ABAQUS viscoelastic model
I am using ABAQUS to model a bitumen slightly above its glass transition temperature (i.e. the bitumen is therefore a very viscous liquid).
I would like to use the viscoelastic model provided with ABAQUS, but this requires that the long-term modulus is non-zero (i.e. no permanet set). This is not true for liquids.
Given that the loading rate we wish to model is high (60Hz), do I need to write a UMAT with the long-term modulus as zero, or can I use the standard viscoelastic ABAQUS model and asume that the long-term modulus is non-zero?
Hmm, I see your problem, you are right that ABAQUS requires that the long-term modulus is non-zero.
I also agree with you about the two possible solutions: (1) use what is currently available in ABAQUS; (2) write a new umat that better predicts the material that you are interested in.
In order to decide if the built-in viscoelastic material model is good enough I would compare the model predictions with any experimental material data that you might have. In addition, I would examine the characteristic relaxation time of the material at temperature conditions that you are interested in. That way you can better decide if 60 Hz is fast enough for the visous resistance to dominate the material response.
Also, FYI, it is not that difficult to write a UMAT for a visous material. If you have the time, if could be a good learning experience.
Best of luck,
A very long time replying, but after many experiments I think I need to have a go at writing a UMAT to properly simulate viscoelastic materials with a zero long-term modulus.
My knowledge of UMATs is currently very limited and this is perhaps a silly question: should the UMAT be based on a mechanical model that permits viscous behaviour (e.g. the Burgers model) or should I take some other approach to predicting stresses and strains?
There are different approaches for formulating a UMAT. For example, you can use differential equation approach, or you could use an integral equation approach. These days I think that the differential equation approach is more common for finite element applications.
There is no fundamental reason why some UMAT formulation approach is better than some other as long as the models fulfill all rules of physics, thermodynamics, etc.
It is often useful to derive the model based on experimental insight into the behavior of the material that you are interested in.
another possibility is to use ANSYS: It does NOT require that the long-term modulus must be non-zero. Of course, that does not mean that the analysis converges all the time due to losing shear stiffness.
for writing a UMAT for viscoelasticity, you might find my PhD thesis useful:
Upsetting and Viscoelasticity of Vitreous SiO2: Experiments, Interpretation and Simulation