Sliding Interface

[ambaus]

Hello computational hobbyists,

I have 2 parts that I want to enforce a non-penetration constraint between. However, when I run the simulation with a sliding interface contact, penetration still occurs. What parameters should I play around with (penalty, tolerance) to make this a harsher constraint? One surface is a soft tissue (picked as the slave) and the other is a rigid bone (picked as the master). Any help would be much appreciated.

Thanks,

Anne

[maas]

Hi Anne,

There a couple of things you can do to reduce penetration. My first question to you is are you using augmented lagrangian or not? The aug lag method will often reduce the gap further than can be obtained with the penalty method. If you are not, then I would definitely give that a try. In any case, increasing the penalty will almost always decrease penetration. I hope this is useful. Let me know if you have any further questions.

Cheers,

Steve.

[ambaus]

How do I know if augmented lagrangian is on? I have the 2 pass method on. Is that the same thing?

Anne

[maas]

Well, actually, it is on by default so you'll probably have it on. You can explicitly turn it on or off using the laugon parameter (1=on, 0=off).
Regarding the 2 pass option, the contact algorithm differentiates between the two contacting surfaces: one surfaces determines the contact traction, and the other one is used to measure the gap. The two-pass algorithm applies the algorithm twice, where in the second pass the role of the surfaces switches.
Did increasing the penalty help at all?

Steve.

[ambaus]

Hi Steve,

Yes, increasing the penalty did help. Thanks a bunch.

Anne

[Rosie Richards]

Hi Steve,
I'm also having a problem constraining my parts so that they won't penetrate through each other. I have tried increasing the penalty as you mentioned above and also have tried using the augmented langragian approach. Even with a penalty factor of 100000 there is no change. I am wondering what else to try. I have attached the folder in case you can spot something that is obviously wrong with the contacts.
Thanks very much in advance for your help.

Rosie

[maas]

Hi Rosie,

I think the problem lies with the way you have your boundary conditions set up. I've made some changes to your file and attached the modified PreView and FEBio input file. I did not attach a plot file, but the problem should run now. The most important changes are:

- I've replaced your very stiff isotropic elastic materials with rigid bodies.
- I've replaced your nodal prescribed displacements with a rigid body displacement.
- I've made the problem symmetric. Instead of moving one block, I've moved both blocks in opposite directions and half the distance of the original displacement. This allowed me to add some additional boundary conditions that help stabilize contact (these are the fixed BC's on the side edges of the cylinder).
- I've redefined your contact surfaces. You had both sides of each block as part of the contact interface although only the side opposite the cylinder participates.
- Finally, I tweaked your contact parameters so the problem runs reasonably and with a fixed time step.

As you will see, the problem now runs all the way, but at the end of the simulation you are still getting a lot of penatration. I will leave it to you to try to fix that . Oh yes, I also removed the pressure condition. I removed it just to simplify the problem a bit. I guess the first thing you should try is to add it back and see if the problem still runs well.

Let me know if you have any further questions.

Cheers,

Steve.

[Rosie Richards]

Thank you so much! That's really very helpful info. Although when I opened the attachment it was the same one as I had posted. So I have been trying to re-do the model myself using your instructions. For some reason tho, Preview now keeps crashing when I try to save it.

Rosie

[maas]

Hi Rosie,

Sorry about uploading the wrong files. I uploaded the right files this time (in the previous post).

Can you upload the PreView that's crashing?

Thanks,

Steve.

[Rosie Richards]

Thanks for uploading those files so quickly. Very much appreciated. I haven't uploaded the preview file as its just blank, because it crashes before it saves, so I'm guessing its of no use to you.

Rosie

[StudentPaul]

penetrating objects

Hi Steve,

i also have problems with penetrating objects in my foot model. I tried varying the penalty factor from 0.01 to 100 but none of these values resulted in contact between the the foot and the plate. Even with auto penalty on or with laugon=1 i had no contact, so i don't know what to do now. Hopefully you can help me. As you can see in the .plt file there is also some movement in the y direction, perhaps you know why this movement occurs?
Thanks in advance.
P.S. my zipfile is too large so i'll try to mail it to you.

greetings, Paul

[maas]

Hi Paul,

The first thing you need to add is a constraint in the y-direction. Currently, your model is underconstraint and this will make convergence extremely difficult. Second, turn of the auto-penalty flag. Auto-penalty does not work very well when one of the contacting bodies is rigid. With the auto-penalty turned of, you will have to play with the penalty parameter to find a good result. Try these changes and let me know what you find.

Cheers,

Steve.

[StudentPaul]

Hi Steve,

I constrained the y-direction and turned the auto-penalty off. I played with the penalty parameter and found that values varying from 0.05 to 5.9 result in penetration. However, a penalty of 6 resulted in a negative Jacobian and failure to converge. It looks like the foot is in contact with the plate at this moment. Any suggestions?

greetings, Paul

[maas]

Hi Paul,

You have your augmentation tolerance set pretty tight. Try changing the tolerance parameter from 0.01 to 0.1. That worked for me.

Cheers,

Steve.

[StudentPaul]

Yes, that works! Thank you very much!