The work presented here is part of an ongoing effort to develop a highly accurate and numerically efficient turbulence simulation technique. The paper consists of four main parts, viz., the general discussion of the procedure known as Additive Turbulent Decomposition, the derivation of the "synthetic velocity" subgrid-scale model of the high wavenumber turbulent fluctuations necessary for its implementation, the numerical investigation of this model and a priori tests of said models physical validity. Through these investigations we have demonstrated that this procedure, coupled with the use of the "Poor Mans Navier-Stokes" equation subgrid-scale model, has the potential to be a faster, more accurate replacement of currently popular turbulence simulation techniques since: 1. The procedure is consistent with the direct solution of the Navier-Stokes equations if the subgrid-scale model is valid, i.e, the equations to be solved are never filtered, only solutions. 2. Model parameter values are "set" by their relationships to N.S. physics found from their derivation from the N.S. equation and can be calculated "on the fly" with the use of a local high-pass filtering of grid-scale results. 3. Preliminary studies of the PMNS equation model herein have shown it to be a computationally inexpensive and a priori valid model in its ability to reproduce high wavenumber fluctuations seen in an experimental turbulent flow.

Fluid-structure interaction (FSI) is an important physical phenomenon in many applications and across various disciplines including aerospace, civil and bio-engineering. In civil engineering, applications include the design of wind turbines, pipelines, suspension bridges and offshore platforms. Ocean structures such as drilling risers, mooring lines, cables, undersea piping and tension-leg platforms can be subject to strong ocean currents, and such structures may suffer from Vortex-Induced Vibrations (VIV's), where vortex shedding of the flow interacts with the structural properties, leading to large amplitude vibrations in both in-line and cross-flow directions. Over the past years, many experimental and numerical studies have been conducted to comprehend the underlying physical mechanisms. However, to date there is still limited understanding of the effect of oscillatory interactions between fluid flow and structural behavior though such interactions can cause large deformations. This research proposes a mathematical framework to accurately predict FSI for elastically supported rigid structures. The numerical method developed solves the Navier-Stokes (NS) equations for the fluid and the Equation of Motion (EOM) for the structure. The proposed method employs Finite Differences (FD) on Cartesian grids together with an improved, efficient and oscillation-free Immersed Boundary Method (IBM), the accuracy of which is verified for several test cases of increasing complexity. A variety of two and three dimensional FSI simulations are performed to demonstrate the accuracy and applicability of the method. In particular, forced and a free vibration of a rigid cylinder including Vortex-Induced Vibration (VIV) of an elastically supported cylinder are presented and compared with reference simulations and experiments. Then, the interference between two cylinders in tandem arrangement at two different spacing is investigated. In terms of VIV, three different scenarios were studied for each cylinder arrangement to compare resonance regime to a single cylinder. Finally, the IBM is implemented into a three-dimensional Large-Eddy Simulation (LES) method and two high Reynolds number (Re) flows are studied for a stationary and transversely oscillating cylinder. The robustness, accuracy and applicability of the method for high Re number flow is demonstrated by comparing the turbulence statistics of the two cases and discussing differences in the mean and instantaneous flows.

Acoustic noise pollution is an environmental aggressor in everyday life. Aero- dynamically generated noise annoys and was linked with health issues. It may be caused by high-speed turbulent free flows (e.g. aircraft jet exhausts), by airflow interacting with solid surfaces (e.g. fan noise, wind turbine noise), or it may arise within a confined flow environment (e.g. air ventilation systems, refrigeration systems). Hence, reducing the acoustic noise levels would result in a better life quality, where a systematic approach to decrease the acoustic noise radiation is required to guarantee optimal results. Computational predic- tion methods able to provide all the required flow quantities with the desired temporal and spatial resolutions are perfectly suited in such application areas, when supplementing restricted experimental investigations. This thesis focuses on the use of numerical methodologies in compressible flow applications to understand aerodynamically noise generation mechanisms and to assess technologies used to suppress it. Robust and fast steady-state Reynolds Averaged Navier-Stokes (RANS) based formulations are employed for the optimal design process, while the high fidelity Large Eddy Simulation (LES) approach is utilized to reveal the detailed flow physics and to investigate the acoustic noise production mechanisms. The employment of fast methods on a wide range of cases represents a brute-force strategy used to scrutinize the optimization parameter space and to provide general behavioral trends. This in combination with accurate simulations performed for particular condi- tions of interest becomes a very powerful approach. Advance post-processing techniques (i.e. Proper Orthogonal Decomposition and Dynamic Mode Decomposition) have been employed to analyze the intricate, highly turbulent flows. The impact of using fluidic injection inside a convergent-divergent nozzle for acoustic noise suppression is analyzed, first using steady-state RANS simulations. More than 250 cases are investigated for the optimal injection location and angle, amount of injected flow and operating conditions. Based on a-priori established criteria, a few optimal candidate solutions are detected from which one geometrical configuration is selected for being thoroughly investigated by using detailed LES calculations. This allows analyzing the unsteady shock pattern movement and the flow structures resulting with fluidic injec- tion. When investigating external fluidic injection configurations, some lead to a high amplitude shock associated noise, so-called screech tones. Such unsteady phenomena can be captured and explained only by using unsteady simulations. Another complex flow scenario demonstrated using LES is that of a high ve- locity jet ejected into a confined convergent-divergent ejector (i.e. a jet pump). The standing wave pattern developed in the confined channel and captured by LES, significantly alters the acoustic noise production. Steady-state methods failed to predict such events. The unsteady highly resolved simulations proved to be essential for analyzing flow and acoustics phenomena in complex problems. This becomes a very powerful approach when is used together with steady-state, low time-consuming formulations and when complemented with experimental measurements. / <p>QC 20141202</p>

A new combustor, referred to as Stagnation Point Reverse Flow (SPRF) combustor has been developed at Georgia Tech to meet increasingly stringent emission regulations. The combustor incorporates a novel design to meet the conflicting requirements of low pollution and high efficiency in both premixed and non-premixed modes. The objective of this thesis is to perform Large Eddy Simulations (LES) on this lab-scale combustor and explain the underlying physics. To achieve this, numerical simulations are performed in both the premixed and non-premixed combustion modes. The velocity field, species field, entrainment characteristics, flame structure, emissions and mixing characteristics are then analyzed. Simulations have been carried out first for a non-reactive case and the flow features in the combustor are analyzed. Next, the simulations have been extended for the premixed reactive case by employing different sub-grid scale combustion chemistry closures - Eddy Break Up (EBU), Artificially Thickened Flame (TF) and Linear Eddy Mixing (LEM) models. Only LEMLES which is an advanced scalar approach is able to accurately predict both the velocity and species field in the combustor. The results from LEM with LES (LEMLES) using a reduced chemical mechanism have been analyzed in the premixed mode. The results showed that mass entrainment occurs along the shear layer in the combustor. The entrained mass carried products into the reactant stream and provided preheating. The product entrainment enhances the reaction rates and stabilizes the flame even at very lean conditions. These products are shown to enter into the flame through local extinction zones present on the flame surface. The flame structure is further analyzed and the combustion mode is found to be primarily in thin reaction zones. The emissions in the combustor are studied using simple global mechanisms for NOx. Computations show extremely low NOx values comparable to the measured emissions. These low emissions are shown to be primarily due to the low temperatures in the combustor. LEMLES computations are also performed with detailed chemistry to capture more accurately the flame structure. The flame in the detailed chemistry case is more sensitive to strain effects and show more extinction zones very near to the injector. LEMLES approach is also used to resolve the combustion mode in the non-premixed case. The studies indicate that mixing of fuel and air close to the injector controls the combustion process. The predictions in the near field are shown to be very sensitive to the inflow conditions. Analysis shows that fuel and air mixing occurs to lean proportions in the combustor before any burning takes place. The flame structure in the non-premixed mode is very similar to the premixed mode. Along with fuel-air mixing, the products also mix with the reactants and provide the preheating effects to stabilize the flame in the downstream region of the combustor.

[pt] Uma metodologia híbrida envolvendo simulação de grandes escalas e função densidade probabilidade transportada (LES-PDF) é desenvolvida para realizar simulações de escoamentos turbulentos reativos a baixo número de Mach. Equações de transporte de massa, da quantidade de movimento e de um escalar são resolvidas em conjunto com uma equação de estado no contexto do método LES. A modelagem da turbulência é realizada pelo modelo clássico de Smagorinsky e a taxa de produção química é representada pela lei de Arrhenius, para reação de combustão única, global e irreversível. As equações de transporte são discretizadas no espaço e no tempo mediante o uso de esquemas de segunda ordem, sobre malhas cartesianas uniformes, no âmbito do método dos volumes finitos. Os efeitos da turbulência sobre a combustão na escala sub-filtro são determinados por uma abordagem lagrangeana da PDF, a qual faz uso da técnica de Monte Carlo: equações diferenciais estocásticas (SDE), equivalentes a equação de Fokker-Plank, são utilizadas para a variável de progresso da reação química. LES e PDF evoluem simultaneamente, trocando informações a cada passo de integração no tempo, de modo que o campo de velocidade filtrado, a freqüência turbulenta e o coeficiente de difusão são fornecidos por LES, enquanto o modelo PDF retorna a taxa de reação química filtrada. Devido ao elevado número de partículas empregado no modelo PDF, a paralelização do programa lagrangeano é realizada, com base na estratégia de decomposição de domínios, implementada no programa euleriano. O modelo final é usado para simular uma configuração experimental que consiste de uma chama de metano e ar, estabilizada entre escoamentos paralelos de gases queimados e gases frescos em um canal de seção transversal quadrada constante. Uma comparação detalhada entre os resultados obtidos e os dados experimentais é realizada. / [en] A hybrid Large Eddy Simulation / transported Probability Density Function (LES-PDF) computational model is developed to perform the numerical simulation of variable-density low Mach number turbulent reactive flows. Transport equations for mass, momentum, and scalars are solved together with an equation of state within the LES framework. Turbulence is modeled using the classical Smagorinsky closure whereas chemical reaction is first addressed thanks to a global single-step chemistry scheme. The governing equations are discretized using second order accuracy spatial and temporal approximations applied to uniform Cartesian meshes within a finite volume framework. The effects of subgrid scale (SGS) turbulence on the combustion processes are accounted for by means of a Lagrangian transported PDF model which is coupled with the LES solver. The PDF model relies on the use of a Monte Carlo technique: Stochastic Differential Equations (SDE), equivalent to the Fokker- Planck equations are considered for the progress variable. LES and PDF models are solved simultaneously, exchanging information at each integration time step, the velocity field, turbulence frequency and diffusion coefficient being provided by LES, whereas the PDF model returns the filtered chemical reaction rate. Parallelization of the Lagrangian solver has been performed based on the domain decomposition strategy, the same strategy being already implemented for the eulerian LES solver. The resulting computational model is used to perform the simulation of an experimental test case consisting of a CH4-air flame established between two streams of fresh and burnt pilot gases in a constant area square cross section channel. The accuracy of the numerical solutions provided by the hybrid LESPDF approach is assessed by detailed comparisons with experimental data.

La turbulence homogène et isotrope fut formalisée par Kolmogorov (1941), à l'aide d'une analyse dimensionnelle. Il parvint à démontrer que la densité spectrale de l'énergie cinétique, $E(k)$ suivait une loi en $k^{-5/3}$. Ce comportement est connu sous le nom de cascade de Kolmogorov. Dans de nombreux contexte géophysique ou astrophysiques, l'hélicité cinétique joue un rôle important. Parker (1955) a notamment démontré que l'hélicité cinétique pouvait contribuer à l'amplification d'un champ magnétique pour des écoulements conducteurs. Brissaud {it et al} (1973) ont alors tenté de déterminer l'influence que l'hélicité cinétique pouvait avoir sur les spectres d'énergie cinétique. Brissaud {it et al} (1973) suggèrent l'existence d'une cascade pour laquelle les spectres d'énergie cinétique suivent une loi en $k^{-7/3}$. Dans la première partie de ce manuscrit nous allons confirmer à l'aide de simulations numériques directes (DNS) l'existence d'une loi asymptotique en $k^{-7/3}$. Nous aurons également recourt à la décomposition en modes hélicitaires afin d'analyser de manière approfondie la physique qui régit ces écoulements. Dans de nombreux écou-le-ments géophysique ou astrophysiques, la turbulence est très forte, et une très large gamme d'échelles est impliquée. Bien que la puissance des calculateurs ait considérablement augmentée ces dernières années, il n'est toujours pas possible de simuler l'ensemble de cette gamme d'échelles pour des configurations réalistes. Une solution connue sous le nom de Large Eddy Simulations (LES) permet de réaliser des simulations de ce type d'écoulement. Concrètement, lors de la réalisation d'une LES, les grandes échelles de l'écoulement sont résolues, et les interactions entre les grandes et les petites échelles de l'écoulement sont modélisées. Divers modèles de turbulence existent déjà pour la réalisation de LES en turbulence. Néanmoins leurs limites ne sont pas toujours bien connues dans le cadre de la turbulence magnétohydrodynamique (MHD), c'est-à-dire pour les fluides conducteurs de l'électricité que l'on rencontre en géophysique ou astrophysique. Dans la seconde partie de ce manuscrit nous allons donc évaluer les performances fonctionnelles (voir Sagaut (2002)) de ces différents modèles dans des configurations correspondant à des dynamos turbulentes, c'est-à-dire à des régimes où un champ magnétique est généré par un fluide conducteur animé d'un mouvement turbulent. Nous étudierons notamment la capacité des modèles LES à reproduire les échanges énergétiques entre grandes et petites échelles. Pour ce faire, nous réaliserons plusieurs DNS, pour des écoulements non-hélicitaires (menant à des dynamos de petites échelles) et des écoulements hélicitaires (menant à des dynamos de grandes échelles). `A l'aide d'une opération de filtrage, nous calculerons les transferts sous-mailles exacts, puis les comparerons aux prédictions fournies par les modèles. Finalement nous réaliserons des LES à l'aide des différents modèles et nous les comparerons aux DNS filtrées. / Homogeneous and isotropic turbulence was first formalized by Kolmogorov (1941), through dimensional analysis. He managed to show that the spectral density of kinetic energy, $E(k)$, was following a $k^{-5/3}$ law. This behaviour is known as Kolmogorov's cascade. For many geophysical and astrophysical flow, kinetic helicity plays an important role. For instance, Parker (1955) showed that for conductive fluids such as Sun, kinetic helicity could contribute to amplify the magnetic field. Brissaud {it et al} (1973) tried to show that kinetic helicity could have an influence on the spectral density of kinetic energy. Through dimensional analysis they suggested the existence of a cascade for which the kinetic energy spectra would follow a $k^{-7/3}$ law. In the first part of this thesis we will confirm thanks to Direct Numerical Simulations (DNS) the existence of such an asymptotic limit in $k^{-7/3}$. We will also use helical decomposition to perform a deep analysis of the physics encountered within such flows. In several geophysical and astrophysical fluids, turbulence is very strong, and involves a large range of scales. Despite the strong development of computational resources the last few decades, it remains impossible to simulate this range of scales for realistic configurations. One solution is known as Large Eddy Simulations (LES). While a LES is performed, only the large scales of the flow are resolved, and the interactions between large and small scales are modeled. Several turbulence models have been developed for LES of turbulence. Nevertheless, the limitations of these models are not always well known for magnetohydrodynamic (MHD) turbulence, i.e for conductive fluids that can be encoutered in geophysics and astrophysics. In the second part of this thesis we will evaluate the functional performances (see Sagaut (2002)) of these models for several flow configurations involving turbulent dynamo action, i.e when a magnetic field is amplified though the action of a turbulent conductive fluid. In particular we will study the capabilities of LES models to reproduce energy exchanges between large and small scales. In order to do so, we will perform several DNS, for both non-helical flows (i.e leading to small scale dynamo) and helical flows (i.e leading to large scale dynamo). Thanks to a filtering operation we will compute the exact subgrid-scale transfers and compare them to the predictions given by several models. Finally we will achieve LES using subgrid-scale models and we will compare them to filtered DNS.

This thesis aims to explore the potential for using large eddy simulation (LES) as a predictive tool for gas-turbine compressor flows. Compressors present a significant challenge for the Reynolds Averaged Navier-Stokes (RANS) based CFD methods commonly used in industry. RANS models require extensive calibration to experimental data, and thus cannot be used predictively. This thesis explores how LES can offer a more predictive alternative, by exploring the sensitivity of LES to sources of uncertainty. Specifically, the importance of the numerical scheme, the Sub-Grid Scale (SGS) model, and the correct specification of inflow turbulence is examined. The sensitivity of LES to the numerical scheme is explored using the Taylor-Green vortex test case. The numerical smoothing, controlled by a user defined smoothing constant, is found to be important. To avoid tuning the numerical scheme, a locally adaptive smoothing (LAS) scheme is implemented. But, this is found to perform poorly in a forced isotropic turbulence test case, due to the intermittency of the dispersive error. A novel scheme, the LAS with windowing (LASW) scheme, is thus introduced. The LASW scheme is shown to be more suitable for predictive LES, as it does not require tuning to a known solution. The LASW scheme is used to perform LES on a compressor cascade, and results are found to be in close agreement with direct numerical simulations. Complex transition mechanisms, combining characteristics of both natural and bypass modes, are observed on the pressure surface. These mechanisms are found to be sensitive to numerical smoothing, emphasising the importance of the LASW scheme, which returns only the minimum smoothing required to prevent dispersion. On the suction surface, separation induced transition occurs. The flow here is seen to be relatively insensitive to numerical smoothing and the choice of SGS model, as long as the Smagorinsky-Lilly SGS model is not used. These findings are encouraging, as they show that, with the LASW scheme and a suitable SGS model, LES can be used predictively in compressor flows. In order to be predictive, the accurate specification of inflow conditions was shown to be just as important as the numerics. RANS models are shown to over-predict the extent of the three dimensional separation in the endwall - suction surface corner. LES is used to examine the challenges for RANS in this region. The LES shows that it is important to accurately capture the suction surface transition location, with early transition leading to a larger endwall separation. Large scale aperiodic unsteadiness is also observed in the endwall region. Additionally, turbulent anisotropy in the endwall - suction surface corner is found to be important. Adding a non-linear term to the RANS model leads to turbulent stresses that are in better agreement with the LES. This results in a stronger corner vortex which is thought to delay the corner separation. The addition of a corner fillet reduces the importance of anisotropy, thereby reducing the uncertainty in the RANS prediction.

L'écoulement de jeu est un phénomène très important dans les turbomachines. Il provient du mouvement relatif entre la pale et la paroi d'extrémité, et la différence de pression à travers la pale. L'écoulement de jeu est extrêmement complexe pour sa nature tridimensionnelle et instable, et son existence conduit à de nombreux effets défavorables, par exemple, les pertes de performance aérodynamique et les émissions de bruit. C'est pourquoi l'écoulement de jeu a motivé de nombreuses recherches expérimentales et numériques. Afin d'améliorer la compréhension du écoulement de jeu et le bruit de large bande associé, une campagne de recherche a été menée au LMFA. En ce qui concerne l'écoulement de jeu, cette campagne de recherche comprend une expérience avec des technologies de mesure avancées, un calcul zonal LES et une série de calculs RANS / URANS. L'expérience et les simulations considèrent une configuration simple de l'écoulement de jeu à un faible nombre de Mach. Les résultats expérimentaux et numériques sont analysés de façon systématique et approfondie dans la présente étude. Enfin, des efforts sont déployés pour la modélisation et la prédiction du bruit à large bande avec des résultats expérimentaux et numériques. On observe dans l'expérience un système à multiple-tourbillon, avec une tourbillon de jeu intense. Les différentes analyses sur les caractéristiques d'écoulement montrent un bon accord entre l'expérience et le ZLES dans la région du écoulement de jeu. L'approche zonale (RANS-LES) s'avère être un outil puissant pour fournir une description détaillée du écoulement de jeu, avec un coût de calcul limité. Cependant, les calculs RANS et URANS surestiment globalement la diffusion de la tourbillon. En outre, l'oscillation du tourbillon de jeu est étudiée en utilisant des champs instantanés de PIV et l'amplitude d'oscillation est évaluée. La réponse dynamique de la tourbillon de jeu est également étudiée avec URANS aux fréquences choisies. Deux modèles de prédiction du bruit en champ lointain, correspondant à deux sources acoustiques différentes, sont reformulés et mis en oeuvre avec les données de champ proche des simulations numériques. Ces prédictions sont comparées aux mesures à champ lointain. En utilisant les données ZLES, le modèle de l’écoulement de jeu sur-estime le bruit généré dans la région de jeu. Le modèle de bruit de bord de fuite est implémenté avec les données ZLES et les données RANS et fournit une très bonne prédiction dans une large bande de fréquence. / The tip-leakage flow is a common flow feature in turbomachines. It originates from the relative motion between the blade tip and the end-wall, and the pressure difference across the blade. The tip-leakage flow is extremely complex for its three-dimensional unsteady nature, and its existence leads to many unfavourable effects, such as aerodynamic performance losses and noise emissions. These issues have motivated extensive experimental and numerical researches from both aerodynamic and aeroacoustic points of view. In order to improve the understanding of the tip-leakage flow and its associated broadband noise, a research campaign has been carried out at LMFA. Regarding the tip-leakage flow, this research campaign includes an experiment with advanced measurement technologies, a zonal LES computation and a series of RANS/URANS computations. Both the experiment and the simulations consider a single-airfoil configuration at low Mach number. Experimental and numerical results are analysed systematically and thoroughly in the current study. Finally, efforts are put on the broadband noise modelling and prediction based on the experimental and numerical results. A multi-vortex system with an intense tip-leakage vortex is observed in the experiment. The various analyses of the flow characteristics show a good agreement between the experiment and the ZLES in the blade tip region. The zonal (RANS-LES) approach proves itself to be a powerful tool to provide a detailed description of the tip-leakage flow, with a limited computational cost. However, the RANS and URANS computations globally over-estimate the diffusion of the tip-leakage vortex. Furthermore, the random oscillation of the tip-leakage vortex is investigated using PIV instantaneous flow fields and the wandering amplitude is evaluated. The dynamic response of the tip-leakage vortex is also studied with URANS at selected frequencies. Two far-field noise prediction models, corresponding to two different acoustic sources, are reformulated and implemented with the near-field data from the numerical simulations. These predictions are compared to the far-field measurements. Using the ZLES data as input, the blade-tip self-noise model is found to over-estimate the noise generated in the blade-tip region. The trailing-edge noise model is implemented with the time-averaged ZLES and the RANS near-field data, and yields a very good prediction within a broad range of frequency.

Les écoulements en turbomachines (et notamment en turboréacteurs) sont caractérisés par de larges structures tourbillonnaires et de fortes intensités turbulentes. Ainsi, l’écoulement secondaire dans la région du jeu, en tête d’aube, est l’origine de pertes d’énergie, d’instabilités et de nuisances sonores. Une simulation fine de ces écoulements peut être obtenue par l’emploi de méthodes LES (Large-Eddy Simulation), qui permettent de capturer les fluctuations turbulentes majeures. Compte tenu des phénomènes rencontrés, le modèle de sous-maille SISM (shear-improved Smagorinsky model) est retenu ici. Ce modèle est local dans son écriture, et prend en compte l’influence du cisaillement moyen. Nous proposons ici deux méthodes de filtrage (locales en espace elles aussi) pour obtenir une évaluation du champ moyen requis par le modèle. Ces méthodes sont, dans un premier temps, testées sur une configuration de canal plan. L’écoulement en régime sous-critique autour d’un barreau cylindrique (Re = 4, 7 ×104) est proposé comme cas-test académique sélectif pour ces méthodes : cet écoulement présente de larges structures tourbillonnaires ainsi qu’une turbulence intense, tout comme l’écoulement de jeu. Les simulations permettent l’obtention de résultats très proches des données expérimentales. Une étude comparée des deux algorithmes d’extraction du champ moyen montre que l’adaptativité du filtrage de Kalman offre toutefois des résultats légèrement meilleurs. Enfin, l’analyse d’un écoulement de jeu par une méthode zonale est réalisée (approche LES en tête d’aube, RANS en pied). La simulation de référence obtient des résultats remarquables dans la zone de tête d’aube, en retrouvant notamment les spectres de vitesse expérimentaux. Une seconde simulation avec l’emploi d’un dispositif de contrôle par aspiration au niveau du carter montre deux conséquences principales à ce dispositif: une réduction des niveaux de turbulence aux environs de la tête d’aube, et une modification de la trajectoire du tourbillon de jeu. Celui-ci rencontre l’aube suivante dans la configuration de référence, ce qui n’est plus le cas dans la configuration avec contrôle. Ces deux observations ont une importance certaine dans la réduction des sources acoustiques. / Flows in turbomachines such as jet engines are subject to large vortical structures and strong turbulent intensities. In particular, secondary flows generated in the fan tip region result in energy losses, instabilities and noise radiation. An accurate simulation of such flows can be achieved with large-eddy simulation(LES), which reproduces the most energetic turbulent eddies. In regard of the flowphysics of highly unsteady wall-bounded flows, the SISM (shear-improved Smagorinsky model) is selected to model the sub-grid scales in the present study. This model is local and takes into account the influence of the mean shear. Two smoothing algorithms that are local in space are developed to evaluate the mean flow required by the model : an exponential averaging and an adaptative Kalman filter. These methods are first tested in a channel flow configuration. The numerical approaches are then evaluated on a relevant academic test case :the flow past a circular cylinder in the sub-critical regime (Re = 4, 7 × 104). This flow is dominated by large quasi-periodic vortical structures together with high intensity turbulent fluctuations; quite similarly but much simpler than those found in the tip gap flow. The aerodynamic as well as the acoustic results of the simulations are in very good agreement with the experimental data. A comparative study of the two smoothing algorithms for mean-flow extraction shows that the adaptability of the Kalman filtering leads to slightly better results. Finally, the study of a fan tip-gap flow is carried out with a zonal approach (LES in the tip region, RANS in the hub and midspan region). The reference simulation gives remarkable results in the blade-tip region, particularly for the velocity spectra. A second simulation with a control device by suction through the casing close to the blade leading edge shows two interesting features : a reduction of the turbulence level around the blade tip, and a modification of the tip-vortex trajectory (thus preventing impingement on the adjacent blade). These effects induce a notable reduction of the noise sources.

O presente trabalho tem como objetivo o desenvolvimento de um sistema voltado à solução de problemas relacionados à Engenharia do Vento Computacional. Para o tratamento das estruturas turbulentas, a Simulação das Grandes Escalas é empregada. Esta metodologia resolve diretamente as estruturas que governam a dinâmica local do escoamento (grandes escalas) e utiliza modelos para resolver as escalas com características mais universais (pequenas escalas). Neste estudo, os efeitos sub-malha são obtidos a partir do modelo clássico de Smagorinsky. Na análise numérica, o método dos elementos finitos é avaliado a partir da utilização de elementos hexaédricos e uma formulação baseada nas equações governantes de escoamentos quase-incompressíveis. Para reduzir o requerimento de memória computacional, esquemas explícitos para solução de sistemas de equações são empregados. O primeiro aspecto a ser abordado para o desenvolvimento do sistema proposto é a redução do tempo de processamento. Partindo do algoritmo desenvolvido por [Petry, 2002], desenvolvese um estudo a cerca de técnicas computacionais de alto desempenho visando acelerar o processamento dos problemas. Assim, apresenta-se um comparativo entre alocações estática e dinâmica de vetores e matrizes, juntamente a implementação do paralelismo de memória compartilhada utilizando diretivas OpenMP. A verificação do aumento da velocidade de processamento é desenvolvida simulando o escoamento em um domínio contendo um corpo imerso aerodinamicamente rombudo. As técnicas utilizadas permitiram a obtenção de um aumento de aproximadamente cinco vezes em relação ao código originalmente avaliado. Uma importante dificuldade na avaliação de escoamentos externos está na solução numérica de problemas advectivo-dominantes. O esquema de Taylor-Galerkin explícito-iterativo, originalmente presente no código e validado para escoamentos internos, mostrou-se inadequado para avaliação do escoamento externo proposto, apresentando perturbações no campo de pressões e não convergindo para a solução correta do problema. Estas instabilidades persistiram em uma versão alternativa desenvolvida, a qual utilizava funções de interpolação de igual ordem para solução da pressão e velocidade. Para uma análise de escoamentos não confinados, é implementado o esquema temporal de dois passos utilizando funções de interpolação para velocidade e pressão de mesma ordem. Esta configuração apresentou resultados físicos de boa qualidade e importante redução no tempo de processamento. Após a identificação da alternativa que permitiu a avaliação dos resultados sem a presença de perturbações, apresenta-se a análise do escoamento sobre um prisma quadrado bidimensional, privilegiando o monitoramento da velocidade, pressão e energia cinética total da turbulência na linha central do domínio e nas proximidades do obstáculo. Esta avaliação é efetuada em malhas com configurações uniformes e irregulares para um número de Reynolds igual a 22000. / Development of a system to solve problems related to Computational Wind Engineering is the main aim of this work. In order to treat turbulent structures, Large Eddy Simulation is employed. This methodology compute directly scales governing local flow dynamics (large eddies) and it use models to solve those with universal character (small eddies). In this study, the sub-grid effects are considered using the standard Smagorinsky model. In the numerical analysis, hexahedral finite elements are used and a formulation based on the governing equations of quasi-compressible flows. To reduce the computational memory request, explicit schemes to solve the equations system are used. In order to reduce CPU time, an algorithm developed by [Petry, 2002] is evaluated and high-performance techniques aiming to accelerate the problem solution are studied. Thus, it is showed a comparison between dynamic and static allocations of vectors and matrices associated to the implementation of shared-memory parallelization using OpenMP directives. The speed up verification is developed simulating the flow around an immersed bluff body. As a consequence of the techniques employed here, an acceleration of approximately five times with respect of the original computational code is obtained. An important difficulty in the external flow evaluation is the numerical solution of convection dominated flows. The Taylor-Galerkin explicit-iterative scheme, (originally used by the program), which was validated for confined flows, did not present good results for external flows simulations and pressure field perturbations were observed. These instabilities were persevered even in an alternative version, where interpolations functions with the same order were used to compute velocity and pressure (in the original version, constant pressure field at element level were employed). To analyze unbounded flows accurately, a two-step explicit scheme using velocity and pressure interpolation functions with the same order was implemented. This configuration presented physical results with good quality and achieve an important reduction in the processing time. After identification of the best alternative without perturbations of the pressure field, the numerical simulation of the flow around a two-dimensional square cylinder was investigated favoring velocity, pressure and total kinetic energy evaluations along the mid line of the domain and in the obstacle vicinity. These evaluations were effectuated with uniform and stretched meshes for a Reynolds number equal to 22000.