Mechanical Vibration

436-431 MECHANICS 4
UNIT 2
MECHANICAL VIBRATION
J.M. KRODKIEWSKI
2004
THE UNIVERSITY OF MELBOURNE
Department of Mechanical and Manufacturing Engineering
1
CONTENTS
0.1 INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
I MODELLING AND ANALYSIS 6
1 MECHANICAL VIBRATION OF ONE-DEGREE-OF-FREEDOM
LINEAR SYSTEMS 8
1.1 MODELLING OF ONE-DEGREE-OF-FREEDOM SYSTEM . . . . 8
1.1.1 Physical model . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.1.2 Mathematical model . . . . . . . . . . . . . . . . . . . . . 11
1.1.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.2 ANALYSIS OF ONE-DEGREE-OF-FREEDOM SYSTEM . . . . . . 27
1.2.1 Free vibration . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.2.2 Forced vibration . . . . . . . . . . . . . . . . . . . . . . . . 33
1.2.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2 MECHANICAL VIBRATIONOF MULTI-DEGREE-OF-FREEDOM
LINEAR SYSTEMS 63
2.1 MODELLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.1.1 Physical model . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.1.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
2.2 ANALYSIS OF MULTI-DEGREE-OF-FREEDOM SYSTEM . . . . . 87
2.2.1 General case . . . . . . . . . . . . . . . . . . . . . . . . . . 87
2.2.2 Modal analysis - case of small damping . . . . . . . . . . 96
2.2.3 Kinetic and potential energy functions - Dissipation
function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
2.2.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
2.3 ENGINEERING APPLICATIONS . . . . . . . . . . . . . . . . . . . 126
2.3.1 Balancing of rotors . . . . . . . . . . . . . . . . . . . . . . 126
2.3.2 Dynamic absorber of vibrations . . . . . . . . . . . . . . 132
3 VIBRATION OF CONTINUOUS SYSTEMS 137
3.1 MODELLING OF CONTINUOUS SYSTEMS . . . . . . . . . . . . . 137
3.1.1 Modelling of strings, rods and shafts . . . . . . . . . . . 137
3.1.2 Modelling of beams . . . . . . . . . . . . . . . . . . . . . . 141
CONTENTS 3
3.2 ANALYSIS OF CONTINUOUS SYSTEMS . . . . . . . . . . . . . . 143
3.2.1 Free vibration of strings, rods and shafts . . . . . . . . . 143
3.2.2 Free vibrations of beams . . . . . . . . . . . . . . . . . . . 149
3.2.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
4 APPROXIMATION OF THE CONTINUOUS SYSTEM BY DIS-
CRETE MODELS 174
4.1 DISCRETE MODEL OF THE FREE-FREE BEAMS . . . . . . . . . 175
4.1.1 Rigid Elements Method. . . . . . . . . . . . . . . . . . . . 175
4.1.2 Finite Elements Method. . . . . . . . . . . . . . . . . . . . 179
4.2 BOUNDARY CONDITIONS . . . . . . . . . . . . . . . . . . . . . . . 186
4.3 CONDENSATION OF THE DISCREET SYSTEMS . . . . . . . . . 187
4.3.1 Condensation of the inertia matrix. . . . . . . . . . . . . 188
4.3.2 Condensation of the damping matrix. . . . . . . . . . . . 189
4.3.3 Condensation of the stiness matrix. . . . . . . . . . . . 189
4.3.4 Condensation of the external forces. . . . . . . . . . . . . 189
4.4 PROBLEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
II EXPERIMENTAL INVESTIGATION 198
5 MODAL ANALYSIS OF ASYSTEMWITH3 DEGREES OF FREE-
DOM 199
5.1 DESCRIPTION OF THE LABORATORY INSTALLATION . . . . . 199
5.2 MODELLING OF THE OBJECT . . . . . . . . . . . . . . . . . . . . 200
5.2.1 Physical model . . . . . . . . . . . . . . . . . . . . . . . . . 200
5.3 ANALYSIS OF THE MATHEMATICAL MODEL . . . . . . . . . . . 202
5.3.1 Natural frequencies and natural modes of the undamped
system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
5.3.2 Equations of motion in terms of the normal coordinates
- transfer functions . . . . . . . . . . . . . . . . . . . . . . 202
5.3.3 Extraction of the natural frequencies and the natural
modes from the transfer functions . . . . . . . . . . . . . 203
5.4 EXPERIMENTAL INVESTIGATION . . . . . . . . . . . . . . . . . 204
5.4.1 Acquiring of the physical model initial parameters . . 204
5.4.2 Measurements of the transfer functions . . . . . . . . . . 205
5.4.3 Identication of the physical model parameters . . . . 206
5.5 WORKSHEET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
INTRODUCTION. 4
0.1 INTRODUCTION.
The purpose of this text is to provide the students with the theoretical background
and engineering applications of the theory of vibrations of mechanical systems. It is
divided into two parts. Part one, Modelling and Analysis, is devoted to this solu-
tion of these engineering problems that can be approximated by means of the linear
models. The second part, Experimental Investigation, describes the laboratory
work recommended for this course.
Part one consists of four chapters.
The rst chapter, Mechanical Vibration of One-Degree-Of-Freedom
Linear System, illustrates modelling and analysis of these engineering problems
that can be approximated by means of the one degree of freedom system. Infor-
mation included in this chapter, as a part of the second year subject Mechanics 1,
where already conveyed to the students and are not to be lectured during this course.
However, since this knowledge is essential for a proper understanding of the following
material, students should study it in their own time.
Chapter two is devoted to modeling and analysis of these mechanical systems
that can be approximated by means of the Multi-Degree-Of-Freedom models.
The Newtons-Eulers approach, Lagranges equations and the inuence coecients
method are utilized for the purpose of creation of the mathematical model. The
considerations are limited to the linear system only. In the general case of damping
the process of looking for the natural frequencies and the system forced response
is provided. Application of the modal analysis to the case of the small structural
damping results in solution of the initial problem and the forced response. Dynamic
balancing of the rotating elements and the passive control of vibrations by means of
the dynamic absorber of vibrations illustrate application of the theory presented to
the engineering problems.
Chapter three, Vibration of Continuous Systems, is concerned with the
problems of vibration associated with one-dimensional continuous systems such as
string, rods, shafts, and beams. The natural frequencies and the natural modes are
used for the exact solutions of the free and forced vibrations. This chapter forms a
base for development of discretization methods presented in the next chapter
In chapter four, Approximation of the Continuous Systems by Dis-
crete Models, two the most important, for engineering applications, methods of
approximation of the continuous systems by the discrete models are presented. The
Rigid Element Method and the Final Element Method are explained and utilized to
produce the inertia and stiness matrices of the free-free beam. Employment of these
matrices to the solution of the engineering problems is demonstrated on a number of
examples. The presented condensation techniques allow to keep size of the discrete
mathematical model on a reasonably low level.
Each chapter is supplied with several engineering problems. Solution to some
of them are provided. Solution to the other problems should be produced by students
during tutorials and in their own time.
Part two gives the theoretical background and description of the laboratory
experiments. One of them is devoted to the experimental determination of the nat-
ural modes and the corresponding natural frequencies of a Multi-Degree-Of-Freedom-
INTRODUCTION. 5
System. The other demonstrates the balancing techniques.
Part I
MODELLING AND ANALYSIS
6
7
Modelling is the part of solution of an engineering problems that aims to-
wards producing its mathematical description. This mathematical description can
be obtained by taking advantage of the known laws of physics. These laws can not
be directly applied to the real system. Therefore it is necessary to introduce many
assumptions that simplify the engineering problems to such extend that the physic
laws may be applied. This part of modelling is called creation of the physical model.
Application of the physics law to the physical model yields the wanted mathematical
description that is called mathematical model. Process of solving of the mathematical
model is called analysis and yields solution to the problem considered. One of the
most frequently encounter in engineering type of motion is the oscillatory motion of
a mechanical system about its equilibrium position. Such a type of motion is called
vibration. This part deals with study of linear vibrations of mechanical system.
Chapter 1
MECHANICAL VIBRATION OF ONE-DEGREE-OF-FREEDOM
LINEAR SYSTEMS
DEFINITION: Any oscillatory motion of a mechanical system about its
equilibrium position is called vibration.
1.1 MODELLING OF ONE-DEGREE-OF-FREEDOM SYSTEM
DEFINITION: Modelling is the part of solution of an engineering problem
that aims for producing its mathematical description.
The mathematical description of the engineering problem one can obtain by
taking advantage of the known lows of physics. These lows can not be directly
applied to the real system. Therefore it is necessary to introduce many assumptions
that simplify the problem to such an extend that the physic laws may by apply. This
part of modelling is called creation of the physical model. Application of the physics
law to the physical model yields the wanted mathematical description which is called
mathematical model.
1.1.1 Physical model
As an example of vibration let us consider the vertical motion of the body 1 suspended
on the rod 2 shown in Fig. 1. If the body is forced out from its equilibrium position
and then it is released, each point of the system performs an independent oscillatory
motion. Therefore, in general, one has to introduce an innite number of independent
coordinates x
i
to determine uniquely its motion.
t
x
i
i
1
2
Figure 1
MODELLING OF ONE-DEGREE-OF-FREEDOM SYSTEM 9
DEFINITION: The number of independent coordinates one has to use to
determine the position of a mechanical system is called number of degrees of
freedom
According to this denition each real system has an innite number of degrees
of freedom. Adaptation of certain assumptions, in many cases, may results in reduc-
tion of this number of degrees of freedom. For example, if one assume that the rod
2 is massless and the body 1 is rigid, only one coordinate is sucient to determine
uniquely the whole system. The displacement x of the rigid body 1 can be chosen as
the independent coordinate (see Fig. 2).
t
x
i
x
i
x
1
2
Figure 2
Position x
i
of all the other points of our system depends on x. If the rod
is uniform, its instantaneous position as a function of x is shown in Fig. 2. The
following analysis will be restricted to system with one degree of freedom only.
To produce the equation of the vibration of the body 1, one has to produce
its free body diagram. In the case considered the free body diagram is shown in Fig.
3.
t
x
1
R
G
Figure 3
The gravity force is denoted by G whereas the force R represents so called
restoring force. In a general case, the restoring force R is a non-linear function of
the displacement x and the instantaneous velocity x of the body 1 (R = R(x, x)).
The relationship between the restoring force R and the elongation x as well as the
velocity x is shown in Fig. 4a) and b) respectively.
x
x
.
R R
0
0
a)
b)
Figure 4
If it is possible to limit the consideration to vibration within a small vicinity
of the system equilibrium position, the non-linear relationship, shown in Fig. 4 can
be linearized.
R=R(x, x) kx +c x (1.1)
The rst term represents the system elasticity and the second one reects the systems
ability for dissipation of energy. k is called stiness and c is called coecient of
damping. The future analysis will be limited to cases for which such a linearization
is acceptable form the engineering point of view. Such cases usually are refer to as
linear vibration and the system considered is call linear system.
Result of this part of modelling is called physical model. The physical model
that reects all the above mention assumption is called one-degree-of-freedom linear
system. For presentation of the physical model we use symbols shown in the Fig. 5.
m
x
m,I
m
k
A, J, E
k
c
.
c
rigid block of mass
rigid body of mass
m
m and moment of inertia I
particle of of mass m
massless spring of stifness k
massless beam area second moment of area and Young modulus A, J E
(linear motion)
(angular motion)
(linear motion)
massless spring of stifness k (angular motion)
massless damper of damping coefficient c (linear motion)
massless damper of damping coefficient c (angular motion)
Figure 5
1.1.2 Mathematical model
To analyze motion of a system it is necessary to develop a mathematical description
that approximates its dynamic behavior. This mathematical description is referred to
as the mathematical model. This mathematical model can be obtained by application
of the known physic lows to the adopted physical model. The creation of the phys-
ical model, has been explained in the previous section. In this section principle of
producing of the mathematical model for the one-degree-of-freedom system is shown.
Let us consider system shown in Fig. 6.
k
m
c
x
mg
s
k x
Figure 6
Let as assume that the system is in an equilibrium. To develop the mathe-
matical model we take advantage of Newtons generalized equations. This require
introduction of the absolute system of coordinates. In this chapter we are assuming
that the origin of the absolute system of coordinates coincides with the centre of
gravity of the body while the body stays at its equilibrium position as shown in Fig.
6. The resultant force of all static forces (in the example considered gravity force
mg and interaction force due to the static elongation of spring kx
s
) is equal to zero.
Therefore, these forces do not have to be included in the Newtons equations. If the
system is out of the equilibrium position (see Fig. 7) by a distance x, there is an
increment in the interaction force between the spring and the block. This increment
is called restoring force.
k
m
c
x
mg
s
k x
-k\x\=-kx
x>0
x<0
k\x\=-kx
Figure 7
In our case the magnitude of the restoring force is |F
R
| = k |x|
If x > 0, the restoring force is opposite to the positive direction of axis x.
Hence F
R
= k |x| = kx
If x < 0, the restoring force has the same direction as axis x. Hence F
R
=
+k |x| = kx
Therefore the restoring force always can be represented in the equation of motion by
term
F
R
= kx (1.2)
k
m
c
x
mg
s
k x
-c\x\=-cx
x>0
x<0
.
.
.
+c\x\=-cx
. .
.
Figure 8
Creating the equation of motion one has to take into consideration the interac-
tion force between the damper and the block considered (see Fig. 8). This interaction
force is called damping force and its absolute value is |F
D
| = c | x| . A very similar to
the above consideration leads to conclusion that the damping force can be represented
in the equation of motion by the following term
F
D
= c x (1.3)
k
m
c
x
F (t)
ex
Figure 9
The assumption that the system is linear allows to apply the superposition
rules and add these forces together with the external force F
ex
(t) (see Fig. 9). Hence,
the equation of motion of the block of mass m is
m x = kx c x +F
ex
(t) (1.4)
Transformation of the above equation into the standard form yields
x + 2
n
x +
2
n
x = f(t) (1.5)
where
n
=
r
k
m
; 2
n
=
c
m
; f(t) =
F
ex
(t)
m
(1.6)
n
- is called natural frequency of the undamped system
- is called damping factor or damping ratio
f(t) - is called unit external excitation
The equation 1.5 is known as the mathematical model of the linear vibration
of the one-degree-of-freedom system.
1.1.3 Problems
Problem 1
A
c
y
k
1
k
2
m
Figure 10
The block of mass m (see Fig. 10)is restricted to move along the vertical axis.
It is supported by the spring of stiness k
1
, the spring of stiness k
2
and the damper
of damping coecient c. The upper end of the spring k
2
moves along the inertial axis
y and its motion is governed by the following equation
y
A
= a sint
were a is the amplitude of motion and is its angular frequency. Produce the equation
of motion of the block.
Solution
A
c
y
k
1
k
2
m
x
Figure 11
Let us introduce the inertial axis x in such a way that its origin coincides with
the centre of gravity of the block 1 when the system is in its equilibrium position (see
Fig. 11. Application of the Newtons low results in the following equation of motion
m x = k
2
x k
1
x +k
2
y c x (1.7)
Its standard form is
x + 2
n
x +
2
n
x = q sint (1.8)
where
2
n
=
k
1
+k
2
m
2
n
=
c
m
q =
k
2
a
m
(1.9)
Problem 2
r
R
1
2
Figure 12
The cylinder 1 (see Fig. 12) of mass m and radius r is plunged into a liquid
of density d. The cylindric container 2 has a radius R. Produce the formula for the
period of the vertical oscillation of the cylinder.
Solution
r
R
x
x
G
z
V
V
1
2
Figure 13
Let us introduce the inertial axis x in such a way that its origin coincides with
the centre of gravity of the cylinder 1 when the system is in its equilibrium position
(see Fig. 13. If the cylinder is displaced from its equilibrium position by a distance
x, the hydrostatic force acting on the cylinder is reduced by
H = (x +z) dgr
2
(1.10)
Since the volume V
1
must be equal to the volume V
2
we have
V
1
= r
2
x = V
2
=
R
2
r
2
z (1.11)
Therefore
z =
r
2
R
2
r
2
x (1.12)
Introducing the above relationship into the formula 1.10 one can get that
H =

x +
r
2
R
2
r
2
x
dgr
2
= dg
R
2
r
2
R
2
r
2
x (1.13)
According to the Newtons law we have
m x = dg
R
2
r
2
R
2
r
2
x (1.14)
The standard form of this equation of motion is
x +
2
n
x = 0 (1.15)
where
2
n
=
dg
m
R
2
r
2
R
2
r
2
(1.16)
The period of the free oscillation of the cylinder is
T
n
=
2
n
=
2
Rr
s
m(R
2
r
2
)
dg
=
2
Rr
s
m (R
2
r
2
)
dg
(1.17)
Problem 3
c
R
G
D
L
m
1
Figure 14
The disk 1 of mass mand radius R(see Fig. 14) is supported by an elastic shaft
of diameter D and length L. The elastic properties of the shaft are determined by
the shear modulus G. The disk can oscillate about the vertical axis and the damping
is modelled by the linear damper of a damping coecient c. Produce equation of
motion of the disk
Solution
c
R
G
D
L
m
1

Figure 15
Motion of the disk is governed by the generalized Newtons equation
I = k
s
cR
2
(1.18)
where
I =
mR
2
2
- the moment of inertia of the disk
k
s
=
T
=
T
TL
JG
=
JG
L
=
D
4
G
32L
the stiness of the rod
Introduction of the above expressions into the equation 1.18 yields
I +cR
2
+
D
4
G
32L
= 0 (1.19)
or
+ 2
n
+
2
n
= 0 (1.20)
where
2
n
=
D
4
G
32LI
2
n
=
cR
2
I
(1.21)
Problem 4
O
k
c
l
b
a
1
m
Figure 16
The thin and uniform plate 1 of mass m (see Fig. 16) can rotate about
the horizontal axis O. The spring of stiness k keeps it in the horizontal position.
The damping coecient c reects dissipation of energy of the system. Produce the
equation of motion of the plate.
Solution

O
k
c
l
b
a
1
m

Figure 17
Motion of the plate along the coordinate (see Fig. 17) is govern by the
generalized Newtons equation
I = M (1.22)
The moment of inertia of the plate 1 about its axis of rotation is
I =
mb
2
12
(1.23)
The moment which act on the plate due to the interaction with the spring k and the
damper c is
M = kl
2
cb
2
(1.24)
Hence
mb
2
12
+kl
2
+cb
2
= 0 (1.25)
or
+ 2
n
+
2
n
= 0 (1.26)
where
2
n
=
12kl
2
mb
2
2
n
=
12c
m
(1.27)
Problem 5
E,I
M
m
t
l
c
Figure 18
The electric motor of mass M (see Fig. 18)is mounted on the massless beam of
length l, the second moment of inertia of its cross-section I and the Young modulus
E. The shaft of the motor has a mass m and rotates with the angular velocity . Its
unbalance (the distance between the axis of rotation and the shaft centre of gravity)
is . The damping properties of the system are modelled by the linear damping of
the damping coecient c. Produce the equation of motion of the system.
Problem 6
y
A
0
c
k
d D
l
L
Figure 19
The wheel shown in the Fig. 19 is made of the material of a density . It
can oscillate about the horizontal axis O. The wheel is supported by the spring of
stiness k and the damper of the damping coecient c. The right hand end of the
damper moves along the horizontal axis y and its motion is given by the following
equation
y = a sin t
Produce the equation of motion of the system
Problem 7
L
r
1
2
Figure 20
The cylinder 1 of mass m is attached to the rigid and massless rod 2 to form
the pendulum shown in the Fig. 20. Produce the formula for the period of oscillation
of the pendulum.
Problem 8
O
k
c
l
b
a
1
m
Figure 21
The thin and uniform plate 1 (see Fig. 21) of mass m can rotate about the
horizontal axis O. The spring of stiness k keeps it in the horizontal position. The
damping coecient c reects dissipation of energy of the system. Produce the formula
for the natural frequency of the system.
ANALYSIS OF ONE-DEGREE-OF-FREEDOM SYSTEM 27
1.2 ANALYSIS OF ONE-DEGREE-OF-FREEDOM SYSTEM
1.2.1 Free vibration
DEFINITION: It is said that a system performs free vibration if there are
no external forces (forces that are explicitly dependent on time) acting on this
system.
In this section, according to the above denition, it is assumed that the resul-
tant of all external forces f(t) is equal to zero. Hence, the mathematical model that
is analyzed in this section takes form
x + 2
n
x +
2
n
x = 0 (1.28)
The equation 1.28 is classied as linear homogeneous ordinary dierential equation of
second order. If one assume that the damping ratio is equal to zero, the equation
1.28 governs the free motion of the undamped system.
x +
2
n
x = 0 (1.29)
Free vibration of an undamped system
The general solution of the homogeneous equation 1.29 is a linear combination of its
two particular linearly independent solutions. These solutions can be obtained by
means of the following procedure. The particular solution can be predicted in the
form 1.30.
x = e
t
(1.30)
Introduction of the solution 1.30 into the equation 1.29 yields the characteristic equa-
tion
2
+
2
n
= 0 (1.31)
This characteristic equation has two roots
1
= +i
n
and
2
= i
n
(1.32)
Hence, in this case, the independent particular solution are
x
1
= sin
n
t and x
2
= cos
n
t (1.33)
Their linear combination is the wanted general solution and approximates the free
vibration of the undamped system.
x = C
s
sin
n
t +C
c
cos
n
t (1.34)
The two constants C
s
and C
c
should be chosen to fulll the initial conditions which
reect the way the free vibrations were initiated. To get an unique solution it is nec-
essary to specify the initial position and the initial velocity of the system considered.
Hence, let us assume that at the instant t = 0 the system was at the position x
0
and
was forced to move with the initial velocity v
0
. Introduction of these initial conditions
into the equation 1.34 results in two algebraic equation that are linear with respect
to the unknown constants C
s
and C
c
.
C
c
= x
0
C
s
n
= v
0
(1.35)
According to 1.34, the particular solution that represents the free vibration of the
system is
x =
v
0
n
sin
n
t +x
0
cos
n
t =
= C sin(
n
t +) (1.36)
where
C =
s
(x
0
)
2
+
v
0
2
; = arctan
x
0
v
0
n
!
(1.37)
For
n
= 1[1/s], x
0
= 1[m] v
0
= 1[m/s] and = 0 the free motion is shown in Fig.
22 The free motion, in the case considered is periodic.
40
-1.5
-1
-0.5
0
0.5
1
10 20 30 50
x[m]
t[s]
C
T
n
x
o
v
o
Figure 22
DEFINITION: The shortest time after which parameters of motion repeat
themselves is called period and the motion is called periodic motion.
According to this denition, since the sine function has a period equal to 2,
we have
sin(
n
(t +T
n
) +) = sin(
n
t + + 2) (1.38)
Hence, the period of the undamped free vibrations is
T
n
=
2
n
(1.39)
Free vibration of a damped system
If the damping ratio is not equal to zero, the equation of the free motion is
x + 2
n
x +
2
n
x = 0 (1.40)
Introduction of the equation 1.30 into 1.40 yields the characteristic equation
2
+ 2
n
+
2
n
= 0 (1.41)
The characteristic equation has two roots
1,2
=
2
n

p
(2
n
)
2
4
2
n
2
=
n

n
p
2
1 (1.42)
The particular solution depend on category of the above roots. Three cases are
possible
Case one - underdamped vibration
If < 1, the characteristic equation has two complex conjugated roots and
this case is often referred to as the underdamped vibration.
1,2
=
n
i
n
p
1
2
=
n
i
d
(1.43)
where
d
=
n
p
1
2
(1.44)
The particular solutions are
x
1
= e
n
t
sin
d
t and x
2
= e
n
t
cos
d
t (1.45)
and their linear combination is
x = e
n
t
(C
s
sin
d
t +C
c
cos
d
t) (1.46)
For the following initial conditions
x |
t=0
= x
0
x |
t=0
= v
0
(1.47)
the two constants C
s
and C
c
are
C
s
=
v
0
+
n
x
0
d
C
c
= x
0
(1.48)
Introduction of the expressions 1.48 into 1.46 produces the free motion in the following
form
x = e
nt
(C
s
sin
d
t +C
c
cos
d
t) = Ce
nt
sin(
d
t +) (1.49)
where
C =
s
v
0
+
n
x
0
2
+ (x
0
)
2
; = arctan
x
0
d
v
0
+
n
x
0
;
d
=
n
p
1
2
(1.50)
For
n
= 1[1/s], x
0
= 1[m] v
0
= 1[m/s] and = .1 the free motion is shown in Fig.
23In this case the motion is not periodic but the time T
d
(see Fig. 23) between every
-1.5
-1
-0.5
0
0.5
1
10 20 30 40 50
x[m]
t[s]
T
d
T
d
x(t)
x(t+ T
d
t
x
o
v
o
)
Figure 23
second zero-point is constant and it is called period of the dumped vibration. It is easy
to see from the expression 1.49 that
T
d
=
2
d
(1.51)
DEFINITION: Natural logarithm of ratio of two displacements x(t) and
x(t + T
d
) that are one period apart is called logarithmic decrement of damping
and will be denoted by .
It will be shown that the logaritmic decrement is constant. Indeed
= ln
x(t)
x(t +T
d
)
= ln
Ce
nt
sin(
d
t +)
Ce
n(t+T
d
)
sin(
d
(t +T
d
) +)
=
= ln
Ce
n
t
sin(
d
t +)
Ce
nt
e
nT
d
sin(
d
t + 2 +)
=
n
T
d
=
2
n
d
=
2
n
1
2
=
=
2
1
2
(1.52)
This formula is frequently used for the experimental determination of the damping
ratio .
=

p
4
2
+
2
(1.53)
The other parameter
n
that exists in the mathematical model 1.40 can be easily
identied by measuring the period of the free motion T
d
. According to the formula
1.44 and 1.51
n
=

d
1
2
=
2
T
d
1
2
(1.54)
Case two - critically damped vibration
If = 1, the characteristic equation has two real and equal one to each other
roots and this case is often referred to as the critically damped vibration
1,2
=
n
(1.55)
x
1
= e
n
t
and x
2
= te
n
t
(1.56)
x = C
s
e
n
t
+C
c
te
n
t
(1.57)
x |
t=0
= x
0
x |
t=0
= v
0
(1.58)
the two constants C
s
and C
c
are as follow
C
s
= x
0
C
c
= v
0
+x
0
n
(1.59)
Introduction of the expressions 1.59 into 1.57 produces expression for the free motion
in the following form
x = e
n
t
(x
0
+t(v
0
+x
0
n
)) (1.60)
For
n
= 1[1/s], x
0
= 1[m] v
0
= 1[m/s] and = 1. the free motion is shown in
Fig. 24. The critical damping oers for the system the possibly faster return to its
equilibrium position.
-1.5
-1
-0.5
0
0.5
1
10 20 30 40 50
x[m]
t[s]
v
o
x
o
Figure 24
Case three - overdamped vibration
If > 1, the characteristic equation has two real roots and this case is often
referred to as the overdamped vibration.
1,2
=
n

n
p
2
1 =
n
(
p
2
1) (1.61)
x
1
= e
n
(+
2
1)t
and x
2
= e
n
(
2
1)t
(1.62)
x = e
n
t
C
s
e
2
1)t
+C
c
e
2
1)t
(1.63)
x |
t=0
= x
0
x |
t=0
= v
0
(1.64)
the two constants C
s
and C
c
are as follow
C
s
=
+
v
0
n
+x
0
(+ +
2
1)
2
2
1
C
c
=

v
0
n
+x
0
( +
2
1)
2
2
1
(1.65)
For
n
= 1[1/s], x
0
= 1[m] v
0
= 1[m/s] and = 5. the free motion is shown in Fig.
25
-1.5
-1
-0.5
0
0.5
1
10 20 30 40 50
x[m]
t[s]
v
o
x
o
Figure 25
1.2.2 Forced vibration
In a general case motion of a vibrating systemis due to both, the initial conditions and
the exciting force. The mathematical model, according to the previous consideration,
is the linear non-homogeneous dierential equation of second order.
x + 2
n
x +
2
n
x = f(t) (1.66)
where
n
=
r
k
m
; 2
n
=
c
m
; f(t) =
F
ex
(t)
m
(1.67)
The general solution of this mathematical model is a superposition of the general
solution of the homogeneous equation x
g
and the particular solution of the non-
homogeneous equation x
p
.
x = x
g
+x
p
(1.68)
The general solution of the homogeneous equation has been produced in the previous
section and for the underdamped vibration it is
x
g
= e
n
t
(C
s
sin
d
t +C
c
cos
d
t) = Ce
n
t
sin(
d
t +) (1.69)
To produce the particular solution of the non-homogeneous equation, let as assume
that the excitation can be approximated by a harmonic function. Such a case is
referred to as the harmonic excitation.
f(t) = q sin t (1.70)
In the above equation q represents the amplitude of the unit excitation and is the
excitation frequency. Introduction of the expression 1.70 into equation 1.66 yields
x + 2
n
x +
2
n
x = q sint (1.71)
In this case it is easy to predict mode of the particular solution
x
p
= A
s
sin t +A
c
cos t (1.72)
where A
s
and A
c
are constant. The function 1.72 is the particular solution if and
only if it fulls the equation 1.71 for any instant of time. Therefore, implementing it
in equation 1.71 one can get
(
2
n
2
)A
s
2
n
A
c
sin t+
2
n
A
s
+ (
2
n
2
)A
c
cos t = q sint (1.73)

This relationship is fullled for any instant of time if
(
2
n
2
)A
s
2
n
A
c
= q
2
n
A
s
+ (
2
n
2
)A
c
= 0 (1.74)
Solution of the above equations yields the expression for the constant A
s
and A
c
A
s
=
q 2
n
0 (
2
n
2
)
(
2
n
2
) 2
n
2
n
(
2
n
2
)
=
(
2
n
2
)q
(
2
n
2
)
2
+ 4(
n
)
2
2
A
c
=
(
2
n
2
) q
2
n
0
(
2
n
2
) 2
n
2
n
(
2
n
2
)
=
2(
n
)q
(
2
n
2
)
2
+ 4(
n
)
2
2
(1.75)
Introduction of the expressions 1.75 into the predicted solution 1.72 yields
x
p
= A
s
sin t +A
c
cos t = Asin(t +) (1.76)
where
A =
p
A
2
s
+A
2
c
=
q
p
(
2
n
2
)
2
+ 4(
n
)
2
2
= arctan
A
c
A
s
= arctan
2(
n
)
2
n
2
(1.77)
or
A =
q
2
n
q
(1 (

n
)
2
)
2
+ 4
2
(

n
)
2
= arctan
2

n
1 (

n
)
2
(1.78)
Introducing 1.69 and 1.76 into the 1.68 one can obtain the general solution of the
equation of motion 1.71 in the following form
x = Ce
n
t
sin(
d
t +) +Asin(t +) (1.79)
The constants C and should be chosen to fulll the required initial conditions.
x |
t=0
= x
0
x |
t=0
= v
0
(1.80)
one can get the following set of the algebraic equations for determination of the
parameters C and
x
0
= C
o
sin
o
+Asin
v
0
= C
o
n
sin
o
+C
o
d
cos
o
+A cos (1.81)
Introduction of the solution of the equations 1.81 (C
o,
o
) to the general solution,
yields particular solution of the non-homogeneous equation that represents the forced
vibration of the system considered.
x = C
o
e
nt
sin(
d
t +
o
) +Asin(t +) (1.82)
This solution, for the following numerical data = 0.1,
n
= 1[1/s], = 2[1/s],
C
o
= 1[m],
o
= 1[rd], A = 0.165205[m], = 0.126835[rd] is shown in Fig. 26
(curve c).The solution 1.82 is assembled out of two terms. First term represents an
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
20 40 60
x[m]
t[s]
transient state of the forced vibration
steady state of the forced vibration
a b
c
A
Figure 26
oscillations with frequency equal to the natural frequency of the damped system
d
.
Motion represented by this term, due to the existing damping, decays to zero (curve
a in Fig. 1.82) and determines time of the transient state of the forced vibrations.
Hence, after an usually short time, the transient state changes into the steady state
represented by the second term in equation 1.82 (curve b in Fig. 1.82)
x = Asin(t +) (1.83)
This harmonic term has amplitude A determined by the formula 1.77. It does not
depend on the initial conditions and is called amplitude of the forced vibration. Mo-
tion approximated by the equation 1.83 is usually referred to as the system forced
vibration.
Both, the exciting force f(t) = q sin t (1.70) and the (steady state) forced
vibration x = Asin(t +) (1.83) are harmonic. Therefore, they can be represented
by means of two vectors rotating with the same angular velocity (see Fig. 27).One
x
A
t
q
Asin( t+ )
qsin( t )
Figure 27
can see from the above interpretation that the angular displacement is the phase
between the exciting force and the displacement it causes. Therefore is called phase
of the forced vibration.
Because the transient state, from engineering point of view play secondary
role, in the following sections the steady state forced vibration will be considered
only.
Forced response due to rotating elements - force transmitted to foundation.
x
t
m
2
sin t x
M
m
m
2
sin t
M
k c
R
m
2
a)
b)
Figure 28
One of many possible excitation of vibrations is excitation caused by inertia
forces produced by moving elements. The possibly simplest case of vibration cased
by this type of excitation is shown in Fig. 28. The rotor of an electrical motor rotates
with the constant angular velocity . If represents the static imbalance of the rotor
and m is its mass, then the rotor produces the centrifugal force
F = m
2
(1.84)
Its component along the vertical axis x is
F
x
= m
2
sint (1.85)
The motor of mass M is supported by means of a beam of the stiness k. The
damping properties are approximated by the damping coecient c. Let us model
vibration of the system. The physical model of the problem described is shown in
Fig. 28b). Taking advantage of the earlier described method of formulation the
mathematical model we have
M x = kx c x +m
2
sin t (1.86)
Transformation of this equation into the standard form yields
x + 2
n
x +
2
n
x = q sint (1.87)
where
n
=
r
k
M
2
n
=
c
M
q =
m
2
M
(1.88)
Hence, the steady state forced vibration are
x = Asin(t +) (1.89)
where according to 1.77
A =
q
2
n
q
(1 (

n
)
2
)
2
+ 4
2
(

n
)
2
= arctan
2

n
1 (

n
)
2
(1.90)
or, taking into consideration 1.88
A =
m
M
(

n
)
2
q
(1 (

n
)
2
)
2
+ 4
2
(

n
)
2
= arctan
2

n
1 (

n
)
2
(1.91)
The ratio
A
m
M
, is called the magnication factor, Its magnitude and the phase as

a function of the ratio

n
for dierent damping factor is shown in Fig. 29.If the
frequency of excitation changes from zero to the value equal to the natural frequency
n
, the amplitude of the forced vibration is growing. Its maximum depends on the
damping ratio and appears for >
n
. The phenomenon at which amplitude of
the forced vibration is maximum is called amplitude resonance. If the frequency of
excitation tends towards innity, the amplitude of the forced vibration tends to
m
M
.
For =
n
, regardless the damping involved, phase of the forced vibration is equal
to 90
o
. This phenomenon is called phase resonance. If the frequency of excitation
0
2
4
6
1
2
3
-180
-135
-90
-45
0
1 2 3
=0
=0.1
=0.25
=0.5
=1.0
=1.5
m
M
1
=0
=0.1
=0.25
=0.5
=1.0
=1.5
Figure 29
tends to innity, the phase tends to 180
o
. Hence the response of the system tends to
be in the anti-phase with the excitation.
The force transmitted to the foundation R, according to the physical model
shown in Fig. 28b) is
R(t) = kx +c x = kAsin(t +) +cA cos(t +) = A
k
2
+c
2
2
sin(t + +)
(1.92)
The amplitude of the reaction is
|R| = A
k
2
+c
2
2
= AM
p
4
n
+ 4
2
2
n
2
=
= m
2
q
1 + 4
2
(

n
)
2
q
(1 (

n
)
2
)
2
+ 4
2
(

n
)
2
The amplication ratio
|R|
m
2
of the reaction as a function of the ratio

n
is shown in
Fig. 30.For the frequency of excitation < 1.4
n
the force transmitted to foundation
0
2
4
6
0 1 2 3
n
R
m
2
1
1.4
=0
=0.1
=0.25
=0.5
=1.0
=1.5
Figure 30
is greater then the centrifugal force itself with its maximum close to frequency
n
.
For > 1.4
n
this reaction is smaller then the excitation force and tends to zero
when the frequency of excitation approaches innity.
Forced response due to the kinematic excitation - vibration isolation
The reaction force R transmitted to foundation causes its vibration. In turn, this
vibration are transmitted to another objects produces its own vibration which very
L
y
M
k
c
y
x
a
v
R
G
B
a)
b)
Figure 31
often are considerably larger then vibration of the foundation itself. Such a type of
excitation is called kinematic excitation.
Aphysical model of systemwith the kinematic excitation is shown in Fig. 31b).
Motion of the point B along the axis y causes vibration of the block M. This physical
model can be used to analyze vibration of a bus caused by the roughness of the
surface of the road shown in Fig. 31a). The stiness k of the spring and the damping
coecient c represent the dynamic properties of the bus shock-absorbers. The block
of mass M stands for the body of the bus. If the surface can be approximated by the
sine-wave of the amplitude a and length L and the bus is travelling with the constant
velocity v, the period of the harmonic excitation is
T =
L
v
(1.93)
Hence, the frequency of excitation, according to 1.39 is
=
2v
L
(1.94)
and the motion of the point B along the axis y can approximated as follows
y = a sin t (1.95)
The equation of motion of the bus is
M x = kx c x +ky +c y (1.96)
Introduction of 1.95 yields
M x +c x +kx = ka sin t +ca cos t
or
x + 2
n
x +
2
n
x =
2
n
a sin t + 2
n
a cos t = q sin(t +) (1.97)
where
q = a
2
n
r
1 + 4
2
(

n
)
2
(1.98)
Without any harm to the generality of the considerations one can neglect the phase
and adopt the mathematical model in the following form
x + 2
n
x +
2
n
x = q sint (1.99)
Motion of the block along axis x is governed by the equation 1.83
x = Asin(t +)
where
A =
q
2
n
q
(1 (

n
)
2
)
2
+ 4
2
(

n
)
2
= arctan
2

n
1 (

n
)
2
(1.100)
Introduction of equation 1.98 gives
A =
a
q
1 + 4
2
(

n
)
2
q
(1 (

n
)
2
)
2
+ 4
2
(

n
)
2
= arctan
2

n
1 (

n
)
2
(1.101)
The magnifying factor
A
a
and the phase as a function of

n
is shown in Fig. 32For
< 1.4
n
it is possible to arrange for the bus to have vibration smaller than the
amplitude of the kinematic excitation
The expression for the reaction force transmitted to the foundation is
R = kx +c x ky c y = kAsin(t +) +cAcos(t +) ka sin t ca cos t
= |R| sin(t +) (1.102)
Problem of minimizing the reaction force R (e.g. 1.92) or the amplitude A (e.g..
1.101) is called vibration isolation.
-180
-135
-90
-45
0
1 2 3
A
a
0
2
4
6
0 1 2 3
=0
=0.1
=0.25
=0.5
=1.0
=1.5
=0
=0.1
=0.25
=0.5
=1.0
=1.5
Figure 32
1.2.3 Problems
Free vibrations
Problem 9
1
2
H
k c
Figure 33
The carriage 1 of the lift shown in Fig. 33 operates between oors of a building.
The distance between the highest and the lowest oor is H = 30m. The average mass
of the carriage is m = 500kg. To attenuate the impact between the carriage and the
basement in the case the rope 3 is broken, the shock absorber 2 is to be installed.
Calculate the stiness k and the damping coecient c of the shock-absorber
which assure that the deceleration during the impact is smaller then 200m/s
2
.
Solution
In the worst case scenario, the lift is at the level H when the rope brakes.
H
k c
x
x
mg
Figure 34
Due to the gravity force the lift is falling down with the initial velocity equal to zero.
Equation of motion of the lift is
m x = mg (1.103)
By double side by side integrating of the above equation one can get
x = A+Bt +
g
2
t
2
(1.104)
Introduction of the following initial conditions
x |
t=0
= 0 x |
t=0
= 0
yields A = 0 and B = 0 and results in the following equation of motion
x =
g
2
t
2
(1.105)
Hence, the time the lift reaches the shock-absorber is
t
o
=
s
2H
g
(1.106)
Since
v = x = gt (1.107)
the velocity of the lift at the time of the impact with the shock-absorber is
v
o
= x |
t=to
=
p
2Hg (1.108)
To analyze the motion of the lift after impact let us introduce the inertial axis y
in such a way that its origin coincides with the upper end of the shock-absorber at
the instant of impact (see Fig. 35).Since at the instant of impact the spring k is
H
k
c
x
y
mg
y
Figure 35
uncompressed, the equation of motion after the lift has reached the shock-absorber is
m y +c y +ky = mg (1.109)
or in the standardized form
y + 2
n
y +
2
n
y = g (1.110)
where
n
=
r
k
m
; 2
n
=
c
m
(1.111)
It is easy to see that in the case considered the particular solution of the non-
homogeneous equation is
y
p
=
g
2
n
(1.112)
The best performance of the shock-absorber is expected if the damping is critical
( = 1). In this case, there exists one double root and the general solution of the
homogeneous equation is
y
g
= C
1
e
n
t
+C
2
te
n
t
(1.113)
Therefore the general solution of the non-homogeneous equation as the sum of y
p
and
y
g
is
y = C
1
e
nt
+C
2
te
nt
+
g
2
n
(1.114)
This equation has to fulll the following initial conditions
y |
t=0
= 0 y |
t=0
= v
o
(1.115)
Introduction of these initial conditions into the equation 1.113 yields
C
1
=
g
2
n
C
2
= v
o
n
(1.116)
and results in the following equation of motion
y =

2
n
n
t
+
v
o
te
n
t
+
g
2
n
=
g
2
n
1 e
n
t
v
o
te
n
t
= D
1 e
n
t
+Ete
n
t
(1.117)
where
D =
g
2
n
E = v
o
n
(1.118)
Double dierentiation of the function 1.117 yields acceleration during the impact
y =

D
2
n
2E
n
n
t
+E
2
n
te
n
t
(1.119)
By inspection of the function 1.118, one can see that the maximum of the deceleration
occurs for time t = 0. Hence the maximum of deceleration is
a
max
= y |
t=0
=

D
2
n
2E
n
(1.120)
If
v
o
>
g
n
(1.121)
both constants E and D are positive. Hence
a
max
= D
2
n
+ 2E
n
= g + 2v
o
n
2g = 2v
o
n
g (1.122)
This deceleration has to be smaller then the allowed deceleration a
a
= 200ms
2
.
2v
o
n
g < a
a
(1.123)
It follows
n
<
a
a
+g
2
2Hg
=
200 + 10
2
2 30 10
= 4.28s
1
(1.124)
Since
n
=
q
k
m
, the stiness of the shock-absorber is
k =
2
n
m = 4.28
2
500 = 9160N/m (1.125)
and the damping coecient
c = 2
n
m = 2 1 4.28 500 = 4280Ns/m (1.126)
Our computation can be accepted only if the inequality 1.121 is fulll. Indeed
v
o
=
p
2Hg =
2 30 10 = 24.5 >
g
n
=
10
4.28
= 2.4m/s (1.127)
The displacement of the lift, its velocity and acceleration during the impact as a
function of time is shown in Fig. 36
displacement
time [s]
y [m]
0
0.5
1
1.5
2
2.5
0 0.25 0.5 0.75 1
v [m/s]
velocity
time [s]
-5
0
5
10
15
20
25
0.25 0.5 0.75 1
acceleration
time [s]
-200
-150
-100
-50
0
50
0.25 0.5 0.75 1
a [m/s ]
2
Figure 36
Problem 10
The power winch W was mounted on the truss T as shown in Fig. 37a) To
T
R
W k
m
c
x
a)
b)
Figure 37
analyze the vibrations of the power winch the installation was modelled by the one
degree of freedom physical model shown in Fig. 38b). In this gure the equivalent
mass, stiness and damping coecient are denoted by m, k and c respectively. Origin
of the axis x coincides with the centre of gravity of the weight m when the system
rests in its equilibrium position.
To identify the unknown parameters m, k, and c, the following experiment was
carried out. The winch was loaded with the weight equal to M
1
= 1000kg as shown in
Fig. 38. Then the load was released allowing the installation to perform the vertical
T
R
W
M
l
L
Figure 38
oscillations in x direction. Record of those oscillations is presented in Fig. 39.
Calculate the parameters m, k, and c.
4
3
1 2
time [s]
-0.003
-0.002
-0.001
0
0.001
0.002
0.003
x[m]
Figure 39
Answer
m = 7000kg; k = 3000000Nm
1
; c = 15000Nsm
1
Problem 11
T
R
W k
m
c
x
a)
b)
Figure 40
The winch W shown in Fig. 40 is modelled as a system with one degree of
freedom of mass m stiness k and the damping coecient c. The winch is lifting the
block of mass M with the constant velocity v
o
(see Fig. 41).Assuming that the rope
T
R
W
M
Figure 41
R is not extendible produce expression for the tension in the rope R before and after
the block will lose contact with the oor.
Solution
Tension in the rope R before the contact is lost
In the rst stage of lifting the block M, it stays motionlessly at the oor
whereas the lift itself is going down with respect to the inertial axis x with the
constant velocity v
o
. The tension T in the rope R varies between 0 and Mg.
0 < T 0 Mg (1.128)
If origin of the inertial axis x coincides with the gravity centre when the unloaded
winch is at its equilibrium, the equation of motion of the winch is
m x +c x +kx = T (1.129)
In the above equation x = 0 (the winch is moving with the constant velocity v
o
),
x = v
o
and x = v
o
t. Hence
T = c(v
o
) +k(v
o
t) (1.130)
The equation 1.130 governs motion of the winch till the tension T will reach value
Mg. Therefore the equation 1.130 allows the time of separation t
s
to be obtained.
t
s
=
Mg cv
o
kv
o
(1.131)
At the instant of separation the winch will be at the position determined by the
following formula
x
s
= v
o
t
s
=
Mg cv
o
k
(1.132)
If Mg < cv
o
then x
s
= t
s
= 0.
If Mg > cv
o
T = cv
o
+kv
o
t for 0 < t < t
s
(1.133)
Tension in the rope R after the contact of the weight with the oor is lost
Without any harm to the generality of the further consideration one may
assume that the time corresponding to the instant of separation is equal to 0.
For t > 0, the equation of motion of the winch and the block (see Fig. 42) are
as following
m x +c x +kx = T
M x
b
= T Mg (1.134)
Since the rope R is not extendible, the instantaneous length of the rope L is
L = L
o
v
o
t (1.135)
k
m
c
x
T
T
Mg
x
x
b
L
Figure 42
Where L
o
stands for the initial length of the rope (the lenght the rope had at the
instance t = 0). Taking into account that
L = x x
b
(1.136)
we have
x
b
= x L = x L
o
+v
o
t (1.137)
Introduction of the equation 1.137 into 1.134 yields
m x +c x +kx = T
M x = T Mg (1.138)
Elimination of the unknown tension force allows the equation of motion of the winch
to be formulated
(m+M) x +c x +kx = Mg (1.139)
The standardized form is as following
x + 2
n
x +
2
n
x = q (1.140)
where
n
=
r
k
m+M
; 2
n
=
c
m+M
; q =
Mg
m+M
(1.141)
The particular solution of the non-homogeneous equation can be predicted as a con-
stant magnitude A. Hence
2
n
A = q; A =
q
2
n
(1.142)
The general solution of the mathematical model 1.140 is
x = e
nt
(C
s
sin
d
t +C
c
cos
d
t) +A (1.143)
where
d
=
n
p
1
2
(1.144)
This solution has to fulll the following initial conditions
for t = 0 x = x
s
x = v
o
(1.145)
Introduction of these initial conditions into the solution 1.143 yields the following
expressions for the constants C
s
and C
c
C
s
=
v
o
+
n
(x
s
A)
d
C
c
= x
s
+A (1.146)
Hence,
x = e
n
t
v
o
+
n
(x
s
A)
d
sin
d
t + (x
s
+A) cos
d
t
+A (1.147)
The time history diagram of the above function is shown in Fig.43 The tension is
x
t
x
s
T
d
x
max
A
O
Figure 43
determined by the equation 1.138
T = M x +Mg (1.148)
Double dierentiation of the function 1.147 yields the wanted tension as a function
of time
T = Mg +Me
nt
C
s
(
n
)
2
+ 2C
c
d
C
s
2
d
sin
d
t
+Me
n
t
C
c
(
n
)
2
2C
s
d
+C
c
2
d
cos
d
t (1.149)
Forced vibration
Problem 12
E,I
M
m
t
l
c
B
A
Figure 44
The electric motor of mass M (see Fig. 44) is mounted on the massless beam
of length l, the second moment of inertia of its cross-section I and Young modulus E.
Shaft of the motor, of mass m, rotates with the constant angular velocity and its
unbalance (distance between the axis of rotation and the shaft centre of gravity) is
. The damping properties of the system are modelled by the linear damping of the
damping coecient c. Produce expression for the amplitude of the forced vibration
of the motor as well as the interaction forces transmitted to the foundation at the
points A and B.
Solution
E,I
M
m
t
l
c
x
m
2
A
B
Figure 45
Application of the Newtons approach to the system shown in Fig. 45 results
in the following dierential equations of motion.
(m+M) x = kx c x +m
2
sin t (1.150)
where k stands for the stiness of the beam EI.
k =
48EI
l
3
(1.151)
Its standardized form is
x + 2
n
x +
2
n
x = q sint (1.152)
where
n
=
r
k
m+M
2
n
=
c
m+M
q =
m
2
m+M
(1.153)
The particular solution of the equation 1.152
x = Asin (t +) (1.154)
where
A =
q
2
n
q
(1 (

n
)
2
)
2
+ 4
2
(

n
)
2
= arctan
2

n
1 (

n
)
2
(1.155)
represents the forced vibrations of the system. In the above formula A stands for the
amplitude of the forced vibrations of the motor. The interaction force at the point
A can be determined from equilibrium of forces acting on the beam at an arbitrarily
chosen position x (see Fig 46).
E,I
0.5k
l
x
x
x 0.5k
x
kx
A
D
Figure 46
The force needed to displace the point D by x is equal to kx. Hence, the
reaction at the point A is
R
A
= 0.5kx = 0.5kAsin (t +) (1.156)
B
x
x
c
c
D
x c
x
Figure 47
To move the point D (see Fig. 47) with the velocity x the force c x is required.
Hence, from the equilibrium of the damper one can see that the reaction at the point
B is
R
B
= c x = cAsin(t +)
Problem 13
x
z
y
z=x-y
1
2
3
4
5
Figure 48
Figure 48 presents a seismic transducer. Its base 2 is attached to the vibrating
object 1. The seismic weight 3 of mass m is supported by the spring 4 of stiness k and
the damper 5 of the damping coecient c.This transducer records the displacement
z = x y (1.157)
where y is the absolute displacement of the vibration object 1 and x is the absolute
displacement of the seismic weight 3. Upon assuming that the object 1 performs a
harmonic motion
y = a sin t (1.158)
derive the formula for the amplication coecient of the amplitude of vibration of
the object 1 of this transducer ( =
amplitude of z
amplitude of y
) as a function of the non-dimensional
frequency

n
.
Solution
The equation of motion of the system shown in Fig. 48 is
m x +c x +kx = c y +ky (1.159)
Its standardize form is
x + 2
n
x +
2
n
x = aq
c
cos t +aq
s
sin t (1.160)
where
n
=
r
k
m
2
n
=
c
m
q
c
=
c
m
q
s
=
k
m
(1.161)
Simplication of the right side of the above equation yields
x + 2
n
x +
2
n
x = aq sin (t +) (1.162)
where
q =
p
q
2
c
+q
2
s
=
2
n
s
4
2
2
+ 1 = arctan
q
c
q
s
= arctan 2

n
(1.163)
According to equation 1.76 (page 34) the particular solution of the equation 1.162 is
x
p
= aAsin(t + +) (1.164)
where
A =
r
4
2
2
+ 1
s
2
+ 4
2
2
= arctan
2

n
1
2
(1.165)
Hence the record of the transducer is
z = x y = aAsin(t + +) a sin t =
= aAcos ( +) sint +aAsin( +) cos t a sint =
= (aAcos ( +) a) sint +aAsin( +) cos t (1.166)
The amplitude of this record is
amp
z
=
q
(aAcos ( +) a)
2
+ (aAsin( +))
2
= a
p
A
2
+ 1 2Acos( +)
(1.167)
Therefore, the coecient of amplication is
=
amp
z
amp
y
=
p
A
2
+ 1 2Acos( +) (1.168)
The diagram presented in Fig.49 shows this amplication coecient as a function
of the ratio

n
.If the coecient of amplication is equal to one, the record of the
amplitude of vibration (amp
z
) is equal to the amplitude of vibration of the object
(amp
y
= a). It almost happends, as one can see from the diagram 49, if the frequency
of the recorded vibrations is twice greater than the natural frequency
n
of the
transducer and the damping ratio is 0.25.
0
1
2
3
4
5
6
1 2 3 4 5
/
n
=0.1
=0.25
=0.5
Figure 49
Problem 14
O
A B

x
a
b
c
d
k
G
C
Figure 50
The physical model of a vibrating table is shown in Fig. 50. It can be con-
sidered as a rigid body of the mass m and the moment of inertia about axis through
its centre of gravity I
G
. It is supported with by means of the spring of the stiness
k and the damper of the damping coecient c. The motion of the lower end of the
spring with respect to the absolute coordinate x can be approximated as follows
x = X cos t
where X stands for the amplitude of the oscillations of the point C and stands for
the frequency of these oscillations.
Produce:
1. the dierential equation of motion of the vibrating table and present it in
the standard form
2. the expression for the amplitude of the forced vibrations of the table caused
by the motion of the point C
3. the components of the interaction force at the point A
4. the expression for the driving force that has to be applied to the point C
Chapter 2
MECHANICAL VIBRATION OF MULTI-DEGREE-OF-FREEDOM
LINEAR SYSTEMS
Since in the nature massless or rigid elements do not exist, therefore each of the
particle the real element is made of can moves independently. It follows that to
determine its position with respect to the inertial space one has to introduce innite
number of coordinates. Hence, according to the previously introduce denition, the
number of degrees of freedom of each real element is equal to innity. But in many
vibration problems, with acceptable accuracy, the real elements can be represented
by a limited number of rigid elements connected to each other by means of massless
elements representing the elastic and damping properties. This process is called
discretization and the nal result of this process is called multi-degree-of-freedom
system. In this chapter it will be assumed that forces produced by these massless
m
i
k
i
c
i
m
j
x
i
k
ij
c
ij
x
j
y (t)
i
Figure 1
elements (springs and dampers) are linear functions of displacements and velocities
respectively.
2.1 MODELLING
Fig. 1 shows part of a multi-degree-of-freedom system. Usually, to describe motion of
such a system a set of local generalized coordinates is introduced. These coordinates
(x
i
, x
j
, y
i
(t)) are motionless with respect to a global (inertial) system of coordinates
MODELLING 64
(not shown in the Fig. 1). The coordinate y
i
(t) is not independent (is an explicit
function of time) whereas the coordinates x
i
, and x
j
are independent and their number
determines the number of degree of freedom of the system. Origin of each coordinate
coincides with the centre of gravity of individual bodies when the whole system is at
its equilibrium position. For this equilibrium position all the static forces acting on
individual bodies produces the resultant force equal to zero.
It will be shown in this section that the equation of motion of the multi-degree of
freedom linear system has the following form
m x+c x+kx=F(t)
where
m - is the inertia matrix
c - is the damping matrix
k- is the stiness matrix
F - is the external excitation matrix
x- is the displacement matrix
There are many methods that allow the mathematical model to be formulated.
In the following sections a few of them are presented.
Newton-Euler method of formulation of the mathematical model
To develop the equations of motion of the system described, one may utilize the
Newtons or Eulers equations. Since in case considered the body of mass m
i
performs
a plane motion hence the Newtons equations may be used.
m
i
x
i
= F (2.1)
If the system stays at its equilibrium position, as it was mention earlier, the resultant
of all static forces is equal zero. Therefore, the force F must contains forces due to
the displacement of the system from its equilibrium position only. To gure these
forces out let us move the mass m
i
out of its equilibrium position by the displacement
x
i
. The conguration a) shown in the gure below is achieved.
MODELLING 65
a)
x
i
6= 0
x
j
= 0
y
i
= 0
x
i
= 0
x
j
= 0
y
i
= 0
b)
x
i
= 0
x
j
6= 0
y
i
= 0
x
i
= 0
x
j
= 0
y
i
= 0
c)
x
i
= 0
x
j
= 0
y
i
6= 0
x
i
= 0
x
j
= 0
y
i
= 0
d)
x
i
= 0
x
j
= 0
y
i
= 0
x
i
6= 0
x
j
= 0
y
i
= 0
e)
x
i
= 0
x
j
= 0
y
i
= 0
x
i
= 0
x
j
6= 0
y
i
= 0
f)
x
i
= 0
x
j
= 0
y
i
= 0
x
i
= 0
x
j
= 0
y
i
6= 0
m
i
k
i
c
i
m
j
x
i
k
ij c
ij
x
j
y (t)
i
x
i
kij i
- x
ki i
- x
m
i
k
i
c
i
m
j
x
i
k
ij c
ij
x
j
y (t)
i
x
j
kij j
+ x
m
i
k
i c
i
m
j
x
i
k
ij c
ij
x
j
y (t)
i
y
i
k
i i + y m
i
k
i
c
i
m
j
x
i
k
ij c
ij
x
j
y (t)
i
x
i
c
ij i
- x
ci i
- x
m
i
k
i
c
i
m
j
x
i
k
ij c
ij
x
j
y (t)
i
x
j
c
ij j
+ x
m
i
k
i
c
i
m
j
x
i
k
ij c
ij
x
j
y (t)
i
y
j
m
i
x
i
= k
i
x
i
k
ij
x
i
+k
ij
x
j
+k
i
y
i
(t) c
i
x
i
c
ij
x
i
+c
ij
x
j
+0
(2.2)
Due to this displacement there are two forces k
i
x
i
and k
ij
x
i
acting on the considered
mass m
i
. Both of them must be taken with sign - because the positive displacement
x
i
causes forces opposite to the positive direction of axis x
i
. Similar consideration
carried out for the displacements along the axis x
j
(conguration b)) and axis y
i
(
conguration c)) results in the term +k
ij
x
j
. and +k
i
y
i
(t) respectively. Up to now it
has been assumed that the velocities of the system along all coordinates are equal
to zero and because of this the dampers do not produce any force. The last three
congurations (d, e, and f) allow to take these forces into account. Due to motion of
the system along the coordinate x
i
with velocity x
i
two additional forces are created
by the dampers c
i
and c
ij.
they are c
i
x
i
and c
ij
x
i
. Both of them are caused by
positive velocity and have sense opposite to the positive sense of axis x
i
. Therefore
they have to be taken with the sign -. The forces caused by motion along the axis
x
j
(conguration e)) and axis y
i
( conguration f)) results in the term +c
ij
x
j
. and 0
respectively. Since the system is linear, one can add all this forces together to obtain
m
i
x
i
= k
i
x
i
k
ij
x
i
+k
ij
x
j
+k
i
y
i
(t) c
i
x
i
c
ij
x
i
+c
ij
x
j
(2.3)
After standardization we have the nal form of equation of motion of the mass m
i
.
m
i
x
i
+ (c
i
+c
ij
) x
i
c
ij
x
j
+ (k
i
+k
ij
)x
i
k
ij
x
j
= k
i
y
i
(t) (2.4)
To accomplished the mathematical model, one has to carry out similar consideration
for each mass involved in the system. As a result of these consideration we are getting
set of dierential equation containing as many equations as the number of degree of
freedom.
MODELLING 66
Lagrange method of formulation of the mathematical model
The same set of equation of motion one can get by utilization of the Lagranges
equations
d
dt
(

q
m
T)

q
m
T +
V
q
m
= Q
m
m = 1, 2, ....M (2.5)
where
T - is the system kinetic energy function
V - stands for the potential energy function
Q
m
- is the generalized force along the generalized coordinate q
m
The kinetic energy function of the system considered is equal to sum of the
kinetic energy stored in the individual rigid bodies the system is made of. Hence
T =
I
X
i=1
_
_
1
2
m
i
v
2
i
+
1
2

ix

iy

iz
_
_
I
ix
0 0
0 I
iy
0
0 0 I
iz
_
_
_
_

ix
iy
iz
_
_
_
_
(2.6)
where
m
i
- mass of the rigid body
v
i
- absolute velocity of the centre of gravity of the body
ix,
iy,
iz,
- components of the absolute angular velocity of the body
I
ix
, I
iy
, I
iz
- The principal moments of inertia of the body about axes through
its centre of gravity
Potential energy function V for the gravity force acting on the link i shown in Fig. 2
is
V
i
= m
i
gr
GiZ
(2.7)
Z
X
Y
O
r
G
i
G
i
i
r
GiZ
Figure 2
Potential energy for the spring s of stiness k
s
and uncompressed length l
s
(see Fig. 3) is
V
s
=
1
2
k
s
(|r
A
r
B
| l
s
)
2
(2.8)
MODELLING 67
Z
X
Y
O
A
B
s
r
A
r
B
Figure 3
Potential energy function for all conservative forces acting on the system is
V =
I
X
i=1
V
i
+
S
X
s=1
V
s
(2.9)
In a general case the damping forces should be classied as non-conservative ones
and, as such, should be included in the generalized force Q
m
. It must be remembered
that the Lagranges equations yield, in general case, a non-linear mathematical model.
Therefore, before application of the developed in this chapter methods of analysis, the
linearization process must be carried out. The following formula allows for any non-
linear multi-variable function to be linearized in vicinity of the system equilibrium
position q
o
1
, ...q
o
m
, ...q
o
M
f(q
1
, ...q
m
, ...q
M
, q
1
, ... q
m
, ... q
M
) = f(q
o
1
, ...q
o
m
, ...q
o
M
, 0, ...0, ...0)+
+
P
M
m=1
f
q
m
(q
o
1
, ...q
o
m
, ...q
o
M
, 0, ...0, ...0) q
m
+
P
M
m=1
f
q
m
(q
o
1
, ...q
o
m
, ...q
o
M
, 0, ...0, ...0) q
m
(2.10)
In the case of the system shown in Fig. 1 the kinetic energy function is
T =
1
2
m
i
x
2
i
+
1
2
m
j
x
2
j
+ (2.11)
Dots in the above equation represents this part of the kinetic energy function that
does not depend on the generalized coordinate x
i
.
If the system takes an arbitral position that is shown in Fig. 4, elongation of
the springs k
i
and k
ij
are respectively
l
i
= x
i
y
i
l
ij
= x
j
x
i
(2.12)
MODELLING 68
m
i
k
i
c
i
m
j
x
i
k
ij
c
ij
x
j
y (t)
i
m
i
k
i
c
i
m
j
x
i
k
ij
c
ij
x
j
y (t)
i
x
j
x
i
y
i
Figure 4
Therefore, the potential energy function is
V =
1
2
k
i
(x
i
y
i
)
2
+
1
2
k
ij
(x
j
x
i
)
2
+ (2.13)
Again, dots stands for this part of the potential energy function that does not depend
on the generalized coordinate x
i
. It should be noted that the above potential energy
function represents increment of the potential energy stored in the springs due to
the displacement of the system from its equilibrium position. Therefore the above
function does not include the potential energy due to the static deection of the
springs. It follows that the conservative forces due to the static deections can not
be produced from this potential energy function. They, together with the gravity
forces, produce resultant equal to zero. Hence, if the potential energy due to the
static deections is not included in the function 2.13 the potential energy due to
gravitation must not be included in the function 2.13 either. If the potential energy
due to the static deections is included in the function 2.13 the potential energy due
to gravitation must be included in the function 2.13 too.
Generally, the force produced by the dampers is included in the generalized
force Q
m
. But, very often, for convenience, a damping function (dissipation function)
D is introduced into the Lagranges equation to produce the damping forces. The
function D does not represent the dissipation energy but has such a property that its
partial derivative produces the damping forces. The damping function is created by
analogy to the creation of the potential energy function. The stiness k is replaced
by the damping coecient c and the generalized displacements are replaced by the
generalized velocities. Hence, in the considered case, since the lower end of the damper
is motionless, the damping function is
D =
1
2
c
i
( x
i
)
2
+
1
2
c
ij
( x
j
x
i
)
2
+ (2.14)
The Lagranges equation with the damping function takes form
d
dt
(

q
m
T)

q
m
T +
V
q
m
+
D
q
m
= Q
m
m = 1, 2, ....M (2.15)
MODELLING 69
Introduction of the equations 2.11, 2.13 and 2.14 into equation 2.15 yields the equation
of the motion of the mass m
i
.
m
i
x
i
+ (c
i
+c
ij
) x
i
c
ij
x
j
+ (k
i
+k
ij
)x
i
k
ij
x
j
= k
i
y
i
(t) (2.16)
The inuence coecient method
m
j
m
i
F
j
x
j
x
i
x
ij
Figure 5
Let us consider the exible structure shown in Fig. 5. Let us assume that the masses
m
i
and m
j
can move along the coordinate x
i
and x
j
respectively. Let us apply along
the coordinate x
j
the force F
j
. Let x
ij
be the displacement of the system along the
coordinate x
i
caused by the force F
j
.
DEFINITION: The ratio
ij
=
x
ij
F
j
(2.17)
is called the inuence coecient
It can be easily proved that for any structure
ij
=
ji
(2.18)
If one apply forces along all I generalized coordinates x
i
along which the system
is allowed to move, the displacement along the i th coordinate, according to the
superposition principle, is.
x
i
=
I
X
j=1
ij
F
j
i = 1, 2, ......I (2.19)
These linear relationships can be written in the matrix form
x = F (2.20)
MODELLING 70
The inverse transformation permits to produce forces that act on the system along
the individual coordinates if the system is at an arbitrarily chosen position x.
F =
1
x (2.21)
The inverse matrix
1
is called stiness matrix and will be denoted by k.
k =
1
(2.22)
Hence, the force that act on the particle m
i
, according to equation 2.21 is
F
i
=
I
X
j=1
k
ij
x
j
(2.23)
Application of the third Newtons equation yields the equation of motion of the
particle i in the following form
m
i
x
i
+
I
X
j=1
k
ij
x
j
= 0 (2.24)
MODELLING 71
2.1.3 Problems
Problem 15
l
1 2
R
Figure 6
The disk 1 of radius R, and mass m is attached to the massless beam 2 of
radius r, length l and the Young modulus E as shown in Fig. 6 Develop equations of
motion of this system.
MODELLING 72
Solution.
l
1 2
R
y
y
z
F
d
M
d
Figure 7
The motion of the disk shown in Fig. 7 is governed by Newtons equations
m y = F
d
I
y
= M
d
(2.25)
In the above mathematical model
I =
1
4
mR
2
- moment of inertia of the disk about axis x
F
d
, M
d
- forces acting on the disk due to its interaction with the beam
The interaction forces F
d
and M
d
can be expressed as a function of the dis-
placements y and
y
by means of the inuence coecient method.
l
2
z
y
y
y
M
F
s
s
Figure 8
If the beam is loaded with force F
s
(see Fig. 8), the corresponding displace-
ments y and
y
are
y =
l
3
3EJ
F
s

y
=
l
2
2EJ
F
s
(2.26)
If the beam is loaded with force M
s
(see Fig. 8), the corresponding displacements y
and
y
are
y =
l
2
2EJ
M
s
,
y
=
l
EJ
M
s
(2.27)
MODELLING 73
Hence the total displacement along coordinates y and
y
are
y =
l
3
3EJ
F
s
+
l
2
2EJ
M
s
y
=
l
2
2EJ
F
s
+
l
EJ
M
s
(2.28)
or in matrix form

y
=

l
3
3EJ
l
2
2EJ
l
2
2EJ
l
EJ

F
s
M
s
(2.29)
where
J =
r
4
4
(2.30)
The inverse transformation yields the wanted forces as function of the displacements
F
s
M
s
=

l
3
3EJ
l
2
2EJ
l
2
2EJ
l
EJ
=

k
11
k
12
k
21
k
22

y
(2.31)
Since, according to the second Newtons law
F
d
M
d
F
s
M
s
(2.32)
the equation of motion takes the following form
m 0
0 I

y

y
k
11
k
12
k
21
k
22

y
(2.33)
Hence, the nal mathematical model of the system considered is
m x +kx = 0 (2.34)
where
m =
m 0
0 I
; k =
k
11
k
12
k
21
k
22
; x =
(2.35)
MODELLING 74
Problem 16
k
G
k
1
l1
2
l2
Figure 9
A rigid beam of mass m and the moments of inertia I about axis through
its centre of gravity G is supported by massless springs k
1
, and as shown in Fig. 9.
Produce equations of motion of the system.
MODELLING 75
Solution.
k
G
y
1
l1
l2
k2
y
y =y+
l
2
2
l
1
1
O
F
M
y =y-
Figure 10
The system has two degree of freedom. Let us then introduce the two coordi-
nates y and as shown in Fig. 10.
The force F and the moment M that act on the beam due to its motion along
coordinates y and are
F = y
1
k
1
y
2
k
2
= (y l
1
)k
1
(y +l
2
)k
2
= [(k
1
+k
2
)y + (k
2
l
2
k
1
l
1
)]
M = +y
1
k
1
l
1
y
2
k
2
l
2
= +(y l
1
)k
1
l
1
(y +l
2
)k
2
l
2
= [(k
2
l
2
k
1
l
1
)y + (k
1
l
2
1
+yk
2
l
2
2
) (2.36)
Hence, the generalized Newtons equations yield
m y = F = [(k
11
+k
2
)y + (k
2
l
2
k
1
l
1
)]
I = M = [(k
2
l
2
yk
1
l
1
)y + (k
1
l
2
1
+yk
2
l
2
2
) (2.37)
The matrix form of the system equations of motion is
m x +kx = 0 (2.38)
where
m =
m 0
0 I
; k =
k
1
+k
2
k
2
l
2
k
1
l
1
k
2
l
2
k
1
l
1
k
1
l
2
1
+yk
2
l
2
2
; x =
(2.39)
MODELLING 76
Problem 17
1
2
k k
R r 3
4
A
Figure 11
The link 1 of a mass m
1
, shown in Fig. 11), can move along the horizontal
slide and is supported by two springs 3 each of stiness k. The ball 2 of mass m
2
and a radius r and the massless rod 4 form a rigid body. This body is hinged to the
link 1 at the point A. All motion is in the vertical plane. Use Lagranges approach
to derive equations of small vibrations of the system about its equilibrium position.
I =
2
5
m
2
r
2
moment of inertia of the ball about axis through its centre of gravity.
MODELLING 77
Solution
1
2
k
k
R
r
3
x
y
x
o
G
r
G
Figure 12
The system has two degree of freedom and the two generalized coordinates x
and are shown in Fig. 12. The kinetic energy of the system T is equal to the sum
of the kinetic energy of the link 1 T
1
and the link 2 T
2
.
T = T
1
+T
2
=
1
2
m
1
x
2
+
1
2
m
2
v
2
G
+
1
2
I
2
(2.40)
The absolute velocity of the centre of gravity of the ball v
G
can be obtained by
dierentiation of its absolute position vector. According to Fig .12, this position
vector is
r
G
= i(x +Rsin ) +j( Rcos ) (2.41)
Hence
v
G
= r
G
= i( x +R cos ) +j(R sin ) (2.42)
The required squared magnitude of this velocity is
v
2
G
= ( x +R cos )
2
+ (R sin)
2
= x
2
+ 2 xR cos +R
2

2
(2.43)
Introduction of Eq. 2.43 into Eq. 2.40 yields the kinetic energy function of the system
as a function of the generalized coordinates x and .
T =
1
2
m
1
x
2
+
1
2
m
2
( x
2
+ 2 xR cos +R
2

2
) +
1
2
I
2
=
1
2
(m
1
+m
2
) x
2
+m
2
R x cos +
1
2
(m
2
R
2
+I)
2
(2.44)
The potential energy function is due the energy stored in the springs and the energy
due to gravitation.
V = 2
1
2
kx
2
m
2
gRcos (2.45)
MODELLING 78
In the case considered, the Lagranges equations can be adopted in the following form
d
dt
T
x
T
x
+
V
x
= 0
d
dt
+
V
= 0 (2.46)
The individual terms that appeare in the above equation are
d
dt
T
x
=
d
dt
((m
1
+m
2
) x +m
2
R cos ) =
= (m
1
+m
2
) x +m
2
R cos m
2
R
2
sin (2.47)
T
x
= 0 (2.48)
V
x
= 2kx (2.49)
d
dt
=
d
dt
m
2
R xcos + (m
2
R
2
+I)
=
= (m
2
R
2
+I) +m
2
R xcos m
2
R x sin (2.50)
T
= m
2
R x sin (2.51)
V
= m
2
gRsin (2.52)
Hence, according to Eq. 2.46, we have the following equations of motion
(m
1
+m
2
) x +m
2
R cos m
2
R
2
sin + 2kx = 0
(m
2
R
2
+I) +m
2
R xcos +m
2
gRsin = 0 (2.53)
For small magnitudes of x and , sin

= , cos

= 1,
2

= 0. Taking this into
account the linearized equations of motion are
(m
1
+m
2
) x +m
2
R + 2kx = 0
(m
2
R
2
+I) +m
2
R x +m
2
gR = 0 (2.54)
Their matrix form is
m x +kx = 0 (2.55)
where
m =
m
1
+m
2
m
2
R
m
2
R m
2
R
2
+I
, k =
2k 0
0 m
2
gR
, x =
(2.56)
MODELLING 79
Problem 18
k
k
k k
q
1
q
2
l
l
A
1
A
2
Figure 13
Two identical and uniform rods shown in Fig. 13, each of mass m and length l,
are joined together to form an inverse double pendulum. The pendulum is supported
by four springs, all of stiness k, in such way that its vertical position (q
1
= 0 and
q
2
= 0) is its stable equilibrium position. Produce equation of small vibrations of the
pendulum about this equilibrium position.
MODELLING 80
Problem 19
l
1
2
EI
4
3 4
R
l
3
GJ
o
Figure 14
The disk 1 of mass m
1
and radius R shown in Fig. 14, is fasten to the massless
and exible shaft 3. The left hand end of the massless and exible beam 4 is rigidly
attached to the disk 1. At its right hand side the particle 2 of m
2
is placed. Derive
equations for analysis of small vibrations of the system.
MODELLING 81
Problem 20
k
R
I
2
R
k
I
1
I
3
J
2
l
2
G
2 J
1
l
1
G
1
Figure 15
A belt gear was modelled as shown in Fig. 15. The shafts are assumed to
be massless and their length the second moment of inertia and the shear modulus is
denoted by l, J, and G respectively. The disks have moments of inertia I
1
, I
2
, and I
3
.
The belt is modelled as the spring of a stiness k. Derive the dierential equations
for the torsional vibrations of the system.
MODELLING 82
Problem 21
D
I
1
I
2
I
3
J
2
l
2
G
2
J
1
l
1
G
1
1
D
2
Figure 16
In Fig. 16 the physical model of a gear box is presented. Derive equations for
the torsional vibrations of the gear box. The shafts the gears are mounted on are
massless.
MODELLING 83
Problem 22
B
O
Y
X
C
l
l
l
l
l
l
A
1 2 3 4 5 6 7
Figure 17
Fig. 17 shows a mechanical system. Link 1 of the system is motionless with
respect to the inertial system of coordinates XY . The links 2 and 3 are hinged to
the link 1 at the point O. The links 4 and 5 join the links 2 and 3 with the collar 6.
The spring 7 has a stiness k and its uncompressed length is equal to 2l. The system
has one degree of freedom and its position may be determined by one generalized
coordinate . The links 4 5 and 6 are assumed to be massless. The links 2 and 3
can be treated as thin and uniform bars each of length 2l and mass m.
Derive equations of the small vibration of the system about its equilibrium
position.
MODELLING 84
Problem 23
m m m
T
l
4
l
4
l
4
l
4
Figure 18
Three beads, each of mass m are attached to the massless string shown in Fig.
18. The string has length l and is loaded with the tensile force T. Derive equation of
motion of the beads
MODELLING 85
Problem 24
R q
1
q
2
l
m
Figure 19
On the massless string of length l the ball of mass m and radius R is suspended
(see Fig. 19). Derive equation of motion of the system.
MODELLING 86
Problem 25
k
c
m
k
s1 s2
k
2
I
1
I
1
I
2
I
i i
R
Figure 20
Fig. 20 presents the physical model of a winch. The shafts of the torsional
stiness k
s1
and k
s2
as well as the gear of ratio i are massless. To the right hand end
of the shaft k
s2
the rotor of the moment of inertia I
2
is attached. The left hand end
of the shaft k
s1
is connected to the drum of the moment of inertia I
1
. The rope is
modelled as a massless spring of the stiness k. At its end the block of mass m is
fastened. The damper of the damping coecient c represents the damping properties
of the system.
Produce the dierential equation of motion of the system.
ANALYSIS OF MULTI-DEGREE-OF-FREEDOM SYSTEM 87
2.2 ANALYSIS OF MULTI-DEGREE-OF-FREEDOM SYSTEM
The analysis carried out in the previous section leads to conclusion that the mathe-
matical model of the linear multi-degree-of -freedom system is as follows
m x +c x +kx = F(t) (2.57)
where
m - matrix of inertia
c - matrix of damping
k - matrix of stiness
F(t)- vector of the external excitation
x- vector of the generalized coordinates
2.2.1 General case
In the general case of the multi-degree-of-freedom system the matrices c and k do not
necessary have to be symmetrical. Such a situation takes place, for example, if the
mechanical structure interacts with uid or air (oil bearings, atter of plane wings
etc.). Since the equation 2.57 is linear, its general solution is always equal to the sum
of the general solution of the homogeneous equation x
g
and the particular solution
of the non-homogeneous equation x
p
.
x = x
g
+x
p
(2.58)
The homogeneous equation
m x +c x +kx = 0 (2.59)
corresponds to the case when the excitation F(t) is not present. Therefore, its gen-
eral solution represents the free (natural) vibrations of the system. The particular
solution of the non-homogeneous equation 2.57 represents the vibrations caused by
the excitation force F(t). It is often refered to as the forced vibrations.
Free vibrations - natural frequencies- stability of the equilibrium position
To analyze the free vibrations let us transfer the homogeneous equation 2.59 to so
called state-space coordinates. Let
y = x (2.60)
be the vector of the generalized velocities. Introduction of Eq. 2.60 into Eq. 2.59
yields the following set of the dierential equations of rst order.
x = y
y = m
1
kx m
1
cy (2.61)
The above equations can be rewritten as follows
z = Az (2.62)
where
z =

x
y
, A =

0 1
m
1
k m
1
c
(2.63)
Solution of the above equation can be predicted in the form 2.64.
z = z
0
e
rt
(2.64)
Introduction of Eq. 2.64 into Eq. 2.62 results in a set of the homogeneous algebraic
equations which are linear with respect to the vector z
0
.
[A1r] z
0
= 0 (2.65)
The equations 2.65 have non-zero solution if and only if the characteristic determinant
is equal to 0.
|[A1r]| = 0 (2.66)
The process of searching for a solution of the equation 2.66 is called eigenvalue problem
and the process of searching for the corresponding vector z
0
is called eigenvector
problem. Both of them can be easily solved by means of the commercially available
computer programs.
The roots r
n
are usually complex and conjugated.
r
n
= h
n
i
n
n = 1.....N (2.67)
Their number N is equal to the number of degree of freedom of the system considered.
The particular solutions corresponding to the complex roots 2.67 are
z
n1
= e
hnt
(Re(z
0n
) cos
n
t Im(z
0n
) sin
n
t)
z
n2
= e
h
n
t
(Re(z
0n
) sin
n
t + Im(z
0n
) cos
n
t) n = 1.....N (2.68)
In the above expressions Re(z
0
n
) and Im(z
0
n
) stand for the real and imaginary part
of the complex and conjugated eigenvector z
0n
associated with the n
th
root of the
set 2.67 respectively. The particular solutions 2.68 allow to formulate the general
solution that approximates the system free vibrations..
z =[z
11
, z
12,
z
21
, z
22,
z
31
, z
32,
.....z
n1
, z
n2,
.........z
N1
, z
N2
] C (2.69)
As one can see from the formulae 2.68, the imaginary parts of roots r
n
represent
the natural frequencies of the system and their real parts represent rate of
decay of the free vibrations. The system with N degree of freedom possesses
N natural frequencies. The equation 2.69 indicates that the free motion of a multi-
degree-of-freedom system is a linear combination of the solutions 2.68.
A graphical interpretation of the solutions 2.68 is given in Fig. 21 for the
positive and negative magnitude of h
n
.The problem of searching for the vector of the
constant magnitudes C is called initial problem. In the general case, this problem is
dicult and goes beyond the scope of this lectures.
The roots 2.67 allow the stability of the system equilibrium position to be
determined.
If all roots r
n
of the equation 2.67 have negative real parts then the
equilibrium position of the system considered is stable.
If at least one root of the equation 2.67 has positive real part then
the equilibrium position of the system considered is unstable.
-20
-15
-10
-5
0
5
10
15
20
25
30
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 t
z
h>0
/
n
n
T = 2
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 t
z
/
n
n
T = 2
h<0
Figure 21
Forced vibrations - transfer functions
The response to the external excitation F(t) of a multi-degree-of-freedom system is
determined by the particular solution of the mathematical model 2.57.
m x +c x +kx = F(t) (2.70)
Let us assume that the excitation force F(t) is a sum of K addends. For the further
analysis let us assume that each of them has the following form
F
k
= F
k
o
cos(t +
k
o
) (2.71)
To facilitate the process of looking for the particular solution of equation 2.70, let us
introduce the complex excitation force by adding to the expression 2.71 the imaginary
part.
f
k
= F
k
o
cos(t +
k
o
) +iF
k
o
sin(t +
k
o
) (2.72)
The relationship between the complex excitation f
k
and the real excitation is shown
in Fig. 22. According to Eulers formula the complex excitation may be rewritten as
follows
f
k
= F
k
o
e
i(t+
k
o
)
= F
k
o
e
i
k
o
e
it
= f
k
o
e
it
(2.73)
Here, f
k
o
is a complex number that depends on the amplitude and phase of the external
excitation. Introduction of Eq. 2.73 into Eq. 2.70 yields
m x +c x +kx = f
o
e
it
(2.74)
Now, the particular solution of Eq. 2.74 can be predicted in the complex form 2.75
x
c
= ae
it
(2.75)
Introduction of Eq. 2.75 into Eq. 2.74 produces set of the algebraic equations which
are linear with respect to the unknown vector a.
2
m+ic +k
a = f
o
(2.76)
Its solution is
a =

2
m+ic +k
1
f
o
(2.77)
k
f
t
0
k
F
0
k
Re
Im
0
k
F cos( ) t+
0
k
k
F
t
0
k
t
0
k
F
Figure 22
Therefore, according to Eq. 2.75, the response of the system x
c
due to the complex
force f is
x
c
= (Re(a) +i Im(a))(cos t +i sin t) (2.78)
Response of the system x due to the real excitation F is represented by the real part
of the solution 2.78.
x = Re(a) cos t Im(a) sin t (2.79)
Motion of the system considered along the coordinate x
k
, according to 2.79 is
x
k
= x
k
o
cos(t +
k
) (2.80)
where
x
k
o
=
p
Re(a
k
)
2
+ Im(a
k
)
2

k
= arc tan
Im(a
k
)
Re(a
k
)
(2.81)
It is easy to see from 2.78 that the amplitude of the forced vibration x
k
o
is equal to the
absolute value of the complex amplitude a
k
, and its phase
k
is equal to the phase
between the complex amplitudes a
k
and the vector e
it
. This ndings are presented
in Fig. 23.
The complex matrix
2
m+ic +k
1
(2.82)
will be denoted by R(i) and it is called matrix of transfer functions. It transfers,
according to 2.77, the vector of the complex excitation f
o
e
it
into the vector of the
complex displacement x
c
= ae
it
.
x
c
= ae
it
= R(i)f
o
e
it
(2.83)
It is easy to see that the element R
pq
(i) of the matrix of transfer functions represents
the complex displacement (amplitude and phase) of the system along the coordinates
x
p
caused by the unit excitation 1e
it
along the coordinate x
q
. Example of three
elements of a matrix of the transform functions are presented in Fig. 24. The rst
two diagrams present the real and the imaginary parts of the complex transform
functions whereas the last two present its absolute value (amplitude) and phase.
k
f
t
0
k
F
0
k
Re
Im
0
k
F cos( ) t+
0
k
k
F
t
0
k
t
0
k
F
0
k
0
k
0
k
x
0
k
x
0
k
x cos( ) t+
0
k
k
x
k
a
Figure 23
-0.0003
-0.0002
-0.0001
0
0.0001
0.0002
0.0003
0.0004
0.0005
0.0006
500 1000 1500 frequency rad/s
transfer function m/N (imaginary part)
R(1,1)
R(1,2)
R(1,3)
-0.004
-0.003
-0.002
-0.001
0
0.001
0.002
0.003
transfer functions m/N (real parts)
R(1,1)
R(1,2)
R(1,3)
0
0.0005
0.001
0.0015
0.002
0.0025
0.003
0.0035
0
500 1000 1500
frequency rad/s
transfer function m/N (modulus)
R(1,1)
R(1,2)
R(1,3)
-4
-2
0
2
4
transfer functions m/N (phase)
R(1,1)
R(1,2)
R(1,3)
Figure 24
Experimental determination of the transfer functions
F(t)
x(t)
t
t
Fourier
transformation
F
x(
p
q
p
q
( )
)
i
i
R
i pq
(i
)
___
x(
p
)
i
F
q
( )
i
F
q
x
q
x
p
x
p
Figure 25
In order to produce the transfer function between the coordinate x
p
and the coor-
dinate x
q
(see Fig. 25) let us apply force F
q
(t) along the coordinate x
q
and record
it simultaneously with the system response x
p
(t) along the coordinate x
p
. Fourier
transformation applied to these functions
F
q
(i) =
Z
+
e
it
F
q
(t)dt
x
p
(i) =
Z
+
e
it
x
p
(t)dt (2.84)
yields the Fourier transforms in the frequency domain x
p
(i) and F
q
(i). The am-
plitude of the complex functions x
p
(i) and F
q
(i)
|x
p
(i)| =
q
Re(x
p
(i))
2
+ Im(x
p
(i))
2
|F
q
(i)| =
q
Re(F
q
(i))
2
+ Im(F
q
(i))
2
(2.85)
represents the amplitude of displacement and force respectively as a function of the
frequency . The corresponding phases are determined by the following formulae.
xp
= arctan
Re(x
p
(i))
Im(x
p
(i))
Fq
= arctan
Re(F
q
(i))
Im(F
q
(i))
(2.86)
These Fourier transforms allow the transfer function R
pq
(i) to be computed.
R
pq
(i) =
x
p
(i)
F
q
(i)
(2.87)
The above formula determines response of the system along coordinate x
p
caused by
the harmonic excitation F
q
q
.
x
p
(i) = R
pq
(i)F
q
(i) (2.88)
Since the systemconsidered is by assumption linear, the response along the coordinate
x
p
caused by set of forces acting along coordinates N coordinates x
q
, according to
the superposition principle, is
x
p
(i) =
q=N
X
q=1
R
pq
(i)F
q
(i) q = 1....N (2.89)
Application of the above described experimental procedure to all coordinates involved
in the modelling (p = 1....N) allows to formulate the matrix of the transfer functions
R
pq
(i).The relationship above can be rewritten in the following matrix form
x(i) = R
pq
(i)F(i) p = 1....N, q = 1....N (2.90)
2.2.2 Modal analysis - case of small damping
In the following analysis it will be assumed that the matrices m, c and k are square
and symmetrical. Size of these matrices is N N where N is the number of the
system degree of freedom. If the vector of the external excitation F(t) is equal to
zero, it is said that the system performs free vibrations. According to the above
denition the free vibrations are governed by the homogeneous set of equations
m x +c x +kx = 0 (2.91)
Free vibration of the undamped system - eigenvalue and eigenvector prob-
lem
If the damping is neglected the equation of the free vibrations is
m x +kx = 0 (2.92)
It is easy to see that
x = Xcos t (2.93)
is a particular solution of the equation 2.92. Indeed, introduction of Eq. 2.93 into
2.92 yields
(
2
m+k)Xcos t = 0 (2.94)
and the dierential equation 2.92 is fullled for any instant of time if the following
set of the homogeneous algebraic equations is fullled.
(
2
m+k)X = 0 (2.95)
In turn, the above set of equations has the non-zero solutions if and only if its char-
acteristic determinant is equal to zero
2
m+k
= 0 (2.96)
The above characteristic equation, for any physical system, has N positive roots with
respect to the parameter
2
. Hence, the parameter can take any of the following
values
1
,
2
,
3
, .....
n
, ....
N
(2.97)
As one can see from Eq. 2.93, these parameters have the physical meaning only for
positive values. They represent frequencies of the system free vibrations. They are
called natural frequencies. The number of dierent natural frequencies is therefore
equal to the number of degree of freedom. For each of the possible natural frequencies
n
the system of equations 2.95 becomes linearly dependent and therefore has innite
number of solution X
n
. Its follows that if X
n
is a solution of Eq. 2.95, the vector
C
n
X
n
(2.98)
where C
n
is arbitrarily chosen constant, is solution of the Eq. 2.95 too. The vector X
n
represents so called natural mode of vibration associated with the natural frequency
n
. It determines the shape that the system must possess to oscillate harmonically
with the frequency
n
.
For example, if a beam with four concentrated masses is considered (see Fig.
26) the vector X
n
contains four numbers
X
n
= [X
1n
, X
2n
, X
3n
, X
4n
]
T
(2.99)
If the system is deected according to the this vector and allowed to move with the
Tn
X
1n
X
2n
X
3n
X
4n
x
3n
x =X cos t
3n
n
t
Figure 26
initial velocity equal to zero, it will oscillate with the frequency
n
. There are four
such a natural modes and four corresponding natural frequencies for this system.
The problem of the determination of the natural frequencies is called eigen-
value problem and searching for the corresponding natural modes is called eigenvector
problem. Therefore the natural frequencies are very often referred to as eigenvalue
and the natural modes as eigenvectors.
Now, one can say that the process of determination of the particular solution
x
n
= X
n
cos
n
t (2.100)
of the equation 2.92 has been accomplished. There are N such particular solutions.
In similar manner one can prove that
x
n
= X
n
sin
n
t (2.101)
is a particular solution too. Since the solutions ?? and 2.101 are linearly independent,
their linear combination forms the general solution of the equation 2.92
x
n
=
N
X
n=1
(S
n
X
n
sin
n
t +C
n
X
n
cos
n
t) (2.102)
The 2N constants S
n
and C
n
should be chosen to satisfy the 2N initial conditions.
Properties of the natural modes.
Each eigenvector has to fulll the Eq. 2.95. Hence,
2
n
mX
n
+kX
n
= 0
2
m
mX
m
+kX
m
= 0 (2.103)
Primultiplying the rst equation by X
T
m
and the second equation by X
T
n
one can get
2
n
X
T
m
mX
n
+X
T
m
kX
n
= 0
2
m
X
T
n
mX
m
+X
T
n
kX
m
= 0 (2.104)
Since matrices m and k are symmetrical
X
T
m
kX
n
= X
T
n
kX
m
and X
T
m
mX
n
= X
T
n
mX
m
(2.105)
Now, primultiplying the rst equation of set 2.104 by -1 and then adding them to-
gether we are getting
(
2
n
2
m
)X
T
n
mX
m
= 0 (2.106)
Since for n 6= m (
2
n
2
m
) 6= 0,
X
T
n
mX
m
= 0 for n 6= m (2.107)
If n = m, since (
2
n

2
n
) = 0, the product X
T
n
mX
n
does not have to be equal to
zero. Let this product be equal to
2
n
X
T
n
mX
n
=
2
n
(2.108)
Division of the above equation by
2
n
yields
(
1
n
X
T
n
)m(
1
n
X
n
) = 1 (2.109)
But according to 2.98
1
n
X
n
is eigenvector too. Let us denot it by
n
n
=
1
n
X
n
(2.110)
The process of producing of the eigenvectors
n
is called normalization and the
eigenvector
n
is called normalized eigenvector or normalized mode. According to
2.109,
T
n
m
n
= 1 (2.111)
Taking into account Eqs 2.107 and 2.111 one can conclude that
T
n
m
m
=

0 if n 6= m
1 if n = m
(2.112)
It is said that eigenvectors
n
and
m
that fulll the above conditions are orthogonal
with respect to the inertia matrix m.
Owning to the above orthogonality condition, the second of the equations
2.104 yields
T
n
k
m
=

0 if n 6= m
2
n
if n = m
(2.113)
It means that the normalized modes are orthogonal with respect to the matrix of
stiness.
The modal modes
n
can be arranged in a square matrix of order N known
as the modal matrix .
=[
1
,
2
, .....
n
, ......
N
] where N is number of degrees of freedom (2.114)
It is easy to see that the developed orthogonality conditions yields
T
m = 1
T
k =
2
(2.115)
where
2
is a square diagonal matrix containing the squared natural frequencies
2
n
2
=
_
2
1
0 . 0 . 0
0
2
2
. 0 . 0
. . . . . .
0 0 .
2
n
. 0
. . . . . .
0 0 . 0 .
2
N
_
_
(2.116)
Normal coordinates - modal damping
Motion of any real systemis always associated with a dissipation of energy. Vibrations
of any mechanical structures are coupled with deections of the elastic elements.
These deections, in turn, cause friction between the particles the elements are made
of. The damping caused by such an internal friction and damping due to friction of
these elements against the surrounding medium is usually referred to as the structural
damping. In many cases, particularly if the system considered is furnished with
special devices design for dissipation of energy called dampers, the structural damping
can be omitted. But in case of absence of such devices, the structural damping has
to be taken into account. The structural damping is extremely dicult or simply
impossible to be predicted by means of any analytical methods. In such cases the
matrix of damping c (see Eq. 2.91) is assumed as the following combination of the
matrix of inertia m and stiness k with the unknown coecients and .
c =m+k (2.117)
This coecients are to be determined experimentally.
It will be shown that application of the following linear transformation
x = (2.118)
to the mathematical model
m x +c x +kx = F(t) (2.119)
results in its decoupling. Indeed, introduction of the transformation 2.118 into 2.119
yields
m +c +k = F(t) (2.120)
Primultiplying both sides of the above equation by
T
we obtain
T
m +
T
(m+k) +
T
k =
T
F(t) (2.121)
Taking advantage of the orthogonality conditions 2.115 we are getting set of indepen-
dent equations
1 + +
2
n
=
T
F(t) (2.122)
where
1 - the unit matrix
2
n
and = (1+
2
n
) - diagonal matrices
Hence, each equation of the above set has the following form

n
+ 2
n
n

n
+
2
n
n
=
T
n
F(t) n = 1, 2, ...N (2.123)
The coecients
n
= (+
2
n
)/2
n
are often referred to as the modal damping ratio.
Solution of each of the above equations can be obtained independently and
according to the discussion carried out in the rst chapter (page 29, Eq. 1.46) can
be written as follows
n
= e
n
t
(C
sn
sin
dn
t +C
cn
cos
dn
t) +
pn
(2.124)
where
dn
=
n
p
1
2
n
and
pn
stands for the particular solution of the non-
homogeneous equation 2.123. Problem of determination of this particular solution is
considered in the next section.
Introduction of the solutions 2.124 into equation 2.118 yields motion of the
system along the physical coordinates x.
Response to the harmonic excitation - transfer functions
Let us solve the Eq. 2.123 for response of the system due to the harmonic excitation
along coordinate x
q
. In this case the right hand side of the equation 2.123 takes form
T
n
F(t) =
T
n
_
_
0
.
F
q
e
it
.
0
_
_
=
qn
F
q
e
it
(2.125)
Hence

n
+ 2
n
n

n
+
2
n
n
=
qn
F
q
e
it
n = 1, 2, ...N (2.126)
Therefore
n
=

qn
F
q
2
n
2
+ 2
n
n
i
e
it
n = 1, 2, ...N (2.127)
Since
x = (2.128)
response along coordinate x
p
x
p
= e
it
N
X
n=1
pn
qn
F
q
2
n
2
+ 2
n
n
i
(2.129)
transfer function between coordinate p and the others
x
p
F
q
e
it
=
N
X
n=1
pn
qn
2
n
2
+ 2
n
n
i
q = 1, 2, .......N (2.130)
R
pq
(i) =
x
p
F
q
e
it
=
P
N
n=1

pn
qn
((
2
n
2
)2
n
n
i)
(
2
n
2
)
2
+4
2
n
2
n
2
=
=
P
N
n=1
pnqn(
2
n
2
)
(
2
n
2
)
2
+4
2
n
2
n
2
+
2pnqn
n
ni
(
2
n
2
)
2
+4
2
n
2
n
q = 1, 2, .......N
(2.131)
if
=
n
R
pq
(i)
=

pn
qn
(
2
n
2
)
4
2
n
2
n
2
+

pn
qn
i
2
n
q = 1, 2, .......N (2.132)
Determination of natural frequencies and modes from the transfer func-
tions
The transfer functions R
pq
(i) can be easily obtained by means of a simple experi-
ment (see page 94). They allow the natural frequencies, natural modes and the modal
damping to be identied. It can be seen from the equation 2.132 that the real part of
the transfer function R
pq
(i) is equal to zero for the frequency equal to the natural
frequency
n
. Hence the zero-points of the real part of the transfer functions deter-
mine the system natural frequencies. From the same equation it is apparent that the
imaginary parts corresponding to
=
n
and measured for dierent q = 1, 2, .......N,
yield the natural modes with accuracy to the constant magnitude C =
2
n
2
n
np
nq
= C Im(R
pq
(i
n
)) q = 1, 2, .......N (2.133)
Alternatively, The natural frequencies and the natural modes can be extracted from
diagrams of the magnitudes and phases of the transfer function.
The phase since the real part of the transfer function is equal to zero for
=
n
is equal to 90
o
= arctan
Im(R
pq
(i
n
))
Re (R
pq
(i
n
))
= arctan = 90
o
(2.134)
This property allows the natural frequencies to be determined.
Since the real part of the transfer function is equal to zero for =
n
, its mod-
ulus is equal to the imaginary part. Therefore the modulus of the transfer functions
corresponding =
n
determine the natural modes
|R
pq
(i)| =

pn
2
n
2
n
qn
(2.135)
An example of extracting the natural frequency and the corresponding natural mode
from the transfer functions is shown in Fig. 27

0
0.00005
0.0001
0.00015
0.0002
0.00025
1500 1600 1700 1800 frequency rad/s
transfer functions m/N (modulus)
R(1,1)
R(1,2)
R(1,3)
-4
-3
-2
-1
0
1
2
3
4
1500 1600 1700 1800 frequency rad/s
R(1,1)
R(1,2)
R(1,3)
/2
/2
1
2
3
natural frequency
natural mode
Figure 27
Determination of the natural frequencies and the natural modes from the
transfer functions is referred to as modal analysis.
2.2.3 Kinetic and potential energy functions - Dissipation function
In this section the kinetic energy function, the potential energy function and the
dissipation function are formulated for a linear system governed by the equation
m x +c x +kx = F(t) (2.136)
where the matrices m, c and k are symmetric and positive denite matrices.
Kinetic energy function
Let us consider function
T =
1
2
x
T
m x x ={ x
1
, ...... x
n
........ x
N
}
T
(2.137)
Performing the matrix multiplication we are getting
T =
1
2
{ x
1
, .. x
n
.. x
N
}
_
_
P
m=N
m=1
m
1m
x
m
.................
P
m=N
m=1
m
nm
x
m
..................
P
m=N
m=1
m
Nm
x
m
_
_
=
1
2
n=N
X
n=1
x
n
m=N
X
m=1
m
nm
x
m
!
(2.138)
=
1
2
n=N
X
n=1
m=N
X
m=1
m
nm
x
n
x
m
If this function is positive denite (is always positive and is equal to zero if and only
if all variables x
n
are equal to zero) the corresponding matrix m is called positive
denite matrix.
If T is the kinetic energy function, according to Lagranges equations should
be
d
dt
T
x
n
T
x
n
= {m
n1
, ..m
nn
..m
nN
}
_
_
x
1
....
x
n
.....
x
N
_
_
(2.139)
Let us prove that the function 2.137 fullls the requirement 2.139.
d
dt
T
x
n
T
x
n
=
d
dt
_
_
_
_
_
_
1
2
_
_
_
_
_
_
{0, 0..1..0}
_
_
P
m=N
m=1
m
1m
x
m
.................
P
m=N
m=1
m
nm
x
m
..................
P
m=N
m=1
m
Nm
x
m
_
_
+ { x
1
, .. x
n
.. x
N
}
_
_
m
1n
.......
m
mn
.......
m
NN
_
_
_
_
_
_
_
_
_
_
_
_
_
_
=
=
d
dt
1
2
m=N
X
m=1
m
nm
x
m
+
m=N
X
m=1
m
mn
x
m
!!
(2.140)
Since m
nm
= m
mn
d
dt
T
x
n
T
x
n
=
d
dt
1
2
2
m=N
X
m=1
m
nm
x
m
!
=
d
dt
m=N
X
m=1
m
nm
x
m
!
(2.141)
=
m=N
X
m=1
m
nm
x
m
= {m
n1
, ......m
nn
........m
nN
}
_
_
x
1
....
x
n
.....
x
N
_
_
Now we may conclude that the function 2.137 is the kinetic energy function if the
matrix m is symmetric and positive denite.
Potential energy function
Let us consider function
V =
1
2
x
T
kx x ={x
1
......x
n
........x
N
}
T
(2.142)
Performing the matrix multiplication we are getting
V =
1
2
{x
1
, x
2
..x
n
..x
N
}
_
_
P
m=N
m=1
k
1m
x
m
.................
P
m=N
m=1
k
nm
x
m
..................
P
m=N
m=1
k
Nm
x
m
_
_
=
1
2
n=N
X
n=1
x
n
m=N
X
m=1
k
nm
x
m
!
(2.143)
=
1
2
n=N
X
n=1
m=N
X
m=1
k
nm
x
n
x
m
If V is the potential energy function, it must be positive denite and according to
Lagranges equations should fullls the following relationship
V
x
n
= {k
n1
, ......k
nn
........k
nN
}
_
_
x
1
....
x
n
.....
x
N
_
_
1 (2.144)
Let us prove that the function 2.142 fullls the requirement 2.144.
V
x
n
=
1
2
_
_
_
_
_
_
{0, 0..1..0}
_
_
P
m=N
m=1
k
1m
x
m
.................
P
m=N
m=1
k
nm
x
m
..................
P
m=N
m=1
k
Nm
x
m
_
_
+ {x
1
, ..x
n
..x
N
}
_
_
k
1n
.......
k
mn
.......
k
NN
_
_
_
_
_
_
_
_
=
=
1
2
m=N
X
m=1
k
nm
x
m
+
m=N
X
m=1
k
mn
x
m
!
(2.145)
Since m
nm
= m
mn
V
x
n
=
1
2
2
m=N
X
m=1
k
nm
x
m
=
m=N
X
m=1
k
nm
x
m
= {k
n1
, ......k
nn
........k
nN
}
_
_
x
1
....
x
n
.....
x
N
_
_
(2.146)
Now we may conclude that the function 2.142 is the kinetic energy function if the
matrix k is symmetric and positive denite.
Dissipation function
It is easy to notice, having in mind the previous consideration, that the function
D =
1
2
x
T
c x x ={ x
1
, ...... x
n
........ x
N
}
T
(2.147)
fullls the following relationship
D
x
n
= {c
n1
, ......c
nn
........c
nN
}
_
_
x
1
....
x
n
.....
x
N
_
_
(2.148)
It follows that if the matrix of damping is symmetrical and positive denite, such a
damping can be included in the Lagranges equation in the following way
d
dt
T
x
n
T
x
n
+
V
x
n
+
D
x
n
= Q
n
(2.149)
The function D is called dissipation function. It must be noted that the dissi-
pation function does not represent the dissipation energy.
The damping forces, in a general case, are not conservative and have to be
included in the generalized force Q
n
.
2.2.4 Problems
Problem 26
m
1 J
1
l
1
E
1
J
2
E
2
m
2
k
l
2
2
l
2
2
A
c
y
B
Figure 28
The point A of the system shown in Fig. 28 moves according to the following
equation
y = A
1
sin(f
1
t) +A
2
sin(f
2
t) (2.150)
where A
1
and A
2
are amplitudes of this motion and f
1
and f
2
are the corresponding
frequencies.
Produce
1. the dierential equations of motion
2. the natural frequencies
3. the steady state motion of the system due to the kinematic excitation y
4. the exciting force at the point A required to maintain the steady state
motion
5. the reaction force and the reaction moment at the point B. .
Given are:
l
1
= 1m E
1
= 0.2 10
12
N/m
2
J
1
= 1 10
8
m
4
m
1
= 10kg
l
2
= 2m E
2
= 0.2 10
12
N/m
2
J
2
= 1 10
8
m
4
m
2
= 20kg
k = 10000N/m
c = 100Ns/m
A
1
= 0.01m f
1
= 30rad/s
A
2
= 0.01m f
2
= 35rad/s
Solution
1. The dierential equations of motion
m
1 J
1
l
1
E
1
J
2
E
2
m
2
y
1 k
l
2
2
l
2
2
A
c
y
B
y
2
Figure 29
Utilization of the Newton-Euler approach for modelling of the system shown
in Fig. 29 allows to develop its mathematical model.
m
1
y
1
= k
1
y
1
ky
1
+ky
2
m
2
y
2
= k
2
y
2
ky
2
+ky
1
c y
2
+c y (2.151)
Its matrix form is
m
1
m
2

y
1
y
2
0
c

y
1
y
2
k +k
1
k
k k +k
2

y
1
y
2
=

0
c y
(2.152)
or shorter
m y +c y +ky = F(t) (2.153)
where
m =
m
1
0
0 m
2
; c =
0 0
0 c
; k =
k +k
1
k
k k +k
2
; F(t) =

0
c y
(2.154)
Taking into consideration Eq. 2.150, the excitation c y is
c y = cA
1
f
1
cos(f
1
t) +cA
2
f
2
cos(f
2
t) = a
1
cos(f
1
t) +a
2
cos(f
2
t) (2.155)
where
a
1
= cA
1
f
1
; a
2
= cA
2
f
2
Introduction of Eq. 2.155 into the equation of motion 2.153 yields
m y +c y +ky = F
1
(t) +F
2
(t) (2.156)
where
F
1
(t) =

0
a
1
cos(f
1
t)
; F
2
(t) =

0
a
2
cos(f
2
t)
(2.157)
For the given numerical data the stiness of the beam 1 at the point of attachment
of the mass 1 is
k
1
=
3E
1
J
1
l
3
1
=
3 0.2 10
12
1 10
8
1
3
= 6000N/m (2.158)
The stiness of the beam 2 at the point of attachment of the mass 2 is
k
2
=
48E
2
J
2
l
3
2
=
48 0.2 10
12
1 10
8
2
3
= 12000N/m (2.159)
Hence
m =
10 0
0 20
; c =
0 0
0 100
; k =
16000 10000
10000 22000
F
1
(t) =

0
30 cos(30t)
; F
2
(t) =

0
35 cos(35t)
(2.160)
2. Free motion - the natural frequencies
To analyze the free vibrations let us transfer the homogeneous equation 2.156
to the state-space coordinates. The substitution
w = y (2.161)
results in the following set of equations
z = Az (2.162)
where
z =

y
w
, A =

0 1
m
1
k m
1
c
=
_
_
0 0 1 0
0 0 0 1
1600.0 1000 0 0
500 1100 0 5
_
_
(2.163)
Solution of the eigenvalue problem yields the following complex roots
1
= 1. 6741 24. 483i
2
= 0. 8259 45. 734i (2.164)
For underdamped system the imaginary part of the above roots represents the natural
frequency of the damped system. The real part indicates the rate of decay of the free
vibrations.
Solution of the eigenvector problem produces the following complex vectors.
Re z
01
=
_
_
1. 6392 10
2
1. 7637 10
2
. 38263
. 35302
_
_
, Imz
01
=
_
_
1. 4508 10
2
1. 3213 10
2
. 37705
. 40969
_
_
Re z
02
=
_
_
1. 9755 10
2
1. 0154 10
2
. 26033
5. 9162 10
2
_
_
, Imz
02
=
_
_
6. 049 10
3
1. 477 10
3
. 90847
. 46561
_
_
(2.165)
According to 2.68, the particular solutions are
z
11
= e
h
1
t
(Re(z
01
) cos
1
t Im(z
01
) sin
1
t) =
= e
1. 6741t
_
_
_
_
_
_
1. 6392 10
2
1. 7637 10
2
. 38263
. 35302
_
_
cos 24. 483t
_
_
1. 4508 10
2
1. 3213 10
2
. 37705
. 40969
_
_
sin24. 483t
_
_
_
_
z
12
= e
h
1
t
(Re(z
01
) sin
1
t + Im(z
01
) cos
1
t) =
= e
1. 6741t
_
_
_
_
_
_
1. 6392 10
2
1. 7637 10
2
. 38263
. 35302
_
_
sin24. 483t +
_
_
1. 4508 10
2
1. 3213 10
2
. 37705
. 40969
_
_
cos 24. 483t
_
_
_
_
z
21
= e
h
2
t
(Re(z
02
) cos
2
t Im(z
02
) sin
2
t) =
= e
0. 8259t
_
_
_
_
_
_
1. 9755 10
2
1. 0154 10
2
. 26033
5. 9162 10
2
_
_
cos 45. 734t
_
_
6. 049 10
3
1. 477 10
3
. 90847
. 46561
_
_
sin 45. 734t
_
_
_
_
z
22
= e
h
2
t
(Re(z
02
) sin
2
t + Im(z
02
) cos
2
t) =
= e
0. 8259t
_
_
_
_
_
_
1. 9755 10
2
1. 0154 10
2
. 26033
5. 9162 10
2
_
_
sin45. 734t +
_
_
6. 049 10
3
1. 477 10
3
. 90847
. 46561
_
_
cos 45. 734t
_
_
_
_
(2.166)
The two rst rows in the above solutions represent displacement along the coordinates
y
1
and y
2
respectively. The two last rows represents the generalized velocities along
the coordinates y
1
and y
2
. Example of the motion along the coordinate y
1
, associated
with the particular solution z
11
(y
111
) and z
21
(y
211
)
y
111
= e
1. 6741t
(1. 6392 10
2
cos 24. 483t + 1. 4508 10
2
sin 24. 483t)
y
211
= e
0. 8259t
(1. 9755 10
2
cos 45. 734t + 6. 049 10
3
sin 45. 734t) (2.167)
are presented in Fig 30.
t[s]
4 3
2 1
0
0.01
-0.01
-0.02
y
111
[m]
t[s]
4
3 2
0
0.01
-0.01
-0.02
y
211
[m]
1
Figure 30
3. The steady state motion of the system due to the kinematic excitation
According to the given data, motion of the point A is
y = 0.01 sin(30 t) + 0.01 sin(35 t) (2.168)
t[s] 1.5 1 0.5
y[m]
0.01
0
-0.01
-0.02
Figure 31
The time history diagram of this motion is given in Fig. 31
The particular solution y, which represents the forced vibration, according to
the superposition rule, is
y = y
1
+y
2
(2.169)
where y
1
is the particular solution of the equation 2.170
m y +c y +ky = F
1
(t) (2.170)
and y
2
is the particular solution of the equation 2.171
m y +c y +ky = F
2
(t) (2.171)
To produce the particular solution of the equation 2.170 let us introduce the complex
excitation
F
c
1
(t) =

0
a
1
cos(f
1
t) +ia
1
sin(f
1
t)
=

0
a
1
e
if
1
t
=

0
a
1
e
if
1
t
=
= F
10
e
if
1
t
=

0
30
e
i30t
(2.172)
Hence the equation of motion takes form
m y +c y +ky = F
10
e
if
1
t
(2.173)
Its particular solution is
y
c
1
= y
c
10
e
if
1
t
(2.174)
where
y
c
10
= (f
2
1
m+if
1
c +k)
1
F
10
=
=

30
2
10 0
0 20
+ 30i
0 0
0 100
16000 10000
10000 22000
0
30
=
=

.00 384 .00 112i
2. 688 10
3
7. 84 10
4
i
(2.175)
The motion of the system, as the real part of 2.174 is
y
1
= Re
.00 384 .00 112i

2. 688 10
3
7. 84 10
4
i
e
i30t
=

.00 384 cos 30t +.00 112 sin 30t
2. 688 10
3
cos 30t + 7. 84 10
4
sin 30t
Similarly, one can obtained motion due to the excitation F

2
(t)
y
c
20
= (f
2
2
m+if
2
c +k)
1
F
20
=
=

35
2
10 0
0 20
+ 35i
0 0
0 100
16000 10000
10000 22000
0
35
=
=

3. 1546 10
3
3. 7855 10
4
i
1. 183 10
3
1. 4196 10
4
i
(2.176)
Hence
y
2
= Re
3. 1546 10
3
3. 7855 10
4
i
1. 183 10
3
1. 4196 10
4
i
e
i35t
=

3. 1546 10
3
cos 35t + 3. 7855 10
4
sin 35t
1. 183 10
3
cos 35t + 1. 4196 10
4
sin35t
(2.177)
The resultant motion of the system due to both components of excitation is
y = y
1
+y
2
=

.00 384 cos 30t +.00 112 sin30t
2. 688 10
3
cos 30t + 7. 84 10
4
sin30t
+
+
3. 1546 10
3
cos 35t + 3. 7855 10
4
sin35t
1. 183 10
3
cos 35t + 1. 4196 10
4
sin35t
.00 38 cos 30t +.00 11 sin 30t 3. 15 10

3
cos 35t + 3. 78 10
4
sin35t
2. 6 10
3
cos 30t + 7. 8 10
4
sin 30t 1. 1 10
3
cos 35t + 1. 41 10
4
sin 35t
(2.178)
This resultant motion of the system along the coordinates y
1
and y
2
, computed
according to the equation 2.178, is shown in Fig. 32 and 33 respectively.
t[s]
1.5
1
0.5
y
1
[m]
0.005
0
-0.005
-0.01
Figure 32
t[s]
1.5 1
0.5
0.005
0
-0.005
-0.01
y
2
[m]
Figure 33
4. The exciting force at the point A required to maintain the steady
state motion
m
2
A
c
y
y
2
y
2
y
.
.
Figure 34
To develop the expression for the force necessary to move the point A according
to the assumed motion 2.168, let us consider the damper c shown in Fig. 34. If the
point A moves with the velocity y and in the same time the mass m
2
moves with the
velocity y
2
, the relative velocity of the point A with respect to the mass m
2
is
v = y y
2
(2.179)
Therefore, to realize this motion, it is necessary to apply at the point A the following
force
F
A
= c ( y y
2
) (2.180)
Hence, according to the equation 2.168 and 2.178 we have
F
A
= 100(
d
dt
(0.01 sin 30t + 0.01 sin35t) +
d
dt
(2. 6 10
3
cos 30t + 7. 8 10
4
sin 30t+
1. 1 10
3
cos 35t + 1. 4 10
4
sin 35t)) =
= 27. 648 cos 30t + 34. 503 cos 35t 8. 064 sin 30t 4. 1405 sin 35t[N]
(2.181)
Diagram of this force is presented in Fig.35
t[s] 1.5
1 0.5
F
A
[N]
40
20
0
-20
-40
-60
Figure 35
5. The reaction force and the reaction moment at the point B.
R
B
M
B
P
J
1
l
1
E
1
B
y
1
Figure 36
According to Fig. 36
R
B
= P
M
B
= Pl
1
(2.182)
where P is dependent on the instantaneous displacement y
1
. This relationship is
determined by the formula 2.158
P = k
1
y
1
=
3E
1
J
1
l
3
1
y
1
=
3 0.2 10
12
1 10
8
1
3
= 6000y
1
(2.183)
The motion along the coordinate y
1
is determined by the function 2.178
y
1
= .00 38 cos 30t +.00 11 sin30t3. 1510
3
cos 35t +3. 7810
4
sin35t (2.184)
Hence
R
B
=6000
.00 38 cos 30t+.00 11 sin30t3. 1510

3
cos 35t + 3. 7810
4
sin 35t
M
B
=60001
.00 38 cos 30t+.00 11 sin 30t3. 1510

3
cos 35t+3. 7810
4
sin 35t
(2.185)
Problem 27
The link 1 of a mass m
1
, shown in Fig. 37, can move along the horizontal
slide and is supported by two springs 3 each of stiness k. The ball 2 of mass m
2
and a radius r is hinged to the link 1 at the point A by means of the massless and
rigid rod 4. All motion is in the vertical plane. The equation of motion, in terms of
the coordinates x and (see Fig. 38) have been formulated in page 78 to be
m x +kx = 0 (2.186)
where
m =
m
1
+m
2
m
2
R
m
2
R m
2
R
2
+I
, k =
2k 0
0 m
2
gR
, x =
, I =
2
5
m
2
r
2
(2.187)
At the instant t = 0, the link 1 was placed to the position shown in Fig. 39 and
released with the initial velocity equal to zero.
For the following data:
m
1
= 2 kg
m
2
= 1 kg
R = 0.1 m
r = .05 m
k = 1000 N/m
a = 0.01 m
Produce:
1. the natural frequencies of the system
2. the normalized natural modes
3. the dierential equation of motion in terms of the normal coordinates
4. the equation motion of the system along the coordinates x and due to
the given initial conditions.
1
2
k k
R
r 3
4
A
Figure 37
1
2
k
k
R
r
3
x
y
x
o
G
Figure 38
1
2
k
k
R
r
3
x
y
a
o
Figure 39
Solution
1. The natural frequencies and the natural modes
According to the given numerical data the moment of inertia of the ball, the
inertia matrix and the stuness matrix are
I =
2
5
1 0.05
2
= 0.001 kgm
2
m =

2 + 1 1 0.1
1 0.1 1 0.1
2
+ 0.001
=

3.0 . 1
. 1 .0 11
(2.188)
k =

1000 0
0 1 10 0.1
=

1000.0 0
0 1.0
According to 2.95 (page 96) one can write the following set of equations
(
2
n
m+k)X = 0 (2.189)
where stands for the natural frequency and X is the corresponding natural mode.
Hence for the given numerical data we are getting
3.0
2
n
+ 1000.0 . 1
2
n
. 1
2
n
.0 11
2
n
+ 1.0

X
=

0
0
(2.190)
This set of equations has non-zero solution if and only if its determinant is equal to
zero. Hence the equation for the natural frequencies is.
3.0
2
n
+ 1000.0 . 1
2
n
. 1
2
n
.0 11
2
n
+ 1.0
= .0 23
4
n
14.0
2
n
+ 1000.0 = 0 (2.191)
Its roots:

22. 936 22. 936 9. 091 3 9. 091 3

yield the wanted natural frequencies
1
= 9. 0913
2
= 22. 936 [s
1
] (2.192)
For
n
=
1
= 9. 0913 the equations 2.190 become linearly dependent. Therefore,
one of the unknown can be chosen arbitrarily (e.g. X
1
= 1) and the other may be
produced from the rst equation of the set 2.190.
X
1
= 1
3.0X
1
2
1
+ 1000.0X
1
. 1
2
1
1
= 0 (2.193)
1
=
1
. 1 9.0913
2
3.0 9.09
2
+ 1000.0) = 90.99
These two numbers form the rst mode of vibrations corresponding to the rst
natural frequency
1
.Similar consideration, carried out for the natural frequency
2
= 22. 936,yields the second mode.
X
2
= 1
2
=
1
. 1 22.936
2
3.0 22.936
2
+ 1000.0
= 10.991 (2.194)
Now, one can create the modal matrix
X = [X
1
, X
2
] =

1 1
90.99 10.991
(2.195)
In this case the modal matrix has two eigenvectors X
1
and X
2
.
X
1
=

1
90.99
; X
2
=

1
10.991
(2.196)
2. Normalization of the natural modes
According to 2.108 the normalization factor is
X
T
n
mX
n
=
2
n
(2.197)
Hence
2
1
=

1 90. 99

3.0 . 1
. 1 .0 11

1
90.99
= 112. 27
1
=
112. 27 = 10. 596 (2.198)

Division of the eigenvector X
1
by the factor
1
yields the normalized mode
1
.
1
=
1
10. 596
1
90.99
=

9. 4375 10
2
8. 5872
(2.199)
Similar procedure allows the second normalized mode to be obtained
2
2
=

1 10.991

3.0 . 1
. 1 .0 11

1
10.991
= 2. 1306
2
=
2. 1306 = 1. 4597
2
=
1
1. 4597
1
10.991
=

. 68507
7. 5296
(2.200)
These two vectors forms the normalized modal matrix .
=
9. 4375 10
2
. 68507
8. 5872 7. 5296
(2.201)
The normalized eigenvectors must be orthogonal with respect to both the inertia
matrix and the stiness matrix. Indeed.
T
m =
=

9. 4375 10
2
8. 5872
. 68507 7. 5296

3.0 . 1
. 1 .0 11

9. 4375 10
2
. 68507
8. 5872 7. 5296
=

1 0
0 1
(2.202)
and
T
k =
=

9. 4375 10
2
8. 5872
. 68507 7. 5296

1000 0
0 1 10 0.1

9. 4375 10
2
. 68507
8. 5872 7. 5296
=

82. 647 0
0 526. 02
=

(9. 091)
2
0
0 (22. 935)
2
=

2
1
0
0
2
2
(2.203)
3. The dierential equation of motion in terms of the normal coordinates
Introducing the substitution 2.118
x = (2.204)
that in the case considered has the following form
= =
(2.205)
into 2.186 and premultiplying them from the left hand side by
T
we are getting the
dierential equations of motion in terms of the normal coordinates .
(
T
m) +(
T
k) = 0 (2.206)
Taking advantage of the orthogonality conditions, the equations of motion are of the
following form

1 0
0 1
(9. 091)
2
0
0 (22. 935)
2
= 0 (2.207)
or

1
+ (9. 091)
2
1
= 0

2
+ (22. 935)
2
2
= 0 (2.208)
The general solution of the above set of the dierential equations, according to 1.36
is
1
=
v
01
1
sin
1
t +
01
cos
1
t
2
=
v
02
2
sin
2
t +
02
cos
2
t (2.209)
Where
01
and
02
stand for the initial position whereas v
01
and v
02
stand for the
initial velocity of the system along the normal coordinates. These initial conditions
must be formulated along the normal coordinates. It can be obtained by transforming
the initial conditions from the physical coordinates to the normal coordinates.

01
02
=
1
X
o
=
1
a
0
=

1.142 .1039
1.3024 1.4313 10
2

0.01
0
=

.01142
1.3024 10
2
(2.210)
v
01
v
02
=

0
0
(2.211)
Introduction of the above initial conditions into the equations 2.209 results in motion
of the system along the normal coordinates
1
=
01
cos
1
t = .01142 cos 9.091t
2
=
02
cos
2
t = 1.3024 10
2
cos 22.935t (2.212)
4. The equations of motion of the system along the coordinates x and
To produce equation of motion along the physical coordinates, one has to
transform the motion along the normal coordinates beck to the physical ones. Hence,
using the relationship 2.204, we are getting
= =
9. 4375 10
2
. 68507
8. 5872 7. 5296

.01142 cos 9.091t
1.3024 10
2
cos 22.935t
=
=

1. 0778 10
3
cos 9. 091t + 8. 9224 10
3
cos 22. 935t
9. 8066 10
2
cos 9. 091t 9. 8066 10
2
cos 22. 935t
(2.213)
This motion is presented in Fig. 40 and 41
-0.01
-0.005
0
0.005
0.01
0.5 1 1.5 t[s]
X[m]
Figure 40
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.5 1 1.5 t[s]
[rad]
Figure 41
Problem 28
1
A
B
C
D
F=F cos t
o

2
L
l
3
k
c
M, I
A
E,J
Figure 42
The rigid beam 1 of mass M, length L and the moment of inertia about
its point of rotation I
A
, is supported by means of the spring of stiness k and the
damper of the damping coecient c as shown in Fig. 42. The beam 2 is massless
and the Youngs modulus E and the second moment of area J determine its dynamic
properties. Its end D is xed and the particle 3 of mass m is attached to the end C.
Derive an expression for the xing moment and the xing force at the point
D due to the exciting force F that is applied to the system at the point B.
Problem 29
1 2
O
1
O
2
a
b
k
1
k
2
k
c
F cos t
1 1
F cos t
2 2
A
I
O1
m
1
I
O2
m
2
G
1
G
2
s
1 s
2
Figure 43
The two rods, 1 and 2, are suspended in the vertical plane as shown in Fig. 43.
Their mass and their moment of inertia about their points of rotation are respectively
m
1
, I
O1
, and m
2
, I
O2
. These rods are connected to each other by means of springs
of the stiness k, k
1
, k
2
and as well as the damper of the damping coecient c. The
centres of gravity of these rods are denoted by G
1
and G
2
respectively. Vibrations of
the systemare excited by the two harmonic forces of amplitudes F
1
, F
2
and frequencies
1
and
2
.
Produce
1. the dierential equation of the small vibrations of the system in the matrix
form
2. the expression for the forced vibrations of the rods
3. the expression for the dynamic reaction at the point A.
Problem 30
A
G
B
C
c
1
k
1
k
2
c
2
m
2
D
m
1
,I
A
M
m
a
b
h

1
2
3
Figure 44
The rigid beam 1 (see Fig. 44) is hinged at the point A and is supported at
the point C by means of the spring of stiness k
1
and the damper of the damping
coecient c
1
. Its mass and its moment of inertia about A are m
1
and I
A
respectively.
The motor 3 is mounted on this beam. It can be approximated by a particle of the
mass M that is concentrated at the point G that is located by the dimensions h and
a. The rotor of this motor rotates with the constant velocity . Its mass is equal
to m and its unbalance is . To attenuate the vibrations of the beam the block 2
of mass m
2
was attached. The damping coecient of the damper between the beam
and the block is denoted by c
2
and the stiness of the supporting spring in denoted
by k
2
.
1. Produce the dierential equation of motion of the system and present it in
the standard matrix form.
2. Produce the expression for the interaction forces at the point A and D.
Problem 31
l l
C D
A
B
l
k

k

F
G
Figure 45
Three uniform platforms each of the length l, the mass m and the moment
of inertia about axis through its centre of gravity I
G
are hinged together to form a
bridge that is shown in Fig. 45. This bridge is supported by means of two springs
each of the stiness k. This system has two degree of freedom and the two generalized
coordinates are denoted by and . There is an excitation force F applied at the
hinge C. This force can be adopted in the following form
F = F
o
cos t
Produce:
1. the dierential equations of motion of the system and present them in the
standard form
2. the equation for the natural frequencies of the system
3. the expression for the amplitude of the forced vibrations of the system
4. the expression for the interaction force between the spring attached to the
hinge B and the ground
ENGINEERING APPLICATIONS 126
2.3 ENGINEERING APPLICATIONS
2.3.1 Balancing of rotors
Let us consider a rigid rotor that rotates with an angular velocity about the axis
AA (see Fig. 46).

U
i
1 2
U
i1
U
i2
F
1
F
2
2 A
A
B
B
m
1 m
2
U
1
U
2
x
1
x
2
t
t
Figure 46
In a general case, due to the limited accuracy of manufacturing, the centres
of gravity of the individual cross-sections do not have to coincide with this axis of
rotation. They are distributed along, usually unknown, line B B. Its follows that
due to rotation of this body at each cross-section i there exists the centrifugal force U
i
(see Fig. 46). Each of this forces can be replaced by two forces U
i1
and U
i2
acting in
two arbitrarily chosen planes. Each of them is perpendicular to the axis of rotation,
therefore their resultants U
1
and U
2
are perpendicular to the axis of rotation too.
Hence, one can eliminate this unbalance of the rotor by means of two weights of
mass m
1
and m
2
attached at such a position that the centrifugal forces F
1
and F
2
balance the resultant forces U
1
and U
2
. The process of searching for magnitude of
the unbalance forces U
1
and U
2
and their phases
1
and
2
is called balancing. The
balancing of a rotor can be performed with help of a specially design machines before
it is installed or can be carried out after its installation in its own bearings. The
second approach for balancing rotors is consider in this section.
U
2
2
x
2
t
1 2
t
1
x
1
U
1
t
3 4
x
1
x
2
x
3
x
4
t
x
3
x
3
3
a
3
U
1o
U
2o
t t
t
a
4
1
2
3
4
5
a
3o
a
4o
Figure 47
Let us consider the rotating machine shown in Fig. 47. According to the
above discussion, if the rotor of this machine can be approximated by a rigid body,
the unbalance forces can be represented by forces U
1
and U
2
in two arbitrarily chosen
plane. These two arbitrarily chosen planes,denoted in Fig. 47 by nubers 1 and 2, are
called balancing plane. Although the selection of the balancing planes is arbitrary,
there are numerous practical considerations for proper selection. For long rotors, for
example, the balancing planes should be chosen as far apart as possible. Furthermore,
these plane should oer an easy access and allow additional weights to be attached.
These unbalance forces excite vibrations of this machine. Let us arrange for these
vibrations to be recorded in two arbitrarily chosen planes. These planes, marked in
Fig. 47 by numbers 3 and 4, are called measurement planes. Let a
3
and a
4
be the
complex displacements measured in the measurement plasen along the coordinates
x
3
and x
4
with help of the two transducers 3 and 4. The transducer 5, which is
called key phasor, creates a timing reference mark on the rotor. This mark, shown
in Fig. 47 by N, allows the phases of the unbalance forces (
1
,
2
) and the phases
of the recorded displacements (
1
,
2
) to be measured. The equation 2.83 oers
the relationship between the unknown unbalance forces U
1
and U
2
and the measured
displacements a
3
and a
4
.
a
3
a
4
=

R
31
(i) R
32
(i)
R
41
(i) R
42
(i)

U
1
U
2
(2.214)
where
a
3
= a
3o
e
i
3
, a
4
= a
4o
e
i
4
, U
1
= U
1o
e
i
1
, U
2
= U
2o
e
i
2
(2.215)
If the transfer functions R
i,j
(i) would be known, this relation would allow the un-
know magnitudes of the unbalance as well as their phases to be determined. In order
to identify the transfer functions two additional tests are required.
Test (1)
U
2
2
x
2
t
1 2
t
1
x
1
U
1
t
3 4
x
3
3
a
3
U
1o
U
2o
t t
a
4
a
3o
a
4o
(1)
m
(1)
(1)
U
U
o
(1)
(1)
(1)
(1)
(1)
(1)
(1)
(1)
Figure 48
An additionl trial weight of mass m
(1)
(see Fig 48) is attached in the balancing
plane 1 at the known (with respect to the key phasors mark) phase
(1)
and the know
distance
(1)
. The system is now excited by both the residual unbalance forces (U
1
and U
2
) and the centrifugal force produced by the trial weight U
(1)
. The amplitude
of this force U
(1)
is
U
(1)
o
= m
(1)
(1)
2
(2.216)
The response of the system is recorded in both measurment planes so the amplitudes
a
(1)
3o
and a
(1)
4o
as well as the phases
(1)
3
and
(1)
4
can be obtained. There is the following
relationship between the measured parameters and the transfer functions.
"
a
(1)
3
a
(1)
4
#
=

R
31
(i) R
32
(i)
R
41
(i) R
42
(i)

U
1
+U
(1)
U
2
(2.217)
a
(1)
3
= a
(1)
3o
e
i
(1)
3
, a
(1)
4
= a
(1)
4o
e
i
(1)
4
, U
1
= U
1o
e
i
1
, U
2
= U
2o
e
i
2
, U
(1)
= U
(1)
o
e
i
(1)
(2.218)
Test (2)
U
2
2
x
2
t
1 2
t
1
x
1
U
1
t
3 4
x
3
3
a
3
U
1o
U
2o
t t
a
4
a
3o
a
4o
(2)
m
(2)
(2)
U
U
o
(2)
(2)
(2)
(2)
(2)
(2)
(2)
(2)
Figure 49
An additional trial weight of mass m
(2)
(see Fig 49) is attached in the balancing
plane 2 at the known (with respect to the key phasors mark) phase
(2)
and the know
distance
(2)
. The system is now excited by both the residual unbalance forces (U
1
and U
2
) and the centrifugal force produced by the trial weight U
(2)
. The amplitude
of this force U
(2)
is
U
(2)
o
= m
(2)
(2)
2
(2.219)
The response of the system is recorded in both measurement planes so the amplitudes
a
(2)
3o
and a
(2)
4o
as well as the phases
(2)
3
and
(2)
4
can be obtained. There is the following
relationship between the measured parameters and the transfer functions.
"
a
(2)
3
a
(2)
4
#
=

R
31
(i) R
32
(i)
R
41
(i) R
42
(i)

U
1
U
2
+U
(2)
(2.220)
a
(2)
3
= a
(2)
3o
e
i
(2)
3
, a
(2)
4
= a
(2)
4o
e
i
(2)
4
, U
1
= U
1o
e
i
1
, U
2
= U
2o
e
i
2
, U
(2)
= U
(2)
o
e
i
(2)
(2.221)
The formulated equations 2.214, 2.217 and 2.220 allowthe unknown transfer functions
and the wanted unbalances U
1
and U
2
to be computed. To achieve that let us subtract
the equations 2.214 from 2.217
"
a
(1)
3
a
(1)
4
#
a
3
a
4
!
=

R
31
(i) R
32
(i)
R
41
(i) R
42
(i)

U
1
+U
(1)
U
2
U
1
U
2
"
a
(1)
3
a
3
a
(1)
4
a
4
#
=

R
31
(i) R
32
(i)
R
41
(i) R
42
(i)

U
(1)
0
"
a
(1)
3
a
3
a
(1)
4
a
4
#
=

R
31
(i)U
(1)
R
41
(i)U
(1)
R
31
(i) =
a
(1)
3
a
3
U
(1)
=
a
(1)
3o
e
i
(1)
3
a
3o
e
i
3
U
(1)
o
e
i
(1)
R
41
(i) =
a
(1)
4
a
4
U
(1)
=
a
(1)
4o
e
i
(1)
4
a
4o
e
i
4
U
(1)
o
e
i
(1)
(2.222)
Similarly, if one subtracts equations 2.214 from 2.220 one can get
R
32
(i) =
a
(2)
3
a
3
U
(2)
=
a
(2)
3o
e
i
(2)
3
a
3o
e
i
3
U
(2)
o
e
i
(2)
R
42
(i) =
a
(2)
4
a
4
U
(2)
=
a
(2)
4o
e
i
(2)
4
a
4o
e
i
4
U
(2)
o
e
i
(2)
(2.223)
Now, the wanted complex imbalances U
1
and U
2
in the plane 1 and 2 may be computed
from the equation 2.214
U
1
U
2
=

R
31
(i) R
32
(i)
R
41
(i) R
42
(i)
a
3
a
4
(2.224)
where a
3
and a
4
represent the know response of the system without the additional
weights.
U
1
U
2
=

R
31
(i) R
32
(i)
R
41
(i) R
42
(i)
a
3o
e
i
3
a
4o
e
i
4
=

U
1o
e
i
1
U
2o
e
i
2
(2.225)
The amplitudes U
1o
and U
2o
determine the weights m
1
and m
2
that should be attached
in the balancing planes
m
1
=
U
1
r
1
2
m
2
=
U
2
r
2
2
(2.226)
These weights, to balance the rotor, should be place at angular position (see Fig. 50)
1
= 180
o
+
1
2
= 180
o
+
2
(2.227)
U
2
2
x
2
t
1 2
t
1
x
1
U
1
t
3 4
x
1
x
2
x
3
x
4
U
1o
U
2o
1
2
3
4
5
r
2
r
1
m
2
m
1
1
Figure 50
2.3.2 Dynamic absorber of vibrations
Let us consider vibration of the ventilator shown in Fig. 51a). Vibration of this
ventilator are due to the imbalance u if its rotor.
x
t
m
2
cos t x
M
m
m
2
cos t
M
K
m
2
a)
b)
r
r
r
r
Figure 51
Let us assume that the system has the following parameters:
M = 100 kg - total mass of the ventilator
m
r
= 20 kg - mass of rotor of the ventilator
K = 9000000 N/m - stiness of the supporting beam
= 314 rad/s the ventilators operating speed
= .0001 m - distance between the axis of rotation and the centre of gravity
u = m
r
= 20 .0001 = .00 2 kgm - imbalance of the rotor
The natural frequency of the system is
n
=
r
K
M
=
r
9000000
100
= 300 (2.228)
Hence, within the range of the rotor angular speed 0 < < 500 the system can be
approximated by system with one degree of freedom. Its physical model is shown in
Fig. 51b). The following mathematical model
M x +Kx = m
r
2
cos t (2.229)
x +
2
n
x = q cos t (2.230)
q =
u
M
2
=
0.002
100

2
= .0000 2
2
(2.231)
allows the amplitude of the forced vibrations of the ventilator A to be predicted.
A =
2
n
.0000 2
2
300
2
(2.232)
Its values, as a function of the angular speed of the rotor is shown in Fig. 52
0
0.0001
100 200 300 400
A
1
[m]
[ 1/s]

Figure 52
As it can be seen from this diagram, the ventilator develops large vibration in
vicinity of its working speed = 314 rad/s and has to pass the critical speed during
the run up. Such a solution is not acceptable. One of a possible way of reducing
these vibration is to furnish the ventilator with the absorber of vibration shown in
g 53
x
t
m
2
cos t x
M
m
m
2
cos t
M
K
m
2
a)
b)
c k
r
r
m
r
y
y
c
m
k
r
Figure 53
It comprises block of mass m, elastic element of stiness k and damper of the
damping coecient c. Application of the Newtons - Eulers method, results in the
following mathematical model.
M x + (K +k)x ky +c x c y = u
2
cost
m y kx +ky c x +c y = 0 (2.233)
Its matrix form is
M 0
0 m

x
y
+c c
c +c

x
y
K +k k
k +k

x
y
=

u
2
cos t
0
(2.234)
To analyze the forced vibrations let us introduce the complex excitation
M 0
0 m

x
y
+c c
c +c

x
y
K +k k
k +k

x
y
=
=

u
2
cos t +iu
2
sin t
0
=

u
2
e
it
0
=

u
2
0
e
it
(2.235)
Introducing notations
m =
M 0
0 m
; c =
+c c
c +c
; k =

K +k k
k +k
; q =
u
2
0
; x =
x
y
(2.236)
The above equations takes form
m x +c x +kx = qe
it
(2.237)
If one predicts the particular solution as
x = Ae
it
(2.238)
and than introduces it into the equation 2.237 one obtains the formula for the am-
plitude of the forced vibration
A =
(
2
m+ic +k)
1
q
(2.239)
Remarkable results we are getting if parameters k and m of the absorber fulll the
following relationship
r
k
m
= = 314 (2.240)
To show it let us assume
m = 25 kg
and compute the value of the stiness k from the formula 2.240
k = m
2
= 25 314
2
= 2. 4649 10
6
N/m
Introduction of this data into equation 2.239 and the zero damping results in the
following response A
1
and A
2
of the system along the coordinates x and y respectively.
A =

A
1
A
2
100 0
0 25
+i
0 0
0 0
11.5 2. 46
2. 46 2. 46
10
6
0.002
2
0
(2.241)
Amplitude A
1
, representing vibrations of the ventilator, as a function of the angular
speed of its rotor is presented in Fig. 54:

0
0.0001
100
200 300
400
A
1
[m]
[1/s]
Figure 54
One can notice that the amplitude of vibration for the working speed = 314
rad/s is equal to zero. But the ventilator still has to pass resonance in vicinity of
= 240 rad/s. To improve the dynamic response, let us analyze the inuence of the
damping coecient c.
A
1
A
2
=
=

100 0
0 25
+i
c c
c c
10
3
+
11.5 2. 46
2. 46 2. 46
10
6
0.002
2
0
(2.242)
The amplitudes of the forced vibration of the ventilator for dierent values of the
damping coecient c, computed according to the formula 2.242 are collected in the
Table 1. It can be noticed, that by increasing the damping coecient c one can lower
amplitude of vibrations in all region of frequency. The best results of attenuation of
vibrations can be achieved if the two local maxima are equal to each other. This case
is shown in the last raw of the table 1. Application of the absorber of vibrations oers
a safe operation in region of the angular speed 0 < < 500 rad/s. The amplitude is
less than 0.00004 m. Damping coecient lager then 5000 results in increment of the
amplitude of the ventilators forced vibrations. If the damping tends to innity, The
relative motion is ceased and the system behaves like the undamped system with one
degree of freedom.
.
Table 1
m = 25
k = 2. 46 10
6
c = 0

0
0.0001
100 200
300 400 [1/s]
A
1
[m]
m = 25
k = 2. 46 10
6
c = 1000

400 300 200 100 0
0.0001
0
A
1
[m]
[1/s]
m = 25
k = 2. 46 10
6
c = 2500

400 300 200 100 0
0.0001
0
A
1
[m]
[1/s]
m = 25
k = 2. 46 10
6
c = 5000

0
0.0001
100 200 300 400
0
A
1
[m]
[1/s
]
m = 25
k = 1.75 10
6
c = 5000

0
0.0001
100 200 300 400
0
A
1
[m]
[1/s]
Chapter 3
VIBRATION OF CONTINUOUS SYSTEMS
3.1 MODELLING OF CONTINUOUS SYSTEMS
3.1.1 Modelling of strings, rods and shafts
Modelling of stings
y(z,t)
z
y(z,t)
y(z,t)
z
2
y(z,t)
z
2
dz +
T
dz
T
T
f(z,t)
z
z
y
A(z), (z)
Figure 1
Strings are elastic elements that are subjected to tensile forces (see Fig. 1).
It is assumed that the tensile force T is large enough to neglect its variations due to
small motion of the string around its equilibrium position. In the Fig. 1 A(z) stands
for area of cross-section of the string and (z) is its density. Motion of the string is
caused by the unit vertical load f(z, t) that in a general case can be a function of
time t and the position z. Let us consider element dz of the sting. Its position is
determined by the coordinate z and its mass dm is
dm = A(z)(z)dz (3.1)
The free body diagram of this element is shown in Fig. 1. According to the second
Newtons law
dm
2
y(z, t)
t
2
= T
y(z, t)
z
+T

y(z, t)
z
+

2
y(z, t)
z
2
dz
+f(z, t)dz (3.2)

MODELLING OF CONTINUOUS SYSTEMS 138
Introduction of Eq. 3.1 into Eq. 3.2 and its simplication yields
A(z)(z)
2
y(z, t)
t
2
dz T

2
y(z, t)
z
2
dz = f(z, t)dz (3.3)
If one divide this equation by A(z)(z)dz it takes form
2
y(z, t)
t
2

T
A(z)(z)
2
y(z, t)
z
2
=
f(z, t)
A(z)(z)
(3.4)
If the string is uniform ( A and are independent of z ) the equation of motion is
2
y(z, t)
t
2

2
2
y(z, t)
z
2
= q(z, t) (3.5)
where
2
=
T
A
; q(z, t) =
f(z, t)
A
(3.6)
Modelling of rods
y(z,t)
dz
f(z,t)
z
z
E(z), A(z), (z)
dz
F(z,t)
F(z,t)
z
dz
+ F(z,t)
z
F(z,t)
F(z,t)
F(z,t)
z
dz
+ F(z,t)
Figure 2
Rods are elastic elements that are subjected to the axial forces. Let as consider
a rod of the cross-section A(z), Youngs modulus E(z) and the density (z). Motion
of the rod is excited by the axial force f(z, t) that, in a general case, can be a function
of position z and time t. Let us consider the highlighted in Fig. 2 element dz. Its
instantaneous position is determined by the displacement y(z, t). Application of the
second Newtons law to the free body diagram of the element yields.
dm
2
y(z, t)
t
2
= F(z, t) +F(z, t) +
F(z, t)
z
dz +f(z, t)dz (3.7)
The axial force F(z, t) is related to the elongation of the element by Hookes law
F(z, t) = A(z)E(z)
y(z,t)
z
dz
dz
= A(z)E(z)
y(z, t)
z
(3.8)
Upon introducing the above expression into Eq. 3.7 one may obtain
dm
2
y(z, t)
t
2

z
A(z)E(z)
y(z, t)
z
dz = f(z, t)dz (3.9)

Since mass of the element is
dm = A(z)(z)dz (3.10)
the equation of motion of the element is
A(z)(z)
2
y(z, t)
t
2

z
A(z)E(z)
y(z, t)
z
= f(z, t)dz (3.11)

If the rod is uniform ( A, E, are constant) one can get
2
y(z, t)
t
2

2
2
y(z, t)
z
2
= q(z, t) (3.12)
where
2
=
E
; q(z, t) =
f(z, t)
A
(3.13)
Modelling of shafts
Shafts are elastic elements that are subjected to torques. Let us assume that the
torque (z, t) is distributed along the axis z and is a function of time t (see Fig. 3).
The shaft has the shear modulus G(z), the density (z), the cross-section area A(z)
and the second moment of area J(z). Due to the moment (z, t), the shaft performs
the torsional vibrations and the instantaneous angular position of the cross-section
at z is (z, t). The angular position at the distance z +dz is by the total dierential
(z,t)
z
dz greater. Let us consider the element dz of the shaft. Its moment of inertia
about the axis z is
dI =
Z
A
r
2
dA(z)dz = (z)dz
Z
A
r
2
dA = J(z)(z)dz (3.14)
dz z
z
G(z), E(z), A(z), (z)
T(z,t)
z
(z,t)
(z,t)
(z,t)
T(z,t)
z
dz
+ T(z,t)
(z,t)
z
dz
+ (z,t)
(z,t)
z
dz
+ (z,t)
(z,t)
Figure 3
Owning to the generalized Newtons law we can write the following equation
dI
2
(z, t)
t
2
= T(z, t) +T(z, t) +
T(z, t)
z
dz +(z, t)dz (3.15)
After introduction of Eq. 3.14 and an elementary simplication the equation 3.15
takes form
J(z)(z)
2
(z, t)
t
2

T(z, t)
z
= (z, t) (3.16)
If we introduce the relationship between the torque T(z, t) and the deection (z, t)
(z, t)
z
dz =
T(z, t)dz
G(z)J(z)
(3.17)
into Eq. 3.16 we are getting
J(z)(z)
2
(z, t)
t
2

z
G(z)J(z)
(z, t)
z
= (z, t) (3.18)
If J(z), (z) and G(z) are constant, the equation of motion takes form.
2
(z, t)
t
2

2
2
(z, t)
z
2
= q(z, t) (3.19)
where
2
=
G
; q(z, t) =
(z, t)
J
(3.20)
3.1.2 Modelling of beams
dz
f(z,t)
z
z
E(z),J(z), A(z), (z)
y(z,t)
y
M(z,t)
V(z,t)
V(z,t)
z
dz
+ V(z,t)
M(z,t)
z
dz
+ M(z,t)
G
f(z,t)
dz/2
Figure 4
Beams are elastic elements that are subjected to lateral loads (forces or mo-
ments that have their vectors perpendicular to the centre line of a beam). Let us
consider a beam of the second moment of area J(z), cross-section A(z), density (z)
and the Youngs modulus E(z). The beam performs vibrations due to the external dis-
tributed unit load f(z, t). The instantaneous position of the element dz is highlighted
in Fig. 4. The equation of motion of the beam in the z direction is
dm
2
y(z, t)
t
2
= +V (z, t) V (z, t)
V (z, t)
z
dz +f(z, t)dz (3.21)
If one neglect the inertia moment associated with rotation of the element dz, sum of
the moments about the point G has to be equal to zero
V (z, t)
dz
2
+
V (z, t) +
V (z, t)
z
dz
dz
2
+M(z, t)
M(z, t) +
M(z, t)
z
dz
= 0
(3.22)
Simplication of the above equation and omission of the terms of order higher then
one with respect to dz, yields the relationship between the bending moment M and
the shearing force V.
V (z, t) =
M(z, t)
z
(3.23)
Since mass of the element dz is
dm = A(z)(z)dz (3.24)
and taking into account Eq. 3.23, one can get the equation of motion in the following
form
A(z)(z)
2
y(z, t)
t
2
+
M
2
(z, t)
z
2
= f(z, t) (3.25)
The mechanics of solids oers the following relationship between the deection of the
beam y(z, t) and the bending moment M(z, t).
M(z, t) = E(z)J(z)
2
y(z, t)
z
2
(3.26)
Introduction of equation 3.26 into equation 3.25 yields
A(z)(z)
2
y(z, t)
t
2
+

2
z
2
E(z)J(z)
2
y(z, t)
z
2
= f(z, t) (3.27)
If the following parameters of the beam A, J, E and are constant, motion of the
beam is governed by the following equation
2
y(z, t)
t
2
+
2
4
y(z, t)
z
4
= q(z, t) (3.28)
where
2
=
EJ
A
; q(z, t) =
f(z, t)
A
(3.29)
ANALYSIS OF CONTINUOUS SYSTEMS 143
3.2 ANALYSIS OF CONTINUOUS SYSTEMS
As could be seen from the previous section, vibrations of strings, rods and shafts are
described by the same mathematical model. Therefore, its analysis can be discussed
simultaneously. The strict solution can be produced only if parameters of the system
considered are constant. In this case the governing equation
2
y(z, t)
t
2

2
2
y(z, t)
z
2
= q(z, t) (3.30)
is classied as linear partial dierential equation of two variables ( z and t ) with
constant coecients (
2
). The general solution, a function of two variables, is sum
of the general solution of the homogeneous equation and the particular solution of
the non-homogeneous equation. If the external excitation q(z, t) = 0, the equation
3.30 describes the free vibration of the system due to a non-zero initial excitation
determined by the initial conditions.
3.2.1 Free vibration of strings, rods and shafts
The free vibrations (natural vibrations) are governed by the homogeneous equation
of Eq. 3.30
2
y(z, t)
t
2

2
2
y(z, t)
z
2
= 0 (3.31)
Boundary conditions - natural frequencies and natural modes
Let us predict the particular solution of the above equation in a form of the product
of two functions. One of them is a function of the position z and the other one is a
harmonic function of time t.
y(z, t) = Y (z) sin
n
t (3.32)
Introduction of the predicted solution 3.32 into equation 3.31 yields the following
ordinary dierential equation
2
n
Y (z)
2
Y
II
(z) = 0 (3.33)
or
Y
II
(z) +
2
n
Y (z) = 0 (3.34)
where
n
=

n
(3.35)
The general solution of this equation is
Y
n
(z) = S
n
sin
n
z +C
n
cos
n
z (3.36)
where
n
=
n
r
A
T
for strings (3.37)
n
=
n
r
E
for rods (3.38)
n
=
n
r
G
for shafts (3.39)
The values for the parameter
n
as well as the constants S
n
and C
n
should be chosen
to fulll the boundary conditions. Some of the boundary conditions for strings, rods
and shafts are shown in the following table.
Table 3.1
y
z
l
string
for z = 0 Y
n
= 0
for z = l Y
n
= 0
z
l
free-free rod
for z = 0 F = AEY
I
n
= 0
for z = l F = AEY
I
n
= 0
z
l
xed-free rod
for z = 0 Y
n
= 0
for z = l F = AEY
I
n
= 0
z
l
k
xed-elasticaly supported rod
for z = 0 Y
n
= 0
for z = l F = AEY
I
n
= kY
n
z
l
xed-xed rod
for z = 0 Y
n
= 0
for z = l Y
n
= 0
z
l
free-free shaft
for z = 0 T = GJY
I
n
= 0
for z = l T = GJY
I
n
= 0
z
l
xed free shaft
for z = 0 Y
n
= 0
for z = l T = GJY
I
n
= 0
z
l
xed-xed shaft
for z = 0 Y
n
= 0
for z = l Y
n
= 0
To demonstrate the way of the determination of the natural frequencies and
the corresponding natural modes, let us consider the xed on both ends shaft (last
row of the above Table). For this case the boundary conditions are
for z = 0 Y
n
= 0
for z = l Y
n
= 0 (3.40)
Introduction of this boundary conditions into the solution 3.36, results in a set of two
homogeneous algebraic equations linear with respect to the constants S
n
and C
n
.
0 = 0 S
n
+ 1 C
n
0 = (sin
n
l)S
n
+ (cos
n
l)C
n
(3.41)
Its matrix form is

0 1
sin
n
l cos
n
l

S
n
C
n
=

0
0
(3.42)
This set of equations has non-zero solutions if and only if its characteristic determinant
is equal to zero.

0 1
sin
n
l cos
n
l
= 0 (3.43)
Hence, in this particular case we have
sin
n
l = 0 (3.44)
This equation is called characteristic equation and has innite number of solution.
Since
n
and l are always positive, only positive roots of the above equation has the
physical meaning
1
=

l
,
2
=
2
l
, ........
n
=
n
l
, ........... n = 1, 2, .... (3.45)
Taking advantage of equation 3.39 one can compute the natural frequencies to be
n
=
n
p
n
=
n
s
G
=
n
l
s
G
n = 1, 2, .... (3.46)
For each of this natural frequencies the set of equations 3.41 becomes linearly de-
pendant and one of the constants can be chosen arbitrarily. If one choose arbitrarily
S
n
,say S
n
= 1, according to the rst equation of the set 3.41, C
n
has to be equal to
0. Therefore we can conclude that the predicted solution, according to 3.36, in the
case considered is
Y
n
(z) = sin
n
z = sin
n
l
z n = 1, 2, .... (3.47)
The functions Y
n
(z)are called eigenfunctions or natural modes and the corresponding
roots
n
are called eigenvalues or natural frequencies. The above analysis allows to
conclude that a continuous systempossesses innite number of the natural frequencies
and innite number of the corresponding natural modes. The rst mode is called
fundamental mode and the corresponding frequency is called fundamental natural
frequency. In the case of free vibrations of the shaft, the natural modes determine
the angular positions of the cross-section of the shaft (z). A few rst of them are
shown in Fig. 5
z
-1
-0.5
0
0.5
1
0.2 0.4 0.6 0.8 1
z
l=1
(z)
Figure 5
Orthogonality of the natural modes
Let us consider two arbitrarily chosen natural modes Y
i
(z) and Y
j
(z). Both of them
must fulll the equation 3.33
2
n
Y (z)
2
Y
II
(z) = 0
Hence
2
i
Y
i
(z)
2
Y
II
i
(z) = 0 (3.48)
2
j
Y
j
(z)
2
Y
II
j
(z) = 0 (3.49)
Premultiplying the equation 3.48 by Y
j
(z) and the equation 3.49 by Y
i
(z) and then
integrating them side by side one can get
2
i
Z
l
0
Y
i
(z)Y
j
(z)dz +
2
Z
l
0
Y
II
i
(z)Y
j
(z)dz = 0
2
j
Z
l
0
Y
j
(z)Y
i
(z)dz +
2
Z
l
0
Y
II
j
(z)Y
i
(z)dz = 0 (3.50)
The second integrals can be integrated by parts. Hence
2
i
Z
l
0
Y
i
(z)Y
j
(z)dz +
2
Y
I
i
(z)Y
j
(z)
l
0
2
Z
l
0
Y
I
i
(z)Y
I
j
(z)dz = 0
2
j
Z
l
0
Y
j
(z)Y
i
(z)dz +
2
Y
I
J
(z)Y
i
(z)
l
0
2
Z
l
0
Y
I
i
(z)Y
I
j
(z)dz = 0 (3.51)
Substraction of the second equation from the rst one yields
2
i

2
j
Z
l
0
Y
j
(z)Y
i
(z)dz +
2
Y
I
i
(z)Y
j
(z)
l
0

Y
I
J
(z)Y
i
(z)
l
0
= 0 (3.52)
It is easy to show that for any boundary conditions the second expression is equal to
zero
Y
I
i
(z)Y
j
(z)
l
0
Y
I
J
(z)Y
i
(z)
l
0
= Y
I
i
(l)Y
j
(l)Y
I
i
(0)Y
j
(0)Y
I
J
(l)Y
i
(l)+Y
I
J
(0)Y
i
(0) = 0
(3.53)
Hence,
Z
l
0
Y
j
(z)Y
i
(z)dz = 0 (3.54)
The above property of the eigenfunctions is called orthogonality condition.
General solution of the homogeneous equation
According to 3.32 one of the particular solution of the equation 3.31 can be adopted
in the following form
y(z, t) = Y (z) sin
n
t (3.55)
At this stage of consideration the function Y (z) is known and we are able to produce
innite number of such particular solutions.
y
n
(z, t) = Y
n
(z) sin
n
t n = 1, 2, 3..... (3.56)
Since the equation 3.31 is of second order with respect to time, to fulll initial con-
ditions we need second set of linearly independent solution. It the same manner as
it was done in the previous section one may prove that the following functions form
the required linearly independent set of solution.
y
n
(z, t) = Y
n
(z) cos
n
t n = 1, 2, 3..... (3.57)
Hence, the general solution of the equation 3.31 eventually may be adopted in the
following form.
y(z, t) =

X
n=1
S
n
Y
n
(z) sin
n
t +C
n
Y
n
(z) cos
n
t (3.58)
This solution has to fulll the initial conditions. The initial conditions determine the
initial position Y
0
(z) and the initial velocity V
0
(z) of the system considered for the
time t equal to zero.
y(z, 0) = Y
0
(z)

t
y(z, 0) = V
0
(z) (3.59)
To produce the constant S
n
and C
n
let us introduce the solution 3.58 into the above
initial conditions. This operation results in the following two equations.
Y
0
(z) =

X
n=1
C
n
Y
n
(z)
V
0
(z) =

X
n=1
S
n
n
Y
n
(z) (3.60)
To determine the unknown constants S
n
and C
n
, let us multiply the above equations
by Y
m
(z) and then integrate them side by side
Z
l
0
Y
0
(z)Y
m
(z)dz =

X
n=1
C
n
Z
l
0
Y
n
(z)Y
m
(z)dz
Z
l
0
V
0
(z)Y
m
(z)dz =

X
n=1
S
n
n
Z
l
0
Y
n
(z)Y
m
(z)dz (3.61)
Taking advantage of the developed orthogonality conditions 3.54 the wanted constants
S
n
and C
n
are
C
n
=
R
l
0
Y
0
(z)Y
n
(z)dz
R
l
0
Y
2
n
(z)dz
S
n
=
1
n
R
l
0
V
0
(z)Y
n
(z)dz
R
l
0
Y
2
n
(z)dz
(3.62)
For the example considered in the previous section the above formulae, according to
3.47, take form
C
n
=
R
l
0
Y
0
(z) sin
n
l
zdz
R
l
0
sin
n
l
z
2
dz
S
n
=
1
n
R
l
0
V
0
(z) sin
n
l
zdz
R
l
0
sin
n
l
z
2
dz
(3.63)
3.2.2 Free vibrations of beams
For the uniform beam the equation of motion was derived to be
2
y(z, t)
t
2
+
2
4
y(z, t)
z
4
= q(z, t) (3.64)
This equation can be classied as linear partial dierential equation of two variables
( z and t ) with constant coecients (
2
). Its order with respect to time is 2 and
with respect to z is equal to 4. The general solution, a function of two variables, is
sum of the general solution of the homogeneous equation and the particular solution
of the non-homogeneous equation. If the external excitation q(z, t) = 0, the equation
3.64 describes the free vibration of the beam due to a non-zero initial conditions.
The free vibrations (natural vibrations) are governed by the homogeneous
equation of 3.64.
2
y(z, t)
t
2
+
2
4
y(z, t)
z
4
= 0 (3.65)
Boundary conditions - natural frequencies and natural modes
Similarly to the analysis of strings and shafts, let us predict the solution of the above
equation in the form of a product of two functions. One of them is a function of the
position z and the other is the harmonic function of time t.
y(z, t) = Y (z) sin
n
t (3.66)
Introduction of the predicted solution 3.66 into equation 3.65 yields the following
ordinary dierential equation
2
n
Y (z) +
2
Y
IV
(z) = 0 (3.67)
or
Y
IV
(z)
4
n
Y (z) = 0 (3.68)
where
4
n
=

2
n
2
=
A
EJ
2
n
(3.69)
The standard form of its particular solution is
Y (z) = e
rz
(3.70)
Introduction of this solution into the equation 3.68 yields the characteristic equation
r
4
=
4
n
(3.71)
Its roots
r
1
=
n
r
2
=
n
r
3
= i
n
r
2
= i
n
(3.72)
determine the set of the linearly independent particular solution.
Y
1
(z) = e
n
z
Y
2
(z) = e
n
z
Y
3
(z) = e
i
n
z
Y
1
(z) = e
i
n
z
(3.73)
Alternatively, one can choose their combinations as the set of the independent solu-
tions
Y
1
(z) =
e
n
z
e
n
z
2
= sinh
n
z Y
2
(z) =
e
n
z
+e
n
z
2
= cosh
n
z
Y
3
(z) =
e
i
n
z
e
i
n
z
2
= sin
n
z Y
2
(z) =
e
i
n
z
+e
i
n
z
2
= cos
n
z (3.74)
A graphical interpretation of these functions for
n
= 1 is given in Fig. 6.
-4
-2
0
2
4
-4 -2 2 4 z
Y (z)
1
Y (z)
2
Y (z)
3
Y (z)
4
Figure 6
The general solution of the equation 3.68, as a linear combination of these
particular solutions is
Y
n
(z) = A
n
sinh
n
z +B
n
cosh
n
z +C
n
sin
n
z +D
n
cos
n
z
(3.75)
Values for the parameter
n
as well as for the constants A
n
, B
n
, C
n
and D
n
should
be chosen to fulll boundary conditions. Since this equation is of fourth order, one
has to produce four boundary conditions reecting the conditions at both ends of the
beam. They involve the function Y (z) and its rst three derivatives with respect to
z.
Y
n
(z) = A
n
sinh
n
z +B
n
cosh
n
z +C
n
sin
n
z +D
n
cos
n
z
(3.76)
Y
I
n
(z) = A
n
n
cosh
n
z +B
n
n
sinh
n
z +C
n
n
cos
n
z D
n
n
sin
n
z
(3.77)
Y
II
n
(z) = A
n
2
n
sinh
n
z +B
n
2
n
cosh
n
z C
n
2
n
sin
n
z D
n
2
n
cos
n
z
(3.78)
Y
I
n
(z) = A
n
3
n
cosh
n
z +B
n
3
n
sinh
n
z C
n
3
n
cos
n
z = D
n
3
n
sin
n
z
(3.79)
The boundary conditions for some cases of beams are shown in Table 3.2.
Table 3.2
z
l
Y
O
free-free beam
for z = 0 M(0) = EJY
II
(0) = 0
for z = 0 V (0) = EJY
III
(0) = 0
for z = l M(l) = EJY
II
(l) = 0
for z = l V (l) = EJY
III
(l) = 0
z
l
Y
O
xed-free beam
for z = 0 Y (0) = 0
for z = 0 Y
I
(0) = 0
II
(l) = 0
III
(l) = 0
z
l
Y
O
pined-xed beam
for z = 0 Y (0) = 0
II
(0) = 0
for z = l Y (l) = 0
for z = l Y
I
(l) = 0
z
l
Y
O
k
k
elasticaly supported beam
II
(0) = 0
for z = 0 V (0) = EJY
III
(0) = kY (0)
II
(l) = 0
III
(l) = +kY (l)
Let us take advantage of the boundary conditions corresponding to the free-
free beam in order to determine the natural frequencies and the natural modes.
II
(0) = 0
for z = 0 V (0) = EJY
III
(0) = 0
II
(l) = 0
III
(l) = 0
(3.80)
Introduction of the functions 3.78 and 3.79 into the above boundary conditions results
in the following set of algebraic equations that are linear with respect to the constants
A
n
, B
n
, C
n
and D
n
.
_
_
0
2
n
0
2
n
3
n
0
3
n
0
2
n
sinh
n
l
2
n
cosh
n
l
2
n
sin
n
l
2
n
cos
n
l
3
n
cosh
n
l
3
n
sinh
n
l
3
n
cos
n
l
3
n
sin
n
l
_
_
_
_
A
n
B
n
C
n
D
n
_
_
=
_
_
0
0
0
0
_
_
(3.81)
They have a non-zero solution if and only if their characteristic determinant is equal
to zero. This condition forms the characteristic equation
0 1 0 1
1 0 1 0
sinh
n
l cosh
n
l sin
n
l cos
n
l
cosh
n
l sinh
n
l cos
n
l sin
n
l
= 0 (3.82)
that, after simplication, takes the following form
cosh
n
l cos
n
l 1 = 0 (3.83)
This characteristic equation is transcendental and therefore has innite number of
roots. Solution of this equation, within a limited range of the parameter
n
l is shown
in Fig. 7
The rst few roots are
0
l = 0
1
l = 4.73
2
l = 7.85
3
l = 11......... (3.84)
As one can see from the diagram 7, the characteristic equation has double root of zero
magnitude. Since the beamconsidered is free-free in space, this root is associated with
the possible translation and rotation of the beam as a rigid body. These two modes,
corresponding to the zero root are shown in Fig. 8a) and b). Modes corresponding
to the non-zero roots can be produced according to the following procedure.
For any root of the characteristic equation the set of equations 3.81, since
its characteristic determinant is zero, becomes linearly dependant. Therefore, it is
possible to choose arbitrarily one of the constants (for example A
n
) and the other
can be obtained from three arbitrarily chosen equations 3.81. If we take advantage
of the second, third and fourth equation we are getting
_
_
0 1 0
cosh
n
l sin
n
l cos
n
l
sinh
n
l cos
n
l sin
n
l
_
_
_
_
B
n
C
n
D
n
_
_
=
_
_
1
sinh
n
l
cosh
n
l
_
_
A
n
(3.85)
-20
-10
0
10
20
5 10 15
l
n
l
1
l
2
l
3
l
0
cosh cos -1=0
l
n
l
n
Figure 7
Hence, for A
n
= 1 we have
_
_
B
n
C
n
D
n
_
_
=
_
_
0 1 0
cosh
n
l sin
n
l cos
n
l
sinh
n
l cos
n
l sin
n
l
_
_
1
_
_
1
sinh
n
l
cosh
n
l
_
_
(3.86)
For the rst non-zero root
1
l = 4.73 the above set of equations yields values for
constants B
1
, C
1
and D
1
_
_
B
1
C
1
D
1
_
_
=
_
_
0 1 0
cosh4.73 sin4.73 cos 4.73
sinh 4.73 cos 4.73 sin 4.73
_
_
1
_
_
1
sinh 4.73
cosh 4.73
_
_
=
_
_
1.017 8
1.0
1.0177
_
_
(3.87)
Hence, the corresponding mode, according to Eq. 3.75 is
Y
1
(z) = 1.0 sinh4.73z + 1.0178 cosh4.73z 1.0 sin4.73z + 1.0177 cos 4.73z (3.88)
Its graphical representation is shown in Fig. 8c).
In the same manner one can produce modes for all the other characteristic
roots. Modes for
2
l = 7.85 and
3
l = 11 are shown in Fig. 8d) and e) respectively.
The formula 3.69 allows the natural frequencies to be computed.
n
=
2
n
s
EJ
A
=
(
n
l)
2
l
2
s
EJ
A
(3.89)
Eventually, taking into account the predicted solution 3.66, the particular solution is
y(z, t) = S
n
Y
n
(z) sin
n
t (3.90)
-2
-1
0
1
2
0.2 0.4 0.6 0.8 1
z
-2
-1
0
1
2
0.2 0.4 0.6 0.8 1
z
-2
-1
0
1
2
0.2 0.4 0.6 0.8 1
z
-2
-1
0
1
2
0.2 0.4 0.6 0.8 1
z
-2
-1
0
1
2
0.2 0.4 0.6 0.8 1
z
0
Y (z)
0
Y (z)
1
Y (z)
2
Y (z)
3
Y (z)
a)
b)
c)
d)
e)
Figure 8
where Y
n
(z) and
n
are uniquely determined and S
n
is an arbitrarily chosen constant.
In the same manner we can show that the function
y(z, t) = C
n
Y
n
(z) cos
n
t (3.91)
is the linearly independent particular solution too. It follows that the following linear
combination
n=
X
n=1
Y
n
(z) (S
n
sin
n
t +C
n
cos
n
t) (3.92)
where
Y
1
(z) = 1.0 sinh4.73z + 1.0178 cosh 4.73z 1.0 sin4.73z + 1.0177 cos 4.73z
Y
2
(z) = ..................... (3.93)
.....................
is the general solution of the equation 3.65. The constants S
n
and C
n
should be
chosen to fulll the initial conditions.
Orthogonality of the natural modes Let us consider two arbitrarily cho-
sen natural modes Y
i
(z) and Y
j
(z). Both of them must fulll the equation 3.67
2
n
Y (z) +
2
Y
IV
(z) = 0
Hence
2
i
Y
i
(z) +
2
Y
IV
i
(z) = 0 (3.94)
2
j
Y
j
(z) +
2
Y
IV
j
(z) = 0 (3.95)
Premultiplying the equation 3.94 by Y
j
(z) and the equation 3.95 by Y
i
(z) and then
integrating them side by side one can get
2
i
Z
l
0
Y
i
(z)Y
j
(z)dz +
2
Z
l
0
Y
IV
i
(z)Y
j
(z)dz = 0
2
j
Z
l
0
Y
j
(z)Y
i
(z)dz +
2
Z
l
0
Y
IV
j
(z)Y
i
(z)dz = 0 (3.96)
The second integrals can be integrated by parts. Hence
2
i
Z
l
0
Y
i
(z)Y
j
(z)dz +
2
Y
III
i
(z)Y
j
(z)
l
0
2
Z
l
0
Y
III
i
(z)Y
I
j
(z)dz = 0
2
j
Z
l
0
Y
j
(z)Y
i
(z)dz +
2
Y
III
J
(z)Y
i
(z)
l
0
2
Z
l
0
Y
III
i
(z)Y
I
j
(z)dz = 0
(3.97)
Let us apply the same procedure to the last integral again
2
i
R
l
0
Y
i
(z)Y
j
(z)dz +
2
Y
III
i
(z)Y
j
(z)
l
0
Y
II
i
(z)Y
I
j
(z)
l
0
+
2
R
l
0
Y
II
i
(z)Y
II
j
(z)dz = 0
(3.98)
2
j
R
l
0
Y
i
(z)Y
j
(z)dz +
2
Y
III
j
(z)Y
i
(z)
l
0
Y
II
j
(z)Y
I
i
(z)
l
0
+
2
R
l
0
Y
II
i
(z)Y
II
j
(z)dz = 0
Substraction of the second equation from the rst one yields
2
i

2
j
R
l
0
Y
j
(z)Y
i
(z)dz+
Y
III
i
(z)Y
j
(z)
l
0
+
2
Y
II
i
(z)Y
I
j
(z)
l
0
+
2
Y
III
j
(z)Y
i
(z)
l
0
Y
II
j
(z)Y
I
i
(z)
l
0
= 0
(3.99)
The expression
Y
III
i
(z)Y
j
(z)
l
0
+
2
Y
II
i
(z)Y
I
j
(z)
l
0
+
2
Y
III
j
(z)Y
i
(z)
l
0
Y
II
j
(z)Y
I
i
(z)
l
0
(3.100)
depends exclusively on boundary conditions. It is easy to show that for any possible
boundary conditions this expression is equal to zero. Hence,
Z
l
0
Y
j
(z)Y
i
(z)dz = 0 (3.101)
The above property of the natural modes is called orthogonality condition and play
a very important role in further development of the theory of vibrations.
3.2.3 Problems
Free vibrations of strings, rods and shafts
Problem 32
z
G
1
, A
1
,
1
, J
1
G
2
, A
2
,
2
, J
2
l
1
l
2
Figure 9
For the shaft shown in Fig. 9 produce equation for its natural frequencies.
Solution
For
0 < z < l
1
(3.102)
motion of the system, according to 3.19, is governed by the following equation
1
(z, t)
t
2

2
1
1
(z, t)
z
2
= 0 (3.103)
where
2
1
=
G
1
1
(3.104)
Similarly, one may say that within range
l
1
< z < l
1
+l
2
(3.105)
motion of the shaft is governed by
2
(z, t)
t
2

2
2
2
(z, t)
z
2
= 0 (3.106)
where
2
2
=
G
2
2
(3.107)
Both parts of the shaft must have the same natural frequencies. Therefore the par-
ticular solution of the above equations must be of the following form
1
(z, t) =
1
(z) sin
n
t (3.108)
2
(z, t) =
2
(z) sin
n
t (3.109)
Introduction of these solutions into the equations of motion yields, according to 3.34,
II
1
(z) +
2
n1
1
(z) = 0 (3.110)
II
2
(z) +
2
n2
2
(z) = 0 (3.111)
where
n1
=

n
1

n2
=

n
2
(3.112)
These two equations are coupled together by the following boundary conditions
for z = 0
1
(0) = 0
for z = l
1

1
(l
1
) =
2
(l
1
)
for z = l
1
G
1
J
1
I
1
(l
1
) = G
2
J
2
I
2
(l
1
)
for z = l
1
+l
2

I
2
(l
1
+l
2
) = 0 (3.113)
The rst boundary condition reects the fact that the left hand end of the shaft is
xed. The second and the third condition represent the continuity of the angular
displacement and continuity of the torque. The last condition says that the torque
at the free end is zero. Since the general solution of equation 3.110 and 3.111are
1
(z) = S
n1
sin

n
1
z +C
n1
cos

n
1
z (3.114)
2
(z) = S
n2
sin

n
2
z +C
n2
cos

n
2
z (3.115)
the formulated boundary conditions results in the following set of equations
C
n1
= 0
S
n1
sin

n
1
l
1
+C
n1
cos

n
1
l
1
S
n2
sin

n
2
l
2
C
n2
cos

n
2
l
2
= 0
S
n1
G
1
J
1
n
1
cos
n
1
l
1
C
n1
G
1
J
1
n
1
sin
n
1
l
1
+
S
n2
G
2
J
2
2
cos

n
2
l
1
+C
n2
G
2
J
2
2
sin

n
2
l
1
= 0
+S
n2
2
cos

n
2
(l
1
+l
2
) C
n2
2
sin

n
2
(l
1
+l
2
) = 0
(3.116)
Its matrix for is
[A]
_
_
S
n1
C
n1
S
n2
C
n2
_
_
= 0 (3.117)
where
[A] =
_
_
0 1 0 0
sin
n
1
l
1
cos
n
1
l
1
sin
n
2
l
2
cos
n
2
l
2
G
1
J
1
n
1
cos
n
1
l
1
G
1
J
1
n
1
sin
n
1
l
1
G
2
J
2
n
2
cos
n
2
l
1
G
2
J
2
n
2
sin
n
2
l
1
0 0

n
2
cos

n
2
(l
1
+l
2
)
2
sin

n
2
(l
1
+l
2
)
_
_
This homogeneous set of equations has the non-zero solutions if and only if its char-
acteristic determinant is equal to zero.
0 1 0 0
sin

n
1
l
1
cos

n
1
l
1
sin

n
2
l
2
cos

n
2
l
2
G
1
J
1
n
1
cos
n
1
l
1
G
1
J
1
n
1
sin
n
1
l
1
G
2
J
2
n
2
cos
n
2
l
1
G
2
J
2
n
2
sin
n
2
l
1
0 0
n
2
cos
n
2
(l
1
+l
2
)
n
2
sin
n
2
(l
1
+l
2
)
= 0
(3.118)
Solution of this equation for the roots
n
yields the wanted natural frequencies of the
shaft.
Problem 33

z
l
E , A ,
m
1
2
Figure 10
The uniform rod 1, shown in Fig. 10, is connected to the block 2 of mass
m. Compute the natural frequencies and the corresponding natural modes of this
assembly.
Solution
The equation of motion of the rod, according to Eq. 3.12, is
2
y(z, t)
t
2

2
2
y(z, t)
z
2
= 0 (3.119)
where
2
=
E
(3.120)
To produce the boundary conditions let us consider the block 2 with the adjusted
innitesimal element (see Fig. 11).
z
l
y(l,t)
y
m
F(l,t)
Figure 11
Equation of motion of the block, according to the Newtons law, is
m
2
y(l, t)
t
2
= F(l, t) (3.121)
or, taking advantage of the relationship 3.8
m

2
y(z, t)
t
2
z=l
= AE
y(z, t)
z
z=l
(3.122)
This equation together with the condition corresponding to the upper end of the rod
y(0, t) = 0 (3.123)
forms boundary conditions for the equation 3.119.
(
y(0, t) = 0
m

2
y(z,t)
t
2
z=l
= AE
y(z,t)
z
z=l
(3.124)
Introduction of the particular solution (see Eq. 3.32)
y(z, t) = Y (z) sin
n
t (3.125)
into the equation of motion 3.119 and the boundary conditions 3.124 yields the ordi-
nary dierential equation
Y
II
(z) +
2
n
Y (z) = 0;
n
=

n
(3.126)
with boundary conditions
Y (0) = 0
m
2
n
Y (l) AEY
I
(l) = 0
(3.127)
The general solution of the equation 3.126, according to 3.36, is
Y
n
(z) = S
n
sin
n
z +C
n
cos
n
z (3.128)
Introduction of this solution into boundary conditions yields
C
n
= 0
m
2
n
(S
n
sin
n
z +C
n
cos
n
z) AE(S
n
n
cos
n
z C
n
n
sin
n
z) = 0
(3.129)
or

m
2
n
sin
n
z AE
n
cos
n
z
S
n
= 0 (3.130)
Hence, the characteristic equation, after taking advantage of 3.126, is
m
2
n
sin

n
z AE
cos

n
z = 0 (3.131)
or after simplication
tan

n
l
AE
n
m
= 0 (3.132)
For the following numerical data
l = 1m
E = 2.1 10
11
N/m
2
A = 25 10
4
m
2
= 7800kg/m
3
=
q
E
= 5188.7m/s
m
r
= Al = 19.5kg - mass of the rod
m = 20kg - mass of the block
the characteristic equation takes the following form
f(
n
) = tan

n
5188.7
5059.0
n
= 0 (3.133)
Its solution f( = 0 is shown in Fig. 12.
-1
-0.5
0
0.5
1
5000 10000 15000 20000 25000 30000
n
f( )
n
Figure 12
The rst three natural frequencies, according to the diagram 12 are
1
= 4400,
2
= 17720,
1
= 33400 s
1
(3.134)
The corresponding natural modes, according to 3.128, are
Y
n
(z) = S
n
sin
n
z = S
n
sin

n
z = S
n
sin

n
5188.7
z (3.135)
For the rst three natural frequencies the corresponding natural modes
Y
1
(z) = sin
4400
5188.7
z Y
2
(z) = sin
17720
5188.7
z Y
3
(z) = sin
33400
5188.7
z (3.136)
are presented in Fig. 13, 14 and 15 respectively.If we neglect the mass of the rod, the
system becomes of one degree of freedom and its the only one natural frequency is
1
=
r
k
m
=
r
EA
lm
=
r
2.1 10
11
25 10
4
20
= 5123s
1
(3.137)
and the corresponding mode is a straight line.
-1
-0.5
0
0.5
1
0.2 0.4 0.6 0.8 1
z
Y (z)
1
Figure 13
-1
-0.5
0
0.5
1
0.2 0.4 0.6 0.8 1
z
Y (z)
2
Figure 14
-1
-0.5
0
0.5
1
0.2 0.4 0.6 0.8 1
z
Y (z)
3
Figure 15
Problem 34
Produce natural frequencies and the corresponding natural modes for the xed-
elastically supported beam shown in Fig. 16
z
l
k
y
O
Figure 16
Solution
According to the equation 3.65, the equation of motion of the beam is
2
y(z, t)
t
2
+
2
4
y(z, t)
z
4
= 0 (3.138)
Its particular solution can be sought in the following form
y(z, t) = Y
n
(z) sin
n
t (3.139)
the above solution has to fulll boundary conditions. At the left hand end the dis-
placement and gradient of the beam have to be equal to zero. Hence,
y(z, t)|
z=0
= 0 (3.140)
y(z, t)
z
z=0
= 0 (3.141)
The right hand end, with the forces acting on it, is shown in Fig. 17. Equilibrium
z
l
k
y
O
z
dz
k y(z,t)|
z=l
V(z,t)|
z=l
M(z,t)|
z=l
Figure 17
conditions for the element dz which have to be fulll for any instant of time, forms
the boundary conditions associated with the right hand end
M(z, t)|
z=l
= E(z)J(z)

2
y(z, t)
z
2
z=l
= 0 (3.142)
V (z, t)|
z=l
=
M(z, t)
z
z=l
=

z
EJ

2
y(z, t)
z
2
z=l
= ky(z, t)|
z=l
(3.143)
Introduction of the solution 3.139 into the above boundary conditions yields
Y
n
(z)|
z=0
= 0
Y
I
n
(z)
z=0
= 0
Y
II
n
(z)
z=l
= 0
Y
III
n
(z)
z=l
Y
n
(z)|
z=l
= 0 (3.144)
where
=
EJ
k
(3.145)
According to the equations 3.76 to 3.79 the expressions for the natural modes Y
n
(z)
and their derivatives are
Y
n
(z) = A
n
sinh
n
z +B
n
cosh
n
z +C
n
sin
n
z +D
n
cos
n
z
Y
I
n
(z) = A
n
n
cosh
n
z +B
n
n
sinh
n
z +C
n
n
cos
n
z D
n
n
sin
n
z
Y
II
n
(z) = A
n
2
n
sinh
n
z +B
n
2
n
cosh
n
z C
n
2
n
sin
n
z D
n
2
n
cos
n
z
Y
III
n
(z) = A
n
3
n
cosh
n
z +B
n
3
n
sinh
n
z C
n
3
n
cos
n
z +D
n
3
n
sin
n
z
(3.146)
where (see Eq. 3.69)
4
n
=

2
n
2
=
A
EJ
2
n
(3.147)
Introduction of the above expressions into the boundary conditions 3.144 results in
the following set of algebraic equations that is linear with respect to the constants
A
n
, B
n
, C
n
and D
n
.
B
n
+D
n
= 0
A
n
+C
n
= 0
A
n
2
n
sinh
n
l +B
n
2
n
cosh
n
l C
n
2
n
sin
n
l D
n
2
n
cos
n
l = 0
A
n
3
n
cosh
n
l +B
n
3
n
sinh
n
l C
n
3
n
cos
n
l +D
n
3
n
sin
n
l
(A
n
sinh
n
l +B
n
cosh
n
l +C
n
sin
n
l +D
n
cos
n
l) = 0
(3.148)
The matrix form of these equations is presented below
_
_
0 1 0 1
1 0 1 0
sinh
n
l cosh
n
l sin
n
l cos
n
l
3
n
cosh
n
l
sinh
n
l

3
n
sinh
n
l
cosh
n
l

3
n
cos
n
l
sin
n
l

3
n
sin
n
l
cos
n
l
_
_
_
_
A
n
B
n
C
n
D
n
_
_
=
_
_
0
0
0
0
_
_
(3.149)
The non-zero solution of this set of equations exists if and only if its characteristic
determinant is equal to zero.
0 1 0 1
1 0 1 0
sinh
n
l cosh
n
l sin
n
l cos
n
l
3
n
cosh
n
l
sinh
n
l

3
n
sinh
n
l
cosh
n
l

3
n
cos
n
l
sin
n
l

3
n
sin
n
l
cos
n
l
= 0 (3.150)
For the following data
E = 2.1 10
11
N/m
2
= 7800kg/m
3
A = 0.03 0.01 = 0.0003m
2
J =
0.030.01
3
12
= 2.5 10
9
m
4
k = 10000N/m
l = 1m
=
EJ
k
=
2.110
11
2.510
9
10000
= 0.0525
the characteristic equation takes form
0 1 0 1
1 0 1 0
sinh
n
cosh
n
sin
n
cos
n
.0525
3
n
cosh
n
sinh
n
.0525
3
n
sinh
n
cosh
n
.0525
3
n
cos
n
sin
n
.0525
3
n
sin
n
cos
n
= 0
(3.151)
Solution of this equation for its roots
n
is presented in Fig.18
-200
-100
0
100
200
2 4 6 8
n
Figure 18
From this diagram the rst three roots are
1
= 2.942m
1
2
= 4.884m
1
(3.152)
3
= 7.888m
1
The relationship 3.147
4
n
=
A
EJ
2
n
oers values for the wanted natural frequencies
1
=
s
4
1
EJ
A
= 129.6s
1
2
=
s
4
2
EJ
A
= 357.3s
1
(3.153)
2
=
s
4
3
EJ
A
= 932.0s
1
For each of these roots the set of equations 3.149 becomes linearly dependant. Hence
one of the unknown constants can be chosen arbitrarily (e.g. D
n
= 1) and the last
equation can be crossed out. The three remaining equations allow the constants A
n
,
B
n
, and C
n
to be computed.
_
_
0 1 0 1
1 0 1 0
sinh
n
l cosh
n
l sin
n
l cos
n
l
_
_
_
_
A
n
B
n
C
n
1
_
_
=
_
_
0
0
0
_
_
(3.154)
_
_
0 1 0
1 0 1
sinh
n
l cosh
n
l sin
n
l
_
_
_
_
A
n
B
n
C
n
_
_
+
_
_
1
0
cos
n
l
_
_
=
_
_
0
0
0
_
_
(3.155)
_
_
A
n
B
n
C
n
_
_
=
_
_
0 1 0
1 0 1
sinh
n
l cosh
n
l sin
n
l
_
_
1
_
_
1
0
cos
n
l
_
_
(3.156)
For the rst three roots the numerical values of these constants are
_
_
A
1
B
1
C
1
_
_
=
_
_
.883
1.0
.883
_
_
;
_
_
A
2
B
2
C
2
_
_
=
_
_
1.02
1.0
1.02
_
_
;
_
_
A
3
B
3
C
3
_
_
=
_
_
1.0
1.0
1.0
_
_
(3.157)
Introducing them into the rst function of 3.146 and remembering that D
n
= 1, we
are getting the corresponding natural modes
Y
1
(z) = .883 sinh 2.942z 1.0 cosh 2.942z .883 sin 2.942z + 1 cos 2.942z
Y
2
(z) = 1.02 sinh 4.884z 1.0 cosh 4.884z 1.02 sin 4.884z + 1 cos 4.884z
Y
3
(z) = 1.0 sinh 7.888z 1.0 cosh 7.888z 1.0 sin7.888z + 1 cos 7.888z (3.158)
The graphical interpretation of these natural modes is given in Fig. 19

-2
-1
0
1

2
0.2 0.4 0.6 0.8 1
z
Y (z)
2
1 3
Y (z)
Y (z)
Figure 19
Problem 35
, G, J
o
l

k

L

k
1
2
z
2

Figure 20
The left hand end of the shaft 1 shown in Fig. 20 is xed. Its right hand end
is supported by means of the massless and rigid beam 2 of length L that is connected
to two springs each of the stiness k.
Produce the equation for the natural frequencies of the shaft.
Problem 36
, G, J
o
l
1
2
z
2
Figure 21
Two rigid discs 2 (see Fig. 21) are joined together by means of the shaft 1 of
the length l. The moment of inertia of each disc about the axis z is I.
Produce the equation for the natural frequencies of the assembly.
Problem 37
y
z
l, E, J, A,
l
Figure 22
The uniform beam is supported as shown in Fig. 22.
Produce the equations for the natural vibrations of this beam
Chapter 4
APPROXIMATION OF THE CONTINUOUS SYSTEM BY
DISCRETE MODELS
The theory developed in the previous chapter provides a deep inside into the dynamic
behavior of the continuous system. However, the utilization of this theory for the
solution of an engineering problem encounters often a great deal of diculties. In
the engineering design such elements like shafts or beams has a variable cross-section
along its length (see Fig. 1) and can be divided into a number of uniform sections.It
y
z
n
m
z=l
n
Figure 1
follows that the equation of motion for each uniform section has to be joint together
with help of the continuity conditions. In case of a beam, the equations of motion of
the uniform elements n and m can be described by two dierential equations of the
form 3.28 (page 142
2
y
n
(z, t)
t
2

2
n
4
y
n
(z, t)
z
4
= q
n
(z, t) (4.1)
2
y
m
(z, t)
t
2

2
m
4
y
m
(z, t)
z
4
= q
m
(z, t) (4.2)
where, according to 3.29
2
n
=
E
n
J
n
A
n
n
; q
n
(z, t) =
f
n
(z, t)
A
n
n

2
m
=
E
m
J
m
A
m
m
; q
m
(z, t) =
f
m
(z, t)
A
m
m
(4.3)
Each of them could be solved within their uniformity range. These equations have to
fulll the following continuity conditions
y
n
(z, t)|
z=ln
= y
m
(z, t)|
z=lm
y
n
(z, t)
z
z=l
n
=
y
m
(z, t)
z
z=l
m
(4.4)
DISCRETE MODEL OF THE FREE-FREE BEAMS 175
In the engineering applications number of these conditions is large and the solution
of equations of motion of the beam, although still possible, is unpractical and te-
dious. Therefore, in most of the engineering problems the continuous systems are
approximated by the multi-degree-of-freedom systems. Process of determination of
the inertia matrix and the stiness matrix of a continuous system is called discretiza-
tion. This process is demonstrated in the next sections taking beam as an example
but it can be apply to any continuous element.
4.1 DISCRETE MODEL OF THE FREE-FREE BEAMS
The commercially available computer packages allow to produce stiness and inertia
matrix of free in space beam along coordinates y
n
and
yn
(see Fig. 2) associated
with an arbitrary chosen number N of points. These distinguished points n are called
nodes.The most popular approaches for creation of the stiness and inertia matrices
1 2 ......... n .............................................................. N
y
n
n
y
Figure 2
are called Rigid Element Method and Finite Element Method.
4.1.1 Rigid Elements Method.
Inertia and stiness matrix for the free-free beam
According to the Rigid Element Method, the beam is divided into a sucient, for
necessary accuracy, number of segments I of constant cross-section (Fig . 3a))The
bending and shearing properties of each segment are represented by two springs of
stiness k
M
i
and k
T
i
respectively (Fig. 3.b). Equivalence of both, the actual element
(Fig. 4a) and its model (Fig. 4b) requires equal angular deection (
r
i
=
e
i
) caused
by the same bending moment M
i
.
Since
r
i
=
M
i
l
i
2EJ
i
and
e
i
=
M
i
2k
M
i
(4.5)
the bending stiness is
k
M
i
=
EJ
i
l
i
(4.6)
Similarly, the equivalence of shearing deections (y
r
i
= y
e
i
) caused by the same
shearing force T
i
(Fig. 4c and Fig. 4d)
y
r
i
= l
i
r
i
=
T
i
l
i
GA
i
and y
e
i
=
T
i
k
T
i
(4.7)
.....
..... .....
.....
l
l l
l
i-1
i i+1
i
k
T
i
k
M
i
G
i
m
i
i
m
i
i
y
k
M
i
k
M
i -1
k
M
i+1
k
T
i k
T
i
-1
+1
m
i
i
+1
+1
l
i
r
i
l
i+1
r
i+1
a)
b)
c)
d)
z z z z
z
l
i-1
/2 l /
i
2
I
I
I
Figure 3
k
T
i
T
i
T
i y
e
i
y
r
i
T
i
T
i
l
i l
i
M
i M
i
r
i
k
M
i
M
i
M
i
i
e
E,J
i
G,A
i
a)
b)
c)
d)
i
e
2
Figure 4
yields
k
T
i
=
GA
i
l
i
(4.8)
The right hand part of the segment l
i1
and the left hand part of the subsequent
segment l
i
, form a section (Fig. 3c). Each section is considered rigid and its inertia
properties are represented by mass m
i
and moments of inertia I
i
. In this way a
complete symmetry is obtained that gives simple programming for computer analysis.
Application of the Lagranges equations to the physical model is shown in Fig.
3d) results in the following equations of motion.
m y +ky = 0 (4.9)
where
m =
_
_
M
1,1
.. 0 0 0 .. 0
. .. . . . .. .
0 .. M
i1,i1
0 0 .. 0
0 .. 0 M
i,i
0 .. 0
0 .. 0 0 M
i+1,i+1
.. 0
. .. . . . .. .
0 .. 0 0 0 .. M
N,N
_
_
(4.10)
k =
_
_
K
1,1
.. 0 0 0 .. 0
. .. . . . .. .
0 .. K
i1,i1
K
i1,i
0 .. 0
0 .. K
i,i1
K
i,i
K
i,i+1
.. 0
0 .. 0 K
i+1,i
K
i+1,i+1
.. 0
. .. . . . .. .
0 .. 0 0 0 .. K
N,N
_
_
(4.11)
y
b
=
_
_
y
1
:
y
i1
y
i
y
i+1
:
y
N
_
_
(4.12)
N = I + 1 (4.13)
M
i,i
=

m
i
0
0 I
i
(4.14)
K
i,i1
=

k
T
i1
k
T
i1
z
r
i1
+k
T
i1
z
l
i
k
M
i1
+k
T
i1
z
r
i1
z
l
i
(4.15)
K
i,i
=

+k
T
i1
+k
T
i
+k
T
i1
z
l
i
+k
T
i
z
r
i
+k
T
i1
z
l
i
+k
T
i
z
r
i
+k
M
i1
+k
T
i1
z
2
l
i
+k
T
i
z
2
r
i
(4.16)
K
i,i+1
=

k
T
i
+k
T
i
z
l
i+1
k
T
i
z
r
i
k
M
i
+k
T
i
z
l
i+1
z
r
i
(4.17)
y
i
=

y
i
y
i
(4.18)
The geometrical interpretation of the vector of coordinates 4.18 is given in Fig. 5.The
z
y
element 1 element N element i
node 1 node N node i
y
i

yi
O
i
Figure 5
coordinates y
i
,
yi
are associated with nodes which are located at the centre of gravity
of the rigid elements.
Introduction of the external forces
If there is a set of forces acting on the rigid element, each of them (e.g. F
i
) can
be equivalently replaced by the force F
i
applied to the node O
i
and the moment
M
i
= a
i
F
i
as shown in Fig. 6. This equivalent set of forces along the nodal
z
y
element i
node i
y
i

yi
O
i
a
i
F
i
F
i
M
i
Figure 6
coordinates y
i
,
yi
should be added to the mathematical model 4.19. In a general
case these forces can be independent of time (static forces) or they can depend on
time (excitation forces). Introducing notations F
s
for the static forces and F(t) for
the excitation forces, the equation of motion of the free-free beam takes the following
form.
m y +ky = F
s
+F(t) (4.19)
4.1.2 Finite Elements Method.
Inertia and stiness matrix for the free-free beam
According to the Finite Elements Method, the shaft is divided into a number of the
uniform and exible elements. The i th element is shown in Fig. 7. In this gure
z
y
l
y
y
i
i2
i1
yi1
yi2
E J
i i
A
i
i
1
2
Figure 7
E
i
, J
i
, A
i
, and
i
stand for Young modulus, second moment of area about the neutral
axis, area of cross-section and the unit mass of the element. The dierential equation
of the statically deected line of the element in the plane yz is
E
i
J
i
d
4
y(z)
dz
4
= 0 (4.20)
Integration of the above equation four times yields
y(z) =
1
6
C
1
z
3
+
1
2
C
2
z
2
+C
3
z +C
4
(4.21)
The constants of integration C
j
(j = 1, 2, 3, 4) must be chosen to fulll the following
boundary conditions
y(z)|
z=0
= y
i1
;
dy(z)
dz
z=0
=
yi1
y(z)|
z=l
i
= y
i2
;
dy(z)
dz
z=l
i
=
yi2
(4.22)
The parameters y
i1
and y
i2
are called nodal displacements and the parameters
i1
and
i2
are called nodal rotations. The nodes are denoted by numbers 1 and 2.
Introduction of the solution 4.21 into the above boundary conditions results in the
following set of algebraic equations that is linear with respect to the constants C
j
.
_
_
0 0 0 1
0 0 1 0
1
6
l
3
i
1
2
l
2
i
l
i
1
1
2
l
2
i
l
i
1 0
_
_
_
_
C
1
C
2
C
3
C
4
_
_
=
_
_
y
i1
yi1
y
i2
yi2
_
_
(4.23)
Its solution yields the integration constants C
j
.
_
_
C
1
C
2
C
3
C
4
_
_
=
_
_
0 0 0 1
0 0 1 0
1
6
l
3
i
1
2
l
2
i
l
i
1
1
2
l
2
i
l
i
1 0
_
_
1
_
_
y
i1
yi1
y
i2
yi2
_
_
=
_
_
6
l
3
i
(2y
i1
+l
i
yi1
2y
i2
+l
i
yi2
)
2
l
2
i
(3y
i1
2l
i
yi1
+ 3y
i2
l
i
yi2
)
yi1
y
i1
_
_
(4.24)
After introduction of Eq. 4.24 into the equation of the deected line 4.21 one may
get it in the following form.
y(z) =
"
1 3
z
l
i
2
+ 2
z
l
i
3
#
y
i1
+
"
z
l
i
z
l
i
2
+
z
l
i
3
#
l
i
yi1
+
"
3
z
l
i
2
2
z
l
i
3
#
y
i2
+
"
z
l
i
2
+
z
l
i
3
#
l
i
yi2
= {H(z)}
T
{y} (4.25)
where
{H(z)} =
_
_
H
1
H
2
l
i
H
3
H
4
l
i
_
_
=
_
_
1 3
z
l
i
2
+ 2
z
l
i
z
l
i
z
l
i
2
+
z
l
i
l
i
3
z
l
i
2
2
z
l
i
z
l
i
2
+
z
l
i
l
i
_
_
(4.26)
{y} =
_
_
y
i1
yi1
y
i2
yi2
_
_
(4.27)
Functions H
1
, H
2
, H
3
, H
4
(see Eq. 4.26) are known as Hermite cubics or shape func-
tions. The matrix {y} contains the nodal coordinates. As it can be seen from Eq.
4.25 the deected line of the nite element is assembled out of terms which are linear
with respect to the nodal coordinates.
If the nite element performs motion with respect to the stationary system of
coordinates xyz, it is assumed that the motion in the plane yz can be approximated
by the following equation.
y(z, t) = {H(z)}
T
{y (t)} (4.28)
As one can see from the equation 4.28, between two subsequent nodes, the dynamic
deection line is approximated by the static deection line. It should be noted that
this assumption is acceptable only if the considered element is reasonably short.
The following mathematical manipulations are aimed to replace the continues
mathematical model of the element considered
E
i
J
i
4
y(z, t)
z
4

i
2
y(z, t)
t
2
= 0 (4.29)
by its discreet representation along the nodal coordinates
[m
i
] { y (t)} + [k
i
] {y (t)} = 0. (4.30)
In the above equations
i
stands for the unit mass of the nite element and [m
i
] and
[k
i
] stands for the inertia and stiness matrix respectively. These two matrices are
going to be developed from the two following criteria:
1. The kinetic energy stored in the continues physical model of the nite
element must be equal to the kinetic energy stored in its discreet physical model.
2. The potential energy stored in the continues physical model of the nite
element must be equal to the potential energy stored in its discreet physical model.
The kinetic energy stored in the continues physical model of the nite element
is
T =
1
2
Z
l
i
0

i
y(z, t)
t
2
dz
=
1
2
Z
l
i
0

i
y(z, t)
t
y(z, t)
t
dz
=
1
2
Z
l
i
0

i
{ y (t)}
T
{H(z)}
{H(z)}
T
{ y (t)}
dz
=
1
2
{ y (t)}
T

i
Z
l
i
0
{H(z)} {H(z)}
T
dz
{ y (t)}
=
1
2
{ y (t)}
T
_
_
_
_
i
Z
l
i
0
_
_
H
2
1
H
1
H
2
l
i
H
1
H
3
H
1
H
4
l
i
H
2
H
1
l
i
H
2
2
l
2
i
H
2
H
3
l
i
H
2
H
4
l
2
i
H
3
H
1
H
3
H
2
l
i
H
2
3
H
3
H
4
l
i
H
4
H
11
l
i
H
4
H
2
l
2
i
H
4
H
3
l
i
H
2
4
l
2
i
_
_
dz
_
_
_
_
{ y (t)}
(4.31)
It is easy to see that the last row of Eq. 4.31 represents kinetic energy function
of the discreet physical model (compare with Eq. 2.137 page 103) along the nodal
coordinates y
i1

yi1
y
i2

yi2
with the following matrix of inertia.
m
i
=
i
Z
l
i
0
_
_
H
2
1
H
1
H
2
l
i
H
1
H
3
H
1
H
4
l
i
H
2
H
1
l
i
H
2
2
l
2
i
H
2
H
3
l
i
H
2
H
4
l
2
i
H
3
H
1
H
3
H
2
l
i
H
2
3
H
3
H
4
l
i
H
4
H
1
l
i
H
4
H
2
l
2
i
H
4
H
3
l
i
H
2
4
l
2
i
_
_
dz
=
i
l
i
_
_
13
35
11
210
l
i
9
70

13
420
l
i
11
210
l
i
1
105
l
2
i
13
420
l
i

1
140
l
2
i
9
70
13
420
l
i
13
35

11
210
l
i
13
420
l
i

1
140
l
2
i

11
210
l
i
1
105
l
2
i
_
_
= m
i
_
_
13
35
11
210
l
i
9
70

13
420
l
i
11
210
l
i
1
105
l
2
i
13
420
l
i

1
140
l
2
i
9
70
13
420
l
i
13
35

11
210
l
i
13
420
l
i

1
140
l
2
i

11
210
l
i
1
105
l
2
i
_
_
(4.32)
In the last formula m
i
stands for mass of the nite element.
To take advantage of the second criterion let us produce expression for the
potential energy function for the continues physical model of the nite element.
V =
1
2
Z
l
i
0
E
i
J
i
2
y(z, t)
z
2
2
dz
=
1
2
Z
l
i
0
E
i
J
i
2
y(z, t)
z
2
2
y(z, t)
z
2
dz
=
1
2
Z
l
i
0
E
i
J
i
{y (t)}
T

d
2
H(z)
dz
2
d
2
H(z)
dz
2
T
{y (t)}
!
dz
=
1
2
{y (t)}
T
"
E
i
J
i
Z
l
i
0
d
2
H(z)
dz
2
d
2
H(z)
dz
2
T
dz
#
{y (t)}
=
1
2
{y (t)}
T
_
_
_
_
E
i
J
i
Z
l
i
0
_
H
00
1
2
H
00
1
H
00
2
l
i
H
00
1
H
00
3
H
00
1
H
00
4
l
i
H
00
2
H
00
1
l
i
(H
00
2
)
2
l
2
i
H
00
2
H
00
3
l
i
H
00
2
H
00
4
l
2
i
H
00
3
H
00
1
H
00
3
H
00
2
l
i
(H
00
3
)
2
H
00
3
H
00
4
l
i
H
00
4
H
00
1
l
i
H
00
4
H
00
2
l
2
i
H
00
4
H
00
3
l
i
(H
00
4
)
2
l
2
i
_
_
dz
_
_
_
_
{y (t)}
(4.33)
As one can see from Eq. 4.33, to fulll the second criterion, the stiness matrix along
the nodal coordinates y
i1

yi1
y
i2

yi2
must be as follows.
k
i
= E
i
J
i
Z
l
i
0
_
H
00
1
2
H
00
1
H
00
2
l
i
H
00
1
H
00
3
H
00
1
H
00
4
l
i
H
00
2
H
00
1
l
i
(H
00
2
)
2
l
2
i
H
00
2
H
00
3
l
i
H
00
2
H
00
4
l
2
i
H
00
3
H
00
1
H
00
3
H
00
2
l
i
(H
00
3
)
2
H
00
3
H
00
4
l
i
H
00
4
H
00
1
l
i
H
00
4
H
00
2
l
2
i
H
00
4
H
00
3
l
i
(H
00
4
)
2
l
2
i
_
_
dz
=
E
i
J
i
l
3
i
_
_
12 6l
i
12 6l
i
6l
i
4l
2
i
6l
i
2l
2
i
12 6l
i
12 6l
i
6l
i
2l
2
i
6l
i
4l
2
i
_
_
(4.34)
Hence, the mathematical model of the element considered can be written as
m
i
y
i
+k
i
y
i
= R
yi
(4.35)
The vector R
i
represents the interaction forces between the neighborhood elements.
R
yi
=

R
yi1
R
yi1
R
yi2
R
yi2
T
(4.36)
In exactly the same manner one can create mathematical model for the next
to the right hand side element of the shaft, say element j.
m
j
y
j
+k
j
y
j
= R
yj
(4.37)
where
m
j
= m
j
_
_
13
35
11
210
l
j
9
70

13
420
l
j
11
210
l
j
1
105
l
2
j
13
420
l
j

1
140
l
2
j
9
70
13
420
l
j
13
35

11
210
l
j
13
420
l
j

1
140
l
2
j

11
210
l
j
1
105
l
2
j
_
_
k
j
=
E
j
A
j
l
3
j
_
_
12 6l
j
12 6l
j
6l
j
4l
2
j
6l
j
2l
2
j
12 6l
j
12 6l
j
6l
j
2l
2
j
6l
j
4l
2
j
_
_
(4.38)
y
j
=

y
j1

yj1
y
j2

yj2
T
(4.39)
R
yi
=

R
yj1
R
yj1
R
yj2
R
yj2
T
(4.40)
These two equations of motion (4.35 4.37), associated with the two elements i and j,
have to fulll the compatibility (continuity and equilibrium) conditions. These condi-
tions allow to join those two elements to create one mathematical model representing
both elements. In the case considered here, the compatibility conditions between the
two elements i and j correspond to the left hand side node of the element i and the
right hand side node of the element j. For these nodes the continuity conditions take
form

y
i2
yi2
=

y
j1
yj1
=

y
ij
yij
(4.41)
and the equilibrium conditions are
R
yi2
R
yi2
R
yj1
R
yj1
=

0
0
(4.42)
They results in the following mathematical model of the joint elements.
m
ij
y
ij
+k
ij
y
ij
= R
yij
(4.43)
where
m
ij
=
_
_
13
35
m
i
11
210
l
i
m
i
9
70
m
i

13
420
l
i
m
i
0 0
11
210
l
i
m
i
1
105
l
2
i
m
i
13
420
l
i
m
i

1
140
l
2
i
m
i
0 0
9
70
m
i
13
420
l
i
m
i
13
35
(m
i
+m
j
)
11
210
(l
i
m
i
+l
j
m
j
)
9
70
m
j

13
420
l
j
m
j
13
420
l
i
m
i

1
140
l
2
i
m
i
11
210
(l
i
m
i
+l
j
m
j
)
1
105
(l
2
i
m
i
+l
2
j
m
j
)
13
420
l
j
m
j

1
140
l
2
j
m
j
0 0
9
70
m
j
13
420
l
j
m
j
13
35
m
j

11
210
l
j
m
j
0 0
13
420
l
j
m
j

1
140
l
2
j
m
j

11
210
l
j
m
j
1
105
l
2
j
m
j
_
_
(4.44)
k
ij
=
_
_
12
E
i
A
i
l
3
i
6
E
i
A
i
l
2
i
12
E
i
A
i
l
3
i
6
E
i
A
i
l
2
i
0 0
6
E
i
A
i
l
2
i
4
E
i
A
i
l
i
6
E
i
A
i
l
2
i
2
E
i
A
i
l
i
0 0
129
E
i
A
i
l
3
i
6
E
i
A
i
l
2
i
12(
E
i
A
i
l
3
i
+
E
j
A
j
l
3
j
) 6(
E
i
A
i
l
2
i
+
E
j
A
j
l
2
j
) 12
E
j
A
j
l
3
j
6
E
j
A
j
l
2
j
6
E
i
A
i
l
2
i
2
E
i
A
i
l
i
6(
E
i
A
i
l
2
i
+
E
j
A
j
l
2
j
) 4(
E
i
A
i
l
i
+
E
j
A
j
l
j
) 6
E
j
A
j
l
2
j
2
E
j
A
j
l
j
0 0 12
E
j
A
j
l
3
j
6
E
j
A
j
l
2
j
12
E
j
A
j
l
3
j
6
E
j
A
j
l
2
j
0 0 6
E
j
A
j
l
2
j
2
E
j
A
j
l
j
6
E
j
A
j
l
2
j
4
E
j
A
j
l
j
_
_
(4.45)
y
ij
=

y
i1

yi1
y
ij

yij
y
j2

yj2
T
(4.46)
R
ij
=

R
yi1
R
yi1
0 0 R
yj2
R
yj2
T
(4.47)
Repetition of the described procedure to all elements of the shaft results in
the mathematical model of the shaft in the plane yz.
m y +ky = 0 (4.48)
The geometrical interpretation of the nodal coordinates appearing in the Eq. 4.46 is
given in Fig. 8.
z
y
element 1 element N-1 element i
node 1 node N node i
y
i

yi
O
i
Figure 8
The coordinates y
i
and
yi
are associated with nodes that are located at the
ends of the nite elements.
Introduction of the external forces
Since the nite element is considered elastic, the treatment of the external forces
presented in the previous section can not be applied. In this case one has to take
advantage of the principle of the virtual work. It says that the virtual work produced
by a force F
i
(see Fig. 9) on the displacement y
i
is equal to the virtual work produced
by a set of forces along the coordinates y
i1

yi1
y
i2

yi2
. Hence
W
i
= F
i
y
i
(a
i
, t) = {F
i1
, M
i1
, F
i2
, M
i2
}
_
_
y
i1
yi1
y
i2
yi2
_
_
(4.49)
But according to 4.28
z
y
l
y
y
i
i2
i1

yi1

yi2
E J
i i
A
i

i
1
2
F
i
a
i
Figure 9
y(a
i
, t) = {H(a
i
)}
T
{y (t)} =
_
_
1 3
a
i
l
i
2
+ 2
a
i
l
i
a
i
l
i
a
i
l
i
2
+
a
i
l
i
l
i
3
a
i
l
i
2
2
a
i
l
i
a
i
l
i
2
+
a
i
l
i
l
i
_
_
T
_
_
y
i1
yi1
y
i2
yi2
_
_
(4.50)
Introduction of the above expression into the expression for the virtual work yields
W
i
= F
i

_
_
1 3
a
i
l
i
2
+ 2
a
i
l
i
a
i
l
i
a
i
l
i
2
+
a
i
l
i
l
i
3
a
i
l
i
2
2
a
i
l
i
a
i
l
i
2
+
a
i
l
i
l
i
_
_
T
_
_
y
i1
yi1
y
i2
yi2
_
_
(4.51)
Hence, the vector of forces along the nodal coordinates is
_
_
F
i1
M
i1
F
i2
M
i2
_
_
= F
i

_
_
1 3
a
i
l
i
2
+ 2
a
i
l
i
a
i
l
i
a
i
l
i
2
+
a
i
l
i
l
i
3
a
i
l
i
2
2
a
i
l
i
a
i
l
i
2
+
a
i
l
i
l
i
_
_
(4.52)
This forces have to be introduced into the equation of motion 4.48
m y +ky = F
s
+F(t) (4.53)
where, similarly as before, F
s
stands for the static forces and F(t) stands for the
excitation forces.
BOUNDARY CONDITIONS 186
4.2 BOUNDARY CONDITIONS
O
z
O
i
y
y
i

y i
B
i
b
y
i
Figure 10
Let us assume that the free-free beam is rigidly supported upon several sup-
ports B
i
(see Fig. 10). The instantaneous position of these supports is determine
with respect to the stationary system of coordinates xyz by coordinates b
y
i
(t) . Let
us denote by b vector of such coordinates.
b =
_
_
.
.
.
b
y
i
(t)
.
.
.
_
_
(4.54)
Let us reorganize vector of coordinates of the shaft
y =

xN
, y
1
y1
, ......y
N
yN
T
(4.55)
in such a way that its upper part y
b
contains coordinates along which the shaft is
rigidly supported and its lower part y
r
contains all the remaining coordinates
y =

y
b
y
r
(4.56)
Let us assume that the mathematical model of the beam
m y +ky = F (4.57)
is organized with respect to the above vector y of coordinates.
m
bb
m
br
m
rb
m
rr

y
b
y
r
k
bb
k
br
k
rb
k
rr

y
b
y
r
=

F
b
F
r
(4.58)
Partitioning of the above equations results in the following set of equations
m
bb
y
b
+m
br
y
r
+k
bb
y
b
+k
br
y
r
= F
b
m
rb
y
b
+m
rr
y
r
+k
rb
y
b
+k
rr
y
r
= F
r
(4.59)
CONDENSATION OF THE DISCREET SYSTEMS 187
Motion of the beam along the coordinates y
b
is determined by the boundary condi-
tions 4.54. Hence, the vector y
b
in the mathematical model 4.59 must be replaced by
b.
m
bb
b +m
br
y
r
+k
bb
b +k
br
y
r
= F
b
m
rb
b +m
rr
y
r
+k
rb
b +k
rr
y
r
= F
r
(4.60)
The second equation governs motion of the supported beam and can be rewritten as
follows
m
rr
y
r
+k
rr
y
r
= F
r
m
rb
b
b
k
rs
b
b
(4.61)
The last two terms represent the kinematic excitation of the beam cause by motion
of its supports. The vector b, in a general case, contains elements that are known
functions of time. Hence the above equation can be solved for motion along the
coordinates y
r
. Let
y
r
= Y
r
(t) (4.62)
be a solution of this equation. This solution approximate motion of the beam along
the remaining coordinates y
r
.
The vector F
b
in the rst equation of the set 4.60 represents forces of the
interaction between the moving beam and its supports. These interaction forces can
be now determined.
F
b
= m
bb
b +m
br
Y
r
+k
bb
b +k
br
Y
r
(t) (4.63)
4.3 CONDENSATION OF THE DISCREET SYSTEMS
In many engineering problems, due to a large number of the uniform sections of the
element to be modelled, number of the nal elements is large too. It follows that the
size of the matrices involved in the discreet mathematical model
m y +c y +ky = F (4.64)
can be too large to carry out the necessary analysis of the mathematical model. In this
section a procedure for reducing the size of mathematical models will be developed.
Let us assume, that the equation 4.64 is arranged in such a way that the
coordinates, which are to be eliminated due to the condensation procedure y
e
, are
located in the upper part of the vector y and these, which are to be retained for
further consideration y
r
, are located in its lower part.
y =

y
e
y
r
(4.65)
Partitioning of the equations 4.64 yields
m
ee
m
er
m
re
m
rr

y
e
y
r
c
ee
c
er
c
re
c
rr

y
e
y
r
k
ee
k
er
k
re
k
rr

y
e
y
r
=

F
e
F
r
(4.66)
To eliminate the coordinates y
e
from the mathematical model 4.66, one have to
determine the relationship between the coordinates y
e
and the coordinates y
r
. One
of many possibilities is to assume that the coordinates y
e
are obeyed to the static
relationship.

k
ee
k
er
k
re
k
rr

y
e
y
r
=

0
0
(4.67)
Hence, upon partitioning equation 4.67 one may obtain
k
ee
y
e
+k
er
y
r
= 0 (4.68)
Therefore the sought relationship is
y
e
= hy
r
(4.69)
where
h = k
1
ee
k
er
(4.70)
Once the relationship is established, one may formulate the following criteria of con-
densation:
1. Kinetic energy of the system before and after condensation must be the
same.
2. Dissipation function of the system before and after condensation must be
the same.
3. Potential energy of the system before and after condensation must be the
same.
4. Virtual work done by all the external forces before and after condensation
must be the same.
4.3.1 Condensation of the inertia matrix.
According to the rst criterion, the kinetic energy of the system before and after
condensation must be the same. The kinetic energy of the system before condensation
is
T =
1
2
y
T
e
y
T
r
m
ee
m
er
m
re
m
rr

y
e
y
r
=
1
2
y
T
e
m
ee
y
e
+ y
T
e
m
er
y
r
+ y
T
r
m
re
y
e
+ y
T
r
m
rr
y
r
(4.71)
Introduction of 4.69 yields
T =
1
2
[h y
r
]
T
m
ee
h y
r
+ [h y
r
]
T
m
er
y
r
+ y
T
r
m
re
h y
r
+ y
T
r
m
rr
y
r
=
1
2
y
T
r
h
T
m
ee
h y
r
+ y
T
r
h
T
m
er
y
r
+ y
T
r
m
re
h y
r
+ y
T
r
m
rr
y
r
=
1
2
y
T
r
h
T
m
ee
h +h
T
m
er
+m
re
h +m
rr
y
r
(4.72)
Hence, if the kinetic energy after condensation is to be the same, the inertia matrix
after condensation m
c
must be equal to
m
c
= h
T
m
ee
h +h
T
m
er
+m
re
h +m
rr
(4.73)
4.3.2 Condensation of the damping matrix.
Since formula for the dissipation function is of the same form as formula for the
kinetic energy, repetition of the above derivation leads to the following denition of
the condensed damping matrix
c
c
= h
T
c
ee
h +h
T
c
er
+c
re
h +c
rr
(4.74)
4.3.3 Condensation of the stiness matrix.
Taking advantage from denition of potential energy of the system considered
V =
1
2
y
T
e
y
T
r
k
ee
k
er
k
re
k
rr

y
e
y
r
=
1
2
y
T
e
k
ee
y
e
+y
T
e
k
er
y
r
+y
T
r
k
re
y
e
+y
T
r
k
rr
y
r
(4.75)
one can arrive to conclusion that the condensed stiness matrix is of the form 4.76
k
c
= h
T
k
ee
h +h
T
k
er
+k
re
h +k
rr
(4.76)
It is easy to show that sum of the rst two terms in the above expression is equal to
zero. Indeed, according to 4.70, they can be transformed as following.
h
T
k
ee
h +h
T
k
er
= (k
1
ee
k
er
)
T
k
ee
(k
1
ee
k
er
) + (k
1
ee
k
er
)
T
k
er
= (k
1
ee
k
er
)
T
k
er
+ (k
1
ee
k
er
)
T
k
er
= 0 (4.77)
Hence,
k
c
= k
re
h +k
rr
(4.78)
4.3.4 Condensation of the external forces.
The virtual work performed by external forces F on the displacements y is
W =

y
T
e
y
T
r
F
e
F
r
= y
T
e
F
e
+y
T
r
F
r
(4.79)
Introduction of 4.69 into the above equation yields
W = (hy
r
)
T
F
e
+y
T
r
F
r
=

y
T
r
h
T
F
e
+y
T
r
F
r
= y
T
r
h
T
F
e
+F
r
(4.80)
Hence,
F
c
=
h
T
F
e
+F
r
(4.81)
The condensed mathematical model, according to the above consideration, can
be adopted as follows
m
c
y
c
+c
c
y
c
+k
c
y
c
= F
c
(4.82)
where
y
c
= y
r
(4.83)
After the equation 4.82 is solved, the relationship 4.69
y
e
= hy
c
permits to produce displacement along the cut o coordinates.
PROBLEMS 190
4.4 PROBLEMS
Problem 38
Produce the natural frequencies and the corresponding natural modes for the
xed-elastically supported uniform beam shown in Fig. 11.
z
l
k
y
O
Figure 11
The exact solution of this problem is presented in page 166 for the following
data
E = 2.1 10
11
N/m
2
= 7800kg/m
3
A = 0.03 0.01 = 0.0003m
2
J =
0.030.01
3
12
= 2.5 10
9
m
4
k = 10000N/m
l = 1m
=
EJ
k
=
2.110
11
2.510
9
10000
= 0.0525
Use this data to produce the solution by means of approximation of this beam with
10 nite elements.
PROBLEMS 191
Solution
To create the mathematical model of the free-free beam, it was divided into
ten nite elements as shown in Fig. 12. The computed mathematical model is

z
l
y
O
y
1

1
y
2

2
y
11

11
Figure 12
m y +ky = 0; y ={y
1
,
1
, y
2
,
2
, ......, y
11
,
11
}
T
(4.84)
The inuence of the spring can be represented by the force ky
11
acting along the
coordinate y
11
(see Fig. 13).

z
l
-ky
11
y
O
y
11
11

Figure 13
This force should be introduced to the right hand side of the equation 4.84.
m y +ky =
_
_
0
0
....
ky
11
0
_
_
=
_
_
0 0 .... .... 0
0 0 .... .... 0
.... .... .... .... ....
0 0 .... k 0
0 0 .... 0 0
_
_
_
_
y
1
1
....
y
11
11
_
_
= k
1
y (4.85)
Therefore, the equation of the beam supported by the spring is
m y +k
s
y = 0 (4.86)
PROBLEMS 192
where
k
s
= k +k
1
= k+
_
_
0 0 .... .... 0
0 0 .... .... 0
.... .... .... .... ....
0 0 .... k 0
0 0 .... 0 0
_
_
(4.87)
To introduce the boundary conditions associated with the left hand side of the beam,
let us partition the above mathematical model in such a manner that all the coordi-
nates involved in this boundary conditions are included in the vector y
1
.
m
11
m
12
m
21
m
22

y
1
y
2
k
s11
k
s12
k
s21
k
s22

y
1
y
2
=0 (4.88)
where
y
1
= {y
1
,
1
}
T
y
2
= {y
2
,
2
, ...., y
11
,
11
}
T
(4.89)
z
l
-ky
11
y
O
y
1
=0
1
=0
y
2
2

y
11
11

R
M
Figure 14
According to the boundary conditions (see Fig. 14)
y
1
= {y
1
,
1
}
T
= {0, 0}
T
(4.90)
and
R
1
= {R, M}
T
(4.91)
Introduction of 4.90 and 4.91 into 4.88 yields
m
11
m
12
m
21
m
22

0
y
2
k
s11
k
s12
k
s21
k
s22

0
y
2
R
1
0
(4.92)
This equation is equivalent to two equations as follows
m
12
y
2
+k
s12
y
2
= R
1
(4.93)
m
22
y
2
+k
s22
y
2
= 0 (4.94)
PROBLEMS 193
The second equation 4.94 is the equation of motion of the supported beam. It was
solved for the natural modes and the natural frequencies. Results of this computation
is shown in Fig. 15 by boxes and in the rst column of the Table below. This results
are compare with natural modes (continuous line in Fig. 15) and natural frequencies
(second column in the Table) obtained by solving the continuous mathematical model
( see problem page 165). The equation 4.93 allows the vector of the interation forces
R
1
to be computed.

- 2
- 1
0
1
2

0.2

0.4 0.6 0.8 1
z

Y
1
(z)
z
l
k
y
O
Y
3
(z) Y
2
(z)
Figure 15
Table
natural frequenciesof
the descreet system
[1/sec]
natural frequencies of
the continuous system
[1/sec]
1 129.5 129.65
2 357.6 357.3
3 933.4 932.0
PROBLEMS 194
Problem 39
The mathematical model of a free-free beam shown in Fig. 16 along coordi-
nates x
1
, x
2
, x
3
, x
4
is as follows
m x +kx = 0 (4.95)
m =
_
_
m
11
m
12
m
13
m
14
m
21
m
22
m
23
m
24
m
31
m
32
m
33
m
34
m
41
m
42
m
43
m
44
_
_
; k =
_
_
k
11
k
12
k
13
m
14
k
21
k
22
k
23
k
24
k
31
k
32
k
33
k
34
k
41
k
42
k
43
k
44
_
_
; x =
_
_
x
1
x
2
x
3
x
4
_
_
(4.96)

x
1
x
2
x
3
x
4
1 2
3 4
Figure 16
This beam is supported upon three rigid pedestals along coordinates x
1
, x
2
, x
3
as shown in Fig. 17.
1 2
3
Z
X
a
3
a sint
2
Figure 17
The motion of these supports with respect to the inertial system of coordinate
XZ is given by the following equations
X
1
= 0
X
2
= a
2
sint
X
3
= a
3
(4.97)
Derive expressions for :
1. the static deection curve,
2. the interaction forces between the beam and the supports
PROBLEMS 195
Solution
Partitioning of the equations 4.95 with respect to the vector of boundary
conditions 4.97 results in the following equation
m x +kx = R (4.98)
where

m
11
m
12
m
21
m
22

x
1
x
2
k
11
k
12
k
21
k
22

x
1
x
2
=

R
1
R
2
(4.99)
m
11
=
_
_
m
11
m
12
m
13
m
21
m
22
m
23
m
31
m
32
m
33
_
_
; m
12
=
_
_
m
14
m
24
m
34
_
_
; m
21
=
m
41
m
42
m
43
; m
22
= m
44
(4.100)
k
11
=
_
_
k
11
k
12
k
13
k
21
k
22
k
23
k
31
k
32
k
33
_
_
; k
12
=
_
_
k
14
k
24
k
34
_
_
; k
21
=
k
41
k
42
k
43
; k
22
= k
44
(4.101)
x
1
=
_
_
x
1
x
2
x
3
_
_
; x
2
=x
4
; R
1
=
_
_
R
1
R
2
R
3
_
_
; R
2
=0 (4.102)
or
m
11
x
1
+m
12
x
2
+k
11
x
1
+k
12
x
2
= R
1
(4.103)
m
21
x
1
+m
22
x
2
+k
21
x
1
+k
22
x
2
= 0 (4.104)
Introduction of boundary conditions 4.97 into the equation 4.104 yields
m
44
x
4
+k
44
x
4
=
m
41
m
42
m
43
_
_
x
1
x
2
x
3
_
_
k
41
k
42
k
43
_
_
x
1
x
2
x
3
_
_
(4.105)
where
_
_
x
1
x
2
x
3
_
_
=
_
_
0
a
2
2
sint
0
_
_
;
_
_
x
1
x
2
x
3
_
_
=
_
_
0
a
2
sin t
a
3
_
_
(4.106)
or
m
44
x
4
+k
44
x
4
= (m
42
a
2
2
k
42
a
2
) sin t k
43
a
3
(4.107)
The static deection is due to the time independent term k
43
a
3
in the right hand
side of the equation 4.107.
m
44
x
4
+k
44
x
4
= k
43
a
3
(4.108)
The particular solution of the equation 4.108 is
x
4
= x
s
(4.109)
k
44
x
s
= k
43
a
3
(4.110)
PROBLEMS 196
x
s
=
k
43
a
3
k
44
(4.111)
Its graphical representation is given in Fig. 18
1
2
3
Z
X
a
3
4
x
s
Figure 18
The forced response due to motion of the support 2 (X
2
= a
2
sint) is repre-
sented by the particular solution due to the time dependant term.
m
44
x
4
+k
44
x
4
= (m
42
a
2
2
k
42
a
2
) sin t (4.112)
For the above equation, the particular solution may be predicted as follows
x
4
= x
d
sint (4.113)
Implementation of the solution 4.113 into the equation 4.112 yields the wanted am-
plitude of the forced vibration x
d
.
x
d
=
(m
42
a
2
2
k
42
a
2
)
2
m
44
+k
44
(4.114)
The resultant motion of the system considered is shown in Fig. 19
1 2
3
Z
X
a
3 4
x
s
a sin t
2
x sin t
d
Figure 19
This motion causes interaction forces along these coordinates along which the
system is attached to the base. These forces can computed from equation 4.103.
m
11
x
1
+m
12
x
2
+k
11
x
1
+k
12
x
2
= R
1
(4.115)
In this equation x
1
stands for the given boundary conditions
x
1
=
_
_
x
1
x
2
x
3
_
_
=
_
_
0
a
2
sint
a
3
_
_
; x
1=
_
_
x
1
x
2
x
3
_
_
=
_
_
0
a
2
2
sin t
0
_
_
(4.116)
PROBLEMS 197
and x
2
represents, known at this stage, motion of the system along the coordinate 4
x
2
= x
4
= x
s
+x
d
sin t x
2
= x
4
=
d
2
dt
2
(x
s
+x
d
sin t) = x
d
2
sint; (4.117)
Hence, the wanted vector of interaction forces is as follows
R
1
= m
11
_
_
0
a
2
2
sint
0
_
_
+k
11
_
_
0
a
2
sin t
a
3
_
_
+m
12
(x
d
2
sin )+k
12
(x
s
+x
d
sin t)
(4.118)
Part II
EXPERIMENTAL
INVESTIGATION
198
Chapter 5
MODAL ANALYSIS OF A SYSTEM WITH 3 DEGREES OF
FREEDOM
5.1 DESCRIPTION OF THE LABORATORY INSTALLATION
1 2 3 4 5 6 7 8
9 10 12
Figure 1
The vibrating object 2, 3, and 4 (see Fig.1) is attached to the base 1. It consists
of the three rectangular blocks 2 joint together by means of the two springs 3. The
spaces between the blocks 2 are lled in with the foam 4 in order to increase the
structural damping. The transducer 5 allows the acceleration of the highest block to
be measured in the horizontal direction. The hammer 6 is used to induce vibrations
of the object. It is furnished with the piezoelectric transducer 7 that permits the
impulse of the force applied to the object to be measured. The rubber tip 8 is used
MODELLING OF THE OBJECT 200
to smooth and extend the impulse of force. Both, the acceleration of the object and
the impulse of the force can be simultaneously recorded and stored in the memory
of the spectrum analyzer 10. These data allow the transfer functions to be produced
and sent to the personal computer 11 for further analysis.
5.2 MODELLING OF THE OBJECT

x
3
m
3
k
3
c
3
x
2
m
2
k
2
c
2
x
1
m
1
k
1
c
1
Figure 2
The base 1, which is considered rigid and motionless, forms a reference system
for measuring its vibrations. The blocks 2 are assumed to be rigid and the springs 3 are
by assumption massless. Motion of the blocks is restricted to one horizontal direction
only. Hence, according to these assumptions, the system can be approximated by
three degrees of freedom physical model. The three independent coordinates x
1
, x
2
and x
3
are shown in Fig.2. Magnitudes of the stuness k
1
, k
2
and k
3
of the springs
can be analytically assessed. To this end let us consider one spring shown in Fig. 3
The dierential equation of the deection of the spring is
EJ
d
2
x
dz
2
= M Fz =
FH
2
Fz (5.1)
Double integration results in the following equation of the bending line.
EJ
dx
dz
=
FH
2
z
F
2
z
2
+A (5.2)
EJx =
FH
4
z
2
F
6
z
3
+Az +B (5.3)
MODELLING OF THE OBJECT 201
z
x
H
x(H)
z
w
t
F
F
M=FH/2
M=FH/2
EJ
Figure 3
Taking advantage of the boundary conditions associated with the lower end of the
spring, one can arrived to the following expression for the bending line.
x =
1
EJ
FH
4
z
2
F
6
z
3
(5.4)
Hence, the deection of the upper end is
x(H) =
1
EJ
FH
4
H
2
F
6
H
3
=
1
12EJ
FH
3
(5.5)
Therefore the stiness of one spring is
k =
F
x(H)
=
12EJ
H
3
(5.6)
where
J =
wt
3
12
(5.7)
Since we deal with a set of two springs between the blocks, the stiness k
i
shown in
the physical model can be computed according to the following formula.
k
i
=
24E
i
J
i
H
3
i
(5.8)
Application of the Newtons equations to the developed physical model results in the
following set of dierential equations
m
1
x
1
+ (c
1
+c
2
) x
1
+ (c
2
) x
2
+ (k
1
+k
2
)x
1
+ (k
2
)x
2
= F
1
m
2
x
2
+ (c
2
) x
1
+ (c
2
+c
3
) x
2
+ (c
3
) x
3
+ (k
2
)x
1
+ (k
2
+k
3
)x
2
+ (k
3
)x
3
= F
2
m
3
x
3
+ (c
3
) x
2
+c
3
x
3
+ (k
3
)x
2
+k
3
x
3
= F
3
ANALYSIS OF THE MATHEMATICAL MODEL 202
These equations can be rewritten as following
m x +c x +kx = F (5.9)
where
m =
_
_
m
1
0 0
0 m
2
0
0 0 m
3
_
_
; c =
_
_
c
1
+c
2
c
2
0
c
2
c
2
+c
3
c
3
0 c
3
c
3
_
_
k =
_
_
k
1
+k
2
k
2
0
k
2
k
2
+k
3
k
3
0 k
3
k
3
_
_
; x =
_
_
x
1
x
2
x
3
_
_
; F =
_
_
F
1
F
2
F
3
_
_
(5.10)
The vector F represents the external excitation that can be applied to the system.
5.3 ANALYSIS OF THE MATHEMATICAL MODEL
5.3.1 Natural frequencies and natural modes of the undamped system.
The matrix of inertia and the matrix of stiness can be assessed from the dimensions
of the object. Hence, the natural frequencies and the corresponding natural modes
of the undamped system can be produced. Implementation of the particular solution
x = Xcos t (5.11)
into the equation of the free motion of the undamped system
m x +kx = F (5.12)
results in a set of the algebraic equations that are linear with respect to the vector
X.

2
m+k
X = 0 (5.13)
Solution of the eigenvalue and eigenvector problem yields the natural frequencies and
the corresponding natural modes.
1
,
2
,
3
(5.14)
=[
1
,
2
,
2
] (5.15)
For detailed explanation see pages 96 to 99
5.3.2 Equations of motion in terms of the normal coordinates - transfer
functions
If one assume that the damping matrix is of the following form
c =m+k (5.16)
the equations of motion 5.9 can be expressed in terms of the normal coordinates
=
1
x (see section normal coordinates - modal damping page 99)

n
+ 2
n
n

n
+
2
n
n
=
T
n
F(t), n = 1, 2, 3 (5.17)
ANALYSIS OF THE MATHEMATICAL MODEL 203
The response of the system along the coordinate x
p
due to the harmonic excitation
F
q
e
it
q
, according to the formula 2.129 (page 101), is
x
p
= e
it
N
X
n=1
pn
qn
F
q
2
n
2
+ 2
n
n
i
(5.18)
Hence the acceleration along the coordinate x
p
as the second derivative with respect
to time, is
x
p
=
2
e
it
N
X
n=1
pn
qn
F
q
2
n
2
+ 2
n
n
i
(5.19)
It follows that the transfer function between the coordinate x
p
and x
q
, according to
2.131 is
R
pq
(i) =
x
p
F
q
e
it
=
2
x
p
F
q
e
it
=
=
2
N
X
n=1

pn
qn
(
2
n
2
)
(
2
n
2
)
2
+ 4
2
n
2
n
2
+
2
pn
qn
n
i
(
2
n
2
)
2
+ 4
2
n
2
n
q = 1, 2, 3
(5.20)
The modal damping ratios
1
,
2
and
3
are unknown and are to be identied by tting
the analytical transfer functions into the experimental ones. Since the transducer
5 (Fig. 1) produces acceleration, the laboratory installation permits to obtain the
acceleration to force transfer function. The theory on the experimental determination
of the transfer functions is given in the section Experimental determination of the
transfer functions (page 94).
5.3.3 Extraction of the natural frequencies and the natural modes from
the transfer functions
The problem of determination of the natural frequencies and the natural modes from
the displacement - force transfer functions was explained in details in section
Determination of natural frequencies and modes from the transfer functions (page
101). Let us do similar manipulation on the acceleration - force transfer function.
First of all let us notice that
if
=
n
R
pq
(i
n
)
=
2
pn
qn
(
2
n
2
)
4
2
n
2
n
2
+

pn
qn
i
2
n
q = 1, 2, 3 (5.21)
Since the real part of the transfer function is equal to zero for =
n
, its absolute
value is equal to the absolute value of the imaginary part.
|R
pq
(i
n
)|
pn
qn
2
n
q = 1, 2, 3 (5.22)
and phase for =
n
= arctan
Im(R
pq
(i
n
))
Re(R
pq
(i
n
))
= arctan = 90
o
(5.23)
EXPERIMENTAL INVESTIGATION 204
Hence, the frequencies corresponding to the phase 90
o
are the wanted natural
frequencies
n
.
Because
n
and
pn
are constants, magnitudes of the absolute value of the
transfer functions for =
n
represents the modes
1n
,
2n
,
3n
associated with the
n th natural frequency. An example of extracting the natural frequency and the
corresponding natural mode from the transfer function is shown in Fig. 4
0
0.00005
0.0001
0.00015
0.0002
0.00025
1500 1600 1700 1800 frequency rad/s
transfer functions m/N (modulus)
R(1,1)
R(1,2)
R(1,3)
-4
-3
-2
-1
0
1
2
3
4
1500 1600 1700 1800
frequency rad/s
R(1,1)
R(1,2)
R(1,3)
/2
/2
natural frequency
natural mode
1
2
Figure 4
5.4 EXPERIMENTAL INVESTIGATION
5.4.1 Acquiring of the physical model initial parameters
The physical model is determined by the following parameters
m
1
, m
2
m
3
- masses of the blocks
k
1
, k
2
k
3
- stiness of the springs
c
1
, c
2
c
3
- damping coecients
The blocks were weighted before assembly and their masses are
m
1
= 0.670kg
m
2
= 0.595kg
m
3
= 0.595kg
The formula 5.7 and 5.8
k
i
=
24EJ
H
3
i
; J =
wt
3
12
(5.24)
allows the stiness k
i
to be computed.
The following set of data is required
E = 0.21 10
12
N/m
2
w = ......................m to be measured during the laboratory session
t = ......................m to be measured during the laboratory session
H
1
= ......................m to be measured during the laboratory session
H
2
H
3
The damping coecients c
i
are dicult to be assessed. Alternatively the damping
properties of the system can be uniquely dened by means of the three modal damp-
ing ratios
1
,
2
and
3
(see equation 5.17). = 1 corresponds to the critical damping.
Inspection of the free vibrations of the object lead to the conclusion that the damp-
ing is much smaller then the critical one. Hence, as the rst approximation of the
damping, let us adopt the following damping ratios
1
= 0.01
2
= 0.01
2
= 0.01
5.4.2 Measurements of the transfer functions
According to the description given in section Experimental determination of the trans-
fer functions (page 94) to produce the transfer function R
pq
(i) you have to measure
response of the system along the coordinate x
p
due to impulse along the coordinate
x
q
. Since the transducer 5 (Fig. 1) is permanently attached to the mass m
3
and the
impulse can be applied along the coordinates x
1
, x
2
or x
3
, the laboratory installation
permits the following transfer functions to be obtained.
R
31
(i) R
32
(i) R
33
(i) (5.25)
The hammer 6 should be used to introduce the impulse. To obtain a reliable result,
10 measurements are to be averaged to get one transfer function. These impulses
should be applied to the middle of the block. The spectrum analyzer must show the
waiting for trigger sign before the subsequent impulse is applied.
As the equipment used is delicate and expensive, one has to observe the fol-
lowing;
1. always place the hammer on the pad provided when it is not used
2. when applying the impulse to the object make sure that the impulse is not
excessive
Harder impact does not produce better results.
5.4.3 Identication of the physical model parameters
In a general case, the identication of a physical model parameters from the transfer
functions bases on a very complicated curve tting procedures. In this experiment,
to t the analytical transfer functions into the experimental one, we are going to use
the trial and error method. We assume that the following parameters
m
1
, m
2
, m
3
, H
1
, H
2
, H
3
, w, E (5.26)
were assessed with a sucient accuracy. Uncertain are
t,
1
,
2
,
3
(5.27)
Use the parameter t to shift the natural frequencies (increment of t results in shift
of the natural frequencies to the right). Use the parameters
i
to align the picks of
the absolute values of the transfer functions (increment in the modal damping ratio
results in lowering the pick of the analytical transfer function). Work on one (say
R
33
(i) transfer function only.
WORKSHEET 207
5.5 WORKSHEET
1. Initial parameters of physical model
Measure the missing parameters and insert them to the table below
m
1
= 0.670kg
Mass of the block
1
H
1
= ..................m
length of the spring
1
m
2
= 0.595kg
Mass of the block
2
H
2
= ..................m
2
m
1
= 0.595kg
Mass of the block
3
H
3
= ..................m
3
1
= 0.01
damping ratio
of mode 1
E = 0.21 10
12
N/m
2
Youngs
modulus
2
= 0.01
damping ratio
of mode 2
w = .....................m
width of the
springs
3
= 0.01
damping ratio
of mode 3
t = .....................m
thickness of the
springs
Run program Prac3
and choose menu Input data to enter the above data.

Set excitation coordinate 3, response coordinate 3.
Save the initial data.
2. Experimental acceleration-force transfer functions R
33
(i)
Choose menu Frequency response measurements
Set up the spectrum analyzer by execution of the sub-menu Setup analyzer
Choose sub-menu Perform measurement, execute it and apply 10 times im-
pulse along the coordinates 3
Choose sub-menu Time/Frequency domain toggle to see the measured trans-
fer function
Choose sub-menu Transfer TRF to computer and execute it
Exit menu Frequency response measurements
Choose Response display/plot to display the transfer functions
3. Identication of the thickness t and the modal damping ratios
i
You can see both the experimental and analytical transfer function R
33
(i).
By varying t,
1
,
2
,
3
in the input data, try to t the analytical data into the
experimental one. Use the parameter t to shift the natural frequencies (increment
of t results in shift of the natural frequencies to the right). Use the parameters
i
to align the picks of the absolute values of the transfer functions (increment in the
modal damping ratio results in lowering the pick of the analytical transfer function).
program designed by Dr. T. Chalko

WORKSHEET 208
Record the identied parameters in the following table
m
1
= 0.670kg
Mass of the block
1
H
1
= ..................m
1
m
2
= 0.595kg
Mass of the block
2
H
2
= ..................m
2
m
1
= 0.595kg
Mass of the block
3
H
3
= ..................m
3
1
= 0.01
damping ratio
of mode 1
E = 0.21 10
12
N/m
2
Youngs
modulus
2
= 0.01
damping ratio
of mode 2
w = .....................m
width of the
springs
3
= 0.01
damping ratio
of mode 3
t = .....................m
thickness of the
springs
Save the identied parameters.
Plot the analytical and the experimental transfer function R
33
(i)
4. Experimental and analytical transfer functions R
31
(i) and R
32
(i)
Choose menu Input data and set the excitation coordinate to 1 and the
response coordinate to 3
Repeat all steps of the section 2
Plot the transfer function R
31
(i)
Choose menu Input data and set the excitation coordinate to 2 and the
response coordinate to 3
Repeat all steps of the section 2
Plot the transfer function R
32
(i)
5. Natural frequencies and the corresponding natural modes
Choose menu Mode shapes display/plot to produce the analytical frequencies
and modes
Plot the natural modes
From plots of the experimental transfer functions R
31
(i), R
32
(i), R
33
(i)
determine the natural frequencies and the natural modes
Insert the experimental and analytical frequencies into the table below
natural frequency
1
natural frequency
2
natural frequency
3
analytical ......................... ......................... .........................
experimental ......................... ......................... .........................
6. Conclusions

Mechanical Vibration

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mechanical Vibration

Uploaded by

Copyright:

Available Formats

436-431 MECHANICS 4

cos t = q sint (1.73)

, is called the magnication factor, Its magnitude and the phase as

.00 384 .00 112i

Similarly, one can obtained motion due to the excitation F

.00 38 cos 30t +.00 11 sin 30t 3. 15 10

.00 38 cos 30t+.00 11 sin30t3. 1510

.00 38 cos 30t+.00 11 sin 30t3. 1510

112. 27 = 10. 596 (2.198)

+f(z, t)dz (3.2)

dz = f(z, t)dz (3.9)

= f(z, t)dz (3.11)

and choose menu Input data to enter the above data.

program designed by Dr. T. Chalko

You might also like