FEA with Viton

[PaulF]

Hi,

First of all, congretulation for this website ... GREAT one.

I'm used to performe FEA with metallic parts and for the fist time, I've to make a calculation with Viton material and I don't know how to do it !

I'm using SAMCEF code with some internal constitutive equations :
- Maxwell/Zener/ Kelvin for viscoelastic material
- Mooney-Rivlin / Hartsmith / ogden /hyperfoam for hyperelastic one

Except the "classical" viscoelastic material equations (seen at school), I've never used such law (and laws for rubbers/polymers ... materials)

What is the best law I can use ? Where canI find the different parameters (determination from the curves herein the website) ? Do I've to considere a "pur plastic" behaviour using the sigma-epsilon curve and to use a non linear calculation ?


Thanks for your advices

Paul.

[Jorgen]

Welcome to the world of polymers :wink:

Viton is a fluoroelastomer and behaves similar to many other types of elastomers. It is difficult to give specific advice without knowing more about your problem. For example, do you have experimental stress-strain curves for the Viton material that you are intrested in? What strain magnitudes are you interested in? Do you want to predict stress relaxation and other viscous/time-dependent effects?

The simplest model that you can try is the Neo-Hookean or Mooney-Rivlin model. These models work reasonable well for simple simulations. You can also activate viscoelasticity with these models.

- Jorgen

[PaulF]

Hi,

Thanks for your first reply !

As you've understood, I'm novice in polymer field ...


About my problem, I'm not interested in the stress/strain mapping in the viton part but I want to take into account its viscoelastic behavior in dynamic calculations .... i want to validate the design of the metallic parts arount the viton one !

I don't find any viton data except the curves in your website (no young modulus, poisson ratio, no density or no damping coefficient for exemple) so I've different possible scenario :
- either I'm able to determine the parameters from your curves by simulating the compression test on a single axisymetric element for example ... but it's a bit boringly since I've to manually iterate !
- either I determine the parameters with the square mean root method (using Marquard-Levenberg method) .. then I've to calculate the simplest analytical formula

If you've any other suggestions, I'm well interested in !

Thanks for your advices

Paul

[sq]

The answer to how complex a model you need depends on how big the range of conditions you want to simulate is.

First, what is "dynamic"? Are you talking about 1 Hz, 10 Hz, 1000 Hz? Do you need accurate simulations over a large range of frequencies? Second, how large a strain are you worried about?

Now, here's the real trick- what kind of test can you use to identify the model parameters, and what kind of time-dependence do you need to figure on?

If your strains are small, Dynamic Mechanical Analysis is for you. As a bonus, linear viscoelaticity may be sufficient for your modeling needs. If your strains are large, then you need either multi-speed tensile tests (if your frequencies of interest are low), or something like Split-hopkinson pressure bar testing (if your frequencies are high). For large strain, neohookean will not suffice, as discussed above. Linear viscoelasticity may also be insufficient, and you'll need to use a nonlinear model such as Bergstrom-Boyce or similar.

P.S.,

I'm always a big fan of Marquardt-Levenberg, though I urge caution in using it with some of Dr. Bergstrom's models, for which robust optimization measures usually work better than quickly-converging ones. At very least, you will need to restrict your search domains lest you quickly find yourself seeking out aphysical (and numerically difficult) solutions.

[PaulF]

Dear Sq,

All you're saying makes good sense ! As you've beeing thinking, I want to calculate the stress/train mapping of my system that includes a Viton seal (modal analysis + dynamic response of my system in a 5 to 4000 Hz frequency range)

Then I think so linear viscoelasticity is sufficient (small strain) ... then I've to determine the parameters from the experimental curves herein.

About the damping ratio of the material, have you any data ?

Regards

Paul

PS : I'm using Gnuplot (GPL licence) to determine the parameters by Marquardt-Levenberg method (mean square root method ? .... i don't remember the name in english)