Compression Simulation

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • cgardner
    Junior Member
    • Nov 2009
    • 3

    Compression Simulation

    Hi! I'm an undergraduate student at Rutgers currently working with a group on a project, and we would like to use FEM to test some of our methods, but are running into difficulties with this package:

    What we would like to simulate is compression of a soft cylinder between two plates. I've constructed this model in the attached file. I've set (what I believe are) appropriate materials on everything, and have constrained one of the plates while applying a pressure force on the other. I also constrained the nodes along the bottom of both plates to keep them from moving in the z-direction. Whenever I try to evaluate it however (and I've tried a lot of different things), the very first element generates a negative jacobian.

    Now, I've never used anything like FEM before, and I have very little experience even in 3-d modeling, so I'm aware there's probably a lot of stuff I'm not accounting for. I understand everything about defining the elements and setting boundary conditions and all that, but I'm having some trouble grasping exactly what it is I need to do to even compress a cylinder like this - I've gone through the examples, but I can't figure out what it is I'm doing wrong.

    Any help would be enormously appreciated!

    PS - In addition to compressing the cylinder, is it possible to construct a soft object with a stiff object buried inside of it? I'm not sure how to go about this in PreView, other than creating the stiff object and simply dragging it down into the middle of the soft object - though I don't know if that's acceptable.

    edit - sorry about that, file's the .feb now
    Last edited by cgardner; 11-12-2009, 10:42 PM.
  • maas
    Lead Code Developer
    • Nov 2007
    • 3400

    #2
    Hi,

    Can you please attach the .feb file instead of the .prv file.

    Thanks,

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

    Comment

    • cgardner
      Junior Member
      • Nov 2009
      • 3

      #3
      Ok, I've worked out just compressing a cylinder between two plates, and now I'm attempting to compress a cylinder with a small stiff block in the middle (a crude representation of a tumor).

      I've cut out some elements from the square inner radius of the cylinder, and dropped a small block into the space. When evaluating however, the block slides right out of the space into the body of the cylinder. Increasing the penalty for the block-cylinder sliding interface helps a little, but any serious increase causes a bunch of negative jacobians.

      Is there a better way to create a cylinder with something in the middle?
      Also, what else can I do to get the block to keep from pushing into the cylinder?

      Comment

      • maas
        Lead Code Developer
        • Nov 2007
        • 3400

        #4
        Hi,

        I think that FEBio contact's detection algorithm might be confused by the fact that you chose all four sides of the cube to be part of a single contact interface. I will use this problem as a test case for the contact detection code, but in the mean time, there are a couple things to get around the issue. First, you can create four separate contact interfaces, one for each side of the cube. Alternatively, you can create a conforming mesh, meaning that the cylinder and box are one mesh (not two distinct meshes as you have it now) and simply assign different material parameters to the part of the mesh that corresponds to the box. It think you should ask yourself if a sliding contact is the right type of contact that you want. It seems to me that the box isn't really sliding with respect to the cylinder, so if you can't create a conforming mesh, try using a tied interface instead. I hope this helps. Let me know if you have any more questions.

        Cheers,

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

        Comment

        • cgardner
          Junior Member
          • Nov 2009
          • 3

          #5
          Thanks for the advice!

          I had some success using a tied interface, but the stiff block was only a single element - making the block from a few hundred or even just a few dozen elements generated negative jacobians.

          I've tried making a conforming mesh, but it generates negative jacobians almost immediately. I also separated the elements in question to form their own object, but this didn't work either.

          At this point, I have tried the following:

          - Defining the block as a rigid body and using rigid interfaces
          - separating the sides of the block into separate interfaces
          - using only two of the four sides as interfaces
          - and each of the last two, both as a separate object and as a conforming mesh


          I have also adjusted the maximum amount of force being applied from 1 to 2 to 20, with the same results. My question now is the same as at first - what am I doing wrong? Perhaps my formulation of the problem is flawed somehow?

          I appreciate all the help!

          Comment

          • maas
            Lead Code Developer
            • Nov 2007
            • 3400

            #6
            I think I may not have been entirely clear. By conforming I meant that the mesh of the block shares its surface nodes with the mesh of the cylinder, such that there is no need to define a contact interface. In other words, there is only one mesh, not two separate ones.

            However, I think the main reason why this problem isn't running, is because you are applying a force driven boundary condition. Force driven problems with contact are usually problematic. The reason is that intially there is no contact, and the rigid body is not constrained to resist the pressure force. Without the contact tractions the rigid body will immediately fly of in space. Since it seems that this model is just a test case, I would suggest to apply a prescribed displacement to the rigid body instead of a force. That takes one problem out of the equation.

            Another thing is that this model shows symmetry. Perhaps you can model only half of the model and apply a symmetry boundary condition to the symmetry plane. That will remove one more contact interface and one more source of problems.

            Cheers,

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

            Comment

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