Dear Experts,

I am solving a natural convection problem(Reyleigh No.=10^6) under steady state condition using cfx5. The residual plots of momentum equations are converging but in oscillatotory manner. The period of oscillation changes with the physical time scale in proportioal way. So should i go ahead with it or try to get rid of this. Is there any possibility of problem being unsteady state problem.

please give me your valuable suggestions.

thanks.

amit

Hi Amit,

I think you have solved your own problem. It is highly likely that your problem has a transient solution, and if you run these type of problems steady state then often you get poor oscillatory convergence.

Glenn

Thanks Glenn.

I will try the problem in unsteady state condition.

Amit

Hi, amit

Do you solve your problem? I am solving the same problem recently. It is a transient problem. It took me about ten hours to finish. I am wondering if there is something wrong and would like to know how long is your solving.

I would like to discuss more with you guys about this problem. I have been struggling on it for quite a while. I couldn't find a steady state solution because the resiudals are oscilating, convergence is never appeared. But using transient procedure, it takes ages for CFX5.6 because of the small real time step to ensure the convergence and the long term convection process. May I ask you how you control the convergence, amit? Regards Forrest

hello Anderw,

It does take a long time for my problem. I think there is no solution for this, because cfx experts says that natural convection problems are often diffcult to converge. So better to a have a properly converged solution in longer time instesd of anything wrong.

regards Amit

Dear Forrest,

The things are very similar with me. So I prefer to wait for sufficient time to get properly converged solution. Right now I don't have any better option. I will inform you if i get anything better.

regards Amit