Stopping of simulation when turning a CVT system 2

[sandarul]

Hi,

I'm trying to learn abaqus because I want to simulate en CVT system with it. After a lot of trying a got finally me model in the starting position. I made a belt which is clamped between 2 sets of poulies. When I'm turning one of the poulies, abaqus runs into simulation problems (to small increments).
I have already tried finer mesh and added some damping to the step. The last one seems to help, only then my poulie goes trough the belt. Also with damping after a while it also stops simulating. I've already changed a lot of parameters but it don't work.
Is there anyone who can give me a push in the right direction?
I've added the file as attachment.

 

[rstupplebeen]

I suggest using a symmetry plane and significantly increasing the number of elements on the belt. Also, try having 1 step to pretension the system then a rotation step instead of 4 steps. I hope this helps.

Rob Stupplebeen

 

[sandarul]

Hi,

Thanks for your help.
Officialy I may not use a symmetry plane, because the belt will be a bit out of the centre normally. But better having a working model with symmetry than have nothing. So I changed my model to your comments but it is still not working. I also tuned a bit with damping but this results in a pulley which goes trough the belt.

the odb output file can be download here:

http://wtrns.fr/Xl4IhedGeFmpv-

 

[rstupplebeen]

Here is a running list:
1. turn on NLgeom (If your problem has large deformations or contact it probably needs this)
2. The pulley in the +X has a slight penetration with the belt (moved -1unit in the X dir)
3. Fix belt is not needed in the initialization step
4. Split belt and modified mesh
5. Z constraint for split in belt for initial and step 1
6. Changed rubber's Poisson's ratio to 0.45 (0.5 produces a numerical singularity) you probably need to model as a hyper-elastic material.
7. Modified step sizes (a lower bound of 1e-11 would take forever to solve and probably has other issues if this is needed)
8. Quadratic elements
9. Split the first step into one to ensure that contact is occurring with stabilization (automatic) on and the second with the rest of the deformation.
10. Reduced initial pretensioning (a hyperelastic material model is probably needed)


I hope this helps.


Rob Stupplebeen

 

[rstupplebeen]

Below is an image of the pretensioned state.

I have also created an animation that is not uploading so I have put it on my webpage at the bottom.

https://sites.google.com/site/robertkstupplebeen/home/engineering/fea

Rob Stupplebeen

 

[sandarul]

Thank you very much. It is finally working. I'm now doing some fine tuning and after that I will upload my result.
I understood all of your comments except 2 of them:
4. Split belt and modified mesh
5. Z constraint for split in belt for initial and step 1
Did you mean the symmetric split (split in Z-X plane)? Than I can't remove the Z constraint right?
I'm so grateful for your help. I was really desperate after 3 weeks of work. But thanks to you I have finally a working model.
Also thanks for you movie it helped me with making it work.

Sorry for my typo's in my previous messages but I can't find the edit button.

 

[rstupplebeen]

That's great I am glad I could help.
4. I used "partition cell" to cut the belt into 2 pieces. This served 2 purposes: a. aided meshing b. provided surfaces for #5
5. This was applied to the new faces and prevented the belt from rotating while pretensioning. It may not be needed though.

I hope this helps.

Rob Stupplebeen

 

[sandarul]

Thanks, I Understand it now.
I first made the same symmetric model and it works.
Then I changed my first model with 4 pulleys and it also works.
Movie of that:

http://www.filedropper.com/nosymetricsimplebelt

(sorry I misspelled symmetric)

Then I made a better v-belt (called belt2) but it had to many elements and it stopped after 10 hours in step 2.
After that I simplified this belt (belt3) but also this belt stops on step 2.
I don't really need the more detailed belts but it would be nice if it works and I want to know if this influences the efficiency.
Also there is something strange with the transmitting efficiency, the efficiency is around 40%.
So I have a lot to try but I am glad that the model finally works. Thanks for your help.

CAE file:

http://www.filedropper.com/simpelpoelie83

 

[sandarul]

I have a new question, I want to improve the model of the rubber v-belt but I'm not sure how to model.
The v-belt excist of diferent layers, to simplify it excist of rubber and some fibres like kevlar in the longitudinal direction. Should I use a hyperelastic model or use an elastic anisotropic model. Because the belt has a different behaviour in the transverse direction compared to the behaviour in the longitudinal direction.
I read a lot about material properties in the ABAQUS books but could not find a proper solution for this problem.

A picture of a cogged V-belt:

http://www.ebmmill.com/images/Dayco/Gold Label Cog Belt Cut Away copy.gif

I still have to do the tests on the belt but before I do them I want to know which tests I have to do.

Thank you in advance.