4 Replies Latest reply on May 18, 2011 4:08 PM by Bill McEachern

    Solid/Shell Error

    Matt Medford

      Using Simulation 2010, I'm running a static analysis of a clamp fixture (1/8th sector - see attached picture).  There are 4 bolts around the perimeter to hold the fixture together, and 5th bolt in the center which applies pressure to a plunger.  The plunger sandwiches a copper disk with a thin (.0003" thick) titanium foil.  You can't see the foil in the attached picture because it's so thin relative to the rest of the parts.  The loads in the bolts are accomplished by defining interference contacts at the bolt heads.  (Of course, this is a pain in SWX because you have to iterate on the bolt lengths - which requires bolt models of various lengths - until you get the right load values.  In Ansys, you would simply define the contact and change the interference as an option in the contact element.  Much easier to change a value in a dialog box than it is to "Replace Components".  But this is all a separate discussion and I'll get off my rant now...)

       

      Most axial interfaces are defined as No Penetration.  The bolt thread contacts are bonded.  As mentioned, the bolt heads have intereference fits.

       

      The foil is modeled as a surface-plane that is then converted to sheet metal with the .0003" thickness.  This then automatically meshes as a shell.  (If I model it as a solid, Simulation chokes on appying the symmetry constraints to the symmetry planes of the foil and also bungs up the meshing of the foil - obviously because the aspect ratio is so severe with such a thin element.)  When I try to Run the analysis, I get the attached cryptic error.  In addition to lousy English, it's difficult to make out what the error is suggesting to try to remedy the problem.

       

      And, yes, it's necessary to have the foil in the analysis.  It's obviously too thin to contribute to the stiffness of the structure, but I need to look at how the pressure profile on the foil/copper interface changes when temperatures are applied (very different CTE's) and friction is considered.

       

      I'd be much obliged for any suggestions and or commentary.

       

      Thanks.

      cryptic.jpgclamp-model.jpg

        • Re: Solid/Shell Error
          Jerry Steiger

          Matt,

           

          I'm just guessing here, but it might have to do with how the thickness is added to your surface model. I can imagine that the thickness could be added symetrically up and down, all up or all down. If your model was built assuming one of the choices but SW uses another, then your parts are going to be interfering on one side and have unwanted clearance on the other.

           

          It seems like you would be better off to run the analysis in two steps. In the first step, ignore the foil. In the second, just look at the anvil, copper, foil, plunger locally (just fairly thin slices of the anvil and plunger) and appy the loads from the first step. I think this is called substructure analysis in ANSYS. I've never run one, but it seems like it would make the problem much easier to solve.

           

          Jerry Steiger

            • Re: Solid/Shell Error
              Matt Medford

              Thanks for the response, Jerry.  First, there does not seem to be an option to change the direction that SW applies the thickness when converting a plane to sheet metal.  And when it's meshed by Simulation, it seems to automatically apply it to the midplane.

               

              So I played around with the shell definition by selecting sides to apply the shell thickness in the analysis tree, and that eliminated the error.  However, the model now doesn't solve because the plunger flies off to infinity (presumabley it can't figure out the no penetration contact).  I frickin' hate Solidworks Simulation.  Too bad Ansys is so bloody expensive - if I had access to it I'd have been done with this analysis a week ago.

                • Re: Solid/Shell Error
                  Jerry Steiger

                  Matt,

                   

                  Well, I often have problems with ANSYS as well, with parts flying through one another. I'm afraid I can't help with SW Simulation. I assume they have something similar to contact normal stiffness and pinball radius. The other thing I can think of that might help would be to minmize the time step, but if you are doing a linear analysis, you probably can't control that.

                   

                  Jerry Steiger

                  • Re: Solid/Shell Error
                    Bill McEachern

                    Best practice is to get rid of all solids you want to treat with shell elements and replace them with surfaces (offset the surface to wherever you want). In 2011 you have the abaility to offset the surface to put the thickness wherever you want it (I don't like the specification myself - percetange of thickness - it would be far more convient to use an offset for the center). The above method makes it obvious what shell has a chance of being automatically bonded to somehting else (if the edges are coincident to another shell - not that it guarantees anything - with bonded as the default). The basic rule is you have to manually apply a bonded contact to connect a shell to anything other than maybe another shell. Even then it is no guarantee.