FEM simulation of Polypropylene / polypropene (PP) and PET

[PerHS]

I am trying to simulate large deflections on thin PP and PET parts.

I am currently using ANSYS Professional since that is the only FEM software I have, but this software does not support the required hyperelasticity material models which I believe is needed to do accurate PP and PET simulation.

Can anyone recommend FEM software and material data/models suitable for single and multiple PP/PET part simulations of large deflections on PP and PET?

Priorities in given order: cost, ease of use, accuracy

Thanks for any input!

[sq]

Well, if cost is the first concern, I think you're going to have to be satisfied with ANSYS. ABAQUS (which I think most here would recommend) is not inexpensive.

Doesn't ANSYS give you the capability of programming constitutive models?

However, I seem to remember that ANSYS already supports hyperelastic models, both hyperelastic elements and material models.

[aarriaga]

Hello, well I think that ANSYS can be as suitable as Abaqus even for developing your own material models. The problems is that maybe you are considering the wrong material behaviour since thermoplastics like PP or PET do not follow hyperelasticity. The usual approximation in ANSYS (not always precise) is to consider elasto-plastic behaviour and to use a isotropic or kinematic hardening behaviour with Von Mises yielding. This is metal behaviour but with some plastic materials gives susprisingly good results.
Good luck!

[Jorgen]

I agree with aarriaga that ANSYS can be as suitable as ABAQUS. I also agree that PP and PET cannot accurately be represented using a hyperelastic model.

Do do not agree, however, that metal plasticity models are suitable for thermoplastics. In my opinion, metal plasticity should only be used as an absolute last resort. One of the main problems with traditional metal plasticity models (isotropic or kinematic hardening with J2-flow theory) is that this type of model can often be accurately calibrated to a limited set of uniaxial data, giving a false sense of "validation", when in fact the predictions in other multiaxial loading modes can be way off. This should not be surprising since the micromechanisms of deformation are totally different in polymers and in metals. For a more detailed discussion on this see the following paper (http://www.polymerfem.com/modules.php?name=Downloads&d_op=getit&lid=12).

- Jorgen

[aarriaga]

Yes, Jorgen is right, the material model validation should not only be done in uniaxial tensile mode but also in compression and shear for calibrating multiaxial loading modes.
My viewpoint was from an engineering or industrial use, the usual way of working with plastics that undergo large deformation is treating them as elasto-plastic materials, have no doubt about it. We work in collaboration with different thermoplastic product design centers in static and crash applications and for example for crash applications the "standard" material model is MAT24 (strain rate dependent elasto-plastic with Von Mises), even knowing that it does not represent real polymer behaviour. First the material model is checked with simple geometries (specimens, plates) under flexural or dart perforation tests and then complex parts are studied. Experimental and simulation errors have been acceptable in many cases.
Regards,

Aitor Arriaga

[PerHS]

Thanks to sq, aarriaga and Jorgen_Bergstrom for valuable feedback :)

It seems that ANSYS should work, but I have to verify that my license can handle custom material models. COMSOL Multiphysics is also available, would that work, at least for single part models?

Regarding material test data - I have found several articles publicised on Internet describing experimental behaviour of PP that I would like to use for verifying material model. Two articles come to my mind here:
"The effect of strain rate on the viscoplastic behavior of isotactic polypropylene at finite strains" by Drozdov, A. D.; deC. Christiansen, J.:
http://arxiv.org/PS_cache/cond-mat/pdf/0207/0207741.pdf

and a similar by the same authors for annealed PP material:
http://arxiv.org/PS_cache/cond-mat/pdf/0205/0205362.pdf

Would it be necessary to contact the authors to add their data to your material model base, or would it be enough to mention them as a source for verification of theoretical model?

[sq]

Hey- my work (incl. experimental results) has been used publicly without my even knowing about it (the work was in aerodynamics, not polymer physics). But, you can't go wrong contacting them- they might even be happy to give you the data in a more convenient form!

[PerHS]

Good idea! As you say, they might have the data available in a more suitable format making verification much easier.

-Per

[hhspiny]

In my work, I also use metal plasticity for glassy polymers. this has to do with the fact more involved glassy plastic models are not easily obtainable, not presenting in abaqus/ansys etc. More importantly, without very careful calibration, all the "advanced" models do not give better result than simplest metal model. For glassy plastics, it is not usual in practical design to load them much into "plastic" region, which is normally considered failure. So an accurate prediction is only important during failure propagation prediction. However, from my experience, during the failure propagation prediction, and accurate multi-axial prediction is not that important, as first such process either happens very fast or confined in small region, in addition, other phenomenon starts to kick in, such as small cracks, crazing.
as a simulation group, we always hate to take any failure prediction cases presented by out polymer colleagues, and often feel powerless in such cases. On the other hand, our life is breezy when we deal with ceramics colleagues.

Experimental and simulation errors have been acceptable in many cases.
Regards,

Aitor Arriaga

[PerHS]

PET could probably be simulated using metal plasticity but realistic material models is in my case required for PP due to large and repetitive deflections beyond yield.
Regarding ceramics; I have done a lot of glass and alumina simulations, and have had the oportunity to follow up the results with lab tests. Mohr-Coloumb models in ANSYS have consistently given acceptable error - less than 20% deviation for short term loads. In these applications this is good knowing the complex failure mechanisms of brittle materials.

[mollard]

Hi Everyone,

I read with interest the posts in this thread. Would anyone out there have some generic tensile stress-strain curves for isotactic PP? I'd prefer 20 - 30% talc filled, but even unfilled would be great. It should be a relatively high strength molding compound, preferably intended for a piping application.

Can anyone help or point me to a source?

Mollard

[Jorgen]

I am afraid that I don't have any data.

Check this thread (http://polymerfem.com/forums/showthread.php?t=242), it has a discussion of where to find experimental data.