EPP foam material model and Failure Criterion

  • 66 Views
  • Last Post 13 June 2018
Miguel posted this 03 June 2018

Hi, I am trying to simulate the structural behaviour (2 cases: static and dynamic) of an EPP(expanded polypropylene)-foam platform being part of an assembly. I have stress-strain data from both the uniaxial tension and compression tests. I tried to approximate it with hyperelastic models (not sure though if this is the best choice) but I could not get an acceptable curve-fit with any of them (separately, several models like Polynomial or Ogden make a pretty good job "curve-fitting" the test data but when I try to fit the whole range of available data combining both tests data nothing works...).

A second doubt is about the right Failure Criterion for this kind of materials: I opted for Mohr-Coulomb as a conservative choice for brittle materials but again, not sure if it is the best option.

Anybody experienced with this kind of polymer foams or with any idea/suggestion? Thanks, it´ll be appreciate it.

Order By: Standard | Newest | Votes
peteroznewman posted this 03 June 2018

Hi Miguel,

Can you share your stress-strain data for tension and compression?  Put whatever document files you have into a zip file and attach it to your reply.

I don't think a Mohr-Coulomb material is going to be a good material model choice.

What is your definition of failure?

Miguel posted this 04 June 2018

Hi Peter,

Thanks for your quick reply. Please find attached the zip file containing the material data (the grade I am interested in is 60g/l - 5135) I have as well as a wbpz file where I tried to reproduce the tensile and compression tests according to the corresponding Standards as a way to validate a model to be used later in the real case. As you will see, I tried to fit the data to several different models in different ways but, unfortunately, I did not manage to get anyone right: the best I could achieve was a pretty good "replay" of the tensile test with a polynomial-2nd order model fit (System J) but no success with the compression one and no clue about how to include the whole data range (both tension and compression) in only one model... Not sure whether I am doing something wrong or it cannot be done just with these input data (the material supplier does not offer any additional tests data) or hyperelastic models are not suitable for this case...

Regarding definition of failure, since there is no yielding associated with this type of materials and they respond with high strains with some level of strain rate dependency (according to the manufacturer), the goal I set is to withstand the applied loads in both cases without the material breaking (a small residual permanent deformation might be admissible provided it won´t cause interference problems with the rest of the assembly components).

Looking forward to hearing your take on this. 

Attached Files

peteroznewman posted this 04 June 2018

Hi Miguel, I downloaded your zip file. Unfortunately, ANSYS 19 generated an error restoring your archive. Please create a fresh Workbench Project Archive .wbpz file and Attach it to your reply.  Also say what version of ANSYS you are using 18.2, 19.0 or other.

Regards,
Peter

Miguel posted this 04 June 2018

Hi Peter,

I don´t know why the wbpz file got corrupted... I created it again (system J is still the only one producing reasonable/correct result; I find all other results inconsistent...  would you agree?). Hope this time works OK.

I am using ANSYS 18.0

Regards, Miguel

Attached Files

peteroznewman posted this 04 June 2018

Hi Miguel,

Wow!  You have done a lot of work!

What I plan to do is to use your experimental data with the ANSYS Curve Fit solver to see if I get a good fit. Did you use this already? While the tensile data is useful, the compression data is not in the format that the Curve Fit solver can use, which is volume ratio and hydrostatic pressure.

This may take some time, and I have a 3-day conference to attend starting tomorrow, so I may not be able to provide more information for 5 days. Maybe another member will have some comment in the mean time.

Regards,
Peter

 

Miguel posted this 04 June 2018

Thank you Peter,

This is really just part of the work!: it is only the material model validation through the available tests data to be able to apply it to the real model... Yes, I tried several CurveFits: some worked fine although just partially as I said, others not at all. You can see them in the EngineeringData window.

I´ll wait for you to see if you can get anything better than me (of course any other comments from members are welcome!).

About the Failure Criterion, can you tell/recommend me something?

Regards, Miguel

peteroznewman posted this 05 June 2018

Hi Miguel,

I think I will be learning more from you than the other way around.

The results might allow you to predict whether the material would have failed, without simulating actual element failure.

Regards, Peter

Miguel posted this 13 June 2018

Hi again Peter,

Did you have a chance to figure out anything on how to go about my issue? What did you mean exactly with your last suggestion? I didn´t really get it...

Thanks. Regards, Miguel

peteroznewman posted this 13 June 2018

Hi Miguel,

I haven't been able to improve the fit you made to the experimental data.

What I meant by the above comment relates to my experience with materials where I added plasticity. Ductile materials have an elongation material property that is the value of total equivalent strain where the material fractures. I can take a model way past that point and the material just keeps stretching, but I know by looking at the total equivalent strain that I am beyond the point where the material has failed. So the model was valid up to that point, then was not.

I have been investigating how to automatically deactivate the elements that have failed so I can continue to apply load and have a valid model. Explicit Dynamics automatically will take out elements that have failed and keep going. That is not so simple in Static Structural, but many students want to do this.

Kind regards, Peter

Close