Semi–Empirical Predictions and Correlations of NOx Emissions From Utility Combustion Turbines

Abstract

Semi–empirical equations model the dominant subprocesses involved in pollutant emissions by assigning specific times to the fuel evaporation, chemistry, and turbulent mixing. They then employ linear ratios of these times with model constants established by correlating data from combustors with different geometries, inlet conditions, fuels, and fuel injectors to make a priori predictions. In this work, thermal NOx emissions from two heavy–duty, dual fuel (natural gas and fuel oil #2) diffusion flame combustors designated A and B operating without inert injection are first predicted, and then correlated using three existing semi–empirical approaches termed the Lefebvre (AHL) model, the Rizk–Mongia (RM) model, and the characteristic time model (CTM). Heterogeneous effects were found to be significant, as fuel droplet evaporation times were required to align the natural gas and fuel oil data. Only the RM model and CTM were employed to study this phenomenon. The CTM achieved the best overall prediction and correlation, as the data from both combustors fell within one standard deviation of the predicted line. The AHL and RM models were not able to account for the geometries of the two combustors. For Combustor A the CTM parameter correlated the data in a highly linear manner, as expected, but for Combustor B there was significant curvature. Using the CTM this was shown to be a residence time effect.