No announcement yet.

Rate-dependent Hyperelastic Model

  • Filter
  • Time
  • Show
Clear All
new posts

  • Rate-dependent Hyperelastic Model

    Hello All!

    I'm trying to find a model that can describe the stress-strain behaviour of a polyurethane which shows the behaviour attached herewith. I am using LS DYNA.

    Can anybody suggest a material model ?
    Thanks in advance

  • #2
    For clarity, the image shows the behaviour of just one material at different strain rates. Please ignore the legend of the graph


    • #3
      In LS-Dyna, you can do that directly with MAT181 - Simplified rubber/foam. It allows you to define a table containing several lists of force-displacement data. Each list matches a constant strain rate ('value' for LCID in table definition). Data is given as force-displacement and recalculated internally to stress and strain with the test specimen dimensions. When the dimensions are all set to '1', data in the lists can be entered as stress vs. strain. But as always, use consistent units! I'd recommend to read the manual very carefully and run a few tests on a unit element to be sure that the material model matches the expected behaviour.

      A few small questions:
      - Is that strain axis going from 0-8% or 0-800% ?
      - Which strain rates are we talking about? Are you sure the machine maintained a constant crosshead speed from start to end of the test? This is most often not true for speeds above 1m/s, depending on the machine and stiffness of the specimen.
      - Some of the curves are crossing others. Is this expected or is there an explanation? I haven't seen that in my high-speed tests of elastomers; the stress-strain curves always showed higher stress level for higher strain rate. It might also cause an instability in your simulation when an element in the mesh jumps from one strain rate to another in the calculation.

      In the end, I chose for a coupling of hyperelasticity and (linear) viscoelasticity, which I calibrated by inverse modelling of the tensile test at different crosshead speeds (in a time table as recorded on the machine). This was very tedious, but gave a reasonable result in the end.
      But I guess that MCalibration would be very suited for this task. If anyone would be so kind to explain how to use it in this case?



      • #4
        Hello Joren!

        Many thanks for your reply.

        The curves I'm trying to replicate using LS DYNA are attached herewith. In the picture attached previously was something similar on my phone at the time. Sorry about it.

        I am indeed trying to use user-definedClick image for larger version

Name:	mcal.jpg
Views:	5
Size:	45.0 KB
ID:	25946 material subroutines. I used MCalibration to calibrate the 9 Parameter Mooney Rivlin model. But it was very hard. I will attach a photo of the best match I got by partially manipulating the coefficients. You will see it does not match very well at low strain range.

        I managed to compile my UMAT and run some simulations.
        This work is based on the following paper

        I managed to recreate the stress strain curves but my curves are shifted downwards instead of upwards as can be seen in the attached image. I cannot figure out why.
        Any input from you is highly appreciated.

        Moreover, my material model assigns the failure stress and strain based on the strain rate applied.

        DakshithaClick image for larger version

Name:	my simulations.JPG
Views:	5
Size:	19.2 KB
ID:	25945
        Attached Files