Large eddy simulation/thickened flame model simulations of a lean partially premixed gas turbine model combustor

Abstract

Large-eddy simulation with the thickened flame model (LES/TFM) is conducted to simulate a three-dimensional dual swirl partially premixed methane/air gas turbine model combustor. Finite-rate chemistry is described by a skeletal chemical mechanism consisting of 16 species and 41 reactions. Flame sensors based on formyl radical (HCO) and chemical explosive mode analysis (CEMA) are proposed and implemented within the TFM framework. The two sensors are designed for multi-step chemical kinetic models to avoid thickening the low-intensity heat release rate (HRR) region. One-dimensional freely-propagating laminar premixed flames are first employed to assess the two new sensors. The HCO-based sensor can successfully avoid the low-intensity HRR region for lean flames, but can fail under fuel-rich conditions. The CEMA-based sensor can robustly avoid low-intensity HRR regions under both fuel-lean and fuel-rich conditions. A second test case using a hydrogen/air reheat burner further demonstrates the robustness of the CEMA-based sensor. The two new sensors are subsequently applied to the gas turbine model combustor, and effects of different flame sensors are studied. Baseline results from the HCO-based sensor are first compared with experimental measurements to validate the LES/TFM solver. The mean and r.m.s. velocity, temperature, and mass fractions of O and CO agree reasonably well with the experiment, although the mixture fraction within the inner recirculation zone (IRZ) is under-predicted. The predicted mean and r.m.s. temperature and species profiles are comparable at most locations, except near the IRZ where the CEMA-based sensor predicts the largest fluctuations by only thickening the chemically explosive regions. After optimisation, only 15% of overhead in computational cost is imposed when the CEMA sensor is employed on-the-fly. Future work includes further reduction of the computational cost of the CEMA-based sensor.