OCR Text |
Show 7 5. COMPARISON OF NOx FOR DIFFERENT BURNERS Stirred Reactor Compared to Other Burners It is infonnative to compare the NOx measurements of a wide range of lean-premixed burners, including those of the stirred reactor, those of other laboratory research burners, those of prototype burners, and those of industrial-scale burners run at simulated engine conditions. These data cover a significant range of conditions, and include the effects of pressure and degree of premixing not covered above in the stirred reactor experiments. The data are taken from the recent literature, and include the work of individuals associated with either the gas turbine engine manufacturers or the natural gas industry. For all cases the fuel is either natural gas or methane. The flame temperature is the measured temperature in the case of the stirred reactor data; and in the case of the other studies, it is either stated or assumed to be the adiabatic equilibrium flame temperature. In Figure 4, the stirred reactor results for the nominal residence times of 1.7/1. 9ms and 6.0/6.9ms are compared to the data from the literature. Except for one of the literature cases, the data are for air and mixture inlet temperatures in the 580 to 690K range. The first literature data plotted are those of Bahlmann and Visser (1994). These data are for a prototype Annular Low Emission Combustion run at atmospheric pressure with the fuel added to only the primary mixer. We computed the adiabatic equilibrium flame temperature assuming Groningen gas and using the values of fuel-air ratio given in the paper. The low values of NOx obtained in this case relative to those of the other burners indicate that the true flame temperature may have been lower than the computed flame temperature. The possibility of this is suggested by the convective cooling used for the combustor. There is also the possibility of an error in our interpretation of the data. The data of Sattelmayer et al. (1990) are for the ABB double-cone burner (late-1980s model) run at atmospheric pressure. In this case we computed the adiabatic equilibrium flame temperature assuming methane and using the fuel-air equivalence ratios given in the paper. The exponential curve fit is: NOx ppmvd (15% 02) = (1.03E+ 7)exp(-25375!f(K)) The curve fit of Leonard and Stegmaier (1993) pertains to laboratory burner results obtained with a nearly perfect premixer and a variety of flame holders. A wide range of conditions was tested, including: pressures of 1 to 30bar, inlet air temperatures of 300 to 800K, and residence times of 2 to lOOms. The NOx depended almost exclusively on the flame temperature, which is assumed to be the adiabatic equilibrium temperature. Leonard and Stegmaier do not indicate a significant sensitivity of the NOx to the other variables: pressure, inlet temperature, residence time, and flame holder design. The Leonard and Stegmaier results are close to those of the earlier work by Leonard and Correa (1990). We have matched the following exponential equation to the Leonard and Stegmaier curve fit: NOx ppmvd (15% 02) = (2.635E+ 7)exp( -2869S!f(K)) Visser and Levinsky (1993) examined NOx data from a number of lean-premixed gas burners and developed a curve fit (based on 0% 02). Over the temperature range plotted |