Abstract:
Numerical investigations have been performed on an incompressible fluid contained in a two
dimensional enclosure. Five identical heat sources were mounted on one vertical wall of the
enclosure with uniform vertical spacing. Two different types of heat sources i.e. protruding
and flush were studied ..The vertical wall opposite to the heated sections and the bottom wall
were adiabatic. The top surface of the test enclosure was an isothermal heat sink.. The
horizontal and vertical dimensions of the protruding heaters were Lg = 9 mrn and L1 = 15 mrn
respectively. The vertical spacing between the heaters was Lz = 15 mill. For flush heater Lg =
o mrn. Investigations were performed with different values of cavity width, resulting in the
variation of aspect ratio (height-ta-width ratio) and cavity width -to-protruding heater height
ratios of 3.67 to 9.17 and 2.0 to 5.0, respectively. The effect of variation of heat flux from the
heat sources, resulting in variation of modified Rayleigh numbers ranging from 1.9 x 106 to
4.9 x 1010 was studied. The investigations were performed with ethylene glycol as the
working fluid.
The governing transport equations for turbulent flow under investigation are closed using the
two equation model of turbulence (k-E), which includes gravity-density interaction. The
equations are discretised using a finite volume technique. A non-uniform grid arrangement of
62 X 62 was used. The governing equations were solved incorporating the Pressure Implicit
with Splitting of Operators (PISO) method.
Results indicate that heat transfer and fluid motion within the enclosure are influenced by the
Rayleigh number, aspect ratio and heat input. At larger width (W=45 & 36mrn) a single vortex
is observed occupying the flow domain and at smaller width (W=22.5 & 18 mrn) the single
vortex structure disintegrates and several secondary flow cells appear within it. Increase in the
width results in a decrease in the difference between the temperatures of the top heated sections.
The temperature gradient in the core flow region are found to be considerably small. The
correlation of local Nusselt number versus local modified Rayleigh number is independent of
the number of heaters in the vertical array and vertical height location of the heaters. The results
show reasonably good agreement with the experimental data.
