Search

A thickened-flame (TF) modeling approach is combined with a large eddy simulation (LES) methodology to model premixed combustion, and the accuracy of these model predictions is evaluated by comparing with the piloted premixed stoichiometric methane-air flame data of Chen et al. (1996, 'The Detailed Flame Structure of Highly Stretched Turbulent Premixed Methane-Air Flames,' Combust. Flame, 107, pp. 233-244) at a Reynolds number Re = 24,000. In the TF model, the flame front is artificially thickened to resolve it on the computational LES grid and the reaction rates are specified using reduced chemistry. The response of the thickened-flame to turbulence is taken care of by incorporating an efficiency function in the governing equations. The efficiency function depends on the characteristics of the local turbulence and on the characteristics of the premixed flame such as laminar flame speed and thickness. Three variants of the TF model are examined: the original thickened-flame model, the power-law flame-wrinkling model, and the dynamically modified TF model. Reasonable agreement is found when comparing predictions with the experimental data and with computations reported using a probability distribution function modeling approach. The results of the TF model are in better agreement with data when compared with the predictions of the G-equation approach. [DOI: 10.1115/1.3094021] Keywords: thickened-flame, flame-wrinkling, turbulence, gas turbine emission, large eddy simulation

The present paper addresses real-time prediction of incipient lean blowout in partially premixed, liquid-fueled gas turbine combustors. Near-lean-blowout combustion is characterized by the intensified, low-frequency combustion oscillations (typically, below 30 Hz). Two indices, namely, the normalized chemiluminescence root mean square and the normalized cumulative duration of lean blowout precursor events, are recommended for lean blowout prediction. Both indices are associated with the statistical characteristics of the flame structure, which changes from the normal distribution to the Rayleigh distribution at the approach of lean blowout. Both indices change little within a large range of equivalence ratios and start to shoot up only when lean blowout is approached. To use the two indices for lean blowout prediction, one needs to perform a detailed a priori lean blowout mapping undersimulated engine operating conditions. However, the mapping can be done without running the engines very close to lean blowout.