Tetra and tria element 1

[izax1]

After using FE methods for 15 years, I have been inactive from FE-analysis for some years, but are now again invovled in a composite material analysis project of a marine structure. I remember when using MSC/NASTRAN some years ago, a lot of scepticism arose from using tet4 and even tet10 elements. (Today I am using NX/NASTRAN) I have recieved a FE model from a customer, with almost only Tet10 and tria6 elements, and I have noticed that this is not unusual today to use these elements. My question is: Are tet and tria now accepted because of convenience or is the quality of the elements so much improved the last 10 years that stress results are acceptable? Or are the same elements used with just denser meshes?
Thanks for any input. Especially from long time FE specialists.

 

[rb1957]

tet4s (and tri3s) bad, very bad ...

tet10s (and tri6) good

Quando Omni Flunkus Moritati

 

[corus]

In general I'd avoid them as the results at high stress gradients tend to be irregular, ie. around the mid-side nodes you'll find the results stagger about. I'm not sure why that is but the results extrapolated to the mid-side node are a little 'stiff' for some reason. This doesn't give you much confidence in the results at these positions. Unless the geometry is extraordinary difficult, it's much better to partition the strucutre and aim for brick, or four sided elements in 2D. These days it's just lazy software that allows someone to just point and click and hey presto you have a mesh and pretty pictures. In addition, you need a lot more elements and nodes (nad so memory) to describe the model than you would need for brick type elements.

 

[MeshNinja]

Fully agree with previous posts, I would not touch tet4s under any circumstances.

It takes a bit of time to prepare geometry but it's worth doing in order to generate a brick rather than tet10 mesh.

 

[izax1]

Thanks guys. I can see that you are also concerned. And I see that almost any CAD provider is marketing the quick fix meshing solution. 3D geometry and tet meshing. Hmmm.