| OCR Text |
Show for having this wall in the furnace is obviously not related to this study but we do not want to get into this story in that paper. It was anyway an available experimental set up for studying NOx • ~e ran the model for 6 different 02 contents and we observed as expected that temperature profiles are very different from one case to an other (figure 4). high oxygen ratio makes the flame burning closer to the burner. Maximum temperature increases with higher oxygen content (from 1950 to 2550 K ) and fumes temperature too ( from 1890 to 2140 K ). The residence time in the furnace is increasing with increasing 02% since temperatures are increasing and fumes volumes decreasing (table 1). Table 1 : Residence time for the 6 different cases 02% in the comburent 21.8 36.7 52.6 66.7 93.0 100. Residence time ( ms ) 110 156 206 261 294 295 Now looking at the NOx concentration profiles (figure 5), one can see that NOx formation appears earlier when 02 content and temperature increase (between cell 8 and 15 or between 56 and 105 cm ) but the remaInIng NOx amount in the exhaust gas ( cell 34 ) decreases when the 02 content in the comburent goes over a value close to 60%. ~hat is interesting here is that using the AIR LIQUIDE simulation, we were able to predict this bell-shaped curve with a very good agreement (figure 6). The differences between computed predictions and experimental results were always less than 40%. The computed and experimental maxima of NOx concentration were found to be in both cases around 60%. More the simulation allows us to put numbers on temperature profiles and residence times and gives us information for a better understanding of the experimental results. IV NOx emission predictions for a glass melting furnace Since our computational method has been validated, we tried to simulate a glass melting furnace in order to compare NOx production from preheated air and pure oxygen processes. An advantage of modeling is that one can change only one parameter ( comburent 02 content here ) and keep any othei parameters constant (heat trasfered to the melt for instance ). It is Quiet impossible to do it neither on an actual furnace, neither on an experimental set up. As shown in the previous section, feeding the furnace with 25°C pure oxygen instead of 25°C air makes only the NOx level doubled (from 235 to 456 ppm). Knowing that NO formation is very sensitive to the gas temperature, we thought that it was interesting to look at the air preheating effect. In order to make a comparison possible between the 3 different cases ( 25°C 100% 02' 600°C air and 1100 0C air) we recalculated each time the burner power needed to transfer the same energy to the melt. For all calculations the glass melt was 1550°C (1823 K). Figure 7 and 8 show respectively the flame temperature profils and the NOx concentration profils as a function of the comburent type. All computed results obtained are summarized in table 1. Looking at these values, one can state that preheated air is much more NOx |
