Orthotropic panel vs shell vs beam models

[Anonymous]

Good morning:

I am testing the orthotropic panel option to model a trapezoidal sheet metal member.

For some reason, there are big differences between deformation and stress distribution between the panel and a shell element model. Am I doing something wrong or is there some limitation on the orthotropic rigidity modelization on Robot?

I have also created a beam element and stress results are quite different from the shell FEA model.

Thanks a lot and regards, Rick

[Krzysztof_Wasik · Accepted solution]

Hi @Anonymous 

I have the following remarks to your model. I am comparing results for case 2 Pz =2000.

- when real trapezoidal shell shape is defined supports defined below neutral plane generates significant horizontal forces. It makes trapezoidal shell structure stiffer than orthotropic (orthotropic model does not produce horizontal forces - see screenshot) To compare results define supports with UY released on one side of both shells.

- trapezoidal shell is loaded with higher loads. Load should be defined as projected to obtain results similar to orthotropic shell sum of reactions 2000 on each edge)

When those corrections will be made, deflection in the middle of slab (when we neglect side effects on trapezoidal shell) are similar for case 2  (2.3 vs 2.5). See attached screenshot.

Bar simulation does not simulate orthotropic slab stiffness correctly. Bar simulation can be comparable to edge effect which gives similar deformation (see attached file).

Orthotropic plate are used to simplify model definition and mainly to transfer loads to other elements. Stresses cannot be used for design. It was discussed in this thread  Please refer to this discussion

[Anonymous]

Thanks a lot @Krzysztof_Wasik, great explanation.

[Anonymous]

Hi again.

A few months ago I used Robot Orthotropic Panel to check stress on a square open GFRP.

Obviously after the information regarding the problem with accuracy of this sort of calculation, I went back and check again using a shell model and Ansys.

Fortunately the stress distribution on the panel model is very similar to the results on the shell model. Phew.

Just thought it was interesting and wanted to share.

@Krzysztof_Wasik