ECCOMAS 2024

Turbulent Flame Topology and Propagation in Ammonia/Hydrogen Premixed Combustion

  • Im, Hong G (King Abdullah University of Science and Techn)
  • Khamedov, Ruslan (King Abdullah University of Science and Techn)
  • Hernandez Perez, Francisco (King Abdullah University of Science and Techn)
  • Malik, Rafi (King Abdullah University of Science and Techn)
  • Malpica Galassi, Riccardo (Sapienza University of Rome)
  • Valorani, Mauro (Sapienza University of Rome)

Please login to view abstract download link

Combustion of ammonia is considered a viable approach to achieve carbon-free power generation. Ammonia also serves as an effective solution for compact storage of hydrogen. There is growing interest in fundamental understanding of distinct combustion characteristics of ammonia due to its low flammability and high propensity to form NOx emissions. This study aims to provide insights into distinct characteristics of turbulence-flame interaction in premixed combustion of ammonia and its blends at different parametric conditions representing various turbulent combustion regimes. High fidelity direct numerical simulations in a rectangular channel configuration are conducted and the results are analyzed to understand the effect of the unique ammonia flame structure on the turbulent flame topology, the resulting burning rate enhancement, and pollutant emissions. The impact of the diffusive-thermal imbalances on the flame dynamics for different levels of ammonia-hydrogen ratios is also investigated, where the flame structure analysis revealed a distinctive variation in H2 and H atom distributions, leading to unique oscillatory mode of flame propagation. The latter part of the presentation discusses a new approach to achieve substantial acceleration of high fidelity simulation of reacting flows by developing a reduced order model. A significant reduction in scalar dimensionality and temporal stiffness is achieved by identifying slow invariant manifolds using computational singular perturbation method combined with principal component analysis. The computational algorithm fully utilizes modern CPU-GPU hybrid computing architecture and the data-based construction of chemical basis vectors. A significant level of speed-up for multi-dimensional reacting flow simulations is demonstrated.