Experimental and Numerical Investigation of Spherical Flame Propagation for Hydrogen-Methane-Air Mixtures

The addition of hydrogen to natural gas offers several advantages: (i) reducing carbon
emissions, (ii) extending the flammability limit by enhancing the flame resistance to strain-induced extinction, thereby increasing flame stability, and (iii) increasing the gas turbine's ability to operate at low loads, which expands the feasible load profile. Few previous experimental studies have focused on premixed spherical flame propagation for hydrogen/air and methane/hydrogen/air mixtures, especially at the high-pressure conditions most relevant to a hydrogen-fuelled spark�ignition car engine or an industrial gas turbine. This work employed a Schlieren technique to measure flame speeds for such mixtures in a spherical stainless steel combustion vessel, from which turbulent and laminar burning velocities were derived. The hydrogen volume fractions in methane were 30, 50, 70 and 100%. The initial pressures were 0.1, 0.5 and 1.0 MPa, and the initial temperatures were 303 and 360 K. The equivalence ratio (ϕ) was varied between 0.5 to 2 for pure hydrogen and from 0.8 to 1.2 for methane/hydrogen mixtures. The root mean square (rms) turbulent velocity (u’) for turbulent measurements varied from 2.0 to 10.0 ms-1.
The unstretched laminar burning velocities, ��, are derived and presented. The Markstein
length, Lb, Markstein number, Mab, and strain rate Markstein number, Masr, have also been derived. The results show that the maximum ul occurs on the rich side of stoichiometric conditions. For example, for 30% and 50% H2, it occurs at ϕ = 1.1. However, it shifts to ϕ = 1.2 for 70% H2 and to ϕ = 1.7 for a pure H2 explosion. The ul increased with hydrogen fraction and temperature and decreased with pressure. Unexpected behaviour was recorded for pure H2 explosions at low temperatures and ϕ=1.5, 1.7, wherein ul did not decrease when the pressure increased from 0.1 to 0.5 MPa.

Simulations of methane-hydrogen-air freely-propagating premixed laminar 1D flame using three recent chemical kinetic mechanisms were compared against the experimentally derived laminar burning velocities. There was generally good agreement with experimentally derived laminar burning velocities; however, for explosions of rich-pure hydrogen at high initial pressures, the level of agreement decreased but remained within the limits of experimental uncertainty.
Following the laminar measurements, turbulent flame investigations were conducted with the following aims: (a) to present an extensive experimental database of turbulent burning velocities for these mixtures over a wide range of conditions, (b) to establish a new correlation forturbulent burning velocity (ut) for a flame with Lewis numbers, Le, not equal to unity, and (c) to quantify the dependence of ut on pressure, temperature, stretch rate, laminar flame instability and rms velocity. As the pressure increased, the Taylor length scales decreased, increasing flame wrinkling and ut. The ut also increased as the temperature and u’ increased. The fuel/air mixture with high laminar flame instability (Le<1) has higher ut than those with higher Le. However, the ut/ul peaked in the high laminar burning velocity region. The turbulent flame investigation concluded that the increase in ut resulting from flame reactivity (laminar burning velocity) is more
significant than positive stretch (negative Mab) and flame instability.
A numerical study followed the experimental investigation using OpenFOAM software. Theturbulent flame wrinkling combustion model associated with the k-ε and dynamic-k-equation LES turbulence models are used in the present work. The empirical correlations from the current measurements are used instead of the Gulder correlation to calculate the wrinkling factor (Ξ) and solve the transport equation for the density-weighted mean reaction regress variable (b). The results showed that the k-ε and LES models agreed well with the experimental result, but LES provides more information about flame surface wrinkling.
The LES simulation with the combustion model that uses the transport equation for laminar burning velocity exhibits better agreement with the experiments than for the use of a constant laminar burning velocity. In addition, The maximum velocity magnitude is within the flame at the point where the mixture starts burning (b =0.8-0.95). This velocity wave was associated with the pressure wave in the flame propagation. The peak value of pressure was more significant for higher turbulent kinetic energy.