Viscoelasticity & Hyperelasticity in Abaqus

[jydmarsh]

I am trying to model several carbon-black filled rubbers (natural and butyl) in Abaqus. I have test data from static simple tension, static planar tension, simple shear stress relaxation and simple shear sinusoidal cyclic tests. The hyperelastic models have been successfully fitted using the Arruda-Boyce model.

I have tried several options in the viscoelastic material and none of them are producing results that match the test data. I first tried defining the stress relaxation curve. I was able to get a good prony series fit and a stress relaxation test in Abaqus matched my test data. When I ran a sinusoidal displacement pattern using this material, the energy dissipation of the sample was much much lower than the test data (.05 loss factor versus 0.35). I then tried the frequency dependent data from the cyclic tests and got my prony series fit from that with the result being same low damping. Any suggestions or assistance would be greatly appreciated.

Thanks,
Justin

[Jorgen]

The linear viscoelastic (LVE) model is sensitive to the type of experimental data that was used for calibration. It sounds like you tried a few different approached, which is good. It might be possible, however, that the calibrated Prony series that is used to represent the LVE model is too narrow, and the experimental characteristic time is outside the range covered by the calibrated LVE model.

Or another option is that linear viscoelasticity is simply not an accurate material model your material and loading conditions. You could try the Bergstrom-Boyce *Hysteresis model.

[jydmarsh]

I have overcome to a degree the issue and am now using the Hysteresis option with the Arruda-Boyce Hyperelastic function. The problem is that I am getting negative eigenvalues as part of my analysis and it is telling me that my Hysteresis Jacobian cannot be invertable. I have tried changing from incompressible to compressible and refining my mesh but neither has worked to overcome the problem. If anyone has experienced this before or has an idea I would be very grateful.

[Jorgen]

It could be caused by ABAQUS' implementation of the Hysteresis model. I have created my own implementation (UMAT and VUMAT) that appears to be more stable.

If you want I could try to run one of your simulations using my implementation (http://usermaterial.polymerfem.com/) of the Hysteresis model (or as I call it, the BB model).

- Jorgen