No abstract

Natural convection currents in a rectangular two-dimensional enclosure representative of the longitudinal section of an industrial glass-melting furnace have been established by both model experiments and numerical calculation. For the latter a finite-difference method has been employed to solve the time-dependent coupled flow and energy equations. The highly generalized mathematical model makes allowance for buoyancy, temperature-dependent viscosity, and diffusive radiation. Generalized boundary conditions are employed to permit specification of any combination of temperature, flux, or mixed thermal boundary conditions. Representative temperature and flow contour maps obtained from the calculations are shown to agree well with experimental results obtained with a 1/20 scale model in which glycerine was employed as the modeling fluid.

No abstract

V(7) = matrix of radially dependent velocities V, = effective y-phase velocity when phase interactions are = matrix of effective velocities (Part I, Eq. 21) = matrix of effective velocities in the constant coefficient axial dispersion model ignored Greek Letters E = matrix of hold-up ratios { = normalised axial coordinate 7 = normalised radial coordinate h = Fourier transform variable p = viscosity p =density T = dimensionless time w = matrix of weighting functions for concentration averaging

Laminar natural convection in a two-dimensional enclosure with both source (uniform heat flux density) and sink (temperature specified) located on the top horizontal boundary is investigated numerically. Temperature and velocity profiles are presented for a high Prandtl number fluid for length Rayleigh numbers in the range 107 to 109 for length to depth ratios of 1:1 to 4:1. To generalize the results, an order of magnitude analysis is used to determine the dependence of temperature, velocity, and boundary-layer thickness scales on aspect ratio and Rayleigh number. The numerical data are well correlated using these suggested scales. The analysis shows the Nusselt number and the maximum horizontal velocity to depend on the 1/6 and 1/3 powers of the Rayleigh number, independent of aspect ratio.