Wrong contact stress calculation with 3D model, good with 2D

[Freemesh]

Hello,

I am a Ideas 9m2 user and I want to do a very simple Hertz Stresses calculation with the following parts (made of generic isotropic steel) :

a sphere (radius = 5 mm) is contacting a block, the block is blocked, a load of 60N is applied to the sphere. It is a classical problem.

A 2D axisymmetric model gives the exact values of contact stress, contact surface area and sphere displacement compared to analytical calculations.

With 3D model, FE stress is half the analytical stress, contact surface area and displacement are completely wrong. Elements size around contact location is close to software tolerances, I can't mesh finer.

Does anybody faces this kind of trouble? Do I do an enormous mistake or is there problem or bug with I-deas?

Thanks in advance.

 

[feadude]

This type of contact is nonlinear because contact area increases as load is increased. Are you doing a linear or nonlinear analysis?

good57morning@netzero.com

 

[Freemesh]

In 2D axisymmetric or 3D analysis, I am using linear static boundary conditions.

 

[feadude]

Are you comparing your results to :
Po=0.388[(PE^2)/(R^2)]^(1/3)??
This is from "advanced mechanics of materials" by Cook and Young page 167

good57morning@netzero.com

 

[Freemesh]

Yes I am.

With R = 5mm, E = 196000MPa and P = 60 N, I should have 1750 MPa.

The 2D model gives 1790 MPa and the 3D model around 800 MPa.

Two possibilities : I do not click on the right button or there is a bug.

 

[feadude]

Can you do a nonlinear analysis for your solid model?
What type and order of solid elements are you using?


good57morning@netzero.com

 

[Freemesh]

My solid elements are parabolic. I did two kind of meshing (free and mapped). Results are identical.

I do the non-linear analysis.