The objective of our proposal is to design and develop a multi-domain/multi-method computation method that is a suitable response to the great simulation challenges posed by the complexity of industrial electromagnetic applications. The areas of application are complex on-board antennas, electromagnetic compatibility and radar signatures. We need to solve Maxwell’s equations efficiently in the harmonic regime in a domain that is unbounded and at a potentially very high frequency. The number of unknowns to be determined is considerable. On the other hand, the heterogeneous scales typical of complex industrial objects prevent the employment of a “universal” method.. The proposal is on a new method of calculation by sub-domains that should remove these difficulties by fully exploiting the massively parallel architectures of future computers. Each domain will be defined so as to be able to use a specialized solver dedicated to its treatment. The design of well conditioned solvers will enable to use efficiently iterative methods The scientific heart of our proposal is the preconditioning of large linear electromagnetic systems. The conditioning problem will be taken into account from the design of the equations, meaning at the continuous problem formulation level rather than after its discretisation.

The aim of this project is to develop a theoretical approach and associated computer code for Large-Eddy Simulation (LES) of liquid jet fragmentation and atomization in all thermodynamic conditions, from subcritical to supercritical pressure and temperature conditions. The transition between sub and supercritical conditions will be considered through a new and unified model, able to deal with capillary effects, phase transition between a liquid and its vapour as well as the dynamics of supercritical fluids. These different regimes of fuel jet dynamics are found in liquid rocket as well as Diesel engines depending on the combustion chamber temperature and pressure conditions. The situation is possibly already prevailing in current aircraft engines or will be reached in the future as the chamber pressure is being increased. For sufficiently low pressure and temperature with respect to the critical point, the fluid undergoes a classical break-up process. The interface separating liquid and gas phases is discontinuous and droplets appear as the final stage of fragmentation. As pressure is increased to supercritical conditions, the phase discontinuity is no longer present. The jet evolves in the presence of a continuous interface and mass transfer takes place through turbulent fluxes and the process is analogous to mixing of variable density jets. It is planned to deal with all these aspects in LES and DNS frameworks. Such a unified approach has never been considered in the author’s knowledge as phase transition modeling in arbitrary flow conditions is a true scientific challenge.

The project FLOCCON is dedicated to the development and validation of an innovative strategy to accelerate solvers for fluid mechanics. The project focuses on incompressible solvers, containing two parts: (1) a linear Poisson equation, and (2) a non-linear advection equation. The key idea of this project is to use deep learning to train neural networks based on solutions of these two equations. To go further, the project will examine learning methods which can guarantee a target accuracy. To do so, physical-based and long-term 'loss functions' will be introduced, in order to ensure a limited error accumulation in time. Moreover, an hybrid strategy will be proposed to obtain a robust solver. finally, an optimisation of this new network-based solver will be carried out on CPUs/GPUs. In addition to classical validation cases, a target application will be simulated on the pollutant dispersion in a large city, which is a challenging case for classical HPC solvers.

Combustion instabilities constitute a severe challenge in the development of efficient low-emission combustion systems, such as gas turbines for aeronautical and power generation applications. Previous work has shown the principle capability of nanosecond repetitively pulsed (NRP) plasma discharges to mitigate this undesirable unsteady combustion phenomenon in academic configurations; however, essential physical effects associated with the application of this technology in real gas turbine engines for aircraft propulsion and power generation have not been considered yet. These effects are related to 1.) liquid-fueled spray flames, 2.) elevated operating pressure, and 3.) high-frequency non-planar modes. The GECCO project will tackle these three aspects with dedicated experiments and high fidelity simulations. A common swirl-burner platform will be used for all three aspects to maximize synergy effects between the individual work packages. The AVBP code, well established for turbulent combustion simulations of academic and industrial configurations, will be combined to a plasma code to account for the effects of NRP plasma discharges on turbulent flames, taking into account ultrafast heating as well as slower thermal and chemical effects. Once validated on the basis of experimental data, this numerical tool will be essential in achieving a comprehensive understanding of the plasma-flame-acoustic interaction related to the 3 effects mentioned above. The effect of NRP discharges on the dynamics of spray flames will be assessed in detailed measurements (phase-Doppler anemometry, light-sheet tomography, particle image velocimetry), investigating the influence on the cold spray, the flame shape, and the dynamic response to acoustic perturbations (flame transfer function, FTF). To assess and demonstrate the potential of NRP discharges in the mitigation of high-frequency azimuthal instabilities, the swirl burner will be equipped with circumferentially distributed plasma actuation. The response of the flame to this type of forcing will be experimentally assessed using azimuthally resolved measurements (pressure, chemiluminescence). Plasma-flame-acoustic interaction at elevated pressures will be investigated in a high-pressure facility. The effect of NRP on the FTF and the response of the flame to low-frequency modulated harmonic plasma forcing will be measured up to 10 bar. All experimental tasks are accompanied by corresponding simulations that will provide a more detailed understanding of the interaction mechanisms than accessible by measurements only. In the final part of the project, NRP discharge forcing will be utilized to control acoustically coupled combustion oscillations in the three experimental facilities (spray flames, high-frequency modes, elevated pressure). GECCO may, thus, increase the fundamental understanding of dynamic plasma flame interaction and, on the other hand, bring this technology significantly closer to real applications.

During the design of cleaner and more efficient aerojet engines, reliability of re-ignition in high altitude must be demonstrated. The TIMBER project involves three laboratories (CERFACS-CORIA-EM2C) and the SAFRAN Group to study and accurately simulate the ignition process in annular multi-burner combustors similar to the ones found in aeronautical gas turbines. On the one hand, the large-scale simulations considered in this investigation will require the use of the latest massively parallel clusters which make such supercomputing efforts affordable. On the other hand, complex multi-physics simulations are necessary to account for the wide range of phenomena that are involved (two-phase flows, turbulence, combustion, conjugate heat transfer, radiation). Several experimental studies in unique multi-burner combustors will provide data to validate the different steps of the project.