Modelling laminar diffusion flames using a fast convergence three-dimensional CVFEM code

Abstract

Laminar diffusion flames have been employed extensively to investigate the effects of physical and chemical-related parameters on combustion phenomena. Attention has been given to improve combustion models while keeping the two-dimensional (2D) axisymmetric approach and simplified boundary conditions. The flow structure due to the discrete jet nature of the reactants claims a more realistic inlet boundary condition for the burner plate which can only be accomplished if the solution domain is 3D. In this work, we present a novel 3D simulation code for the analysis of non-premixed laminar flames of methane and air, considering detailed inlet boundary conditions. The code is based on the control volume finite element method, written in MATLAB environment. To accomplish that a novel flow-oriented upwind scheme was implemented for the advection terms of the conservation equations. Combustion modelling can be carried out by either an enhanced flame sheet model in which an enthalpy equation is integrated in combination with the mixture fraction equation or detailed reaction schemes. Radiative heat losses were accounted for in the added energy equation. Three-dimensional numerical predictions, based on the flame sheet model, were able to simulate the near-wall gas temperature and inflow velocity field of a set of perforated plate burner with different porosities. It was found that the discrete nature of the injection of fuel and oxidiser, inflow boundary condition, plays a significant role on flame shape and length. Much better agreement between published experimental data and numerical predictions was obtained, particularly regarding flame structure, species mass fractions and temperature distribution of non-premixed laminar flames regardless of the adopted simplified combustion model. It was observed that flame length decreased as the burner porosity increased, up to a certain limit. The modified flow-oriented scheme greatly improved solution stability and computational time of the reacting system predictions.