3 Replies Latest reply on Dec 6, 2011 2:06 PM by Bill McEachern

    Solid mesh volume failure

    Will Swanson

      In attempting to create a solid-body curvature-based mesh, I have encountered an error that I can neither trouble-shoot nor explain.

       

      Attached is a PDF that shows the model and the issue.  There are (5) intermediate horizontal members that frame into a vertical member, with sheeting on one side.  All bodies are fully combined.  When any (4) of the (5) intermediate members are present, the mesh is successful.  However, when all (5) are present, the mesh fails at the volume level.  Mesh parameters are given in the PDF, and all parameters were kept constant in each mesh.  I also do not believe that global size of the mesh is the cause, because I repeated the same procedure using the full model (essentially a mirror of what is seen here) and, again, (4) of (5) meshed correctly for identical element size parameters.

       

      Has anyone seen anything like this?  Any additional advice?

       

      Thank you.

        • Re: Solid mesh volume failure
          Bill McEachern

          I would avoid the whole issue and get a better analysis suing shell elements. Would take about 10 min or so to pull off by the looks of it. Are you trying to do this with Xpress product?

            • Re: Solid mesh volume failure
              Will Swanson

              I agree that shells elements may be better suited, but the observed peculiarity is the underlying question. I have had a nearly identical experience with bolt groups and am curious if there are any explanations.

               

              In the present case, I have found a fix: If you notice in the PDF, the max-to-min element size ratio is 16-to-1 (4in to 0.25in).  I had adjusted each of these parameters individually, but when I allowed SW to default to the standard 5-to-1 ratio (now using 2.2in to 0.44in), it was able to mesh. 

               

              Curioser and curioser. If anyone has an explanation on the importance of this ratio in curvature-based mesh generation, I would be interested in further information.

                • Re: Solid mesh volume failure
                  Bill McEachern

                  Sorry thought you were trying to solve a particular problem.

                  On meshing: I would be careful when using hte curvature based mesher. They made it the default mesher in 2011. However, in my experience, it produces some element with very bad aspect ratios - you could get like 10^6 aspect ratios as a maximum. What's with that? Did you notice they took out the max apsect ratio and jacobian out mesh details? Not a great move in my humble opinion.

                  On the other hand the curvature mesher is much easier to get a mesh with. It is worth spending the time to get a mesh with the standard (Distenne) mesher? Depends on the situation really and where the bad elements end up.

                  By specifying a large element size variation, it can sometimes be difficult with small surface patches to make the transitions. Further if it does manage to get the surface mesh, if there is little room in the volume making the transitions across the thickness can pretty hard to pull off. I would hazard a guess that if your parts where thicker you could probably get it done. Or as you found out reduce the biggest element or limiting the range could also make the volume mesh easier to pull off with much less variation to deal with.

                  Some one from solidworks maybe able to shed more light on the details of how the mesher really works.