Hertz Contact Stresses

[bigphikl]

Hi,

I have been trying to model a hemi-cylinder in contact with a brick in 2D, as in the hertz solution for contact stresses.

I fix the bottom of the brick from moving and i apply a pressure to the top of the hemi-cylinder.

But I keep getting two peak tensile stresses at the edge of the contact area.

I am new to Abaqus, and feel as if I am missing something.

Does anybody know how I can get rid of these peaks and get the proper hertz solution.


Any help is much appreciated,
Philip.

 

[corus]

A picture of the deformation and stresses might be useful.

corus

 

[bigphikl]

I've attached a picture of the von Mises Stresses that I'm getting.

The top object is the hemi-cylinder, and the bottom is the brick.

Thanks for your help.

 

[bigphikl]

Here's a picture of the deformation scaled by 100.


If you need any other of the stresses, just let me know.

 

[corus]

It appears to be related to the mesh. Use symmetry conditions on the mid-line and halve your model, and then refine the mesh at the contact point. You can also define your contact condition twice, reversing master and slave surfaces, so that no surface can pass through another.

corus

 

[bigphikl]

Thanks for your help.

I'll try halving my model and applying symmetry conditions and hopefully that'll help. I'll let you know.

I really don't think it's a mesh problem though, as the mesh is very fine at the contact, and it is perfect quads.

The cylinder (slave) mesh is far more refined than the brick (master), so I don't think the surface are passing into each other. I am not particularly concerned with the stresses in the brick.

I'll let you know how I get on,
Thanks

 

[bobpozzi]

Hi there,
Is there anyone who can help in using abaqus for a roller bearing simulation?
I used a bearing portion with one ball and inner outer collar segments; with increased mesh size I got non converging results.
Any idea?
Many thanks
Bob

 

[corus]

Well.. the deformed shape of the brick seems a little coarse to me. You may have hour glassing too, which I think you can get rid of by not using the reduced integration element, or is it vice versa?

corus

 

[Noirom84]

Hi!
Maybe you can attached cae file. I've made such type of analysis. You should get such stress distribution:

http://www.youtube.com/watch?v=OfVRTKDoQHg

There are 2 main theories about "hard contact"
1. Hertz theory
2. Bielajew's theory

In a nutshell...First says that maximum stress value should be in area of contact. Second says that maximum stress values are a little bit shifted from contact area.

Regards

 

[bigphikl]

I found in the abaqus benchmark solution for hertz contact solution in the interactive manual at:

http://boston.we.mtu.edu:2080/v6.7/books/bmk/default.htm

I downloaded the first 2D input file and then ran it, but I didn't get the same solution as the one shown, does anyone know why?