Hertz contact: stress distribution on inner ring problem.

[Fabi0]

Hi,

it's second time i post because of this problem: sorry, but i can't solve it.

I've model a single roll between a portion of inner and outer ring: on outer, the distribution is ok (figure one) but on inner isn't ok.

So i tryed with submodeling, to get some more omegeneous mesh element, but nothing get better (figure 3)

Than, i tryed with a modeling of a quarter of the bearing, and the result gets to the opposite: inner ring with nice distribution and outer with a too much linear distribution. (figure 4 & figure 5).

All analisys are setted with:
small-sliding
hard-contact
frictionless
Surface to Surface contact

My doubt are about boundary condition and contact properties....I costrained inner ring with pin all over lateral surface and bottom surface (everything except surface on top, where contact occours), while outer and roll are costrained to move on onlye one of the three axes (the one that go from outer to inner ring by roll axis).

Why in the analisys of the quarter, i got a specular solution?

I'me trying now with penalty non linear mode and with differ boundary condion on inner ring. May be usefull to try with a model that take into consideration a bigger portion of inner ring? I bleave not, because hertz contact take place in a small area....but i'd like to have a your opinion.

Considering that every "test" take me about 4 machin's hours, every help will be very appreciated.

FB

 

[Fabi0]

Sorry but...upup!! I really can't solve it by myself

FB

 

[corus]

For smooth results try and have the mesh the same either side of the contact surfaces. You have some scattering where nodes have slipped through the other surface.

Use the symmetry between each of the bearings to reduce the model size down. There's no need to do a quarter model.

Tata

 

[Fabi0]

Thx for your reply Corus!

But i don't undestand what do u mean with "try to have the mesh the same".....i mean, with submodeling i've got a very regular mesh element shape, how can i do better?
Also, abaqus help teach that slave mesh must be more refined than master surface....maybe you're saying that my slave mesh is too big?

FB

 

[corus]

No, Abaqus doesn't say that the slave surface must be more refined, it says that the more refined mesh should be the slave surface. A subtle difference. You are wanting to achieve smooth contours so I'm suggesting that you make the mesh size the same on both sides of the contact zone. In that case it wouldn't matter which was the slave or master surface.

Tata