exporting yarns and matrix mesh to abaqus

[yanxueotft]

I'm a master student studying triaxially braided composites recently. TexGen is really a great CAD software modeling textile composites and I have built a model using script. But now I get a trouble when I want to export tetrahedral elements. I have tried to export iges file but failed. I also have tried to export inp file directly, but i can only export abaqus voxel file and dry fibre file successfully. I need to mesh both yarns and matrix using tetrahedral elements. This have troubled me for several days, I really hope you can help me, thank you .

[louisepb]

Hi,

Glad you're finding TexGen useful.
To get tetrahedral elements you need to use the File->Export->Volume Mesh option. Then choose to export as a .inp in the file select dialog.

Louise

[yanxueotft]

Hi,

Glad you're finding TexGen useful.
To get tetrahedral elements you need to use the File->Export->Volume Mesh option. Then choose to export as a .inp in the file select dialog.

Louise

thank you for replying so quickly. I have tried exporting volume mesh for triaxial braid architecture, but it didn't work, the texgen had no response and then error. But if i mesh an simple textile, volume mesh works. I don't know why...I'm totally confused...by the way, can the so called voxel elements take the place of normal elements? I have never heard of it...
Thank you very much!
Yanxue

[Flo_TUM]

To me that sounds as if your model contains unsolved interpenetrations, which is not easy to get rid of when modeling triaxial braids.
I don't know how the meshing-algorithm in texgen works but I don't think it's possible to mesh interpenetrating volumes with thedradrals so you should check if that's the problem.

A voxel-mesh consits of regular hexaedral elements.
It's an easy way to get a highquality mesh but with the disadvantage of a less exact representation of the geometry depending on the elementsize.
I've attached a screenshot of how the yarns could look in a voxel-meshed cell.

greets,
Flo

[yanxueotft]

To me that sounds as if your model contains unsolved interpenetrations, which is not easy to get rid of when modeling triaxial braids.
I don't know how the meshing-algorithm in texgen works but I don't think it's possible to mesh interpenetrating volumes with thedradrals so you should check if that's the problem.

A voxel-mesh consits of regular hexaedral elements.
It's an easy way to get a highquality mesh but with the disadvantage of a less exact representation of the geometry depending on the elementsize.
I've attached a screenshot of how the yarns could look in a voxel-meshed cell.

greets,
Flo

Thanks for your advice. I checked my model, it did contain several interpenetrations and I modified it. But the problem still exists...no response...!!!
And for voxel-mesh, I just wonder whether it can provide accurate simulation result?
thanks again~!
Yanxue

[louisepb]

Hi Yanxue,

The volume mesh export can cope with interpenetrations so this shouldn't be your problem. If you send me the file you're working on and the parameters you use in the export then I could run in debug and try and see where it's failing.

With the voxel mesh you need to be careful that you use a number of voxels which captures all the details in your textile. For instance if you have any fine yarns then you need to make sure that the mesh is fine enough for these to be continuous. These have been used in our group to produce some good results for complex 3D weaves which would have been problematic to mesh any other way at the moment.

Louise

[yanxueotft]

Hi Yanxue,

The volume mesh export can cope with interpenetrations so this shouldn't be your problem. If you send me the file you're working on and the parameters you use in the export then I could run in debug and try and see where it's failing.

With the voxel mesh you need to be careful that you use a number of voxels which captures all the details in your textile. For instance if you have any fine yarns then you need to make sure that the mesh is fine enough for these to be continuous. These have been used in our group to produce some good results for complex 3D weaves which would have been problematic to mesh any other way at the moment.

Louise

Hi Louise,
Thanks for your replying. You are right, when I change the domain size, I can export volume mesh now, amazing...And my test model is the same as http://texgen.sourceforge.net/phpBB3/vi ... p?f=1&t=64. I just change its domain to an RUC. Now my new problem is that the meshing is strange. In the most of the area of the yarns and matrix, the tetrahedral elements is good, but in some area, the element size is too small. This appearance will be mended a little if I choose period meshing.


And by the way, if I want to add period boundary condition, can I only choose voxel mesh? Volume mesh seems can't generate same mesh in master surface and slave surface...
Regards,
Yanxue

[louisepb]

Hi Yanxue,

The areas with small mesh elements are because these are small areas between the yarns. The top surface, containing these small areas, is triangulated and then these triangles are projected down through the volume leading to the areas you highlighted.

The volume mesh option doesn't produce matching meshes on opposite faces. The voxel mesh and dry fibre mesh will provide matching node sets. I have been experimenting with the use of Tetgen which is an open source meshing program. This gives the ability to produce matching meshes on opposite faces but doesn't always manage to produce a mesh. The code to produce Tetgen meshes is included in the source code in the SVN repository so that you could build TexGen and experiment with this if you wanted to. I haven't included it in the release build because of its unpredictability.

Louise

[yanxueotft]

Hi Yanxue,

The areas with small mesh elements are because these are small areas between the yarns. The top surface, containing these small areas, is triangulated and then these triangles are projected down through the volume leading to the areas you highlighted.

The volume mesh option doesn't produce matching meshes on opposite faces. The voxel mesh and dry fibre mesh will provide matching node sets. I have been experimenting with the use of Tetgen which is an open source meshing program. This gives the ability to produce matching meshes on opposite faces but doesn't always manage to produce a mesh. The code to produce Tetgen meshes is included in the source code in the SVN repository so that you could build TexGen and experiment with this if you wanted to. I haven't included it in the release build because of its unpredictability.

Louise

Hi Louise,
Thank you very much!
I think I can do the next step for my study, in which I'll try to defining local material directions for yarns in my model. It seems Texgen can define local material direction because I find this code in *inp file exported from texgen:

Code: Select all

********************
*** ORIENTATIONS ***
********************
** Orientation vectors
** 1st vector represents the fibre direction
** 2nd vector is an arbitrary vector perpendicular to the first
*Distribution Table, Name=TexGenOrientationVectors
COORD3D,COORD3D
*Distribution, Location=Element, Table=TexGenOrientationVectors, Name=TexGenOrientationVectors, Input=voex.ori
*Orientation, Name=TexGenOrientations, Definition=coordinates
TexGenOrientationVectors
1, 0

Code: Select all

*Solid Section, ElSet=Yarn0, Material=Mat1, Orientation=TexGenOrientations
1.0,
*Solid Section, ElSet=Yarn1, Material=Mat1, Orientation=TexGenOrientations
1.0,
*Solid Section, ElSet=Yarn2, Material=Mat1, Orientation=TexGenOrientations
1.0,

I checked the *ori file and found that this file defined each element's direction. Does this define material direction follow the yarn curve? If it does, how can I demonstrate it in abaqus? Or I should define material orientation in abaqus with User Subroutine Defined Orientation?
Regards,
Yanxue

[louisepb]

Hi Yanxue,

The .ori file gives the fibre orientations (the yarn tangent and normal) for the centre point of each element.

I'm afraid that I'm not an Abaqus expert so you'll have to refer to the Abaqus User Manual to see how to use the orientation information.

Best regards,
Louise

[yanxueotft]

Hi Yanxue,

The .ori file gives the fibre orientations (the yarn tangent and normal) for the centre point of each element.

I'm afraid that I'm not an Abaqus expert so you'll have to refer to the Abaqus User Manual to see how to use the orientation information.

Best regards,
Louise

Hi Louise,
I have found how to display material orientation in abaqus, thanks for your help~! I love you and your group!
Rergards,
Yanxue

[George_Bal]

Hello,

I have a triaxial braided model and I 'm trying to export a Volume mesh from it (.inp file) with periodic BCs.
The meshing algorithm keeps failing, even though for some other models I tried, it works.
I also tried to play around with the domain and the merge tolerance (saw that in the forum), but I still got nothing.

Could you please give me a hint on what might be the problem? I am using TexGen 3.7.0

Thank you in advance,
George.

[louisepb]

Hi George,
The added constraints necessary for creating the matched elements on the opposite sides of the domain mean that in certain cases the algorithm fails. You could try changing the seed size which governs the overall mesh density.

I can't remember if there have been any bug fixes to the volume meshing recently but you could also try updating your TexGen version.

If you don't have any I could take a look at it if you sent me details of what you're trying to do.

Best regards,
Louise