CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Phoenics (http://www.cfd-online.com/Forums/phoenics/)
-   -   convergence with migal (http://www.cfd-online.com/Forums/phoenics/51847-convergence-migal.html)

TN January 6, 2005 07:00

convergence with migal
 
Hi all, I am currently trying to use migal to model flows within and around a building with various k-e models. My grid is highly stretched and dense around the structure and I am having trouble with diverging solutions particularly with the higher order schemes and k-e rng model. could someone help?

leon mills January 11, 2005 01:09

Re: convergence with migal
 
TN,

I would suggest that the problem is most likely in the grid. I have had problems getting a converged solution with highly streched grids, and when grid size changes dramically in the domain.

I try to keep the length of the grid to less than 4 times the width / height of the cell. This number seems to work most of the time for me.

I also tend to create transition zones in the mesh. Dense near the building, more coarse for a zone around the dense mesh, then the coarse mesh away from the building.

For me this approach tends to minimise the number of cells used and maintains good convergence and captures the information where it is important.

Hope this helps.

Leon

TN January 14, 2005 05:45

Re: convergence with migal
 
Thanks for your advice Leon,

Your suggestion has helped. My problem lies moreso with the actual tuning of Migal. When I'm approaching convergence the solution diverges dramatically. It seems that there are so much parameters within this solver that it is hard to differentiate between those that will help and those that will just slow down divergence. I can't avoid high cell aspect ratio in some regions due to the use of an inclined narrow object. Maybe this is the reason. Has anyone experience with migal for this type of problem?


All times are GMT -4. The time now is 00:36.