5 Replies Latest reply on Mar 21, 2011 11:30 AM by David Suelflow

    Shell to Solid bonding Errors

    David Suelflow

      We have a truss that we want to analyze just one joint so to save time we have most of it as shell elements and the joint in question as solid.  When we try to run the simulation we get an error stating that the contact between the solid and shell is no good.  Does anyone have a suggestion as to best way to connect these two?

      Attached are some screen shots with the solids highlighted…

        • Re: Shell to Solid bonding Errors
          Jerry Steiger

          David,

           

          Did you ever work out how to mate the shells and solids?

           

          Another possible solution would be to do the analysis of the complete truss as shells, then analyze just the joint in question with loads from the first analysis. I know that ANSYS has tools to make this sub-structure analysis relatively easy, although I've never actually done one. I don't know if Simulation does.

           

          Jerry Steiger

          • Re: Shell to Solid bonding Errors
            Bill McEachern

            If hte blue highlighted bits are the extents of hte solid bodies you are in good shape. Just bond the egde of the shell to the face of solid - this would be the face ot hte solid that is normal tot he axis of the member. It shoudl work fine.  See attached image for the general idea.

            • Re: Shell to Solid bonding Errors
              David Suelflow

              Yea, we ended up bonding the edge of the shell to the face of the solid… the problem we then ran into is Simulations’ use of a tetra mesh rather than the more accurate hex.  We ended up doing the full truss analysis in RISA 3D and then applying the forces in Abaqus.  We still run into some meshing issues that we have to work through but in the end it gives us the best results.

                • Re: Shell to Solid bonding Errors
                  Bill McEachern

                  I disagree withthe statement that HEX's are more accurate than tet's though I would certainly agree that they are more efficient. What made the RISA/ABAQUS solution better/more accurate and how did you know it was more accurate? Why wouldn't you have done the whole thing in ABAQUS as it is a pretty good code? I am not trying to defend simulatin here in anyway shape or form - it has lots of issues. Just curious about how you determined what was what on the problem.

                    • Re: Shell to Solid bonding Errors
                      David Suelflow

                      I’ll leave the tet vs. hex debate to the engineers…  I’m just the recovering architect modeler and am just going from what they tell me.  Same with the accuracy… I’m just repeating what I hear.

                       

                      We initially wanted to do it all in SW but ran into meshing issues, so we switched to Abaqus.  We were already doing an analysis in RISA for other purposes so we co-opted their results for our tests.  If we did not have them, we would have done a stick model in Abaqus and replaced one of the stick joints with a solid one.

                       

                      The reason we model in SW is that the shape of the joints and welds are complicated and we are new to Abaqus and not well versed in its modeling tools.  Attached is a sample… Isolate one of the “Weld Body” and you’ll see what I mean.