Linear analysis and mesh refinement

[transall]

Hello,

I am doing a linear elastic analysis. My method usual is to create a free mesh in order to locate the high stress area and then create mapped meshes in order to have smallest element in these high stress areas. But in my case I'am analysing the maximum Von Mises equivalent stress and it didn't converged as I refine the mesh. The high is around a circle and it is a very tiny area, one half of the circle has 0 DOF and the other is free in order to simulate the contact with a bolt when a force is acting on one side of the circle. Could you explain me why the eqivalnet stress doesn't converge?

Many thanks for your help,

Mickaël

 

[Transient1]

From what you have stated:

You have a fastener hole that is constrained along half of the interior face (fixed constraint). When you say the solution does not converge, do you mean that

(A) The solver does not complete.

or

(B) No matter how fine the mesh gets, the peak von mises stress at the hole increases or moves around and does not converge upon a steady value.

If it is B, the answer has to do with the method of constraint not accurately representing a bolted fastener. In reality, the faces of the hole are not fixed as in the simulation. It is the pressure from the fastener that fixes the hole. If you search

http://www.ansys.net

for bolts, you'll find a few interesting articles. The stresses are artificial, not to mention they increase dramatically due to the stress concentrations a hole geoemtry gives. When constraining fasteners in this manner, I wouldn't trust the stresses. However, the reaction loads should be accurate enough. Also, monitor how the principal stresses change, you may find the principal stresses in one direction level off while the principal stresses in another direction are varying with each refinement.

 

[transall]

thank you for your help