Module 9

Stability and Buckling

Readings: BC Ch 14

Learning Objectives
• Understand the basic concept of structural instability and bifurcation of equilibrium.
• Derive the basic buckling load of beams subject to uniform compression and different
displacement boundary conditions.
• Understand under what conditions structural design is limited by buckling considera-
tions.
• Understand the response of beam structures under a combination of tranverse loads
and intense compressive loads.
• Understand the postbuckling behavior of beam structures.

9.1 Introduction to bifurcation of equilibrium and struc-

tural instability

Concept Question 9.1.1. Buckling of a rigid bar on a torsional spring

Consider a rigid bar with a torsional spring at one end and a compressive axial load at
the other end (Figure 9.1(a)). We consider the possibility that the bar can be in equilibrium
not just in the undeformed configuration θ = 0, but perhaps in a deformed configuration as
well, Figure 9.1(b).
1. State the equilibrium equation in the deformed configuration.
Solution: To determine the critical value of P , we consider the equilibrium of the
moment with respect to point O in Figure 9.1(b)
P L sin(θ) = Kθ θ

209
210 MODULE 9. STABILITY AND BUCKLING

P P
L sin(θ)

θ
L

bcbcbc
bc bcbcbc
bc Kθ θ
O O
(a) θ = 0 (b) θ 6= 0

Figure 9.1: Equilibrium positions of a rigid bar on a torsional spring for a trivial solution
(θ = 0) and a non-trivial solution (θ 6= 0).

2. Rewrite the equations in the case of small angles

Solution: For small angles, sin(θ) ' θ, hence:
(P L − Kθ )θ = 0

3. Interpret this equation. Under what conditions is it satisfied? Solution: This

equation can be satisfied in any of the two scenarios:
(a) θ = 0. This is the trivial solution.
Kθ
(b) the parenthesis is zero, which required P = Pcr = L

4. If the first solution is satisfied θ = 0, what are the restrictions on the load P ?
Solution: There are no restrictions, P can adopt any value (of course we are ignoring
plastic yield or other material failure modes under compression, but for the purpose of
this exercise we assumed the bar to be rigid).
5. If the second solution is satisfied P = Pcr , what are the restrictions on the angle of
rotation θ? Solution: There are no restrictions, θ can adopt any value.
6. What is the implication? Solution: The implication is that once the
load reaches the critical value, the rotation is unbounded (unstable). Increases in the
rotation angle leads to growth of the loading moment which is equal to the growth of
the internal resisting moment. Then, any angle θ is an equilibrium position.
7. Challenge: what happens for large angles? Solution:
The loading moment P L sin θ grows slower than the resisting moment Kθ θ and in the
large deformation case a second (stable) equilibrium configuration is obtained
9.1. INTRODUCTION TO BIFURCATION OF EQUILIBRIUM AND STRUCTURAL INSTABILITY211

Concept Question 9.1.2. Euler buckling load for a cantilever beam

e3 P

ū3 (x1 ) δ

e1

Figure 9.2: Bifurcation of equilibrium in a compressed cantilever beam

Consider a cantilever beam of length L made of a material with Young’s modulus E and
whose uniform cross section has a moment of inertia with respect to the x2 axis I22 . The
beam is subjected to a compressive load P , as shown in the figure.
We seek to find conditions under which the beam will buckle, i.e. the beam can be in
equilibrium under the load P in a configuration involving non-trivial (non-zero) lateral de-
flections v(x). To this end, we enforce equilibrium of the beam in the deformed configuration.

1. At a position x1 along the axis, the deflection of the beam is u3 (x1 ) and the moment
produced by the force P with respect to that point on the beam in the deformed
configuration is given by....
Solution:


M2 = P δ − u3 (x1 ) (9.1)

where δ is the deflection at the cantilever’s tip:

δ = u3 (L) (9.2)

and is an unknown of the problem

2. Write the expression for the internal moment produced by the ensuing bending stresses
in terms of the curvature at that point
Solution: From Euler-Bernoulli beam theory, the internal moment produced by
the ensuing bending stresses are given by:

M2 = EI22 u003 (x1 ) (9.3)

212 MODULE 9. STABILITY AND BUCKLING

3. Show that enforcing equilibrium of internal and external moments leads to an ODE of
the type:
u003 (x1 ) + k 2 u3 (x1 ) = k 2 δ
and find k Solution: Setting Mint = Mext , we obtain:


EI22 u003 (x1 ) = P δ − u3 (x1 )

dividing by EI22 and rearranging we obtain the sought expression with:

P
k2 = (9.4)
EI22

4. The general solution of this ODE is:

u3 (x1 ) = A sin (kx1 ) + B cos (kx1 ) + δ

Apply the appropriate boundary conditions to this problem to obtain the solution for
the deflection in terms of δ Solution: Applying the boundary condition u03 = 0 at
x1 = 0, we find:
kA cos (k0) − B sin (k0) = kA = 0 (9.5)
from which we conclude that A = 0. From the boundary condition u3 = 0 at x1 = 0,
we obtain:
B cos (k0) + δ = 0 (9.6)
which gives the value of B in terms of δ:

B = −δ (9.7)

The solution is finally:

u3 (x) = δ(1 − cos (kx1 )) (9.8)
which is given in terms of the unknown value of the deflection at the tip δ.
5. From the solution obtained, use the condition that u3 (L) = δ and derive two possible
solutions to this problem: 1) the trivial solution where there is no deformation, 2) a non-
trivial solution where equilibrium can occur in the deformed configuration providing
that the load is large enough.
Solution: 1) this case is simply 0 = δ, → u3 (x1 ) = δ(1 − cos (kx1 ) = 0 everywhere.
In this case, k and therefore the load can adopt any value.
2) 0 6= δ = u3 (L) = δ(1 − cos (kL), which requires:

cos (kL) = 0 (9.9)

This condition is satisfied when the argument of the cosine is an odd multiple of π2 :
π
kL = (2n + 1) (9.10)
2
9.1. INTRODUCTION TO BIFURCATION OF EQUILIBRIUM AND STRUCTURAL INSTABILITY213

6. Express the non-trivial condition in terms of the applied load to obtain the critical
loads Solution: Replacing the value of k from equation ?? we obtain:
r
P π
L = (2n + 1) (9.11)
EI22 2

from where we finally obtain the critical loads:

π 2 EI22
Pcrn = (2n + 1)2 (9.12)
4L2

7. What is the minimum value of the load P for which a non-trivial solution is found?
Solution: The minumum value of this expression is attained for n = 0 with the
result:
π 2 EI22
Pcr0 = (9.13)
4L2
which is known as Euler’s buckling load.

8. Find the mode shapes of the deformed cantilever for each value of the critical load
Solution: From the solution given by equation ?? we obtain the corresponding
buckling modes:
n h π x1 io
un3 (x1 ) = δ 1 − cos (2n + 1) (9.14)
2L

9. Sketch the first three buckling modes of the beam Solution: Shown in Figure ??

(a) First buckling mode, n=0 (b) Second buckling mode, n=1 (c) Third buckling mode, n=2

Figure 9.3: Buckling modes of a cantilever beam

214 MODULE 9. STABILITY AND BUCKLING

e2 p2
P

p1
ū2

e1

Figure 9.4: Deformed beam with lateral and axial loads

9.2 Equilibrium equations

As discussed in previous sections, they key ingredient in the analysis of bifurcation of equi-
librium is to allow for the possibility that the structure will have additional equilibrium
configurations in the deformed state. In order to express this in mathematical terms, we
need to restate the differential equations of equilibrium of the beam in the deformed config-
uration, Figure 9.3. Consider the equilibrium of an infinitesimal slice of beam of size dx1 ,
Figure 9.4. Since we are interested in computing the critical buckling load, we will consider
the beam to be at the onset of buckling. Accordingly, we will assume that the deflection is
very small (ū02  1) and that the transverse shear force V2 is very small compared to the
normal force N1 (V2  N1 ).
Force equilibrium in the e1 direction gives:

− N1 cos(ū02 ) + (N1 + N10 dx1 ) cos(ū02 + ū002 dx1 )

+ V2 sin(ū02 ) − (V2 + V20 dx1 ) sin(ū02 + ū002 dx1 ) + p1 dx1 = 0

According to the assumption of small deflection, it follows that

cos(ū02 ) ≈ 1 and cos(ū02 + ū002 dx1 ) ≈ 1

sin(ū02 ) ≈ ū02 and sin(ū02 + ū002 dx1 ) ≈ ū02

and we obtain:

N10 − V20 ū02 = −p1

0
(N1 − V2 ū02 ) + V2 ū002 = −p1

The term in ū002 is a second order differential term which can be neglected. The term V2 ū02 is
very small compared to N1 because ū02  1 and V2  N1 ; it is thus neglected as well. The
previous equation can thus be re-written as follows:

N10 = −p1 (9.15)

where p1 is the distributed force in the e1 -direction.

9.3. GOVERNING EQUATION 215

N1 + N1′ dx1
M3 + M3′ dx1
ū′2 + ū′′2 dx1
M3 A b

π − ū′2 V2 + V2′ dx1

N1 V2
dx1

Figure 9.5: Free body diagram of an infinitesimal slice of the deformed beam

Force equilibrium in the e2 direction gives:

− V2 cos(ū02 ) + (V2 + V20 dx1 ) cos(ū02 + ū002 dx1 )

− N1 sin(ū02 ) + (N1 + N10 dx1 ) sin(ū02 + ū002 dx1 ) + p2 dx1 = 0

Using the same simplifications of the sines and cosines introduced above, the equation be-
comes:

V20 + N10 ū02 = −p2

0
(V2 + N1 ū02 ) − N1 ū002 = −p2

The term in ū002 is of second differential order and is thus neglected. However, both V2 and
N1 ū02 are of the same order of magnitude. p2 is the distributed force in the direction e2 . We
then obtain:
0
V20 + (N1 ū02 ) = −p2 (9.16)

Moment equilibrium in the e3 direction with respect to point A shown in Figure 9.4 gives:

−M3 + (M3 + M30 dx1 ) + V2 cos(ū002 dx1 )dx1 + N1 sin(ū002 dx1 )dx1 = 0

After applying the previously introduced sines and cosines simplifications and neglecting
higher order terms, the equation becomes:

M30 + V2 = 0 (9.17)

9.3 Governing equation

The general beam-column equation can be derived by differentiating (9.3) with respect to
x1 and using the expression of V20 from (9.2):
0
(M30 + V2 ) = M300 + V20
= M300 − (N1 ū02 )0 − p2 = 0
216 MODULE 9. STABILITY AND BUCKLING

Then, using the moment-curvature relationship (7.13), we arrive at:

M300 − (N1 ū02 )0 = p2

c 00 00
(H33 ū2 ) − (N1 ū02 )0 = p2
(IV )
c
H33 ū2 − (N1 ū02 )0 = p2
c
Finally in the case of homogeneous cross sections, we have H33 = EI33 and the beam column
equation becomes:
(IV )
EI33 ū2 − (N1 ū02 )0 = p2 (9.18)
which is a fourth-order differential equation, that depends on N1 . Hence, in order to solve
(9.4), one needs to solve first (9.1) with the appropriate boundary condition: N1 (L) = −P.
In the case of no axial distributed force, (9.4) becomes:

(IV )
EI33 ū2 + Pū002 = p2 (9.19)

Solutions of (9.5) are of the form:

r r
P P
ū2 (x1 ) = A sin( x1 ) + B cos( x1 ) + Cx1 + D
EI33 EI33
In order to solve this fourth-order differential equation we need four boundary conditions,
two at each end.

9.4 Buckling loads and shapes for different beam bound-

ary conditions

Concept Question 9.4.1. Buckling of a uniform beam simply supported at both ends
Consider the case of a uniform beam (i.e, the product EI is constant along the beam) of
length L as shown in Figure 9.4.1. The beam is simply supported at both ends and loaded by
a uniform axial force P which acts on the beam neutral axis. The displacement ū22 satisfies
the governing equation (9.4) and the solution is given by (9.3).
e2

b
P b P

L e1

Figure 9.6: Simply supported uniform beam at both ends.

1. Write the boundary conditions needed to determine the constants A, B, C and D in

the solution of equation (9.3). Solution:
9.4. BUCKLING LOADS AND SHAPES FOR DIFFERENT BEAM BOUNDARY CONDITIONS217

at x1 = 0 and x1 = L:
 
u2 (x1 = 0) = 0 u2 (x1 = L) = 0
and
M3 = EI33 u002 (x1 = 0) = 0 M3 = EI33 u002 (x1 = L) = 0

2. Using these boundary conditions, compute the three constants B, C and D to obtain
the non-trivial solution ū2 as a function of the constant A.
Solution:
for ū2 (x1 = 0) = 0:
r r
P P
ū2 (x1 = 0) = A sin( × 0) + B cos( × 0) + C × 0 + D
EI33 EI33
= B+D =0

for M3 (x1 = 0) = 0:
r r
P P
ū002 (x1 = 0) = −P A sin( × 0) − P B cos( × 0)
EI33 EI33
= B=0

hence, D = 0.
for ū2 (x1 = L) = 0:
r
P
ū2 (x1 = L) = A sin( × L) + C × L = 0
EI33

for M3 (x1 = L) = 0:
r
P
ū002 (x1 = L) = −AP sin( × L) = 0
EI33

The last two equations can be satisfied under the following two conditions: 1) C =
A = 0 → u2 (x1 ) = 0 ∀x1 (trivial solution), in which
q case there is no restriction on the
load P , 2) C = 0, A 6= 0, which will require sin( EIP33 × L) = 0. The deflection of the
beam is then given as a function of the undetermined constant A:
r
P
ū2 (x1 ) = A sin( × x1 )
EI33

3. Using the boundary condition ū2 (x1 = L) = 0, determine the condition on the load P
for which we have a non-trivial solution for ū2 (i.e ū2 6= 0).
218 MODULE 9. STABILITY AND BUCKLING

Solution: As we saw, the non-trivial solution requires

r
P
sin L = 0, →
EI33
r
P
L = nπ
EI33
The buckling loads are then given by:

n2 π 2 EI33
Pcrn =
L2

which is similar to the equation (??).

And the corresponding buckling deflection modes by
nπx1
ū2 (x1 ) = A sin( )
L

4. Determine the lowest (Euler) buckling load Pcr Solution: This is obtained for
n = 1:
π 2 EI33
Pcr1 = (9.20)
L2

5. Compare the Euler buckling load for a simply supported beam with the one obtained
previously for a cantilever beam (equation (??)). Solution: In the case of a
cantilever beam, we found:
π 2 EI22
Pcr0 =
4L2
By comparing the two equations we see that the Euler buckling load for a simply
supported beam is 4 times higher than that for a cantilever beam.

Concept Question 9.4.2. Buckling of a uniform beam clamped at both ends

Consider the case of a uniform beam of length L as shown in Figure 9.4.2. The beam is
clamped at both ends and loaded by a uniform axial force P at (x1 = L) which acts on
the beam neutral axis. The displacement ū22 satisfies the governing equation (9.4) and the
solution is given by (9.3).

1. Write the boundary conditions needed to determine the constants A, B, C and D in

the solution:
r r
P P
ū2 (x1 ) = A sin( x1 ) + B cos( x1 ) + Cx1 + D
EI33 EI33
9.4. BUCKLING LOADS AND SHAPES FOR DIFFERENT BEAM BOUNDARY CONDITIONS219

0.8

0.6

0.4

0.2
ū2 /A

−0.2

−0.4

−0.6

−0.8 k = π/L
k = 2π/L
−1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x1 /L

Figure 9.7: Deformation modes of the simple supported beam.

220 MODULE 9. STABILITY AND BUCKLING

e2

L e1

Figure 9.8: Uniform beam clamped at both ends.

Solution:
at x1 = 0 and x1 = L:
 
u2 (x1 = 0) = 0 u2 (x1 = L) = 0
and (9.21)
u02 (x1 = 0) = 0 u02 (x1 = L) = 0

2. Using these boundary conditions, determine the condition on the load P for which the
beam can be in equilibrium in a deformed configuration, (i.e. we have a non-trivial
solution ū2 6= 0).
Solution:
for ū2 (x1 = 0) = 0:
r r
P P
ū2 (x1 = 0) = A sin( × 0) + B cos( × 0) + C × 0 + D
EI33 EI33
= B+D =0

for ū02 (x1 = 0) = 0:

r r r r
P P P P
ū02 (x1 = 0) = A cos( × 0) − B sin( × 0) + C
EI EI33 EI33 EI33
r 33
P
= A+C =0
EI33

for ū2 (x1 = L) = 0:

r r
P P
ū2 (x1 = L) = A sin( × L) + B cos( × L) + C × L + D
EI33 EI33

for ū02 (x1 = L) = 0:

r r r r
P P P P
ū02 (x1 = L) = A cos( × L) − B sin( × L) + C
EI33 EI33 EI33 EI33

We obtain the following system:

9.4. BUCKLING LOADS AND SHAPES FOR DIFFERENT BEAM BOUNDARY CONDITIONS221

 
q0 1 0 1
    P
0

1 0  
 0   EI33  A 

     

0  q q  B
= 

 0   sin( EIP33 × L) cos( EIP33 × L) L 1  C 
     

0   D
 q q q q 
P P P P
EI33
cos( EI33
× L) − EI33
sin( EI33
× L) 1 0

For a non trivial solution:

   

 A 
  0 
    

B 0
6=

 C 
  0 
    

D 0

which requires the matrix to be

q singular, i.e. its determinant must vanish. Let’s call
this matrix H and define k = EIP33 .

k 1 0 k 0 1
det(H) = − sin(kL) L 1 − sin(kL) cos(kL) L
k cos(kL) 1 0 k cos(kL) −k sin(kL) 1
k 1 cos(kL) L sin(kL) cos(kL)
= + −k −
k cos(kL) 1 −k sin(kL) 1 k cos(kL) −k sin(kL)
= k(1 − cos(kL)) − k(cos(kL) + Lk sin(kL)) + k sin2 (kL) + k cos2 (kL)
= k − k cos(kL) − k cos(kL) − Lk 2 sin(kL) + k
= 2k − 2k cos(kL) − Lk 2 sin(kL)
 
kL
= 2k 1 − cos(kL) − sin(kL)
2
 
2 kL kL kL kL
= 2k 1 − [1 − 2 sin ( )] − × 2 sin( ) cos( ))
2 2 2 2
 
kL kL kL kL
= 4k sin( ) sin( ) − cos( )
2 2 2 2
= 0,

which implies the three cases:

(a) k = 0
(b) sin( kL
2
) = 0, or
2nπ
k= (9.22)
L
222 MODULE 9. STABILITY AND BUCKLING

(c) sin( kL
2
)− kL
2
cos( kL
2
) = 0, or

kL kL
tan( )= . (9.23)
2 2

In Case (??), we can replace k by its original expression and obtain

r
P 2nπ
=
EI33 L
hence:
4n2 π 2 EI33
Pcrn =
L2
In this case, the displacement ū2 reads:
 
2nπ
ū2 = B (cos(kx1 ) − 1) = B cos( x1 ) − 1
L

In Case (??), the solution is also a series of numbers due to the periodicity of tangential
function. The solutions can be obtained numerically, and the first two are 8.97L
(= 2.85π
L
),
15.45 4.92π
L
(= L
), which lead to

80.76EI33 8.18π 2 EI33 238.72EI33 24.19π 2 EI33

Pcr = k 2 EI33 = (= ), (= ), ...
L2 L2 L2 L2
Set B = 1 and then coefficients A, C, D can be determined by solving the reduced
linear system.
The first two deformation modes from Case (??) and Case (??) are plotted in Figure
9.4.2.

3. Determine the Euler critical load Pcr0 and compare the expression obtained with those
found for the simply-supported and the cantilever beam. Solution:

4π 2 EI33
Pcr0 =
L2
we obtain:

(Pcr0 )clamped = 4 × (Pcr0 )simply−supported = 4 × (Pcr0 )cantilever

9.4. BUCKLING LOADS AND SHAPES FOR DIFFERENT BEAM BOUNDARY CONDITIONS223

2
k = 2π/L
1.5 k = 2.85π/L
k = 4π/L
1 k = 4.92π/L

0.5
ū2 /B

−0.5

−1

−1.5

−2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x1 /L

Figure 9.9: Deformation modes of the clamped beam.

224 MODULE 9. STABILITY AND BUCKLING

e2
a

b P

L e1

Figure 9.10: Clamped beam at both ends with an intermediate support at x = a.

Concept Question 9.4.3. Buckling of a uniform beam clamped at both ends with an inter-
mediate support
Consider the uniform beam of length L, clamped at both ends (Figure 9.8) loaded by a force
P at the right end (x1 = L) along the beam neutral axis. An additional support is placed
at the cross-section x1 = a, as shown in the figure.

1. The analysis is done considering the left and right regions as separate solutions and
then enforcing compatibility at the support. The transverse displacement is denoted
ū2 and ũ2 in the first and second region, respectively.
Determine the general form of the transverse displacement ū2 and ũ2 in both regions.
For convenience, we introduce k 2 = P/EI33 . Solution: The differential equations
governing the transverse displacement in both regions, 1 and 2 are the following:

(IV ) P 00
ū2 + ū = 0 for 0 ≤ x1 ≤ a
EI33 2
(IV ) P 00
ũ2 + ũ = 0 for a ≤ x1 ≤ L
EI33 2
Thus, both, ū2 and ũ2 have the same following form:

ū2 (x1 ) = A1 sin(kx1 ) + B1 cos(kx1 ) + C1 x1 + D1

ũ2 (x1 ) = A2 sin(kx1 ) + B2 cos(kx1 ) + C2 x1 + D2

2. Determine the boundary conditions on the beam: Solution: The boundary

conditions read for ū2 :

ū02 (x1 = 0) = 0 ⇒ kA1 + C1 = 0

ū2 (x1 = 0) = 0 ⇒ B1 + D1 = 0
ū2 (x1 = a) = 0 ⇒ A1 sin(ka) + B1 cos(ka) + C1 a + D1 = 0

and for ũ2 :

ũ02 (x1 = L) = 0 ⇒ kA2 cos(kL) − kB2 sin(kL) + C2 = 0

ũ2 (x1 = L) = 0 ⇒ A2 sin(kL) + B2 cos(kL) + C2 L + D2 = 0
ũ2 (x1 = a) = 0 ⇒ A2 sin(ka) + B2 cos(ka) + C2 a + D2 = 0
9.4. BUCKLING LOADS AND SHAPES FOR DIFFERENT BEAM BOUNDARY CONDITIONS225

3. Are the previously found boundary conditions enough to compute the solution on both
sides of the additional support? If not, what other conditions must be satisfied by ū2
and ũ2 on both sides of the additional support? Solution: At this
stage we are two equations short to fully compute the transverse displacement on both
side of the additional support. We are still missing continuity conditions of the slope
and the bending moments – We have already enforced continuity of displacements at
the intermediate support by applying the null displacement condition at x1 = a in the
previous question. These continuity conditions read:
ū02 (x1 = a) = ũ02 (x1 = a) (continuity of the slopes)
ū002 (x1 = a) = ũ002 (x1 = a) (continuity of the bending moments)

4. Apply the boundary conditions only and show that the displacements: ū2 and ũ2 can
respectively be written as:
ū2 = A ((cos(ka) − 1)(sin(kx1 ) − kx1 ) − (sin(ka) − ka)(cos(kx1 ) − 1))
ũ2 = C ((cos(k(L − a)) − 1)(sin(k(L − x1 )) − k(L − x1 ))
− (sin(k(L − a)) − k(L − a))(cos(k(L − x1 )) − 1))

Solution:
Replacing Bi , Ci , Di by Ai , we can obtain:
 
sin(ka) − ka
ū2 = A1 sin(kx1 ) − kx1 − (cos(kx1 ) − 1)
cos(ka) − 1
 
sin(k(L − a)) − k(L − a)
ũ2 = A2 sin(k(L − x1 )) − k(L − x1 ) − (cos(k(L − x1 )) − 1)
cos(k(L − a)) − 1
Then we can define A = A1 /(cos(ka) − 1) and C = A2 /(cos(k(L − a)) − 1) to simplify
the expression.
5. Apply the additional conditions to both ū2 and ũ2 and derive a system of two equations
depending on: A, C. What condition should satisfy the system of equation so that
non-trivial solutions are found? Solution:
Apply the two continuity conditions and use trigonometric idenities, we can obtain:
A[2 − 2 cos(ka) − ka sin(ka)] + C[2 − 2 cos(kâ) − kâ sin(kâ)] = 0
A[− sin(ka) + ka cos(ka)] − C[− sin(kâ) + kâ cos kâ] = 0
where â = L − a. The non-trivial solution can be found when the determinant of the
2-by-2 linear system is zero, i.e. when
0 =[2 − 2 cos(ka) − ka sin(ka)] × [sin(kâ) − kâ cos kâ]
+ [2 − 2 cos(kâ) − kâ sin(kâ)] × [sin(ka) − ka cos(ka)]
226 MODULE 9. STABILITY AND BUCKLING

6. Let us introduce the following non-dimensional quantities u = a/L and α = kL, and
rewrite the previously found condition.
Solution: With definition u = a/L,
α = kL, and the â = L − a we just defined above, we can get identities: ka = αu and
kâ = α(1 − u). The previous condition now can be written as

0= [2 − 2 cos(αu) − αu sin(αu)] × [sin(α(1 − u)) − α(1 − u) cos(α(1 − u))]

+[2 − 2 cos(α(1 − u)) − α(1 − u) sin(α(1 − u))] × [sin(αu) − αu cos(αu)]

7. Determine the location of the intermediate support (a = a∗ ) for which the buckling
load will attain a maximum, hence the best location of the intermediate support to
avoid buckling.
Solution:
For a = a∗ , the buckling load will attain a maximum value and corresponds herein to
u∗ = a∗ /L. The previously found condition is an equation of α and u. In other word,
α may be viewed as an implicit function of u. The previous equation can be written
as follows using trigonometric identities:

f (α(u), u) =[2 − 2 cos(αu) − αu sin(αu)] × [sin(α(1 − u)) − α(1 − u) cos(α(1 − u))]

+ [2 − 2 cos(α(1 − u)) − α(1 − u) sin(α(1 − u))] × [sin(αu) − αu cos(αu)]
 2 
= α u(1 − u) − 2 sin(α) + α cos(α) + α cos(αu) cos(α(1 − u))
+ 2 sin(α(1 − u)) + 2 sin(αu) − 2α(1 − u) cos(α(1 − u)) − 2αu cos(αu)
=0 .

Let’s derive the function f with respect to u:

∂f dα ∂f
+ =0.
∂α du ∂u
Thus,
 −1  
dα ∂f ∂f
=− ,
du ∂α ∂u
assuming that ∂f /∂α 6= 0. The condition:

dα
=0,
du
is then equivalent to:
∂f
=0,
∂u
which leads to:

(1 − 2u)α2 sin(α) + α2 sin(α(1 − 2u)) − 2α2 (1 − u) sin(α(1 − u)) + 2α2 u sin(αu) = 0 .

9.5. BUCKLING OF BEAMS WITH IMPERFECTIONS 227

We immediately identify that u∗ = 1/2 satisfies this equation, and therefore dα

du
(u∗ ) = 0.
∗
By substituting u = u = 1/2 into the previous equation f (α(u), u) = 0, we can obtain:

α = 4π .

In sum, the location of the intermediate support that maximizes the buckling load is
in the middle of the beam. The buckling load is equal to:

16π 2 EI
Pcr = .
L2

9.5 Buckling of beams with imperfections

So far, we have assumed idealized beams with mathematically exact geometries, made of
perfectly homogeneous materials and loads perfectly aligned with the centroid axis. In re-
ality, beams have imperfections due to the fabrication process and cannot be considered as
homogeneous or geometrically exact. In this section, we study the effect of these imperfec-
tions.

Concept Question 9.5.1. Buckling of a simply supported beam with an imperfection

We will account for any geometric imperfection in the material as an eccentricity in the
application of the load. Consider a simply-supported beam of length L as shown in Figure
9.5.1. The uniform compressive load applied at the free end has an eccentricity e.

e2

e1 b
b
e
P
L

Figure 9.11: Simply supported beam with eccentric end load

1. what do you think is the main difference with the idealized buckling problem? How
does the influence of the eccentricity affect the analysis? Solution: The main
difference is that the load P now produces a bending moment even in the undeformed
configuration. We will call this the primary bending moment. This is the moment that
we would need to consider in the absence of structural instability considerations, i.e.
in linear beam theory. The analysis changes in a fundamental way, as now the problem
has non-homogeneous boundary conditions.
228 MODULE 9. STABILITY AND BUCKLING

2. how do you think the governing equation changes with respect to the idealized buckling
problem? Solution: It doesn’t change at all, the differential equation is still the
homogeneous equation used in the idealized case, as there is no distributed transverse
force, i.e. p2 = 0

3. Write the boundary conditions needed to determine the constants A, B, C and D in

the solution:
r r
P P
ū2 (x1 ) = A sin( x1 ) + B cos( x1 ) + Cx1 + D
EI33 EI33
Solution:
at x1 = 0 and x1 = L:
 
u2 (x1 = 0) = 0 u2 (x1 = L) = 0
and
M3 (x1 ) = EI33 u002 (x1 = 0) = −P e M3 (x1 ) = EI33 u002 (x1 = L) = −P e

4. Apply the boundary conditions and find the solution ū2 .

Solution: for ū2 (x1 = 0) = 0:
r r
P P
ū2 (x1 = 0) = A sin( × 0) + B cos( × 0) + C × 0 + D
EI33 EI33
= B+D =0

for M3 (x1 = 0) = −P e:
r r
P P
ū002 (x1 = 0) = −P A sin( × 0) − P B cos( × 0)
EI33 EI33
= −P B = −P e
→ B=e

hence D = −e.
for ū2 (x1 = L) = 0:
r r
P P
ū2 (x1 = L) = A sin( × L) + e cos( × L) + C × L − e = 0
EI33 EI33

for M3 (x1 = L) = −P e:
r r
P P
M3 (x1 = L) = −AP sin( × L) − eP cos( × L) = −P e
EI33 EI33
q
(1 − cos( EIP33 × L))
→ A=e q
sin( EIP33 × L)
9.5. BUCKLING OF BEAMS WITH IMPERFECTIONS 229

hence, C = 0. we finally obtain the displacement ū2 as a function of e:

 q 
 (1 − cos( EIP × L)) r r 
33 P P
ū2 (x1 ) = e q sin( x1 ) + cos( x1 ) − 1
 sin( P
× L) EI33 EI33 
EI33

5. Notice that we did obtain a fully defined solution in this case. No bifurcation of
equilibrium in this case? How come? What happens to the solution as P approaches
the critical load? Solution: We
found a solution because this is a non-homogeneous problem, i.e. solving the governing
equation responds to the question: how does the beam deform under the application
of the load p0 (and how does the influece of P modify that response. This is a very
different question from: under what conditions (what values of P ) would this beam be
able to be in equilibrium in the deformed configuration (in addition to the undeformed
configuration)? So there is no bifurcation of equilibrium.
It can be seen in the solution that when P approaches the critical load the displacements
grow unboundedly, i.e., the beam would fail.

6. Determine the relation P = f (ū2 (L/2)) at the middle of the beam and plot this ex-
pression for different ratios e/L
q
Solution: Let’s use k = EIP33
 
L (1 − cos(k × L)) L L
ū2 (x1 = ) = e sin(k ) + cos(k ) − 1
2 sin(k × L) 2 2
( )
L
(1 − cos(2k × 2 )) L L
= e sin(k ) + cos(k ) − 1
sin(2k × L2 ) 2 2
( )
(1 − cos(2k × L2 )) L L
= e sin(k ) + cos(k ) − 1
2 sin(k × L2 ) cos(k × L2 ) 2 2
( )
(1 − cos(2k × L2 )) L
= e L
+ cos(k ) − 1
2 cos(k × 2 ) 2
( )
2 sin2 (k × L2 )) L
= e + cos(k ) − 1
2 cos(k × L2 ) 2
( )
1
= e −1
cos(k × L2 ))

and we have: !
4 e
P = EI33 2 arccos2 L
L ū2 ( 2 ) + e
230 MODULE 9. STABILITY AND BUCKLING

so !
P 4 e
= 2 arccos2 L
Pcr π ū2 ( 2 ) + e

π 2 EI33
with Pcr =
L2
7. Draw the function f for several values of the ratio e/L and interpret the result.
Solution: For low values of the load P : 1) an increase of the load leads to an
essentially linear growth of the deflection at the center, 2) increasing the eccentricity
e increases the primary moment (for a fixed P ) and thus the deflection.
As the load increases and becomes a significant fraction of the critical load, the behavior
deviates from the solution of linear elasticity and the secondary moment produced by
the load P with the extra moment arm corresponding to the deflection further increases
the deflection. When the load gets close to the critical value, the deflections grow
unbounded.
The plots also bears the interpretation that as the imperfections disappear there is a
smooth transition to the solution of the bifurcation problem.

1.2

0.8
P/Pcr

0.6

0.4 e/L = 0.01

e/L = 0.05
e/L = 0.1
0.2 e/L = 0.2
e/L = 0.5
e/L = 1.0
0
0 2 4 6 8 10
ū2 (L/2)/L

Figure 9.12: simply supported beam with eccentric end load

8. Find the distribution of the bending moment Solution: It follows directly that:
q
h 1 − cos EIP L r r i
33 P P
M3 = EI33 u002 = −P e q sin x1 + cos x1
sin EIP33 L EI33 EI33
9.6. OTHER ISSUES IN BUCKLING INSTABILITY 231

9. Interpret the result in the limits P → 0, Pcr Solution:

P → 0, M3 → −P e

i.e. the primary bending moment obtained when equilibrium in the undeformed con-
figuration is considered

P
P → Pcr , L → π M3 → ∞
EI33
i.e. as the load reaches the critical load, the moment grows unbounded.

9.6 Other issues in buckling instability

Concept Question 9.6.1. We saw that beams and columns under states of strong com-
pression buckle.

1. Is this always true? If not, what other considerations come into play and when would
that happen? Solution: Buckling occurs if the compressive force approaches the
critical value which in all cases scales as π 2 EI
L2
. If the beam is long, the critical load is
low and the beam buckles under fairly low loads which implies that the stresses in the
material are low as well (poor structural efficiency).
For short beams, the critical load increases quadratically with the reduction in length,
which means that the likelihood of buckling decreases, whereas the stress σ11 = PA can
increase with P to high values and reach material limits.

2. In order to start looking at this problem, let’s write the critical load for general bound-
ary conditions as:
EI EI EI
Pcr = cπ 2 2 = π 2 = π 2 02
L L L
( √ )2
c
|{z}
L0

where we define c as the coefficient of fixity which depends on the boundary condition
(e.g. c = 1 for simply supported, c = 4 for clamped-clamped, c = 1/4 for cantilever,
etc). L0 = √Lc as the equivalent length for buckling.

3. In order to compare the competition between buckling and material failure by compres-
sion, evaluate give an expression for the stress in the material when the load approaches
the critical value Solution: This is simply:
   ρ 2
Pcr 2 E I
σ11 = = π 02 = π2E
A L A L0
| {z }
ρ2
232 MODULE 9. STABILITY AND BUCKLING

where we have defined: r

I
ρ≡
A
as the radius of gyration. (Interpretation?)
0
4. Define the beam slenderness ratio as λ = Lρ and plot the “buckling stress” as a
function of λ. Superimpose in your plot the material limiting stress (yielding, crushing)
and define regions of beam response as a function of the slenderness ratio (buckling,
crushing or squashing and transition between the two. Solution: