7 Replies Latest reply on Dec 4, 2018 2:46 PM by Bill McEachern

    Mooney-Rivlin Testing

    PJ Pisczak

      I have read comments many comments that it is better to test for elastomeric data and place the information into Solidworks® Non-linear, M-R (Mooney-Rivlin), Simulation™.  It is my understanding that this will yield more accurate results.  Also, the M-R constants, C1 & C10, could be generated from this supplied data for future use.  I have also heard that the part should be tested.

       

      This is well and good.  However, we not only use standard round O-rings, but more importantly some complex O-ring cross-sections, which I might describe as a 'spider's web'.  So, not why, but how do I test to get the data to place into Solidworks Simulation.  With metal or plastic, we make dogbone tensile samples.  With elastomers, how do we get this data?  Compression testing of a cube of the material; the actual part?  All thoughts appreciated.

       

      Phil

        • Re: Mooney-Rivlin Testing
          Bill McEachern

          You should look up the test procedures for the various test to characterize a hyper elastic material. The answers are then pretty obvious. Too much work for me to provide a full education here but simple tension test, compression test, biaxial test and shear test might give terms to enable a search that gets you there.

          • Re: Mooney-Rivlin Testing
            PJ Pisczak

            Bill - tx for your response.  I might have had an epiphany, or just bad idea .  I went to the Case Tech (Case Western Reserve University) library this morning to pull some books on Mooney-Rivlin, o-ring gaskets, etc..  When it dawned on me that I could test the o-ring as a function of force vs displacement and then FEA the same part placing the test results in that model.  I would load the model the same way that I had it tested.  From there I could use SW to extract the C1 & C10 and use it for future models of O-rings, presumably using the same cross-sectional diameter and material.  I'm not sure it would work with different cross-sections.  But it seems possible.  What are your thoughts?

             

            Phil

              • Re: Mooney-Rivlin Testing
                Bill McEachern

                The shape doesn't matter. Its the material you need to characterize. For low strain levels the two coefficient mooney rivlin model provides a reasonable estimate and you can get this from a simple tension test - the sample shape is pretty trivial. See the link provided by Bence Rivasz above. You idea seems complicated and a trail and error thing to guess at the right coefficients - might not be impossible but sounds difficult to me but haven't thought about it all that much so I might not understand it all that well. Don't let me discourage you on that one. The other thing is ask the supplier for a simple tension test for the resin.

                  • Re: Mooney-Rivlin Testing
                    PJ Pisczak

                    Bill,

                     

                    I don't understand trial and error as I don't think there need be any.  If one is loading material test data into the Simulation, and your point that the shape doesn't matter, then one should be able to calculate the C1 & C10 as long as the model is the same on which one tests.  Since we are dealing with compressive O-rings, a tension test seems to run counter to the o-ring application.

                     

                    I guess we are spoiled because we test both ways [1] absolute, e.g. dog-bone samples and [2] relative - the actual part.  I tend to use the dog-bone information for my iterative models and then test for how well the finished, best estimated, parts perform.  Lastly I compare that with our FEA model to close the information loop and see how close we came and whether any adjustments need be made in the future.  But everything I've read and done previously with Mooney-Rivlin (Algor, Ansys, Mechanica, Rand-Micas) indicates that it is better to test (hence this latest effort) as C1/C10 are sensitive with variations created a wider range of analysis results.

                     

                    I will take ten (10) of our o-ring parts, put it on a Tinius-Olsen and determine what the compressive response curve looks like, up and down.  I'll try this approach and see how the C1/C10 constants compare to those already floating around the internet.  Whatever the results I will report that to the SW Portal.  My hope is to determine accurate constants by using parts which are close to what we are investigating.  Appreciate your thoughts.

                     

                    Phil