Material Properties for BB-Model

[gerhaher]

Hi everybody!

I do practical orientated research on elastomeric bearings so I´m not so much in the materia theory thing. :)
To calculate the damping of the bearings it is important to consider the hysteretic behavior of the elastomer. I did some first tries with the BB-Model but I miss a tool to determine the material parameters. If I don´t want to care about the time dependency (rate effect, creep...), only the damping and non-linear strain-stress behavior is important to me.

Can anybody tell me how to get appropriate material parameters from given test data (uniaxial, biaxial, pure shear):?:

Which of the parameters of the BB-Model do I have to define and which are not important to me if I only want to consider the hysteretic behavior:?:

Thanks in advance!
Ulrich

[Jorgen]

Hello Ulrich,

The BB-model can handle the hysteretic behavior of elastomers well. Note, however, that the reason for hysteresis in an elastomer is the rate-dependence. That is, the response of elastomers is mainly viscoelastic, not viscoplastic.

There are basically two ways to find the material parameters:
(1) use an automatic tool (http://polymerfem.com/index.php?pageid=umats)
(2) use trial-and-error

Perhaps in your case the easiest solution is to start with a reasonable initial set of material parameters, and then manually adjusting the parameters until you get the results that you like.

- Jorgen

[gerhaher]

Hello Jorgen!

Thanks for your quick reply!

About the note: My experiences in tests on elastomeric bearings but also in simple uniaxial tension tests showed that the time dependency of the material behavior seems to be very few. Test on bearings with a shear deformation of 100% showed nearly the same results for 0,5 Hz and 1,0 Hz for the deformation. Also see test results on a 50 mm test specimen with a difference in the deformation velocity of factor 100 (see file attached). So does the hysteretic behavior arise from a "inner friction" (with no time dependency) or from a viscous behavior like a dashpot? Maybe the velocities in my tests are just too low? (the correlation of the reaction force of a damper and the velocity should be quadratic, I guess?)

About the matlab tool: Could you send me the matlab files including the subroutines for a try?

Triel and error: I tried the material parameters you gave in "PolyUMod_Manual_ABAQUS.pdf" as an example in a simple uniaxial tension model (see inp file). Energy results (see pdf-file): ALLAE and ALLIE are just the same. canging C3D8R to C3D8 made those two energies disappear but still the elements got awfully distorted and still no strain energy appeared. The rest of ALLWK goes into ETOTAL which I interpreted as a missfunktional loss of energy. Could you tell me what is wrong with that model? did I use the material model in the right way? is the material model maybe not installed right? or are the parameteres just irrational guesses and do not describe a real material?

Thanks in advance for your help!
Ulrich

[gerhaher]

Hello Jorgen!

One more question about guessing the parameters, described in your paper:
I tried to get the 9 parameters like that:
1) mu = C_R(A) (as described in appendix A)
2) lambdaL ???
3) kappa = estimated as 2000 MPa or kappa = 500 x C_R(A) (as described in appendix A)
4) s = C_R(B) / C_R(A)
5) xi = ???
6) C = ???
7) tauBase = ???
8) m = estimated as 4 (as described in appendix A)
9) tauCut = ???

How can I estimate the missing parameters? :confused:
And what to do with C_1 and C_2?

About the tests in PolyUMod_Manual_ABAQUS.pdf:
The results form abaqus/explicit are often very different to those from abaqus/standard. Especially with the BB-Model most of the result with explicit are quite bad. What does that mean for the usability of the BB-Model with explicit? Do you think I will get appropriate results for the elastomeric bearings?

Thanks for your help!
Ulrich

[Jorgen]

You are asking a lot of questions, and since some of your statements are wrong, I feel the need for a careful answer ;)

- There are some people who believe that hysteresis during cyclic loading is due to "internal friction", i.e. plasticity. And there are some people who believe that the hysteresis is due to viscous dissipation. And then there are some who think that both of these mechanisms can be active. These days I think that most rational researchers fall into the third category. In all of the work that I have done on various types of elastomers it has been clear that the viscous dissipation dominates over the plasticity. In the data file that you presented there was a large amount of Mullins effect, and surprisingly little rate-dependence. This suggests that for that particular type of elastomer (do you know what it is?) "internal friction" may play an important role in addition to the viscous dissipation.

- No, I cannot send the Matlab files. They are only commercially available. However, if you provide your experimental data I might be able to help guide the calibration.

- The reason your input file does not work is that you are not using any hourglass control!! If you add a statement:
*Section Controls, name=..., hourglass=enhanced
the simulation works fine. There is nothing wrong with the model.

- You also need to be carful when you analyze energy output from an explicit simulation that is using mass scaling. Also, you had massive hourglassing in your simulation. In a case like that it makes no sense to try to analyze the energy output.

- Is the material that you are interested in the same as the experimental data set that you provided? If so, you might want to use a model that includes Mullins effect.

- About the material parameters:

mu = shear modulus of the equilibrium response (what do you mean by C_R(A)?)

lambdaL = locking stretch. I am guessing this is not important for your material

s = ratio of shear modulus of network B to the shear modulus of network A

xi = 0.05. You do not have to change that value

tauBase = flow stress (important parameter for the viscous flow)

m = flow stress exponent (important parameter for the viscous flow)

tauCut = 0.01. You do not have to change that value


- No, the values for Abaqus/Explicit are not different from Abaqus/Standard. The difference that you are referring to is an unfortunate miscommunication in the documentation about what was actually done in the various test case. It will be fixed in the next version of the document.

- There is no usability issue with the BB model. As with anything new, it takes some effort to learn how to do it right. I still think that the BB model could be a useful tool for your work.

Now I have request for you: if you have any follow up questions or comments, can you please send them directly to my email.

Thanks,
Jorgen