| OCR Text |
Show 9 that the zones of highly non-equilibrium chemistry had residence times greater than about 2ms; also, it is likely that any post-flame NOx fonned slowly_ These characteristics should yield a weak sensitivity to residence time. These conditions and characteristics also are seen in the data of Snyder et ale (1994) discussed below. Additional Burner NOx Data Additional burner NOx data are shown in Figure 5. For reference, the curve fit of Leonard and Stegmaier (1993) is shown again. These data show the effects of different flame holders and premixers. The laboratory burner data of Leonard and Correa (1990), Lovett and Abuaf (1992), and Leonard and Stegmaier (1993) have similarities and are discussed together. Leonard and Correa obtained their data with a perforated plate flame holder in a tubular combustor. Their results for 615K inlet temperature and 10.2atm are plotted in Figure 5; the results which they obtained at lower pressures showed about the same NOx behavior as for the 10.2atm data. Lovett and Abuaf, working at the same conditions as Leonard and Correa, tested v-gutter and swirler flame holders. The v-gutter gave nearly identical NOx to that of the perforated plate flame holder (of Leonard and Correa). The swirler gave slightly lower NOx. These results of Lovett and Abuaf (and Leonard and Correa) lie close to the curve fit of Leonard of Stegmaier. However, when Lovett and Abuaf increased the pressure to 13.6atm and the inlet temperature to 650K, the NOx increased. Also, the slope of the NOx versus flame temperature increased, indicating an increased apparent activation energy of the NOx formation. Although these 650K/13.6atm NOx data lie within the range of the data used by Leonard and Stegmaier to develop their curve fit of NOx versus flame temperature, the 650K/13.6atm data suggest that the NOx is determined not only by flame temperature, but is influenced also by pressure and/or inlet temperature. The results of Snyder et al. (1994) are examined next. They examined a prototype gas fuel injection/premixing system, using different penetration patterns of the fuel. The paper indicates that the best mixing of the fuel into the air was obtained with the high penetration system. It appears that the low penetration system did not uniformly mix the fuel into the air prior to combustion. This resulted in a significant increase in the NOx compared to the high penetration system, which yielded quite low NOx for the 19atm pressure of the tests. The high and low penetration systems show different slopes of NOx versus flame temperature. The reason for this behavior is not clear. Whereas the low penetration data have a slope similar to that of other data on Figures 4 and 5, the high penetration data have the highest slope of any of the data examined in this study. The apparent activation energy of NOx formation of the high penetration data is 80.8kcal/gmol (and is treated as an outlier in this study), and that of the low penetration data is 59.3kcaVgmol. Snyder et al. also tested their high penetration system as a function of pressure, at 710K inlet temperature. (Some of these data are included as part of Figure 7 in the modeling section below.) If the Snyder et ale NOx data for constant flame temperature are fit with an expression of the type NOx = apb, we find that b = 0.32 for the 1800K flame temperature, and 0.52 for 1900K. These results support the hypothesis that the pressure sensitivity of lean-premixed NOx increases as the flame temperature increases. Also plotted in Figure 5 are results of McVey et al. (1992). In this work, a perforated plate flame holder with a pilot flame, and an aero-vane flame holder were used. These |
