Assigning Thickness to 2-D Shell Elements

[struclearner]

Dears,

When the shell elements are created at the mid surface of a plate, the thickness value of the plate, assigned to shell elements goes to both ways, i.e. 1/2 value is assigned to either side of the mid surface. When the shells elements are placed on the solids elements, as a surface coating to extract the stresses at the outer surface, or for quick evaluation of the structure thickness sensitivity upon the stress/strength, the thickness assigned to shell elements laying on the solid elements should go on the outside direction only.
Please comment if my understanding is correct and how we can assign the thickness value going in one direction only in a FE program/solver.

Thanks for the help.

 

[ESPcomposites]

Look into what is called a "shell offset".

Brian

http://www.espcomposites.com

 

[SWComposites]

Never use shell offsets; nothing but trouble/rubbish.

If using shells on solid surfaces, the thickness should be very small, so just center thickness on the midplane.

 

[ESPcomposites]

I have run across a few problems where the position of the shell can have a significant affect to the overall section properties (general applications beyond just sandwich structures with thin facesheets). For such a case, it may not be sufficiently accurate to place the shell at the solid's face. There are few ways to do this without a shell offset, but with a shell offset you can easily adjust the shell thickness (great for parametric/sensitivity studies and also if the shell thickness is varying). I have demonstrated excellent correlation to experimental data for deflection and natural frequency using shell offsets (provided they are used within reason). Placing the shell at the solid's face would have led to an unacceptable amount of error.

If you do use them, create some simple test case models to ensure you are getting the expected input/output (and make sure the element normals are pointing in the correct direction w.r.t. how the offset is defined). The downside is that offsets are easy to do incorrectly and difficult to check/debug (different solvers also implement them in different ways, which can further increase the potential for improper use). So if your problems allows for you to avoid them, then great...but a shell offset is one way of answering the original question as stated.

Brian

http://www.espcomposites.com

 

[goeasyon]

The mid-surface assumption seems to hold for most of the shell theories.

https://welsim.com

 

[rb1957]

whether your shells are centered about the mid-surface or the OML would matter to a design model but not to a stress FEA model (negligible change in stiffness).

Using the OML may be convenient, taking surfaces from the design model, and would be valid without off-setting the thickness.
If you Really wanted to you could offset the surface t/2 but that'll be a pile of hurt in all but the simplest geometries.

If using 3D design model, then you can import the solid and extract the mid-surface (if you want to fuss it that much) or use the OML surface.

another day in paradise, or is paradise one day closer ?

 

[ESPcomposites]

"whether your shells are centered about the mid-surface or the OML would matter to a design model but not to a stress FEA model (negligible change in stiffness)."

For many applications and objectives this is true, but not all. Simple hand calcs will give you an idea of how good the assumption is. You can compare this to your threshold of acceptability and decide from there.

Brian

http://www.espcomposites.com