8 Replies Latest reply on Jan 23, 2014 5:08 PM by Aaron Gradeen

    Large Displacement Issue

    Aaron Gradeen

      I'm analyzing the bending stress on a 40 mm pin in a pin/clevis assembly. The load on the pin is applied through a spherical plain bearing. I have inserted an image of the assembly below (bearing is not shown, those are just spacers on the pin which locate the bearing. I have also inserted an image of the assembly cut in half for symmetry in the simulation study. Here you can see the inner race of the spherical plain bearing is modelled. I am applying a force on the flat which I have created. Typically I use flats on cylindrical parts that are not rotationally constrained since this seems to stabilize them from spinning infinitely in the model. I have "no penetration" contacts setup between the bearing and the pin as well as between the clevis bore and the pin. A global friction value of 0.33 is applied. Symmetry is applied to the three cut faces on the clevis, pin and bearing. The face of the clevis which is parallel to the plane of symmetry is set as fixed.

      Pin Clevis Assembly.JPGPin Clevis Simulation.JPG

      Even with the soft spring effect enabled I am getting large displacements.  I believe the bearing is spinning and expanding a large amount under the infinite centripetal force. I don't want to constrain this bearing more than I need to since overconstraining a model adds unrealistic stresses. The maximum bending stress in the pin is very important in this case since it is used in a fatigue calculation. I am using a standard high quality mesh (oddly enough it ran fine with draft quality mesh...perhaps something to do with fewer degrees of freedom?)  Any input would be greatly appreciated   

        • Re: Large Displacement Issue
          Jared Conway

          your pictures didn't come through

          can you reattach them

           

          i think your assumptions are correct about the cause of the large displacement warning. my recommendation would be to go back to the beginning of this analysis, start simple with bonded contacts, determine exactly what is causing it, then look at whether your restraints are correct. or if you can add more to stabilize the problem. numerically if something is offbalance, you may have no choice but to add a restraint. sometimes that oculd be as simple as adding friction.

           

          if the large displacement turns out to be true, you may need to go to nonlinear. but first rule out that it isn't because of your restraints.

            • Re: Large Displacement Issue
              Aaron Gradeen

              Hopefully a .png will work then.

              Pin Clevis Assembly.pngPin Clevis Simulation.png

                • Re: Large Displacement Issue
                  Jared Conway

                  what are you trying to learn? how critical is it that you're pulling up at an angle?

                    • Re: Large Displacement Issue
                      Aaron Gradeen

                      I'm interested in the maximum stress on the pin and the clevis. I will use this study along with another loading scenario to run a fatigue study that uses "find cycle peaks". The direction of the force will have a small effect on the clevis stress. It sounds like you're eluding to the fact that pulling up vertically will give me the same stress in the pin. You make a good point. Do you think the direction of the force will make a difference in the stability of the model? I figured it wouldn't matter since it was a radial load. I also just recently tried running the model with a bearing load applied directly to the pin and I got the same result.

                      Pin Bearing Scenario.PNG

                      The displacement plot appears to show that the pin is still rotating. I've applied a bearing force to a pin before and I don't remember having the same issue. Maybe the bearing force is better suited for an internal diameter rather than an external one?

                      Displacement Plot.PNG

                      • Re: Large Displacement Issue
                        Aaron Gradeen

                        Hey Jared,

                          Thanks for the input. I got some help from our tech support and it looks like simply changing the results folder location fixed the issue.  Here's what their response was:

                        "I ran the study "Ultimate Load" with only the settings you had and it completely successfully without any Large Displacement messages.  Soft spring was enabled.Initially I had a STAR.exe error and it failed to solve.  Then I noticed that the hard drive where the results were saving only had about 30GB.  During the solving process, it looks like it peaks higher than that due to the large DOF.I pointed the Results folder of the study to a new folder on a different drive and ran it.  Attached are the screenshots after the successful run.  The Large Displacement option is still disabled.  Displacement plot shows small displacements (0.336mm).I'd recommend pointing your Results folder of the study (in the Ultimate Load study properties) to a new folder on a hard drive with sufficient storage.  This will generate a brand new mesh and results.  It could be that old results were causing issues."

                        I also noticed I was getting two solid bodies in the simulation for the bearing part I was using. You can see in the picture below that one was deforming and one was just floating so maybe that was the issue. Two Bodies.png

                        Current results look much more promising

                        Solver Finishes.png

                        Thanks again for the help

                        -A

                  • Re: Large Displacement Issue
                    Shaun Densberger

                    I think the issue you're running into is due to rigid body motion in your model. Using "soft springs" does prevent rigid body motion, but it will essentially invalidate any displacement results. If I understand your model correctly and what you want to simulate, then I'd suggest changing the loading direction to be purely in the vertical direction, and cutting you model in half again (such that you now have a quarter-model) and applying symmetry constrains to the new cut surfaces.

                    Aaron Gradeen wrote:

                     

                    Even with the soft spring effect enabled I am getting large displacements.  I believe the bearing is spinning and expanding a large amount under the infinite centripetal force.

                    Just for you own edification, the large displacements you're seeing are not due to an infinite centripetal force (if the centripetal acceleration was infinite, then then displacements would be infinite as well). Instead, what is happening is that the "soft spring" option is adding a spring with very low stiffness to the pin to try and prevent rigid body rotation. The amount the pin rotates depends on several different factors (spring stiffness, attachment location, magnitude of unbalanced loads, etc), but the lower the stiffness of the spring, the higher the displacement (i.e. the more the pin rotates).

                     

                    Now, what might have thrown you off is that if you look at a deformed plot of the displacement, you'll see that the pin increased in diameter (and hence you'll think that there is some centripetal load causing the radial expansion). However, the radial expansion that you're seeing is an effect of linear interpolation. When you create a deformed plot, you define a scale factor for the deformation (1 being the true deformation), in which the software linearly scales the deformed shape. What happens for objects undergoing rotation is that the any given point on the rotating object is scaled along a line that is tangent to the original point; the result of this is that the object appears to expand in diameter. As an example, if we take a solid rod and fix on end (like a cantilever beam) and apply a torsional load to the other end, we'll see the an artificial expansion of the free end.

                     

                    Linear Expansion.PNG

                      • Re: Large Displacement Issue
                        Aaron Gradeen

                        Thanks Shaun. Your explanation of how the deformation is linearly interpolated tangentially from the cylindrical surface cleared up what I was seeing. I'm not too concerned about the pin displacement so perhaps using the soft spring option is okay initially. Just for verification though I will try a run with just a quarter of the model and apply symmetry to both faces. I'll run the case without the soft spring enabled as Jared suggested. I'm interested to see the deviations in stress and displacement for the two cases.