You are on page 1of 6

Predicting Buckling Failure of the Compressive Face Sheet of a Sandwich

Structure in Pure Bending


Consider a sandwich structure whose face sheets have a span-wise (horizontal direction) Young's modulus E f and whose core
has a transverse (vertical direction) Young's modulus of Ec . In pure bending, if E f >> Ec , then the face sheets will take all the
bending loads, and the core none.

Figure 1. Sandwich structure, with E f >> Ec , loaded in pure bending.

Since the tensile face sheet cannot buckle, this is equivalent to an elastic column on an elastic foundation.

Figure 2. Elastic column on an elastic foundation.

Assuming perfect bonding to the core, the (somewhat exaggerated) buckled shape of the face sheet would look something like
this:

Figure 3. Buckled shape of face sheet and core.

In reality, the onset of buckling would probably cause catastrophic failure.

The easiest way to attack this is using energy methods. Because energy is conserved as the face sheet buckles, the work done in
deforming the face sheet must be equal to the strain energy stored in the face sheet and in the core. This analysis is based on
arguments for fiber buckling in a composite, by Rosen1 .

Let's start by calculating the work P does when the face sheet buckles.

External work to buckle face sheet


Since before buckling the axial deflection of the face sheet is very small, any deflection that does occur is due to buckling. The
work done by the force P is therefore W º P u, where u is the deflection in the x-direction (axial deflection), and v is the deflec-
tion in the y-direction (transverse deflection) that occur upon buckling.

Figure 4. Buckled face sheet.


2 Buckling 2.nb

Figure 4. Buckled face sheet.

Note from Figure 4 that:

du = ds - dx (1)

From the triangle:

2
dv
ds = dx 2 + dv 2 = 1 + dx (2)
dx

So, substituting this into Eq. 1, and simplifying:

2
dv
du = 1 + | 1 dx. (3)
dx

But, in order to find the value of u, we need to know what shape the face sheet buckles into. For small deflections, we can
assume the face sheet is a sine wave of undetermined amplitude a and number of half waves m.
x
v HxL = a sin m p (4)
L
Since we have a proposed v(x), we can take its derivative, and sub this into Eq. 3, to get:

amp x 2
du = 1+ CosBm p F - 1 dx (5)
L L

To get the total deflection u, we would just integrate du over the length of the face sheet.

L amp x 2
u = ‡ 1+ CosBm p F - 1 dx (6)
0 L L

Problem is, this is not an easy integral to evaluate, so go to plan B. This involves approximating 1 + Hv ' HxLL2 - 1 as a Taylor

series expansion about x = 0, and taking only the significant terms. The resulting function is easier to integrate, and gives:

a2 m 2 p2
u= (7)
4L
To get this, we had to assume that m can only take on integer values. Finally, the work done by P as the face sheet buckles:

W =Pu (8)

or,
Buckling 2.nb 3

or,

a2 m 2 p2 P
W = (9)
4L

Strain energy stored in face sheet


For bending, the strain energy density  is the integral of the normal stress/strain curve. The face sheet shows linear properties
until failure, so:
1
f = sf ef . (10)
2
But Hook's law states:
sf
ef = (11)
Ef
The bending stress is:

M y*
sf = (12)
If
where y* is measured from the face sheet's neutral axis.

And the bending governing equation is:


M= Ef If v '' (13)

So:

1
f = Ef Hv ''L2 Hy * L2 . (14)
2
The strain energy in the face sheet is then just the integral of the strain energy density throughout the volume.

1 L
Uf = ‡ f „ V = Ef ‡ Hv ''L2 ‡ Hy * L2 dA dx (15)
V 2 0 A

But, Ÿ A Hy* L2 „ A is just the area moment of inertia I f of the face sheet, so:

Ef If L
2
Uf = ‡ Hv ''L dx (16)
2 0

Let's call the thickness of the core h2 , and the thickness of the compressive face sheet h1 .

Figure 5. Geometry of sandwich structure,


where x is the span-wise direction.
4 Buckling 2.nb

Figure 5. Geometry of sandwich structure,


where x is the span-wise direction.

The face sheet is rectangular in cross section. Let's say that the width, in and out of the page, is b, so:

1
If = b h31 (17)
12
Taking the second derivative of v(x), substituting into Eq. 16, and simplifying gives us:

a2 b h31 m4 p4 Ef
Uf = (18)
48 L3

Strain energy stored in core

Now that we have the external work done during buckling, and the internal strain energy stored in the face sheet, it is time to
consider the strain energy stored in the core.

Strain energy density for linear stress/strain in the core.

1
c = sc e c (19)
2
Strain is just the change in length divided by the original length. But we know that if the core is bonded to the face sheet, the
strain in the core will be proportional to the deformation of the face sheet. Figures 3 and 5 lead to the following expression for
the transverse strain in the core:
vHxL
ec = (20)
h2
And Hooke's Law can be applied.

sc = Ec e c (21)

Combining Eqs. 19 - 21, and recognizing that the strain energy in the core is then the integral of the strain energy density over
the volume of core, we get:
b Ec L
Uc = ‡ v 2 HxL dx (22)
2 h2 0

If we again require m to be an integer number of half-waves, this simplifies to:

a2 b Ec L
Uc = (23)
4 h2

Buckling load
Buckling 2.nb 5

Buckling load
We can now solve for the buckling load on the face sheet using conservation of energy.

W = Uf + Uc (24)

Substitute all that garbage in:

a2 m 2 p2 P a2 b h31 m4 p4 Ef a2 b Ec L
= + (25)
4L 48 L3 4 h2
and solve for P:

2 Ec L 2 Ef h31 m2 p2
P=b + (26)
h2 m 2 p2 12 L2

Notice that a canceled out, meaning that the amplitude is arbitrary. We now have the buckling force in the face sheet, as a
function of material properties, geometry, and the parameter m, which identifies the number of half-waves the face sheet exhibits
over its length. Let's stick in some numbers and plot P with respect to m, for a given length of spar.
P HlbL
2.0 µ 106

1.5 µ 106

1.0 µ 106

500 000

m
0 5 10 15 20

Figure 6. P vs. m for a given set of geometry and material properties.

If m is too small (the period is large), too much energy is stored in the core, whereas if m is too large (the period is small), then
the face sheet wiggles back and forth too much, and stores too much energy. The actual value of m that the face sheet will buckle
into will be that for which P is lowest. By finding the value of m that leads to the lowest P, it can be shown that the critical
buckling load simplifies to:

h31 Ec Ef
Pcr = b (27)
3 h2

But, what we really want is the internal bending moment that would cause buckling (because we have already developed equa-
tions for the shear force and bending moment for the whole wing).

For a spar where E f >> Ec (a conservative approximation), Figures 1 and 5 lead to a bending moment in the spar of
PHh2 + Hh1 + h3 L ê 2L. Therefore, the bending moment that will cause buckling is:

h1 + h3 h31 Ec Ef
Mcr = b h2 + (28)
2 3 h2

So, the only things you need to know are the dimensions of your spar (Figure 5), and Young's modulus (stiffness) of the face
sheet and core. For composites, E f is the stiffness along the longitudinal direction of the face sheet (along the wingspan), but Ec
is the stiffness through the thickness of the core (top to bottom).

D.P. Johnson
2/4/2010
6 Buckling 2.nb

So, the only things you need to know are the dimensions of your spar (Figure 5), and Young's modulus (stiffness) of the face
sheet and core. For composites, E f is the stiffness along the longitudinal direction of the face sheet (along the wingspan), but Ec
is the stiffness through the thickness of the core (top to bottom).

D.P. Johnson
2/4/2010

1
B. W. Rosen, "Mechanics of Composite Strengthening," Ch. 3, in Fiber Composite Materials, ASM, Metal Park, OH, 1965.

You might also like