10 Replies Latest reply on Nov 3, 2013 1:54 PM by Jared Conway

    Non-linear: Hyperelastic-Mooney-Rivlin Analysis

    Christopher Hardy

      Hi,

       

      I recently did some compression tests on a rubber sample. From this I was able to attain experimental data for Solidworks to be used in the Hyperelastic Mooney-Rivlin Model.

       

      Using the "Simple Tension" graph, ie Stretch Ratio (new length/original length) in relation to Stress (Force over original area). I was able to specify non-linear material properties to my rubber sample.

       

      I have a problem when I try to model various runs of the experiment (same experimental testing/scenario, slightly different results). Some of the models are able to solve 100% according to its full time duration, others fail within a fraction of the duration. (From this, I know that the model is working correctly)

       

      Does anyone have any ideas to why this might be occuring ? It seems logical that it all works right, considering the numbers might change ever so slightly witin like few % of each other

       

      Is it some bad data in the results that can throw this analysis off, e.g like repeated stretch ratio values with different stress components ?  Or is it more an analysis factor in study properties?

        • Re: Non-linear: Hyperelastic-Mooney-Rivlin Analysis
          Jared Conway

          cross post https://forum.solidworks.com/message/388915#388915

           

          elaborating on what you changed between the studies would be helpful for troubleshooting

            • Re: Non-linear: Hyperelastic-Mooney-Rivlin Analysis
              Christopher Hardy

              I changed the stretch-stress graph associated with each hyperelastic-mooney-rivlin analysis. I also changed the time duration of the analysis to meet the data. (nothing else)

               

              pretty much just used another set of experimental data, but all data follow the same trend and only vary slightly in results. Just confused as to why a premature error message shows up.

               

              "displacement grow to large"

                • Re: Non-linear: Hyperelastic-Mooney-Rivlin Analysis
                  Jared Conway

                  what is the exact error message.

                   

                  if you set up a simple problem (cantilever beam), does the changed curve work? changing the curve could make a huge difference if you're using the same loading condition. have you tried reducing the load?

                    • Re: Non-linear: Hyperelastic-Mooney-Rivlin Analysis
                      Jared Conway

                      note, reducing the load is for troubleshooting to determine if it is your curve, the load, your setup or the geometry that are causing issues.

                      • Re: Non-linear: Hyperelastic-Mooney-Rivlin Analysis
                        Christopher Hardy

                        I am actually using a prescribed displacement through "Reference Geometry" fixture. That is how I am loading my model.

                        I use varying displacement-time graphs to change my loading scenario (from experiment) and have corresponding stretch-stress curves to match (also from experiment). From this I would like to analyse the stresses that result of this deformation rate.

                         

                        I am not able to compare the loading in a cantilever beam scenario, as I would have no real data or expectations of what would happen in such a case. The material is custom fabricated.

                         

                        Just really curious as to why the program is struggling to model this particular analysis, if it works previously for a very similar case? (just slower loading condition) Is it because of the steep increase in Force, and large deformation over short period of time?

                         

                        Error message is shown below.

                         

                        Thanks for the help Jared

                         

                        Solidworks Error.jpg

                          • Re: Non-linear: Hyperelastic-Mooney-Rivlin Analysis
                            Jared Conway

                            Not clear on what you are changing now. My understanding was that you changed the material properties. Ie the curves.

                             

                            Now you're saying that the only change is the time curve?

                             

                            What exactly are you changing that is causing the failure?

                             

                            Also in your post you say you entered a compression curve. That isn't a recommended input. Do you have the ones suggested by the kb article?

                             

                            And for the cantilever beam, it can really be anything. A cube in compression is fine. This is for troubleshooting to determine if it is a software, model or setup issue. Even if you don't have absolute answers I expect that you know the relative effect of the changes and how the model should behave based on them. This test model also gives you something that you can post that is easier to look at.

                             

                            The other option here is to start over from the beginning and explain your application, what your inputs are, what your expected results are and what you're getting.

                              • Re: Non-linear: Hyperelastic-Mooney-Rivlin Analysis
                                Christopher Hardy

                                This is under all of the same anlysis setup.

                                • I changed both the SS curve and the time curve to replicate the experiment conducted.

                                 

                                • This has caused a failure in the results.

                                 

                                • When I do this with another matching pair of SS and time curve, the analysis is able to solve.

                                 

                                Therefore, my question is that, why does this particular SS curve and matching time curve fail to solve ?

                                 

                                 

                                Example:

                                Case 1 - SS curve(1) + time curve(1) + setup (1) = solved

                                 

                                Case 2 - SS curve(2) + time curve(2) + setup (1) = unsolved

                                  • Re: Non-linear: Hyperelastic-Mooney-Rivlin Analysis
                                    Jared Conway

                                    Nonlienar is pseudo time. The actual length of time is irrelevant.

                                     

                                    So by changing it, you might be affecting the way the problem solves. But also it could just be the second curve causes numerical difficulties. Set them back to the standard 1s. Then follow the instructions in the error messages to try and get the solve to go through. There are some other articles about that in the kb.

                                     

                                    If you are still having trouble, check the kb article I referenced and the suggestions it makes. Test the simple part.

                                     

                                    At that point, I would contact your reseller and have them check with the developers that the compression curve is supported and check if they have any other suggestions for achieving convergence.

                                      • Re: Non-linear: Hyperelastic-Mooney-Rivlin Analysis
                                        Christopher Hardy

                                        By using this time profile, I thought I was controlling the displacement with respect to time (even if pseudo)? If i use the single 1s, wouldn't that mean i just be splitting the event over 1s of pseudo time (same thing)?

                                         

                                        Unless what you are saying is that increased pseudo time might have implications. is that correct?

                                          • Re: Non-linear: Hyperelastic-Mooney-Rivlin Analysis
                                            Jared Conway

                                            If you are using the auto time step it isn't as important but the total time curve could effect the initial guess and cause problems. By using the same time curve, you eliminate that variable. If the problem still happens, you know it is the material curve assuming the magnitude of the load is the same. Then you're back to the validity of the curve and modifying the nonlinear properties as per the kb.

                                             

                                             

                                            For reference, the time curve in nonlinear is more for the solver than for you unless you are sequencing events, using creep or viscoelasticity.