Carbon Composite vs Aluminium Frame FEA

[Ld3Ake]

Hi

I want to determine the optimal layup for a carbon composite frame structure but I am new to composite finite element modelling. So far I have just been setting up simple geometries analysing static and buckling load cases that I have been comparing against an aluminium model.

The geometry I am using is essentially a hollow square frame which has been modelled as surfaces. For both the composite and aluminium models the material thickness, loads, constraints and mesh size are identical. The buckling load that is calculated for the aluminiium model is 52kN, whilst the best I can get with the composite model is 43kN. I have definied my material orientation such that the 0' fibers follow the length of the bar, e.g. horizontal orientation definied in x direction, vertical in y direction. I know that to counteract buckling you need +-45 on the outside, which I have done, but I still can't get a frame that is stiffer than the aluminium.

My question is: for a given geometry and single load case, is it possible to determine a layup which results in a higher buckling load than the aluminium model?

 

[Ld3Ake]

Thanks for any info you may have.

 

[Ld3Ake]

Oh and another question. This is the buckling mode that my model produces. Is this physically possible?

Ideally I would have some physical data to validate my model but in this case I have to design the structure for the test specimen. I just want to get something that is a reasonable first iteration.

 

[SWComposites]

Yes, that is a "local" buckling mode; quite possible and common.

What is your exact layup?
What are the ply stiffness properties of your composite material?

 

[SWComposites]

Also, what is the thickness of the aluminum material? and what is the thickness of your composite ply?

 

[Ld3Ake]

I am just using an arbitary Carbon/Epoxy Composite:

E11 = 128Gpa,
E22 = 8.5Gpa
G12 = 3.4GPa,

The laminate is 2.2mm thick, with an 11ply layup [45/-45/0/-45/45/90]s (the 90 ply is the middle). So far I have used this layup uniformly across the whole structure.

The aluminium is 7050 with E=70.3GPa. Thickness is also 2.2mm

 

[RPstress]

That's ordinary HS UD fibre. 16.7% 0° 66.7%+-45° 16.7% 90°...equivalent in-plane E ~42 GPa, maybe 10% extra ability to resist buckling for putting the 45s on the outside: expected buckling ratio composite/Al 46/72 ~= 65%. 43/52 = 80%; I'm surprised it's that much.

You need to orientate the layup to resist the compression causing the buckling that's occuring.

Note: just because 45s on the outside are better than 45s on the inside doesn't make 45s good for preventing buckling; for onset of buckling it's effective bending modulus that counts most.

According to my notes, for a specially orthotropic laminate which is long compared with its width, compression buckling load Pcr is
Pcr = 2*pi^2/b * ( sqrt(D11*D22) + D12 + 2*D66 )
or in terms of equivalent bending moduli (subscript f for flex)
Pcr = t^3/b * pi^2/(6*(1-nufXY*nufYX)) * ( sqrt(EfX*EfY) + EfY*nufXY + 2*GfXY*(1-nufXY*nufYX) )

Sticking the 45s on the outside boosts bending G (GfXY above) which helps a bit and generally increases Pcr a bit compared with putting the 45s on the inside.

Compare Al Pcr
Pcr = pi^2/(1-nu^2).E.t^3/b

Of course, if you make the laminate the same weight as the Al its thickness goes up by about 2.8/1.55 = 1.8 times, so the bending stiffness and buckling load go up a bunch, which is one of the main things that is attractive about carbon/epoxy. t^3 soon overcomes differences in E of maybe 1-1/3 or 1.5 times (for a sensible layup).

Using intermediate modulus fibre as is commonly done these days for weight critical items gives about 20 or 30% extra E with no reduction in strength. If you really want a buckling or bending-vibration resistant structure such as satellite bits then high modulus or even ultra high modulus is used. The last time I worked on such a thing (1995) we used Torayca M55J, fibre E = 340 GPa (35 Msi). Even quasi-isotropic laminate E is 70+ GPa. That got the plate natural frequency well up for the weight. These days, if you could put up with a laminate allowable strain <1/4%, you could use M70J (E = 670 GPa) or maybe some seriously stiff pitch-based fibres.

That's enough war stories. Sorry.

 

[Ld3Ake]

Ok, thank you very much for replying RPstress. Obviously the specific stiffness of the composite is greater than aluminium, and when I ran the buckling analysis again using a thickness for the composite that is equivalent in mass to aluminium the buckling load was far superior.

So, as a very broad generalisation, it looks like that for a given cross section and thickness a carbon composite is not expected to match the performance of aluminium, unless of course an ultra high modulus fibre is used.

 

[SWComposites]

depends on your definition of "performance"; static tension strength, static compression strength, buckling stress, fatigue strength, etc .... all have different relative characteristics, and it depends on whether you compare on constant thickness, constant weight or something else. Broad generalizations are generally worthless.

 

[RPstress]

As SW implies, although weight is very important, it's the overall lifecycle cost which is often most important.

Carbon/epoxy's relative lack of fatigue sensitivity, lack of corrosion and lack of necessity for inspection for cracks make Boeing claim that 787 service costs will be quite a bit less even than a pretty modern airframe like the 777.

Mr. B. claims that 787 yearly maintenance cost will be 30% less than a 767 and will be maintained at that low level for at least twice the time after which the 767 costs zoom upwards.

That's not to say that looking after carbon airframe is problem free; the problems are different from ally and Boeing thinks they can be made significantly less costly.

 

[Ld3Ake]

Ha bringing the 787 into the discussion is worthy of a whole new thread entirely! Sure, in perfect operation the maintenance costs will be lower as there are fewer parts. The obvious flaw is what do you do if your one-piece fuselage gets hit? As far as I'm aware there is no way to accurately tell the extent of the damage. And you can't easily put a patch over the inflicted area. It will be interesting to see how this aircraft performs with the airlines over the coming years...

As for my initial problem, I decided to move to the next stage of modelling which is adding a foam core. I am actually creating a sandwich structure but I was trying to work in a step by step process gradually adding more complexity to my model. Turns out that the foam core significantly increases the performance of my structure. Instead of the local buckling on the top plate of the top beam I now get global buckling of the structure with the 0 plies taking most of the load. This structure massively outperforms an equivalent aluminium-foam core sandwich model.

The final structure I want to model is actualy one with a curved geometry but I am struggling to get my FEA software to align the material orientation correctly on the curved corners. Therefore at the moment I am having to work with straight edged geometries.