Geometry Check for Uniaxial/Corrugated panels

[ULBsha]

Hi,

I'm sizing a wing structure with corrugated panels. I use HyperFEA, and after a set of iterations, I update all the bounds, then re-run HyperFEA. When I update the bounds (based on the found result and Margin of safety), for some components, I get 'INVLD GEOM'.

I found out that I get rid of it if I increase the Y span panel buckling. So, I guess this INVLD GEOM is due to the violation of the Geometry Rule 40, i.e. 'Panel Width to Stiffener Spacing [Min # of stiffeners]'. Is that right ? (note that the analysis finishes without any error).


Therefore, I would like to know what would be the best (realistic & clever) way to solve this issue, knowing that I can't increase the panel buckling, since it is fixed by the position of ribs/spars ?

Should I just impose a maximum value when I'm updating the bounds, so that the stiffener spacing is not higher than 1/3rd of the panel buckling (to get at least 3 stiffeners) ?
If yes, I should also impose maximum values for other properties since this spacing should also be higher than the sum of all the parts between two stiffeners (e.g. flanges)... And therefore, what should I do if all these upper bounds are hit by HyperSizer during the analysis ?


PS: I also had a similar problem when sizing uniaxial stiffened panels.

Thanks for your help,
Regards

[Ryan]

Geometry checks can be disabled using the failure tab. The criteria for each geometry can also be modified using the Backdoor Data form.

http://hypersizer.com/help/#Failure/stiff-geometry.php

[ULBsha]

Thank you Ryan.

I saw that I can disable the geometry checks, however, I didn't do it because I was wondering if that's a good practice since the smeared stiffeners approach is not valid anymore ?

[James]

The smeared approach is an approximation that works best when the panel has more stringers. With fewer stringers, the iso-strain plate theory assumptions may be less accurately determining the stress distribution in the panel objects (skin, stringer web, etc.). This all depends on the surrounding structure and boundary conditions. The geometry check (>3 stiffeners per Y-span) is kind of an arbitrary rule based on previous experience. I'd have to see the application to judge if the smeared approach is going to be accurate.

Remember, HyperSizer also supports discrete modeling approaches where stringers are explicitly modeled with bar and/or shell elements.

[ULBsha]

Thanks for the explanations.

The smeared approach is an approximation that works best when the panel has more stringers. With fewer stringers, the iso-strain plate theory assumptions may be less accurately determining the stress distribution in the panel objects (skin, stringer web, etc.). This all depends on the surrounding structure and boundary conditions. The geometry check (>3 stiffeners per Y-span) is kind of an arbitrary rule based on previous experience. I'd have to see the application to judge if the smeared approach is going to be accurate.

The application is about sizing a complete aircraft. The problem is either on the suction side of the wing skin, or one part of the fuselage skin. These are defined as single component inside a single groupe. However, since they are supported by the ribs-spars or ringframes-longerons, I assumed that the X-span and Y-span buckling lengths are defined by the rib-spar and frame-longerons limitations (see pictures in attachments).

I know that the best way would be to split the skin and define different components for all these parts located between ribs-spars (or frames-longerons), but is my approach acceptable and not problematic ?

If I disable the geometry check, the results would just less accurate, or it is also possible that they become not correct at all ?

Remember, HyperSizer also supports discrete modeling approaches where stringers are explicitly modeled with bar and/or shell elements.

Yes sure. This solution would be good. But it would be much more time consuming since I also have to find the optimal configuration for ribs/spars and frames/longerons.
On the other hand, the solution of disabling the geometry check doesn't make me feel to be a safe solution (as you said, and also explained in the manuals based on previous experience, this smeared approach has been used for at least 3 stiffeners). So, I wanted to ask you again if there is no other solution/way/trick to get rid of my problem, which can be safer than disabling the geometry check, and less time consuming than modeling all the stiffeners ?

Thanks

[James]

Based on the images of your structure, I think the smeared approach will return accurate results for panel designs with as few as 1 stringer. These panel bays are small and clearly bounded by frames which would enforce strain compatibility on the edges. I think you can disable the geometry check and feel safe ignoring it. This check was implemented to prevent users from analyzing large panels with few stringers since panels like that will likely not behave as a smeared panel.