Cartilage Repair Indentation - Zero Diagonals

[sajids]

Hello, I am a graduate student at RPI using FEbio to simulate the effects of inclusions or defects on a repair model of cartilage. My model is a tube with hole (so a short cylinder with the center removed), then another cylinder to fill that exact space. Then, I have a 3rd short cylinder which I am treating as an indenter. My analysis is biphasic. I have modeled the indenter as a rigid body and constrained it so it can only move in the z direction. at time = 1, a 0.01N force is applied and meant to compress the cartilage underneath. Both cartilage samples are modeled as biphasic materials, neo-hookean with perm-const-iso permeability. I have boundary conditions on the underside of the cartilage pieces, so no fluid flow there and no movement in any axis. I have 2 contacts set up, a biphasic contact between the indenter and cartilage surface, and sliding elastic between the two pieces of cartilage themselves. Yet every time I run it in FEBio, it runs for a little bit, then just reports Zero Diagonal, aborting run. I suspect the problem is with my time points or amount of load, but I am relatively new to FEA and am unsure.

I have attached the .feb file if anyone can help and take a look.

[ateshian]

Hi,

I think the main issue with your model is that the force on the indenter is going upward (positive) when it should be pointing downward. So you need to negate the load you are applying, to maintain contact between the indenter and the biphasic material.

Other things I noticed in your model: (1) The load you prescribe is 0.01 and the load curve you use is a step function equal to 0.01; keep in mind that this means you are applying 0.01x0.01 = 0.0001 amount of load. If that's not your intention you should change one of these two values. (2) You are using a sliding-elastic contact between the inner and outer cores, though they are both biphasic materials. If you want to maintain continuity of the fluid pressure between these two domains you should use a sliding-biphasic (or tied-biphasic) contact interface. (3) For biphasic contact analyses it is probably best to use non-symmetric stiffness (instead of the default, which is symmetric). In that case use Broyden instead of BFGS, or even better, use full Newton (max_ups = 0).

Best,

Gerard

[sajids]

Gerard,

Thank you so much for your help. Yes I had tried using the load to be negative/positive, but didn't observe a difference. However, i did not know that the load on the UI vs curve editor both applied, I thought they had to be the same. As for the contact, I have altered that as well, along with the non-linear solver. Worked like a charm! Thanks again : )

Another question- how can I mesh my models in 3D, through the full depth? I'm ultimately interested in exporting the nodal displacements to Matlab for analysis there. If you had any other insight, I'd appreciate it very much.

[ateshian]

Hi,

Can you please clarify the question "how can I mesh my models in 3D, through the full depth?"

Best,

Gerard