Vous êtes sur la page 1sur 57

S.

Ilanko 2007





3.1 Equations of Motion using Newtons 2
nd
Law & Calculation of Natural Frequencies

Even in a free country there are laws that its people must abide by. In vibration, the motion
along any degree of freedom is also subject to a law, which may be expressed as Newtons
second law of motion. We were able to write an equation of motion for a single degree of
freedom system by summing all the actions (forces or moments) along the direction of motion
in which the system is free to move, and equating the sum of the actions to the product of
acceleration and an inertial resistance factor (this factor being mass for a translational degree
of freedom, or moment of inertia for a rotational degree of freedom). A discrete system with n
degrees of freedom will have n equations of motion. This means more work for us. In
vibration, freedom seems to come with a price! These equations may be obtained by applying
Newtons second law of motion to the discrete masses along each of the n independent
translational/rotational co-ordinates. This is not the only way to obtain the equations of
motion. In another section, we will study an alternative method based on an energy principle
to derive these equations. For now, let us consider the application of Newtons second law of
motion to find the natural frequencies and modes, using a very simple 2-DOF spring-mass
system (throughout this text, the term mass will be used to indicate a particle possessing
mass) as shown in Figure 3.1.1, as an illustration. The 2-DOF system is only an illustration
and we will present the steps using matrix notation in a general way so that one can see how
this could be done for any discrete system.

State 2 (at time t during vibration)
State 1 (equilibrium, springs unstressed)
T
1

T
2

T
2

u
1

u
2

m
2

k
1
k
2

Figure 3.1.1 A 2 DOF system
m
1

3. Systems with Several Degrees of Freedom
S. Ilanko 2007
50

Let the dynamic displacement of the masses be u
1
, u
2
and their amplitudes be
1
,
2
. Since x,y
are used for representing the static co-ordinates of continuous systems, the dynamic
displacements will be denoted by u,v, and to be consistent we will use the same notation in
this book.

For determining the natural modes and frequencies, the motion may be assumed to be simple
harmonic.

If u
1
=
1
sin(t+) and u
2
=
2
sin(t+)
then the accelerations are given by
( )
1
2
1
2
1
sin u t u u = + = and
2
2
2
u u = .
Applying Newtons second law to m
1
along u
1
(see Figure 3.1.2),
1 1 1 2
u m T T =
Using the constitutive equations for the springs, ( )
1
2
1 1 1 1 1 1 2 2
u m u m u k u u k = =
Rearranging this we get, (k
1
+k
2
)u
1
-k
2
u
2
= m
1

2
u
1
...(3.1.1)
This is the first equation of motion.
Similarly for the second mass, k
2
(u
2
-u
1
) = m
2

2
u
2
...(3.1.2)
These equations of motion may be given in the following matrix form,



...(3.1.3)

where the elements of the generalized stiffness and mass matrices are as follows:
(

+
=
2 2
2 2 1
) (
] [
k k
k k k
K and
(

=
2
1
0
0
] [
m
m
M

...(3.1.4a,b)
This is a standard eigenvalue problem, and for any linear, nDOF discrete system, one can
derive n equations of motion in the form of (3.1.3). That is the resulting matrix equation will
contain an n n stiffness matrix and an n n mass matrix. This eigenvalue equation may be
solved using iterative methods, or by finding the roots of a determinantal frequency equation
giving n natural frequencies and modes. For a 2-DOF system, explicit expressions for the
natural frequencies can be obtained as follows.

[K]{u}=
2
[M]{u}
m
1

1
u
T
1

T
2

Figure 3.1.2 Freebody diagram
for

particle m
1

S. Ilanko 2007
51
From equation (3.1.1),
2 2
1 2 1
2
1
) (
u
m k k
k
u
+
= ...(3.1.4c)
Substituting this into equation (3.1.2) gives
2
2
2 2 2
1 2 1
2
2 2 2
) (
u m u
m k k
k
k u k

=
+

0
) (
2
2
2 2
1 2 1
2
2 2
=
|
|
.
|

\
|

+
u m
m k k
k
k k


This is true if either 0
2
= u
or 0
) (
2
2 2
1 2 1
2
2 2
=
+

m
m k k
k
k k
However, if 0
2
= u then from equation (3.1.4c), 0
1
= u and the system will not vibrate. This is
a trivial solution. For non-trivial solution, 0
) (
2
2 2
1 2 1
2
2 2
=
+

m
m k k
k
k k
This is the frequency equation. The roots of this frequency equation are the two natural
frequencies. Substituting the natural frequencies into either equation (3.1.1) or (3.1.2) gives
the ratio of the displacements (u
1
/ u
2
)
.

As an example, let k
1
= 100N/m, k
2
=200 N/m, m
1
= 0.2 kg and m
2
= 0.3 kg. This yields the
following values for the natural frequencies (eigenvalues) and modes (eigenvectors):
First mode:
1
= 12.91 rad/sec, (
1
/
2
)
1
= 0.75 ..(3.1.5a,b)
Second mode:
2
= 44.72 rad/sec, (
1
/
2
)
2
= -2.0 ..(3.1.6a,b)
The system can vibrate freely at any of the two frequencies without any external dynamic
force. If it vibrates purely in one of the above modes at the corresponding frequency, it is
then called a principal vibration. The actual dynamic displacement of a system depends on
initial conditions, and is generally a combination of all its natural modes as described in the
next section.

For an n-DOF system equation (3.1.3) may be solved using iteration. The following procedure
may also be used, but it is often not convenient.
Rewriting equation (3.1.3),
Let [K
d
] =[K]-
2
[M],
then [K
d
] {u}={0}.
For non-trivial solution, |K
d
|=0
This is the frequency equation.
S. Ilanko 2007
52
From here onwards, the following matrix notation will also be used to represent the principal
modes as a matrix called modal matrix:



..(3.1.7)


In this matrix the i
th
row represents the i
th
degree of freedom, and the j
th
column represents the
j
th
mode. i.e.
j i
a
,
is the displacement (or rotation) corresponding to the i
th
degree of freedom
in the j
th
principal mode. The modes are normalized either by setting the maximum amplitude
to unity, or by setting energy terms to unity. In the above example, by setting the amplitudes
to unity, the modal matrix may be written as:

(

0.5 1.00
1.00 0.75
..(3.1.8)

The normalized modes are referred to as normal modes. This is not the only way to normalise
the modal matrix. Another more useful normalisation method will be discussed later.

Derivation of the Equations of Motion using Newtons Second Law
A freebody diagram of each particle of rigid body must be sketched. In addition to the forces
and moments acting on each freebody, the accelerations and the relevant mass or moment of
inertia must also be shown.
Then Newtons second law should be applied either in a translational direction or a rotational
direction.
The simple harmonic relationship may be used to express the accelerations in terms of the
frequency and displacement.
The resulting equations are arranged in a matrix form to obtain the eigenvalue equation.

The derivation of the equations of motion by applying Newtons second law of motion for
many practical problems can be very difficult and a more convenient method based on the
potential and kinetic energies of the system will be discussed in another chapter.
(
(
(
(

n,n
i,j
, ,
, ,
a
a
a a
a a
a
2 2 1 2
2 1 1 1
] [ matrix modal The =
S. Ilanko 2007
53
m
b,
I
b
m
b
,

I
b

Example 1

Two rigid bars are hinged at points O and
P, and connected by means of an elastic
spring of stiffness k as shown in Figure
3.1.3. They also carry four concentrated
masses at their ends. The bars are
identical having a mass m
b
(which is
equal to 0.2m) and a moment of inertia of
I
b
(which is 0.2mL
2
/12) about their
centroidal axis perpendicular to the plane
of oscillation. This system is in static
equilibrium under gravity field g.




P
O
g
0.2L
0.3L
0.5L
0.2L
0.7L
0.1L
k
m
m
2 m
2.5 m
Figure 3.1.3
S. Ilanko 2007
54
Example 2

Figure 3.1.4 is a schematic representation of a power transmission system after the sudden
application of a brake on the left end of the top shaft.
The relevant moments of inertia of the gears and the rotor are:
I
2
= 1.2 kgm
2
;
I
3
= 0.15 kgm
2
; and
I
4
= 3 kgm
2

(a.) Obtain the equations of motion for the torsional vibration of the above system
neglecting the mass of the shafts (treating the system as discrete).
(b.) Show that the natural frequencies and modes of the above system are as follows:

1
= 18.34 rad/s, and for the first mode,
2
/
4
= 0.374

2
= 115.0 rad/s, and for the second mode,
2
/
4
= 4.458
I
2
k
1
=6 kNm/rad
I
3
I
4
k
2
= 4 kNm/rad
2r
r
Figure 3.1.4

4
(t)

2
(t)
S. Ilanko 2007
55
3.2 Response of a 2-DOF Undamped System

Transient Response

In the preceding example, if the masses were initially displaced by 5 mm from their
equilibrium state, and then released, the response may be calculated as follows:
Initial conditions:
u
1
(0) = 5 mm ; u
2
(0) = 5 mm.
Also 0 ) 0 (
1
= u ; 0 ) 0 (
2
= u
Let ( ) ( )
2 2
2
1 1 1
2
1
sin
5 0
00 1
sin
00 1
75 0
t+
. -
.
G t+
.
.
G
u
u
)
`

+
)
`

=
)
`


Substituting the first two initial conditions we get,

) + sin(0
0.5 -
1.00
G ) + sin(0
1.00
0.75
G
5
5
2
2
1 1

)
`

+
)
`

=
)
`

..(3.2.1)

Using the last two initial conditions,

) + cos(0
0.5 -
1.00
G ) + cos(0
1.00
0.75
G
0
0
2 2 1 1 1 2

)
`

+
)
`

=
)
`

..(3.2.2)

The above equation is satisfied if
1
=
2
=/2. Substituting these into equation (3.2.1) gives:

)
`

+
)
`

=
)
`

5 0
00 1
00 1
75 0
5
5
2
1
. -
.
G
.
.
G
Solving the two simultaneous equations gives: G
1
= 5.45 and G
2
= 0.90
Hence the response is:
( ) ( ) 2 72 44 sin
45 0
91 . 0
2 91 12 sin
45 . 5
09 . 4
2
1
/
)
`

+ /
)
`

=
)
`

t+ .
. -
t+ .
u
u

i.e. ( ) ( ) t . t .
u
u
72 44 cos
45 . 0
91 . 0
91 12 cos
45 . 5
09 . 4
2
1
)
`

+
)
`

=
)
`

..(3.2.3)

This is the required dynamic response. To develop a more convenient general procedure for
obtaining the dynamic displacement, it is necessary to consider an important relationship
S. Ilanko 2007
56
between the natural modes called the orthogonality relationship. This is done in section 3.3
and a more convenient procedure for calculating the transient response of a discrete system is
described in section 3.4.

Steady State Response Due to Simple Harmonic Force

Let us consider the vibration of a 2-DOF spring-mass system subject to a harmonic excitation
as shown in Figure 3.2.1.
Applying Newtons second law of motion to the two particles gives the following equations:
1 1 1 2 1
sin u m T T t F = +
2 2 2
u m T =
Substituting the constitutive equations for the springs into the above equations and
rearranging leads to the following equations of motion:
t F u k u m u k k = + + sin ) (
1 2 2 1 1 1 2 1

0
2 2 2 2 1 2
= + + u m u k u k
By inspection, it may be seen that the forms pt u u sin
1 1
= and pt u u sin
2 2
= satisfy the above
equations of motion. Substitution gives:
t F u k u p m k k = + sin ) (
1 2 2 1
2
2 2 1
..(3.2.4a)
State 2 (at time t during vibration)
State 1 (equilibrium, springs unstressed)
T
1

T
2

T
2

u
1

m
2

k
1
k
2

Figure 3.2.1 A 2 DOF system Subject to a SH Force
m
1

F
1
sin t
F
2
sin t
T
2

F
1
sin t

T
1
T
2

2

FBD of m
1

FBD of m
2

S. Ilanko 2007
57
0 ) (
2
2
2 2 1 2
= + u m k u k ..(3.2.4b)
From equation (3.2.4b),
) (
2
2 2
1 2
2

=
m k
u k
u ..(3.2.4c)
Substituting this into equation (3.2.4a) gives:
( )
t F u
m k
k m k m k k
=

+
sin
) (
) )( (
1 1 2
2 2
2
2
2
2 2
2
2 2 1

( )
2
2
2
2 2
2
2 2 1
1
2
2 2
1
) )( (
sin ) (
k m k m k k
t F m k
u
+

= ..(3.2.4d)
Substituting this into equation (3.2.4c) gives:
( )
2
2
2
2 2
2
2 2 1
1 2
2
) )( (
sin
k m k m k k
t F k
u
+

= ..(3.2.4d)
It may be noted that if 0
2
2 2
= m k then 0
1
= u
This is significant and can be used to control the vibration due to harmonic excitation.

An undamped vibration absorber
For example, if a SDOF is subject to a harmonic excitation as shown in Figure 3.2.2, then
adding another SDOF consisting of a mass m
2
and stiffness k
2
such that 0
2
2 2
= m k will
ensure that the original system will remain undisturbed by the excitation as all the excitation
force will be absorbed by the subsystem.

Consider the attached spring-mass system as a separate system. The natural frequency of this
system is given by
2 2
/ m k . This means that if the attached system has a natural frequency
equal to the excitation frequency of the main system then the main system will not vibrate.
This is called an undamped vibration absorber or a detuner.
u
1

m
2

Figure 3.2.2 An undamped vibration absorber
m
1

F
1
sin t
Original SDOF system
Detuned subsystem
k
1

k
2

m
2

u
1

m
1

F
1
sin t
k
2

k
1

u
2

Added subsystem
S. Ilanko 2007
58
3.3 Orthogonality of Modes.

In the preceding example, it may be seen that m
1.

1 , 1
a .
2 , 1
a + m
2.

1 , 2
a
.
2 , 2
a = 0.
For any discrete system with n degrees of freedom, we will show that there is a similar
relationship between any two different natural modes. We will initially restrict our arguments
to systems that have a diagonal mass matrix. If all displacements share the same vector
direction then we will prove that


...........(3.3.1a)


The above relationship is called the orthogonality property of the natural modes. In a situation
where the displacements are in different directions, the above equation still holds, if care is
taken to ensure that a mass that has two or three degrees of freedom, appears twice or three
times in the expression, the number of such appearances being equal to the number of degrees
of freedom. That is to say that the subscript i strictly refers to a degree of freedom and not a
mass. For example if a particle having mass m
0
is associated with the third and fourth degrees
of freedom, then both m
1
and m
2
should be set to m
0
. This is explained through a 2-DOF
spring-mass system at the end of this section.

In general the mass terms m
i
may include moments of inertia
i
in which case the
displacement co-ordinates a
i,j
would be angle of rotations along the degrees of freedom.

It is convenient to express the orthogonality relationship in matrix form:

...........(3.3.1b)

where [ M ] is a diagonal matrix which is sometimes referred to as the generalised mass
matrix. In this book, we will simply call it the diagonalised inertia matrix. The elements of
this matrix may be associated with either rotational or translational motion, and inertia matrix
is a more suitable term.

, 0
1
=

=
i,k i,j i
a a m
n
i
for j k
[a]
T
[M][a] = [ M ]
S. Ilanko 2007
59
For a n DOF system, there are n(n-1) number of orthogonality relationships which are
represented by the equations from (3.3.1b) that give rise to the off-diagonal terms in [ M ].
The modes may be normalised by making this an identity matrix.

For example, to normalise the first mode in the previous example,
m
1
(
1 , 1
a )
2
+ m
2
(
1 , 2
a )
2
=1 .............(3.3.2)
But
1 , 1
a /
1 , 2
a = 0.75/1.0 .............(3.3.3)
Solving these two equations gives:
1 , 2
a = 1.557 and
1 , 1
a = 1.168
Similarly the second mode may be normalized to obtain:
2 , 1
a = 1.907 and
2 , 2
a = -0.953
Hence the alternative to normalised modes in equation (3.1.8) is:
(

0.953 1.557
1.907 1.168
.............(3.3.4)
This would give an identity matrix on the right hand side of equation (3.3.1b). This type of
normalisation is often used in response calculations as explained later.

Proof of Orthogonality:
Consider the free vibration of an n degree of freedom
system in its j
th
mode. The freebody diagram of a
typical mass with forces and acceleration along the i
th

degree of freedom is shown in Figure 3.3.1.

The net elastic restoring force on the mass is
F
i,j
= - m
i

j
2
a
i,j


By d'Alambert's principle, if each of the
masses were subject to a static force F
i,j
= m
i

j
2
a
i,j
then the resulting displacements of the
masses will be given by a
i,j
. Now consider
the forces associated with the k
th
mode of
vibration. Similarly we can say that forces
F
i,k
= m
i

k
2
a
i,k
will result in displacements
F
i,j
F
i,k
a
i,j a
i,k
Force

Displacement

Figure 3.3.2
Acceleration
-
j
2
a
i,j

m
i

F
i,j

Figure 3.3.1 Freebody
Diagram of a typical mass
S. Ilanko 2007
60
a
i,k
.

From Betty-Maxwell's reciprocal theorem, for linear elastic structures, the work done by the
forces F
i,j
while undergoing displacements a
i,k
is equal to the work done by the second set of
forces F
i,k
over displacements a
i,j
. This can be explained by equating the work done by the
two sets of forces applied in different order.

First let us consider the work done on mass m
i
by forces F
i,j
while causing displacement a
i,j
.
This is given by the dotted triangular area under the force-displacement curve shown in
Figure 3.3.2 and is equal to
j i j i
a F
, ,
2
1


Now let us apply forces F
ik
which cause displacements a
i,k
. The work done by the force F
ik
on
mass m
i
is given by the hatched triangular area and is equal to
k i k i
a F
, ,
2
1
. However as
displacements a
i,k
. are being caused by F
ik
the force F
i,j
that is already acting on the system
would also do work equal to F
i,j
a
i,k
. shown by the shaded rectangular area in Figure 3.3.2. The
net work done on the ith mass is therefore
j i j i
a F
, ,
2
1
+
k i k i
a F
, ,
2
1
+ F
i,j
a
i,k
.
The total work done on the system is given by summing the work done for all masses. This is

=
|
.
|

\
|
+ +
n
i
k i j i k i k i j i j i
a F a F a F
1
, , , , , ,
2
1
2
1

If the order of application of the forces were reversed (that is forces F
i,k
is applied first
followed by F
i,k
) then the net work done is

=
|
.
|

\
|
+ +
n
i
j i k i k i k i j i j i
a F a F a F
1
, , , , , ,
2
1
2
1

For a linear elastic structure, the order of application of force has no effect on its behaviour.
Therefore the total work done may be equated.

=
|
.
|

\
|
+ +
n
i
k i j i k i k i j i j i
a F a F a F
1
, , , , , ,
2
1
2
1
=

=
|
.
|

\
|
+ +
n
i
j i k i k i k i j i j i
a F a F a F
1
, , , , , ,
2
1
2
1

This is a statement of the reciprocal theorem.
Cancelling the common terms, we get

= =
=
n
i
j i k i
n
i
k i j i
a F a F
1
, ,
1
, ,

Using dAlemberts principle,
( )

=
n
i
k i j i j i
a a m
1
, ,
2
= ( )

=
n
i
j i k i k i
a a m
1
, ,
2

S. Ilanko 2007
61
Rearranging gives: ( )( )
2 2
1
, , k j
n
i
k i j i i
a a m

=
= 0
If
j

k
, then ( )

=
n
i
k i j i i
a a m
1
, ,
= 0.

This means that if two modes have different natural frequencies then the sum of the product
of the masses and the displacements/rotations corresponding to those modes will be zero. The
modes are said to be orthogonal. In cases where the co-ordinates a
i,j
and a
j,k
are in different
directions, the work done is obtained by taking the dot product of the force vector (and
displacement vector. An alternative matrix approach for a general proof of orthogonality is
given below:

When a discrete system vibrates in the j
th
mode, equation (3.1.3) reduces to:
[K]{a}
j
= (
j
)
2
[M]{a}
j

where {a}
j
is the j
th
mode which is the j
th
column of the modal matrix.

Pre-multiplying both sides of this equation by
T
} {
k
a gives:
T
} {
k
a [K]{a}
j
= (
j
)
2 T
} {
k
a [M]{a}
j


Similarly pre-multiplying the equations of motion corresponding to the k
th
mode by the
transpose of the j
th
mode gives:
T
} {
j
a [K]{a}
k
= (
k
)
2 T
} {
j
a [M]{a}
k


Since the stiffness matrix [K] is symmetrical the left hand side of the last two equations are
equal. Therefore we can equate the right hand side of these equations to obtain:
(
j
)
2 T
} {
k
a [M]{a}
j
= (
k
)
2 T
} {
j
a [M]{a}
k
..(3.3.5)
Since the mass matrix is also symmetrical
T
} {
k
a [M]{a}
j
=
T
} {
j
a [M]{a}
k

Substituting this into equation (3.3.5) gives:
(
j
)
2 T
} {
j
a [M]{a}
k
= (
k
)
2 T
} {
j
a [M]{a}
k

If the frequencies
j
and
k
are not equal, for the above equation to be true,

..(3.3.6) T
} {
j
a [M]{a}
k
= 0

S. Ilanko 2007
62

This is a matrix representation of the orthogonality property of natural modes. Since j and k
are arbitrary, a series of the above equations for any combination of j and k (jk) may be
written.
If j = k, then

..(3.3.7a)


Therefore we can write a matrix equation of the form (3.3.1b) which we will restate now.
[a]
T
[M][a] = [ M ] ....(3.3.1b)

where [ M ] is a diagonal matrix.

The elements of the diagonal matrix are defined by equation (3.3.7a) and in systems where
the original mass matrix is diagonal, they may also be expressed in the following form:

..(3.3.7b)


For an n-DOF system, it is possible to write n equation of the form (3.3.7). This gives us
another way to normalise the modes. We can make [ M ]an identity matrix [I] by scaling the
modes such that
T
} {
j
a [M]{a}
j
=1
Such a normalized mode is called the normal mode, and may be denoted by [N]

..(3.3.8)


Orthogonality with respect to stiffness matrix

We will now show that a similar orthogonal relationship with respect to stiffness also exists.
Applying equation (3.1.3) to a system that is vibrating in its k
th
natural mode we get
[K]{a}
k
= (
k
)
2
[M]{a}
k

[N]
T
[M][N] = [I]
2
, ,
) (
r i
n
i
i r r
a m M

=
T
} {
j
a [M]{a}
j
=
j j
M
,
0

S. Ilanko 2007
63
Substituting this into equation (3.3.6) gives
T
} {
j
a
2
1
k

[K]{a}
k
= 0
This gives
T
} {
j
a [K]{a}
k
= 0 ..(3.3.9)
It also follows that if j = k, then
T
} {
j
a [K]{a}
j
=
j j
K
,
= (
j
)
2

j j
M
,
..(3.3.10)
and [a]
T
[K][a] = [ K ], where
j j
K
,
= (
j
)
2

j j
M
,
..(3.3.11)
In terms of the normal modes,
[N]
T
[K][N] = [ K ] ..(3.3.12)
where [ K ] is a diagonal matrix whose elements contain the eigenvalues.
That is
2
, j j j
K = ..(3.3.13)

These relationships will be very useful in the calculation of dynamic displacements of
vibratory systems as explained in the following sections.

A geometrical and vectorial interpretation of orthogonality

To illustrate the situation where one needs to be careful in applying the orthogonality
condition to a system where a mass may be associated with more than one degree of
freedom, let us consider a spring mass system that is free to vibrate in x-y plane.

This system consists of a particle of mass m, restrained by two springs, one oriented along the
y axis and another at angle to the y axis as shown in Figure 3.3.3. Let the stiffness of the
springs in the y and the inclined directions be k
1
and k
2
and the force in these springs be F
1

Figure 3.3.3
m

v

k
1
k
2
u


F
1
F
2
u
v
m


S. Ilanko 2007
64
and F
2
respectively. Let the independent coordinates be u and v, these being measured in the x
and y directions respectively.
For small displacements, the elongation of the springs 1 and 2 would be v and (u sin + v cos
) respectively.
From these we can write the constitutive equations in terms of the displacements:
v k F
1 1
= and ( ) cos sin
2 2
v u k F + =

Writing Newtons second law in x and y directions, we get:
u m F = sin
2
= u m
2

and v m F F = + cos
2 1
= v m
2

Substituting the constitutive equations into the above we get,
[ ]
)
`

=
)
`

v
u
M
v
u
K
2
] [ (3.3.14)
where[ ]
(

+
=


2
2 1 2
2
2
2
cos cos sin
cos sin sin
k k k
k k
K and [ ]
(

=
1 0
0 1
m M
For convenience, let k k k = =
2 1

[ ]
(

+
=


2
2
cos 1 cos sin
cos sin sin
k K
For non-trivial solution, | K- M
2
| = 0 gives the following roots for
2

( )( ) m k / cos 1
2
1
= and ( )( ) m k / cos 1
2
2
+ =
The corresponding modes may be found by evaluation ( ) v u / which gives the tangent of
direction of motion of the mass. From the first row of equation (3.3.14),
|
.
|

\
|
v
u
=
2 2
sin
cos sin


m k
k

which for the fits and second natural frequencies give the following
results:
) cos 1 ( sin
cos sin
2
1



=
|
.
|

\
|
k k
k
v
u
=


2
cos cos
cos sin

cos 1
sin


and
) sin 1 ( sin
cos sin
2
2


+
=
|
.
|

\
|
k k
k
v
u
=


2
cos cos
cos sin
) cos 1 (
sin

+
=

Multiplying these tangents give 1 showing that these directions are perpendicular to each
other. This is a situation where one can see that two modes are geometrically orthogonal
(perpendicular). The dot product of two non-zero vectors would be zero only if they are
S. Ilanko 2007
65
perpendicular, and for a situation involving vibration of one particle possessing mass in
possibly more than one translational sense, then the modal directions must be perpendicular to
each other. It should be mentioned here that this statement is only true in the absence of
rotational motion. The orthogonality statement for a rigid body that is capable of rotational
motion would include a term associated with its moment of inertia and two rotational
displacement values corresponding to two modes, and in such a situation the geometric
interpretation of orthogonality of modes does not apply.

Let us write down the modal matrix for this problem in terms of its vector components in the
x and y directions. i.e. [ ]
(

=
2 1
2 1
v v
u u
a =
(
(

|
.
|

\
|
+

|
.
|

\
|

1 1
cos 1
sin
cos 1
sin


The mass matrix is[ ]
(

=
m
m
M
0
0

[ ] [ ][ ] = a M a
T
(
(
(
(

|
.
|

\
|
+

|
.
|

\
|

1
cos 1
sin
1
cos 1
sin

m
(
(

|
.
|

\
|
+
|
.
|

\
|

1 1
cos 1
sin
cos 1
sin


= m
( )
( )
(
(
(
(
(

|
|
.
|

\
|
+
+
|
|
.
|

\
|
+

1
cos 1
sin
0
0 1
cos 1
sin
2
2
2
2

=[ ] M - a diagonal matrix.

Note that although there is only one particle there are two mass terms, each referring to the
inertial resistance of the particle against displacements in two directions. We will now look at
an interesting example where the vibration of a single mass in two orthogonal directions is
affected by different system properties.

Lateral and axial vibration of a spring-mass system

Consider the vibration of the spring mass system shown in Figure 3.3.4. The springs have
stiffness coefficients k
1
,k
2
and are under a static tensile force T
S
. Let us say that the mass and
springs lie on a smooth horizontal table, and that we are interested in the natural frequencies
and modes of this system when it vibrates in the horizontal plane.
S. Ilanko 2007
66

Figure 3.3.54. Spring Mass System Under Static Force


Let the static tension in the springs be T
S
and the dynamic tensions be T
D1
, T
D2
. Now,
Newtons second law may be applied in u, v directions:
In the u direction,
( ) ( ) u m T T T T
D S D S
= + +
1 1 2 2
cos cos
For small displacements, cos
1
, cos
2
1, giving u m T T
D D
=
1 2

Using the spring properties, this becomes u m e k e k =
1 1 2 2

where e
1
,e
2
are the elongation of the springs 1,2 respectively. For small displacements, it can
be shown that e
2
= -u and e
1
= u. Equation of motion thus reduces to:
( ) u m u k k = +
2 1

But u u
2
=
Therefore ((k
1
+k
2
) -m
2
)u = 0.
The non-trivial solution of this frequency equation is
m
k k
2 1
+
= (3.3.15)

It is interesting to note that this natural frequency does not depend on T
D
and corresponds to a
non-trivial solution of u. This means that there is a natural mode of vibration that is entirely
axial, and the corresponding frequency is independent of the static force in the axial direction.
Now let us obtain the equation of motion in the v direction.

T
S

T
S

T
S
+T
D1

T
S
+T
D2


T
S
+T
D1

T
S
+T
D2

Freebody Diagram
m

2

L
1
L
2

k
2

k
1

At equilibrium state
During vibration


v
v
S. Ilanko 2007
67
( ) ( ) v m T T T T
D S D S
= + +
2 2 1 1
sin sin
For small displacements, sin
1
v/L
1
and sin
2
v/L
2

Also the dynamic forces T
D1
,T
D2
are negligible compared to T
S
. Using the above into the
equation of motion gives:
v m
dt
v d
m
L L
v T
S
2
2
2
2 1
1 1
= =
|
|
.
|

\
|
+
For non-trivial solution of v,
m
L L
T
S
|
|
.
|

\
|
+
=
2 1
1 1
..(3.3.16)

This is a pure lateral mode, and its frequency is not dependent on the spring properties.
However, it is a function of the axial force (geometric stiffness) and will be zero in the
absence of any axial tension. The frequency increases with axial tension. This is true for all
mechanical systems. That is, the natural frequency in lateral vibration depends on the axial
force, and increases with tension. A good illustration of this is the well-known fact that the
pitch of string instruments increases with tension.

It is not possible to give a geometric interpretation of the orthogonality principle in general.
The main application of the orthogonality property of the modes is in response calculations.

3.4 General procedure for finding the transient response of an undamped system

Having established the orthogonality relationship, we can now set up a general procedure for
solving an initial value problem. Let us suppose, we have the frequencies and modes of an n
degree of freedom system that is subject to prescribed initial conditions. i.e. { } u and { } u at t
= 0 are known.

Due to the orthogonality of modes with respect to mass,
0 =

n
i
s , i r , i i
a a m for r s (3.4.1)
where, n is the number of degrees of freedom.

S. Ilanko 2007
68
This is to say that the r
th
mode and s
th
mode are orthogonal with respect to mass. The modes
may be arranged in a matrix called modal matrix, which gives the j
th
mode in the j
th
column
and the rows correspond to mass (the term 'mass' is used here generically to include mass and
moment of inertia) numbers. The modal matrix may be denoted by [a], and it is not unique
for any given system.

For the two degree of freedom system considered in this chapter, any of the following may be
used as the modal matrix, as long as the relative ratios of the displacement (translation or
rotation) of various masses for a given mode remain unchanged thus defining a unique shape
or pattern for each mode.

(

1 4
2 3
or
1 4
2 3
or
5 . 0 1
1 75 . 0
could be used as the modal matrix.

In terms of the modal matrix, the orthogonality property leads to a matrix manipulation
technique which can be used to simplify vibration analysis.

Recalling from equation (3.3.1b) the diagonalised inertia matrix ] [M is obtained from

[a]
T
[M][a] = ] [M (3.3.1b)

The r
th
diagonal element of ] [M is given by equation (3.3.7)
2
, ,
) (
r i
n
i
i r r
a m M

= (3.3.7b)

Solving an initial Value Problem:

If an undamped vibratory system is given an initial displacement or velocity in the form of
one of its natural modes and then released, it will continue to vibrate in that particular mode.
For any other initial condition, its response will consist of more than one mode. The
contribution from each mode to the vibration of the system may be expressed in the following
form:

S. Ilanko 2007
69
(3.4.2)


where {u} gives the dynamic displacement, [a] is the modal matrix and {p} is a vector
consisting of the contribution from each mode in the form of a scaled time function. These
contribution factors {p} are called the principal co-ordinates.
i.e. p
r
= A
r
sin
r
t + B
r
cos
r
t (3.4.3a)
We can also obtain the time derivative

r
p =
r
A
r
cos
r
t -
r
B
r
sin
r
t (3.4.3b)

Use of initial conditions:

Note that at t = 0, p
r
= B
r
(3.4.3c)
and
r
p =
r
A
r
(3.4.3d)
The actual displacement of the i
th
mass would be given by
u
i
=

=
l
r
r r i
p a
1
,
(3.4.4)
where l is the number of modes.
Pre-multiplying both sides of equation (3.4.2) by [a]
T
[M] gives:
[a]
T
[M]{u} = [a]
T
[M][a] {p} (3.4.5)

Substituting equation (3.3.1b) into this gives:

(3.4.6)


Since ] [M is diagonal, these equations are decoupled, and for any given initial conditions,
{p} may be easily determined as follows:

If the original mass matrix is also diagonal, then expanding the r
th
row of equation (3.4.6), and
substituting equation (3.3.7b) into the resulting equation, we get,

[a]
T
[M]{u} = ] [M {p}
{u} = [a]{p}
S. Ilanko 2007
70
=

=
n
i
i r i i
u a m
1
,
p
r
2
) a ( m
r , i
n
i
i
(3.4.7)
Therefore p
r
= /
1
,
|
.
|

\
|

=
n
i
i r i i
u a m |
.
|

\
|

2
) a ( m
r , i
n
i
i
(3.4.8)

Since only the initial values of u
i
would be known, the components of p
r
(A
r
and B
r
) need to
be determined using the initial conditions and equations (3.4.3c) and (3.4.3d). This gives:


(3.4.9a)

(3.4.9b)


If the original mass matrix contains off-diagonal elements then the matrix equation (3.4.6)
must be used to obtain the principal coordinates.

Once the principal co-ordinates are found, the displacement is given by equation (3.4.2). So
equations (3.4.9a), (3.4.9b), (3.4.3a) and (3.4.2) give the response of a general multi-degree of
freedom system subject to any initial conditions, without requiring any simultaneous
equations solver. Another advantage of this approach is that one does not need the knowledge
of all modes of a system to estimate the response. In practice, the use of the first few modes
(equation (3.4.2)) may be sufficient. That is one could use a narrow rectangular matrix
consisting of as many modes as necessary to reach the desired accuracy. However, all masses
must be included to get correct results. (i.e. the number of modes l may be smaller than the
degree of freedom n
f
, but the number of masses (n) must be at least equal to n
f
, and in some
cases it may exceed n
f
. The latter occurs when there are several masses whose relative motion
is constrained for example as in gears. Using a different modal matrix would give different
{p} but the final response will be the same.

Example 1:
Let us solve the example in Section 3.2 using the general approach.

B
r
= / ) 0 (
1
,
|
.
|

\
|

=
n
i
i r i i
u a m |
.
|

\
|

2
) a ( m
r , i
n
i
i

A
r
= / ) 0 (
1
,
|
.
|

\
|

=
n
i
i r i i
u a m |
.
|

\
|

2
) a ( m
r , i
n
i
i r


S. Ilanko 2007
71
Let [a] =
(

5 . 0 1
1 75 . 0

[M] =
(

3 . 0 0
0 2 . 0

Firs we can express initial conditions in terms of principal coordinates.
{u} = [a]{p}
[a]
T
[M]{u} = [a]
T
[M][a]{p}= ] [M {p}

(

5 . 0 1
1 75 . 0
(

3 . 0 0
0 2 . 0
{u}=
(

5 . 0 1
1 75 . 0
(

3 . 0 0
0 2 . 0
(

5 . 0 1
1 75 . 0
{p}

(

15 . 0 2 . 0
3 . 0 15 . 0
{u}=
(

275 . 0 0
0 4125 . 0
{p}

This is true at any time. Therefore at t=0,

(

15 . 0 2 . 0
3 . 0 15 . 0
{u(0)} =
(

275 . 0 0
0 4125 . 0
{p(0)}
But we have {u(0)} =
)
`

5
5
mm
Substituting this into the previous equation gives:
)
`

275 . 0
4125 . 0
{p(0)}=
)
`

25 . 0
25 . 2
mm giving
)
`

=
909 . 0
45 . 5
)} 0 ( {p
Similarly,
(

15 . 0 2 . 0
3 . 0 15 . 0
{ ) 0 ( u } =
(

275 . 0 0
0 4125 . 0
{ ) 0 ( p }
and { ) 0 ( u } ={ } 0
gives { ) 0 ( p }={ } 0
From equations (3.4.3c,d) we get
)
`

= =
909 . 0
45 . 5
)} 0 ( { } { p B mm and { } 0 }
) 0 (
{ } { = =

p
A


The frequencies are given in equations (3.1.5,6).

S. Ilanko 2007
72
} ]{ [ } { p a u = =
(

5 . 0 1
1 75 . 0
)
`

) 72 . 44 cos( 909 . 0
) 91 . 12 cos( 45 . 5
t
t
mm

=
)
`

+
) 72 . 44 ( cos 45 . 0 ) 91 . 12 ( cos .45 5
) 72 . 44 ( cos 91 . 0 ) 91 . 12 ( cos 09 . 4
t t
t t
mm

These results could have been obtained by substituting the elements of the modal matrix and
mass matrix into equations (3.4.9a,b), and then using equation (3.4.2), but we have gone
through the problem step-by-step. Using a normal mode [N] instead of [a] would have given
the same result for the response.

[N]
T
[M]{u} = [N]
T
[M][N]{p}=[I]{p}

{p}=[N]
T
[M]{u}

{p(0)}= [N]
T
[M] {u(0)}
From equation (3.3.4) [N] =
(

0.953 1.557
1.907 1.168


{p(0)}=

(

0.953 1.907
1.557 1.168
(

3 . 0 0
0 2 . 0
)
`

5
5
=
(

2859 . 0 3814 . 0
4671 . 0 2336 . 0
)
`

5
5
=
)
`

4775 . 0
5035 . 3

Since { ) 0 ( u } ={ } 0 , { ) 0 ( p }={ } 0 giving { } 0 }
) 0 (
{ } { = =

p
A


= = )} 0 ( { } { p B
)
`

4775 . 0
5035 . 3
mm
The principal coordinates are different now, but we will get the same displacement for the
masses.
} ]{ [ } { q P u = =
(

0.953 1.557
1.907 1.168
)
`

) 72 . 44 cos( .4775 0
) 91 . 12 cos( .5035 3
t
t
mm
=
)
`

+
) 72 . 44 ( cos 45 . 0 ) 91 . 12 ( cos .45 5
) 72 . 44 ( cos 91 . 0 ) 91 . 12 ( cos 09 . 4
t t
t t
mm

S. Ilanko 2007
73
Compared to the way in which this was solved in Section 3.2, we should note that using the
modal matrix and principal coordinates, we did not have to invert a matrix nor solve a set of
simultaneous equations. The use of orthogonality relationship has enabled us to solve the
problem using matrix multiplications and additions only.
Example 2
The response of a four dof undamped spring-mass system with the following properties is
required.
Mass
(kg)
Mode Number Mode 1 Mode 2 Mode 3 Mode 4
Nat. frequencies (rad/s) 0.9031 2.6312 4.8944 6.9064
Displacements
1.0 u
1
0.2390 0.5342 1.0000 1.0000
1.2 u
2
0.5445 1.0000 0.7364 -0.8466
2.5 u
3
0.7470 0.9340 -0.5198 0.1913
3.1 u
4
1.0000 -0.8148 0.0809 -0.0138

Initially the system is given the following displacements and then released:
Show that the dynamic displacement u
1
is given by:
u
1
= 0.119 cos .9031t - 0.020 cos 2.6312t + 0.002 cos 4.8944 t + 0.003 cos 6.9064 t mm

Solution: See Transient response.ppt

3.5 Free vibration analysis of damped systems

The analysis will be confined to viscous damping only. The differential equation of motion
will be of the form:

(3.5.1)
[ ]{ } [ ]{ } [ ]{ } { } 0 = + + u K u C u M

53 . 0
34 . 0
23 . 0
1 . 0
) 0 (
) 0 (
) 0 (
) 0 (
4
3
2
1
u
u
u
u
S. Ilanko 2007
74
[ ] [ ] [ ] K M C + =
This may be obtained by applying Newtons law of motion or using an energy method (see
Chapter 6).
In terms of principal coordinates,
{ } [ ]{ } p a u = (3.5.2a)
Therefore, { } [ ]{ } p a u = and { } [ ]{ } p a u = (3.5.2b, c)

Substituting these into equation (3.5.1) and pre-multiplying each term by [ ]
T
a gives:

[ ][ ]{ } [ ][ ]{ } [ ][ ]{ } { } 0 ] [ ] [ ] [ = + + p a K a p a C a p a M a
T T T
(3.5.3)

Usually, [ ][ ] a C a
T
] [ is not a diagonal matrix, and the above equations are therefore coupled. A
practical approach to overcome this problem is to introduce the concept of proportional
damping in which the damping matrix [ ] C is assumed proportional to either the mass matrix
or the stiffness matrix, or a linear combination of the two.

i.e. Let (3.5.4)

Then [ ][ ] a C a
T
] [ = [ ][ ] [ ][ ] a K a a M a
T T
] [ ] [ + = [ ]
r r r
M
,
2
) ( + (3.5.5)

Here we are using the notation that [ ]
r r r
M
,
2
) ( + is a diagonal matrix whose r
th
diagonal
element is given by (
r r r
M
,
2
) +
Substituting equation (3.5.5) into equation (3.5.1), we get the following decoupled equations:

[ ]{ }+ p M
r r

,
[ ]{ } [ ]{ } { } 0 2
,
2
,
= + p M p M
r r r r r r r
(3.5.6)
where ) ( 2
2
r r r
+ = (3.5.7)

This gives the uncoupled equations

(3.5.8)


0 2
2
= + +
r r r r r r
p p p , r = 1,2,3l,

S. Ilanko 2007
75
which are of the same form as the damped, single degree of freedom system for which the
solution is
(3.5.9)

{ } u may be found using equation (3.5.2a)

Usually it is no significance whether modal damping
r r r r
M
,
2 is based upon [ ] C being
proportional to [ ] M or [ ] K , or a combination of both. From a practical point of view,
r
and
r
are interpreted as properties inherent in the system. The frequency
r
can be
calculated or measured experimentally, and
r
is usually obtained by experiment or assumed
from experience. A typical value for mechanical or civil structures is
r
= 0.05 or less.

3.6 Forced Vibration

Response of a multi-DOF system due to an excitation (dynamic force/s) involves the solution
of a system of simultaneous non-homogeneous differential equations. Let us first study a
simple example and then develop a general procedure.
Example: A 2 DOF undamped system subject to simple harmonic excitation

Figure 3.6.1 Forced Vibration of a 2 D.O.F. System

d
2
u
1
/dt
2

F
1

State 1 (equilibrium, springs unstressed)
State 2 (at time t during vibration)
T
1

T
2

T
2

u
1

u
2

m
2

k
1

k
2

m
1

F
1
+ T
2

T
1

m
1

d
2
u
2
/dt
2

T
2

m
2

F
2

F
2

( ) t B t A e p
r r r r r r
t
r
r r
2 2
1 sin 1 cos

+ =


S. Ilanko 2007
76
The system shown in Figure 3.6.1 is the same spring-mass system in Figure 3.1.1, but is
subjected to two dynamic forces F
1
, F
2
. Application of Newtons second law to the masses
yields the following equations, which are similar to equations (3.1.1) and (3.1.2) except for
the addition of the forces F
1
and F
2
.

Applying Newtons second law to mass m
1
,
1 1 1 2 1
u m T T F = +

which after rearranging gives:
1 1 1 1 2
F u m T T = + +
Using the constitutive relationships we get:
( )
1 1 1 2 2 1 2 1
F u m u k u k k = + + ...(3.6.1)

Similarly for the second mass we get,
2 2 2 2 2 1 2
F u m u k u k = + ...(3.6.2)

These equations of motion may be given in the following matrix form,

[ ]{ } [ ]{ } { } F u M u K = +

...(3.6.3)

The solution for {u} consists of a particular integral {u
p
} and a complimentary function {u
c
}.

ie. {u} = {u
p
} + {u
c
} ...(3.6.4)

the complimentary function is the solution to the free vibration problem, and depends on
initial conditions. It can be found as described earlier. The particular integral is a function of
the applied force vector {F}. Inclusion of damping (see next section) leads to a
complimentary function that decays with time, and this function is then called the transient
response. In all practical problems, there will be some damping and with time the response
becomes dominated by the particular integral. For this reason it is called the steady state
response. The remainder of this section deals only with the steady state response. Therefore
for convenience we will use {u} instead of {u
p
}

General steady state response due to harmonic excitation

S. Ilanko 2007
77
If the applied forces are simple harmonic then the following simple method gives the
response.
Let {F} = {Q } sin(t+) ..(3.6.5)
where is the frequency of excitation.
Then the particular integral will take the form {u} = {} sin (t+) ..(3.6.6)
Substituting the above equations into equation (3.6.3) and cancelling sin (t+) from both
sides give:
[K]{}-
2
[M]{} = {Q}

...(3.6.7)
Since p and {Q} are known {
p
} may be calculated by solving the following equation:
[K
d
] {} = {Q}

...(3.6.8)
where [K
d
] is a dynamic stiffness matrix given by [K
d
] = [K] -
2
[M] ...(3.6.9)

Let us consider the variation of the amplitudes of vibration with frequency. This is called the
response. For the above example let the excitation forces be given by:
{ }
)
`


=
0
N ) ( cos

t F
F
[ ]
(

+
=
2
2 2 2
2
2
1 2 1
-m k -k
-k m k k
K
d

So we now need to solve
)
`

=
)
`

+
0

2
1
2
2 2 2
2
2
1 2 1
F
u
u
-m k -k
-k m k k

By elimination or substitution we get: F
K
m k
u
d

) (

2
2 2
1

=
and F
K
k
u
d

) (

2
2
=
The results for this problem are given in Figure 3.6.2. in a non-dimensional form. The
displacement amplitudes are non-dimensionalised by dividing them by the static deflection
corresponding to = 0.

It is important to note that replacing in [K
d
] to would make |K
d
| =0 the frequency
equation for the 2-DOF system. This means that |K
d
| =0 if
1
= or
2
= . This implies that
the response of both masses would tend to infinity as the frequency of the forcing function
S. Ilanko 2007
78
approaches any one of its two natural frequencies, as to be expected since the system is
undamped. What is more interesting to observe is that if
2
2 2
m k = then
1
=0. This means
that if the excitation force on m
1
had a frequency equal to the natural frequency of a single
DOF spring-mass system consisting of k
2
and m
2
, then the first mass would be stationary. This
is the basis of the undamped vibration absorber discussed earlier. We will discuss this in
more detail at another section, but let us turn our attention to the problem of solving the
equations of motion. Although we were able to find the solution by elimination or
substitution, we can use a more convenient procedure that only requires matrix additions and
multiplications as has been done for the free vibration problem.
We would consider the effect of one harmonic excitation at a time and finally superimpose
the results. For a simple harmonic excitation of the form {F sin (t+)} we can express the
displacement vector {p} in the following form.
{p} = {b sin(t+)}={ p }sin(t+) ...........(3.6.10)
Then {} = [a]{ q }and { p }= [a]
-1
{} ...........(3.6.11a,b)
Substituting this into equation (3.6.7) gives: [K][a]{ p } -
2
[M][a]{ p } = {Q} ...(3.6.12)
Multiplying by [a]
T
gives: [a]
T
[K][a]{ p } -
2
[a]
T
[M][a]{ p }=[a]
T
{Q} ......(3.6.13)
Due to orthogonality, the product terms [a]
T
[K][a] and [a]
T
[M][a] are both diagonal matrices.
This means the solution to equation (3.6.13) may be obtained without any further matrix
algebra. We must note here that the modal matrix is not unique and can be scaled. To make
the analysis more convenient, let us consider scaling the modal matrix in the following way.
For the free vibrational analysis, {Q} = {0}, and = results in
[a]
T
[K][a]{ p } -
2
[a]
T
[M][ a]{ p } = {0} (3.6.14)

Since the natural modes are orthogonal with respect to the mass matrix, the modes may be
normalized by setting [N]
T
[M][ N] = [I]. (3.6.15)
Putting this into equation (3.6.14) gives:
[N]
T
[K][ N]{ q } = [ K ], (3.6.16)
where [ K ] is a diagonal matrix containing the square of natural frequencies, i.e.
i i
K
,
=
i
2

Substituting equations (3.6.15) and (3.6.16) into equation (3.6.13) gives:
[ K ]{ p }-
2
[I]{ p }= [a]
T
{Q}
These equations are decoupled, and elements of { p } are therefore given by:
S. Ilanko 2007
79

( )
2 2
... 2 , 1
j,i


=

=
i
n j
j
i
Q a
p

(3.6.17)
{} can then be found from equation (3.6.11)
In our example, Q
1
= ) sin(

t F and Q
2
=0
Therefore
( )
2 2
1,i
) sin(


=
i
i
t F a
p


2 2 , 1 1 1 , 1 1
p a p a u + = = F
a a

) ( ) (
2 2
2
2
2 , 1
2 2
1
2
1 , 1
|
|
.
|

\
|
|
|
.
|

\
|

+
|
|
.
|

\
|


S. Ilanko 2007
80
3.7 Forced Damped Vibration

In general a discrete system may possess damping properties and may be subject to any set of
dynamic forces. Assuming the damping to be proportional, we will now reduce the problem to
a set of decoupled differential equations which may be solved in a way similar to that of a
single DOF system. Once the solution to each decoupled equation is found, the system
response may be found by superposition.

The equation of motion is:

(3.7.1)


Substituting { } [ ]{ } p a u = into the above equation and premultiplying by [a]
T
gives

{ } { } { } { } F a p a K a p a C a p a M a
T T T T
] [ ] ][ [ ] [ ] ][ [ ] [ ] ][ [ ] [ = + +

Going through the same algebraic manipulations as in Section 3.4, the following decoupled
equations is obtained:

[ ]{ }+ p M
r r

,
[ ]{ } [ ]{ } { } F a p M p M
r r r r r r r
T
,
2
,
] [ 2 = +

where [ M ] is a diagonal matrix.

The r
th
equation of this matrix is:


(3.7.2)


If the modal matrix is normalised by setting the diagonal terms of [ ] M to unity, then equation
(3.7.2) reduces to:

{ } { } { } { } F u K u C u M = + + ] [ ] [ ] [

r r
n
i
i r i
r r r r r r
M
F a
p p p
,
1
,
2
2

=
= + +
S. Ilanko 2007
81
(3.7.3)



p
r
may be found by solving this decoupled equation in the same way as a single DOF system
and once p
r
are determined, u may be found by using
{u} = [a]{p} (3.7.4)

This may also be written as


(3.7.5)



These derivations are applicable, even for continuous systems if they are numerically
discretised as explained in Chapter 6.

It is worth noting that in finding the actual response it is not always possible, nor necessary to
consider all modes. For example, in a 100 DOF structural system (n=100), if it is possible to
get a good estimate of the response by considering only the first 10 terms. Hence the
summation in equation (3.6.22) may be terminated after 10 terms instead of 100 terms.

=
= + +
n
i
i r i r r r r r r
F a p p p
1
,
2
2

=
=
n
r
r r j k
p a u
1
,

S. Ilanko 2007
82
3.8. Lagranges Equations of Motion

Application of Newtons second law in finding equations of motion often requires careful
handling of the sign convention for forces, displacements etc. A more convenient technique,
particularly useful in complicated systems, is often used to obtain equations of motion based
on scalar energy terms. This is due to Lagrange, and the general form of the equation is:

k
k k k k
Q
q
V
q
D
q
T
)
q
T
(
dt
d
=



Where T is the kinetic energy of the system, which usually takes the form
V is the potential energy, which normally takes the form: .......
2
1
2
i i
i
x k V

=
D is the dissipation function due to factors such as damping and is of the form,
2
2
1
i i
i
x c D

=
Q
k
are the generalised forces corresponding to the generalised co-ordinates q
k
, and t is the
time variable, k positive integer which is less than or equal to the number of degrees of
freedom n (1 k n). The generalised co-ordinates q
k
must be independent. It should be
noted here, that unlike in the Rayleighs method, V, T, etc. are not the maximum values of
potential, kinetic energies. They are time dependent functions.

Derivation of Lagranges Equation of Motion

According to the principle of virtual work, for a system of masses that is in static equilibrium,
0 =
i
i
i
r F ,
where
i
F is the net force corresponding to the i
th
mass, and
i
r is a small virtual displacement
along the position vector r
k
that is compatible with the system constraints.

dAlemberts Principle:
From Newtons second law of motion, for a system vibrating freely,
. r m F
i i i
0 =
2 2
2
1
2
1
i i
i
i i
i
I T ..... or ......... x m T


= =
S. Ilanko 2007
83
If the system is subject to forces of magnitude
i i
r m then it would be in equilibrium, and the
resulting displacements would be equal in magnitude to the vibratory displacements. This
means we may treat the vibratory system as an equivalent static problem, by adding negative
inertia forces. This is dAlemberts principle. This enables us to obtain an equivalent
statement of virtual work theorem for a vibratory system in the following form:
( )

=
i
i i i i
. r r m F 0 (3.9.1)
Let r
i
be described by a set of independent generalised co-ordinates q
k
.
r
i
= r
i
( q
1,
q
2,
q
3, .
q
n
)
Taking its first variation,
k
k k
i
i
q
q
r
r

= ..(3.8.2)
Substituting this into equation (3.8.1) gives:
( )

=


i
k
i k
i
i i i
. q
q
r
r m F 0
Since the generalised co-ordinates q
k
are independent,
( )


i k
i
i i i
.
q
r
r m F 0 .(3.8.3)
First let us transform the terms associated with mass as a function of the kinetic energy.
Dividing both sides of equation (3.8.2) by t and taking the limit ast 0 gives:

=
k
k
k
i
i
q
q
r
r (3.8.4)
Differentiating the above equation with respect to
k
q gives:
k
i
k
i
q
r
q
r

.(3.8.5)
Therefore,
|
|
.
|

\
|

=
|
|
.
|

\
|

k
i
k
i
q
r
dt
d
q
r
dt
d


j
j k
i
j
q
q
r
q

|
|
.
|

\
|

= (using the chain rule)



j
j j
k
i
q
q
r
q

|
|
.
|

\
|

= (interchanging the order of operations)


k
i
i
k
q
r
r
q

=

(using equation (3.8.4))
S. Ilanko 2007
84
i.e.
|
|
.
|

\
|

=
|
|
.
|

\
|

k
i
k
i
q
r
q
r
dt
d

..(3.8.6)
Therefore,
|
|
.
|

\
|

=
|
|
.
|

\
|

k
i
i
k
i
i
q
r
r
q
r
dt
d
r


Multiplying the equation by m
i
and taking all terms to the left-hand side and adding
k
i
i i
q
r
r m

gives:
k
i
i i
k
i
i i
k
i
i i
k
i
i i
q
r
r m
q
r
r m
q
r
dt
d
r m
q
r
r m

=
|
|
.
|

\
|


|
|
.
|

\
|


Summing over i results in:

=
|
|
.
|

\
|


|
|
.
|

\
|
|
|
.
|

\
|

i k
i
i i
i k
i
i i
i k
i
i i
k
i
i i
q
r
r m
q
r
r m
q
r
dt
d
r m
q
r
r m

(3.8.7)
The kinetic energy is of the form:

|
.
|

\
|
=
i
i i
r m T
2
2
1
(3.8.8)
Therefore
k
i
i
i
i
k
q
r
r m
q
T


Differentiating this with respect to t gives:
|
|
.
|

\
|

=
|
|
.
|

\
|

k
i
i
i
i
k
q
r
r
dt
d
m
q
T
dt
d


|
|
.
|

\
|
|
|
.
|

\
|

=

k
i
i
k
i
i
i
i
q
r
dt
d
r
q
r
r m

(3.8.8a)
Also from equation (3.8.8)

i k
i
i i
k
q
r
r m
q
T
(3.8.8b)
Substituting equations (3.8.8a) and (3.8.8b) into equation (3.8.7) gives:

|
|
.
|

\
|

i k
i
i i
k k
q
r
r m
q
T
q
T
dt
d


and substituting equation (3.8.5) for the last term on the right-hand side of the above equation
gives:

|
|
.
|

\
|

i k
i
i i
k k
q
r
r m
q
T
q
T
dt
d

...(3.8.9)

S. Ilanko 2007
85
This forms a part of the virtual work equation (3.8.3). The remaining terms

i k
i
i
q
r
F come
from the work done by restoring forces, frictional forces and any external forces and it may be
shown that
|
|
.
|

\
|

k k i k
i
i
q
D
q
V
q
r
F

+Q
k
..(3.8.10)
Substituting equations (3.8.9) and (3.8.10) into equation (3.8.3) gives:
k
k k k k
Q
q
V
q
D
q
T
)
q
T
(
dt
d
=


..(3.8.11)

3.9 Application of Lagranges Equations of Motion:

3.9.1. A single degree of freedom system






This is a single degree of freedom system.
Generalised co-ordinate q
k
= x.
Generalised force Q
k
= F
0
sin(pt)
Kinetic Energy
2
2
1
x m T =
Potential Energy
2
2
1
kx V =
Dissipation function
2
2
1
x c D =
Using the above
kx
x
V
=


x m
x
T


( ) x m x m
dt
d
x
T
dt
d

= =

) (
F
0
sin(pt)
k
c
m
x
S. Ilanko 2007
86
0 =

x
T

x c
x
D



Substituting the above equations into Lagranges formula yields the following equation of
motion:
) sin(
0
pt F x m x c kx = + +
For free vibration, F
0
= 0 which gives
0 = + + x m x c kx
And for undamped free vibration,
0 = + x m kx

3.9.2. Simple Pendulum




















(g/l) solution, trivial - non For
0 l - Therefore,
- motion, harmonic simple For
0
0
0 ) (
formula, s Lagrange' Using
Therefore
sin , 0 as , vibration amplitude small For
) (sin
) cos 1 (
0
0
) (
2
1
2
) (
2
1
, 1
system. freedom of degree single a is This
2
2
2
2
2
2
1
=
= +
=
= +
= +
=

=
=
=

=
= =

g
g l
mgl ml
V D T T
dt
d
mgl
V
mgl
V
mgl V
D
T
ml
T
dt
d
ml
T
l m T
q n



l

m
mg
Length = l
S. Ilanko 2007
87
3.9.3. Two Degree of Freedom System





This is a two degree of freedom system. i.e. n = 2
Let q
1
= x
1
and q
2
= x
2
Since there are no forces, Q
1
= Q
2
= 0.
( )
1 2
2
2 1 2 1 2 2 1 2 1 1
2
2
2 2
2
1
1 1
1
1 1
1
2
1 2 2
2
1 1
2 2
2
) (
0
) (
0
) (
0
) (
2
1
x k x k
x
V
x k x k k x k x k x k
x
V
x
T
x m
x
T
dt
d
x
T
x m
x
T
dt
d
x m
x
T
D
x x k x k V
=

+ = + =

=
+ =


Using Lagranges formula:
0 0 ) ( and
0 ) ( 0 ) (
2 2 2 2 1 2
2 2 2 2
1 1 2 2 1 2 1
1 1 1 1
= + + =

= + + =

x m x k x k
x
V
x
D
x
T
x
T
dt
d
x m x k x k k
x
V
x
D
x
T
x
T
dt
d





These are the equations of motion.

k
1

m
1

x
1

k
2

m
2

x
2

( )
2
2 2
2
1 1
2
2
1
2
1
x m x m x m T
i i
i
+ = =

S. Ilanko 2007
88
3.9.4. Torsional vibration of a Geared System

Consider the torsional vibration of
the geared shaft system shown in
the figure. It may be assumed
that the mass of the shafts is
negligible, and the system may be
treated as a discrete system.
Although there are four rotors,
since the rotation of the gears in contact are related by the gear ratio, there are only three
degrees of freedom. Therefore, in choosing the generalised co-ordinates, the rotation of only
one of the two gears should be taken. In this example, let us choose the rotations of rotors 1,2
and 4 as the generalised co-ordinates.
{q
k
} = {
1
,
2
,
4
}

3
is related to
2
since

r
2

2
= - r
3

3
or
3
2
2 3
where ,
r
r
= = (1)
The potential energy is given by:
2
3 4
2
2 2
1 2
1
1
) (
2
1
) (
2
1

|
|
.
|

\
|
+
|
|
.
|

\
|
=
L
GJ
L
GJ
V
Using equation (1) this may be rewritten
2
2 4
2
2 2
1 2
1
1
) (
2
1
) (
2
1
+
|
|
.
|

\
|
+
|
|
.
|

\
|
=
L
GJ
L
GJ
V .(2)
2
4 4
2
3 3
2
2 2
2
1 1
2
1
2
1
2
1
2
1


I I I I T + + + =
Substituting equation (1) into the above gives:
2
4 4
2
2
2
3
2
2 2
2
1 1
2
1
2
1
2
1
2
1


I I I I T + + + = ..(3)
From equation (2), we have,
) (
2 1
1
1
1


|
|
.
|

\
|
=

L
GJ V
(3a)

4
r
3
r
2
S. Ilanko 2007
89
) ( ) (
2 4
2
2
1 2
1
1
2

+
|
|
.
|

\
|
+
|
|
.
|

\
|
=

L
GJ
L
GJ V
(4a)
and ) (
2 4
2
2
4

+
|
|
.
|

\
|
=

L
GJ V
(5a)
From equation (3), we have,
( )
1 1 1 1
1

I I
dt
d T
dt
d
= =
|
|
.
|

\
|

..(3b)
( )
2 3
2
2 2
2
3 2 2
2
) (

I I I I
dt
d T
dt
d
+ = + =
|
|
.
|

\
|

(4b)
and ( )
4 4 4 4
4

I I
dt
d T
dt
d
= =
|
|
.
|

\
|

..(5b)
also, 0
4 2 1
=


T T T
(3c), (4c), (5c)
Substituting equations (3a), (3b) and (3c) into Lagranges general form of equation of motion
gives:
0 ) (
1 1 2 1
1
1
= +
|
|
.
|

\
|


I
L
GJ
..(3d)
Similarly, using equations (4a), (4b) and (4c) we can get
0 ) ( ) ( ) (
2 3
2
2 2 4
2
2
1 2
1
1
= + + +
|
|
.
|

\
|
+
|
|
.
|

\
|


I I
L
GJ
L
GJ
.(4d)
And from equations (5a,b,c) we have,
0 ) (
4 4 2 4
2
2
= + +
|
|
.
|

\
|


I
L
GJ
..(5d)
Equations (3d), (4d) and (5d) are the Lagranges equations of motion for the geared torsional
vibratory system. This problem may be used to illustrate the importance of ensuring that the
generalised co-ordinates are independent. If we had taken four generalised co-ordinates
1
,

2
,
3
, and
4
, then equation (4d) would have become
0 ) (
2 2 1 2
1
1
= +
|
|
.
|

\
|


I
L
GJ

which is incorrect. The best way to choose the correct generalised co-ordinates is to see if it
is possible to displace the system along each of the generalised co-ordinates without violating
the constraints of the system and its supports. For example, it is not possible to give a rotation
S. Ilanko 2007
90
of one of the gears without rotating the other, as the constraint equation (1) cannot be
violated. Therefore, only one of the gear rotations (
2
,
3
) may be regarded as a generalised
co-ordinate.

3.9.5. Compound Pendulum on a Roller



















The system shown here consists of a
compound pendulum hanging from a
moving platform which rests on two
rollers and is connected to a rigid wall
by means of a spring of stiffness k. It
has three degrees of freedom (n = 3).
It is convenient to take the
displacement of the platform (x) and
the rotations of the two strings (
1
)
and (
2
) as the three generalised co-
ordinates.

i.e Let q
1
= x, q
2
=
1
, q
3
=
2
For small amplitude oscillations,
2 2 2
) cos 1 ( ) cos 1 (
m of anslation verticaltr
2
) cos 1 (
m of on translati vertical
) ( m of velocity and
m of velocity
2
2 2
2
1 1
2
2 2
2
1 1
2 2 1 1
2
2
1 1
1 1
1
2 2 1 1 2
1 1 1

L L L L
L L
L
L
L L x
L x
+
= +
+ =

=
+ =
=


( )
( )
2
0 0
2
2 2
2
1 1 2
2
1 1 1
2
1 1 3 2 1
2
2 2
2
1 1
2 3 2
2
1 1
1 2 1
2
1 1 1
2
1
m of energy Kinetic
) 1 ....( ) (
2
1
energy potential Net
2
: m on gravity to due energy Potential
2
: m on gravity to due energy Potential
2
1
: spring the to due Energy Potential
x m
L L g m gL m x k V V V
L L
g m V
L
g m V
x k V
=
+ + + = + + =
+
=
=
=

k
x
L
1
L
2
m
1
m
2

2
g
Rollers have
mass m, radius
a and radius of
gyration r
m
0
S. Ilanko 2007
91
( )
( ) ) 2 ....( ) ( ) (
2
2
2
1
Energy Kinetic Net
) (
2
1
m mass of Energy Kinetic and
) (
2
1
m mass of Energy Kinetic
2 2
1
2 rollers both of energy Kinetic
2
2 2 1 1 2
2
1 1 1
2
2 2 2
0
2
2 2 1 1 2 2
2
1 1 1 1
2
2 2
|
|
.
|

\
|
+ +
|
|
.
|

\
|
|
.
|

\
|
+ + =
=
=
|
|
.
|

\
|
|
.
|

\
|
+ =

L L x m L x m
a
x
r a m x m
L L x m
L x m
a
x
r a m
From equation (1) we can obtain the following expressions, required in Lagranges formula:
2 2 2
2
1 1 2 1
1
) (

gL m
V
gL m m
V
kx
x
V
=

+ =


Similarly, from equation (2) we can obtain the following expressions:
motion. of equations s Lagrange' the are These
0
0 ) ( ) ( ) (
0 ) ( ) 1 (
2
: motion of equations following the gives formula s Lagrange' into s expression above the ng Substituti
0 have also We
) (
) ( ) ( ) (
) ( ) 1 (
4
) (
2
2
2 2 1 2 1 2 2 2 2 2 2
2 2 1 2 1
2
1 2 1 1 2 1 1 1 2 1
2 2 2 1 1 2 1 2 1 2
2
0
2 1
2
2
2 2 1 2 1 2 2 2
2
2 2 1 2 1
2
1 2 1 1 2 1
1
2 2 2 1 1 2 1 2 1 2
2
0
= + +
= + + + + +
= +
|
|
.
|

\
|
+ +
|
|
.
|
+ + +
=

+ + =

+ + + + =

+
|
|
.
|

\
|
+ +
|
|
.
|
+ + =

L m L L m x L m gL m
L L m L m m x L m m gL m m
L m L m m x m m
a
r m
m kx
T T
x
T
L m L L m x L m
T
dt
d
L L m L m m x L m m
T
dt
d
L m L m m x m m
a
r m
m
x
T
dt
d
S. Ilanko 2007
92
3.9.6. Two-Link Robot Mechanism














In the first five examples, the kinetic energy T, was a function of the time derivative of the
generalised co-ordinates, but it was independent of the generalised co-
ordinates ) 0 / ( =
k
dq dT . In this example the kinetic energy is a function of generalised co-
ordinates as well as their time derivatives.

Potential Energy due to gravity:
|
.
|

\
|
+ = cos
2
cos cos
2
2
1 2
1
1 1
L
L g m
L
g m V .(1)
Potential Energy due to the springs: ) 2 .( .......... ) (
2
1
2
1
2
2
2
1 1
+ = k k V
Net potential energy
2 1
V V V + = ....(3)
Kinetic Energy of the first arm: ,
2 2
1
2
1
2
1
1
2
1 1
|
.
|

\
|
+ =

L
m I T .(4)
Where
1
I is the moment of inertia of the first arm about its centroidal axis and is given by:
12
2
1 1
1
L m
I = (4a)
Similarly, kinetic energy of the second arm is:
The two arm robot mechanism shown in
the figure consists of two rigid arms having
masses m
1
, m
2
and lengths L
1
,L
2

respectively. They are hinged together at
B, and one arm is hinged to a rigid frame at
A. Arm AB is connected to the support by
a coil spring of stiffness k
1
, and the two
arms are also connected by a coil spring of
stiffness k
2
at the hinge B. The system is
under a uniform gravity field g. In this
problem we need to obtain the equations of
motion, for large amplitude rotations.
k
1
k
2
m
1
,L
1
m
2
,L
2


g
B

A

C

S. Ilanko 2007
93
|
|
.
|

\
|
|
|
.
|

\
|
|
.
|

\
|
+ +
|
|
.
|

\
|
|
.
|

\
|
+ + =
2
2
1
2
2
1 2
2
2 2
) sin(
2
) sin( ) cos(
2
) cos(
2
1
2
1

L
L
dt
d L
L
dt
d
m I T . (5)
where
12
2
2 2
2
L m
I = (5a)
After differentiating with respect to t, equation (5) becomes:
|
|
.
|

\
|
|
|
.
|

\
|
|
.
|

\
|
+ +
|
|
.
|

\
|
|
.
|

\
|
+ + =
2
2
1
2
2
1 2
2
2 2
) cos(
2
) cos( ( ) sin(
2
) sin( (
2
1
2
1

L
L
L
L m I T

Expanding this, and simplifying yields:
( ) ( )
|
|
.
|

\
|
+ |
.
|

\
|
+ + =

) cos( L L
2
L
L m
2
1
I
2
1
T
2 1
2
2
2
1 2
2
2 2
(5b)
Total kinetic energy is T = T
1
+ T
2
..(6)
Using equations (1), (2) and (3), we get

sin )
2
( ) (
sin sin
2
) (
1 2
1 1
2 2 1
1 2
1 1
2 2 1
g L m
L m
k k k
gL m
gL m
k k k
V
+ + + =
+ + + =


..(7a)
From equations (4), (5) and (6)
) sin(
2
1
2 1 2

L L m
T
.(7b), and

) cos(
2
1
)
4
(
2 1 2
2
1 2
2
1 1
1
+ + + =

L L m L m
L m
I
T
..(7c)
Differentiating equation (7c) with respect to t gives:

) cos(
2
1
)
4
(
2 1 2
2
1 2
2
1 1
1
+ + + = |
.
|

\
|

L L m L m
L m
I
T
dt
d

) )( sin(
2
1
2 1 2


+ L L m .(7d)
Substituting equations (7a)..(7d) into Lagranges formula yields:



) cos(
2
1
)
4
(
) sin(
2
1
sin )
2
( ) (
2 1 2
2
1 2
2
1 1
1
2 1 2 1 2
1 1
2 2 1
+ + + +
+ + + +
L L m L m
L m
I
L L m g L m
L m
k k k

S. Ilanko 2007
94
0 ) )( sin(
2
1
2 1 2
= +

L L m .(8)
Similarly, we can operate with respect to .
) ( sin
2
1
2 2 2

+ =

k gL m
V
(9a)
) sin(
2
1
2 1 2

L L m
T
(9b) and
|
.
|

\
|
+ + =

) cos(
2 2
1
2 2 2
2

L
L L m
I
T
(9c)
Differentiating equation (9c) with respect to t gives:
|
.
|

\
|
+ + + =
|
.
|

\
|

) )( sin( ) cos(
2 2
1 1
2 2 2
2

L L
L L m
I
T
dt
d
9(d)

Substituting equations (9a)..(9d) into Lagranges formula yields:

0 ) )( sin( ) cos(
2 2
) sin(
2
1
sin
2
1
) (
1 1
2 2 2
2
2 1 2 2 2 2
= |
.
|

\
|
+ + +
+

L L
L L m
I
L L m gL m k

.(10)
3.9.3.8. Radial Vibration of a Rotating Spring-Mass System

Another example in which the kinetic energy is a function
of a generalised co-ordinate is a rotating spring-mass
system. Consider the radial vibration of the spring-mass
system that is rotating at a constant angular speed of
radians per second, as shown in the figure. The length of
the unstretched spring is L. The extension of the spring
due to the centripetal acceleration during the steady state
rotation is u
s
. A further radial vibratory motion is denoted
by u.

The potential energy of the spring during vibration is given by:
L
u
s u
S. Ilanko 2007
95
2
) (
2
1
u u k V
s
+ = ,
and the kinetic energy of the mass is given by:
( ) ( )
2 2
) (
2
1
u u u L m T
s
+ + + =
( ) ( )
.......(2) 0 ) (
.....(1) 0 i.e.
h. must vanis equation the of parts two the , any for valid is equation above the Since
0 ) (
: get we , separately with associated terms the Grouping
0 ) ( ) (
equation, s Lagrange' Using
) (
) (
2
2
2 2
2
2
= +
= +
= + + +
= + + + + =

|
.
|

\
|

=
|
.
|

\
|

+ + =

+ =

s s
s s
s s
s
s
u L m ku
u m u m ku
u
u L m ku u m u m ku
u
u u L m u m u u k
u
T
u
T
dt
d
u
V
u m
u
T
dt
d
u u L m
u
T
u u k
u
V



Equation (1) is the Lagranges equation for the vibratory motion, and may be used to obtain
the natural frequency, while equation (1) gives the steady state radial displacement u
s
.
From equation (2),
|
.
|

\
|

=
1
2
m
k
L
u
s
(3)
And substituting the simple harmonic relationship u u
2
= into equation (1) gives:
0 ) (
2 2
= + mu ku .
For non-trivial solution,
|
|
.
|

\
|

|
.
|

\
|
=
2
m
k
..(4)
It should be noted here that (k/m) represents the square of the natural frequency of the non-
rotating system (
s
) and hence equation (4) may also be written:
) (
2 2
=
s
..(4a)
It may be noted that when


s
, the natural frequency of the rotating system would
approach zero, thus indicating a state of critical equilibrium. This is also evident from
equation (3) as the steady state displacement u
s
would also be infinite as
2


(k/m). This
type of instability occurs in shafts also, when the speed of rotation reaches the natural
S. Ilanko 2007
96
frequencies of the shaft in flexural vibration. This is called whirling of shafts, and the speeds
at which whirling occurs are referred to as the whirling speeds.

3.9.8 Two DOF System with Translation and Rotation


2
2
2
1
) )( 2 / ( ) )( 2 / (
G G G G
b y k a y k V + + =
( )
2 2
) 2 / 1 ( I y m T
G
+ =
2
2
2
1
) )( 2 / ( ) )( 2 / (
G G G G
b y c a y c D

+ + =
) ( ) (
2 1 G G G G
G
b y k a y k
y
V
+ + =

;
) ( ) (
2 1 G G G G
G
b y bk a y ak
V

+ + =


G
G
y m
y
T
dt
d

=
|
|
.
|

\
|

;
G
G
I
T
dt
d

=
|
|
.
|

\
|


) ( ) (
2 1 G G G G
G
b y c a y c
y
D

+ + =


Figure 3.9.8
G
y
k
2

k
1
m, I
G



G
a b
G
y
k
2

k
1
m, I
G


G
a b
c
2

c
2

c
1
c
1

S. Ilanko 2007
97
) ( ) (
2 1 G G G G
G
b y bc a y ac
D

+ + =


Equations of Motion:
0 ) ( ) ( ) ( ) (
2 1 2 1
= + + + + + +
G G G G G G G G G
y m b y c a y c b y k a y k


Rearranging gives:
0 ) ( ) ( ) ( ) (
1 2 2 1 1 2 2 1
= + + + + + +
G G G G G
y m a c b c y c c a k b k y k k

(1)
0 ) ( ) ( ) ( ) (
2 1 2 1
= + + + + +
G G G G G G G G G
I b y bc a y ac b y bk a y ak


Rearranging gives:
0 ) ( ) ( ) ( ) (
2
2
2
1 1 2
2
2
2
1 1 2
= + + + + + +
G G G G G
I b c a c y a c b c b k a k y a k b k

(2)
If the system is undamped, these reduce to:
0 ) ( ) (
1 2
2
2 1
= + +
G G
a k b k y m k k
0 ) ( ) (
2 2
2
2
1 1 2
= + +
G G
I b k a k y a k b k
This may be written as:
)
`

=
)
`

+
+
0
0
) ( ) (
) ( ) (
2 2
2
2
1 1 2
1 2
2
2 1
G
G
y
I b k a k a k b k
a k b k m k k


For N.T.S., 0
) ( ) (
) ( ) (
2 2
2
2
1 1 2
1 2
2
2 1
=
+
+

I b k a k a k b k
a k b k m k k

i.e. 0 ) ( ) )( (
2
1 2
2 2
2
2
1
2
2 1
= + + a k b k I b k a k m k k is the frequency equation the roots
of which give the first and second natural frequencies.
Note: If ) (
1 2
a k b k =0 then the system is decoupled!

S. Ilanko 2007
98
m

l

l/2


Figure 3.8.2.9
k
mg

c
3.9.9 Inverse Pendulum
cos
2 2 2
1
2
l
mg
l
k V +
|
.
|

\
|
=

2
2
2
2
2 2
6 2 12 2
1

ml l
m
ml
T =
|
|
.
|

\
|
|
|
.
|

\
|
+ =

2
2 2
1
|
.
|

\
|
=
l
c D

Applying Lagranges formula
0 ) ( =

k k k k
q
V
q
D
q
T
q
T
dt
d

,
the equation of motion is:
0 sin
2 4 4
0
3
2 2 2
= + +
mgl kl cl ml

For small amplitude motion, as sin this reduces to:
0
2 4 4 3
2 2 2
=
|
|
.
|

\
|
+ +
mgl kl cl ml

Note, if l mg k / 2 the system is unstable!

Substituting l mg k / 10 = and dividing by l
2
gives the following simplified equation.
0
2
4 3
= + +
l
mg c m

S. Ilanko 2007
99
From the formula supplied, l g
m
l mg
n
/ 6
) 3 / (
) / 2 (
= = = 2.45 rad/s
Damping ratio 031 . 0
) 3 /( 2 8
/ 2 . 0
) 2 /(
2
= = =
l g m
l g m
km c
2
031 . 0 1 6 =
d
= 2.45 rad/s
t
d d
n
e t B t A t



+ = ) cos sin ( ) (
20 20 ) 0 ( = = B radians
t
d d d
t
d d n
n n
e t B t A e t B t A t



+ + = ) sin cos ( ) cos sin ( ) (
d n
d n
B A
A B


/
0 ) ( ) ( 0 ) 0 (
=
= + =

= 0.62 radians
t
e t t t
076 . 0
) 45 . 2 cos 20 45 . 2 sin 62 . 0 ( ) (

+ =


S. Ilanko 2007
100
3.10 The Energy Terms in the Rayleigh-Ritz Method and Lagranges Equations

It is important to know the differences between the energy terms used in the Rayleigh-Ritz
method and the energy terms in Lagranges equations of motion. In the Rayleigh-Ritz
method, which is based on the principle of conservation of energy, the maximum kinetic
energy associated with vibration and the corresponding maximum total potential energy terms
are used. In Lagranges equations, the energy terms used are time dependent functions.

Another difference comes in the consideration of some potential energy terms due to external
forces. This may be illustrated by considering the static deflection of a spring mass system.
Static case may be treated as a special dynamic case with the applied force in the form Q
0

cos(pt) where the excitation frequency p = 0. Both Lagranges equations of motion and the
Rayleigh-Ritz procedure are applicable in static equilibrium problems too.

In this single degree of freedom system, the only generalised co-ordinate (q
k
) is x.
For Lagranges equation of motion, the potential energy may be taken as
V= (1/2) k x
2

and the kinetic energy
T= (1/2)m(dx/dt)
2
.
The generalised force Q
1
= {Q
0
cos(pt)}.
Applying Lagranges equation we get the following equation of motion:
) cos(
0
pt Q x m kx = +
For the steady state solution, we may use x=X cos pt which gives
kx-mp
2
x = Q
0
cos(pt)
And since p=0 this reduces to kX = Q
0
and X = Q
0
/k as expected.

In using the Rayleigh-Ritz procedure for static analysis, the total potential energy should
include the potential of the applied force.
V
m
= (1/2) k X
2
- Q
0
X
T
m
= 0
Rayleigh-Ritz equation is V
m
/X =0 gives kX-Q
0
=0.
Therefore X=Q
0
/k
S. Ilanko 2007
101
What is important here is that the total potential energy used in the Rayleigh-Ritz method
includes the potential of the applied load where as in Lagranges equation the applied load
comes in as a generalised force.

S. Ilanko 2007
102
m
b,
I
b
m
b,
I
b
Multi DOF Systems: Problems


1. A rigid rod of length l
2
and mass m is hinged to another
rigid but light (of negligible mass) rod of length l
1
which
is hinged to a fixed point O as shown in Figure 1. The
system is under gravity field g.

a) Derive the linearized equations of motion of this
system giving the mass and stiffness matrices.
b) If l
2
= 6 l
1
= 6 l, calculate the natural frequencies
and eigenvectors (modes) of the system.


Note: The moment of inertia of a rod of mass m and
length l, about a centroidal axis perpendicular to its
axis is given by
12
2
ml
I =

2. Two rigid bars are hinged at points O and P, and connected by means of an elastic spring of
stiffness k as shown in Figure 2.
They also carry four concentrated
masses at their ends. The bars are
identical having a mass m
b

(which is equal to 0.2m) and a
moment of inertia of I
b
(which is
0.2mL
2
/12) about their centroidal
axis perpendicular to the plane of
oscillation. This system is in
static equilibrium under gravity
field g.

Figure 1

2

l
1
l
2
, m
g

O
P
O
g
0.2L
0.3L
0.5L
0.2L
0.7L
0.1L
k
m
m
2 m
2.5 m
Figure 2
S. Ilanko 2007
103
3. (a) Derive the equations of motion for the
torsional vibration of the system in
Figure 3a, neglecting the mass of the
shafts (treating the system as discrete).
Do not solve the equations. Note r
3
=
r
2
.
(b) If
4 1
I I = =
3 2
I I = = I and
2 1
k k = = k,
write down the expressions for any
two natural frequencies of this system.
(c)
Would your answers to the above
questions be any different, if the
system had a geometric arrangement
as shown in Figure 3b.
4. Figure 4 is a schematic
representation of a
power transmission
system after the
sudden application of
a brake on the left end
of the top shaft.
The relevant moments
of inertia of the gears
and the rotor are:
I
2
= 1.2 kgm
2
;
I
3
= 0.15 kgm
2
; and
I
4
= 3 kgm
2

(a.) Obtain the equations of motion for the torsional vibration of the above system
neglecting the mass of the shafts (treating the system as discrete).
(b.) Show that the natural frequencies and modes of the above system are as follows:

1
= 18.34 rad/s, and for the first mode,
2
/
4
= 0.374

2
= 115.0 rad/s, and for the second mode,
2
/
4
= 4.458

1
r
3
= r
2
r
2

3
k
2
k
1
I
2
I
1
I
4
I
3
Figure 3a

2
r
2
k
1
I
2
I
1
r
3
= r
2

4
k
2
I
3
I
4
Figure 3b
I
2
k
1
=6 kNm/rad
I
3
I
4
k
2
= 4 kNm/rad
2r
r
Figure 4

4
(t)

2
(t)
S. Ilanko 2007
104
(c.) Obtain an expression for the subsequent torsional oscillation,
2
(t) of I
2
for the
following initial conditions:

2
(0) =
4
(0) = 0 and rad/s 120 - ) 0 ( and rad/s 60 ) 0 (
4 2
= =



5. The natural frequencies and some values of a modal matrix for the three-degree of freedom
spring-mass system shown in Figure 5 are given below.
Modal matrix:
(
(
(

1 1 ?
? 0 ?
1 1 1

The natural frequencies are:

82 . 20
00 . 10
0
rad/sec.

(a) Complete the missing values in the modal matrix.

(b) Obtain an expression for the displacement of the second mass as a function of time,
for the following initial conditions:
Displacement at t = 0, are

) 0 ( u
) 0 ( u
) 0 ( u
3
2
1
= mm
7
2
5

and
zero initial velocity. i.e.

) 0 ( u
) 0 ( u
) 0 ( u
3
2
1

0
0
0

(c) Comment on how these results could have been obtained more conveniently by
making use of symmetry.

6. The two degree of freedom system shown in Figure 6 has the
following natural frequencies and modal matrix:
=
1
5.41 rad/s, =
2
13.07 rad/s; [a] =
(


462 . 0 1913 . 0
1913 . 0 462 . 0

The first and second degrees of freedom correspond to
translations in x and y directions respectively.
(a) If the mass matrix [M] is
(

4 0
0 4
kg, find [ ] M .
(b) If the 4 kg mass is given an initial displacement of 5 mm in
the x direction, and then released, find an expression for the subsequent displacement
of the mass in the y direction. i.e. { }
)
`

=
0
5
) 0 ( u mm, and { } { } 0 ) 0 ( = u , ? ) ( ) (
2
= = t y t u

3 kg
5 kg

5 kg
u
1 u
2
u
3
Figure 5
Figure 6
m= 4 kg

y

k

k

x

45

S. Ilanko 2007
105
7. Derive the equations of
motion for the system
shown in Figure 7. This
consists of two rigid
bodies of mass m
0
,
connected by a spring of
stiffness k, and
supported on rollers
resting on a non-slip
surface. One of the
bodies is partly
restrained by a light
elastic cable of length L.
The cable is under a
tension T and at
equilibrium state it is
perpendicular to the
direction of motion of
the bodies. All
displacements may be assumed to be small.

8. In the 2-DOF oscillator shown in Figure 8, the spring is unextended for 0
2 1
= = . For
the small vibration studied here, the spring may be assumed to remain horizontal during
the motion. The stiffness coefficient of the
spring is k.


(a.) Give the expressions for the kinetic
energy T and the potential energy V in
terms of
2 1 2 1
, , , .
(b.) From these energy expressions, find
the elements of the mass and stiffness
matrices (M and K) describing the
linearised system.
(c.) For, m m m = =
2 1
and l l l = =
2 1
, find
the natural frequencies and the modes (intermediate steps may be skipped). Hint: You
may make use of the symmetry of the system.




k
L

T

m
0 m
0
x
2
Figure 7
All rollers have
mass m, radius
a and radius of
gyration r
m
1
1

l
1
m
2
l
2
s

2

Figure 8
k

Vous aimerez peut-être aussi