436-439 DYNAMICS OF ROTORS

J.M. KRODKIEWSKI

2007

THE UNIVERSITY OF MELBOURNE

Department of Mechanical and Manufacturing Engineering

1
 2

DYNAMICS OF ROTORS

Copyright C 2007 by J.M. Krodkiewski

ISBN 0-7325-1535-8

The University of Melbourne

Department of Mechanical and Manufacturing Engineering
 CONTENTS

I MODELLING 11
1 MATHEMATICAL MODELS OF SUB-SYSTEMS 12
1.1 MATHEMATICAL MODELS OF RIGID ELEMENTS . . . . . . . . 12
1.1.1 Equations of motion of symmetric rigid bodies . . . . . 12
1.1.2 Equations of motion of non-symmetric bodies . . . . . . 17
1.1.3 Modelling of residual unbalance . . . . . . . . . . . . . . 23
1.1.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

1.2 MATHEMATICAL MODELS OF SHAFTS . . . . . . . . . . . . . . 38

1.2.1 Modelling of dynamic properties of the free-free shaft 38
1.2.2 Modelling of residual permanent deformation . . . . . . 49
1.2.3 Modelling of residual unbalance . . . . . . . . . . . . . . 52
1.2.4 Interaction forces . . . . . . . . . . . . . . . . . . . . . . . 53
1.2.5 Boundary conditions . . . . . . . . . . . . . . . . . . . . . 54
1.2.6 Constraints imposed by rigid bodies . . . . . . . . . . . . 55
1.2.7 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . 61
1.3 MATHEMATICAL MODELS OF BEARINGS . . . . . . . . . . . . . 71
1.3.1 Oil journal bearings . . . . . . . . . . . . . . . . . . . . . . 71
1.3.2 Magnetic bearings . . . . . . . . . . . . . . . . . . . . . . . 84
1.3.3 Rolling-elements bearings . . . . . . . . . . . . . . . . . . 91
1.3.4 Squeeze film dampers . . . . . . . . . . . . . . . . . . . . . 94
1.3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
1.3.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

1.4 MATHEMATICAL MODELS OF FOUNDATIONS (ENVIRONMENT)108

1.4.1 Determination of dynamic properties by means of ex-
perimental investigation . . . . . . . . . . . . . . . . . . . 108
1.4.2 Modelling of environmental excitation . . . . . . . . . . 110
1.4.3 Modelling of configuration . . . . . . . . . . . . . . . . . . 111
1.4.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

2 COMPOSITION OF MATHEMATICAL MODEL OF ROTOR SYS-

TEMS 115
2.1 MATHEMATICAL MODEL OF TWO SUB-SYSTEMS . . . . . . . 115
2.2 CONDENSATION OF MATHEMATICAL MODELS . . . . . . . . . 117
CONTENTS 4

2.2.1 Condensation of the inertia matrix . . . . . . . . . . . . 118

2.2.2 Condensation of the damping matrix . . . . . . . . . . . 119
2.2.3 Condensation of the stiffness matrix . . . . . . . . . . . . 119
2.2.4 Condensation of the external forces . . . . . . . . . . . . 119
2.3 CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
2.4 PROBLEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

II ANALYSIS 129
3 ANALYSIS OF MATHEMATICAL MODELS - GENERAL RE-
MARKS 130
3.1 EQUILIBRIUM POSITION . . . . . . . . . . . . . . . . . . . . . . . 130
3.2 RESPONSE TO EXTERNAL EXCITATION . . . . . . . . . . . . . 131
3.3 STABILITY OF MOTION - FREE VIBRATIONS . . . . . . . . . . 132
3.4 PROBLEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134

4 ANALYSIS OF ROTOR WITH ONE CONCENTRATED MASS 152

4.1 PHYSICAL MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
4.2 MATHEMATICAL MODEL . . . . . . . . . . . . . . . . . . . . . . . 153
4.3 ANALYSIS OF MOTION IN TERMS OF THE STATIONARY CO-
ORDINATES SYSTEM . . . . . . . . . . . . . . . . . . . . . . . . . 156
4.3.1 Free vibrations (general solution of the homogeneous
equation) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
4.3.2 Forced vibrations (particular solution of the non-homogeneous
equation) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
4.4 ANALYSIS OF MOTION IN TERMS OF THE ROTATING COOR-
DINATES SYSTEM . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
4.4.1 Free vibrations . . . . . . . . . . . . . . . . . . . . . . . . . 159
4.4.2 Forced vibrations . . . . . . . . . . . . . . . . . . . . . . . . 159
4.5 INFLUENCE OF DAMPING . . . . . . . . . . . . . . . . . . . . . . 160
4.5.1 Analysis of influence of the external damping . . . . . . 162
4.5.2 Analysis of the internal damping . . . . . . . . . . . . . . 165
4.5.3 Influence of the external and internal damping . . . . . 170
4.6 INFLUENCE OF AXIAL FORCES . . . . . . . . . . . . . . . . . . . 173
4.6.1 Description of physical model . . . . . . . . . . . . . . . . 173
4.6.2 Development of mathematical model and its analysis . 174

5 ANALYSIS OF MOTION OF ROTOR WITH UNSYMMETRICAL

CROSS-SECTION 177
5.1 MODELLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
5.1.1 Description of the physical model . . . . . . . . . . . . . 177
5.1.2 Development of the mathematical model . . . . . . . . . 178
5.2 ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
5.2.1 Analysis of the free vibrations - stability analysis . . . 180
CONTENTS 5

5.2.2 Analysis of the forced vibrations . . . . . . . . . . . . . . 185

6 ANALYSIS OF MOTION OF ROTOR WITH ONE DISK 191

6.1 MODELLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
6.2 ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
6.2.1 Free vibrations . . . . . . . . . . . . . . . . . . . . . . . . . 193
6.2.2 Influence of residual unbalance . . . . . . . . . . . . . . . 195
6.2.3 Influence of the ratio Iz /I on the natural frequencies
and the critical speeds . . . . . . . . . . . . . . . . . . . . 196

7 ANALYSIS OF MOTION OF ROTOR WITH ONE UNSYMMET-

RICAL DISK 202
7.1 MODELLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
7.1.1 Description of the physical model . . . . . . . . . . . . . 202
7.1.2 Development of the mathematical model . . . . . . . . . 203
7.2 ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
7.2.1 Free vibrations- Stability of motion . . . . . . . . . . . . 205
7.2.2 Particular solution - Equilibrium position . . . . . . . . 209
7.2.3 Forced vibration caused by the residual imbalance . . . 210

8 ANALYSIS OF INFLUENCE OF THE OIL BEARINGS ON MO-

TION OF ROTORS 213
8.1 MODELLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
8.1.1 Physical model . . . . . . . . . . . . . . . . . . . . . . . . . 213
8.1.2 Mathematical model . . . . . . . . . . . . . . . . . . . . . 214
8.2 ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
8.2.1 Description of the numerical simulation . . . . . . . . . 214
8.2.2 Results of the numerical simulation . . . . . . . . . . . . 215

9 ANALYSIS OF THE INFLUENCE OF THE ROLLING ELEMENTS

BEARINGS ON A ROTOR VIBRATIONS 224
9.1 CASE OF THE VERTICAL SHAFT . . . . . . . . . . . . . . . . . . 224
9.1.1 Physical model . . . . . . . . . . . . . . . . . . . . . . . . . 224
9.1.2 Mathematical model and its analysis . . . . . . . . . . . 224
9.2 CASE OF THE HORIZONTAL SHAFT . . . . . . . . . . . . . . . . 228
9.2.1 Physical model . . . . . . . . . . . . . . . . . . . . . . . . . 228
9.2.2 Mathematical model and its analysis . . . . . . . . . . . 229

10 ANALYSIS OF THE INFLUENCE OF THE SUPPORTING STRUC-

TURES ON A ROTOR VIBRATIONS. 234
10.1 PROPERTIES OF SUPPORTING STRUCTURES . . . . . . . . . . 235
10.1.1 Isotropic structures . . . . . . . . . . . . . . . . . . . . . . 235
10.1.2 Anisotropic structures . . . . . . . . . . . . . . . . . . . . 236
10.2 INFLUENCE OF THE SUPPORTING STRUCTURES ON A RO-
TOR WITH CONCENTRATED MASS . . . . . . . . . . . . . . . . 237
10.2.1 Influence of the isotropic structure. . . . . . . . . . . . . 238
CONTENTS 6

10.2.2 Influence of the anisotropic structure. . . . . . . . . . . 242

10.3 INFLUENCE OF THE SUPPORTING STRUCTURES ON A RO-
TOR WITH DISTRIBUTED MASS . . . . . . . . . . . . . . . . . . 248
10.4 INFLUENCE OF THE SUPPORTING STRUCTURES ON A RO-
TOR WITH CONCENTRATED DISK . . . . . . . . . . . . . . . . . 252
10.5 Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
10.6 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
10.6.1 Influence of rigid support . . . . . . . . . . . . . . . . . . . . . 253
10.6.2 Influence of isotropic support . . . . . . . . . . . . . . . . . . 254
10.6.3 Influence of anisotropic support . . . . . . . . . . . . . . . . . 255
10.6.4 Influence of mass of the supporting structure . . . . . . . . . . 260

11 ANALYSIS OF THE INFLUENCE OF THE SQUEEZE FILM DAMPER

ON A ROTOR VIBRATIONS 262
11.1 MODELLING AND ANALYSIS OF A RIGIDLY SUPPORTED RO-
TOR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
11.2 MODELLING AND ANALYSIS OF THE ROTOR WITH SQUEEZE
FILM DAMPER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266

12 OPTIMIZATION OF MOTION OF ROTOR SUPPORTED UPON

MAGNETIC BEARINGS 270
12.1 MODELLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
12.2 ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274

13 ANALYSIS OF THE INFLUENCE OF THE NON-LINEAR PROP-

ERTIES OF ELASTIC SUPPORTS ON ROTOR VIBRATIONS. 277
13.1 MODELLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
13.1.1 Physical model . . . . . . . . . . . . . . . . . . . . . . . . . 277
13.1.2 Mathematical model . . . . . . . . . . . . . . . . . . . . . 279
13.2 ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
13.2.1 Approximate analytical analysis of the forced vibrations 279
13.2.2 Numerical simulation . . . . . . . . . . . . . . . . . . . . . 284

14 ’CHATTER’ IN ROTATING MACHINERY 289

14.1 MODELLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
14.1.1 Physical model . . . . . . . . . . . . . . . . . . . . . . . . . 290
14.1.2 Mathematical model . . . . . . . . . . . . . . . . . . . . . 290
14.1.3 Quantitative analysis of stability of the system equilib-
rium position . . . . . . . . . . . . . . . . . . . . . . . . . . 292
14.1.4 Qualitative analysis of the system motion . . . . . . . . 294

15 ANALYSIS OF INFLUENCE OF THE ANISOTROPY OF SUP-

PORTS ON MOTION OF ROTOR WITH UNSYMMETRICAL
CROSS-SECTION 300
15.1 MODELLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
15.1.1 Physical model . . . . . . . . . . . . . . . . . . . . . . . . . 300
CONTENTS 7

15.1.2 Mathematical model . . . . . . . . . . . . . . . . . . . . . 302

15.1.3 Analysis of the free vibrations - stability of the equi-
librium position . . . . . . . . . . . . . . . . . . . . . . . . 303
15.1.4 Analysis of the forced vibrations . . . . . . . . . . . . . 315

III EXPERIMENTAL INVESTIGATIONS 325

16 INFLUENCE OF THE GYROSCOPIC EFFECT ON A ROTOR
VIBRATIONS 326
16.1 DESCRIPTION OF THE LABORATORY INSTALLATION . . . . . 326
16.2 MODELLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
16.2.1 Physical model . . . . . . . . . . . . . . . . . . . . . . . . . 327
16.2.2 Mathematical model . . . . . . . . . . . . . . . . . . . . . 327
16.3 ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
16.3.1 Natural vibrations . . . . . . . . . . . . . . . . . . . . . . . 330
16.3.2 Response to residual unbalance . . . . . . . . . . . . . . . 331
16.4 EXPERIMENTAL INVESTIGATION. . . . . . . . . . . . . . . . . . 332
16.4.1 Identification of parameters of the laboratory installation332
16.4.2 Measurement of the natural frequencies of the rotor . 334
16.4.3 Measurement of amplitude of the forced vibration . . . 335
16.4.4 Verification of the mathematical model . . . . . . . . . . 335

17 MODELLING OF ROTOR SYSTEMS 336

17.1 DESCRIPTION OF THE LABORATORY INSTALLATION . . . . . 336
17.2 DEVELOPMENT OF THE MATHEMATICAL MODELS OF SUB-
SYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
17.2.1 Mathematical model of the free-free shaft . . . . . . . . 338
17.2.2 Mathematical models of the rigid elements . . . . . . . 339
17.2.3 Mathematical models of the elastic elements . . . . . . 340
17.2.4 Mathematical model of foundation . . . . . . . . . . . . . 341
17.3 COMPOSITION OF THE SYSTEM MATHEMATICAL MODEL . . 341
17.4 ANALYSIS OF THE MATHEMATICAL MODEL. . . . . . . . . . . 341
17.5 EXPERIMENTAL INVESTIGATION . . . . . . . . . . . . . . . . . 343
17.5.1 Measurement of the angular speed of the rotor . . . . . 343
17.5.2 Measurement of the natural frequencies . . . . . . . . . 343
17.5.3 Measurement of amplitude of the forced vibration . . . 344
17.6 VERIFICATION OF THE DEVELOPED MATHEMATICAL MODEL344

18 BALANCING OF ROTOR 345

18.1 DESCRIPTION OF THE LABORATORY INSTALLATION . . . . . 345
18.2 THEORETICAL BACKGROUND . . . . . . . . . . . . . . . . . . . 345
18.3 EXPERIMENTAL INVESTIGATION. . . . . . . . . . . . . . . . . . 351
18.3.1 Description of the balancing procedure. . . . . . . . . . . . . . 352
CONTENTS 8

IV PROBLEMS 353
19 PREVIOUS YEARS EXAMINATION QUESTIONS 354
19.1 SAMPLES OF QUESTIONS WITH SOLUTIONS . . . . . . . . . . . 354
19.2 SAMPLES OF QUESTIONS WITHOUT SOLUTION . . . . . . . . 370
CONTENTS 9

INTRODUCTION

Ωn Ωm

Figure 1

The mechanical system (Fig 1) that contains rotating elements is usually re-
ferred to as rotor system. Kinetic energy of these rotating elements forms an internal
source of energy. This internal source of energy can be considered infinite since re-
gardless of the amount of energy taken out from this source, due the ever existing
connection of the rotor system with the driving system, the kinetic energy accumu-
lated in the rotating elements is unchanged. It means that, in certain circumstances,
unlimited amount of energy can be transferred into vibration of the rotor - system.
These vibrations disturb the technological processes the machine is design for, results
in the shorter live-time and very often lead to its destruction. These vibrations can
be developed even if there is no external forces acting on the rotor - system. There-
fore prediction and attenuation of vibrations of the rotor - system are very important
from engineering point of view on bath: the design stage and during maintenance.
Prediction of the dynamic behavior of the rotor - system requires analysis of the
mathematical model that quantitatively reflects its dynamic properties.
This lectures give theoretical bases for modelling and analysis of vibrations
of rotating machinery. The described procedures allow to develop the mathematical
models for the rotor sub-systems (rigid elements, shafts, bearings, foundations (see
Fig. 2).
CONTENTS 10

Rigid elements
Ωn Ωm Ωn Ωm

Shafts
Ωn

Ωm

Bearings

Foundations

Figure 2

Since after linearization these mathematical models are of the same form,
process of composition of the final mathematical model can be standardized to facil-
itate the necessary programming. At each step of composition of the mathematical
model a reduction of its size can be obtained by means of the condensation technique.
Environment, the system operates in, determines boundary conditions for the rotor
system. The described methods of analysis of the final mathematical model yield
answer to the basic engineering problems such as system equilibrium position and its
stability, response to the environmental excitation, response to the residual unbalance
and critical speeds.
The following chapter give theoretical basis for creation of the mathematical
model of the rotor system.
 Part I
MODELLING

11
 Chapter 1
MATHEMATICAL MODELS OF SUB-SYSTEMS

To obtain quantitatively accurate mathematical model of a rotor system, one have

to take into consideration influence of the attached to the shaft elements, mass and
stiffness distribution of shafts, dynamic properties of foundations, bearings and, in
many cases, environment the rotor works in. These parts of the rotor system are
called sub-systems and development of their mathematical models is given in this
paragraph.

1.1 MATHEMATICAL MODELS OF RIGID ELEMENTS

Elements that, by assumption, can not be deflected are called rigid elements.
1.1.1 Equations of motion of symmetric rigid bodies
Physical model

O,C G Ω z

y
x

Figure 1

Consider a rigid body shown in Fig. 1 that is to be attached to a shaft at the point C.
Let us introduce the stationary system of coordinates xyz through the point C. Its
axis y is vertically down and axis z is along axis of symmetry of the body. Centre of
gravity of the body G is located at this axis of symmetry. The body rotates about its
axis of symmetry with a constant velocity Ω and can perform small general motion
with respect to the stationary system of coordinates. The instantaneous position with
respect to the stationary system of coordinates is shown in Fig. 2
MATHEMATICAL MODELS OF RIGID ELEMENTS 13

r r
G
y zr zp
G
C a _ Ω
x z F M zr
mg zr x

F xp
yr
φ Fxr
_
_ xr
y M yr a M xr

yr yp

Figure 2

Its small linear displacement can be determined by the position vector r and its
small angular displacement is denoted by vector φ. Let us assume that these two vec-
tors are determined by their components along the stationary system of coordinates
as follows

r = ix + jy + kz (1.1)

φ = iφx + jφy + kφz (1.2)

These two vectors uniquely define instantaneous position of the rotating system of
coordinates xr yr zr . Principal axes of the body considered, through centre of gravity
G, are parallel to the rotating system of coordinates and the principal moments of
inertia of the body about these axes are Ix = Iy = I and Iz respectively. The body
rotates with the constant angular speed Ω about axis zr .

Matrix of direction cosines

To derive the matrix of direction cosines between the rotating system of coordinates
xr yr zr and the stationary system of coordinates xyz, let us rotate subsequently the
rotating system of coordinates with respect to the stationary one by small angles
φx , φy and φz .
MATHEMATICAL MODELS OF RIGID ELEMENTS 14

z1 yr
y1 y2
x x1 z z1 z2 z r y2
φy
z2
φx φz

y y1 x2 x1 x2 xr
b) c)
a)

Figure 3

Rotation of the system xr yr zr about axis x by the small angle φx results in

the new position of the rotating frame x1 y1 z1 (see Fig. 3a). Matrix of the direction
cosines between those two systems of coordinates is as following:
⎡ ⎤ ⎡ ⎤⎡ ⎤
x 1 0 0 x1
⎣ y ⎦ = ⎣ 0 1 −φx ⎦ ⎣ y1 ⎦ (1.3)
z 0 φx 1 z1
Rotation of the system xr yr zr about axis y1 by the small angle φy results in the
new position of the rotating frame shown in Fig. 3b. The corresponding matrix of
direction cosines has the following form
⎡ ⎤ ⎡ ⎤⎡ ⎤
x1 1 0 φy x2
⎣ y1 ⎦ = ⎣ 0 1 0 ⎦ ⎣ y2 ⎦ (1.4)
z1 −φy 0 1 z2

Rotation of the system of coordinates xr yr zr about axis z2 by the small angle φz results
in the final position of the rotating frame shown in Fig. 3c. The corresponding matrix
of direction cosines has the following form:
⎡ ⎤ ⎡ ⎤⎡ ⎤
x2 1 −φz 0 xr
⎣ y2 ⎦ = ⎣ φz 1 0 ⎦ ⎣ yr ⎦ (1.5)
z2 0 0 1 zr
Hence, the matrix of direction cosines between the stationary and the rotating system
of coordinates, with accuracy to linear terms with respect to the components φx , φy
and φz of the small vector φ, can be adopted in the following form:
⎡ ⎤ ⎡ ⎤⎡ ⎤
x 1 −φz φy xr
⎣ y ⎦ = ⎣ φz 1 −φx ⎦ ⎣ yr ⎦ (1.6)
z −φy φx 1 zr
Since the angles φx , φy and φz are small by assumption, during multiplication of the
above matrices, the magnitudes of order greater then one have been omitted.
MATHEMATICAL MODELS OF RIGID ELEMENTS 15

Angular and linear velocities

The angular speed of the rotating system of coordinates is equal to the angular speed
of the principal axes and according to Eq. 1.2 is
ωr = ωp = φ̇ = iφ̇x + jφ̇y + kφ̇z (1.7)
According to Eq. 1.6, components of the angular speed of the principal axes along
system of coordinates xp yp zp are
ωr = ωp = ip φ̇x + jp φ̇y + kp φ̇z = φ̇ (1.8)
The linear velocity of the centre of gravity G as the first time derivative of its position
vector
rG = r + a = ix + jy + kz + kp a = ip x + jp y + kp (z + a) (1.9)
is
vG = ṙG = ip ẋ + jp ẏ + kp ż + ωp × rG = ip (ẋ + aφ̇y ) + jp (ẏ − aφ̇x ) + kp ż (1.10)

Equations of motion
Motion of the body is governed by Euler’s modified equations
m(v̇Gxp + vGzp ωpyp − vGyp ωpzp ) = Fxr + mgφz
m(v̇Gyp + vGxp ω pzp − vGzp ω pxp ) = Fyr + mg
m(v̇Gzp + vGyp ωpxp − vGxp ω pyp ) = Fzr − mgφx

I ω̇ pxp + (Iz − I)ωpyp ωpzp + Iz Ωω pyp = M̄xr + Fyr a

I ω̇pyp − (Iz − I)ωpxp ωpzp − Iz Ωω pxp = M̄yr − Fxr a
Iz (ω̇ pzp + Ω̇) = M̄zr (1.11)
Components of the interaction forces F and M̄ along the stationary system of coordi-
nates, according to Eq. 1.6, are equal to their components along the rotating system
of coordinates with accuracy to small magnitudes of order one.
Fxr = Fx , Fyr = Fy , Fzr = Fz , M̄xr = M̄x , M̄yr = M̄y , M̄zr = M̄z (1.12)
Introducing Eqs. 1.8, 1.10 and 1.12 into equations 1.11, with accuracy to small
magnitudes of order one, one can obtain the equations of motion in the following
form

mẍ + maφ̈y = Fx + mgφz

mÿ − maφ̈x = Fy + mg
mz̈ = Fz − mgφx

.
I φ̈x + Iz Ωφy = M̄x + Fy a
.
I φ̈y − Iz Ωφx = M̄y − Fx a
Iz φ̈z = M̄z (1.13)
MATHEMATICAL MODELS OF RIGID ELEMENTS 16

The sixth equation of 1.13 represents torsional vibrations

The graphical interpretation of the angular displacements and moments is
given in Fig. 4.

_
My My Mx
φx
O _ z O z
ϕy Mx ϕx φy
φy φx
_ _
My Mx
y x

Figure 4

In Fig. 4 new notations for the angular displacements ϕ and moments M,

which are commonly used mechanics of solids, are shown also. There exists the
following relationship between the angular displacements φ and ϕ as well as moments
M̄ and M.

φx = −ϕy , φy = ϕx , φz = ϕz , M̄x = −My , M̄y = Mx , M̄z = Mz (1.14)

Introduction of Eq. 1.14 into Eq. 1.13 yields equations of motion in the form

mẍ + maϕ̈x = Fx + mgϕz

mÿ + maϕ̈y = Fy + mg
mz̈ = Fz + mgϕy

.
I ϕ̈x + Iz Ωϕy = Mx − Fx a
.
I ϕ̈y − Iz Ωϕx = My − Fy a
Iz ϕ̈z = Mz (1.15)

If one assume that shaft, the body is attached to, can not perform axial vibrations
(z = 0) and torsional vibration (ϕz = 0), the mathematical model for the body can
be simplified to the following form

mẍ + cẋ = R + FG (1.16)

MATHEMATICAL MODELS OF RIGID ELEMENTS 17

where:
⎡ ⎤
m 0 ma 0
⎢ 0 m 0 ma ⎥
m = ⎢
⎣ ma 0 I + ma2
⎥
⎦
0
0 ma 0 I + ma2
⎡ ⎤
0 0 0 0
⎢ 0 0 0 0 ⎥
c = ⎢
⎣ 0 0
⎥
0 Iz Ω ⎦
0 0 −Iz Ω 0
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x Fx 0
⎢ y ⎥ ⎢ Fy ⎥ ⎢ mg ⎥
x = ⎢ ⎥ ⎢ ⎥
⎣ ϕx ⎦ , R = ⎣ Mx ⎦ , FG =⎢
⎣ 0 ⎦
⎥ (1.17)
ϕy My mga

As one can see from the third of equation 1.15, this assumption results in existence
of axial force
Fz = −mgϕy (1.18)
1.1.2 Equations of motion of non-symmetric bodies
Physical model
In the considerations carried out in the previous paragraph, it was assumed that axis
of rotation of the rigid element coincides with its axis of symmetry ( Ix = Iy = I ).
Now, let us assume that the principal moment of inertia Ix and Iy are not equal to
each other but its axis of rotation coincides the principal axis of inertia zp through
the centre of gravity G associated with the moment of inertia Iz (Fig 5).
MATHEMATICAL MODELS OF RIGID ELEMENTS 18

_
Mz
O Fz

y rG zr zp
r G
x a Ω
z C mg x

Fx _
φ Mx Ωt
xr
y a xp

yr
Fy Ωt
yp
_
My

Figure 5

In Fig. 5, xyz is the absolute system of coordinates. With respect to the

absolute system of coordinates xyz, the rigid element performs small motion deter-
mined by the small position vector r and the small angular displacement φ. These
two vectors defines position of the rotating system of coordinates xr yr zr . The body
rotates about axis zr with the constant velocity Ω. Therefore its principal axis zp
coincides axis principal zr and the principal axis xp yp are displaced by the angular
displacement Ωt with respect to xr yr
Let us assume that the vectors r and φ are given by their components along
the rotating system of coordinates xyz.

r = ix + jy + kz (1.19)

φ = iφx + jφy + kφz (1.20)

Matrices of direction cosines

Matrix of direction cosines between system of coordinates xyz and xr yr zr was derived
previously to be ⎡ ⎤ ⎡ ⎤⎡ ⎤
x 1 −φz φy xr
⎣ y ⎦ = ⎣ φz 1 −φx ⎦ ⎣ yr ⎦ (1.21)
z −φy φx 1 zr
MATHEMATICAL MODELS OF RIGID ELEMENTS 19

Matrix of direction cosines between the rotating system of coordinates xr yr zr and

principal axes xp yp zp is as follows
⎡ ⎤ ⎡ ⎤⎡ ⎤
xr cos Ω t − sin Ω t 0 xp
⎣ yr ⎦ = ⎣ sin Ω tz cos Ω t 0 ⎦ ⎣ yp ⎦ (1.22)
zr 0 0 1 zp

Hence , matrix of direction cosines between the absolute system of coordinates xyz
and the principal axis xp yp zp is
⎡ ⎤ ⎡ ⎤⎡ ⎤⎡ ⎤
x 1 −φz φy cos Ωt − sin Ωt 0 xp
⎣ y ⎦ = ⎣ φz 1 −φx ⎦ ⎣ sin Ωt cos Ωt 0 ⎦ ⎣ yp ⎦
z −φy φx 1 0 0 1 zp
⎡ ⎤⎡ ⎤
cos Ωt − φz sin Ωt − sin Ωt − φz cos Ωt φy xp
= ⎣ + sin Ωt + φz cos Ω −φz sin Ωt + cos Ωt −φx ⎦ ⎣ yp ⎦
−φy cos Ωt + φx sin Ωt φy sin Ωt + φx cos Ωt 1 zp
⎡ ⎤
xp
= [Cp→a ] yp ⎦
⎣ (1.23)
zp

Angular and linear velocities

The angular speed of the rotating system of coordinates according to Eq. 1.20 is

ωr = φ̇ = iφ̇x + jφ̇y + kφ̇z (1.24)

Its components along the principal axes xp yp zp are

⎡ ⎤ ⎡ ⎤ ⎡ ⎤
ωrpx φ̇x φ̇x cos Ωt + φ̇y sin Ωt
⎣ ω rpy ⎦ = [Cp→a ]T ⎣ φ̇y ⎦ = ⎣ −φ̇x sin Ωt + φ̇y cos Ωt ⎦ (1.25)
ωrpz φ̇z φ̇z

Hence, the angular speed of the principal axes is

ωp = ωr + kp Ω
(1.26)
= ip (φ̇x cos Ωt + φ̇y sin Ωt) + jp (−φ̇x sin Ωt + φ̇y cos Ωt) + kp (φ̇z + Ω)

In a similar manner one can develop expression for components of the position vector
r along the principal axes xp yp zp .
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
xp x x cos Ωt + y sin Ωt
⎣ yp ⎦ = [Cp→a ]T ⎣ y ⎦ = ⎣ −x sin Ωt + y cos Ωt ⎦ (1.27)
zp z z

Hence, the position vector of the centre of gravity G is

rG = r + kp a
= ip (x cos Ωt + y sin Ωt) + jp (−x sin Ωt + y cos Ωt) + kp (z + a) (1.28)
MATHEMATICAL MODELS OF RIGID ELEMENTS 20

Its first time derivative yields the absolute velocity of the centre of gravity G.

vG = ṙG = r0G + ω p × rG
= ip (ẋ cos Ωt + ẏ sin Ωt + a(−φ̇x sin Ωt + φ̇y cos Ωt))
(1.29)
+jp (−ẋ sin Ωt + ẏ cos Ωt − a(φ̇x cos Ωt + φ̇y sin Ωt))
+kp (ż)

Equations of motion
Motion of the body is governed by Euler’s Equations of Motion that have the following
form

m(v̇Gxp + vGzp ω pyp − vGyp ω pzp ) = Fxp + Gxp

m(v̇Gyp + vGxp ωpzp − vGzp ω pxp ) = Fyp + Gyp
m(v̇Gzp + vGyp ω pxp − vGxp ωpyp ) = Fzp + Gzp

Ix ω̇ pxp + (Iz − Iy )ω pyp ω pzp = M̄xp + Fyp a

Iy ω̇ pyp + (Ix − Iz )ω pxp ω pzp = M̄yp − Fxp a
Iz ω̇pzp + (Iy − Ix )ωpxp ω pyp = M̄zp (1.30)

In the above equations Gxp , Gyp , Gzp stand for components of the gravity force

G = jmg (1.31)
that can be obtained with help of the matrix of direction cosines Eq. 1.23.

⎡ ⎤ ⎡ ⎤ ⎡ ⎤
Gxp 0 φz mg cos Ωt + mg sin Ωt
⎣ Gyp ⎦ = [Cp→a ]T ⎣ mg ⎦ = ⎣ −φz mg sin Ωt + mg cos Ωt ⎦ (1.32)
Gzp 0 −mgφx

Similarly, the components of interaction forces along the principal axes

Fxp , Fyp , Fzp , M̄xp , M̄yp , M̄zp

can be expressed in terms of their components along the absolute system of coordi-
nates xyz.
⎡ ⎤
Fxp
⎣ Fyp ⎦
Fzp
⎡ ⎤⎡ ⎤
cos Ωt − φz sin Ωt + sin Ωt + φz cos Ω −φy cos Ωt + φx sin Ωt Fx
= ⎣ − sin Ωt − φz cos Ωt −φz sin Ωt + cos Ωt φy sin Ωt + φx cos Ωt ⎦ ⎣ Fy ⎦
φy −φx 1 Fz
⎡ ⎤
Fx cos Ωt + Fy sin Ωt
= ⎣ −Fx sin Ωt + Fy cos Ωt ⎦ (1.33)
Fz
MATHEMATICAL MODELS OF RIGID ELEMENTS 21
⎡ ⎤
M̄xp
⎣ M̄yp ⎦
M̄zp
⎡ ⎤⎡ ⎤
cos Ωt − φz sin Ωt + sin Ωt + φz cos Ω −φy cos Ωt + φx sin Ωt M̄x
= ⎣ − sin Ωt − φz cos Ωt −φz sin Ωt + cos Ωt φy sin Ωt + φx cos Ωt ⎦ ⎣ M̄y ⎦
φy −φx 1 M̄z
⎡ ⎤
M̄x cos Ωt + M̄y sin Ωt
= ⎣ −M̄x sin Ωt + M̄y cos Ωt ⎦ (1.34)
M̄z

Introduction of Eqs. 1.26, 1.29, 1.32, 1.33 and 1.34 into 1.30 yields the following
equations.

m(+ cos Ωtẍ + sin Ωtÿ − a sin Ωtφ̈x + a cos Ωtφ̈y )

= +Fx cos Ωt + Fy sin Ωt + φz mg cos Ωt + mg sin Ωt

m(− sin Ωtẍ + cos Ωtÿ − a cos Ωtφ̈x − a sin Ωtφ̈y )

= −Fx sin Ωt + Fy cos Ωt − φz mg sin Ωt + mg cos Ωt

mz̈ = Fz − mgφx

Ix (−Ω sin Ωt φ̇x + cos Ωtφ̈x + Ω cos Ωt φ̇y + sin Ωtφ̈y )

+(Iz − Iy )(−Ω sin Ωt φ̇x + Ω cos Ωt φ̇y )
= +M̄x cos Ωt + M̄y sin Ωt + (−Fx sin Ωt + Fy cos Ωt)a

Iy (−Ω cos Ωt φ̇x − sin Ωtφ̈x − Ω sin Ωt φ̇y + cos Ωtφ̈y )

+(Ix − Iz )(+Ω cos Ωt φ̇x + Ω sin Ωt φ̇y )
= −M̄x sin Ωt + M̄y cos Ωt − (−Fx cos Ωt + Fy sin Ωt)a

Iz φ̈z = M̄z (1.35)

These equations can be sinplify to the form 1.36.

mẍ + maφ̈y = Fx + mgφz

mÿ − maφ̈x = Fy + mg
mz̈ = Fz − mgφx

(Io − ∆I cos 2Ωt)φ̈y + ∆I sin 2Ωtφ̈x + (−Iz Ω + 2∆IΩ cos 2Ωt)φ̇x

+2∆IΩ sin 2Ωtφ̇y
= M̄y − Fx a
MATHEMATICAL MODELS OF RIGID ELEMENTS 22

(Io + ∆I cos 2Ωt)φ̈x + ∆I sin 2Ωtφ̈y + (+Iz Ω + 2∆IΩ cos 2Ωt)φ̇y

−2∆IΩ sin 2Ωtφ̇x
= M̄x + Fy a

Iz φ̈z = M̄z (1.36)

where:
Ix + Iy Ix − Iy
Io = ∆I = (1.37)
2 2
The first equation of 1.36 was obtained by multiplication of the first equation 1.35
by cos Ωt, multiplication of the second equation of 1.35 by -sin Ωt and then addition
of these two equations together. The second equation of 1.36 was obtained by multi-
plication of the first equation 1.35 by sin Ωt, multiplication of the second equation of
1.35 by cos Ωt and then addition of these two equations together. In similar manner
the fourth and fifth equation were obtained.
Elimination of Fx and Fy from the fourth and fifth equation of 1.36 with help
of the first and second equation 1.36 and introduction of notations 1.14 yields the
following equations of motion.

mẍ + maϕ̈x = Fx + mgϕz

mÿ − maϕ̈y = Fy + mg
mz̈ = Fz + mgϕy

maẍ + (Io + ma2 − ∆I cos 2Ωt)ϕ̈x − ∆I sin 2Ωtϕ̈y + (+Iz Ω − 2∆IΩ cos 2Ωt)ϕ̇y
+2∆IΩ sin 2Ωtϕ̇x
= Mx − mgaϕz

maÿ + (Io + ma2 + ∆I cos 2Ωt)ϕ̈y − ∆I sin 2Ωtϕ̈x + (−Iz Ω − 2∆IΩ cos 2Ωt)ϕ̇x
−2∆IΩ sin 2Ωtϕ̇y
= My + mga

Iz ϕ̈z = Mz (1.38)
If one assume that the rotor the body is attached to, can not perform axial
vibrations (z = 0) and torsional vibration (ϕz =0), the mathematical model for the
body can be simplified to the following form

mẍ + cẋ = R + FG (1.39)

MATHEMATICAL MODELS OF RIGID ELEMENTS 23

where
⎡ ⎤
m 0 ma 0
⎢ 0 m 0 ma ⎥
m = ⎢
⎣ ma 0 Io + ma − ∆I cos 2Ωt
2
⎥
⎦
−∆I sin 2Ωt
2
0 ma −∆I sin 2Ωt Io + ma + ∆I cos 2Ωt
⎡ ⎤
0 0 0 0
⎢ 0 0 0 0 ⎥
c = ⎢
⎣ 0 0
⎥
2∆IΩ sin 2Ωt Iz Ω − 2∆IΩ cos 2Ωt ⎦
0 0 −Iz Ω − 2∆IΩ cos 2Ωt −2∆IΩ sin 2Ωt
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x Fx 0
⎢ y ⎥ ⎢ Fy ⎥ ⎢ mg ⎥
x = ⎢ ⎥ ⎢ ⎥ ⎢
⎣ ϕx ⎦ , R = ⎣ Mx ⎦ , FG = ⎣ 0 ⎦
⎥ (1.40)
ϕy My mga

1.1.3 Modelling of residual unbalance

In modelling, presented in the previous section, it was assumed that the centre of
gravity of the body considered is at axes of rotation zr of the rotating system of
coordinates xr yr zr . Position of this system of coordinates is determined by the linear
displacement vector r and the angular displacement φ. Moreover, it was assumed
that one of its principal axis through this centre coincides the axis of rotation. Due
to the ever existing residual unbalance, the above assumptions are not fulfilled. To
derive the inertia forces that are caused by the residual unbalance, let us introduce
the rotating system of coordinates xR yR zR that rotates with respect to the rotating
system of coordinates xr yr zr with the angular velocity Ω equal to the angular velocity
of the shaft (Fig. 6).
MATHEMATICAL MODELS OF RIGID ELEMENTS 24

O x

y zr z R
r Ω
Or
a
C
µ Ωt xr
µy
φ R zp
µx xR
R
G
∆
xp

yR yr

yp

Figure 6

Due to the residual unbalance, the centre of gravity G as well as the principal
axes xp yp zp do not coincide the xR yR zR frame. The centre of gravity G is displaced
from the origin Or by a small distance µ and the principal axes of the body xp yp zp
are turned by a small angular displacement ∆ with respect to the rotating system of
coordinates xR yR zR . Hence
µ = iR µxR + jR µyR
∆ = iR ∆xR + jR ∆yR + kR ∆zR (1.41)
The absolute angular velocity of the rotating system of coordinates xR yR zR is
ω R = φ̇ + kR Ω (1.42)
The absolute angular velocity of the principal axes is
˙ = φ̇ + kR Ω + ∆
ωp = ωR + ∆ ˙ (1.43)
Since ∆ is a constant angle (∆0 = 0) , the absolute velocity of the principal axes is
ωp = φ̇ + kR Ω + ∆0 + ωR × ∆ = φ̇ + kR Ω + (φ̇ + kR Ω) × ∆ (1.44)
With accuracy to the small magnitudes of order one (φ̇×∆ = 0), the above expression
takes form
¯ ¯
¯ iR j k ¯
¯ R R ¯
ω p = φ̇ + kR Ω + kR Ω × ∆ = φ̇ + kR Ω + ¯¯ 0 0 Ω ¯¯
¯ ∆x ∆y ∆z ¯
R R R

= φ̇ + iR (−∆yR Ω) + jR (∆xR Ω) + kR Ω (1.45)

MATHEMATICAL MODELS OF RIGID ELEMENTS 25

The absolute velocity vG of the centre of gravity G, as the first derivative of its
position vector (r + a + µ) with accuracy limited to the small magnitudes of order
one is as follows
d
vG = (r + a + µ) = ṙ + ȧ + µ̇ = ṙ + ȧ + µ0 + ωR × µ
dt
= ṙ + ȧ + (φ̇ + kR Ω) × µ ∼= ṙ + ȧ + kR Ω × µ
¯ ¯
¯ iR j k ¯
¯ R R ¯
= ṙ + ȧ+ ¯¯ 0 0 Ω ¯¯
¯ µx µy µz ¯
R R R

= ṙ + ȧ + iR (−ΩµyR ) + jR (ΩµxR ) (1.46)

The inertial terms (inertial forces) which are caused by the residual unbalance µ and
∆ can be obtained by introduction of the above expressions into Euler’s equations
of motion and selection of these terms which does not appear in equations 1.15. The
same result can be obtained by retaining in the above expressions only these terms
which correspond to the vector µ and ∆ and then introducing them into Euler’s
equations. Comparison of Eqs. 1.45 and 1.46 with equations 1.8 and 1.10 allows to
identify these terms to be:

ω = iR (−∆yR Ω) + jR (∆xR Ω) + kR Ω (1.47)

v = iR (−ΩµyR ) + jR (ΩµxR ) (1.48)

To make use from Euler’s equations, it is necessary to know expressions for compo-
nents of the above vectors along the body system of coordinates. Since the body
system of coordinates is turned with respect to the rotating system of coordinates
xR yR zR by the small angle ∆ , the matrix of direction cosines between those two
systems, according to Eq. 1.5 is
⎡ ⎤
1 ∆xR −∆yR
[CR→p ] = [Cp→R ]T = ⎣ −∆zR 1 ∆xR ⎦ (1.49)
∆yR −∆xR 1

Hence:
⎡ ⎤ ⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤
ωxp ω xR 1 ∆xR −∆yR −∆yR Ω −2∆yR Ω
⎣ ωyp ⎦ = [CR→p ] ⎣ ω yR ⎦ = ⎣ −∆zR 1 ∆xR ⎦ ⎣ ∆xR Ω ⎦ = ⎣ 2∆xR Ω ⎦
ωzp ω zR ∆yR −∆xR 1 Ω Ω
⎡ ⎤ ⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ (1.50)
⎤
vxp vxR 1 ∆xR −∆yR −µyR Ω −µyR Ω
⎣ vyp ⎦ = [CR→p ] ⎣ vyR ⎦ = ⎣ −∆zR 1 ∆xR ⎦ ⎣ µxR Ω ⎦ = ⎣ µxR Ω ⎦
vzp vzR ∆yR −∆xR 1 0 0
(1.51)
MATHEMATICAL MODELS OF RIGID ELEMENTS 26

Introduction of these above components into the left hand side of Euler’s equations
m(v̇xp + vzp ω yp − vyp ωzp ) = Fxp
m(v̇yp + vxp ωzp − vzp ωxp ) = Fyp
m(v̇zp + vyp ω xp − vxp ω yp ) = Fzp

Ix ω̇xp + (Iz − Iy )ω yp ω zp = M̄xp

Iy ω̇yp + (Ix − Iz )ωxp ω zp = M̄yp
Iz ω̇ zp + (Iy − Ix )ωxp ω yp = M̄zp (1.52)
yields components of the inertial moment M̄ and the inertial force F caused by the
residual unbalance µ and ∆.
Fxp = m(−µxR )Ω2
Fyp = m(−µyR )Ω2
Fzp = 0 (1.53)

M̄xp = (Iz − Iy )(2∆xR )Ω2

M̄yp = (Ix − Iz )(−2∆yR )Ω2
M̄zp = 0 (1.54)
These components are independent of the generalized coordinates x, y, φx and φy
therefore taken with sign ’-’ can be considered as the excitation forces caused by the
residual unbalance.
UFxp = mµxR Ω2
UFyp = mµyR Ω2 (1.55)

U M̄xp = 2(Iy − Iz )∆xR Ω2

U M̄yp = 2(Ix − Iz )∆yR Ω2 (1.56)
Multiplication of Eq. 1.55 and 1.56 by matrix of direction cosines [Cp→R ] (see Eq.
1.49) yields components along the rotating system of coordinates xR yR zR .
UFxR = mµxR Ω2
U FyR = mµyR Ω2 (1.57)

U M̄xR = 2(Iy − Iz )∆xR Ω2

U M̄yR = 2(Ix − Iz )∆yR Ω2 (1.58)
Multiplication of Eq. 1.57 and 1.58 by matrix of direction cosines between system of
coordinates xR yR zR and xr yr zr
∙ ¸
cos(Ω t) − sin(Ω t)
[CR→r ] = (1.59)
sin(Ω t) cos(Ω t)
MATHEMATICAL MODELS OF RIGID ELEMENTS 27

yields the components along the rotating system of coordinates xr yr zr

UFxr = SΩ2 cos(Ωt + ϕS )

UFyr = SΩ2 sin(Ωt + ϕS ) (1.60)

where:
q µy
S = m µ2x + µ2y ϕS = arc tan (1.61)
µx
and

U M̄xr = −DΩ2 sin(Ωt + ϕD )

U M̄yr = +DΩ2 cos(Ωt + ϕD ) (1.62)

where:

q
(Iy − Iz )∆xR
D = (2(Ix − Iz )∆yR )2 + (2(Iy − Iz )∆xR )2 ϕD = −arc tan (1.63)
(Ix − Iz )∆yR
The magnitudes S and D are called static and dynamic unbalance respectively.
Taking into account matrix of direction cosines between the rotating system
of coordinates xr yr zr and the stationary system of coordinates xyz (Eq. 1.49) one
can see that with accuracy to small magnitudes of order one

U Fx = UFxr UFy = UFyr U M̄x = U M̄xr U M̄y = U M̄yr . (1.64)

Upon introducing notations (see formula 1.14 and Fig. 4):
UMx = U M̄y - for unbalanced moment in plane xz
UMy = −U M̄x - for unbalanced moment in plane yz
and
δxR = ∆yR - for angle of unbalance in plane xz
δyR = −∆xR - for angle of unbalance in plane yz
the following vector of the external excitation can be produced.
⎡ ⎤ ⎡ ⎤
UFx SΩ2 cos(Ω t + ϕS )
⎢ U Fy ⎥ ⎢ SΩ2 sin(Ω t + ϕS ) ⎥
UF = ⎢ ⎥ ⎢
⎣ UMx ⎦ = ⎣ DΩ2 cos(Ω t + ϕD ) ⎦
⎥ (1.65)
UMy DΩ2 sin(Ω t + ϕD )
It can be now inserted into equations of motion of the rigid element 1.15

mẍ + maϕ̈x = Fx + UFx

mÿ + maϕ̈y = Fy + mg + U Fy (1.66)

.
I ϕ̈x + Iz Ωϕy = Mx − Fx a + UMx
.
I ϕ̈y − Iz Ωϕx = My − Fy a + U My (1.67)
MATHEMATICAL MODELS OF RIGID ELEMENTS 28

After elimination of Fx and Fy from equations 1.67 with help of equations 1.66, The
mathematical model of the unbalanced rigid body takes the following form.

mẍ + cẋ = R + FG +FU (1.68)

where:
⎡ ⎤
m 0 ma 0
⎢ 0 m 0 ma ⎥
m = ⎢
⎣ ma 0 I + ma2
⎥
⎦
0
0 ma 0 I + ma2
⎡ ⎤
0 0 0 0
⎢ 0 0 0 0 ⎥
c = ⎢
⎣ 0 0
⎥
0 Iz Ω ⎦
0 0 −Iz Ω 0
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x Fx 0
⎢ y ⎥ ⎢ Fy ⎥ ⎢ mg ⎥
x = ⎢ ⎥ ⎢ ⎥
⎣ ϕx ⎦ , R = ⎣ Mx ⎦ , FG =⎢
⎣ 0 ⎦,
⎥

ϕy My mga
⎡ ⎤
UFx
⎢ UFy ⎥
FU = ⎢
⎣ UMx + aU Fx ⎦
⎥ (1.69)
UMy + aU Fy

U Fx = SΩ2 cos(Ω t + ϕS )
U Fy = SΩ2 sin(Ω t + ϕS )
UMx = DΩ2 cos(Ω t + ϕD )
UMy = DΩ2 sin(Ω t + ϕD ) (1.70)
q µy
S = m µ2x + µ2y ϕS = arc tan
µx
q
(Iy − Iz )∆xR
D= (2(Ix − Iz )∆yR )2 + (2(Iy − Iz )∆xR )2 ϕD = −arc tan (1.71)
(Ix − Iz )∆yR
Figure 7 offers interpretation of symbols appearing in equations 1.69, 1.70 and 1.71.
In a similar manner it is possible to introduce the residual unbalance to the
mathematical model 1.39
MATHEMATICAL MODELS OF RIGID ELEMENTS 29

z
o
x OR
zR
C G µxR ϕx
a ∆ yR
zp

x xp xR

z
o
y zp
OR
µ yR ∆ xR zR
C G ϕy

yR yp
y

Figure 7
MATHEMATICAL MODELS OF RIGID ELEMENTS 30

1.1.4 Problems

Problem 1

2 1

G R
z

l L
a
y

Figure 8

The cylindrical rigid body 1 of radius R, length L and density d is attached

to the massless shaft 2 of radius r, length l and Young modulus E as shown in Fig.
8 The assembly rotates with the constant angular velocity Ω. Develop equations of
motion of this rotor.
MATHEMATICAL MODELS OF RIGID ELEMENTS 31

Solution.
According to Eq. 1.16, the mathematical model of the rigid body is:

mẍ + cẋ = R + FG (1.72)

where:
⎡ ⎤ ⎡ ⎤
m 0 ma 0 0 0 0 0
⎢ 0 m 0 ma ⎥ ⎢ 0 0 0 0 ⎥
m=⎢
⎣ ma 0 Ip + ma2
⎥
⎦ c=⎢
⎣ 0
⎥
0 0 0 Iz Ω ⎦
2
0 ma 0 Ip + ma 0 0 −Iz Ω 0
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x Fx 0
⎢ y ⎥ ⎢ ⎥ ⎢ mg ⎥
x=⎢ ⎥ , R = ⎢ Fy ⎥ , FG = ⎢ ⎥ (1.73)
⎣ ϕx ⎦ ⎣ Mx ⎦ ⎣ 0 ⎦
ϕy My mga
In the above mathematical model
1 1 1
a = L/2, m = πR2 Ldg, I = mR2 + mL2 , Iz = mR2 (1.74)
4 12 2
The interaction forces R can be expressed as function of the displacements x by
consideration of the shaft 2.

2
Fsy
Msy
z
y
ϕy
l

Figure 9

If the shaft is loaded with force Fsy and moment Msy (see Fig. 9), the corre-
sponding displacements y and ϕy are respectively

l3 l2 l2 l
y= Fsy + Ms , ϕy = Fsy + Ms (1.75)
3EJ 2EJ y 2EJ EJ y
or in matrix form ∙ ¸ ∙ ¸∙ ¸
l3 l2
y 3EJ 2EJ
Fsy
= l2 l (1.76)
ϕy 2EJ EJ
Msy
MATHEMATICAL MODELS OF RIGID ELEMENTS 32

where
π r4
J= (1.77)
4
The inverse transformation yields the wanted forces as function of the displacements
∙ ¸ ∙ l3 l2
¸−1 ∙ ¸ ∙ ¸∙ ¸
Fsy 3EJ 2EJ
y k11 k12 y
= l2 l = (1.78)
Msy 2EJ EJ
ϕy k21 k22 ϕy

Similarly
∙ ¸ ∙ l3 l2
¸−1 ∙ ¸ ∙ ¸∙ ¸
Fsx 3EJ 2EJ
x k11 k12 x
= l2 l = (1.79)
Msx 2EJ EJ
ϕx k21 k22 ϕx

Hence the vector of interaction forces that acts on the shaft is

⎡ ⎤ ⎡ ⎤⎡ ⎤
Fsx k11 0 k12 0 x
⎢ Fsy ⎥ ⎢ 0 k11 0 k12 ⎥⎢ y ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥ (1.80)
⎣ Msx ⎦ = ⎣ k21 0 k22 0 ⎦ ⎣ ϕx ⎦ = kx
Msy 0 k21 0 k22 ϕy

Since the vector of interaction forces that acts on the body

R = −Fs (1.81)

the final mathematical model of the system considered is

mẍ + cẋ + kx = FG (1.82)

MATHEMATICAL MODELS OF RIGID ELEMENTS 33

Problem 2

G z x k 2x k 2x

k 1y k 2y k 2y
l1 l2
y y

Figure 10

Rotor of mass m, the principal moments of inertia about axis x equal to I

and the principal moment of inertia about axis z equal to Iz is supported by massless
springs ky1 , kx1 , ky2 , kx2 as shown in Fig. 10. This rotor rotates with the angular
velocity Ω. Develop equations of motion of the rotor.
MATHEMATICAL MODELS OF RIGID ELEMENTS 34

Solution

My
G O z
y y2=y+ϕy l 2
y1 =y- ϕy l1
k 1y
l1 Fy l2 ϕy
k 2y
y

Figure 11

According to Eq. 1.16, the mathematical model of the rigid body is

mẍ + cẋ = R + FG (1.83)
where: ⎡ ⎤ ⎡ ⎤
m 0 0 0 0 0 0 0
⎢ 0 m 0 0 ⎥ ⎢ 0 0 0 0 ⎥
m=⎢
⎣ 0 0 I 0 ⎦,
⎥ c=⎢⎣ 0
⎥
0 0 Iz Ω ⎦
0 0 0 I 0 0 −Iz Ω 0
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x Fx 0
⎢ y ⎥ ⎢ Fy ⎥ ⎢ mg ⎥
x=⎢ ⎥ ⎢
⎣ ϕx ⎦ , R = ⎣
⎥, FG = ⎢ ⎥ (1.84)
Mx ⎦ ⎣ 0 ⎦
ϕy My 0
Components Fy and My of the interaction force R, according to the Fig. 11, are
Fy = −y1 ky1 − y2 ky2 = −(y − ϕy l1 )ky1 − (y + ϕy l2 )ky2
= −[(ky1 + ky2 )y + (ky2 l2 − ky1 l1 )ϕy ]
My = +y1 ky1 l1 − y2 ky2 l2 = +(y − ϕy l1 )ky1 l1 − (y + ϕy l2 )ky2 l2
= −[(ky2 l2 − yky1 l1 )y + (ky1 l12 + yky2 l22 )ϕy (1.85)
Similarly
Fx = −x1 2kx1 − x2 2kx2 = −(x − ϕx l1 )2kx1 − (x + ϕx l2 )2kx2
= −[(2kx1 + 2kx2 )x + (2kx2 l2 − 2kx1 l1 )ϕx ]
Mx = +x1 2kx1 l1 − x2 2kx2 l2 = +(x − ϕx l1 )2kx1 l1 − (x + ϕx l2 )2kx2 l2
= −[(2kx2 l2 − 2kx1 l1 )x + (2kx1 l12 + 2kx2 l22 )ϕ (1.86)
The above equations can be rewritten in the following matrix form:

⎡ ⎤ ⎡ ⎤⎡ ⎤
Fx 2kx1 + 2kx2 0 2kx2 l2 − 2kx1 l1 0 x
⎢ Fy ⎥ ⎢ 0 k l − k l 0 k l − ky1 l1 ⎥⎢ y ⎥
⎢ ⎥ ⎢ y2 2 y1 1 y2 2 ⎥⎢ ⎥
⎣ Mx ⎦ = − ⎣ 2kx2 l2 − 2kx1 l1 0 2 2
2kx1 l1 + 2kx2 l2 0 ⎦ ⎣ ϕx ⎦
My 0 ky2 l2 − ky1 l1 0 ky2 l2 − ky1 l1 ϕy
= −k · x (1.87)
MATHEMATICAL MODELS OF RIGID ELEMENTS 35

Hence the equation of motion of the rotor system is

mẍ + cẋ + kx = +FG (1.88)

MATHEMATICAL MODELS OF RIGID ELEMENTS 36

Problem 3

2 1

G R
O z

δy µy
R
l L
a
y

Figure 12

Consider rotor described in problem 1. Assume that the angle between axis
of symmetry of the body 1 and axis of rotation has small magnitude δ y as shown in
Fig. 12. Develop mathematical model of the rotor.
MATHEMATICAL MODELS OF RIGID ELEMENTS 37

Solution.
According to notations adopted in section 1.3, the residual unbalances of the
rotor considered are:

∆xR = −δ y ∆yR = 0 µxR = 0 µyR = a δ y (1.89)

Hence the vector of unbalanced forces at the centre of gravity G, according to Eq.
1.65 is: ⎡ ⎤ ⎡ ⎤
U FGx SΩ2 cos(Ω t + ϕS )
⎢ UFGy ⎥ ⎢ SΩ2 sin(Ω t + ϕS ) ⎥
FUG = ⎢ ⎥ ⎢
⎣ UMGx ⎦ = ⎣ DΩ2 cos(Ω t + ϕD ) ⎦
⎥ (1.90)
UMGy DΩ2 sin(Ω t + ϕD )
where

S = mµyR = ma δ y , ϕS = π/2, D = 2(I − Iz )δ y , ϕD = π/2 (1.91)

The equivalent set of forces at the origin of the system of coordinates O (see Fig. 13)
is:

2 1

G R
O z

UMy y δ µy
l UFy UMG R
L y
a UFGy
y

Figure 13

⎡ ⎤ ⎡ ⎤
U Fx U FGx
⎢ UFy ⎥ ⎢ UFGy ⎥
FU = ⎢ ⎥ ⎢
⎣ UMx ⎦ = ⎣ UMGx + aUFGx
⎥
⎦ (1.92)
UMy UMGy + aUFGy
This vector has to be added to the mathematical model 1.82.

mẍ + cẋ + kx = FG +FU (1.93)

MATHEMATICAL MODELS OF SHAFTS 38

1.2 MATHEMATICAL MODELS OF SHAFTS

Shaft of rotor, in many cases, has to be considered as flexible element of continuously
varied cross-section and mass distribution. The shaft, similarly as the rigid elements,
rotates with the angular velocity Ω.
1.2.1 Modelling of dynamic properties of the free-free shaft
Commercially available computer packages allow to produce stiffness and inertia ma-
trix of free in space beam along arbitrary chosen number N of coordinates yn and
ϕyn (see Fig. 14). These distinguished points n are called nodes.

1 2 ......... n .............................................................. N

yn

ϕy
n

Figure 14

There are two the most popular approaches for creation of the stiffness and
inertia matrices called Rigid Element Method and Finite Element Method.

Rigid Elements Method

According to the Rigid Element Method, the beam is divided into a sufficient, for
necessary accuracy, number of segments I of constant cross-section (Fig . 15a))The
bending and shearing properties of each segment are represented by two springs of
stiffness kMi and kTi respectively (Fig. 15.b). Equivalence of both, the actual element
(Fig. 16a) and its model (Fig. 16b) requires equal angular deflection ( δri = δ ei )
caused by the same bending moment Mi .
Since:
Mi li Mi
δ ri = and δ ei = (1.94)
2EJi 2kMi
the bending stiffness is
EJi
kMi = (1.95)
li
Similarly, the equivalence of shearing deflections (yri = yei ) caused by the same
shearing force Ti (Fig. 16c and Fig. 16d)
Ti li Ti
yri = li γ ri = and yei = (1.96)
GAi kTi
yields
GAi
kTi = (1.97)
li
MATHEMATICAL MODELS OF SHAFTS 39

l i-1 li l i+1

..... .....
a)

li
kT i
b)
k Mi

l i-1 /2 l i/ 2
mi c)
Gi
Ii

mi m i+1
k T i -1 k T i+1
Ii I i +1
.....

k Mi-1 k Mi k Mi+1
.....

z d)

zli z ri z l i+1 z ri+1

y

Figure 15

E,Ji G,Ai
li li
Ti y
Mi Mi a) r i c)
Ti
δri

δe i Ti ye
i
Mi Mi
b) Ti kT i d)
2δe i
k Mi

Figure 16
MATHEMATICAL MODELS OF SHAFTS 40

The right hand part of the segment li−1 and the left hand part of the subsequent
segment li , form a section (Fig. 15c). Each section is considered rigid and its
inertia properties are represented by mass mi and moments of inertia Ii . In this
way a complete symmetry is obtained that gives simple programming for computer
analysis.
Equations of motion that govern motion of the beam which physical model is
shown in Fig. 15d) have the following form,

mb ÿb +kb yb = 0 (1.98)

where: ⎡ ⎤
M1,1 .. 0 0 0 .. 0
⎢ .̇ .̇.̇ .̇ .̇ .̇ .̇.̇ .̇ ⎥
⎢ ⎥
⎢ 0 .. Mi−1,i−1 0 0 .. 0 ⎥
⎢ ⎥
mb = ⎢
⎢ 0 .. 0 Mi,i 0 .. 0 ⎥
⎥ (1.99)
⎢ 0 .. 0 0 Mi+1,i+1 .. 0 ⎥
⎢ ⎥
⎣ .̇ .̇.̇ .̇ .̇ .̇ .̇.̇ .̇ ⎦
0 .. 0 0 0 .. MN,N
⎡ ⎤
K1,1 .. 0 0 0 .. 0
⎢ .̇ .̇.̇ .̇ .̇ .̇ .̇.̇ .̇ ⎥
⎢ ⎥
⎢ 0 .. Ki−1,i−1 Ki−1,i 0 .. 0 ⎥
⎢ ⎥
kb = ⎢
⎢ 0 .. Ki,i−1 Ki,i Ki,i+1 .. 0 ⎥
⎥ (1.100)
⎢ 0 .. 0 Ki+1,i Ki+1,i+1 .. 0 ⎥
⎢ ⎥
⎣ .̇ .̇.̇ .̇ .̇ .̇ .̇.̇ .̇ ⎦
0 .. 0 0 0 .. KN,N
⎡ ⎤
y1
⎢ : ⎥
⎢ ⎥
⎢ yi−1 ⎥
⎢ ⎥
yb = ⎢
⎢ yi
⎥
⎥ (1.101)
⎢ yi+1 ⎥
⎢ ⎥
⎣ : ⎦
yN
N =I +1 (1.102)
∙ ¸
mi 0
Mi,i = (1.103)
0 Ii
∙ ¸
−kTi−1 −kTi−1 zri−1
Ki,i−1 = (1.104)
+kTi−1 zli −kMi−1 + kTi−1 zri−1 zli
∙ ¸
+kTi−1 + kTi +kTi−1 zli + kTi zri
Ki,i = (1.105)
+kTi−1 zli + kTi zri +kMi−1 + kTi−1 zl2i + kTi zr2i
∙ ¸
−kTi +kTi zli+1
Ki,i+1 = (1.106)
−kTi zri −kMi + kTi zli+1 zri
∙ ¸
yi
yi = (1.107)
ϕyi
MATHEMATICAL MODELS OF SHAFTS 41

If the shaft is symmetric, its motion in plane xz and yz is governed by equations that
can be composed from matrices mb and kb .

mẍ + kx = 0 (1.108)
where:
∙ ¸
mb 0
m= (1.109)
0 mb
∙ ¸
kb 0
k= (1.110)
0 kb
© ªT
x = x1 , ϕx1 , ......xi , ϕxi , ......xN , ϕxN , y1 ϕy1 , ......yN ϕyN (1.111)
The geometrical interpretation of the vector of coordinates 1.111 is given in Fig. 17.

element 1 element i element N

Oi
z
node 1 node i node N
x
x i yi ϕ xi ϕ yi

Figure 17

The coordinates xi , ϕxi , yi , ϕyi 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. Fi ) can be equivalently replaced by the force Fi applied
to the node Oi and the moment Mi = ai · Fi as shown in Fig. 18This equivalent set of
forces along the nodal coordinates yi , ϕyi should be added to the mathematical model
1.108. In a general case these forces can be independent of time (static forces) or
they can depend on time (excitation forces). Introducing notations Fs for the static
forces and F(t) for the excitation forces, the equation of motion of the free-free beam
takes the following form.
mẍ + kx = Fs + F(t) (1.112)

Finite Elements Method

According to the Finite Elements Method, the shaft is divided into a number of the
uniform and flexible elements. The i − th element is shown in Fig. 19.
MATHEMATICAL MODELS OF SHAFTS 42

Fi
Mi element i

Oi z
ai node i
yi ϕ
Fi yi

Figure 18

z
1
yi1
E i Ji Ai ρi yi2

ϕyi1
li 2
ϕyi2
y

Figure 19

In this figure Ei , Ji , Ai , and ρi stand for Young modulus, second moment of

area about the neutral axis, area of cross-section and the unit mass of the element.
The differential equation of the statically deflected line of the element in the plane
yz is
d4 y(z)
Ei Ji =0 (1.113)
dz 4
Integration of the above equation four times yields
1 1
y(z) = C1 z 3 + C2 z 2 + C3 z + C4 (1.114)
6 2
The constants of integration Cj (j = 1, 2, 3, 4) must be chosen to fulfill the following
boundary conditions
¯ ¯
dy(z) ¯¯ dy(z) ¯¯
y(z)|z=0 = yi1 ; = ϕyi1 y(z)|z=li = yi2 ; = ϕyi2 (1.115)
dz ¯z=0 dz ¯z=li

The parameters yi1 and yi2 are called nodal displacements and the parameters ϕi1
and ϕi2 are called nodal rotations. The nodes are denoted by numbers 1 and 2. Intro-
duction of solution 1.114 into the above boundary conditions results in the following
MATHEMATICAL MODELS OF SHAFTS 43

set of algebraic equations linear with respect to the constants Cj .

⎡ ⎤⎡ ⎤ ⎡ ⎤
0 0 0 1 C1 yi1
⎢ 0 0 ⎥ ⎢ C2 ⎥ ⎢ ϕyi1 ⎥
⎢ 1 3 102 1 ⎥⎢ ⎥=⎢ ⎥ (1.116)
⎣ li l li 1 ⎦ ⎣ C3 ⎦ ⎣ yi2 ⎦
6 2 i
1 2
l
2 i
li 1 0 C4 ϕyi2

Its solution yields the integration constants Cj .

⎡ ⎤ ⎡ ⎤−1 ⎡ ⎤ ⎡ 6
⎤
C1 0 0 0 1 yi1 li3
+ li ϕyi1 − 2yi2 + li ϕyi2 )
(2yi1
⎢ C2 ⎥ ⎢ 0 ⎥ ⎢ ⎥ ⎢ 2 ⎥
⎢ ⎥ = ⎢ 1 3 102 1 0 ⎥ ⎢ ϕyi1 ⎥ = ⎢ ⎢ li2
(−3yi1 − 2li ϕyi1 + 3yi2 − li ϕyi2 ) ⎥
⎥
⎣ C3 ⎦ ⎣ li l li 1 ⎦ ⎣ yi2 ⎦ ⎣ ϕyi1 ⎦
6 2 i
1 2
C4 l
2 i
li 1 0 ϕyi2 yi1
(1.117)
After introduction of Eq. 1.117 into the equation of the deflected line 1.114 one may
get it in the following form.
" µ ¶2 µ ¶3 # "µ ¶ µ ¶2 µ ¶3 #
z z z z z
y(z) = 1 − 3 +2 yi1 + −2 + li ϕyi1
li li li li li
" µ ¶ µ ¶3 # " µ ¶ µ ¶3 #
2 2
z z z z
+ 3 −2 yi2 + − + li ϕyi2
li li li li
= {H(z)}T {y} (1.118)

where: ⎧ ³ ´2 ³ ´3 ⎫
⎪
⎪ 1 − 3 z
+ 2 z ⎪
⎪
⎧ ⎫ ⎪ ⎪
⎪ ∙ ³ ´
li li
³ ´2 ³ ´3 ¸
⎪
⎪
⎪
⎪ H ⎪ ⎪
⎪ ⎪
⎪
⎪
⎨
1
⎬ ⎪
⎪ ⎨ z
− 2 lz + lz li ⎪
⎬
H2 li li i i
{H(z)} = = ³ ´2 ³ ´3 (1.119)
⎪
⎪ H3 ⎪ ⎪ ⎪ ⎪
⎩ ⎭ ⎪ ⎪
⎪ 3 lz − 2 lz ⎪
⎪
⎪
H4 li ⎪
⎪ ∙ ³ i ´2 ³ ´i 3 ¸ ⎪
⎪
⎪
⎪ ⎪
⎪
⎩ − z + z li
l li i
⎭
⎧ ⎫
⎪
⎪ yi1 ⎪
⎪
⎨ ⎬
ϕyi1
{y} = (1.120)
⎪
⎪ yi2 ⎪
⎪
⎩ ⎭
ϕyi2
Functions H1 , H2 , H3 , H4 (see Eq. 1.119) are known as Hermite cubics or shape func-
tions. The matrix {y} contains the nodal coordinates. As it can be seen from Eq.
1.118 the deflected line of the finite element is assembled of terms which are linear
with respect to the nodal coordinates.
If the finite 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)} (1.121)

MATHEMATICAL MODELS OF SHAFTS 44

As one can see from the equation 1.121, the dynamic deflection line is approximated
by the static deflection 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
∂ 4 y(z, t) ∂ 2 y(z, t)
Ei Ji − ρi =0 (1.122)
∂z 4 ∂t2
by its discreet representation along the nodal coordinates

[mi ] {ÿ (t)} + [ki ] {y (t)} = 0. (1.123)

In the above equations ρi stands for the unit mass of the finite element and [mi ] and
[ki ] stands for the inertia and stiffness 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 finite
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 finite
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 finite element
is
Z µ ¶2
1 li ∂y(z, t)
T = ρ dz
2 0 i ∂t
Z µ ¶µ ¶
1 li ∂y(z, t) ∂y(z, t)
= ρ dz
2 0 i ∂t ∂t
Z
1 li ³ T
´³
T
´
= ρ {ẏ (t)} {H(z)} {H(z)} {ẏ (t)} dz
2 0 i
∙ Z li ¸
1 T T
= {ẏ (t)} ρi {H(z)} {H(z)} dz {ẏ (t)}
2 0
⎛ ⎡ 2
⎤ ⎞
Z li H 1 H 1 H 2 li H1 H 3 H1 H4 l i
1 ⎜
T ⎜
⎢ H2 H1 li H22 li2 H2 H3 li H2 H4 li2 ⎥ ⎟
{ẏ (t)} ⎝ρi ⎢ ⎥ ⎟
=
2 ⎣ H3 H1 H3 H2 li H 2
3 H3 H4 l i
⎦ ⎠ {ẏ (t)}
dz
0
H4 H11 li H4 H2 li2 H4 H3 li H42 li2
(1.124)

It is easy to see that the last row of Eq. 1.124 represents kinetic energy function
of the discreet physical model along the nodal coordinates yi1 ϕyi1 yi2 ϕyi2 with
the following matrix of inertia.
MATHEMATICAL MODELS OF SHAFTS 45

⎡ 2
⎤
Z li H 1 H 1 H2 li H1 H 3 H 1 H4 li
⎢ H2 H1 li H22 li2 H2 H3 li H2 H4 li2 ⎥
mi = ρi ⎢ ⎥ dz
⎣ H3 H1 H3 H2 li H32 H3 H4 li ⎦
0
H4 H1 li H4 H2 li2 H4 H3 li H42 li2
⎡ 13 11 9 13
⎤
35 210
l i 70
− 420
li
⎢ 11 li 1 2
li 13
li − 140 1 2 ⎥
li ⎥
⎢
= ρi li ⎣ 210 105 420
9
70
13
420 i
l 13
35
− 210 li ⎦
11
13 1 2 11 1 2
− 420 li − 140 li − 210 li 105 li
⎡ 13 11 9 13
⎤
35 210 i
l 70
− 420 li
⎢ 11 li 1 2
l 13
l − 140 1 2 ⎥
li ⎥
= mi ⎢ ⎣
210 105 i 420 i (1.125)
9
70
13
420 i
l 13
35
− 11
210 i
l ⎦
13 1 2 11 1 2
− 420 li − 140 li − 210 li 105 li
In the last formula mi stands for mass of the finite element.
To take advantage of the second criterion let us produce expression for the
potential energy function for the continues physical model of the finite element.
Z µ 2 ¶2
1 li ∂ y(z, t)
V = Ei Ji dz
2 0 ∂z 2
Z µ 2 ¶µ 2 ¶
1 li ∂ y(z, t) ∂ y(z, t)
= Ei Ji dz
2 0 ∂z 2 ∂z 2
Z µ ½ 2 ¾¶ ý 2 ¾T !
1 li d H(z) d H(z)
= Ei Ji {y (t)}T {y (t)} dz
2 0 dz 2 dz 2
" Z li ½ 2 ¾½ 2 ¾T #
1 d H(z) d H(z)
= {y (t)}T Ei Ji dz {y (t)}
2 0 dz 2 dz 2
⎛ ⎡ ¡ 00 ¢2 00 00 00 00 00 00 ⎤ ⎞
Z li H1 H1 H2 li H1 H3 H1 H4 li
1 ⎜ ⎢ H 00 H 00 li (H 00 )2 l2 H 00 H 00 li H 00 H 00 l2 ⎥ ⎟
= {y (t)}T ⎜ E J
⎝ i i
⎢ 200 100
⎣ H H
2 i 2 3 2 4 i ⎥ dz ⎟ {y (t)}
H3 H2 li (H3 )2 H3 H4 li ⎦ ⎠
00 00 00 00 00
2 0 3 1
00 00 00 00 2 00 00 00 2 2
H4 H1 li H4 H2 li H4 H3 li (H4 ) li
(1.126)
As one can see from Eq. 1.126, to fulfill the second criterion, the stiffness matrix
along the nodal coordinates yi1 ϕyi1 yi2 ϕyi2 must be as follows.
⎡ ¡ 00 ¢2 00 00 00 00 00 00 ⎤
Z li H 1 H1 H 2 li H 1 H 3 H1 H 4 li
⎢ H 00 H 00 li (H 00 )2 l2 H 00 H 00 li H 00 H 00 l2 ⎥
ki = Ei Ji ⎢ 200 100 2 i 2 3 2 4 i ⎥ dz
⎣ H H H
00
H
00
l (H
00 2
) H
00 00
⎦
0 3 1 3 2 i 3 3 H4 li
00 00 00 00 2 00 00 00 2 2
H4 H1 li H4 H2 li H4 H3 li (H4 ) li
⎡ ⎤
12 6li −12 6li
Ei Ji ⎢
⎢ 6li 4li2 −6li 2li2 ⎥ ⎥
= (1.127)
li3 ⎣ −12 −6li 12 −6li ⎦
6li 2li2 −6li 4li2
MATHEMATICAL MODELS OF SHAFTS 46

Hence, the mathematical model of the element considered can be written as

mi ÿi + ki yi = Ryi (1.128)

The vector Ri represents the interaction forces between the neighborhood elements.
£ ¤T
Ryi = Ryi1 Rϕyi1 Ryi2 Rϕyi2 (1.129)

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.

mj ÿj + kj yj = Ryj (1.130)

where:
⎡ 13 11 9 13
⎤
35
l
210 j 70
− 420 lj
⎢ 11
l 1 2
l 13
l 1 2 ⎥
− 140 lj ⎥
mj = mj ⎢
⎣
210 j 105 j 420 j
9
70
13
l
420 j
13
35
− 210 lj ⎦
11
13 1 2 11 1 2
− 420 lj − 140 lj − 210 lj l
105 j
⎡ ⎤
12 6lj −12 6lj
⎢
Ej Aj ⎢ 6lj 4lj2 −6lj 2lj2 ⎥
kj = ⎥ (1.131)
lj3 ⎣ −12 −6lj 12 −6lj ⎦
6lj 2lj2 −6lj 4lj2
£ ¤T
yj = yj1 ϕyj1 yj2 ϕyj2 (1.132)
£ ¤T
Ryj = Ryj1 Rϕyj1 Ryj2 Rϕyj2 (1.133)
These two equations of motion (1.128 1.130) associated with the two elements i and
j may be joint one to the other to create one mathematical model representing both
elements. Details of this process of composition of the mathematical model is given
in chapter 2 (page 115). In the case considered here, the compatibility (continuity
and equilibrium) 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
∙ ¸ ∙ ¸ ∙ ¸
yi2 yj1 yij
= = (1.134)
ϕyi2 ϕyj1 ϕyij

and the equilibrium conditions are

∙ ¸ ∙ ¸ ∙ ¸
Ryi2 Ryj1 0
+ = (1.135)
Rϕyi2 Rϕyj1 0

They results in the following mathematical model of the joint elements.

mij ÿij + kij yij = Ryij (1.136)

where:
MATHEMATICAL MODELS OF SHAFTS 47

⎡ 13 11 9 13
⎤
m
35 i
lm
210 i i
m
70 i
− 420 li mi 0 0
⎢ 11
lm 1 2
l mi 13
lm 1 2
− 140 li mi 0 0 ⎥
⎢ 210 i i 105 i 420 i i ⎥
⎢ 9
m 13
lm 13
(m + m ) 11
(−l m +l m ) 9
m − 13
lm ⎥
mij = ⎢
⎢
70 i
13
420 i i
1 2 11
35 i j 210
1 2
i i j j
2
70 j
13
420 j j
1 2
⎥
⎥
⎢ − 420 li mi − 140 li mi 210
(−li mi +lj mj ) 105 (li mi + lj mj ) 420 lj mj − 140 lj mj ⎥
⎣ 0 0 9
m 13
lm 13
m 11
− 210 lj mj ⎦
70 j 420 j j 35 j
13 1 2 11 1 2
0 0 − 420 lj mj − 140 lj mj − 210 lj mj 105 lj mj

(1.137)
⎡ ⎤
12 Eli3Ai 6 Eli2Ai −12 Eli3Ai 0 6 Eli2Ai 0
⎢ i i i i
⎥
⎢ 6 Eli2Ai 4 EilAi −6 Eli2Ai 0 2 EilAi 0 ⎥
⎢ i i i i ⎥
⎢ −129 Ei3Ai −6 Ei2Ai 12( Ei3Ai + j3 j ) 6(− Ei2Ai + j2 j ) −12 Ej3Aj
E A E A Ej Aj ⎥
6 l2
⎢ li li li lj li lj lj ⎥
kij = ⎢ Ej Aj ⎥
j
⎢ 6 Ei2Ai Ei Ai Ei Ai Ej Aj Ei Ai Ej Aj Ej Aj ⎥
⎢ li
2 l 6(− l2 + l2 ) 4( l + l ) −6 l2 2 l ⎥
i i j i j j j
⎢ E A E A E A Ej Aj ⎥
⎢ 0 0 −12 jl3 j −6 jl2 j 12 jl3 j −6 l2 ⎥
⎣ j j j j ⎦
E A E A E A E A
0 0 6 jl2 j 2 jl j −6 jl2 j 4 jl j
j j j j

(1.138)
£ ¤T
yij = yi1 ϕyi1 yij ϕyij yj2 ϕyj2 (1.139)
£ ¤T
Rij = Ryi1 Rϕyi1 0 0 Ryj2 Rϕyj2 (1.140)

Repetition of the described procedure to all elements of the shaft results in

the mathematical model of the shaft in the plane yz.

mb ÿb +kb yb = 0 (1.141)

If the shaft is symmetric, its motion in the plane xz and yz is governed by equation
that can be composed of the derived matrices mb and kb .

mẍ + kx = 0 (1.142)

where ∙ ¸ ∙ ¸
mb 0 kb 0
m= k= (1.143)
0 mb 0 kb
© ªT
x = x1 , ϕx1 , ......xi , ϕxi , ......xN , ϕxN , y1 ϕy1 , ......yN ϕyN (1.144)
Geometrical interpretation of the nodal coordinates appearing in the Eq. 1.144 is
given in Fig. 20.
MATHEMATICAL MODELS OF SHAFTS 48

element 1 element i element N-1

Oi
z
node 1 node i node N
x
x i yi ϕ xi ϕ yi

Figure 20

The coordinates xi , ϕxi , yi , ϕyi are associated with nodes which are located at
the ends of the finite elements.

Introduction of the external forces Since the finite 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 Fi (see Fig. 21) on the displacement
yi is equal to the virtual work produced by a set of forces along the coordinates
yi1 ϕyi1 yi2 ϕyi2 . Hence
⎧ ⎫
⎪
⎪ yi1 ⎪⎪
⎨ ⎬
ϕyi1
∂Wi = Fi · yi (ai , t) = {Fi1 , Mi1 , Fi2 , Mi2 } (1.145)
⎪
⎪ y ⎪
⎩ i2 ⎪ ⎭
ϕyi2

But according to 1.121

ai Fi
z
1
yi1
Ei Ji Ai ρi yi2

ϕyi1
li 2
ϕyi2
y

Figure 21
MATHEMATICAL MODELS OF SHAFTS 49

⎧ ³ ´2 ³ ´3 ⎫T
⎪
⎪ 1−3 l ai
+ 2 al i ⎪
⎪
⎪
⎪ ∙³ ´ ⎪
⎪ ⎧ ⎫
⎪
⎪ ³ ´2 ³ i ´3 ¸
i
⎪
⎪
⎪
⎪ ⎪
⎪ ⎪ yi1 ⎪
⎨ ai
− 2 al i + al i li ⎬ ⎪⎨ ⎪
⎬
T
li i i ϕyi1
y(ai , t) = {H(ai )} {y (t)} = ³ ´ 2 ³ ´ 3 (1.146)
⎪
⎪ ai ai ⎪
⎪ ⎪
⎪ y ⎪
⎪
⎪
⎪ 3 − 2 ⎪
⎪ ⎩ i2 ⎭
⎪
⎪ ∙ ³ li ´ l
³ ´i 3 ¸ ⎪
⎪ ϕyi2
⎪
⎪ 2 ⎪
⎪
⎩ − al i + al i li ⎭
i i

Introduction of the above expression into the expression for the virtual work yields
⎧ ³ ´2 ³ ´3 ⎫T
⎪
⎪ 1 − 3 ai
+ 2 ai ⎪
⎪
⎪
⎪ ∙³ ´ li l ⎪ ⎧
⎪ ⎫
⎪
⎪ ³ ´2 ³ i ´3 ¸ ⎪
⎪
⎪
⎪ ⎪
⎪ ⎪ yi1 ⎪
⎨ ai
li
− 2 al i + al i li ⎬ ⎪⎨ ⎪
⎬
i i ϕyi1
∂Wi = Fi · ³ ´2 ³ ´3 (1.147)
⎪
⎪ 3 al i − 2 al i ⎪
⎪ ⎪
⎪ yi2 ⎪
⎪
⎪
⎪ ∙ ³i´ ⎪
⎪ ⎩ ⎭
⎪
⎪
⎪ 2 ³ ´i 3 ¸ ⎪
⎪
⎪
ϕyi2
⎪
⎩ − al i + al i li ⎪
⎭
i i

Hence, the vector of forces along the nodal coordinates is

⎧ ³ ´2 ³ ´3 ⎫
⎪
⎪ 1−3 l ai
+ 2 al i ⎪
⎪
⎧ ⎫ ⎪
⎪ ∙³ ´ ⎪
⎪
F
⎪
⎪ ³ ´2 ³ i ´3 ¸
i
⎪
⎪
⎪
⎪ ⎪
i1 ⎪ ⎪
⎪ ai ⎪
⎪
⎨ ⎬ ⎨ li
− 2 al i + al i li ⎬
Mi1 i i
= Fi · ³ ´ 2 ³ ´ 3 (1.148)
⎪
⎪ F ⎪ ⎪ ⎪
⎩ i2 ⎪ ⎪ ⎪
ai ai
⎭ ⎪
⎪ 3 − 2 ⎪
⎪
Mi2 ⎪
⎪ ∙ ³ li ´ l
³ ´i 3 ¸ ⎪
⎪
⎪
⎪ 2 ⎪
⎪
⎩ − al i + al i li ⎭
i i

This forces have to be introduced into the equation of motion 1.142

mẍ + kx = Fs + F(t) (1.149)

where, similarly as before, Fs stands for the static forces and F(t) stands for the
excitation forces.
1.2.2 Modelling of residual permanent deformation
The mathematical model 1.108 was developed assuming that the centres of cross-
section of the shaft Ci coincide origins Oi of all the local systems of coordinates
xRi yRi zRi (see Fig. 22a). Due to the manufacturing accuracy and the thermal treat-
ment, the shaft is permanently deflected as it is shown in Fig. 22b.
MATHEMATICAL MODELS OF SHAFTS 50

O Ci Oi
z

xR xR
a)
i
yR yRi

O Oi
ayi axi z
α xi
Ci α yi
xR xR i b)
yR yR i

O Ci Oi
Fa xRi z

Fa c)
yRi

xR xR i
yR yR i

Figure 22

This deformation can be determined by two vectors ax and ay .

⎡ .. ⎤ ⎡ .. ⎤
. .
⎢ ∙ ¸ ⎥ ⎢ ∙ ¸ ⎥
⎢ axi ⎥ ⎢ ayi ⎥
ax = ⎢ ⎥ ay = ⎢ ⎥ (1.150)
⎣ αxi ⎦ ⎣ αyi ⎦
.. ..
. .
To include this permanent deformation of the shaft in its mathematical model let us
determine forces that cause this deformations along the local systems of coordinates
xRi yRi zRi . According to Eq. 1.98, these forces are as follows

FaxR = kb ax
(1.151)
FayR = kb ay
MATHEMATICAL MODELS OF SHAFTS 51

Fa x
Ci Oi i
xi
Fa xRi
Fa yRi Ωt

Fa y i
xRi

yR i yi

Figure 23

These forces rotate with the shaft with the angular velocity Ω. Hence, their
components along the stationary system of coordinates xyz, according to Fig. 23 are
as follows
∙ ¸ ∙ ¸ ∙ ¸
Fax FaxR cos Ωt − FayR sin Ωt kb ax cos Ωt − kb ay sin Ωt
Fa (t) = = =
Fay FaxR sin Ωt + FayR cos Ωt kb ay cos Ωt + kb ax sin Ωt
⎡ .. ⎤
.
⎢ ⎥
= ⎣ Fan cos(Ωt + ϕan ) ⎦ (1.152)
..
.

The above vector represents excitation force due to the residual permanent deforma-
tion and has to be included in the mathematical model 1.108.

mẍ + kx = Fa (t) (1.153)

In the consideration above, the number of coordinates the permanent deformation was
determined along was equal to number of coordinates adopted for the mathematical
model. Not always this requirement is fulfil. Therefore let us now consider case when
the permanent deformation of a shaft given by mathematical model 1.108
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸
mb 0 ẍ kb 0 x 0
+ = (1.154)
0 mb ÿ 0 kb y 0
is determined along some of the coordinates xR and yR of the rotating system of
coordinates.
∙ ¸ ∙ ¸
xRa yRa
xR = yR = (1.155)
xRr yRr
Let xRa and yRa be these coordinates the permanent deformation is determined along.
By ax and ay let us denote vectors of the permanent deformation in plane xR z and
yR z of the rotating system of coordinates respectively (see Fig 22). The equivalent
set of forces FaxR , which causes the deformation in the plane xR z , has form
MATHEMATICAL MODELS OF SHAFTS 52

∙ ¸
FxRa
FaR = (1.156)
0
The vector FxRa has to fulfil the following relationship:
∙ ¸∙ ¸ ∙ ¸
kbaa kbar ax FxRa
= (1.157)
kbra kbrr xRr 0
which is equivalent to

kbaa ax + kbar xRr = FxRa

kbra ax + kbrr xRr = 0 (1.158)
Second equation yields

xRr = −k−1
brr (kbra ax ) (1.159)
Introduction of equation 1.159 into the first equation of the system 1.158 produces
the wanted expression for FxRa
FxRa = kbaa ax + kbar (k−1
brr (−kbra ax )) (1.160)
Similarly, the equivalent vector of forces in plane yR z is
∙ ¸
FyRa
FayR = (1.161)
0
where
FyRa = kbaa ay + kbar (k−1
brr (−kbra ay )) (1.162)
Equivalent vector of forces along the stationary system of coordinates, according to
Eq. 1.152 is ∙ ¸ ∙ ¸
Fax FaxR cos Ωt − FayR sin Ωt
Fa (t) = = (1.163)
Fay FaxR sin Ωt + FayR cos Ωt
1.2.3 Modelling of residual unbalance
In the above modelling we have assumed that the centre of gravity of each section Gi
coincides axis of rotation of the shaft Oi and principal axes of each section coincide
axes of rotation. This assumption is never strictly fulfilled.

O Gi Oi
ϕx z
i
ϕy
xi yi i

x xi
y yi

Figure 24
MATHEMATICAL MODELS OF SHAFTS 53

The residual unbalance of each rotor section i, according to the consideration

carried out section 1.3 of the chapter 1, along coordinates xi , ϕxi , yi , ϕyi (see Fig. 24),
results in vector of external excitation
⎡ ⎤
Si cos(Ωt + ϕSi )
⎢ Di cos(Ωt + ϕD ) ⎥
FUi = Ω2 ⎢ i
⎣ Si sin(Ωt + ϕS ) ⎦
⎥ (1.164)
i
Di sin(Ωt + ϕDi )
Hence, vector of excitation of the whole shaft is of the following form
⎡ ⎤
.̇
⎢ Si cos(Ωt + ϕS ) ⎥
⎢ i ⎥ ⎡ ⎤
⎢ Di cos(Ωt + ϕD ) ⎥ .̇
⎢ i ⎥
FU (t) = Ω2 ⎢
⎢ .̇ ⎥ = Ω2 ⎣ FUn cos(Ωt + ϕU ) ⎦
⎥ n
(1.165)
⎢ Si sin(Ωt + ϕS ⎥ .̇
⎢ i ⎥
⎣ Di sin(Ωt + ϕD ) ⎦
i
.̇

Adding this vector to the mathematical model 1.153 one can modify these equations
to the following form
mẍ + kx = Fa (t) + FU (t) (1.166)
1.2.4 Interaction forces

O Oi
ϕx z
i
Mx i ϕy
i
My i Rx
i
x Ryi xi
y yi

Figure 25

If the shaft is to be connected along a number of coordinates with other elements or

foundation (see Fig. 25), one have to introduce interaction forces which act on the
shaft. If the interaction forces are denoted by R, the governing equations take the
following form
mẍ + kx = R + Fa (t) + FU (t) + FG
MATHEMATICAL MODELS OF SHAFTS 54

1.2.5 Boundary conditions

O Oi
ϕx z
i
ϕy
bxi byi i

x Bi xi
y yi

Figure 26

Let us assume that the shaft is rigidly supported upon several supports Bi (see Fig.
26) which instantaneous position is determine with respect to the stationary system
of coordinates xyz by coordinates bxi (t) and byi (t) . Let us denote by b vector of such
coordinates. ⎡ ⎤
..
.
⎢ ⎥
⎢ bxi (t) ⎥
b=⎢ ⎥ (1.167)
⎣ byi (t) ⎦
..
.
Let us reorganize vector of coordinates of the shaft
© ªT
x = x1 , ϕx1 , ......xn , ϕxn , ......xN , ϕxN , y1 ϕy1 , ......yN ϕyN (1.168)
in such a way that its upper part xb contains coordinates along which the shaft is
rigidly supported and its lower part xr contains all the remaining coordinates
∙ ¸
xb
x= (1.169)
xr
and let us assume that the mathematical model of the shaft is organized with respect
to the above vector of coordinates.
mẍ + kx = R + F (1.170)
where:
F = Fa (t) + FU (t) + FG (1.171)
Upon partitioning the above equation
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸ ∙ ¸
mbb mbr ẍb kbb kbr xb Rb Fb
+ = + (1.172)
mrb mrr ẍr krb krr xr Rr Fr
one can obtain two sets of equations
mbb ẍb + mbr ẍr + kbb xb + kbr xr = Rb + Fb
mrb ẍb + mrr ẍr + krb xb + krr xr = Rr + Fr (1.173)
MATHEMATICAL MODELS OF SHAFTS 55

Since vector xb has to be, in the case considered, equal to the vector b, substitution
of this relationship into the second equation of Eqs. 1.173 yields equations of motion
of the rotor in the following form.

mrr ẍr + krr xr = Rr + Fr − mrb b̈b − krs bb (1.174)

Motion of the system supported upon the supports Bi along the coordinates xr is
governed by solution of the equation 1.174. If one denote this motion by Xr (t), the
first equation of Eqs. 1.173 offers the dynamic reactions along coordinates the shaft
is supported upon.

Rb = mbb b̈ + mbr Ẍr + kbb b + kbr Xr (t) − Fb (1.175)

It is easy to notice that the equation of motion of the supported shaft Eq. 1.174 has
the same form as the unsupported one.

mẍ + kx = R + F (1.176)

where:
F = Fa (t) + FU (t) + FG − mrb b̈b − krs bb (1.177)
The last two terms represent the external excitation caused by motion of the supports
Bi .
1.2.6 Constraints imposed by rigid bodies
According to the adopted assumptions, motion of the elements along the axis z is
neglected and rotation about axis z is assumed to be known. Therefore a rigid body
possesses two degrees of freedom in the plane xz and has another two degrees of
freedom in the plane yz. E.g. coordinates yR1 and ϕR1 uniquely determine position
of the rigid body in the yR z plane of the rotating system of coordinates xR yR z (see Fig.
27). Hence, if one choose them as the independent coordinates, position of the body
along the coordinates yR2 , ϕR2 , yR3 , ϕR3 , yR4 , ϕR4 can be expressed as functions of yR1
and ϕR1 , dimensions of the body l2−1 , l1 , ... and parameters reflecting the accuracy
of its manufacturing αRy .
MATHEMATICAL MODELS OF SHAFTS 56

x
1 2 3 4 xR z

yR1 y R2 yR3 yR 4 z
y
yR
ϕ = ϕR
R1 2 ϕ ϕ xR
R3 = R4
x
l 1-2 αRy
l1
l2
l 3-4
yR Ωt
y yR

Figure 27

yR2 = yR1 + l2−1 ϕR1

ϕR2 = ϕR1
yR3 = yR2 + l1 ϕR1 + l2 (ϕR1 − αRy ) = yR1 + (l2−1 + l1 + l2 )ϕR1 + (−l2 )αRy
ϕR3 = ϕR1 − αRy
yR4 = yR3 + l4−3 (ϕR1 − αRy ) = yR1 + (l2−1 + l1 + l2 + l4−3 )ϕR1 + (−l2 − l4−3 )αRy
ϕR4 = ϕR1 − αRy (1.178)

The above constraints may be rewritten in the matrix form.

⎡ ⎤ ⎡ ⎤ ⎡ ⎤
yR2 1 l2−1 0
⎢ ϕR2 ⎥ ⎢ 0 1 ⎥ ⎢ 0 ⎥
⎢ ⎥ ⎢ ⎥∙ ¸ ⎢ ⎥
⎢ yR3 ⎥ ⎢ 1 l2−1 + l1 + l2 ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥ yR1 + ⎢ (−l2 )αRy ⎥ (1.179)
⎢ ϕR3 ⎥ = ⎢ 0 0 ⎥ ⎢ −αRy ⎥
⎢ ⎥ ⎢ ⎥ ϕR1 ⎢ ⎥
⎣ yR4 ⎦ ⎣ 1 l2−1 + l1 + l2 + l4−3 ⎦ ⎣ (−l2 − l4−3 )αRy ⎦
ϕR4 0 1 −αRy
Generally, the constraints imposed on shaft by rigid bodies attached to it may be
written in the following form.

yRd = adi yRi + ady (1.180)

where:
yRd - vector of the dependent coordinates
yRi - vector of the independent coordinates
adi - rectangular matrix of parameters
MATHEMATICAL MODELS OF SHAFTS 57

ady - vector of parameters.

It is easy to notice, by examination of Eq. 1.179, that the rectangular matrix
which expresses the constraints in the plane xR z is identical with that in the plane
yR z. Hence
xRd = adi xRi + adx (1.181)
Upon taking into account that according to Fig. 27,
x = xR cos Ωt − yR sin Ωt
y = yR cos Ωt + xR sin Ωt (1.182)
the constraints along the stationary system of coordinates xyz are
xd = (adi xRi + adx ) cos Ωt − (adi yRi + ady ) sin Ωt
= (adi (xi cos Ωt + yi sin Ωt) + adx ) cos Ωt − (adi (yi cos Ωt − xi sin Ωt) + ady ) sin Ωt
= adi xi + adx cos Ωt − ady sin Ωt

yd = (adi yRi + ady ) cos Ωt + (adi xRi + adx ) sin Ωt

= (adi (yi cos Ωt − xi sin Ωt) + ady ) cos Ωt + (adi (xi cos Ωt + yi sin Ωt) + adx ) sin Ωt
= adi yi + adx sin Ωt + ady cos Ωt (1.183)
The above relationships permit the dependant coordinates to be eliminated from the
mathematical model of the rotor. To do it let us assume that the mathematical model
of the rotor before the constraints are introduced can be adopted as follows
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸ ∙ ¸
m 0 ẍ c 0 ẋ k 0 x Rx Fx
+ + = + (1.184)
0 m ÿ 0 c ẏ 0 k y Ry Fy
where:
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
mdd mdi mdr cdd cdi cdr kdd kdi kdr
m = ⎣ mid mii mir ⎦ ; c = ⎣ cid cii cir ⎦ ; k = ⎣ kid kii kir ⎦
mrd mri mrr crd cri crr krd kri krr
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
xd yd Rxd Ryd
x = ⎣ xi ⎦ ; y = ⎣ yi ⎦ ; Rx = ⎣ Rxi ⎦ ; Ry = ⎣ Ryi ⎦
xR yR Rxr Ryr
⎡ ⎤ ⎡ ⎤
Fxd Fyd
⎣
Fx = Fxi ; ⎦ Fy = Fyi ⎦
⎣ (1.185)
Fxr Fyr
In the formulae above, the subscripts d, i, and r correspond to the dependant,
independent and remaining coordinates respectively. The potential energy function
for the rotor is:
⎡ ⎤⎡ ⎤
£ ¤ kdd kdi kdr xd
1
V = ( xTd xTi xTr ⎣ kid kii kir ⎦ ⎣ xi ⎦ +
2
krd kdi krr xr
⎡ ⎤⎡ ⎤
£ T ¤ kdd kdi kdr yd
+ yd yiT yrT ⎣ kid kii kir ⎦ ⎣ yi ⎦) (1.186)
krd kdi krr yr
MATHEMATICAL MODELS OF SHAFTS 58

Introduction of Eq. 1.183 into 1.186 allows to eliminate the dependent coordinates
from the expression of the potential energy function.
∙ ¸∙ ¸
1
£ ¤
aTdi kdd adi + aTdi kdi + kid adi + kii , aTdi kdr + kir xi
V = 2
( xTi xTr +
∙ T krd adi + kri , krr ¸ ∙xr ¸
£ ¤ adi kdd adi + aTdi kdi + kid adi + kii , aTdi kdr + kir yi
+ yiT yrT )+
krd adi +kri , krr yr
+(xTi (aTdi kdd adx +kid adx ) + yTi (aTdi kdd ady +kid ady )+
+xTr (krd adx ) + yTr (krd ady )) cos Ωt+
+(−xTi (aTdi kdd ady +kid ady ) + yTi (aTdi kdd adx +kid adx )+
−xTr (krd ady ) + yTr (krd adx )) sin Ωt
(1.187)
Similarly, the kinetic energy of the not constrained rotor is
⎡ ⎤⎡ ⎤
m m m ẋ
1 £ T ¤ dd di dr d
E = ( ẋd ẋTi ẋTr ⎣ mid mii mir ⎦ ⎣ ẋi ⎦ + (1.188)
2
mrd mdi mrr ẋr
⎡ ⎤⎡ ⎤
£ T ¤ mdd mdi mdr ẏd
+ ẏd ẏiT ẏrT ⎣ mid mii mir ⎦ ⎣ ẏi ⎦) (1.189)
mrd mdi mrr ẏr
Differentiation of the constraints 1.183 yields
ẋd = adi ẋi − adx Ω sin Ωt − ady Ω cos Ωt
ẏd = adi ẏi + adx Ω cos Ωt − ady Ω sin Ωt (1.190)
Introduction of the above constraints into the expression for the kinetic energy func-
tion results in the following kinetic energy function of the constrained rotor.
∙ ¸∙ ¸
1
£ ¤
aTdi mdd adi +aTdi mdi +mid adi +mii aTdi mdr +mir ẋi
E= 2
( ẋTi ẋTr +
∙ T mrd adi +mri mrr ¸ ∙ẋr ¸
£ ¤ adi mdd adi +aTdi mdi +mid adi +mii aTdi mdr +mir ẏi
+ ẏiT ẏrT )+
mrd adi +mri mrr ẏr
−(ẋTi (aTdi mdd adx +mid adx ) + ẏTi (aTdi mdd ady +mid ady )+
+ẋTr (mrd adx ) + ẏTr (mrd ady ))Ω sin Ωt+
+(−ẋTi (aTdi mdd ady +mid ady ) + ẏTi (aTdi mdd adx +mid adx )+
−ẋTr (mrd ady ) + ẏTr (mrd adx ))Ω cos Ωt
(1.191)
Introduction of the constraints 1.190 into the dissipation energy function
⎡ ⎤⎡ ⎤
£ ¤ cdd cdi cdr ẋd
1 ⎣
D = T T T
( ẋd ẋi ẋr cid cii cir ⎦ ⎣ ẋi ⎦ +
2
crd cdi crr ẋr
⎡ ⎤⎡ ⎤
£ T ¤ cdd cdi cdr ẏd
T
+ ẏd ẏi ẏr T ⎣ cid cii cir ⎦ ⎣ ẏi ⎦ (1.192)
crd cdi crr ẏr
MATHEMATICAL MODELS OF SHAFTS 59

yields:

∙ ¸∙ ¸
1
£ ¤
aTdi cdd adi +aTdi cdi +cid adi +cii , aTdi cdr +cir ẋi
D= 2
( ẋTi ẋTr +
∙ T crd adi +cri , crr ¸ ∙ẋr ¸
£ ¤ adi cdd adi +aTdi cdi +cid adi +cii , aTdi cdr +cir ẏi
+ ẏiT ẏrT )+
crd adi +cri crr ẏr
−(ẋTi (aTdi cdd adx +cid adx ) + ẏiT (aTdi cdd ady +cid ady )+
+ẋTr (crd adx ) + ẏTr (crd ady ))Ω sin Ωt+
+(−ẋTi (aTdi cdd ady +cid ady ) + ẏiT (aTdi cdd adx +cid adx )+
−ẋTr (crd ady ) + ẏTr (crd adx ))Ω cos Ωt
(1.193)
Introduction of the constraints 1.183 into the virtual work expression
⎡ ⎤ ⎡ ⎤
£ ¤ F xd £ ¤ F yd
δW = δxTd δxTi δxTr ⎣ Fxi ⎦ + δydT δyiT δyrT ⎣ Fyi ⎦ (1.194)
Fxr Fyr
yields the virtual work done by the external forces of the constrained system along
the coordinates xi , xR , yi and yR .
⎡ ⎤ ⎡ ⎤
£ T T ¤ Fxd £ ¤ F yd
δW = δxi adi δxTi δxTr ⎣ Fxi ⎦ + δyiT aTdi δyiT δyrT ⎣ Fyi ⎦
Fxr Fyr
∙ T ¸ ∙ T ¸
£ T ¤ adi Fxd + Fxi £ ¤ adi Fyd + Fyi
= xi xTr + yiT yrT (1.195)
Fxr Fyr
Introduction of Eq’s. 1.187, 1.191, 1.193, and 1.195 into Lagrange equations of motion
µ ¶
d ∂E ∂E ∂D ∂V £ ¤
− + + = Qn qn ⊂ xTi , xTr , yiT , yrT (1.196)
dt ∂ q̇n ∂ qn ∂ q̇n ∂ qn
results in the following equations of motion of the constrained rotor
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
mc 0 ẍc cc 0 ẋc kc 0 xc
+ +
0 mc ÿc 0 cc ẏc 0 kc yc
∙ ¸ ∙ ¸ ∙ c ¸ ∙ s ¸
Rx Fcx Fx Fx
= + + c cos Ωt + sin Ωt (1.197)
Ry Fcy Fy Fsy
where
∙ ¸
aTdi mdd adi +aTdi mdi +mid adi +mii, aTdi mdr +mir
mc =
mrd adi +mri mrr
∙ ¸
aTdi cdd adi +aTdi cdi +cid adi +cii , aTdi cdr +cir
cc = (1.198)
crd adi +cri crr
∙ ¸
aTdi kdd adi +aTdi kdi +kid adi +kii , aTdi kdr +kir
kc =
krd adi +kri krr
MATHEMATICAL MODELS OF SHAFTS 60
∙ ¸ ∙ ¸
xi yi
xc = ; yc = ;
xr yr
∙ ¸ ∙ ¸ (1.199)
aTdi Fxd +Fxi aTdi Fyd +Fyi
Fcx = ; Fcy =
Fxr Fyr
∙ ¸ ∙ T ¸
aTdi mdd adx +mid adx adi cdd ady +mid ady
Fcx = (−Ω2 ) + (−Ω )+
mrd adx crd ad∙y ¸
aTdi kdd adx +kid adx
+(+ )
krd adx
∙ ¸ ∙ T ¸
aTdi mdd ady +mid ady adi cdd adx +mid adx
Fcy = (−Ω 2
) + (+Ω )+
mrd ady crd ad∙x ¸
aTdi kdd ady +kid ady
+(+ )
krd ady
∙ ¸ ∙ T ¸
aTdi mdd ady +mid ady adi cdd adx +mid adx
Fsx = (+Ω 2
) + (−Ω )+
mrd ady crd a∙dx ¸
aTdi kdd ady +kid ady
+(− )
krd ady
∙ ¸ ∙ T ¸
aTdi mdd adx +mid adx a cdd ad +m id a d
Fsy = (−Ω2 ) + (−Ω di y y
)+
mrd adx crd ad∙y ¸
aTdi kdd adx +kid adx
+(+ )
krd adx
(1.200)
The expressions 1.200 represents excitation forces due to the inaccuracy of the rigid
element ( adx , ady ).
For Ω = 0 the above formulae yield mathematical model of the constrained
and not rotating element.
MATHEMATICAL MODELS OF SHAFTS 61

1.2.7 Problems
Problem 4

1 2 3 4 5
z
a1=0 a2 a3 a4 a5=0 a)

Z
b1=0 b2 b4=0

b)
X

Figure 28

The dynamic properties of the free-free beam are determined by the following
mathematical model
m ẍ + k x = 0 (1.201)
where:
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
m11 m12 m13 m14 m15 k11 k12 x1
k13 k14 k15
⎢ m21 m22 m23 m24⎥ m25 ⎢ k21 ⎥k22 ⎢ x2
k23 k24 k25 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
m=⎢ ⎢ m31 m32 m33 m34⎥, k = ⎢
⎥ m35 ⎢ k31 ⎥ , x = ⎢ x3
⎥k32 ⎢k33 k34 k35 ⎥
⎥
⎣ m41 m42 m43 m44⎦ m45 ⎣ k41 ⎦k42 ⎣ x4
k43 k44 k45 ⎦
m51 m52 m53 m54 m55 k51 k52 k53
x5 k54 k55
(1.202)
The beam is permanently deflected in plane xz as shown in Fig. 33a) The permanent
deformation is determined by the vector a.

a = [a1 , a2 , a3 , a4 , a5 ]T (1.203)

The beam is supported rigidly upon three supports. Configuration of these supports
is shown in Fig. 33b).
£ ¤T
b = b1 b2 b4 (1.204)
Calculate the static deflection line of the supported beam and its natural frequencies.
MATHEMATICAL MODELS OF SHAFTS 62

Solution
Since the equivalent to the permanent deformation set of forces is

Fa = ka (1.205)

the equation of the free-free and deflected beam is

m ẍ + k x = Fa (1.206)

Reorganization of this mathematical model with respect to the vector b yields

⎡ ⎤⎡ ⎤ ⎡ ⎤⎡ ⎤
m11 m12 m14 m13 m15 ẍ1 k11 k12 k14 k13 k15 x1
⎢ m21 m22 m24 m23 m25 ⎥ ⎢ ẍ2 ⎥ ⎢ k21 k22 k24 k23 k25 ⎥ ⎢ x2 ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥
⎢ m41 m42 m44 m43 m45 ⎥ ⎢ ẍ4 ⎥ + ⎢ k41 k42 k44 k43 k45 ⎥ ⎢ ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ x4 ⎥
⎣ m31 m32 m34 m33 m35 ⎦ ⎣ ẍ3 ⎦ ⎣ k31 k32 k34 k33 k35 ⎦ ⎣ x3 ⎦
m51 m52 m54 m53 m55 ẍ5 k51 k52 k54 k53 k55 x5
⎡ ⎤⎡ ⎤ ⎡ ⎤
k11 k12 k14 k13 k15 a1 R1
⎢ k21 k22 k24 k23 k25 ⎥ ⎢ a2 ⎥ ⎢ R2 ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥
= ⎢⎢ k41 k42 k44 k43 k45
⎥ ⎢ a4 ⎥ + ⎢ R4 ⎥
⎥⎢ ⎥ ⎢ ⎥ (1.207)
⎣ k31 k32 k34 k33 k35 ⎦ ⎣ a3 ⎦ ⎣ 0 ⎦
k51 k52 k54 k53 k55 a5 0
or
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸
mbb mbr ẍb kbb kbr xb kbb kbr ab Rb
+ = + (1.208)
mrb mrr ẍr krb krr xr krb krr ar 0

where: ⎡ ⎤ ⎡ ⎤
m11 m12 m14 m13 m15
mbb = ⎣ m21 m22 m24 ⎦ , mbr = ⎣ m23 m25 ⎦ ,
m41 m42 m44 m43 m45
∙ ¸ ∙ ¸
m31 m32 m34 m33 m35
mrb = , mrr = ,
m51 m52 m54 m53 m55
⎡ ⎤ ⎡ ⎤
k11 k12 k14 k13 k15
kbb = ⎣ k21 k22 k24 ⎦ , kbr = ⎣ k23 k25 ⎦ ,
k41 k42 k44 k43 k45
∙ ¸ ∙ ¸
k31 k32 k34 k33 k35
krb = , krr = ,
k51 k52 k54 k53 k55
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x1 ∙ ¸ a1 ∙ ¸ R1
x3 a3
xb = ⎣ x2 ⎦ , xr = , ab = ⎣ a2 ⎦ , ar = , Rb = ⎣ R2 ⎦
x5 a5
x4 a4 R4
(1.209)
Since
xb = b (1.210)
MATHEMATICAL MODELS OF SHAFTS 63

the above equations after partitioning takes the following form

mbb 0 + mbr ẍr +kbb b + kbr xr = kbb ab +kbr ar +Rb (1.211)

mrb 0 + mrr ẍr +krb b + krr xr = krb ab +krr ar (1.212)

Now, the equation 1.212 can be rewritten in form

mrr ẍr +krr xr = krb ab +krr ar −krb b (1.213)

The equation 1.213 governs the free motion of the beam. Its particular solution which
represents the deflection line of the beam can be predicted as a constant vector Xr

xr = Xr (1.214)

Introduction of Eq. 1.214 into 1.213 yields

Xr = k−1
rr (krb ab +krr ar −krb b) (1.215)

The static reactions at the supports can be computed from equation 1.211.

Rb = kbb b + kbr Xr −kbb ab −kbr ar (1.216)

Solution of the homogeneous equation 1.213

mrr ẍr +krr xr = 0 (1.217)

offers the characteristic equation for the natural frequencies of the beam ω.
¯ −1 ¯
¯mrr krr − 1ω 2 ¯ = 0 (1.218)
MATHEMATICAL MODELS OF SHAFTS 64

Problem 5

The mathematical model of a free-free shaft, shown in Fig. 29, along coordi-
nates x1 , x2 , x3 , x4 is as follows:

mẍ + kx = 0 (1.219)
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
m11 m12 m13 m14 k11 k12 k13 m14 x1
⎢ m21 m22 m23 m24 ⎥ ⎢ k21 k22 k23 k24 ⎥ ⎢ x2 ⎥
m =⎢
⎣ m31
⎥; k =⎢
⎦ ⎣ k31 k32
⎥; x =⎢ ⎥
m32 m33 m34 k33 k34 ⎦ ⎣ x3 ⎦
m41 m42 m43 m44 k41 k42 k43 k44 x4
(1.220)

1 2 3 4

x1 x2 x3 x4

Figure 29

This shaft is supported upon three rigid pedestals along coordinates x1 , x2 , x3

as shown in Fig. 30.

1 2 3
a3 Z

X a2 sin ω t

Figure 30

Their motion with respect to the inertial system of coordinate system is given
by the following equations:

X1 = 0
X2 = a2 sin ωt
X3 = a3 (1.221)

Derive expressions for :

1. the static deflection curve,
2. the interaction forces between the shaft and the pedestals with respect to
the inertial system of coordinates
MATHEMATICAL MODELS OF SHAFTS 65

Solution:
Partitioning of the equations 1.219 with respect to the vector of boundary
conditions 1.221 results in the following

mẍ + kx = R (1.222)
∙ ¸∙ ¸ ∙ ¸ ∙ ¸ ∙ ¸
m11 m12 ẍ1 k11 k12 x1 R1
+ = (1.223)
m21 m22 ẍ2 k21 k22 x2 R2
⎡ ⎤ ⎡ ⎤
m11 m12 m13 m14 £ ¤
m11 = ⎣ m21 m22 m23 ⎦ ; m12 = ⎣ m24 ⎦ ; m21 = m41 m42 m43 ; m22 = m44
m31 m32 m33 m34
⎡ ⎤ ⎡ ⎤ (1.224)
k11 k12 k13 k14 £ ¤
⎣
k11 = k21 k22 k23 ⎦ ; k12 = ⎣ k24 ⎦ ; k21 = k41 k42 k43 ; k22 = k44
k31 k32 k33 k34
⎡ ⎤ ⎡ ⎤ (1.225)
x1 R1
x1 = ⎣ x2 ⎦ ; x2 =x4 ; R1 = ⎣ R2 ⎦ ; R2 =0 (1.226)
x3 R3
or
m11 ẍ1 + m12 ẍ2 + k11 x1 + k12 x2 = R1 (1.227)
m21 ẍ1 + m22 ẍ2 + k21 x1 + k22 x2 = 0 (1.228)
Introduction of boundary conditions 1.221 into the equation 1.228 yields
⎡ ⎤ ⎡ ⎤
£ ¤ ẍ1 £ ¤ x1
m44 ẍ4 + k44 x4 = − m41 m42 m43 ⎣ ẍ2 ⎦ − k41 k42 k43 ⎣ x2 ⎦ (1.229)
ẍ3 x3

where
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
ẍ1 0 x1 0
⎣ ẍ2 ⎦ = ⎣ −a2 ω2 sin ωt ⎦ ; ⎣ x2 ⎦ = ⎣ a2 sin ωt ⎦ (1.230)
ẍ3 0 x3 a3
or
m44 ẍ4 + k44 x4 = (m42 a2 ω 2 − k42 a2 ) sin ωt − k43 a3 (1.231)
The static deflection is due to the time independent term −k43 a3 in the right hand
side of equation 1.231.
m44 ẍ4 + k44 x4 = −k43 a3 (1.232)
The particular solution of 1.232 is

x4 = xs (1.233)

k44 xs = −k43 a3 (1.234)

−k43 a3
xs = (1.235)
k44
MATHEMATICAL MODELS OF SHAFTS 66

Its graphical representation is given in Fig. 31

1 2 3 4
a3 Z

X xs

Figure 31

The forced response due to the motion of the support 2 (X2 = a2 sin ωt) is
represented by the particular solution due to the time dependant term.

m44 ẍ4 + k44 x4 = (m42 a2 ω 2 − k42 a2 ) sin ωt (1.236)

For the above equation, the particular solution may be predicted as follows

x4 = xd sin ωt (1.237)

Implementation of the solution 1.237 into the equation 1.236 yields the wanted am-
plitude of the forced vibration xd .

(m42 a2 ω 2 − k42 a2 )
xd = (1.238)
−ω 2 m44 + k44
The resultant motion of the system considered is shown in Fig. 32

1 2 3 4
a3 xs Z
X a2 sinω t xd sinω t

Figure 32

This motion causes interaction forces along these coordinates the system is
attached to the base. These forces can be computed from equation 1.227.

m11 ẍ1 + m12 ẍ2 + k11 x1 + k12 x2 = R1 (1.239)

In this equation x1 stands for the given boundary conditions

⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x1 0 ẍ1 0
x1 = ⎣ x2 ⎦ = ⎣ a2 sin ωt ⎦ ; ẍ1= ⎣ ẍ2 ⎦ = ⎣ −a2 ω 2 sin ωt ⎦ (1.240)
x3 a3 ẍ3 0
MATHEMATICAL MODELS OF SHAFTS 67

and x2 represents, known at this stage, motion of the system along the coordinate 4

d2
x2 = x4 = xs + xd sin ωt ẍ2 = ẍ4 = (xs + xd sin ωt) = −xd ω 2 sin ωt; (1.241)
dt2
Hence, the wanted vector of interaction forces is as follows

⎡ ⎤ ⎡ ⎤
0 0
R1 = m11 ⎣ −a2 ω 2 sin ωt ⎦+k11 ⎣ a2 sin ωt ⎦+m12 (−xd ω2 sin ω)+k12 (xs +xd sin ωt)
0 a3
(1.242)
MATHEMATICAL MODELS OF SHAFTS 68

Problem 6

The dynamic properties of the free-free shaft are determined by the following
mathematical model
m ẍ + k x = 0 (1.243)
where:
⎡ ⎤ ⎡ ⎤ ⎤ ⎡
m11 m12 m13 m14 k11 k12 k13 xR1
k14
⎢ m21 m22 m23 m24 ⎥ ⎢ k21 k22 k23 ⎢ xR2 ⎥
k24 ⎥
m=⎢
⎣ m31
⎥, k=⎢ x=⎢ ⎥, ⎥
m32 m33 m34 ⎦ ⎣ k31 k32 k33 ⎣ xR3 ⎦
k34 ⎦
m41 m42 m43 m44 k41 k42 k43 k44
xR4
(1.244)
The shaft is permanently deflected in one plane as shown in Fig. 33. The permanent
deformation is determined by the vector a.

a = [a1 , a2 , a3 , a4 ]T (1.245)

1 2 3 4
z
a1 a2 a3 a4

yR

Figure 33

The shaft is rigidly supported at node 1 and 2 and rotates with the angular
velocity Ω. Calculate response of this shaft due to the permanent deformation and
the dynamic reactions at the support 1 and 2.
MATHEMATICAL MODELS OF SHAFTS 69

Solution.

O x
Ωt
Fy
R
xR

yR y

Figure 34

Let us introduce the rotating system of coordinates xR yR zR and the stationary

system of coordinates xyz as it is shown in Fig. 34. Assume that the permanent
deformation occurs in the plane yR z. Hence, the vector of equivalent forces is
FyR = ka (1.246)
Its components along the stationary system of coordinates are
Fx = −FyR sin Ωt = −ka sin Ωt (1.247)
Fy = FyR cos Ωt = ka cos Ωt (1.248)
Therefore, the equation of motion of the shaft in the plane xz of the stationary system
of coordinates is
m ẍ + k x = Rx +Fx (1.249)
The shaft is supported at node 1 and 2. Hence,
£ ¤T
Rx = R1x R2x 0 0 (1.250)
After partition of the equation 1.249 with respect to the supported coordinates, equa-
tions of the supported shaft are
mbb ẍb +mbr ẍr +kbb xb +kbr xr = Rxb − (kbb ab +kbr ar ) sin Ωt (1.251)
mrb ẍb +mrr ẍr +krb xb +krr xr = −(krb ab +krr ar ) sin Ωt (1.252)
where:
∙ ¸ ∙ ¸
m11 m12 m13 m14
mbb = , mbr = ,
m21 m22 m23 m24
∙ ¸ ∙ ¸
m31 m32 m33 m34
mrb = , mbr = , (1.253)
m41 m42 m43 m44
∙ ¸ ∙ ¸
k11 k12 k13 k14
kbb = , kbr = ,
k21 k22 k23 k24
∙ ¸ ∙ ¸
k31 k32 k33 k34
krb = , kbr = , (1.254)
k41 k42 k43 k44
MATHEMATICAL MODELS OF SHAFTS 70
∙ ¸ ∙ ¸ ∙ ¸ ∙ ¸ ∙ ¸
x1 0 x3 a1 a3
xb = = , xr = , ab = , ar = (1.255)
x2 0 x4 a2 a4
Introduction of Eq. 1.255 into 1.251 and 1.252results in their simplification.

mbr ẍr +kbr xr = Rxb − (kbb ab +kbr ar ) sin Ωt (1.256)

mrr ẍr +krr xr = −(krb ab +krr ar ) sin Ωt (1.257)

Similarly, the equations of motion in the plane yz are

mbr ÿr +kbr yr = Ryb + (kbb ab +kbr ar ) cos Ωt (1.258)

mrr ÿr +krr yr = (krb ab +krr ar ) cos Ωt (1.259)

The particular solution of Eq. 1.257

xr = Xr sin Ωt (1.260)

represents the forced vibrations in plane xz due to the permanent deformation a.

Introduction of Eq. 1.260 into 1.257 yields

(−Ω2 mrr + krr )Xr = −(krb ab + krr ar ) (1.261)

Hence,
Xr = −(−Ω2 mrr +krr )−1 (krb ab +krr ar ) (1.262)
Similarly, the forced vibrations in plane yz are represented by the particular solution
of Eq. 1.259
yr = Yr cos Ωt (1.263)
where
Yr = (−Ω2 mrr +krr )−1 (krb ab +krr ar ) (1.264)
The equations 1.256 and 1.258 allow the dynamic reactions at the supports 1 and 2
to be computed.

Rxb = −mbr Xr Ω2 sin Ωt + kbr Xr sin Ωt + (kbb ab +kbr ar ) sin Ωt (1.265)

Ryb = −mbr Yr Ω2 cos Ωt + kbr Yr cos Ωt − (kbb ab +kbr ar ) cos Ωt (1.266)
MATHEMATICAL MODELS OF BEARINGS 71

1.3 MATHEMATICAL MODELS OF BEARINGS

1.3.1 Oil journal bearings

6 4 1 3 2

5 3 5

Figure 35 Cross-section of the journal bearing: 1 - housing, 2 - journal, 3 - sleeves, 4 -

inlet, 5 - outlet, 6 - seals

Figure 35 shows the cross-section of an oil journal bearing. Geometry of the commonly
used sleeves 3 are presented in Figure 36.
MATHEMATICAL MODELS OF BEARINGS 72

Fo Fo
Ω Ω
20 o
R R
r r

a) b)
R+r
Fo R+r Fo
Ω
Ω
R+r
ex R+r
R+r

c) d)

Fo R+r Fo
Ω R+r Ω
R+r
R+r
R+r

e) f)

Figure 36 Types of journal bearings: a) partial arc, b) two-axial groove, c) & d)

excentrical, e) offset cylindrical, f) tilting pad

In this figure R stands for radius of the journal, r stands for the radial clear-
ance, Fo for the static load and Ω represents the angular speed of the rotor.
The dynamic properties of the oil bearing (Fig. 37a)) are determined by the
following parameters:
R - radius of sleeve
Rj - radius of journal
L - length of the bearing
Fo - static load
Ω - angular velocity of the journal
η - viscosity of the oil film
MATHEMATICAL MODELS OF BEARINGS 73

r = R − Rj - radial clearance
These parameters determine the Sommerfeld number (non-dimensional load)
S
Fo r2
S= (1.267)
2R3 LηΩ

bearing
R
sleeve
Ω Ω
journal
Rj e
Os x Os
ϕ

y x
Oj x oil film Oj
α h
Fo

p
y
a) b) H

Figure 37

The pressure distribution within the oil film p (see Fig. 37a)) depends on the
instantaneous position (x, y) and the instantaneous velocity (ẋ, ẏ) of the journal with
respect to the sleeves. Taking into account that according to Fig. 37b)

x = −e sin α
y = e cos α (1.268)

the relative position of journal with respect to sleeve can be alternatively determined
by two coordinates e and α as shown in Fig. 37b). For the assumed relative position
and velocity, the Reynolds equation 1.269, which governs flow of the oil between
journal and sleeve, yields the pressure distribution p.
µ ¶ µ ¶
1 ∂ h3 ∂p ∂ h3 ∂p ∂h ∂h
2
+ = 6Ω + 12 (1.269)
R ∂ϕ η ∂ϕ ∂z η ∂z ∂ϕ ∂t
In the above equation h is the thickness of the oil film. For a small clearance r it is
as follows

h = r + e cos(ϕ − α) = r + e cos α cos ϕ + e sin α sin ϕ = r − x sin ϕ + y cos ϕ (1.270)

Hence
∂h
= −x cos ϕ − y sin ϕ (1.271)
∂ϕ
MATHEMATICAL MODELS OF BEARINGS 74

And eventually
∂h
= −ẋ cos ϕ − ẏ sin ϕ (1.272)
∂t
If one assume that the viscosity of the oil η is independent of the variable ϕ (the
influence of the temperature distribution inside the oil film is neglected) and the
thickness of the oil film is independent of z, the equation 1.269 can be rewritten as
follows µµ ¶ ¶ µ ¶
2
1 ∂ 3 ∂p 3∂ p h3 ∂ 2 p ∂h ∂h
2
h +h 2
+ 2
= 6Ω + 12
Rη ∂ϕ ∂ϕ ∂ϕ η ∂z ∂ϕ ∂t
or
µ ¶ µ 2 ¶ µ ¶µ ¶
h3 ∂ 2 p h3 ∂ p 1 ∂ 3 ∂p ∂h ∂h
+ + h = 6Ω + 12 (1.273)
η ∂z 2 R2 η ∂ϕ2 R2 η ∂ϕ ∂ϕ ∂ϕ ∂t

where according to 1.270

∂ 3
h = 3 (r − x sin ϕ + y cos ϕ)2 (−x cos ϕ − y sin ϕ) (1.274)
∂ϕ
The equation 1.274 should be integrated over the bearing surface formed by
the two variable z and ϕ. This surface is shown in Fig.38a).
MATHEMATICAL MODELS OF BEARINGS 75

z F H
C
D
L/2

0 a)
ϕ
L/2

A B
E 360 deg G
p

b)
pin

ϕ
p

pin c)

ϕ
pcav

Figure 38

The edges AB, BC, CD and DA, form the external boundary condition. If
oil is released to atmosphere the pressure along the edge AB and DC can be assume
equal to the atmospheric pressure. The pressure along the edge AD and BC is equal
to the inlet pressure pin .
The process of integration of the equation 1.269 starts from the boundary
conditions associated with the edge AD (p = pin ) and an arbitrarily chosen pressure
gradient (see Fig. 38a)). It is progressing towards the edge BC (see Fig. 38b). If at
the edge BC the boundary conditions are met (p = pin ), the process of integration
is finished. If the boundary conditions are not met (p 6= pin ) the integration process
starts from the beginning with a different pressure gradient. To include the cavitation,
the oil film pressure is controlled at each step of integration. If the current pressure
drops below the level of cavitation (see Fig. 38c), the cavitation pressure pcav replaces
it and the process of integration is continued till the other external boundary BC
is reached. The last integration yields the internal boundary EF and GH (see Fig.
38a)) for the Reynolds equation 1.269.
Very often the axis ϕ (see Fig. 38a)) is the axis of symmetry and the process of
integration can be limited to the positive quarter of the plane ϕz. A typical solution
MATHEMATICAL MODELS OF BEARINGS 76

of the Reynolds equation over this positive quarter of the plane ϕz is shown in Fig.
39

0
z
ϕ

Figure 39

The following integrals produces the components of the hydrodynamic force

H that acts on the journal.
+L ⎛ ⎞
Z 2 Z2π
Hx = − ⎝ p(ϕ, z) sin ϕdϕ⎠ dz
−L 0
2

+L ⎛ 2π ⎞
Z2 Z
Hy = + ⎝ p(ϕ, z) cos ϕdϕ⎠ dz (1.275)
−L 0
2

The following results of the described above computation are corresponding

to the three sleeves concentric journal bearing shown in Fig.40
ϕon

ϕn

Figure 40 ϕ01 =18.50 ,ϕ1 =470 , ϕ02 =97.70 ,ϕ2 =1500 , ϕ03 =282.50 ,ϕ1 =65.50 , L/D = 0.8
MATHEMATICAL MODELS OF BEARINGS 77

Equilibrium position - Static characteristic of oil bearings

To find the equilibrium position of the journal for a given load Fo and an angular
velocity Ω, one is supposed to assumed it arbitrarily as (x, y) and ẋ = ẏ = 0. If the
computed pressure distribution does not produces the hydrodynamic force H equal
to the static load Fo , the assumed position does not corresponds to the equilibrium
position. The process should be repeated for differently assumed equilibrium position
(x, y and ẋ = ẏ = 0). Any procedure that minimize the objective function OF =
|Fo − H| leads quickly to the wanted equilibrium position xo , yo . Result of such a
computations is usually presented as a function of the non-dimensional load S.
Locus of equilibrium positions for different magnitudes of the non-dimensional
load S is given in Fig. 41b). In this diagram c corresponds to so called relative
eccentricity
e
c= (1.276)
r
In this diagram xo and y o are standing for the non-dimensional equilibrium position

xo = xo r and yo = y o r (1.277)

Fig. 41a) gives the relationship between the non-dimensional load S and the relative
eccentricity c.
Hr2
S= (1.278)
2R3 LηΩ
Those two diagrams, which can be found in engineering literature for different design
of the oil bearing, determine uniquely the equilibrium position (xo , yo ) of journal with
respect to sleeve (follow the arrows). They are called the static characteristic of oil
bearing.
The pressure distribution in the central cross-section (z = 0) is presented in
Fig. 42 and 43. It corresponds to the equilibrium position e = 0.93 and e = 0.586
computed for load 200N and the angular velocities 30 and 300rad/s respectively
MATHEMATICAL MODELS OF BEARINGS 78

10

6
S 5
a)
4

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 c
-1

-0.8

-0.6

b) -0.4

-0.2
_
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
xo

0.2

0.4

0.6

0.8

_1
yo

Figure 41
MATHEMATICAL MODELS OF BEARINGS 79

400000

350000

300000

250000

200000
2
p [N/m]
150000

100000

50000

0 50 100 150 200 250 300 350

-50000
ϕ [deg]

Figure 42

200000

150000

100000
p [N/m2
50000

0
0 50 100 150 200 250 300 350
-50000
ϕ [deg]

Figure 43
MATHEMATICAL MODELS OF BEARINGS 80

Linearized mathematical model - dynamic characteristic of the oil bearing

The solution of the Reynolds equation allows the hydrodynamic forces Hx and Hy
to be computed for any instantaneous position and velocity of the journal bearing
x, y, ẋ, ẏ. Let us utilize this equation to produce the linearized equations of motion of
the journal in vicinity of the equilibrium position xo , yo (see Fig.44).

Ob xb
Oj
yj ∆ Hx xj
xo xj
yo
∆ Hy

yj yb

Figure 44

If journal performs small motion xj , yj in vicinity of its equilibrium position

Oj , the increments of hydrodynamic forces ∆Hx and ∆Hy acting on the oil film can
be approximated by the following linear function
∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
∆Hx kxx kxy xj cxx cxy ẋj
= + (1.279)
∆Hy kyx kyy yj cyx cyy ẏj
The stiffness coefficients kxx , kxy , kyx , kyy and the damping coefficients cxx , cxy , cyx , cyy
can be computed with help of the Reynolds equations according to the expressions
1.280 and 1.281
Hx (xo + ∆x, yo , 0, 0) − Hx (xo , yo , 0, 0)
kxx =
∆x
Hx (xo , yo + ∆y, 0, 0) − Hx (xo , yo , 0, 0)
kxy =
∆y
Hy (xo + ∆x, y, 0, 0) − Hx (xo , yo , 0, 0)
kyx = (1.280)
∆x
Hy (xo , yo + ∆y, 0, 0) − Hx (xo , yo , 0, 0)
kyy =
∆y
MATHEMATICAL MODELS OF BEARINGS 81

Hx (xo , yo , ẋ, 0) − Hx (xo , yo , 0, 0)

cxx =
ẋ
Hx (xo , yo , 0, ẏ) − Hx (xo , yo , 0, 0)
cxy =
ẏ
Hy (xo , yo , ẋ, 0) − Hy (xo , yo , 0, 0)
cyx = (1.281)
ẋ
Hy (xo , yo , 0, ẏ) − Hy (xo , yo , 0, 0)
cyy =
ẏ
Their non-dimensional magnitudes, denoted here with dash,
Sr SrΩ
k̄ij = kij c̄ij = cij (1.282)
Fo Fo
are usually plotted as a function of the relative eccentricity c. For the bearing shown
in Fig. 40 the non-dimensional stiffness and damping coefficients are given in Fig.
45.
MATHEMATICAL MODELS OF BEARINGS 82

10
9
8
7
6
S 5
4
a)
3
2
1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 c
30
25 yy
20
15
_ xx
10
yx xy
k 5 b)
0
-5
-10
-15
-20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 c
35
30 yy
25
_ 20
c 15
10
yx xx c)
5 xy
0
-5
-10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 c

Figure 45

The diagrams presented in Fig. 45 are called the dynamic characteristic of

journal bearing. They allow the stiffness and damping coefficients to be determined.
The equation 1.279can be presented in the following matrix form.

∆H = kf xj + cf ẋj (1.283)
MATHEMATICAL MODELS OF BEARINGS 83

Ω
xb
Ob yb xb
Oj
yj ∆ Hx xj
xo xj
yo
∆ Hy

yj yb

Figure 46

If both, the journal and sleeve are moving in vicinity of their equilibrium
positions Oj and Ob respectively as is shown in Fig. 46, the relationship between the
hydrodynamic forces ∆Hx , ∆Hy and displacements xj , yj , xb , yb is as follows:

∆H = kf (xj −xb ) + cf (ẋj −ẋb ) (1.284)

Since the oil film is considered massless, the interaction force between the oil film and
the sleeve is equal −∆H.

−∆H = −kf (xj −xb ) − cf (ẋj −ẋb ) (1.285)

The two last equation can be rewritten in form of one matrix equation
∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
∆H kf −kf xj cf −cf ẋj
= + (1.286)
−∆H −kf kf xb −cf cf ẋb
The vector of hydrodynamic forces plays role of the vector of interaction forces be-
tween the oil film and the journal and sleeve R. Therefore, the final mathematical
model of the oil film can be adopted in form 1.287

kx + cẋ = R (1.287)

where: ∙ ¸ ∙ ¸
kf −kf cf −cf
k= , c=
−kf kf −cf cf
∙ ¸ ∙ ¸
xj ẋj
x= , ẋ = (1.288)
xb ẋb
It should be noticed that in order to formulate the mathematical model of the oil film
(Eq. 1.287), the static load Fo has to be determined earlier .
Exactly the same form has the mathematical model of elastic elements (e.g.
rubber blocks, shock absorbers, squeeze film dampers etc.) that can be considered as
massless.
MATHEMATICAL MODELS OF BEARINGS 84

1.3.2 Magnetic bearings

1 2 3

x x

Figure 47

i yo
i xo + + x

ix iy y x y

.
POWER Vx (x,x) BEARING
AMPLIFIER . CONTROLLER
Vy (y,y)

Figure 48

The magnetic bearing (see Fig. 47) levitates the journal 4 of the rotor by attraction
created by the electromagnet 1. The core of the poles 1 are made of the permanent
magnets. This attraction force makes the bearing equilibrium position unstable and
,eventually, pull the journal to contact with the electromagnet. For this reason, the
magnetic bearing must be furnished with system that controls the designed equilib-
rium position of the journal. An example of such a control system is shown in Fig.
48. If the sensor 2 (Fig. 47) detects the instantaneous position of the journal 4 to be
to the right of the required equilibrium position, the bearing controller shown in Fig.
48 makes the attraction force in the right hand side electromagnet smaller and in the
MATHEMATICAL MODELS OF BEARINGS 85

left hand side one larger. This, in turn, causes the journal to move back towards the
equilibrium position.
The following parameters determine the dynamic properties of a magnetic
bearing.
Journal diameter D[m]
Radial stator-rotor air gap c = 0.001[m]
Bearing actuator force coefficient κ = 1.49 × 10−4 [N/(A − t)2 ]
Control winding turns N = 140[turn]
Permanent magnet air-gap MMF ϑ = 299[A − t]
o
Pole face angle (see Fig.47) α = 45
Max. control current ¡ diimax
¢ = 2[A]
Max. control current slew rate dt max
= 5000[A/s]
Actuator saturation current isat = 3[A]
Bias current i0 = 0[A]
The non-dimensional force F̄x that act on the journal in the x direction is the sum
of forces produced by both the left hand side electromagnet F̄x1 and the right hand
side electromagnet F̄x2 . It can be determined from the following formula∗

F̄x = F̄x1 + F̄x2 = (1 + ı̄o + ı̄)2 Ψ̄1 (x̄) + (1 − ı̄o − ı̄)2 Ψ̄2 (x̄) (1.289)

where
µ ¶ Ãr !
1 sin α 2x̄ 1 + x̄ α
Ψ̄1 (x̄) = + 3 arctan tan (1.290)
1 − x̄2 1 − x̄ cos α (1 − x̄2 ) 2 1 − x̄ 2
µ ¶ Ãr !
1 sin α 2x̄ 1 − x̄ α
Ψ̄2 (x̄) = − + 3 arctan tan (1.291)
1 − x̄2 1 + x̄ cos α (1 − x̄2 ) 2 1 + x̄ 2

Here
F̄ stands for the non-dimensional force
F
F̄ = (1.292)
κϑ2
x̄ stands for the non-dimensional displacement
x
x̄ = (1.293)
c
ı̄, ı̄o stands for the non-dimensional control and bias current respectively
Ni Nio
ı̄ = ı̄o = (1.294)
ϑ ϑ
For i equal to zero the formula 1.289 yields the relationship between the static force,
position of the journal and the bias current. This relationship is called the static
∗
Krodkiewski J., Zmood R.B., Use of programmed magnetic bearing stiffness and damping to
minimise rotor vibration, Proceedings of the Third International Symposium on magnetic bearings,
ed Allaire E., pub. Technomic Publishing Co. Inc., pp 35-44, 1992.
MATHEMATICAL MODELS OF BEARINGS 86

characteristic of the magnetic bearing. An example of such a characteristic in its

non-dimensional form is presented in Fig.49

static force (non- io=-0.9 6

dimensional
io=-0.6 5

4
io=-0.3
3

0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-1

-2

-3 io=0.0

-4 io=0.3
-5 io=0.6
-6 io=0.9
-7
displacement (non-dimensional)

Figure 49

This characteristic allows the static equilibrium position to be designed. For

example if the non-dimensional static force is equal to F̄ = 0.9 and one assume the
bias current io = 0 then the equilibrium is at x̄o = −0.3.
To maintain the bearing journal in the vicinity of the equilibrium position
x̄o it is assumed that the current ı̄ depends upon the journal displacement from its
equilibrium position ξ (see Fig.50) and its instantaneous velocity vξ . The velocity is
non-dimesionalized with help of the following definition

dξ̄
v̄ξ = t̄ = Ωt (1.295)
dt̄
Here, Ω stands for the rotating speed of the journal and t for time
MATHEMATICAL MODELS OF BEARINGS 87

x
ξ

xo x

Figure 50

The controller produces the control current ı̄ that in the general case can be
approximated by the Taylor series of the two variables ξ̄ and v̄ξl .
X
K X
L
k
ı̄(ξ̄, v̄ξ ) = ākl ξ̄ v̄ξl (1.296)
k=0 l=0
If we introduce the above relationships into 1.289 and take into account that
x̄ = x̄o + ξ̄ (1.297)
we are getting expression for the instantaneous force that acts on the journal as a
function of its parameters of motion.
X
K X
L
k
X
K X
L
k
F̄x = (1 + ākl ξ̄ v̄ξl )2 (ı̄o + ākl ξ̄ v̄ξl )Ψ̄(x̄o + ξ̄)
k=0 l=0 k=0 l=0
X
K X
L
k
X
K X
L
k
+(1 − ākl ξ̄ v̄ξl )2 (ı̄o + ākl ξ̄ v̄ξl )Ψ̄(x̄o + ξ̄) =
k=0 l=0 k=0 l=0
X
K X
L
k
= b̄kl ξ̄ v̄ξl )
k=0 l=0
(1.298)
Arranging for this force to be a linear function of ξ̄ and v̄ξl in vicinity of the equilibrium
position x̄o could be one of possible control strategy.
F̄x = k̄x ξ̄ + c̄x v̄ξ (1.299)
where
c
k̄x = kx
κϑ2
cΩ
c̄x = cx (1.300)
κϑ2
MATHEMATICAL MODELS OF BEARINGS 88

To make it happen

b10 = k̄x
b01 = c̄x
bkl = 0 for k + l > 1 (1.301)

For any equilibrium position x̄o and any arbitrarily chosen stiffness k̄x and damping
c̄x , the last set of equations permits the coefficients akl to be computed. Now, the
force F̄x is uniquely determined as a function of ξ̄ and v̄ξ or alternatively as a function
of x̄ and v̄x . This function is called dynamic characteristic.

F̄x = F̄x (x̄, v̄x ) (1.302)

An example of such a dynamic characteristic for x̄o = 0.3, k̄x = 6, and c̄x = 2 is
presented in Fig. 51. It was computed for k + l ≤ 3

2
force (non- operating clearance
dimensional)
1

0
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1
displacement (non-dimensional)
-1
vx=0.6
vx=0.4
vx=0.2
-2
vx=0
equilibrium position
vx=-0.2
-3 vx=-0.4
vx=-0.6

-4

-5

Figure 51

It can be seen that this characteristic is linear over a wide range of the bearing
gap. The region where the bearings can be consider linear is called the operating
clearance and will be denoted by co . By inspection of the diagram provided in Fig.
51 one can assess the non-dimensional operation clearance to be c̄o = 0.2. The touch
MATHEMATICAL MODELS OF BEARINGS 89

down bearing 3 (see Fig. 47) should prevent the journal to go outside this operating
clearance.
At each instant the total current in both coils has to be positive and fulfill the
following criteria

0 < ı̄min < 1 + ı̄0 + ı̄ < ı̄sat and 0 < ı̄min < 1 − ı̄0 − ı̄ < ı̄sat (1.303)

Another criterion is due to the limitation of the current slew rate

µ ¶
dı̄ dı̄
< (1.304)
dt̄ dt̄ max

The domain of parameters kx and cx that fulfill the above criteria can be
produced according to the following procedure:
1. assume k̄x and c̄x arbitrarily
2. compute the coefficients ākl from the equations 1.301
3. compute the displacement and velocity of journal from the following equa-
tion

ξ̄|t̄ = c̄o sin t̄ (1.305)

v̄ξ |t̄ = c̄o cos t̄ (1.306)

where c̄o is the non-dimensional operating clearance

4. compute the displacement and velocity of journal for t̄+dt̄ from the following
equation

ξ̄|t̄+dt̄ = c̄o sin(t̄ + dt̄) (1.307)

v̄ξ |t̄+dt̄ = c̄o cos (t̄ + dt̄) (1.308)

5. with help of the formulae 1.296 compute ı̄|t̄ and ı̄|t̄+dt̄

ı̄| −ı̄|
6. If ı̄|t̄ fulfill the criteria 1.303 and t̄+ddt̄t̄ t̄ fulfill the criterion 1.304 the
assumed stiffness and damping coefficient is available. An example of the available
stiffness and damping coefficients for x̄o = 0.3, c̄o = 0.2 is shown in Fig.52
MATHEMATICAL MODELS OF BEARINGS 90

15

10
cx (non-
dimensional)

0
-15 -10 -5 0 5 10 15
kx (non-dimansional

-5

-10

-15

Figure 52

The magnitudes of the stiffness k and damping c could be varied during opera-
tion of the rotor system. This feature of the magnetic bearings allow for active control
of stability of the rotor system equilibrium position as well as for control of its forced
response. The available magnitudes of k and c are usually supply by manufacturer.
This characteristic allows the mathematical model of the magnetic bearing to
be established in the following form

cẋ + kx = R (1.309)

where
⎡ ⎤ ⎡ ⎤
cx −cx 0 0 kx −kx 0 0
⎢ −cx cx ⎥ ⎢ 0 ⎥
c=⎢
0 0 ⎥ ; k = ⎢ −kx kx 0 ⎥
⎣ 0 0 cy −cy ⎦ ⎣ 0 0 ky −ky ⎦
0 0 −cy cy 0 0 −ky ky
⎡ ⎤ ⎡ ⎤
xj Rxj
⎢ xh ⎥ ⎢ ⎥
x=⎢ ⎥ ; R = ⎢ Rxh ⎥
⎣ yj ⎦ ⎣ Ryj ⎦
yh Ryh
The indexes j and h refers to motion of the journal and housing respectively. The
magnetic bearings do not require any medium, therefore very often they are used in
space craft and to suspend flywheels for energy storage that work usually in vacuum.
MATHEMATICAL MODELS OF BEARINGS 91

1.3.3 Rolling-elements bearings

Figure 53

Rolling-elements bearings ( see Fig. 62) are used where the bearing capability of
sustaining full-load capacity for low or even zero rotating speed is required. They
are compact and posses high load capacity. As compare with the oil bearing and the
magnetic bearings, the rolling-element bearing have the following disadvantages:
1. There is always rolling contact between rollers, stator, rotor races and the
cages. Therefore the life time of the rolling-elements bearings is considerably lower.
2. Rolling elements have no internal damping, therefore very often additional
dampers are required to improve stability of the rotor bearing system.

δc δh

a) b)

δo

c) d)

Figure 54
MATHEMATICAL MODELS OF BEARINGS 92

To develop a mathematical model for rolling-elements bearings one have to

take into account the following aspects.
1. In order not to overload the rolling elements and the races there is always
clearance between the outer and inner rings. It results in the deflection δc shown in
Fig. 54a)
2. Due to the external load, one has to take into account the Hertzian de-
flection of the rolling elements and the races. This deflection δ h is shown in Fig.
54b).
3. Due to flexibility of the housing, the rolling-element bearing and not uniform
load transmitted by the rolling elements, the outer ring (see Fig. 54c)) as well as the
inner became elliptical. This effect, called ovalization, results in deflection δ o shown
in Fig. 54d).

F
δc δh δo

Fo

FA

A δ

Figure 55

To compute these deflection one has to assume magnitude of the load Fo .

These computed deflection allow the stiffness characteristic for a bearing to be drawn.
Example of such a characteristic is shown in Fig. 55. This stiffness characteristic
(solid line) is non-linear. For the given amplitude of vibration the diagram in Fig. 55
allows to obtain the linearized stiffness. One of the simplest method of linearization is
shown in Fig. 55 with broken line. According to this method, the linearized stiffness
is:
F Fo (A − δ c )
kA = = for A > δ c (1.310)
A A(δ h + δ o )
The linearized stiffness is dependant of the amplitude of vibration of the rotor system
A at the bearing and declines to zero when the amplitude of vibration approaches
MATHEMATICAL MODELS OF BEARINGS 93

the clearance δ c . It is zero for the amplitude of vibrations less then the clearance δ c .
Since rolling-element bearings introduces a negligible damping, The final math-
ematical model can be adopted as follows

kx = R (1.311)

where: ⎡ ⎤ ⎡ ⎤
kA −kA 0 0 xj
⎢ −kA kA 0 0 ⎥ ⎢ xb ⎥
k=⎢
⎣ 0
⎥, x=⎢ ⎥ (1.312)
0 kA −kA ⎦ ⎣ yj ⎦
0 0 −kA kA yb
xj , xb , yj , yb in the above formulae stands for displacement of the journal and housing
of the bearing respectively..
Because the linearized stiffness depends on the amplitude of vibration A, a
method of the successive approximations must be used to solve the problem of the
forced vibrations of the rotor system. To this end, one should
1 arbitrarily assume the relative amplitude of vibration A1 .
2 calculate the linearized stiffness according to the formula 1.310.
3 compute amplitudes of the forced vibrations for the whole system to obtain
a better approximation of the amplitude of vibration at the bearing A2
4 repeat 2 and 3 till the improvement in amplitudes of vibration is negligible.
MATHEMATICAL MODELS OF BEARINGS 94

1.3.4 Squeeze film dampers

To improve stability of the rotor system and its force response very often it is necessary
to introduce devices design especially for dissipation of energy. Such a devices are
called dampers. This devices may be passive or active. The magnetic bearings can
be considered as active dampers if the force produced is of the following form

Fx = cẋ
Fy = cẏ (1.313)

Another damper that can be classified as passive one is shown in Fig. 56.

6 5

Figure 56

The rotating shaft 1 is connected to the flexible bush 3 via rolling element
bearing 2. Between the bush 3 and the housing 4 there is the oil film 5. The oil from
between the housing and the bush 3 is released to atmosphere (open ends dampers)
or is prevented from discharge by means of the O-rings shown in the Fig.56 (closed
ends dampers) The oil is supply through a center hole 6 as shown in Fig.56 or center
groove. The four possible designs of the dampers are shown in Fig.57. The dotted
area shows the boundary conditions and an approximate pressure distribution in axial
direction for the individual design. The flexible bush 3 allows to keep the shaft in
central position with respect to the housing.
MATHEMATICAL MODELS OF BEARINGS 95

L L/2 L/2 L L L

a) closed ends b) closed ends c)open ends - d)open ends -

- centre hole - centre groove centre hole center grove

Figure 57

ϕ
R

e
h
. y
x x
. Hx
y
v x H
Hy

β α=ω t
y

Figure 58

Due to the ever existing imbalance the bush 3 performs precession (translation
along closed curve) along a circular orbit of radius e with speed ω equal to the angular
MATHEMATICAL MODELS OF BEARINGS 96

velocity of the rotor Ω (see Fig. ??). Hence its instantaneous position and velocities
are

x = e cos ωt = e cos α
y = e sin ωt = e sin α (1.314)
ẋ = −eω sin ωt = −eω sin α
ẏ = eω cos ωt = eω cos α (1.315)

To compute the pressure distribution for the above position and velocity of the centre
of the bush 3 we can adopt the Reynolds equation 1.269 in the following form
µ ¶ µ ¶
1 ∂ h3 ∂p ∂ h3 ∂p ∂h
2
+ = 12 (1.316)
R ∂ϕ η ∂ϕ ∂z η ∂z ∂t

The first term in the right hand side is neglected since the angular velocity Ω of
the bush is equal to zero. Integration of the pressure yields the components of the
hydrodynamic force along the axis x and y.
+L ⎛ 2π ⎞
Z2 Z
Hx = − ⎝ p(ϕ, z) sin ϕdϕ⎠ dz
−L 0
2

+L ⎛ 2π ⎞
Z2 Z
Hy = + ⎝ p(ϕ, z) cos ϕdϕ⎠ dz (1.317)
−L 0
2

On the other hand for the harmonic motion (see Eq. 1.314) and the same properties
of the damper oil film in x and y direction one can see that the angle β and the
magnitude of the hydrodynamic force are independent of the position (x, y). Hence

Hx = H cos(α + β) = H cos β cos α − H sin β sin α

Hy = H sin(α + β) = H cos β sin α + H sin β cos α (1.318)

or, taken into consideration the equation 1.314 and 1.315

H cos β H sin β
Hx = x+ ẋ
e eω
H cos β H sin β
Hy = y+ ẏ (1.319)
e eω
From the last equation one can see that the components of the hydrodynamic force
are linear functions of the journal displacement and its velocity.

Hx = kx + cẋ
Hy = ky + cẏ (1.320)
MATHEMATICAL MODELS OF BEARINGS 97

The coefficients k and c are constant and are called stiffness and damping coefficient
respectively.
H cos β
k(e, ω) =
e
H sin β
c(e, ω) = (1.321)
eω
As one can see from the procedure presented, the stiffness and damping coefficients
are function of the the eccentricity e and the speed of the precession ω.In some cases
it is possible to predict the pressure distribution in the axial direction z. For example
given in Fig. 57a) due to lack of flow in z direction the pressure is constant in this
direction. In this case the Reynolds equation can be solved analytically (the long
bearing theory). This solution results in the following expressions for the damping
and stiffness coefficients†
24R3 Lηω
k = (1.322)
r3 (2 + 2 )(1 − 2 )
12πR3 Lη
c = p (1.323)
r3 (2 + 2 ) (1 − 2 )

In case b), c), and d) presented in Fig. 57 so called short bearing theory yields
the approximate formulae for the stiffness and damping coefficient as a function of
the eccentricity e speed ω.‡
2RL3 ηω
k = p (1.324)
r 3 (1 − 2 )
πRL3 η
c = p (1.325)
2r3 (1 − 2 )
In the above formulae:
R - outer radius of the bush
L - length of the damper as defined in Fig. 57
η - oil viscosity
ω - speed of the precession (usually equal to the angular velocity of the shaft
Ω)
= re - eccentricity ration
e - radius of the orbit
r - radial clearance
Similar considerations to these presented in the previous sections lead to con-
clusion that the mathematical model of the squeeze film damper is as follows

cẋ + kx = R (1.326)
†
Handbook of Rotordynamics, Fredric F. Ehrich, Krieger Publishing Company Inc.1999 page
1.55
‡
Handbook of Rotordynamics, Fredric F. Ehrich, Krieger Publishing Company Inc.1999 page
1.55
MATHEMATICAL MODELS OF BEARINGS 98

where
⎡ ⎤ ⎡ ⎤
c −c 0 0 k −k 0 0
⎢ −c c 0 ⎥
0 ⎥ ⎢ −k k 0 0 ⎥
c=⎢
⎣ 0 ; k=⎢ ⎥ (1.327)
0 c −c ⎦ ⎣ 0 0 k −k ⎦
0 0 −c c 0 0 −k k
⎡ ⎤ ⎡ ⎤
xb Rxb
⎢ xh ⎥ ⎢ Rxh ⎥
x=⎢ ⎥ ⎢ ⎥
⎣ yb ⎦ ; R = ⎣ Ryb ⎦ (1.328)
yh Ryh
Where xb , yb represents motion of the bush and xh , yh represent motion of the housing.
The vector R represents the interaction forces between the oil film and the bush as
well as between the oil film and the housing.
Since the matrices k and c are functions of the unknown eccentricity e the prob-
lem of vibrations can be solved by means of the subsequent approximation method.
MATHEMATICAL MODELS OF BEARINGS 99

1.3.5 Conclusions
The above consideration lead to conclusion that for analysis of small vibrations around
the equilibrium position of the rotor - bearing system, the dynamic properties of each
sub-system can be approximated by means of mathematical model that has the same
general form 1.329
mẍ + cẋ + kx = R + F (1.329)
where R stands for vector of the interaction forces between the sub-systems involved
and F stands for vector of the external load.
The formulated mathematical models of sub-systems forms bases for creation
of the mathematical model of the rotor - bearing system.
MATHEMATICAL MODELS OF BEARINGS 100

ϕ n ϕ
o n

x1
r
x
R
Fo

y1
y

Figure 59

1.3.6 Problems

Problem 7

Fig. 59 shows a three partial sleeves bearing. Its dynamic properties are
determined by the following parameters:
ϕ01 =18.50 ,ϕ1 =470 , ϕ02 =97.70 ,ϕ2 =1500 , ϕ03 =282.50 ,ϕ1 =65.50 - angles deter-
mining position of the partial sleeves
L/D = 0.8 - length to diameter ratio
R = 0.05[m] - radius of sleeve
r = 0.0005[m] - radial clearance
L = 0.08[m]- length of the sleeve
η = 0.02[Ns/m2 ] - viscosity of oil
Fo = 1000[N] - static load.
β = 30o - angle between direction of the load and axis y.
Ω = 300[rad/s] - angular velocity of the journal.
Develop mathematical model of the bearing along system of coordinates xy.
MATHEMATICAL MODELS OF BEARINGS 101

Solution.
The non-dimensional load along axis y1 , according to Eq. 1.267 is
Fo r2 1000 × 0.00052
S= = = 2.1 (1.330)
2R3 LηΩ 2 × 0.053 × 0.08 × 0.02 × 300
According to Fig. 41a), the non-dimensional eccentricity c is
c = 0.58 (1.331)
According to Fig. 41b the non-dimensional coordinates of the journal’s equilibrium
position with respect to the system of coordinates x1 y1 are
x̄1o = −0.43, ȳ1o = 0.4 (1.332)

x1
xo
x1 x
o yo y1
β o

Fo

y1
y

Figure 60

Hence, the dimensional coordinates of the journal equilibrium position with

respect to the system of coordinates x1 y1 are
x1o = x̄1o r = −0.43 × 0.0005 = −0.000215[m],
y1o = ȳ1o r = 0.4 × 0.0005 = 0.0002[m] (1.333)
Matrix of direction cosines between system of coordinates xy and x1 y1 , according to
Fig. 60, is
∙ ¸ ∙ ¸∙ ¸ ∙ ¸
x cos β sin β x1 x1
= = [C]
y − sin β cos β y1 y1
∙ ¸∙ ¸
0.87 0.5 x1
= (1.334)
−05 0.87 y1
Hence, the dimensional coordinates of the journal equilibrium position with respect
to the system of coordinates xy are
∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
xo 0.87 0.5 x1o 0.87 0.5 −0.000215
= =
yo −0.5 0.87 y1o −0.5 0.87 0.0002
∙ ¸
−8.705 × 10−5
= m (1.335)
2.815 × 10−4
MATHEMATICAL MODELS OF BEARINGS 102

Diagrams, shown in Fig. 45, allow the non-dimensional stiffness and damping matri-
ces to be obtained.
∙ ¸ ∙ ¸
1.2 −0.8 1.1 1.3
k̄1 = , c̄1 = (1.336)
6.4 3.6 0.9 6.8

According to formula 1.282 the dimensional stiffness and damping matrices are
∙ ¸ ∙ ¸
Fo 1000 1.8 0.8 1.714 3 × 106 7.619 × 105
k1 = k̄1 = = [N/m]
Sr 2.1 × 0.0005 −3.6 3.1 −3.428 6 × 106 2.952 4 × 106
∙ ¸ ∙ ¸
Fo 1000 1.2 −2.1 3809.5 −6666.7
c1 = c̄1 = = [Ns/m]
SrΩ 2.1 × 0.0005 × 300 −2.1 5.5 −6666.7 17460.0
(1.337)

Their dimensional magnitudes, in the system of coordinates xy, can be produced with
help of the matrix of direction cosines (Eq. 1.334).
The restoration forces along system of coordinates xy are
∙ ¸ ∙ ¸ ∙ ¸
x1 T x x
Fk = [C] Fk1 = [C] k1 = [C] k1 [C] =k (1.338)
y1 y y

where
∙ ¸∙ ¸∙ ¸
T 0.87 0.5 1.714 3 × 106 7.619 × 105 0.87 −0.5
k = [C] k1 [C] =
−0.5 0.87 −3.428 6 × 106 2.952 4 × 106 +0.5 0.87
∙ ¸
8.75 × 105 1.97 × 106
= [N/m] (1.339)
−2.24 × 106 3. 82 × 106

Similarly
∙ ¸∙ ¸∙ ¸
T0.87 0.5 3809.5 −6666.7 0.87 −0.5
c = [C] c1 [C] =
−0.5 0.87 −6666.7 17460.0 −0.5 0.87
∙ ¸
−1481 2558
= [Ns/m] (1.340)
−15965 19968

Hence, the mathematical model of the oil film, according to Eq. 1.287 is
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸
k −k xj c −c ẋj H
+ = (1.341)
−k k xb −c c ẋb −H

where: ∙ ¸
xj
xj = stands for displacements of journal from its equilibrium position.
∙ yj ¸
xb
xb = stands for displacements of the bearing case from its equilibrium
yb
position. ∙ ¸
Hx
H= stands for the dynamic interaction forces.
Hy
MATHEMATICAL MODELS OF BEARINGS 103

Problem 8

Figure 61

Rotor of the electrical motor shown in Fig. 61, is supported by means of two
identical roller bearings. The manufacturer specification for these bearing is:

Races Rollers
Bore diameter D1 [m] 0.11 Number of rollers N 28
Outer diameter D2 [m] 0.15 Roller diameter D[m] 0.011
With of races [m] 0.020 Overall length [m] 0.011
Effective length l[m] 0.00827
Diametral clearance c[m] 0.00001493
Young modulus E1,2 [Nm−2 ] 0.2×1012 Young modulus E[Nm−2 ] 0.25×1012
Poison ratio ν 1,2 0.3 Poison ratio ν 0.3

Produce the mathematical model for the roller bearing.

MATHEMATICAL MODELS OF BEARINGS 104

Solution
The linearized stiffness of the roller bearing
a) Clearance deflection
The clearance deflection is equal to half of the diametral clearance

δ c = 7.47 · 10−6 m (1.342)

b) Hertzian deflection

Fo

α F'cosα sinα

F'cosα
F' 2
F'cos α

Figure 62

The following equilibrium condition (see Fig.62) yields the relationship be-
tween the arbitrarily assumed load

Fo = 5000N (1.343)

and the load taking by the lowest roller F 0 .

Fo = (1 cos2 0o + 2 cos2 12.86o + 2 cos2 25.72o + 2 cos2 38.58o

+2 cos2 51.44o + 2 cos2 64.3 + 2 cos2 77.16 + 2 cos2 90o )
= 5.57F 0 (1.344)

Hence
Fo 5000N
F0 = = ≈ 900N (1.345)
5.57 5.57
Dividing this number by the effective length l one can get the unit force F

F0 900
F = = ≈ 1 · 105 N (1.346)
0.00827 0.00827
MATHEMATICAL MODELS OF BEARINGS 105

D2
D1
b1 c1

b2 c2

Figure 63

According to the Hertz formula, the with of the contact area is equal to
s µ ¶
D1 D 1 − ν 21 1 − ν 2
b1 = 0.8 F + (1.347)
D1 + D E1 E2
s µ ¶
5
0.11 · 0.011 1 − 0.32 1 − 0.32
= 0.8 1 · 10 + (1.348)
0.11 + 0.011 0.2 · 1012 0.25 · 1012
√
= 0.8 1 · 105 · 8.19 · 10−14 = 72 · 10−6 m (1.349)

s µ ¶
D2 D 1 − ν 22 1 − ν 2
b2 = 0.8 F + (1.350)
D2 − D E2 E
s µ ¶
5
0.15 · 0.011 1 − 0.32 1 − 0.32
= 0.8 1 · 10 + (1.351)
0.15 − 0.011 0.2 · 1012 0.25 · 1012
√
= 0.8 1 · 105 · 9.72 · 10−14 = 79 · 10−6 m (1.352)

Neglecting the deflection of races (their diameter is much higher then roller diameter)
the total Hertzian deflection, according to Fig. 63 is
sµ ¶ sµ ¶
2 2
D D D D 1 ¡ 2 ¢
δh = c1 + c2 = − 2
− b1 + − − b22 = b1 + b22 = 1 · 10−6 m
2 2 2 2 D
(1.353)
c) Ovalization deflection
The deflection due to ovalization for the load Fo = 5000N was computed to
be
δ o = 5 · 10−6 m (1.354)
The above computations results in the stiffness characteristic of the bearing shown
in Fig. 64.
MATHEMATICAL MODELS OF BEARINGS 106

F [N]
δc δh δo

5000

FA

A 10 .10 -6 δ [m]

Figure 64

From this diagram one can see that the linearized stiffness of the bearing as a
function of amplitude of vibration A is

FA Fo (A − δ c ) 5000(A − 7.47 · 10−6 ) A − 7.47 · 10−6

kA = = = = (1.355)
A A(δ h + δ o ) A(1 · 10−6 + 5 · 10−6 ) 12 · 10−10 A

yj

yh
kA

Figure 65

Hence, the mathematical model for the bearing represented in Fig. 1 by the
spring of stiffness kA can be written as follows

kx = R (1.356)
MATHEMATICAL MODELS OF BEARINGS 107

where
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
kA −kA 0 0 xj Rxj
⎢ −kA kA 0 0 ⎥ ⎢ xh ⎥ ⎢ Rxh ⎥
k=⎢
⎣ 0
⎥; x=⎢ ⎥
⎣ yj ⎦ ; R =⎢ ⎥
0 kA −kA ⎦ ⎣ Ryj ⎦
0 0 −kA kA yh Ryh

Their index j and h refers to motion of the journal and housing respectively.
MATHEMATICAL MODELS OF FOUNDATIONS (ENVIRONMENT) 108

1.4 MATHEMATICAL MODELS OF FOUNDATIONS (ENVIRONMENT)

Foundations the rotor is supported upon due to a very complex shape and not neg-
ligible damping usually can not be modelled by means of the analytical methods. In
this case the dynamic properties, very often, are determined from experiment.
1.4.1 Determination of dynamic properties by means of experimental in-
vestigation
For analysis of small vibrations in vicinity of equilibrium position the mathematical
model of the foundation can be assumed in the following form

m ẍ + c ẋ + k x = 0 (1.357)

where
x - is the displacement vector
m, c, k, - inertia, damping and stiffness matrix respectively
The unknown matrices m, c, and k can be identified by means of experimental
investigation. To this end, let us assume that for the future composition and the
necessary accuracy of modelling, the mathematical model has to be produced along
the following vector of coordinates x

x = [x1 , x2 , .....xi , .....xj , ......xN ]T (1.358)

F(t) Fj ( iω )
j Ri j (iω )
t Fourier ω x( iω )
i___
x(t) x(i iω )
i transformation
Fj ( iω ) ω
t ω

xj xi

Figure 66

In order to produce the transfer function between the coordinate xi and the
coordinate xj let us apply force Fj (t) and record it simultaneously with the foundation
response xi (t) along the coordinate xi (see Fig. 66). Fourier transformation applied
to these functions
Z +∞
Fj (iω) = e−iωt Fj (t)dt
−∞
Z +∞
xi (iω) = e−iωt xi (t)dt (1.359)
−∞
MATHEMATICAL MODELS OF FOUNDATIONS (ENVIRONMENT) 109

yields the Fourier transforms in the frequency domain xi (iω) and Fj (iω). The mod-
ulus of the complex numbers xi (iω) and Fj (iω)
p
|xi (iω)| = Re(xi (iω))2 + Im(xi (iω))2
q
|Fj (iω)| = Re(Fj (iω))2 + Im(Fj (iω))2 (1.360)

represent the amplitude of displacement and force respectively as a function of the

frequency ω. The corresponding phases are determined by the following formulae.
Re(xi (iω))
ϕxi = arctan
Im(xi (iω))
Re(Fj (iω))
ϕF j = arctan (1.361)
Im(Fj (iω))
These Fourier transforms allow the transfer function Rij (iω) to be computed.
xi (iω)
Rij (iω) = (1.362)
Fj (iω)
The above formula determine response of the system along coordinate xi caused by
the harmonic excitation Fj along the coordinates xj .

xi (iω) = Rij (iω)Fj (iω) (1.363)

Since the system considered is by assumption linear, the response along the coordinate
xj caused by set of forces acting along coordinates 1.358 according to the superposition
principle is
j−N
X
xi (iω) = Rij (iω)Fj (iω) i = 1....N (1.364)
j−1

The relationship above can be rewritten in the following matrix form

x(iω) = Rij (iω)F(iω) (1.365)

or
F(iω) = R−1
ij (iω)x(iω) (1.366)
Application of the above described experimental procedure to all coordinates involved
in the modelling allows to formulate this matrix of the transfer functions Rij (iω).
There exists a simple relationship between matrix of the transfer functions a
Rij (iω) and the unknown matrices m, c, and k. To develop this relationship let us
consider the mathematical model of the foundation 1.357. Assume that, similarly as
it was in the experiment described, there is a set of the harmonic forces acted along
the coordinates 1.358.
m ẍ + c ẋ + k x = F(iω)eiωt (1.367)
The response due to this excitation can be predicted in the following form

x(t) = x(iω)eiωt (1.368)

MATHEMATICAL MODELS OF FOUNDATIONS (ENVIRONMENT) 110

Introduction of 1.368 into 1.367 yields

((−ω 2 m + k) + iωc)x(iω) = F(iω) (1.369)

Comparison of with 1.369 with 1.366 allow to conclude that

R−1 2
ij (iω) = (−ω m + k) + iωc (1.370)

The matrix R−1

ij (iω) will be denoted by Sij (iω) and is called matrix of the dynamic
stiffness.
R−1 2
ij (iω) = Sij (iω) = (−ω m + k) + iωc (1.371)
This relationship above is often used for identification the matrices m, c, and k. The
simplest procedure of identification is as follow
For ω = 0 , the Eq. 1.370 yields directly the stiffness matrix.

k = Re Sij (i0) (1.372)

For arbitrarily chosen ω = ω 1 the same equation yields

Sij (iω 1 ) = (−ω21 m + k) + iω 1 c (1.373)

From Eq. 1.373 one can see that the matrix ω 1 c has to be equal to the imaginary
part of Sij (iω1 ). Hence,
1
c= Im Sij (iω 1 ) (1.374)
ω1
The real part of Sij (iω 1 ) has to be equal to −ω 21 m + k. Therefore
1
m= (k − Re Sij (iω 1 )) (1.375)
ω 21
Taking into consideration Eq. 1.372 one may produce the inertia matrix in the
following form
1
m= (Re Sij (i0) − Re Sij (iω 1 )) (1.376)
ω 21
This procedure, due to ill conditioning of the algebraic equations involved, is not
reliable. Process of identification of parameters of the mathematical model is still
under intensive research.
1.4.2 Modelling of environmental excitation
The foundation can have its own motion due to the environment the machine operates
in. This motion can be measured along all the coordinates involved as shown in Fig.
67.
MATHEMATICAL MODELS OF FOUNDATIONS (ENVIRONMENT) 111

ex i (t)

xi
x

Figure 67

Let us denote the measured vector of environmental excitation by e.

⎡ . ⎤
..
⎢ ⎥
e = ⎣ exi (t) ⎦ (1.377)
..
.
If the mathematical model of the foundation considered is available, the environmen-
tal motion can be simulated by vector of forces computed according to the following
formula
Fe (t) = më + cė + ke (1.378)
This vector has to be added to the mathematical model 1.357

mẍ + cẋ + kx = Fe (t) (1.379)

1.4.3 Modelling of configuration

Due to manufacturing errors the position of bearings of the foundation not necessarily
have to coincide the axis z (see Fig. 68). Vector of their positions with respect to the
stationary system of coordinates xyz is called vector of configuration and is denoted
here by a. ⎡ . ⎤
..
⎢ ⎥
a = ⎣ axi ⎦ (1.380)
..
.

ax i
z

xi
x

Figure 68
MATHEMATICAL MODELS OF FOUNDATIONS (ENVIRONMENT) 112

The mathematical model 1.357 allows to compute vector of forces which must
be applied to the perfectly align foundation to keep it in the position determined by
the vector of configuration.
Fa = k · a (1.381)
If the above vector is added to the mathematical model of the foundation, the foun-
dation can be considered as perfectly align.

m ẍ + c ẋ + k x = Fe (t) + Fa (1.382)

To accomplish the modelling of foundation the interaction forces must be added

along these coordinates along which the foundation is connected to all the other
sub-systems.
m ẍ + c ẋ + k x = R + F (1.383)
where:
F = Fe (t) + Fa (1.384)
MATHEMATICAL MODELS OF FOUNDATIONS (ENVIRONMENT) 113

1.4.4 Problems

Problem 9

O
x

Figure 69

The dynamic properties of the support shown in Fig. 69 are determined by

means of the inertia matrix m and the stiffness matrix k along the coordinates x and
y.

∙ ¸ ∙ ¸ ∙ ¸
mxx mxy kxx kxy cxx cxy
m= , k= , c= (1.385)
myx myy kyx kyy cyx cyy

Motion of this support along the coordinates x and y was measured to be

ex = Sx sin ωt + Cx cos ωt
ey = Sy sin ωt + Cy cos ωt (1.386)

Produce the mathematical model of the support.

MATHEMATICAL MODELS OF FOUNDATIONS (ENVIRONMENT) 114

Solution.
Forces that simulate motion of the support considered, according to Eq. 1.378,
are

Fe (t) = më + cė + ke

∙ ¸ ∙ ¸
Sx 2 Cx
= −m ω sin ωt − m ω2 cos ωt
Sy Cy
∙ ¸ ∙ ¸
Sx Cx
+c ω cos ωt − c ω sin ωt
Sy Cy
∙ ¸ ∙ ¸
Sx Cx
+k sin ωt + k cos ωt
Sy Cy
∙ ¸ ∙ ¸
2 Sx Cx
= ((−mω + k) − cω ) sin ωt
Sy Cy
∙ ¸ ∙ ¸
2 Cx Sx
+((−mω + k) + cω ) cos ωt
Cy Sy
= FS sin ωt + FC cos ωt (1.387)
Hence, the mathematical model of the support may be adopted in the following form

mẍ + cẋ + kx = FS sin ωt+FC cos ωt+R (1.388)

where:
∙ ¸ ∙ ¸ ∙ ¸ ∙ ¸
2 Sx Cx 2 Cx Sx
FS = (−mω + k) − cω , FC = (−mω + k) + cω
Sy Cy Cy Sy
(1.389)
 Chapter 2
COMPOSITION OF MATHEMATICAL MODEL OF ROTOR
SYSTEMS

2.1 MATHEMATICAL MODEL OF TWO SUB-SYSTEMS

Let us consider two sub-systems n and m that are to be connected to each other
along vector of coordinates xc ( Fig. 1)

x 1n x2n x n3 x4n

x9m x 1m x 2m xm3

x1 x2 x3 x4
x9

Figure 1

In this case the connecting coordinates for the sub-system n and m are as
follows

xnc = [xn1 , xn2 , xn3 ]T (2.1)

m T
xm
c = [xm m
1 , x2 , x3 ] (2.2)

Let us denote vectors of the not engaged coordinates by xnu and xm

u for the sub-system
n and m respectively. Hence,

xnu = [xn4 , .......]T (2.3)

T
xm
u = [xm9 , .....] (2.4)

The mathematical models of the sub-system n and m, after reorganization according

MATHEMATICAL MODEL OF TWO SUB-SYSTEMS 116

to the following displacement vectors

∙ ¸
n xnu
x = (2.5)
xnc
∙ ¸
m xm
c
x = (2.6)
xm
u

can be adopted as follows

∙ n ¸∙ n ¸ ∙ n ¸∙ n ¸ ∙ n ¸∙ n ¸
muu mnuc ẍu cuu cnuc ẋu kuu knuc xu
n n n + n n n + n n
m m ẍ ccu ccc ẋc kcu kcc xnc
∙ ncu¸ ∙cc n ¸ c
Ru Fu
= n + (2.7)
Rc Fnc

∙ ¸∙ m ¸ ∙ m m ¸∙ m ¸ ∙ m ¸∙ m ¸
mm m
cc mcu ẍc ccc ccu ẋc kcc km
cu xc
m m m + n m m + m m
muc muu ẍu cuc cuu ẋu kuc kuu xm
u
∙ m ¸ ∙ m ¸
Rc Fc
= m + (2.8)
Ru Fm
u

Partition of those above equations yields

mnuu ẍnu + mnuc ẍnc + cnuu ẋnu + cnuc ẋnc + knuu xnu + knuc xnc = Rnu + Fnu
mncu ẍnu + mncc ẍnc + cncu ẋnu + cncc ẋnc + kncu xnu + kncc xnc = Rnc + Fnc (2.9)

mm m m m m m m m m m m m
cc ẍc + mcu ẍu +ccc ẋc + ccu ẋu + kcc xc + kcu xu = Rm m
c + Fc
mm m m m n m m m m m m m
uc ẍc + muu ẍu + cuc ẋc + cuu ẋu + kuc xc + kuu xu = Rm m
u + Fu (2.10)

The combined system must fulfil so called continuity conditions

xnc = xc
(2.11)
xmc = xc

and equilibrium conditions

Rnc +Rmc = 0
(2.12)
Fnc +Fm
c = Fc

Upon introducing the continuity and equilibrium conditions into equations 2.9 and
2.10, one may obtain the following equations
mnuu ẍnu +mnuc ẍc +cnuu ẋnu +cnuc ẋc
+knuu xnu +knuc xc = Rnu +Fnu

mncu ẍnu +(mncc +mm m m n n n m m m

cc )ẍc +mcu ẍu +ccu ẋu +(ccc +ccc )ẋc +ccu ẋu
n n n m m m (2.13)
+kcu xu + (kcc +kcc )xc + kcu xu =0 + Fnc +Fm
c

+mm m m
uc ẍc +muu ẍu +cm m m
uc ẋc +cuu ẋu
m m m
+kuc xc + kuu xu = Rm u +Fm
u
CONDENSATION OF MATHEMATICAL MODELS 117

The above equations, which govern motion of the created system, can be written in
the matrix form
⎡ n ⎤⎡ n ⎤ ⎡ ⎤⎡ n ⎤
muu mnuc 0 ẍu cnuu cnuc 0 ẋu
⎣ mncu mncc +mm cc m m ⎦⎣
cu ẍc ⎦ + ⎣ cncu cncc +cmcc cm ⎦⎣
cu ẋc ⎦
m m
0 muc muu ẍmu 0 m
cuc m
cuu ẋm
u

⎡ ⎤⎡ n ⎤ ⎡ ⎤ ⎡ ⎤
knuu knuc 0 xu Rnu Fnu
⎣ n n m m ⎦⎣
+ kcu kcc +kcc kcu xc ⎦ = ⎣ 0 ⎦ + ⎣ Fnc +Fmc
⎦
m m
0 kuc kuu xmu Rum
Fum

(2.14)
or shortly
mẍ + cẋ + kx = R + F (2.15)
As one can notice, the resultant matrices can be obtained by overlapping matrices of
individual sub-systems
n
n 0 Fn R
m

m= m
F= R= 0
m
0 F
m
Rm

(2.16)
The created system is of the same form as each sub-system. Therefore it can be, in
the next stage of composition, considered as sub-system.
During composition of the mathematical model the size of the matrices m, c,
and k is getting larger. On the other hand, some of those coordinates which originally
were introduced to increase accuracy of modelling of the sub-systems or to join them
together are not any more important for future analysis. To facilitate the future
analysis, those unnecessary coordinates can be eliminated from the mathematical
model at each stage of composition with little harm to the final results. The procedure
which allows to squeeze the matrices involved is called condensation.

2.2 CONDENSATION OF MATHEMATICAL MODELS

According to the previous consideration, at any stage of composition, it is possible
to adopt the mathematical model in the following form

mẍ + cẋ + kx = R + F (2.17)

Let us assume, that this equation is arranged in such a way that the coordinates
which are to be eliminated due to the condensation procedure xe are located in the
upper part of the vector x and these which are to be retained for further consideration
xr are located in its lower part.
CONDENSATION OF MATHEMATICAL MODELS 118

∙ ¸
xe
x= (2.18)
xr
Partitioning of the equations 2.17 yields
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸ ∙ ¸
mee mer ẍe cee cer ẋe kee ker xe Re Fe
+ + = +
mre mrr ẍr cre crr ẋr kre krr xr Rr Fr
(2.19)
It is obvious that the vector of interaction forces Re along the coordinates to be
eliminated is equal to 0.
Re = 0 (2.20)
To eliminate the coordinates xe from the mathematical model 2.19, one have to
determine the relationship between the coordinates xe and the coordinates xr . One
of many possibilities is to assume that the coordinates xe are obeyed to the static
relationship that can be obtained from 2.19.
∙ ¸∙ ¸ ∙ ¸
kee ker xe 0
= (2.21)
kre krr xr Rr
Hence, upon partitioning equation 2.19 one may obtain
kee xe +ker xr = 0 (2.22)
Therefore the sought relationship is
xe = hxr (2.23)
where
h = −k−1
ee ker (2.24)
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.
2.2.1 Condensation of the inertia matrix
According to the first criterion, the kinetic energy of the system before and after
condensation must be the same. The kinetic energy of the system before condensation
is
∙ ¸∙ ¸
1£ T T
¤ mee mer ẋe
T = ẋe ẋr
2 mre mrr ẋr
1¡ T ¢
= ẋe mee ẋe + ẋTe mer ẋr + ẋTr mre ẋe + ẋTr mrr ẋr (2.25)
2
CONDENSATION OF MATHEMATICAL MODELS 119

Introduction of 2.23 yields

1³ ´
T = [hẋr ]T mee hẋr + [hẋr ]T mer ẋr + ẋTr mre hẋr +ẋTr mrr ẋr
2
1¡ T T ¢
= ẋr h mee hẋr +ẋTr hT mer ẋr +ẋTr mre hẋr +ẋTr mrr ẋr
2
1¡ T£ T ¤ ¢
= ẋr h mee h + hT mer +mre h + mrr ẋr (2.26)
2
Hence, if the kinetic energy after condensation is to be the same, the inertia matrix
after condensation mc must be equal to

mc = hT mee h + hT mer +mre h + mrr (2.27)

2.2.2 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 definition of
the condensed damping matrix

cc = hT cee h + hT cer +cre h + crr (2.28)

2.2.3 Condensation of the stiffness matrix

Taking advantage from definition of potential energy of the system considered
∙ ¸∙ ¸
1£ T ¤ kee k er x e
V = xe xTr
2 kre krr xr
1¡ T ¢
= xe kee xe + xTe ker xr + xTr kre xe + xTr krr xr (2.29)
2
one can arrive to conclusion that the condensed stiffness matrix is of the form 2.30

kc = hT kee h + hT ker +kre h + krr (2.30)

It is easy to show that sum of the first two terms in the above expression is equal to
zero. Indeed, according to 2.24, they can be transformed as following.

hT kee h + hT ker = (−k−1 T −1 −1 T

ee ker ) kee (−kee ker ) + (−kee ker ) ker
= −(−k−1 T −1 T
ee ker ) ker + (−kee ker ) ker = 0 (2.31)

Hence,
kc = kre h + krr (2.32)
2.2.4 Condensation of the external forces
The virtual work performed by external forces F on the displacements x is
∙ ¸
£ T ¤ Fe
δW = xe xr T
= xTe Fe +xTr Fr (2.33)
Fr
Introduction of 2.23 into the above equation yields
¡ ¢ ¡ ¢
δW = (hxr )T Fe +xTr Fr = xTr hT Fe +xTr Fr = xTr hT Fe +Fr (2.34)
CONCLUSIONS 120

Hence, ¡ ¢
Fc = hT Fe +Fr (2.35)
The condensed mathematical model, according to the above consideration, can be
adopted as follows
mc ẍc +cc ẋc +kc xc = Rc +Fc (2.36)
where
xc = xr (2.37)
The relationship 2.23
xe = hxc
permits to produce displacement along the ’cut off’ coordinates on the base of solution
of the equation 2.36

2.3 CONCLUSIONS
The procedures, derived in this chapter, permit the mathematical model of rotor
- bearing system to be created. In the case considered all its sub-systems can be
linearized in vicinity of its equilibrium position. The concept of composition, and
condensation can be applied not only to linearized system but to non linear system
as well.
During composition the interaction forces gradually disappear and the final
mathematical model of the rotor-bearing system can be adopted in the following
form
mẍ + cẋ + kx = F (2.38)
PROBLEMS 121

2.4 PROBLEMS

Problem 10

a l b
k x3

x1 x2 x3

x4

Figure 2

. The mathematical model of the free-free beam, shown in Fig. 2a, has been
condensed to three coordinates x1 , x2 , x3 . The inertia and stiffness matrix as well as
the vector of gravity forces are as follow
⎡ ⎤ ⎡ ⎤
m11 m12 m13 k11 k12 k13
m = ⎣ m21 m22 m23 ⎦ ; k = ⎣ k21 k22 k23 ⎦ ;
m31 m32 m33 k31 k32 k33
⎡ ⎤ ⎡ ⎤
G1 x1
G = ⎣ G2 ⎦ ; x = ⎣ x2 ⎦ (2.39)
G3 x3
The beam is connected with the massless spring of the uncompressed length l and
stiffness k (Fig. 2b) along the coordinate x3 . The assembly is supported as shown in
Fig. 3.
Develop mathematical model of the system and compute the static deflection
line of the beam .

x3 l
x1 x2
a

x4

Figure 3
PROBLEMS 122

Solution.
Mathematical model of the spring, according to Eq. 1.287, can be adopted in
the following form ∙ ¸∙ ¸
k −k x3
=0 (2.40)
−k k x4

x3 l
x1 x2

x4

Figure 4

. According to the composition principle, the mathematical model of the free-

free beam-spring system, shown in Fig. 4 is
⎡ ⎤⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤ ⎡ ⎤
m11 m12 m13 0 ẍ1 k11 k12 k13 0 x1 G1 R1
⎢ m21 m22 m23 0 ⎥ ⎢ ẍ2 ⎥ ⎢ k21 k22 k 0 ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ ⎥⎢ ⎥ ⎢ 23 ⎥ ⎢ 2 ⎥ ⎢ 2 ⎥ ⎢ R2
x G ⎥
⎣ m31 m32 m33 0 ⎦ ⎣ ẍ3 ⎦+⎣ k31 k32 k33 + k −k ⎦ ⎣ x3 ⎦ = ⎣ G3 ⎦+⎣ 0 ⎦
0 0 0 0 ẍ4 0 0 −k k x4 0 R4
(2.41)
According to Fig. 3, the boundary conditions results in the following vector of dis-
placements. ⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x1 0 ẍ1 0
⎣ x2 ⎦ = ⎣ 0 ⎦ , ⎣ ẍ2 ⎦ = ⎣ 0 ⎦ (2.42)
x4 a ẍ4 0
Reorganization of the mathematical model 2.41 with respect to the boundary condi-
tions 2.42 yields
⎡ ⎤⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤
m11 m12 0 m13 ẍ1 k11 k12 0 k13 x1 G1
⎢ m21 m22 0 m23 ⎥ ⎢ ẍ2 ⎥ ⎢ k21 k22 0 k ⎥ ⎢ ⎥ ⎢ G2 ⎥
⎢ ⎥⎢ ⎥+⎢ 23 ⎥ ⎢ x2 ⎥=⎢ ⎥
⎣ 0 0 0 0 ⎦ ⎣ ẍ4 ⎦ ⎣ 0 0 k −k ⎦ ⎣ x4 ⎦ ⎣ 0 ⎦
m31 m32 0 m33 ẍ3 k31 k32 −k k33 + k x3 G3
(2.43)
Introduction of the boundary conditions 2.42 into 2.43 and then partition it results
in the following set of equations
⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
m13 k11 k12 0 0 k13 G1 R1
⎣ m23 ⎦ ẍ3 + ⎣ k21 k22 0 ⎦ ⎣ 0 ⎦ + ⎣ k23 ⎦ x3 = ⎣ G2 ⎦ + ⎣ R2 ⎦ (2.44)
0 0 0 k a −k 0 R4
⎡ ⎤
£ ¤ 0
m33 ẍ3 + k31 k32 −k ⎣ 0 ⎦ + (k33 + k)x3 = G3 (2.45)
a
PROBLEMS 123

The equation 2.45 governs motion of the system along the coordinate x3 .

m33 ẍ3 + (k33 + k)x3 = G3 + ka (2.46)

Its particular solution yields the static deflection of the system along the coordinate
x3 .
G3 + ka
x3s = (2.47)
k33 + k

x3s

l
x3
x1 x2
a

x4

Figure 5

. The homogeneous equation 2.46

m33 ẍ3 + (k33 + k)x3 = 0 (2.48)

governs the free motion of the system in vicinity of the static deflection line. Its
general solution is
x3o = A sin ωt + B cos ωt (2.49)
where the natural frequency ω is determine by
r
k33 + k
ω= (2.50)
m33
The equations 2.44 allow the reactions at the supports to be computed
⎡ ⎤ ⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤ ⎡ ⎤
R1 m13 k11 k12 0 0 k13 G1
⎣ R2 ⎦ = ⎣ m23 ⎦ ẍ3o + ⎣ k21 k22 0 ⎦⎣ 0 ⎦ + ⎣ k23 ⎦ (x3s + x3o ) − ⎣ G2 ⎦
R4 0 0 0 k a −k 0
(2.51)
PROBLEMS 124

Problem 11

To create the mathematical model of the system shown in Fig. 6a), it was
split into four sub-systems shown in Fig. 6b).

a)

x1 x2 x3
2 3

x1 x3
k c k c b)
x4 4 x5

x6 x4 x5 x7

Figure 6

The mathematical model of the free-free beam 1, condensed to the coordinate

vector x1 x2 x3 , is
m1 ẍ1 +k1 x1 = G1 +R1 (2.52)
where ⎡ ⎤ ⎡ ⎤
m11 m12 m13 k11 k12 k13
m1 = ⎣ m21 m22 m23 ⎦ , k1 = ⎣ k21 k22 k23 ⎦ ,
m31 m32 m33 k31 k32 k33
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x1 G1 R1
⎣
x1 = x2 ⎦, G1 = ⎣ G2 ⎦ , R1 = ⎣ 0 ⎦ (2.53)
x3 G3 R3
The mathematical model of the free-free beam 4, condensed to the coordinate vector
x6 x4 x5 x7 , is
m4 ẍ4 +k4 x4 = G4 +R4 (2.54)
where
⎡ ⎤ ⎡ ⎤
m44 m45 m46 m47 k44 k45 k46 k47
⎢ m54 m55 m56 m57 ⎥ ⎢ k54 k55 k56 k57 ⎥
m4 = ⎢
⎣ m64
⎥, k4 = ⎢ ⎥,
m65 m66 m67 ⎦ ⎣ k64 k65 k66 k67 ⎦
m74 m75 m76 m77 k74 k75 k76 k77
PROBLEMS 125
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x4 G4 R4
⎢ x5 ⎥ ⎢ G5 ⎥ ⎢ R5 ⎥
x4 = ⎢ ⎥
⎣ x6 ⎦ , G4 = ⎢ ⎥
⎣ G6 ⎦ , R4 = ⎢ ⎥
⎣ R6 ⎦ (2.55)
x7 G7 R7
The elements 2 and 3 may be considered as massless of stiffness k and damping
coefficient c. Develop the mathematical model of the system.
PROBLEMS 126

Solution.

x4 x5
x6 x7

Figure 7

Since the beam 4 is supported along coordinates x6 and x7 , mathematical

model of the supported beam 4 is
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸ ∙ ¸
m44 m45 ẍ4 k44 k45 x4 G4 R4
+ = + (2.56)
m54 m55 ẍ5 k54 k55 x5 G5 R5

2
x1
k c

x4

Figure 8

The mathematical model of the element 2 can be adopted in the following

form ∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸
k −k x1 c −c ẋ1 −R1
+ = (2.57)
−k k x4 −c c ẋ4 −R4

3
x3
k c

x5

Figure 9

Similarly, the mathematical model of the element 3 can be adopted as follows

∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸
k −k x3 c −c ẋ3 −R3
+ = (2.58)
−k k x5 −c c ẋ5 −R5
PROBLEMS 127

2 3

k
x1
c + k
x3
c

x4 x5

Figure 10

Hence, the combined mathematical model of element 2 and 3 (see Fig. 10) is
⎡ ⎤⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤
k −k 0 0 x1 c −c 0 0 ẋ1 −R1
⎢ −k k 0 0 ⎥ ⎢ x4 ⎥ ⎢ −c c 0 0 ⎥ ⎢ ẋ4 ⎥ ⎢ −R4 ⎥
⎢ ⎥⎢ ⎥+⎢ ⎥⎢ ⎥=⎢ ⎥
⎣ 0 0 k −k ⎦ ⎣ x3 ⎦ ⎣ 0 0 c −c ⎦ ⎣ ẋ3 ⎦ ⎣ −R3 ⎦ (2.59)
0 0 −k k x5 0 0 −c c ẋ5 −R5

After reorganization with respect to the vector of connecting coordinates x4 and x5

the above mathematical model takes form
⎡ ⎤⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤
k 0 −k 0 x1 c 0 −c 0 ẋ1 −R1
⎢ 0 k 0 −k ⎥ ⎢ ⎥ ⎢ 0 −c ⎥ ⎢ ⎥ ⎢ ⎥
⎢ ⎥ ⎢ x3 ⎥ + ⎢ 0 c ⎥ ⎢ ẋ3 ⎥ = ⎢ −R3 ⎥ (2.60)
⎣ −k 0 k 0 ⎦ ⎣ x4 ⎦ ⎣ −c 0 c 0 ⎦ ⎣ ẋ4 ⎦ ⎣ −R4 ⎦
0 −k 0 k x5 0 −c 0 c ẋ5 −R5

2 3

x1 x3
k c 4 k c

x4 x5
x6 x7

Figure 11

Composition of the above mathematical model with the sported beam 4 (Fig.
11) results in the following
⎡ ⎤⎡ ⎤ ⎡ ⎤⎡ ⎤
0 0 0 0 ẍ1 k 0 −k 0 x1
⎢ 0 0 0 0 ⎥ ⎢ ⎥ ⎢ −k ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ẍ3 ⎥ + ⎢ 0 k 0 ⎥ ⎢ x3 ⎥
⎣ 0 0 m44 m45 ⎦ ⎣ ẍ4 ⎦ ⎣ −k 0 k + k44 k45 ⎦ ⎣ x4 ⎦
0 0 m54 m55 ẍ5 0 −k k54 k + k55 x5
⎡ ⎤⎡ ⎤ ⎡ ⎤ ⎡ ⎤
c 0 −c 0 ẋ1 0 −R1
⎢ 0 c 0 −c ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
+⎢ ⎥ ⎢ ẋ3 ⎥ = ⎢ 0 ⎥ + ⎢ −R3 ⎥ (2.61)
⎣ −c 0 c 0 ⎦ ⎣ ẋ4 ⎦ ⎣ G4 ⎦ ⎣ 0 ⎦
0 −c 0 c ẋ5 G5 0
PROBLEMS 128

Reorganization of the mathematical model of the free-free beam 1 with respect to

the connecting coordinates x1 and x3 yields
⎡ ⎤⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
m22 m21 m23 ẍ2 k22 k21 k23 x2 G2 0
⎣ m12 m11 m13 ⎦ ⎣ ẍ1 ⎦ , + ⎣ k12 k11 k13 ⎦ + ⎣ x1 ⎦ = ⎣ G1 ⎦ + ⎣ R1 ⎦
m32 m31 m33 ẍ3 k32 k31 k33 x3 G3 R3
(2.62)

x1 x2 x3

x4 x5
x6 x7

Figure 12

Composition of the mathematical model 2.61 with 2.62 (Fig. 12) produces the
final mathematical model of the whole structure
⎡ ⎤⎡ ⎤ ⎡ ⎤⎡ ⎤
m22 m21 m23 0 0 ẍ2 0 0 0 0 0 ẋ2
⎢ m12 m11 m13 0 0 ⎥ ⎢ ⎥ ⎢ 0 −c 0 ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ẍ1 ⎥ ⎢ 0 c ⎥ ⎢ ẋ1 ⎥
⎢ m32 m31 m33 0 ⎥ ⎢ ⎥ ⎢
0 ⎥ ⎢ ẍ3 ⎥ + ⎢ 0 0 c ⎥ ⎢
0 −c ⎥ ⎢ ẋ3 ⎥
⎢ ⎥
⎣ 0 0 0 m44 m45 ⎦ ⎣ ẍ4 ⎦ ⎣ 0 −c 0 c 0 ⎦ ⎣ ẋ4 ⎦
0 0 0 m54 m55 ẍ5 0 0 −c 0 c ẋ5
⎡ ⎤⎡ ⎤ ⎡ ⎤
k22 k21 k23 0 0 x2 G2
⎢ k12 k11 + k k13 −k 0 ⎥ ⎢ x1 ⎥ ⎢ G1 ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥
+⎢⎢ k32 k31 k32 + k 0 −k ⎥ ⎢ ⎥ ⎢
⎥ ⎢ x3 ⎥ = ⎢ G3 ⎥
⎥ (2.63)
⎣ 0 −k 0 k + k44 k45 ⎦ ⎣ x4 ⎦ ⎣ G4 ⎦
0 0 −k k54 k + k55 x5 G5
 Part II
ANALYSIS

129
 Chapter 3
ANALYSIS OF MATHEMATICAL MODELS - GENERAL
REMARKS

The external force F in formula 2.38 can be always resolved into two parts.
mẍ + cẋ + kx = Fs + Fd (t) (3.1)
One of them Fs represents static external load applied to the system. It does not
depend on time and determines the equilibrium position of the rotor system. The
second one Fd (t) represents excitation of the rotor system caused by the residual
imbalance forces, the residual permanent deflection and the environment motion.
Since the equations 3.1 is linear, solution of the homogeneous equation represents
free motion of the system whereas the particular solution of the non-homogeneous
equation represents equilibrium position and the response to the external excitation.

3.1 EQUILIBRIUM POSITION

x
xs3

x
s4 x s1 x s5 x s6 x x s2

Figure 1

If one retain in the right hand side of equation 3.1 only the static load ( forces that
do not depend on time) Fs ,
mẍ + cẋ + kx = Fs (3.2)
its particular solution can be predicted as vector of constant magnitudes.
x = xs (3.3)
Introduction of Eq. 7.2 into Eq. 7.1 yields
xs = k−1 Fs (3.4)
RESPONSE TO EXTERNAL EXCITATION 131

This vector contains coordinates that determine the system equilibrium position with
respect to the adopted absolute frame xyz (Fig. 1). Around this equilibrium position
the system performs vibration due to the unbalance of the rotating parts, residual
permanent deflection and motion of its environment.

3.2 RESPONSE TO EXTERNAL EXCITATION

Response to the external excitation Fd (t) of the system considered is determined by
the particular solution of the mathematical model.

mẍ + cẋ + kx = Fd (t) (3.5)

The force Fd (t) can be represented by the harmonic series.

X
n=N X
n=N
Fd (t) = Fn = Fcn cos(ω n t) + Fsn sin(ωn t) (3.6)
n=1 n=1

The element k of the vector Fn can be assumed in the following form

Fnk = Fnko cos(ω n t + ϕkno ) (3.7)

To facilitate process of looking for the particular solution of 3.5, let us introduce
complex excitation force by adding to the expression 3.7 the imaginary part.

fnk = Fnko cos(ω n t + ϕkno ) + iFnko sin(ωn t + ϕkno ) (3.8)

Hence, according to Euler formula one can obtain

k k
fnk = Fnko ei(ωn t+ϕno ) = Fnko eiϕno eiωn t = fnko eiωn t (3.9)

where fnko is a complex number that depends on amplitude and phase of the external
excitation. Introduction of Eq. 3.9 into Eq. 3.5 yields

mẍ + cẋ + kx = fno eiωn t (3.10)

Now, the particular solution of Eq. 3.10 can be predicted in the complex form 3.11

xcn = an eiωn t (3.11)

Introduction of Eq. 3.11 into Eq. 3.10 produces set of algebraic equations linear with
respect to the unknown vector an .
¡ 2 ¢
−ωn m + iωn c + k an = fno (3.12)

Its solution is ¡ ¢− 1
an = −ω2n m + iω n c + k fno (3.13)
Therefore, according to Eq. 3.11, the response of the system xcn due to the complex
force fn is
xcn = (Re(an ) + i Im(an ))(cos ω n t + i sin ωn t) (3.14)
STABILITY OF MOTION - FREE VIBRATIONS 132

Hence, response of the system xn due to the real excitation Fn is the real part of the
solution 3.14.
xn = Re(an ) cos ωn t − Im(an ) sin ω n t (3.15)
Motion of the system considered along the coordinate xkn , according to 3.15 is

xkn = xkno cos(ωn t + β kno ) (3.16)

where
p Im(akn )
xkno = Re(akn )2 + Im(akn )2 β kno = arc tan (3.17)
Re(akn )

3.3 STABILITY OF MOTION - FREE VIBRATIONS

To analyze stability of motion of the system considered let us transfer the homoge-
neous equation of equation 3.1

mẍ + cẋ + kx = 0 (3.18)

to the state - space coordinates. Let

y = ẋ (3.19)

be the vector of generalized velocities. Introduction of 3.19 into 3.18 yields set of
differential equations of first order.

ẋ = y
ẏ = −m−1 kx − m−1 cy (3.20)

The above equations can be rewritten as follows

ż = Az (3.21)

where ∙ ¸ ∙ ¸
x 0 1
z= , A= (3.22)
y −m k −m−1 c
−1

Solution of the above equation can be predicted in form 3.23.

z = z0 ert (3.23)

Introduction of 3.23 into 3.21 results in set of the homogeneous algebraic equations
linear with respect to the vector zo .

[A − 1r] z0 = 0 (3.24)

Equations 3.24 have non-zero solution if and only if the characteristic determinant is
equal to 0.
|[A − 1r]| = 0 (3.25)
Process of searching for solution of equation 3.25 is called eigenvalue problem and
process of searching for the corresponding vector z0 is called eigenvector problem.
STABILITY OF MOTION - FREE VIBRATIONS 133

Both of them can be easily solved be means of commercially available computer

programs.
The roots rn are usually complex and conjugated.

rn = hn ± iω n n = 1.....N (3.26)

where N is number of coordinates (number of degree of freedom) in the final mathe-

matical model 2.38. The particular solutions corresponding to the complex roots 3.26
are

zn1 = ehn t (Re(zOn ) cos ωn t − Im(zOn ) sin ωn t)

zn2 = ehn t (Im(zOn ) cos ω n t + Re(zOn ) sin ω n t) n = 1.....N (3.27)

Re(z0n ) and Im(z0n ) are real and imaginary part of the complex and conjugated
eigenvector z0n associated with the roots 3.26 respectively. As one can see from the
formulae 3.26, the imaginary part of roots rn represents natural frequency of the
system free vibrations and its real part represents rate of decay of the free vibrations.
Graphical interpretation of these solutions is given in Fig. 2 for positive and
negative magnitude of hn .
If all roots rn of the equation 3.25 have their real part less then zero, the
system considered has stable equilibrium position. If at least one root has its real
part positive, the equilibrium position of the system considered is unstable.

30 z z
1
25
0.8
20
15
h>0 0.6 h<0
10 0.4

5 0.2
0 0
-5 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 -0.2 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
-10 -0.4
-15 Tn =2π /ωn -0.6
-20 Tn =2π /ωn
-0.8

Figure 2
PROBLEMS 134

3.4 PROBLEMS
Problem 12

x
m=50kg
2
Ixx = I
0.05 yy = I =15kgm
z z 2
I z =10kgm
b
b = 0.0001m
0.3 0.7 0.3 a = 0.65 m
a
x

Figure 3

0.05[m] ϕ ϕ ϕ ϕ
y6 y7 y14 y15

0.1 0.1 0.1 0.1 0.1 0.3 0.1 0.10.1 0.1 0.1 0.1 0.1 0.3 0.1 0.1 0.1 0.1 0.1 [m]

y y y y y y y y y y y y y y y y y y y y
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Figure 4

The drum shown in Fig. 3, can be considered as rigid body. The tolerance
of alignment of the holes is b = 0.0001[m]. This drum is connected to the flexible
shaft and supported as shown in Fig. 4. The dynamic properties of the shaft are as
follows:
d = 0.05m - diameter of the shaft
E = 0.2 × 1012 N/m2 - Young modulus
G = 0.0666 × 1012 N/m2 - shear modulus
ν = 0.333 - Poisson number
ρ = 7800kg/m3 - density
Create the mathematical model of the system and compute the forced response of
the system due to the misalignment b.
PROBLEMS 135

Solution.
The mathematical model of the unconstrained and free-free rotor can be
adopted as follows
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸ ∙ ¸
m 0 ẍ k 0 x Rx 0
+ = + (3.28)
0 m ÿ 0 k y Ry G

where ⎡ ⎤ ⎡ ⎤
mdd mdi mdr kdd kdi kdr
m = ⎣ mid mii mir ⎦ ; k = ⎣ kid kii kir ⎦
mrd mdi mrr krd kdi krr
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
xd yd Rxd Ryd Gd
x = ⎣ xi ⎦ ; y = ⎣ yi ⎦ ; Rx = ⎣ Rxi ⎦ ; Ry = ⎣ Ryi ⎦ ; G = ⎣ Gi ⎦
xr yr Rxr Ryr Gr
(3.29)

xd = [x7 , ϕx7 , x14 , ϕx14 , x15 , ϕx15 ]T

£ ¤T - vectors of the dependent coordinates
yd = y7 , ϕy7 , y14 , ϕy14 , y15 , ϕy15
xi = [x6 , ϕx6 ]T
£ ¤T - vectors of the independent coordinates
yi = y6 , ϕy6
xr = [x1 , x2 , x3 , x4 , x5 , x8 , x9 , x10 , x11 , x12 , x13 , y16 , x17 , x18 , y19 , y20 ]T
- vectors of the
yr = [y1 , y2 , y3 , y4 , y5 , y8 , y9 , y10 , y11 , y12 , y13 , y16 , y17 , y18 , y19 , y20 ]T
not engaged coordinates.
Solution of the eigenvalue and eigenvector problem for the above mathematical
model results in the following natural modes and natural frequencies of the uncon-
strained and free - free rotor.

α = 266.9 rad/s α =735.5rad/s

a) b)

Figure 5

According to Fig. 3, the constraint equations can be adopted in the following

form:
y7 = y6 + 0.3ϕy6
ϕy7 = ϕy6
y14 = y6 + (0.3 + 0.7)ϕy6 + 0.0001
(3.30)
ϕy14 = ϕy6
y15 = y6 + (0.3 + 0.7 + 0.3)ϕy6 + 0.0001
ϕy15 = ϕy6
PROBLEMS 136

or in the matrix form

yd = adi yi + ady (3.31)
where ⎡ ⎤ ⎡ ⎤
1 0.3 0
⎢ 0 1 ⎥ ⎢ 0 ⎥
⎢ ⎥ ⎢ ⎥
⎢ 1 1 ⎥ ⎢ 0.0001 ⎥
adi = ⎢
⎢
⎥, ady =⎢ ⎥ (3.32)
⎢ 0 1 ⎥ ⎥
⎢
⎢ 0 ⎥
⎥
⎣ 1 1.3 ⎦ ⎣ 0.0001 ⎦
0 1 0
One may assume that in plane xz the misalignment is equal to zero. Hence,
adx = 0 (3.33)
According to considerations in sub-section 1.2.6 (page 55), the equations of motion
of the shaft after introduction of the constraints 3.31 and 3.33 are
∙ ¸∙ ¸ ∙ ¸∙ ¸
mc 0 ẍc kc 0 xc
+
0 mc ∙ ÿc ¸ ∙ 0 k ¸ c ∙ ycc ¸ ∙ s ¸ (3.34)
Rxc 0 Fxc Fxc
= + + cos Ωt + sin Ωt
Ryc Gc Fcyc Fsyc
where ∙ ¸ ∙ ¸
xi yi
xc = ; yc =
xr yr
Fig. 6 shows two first natural modes of the shaft (without the drum) in the plane yz
after the constraint equations have been introduced.

a) b)

α = 1027 rad/s α = 1427 rad/s

Figure 6

According to Eq. 1.17 (page 17), the equation of motion of the drum along
the independent coordinates xi and yi are
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸ ∙ ¸
md 0 ẍi 0 cd ẋi Rxi 0
+ = + (3.35)
0 md ÿi −cd 0 ẏi Ryi Gd
where
∙ ¸ ∙ ¸ ∙ ¸
m ma 0 0 mg
md = , cd = , Gd = (3.36)
ma I + ma2 0 Iz Ω mga
PROBLEMS 137

Application of the composition rules to the equation 3.34 and 3.34 yields equation of
motion of the free-free shaft - drum mathematical model.
∙ ¸ ∙ ¸ ∙ ¸
ẍc ẋc xc
mcd +g + kcd
ÿc ∙ cd ẏ ¸c ∙ ¸ yc ∙ c ¸ ∙ s ¸ (3.37)
Rxcd 0 Fxc Fxc
= + + cos Ωt + sin Ω
Rycd Gcd Fcyc Fsyc
The natural modes of the free-free shaft - drum sub-system are shown in Fig. 7.

a) b)

α = 937 rad/s α = 1069 rad/s

Figure 7

Introduction of the boundary conditions (see page 54) results in the following
mathematical model.
∙ ¸ ∙ ¸ ∙ ¸ ∙ ¸ ∙ c ¸ ∙ s ¸
ẍc ẋc xc 0 Fx Fx
m +g +k = + cos Ωt + sin Ωt (3.38)
ÿc ẏc yc G Fcy Fsy
The natural modes and the natural frequencies of the supported shaft-drum sub-
system are shown in Fig. 8.

α = 420.4 rad/s

α =194.3 rad/s

a) b)

Figure 8

The particular solution of the following non-homogeneous equation

∙ ¸ ∙ ¸ ∙ ¸ ∙ ¸
ẍc ẋc xc 0
m +g +k = (3.39)
ÿc ẏc yc G
PROBLEMS 138

represents the static deflection line due to the gravity forces. It is shown in Fig. 9.

1x10 -4
static
deflection
[m] 2x10 -4

Figure 9

Around this equilibrium position the rotor performs the forced vibration due
to the non zero tolerance b (see Fig. 3). The forced vibrations, due to the inac-
curacy of the drum, are approximated by the particular solution of the following
non-homogeneous equation.
∙ ¸ ∙ ¸ ∙ ¸ ∙ c ¸ ∙ s ¸
ẍc ẋc xc Fx Fx
m +g +k = c cos Ωt + sin Ωt (3.40)
ÿc ẏc yc Fy Fsy

Amplitude of the forced vibration along the coordinate y6 (see Fig. 4) as a function
of the angular velocity Ω is presented in Fig. 10.

10

amplitude x 10 4 [m]

0
100 200 300 400 500 600
Ω [rad/s]

Figure 10
PROBLEMS 139

Due to the geometrical symmetry of the system considered, the first critical
speed (Ωcr1 = 194.3[rad/s]) coincides with the first natural frequency whereas the
second critical speed ( Ωcr2 = 568[rad/s] ) is due to the gyroscopic effect, higher.
PROBLEMS 140

Problem 13

φ 0.08 φ 0.05 φ 0.04 φ 0.1 φ 0.04 φ 0.06 φ 0.05

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

a)
24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46

b)

c)
0.1
2.4

1 2 0.05 3 4 z
d)

Figure 11

Two beams: beam a and beam b, shown in Fig. 11a) and Fig. 11b) respec-
tively, are joined together along the coordinates y24 by the universal joint (see Fig.
11c)). The dynamic properties of these beams are as follows
E = 0.2 × 1012 N/m2 - Young modulus
G = 0.0666 × 1012 N/m2 - shear modulus
ν = 0.333 - Poisson number
ρ = 7800kg/m3 - density
The assembly is supported upon four supports shown in Fig .2ad). Motion of the
supports with respect to the inertial frame yz is as follows

y1 = 0.005 cos 100t

y7 = −0.003 cos 100t
(3.41)
y30 = 0.05
y44 = 0.05

Compute:
1. Static equilibrium position of the rotor system
2. Natural modes and corresponding natural frequencies
3. Response of the rotor system due to motion of the supports:
PROBLEMS 141

Solution.
The mathematical model of the beams a and b are as follows
∙ a ¸∙ a ¸ ∙ a ¸∙ a ¸ ∙ a ¸
m11 ma12 ÿ1 k11 ka12 y1 G1
+ = (3.42)
ma21 ma22 a
ÿ24 ka21 k22
a a
y24 Ga24
∙ b ¸∙ b ¸ ∙ b ¸∙ b ¸ ∙ b ¸
m11 mb12 ÿ24 k11 kb12 y24 G24
b b b + b b b = (3.43)
m21 m22 ÿ2 k21 k22 y2 Ga2
where:

y1a = [y1 , y2 , ...., y22 , y23 ]T , y2b = [y25 , y26 , ...., y43 , y44 ] (3.44)
Natural modes and the corresponding natural frequencies of the beam a and beam
b are shown in Fig. 12 and 13 respectively. Application of the composition rules

ω 1=230.3rad/s

ω 2=757.8rad/s

Figure 12

yields mathematical model of the two beams joined as shown in Fig. 11c)
⎡ a ⎤⎡ ⎤ ⎡ a ⎤⎡ ⎤ ⎡ ⎤
m11 ma12 0 ÿ1 k11 ka12 0 y1 Ga1
⎣ ma21 ma22 + mb11 mb12 ⎦ ⎣ ÿ24 ⎦+⎣ ka21 k22
a
+ k11b
kb12 ⎦ ⎣ y24 ⎦ = ⎣ Ga24 + Gb24 ⎦
0 mb21 mb22 ÿ2 0 kb21 kb22 y2 Gb2
(3.45)
Fig. 14 presents the natural modes and the corresponding natural frequencies of the
sub-system that is governed by equation 3.45.
PROBLEMS 142

ω 1=142.8rad/s
ω =217.9rad/s
2

Figure 14

After reorganization of equations 3.45 with respect to the supported coordi-

nates
y3 = [y1 , y7 , y30 , y44 ] (3.46)
they can be rewritten as follows
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸ ∙ ¸
m33 m34 ÿ3 k33 k34 y3 G3 R3
+ = + (3.47)
m43 m44 ÿ4 k43 k44 y4 G4 0

Boundary conditions for the equation 3.47 can be formulated as follows

y3 = b0 + bc cos 100t (3.48)

where ⎡ ⎤ ⎡ ⎤
0 0.005
⎢ 0 ⎥ ⎢ −0.003 ⎥
b0 = ⎢ ⎥
⎣ 0.05 ⎦ , bc = ⎢
⎣
⎥
⎦ (3.49)
0
0.05 0
Introduction of the boundary conditions 3.48 into the equations 3.47 yields the math-
ematical model of the supported system.

−m33 bc 1002 cos 100t + m34 ÿ4 + k33 bo + k33 bc cos 100t + k34 y4 − G3 = R3 (3.50)

m44 ÿ4 + k44 y4 = G4 − k43 b0 − k43 bc cos 100t + m43 bc 1002 cos 100t (3.51)
Solution of the homogeneous equation 3.51

m44 ÿ4 +k44 y4 = 0 (3.52)

offers the natural modes and the natural frequencies that are shown in Fig. 15.
PROBLEMS 143

ω 1 =113.9rad/s

ω 2 =388.6rad/s

Figure 15

The particular solution of equation:

m44 ÿ4 +k44 y4 = G4 −k43 b0 (3.53)

produces equilibrium position of the system considered. It is shown in Fig. 16. The
particular solution of equation

m44 ÿ4 +k44 y4 = −k43 bc cos 100t+m43 bc 1002 cos 100t (3.54)

allows amplitudes of the forced vibrations due to the environmental excitation to be

computed. They are shown in Fig. 17.
PROBLEMS 144

ω 1 =215.8rad/s
ω 2 =719.1rad/s

Figure 13

0.00

0.01

0.02

0.03
0.04

0.05
0.06

Figure 16
PROBLEMS 145

Problem 14
To model the rotor system shown in Fig. 18a), the system has been split into
two sub-systems shown in Fig. 18b).

a)
1

z Φ 0.02
y11 y21 y31 y41 y51
2 b)

a z
y32 y42 y52 Φ 0.04
Φ 0.05
Φ 0.06

0.04 0.01 0.03

mass of bearings 3 - 0.15 [kg]

missalignment a=0.00001 [m]
Young modulus E= 0.2 x 10 12 [N/m 2 ]
Shear modulus G=0.066 x 10 12 [N/m 2 ]
dencity d=7800 [kg/m 3]
all dimension are in meters

Figure 18

The mathematical model of the free-free shaft 1 (together with bearings 3 )

has been condensed along coordinates to the nodes 1,2,3,4,5. Along coordinates
x11 , x12 , x13 , x14 , x15 and y11 , y21 , y31 , y41 , y51 the mathematical models are respectively
∙ 1 ¸∙ 1 ¸ ∙ 1 ¸∙ 1 ¸ ∙ 1 ¸
m11 m112 ẍ1 k11 k112 x1 R1x
+ = (3.55)
m121 m122 ẍ12 k121 k122 x12 R12x
∙ 1 ¸∙ 1 ¸ ∙ 1 ¸∙ 1 ¸ ∙ 1 ¸∙ 1 ¸
m11 m112 ÿ1 k11 k112 y1 G1 R1y
+ = (3.56)
m121 m122 ÿ21 k121 k122 y21 G12 R12y
PROBLEMS 146

where: ⎡ ⎤ ⎡ ⎤
∙ ¸ x13 ∙ ¸ y31
x11 y11
x11 = , x12 = ⎣ x14 ⎦ , y11 = , y21 = ⎣ y41 ⎦ (3.57)
x12 y21
x15 y51
The mathematical model of the shaft 2, condensed to the nodes 3,4,5, is as follows

m2 ẍ2 +k2 x2 = R2x (3.58)

m2 ÿ2 +k2 y2 = G2 +R2y (3.59)

where: ⎡ ⎤ ⎡ ⎤
x23 y32
x2 = ⎣ x24 ⎦ , y2 = ⎣ y42 ⎦ (3.60)
x25 y52
The tolerance of manufacturing housings of the bearings 3 is a (see Fig. 18b)).
Hence the vector of the permanent deformation in the plane xz an yz may be assumed
in the following form ⎡ ⎤ ⎡ ⎤
0 0
ax = 0 , ay = a ⎦
2 ⎣ ⎦ 2 ⎣ (3.61)
0 0
The shaft 2 rotates with the angular velocity Ω and the shaft 1 is stationary. Develop
mathematical model of the system.
PROBLEMS 147

Solution.
Since the shaft 1 is supported along coordinates x11 , let us partition the equa-
tions 3.55 and with respect to this vector and introduce the boundary conditions
x11 = 0.

m112 ẍ12 +k112 x12 = R11x

m122 ẍ12 +k122 x12 = R12x (3.62)

A similar treatment of equations 3.56 yields

m112 ÿ21 +k112 y21 = G11 +R11y

m122 ÿ21 +k122 y21 = G12 +R12y (3.63)

The lower equations 3.62 and 3.63 form the mathematical model of the supported
shaft 1.

m122 ẍ12 +k122 x12 = R12x

m122 ÿ21 +k122 y21 = G12 +R12y (3.64)

At this stage of composition they are not coupled and the particular solution of

m122 ÿ21 +k122 y21 = G12 (3.65)

yields static deflection of the shaft 1 and the general solution produces its natural
frequencies and natural modes. They are shown in Fig. 19 and Fig. 20 respectively.
The shaft 2 should be considered as permanently deflected shaft. The permanent
deformation (see page 49) causes vector of external excitation determined by the
formula 1.152. In the case considered this vector of excitation is

Fx = −k2 a2y sin Ωt

Fy = k2 a2y cos Ωt (3.66)

and the mathematical model of the shaft 2 takes the following form

m2 ẍ2 +k2 x2 = −k2 a2y sin Ωt + R2x

m2 ÿ2 +k2 y2 = G2 + k2 a2y cos Ωt + R2y (3.67)

Solution of the homogeneous equation 3.67 yields the natural frequencies and natural
modes. They are shown in Fig 21. Composition of the two mathematical models
3.64 and 3.67, results in mathematical model of the rotor system.

(m122 +m2 )ẍ + (k122 +k2 )x = −k2 a2y sin Ω2x t

(m122 +m2 )ÿ + (k122 +k2 )y = G12 +G2 +k2 a2y cos Ωt (3.68)

The particular solution of equation

(m122 + m2 )ÿ + (k22

1
+ k2 )y = G12 + G2 (3.69)
PROBLEMS 148

amplitudes of the kinematic

excitation
amplitudes
of
vibrations
[m] x 10 2

Figure 17

0.0

y [m]

1.0 x10 -4

0.0 0.1 0.2 0.3 0.4 0.5 z [m]

Figure 19
PROBLEMS 149

1.0

338.7 [rad/s]
2073 [rad/s]

0.0

-0.1
0.0 0.1 0.2 0.3 0.4 0.5 z [m]

Figure 20

1.0
17920 [rad/s]

41720 [rad/s]

0.0

-1.0
0.2 0.3 0.4 0.5 z [m]

Figure 21
PROBLEMS 150

represents static deflection line in the vertical plane yz. It is shown in Fig. 22.

0.0

1.0 x 10 -3

2.0 x 10 -3

0.0 0.1 0.2 0.3 0.4 0.5 z [m]

Figure 22

Solution of the homogeneous equation offers the natural frequencies and the
natural modes. The first three natural modes and the corresponding frequencies are
shown in Fig. 23a),b),c) respectively.
If in the right hand side of equation 3.68 one keeps the time dependent terms
only,

(m122 +m2 )ẍ + (k122 +k2 )x = −k2 a2y sin Ω2x

(m122 +m2 )ÿ + (k122 +k2 )y = k2 a2y cos Ωt (3.70)

its particular solution represents the forced vibrations of the assembly due to the
manufacturing tolerances. Amplitude of these forced vibrations ,along coordinates
y3 , y4 and y5 , as a function of the angular velocity of the rotor is presented in Fig.
24.
PROBLEMS 151

1.0 1.0

b)
a) 1653 [rad/s]
87.65 [rad/s]
0.0 0.0

-1.0 -1.0
0.1 0.2 0.3 0.4 0.5 z [m] 0.1 0.2 0.3 0.4 0.5 z [m]
1.0

c)
11920 [rad/s]

0.0

-1.0
0.1 0.2 0.3 0.4 0.5 z [m]

Figure 23

1.0 x 10 -4

y3

A [m] y5
y4

0.0
0.0 Ω [rad/s] 10 000

Figure 24
 Chapter 4
ANALYSIS OF ROTOR WITH ONE CONCENTRATED MASS

4.1 PHYSICAL MODEL

S E

C
G
Ω
µ

Figure 1

R
O Y Z
X
rC
C C
X rGC Ω
rG Ωt G
µ

Fs
Y
Y

Figure 2

Let us consider rotor shown in Fig. 1which by assumption rotates with a constant
angular speed Ω. The shaft S of the rotor is supported rigidly at its ends. Assume
that the shaft can be considered massless and flexible whereas the element E can
be approximated by a particle of mass m. This particle is attached to the shaft at
the centre of gravity G of the element E. The centre of gravity G is displaced by µ
from the geometrical centre of the shaft cross-section C. The distance µ represents
imbalance of the element E and can be considered as a small magnitude. To analyze
motion of this system let us introduce the inertial system of coordinates XY Z as it
is shown in Fig. 2. The instantaneous position of the centre C is determined by the
position vector rC . The centre of gravity G rotates with respect to this centre with the
angular velocity Ω. Since the angular velocity is constant, the relative instantaneous
MATHEMATICAL MODEL 153

position of the centre of gravity G is determined by the angle Ωt and the imbalance
µ (vector rGC ). The absolute position of the centre of gravity G, in Fig. 2 is denoted
by rG . The vector Fs represents the static resultant force acting on the element E. R
stands for the interaction force between the element considered and the shaft.

4.2 MATHEMATICAL MODEL

Motion of the centre of gravity G is governed by the Newton law

mr̈G = R + Fs (4.1)

where according to Fig. 2

rG = I(X+µ cos Ωt) + J(Y +µ sin Ωt)

R = −IkX − JkY
Fs = IFXs + JFY s (4.2)

In the above formula k stands for stiffness of the shaft at the point C and XY are
its coordinates. Introduction of Eq’s. 4.2 into Eq. 4.1, results in the following set of
differential equations.

m(Ẍ−µΩ2 cos Ωt) = −kX + FXs

m(Ÿ −µΩ2 sin Ωt) = −kY + FY s (4.3)

or after reorganization

mẌ + kX = FXs + mµΩ2 cos Ωt

mŸ + kY = FY s + mµΩ2 sin Ωt (4.4)

The particular solution of the non-homogeneous Eq. 4.5

mẌ + kX = FXs
mŸ + kY = FY s (4.5)

yields the equilibrium position (Xs , Ys ). Upon assuming the particular position in
form

X = Xs
Y = Ys (4.6)

one may obtain the following formulae for coordinates of the equilibrium position
which are usually referred to as the static deflection of the shaft.
FXs FY s
Xs = , Ys = (4.7)
k k
MATHEMATICAL MODEL 154

The total deflection of the shaft X,Y are sum of the static deflection Xs , Ys and the
dynamic deflection x,y (see Fig. 3).

X = Xs + x
Y = Ys + y (4.8)

Ys
O X
x
o y
Xs x C
G Ω t
µ

Y
y

Figure 3

Introduction of Eq. 4.8 into the mathematical model 4.4 in equations which
govern the dynamic deflections xy.

mẍ + kx = mµΩ2 cos Ωt

mÿ + ky = mµΩ2 sin Ωt (4.9)

or

ẍ + ω 2 x = q cos Ωt (4.10)
ÿ + ω 2 y = q sin Ωt (4.11)

where r
k
ω= , q = µΩ2 (4.12)
m
Upon multiplying the equation 4.11 by the imaginary unit i and adding the equations
4.10and 4.11, one may obtain the equations of motion of the rotor in the following
form
z̈ + ω 2 z = qeiΩt (4.13)
where
z = x + iy (4.14)
The above equation governs motion of the rotor in the stationary system of coordi-
nates xyz.
MATHEMATICAL MODEL 155

Let us introduce the rotating system of coordinates xR , yR , zR , shown in Fig.

4.

O
z
ϕ
Ωt x
rC
C
xR

i yR iy

Figure 4

Axis zR coincides axis z and axes xR and yR rotates with the constant angular
velocity Ω. In terms of the complex notations, position of the point C in the stationary
system of coordinates xyz is
z = |z| eiϕ (4.15)
and in the rotating system of coordinates is

zR = |zR | ei(ϕ−Ωt) = |zR | eiϕ e−iΩt (4.16)

Introduction of Eq. 4.15 into Eq. 4.16 yields the relationship between coordinates of
the same point in the stationary (x, iy) and the rotating (xR ,yR ) system of coordi-
nates.
zR = ze−iΩt (4.17)
The inverse transformation is
z = zR e+iΩt (4.18)
Differentiating of Eq. 4.18 with respect to time one can obtain

ż = żR eiΩt + zR iΩeiΩt

z̈ = z̈R eiΩt + 2żR iΩeiΩt − zR Ω2 e+iΩt (4.19)

Introduction of Eq. 4.19 into the mathematical model 4.10 produces equation of
motion of the rotor in terms of the rotating system of coordinates.

z̈R + 2żR iΩ + zR (ω 2 − Ω2 ) = q (4.20)

ANALYSIS OF MOTION IN TERMS OF THE STATIONARY COORDINATES SYSTEM 156

4.3 ANALYSIS OF MOTION IN TERMS OF THE STATIONARY CO-

ORDINATES SYSTEM
The equation 4.13 can be classified as the linear non-homogeneous differential equa-
tion. Its solution is a superposition of the general solution of homogeneous equation
zg and the particular solution of non-homogeneous equation zp . Since the excitation
force qeiΩt in the homogeneous equation is omitted, the general solution of the homo-
geneous equation represents so called free vibrations of the rotor. The general solution
of the non-homogeneous equation z = zg + zp represents so called forced vibration of
the rotor. The particular solution zp represents steady state forced vibration.
4.3.1 Free vibrations (general solution of the homogeneous equation)
The particular solution of the homogeneous equation 4.13
z̈ + ω 2 z = 0 (4.21)
can be predicted in the following form.
z = aert (4.22)
Introduction of Eq. 4.22 into 4.21 yields the characteristic equation.
r2 = −ω2 (4.23)
Its roots
r = ±iω (4.24)
allow to formulate the general solution.
zg = Aeiωt + Be−iωt (4.25)
Since A and B are complex, the solution 4.25 can be rewritten as follows.
zg = |A| eia eiωt + |B| eib e−iωt = |A| ei(a+ωt) + |B| ei(b−ωt) (4.26)
The constant magnitude of |A| , |B| , a and b have to be chosen to fulfil the following
initial conditions.
z(0) = zo , ż(0) = żo (4.27)
If |B| = 0, it is said that the rotor performs positive (simple) precession. If
|A| = 0, it is said that the rotor performs negative (inverse) precession. In a general
case, the free vibration, as one can see from 4.26), are superposition of both, the
negative and positive precession. The geometrical interpretation is given in Fig. 5.

O x

vector performing positive precession

a
vector performing negative precession
ωt
|A| Ω
|B| C
ωt
iy b

Figure 5
ANALYSIS OF MOTION IN TERMS OF THE STATIONARY COORDINATES SYSTEM 157

The parameter ω is called the natural frequency of the rotor in the stationary
system of coordinates. In the case considered, the natural frequency of the rotor is
independent of its angular velocity Ω as it is shown in Fig. 6.

0
Ω
−ω

Figure 6

4.3.2 Forced vibrations (particular solution of the non-homogeneous equa-

tion)
The particular solution of the non-homogeneous equation of motion
z̈ + ω 2 z = qeiΩt (4.28)
may be predicted as following
z = CeiΩt (4.29)
Introduction of Eq. 4.29 into Eq. 4.28 yields the following linear algebraic equation
−Ω2 C + ω 2 C = q (4.30)
Hence, the constant C is
q µΩ2
C= 2 = 2 (4.31)
ω − Ω2 ω − Ω2
Magnitude of C is called amplitude of forced vibration. According to 4.29, the centre
of the shaft moves along the circle of radius C performing the positive precession with
the angular velocity Ω (see Fig. 7).

o
Ωt x
C

iy

Figure 7
ANALYSIS OF MOTION IN TERMS OF THE STATIONARY COORDINATES SYSTEM 158

The relationship between the amplitude of the forced vibrations C and the
angular velocity Ω is given in Fig. 8.

12
10
8
C /µ 6
4
2 Ω/ω
0.5 1 1.5 2 2.5 3
0

-2
-4
-6
-8
-10
-12

Figure 8

Angular velocity of the rotor for which the amplitude of the forced vibrations
reaches its local or absolute maximum is called critical speed. In the case considered
the critical speed Ωcr is equal to the natural frequency ω.

C
C
µ
G
µ G
o o
o
x x
C x
Ωt Ωt
Ωt
µ G
iy iy
iy
a) b) c)

Figure 9

The instantaneous position of the centre of gravity G and the centre of the
shaft C for Ω < Ωcr , Ω > Ωcr , and Ω = ∞ are shown in Fig. 9a)b) and c) respectively.
ANALYSIS OF MOTION IN TERMS OF THE ROTATING COORDINATES SYSTEM 159

4.4 ANALYSIS OF MOTION IN TERMS OF THE ROTATING COOR-

DINATES SYSTEM
4.4.1 Free vibrations
According to Eq. 4.20 the free motion of the rotor in terms of coordinates of the
rotating axes is
z̈R + 2żR iΩ + zR (ω 2 − Ω2 ) = 0 (4.32)
The particular solution may be predicted in the following form.
zR = AR eirt (4.33)
Introduction of Eq. 4.33 into 4.32 yields the characteristic equation 4.34
−r2 − 2Ωr + (ω2 − Ω2 ) = 0 (4.34)
Its roots are determined by the following expression.
p
2Ω ± 4Ω2 + 4(ω 2 − Ω2 )
r= = −Ω ± ω (4.35)
−2
Taking into account Eq. 4.33, represents the natural frequency of the rotor in the
rotating system of coordinates.. Hence, the natural frequency with respect to the
rotating system are
ω R = ±ω − Ω (4.36)
The graphical interpretation of the expression 4.36 is given in Fig. 10.

ωR

0 Ω
−ω

Figure 10

4.4.2 Forced vibrations

The steady state forced vibrations are represented by the particular solution of the
non-homogeneous equation 4.20. It may be sought in form
zR = C (4.37)
Introduction of 4.37 into 4.20 yields the expression 4.31 for the amplitude of forced
vibrations which is rewritten here.
q µΩ2
C= = (4.38)
ω 2 − Ω2 ω 2 − Ω2
INFLUENCE OF DAMPING 160

The last formula indicate that if the rotor performs the steady state motion , the
points C and G are motionless with respect to the rotating system of coordinates.
Their positions for Ω < ω, Ω > ω and Ω = ∞ are shown in Fig. 11a)b) and c)
respectively.

o C G C G o C µ G
µ x µ x o x
t

iy iy iy
a) b) c)

Figure 11

4.5 INFLUENCE OF DAMPING

There are two main causes of dissipation of energy due to the rotation and vibratory
motion of rotors.
1. Friction of the rotor against the surrounding medium. The mathematical
model of dissipation of energy caused by this kind of friction is known as the external
damping.
2. Friction between particles the rotor is made up. Mathematical model of
dissipation of energy caused by the friction between the individual particles is called
internal damping.

o x
y
Ω
C cex
x
cex

Figure 12

The damping force which acts on the shaft, reflecting the external damping,
can be modelled to be proportional to the translational motion of the centre of the
shaft with respect to the stationary system of coordinates (see Fig. 12).
x
Fex = −cex ẋ
y
Fex = −cex ẏ (4.39)
INFLUENCE OF DAMPING 161

In terms of the complex notation the above forces take form

Fex = −cex ż (4.40)

Introduction of Eq. 4.40 into equation of motion 4.13 yields

z̈ + 2hex ż + ω2 z = qeiΩt (4.41)

where:
cex
2hex = (4.42)
m
Forces due to the internal damping can be modelled as proportional to the
rate of change of the curvature of the shaft as it can be seen from the rotating system
of coordinates (see Fig. 13).

Ys
O X
o x
y
Xs x C
G Ω t
µ
yR

Y xR
y

cin

Figure 13

One may assume that the instantaneous curvature κ in the plane XZ and Y Z
of the stationary system of coordinates is proportional to the absolute deflection in
X and Y direction.

κX ≈ X = Xs + x
κY ≈ Y = Ys + y (4.43)

Using the complex notations the curvature can be rewritten as follows.

κz ≈ Z = Zs + z (4.44)
INFLUENCE OF DAMPING 162

This curvature with respect to the rotating system of coordinates, according to 4.17
is
κzR = κz e−iΩt ≈ (Zs + z)e−iΩt (4.45)
Hence, the damping force along the rotating axes is
d
FinR = −cin ((Zs + z)e−iΩt ) = −cin (−iΩZs − iΩz + ż)e−iΩt (4.46)
dt
Therefore, the damping force in terms of the stationary system of coordinates is
Fin = FinR eiΩt = −cin (−iΩZs − iΩz + ż) (4.47)
Introduction of Eq. 4.47 into equation of motion 4.41 yields
z̈ + 2hex ż + 2hin (−iΩZs − iΩz + ż) + ω2 z = qeiΩt (4.48)
where:
cin
2hin = (4.49)
m
The above equation determine motion of the rotor in case when both type of damping
is included.
4.5.1 Analysis of influence of the external damping
For hin = 0 the equations of motion take the following form
z̈ + 2hex ż + ω2 z = qeiΩt (4.50)

Free vibrations
If one assume that the imbalance µ is equal zero (q = 0), the above mathematical
model describe the free motion of the rotor.
z̈ + 2hex ż + ω2 z = 0 (4.51)
Upon adopting the particular solution of the homogeneous equation 4.51 in form
z = Aert (4.52)
the characteristic equation
r2 + 2hex r + ω2 = 0 (4.53)
for hex < ω (case of small damping) yields the following roots.
r = −hex ± iλex (4.54)
where p
λex = ω 2 + h2ex (4.55)
Since hex is always positive, the real part of roots r is negative. Therefore, the equilib-
rium position (Xs , Ys ) as well as all solutions of the equation 4.51 are asymptotically
stable. They have the following form
zg = Ae−hex t eiλex t + Be−hex t e−iλex t (4.56)
INFLUENCE OF DAMPING 163

As one can see from 4.56 the imaginary part of the characteristic roots λex represents
the natural frequency of the rotor. The complex constants A and B must be chosen
to fulfil initial conditions which have in general, the following form.

z(0) = z0 = x0 + iy0 , ż(0) = ż0 = ẋ0 + iẏ0 (4.57)

From 4.56 one can see that the instantaneous position of the centre of the shaft
is determine by two vectors. One of them performs the positive precession with
amplitude Ae−hex t . The other one performs the negative precession with amplitude
Be−hex t . Both amplitudes decay to zero if time approaches infinity. The geometrical
interpretation of the solution 4.56 is give in Fig. 14.

x
o
Ω -Ω
A e -hex t
Be -hex t

Figure 14

Forced vibrations
The forced vibrations are determine by the particular solution of the non-homogeneous
equation Eq. 4.50. This solution can be predicted in the following form

zp = CeiΩt (4.58)

Introduction of Eq. 4.58 into Eq. 4.50 yields the algebraic equation which permits
the unknown constant C to be determined.

(−Ω2 + ω 2 + 2ihex Ω)C = q (4.59)

Hence,
q
C= = |C| eiϕ (4.60)
−Ω2 + ω2 + 2ihex Ω
where
q −2hex Ω
|C| = p , ϕ = arctan (4.61)
(ω − Ω )2 + 4h2ex Ω2
2 2 (ω2 − Ω2 )2
The final form of the particular solution can be obtained by introduction of 4.60 into
4.58
zp = |C| ei(Ωt+ϕ) (4.62)
INFLUENCE OF DAMPING 164

It indicates that the centre of the shaft performs positive precession around circle of
radius |C|. The centre of gravity G stays on parallel to the rotating axis xR line (see
Fig. 15)

x
o
Ω t
C
ϕ
yR G
x
R
y

Figure 15

The amplitude of the forced vibrations |C| and the phase ϕ as a function of
the angular velocity Ω are presented in Fig. 16.

12
for h ex/ω = 0.0
10
for h ex/ω = 0.05
8
for h ex/ω = 0.2
C /µ 6
for h ex/ω = 1
4
for h ex/ω = 2
2
0
0 0.5 1 1.5 2 2.5 3
Ω/ω
0
-0.5 0.5 1 1.5 2 2.5 3

-1
-1.5 π / 2
ϕ -2
[rad]-2.5
-3 π
-3.5

Figure 16
INFLUENCE OF DAMPING 165

4.5.2 Analysis of the internal damping

For hex = 0 the equation of motion 4.48 yields

z̈ + 2hin (−iΩZs − iΩz + ż) + ω 2 z = qeiΩt (4.63)

For further analysis let as assume that the static forces are due to gravity only and
therefore the static deflection exists only in plane Y Z (see Fig. 17).

O
Ys X

x
o y
x C
G Ω t
µ
yR

Y y xR

Figure 17

Zs = iYs + 0 (4.64)
Introduction of Eq. 4.64 into results in the following equation of motion

z̈ + 2hin (ΩYs − iΩz + ż) + ω 2 z = qeiΩt (4.65)

For q = 0 the above equation describes the free vibration of the shaft.

Free vibration.
According to 4.65, motion of the perfectly balanced rotor (q = 0) is govern by the
following equation.

z̈ + 2hin ż − 2ihin Ωz + ω2 z = −2hin ΩYs (4.66)

Solution of the above equation has two parts: the general solution of the homogeneous
equation zg and the particular solution of the non-homogeneous equation zp .

z = zg + zp (4.67)

The particular solution of the homogeneous equation

z̈ + 2hin ż − 2ihin Ωz + ω 2 z = 0 (4.68)

INFLUENCE OF DAMPING 166

can be predicted as
z = Arrt (4.69)
Introduction of Eq. 4.69 into 4.68 results in the following characteristic equation

r2 + 2hin r + ω2 − 2ihin Ω = 0 (4.70)

Its roots are:

q p
r = −hin ± h2in − ω 2 + 2ihin Ω = −hin ± ρ(cos ϕ + i sin ϕ) (4.71)

where
q
ρ = (h2in − ω 2 )2 + 4h2in Ω2
2hin Ω
sin ϕ =
ρ
h − ω2
2
cos ϕ = in (4.72)
ρ
Application of the deMoivere’a formula
p
n √ ϕ + 2kπ ϕ + 2kπ
ρ(cos ϕ + i sin ϕ) = n ρ(cos + i sin ); k = 0, 1, 2....n − 1
n n
(4.73)
to the expression 4.71 yields roots of the characteristic equation in form
√ ϕ √ ϕ
r1 = −hin + ρ cos + i ρ sin = Re(r1 ) + i Im(r1 )
2 2
√ ϕ √ ϕ
r2 = −hin − ρ cos − i ρ sin = Re(r2 ) + i Im(r2 ) (4.74)
2 2
The imaginary part and real part of the characteristic roots as a function of the rotor
angular speed is shown in Fig. 18 and 19 respectively.

Im(r) Im(r1 )

ω 2 -hin
2

- ω 2 -hin
2

Im(r2 )

Figure 18
INFLUENCE OF DAMPING 167

Re (r)

Re(r1 )

Ωt Ω
-h in

Re (r2 )

Figure 19

The general solution, as the linear combination of the particular solutions, has
the following form.

zg = Ae(Re(r1 )+i Im(r1 ))t + Be(Re(r2 )−i Im(r1 ))t = AeRe(r1 )t ei Im(r1 )t + BeRe(r2 )t e−i Im(r1 )t
(4.75)
From Eq. 4.75 one can see that the rotor performs positive and negative precession
with the same frequency Im(r1 ). For Ω < Ωt the real part of both roots is negative.
Therefore the amplitudes of the precession AeRe(r1 )t and BeRe(r2 )t decays to zero. It
follows that the equilibrium position of the rotor is stable. For Ω > Ωt the real part
which corresponds to the positive precession is positive. Hence, the amplitude of the
positive precession grows to infinity. In this case the equilibrium position is unstable.
The angular velocity Ωt determines so called stability threshold. The first equation
4.74 allows the threshold velocity to be computed.
√ ϕ
−hin + ρ cos = 0 (4.76)
2
The particular solution of the non-homogeneous equation 4.66

z̈ + 2hin ż − 2ihin Ωz + ω2 z = −2hin ΩYs (4.77)

can be predicted as a constant (time independent) and complex magnitude.

zp = E (4.78)

To obtain an expression for E, let us introduce Eq. 4.78 into 4.77.

−2ihin ΩE + ω2 E = −2hin ΩYs (4.79)

Hence,
−2hin ΩYs
E= = |E| eiη (4.80)
ω 2 − 2ihin Ω
INFLUENCE OF DAMPING 168

where

2hin ΩYs
|E| = p (4.81)
ω 4 + 4h2in Ω2
−2hin Ω
sin η = p
ω 4 + 4h2in Ω2
−ω 2
cos η = p (4.82)
ω 4 + 4h2in Ω2

The modulus |E| and the phase η are presented in Fig. 20 as a function of
the angular speed Ω. They determines the equilibrium position of the rotor which, in
this case, depends on its rotating speed. Since, according to 4.82, both sin η and cos η
are negative, all possible equilibrium positions are located in the third quarter of the
plane xy as shown in Fig. 21. Locus of the stable equilibrium positions (Ω < Ωt ) is
marked by continues line. Locus of the unstable equilibrium positions (Ω > Ωt ) is
marked by the broken line.
INFLUENCE OF DAMPING 169

|E|

Ys

0
0 Ωt Ω
η

-2π
3

π
0 Ωt Ω

Figure 20

X
0 Ω=
Ω=Ωt
Ω
Ys
Ω=0
|E |
x
o
η

y Y

Figure 21
INFLUENCE OF DAMPING 170

Forced vibration.
According to Eq. 4.63, the forced vibration caused by the imbalance of the rotor is
determined by the particular solution of the following equation.
z̈ + 2hin (−iΩz + ż) + ω 2 z = qeiΩt (4.83)
It can be predicted as follows
z = CeiΩt (4.84)
Introduction of 4.84 into 4.83 yields equation
(−Ω2 + ω 2 )C = q (4.85)
which allows the constant C to be obtained.
q
C= 2 (4.86)
ω − Ω2
The geometrical interpretation of this particular solution, for Ω < ω is given in Fig.
22
X
0

|E | C
E
x
o
η Ω t
C
G
y Y
x
R
yR

Figure 22

The centre of the shaft C performs the positive precession along circle of radius
C. The centre of this circle is at the equilibrium position E, determined by the angle
η and modulus |E|. Both the centre of the shaft and the centre of gravity stays on
line which is parallel to the rotating axis xR .
4.5.3 Influence of the external and internal damping
In a general case, when both the external and internal damping are present, motion
of the rotor is governed by equation 4.48 which may by rewritten in the following
form.
z̈ + 2(hex + hin )ż + (ω 2 − 2ihin Ω)z = 2ihin ΩZs + qeiΩt (4.87)
The general solution of the homogeneous equation, according to the previous consid-
eration has the following form.
z = A1 er1 t + A2 er2 t (4.88)
INFLUENCE OF DAMPING 171

The characteristic roots r1 and r2 which determine stability of the system equilibrium
position, in this case, are
√ ϕ √ ϕ
r1 = −(hin + hex ) + ρ cos + i ρ sin = Re(r1 ) + i Im(r1 )
2 2
√ ϕ √ ϕ
r2 = −(hin + hex ) − ρ cos − i ρ sin = Re(r2 ) + i Im(r2 ) (4.89)
2 2
where
q
ρ = ((hin + hex )2 − ω 2 )2 + 4h2in Ω2
2hin Ω
sin ϕ =
ρ
h − ω2
2
cos ϕ = in (4.90)
ρ
These roots as a function of the angular speed are presented in Fig. 23 and 24.

Im(r) Im(r1 )

ω 2 -(hin+ h ex )2

2
- ω -(hin+ h ex )2
Im (r2 )

Figure 23
INFLUENCE OF DAMPING 172

Re(r)

Re(r1 )

Ωt Ω
-(hin+hex)

Re(r2 )

Figure 24

The steady state motion is represented by the particular solution of the non-
homogeneous equation. Repetition of the previous consideration results in the fol-
lowing particular solution.

z = |E| eiη + |C| ei(Ωt+ϕ) (4.91)

where
2hin ΩYs
|E| = p
ω 2 + 4h2in Ω2
2hin Ω
η = arctan +π
ω2
q
|C| = p
(ω 2 − Ω2 )2 + 4h2ex Ω2
2hex Ω
ϕ = − arctan 2 (4.92)
ω − Ω2
INFLUENCE OF AXIAL FORCES 173

X
0

|E | |C |
E
x
o
η C
Ω t
ϕ
y Y
G
yR x
R

Figure 25

The graphical interpretation of the above solution is presented in Fig. 25.

Due to the internal dumping, the equilibrium position of the rotor E depends on its
angular velocity. The centre of gravity of the shaft C performs positive precession
about that point. The centre of gravity G is by µ apart from the centre C and
remains on the straight line which is parallel to the axis xR of the rotating system of
coordinates.

4.6 INFLUENCE OF AXIAL FORCES

4.6.1 Description of physical model
In many cases, which may be encountered in engineering, it is necessary to include
axial forces in the mathematical model of rotor system. To analyze influence of these
axial forces let us consider physical model of the rotor shown in Fig. 26To simplify
the mathematical model let us assume that the shaft is uniform and its mass is
concentrated in the middle of its span. The shaft is rigidly supported at its ends. Let
FZ be the axial compressive force. The shaft rotates with a constant speed Ω. To
develop expression for stiffness of the shaft at position of the concentrated mass let
us apply the static force S.

S /2
S
FZ

Ys s Z

G
Z Ω
l/2
Y

Figure 26
INFLUENCE OF AXIAL FORCES 174

4.6.2 Development of mathematical model and its analysis

The bending moment at the arbitrarily chosen cross-section Z is:
S
Mb = −FZ Ys − Z (4.93)
2
Introduction of the above expression into the differential equation of deflected line
00
EIYs = Mb (4.94)

results in the following differential equation

00
Ys + λ2 Ys = −κZ (4.95)

where:
FZ S
λ2 = κ= (4.96)
EI 2EI
The general solution of the above equation is a sum of the general solution of the
homogeneous equation and the particular solution of the nonhomogeneous equation.
The general solution is
Ysg = C1 sin λZ + C2 cos λZ (4.97)
The particular solution may be predicted in the following form

Ysp = A + BZ (4.98)

Implementation of the above solution into the Eq. 4.95yields the algebraic equation

λ2 A + λ2 BZ = −κZ (4.99)

It is fulfilled if and only if

κ
A = 0 and B = − (4.100)
λ2
Therefore, the general solution of the nonhomogeneous equation is
κ
Ys = Ysg + Ysp = C1 sin λZ + C2 cos λZ − Z (4.101)
λ2
The constant magnitudes C1 and C2 have to be chosen to fulfill the following boundary
conditions
0
Ys (0) = 0 Ys (l/2) = 0 (4.102)
The first condition yields
C2 = 0 (4.103)
Hence, the first derivative of the deflected line is
0 κ
Ys = C1 λ cos λZ − (4.104)
λ2
INFLUENCE OF AXIAL FORCES 175

Upon taking advantage from the second boundary condition we have

0 l κ
Ys (l/2) = 0 = C1 λ cos λ − 2 (4.105)
2 λ
and
κ
C1 = 3 (4.106)
λ cos λ 2l
According to Eq. 4.101, the deflected line can be approximated by the following
equation à !
κ sin λZ l
Ys = 2 l
−Z for 0 < Z < (4.107)
λ λ cos λ 2 2
Therefore, the displacement due to the applied force S at the point of its application
is equal to µ ¶
κ 1 l l
Ys (l/2) = s = 2 tan λ − (4.108)
λ λ 2 2
This displacement tends to infinity if the argument of the function tan tends to π/2.
Hence,
λl = π (4.109)
yields the Euler’s formula for the buckling critical magnitude of the axial force FZ .
π2
FZcr = EI (4.110)
l2
For FZ < FZcr , the ratio of force S and the corresponding displacement determines
the relationship between stiffness of the shaft k and the axial force FZ .
S S 4EIλ3
k= = S
¡1 ¢ = (4.111)
s EIλ2 λ
tan λ 2l − 2l 2 tan λ 2l − lλ
This relationship is shown in Fig. 27.

k
48EI
l3

π λ
l

Figure 27

According to Eq. 4.13, the forced vibrations due to imbalance of the rotor µ
are governed by
z̈ + 2ςωn ż + ω 2n z = qeiΩt (4.112)
INFLUENCE OF AXIAL FORCES 176

where

k 1 4EIλ3 c
ω 2n = = ; 2ςω n = and q = µΩ2 (4.113)
m m 2 tan λ 2l − lλ m

and, according to Eq. 4.96 r

Fz
λ=
EI
The particular solution of Eq.4.112 produces formula for amplitude of the forced
vibrations
q
A= q
(ω 2n − Ω2 )2 + 4ς 2 ω 2n Ω2
This amplitude, as a function of the axial force Fz and the angular velocity of the
rotor considered Ω is presented in Fig. 28 for rotor of length 5m and the outer and
inner diameter 0.05 and 0.045m respectively. The computation was carried out for
the damping ratio ζ = 0.003. As one can see from this diagram, the axil force results
in lowering of the critical speeds and increases the amplitude of vibrations in vicinity
of the resonance. If its magnitude is greater than FZcr = 1.57 × 104 N, the equilibrium
position of the system considered looses its stability.

20

A/ µ

10

15000

Fz [N]
0.0 0
30
0 Ω [rad/s] 60

Figure 28
 Chapter 5
ANALYSIS OF MOTION OF ROTOR WITH UNSYMMETRICAL
CROSS-SECTION

5.1 MODELLING
5.1.1 Description of the physical model

S E

C
G
Ω
µ

Figure 1

Let us consider rotor shown in Fig. 1 which by assumption rotates with a constant
angular speed Ω. The shaft S of the rotor is supported rigidly at its ends. Assume
that the shaft can be considered massless and flexible whereas the element E can
be approximated by a particle of mass m. This particle is attached to the shaft at
the centre of gravity G of the element E. The centre of gravity G is displaced by µ
from the geometrical centre of the shaft cross-section C. The distance µ represents
imbalance of the element E and can be considered as a small magnitude. The shaft
S has different second moments of area of its cross-section.
To analyze motion of this system let us introduce the global inertial system
of coordinates XY Z as it is shown in Fig. 2. Motion of the point C and the
point G is constrained to the plane xy of the inertial system of coordinates xyz (see
Fig. 2). Cross-section of the shaft has different second moments of area. Since
the shaft rotates, the principal axes of its cross-section rotates too. The rotating
system of coordinates xR , yR , zR rotates about its axis zR in such way that its axes
xR and yR are always parallel to the principal axes of the shaft cross-section. By
assumption the shaft rotates with a constant angular speed Ω, hence the system of
coordinates xR , yR , zR rotates with the same angular velocity and therefore its position
is determined by the angle Ωt.
The instantaneous position of the centre C is determined by the position
vector rC . The relative position of the centre of gravity G with respect to this centre
MODELLING 178

is given by vector rGC . Its components µxR and µyR along the rotating system of
coordinates determine imbalance of the rotor. The absolute position of the centre of
gravity G, in Fig. 2 is denoted by rG . The vector R represents reaction of the shaft S
on the attached element E. Its components along the rotating system of coordinates
xR , yR , zR are denoted by RxR and RyR respectively.

Ry R

O,o x O o Z , z , zR

Rx rC
Ωt
xR
R

C Ω
C r yR
xR
rG GCµ
y
µx G
R

yR y G Y y

Figure 2

5.1.2 Development of the mathematical model

rG = iR (xR + µxR ) + jR (yR + µyR ) (5.1)
Since the vector rG is resolved along the rotating system of coordinates, the Coriolis
statement has to be used to obtained components of the absolute velocity vector.

¯ ¯
¯ i j z ¯
¯ R R R ¯
= ṙG = rG + Ω × rG = iR ẋR + jR ẏR + ¯¯ Ω ¯¯
0
vG 0 0
¯ xR + µx yR + µy 0 ¯
R R

= iR (ẋR − ΩyR − ΩµyR ) + jR (ẏR + ΩxR + ΩµxR ) (5.2)

Components of acceleration of the gravity centre, according to the Coriolis statement,

are:
0
aG = v̇G = vG + Ω × vG = iR (ẍR − ΩẏR ) + jR (ÿR + ΩẋR )
¯ ¯
¯ iR jR zR ¯¯
¯
+ ¯¯ 0 0 Ω ¯¯
¯ ẋR − ΩyR − Ωµy ẏR + ΩxR + Ωµx 0 ¯
R R

= iR (ẍR − 2ΩẏR − Ω2 xR − Ω2 µxR ) + jR (ÿR + 2ΩẋR − Ω2 yR − Ω2 µyR ) (5.3)

Taking advantage of the Newton’s law, one can write that

m(ẍR − ΩẏR − Ω2 xR − Ω2 µxR ) = FxR

m(ÿR + ΩẋR − Ω2 yR − Ω2 µyR ) = FyR (5.4)
MODELLING 179

where FxR and FyR represent components of the resultant force FR acting on the
particle considered. This force can be adopted as follows

FR = RR +FexR + FinR + GR (5.5)

where:
RR represents the reaction of the shaft,
FexR and FinR stands for forces due to the external and the internal damping
respectively,
GR represents the gravity force acting on the element.
The reaction RR depend on the principal stiffness of the shaft kx and ky .

RR = −iR kx xR − jR ky yR (5.6)

According to Eq. 4.39, the components of the external damping force along the
stationary coordinates xy are
x
Fex = −cex ẋ
y
Fex = −cex ẏ (5.7)

Using the complex notations the above expression takes form:

Fex = −cex ż (5.8)

According to the formulae 4.17 and 4.18 (page 155), the complex force along the
rotating system of coordinates is

FexR e+iΩt = −cex (żR e+iΩt + zR iΩe+iΩt ) (5.9)

Therefore, components of this force along the rotating system of coordinates are:

FexR = −iR cex (ẋR − ΩyR ) − jR cex (ẏR + ΩxR ) (5.10)

According to the consideration carried out in page 161, the force due to the internal
damping can be adopted as following

FinR = −iR cinx ẋR − jR ciny ẏR (5.11)

The gravity force GR , according to the drawing in Fig. 2, along the rotating system
of coordinates has the following components

GR = iR mg sin Ωt + jR mg cos Ωt (5.12)

Introduction of Eq’s. 5.12, 5.11, 5.10 and 5.6 into 8.4 yields the equation of motion
of the system considered along the rotating system of coordinates.

ẍR − 2ΩẏR + (ω2x − Ω2 )xR + 2(hex + hinx )ẋR − 2hex ΩyR = qx + g sin Ωt
ÿR + 2ΩẋR + (ω2y − Ω2 )yR + 2(hex + hiny )ẏR + 2hex ΩxR = qy + g cos Ωt(5.13)
(5.14)
ANALYSIS 180

where:
kx ky
ω 2x = , ω 2y =
m m
cinx cin cex
2hinx = , 2hinx = x , 2hex =
m m m
2 2
qx = Ω µxR , qy = Ω µyR (5.15)
Very often, instead of the squared natural frequencies of the stationary shaft ω 2x and
ω2y , their average magnitude is introduced

ω2x + ω 2y
ω2o = (5.16)
2
and so called coefficient of parametric excitation ε
ω 2x − ω2y
ε= 2 (5.17)
ωx + ω 2y

Introduction of the above expressions into Eq. 5.13 yields

ẍR − 2ΩẏR + [ω2o (1 + ε) − Ω2 ]xR + 2(hex + hinx )ẋR − 2hex ΩyR = qx + g sin Ωt
ÿR + 2ΩẋR + [ω 2o (1 − ε) − Ω2 ]yR + 2(hex + hiny )ẏR + 2hex ΩxR = qy + g cos Ωt
(5.18)

Now, one can easily arrange for the above equations to be organized in the standard
matrix form.
mR ẍR +cR ẋR +kR xR = F (5.19)
where: ∙ ¸ ∙ ¸
1 0 2(hex + hinx ) −2Ω
mR = ; cR =
0 1 +2Ω 2(hex + hiny )
∙ 2 ¸ ∙ ¸
ω o (1 + ε) − Ω2 −2hex Ω xR
kR = ; xR =
+2hex Ω ω 2o (1 − ε) − Ω2 yR
∙ ¸ ∙ ¸
qx g sin Ωt
F = Fq + Fg = + (5.20)
qy g cos Ωt

5.2 ANALYSIS
5.2.1 Analysis of the free vibrations - stability analysis
The homogeneous part of equation 5.28 governs the free motion of the rotor considered
and has the following form

mR ẍR +cR ẋR +kR xR = 0 (5.21)

The complex natural frequencies r = h + iω R , may be obtain from the characteristic

equation which, according to consideration provided in part2 chap.1 (see page 132),
has the following form.
|[A − 1r]| = 0 (5.22)
ANALYSIS 181

where ∙ ¸
0 1
A= (5.23)
−mR kR −m−1
−1
R cR

The imaginary part ω R and the real part h of the complex roots r as a function of
the rotating speed Ω are presented in figures 3 4 5.
ANALYSIS 182

In Fig. 3 results of analysis of the undamped case (hex = hinx = hiny = 0)

are shown for ω o = 15[1/s] and ε = 0.6. For the angular speed within the range
0 < Ω < Ωln there exists four natural frequencies (imaginary parts) which correspond
to the real part equal to 0. Therefore within this region the equilibrium position of
the rotor is stable. For range Ωln < Ω < Ωun there exist two non-zero real parts h
which correspond to the imaginary parts equal to zero. One of these real parts is
positive. Therefore the vibration of the rotor will grow to infinity and its equilibrium
position is unstable. Frequency of this vibration, as it can be seen from the rotating
system of coordinates, is equal to zero (ω R = 0). It means that this frequency with
respect to the stationary coordinates ω is equal to the angular speed of the rotor
Ω (see Eq. 4.36 page 159). Region of the angular speed within which frequency of
vibration growing to infinity is equal to the angular speed is called simple instability
region. For the angular speed greater than Ωun the equilibrium position of the rotor
is stable.

40

20
ωR
[1/s]
0 a)
-20

-40

0 10 20 30 Ω [1/s]

2
h
[1/s] 0 b)
-2

-4

0 10 20 30 Ω [1/s]
Ω nl Ω u
n
simple instability region

Figure 3
ANALYSIS 183

The influence of the external damping on stability of the equilibrium posi-

tion is presented in Fig. 4 The diagram of the natural frequencies, computed for
ωo = 15[1/s], ε = 0.6, hex = 2[1/s], hinx = hiny = 0 and shown in Fig. 4a), is
almost unchanged but the width of the instability region (see Fig. 4b)) is greatly
affected. The width of the simple instability region as a function of the coefficient
of parametric excitation ε is presented in Fig. 4c) for different magnitudes of the
external damping. From this diagram one can see that for sufficiently large external
damping the instability region is suppressed to zero.

40

20

ωR 0 a)
[1/s]
-20

-40

0 10 20 30 Ω [1/s]

0
h
[1/s] -2 b)
-4

-6

0 20 u 30 Ω [1/s]
Ω nl 10
Ω n
simple instability region

0.8

0.6

ε 0.4
h ex= 0 [1/s] c)
h ex= 2 [1/s]
0.2 h ex= 5 [1/s]
0
0 10 20 30 Ω [1/s]
.

Figure 4
ANALYSIS 184

The internal damping (hinx = 2[1/s], hiny = 1[1/s]) (Fig. 5) affects the simple
instability region and causes the instability threshold for higher angular speed of the
rotor. (compare with results of analysis in page 167 Fig. 19)

40

20

ωR 0 a)
[1/s]
-20

-40

0 10 20 30 Ω [1/s]

0
h
[1/s] -2
u
b)
-4 Ω n Ωt
l
-6 Ωn
0 10 20 30 Ω [1/s]
simple instability region
instability threshold

0.8

0.6
ε 0.4
c)
0.2

0
0 10 20 30 Ω [1/s]

Figure 5
ANALYSIS 185

5.2.2 Analysis of the forced vibrations

Forced vibrations caused by imbalance
According to Eq. 5.20 the forced vibration caused by imbalance are governed by the
following equation.
mR ẍR +cR ẋR +kR xR = F (5.24)
where: ∙ ¸
qx
F = Fq = (5.25)
qy
The particular solution of the above equation can be predicted as a vector of constant
magnitudes
xRq = const (5.26)
Introduction of Eq. 5.25 into Eq. 5.24 yields the particular solution in form
∙ ¸ ∙ ¸
−1 qx xRq
xRq = kR = (5.27)
qy yRq
where ∙ ¸
ω 2o (1 + ε) − Ω2 −2hex Ω
kR = (5.28)
+2hex Ω ω o (1 − ε) − Ω2
2

Using the complex notation (see page 154) the forced vibration are:

zRq = xRq + iyRq (5.29)

Hence, the complex vibration as they can be observed from the stationary system of
coordinates, according to 4.18 page155, are as follows.

zq = zRq eiΩt = xRq cos Ωt − yRq sin Ωt + i(xRq sin Ωt + yRq cos Ωt) (5.30)

Eventually, motion of the rotor considered along the stationary system of coordinates
xy due to its imbalance is:

xq = xRq cos Ωt − yRq sin Ωt

yq = xRq sin Ωt + yRq cos Ωt (5.31)

where xRq and yRq are given by Eq. 5.27. The above equations are equivalent to

xq = A sin(Ωt + ϕ)
yq = −A cos(Ωt + ϕ) (5.32)

where:
A - stands for amplitude of the forced vibration,
ϕ - is the phase between the excitation force and the system response.
q
Aunb = (xRq )2 + (yRq )2
µ ¶
−1 xRq
ϕ = tan − (5.33)
yRq
ANALYSIS 186

As one can see from 5.31, the frequency of the forced vibration caused by the residual
imbalance is equal to the angular speed Ω of the rotor and the amplitude of this
vibration is independent of the internal damping.
The amplitude as a function of the angular speed of the rotor is presented in
Fig. 6 for the following numerical data
ω o = 15[rad/s], ε = 0.6, hex = 2[1/s], hinx = 2[1/s], hiny = 1[1/s] (5.34)
The two critical speeds coincide the left and right instability region within which the
solution 5.32 are unstable.

0
h
[1/s] -2 a)
-4 Ω nu Ωt
-6 Ω nl
0 10 20 30 Ω [1/s]
simple instability region
instability threshold

0.1

0.08 stable solutions

0.06
A unb
unstable solutions
[m]
0.04 b)

0.02

0 10 20 30 Ω [1/s]
l
Ω cr = Ω nl Ω cru = Ω nu

Figure 6

Forced vibrations caused by gravitation

The forced vibrations caused by gravitation are governed by the following equation.

mR ẍR +cR ẋR +kR xR = F (5.35)

ANALYSIS 187

where:
∙ ¸ ∙ ¸ ∙ ¸
g sin Ωt g 0
F = Fg = = sin Ωt + cos Ωt = gs sin Ωt + gc cos Ωt (5.36)
g cos Ωt 0 g

The particular solution of the equation 5.35 can be predicted in the following form.

xRg = S sin Ωt + C cos Ωt (5.37)

Introduction of the above expression into the equation 5.35 yields algebraic set of
equations linear with respect to the constant S and C. Their solution offers magni-
tudes of elements of the matrices S and C.
⎡ ⎤
∙ ¸ S x ∙ ¸−1 ∙ ¸
S ⎢ Sy ⎥ −Ω2 mR + kR −ΩcR gs
=⎣⎢ ⎥ = (5.38)
C Cx ⎦ +ΩcR −Ω2 mR + kR gc
Cy

Hence, the solution along the rotating system of coordinates can be written as follows

xRg = Sx sin Ωt + Cx cos Ωt

yRg = Sy sin Ωt + Cy cos Ωt

or in the complex form.

zRg = xRg + iyRg = (Sx + iSy ) sin Ωt + (Cx + iCy ) cos Ωt (5.39)

Since according to the Euler formulae

eiΩt − e−iΩt eiΩt + e−iΩt

sin Ωt = ; cos Ωt = (5.40)
2i 2
the complex solution can be transformed to be
µ ¶ µ ¶
Cx + Sy Cy − Sx iΩt Cx − Sy Cy + Sx −iΩt
zRg = +i e + +i e (5.41)
2 2 2 2
The complex particular solution along the stationary system of coordinates, according
to 4.18 page 155, are as follows.
µ ¶ µ ¶
iΩt Cx + Sy Cy − Sx i2Ωt Cx − Sy Cy + Sx
zg = zRg e = +i e + +i (5.42)
2 2 2 2
Separation of the real part from the imaginary part in the expression 5.42 yields
motion of the rotor along the x and y axes of the stationary system of coordinates.
Cx − Sy Cy − Sx Cx + Sy
xg = − sin 2Ωt + cos 2Ωt = xs + Agr sin(2Ωt + ϕ)
2 2 2
Cy + Sx Cx + Sy Cy − Sx
yg = + sin 2Ωt + cos 2Ωt = ys − Agr cos(2Ωt + ϕ)
2 2 2
(5.43)
ANALYSIS 188

In the above formula xs and ys are referred to as the equilibrium position of the rotor.
In general this equilibrium position depends on the internal damping and the rotor’s
angular speed Ω (compare with results of analysis of influence of internal damping
on page 165).
Cx − Sy
xs =
2
Cy + Sx
ys = (5.44)
2
Agr stands for the amplitude of the forced vibrations caused by the gravitation.
sµ ¶2 µ ¶2
Cy − Sx Cx + Sy
Agr = +
2 2

Phase of this vibrations is denoted by ϕ.

µ ¶
−1 Cx + Sy
ϕ = tan − (5.45)
Cy − Sx
ANALYSIS 189

Results of the numerical computation of the forced vibrations caused by grav-

itation, presented in Fig’s. 7 and 8, correspond to the following data:

ω o = 15[rad/s], ε = 0.6, hex = 2[1/s], hinx = 2[1/s], hiny = 1[1/s] (5.46)

In Fig. 7 amplitude of the forced vibrations is presented as a function of the rotor’s

angular speed. Frequency of these vibrations, according to Eq. 5.43 is twice the
angular speed Ω. Critical speed appears at the angular speed equal to half of+ the
average natural frequency of the rotor ωo . Such a resonance (critical speed) is called
sub-harmonic resonance (sub-harmonic critical speed).

0
h
[1/s] -2
-4 Ω nu Ωt a)
-6 Ω nl
0 10 20 30 Ω [1/s]
simple instability region
instability threshold

0.1

0.08 stable solutions

0.06
A gr unstable solutions
[m]
0.04 b)

0.02

0 10 20 30 Ω [1/s]
_1
Ω = ωo
cr 2

Figure 7
ANALYSIS 190

Fig. 8 shows trajectory of the equilibrium position as a function of the rotor’s

angular speed Ω.

-0.06 -0.05 -0.04 -0.03 -0.02 -0.01

x [m]
0.01

0.02

0.03

0.04

0.05
Ω = 6 [rad/s] Ω = 40 [rad/s] Ω = 0 [rad/s]
0.06

0.07

0.08
y [m]

Figure 8
 Chapter 6
ANALYSIS OF MOTION OF ROTOR WITH ONE DISK

6.1 MODELLING

S E

O G Z
a
Ω

Figure 1

Let us consider rotor shown in Fig. 1. Assume that the shaft S of this rotor is
flexible and massless. The element E can be considered as a rigid body attached
to the shaft at the point O. The point O is chosen to be an origin of the absolute
system of coordinates XYZ. The centre of gravity G is by a apart from the origin O.
This element possesses mass m and the axis Z is its axis of symmetry. Therefore the
moments of inertia about axis X and Y are equal to each other and are denoted by
Ix = I and Iy = I respectively. The rotor rotates with the constant angular velocity
Ω. The mathematical model, for the above stated assumption, were derived in part
1 (Eq. 1.82) to be as follows.

mẌ + cẊ + kX = FG (6.1)

where:
⎡ ⎤ ⎡ ⎤
m 0 ma 0 k11 0 k12 0
⎢ 0 m 0 ma ⎥ ⎢ 0 k11 0 k12 ⎥
m=⎢
⎣ ma
⎥
⎦ k =⎢⎣ k21
⎥
0 I + ma 2
0 0 k22 0 ⎦
2
0 ma 0 I + ma 0 k21 0 k22
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
0 0 0 0 X 0
⎢ 0 0 0 0 ⎥ ⎢ Y ⎥ ⎢ mg ⎥
c=⎢⎣ 0
⎥ X=⎢ ⎥ FG =⎢ ⎥ (6.2)
0 0 Iz Ω ⎦ ⎣ ϕX ⎦ ⎣ 0 ⎦
0 0 −Iz Ω 0 ϕY mga
MODELLING 192

For further analysis let us assume that the element E is statically and dynamically
unbalance. Therefore, according to the considerations carried out in part 1, it is
necessary to include in the mathematical model the following vector representing the
external excitation FU (Eq. 1.65).
⎡ ⎤ ⎡ ⎤
UFx SΩ2 cos(Ω t + ϕS )
⎢ UFy ⎥ ⎢ SΩ2 sin(Ω t + ϕS ) ⎥
FU = ⎢ ⎥ ⎢
⎣ U Mx ⎦ = ⎣ DΩ2 cos(Ω t + ϕD ) ⎦
⎥ (6.3)
UMy DΩ2 sin(Ω t + ϕD )
Hence the mathematical model of the system considered can be adopted in form
mẌ + cẊ + kX = FG + FU (6.4)
The particular solution of the following equation
mẌ + cẊ + kX = FG (6.5)
yields coordinates of the equilibrium position of the system. It is
⎡ ⎤
Xs
⎢ Ys ⎥
Xs = ⎢ ⎥
⎣ ϕX ⎦ = k FG
−1
(6.6)
s
ϕYs

S E

Ys
O Z

o ϕYs
Ω
z

y Y

Figure 2

If one introduce new notations (see Fig. )

X = Xs + x (6.7)
the equations of motion can be rewritten in the following form
mẍ + cẋ + kx = FU (6.8)
where (see Fig. 2) ⎡ ⎤
x
⎢ y ⎥
x=⎢ ⎥
⎣ ϕx ⎦ (6.9)
ϕy
ANALYSIS 193

The vector x determines the instantaneous position of the disk with respect to the
stationary system of coordinates xyz shown in Fig. 2. Introduction of 6.9 to 6.8
yields

mẍ + maϕ̈x + k11 x + kϕx = SΩ2 cos(Ω t + ϕS )

mÿ + maϕ̈y + k11 y + kϕy = SΩ2 sin(Ω t + ϕS ) (6.10)

maẍ + (I + ma2 )ϕ̈x + Iz Ωϕ̇y + kx + k22 ϕx = DΩ2 cos(Ω t + ϕD )

maÿ + (I + ma2 )ϕ̈y − Iz Ωϕ̇x + ky + k22 ϕy = DΩ2 sin(Ω t + ϕD ) (6.11)

Upon pre-multiplying the second of Eq. 6.10 and Eq. 6.11 by the imaginary unit
i, and then adding equations 6.10 and equations 6.11 in pairs, one may obtain the
complex form of equations of motion.

mz̈ + maϕ̈z + k11 z + kϕz = SΩ2 ei(Ωt+ϕS )

maz̈ + (I + ma2 )ϕ̈z − iIz Ωϕ̇z + kz + k22 ϕz = DΩ2 ei(Ωt+ϕD ) (6.12)

where

z = x + iy
ϕz = ϕx + iϕy (6.13)

6.2 ANALYSIS
6.2.1 Free vibrations
If one neglects in Eq. 6.12 the external excitation due to the residual unbalance
(S = 0 and D = 0), the following homogeneous equations will govern the free motion
of the rotor considered.

mz̈ + maϕ̈z + k11 z + kϕz = 0

2
maz̈ + (I + ma )ϕ̈z − iIz Ωϕ̇z + kz + k22 ϕz = 0 (6.14)

Its particular solution can be predicted in the following form.

z = Aeiωt
ϕz = Beiωt (6.15)

In the above equation, ω represent the natural frequencies of the rotor. To establish
the relationship between these natural frequencies and the angular velocity of the
rotor, let us introduce the predicted solution 6.15 into the equation of motion 6.14.
This introduction yields a set of the algebraic equations linear with respect to the
constants A and B.

(−ω 2 m + k11 )A + (−ω 2 ma + k)B = 0

(−ω 2 ma + k)A + (−ω2 (I + ma2 ) + Iz Ωω + k22 )B = 0 (6.16)
ANALYSIS 194

The above set of equation has non-zero solutions if and only if its characteristic
determinant is equal to zero.
¯ ¯
¯ −ω 2 m + k11 −ω2 ma + k ¯
|∆(ω, Ω)| = ¯¯ ¯=0
¯ (6.17)
−ω 2 ma + k −ω 2 (I + ma2 ) + Iz Ωω + k22

For positive m, I, Iz , k11 , and k22 , the equation 6.17 possesses four real roots. Two
of them are positive (ω + + − −
1 , ω 2 ) and the other two are negative (ω 1 , ω 2 ). These four
natural frequencies, as a function of the rotating speed Ω, are shown in Fig. 3.

ω 2+

ω 1+
-
ω 1
Ω
-
ω 2

Figure 3

The linear combination of the particular solutions 6.15 yields motion of the
centre of shaft C.
+ + − −
z = A+
1e
iω 1 t
+ A+
2e
iω2 t
+ A−
1e
iω1 t
+ A−
2e
iω 2 t
(6.18)

Its instantaneous position (see Fig. 4) is indicated by the resultant R of four vectors.
Two of them perform the positive precession and the other two perform the negative
precession.

x
Ω
+
ω 1 R
+
ω 2 C,G
-
ω 1 ω -
2

Figure 4
ANALYSIS 195

6.2.2 Influence of residual unbalance

The non-homogeneous equations which govern vibrations of rotor due to the residual
unbalance are, according to Eq. 6.12, as follows

mz̈ + maϕ̈z + k11 z + kϕz = SΩ2 eiϕS eiΩt

maz̈ + (I + ma2 )ϕ̈z − iIz Ωϕ̇z + kz + k22 ϕz = DΩ2 eiϕD eiΩt (6.19)

The particular solution of the above equations can be predicted in the following form

z = AeiΩt
ϕz = BeiΩt (6.20)

where A and B are the complex constant quantities which can be obtained by intro-
duction of Eq. 6.20 into 6.12.

(−mΩ2 + k11 )A + (−maΩ2 + k)B = SΩ2 eiϕS

(−maΩ2 + k)A + (−(I + ma2 )Ω2 + Iz Ω2 + k22 )B = DΩ2 eiϕD (6.21)

Solution of the above equation with respect to the unknown A and B, according to
Cramer’s formulae, are

|∆A (Ω)|
A =
|∆(Ω)|
|∆B (Ω)|
B = (6.22)
|∆(Ω)|

where
¯ ¯
¯ −mΩ2 + k11 −maΩ2 + k ¯
|∆(Ω)| = ¯¯ ¯ (6.23)
−maΩ + k −(I + ma )Ω + Iz Ω + k22 ¯
2 2 2 2
¯ ¯
¯ SΩ2 eiϕS −maΩ2 + k ¯
|∆A (Ω)| = ¯¯ ¯ (6.24)
2 iϕD
DΩ e −(I + ma )Ω + Iz Ω + k22 ¯
2 2 2
¯ ¯
¯ −mΩ2 + k11 SΩ2 eiϕS ¯¯
|∆B (Ω)| = ¯¯ (6.25)
−maΩ2 + k DΩ2 eiϕD ¯

The amplitudes of the forced vibrations have infinite magnitudes if the denominator
|∆(Ω)| is equal to 0. Hence the equation

|∆(Ω)| = 0 (6.26)

yields the critical speeds of the rotor. It is easy to notice that the determinant
|∆(Ω)| is equal to |∆(ω, Ω)| (see Eq. 6.17) for Ω = ω. Therefore the critical speeds
can be alternatively obtain as the intersection points of the diagram of the natural
frequencies of the rotor (see Fig. 3) with the straight line Ω = ω. These intersection
points are shown in Fig. 5
ANALYSIS 196

¯ ¯
¯ −ω 2 m + k11 −ω2 ma + k ¯
|∆(ω, Ω)| = ¯¯ ¯=0
¯ (6.27)
−ω ma + k −ω (I + ma2 ) + Iz Ωω + k22
2 2

ω
ω= Ω

Ω cr1 Ω cr2
Ω cr2 Ω cr1 Ω

Figure 5

6.2.3 Influence of the ratio Iz /I on the natural frequencies and the critical
speeds
Let us considered the rotor shown in Fig.6

S E

R
z

Figure 6

The dynamic behavior of the rotor depends to a large extend on the ratio of
the moment of inertia Iz of the element E about the axis z and the moment of inertia
I about axis perpendicular to the axis of rotation z. This ratio is determine by the
shape of the element E. This shape in turn can be defined by the ratio w/R. Here
ANALYSIS 197

w stands for the width of the element and R is its radius. The moment of inertia of
the disk about the axis z is
1
Iz = mR2 (6.28)
2
The moment of inertia of the disk about axis perpendicular to the axis z is
1 1
I = m( R2 + w2 ) (6.29)
4 12
Hence the ratio is
1
Iz 2
mR2 1
= 1 2 = (6.30)
I m( 14 R2 + 12 w ) 1
2
w 2
+ 16 ( R )
The relation between the ratio Iz /I and the ratio w/R is given in Fig. 7.

1.8
Iz/I
1.6

1.4

1.2

0.8

0.6

0.4

0.2

0
0 1 2 3 4 5 6
w/R

Figure 7

As one can see from this diagram the ratio Iz /I changes from 2 to 0. The ratio
equal to 2 corresponds to a thin disk (w = 0) and 0 corresponds to case when the
gyroscopic effect is neglected. (R = 0). Let us produce the diagram of the natural
frequency for Iz /I = 2, 1 and 0.5 respectively. The following diagrams were produced
for the following numerical data
m = 6kg mass of the disk
ρ = 7850kg/m3 density of the disk
E = 2.05E11N/m2
l = 0.5m length of half of the shaft (see Fig. 6)
ANALYSIS 198

d = 0.02m diameter of the shaft

For the above data the magnitude of the moments of inertia involved as a function
of the ratio w/R are shown in Fig 8.

0.1
Iz,I
0.09

0.08

0.07

0.06 Iz
I
0.05

0.04

0.03

0.02

0.01

0
0 1 2 3 4 5 6 7
w/R

Figure 8

For
w/R = 0.05 the moments of inertia are Iz = 0.086kgm2 and I = 0.043kgm2
Iz /I = 2
w/R = 1.7 the moments of inertia are Iz = 0.0081kgm2 and I = 0.0081kgm2
Iz /I = 1
w/R = 3 the moments of inertia are Iz = 0.0056kgm2 and I = 0.0112kgm2
Iz /I = 0.5
ANALYSIS 199

Results of computation for Iz /I = 2 are presented in Fig.9. It is easy to see

from the drawing 6 that the linear vibrations of the disk are uncoupled with the angu-
lar one. For this reason the gyroscopic effect does not influence the linear vibrations
and the lower natural frequency is not affected by the angular velocity of the rotor.
The natural frequency associated with the angular mode of vibration (particularly
for the positive precession) changes significantly with the angular velocity of the ro-
tor. By inspection of this diagram is easy to notice that the straight line ω n = 2Ω
is the asymptote for the upper branch of this diagram. Therefore there is only one
cross-section of the diagram of the natural frequencies with the straight line ω n = Ω.
It means that there is only one critical speed caused by the residual unbalance. It is
denoted in Fig.9 by Ωcr1 .

6.0E+3

ωn [rad/s]
5.0E+3

4.0E+3

ωn =Ω
3.0E+3

2.0E+3

Ωcr1
1.0E+3

-1.0E+4 -5.0E+3 0.0E+0 5.0E+3 1.0E+4

Ω [rad/s]

Figure 9
ANALYSIS 200

The case Iz /I = 1 is presented in Fig. 10. The upper branch of the natural
frequency tends towards the asymptote ωn = Ω. Hence, similarly to the previous case
we deal with only one critical speed associated with the linear vibrations of the rotor..

1.2E+4

1.0E+4
ωn [rad/s]

8.0E+3

6.0E+3

ωn =Ω

4.0E+3

Ωcr1
2.0E+3

0.0E+0
-1.0E+4 -5.0E+3 0.0E+0 5.0E+3 1.0E+4
Ω [rad/s]

Figure 10
ANALYSIS 201

Different number of the critical speed exists if the ratio Iz /I < 1.The diagram
presented in Fig.11 corresponds to the ratio Iz /I = 0.5. In this case there are two
cross-section points of the curves representing magnitudes of the rotor natural fre-
quencies and the straight line ω n = Ω. Each of them determines the critical speed.
They are denoted in Fig.11 by Ωcr1 and Ωcr2 . The lower one is associated with the
transverse vibration of the rotor and the higher one with the angular vibrations. This
phenomenon is due to symmetry of the rotor considered. In a general case the trans-
verse vibrations are coupled with the angular one and the the low natural frequencies
are dependent on the angular speed of the rotor.

6.0E+3

ωn [rad/s]]
5.0E+3

4.0E+3
ωn=Ω

3.0E+3

2.0E+3 Ωcr1 Ωcr2

1.0E+3

0.0E+0
-1.0E+4 -8.0E+3 -6.0E+3 -4.0E+3 -2.0E+3 0.0E+0 2.0E+3 4.0E+3 6.0E+3 8.0E+3 1.0E+4

Ω [rad/s]]

Figure 11
 Chapter 7
ANALYSIS OF MOTION OF ROTOR WITH ONE
UNSYMMETRICAL DISK

7.1 MODELLING
7.1.1 Description of the physical model

S a E

y
o G z x
o
ϕy xR Ω t
Ω

y yR
y

Figure 1

Let us consider rotor shown in Fig. 1. The element E can be considered as a rigid
body attached to the shaft S at the point o. The point o is chosen to be at origin of
the absolute system of coordinates xyz. The centre of gravity G is by a apart from
the origin o. This element possesses mass m and rotates about axis z with a constant
angular velocity Ω. The system of coordinates xR yR zR rotates about axis z of the
xyzsystem of coordinates with the angular speed equal to the angular speed of the
element E. Therefore its angular position is determined by the angular displacement
Ωt. The axis z is not its the axis of symmetry of the element E. Therefore the
moments of inertia about axis xR and yR of the rotating system of coordinates are
not equal to each other. Their magnitudes about axes through the centre of gravity
G are denoted here by Ix and Iy respectively. Iz stands for moment of inertia about
axis z. For simplicity of the further consideration, let us assume that the shaft S is
massless.
MODELLING 203

7.1.2 Development of the mathematical model

The mathematical model of the element E, were derived in part 1 (Eq. 1.39, page
22) to be as follows.
me ẍ + ce ẋ = Re +FeG (7.1)
where
⎡ ⎤
m 0 ma 0
⎢ 0 m 0 ma ⎥
me = ⎢
⎣ ma 0 Io + ma − ∆I cos 2Ωt
2
⎥
⎦
−∆I sin 2Ωt
2
0 ma −∆I sin 2Ωt Io + ma + ∆I cos 2Ωt
⎡ ⎤
0 0 0 0
⎢ 0 0 0 0 ⎥
ce = ⎢
⎣ 0 0
⎥
2∆IΩ sin 2Ωt Iz Ω − 2∆IΩ cos 2Ωt ⎦
0 0 −Iz Ω − 2∆IΩ cos 2Ωt −2∆IΩ sin 2Ωt
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x Fx 0
⎢ y ⎥ ⎢ ⎥ ⎢ ⎥
x = ⎢ ⎥ , Re = ⎢ Fy ⎥ , FeG = ⎢ mg ⎥ (7.2)
⎣ ϕx ⎦ ⎣ Mx ⎦ ⎣ 0 ⎦
ϕy My mga
Ix + Iy Ix − Iy
Io = ∆I = (7.3)
2 2
The mathematical model of the shaft S, along coordinates x, may be adopted in the
following form
ks x = Rs (7.4)
where ⎡ ⎤
k11 0 k12 0
⎢ 0 k11 0 k12 ⎥
ks = ⎢
⎣ k12 0 k22 0 ⎦
⎥ (7.5)
0 k12 0 k22
Application of the developed in part1 composition rules, results in the following math-
ematical model of the rotor system.

mẍ + cẋ + kx = FG (7.6)

where
m = me c = ce k = ks FG = FeG (7.7)
Introduction of the complex notations

x + iy = z and ϕx + iϕy = ϕz (7.8)

results in the following set of the system equations of motion.

mz̈ + maϕ̈z + k11 z + k12 ϕz = img

_ _
2 i2Ωt
maz̈ + (Io + ma )ϕ̈z − ∆Ie ϕ̈z − Iz Ωiϕ̇z − 2∆IΩiei2Ωt ϕ̇z + k12 z + k22 ϕz = imga
(7.9)
MODELLING 204

These equation, although linear, have coefficients that depend upon time. It will
be shown that if the stiffness matrix ks of the shaft considered possesses the same
dynamic properties in both the xz and yz plane (as it is in the considered case),
transformation of the above mathematical model to the rotating system of coordi-
nates xR yR zR yields equations with constant coefficients. Therefore, let us apply the
developed earlier transformation (see Eq’s 4.18 and 4.19)

z = zR eiΩt (7.10)

ż = żR eiΩt + zR iΩeiΩt

z̈ = z̈R eiΩt + 2żR iΩeiΩt − zR Ω2 eiΩt

ϕz = ϕzR e+iΩt (7.11)

ϕ̇z = ϕ̇zR eiΩt + ϕzR iΩeiΩt

ϕ̈z = ϕ̈zR eiΩt + 2ϕ̇zR iΩeiΩt − ϕzR Ω2 eiΩt
_ _
ϕz = ϕzR e−iΩt (7.12)
_ _ _
ϕ̇z = ϕ̇zR e−iΩt − ϕzR iΩe−iΩt
_ _ _ _
ϕ̈z = ϕ̈zR e−iΩt − 2ϕ̇zR iΩe−iΩt − ϕzR Ω2 e−iΩt

to the Eq’s 7.9. As a result of application of the above transformation, one can get
the following set of equations.

m(z̈R + 2żR iΩ − zR Ω2 ) + ma(ϕ̈zR + 2ϕ̇zR iΩ − ϕzR Ω2 ) + k11 zR + k12 ϕzR = imge−iΩt

_ _ _
ma(z̈R +2żR iΩ−zR Ω2 )+(Io +ma2 )(ϕ̈zR +2ϕ̇zR iΩ−ϕzR Ω2 )−∆I(ϕ̈zR −2ϕ̇zR iΩ−ϕzR Ω2 )
_ _
−Iz Ωi(ϕ̇zR + ϕzR iΩ) − 2∆IΩi(ϕ̇zR − ϕzR iΩ) + k12 zR + k22 ϕzR = imgae−iΩt (7.13)
Separation of the real part from the imaginary part yields the final mathematical
model in the following matrix form.

mR ẍR +cR ẋR +kR xR = FGR (7.14)

⎡ ⎤
m 0 ma 0
⎢ 0 m 0 ma ⎥
mR = ⎢
⎣ ma 0 Io + ma2 − ∆I
⎥
⎦
0
0 ma Io + ma2 + ∆I
⎡ ⎤
0 −2mΩ 0 −2maΩ
⎢ 2mΩ 0 2maΩ 0 ⎥
cR = ⎢
⎣ 0
⎥
−2maΩ 0 −[2(Io + ma ) − Iz ]Ω ⎦
2

2maΩ 0 [2(Io + ma2 ) − Iz ]Ω 0

ANALYSIS 205
⎡
k11 − mΩ2 0
⎢ 0 k − mΩ2
kR = ⎢
⎣ k12 − maΩ2
11
0
0 k12 − maΩ2
⎤
k12 − maΩ2 0
0 k12 − maΩ2 ⎥
⎥
2
k22 − (Io + ∆I + ma − Iz )Ω2
0 ⎦
2 2
0 k22 − (Io − ∆I + ma − Iz )Ω
⎡ ⎤ ⎡ ⎤
xR mg sin Ωt
⎢ yR ⎥ ⎢ mg cos Ωt ⎥
xR = ⎢ ⎥ ⎢
⎣ ϕxR ⎦ , FGR = ⎣ mga sin Ωt ⎦
⎥ (7.15)
ϕyR mga cos Ωt

7.2 ANALYSIS
7.2.1 Free vibrations- Stability of motion
Undamped case
The free motion of the rotor, with respect to the rotating system of coordinates, is
govern by the homogeneous equation of Eq. 7.14

mR ẍR +cR ẋR +kR xR = 0 (7.16)

The complex natural frequencies r, may be obtain from the characteristic equation
which according to consideration provided in part2 chap.1 (page 132), has the follow-
ing form.
|[A − 1r]| = 0 (7.17)
where ∙ ¸
0 1
A= (7.18)
−m−1
R Rk −m−1
R cR

Results of searching for eigenvalues r = h±iωR of Eq. 7.17 are shown in figures 2 and
3. Fig. 2(a) and 3(a) presents the imaginary parts of ωR eigenvalues r, which repre-
sents the frequencies of the system natural frequencies with respect to the rotating
system of coordinates. Fig. 2(b) and 3(b) shows the real parts h of the eigenvalues
r. Fig. 2(c) and 3(c) presents the system natural frequencies with respect to the
stationary system of coordinates ω. The diagrams shown in Fig. 2(c) and 3(c) was
created from the diagrams shown in figures (a) by application of the developed in
chapter 2 formula 4.36 (page 159)

ω = ωR + Ω (7.19)

Fig. 2 was obtained for the following set of the numerical data:

k11 = 10 000N/m, k22 = 1 000N/m, k12 = 2 000N/m, a = 0, m = 50kg,

ANALYSIS 206

30
20

10
ω R [1/s] 0 (a)
-10

-20

-30

0.3
0.2
0.1
h [1/s] 0 (b)
-0.1
-0.2

-0.3

60

40

ω [1/s] 20
(c)
Ω=ω
0

-20
0 4 8 12 16 Ω [1/s]
l u
Ω n1 Ω n1
simple instability region

Figure 2
ANALYSIS 207

30

20

10

ω R [1/s] 0
(a)

-10

-20

-30

0.8
0.6
0.4
0.2
h [1/s] 0 (b)
-0.2
-0.4
-0.6
-0.8

50

40

30

20 (c)
ω [1/s]
10

0 Ω=ω

-10

0 4 8 12 16 Ω [1/s]
l u l u u l
Ω n1 Ω n1 Ω n2 Ω n2 Ω n3 Ω n3
combined instability region
simple instability regions

Figure 3
ANALYSIS 208

Io = 10kgm2 , Iz = 20kgm2 , ∆I = 4kgm2 (7.20)

In this case (Io < Iz ), for Ω < Ωln1 and Ω < Ωun1 , there exists eight roots with the real
parts equal to zero. Therefore, in these regions of the angular speed the equilibrium
position of the rotor considered is stable. For Ωln1 < Ω < Ωun1 , there exists two roots
with the non-zero real part equal to ±h (see Fig. 2(b)). Therefore, in this region of
stability the equilibrium position is unstable. The corresponding imaginary part with
respect to the stationary system of coordinates is equal to the rotor angular speed Ω
(see Fig. 2(c)). Hence the disturbances that grows to infinity have frequency equal to
the rotor angular speed. Region of instability having such a property is called simple
region of instability.
Different picture can be observed if Io > Iz . This case is demonstrated in Fig.
3 for the following data:
k11 = 10 000N/m, k22 = 1 000N/m, k12 = 2 000N/m, a = 0, m = 50kg,
Io = 20kgm2 , Iz = 10kgm2 , ∆I = 4kgm2 (7.21)
In this case there exists three instability regions for Ωln1
<Ω< Ωun1 , Ωln2
< Ω < Ωun2 ,
Ωln3 < Ω < Ωun3 . Two of them, namely the region of stability 1 and 3, can be classified
as simple because the natural frequencies of the corresponding disturbances are equal
to the rotor angular speed (see Fig. 3(c)). The growing to infinity disturbances within
the instability region denoted by subscript 2, are a superposition of two different
frequencies. One of them is smaller then the rotor angular speed, the other is greater.
Region of instability having such a property is called combined region of instability.

Influence of the external damping

According to the consideration carried out in chapter 2, (page 160) the damping forces
due to the external damping along coordinates x, y, ϕx , ϕy of the stationary system of
coordinates may be adopted as follow
x y
Fex = cex ẋ Fex = cex ẏ
x y
Mex = cϕex ϕ̇x Mex = cϕex ϕ̇y (7.22)
or in the complex form
z
Fex = cex ż
z
Mex = cϕex ϕ̇z (7.23)
where the complex function z and ϕz are defined by Eq. 7.8. Application of the Eq.
7.10 and 7.11 yields the damping forces along the rotating system of coordinates.
zR
Fex = cex (ẋR − yR Ω) + i(ẏR + xR Ω))
zR
Mex = cϕex (ϕ̇xR − ϕyR Ω) + i(ϕ̇yR + ϕxR Ω)) (7.24)
Separation of the real part from the imaginary part, results in the damping forces
along the rotating axes.
xR yR
Fex = cex (ẋR − yR Ω) Fex = cex (ẏR + xR Ω)
xR yR
Mex = cϕex (ϕ̇xR − ϕyR Ω) Mex = cϕex (ϕ̇yR + ϕxR Ω) (7.25)
ANALYSIS 209

Implementation of the above forces in the mathematical model 7.14, eventuates in

the following modification of the stiffness and inertia matrices.
⎡ ⎤
cex −2mΩ 0 −2maΩ
⎢ 2mΩ cex 2maΩ 0 ⎥
cR = ⎢
⎣ 0
⎥
−2maΩ cϕex −[2(Io + ma ) − Iz ]Ω ⎦
2

2maΩ 0 [2(Io + ma2 ) − Iz ]Ω cϕex

⎡
k11 − mΩ2 −cex Ω
⎢ c Ω k11 − mΩ
2
kR = ⎢ ex
⎣ k12 − maΩ2 0
0 k12 − maΩ2
⎤
k12 − maΩ2 0
0 k12 − maΩ2 ⎥
⎥
2
k22 − (Io + ∆I + ma − Iz )Ω2
−cϕex Ω ⎦
2 2
cϕex Ω k22 − (Io − ∆I + ma − Iz )Ω
(7.26)

The simple and combined instability regions as a function of ∆I for the set of data
7.21 and different magnitudes of the damping coefficients are shown in Fig. 4. As
one can see from this diagram, the external damping narrowing the instability regions
and eventually liquidates them completely.
10

6
2
∆ I[kgm ]
4

0
0 2 4 6 8 10 12 14 16 18 Ω [1/s]

-1
c=0, cϕ =0 c=100 [Nm s], cϕ =10[Nms]
-1 -1
c=200 [Nm s], cϕ =20[Nms] c=300 [Nm s], cϕ =30[Nms]

Figure 4

7.2.2 Particular solution - Equilibrium position

The vector ⎡ ⎤
mg sin Ωt
⎢ mg cos Ωt ⎥
FGR = ⎢
⎣ mga sin Ωt ⎦
⎥ (7.27)
mga cos Ωt
(see Eq. 7.15) contains forces due to gravitation, therefore the particular solution of
Eq. 7.14 generated by this vector represents static deflection of the system as it can
ANALYSIS 210

be seen from the rotating system of coordinates. It can be predicted as follows

xR = S sin Ωt
yR = S cos Ωt
ϕxR = Φs sin Ωt
ϕyR = Φs cos Ωt (7.28)

Implementation of the above particular solution in Eq. 7.14 ,results in the following
set of the non-homogeneous algebraic equations linear with respect to the unknown
magnitudes of S and Φs .
∙ ¸∙ ¸ ∙ ¸
k11 k12 S mg
= (7.29)
k12 k22 Φs mag

Hence: ∙ ¸ ∙ ¸−1 ∙ ¸
S k11 k12 mg
= (7.30)
Φs k12 k22 mag
Application of the transformation 7.10 (page 204) yields static equilibrium position
of the system with respect to the stationary system of coordinates.

x = 0
y = S
ϕx = 0
ϕy = Φs (7.31)

Obviously, this equilibrium position exists for only these angular speeds of the rotor
which belong to the stable regions.
7.2.3 Forced vibration caused by the residual imbalance
According to the consideration carried out in part 1 chapt. 1 (page 26), the residual
imbalance can be modelled by the following vector of forces along the rotating system
of coordinates. ⎡ ⎤
mµxR Ω2
⎢ mµyR Ω2 ⎥
UFR = ⎢ ⎣ 2(Io − ∆I − Iz )∆xR Ω2 ⎦
⎥ (7.32)
2(Io + ∆I − Iz )∆yR Ω2
Hence, the forced vibrations of the rotor are governed by

mR ẍR +cR ẋR +kR xR = FUR (7.33)

Since the vector UFR contains constant magnitudes, the particular solution may be
predicted as a vector with constant elements too.
⎡ ⎤
X
⎢ Y ⎥
xR = ⎢
⎣ ΦX ⎦
⎥ (7.34)
ΦY
ANALYSIS 211

This prediction leads to the following expression for the amplitudes of the forced
vibrations xR
xR = k−1
R FUR (7.35)
where
⎡
k11 − mΩ2 −cex Ω
⎢ cex Ω k11 − mΩ2
kR = ⎢
⎣ k12 − maΩ2 0
0 k12 − maΩ2
⎤
k12 − maΩ2 0
0 k12 − maΩ2 ⎥
⎥
2
k22 − (Io + ∆I + ma − Iz )Ω2
−cϕex Ω ⎦
cϕex Ω k22 − (Io − ∆I + ma2 − Iz )Ω2
(7.36)

Results of computation of the amplitudes of forced vibrations X and Y for set of data
7.21 and the unit imbalance (µxR = µyR = 1[m], ∆xR = ∆yR = 1[rad]) are shown in
Fig. 5. If damping is omitted, the amplitude of forced vibrations is growing to infinity
at the boundaries of the simple instability regions. Therefore, the instability boundary
of the simple instability regions coincide the critical speeds of the rotor considered.
Inside the instability regions the solutions 7.35 are unstable and therefore the steady
state forced vibrations can not be observed. In vicinity of the combined instability
region there is no critical speeds.
ANALYSIS 212

30

20

10

ω R [1/s] 0
(a)

-10

-20

-30

10

unstable solutions stable solutions

8

X, Y [m]
4 (b)

2
Ω cr1
l
Ω ucr1
l
Ω cr3 Ω ucr3
0
0 4 8 12 16 Ω [1/s]
l u l
Ω n1 Ω n1
l
Ω n2 Ω u
n2 Ω n3 Ω un3
combined instability region
simple instability regions
X Y

Figure 5
 Chapter 8
ANALYSIS OF INFLUENCE OF THE OIL BEARINGS ON
MOTION OF ROTORS

8.1 MODELLING
8.1.1 Physical model

ϕon
ϕn

Figure 1

Let us consider the rotor shown in Fig. 1. To simplify the following analysis it is
assumed that the rotor is symmetrical and rigid. The rotor possesses mass m =20 kg
and its unbalance is µ =0.00005 m. Each of its ends are supported by means of two
identical journal oil bearings. The bearing properties are as follows
L/D =0.830000 - length to diameter ratio
D =0.5E-01 m - diameter of the journal
r =0.3E-03 m - radial clearance
η =0.320000E-01 Pa.s - viscosity of oil
Position of the three sleeves with respect to the system of coordinates xy (see Fig 1)
are
ϕo1 =18.5o ϕ1 =47.0o
ϕo2 =97.7o ϕ2 =150.0o
o
ϕo3 =282.5 ϕ3 =65.5o
The bearings are loaded by gravity forces. Therefore the load is
Fx = 0.0[N] and Fy = 100[N].
The boundary conditions were adopted according to the description given in page 75.
They are in terms of the atmospheric pressure equal to 101300[P a] and the cavitation
ANALYSIS 214

pressure equal to 70000.0[P a]. The static and dynamic characteristic produced in
section ’Mathematical model of oil journal bearings’ (see page 71) allows for prediction
of the system equilibrium position and analysis of its motion in small vicinity of this
equilibrium position. To analyze motion in full clearance it is necessary to integrate
the non-linear equation of motion of the system.
8.1.2 Mathematical model
The mathematical model of the system shown in Fig. 1 can be adopted in the
following form.

mẍ = 2Hx (x, y, ẋ, ẏ) + µmΩ2 sin Ωt

mÿ = 2Hy (x, y, ẋ, ẏ) + mg + µmΩ2 cos Ωt (8.1)

or after dividing it by m
2
ẍ = Hx (x, y, ẋ, ẏ) + µΩ2 sin Ωt
m
2
ÿ = Hx (x, y, ẋ, ẏ) + g + µΩ2 cos Ωt (8.2)
m
The components of the hydrodynamic force can be computed from Reynolds equation
according to equations 1.269 (page 73). To solve the equations of motion 8.2 they
were transfer to the state-space coordinates

ẋ = vx
ẏ = vy
2
v̇x = Hx (x, y, vx , vy ) + µΩ2 sin Ωt
m
2
v̇y = Hy (x, y, vx , vy ) + g + µΩ2 cos Ωt (8.3)
m

8.2 ANALYSIS
8.2.1 Description of the numerical simulation
The mathematical model 8.3 was solved by means of the Runge Kuta method with
the step of integration equal to ∆t =0.5E-03[s] for zero initial conditions (x(0) =
(0) (0)
y (0) = vx = vy = 0).
The procedure of solving the equation 8.3 was as follows:
(0) (0)
1. for the given initial conditions x(0) = 0 y (0) = 0 vx = 0 vy = 0 the
Reynolds equations can be solved for the pressure distribution p(0) (see equation ??
page 73).
2. the pressure p(0) can be integrated according to equation 1.275 to produce
(0) (0)
the components of the hydrodynamic forces Hx and Hy .
(0) (0)
3. the derivative ẋ(0) ẏ (0) v̇x v̇y can be computed according to the equa-
tions 8.3
ANALYSIS 215

(1) (1)
4. next the new position x(1) y (1) and new velocities vx vy can be pro-
duced

x(1) = x(0) + ẋ(0) ∆t

y (1) = y (0) + ẏ (0) ∆t
vx(1) = vx(0) + v̇x(0) ∆t
vy(1) = vy(0) + v̇y(0) ∆t (8.4)

Now the steps 1, 2, 3 and 4 can be repeated to find position of the system at
the instant equal to 2∆t. This procedure was repeated N =500 times.
8.2.2 Results of the numerical simulation
For an angular velocity Ω the the numerical integration of the equation 8.3 permitted
to produce the trajectory of motion of the journal (top diagram) and the time history
diagram (middle diagram). Fast Fourier Transformation (FST) of the time history
diagrams yields the spectrum of vibration (Fourier transforms) shown in the bottom
diagram. It allows the frequencies involved in the motion to be figured out.
ANALYSIS 216

Fig 2 to Fig. 4 were produce for low angular velocity Ω = 50[rad/s]. The
centre of the journal moves quickly to the equilibrium position (xo ∼
= −0.00013m, yo ∼
=
0.00011m) that coincides with the one in the static characteristic shown Fig. 41 (page
78). Then it oscillates about this equilibrium position due to the residual unbalance
µ. The frequency of these oscillations can be read from the figure 4. It is equal to
the rotating speed Ω = 50[rad/s].

x [m]
-0.0003

-0.0002

-0.0001

-0.0003 -0.0002 -0.0001 0 0.0001 0.0002 0.0003

y [m]

0.0001

0.0002

0.0003

Figure 2 Ω = 50rad/s

0 .0 0 0 3

0 .0 0 0 2

0 .0 0 0 1
x [m ]
y [m ]
0
0 0 .2 0 .4 0 .6 0 .8 1 1 .2 1 .4 1 .6 1 .8 2

-0 .0 0 0 1

-0 .0 0 0 2

-0 .0 0 0 3
tim e [s ]

Figure 3 Ω = 50rad/s

0.0009

0.0008

0.0007

0.0006

0.0005
x
0.0004
y
0.0003

0.0002

0.0001

0
0 100 200 300 400 500
frequency [rad/s]

Figure 4 Ω = 50rad/s
ANALYSIS 217

Fig. 5 to 7 correspond to the angular speed of the rotor equal to 100[rad/s].

Similarly, the journal moves to the equilibrium position and oscillates around it. This
time the orbit is larger. The ultra harmonics visible on the spectrum are caused by
non-linearity of the oil film. The equilibrium position is stable and the vibrations are
due to the unbalance only.

x [m]
-0.0003

-0.0002

-0.0001

-0.0003 -0.0002 -0.0001 0 0.0001 0.0002 0.0003

y [m]

0.0001

0.0002

0.0003

Figure 5 Ω = 100[rad/s]

0 .0 0 0 3

0 .0 0 0 2

0 .0 0 0 1
x [m ]
y [m ]
0
0 0 .5 1 1 .5 2

-0 .0 0 0 1

-0 .0 0 0 2

-0 .0 0 0 3
tim e [s ]

Figure 6 Ω = 100[rad/s]

0.0025

0.002

0.0015
x
y
0.001

0.0005

0
0 100 200 300 400 500

frequency [rad/s]

Figure 7 Ω = 100[rad/s]
ANALYSIS 218

The results of simulation for the angular speed equal to 200[rad/s] is shown
in Fig. 8 to 10. It can be seen that for this speed it takes the system more time to
reach the steady-state vibrations of the frequency equal to the angular speed.

x [m]
-0.0003

-0.0002

-0.0001

y [m] -0.0003 -0.0002 -0.0001 0 0.0001 0.0002 0.0003

0

0.0001

0.0002

0.0003

Figure 8 Ω = 200[rad/s]

0 .0 0 0 3

0 .0 0 0 2
x [m ]
y [m ]
0 .0 0 0 1

0
0 0 .5 1 1 .5 2

-0 .0 0 0 1

-0 .0 0 0 2

-0 .0 0 0 3

tim e [s ]

Figure 9 Ω = 200[rad/s]

0.006

0.005

0.004 x
0.003 y

0.002

0.001

0
0 100 200 300 400 500
frequency [rad/s]

Figure 10 Ω = 200[rad/s]
ANALYSIS 219

For the rotating speed Ω = 250[rad/s] the pattern of vibrations are signifi-
cantly different. Apart of the ultra harmonics the spectrum of vibrations reveals large
vibrations of frequency equal to half of the angular speed. They represent the free
vibrations that very slowly decay to zero. Such a response is observed in vicinity of
the instability threshold.

x [m]
-0.0003

-0.0002

-0.0001
-0.000 -0.000 -0.000
3 2 1 0 0.0001 0.0002 0.0003
y [m]

0.0001

0.0002

0.0003

Figure 11 Ω = 250[rad/s]

0 .0 0 0 3

0 .0 0 0 2
x [m ]
y [m ]
0 .0 0 0 1

0
0 0 .2 0 .4 0 .6 0 .8 1

-0 .0 0 0 1

-0 .0 0 0 2

-0 .0 0 0 3
tim e [s ]

Figure 12 Ω = 250[rad/s]

0.005

0.004

0.003 x
y
0.002

0.001

0
0 100 200 300 400 500
frequency [rad/s]

Figure 13 Ω = 250[rad/s]
ANALYSIS 220

For Ω = 300[rad/s] (see Fig. 14 to 16) the transient state is similar to the
one for Ω = 250[rad/s]. But this time the vibrations of the frequency equal to half of
the rotating speed are gradually growing. After a long time the journal reaches the
steady state motion that is shown in Fig. 17 to 19

x [m]
-0.0003

-0.0002

-0.0001

-0.0003 -0.0002 -0.0001 0 0.0001 0.0002 0.0003

y [m]

0.0001

0.0002

0.0003

Figure 14 Ω = 300[rad/s] transient state

0 .0 0 0 3

0 .0 0 0 2
x [m ]
y [m ]
0 .0 0 0 1

0
0 0 .2 0 .4 0 .6 0 .8 1

-0 .0 0 0 1

-0 .0 0 0 2

-0 .0 0 0 3

tim e [s ]

Figure 15 Ω = 300[rad/s] transient state

0.009
0.008
0.007
0.006
0.005 x
0.004 y
0.003
0.002
0.001
0
0 100 200 300 400 500

frequency [rad/s]

Figure 16 Ω = 300[rad/s] transient state

ANALYSIS 221

The steady state motion presented in Fig. 17 to 19 possesses in its spectrum

two distinctive frequency. The vibration of the frequency equal to the rotating speed
are caused by the unbalance of the rotor and therefore can be classified as the forced
vibrations. The vibrations of frequency equal about half of the rotating speed can be
classified as the self excited vibrations. They are due to the fact that for this rotating
speed the equilibrium position is not stable any more. This vibration exist even the
unbalance is equal to zero. This situation is simulated in Fig. 20 to 22.

x [m]
-0.0003

-0.0002

-0.0001
-0.000 -0.000 -0.000
3 2 1 0 0.0001 0.0002 0.0003
y [m]

0.0001

0.0002

0.0003

Figure 17 Ω = 300 steady state motion

0 .0 0 0 3

0 .0 0 0 2 x [m ]
y [m ]
0 .0 0 0 1

0
10 1 0 .1 1 0 .2 1 0 .3 1 0 .4 1 0 .5 1 0 .6 1 0 .7 1 0 .8 1 0 .9 11

-0 .0 0 0 1

-0 .0 0 0 2

-0 .0 0 0 3

tim e [s ]

Figure 18 Ω = 300[rad/s] - steady state motion

0.006

0.005

0.004 x
0.003 y
0.002

0.001

0
0 50 100 150 200 250 300 350 400 450 500
-0.001
frequency [rad/s]

Figure 19 Ω = 300[rad/s] - steady state motion

ANALYSIS 222

Motion of the journal of the perfectly balanced rotor rotating with speed Ω =
300[rad/s] is presented in Fig. 20 to 22. It performs the steady state vibrations
with frequency about 225 [rad/s]. The trajectory of this motion is called limit cycle.
Amplitude of these vibrations depends on the rotating speed.

x [m]
-0.0003

-0.0002

-0.0001

-0.0003 -0.0002 -0.0001 0 0.0001 0.0002 0.0003

y [m]

0.0001

0.0002

0.0003

Figure 20 Ω = 300[rad/s] - µ = 0

0 .0 0 0 3

0 .0 0 0 2

0 .0 0 0 1 x [m ]
y [m ]
0
0 0 .2 0 .4 0 .6 0 .8 1

-0 .0 0 0 1

-0 .0 0 0 2

-0 .0 0 0 3
tim e [s ]

Figure 21 Ω = 300[rad/s] - µ = 0

0.01

0.008

0.006 x
y

0.004

0.002

0
0 50 100 150 200 250 300 350 400 450 500
frequency [rad/s]

Figure 22 Ω = 300[rad/s] - µ = 0
ANALYSIS 223

Results of increment of the rotating speed by next 50 [rad/s] is presented in

Fig. 23 to 25. From the spectrum shown one can see that the amplitude of the self
excited vibrations is much lager the amplitude of the forced vibrations caused by the
unbalance.

x [m]
-0.0003

-0.0002

-0.0001

-0.0003 -0.0002 -0.0001 0 0.0001 0.0002 0.0003

y [m]

0.0001

0.0002

0.0003

Figure 23 Ω = 350[rad/s]

0 .0 0 0 3

0 .0 0 0 2

0 .0 0 0 1
x [m ]
y [m ]
0
0 0 .2 0 .4 0 .6 0 .8 1

-0 .0 0 0 1

-0 .0 0 0 2

-0 .0 0 0 3

tim e [s ]

Figure 24 Ω = 350[rad/s]

0.02
0.018
0.016
0.014
0.012 x
0.01 y
0.008
0.006
0.004
0.002
0
0 100 200 300 400 500

frequency [rad/s]

Figure 25 Ω = 350[rad/s]
 Chapter 9
ANALYSIS OF THE INFLUENCE OF THE ROLLING ELEMENTS
BEARINGS ON A ROTOR VIBRATIONS

9.1 CASE OF THE VERTICAL SHAFT

9.1.1 Physical model
For the vertical model of a rotor one can assume that due to the ever existing un-
balance the rolling elements are always in contact with the outer ring (see Fig. 1).
The centre of the journal (inner ring) performs precession along a circle. Radius of
this circle depends of the radial clearance, Hertzian and ovalization deflections. Let
us assume that the rotor is symmetrical with respect to the plane A − A and its
mass is m is concentrated in the middle of the massless shaft of stiffness kr . For this
assumptions the physical model of the rotor plane motion can be accepted as shown
in Fig. 2.In this diagram kB stands for the stiffness of the bearing due to the Hertzian
and ovalization deflection.
Fo
kB = (9.1)
δh + δo
The stiffness of the rotor is denoted by kr . The combined stiffness of the rotor kr and
bearings 2kB is
2kB kr
k= (9.2)
2kB + kr
The point B represents the instantaneous position of the centre of the journal whereas
C indicates the position of the centre of the rotor. If µ represents the unbalance of
the rotor, the centre of mass of the rotor is at the point G.
9.1.2 Mathematical model and its analysis
The absolute position vector of the centre of gravity G is

rG = (x + µ cos Ωt) i+ (y + µ sin Ωt) j (9.3)

The interaction force between the concentrated mass and the shaft is
³p ´
F = kr = k x2 + y 2 − δc (9.4)

Double differentiation of the position vector 9.3.with respect to time yields the ab-
solute acceleration
¡ ¢ ¡ ¢
r̈G = ẍ − µΩ2 cos Ωt i+ ÿ − µΩ2 sin Ωt j (9.5)
CASE OF THE VERTICAL SHAFT 225

A A

Figure 1

O A
x
δc 2kB F=kr
B y
kr
x C
Ωt m
rG
r G
y
µ

Figure 2
CASE OF THE VERTICAL SHAFT 226

Application of the Newton’s law results in the following equations of motion

m ((ẍ − µΩ2 cos³Ωt) i+ (ÿ − µΩ2´sin Ωt) j) = ³ ´

p p (9.6)
= −ik x2 + y 2 − δ c cos Ωt − jk x2 + y 2 − δ c sin Ωt

Taking into account that

y x
sin Ωt = p cos Ωt = p (9.7)
x2 + y 2 x2 + y 2

the equation of motion can be rewritten as follows

¡ ¢
mẍ + kx = mµΩ2 + kδ c cos Ωt
¡ ¢
mẍ + kx = mµΩ2 + kδ c sin Ωt (9.8)

or in the complex form

µ ¶
k k
z̈ + z = µΩ + δ c eiΩt
2
(9.9)
m m
where
z = x + iy (9.10)
If the external damping is included, the equation of motion is
µ ¶
c k k
z̈ + ż + z = µΩ + δ c eiΩt
2
(9.11)
m m m
The particular solution of the above equation is

z = AeiΩt (9.12)

where
k
µΩ2 + m δc
A = q¡ ¢2 ¡ c ¢2 (9.13)
k
m
− Ω2 + m Ω2
is the distance OC.(see Fig. 2).
The force F that acts on the concentrated mass is

F = k(A − δ c ) (9.14)

Half of this force acts on the journal, therefore its deflection is

F k(A − δ c )
AB = = (9.15)
2kB 2kB
Eventually, The radius of precession of the centre of the journal is
k(A − δ c )
R = AB + δ c = + δc (9.16)
2kB
CASE OF THE VERTICAL SHAFT 227

This radius R as well the radius of precession of the centre of the shaft A, as a function
of the rotating speed, is presented in Fig.3 for the following numerical data
m = 250kg
µ = 1.00E − 04[m]
kr = 5.00E + 07[N/m]
c = 10000[Ns/m]
The bearing characteristic was produced in section ’Problems’ (see page 103)
δc = 4.47e − 5[m]
δh = 1e − 6[m] for Fo = 5000[N]
δo = 5e − 6[m] for Fo = 5000[N]

0.0011

0.001

0.0009

0.0008

0.0007

0.0006
A [m]
0.0005 R [m]

0.0004

0.0003

0.0002

0.0001
µ
δc
0
0 200 400 600 800 1000 1200
rotating speed [rad/s]

Figure 3
CASE OF THE HORIZONTAL SHAFT 228

9.2 CASE OF THE HORIZONTAL SHAFT

9.2.1 Physical model

Figure 4

kB kB
2δc
kB m
kr
x
kB
kB kB

y y
δc δc

Figure 5

The horizontally situated rotor (Fig. 4) behaves differently in the horizontal x and
differently in the vertical y direction. In the vertical direction (see Fig.5) the grav-
ity forces keep the journal of the bearing in constant contact with rolling elements.
Therefore motion of the rotor in this direction is similar to that for the vertical rotor
described in the previous section. In the horizontal direction, in vicinity of its central
position, the journal is not in contact with the bearing. Therefore its motion for
−δ c < x < δ c can be approximated by the physical model presented in Fig. 6.
CASE OF THE HORIZONTAL SHAFT 229

x=xj
µmΩ 2cosΩ t
c x
m

Figure 6

To predict the forced vibrations of the rotor in the x direction for |x| > δ c one
can linearize the characteristic of the bearing and adopt it in form 1.355(see page
106).

FA Fo (A − δ c )
kA = =
A A(δ h + δ o )
5000(A − 7.47 · 10−6 ) A − 7.47 · 10−6
= = for A > δ c = 7.47 · 10−6 (9.17)
A(1 · 10−6 + 5 · 10−6 ) 12 · 10−10 A
In this case the physical model can be adopted as it is shown in Fig. 7

x
xj
2kA m
kr µmΩ 2cosΩ t
x
c

Figure 7

9.2.2 Mathematical model and its analysis

The equations of motion in the x direction, according to the developed in the previous
section physical models, were adopted in the following form
mẍ + cẋ = µmΩ2 cos Ωt for |x| < δ c (9.18)
mẍ + cẋ + kx = µmΩ2 cos Ωt for |x| > δ c (9.19)
where k stand for the combined bearing-rotor stiffness
2kA kr
k= (9.20)
2kA + kr
The particular solution of the above equations yields the formulae for the amplitudes
of the forced vibration of the rotor.
µΩ
A = q ¡ ¢2 for |x| < δc (9.21)
Ω2 + mc
µΩ2
A = q¡ ¢2 ¡ ¢2 for |x| > δ c (9.22)
k
m
− Ω2 + mc Ω2
CASE OF THE HORIZONTAL SHAFT 230

It must be noticed that in the last formula the stiffness k is a function of the wanted
amplitude of vibrations A.
−6
2kA kr 2 A−7.47·10 k
12·10−10 A r
k= = A−7.47·10 −6 (9.23)
2kA + kr 2 12·10−10 A + kr

Since the amplitude of the interaction force between the particle m and the shaft is

F = 0 for |x| < δ c (9.24)

F = Ak for |x| > δ c (9.25)

the amplitude of vibration of the journal along the coordinate xj is

Aj = A for |x| < δ c (9.26)

2kA
Aj = A for |x| > δ c (9.27)
k
The equations 9.21 and 9.22 were solved by means of the subsequent approx-
imation method for different angular speeds of the rotor. Then the amplitudes of
the journal were computed according to equations 9.26 and 9.27. Results of this
computation (the resonance diagram) are presented in Fig. 8. with help of the
thickest line (clearance 7.47e − 6[m]). As one can see from this diagram the ampli-
tude of the journal grows very quickly to reach the magnitude equal to the clearance
δ c = 7.47e − 6[m]. Then the amplitude grows slowly and reaches its maximum for the
critical speed equal to 425[rad/s]. When the rotating speed goes towards infinity the
amplitude of the forced vibrations tends to the value of the unbalance of the rotor
µ = 0.00001[m].
CASE OF THE HORIZONTAL SHAFT 231

0.00012

amplitude [m]

0.0001

clearance 90e-6[m]
0.00008
clearance 90e-6[m]
clearance 50e-6[m]
clearance 7.47e-6[m]
0.00006

0.00004

0.00002

0
0 100 200 300 400 500 600 700 800
rotating speed [rad/s]

Figure 8

To analyze the influence of the bearing clearance on the forced vibrations

of the rotor, the above computation was carried out for two lager clearance δ c =
50e − 6[m].and δ c = 90e − 6[m]. Results of this computation are also presented in
Fig. 8. For the clearance equal to 50e−6[m] the right hand side slope of the resonance
curve is much steeper than the left hand side one. The critical speed is shifted to
the left and the vibration are generally larger. The thin line represents the resonance
curve for the bearing clearance equal to 90e − 6[m]. In the range of the rotating speed
between 340 and 390[rad/s] the journal can performs the forced vibration with two
different amplitudes.
To evaluate accuracy of the above method of prediction of the amplitude of
the forced vibration, a nonlinear simulation of the system motion was performed.
The following non-linear equation

mẍ + cẋ + F (x) = µmΩ2 cos Ωt (9.28)

where according to the diagram in Fig. 9

F (x) = 0 for |x| < δ c (9.29)

5000
F (x) = (−δ c + x) for x > δ c (9.30)
δh + δo
5000
F (x) = (δ c + x) for x < −δ c (9.31)
δh + δo
CASE OF THE HORIZONTAL SHAFT 232

was numerically integrated by means of the Runge Kuta method.

F [N]
δc δh δo

5000

FA

A 10 .10 -6 δ [m]

Figure 9

The simulation corresponding to the rotating speed Ω = 400[rad/s] and clear-

ance δc = 50E − 6[m] is presented in Fig. 10. The thick line represents motion of
the centre of the journal and the thin one corresponds to the motion of the centre of
the rotor. It can be noticed that magnitude of the amplitude of the forced vibrations
coincides with that shown in Fig. 3. Spectrum of this vibrations is shown in Fig. 11.
The frequencies involved in this vibrations are equal to Ω + 2nΩ. Such a spectrum is
frequently monitored for rotor systems with worn down bearings.
CASE OF THE HORIZONTAL SHAFT 233

0.0008

0.0006

displacement [m]
0.0004
centre od the rotor
centre of the bearing
0.0002

0
1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4

-0.0002

-0.0004

-0.0006

-0.0008

time [s]

Figure 10

0.04

0.035

0.03

0.025

centre of bearing
0.02

0.015

0.01

0.005

0
0 500 1000 1500 2000 2500 3000 3500

frequency [rad/s]

Figure 11
 Chapter 10
ANALYSIS OF THE INFLUENCE OF THE SUPPORTING
STRUCTURES ON A ROTOR VIBRATIONS.

In the analysis so far it was assumed that the rotor is supported by rigid supports.
In many cases such an assumption cannot be adopted. The supporting structure
influences the dynamic behavior not only quantitatively but qualitatively as well.
The aim of this chapter is to provide qualitative analysis of the supporting structures
on vibrations of rotors.
In general, the interaction forces between the rotor and the supporting struc-
tures are function of all coordinates involved in their modelling and their derivative.
In this chapter we will assume that the interaction forces are function of these coor-
dinates only along which the supporting structure is attached to the rotor (see Fig.
1).

Rx = Rx (x, y, ẋ, ẏ, ẍ, ÿ)

Ry = Ry (x, y, ẋ, ẏ, ẍ, ÿ) (10.1)

z
x y

y Rx
Ry

Figure 1

In vicinity of the equilibrium position the generally non-linear functions 10.1

can be linearize and adopted for the future analysis as follows

R = mz̈ + cż + kz (10.2)

where ∙ ¸ ∙ ¸
Rx x
R= and z = (10.3)
Ry y
PROPERTIES OF SUPPORTING STRUCTURES 235

If the matrices m, c and k are of the following form

∙ ¸
a 0
a= (10.4)
0 a
The supporting structure is called isotropic. Otherwise the supporting structure is
called anisotropic

10.1 PROPERTIES OF SUPPORTING STRUCTURES

10.1.1 Isotropic structures

x
α

x1

y1 y

Figure 2

Let ∙ ¸ ∙ ¸∙ ¸
Rx a 0 x
= (10.5)
Ry 0 a y
be the relationship between the forces and displacements along the xy system of
coordinates. Let us produce this relationship along the system of coordinates x1 y1 .
The matrix of directional cosines is
∙ ¸ ∙ ¸ ∙ ¸∙ ¸
x x1 cos α − sin α x1
= [C] = (10.6)
y y1 sin α cos α y1
Utilization of this equation in transferring the equation 10.5 to the rotating
system of coordinates results in the following equation
∙ ¸∙ ¸ ∙ ¸∙ ¸∙ ¸
cos α − sin α Rx1 a 0 cos α − sin α x1
= (10.7)
sin α cos α Ry1 0 a sin α cos α y1
Therefore the wanted relationship is
∙ ¸ ∙ ¸T ∙ ¸∙ ¸∙ ¸
Rx1 cos α − sin α a 0 cos α − sin α x1
= =
Ry1 sin α cos α 0 a sin α cos α y1
∙ ¸∙ ¸
a 0 x1
= (10.8)
0 a y1
The above analysis allows us to say that the dynamic properties of the isotropic
supported structure are the same regardless the direction determined by the angle α.
PROPERTIES OF SUPPORTING STRUCTURES 236

10.1.2 Anisotropic structures

x
α

x1

y1 y

Figure 3

Let ∙ ¸ ∙ ¸∙ ¸
Rx axx axy x
= (10.9)
Ry ayx ayy y
be the relationship between the forces and displacements along the xy system of
coordinates. Let us apply the procedure from the previous section to produce this
relationship along the system of coordinates x1 y1 .According to equation 10.6 we have
∙ ¸ ∙ ¸T ∙ ¸∙ ¸∙ ¸
Rx1 cos α − sin α a11 a12 cos α − sin α x1
= =
Ry1 ∙ sin α cos α a21 a22 sin α cos α y1
1
(a + a22 + (a11 − a22 ) cos 2α + (a12 + a21 ) sin 2α)
= 12 11
(a − a12 + (a12 + a21 ) cos 2α + (−a11 + a22 ) sin 2α)
2 21 ¸∙ ¸
1
2
(a12 − a21 + (a12 + a21 ) cos 2α + (−a11 + a22 ) sin 2α) x1
1
(a + a22 + (−a11 + a22 ) cos 2α − (a12 + a21 ) sin 2α)
2 11
y1
(10.10)
It is easy to see that if
(a12 + a21 ) cos 2α − (a11 − a22 ) sin 2α = 0 (10.11)
the cross-coefficients have equal absolute value but the opposite sign. Solution αp of
the above equation
a12 + a21
tan 2αp = (10.12)
a11 − a22
determine the so called principal axes of anisotropy. Along these principal axes the
relationship between forces and displacements is
∙ ¸ ∙ ¸∙ ¸
Fxp b11 b12 xp
= (10.13)
Fyp b21 b22 yp
where
1
b11 = (a11 + a22 + (a11 − a22 ) cos 2αp + (a12 + a21 ) sin 2αp ) (10.14)
2
1
b12 = −b21 = (a12 − a21 ) (10.15)
2
1
b22 = (a11 + a22 + (−a11 + a22 ) cos 2αp + (−a12 − a21 ) sin 2αp ) (10.16)
2
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH CONCENTRATED
MASS 237
Such a properties possesses journal oil film that was discussed in section "Oil journal
bearings." (see page.71).
In most engineering cases the cross-coefficients are equal to each other. There-
fore along the principal axes the cross-coeficients, according to 10.15, are equal to zero.
and the force-displacement relationship is as following
∙ ¸ ∙ ¸∙ ¸
Rxp b11 0 xp
= (10.17)
Ryp 0 b22 yp

where
1
b11 = (a11 + a22 + (a11 − a22 ) cos 2αp + 2a12 sin 2αp )
2
1
b22 = (a11 + a22 + (−a11 + a22 ) cos 2αp − 2a12 sin 2αp ) (10.18)
2

10.2 INFLUENCE OF THE SUPPORTING STRUCTURES ON A RO-

TOR WITH CONCENTRATED MASS
To analyze the influence of the supporting structure let us assume that the rotor
possesses mass mr . This mass is concentrated in the middle of its span (see Fig. 4).

yj z
kr y
mr

Figure 4

Ry

x
yj
y
Rx Ωt
xj kr mr
µ
x
y

Figure 5
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH CONCENTRATED
MASS 238
The shaft of this rotor is uniform and its stiffness is kr . The rotor is supported
at its ends by two identical supports. For this assumption the physical model of the
rotr-support system is shown in Fig. 5). Here µ stands for the unbalance of the
rotor and Ω stands for its rotating speed. The interaction forces between the rotor
and both supporting structures are denoted by Rx and Ry . They are function of
the instantaneous position and velocity of the journal of the rotor xj and yj . The
mathematical model of the rotor may be adopted in the following form.

mr z̈ + kr z + R = Fc cosΩt + Fs sinΩt (10.19)

where
⎡ ⎤ ⎡ ⎤
mr 0 0 0 kr 0 −kr 0
⎢ 0 mr 0 0 ⎥ ⎢ 0 kr 0 −kr ⎥
mr = ⎢
⎣ 0
⎥ kr ⎢
⎦ ⎣ −kr
⎥
0 0 0 0 kr 0 ⎦
0 0 0 0 0 −kr 0 kr
⎡ ⎤ ⎡ ⎤ ⎡ 2
⎤ ⎡ ⎤
0 x µΩ mr 0
⎢ 0 ⎥ ⎢ y ⎥ ⎢ 0 ⎥ ⎢ µΩ2 mr ⎥
R = ⎢ ⎥ z=⎢
⎣ Rx ⎦ ⎣
⎥ Fc = ⎢ ⎥ Fs = ⎢ ⎥(10.20)
xj ⎦ ⎣ 0 ⎦ ⎣ 0 ⎦
Ry yj 0 0

The results of the following numerical computations was produced for mr = 10[kg],
kr = 1E5[N/m] and µ = 0.0001[m]
10.2.1 Influence of the isotropic structure.
Influence of support’s stiffness
Let us assume that the inertia matrix is equal to zero and the force-displacement
relationship takes form ∙ ¸
Rx
Rs = = kzs + cżs (10.21)
Ry
where ∙ ¸ ∙ ¸ ∙ ¸
k 0 c 0 xj
k= c= zs = (10.22)
0 k 0 c yj
Introduction of the above equation into the mathematical model of the rotor results
in the mathematical model of the whole system. The resonance diagram for different
magnitudes of the stiffness of the supports k and damping coefficient c = 100[Ns/m]
is presented in Fig. 6
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH CONCENTRATED
MASS 239
For a large stiffness of the support (see Fig. 6 (k = 1E12)) the supports as
well as the rotor’s journal hardly move (tick line). Therefore the damping provided
by the supporting structure does not attenuate the vibrations of the rotor (thin line
k = 1E12)). Lowering of the stiffness of the supports results in lowering the critical
speeds. The ratio of support to rotor amplitude of vibrations vibration is growing.
Due to the motion of the supports and its damping properties the energy is dissipated.
This results in lowering the amplitude of vibrations of the rotor.

0.004

A for k=1E12
Aj for k=1E12
A for k=1E6
Amplitude [m]
Aj for k=1E6
0.002 A for k=1E5
Aj for k=1E5
A for k=1E4
Aj for k=1E4

0
0 20 40 60 80 100 120
rotating speed [rad/s]

Figure 6
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH CONCENTRATED
MASS 240
Influence of support’s inertia
Let us approximate the inertia properties of the supports as it is shown in the equation
10.23 and 10.24. ∙ ¸
Rx
Rs = = mz̈s +cżs + kzs (10.23)
Ry
where
∙ ¸ ∙ ¸ ∙ ¸ ∙ ¸
m 0 c 0 k 0 xj
m= c= k= zs = (10.24)
0 m 0 c 0 k yj

Now, three modes of vibration are possible. They are shown in Fig. 7

z
kr mr m
m

mr
z
kr m
m

m
z
mr
kr
m

Figure 7

The third mode is independent of the rotor mass and stiffness. This mode is
not included in the adopted mathematical model of the rotor. The resonance diagram
of the rotor-support system is presented in Fig. 8 for c = 100[Ns/m], k = 1.E5 and
varied mass of the supports. The thin line represents the amplitude of vibrations
of the concentrated in the middle of the shaft mass. The thick line represents the
amplitude of vibration of the rotor journals.
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH CONCENTRATED
MASS 241
Due to the isotropic properties the system response is identical in both the xz
plane and the yz plane. If mass of the supporting structure is zero,see Fig.8, we deal
with one critical speed about 70[rad/s]. Introduction of the mass of the supports
results in two critical speeds corresponding to the first two modes shown in Fig. 7.
One of them is lower then 70[rad/s] and the other is higher. Both of them are shifted
to the left when the mass increases. The lower one is damped increasingly better
whereas the higher one is damped gradually less.

0.003

amplitude [m]
0.0025 m=0[kg]
m=0[kg]
m=10[kg]
0.002 m=10[kg]
m=20[kg]
m=20[kg]
0.0015 m=50[kg]
m=50[kg]

0.001

0.0005

0
0 20 40 60 80 100 120 140 160 180
rotating speed [ras/s]

Figure 8
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH CONCENTRATED
MASS 242
10.2.2 Influence of the anisotropic structure.
Influence of support’s stiffness
Let us assume that the inertia matrix and the damping matrix is equal to zero. And
the force-displacement relationship is
∙ ¸
Rx
Rs = = kzs + cżs (10.25)
Ry

where ∙ ¸ ∙ ¸ ∙ ¸
kxx kxy c 0 xj
k= c= zs = (10.26)
kyx kyy 0 c yj

Case of symmetric matrix of stiffness (kxy = kyx ) In this case there always
exists such a system of coordinates xj , yj along which the cross coefficients are equal
zero and the coefficients on the main diagonal can be computed according to the
equations 10.12 and 10.14 to 10.16.
1
kxp = (kxx + kyy + (kxx − kyy ) cos 2αp + 2kxy sin 2αp )
2
1
kyp = (kxx + kyy + (−kxx + kyy ) cos 2αp − 2kxy sin 2αp )
2
As it can be seen from the above equation the elements of the main diagonal are of
the form
k0 + ∆k and k0 − ∆k (10.27)
where

k0 = kxx + kyy and ∆k = +(kxx − kyy ) cos 2αp + 2kxy sin 2αp ) (10.28)

Therefore for the analysis of influence of the cross-coefficients it is sufficient to consider

the stiffness matrix as follows
∙ ¸
k0 + ∆k 0
k= (10.29)
0 k0 − ∆k

The following diagram were produced for c = 100[Ns/m and k0 = 1E5[N/m]

INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH CONCENTRATED
MASS 243
In this case the homogeneous equation of motion of the whole system

mr z̈ + kr z + R = Fc cosΩt + Fs sinΩt (10.30)

yields the roots of the characteristic equation. The imaginary parts of these roots
represents the natural frequencies of the system. They are shown in Fig.9. The corre-
sponding real parts (see Fig.10) determine the stability of solutions of the equations
10.30. One can see from the Fig.9 that the parameter ∆k splits the natural fre-
quency into two frequencies different in xp and yp direction. One of them grows with
the parameter ∆k the other drops down to zero. When the lower natural frequency
reaches zero (∆k = k0 ), one can see from the diagram in Fig.10 that the solution
of the equations 10.30 are unstable and the free vibration of the system tends to
infinity. Therefore detection of a low frequency in a spectrum of a rotor vibration
may indicates that the rotor approaches the instability threshold.
100

I m m a g in a r y p a r t o f r o o t s [ r a d /s ]
80

60

40

20

0 .0 0 E + 0 0 2 .0 0 E + 0 4 4 .0 0 E + 0 4 6 .0 0 E + 0 4 8 .0 0 E + 0 4 1 .0 0 E + 0 5

-2 0
∆ k [N /m ]
-4 0

-6 0

-8 0

-1 0 0

Figure 9

10
R e a l p a r t o f r o o t s [ r a d /s ]

0 .0 E + 0 2 .0 E + 4 4 .0 E + 4 6 .0 E + 4 8 .0 E + 4 1 .0 E + 5

-5

∆ k [ N /m ]

Figure 10
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH CONCENTRATED
MASS 244
The diagram in Fig.11 presents the resonance curve of the rotor-support sys-
tem for c = 100[Ns/m], k0 = 1E5 and two different values for ∆k. The boxed line
corresponds to ∆k = 0. In this case we deal with one critical speed close to 70[rad/s]
(see the diagram of the natural frequencies). For ∆k = 8E4[N] the system pos-
sesses two critical speeds. One at 42[ra/s] and the second one at about 80[rad/s]. If
∆k = k0 the lower critical speed is equal to zero. If ∆k > k0 the equilibrium position
of the system is unstable.

0.007

amplitude [m]
0.006

0.005 x,y, for dk=0[N]

xj,yj for dk=0[N]
x for dk=8E4[N]
0.004
y for dk=8E8[N]
xj for dk=8E4[N]
0.003 yj for dk=8E4[N]

0.002

0.001

0
20 30 40 50 60 70 80 90 100

rotating speed [rad/s]

Figure 11
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH CONCENTRATED
MASS 245
Case of antisymmetric matrix of stiffness (kxy = −kyx ) In a general case,
along the principal axis of anisotropy, the stiffness matrix has the following form
∙ ¸
ko + ∆k kxy
k= (10.31)
−kxy ko − ∆k

For analysis of the influence of the cross-coefficients kxy the force -displacement rela-
tionship for the supporting structure was adopted in the following form
∙ ¸
Rx
Rs = = kzs + cżs (10.32)
Ry

where ∙ ¸ ∙ ¸
c 0 xj
c= zs = (10.33)
0 c yj
The numerical computation was carried out for c = 100[Ns/m], k0 = 1E5[N/m] and
∆k = 8E4[N].
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH CONCENTRATED
MASS 246
Fig. 12 presents the natural frequencies of the system obtained as the imagi-
nary part of the characteristic roots. Due to the increment in the cross-coefficients kxy
the two natural frequencies (40[rad/s] and 80[rad/s] see Fig.12) tends to each other.
They are associated with the negative real parts of the characteristic roots (see Fig.
13). Therefore in the range of the cross-coefficients 0 to 80000N/m the equilibrium
position as well as all solution of the mathematical model considered are stable. The
situation changes radically when the two natural frequencies meet each other. From
this stage on we have only one natural frequency (±72000[rad/s]) associated with
two different real parts of the characteristic roots. One of them is growing rapidly
whereas the other is getting smaller. When the higher root becomes positive the
equilibrium position is loosing its stability.
Detection of two close to each other natural frequencies in a spectrum of vibra-
tions of a rotor system may indicates that the system approaches stability threshold.
100
Im m a g in a r y p a rt o f ro o ts [r a d /s ]
80

60

40

20

2 .E + 0 4 4 .E + 0 4 6 .E + 0 4 8 .E + 0 4

-2 0 k xy [N /m ]

-4 0

-6 0

-8 0

Figure 12

8
R ea l p a r ts o f r o o ts
6

2
2 .E + 0 4 4 .E + 0 4 6 .E + 0 4 8 .E + 0 4
0

-2

-4 k xy [N /m ]

-6

-8

-1 0

Figure 13
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH CONCENTRATED
MASS 247
The forced response of the rotor due to the residual excitation is presented in
Fig. 14. For kxy = 0 we deal with two critical speeds which coincide with the two
natural frequencies shown in Fig. 12. The distance between them tends to zero as
the cross-coefficient is growing. For a large enough magnitude of the cross-coefficient
this two critical speeds coincide each other. This situation in Fig.14 is marked with
x.

0 .0 0 7
a m p litu d e [m ]
0 .0 0 6

0 .0 0 5 x d ire c tio n k x y = 0 (s ta b le )
y d ire c tio n k x y = 0 (s ta b le )
0 .0 0 4 x d ire c tio n k x y = 5 E 4 (s ta b le )
y d ire c tio n k x y = 5 E 4 (s ta b le )
x d ire c tio n k x y = 9 E 4 (u n s ta b le )
0 .0 0 3
y d ire c tio n k x y = 9 E 4 (u n s ta b le )

0 .0 0 2

0 .0 0 1

0
20 30 40 50 60 70 80 90 100
r o ta tin g s p e e d [r a d /s ]

Figure 14
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH DISTRIBUTED
MASS 248
10.3 INFLUENCE OF THE SUPPORTING STRUCTURES ON A RO-
TOR WITH DISTRIBUTED MASS

O
z
k k
l

Figure 15

Let us consider the uniform beam of length l Young’s modulus E, area A, the second
moment of area J and density ρ shown in Fig. 15. This beam is supported at both
ends by means of springs of the same stiffness k. The equation of motion of the beam
is
∂ 2 y(z, t) 4
2 ∂ y(z, t)
+λ =0 (10.34)
∂t2 ∂z 4
where
EJ
λ2 = (10.35)
Aρ
Its particular solution can be sought in the following form

y(z, t) = Yn (z) sin ω n t (10.36)

The above solution has to fulfill boundary conditions. The left hand side end of the
beam, with the forces acting on it, is shown in Fig.16.

dz dz
M(z,t)|z=0
y z
V(z,t)| z=0
V(z,t)| z=l
M(z,t)|z=l
k
k
l
k y(z,t)|z=l
y k y(z,t)| z=0

Figure 16

The equilibrium conditions applied to the element dz yields the boundary

INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH DISTRIBUTED
MASS 249
conditions associated with the left hand end of the beam in the following form.
¯
∂ 2 y(z, t) ¯¯
M(z, t)|z=0 = E(z)J(z) =0 (10.37)
∂z 2 ¯z=0
¯ ¯
∂M(z, t) ¯¯ ∂ ∂ 2 y(z, t) ¯¯
V (z, t)|z=0 = = EJ = − ky(z, t)|z=0 (10.38)
∂z ¯z=0 ∂z ∂z 2 ¯z=0
Similarly, at the right hand end according to Fig.16 we have
¯
∂ 2 y(z, t) ¯¯
M(z, t)|z=l = E(z)J(z) =0 (10.39)
∂z 2 ¯z=l
¯ ¯
∂M(z, t) ¯¯ ∂ ∂ 2 y(z, t) ¯¯
V (z, t)|z=l = = EJ = ky(z, t)|z=l (10.40)
∂z ¯z=l ∂z ∂z 2 ¯z=l
Introduction of the solution 10.36 into the above boundary conditions yields
¯
YnII (z)¯z=0 = 0
¯
α YnIII (z)¯z=0 + Yn (z)|z=0 = 0
¯
YnII (z)¯z=l = 0
¯
α YnIII (z)¯z=l − Yn (z)|z=l = 0 (10.41)
where
EJ
α= (10.42)
k
The expressions for the natural modes Yn (z) and their derivatives are
Yn (z) = An sinh β n z + Bn cosh β n z + Cn sin β n z + Dn cos β n z
YnI (z) = An β n cosh β n z + Bn β n sinh β n z + Cn β n cos β n z − Dn β n sin β n z
YnII (z) = An β 2n sinh β n z + Bn β 2n cosh β n z − Cn β 2n sin β n z − Dn β 2n cos β n z

YnIII (z) = An β 3n cosh β n z + Bn β 3n sinh β n z − Cn β 3n cos β n z + Dn β 3n sin β n z

(10.43)
where
ω2n Aρ 2
β 4n =
2 = ω (10.44)
λ EJ n
Introduction of the above expressions into the boundary conditions 10.41 results in
the following set of four algebraic equations that is linear with respect to the constants
An , Bn , Cn and Dn .
β 2n Bn − β 2n Dn = 0
β 3n An + 1Bn − β 3n Cn + 1Dn = 0
An β 2n sinh β n l + Bn β 2n cosh β n l − Cn β 2n sin β n l − Dn β 2n cos β n l = 0
An β 3n α cosh β n l + Bn β 3n α sinh β n l − Cn β 3n α cos β n l + Dn β 3n α sin β n l
−(An sinh β n l + Bn cosh β n l + Cn sin β n l + Dn cos β n l) = 0
(10.45)
The matrix form of these equations is presented below
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH DISTRIBUTED
MASS 250

⎡ ⎤
0 1 0 -1 ⎡ ⎤ ⎡ ⎤
⎢ 3
β nα 1 -β 3n α 1 ⎥ An 0
⎢ ⎥ ⎢ Bn ⎥ ⎢ 0 ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢ sinh β n l cosh β n l − sin β n l − cos β n l ⎥ ⎣ Cn ⎦ = ⎣ 0 ⎦
⎣ β 3n α cosh β n l β 3n α sinh β n l −β 3n α cos β n l β 3n α sin β n l ⎦
Dn 0
− sinh β n l − cosh β n l − sin β n l − cos β n l

(10.46)

The non-zero solution of this set of equations exists if and only if its charac-
teristic determinant is equal to zero.
¯ ¯
¯ 0 1 0 -1 ¯
¯ ¯
¯ 3
βn 1 3
-β n 1 ¯
¯ ¯
¯ sinh β l cosh β l − sin β l − cos β l ¯ = 0 (10.47)
¯ n n n n ¯
¯ β α cosh β l
3
β 3
α sinh β l −β 3
α cos β l β 3
α sin β l ¯
¯ n n n n n n n n ¯
¯ − sinh β l − cosh β n l − sin β n l − cos β n l ¯
n

For k = 0 the above equation yields the characteristic equation for the free-free beam.
It is easy to show that in this case the equation 10.47 can be simplified to the following
form
cosh β n l cos β n l − 1 = 0 (10.48)
For k = ∞ we are getting the characteristic equation of the rigidly supported beam

sin β n l = 0 (10.49)

Solution of the equations 10.47, 10.48 and 10.49 for their roots are presented in
Fig.17. As one can see a replacement of the rigid supports with the elastic one results
in lowering of the natural frequencies. The first two tends to zero and their modes
tends to the ’rigid body modes’. The third mode of the rigidly supported beam tends
towards the first mode of the free-free beam.
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH DISTRIBUTED
MASS 251

arctan(k)
3.14 6.24 9.42
π
k=∞

4.7 7.8 10.9 β nl

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Figure 17
INFLUENCE OF THE SUPPORTING STRUCTURES ON A ROTOR WITH CONCENTRATED
DISK 252
10.4 INFLUENCE OF THE SUPPORTING STRUCTURES ON A RO-
TOR WITH CONCENTRATED DISK
10.5 Modelling
Let us consider rotor shown in Fig.18. It consists of the shift 1 and the rigid disk
2. The left hand end of this rotor is supported by rigid support. Between this
support and the disk there is additional flexible support 3. Properties of this flexible
support influences to large extend the dynamic behavior of the system. To analyze
this influence the physical model shown in Fig.19 was adopted.

0.5m 0.2m
A 5 A-A
1 3 2

0.02m
Ω
4 4
A

Figure 18

Ω
y1 x1 y2 x2

z
x1
x2

Rx
x y1
Ry
y y2 ϕx2

ϕy2

Figure 19
Analysis 253

The shaft 1 is massless. After introduction of the boundary condition associ-

ated with its left hand end, its matrix of stiffness along the coordinates
£ ¤
x1 x2 ϕx2
was computed to be:

⎡ ⎤
0.123991E + 07 −0.110226E + 07 151627
ks = ⎣ −0.110226E + 07 993952. −144639. ⎦ (10.50)
151627. −144639. 25433.

The disk 2 possesses the following dynamic properties

m = 6[kg], Ix = Iy = 0.081[kgm2 ], Iz = 0.081[kgm2 ], µ = 1.E − 3[m] (10.51)

where m is mass of the disk, I stands for its moments of inertia and µ represents the
unbalance.
The mathematical model of the supporting structure was adopted as follows
∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
Rx ko + ∆k kxy x1 Cx 0 ẋ1 M 0 ẍ1
= + +
Ry −kxy ko − ∆k y1 0 Cy ẏ1 0 M ÿ1
(10.52)
The mass M of the support, in the following simulation, was chosen to be small
enough (M = 1.E − 4[kg]) to influence the demonstrated in this chapter first four
natural frequencies associated with the disk.

10.6 Analysis
10.6.1 Influence of rigid support
The dynamic behavior of the rigidly supported rotor with disk was considered in
chapter ANALYSIS OF MOTION OF ROTOR WITH ONE DISK (see page 191).
To create a reference for the following analysis of influence of the supporting structure
some of the computation are repeated here for the numerical data 10.50 and 10.51.
Fig. 20 shows the imaginary parts of the eigenvalues. They can be interpreted as
the system natural frequencies. Since rigid support does not introduce damping, the
corresponding real parts are equal to zero. According to 10.51 Ix = Iy = Iz . Hence
there is only one critical speed caused by the residual imbalance. This critical speed
is the abscissa of the intersection point of the line Ω = ω n with the natural frequency
diagram. The presented in Fig.20 diagram was computed for the dynamic properties
of the supporting structure determined by the equation 10.53. Its stiffness (1.0E8) is
much lager that the stiffness of the shaft (see matrix 10.50).
∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
Rx 1.E8 0 x1 0 0 ẋ1
= + (10.53)
Ry 0 1.E8 y1 0 0 ẏ1
Analysis 254

900
n a t u r a l fr e q u e n c ie s
(im a g in a r y p a r t s o f e ig e n v a lu e s ) 800
[r a d /s]

700

600

500

400

300 Ω =ω n

200

100
C ritic a l s p e e d Ω cr

0
-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 100 200 300 400
r o ta tin g sp e e d Ω [r a d /s]

Figure 20

10.6.2 Influence of isotropic support

If kxx = kyy the supporting structure is called isotropic. Its influence on the dynamic
behavior of the rotor is shown in Fig. 21-23. The introduction of the stiffness and
damping of the supporting structure
∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
Rx 1.E5 0 x1 100 0 ẋ1
= + (10.54)
Ry 0 1.E5 y1 0 100 ẏ1

does not bring in a new qualitative phenomenons. From the Fig.21 one can see that
the natural frequencies are lower but still the system possesses only one critical speed.
Due to the positive damping the real parts of the complex eigenvalues are negative for
any angular speed Ω (see Fig.22). It follows that the equilibrium position as well as
the forced response is stable. The forced response caused by the residual imbalance
is shown in Fig.23.
600
n a t u r a l fr e q u e n c ie s
(im a g in a r y p a r ts o f e ig e n v a lu e s )
[r a d /s] 500

400

300

Ω =ω n

200

100

C ritic a l s p e e d Ω cr

0
-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 100 200 300 400
r o ta tin g s p e e d Ω [r a d /s ]

Figure 21
Analysis 255

r o ta tin g s p e e d Ω [r a d /s ]

-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 100 200 300 400

-2

-4

-6

-8

-1 0

-1 2

r e a l p a r ts o f e ig e n v a lu e s -1 4
[r a d /s]

-1 6

Figure 22

a m p litu d e [ m ]

1 .2 E -3

x 1 = y 1
x 2 = y 2

8 .0 E -4

4 .0 E -4

-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0

r o ta tin g s p e e d Ω [r a d /s ]

Figure 23

10.6.3 Influence of anisotropic support

Case kxy = −kyx = 0
Let us adopt the properties of the supporting structure as they are shown in the
expression 10.55
∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
Rx 1.5E5 0 x1 0 0 ẋ1
= + (10.55)
Ry 0 0.5E5 y1 0 0 ẏ1

The stiffness coefficients kxx and kyy are not equal to each other but the cross-
coefficients are equal to zero (kxy = 0). Because the stiffness of the supporting
structure is different in the xz plane and different in the yz plane the natural fre-
quencies in these two planes are different as well. Therefore for the rotating speed
Ω = 0 the system possesses four natural frequencies ( for the isotropic supports, due
to symmetry, the system possesses two different natural frequencies.). As can be seen
Analysis 256

from the Fig. 24 the natural frequencies associated with the positive precession are
equal to that associated with the negative precession. There are three intersection
points between the diagram of the natural frequencies and the line Ω = ωn . There-
fore the system has three critical speeds. Since the damping in this consideration was
neglected, the real parts of the eigenvalues are equal to zero.
600

n a t u r a l fr e q u e n c ie s
500 (im a g in a r y p a r t s o f e ig e n v a lu e s )
[r a d /s]

400

Ω =ω n
300

200

C ritic a l sp e e d Ω c r

100

0
-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 100 200 300 400
r o ta tin g sp e e d Ω [r a d /s]

Figure 24

The influence of the damping is illustrated by the following diagrams. Let

us introduce damping to the mathematical model of the supporting structure. Its
parameters are shown in the expression 10.56

∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
Rx 1.5E5 0 x1 150 0 ẋ1
= + (10.56)
Ry 0 0.5E5 y1 0 50 ẏ1
The introduced damping has negligible influence on the imaginary parts of the charac-
teristic roots shown in Fig.25. They represents the natural frequencies of the system.
The real parts of the characteristic roots are significantly affected by the introduced
damping. This influence is shown in Fig. 26. All real parts are negative therefore
the equilibrium position of the rotor is asymptotically stable. The critical speeds of
the rotor can be obtained by crossing the diagram of the natural frequencies with the
straight line Ω = ωn .As one can see from the diagram in Fig.25 the rotor possesses
three critical speeds. This could be confirmed by computation of the forced response
of the system due to the residual unbalance µ. Results of such a computation are
presented in Fig.27. The introduced damping limits the amplitudes of the forced vi-
brations for all angular velocities of the rotor. It should be noticed from the diagram
shown in Fig.27 that the low critical speed (about 85 rad/s see Fig.23) due to the
difference between the coefficients kxx and kyy is split into two close to each other
Analysis 257

critical speeds. The distance between them depends on this difference. Amplitudes
of the forced vibrations in the vicinity of the third ("new") critical speed (about
250 rad/s see Fig.27) are much smaller that these in vicinity of the first two critical
speeds. They depend on the difference between the coefficient kxx and kyy . For a large
difference |kxx − kyy | the amplitudes associated with this critical speed are of much
greater significance..
600
n a tu r a l fr e q u e n c ie s
(im a g in a r y p a r ts o f e ig e n v a lu e s )
500 [r a d /s]

400

Ω =ω n

300

200
C ritic a l s p e e d Ω cr

100

0
-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 100 200 300 400
r o ta tin g s p e e d Ω [r a d /s ]

Figure 25

r o ta tin g sp e e d Ω [r a d /s ]

-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 100 200 300 400

-2

-4

-6

-8

-1 0

-1 2

-1 4

-1 6 r e a l p a r ts o f e ig e n v a lu e s
[r a d /s]
-1 8

Figure 26

2 .0 E -3

a m p litu d e [m ]

1 .5 E -3

x 1
x 2
1 .0 E -3 y 1
y 2

5 .0 E -4

0 .0 E + 0
-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0
r o ta tin g sp e e d Ω [r a d /s ]

Figure 27
Analysis 258

Case kxy = −kyx 6= 0

As it was shown in section Influence of Anisotropic Structure (see Page 245) the lack
of the symmetry of the stiffness matrix causes instability of the system equilibrium
position. In case of the system considered in this section and if the damping is equal
to zero (C = 0), equilibrium position is stable for the angular velocity Ω = 0. This
case is presented in Fig. 28 and Fig.30.

∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
Rx 1.5E5 0.45E5 x1 0 0 ẋ1
= + (10.57)
Ry −0.45E5 0.5E5 y1 0 0 ẏ1

600

n a t u r a l fr e q u e n c ie s
(im a g in a r y p a r t s o f e ig e n v a lu e s )
[r a d /s]
500

400

300

200

100

0
-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 100 200 300 400
r o ta tin g s p e e d Ω [r a d /s ]

Figure 28

r e a l p a r ts o f e ig e n v a lu e s
4 [r a d /s]

0
-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0
-1
r o ta tin g sp e e d Ω [r a d /s ]

-2

-3

-4

-5

Figure 29

The existing in the supporting structure damping makes the equilibrium posi-
tion stable in a certain range of the angular velocities. The combination of the gyro-
scopic effect, damping of the supporting structure (C 6= 0) and the angular speed of
Analysis 259

the rotor is investigated here for the parameters determined by the expression 10.58.
∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
Rx 1.5E5 0.45E5 x1 150 0 ẋ1
= + (10.58)
Ry −0.45E5 0.5E5 y1 0 50 ẏ1

Results of this investigation are presented in Figures .30 to 32. It can be seen from
Fig.31 that outside the range of the angular speed from -300 to +280rad/s there
exists eigenvalue with a positive real part. It follows that the equilibrium position
outside this range is unstable. The range the angular velocity associated with a stable
equilibrium position depends on direction of rotation of the rotor. For the positive
direction the threshold of stability is lower then for the negative direction.
Number of the critical speeds is unchanged but the amplitudes of the forced
vibrations of the rotor depends on direction of rotation of the rotor (see Fig.32).

in s ta b ility re g io n s

6 0 0

n a t u r a l fr e q u e n c ie s
(im a g in a r y p a r t s o f e ig e n v a lu e s )
[r a d /s]
5 0 0

4 0 0

Ω = ω n
3 0 0

2 0 0

C ritic a l s p e e d Ω cr

1 0 0

0
-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0

r o ta tin g sp e e d Ω [r a d /s ]

Figure 30

in s ta b ility re g io n s

r o ta tin g sp e e d Ω [r a d /s ]

0
-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 100 200 300 400
-2

-4

-6

-8

-1 0

-1 2

r e a l p a r ts o f e ig e n v a lu e s
-1 4 [r a d /s]

-1 6

-1 8

Figure 31
Analysis 260

in s ta b ility re g io n s

3 .0 E -3

a m p litu d e [ m ]

2 .5 E -3

2 .0 E -3 x 1
x 2
y 1
y 2
1 .5 E -3

1 .0 E -3

5 .0 E -4

0 .0 E + 0
-4 0 0 -3 0 0 -2 0 0 -1 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0
r o ta tin g sp e e d Ω [r a d /s ]

Figure 32

10.6.4 Influence of mass of the supporting structure

If matrix of inertia of the supporting structure is not equal to zero (see expression
10.59), the system has two more degree of freedom.
∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
Rx 1.5E5 0 x1 150 0 ẋ1 10 0 ẍ1
= + +
Ry 0 0.5E5 y1 0 50 ẏ1 0 10 ÿ1
(10.59)
It follows that the number of the natural frequencies (eigenvalues) is by two larger.
This could be observed from the Fig. 33. Due to mass of the support the first four
natural frequencies (70,95,315,and 350rad/s see Fig. 25) are getting lower. Apart
of them we deal with two more natural frequencies. Furthermore the straight line
Ω = ω n is not any more an asymptote for the curve that corresponds to the fourth
natural frequency. Only the curve associated with the highest natural frequency
possesses the asymptote Ω = ω n . All others has cross-section with this line. Hence,
the system possesses five critical speeds. They are shown in Fig. 35.
Analysis 261

1500

n a tu r a l fr e q u e n c ie s
(im a g in a r y p a r ts o f e ig e n v a lu e s )
[r a d /s]

1000

Ω =ω n

500

C ritic a l s p e e d Ω cr

0
-1 0 0 0 -5 0 0 0 500 1000
r o ta tin g s p e e d Ω [r a d /s ]

Figure 33

r o ta tin g sp e e d Ω [r a d /s ]

-1 0 0 0 -5 0 0 0 500 1000

-1

-2

-3

r e a l p a r ts o f e ig e n v a lu e s
[r a d /s]

-4

Figure 34

4 .0 E - 4

a m p l i t u d e [ m ]

3 .0 E - 4

C r i t i c a l s p e e d Ω c r

2 .0 E - 4

1 .0 E - 4

0 .0 E + 0
0 5 0 0 1 0 0 0

r o t a t i n g s p e e d Ω [ r a d / s ]

Figure 35
 Chapter 11
ANALYSIS OF THE INFLUENCE OF THE SQUEEZE FILM
DAMPER ON A ROTOR VIBRATIONS

11.1 MODELLING AND ANALYSIS OF A RIGIDLY SUPPORTED

ROTOR
Due to the permanent deformation and the residual imbalance the disk of the rotor
shown in Fig.1 develops extensive vibration in the plane 6. . Result of measurement
of the amplitude of these vibration as a function of the rotating speed of the rotor is
shown in Fig. 2

1 2 3 4 5 6

Figure 1

0.0006

amplitide [m]
0.0005

0.0004

0.0003

0.0002

0.0001

0
0 100 200 300 400 500 600 700 800 900 1000

rotating speed [rad/s]

Figure 2
MODELLING AND ANALYSIS OF A RIGIDLY SUPPORTED ROTOR 263

To consider possible remedies a mathematical model of the free-free rotor was

created.
mf ẍ + kf x = 0 (11.1)
The matrix of inertia m and stiffness k of the free-free rotor was produced with help
of the FEM and then were condensed to the following vector of coordinates

x = [x1 , x2 , x3 , x4 , x5 , x6 , y1 , y2 , y3 , y4 , y5 , y6 ]T (11.2)

associated with the six planes shown in Fig. 1.

Then, the damping matrix and the excitation were introduced into the model
of the rigidly supported rotor.

mf ẍ + cf ẋ + kf x = Ff a (t)+Ff u (t) (11.3)

In the next stage, mathematical model of the rigidly supported rotor was produce by
introducing an appropriate boundary conditions into the model 11.1

mẍ + cẋ + kx = Fa (t)+Fu (t) (11.4)

Now, the damping matrix cf and the vectors Ff a and Ff u that model the permanent
deformation and unbalance of the rotor respectively were chosen by means of the
trial and error method to fit the experimental data shown in Fig. 2. The resonance
diagram produced from the mathematical model 11.3 is presented in Fig. 3. The
mode of vibrations corresponding to the critical speed Ω = 645[rad/s] is shown in
Fig. 4
The resonance diagram shown in Fig. 5 was obtained with help of the math-
ematical model 11.4 that corresponds to the rigidly supported rotor The mode of
vibrations corresponding to the critical speed is shown in Fig. 6. It can be seen that
the resonance curve that corresponds to the amplitudes of the forced vibrations along
the coordinate x6 fits very well the experimental data. The comparison between the
experimental data and the created mathematical model is presented in Fig. 7.
MODELLING AND ANALYSIS OF A RIGIDLY SUPPORTED ROTOR 264

0.004

amplitude [m]
0.0035

0.003

x1
0.0025 x2
x3
x4
0.002 x5
x6

0.0015

0.001

0.0005

0
0 100 200 300 400 500 600 700 800 900 1000
rotating speed [rad/s]

Figure 3

1
0.8
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
-0.2

-0.4
-0.6
-0.8
-1
lenght [m]

Figure 4
MODELLING AND ANALYSIS OF A RIGIDLY SUPPORTED ROTOR 265

0.0045

amplitude [m]
0.004

0.0035

0.003

x2
x3
0.0025
x4
x6

0.002

0.0015

0.001

0.0005

0
0 100 200 300 400 500 600 700 800 900 1000

rotating speed [rad/s]

Figure 5

1
0.8 amplitude
0.6
0.4
0.2
0
-0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
-0.4
-0.6
-0.8
-1
lenght [m]

Figure 6

0.0025

0.002 amplitide [m]

0.0015

0.001

0.0005

0
0 100 200 300 400 500 600 700 800 900 1000
rotating speed [rad/s]

Figure 7
MODELLING AND ANALYSIS OF THE ROTOR WITH SQUEEZE FILM DAMPER 266

Since the mathematical model 11.2 approximate very well the dynamic prop-
erties of the rotor system it can be used for the design of the squeeze film damper for
this rotor.

11.2 MODELLING AND ANALYSIS OF THE ROTOR WITH SQUEEZE

FILM DAMPER
Both bearings (in the plane 1 and 5 see Fig. 1) were furnished with an identical
damper. The drawing of the damper is shown in Fig. 8.

6 5

4
L
3

2
R
1

Figure 8

The flexible bush 3 was modelled with help of the FEM and then condensed
to two coordinates x1 , y1 .
mb ẍb +kb xb = 0 (11.5)
where

xb = [x1 , y1 ]T for the left hand side bearing

xb = [x5 , y5 ]T for the right hand side bearing (11.6)

Composition of the mathematical model of the free-free rotor 11.1 with the mathe-
matical model of the two bushes 11.5 yields the mathematical model of the rotor-bush
system
mrb ẍ + crb ẋ + krb x = Farb (t)+Furb (t) (11.7)
x = [x1 , x2 , x3 , x4 , x5 , x6 , y1 , y2 , y3 , y4 , y5 , y6 ]T (11.8)
The influence of the dynamic properties of these bushes is shown in Fig.9. Two
new low critical speed may be noticed and an increment of the critical speed from
Ω = 650[rad/s] (see Fig.3) to Ω = 740[rad/s] may be observed. The modes of
vibrations corresponding to these three critical speeds are shown in Fig.10.
MODELLING AND ANALYSIS OF THE ROTOR WITH SQUEEZE FILM DAMPER 267

0.016

0.014
amplitude [m]
0.012
x1
0.01 x2
x3
0.008 x4
x5
0.006
x6

0.004

0.002

0
0 100 200 300 400 500 600 700 800 900 1000
rotating speed [rad/s]

Figure 9

1
0.8
first mode
0.6
second mode
0.4
third mode
0.2

0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40

-0.2

-0.4

-0.6

-0.8

-1
lenght [m]

Figure 10
MODELLING AND ANALYSIS OF THE ROTOR WITH SQUEEZE FILM DAMPER 268

Parameters of the squeeze film damper adopted are as follows

R = 0.045[m] bush outer radius
L = 0.03[m] effective length of the bush
r = 2.00E − 04[m] radial clearance
2
η = 0.032[Ns/m ] viscosity of oil
According to the theory provided in the section ’Squeeze film dampers’ (see
page 1.313) the so called short bearing theory yields the approximate formulae for
the stiffness and damping coefficient as a function of the eccentricity e and speed ω.
2RL3 ηω
k = p (11.9)
r 3 (1 − 2 )
πRL3 η
c = p (11.10)
2r3 (1 − 2 )
where
ω - speed of the precession (equal to the angular velocity of the shaft Ω)
= re - eccentricity ration
e - radius of the orbit
The mathematical model of the squeeze film dampers is

cf ẋ + kf x = 0 (11.11)

where
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
k1 0 0 0 c1 0 0 0 x1
⎢ 0 k5 0 0 ⎥ ⎢ 0 c5 0 0 ⎥ ⎢ ⎥
kf = ⎢ ⎥ cf = ⎢ ⎥ xf = ⎢ x5 ⎥ (11.12)
⎣ 0 0 k1 0 ⎦ ⎣ 0 0 c1 0 ⎦ ⎣ y1 ⎦
0 0 0 k5 0 0 0 c5 y5
2RL3 ηω 1 2RL3 ηω 5 πRL3 η πRL3 η
k1 = p 2
k5 = p c1 = p c5 = p
r 3 (1 − 1) r 3 (1 − 25 ) 2r3 (1 − 21 ) 2r3 (1 − 25 )
(11.13)

The resonance diagram of the rotor supported upon the rolling elements bearing with
the damper can be produce now making use of the following subsequent approxima-
tion method
1 assume the rotating speed Ω = ω
1.1 assume arbitrarily the radius of the orbits e1 and e5
1.2 compute the eccentricities 1 and 5
1.3 compute the stiffness k1 , k5 and the damping coefficients c1 , c5 according
to 11.13
1.4 produce the matrices kf , cf according to the expressions 11.12
1.5 combine the mathematical model 11.11 with 11.7 to produce the mathe-
matical model of the whole system

Mẍ + Cẋ + Kx = Fa (t)+Fu (t) (11.14)

1.6 produce the amplitude of the forced vibrations of the system.

MODELLING AND ANALYSIS OF THE ROTOR WITH SQUEEZE FILM DAMPER 269

1.7 extract the following amplitudes

Ax1 = Ay1 = A1 and Ax5 = Ay5 = A5 (11.15)

1.8 if e1 6= A1 or e5 6= A5 then e1 = A1 and e5 = A5

1.9 repeat 1.2 to 1.8 till e1 = A1 and e5 = A5
1.10 the amplitudes A1 , A2 , A3 , A4 , A5 , A6 , computed in the last iteration
can be considered as the forced response of the system for the assumed in the step 1
rotating speed Ω = ω
Repetition of the procedure for several rotating speeds from the region 0 −
1000[rad/s] results in the resonance diagram shown in Fig. 11

x1
amplitude [m]
0.0002 x2
x3
x4
x5
x6

0.0001

0.0
0 100 200 300 400 500 600 700 800 900
rotating speed [rad/s]

Figure 11

Comparison between the resonance diagram along the coordinate 6 for rotor
with and without the squeeze film damper is display in Fig. 12

0.0025

amplitude [m]
0.002

x6 with damper
0.0015 x6 without damper

0.001

0.0005

0
0 200 400 600 800 1000
rotating speed [rad/s]

Figure 12
 Chapter 12
OPTIMIZATION OF MOTION OF ROTOR SUPPORTED UPON
MAGNETIC BEARINGS

12.1 MODELLING

1 2 3 4

Φ20
Φ45

25 100 50 250 50 100 25

600

Figure 1

The shaft whose main dimensions are shown in Fig. 1 was modelled with help of the
FEM and then condensed to four coordinates. Its matrix of inertia and stiffness are
as follows∗
⎡ ⎤
0.336238 -0.662154E -02 -0.760269E-01 0.319814E-01
⎢ -0.662149E-02 1.10615 0.182176 -0.760268E-01 ⎥
m=⎢ ⎣ -0.760267E-01
⎥
0.182176 1.10615 -0.662240E-02 ⎦
⎡ 0.319814E-01 -0.760268E-01 -0.662279E-02 0.336239 ⎤
809938. -0.139124E+07 934131 .-352786.
⎢ -0.139124E+07 0.255372E+07 -0.209682E+07 934131. ⎥
k=⎢ ⎣ 934128. -0.209681E+07 0.255379E+07 -0.139127E+07 ⎦
⎥

-352786. 934130. -0.139125E+07 809931.

The total mass of this rotor is m = 2.48[kg]. The natural modes of the free-free
rotor are presented in Fig 2.
∗
Krodkiewski J., Zmood R.B., Use of programmed magnetic bearing stiffness and damping to
minimise rotor vibration, Proceedings of the Third International Symposium on magnetic bearings,
ed Allaire E., pub. Technomic Publishing Co. Inc., pp 35-44, 1992.
MODELLING 271

0.8 first mode - 0[rad/s]

second mode 0[rad/s]
0.6 third mode 1207[rad/s]
fourth mode 3190[rad/s]
0.4

0.2

0
0 0.1 0.2 0.3 0.4 0.5 0.6
-0.2

-0.4

-0.6

-0.8

-1

Figure 2

The rotor is supported by means of two identical magnetic bearings at the

node 2 and 3. The mathematical model of the rotor in the plane xz and yz can be
adopted as follows

mẍ + kx+ = Fu cos Ωt

mÿ + ky+ = Fu sin Ωt (12.1)

where
£ ¤T £ ¤T
x= x1 x2 x3 x4 y= y1 y2 y3 y4 (12.2)
The vector Fu is the vector of the external excitation due to the residual unbalance.

Fu = mµΩ2 (12.3)

For this modelling the unbalance was adopted as a linear combination of the natural
modes with equal weights

µ = [0.00002. − 0.0000038 0.0000189 0.00002.]T (12.4)

The interaction forces Rx and Ry depend on the properties of the magnetic bearings
and their controller. The bearings used have the following parameters.
Journal diameter D = 0.045[m]
Radial stator-rotor air gap c = 0.001[m]
Bearing actuator force coefficient κ = 1.49 × 10−4 [N/(A − t)2
Control winding turns N = 140[turn]
Permanent magnet air-gap MMF ϑ = 299[A − t]
Pole face angle α = 45o
Max. control current imax = 2[A]
MODELLING 272

¡ di ¢
Max. control current slew rate dt max
= 5000[A/s]
Actuator saturation current isat = 3[A]
Bias current i0 = 0[A]
The static characteristic for this type of bearing is shown in Fig.49 (page 86) in the
non-dimensional form. It determines the static force Fs as a function of the static
position of the journal and the bias current. The non-dimensional displacement is
defined as the ratio of the dimensional displacement and the gap c

x̄ = x/c
ȳ = y/c (12.5)

The non-dimensional force is

F
F̄ = (12.6)
κϑ2
and the non-dimensional current is
Ni
ı̄ = (12.7)
ϑ
G 12.2
To balance the gravity force G = mg/2 = 12.2[N] (Ḡ = κϑ 2 = 1.49×10−4 2992 = 0.9) for

the bias current io = 0 the equilibrium position in the y direction,according to this

diagram, must be

ȳo = 0.3
yo = 0.0003[m] (12.8)

Since the static force in the x direction is equal to zero, the coordinate x of the
equilibrium position for the bias current equal to zero is

x̄o = 0
xo = 0 (12.9)

The controller was design to produce a linear dynamic characteristic around the
required equilibrium position.

Rx = kx x + cx ẋ
Ry = ky y + cy ẏ (12.10)

For the operating clearance co = 0.0002[m] (c̄o = 0.2) the region of the available coeffi-
cients kx , cx , ky and cy is presented in Fig.3 and Fig. 4 respectively.The mathematical
model of the rotor-bearing system was adopted as follows

Mẍ + Cx ẋ + Kx x+ = Fu cos Ωt
Mÿ + Cy ẏ + Ky y+ = Fu sin Ωt (12.11)

where
MODELLING 273

15

10
Cx (non-
dimensional)
5

0
-15 -10 -5 0 5 10 15
-5 Kx (non-dimensional)

-10

-15

Figure 3

15

10
Cy (non-
dimensional)
5

0
-15 -10 -5 0 5 10 15
Ky (non-dimansional
-5

-10

-15

Figure 4
ANALYSIS 274
⎡ ⎤
0.336238 -0.662154E -02 -0.760269E-01 0.319814E-01
⎢ -0.662149E-02 1.10615 0.182176 -0.760268E-01 ⎥
M=⎢⎣ -0.760267E-01
⎥
0.182176 1.10615 -0.662240E-02 ⎦
0.319814E-01 -0.760268E-01
⎡ -0.662279E-02
⎤ 0.336239
0 0 0 0
⎢ 0 cx 0 0 ⎥
Cx = ⎢ ⎣ 0 0 cx 0 ⎦
⎥

⎡ 0 0 0 0 ⎤
809938. -0.139124E+07 934131 .-352786.
⎢ -0.139124E+07 0.255372E+07+kx -0.209682E+07 934131. ⎥
Kx =⎢
⎣ 934128. -0.209681E+07 0.255379E+07+kx -0.139127E+07
⎥
⎦
-352786. 934130.⎡ -0.139125E+07
⎤ 809931.
0 0 0 0
⎢ 0 cy 0 0 ⎥
Cy = ⎢ ⎣ 0 0 cy 0 ⎦
⎥

⎡ 0 0 0 0 ⎤
809938. -0.139124E+07 934131 .-352786.
⎢ -0.139124E+07 0.255372E+07+ky -0.209682E+07 934131. ⎥
Ky =⎢
⎣ 934128. -0.209681E+07 0.255379E+07+ky -0.139127E+07
⎥
⎦
-352786. 934130. -0.139125E+07 809931.

(12.12)

12.2 ANALYSIS
The vector of amplitudes of the response of this system due to the excitation Fu is
£ ¤−1
Ax = abs( MΩ2 + iCx Ω + Kx Fu )
£ ¤−1
Ay = abs( MΩ2 + iCy Ω + Ky Fu ) (12.13)

It depends on the stiffness and damping coefficients of the magnetic bearing. In

order to search for the optimal magnitudes of the bearing’s stiffness and damping
coefficients the following goal functions was introduced
v
u 4
u1 X
gx = t (Axl )2
4 l=1
v
u 4
u1 X
gy = t (Ayl )2 (12.14)
4 l=1

where Axl and Ayl are elements of the amplitude vectors 12.13.
Such a goal function should reflect the aim of optimization. For example
if elimination of vibration of the system in a certain plane is of a greater concern
then in the others, a weighting function which weight vibrations at certain locations
differently from others may be used. If the force transmitted to the foundation is of
ANALYSIS 275

a greater importance that the amplitude of vibrations the goal function should be
created from expression that determine the interaction forces between the bearings
and the foundation.
For each rotating speed Ω from the region 0 − 1400[rad/s] the global search
for the coefficients kx and cx that minimize the goal function was carried out. Result
of this search is presented in Fig. 5.

16

14

12

10 kx (non-dimensional)
cx (non-dimensional)
ky (non-dimensional)
8
cy (non-dimensional)

0
0 200 400 600 800 1000 1200 1400

Figure 5

For the optimal stiffness k and damping c the coefficients alk can be computed
according to the formula 1.296 page 87 to accomplish the control law for the given
speed. To present benefits of this optimization the goal function for the rotor-bearing
system response was computed for constant stiffness kx = ky = 6000[N/m] and
constant damping cx = cy = 20[Ns/m]. This goal function is shown in Fig. 6 and
Fig. 7 with the broken line. The goal function computed for the optimized coefficients
k and c are shown in continuous line. The response for the optimized coefficients k
and c along the coordinates through the nodes 2 and 3 should be within the operating
clearance.
ANALYSIS 276

0.00016

amplitude [m]
0.00014

0.00012 x optimized goal function

0.0001 y optimized goal function

0.00008 x,y unoptimized goal function

0.00006

0.00004

0.00002

0
0 100 200 300 400 500

rotating speed [rad/s]

Figure 6

0.0030
amplitude [m]

0.0025

x optimized goal
function
0.0020
y optimized goal
function
0.0015 x,y unoptimized goal
function

0.0010

0.0005

0
1100 1150 1200 1250 1300

rotating speed [rad/s]

Figure 7
 Chapter 13
ANALYSIS OF THE INFLUENCE OF THE NON-LINEAR
PROPERTIES OF ELASTIC SUPPORTS ON ROTOR
VIBRATIONS.

13.1 MODELLING
13.1.1 Physical model
The main cause of the forced vibration of rotors is the residual unbalance or the
residual permanent deformation of the shaft. In many cases these sources of excitation
can be minimize by proper balancing. But there are cases when the unbalance can
not be control. For example the unbalance of the rotor of a washing machine (see
Fig.1) depends on the load distribution and in many cases is very large.

Figure 1
MODELLING 278

This, sometimes huge, unbalance results in a very large amplitude of vibration.

Therefore this type of machinery is designed to survive a massive vibrations. Usually
rotors of such machines is supported upon very flexible supports that keep the inter-
action forces on reasonably low level even if the amplitude of vibrations is big. For
such a large motion it is difficult to assume that the stiffness and damping properties
are linear. Furthermore, these elastic supports are designed to be non-linear. In this
section the influence of the non-linearity of supports on a rotor vibration is investi-
gated. Let us consider the washing machine shown in Fig.1. To analyze its vibrations
the in the vertical plane the physical model depicted in Fig. 2 was adopted.

F = m µ Ω 2c o s Ω t

k (x ) c

Figure 2

Very often the nonlinearity of the spring is approximated by the algebraic

function of the third order.
F = ko x + ∆kx3 (13.1)
Diagram of this function is shown in Fig.3

x
A

Figure 3
ANALYSIS 279

If ∆k > 0 it is said that the spring has hard characteristic. If ∆k < 0 it is

said that spring is of a soft characteristic. The soft characteristic is shown in Fig. 4.

x
A

Figure 4

13.1.2 Mathematical model

The equation of motion of the above physical model is
M ẍ + cẋ + ko x + ∆kx3 = mµΩ2 cos Ωt (13.2)

13.2 ANALYSIS
13.2.1 Approximate analytical analysis of the forced vibrations
It is a non-linear differential equation and a strict solution can not be produced by
means of the analytical methods To predict an approximate response of this system
due to the external excitation let us linearize the force-displacement relationship. One
of the possibility of the linearization is to take advantage of the criterion that the
potential energy stored in the nonlinear spring
ZA
¡ 3
¢ A2 A4
V = ko x + ∆kx dx = ko + ∆k (13.3)
2 4
0

and the one stored in the linear spring

1
V = kA A2 (13.4)
2
are equal to each other for the same amplitude of oscillations A.
A2 A4 1
ko+ ∆k = kA A2 (13.5)
2 4 2
The last equation allows the linearized stiffness kA to be produced as a function of
the amplitude A
1
kA = ko + ∆kA2 (13.6)
2
ANALYSIS 280

Such a linearization results in the following mathematical model

M ẍ + cẋ + kA x = mµΩ2 cos Ωt (13.7)
After standardization we have
ẍ + 2ξωẋ + ω2 x = q cos Ωt (13.8)
where

c
2ξω = (13.9)
M
r
kA
ω = (13.10)
M
mµΩ2
q = (13.11)
M
It must be noticed that the stiffness kA is a function of the amplitude of vibrations.
Therefore the natural frequency (that is determined by equation 13.10) is function of
the amplitude of the free vibration
r s
kA ko + 12 ∆kA2
ω= = (13.12)
M M
The forced vibration are determined by the particular solution of the equation 13.8.
x = A cos (Ωt + ϕ) (13.13)
where
q
A= q (13.14)
(ω 2 − Ω2 )2 + 4ξ 2 ω2 Ω2
Introduction of the expression 13.12 into the formula 13.14 yields
q
A = r³ ´ (13.15)
ko + 12 ∆kA2 2 k + 1 ∆kA2 2
M
− Ω2 + 4ξ 2 o 2M Ω

Solving this equation with respect to Ω one can get

s √
−b ± b2 − 4ac
Ω= (13.16)
2a
where
³ m ´2
2
a = A − µ
µ M
k 1 2 ³ c ´2 ¶
2 o + 2 ∆kA
b = A −2 +
M M
µ 1 2 ¶2
ko + 2 ∆kA
c = A2 (13.17)
M
ANALYSIS 281

The last equation allows the resonance diagram to be produced.

The following numerical data was accepted for creation of the resonance dia-
grams
M = 100kg
m = 50kg
µ = 0.005m
The resonance diagram presented in Fig. 5 corresponds to the hard characteristic of
the supporting springs determined by the following data.
ko = 1.00E + 05[N/m]
∆k = 1.00E + 09[N/m]
The soft characteristic is determined by
ko = 1.00E + 05[N/m]
∆k = −1.00E + 09[N/m]
and the resonance diagram for this characteristic is presented in Fig. 6 for different
values of damping.
The skeleton line in these diagrams presents the relationship between the nat-
ural frequency and the amplitude of vibration of the system. Not all parts of the
resonance curve represent the stable solution. It could be shown that the solution
corresponding to the amplitudes marked with the broken line are unstable. In the
regions of the rotating speed that are related to the unstable solution there are two
stable solutions. One of a large amplitude of vibrations and the other is of a small
amplitude. The initial conditions decide about the amplitude of the forced vibrations.
ANALYSIS 282

0.02
amplitude [m]
c=0[Ns/m]

0.015

c=800Ns/m]

c=1000Ns/m]
0.01

0.005

natural frequency as a function

of amplitude

0 20 40 60 80 100 120
rotating speed [rad/s]

Figure 5

0.016
amplitude [m]

0.014

0.012
c=0.0 [Ns/m]

0.01
c=500 [Ns/m]

0.008
c=1000 [Ns/m]

0.006

0.004

0.002
natural frequency as a fuction
of amplitude

0
0 20 40 60 80 100 120
rotating speed [rad/s]

Figure 6
ANALYSIS 283

The possible run up and shut down for the system with the hard and soft
characteristic is shown in Fig.7 and Fig.8 respectively.
0.02
amplitude [m]
c=0[Ns/m]

0.015

c=800Ns/m]

c=1000Ns/m]
0.01

0.005

natural frequency as a function

of amplitude

0 20 40 60 80 100 120
rotating speed [rad/s]

Figure 7

0.016
amplitude [m]

0.014

0.012
c=0.0 [Ns/m]

0.01
c=500 [Ns/m]

0.008
c=1000 [Ns/m]

0.006

0.004

0.002
natural frequency as a fuction
of amplitude

0
0 20 40 60 80 100 120
rotating speed [rad/s]

Figure 8
ANALYSIS 284

13.2.2 Numerical simulation

To assessed the accuracy of the approximate analysis of the forced vibrations of the
non-linear system, the non-linear equation was solved by means of the Runge-Kutt
method for different rotating speed of the rotor. Some of these computer simulations
are presented in Fig. 9 to Fig.12. All of the following simulation correspond to the
hard characteristic and the damping coefficient c = 800[Ns/m]
For the rotating speed Ω = 40[rad/s] (see Fig 9) and the zero initial conditions
the vibrations of the system due to the residual unbalance after a short transient state
are reaching the steady state amplitude A = 1.1E −2[m]. The amplitude of the forced
vibrations taken from the diagram in Fig. 5 is A = 1.18E − 2[m].

1.5E-2
displacement
1.0E-2 [m]

5.0E-3

0.0E+0
0 0.5 1 1.5 2
time [s]
-5.0E-3

-1.0E-2

-1.5E-2

Figure 9

For the rotating speed Ω = 45[rad/s] (see Fig.10 and Fig. 11) for the initial
conditions x|t=0 = 1.3E − 2[m] and ẋ|t=0 = 0 the steady state vibrations are of the
amplitude equal to A = 1.35E − 2[m]. This corresponds to the vibrations of the large
amplitude predicted to be A = 1.3E − 2[m] (see Fig.5).

1.5E-2

displacement
1.0E-2 [m]

5.0E-3

0.0E+0
0 0.5 1 1.5 time [s] 2

-5.0E-3

-1.0E-2

-1.5E-2

Figure 10

For the zero initial conditions (see Fig.11) the amplitude of the forced vibra-
tions is A = 6.E − 3[m]. Reading from the diagram in Fig. 5 is A = 6.1E − 3[m].
The unstable solution can not be confirmed by the numerical simulation.
ANALYSIS 285

1.0E-2

8.0E-3 displacement [m]

6.0E-3

4.0E-3

2.0E-3

0.0E+0
0 0.5 1 1.5 2
-2.0E-3 time [s]
-4.0E-3

-6.0E-3

-8.0E-3

-1.0E-2

Figure 11

The last simulation demonstrates the forced vibrations that takes place for
rotating speed larger that the critical one. The rotating speed is Ω = 50[rad/s]. The
amplitudes of the steady state vibrations taken from the diagram in Fig. 12 and that
in Fig.5 are 4.1E − 3[m] and 4.3E − 3[m]. These simulations allows to conclude that
even in case of large non-linearity the presented method can be used for prediction
of the forced vibration of an non-linear systems.

2.0E-2

1.5E-2 displacement
[m]
1.0E-2

5.0E-3

0.0E+0
0 0.5 1 1.5 time [s] 2
-5.0E-3

-1.0E-2

-1.5E-2

-2.0E-2

Figure 12

It must be stressed that such a good agreement between the non-linear sim-
ulation and the approximate solution is due to the fact that the amplitude of the
forced vibration is reasonable small.
ANALYSIS 286

Fig. 13 present the steady state vibrations caused by the unbalance µ =

0.005[m]. Spectrum of this motion is shown in Fig.14. It can be seen that the contri-
bution of sub and ultra charmonics is small.

1.5E-2

displacement
[m]
1.0E-2

5.0E-3

0.0E+0
5.0E+0 5.2E+0 5.4E+0 5.6E+0 5.8E+0 6.0E+0

-5.0E-3

-1.0E-2

-1.5E-2
time [s]

Figure 13

4.5E+0

4.0E+0

3.5E+0

3.0E+0

2.5E+0

2.0E+0

1.5E+0

1.0E+0

5.0E-1

0.0E+0
0.0E+0 5.0E+1 1.0E+2 1.5E+2 2.0E+2
-5.0E-1
frequency [rad/s]

Figure 14

Let us simulate this motion for larger unbalance µ

ANALYSIS 287

Fig. 15 present the steady state vibrations caused by the unbalance µ =

0.05[m]. Spectrum of this motion is shown in Fig.16. It can be seen that the ratio of
the first charmonic and the third one is about ten times greater. Approximation of
this motion by means of the demonstrated above linearization may turns out to be
not sufficient. In such a case it is necessary to take advantage of more sophisticated
methods∗

2.5E-2

2.0E-2 displacement
[m]
1.5E-2

1.0E-2

5.0E-3

0.0E+0
5.0E+0 5.2E+0 5.4E+0 5.6E+0 5.8E+0 6.0E+0
-5.0E-3

-1.0E-2

-1.5E-2

-2.0E-2

-2.5E-2
time [s]

Figure 15

7.0E+0

6.0E+0

5.0E+0

4.0E+0

3.0E+0

2.0E+0

1.0E+0

0.0E+0
0.0E+0 5.0E+1 1.0E+2 1.5E+2 2.0E+2
frequency [rad/s]

Figure 16

∗
eg. Krylov-Bogolubov-Mitropolski methodes
ANALYSIS 288

The simulation shown in Fig.13 and Fig. 14 was carried out for many angular
speeds of the rotor. The magnitude of the pick corresponding to the rotating speed
as a function of the rotating speed is shown in Fig. 17. It can be seen that apart
of the main resonance associated with the skeleton line started at the rotating speed
Ω = 32[rad/s] there exists a small resonance associated with the skeleton curve
started at Ω = 32/2 = 16[rad/s]. This resonance is called ultra-charmonic resonance.
Generally such a ultra-charmonic resonances can appear in vicinity of Ω = ω/N. Due
to the ever existing damping the resonances which correspond only to low numbers
N can be observed in real systems. They can not be detected by means of the
linearization method presented at the beginning of this chapter.

4.5

3.5

2.5

1.5

0.5

0
0 10 20 30 40 50 60 70 80
angularspeed [rad/s]

Figure 17
 Chapter 14
’CHATTER’ IN ROTATING MACHINERY

To control the rotating speed of elements of the rotating machinery, they are fur-
nished with such devices like brakes or clutches. They play a very important role
during the run up or shut down. These devices are particularly often met in lifting
equipment and vehicles. In all these cases the rotating element is in contact with
a motionless element. The interaction force is dependent on the relative velocity of
these two elements and the normal contact force. This interaction force in certain cir-
cumstances may damp the vibration or, in other circumstances, may promote them.
For solution of a great deal of the engineering problems it is sufficient to assume
that the relationship between the friction force and the velocity is according to the
Newton’s hypothesis. It assume that the friction force is opposite to the relative
velocity. For the relative velocity equal to zero the so call static force is different
from the friction force corresponding to the non-zero velocity. The friction force for
the non-zero velocity is referred to as the kinetic friction force and according to the
Newton’s hypothesis is constant. This hypothesis is presented in Fig.1

F(vr)

Fs
Fk

vr

Figure 1

Not all the phenomena that can be observed in a mechanical system can
be explained in terms of such an assumption. One has to remember that such a
discontinuity never appear in the real systems and the friction force is not perfectly
independent of the velocity. The relationship between the friction force and the
relative velocity shown in Fig.2 is much more closer to what can be observed in the
real mechanical system.
MODELLING 290

F(vr)

Fs
Fm Fk

vr

vrm

Figure 2

The position of the minimum of the friction force appears to play a very
important role in the dynamic behavior of the mechanical systems. With a similar
situation we deal with in machining materials. The interaction force between the
rotating piece and the cutting tool may cause heavy vibrations. This type of vibration
is often referred to as ’chatter’ and is an example of so called self-excited vibration.

14.1 MODELLING
14.1.1 Physical model
To model this type of vibration let us consider rotor system shown in Fig.3. Let
us assume that due to the force N applied to the break 5, the load 1 moves with
constant velocity. Hence the drum 2 has a constant angular velocity Ω. This drum
is connected to the brake via the elastic shaft 3 of the torsional stiffness kt . The disk
4 of the brake has the moment of inertia I. Since the shaft 3 is flexible, the disk 4
may move with respect to the drum 2. Let denote the relative angular displacement
of the disk with respect to the drum by ϕ. The absolute angular velocity of the disk
is then Ω + ϕ̇. The pad 5 of the brake is motionless. Therefore the relative velocity
between the pad and disk is equal to the absolute velocity of the rim of the disk 4.

vr = va = R(Ω + ϕ̇) (14.1)

The force F (vr ) represents the interaction force between the disk and the pad.
14.1.2 Mathematical model
Application of the generalized Newton’s equation

Iε = M (14.2)

to the disk 4 yields its equation of motion in the following form.

d
I (Ω + ϕ̇) = −kt ϕ − RF [R(Ω + ϕ̇)] (14.3)
dt
MODELLING 291

N
Ω
kt I
.
F(vr) vr=va=R( Ω +ϕ)

x
R . Ωt
Ω+ϕ

y
1 2 3 4 5

Figure 3

where
kt ϕ is the interaction moment between the shaft 3 and the disk 4
RF [R(Ω + ϕ̇)] is the interaction moment between the disk 4 and the pad 5.
After simple manipulation we are getting

I ϕ̈ + kt ϕ = −RF [R(Ω + ϕ̇)] (14.4)

Introducing substitution
z = Rϕ (14.5)
the equation of motion becomes
I kt
2
z̈ + 2 z = −F (vo + ż) (14.6)
R R
where
vo = RΩ (14.7)
or
mz̈ + kz = −F (vo + ż) (14.8)
where
I kt
m= k= (14.9)
R2 R2
MODELLING 292

14.1.3 Quantitative analysis of stability of the system equilibrium position

For ż = z̈ = 0 the equation 14.8 yields equilibrium position zo
F (vo )
zo = − (14.10)
k
Introduction of the following substitution
z = zo + ς (14.11)
allows the equation 14.8 to be rewritten as
mς̈ + kς = −F (vo + ς̇) + F (vo ) (14.12)
The above equation determine the motion of the system about the equilibrium po-
sition zo . Therefore if the system oscillate in vicinity of the equilibrium position its
energy is
1 1
E = mς̇ 2 + kς 2 (14.13)
2 2
Its time derivative is
µ ¶
d d 1 2 1 2
E= mς̇ + kς = mς̈ ς̇ + kς ς̇ (14.14)
dt dt 2 2
If the equation 14.12 is multiplied side by side by ς̇ its left hand side represents the
time derivative of the system energy. Therefore
d
E = − (F (vo + ς̇) − F (vo )) ς̇ (14.15)
dt
Case 1
Let us consider a case when
vo < vrm (14.16)
This case is illustrated in Fig.4

F(vr)

A B
. C . F(vo)
F(vo+ζ) ζ Fm

vr
vo

vrm

Figure 4
MODELLING 293

If the system stays in equilibrium the relative velocity is equal to vo and ς̇ = 0

(point A in Fig.4). If the system starts to oscillate around its equilibrium position the
relative velocity oscillates about the velocity vo within a certain range CB. At each
instant of time from the region AB magnitude of the expression (F (vo + ς̇) − F (vo )) <
0 and ς̇ > 0. Therefore the right hand side of the equation 14.15 is greater then zero.
It follows that dtd E > 0. Within the region CA (F (vo + ς̇) − F (vo )) > 0 but ς̇ < 0.
Hence again the right hand side is positive and the derivative of the energy dtd E > 0.
We can conclude that for any instant of time the energy of the system is growing.
Consequently the amplitude of velocity is getting larger. Such a situation last till the
amplitude of velocity is as large as is shown in Fig. 5

F(v r)

D A B
. C . F(v o)
F(v o+ζ) ζ Fm

vr
vo

v rm

Figure 5

Now, for the velocities from the region CB the derivative of the energy is still
positive and the energy is flowing into the system. But for velocities from region DC
(F (vo + ς̇) − F (vo )) < 0 and ς̇ < 0. Therefore the product (F (vo + ς̇) − F (vo )) ς̇ > 0.
It follows that the right hand side of the equation 14.15 is negative so in this rage
of velocity the energy is flowing out of the system. Therefore for the pick-to-pick
velocity amplitude DB the total increment of energy for a one oscillation can be
zero. Consequently the system can perform steady state vibrations with the velocity
pick-to-pick amplitude DB. It should be noticed that the velocity amplitude depends
of the velocity vo . The larger velocity vo the lager amplitude of velocity of the steady
state vibrations. We can conclude that for the velocity vo < vrm the equilibrium
position zo is unstable and the system performs a steady state vibration around it.

Case 2

Let us consider case when the velocity vo > vrm . This situation is presented
in Fig. 6.
MODELLING 294

F(v r)

B
C
. A. F(v o)
F(v o+ζ) Fm ζ
vr
vo

v rm

Figure 6

It can be seen that for any velocity from the range CB the product
(F (vo + ς̇) − F (vo )) ς̇ > 0. Therefore the derivative of energy (see equation 14.15)
is negative. Since the energy is all the time dissipated, the equilibrium position is
stable. In this case we can deal with a different scenario too. If the initial veloc-
ity in sufficiently large the system may perform large oscillations around the stable
equilibrium position. This case is illustrated in Fig. 7

F(v r)

B
. E D C A.
F(v o+ζ) Fm ζ F(v o)

vr
vo

v rm

Figure 7

14.1.4 Qualitative analysis of the system motion

The mathematical model 14.8 was solved by means of the Runge-Kutt method for
the following numerical data;
MODELLING 295

m = RI2 = 1
k = Rkt2 = 40000
The natural frequency of the shaft is
r r
k 40000
ωn = = = 200.0[rad/s] (14.17)
m 1
The friction force adopted for these computations is shown in Fig.8

1 .5

f r ic t i o n f o r c e [ N ]
1 vr [m/s] F (vr) [N]
0 0
0.0001 0.9999
0 .5
0.0092 0.6428
0.0149 0.5801
0 0.0293 0.5307
- 0 .6 - 0 .4 - 0 .2 0 0 .2 0 .4 0 .6
0.0546 0.4924
-0 .5 0.1011 0.4509
0.158 0.454
0.2052 0.4598
-1
0.589 0.557

-1 .5
r e l a t i v e v e l o c it y [ m / s ]

Figure 8

The table in Fig.8 provides with coordinates of points that belong to the
diagram of the friction force. In each step of the ’marching’ procedure of integration of
the equation 14.8 the magnitude of this force was interpolated from these coordinates.
From the Table in Fig. 8 one can see that in vicinity of velocity equal to zero the
friction force changes linearly with a large gradient ( Fv = 0.9999
0.0001
= 9999.0). This
modelling of the ’static’ friction is equivalent to connection of the disk and pad by
means of a damper with a very large damping coefficient. This approximation reflects
better the real interaction at the contact point.
The diagrams of the following simulations present:
a) the displacement z as a function of time
b) the velocity ż as a function of time
c) the motion in terms of the state-space coordinates (phase plane)
MODELLING 296

Case 1a)
velocity vo = 1.E − 03[m/s] < vrm = 0.1[m/s]
initial conditions z|t=0 = 0 ż|t=0 = vo = 0
the friction force F (vo ) = 0.92[N]
equilibrium position zo = − F (vk o ) = − 40000
0.92
= −2.3E − 05[m]

0.0E+0
1.5E-3
0.0E+0 2.0E-2 4.0E-2 6.0E-2 8.0E-2 1.0E-1 velocity [m/s]
-5.0E-6 time [s]
1.0E-3
-1.0E-5
5.0E-4
-1.5E-5
0.0E+0
-2.0E-5
0.0E+0 2.0E-2 4.0E-2 6.0E-2 8.0E-2 1.0E-1
-5.0E-4 time [s]
-2.5E-5

displacement
-3.0E-5 -1.0E-3
[m]
-3.5E-5 -1.5E-3

a) b)
1.5E-3
velocity [m/s]
1.0E-3

5.0E-4

0.0E+0
-3.5E-5 -3.0E-5 -2.5E-5 -2.0E-5 -1.5E-5 -1.0E-5 -5.0E-6 0.0E+0
displacement [m] -5.0E-4

-1.0E-3

-1.5E-3

c)

This simulation shows the pattern of the system vibrations in case of a very
small relative velocity vo . The equilibrium position is unstable and the system, after
a very short transient state, develops a periodic vibrations around this position. One
can notice two phases of this periodic motion. During one phase the absolute velocity
of the disk ω a = Ω + ϕ̇ ∼ = 0. It means that the disk is stuck to the pad. During
the other there is slipping between the disk and the pad. Therefore this mode of
oscillations very often is referred to as the ’stick-slip’ vibrations. This imperfect stick
visible from the diagram b) is caused by the fact that the friction force in vicinity of the
velocity equal to zero is modelled linearly with a very large gradient (see Table. in Fig.
8). It can be seen from Fig. a) and b) that the stable periodic motion is formed for
the velocity amplitude larger that the velocity vo = 1.E − 03[m/s] < 1.2E − 03[m/s].
The frequency of this periodic motion is equal to the natural frequency of the shaft
ωn = 200[1/s]
MODELLING 297

Case 1b)

velocity vo = 1.E − 02[m/s] < vrm = 0.1[m/s]

initial conditions z|t=0 = 0 ż|t=0 = vo = 1.E − 02[m/s]
the friction force F (vo ) = 0.6[N]
equilibrium position zo = − F (vk o ) = − 40000
0.6
= −1.5E − 05[m]

6.0E-5 1.5E-2
velocity [m/s]
4.0E-5
displacement [m] 1.0E-2

2.0E-5
5.0E-3
0.0E+0
0.0E+0 2.0E-2 4.0E-2 6.0E-2 8.0E-2 1.0E-1
0.0E+0
-2.0E-5 time[s] 0.0E+0 2.0E-2 4.0E-2 6.0E-2 8.0E-2 1.0E-1

-4.0E-5 -5.0E-3 time [s]

-6.0E-5 -1.0E-2

-8.0E-5
-1.5E-2

a) b)
1.5E-2
velocity [m/s]

1.0E-2

5.0E-3

0.0E+0
-8.0E-5 -6.0E-5 -4.0E-5 -2.0E-5 0.0E+0 2.0E-5 4.0E-5 6.0E-5

displacement [m]
-5.0E-3

-1.0E-2

-1.5E-2

c)

This simulation presents similar case to the previous one but for ten times
larger velocity vo = 1.E − 02[m/s]. Similarly, we deal with the stick-slip periodic
motion but this time the amplitude of oscillation is about ten times greater. The
frequency of oscillation is unchanged.
MODELLING 298

Case 2a)

velocity vo = 2.E − 01[m/s] > vrm = 0.1[m/s]

initial conditions z|t=0 = 0 ż|t=0 = vo = 0[ m/s]
the friction force F (vo ) = 0.456[N]
equilibrium position zo = − F (vk o ) = − 40000
0.456
= −1.14E − 05[m]

0.0E+0 3.0E-3
0.0E+0 2.0E-1 4.0E-1 6.0E-1 8.0E-1 1.0E+0
velocity [m/s]
time [s]
-5.0E-6 2.0E-3

1.0E-3
-1.0E-5

0.0E+0
0.0E+0 2.0E-1 4.0E-1 6.0E-1 8.0E-1 1.0E+0
-1.5E-5
time [s]
-1.0E-3

-2.0E-5
-2.0E-3
displacement [m]
-2.5E-5 -3.0E-3

a) b)
3.0E-3

velocity [m/s]

2.0E-3

1.0E-3

0.0E+0
-2.5E-5 -2.0E-5 -1.5E-5 -1.0E-5 -5.0E-6 0.0E+0

displacement [m]
-1.0E-3

-2.0E-3

-3.0E-3

c)

A dramatic change can be observed in the dynamic behavior of the system if

we cross the minimum point. Now could be seen from the diagrams above that the
equilibrium position is stable. For the initial conditions z|t=0 = 0 ż|t=0 = vo = 0[
m/s] the vibrations of the system slowly decay to zero. Frequency of these oscillations
again is equal to the natural frequency of the shaft. The stick-slip motion can not be
observed any more.
As it was predicted by the quantitative analysis, not always the free vibrations
tends to the stable equilibrium position. There are such initial conditions for which
the system can perform large periodic vibration. This case is demonstrated in the
following simulation.
MODELLING 299

Case 2b)

velocity vo = 2.E − 01[m/s] > vrm = 0.1[m/s]

initial conditions z|t=0 = 0 ż|t=0 = vo = 2.E − 01[m/s]
the friction force F (vo ) = 0.456[N]
equilibrium position zo = − F (vk o ) = − 40000
0.456
= −1.14E − 05[m]

1.5E-3 2.5E-1
displacement [m] velocity [m/s]
2.0E-1
1.0E-3
1.5E-1

1.0E-1
5.0E-4
5.0E-2

0.0E+0 0.0E+0
9.0E-1 9.2E-1 9.4E-1 9.6E-1 9.8E-1 1.0E+0 9.0E-1 9.2E-1 9.4E-1 9.6E-1 9.8E-1 1.0E
-5.0E-2
time [s]
-5.0E-4 time [s]
-1.0E-1

-1.0E-3 -1.5E-1

-2.0E-1

-1.5E-3
-2.5E-1

a) b)
2.5E-1
velocity [m/s]
2.0E-1

1.5E-1

1.0E-1

5.0E-2

0.0E+0
-1.5E-3 -1.0E-3 -5.0E-4 0.0E+0 5.0E-4 1.0E-3 1.5E-3
-5.0E-2
displacement [m]
-1.0E-1

-1.5E-1

-2.0E-1

-2.5E-1

c)

The diagrams above present motion of the system after a long transient state
(0.9s). As can be seen the amplitude of velocity of the steady state motion of the
system is by little greater then vo . This observation coincides with the quantitative
analysis.
 Chapter 15
ANALYSIS OF INFLUENCE OF THE ANISOTROPY OF
SUPPORTS ON MOTION OF ROTOR WITH UNSYMMETRICAL
CROSS-SECTION

15.1 MODELLING
15.1.1 Physical model

1 2 3 A-A 5
A
Ω
C

4 4

Figure 1

Let us consider rotor shown in Fig. 1 that rotates with a constant angular speed
Ω. The shaft 1 of the rotor is supported rigidly at its ends. Assume that shaft 1 is
massless and has different second moments of area of its cross-section. The element 2
of mass m is connected to the shaft via the ball bearing 3. This element is supported
by elastic elements 4 and 5 in the x and y direction.
To analyze motion of this system let us introduce the global inertial system of
coordinates XY Z as it is shown in Fig. 2.
MODELLING 301

R yR
O ,o
Sy
x O o Z , z , zR
y
RxR rG Ωt
xR
x C Ω
Sx G yR G
xR

G
yR y Y y

Figure 2

Motion of the centre of gravity G is constrained to the vertical plane xy of

the local absolute system of coordinates xyz. The system of coordinates xR , yR , zR
rotates with respect to the stationary one with the angular velocity equal to the
velocity of the rotor Ω. Its axes xR and yR are always parallel to the principal axes
of the cross-section of the shaft as shown in Fig. 2. The centre of gravity of the
element 2 coincides with the centre of the cross-section of the shaft C and its position
is determined by the position vector rG . The forces RxR and RyR components of the
reaction force that act on the element 2 due to its interaction with the shaft. They
depend on the stiffness of the shaft kx and ky in the xR and yR direction respectively.

RxR = kx xR (15.1)
RyR = ky yR (15.2)

Apart of the shaft, the element 2 interacts with the supports 4 and 5. The force S
represents this reaction acting on the element 2. Its components can be adopted in
the most general form as follows

Sx = Kx x + Kxy y + Cx ẋ (15.3)
Sy = −Kxy x + Ky y + Cy ẏ (15.4)

where Kx , Ky and Kxy as well as Cx and Cy determine the dynamic properties of the
supports 5 and 4.
If Kxy = 0 and Kx = Ky = K the support is isotropic and a strict solution may
be obtained with help of the methods provided in chapter ANALYSIS OF MOTION
OF ROTOR WITH UNSYMMETRICAL CROSS-SECTION (see page213). Such
analysis lead to conclusion that there is no qualitative differences between motion of
a rotor with rigid supports and isotropic supports.
The following modeling and analysis correspond to case when the supports
exhibits the anisotropic properties.
MODELLING 302

15.1.2 Mathematical model

Application of the Newton’s law to the element 2 yields
mẍ + kx xR cos Ωt − ky yR sin Ωt + Kx x + Kxy y + (Cx + c) ẋ = 0 (15.5)
mÿ + kx xR sin Ωt + ky yR cos Ωt − Kxy x + Ky y + (Cy + c) ẏ = G
where c represents the external damping of the shaft and according to Fig.2
xR = x cos Ωt + y sin Ωt (15.6)
yR = −x sin Ωt + y cos Ωt
Introduction of the relationship 15.6 into 15.5 after a simple manipulations yields
1 1 1
mẍ+ (kx +ky )x+ (kx − ky ) y sin 2Ωt+ (kx − ky ) x cos 2Ωt+ (Cx + c) ẋ + K x x + K xy y= 0
2 2 2
1 1 1
mÿ+ (kx +k y )y− (kx − ky ) y cos 2Ωt+ (kx − ky ) x sin 2Ωt+ (Cy + c) ẏ − K xy x + K y y= G
2 2 2
(15.7)
Introducing the following notations
1
ko = (kx + ky ) (15.8)
2
1
∆k = (kx − ky ) (15.9)
2
The equations of motion of the system considered take the form 15.7
mẍ + ko x + ∆ky sin 2Ωt + ∆kx cos 2Ωt + (Cx + c) ẋ + Kx x + Kxy y = 0
mÿ + ko y − ∆ky cos 2Ωt + ∆kx sin 2Ωt + (Cy + c) ẏ − Kxy x + Ky y = G
(15.10)
For further consideration they are arranged in the following matrix structure
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸
m 0 ẍ Cx + c 0 ẋ Kx Kxy x
+ + +
0 µ∙
m ÿ ¸ ∙ 0 Cy¸ + c ẏ ∙ −Kxy ¸ Ky ¶y∙ ¸ ∙ ¸
ko 0 0 ∆k ∆k 0 x 0
+ + sin 2Ωt + cos 2Ωt =
0 ko ∆k 0 0 −∆k y G
(15.11)
mz̈ + Cż + Kz + (ko + ∆ks sin 2Ωt + ∆kc cos 2Ωt) z = Q (15.12)
Takin advantage of the Euler formulae
ei2Ωt − e−i2Ωt
sin 2Ωt =
2i
i2Ωt
e + e−i2Ωt
cos 2Ωt = (15.13)
2
the equations of motion can be put into the following shape
µ ¶
1 i2Ωt 1 −i2Ωt
mz̈ + Cż + Kz + ko + (∆kc + i∆ks ) e + (∆kc − i∆ks ) e z=Q
2 2
(15.14)
MODELLING 303

These equations can be classified as the set of ordinary linear differential equations
with periodic coefficients. In a general case such a set of the differential equation has
no strict sulution. Here the Floquet’s theory is used to produce approximate solution
of the above set of equations.
15.1.3 Analysis of the free vibrations - stability of the equilibrium position
The free motion of the rotor is governed by the homogeneous equation 15.14. They
are as following
µ ¶
1 i2Ωt 1 −i2Ωt
mz̈ + Cż + Kz + ko + (∆kc + i∆ks ) e + (∆kc − i∆ks ) e z=0
2 2
(15.15)
The period of coefficients of the above equation is
2π π
T = = (15.16)
2Ω Ω
According to the Floquet’s theory, the particular solution of the equation 15.15can
be predicted in form of the product of a periodic function of the period T and the
exponential function ert .
z = Φ(t)ert (15.17)
Here Φ(t) = Φ(t + T ) and r is a complex number called Floquet’s exponent. If the
real part of the Floquet’s exponent is less then zero, the particular solution decay
to zero. If the real part of the Floquet’s exponent is greater then zero, than the
particular solution are growing to infinity. The free motion around the equilibrium
position is always a linear combination of such a particular solution. Therefore at
the boundary of stability of the system equilibrium position there must be particular
solution with the real part equal to zero. Hence, limited our consideration to the
stability boundary only we can predict the particular solution as follows.

z = Φ(t)eiµt (15.18)

where µ is a real number that represents the imaginary part of the Floquet’s exponent.
The periodic function Φ(t) can be sough in the following Fourier series.

X
l=L
Φ(t) = (Sl sin 2lΩt + Cl cos 2lΩt) (15.19)
l=0

Hence, at the boundary the particular solution is

X
l=L
z= (Sl sin 2lΩt + Cl cos 2lΩt) eiµt (15.20)
l=0

or, after implementation of the Euler’s formulae

X
l=+L
z= Zl ei2lΩt eiµt (15.21)
l=−L
MODELLING 304

The predicted solution 15.21 fulfills the equation 15.15 for any instant of time if the
factors associated with the following functions

.. ei(ψ−1 )t ei(ψo )t ei(ψ+1 )t ..... (15.22)

where
ψl = (2lΩ + µ) l = ...... − 1, 0, 1, ....... (15.23)
are equal to zero. This condition allows the following set of linear and homogeneous
algebraic equations to be formulated.
⎡ ⎤⎡ ⎤
. ... ... ... . .
⎢ . −ψ2−1 m+iψ −1 C+K+ko 1
(∆k − i∆k ) 0 . ⎥ ⎢ Z ⎥
⎢ 2 c s ⎥ ⎢ −1 ⎥
⎢ . 1
(∆k + i∆k ) −ψ 2
m+iψ C+K+k 1
(∆k − i∆k ) . ⎥ ⎢ Zo ⎥=
⎢ 2 c s o o o 2 c s ⎥⎢ ⎥
⎣ . 0 1
(∆k c + i∆ks ) −ψ 2
+1 m+iψ +1 C+K+k o . ⎦ ⎣ Z+1 ⎦
2
. ... ... ... . .
⎡ ⎤ ⎧
. ⎪
⎪ ...
⎪
⎪
⎢ 0 ⎥ ⎨ e( −1 )
⎪ ψ t
⎢ ⎥
=⎢⎢ 0 ⎥
⎥ e(ψo )t
⎣ 0 ⎦ ⎪
⎪
⎪
⎪
⎪ e(ψ+1 )t
. ⎩
...
(15.24)
This set has the non-zero solution if and only if its characteristic determinant is equal
to zero.
¯ ¯
¯ . ... . ¯
¯ ¯
¯ ¯
¯ 2
−ψ−1 m+iψ −1 C+K+ko 1
(∆kc − i∆ks ) 0 ¯
¯ 2 ¯
¯ 1 2 1 ¯
¯ . (∆kc + i∆ks ) −ψo m+iψ o C+K+ko (∆kc − i∆ks ) . ¯=0
¯ 2 2 ¯
¯ 0 1
(∆kc + i∆ks ) −ψ2+1 m+iψ +1 C+K+ko ¯
¯ 2 ¯
¯ ¯
¯ . ... . ¯
(15.25)
This infinite determinant has the band structure and the elements outside the diag-
onal elements are associated with the parameters ∆k. Therefore, for relatively small
∆k, it is possible to cut out limited determinant to produce the wanted boundary of
stability. The smallest determinant that is dependent on the parameter ∆k is inside
the double frame in equation 15.25. For this first approximation the characteristic
equation takes the following form.
¯ ¯
¯ −ψ2 m+iψ C+K+ko 1
(∆kc − i∆ks ) 0 ¯
¯ −1 −1 2 ¯
¯ 1 2 1 ¯
D(µ, Ω) = ¯ (∆kc + i∆ks ) −ψo m+iψ o C+K+ko (∆kc − i∆ks ) ¯=0
¯ 2 2 ¯
¯ 0 1
(∆kc + i∆ks ) −ψ+1 m+iψ +1 C+K+ko ¯
2
2
(15.26)
MODELLING 305

The elements of the determinant D(µ, Ω) are complex. Hence both the real and
imaginary part of the determinant D(µ, Ω) must be equal to zero.

Re (D(µ, Ω)) = 0
Im (D(µ, Ω)) = 0 (15.27)

The above equations form set of two equations with two unknown µ and Ω. The in-
tersection points of solution of the equation 15.27 determine the boundary of stability
(see Fig.3).

zero-points of the real part

µ
of the determinant D

zero-points of the imaginary part

of the determinant D

Floquet’s exponent

stability boundary

Figure 3
MODELLING 306

To assess accuracy of the determination of the stability boundary of the equi-

librium position let us solve the equation 15.26.in case for which the strict solution
is available.
For m = 50[kg], K = 0, ko = 11250.0[N/m], ∆k = 6750.0[N/m], c =
200[Ns/m], (ω o = 15[rad/s], m = 50[kg], ε = 0.6 hex = 2) the strict solution
was produced in chapter ANALYSIS OF MOTION OF ROTOR WITH UNSYM-
METRICAL CROSS-SECTION (page 184). It is repeated if Fig.4. The stability
boundary are 10.1 − 19.2[rad/s]. The solution of the equation 15.26.is presented in
Fig.5. Boundary of stability obtained by means of this approach are 9.6−18.4[rad/s].
As one can see, even for large ∆k = 0.6ko (ε = 0.6) the assessment of the boundary
of stability is within 5%.

40

20
ωR 0 a)
[1/s]
-20

-40

0 10 20 30 Ω[1/s]

2
0
h
[1/s] -2 b)
-4
-6
0 Ωnl 10 20 Ω u 30 Ω[1/s]
n
simple instability region

Figure 4

40

Floquet's exponent
35
[rad/s]

30

25
Re(D)
Im(D)
20

15

10

0
0 5 10 15 20 25 30 35 40
Rotating speed [rad/s]

Figure 5
MODELLING 307

Influence of anisotropy of the supporting structure - case Kxy = 0

Undamped motion (C = 0)

The solution of the equation 15.26 was produced here for the following numer-
ical data.

m = 50kg ko = 11250N/m ∆k = 6750N/m Kx = 5000N/m Ky = 30000N/m c = Cx = Cy = 0

(15.28)
It can be seen from the Fig.6 that if the damping is neglected in the mathematical
model, the zero-points of the real and the imaginary part of the equation 15.26
coincide each other. It means that for any angular speed Ω there exist solutions
that are non growing and not decaying. In this case the diagram presented in Fig. 6
can be interpreted as the diagram of the natural frequencies of the system. As one
can see from this diagram, the natural frequencies depend on the angular speed of
the rotor. For example for the rotating speed Ω = 5rad/s the frequencies of the free
motion of the shaft are

6.9 16.6 19.4 27.1 29.8 38.5 (15.29)

primary simple primary combined primary simple

instability region instability region instability region

40

Floquet's exponent
35 [rad/s] Ω =µ

30

25 Re(D)=0
Im(D)=0

20

15

10

0
0 5 10 15 20 25 30 35 40
rotating speed [rad/s]

Figure 6

To verify these natural frequencies the equations 15.15 were numerically inte-
grated by means of the Runge-Kuta method. Results of this integration is presented
in Fig. 7. Fig. 8 displays the spectrum of this free vibration. In spite of the fact
MODELLING 308

that the system possesses only two degree of freedom, the spectrum reveals at least
ten frequencies. They are

6.76 9.83 16.6 19.7 27. 29.5 37 40 (15.30)

Most of them were predicted by the fist order approximation with a very high accu-
racy. The other can be predicted by the second order approximation of the solution
of equation of motion.
2 .0 E + 0

d is p la c e m e n t [m ] x
1 .5 E + 0
y

1 .0 E + 0

5 .0 E - 1

0 .0 E + 0
1 .0 E + 0 2 .0 E + 0 3 .0 E + 0 4 .0 E + 0

-5 .0 E -1

-1 .0 E + 0

-1 .5 E + 0

-2 .0 E + 0
tim e [s]

Figure 7

4.0E +2

3.0E+2

x
y
2.0E+2

1.0E+2

0.0E +0
0.0E+0 1.0E+1 2.0E+1 3.0E+1 4.0E+1 5.0E+1
frequency [rad/s]

Figure 8

Investigating the diagram in Fig.6 it is easy to notice that there are some
regions when the number of the natural frequencies is smaller. They are highlighted
in Fig.6 In these regions the vibration with the missing frequency can not be limited
therefore they must decay or grow. For this reason inside these regions the equilibrium
position of the rotor can be unstable. The three large instability regions shown in
Fig.6 are called primary instability regions.
Magnification of the region of the rotating speed 7 to 15rad/s, shown in Fig.9
reveals more instability regions. they are called secondary instability regions.
MODELLING 309

s e c o n d a r y s im p le s e c o n d a r y c o m b in e d s e c o n d a r y s i m p le
i n s t a b i lity r e g i o n s i n s ta b ility r e g i o n s i n s ta b ilit y r e g i o n s

10

9
F lo q u e t's e x p o n e n t
[ r a d /s ]
8

R e (D )= 0
6 Im ( D ) = 0

0
7 8 9 10 11 12 13 14 15
r o ta t in g s p e e d [ r a d / s ]

Figure 9

There is a certain relation ship between the position of the instability regions
and the system natural frequencies. If ∆k is equal to zero, the system possesses two
natural frequencies
r r
ko + Kx 11250 + 5000
ω1 = = = 18.0rad/s (15.31)
m 50
r r
ko + Ky 11250 + 30000
ω2 = = = 28.7rad/s (15.32)
m 50
By inspection of the diagrams in the Fig. 6 and 9 one can see that the instability
regions shaded in the above diagrams exists in the vicinity of the rotating speed
determined by the following formulae
ωn
Ωnk = k = 1, 2, ...... (15.33)
k
In the case considered
ω1 18.0 ω2 28.7
Ω11 = = = 18.0, Ω21 = = = 28.7, (15.34)
1 1 1 1
ω1 18.0 ω2 28.7
Ω12 = = = 9.0, Ω22 = = = 14.35 (15.35)
2 2 2 2
These instability regions associated with only one natural frequency are called simple
instability regions.
A different type of instability regions can be observed in Fig.6 and 9. They
are dotted to distinguish them of the simple instability regions. They exists in the
vicinity of the rotating speeds determined by the following formula
(+) ωn + ωm
Ωnmk = k = 1, 2, ...... (15.36)
2k
MODELLING 310

In the case considered

(+) ω1 + ω2 18.0 + 28.7 (+) ω1 + ω2 18.0 + 28.7
Ω121 = = = 23.35, Ω122 = = = 11.675
2 2 2·4 4
(15.37)
The rotating speeds Ωnmk are a combination of two natural frequencies and for this
reason are called combined instability regions.
The instability regions associated with the natural numbers k higher that 2
are observed in a real systems very seldom. As it is going to be shown in the next
section, due to the ever existing damping the width of these regions is quickly reduced
to zero.
Influence of damping
To the system considered in the previous section
m = 50kg ko = 11250N/m ∆k = 6750N/m Kx = 5000N/m Ky = 30000N/m
(15.38)
the shaft external damping c and the support damping Cxx Cyy was added.
c = 20Ns/m, Cxx = 10Ns/m, Cyy = 60Ns/m (15.39)
The zero-points of the real and imaginary parts of the determinant D (see equation
15.26) are shown in Fig.10. The cross-section determine the boundary of stability.
It can be notice that the secondary instability regions do not exist any more. The
primary instability regions are slightly narrower. They are shaded in Fig.10.

prim ary sim ple prim ary com bined prim ary sim ple
instability region instability region instability region

40

F loq ue t’s exp on ent [ra d/s]

35

30

25

R e (D )= 0
20 Im (D )=0

15

10

0
0 5 10 15 20 25 30 35 40
rota ting spe ed [rad/s]

Figure 10
MODELLING 311

Outside these regions the equilibrium position should be stable. To demon-

strate this the equations of motion were integrated for the rotating speed equal to
15rad/s. Results of this integration is shown in Fig.11a). The free vibrations of the
system decay to zero. Spectrum of these vibrations is shown in Fig. 11b)
3.0E+0
displacement
[m]
2.0E+0
x 6.0E+
y 2

1.0E+0
4.0E+2

x
0.0E+0 y
0.0E+0 5.0E+0 1.0E+1 1.5E+1
a) b)
2.0E+2
-1.0E+0

0.0E+0
-2.0E+0 0.0E+0 1.0E+1 2.0E+1 3.0E+1 4.0E+1 5.0E+1
frequency [rad/s

-3.0E+0
time [s]

Figure 11

Different pattern of vibration can be observed inside the stability regions. The
following simulation corresponds to the rotating speed is equal to 16rad/s) taken from
the first simple instability region (see Fig.12). The free vibrations grow with time.
Frequency of this vibration is equal to the rotating speed as one can see from the
spectrum displayed in Fig.12b).
6.0E+1
displacement
[m]
7.0E+4
4.0E+1
x 6.0E+4
y
2.0E+1 5.0E+4 x
y
4.0E+4

a) 0.0E+0
1.0E+0 2.0E+0 3.0E+0 4.0E+0 b) 3.0E+4

2.0E+4
-2.0E+1
1.0E+4

-4.0E+1 0.0E+0
0.0E+0 1.0E+1 2.0E+1 3.0E+1 4.0E+1
frequency [rad/s]

-6.0E+1
time [s]

Figure 12

The next time history diagram (Fig.13) presents vibration of the rotor for the
rotating speed equal to 22rad/s. It is chosen from inside of the combined instability
MODELLING 312

region. The spectrum shown in Fig.13b) reveals two frequencies. They are equal to
ω1 = 17.8 and ω 2 = 26.4rad/s. It is easy to notice that ω12·1
+ω 2
= 17.8+26.4
2
= 22.1 is
equal to the rotating speed Ω121 = 22.
8.0E+0
displacement [m]
3.5E+3
6.0E+0
x 3.0E+3
y
4.0E+0
2.5E+3 x
y
2.0E+0
2.0E+3

a) 0.0E+0 1.0E+0 2.0E+0 3.0E+0 4.0E+0 b) 1.5E+3

-2.0E+0 1.0E+3

5.0E+2
-4.0E+0

0.0E+0
-6.0E+0 0.0E+0 1.0E+1 2.0E+1 3.0E+1 4.0E+1 5.0E+1

frequency [rad/s]

-8.0E+0
time [s]

Figure 13

For the rotating speed taken from the second simple instability region Ω21 =
29rad/s (see Fig.4) the pattern of vibrations is similar to the one shown in Fig.12.
The frequency of oscillations that grows in time is equal to the rotating speed.
1.5E+1
displacement [m]
1.8E+4
1.0E+1
1.6E+4
x
y 1.4E+4
5.0E+0
1.2E+4
x
1.0E+4
a) 0.0E+0
1.0E+0 2.0E+0 3.0E+0 4.0E+0 b) 8.0E+3 y

6.0E+3
-5.0E+0
4.0E+3

2.0E+3
-1.0E+1
0.0E+0
0.0E+0 1.0E+1 2.0E+1 3.0E+1 4.0E+1
frequency [rad/s]
-1.5E+1
time [s]

Figure 14

The results of the computation depicted in Fig. 15 corresponds to the damp-

ing determined by the following parameters c = 20Ns/m Cxx = 20Ns/m Cyy =
120Ns/m. This level of damping eliminates the higher simple instability region. and
results in further narrowing the two still existing instability regions.
MODELLING 313

primary simple primary combined

instability region instability region

40

Floquet’s exponent [rad/s]

35

30

25 Re(D)=0
Im(D)=0
20

15

10

0
0 5 10 15 20 25 30 35 40

rotating speed [rad/s]

Figure 15

When the damping were increased to c = 20Ns/m Cxx = 80Ns/m and Cyy =
480Ns/m the combined instability region was reduced to zero. This case is shown in
Fig.16.
MODELLING 314

primary simple
instability region

40

35 Floquet’s exponent [rad/s]

30

25
Re(D)=0
Im(D)=0
20

15

10

0
0 5 10 15 20 25 30 35 40
rotating speed [rad/s]

Figure 16

The damping parameters c = 20Ns/m Cxx = 320Ns/m and Cyy = 960Ns/m

results in stabilization of the equilibrium position for all rotating speeds of the rotor
(see Fig.17).
MODELLING 315

40

Floquet’s exponent [rad/s]

35

30

Re(D)=0
25 Im(D)=0

20

15

10

0
0 5 10 15 20 25 30 35 40
rotating speed [rad/s]

Figure 17

15.1.4 Analysis of the forced vibrations

Response due to gravitation
To analyze the forced vibrations caused by the gravitation let us recall the equation
15.14

µ ¶
1 i2Ωt 1 −i2Ωt
mz̈ + Cż + Kz + ko + (∆kc + i∆ks ) e + (∆kc − i∆ks ) e z=Q
2 2
(15.40)
where according to 15.11
∙ ¸ ∙ ¸ ∙ ¸
m 0 Cx + c 0 Kx Kxy
m = ; C= ; K=
0 m 0 Cy + c −Kxy Ky
∙ ¸ ∙ ¸ ∙ ¸
ko 0 0 ∆k ∆k 0
ko = ; ∆kc = ; ∆ks =
0 ko ∆k 0 0 −∆k
∙ ¸ ∙ ¸
x 0
z = ; Q= (15.41)
y G

The vector Q represents the forces caused by gravitation. Since the equation 15.40
is linear, the forced vibrations caused by the gravitation can be approximated by the
MODELLING 316

particular solution of the non-homogeneous equation. Mathematics offers no methods

for searching for the strict particular solution. Here, so called small parameter method
is used to produce an approximated particular solution.
To take advantage of this method we have to assume that the periodic coeffi-
cients (∆kc and ∆ks ) are small with respect to the time independent ones. Therefore
the equation 15.40 can be rewritten as follows
µ ¶
1 i2Ωt 1 −i2Ωt
mz̈ + Cż + (K + ko ) z + (∆kc + i∆ks ) e + (∆kc − i∆ks ) e z=Q
2 2
(15.42)
where stands a parameter that is small enough to arrange for the method of solution
to be converged.
The particular solution is sought in the following form

X
N
n
z= zn (t) (15.43)
n=0

where zn (t) are unknown functions of time. To figure them out let us introduce the
predicted solution 15.43 into the equation 15.42 and equate terms associated with
the same power of the small parameter . As a result of these manipulations one can
obtained the following set of recurrent equations

mz̈0 +Cż0 + (K+ko ) z0 = Q ............................................................................... 0

1¡ ¢
mz̈1 +Cż1 + (K+ko ) z1 = − (∆kc +i∆ks ) ei2Ωt + (∆kc −i∆ks ) e−i2Ωt z0 ....... 1
2
1¡ ¢
mz̈2 +Cż2 + (K+ko ) z2 = − (∆kc +i∆ks ) ei2Ωt + (∆kc −i∆ks ) e−i2Ωt z1 ....... 2
2
...................................
................................... (15.44)

The first equation yields the function z0

(0)
z0 = Z0 (15.45)
(0)
Z0 = (K+ko )−1 Q (15.46)
Introduction of this function into the second equation of the set 15.44 allows the right
hand side of this equation to be expressed as an explicit function of time.
(+2) i2Ωt (−2) −i2Ωt
mz̈1 +Cż1 + (K+ko ) z1 = F1 e + F1 e (15.47)

(+2) 1 (0)
F1 = − (∆kc +i∆ks ) Z0
2
(−2) 1 (0)
F1 = − (∆kc −i∆ks ) Z0 (15.48)
2
MODELLING 317

The particular solution of the above equation is

(+2) (−2)
z1 = Z1 ei2Ωt + Z1 e−i2Ωt (15.49)
(+2) ¡ ¢−1 (+2)
Z1 = −4Ω2 m + 2ΩiC + (K+ko ) F1 (15.50)
(−2) ¡ ¢−1 (−2)
Z1 = −4Ω2 m − 2ΩiC + (K+ko ) F1 (15.51)
To get the third term of the particular solution let us introduce the solution 15.49
into the third equation of 15.44

mz̈2 +Cż2 + (K+ko ) z2 =

¡ ¢ ³ (+2) (−2)
´
= − 12 (∆kc +i∆ks ) ei2Ωt + (∆kc −i∆ks ) e−i2Ωt Z1 ei2Ωt + Z1 e−i2Ωt
(15.52)
After a simple manipulation one can obtained the following differential equation

mz̈2 +Cż2 + (K+ko ) z2 =

(+2) (−2)
= − 12 ((∆kc −i∆ks ) Z1 + (∆kc +i∆ks ) Z1 (15.53)
(+2) (−2)
+ (∆kc +i∆ks ) Z1 ei4Ωt + (∆kc −i∆ks ) Z1 e−i4Ωt )

It can be rewritten in the following, more compact, form

(0) (+4) i4Ωt (−4) −i4Ωt
mz̈2 +Cż2 + (K+ko ) z2 = F2 + F2 e + F2 e (15.54)

where
(0) 1³ (+2) (−2)
´
F2 = − + (∆kc −i∆ks ) Z1 + (∆kc +i∆ks ) Z1
2
(+4) 1 (+2)
F2 = − (∆kc +i∆ks ) Z1 (15.55)
2
(−4) 1 (−2)
F2 = − (∆kc −i∆ks ) Z1
2
The particular solution of the above equation is
(0) (+4) i4Ωt (−4) −i4Ωt
z2 = Z2 + Z2 e + Z2 e (15.56)

where
(0) (0)
Z2 = (K+ko )−1 F2
(+4) ¡ ¢−1 (+4)
Z2 = −16Ω2 m + 4ΩiC + (K+ko ) F2 (15.57)
(−4) ¡ ¢−1 (+4)
Z2 = −16Ω2 m − 4ΩiC + (K+ko ) F2

Neglecting terms of order higher or equal then 3 the particular solution according to
15.43 is
z = 0 z0 + 1 z1 + 2 z2 + ...... (15.58)
Since in the original equation 15.40

=1 (15.59)
MODELLING 318

its solution is
³ ´
(0) (0) (+2) (−2) (+4) (−4)
z = z0 + z1 + z2 .... = Z0 + Z2 + Z1 ei2Ωt + Z1 e−i2Ωt + Z2 ei4Ωt + Z2 e−i4Ωt
(15.60)
According to Euler’s formula we have that
³ ´ ³ ´
(+2) i2Ωt (−2) −i2Ωt (+2) (−2) (+2) (−2)
Z1 e +Z1 e = Z1 + Z1 cos 2Ωt+i Z1 − Z1 sin 2Ωt (15.61)
³ ´ ³ ´
(+4) i4Ωt (−4) −i4Ωt (+4) (−4) (+4) (−4)
Z2 e +Z2 e = Z1 + Z1 cos 4Ωt+i Z1 − Z1 sin 4Ωt (15.62)
Implementation of the expressions 15.62 and 15.63 into 15.60 yields the solution as
follows
³ ´ ³ ´ ³ ´
(0) (0) (+2) (−2) (+2) (−2)
z = Z0 + Z2 + Z1 + Z1 cos 2Ωt + i Z1 − Z1 sin 2Ωt +
³ ´ ³ ´
(+4) (−4) (+4) (−4)
+ Z1 + Z1 cos 4Ωt + i Z1 − Z1 (15.63)
sin 4Ωt

In spite of the fact that the matrices Z are generally complex the factors in front of
the function of time are real. It is obvious if one take into account that z is solution
of a real differential equation. Therefore the solution above can be rewritten in the
following form
" # ⎡ (2) ³
(2)
´ ⎤ ⎡
(4)
³
(4)
´ ⎤
Ax
(0) A x cos 2Ωt + ϕ x A x cos 4Ωt + ϕx
z= (0) + ⎣ (2) ³
(2)
´ ⎦+⎣
(4)
³
(4)
´ ⎦ (15.64)
Ay Ay cos 2Ωt + ϕy Ay cos 4Ωt + ϕy

where
" #
(0)
Ax (0) (0)
(0) = Z0 + Z2 (15.65)
Ay
" # r³
(2)
Ax ´2 h ³ ´i2
(+2) (−2) (+2) (−2)
(2) = Z1 + Z1 + i Z1 − Z1 (15.66)
Ay
" # r³
(4)
Ax ´2 h ³ ´i2
(+4) (−4) (+4) (−4)
(4) = Z1 + Z1 + i Z1 − Z1 (15.67)
Ay
" #
(0)
Ax
The first term (0) in this solution determines the equilibrium position of the
Ay
rotor system due to the gravity force. Around this equilibrium the rotor performs the
forced vibrations with frequencies that, in general, are products of the angular speed
and even natural numbers (2, 4, .....).

ωk = kΩ k = 2, 4, ......... (15.68)

The formulae 15.66 and 15.67 identify the amplitudes of these vibrations as a function
of the angular velocity of the rotor Ω.
MODELLING 319

The presented here diagrams are produced for the numerical data used in the
previous section (see 15.38 and 15.39).
m = 50kg ko = 11250N/m ∆k = 6750N/m Kx = 5000N/m Ky = 30000N/m Kxy = 0
c = 20Ns/m, Cxx = 10Ns/m, Cyy = 60Ns/m (15.69)
Hence∙ the matrices
¸ involved
∙ in the mathematical
¸ ∙ model ¸considered∙are ¸
50 0 5000 0 30 0 11250 0
m= K= C= ko =
∙0 50 ¸ 0 30000
∙ ¸ 0 80∙ ¸ 0 11250
6750 0 0 6750 0
∆kc = ∆ks = Q=
0 −6750 6750 0 500
The computed in the previous section natural frequencies are
r r
ko + Kx 11250 + 5000
ω1 = = = 18.0rad/s
m 50
r r
ko + Ky 11250 + 30000
ω2 = = = 28.7rad/s (15.70)
m 50
The amplitudes of the forced vibrations of the centre of the rotor along axes x and y
(see Fig. 2) are depicted in Fig. 18. By inspection of this diagram one can arrived
to conclusion that the critical speeds of the system are dependent on the natural
frequencies and are determined by the following formula
(k) ωn
Ωcr = k = 2, 4, ....... (15.71)
k
The significance of these critical speeds is getting lower with increment of the number
k.

Response due to the residual unbalance

Let us consider a case when the centre of the shaft C does not coincides with the
centre of gravity G. This situation is shown in Fig. 19. The components of the
residual unbalance µx and µy identify position of the centre of gravity with respect
to the rotating system of coordinates xR yR .

R yR
Sy O Z , z , zR
O ,o x o
y
RxR rG Ωt
xR
x C C Ω
Sx yR
xR α
µ µy G
µx
G µxmΩ 2

µymΩ 2
yR y µmΩ2 Y y
G

Figure 19
MODELLING 320

0.1 Ax
(2)
Ax(2) Ay(2)
(2)
Ay
Ax(4)
Ay(4)

Ay(4)
0.05 Ax(4)

0
0 10 20 Ω rad/s 30 40

ω1 ω2

ω1 /4 ω2 /2

ω2 /4 ω1 /2

Figure 18

The centrifugal force µmΩ2 produces two components along the stationary
system of coordinates xy.
qx = µmΩ2 cos (Ωt + α) = µx mΩ2 cos Ωt − µy mΩ2 sin Ωt
qy = µmΩ2 sin (Ωt + α) = µx mΩ2 sin Ωt + µy mΩ2 cos Ωt (15.72)
Replacement of the trigonometric functions by their Euler equivalent results in the
following real components expressed by complex functions.
iΩt
+ e−iΩt
2e 2e
iΩt
− e−iΩt
qx = µx mΩ − µy mΩ
2 2i
mΩ2 ¡ ¢ iΩt mΩ2 ¡ ¢
= µx + iµy e + µx − iµy e−iΩt
2 2
iΩt −iΩt iΩt
2e −e 2e + e−iΩt
qy = µx mΩ + µy mΩ (15.73)
2i 2
mΩ2 ¡ ¢ iΩt mΩ2 ¡ ¢
= −iµx + µy e + iµx + µy e−iΩt
2 2
Now, they can be arranged in a matrix form
∙ ¸
qx
= q(+1) eiΩt + q(−1) e−iΩt (15.74)
qy
MODELLING 321

where
∙ ¡ ¢ ¸ ∙ ¡ ¢ ¸
(+1) mΩ2 µ¡x + iµy ¢ (−1) mΩ2 ¡µx − iµy ¢
q = q = (15.75)
2 −i µx + iµy 2 i µx − iµy

This vector of excitation 15.74 has to be included in the right hand side of the equation
15.14

¡1 ¢
mz̈ + Cż + (K + ko ) z + 2
(∆kc + i∆ks ) ei2Ωt + 12 (∆kc − i∆ks ) e−i2Ωt z =
= Q + q(+1) eiΩt + q(−1) e−iΩt
(15.76)
where according to 15.11
∙ ¸ ∙ ¸ ∙ ¸
m 0 Cx + c 0 Kx Kxy
m = ; C= ; K=
0 m 0 Cy + c −Kxy Ky
∙ ¸ ∙ ¸ ∙ ¸
ko 0 0 ∆k ∆k 0
ko = ; ∆kc = ; ∆ks =
0 ko ∆k 0 0 −∆k
∙ ¸ ∙ ¸
x 0
z = ; Q= (15.77)
y G

Leaving in the right hand side of the above equation terms associated with the exci-
tation due to the residual unbalance only, we have
¡ ¢
mz̈ + Cż + (K + ko ) z + 12 (∆kc + i∆ks ) ei2Ωt + 12 (∆kc − i∆ks ) e−i2Ωt z =
= q(+1) eiΩt + q(−1) e−iΩt
(15.78)
The particular solution of the nonhomogeneous equation 15.78 approximate response
of the system due to its unbalance. To figure it out let us utilize the small parameter
method described in the previous section. To this end let us introduce the small
parameter
¡ ¢
mz̈ + Cż + (K + ko ) z + 12 (∆kc + i∆ks ) ei2Ωt + 12 (∆kc − i∆ks ) e−i2Ωt z =
= q(+1) eiΩt + q(−1) e−iΩt
(15.79)
and predict the particular solution as follows

X
N
n
z= zn (t) (15.80)
n=0

where zn (t) are unknown functions of time. Introducing of the solution 15.80 into
the equation 15.79 and equating terms associated with the same power of the small
parameter one can get the following set of recurrent equations
MODELLING 322

mz̈0 +Cż0 + (K+ko ) z0 = q(+1) eiΩt + q(−1) e−iΩt ......................................................... 0

1¡ ¢
mz̈1 +Cż1 + (K+ko ) z1 = − (∆kc +i∆ks ) ei2Ωt + (∆kc −i∆ks ) e−i2Ωt z0 ....... 1
2
1¡ ¢
mz̈2 +Cż2 + (K+ko ) z2 = − (∆kc +i∆ks ) ei2Ωt + (∆kc −i∆ks ) e−i2Ωt z1 ....... 2
2
...................................
................................... (15.81)

The first equation yields the function z0

(+1) iΩt (−1) −iΩt
z0 = Z0 e + Z0 e (15.82)

where
(+1) ¡ 2 ¢−1 (+1)
Z0 = −Ω m + ΩiC + (K+ko ) q
(−1) ¡ 2 ¢−1 (−1)
Z0 = −Ω m − ΩiC + (K+ko ) q (15.83)

Introduction of this function into the second equation of the set 15.81 allows the right
hand side of this equation to be expressed as an explicit function of time.
(+1) iΩt (−1) −iΩt (+3) i3Ωt (−3) −i3Ωt
mz̈1 +Cż1 + (K+ko ) z1 = F1 e + F1 e + F1 e + F1 e (15.84)

where
(+1) 1 (−1)
F1 = − (∆kc +i∆ks ) Z0
2
(−1) 1 (+1)
F1 = − (∆kc −i∆ks ) Z0
2
(+3) 1 (+1)
F1 = − (∆kc +i∆ks ) Z0
2
(−3) 1 (−1)
F1 = − (∆kc −i∆ks ) Z0 (15.85)
2
The particular solution of the above equation is
(+1) iΩt (−1) −iΩt (+3) i3Ωt (−3) −i3Ωt
z1 = Z1 e + Z1 e + Z1 e + Z1 e (15.86)

where
(+1) ¡ 2 ¢−1 (+1)
Z1 = −Ω m + ΩiC + (K+ko ) F1
(−1) ¡ 2 ¢−1 (−1)
Z1 = −Ω m − ΩiC + (K+ko ) F1
(+3) ¡ ¢−1 (+3)
Z1 = −9Ω2 m + 3ΩiC + (K+ko ) F1
(−3) ¡ ¢ −1 (−3)
Z1 = −9Ω2 m − 3ΩiC + (K+ko ) F1 (15.87)

Neglecting terms of order higher or equal then 2 the particular solution according to
15.80 is
z = 0 z0 + 1 z1 + ...... (15.88)
MODELLING 323

Since in the original equation 15.40

=1 (15.89)

its solution is
³ ´ ³ ´
(+1) (+1) (−1) (−1) (+3) (−3)
z = z0 +z1 +.... = Z0 + Z1 eiΩt + Z0 + Z1 e−iΩt +Z1 ei3Ωt +Z1 e−i3Ωt
(15.90)
In similar manner as it was done in the previous section the above solution can be
transformed to the following form
⎡ ³ ´ ⎤ ⎡ ³ ´ ⎤
(1) (1) (3) (3)
Ax cos Ωt + ϕx Ax cos 3Ωt + ϕx
z = ⎣ (1) ³
(1)
´ ⎦+⎣
(3)
³
(3)
´ ⎦ (15.91)
Ay cos Ωt + ϕy Ay cos 3Ωt + ϕy
" # r³
(1)
Ax ´2 h ³ ´i2
(+1) (+1) (−1) (−1) (+1) (+1) (−1) (−1)
(1) = Z0 + Z1 + Z0 + Z1 + i Z0 + Z1 − Z0 − Z1
Ay
" # r³ (15.92)
(3)
Ax ´2 h ³ ´i2
(+3) (−3) (+3) (−3)
(3) = Z1 + Z1 + i Z1 − Z1 (15.93)
Ay
It can be seen from the equation 15.90 that the response of the system to the residual
unbalance is not harmonic. It is a superposition of the harmonic terms of frequency
being product of the working angular velocity of the rotor and the odd natural num-
bers.
ω k = kΩ k = 1, 3, ..... (15.94)
The formulae 15.92 and 15.93 allow for determination of the amplitudes of the indi-
vidual terms. They are presented in Fig. 20 as a function of the angular velocity Ω.
It can be noticed from this diagrams that the system possesses several critical speeds.
The are associated with the natural frequencies of the system.
r r
ko + Kx 11250 + 5000
ω1 = = = 18.0rad/s
m 50
r r
ko + Ky 11250 + 30000
ω2 = = = 28.7rad/s (15.95)
m 50
In general the critical speeds are equal to
(k) ωn
Ωcr = k = 1, 3, ....... (15.96)
k
The critical speeds associated with
k = 1 are called primary critical speeds
k = 2 are called secondary critical speeds
k = 3 are called tertiary critical speeds
MODELLING 324

Ay(1)
0.1 Ax
(1) Ax(1)
Ay(1)
Ax(3)
Ay(3)

Ay(3)

0.05
Ax(3)

0
0 10 20 Ω rad/s 30 40

ω1 ω2
ω1 /3 ω2 /3

Figure 20
 Part III
EXPERIMENTAL
INVESTIGATIONS

325
 Chapter 16
INFLUENCE OF THE GYROSCOPIC EFFECT ON A ROTOR
VIBRATIONS

16.1 DESCRIPTION OF THE LABORATORY INSTALLATION

F1
8 16 10 6 5 4 3 1 5 2 11 7

d2
c

r
Ω b

b/2
F2
a=0.015m L

Ω Ω d1 d2
[Hz] [V] [V] [Hz] [V]

9 17 12 13 14 18

Figure 1

The disk 2 of the rotor shown in Fig. 1 is supported by means of the self-aligning
bearing. The left hand side of the rotor is supported upon the flexible support which
is assembled of the block 3 springs 4 and the self-aligning ball bearing. The shaft 1
of the rotor is driven by the electric motor 8. Its angular velocity can be controlled
by the speed regulator 9.
MODELLING 327

16.2 MODELLING
16.2.1 Physical model

3 1 2

z
x ϕx

m I,Iz
x
z
y ϕy

Figure 2

The physical model of the described above rotor system is shown in Fig. 2.
The disk 2 is considered as a rigid body. Its moments of inertia about axes through
the point of rotation O are Ix = Iy = I and Iz . The block 3 can translate in plane
xy and its mass is m. The shaft 1 is assumed to be massless. The rotor rotates with
the constant angular velocity Ω. Upon adopting the above assumptions, the system
has four degree of freedom. Therefore its instantaneous position with respect to the
inertial system of coordinates xyz is determined by four generalized coordinates x, y,
ϕx and ϕy .
16.2.2 Mathematical model

z
ϕx
I,Iz
xd

Figure 3

The mathematical model of the disk shown in Fig. 3 can be adopted as follows

md ẍd + gd ẋd = Rd +FGd +FUd (16.1)

MODELLING 328

where:
⎡ ⎤
m 0 ma 0
⎢ 0 m 0 ma ⎥
md = ⎢
⎣ ma 0 Ip + ma 2
⎥
⎦
0
2
0 ma 0 Ip + ma
⎡ ⎤
0 0 0 0
⎢ 0 0 0 0 ⎥
cd = ⎢
⎣ 0 0
⎥
0 Iz Ω ⎦
0 0 −Iz Ω 0
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
xd Fx 0
⎢ yd ⎥ ⎢ Fy ⎥ ⎢ mg ⎥
xd = ⎢ ⎥ ⎢ ⎥
⎣ ϕx ⎦ , Rd = ⎣ Mx ⎦ , FGd = ⎣ 0 ⎦
⎢ ⎥

ϕy My mga
⎡ ⎤ ⎡ 2
⎤
U Fx SΩ cos(Ω t + ϕS )
⎢ UFy ⎥ ⎢ SΩ2 sin(Ω t + ϕS ) ⎥
FU d = ⎢ ⎥ ⎢
⎣ UMx ⎦ = ⎣ DΩ2 cos(Ω t + ϕD ) ⎦
⎥ (16.2)
UMy DΩ2 sin(Ω t + ϕD )

In the case considered a = 0 and xd = 0 as well as yd = 0. Therefore, the

matrices 16.2 can be simplified to the following form:
∙ ¸
Ip 0
md =
0 Ip
∙ ¸
0 Iz Ω
cd =
−Iz Ω 0
∙ ¸ ∙ ¸ ∙ ¸
ϕx Mx 0
xd = , Rd = , FGd =
ϕy My 0
∙ ¸ ∙ 2
¸
UMx DΩ cos(Ω t + ϕD )
FU d = = (16.3)
UMy DΩ2 sin(Ω t + ϕD )

3
z
x k

Figure 4

The equation of motion of the block 3, shown in Fig. 4can be adopted in the
following form
MODELLING 329

mẍ + kx = Rx
mÿ + ky = Ry (16.4)

Upon combining these two mathematical models together one may obtain
⎡ ⎤⎡ ⎤ ⎡ ⎤⎡ ⎤
m 0 0 0 ẍ 0 0 0 0 ẋ
⎢ 0 I 0 0 ⎥ ⎢ ϕ̈x ⎥ ⎢ 0 0 0 Iz Ω ⎥ ⎢ ⎥
⎢ ⎥⎢ ⎥+⎢ ⎥ ⎢ ϕ̇x ⎥
⎣ 0 0 m 0 ⎦ ⎣ ÿ ⎦ ⎣ 0 0 0 0 ⎦ ⎣ ẏ ⎦
0 0 0 I ϕ̈y 0 −Iz Ω 0 0 ϕ̇y
⎡ ⎤⎡ ⎤ ⎡ ⎤ ⎡ ⎤
k 0 0 0 x Rx 0
⎢ 0 0 0 0 ⎥ ⎢ ϕx ⎥ ⎢ Mx ⎥ ⎢ DΩ2 cos(Ω t + ϕD ) ⎥
+⎢ ⎥⎢ ⎥ ⎢
⎣ 0 0 k 0 ⎦ ⎣ y ⎦ = ⎣ Ry ⎦ + ⎣
⎥ ⎢ ⎥
⎦ (16.5)
0
2
0 0 0 0 ϕy Mx DΩ sin(Ω t + ϕD )
The vector of interaction forces between the rigid elements 2 and 3 and the massless
shaft 1 can be always approximated by product of the unknown stiffness matrix and
the vector of the generalized coordinates.
⎡ ⎤ ⎡ ⎤⎡ ⎤
Rx k11 k12 0 0 x
⎢ Mx ⎥ ⎢ k21 k22 0 0 ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ϕx ⎥
⎣ Ry ⎦ = − ⎣ 0 0 k11 k12 ⎦ ⎣ y ⎦
(16.6)
Mx 0 0 k21 k22 ϕy

Introduction of Eq.. 16.6 into 16.5 yields equation of motion of the system in the
following form
⎡ ⎤⎡ ⎤ ⎡ ⎤⎡ ⎤
m 0 0 0 ẍ 0 0 0 0 ẋ
⎢ 0 I 0 0 ⎥ ⎢ ϕ̈x ⎥ ⎢ 0 0 0 Iz Ω ⎥ ⎢ ⎥
⎢ ⎥⎢ ⎥+⎢ ⎥ ⎢ ϕ̇x ⎥
⎣ 0 0 m 0 ⎦ ⎣ ÿ ⎦ ⎣ 0 0 0 0 ⎦ ⎣ ẏ ⎦
0 0 0 I ϕ̈y 0 −Iz Ω 0 0 ϕ̇y
⎡ ⎤⎡ ⎤ ⎡ ⎤
K11 K 0 0 x 0
⎢ K K22 0 0 ⎥ ⎢ ⎥ ⎢ 2 ⎥
+⎢ ⎥ ⎢ ϕx ⎥ = ⎢ DΩ cos(Ω t + ϕD ) ⎥ (16.7)
⎣ 0 0 K11 K ⎦ ⎣ y ⎦ ⎣ 0 ⎦
0 0 K K22 ϕy DΩ2 sin(Ω t + ϕD )
Implementation of the complex function

z = x + iy
ϕz = ϕx + iϕy (16.8)

allows the equations 16.7 to be simplified.

mz̈ + K11 z + Kϕz = 0

I ϕ̈z + Kz + K22 ϕz − iIz Ωϕ̇z = Dei(Ω t+ϕD ) (16.9)
ANALYSIS 330

16.3 ANALYSIS
16.3.1 Natural vibrations
If we neglect the excitation forces the equations 16.9 becomes homogeneous and
govern natural (free) motion of the rotor system.

mz̈ + K11 z + Kϕz = 0

I ϕ̈z + Kz + K22 ϕz − iIz Ωϕ̇z = 0 (16.10)

Its particular solution may be predicted in the following form

z = Aeiωt
ϕz = Beiωt (16.11)

In the above solution α stands for the natural frequency and A and B are time inde-
pendent complex magnitudes which have to be chosen to fulfill the initial conditions.
Introduction of Eq. 16.11 into Eq. 16.10 yields the following set of the homogeneous
algebraic equations linear with respect to the constants A and B.

(−mω2 + K11 )A + ( K )B = 0
2
( K )A + (−Iα + Iz Ωω + K22 )B = 0 (16.12)

The non-trivial solution of the above set of equation exists if and only if the charac-
teristic determinant is equal to zero.
¯ ¯
¯ −mω 2 + K11 K ¯
¯ ¯=0 (16.13)
¯ K −Iω + Iz Ωω + K22 ¯
2

Equation 16.13 is called the characteristic equation and its roots are the wanted
natural frequencies ω of the system considered. They are presented in Fig. 5a) as
functions of the angular velocity of the rotor Ω .
ANALYSIS 331

Natural frequencies
100 positive precession
negative precession

ω
50 a)
ω=Ω

10
0 10 20 30 40 50
angular speed Ω

Amplitude of the forced vibration

5
A b)

0 0 10 20 30 40 50
angular speed Ω

Figure 5

16.3.2 Response to residual unbalance

The particular solution of the non-homogeneous equations 16.9 represents the forced
vibration of the rotor. It can be predicted as follows.

z = AeiΩt
ϕz = BeiΩt (16.14)

Here A and B are the complex amplitudes of vibration caused by the residual im-
balance. Introduction of Eq. 16.14 into the equations of motion 16.9 yields the
non-homogeneous set of the algebraic equations linear with respect to the constants
A and B.

(−mΩ2 + K11 )A + ( K )B = 0
( K )A + (−Iα + Iz Ω + K22 )B = DeiϕD
2 2
(16.15)
EXPERIMENTAL INVESTIGATION. 332

The absolute values of solution of the above equations

q
|A| = Re2 (A) + Im2 (A)
q
|B| = Re2 (B) + Im2 (B) (16.16)

represents the amplitudes of vibrations whereas the angles

Im(A)
ϕA = arctan( )
Re(A)
Im(B)
ϕB = arctan( ) (16.17)
Re(B)

represents phases of those vibrations with respect to the base used for measuring
phases of the imbalance. Fig. 5b) shows the amplitude A as a function of the angular
speed of the rotor. If the angular speed of the rotor is equal to the natural frequency
corresponding to the positive precession (see Fig. 5) the amplitude of the forced
vibrations is not limited. This angular speed is called the critical speed.

16.4 EXPERIMENTAL INVESTIGATION.

The main purpose of the experimental investigations is to verify the developed in
the previous section mathematical model. This verification may be accomplished
by comparison of results of analysis of the mathematical model with results of the
experimental investigation.
16.4.1 Identification of parameters of the laboratory installation
The mathematical model 16.9 has the following parameters which have to be identified
K11 , K, K22 - coefficients of the stiffness matrix.
m, I, Iz - the inertia parameters

Identification of the stiffness matrix coefficients K11, K22, K.

3 1 2

M z
x ϕx

F
x

Figure 6

The total displacement of the system shown in Fig. 6 along the coordinates x and
ϕx caused by the force F and moment M, according to the superposition principle,
may be expressed as follows.
EXPERIMENTAL INVESTIGATION. 333

x = δ 11 F + δ 12 M
ϕx = δ 21 F + δ 22 M (16.18)

The coefficients δ11 , δ 12 , δ 21 and δ 22 are called influence coefficients. If M = 0, the

Eqs. 16.18 yield formulae for the coefficients δ 11 and δ 21 .

x|M=0
δ11 =
F
ϕx |M=0
δ21 = (16.19)
F
Similarly the coefficients δ 12 and δ22 can be computed according to

x|F =0
δ 12 =
M
ϕx |F =0
δ 22 = (16.20)
M
The above formulae permit the influence coefficients to be determined by means of a
simple experiment.
To determine the influence coefficients δ 11 and δ 21 it is sufficient to apply the
known load F1 (see Fig. 1) and measure displacements due to this load along coordi-
nates x and ϕx . These displacements can be measured with help of the transducers 10
and 11 (see Fig. 1). If d1 and d2 are the displacements measured by the transducers
10 and 11 respectively, the displacements x and ϕx are:

x|M=0 = d1
d2
ϕx |M=0 = (16.21)
c
To determine the influence coefficients δ 12 and δ 22 it is sufficient to apply the
known load F2 (see Fig. 1) and measure displacements along coordinates x and ϕx
caused by this load. The moment M applied along the coordinate ϕx , according to
Fig. 1, is.
M = F2 (L + a) (16.22)
According to Maxwell’s reciprocity theorem

δ 12 = δ 21 = δ (16.23)

The relationship 16.18 can be rewritten in the following matrix form.

∙ ¸ ∙ ¸∙ ¸
x δ 11 δ F
= (16.24)
ϕx δ δ 22 M
EXPERIMENTAL INVESTIGATION. 334

Upon inverting the above equations one may obtain the system stiffness matrix.
∙ ¸ ∙ ¸∙ ¸
F K11 K x
= (16.25)
M K K22 ϕx

where ∙ ¸ ∙ ¸−1
K11 K δ 11 δ 12
= (16.26)
K K22 δ 21 δ 22

Identification of mass m and moments of inertia I, Iz

If the rotor does not rotate (Ω = 0), according to Eq. 16.13, the equation for the
natural frequencies is
¯ ¯
¯ −mω 2 + K11 K ¯
¯ ¯=0 (16.27)
¯ K −Iω + K22 ¯
2

or
−(K22 ω 2 )m − (K11 ω2 )I + (ω 4 )Im + K11 K22 − K 2 = 0 (16.28)
The above equation must be fulfilled by both of the natural frequencies ω1 and ω 2 .
They can be measured with help of the spectrum analyzer 14 (see Fig. 1). Introduc-
tion of these natural frequencies into Eq. 16.28 yields set of two non-linear algebraic
equations with two unknowns m and I.

−(K22 ω 21 )m − (K11 ω 21 )I + (ω 41 )Im + K11 K22 − K 2 = 0

−(K22 ω 22 )m − (K11 ω 22 )I + (ω 42 )Im + K11 K22 − K 2 = 0 (16.29)

Their solution offers the wanted parameters m and I.

Analytical analysis of dimensions of the disc leads to conclusion that its mo-
ment of inertia about the axis z is

Iz = 0.9(2I) (16.30)

16.4.2 Measurement of the natural frequencies of the rotor

The regulator 9 (see Fig. 1) allows the rotating speed Ω of the motor 8 to be adjusted.
The transducer 16 produces 10 impulses per each revolution of its rotor. Frequency
of these pulses fΩ Hz may be read from the frequency meter 17. This reading permits
the current angular speed to be determined.

Ω = 0.1fΩ Hz (16.31)

Undisturbed rotating rotor, due to the residual imbalance, performs the forced vi-
brations. Frequency of this vibration is equal to the rotating speed. In this case,
spectrum has only one pick corresponding to the current rotating speed. If one apply
an impact to the rotating rotor, its vibration will be superposition of both the forced
vibration and natural vibrations.. Therefore on screen of the spectrum analyzer 14
four additional picks should be observed. These four picks correspond to the wanted
natural frequencies of the rotating rotor.
EXPERIMENTAL INVESTIGATION. 335

16.4.3 Measurement of amplitude of the forced vibration

For the steady state motion the rotor performs forced vibrations due to the residual
imbalance. The oscilloscope 18 (see Fig. 1) permit the amplitude of the forced
vibration to be measured for any rotating speed Ω. This experimental data (amplitude
of the forced vibration as a function of the rotating speed) allows the resonance
diagram to be created. Local maxima of this diagram indicates critical speed of the
rotor.
16.4.4 Verification of the mathematical model
Verification of the mathematical model of the rotor should be obtained by comparison
of diagram of the measured natural frequencies as a function of the rotor’s speed with
these analytically predicted and comparison of the experimental resonance diagram
with the analytical one.
 Chapter 17
MODELLING OF ROTOR SYSTEMS

17.1 DESCRIPTION OF THE LABORATORY INSTALLATION

Fig. 1 presents a rotor system.

X E

L1 L2 L3 L4 L5

EM C S BB R E F
.

Figure 1

Its shaft S is driven by the electric motor EM. The flexible coupling C
prevents the vibration of the electric motor to be transmitted to the shaft. The shaft
is connected to the supporting structure by means of the self- aligning ball bearings
BB. The housings of the ball bearings R are connected to the foundation F by means
of system of spring E.
DEVELOPMENT OF THE MATHEMATICAL MODELS OF SUB-SYSTEMS 337

17.2 DEVELOPMENT OF THE MATHEMATICAL MODELS OF SUB-

SYSTEMS
To create the mathematical model of the rotor system, it was divided into number
of sub-systems. This division is shown in Fig. 2. The shaft S can be considered as
continuous system. The housings R may be treated as rigid bodies. The springs E
are assumed to be massless..

R1 R2 R3

E1 E2 E3

Figure 2
DEVELOPMENT OF THE MATHEMATICAL MODELS OF SUB-SYSTEMS 338

17.2.1 Mathematical model of the free-free shaft

L1 L2 L3 L4 L5

1 2 3 4 5 6 7 8 9 10 11

1 2 3 4 5 6 7 8 9 10 11 12

X
Z

51 52 53 54 55 56 57 58 59 60 61 62

Figure 3

The mathematical model of the shaft is to be created by means of the Finite Element
Method. Therefore the shaft has to be divided into several elements. The proposed
division of the shaft is shown in Fig. 3 To produce the matrix of inertia and the
matrix of stiffness the following parameters must be assessed for each element.
density ρ [kg/m3 ]
Young modulus E [N/m2 ]
shear modulus G [N/m2 ]
Poisson ratio ν
length of element l [m]
outer diameter D [m]
inner diameter d [m]
The mathematical models of the shaft SX and SY can be created separately in the
plane xz and yz. These two mathematical models SX and SY are associated with
the following vectors of coordinates

[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]

[51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62] (17.1)
respectively (see Fig. 3).
DEVELOPMENT OF THE MATHEMATICAL MODELS OF SUB-SYSTEMS 339

17.2.2 Mathematical models of the rigid elements

If the rigid elements R (see Fig. 2) may be considered as a particle (a = 0, Ip = 0),
their mathematical model, according to Eq. 1.16 (page 16) may be adopted in the
following form
mi ẍi = Ri +FGi (17.2)
where
∙ ¸ ∙ ¸ ∙ ¸ ∙ ¸
mi 0 xi Fxi 0
mi = ; xi = ; Ri = ; FGi = (17.3)
0 mi yi Fyi mi g

The mathematical models of the three rigid elements R1, R2 and R3, according to
Fig. 4, are identified with the following vectors of coordinates:

[4, 54]
[6, 56]
[9, 59] (17.4)

1 2 3 4 5 6 7 8 9 10 11 12

X
Z

51 52 53 54 55 56 57 58 59 60 61 62

4 6 9

54 56 59

Figure 4
DEVELOPMENT OF THE MATHEMATICAL MODELS OF SUB-SYSTEMS 340

17.2.3 Mathematical models of the elastic elements

X
α xj xi
yi
ξ

yj
η
Y

Figure 5

The elastic element shown in Fig. 5 is considered to be massless. Its dynamic

properties, according to considerations provided in section 1.4.4 (page 114), may be
determined by stiffness matrix k, damping matrix c. Let as assume that along the
principal axes ξ and η, determined by the angle α with respect to the global system
of coordinates XY, these matrices have the following form
∙ ¸ ∙ ¸
kξξ kξη cξξ cξη
k= , c= (17.5)
kηξ kηη cηξ cηη
The mathematical model of this free - free element along vector of coordinates
£ ¤
x = xi yi xj yj (17.6)

(see Fig. 5) has the following form

∙ T ¸ ∙ T ¸
C kC −CT kC C cC −CT cC
x+ ẋ = R (17.7)
−CT kC CT kC −CT cC CT cC

where ∙ ¸
cos α sin α
C= (17.8)
− sin α cos α
is the matrix of direction cosines between the principal axes and the global axes.
According to Fig. 6 the following vectors of coordinates are link to the math-
ematical models E1, E2 and E3 of the three elastic elements.

[4, 54, 24, 74]

[6, 56, 26, 76]
[9, 59, 29, 79] (17.9)
COMPOSITION OF THE SYSTEM MATHEMATICAL MODEL 341

4 6 9

54 56 95

24 4 26 6 29 9
54 56 59

74 76 79

Figure 6

17.2.4 Mathematical model of foundation

If the foundation F (see Fig. 2) may be considered rigid, motion of the system
considered along vector of coordinates:

[24, 74, 26, 76, 29, 79] (17.10)

is not possible. This assumption results in boundary conditions associated with math-
ematical models E1, E2 and E3.

17.3 COMPOSITION OF THE SYSTEM MATHEMATICAL MODEL

The sequence the mathematical models are join together is absolutely arbitrary. Fig.
7 presents one of the possible sequence of producing the final mathematical model of
the rotor system SYS and SYSC.

17.4 ANALYSIS OF THE MATHEMATICAL MODEL.

The discussed in part II (page 129) methods of analysis of the mathematical model
allow to develop programs which permit the following parameters to be computed:
1. natural frequencies of the system.
2. natural modes
3. equilibrium position
4. roots of the characteristic equation which permit the stability of the equilibrium
position to be assessed.
5. amplitudes of the forced vibration.
Computation of these parameters for the rotor bearing system before conden-
sation (mat. mod. SYS) and after condensation (mat. mod. SYSC) forms set of
data which allows to analyze the influence of condensation on accuracy of the math-
ematical models. Results of this analysis if confronted with experimentally measured
parameters of motion of the rotor system, allows to verify the developed methods of
composition and analysis of the rotor systems.
ANALYSIS OF THE MATHEMATICAL MODEL. 342

SUB-SYSTEM MODELLING
SX SY R1 R2 R3 E1 E2 E3

ADDITION ADDITION ADDITION

OF TWO OF TWO OF TWO
SUB-SYST'S SUB-SYST'S SUB-SYST'S

S R12 E12
ADDITION ADDITION
OF TWO OF TWO
INTRO. OF SUB-SYST'S SUB-SYST'S
PERMANENT
DEFLECTION

R E
ADDITION
SD OF TWO
SUB-SYST'S

COMPOSITION RE

AND INTRO. OF
BOUNDARY
CONDITIONS

CONDENSATION SUP
ADDITION
OF TWO
SUB-SYST'S

SYS

CONDENSATION

SYSC

MODAL STATIC STABILITY FORCED

ANALYSIS ANALYSIS ANALYSIS RESPONSE

NATURAL EQUILIBRIUM STABILITY CRITICAL

FREQUENCIES POSITION OF SPEEDS
EQUILIBRIUM
NATURAL AMPLITUDES
POSITION
MODES OF FORCED
VIBRATIONS

ANALYSIS

Figure 7
EXPERIMENTAL INVESTIGATION 343

17.5 EXPERIMENTAL INVESTIGATION

1
Y

Ω Ω
[Hz] [Hz] [V]

4 5 6 7

Figure 8

Fig. 8 presents equipment which allows to measure the following parameters:

1. the angular velocity of the rotor
2. natural frequencies
3. amplitude of the vibration of the system along coordinates 6 and 56.
17.5.1 Measurement of the angular speed of the rotor
The transducer 3 (see Fig. 8) produces 10 pulses per each revolution of the rotor.
Frequency of theses pulses f is measured by the frequency counter 5. Therefore the
instantaneous angular velocity of the rotor is
f
Ω = 2π [rad/s] (17.11)
10
where: f is the frequency of pulses in [Hz]
17.5.2 Measurement of the natural frequencies
The free motion of any mechanical system, e.g.. due to an impulse of force, is a
superposition of its natural frequencies. This motion (displacement) of the rotor
along the coordinates 6 and 56 is proportional to the voltage produced by the eddy-
current transducers 1 and 2. The spectral analysis of this voltage, performed by the
spectrum analyzer 6, yields directly the wanted natural frequencies in Hz.
VERIFICATION OF THE DEVELOPED MATHEMATICAL MODEL 344

17.5.3 Measurement of amplitude of the forced vibration

The oscilloscope 7 allows to record the trajectory of the rotor in plane 6, 56. For
the steady state motion the free motion is damped out and the rotor performs the
harmonic motion with frequency equal to the angular velocity of the rotor. Therefore
the trajectory of motion is, supposed to be an ellipse (see Fig. 9). Ax and Ay are the
wanted amplitudes of the forced vibrations in the x and y direction respectively.

2Ay
o
x

2Ax

Figure 9

17.6 VERIFICATION OF THE DEVELOPED MATHEMATICAL MODEL

The verification of the final mathematical models SYS and SYSC (see Fig.7) can be
obtained by comparison of the measured natural frequencies, measured critical speeds
and the measured amplitudes of the forced vibrations with the predicted magnitudes
of these parameters.
 Chapter 18
BALANCING OF ROTOR

18.1 DESCRIPTION OF THE LABORATORY INSTALLATION

3 2 1 4 5 3 6

Figure 1

18.2 THEORETICAL BACKGROUND

Let us consider a rigid rotor that rotates with an angular velocity ω about the axis
A − A (see Fig. 2).
THEORETICAL BACKGROUND 346

1 2
U1
ω U2

B
ϕ1 B
A ϕ2
m1 A
m2
F2
ωt ωt
Ui2
Ui1
F1 Ui
x1 x2

Figure 2

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
Ui (see Fig. 2). Each of this forces can be replaced by two forces Ui1 and Ui2 acting
in two arbitrarily chosen planes. Each of them is perpendicular to the axis of rotation,
therefore their resultants U1 and U2 are perpendicular to the axis of rotation too.
Hence, one can eliminate this unbalance of the rotor by means of two weights of
mass m1 and m2 attached at such a position that the centrifugal forces F1 and F2
balance the resultant forces U1 and U2 . The process of searching for magnitude of
the unbalance forces U1 and U2 and their phases ϕ1 and ϕ2 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.
THEORETICAL BACKGROUND 347

3 2 1 4 5 3

ωt

3 1 2 4

U1o U2o
ϕ1 ϕ2
U1 ωt
x1 ω t U2
x2
a4
β3 ωt a4
β3 t
a3 β4
a3
ωt
x3 ω t x3

Figure 3

Let us consider the rotating machine shown in Fig. 3. 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 U1 and U2 in two arbitrarily chosen
plane. These two arbitrarily chosen planes,denoted in Fig. 3 by numbers 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 offer 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. 3 by numbers 3 and 4, are called measurement planes. Let a3 and a4 be the
THEORETICAL BACKGROUND 348

complex displacements measured in the measurement planes along the coordinates

x3 and x4 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. 3 by N, allows the phases of the unbalance forces (ϕ1 , ϕ2 ) and the phases of
the recorded displacements (β 1 , β 2 ) to be measured. The relationship between the
unknown unbalance forces U1 and U2 and the measured displacements a3 and a4 is.
∙ ¸ ∙ ¸∙ ¸
a3 R31 (iω) R32 (iω) U1
= (18.1)
a4 R41 (iω) R42 (iω) U2

where
a3 = a3o eiβ 3 , a4 = a4o eiβ 4 , U1 = U1o eiϕ1 , U2 = U2o eiϕ2 (18.2)
If the transfer functions Ri,j (iω) would be known, this relation would allow the un-
known 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)

3 1 2 4
µ (1) U (1)
m (1)
(1)
Uo
ϕ (1)
U1o U2o
ϕ1
ωt ϕ2
ωt U1 ωt
x1 U2
x2
a (1)
4
(1) a (1)
4o
β3
(1)
a 3o β 4(1)
(1)
a3

x3 ωt ωt

Figure 4

An additional trial weight of mass m(1) (see Fig 4) is attached in the balancing
plane 1 at the known (with respect to the key phasor’s mark) phase ϕ(1) and the know
distance µ(1) . The system is now excited by both the residual unbalance forces (U1
and U2 ) and the centrifugal force produced by the trial weight U (1) . The amplitude
of this force U (1) is
Uo(1) = m(1) µ(1) ω2 (18.3)
The response of the system is recorded in both measurement planes so the amplitudes
(1) (1) (1) (1)
a3o and a4o as well as the phases β 3 and β 4 can be obtained. There is the following
THEORETICAL BACKGROUND 349

relationship between the measured parameters and the transfer functions.

" # ∙ ¸∙ ¸
(1)
a3 R31 (iω) R32 (iω) U1 + U (1)
(1) = (18.4)
a4 R41 (iω) R42 (iω) U2

(1) (1) (1)

(1) (1) (1) (1)
a3 = a3o eiβ 3 , a4 = a4o eiβ 4 , U1 = U1o eiϕ1 , U2 = U2o eiϕ2 , U (1) = Uo(1) eiϕ
(18.5)
Test (2)

3 1 2 4

µ (2) U
(2)
m (2)
(2)
Uo

ϕ (2)
U1o U2o
ϕ1
ωt ϕ2
ωt U1 ωt
x1 U2
x2
a (2)
4
(2) a (2)
4o
β3
(2)
a 3o β 4(2)
(2)
a3

x3 ωt ωt

Figure 5

An additional trial weight of mass m(2) (see Fig 5) is attached in the balancing
plane 2 at the known (with respect to the key phasor’s mark) phase ϕ(2) and the know
distance µ(2) . The system is now excited by both the residual unbalance forces (U1
and U2 ) and the centrifugal force produced by the trial weight U (2) . The amplitude
of this force U (2) is
Uo(2) = m(2) µ(2) ω2 (18.6)
The response of the system is recorded in both measurement planes so the amplitudes
(2) (2) (2) (2)
a3o and a4o as well as the phases β 3 and β 4 can be obtained. There is the following
relationship between the measured parameters and the transfer functions.
" # ∙ ¸∙ ¸
(2)
a3 R31 (iω) R32 (iω) U1
= (18.7)
(2)
a4 R41 (iω) R42 (iω) U2 + U (2)

(2) (2) (2)

(2) (2) (2) (2)
a3 = a3o eiβ 3 , a4 = a4o eiβ 4 , U1 = U1o eiϕ1 , U2 = U2o eiϕ2 , U (2) = Uo(2) eiϕ
(18.8)
THEORETICAL BACKGROUND 350

The formulated equations 18.1, 18.4 and 18.7 allow the unknown transfer functions
and the wanted unbalances U1 and U2 to be computed. To achieve that let us subtract
the equations 18.1 from 18.4
Ã" # ∙ ¸! ∙ ¸ µ∙ ¸ ∙ ¸¶
(1)
a3 a3 R31 (iω) R32 (iω) U1 + U (1) U1
(1) − = −
a4 a4 R41 (iω) R42 (iω) U2 U2
" # ∙ ¸ ∙ (1) ¸
(1)
a3 − a3 R31 (iω) R32 (iω) U
(1) =
a4 − a4 R41 (iω) R 42 (iω) 0
" # ∙ ¸
(1)
a3 − a3 R31 (iω)U (1)
(1) =
a4 − a4 R41 (iω)U (1)

(1) (1) (1)

a − a3 a eiβ 3 − a3o eiβ 3
R31 (iω) = 3 (1) = 3o (1) (1)
U Uo eiϕ
(1) (1) (1)
a − a4 a eiβ 4 − a4o eiβ 4
R41 (iω) = 4 (1) = 4o (1) (1) (18.9)
U Uo eiϕ
Similarly, if one subtracts equations 18.1 from 18.7 one can get
(2) (2) (2)
a − a3 a eiβ 3 − a3o eiβ 3
R32 (iω) = 3 (2) = 3o (2) (2)
U Uo eiϕ
(2) (2) (2)
a − a4 a eiβ 4 − a4o eiβ 4
R42 (iω) = 4 (2) = 4o (2) (2) (18.10)
U Uo eiϕ
Now, the wanted complex imbalances U1 and U2 in the plane 1 and 2 may be computed
from the equation 18.1
∙ ¸ ∙ ¸−1 ∙ ¸
U1 R31 (iω) R32 (iω) a3
= (18.11)
U2 R41 (iω) R42 (iω) a4

where a3 and a4 represent the know response of the system without the additional
weights.
∙ ¸ ∙ ¸−1 ∙ ¸ ∙ ¸
U1 R31 (iω) R32 (iω) a3o eiβ 3 U1o eiϕ1
= = (18.12)
U2 R41 (iω) R42 (iω) a4o eiβ 4 U2o eiϕ2

The amplitudes U1o and U2o determine the weights m1 and m2 that should be attached
in the balancing planes
U1
m1 =
r1 ω 2
U2
m2 = (18.13)
r2 ω 2
EXPERIMENTAL INVESTIGATION. 351

These weights, to balance the rotor, should be place at angular position (see Fig. 6)

β 1 = 180o + ϕ1
β 2 = 180o + ϕ2 (18.14)

3 2 1 4 3

ωt

3 1 2 4

m1 m2
β1 β2
r1 r2
U1o U2o
ϕ1
ϕ2
x1 ωt U1 ωt U2
x2

Figure 6

18.3 EXPERIMENTAL INVESTIGATION.

The purpose of the experimental investigations is to carry out the dynamic balancing
of the grinder shown in Fig.1 and assess the residual unbalance.
EXPERIMENTAL INVESTIGATION. 352

18.3.1 Description of the balancing procedure.

Plane of correction 1 Plane of correction 2

Amplitude Phase Amplitude Phase

Initial run
Test run1
Test weight 1

Test run 2
Test weight 2

Correction 1
Check run 1
Correction 2

Check run 2
Correction 3
Check run 3

Correction 4
Check run 4

Correction 5
Check run 5
Correction1

Check run1

Figure 7
 Part IV
PROBLEMS

353
 Chapter 19
PREVIOUS YEARS EXAMINATION QUESTIONS

19.1 SAMPLES OF QUESTIONS WITH SOLUTIONS

Problem 15

The dynamic properties of two identical free-free beams, shown in Fig. 1 are
determined by the following inertia and stiffness matrices.

⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
m11 m12 m13 k11 k12 k13 xa1 xb1
m = ⎣ m21 m22 m23 ⎦ ; k = ⎣ k21 k22 k23 ⎦ ; xa = ⎣ xa2 ⎦ ; xb = ⎣ xb2 ⎦
m31 m32 m33 k31 k32 k33 xa3 xb3
(19.1)

b
x1a x2a x3a x1b x2b x3

Figure 1

These two beams were joined together by means of a spring of stiffness k and
a damper of the damping coefficient c to form system shown in Fig.2.

1 2 3

k c

Figure 2

Produce equations of motion of the system and derive equation for natural
frequencies of the system.
SAMPLES OF QUESTIONS WITH SOLUTIONS 355

Solution.
Equations of the beam a are

mẍa + kxa = Ra (19.2)

To introduce the boundary conditions the above equations should be partitioned as

follows
∙ a ¸
a x1 £ ¤
x1 = ; xa2 = xa3
xa2
∙ ¸ ∙ ¸
m11 m12 m13 £ ¤ £ ¤
m11 = ; m12 = ; m21 = m31 m32 ; m22 = m33
m21 m22 m23
∙ ¸ ∙ ¸
k11 k12 k13 £ ¤ £ ¤
k11 = ; k12 = ; k21 = k31 k32 ; k22 = k33
k21 k22 k23
∙ a ¸
a R1 £ ¤
R1 = ; Ra2 = R3a (19.3)
R2a

m11 ẍa1 + m12 ẍa2 + k11 xa1 + k12 xa2 = Ra1

m21 ẍa1 + m22 ẍa2 + k21 xa1 + k22 xa2 = Ra2 (19.4)

Boundary conditions in the case considered are

∙ ¸
a 0
x1 = (19.5)
0

Introducing them into equations 19.4 one may obtain

m12 ẍa2 + k12 xa2 = Ra1 (19.6)

m22 ẍa2 + k22 xa2 = Ra2 (19.7)

The equation 19.6 represents the relationship between motion of the beam a along the
coordinates xa2 and the interaction forces Ra1 whereas the equation 19.7 offers equation
of motion of the beam a supported along coordinates xa1 . It could be adopted for
further consideration in the following form

m33 ẍa3 + k33 xa3 = R3a (19.8)

In the same manner one can obtain equation of motion of the supported beam b

m33 ẍb3 + k33 xb3 = R3b (19.9)

Combining these two equations together, we are getting equations of motion of the
system shown in Fig. 3
SAMPLES OF QUESTIONS WITH SOLUTIONS 356

1 2 3

xa3

x3b

Figure 3

∙ ¸∙ a ¸ ∙ ¸∙ a ¸ ∙ a ¸
m33 0 ẍ3 k33 0 x3 R3
b + b = (19.10)
0 m33 ẍ3 0 k33 x3 R3b
Motion of the element shown in Fig. 4 is governed by equation 19.11

x3a
k c

x3b

Figure 4

∙ ¸∙ a ¸ ∙ ¸∙ a ¸ ∙ a ¸
c −c ẋ3 k −k x3 R3
+ = (19.11)
−c c ẋb3 −k k xb3 R3b
Application of the composition rules to the equations 19.10 and 19.11 yields equations
of motion of the system shown in Fig. 2
∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸∙ ¸ ∙ ¸
m33 0 ẍa3 c −c ẋa3 k33 + k −k xa3 0
+ + =
0 m33 ẍb3 −c c ẋb3 −k k33 + k xb3 0
(19.12)
They may be rewritten in the matrix form

MẌ + CẊ + KX = 0 (19.13)

where
∙ ¸ ∙ ¸ ∙ ¸ ∙ ¸
m33 0 c −c k33 + k −k xa3
M= ; C= ; K= ; X=
0 m33 −c c −k k33 + k xb3
(19.14)
SAMPLES OF QUESTIONS WITH SOLUTIONS 357

Substitution
Ẋ = Y (19.15)
allows the equations of motion to be formulated in terms of the state space coordinates
∙ ¸ ∙ ¸∙ ¸
Ẋ 0 1 X
= (19.16)
Ẏ −M−1 K −M−1 C Y
If ∙ ¸ ∙ ¸
0 1 X
A= and Z = (19.17)
−M K −M−1 C
−1
Y
the equations of motion take form

Ż = AZ (19.18)
Hence, the equation for the natural frequencies is

det(A − 1ω) = 0 (19.19)

SAMPLES OF QUESTIONS WITH SOLUTIONS 358

Problem 16

a)
y1f y3f

b)
y1b y2b y3b

c)

y1 y2 y3

Figure 5

The mathematical model of the foundation shown in Fig. 5a) was identified
along the following vector of coordinates [y1f , y3f ]. Its inertia matrix, damping matrix
and the stiffness matrix associated with this vector of coordinates are:
∙ f ¸ ∙ f ¸ ∙ f f ¸
m11 mf13 c11 cf13 k11 k13
m= ; c= ; k= (19.20)
mf31 mf33 cf31 cf33 f
k31 f
k33

Motion of this foundation caused by the environmental excitation was measured along
the coordinates [y1f , y3f ] and approximated by the following function of time:

ef1 = E1f sin ωt; ef3 = E3f cos ωt (19.21)

The mathematical model of the free-free beam shown in Fig. 5b) may be adopted in
the following form:
⎡ s ⎤⎡ s ⎤ ⎡ ⎤⎡ s ⎤ ⎡ ⎤
m11 ms12 ms13 ÿ1 k11 k12 k13 y1 0
⎣ ms21 ms22 ms23 ⎦ ⎣ ÿ2s ⎦ + ⎣ k21 k22 k23 ⎦ ⎣ y2s ⎦ = ⎣ 0 ⎦ (19.22)
s s s s s
m31 m32 m33 ÿ3 k31 k32 k33 y3 0

Produce the mathematical model of the system shown in Fig. 5c) and derive an ex-
pression for amplitude of the forced vibration of the system due to the environmental
excitation.
SAMPLES OF QUESTIONS WITH SOLUTIONS 359

Solution:
The equivalent forces due to the environmental excitation are
∙ f ¸ ∙ t ¸∙ f ¸ ∙ t ¸∙ f ¸ ∙ t ¸∙ f ¸
F1 m11 mt13 ë1 c11 ct13 ė1 t
k11 k13 e1
f = f + f +
F3 t
m31 m33 t
ë3 t
c31 c33t
e3 t t
k31 k33 ef3
∙ ¸
−mt11 E1f ω 2 ct11 − ct13 E3f ω + k11
t
E1f
= sin ωt
−mt31 E1f ω 2 − ct33 E3f ω + k31t
E1f
∙ ¸
−mt13 E3f ω2 + ct11 E1f ω + k13
t
E3f
+ cos ωt
−mt33 E3f ω2 + ct31 E1f ω + k33
t
E3f
∙ ¸ ∙ ¸
S1 C1
= sin ω + cos ωt (19.23)
S3 C3

Implementation of these forces in the equations of motion of the foundation yields

∙ f ¸∙ f ¸ ∙ f ¸∙ f ¸ ∙ f f ¸∙ f ¸
m11 mf13 ÿ1 c11 cf13 ẏ1 k11 k13 y1
f f f + f f f + f f
m31 m33 ÿ3 c31 c33 ẏ3 k31 k33 y3f
∙ ¸ ∙ ¸ ∙ ¸
S1 C1 R1f
= sin ω + cos ωt + (19.24)
S3 C3 R3f
Reorganization of the mathematical model of the beam results in the following equa-
tion
⎡ s ⎤⎡ s ⎤ ⎡ ⎤⎡ s ⎤ ⎡ s ⎤
m11 ms13 ms12 ÿ1 k11 k13 k12 y1 R1
⎣ ms31 ms33 ms32 ⎦ ⎣ ÿ3s ⎦ + ⎣ k31 k33 k32 ⎦ ⎣ y2s ⎦ = ⎣ R3s ⎦ (19.25)
s s s s s
m21 m23 m22 ÿ2 k21 k23 k22 y3 0

Application of the composition rules to the equations of motion of the foundation

19.23 and the beam 19.25 yields the mathematical model of the whole system
⎡ s ⎤⎡ ⎤ ⎡ f ⎤⎡ ⎤
m11 + mf11 ms13 + mf13 ms12 ÿ1 c11 cf13 0 ẏ1
⎣ ms + mf ms + mf ms ⎦ ⎣ ÿ3 ⎦ + ⎣ cf cf 0 ⎦ ⎣ ẏ2 ⎦
31 31 33 33 32 31 33
ms21 ms23 ms22 ÿ2 0 0 0 ẏ3
⎡ f f ⎤ ⎡ ⎤
k11 + k11 k13 + k13 k12 y1
+ ⎣ k31 + k31f f
k33 + k33 k32 ⎦ ⎣ y2 ⎦
k21 k23 k22 y3
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
S1 C1 A1 cos(ωt + ϕ1 )
= ⎣ S3 ⎦ sin ωt + ⎣ C3 ⎦ cos ωt = ⎣ A3 cos(ωt + ϕ2 ) ⎦ (19.26)
0 0 0

where µ ¶
p Si
Ai = (Si )2 + (Ci )2 ; ϕi = − arctan (19.27)
Ci
SAMPLES OF QUESTIONS WITH SOLUTIONS 360

The above equations in terms of the complex excitation may be adopted in the fol-
lowing form
⎡ ⎤ ⎡ ⎤
A1 cos(ωt + ϕ1 ) A1 sin(ωt + ϕ1 )
mÿ + cẏ + ky = ⎣ A3 cos(ωt + ϕ2 ) ⎦ +i ⎣ A3 sin(ωt + ϕ2 ) ⎦
0 0
⎡ ⎤ ⎡ ⎤
A1 ei(ωt+ϕ1 ) A1 eiϕ1 eiωt
= ⎣ A3 ei(ωt+ϕ3 ) ⎦ = ⎣ A3 eiϕ3 eiωt ⎦ (19.28)
0 0
⎡ ⎤ ⎡ ⎤
B1 eiωt B1
= ⎣ B3 eiωt ⎦ = ⎣ B3 ⎦ eiωt
0 0
= Beiωt

The forced vibration of the system is approximated by the real part of the particular
solution of the above equation. It can be predicted as

yc = Yc eiωt (19.29)

Introduction of the predicted solution into equation 19.28 yields the complex ampli-
tudes

Yc = (−ω2 m + ωc + k)−1 B (19.30)

Hence, the complex solution is

yc = (Re(Yc ) + i Im(Yc ))(cos ωt + i sin ωt) (19.31)

Its real part represents the wanted forced vibration of the system

y = Re(yc ) = (Re(Yc ) cos ωt − Im(Yc ) sin ωt (19.32)

SAMPLES OF QUESTIONS WITH SOLUTIONS 361

Problem 17

a)

z
b)
ys ys ys ys
1 2 3 4

xd
G z 3
c)
a yd
4
yd
yd 3
3

Figure 6

In order to analyze motion of the rotor shown in Fig. 6a) it was split into
sub-systems shown in Fig. 6b) and Fig. 6c) respectively. The mathematical model
of the free-free rotor shown in Fig. 6b) along the system of coordinates [y1s , y2s , y2s , y2s ]
is
⎡ s ⎤⎡ s ⎤ ⎡ s ⎤⎡ s ⎤ ⎡ ⎤
m11 ms12 ms13 ms14 ÿ1 k11 k12s s
k13 s
k14 y1 G1
⎢ ms21 ms22 ms23 ms24 ⎥ ⎢ ÿ2s ⎥ ⎢ k21 s s
k22 s
k23 s ⎥⎢ s ⎥
k24 ⎢ ⎥
⎢ s ⎥ ⎢ ⎥ + ⎢ ⎥ ⎢ y2 ⎥ ⎢ G2 ⎥
s ⎦⎣ s ⎦ = ⎣
⎣ m31 ms32 ms33 ms34 ⎦ ⎣ ÿ3s ⎦ ⎣ k31 s s s
k32 k33 k34 y3 G3 ⎦
ms41 ms42 ms43 ms44 ÿ4s s
k41 s
k42 s
k43 s
k44 y4s G4
(19.33)
The mathematical model of the disk shown in Fig. 6c) along system of coordinates
[xd3 , xd4 , y3d , y4d ] is
⎡ ⎤⎡ d ⎤ ⎡ ⎤⎡ d ⎤ ⎡ ⎤
m ma 0 0 ẍ3 0 0 0 0 ẋ3 0
⎢ ma I 0 0 ⎥ ⎢ d ⎥ ⎢ 0 Ix Ω ⎥ ⎢ d ⎥ ⎢ ⎥
⎢ ⎥ ⎢ xd4 ⎥ + ⎢ 0 0 ⎥ ⎢ ẋd3 ⎥ = ⎢ 0 ⎥
⎣ 0 0 m ma ⎦ ⎣ ÿ3 ⎦ ⎣ 0 0 0 0 ⎦ ⎣ ẏ3 ⎦ ⎣ mg ⎦
d
0 0 ma I ÿ4 0 −Iz Ω 0 0 ẏ4d mga
(19.34)
Derive an equation for the natural frequencies of the rotor shown in Fig. 6a).
SAMPLES OF QUESTIONS WITH SOLUTIONS 362

Solution
To introduce the boundary conditions, let us partition the equation of motion
of the shaft.
∙ s ¸∙ s ¸ ∙ s ¸∙ s ¸ ∙ s ¸∙ s ¸ ∙ s ¸∙ s ¸
m11 ms12 ÿ1 m13 ms14 ÿ3 k11 k12s
y1 k13 k14s
y3
s s s + s s s + s s s + s s
m21 m22 ÿ2 m23 m24 ÿ4 k21 k22 y2 k23 k24 y4s
∙ s s
¸∙ s ¸ ∙ s ¸ ∙ s ¸
k13 k14 y3 G1 R1y
+ s s s = s + s (19.35)
k23 k24 y4 G2 R2y
∙ s ¸∙ s ¸ ∙ s ¸∙ s ¸ ∙ s ¸∙ s ¸
m31 ms32 ÿ1 m33 ms34 ÿ3 s
k31 k32 y1
+ +
ms41 ms42 ÿ2s ms43 ms44 ÿ4s s
k41 s
k42 y2s
∙ s s
¸∙ s ¸ ∙ s ¸ ∙ s ¸
k33 k34 y3 G3 R3y
+ s s s = s + s (19.36)
k43 k44 y4 G4 R4y
Introduction of boundary conditions
∙ ¸ ∙ ¸
y1s 0
= (19.37)
y2s 0
into the equations 19.35 results in relationship between the interaction forces and
motion of the shaft. The equations of motion of the supported shaft can be obtained
by substitution of the boundary conditions 19.37 into equations 19.36.
∙ s ¸∙ s ¸ ∙ s ¸∙ s ¸ ∙ s ¸ ∙ s ¸
m13 ms14 ÿ3 k13 k14s
y3 G1 R1y
s s s + s s s = s + (19.38)
m23 m24 ÿ4 k23 k24 y4 G2 R2y
∙ s ¸∙ s ¸ ∙ s ¸∙ s ¸ ∙ s ∙ s ¸¸
m33 ms34 ÿ3 k33 k34s
y3 G3 R3y
+ = + (19.39)
ms43 ms44 ÿ4s s
k43 s
k44 y4s Gs4 s
R4y
In the same manner one can produce equations of motion of the shaft in plane xz.
∙ s ¸∙ s ¸ ∙ s ¸∙ s ¸ ∙ ¸ ∙ s ¸
m33 ms34 ẍ3 s
k33 k34 x3 0 R3x
s s s + s s s = + s (19.40)
m43 m44 ẍ4 k43 k44 x4 0 R4x
Combining the equations 19.39 and 19.40 we are getting equations of motion of the
shaft.

⎡ ⎤⎡ s ⎤ ⎡ ⎤⎡ s ⎤ ⎡ ⎤ ⎡ s ⎤
ms33 ms34 0 0 ẍ3 s
k33 s
k34 0 0 x3 0 R3x
⎢ ms43 ms44 0 0 ⎥ ⎢ ẍs4 ⎥ ⎢ k43
s
k s
0 0 ⎥ ⎢ x s ⎥ ⎢ 0 ⎥ ⎢ R4x s ⎥
⎢ ⎥⎢ ⎥+⎢ 44 ⎥ ⎢ 4 ⎥=⎢ ⎥+⎢ s ⎥
⎣ 0 0 ms33 ms34 ⎦ ⎣ ÿ3s ⎦ ⎣ 0 s
0 k33 s ⎦⎣ s
k34 y3 ⎦ ⎣ G3 ⎦ ⎣ R3y ⎦
0 0 ms43 ms44 ÿ4s 0 s
0 k43 s
k44 y4s G4 R4ys

(19.41)
Now, combining equations of motion of the shaft 19.41 with the equations of motion
of the disk
⎡ ⎤⎡ d ⎤ ⎡ ⎤⎡ d ⎤ ⎡ ⎤ ⎡ d ⎤
m ma 0 0 ẍ3 0 0 0 0 ẋ3 0 R3x
⎢ ma I 0 ⎥ ⎢ d ⎥ ⎢
0 ⎥ ⎢ x4 ⎥ ⎢ 0 0 ⎥ ⎢ d ⎥ ⎢
0 Ix Ω ⎥ ⎢ ẋ3 ⎥ ⎢ 0 ⎥ ⎢ R4x ⎥ ⎢ d ⎥
⎢ + = + ⎥
⎣ 0 0 m ma ⎦ ⎣ ÿ3d ⎦ ⎣ 0 0 0 0 ⎦ ⎣ ẏ3d ⎦ ⎣ mg ⎦ ⎣ R3y d ⎦

0 0 ma I ÿ4d 0 −Iz Ω 0 0 ẏ4d mga d

R4y
(19.42)
SAMPLES OF QUESTIONS WITH SOLUTIONS 363

we are getting equations of motion of the whole rotor in the following form.
⎡ ⎤⎡ ⎤
ms33 + m ms34 + ma 0 0 ẍ3
⎢ ms43 + ma ms44 + I 0 0 ⎥ ⎢ ẍ4 ⎥
⎢ ⎥⎢ ⎥
⎣ 0 0 ms33 + m ms34 + ma ⎦ ⎣ ÿ3 ⎦
0 0 ms43 + ma ms44 + I ÿ4
⎡ ⎤⎡ ⎤ ⎡ s s
⎤⎡ ⎤ ⎡ ⎤
0 0 0 0 ẋ3 k33 k34 0 0 x3 0
⎢ 0 0 0 Ix Ω ⎥ ⎢ ⎥ ⎢ s s
0 ⎥ ⎢ ⎥ ⎢ ⎥
+⎢ ⎥ ⎢ ẋ3 ⎥ + ⎢ k43 k44 0s ⎥ ⎢ x4 ⎥ = ⎢ 0 ⎥
⎣ 0 0 0 0 ⎦ ⎣ ẏ3 ⎦ ⎣ 0 0 k33 k34 ⎦ ⎣ y3 ⎦ ⎣ G3 + mg ⎦
s
s s
0 −Iz Ω 0 0 ẏ4 0 0 k43 k44 y4 G4 + mga
(19.43)
The above equations cam be written in shorter form

mz̈ + cż + kz = G (19.44)

The homogeneous equation

mz̈ + cż + kz = 0 (19.45)
yields the equation for the natural frequencies in the following form

det(A − 1ω) = 0 (19.46)

where ∙ ¸
0 1
A= (19.47)
−m−1 k −m−1 c
SAMPLES OF QUESTIONS WITH SOLUTIONS 364

Problem 18

a)
A B C
a
x1
a
x2
a
x3 z
z
A B C
b)
b
x1 xb xb b
2 3

c) A B
C z
xC3

x1 x2 c

x3

Figure 7

The mathematical model of the free-free beam a in the horizontal plane xz

(see Fig. 7 a) is as follows
Ma ẍa + Ka xa = 0 (19.48)
where
⎡
a a a
⎤ ⎡ a a a
⎤ ⎡ a ⎤
M11 M12 M13 K11 K12 K13 x1
a ⎣ a a a ⎣
M = M21 M22 M23 ; K = K21 K22 K23 ; x = xa2 ⎦
a ⎦ a a a ⎦ a ⎣ (19.49)
a a a a a a
M31 M32 M33 K31 K32 K33 xa3
This beam is permanently deformed and the vector of its permanent deformation is
£ ¤T
aa = 0 aa 0 (19.50)
The mathematical model of the free-free beam b in the horizontal plane xz (see Fig.
7 b) is as follows
Mb ẍb + Kb xb = 0 (19.51)
where
⎡ b b b
⎤ ⎡ b b b
⎤ ⎡ b ⎤
M11 M12 M13 K11 K12 K13 x1
Mb = ⎣ M21
b b
M22 b ⎦
M23 ; Kb = ⎣ K21
b b
K22 b ⎦
K23 ; xb = ⎣ xb2 ⎦ (19.52)
b b b b b b
M31 M32 M33 K31 K32 K33 xb3
This beam is also permanently deformed and the vector of its permanent deformation
is £ ¤T
ab = 0 ab 0 (19.53)
SAMPLES OF QUESTIONS WITH SOLUTIONS 365

These two beams were joined together as is shown in Fig. 7 c and supported by the
two supports A and B and the damper of the damping coefficient c.
The support B is stationary whereas the support A performs the harmonic
oscillations governed by the following equation

xA = A sin Ωt (19.54)

Produce:
1. the mathematical model of the system shown in Fig. 7 c)
2. the expression for the equilibrium position xC3 of the point C after assembly.
3. the expression for the amplitude of the forced vibration of the point C
caused by the motion of the support A.
4. the expression for the interaction force at the support A
SAMPLES OF QUESTIONS WITH SOLUTIONS 366

Solution
1. the mathematical model of the system
Vector of the equivalent forces acting on the beam a is
⎡ a a ⎤ ⎡ a ⎤
K21 a F1
Fa = Ka aa = ⎣ K22 a a ⎦
a = ⎣ F2a ⎦ (19.55)
a a
K23 a F3a

Vector of the equivalent forces acting on the beam b is

⎡ b b ⎤ ⎡ b ⎤
K21 a F1
b b b ⎣
F = K a = K22 a b b ⎦
= F2b ⎦
⎣ (19.56)
b b
K23 a F3b

The mathematical model of the permanently deformed and free-free beam a is

Ma ẍa + Ka xa = Fa + Rab + Rao (19.57)

The mathematical model of the permanently deformed and free-free beam b is

Mb ẍb + Kb xb = Fb + Rba + Rbo (19.58)

The vector Rab represents these interaction forces that act on the beam a due to its
interaction with the beam b. The vector Rba represents these interaction forces that
act on the beam b due to its interaction with the beam a. Obviously
⎡ ab ⎤ ⎡ ba ⎤
R1 R1
Rab = ⎣ R2ab ⎦ = −Rba = − ⎣ R2ba ⎦ (19.59)
ab ba
R3 R3

The vectors Rao and Rbo represent the interaction forces between the base o (inertial
space) and the beam a and b respectively.
⎡ ao ⎤ ⎡ bo ⎤
R1 R1
Rao = ⎣ R2ao ⎦ ; Rbo = ⎣ R2bo ⎦
0 0

These two beams are joint together along coordinates x1 x2 and x3 . Therefore
the mathematical model of the assembly is
¡ a ¢ ¡ ¢
M + Mb ẍ + Ka + Kb x = Fa + Fb + Rbo (19.60)

where ⎡ ⎤
x1
x = xa = xb = ⎣ x2 ⎦ (19.61)
x3
According to the description of the problem the boundary conditions are

x1 = A sin Ωt x2 = 0 (19.62)
SAMPLES OF QUESTIONS WITH SOLUTIONS 367

Introduction of these boundary conditions into the mathematical model of the free-
free assembly 19.60 one can get the mathematical model of the supported assembly
in the following form
⎡ a b a b a b
⎤⎡ ⎤
M11 + M11 M12 + M12 M13 + M13 −AΩ2 sin Ωt
⎣ M21a b
+ M21 M22a b
+ M22 M23 a b ⎦⎣
+ M23 0 ⎦+
a b a b a b
M⎡31 + M31 M32 + M32 M33 + M33 ⎤ ⎡ ẍ3 ⎤ ⎡ a ⎤ ⎡ o ⎤
a b a b a b
K11 + K11 K12 + K12 K13 + K13 A sin Ωt F1 + F1b R1
⎣ a b a b
+ K21 + K21 K22 + K22 K23 + K23 a b ⎦⎣
0 ⎦ = F2 + F2 + R2o ⎦
⎣ a b ⎦ ⎣
a b a b a b
K31 + M31 K32 + K32 K33 + K33 x3 F3a + F3b 0
(19.63)
where ⎡ o ⎤ ⎡ ao ⎤ ⎡ bo ⎤
R1 R1 R1
⎣ R2o ⎦ = ⎣ R2ao ⎦ + ⎣ R2bo ⎦ (19.64)
0 0 0
After partition of the above equation we have
∙ ¸ ∙ ¸
ab −AΩ2 sin Ωt ab ab A sin Ωt
M11 + M12 ẍ3 + K11 + Kab ab
12 x3 = F1 + R
o
(19.65)
0 0
∙ ¸ ∙ ¸
ab −AΩ2 sin Ωt ab ab A sin Ωt
M21 + M22 ẍ3 + K21 + Kab22 x3 = F2
ab
(19.66)
0 0
where
∙ a b a b
¸ ∙ a b
¸
M11 + M11 M12 + M12 M13 + M13
Mab
11 = a b a b ; Mab
12 = a b
M21 + M21 M22 + M22 M23 + M23
£ a ¤ £ ¤
Mab
21 = M31 + M31b a
M32 + M32 b
; Mab a
22 = M33 + M33
b
∙ a b a b
¸ ∙ a b
¸
K11 + K11 K12 + K12 K13 + K13
Kab
11 = a b a b ; ab
K12 = a b
K21 + K21 K22 + K22 K23 + K23
£ a ¤ £ ¤
Kab
21 = K31 + M31b a
K32 b
+ K32 ; Kab a
22 = K33 + K33
b
∙ a ¸
F1 + F1b £ a ¤
; Fab
1 = a b ; Fab
2 = F3 + F3
b
(19.67)
F2 + F2
The equation 19.66 can be rewritten as follows
¡ a b
¢ ¡ a b
¢
M33 + M¡33 ẍ3 + ¢K33 +
¡ a K33 x3 =
¢ ¡ a ¢ (19.68)
= F3a + F3b + M31 + M31b
AΩ2 sin Ωt − K31 b
+ M31 A sin Ω
It represents mathematical model of the assembly without the damper. Application
of the composition procedure to the subsystem 19.68 and the damper results in the
following final mathematical model.
¡ a b
¢ ¡ a b
¢
M33 + M33 ẍ3¡+ cẋ + K
¢ ¡¡ a x3 =b ¢ 2 ¡ a
33 + K33 ¢¢ (19.69)
= F3a + F3b + M31 b
+ M31 Ω − K31 + M31 A sin Ωt
2. the expression for the equilibrium position xC3 of the point C
after assembly.
SAMPLES OF QUESTIONS WITH SOLUTIONS 368

The equilibrium position is represented by the particular solution of the fol-

lowing non-homogeneous equation
¡ a b
¢ ¡ a b
¢ ¡ ¢
M33 + M33 ẍ3 + cẋ + K33 + K33 x3 = F3a + F3b (19.70)
It can be predicted as a constant value
x = xC3 (19.71)
Hence
F3a + F3b
xC3=a b
(19.72)
K33 + K33
3. the expression for the amplitude of the forced vibration of the
point C caused by the motion of the support A.
The forces vibrations are governed by solution of the following equation
¡ a b
¢ ¡ a b
¢
M33 + M33 ẍ3 + cẋ3 + K33 + K33 x3 = sin Ωt (19.73)
After standardization it takes form
ẍ3 + 2ςωn ẋ3 + ω 2n x3 = q sin Ωt
where
¡¡ a b
¢ 2 ¡ a b
¢¢
a b M31 + M31 Ω − K31 + M31 A
c K33 + K33
2ςωn = a b
; ω2n = a b
; q= a b
M33 + M33 M33 + M33 M33 + M33
(19.74)
The particular solution is
x3 = AC sin (Ωt + ϕ) (19.75)
where
q −2ςωn Ω
AC = q ; ϕ arctan (19.76)
2 2 2 2 2 ω2n − Ω2
(ω n − Ω ) + 4ς ω n
4. the expression for the interaction force at the support A
The equation 19.65 offers the relationship between the motion of the system
and the interaction forces Ro along the supported coordinates x1 and x2 .
∙ o ¸ ∙ ¸
o RA ab −AΩ2 sin Ωt
R = o M11 + Mab 12 ẍ3
RB 0 ∙ ¸
ab A sin Ωt
+ K11 + Kab ab
12 x3 − F1 =
∙ a ¸∙ 0 ¸
b a b
M11 + M11 M12 + M12 −AΩ2 sin Ωt
= a b a b +
M21 +∙M21 M22 +M ¸ 22 0
a b
M13 + M13
+ a b (−AC Ω2 sin (Ωt + ϕ)) +
M23 +∙M23 ¸∙ ¸
a b a b
K11 + K11 K12 + K12 A sin Ωt
+ a b a b +
K21 + ∙K21 K22 + K¸22 0 ∙ a ¸
a b
K13 + K13 F1 + F1b
+ a b AC sin (Ωt + ϕ) −
K23 + K23 F2a + F2b
(19.77)
SAMPLES OF QUESTIONS WITH SOLUTIONS 369

Hence
o
¡ a b
¢ 2
¡ a b
¢
RA = M11 + M11 ¡(−AΩ sin Ωt)
¢ + M13 +
¡ M13 (−A¢C Ω2 sin (Ωt + ϕ)) ¡ ¢
a b a b
+ K11 + K11 A sin Ωt + K13 + K13 AC sin (Ωt + ϕ) − F1a + F1b
SAMPLES OF QUESTIONS WITHOUT SOLUTION 370

19.2 SAMPLES OF QUESTIONS WITHOUT SOLUTION

Problem 19

a) 1 2 3 b)
1
2

k c
3
m
c)
1 2
1

y1 , ϕ y1 y 2 ,ϕ y 2

Figure 8

Fig. 8 a) shows a typical design of the support of a rotor by means of a rolling

elements bearing. The sub-assembly comprising the beam 1, the squeeze-film 2 and
the bearing sleeve 3 may be modelled as shown in Fig. 8b). In this figure, k and c
stand for the stiffness and damping of the squeeze film and m stands for mass of the
bearing sleeve. The mathematical model of the free-free beam 1, shown in Fig. 8c),
is as follows
Mÿ + Ky = 0
where
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
M11 M12 M13 M14 K11 K12 K13 K14 y1
⎢ M21 M22 M23 M24 ⎥ ⎢ K21 K22 K23 K24 ⎥ ⎢ ϕy1 ⎥
M =⎢
⎣ M31
⎥; K =⎢ ⎥; y =⎢ ⎥
M32 M33 M34 ⎦ ⎣ K31 K32 K33 K34 ⎦ ⎣ y2 ⎦
M41 M42 M43 M44 K41 K42 K43 K44 ϕy2

Produce the mathematical model of the sub-assembly shown in Fig. 8b)

SAMPLES OF QUESTIONS WITHOUT SOLUTION 371

Problem 20

α = A sin ω t 1 2
a)
R A B
O

A
b)
B

y1 , ϕ y1 y2 ,ϕ y2

Figure 9

The rigid shaft 1 of radius R performs the small oscillatory motion governed
by the following equation.
α = A sin ωt (19.78)
The flexible beam 2 is rigidly attached to this shaft at the point A and it is simply
supported at the point B. The mathematical model of the free-free beam 2, shown
in Fig. 9b), is as follows
Mÿ + Ky = 0
where
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
M11 M12 M13 M14 K11 K12 K13 K14 y1
⎢ M21 M22 M23 M24 ⎥ ⎢ K21 K22 K23 K24 ⎥ ⎢ ϕy1 ⎥
M =⎢
⎣ M31
⎥; K =⎢ ⎥; y =⎢ ⎥
M32 M33 M34 ⎦ ⎣ K31 K32 K33 K34 ⎦ ⎣ y2 ⎦
M41 M42 M43 M44 K41 K42 K43 K44 ϕy2

Produce expression for the interaction force at the support B.

SAMPLES OF QUESTIONS WITHOUT SOLUTION 372

Problem 21

Figure 10 shows a design of the rig for testing the oil film journal bearing. The
journal 1 is rigidly supported and rotates with the angular velocity Ω=1000RP M.
The radial clearance r of the oil bearing is equal to 0.0005m and oil viscosity η is equal
to 0.02N s/m2 . The bearing is loaded by means of the arm 2. Its mass is m=6kg
and its moment of inertia about the point O is IO =0.2kgm2 . The position of the
centre of gravity of the arm is determined by the distance b. The static and dynamic
characteristic of this bearing is given in Fig. 11 and 12 respectively.
In these figures:
c = e/r - non-dimensional eccentricity
r2
S = 2RF3oLηΩ Sommerfeld number
k̄ij = Fo kij ; c̄ij = SrΩ
Sr
c - non-dimensional stiffness and damping coefficient
Fo ij
respectively
Calculate the equilibrium position (x, ϕ) of the arm 2 and produce equation
for the natural frequencies of the rig.

L=0.025[m]

1 2

r
O
Ω G φ
R=0.025[m]
x
b=.3[m]
a=.5[m]

Figure 10
SAMPLES OF QUESTIONS WITHOUT SOLUTION 373

10

6
S 5
a)
4

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 c
-1

-0.8

-0.6

b) -0.4

-0.2
_
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
xo

0.2

0.4

0.6

0.8

_1
yo

Figure 11

10
9
8
7
6
S 5
4
a)
3
2
1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 c
30
25 yy
20
15
_ xx
10
yx xy
k 5 b)
0
-5
-10
-15
-20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 c
35
30 yy
25
_ 20
c 15
10 xx c)
5
yx xy
0
-5
-10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 c

Figure 12
SAMPLES OF QUESTIONS WITHOUT SOLUTION 374

Problem 22

a)
1 2

l l Ω

b)
G
ϕy

y
c)

y1 y2

Figure 13

Fig. 13a) shows a rotor system used for storage of energy. The flywheel 1
rotates with a constant angular velocity Ω and may be approximated by a rigid body.
The mathematical model of the free-free flywheel 1 shown in Fig. 13b) may be
adopted in the following form.
⎡ ⎤⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤
m ma 0 0 ẍ 0 0 0 0 ẋ 0
⎢ ma I 0 ⎥ ⎢ ⎥ ⎢ 0 Iz Ω ⎥ ⎢ ⎥ ⎢ ⎥
⎢ 0 ⎥ ⎢ ϕ̈x ⎥ + ⎢ 0 0 ⎥ ⎢ ϕ̇x ⎥ = ⎢ 0 ⎥
⎣ 0 0 m ma ⎦ ⎣ ÿ ⎦ ⎣ 0 0 0 0 ⎦ ⎣ ẏ ⎦ ⎣ mg ⎦
0 0 ma I ϕ̈y 0 −Iz Ω 0 0 ϕ̇y mga

The shaft 2, shown in Fig. 13c) may be treated as massless and its stiffness matrix
k along the coordinates y1 and y2 is the same as along coordinates x1 and x2 .
∙ ¸
K k
k=
k K

Produce expression for natural frequencies of the rotor system as a function of the
flywheel angular velocity Ω.
SAMPLES OF QUESTIONS WITHOUT SOLUTION 375

Problem 23

a)

y2a y3a y 3b y 4b y5b

y1a

b)
b
y2a y3a y 3b y5b
y1a a y 4b

c)
A B C

Ω
y

Figure 14

The mathematical models of the two free-free shafts shown in Fig. 14a) are
condensed along vector of coordinates ya and yb respectively.

ma ÿa + ka ya = FaG mb ÿb + kb ya = FbG

where:
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
ma11 ma12 ma13 a
k11 a
k12 a
k13 Ga1 y1a
ma = ⎣ ma21 ma22 ma23 ⎦ ; ka = ⎣ a
k21 a
k22 a ⎦
k23 ; FaG = ⎣ Ga2 ⎦ ; ya = ⎣ y2a ⎦
ma31 ma32 ma33 a a a
k31 k32 k33 Ga3 y3a
⎡ b ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
m11 mb12 mb13 b
k11 b
k12 b
k13 Gb1 y1b
mb = ⎣ mb21 mb22 mb23 ⎦ ; kb = ⎣ b
k21 b
k22 b ⎦
k23 ; FbG = ⎣ Gb2 ⎦ ; ya = ⎣ y2b ⎦
mb31 mb32 mb33 b b b
k31 k32 k33 Gb3 y3b
These shafts are permanently deflected and their permanent deformation is deter-
mined by the parameters a and b (see Fig. 14b)). These shafts were connected to
each other by means of a universal joint and supported as shown in Fig. 14c). This
rotor system rotates with a constant angular velocity Ω. Produce expression for the
reaction force at the support C due to the permanent deformation of these shafts.
SAMPLES OF QUESTIONS WITHOUT SOLUTION 376

Problem 24

a)
A

M
b)

C
K

c)

y1 y2 y3

Figure 15

To reduce the interaction force transmitted to foundation through the bearing

A, the rotor shown in Fig. 15a) was supply with a damper. The modified system is
shown in Fig. b). Produce the mathematical model of the modified system
Given are:
1. The mathematical model of the free-free rotor shown in Fig. 15c)

mẍ + kx = 0 mÿ + ky = G

where
⎡ ⎤ ⎡ ⎤
m11 m12 m13 k11 k12 k13
m = ⎣ m21 m22 m23 ⎦ ; k = ⎣ k21 k22 k23 ⎦
m31 m32 m33 k31 k32 k33
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x1 y1 G1
x = ⎣ x2 ⎦ ; y = ⎣ y2 ⎦ ; G = ⎣ G2 ⎦
x3 y3 G3

2. M, K, C - mass, stiffness and damping coefficient of the damper

3. Ω - angular velocity of the rotor
4. m - static unbalance of the rotor in plane 3.
SAMPLES OF QUESTIONS WITHOUT SOLUTION 377

Problem 25

a) 3
1 2

0.3m 0.5m

b) 3 R=0.025m
1 2

L=0.025m
0.3m 0.5m

Figure 16

The inertia matrix m, the stiffness matrix k and the gravity vector G deter-
mine the dynamic properties of the free-free rotor in the vertical plane yz (see Fig.
16a)
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
0.5 0.1 0.2 6 −2 −1 4
m = ⎣ 0.1 1 0.3 ⎦ kg; k = ⎣ −2 3 −1 ⎦ · 105 N; G = ⎣ 8 ⎦ N
0.2 0.3 1 −1 −1 5 10
⎡ ⎤
y1
y = ⎣ y2 ⎦
y3
This rotor is supported by means of the self-aligning ball bearing at its node 1 and by
means of the oil bearing at the node 3 as shown in Fig. 16b). Radial clearance r of the
oil bearing is equal to 0.0005m and oil viscosity η is equal to 0.02Ns/m2 . The shaft
rotates with angular velocity Ω=1000RP M. The static and dynamic characteristics
of this bearing are given in Fig. 11 and 12 respectively.
In these figures:
c = e/r - non-dimensional eccentricity
r2
S = 2RF3oLηΩ Sommerfeld number
k̄ij = Sr k ; c̄ij = SrΩ
Fo ij
c - non-dimensional stiffness and damping coefficient
Fo ij
respectively
Calculate the equilibrium position and produce the mathematical model of the rotor-
bearing system.
SAMPLES OF QUESTIONS WITHOUT SOLUTION 378

Problem 26

a)

y1a y2a y3a y 2b y 3b y4b

b)

y2 y3

c)
M

y2 y3

Figure 17

The mathematical models of the two free-free beam elements shown in Fig.
17a) are condensed along vector of coordinates ya and yb respectively. The corre-
sponding matrices of inertia, stiffness and gravity are as follows
⎡ a ⎤ ⎡ a ⎤ ⎡ a ⎤ ⎡ a ⎤
m11 ma12 ma13 a
k11 k12 a
k13 G1 y1
ma = ⎣ ma21 ma22 ma23 ⎦ ; ka = ⎣ k21 a a
k22 a ⎦
k23 ; FaG = ⎣ Ga2 ⎦ ; ya = ⎣ y2a ⎦
ma31 ma32 ma33 a
k31 a
k32 a
k33 Ga3 y3a
⎡ b ⎤ ⎡ b ⎤ ⎡ b ⎤ ⎡ b ⎤
m11 mb12 mb13 b
k11 k12 b
k13 G1 y1
b
m = ⎣ m b
21 m b
22 m b ⎦
23
b
; k = ⎣ k b
21 k b
22 k b ⎦
23
b
; FG = ⎣ G b ⎦
2 ; y = y2b ⎦
a ⎣
mb31 mb32 mb33 b
k31 b
k32 b
k33 Gb3 y3b
These two elements were joined together to form a bridge shown in Fig. 17b). If
a person of mass M is standing at the position shown in Fig. 17c), determine the
static deflection of the bridge along the coordinates y2 and y3 as well as its natural
frequencies.
SAMPLES OF QUESTIONS WITHOUT SOLUTION 379

Problem 27

a)
Z

y1a y2a y3a

b) b
Z

y 1b y 2b y3b
c) Z

y1 c1 y2 c2 c3
y3

Figure 18

The mathematical model of the foundation shown in Fig. 18a) is

ma ÿa + ca ẏa +ka ya = 0

where
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
ma11 ma12 ma13 a
k11 a
k12 a
k13 ca11 ca12 ca13
ma = ⎣ ma21 ma22 ma23 ⎦ ; ka = ⎣ k21
a a
k22 a ⎦
k23 ; ca = ⎣ ca21 ca22 ca23 ⎦
ma31 ma32 ma33 a a
k31 k32 k33a
ca31 ca32 ca33
⎡ a ⎤
y1
y = y2a ⎦
a ⎣
y3a
The motion of this foundation along the vector coordinates ya is
⎡ ⎤ ⎡ ⎤
es1 ec1
e = ⎣ es2 ⎦ sin ωt + ⎣ ec2 ⎦ cos ωt = es sin ωt + ec cos ωt
es3 ec3

The mathematical model of the free- free beam shown in Fig. 18b) is condensed along
the vector of coordinates yb and has the following form

mb ÿb +kb yb = FG
SAMPLES OF QUESTIONS WITHOUT SOLUTION 380

where
⎡ ⎤ ⎡ b ⎤ ⎡ ⎤ ⎡ b ⎤
mb11 mb12 mb13 b
k11 k12 b
k13 G1 y1
b ⎣ b b b ⎣
m = m21 m22 m23 ; k = k21 k22 k23 ; FG = G2 ; y = y2b ⎦
b ⎦ b b b ⎦ ⎣ ⎦ a ⎣
mb31 mb32 mb33 b
k31 b
k32 b
k33 G3 y3b

Its permanent deformation is determined by the misalignment b shown in Fig. 18b).

This beam was joined with the foundation as shown in Fig. 18c).
Produce the expression for
1. the static equilibrium position of the system shown in Fig. 18c)
2. the vector of amplitudes of vibration of the system caused by the environ-
mental excitation e
SAMPLES OF QUESTIONS WITHOUT SOLUTION 381

Problem 28

a)

1 2
3 5 4

b)

3 5 4
1 k 2
k5 k
4
y3 y
5 y
4
2 3
c)
2
y1 1
y2 y1 y2 y2 y
1 3 3 4 5

Figure 19

Fig. 19 shows the design of a bridge suspension bridge. Its physical model
was accepted in the form shown in Fig. 19 b). To produce the mathematical model
of the bridge it was split into 5 sub-systems shown in Fig. 19c). The inertia matrix
and the stiffness matrix for the free-free sub-system 1 and 2 are identical

Mÿ1 + Ky1 = 0 Mÿ2 + Ky2 = 0

where
⎡ ⎤ ⎡ ⎤ ⎡⎤ ⎡ ⎤
M11 M12 M13 K11 K12 K13 y11 y12
M = ⎣ M21 M22 M23 ⎦ K = ⎣ K21 K22 K23 ⎦ y1 = ⎣ y21 ⎦ y2 = ⎣ y22 ⎦
M31 M32 M33 K31 K32 K33 y21 y22

The stiffness of the springs 3, 4 and 5 are respectively k, k, and k5 .. Produce the
mathematical model of the bridge and the equation for its natural frequencies.
SAMPLES OF QUESTIONS WITHOUT SOLUTION 382

Problem 29

2
1

a)
z

1
y y5

b)

1
y1 2 y21 y31 y41

c)

2
y22 y52 y3

Figure 20

The rotor 1 of the assembly shown in Fig. 20a) rotates with the constant
angular velocity ω 1 and the rotor 2 rotates with the constant angular velocity ω 2 .
Motion of the free-free shaft 1 in the yz plane (see Fig. 20b)) is governed by the
following equation.
M1 ÿ1 + K1 y1 = 0
where
1
⎡ 1 1 1
⎤ ⎡
1 1 1 1
⎤ ⎡⎤
M11 M12 M13 M14 K11 K12 K13 K14 y11
⎢ M21
1 1
M22 1
M23 1 ⎥
M24 ⎢ K21
1 1
K22 1
K23 1 ⎥
K24 ⎢ y21 ⎥
M1 = ⎢
⎣ M31
1 1 1 1 ⎦
⎥ K1 = ⎢
⎣ K31
1 1 1 1 ⎦
⎥ y1 = ⎢ ⎥
⎣ y31 ⎦
M32 M33 M34 K32 K33 K34
1 1 1 1 1 1 1 1
M41 M42 M43 M44 K41 K42 K43 K44 y41

This shaft is permanently deflected in the plane

£ yz and the vector
¤t of the permanent
deformation along the coordinates y1 is a1 = 0 a12 a13 0 .
Motion of the free-free shaft 2 (see Fig. 20c)) is governed by the following
equation
M2 ÿ2 + K2 y2 = 0
SAMPLES OF QUESTIONS WITHOUT SOLUTION 383

where
⎡ 2 2 2
⎤ ⎡ 2 2 2
⎤ ⎡ ⎤
M22 M23 M25 K22 K23 K25 y12
M2 = ⎣ M32
2 2
M33 2 ⎦
M35 K1 = ⎣ K32
2 2
K33 2 ⎦
K35 y1 = ⎣ y32 ⎦
2 2 2 2 2 2
M52 M53 M55 K52 K53 K55 y52

This shaft is permanently deflected in the plane

£ yz and the
¤t vector of the permanent
deformation along the coordinates y2 is a2 = 0 a23 0 .
Produce the expression for the forced vibrations of the assembly along the
coordinatey5 .
SAMPLES OF QUESTIONS WITHOUT SOLUTION 384

Problem 30

a)

w
1
2
z
b)

2
y z
c)
y1 y3
y2

Figure 21

In order to model the rotor shown in Fig.21a), it was approximated by the

thin and rigid disk 1 (w = 0) and the shaft 2 (see Fig.21b)). The disk has mass m
and its principal moments of inertia are Ix = Iy = I, Iz . The mathematical model
of the free-free shaft 2 was condensed along the set of three coordinates shown in
Fig.21c). It is as follows

Mÿ + Ky = G
where
⎡ ⎤ ⎡ ⎤ ⎡
⎤ ⎡ ⎤
M11 M12 M13 K11 K12 K13 G1 y1
M = ⎣ M21 M22 M23 ⎦ K = ⎣ K21 K22 K23 ⎦ G = ⎣ G2 ⎦ y = ⎣ y2 ⎦
M31 M32 M33 K31 K32 K33 G3 y3

The shaft 2 is permanently deformed and the vector of the permanent deflec-
tion along the coordinates y is

a = [0, 0, α]T
The rotor rotates with the constant angular speed Ω.
Produce
SAMPLES OF QUESTIONS WITHOUT SOLUTION 385

1. the mathematical model of the rotor

2. the equation for the determination of the natural frequencies of the rotor
as a function of its rotating speeds Ω
3. the expression for the amplitudes of the forced vibrations due to the per-
manent deflection a of the shaft 2
SAMPLES OF QUESTIONS WITHOUT SOLUTION 386

Problem 31

F 2

B
k k l
A
C

c 1 c a)

b)
1
y11 y2 1
y3 1

Figure 22

The two identical beams 1 and 2 are connected to each other by means of two
massless springs of the stiffness k and two dampers of the damping coefficient c as
shown in Fig.22a). The uncompressed length of the springs is l. The mathematical
model of the free-free beam 1 (see Fig.22b) is as follows

Mÿ1 + Ky1 = G
where
⎡ ⎤ ⎡ ⎤ ⎡⎤ ⎡ ⎤
M11 M12 M13 K11 K12 K13 G1 y11
M = ⎣ M21 M22 M23 ⎦ K = ⎣ K21 K22 K23 ⎦ G = ⎣ G2 ⎦ y1 = ⎣ y21 ⎦
M31 M32 M33 K31 K32 K33 G3 y31

Produce:
1. the mathematical model of the system shown in Fig.22a)
2. the vector of the coordinates that determines the equilibrium position of
the system due to the gravity forces G.
3. the expression for the amplitude of the forced vibration at the point C
caused by the exciting force F
F = Fo cos ωt
4 the expression for the interaction force at the support A