Shell Thickness

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • ambaus
    Member
    • Jun 2008
    • 43

    Shell Thickness

    Hi Steve,

    This is to summarize our emails for anyone else who has the same problem. If FEBio terminates with a RunTime error, double check that shell thickness is not 0.

    Thanks,

    Anne
  • ambaus
    Member
    • Jun 2008
    • 43

    #2
    Hi Steve,

    This did not fix the problem. I even took out the shell elements (the bones) and just tried pulling on a neo-hookean ligament (I've attached this model make in PreView 1.2). I got the same error. Is this a bug in the program?

    Thanks,

    Anne

    Comment

    • maas
      Lead Code Developer
      • Nov 2007
      • 3400

      #3
      Hi Anne,

      Some of your elements have a really bad aspect ratio. You can see this in PreView if you select the Contour plot feature. In the dialog box that appears, select Jacobian and then select update (make sure the mesh is selected). You'll see that a lot of the elements actually have a negative jacobian (all the dark blue elements). So I'm afraid you'll have to go back to your meshing program and see if you can create better shaped elements.

      Cheers,

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

      Comment

      • ambaus
        Member
        • Jun 2008
        • 43

        #4
        So does this mean that I can have no negative jacobian elements?

        Comment

        • maas
          Lead Code Developer
          • Nov 2007
          • 3400

          #5
          Hi Anne,

          Sorry for the late reply, but technically yes. You cannot have any negative jacobian elements. However, if this is a nuisance I can see if I can relax this requirement a bit for rigid elements that are not used for contact.

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

          Comment

          • ambaus
            Member
            • Jun 2008
            • 43

            #6
            Hi Steve,

            Can you relax this requirement for deformable elements as well? I don't have negative Jacobians in contact regions. I know ABAQUS will solve a problem as long as no elements have 0 volume. Also, some on my negative Jacobians are very small when the simulation runs.

            Thanks,

            Anne

            Comment

            • weiss
              Moderator
              • Nov 2007
              • 124

              #7
              Hi Anne,

              If your deformable elements have a negative jacobian before the start of an analysis, there is no possible way to proceed in any finite element solver. this is because there is no unique mapping from the global coordinate system to the local coordinate system of the finite element, so none of the finite element based calculations can be performed.

              Physically, it means that your element has negative volume before the analysis begins.

              Cheers,

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

              Comment

              • ambaus
                Member
                • Jun 2008
                • 43

                #8
                Hi Jeff,

                Actually, I can get a mesh with negative Jacobians but still have non-negative volume elements (in IA-FEMesh). The writer of this program says ABAQUS will solve volumes with negative Jacobians.

                Thanks,

                Anne

                Comment

                • weiss
                  Moderator
                  • Nov 2007
                  • 124

                  #9
                  Originally posted by ambaus View Post
                  Hi Jeff,

                  Actually, I can get a mesh with negative Jacobians but still have non-negative volume elements (in IA-FEMesh). The writer of this program says ABAQUS will solve volumes with negative Jacobians.

                  Thanks,

                  Anne
                  Hi Anne - I don't understand parts of your statement above. Please clarify. Who is "the writer of this program", and which program do you refer to? I can assure you that, without modifying the topology of the element before the analysis, ABAQUS cannot perform an implicit nonlinear FE analysis on a deformable element that starts out initially with an negative jacobian.

                  Keep in mind that although the element may have a positive overall volume, if the jacobian is negative at any of the integration points, the volume at those points is negative in terms of the shape functions and the mapping from global to local coordinate systems. Thus the integration point operations based on the shape function mappings are ill defined and cannot be performed.

                  Cheers,

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

                  Comment

                  • ambaus
                    Member
                    • Jun 2008
                    • 43

                    #10
                    Hi Jeff,

                    Dr. Nicole Grosland is the developer of IA-FEMesh that I have been in contact with.

                    Your explanation makes sense as to why I would have an overall positive volume but still a negative Jacobian. Thank you so much for your input.

                    What I don't understand is that some of my parts have negative Jacobians but my model still runs in FEBio. They are not at contact areas. Would you expect that I would get non reliable results?

                    Thanks,

                    Anne

                    Comment

                    • maas
                      Lead Code Developer
                      • Nov 2007
                      • 3400

                      #11
                      Hi Anne,

                      So I have to agree with Jeff that initial negative jacobians are indicative of a problem in the mesh and any reliable FE solver should throw a fit when this occurs.

                      Now, why your model might run in FEBio although PreView says that you have negative jacobians, can be explained because PreView does not sample the element in the same way as FEBio does. FEBio samples the element at the Gauss-points of a 2x2x2 integration rule (at least for hex-elements), where PreView samples them at the element nodes. The reason for this difference is that I did not want to enforce a particular integration rule in PreView, and for that reason decided to sample the nodes, where the jacobians will always have their worst value. In other words, if PreView says you have no negative Jacobians, then you can be pretty confident that FEBio won't complain either about initially inverted elements (elements can still invert due to deformation though). On the other hand, if PreView says that you have negative jacobians, FEBio might or might not complain. But in any case, you should be very suspicious about your results in the areas of poor quality elements.

                      We understand that meshing a complex geometry is always a very difficult part of the modeling. We try to come up with element formulations that give reasonable answers even for poor quality elements, but you have to understand that there are certain bounds. A finite element solver can simply not proceed if the jacobian is zero or negative at the integration points, since this leads to non-physical solutions, and again, I would be very suspicious about any FE solver that claims otherwise. I hope this helps. Let me know if you need any further assistance.

                      Cheers,

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

                      Comment

                      • ambaus
                        Member
                        • Jun 2008
                        • 43

                        #12
                        Hi Steve,

                        Ok. This is starting to make more sense now. Thank you for the explanation.

                        Thanks, :-)
                        Anne

                        Comment

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