CHAPTER 12 TWO-DEGREE- OF-FREEDOM-SYSTEMS

Introduction to two degree of freedom systems:

• The vibrating systems, which require two coordinates to describe its motion, are
called two-degrees-of –freedom systems.
• These coordinates are called generalized coordinates when they are independent
of each other and equal in number to the degrees of freedom of the system.
• Unlike single degree of freedom system, where only one co-ordinate and hence
one equation of motion is required to express the vibration of the system, in two-
dof systems minimum two co-ordinates and hence two equations of motion are
required to represent the motion of the system. For a conservative natural system,
these equations can be written by using mass and stiffness matrices.
• One may find a number of generalized co-ordinate systems to represent the
motion of the same system. While using these co-ordinates the mass and stiffness
matrices may be coupled or uncoupled. When the mass matrix is coupled, the
system is said to be dynamically coupled and when the stiffness matrix is coupled,
the system is known to be statically coupled.
• The set of co-ordinates for which both the mass and stiffness matrix are
uncoupled, are known as principal co-ordinates. In this case both the system
equations are independent and individually they can be solved as that of a single-
dof system.
• A two-dof system differs from the single dof system in that it has two natural
frequencies, and for each of the natural frequencies there corresponds a natural
state of vibration with a displacement configuration known as the normal mode.
Mathematical terms associated with these quantities are eigenvalues and
eigenvectors.
• Normal mode vibrations are free vibrations that depend only on the mass and
stiffness of the system and how they are distributed. A normal mode oscillation is
defined as one in which each mass of the system undergoes harmonic motion of
same frequency and passes the equilibrium position simultaneously.
• The study of two-dof- systems is important because one may extend the same
concepts used in these cases to more than 2-dof- systems. Also in these cases one
can easily obtain an analytical or closed-form solutions. But for more degrees of

209
 freedom systems numerical analysis using computer is required to find natural
frequencies (eigenvalues) and mode shapes (eigenvectors).

The above points will be elaborated with the help of examples in this lecture.
Few examples of two-degree-of-freedom systems

Figure 1 shows two masses m1 and m2 with three springs having spring stiffness
k1 , k2 and k3 free to move on the horizontal surface. Let x1 and x2 be the displacement of
mass m1 and m2 respectively.
x1 x2
k1 k2 k3
m1 m2

Figure 1

As described in the previous lectures one may easily derive the equation of motion by
using d’Alembert principle or the energy principle (Lagrange principle or Hamilton’s
principle)
m1 x1

Using d’Alembert principle for mass m1 , from the k2 ( x1 − x2 )

k1 x1
free body diagram shown in figure 1(b) m1
x1 + (k1 + k2 ) x1 − k2 x2 = 0
m1  (1)
and similarly for mass m2
x2 − k1 x1 + (k2 + k3 ) x2 = 0
m2  (2)
m2 x2

Important points to remember k2 ( x2 − x1 ) k3 x2

m2

• Inertia force acts opposite to the

direction of acceleration, so in both the
free body diagrams inertia forces are shown
Figure 1 (b), Free body diagram
towards left.
• For spring k2 , assuming x2 > x1 ,

210
The spring will pull mass m1 towards right by k2 ( x2 − x1 ) and it is stretched by x2 − x1
(towards right) it will exert a force of k2 ( x2 − x1 ) towards left on mass m2 . Similarly
assuming x1 > x2 , the spring get compressed by an amount x2 − x1 and exert tensile force
of k2 ( x1 − x2 ) . One may note that in both cases, free body diagram remain unchanged.

Now if one uses Lagrange principle,

1 1
The Kinetic energy = T = m1 x12 + m2 x22 and (3)
2 2
1 1 1
Potential energy = U = k1 x12 + k2 ( x1 − x2 ) 2 + k3 x22 (4)
2 2 2
So, the Lagrangian

⎛1 1 ⎞ ⎛1 1 1 ⎞
L = T − U = ⎜ m1 x12 + m2 x22 ⎟ − ⎜ k1 x12 + k2 ( x1 − x2 ) 2 + k3 x22 ⎟ (5)
⎝2 2 ⎠ ⎝2 2 2 ⎠
The equation of motion for this free vibration case can be found from the Lagrange
principle

d ⎛ ∂L ⎞ ∂L
⎜ ⎟− =0, (6)
dt ⎝ ∂qk ⎠ ∂qk

and noting that the generalized co-ordinate q1 = x1 and q2 = x2

which yields
x1 + (k1 + k2 ) x1 − k2 x2 = 0
m1  (7)
x2 − k1 x1 + (k2 + k3 ) x2 = 0
m2  (8)
Same as obtained before using d’Alembert principle.

Now writing the equation of motion in matrix form

x1 ⎞ ⎡ k1 + k2
⎡ m1 0 ⎤ ⎛  −k2 ⎤ ⎛ x1 ⎞ ⎛ 0 ⎞
⎢ 0 m ⎥ ⎜  ⎟+⎢ ⎜ ⎟=
k2 + k3 ⎦⎥ ⎝ x2 ⎠ ⎜⎝ 0 ⎟⎠
. (9)
⎣ 2 ⎦ ⎝ x2 ⎠ ⎣ −k2

Here it may be noted that for the present two degree-of-freedom system, the system is
dynamically uncoupled but statically coupled.

211
Example 2.

Consider a lathe machine, which can be modeled as a rigid bar with its center of mass not
coinciding with its geometric center and supported by two springs, k1 , k2 .

xc
J cθc xc + eθc )
m ( 

C θc
k1 ( xc − l1θ c ) e G
k1 k2 k2 ( xc + l2θ c )
l1 l2
k

Figure 3: Free body diagram of the system

Figure 2

In this example, it will be shown, how the use of different coordinate systems lead to
static and or dynamic coupled or uncoupled equations of motion. Clearly this is a two-
degree-of freedom system and one may express the co-ordinate system in many different
ways. Figure 3 shows the free body diagram of the system where point G is the center of
mass. Point C represents a point on the bar at which we want to define the co-ordinates of
this system. This point is at a distance l1 from the left end and l2 from right end. Distance
between points C and G is e . Assuming xc is the linear displacement of point C and θc
the rotation about point C, the equation of motion of this system can be obtained by using
d’Alember’s principle. Now summation of all the forces, viz. the spring forces and the
inertia forces must be equal to zero leads to the following equation.
mx + meθ + k ( x − l θ ) + k ( x + l θ ) = 0
c c 1 c 1 c 2 c 2 c (10)

Again taking moment of all the forces about point C

J θ + ( mx + meθ )e − k ( x − l θ )l + k ( x + l θ )l = 0
G c c c 1 c 1 c 1 2 c 2 c 2 (11)
Noting J c = J G + me 2 , the above two equations in matrix form can be written as
xc ⎞ ⎡ k1 + k2
⎡ m me ⎤ ⎛  k2l2 − k1l1 ⎤ ⎛ xc ⎞ ⎛ 0 ⎞
⎢ me J ⎥ ⎜ θ ⎟ + ⎢ k l − k l k l 2 + k l 2 ⎥ ⎜ θ ⎟ = ⎜ 0 ⎟ (12)
⎣ c ⎦⎝ c ⎠ ⎣ 2 2 11 11 2 2 ⎦⎝ c ⎠ ⎝ ⎠

212
Now depending on the position of point C, few cases can are studied below.

Case 1 : Considering e = 0 , i.e., point C and G coincides, the equation of motion can be
written as

x
θ
k1 ( x − l1θ )
k2 ( x + l2θ )

x ⎞ ⎡ k1 + k2
⎡ m 0 ⎤ ⎛  k2l2 − k1l1 ⎤ ⎛ x ⎞ ⎛ 0 ⎞
⎢ 0 J ⎥ ⎜ θ⎟ + ⎢ k l − k l k l 2 + k l 2 ⎥ ⎜ θ ⎟ = ⎜ 0 ⎟ (13)
⎣ G ⎦⎝ ⎠ ⎣ 2 2 11 11 2 2 ⎦⎝ ⎠ ⎝ ⎠

So in this case the system is statically coupled and if k1l1 = k2l2 , this coupling disappears,
and we obtained uncoupled x and θ vibrations.

Case 2 : If, k2l2 = k1l1 , the equation of motion becomes

xc ⎞ ⎡ k1 + k2
⎡ m me ⎤ ⎛  0 ⎤ ⎛ xc ⎞ ⎛ 0 ⎞
⎢ me J ⎥ ⎜ θ ⎟ + ⎢ 0 2 ⎜ ⎟ =⎜ ⎟.
k l + k l ⎦⎥ ⎝ θ c ⎠ ⎝ 0 ⎠
2
(14)
⎣ c ⎦⎝ c ⎠ ⎣ 11 2 2

Hence in this case the system is dynamically coupled but statically uncoupled.

Case 3: If we choose l1 = 0 , i.e. point C coincide with the left end, the equation of
motion will become
xc ⎞ ⎡ k1 + k2
⎡ m me ⎤ ⎛  k2l2 ⎤ ⎛ xc ⎞ ⎛ 0 ⎞
⎢ me J ⎥ ⎜ θ ⎟ + ⎢ k l ⎜ ⎟ = ⎜ ⎟.
k2l22 ⎦⎥ ⎝ θ c ⎠ ⎝ 0 ⎠
(15)
⎣ c ⎦⎝ c ⎠ ⎣ 22
Here the system is both statically and dynamically coupled.

Normal Mode Vibration

Again considering the problem of the spring-mass system in figure 1 with m1 = m ,
m2 = 2m , k1 = k2 = k3 = k , the equation of motion (9) can be written as
mx1 + k ( x1 − x2 ) + kx1 = 0
(16)
2mx2 − k ( x1 − x2 ) + kx2 = 0

213
We define a normal mode oscillation as one in which each mass undergoes harmonic
motion of the same frequency, passing simultaneously through the equilibrium position.
For such motion, we let
x1 = A1eiωt , x2 = A2 eiωt (17)
Hence,
(2k − mω 2 ) A1 − kA2 = 0
(18)
−kA1 + (2k − 2mω 2 ) A2 = 0
or, in matrix form
⎡ 2k − mω 2 −k ⎤ ⎛ A1 ⎞ ⎛ 0 ⎞
⎢ 2⎥⎜ ⎟=⎜ ⎟ (19)
⎣ −k 2k − 2mω ⎦ ⎝ A2 ⎠ ⎝ 0 ⎠
Hence for nonzero values of A1 and A2 (i.e., for non-trivial response)
2k − mω 2 −k
= 0. (20)
−k 2k − 2mω 2
Now substituting ω 2 = λ , equation 6.1. yields
k 3 k
λ 2 − (3 )λ + ( ) 2 = 0 (21)
m 2 m
3 1 k k 3 1 k k
Hence, λ1 = ( − 3) = 0.634 and λ2 = ( + 3) = 2.366
2 2 m m 2 2 m m
k k
So, the natural frequencies of the system are ω1 = λ1 = 0.634 and ω2 = 2.366
m m
Now from equation (1)., it may be observed that for these frequencies, as both the
equations are not independent, one can not get unique value of A1 and A2 . So one should
find a normalized value. One may normalize the response by finding the ratio of A1 to A2 .
From the first equation (19) the normalized value can be given by
A1 k k
= = (22)
A2 2k − mω 2
2k − mλ
and from the second equation of (19), the normalized value can be given by
A1 2k − 2mω 2 2k − 2mλ
= = (23)
A2 k k
k
Now, substituting ω12 = λ1 = 0.634 in equation (22) and (23) yields the same values,
m
as both these equations are linearly dependent. Here,

214
⎛ A1 ⎞ 0.732
⎜ ⎟ = (24)
⎝ A2 ⎠ λ =λ1 1
k
and similarly for ω2 2 = λ2 = 2.366
m
⎛ A1 ⎞ −2.73
⎜ ⎟ = (25)
⎝ A2 ⎠λ =λ2 1

It may be noted

• Equation (19) gives only the ratio of the amplitudes and not their absolute values,
which are arbitrary.
• If one of the amplitudes is chosen to be 1 or any number, we say that amplitudes
ratio is normalized to that number.
• The normalized amplitude ratios are called the normal modes and designated
by φn ( x) .
From equation (24) and (25), the two normal modes of this problem are:

⎧0.731⎫ ⎧−2.73⎫
φ1 ( x) = ⎨ ⎬ φ2 ( x ) = ⎨ ⎬
⎩1.00 ⎭ ⎩1.00 ⎭

In the 1st normal mode, the two masses move in the same direction and are said to be in
phase and in the 2nd mode the two masses move in the opposite direction and are said to
be out of phase. Also in the first mode when the second mass moves unit distance, the
first mass moves 0.731 units in the same direction and in the second mode, when the
second mass moves unit distance; the first mass moves 2.73 units in opposite direction.

Free vibration using normal modes

When the system is disturbed from its initial position, the resulting free-vibration of the
system will be a combination of the different normal modes. The participation of
different modes will depend on the initial conditions of displacements and velocities. So
for a system the free vibration can be given by

x = φ1 A sin(ω1t +ψ 1 ) + φ2 B sin(ω2t + ψ 2 ) (27)

215
Here A and B are part of participation of first and second modes respectively in the
resulting free vibration and ψ 1 and ψ 2 are the phase difference. They depend on the
initial conditions. This is explained with the help of the following example.

Example: Let us consider the same spring-mass problem (figure 4) for which the natural
frequencies and normal modes are determined. We have to determine the resulting free
vibration when the system is given an initial displacement x1 (0) = 5, x2 (0) = 1 and initial
velocity x1 (0) = x2 (0) = 0 .
x1 x2
k k k
m 2m

Figure 4
Solution:
Any free vibration can be considered to be the superposition of its normal modes. For
each of these modes the time solution can be expressed as:

⎧ x1 ⎫ ⎧0.731⎫
⎨ ⎬ =⎨ ⎬ sin ω1t
⎩ x2 ⎭1 ⎩1 ⎭
⎧ x1 ⎫ ⎧ −2.731⎫
⎨ ⎬ =⎨ ⎬ sin ω2t
⎩ x2 ⎭2 ⎩1.00 ⎭

The general solution for the free vibration can then be written as:

⎧ x1 ⎫ ⎧0.731⎫ ⎧−2.73⎫
⎨ ⎬ = A⎨ ⎬ sin(ω1t +ψ 1 ) + B ⎨ ⎬ sin(ω2t + ψ 2 )
⎩ x2 ⎭ ⎩1.00 ⎭ ⎩1 ⎭

where A and B allow different amounts of each mode and ψ 1 and ψ 2 allows the two
modes’ different phases or starting values.

Substituting:

216
⎧ x1 (0) ⎫ ⎧5⎫ ⎧0.731⎫ ⎧−2.731⎫
⎨ ⎬ = ⎨ ⎬ = A⎨ ⎬ sinψ 1 + B ⎨ ⎬ sinψ 2
⎩ x2 (0) ⎭ ⎩1 ⎭ ⎩1 ⎭ ⎩1 ⎭
⎧ x1 (0) ⎫ ⎧0 ⎫ ⎧0.731⎫ ⎧−2.731⎫
⎨ ⎬ = ⎨ ⎬ = ω1 A ⎨ ⎬ cosψ 1 + ω2 B ⎨ ⎬ cosψ 2
⎩ x2 (0) ⎭ ⎩0 ⎭ ⎩1 ⎭ ⎩1 ⎭

cosψ 1 = cosψ 2 = 0 =>ψ 1 = ψ 2 = 900

Substituting in 1st set:

⎧5⎫ ⎧0.731⎫ ⎧−2.731⎫
⎨ ⎬ = A⎨ ⎬+ B⎨ ⎬
⎩1 ⎭ ⎩1 ⎭ ⎩1 ⎭

0.731A-2.731B= 5 A= 2.233
A+B =1 B=-1.233

Hence the resulting free vibration is

⎧ x1 ⎫ ⎧0.731⎫ ⎧ −2.731⎫
⎨ ⎬ = 2.233 ⎨ ⎬ cos ω1t − 1.233 ⎨ ⎬ cos ω2t
⎩ x2 ⎭ ⎩1.00 ⎭ ⎩1.000 ⎭

Normal modes from eigenvalues

The equation of motion for a two-degree-of freedom system can be written in matrix
form as
x + K x=0
M  (28)
where M and K are the mass and stiffness matrix respectively; x is the vector of
generalized co-ordinates. Now pre-multiplying M −1 in both side of equation 6.2. one
may get
x + M −1 K x = 0
I  (29)
x + Ax=0
or, I  (30)
Here A = M −1 K is known as the dynamic matrix. Now to find the normal modes,

x1 = X 1eiωt , x2 = X 2 eiωt , the above equation will reduce to

[ A − λI ] X = 0 (31)

217
where X = { x1 x2 } and λ =ω 2 .
T

From equation (31) it is apparent that the free vibration problem in this case is reduced to
that of finding the eigenvalues and eigenvectors of the matrix A.

Example: Determine the normal modes of a double pendulum.

Solution

Kinetic energy of the system =

1 1
T = m1l12θ12 + m2 (l12θ12 + l22θ22 + 2l1l2θ1θ 2 cos(θ 2 − θ1 )) θ1 l1θ1
2 2 l2θ 2
Potential energy of the system =
U = m1 g (1 − cos θ1 ) + m2 g {l1 (1 − cos θ1 ) + l2 (1 − cos θ 2 )} θ 2 − θ1
θ2
l1θ1
= g {( m1 + m2 )l1 (1 − cos θ1 ) + m2l2 (1 − cos θ 2 )}
So Lagrangian of the system = Figure 5
L = T −U
1 1
= m1l12θ12 + m2 (l1θ1 + l2θ2 + 2l1l2 cos(θ 2 − θ1 )) − g {(m1 + m2 )(1 − cos θ1 ) + m2 (1 − cos θ 2 )}
2 2
So using Lagrange principle, and assuming small angle of rotation, the equation of
motion can be written in matrix form as
⎡(m1 + m2 )l12 m2l1l2 ⎤ ⎛ θ1 ⎞ ⎡(m1 + m2 )l1 g 0 ⎤ ⎛ θ1 ⎞ ⎛ 0 ⎞
⎢ 2 ⎥ ⎜  ⎟
+⎢ ⎜ ⎟=⎜ ⎟
⎣ m2l1l2 m2l2 ⎦ ⎝ θ 2 ⎠ ⎣ 0 m2l2 g ⎥⎦ ⎝ θ 2 ⎠ ⎝ 0 ⎠

Now considering a special case when m1 = m2 = m and l1 = l2 = l , the above equation

becomes
⎡ 2ml 2 ml 2 ⎤ ⎛ θ1 ⎞ ⎡ 2mlg 0 ⎤ ⎛ θ1 ⎞ ⎛ 0 ⎞
⎢ 2 ⎥ ⎜ ⎟ + ⎢ ⎜ ⎟=⎜ ⎟
⎣ ml ml 2 ⎦ ⎝ θ2 ⎠ ⎣ 0 mlg ⎥⎦ ⎝ θ 2 ⎠ ⎝ 0 ⎠

⎡ 2 1⎤ ⎛ θ1 ⎞ ⎡ 2 0⎤ ⎛ θ1 ⎞ ⎛ 0 ⎞
or, ml 2 ⎢ ⎥ ⎜  ⎟ + mlg ⎢ ⎥⎜ ⎟ = ⎜ ⎟
⎣1 1⎦ ⎝ θ 2 ⎠ ⎣0 1⎦ ⎝θ2 ⎠ ⎝ 0 ⎠

1 ⎛ 1 −1⎞ ⎡ 2 0 ⎤ g ⎛ 2 −1⎞
Now A = 2 ⎜ ⎟ mlg ⎢ ⎥= ⎜ ⎟
ml ⎝ −1 2 ⎠ ⎣ 0 1 ⎦ l ⎝ −2 2 ⎠
To find eigenvalues of A,

218
 g g
−λ 2 −
l l
A − λI = 0 ⇒ =0
g g
−2 2 −λ
l l
2 2
⎛g⎞ g ⎛g⎞
Or, 4 ⎜ ⎟ − 4 λ + λ 2 − 2 ⎜ ⎟ = 0
⎝l ⎠ l ⎝l⎠
2
g ⎛g⎞
Or, λ − 4 λ + 2 ⎜ ⎟ = 0
2

l ⎝l ⎠
2 2
g ⎛ g⎞ ⎛g⎞
4 ± ⎜ 4 ⎟ − 8⎜ ⎟
l ⎝ l ⎠ ⎝l⎠ g
Or, λ = = (2 ± 2)
2 l
g g
Hence natural frequencies are ω1 = 0.7653 , ω2 = 1.8478
l l
The normal modes can be determined from the eigenvalues.
The corresponding principal modes are obtained as

g
⎛ θ1 ⎞ l 1
⎜ ⎟ = =
⎝ θ 2 ⎠λ =λ1 (2 − 2 + 2) g 2
l
g
⎛ θ1 ⎞ l 1
⎜ ⎟ = =−
⎝ θ 2 ⎠λ =λ2 (2 − 2 − 2) g 2
l
It may be noted that while in the first mode
Both the pendulum moves in the same direction, Figure 6
In the second mode they move in opposite direction
One may solve the same problem by taking x1 and x2 as the generalized coordinates.
Here x1 is the horizontal distance moves by mass m1 and x2 is the distance move by
mass m2 . Figure 7 show the free body diagram of both the masses.
T1 y y
T2
θ1 θ2
m2 x2 x
m1 x1 x
θ2
m2 g
m1 g Figure 7
T2

219
From the free body diagram of mass m2 ,
T2 cos θ 2 = m2 g
T2 sin θ 2 = − m2 
x2
Also from the free body diagram of mass m1 ,
T1 cos θ1 − T2 cos θ 2 = m1 g
T1 sin θ1 − T2 sin θ 2 + m1 
x2 = 0
Assuming θ1 and θ 2 to be small,
sin θ1 = tan θ1 = θ1 = x1 / l
and sin θ 2 = tan θ 2 = θ 2 = ( x2 − x1 ) / l
Hence
T2 = m2 g , and T1 = (m1 + m2 ) g
⎛ ⎛ ( m + m2 ) g ⎞ m2 g ⎞ − m2 g
x1 + ⎜ ⎜ 1
m1  ⎟+ ⎟⎟ x1 + x2 = 0
⎜
⎝⎝ l1 ⎠ l2 ⎠ l2

⎛x −x ⎞
x2 + m2 g ⎜ 2 1 ⎟ = 0
m2 
⎝ l2 ⎠
Hence in matrix form
⎡⎛ ⎛ (m1 + m2 ) g ⎞ m2 g ⎞ −m2 g ⎤
⎢⎜ ⎜ ⎟+ ⎟ ⎥
⎡ m1 0 ⎤ ⎛ x1 ⎞ ⎢⎜⎝ ⎝ l1 ⎠ l2 ⎟⎠ l2 ⎥ ⎛ x1 ⎞ ⎛ 0 ⎞
⎢0 ⎜ ⎟+⎢ ⎜ ⎟=⎜ ⎟
⎣ m2 ⎥⎦ ⎝ 
x2 ⎠ −m2 g m2 g ⎥⎥ ⎝ x2 ⎠ ⎝ 0 ⎠
⎢
⎢⎣ l2 l2 ⎥⎦
Considering the case in which m1 = m2 = m and l1 = l2 = l , the above equation becomes

x1 ⎞ g ⎡ 3 −1⎤ ⎛ x1 ⎞ ⎛ 0 ⎞
⎡1 0 ⎤ ⎛ 
⎢0 1 ⎥ ⎜  ⎟+ ⎢ ⎥⎜ ⎟ = ⎜ ⎟
⎣ ⎦ ⎝ x2 ⎠ l ⎣ −1 1 ⎦ ⎝ x2 ⎠ ⎝ 0 ⎠

220
 g ⎡ 3 −1⎤
A= ⎢ −1 1 ⎥
l ⎣ ⎦
g g
−λ
3 −
l l
and A − λ I = 0 ⇒ =0
g g
Hence − −λ
l l
2
g ⎛g⎞
Or, λ 2 − 4 λ + 2⎜ ⎟ = 0
l ⎝l ⎠

( ) g
(
or , λ1 = 2 − 2 and λ2 = 2 + 2
l
g
l
)
Same as those obtained by taking θ1 and θ 2 as the generalized coordinates.
g
⎛ X1 ⎞ l 1 1
⎜ ⎟ = = = = 0.4142
⎝ X 2
g
⎠λ =λ 3 − λ1 3 − 2 + 2 2.4142
1 l
Now
g
⎛ X1 ⎞ l 1 1
⎜ ⎟ = = =− = −2.4142
3 − λ2 3 − 2 − 2
⎝ X 2 ⎠λ =λ g 0.4142
2 l

1
1

-0.4142
2.4142
Figure 8

The different modes are as shown in the above figure.

Example Determine the equation of motion if the double pendulum is started with initial
conditions x1 (0) = x2 (0) = 0.5, x1 (0) = x2 (0) = 0.
Solution:
The resulting free vibration can be considered to be the superposition of the normal
modes. For each of these modes, the time solution can be written as

221
⎛ x1 ⎞ ⎛ X 1 ⎞ ⎛ x1 ⎞ ⎛ X 1 ⎞
⎜ ⎟ = ⎜ ⎟ sin ω1t ⎜ ⎟ = ⎜ ⎟ sin ω2t
⎝ x2 ⎠1 ⎝ X 2 ⎠1 ⎝ x2 ⎠2 ⎝ X 2 ⎠ 2
The general solution for the free vibration can be written as
⎛ x1 ⎞ ⎛ 0.4142 ⎞ ⎛ −2.4142 ⎞
⎜ ⎟= A⎜ ⎟ sin(ω1t +ψ 1 ) + B ⎜ ⎟ sin(ω2t + ψ 2 )
⎝ x2 ⎠ ⎝ 1 ⎠ ⎝ 1 ⎠

where A and B are the amounts of first and second mode’s participation and ψ 1 andψ 2
are the starting values or phases of the two modes. Substituting the initial conditions in
the above equation
⎛ 0.5 ⎞ ⎛ 0.4142 ⎞ ⎛ −2.4142 ⎞
⎜ ⎟= A⎜ ⎟ sinψ 1 + B ⎜ ⎟ sinψ 2
⎝ 0.5 ⎠ ⎝ 1 ⎠ ⎝ 1 ⎠

and

⎛0⎞ ⎛ 0.4142 ⎞ ⎛ −2.4142 ⎞

⎜ ⎟ = Aω1 ⎜ ⎟ cosψ 1 + Bω2 ⎜ ⎟ cosψ 2
⎝0⎠ ⎝ 1 ⎠ ⎝ 1 ⎠
For the second set of equations to be satisfied, cosψ 1 = cosψ 2 = 0 , so that ψ 1 = ψ 2 = 900 .
Hence A = 0.6035 and B = −0.1036 . So the equation for free vibration can be given by
⎛ x1 ⎞ ⎛ 0.4142 ⎞ ⎛ −2.4142 ⎞
⎜ ⎟ = 0.6035 ⎜ ⎟ cos ω1t − 0.1036 ⎜ ⎟ cos ω2t
⎝ x2 ⎠ ⎝ 1 ⎠ ⎝ 1 ⎠

Damped-free vibration of two-dof systems

Consider a two degrees of freedom system with damping as shown in figure

x1 x2 k3
k1 k2
m1 m2 c3
c1 c2

Figure 9
Now the equation of motion of this system can be given by

x1 ⎞ ⎡ k1 + k2
⎡ m1 0 ⎤ ⎛  −k2 ⎤ ⎛ x1 ⎞ ⎡ c1 + c2 −c2 ⎤ ⎛ x1 ⎞ ⎛ 0 ⎞
⎢ 0 m ⎥ ⎜  ⎟+⎢ ⎜ ⎟+ ⎜ ⎟=
k2 + k3 ⎥⎦ ⎝ x2 ⎠ ⎢⎣ −c2 c2 + c3 ⎥⎦ ⎝ x2 ⎠ ⎜⎝ 0 ⎠⎟
(32)
⎣ 2 ⎦ ⎝ x2 ⎠ ⎣ − k2

222
As in the previous case, here also the solution of the above equations can be written as
x1 = A1e st and x2 = A2 e st (33)
where A1 , A2 and s are constant . Substituting (33) in (32) , one may write

⎡ m1s 2 + (c1 + c2 ) s + k1 + k2 −c2 s − k2 ⎤ ⎛ A1 ⎞ ⎛ 0 ⎞

⎢ ⎥⎜ ⎟ = ⎜ ⎟ (34)
⎣ −c2 s − k2 m2 s + (c2 + c3 ) s + k2 + k3 ⎦ ⎝ A2 ⎠ ⎝ 0 ⎠
2

Now for a nontrivial response i.e., for non-zero values of A1 and A2 , the determinant of
their coefficient matrix must vanish. Hence

m1s 2 + (c1 + c2 ) s + k1 + k2 −c2 s − k2

=0 (35)
−c2 s − k2 m2 s 2 + (c2 + c3 ) s + k2 + k3
or, ( m1s 2 + (c1 + c2 ) s + k1 + k 2 )( m2 s 2 + (c2 + c3 ) s + k2 + k3 ) + (c2 s + k 2 ) 2 = 0 (36)
which is a fourth order equation in s and is known as the characteristic equation of the
system. This equation is to be solved to get four roots. The general solution of the system
can be given by
x1 = A11e s1t + A12 e s2t + A13e s3t + A14 e s4t
(37)
x2 = A21e s1t + A22 e s2t + A23e s3t + A24 e s4t
Here A1i , i = 1, 2,3, 4 are four arbitrary constants to be determined from the initial
conditions and the coefficients A2i , i = 1, 2,3, 4 are related to A2i and can be determined
from equation (34) as

A1i c2 si + k2
= (38)
A2i m1si + (c1 + c2 ) si + k1 + k2
2

For a physical system with damping, the motion will die out with time. For a stable
system, all the four roots must be either real negative numbers or complex number with
negative real parts. It may be recalled that, if the roots contain complex conjugate
numbers, the motion will be oscillatory.

Example: Find the response of the system as shown in figure 9 considering m1 = m2 = m,

k1 = k2 = k3 = k and c1 = c3 = 0 and c2 = c .
Solution.
In this case the characteristics equation becomes
(ms 2 + cs + 2k )(ms 2 + cs + 2k ) − (cs + k ) 2 = 0

223
(ms 2 + cs + 2k ) 2 − (cs + k ) 2 = 0
or , m 2 s 4 + 2mcs 3 + (4mk + c 2 − c 2 ) s 2 + (4kc − 2kc) s + 4k 2 − k 2 = 0
or , m 2 s 4 + 2mcs 3 + 4mks 2 + 2kcs + 3k 2 = 0
or , ms 2 (ms 2 + 2cs + 3k ) + k (ms 2 + 2cs + 3k ) = 0
or , (ms 2 + k )(ms 2 + 2cs + 3k ) = 0
or , (ms 2 + k )(ms 2 + 2cs + 3k ) = 0
Hence the roots are
2
k c ⎛c⎞ k
s1,2 = ±i and s3,4 = − ± ⎜ ⎟ − 3
m m ⎝m⎠ m
So the system has a pair of complex conjugate

SEMI-DEFINITE SYSTEMS

The systems with have one of their natural frequencies equal to zero are known as semi-
definite or degenerate systems. One can show that the following two systems are
degenerate systems.

x1 x2 θ1 θ2
k
m1 m2 I1 I2

Figure 10
Figure 11
From figure 10 the equation of motion of the system is
⎡ m1 0 ⎤ ⎛ 
x1 ⎞ ⎡ k −k ⎤ ⎛ x1 ⎞ ⎛ 0 ⎞
⎢ 0 m ⎥ ⎜  ⎟+⎢ ⎜ ⎟=⎜ ⎟
k ⎥⎦ ⎝ x2 ⎠ ⎝ 0 ⎠
(39)
⎣ 2 ⎦ ⎝ x2 ⎠ ⎣ −k

Assuming the solution x1 = A1eiωt and x2 = A2 eiωt (40)

⎡ k − m1ω 2 −k ⎤ ⎛ A1 ⎞ ⎛ 0 ⎞
⎢ ⎥⎜ ⎟ = ⎜ ⎟ (41)
⎣ −k k − m2ω ⎦ ⎝ A2 ⎠ ⎝ 0 ⎠
2

So for non-zero values of A1 , A2 ,

224
k − m1ω 2 −k
=0 (42)
−k k − m2ω 2

(
or, k − m1ω 2 )( k − m ω ) − k
2
2 2
=0 (43)

or, k 2 − k ( m1 + m2 )ω 2 + m1m2ω 4 − k 2 = 0 (44)

or, ω 2 ( m1m2ω 2 − k ( m1 + m2 )) = 0 (45)
k (m1 + m2 )
⇒ ω1 = 0, and, ω2 = (46)
m1m2
Hence, the system is a semi-definite or degenerate system. Corresponding to the first
mode frequency, i.e., ω1 = 0, A1 = A2 . So the system will have a rigid-body motion. For
the second mode frequency
A1 k km1m2 km1m2 m
= = = =− 2 (47)
A2 k − m1ω 2
km1m2 − m1k ( m1 + m2 ) − m1km1 m1
amplitude ratio is inversely proportional to the mass ratio the system.
Similarly one may show for the two-rotor system,
θ1 I
=− 2 (48)
θ2 I1
the ratio of angle of rotation inversely proportional to the moment of inertia of the rotors.

Forced harmonic vibration, Vibration Absorber

Consider a system excited by a harmonic force F1 sin ωt expressed by the matrix equation
⎡ m11 m12 ⎤ ⎛ 
x1 ⎞ ⎡ k11 k12 ⎤ ⎛ x1 ⎞ ⎛ F ⎞
⎢m ⎥ ⎜  ⎟ + ⎢ ⎥ ⎜ ⎟ = ⎜ ⎟ sin ωt (49)
⎣ 21 m22 ⎦ ⎝ x2 ⎠ ⎣ k21 k22 ⎦ ⎝ x2 ⎠ ⎝ 0 ⎠

Since the system is undamped, the solution can be assumed as

⎛ x1 ⎞ ⎛ X 1 ⎞
⎜ ⎟ = ⎜ ⎟ sin ωt (50)
⎝ x2 ⎠ ⎝ X 2 ⎠
Substituting equation (50) in equation (49), one obtains
⎡ k11 − m11ω 2 k12 − m12ω 2 ⎤ ⎛ X 1 ⎞ ⎛ F1 ⎞
⎢ ⎥ ⎜ ⎟ sin ω t = ⎜ ⎟ sin ωt
⎣ k21 − m21ω k22 − m22ω ⎦ ⎝ X 2 ⎠
2 2
⎝0⎠
(51)
⎡ k11 − m11ω 2 k12 − m12ω 2 ⎤ ⎛ X 1 ⎞ ⎛ F ⎞
or , ⎢ 2⎥⎜ ⎟=⎜ ⎟
⎣ k21 − m21ω k22 − m22ω ⎦ ⎝ X 2 ⎠ ⎝ 0 ⎠
2

225
 −1
⎛ X 1 ⎞ ⎡ k11 − m11ω 2 k12 − m12ω 2 ⎤ ⎛ F ⎞
⎜ ⎟=⎢ ⎥ ⎜ ⎟
⎝ X 2 ⎠ ⎣ k21 − m21ω k22 − m22ω 2 ⎦ ⎝ 0 ⎠
2

⎡ k22 − m22ω 2 − k12 + m12ω 2 ⎤ ⎛ F ⎞

⎢ ⎥⎜ ⎟ (52)
⎣ −k21 + m21ω 2 k11 − m11ω 2 ⎦ ⎝ 0 ⎠
=
k11 − m11ω 2 k12 − m12ω 2
k21 − m21ω 2 k22 − m22ω 2
Hence

X1 =
(k 22 − m22ω 2 F ) , (53)
Z (ω )
⎡ k11 − m11ω 2 k12 − m12ω 2 ⎤
where [ Z (ω ) ] = ⎢ ⎥
⎣ k21 − m21ω k22 − m22ω 2 ⎦
2

X2 =
(k 21 − m21ω 2 F ) (54)
Z (ω )
Example Consider the system shown in figure 12 where the mass m1 is subjected to a
force F sin ωt . Find the response of the system when m1 = m2 and k1 = k2 = k3 .

x1 x2
F sin ωt
k1 k2 k3
m1 m2

Figure 12
Solution:
The equation of motion of this system can be written as

⎡ m1 x1 ⎞ ⎡ k1 + k2
0 ⎤ ⎛  −k2 ⎤ ⎛ x1 ⎞ ⎛ F sin ωt ⎞
⎜  ⎟ + ⎢ ⎜ ⎟=
⎢0
⎣
⎥
m2 ⎦ ⎝ x2 ⎠ ⎣ −k2 k2 + k3 ⎥⎦ ⎝ x2 ⎠ ⎝⎜ 0 ⎠⎟

⎡ m 0 ⎤ ⎛ 
x1 ⎞ ⎡ 2k −k ⎤ ⎛ x1 ⎞ ⎛ F sin ωt ⎞
⎢ 0 m ⎥ ⎜  ⎟+⎢ ⎜ ⎟=⎜ ⎟
⎣ ⎦ ⎝ x2 ⎠ ⎣ − k 2k ⎥⎦ ⎝ x2 ⎠ ⎝ 0 ⎠

So assuming the solution

226
⎛ x1 ⎞ ⎛ X 1 ⎞
⎜ ⎟ = ⎜ ⎟ sin ωt and proceeding as explained before
⎝ x2 ⎠ ⎝ X 2 ⎠
⎡ 2k − mω 2 −k ⎤
[ Z (ω ) ] ⎢
= ⎥
⎣ −k 2k − mω 2 ⎦
k2
Z (ω ) = ( 2k − mω ) − k = m ω − 4mkω + 3k = m (ω − 4 ω + 3 2 )
2 2 2 2 4 2 2 2 4k 2
m m
k k
or, Z (ω ) = m 2 (ω 2 − )(ω 2 − 3 ) = m 2 (ω 2 − ω12 )(ω 2 − ω22 )
m m
k k
where, ω12 = and ω22 = 3 are normal mode frequencies of this system.
m m
Hence,

X1 =
( 2k − mω ) F
2

m 2 (ω 2 − ω12 )(ω 2 − ω22 )

kF
X2 =
m (ω − ω12 )(ω 2 − ω22 )
2 2

So it may be observed that the system will have maximum vibration

when ω = ω1 or , ω = ω2 . Also it may be observed that X 1 = 0, when ω 2 = 2k / m .
Tuned Vibration Absorber
Consider a vibrating system of mass m1 , stiffness k1 , subjected to a force F sin ωt . As
studied in case of forced vibration of single-degree of freedom system, the system will
have a steady state response given by
F sin ωt
x= , where ωn = k1 / m1
m(ωn2 − ω 2 ) (55)
which will be maximum when ω = ωn. Now to absorb this k1
vibration, one may add a secondary spring and mass F sin ωt
system as shown in figure 13.
m1 x1

k2

m2
x2

Figure 13

227
The equation of motion for this system can be given by

x1 ⎞ ⎡ k1 + k2
⎡ m1 0 ⎤ ⎛  − k2 ⎤ ⎛ x1 ⎞ ⎛ F sin ωt ⎞
⎢ 0 m ⎥ ⎜ ⎟+⎢ ⎜ ⎟=⎜ ⎟
k2 ⎦⎥ ⎝ x2 ⎠ ⎝ 0 ⎠
(56)
⎣ 2 ⎦ ⎝ x2 ⎠ ⎣ − k2
Comparing equation (49) and (56),
m11 = m1 , m12 = 0, m21 = 0, m22 = m2, k11 = k1 + k2 , k12 = − k2 , k21 = − k2 , and k 22 = k .
Hence,
k1 + k2 − m1ω 2 − k2
Z (ω ) = = k1k2 − m1k2ω 2 − k1m2ω 2 − k2 m2ω 2 + m1m2ω 4
−k2 k2 − m2ω 2

= m1m2 (λ1 − ω 2 )(λ2 − ω 2 ) (57)

where λ1 and λ2 are the roots of the characteristic equation Z (ω ) = 0 of the free-
vibration of this system., which can be given by
⎧ 2 ⎫
⎪⎛ k1 k2 k2 ⎞ ⎛ k1 k2 k2 ⎞ kk ⎪
λ1,2 = 0.5 ⎨⎜ + + ⎟± ⎜ + + ⎟ −4 1 2 ⎬ (58)
⎪⎩⎝ m1 m2 m1 ⎠ ⎝ m1 m2 m1 ⎠ m1m2 ⎪⎭
Now from equation (53) and (54)

X1 =
(k 22 − m22ω 2 F) =
(k 2 − m2ω 2 F ) , (59)
Z (ω ) Z (ω )
− k2 F
X2 = (60)
Z (ω )
k2
From equation (59), it is clear that, X 1 = 0, when ω 2 = . Hence, by suitably choosing
m2
the stiffness and mass of the secondary spring and mass system, vibration can be
k
completely eliminated from the primary system. For ω 2 = 2 ,
m2
k2 k k k k
Z (ω ) = k1k2 − m1k2 − k1m2 2 − k2 m2 2 + m1m2 2 2
m2 m2 m2 m2 m2
(61)
k2 k2
= k1k2 − m1 2 − k1k2 − k22 + m1 2 = − k22
m2 m2
− k2 F F
and X 2 = = (62)
− k22 k2

228
Centrifugal Pendulum Vibration Absorber

The tuned vibration absorber is only effective when the frequency of external excitation
equals to the natural frequency of the secondary spring and mass system. But in many
cases, for example in case of an automobile engine, the exciting torques are proportional
to the rotational speed ‘n’ which may vary over a wide range. For the absorber to be
effective, its natural frequency must also be proportional to the speed. The characteristics
of the centrifugal pendulum are ideally suited for this purpose.

Placing the coordinates through point O’, parallel and normal to r, the line r rotates with
angular velocity ( θ + φ )

ĵ
iˆ

R
O′
O

The acceleration of mass m

am = ⎡⎣ − Rθ 2 cos φ + Rθ sin φ − r (θ + φ) 2 ⎤⎦ iˆ + ⎡⎣ Rθ 2 sin φ + Rθ cos φ + r (θ + φ) 2 ⎤⎦ ˆj (63)
Since the moment about O′ is zero,
M O′ = m ⎡⎣ Rθ 2 sin φ + Rθ cos φ + r (θ + φ) 2 ⎤⎦ r = 0 (64)
Assuming φ to be small, cos φ = 1, sin φ = φ , so
⎛R ⎞ ⎛ R+r ⎞
φ + ⎜ θ 2 ⎟ φ = − ⎜ ⎟θ
 (65)
⎝r ⎠ ⎝ r ⎠
If we assume the motion of the wheel to be a steady rotation n plus a small sinusoidal
oscillation of frequency ω , one may write
θ = nt + θ 0 sin ωt (66)
θ = nt + ωθ 0 cos ωt ≅ n (67)
θ = −θ 0ω 2 sin ωt (68)
Substituting the above equations in equation (65) yields,

229
 ⎛R 2⎞ ⎛ R+r ⎞ 2
φ + ⎜n ⎟φ = ⎜ ⎟ ω θ 0 sin ωt (69)
⎝r ⎠ ⎝ r ⎠
Hence the natural frequency of the pendulum is
R
ωn = n (70)
r
and its steady-state solution is
(R + r) / r
φ= ω 2θ 0 sin ωt (71)
−ω 2 + ( Rn 2 / r )
It may be noted that the same pendulum in a gravity field would have a natural
g
frequency of . So it may be noted that for the centrifugal pendulum the gravity
r
field is replaced by the centrifugal field Rn 2 .

Torque exerted by the pendulum on the wheel

With the ĵ component of am equal to zero, the pendulum force is a tension along r ,
given by m times the iˆ component of am .

( )
T = R cos φ iˆ + R sin φ ˆj × m ⎡⎣ − Rθ 2 cos φ + Rθsin φ − r (θ + φ) 2 ⎤⎦ iˆ
(72)
= − mRφ ⎡⎣ − Rω 2θ 0 sin ωt sin φ − Rn 2 − rn 2 − rφ2 − 2rθφ
 ⎤
⎦
Now assuming small angle of rotation
T = −m ( R + r ) n 2 Rφ (73)
Now substituting the (73) in (72),
−mR( R + r ) 2 n 2 / r 2
T= ω θ0 sin ωt
( Rn 2 / r ) − ω 2
⎡ m( R + r ) 2 ⎤ 
= −⎢ 2⎥
θ = J eff θ (74)
⎣ 1 − rω 2
/ Rn ⎦
Hence the effective inertia can be written as
⎡ m( R + r ) 2 ⎤ m( R + r ) 2
J eff = − ⎢ 2 ⎥
= − (75)
⎣1 − rω / Rn ⎦ 1 − (ω / ωn )
2 2

which can be ∞ at its natural frequency. This possesses some difficulties in the
design of the pendulum. For example to suppress a disturbing torque of frequency
equal to four times the natural speed n, the pendulum must meet the requirement
ω 2 = (4n) 2 = n 2 R / r . Hence, as the length of the pendulum r = R /16 becomes very

230
 small it will be difficult to design it. To avoid this one may go for Chilton bifilar
design.

Exercise problems
1. In a certain refrigeration plant, a section of pipe carrying the refrigerant vibrated violently
at a compressor speed of 232 rpm. To eliminate this difficulty, it was proposed to clamp a
cantilever spring mass system to the pipe to act as an absorber. For a trial test, for a 905
gm. Absorber tuned to 232 cpm resulted in two natural frequencies of 198 and 272 cpm.
If the absorber system is to be designed so that the natural frequencies lie outside the
region 160 to 320 cpm, what must be the weight and spring stiffness?
2. Derive the normal modes of vibration of a double pendulum with same length and
mass of the pendulum.
3. Develop a matlab code for determination of free-vibration of a general two-degree
of freedom system.
4. Derive the equation of motion for the double pendulum shown in figure p1 in terms of θ1
and θ2 using Lagrange principle. Determine the natural frequencies and mode shapes of
the systems. If the system is started with the following initial conditions: x1(0) =x2(0) = X,
v1(0) =v2(0)=0, (v1 and v2 are velocity) determine the equation of motion. If the lower
mass is given an impulse F0 δ (t), determine the response in terms of normal modes.

θ1 L1

m1
L2
x1 θ2
m2

x2

Figure P1
5. A centrifugal pump rotating at 500 rpm is driven by an electric motor at 1200 rpm
through a single stage reduction gearing. The moments of inertia of the pump impeller
and the motor are 1600 kg.m2 and 500 kg.m2 respectively. The lengths of the pump shaft
and the motor shaft are 450 and 200 mm, and their diameters are 100 and 50 mm
respectively. Neglecting the inertia of the gears, find the frequencies of torsional
oscillations of the system. Also determine the position of the nodes.

231