Impact of a discretisation method and chemical kinetics on the accuracy of simulation of a lifted hydrogen flame

This paper presents an analysis of numerical and modelling issues based on a simulation of nitrogen-diluted hydrogen lifted flame evolving in a hot co-flow. We apply the large-eddy simulations (LES) method with the so-called 'no combustion model' and concentrate on the impact of chemical mechanisms and various discretisation schemes on the obtained results. The main attention is put to the latter issue in which we analyse to what extent a discretisation method of the convective terms of the scalar equations for the species and enthalpy affects the solutions. We consider eight commonly known total variation diminishing (TVD) schemes and three upwind schemes of the second order. The remaining terms in the scalar equations and all the terms in the Navier-Stokes equations are discretised applying the sixth-order compact difference method. Such an approach makes the discretisation errors of the convective terms the main factor affecting the solutions from the numerical point of view. Prior to the main analysis, the differences caused by the use of various TVD or upwind schemes are highlighted based on a single scalar transport equation with a known analytical solution. The results obtained for the flame are compared to experimental data taken from the literature. It is shown that the differences due to the application of a particular discretisation method are of similar magnitude as the differences between the simulation results and experimental data. Moreover, analysis of the impact of the chemical mechanism showed that observed differences are comparable to these originating from the use of different discretisation methods.