Determination of NOx emissions from strong swirling confined flames with an integrated CFD-based procedure

The present work is focused on a new procedure for the determination of NOx emission from combustion processes, which allow using very detailed and comprehensive reaction schemes, on the basis ofthe results obtained from CFD computations. This procedure is validated in the case of high swirled confined natural gas diffusion flames. The experimental data refer to the work developed within the German TECFLAM cooperation concerning a swirl burner (0.6<S<1.4) fed with natural gas characterized by 150kW thermal load and 0.8 equivalence ratio (TECFLAM webpage, www.tu-darmstadt.de/fb/mb/ekt/tecflam; Schmittel et al., Proceedings ofthe Combustion Institute 28 (2000) 303–309). The CFD analysis represents a useful technology to provide the flow and temperature fields. The high swirling configuration makes the solution an interesting and difficult task for commercial codes. In particular, previous attempts of predicting the flow and temperature fields with the FLUENT code failed (Meier et al., Applied Physics B 71 (2000a) 725–731). Only a careful attention to the boundary conditions and converging strategy allows to reach a satisfactory modeling ofthe main characteristics ofthe flame. The numerical calculation was performed using the commercial code FLUENT6.0. NOx formation is a chemical process whose time-scale is ofthe same order ofmixing fluid dynamics. For this reason, comprehensive modeling ofNO x reaction processes in combustion systems requires simulation ofboth the turbulent fluid dynamics and chemical kinetics in the system being modeled. Hundreds ofelementary reactions are required to provide a detailed description ofthe formation and depletion ofoxides ofnitrogen in combustion systems. However, it is not currently feasible to use such detailed reaction mechanisms to model a turbulent reacting system in which large reaction kinetics schemes are coupled with the turbulent fluid dynamics. Consequently, the difficulties in coupling detailed chemistry and detailed fluid dynamics force to adopt proper simplifications. The prediction ofNO x formation is then obtained by postprocessing the flow and temperature fields, as predicted by CFD, and lumping together computational cells similar in terms of NO x formation. The resulting mìcrocells are assumed to be a network of ideal reactors, which are simulated adopting a detailed kinetics. The characteristics and operating conditions of each reactor are defined by a procedure already developed for furnaces and named SFIRN (Faravelli et al., Computers and Chemical Engineering (2001) 613–618). The predictions have been tested on flames at different swirl numbers. Both the CFD and the chemical analysis show a satisfactory agreement with the measured data.