7 Replies Latest reply on Mar 20, 2017 8:35 AM by John Willett

    Beam Cross-Section Dimensions -- Meaning for Simulations?

    John Willett

      1) Can a beam or a shell me made rigid for purposes of simulation?

      2) Does the beam cross section have any significance in a simulation beyond determining its stiffness and mass?

      3) Can I bond the beam joint to any size area on a shell, or must it be at least the projection of the beam cross section?

      4) What is the function of the "for construction" diameter in a "pipe" beam?

       

      Here are more details:  I needed a pivot for planar triangle, modeled as a shell, so I was kindly advised to use a beam with a "hinge" joint at one end and to bond that hinge to the appropriate area on the shell, defined by a split line.  Then I wanted to make the beam stiff enough without too large a cross section relative to my triangle -- apparently SW Premium (linear static simulation) doesn't allow a beam to be made rigid.  (I don't' even know if the beam cross section matters in a simulation except to determine the stiffness; but I needed to define the split line on the triangle somehow, both for bonding to the joint and for specification of a contact set to get accurate free-body forces/moments, so I used the beam cross section.)  In defining the "Weldments/Structural Member" for the beam, I chose "pipe" for circular symmetry and was offered only three choices, all to large.  The main puzzle here is that there are three diameters in a pipe cross section, OD, ID, and a smaller "for construction" circle of unknown function.  Can I edit all three of these diameters any way I want without disturbing the function of the beam (except for its stiffness and mass, of course)?

        • Re: Beam Cross-Section Dimensions -- Meaning for Simulations?
          Bill McEachern

          Just extrude a beam of the size and length you want - you don't have to use the structural member command. In the tree you can RMB on the body and pick "treat as beam". You might get a warning about the beam length to diameter ratio - just ignore it. I would make the beam about the size of the area you want to bond it to - you could even make it triangular. Just make it steel. Then you need another beam to connect it to and at the joint of the two beams you can get the joint you want (i.e. translations matched and rotations free). This of course would be tremendously simple if they just offered couplings. Couplings would be so handy for so many things.......

            • Re: Beam Cross-Section Dimensions -- Meaning for Simulations?
              John Willett

              Bill -- Thanks!  I didn't know you could avoid the "Weldment" approach.  Your method sounds simpler, but...

               

              With the approach I've taken (Insert/Weldments/Structural Member/Pipe/[chosen size]), I only need one beam -- essential in my application because the "pivot" must be right at the triangle face.  In the resulting Parts/Cut List/Cut-List-Item[X]/[Name of beam] the simulation study, RMB/Edit Definition... gives me the opportunity to define the "joint" at each end of the beam as "Rigid," "Hinge" (effectively translations matched and rotations free as you said), "Slide," or "Manual."  After that, the selected joint can be "Bonded" (not sure if the other contact types work) directly to a face of a shell or other type of body in a Contact Set.  I get my pivot where I want it with no second beam!  (This recipe came from Ryan Dark at Simplified Simulation of a Pivot (or Ball Joint) Between Parts?.)

               

              I haven't checked so far whether all of this mechanism is available with your approach, but in any case I still don't really understand how the beam cross section is interpreted by Simulation in terms of contact areas, nor how to legitimately modify that cross section for a "Structural Member." -- John Willett

                • Re: Beam Cross-Section Dimensions -- Meaning for Simulations?
                  Bill McEachern

                  HI John,

                  I would make up a little test model and see if you can release the degrees of freedom at an end with a bonded connection. I took a look at the other post ....seems like you can. The beam cross section ends up as numbers (2nd moments of area: Ixx, Iyy, & Ixy & the area) in the beam definition. Sometimes, at least in the past, it would get confused on what direction was the axis when the beams are really short. You can also load a large selection of beam cross sections into the structural member command from the libraries. You can define any cross section and add it into the selection list. I am sure you can you tube it to find out how. I don't think I really understand what the subtleties are of the problem.

                  What the program does, as far as I know, is project the beam cross section onto the face to be bonded. It then puts a bonded connection between the joint and the projected area of the beam on the face. It is key that projected beam cross section fits in the area (I think - do a test if you want it bigger). They do this to match the stiffness of the connection to the beam dimensions - otherwise your connection can be much to stiff. Does that help?

                    • Re: Beam Cross-Section Dimensions -- Meaning for Simulations?
                      John Willett

                      >>What the program does, as far as I know, is project the beam cross section onto the face to be bonded. It then puts a bonded connection between the joint and the projected area of the beam on the face.It is key that projected beam cross section fits in the area...<<

                       

                      Bill -- Thanks again.  I think you're zeroed in on my main question.

                      Stress around Beam.png

                      The graphic above shows an example of stress distribution in a planar shell around the contact point with a "hinged" beam oriented normal to the shell.  The three circles delineate the beam cross section that I described above for my (edited) pipe "structural member."  Notice that the stress jumps up exactly at the ID of the pipe, might or might not have another discontinuity at the "for construction" circle, and appears not to notice the OD, which is also my split line for the contact set.

                       

                      Tentative conclusion is that the beam applies all its force (perhaps also its torque, except that this one is hinged) exactly at the pipe ID and ignores the rest of the cross-section dimensions.  Make any sense? -- John WIllett

                      • Re: Beam Cross-Section Dimensions -- Meaning for Simulations?
                        John Willett

                        >>What the program does, as far as I know, is project the beam cross section onto the face to be bonded. It then puts a bonded connection between the joint and the projected area of the beam on the face. It is key that projected beam cross section fits in the area (I think - do a test if you want it bigger). They do this to match the stiffness of the connection to the beam dimensions - otherwise your connection can be much to stiff. Does that help?<<

                         

                         

                        This helps a lot, now that have a bit more information through my VAR.  As you may recall, I was placing a split line at the outer edge of the beam (actually converting the sketch defining the beam cross section into a new sketch in which the OD defines the split line) in order to define both a mesh control and a contact set (to improve accuracy of free-body forces/moments).  This appears to have caused free-body forces on some joints like this to be wrong while most were right (hence my question to the VAR), and it may also have been the problem in the peculiar image above.

                         

                        I was advised to bond to a larger area than the beam cross section, so I expanded the split line for both contact set and mesh control slightly beyond the OD of the beam.  Now I'm getting both correct free-body forces and a more reasonable stress distribution that seems to "see" both the ID and the OD of the beam.

                         

                        Here are stress distributions for the two cases that compare apples to apples.  First is with the beam OD defining the split line.  The beam cross section is highlighted orange:

                        Split Line Coincides with Pipe Hinged.png

                        Clearly this makes no sense, as the stress should not be concentrated just outside the ID of the pipe.  Next is a similar image with the larger split line, highlighted in orange but the with beam cross section still visible:

                        Larger Split Line Hinged.png

                        This looks much more reasonable, to me at least, as it shows effects of both the ID and the OD of the beam.

                         

                        Leaving aside the expected inaccuracy of the stresses shown, do you agree that this is the way I was supposed to do it in the first place? -- John Willett