Physics-guided laminar flame speed correlation for methane-hydrogen-air mixtures with varying dilution

Abstract

Fuel-flexible, low-carbon combustion systems need to accommodate methane/hydrogen mixtures with air and exhaust-gas dilution. To develop these, we require accurate and efficient correlations for laminar flame speed (LFS). In this work, we introduce a physics-guided LFS correlation that applies to burners, gas engines, and turbines. Our model uses a core-kinetic approach based on flame temperatures, an algebraic function for the equivalence ratio, and a mass-flux-based blending law. This allows for accurate predictions with any methane/hydrogen blend. We set the model parameters using one-dimensional flame simulations with C3Mech v4.0.1, chosen for its high prediction accuracy for a wide range of experimental data, including new results from our spherical combustion chamber. The new correlation provides accuracy comparable to a machine learning approach (Gaussian process regression), yet remains physically consistent, differentiable, and extrapolates well. This makes it suitable for computational fluid dynamics and control of fuel-flexible combustion systems.

Figures (12)

• Figure 1: Overview of the LFS modeling framework.
• Figure 2: Spherical flame images for $\phi\!= 1.0$ mixtures, preheat temperatures of 373K, and a radius of 18mm recorded in the RWTH setup: Example flames from the present study and Beeckmann et al. Beeckmann2017CI_H2.
• Figure 3: Comparison of chemical-kinetic mechanisms with selected LFS experiments for (a)–(c) CH$_4$/air Beeckmann2019CI_flame_propagationHesse2019simulationDirrenberger2011laminar_naturalgasPark_2011Hu2015ignitiondelayHalter2005characterizationWang2018lbv_ch4_dme_h2Okafor2018methaneammoniaVarghese2019methaneLuo_2021Liao2004naturalgasWu2016lbv_pressureEckart2022ch4_h2Duva2020methaneMohammad2019dme_methane, (d)–(f) CH$_4$/H$_2$/air PrataliMaffei2025C3MechV4ECM_2017_Kruse_H2_CH4_mixBeeckmann2019CI_flame_propagationICDERS_2017_H2_FlameSpeedsHu2009CH4_H2_air_studyHuang2006LBV_H2Bradley2007hydrogenVarea2015_hydrogenAung1998Pressure_effect_H2Verhelst2005hydrogendayma2014peculiarHermanns2007phdTanoue2003JSMEKrejci2013hydrogenYu1986_methane_hydrogenWang2018lbv_ch4_dme_h2Okafor2018methaneammonia, and (g)–(i) H$_2$/air mixtures Hesse2019simulationICDERS_2017_H2_FlameSpeedsBradley2007hydrogenVarea2015_hydrogenLamoureux2003bombDahoe2005_laminarAung1998Pressure_effect_H2Qin2000LBV_H2Verhelst2005hydrogendayma2014peculiarHu2009CH4_H2_air_studyKrejci2013hydrogen. Panel titles indicate $T_\mathrm{u}$, $p$, $\phi$, and $X_{\mathrm{f,H}_2}$.
• Figure 4: Comparison of chemical-kinetic mechanisms with LFS experiments using coefficients of determination, $R^2$, and mean absolute percentage errors, f (MAPE).
• Figure 5: Example adiabatic flame temperature profiles from detailed chemistry (symbols) and the present study's model (lines), for atmospheric burner (top row) and engine (bottom row) representative conditions with varying $Y_\mathrm{ed} = 0$, 0.1, 0.2, and 0.3 shown in each graph from top to bottom.