3 Replies Latest reply on Sep 25, 2013 3:47 PM by Jared Conway

    Beam elements corrupt behaviour

    George Bon

      Hi all,

       

      I am doing some static stress analysis on ducts involving shell and beam elements.

       

      Overall I can manage to get good results using workarounds, but there is one time consuming issue that I would like to solve:

       

      After modifying the model slightly as part of my design iterations (e.g. some member sizes, load, etc, ) and re-running the analysis, the model will solve but in some parts of the model the beam elements will become much stiffer (it seems) to the point where there is no deformation in them. Typically if I am analyzing a square duct, one side will behave fine and the other will not (see example below).

       

      The only way I found to keep working on the model is to create a new study and copy all the simulation features (bodies, joint group, connections, fixtures, external loads, and even mesh using the drag and drop method). Solving this new study without any modification works fine and gives correct results.

       

      Therefore there seems to be a corruption of the simulation file (which happens all the time, I have to follow this 'new study/copy paste' process dozens of time for each analysis, which is causing the analysis to take 10 times the time it should).

       

      At some point I thought this was caused by using mirror features in the model but removing them does not make a difference.

       

      Example:

      model.png

      The main load on that duct is a downward pressure applied on the underside to represent dust filling.

       

      Here are shots of the problematic results:

      The left side behaves OK and deforms as expected:

      2.png

      The right side seems to have stiffened and hardly moves, and stresses on the ride side are 10 to 50 times lower than the left side:

      3.png

      Creating a new study and copying all simulation features solves the issue:

      4.png

       

      Please note I do not use the symmetry even though I could for other reasons.

       

      Any ideas appreciated!

       

      Thanks

        • Re: Beam elements corrupt behaviour
          Jared Conway

          pretty cool model.

           

          not sure what version of solidworks you're running but have you tried showing the beams (render beams) to see if they are the sizes you expect them to be?

           

          overall, what you're describing, i have never seen. have you spoken with your reseller about this? i think what they will need (and if you want more help on this through the forum as well) is the assembly and simulation results in the "bad" state and also in the good state. more information about the workflow in between iterations would also be helpful. what specifically are you changing, is the area that isn't working being affected by those changes..etc.

            • Re: Beam elements corrupt behaviour
              George Bon

              Thanks for your reply Jared. To elaborate:

               

              • I am running 2013.
              • Rendering beams does not work for me (less than a 10th of them show up, but I will look into that.
              • I believe any change involving a remesh (wether in the problematic area or not) will cause a problem. For example, just changing the value of a load is fine, but modifying a mesh control value is not.
              • Right after posting this I found that switching back and forth between solvers (FFEPlus and Direct sparse) for each consecutive run makes it work. This is much less time consuming that creating a new analysis every time. I'm making progress...
              • I am going to see what can be done through the reseller.