CFD Online Logo CFD Online URL
Home > Forums > Software User Forums > ANSYS > FLUENT

Convergence Criterion

Register Blogs Members List Search Today's Posts Mark Forums Read

LinkBack Thread Tools Display Modes
Old   August 26, 2005, 13:29
Default Convergence Criterion
Posts: n/a
Hello I wonder if the convergence criterion ( in Residual Monitors console) affects the results or it just monitors whether the calculation is converged.

  Reply With Quote

Old   August 26, 2005, 17:22
Default Re: Convergence Criterion
Posts: n/a
It just monitors whether you solution is converged with the convergence criterion you set
  Reply With Quote

Old   August 26, 2005, 20:15
Default Re: Convergence Criterion
Posts: n/a
As you know in an ideally converged solution it will not matter. But the way you monitor your solution defintely affects the results. This sounds bit out from the first thing i wrote (a converged solution shall be unique). But what i want to say is the way you decide on convergence may affect your final results. Why so. The main thing is by default in fluent we see scaled residuals. that is

res = sum all cells (Src - ap * phi_p - sum(ai_neigh * phi_neigh) ) / sum all cells ( ap * phi_p )

instead of absolute residual as

res = sum all cells (Src - ap * phi_p - sum(ai_neigh * phi_neigh) )

i have observed that when the divergence occurs and absolute residuals rises very much , you may not see so much rise in scaled residual.

in otherword when scaled residual has moved from 0.1 to 0.2 absolute residual moght have moved from 0.01 to 100.

So when we base our residual monitoring based on scaled residuals , the solution might be changing eventhough the scaled residuals have fallen nicely. Now a decision based on it, could be wrong and you end up seeing the results you were not aspecting.

Some people say that monitor the solution at some particular point in solution domain, but that still is not completely error free. The reason is, in complecated cases where its difficult to get converged solution, in those cases lot of solution domain quickly settles to some near fixed values. but some part of the solution keeps on changing and that slowly settles. so if you chose the point in a place where solution becomes stable soon, and you say i got converged solution, then again you are wrong.

To check this, what you can do is take fairly complecated mesh and and plot the vector plot at each iterations, then in some cases you will be able to see that the velocity vector just rotates around before it settles. where as other parts have settles easily (in otherwords a difficult to converge case).

in the end, its mostly a personal dicision when you want to stop the solution from iterating.

  Reply With Quote


Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On

Similar Threads
Thread Thread Starter Forum Replies Last Post
Convergence Centurion2011 FLUENT 44 June 20, 2017 09:57
problem with Min/max rho tH3f0rC3 OpenFOAM 7 February 23, 2013 06:37
Force can not converge colopolo CFX 13 October 4, 2011 22:03
convergence criterion Dominique FLUENT 5 November 24, 2006 03:36
Setting convergence criterion Vidya FLUENT 5 August 6, 2006 16:46

All times are GMT -4. The time now is 05:46.