Material modeling on toughened epoxy

[braided]

Dear Jorgen,

I learn a lot from this great website. Thank you very much! Now it’s for me to ask the ‘stupid’ question as follows: :lol:

I am working on a project involving high velocity impact on braided composites, and I try to do some micromechanics analysis. One step is I need the material law for matrix. In my case, the matrix is toughened epoxy, such as 977-2, And I have tension, compression and shear data at the different strain rates (from low 10-5/s to high 500 /s). However, I don’t have any unloading data for those materials. Can you give me some suggestion on how to model this material?

1. I already programmed a UMAT based on the combination of Drucker-Prager yielding model and Bodner’s state variable theory in LSDYNA. And I use explicit method. In the one element test, the simulation can simultaneously match test data in shear and tension at different strain rates using one group of material parameters, but I need to adjust parameters for compression. (is this strange?) The more serious thing is I do multi-axial simulation; the simulation results are very very :roll: strange, such as ‘softening’. What could cause the above problem?

2. What’s kind of unloading curve shape should be for toughened epoxy? In my understanding, epoxy is thermosets, my stupid question is whether thermosets have plastic deformation from the viewpoint of micromechanics? :cry: The idea using Drucker-Prager model is from other researcher’s work on thermoplastics.

Thanks!

Jim

[Jorgen]

Hi Jim,

It sounds like you are working on a very interesting project and that you are making good progress.

Here are a few comments:

:arrow: The unloading behavior of thermoset polymers is different from the "up-loading" behavior. Even a material such as epoxy will exhibit hysteresis during cyclic loading, if the strain is large enough. Hence, you can experience a certain degree of plasticity also with these stiff materials.

:arrow: I am somewhat surprised that you were not able to simultaneously predict both tension and compression. With the appropriate material model (and material parameters) that should be possible. The softening that you see during multiaxial loading is that from monotonic loading? If so, your material model is not Drucker's stable and I would be hesitant to use it for general multiaxial simulations :?

:arrow: How different is the experimental response in tension and compression? I suspect that the initial response of an advanced material model specifically developed for thermoplastics can be made to accurately fit the response also of toughned epoxy. Since the strain to failure is rather small, all you really need is a model that captures the influence of strain rate on the stiffness, the influence of hydrostatic pressure on the yield behavior, the distributed onset of localized yielding, and a realistic I1-based hyperelastic network response. The Hybrid Model (http://www.polymerfem.com/modules.php?name=Downloads&d_op=viewdownload&cid=5), for example, contains all of these characteristics.

Best of luck,
Jorgen