6 Replies Latest reply on Feb 17, 2012 3:31 PM by Adrian Tayne

    Bonded surfaces are moving with respect to one another

    Adrian Tayne

      I am attempting to analyze a frame structure that is comprised of C-channel beams that are bolted together along each side using solid blocks while corners are welded together with gusseting. To run the simulation, I excluded the fillet welds attaching the gussets to beams and manually created bonded contact sets between them.

       

      My understanding is that when you apply a bonded contact between two surfaces, the surfaces should not shear w.r.t. one another. However, when I run the simulation and create an exaggerated displacement plot, the gussets are shown completely shearing off of one of the beams that they are bonded to. All of the contact sets between gusset and beam surfaces have been created manually. I attempted to add edge weld contact sets but since these aren't sheet metal parts, I get an error. I also tried defining the bond contact between gusset and the two beam surfaces as one contact (gusset-beam and gusset-other beam) set as well as two separate contact sets (beam-gusset-beam). Both gave the same result.

       

      This simulation was run with only a 1 lb load just to test the setup, so displacements are very small.

       

      Any insights are appreciated. Thanks

       

       

       

      1.jpg

      2.jpg

      3.jpg

        • Re: Bonded surfaces are moving with respect to one another
          Adrian Tayne

          Could it be a case of too few solid elements in those gussets causing a meshing issue between it and the beam? I could really use some help on this.

          • Re: Bonded surfaces are moving with respect to one another
            Paul Kellner

            Is it possible that even though you told them to bond, they didn't?

             

            Is there a gap between the gusset and the C-section?

             

            Is there a split line on the gusset and C-section where they join to help the mesher figure out what should be bonded?

              • Re: Bonded surfaces are moving with respect to one another
                David Maxham

                I think Paul is on to something here, especially since some of the gussets look like they are bonded correctly.  I understand that you're using solid parts, but I'm wondering if SW is getting confused on some bonds - maybe if you re-did the problem ones?  For example with a shell to solid contact, at least as described in my 2009 training manual, page 338, "a shell entity (face or edge) must be Set 1, while faces on solid components must be Set 2."  Probably doesn't apply here, but maybe there's something similar going on?

              • Re: Bonded surfaces are moving with respect to one another
                Adrian Tayne

                Paul, thanks for the response. The parts have coincident faces and there is no split line on either part. I haven't used split lines to date and am not totally sure when they are necessary because after I get all of my restraints and contacts established, I haven't had any real inexplicable issues except for this one.

                 

                David, The way I set up the contact was using the face of the gusset as set 1 and the faces of the beams as set 2. I'm pretty sure I tried setting them up as completely separate contact sets as well thinking I was confusing it by using multiple parts in one contact set. I'm still not really sure if I can do this or not.

                 

                I was working on the frame alone to simplify my problem so I could learn assembly simulation without getting too bogged down in contact sets. I actually popped the frame into the case that it was meant for and setup the simulation again and this issue did not arrise. That time I did the contact sets separately and they seemed to stay bonded.

                 

                Thanks for the help guys.

                  • Re: Bonded surfaces are moving with respect to one another
                    Samuel Haworth

                    Bonded contact is all screwed up. Here are the rules I've worked out:

                     

                    Free body forces will never report correctly is you use global bonded contact.

                     

                    Manually defined contact is always incompatible mesh, which adds stiffness to the connection.

                     

                    Using the "more accurate" incompatible mesh option in the study properties will add loads of time to the solve, and often fails to bond the surfaces together.

                     

                    Also, copying a mesh that has "more accurate" bonding from one study to another will not work. The mesh has to be recreated in the new study.

                     

                    "simplified" incompatible bonded contact actually works in all the cases I've tried. You can copy the mesh, you can recover free-body forces, bonded faces stay bonded.

                     

                    So, in conclusion: use global bonded compatible mesh when you don't need free body results, use simplified bonded if you do need free body forces.