Simulation and Modeling of Intrinsic Instabilities in Laminar and Turbulent Premixed Ammonia/Hydrogen/Air Flames

Methods

The adaptive mesh solver PeleLMeX was used for the simulations. PeleLMeX solves equations based on a low-Mach-number formulation of the reacting Navier-Stokes equations, where the fluid is treated as a mixture of ideal gases. A mixture-averaged model for differential species diffusion and a thermal diffusion model to incorporate Soret effects is used. The discretization couples a multi-implicit spectral deferred correction approach for the integration of mass, species and energy equations with a density-weighted approximate projection method, which incorporates the equation of state through a velocity divergence constraint. The resulting discretisation is integrated with timesteps determined by an advective CFL number, with the faster diffusion and chemical processes treated implicitly. Note, that the timesteps for slow flames such as ammonia are still limited by the chemistry due to numerical stiffness of the reaction mechanisms. The scheme is embedded in a parallel adaptive mesh refinement (AMR) algorithm framework based on a hierarchical system of rectangular grid patches. The complete integration algorithm is second order accurate in space and time. The chemistry integration is offloaded to CVODE provided by SUNDIALS. For chemical reactions a mechanism with 30 species and 243 reactions was used, which has shown to give the best overall agreement with experimental data for flame speed, extinction strain rates, and species concentrations.

Results

In this phase of the project, the flame propagation mechanisms of lean, premixed ammonia/hydrogen/air flames in a prechamber system under elevated pressure and temperature conditions ($\varphi =0.6$, $T_{\mathrm{u}}=500\mathrm{K}$ $p=4\mathrm{bar}$) were investigated. For the study, three-dimensional DNS were performed for fuel mixtures with hydrogen contents of 0, 20 and 30% (X0, X20, X30) by volume. The mixtures were ignited inside the prechamber. Different stages in flame propagation were observed and the flame surface area generation mechanism was investigated through the analysis of combined curvature and strain rate terms. The balance between these terms was shifted as the hydrogen content increased. For pure ammonia and low-hydrogen content mixtures, some quenching was observed, resulting in a stall of the flame surface area generation. For hydrogen-enriched mixtures a significant enhancement of the local flame propagation speed was observed. In contrast, for the X0 case, the normalized reactivity does not exhibit the same dynamic increase.

Discussion

The different stages of the flame development showed a transition from the hot jet to turbulent flame propagation for cases X20 and X30, showing good agreement with similar experimental studies. The analysis of the flame surface generation mechanism showed that the impact of the tangential strain rate increases with higher hydrogen content due to increased velocities and enhanced turbulence generation behind the nozzle. Initially, positively curved regions had the greatest impact on flame surface generation. For cases X20 and X30, the contribution of positively curved regions peaked when the hot reactive jet entered the main chamber. A subsequent decrease in flame stretch indicated the transition from hot jet-dominated to turbulent flame propagation. This decrease resulted from reduced contributions from positively curved regions and increased flame surface destruction in negatively curved regions. These effects essentially balance each other, leading to steady flame stretch during the turbulent propagation phase. Additionally, in cases X0 and X20, the flame surface area generation stalled upon entrance into the domain. This was shown to be due to at least partial quenching observed in the OH field. While case X0 exhibited enhanced local reactivity, the penetration of the hot reactive jet into the main chamber did not produce a local maximum in normalized reactivity. This behavior changed significantly for the hydrogen-enriched cases X20 and X30, which showed local maxima in normalized reactivity after the hot reactive jet entered the main chamber. This dynamic was closely linked to the increase of the mean curvature caused by the breaking up of the jet and due to the differential diffusion of hydrogen in positively curved regions. For both mixtures, the enhancement declined and nearly saturated at a single value when the transition to turbulent flame propagation occurred. The enhanced reactivity in case X0 was linked to an initially increased specific enthalpy, caused by the ignition spark, thereby enhancing the normalized reactivity after ignition due to elevated temperatures. For X0, the subsequent flame development within the main chamber showed some evidence of differential diffusion through variations in enthalpy below the adiabatic temperature.

Additional Project Information

DFG classification: 404 Fluid Mechanics, Technical Thermodynamics and Thermal Energy Engineering
Software: PeleLMeX
Cluster: CLAIX

Publications

Marvin Schüpfer, Thomas L. Howarth, Michael Gauding, Heinz Pitsch,
Flame propagation dynamics in an ammonia/hydrogen combustor with a passive prechamber,
White paper, April 2025

Svenja Nerzak, Thomas L. Howarth, Michael Gauding, Heinz Pitsch,
Transport terms in LES flamelet combustion model: A priori analysis of turbulent premixed hydrogen slot flames,
White Paper, April 2025

Thesis:

Marvin Schüpfer,
Analysis of the turbulent combustion of hydrogen/ammonia fuels in prechamber combustion systems by means of direct numerical simulation,
Master thesis, March 2025

Tom Jose,
Investigation of partially premixed lifted turbulent H2 flames,
Master thesis, January 2025