The objective of SIMUMOPS is to develop high-fidelity LES (Large Eddy Simulation) tools, in the field of propulsion (helicopter, aircraft, missile) for reacting two phase flows, in moving geometries, taking into account liquid film formation on walls. It is an HPC project because the resources needed for such simulations will require the largest parallel computers. Developments will be based on the AVBP LES software of CERFACS and IFPEN, which is the leading LES tool for reacting flows worldwide. Additional developments are needed in AVBP to handle moving meshes and liquid films on walls as well as the impact of droplets on films and walls. Two target applications have been identified for the second part of the project: (1) Constant Volume Combustion (CVC) chambers and (2) Oil-separation devices for helicopter engines. Application 1 corresponds probably to the last drastic efficiency improvement, which can be expected for all civil or military engines: burning at constant volume can improve the cycle efficiency from 10 to 20 %. Simulation is required to understand how a combustor can operate at constant volume in practice: here valves will be used at the inlet and outlet of the chamber, thereby requiring LES capabilities in moving systems. LES first tests performed in 2015 at CERFACS in collaboration with Pprime in Poitiers where an experiment has been installed, also show that liquid fuel injection will be a major issue, especially because significant amounts of kerosene will flow along the walls. Interestingly, the numerical machinery required for CVC systems is exactly the one needed for oil-separation devices in helicopters which are used to centrifuge oil and send it back to the engine: this is done in a rotating system and oil flow along the walls is a controlling phenomenon. Therefore the same numerical tool will be used for both cases (CVC and oil-separators). An additional issue which will be tackled in SIMUMOPS is 'integrated' LES. Up to now, most engines are computed in a 'modular' way: compressor, chamber and turbine are simulated separately, exchanging very limited information (flow rates, mean temperatures). In a CVC chamber, however, the whole flow is pulsated and the turbine is fed with an unsteady inlet: turbine and chamber must be computed together to provide reliable results. CERFACS has developed an overset-grid method that allows splitting the domain in multiple elements that run on the same machine using multiple instances. This efficient method allows 'integrated' LES. It will be applied here for the first time to the simulation of a CVC chamber coupled to a turbine. It is also used for all moving geometry cases as for the oil-separator. Optimizing it on HPC systems will be another task of SIMUMOPS. SAFRAN TECH and CERFACS bring complementary expertise for SIMUMOPS: SAFRAN TECH is leading a research project on CVC chambers and will provide industry guidelines for design and objectives, as well as experimental results. CERFACS will work on the physical modeling and the numerical HPC aspects of the project.

Les méthodes actuelles de simulation 3D appliquées aux moteurs permettent la sélection qualitative et à moindre frais des géométries et solutions techniques les plus prometteuses en terme de réduction de la consommation et des émissions de polluants. Malgré leur emploi de plus en plus systématique, elles sont limitées à la prédiction de points moteur stables. Or, l'avènement de nouvelles technologies, tels les procédés de combustion CAI et HCCI, ou l'injection directe du carburant, entraîne souvent l'apparition de variabilités cycliques importantes, dont la source et l'amplitude ne sont pas accessibles aux méthodes de simulation actuelles._x000D_ SGEmac propose de développer une méthodologie de simulation de l'écoulement et de la combustion dans les moteurs à piston en rupture avec celles utilisées actuellement, et basée sur la Simulation aux Grandes Echelles (SGE). Celle-ci- est par nature adaptée à la simulation d'instationnarités hautes fréquences, permettant ainsi d'agrandir le champ d'application de la simulation moteur à la prédiction de l'apparition de variations cycliques, et de contribuer ainsi à la limitation de leur impact sur les émissions et la consommation des moteur à piston dès la phase de conception_x000D_ Pour atteindre cet objectif, une expérience de référence sur un monocylindre fonctionnant en allumage commandé à injection directe et indirecte homogène sera réalisée. Une instrumentation combinant capteurs de pression et diagnostics optiques avancés permettra de caractériser les détails du fonctionnement du moteur cycle à cycle.