Dû aux efforts apportés à la réduction des émissions de NOx dans des chambres de combustion aéronautiques il y a une tendance récente vers des systèmes à combustion pauvre. Cela résulte dans l'apparition de nouveaux types d'injecteur qui sont caractérisés par une complexité géométrique accrue et par des nouvelles stratégies pour l'injection du carburant liquide, comme des systèmes multi-point. Les deux éléments créent des exigences supplémentaires pour des outils de simulation numériques. La simulation à grandes échelles (SGE ou LES en anglais) est aujourd’hui considérée comme la méthode la plus prometteuse pour capturer les phénomènes d'écoulement complexes qui apparaissent dans une telle application. Dans le présent travail, deux sujets principaux sont abordés : Le premier est le traitement de la paroi ce qui nécessite une modélisation qui reste délicate en SGE, en particulier dans des géométries complexes. Une nouvelle méthode d'implementation pour des lois de paroi est proposée. Une étude dans une géométrie réaliste démontre que la nouvelle formulation donne de meilleurs résultats comparé à l’implémentation classique. Ensuite, la capacité d'une approche SGE typique (utilisant des lois de paroi) de prédire la perte de charge dans une géométrie représentative est analysée et des sources d'erreur sont identifiés. Le deuxième sujet est la simulation du carburant liquide dans une chambre de combustion. Avec des méthodes Eulériennes et Lagrangiennes, deux approches sont disponibles pour cette tâche. La méthode Eulérienne considère un spray de gouttelettes comme un milieu continu pour lequel on peut écrire des équations de transport. Dans la formulation Lagrangienne, des gouttes individuelles sont suivies ce qui mène à des équations simples. D’autre part, sur le plan numérique, le grand nombre de gouttes à traiter peut s’avérer délicat. La comparaison des deux méthodes sous conditions identiques (solveur gazeux, modèles physiques) est un aspect central du présent travail. Les phénomènes les plus importants dans ce contexte sont l'évaporation ainsi que le problème d'injection d'un jet liquide dans un écoulement gazeux transverse ce qui correspond à une version simplifiée d’un système multi-point. Le cas d'application final est la configuration d’un seul injecteur aéronautique, monté dans un banc d'essai expérimental. Ceci permet d'appliquer de manière simultanée tous les développements préliminaires de ce travail. L'écoulement considéré est non-réactif mais à part cela il correspond au régime ralenti d'un moteur d'avion. Dû aux conditions préchauffées, le spray issu du système d'injection multi-point s'évapore dans la chambre. Cet écoulement est simulé utilisant les approches Eulériennes et Lagrangiennes et les résultats sont comparés aux données expérimentales. / Due to efforts to reduce NOx emissions of aeronautical combustors, there is a recent trend towards lean combustion technologies. This results in novel injector designs, which are characterized by increased geometrical complexity and new injection strategies for the liquid fuel, such as multipoint systems. Both elements create additional challenges for numerical simulation tools. Large-Eddy simulation (LES) is regarded as the most promising method to capture complex flow phenomena in such an application. In the present work, two main areas of interest are considered: The first is wall modeling, which remains a challenging field in LES, in particular for complex geometries. A new implementation method for wall functions that uses a no-slip condition at the wall is proposed. It is shown that in a realistic burner geometry the new formulation yields improved results compared to a classical implementation. Furthermore, the capability of a typical LES with wall models to predict the pressure drop in a representative geometry is assessed and sources of error are identified. The second topic is the simulation of liquid fuel in a combustor. With Eulerian and Lagrangian methods, two different approaches are available for this task. The Eulerian approach considers a droplet spray as a continuum for which transport equations can be formulated. In the Lagrangian formulation, individual droplets are tracked, which leads to a simple formulation but can be challenging in terms of numerics due to the large number of particles to be treated. The comparison of these methods under identical conditions (gaseous flow solver, physical models) is a central aspect of the present work. The most important phenomena that are studied in view of the final application are evaporation and the problem of transverse liquid jets in a gaseous crossflow as a simplified representation of a multipoint system. The final application case is the configuration of a single aeronautical injector mounted in an experimental test bench. It allows to simultaneously apply all preliminary developments. The flow considered is non-reactive but otherwise corresponds to a partial load regime in an aeroengine Due to the pre-heated conditions, the spray issued by the multi-point injection undergoes evaporation. This flow is simulated using Eulerian and Lagrangian methods and the results are compared to experimental data.

Trapped Vortex Combustor (TVC) is a relatively new concept for potential use in gas turbine engines addressing ever increasing demands of high efficiency, low emissions, low pressure drop, and improved pattern factor. This concept holds promise for future because of its inherent advantages over conventional swirl-stabilized combustors. The main difference between TVC and a conventional gas turbine combustor is in the way combustion is stabilized. In conventional combustors, flame is stabilized because of formation of toroidal flow pattern in the primary zone due to interaction between incoming swirling air and fuel flow. On the other hand, in TVC, there is a physical cavity in the wall of combustor with continuous injection of air and fuel leading to stable and sustained combustion. Past work related to TVC has focussed on use of two cavities in the combustor liner. In the present study, a single cavity combustor concept is evaluated through simulation and experiments for applications requiring compact combustors such as Unmanned Aerial Vehicles (UAVs) and cruise missiles. In the present work, numerical simulations were initially performed on a planar, rectangular single-cavity geometry to assess sensitivity of various parameters and to design a single-cavity TVC test rig. A water-cooled, modular, atmospheric pressure TVC test rig is designed and fabricated for reacting and non-reacting flow experiments. The unique features of this rig consist of a continuously variable length-to-depth ratio (L/D) of the cavity and optical access through quartz plates provided on three sides for visualization. Flame stabilization in the single cavity TVC was successfully achieved with methane as fuel, and the range of flow conditions for stable operation were identified. From these, a few cases were selected for detailed experimentation. Reacting flow experiments for the selected cases indicated that reducing L/D ratio and increasing cavity-air velocity favour stable combustion. The pressure drop across the single-cavity TVC is observed to be lower as compared to conventional combustors. Temperatures are measured at the exit using thermocouples and corrected for radiative losses. Species concentrations are measured at the exit using an exhaust gas analyzer. The combustion efficiency is observed to be around 98-99% and the pattern factor is observed to be in the range of 0.08 to 0.13. High-speed imaging made possible by the optical access indicates that the overall combustion is fairly steady, and there is no major vortex shedding downstream. This enabled steady-state simulations to be performed for the selected cases. Insight from simulations has highlighted the importance of air and fuel injection strategies in the cavity. From a mixing and combustion efficiency standpoint, it is desirable to have a cavity vortex that is anti-clockwise. However, the natural tendency for flow over a cavity is to form a vortex that is clockwise. The tendency to blow-out at higher inlet flow velocities is thought to be because of these two opposing effects. This interaction helps improve mixing, however leads to poor flame stability unless cavity-air velocity is strong enough to support a strong anti-clockwise vortex in the cavity. This basic understating of cavity flow dynamics can be used for further design improvements in future to improve flame stability at higher inlet flow velocities and eventually lead to the development of a practical combustor.

Une méthode de frontières immergées est développée pour la simulation d’écoulements compressibles et validée au travers de cas-tests spécifiques (réflexion d’ondes acoustiques et quantification de la conservation de la masse dans des canaux inclinés). La simulation aux grandes échelles (LES) d’une cavité transsonique est ensuite présentée. Le bouclage aéro-acoustique, très sensible aux conditions aux limites, est reproduit avec précision par la LES dans le cas où les parois sont immergées dans un maillage structurée. La comparaison des stratégies de modélisation de sous-maille pour cet écoulement transsonique et l’adaptation des filtres en présence de frontières immergées sont également discutées. Le rôle, souvent sous-estimé, du schéma de viscosité artificiel, est quantifié.Dans la dernière partie du manuscrit, des études sont réalisées pour aider au dimensionnement d’un nouveau concept de chambre de combustion où la flamme est stabilisée par la recirculation de gaz brûlés dans une cavité (chambre TVC pour Trapped Vortex Combustor). La modélisation de la combustion turbulente est basée sur une chimie tabulée, couplée à une fonction densité de probabilité présumée (PCM-FPI). L’étude de la dynamique de la flamme est réalisée pour diverses conditions de fonctionnement (débit de l’écoulement principal et présence ou non d’un swirl). Les spécificités de mise en œuvre de la simulation d’un écoulement de ce type sont discutées et un soin particulier est apporté au traitement de la condition de sortie, qui constitue un point sensible de la chaîne de modélisation. Les phénomènes d’instabilités et de retour de la flamme sont mis en évidence ainsi que les modifications à apporter au dispositif afin de minimiser ces effets. L’existence d’un cycle limite acoustique est souligné et une formule permettant d’anticiper le niveau des fluctuations de pression est proposée et validée. Une correction au modèle PCM-FPI est présentée afin de préserver la vitesse de flamme et d’assurer une reproduction plus précise de la dynamique de flamme. / An immersed boundary method has been developed for the simulation of compressible flow and validated with reference test cases (pressure wave reflection and quantification of mass conservation for various inclined channels). Large Eddy Simulation (LES) of a transonic cavity is then presented. The aeroacoustic feedback loop, which is highly sensitive to the boundary conditions, was accurately reproduced where the walls are immersed inside a structured grid. The comparison between the modeling approaches for this transonic flow and the correction of the filtering operation near immersed boundaries are also discussed. The often underestimated role of the numerical artificial dissipation is also quantified.In the last part of this manuscript, many studies are realized to help in the design of a new combustion chamber for Trapped Vortex Combustor (TVC). The turbulent combustion model is based on tabulated chemistry and a presumed probability density function (PCM-FPI) method.The flame dynamics is studied for various operating conditions (flowrate of the main flow and presence of swirl motion). Details concerning the realization of such a flow are discussed and special care is taken for the treatment of the most sensitive outlet boundary condition. The phenomena of combustion instabilities and of flame backflow are highlighted along with the modifications to be made for the device to minimize these effects. The existence of a acoustic limit cycle is emphasized and a formula is proposed and validated to anticipate the level of pressure fluctuations. Finally a correction to the PCM-FPI model is suggested to preserve the flame front speed and to ensure a more accurate description of the flame dynamics.

This thesis aims at improving our knowledge on swirl combustors. The work presented here is based on Large Eddy Simulations (LES) coupled to an advanced combustion model: the Conditional Moment Closure (CMC). Numerical predictions have been systematically compared and validated with detailed experimental datasets. In order to analyze further the physics underlying the large numerical datasets, Proper Orthogonal Decomposition (POD) has also been used throughout the thesis. Various aspects of the aerodynamics of swirling flames are investigated, such as precession or vortex formation caused by flow oscillations, as well as various combustion aspects such as localized extinctions and flame lift-off. All the above affect flame stabilization in different ways and are explored through focused simulations. The first study investigates isothermal air flows behind an enclosed bluff body, with the incoming flow being pulsated. These flows have strong similarities to flows found in combustors experiencing self-excited oscillations and can therefore be considered as canonical problems. At high enough forcing frequencies, double ring vortices are shed from the air pipe exit. Various harmonics of the pulsating frequency are observed in the spectra and their relation with the vortex shedding is investigated through POD. The second study explores the structure of the Delft III piloted turbulent non-premixed flame. The simple configuration allows to analyze further key combustion aspects of combustors, with further insights provided on the dynamics of localized extinctions and re-ignition, as well as the pollutants emissions. The third study presents a comprehensive analysis of the aerodynamics of swirl flows based on the TECFLAM confined non-premixed S09c configuration. A periodic component inside the air inlet pipe and around the central bluff body is observed, for both the inert and reactive flows. POD shows that these flow oscillations are due to single and double helical vortices, similar to Precessing Vortex Cores (PVC), that develop inside the air inlet pipe and whose axes rotate around the burner. The combustion process is found to affect the swirl flow aerodynamics. Finally, the fourth study investigates the TECFLAM configuration again, but here attention is given to the flame lift-off evident in experiments and reproduced by the LES-CMC formulation. The stabilization process and the pollutants emission of the flame are investigated in detail.