[djchris]

Hi, the last 4 months I try to simulate a solar chimney which is installed in a pavilion/shed of a square. The problem is that even I tried to run 18000 iterations, the problem doesn't converge.
- I tried Bousinesq model for air, I also tried the incompressible ideal gas,
- I tried coupled solver and SIMPLE
- In solution controls, I change the density to standard, body forced weighted etc. but nothing.
- I turned off the multigrid
- changed the second order upwind to first order...

What I have to do?

The residuals are under the limits for example the energy is 4.5e-07 but doesn't converge.
Can everyone help, as I need to finish the project until 1 December, it's my M.Sc. Thesis?

[sagila]

Hi,

One of the solutions is increasing gravitational acceleration step by step.
Another solution is defining a good initialization (instead of increasing gravitational acceleration step by step).
Also, often you need to change under-relaxation factors.

[djchris]

Originally Posted by sagila

Hi,

One of the solutions is increasing gravitational acceleration step by step.
Another solution is defining a good initialization (instead of increasing gravitational acceleration step by step).
Also, often you need to change under-relaxation factors.

Sagila thanks for the quick reply,
About the gravitational acceleration, do you mean that I have to start with, for example, 2 m/s^2 and increase it every 100 iterations by 1 point unless 9.8 m/s^2?

I use the hybrid initialization, I can't use the FMG as I use the shell-conduction option.

[sagila]

About gravitational acceleration: yes, you are right, but sometimes you need to start with a value of a lower order. Please be careful about increasing gravitational acceleration after your solution converges.
Also, as I mentioned before, often you need to change under-relaxation factors.

[djchris]

I tried both of them but the solution doesn't converge. The residuals are under the limits, but the solution reaches 12000 iterations with no converge. Also, the variation of the residuals is quiet small but they don't converge..