Thèse Simulations Numériques Détaillées de la Combustion en Milieux Poreux Utilisant l'Ammoniac et l'Hydrogène Comme Combustibles Couplage Multiphysique et Modélisation Physique H/F - Doctorat.Gouv.Fr

Établissement : Université Paris-Saclay GS Sciences de l'ingénierie et des systèmes École doctorale : Sciences Mécaniques Energétiques et MAtériaux pour l'InGénierie Laboratoire de recherche : IFP Energies Nouvelles - Sciences et Technologies du Numérique Direction de la thèse : Olivier COLIN ORCID 0000000289473490 Début de la thèse : 2026-11-16 Date limite de candidature : 2026-11-15T23:59:59 Pour lutter contre le changement climatique, le développement de brûleurs sans carbone et à faibles émissions devient une nécessité pour l'industrie et un défi technique majeur pour les chercheurs. L'ammoniac (NH3) et l'hydrogène (H2) sont les deux principaux combustibles sans carbone considérés aujourd'hui. Si H2 brûle facilement, son stockage et son transport sont difficiles, tandis que NH3 possède une vitesse de flamme faible mais bénéficie d'une infrastructure existante. Une solution prometteuse est le craquage partiel de l'ammoniac proche du brûleur, créant des mélanges NH3/H2 à combustion plus rapide. Toutefois, ces mélanges produisent beaucoup de NOx, sauf à très faibles rapports carburant/air (FAER), difficiles à contrôler dans les brûleurs turbulents.

Une alternative étudiée dans cette thèse est l'utilisation des brûleurs poreux. Grâce à la recirculation de chaleur dans le matériau poreux, les flammes peuvent se stabiliser à faible FAER, réduisant potentiellement les émissions de NOx. Le développement des brûleurs poreux nécessite une compréhension approfondie de la physique couplée impliquée : conduction thermique, transfert radiatif, stabilisation des flammes, chimie de surface... Ces phénomènes sont difficiles à étudier expérimentalement en raison des petites échelles impliquées et de l'opacité de la matrice poreuse. La simulation numérique (CFD) semble donc être la meilleure approche pour améliorer cette technologie.

Cette thèse vise à étudier les flammes NH3/H2 dans les brûleurs poreux par des simulations numériques directes (DNS) afin de comprendre les mécanismes de formation des NOx. Le code CFD CONVERGE sera utilisé, avec un raffinement de maillage adaptatif pour améliorer la résolution des transferts thermiques et des réactions chimiques. Le rôle de la chimie de surface dans la formation des NOx sera également exploré. Les résultats innovants de ces recherches seront valorisés par des publications dans des revues scientifiques à comité de lecture.
Hydrogen and ammonia as carbon-free energy vectors
According to the French Low-Carbon National Strategy, achieving carbon neutrality by 2050 requires the integration of low-carbon energy vectors into industrial processes. This transition is essential for mitigating climate change, strengthening energy security, and promoting economic growth through sustainable methods. Among the energy carriers identified, hydrogen (H2) and ammonia (NH3) stand out as key candidates [1]. While public awareness of ammonia remains limited, it plays a vital role in various industries, including fertilizer and plastic production, pharmaceuticals, textiles, and water treatment.
Ammonia offers several advantages as a fuel - it is carbon-free, has a higher volumetric energy density than hydrogen, and despite its toxicity, can be safely stored and transported due to existing infrastructure. However, ammonia combustion presents significant challenges, including a narrow flammability range and low flame speed, making it unsuitable for direct use in conventional combustion systems. Ammonia combustion produces high nitrogen oxides (NOx) emissions, and unburnt NH3 can harm air quality through direct effects or complex atmospheric reactions. Hydrogen, for its part, poses well-known safety issues due to its high flammability and detonability. Interestingly, ammonia can be entirely or partially cracked back into hydrogen [2], yielding NH3/N2/H2 mixtures whose energy efficiency and intermediate combustion properties can be close to well-known hydrocarbons and are worth considering [3].
Harnessing NH3/H2/N2 blends using porous burners
In this context, it seems necessary to pursue the development of combustion devices capable of handling both NH3, H2 and their blends, to ensure a broad, fuel-flex, low-emission and safe operation. Porous burners, in which combustion occurs within the voids of a solid matrix, show unique potential to harness these blends [4]. Recent works showed that porous burners are able to reduce unburnt NH3 and NOx emissions [5]. Thanks to a process of heat recirculation, where enthalpy travels upstream of the reaction front through conduction and radiation, combustion temperatures are increased thus chemical kinetics are accelerated (see Fig. 1) [6]. In practice, this enables a broadening of the flammability limits, a large down-turn ratio and a certain fuel-flexibility [7]. On the lean side, this enables the combustion of normally non-flammable mixtures, typically showing lower pollutant emissions. This is particularly relevant for H2, where the inlet mixture could be non-flammable outside the combustion device [8].
To date, porous burners are mostly used in the industry in the so-called radiant mode', where maximal heat loss to the outside of the burner is sought. This is particularly useful for the heating of large surfaces, glass manufacture or the drying of paper [9]. In special cases, up to 30-40% of the chemical enthalpy can be converted to radiative flux [10]. However, few applications are found in the domestic sector for heat generation, such as home cooking devices or boilers, where external heat loss rather has to be avoided. In practice, the TRL of porous burners remains quite low, due to the low durability of the matrix materials: ceramics are brittle and can break due to thermal shocks [11], while metals have a low melting point which restrains the operability [12].
The vast majority of existing studies concern the combustion of hydrocarbons within porous burners [13]. In many studies, low levels of CO and NOx emissions are reported [14] - though some operating points and systems show on the contrary high emissions [15]. 1D macroscopic models known as Volume-Averaged Models (VAMs) are sometimes unable to predict accurately these low emissions [16]. Although some studies report 2/3D pore-level simulations of porous burners, the mechanisms leading to these low emissions were not considered due to the high numerical costs of using detailed kinetics. Also, it is likely that these low emissions arise from the intricate coupling of the flow at the pore level, the Conjugate Heat Transfer (CHT), detailed kinetics, radiative heat transfer (RHT) and losses, side wall quenching and the physical size of the burner - which require the use of Direct Numerical Simulations (DNS), an efficient physical modelling and an accurate meshing representation. Of course, a direct understanding of the mechanisms leading to pollutant formation within porous burners is necessary to develop clean NH3/H2 combustion systems.
Thermo-diffusive effects
Thermo-diffusive effects significantly impact flame propagation in ammonia (NH) and hydrogen (H) mixtures, affecting global flame speed and structure. Khamedov et al. [17] used direct numerical simulations (DNS) to study turbulent NH-H-N-air flames, finding that increasing the H fraction elevates the mean turbulent flame speed and induces oscillations due to the preferential diffusion of H mixed with low-reactivity NH. These mixtures show complex dynamics. Similarly, Gaucherand et al. [18] examined instabilities in premixed hydrogen and hydrogen/ammonia flames under various pressures and equivalence ratios. Their study showed that adding ammonia reduces the growth rate of perturbations due to lower reactivity, resulting in a higher Zeldovich number. They also found that very lean mixtures (equivalence ratios of 0.4 and 0.5) remain thermo-diffusively unstable due to hydrogen's preferential diffusion.

Flame-wall interactions (FWI) and wall fluxes also play a critical role in pollutant formation and combustion stability in NH/H mixtures. Tamadonfar et al. [19] studied laminar FWI and found that wall heat fluxes are significantly influenced by the flame's proximity to the wall; as laminar flames approach the wall, NO production decreases and NO production increases due to lower combustion efficiency and reduced H and OH radicals. Building on this, Chi [20] investigated turbulent FWI for NH/H mixtures using DNS. Turbulent flames showed enhanced mixing and curvature effects, leading to a smaller quenching distance and less unburned NH. However, turbulence also resulted in higher NO production near the wall due to frequent flame-flame interactions and thickened flame fronts, especially at low Damköhler numbers.

Overall, thermo-diffusive instabilities in NH-H mixtures are strongly affected by fuel composition and equivalence ratio, impacting pollutant formation and flame stability both during propagation and wall interaction. Understanding these effects is vital for optimizing combustion technologies using these mixtures, particularly under lean conditions where instabilities are more pronounced.
1) Develop a reliable DNS framework adapted to porous media combustion that is easy to implement on various burner-scale configurations. This objective comes from the observation that DNS of porous burners are today very complex to set up, because the internal convoluted flame front must be fully resolved at an unknown position, and CHT and RHT are necessary. This is especially challenging to simulate large systems at burner scale. The CFD tool CONVERGE will be used, which includes a unique automatic meshing at pre-processing, Adaptive Mesh Refinement (AMR) capability, CHT and RHT handling, and the availability of complex chemistry.
2) Understand the pore-level mechanisms leading to low or high NOx emissions within porous burners (quenching, near-wall reactions, chemical pathways, influence of boundary conditions, CHT and RHT). This is likely the main challenge of H2/NH3 porous media combustion, where low levels of NOx at the burner exit must be ensured for societal acceptability.
3) Explore the influence of topological variations of the porous structure (pore size and geometry, porosity and spatial gradation of these parameters) in order to understand more complex pore-level stabilization mechanisms within a variety of solid structures, with the aim of improving combustion efficiency and reduce pollutants emissions.
IFPEN has the appropriate CFD tool to study the complex multi-physics system of a porous burner. One of our partners, CEA Saclay, has a new experimental bench adapted to the premixed combustion (TREASURE), allowing a direct visualization of submerged combustion within porous media (see Fig. 1). This offers the necessary ecosystem for large-scale simulation and adequate validation.
CFD tool. The PhD will focus on the CFD tool CONVERGE [21]. This code is commercialized by Convergent Science (CSI) worldwide for various applications like piston engines, gas turbines, pumps, or wind turbines. IFPEN has a strong partnership with CSI since 2015 and has full access the code, allowing efficient model development. Unlike many codes, only a surface discretization needs to be provided by the user. The meshing is performed automatically by the code using regular hexahedra. These hexahedra are cut at boundaries following the exact surface topology, thus ensuring correct modeling of wall phenomena. Mesh refinement is performed by successive divisions of the hexahedral mesh by a factor 8. This refinement can be performed by the user prior to the simulation by defining zones of refinement and on-the-fly through Adaptive Mesh Refinement (AMR). The TFM-AMR combustion model developed at IFPEN [22] takes advantage of this AMR feature to automatically refine the mesh in the flame front. While the meshing of a porous medium is particularly tedious with standard codes, this work should be extremely simplified with CONVERGE during this PhD. In order to avoid the strong time step limitation of the compressible Navier-Stokes equations, low-Mach models will be used, which are compatible with the small pressure drops and velocities attained within stationary porous media combustion.

Profil de Master recherché - MSc
Informatique : modélisation 3D, programmation ;
Mécanique des fluides : combustion, écoulements turbulents ;
Numérique : schémas numériques, langage C++, Fortran ;
Une très bonne maîtrise de l'analyse numérique et de la programmation est exigée.
Bon niveau d'anglais (capacité à lire des articles scientifiques et à échanger oralement au quotidien).

• Programmation
• Anglais
• Rédaction d'articles scientifiques
• Mécanique des fluides
• Fortran
• Modélisation 3D
• Chimie