Comparison of NOx prediction methodologies for gas turbine combustor simulations

Requirements to reduce the emissions of nitrogen oxides (NOx) from gas turbines used in aircraft propulsion and stationary power generation have led to consideration of several low-emission combustor concepts, e.g., lean premixed/prevaporized (LPP), lean direct injection (LDI), and rich burn/quick mix/lean burn (RQL). As part of the engineering research, design, and development of these combustors, multi-dimensional combustion computational fluid dynamics (CFD) calculations are increasingly being utilized. To examine the ability of combustion CFD to accurately predict emissions, a series of calculations have been performed with STAR-CD, a general purpose commercial CFD package capable of treating three-dimensional flows with turbulence, sprays, and chemical reactions. Some of the calculations shown in this paper make use of a new coupled n-step chemistry solver combining features of CHEMKIN with STAR-CD which is being developed jointly by Reaction Design and adapco. The NOx predictions are made via two techniques. In the first, a joint scalar PPDF thermal NOx postprocessor is employed in conjunction with a two-step combined timescale combustion model. In the second approach, a reduced 26-step finite rate kinetics mechanism is utilized. Here, the combustion chemistry is combined with the NOx formation chemistry based on the extended Zel'dovich mechanism for thermal NOx and a surrogate species approach for prompt NOx. Comparisons to experimental data show that both approaches can produce reasonable predictions.