Effects of molecular diffusion modeling on turbulent premixed NH[formula omitted]/H[formula omitted]/air flames

Abstract

Ammonia (NH3) is a promising carbon-free energy carrier and can be used as a fuel in gas turbines for power generation. Using a blend of NH3 and hydrogen (H2) has the potential to address the poor combustion properties of NH3/air mixtures. Predictive modeling of the NH3/H2 combustion process requires a clear understanding of molecular diffusion due to the high diffusivity of H2. While the effects of molecular diffusion modeling (MDM) have been extensively investigated in laminar/turbulent hydrocarbon and pure H2 flames, comparatively less investigations have been made to explore the impact of MDM on turbulent premixed NH3/H2/air combustion characteristics as well as NO emission. To this end, four three-dimensional direct numerical simulations of temporally evolving turbulent premixed NH3/H2/air jet flames are performed in this work considering MDM methods of increased complexity: 1) unity Lewis number (UL); 2) unity Lewis number with Soret effect (UL+S); 3) mixture-averaged diffusion with Soret effect (MA+S); and 4) full multicomponent diffusion with Soret effect (MC+S), respectively. The emphasis is placed on assessing MDM effects on flame structure, NO production, and combined role of MDM and stretch rate. Comparing MA+S with UL+S under both laminar and turbulent conditions indicates that the differential diffusion effect is significantly suppressed in the turbulent NH3/H2/air flames considered here. While MDM can affect the local equivalence ratio, it is found that the choice of the MDM method has a negligible impact on the major species and NO distribution conditioned on progress variable, as well as on critical reactions. However, MDM still shows a notable impact on NO production and turbulent flame speed. Specifically, from the most detailed model (MC+S) to the most simplified model (UL), the NO production speed decreases by 30% for the fuel pathway but increases by 338% for the thermal pathway; simultaneously, the global turbulent flame speed decreases by 18%. Furthermore, it is found that both NO mass fraction and local displacement speed are very sensitive to flame curvature and strain, while MDM has a negligible impact on the responses of normalized NO mass fraction and local displacement speed to the change in flame curvature or strain. This suggests that the effects of curvature and strain should be included in flamelet-based models to accurately predict NO emission and NH3/H2/air flame propagation, even when using the assumption of unity Lewis number.