Detailed Kinetic Model for Combustion of NH3/H2 Blends

TL;DR

Ammonia/hydrogen combustion presents ignition, instability, and NOx challenges that hinder practical use. The authors develop a comprehensive NH3/H2 kinetic mechanism using the Reaction Mechanism Generator (RMG) with updated thermochemistry, high-level kinetics, and bath-gas mixture rules, avoiding parameter tuning. The resulting mechanism (53 species, 704 reactions) demonstrates improved predictive robustness across ignition delay times, laminar burning velocities, and species profiles in jet-stirred and flow reactors, though NOx remains a target for refinement. This work provides a transferable, untuned framework that captures bath-gas effects and reduces mechanism-truncation errors, advancing reliable CFD-enabled design of burners and engines for NH3/H2 fuels.

Abstract

Ammonia is a promising zero-carbon fuel for industrial and transport applications, but its combustion is hindered by flame instabilities, incomplete oxidation, and the formation of nitrogen oxides. Accurate and detailed kinetic models are critical for designing optimal burners and engines. Despite numerous mechanisms published in recent years, large discrepancies remain between model predictions and experimental data, particularly for NOx species. In this work, we have reviewed the literature to obtain the most up-to-date and reliable thermochemical and kinetic parameters for most reactions present in ammonia combustion, and for reactions for which these parameters are not available, we performed high-level calculations to determine them. The purpose of this was to minimize the number of estimated parameters used in model development. A new, detailed kinetic mechanism was then generated with the Reaction Mechanism Generator (RMG). To ensure physical consistency, geometry optimizations were carried out for all hypothesized 'edge' species, and any non-convergent or non-physical structures were excluded. The resulting mechanism was tested against experimental laminar burning velocities, ignition delay time, flow reactor species profiles, and jet-stirred reactor data, and compared with five recent representative mechanisms. Recently developed bath-gas-mixture rules were applied to a number of key reactions in the mechanism, and we found this to result in better agreement with experiment for a number of modeling targets. While the mechanism does not reproduce all experimental results, it demonstrates improved robustness without parameter tuning, thereby reducing the risk of over-fitting and enhancing predictive reliability under conditions relevant to practical applications.

Figures (12)

• Figure 1: Computed rate coefficients for the O(^3P) + HNO -> ^2OH + ^2NO reactions shown together with the experimental determination by Inomata and Washida.INOMATAWASHIDA1999. The submerged saddle-point for this reaction was lowered by 2.4 kJ.mol^-1 to obtain better agreement with experiment.
• Figure 2: Computed rate coefficients for the well-skipping ^2H + ^2H2NO -> (NH2OH/NH3O) -> H2 + HNO and ^2H + ^2H2NO -> (NH2OH/NH3O) -> ^2NH2 + ^2OH reactions shown together with the direct-abstraction reaction ^2H + ^2H2NO -> H2 + HNO. The first two reactions were found to still be at the low-pressure limit at 100 bar
• Figure 3: Computed rate coefficients for the well-skipping ^2HO2 + ^3NH -> (^2HNOOH) -> ^2OH + HNO, ^2HO2 + ^3NH -> (^2HNOOH/^2H2NOO) -> ^3O2 + ^2NH2 and ^2HO2 + ^3NH -> (^2HNOOH/^2H2NOO) -> ^3O + ^2H2NO. The reactions were found to still be at the low-pressure limit at 100 bar. Only the ^2OH + HNO forming channel is relevant in practice.
• Figure 4: The temperature dependencies of the abstraction rate coefficients updated and/or computed in this work.
• Figure 5: Standard enthalpies of formation ($\Delta_\mathrm{f}H_\mathrm{298~K}^{\ominus}$) calculated at DLPNO-CCSD(T)-F12/cc-pVTZ-F12//$\omega$B97X-D/def2-TZVP are compared to estimates by RMG RMG2021RMG2022, including group additivity (blue circles) and hydrogen bond increment (HBI) corrections Pang2024 (red squares).