RVE with periodic BC

[ezgi]

Hello everyone,

I am trying to simulate tensile test on short fiber composite(fiber modeled as ellipsoid) RVE  with periodic boundary conditions. I am using tetrahedron elements, I have generated my mesh on lsprepost. When I run the program, on my output file I am getting results saying "TETSHP: Determinant =   0.0000000000000000 " and not getting any stress values. I couldn't figure out what might be the reason. I would really appreciate any help.

I am attaching my input file and files for coordinates and elements.

Thanks,
Ezgi

[Prof. R.L. Taylor]

What version of feap are you using; what element types; more general information please

[ezgi]

Thank you Prof. Taylor for your quick reply .

I am using FEAP 8.4.1d.  Element type is 4 node tetrahedron.  First I did the analysis with 8-node brick elements using feap commands BLOCK and COORDINATES with single material, it worked fine. But after changing the mesh by including coordinate and element file, I didn't get any stress results. My RVE consists of single ellipsoid fiber inside a rectangular prism. I thought the reason might be with the different nodal layout on the opposing faces, but I checked it and they are the same.

[Prof. R.L. Taylor]

The program recognizes the tet.  You will have to debug the error.

Post the input file to make sure it works for us.

If one of the elements is incorrectly numbered then perhaps the routine will throw the error.  If every element then something is wrong in the interface to the shape function

The shape function is called from interp3d.f (./elements/solid3d directory).  You need to check that the passing of the "xl" values is good for all elements -- add a print to the output file is the easiest way to check. 

[ezgi]

Thank you again Prof. Taylor for your quick reply.

I am attaching the input file along with the mesh files.

Ezgi

[Prof. R.L. Taylor]

I checked hour mesh and it seems the tets do not have volume -- I did not check them all but checked the last one and the z coordinate for all 4 nodes is the same.  So the problem is the mesh generator somehow.  They are not tets.

[ezgi]

Indeed there is something wrong with my meshes they are without volume, thank you so much for your observations and input Prof. Taylor,

[Prof. R.L. Taylor]

Yes, if you do PLOT PERS and then PLOT MESH all you see is surfaces.  If you had a mesh of filled tets you would need to add PLOT HIDE to make a plot of the surfaces.  So is seems the mesh generator made elements like those for shells or membranes.

[Prof. R.L. Taylor]

One other thing.  If you are using periodic b.c. on the RVE then you should only have boundary conditions for the 8 corner points, not the entire outer surface.