8 Replies Latest reply on Jul 18, 2018 3:12 AM by Peter Dratwa

    Simulation of Electrical Contact Spring

    Peter Dratwa

      Hello there!!

      I am designing a electrical contact spring. And also comparing it with actual result from testing Setup. From simulation i am getting Resultant force as 1.4 N and from actual testing of same Contact spring with measuring device i got the 0.8 N. Can anyone explain why is the difference so much ?

      I know there is some tolerance between actual geometry and input geometry in solidworks, because of manufacturing tolerance. But i measure it from optical device which gives you accurate result. Also friction between punch and contact is difference (In simulation and in actual case). Can anyone  help me out with it? how much difference should i expect ? Or how can i improve my model ? I am using stress strauin curve as input material model for Von Mises Plasticity. Which i measure from tension test experiments.

      Thanks in advance.

        • Re: Simulation of Electrical Contact Spring
          Ryan Dark

          Hi Peter,

          Have you tested your results for convergence?  That is to say, have you duplicated your study and run the duplicate study with a smaller mesh element size to see if the 1.4 N result is maintained with smaller elements?

           

          If you have already done that I might expect that there is a boundary/assumption in the FEA setup that causes it to deviate from the real life test results.  As I often say to people in the forums, is this a model you can attach to your post so we can look through it?

            • Re: Simulation of Electrical Contact Spring
              Peter Dratwa

              Hi Ryan,

              Thanks for the reply. What do you mean by convergence? Do you mean convergence tolerance ? I am just comparing the simulation result with real life result. Here is the model. I am pressing the spring with top plate , while the bottom of the spring is fixed. The material model i am using is Plasticity von Mises. Load type is prescribed displacement with loading and unloading.

                • Re: Simulation of Electrical Contact Spring
                  Ryan Dark

                  Peter,

                  Convergence testing is where you compare FEA results to other FEA results to determine how the results are trending.  This allows you to make a determination whether the results are internally consistent.  This is typically done by running several successive studies with the same boundaries but with different mesh element sizes.  I might expect the results to change as the element sizes change but there will become a point where reducing the mesh element size doesn't change the results.  That is where the setup would be considered converged.

                   

                  Also, looking at your model, have you created the mesh such that there are at least two elements across the thickness of the thin wall?

              • Re: Simulation of Electrical Contact Spring
                Bill McEachern

                I would say friction is a big deal here and misses in friction will translate to errors in the computed force assuming sliding actually occurs which may not. I would suggest bounding the problem by running the analysis with zero friction and by bonding the contact surface and assessing the response. Ideally you actual answer is between the two values produced. It would be more helpful if you could provide the amount of displacement that is being reported. Force numbers in an NL analysis are unlikely to be strongly affected by discretization convergence issues unlike say stress values.

                  • Re: Simulation of Electrical Contact Spring
                    Peter Dratwa

                    Hi Bill,

                    Thanks for the reply. In this type of spring, there is always little bit of sliding movement on the top (Because Bottom of the spring is fixed). In compression test on testing machine also I can see some sliding displacement of top surface of spring and plate. So I cant nullified the friction. But yes I will try with 0 friction coefficient and Lets see what comes out. I am pressing plate just 1 mm and time i have taken is 5 sec( loading and unloading).

                    Is the input time also plays big role?

                      • Re: Simulation of Electrical Contact Spring
                        Bill McEachern

                        Are you running a NL dynamic study? If so, then I would use the time it actually takes on the test machine. If you are running a static study the time values have no impact what so ever. The friction models in SWX sim are not the most sophisticated ones out there - not by a long shot. What does you time versus force plot look like?

                          • Re: Simulation of Electrical Contact Spring
                            Peter Dratwa

                            I am running Nonlinear Static study. Here I dont have consider friction , like you have said. so difference is just 0.1N between with and without friction. Time required is really less for without friction. So this mean friction is not that important ,when it comes to solving time for simulation? but in actual case there is friction for sure. In graph ,resultant force  is on plate.