UMAT path dependent linear viscoelastic model

[Wenani34]

Hey Jørgen and fellow members,

I'm working on implementing a path dependent linear viscoelastic UMAT subroutine which accounts for relaxation behaviour during the curing of a thermoset polymer (as implemented by Svanberg et. al 2002). To accompany the UMAT subroutine, UEXPAN is also used to calculate thermal expansion and chemical shrinkage strains.

When carrying out a simple tensile test modelling example with thermal effects, i have a problem with excessive distortions once the bar is stressed by a load. In the model i'm changing between two stiffness matrices (for the polymer in rubber and glassy state depending on T_g).

Is there something i might be overlooking? All my expansion strains passed on from UEXPAN are correct.

Any comments etc. are welcome )

Wenani34

[Jorgen]

Interesting project - I have wanted to work on that issue as well for a while now...

My guess is that you UMAT is not quite correct. Have you debugged it?

-Jorgen

[Wenani34]

Hej Jorgen,

I debugged the code and found the errors so now the model works fine for a neat resin. This leads me to the next problem (challenge).

I have included fibre material properties now using micromechanics as well. The whole idea with the model is replacing the time dependent relaxation modulus with a thermal expansion (different in glassy and rubbery phase), chemical shrinkage and frozen-in deformations. Making a long story short, this means including a switch using internally stored state variables, which trigger relaxation and the change in CTE, chemical shrinkage coefficients and equivalent laminate stiffness depending on the cure degree, temperature and T_g.

As you can imagine this involves sudden changes in the behaviour of the material resulting in convergence problems. Is there some kind of damping effect that you could recommend to avoid this when the changes are triggered?

- Michael (Wenani34)