CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > FLUENT

Convergence Question

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

Reply
 
LinkBack Thread Tools Display Modes
Old   May 7, 2003, 18:59
Default Convergence Question
  #1
Colin
Guest
 
Posts: n/a
Hi, there. If the residual of continuity doesn't fall below 0.001, what should I do to get convergence? Adjust relaxation factor? Or adjust the mesh? Anything else? What the "thumb of rules" of these adjustments?
  Reply With Quote

Old   May 7, 2003, 20:51
Default Re: Convergence Question
  #2
ap
Guest
 
Posts: n/a
Hi Colin, I don't know the problem you're solving, so I'll try to give some general advice.

- First of all you should check if your problem is well-posed. Verify all boundary conditions.

- Check your grid and refine it, if necessary.

- If you're doing unsteady calculations, try to reduce time step and/or increase the number of iteration per time step.

- Try to reduce under-relaxation factors.

- Try different discretization schemes.

- Change solver settings in the Solve->Multigrid dialog box (read the manual for more info).

Hi

ap
  Reply With Quote

Old   May 7, 2003, 21:13
Default Re: Convergence Question
  #3
raintung
Guest
 
Posts: n/a
I met the same problem, convergence approach steadily at first, but then the residuals will level off(explict solver). I don't know what should I do. I set the some of the parameters under FAS Multigrid Controls , but divergence appears.
  Reply With Quote

Old   May 7, 2003, 21:47
Default Re: Convergence Question
  #4
Colin
Guest
 
Posts: n/a
I think AP has given us some good ideas. Just try, maybe it works.

Thanks Ap.
  Reply With Quote

Old   May 8, 2003, 09:34
Default Re: Convergence Question
  #5
ap
Guest
 
Posts: n/a
If you give some datail of the problem you're solving, maybe I can be more specific.

Hi
  Reply With Quote

Old   May 8, 2003, 21:13
Default Re: Convergence Question
  #6
raintung
Guest
 
Posts: n/a
Can you give me some advice? I am simulating the 3Dflow field in compressor. The mesh is semi-structured and the boundary conditions obtain from the experiment. Because there is a thin fence on the endwall, it is very difficult to get desirable mesh and convergence is not easy. I use k-eplison Realizable model. What should I do? Thanks.
  Reply With Quote

Old   May 9, 2003, 08:59
Default Re: Convergence Question
  #7
ap
Guest
 
Posts: n/a
It seems a complex situation...

I assume your boundary conditions are good, because you have them from experiments.

So try to check mesh quality and, if possible, refine it where quality is really low. You can try to remesh your domain using unstructured grids, if it's too difficult to refine the semi-structured one.

If this doesn't work, try to reduce under-relaxation factors and to increase the number of iteration on the internal cycle (Multigrid panel).

However, changing parameters of the solver may help, but doesn't solve problems due to a poor quality mesh

Good work

ap
  Reply With Quote

Old   May 9, 2003, 21:13
Default Re: Convergence Question
  #8
raintung
Guest
 
Posts: n/a
Thank you. I will try to do that.
  Reply With Quote

Old   May 9, 2003, 21:47
Default Re: Convergence Question
  #9
raintung
Guest
 
Posts: n/a
Can I ask you a simple question? We often say ˇ°refining meshˇ±, but how many means can we do that effectively? Thank you.
  Reply With Quote

Old   May 10, 2003, 18:21
Default Re: Convergence Question
  #10
ap
Guest
 
Posts: n/a
I'm not sure I understood your question.

You want to know how you can refine a mesh?

Personally, I don't like grid adaption tools which are integrated in FLUENT. Usually they increase too much the number of cells.

I prefer to use Gambit and to remesh the system by hand, choosing the number of nodes on the edges of my domain where I want to improve mesh quality. It requires more time, but I think it can give better results without increasing the number of cells too much.

Hi
  Reply With Quote

Old   May 10, 2003, 21:02
Default Re: Convergence Question
  #11
raintung
Guest
 
Posts: n/a
Thank you! I think it is a good idea. I quite agree with you.
  Reply With Quote

Old   May 16, 2003, 04:41
Default Re: Convergence Question
  #12
phil
Guest
 
Posts: n/a
when you remesh in gambit, is it possible to then effectively patch the data from the previous analysis even though the mesh has changed?
  Reply With Quote

Old   May 16, 2003, 08:35
Default Re: Convergence Question
  #13
ap
Guest
 
Posts: n/a
If the geometry doesn't change, you could use the Grid-to-grid solution interpolation function.

See the post: GRID TO GRID INTERPOLATION in FLENT for a link to the page of the manual.

Hi

ap
  Reply With Quote

Old   May 16, 2003, 11:41
Default Re: Convergence Question
  #14
phil
Guest
 
Posts: n/a
cheers ap i was fairly sure it could be done but didnt know what it was called!
  Reply With Quote

Reply

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
Time step dependence of convergence behavior of steady state simulations in CFX Chander Main CFD Forum 5 December 23, 2013 06:31
Convergence Question racerdude777 Main CFD Forum 2 March 14, 2011 23:07
Convergence Question James FLUENT 2 October 29, 2005 14:59
Question about convergence criteria? Thomas FLUENT 7 April 8, 2005 19:38
help! cyclone convergence question Lcw FLUENT 1 January 26, 2005 16:29


All times are GMT -4. The time now is 10:20.