curve fiting

[christellegarcia]

Hello
I know that if I want to use the curve fiting mode, I must do any tests on differents materials. I would like to know if the tensile strengh VS strain and the compressive test is sufficient to caracterise materials (elastomers). I know that if I have more tests, calculation will be more precise. But the precision brought by the others are improved of how pourcent?
Thanks
Regards

 

[Drej]

Good question.

1/ If only limited data is available, consider using some of the constitutive models based on first strain invariant only.

2/ The Yeoh, Arruda-Boyce, and Gent models may provide realistic responses for rubbers in other modes of deformation if only uniaxial test data is available.

There is no simple answer to this, but have a look at the following articles:

http://www.padtinc.com/epubs/focus/common/focus.asp?I=12&P=article3.htm

http://www.padtinc.com/epubs/focus/common/focus.asp?I=27&P=article2.htm

The two tests you describe could possibly be sufficient, but it depends very much if your model is loaded simply or not, or whether the geometry itself creates complex loading not covered by the test results and curve fit.

Cheers,

-- drej --


------------
See faq569-1083 for details on how to make best use of Eng-Tips.com

 

[cloche]

Well done, Drej!

I'd add three more points:
1- the "best" constitutive law is not univocally defined; that is, some materials are best described with M-R, some others with Yeoh, etc.
Moreover, the best capture of the behaviour of rubbers is achieved with "mixed mode" laws that can not be implemented in Ansys without programming user-defined routines, so some approximation must be taken into account anyway.
2- the "precision" of the curve-fit based on one test type only is as better as closer the real stress state of the part is to the one used in the test. With regard to this, an analysis of a rubber seal I recently made (NBR-70...!) using a 2-coeffs M-R law (did not have more precise data) showed a 7% spread with respect to the physical tests made on a prototype, which is not so bad in the end...
3- be sure to express the data for curve-fit in the form "engineering stress - engineering strain" and not "true stress - true strain"; at least, I remember it like that, maybe it's the opposite but you can easily verify in the help (in my case, I had the coeffs, not the sigma-epsilon data)

Saluti

Claudio

 

[Drej]

The test data is collected as engineering stress & strain, and hence engineering stress and engineering strain are used for the curve fitting. The exception is volumetric data where true stress is required (this is different from curve-fitting for metal plasticity, where collected data is converted to true stress and true strain). Remember also that collected data may need to be adjusted to account for effects such as hysteresis and stress-softening behaviour.

Cheers,

-- drej --


------------
See faq569-1083 for details on how to make best use of Eng-Tips.com