[fran.cesc]

Good day,

Would you consider the attached plots as those of a converged solution?

My concerns are:
- Is it normal that I need like 200 iterations for p, while only a few for the other variables?
- Is it normal that the p residual is kind of fluctuating?
- Is it normal that the p residual is like 2 order of magnitude higher than the other residuals?

Note:
- the discontinuities at about 1.4s is due to the fact that I changed the time step on the go.
- I have got a time variant BC, with the input velocity that raises from 0 to its final value at 2s.

Thanks

[fedenr]

Hello fran.cesc,
In my opinion the residuals look good. Regarding your doubts, it is normal that the residuals fluctuate as long as the amplitude is small, just as the number of iterations for each variable changes between cases and solvers. At least that has been my experience, someone correct me if I am wrong.
As for the convergence, I think it is better to check it by analyzing how much some variable of interest of your problem varies, such as CL and CD if it is an airfoil, Instead of looking only at residuals.

[DevilX]

Hey there,


let me give my 2 cents about your case.

- without a log file, case description, solver/schemes settings its hard to evaluate your case
- p-Iterations of 200 is really high - shouldn´t be so, mostly due to fvSolution settings (tolerance 0.0 maybe?)
- p residual fluctuating can be okay, but not in this range of 0.01, this is to huge to let it be okay with it. What solver are you using?
- so pressure boundary conditions maybe also a problem here
- transient cases are robust and stable, you should go to 1e-3 with residuals at least (its possible, look at your other residuals)
- timeStepErrors are fine
- monitor your variables of interest (massflow, forces, liftcoefficent...) and evaluate convergence on these, do not focus only on residuals

[fran.cesc]

Thanks for your reply

Originally Posted by DevilX

- without a log file, case description, solver/schemes settings its hard to evaluate your case

It is an external flow case. The fluid is air. Please see attached for a screenshot of the geometry. I am also attaching the main case files.

Originally Posted by DevilX

- p-Iterations of 200 is really high - shouldn´t be so, mostly due to fvSolution settings (tolerance 0.0 maybe?)

I have got tolerance 1e7 and relTol 0.01; but I have set that for pFinal relTol goes to 0.

Originally Posted by DevilX

- p residual fluctuating can be okay, but not in this range of 0.01, this is to huge to let it be okay with it. What solver are you using?

In order to solve p, I am using the GAMG solver.

Originally Posted by DevilX

- so pressure boundary conditions maybe also a problem here

For the pressure BC, I am using zeroGradient at the inlet and fixedValue (0) at the outlet. At the top and bottom, I am using a symmetry condition; whereas I use zeroGradient on the walls.

Originally Posted by DevilX

- transient cases are robust and stable, you should go to 1e-3 with residuals at least (its possible, look at your other residuals)

I have also selected very robust and 1st order settings, but still...

Originally Posted by DevilX

- timeStepErrors are fine
- monitor your variables of interest (massflow, forces, liftcoefficent...) and evaluate convergence on these, do not focus only on residuals

Thanks for your suggestion.



I have attached the main files of my case. Any suggestion about what the problem you were mentioning could be please?

Thanks

Kind Regards