| OCR Text |
Show 5 and reactor mean residence times of 1.6 to 7.2ms actual. The corresponding nominal residence time range is 1.7 to 6.9ms. The temperature is measured with a Pt/Rh thermocouple, coated to prevent catalytic surface reaction, and corrected by calculation for radiation heat loss. Measurements of temperature in stirred reactors with uncoated thermocouples give erroneous readings (i.e., elevated temperatures because of the surface oxidation) and should not be used. The calculation for thermocouple radiation heat loss, which is based on optical pyrometer readings of the brightness temperature of the thermocouple and the reactor wall, is relatively small (i.e., 30 to 50 degrees C over the conditions of these experiments). The small loss occurs because the stirred reactor approaches a black-body cavity. For the highest combustion loadings (i.e., the shortest residence times) the reactor operates nearly adiabatically, within 20 to 30 degrees C of the adiabatic equilibrium temperature. Gas sampling from the stirred reactor is accomplished with a quartz probe which utilizes both aerodynamic quench and wall water cooling. Gas samples are analyzed for NO, NOx, N20, CO, CO2, 02, and C1-C2 hydrocarbons. The gas sample probe is known to convert CO to C02 and NO to N02. The CO to C02 conversion is accounted for in modeling of the probe chemistry; nominally, about 14% of the CO is converted in the probe of these experiments. Because of the NO to N02 conversion, and the solubility of N02 in water, great care is taken in the experiments to preserve the total NOx. A special dry measurement technique is used from time to time to assure that none of the N02 is absorbed. Accurate measurement of N20 from combustion sources has been a problem in the past. However, those problems were associated with grab samples and the presence of sulfur. The present work utilizes an on-line GC analysis technique (Steele et al., 1994), and sulfur is absent from the gases. In this paper, attention is focused on the NOx data. The other gas sample data are listed by Steele et al. (1994). The temperature and gas sample measurements are taken downstream of the reactor inlet jets, in the bulk recirculation zone of the reactor. This region has uniform temperature and species concentrations. Experimental Results The NOx results for methane are shown in Figure 1. In this figure the NOx is expressed on a wet, actual 02 basis. The data for NOx divided by residence time (tau) are found to correlate well when plotted in Arrhenius-type format versus the reciprocal of the measured temperature. Based on the curve fit shown on the figure, the apparent activation energy of the NOx formation is 48.3kcal/gmol, which is significantly lower than that of thermal NOx. Thermal NOx, including the energy associated with the dissociation of the 02 into O-atom, has an activation energy of about 134kcal/gmol. The data plotted in Figure 1 show that the NOx formed in lean-premixed high intensity combustion zones depends primarily on the combustion temperature and residence time. There is a secondary sensitivity to mixture inlet temperature, which is more clearly seen when the NOx data are adjusted to 15% 02 dry basis. (This is done below in Figures 2 and 3). Insensitivity to reactor nozzle type and reactor size is evident in Figure 1. The insensitivity to reactor size is important, because prior to these experiments, there was concern that the small (2cc) reactor, because of its high surface to volume ratio, would suffer from surface reaction effects. Our work had been criticized because of this possibility (Zelina and Ballal, 1994). However, the fact that NOx data from reactors of 2, 16, and 50cc give essentially identical results gives little weight to this possibility. |
