|
[Sponsors] |
Differences between new and old versions of 1.5-dev |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
August 29, 2010, 17:37 |
Differences between new and old versions of 1.5-dev
|
#1 |
Senior Member
Ben K
Join Date: Feb 2010
Location: Ottawa, Canada
Posts: 140
Rep Power: 19 |
I work with OpenFoam 1.5-dev at work and at home. At work we've had 1.5-dev installed for a long time (I'm not sure exactly when it was last updated but I'm guessing over a year ago).
Today, using my home computer, I installed 1.5-dev by svn co https://openfoam-extend.svn.sourcefo...enFOAM-1.5-dev The weird thing is that my model developed at work, which works with the 1.5-dev at work, doesn't work on my new installation of 1.5-dev at home. Specifically, I'm using the coupledFvScalarMatrix functonality of 1.5-dev (also used in conjugateHeatFoam). Basically when I run my model at home, it works for the first time step and then starts to oscillate and never converges whereas at work, there are no problems. I'm not sure exactly what the problem is...the tutorial problems run on my new 1.5-dev installation. Here's an example of a model which runs at work using an old version of 1.5-dev but doesn't run using 1.5-dev that I installed today: http://dl.dropbox.com/u/12812/OF/test.tgz If anybody has time to run this on your 1.5-dev, just type: tar -zxvf test.tgz cd test/Aug29/testFoam wclean;wmake cd ../case testFoam I'm not sure what the problem is but like I said, this works perfectly on an old version of 1.5-dev but not on my new installation. |
|
August 29, 2010, 19:53 |
|
#2 |
Senior Member
Sandeep Menon
Join Date: Mar 2009
Location: Amherst, MA
Posts: 403
Rep Power: 25 |
Do an 'svn info' on both installations, and track the file changes between the two versions. That should narrow down your search to what went wrong.
|
|
August 29, 2010, 20:11 |
|
#3 |
Senior Member
Ben K
Join Date: Feb 2010
Location: Ottawa, Canada
Posts: 140
Rep Power: 19 |
This is from the old installation at work:
Path: . URL: https://openfoam-extend.svn.sourcefo...enFOAM-1.5-dev Repository Root: https://openfoam-extend.svn.sourcefo...penfoam-extend Repository UUID: e4e07f05-0c2f-0410-a05a-b8ba57e0c909 Revision: 1397 Node Kind: directory Schedule: normal Last Changed Author: hjasak Last Changed Rev: 1397 Last Changed Date: 2009-08-14 05:22:30 -0400 (Fri, 14 Aug 2009) and this is on my home comp: Path: . URL: https://openfoam-extend.svn.sourcefo...enFOAM-1.5-dev Repository Root: https://openfoam-extend.svn.sourcefo...penfoam-extend Repository UUID: e4e07f05-0c2f-0410-a05a-b8ba57e0c909 Revision: 1822 Node Kind: directory Schedule: normal Last Changed Author: deepsterblue Last Changed Rev: 1822 Last Changed Date: 2010-08-26 15:57:35 -0400 (Thu, 26 Aug 2010) I'm guessing a lot of things could have changed between those two revisions. |
|
August 29, 2010, 20:29 |
|
#4 |
Senior Member
Sandeep Menon
Join Date: Mar 2009
Location: Amherst, MA
Posts: 403
Rep Power: 25 |
Yup. What you could do is pick a revision somewhere in between and, in some weird sense, perform your own root-finding algorithm to pin-point a certain revision
|
|
August 29, 2010, 22:04 |
|
#5 |
Senior Member
Ben K
Join Date: Feb 2010
Location: Ottawa, Canada
Posts: 140
Rep Power: 19 |
It appears as if there is a problem with the CrankNicholson scheme in the new 1.5-dev revisions. Using the Euler scheme works though.
|
|
August 29, 2010, 22:08 |
|
#6 |
Senior Member
Sandeep Menon
Join Date: Mar 2009
Location: Amherst, MA
Posts: 403
Rep Power: 25 |
Perhaps you could file a bug-report, with a small suitable test-case that can reproduce the problem. I believe the Mantis system is appropriate for this - but I'm a newbie in that regard. Perhaps Bernhard / Martin can be of some assistance here.
|
|
August 30, 2010, 09:03 |
|
#7 |
Senior Member
Ben K
Join Date: Feb 2010
Location: Ottawa, Canada
Posts: 140
Rep Power: 19 |
I'll have to look a bit more into this. In the conjugateHeatFoam tutorial that comes with the new revision of 1.5-dev, I was able to use the CrankNicholson time scheme but with my model posted above it only solves with the Euler time scheme.
|
|
|
|