Thermal Shrinkage of Polymers

[Jay102]

I would like to model a typical thermal shrinkage of a 10cm x 10cm Polyethylene Terephthalate (PET) film (think "shrink wrap") that is contained by a square metal frame. A temperature field is applied to the film, which shrinks the film and causes the metal frame to cave inwards (the extent of this caving is what I am interested in).

How can I model the PET film such that it shrinks "the right way" under (high) temperature? Entering a negative coefficient of thermal expansion alpha (CTE) seems to do the trick (despite PET having a positive coeff), but is there a more "technically" correct way of modeling it in ABAQUS? I don't see how it can model random molecular coiling of polymer chains...

Thanks.

 

[vumat721]

What do you mean by the "right" way? Can you give a clearer example about what Abaqus is doing and what you expect to see?

 

[Jay102]

Hi vumat721,
What I mean by the "right" way is, if an FEM expert were to model the heat-shrinking polymer film, would s/he simply model it as a 2D sheet with a negative CTE (obtained from empirical data)?

For bonus, is it possible to model the wrinkling of the thin film? What kind of data input (material properties) does ABAQUS require to model wrinkling?

 

[vumat721]

The right way would be to assign temperature dependent values to your elastic constants and thermal expansion coefficients for both materials. You should be able to input these as tables from experimental data. If the experimental data shows a negative CTE, then that's what you should use.

I would try modeling the sheet with as much symmetry as possible. For example, plane strain or axisymmetric, or quarter or half-symmetry for the full 3D. Use at least three elements through the thickness if you want to capture bending accurately. It's hard to say if it will pick up wrinkling, that will depend on the source of the wrinkling. If it's due to plastic deformation of parts of the sheet, then you will need to include plasticity in your material model. If the wrinkling is due purely to kinematics, then you should be able to pick it up in some form using an elastic model.