Cartilage Contact Problem - No deformation at all on one part of the geometry

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • sajids
    Junior Member
    • Nov 2019
    • 7

    Cartilage Contact Problem - No deformation at all on one part of the geometry

    I am in the process of modeling the medial half of the knee for a cartilage contact problem. I have designed idealized geometry of the medial size of the femoral cartilage above the tibial cartilage. As I am examining a steady-state loading plateau, I am neglecting viscoelasticity or any fluid flow, purely matrix stiffness. The cartilage is modeled as neo-hookean with a density of 1100kg/m^3, 3.3MPa stiffness, and 0.4 poissons. Both the femoral and tibial components are treated this way. On the articulating surfaces, I have an sliding elastic contact interface with a penalty of 1000, and frictional coefficient of 0.02. My boundary conditions no movement at all on the lower surface of the tibial cartilage (where it would contact bone) and z only movement on the opposite femoral cartilage side (again, where it wound contact bone.)

    The model runs in a reasonable amount of time, but always ends with error termination as negative jacobians are detected for time points. When examining the results in postview, I notice the femoral component is deforming as expected, but the tibial cartilage exhibits no deformation _at all_. The femoral cartilage eventually pushes "through" the tibial cartilage and they wind up overlapping, hence the negative jacobian and resulting errors.

    Does anyone have any insight on this? It isn't the boundary condition on the tibial cartilage- i've double and triple checked that its ONLY the underside surface, and nowhere else.
  • maas
    Lead Code Developer
    • Nov 2007
    • 3400

    #2
    Hi,

    It sounds like a problem with the contact. I would first double-check that the correct surface selections are assigned to the primary and secondary surfaces. It might also be the penalty value. If you entered the neo-Hookean's elastic modulus (E) as 3.3e6, than a penalty of 1000 might be low, although it also depends on your element size. You can try turning on the "auto-penalty" option, which calculates a default value for the penalty. In that case, the "penalty" parameter becomes a scale factor to this auto-penalty, so I would start by reducing it back to 1, and then increase as necessary to reduce the contact overlap.

    Cheers,

    Steve
    Department of Bioengineering, University of Utah
    Scientific Computing and Imaging institute, University of Utah

    Comment

    • sajids
      Junior Member
      • Nov 2019
      • 7

      #3
      I have the femoral articulating surface (the one pushing down) as the master, and the tibial surface as the slave.

      I had no idea the penalty depended on the loads involved/stiffness, I will try that!

      Comment

      • sajids
        Junior Member
        • Nov 2019
        • 7

        #4
        So i've updated the model with the penalty factor and rechecked the contact surfaces- all look good. I don't see one half of my model moving "into" the other half anymore, but I still see essentially zero displacements on one half of the surface. Would I be able to upload the .feb file so someone can take a look?

        Comment

        • maas
          Lead Code Developer
          • Nov 2007
          • 3400

          #5
          You may email it to steve.maas@utah.edu.

          Cheers,

          Steve
          Department of Bioengineering, University of Utah
          Scientific Computing and Imaging institute, University of Utah

          Comment

          Working...
          X
          😀
          😂
          🥰
          😘
          🤢
          😎
          😞
          😡
          👍
          👎