Viscoelastic materials from DMA and relaxation data

[Jabba]

Hello,

I am trying to simulate a dynamic three point bending experiment in Abaqus, but first, I'm trying to make it work in static. I have done several tests in tension, bending, DMA, stress relaxation.

My material looks like this:

Hyperelastic, Ogden model, n=3, uniaxial test data
Viscoelastic, Frequency, Tabular
and
Viscoelastic, Time, Shear Relaxation test data

I have also done several static simulations (step=visco) with these viscoelastic material models at different strain rates. Although the relaxation based material differs from the DMA based material, the strain-rate has no influence on the DMA material.

My questions are:

1) I have several DMA material models, for different types of tests and different amplitudes. Why is my material behaviour identical, regardless of the input data? (I used data for 0.1Hz, 1Hz, 10Hz and 100Hz in one case)

2) Why aren't the DMA material models influenced by strain rate, as the stress relaxation model is?

3) Can Abaqus/explicit determine strain-rate influenced results, as the "visco" step?

Thanks,

Dan.

[Jorgen]

What material are you studying?

Here are some answers:

1) Different experimental data should give different viscoelastic material models.
2) A viscoelastic material model should be influenced by the applied strain rate.
3) Yes, you can capture the influence of strain-rate also in Explicit simulations (if you use an appropriate material model).

I am not sure what you mean by "DMA material model".

-Jorgen

[Jabba]

Hello Jorgen and thanks for your prompt reply

I am studying a p(amide)-12 based thermoplastic polymer (it is more compliant than regular nylon 12 and it is resistant to humidity changes).

I have solved the first two problems. The viscoelastic material model should have been set in the Time domain/Frequency not in the Frequency domain/Tabular.

"DMA material model" is a short version of "Viscoelastic material model based on DMA data".

Now I'm trying to figure out how do the viscoelastic material models influence the input test data. I have inputted an uniaxial stress-strain curve form my 2mm/min experiments to calibrate the Ogden model. In my experiments, the tensile strength saturates after a given strain rate (about 2000mm/min).

The stress relaxation based material response saturates around the values of the "virgin" hyperelastic material model (w/o viscoelasticity). So in order to get an accurate response, I should input the test data for the saturated stress-strain curve, and let the viscoelastic material model "soften" its response with lower strain rates, right?

However, this approach is not suitable for the viscoelastic materials based on DMA data, as their behaviour seems to be the exact opposite: the stress-strain response saturates at a higher tensile strength than the inputted data, and for small strain rates, its response coincides with the inputted data.

To make this easier to understand, I have attached a picture with the stress-strain response in tension:

"Experiment" - the experimental stress strain curve for 2mm/min
"Relaxation Slow/Fast" Viscoelastic material model based on relaxation data, simulated at slow/fast strain rates
"DMA Slow/Fast" Viscoelastic material model based on DMA data, simulated a slow/fast strain rates

What are your thoughts on these curves? Have you encountered such behaviour for other viscoelastic materiasl? Or I am getting something wrong in transforming experimental data into Abaqus input data?

Thanks,

Dan.