Convergence Issues - Shell Assembly w/contacts

[ryano123]

I am having some interesting results for the convergence of stresses in an assembly modeled with shell elements.

Roughly speaking, the model is constructed as follows:
The model has a remote load applied to one component. Both bonded and no penetration contacts are used to link individual components. The model is fully constrained with the combination of component contacts and a fixed support applied to a face of one of the components.

The convergence results are interesting for two reasons:

1. Stresses seem to decrease as mesh size is increased.

2. Some oscillation in stress is evident as mesh size is increased.

Any thoughts on what might be causing this response?

Convergence Data is attached.

Thanks,
Ryano123

 

[ryano123]

Noticed an error in the spread sheet...

The % element/node differnece is actually showing the change in element stress between runs. The actual data is:

Study % element/node difference
1 3
2 1
3 1
4 1
5 0

 

[kellnerp]

When you model contact it is very easy to create singularities. Singularities occur when the sharp edge of one component comes into contact with another component face. The tip off is that as you decrease the mesh size the stresses go up. This is because, at a singularity, the stress is infinite so reducing the mesh size better captures the infinite stress.

You may need to round sharp edges to help get rid of this or play with the spring constants on the contact elements.


TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."

 

[ryano123]

In my case the stress is decreasing with increased mesh size. Can this behavior also be attributed to singularities?

 

[kellnerp]

When you say increasing mesh size you mean decreasing the number of elements in the model don't you?

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."

 

[ryano123]

Sorry for the confusion, my FEA vocabulary is still developing. I should have said for increased mesh density, i.e. smaller elements, stresses decrease.

In the attached spreadsheet I show von Mises stress probed from a point on the end of a strategically placed split line. I get the same shaped curves when I compare strain energy density between studies.

 

[kellnerp]

Since you have this strategically placed split line, plot the stress along the line for the different mesh sizes and see if the stresses are varying smoothly. I am assuming the split line is in the zone of contact.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."