Natural Convection in closed hydrothermal system

Scott

I am interested in getting some advice/help in attempting to model the fluid flow in a closed hydrothermal vessel using the Fluent CFD software. I am modeling the vessel architecture through a 2D, axisymmetric model, consisting of two different circular symmetric heat zones on the vertical walls. However, I can not gain convergence of the continuity, x and y velocities and turbulence parameter residuals.

My Fluent model uses a pressure based, steady solver and assumes the boussinesq approximation for natural convection. I have applied the heat zones to the outer solid vessel wall and maintain a coupled BC between solid and fluid regions. I have tried both the k-e and k-omega turbulence models with the standard wall functions and the standard model constants. The Ra no. for my flows is 8 * 10^8, which suggests a laminar to turbulent transition.

My current queries are:

1. The hydrothermal system includes a high value of pressure, assumed to be relatively constant throughout the fluid regime. The only place for the input of this pressure is under the operating conditions. However, convergence seems to be detrimentally effected. Is there a better way of dealing with the pressure?

2. From what I have read, natural convection problems are highly sensitive to the meshing arrangements. I have used boundary layers in the fluid region, at the solid/fluid interface in the meshing scheme, but coupling between fluid and solid regimes presents difficulties.

3. Is the steady state solver the way to go in modeling this type of behaviour?

4. Is the Boussinesq approxomation useful for temperature differences of 50 deg C?

Any help or suggestions would greatly appreciated.

Many thanks

Scott