CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   Convergence Criteria in FLUENT (https://www.cfd-online.com/Forums/fluent/97903-convergence-criteria-fluent.html)

mahi007 February 28, 2012 05:44

Convergence Criteria in FLUENT
 
Hello everyone

I am new to CFD. I am doing analysis of flow past circular cylinder in ANSYS 13. I created a mesh in ICEM CFD and I am currently doing mesh sensitivity study by refining mesh. I set the convergence criteria to 1e-06 as I am doing laminar flow. I am interested in cd values and I am monitoring cd too. Even after residuals meet convergence criteria, cd value is getting changed. I mean each cd value is repeating twice or thrice and decreasing to next value. If I leave it as such, each value is repeating as many as 10 times. When can I stop?

Thank you

delaneyluke February 29, 2012 06:58

If the change in Cd over iterations is higher than the expected variation then try reducing your relaxation factors
Regards
Luke

Zigainer February 29, 2012 08:48

Quote:

Originally Posted by mahi007 (Post 346621)
When can I stop?

How big are your changes? Is it changing in one direction or is it oscillating?

Far February 29, 2012 09:55

You should observe the value up to fourth decimal place. I usually keep the convergence criteria up to 10e-12 for laminar flows, like you have, so that drag easily converge to required value.

cfd seeker February 29, 2012 10:51

CL,CD and CM are global parameters. You should iterate until the drag converges up till third decimal place at least. If there are large oscillations in CD monitor then check your boundary conditions and the quality of your mesh.

Regards

mahi007 March 1, 2012 01:16

Quote:

Originally Posted by Zigainer (Post 346895)
How big are your changes? Is it changing in one direction or is it oscillating?

It is not oscillating but decreasing. As I have told, every value is repeating for 7-8 times and then going for next lower value. I mean say from 1.7128 repeating for 8 times and then going to 1.7127 and then repeating again for 8 times and so on.

mahi007 March 1, 2012 01:19

Quote:

Originally Posted by delaneyluke (Post 346875)
If the change in Cd over iterations is higher than the expected variation then try reducing your relaxation factors
Regards
Luke

I do not know how to play with relaxation factors. Most of the people recommended not to change factors as they might destabilize the solution. Can you suggest me any changes?

mahi007 March 1, 2012 01:22

Quote:

Originally Posted by Far (Post 346911)
You should observe the value up to fourth decimal place. I usually keep the convergence criteria up to 10e-12 for laminar flows, like you have, so that drag easily converge to required value.

I am using 1e-06 as my convergence criteria as you suggested. Should I go to as high as 1e-12 for laminar?

Can I assume a converged solution if cd doesnot change over 10-12 iterations?

Far March 1, 2012 01:58

Quote:

I am using 1e-06 as my convergence criteria as you suggested. Should I go to as high as 1e-12 for laminar?
1e-06 is good for turbulent flows, but for laminar flows you should go for tighter convergence e.g 1e-12

Quote:

Can I assume a converged solution if cd doesnot change over 10-12 iterations?
monitor for at-least 100-150 iterations

mahi007 March 1, 2012 06:46

Quote:

Originally Posted by Far (Post 347014)
1e-06 is good for turbulent flows, but for laminar flows you should go for tighter convergence e.g 1e-12


monitor for at-least 100-150 iterations

1e-12 is far away from reach. My residual value is starting from 1. It is taking almost 8000 iterations to reach 1e-06. As mine is academical project which I have to complete in 80 days with limited resources, Can I compromise?

Also i have refined mesh 5 folds and it came upto 60 MB of mesh file. Is it normal? And it is taking lot of time fr each iteration. My computer does not have great processing speed.


All times are GMT -4. The time now is 15:15.