Hertzian contact, path dependency and partial-slip

[Stiffie]

Hello,

I'm working on simulating hertzian (line) contact with very small (micrometer scale) sliding motion typically found in fretting applications. In such applications, a central stick zone is seen, whereas at a certain distance towards the contact edges, sliding will occur, giving rise to singularities. A very fine mesh is indeed important, and using Lagrangian multipliers to get accurately describe the partial slip situation.

Now, most fretting studies follows the "Cattaneo-Mindlin case" where the normal load is applied first, then held constant whilst a tangential force is applied cyclicly. Abaqus gladly reports well-behaving stresses and slip for this case. However, when normal and tangential displacement is applied proportionally, wierd stuff is happening in the contact interface, and I'm not quite sure how to interpret these. Reading through the details of Abaqus Contact algorithms in the manual is not particulary enlightening.

Can someone help me in the direction of understanding these effects? See the attatchments. I've included plots of slip, shear and pressure for the case of sequent application of loads, sequent application of displacements and simultaneous application of displacements.

Any insights, hints, reference to papers or anything would be most helpful!

Thanks

 

[FEA way]

The reason to apply normal displacement first, hold it and then activate tangential displacement is that it's highly recommended to initialize contact in the first step so that the surfaces touch each other with bigger contact area. This may be very important if you have some convergence issues (displacement control itself helps a lot in this matter too). However one should remember that such situation is different than the one when both displacements are prescribed at the same time. That's because in case of the first approach the sliding will start when contact is already fully initialized while in case of the second approach the sliding will start before the surfaces are in full contact. Thus significant differences in results for these cases can be expected. To interpret the results you can run animation and watch it in slow motion. You should be able see when some stepwise movement happens. Contact status variable may be useful too.

 

[Stiffie]

Thank you for your reply! Yes I expected quite different results since the contact area and friction force will not be the same for the two load cases.

I'd think the stresses are physical, but i'm struggling to interpret the slip patterns (see attached figure here). This pattern seems non-physical, but maybe its a natural oscillation of the slip pattern on the surface governed by the contact normal and tangential stiffnesses? Each peak consists of many elements.

Hmm i wonder what i will get if i run a fast fourier transform on the slip pattern...
Edit: I forgot to mention that in these cases, there is a bulk load applied to the flat specimen aswell. But this shouldn't change much i believe.

(Contact is R25 line contact on flat, element size approx. 0.005, E=70,000)

 

[DanStro]

Could you be seeing the contact algorithm changing the elements that are in contact as the slipping is happening?

 

[Stiffie]

Hi DanStro, thank you for your reply! Much appreciated.
What change exactly are you thinking of? In the contact algorithm itself or in the nodal displacements?

 

[DanStro]

Both, sort of. When parts that are in contact start sliding a 'large' distance the algorithm has to keep track of which nodes are in contact, which will likely change if the sliding distance is large when compared to the element size. I've found that when this large sliding happens there are often these kind of jagged types of plots. Look up large sliding contact in the help documents to get a good description of how the algorithm works.

 

[Stiffie]

Ah yes, good point. I will make a deep dive into the solver details about contact node tracking and large sliding.
Also, thinking about making some parameter studies to visualize the effects of friction, element size, sliding distance etc.