Combustion in cryogenic engines is a complex phenomenon, involving either liquid or supercritical fluids at high pressure, strong and fast oxidation chemistry, and high turbulence intensity. Due to extreme operating conditions, a particularly critical issue in rocket engine is wall heat transfer which requires efficient cooling of the combustor walls. The concern goes beyond material resistance: heat fluxes extracted through the chamber walls may be reused to reduce ergol mass or increase the power of the engine. In expander-type engine cycle, this is even more important since the heat extracted by the cooling system is used to drive the turbo-pumps that feed the chamber in fuel and oxidizer. The design of rocket combustors requires therefore an accurate prediction of wall heat flux. To understand and control the physics at play in such combustor, the Large Eddy Simulation (LES) approach is an efficient and reliable numerical tool. In this thesis work, the objective is to predict wall fluxes in a subcritical rocket engine configuration by means of LES. In such condition, ergols may be in their liquid state and it is necessary to model liquid jet atomization, dispersion and evaporation.The physics that have to be treated in such engine are: highly turbulent reactive flow, liquid jet atomization, fast and strong kinetic chemistry and finally important wall heat fluxes. This work first focuses on several modeling aspects that are needed to perform the target simulations. H2/O2 flames are driven by a very fast chemistry, modeled with a reduced mechanism validated on academic configurations for a large range of operating conditions in laminar pre- mixed and non-premixed flames. To form the spray issued from the atomization of liquid oxygen (LOx) an injection model is proposed based on empirical correlations. Finally, a wall law is employed to recover the wall fluxes without resolving directly the boundary layer. It has been specifically developed for important temperature gradients at the wall and validated on turbulent channel configurations by comparison with wall resolved LES. The above models are then applied first to the simulation of the CONFORTH sub-scale thrust chamber. This configuration studied on the MASCOTTE test facility (ONERA) has been measured in terms of wall temperature and heat flux. The LES shows a good agreement compared to experiment, which demonstrates the capability of LES to predict heat fluxes in rocket combustion chambers. Finally, the JAXA experiment conducted at JAXA/Kakuda space center to observe heat transfer enhancement brought by longitudinal ribs along the chamber inner walls is also simulated with the same methodology. Temperature and wall fluxes measured with smooth walls and ribbed walls are well recovered by LES. This confirms that the LES methodology proposed in this work is able to handle wall fluxes in complex geometries for rocket operating conditions.

La combustion cryogénique dans les moteurs de fusée dits à propulsion liquide utilise généralement un couple d'ergols, le plus couramment composé d'hydrogène/oxygène (H2/O2). Privilégiée pour le fort pouvoir calorifique du dihydrogène, cette combustion à haute pression, induit des températures de fonctionnement très élevées et nécessite l'intégration d'un système de refroidissement. La prédiction des flux thermiques aux parois est donc un élément essentiel de la conception d'une chambre de combustion de moteur fusée. Ces flux sont le résultat d'écoulements fortement turbulents, compressibles, avec une cinétique chimique violente induisant de forts gradients d'espèces et de température. La simulation de ces phénomènes nécessite des approches spécifiques telles que la Simulation aux Grandes Echelles (SGE) qui réalise un très bon compromis entre précision et coût de calcul. Cette thèse a ainsi pour objectif la simulation par SGE des transferts de chaleur aux parois dans les chambres de combustion de moteurs fusée opérant en régime sous-critique. Le régime sous-critique implique un état liquide pour un des ergols, dont il faut traiter l'injection et l'atomisation. Dans un premier temps ce travail s'intéresse à plusieurs éléments de modélisation nécessaire pour réaliser les simulations visées. Le comportement des flammes H2/O2 est décrit par un schéma cinétique réduit et validé sur des configurations académiques. La prédictivité de ce schéma est évaluée sur une large gamme de fonctionnement dans des conditions représentatives des moteurs fusée. La simulation de l'injection de l'oxygène liquide (LOx) est un autre point critique qui nécessite de décrire l'atomisation et la phase dispersée ainsi que son couplage avec la phase gazeuse. La déstabilisation et l'atomisation primaire du jet liquide, trop complexe à simuler en SGE 3D, sont omises ici pour injecter directement un spray paramétré grâce à des corrélations empiriques. Enfin, la prédiction des flux thermiques utilise un modèle de loi de paroi spécifiquement dédiée aux écoulements à fort gradient de température. Cette loi de paroi est validée sur des configurations de canaux turbulents par comparaison avec des simulations avec résolution directe de la couche limite. La méthodologie basée sur les modèles développés est ensuite employée pour la simulation d'une chambre de combustion représentative du fonctionnement des moteurs cryogéniques. Il s'agit de la configuration CONFORTH testée sur le banc MASCOTTE (ONERA) et pour laquelle des mesures de température de paroi et de flux thermiques sont disponibles. Les résultats des SGE montrent un bon accord avec l'expérience et démontrent la capacité de la SGE à prédire les flux thermiques dans une chambre de combustion de moteur fusée. Enfin, dans un dernier chapitre ce travail s'intéresse à une méthode d'augmentation des transferts thermiques via une expérience de JAXA utilisant des parois rainurées dans la direction axiale. Par comparaison avec une chambre à parois lisses, les résultats démontrent la bonne prédiction par la SGE de l'augmentation du flux de chaleur grâce aux rainures et confirment la validité de la méthode développée pour des géométries de paroi complexes. / Combustion in cryogenic engines is a complex phenomenon, involving either liquid or supercritical fluids at high pressure, strong and fast oxidation chemistry, and high turbulence intensity. Due to extreme operating conditions, a particularly critical issue in rocket engine is wall heat transfer which requires efficient cooling of the combustor walls. The concern goes beyond material resistance: heat fluxes extracted through the chamber walls may be reused to reduce ergol mass or increase the power of the engine. In expander-type engine cycle, this is even more important since the heat extracted by the cooling system is used to drive the turbo-pumps that feed the chamber in fuel and oxidizer. The design of rocket combustors requires therefore an accurate prediction of wall heat flux. To understand and control the physics at play in such combustor, the Large Eddy Simulation (LES) approach is an efficient and reliable numerical tool. In this thesis work, the objective is to predict wall fluxes in a subcritical rocket engine configuration by means of LES. In such condition, ergols may be in their liquid state and it is necessary to model liquid jet atomization, dispersion and evaporation.The physics that have to be treated in such engine are: highly turbulent reactive flow, liquid jet atomization, fast and strong kinetic chemistry and finally important wall heat fluxes. This work first focuses on several modeling aspects that are needed to perform the target simulations. H2/O2 flames are driven by a very fast chemistry, modeled with a reduced mechanism validated on academic configurations for a large range of operating conditions in laminar pre- mixed and non-premixed flames. To form the spray issued from the atomization of liquid oxygen (LOx) an injection model is proposed based on empirical correlations. Finally, a wall law is employed to recover the wall fluxes without resolving directly the boundary layer. It has been specifically developed for important temperature gradients at the wall and validated on turbulent channel configurations by comparison with wall resolved LES. The above models are then applied first to the simulation of the CONFORTH sub-scale thrust chamber. This configuration studied on the MASCOTTE test facility (ONERA) has been measured in terms of wall temperature and heat flux. The LES shows a good agreement compared to experiment, which demonstrates the capability of LES to predict heat fluxes in rocket combustion chambers. Finally, the JAXA experiment conducted at JAXA/Kakuda space center to observe heat transfer enhancement brought by longitudinal ribs along the chamber inner walls is also simulated with the same methodology. Temperature and wall fluxes measured with smooth walls and ribbed walls are well recovered by LES. This confirms that the LES methodology proposed in this work is able to handle wall fluxes in complex geometries for rocket operating conditions.