Currently I am trying to take models out of texgen and use them with ABAQUS. When the files are saved they usually crash or are very large. Has anyone had any luck using the python scripting and importing it into abaqus. I attempted to import a simple pattern but had no luck.
I've seen others have attempted this in the past but I was wondering if any recent advances had been made.
Also does texgen have the ability to show the volume of the fibers?
Thanks
Python/Abaqus
Moderators: Martin, Developers
Hi Frank, I'm glad you found your way to the forum 
.
Yes, exporting to IGES and STEP files can be a bit troublesome with large files sizes unfortunately. One of my colleagues who is working on a separate project created a python script to export geometry to ABAQUS. But he is working on a different project, the sponsors of which haven't quite got their head around the open source idea and won't allow him to share the script with everybody else. This method isn't without problems either though, so don't worry too much about that.
In the last few weeks I've been working on writing code to generate meshes of both yarns and matrix volumes directly in TexGen. It's looking promising, and is very near completion. Hopefully I'll be able to roll that out in the next few weeks. This would remove all the issues with transferring geometry between software packages. Meshes are very easy to transfer because there is only really one way to represent it.
As for showing the volume of the fibres, that's not possible in TexGen. Fibres are not modelled, yet... Maybe sometime in the future they will be but that's not a short-term goal.
.Yes, exporting to IGES and STEP files can be a bit troublesome with large files sizes unfortunately. One of my colleagues who is working on a separate project created a python script to export geometry to ABAQUS. But he is working on a different project, the sponsors of which haven't quite got their head around the open source idea and won't allow him to share the script with everybody else. This method isn't without problems either though, so don't worry too much about that.
In the last few weeks I've been working on writing code to generate meshes of both yarns and matrix volumes directly in TexGen. It's looking promising, and is very near completion. Hopefully I'll be able to roll that out in the next few weeks. This would remove all the issues with transferring geometry between software packages. Meshes are very easy to transfer because there is only really one way to represent it.
As for showing the volume of the fibres, that's not possible in TexGen. Fibres are not modelled, yet... Maybe sometime in the future they will be but that's not a short-term goal.
Do you mean the actual geometry of the fibres or the volume/amount of fibre in the model?Also does texgen have the ability to show the volume of the fibers?
The former is not available, the latter is easily calculated (fibre_area*yarn length) , Martin can probably tell you whether there is a button for it in the GUI, I don't use the GUI at the moment.
W
-
lxtlixuetao
- Posts: 4
- Joined: Tue Nov 20, 2007 9:29 pm
Martin,
I also have this question about outputing meshing file for Abaqus, it is a large file, have no idea to deal with it. I got your first reply for this is about half a year ago, did you figure out the way to get through it? Or is new version updated with improvement?
Another question is that, when you do textile geometry, you do fiber bundles and matrix together or seperately. The examples just show me the way you do undulated fiber bundles. The rest matrix should be added in somehow, and this is my quetion.
Thanks in advance.
Xuetao
I also have this question about outputing meshing file for Abaqus, it is a large file, have no idea to deal with it. I got your first reply for this is about half a year ago, did you figure out the way to get through it? Or is new version updated with improvement?
Another question is that, when you do textile geometry, you do fiber bundles and matrix together or seperately. The examples just show me the way you do undulated fiber bundles. The rest matrix should be added in somehow, and this is my quetion.
Thanks in advance.
Xuetao
Well the TexGen mesh generator is now in the latest version. So you can generate meshes that way rather than trying to export to .step or .iges. Go to File -> Export -> Volume Mesh... and save it as ABAQUS input file. Unless some new development occurs with the .step and .iges files (unlikely) I won't be working on the side of things because it's just not worth it. There are fundamental flaws with using these file formats that can never be fully resolved.lxtlixuetao wrote:Martin,
I also have this question about outputing meshing file for Abaqus, it is a large file, have no idea to deal with it. I got your first reply for this is about half a year ago, did you figure out the way to get through it? Or is new version updated with improvement?
The matrix volume is implicitly defined as being the region of space not occupied by the yarns. So basically, no you don't need to "add them in somehow". But you will want to limit your model to a finite region of space called the "domain". Once you have the yarns and a domain specified then you can mesh it as described above.lxtlixuetao wrote: Another question is that, when you do textile geometry, you do fiber bundles and matrix together or seperately. The examples just show me the way you do undulated fiber bundles. The rest matrix should be added in somehow, and this is my quetion.
-
lxtlixuetao
- Posts: 4
- Joined: Tue Nov 20, 2007 9:29 pm
Thanks for your immediate reply. Is there any chance to control element type? Like, 8 nodes brick element. Seed size should be used to adjust mesh density. How it work?
Actually, I tried to create mesh file follow your instruction. The texgen output command window shows the procedure. At last it will keep saving mesh and element orientation data into *inp and *.ori, and additional *.eld files. Strange things happened that, with same *.py file, sometimes it gives me a large file, while others it just gives 1 or 2kb files without any node and element information in it. Should I wait longer time to get complete information? What should the last message be in texgen ouput window for export model meshing? Thanks.
Actually, I tried to create mesh file follow your instruction. The texgen output command window shows the procedure. At last it will keep saving mesh and element orientation data into *inp and *.ori, and additional *.eld files. Strange things happened that, with same *.py file, sometimes it gives me a large file, while others it just gives 1 or 2kb files without any node and element information in it. Should I wait longer time to get complete information? What should the last message be in texgen ouput window for export model meshing? Thanks.
The meshing algorithm built into TexGen only works with Tetrahedral elements at the moment unfortunately. Building brick element meshes for completely arbitrary geometries is very tricky so it is not currently supported. You can use 10 noded tetrahedral elements though, but to be honest in the initial tests using these elements in ABAQUS I often ran into trouble with them. Some of the elements get quite distorted on the boundaries, I plan to alleviate this issue by having the option to keep the mid-side nodes sit on a straight line rather than getting projected.
About the mesh density, the seed represents the average edge length of the element. So smaller seed size = higher mesh density.
The .ori file contains element orientation data and the .eld file contains other element data such as element volume fraction. Note that in the latest released version of TexGen the .inp file refers to them by absolute path, so if you move the files after you save them you will need to edit the path. I have fixed this in the development release to use relative paths. I guess I should provide a new release soon as I have also made some bug fixes to the TexGen mesher.
I'm not sure about the file sizes, you may indeed have to wait a little longer for it to complete. When the meshing is complete you should see something like this:
"Meshing complete with 1765 elements and 457 nodes"
About the mesh density, the seed represents the average edge length of the element. So smaller seed size = higher mesh density.
The .ori file contains element orientation data and the .eld file contains other element data such as element volume fraction. Note that in the latest released version of TexGen the .inp file refers to them by absolute path, so if you move the files after you save them you will need to edit the path. I have fixed this in the development release to use relative paths. I guess I should provide a new release soon as I have also made some bug fixes to the TexGen mesher.
I'm not sure about the file sizes, you may indeed have to wait a little longer for it to complete. When the meshing is complete you should see something like this:
"Meshing complete with 1765 elements and 457 nodes"
-
lxtlixuetao
- Posts: 4
- Joined: Tue Nov 20, 2007 9:29 pm
For me, the advantage of TexGen is that it is easy to build geometry, while as you mentioned in the last reply, the meshing result always gives distorted element which will cause numerical errors in the following FEM analysis. So, what I want to do is, output geometry file, we say IGES or STEP files, which could be recognized by PATRAN or other codes, use them to manually control the details. Any suggestions on this? Or, any samples available?
Another point I want to discuss with you, due to characteristics of textile composite, the space between undulated fiber tows are quite small, meanwhile actually this interface material control damage type and final failure which is what we want to capture. So, the meshing in this places is critical issue. I read some paper, they do this with adaptive meshing which refine mesh in small gap and tow corners, and make coarse meshing in other places. It will reduce computational load in certain extend. Did you ever consider this? I, myself, is incapable to verify and finish this in short time, so just eager to learn from you and others, for benefit of your experience. Thanks again for sharing.
Another point I want to discuss with you, due to characteristics of textile composite, the space between undulated fiber tows are quite small, meanwhile actually this interface material control damage type and final failure which is what we want to capture. So, the meshing in this places is critical issue. I read some paper, they do this with adaptive meshing which refine mesh in small gap and tow corners, and make coarse meshing in other places. It will reduce computational load in certain extend. Did you ever consider this? I, myself, is incapable to verify and finish this in short time, so just eager to learn from you and others, for benefit of your experience. Thanks again for sharing.
Yes, that would be ideal. Just export to IGES or STEP, load it up in your favourite meshing/analysis software and hey presto. Unfortunately as you may have already discovered or will discover things are not quite so pain-free. After having spent quite a substantial amount of time dealing with these IGES and STEP file formats I have become increasingly frustrated with them. To the point where I've now given up on them, I would rather invest my time in improving the TexGen mesh generator.
But by all means, you can try to use them. The TexGen iges&step export will probably work more or less for very simple 2d weave models. File->Export->IGES File... or File->Export->STEP File...
The advantage of exporting meshes rather than geometry in the form of mathematically defined surfaces (i.e. IGES & STEP) is that a mesh is represented in exactly the same way in every software package. The other advantage is that we have more control over the mesh generation process, i.e. certain things can be adjusted while meshing to avoid issues.
The mesh generation in TexGen produces quite good meshes compared those general algorithms found in other software. In some cases it will produce better quality meshes, in other cases it won't so it depends on the fabric really. I suggest you try to use linear tetrahedral elements if your simulations aren't running.
I haven't implemented mesh refinement in regions of small gaps for the TexGen mesh generator. This may be something to I will consider implementing on the future. Thanks for your input.
But by all means, you can try to use them. The TexGen iges&step export will probably work more or less for very simple 2d weave models. File->Export->IGES File... or File->Export->STEP File...
The advantage of exporting meshes rather than geometry in the form of mathematically defined surfaces (i.e. IGES & STEP) is that a mesh is represented in exactly the same way in every software package. The other advantage is that we have more control over the mesh generation process, i.e. certain things can be adjusted while meshing to avoid issues.
The mesh generation in TexGen produces quite good meshes compared those general algorithms found in other software. In some cases it will produce better quality meshes, in other cases it won't so it depends on the fabric really. I suggest you try to use linear tetrahedral elements if your simulations aren't running.
I haven't implemented mesh refinement in regions of small gaps for the TexGen mesh generator. This may be something to I will consider implementing on the future. Thanks for your input.