Comparison between shell and solid elements

General Questions
1

Attached are two models identical in geometry, constraints, and loads. A static analysis and a buckling analysis were performed on the two models (shell and solid).

The results obtained are comparable for:

_ Displacements

_ Von Mises Stresses

The results obtained are comparable for:

_ Displacements

_ Von Mises Stresses

The results are not comparable for:

_ Buckling Analysis;

very different values ​​are obtained

_ Values ​​of the constraint reactions, particularly given that the structures are symmetrical in both geometry and loads, the constraint reactions in the X (RF1) and Y (RF2) directions are zero only for the solid model (correct solution) but non-zero in the shell model (incorrect solution).

I can’t understand why.

Thank you for your replies.

A.

Link: Unique Download Link | WeTransfer

2

I haven’t checked your model yet, but are you aware of the overconstraint issues when fixing rotational DOFs of shell elements ? Especially the drilling DOF thing: Inconsistent result with CalculiX on a shell model of a plate in bending · Issue #64 · Dhondtguido/CalculiX · GitHub and Contact of two shell pipes - no convergence in newer ccx releases - CalculiX (official versions are on www.calculix.de, the official GitHub repository is at https://github.com/Dhondtguido/CalculiX).

There are also problems when using rigid body constraint with shells - this should be avoided and one way is to extrude at least part of the model (the one involved in RB constraint) to solid using the Thicken Shell Mesh tool in PrePoMax.

3

In the shell model, I have removed the rotational constraints at the base, but the problem is still there. How do you suggest to move forward? Many thanks.

4

For some reason, in the shell model the loads for the buckling step (2,1) do not produce the same result as the static step (1,1). If the static step is deactivated and only the buckling step is run the results are consistent, however not identical, in both models, solid and shell.

5

This problem was solved at a smaller scale in the forum some time ago. FEAnalyst was preparing his nonlinear buckling of a cylindrical shell video at that time. You are probably facing some issue with the BC or you are comparing different type of elements.

By the way , unless you are planning to send your tank to the space , gravity loads (including hydrostatic) do not scale in a Buckling analysis.

6

You don’t necessarily have to remove all of them. You should particularly watch out for the drilling DOF and you can try using cylindrical coordinate system fir the BCs as mentioned in the referenced CalculiX forum thread.

The question to the OP is if the loads have to be the same in both steps. Usually, the first step only introduces preload and includes so-called dead loads such as gravity while buckling step only includes live (operational) loads such as pressure/force. Sometimes, a slightly different approach is used to introduce load imperfections. This thread is a good example: Issues with linear buckling analysis following ASME PTB-2022 example problem