Showing results for 
Search instead for 
Did you mean: 

Thickened L-Model to shell/beam model


Hi all,

I am currently experimenting how to reproduce the result of a solid calculation using shell elements.

I've tried and made several examples of simple L-Connections between two shells, a T-Joint model (like the one that Blas Molero did for the Glue connection), and a simple plate model where I compared methods of joining, like e.g. glueing, stitching and RBE.


My next example case would be a "thickened" L- connection (see attached figure).

Here, I am struggling really much to develop a method to reproduce the solid results using shells.

My approaches/comparisons so far:

  • Partition the shells to 4 segments (2 * low thickness, 2* higher thickness, see different colors)
  • Introduce two "thick" shells
  • Introduce only one thick shell (one time the horizontal sheet segment, one time the vertical sheet segment)
  • Different connections (stitching, RBE, glueing)

None of these experiment results have lead to a way that could represent the structure very well.

My next idea was to introduce a beam in the corner but I am not sure if this is a good way.


Maybe someone of you has encountered a similar problem and/or can help me out or guide me further?Smiley Happy


(Current simulation settings: fixed constrained on the right face/edge and force application in X-Direction on the top left edge)


Best regards,




Re: Thickened L-Model to shell/beam model

Gears Phenom Gears Phenom
Gears Phenom

Not sure what you want to say "to reproduce same results". Displacements, stresses? I mean: if you are looking for displacements to have to be able to reproduce the same stiffness.  Easier than rerpoduce stresses (above all peak stresses).Looking at your model the thick "block".... I would say that shell theory (even thick shell) is not suitable for it. Besides, shell offset must be included.....

Re: Thickened L-Model to shell/beam model


Thank you @jon_morga for your answer.


For the first step, I want to take a look at the displacements, meaning the stiffness behaviour. However, I later want to trim the model in that way, so that the dynamic behaviour is approximately the same.

Yes, I guessed that the theory is not suitable for it. Nevertheless, I still try to find a solution which could be used for trimming the model in the way I would like.


Yes I've already thought about the aspect of the shell offset and implemented in some of my testcases. However, the shell offset did not lead to big deviations in my example case.

Re: Thickened L-Model to shell/beam model

Gears Phenom Gears Phenom
Gears Phenom

Ok.the shell offset could lead to changes in stress values. Not a lot in the displacements. Neverthless, solid approach is the most accurate approach (normally, if the mesh, number elements through thickness ....are correct) as shell is only a mathematical simplification. Do not htink the beam approach, as you said, will improve what you are doing. Even it make it  worse. Hope you find what you want.