Simulation solves with solid mesh but not shell mesh

Hello,

I am working on a static simulation of a formed sheet metal body that is roughly a very large, thin C-channel with flat support ribs placed at intervals inside the C-channel. 

The model is an assembly composed of individual, single-body parts.

The sheet metal body was made using the SolidWorks sheet metal tools and automatically meshes as 2D shells.  The ribs are regular solid bodies, so by default they mesh as solids. 

I made a copy of the simulation in which I converted the ribs to mesh with shells. 

When I run the simulation using the solid ribs, it solves and produces reasonable results.  However, when I initially ran the shell rib version, where the only difference was the type of elements used in the ribs, the study failed, claiming that I had insufficient fixtures in the model.  I am currently trying to debug the shell rib version.

Is there something I am missing that would make the simulation fail?

Thanks,

Katie P.

SolidworksSimulation
Answers 6
SF
SF 2015-08-14 Best Answer

Hi Kathryn,

Have you defined BONDED contact sets between the rib and the rest of the body while using shell mesh?

It could be that the rib is acting as a free body, hence it is not properly restrained leading to that error.

Regards,

David

JS
JS 2015-08-14

Yes, indeed shells and solids must be bonded with a local contact set

If you turn on 'soft springs' you will see that the global bond between the shells and solids fails. The large displacement of the shelled parts shows that.

Don't forget to turn it off again, or else your results are incorrect.

Janko Stellaard

Cadmes BV

The Netherlands

SF
SF 2015-08-14

Hi David and Janko,

Thank you for the advice.

Yes, each rib is bonded using an individual component contact.  I am not using global bonding.  Should I be using contact sets instead of component contacts?

When I added soft springs to the model, the model still would not run.

When the model failed with the soft springs, I activated the non-touching faces option in the bonded contacts dialog with a 3mm maximum gap, which is about 0.5 mm larger than the gap between the midsurface of the body shell and the edge of the rib.  I thought that the gap between the body shell and the ribs might be causing the software to fail to detect the contact.  After making this change, only one of the ribs appeared not to be bonded, dropping through the model about 8 km.  All of the individual component contacts are identical, so I do not know why only one appears to have failed.

Also, why would component contacts work when the ribs were solid, but not when they were shells if the geometry was identical?

Thanks again,

Kathryn

SF
SF 2015-08-14

I should add that I am using SolidWorks 2015 x64 SP1.  My company has two licenses for SW Simulation, one Professional, and one Premium, and I pull whichever one is available. 

Workstation specs:

Windows 7 x64 Professional SP1

Intel Core i5-3470 CPU @ 3.20 GHz

20 GB RAM

Intel HD Graphics 2500

JS
JS 2015-08-14

Don't use a component contact, but a local contact set on each individual bond, so edge to face.

If after that you still have troubles, then go to properties and set 'incompatible bonding options' to 'more accurate', there has been some buggines on the bonded contact lately.

And if all that doesn't work, try to save the model as a parasolid and setup your simulation again. But I believe that shouldn't be necessary.

Janko Stellaard

Cadmes BV

The Netherlands

SF
SF 2015-08-14

Thank you Janko,

I revised the model to have shell edge - solid/shell face pairs for the body shell x rib contacts.

It is working now.

Best regards,

Kathryn