| OCR Text |
Show 10 results are shown for comparison to the data of Snyder et ale (1994), since they represent earlier work by the same group. These results fall between the high and low penetration data of Snyder et ale (1994). Two conclusions from the results of Figure 5 are the following: 1. The behavior of the premixer has a significant effect on the NOx. Data obtained with well premixed systems can serve as a guide to the lowest NOx emissions which are possible for given engine combustor conditions. 2. Some combustors show significant pressure dependency of the NOx. This dependency appears to increase as the flame temperature increases. However, a general understanding of the effect of pressure in lean-premixed combustion is not available from the data examined. 6. RESULTS FROM MODELING Three types of computer code are used in our work to model NOx behavior. These are: 1) chemical reactor code, which consists of selected arrangements of PSR (perfectly stirred reactor) and PFR (Plug flow reactor) zones, 2) one-dimensional flame code, and 3) a stirred reactor code, which includes the effects of finite rate mixing. The chemical reactor code is used in this paper to assess trends in NOx fonnation (Figures 6, 7, and 8), and is used in the analysis of our stirred reactor data (Steele et al., 1994). The I-D flame code (as well as the chemical reactor code) is used to develop and verify reduced and global chemical kinetic mechanisms. The reduced and global mechanisms are used in CFD codes to compute NOx emission, and in the stirred reactor/finite mixing code. The stirred reactor/finite mixing code is our latest effort (Tonouchi and Pratt, 1994). This code is used to analyze the effects of finite rate mixing on the NOx formation in high intensity combustion. In this code, finite rate macromixing is treated by variable backmixing, and finite rate micromixing is treated by coalescence-dispersion theory. Figures 6, 7, and 8 illustrate the application of simple chemical reactor modeling for the examination of NOx trends. In this case the model consists of a PSR zone followed by a PFR zone. The fuel is CH4 , perfectly premixed with the air. In Figure 6, the exponential behavior of NOx with flame temperature is shown for three pressure levels and two inlet mixture temperatures. For these calculations, the effects of the pressure (for 10atm and above) and inlet temperature are secondary compared to that of the flame temperature. For the 1600 to 1900K regime, the 10 to 30atm cases have an average apparent activation energy of NOx formation of 58.7kcal/gmol. In Figure 7, the effect of pressure is shown for 1700 and 1900K flame temperatures, for three combinations of PSR/PFR residence times. This modeling indicates that the NOx pressure dependency is greatest in the 1 to 8atm regime. For the 1700K flames, especially as the PSR component of the residence time is increased, the NOx falls off somewhat as the pressure is increased above 8atm. For all of the cases, the NOx increases as the PSR residence time increases. This is explained below under the discussion of Figure 8. Superimposed on Figure 7 are the experimental results of Snyder et al. (1994). The experimental data are in the range of the model predictions (assuming a 2ms PSR zone), except for the 20atm pressure. Overall, the experimental data show a greater sensitivity to pressure than predicted by the modeling. The modeling, however, does confinn the |
