Describing through-thickness stress state in a thin-shell structure

[mp2333]

Hi everyone,

First post here, so let me know if anything in this post is not as it should be.

I am currently analyzing a thin-shell structure using ANSYS with 2 layers of SOLSH190 (solid-shell) elements as shown below. The thin-shell has the four circular corners that are fully fixed, and is subjected to uniform loading (applied at each top node). The analysis is geometrically nonlinear. I performed a convergence study and benchmarked the model against another model with shell (SHELL181) elements, so I am fairly confident in my results.

Since I have 2 layers of elements, that makes 3 nodes through the thickness. I am trying to back-calculate the stresses in the middle node from the stresses in the top and bottom nodes, but I am struggling to do so. I assumed that the middle node stresses would be equal to the average of the top and bottom stresses but that is not the case.

Is there something I am missing here, or am I wrong in thinking that I can back-calculate the stresses in the middle node ?

Thanks for the help

 

[FEA way]

Try with a simple model first - for example, a cantilever plate behaving like a beam and analyzed using different element types (solid, shell, solid-shell). This way you will see how the stresses distribute and you will be able to compare them with the analytical solution. Keep in mind that in bending you will see a neutral axis with zero bending stress in the middle and that you should check the stress components in addition to von Mises stress. Shell elements can also detect stresses in the through-thcikness direction thanks to additional integration points.

 

[NRP99]

Did you check help documentation for SOLSH190. Link

KEYOPT(8)
Storage of layer data:

0 --
For multi-layer elements, store data for bottom of bottom layer and top of top layer. For single-layer elements, store data for TOP and BOTTOM. (Default)

1 --
For multilayer elements, store data for top and bottom for all layers. (Before using this option, be aware that the amount of data involved can be very large.)

 

[mp2333]

Thank you for the replies.

FEA way - I already did simpler tests as you suggested, but I struggled to make some sense of it. However, I did not look at the analytical solution, which is indeed where I should have started, thank you for the advice! Something interesting though is that I performed the same analysis but with shell elements (SHELL181), and the result was as my assumption: the middle stresses at each node were equal to the average of the top and bottom stresses.

NRP99 - I did look at the documentation, and I am using the default setting for the storage of layer data. Since I am using two layers of elements (and not one element of 2 layers) I have the values for the middle layer. However, I am trying to see if I can back-calculate the middle stresses from the top and bottom ones.