OCR Text |
Show AFRC - 1996 Int. Symposium Baltimore The curve given in figure 6 may be modeled as: NOx (g/GJ) = 9 10-5 exp (6.5 10-3 * TaJ with Tad the adiabatic flame temperature in K. September 30 - October 2 1996 Page 11 of 14 g/GJ (1) It is instructive to compare the curve on the NOx emission data on premixed flames with the one presented by Visser and Levinsky (1993). In that paper on practical premixed combustion systems, the NOx production was also plotted against the adiabatic flame temperature. The data presented here coincide with the maximum values of the band of NOx values given by Visser and Levinsky, but the temperature dependence of NOx is identical. The curve-fit presented in that work yields: NOx (ppm, 0% O~ = 8 10-5 exp (6.5 10-3 * TaJ ppm (2) Since one g/GJ is equivalent to 2 ppm (at ) % ~), it may be concluded that the NOx emissions in the experiments presented in this paper are about twice the emissions gathered in practical combustion equipment. This may be explained by the present absence of flue gas andlor air admixture from the environment into the flames under investigation. As on aside, we mention that, the same factor of 6.5 10-3 has been found in turbulent (non-premixed) natural gas diffusion flames as well (Visser and Bahlmann, 1994). Figure 7 (next page) plots the measured NOx production versus the reciprocal of the adiabatic flame temperature (in Kelvin). It shows that for temperatures above 1550°C, (1/Tad < 5.5 10-4) the NOx production is only dependent on the flame temperature. The logarithmic plot of the NOx production also shows that at lower flame temperatures the data with initial conditions: 400°C and 16% 02, give lower NOx production than the other data points. The data with this minimum NOx production have been obtained in two independent series of experiments with an air factor > 1.5 (see also figure 5). Because these mixtures have the highest O2 concentrations at these low temperatures, it is tempting to explain this difference to different amount of prompt (Fenimore) NOx being formed. Assuming that the thermal NOx production may be written as: (3) the term E/R may be estimated from figure 7 as: E/R = 25000 (11K). This gives E :: 200 kJ/mole, substantially lower than the 319 kJ/mole expected for the Zeldovich mechanism. However, the observed temperature dependence should be considered as an "integral" dependence that contains the effects of the Zeldovich mechanism, including the effects of super-equilibrium oxygen atoms (Bowman, 1992), and a contribution from the Fenimore |