contact between an irregular cylinder and 2 plates PLEASE!!!

Please attach the model.

Do you mean a screen capture of the CAE or the .inp fule?
Thanks!

The model .cae file.

hi Mechfeeney,
I guess the file is too big for ENGINEERING.com coz i get an error message.
i could put it in dropbox but id need an email address...
Let me know what youd prefer!
Carolina

I think that's against site policy. Is there a personal message option on this site?

im trying to find an alternative!

just take a wild guess..... at gmail.com

Try deleting your mesh and do a File-Compress.
A picture or and INP would also work.

Rob Stupplebeen

Sorry, rstupplebeen, ive just seen your message.

There are orphan meshes in my model that come from previous analysis. And I cannot upload even split zip files of the .inp here.
I see youre in the Yahoo group. I can send it to you there if thats ok with you. Please let me know.

Thank you!

Sure you can post it there. Please be sure to provides links between the forums. Alternatively you could place your file in filehippo, dropbox...

Rob Stupplebeen

hi rstupplebeen!

here it is:

Its massive but as i said these are orphan meshes and i did not want to delete any part.

Thank you so much for your help!
Carolina

Here is a running list:
1. Try using 2 analytical surfaces as the compression plates.
2. My guess is that you want to use a hyperelastic material model i have used this link in the past assuming you do not have material test data
3. Try using a displacement on the top plate instead of surface traction
I hope this helps.


Rob Stupplebeen

4. To speed debugging I would create a similar model with a drastically reduced node count or at least only model 1 half.

Rob Stupplebeen

Hi Carolina,
Sorry it has taken so long to get back to you (things have been really busy at work). Upon investigation of your model, I have arrived to similar conclusions as Rob.
1. The 3d scan imported cylinder is not refined enough. It contains several abrupt edges (almost like discontinuities; and the elements come to very sharp inverted spikes). The geometry is thus incredibly complex, but moreover, not realistic. Geometry like you have in the 3D scan is nearly impossible to work with, too many sharp edges and areas that resemble discontinuities.
2. For the purpose of your work, a better solution would be to drastically simplify the cylinder and work out a good material model.
3. Analyzing it with linearly elastic material is indeed a good starting point, but will not yield realistic results (for your application).
4. I noticed that you had several suppressed features here and there throughout the model and redundant material / section definitions. It is really important to keep a model clear, concise, and as simple as possible. Clutter makes things confusing. (This is just advice for the future : ] and I mean no disrespect).

For a first cut analysis, I would model a linearly elastic cylinder between a fixed plate and a plate that has a displacement (not a force). The cylinder should initially be very close, almost tangent to the plates. Do not apply a fixed boundary condition to the backside of the cylinder, as this would add unrealistic stresses and moments. Instead allow the cylinder freedom to move along the same axis of the displacement. Once you have the basic contact model setup and working, you can start to introduce the complexities like a hyperE material config. You can then start modifying the geometry of the cylinder to be a bit more realistic. It is highly impractical and inaccurate to try and model the 3d scan you have. As I said, the geometry is too complex and the scan is not smooth enough. One of the key FEA principles is to simplify as much as possible while maintaining accuracy. Your situation is that you have something so complicated that even if you get a working model, the results will not be trustworthy. For my own interest I will be working on an approach I think best suited for this problem. Good luck in the mean time.

To expand on mechfeeney's post. I would use your mesh to create a representative geometry in a CAD program such as SolidWorks.

Rob Stupplebeen

Dear rstupplebeen and mechfeeney,
Thank you for taking the time.
Ive considered smoothing but i dont have access to cad software that is compatible with Abaqus right here right now.
"Clutter" is mainly related to things i still have to add to the model. And, of course, i already have the hyperelastic version, but if i cant deal with elastic...

I will try using 2 analytical surfaces as the compression plates and displacements instead of tractions.

Thank you again,
Carolina

Any CAD software should be able to create this geometry. Then simply save as a SAT, STEP, IGS in that order of preference.

You can also create a simplified shape with Abaqus's sketcher. I made some measurements on your mesh and believe that a revolved ellipse with a height of 0.120mm and a diameter at the base of 0.135mm would be a good approximation.


I hope this helps.

Rob Stupplebeen

Nice Rob. That's a great starter model. Very kind of you.

Dear friends,
This is very kind of you, but i need to model real breasts.
I need to find the way with the real thing.
Cheers,
Carolina