[xlworn]

I’ve tried to do some calculation of natural convection heat transfer using CFX 12 and FLUENT 12, but it cannot reach convergence. My geometry is similar with the one you described in those papers. It starts from a single vertical tube in a large tank, tube length is 1m, diameters are 19mm and 6mm, tank is 1m×0.2m×0.2m.
.


1、 diameter 19mm, in CFX.
Transient, time step 0.6s
Boundary condition:
Top wall free slip, and adiabatic. Bottom wall no slip, and adiabatic. Sidewall no slip, and adiabatic. Tube surface wall no slip, and constant temperature 323.15K.
The property of water is constant. Boussinesq assumption is applied. Gravity is 9.8m/s2

That is RMS plot of momentum and mass. And also the RMS of heat transfer and turbulence are all below the 1.0e-05.
I have a temperature contour on a clip plane at 200s

It looks good, so I start to calculate the situation of tube diameter is 6mm.
1、 diameter is 6mm, in CFX
all of other settings are the same with the settings in diameter is 19mm except the mesh.
And the result is not good

Although RMS is below 1.0e-05, it has a fluctuation at about 170 time steps. What confused me is the temperature contour on clip plane, it looks like that there is no variations of temperature of the water around the tube.

100s

200s
Temperature contours at 100s and 200s are almost same, it is hard to understand. So could you please give me some advice?

Ps: Y+value for diameter=19mm is about 5, and Y+ value for diameter=9mm is about 0.8

[ghorrocks]

http://www.cfd-online.com/Wiki/Ansys...gence_criteria