Numerical investigations on flashback limits of premixed hydrogen-methane-air laminar flames
Résumé
Injecting hydrogen into the natural gas network to reduce CO2 emissions in the residential sector is considered in EU as an important element of the zero CO2 emissions target for 2050. Assuming that hydrogen is generated using non-CO2 emitting processes, this policy has indeed an important decarbonizing potential. One crucial element is obviously the blending rate of hydrogen in natural gas and the associated potential risks, the main one being the flame flash back phenomenon that could occur in home appliances using premixed laminar burners. Presently, there is no homogenized regulation at the EU level for hydrogen blending rates. Therefore, basic studies are needed to determine the flash back limits for various natural gas and hydrogen premixtures burning in air. This need is recently addressed especially experimentally in various studies. A very comprehensive one is the work by de Vries et al. (2017). In the present study, we numerically analyse and simulate their experimental results concerning the flashback limits, obtained in an experimental set-up using a quartz tube burner of 1 m length and 1 cm diameter. They have carefully regulated the mass flow rates and the hydrogen content of methane air mixtures in order to determine the flashback limits for agiven unburnt mixture. They characterized these limits for different Wobbe indices and hydrogen blending rates. In the present study, the open source CFD code OpenFOAM is used for 2D axisymmetric numerical simulation of the experiments. The reactingFoam solver and the laminar combustion model areutilized in order to model the laminar reacting flow. ReactingFoam uses finite rate chemistry to compute reaction rates and uses PIMPLE algorithm to solve Navier-Stokes equations. GRIMECH 3.0 is used for chemistry computations, consisting of 325 reactions and 53 species. To reduce the computational cost, Tabulated Dynamic Adaptive Chemistry method is used which takes advantage of combining dynamic adaptive chemistry and in-situ adaptive tabulation algorithms. A stable flame is computed with a velocity higher than the experimental flashback limit with local time stepping (LTS) for every hydrogen-methane blending ratio, as a base case. To observe flashback, the inlet bulk velocity is gradually reduced and transient computations are performed, without LTS. Initiation of flashback below the experimental limit is readily observed for pure methane-air flames. The effect of hydrogen addition on flashback limits is studied and compared to the experimental results. The changes of the wall velocity gradients at flashback initiation are discussed.