EVALUATION OF NOX MECHANISMS FOR LEAN, PREMIXED COMBUSTION

The formation of the oxides of nitrogen, NO(x), is examined through experiments and chemical kinetic modeling for lean, premixed combustion in a laboratory, atmospheric pressure, jet-stirred reactor. The experimental conditions are as follows: fuel-air equivalence ratio (phi) of 0.6, temperatures of 1460 to 1730 K, and reactor loadings of 20 to 150 kg/sec-m3-atm2, which correspond to reactor mean residence times of 11.4 to 1.8 milliseconds. Two fuels are examined: ethylene, because of its importance as a combustion intermediate, and methane, because of its importance as a component of natural gas. Besides the premixed operation, the reactor also is operated nonpremixed. For both modes, the NO(x) increases with decreasing loading, from about 3-4 ppmv at the highest loading to about 11-21 ppmv at the lowest loading for the ethylene fuel. This increase in NO(x) occurs because a hot spot develops on centerline when the reactor is lightly loaded. Also for the lowest loading, the nonpremixed mode produces about twice as much NO(x) as the premixed mode, i.e., about 21 versus 11 ppmv. At the other reactor loadings, however, because of the intense mixing, the NO(x) levels are only slightly elevated for the nonpremixed mode compared to the premixed mode. Upon switching to methane fuel, the NO(x) decreases by about 25 percent. The major finding of this study is that prompt NO is the predominant mechanism for the NO(x) formed. The other mechanisms considered are the Zeldovich and nitrous oxide mechanisms. Furthermore, the amount of NO(x) measured and modeled agrees almost exactly with the extrapolation of Fenimore's (1971) original prompt NO data to the present conditions of phi = 0.6. Although Fenimore conducted his experiments with porous plate and Meker-type burners for 0.8 less-than-or-equal-to phi less-than-or-equal-to 1.7, our findings show that his results apply well to high-intensity, lean combustion. In the gas turbine literature, e.g., see Shaw (1974) and Toof (1985), Fenimore's results are expressed as: NO/(NO)equil = P1/2func (phi) It is func (phi) that extrapolates well to our conditions. This finding indicates that func (phi) applies to laboratory burners of widely different mixing intensity, i.e., from Fenimore's burners with structured flame fronts to our high-intensity burner with dispersed reaction. In our opinion, this finding strengthens the justification of using func (phi) for the prediction of NO(x) formation in practical combustors, including lean, premixed combustors.