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

channelFoam streamwise cyclic/processor boundary

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

Reply
 
LinkBack Thread Tools Display Modes
Old   June 15, 2010, 04:24
Default channelFoam streamwise cyclic/processor boundary
  #1
New Member
 
Per Nilsson
Join Date: Mar 2009
Location: Lund, Sweden
Posts: 13
Rep Power: 7
borrbyper is on a distinguished road
Hi!

Some boundaries sometimes do more than transfer data, see attached figure of Q=10.
(Q is computed only from the reconstructed U at this instant.
White lines show processor boundaries.)

Both cyclic and processor boundaries are visible, but only in the streamwise direction.
(I see the same thing at the in/out boundaries also without lengthwise decomposition.)
Therefore I suspect the problem to be related to the implementation of the driving pressure gradient in channelFoam.

OpenFOAM-1.6.x, Ubar~20m/s, nu~1e-5, H=0.02m, ~3.5M vols on 8 cores.

Regards,
Per
Attached Images
File Type: jpg Qcrit10.jpg (59.3 KB, 86 views)
borrbyper is offline   Reply With Quote

Old   June 17, 2010, 07:15
Default
  #2
Super Moderator
 
Mattijs Janssens
Join Date: Mar 2009
Posts: 1,416
Rep Power: 15
mattijs is on a distinguished road
I did a quick test but cannot repeat it.

- this is not the 0 startup solution?
- what exactly are your steps? (run Q in parallel then reconstruct or reconstruct then run Q?, paraview to do isosurfaces?)

Regards,

Mattijs
mattijs is offline   Reply With Quote

Old   June 17, 2010, 07:43
Default
  #3
New Member
 
Per Nilsson
Join Date: Mar 2009
Location: Lund, Sweden
Posts: 13
Rep Power: 7
borrbyper is on a distinguished road
To start new-case:
1. blockMesh channel395-copy with the same number of nodes as new-case.
2. mapFields from channel395/0 to channel395-copy/0
3. copy channel395-copy/0 to new-case

I would have guessed that much of the initial solution should be blown away
by the high Ubar I set compared to channel395, but it appears much better than 0 initial.

The case has run with dt=1e-5s to about 0.1s.
pMean has almost converged.

To generate the plot:
1. reconstruct last time
2. compute Q for reconstructed time (using util I got from Lillberg/Wikström)
3. foamToVTK for reconstructed time
4. create iso-surfaces with paraview

I can compress the new case and upload to e.g. your ftp if you like.


Regards,
Per
borrbyper is offline   Reply With Quote

Old   June 18, 2010, 01:39
Default
  #4
Super Moderator
 
Mattijs Janssens
Join Date: Mar 2009
Posts: 1,416
Rep Power: 15
mattijs is on a distinguished road
Hi Per,

Does the same happen with the 16x Q utility?
mattijs is offline   Reply With Quote

Old   June 18, 2010, 03:01
Default
  #5
New Member
 
Per Nilsson
Join Date: Mar 2009
Location: Lund, Sweden
Posts: 13
Rep Power: 7
borrbyper is on a distinguished road
I will try on Tuesday (out of office now), but I doubt that it will make a difference for this. I think there is some distortion in the U field.


Regards,
Per
borrbyper is offline   Reply With Quote

Old   June 18, 2010, 04:20
Default
  #6
Super Moderator
 
Mattijs Janssens
Join Date: Mar 2009
Posts: 1,416
Rep Power: 15
mattijs is on a distinguished road
If not, do you have a cut-down version of your case that you can send me or I can download from you?
mattijs is offline   Reply With Quote

Old   June 18, 2010, 09:12
Default
  #7
New Member
 
Per Nilsson
Join Date: Mar 2009
Location: Lund, Sweden
Posts: 13
Rep Power: 7
borrbyper is on a distinguished road
Here is a package without the heavy timestep.
A timestep with p, nuSgs, k and especially U is 47MB compressed.
Where can I put it?


Regards,
Per
Attached Files
File Type: zip channelNoTimestep.zip (6.5 KB, 25 views)
borrbyper is offline   Reply With Quote

Old   June 18, 2010, 12:13
Default
  #8
Super Moderator
 
Mattijs Janssens
Join Date: Mar 2009
Posts: 1,416
Rep Power: 15
mattijs is on a distinguished road
No ftp site. Do you have a somewhere I can download it from. Alternatively can you use e.g. transferbigfiles?
mattijs is offline   Reply With Quote

Reply

Tags
channelfoam, cyclic, processor boundary

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
2nd Order Discretization and Streamwise Periodic Will FLUENT 0 August 29, 2008 13:01
Report Streamwise Pressure Gradient? Will FLUENT 0 July 12, 2008 23:01
Streamwise Second Derivatives S. CFX 3 October 4, 2007 03:16
streamwise coefficient multiplier Rosalba Cobos De los Santos CFX 3 July 13, 2006 11:08
Streamwise periodic heat transfer Novice CFX 0 March 4, 2006 09:14


All times are GMT -4. The time now is 00:07.