11 Replies Latest reply on Jun 14, 2018 9:49 AM by Craig Schiller

    Spring in bending condition?

    Bert Bourgeois

      Hi all,

       

      I am looking at analysing what we call at my work a 'spring mount'

       

      1.JPG

       

      Imagine the bottom plate is fixed, while some force is applied on that split strip on the tube part.

       

      the spring you see are screws directly on some threaded boss.

       

      This is how I tried to simulate the assembly:

      2.JPG

      (the fixtures at the end of the pipe are 'sliders', fixture on the bottom plate is 'fixed'))

       

      but even for small force (20N), the assembly wants to solve in large displacement (and fails), or give me massive stresses (+1000Mpa) and displacements. that's for a Static analysis. I tried a non-linear static one, but after 2hrs and no results I gave up.

       

      What did I do wrong? is Simulation on able to analyse a spring in such condition?

        • Re: Spring in bending condition?
          Janko Stellaard

          Hi Bert,

           

          What values did you use for the stiffness of the springs?

           

          Janko Stellaard

          Cadmes BV

          The Netherlands

          • Re: Spring in bending condition?
            Shaun Densberger

            "but even for small force (20N), the assembly wants to solve in large displacement (and fails), or give me massive stresses (+1000Mpa) and displacements. that's for a Static analysis. I tried a non-linear static one, but after 2hrs and no results I gave up."

             

            Well, have you done a hand-calculation to approximate the displacement you'll see? It could easily be that you do in fact need large-displacement formulation, which is a nonlinear problem (i.e. requires a nonlinear solution process). Applying small strain-displacement formulation to a problem that results in large strain-displacement can easily show erroneous results in the stresses and displacements.

              • Re: Spring in bending condition?
                Bert Bourgeois

                I cant say I have done any calcs yet, as I'd need to know a bit more data about the spring first.

                 

                for such a small force and 2x quite stiff springs, I would imagine that assembly would not move by much tho.

                  • Re: Spring in bending condition?
                    Shaun Densberger

                    From a finite element method perspective, how much is too much displacement is relative. What are the dimensions of your parts? Let's assume that you're trying to lift the tube up with 20 N of force and that your springs are 3135 N/m each. Under these assumptions, your total displacement would be ~3.2 mm, which could be "large" depending on the dimensions of your geometry.

                     

                    It's also possible that the spring values you entered in the model are not representative of the real springs.

                      • Re: Spring in bending condition?
                        Bert Bourgeois

                        for example the tube is 170mm long and Ø37mm.

                         

                        i tried to up the spring value by 1000x and it still wanted to use large displacement method.

                         

                        Maybe the spring isn't enough the restrain correctly the pipe part. i could try to build fake lug at each end, with a hole in it so i can applied a fix hinge fixture, so i can simulate the movement that assembly would have in real life.

                          • Re: Spring in bending condition?
                            Shaun Densberger

                            I don't have access to SW simulation, so I can't t confirm this, but here is what I think is happening:

                             

                             

                            It looks as if you're trying to connect the spring elements to solid elements and I'm not sure if SW will resolve the nodal incompatibility (spring elements had 6 dof per node, solid elements only have 3). If this nodal incompatibility isn't resolved, then you'll have rigid body motion of the spring elements, which might cause the solver to flag a LD warning (instead of singular stiffness matrix error). Another possibility is that in the SW help (2013 SOLIDWORKS Help - Connector - Spring) it  specifically states that the tangential stiffness is the, "(s)tiffness in the plane of the faces or tangential to edges of shells." Their specific use of shells may be because they don't resolve the issue in #1.

                             

                            What you can try doing is creating a beam element that was a coil like your spring, and play around with the Young's Modulus and Poisson's ratio until you have an equivalent stiffness response as that given by the manufacture. You'll need to make sure that SW can connect beam elements to solid elements (like springs, beams have 6 dof). While a beam element is more computationally intensive than a spring element (the former has quantities like stress calculated) it should add a very negligible amount to your solve time.

                    • Re: Spring in bending condition?
                      Jet Lim

                      Hi Bert,

                       

                      Is it 20N in total or 20N for each location? In the settings is the option to set total or per location. If its per location the total force will then be 100N which could be large in this case

                      • Re: Spring in bending condition?
                        Craig Schiller

                        +Bert Bourgeois It's been over 2 years, so this may not help the original poster, but I have frequently seen the described behavior when the model is slightly under-constrained. It's tempting to think that applying a couple of "No Penetration" Contact Sets in a default "bonded" setup (in addition to appropriate Fixtures) should constrain things well enough, but for geometries such as that shown, one often needs:

                         

                        • a Pin Connector (or two)
                        • or a combination of "On Cylindrical Faces" no radial translation/no rotation.

                         

                        It's always worth trying a little more constraint to see if it works. Remember, when you look at the geometry, certain things seem "obvious," but you have to explicitly define conditions for the solver. The "On Cylindrical Faces" no radial translation/no rotation constraint pair recently got my model to solve after repeated "Large Displacement" warnings. My geometry was a simple clamp plate acting on a captured bobbin in a counterbore. I was using several Screw Connectors already, but the solver doesn't interpret the force load from the screw as preventing radial motion. I imagine a spring system would be similar.