OCR Text |
Show OF OPERATING PARAMETERS ON N O X PRODUCTION IN A FLAT GLASS FURNACE RESULTS AND DISCUSSION The results and discussion of predictions are now presented. The effect of soot on the simulation of a port module is first presented and discussed. The effect of reducing the overall amount of preheated air and introduction of a oxygen lance is illustrated next and also investigated at two different levels of oxygen enrichment. Finally, exhaust gas temperature, average radiative heat flux, N O x formation, and heat flux uniformity are presented for a full-furnace simulation of oxygen/fuel firing and compared with the same values for the simulation using air. The oxygen-firing condition is further investigated by maintaining the same overall stoichiometry but allowing for some nitrogen to be present. Table 1 shows these figures of merit for the simulations just described. Table 1 also includes average concentration values of C 0 2 , C O , and 0 2 at the exit plane of the simulations as well as an efficiency parameter as previously described (Fuel Utilization Efficiency). This parameter is directly related to the average heat flux to the glass given a fixed enthalpy of formation for the fuel. Effect of Soot As previously explained, the software used in the present study has a subroutine that allows for the inclusion of a soot formation and transport model. This two-step model predicts the generation of soot and subsequent transport by the flow field. The soot parameters were adjusted within acceptable values in the subroutine so as to produce soot concentration levels in the flame zone comparable with experimentally determined values for a methane flame. The presence of soot increases flame radiation to the furnace walls and glass surface, causing the average exhaust gas temperature to decrease from 2229 to 1897 K. The average radiative heat flux to the glass also increased from 107 to 124 kW/m2. Taking as a reference the average exit temperature for the case without soot, This 1 5 % decrease in average gas temperature, as expected, had a significant impact on the N O x formation process. The average N O x concentration value at the exit decreased by more than 5 5 % from 5247 to 2161 p p m while the N O x per pound of glass dropped from 29.4 to 12.1 lb/ton of glass. This significant decrease in N O x formation is associated with the fact that, at the combustion temperature levels present in this problem, the N O x formation rate varies exponentially with temperature as expected from the thermal N O x mechanism. It is interesting to note that the heat flux uniformity also improved for the simulation with the soot. Contours of the soot as well as C O concentrations at the burner level (not shown) clearly showed a much broader and longer flame for the simulation with the soot, causing the improved heat flux uniformity for this and the other simulations shown in Table 1 as compared to the simulations without soot. In fact, the other parameters investigated had a minimum effect on the heat flux uniformity as shown in Table 1. Effect of 02 Lance The existence of oxygen available from other processes in a typical float glass plant makes it an attractive source for the oxygen needed for combustion. In these particular simulations, the amount of preheated air was reduced to accommodate an 0 2 lance supplying 1 0 % of the total oxygen supplied by the preheated air flow while maintaining the same overall stoichiometry. Another case was also investigated by keeping the 1 0 % oxygen split but reducing the overall stoichiometry from 119 to 1 0 7 % theoretical air (TA). Finally, a third simulation increases the oxygen lance fraction to 3 0 % . There are of course some design and installation challenges in the implementation of an oxygen lance in an actual furnace. There are increased costs associated with piping necessary to delivery the oxygen to the furnace; and the lance would most probably have to be water-cooled. The ideal location of the lance is also an important variable which was not parametrically studied here. As shown in Table 1, comparing the lance and air simulations, both simulations including soot formation and transport models with the same values of input parameters, the presence of the lance slightly increases (7%) the average exhaust gas temperature. However, the increase in temperatures in |