CFD assessment of non-premixed turbulent methane flame with reduced chemical kinetics

Abstract Pollutant predictions in turbulent diffusion flames depend on how mixing, chemistry and their coupling are modelled. Using the Brookes–Moss experiments as reference, this study assesses how the Reynolds-averaged Navier–Stokes (RANS) turbulence closure, the chemical-kinetic scheme and the turbulence intensity of the air co-flow affect the computed structure, nitric oxide (NO) and soot of an enclosed methane jet flame at a Reynolds number of 5000. Steady axisymmetric simulations used the eddy-dissipation model, thermal and prompt NO post-processing and the Moss–Brookes soot model. The measured mixture fraction and the fast-chemistry limit were combined to bound the predicted fuel mass fraction. Of the closures tested ( k – ε , k – ω and a Reynolds-stress model, RSM), the RSM lowered the mean absolute percentage error of the centreline temperature from 30.1% to 15.9%. Replacing the one-step scheme by a 16-step augmented reduced mechanism lowered the fuel-decay error from 56.6% to 9.5%, a larger gain than that obtained by changing the closure. Raising the co-flow turbulence intensity from 20% to 80% reduced the peak NO from 183 to 168 ppm (− 8.4%) and the peak soot volume fraction by 43%. Centreline soot levels agreed with the measurements within 7%, although the radial soot extent was under-predicted.