Ansys 9.0 New Solid-Shell Element SOLSH190 1

[mcintya]

Does anyone know how to properly use this element?
According the the literature, it is supposed to help with modelling contact between shells and solids. Can you use this element replace the shell elements in a model so that you will no longer have a gap between adjoining parts, thus making the contact alot easier to set up? Can you use only one of these elements through the thickness of the part? If not, what would be the advantage of using this element over solid elements?

 

[mcintya]

To answer my own questions:
-The Solsh190 element meshes the same way as you would mesh a solid, which also makes contact easier to set up.
-No, you still need to use at least three elements through the thickness of the shell to get accurate results.
-The benefit over solid elements is that the mesh in the other two dimensions can be coarser (solid elements require the mesh to be finer before the solution starts to converge).

 

[barryodonnell]

This doesnt make sense, surely a single element through the thickness is appropiate due to the in-plane response.

Barry

 

[mcintya]

Thanks for your response, could you clarify what exactly you mean by "in plane response".
I had run a test case in which I compared the results of using one element vs three elements through the thickness of a simple rectangular beam. I then compared these results with hand calculations, the three elements thickness resulted in a much more accurate deflection and stress distribution.
I admit that I am still a beginner when it comes to FEA, but I have been told by others in my company that have considerable experience using other FEA packages that using three elements through the thickness of a shell (when using solid elements) is standard practice.

 

[Drej]

> I admit that I am still a beginner when it comes to FEA, but I have been told by others in my company that have considerable experience using other FEA packages that using three elements through the thickness of a shell (when using solid elements) is standard practice.

Be careful about terminology. If your system is membrane dominated i.e. all of your loads are in plane (like plane strain), then you only need a single element through the "thickness" to accurately capture the resulting behaviour. On the other hand, if you have BENDING, and hence your loads are not in plane, you must capture the resulting stress gradient through the "thickness" of your mesh. Shell elements (like 63, 93) have section points (3 by default ususally) through the thickness, hence you can capture both the membrane and bending behaviour. (This is valid for linear material analyses. If you go non-linear, you may need more section points to capture the non-linear stress gradient.) Solid elements (like 45, 95) by their nature cannot capture the bending behaviour accurately with only one element through thickness - more specifically, because the formulation of the element cannot capture the stress gradient. Hence, be careful about terminology.

Cheers,

-- drej --

 

[mcintya]

Thanks Drej,
That makes sense, most of the problems we analyze are bending dominated (stamped brackets and such) vitually none of the loading is in plane. I guess thats why "standard practice" in our company was to use three elements through the thickness. We also do non-linear analysis to evaluate the behaviour of new designs under FMVSS 225 (A destrutive test) loads before we build a prototype. I am trying to determine whether or not I should use the new solid shell elements that have 3 dof, or continue to use shell elements that have 6 dof.

 

[Drej]

No problem.

> I am trying to determine whether or not I should use the new solid shell elements that have 3 dof, or continue to use shell elements that have 6 dof.

Depends on a few things: if you're more comfortable using the standard 6 DOF shells, then I'd use these, especially if you're going non-linear (the results from the 3 DOF shells may not make immediate sense to you or be more difficult to use initially); on the other hand, if your model is large, you may find that using the 3 DOF elements speed your run times up a bit (that's assuming they are true shells with either 4/8 nodes). It's a huge balancing act to be honest, but simple is ALWAYS best. The best of luck to you.

-- drej --