Uniaxial compression test

[Jay]

Dear Dr. Bergstrom,

I have a question about your Ph. D thesis.
You use test data from uniaxial compression, plane strain, and shear tests.
However, compression test is said to be incorrect for that the boundary condition is not easy to be controlled. The friction at boundary greatly influences the test results. Do you agree with saying?
Also, why didn't you use uniaxial tension test?
Does your BB model fit well in tension dominated structure?

[Jorgen]

Hello Jay,

Thanks for your interest. You are right that compression tests are somewhat tricky to perform - you need to make sure that you eliminate friction with the loading machine. In my thesis work I used sheets of teflon to eliminate this problem. I know that this approached worked since I saw no barrelling of the test specimens during the compression tests.

I could have used tension experiments in addition to, or to replace the compression experiments. At that time I had easier access to compression specimens than to tension experiments, and it was easier to accurately measure the applied deformation (strain) in compression than in tension. Also, I wanted to use a temperature chamber to explore the effect of temperature on the response, and this temperature chamber only worked on our compression test machines. In summary, if was for pure practical and experimental reasons that I performed mostly compression experiments and not tension experiments.

I have since performed and worked with tension experiments. It turns out, and this is not surprising, that the BB-model work equally well in tension as in compression. In fact, the BB-model typically works for both tension and compression, even if it is only calibrated in one of these modes.

Jorgen

[Jay]

Thank you for your explanation.
Does BB model predict tension behavior from the result of compression test? or a tension test is required?

[Jorgen]

Hello Jay,

Since the BB-model is micromechanism based, it does a very good job of capturing tension, compression, and multiaxial loading histories, using one set of material parmateres. In fact, I consider this to be one of the key benifits of this model compared to some other more phenomenological models. So yes, if you calibrate the model is compression you can still use the same parameters to predict tension, and vice versa.

Jorgen

[Jay]

Did you put anything (like oil) between the specimen and teflon sheets?

[Jorgen]

Hello Jay,

Good question. I ran some tests with a liquid soap between the Teflon sheets and the test specimen. In other experiments I did not add any additional lubrication. I got very similar results for these two groups of specimens, and I did not see any evidence of barelling of the specimens during the testing. Based on this, I think that interface friction did not play an important role in the test results for my experiments. The frictional influence might be more important for large samples, or at high temperature.

Jorgen