[joakim]

hi

I am trying to re-run a case that I ran in OpenFoam 1.3. When using the setField function, I get the following output on the screen:


Create mesh for time = 0

#0 Foam::error::printStack(Foam:stream&)
#1 Foam::sigFpe::sigFpeHandler(int)
#2 ??
#3 Foam::wedgePolyPatch::initTransforms()
#4 Foam::wedgePolyPatch::wedgePolyPatch(Foam::word const&, Foam::dictionary const&, int, Foam::polyBoundaryMesh const&)
#5 Foam::polyPatch::adddictionaryConstructorToTable<f oam::wedgepolypatch>::New(Foam ::word const&, Foam::dictionary const&, int, Foam::polyBoundaryMesh const&)
#6 Foam::polyPatch::New(Foam::word const&, Foam::dictionary const&, int, Foam::polyBoundaryMesh const&)
#7 Foam::polyBoundaryMesh::polyBoundaryMesh(Foam::IOo bject const&, Foam::polyMesh const&)
#8 Foam::polyMesh::polyMesh(Foam::IOobject const&)
#9 Foam::fvMesh::fvMesh(Foam::IOobject const&)
#10 main
#11 __libc_start_main
#12 __gxx_personality_v0 at ../sysdeps/x86_64/elf/start.S:116
Floating point exception


The domain is an axi-symmetric case but the collapsed surface is replaced by a finite-area surface. What have changed in the setField-function?

regards

/Joakim

[henry]

Is the case small enough to post here for us to test or if not can you make a small case which reproduces the problem?

Henry

[joakim]

Dear Henry

I have constructed two tar-files for the v1.3 and v1.4.

Regards

/Joakim

[joakim]

Sorry!

I can't upload the files, they are too big.

To be a little more desciptive instead:
I generated a mesh in i Icem-tetra, which was saved in star-format and converted via starToFoam. The domain in L-shaped with an inlet at the top and wedge-conditions on the sides. Remaining b.c. are pressure-outlet and wall.

I did some more testing I have found out that the problem is not with setField. I ran checkMesh on the mesh and this one failed for the v1.4 too with an output that is vey similar to the one I got whenI ran setField:


Create polyMesh for time = constant

#0 Foam::error::printStack(Foam:stream&)
#1 Foam::sigFpe::sigFpeHandler(int)
#2 ??
#3 Foam::wedgePolyPatch::initTransforms()
#4 Foam::wedgePolyPatch::wedgePolyPatch(Foam::word const&, Foam::dictionary const&, int, Foam::polyBoundaryMesh const&)
#5 Foam::polyPatch::adddictionaryConstructorToTable<f oam::wedgepolypatch>::New(Foam ::word const&, Foam::dictionary const&, int, Foam::polyBoundaryMesh const&)
#6 Foam::polyPatch::New(Foam::word const&, Foam::dictionary const&, int, Foam::polyBoundaryMesh const&)
#7 Foam::polyBoundaryMesh::polyBoundaryMesh(Foam::IOo bject const&, Foam::polyMesh const&)
#8 Foam::polyMesh::polyMesh(Foam::IOobject const&)
#9 main
#10 __libc_start_main
#11 __gxx_personality_v0 at ../sysdeps/x86_64/elf/start.S:116
Floating point exception

As before, checkMesh for v1.3 did just fine.

Regards

/Joakim

[mattijs]

Hi Joakim,

Is your case actually 2D? Wedges will only work on 2D cases (i.e. 1 cell thick, created/createable by rotational extrusion). So it will not work with e.g. tetrahedra. The checking in 1.4 is a bit more strict than in 1.3.

[joakim]

Hi Mattijs

Yes it is a 2D case. One layer of tetras extruded to a prisms layer.

The reason I worked with the case is that I never got O.F. v1.3 to accept the genuine 2D-axisymmetric meshes created in icem-tetra.
We had a discussion about this onces under the thred "Axisymmetric bodies, wedge-type B.C.'s". You came up with a suggestion to use the collapseEdges utility, but since my mesh doesn't contain quads I guess this approach doesn't help.

As I stated back than, I constructed two meshes. One structured mesh using blockMesh and an unstructured mesh in ICEM-tetra. I noted a difference when running checkMesh on the two cases. For the structured mesh, in the output, it writes out empty when it checks the collapsed surface, whereas in the case of the unstructured mesh it complains that the surface area is 0. I assume the same goes for the solver. The code ignores the empty boundary in the blockMeshed-case where the empty b.c. is accepted, whereas the OpenFoam still thinks by 2D-axisymmetric mesh is a 3D mesh with an empty b.c. which is not tolerated and the solver fails to start.

I have not tested the unstructured mesh in v1.4 yet but I will do that as soon as I can. The structured mesh I mention above worked fine.

Regards

/Joakim

[mattijs]

Can you create a small unstructured testcase which shows the problem and you can send over? The collapseEdges should collapse the zero-area axis faces.

[joakim]

Dear Mattijs

Sorry for the delayed repsons. I have created two very small test-problems. The first case is a wedge-shaped domain with a collapsed interface, see image



The second test case as a geometry with the same shape but the collased interface is replaced with a surface with a small area.



The first problems works in neither O.F 1.3. or O.F 1.4 whereas the second problem works in O.F 1.3.

This is what I did in O.F.1.4:

The meshes where created in icem-cfd v.11. Saved into star-format and translated via starToFoam to O.F-format. The starToFoam application seems to run fine, but when doing checkMesh I get the following outputs

case1: log_before

case2: log_before_2

Note how it complains about the elements with 0-area surface. If we just go a head and run FoamX to subscibe bounday conditions, where the sides are given wedge-bounday conditions and running checkMesh afterword both cases fails. This did NOT happend using O.F. 1.3

case1: log_after

case2: log_after_2

I conclude that here are to problems
1) The seems to be something wrong with the wedge-condition. This part worked in O.F. 1.3.
2) As I stated ealier when I ran a similar geometry with a mesh constructed in blockMesh, checkMesh noted that the collapsed bounday had an empty b.c. and ignored it, so did the solver.
For the icem-tet mesh, checkMesh do not ignore the boundary, neither the solver whish I assume it should.

I hope a have been clear enough about the problems

Best regards

/Joakim

[mattijs]

Can you post the meshes themselves so we can have a look? If they're too big to post (> 50k or so) just send them directly to me.

[joakim]

Hi Mattijs

Sorry, I forgot to post the cases:

case1: VF_1_14.tar.gz

case2: VF_2_14.tar.gz

Regards

/Joakim

[mattijs]

- Switch off FOAM_SIGFPE and 1.4 runs through for me.

- Your wedges do not straddle a coordinate plane. Instead one of them is in the xy plane. See section 6.2.2 of User Guide about wedges.

[joakim]

Hi Mattijs

Tnx for your replay, so I tried what u suggested:

case2:
1) So it seems that by symmetrizing the wedge-mesh around the xy-plane makes things work.

VF_1_14.tar.gz

This was not neccesary in O.F. 1.3. Note that this was regardless the of value of FOAM_SIGFPE. By the way, when you say "turn off", do you mean

export FOAM_SIGFPE=false

in the bashrc file under .OpenFOAM-1.4 ?

2) I did the same symmetrization with case1. Here I still get the problem with the empty b.c.

log_out

and the solver still complain about the 0 face area when I run checkMesh.

Did I do something wrong with FOAM_SIGFPE?
or is this due to something else.

regards

/Joakim

[mattijs]

1) You'll have to

unset FOAM_SIGFPE

Setting it to anything (even 'false') switches on the trapping.

2) empty patches and fields are to be used for front and back of 2D cases. Yours are on the wedge axis if I remember correctly.

[rolando]

Hi,
I got similar problems as Joakim. My case interrupts with simialar messages as above.
I don´t have wedges in my case but empty patches.
If I unset FOAM_SIGFPE, as Mattijs proposed above, it works.
What does unsetting this variable cause?
Why didn´t I have this problem with OpenFOAM-1.3?

Can anybody tell me something about this?

Rolando

[lakeat]

Yes, when I used liftDrag, I got the same problems.
again, What does unsetting this variable cause?

Thanks

Daniel

[openfoam_user]

How can I unset FOAM_SIGFPE ?

Which file do I have to modify ?

I have to modify the file .cshrc located in OpenFOAM/OpenFOAM-1.5/etc/ ?

Thanks,

Stephane

[lostin4ever]

I am trying to re run a tanksloshing case using latestTime utility for further time. It is showing error (pasted below). Can anybody tell how to get rid of this error.


Create time

Create mesh for time = 2.4

Selecting dynamicFvMesh solidBodyMotionFvMesh
Selecting solid-body motion function SKA

Reading g
Reading field p

Reading field alpha1

Reading field U

Reading/calculating face flux field phi

Reading transportProperties

Selecting incompressible transport model Newtonian
Selecting incompressible transport model Newtonian
Selecting turbulence model type laminar


Unable to set reference cell for field p
Reference point pRefPoint (0 0 0.15) found on 0 domains (should be one)


file: /home/akash/OpenFOAM/akash-1.6/run/multiphase/interDyMFoam/ras/sloshingTank_practice1/system/fvSolution::PISO from line 94 to line 103.

From function void Foam::setRefCell
(
const volScalarField&,
const dictionary&,
label& scalar&,
bool
)
in file cfdTools/general/findRefCell/findRefCell.C at line 93.

FOAM exiting

[mattijs]

Your mesh is moving. You'll have to make sure that the reference point is inside the mesh (upon restart).