MODULE 3 – ELASTIC BUCKLING OF COLUMNS

3.1 CRITICAL LOAD OF THE EULER COLUMN

Even as simple a structural element as an axially loaded member behaves in a
fairly complex manner. It is desirable to begin the study of columns with a very idealized
case, the Euler column* (*The Euler column takes its name from the man who, in year
1744, presented the first accurate stability analysis of a column. Although it is customary
today to refer to a simply supported column as an Euler column, Euler in fact analyzed a
member fixed at one end and free at the other end in his famous treatise.)
The axially loaded member shown in Fig 3.1 is assumed to have a constant cross-
sectional area and to be made of homogeneous material. In addition, four assumptions are
made:
1. The ends of the members are simply supported. The lower end is attached to
an immovable hinge, and the upper end is supported so that it can rotate freely
and move vertically but not horizontally.
2. The member is perfectly straight, and the load is applied along its centroidal
axis.
3. The material obeys Hooke’s Law.
2
 dy 
4. The deformations of the member are small enough so that the term   or
 dx 
(y’) is negligible compared to unity in the expression for the
2

y"
curvature . Hence the curvature can be approximated as y” or
 
3
1   y '
2 2

 d2y 
 2  .
 dx 
In accordance with the criterion of neutral equilibrium, the critical load is that load for
which equilibrium in the slightly bent configuration shown in Fig 3.1 is possible.
 P x

○ ○
P
Mx
y”
L y
y

P
P
(a) (b) (c)

Fig. 3.1 Euler Column

d2y
dx 2 d2y
The curvature   3
  y"
dx 2
 dy  2  2
   1
 dx  

If the coordinate axes are taken as shown in the figure, the internal moment at any
section, a distance x from the origin, is

M x  EI ()  EIy"

Equating this expression to the externally applied bending moment, Py, gives

EIy" Py  0 (3.1)

Eq. (3.1) is a linear differential equation with constant coefficients. As such, its solution
is readily obtained. However, before considering this solution, let us digress somewhat
and see what form Eq. (3.1) takes if the foregoing simplifying assumptions are not made.
If the assumptions of elastic behavior and small displacements are not made, the modulus
y"
E in Eq. (3.1) becomes a variable and the curvature y” is replaced by .
 
3
1   y '
2 2
The equation then has neither constant coefficients nor is it linear; as a result its solution
is difficult to obtain. If the assumptions of concentric loading and hinged supports are not
made, there are additional terms on the right-hand side of the equation. This makes the
equation nonhomogeneous, but does not make the task of obtaining a solution difficult.
The solution to Eq. (3.1) will now be obtained. Introducing the notation

P
K2  (3.2)
EI
Eq. (3.1) becomes

y" K 2 y  0 (3.3)

The solution of homogeneous linear differential equations with constant coefficients is

always of the form

y  e mx

Substitution of this expression into Eq. (3.3) leads to

m   iK

Hence, the general solution of Eq. (3.3) is

y  c1e iKx  c 2 e iKx

Making use of the relation

e  iKx  cos Kx  i sin Kx

and the fact that both the real and imaginary parts of a complex function that satisfies Eq.
(3.3) must also satisfy the equation, the general solution can be rewritten in the form

y  A sin Kx  B cos Kx (3.4)

To evaluate the arbitrary constants A and B, we make use of the boundary conditions

y = 0 at x = 0
(3.5)
y = 0 at x = L

The first of these conditions when substituted into Eq. (3.4) leads to

B=0
Consequently

y  A sin Kx (3.6)

From the second condition one obtains

A sin KL  0

This relation can be satisfied in one of two ways; either

A=0

or sin KL  0

If A = 0, K and consequently P can have any value. This result is known as the trivial
solution, because it confirms what is already known, that a column is in equilibrium
under any axial load P as long as the member remains perfectly straight. If sin KL  0

KL  n

Where n = 1, 2, 3, … Substitution of this expression into Eqs. (3.2) and (3.6) leads to

 2 EI
Pn 2
(3.7)
L2

nx
And y  A sin (3.8)
L

At the loads given by Eq. (3.7) the column can be in equilibrium in a slightly bent form.
The shape of the deformation is given by Eq. (3.8). However, its amplitude is
intermediate, since A can have any value when sin(KL)  0 .
The value of P, obtained by setting n equal to 1, is

 2 EI
P (3.9)
L2

This load is known as the Euler load. It is the smallest load at which a state of neutral
equilibrium is possible. Hence it is also the smallest load at which the column ceases to
be in stable equilibrium. It is also called the buckling load.
 P

2
 EI
P
2
L

Fig. 3.2 Behavior of Euler Column

3.2 LINEAR COLUMN THEORY – AN EIGENVALUE PROBLEM

The small-deflection column theory presented in Article 3.1 is based on a linear
differential equation and is, known as the linear column theory. By comparison, the large-
deflection theory of columns to be studied later is based on nonlinear differential
equation and is referred to as a nonlinear column theory. Although the small-deflection
theory can be considered linear when contrasted with large-deflection theory, it is not
linear in the same sense of the word as simple beam bending theory. In the latter,
equilibrium is based on undeformed geometry and deflections are proportional to the
applied loads, which is certainly not true for the Euler column.
Thus linear column theory is evidently different from both flexure and large-
deformation column theory. As a matter of fact, it does belong to an entirely different
class of problems known as eigen value problems. These problems are characterized by
the fact that nonzero solutions for the dependent variable exist only for certain discrete
values of some parameters. The values of the parameters, for which nonzero solutions
exist, are known as eigen values, and the solutions as eigenvectors. Only the shape and
not the amplitude of the eigenvector can be determined in an eigen value problem. In a
n 2  2 EI
stability problem such as the axially loaded column, the loads at which nonzero
L2
deflections are possible are the eigen values, and the deflected shapes that can exist at
these loads are the eigenvectors. The smallest eigen value is the critical load and the
corresponding eigen vector is the buckling mode shape.
3.3 BOUNDARY CONDITIONS
The first step toward generalizing the results obtained in Article 3.1 is to consider
boundary conditions other than hinged-hinged supports.

Case 1. Both ends fixed

x
P
M0
P
y
L

4 M = -EIy”
●

L inflection L equivalent
points 2 Euler column
M 0  M  Py  0
● L

4 y
M0 M0

P
P

(a) (b)

Fig. 3.3 Fixed – Fixed Column

Equating the external moment to the internal moment, at a section a distance x from the
origin, Fig. (3.3) one obtains

EIy"Py  M 0

M0
y" K 2 y  (3.10)
EI

P
where K2 
EI

The solution to Eq. (3.10) consists of a complementary and a particular part. The
complementary part is

y c  A sin Kx  B cos Kx
Let the particular solution yp be

M  K2
y p  m 0   m M0
 EI  P

d2 yp
0
dx 2

Substituting into the differential equation yields

d2 yp  K2 
2
 K 2 y p  M 0  
dx  P 

 K2   K2 
0  K 2 mM 0    M 0  
 P   P 

1
m
K2

 K2  1 K 2M0 M0
Hence y p  m M 0  2 
 P  K P P

The solution is y  y c  y p

M0
or y  A sin Kx  B cos Kx  (3.11)
P

where A and B are arbitrary constants to be evaluated using the boundary conditions.

y’ = 0 , y = 0 at x = 0

y = 0 at x = L

The first two conditions are satisfied if

M0
A = 0 and B
P

M0
Hence y (1  cos Kx )
P
The last boundary condition leads to the transcendental equation

cosKL = 1.0

The smallest nonzero root to this equation is

KL = 2

from which

4 2 EI
Pcr  (3.12)
L2

M0 2x
and y (1  cos ) (3.13)
P L

Eq (3.12) indicates that the critical load of a column with fixed ends is four times as large
as the critical load of a hinged-hinged column.

Determine the point of inflection

M0
y (1  cos Kx )
P

dy M 0
 (0  K sin Kx )
dx P

d2y M0
2
 (0  K 2 cos Kx )  0
dx P


For cosKx = 0; Kx 
2

2
But K
L

2  L
Therefore x or x from A
L 2 4

L
By symmetry, x  from B
4
 L L L
Thus, Le  L     
4 4 2

 2 EI 4 2 EI
Pcr   as derived before (3.14)
L e 2 L2

The length Le of this equivalent Euler column is called the effective length of the
member.

Case 2: One end fixed and one end free

P

P
y Equivalent
L Euler
M = -EIy”
Column
L

y
L
P
P

M 0 0
P  M  Py  0 P
P  EIy" Py  0

Fig. 3.4 Fixed – free column

EIy" Py  P

y"K 2 y  K 2  (3.15)

P
Where K2 
EI
The solution to Eq. (3.15) is

y  yc  yp

y c  A sin Kx  B cos Kx

y p  mK 2 

Find the value of m

d2 yp
0
dx 2

Substituting into Eq (3.15)

 
0  K 2 mK 2   K 2 

1
m
K2

yp 
1
 
K2   
K2

Therefore

y  A sin Kx  B cos Kx   (3.16)

The boundary conditions are

y = 0 and y’ = 0 at x = 0

They are satisfied if

A=0 ; B = - 

Hence y  (1  cos Kx )

The boundary condition at the upper end of the member

y =  at x = L

is satisfied if

cosKL = 0
The smallest nontrivial root of this equation is


KL 
2

which leads to

 2 EI
Pcr  (3.17)
4L2

Eq. (3.17) indicates that the critical load of a column fixed at one end and free at the other
is one fourth the Euler load. In other words, the effective length of the equivalent Euler
column is therefore equal to 2L and the critical load of a fixed-free column can be
expressed in the form

 2 EI
Pcr  (3.18)
2L2
x
and y  (1  cos ) (3.19)
2L

Case 3: One end fixed and one end hinged

x
P
M0
M0
P
y
M
L
M0

L
L
x

M0
M0 L
L
y
M0

P
P

Fig. 3.5
The equilibrium of forces and moments yields

M0
M x  Py  0
L

M0
 EIy"Py   x
L

M0
EIy" Py  x (3.20)
L

M0 x
or y" K 2 y  .
EI L

P
where K2 
EI

The solution to Eq. (3.20) is

y  yc  yp

where y c  A sin Kx  B cos Kx

M x
and y p  m 0 . 
 EI L 

dy p M0 1
m
dx EI L

d2 yp
 m(0)
dx 2

Substituting in the differential equation

 M x M x
0  K2m 0   0
 EI L  EI L

1
m
K2

1 M0 x M0 x
Therefore yp  
K 2 EI L P L
Thus, the solution to Eq. (3.20) is

M0 x
y  A sin Kx  B cos Kx  (3.21)
P L

The boundary conditions are

y = 0 at x = 0

y’ = 0 at x = L

and are satisfied if

M0 1
B= 0 ; A 
P KL cos KL

M0 x sin Kx 
Hence y   
P  L KL cos KL 

The boundary condition

y =  at x = L

leads to the transcendental equation

tanKL = KL

The smallest nonzero root of this equation is

KL = 4.49

which leads to

20.2EI
Pcr  (3.22)
L2

M0 x  x 
and y   1.02 sin 4.49   (3.23)
P L  L 

The corresponding deflection curve has an inflection point at x = 0.7L. Hence the
effective length is 0.70L and the critical can also be expressed as

 2 EI
Pcr  (3.24)
0.7L2
Case 4: Elastically restrained end

P Q
L P+Q
M M
L
EI M

Q
L
EI
x

M
y L

P P+Q

Fig. 3.6

P
y
-EIy”=Mx
M

M0

P
 M
Mx  x  Py  0
L

M
 EIy " Py   x
L

M x
y" K 2 y  . (3.25)
EI L

P
where K2 
EI

The solution to Eq. (3.25) is

Mx
y  A sin Kx  B cos Kx  (3.26)
P L

Note: Please refer to previous case for the particular solution

The boundary condition at the lower end of the member

y = 0 at x = 0

is satisfied if B= 0

and the condition at the upper end

y = 0 at x = L

requires that
M 1
A
P sin KL

M  x sin Kx 
Hence y   
P  L sin KL 

Since the column is rigidly connected to the beam, the rotation of the column at its upper
end must be equal to the rotation of the left-hand end of the beam. For the column, the
shape at x = L is

M  1 K cos KL 
y'    
P L sin KL 
 M1 K 
or y'     (3.27)
P  L tan KL 

M  1 1 
y'    
KEI  KL tan KL 

The rotation  of the left-hand end of the beam is obtained from the slope deflection
equation. Thus

M
2EI
2
L

from which

ML
 (3.28)
4EI

Equating the expressions in (3.27) and (3.28) gives

ML M  1 1 
   
4EI KEI  KL tan KL 

Note the negative sign in front of the right-hand side of the equation is needed because
the slope as given by (3.27) is negative, whereas  is positive.

KL  1 1 
Thus,     (3.29)
4  KL tan KL 

Eq. (3.29) is the transcendental equation for the column being studied. It is convenient to
rewrite the equation in the form

4KL
tan(KL) 
KL 2  4
The smallest root satisfying this expression is

KL = 3.83

from which

14.7EI
Pcr  (3.30)
L2
3.4 HIGHER-ORDER DIFFERENTIAL EQUATION FOR COLUMNS
In the previous section buckling loads were obtained for columns with various
boundary conditions. In each case, a second-order differential equation, valid only for the
member being analyzed, was used. However, as shown by Timoshenko and Gere, a
single fourth order equation applicable to any column regardless of the boundary
conditions can also be employed. This equation will be introduced now, and its solution
obtained for two specific sets of end restraints.
For the general case of a column with unspecified boundary conditions, there
exists a moment and a shear at each end of the member (Fig 3.7). By choosing
appropriate values for these moments and shears, any desired set of boundary conditions
can be satisfied. Equating the external moments to the internal moment, at any section a
distance x from the origin leads to

x
P
MB P
y
Q
Mx
●

L x

Q
●
Q MA
MA
P
P

Fig. 3.7 Column with unspecified boundary conditions

M x  M A  Qx  Py  0

 EIy"Py  Qx  M A  0 (3.31)

EIy"Py  Qx  M A

Differentiating with respect to x

EIy' ' ' Py'  Q  0 (3.32)

 EIy' ' ' ' Py' '  0 (3.33)

and introducing the notation

P
K2 
EI

The equation becomes

y iv  K 2 y' '  0 (3.34)

This fourth-order differential equation is applicable to any set of boundary conditions. Its
general solution is

y  c1 sin Kx  c 2 cos Kx  c 3 x  c 4 (3.35)

The arbitrary constants in the solution are determined from the boundary conditions of
the specific case under investigation

Case 1: Hinged-hinged
For the hinged-hinged column (Fig 3.8) the deflection and bending moment are
zero at both ends of the member. Hence

Fig. 3.8 Hinged-hinged

 y = 0 at x = 0 (1)

EIy” = 0 at x = 0 (2)

y = 0 at x = L (3)

EIy” = 0 at x = L (4)

From Eq (3.35)

y  c1 sin Kx  c 2 cos Kx  c 3 x  c 4

y'  c1K cos Kx  c 2 K sin Kx  c 3

y' '  c1K 2 sin Kx  c 2 K 2 cos Kx

y = 0 at x = 0 : 0  c2  0  c4  0 (1)

y” = 0 at x = 0 : 0  c2K 2  0  0  0 (2)

y = 0 at x = L : c1 sin KL  c 2 cos KL  c 3 L  c 4  0 (3)

y” = 0 at x = L :  c1K 2 sin KL  c 2 K 2 cos KL  0  0  0 (4)

From (2) c2  0

From (1) c 4  0 since c 2  0

From (3) c1 sin KL  c 3 L  0

From (4)  c1K 2 sin KL  0  0

Since K2 is not identically equal to zero, these relations are satisfied either if

c1 = c 3 = 0

or if sinKL = c3 = 0

The first alternative leads to the trivial solution of equilibrium at all loads provided the
member remains straight. From the second conditions, one obtains

KL = n n = 1, 2, 3 …
which for n = 1 leads to the critical load

 2 EI
Pcr  2 (3.36)
L

Another method of finding the critical load:

The four boundary conditions yields the following four equations:

0  c2  0  c4  0

0  c2K 2  0  0  0

c1 sin KL  c 2 cos KL  c 3 L  c 4  0

 c1K 2 sin KL  c 2 K 2 cos KL  0  0  0

A nontrivial solution exists if all four constants are not identically equal to zero. This can
happen only if the determinant of the coefficients of c i’s vanishes or

c1 c2 c3 c4
 0 1 0 1
 0 K 2
0 0
 0
 sin KL cos KL L 1
 
 K sin KL  K cos KL 0
2 2
0

Subtract Row 3 from Row 1

  sin KL 1  cos KL  L 0
 0  K2 0 0
 0
 sin KL cos KL L 1
 
 K sin KL  K cos KL 0
2 2
0

The remaining rows and columns are as follows

  sin KL 1  cos KL  L
 0  K2 0   0

 K 2 sin KL  K 2 cos KL 0 
This can be expanded as:

 sin KL  K 2 0  1  cos KL 0 K 2 sin KL   L K 2 cos KL 0 

(1) (3 4)  0
  K sin KL  K  L    K cos KL 0 sin KL   01  cos KL 0
2 2 2

or 
(1) LK 4 sin KL  0 
The solution of this transcendental equation is

KL = n n = 1, 2, 3 …

and the critical load is

 2 EI
Pcr  for n = 1
L2

Case 2: Fixed-free

x P


P
P

Fig 3.9 Fixed-free

The governing differential equation is

y iv  K 2 y' '  0

The boundary conditions are:

(1) y = 0 at x = 0

(2) y’ = 0 at x = 0
(3) y” = 0 at x = L (Moment = 0)

(4) y’’’ +K2y’= 0 at x = L (Shear = 0 )

Let y  c1 sin Kx  c 2 cos Kx  c 3 x  c 4

From the above equation:

y'  c1K cos Kx  c 2 K sin Kx  c 3  0

y' '  c1K 2 sin Kx  c 2 K 2 cos Kx  0  0

y' ' '  c1K 3 cos Kx  c 2 K 3 sin Kx  0  0

Using the boundary conditions

c1 (0)  c 2 (1)  c 3 (0)  c 4 (1)  0

c1 (1)  c 2 (0)  c 3 (1)  c 4 (0)  0

 c1K 2 sin KL  c 2 K 2 cos KL  c 3 (0)  c 4 (0)  0

 c K1
3
    
cos KL  c1K 3 cos KL  c 2 K 3 sin KL  c 2 K 3 sin KL  c 3 K 2  c 4 (0)  0

The determinant of the above must be zero,

 0 1 0 1
 1 0 1 0
 0
 K sin KL  K cos KL 0
2 2
0
 
 0 0 K2 0

This can be expanded as


(1) (14) 1(K 2 cos KL)K 2  0  0  0  0  0  0

(1)  K 4 cos KL  0 
For a nontrivial solution

cosKL = 0
This leads to


KL  (2n  1) , n = 1, 2, 3 …
2

And for n = 1


KL 
2

 2 EI
And Pcr  (3.37)
4L2

Case 3: Fixed-fixed

x
P
M

●
y
M

The governing differential equation is

y iv  K 2 y' '  0

The boundary conditions are:

(1) y = 0 at x = 0

(2) y = 0 at x = L
(3) y’ = 0 at x = 0

(4) y’ = 0 at x = L

Assuming that y  c1 sin Kx  c 2 cos Kx  c 3 x  c 4

y'  c1K cos Kx  c 2 K sin Kx  c 3  0

Using the four boundary conditions, we have

c1 (0)  c 2 (1)  c 3 (0)  c 4 (1)  0

c1 sin KL  c 2 cos KL  c 3 L  c 4  0

c1 K  c2 (0)  c3 (1)  0  0

c1 K cos KL  c2 K sin KL  c3 (1)  0  0

The characteristic equation for this case is given by

 0 1 0 1
 sin KL cos KL L 1
 0
 K 0 1 0
 
 K cos KL  K sin KL 1 0

Subtract Row 2 from Row 1

  sin KL 1  cos KL  L 0
 sin KL cos KL L 1
 0
 K 0 1 0
 
 K cos KL  K sin KL 1 0

Expansion of this determinant yields the following

  sin KL 1  cos KL  L 
 K 0 1   0

 K cos KL  K sin KL 1 

 
(1) ( 2 4) 0  K cos KL1  cos KL   K 2 L sin KL)  0  K sin 2 KL  K 1  cos KL   0


 (1) K cos KL  K cos 2 KL  K 2 L sin KL  K sin 2 KL  K  K cos KL  0 
2K cos KL  K 2
  
L sin KL  K cos 2 KL  sin 2 KL  K  0

2(cos KL  1)  KL sin KL  0 (3.38)

 KL 
Since (cos KL  1)  2 sin 2  
 2 

KL KL
and sin KL  2 sin cos
2 2

Eq. (3.38) becomes

 KL  KL KL
2  2 sin 2   KL.2 sin cos 0
 2  2 2

 2 KL   KL  KL KL
  sin   sin cos 0
 2   2  2 2

 KL  KL  KL 
sin KL  sin   cos 0
 2  2  2 

Then either

sin KL  0

KL  KL   KL 
or cos   sin  
2  2   2 

The first of these solutions leads to

4n 2 2 EI
P n = 1, 2, 3 …
L2

from which

4 2 EI
Pcr  for n=1 (3.39)
L2
3.5 LARGE-DEFORMATION THEORY FOR COLUMNS
In the preceding articles the behavior of columns was studied using a linear
differential equation. The linear equation was obtained by employing an approximate
d2y
expression , for the curvature of the member. Since this approximation is valid only
dx 2
when the deformations are small, the results obtained from the linear equation are limited
to small deflections. In this article the limitations of small deflections will be removed by
using an exact expression for the curvature.

 P
y
Mx
L 

 x
s
y

P
P
Fig. 3.10 Large deflection of column

Consider the simply supported column shown in Fig. 3.10. Aside from the
assumption of small deflections, which is no longer being made, all other idealizations
made for the Euler column are still valid. The member is initially assumed to be perfectly
straight and loaded along its centroidal axis., and the material is assumed to obey
Hooke’s law.
If the x-y coordinate axes are taken as shown in the figure and the column is in
equilibrium in a bent configuration, then the external moment, Py, at any section, is equal
to the internal resisting moment Mx. Thus

M x  Py (3.40)

EI
But Mx  
R
 1
where is the curvature
R

The solution of the equation is facilitated if the curvature is expressed as the rate of
change of the slope. Letting the slope at any point be given by  and the distance along
the curve from the origin to the point by s, the curvature is

1 d

R ds

Substitution of this expression in Eq. (3.40) leads to

d
EI  Py  0 (3.41)
ds

Eq. (3.41) can be used to study the behavior of columns at large as well as at small
d d2y
deformations. Because is used in place of , Eq. (3.41) is nonlinear, and its
ds dx 2
solution is a good deal more difficult to obtain than that of the equation EIy"Py  0 .
The analysis presented here follows the general outline of the solution given by Wang.
P
Introducing the notation K 2  , Eq. (3.41) takes the form
EI

d
 K2y  0 (3.42)
ds

dy
If the equation is differentiated with respect to s, and is replaced by sin, one obtains,
ds
d 2 dy
2
 K2 0
ds ds

d 2
2
 K 2 sin   0 (3.43)
ds

 d 
Multiplying the terms by 2 ds and integrating both terms gives
 ds 

 d 2   d   d 
  ds 2 2 ds ds   K sin  .2 ds ds  0
2
(3.44)

 d  d    d 
2 2

Since 2  2 ds  d 
 
 ds  ds   ds 
and sin d  d cos 

Eq. (3.44) can be written in the form

 d 
2

 d    2 K 2  d cos   0 (3.45)
 ds 

Carrying out the integration leads to

 d 
2

   2 K cos   c
2
(3.46)
 ds 

d
The constant of integration, c, is evaluated from the condition that  =  and  0 at the
ds
origin. Hence

c  2K 2 cos 

And Eq. (3.46) becomes

 d 
2

   2 K cos  cos 
2

 ds 

Taking the square root of both terms gives

 d 
   K 2 cos  cos 
 ds 

d
ds   (3.47)
K 2 cos  cos 

The negative sign denotes that  decreases as s increases.

If ds is integrated from 0 to L, one obtains the total length of the column.
Thus
L  d
L   ds   (3.48)
0 
K 2 cos  cos 

The minus sign in front of the equation has been eliminated by reversing the limits of
integration.
Making use of the identities
  
cos  1  2 sin 2  
2
(3.49)
 
cos  1  2 sin 2  
2

In Eq. (3.48) leads to

1  d
L 

 2 
(3.50)
2K  
 sin    sin   
 2  2 

To simplify this integral, let

 
P  sin   (3.51)
2

and introduce a new variable  defined by

 
P sin   sin   (3.52)
2

The integral of Eq. (3.50) can be expressed in terms of  and determines the limits on
corresponding to the given limits on . Taking the differential of both sides of Eq.
(3.52), gives

1 
cos d  P cos d
2 2

2 P cosd
d 
 
cos 
2

  
and using the identity cos  1  sin 2  
2 2

one obtains

2 P cosd
d 
1  P 2 sin 2  from Eq. (3.52)
Rewriting Eq. (3.52) in the form

  
sin   sin   sin
2 2
P from Eq. (3.51)


It is evident that  varies from – to , sin varies from -1 to +1, and  varies from 
2

to . In view of Eq. (3.53), and noting that
2

   
sin 2    sin 2    P cos  (3.54)
2 2

Eq. (3.50) can be written in the form


2 d
L
K 0
2

1  P 2
sin 2  
(3.55)

2
or L (3.56)
K


d
in which   2 (3.57)
0
1  P 2
sin 2  
This integral is known as the complete elliptic integral of the first kind. Its values
can be obtained from a table of integrals, which tabulates  for various values of P.
Eq. (3.56) can also be written in the form

2
L
P
EI

P 42
or  2 (3.58)
Pcr 

 2 EI
in which Pcr 
L2

The nonlinear theory to thus leads to the same critical load as the linear theory.
 It is now desirable to determine how the load varies as the deflection of the
member becomes large. Using Eq. (3.56), a curve of load versus  can be plotted, or the
more commonly employed relation between load and mid-height deflection,  can be
obtained as follows.
Noting that dy = sinds and making use of Eq (3.47) gives

sin d
dy  
K 2 cos  cos 
The midheight deflection  is obtained by integrating dy from 0 to . Hence

0  sin d
    dy  
 0
K 2 cos  cos 
The limits for the integral on the right-hand side of the equation were obtained by noting
that  varies from  to 0 as y varies from 0 to . Using the relation in Eq (3.49) gives

1  sin d
   2 
2K  
0
 sin    sin 2   
 2  2 

Except for term sin, this integral is similar the one in Eq. (3.50). As before, a change in
variable from  to  is therefore made, and in view of Eq. (3.51), (3.52), (3.53) and
(3.54), one obtains


1 2 sin (2P cos d)
2K 0
 (3.59)
1  P 2
sin 2 (P cos ) 
It can also be shown that

sin   2 P sin  1  P 2 sin 2 

Substitution of this relation in Eq. (3.59) leads to


2P 2 2P
 
K 0
sin d 
K

from which
  2P
 (3.60)
L P

Pcr

Using Eqs. (3.58) and (3.60), it is possible to compute for various values of  the
P 
corresponding values of and . This is tabulated in Table 3.1
Pcr L
 0 o
20o 40o 60o
  1.583 1.620 1.686
2
P 0 0.174 0.342 0.500
P 1 1.015 1.063 1.152
Pcr
 0 0.110 0.211 0.296
L
3.6 BEHAVIOR OF IMPERFECT COLUMNS
In the derivation of the Euler load, the member is assumed to be perfectly straight
and the loading is assumed to be concentric at every cross section. These idealizations are
made to simplify the problem. However, perfect members do not exist in actual
engineering structures. Both imperfections of shape and small eccentricities of loading
are present in all real columns. It is therefore desirable to investigate the behavior of an
imperfect column and compare it with the behavior predicted by Euler theory.

Case 1: Initially Bent Column

P
y0 y
Mx
L y0 y

P
P

Fig. 3.11 Initially bent column

Let the initial deformation of the member be given by y0 and the additional
deformation due to bending by y.

x
y 0  a sin (3.61)
L

From the free body diagram of the column

M x  P( y 0  y )  0

 EIy" P( y0  y)  0 (3.62)

P
Substituting the value y0 and setting K 2  , one obtains
EI
 y "  K 2 y   K 2 y0

x
y "  K 2 y   K 2 a sin (3.63)
L

The solution of Eq . (3.63) consists of a complementary and particular solutions. Thus

y  yc  y p

y c  A sin Kx  B cos Kx (3.64)

The particular function is any solution of the entire nonhomogeneous equation. When the
right-hand side of the equation consists of a sine or cosine term, the particular function is
of the form

x x
y p  C sin  D cos (3.65)
L L

Substituting Eq. (3.65) into (3.63) and combining terms gives

  2 2   x   2  2  x
C K  2   K 2 a  sin  D K  2  cos 0
  L   L   L   L

This equation can be satisfied for all values of x only if both the coefficientS of the sine
and cosine terms are made to vanish. Hence

a
C
 2 
 2 2 
 K L 1

2
and either D = 0 or K 2 
L2

2  2 EI
If one lets K 2  , the solution for y becomes limited to P  . This is not
L2 L2
the case to be investigated, and D must therefore vanish. Hence, D = 0

Introducing the notation

P

PE
  2 EI
where PE  , the expression for C can be written in the form
L2

a a
C 
1  1   
  1
 

Consequently,

a x
yp  . sin (3.66)
1    L

 x
and y  A sin Kx  B cos Kx  .a sin (3.67)
1 L

The arbitrary constants in (3.67) are evaluated from the boundary conditions

y = 0 at x = 0

This leads to B = 0

and from the condition y = 0 at x = L one obtains

AsinKL = 0

Thus, either A or sinKL must vanish. Letting sinKL = 0 again limits the solution for y to
P = PE. As before, this is undesirable, and consequently

A=0

Substitution of A = 0 and B = 0 into Eq. (3.67) leads to the bending deflection

 x
y a sin (3.68)
1    L

The total deflection from the vertical is obtained by adding this expression to the initial
deflection. Thus

   x
yT  y 0  y  1   a sin
 1    L

a x
yT  sin (3.69)
1    L
The total deflection at midheight is
 a a
  (3.70)
1     P 
1  
 PE 

Behavior of Imperfect Columns (Another Approach)

From Fig. 3.11 considering the free-body diagram

M x  P( y 0  y )  0

 EIy" P( y0  y)  0

P
y "  K 2 y   K 2 y0 where K 2 
EI
x
Let y 0  a sin
L
x
y "  K 2 y   K 2 a sin
L

A series solution may be assumed, each term of which satisfies the boundary conditions

x
y  A sin
L
 2
x
y"   A 2
sin
L L

2 x x x
A 2
sin  K 2 A sin   K 2 a sin
L L L L

 2 
 A 2  K 2    K 2 a
L 

K 2a
A
 2 
 2  K 2 
L 

K 2a x
Therefore y . sin
 2
 L
 2  K 2 
L 
 P
a
EI x
y . sin
 2
P L
 2  
L EI 

 2 EI
Setting PE 
L2

P x
y a sin
  2 EI  L
 2  P 
 L 

P x
y a sin
PE  P  L

P
If 
PE

a x
y sin
1    L
   x
yT  y  y0    1a sin
 1     L

1 x
yT  a sin
1    L

L
at midheight, x 
2

a 
  yt   . sin
L
 x 
 2
1    2

a a
 
1     P 
1  
 PE 
3.7 BEHAVIOR OF ECCENTRICALLY LOADED COLUMN

e+y
P
e y

L Mx

y
e

P
P

Fig. 3.12 Eccentrically loaded column

From the free-body diagram

M x  P (e  y )  0

 EIy"P(e  y)  0

EIy" P(e  y)  0 (3.71)

P
Setting K2 
EI

one obtains y "  K 2 y  K 2 e (3.72)

The general solution is

y  A sin Kx  B cos Kx  e (3.73)

at x = 0 ; y = 0 therefore B = 0

at x = L ; y = 0
  1  cos KL 
A  e 
 sin KL 

Substituting A and B in Eq (3.73) gives

  1  cos KL  
y  e cos Kx    sin Kx  1 (3.74)
  sin KL  

L
at midheight; x  ;y=
2

  KL   1  cos KL   KL  
  e cos   sin   1 (3.75)
  2   sin KL   2  

 KL 
But cos KL  1  2 sin 2  
 2 

 KL   KL 
sin KL  2 sin   cos 
 2   2 

Eq. (3.75) can be reduced to

  KL  
  e sec   1
  2  

  P 
  e sec  1 (3.76)
 2 PE 

 2 EI
Where PE 
L2