Problem with the friction model in an Abaqus simulation

[mkamrani]

Hello,

I have faced a weird problem and would really appreciate any comments. Assume a 2D model e.g. an axisymmetric model, meshed with quad, linear, and reduced integration elements.

As you now if a model contains contact, we will get "CSHEAR1" and "CSHEAR2" in outputs. Assuming that the contact surface is and remains flat and horizontal, one can expect S12 (shear component of stress tensor in outputs) on the contact surface to be equal to CSHEAR1 or CSHEAR2 (shear stress in the outputs which is based on the friction model).

But, my experience shows that if the layer of elements on the contact surface are a little distorted (after some level of deformation applied to the model), these two can be different. I understand that S12 is calculated on the integration points and then interpolated to the nodal points, but still I expect S12 and CSHEAR1 to be equal.

I have tried different friction models and different element types and different mesh distributions and densities, but this problem is always there.

I would appreciate any comments or ideas regarding the reason behind this or how to resolve this issue.

 

[FEAnalyst]

One thing to keep in mind regarding CSHEAR output variable is that Abaqus calculates it with respect to local tangent directions on the surfaces (CTANDIR). They are not always the same as global directions and can rotate if NLGEOM is on. It's also important how CSHEAR is actually calculated. Abaqus evaluates it at each constraint point as a scalar product of this variable's vector and local tangent direction. Due to these facts I think that CSHEAR components don't always correspond to regular shear stress components.

 

[mkamrani]

thanks for the response. I do appreciate it. I am wondering if there is any way to resolve this problem. I really need shear stress e.g. S12 be equal to CSHEAR.
thanks,

 

[FEAnalyst]

If you plot CTANDIR you might be able to adjust shear stress output to correlate with CSHEAR. What I mean is that you can create a user coordinate system and transform results (including stress values) to this system. It might be also easier with NLGEOM=Off.

 

[mkamrani]

Thanks for your help

 

[Jason LIN]

Thanks for your help

Sir,did you solve this problem? Can you kindly explain it in detail?I am facing this problem as you, much thanks to you.

 

[berkeman]

Welcome to PF.

Sir,did you solve this problem? Can you kindly explain it in detail?I am facing this problem as you, much thanks to you.

Can you show your work so far please? Thanks.