Cartilage-Bone Tied Contact

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • OJMorgan
    Junior Member
    • Jun 2017
    • 20

    Cartilage-Bone Tied Contact

    Dear FEBio Forum,

    I'm trying to move from using a rigid body to using a linear elastic material to simulate bone; however, some of the ties between the cartilage-bone interface have not been successful. I've read in the forum that the tied contact in FEBio does not like curved surfaces. Unfortunately, most of the surfaces I need to tie are curved/elipsoidal. Is there a way of improving the tied contact's outcome? I have played with the penalty ranging 0.2-100 but still some of the surfaces are moving apart from oneanother. Is it a matter of playing with the penalty factor to find the optimum setting or are there more suitable options such as, welding the two surfaces (though not ideal). I am currently have the augmented lagrangian activated at value of 1 and the penalty factor at 1, also. Any help or advice is much appreciated.

    Many Thanks,

    Oliver
  • ateshian
    Developer
    • Dec 2007
    • 1824

    #2
    Hi Oliver,

    Tied contact interfaces should work equally well with curved or planar surfaces. Have you turned the auto-penalty on? This will make it easier to calculate a valid penalty parameter. With auto-penalty on, penalty values from 1 to 100 are good initial guesses.

    If you've exhausted your options, you can also try the tied biphasic contact interface (even if the materials you are connecting are just elastic, it will work fine with those too). It uses a slightly different formulation, which may be helpful. Keep in mind that this interface works best when setting the stiffness matrix to non-symmetric (in the control section).

    Best,

    Gerard

    Comment

    • OJMorgan
      Junior Member
      • Jun 2017
      • 20

      #3
      Hello Prof. Ateshian,

      The tied biphasic solution worked; however, I've had to revert back to using rigid bodies as my a new choice of analysis technique has meant the model has become unstable. The model is force driven, made up of a ground and a construct: rigid body1-cartilage1-cartilage2-rigid body2. The rigid body2 has 1 DoF (sagittal rotation) as though it is pinned at the posterior surface. This is to ensure even contact to rigid body1 which should have 6 DoF. The motion of rigid body1 is force driven by tensioning two deformable parts. It seems that unless the rigid body1 is constrained to 1 rotational degree of freedom and 1 translational degree of freedom it will intersect the ground and eventually move through it entirely. Though, even when constrained in this way the model will solve only to 30%. I'm in the process of using a rigid wall to simulate the ground but I feel this will not solve the issue of convergence. Both of the virtual cartilaginous surfaces have initial contact before the analysis commences as I realise a force driven simulation would not solve otherwise. A dynamic analysis may be more suitable for this type of simulation but I cannot use a biphasic material. Do you think creating an initial bolt load for the deformable parts may improve the analysis convergence?

      Many Thanks,

      Oliver

      Comment

      • OJMorgan
        Junior Member
        • Jun 2017
        • 20

        #4
        Hello Prof. Ateshian,

        The model is now converging. The strange results arise from two variations of the model I am running. When I use Linear springs to model ligaments the rigid body intersection is minimal to nothing but when I replace the linear springs with linear tension-only wires the the model construct intersects and passes through the ground. Is this a possible issue with FEBio, or in your opinion, is the lack of compression provided by the linear springs causing instability in the model?

        Many Thanks,

        Oliver

        Oliver

        Comment

        • ateshian
          Developer
          • Dec 2007
          • 1824

          #5
          Hi Oliver,

          Is this a possible issue with FEBio, or in your opinion, is the lack of compression provided by the linear springs causing instability in the model?
          In my experience, using tension-only springs (or tension-only fiber models) can cause instabilities in a model, so I would suspect that this is the issue with your model.

          Best,

          Gerard

          Comment

          • OJMorgan
            Junior Member
            • Jun 2017
            • 20

            #6
            Dear Gerard,

            Thank you for your advice. My final question is whether FEBio is able to introduce discrete materials such as, tension-only linear springs in a 2nd or 3rd step? The reason I ask is that I'm attempting a workaround by bringing each part into "tight" contact using a displacement driven step, and then introducing the the tension-only fibre models, hopefully reducing the possible slack present in the initial step. Currently, this model fails to begin, citing the tag "discrete" as unrecognisable yet the same string used to identify the springs is used when defined in the initial step.

            Thanks,

            Oliver

            Comment

            • ateshian
              Developer
              • Dec 2007
              • 1824

              #7
              Hi Oliver,

              Would you mind attaching a small example file that illustrates this issue? It would be easier for us to debug it with an example.

              Thanks,

              Gerard

              Comment

              • OJMorgan
                Junior Member
                • Jun 2017
                • 20

                #8
                Hi Gerard,

                Do you have an email I can send to as this is ongoing research?

                Thanks,

                Oliver

                Comment

                • ateshian
                  Developer
                  • Dec 2007
                  • 1824

                  #9
                  Hi Oliver,

                  My email is ateshian at columbia.edu

                  Best,

                  Gerard

                  Comment

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