|
[Sponsors] |
December 4, 2013, 21:06 |
Help me understand convergence/residuals
|
#1 |
New Member
Join Date: Nov 2013
Posts: 4
Rep Power: 12 |
Hi
I'm new to CFD and trying to understand convergence criteria and residuals I hear that convergence criteria should be set to ex. 10^-3, 10^-6 and that the solution has converged when it reaches this value. However I find that typically the residuals will start at 0 and then decrease, decreasing more slowly as the number gets smaller, eventually reach a state that is more or less steady. So what is the point of selecting a convergence value at all? It seems that selecting a convergence value lower than this steady state value will cause the residuals to never reach convergence and simply run until the maximum iterations have been performed. Setting a value higher than this steady state value causes it to converge before actual convergence has been reached. So clearly I don't understand something properly, and any help would be appreciated. |
|
December 5, 2013, 11:55 |
|
#2 |
Senior Member
Join Date: Jan 2010
Location: Germany
Posts: 268
Rep Power: 17 |
You have to judge convergence by means of the variations of key variables behaviour in you simulation, the imbalances from the flux report and not only the Residuals. Latter only depict the change of the rate of change of the solution from iteration to iteration
|
|
December 8, 2013, 07:20 |
|
#3 |
Retired Super Moderator
Bruno Santos
Join Date: Mar 2009
Location: Lisbon, Portugal
Posts: 10,975
Blog Entries: 45
Rep Power: 128 |
Greetings to all!
I was browsing by (while hunting down spammers ) and noticed this thread. Have a look at this old thread that should explain a lot about the topic of "convergence": http://www.cfd-online.com/Forums/flu...nvergence.html |
|
|
|