11 Replies Latest reply on Feb 22, 2018 8:16 PM by Ryan Dark

    Stretching silicone rubber

    Trevor Buckner

      I am trying to run a simulation on some elastomers that we are inflating with air.

       

       

      I am running a Nonlinear Study with 2D simplification, and I want to see the stresses at specific points along the geometry, but after reaching about 30% of the solution, I get the classic error:

      Looking around the web it seems like using Arc Length control method should help me over the buckling stages, but it still seems to fail, in fact even sooner at about 25%. I am not sure where to go from here, and there seem to be scant resources on how the Arc Length method actually works and what each of the settings mean. Any suggestions?

        • Re: Stretching silicone rubber
          Ryan Dark

          Hi Trevor,

          My first thought was to use Arc length control... until you said it didn't work.  Are you able to save your model/simulation with us so we can look through it?

            • Re: Stretching silicone rubber
              Trevor Buckner

              Sure. Which files do I need to send? Just the SLDPRT file? Solidworks generates so many pieces as soon as I run a study I'm not sure which ones actually hold the simulation data.

               

              I got it to run a little farther I think? I noticed Arc-Length counts its percent completion based on the max number of steps you give it, rather than how close it actually gets to your target forces. So if I tell it to only run 50 steps it will complete, but obviously that's not the full solution; if I set the max to 200 steps, it keeps running to about 160 before failure, and the stresses are noticeably higher, but since it still didn't finish "gracefully" I assume it didn't converge.  Is there a way to tell in Arc-Length mode how close to a real solution we've gotten? This is critical, because I've noticed that as I increase the pressure or change the shape, the problem becomes more more difficult to solve so that by 50 steps, the higher-pressure analysis actually has less deformation simply because it couldn't get as far into the calculation in just 50 steps.

                • Re: Stretching silicone rubber
                  Ryan Dark

                  Hi Trevor,

                  By the time you get to nonlinear analysis "ending gracefully" at 100% solver completion is not longer what you are after.  By the time you are at a point where you need to be using Arc Length Control usually you are making a study setup that will stress your model to obliteration on purpose.  You expect that the solver will not run to 100% and welcome that fact.  This isn't a bad thing; it just means that you need to query your reaction forces and make sure you understand what force is causing the theoretical failure.  If you do convergence testing (reducing mesh element size with the same load/fixture boundaries) and get the same failure point you have your answer.  If you only make it to 26.7% but do so consistently you are headed a good direction.

                   

                  It sounds like you are on the right path.  I looked through your model and all the studies are setup with "Linear Static Isotropic" material models; you will definitely need to get fully switched over to Mooney-Rivlin material properties before you go any further.

                    • Re: Stretching silicone rubber
                      Trevor Buckner

                      Thanks for your response! I can see where in most cases you might want to see where the buckling point is and stop there. That makes sense if you are trying to avoid catastrophic failure of your part.

                       

                      However, for this particular highly-stretchable elastomer object, buckling is not failure. We have selected this shape with the intention of causing buckling and snap-through effects, and need to see where the shape ends up after that point. If we can only see what happens before the point of buckling, then we gain nothing from the analysis. Is there any way to force it to move past the buckling and continue to see what happens to the shape? I thought that was what Arc-Length control was designed to handle, but despite my best efforts it still just gets right to buckling and no further. Any thoughts? We have also tried with Mooney-Rivlin using some old coefficients someone gave us a while ago, but it just causes the solver to crash on the first step.

                        • Re: Stretching silicone rubber
                          Ryan Dark

                          Trevor,

                          The best advice I could give to you is to not take short-cuts.  My thoughts keep coming back to, "why isn't he sticking with Mooney-Rivlin"?  The results you get from Linear Elastic Isotropic will not be close to reality so you may struggle (and succeed) in getting the snap-through effect to solve but your results will be bunk.  It would be better to get as much of your setup as close to real life as possible then struggle with your solver.

                            • Re: Stretching silicone rubber
                              Trevor Buckner

                              I think you may have missed the last line of my most recent comment:

                              We have also tried with Mooney-Rivlin using some old coefficients someone gave us a while ago, but it just causes the solver to crash on the first step.

                              Seeing that this didn't work, I just now looked up some other silicone rubber coeffecients for Mooney-Rivlin for a material that is similar to what we are using (we have tested this other material in a similar application and it does the snap-through just fine). The solver doesn't crash on the first step anymore, but it still gets no farther than just using linear elastic. The results are almost identical at the point where it fails; right at the buckle point. I think I can now say that our Mooney-Rivlin setup is roughly equivalent to our real life system. So now, having tried Mooney-Rivlin and seen no improvement, I'm starting to think Solidworks just isn't designed to handle this kind of snap-through movement (although I swear I saw an example somewhere that used Arc-Length for a snap-through effect and it actually worked.. can't seem to find it now...).

                               

                              Anyway, thank you for all of your help so far! With Mooney-Rivlin set up, I believe I am at the stage where my setup is "close to reality" and need to keep "struggling with the solver". Unfortunately I don't know enough about the inner workings of the solver to do that last part successfully.

                                • Re: Stretching silicone rubber
                                  Ryan Dark

                                  Trevor,

                                  I didn't miss your last line.  It is just that important that you get a good material model applied to your setup before you move on.  If you have updated the material model with good Mooney-Rivlin data then upload an updated model and lets see what else has to be done.

                                    • Re: Stretching silicone rubber
                                      Trevor Buckner

                                      Alright, thank you much! Here's the updated setup with Mooney-Rivlin.  Assuming the "Load Factor" that Arc-Length control displays is related to how close we are to the desired load (1 being the load is fully-applied), I guess I am actually only getting 3.95% completion.

                                        • Re: Stretching silicone rubber
                                          Ryan Dark

                                          Trevor,

                                          I worked with your latest model for a bit and am only able to get it to deflect up to 9.4 mm before the solver fails to buckling.  This was having loosened both the "Convergence tolerance" and "Maximum incremental strain" to 0.1 (not bueno).  I don't know if you are going to get the kind of behavior you desire from Simulation but I am bumping your thread in case anyone else wants a crack at it.

                                           

                                          Stress:

                                          Displacement:

                          • Re: Stretching silicone rubber
                            Bill McEachern

                            could be your material model. What is it and how were the coefficients determined? If by test curves which ones and how many? What are the strain levels in the piece on the last step completed?