No abstract

The major exhaust gas pollutants from heavy duty gas turbine engines are CO and NOx. The difficulty of predicting the concentration of these combustion products originates from their wide range of chemical time scales. In this paper, a combustion model that includes the prediction of the carbon monoxide and nitric oxide emissions is tested. Large eddy simulations (LES) are performed using a compressible code (OpenFOAM). A modified flamelet generated manifolds (FGM) approach is applied with an artificially thickened flame approach (ATF) to resolve the flame on the numerical grid, with a flame sensor to ensure that the flame is only thickened in the flame region. For the prediction of the CO and NOx emissions, pollutant species transport equations and a second, CO based, progress variable are introduced for the flame burnout zone to account for slow chemistry effects. For the validation of the models, the Cambridge burner of Sweeney et al. (2012, “The Structure of Turbulent Stratified and Premixed Methane/Air Flames—I: Non-Swirling Flows,” Combust. Flame, 159, pp. 2896–2911; 2012, “The Structure of Turbulent Stratified and Premixed Methane/Air Flames—II: Swirling Flows,” Combust. Flame, 159, pp. 2912–2929.) is employed, as both carbon monoxide and nitric oxide [Apeloig et al. (2016, “PLIF Measurements of Nitric Oxide and Hydroxyl Radicals Distributions in Swirl Stratified Premixed Flames,” 18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics, Lisbon, Portugal, July 4–7.)] data are available.

This work contributes to the understanding of mechanisms that lead to increased carbon monoxide (CO) concentrations in gas turbine combustion systems. Large-eddy simulations (LES) of a full scale high pressure prototype Siemens gas turbine combustor at three staged part load operating conditions are presented, demonstrating the ability to predict carbon monoxide pollutants from a complex technical system by investigating sources of incomplete CO oxidation. Analytically reduced chemistry is applied for the accurate pollutant prediction together with the dynamic thickened flame model. LES results show that carbon monoxide emissions at the probe location are predicted in good agreement with the available test data, indicating two operating points with moderate pollutant levels and one operating point with CO concentrations below 10 ppm. Large mixture inhomogeneities are identified in the combustion chamber for all operating points. The investigation of mixture formation indicates that fuel-rich mixtures mainly emerge from the pilot stage resulting in high equivalence ratio streaks that lead to large CO levels at the combustor outlet. Flame quenching due to flame-wall-interaction are found to be of no relevance for CO in the investigated combustion chamber. Post-processing with Lagrangian tracer particles shows that cold air—from effusion cooling or stages that are not being supplied with fuel—lead to significant flame quenching, as mixtures are shifted to leaner equivalence ratios and the oxidation of CO is inhibited.

No abstract

No abstract