CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Convergence problem...again (http://www.cfd-online.com/Forums/fluent/30539-convergence-problem-again.html)

moiami November 19, 2002 03:02

Convergence problem...again
 
Hi,

I know the question has been asked to death b4 but I have difficulties in tracking the answers that can specifically answer my questions:

1. How to know when my solutions have converged? What I did was to run it using the default convergence criterion in the residual pop-up menu and as long as the curves haven't become horizontal I add the zero's in the criterion. Is that the correct way?

2. When to stop the iteration? I read somewhere that as long as the 5 numbers in the console (print off of the residual values) e.g. 5.2661e-06 are not the same for 50 iterations than I have to continue iterating. The problem is my residual curves have long gone into a horizontal mode but the 5 digits are still experiencing +/-0.2000e-06, small in my eyes...but I don't dare stop the iterations. Was what I read true? What's the range of fluctuation that is safe for me to stop my iteration then?

Thanks

Boris November 19, 2002 04:20

Re: Convergence problem...again
 
You should monitor the parameters of your interest,say the drag or average temperature of a certain surface, etc, to decide whether the solution is really converged or not. It is more realible than monitoring the residuals.

Alamgir November 20, 2002 00:15

Re: Convergence problem...again
 
Moiami,

1 if u look at the residuals and all the residuals values decreased by 3 order then it would be considered as converged results. And when any values remain constant ie horizontal after getting few at least 2 order decreased then it would be considered as converged. if not then never user 0 in the covergence criterion but decrease the under reluxation faction in solve>control>solution menu. and see the effect.

2 if within 50-60 the values of all residuals get 3 order decreased then it is not considered as converged then put another 0 in the converge critera to allow run more iterations.

go ur own. goodluck

Alamgir


All times are GMT -4. The time now is 03:13.