Natural convection (Convergence)
 

Hi!

I am studying Natural convection between a steel cylinder to -12.5 ° C (boundary condition:wall) and the atmosphere at standard conditions (pressure outlet boundary condition).

The mesh is in 2D.

I am doing the simulation in Steady and using air as Boussinesq with a thermal expansion coefficient value of 0.00333.
As the Rayleigh number is greater than 10^8 I lowered the gravity and I used the first order scheme. This way I get a convergent solution. I am slowly going to approach the gravity value of 9.81, but when the gravity is 6.5 , I don´t obtain converged solution.

I don´t Know where is the mistake.



Any idea what is wrong?
As I can solve it?

Thank you very much!

Could you describe the computational domain (a picture would be nice)?

Do you use the pressure outlet boundary condition for all exterior boundaries?

What does your mesh look like and in which range are the Y+ values?

Hi,

I send a picture where you can see the boundary conditions.

Thank you very much!

Did you use any prism layers for the mesh around the solid walls?
These are crucial to achieve convergence especially when heat transfer is involved.

I´m sorry, but I don´t Know what prism layers are. I´m learning to use the FLUENT and ICEM with a tutorial.

I created a regular mesh and after I used the command Ogrid in parts close to the walls.

You can see it in the next file:

The Ogrid creates the prism layers, the mesh looks ok.
Although it could be refined close to the solid wall and coarsened further away, but this is not likely cause the convergence problems.

If you want you can upload your file so I can have a closer look at it.

I´ve tryed to send the file, but I can´t send this type of file.

I don´t Know if the problem is to use "Pressure outlet" like boundary condition in the mesh to represent the atmosphere.

Thank you for all.