Learning Forum

[FedeZappa]

Hello everybody,

I apologize if my question is trivial, but I couldn't find any answer to my doubts.

Lately I'm introducing many friction contacts to my simulations, and I'm noticing that there is a discrepancy between the friction coefficient suggested by papers, books, web-sites and the friction coefficient suggested for the numerical simulations.

More precisely the friction coefficient suggested by the books (Steel-Steel, Al-Al, Steel-Al) contacts is way higher than 0.2 (value after that ANSYS warns you that the friction coefficient is high and can cause convergence problem AND value typically used to simulate the friction between dry and clean steel surfaces).

Actually, is also way higher than the cautelative value of 0.1-0.05 suggested here on the forum for dry surfaces. 

So, I would like some clarifications about the selection of the friction coefficient, and maybe some suggestions about the sources or criteria that can be used to select it.
Eventually, if you can suggest me also sources where I can find the friction coefficient in case of lubricated surface it would be great.

Thank you very much for any answer.

[peteroznewman]

Hello FedeZappa,

Engineers doing simulation are sometimes looking to predict a specific result. That is when they should use a value as close to reality as possible. You could calculate friction coefficient from physical measurements on a fixture that directly measures tangential force for a known normal force.

Other times, engineers are doing simulations to predict a worst case situation. For example, to predict the load that could cause a structure to fail, they would use the minimum yield strength for the material. In a physical test, the structure might survive much higher loads because the actual material in that sample has a higher yield strength than the minimum specification for that material.

Friction is different from yield strength, where the minimum is almost always the worst case.  In some cases, the highest possible friction is the worst case, while in other cases, the lowest possible friction is the worst case.  Different responses might require opposite ends of the friction spectrum to expose the worst case for each response. When that happens, two models are solved, one with frictionless contact and another with rough contact.

Actually, is also way higher than the cautelative value of 0.1-0.05 suggested here on the forum for dry surfaces. 

When I translate the Italian "cautelative" to English, it is precautionary. I think in this case, a better translation is conservative. For the case when the worst case is the lowest friction, you can make a good argument that the low end of coefficient of friction for dry surfaces is in the 0.1 to 0.05 range, but as I said above, you could go to frictionless and know you have the worst case.  Some models might not converge with frictionless contact and a small amount of friction can help convergence.

[FedeZappa]

Unluckily I'm looking for a specific result. According to your post I should perform experimental measurements but for me it's technically impossible. Without any further idea or data I will go without specific values.

Anyway your answer was great as always, thank you for the kind help.

 

[ashishchache]

Hello Fedezappa,

               I have also struct with same problem similar to yours, If you have got the answer then will you explain me how do we have to select specific coeffi. of friction when have many values  (or not have any data related to it).