For many decades, turbulence has been the subject of extensive numerical research and experimental work. A bottleneck problem in turbulence research has been to detect and characterize the energetic, space and time-dependent structures and give a mathematical definition to each topology. This research presents a fundamental study of coherent structures, embedded in turbulent flows, by use of Proper Orthogonal Decomposition (POD). The target is to detect dominant features which contain the largest fraction of the total kinetic energy and hence contribute more to a turbulent flow. POD is proven to be a robust methodology in multivariate analysis of non-linear problems. This method also helps to obtain a low-dimensional approximation of a high-dimensional process, like a turbulent flow. This manuscript-based dissertation consists of five chapters. The first chapter starts with a brief introduction to turbulence, available simulation techniques, limitations and practical applications. Next, POD is introduced and the step-by-step approach is explained in detail. Three submitted manuscripts are presented in the subsequent chapters. Each chapter starts with introducing the study case and explaining the contribution of the study to the whole topic and also has its topic-relevant literature review. Each article consists of two parts: flow simulation and verification of the results at the onset, followed by POD analysis and reconstruction of the turbulent flow fields. For flow simulation, Large Eddy Simulation (LES) was performed to obtain databases for POD analysis. The simulations were validated by making comparison with available experimental and numerical studies. For each case, coherent topologies are characterized and the contribution of kinetic energy for each structure is determined and compared with previous literature. The first manuscript focused on investigating the large-scale dynamics in the wake of an infinite square cylinder. This case is the first step towards the targeting study case of this research, i.e. flow over rib roughened walls. The main purpose the first step is to establish a benchmark for comparison to the more complicated cases of a square cylinder with a nearby wall and flow over a rib-roughened surface. For POD analysis, the three-dimensional velocity field is obtained from LES of the flow around an infinite square cylinder at a Reynolds number of Re = 500. The POD algorithm is examined and the total energy of the flow is found to be well captured by only a small number of eigenmodes. From the energy spectrum, it is learned that each eigenmode represents a particular flow characteristic embedded in the turbulent wake, and eigenmodes with analogous characteristics can be bundled as pairs. Qualitative analysis of the dominant modes provided insight as to the spatial distribution of dominant structures in the turbulent wake. Another outcome of this chapter is to develop physical interpretations of the energetic structures by examining the temporal coefficients and tracking their life-cycle. It was observed that the paired temporal coefficients are approximately sinusoidal with similar order of magnitude and frequency and a phase shift. Lastly, it was observed that the turbulent flow field can be approximated by a linear combination of the mean flow and a finite number of spatial modes. The second manuscript analyses the influence of a solid wall on the wake dynamics of an infinite square cylinder. Different cases have been studied by changing the distance between the cylinder and the bottom wall. From the simulation results, it is learned that the value of drag and lift coefficients can be significantly affected by a nearby solid wall. From the energy decay spectrum it is observed that the energy decay rate varies for different gap ratios and accordingly a physical explanation is developed. Visualization of coherent structures for each case shows that for larger gaps, although the structures are distorted and inclined away from the wall, the travelling wave characteristic persists. Lastly, it is observed that as the gap ratio gets smaller, energetic structures originated by the wall begin to appear in the lower index modes. The last manuscript presents a numerical study of the structures in turbulent Couette flow with roughness on one wall, which as mentioned earlier, is the targeting study case of this research. Flow over both roughened and smooth surfaces was examined in a single study. Comparison was made with experiments and other numerical studies to verify the LES results. The mean velocity distribution across the channel shows that the rib roughness on the bottom wall has a strong effect on the velocity profile on the opposite wall. The energetic coherent dynamics of turbulent flow were investigated by the use of POD. The energy decay spectrum was analysed and the influence of a roughened wall and each roughness element on formation of those structures was investigated. Coherent POD modes on a spanwise sampling plane are detected. A secondary swirling motion is visualized, for the first two modes and counter-rotating cells are observed in the lower region of the channel above the rough wall for the higher modes. At the end, a quantitative analysis of the POD temporal coefficients was performed, which characterize the life-cycle of each coherent dynamic. A motivating outcome of this analysis is to decompose the time trace curves into quasi-periodic and fluctuations curves and to detect a linkage between these life cycles and physical meaning and location of each energetic pattern. At the end, in a closuring chapter, concluding remarks of this research work are presented in more detail and some potential extensions have been proposed for future researchers.

The flow characteristics in compound channels with and without vegetation on the floodplain were investigated experimentally and numerically in this thesis. Detailed measurements of velocity and boundary shear stress, using a Pitot tube and an acoustic Doppler velocimeter together with a Preston tube, were undertaken to understand the flow characteristics in compound channels. Eight no-rod cases, two emergent-rod cases and two submerged-rod cases were tested. Unsteady large eddies that occur in the shear layer were explored numerically with Large Eddy Simulation (LES) to identify its generation and its effects on the flow behaviors. Mean flow parameters were predicted using the quasi-2D model by considering the shear effect. Usirgg the data of depth-averaged velocity and boundary shear stress, the contributions of shear-generated turbulence and bed-generated turbulence to the Reynolds shear stress were identified, the apparent shear stress was calculated using the modified method of Shiono and Knight (1991) and the depth-averaged secondary current force was then obtained. Large eddies were important to the lateral momentum exchange in shallow non-vegetated compound channels and even in deep vegetated compound channels. In the compound channel with one-line rods at the floodplain edge, the secondary current forces were of opposite signs in the main channel and on the floodplain and the bed shear stress was smaller than the standard two-dimensional value of yHSo due to the vegetation effect, where y,H,So are the specific weight of water, water depth and bed slope respectively. In vegetated compound channels, the velocity patterns were different to those and the discharges were smaller than those in non-vegetated compound channels under similar relative water depth conditions. The anisotropy of turbulence was the main contribution to the generation of secondary currents in non-vegetated and vegetated compound channels, but the Reynolds stress term was more important in the vegetated compound channels. Results of cross spectra showed the mechanisms of the turbulent shear generation near the main channel-floodplain junction are due to large eddies in the non-vegetated compound channel and owing to wakes in the vegetated compound channel. LES results indicated that large eddies caused significant spatial and temporal fluctuations of velocity and water level in the compound channel and the instantaneousv alues of these flow parameters were significantly higher than the mean values. In vegetated compound channels, the flow moved from the main channel to the floodplain and from the floodplain to the main channel alternately. The characteristic frequencies of the large eddy were less than 1Hz which was consistent with the experimental data. The capability of the quasi-2D model to predict the 2D mean flow parameters in compound channels were assessed under different flow conditions and also improved by using the mean wall velocity as the boundary condition and appropriate values of the lateral gradient of the secondary current force. In the vegetated compound channels, new approaches were proposed to treat the drag force in the cases of oneline emergent rods at the floodplain edge and submerged rods on the floodplain.

In the last 10-15 years, large-eddy simulation (LES) has become well established for non-reacting flows, and several successful models have been developed for the transfer of momentum and kinetic energy to the subfilter-scales (SFS). However, for reacting flows, LES is still undergoing significant development. In particular, for many premixed combustion applications, the chemical reactions are confined to propagating surfaces that are significantly thinner than the computational grids used in practical LES. In these situations, the chemical kinetics and its interaction with the turbulence are not resolved and must be entirely modelled. There is, therefore, a need for accurate and robust physical modelling of combustion at the subfilter-scales. In this thesis, modelled transport equations for progress variable and flame surface density (FSD) were implemented and coupled to the Favre-filtered Navier-Stokes equations for a compressible reactive thermally perfect mixture. In order to reduce the computational costs and increase the resolution of simulating combusting flows, a parallel adaptive mesh (AMR) refinement finite-volume algorithm was extended and used for the prediction of turbulent premixed flames. The proposed LES methodology was applied to the numerical solution of freely propagating flames in decaying isotropic turbulent flow and Bunsen-type flames. Results for both stoichiometric and lean flames are presented. Comparisons are made between turbulent flame structure predictions for methane, propane, hydrogen fuels, and other available numerical results and experimental data. Details of subfilter-scale modelling, numerical solution scheme, computational results, and capabilities of the methodology for predicting premixed combustion processes are included in the discussions. The current study represents the first application of a full transport equation model for the FSD to LES of a laboratory-scale turbulent premixed flame. The comparisons of the LES results of this thesis to the experimental data provide strong support for the validity of the modelled transport equation for the FSD. While the LES predictions of turbulent burning rate are seemingly correct for flames lying within the wrinkled and corrugated flamelet regimes and for lower turbulence intensities, the findings cast doubt on the validity of the flamelet approximation for flames within the thin reaction zones regime.

Turbulente Austauschprozesse in der atmosphärischen Grenzschicht spielen eine Schlüsselrolle beim vertikalen Impuls-, Energie- und Stofftransport in der Erdatmosphäre. In meso- und globalskaligen Atmosphärenmodellen sind turbulente Austauschprozesse jedoch subskalig und müssen unter Verwendung geeigneter Schliessungsansätze parametrisiert werden. Hierbei spielt die Spezifikation der charakteristischen Turbulenzlängenskala in Abhängigkeit vom Stabilitätszustand der Atmosphäre eine entscheidende Rolle. Gegenwärtig verwendete Ansätze, die auf der Verwendung der turbulenten Mischungslänge für neutrale Schichtung sowie dimensionsloser Stabilitätsfunktionen basieren, zeigen vor allem Defizite im oberen Bereich der konvektiven Grenzschicht sowie in der Entrainmentzone, wo starke vertikale Gradienten auftreten. In der vorliegenden Arbeit wurden hochaufgelöste dreidimensionale Grobstruktursimulationen der trockenen und feuchten Grenzschicht für ein weites Spektrum von Labilitätsbedingungen durchgeführt. Erste und zweite Momente atmosphärischer Strömungsvariablen wurden aus den simulierten hydro- und thermodynamischen Feldern berechnet und diskutiert. Die Spektraleigenschaften turbulenter Fluktuationen der Strömungsvariablen, das raumzeitliche Verhalten kohärenter Strukturen sowie charakteristische Turbulenzlängenskalen wurden abgeleitet. Eine Verifizierung der charakteristischen Turbulenzlängenskalen erfolgte durch Vergleich mit Ergebnissen früherer numerischer Simulationen, mit Turbulenzmessungen in der atmosphärischen Grenzschicht sowie mit Laborexperimenten. Mit Hilfe der nichtlinearen Datenmodellierung wurden leicht verwendbare Approximationen der charakteristischen Turbulenzlängenskalen abgeleitet und deren statistische Signifikanz diskutiert. Unter Verwendung dieser Approximationen wurde ein existierendes Parametrisierungsmodell revidiert und mit Hilfe von Grobstruktursimulationen verifiziert. Desweiteren wurde der Einfluß der turbulenten Mischungslänge auf die Prognose mesokaliger Felder untersucht. Hierzu wurde mit dem Lokal-Modell des Deutschen Wetterdienstes eine entsprechende Sensitivitätsstudie durchgeführt. Anhand von Satellitendaten und Analysedaten aus der 4D-Datenassimilation wurden die Simulationsergebnisse verifiziert.

The Large Eddy Simulation approach is employed to predict the flow physics and heat transfer characteristics of a film-cooling problem that is formed from the interaction of a coolant jet with a hot mainstream flow. The film-cooling technique is used to protect turbine blades from thermal failure, allowing the gas inlet temperature to be increased beyond the failure temperature of the turbine blade material in order to enhance the efficiency of gas turbine engines. A coolant fluid is injected into the hot mainstream through several rows of injection holes placed on the surface of a gas turbine blade in order to form a protective coolant film layer on the blade surface. However, due to the complex, unsteady and three-dimensional interactions between the coolant and the hot gases, it is difficult to achieve the desired cooling performance. Understanding of this complex flow and heat transfer process will be helpful in designing more efficiently cooled rotor blades. A comprehensive numerical investigation of a rotating film-cooling performance under different conditions is conducted in this thesis, including film-cooling on a flat surface and film-cooling on a rotating gas turbine blade. The flow-governing equations are discretised based on the finite-volumes method and then solved iteratively using the well-known SIMPLE and PISO algorithms. An in-house FORTRAN code has been developed to investigate the flat plate film-cooling configuration, while the gas turbine blade geometry has been simulated using the STAR-CCM+ CFD commercial code. The first goal of the present thesis is to investigate the physics of the flow and heat transfer, which occurs during film-cooling from a standard film hole configuration. Film-cooling performance is analysed by looking at the distribution of flow and thermal fields downstream of the film holes. The predicted mean velocity profiles and spanwise-averaged film-cooling effectiveness are compared with experimental data in order to validate the reliability of the LES technique. Comparison of adiabatic film-cooling effectiveness with experiments shows excellent agreement for the local and spanwise-averaged film-cooling effectiveness, confirming the correct prediction of the film-cooling behaviour. The film coverage and film-cooling effectiveness distributions are presented along with discussions of the influence of blowing ratio and rotation number. Overall, it was found that both rotation number and blowing ratio play significant roles in determining the film-cooling effectiveness distributions. The second goal is to investigate the impact of innovative anti-vortex holes on the film-cooling performance. The anti-vortex hole design counteracts the detrimental kidney vorticity associated with the main hole, allowing coolant to remain attached to the blade surface. Thus, the new design significantly improves the film-cooling performance compared to the standard hole arrangement, particularly at high blowing ratios. The anti-vortex hole technique is unique in that it requires only readily machinable round holes, unlike shaped film-cooling holes and other advanced concepts. The effects of blowing ratio and the positions of the anti-vortex side holes on the physics of the hot mainstream-coolant interaction in a film-cooled turbine blade are also investigated. The results also indicate that the side holes of the anti-vortex design promote the interaction between the vortical structures; therefore, the film coverage contours reveal an improvement in the lateral spreading of the coolant jet.

O presente trabalho tem por objetivo estudar as Interações Turbulência-Radiação em um escoamento não reativo para meios participantes. Estas interações caracterizam-se por um complexo fenômeno transiente, devido à combinação de dois fenômenos, unindo as características das flutuações da turbulência e da elevada não linearidade do fenômeno da radiação térmica. O estudo consiste em análise numérica do problema por dinâmica de fluidos computacional, através da utilização do Fire Dynamics Simulator (FDS), um software Open-Source, na qual a modelagem da turbulência é feita através da Simulação de Grandes Escalas. Como se trata de um software novo, bem como sendo introduzido no grupo de pesquisa, primeiramente é realizada a simulação de um caso benchmark para verificação e avaliação da formulação numérica. A análise do TRI é realizada em um problema proposto baseado em trocadores de calor reais utilizados em máquinas térmicas, como por exemplo, geradores de vapor ou coletores de escapamento de motores, envolvendo transferência combinada de convecção forçada e radiação térmica. A metodologia de avaliação consiste em comparar o fluxo radiante médio nas fronteiras obtido através da simulação transiente e compará-lo com o fluxo obtido por meio do campo médio temporal de temperaturas. São avaliadas a influência da intensidade de turbulência na entrada do escoamento, assim como a da espessura óptica, ambos relevantes para os efeitos do TRI. Conforme descrito pela literatura, neste tipo de problema as interações podem ser negligenciadas, confirmando os resultados obtidos, da ordem de 2% para o fluxo radiante. / This dissertation has the objective of analyzing the Turbulence-Radiation Interaction for a non-reactive flow with a participating media. These interactions are characterized by complex transient effects, due to the combination of two phenomena, coupling the scalar fluctuations of the turbulence and the highly non-linearity of thermal radiation. The study consists in a numerical analysis through Computational Fluid Dynamics, using the Fire Dynamics Simulator (FDS), an Open-Source software, which employs the Large Eddy Simulation method. Because the software is under development and new in the research group, it will be performed the simulation of a benchmark case for verification and evaluation of the numerical methodology. The TRI analysis will be performed in a proposed problem, based on real heat exchangers, as an example, steam generators or exhaust manifold of combustion engines, involving combined heat transfer between forced convection and radiative heat transfer. The methodology consists in evaluating the radiative mean heat flux obtained by the transient simulation and compare it with the flux obtained with the time-averaged temperature field. It will be evaluated the influence of the turbulence intensity at the inlet and the optical thickness, both very important for the TRI effects. According to the literature, in this case the TRI effects could be neglected, confirming the obtained results, around 2% for the radiative heat flux.

abstract: Next generation gas turbines will be required to produce low concentrations of pollutants such as oxides of nitrogen (NOx), carbon monoxide (CO), and soot. In order to design gas turbines which produce lower emissions it is essential to have computational tools to help designers. Over the past few decades, computational fluid dynamics (CFD) has played a key role in the design of turbomachinary and will be heavily relied upon for the design of future components. In order to design components with the least amount of experimental rig testing, the ensemble of submodels used in simulations must be known to accurately predict the component's performance. The present work aims to validate a CFD model used for a reverse flow, rich-burn, quick quench, lean-burn combustor being developed at Honeywell. Initially, simulations are performed to establish a baseline which will help to assess impact to combustor performance made by changing CFD models. Rig test data from Honeywell is compared to these baseline simulation results. Reynolds averaged Navier-Stokes (RANS) and Large Eddy Simulation (LES) turbulence models are both used with the presumption that the LES turbulence model will better predict combustor performance. One specific model, the fuel spray model, is evaluated next. Experimental data of the fuel spray in an isolated environment is used to evaluate models for the fuel spray and a new, simpler approach for inputting the spray boundary conditions (BC) in the combustor is developed. The combustor is simulated once more to evaluate changes from the new fuel spray boundary conditions. This CFD model is then used in a predictive simulation of eight other combustor configurations. All computer simulations in this work were preformed with the commercial CFD software ANSYS FLUENT. NOx pollutant emissions are predicted reasonably well across the range of configurations tested using the RANS turbulence model. However, in LES, significant under predictions are seen. Causes of the under prediction in NOx concentrations are investigated. Temperature metrics at the exit of the combustor, however, are seen to be better predicted with LES. / Dissertation/Thesis / M.S. Mechanical Engineering 2012

Ce travail a pour but de comparer les approches de simulation numérique des grandes échelles explicite (LES) et implicite (ILES) pour un jet turbulent non-réactif ou réactif d’hydrogène à grande vitesse dans un co-courant d’air, typique d’un super-statoréacteur. La résolution des calculs va de 32 × 32 × 128 à 256 × 256 × 1024, à l’aide d’un schéma WENO d’ordre 5. Les LES explicites emploient les modèles sous-maille de Smagorinsky et de Fonction de Structure Sélective, associés au transport moléculaire. Les LES implicites sont réalisées avec et sans modèle de transport moléculaire, en résolvant les équations de Navier- Stokes ou d’Euler. Dans le cas non-réactif, le modèle de Smagorinsky est trop dissipatif. Le modèle de Fonction de Structure Sélective améliore les résultats, sans faire mieux que l’approche ILES quelle que soit la résolution. Dans le cas réactif, une coupure physique visqueuse est indispensable pour fixer une épaisseur à la flamme, et assurer la convergence en maillage de l’approche ILES. On montre aussi que les résultats LES/ILES sont moins sensibles aux conditions d’injection que ceux de l’approche RANS. Le premier chapitre est une introduction générale au contexte de l’étude. Au second chapitre, on rappelle les équations générales pour un écoulement réactif et on détaille les modèles thermodynamique et de transport retenus. Au troisième chapitre, les équations de la LES et les modèles sous-maille sont présentés. On examine également quelques propriétés du schéma numérique. Le chapitre 4 est consacré à la méthode numérique et au code de calcul. Enfin, on présente les cas-tests et on discute les résultats au chapitre 5. / This work is intended to compare Large Eddy Simulation and Implicit Large Eddy Simulation (LES and ILES) for a turbulent, non-reacting or reacting high speed H2 jet in co-flowing air, typical of scramjet engines. Numerical simulations are performed at resolutions ranging from 32 × 32 × 128 to 256 × 256 × 1024, using a 5th order WENO scheme. Physical LES are carried out with the Smagorinsky and the Selective Structure Function models associated to molecular diffusion. Implicit LES are performed with and without molecular diffusion, by solving either the Navier-Stokes or the Euler equations. In the nonreacting case, the Smagorinsky model is too dissipative. The Selective Structure Function leads to better results, but does not show any superiority compared to ILES, whatever the grid resolution. In the reacting case, a molecular viscous cut-off in the simulation is mandatory to set a physical width for the reaction zone in the ILES approach, hence to achieve grid-convergence. It is also found that LES/ILES are less sensitive to the inlet conditions than the RANS approach. The first chapter is an introduction to the context of this study. In the second chapter, the governing equations for multispecies reacting flows are presented, with emphasis on the thermodynamic and transport models. In the third chapter, physical LES equations and explicit sub-grid modeling strategies are detailed. Some properties of the numerical scheme are also investigated. In chapter four, the numerical scheme and some aspects of the solver are explained. Finally, non-reacting and reacting numerical experiments are presented and the results are discussed.

Un des principaux enjeux de l'industrie aéronautique est la recherche du moteur au meilleur rendement possible, pour satisfaire des contraintes économiques, techniques et environnementales. Les turbomachines bénéficient d'un constant perfectionnement depuis plus de 60 ans, et cette technologie semble avoir atteint un plateau. Une rupture technologique est aujourd'hui nécessaire, comme la combustion à volume constant (CVC). Le gain attendu est suffisant pour tenter de remplacer les systèmes actuels où la combustion se fait à pression constante. La combustion à isovolume fait appel à des mécanismes encore rarement maitrisés dans le contexte aéronautique. Sa compréhension passe par des expérimentations et des modèles théoriques et numériques. L’objectif de cette thèse est de développer une théorie et un outil de simulation LES (Large Eddy Simulation) appliqué au cas du concept ‘thermoréacteur’. Ainsi, la première étape a consisté à mettre en place un outil de simulation 0D traduisant l’évolution d’un cycle moteur de type CVC (Combustion à Volume Constant). Certains modèles utilisés dans cet outil 0D sont basés sur des corrélations expérimentales. D'autres présentent des paramètres à déterminer à partir de simulations numériques. La simulation 3D d’un système de type CVC est envisageable aujourd’hui grâce aux progrès récents des méthodes LES. Ainsi, des simulations du thermoréacteur ont pu être réalisées, et confrontées aux résultats expérimentaux obtenus au laboratoire Pprime sur trois points de fonctionnement. Les variabilités cycle à cycle observées expérimentalement ont été analysées dans les calculs LES. Les vitesses importantes au niveau de l'allumage et le taux de résidus du cycle précédent semblent être les principaux facteurs à l'origine de ces variations cycle à cycle. / A major challenge for the aircraft industry is to improve engine efficiency and to reduce pollutant emissions for economic, technical and environmental reasons. Aeronautical gas turbines have enjoyed a constant improvement for more than 60 years. This technology seems to have reached such efficiency levels that a technological breakthrough is necessary. Constant Volume Combustion (CVC) offers significant gain in consumption and could replace classical constant pressure combustion technologies, currently used in aeronautical engines. Mechanisms involved in isovolume combustion are not accurately controlled in the context of aeronautical chambers. Experimental, theoretical and numerical studies should provide a better understanding of CVC devices. The objective of this thesis is to develop simulation tools to study the thermoreacteur concept. First, a zero-dimensional (0D) simulation tool is developed to describe the evolution of a CVC cycle. Models based on experimental correlations are used to build the 0D tool. Parameters have to be determined from numerical simulations. Today, the 3D simulation of a CVC system is possible thanks to the recent progress of the LES (Large Eddy Simulation) methods developed at CERFACS. Simulations of the thermoreacteur concept have been carried out, and compared to experimental results obtained at the Pprime laboratory. Three operating points have been calculated. The main conclusion is the existence of significant cyclic variations which are observed in the experiment and analyzed in the LES: the local flow velocity at spark timing and the level of residuals gases are the major factors leading to cyclic variations.

Le moteur à allumage commandé downsizé, couplé à une stratégie d’injection directe, est l’une des solutions privilégiées par les constructeurs automobiles afin de réduire les émissions polluantes et d’augmenter le rendement. Toutefois, l’augmentation de la pression d’injection visant à favoriser l’atomisation du spray et donc l’homogénéité du mélange peut engendrer une forte interaction entre le spray et les parois de la chambre de combustion. Cette interaction est à l’origine d’hétérogénéités locales susceptibles d’altérer la combustion. Du fait de son caractère instationnaire, l’interaction spray/paroi (formation et évaporation d’un film liquide) et plus généralement la préparation du mélange en moteur à injection directe essence sont des phénomènes difficiles à analyser expérimentalement. En effet, un moteur muni d’accès optiques ne peut pas fonctionner dans les conditions thermodynamiques réelles (pression, température...). Dans ce contexte, la modélisation et plus particulièrement la Simulation aux Grandes Echelles (“Large Eddy Simulation” LES) est un moyen d’analyse complémentaire et indispensable. L’objectif de cette thèse est de développer les modèles physiques nécessaires à la description de la phase liquide avec une approche Euler-Lagrange pour la simulation dans les moteurs à piston. Dans un premier temps, une modélisation des caractéristiques physiques du spray en sortie d’injecteur, nommée GDI, est proposée et validée par comparaison avec des mesures expérimentales. Les résultats montrent la capacité du modèle GDI à reproduire la dynamique générale d’un spray pour deux types d’injecteurs multi-trous. Dans un deuxième temps, deux modèles sont développés pour traiter respectivement l’interaction entre le spray et les parois et l’évaporation du film liquide. Les premières validations de ces modèles sont faites sur des expériences académiques dédiées, permettant des comparaisons précises avec les mesures. Finalement deux configurations moteur sont simulées. La première, sans combustion, permet d’évaluer l’impact d’une modélisation fine de l’interaction spray/paroi par rapport à une approche simplifiée. Les résultats montrent que la prise en compte de la formation et de l’évaporation du film liquide modifie significativement la formation du mélange, notamment le champ de richesse au Point Mort Haut. La seconde est utilisée pour analyser l’impact de la phase liquide sur le mélange et la combustion. Ces calculs sont comparés à des calculs réalisés sans injection liquide et à des mesures expérimentales. Les résultats mettent en évidence que les stratifications de richesse et de température, causées par l’évaporation du liquide, ont un effet de plissement sur la flamme et diminuent sa vitesse de propagation. / Downsized spark ignition engines coupled with a direct injection strategy, are more and more attractive for car manufacturers in order to reduce pollutant emissions and increase efficiency. However, the high pressure levels used to promote spray atomization and consequently mixing can generate a strong interaction between the spray and the combustion chamber walls. The combustion process may be affected by local heterogeneities caused by this interaction. Spray/walls interaction (formation and evaporation of the liquid film) and mixture preparation are unsteady phenomena, explaining why their experimental studies are limited. In fact, it is difficult to reproduce the thermodynamic conditions (pressure, temperature...) representative of an engine with optical accesses. In this context, numerical simulation, and in particular Large Eddy Simulation (LES) is a complementary mean of analysis. This work aims at developing the necessary models for the two-phase combustion simulation for engines, using an Euler-Lagrange approach. First, a modeling of the spray physics downstream to the injector exit is proposed and validated by comparison with experimental data. Second, two models are proposed and implemented to adress respectively the spray/wall interaction and the liquid film evaporation. These models are first validated on dedicated academics experiments, allowing an accurate comparison with experimental data. Then, two engine configurations are simulated. The first one, without combustion, allows the evaluation of an accurate spray/wall interaction modeling in comparison with a simplified approach. Results show that accounting for the formation and evaporation of the liquid film has a significant impact on the fuel/air mixing, especially on the equivalence ratio distribution at the Top Dead Center. The second one is used to analyze the impact of liquid on the mixing and the combustion. The simulations are compared to experiments data and to simulations assuming a perfect gaseous mixing (without liquid injection). Results show that the temperature and equivalence ratio heterogeneities, created by the liquid evaporation, have a wrinkling effect on the flame and reduce its propagation speed.