Buckling margin not coinciding with FE analysis result.

[iankim]

As seen in the attached pictures to this posting,
the resulting buckling margin (0.115) calculated
by Hypersizer does not coincide with the margin
(0.5791-1=-0.4209) calculated by MSC.Nastran FE
analysis.
I calculated the margin using in-house tool
which uses Rayleigh-Ritz method, but it gave me
the margin -0.4094, which is around FE analysis
result -0.4209.
I don't understand where the difference was originated.
This panel even doesn't have curvature, so I don't
think that big difference is a possible case.

[Ryan]

What are the margins for the flat panel buckling? "Panel Buckling, Flat, Simple BC, Uniaxial or Biaxial" and "Panel Buckling, Flat, Simple BC, Shear". This will give you an idea how much influence the shear load has.

There are also some options in the Backdoor Data form to make the HyperSizer curved panel buckling (Rayleigh-Ritz) search more mode shapes. One thing you could try is activating these options and see of the margin changes.

http://hypersizer.com/help/index.php#BackdoorData/bd-15-panel_buckling.php

-Ryan

[iankim]

I used backdoor option to facilitate shear load into buckling,
and then the load and margin were changed.
The changed margin is 0.3832 as seen in the first attachment.

I calculated the margin using other tool and FE analysis,
and the margin was between 0.1 and 0.15.
So 0.3832 is a little bit too high compared to 0.1~0.15.

I tried the other buckling calculation for that panel independently,
which means the panel is not a component of big model
but just a stand-alone single panel.
And its margin turned out to be 0.04949 as seen in the second attachment.
This number is closer to 0.1~0.15 than 0.3832.

How should I understand the different result (0.3832 and 0.04949) of
a panel when the former is the result of a component of whole wing model and the latter is the result of a single panel?
is

[Ryan]

There may be some mismatch with the boundary conditions of the different analyses. The HS buckling analysis of a component ignores any surrounding components. Idealized boundary conditions are used. In your case, simply supported boundary conditions are used.

Another source of difference may be the moment loads. HS ignores the applied moment when doing the buckling analysis. I'm not sure how your other buckling methods are handling the applied loads.

-Ryan