| OCR Text |
Show • Glass is assumed to be opaque to all thermal radiation, and the glass surface temperature is uniform along the furnace. In the actual modeling work, a set of non-linear equations was solved numerically using a finite difference, iterative procedure. The numerical procedure was verified by conducting a grid independence study. Parametric studies were conducted on the effects of firing rate, fuel and soot burnout length, soot volume fraction, inlet oxidant temperature, furnace geometry such as height and length, and other design factors. A number of conclusions were reached based on this parametric study. • With no soot in the inlet fuel, temperatures and heat flux to the load are strongly peaked. • Oxy-gas flames have more peaked local (gas and crown) temperatures and heat fluxes than air-gas flames. • Soot volume fraction at the furnace inlet and soot burnout length are important parameters affecting furnace performance. • When soot concentrations at the furnace inlet are high enough (10'6), luminous flames can have as much as 25%) more heat transfer to the load compared with non-luminous flames under similar conditions. • Flames covering a larger fraction of the glass surface enhance the thermal performance of the furnace. • Short flames are undesirable because they produce highly peaked and nonuniform heat fluxes at the glass surface along with high glass and refractory temperatures. • High soot concentrations in the flame lower the gas temperature while increasing heat transfer. This is expected to decrease N O x production. Crown temperature is slightly increased. The modeling calculations were extended from one dimension to a quasi two dimensional approach to relax the restriction of axial radiation heat transfer. This was expected to provide more realistic results while also showing lower temperatures and less peaked temperatures and heat fluxes. Comparison of the 1-D and quasi 2-D results (Figure 7 and 8) showed the expected trend. The heat transfer to the load is increased by 25%) in both models. The magnitude of the heat fluxes is the same for the 1-D and quasi 2-D models, but the heat fluxes with the quasi 2-D model are lower and less peaked which is partly accounted for by heat losses through the entrance and exhaust ports. Despite the significantly greater effort required to calculate the quasi 2-D model, the quasi 2-D model did not produce any results which were not observed with the 1-D model. 13 |
