Evaluation of Global Mechanisms for LES Analysis of SGT-100 DLE Combustion System

Abstract

This paper presents the results of Computational Fluid Dynamics (CFD) analyses obtained for the experimental version of the SGT-100 Dry Low Emission (DLE) gas turbine burner provided by Siemens Industrial Turbomachinery Ltd (SIT). A testing and measurement campaign for this burner was previously carried out at the DLR Institute of Combustion Technology, Stuttgart, Germany, for various operating pressure conditions. The present work shows the successful validation of the CFD model in terms of time-averaged temperature and velocity data within measurement errors at an operating pressure of 3 bar. Several well known global mechanisms are tested in this work, namely the Westbrook Dryer 2-step (WD) scheme, the Jones and Lindstedt 4-step (JL4) scheme, the Meredith et al. 3-step (M3) scheme and a recently developed in-house 4-step scheme (M4) for methane-air mixtures. The M4 scheme is optimized by matching the detailed GRI-Mech 3.0 mechanism in terms of 1D laminar flame speed, using the CHEMKIN software for a wide range of pressures (1 to 6 bar), unburned gas temperatures (295 to 650 K) and equivalence ratios range (0.4 to 1.6). CFD simulations are performed using the Eddy Dissipation Model (EDM)/Finite Rate Chemistry (FRC) non-premixed turbulence chemistry interaction model. Both steady-state Reynolds Averaged Navier Stokes (RANS) and hybrid Unsteady Reynolds Averaged Navier Stokes /Large Eddy Simulation (URANS/LES) turbulence models are used. The LES Wall Adaptive Large Eddy-Viscosity (WALE) model with finite rate chemistry is also tested for validation. Velocity profiles, flame temperatures and major species are compared with experiments for different global reaction mechanisms used with different turbulence models. A reasonable agreement is found with the M4 global reaction mechanism in predicting mixing, temperatures and major species. RANS simulations are observed to underpredict the temperature profiles downstream and overpredict in the upstream region, while the velocity profiles are found to be in close agreement with experiments. The SAS-SST turbulence model predicts the velocity profiles in good agreement with experimental data and slightly better than the RANS model. Both the transient simulations slightly overpredict the temperature profiles. The LES-WALE model gives too high and unrealistic temperatures.