3 Replies Latest reply on Apr 7, 2011 12:32 PM by Kevin De Smet

    Weldments and FEA

    Michael Feeney

      Hello All,

       

      I am running natural frequency analysis on a tube frame structure for my thesis. I have been using ABAQUS, but I wanted to compare the results with SolidWorks (not to mention SolidWorks models, in general, look prettier). I was curious if anyone had input on the following:

       

      Here is one of the ABAQUS models:

       

      1.png2.png

      The figure on the top is the whole tripod viewed in the undeformed state. The simulation has been run and you can see the frequency in the bottom text to be 214.38 Hz. The bottom figure is a close up of the joining members.

       

      Here is my SolidWorks model:

      3.png4.png

      5.png

      The top figure is the whole model and the bottom figure is a zoomed view of the node weldment connection. As can be seen the frequencies from ABAQUS and SolidWorks are roughly the same.

       

      In ABAQUS, the user simple draws a wire frame and then applies a beam profile to the frame members. Apparently ABAQUS allows material interference. In Solidworks, I used the weldment feature, but I had to modify the design slightly. Notice at the ends of the tripod legs they don't come to a single point, but rather 4 spaced points. I had to do this because the weldment feature didn't allow that type of geometry. Furthermore, at the upper section area where I zoomed in, that's how the weldment treated the geometry. I'm curious how I can explain this in my paper. Is this just a visual error and both programs are treating the members the same way? How do these end connections effect the system?

       

      Also here is the physical model I made from brass tubing and solder. I have to validate software with experimental results of course!

       

      IMG_1019.JPG

        • Re: Weldments and FEA
          Dustin Biber

          In SolidWorks you should still have a single node point in your sketch that all the members connect to.  There will be some trimming operations that will occur to get the proper fit, but your sketch elements should all connect.  Once that occurs, the analysis should run very similar to your description of abacus.  Post an image of your sketch.

           

          The FEA in either system won't be super accurate as far as the stresses go in the joints.  Deflection as a whole should be pretty good. Your frequency analysis is an offshoot of deflection, so it should be accurate too.  If you need better stress results in the joints, you will need to model just the joint and analyze it (welds and everything... as a solid, not beams).

            • Re: Weldments and FEA
              Michael Feeney

              I'm only concerned about Natural Frequency, not stresses. My sketch is below.

              6.png

              Notice that the upper section vertices go to a single point but the ends of the span members go to 4 evenly spaced points. I did this because SolidWorks couldnt fit the geometry at that steep of an angle. I guess SolidWorks just ignores what the weldment joint looks like and solves it as beam elements attached at a single node (assuming the drawing converges to a single node) **when doing natural frequency analysis**.

                • Re: Weldments and FEA
                  Kevin De Smet

                  Make sure to check your joint groups in the simulation feature tree, you can adjust the distance-apart joints need to be in order to get merged. For your model you will want to set this value quite low so it doesn't bridge the distance and merge the nodes together.