How to avoid weight increase due to buckling failure?

[ULBsha]

I am sizing a hypersonic wing structure, with multiple ribs and spars.

The skin is modeled as stiffened panels, ribs and spars are represented by open beams for their caps, and small panels for their webs (like the AP1 tutorial).

If I only optimise this structure for the pressure load case, I get a wing of 50 tonnes (30 for the skin, 20 for the ribs+spars). However, by turning off the buckling failure mode for beams and panels, my weight is reduced to only 17 tons (15 for the skin and 2 for the ribs+spars !).

Since the weight of this wing has been estimated at 15 tonnes (almost the same weight than the case without buckling), I was wondering how I could reduce considerably the weight (in order to get something close to the estimated value of 15 tons) by allowing the buckling failure mode ?
I tried to change a lot of different parameters, but the buckling mode always imposes a very high weight. Did you have any similar behaviour ? If yes, how did you solve it? How could I get rid of this big weight increase due to the buckling failure ?

P.S: Uniaxial and grid stiffened panel families for my wing skin, with T-shape beams for the spars/ribs caps, have been all tested, but the results give the same order of magnatiude. Note that the number of ribs and spars has also been varied (between 3 and 8 ). The X-span and Y-span buckling lengths have also been verified and modified when not correct. The beam orientation seems also right (the same than AP1 when compared graphically).

Thanks a lot.
Regards

[ULBsha]

No one has experienced this kind of behaviour... ? 

[James]

It's difficult to diagnose what is happening without inspecting your model. When you say "by turning off the buckling failure mode for beams and panels, my weight is reduced to only 17 tons" are you turning off both panel buckling and local buckling?

Here are some suggestions based on your original post...

Check X buckling spans:
Panel buckling of stiffened panels is very sensitive to the X buckling span. Usually in wings the ribs are spaced 18-24 in apart, which creates a short panel (low X buckling span). This is ideal for stiffened panels because the have a high D11 and low D22.

Controlling failure modes for stiffened panels:
For stiffened panels, the controlling failure modes are typically: strength, local buckling of the skin, crippling and panel buckling. By turning off local buckling and panel buckling you are going to miss potential failure modes. If active, the JE crippling-buckling interaction method is likely to control much of the structure. For composites you could make an argument to turn this failure mode off based on the fact that the crippling is using damage tolerant allowables so the crippling-buckling interaction is conservative.

How to design stiffened panels so they don't buckle:
For stiffened panels, it is better to have stiffeners with a high EI. So increase the height and the axial stiffness of the flanges. Also, the skin between the stiffeners is prone to local buckling and typically doesn't handle Ny compression well. For composite skins it is better to force 90s to the outside to prevent transverse local buckling waves. 

Euler beam buckling too conservative:
For beams that are attached to skin (like spar caps and rib posts), I don't think it is appropriate to use the beam buckling analysis in HyperSizer. These methods are assuming simply supported beams that are unsupported down the length. This is similar to the classing Euler buckling approach for beams. It will return conservative results for the beams.