CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   IcoDyMFoam and tetrahedral meshes (http://www.cfd-online.com/Forums/openfoam-solving/59740-icodymfoam-tetrahedral-meshes.html)

ulrich_lange April 13, 2007 02:22

I am an OpenFOAM newbie and cu
 
I am an OpenFOAM newbie and currently taking my first timid steps by expanding the tutorial problems: I try to model a 3D-mixer using icoDyMFoam. It works fine on a hexahedral mesh,but using a tetrahedral mesh for the same mixer geometry results immediately in solution singularities (The first message in the first time step is "ICCG: Solving for pcorr: Solution singularity").
Should I change something in fvSchemes, fvSolution when using tetrahedral meshes? Or can the slider procedure in icoDyMFoam not handle tetrahedral meshes? (I have no problems with my tetrahedral meshes in "icoFoam"-test cases).

I generate the meshes with Gambit and convert them using fluentMeshToFoam.

Thanks in advance for comments/help. And thanks to the OpenFOAM developers for making your impressive work available as open source.

Regards,

Ulrich

olesen April 13, 2007 02:59

I've had problems with sliding
 
I've had problems with sliding interfaces too (for a different topology, but with hexas). See my tinySlider problem that I posted on the forum a while ago.

Rather than waiting for solution singularities to signal a problem, have you tried moveMesh/moveDynamicMesh to test the mesh movement alone?

Beyond any issues of the above issues, I would really try to avoid tet meshes in general, regardless if the solver is open-source or commercial.
If you wish to retain the automatic meshing capabilities of tet meshes, you might try converting it to poly mesh (eg, polyDualMesh) first. The cells in the corners will not be convex, but they run in many cases anyhow.

ulrich_lange April 13, 2007 04:07

Thank for your hints to the mo
 
Thank for your hints to the moveDynamicMesh tool and in particular the polyDualMesh tools, Mark. I will try it.

Meanwhile I seem to have found also another solution for my immediate problem: It suffices to mesh the Slider surfaces with rectangles (so the adjacent cells will be pyramids) before tetmeshing the volumes.
At least for my simple test case, I get the solution on this tetmesh now.

david_borer April 13, 2007 04:33

I,m also new to OpenFOAM and h
 
I,m also new to OpenFOAM and have been playing around with the icoDyMFoam solver and the mixer2D tutorial. I tried a new setup with a simple rotor in a channelflow (in 2D), and somehow the slidingInterface seems to interfere quite a lot with the flow, i get some velocity jumps across the interface. Im wondering if anybody else has experienced similar problems?

hjasak April 13, 2007 05:16

This all works fine for me on
 
This all works fine for me on the patched up version of OpenFOAM-1.3. On 1.4 I reserve any comments - it is likely that my code was not supported or updated.

Hrv

jens_klostermann April 13, 2007 05:55

Hi all, First thank you to
 
Hi all,

First thank you to all OF developer!!

Just a comment:

I WAS eager to test the new OF1.4 version. And I think there is a lot good stuff in it, especially multiphase interfoam, the new GAMG solver (didn't test it yet), etc.
But I stress out WAS, since there is also a whole bunch of good stuff missing which was in the latest Hrv OF1.3 versions e.g. FSI, free surface tracking, etc.
So I ask myself why is this good stuff missing in OF1.4.
Maybe the user is supposed two maintain two versions of OpenFoam one OF-1.4 (from the Weller corner) and one OF-1.3 patched up (from the Jasak corner). Or everybody creates his own version of with the best of the two. But with this life isn't getting easier and there is only one looser: the code itself - OpenFOAM.

Sorry, this is only my opinion and maybe this is not the right place to discuss this, but I hope the discrepancies between the two corners will settle down and in the future there will be one OpenFOAM with ALL the nice features in one code.

Jens

lr103476 April 13, 2007 06:11

I think that the new version,
 
I think that the new version, OpenFoam 1.4, is supposed to be thoroughly tested and mainly bug free, since it's an official release.

The patched-up 1.3 version from prof. Jasak is unofficial and in development. My experience, though, is that the official releases (1.2, 1.3 and 1.4) are pretty OK, but eventually the users (we) need to make some fixes ourselves. Especially the 1.3 version was crappy concerning mesh motion and lift/drag calculation. I will compile / run my user solvers/utilities to the new 1.4 and we will see.

Regards, Frank

david_borer April 13, 2007 08:21

http://www.cfd-online.com/Open
 
http://www.cfd-online.com/OpenFOAM_D...your_image.gif

hjasak April 13, 2007 08:43

True and fair. I do not necess
 
True and fair. I do not necessarily agree that a new number means a better code because of a dysmal experience with 1.3. However, this will not stop me from moving my work into 1.4 stream and releasing all the new stuff as usual.

You will appreciate this will take some time and I've got other stuff to do as well. In any case, my plan is to set up a collaboration with some leading centres of development and use of OpenFOAM and set up a public CVS (-like) access. More on that at the Workshop in Zagreb.

Hrv

david_borer April 13, 2007 08:46

I still have some problems wit
 
I still have some problems with icoDyMFoam. I tried the release 28_03_07 from powerlab.fsb.hr and i still get some strange results at the slidinginterface:

http://www.cfd-online.com/OpenFOAM_D...your_image.gif

the block in the middle is rotating and the flow is going downwards (left and right are walls), one can see some discontinuities at the lower left region. How can I use icoDyMFoam without getting this?

david_borer April 13, 2007 09:04

I still have some problems wit
 
I still have some problems with icoDyMFoam. I tried the release 28_03_07 from powerlab.fsb.hr and i still get some strange results at the slidinginterface:

http://www.cfd-online.com/OpenFOAM_D...ges/1/4205.jpg

the block in the middle is rotating and the flow is going downwards (left and right are walls), one can see some discontinuities right at the interface. How can I use icoDyMFoam without getting this? How do i have to take the rotation into account for setting the timestep?

jdkummer April 16, 2007 16:44

I tried setting up almost the
 
I tried setting up almost the exact same problem, but I have a different outcome. For some reason, the block in the middle does not rotate at the correct rate. I set it to spin at 60 rpm, so after 1 second it should make 1 full rotation; however, it ends up going around about 1 1/4 times. Also, the rotation speed changes slightly when I change the timestep. I originally tried this a few months ago on the standard downloaded version 1.3 from the website, but now just tried it again on the updated 1.3 and got the same result. Unfortunately, I cannot figure out how to run the case in version 1.4. Any suggestions would be greatly appreciated.

Thanks

Joe Kummer


All times are GMT -4. The time now is 22:41.