Coulomb friction: Theory vs Abaqus

[drennon236]

I want to find the equivalent friction force in the first figure, in the two cylinder that are in contact with each other in the second picture. Is it correct to look at S12 in Abaqus? Should I look at something else since S12 is shear (MPa) and units are different than the friction force (N)? I wanna do this to make sure the friction force I get in my cylinders is according to theory.

 

[FEA way]

You can use specialized contact output for that. Check variables CSHEAR (frictional shear stress) and CSHEARF (frictional shear force).

 

[drennon236]

Do you know how to get these specialized contact outputs? Can I get it for Dynamic, Implicit step? And C3DR8 elements?

 

[FEA way]

They can be found in Field Output Requests —> Contact. Just select CSTRESS (CSHEAR is part of it) and CFORCE (CSHEARF is part of it).

You can also select CSTRESS in History Output Requests but CFORCE is not available there (total force due to frictional stress CFS can be used instead).

Sure, they can be used in this type of analysis and with these elements.

 

[drennon236]

FEA way; how would you suggest going about comparing CSHEAR/CSHEARF with analytical results? I want to make sure the friction force/stress I get in the cylinders is according to theory, but I am not sure which theory to compare it with.

 

[FEA way]

Create a contour plot of CSHEARF but make sure that you select proper component of this variable (there are three available: CSHEARF1, CSHEARF2 and CSHEARF3, the choice depends on the orientation of the model in current coordinate system). In addition you can plot CSHEARF —> Magnitude (resultant of these 3 vectors).

These variables can be compared with analytical value of the force of friction (calculated based on applied load and coefficient of friction).

 

[drennon236]

The applied load; if I apply pressure to the cylinders, will the load then be pressure multiplied with the area?

 

[FEA way]

Yes, you have to multiply applied pressure by the area to get force for these friction calculations.

 

[drennon236]

I read online that the change from static to dynamic friction will probably not be close to analytical, but should quickly approach it once the movement is started. Would you happen to know why? And can I simulate movement by having a displacement boundary condition.

 

[FEA way]

There are separate coefficients for static and kinetic friction while in Abaqus you normally define single friction coefficient. However it’s possible to use more advanced friction models (check the Static-Kinetic Exponential Decay option).

Sure, you can use displacement control instead of force control. In fact, the former can significantly aid convergence in analyses with contact.

 

[drennon236]

Force control would be applying normal pressure? For example hydrostatic pressure to the cylinders to activate coulomb friction between the two cylinder. They are supposed to be in close contact. And then make sure the CSHEARF output somewhat match the analytical ?

What would then displacement control be? Displacing the outer cylinder for example vertically downwards? - because then there is going to be minimum amount of CSHEARF, because there is no normal force to activate the coulomb friction.

 

[FEA way]

Generally speaking, displacement control is preferred method in FEA but of course it’s not always possible. It all depends on the goal of the analysis and adopted loading scenario. If you only want to find the friction force then you will most likely need pressure to establish contact (there are other ways to achieve this though, we discussed some of them in your previous thread). Then you will have to cause some slip between these two surfaces. You can use prescribed displacement in vertical direction for this purpose. Or enable gravity.

There is a comprehensive document about modeling of grouted joints that you may find useful. Search for "DNVGL-RP-0419 Analysis of grouted connections using the finite element method"

 

[drennon236]

Thank you.