Detailed emissions prediction for a turbulent swirling nonpremixed flame

This paper describes the validation of a previously described CFD-CRN method (computational fluid dynamics-chemical reactor network) for emissions predictions for the laboratory benchmark TECFLAM S09c flame. It details CFD simulation, solution strategy, validation using, CRN generation, detailed emissions predictions and reaction pathway studies. Steady-state 3D CFD models of a 45 degrees sector of the combustor, employing standard numerical techniques; steady-state k-omega SST turbulence, PI radiation, finite-rate eddy-dissipation turbulence-chemistry interaction, and a three-step methane combustion mechanism, were created in ANSYS FLUENT v14. Steady-state models were used as they are of interest to industrial researchers, who are often limited to their use by extremely complex geometries. The models differ in their handling of pressure-velocity coupling and discretization of the momentum equation. The solution which uses SIMPLE coupling and second-order upwind discretization of the momentum equation was generally found to give better results. Satisfactory agreement with experimental profiles for velocity, turbulent kinetic energy, temperature, and species mass fractions has been achieved. Some errors are seen in temperature and CO mass fractions at the highest temperatures (T > 2000 K) and are due to the fact that the highly simplified three-step kinetic mechanism employed due to practical limitations on computational resources, underestimates CO2 dissociation. The results compare satisfactorily with the state-of-the-art. Nonzero concentrations of CO are predicted in the external recirculation zone, which has not been achieved in previous modeling efforts. The validated solution was used as the basis to generate a CRN using the CFD-CRN method. The CFD-CRN method uses user-defined criteria to divide the CFD domain into a set of interconnected perfectly stirred reactors (PSRs) (i.e., a CRN). This CRN is then solved using detailed chemical kinetic mechanisms in the Kinetic Post-Processor (KPPSMOKE) solver. CRN size-independence studies are performed, and it is determined that 5000+ PSRs were needed to adequately capture pollutant formation in the complex recirculating flow field. Validation of CFD-CRN predictions of major species show similar accuracy to steady-state CFD simulation, with improvements over state-of-the-art CFD in CO profile predictions. Satisfactory agreement with previously published experimental NOx contour plots is also seen. The CFD-CRN is used to study NOx formation pathways in the swirling, turbulent environment of the S09c diffusion flame. As expected, NO seen to form primarily in the high-temperature internal recirculation zone (IRZ). The prompt pathway is predicted to be of greatest importance in this area. Significant NOx reburning is seen in the low temperature fuel air jets immediately adjacent to the IRZ. Overall, the prompt pathway is responsible for 77\% of NOx leaving the system, with 12\% due to thermal and 11\% due to N2O intermediate.The authors are grateful to the Irish Research Council (IRC) and Rolls Royce Canada for cofunding this work with an Enterprise Partnership Scheme Postdoctoral Fellowship Award. Politecnico di Milano acknowledges the financial support of the European Union (EU) as part of the EMICOPTER Project (CS-GA-2009-1261 251798).peer-reviewe