data for ABAQUS' hyperelastic models

[Robert]

Hello,

To date, I have modelled bitumen with ABAQUS’ linear visco-elastic constitutive model. To calculate the Prony series parameters, I have supplied ABAQUS with oscillatory shear test data (frequency versus storage and loss modulii) from Rheometer tests.

I would now like to look at bitumen behaviour for non-linear strains (say up to 100%), using a visco-hyperelastic constitutive model. I am thinking initially I will use a Marlow hyperelastic model, and would like to again use oscillatory shear test data from a Rheometer to calculate the hyperelastic model parameters, but I am unsure about how to go about transforming my shear data into a form suitable for ABAQUS’ hyperelastic models (either uniaxial, biaxial, or planar test data). The bitumen is incompressible.

Thank you.

[Jorgen]

Interesting. What simulations are you planning on performing using the bitumen material?

My recommendation is to use uniaxial compression experiments instead of rheometer experiments to characterize the large strain behavior of the bitumen. That way you get data that is easier to interpret and you can use ABAQUS's built-in methods for fitting a hyperelastic model.

I also recommend that you perform the compression experiments at different deformation rates. That way you can evaluate the rate-dependence of the material.

- Jorgen

[Robert]

Jorgen,

Many thanks for the help.

I am attempting to model the behaviour of bitumen for road surfacing applications.

[sq]

The shear strains that a rheometer are capable of providing (typ <1%) are almost certainly too small to identify any hyperelastic model beyond neoHookean, and even in that case the range of applicability will be limited.

One approach might be to start by identifying the HE model using uniaxial compression data, as suggested already.

I have seen a number of cases of then applying a Prony series as calculated from DMA. This can be a bit sketchy, though, for a couple of reasons. First, it is often very difficult to match DMA moduli (at a given frequency) to tensile test results at the corresponding strain rate. There is often a bit of hand-waving about how DMA or rheometer results should be scaled to match tensile test data. Second, the mechanisms of energy loss are often stretch- and rate- dependent. That is, it may not be correct to infer storage modulus or loss tangent at large strains from small strain data.

Finally, while I agree with Jorgen that you should perform experiments at different strain rates, I think it's also appropriate to determine what strain rates you need to cover to model your situation appropriately. If you're going to model the interaction of tires with the road surface, the contact times may be very short compared with what is possible using traditional uniaxial test rigs.

[Robert]

Thank you.