Evaluation of chemical kinetic models for simulations of hydrogen detonations by comparison with experimental data

Two-dimensional numerical simulations of a weakly unstable detonation mixture 2H2+O2+3.76Ar at 20kPa and 295K were performed using our validated OpenFOAM solver based on reacting-PimpleCentralFoam. This study compared the detonation dynamics obtained with four chemical models, namely Hong 2011, Burke 2012, Mével 2014, and FFCM-2 with recently obtained experimental results. The experimental–numerical comparisons were performed in threefold: (i) quantitative comparisons of the cell sizes (λ) and their distributions (2σ/λ); (ii) qualitative comparisons of the detonation structure based on simultaneous planar laser-induced fluorescence of both nitric oxide (NO-PLIF) and OH radical (OH-PLIF); (iii) qualitative and quantitative comparisons of the detonation dynamics based on combined Rayleigh scattering and NO-PLIF measurements. The simulations conducted with Hong 2011’s, Burke 2012’s, and FFCM-2’s models satisfactorily reproduced the average cell size (within 10%), while it was 1.5 times smaller with Mével 2014’s model. The opposite trends were observed in cell size distributions (2σ/λ) with satisfactory predictions from Mével 2014’s model (within 25%) and almost no cell size variations (2σ/λ < 0.1) for the other models. By comparing the simultaneous NO- and OH-PLIF imaging, the simulations conducted with FFCM-2’s and Mével 2014’s models qualitatively reproduced the reaction zone structure, while more discrepancies were obtained with Hong 2011’s and Burke 2012’s models. Quantitatively, simulations conducted with FFCM-2’s and Mével 2014’s models presented the lowest discrepancy (below two-fold) at reproducing the induction zone dynamics along the cellular cycle, while large discrepancies (approximately three-fold) were observed with Hong 2011’s and Burke 2012’s models. Chemical timescale analyses evidenced the relation between the faster reaction timescales of Mével 2014’s model and the ability to reproduce the experimental variability on both λ and Δi. These detailed comparisons emphasized the importance of the chemical model selection and the need for combined experimental measurements to both validate chemical models and achieve predictive detonation simulations.