CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   STAR-CCM+ (http://www.cfd-online.com/Forums/star-ccm/)
-   -   Increasing performance changing AMG parameters in Star-CCM+ (http://www.cfd-online.com/Forums/star-ccm/98943-increasing-performance-changing-amg-parameters-star-ccm.html)

archeo March 22, 2012 08:51

Increasing performance changing AMG parameters in Star-CCM+
 
Hi. I am simulating a rotating machine with heat tranfer and heat conduction. The pressure takes long before converging. Is there any parameter in the AMG solver I could change to increase the performances? (the BicGstab is already on)

ryancoe March 23, 2012 10:23

For the Multi-Grid, I think that you may be able to increase the group size to improve convergence speed.

I would also experiment with changing the under-relaxation factors.

archeo March 26, 2012 05:13

I saw some improvment in the convergence behavior when increasing the group size. Thanks for your advice. I changed the group size for the pressure. Do you think I could expect further improvments when changing the group size on the momentum? Does the group size affect the memory requirement?

ryancoe March 26, 2012 14:09

I'm not sure whether increasing the group will directly affect the memory requirement.

Multigrid methods take advantage of the ability of coarser grids to damp transients more quickly. Without knowing more about your simulation its tough to say how the momentum performance may change. I guess its worth a try, right?

arjun March 26, 2012 18:45

Quote:

Originally Posted by archeo (Post 350881)
Hi. I am simulating a rotating machine with heat tranfer and heat conduction. The pressure takes long before converging. Is there any parameter in the AMG solver I could change to increase the performances? (the BicGstab is already on)


By pressure do you mean continuity( mass convergence)?

If yes then there are lots of other factors that affect it. The heavy impact comes from skew or non-orthogonality. If possible improve the mesh.


All times are GMT -4. The time now is 20:38.