Contact issues using cell growth material to compress elastic material

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • aksteele16
    Junior Member
    • Jan 2019
    • 10

    Contact issues using cell growth material to compress elastic material

    Hi all:
    I am working on a project trying to use FEBio to model the growth of a cancer tumor around a duct in the body. As the most simple form of the model, I am modeling the cancerous mass using the cell growth material and forcing it to expand in a ring shape by imposing 2 rigid rings around it. To impose boundary conditions in cartesian coordinates, I am modeling a quarter of the setup. The geometry can be seen below (blue is cancerous cell growth, yellow is elastic material "healthy cells" and green and purple are rigid bodies). When I run the model as attached below, the cancerous cell growth behaves as I would expect, but I can not get the healthy cells (yellow) to compress without buckling and subsequently penetrating the surrounding rigid bodies. I have tried increasing the poisson's ratio, using augmented lagrangians, etc. Please let me know if you have any suggestions on how I can improve this situation (or any other general suggestions!). Any help would be much appreciated.

    Thanks,
    Ann

    ductal_carcinoma_v1.feb
  • weiss
    Moderator
    • Nov 2007
    • 124

    #2
    Hmm, this model does not run at all for me. Would you please double check that this file is the one that you ran and saw the yellow area compress and buckle?

    Which version of FEBio2 are you running?
    Jeffrey A. Weiss
    Professor, Department of Biomedical Engineering, University of Utah
    Director, Musculoskeletal Research Laboratories
    jeff.weiss@utah.edu

    Comment

    • aksteele16
      Junior Member
      • Jan 2019
      • 10

      #3
      Thanks for your quick reply. Yes this one works for me - I could attach PreView and/or PostView files if that might be helpful? I am running 2.8.5.

      Comment

      • weiss
        Moderator
        • Nov 2007
        • 124

        #4
        Yes, please attach the .prv file.
        Jeffrey A. Weiss
        Professor, Department of Biomedical Engineering, University of Utah
        Director, Musculoskeletal Research Laboratories
        jeff.weiss@utah.edu

        Comment

        • aksteele16
          Junior Member
          • Jan 2019
          • 10

          #5
          See attached.

          ductal_carcinoma_v1.prv.zip

          Comment

          • brandonz
            Member
            • Oct 2014
            • 48

            #6
            Hi Ann,

            I looked at your PreView model. In the contact named "Sliding-elasticContact1", your master is Surface8 and your slaves are Surface17 and Surface19. However, if you select Surface17, you will notice it is actually the top of the healthy cells, and not the side facing the stroma. Your large penetration is occurring because you haven't defined any contact between the healthy cells and the stroma. If you replace Surface17 with Surface13 (the correct face), your problem goes away.

            When faced with issues like this, I always first plot all the contact variables (contact area, traction, pressure, gap, etc). In PostView, I displayed the contact area and immediately noticed no value was displayed on the healthy cells facing the stroma but an area was defined on the top surface, which led me to realize you must have just clicked the wrong surface. This simple debugging method works because PostView will only display contact variables on surfaces where contact is defined; thus if no variables are displayed, that surface is not a contact surface.

            When I fixed your contact definition problem, your model runs fine, but the penetration is large. I briefly played with this and my best results came with augmentations, a tolerance of 0.05, and a penalty of 0.4 with auto penalty on (for both sliding-elastic interfaces), although the results were still not great.

            I noticed your healthy cells are a neo-Hookean material with v=0.49. Combining a nearly-incompressible material with large swelling in a confined problem is very difficult, and the coarse mesh does not help. (Not to mention that neo-Hookean materials may not be the best choice for v=0.49...a nearly-incompressible Mooney-Rivlin may work better here). You likely will need to refine the mesh or choose a different material model to get better results.

            Brandon

            Comment

            • weiss
              Moderator
              • Nov 2007
              • 124

              #7
              Thanks Brandon!

              Jeff
              Jeffrey A. Weiss
              Professor, Department of Biomedical Engineering, University of Utah
              Director, Musculoskeletal Research Laboratories
              jeff.weiss@utah.edu

              Comment

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