretical analysis is numerically verified by two tests on problems displaying shock-like phenomena. We then present a thorough numerical analysis for the finite element discretization of the variational multiscale proper orthogonal decomposition reduced-order model for convection-dominated convection-diffusion-reaction equations. Numerical tests show the increased numerical accuracy over the standard reduced-order model and illustrate the theoretical convergence rates. We also discuss the use of the new reduced-order models in realistic applications such as airflow simulation in energy efficient building design and control problems as well as numerical simulation of large-scale ocean motions in climate modeling. Several research directions that we plan to pursue in the future are outlined. / Ph. D.

This work is motivated by the question: how much complexity is appropriate for a cloud parameterization used in general circulation models (GCM). To approach this question, cloud parameterizations across the complexity range are explored using general circulation models and theoretical Monte-Carlo simulations. Their results are compared with high-resolution satellite observations and simulations that resolve the GCM subgrid-scale variability explicitly. A process-orientated evaluation is facilitated by GCM forecast simulations which reproduce the synoptic state. For this purpose novel methods were develop to a) conceptually relate the underlying saturation deficit probability density function (PDF) with its saturated cloudy part, b) analytically compute the vertical integrated liquid water path (LWP) variability, c) diagnose the relevant PDF-moments from cloud parameterizations, d) derive high-resolution LWP from satellite observations and e) deduce the LWP statistics by aggregating the LWP onto boxes equivalent to the GCM grid size. On this basis, this work shows that it is possible to evaluate the sub-grid scale variability of cloud parameterizations in terms of cloud variables. Differences among the PDF types increase with complexity, in particular the more advanced cloud parameterizations can make use of their double Gaussian PDF in conditions, where cumulus convection forms a separate mode with respect to the remainder of the grid-box. Therefore, it is concluded that the difference between unimodal and bimodal PDFs is more important, than the shape within each mode. However, the simulations and their evaluation reveals that the advanced parameterizations do not take full advantage of their abilities and their statistical relationships are broadly similar to less complex PDF shapes, while the results from observations and cloud resolving simulations indicate even more complex distributions. Therefore, this work suggests that the use of less complex PDF shapes might yield a better trade-off. With increasing model resolution initial weaknesses of simpler, e.g. unimodal PDFs, will be diminished. While cloud schemes for coarse-resolved models need to parameterize multiple cloud regimes per grid-box, higher spatial resolution of future GCMs will separate them better, so that the unimodal approximation improves.

La modélisation par simulation des grandes échelles (Large-Eddy Simulation - LES) des processus physiques régissant la couche limite atmosphérique (CLA) demeure complexe de part la difficulté des modèles à capter l'évolution de la turbulence entre différentes conditions de stratification. De ce fait, l'étude LES du cycle diurne complet de la CLA comprenant des situations convectives la journée et des conditions stables la nuit est très peu documenté. La simulation de la couche limite stable où la turbulence est faible, intermittente et qui est caractérisée par des structures turbulentes de petite taille est tout particulièrement compliquée. En conséquence, la capacité de la LES à bien reproduire les conditions météorologiques de la CLA, notamment en situation stable, est étudiée à l'aide du code de mécanique des fluides développé par EDF R&D, Code_Saturne. Dans une première étude, le modèle LES est validé sur un cas de couche limite convective quasi stationnaire sur terrain homogène. L'influence des modèles sous-maille de Smagorinsky, Germano-Lilly, Wong-Lilly et WALE (Wall-Adapting Local Eddy-viscosity) ainsi que la sensibilité aux méthodes de paramétrisation sur les champs moyens, les flux et les variances est discutées. Dans une seconde étude le cycle diurne complet de la CLA pendant la campagne de mesure Wangara est modélisé. L'écart aux mesures étant faible le jour, ce travail se concentre sur les difficultés rencontrées la nuit à bien modéliser la couche limite stable. L'impact de différents modèles sous-maille ainsi que la sensibilité au coefficient de Smagorinsky ont été analysés. Par l'intermédiaire d'un couplage radiatif réalisé en LES, les répercussions du rayonnement infrarouge et solaire sur le jet de basse couche nocturne et le gradient thermique près de la surface sont exposées. De plus l'adaptation de la résolution du domaine à l'intensité de la turbulence et la forte stabilité atmosphérique durant l'expérience Wangara sont commentées. Enfin un examen des oscillations numériques inhérentes à Code_Saturne est réalisé afin d'en limiter les effets / Large-eddy simulation (LES) of the physical processes in the atmospheric boundary layer (ABL) remains a complex subject. LES models have difficulties to capture the evolution of the turbulence in different conditions of stratification. Consequently, LES of the whole diurnal cycle of the ABL including convetive situations in daytime and stable situations in the night time is seldom documented. The simulation of the stable atmospheric boundary layer which is characterized by small eddies and by weak and sporadic turbulence is espacialy difficult. Therefore The LES ability to well reproduce real meteorological conditions, particularly in stable situations, is studied with the CFD code developed by EDF R&D, Code_Saturne. The first study consist in validate LES on a quasi-steady state convective case with homogeneous terrain. The influence of the subgrid-scale models (Smagorinsky model, Germano-Lilly model, Wong-Lilly model and Wall-Adapting Local Eddy-viscosity model) and the sensitivity to the parametrization method on the mean fields, flux and variances are discussed.In a second study, the diurnal cycle of the ABL during Wangara experiment is simulated. The deviation from the measurement is weak during the day, so this work is focused on the difficulties met during the night to simulate the stable atmospheric boundary layer. The impact of the different subgrid-scale models and the sensitivity to the Smagorinsky constant are been analysed. By coupling radiative forcing with LES, the consequences of infra-red and solar radiation on the nocturnal low level jet and on thermal gradient, close to the surface, are exposed. More, enhancement of the domain resolution to the turbulence intensity and the strong atmospheric stability during the Wangara experiment are analysed. Finally, a study of the numerical oscillations inherent to Code_Saturne is realized in order to decrease their effects

Submitted by PPG Engenharia e Tecnologia de Materiais (engenharia.pg.materiais@pucrs.br) on 2018-06-05T13:28:29Z No. of bitstreams: 1 frantz2018simulacoes.pdf: 23131075 bytes, checksum: e748910d1820968a07c86be9461b7489 (MD5) / Approved for entry into archive by Sheila Dias (sheila.dias@pucrs.br) on 2018-06-12T12:40:17Z (GMT) No. of bitstreams: 1 frantz2018simulacoes.pdf: 23131075 bytes, checksum: e748910d1820968a07c86be9461b7489 (MD5) / Made available in DSpace on 2018-06-12T12:49:08Z (GMT). No. of bitstreams: 1 frantz2018simulacoes.pdf: 23131075 bytes, checksum: e748910d1820968a07c86be9461b7489 (MD5) Previous issue date: 2018-03-09 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior - CAPES / This work investigates the method of large-eddy simulation (LES) in the context of gravity currents, which is found necessary since it allows a substantial increase in the order of magnitude of the characteristic Reynolds number used in numerical simulations, approaching them with natural scales, in addition to significantly reducing the computational cost. The implicit large eddy simulation (ILES) methodology, based on the spectral vanishing viscosity model, is unprecedentedly employed in the context of gravity currents, is compared against with explicit methods such as the static and dynamic Smagorisnky. The evaluation of the models is performed based on statistics from a direct numerical simulation (DNS). Results demonstrate that the first model based purely on numerical dissipation, introduced by means of the second order derivative, generates better correlations with the direct simulation. Finally, experimental cases of the literature, in different flow configurations, are reproduced numerically showing good agreement in terms of the front position evolution. / Este trabalho investiga o m?todo de simula??o de grandes escalas (LES) no contexto de correntes gravitacionais. O mesmo se faz necess?rio, visto que possibilita um aumento substancial da ordem de grandeza do n?mero de Reynolds caracter?stico utilizado em simula??es num?ricas, aproximando os mesmos de escalas naturais, al?m de reduzir significativamente o custo computacional dos c?lculos. A avalia??o dos modelos ? realizada utilizando uma base de dados de simula??o num?rica direta (DNS). A metodologia de simula??o de grandes escalas impl?cita (ILES), baseada no modelo de viscosidade turbulenta espectral, ? colocado a prova de maneira in?dita no contexto de correntes de gravidade com m?todos expl?citos dispon?veis na literatura. Resultados demonstram que o mesmo, baseado puramente em dissipa??o num?rica introduzida por meio do comportamento dos esquemas de derivada de segunda ordem, gera melhores correla??es com as estat?sticas baseadas em campos m?dios da simula??o direta. Por fim, casos experimentais da literatura, em diferentes configura??es de escoamento, s?o reproduzidos numericamente.