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

Page 4

EFFECT OF OPERATING PARAMETERS ON N O X PRODUCTION IN A FLAT GLASS FURNACE 3_ Numerical simulation of an entire float glass furnace is costly and, despite advances in computing power, provides spatial grid resolution which is inadequate. In this study the preferred computational domain is that illustrated in Fig. 2, a single port module extending from inlet portneck to exhaust portneck, at the same time exploiting the symmetry of the flow along the centerline between port-necks (as shown in the figure). Fuel was introduced at each side of the portneck through burner tubes vectored such that the intersection point would be the portneck exit (furnace wall). Simple pipe burner nozzles were simulated. As a compromise between prediction accuracy and computational effort 102,000 cells were employed in the port module simulations. A 98,000-cell grid was used for the full-furnace simulations, although considerable refinement would be required to achieve grid-independent results. Where a single port module was used in the simulations, identical firing among the multiple portnecks was assumed in the calculation of lb NOx/ton of glass. Although this assump­tion is not realistic, it nevertheless provides some basis for comparison between port module and full furnace simulations. MATHEMATICAL MODEL The combustion chamber model used in the study is able to simulate the turbulent fluid flow, ra­diation heat transfer, gas combustion, and pollutant emissions in the combustion space of typical in­dustrial scale float glass melting furnaces. A well-known commercial code, F L U E N T , was used as the basis for the submodels describing the coupled complex phenomena and the numerical solution of the describing equations. The thermal and fluid behavior of the gas mixture inside the combustion chamber is predicted by solving the three-dimensional governing partial differential equations set in steady-state, time-averaged form. Details regarding the modeling approach, theory behind the sub­models, and numerical method may be found elsewhere ( F L U E N T User's Guide, 1995). The com­bustion chamber model includes several sub-models which are summarized briefly as follows: Turbulence Model. The turbulent flow in the combustion chamber has been modeled by the k-e model (Launder and Spalding, 1973, 1973; Rodi, 1984), in which transport equations for kinetic tur­bulent energy k and its dissipation rate £ are solved, and effective viscosity is calculated. Accepted empirical constants associated with this turbulence closure are used in the simulations. Radiation Model. The discrete transfer radiation model (DTRM) (Shah, 1979: Carvalho, 1991) was used in this study for prediction of surface-to-surface radiation heat transfer. The predicted radiant intensity represents an integrated intensity across all wavelengths. In this model, the change in radiant intensity along a path is integrated along a series of rays emanating from a single point in each dis­crete control volume on a surface. This series of rays defines the hemispherical solid angle about that point. The radiation intensity approaching a point on a wall surface is integrated to yield the incident radiation heat flux. The net radiation heat flux from the surface is computed as a sum of the reflected portion of the incident flux and the emissive power of the surface. Chemical Reaction Model. The mixture fraction/PDF modeling approach was used in the combustion simulations, which involves the solution of transport equations for a single conserved scalar (mixture fraction) (Hill and Smoot, 1993). In this approach, transport equations for individual species are not solved. Rather, individual component concentrations for the species of interest are derived from the predicted mixture fraction distribution. Reacting systems are treated using either chemical equilib­rium calculations or using infinitely fast chemistry (the flame sheet or "mixed-is-burned" ap­proach). The interaction of turbulence and chemistry are accounted for with help of a probability density function (PDF) whose shape is specified.