ligament penetrate into cartilage--contact problem

[YIDAN]

Hi,

I am a master student and work on my thesis by using the knee joint model from SimTK (https://simtk.org/projects/openknee). Now, I tried to apply 1 NM internal rotation on the tibia, but the result showed that the MCL penetrated into the femoral cartilage. Then I defined the sliding-elastic contact between the MCL and cartilage, but it still didn't work. Even I magnify the value of penalty factor up to 1 in augmented lagrangian. It seems nothing changed. Could anyone help me to figure out the issue? Any help, I will appreciate it a lot!

The result (at final step) is shown as attached, where you can see that the MCL penetrated into the cartilage.

Since the Feb file is too big, I will share the link here:https://drive.google.com/open?id=1gN...1TZcGCj2aYQunF.


Sincerely,
Yidan

[ateshian]

Hi Yidan,

I looked at your model and I noticed that the contact interface your created with the MCL includes not only cartilage but part of the tibia, which is a rigid body. In that case, since a rigid body is involved, you should use a single-pass analysis where the rigid surface is the master. I did this change and I still saw a little bit of an overlap between the MCL and cartilage, but it only occurred with a single layer of elements on the articular layer which is contacting the MCL at angles for from 180 degrees (i.e., the unit outward normals of the opposing contact surfaces are not pointing toward each other much). So I increased the moment from 1 N.m to 2 N.m (I also turned off the Lagrangian augmentation for efficiency) to see if the overlap would get worse but it didn't: The contact is indeed engaging as it should, it just looks awkward because the normals are not facing each other.

Keep in mind that the contact algorithm shoots rays from the integration points on the slave surface which intersect the master surface. The contact is strictly enforced only at the integration points of the slave surface. If the mesh on the slave surface is relatively coarse (as it is for the MCL), the integrations points (typically there are four of them in a quad4 surface element) are sparsely distributed on each quad4 and the master surface elements may penetrate the slave surface elements significantly until it encounters one of those integration points. I think that's what's happening in your model.

Best,

Gerard

[YIDAN]

Hi Gerard,

Thanks for your kind reply!

According to your suggestions, I turned off the two_pass, but the overlap situation didn't change (i.e., overlap area became smaller or lager). Also, I changed the rotation moment from 1NM to 2NM, but the overlap became more severe. So I'm curious how you make it didn't get worse.

Also, your suggestions might be where the issue is. The normal of the master and slave surface can't point to each other much. So I am still thinking about how to solve it. If I change the value of the projection tolerance and search radius, will the problem be solved?

Thanks again for your help! I appreciate it a lot!!

Sincerely,
Yidan

[ateshian]

Hi Yidan,

Here is the list of changes I made to your file:
1) Lines 234132-234133: Changed

<laugon>1</laugon>
<tolerance>0.01</tolerance>

to

<laugon>0</laugon>
<tolerance>0</tolerance>

2) Line 234137: Changed

<two_pass>1</two_pass>

to

<two_pass>0</two_pass>

3) Line 234211: Changed

<moment lc="3">1000</moment>

to

<moment lc="3">2000</moment>

I checked the contact force between the MCL and tibia/cartilage surface and it increases steadily after contact is initiated (see tf_joint_IR03_1NM03.png).

Best,

Gerard