Simple convergence test

[StructuresDan]

Hi,

I have modelled a cantilever beam in Workbench 10mm thick, 100mm long and 30mm wide as shown in the attached picture. I am applying a force at the tip of the cantilever of 100N downwards.

I understand that if I don't have a fillet the analysis will result in a stress singularity at the root of the beam and the model will not converge (which I confirmed with ANSYS WB). So, I added a 5mm fillet as shown expecting the model to converge.

Does anyone have any idea why I cannot get the bending stresses to converge? It appears that the edge between the fillet radius and the cantilever is causing some problems as the attached picture shows.

Many thanks,

Dan

 

[Drej]

Presuming this is a fully linear analysis, it could be that you just haven't refined the mesh enough as yet to achieve convergence for the variable of stress you've chosen for convergence. Your model looks like a pure tet mesh, which can make the structure behave overly stiff unless you have enough elements to cover the strain gradient, and hence by throwing more tets at the area under scrutiny may help to provide convergence. Make sure you're not using first order tets, as these are notoriously stiff in structural models.

Also check displacement convergence, which should converge fairly easily, and will be linear. Note that stress is a non-linear convergence variable. Note also that this is not bending stress you are looking at, it is equivalent ie von Mises stress, which is a compound of all of the stress components at a point. It's worth checking other stress components to verify the VM convergence behaviour such as directional stresses at the surface (ie pure bending stress). I would also do a simple hand calc to verify the stresses and hence your model.


------------
See faq569-1083 for details on how to make best use of Eng-Tips.com