CERFACS' research in hydrogen transition and safety addresses the growing need to decarbonize energy systems, driven by climate change and EU goals of an 80–95% reduction in CO₂ emissions by 2050. Hydrogen (H₂), either directly used or blended with other fuels (e.g., methane, ammonia), offers a promising alternative. Other carbon-neutral strategies involve producing synthetic fuels (e-fuels) through green technologies like water electrolysis and methanation. However, using H₂ introduces significant technological and safety challenges, especially for combustion systems in gas turbines, furnaces, and propulsion.
CERFACS brings its expertise in combustion chemistry, turbulent combustion, phase change, and heat transfer to develop Large Eddy Simulation (LES)-based CFD tools for H₂ combustion. The objectives include providing predictive combustion models, updating combustion strategies for H₂-rich fuels, accurately assessing pollutant emissions (especially NOₓ), and ensuring turbines can withstand higher flame temperatures associated with H₂.
Four key challenges are targeted. First, the integration and reduction of complex chemical schemes involving H₂ and its blends are essential for pollutant prediction and simulation efficiency. Second, modeling thermo-diffusive instabilities, common in H₂ combustion, is critical to prevent dangerous flame behavior like flashback or explosions. Third, the multiphysics interactions (turbulence, acoustics, radiation) must be captured to ensure burner operability, requiring tightly coupled simulation approaches. Fourth, simulating large-scale burners (e.g., industrial furnaces) presents computational difficulties due to large disparities in spatial and temporal scales, especially when using explicit time-stepping.
To meet these challenges, CERFACS utilizes advanced tools like ARCANE for chemistry reduction, AVBP for LES/DNS simulations, and CWIPI for multiphysics coupling. Strategic focus areas include sustainable programming—especially optimizing HPC workflows for expensive simulations—and improving numerical algorithms for stiff chemistry and thermal radiation. Data-driven modeling, particularly deep learning, is also being explored to enhance subgrid-scale modeling and develop “smart RANS” methods from high-resolution LES or DNS data.
This work is highly relevant to CERFACS’ industrial and institutional partners, including Airbus, Safran, ONERA, EDF, Total, and CNES, all of whom are engaged in hydrogen research. It also supports cross-sector applications in electricity generation, petrochemistry, and manufacturing industries (e.g., glass and cement), while addressing safety concerns related to explosion risks, as studied within the LEFEX research program.
