NATIONAL INSTITUTE OF TECHNOLOGY AGARTALA

Theory of Elastic Stability

(PCE12B01)

RICHI PRASAD SHARMA

Professor in Civil Engineering
NATIONAL INSTITUTE OF TECHNOLOGY AGARTALA
Email: richisharma.sharma@gmail.com
rps.civil@nita.ac.in
Mobile:+919436463474

2/22/2024 1
Energy Method:
The basic criteria for determining the stability of equilibrium are
derived for conservative linearly elastic systems using energy
method.

Let the total function = U+ (6.1)

Where u is internal strain energy
 is potential energy of the external forces that act on a system.

Disregarding a possible additive constant, =-We, i.e. the loss of

potential energy during the application of the forces is equal to the
work done on the system by external forces.
So equation (6.1) ca be rewritten as
= U-We (6.2)

2/22/2024 2
 As per classical mechanics, for equilibrium, total potential 
must be stationary; therefore, its variation  must equal zero
i.e.

=U- We=0
Using Taylor’s expansion
1 1
∆𝜋 = 𝛿𝜋 + 𝛿 2 𝜋 + 𝛿 3 𝜋+---- (6.3)
2! 3!
For any type of equilibrium =0
So 𝛿 2 𝜋 determines the type of equilibrium
Thus from equation (6.3), the stability criteria are
𝛿 2 𝜋 > 0 for stable equilibrium
𝛿 2 𝜋 < 0 for unstable equilibrium
𝛿 2 𝜋 = 0 for neutral equilibrium associated with critical load.

2/22/2024 3
( Taylor’s expansion
Let a function f(x,y)
𝜕𝑓 𝜕𝑓 1 2 𝜕2 𝑓 2
2𝜕 𝑓 2𝜕 𝑓
Then f(x+h,y+k)=f(x,y)+h𝜕𝑥 + 𝑘 𝜕𝑦 + ℎ 𝜕2 𝑥 + 2ℎ𝑘 𝜕𝑥𝜕𝑦 + 𝑘 𝜕2 𝑦 +⋯
2!
𝜕𝑓 = f x + h, y + k − f x, y

𝜕𝑓 𝜕𝑓 1 2 𝜕2 𝑓 2
2𝜕 𝑓 2𝜕 𝑓
= h 𝜕𝑥 + 𝑘 𝜕𝑦 + ℎ 𝜕2 𝑥 + 2ℎ𝑘 𝜕𝑥𝜕𝑦 + 𝑘 𝜕2 𝑦 +⋯
2!

2/22/2024 4
 Y,𝑣

𝑣 P
P X,𝑢
A B
L, EI
Beam-Column 𝐹𝑖𝑔. 6.1
𝑥
The general differential equation of equilibrium of slightly buckled column with general
boundary condition
𝑑2 𝑑2 𝑣 𝑑2 𝑣
𝑝 𝑥 𝑑𝑥 2 + 𝜆2 𝑑𝑥 2 = 0 (6.4)
𝑑𝑥 2
Where function 𝑣 𝑖𝑠 𝑜𝑟𝑡ℎ𝑜𝑔𝑜𝑛𝑎𝑙
𝐿 𝑑𝑣 𝑑𝑣
i.e. 0 𝑑𝑥𝑖 𝑑𝑥𝑘 𝑑𝑥 = 0 (6.5)
𝐿 𝑑2 𝑣𝑖 𝑑2 𝑣𝑘
𝑝 𝑥 𝑑𝑥
=0 (6.6)
0 𝑑𝑥 2 𝑑𝑥 2
If 𝑖 = 𝑘, the integral becomes
𝐿 𝑑𝑣𝑖 2
0 𝑑𝑥
𝑑𝑥 ≠ 0 Always positive quantity (6.7)
2
𝐿 𝑑2 𝑣𝑖
0
𝑝 𝑥 𝑑𝑥 2 𝑑𝑥 ≠ 0 (6.8)

2/22/2024 5
Theorem of Stationary Potential Energy
When external forces are applied gradually in an elastic body and no dissipation of energy
takes place due to friction etc., work done by external forces should be equal to the
internal elastic energy 𝑈
i.e. 𝑊 = 𝑈
For small virtual Displacement
𝛿𝑊 = −𝛿Ω 𝑑𝑒𝑐𝑟𝑒𝑠𝑒 𝑜𝑓 𝑝𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙 𝑒𝑛𝑒𝑟𝑔𝑦
−𝛿𝑉 − δ𝑊 = 0
⇒ −𝛿Ω − δ𝑈 = 0
⇒𝛿 U+Ω =0 (6.9)
The expression 𝑈 + Ω 𝑖𝑛 equation (6.9) is known as the total potential energy of the
system.
The first order change in the total potential energy must vanish for every virtual
displacement when an elastic body is equilibrium.
𝑈 + Ω = 𝑠𝑡𝑎𝑡𝑖𝑜𝑛𝑒𝑟𝑦 (6.9a)
For stable equilibrium, the total potential energy is minimum

2/22/2024 6
Application to Buckling Problem ∆𝑆

Y,𝑣 ∆𝑣

∆𝑥
∆𝑆
𝑣 ∆𝐿 P
P X,𝑢
A ∆𝑥 B
L, EI
Beam-Column 𝐹𝑖𝑔. 6.2
𝑥

Let the buckled form be expressed by 𝑣 = 𝑓 𝑥

The elastic strain energy is
2
1 𝐿 𝑑2 𝑣
𝑈= 𝐸 𝐼
0 𝑋 𝑑𝑥 2
𝑑𝑥 (6.10)
2
Taking the un-deflected position as datum, the potential energy in the buckled
form is
𝐿
Ω = −𝑃Δ𝐿 = −𝑃 0 (∆𝑠 − ∆𝑥) (6.11)
Where ∆𝐿 axial contraction of column length

2/22/2024 7
 ∆𝑆
∆𝑣

∆𝑥
1
2 2
2
1
2 2
∆𝑣
∆𝑆 = ∆𝑥 + ∆𝑣 = ∆𝑥 1 +
∆𝑥
2
1 ∆𝑣
= ∆𝑥 1 + + ⋯….
2 ∆𝑥
2
1 ∆𝑣
= ∆𝑥 + ∆𝑥
2 ∆𝑥
2
1 ∆𝑣
∆𝑠 − ∆𝑥 = ∆𝑥
2 ∆𝑥
2
1 𝑑𝑣
∆𝑥 → 0 ∆𝑠 − Δ𝑥 = 𝑑𝑥
2 𝑑𝑥
𝐿 1 𝐿 𝑑𝑣 2
Hence potential energy Ω = 𝑃 0
∆𝑠 − Δ𝑥 = 𝑃 0 𝑑𝑥 𝑑𝑥 (6.12)
2

2/22/2024 8
 The total potential energy is therefore , given by
2
1 𝐿 𝑑2 𝑣 1 𝐿 𝑑𝑣 2
Π=𝑈+Ω= 𝐸 0 𝐼𝑋 𝑑𝑥 2 𝑑𝑥 − 𝑃 0 𝑑𝑥 𝑑𝑥 (6.13)
2 2
For equilibrium, the variation of equation (6.10) should be vanish i.e.
2
1 𝐿 𝑑2 𝑣 1 𝐿 𝑑𝑣 2
𝛿Π = 𝛿 𝑈 + Ω = 𝛿 𝐸 0 𝐼𝑋 𝑑𝑥 2 𝑑𝑥 − 𝑃 0 𝑑𝑥 𝑑𝑥 =0 (6.14)
2 2
The equation (6.14) permits the determination of the function 𝑣 = 𝑓 𝑥

The merit of the energy criterion for solution of stability problems becomes in the
Rayleigh-Ritz method

2/22/2024 9
The Rayleigh-Ritz Method:

A direct solution to the extremum problem stated by equation (6.9a) is obtained by the
Rayleigh-Ritz method, revealing the importance of the energy criterion. Same method
can be used to analysis of the buckling problem.

The deflection of the buckled column is expressed in the form of a finite series:
Let 𝑣 = 𝑐1 𝜙1 +𝑐2 𝜙2 + 𝑐3 𝜙3 + ⋯ … … + 𝑐𝑛 𝜙𝑛 (6.15)

The 𝜙 terms are a set of arbitrarily chosen function of 𝑥 , such that each term satisfies
the prescribed boundary conditions of the column. These are call coordinate functions.
The coefficients 𝑐 correspond to a set of parameters, as yet undetermined.
Putting equation (6.15) in to equation (6.9a), we get
𝑈 + Ω = 𝐹1 (𝑐1 , 𝑐2 , 𝑐3 ,………𝑐𝑛 ) − P𝐹2 𝑐1 , 𝑐2 , 𝑐3 , ………+𝑐𝑛 (6.16)

For determine the parameters 𝑐1 , 𝑐2 , 𝑐3 ,………𝑐𝑛

𝜕 𝑈+Ω
=0 𝑖 = 1,2,3 … … … 𝑛 (6.17)
𝜕𝑐𝑖
i.e.

2/22/2024 10
 𝜕𝐹1 𝜕𝐹
− 𝑃 𝜕𝑐2 = 0
𝜕𝑐1 1
𝜕𝐹1 𝜕𝐹
− 𝑃 𝜕𝑐2 =0
𝜕𝑐2 2 𝑐𝑖 =0
𝜕𝐹1 𝜕𝐹2
− 𝑃 𝜕𝑐 =0 (6.18)
𝜕𝑐3 3
⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯
𝜕𝐹1 𝜕𝐹2
−𝑃 =0
𝜕𝑐𝑛 𝜕𝑐𝑛
The above set of equations (6.18) involve only linear functions since these are
derivatives of the quadratic expressions in involved in 𝑈 + Ω. Since equation
(6.18) is a set of homogenous equations, for existence of a non-trivial solution,
the determinant of coefficients should be equal to zero, i.e.
Δ=0 (6.19)

This is an equation of degree n in the unknown P and is the stability condition

from which P can be determined. The smallest of the roots gives the critical load
𝑃𝑐𝑟
Introducing 𝑃 = 𝑃𝑐𝑟 in equation (6.18), a set of n linear homogenous equations
is obtained from which the ratios of the parameter 𝑐 can be determined, calling
𝑐2 𝑐3 𝑐4 𝑐𝑛
= 𝛼 2 , = 𝛼 3 , = 𝛼 4 ,⋯ = 𝛼𝑛 ,
𝑐1 𝑐 1 𝑐1 1 𝑐
Then buckling mode 𝑣 = 𝑐1 ∅1 𝛼1 + ∅2 𝛼2 + ∅3 𝛼3 + ⋯ ⋯ + ∅𝑛 𝛼𝑛 6.20)

2/22/2024 11
The importance of the Rayleigh-Ritz method lies in the fact that it
offers a method of obtaining an approximate solution to the
buckling problem. The method, in many cases, involves less labour
than is involved in solving differential equation and the associate
eigenvalue (i.e. characteristic value problem). In the majority of
cases, a few of the series in equation (6.15)give a sufficiently
accurate result. Success or failure in applying the Rayleigh-Ritz
method to any problem depends largely on the proper choice of
the coordinate functions.
In Majority cases , satisfactory results can be obtained only when
the coordinate function chosen fro a system of orthogonal
functions .
This is the reason Fourier Series play such an important role in the application of
the Rayleigh-Ritz method

2/22/2024 12
Example 6.1: Consider a pin-ended column subjected to an axial compressive load P,
as shown in fig. 6.3
Y,𝑣

𝑣 P
P X,𝑢
A B
L, EI
Beam-Column 𝐹𝑖𝑔. 6.3
𝑥

Solution:
Assume that the buckled shape is given by
𝜋𝑥
𝑣 = 𝑐𝑠𝑖𝑛
𝐿
Where c is an unknown parameter. The coordinate function satisfies the boundary
conditions which are
𝑣 0 = 0, 𝑣 𝐿 = 0,
𝑑2𝑣 𝑑2𝑣
0 = 0, 2 𝐿 = 0
𝑑𝑥 2 𝑑𝑥
2/22/2024 13
From equation (6.10), the strain energy is obtained as
2
1 𝐿 𝑑2 𝑣
𝑈 = 2 𝐸 0 𝐼𝑋 𝑑𝑥 2 𝑑𝑥
1 𝐿 2 𝜋 4 2 𝜋𝑥 1 𝐸𝐼
= 𝐸𝐼 0
𝑐 𝑠𝑖𝑛 = 4 𝜋 4 𝑐 2 𝐿3
𝑑𝑥
2 𝐿 𝐿
From equation (6.12), the potential energy is obtained as
1 𝐿 𝑑𝑣 2
Ω = − 2 𝑃 0 𝑑𝑥 𝑑𝑥
1 𝐿 2 𝜋 2 𝜋𝑥 1 𝑐2
= − 2 𝑃 0 𝑐 𝐿 𝑐𝑜𝑠 2 𝐿 𝑑𝑥 = 2
− 4 𝑃𝜋 𝐿

Thus, the total potential energy is

1 4 2 𝐸𝐼 1 2 𝑐2
Π=𝑈+Ω= 𝜋 𝑐 𝐿3 − 𝑃𝜋 𝐿
4 4
For extremum
𝜕 𝑈+Ω
=0
𝜕𝑐
1 4 𝐸𝐼 1 2 𝑐 =0
𝜋 𝑐 3 − 𝑃𝜋
2 𝐿 2 𝐿
1 2
𝑐 𝜋 𝐸𝐼
⇒ 2 𝜋 2 𝐿 𝐿2 − 𝑃 = 0
𝜋2 𝐸𝐼
For non-trivial solution 2 −𝑃 =0
𝐿
𝜋2 𝐸𝐼
𝑃 = 𝑃𝑐𝑟 = 𝐿2

2/22/2024 14
We have been able to obtain an exact solution since the coordinate function we used
happens to give the exact deflected shape for the column.

2/22/2024 15
Example 6.2: Consider a column fixed at one end and free at the other end(Fig. 6.4). It
is subjected to a compressive load P at the free end. Determine the approximate
𝑥 2 𝑥 3
critical load assuming the deflection curve as 𝑣 = 𝑐1 𝐿 + 𝑐2 𝐿
𝑋, 𝑢 𝑋, 𝑢
The boundary conditions are
P 𝑃𝐶𝑟
𝛿 𝑑𝑣
B 𝑣 0 = 0, 0 = 0,
B 𝑑𝑥
𝑑2𝑣
𝐿 =0
𝑑𝑥 2
(i) Let us ignore the last condition
L,EI 𝑣
for the time being. The first two
conditions are satisfied by the
𝑥 coordinate functions.
The strain energy is equal to
2
1 𝐿 𝑑2 𝑣
𝑈 = 2 𝐸 0 𝐼𝑋 𝑑𝑥 2 𝑑𝑥
Y, 𝑣 A Y, 𝑣 𝐿 2
A 1 2𝑐1 6𝑐2 𝑥
= 𝐸𝐼 + 3 𝑑𝑥
Figure 6.4 2 0 𝐿2 𝐿
2𝐸𝐼 2
= 3
𝑐1 + 3𝑐1 𝑐2 + 3𝑐2 2
𝐿

2/22/2024
16
The potential energy is equal to
1 𝐿 𝑑𝑣 2
Ω= − 2 𝑃 0 𝑑𝑥 𝑑𝑥
𝐿 2
1 2𝑐1 𝑥 3𝑐2 𝑥 2 𝑃
=− 𝑃 2 + 3 𝑑𝑥 = − 20𝑐1 2 + 45𝑐1 𝑐2 + 27𝑐2 2
2 0 𝐿 𝐿 30𝐿
Thus, the total potential energy is
2𝐸𝐼 𝑃
Π = 𝑈 + Ω = 𝐿3 𝑐1 2 + 3𝑐1 𝑐2 + 3𝑐2 2 − 30𝐿 20𝑐1 2 + 45𝑐1 𝑐2 + 27𝑐2 2
2
2𝐸𝐼 2𝑃 6𝐸𝐼 3𝑃 2
6𝐸𝐼 9𝑃
= 𝑐1 − + 𝑐1 𝑐2 − + 𝑐2 −
𝐿3 3𝐿 𝐿3 2𝐿 𝐿3 10𝐿

For an extremum we should have

𝜕 𝑈+Ω 4𝐸𝐼 4𝑃 6𝐸𝐼 3𝑃
=0⇒ − 𝑐 + − 𝑐 =0
𝜕𝑐1 𝐿3 3𝐿 1 𝐿3 2𝐿 2
𝜕 𝑈+Ω 6𝐸𝐼 3𝑃 12𝐸𝐼 9𝑃
=0⇒ − 𝑐 + − 𝑐 =0
𝜕𝑐2 𝐿3 2𝐿 1 𝐿3 5𝐿 2
For the existence of a non-trivial solution, the determinant ∆ 𝑜𝑓 the coefficients should
be equal to zero, Hence
2
4𝐸𝐼 4𝑃 12𝐸𝐼 9𝑃 6𝐸𝐼 3𝑃
∆= − − − − =0
𝐿3 3𝐿 𝐿3 5𝐿 𝐿3 2𝐿

2/22/2024 17
 3𝑃2 𝐿4 − 104𝑃𝐿2 𝐸𝐼 + 240𝐸 2 𝐼 2 = 0
Therefore,
𝐸𝐼 𝐸𝐼
𝑃 = 2.49 𝐿2 𝑜𝑟 32.18 𝐿2

𝐸𝐼
The smaller value is 𝑃𝑐𝑟 = 2.49
𝐿2
𝜋2 𝐸𝐼 𝐸𝐼
𝜋2 𝐸𝐼 2 −2.49 2
4𝐿 𝐿
Compared to the exact value , the error is 𝜋2 𝐸𝐼
= +0.92%
4𝐿2
4𝐿2
(ii) In the above analysis, we have ignored the third boundary condition,
𝑑2 𝑣
𝑑𝑥 2
𝐿 =0
If we use this condition
𝑑2𝑣 2𝑐1 6𝑐2 𝐿
𝐿 = 0 ⇒ + 3 =0
𝑑𝑥 2 𝐿2 𝐿
𝑜𝑟 𝑐1 = −3𝑐2
Using this
𝑥 2 𝑥 3
𝑣 = 𝑐1 + 𝑐2
𝐿 𝐿
𝑥 2 1 𝑥 3
⇒ 𝑣 = 𝑐1 − 𝑐1
𝐿 3 𝐿

2/22/2024 18
Substituting 𝑐1 = −3𝑐2 in the expression for 𝑈 and Ω
2𝐸𝐼
𝑈 = 3 𝑐1 2 + 3𝑐1 𝑐2 + 3𝑐2 2
𝐿

2𝐸𝐼 2 2
1 2 2𝐸𝐼 2
⇒ 𝑈 = 3 𝑐1 − 𝑐1 + 𝑐1 = 3 𝑐1
𝐿 3 3𝐿
𝑃
And Ω = − 30𝐿 20𝑐1 2 + 45𝑐1 𝑐2 + 27𝑐2 2
𝑃 2 2 1 2 1 2 4 𝑃 2
⇒Ω=− 𝑐1 − 𝑐1 + 𝑐1 = − 𝑐
𝐿 3 2 10 45 𝐿 1
Therefore
2𝐸𝐼 2 4 𝑃 2 2𝐸𝐼 4 𝑃 2
𝑈+Ω= 𝑐 1 − 𝑐1 = − 𝑐1
3𝐿3 45 𝐿 3𝐿3 45 𝐿
For extremum
2𝐸𝐼 4 𝑃
− =0
3𝐿3 45 𝐿
𝐸𝐼
Or 𝑃 = 𝑃𝑐𝑟 = 2.5 𝐿3
Which is almost identical with the previous solution but solution has been obtained
with comparative ease.

2/22/2024 19
Timoshenko’s Concept of Solving Buckling Problem:
For stable configuration
𝑃 < 𝑃𝑐𝑟
𝑈−𝑊 >0
For un stable configuration
𝑃 > 𝑃𝑐𝑟
𝑊>𝑈
For equilibrium unstable configuration
𝑈−𝑊 =0
By Rayleigh-Ritz method

2
1 𝐿 𝑑2 𝑣
𝑈 = 2 𝐸𝐼 0 𝑑𝑥 = 𝐹1 (𝑐1 , 𝑐2 , 𝑐3 ,………𝑐𝑛 )
𝑑𝑥 2
1 𝐿 𝑑𝑣 2
𝑊= 𝑃 0 𝑑𝑥 = P𝐹2 𝑐1 , 𝑐2 , 𝑐3 , ………𝑐𝑛
2 𝑑𝑥

When, 𝑈 − 𝑊 = 0
2
𝐿 𝑑2 𝑣
𝐸𝐼 0 𝑑𝑥2 𝑑𝑥 𝐹1 (𝑐1 , 𝑐2 ,𝑐3 ,………𝑐𝑛 )
𝑃= 𝐿 𝑑𝑣 2
= (6.21)
𝑑𝑥 𝐹2 𝑐1 , 𝑐2 ,𝑐3 ,………𝑐𝑛
0 𝑑𝑥
This is Rayleigh-Ritz formula

2/22/2024 20
For a pin ended column or a column with free end
𝑑2𝑣
𝑀 = 𝐸𝐼 2
𝑑𝑥
𝑀 = 𝑃𝑣
𝑑 2 𝑣 𝑀 𝑃𝑣
∴ 2= =
𝑑𝑥 𝐸𝐼 𝐸𝐼
Then, from equation (6.21), we get
2
𝐿 𝑑2𝑣 𝑃 2 𝐿
𝐸𝐼 0 𝑑𝑥 𝑣 2
𝑑𝑥
𝑑𝑥 2 𝐸𝐼 0
𝑃= 2 = 2
𝐿 𝑑𝑣 𝐿 𝑑𝑣
0 𝑑𝑥
𝑑𝑥 0 𝑑𝑥
𝑑𝑥
𝑃 𝐿 2
𝑣 𝑑𝑥
1= 𝐸𝐼 0
2
𝐿 𝑑𝑣
0 𝑑𝑥
𝑑𝑥
𝐿 𝑑𝑣 2
0 𝑑𝑥 𝑑𝑥 𝐹 (𝑐1 , 𝑐2 ,𝑐3 ,………𝑐𝑛 )
𝑃= 𝐿 = 𝐹1 Timoshenko Formula (6.22)
𝐸𝐼 0 𝑣 2 𝑑𝑥 2 𝑐1 , 𝑐2 ,𝑐3 ,………𝑐𝑛

For extremum value

𝜕𝐹1 𝜕𝐹2
−𝑃 =0
𝜕𝑐𝑖 𝜕𝑐𝑖

2/22/2024 21
Columns with variable Cross-Section:
Example 6.3 Consider a column with the moment of inertia of the cross-section area
varying according to the equation. The column is hinged at both ends. Determine the
𝑃𝐶𝑟 using Rayleigh-Ritz method and Tomoshenko’s method .
𝜋𝑥
𝐼 = 𝐼0 1 + 𝑠𝑖𝑛
Y, 𝑣 𝐿

P
P X, 𝑢
A B

Solution: Assume that the deflection curve can be represented by the series
𝑛𝜋𝑥
𝑣= 𝑐𝑛 𝑠𝑖𝑛
𝐿

2/22/2024 22
Since the deflection curve must be symmetrical with respect to the middle point of the
column because the moment of inertia is symmetrical about the middle point.

The even parameter in the above series vanish.

The deflection equation the becomes
𝜋𝑥 3𝜋𝑥 5𝜋𝑥
𝑣 = 𝑐1 𝑠𝑖𝑛 + 𝑐3 𝑠𝑖𝑛 + 𝑐5 𝑠𝑖𝑛
𝐿 𝐿 𝐿
Considering only two term of the series
𝜋𝑥 3𝜋𝑥
𝑣 = 𝑐1 𝑠𝑖𝑛 + 𝑐3 𝑠𝑖𝑛
𝐿 𝐿

2/22/2024 23