5.2.

Post-Buckling

Post-buckled composite stiffened panel under shear

Buckling vs Post-buckling
in general buckling does not imply failure especially
for plates
P
Pcr

wcenter
beam post-buckling curve is flat and failure strains are reached at low
post-buckling loads

Buckling vs Post-buckling
post-buckled (stiffened) skins are significantly lighter
but have additional challenges:
susceptible to skin/stiffener separation failure
where only resin holds the structure (unless
fasteners or other means are used to hold it together)
susceptible to the creation and growth of
delamination under fatigue loading especially at high
post-buckling factors
skin/stiffener
separation

Buckling versus Post-Buckling

If a composite structure is allowed to post-buckle,
the ratio of final failure load to buckling load (post
buckling ratio PB) must be chosen carefully
(especially for shear loads)
Conservative approach (but not very efficient): Do
not allow buckling below limit load (i.e. PB=1.5)
PB>5 great design challenge both under static and
fatigue loads

Post-buckling mode
frame
stiffeners

skin
frame

What buckles and when?

What BCs are the different components supposed to simulate?

Post-Buckling Scenarios
Skin buckling as a whole (stiffeners only
increase the bending stiffness of the skin)
Skin buckling between the bays (stiffeners
act as panel breakers)
Stiffeners buckle as columns or locally
(crippling)
Frames do not buckle!

Post-buckling scenarios
Most efficient:
Skin between stiffeners and frames buckles
first

Stiffeners do not buckle and do not move out

of plane (depending on cross-section they may
rotate => BC implications)
When required PB value is reached, skin fails
in compression and/or stiffeners fail by crippling

stiffeners impose zero deflection

and slope

stiffeners impose zero deflection

only

Post-Buckling Analysis
governing equations: Large deflection von Karman
equations
place moment equilibrium eqn into Fz equilibrium eqn:
2 M xy 2 M y
2M x
2w
2w
2w
w
w
2

N x 2 2 N xy
N y 2 px
py
pz 0
2
2
xy
xy
x
y
x
y
x
y

px, py, pz distributed loads (force/area)

use the moment-curvature relations to substitute for

the moments

2w
2w
M x D11 2 D12 2
x
y
2w
2w
M y D12 2 D22 2
x
y
M xy

2w
2 D66
xy

Bij=0
D16=D26=0

Derivation of von Karman

equations (large deflections)
to obtain:
4w
4w
4w
2w
2w
2w
w
w
D11 4 2( D12 2 D66 ) 2 2 D22 4 N x 2 2 N xy
N y 2 px
py
pz
xy
x
y
x
x y
y
x
y

1st von Karman equation: Predominantly bending

Strain Compatibility
2

2w
2 xo
3u
1 2w 2 w
3u
w 3 w

x xy 2
y 2
xy 2 2 y x xy xy 2 xy

u 1 w

x 2 x

yo

v 1 w

y 2 y

xyo

u v w w

y x x y

xo

2 yo
x 2

2w
3v
1 2w 2 w
3v
w 3 w

y x 2 y
x y 2 x y xy x 2 y xy

2 xyo

3u
3v
w 3 w w 3 w 2 w
2w 2w
2

xy
xy 2 x 2 y y x 2 y x xy 2 xy
x y 2

2 yo

2 xyo

xo

...
2
2
xy
y
x
2

Derivation of von Karman

equations (large deflections)
use (non-linear) strain compatibility
2
2
2 xo yo xyo 2 w 2 w 2 w
2

2
2
xy xy
y
x
x y 2
2

xo, yo, xyo mid-plane

strains

non-linear terms

invert stress-strain eqns to express strains in terms of

Nx, Ny, Nxy
N x A11 xo A12 yo
N y A12 xo A22 yo
N xy A66 xyo
A16=A26=0

xo

A22

A11 A22 A12

yo
xyo

Nx

A12
A11 A22 A12

1
N xy
A66

A12

A11 A22 A12

Nx

Ny

A11
A11 A22 A12

Ny

Derivation of von Karman

equations (large deflections)
substitute in strain compatibility to obtain,
1
2
A11 A22 A12

2
2
2
2

Ny

Nx
y
x
A22

A
12
11
12
2
2
2
2

2
2
1 2 N xy 2 w

2

2
A66 xy xy

introduce Airy stress function and potential V that identically

satisfy stress equilibrium:
2F
Nx 2 V
y
2F
Ny 2 V
x
2F
N xy
xy

potential of distributed
loads (=0 in our postbuckling problem)

Derivation of von Karman

equations (large deflections)
to obtain
2

1
A11 A22 A12

2w 2w 2w
4F
4F
4F 1 4F
A22

2
2 A12 2 2 A11

4
4
2
2
y
x y
x A66 x y
x y 2
xy

2nd von Karman equation: Predominantly stretching

(and non-linear)

Post-buckling of a square anisotropic plate

under compression
y

rigid
a
a

Px
(applied
force)
x

simply-supported with three immovable edges:

w=0 at x=y=0 and x=y=a
u=0 at x=0
v=0 at y=0 and y=a
u=-C at x=a (constant compressive displacement)

Solve the two von Karman equations

approximately
assume
w w11 sin

x
a

sin

w11, K02, K20 unknown

coefficients

a
Px y 2 Py x 2
2x
2y
F

K 20 cos
K 02 cos
a 2
a 2
a
a

substitute in the 2nd von Karman equation

A22
A11 A22 A12

4
4
A11
16 4
2y
16 4
2x
2y
2x
2
2
K 02 4 cos

K
cos

w
cos

w
cos
20
11
11
2
a
a
a
a
a
a4
2a 4
2a 4
A11 A22 A12

K 02

A A A12 w11
11 22
A22
32

K 20

A A A12 w11
11 22
A11
32

matching coefficients
of cos(2y/a) and
cos(2x/a)

and the 2nd von

Karman eqn is
identically satisfied

Solution (contd)
Py (as a function of Px) and deflection C are determined
by integrating stress-strain eqns and using average BCs
on u and v:
xo

also

u 1 w

x 2 x

xo

non-linear strain-displacement eq

A22
A11 A22 A12

Nx

A12
A11 A22 A12

Ny

inverted stress-strain eq

2F
Nx 2 V
y
2F
Ny 2 V
x
2F
N xy
xy

Airy stress function F

A22
u
dxdy

2
0 0 x
A11 A22 A12
a a

A12
2F
dxdy

2
0 0 y 2
A11 A22 A12
a a

2F
1 w
dxdy

dxdy
0 0 x 2
2 0 0 x
a a

a a

Solution (contd)
au (a, y ) u(0, y)

=-C

A22 a 2
A11 A22 A12

A12 a 2
Px

2
a A11 A22 A12

Py

a

=0

x y
( w11 cos sin ) 2 dxdy
a
a
a
2

w11

2
4

from which:
C

aA22
A11 A22 A12

similarly, from

yo

2
Py
Px
aA12
2

w11
2
a A11 A22 A12 a
8a

v 1 w

y 2 y

2
A12
A11 A22 A12
2
Py Px
w11
A11
8a
A11

= Poissons ratio for

isotropic

=0 for in-plane problems

Solution (contd)
substitute now in the 1st von Karman equation
4w
4w
4w 2F 2w
2F 2w 2F 2w
D11 4 2( D12 2 D66 ) 2 2 D22 4 2
2

xy xy x 2 y 2
x
x y
y
y x 2

use the following:

4w
4w
4w
x y

w
sin
sin

11
4
2
2
4
a
a
x
x y
y
a
4

2w 2w
x y

w
sin
sin

11
a
a
a
x 2
y 2

2

Py 2 A11 A22 A12 w11

2F
2x

cos

a a
A11
32
a
x 2
2

Px 2 A11 A22 A12 w11

2F
2y

cos

a a
A22
32
a
y 2
2

Solution (contd)
to substitute, and match coefficients of
sin(x/a)sin(y/a) to obtain the governing eq for w11
2
2 A11 A22 A12 A11 3 A22

16 A11 A22

A
w D11 2( D12 2 D66 ) D22 Px 1 12 w11 0
A11

a
3
11

which for w110 leads to

2nd von Karman eq is satisfied

approximately!

16 A11 A22 D11 2 D12 2 D66 D22

Px
w11
1
2
2
D

2
D

2
D

D
A
A

A
A

3
A

11 22 12 11 22

11
12
66
22
a

A12

A11

=Pcr , buckling load (units of force), exact for square plate

Solution (end)
For out-of-plane deflections to be possible, must have

Px
1
Pcr
the applied load Px must exceed the plate buckling load

Results-Implications
use as example,
(45)/(0/90)/(45) square plate of side 25.4 cm
Material is plain weave fabric with properties:

Ex=Ey=68.94 GPa
xy=0.05
Gxy=5.17 GPa
ply thickness = 0.19 mm

Load versus center deflection

10
9
8
7
6
5
4
3
2
1
0

P/Pcr
a
a

center defl/plate thickness (w11/h)

In-plane compression load

1

P
=1
Pcr

0.9

1.2

0.8
0.7

y/a

0.6
0.5
0.4
0.3
0.2
0.1
0
0

Nx/(Pcr/a)

the center of the plate sees very little load!

Implication
after buckling approximate in-plane load

a
peak Nx value

there is an effective width beff at the edges of the

panel over which the load is concentrated

beff

Determination of effective width

total applied force must equal the force created by
the load applied over the effective width:

N dy 2N
x

x max

beff

P A A A12 w11 2
2F
2y
N x 2 x 11 22
cos
a
A22
32 a
a
y
2

P A A A12 w11 2
x 11 22

a
A22
32 a
2

N x max

N dy P
x

Solving for beff

beff a

A12 Pcr
A11

1
21 21

A
P
A

3
A

11

x
11
22

Note that for quasi-isotropic layup,

A12
12
A11
A11
1

A11 3 A22 4

12 0.3

beff a

1
Pcr
2 1.31
Px

beff=0.303a for Px>>Pcr and QI layup

Effective width beff

0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0

beff/a

A12 =0.05
A11

beff

0.3
0.7
a
a
0

0.1

0.2

0.3

0.4

0.5

Pcr/Px

weak dependence on A12/A11

0.6

0.7

0.8

0.9

Significance for design

beff provides an idea of where failure is expected under
compression and where reinforcement may be needed
For (conservative) failure predictions obtain the
maximum compressive stress
max

N x max

(h is plate thickness)

and compare to allowable ultimate compression stress

max ult

for no failure

plus one more use in crippling of stiffener flanges

How good is the analysis compared to

reality?
For a meaningful comparison, two conditions must be
met:
Boundary conditions in the test case must be the same as in
the model
Sufficient number of terms must be included in the solution (as
opposed to only one term used here for simplicity)

As it is hard to find test results with exactly these

boundary conditions, FE results are used to validate the
analysis
Two different laminates: (a) 45 degree dominated
(b) Quasi-isotropic

Analysis model versus FE (45-dominated

laminate)

end displacement

Results from R. Kroese MS Thesis TUDelft 2013

Excellent agreement between FE and Analysis

Analysis model versus FE (Quasi-isotropic

laminate)

end displacement

Results from R. Kroese MS Thesis TUDelft 2013

Excellent agreement between FE and Analysis

Boundary conditions discussion

If the stiffeners in a stiffened panel have very high EI and
very low GJ, and one end is attached to a frame with
also very high EI and low GJ, this analytical model would
be quite accurate
If not, the model may be conservative or unconservative
Must also use enough terms in the series and account
for rectangular as opposed to square panels

frame

stiffener

skin between
stiffeners