CFD Online Discussion Forums

CFD Online Discussion Forums (
-   OpenFOAM Running, Solving & CFD (
-   -   Convergence difficulties with 2d mixerGGI (

louisgag February 3, 2009 14:24

i am no expert of ggi but perh
i am no expert of ggi but perhaps you can replace p and pfinal brackets by

p ICCG 1e-06 0;
pFinal ICCG 1e-06 0;

christinasmuda February 4, 2009 03:55

Hi Louis, thanks for your s
Hi Louis,

thanks for your suggestion. I changed the solver to ICCG, but the convergence behaviour didn't change at all. Here is my case: rotor2Dggi.rar.

What else could I try in order to improve the convergence?


louisgag February 4, 2009 14:09

Hello Christina, I noticed
Hello Christina,

I noticed your simulation endtime was 0.01 second. It would be surprising to see the velocity distribution you expect show up at 0.01s considering your rpm is 100.

I have relaxed your tolerances and reduced the nonOrtho correctors and am running your case at the moment. I'll update you on this later during the day.



louisgag February 4, 2009 14:44

I was able to get some results
I was able to get some results that somewhat match the velocity distribution you were looking for by changing endtime to 1s, nCorrs to 0 and nonOrthocorrs to 0. Setting those to 0 might be extreme and you might want to play around with these values, but I set them to zero to speed up calculation and considering your mesh was very orthogonal.

Send me an email if you want the modified case along with the results (5mb).

Good luck,


christinasmuda February 6, 2009 03:34

Hi Louis, thanks for sendin
Hi Louis,

thanks for sending the case.

I already tried running the case without any PISO correctors yesterday and I got the same flow field you sent to me. Unfortunately this flow field is wrong, in my opinion. I did the same calculations with fluent and with OpenFoam using sliding interfaces and both times I got the picture, I put in the link.

I'm still not really convinced about the solution you suggested. If I understand the PISO algorithm well, there's no sense in running it without any corrector steps, is there?

But as soon as I set nCorrectors to 1, I get a floating point error quickly. Setting the Courant Number to smaller values (0.25 or 0.1), the calculation is running for a longer time, but still ends up in a floating point error.

What is wrong with my set-up?

Any help is highly appreciated.


louisgag February 6, 2009 13:05

Dear Christina, After sendi
Dear Christina,

After sending you the case yesterday, I also tried to increase the nCorrectors and ran into the same floating point error. It seems that the flow diverges suddenly at some point.

I question whether your mesh could be too small (being 3 mm wide), but checkMesh does not report any cell volume problem...

If I find the problem I'll let you know. Please let me know if you do.

Good luck in the meantime,


All times are GMT -4. The time now is 22:57.