Simulating a compressed EVA pad

suktansuktan
Hello,

I am a beginner of FEA. I am trying to simulating a compressed EVA cylinder.

The EVA was purchased from grocery shop. It is a cylinder of outer diameter 30 mm and height 9.6 mm. It weights 0.78 g. I checked it with a Durometer Shore A meter. The average value is 28. I used the conversion give in this link https://3dvision.com/blog/entry/2011/07/14/convert-durometer-to-youngs-modulus.html to obtain Young's modulus of 1.0179 MPa. Poisson's ration is assumed 0.4.

The EVA pad was compressed by a force meter which was mounted to z-axis of a CNC machine. The meter had a circular metal adapter attached. Diameter of it is nearly 30 mm. When the EVA was compressed for 1 mm, the meter showed a peak value of 39.4 N. I want to use this experiment to verify the simulation.

I created the model by using plate mesh and extrude tools. I added frictionless support and force (39.4 N) for constraints/loads.

Am I doing things both modelling and experiment correctly?

Displacement in Z converged to ~0.44 mm. I can't figure out where I did wrong for not getting a closer value.

Thanks for reading. Please give me some hints.

Comments

  • DaveStuppleDaveStupple
    edited May 2017
    Hi Suktan,

    I am not very expert at this, but I can give you some suggestions for now, and later you will get replies from the real experts.

    I have recently been trying to simulate compression of an O-ring.
    http://mecway.com/forum/discussion/418/compressible-o-ring-with-contact-surfaces#latest

    The first thing that was suggested to me was to define the elastomer as a hyperelastic material. I think this might be appropriate for you with such low modulus. You will need to have CCX software installed (now bundled with Mecway, I think, but also available elsewhere) and select this as your solver in the analysis dialogue. It doesn't need to be any more complicated than that I think, for your analysis, as Mecway should take care of dealing with CCX. With the Neo-Hookean option (as I used), instead of Young's and Poisson you define C10 and D1 parameters, which are defined here (downlaods a document).


    I remember seeing different formulae for the Shore to Youngs conversion, so there might be some uncertainty there. Trying different assumptions for Poisson might help, I know it is very difficult to find listed values.

    Also, you might try a more refined mesh, though this might make no difference other than slowing down the solving. If solving time becomes a problem, you could reduce node numbers by taking a quarter slice of your model, and imposing frictionless support on the 'cut' surfaces, so it still behaves like a full disc. If you remesh with the concentric circle option you could make the slice thinner.

    I hope this makes sense, and you will definitely get more authoritative advice from the other guys, but I have had so much help on here before that I am happy to make an attempt.
  • SergioSergio
    edited May 2017
    For EVA I would choose hyperfoam material, is not available directly from Mecway´s interface, but you can easily setup by means of CCX custom cards. Now stiffness of foams will depend on material and density, so you must perform a kind of material calibration to find the coefficient of your own material. As you have a way to measure load/deflection, my suggestion is to build a curve up to 60~70% of height of the sample, and then start to play with the material coefficients until you find an acceptable agreement.

    Regards
  • suktansuktan
    Thank you, Dave and Sergio. Very useful hints!

    I will try CCX for hyperelastic.

    Regarding conversion from Shore A to Young's modulus, Wikipedia shows three equations. For Shore A 28, I have 1.0512 MPa, 0.633 MPa and 1.0416 MPa. The middle is nearly 50 % less. What do you think?
  • SergioSergio
    True is as a rubber FEA part analyst for years, I never use this kind of coarse aproximation for rubber, I always use hyperelastic with specific material coefficients. I have told to use such a coarse representation of the rubber only in case that is needed to include in an assembly and the exact rubber behavieur is not requested, only as a means of translate or apply loads on some metalic components.

    I'm not sure that you can charaterize accuratly a foam with Shore A hardness, have worked on some poliurethane foams and they don't reffer to Shore but static load deflection of the part itself.

    Regards.
  • suktansuktan
    Thanks a again, Sergio!

    It seems that we better get detailed specs from supplier or test the rubber material by load deflection or tensile test. Shore is not reliable. Correct?

    Sorry, my background is EE. A newbie in this area.
  • suktansuktan
    I came across a page of google book "Physical Testing of Rubber". Although my compression test is not perfect, I think the equation is still useful for me to estimate the Young's modulus (E) in a different way. Well my sample is EVA disc instead of a real rubber. In my case, I have F = 40 N, A = 706.86 mm^2 and Lambda = 0.8958. E is obtained as 0.48453 MPa.

    I used this value in the FEA. Z displacement was 0.9245 mm. It is not too far from the measured (or setting) 1 mm. Do you see anything wrong here about estimating E and using it in FEA?

    Thank you.
  • SergioSergio
    What's the aim of your simulation? All depend on that, if you need an rough estimation of deformation, then maybe is enough. If you need to predict stresses, or durability I suggest you go further with more complex material modeling.

    Regards
  • suktansuktan
    Hi Sergio,

    I currently aim for deformation estimation. I wan to learn some basics of FEA and modelling from this exercise. Since I can do simple compression test as stated, FEA result can be verified. I can then know whether I made something wrong or not.

    My next step is to estimate deformation of rubber roll-surround used in loudspeaker drivers. There are some samples readily for load vs deflection (or displacement) measurement. I will keep the deflection small (linear assumption holds). Again, I will use measured data for verification.

    Cheers
Sign In or Register to comment.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Sign In Register