A new preferential diffusion model applied to FGM simulations of hydrogen flames

Abstract

Preferential diffusion is very important in simulations of hydrogen flames. Flame stretch and curvature induce strong preferential diffusion effects in laminar premixed hydrogen flames, causing strong local deviations from the unburnt mixture fraction in the reaction zone. In tabulated chemistry methods, this necessitates the use of a partially premixed model even if the inlet mixture is purely premixed. Furthermore, in realistic combustion problems heat losses often play a dominant role. In this paper we derive a preferential diffusion model for constant but non-unity species Lewis numbers using three controlling variables, namely mixture fraction, progress variable and enthalpy. The model has been implemented in the Flamelet Generated Manifold (FGM) approach and validated by comparing with detailed chemistry simulations. As a test case we investigate a 2D laminar premixed hydrogen flame stabilised on an isothermal slit burner. Additionally, the model was compared with the standard treatment of preferential diffusion in FGM to show the increase in accuracy of the new model presented in this paper. The new model shows a significant improvement compared to the previous model, which can be attributed to the inclusion of cross-diffusion. The importance of the additional diffusion terms and its variation in mixture fraction for initially purely premixed hydrogen flames is highlighted.

Local manifold learning and its link to domain-based physics knowledge

“…Mass fractions of combustion products, (in the case of hydrogen fuel, H 2 O) can be used to define progress variables [34,35,36,37,38]. In general, a progress variable, c, is defined in the literature as c = ns i=1 w i Y i , where the weighting factors, w i , are typically non-zero only for stable species and such that c changes monotonically across the flame.…”

Flamelet modeling of thermo-diffusively unstable hydrogen-air flames

“…Promising modeling approaches, taking into account detailed kinetics and their interaction with transport, include flamelet-based models such as the Flamelet/Progress Variable (FPV) approach [13], Flamelet Generated Manifolds (FGM) [14], or the Flame Prolongation for ILDM (FPI) model [15]. While manifolds for non-premixed combustion are usually generated from stretched non-premixed flamelets [16], manifolds for premixed combustion can recover moderate stretch effects even when being generated from unstretched premixed flamelets [14,[17][18][19]. Nevertheless, attempts have also been made to generate manifolds from stretched premixed flamelets [14,20,21].…”

“…Many advancements were aiming for improved predictions of the local mixture composition which has been realized via mixture fraction definitions based on elemental mass fractions, such as the Bilger mixture fraction, and efforts have been made to model its diffusivity [14,20,22]. Recently, different model extensions have been developed to improve the predictions of flamelet-based models for hydrogen combustion, where differential diffusion effects need to be captured accurately [17][18][19]. For TD unstable flames, which exhibit large curvature variations due to flame front corrugations, the suitability of these manifolds needs to be further investigated.…”