Numerical Simulations on the Effect of Operating Parameters on NOx Production in an Industrial Flat Glass Furnace

Page 9

OF NOx PRODUCTION IN A FLOAT GLASS FURNACE the furnace caused by the reduction of nitrogen in the furnace is significant enough to increase the N O x formation by more than 9 5 % from 2161 to 4255 p p m as well as from 12.1 to 23.8 lb/ton of glass. Although the presence of the lance may be viewed as a staged combustion approach, which should ameliorate the N O x formation problem, the increase in gas temperature certainly offsets that benefit and the result is a significant increase in the N O x formation with the oxygen lance. Note that the 3 0 % 0 2 lance case suggests that a greater fraction of the oxygen delivered to the flame later in the combustion process may offer benefits in terms of reduced N O x reduction. Average radiative heat flux to the glass also decreased when increasing the overall theoretical air, indicating a decrease in the heat transfer efficiency in the furnace. The effect of the local concentration of oxygen in the overall N O x formation is also illustrated by the lance calculation at a reduced excess air setting (107% T A ) . In this case, although the exit temperature is about the same for both 1 0 % lance cases, the N O x level dropped for the richer case. Effect of Oxy/Fuel Firing The effect of burning with 1 0 0 % oxygen firing versus that of burning with preheated air can be illustrated by comparing the data shown in Table 1 for these two cases. It is important to note that the simulation for oxygen firing is that for a full furnace whereas the case for the preheated air combus­tion is a single port module as previously explained. The N O x generated per ton of glass for a full furnace was calculated for the port-module simulation assuming that those conditions prevailed in the entire furnace. Although that is certainly not the case and grid resolution is also a concern, the com­parisons made provide interesting qualitative trends which have been supported by experimental ob­servations in furnaces burning with pure oxygen or partially enriched situations. Also, in order to illustrate the effect of having nitrogen still present, either due to air leakage into the furnace or amounts present as nitrogen impurity in the oxygen-generation process used. N o significant differences were observed in the average exhaust temperatures for both cases. The average temperature was only slightly lower for the pure oxygen case. The radiative heat flux to the glass was also slightly lower for that case although heat flux uniformity did not change appreciably. N O x formation increased as indicated by the average N O x concentration at the exit of the furnace; however, given the significant reduction in the volume flow in the exhaust due to the fact that the ni­trogen flow has been removed, the amount of N O x formed per ton of glass is reduced by more than 7 0 % . This is a significant impact on the N O x production per ton of glass and explains the reason why most glass furnace operators are seriously considering switching to pure oxygen firing. It is clear that such an approach is not also without its challenges. First, the amount of water vapor in the ex­haust is increased significantly. In general, this makes the corrosion problem more exacerbated, but most importantly related to corrosion of the furnace refractory. The cost of the oxygen generation plant required on site is also significant, but this factor may be offset by the significant reduction in the cost of rebuilding the furnace because the regenerators are not longer needed. As shown by the simulations, if nitrogen concentration in the oxidizer is increased, the amount of N O x generated is also increased appreciably. CONCLUSIONS The effects of several important furnace operating and design parameters in an industrial, flat-glass furnace have been numerically investigated. These include the effect of (1) soot; (2) an oxygen lance supplying part of the oxygen needed for the combustion process; and (3) firing with oxygen. Comparison between the simulations were performed mainly based in figures of merit including the average exhaust gas temperature, average radiative heat flux to the glass, average N O x concentration at the furnace exit as well as N O x formed per ton of glass produced, and heat flux uniformity to the glass.