Numerical simulations of combustion process in a gas turbine with a single and multi-point fuel injection system

The paper presents a numerical study of a medium size model of industrial gas turbine combustor. The research was conducted using the RANS (Reynolds Averaged Navier-Stokes) approach with k-epsilon model and LES (Large Eddy Simulation) with WALE (Wall Adapting Local Eddy viscosity) subgrid model. The simulations were performed in cold and reacting flow conditions. In the latter case, the combustion process was modelled using a steady flamelet model with chemical mechanisms of Smooke with 16 species and 25 elementary reactions, and the GRI-2.11 with 49 species and 277 elementary reactions including NO chemistry. In the first part of the paper, the numerical results were validated against experimental data including velocity field, temperature and species concentrations. The velocity components predicted for the cold flow agree very well with measurements. In the case of the simulations of the reacting flow, some discrepancies were observed in both the temperature field and species concentrations. However, the main flame characteristics were captured correctly. It turned out that the chemical kinetics had a larger impact on the results than the turbulence model. In the second part of the paper, we modified the fuel and air injection method and analysed how the changes introduced affect the flame dynamics. It was shown that: (i) depending on the distribution of air, the velocity, temperature and species composition in the upper part of the combustion chamber can be significantly altered; (ii) more substantial changes can be achieved by shifting the fuel injection points; their location outside the main recirculation zone leads to a dangerous situation resulting in overheating of the walls; (iii) it turns out that substantial differences in the flame characteristics in the upper part of the combustion chamber vanish approaching the outlet plane and the resulting mixture compositions are very similar. (C) 2016 Elsevier Ltd. All rights reserved.