Analysis of NOX Formation in an Axially Staged Combustion System at Elevated Pressure Conditions

Abstract

The objective of this investigation was to study the effect of axially staged injection of methane in the vitiated air cross flow in a two stage combustion chamber on the formation of NOX for different momentum flux ratios. The primary cylindrical combustor equipped with a low swirl air blast nozzle operating with Jet-A liquid fuel generates vitiated air in the temperature range of 1473–1673 K at pressures of 5–8 bars. A methane injector was flush mounted to the inner surface of the secondary combustor at an angle of 30 deg. Oil cooled movable and static gas probes were used to collect the gas samples. The mole fractions of NO, NO2, CO, CO2, and O2 in the collected exhaust gas samples were measured using gas analyzers. For all the investigated operating conditions, the change in the mole fraction of NOX due to the injection of methane (ΔNOX) corrected to 15% O2 and measured in dry mode was less than 15 ppm. The mole fraction of ΔNOX increased with an increase in mass flow rate of methane and it was not affected by a change in the momentum flux ratio. The penetration depth of the methane jet was estimated from the profiles of mole fraction of O2 obtained from the samples collected using the movable gas probe. For the investigated momentum flux ratios, the penetration depth observed was 15 mm at 5 bars and 5 mm at 6.5 and 8 bars. The results obtained from the simulations of the secondary combustor using a RANS turbulence model were also presented. Reaction modeling of the jet flame present in a vitiated air cross flow posed a significant challenge as it was embedded in a high turbulent flow and burns in partial premixed mode. The applicability of two different reaction models has been investigated. The first approach employed a combination of the eddy dissipation and the finite rate chemistry models to determine the reaction rate, while the presumed JPDF model was used in the further investigations. Predictions were in closer agreement to the measurements while employing the presumed JPDF model. This model was also able to predict some key features of the flow such as the change of penetration depth with the pressure.