2010:003

2010:003
CIV CIV
2010:003 CIV

MA
S T E RS THESIS
T H E SI S
M
A
STERS
M A STERS THESIS

Analyses
of
a
Analyses of a
Rotor
Dynamic
Testrigs
Rotor Dynamic Testrigs

Athanasios Nassis
Athanasios Nassis
Athanasios Nassis
MASTER OF SCIENCE PROGRAMME
MASTER
OF SCIENCE
PROGRAMME
Mechanical
Engineering
Mechanical Engineering
Lule University of Technology

University
MASTER
OF
SCIENCE
PROGRAMME
Department
ofLule
Applied
Physics of
andTechnology
Mechanical Engineering
Mechanical
Engineering
Department of Applied
Physics
and
Mechanical Engineering
Division of Solid Mechanics
Division of Solid Mechanics

Lule University of Technology

#)6s)33. s)32.,45 %8  3%
Department of Applied Physics and Mechanical Engineering
#)6s)33. s)32.,45 %8  3%
Division of Solid Mechanics

2010:003 CIV ISSN: 1402 - 1617 ISRN: LTU - EX - - 10/003 - - SE

Preface
Rotating machines and rotor dynamics is a specific area of dynamics and has
been changed very little and in a low pace during the last decades. Early
engineers did not know what would happen in case they could exceed to
operate a system over the first natural frequency. Finally engineers like De
Laval succeed to operate a rotor way above the critical speed. Nowadays,
engineers try to work with more complicated problems or to develop techniques that could determine the dynamics of a real machine more accurately.
A similar study regarding rotating machines took place at the period
Marc-July 2009, at Lule
a University of Technology in Sweden. The main
purpose of this study was to combine theoretical and experimental results
in different systems. Furthermore, the use of Rotor Kit RK4 was part of
this study to evaluate the dynamics of a small scale rotor.
March-July 2009

Acknowledgements
The writer of the report gratefully acknowledges the helpful assistance by:
Professor Jan Olov Aidanp
a
a who was always there to provide his assistance
in any encountered obstacle, PHD students Yogeshwarsing Calleecharan and
Jean-Claude Luneno for their precious feedback which made the whole work
exciting, in order to dive deeper and extract new information that could be
challengingly used for present and future research, and Jan Granstrom who
helped in order to fulfill the connection between the experimental equipment
and the computer, to program and develop a suitable environment for the
user.
Furthermore, the writer dedicates the present report with all his
heart to his family(Dimitrios, Efthymia, Georgios and Barbara)
for their support and encouragement in times of difficulties and
doubt.....

Abstract
This thesis report describes how to develop a mathematical model of a
rotordynamical system, either the reader is a beginner or has a background
in dynamical systems. The theory is described in order to conceive the main
idea how to develop rotor dynamical models. The problems are based on
FEM and described in detail.
In addition, the educational Rotor Kit RK4 will be introduced and
analysed. This rotor equipment is especially developed for measuring and
detecting different phenomena that occur under operation of a rotating system. Many setups and options will be shown in order understand how Rotor
Kit RK4 can be used. That will help one to develop models from a simple
to a more complicate system. Thus, according to each setup the mathematical model will be adapted once with a single mass, once with two masses,
perturbation at different spans and so forth. For each system different weaknesses of the theoretical model is observed. The main purpose is to identify
the reason of having a model unable to calculate the right results. That
will be carried out by changing different values of the model like different
number of elements, and by including other facts which were assumed that
not exist but finally affect the final computation.
The mathematical models which were developed for this thesis are three.
First model is a simple system with rigid points of support. The code
calculates only the rotor regarding its properties and objects attached on it.
The second model is a system that includes bearing properties. The system
now can be slightly move up and down at the support which shows that
the system became more sensitive. The third model includes the extent
part of the rotor out of the bearings. The real rotor kits shaft does not
stop at the bearings, but it has a small part of its length out of them at
both ends. It is understandable that, improving a model by adding other
facts like bolts, nuts and generally the environment in which the real system
works, computations can be achieved to be closer to the real results.
The used equipment has unlimited choices in developing experimental
tasks so long as someone has a vast fantasy. Moreover, the equipment provides the ability to easily adapt any real system, to almost the same in
a smaller scale. Many experimental tasks are shown which were designed
for students in order to get a better understanding about Rotor Kit and
Rotor dynamics.

Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 Rotor dynamic modeling and analysis

1.1 Element matrices . . . . . . . . . . . .
1.1.1 Fundamental beam theory . . .
1.1.2 Finite element method . . . . .
1.1.3 Bearing matrices . . . . . . . .
1.2 Assembly process . . . . . . . . . . . .
1.2.1 System matrix assembly . . . .
1.2.2 Assembly of the rigid disc . . .
1.2.3 Boundary conditions . . . . . .
1.3 The general eigenvalue problem . . . .
1.3.1 General eigenvalues . . . . . .
1.3.2 Campbell diagram & speeds . .
1.3.3 Particular solution . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

4
4
4
6
7
8
8
9
10
12
12
14
14

2 Rotor Kit-Bently Nevada

2.1 Introduction . . . . . . .
2.2 Rotor Kit . . . . . . . .
2.2.1 Oil whirl & whip
2.2.2 Perturbator . . .
2.3 Probes & calibration . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

16
16
16
17
18
18

.
.
.
.
.
.
.
.

22
22
22
26
30
30
31
32
32

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

3 Experimental & theoretical results

3.1 Rotor effect of a disc mass & perturbation . . .
3.1.1 Jeffcott rotor . . . . . . . . . . . . . . .
3.1.2 Disc mass at 2/3 of the span . . . . . .
3.2 Journal bearing . . . . . . . . . . . . . . . . . .
3.2.1 Oil wedge force . . . . . . . . . . . . . .
3.3 Whirl & whip . . . . . . . . . . . . . . . . . . .
3.3.1 Oil pressure control . . . . . . . . . . .
3.3.2 Journal bearing whirl-whip phenomena
A Rotor Kit specifications

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

38

B Proximity probe
C

39

High magnification extensometer calibrator

D Experimental suggestions
D.1 Disc & unbalance mass .
D.2 Preload condition . . . .
D.3 Rub condition . . . . . .
D.4 Oil whirl/whip . . . . .
D.5 Perturbator . . . . . . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

E Perturbation orbits

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

40
.
.
.
.
.

.
.
.
.
.

42
42
43
44
44
44
45

Introduction
During the last 50 years engineers have developed several new techniques
and solved many problems that industry is facing in dynamics of rotating
machines. However the demands and reliability on machines that can operate in hard conditions are expected to increase. In this study one of the most
significant goals was to analyse and evaluate the rotor kit RK4. The rotor is
especially designed for educational purpose and it will be used in dynamic
courses at LTU. Therefore, it is important to identify potential errors or
assumptions that significantly affect the results by comparing experimental
with theoretical results. Thus, a theoretical model was developed for the
rotor kit equipment to extract information not only about the assumptions
done in order to simplify the study of a system, but also to observe how the
same environment in which the system operates affects the results.
In addition, several studies have be done with rotor kit which indicate
the ability of unlimited configurations and experimental tasks that can be
achieved. Some remarkable references are: Active vibration control of rotor
by Kari Tammi with the use of special accessory to develop magnetic field
round the rotor [1]. Response of a warped flexible rotor with a fluid bearing
by Jim Meagher, Ci Wu and Chris Lencioni [2]. Investigation of vibration of
a rotor system supported by absolutely rigid bearings with a shaft containing
a notch by Petr Ferfecki, Jan Ondrouch and Tomas Lukas [3]. The main
purpose was to study a rotor system with a shaft weakened by a notch.
Determination of oil whip phenomena by Agnes Muszynska [4]. Experiment
design for simulating different faults in a rotor kit by Enayet Halim [5].
A magnetorheological fluid damper for rotor applications by P. Forte, M.
Paterno, and E. Rustighi [6].
The report describes the theory, equipment and experimental results
of different setups. In the first section the theory used for developing the
theoretical models is described. The theory shows how to develop matrices,
build a FEM model and calculate the natural frequencies by using General
Eigenvalue Problem. Section two describes the rotor kit RK4 and how it
can be used for educational purpose. Third section presents experimental
results and comparison with theoretical results.
The purpose of this thesis, as mentioned above, is to identify why calculations are not the same or sometimes not even close at all to the real values.
Thus, the main goal is, to track down what more should be taken into consideration in the computation (like bearings, support and so forth). The
thesis describes also several experimental setups that will give undergraduate students the opportunity to compare theory with the reality. Finally,
the results are discussed together with suggestions how the experimental
setups could be improved.

Chapter 1

Rotor dynamic modeling and

analysis
1.1

Element matrices

This section will describe the theoretical method and techniques that were
used to develop a mathematical model. In addition, explanations are given
to better understand how the matrices are built.

1.1.1

Fundamental beam theory

In rotor dynamics, problems can be approached with different methods, like

the use of classical formulas of physics or by FEM method. Even though
FEM is more difficult to define, it is an effective method to simulate dynamical problems. In this section the main purpose is to describe the fundamental
theory on how to build an element.
First, a prismatic beam, Figure 1.1, subjected to a constant force is
described. Then the shape functions of an element can be found for static
conditions. According to the theory [7] a beam has to satisfy the equation
of the elastic line,


2
2u
EI 2 = 0
(1.1)
x2
x
thus, it can be assumed that,

u(l,t)

c1 (t)l3 + c2 (t)l2 + c3 (t)l + c4 (t)

(1.2)

Where, c1 , c2 , c3 and c4 are the constants that can be found from the
boundary conditions, and x1 , 1 , y1 , 1 , x2 , 2 , y2 and 2 for the displacements and rotations respectively. The behavior of a prismatic beam can be
4

2(2)
x(y)

4(4)

2(2)

x1(y1)

()

x3(y3)

1(1)

l1

x(y)

x2(y2)

l2

zi
x1

zi+1
x3

x1(y1)

y1

y3

Figure 1.1: Prismatic beam

described with a shape function graph. From the theory, book [8], of FEM
and prismatic beams, the four shape functions for the element in Figure 1.1,
will be,

h1 (l)
h3 (l)

=
=

1 3s2 + 2s3 ,
3s2 + 2s3 ,

h2 (l)
h4 (l)

=
=

L(s 2s2 + s3 )
L(s2 + s3 )

(1.3)

where for s = l/L and L the length of the entire element. The same shape
functions 1.3 stand for the y direction. From the general solution 1.2 now
can be found that:
u(l,t)
u(l,t)

=
=

h1 (l)x1 + h2 (l)1 + h3 (l)x2 + h4 (l)2

h1 (l)y1 + h2 (l)1 + h3 (l)y2 + h4 (l)2

(1.4)

The kinetic energy, T, can be introduced as,

1
T (t) =
2

ZL


A

du(l, t)
dt

2
dl

(1.5)

which should be equal to,

T (t) =

1  T
u [M e ] {u}

(1.6)

Since it is assumed that the equation 1.5 and 1.6 are equal, the M e mass
matrix can be determined. A similar approach is used for the K e stiffness
matrix. For the stiffness matrix the kinetic energy will be,
1
T (t) =
2

ZL


EI

2 u(l, t)
l2

2
dl

(1.7)

which should be equal to,

T (t) =

1  T e
u [K ] {u}
2
5

(1.8)

After the integration the elements of matrix K e can be identified. Every

part that can be described with matrices, from now on can be called a finite
element. The inverse of the stiffness matrix A(e) = K (e)1 is the flexibility
matrix [9] .

1.1.2

Finite element method

Finite element is today a frequently used method which help engineers to

create models. The accuracy of the results is related on how many elements
are used for a problem. However, that does not imply that one has to use
as many elements as possible. For instance, if it is of interest to find the
1st and 2nd natural frequency of a beam problem it is enough to use only
two-four elements.
Nevertheless, a higher number of elements are necessary, if we are interested in beams of more complex shapes or in finding mode shapes of the
beam. In addition, the element type is selected due to the complexity of the
problem. In this study were chosen to use 88 element matrices.


(e) = 1 y1 1 1 2 y2 2 2

(1.9)

The above vector represents the displacements and rotations at the left
and right node of an element. The following matrix is the stiffness matrix
that was used at the present study.

12
0
0
6L 12
0
0
6L
0
12 6L
0
0
12 6L
0

2
2
0
6L 4L
0
0
6L 2L
0

2
2
EI
6L
0
0
4L
6L
0
0
2L

= 3
0
0
6L 12
0
0
6L
L
12

0
12 6L
0
0
12
6L
0

0
6L 2L2
0
0
6L 4L2
0
6L
0
0
2L2 6L
0
0
4L2

K (e)

(1.10)

This is an 8 8 matrix where, E is the Youngs modulus, I = r4 /4 is

area moment of inertia, r the radius of the element and L the element length.
Assuming LT otal the total length of the uniform rotor and n the number of
the elements, a single element length will be equal to L = LT otal /n.
The following matrix determines the mass of each element. The mass
matrix takes into account only the mass of the rotor, therefore, it will later
be shown how the mass of the disc can be added to the right node.

156
0
0

22L
54

0
0
13L

M (e) =

m
420

0
156
22L
0
0
54
13L
0

0
22L
4L2
0
0
13L
3L2
0

22L
0
0
4L2
13L
0
0
3L2

54
0
0
13L
156
0
0
22L

0
54
13L
0
0
156
22L
0

0
13L
3L2
0
0
22L
4L2
0

13L
0

0
3L2
22L

0
0
4L2

(1.11)

The bearings of the rotor kit were considered to be very stiff, thus,
both cases were included in the final computations, with stiff bearings and
flecible bearings. The gyroscopic effect results in torques on the system at
the node on which a polar inertia is attached. One method is to directly add
the moment of inertia at the right node, something that will be explained
further at the assembly section. Thus, the damping matrix can be of 4 4
size which in fact represents only a single node. This matrix will directly be
placed in the final matrix at the right node.
0
0
Cd =
0
0

0
0
0
0
0
0
0
0
Jp
0 Jp
0

(1.12)

Note that this is the method which was used for this thesis, however,
one can use matrices of 8 8 size, which is suggested in some books, for
instance [9]. Usually those matrices include both nodes which overlap each
other, and therefor they are always multiplied by 1/2. Matrix 1.12 is for
a disc on one node, where = n2/60 is the driving frequency with n
the rotation speed in rpm. The Jp = 21 md R2 is the polar mass moment of
inertia, where md the mass of the disc and R the radius.
The last matrix is the Md matrix that contains information about the
disc. The disc mass matrix is developed to describe only where the disc is
positioned,

md 0
0 0
0 md 0 0

Md =
(1.13)
0
0 Jd 0
0
0
0 Jd


2
m t 2
where, md is the disc mass, Jd = 2[ mR
8 + 6 2 ] the mass moment of
inertia and t the width of the disc.

1.1.3

Bearing matrices

If a rotor boundary conditions are flecible, then the bearing conditions can
be added to the node as a damping and stiffness matrix. The node stiffness
matrix due to concentrated bearing properties is,
7

(n)

Kbearing

Kxx Kxy
Kyx Kyy
=
0
0
0
0

0
0
0
0

0
0

0
0

(1.14)

and the node damping matrix due to the damping of the bearing is

Cxx Cxy 0 0
Cyx Cyy 0 0
(n)

(1.15)
Cbearing =
0
0 0 0
0
0 0 0
These are the matrices which can increase the sensitivity of a system,
since they take into account the journal bearing motion. Including bearing
conditions to a system, will give lower frequency. The final matrices now
will be,

Mf
Cf
Kf

=
=
=

Mr + Mce
Cr
Mr + Mce

(1.16)

where, Mce &Kce are the rotors element mass and stiffness matrix respectively( equation 1.11 and 1.10 ), and Mr , Cr &Kr are the matrices for
disc mass, bearing damping and bearing stiffness matrix respectively. The
Mr , Cr &Kr are the matrices in which are placed the Md (1.13), Cd (1.12),
Cbearing (1.15) and Kbearing matrix(1.14) which are of 4 4 size. In the
following sections the assembly process is described to develop a system.

1.2

Assembly process

Introduction
From the theory is known that a model can be assembled in different ways.
The present study is based on Finite Element Method and in the following
sections the process is described in detail.

1.2.1

System matrix assembly

According to Finite Element Method, a beam or a rotor is considered to

be a system that can be easily described with one or more elements. It is
obvious that more elements results in more realistic solutions. Nevertheless,
it is not always suitable to use extremely high number of elements due to
time consuming computations.

The system matrix assembly can be developed with elements that overlap
each other at the common nodes, as indicated in Figure 1.2. To begin
with, a beam can be divided in smaller elements that are connected at
their ends(nodes). Each element consists of two nodes(n1 and n2 ) and the
element that connects those two. In this case each element is a matrix.
From the figure is obvious that four elements are coupled in the assembled
matrix(Element 1,2,3 and 4). The symbols 1 , 2 , 3 and 4 , are the vectors
for each element and in the present case they represent the displacement
vectors. At the connection point the second node n12 of the first element
must to be linked with the first node n21 of the second element. The same
procedure will be repeated for all nodes.
x1

(1)

x2

(2)

x3

(3)

A
(4)

x4
x5

}
}
}
}
}

n11

n11

e(1)

n21

n12+n21

n12
e(2)

2
n31

n22+n31

n22

n32+n41

n41

n32
e(4)

4
n42

Beam

e(3)

n42

Figure 1.2: Overlap matrices - Beam matrix.

The final result is a matrix with a size related to the number of assembled
elements. The connected parts can be presented as matrices that overlap
at the common nodes in order to make a system that interacts physically,
(A,B and C are the overlap parts). Regarding the previous sections and
the introduced matrices, each node consist of four quantities (i , yi , i , i ).
Thus, the matrix in Figure 1.2 will be of 20 20 size.

1.2.2

Assembly of the rigid disc

There are different ways to include an additional inertia matrix in rotor

dynamics. One simple way is to consider mass and inertia of the disc as
nodal properties and just include them to one (4 4matrices) or two nodes
(8 8matrices). One node gives a point property and two nodes are used
when the thickness is included [9]. Inertia properties can also be included
with consistent matrices as in [9] .
Figure 1.3 shows how a 4 4 disc matrix is placed in the global matrix.
This kind of matrix is diagonal and the imaginary dash lines in the figure
split the main matrix in submatrices of 4 4. Since the node on which the
disc is attached is known, a disc matrix 44 can be placed on the diagonal of
the matrix size[N], at the position which represents the node where the disc
9

Zeros[N]

0000
0000
0000
0000

[Md]

(1)
(2)
(3)

+
(4)

m55
m65
m75
m85

m56 m57 m58

m66 m67 m68
m76 m77 m78
m86 m87 m88

Figure 1.3: Disc matrix.

0000
0000
0000
0000

Diagonalmatrix

is attached. The 4 4 matrix used in this report is similar to 1.13. In this

case the disc is at the second node, and as a result the disc matrix is placed
at the second diagonal dashed square on the global matrix (matrix area:
m55 m88 ). In the figure, the left matrix represents Mce where (1),(2),(3)
and(4) are overlap matrices similar to mass matrix (1.11), and the right the
Mr with Md (1.13) placed at the second diagonal dashed square. According
to equation (1.16) these two matrices can be added together, in order to get
the final matrix Mf .

1.2.3

Boundary conditions

Boundary conditions are important for the final results. From now on,
boundary conditions will be noted as BC. If the BC can be determined with
good accuracy, then a mathematical model will be able to provide more
realistic results.
To be able to determine BC it is important to get some information about
the system. To become more specific, information about the bearings and
the general environment in which a rotor operates. For instance, what kind
of material do the bearings have and which properties, oil or any other kind
of lubrication between the contact surfaces and so forth. On the other hand,
if the support points could be considered stiff, then they can be discarded
form the global matrix, since the final displacement = 0.
To give a better understanding about the BC, a rotor with n elements
is shown in Figure 1.4. The previously defined mass matrix is taken as an
example. The rotor is supported with bearings that enable rotation about
x and y axis. Here it has to be mentioned, that after a laboratory control,
the bearings of Rotor Kit allow the rotor operating in slight rotations about
x and y axis, as indicated in Figure 1.5.a. In the case of Figure 1.4 it is
obvious (if the support is considered to be very stiff) that the rotor is not

10

MassMatrixSize[N]
(1)
x1
y1
1
1
xi
yi
i
i

BC
(i)

[Md]

BC

(i+1)

(n1)

xn
yn
n
n
xn+1
yn+1
n+1
n+1

BC

BC

Figure 1.4: Determination of Boundary conditions

able to move in x and y direction at the 1st and nth node. Thus, the columns
and rows of the matrix that represent those nodes in x and y direction have
to be discarded or replaced with zero(0). In the figure those columns and
rows are shown with BC lines.
There are some cases where a deeper study is necessary and the behavior of the bearings might be of a big interest. In this case the nodes that
represent the BC of the system should not be discarded, but in a contrary a
damper-spring system shall be added at the main system that can describe
the motion of the support. Thus, it is evident that if the same system is supported with a damper-spring system, the values for the natural frequencies
would be lower in relation with a rigidly supported system.

a.
Ky

2 Bearing

i
xi

z
i+1
xi+1

yi

Cxx

i+1

DetailA

Kxx

yi+1

Ff

DetailB
F

Cy

1 Rotor

b.

i+1

Figure 1.5: Cross correlation of a bearing with a damper-spring system

Figure 1.5.a shows an element whose left node is located in a slippingbearing. These couples of damper-spring system in x and y direction, rep11

resent the systems BC. Further more, Figure 1.5.b shows how the damperspring system of a BC can be described. It is known from Rotor Dynamics,
that any friction inside a system develops damping forces [9]. Detail A shows
what occurs at the BC while a system is operating. While the system is operating, bowing of the rotor occurs due to grown vibrations. Thus bearing
surface is under pressure p and that grows friction forces Ff . A simple way
to determine the spring stiffness is by using Hookes law, as:
Kx = F A = F

(1.17)

thus, the spring stiffness can be easily found by applying a known force F
on the bearing and measuring the displacement x.
In Detail B is shown the bearings behavior under pressure of an F force.
Moreover, similar phenomena occur at the attached points. Attached point
is called any point along the span on which a rigid disc or other object is
mounted.

1.3

The general eigenvalue problem

The general eigenvalue problem is a useful method for problems that include damping. This method is easy to use compared with other methods
that can be used on complex systems. Furthermore, the general eigenvalue
method provides more information about a systems behavior. With the
development of programs, this method is today frequently more suitable for
very complex problems. Once the eigenvalues are solved one can directly
get information about natural frequency, damped natural frequency and the
damping ratio.

1.3.1

General eigenvalues

From the classic dynamics is known that a system can be defined from the
Second Law of Newton. According to the 2nd law of Newton, a system can
be described as,
Mx
+ C x + Kx = 0

(1.18)

where, M is the mass matrix, C the damping matrix which can be replaced
with D = C + Jp = Cf , where Jp is the damping matrix due to gyroscopic
effect, K the stiffness matrix, and x
, x & x the vectors for the acceleration,
velocity and displacement respectively. If they state vectors are defined as
y = (
x x)
0 and y = (x x)0 , then the equation 1.18 according to [10] can be
written as,




K 0
0 K
y +
y=0
(1.19)
0
M
K C
12

or





I 0
0 I
y +
y=0
0 M
K C

where, I is the unit diagonal matrix and 0 zero matrix. Assuming that the
first matrix is -S and the second R, the equation 1.19 becomes,
S y + Ry = 0

(1.20)

Assuming a solution as y = CY et , then the problem becomes an eigenvalue

problem, and the equation 1.20 can be turned into,
[R S] Y = 0,
[A I] Y = 0,
or


R

(1.21)


1
S I Y =0

The eigenvalues can now be determined from the equations 1.21, since
it is known that A = S 1 R. With the use of a mathematical program, the
eigenvalues can be extracted directly from A. In rotordynamics the most
interesting is to plot the imaginary part of the eigenvalues to get the Campbell diagram, see Figure 1.6. Since, the eigenvalues are known the Y can be
easily determined.
yh (t) =

n
X

Ci Yi ei t

(1.22)

i=1

The Ci can be solved for known initial conditions with a mathematical

program. If E is the matrix with the eigenvectors,
E = [Y1 , Y2 , ..., Yn ]

(1.23)

then the constants can be determined from the following equation:

C = E 1 yh (0)

(1.24)

Having the solved constants from 1.24, in equation 1.23 the motion of a
unforced system can be analysed.

13

n+1

=
n+2

W1

n+3

W2

Forward

n+4
n+5

W1

n+6

Backward

W2

W3

Figure 1.6: Campbell Diagram

1.3.2

Campbell diagram & speeds

Since the main topic deals with rotor dynamics, it is important to mention
about the Campbell diagram. As shown in Figure 1.6, once the eigenvalues
are solved, then the imaginary part can be plotted in order to see how the
gyroscopic effect acts on the rotor. Moreover, Campbell diagram shows the
resonances of a system, either these are operating forward or backward.
To begin with, Campbell diagram includes, two straight lines which
are the driving frequencies for forward and backward direction, and several lines n+1 , ...., n+6 whose intersection with the driving frequency(dash
lines) shows the resonance. The curves n+1 , n+2 & n+3 are positive which
indicates that the whirl has the same direction as the angular velocity of the
rotor, and is called Forward precession [9]. The curves n+4 , n+5 & n+6
are negative, which indicates that the whirl has opposite direction related
to the rotors, and is called Backward precession [9]. At the right of the
same figure are the mode shapes (Wi ) of the shaft for backward and forward
frequency.
In Figure 1.6, is the Campbell diagram of a Jeffcott rotor. The disc is
placed at the midspan of the shaft and as it can be seen from the diagram
there is no gyroscopic effect at the first resonance speed. The eigenfrequencies n+3 and n+4 are just straight lines with the same value for forward
and backward precession.

1.3.3

Particular solution

The general eigenvalue problem gives solution for systems without external
force. This force can be a small unbalanced mass that develops centrifugal
forces. Those centrifugal forces increase the systems amplitude, especially
when a rotor operates close to resonances. Note that, if one is aiming to
14

exceed the critical speed(which in fact is the most crucial), the rotor should
accelerate fast through this critical area. According to [10] for particular
solution the 2nd law of Newton becomes,
Mx
+ Dx + Kx = fs sin(t) + fc cos(t)

(1.25)

where, the two vectors fs and fc are forces. According to the theory [10], a
harmonic input on a linear system can only result in a harmonic output of
equal frequency, and therefore it can be assumed that,
x = a sin(t) + b cos(t) = as + bc

(1.26)

where, a and b are real vectors. Thus, the equation of motion becomes,
[K 2 M ][as + bc] + D[ac bs] = fs s + fc c

(1.27)

By separating the equation 1.27 in s and c term, it becomes two equations,



2
K 2 M  a Db = fs
(1.28)
K M b Da = fc
= [K 2 M ], so the
To simplify the equations 1.28, it is assumed that K
equation bocomes,
h i
a Db = fs
K
h i
(1.29)
b Da = fc
K
Solving the equations 1.29 by the terms a and b, the problem is solved.
i1 h
i
+ 2 DK
1 D
s
K
fc DKf
1 fs + K
1 Db
a = K
b =

(1.30)

Note that the is the driving frequency of the system. In addition, this
a method that assumes that the system includes damping that is not equal
to zeros, D 6= 0. The amplitude can be found from the resultant of a and b,
and if R is the amplitude then,
p
R = a2 + b2
(1.31)
If the R is plotted, then the resonance frequencies can be found for actual
forcing.

15

Chapter 2

Rotor Kit-Bently Nevada

2.1

Introduction

Bently Nevada Rotor Kit RK-4, is a rotating machine that can be adapted in
different configurations and is suitable for educational purposes. The Rotor
Kit comes along with different accessories that can provide different options
and a wide number of experimental tasks.

Rotor speed
Shaft bow
Rotor stiffness
Amount and angle of unbalance
Shaft rub or hitting condition
Rotor-bearing relationships

The user is able to perform many different experiments and compare the
final results with a theoretical model. The advantage of this experimental
equipment, is that one can verify that theoretical calculations give realistic
values, or values that are close to the real ones. Moreover, this is one way
to verify that assumptions in a theoretical model are valid.

2.2

Rotor Kit

The rotor kit is shown in Figure 2.1. In the mounting blocks probes are
used to measure the displacement of the rotor at different positions of the
shaft. All the outputs will be displayed by a simple program in LabView or
an oscillator. It is important that the rotor is set up on a rigid base to avoid
disturbances that can influence the measurements.
Figure 2.1 shows one of the numerous setups that can be achieved with
this equipment. Two disc are available that can be placed in any position on
the rotor. Further more, constant forces and unbalanced masses can easily

16

Figure 2.1: Rotor Kit

added to develop a more complicate system. Two interesting options that
can be applied are, Oil Whirl/Whip option and Perturbator option, Figure
2.2.

a.

b.

y
x

Figure 2.2: Oil Whirl/Whip and Perturbator option

2.2.1

Oil whirl & whip

The oil whirl/whip option, Figure 2.2.a, can be applied by using the special
equipment Oil Whirl/Whip kit. The Oil kit develops a special environment
for the rotor by operating in an oil reservoir. Due to the transparent material
it is possible to observe the oilfilm during the experiments. For this option
a preload frame can be added which can remove the effect of gravity and
position the journal at a desire eccentricity. With this equipment both oil
whirl and oil whip can be studied. Whirl can be developed or prevented by
applying preload in any direction. The preload can be regulated by changing
the oil flow or pressure(for further information look [11]).

17

2.2.2

Perturbator

The perturbator option, Figure 2.2.b, includes a perturbator disc which

accomplishes asynchronous rotation of an unbalance in both forward and
backward direction. A second motor rotates the perturbator disc in different speeds independently of the rotor speed. In addition, a weight can be
attached on the perturbator disc, in order to develop a constant force that
is pulling the rotor towards one direction.
In appendix D can be found some experimental suggestions for all of three options above.

2.3

Probes & calibration

Figure 2.3: Proximity probes-Stadard Mount

There are eight inductive probes included in the rotor kit which are of
high accuracy and work with DC voltage. The eight probes that constitute
the system are: 2 for measuring the speed, 2 for getting the pulses and 4
for measuring the displacement of the rotor at different positions along the
shaft. The probes which are measuring the displacement can be aligned in
x and y direction, or even in an angle of 45o . A probe is consisted from the
parts shown in Figure 2.3.
The part No.1 is the probe tip that collects all the information directly
from the shaft surface (More information about the probes in Appendix
B). The probe reacts to a displacement with a voltage. Therefore, this
voltage has to be translated in (mm). Thus, according to Table 2.1 which
describes the correlation between voltage and displacement, each output can
be transformed directly in (mm).
Table 2.1 contains the values that describe the correlation between the
voltage and displacement. In fact, more detailed measurements should be
done at the laboratory for shorter magnitudes, in order to get the smallest
possible error. Interesting to note is that the probes behaved linearly for
amplitudes between 0,251,78mm. Thus, the approximation of the displacement could be done by considering a linear behavior of the probes. A polyfit
18

Table 2.1: Calibration Table.

The ordering 1,2,3 & 4
Distance(mm)
a\a
0,00
0,25
0,51
0,76
1,02
1,27
1,52
1,78
2,00
2,25
2,50

Each probe was calibrated against the shaft and the bearing.
is based on the input canals of the monitor for getting the measurements.
Probe-Shaft
Probe-Bearing
No.1
N0.2
No.3
No.4
No.1
No.2
No.3
No.4
0,00
0,00
0,00
0,00
0,00
0,00
0,00
0,00
2,90
2,25
1,69
1,22
2,96
2,25
1,69
1,77
4,97
4,14
3,69
3,24
4,88
4,14
3,69
3,80
7,09
6,23
5,68
5,14
6,87
6,23
5,68
5,83
9,26
8,39
7,85
7,19
8,93
8,39
7,85
8,01
11,30 10,41
9,90
9,18
10,83 10,41
9,90
10,08
13,14 12,31 11,86 11,10 12,60 12,31 11,86 12,04
14,30 14,05 13,70 12,58 14,18 14,05 13,70 13,86
14,39 14,35 14,33 14,20 14,35 14,35 14,33 14,34
14,45 14,43 14,41 14,37 14,42 14,43 14,41 14,42
14,48 14,47 14,46 14,43 14,46 14,47 14,46 14,47

line can be used to describe the displacements from the measured voltage of
each probe. Polyfitline can be easily calculate by using Matlab. Matlab finds
the coefficients of a polynomial (p(x) = p1 xn + p2 xn1 + ... + pn1 x + pn ) of
n degree that fits the data from p(x(i)) to y(i). This method approaches the
least error or the line that fits best through the measuring points. Furthermore, it has to be mentioned about mils or mil, which is a length unit equal
to 0.001 inch(0.0254mm) or milli-inch. Mils are used primarily in the U.S
to express small distances and tolerances. One mil is equal to 25.4 microns.
In addition, the probes were calibrated against different materials to observe how it affects the accuracy and compared with the rotors(material).
Note, that Rotor Kit RK4 comes along with certification (Final Test Results) of each probe. This certification shows the correlation between the
voltage and displacement, and the potential error given by the same probe
at different gaps for a polyfit line. In Table 2.1 lists the values that are interpolated in order to get a characteristic curve from which the program will
calculate the right signals. By generating a polyfit line through the points
0,251,78mm for each probe, then the error will be according to Figure 2.4.
Notification: Figure 2.4 shows the error for a polyfit line. The reason for this
study is to make a comparison between the experimental and factorial error. In fact
the program that treats the raw data, is relying on a calibration table, Table 2.1, that
develops interpolation between those points, from 0,252,50mm. Thus, the error is
even less than the one shown in Figure 2.4.

In this figure the error is shown which affects the accuracy of the measurements. From Figure 2.4 can be observed the calibration curve from
Bently Nevada does not completely match to the Lab error.
To begin with, Rotor Kit RK4 has been manufactured in U.S and one potential reason is that the calibration was done with different voltage(110120
Volts at 60 Hertz). Thus, it is unknown whether or not, the very same calibration equipment should give the same error for the probes in a higher
voltage, similar to European(220230 volts at 50 Hertz). Moreover, the

19

ProbeNo.08102A3A

Error(%)
10

LabC.C

0
10
10

BentlyNevadaC.C
20

30

40

50

60

70 Gap(mils)

ProbeNo.08102A3L

Error(%)
10

LabC.C

0
10
10

BentlyNevadaC.C
20

30

40

50

60

70 Gap(mils)

ProbeNo.08F00W7W

Error(%)
10

LabC.C

0
10
10

BentlyNevadaC.C
20

30

40

50

60

70 Gap(mils)

ProbeNo.08F01ZXL

Error(%)
10

LabC.C

0
10
10

BentlyNevadaC.C
20

30

40

50

60

70 Gap(mils)

Figure 2.4: Final Test Results.Comparison of the error taken directly at the
lab with Bently Nevada results.

20

calibration process was accomplished with different equipment than the one
used by Bently Nevada. The calibrator used for this purpose is described in
Appendix C. At the lab was marked that the probes gave a characteristic
line that was parallel to Bently Nevadas, but slightly higher along the voltage axis. One additional reason can be the environmental temperature and
the wiring from the output to voltmeter. Finally, one reason can be that
the calibration was done against another material.

21

Chapter 3

Experimental & theoretical

results
The present chapter compares experimental and theoretical results. The
purpose is to scale down the problem before will be adapted in real dimensions. The experiments which will be explained in this section, were
accomplished with Rotor Kit RK4 Bently Nevada. Furthermore, discussion will take place for each experiment to evaluate theoretical models
which were developed in order to describe the same experimental systems.

3.1

Rotor effect of a disc mass & perturbation

This task shows how a system responses, while a perturbator attached on the
shaft rolling back or forth. In fact, perturbator is an independent part of the
system, whose main purpose is the determination of the systems dynamic
stiffness [12] related only to the rotational velocity. In this case a system
has constant speed and the perturbator is running gradually. Furthermore,
this is one method to identify potential lower modes of the system that were
muted while using synchronous perturbation for unknown reasons.
Last, according to [12] the given information of a nonsynchronous perturbation, in comparison always with a synchronous perturbation, are more
fruitful since, as it was mentioned above, a synchronous perturbation is affected by the very same system as a result of information that can not be
observable.

3.1.1

Jeffcott rotor

This subsection describes the experimental result of a Jeffcott Rotor and

compares with the theoretical model of the same system. Note that, Jeffcott
rotor is called every system with the disc mass placed at the midspan of the
rotating shaft. To begin with, the demonstration was executed with a disc

22

mass at the midspan of the shaft and the perturbator disc very close to its
inboard side. Note that, the inboard side is the side with the motor and
outboard where the shaft ends.
x(mm)

y(mm)

ForwardPerturbation

0.3

0.3

0.2

0.2

0.1

0.1

0.1

0.1

0.2

0.2

0.3

0.3
1000

500

1500

2000

rpm

2500

(1930)

x(mm)

500

0.1

0.1

0.1

0.1

0.2

0.2

0.3

0.3

y(mm)
0.3

1500

Constantspeed:
1000rpm

2000
(1921)

rpm

2500

Perturbatorspeed:
1930rpm

0.3

0.2

0.2

0.1

0.1

0.1

0.1

0.2

0.2
0.2

0.1

0.1

0.2

0.3

1000

500

y(mm)

2000

2500

x(mm)

rpm

BackwardPerturbation

0.3
0.2

1000

1500

(1930)

0.2

500

1000

y(mm)

BackwardPerturbation

0.3

ForwardPerturbation

1500

Constantspeed:
1000rpm

2000
(1921)

rpm

Perturbatorspeed:
1921rpm

0.2

2500

0.1

0.1

0.2

0.3

x(mm)

Figure 3.1: Experimental results.

Amplitudes for forward & backward perturbation.

Orbits for both cases as they were conceived during the experiment.

Parameters of experiment: shaft length L=0.48m, shaft diameter d=0.01m,

density = 7850kg/m3 , Youngs modulus E = 206 109 P a, unbalance mass
mu = 0.0004kg, unbalance distance from the shaft center u=0.03m, disc
mass md = 0, 80630kg, bearing block stiffness [5] kb = 5.75 105 N/m.
According to the theory, Jefftott Rotor should have equal first critical
23

speed for forward and backward rotation. As it can be seen from Figure
3.1, indeed the first critical speed matches for both cases with a very little
divergence, forward 1930rpm and backward 1921rpm. Thus, for both cases
the first critical speed is approximately 203rad/s. The first natural
frequency is very obvious, due to the fact that suddenly a multiperiodic
solution becomes a large orbit of single frequency as shown in the figure.
In this phase the orbits have the highest amplitude, and therefore, rotating
machines should pass rapidly through this area because the consequences
can be severe.
These experiments as can be seen from the graphs, were executed for
speeds 200rpm-3000rpm. However, interesting was the behavior of the shaft
when it was rapidly accelerated above 5000rpm for a very short period.
The shaft behaves as almost a perfect balanced system in velocities above
the first natural frequency, and the orbit gets smaller and smaller until the
next critical speed. Furthermore, even if the shaft changes amplitude, it is
whirling round the same axial line.
a.
(rad/s)

b.

Simplemodel

(rad/s)

Simplemodel&Bearings

Perturbationguide

(210.6)

(215)

(rad/s)

(rad/s)
c.
(rad/s)

Entiremodel&Bearings

(210.5)

(rad/s)

Figure 3.2: Matlab results.a. Simple model of Rotor Kit, b. including bearing system &
c. including the extent part after the bearings
Regarding the theory and the mathematical model developed for this
purpose, in Figure 3.2 are indicated the results of three different mathe-

24

Table 3.1: Jeffcott Rotor results

Experimental results
203(rad/s)
Simple model(Matlab)
215(rad/s)
Simple model & bearings 210.6(rad/s)
Entire shaft & bearings
210.5(rad/s)
matical codes. To begin with, the campbell diagram in figure a, describes a
simple model of Rotor Kit constituted by a shaft and a disc(bearing stiffness
equal to infinity). The final result of this model is 215rad/s. The second
campbell diagram in figure b, shows a model that includes bearing system
with the first natural frequency being at 210.6rad/s. The last diagram in
figure c, gives slightly lower value, 210.5rad/s, for the first natural frequency
and the mathematical model includes the extent parts after both bearings.
Thus, relying on Table 3.1 ,it can be identified that a simple code model
can give good results, but a more detailed can provide more precise information. The approach is very good, even though a small divergence of
approximately 7rad/s still exists. Unfortunately, it was impossible to reach
the second frequency, since the systems parameters where such that it was
above the speed limit of the experimental facility(10000rpm).
Orbits & shapes
An interesting observation during the experiments is the difference in the
orbits shape during different speeds.
y(mm)
0.3

Constantspeed:
1000rpm

Perturbatorspeed:
1990rpm

y(mm)

0.3

0.2

0.2

0.1

0.1

0.1

0.1

0.2

0.2
0.2

0.1

0.1

0.2

0.3

x(mm)

Constantspeed:
1000rpm

Perturbatorspeed:
1990rpm

0.2

0.1

0.1

0.2

0.3

x(mm)

Figure 3.3: Orbits in double perturbation speed.Forward perturbation to the

left & backward perturbation to the right

In Figure 3.3 the orbits for both cases can be seen while the peturbator is
running almost with double speed(constant:1000rpm-perturbation:1990rpm).
It is obvious why that happens. A system needs some time to create an orbit circle, similarly the perturbator disc needs the same time. Nevertheless,
25

in this case the perturbator is running with double speed, and thus, in the
same time is able to achieve two orbit circles.
a.

b.
y

y
Alignedpoint
b
a

Alignedpoint

Fp

Fs

a
Fs

Fp

120

Figure 3.4: Diagram. Analysis of backward & forward perturbation.

To give a better understanding, in Figure 3.4 it can be seen how two
independent forces are moving. In figure a, is the case of forward perturbation. Both forces are moving at the same direction and they will be met
at the aligned point at a side. At that point grows the highest excitation
Fmax = Fs + Fp , where Fs and Fp are the system and perturbation force
respectively. Since the perturbation speed is double, while Fs will be at the
b side of the aligned point, the Fp will be exactly at the opposite side(side a),
thus at this instant on the system acts the minimum force, Fmin = Fs Fp .
It can be identified that the system will have two orbits with different amplitudes as it is shown in Figure 3.3 forward.
If it is analyzed the orbit of Figure 3.3 backward, then from Figure 3.4.b
can be seen that with backward perturbation, Fs and Fp will be aligned six
times for every single rotation of the system, three Fmax and three Fmin .
In Figure 3.4.b the positive aligned points that abstain 120o to each other
are shown. Now, it is obvious why three peaks grow in backward perturbation with double perturbation speed. For more information about orbits in
different speeds check Appendix E.1. and E.2.

3.1.2

Disc mass at 2/3 of the span

The second experimental task is to displace the mass disc at the 2/3 of
the span. The perturbation disc is as before very close to the mass disc
side towards the inboard bearing. All the experimental parameters for this
experiment are the same with the experiment of Jeffcott rotor.
26

x(mm)

y(mm)

ForwardPerturbation

0.3

0.3

0.2

0.2

0.1

0.1

0.1

0.1

0.2

0.2

0.3

0.3
1000

500

1500

2000

rpm

2500

(2127)

x(mm)

500

0.1

0.1

0.1

0.1

0.2

0.2

0.3

0.3

y(mm)
0.3

1500

Constantspeed:
1000rpm

2000
(2120)

rpm

2500

Perturbatorspeed:
2127rpm

2000

2500

1000

500

y(mm)
0.3

0.2

0.2

0.1

0.1

rpm

BackwardPerturbation

0.3
0.2

1000

1500

(2127)

0.2

500

1000

x(mm)

BackwardPerturbation

0.3

ForwardPerturbation

1500

Constantspeed:
1000rpm

2000
(2120)

2500

rpm

Perturbatorspeed:
2120rpm

0.1
0.2

0.2

0.2

0.1

0.1

0.2

0.3

x(mm)

Figure 3.5: Experimental results.

0.2

0.1

0.1

0.2

x(mm)

Amplitudes for forward & backward perturbation at

2/3 of the span. Orbits for both cases as they were conceived during the experiment.

27

From the theory and since the mass is displaced, it is expected to have
two different first critical speeds for back and fort perturbation. From Figure
3.5 it is obvious that both cases give different critical speeds. Further more,
the experimental result shows higher speed for forward perturbation, as
it was expected. Thus, the first critical speed for forward perturbation is
223rad/s and backward 222rad/s. In addition, it can be seen from
the figure how the orbits in both cases become. At this point the orbits get
the highest amplitude, therefore, as it was mentioned in the previous section
all rotating machines should pass as fast as possible through the first critical
area.
These experiments were executed for speeds 200rpm-3000rpm. However,
it can be seen that the orbits are getting smaller above 5000rpm. That
happens due to the fact that the rotor gets stiffer while running in speeds
much higher the first natural frequency. Further more, the shaft always
moving in ellipsoid orbits(whirling) round the same imaginary axis even if
the shaft changes amplitude.
a.
(rad/s)

b.

Simplemodel

(rad/s)

Simplemodel&Bearings

Perturbationguide

(231.5)

(238)
(237)

(231.0)

(rad/s)
c.
(rad/s)

(rad/s)

Entiremodel&Bearings

(231.4)

(230.8)

(rad/s)

Figure 3.6: Matlab results.a. Simple model of Rotor Kit, b. including bearing system &
c. including the extent part after the bearings
Adapting the matlab code for mass disc at 2/3 of the span, the outcome
will be Figure 3.6. As before, in figure a stands a simple model of the

28

Table 3.2: 2/3 of the span results

Forward P. Backward P.
Experimental results
223(rad/s)
222(rad/s)
Simple model(Matlab)
238(rad/s)
237(rad/s)
Simple model & bearings 231.5(rad/s)
231(rad/s)
Entire shaft & bearings
231.4(rad/s) 230.8(rad/s)
very same system without any bearing system, with forward frequency at
238rad/s and backward at 237rad/s. The second model includes bearing
system and obviously lower results of 231.5rad/s for forward and 231.0rad/s
for backward frequency. The last code shows a model that describes the
entire system, including the extent parts after both bearings, with forward
frequency at 231.4rad/s and backward at 230.8rad/s. Thus, the results will
be as shown in Table 3.2.
Having now all the results, Table 3.2, it is understandable that including
more components that affect the entire system, the theoretical results can
be closer to the experimental ones. By including the entire shaft and the
bearing system, it was achieved to reduce the difference from 15rad/s to
7rad/s. As it can be seen from Figure 3.6 the second critical speed is quite
high, therefore, it was impossible to reach it.
Orbits & shapes
Similarly with the previous section, here it can be seen how the orbit becomes
when the perturbation speed reach double value in comparison with the
systems speed.
y(mm)
0.3

Constantspeed:
1000rpm

Perturbatorspeed:
1987rpm

y(mm)

0.3

0.2

0.2

0.1

0.1

Constantspeed:
1000rpm

Perturbatorspeed:
1987rpm

0.1
0.2

0.2

0.2

0.1

0.1

0.2

0.3

x(mm)

0.2

0.1

0.1

0.2

x(mm)

Figure 3.7: Orbits in double perturbation speed.Forward perturbation to the

left & backward perturbation to the right

In Figure 3.7 is shown orbits similar of Figure 3.3, and it is known hence-

29

forth why the orbit change in these modes. Something interesting to be mentioned, is that for speed multiple to the constant(three,four,five....times),
these orbits will get proportionally one more sub-orbit. For instance, the
orbits in the figure will get, one more sub-orbit for perturbation speed
3000rpm,or two more sub-orbits for perturbation speed 4000rpm and so
fort.

3.2

Journal bearing

Journal bearing is an option for observing different phenomena that occurs

while a shaft operates in an oil environment. Oil grows forces on a journal surface in erratic orbits. In the following subsections is shown those
phenomena through different tasks by using Rotor Kit RK4 equipment.

3.2.1

a.

Oil wedge force

b.

ClearanceofBearing

0.2
0.1

Rotationoftheshaft

c.

y
0.2

0.2

0.1

F1

0.1

0.1

0.2

a
F2

0.1

0.1

0.2

0.1

0.2
0.2

0.1

0.2
0.2

0.1

0.1

0.2

0.2

0.1

0.1

0.2

Figure 3.8: Oil wedge force

In Figure 3.8.a is shown the journal in a slow roll speed with only the
gravity acting on it. In addition, the preload frame applies a very small load
on the shaft. The rotor rolls counter clock wise and the oil wedge displaces
the shaft from point a to point b.
In Figure 3.8.b the journal is placed at the center of bearing clearance,
point a. In this case the rotor is in a complete immobility. If now a force
F1 and F2 applied in x and y direction respectively, the journal will be
displaced along the same directions. As indicated in the same figure, the
center of journal is moving for point a to point b under the force F1 , and
back at a after the removal of the force. Similarly in x direction, from point
a to point c.
Thus, it is obvious that the journal is moving almost parallel to any force
acting on it, due to the fact that no fluid wedge force exist.
If now speed up the shaft till 1500 rpm and try to apply a vertical force
F as in Figure 3.8.c the new position of the journal will be different than the
30

expected one. Instead, the journal is moving in a new direction which is the
resultant of force F and fluid wedge forces, point e. Thus, fluid forces can
be easily verified, since the applied force and the new direction is known.
y(mm)
0.2

Journal:
700(rpm)
6(psi)

3.3

y(mm)

Journal:
1340(rpm)
6(psi)

Whirl & whip

0.2

y(mm)
0.2

0.1

0.1

0.1

0.1

0.1

0.1

0.2

0.2

0.2

0.2

y(mm)
0.2

0.1

0.1

0.2

x(mm)

Midspan:
700(rpm)
6(psi)

0.2

y(mm)
0.2

0.1

0.1

0.2

x(mm)

Midspan:
1340(rpm)
6(psi)

0.2

y(mm)
0.2

0.1

0.1

0.1

0.1

0.1

0.1

0.2

0.2

0.2

0.2

0.1

0.1

0.2

x(mm)

0.2

0.1

0.1

0.2

x(mm)

Journal:
3000(rpm)
6(psi)

0.1

0.1

0.1

0.1

Figure 3.9: Whirl and whip orbit. The present figures show how the shaft
orbits change under different angular velocities.
Whirl and whip are two phenomena which are of big importance in rotor
dynamics. The experimental task of Figure 3.9, shows the results of JournalBearing demonstration on Rotor Kit RK4 Bently Nevada. The task was
fulfilled in 6psi oil pressure and angular velocities from 30rpm to 3000rpm.
During this task main purpose was to verify all which are known from the
theory.
In the figure can be seen the journal and midspan orbits. At the first
graph is shown the shaft operating at 700rpm. The journal remains almost
at the center of the bearing, with the midspan moving softly in a small orbit.
The oil force is enough to keep the journal align with the bearing.
Increasing the speed gradually at 1340rpm, the shaft starts whirling.
The midspan main orbit has become wider with quasiperiodic orbits. At
that point it was very easy to get back the initial stability, with only a small
tension in one of the preload springs. Increasing now the angular speed above
31

x(mm)

Midspan:
3000(rpm)
6(psi)

0.2

0.2

0.2

x(mm)

1340rpm(which in fact is where the area of critical speed starts) and much
higher of it(f.e 3000rpm) to make sure that the rotor is whipping, even the
double tension was not able to stabilize the shaft in smooth operation. From
the picture can be seen how the orbit became at 3000rpm. In conclusion,
it is more difficult to make the instability to go away while the rotor is
whipping.
Something that should be done comprehensible, is the difference between
whirl and whip. Whirl can take place sometimes even at low speeds, and its
frequency is normally the half running speed(the oil film speed less than 50%
of the journal surface speed, [12]). If this frequency coincides the natural
frequency, then the oil whip occurs. Something remarkable is that whip
locks the shaft at the same orbit due to the high frequency, which is obvious
from the experiments done with Rotor Kit. In addition, the fluid stiffness
is increasing due to lower circumferential velocity [11].

3.3.1

Oil pressure control

Oil pressure is very important to find out when the oil whirl/whip will take
place. The oil pressure determines the size of concentrated radial forces
which direct the journal.
In this experiment main purpose was to make an instability to go away
by increasing the oil pressure. With rolling speed at 4280rpm and 4.5psi, as
it can be seen from Figure 3.10 the journal and midspan starts whipping.
The rolling speed is enough higher of the first natural frequency to unsure
that the shaft is whipping. In the same figure now can be observed that
the instability which whipping the shaft, went away by increasing the oil
pressure from 4.5psi to 13.5psi.
Higher oil pressure increases the fluid film radial stiffness. Thus, the
journal can be stabilized and directed to the center of the bearing. In fact,
the oil pressure is high enough to overcome the dynamic forces of the journal
and obtain higher circumferential velocity. Circumferential velocity usually
0.420.48running speed, [13], makes a journal almost to whirl.

3.3.2

Journal bearing whirl-whip phenomena

An interesting part of detecting the behavioral progress of a running journal

in an oil bearing, is to observe how the midspan and journal orbits gradually
change and how the journal is being aligned to the bearing due to growing
dynamical forces. In this section will be shown the whip effect of an experiment done with Rotor Kit RK4. This and all the previous tasks were
executed according to arrangement Appendix D.4.
As it can be seen from Figure 3.11, the journal operated in 5.5psi for
velocities from 200rpm to 2200rpm. From both graphs is obvious that the
journal is not perfectly aligned to the bearing(it is always difficult to direct

32

y(mm)
0.2

Journal:
4280(rpm)
4.5(psi)

y(mm)
0.2

0.1

0.1

0.1

0.1

0.2

0.2
0.2

y(mm)
0.2

0.1

0.1

0.2

x(mm)

Midspan:
4280(rpm)
4.5(psi)

0.2

y(mm)
0.2

0.1

0.1

0.1

0.1

0.2

0.2

0.2

0.1

0.1

0.2

x(mm)

Journal:
4280(rpm)
13.5(psi)

0.1

0.1

0.2

Midspan:
4280(rpm)
13.5(psi)

0.2

x(mm)

0.1

0.1

0.2

x(mm)

Figure 3.10: Behavior of the shaft by changing oil pressure. Shaft whipping
in 4.5psi, left. Shaft align with the bearing in 13.5psi, right.

x(mm)

y(mm)

Bearing
center
Bearing
clearance

rpm

rpm

rpm

Figure 3.11: Amplitude graph for x and y direction. The task was demonstrated
in 5.5psi from 200rpm to 2200rpm.

33

a rolling journal at the right position.), thus, the journal is running in an

eccentric position inside the bearing.
To begin with, each graph can be divided in three areas. The first area
a, as it can be seen from the graph, is the area where the journal is being
aligned gradually as the angular velocity is increasing. The journal increases
the oil speed as well as the oil stiffness [11], as a result of changing position
in a new equilibrium area inside the bearing.
The second area b, is the area of oil whirl. The journal stars whirling
and rapidly is placed in higher orbit similar to the bearings clearance. In
this area any instability can be easily disappeared with the application of
only a small force, since the dynamic forces of the journal are still weak.
The last area c, is where the journal starts whipping. In fact, whip will
lock whirls frequency and maintain it while operating into higher and higher
speeds [13]. According to [11] the shaft is in a very high position from the
bearing center, therefore, a whipping shaft demands more force to stabilize
its orbit.
As can be seen from the figure, the axial line is the imaginary bearing
center, and the dash lines the bearing clearance. It is obvious that the
journal will never exceed to have contact with the bearing surface due to oil
wedge. It should be perfect lubrication, in order to avoid contact between
the journal and the bearing.
Similarly to Figure 3.11, in Figure 3.12 one can see how the midspan of
the shaft is moving in relation with the journal. According the figure, it is
seen the different positions and orbits the shaft gets during the experiment.
In 200rpm the shaft is almost aligned with the journal obviously in an eccentric position lower from the imaginary center. In 900rpm, with the midspan
almost at the same position, the journal is directed towards the imaginary
center in a smaller orbit. It is remarkable this behavior due to the higher
oil circumferential velocity. In fact, if it can be achieved to have high circumferential velocity, vibration problems can be avoided according to [14].
That can be carried out by preturbating the oil flow inside the bearing.
As the circumferential velocity is increasing the journal orbit is getting
smaller and smaller. That is obvious from the figure with the journal running
in 1460rpm. At the next phase the journal starts whirling in 1590rpm.
According to [13], any instability that exceeds orbit amplitude more than
50% of the clearance, is considered to be excessive.
The journal now is operating at its first natural frequency, 1620rpm.
The journal is whirling in an orbit whose amplitude is growing towards the
clearance limits, and the midspan running in erratic orbits. Last, running
the shaft in a higher velocity, 2200rpm, the journal is henceforth whipping
with orbit amplitude as high as possible.

34

y(mm) Journal:
0.2

200(rpm)
5.5(psi)

y(mm)
z

0.2

y(mm)

Journal:
900(rpm)
5.5(psi)

0.2

0.1

0.1

0.1

0.1

0.1

0.1

0.2

0.2
0.2

0.1

0.1

y(mm)

Midspan:
0.2 200(rpm)
5.5(psi)

x(mm)
y(mm)

0.2

0.1

0.1

Midspan:
0.2 900(rpm)
5.5(psi)
z

x(mm)
y(mm)

0.2
z

0.1

0.1

0.1

0.1

0.1

0.2

0.2

0.2

0.1

0.1

y(mm) Journal:

x(mm)

1590(rpm)
0.2 5.5(psi)

0.2

0.1

0.1

y(mm)
z

Journal:
1620(rpm)
0.2 5.5(psi)

x(mm)

0.1

0.1

0.1

0.1

0.2

y(mm)
0.2

0.1

0.1

x(mm)
y(mm)

Midspan:
1590(rpm)
5.5(psi)

0.2

0.1

0.1

x(mm)
y(mm)

Midspan:
1620(rpm)
5.5(psi)

0.2

0.1

0.1

0.1

0.1

0.1

0.2
0.2

0.1

0.1

x(mm)

0.1

Midspan:
1460(rpm)
5.5(psi)

0.2

0.1

x(mm)

0.1

x(mm)

0.2
0.2

0.1

0.2

0.1

Journal:
2200(rpm)
0.2 5.5(psi)

0.1

0.2

0.2

y(mm)

0.1

0.2

0.2
0

0.1

0.2

Journal:
1460(rpm)
5.5(psi)

0.2

0.1

0.1

Midspan:
2200(rpm)
5.5(psi)

x(mm)

0.2
0.2

0.1

0.1

x(mm)

0.2

0.1

0.1

x(mm)

Figure 3.12: Orbits of the journal and midspan.Behavior of the shaft for different speeds running counter clockwise.

35

Conclusion
The present thesis reports all the experimental tasks done in order to compare and verify the analytical solution of a system. All the experimental
tasks were executed by using Rotor Kit RK4. The interesting part is, why
theoretical results do not match with the experimental.
First of all, in this thesis perturbation was of main importance and that
is why it was probed deep into it. From the experimental results it is seen
that all three theoretical model give higher first critical speeds. It is believed
that small errors as it can be seen from Table 3.1 and 3.2, can occur. For
Jeffcott Rotor the error is: 5,58% for a simple model without bearings,
3,61% for a model that includes bearing stiffness and 3,56% if it will be
taken into account the extent parts out of the bearings. Similarly for Disc
mass at 2/3 of the span the error is: for a simple model 6,30% forward and
6,33% backward, including the bearing stiffness 3,67% forward and 3,90%
backward, and including the extent rotor length out the bearings 3,63%
forward and 3,81% backward. It is natural to have some difference between
the theoretical and experimental results, since the theoretical method does
not take in a consideration surrounding factors that can affect the final
results. Besides, including only the bearing properties it can be realized that
the error is very little. It is still unknown whether or not the friction at the
attach points of the rotor(f.i the place where the disc is attached on the span)
could change the physical properties of the rotor at that point, for instance,
different stiffness of the rotor and damping due to the friction between the
rotor and the disc. Another potential factor for lower experimental first
natural frequency, the transportation of vibrations from the motor via the
coupling.
Many of the reasons which have already been mentioned are: mistakes
during the calibration. For instance, the calibration and the values on which
the program relies on, done under different conditions like, calibration environment and temperature. Further more, one more reason is that the chosen
bearing stiffness was taken according to [5], however, even if the theoretical
model relied on this stiffness it is still not sure whether or not this value
is true. Thus, the bearing stiffness might be higher than the real one, and
therefore the theoretical model gives higher first critical speed.
In addition, the rotors body was considered to be rigid. In fact, all
members of a system get some kind of distortion and obviously if someone
does not include them in the final calculation, the result will be a stiffer
system.
Moreover, it has to be mentioned that Rotor Kit RK4 was adjusted on a
wooden bench whose potential weakness transfered vibrations on the main
system which met the first natural frequency earlier. The wooden bench
add some kind of elasticity on the system, something that was not taken
into account at the theoretical models.
36

Last but not least, it is understandable now that by making a model

that describes exactly the real environment and situation in which a system
operates, the approach can be closer. This is something difficult to make in
reality, since the description of a real system becomes very complicate and
makes it difficult for engineers to compute such a model. Thus, from above
we conclude that the assumptions gave us higher frequency.
In conclusion, the development of different experimental tasks brought
the undergraduate students closer to the topic by raising questions and
points that could be included at the present work or a nice suggestion for
future research.

37

Appendix A

Rotor Kit specifications

ELECTRICAL
Power Input

95 to 125 Vac, 45 to 65 Hz, 3 Amp, single phase, or

190 to 250 Vac, 45 to 65 Hz, 3 Amp, single phase

Proximitor
Power Output
Maximum Speed
Ramp Rate Control

-18.0 Vdc 0.8 Vdc at 140 mA max

10.000 rpm, typical
115.000 rpm/min, typical

MECHANICAL
RPM Range

To 10.000 rpm

DIMENSIONS
Length
Width
Height

780 mm (30.8 in)

340 mm (13.4 in)
165 mm (6.5 in)

WEIGHTS
Rotor Kit

14.5 kg (32 lbs)

Additional Weights
Procimitor Assembly
Motor Speed Control
Oil Whirl Option
Perturbator Option

0.9 kg (2 lbs)
3.5 kg (7.8 lbs)
7.3 kg (16 lbs)
10.0 kg (22lbs)

ENVIRONMENTAL CONDITIONS
Operational Temperatures 25o C 10o C(77o F 18o F )

38

Appendix B

Proximity probe

Figure B.1: Proximity probe-Stadard Mount

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.

Probe tip, 5.26mm(0.207 in)maximum diameter

Hexagonal nut
Case thread
Wrench flats
750 cable, 2.8mm(0.11 in) maximum outside diameter, 7.6mm
(0.30 in) maximum outside diameter of armor
3.23mm (0.127 in)
Uthreaded length A
Miniature male coaxial connector, 7.23mm(0.285 in)maximum
outside diameter D
Case length B
2.92mm (0.115 in) maximum
Total length C, -30%,-0%

These are many specifications for this type of probe. The environment
and other factors that affect the results, are of main importance, and for this
reason someone should consult the manual that provides deeper description
of Transducer Systems. For those who want to have some further information can visit the following link, http://www.gepower.com/prod_serv/
products/oc/en/downloads/147386.pdf.
39

Appendix C
High magnification extensometer calibrator

Instron Extensometer Calibrator.

Catalogue Numbers:
2602-001
(G-55-1)
2602-004
(G-55-1M)
This instrument is a calibrator of high ration of output to displacement.
Accurate and rugged units, this calibrator is adaptable to most types of
extensometers, and has a range of 1-inch(25,4mm-metric) units.

Figure C.1: High Magnification Calibration Scales (Inch System).

The High Magnification Extensometer Calibrator, Figure C.1, consists
of a large micrometer head mounted on a stand. It has two thumbscrew
40

couplings for mounting various size and shape adapter spindles supplied
with the unit to accommodate most extensometers. The dial on the metric
micrometer head indicates directly to 0,002mm and by a four line vernier to
0,0005mm. Its accuracy is 0,00038mm at any setting. Figure C.1 shows
a vernier based in inch measuring system (The vernier that was used to
calibrate the probes was in metric system).

Figure C.2: Calibrator.

In Figure C.2 can be seen the parts that constitute the Calibrator. This
instrument was used to unsure the probe accuracy. The probes were adjusted
on the mobile part of the calibrator and the shaft was stabilized on the
opposite side.

41

Appendix D

Experimental suggestions
The present section introduces some experimental suggestions from Bently
Nevada. Of course there are numerous combinations that someone can accomplish on this equipment in order to achieve a more advanced model. The
experimental models should be carried out with all the necessary instructions for a proper environment. Something very crucial: be sure that
the foundation on which Rotor Kit RK4 is accommodated, is rigid.

D.1

Disc & unbalance mass

a.

b.

c.

d.

Threadholesforunbalance
d

Figure D.1: Disc options:

Rotor with, a) two discs, b) one disc at midspan (Jeffcott rotor),

c) three discs(two regular and one perturbator disc) & d) one disc at any position.

42

In Figure D.1 are indicated many disc configurations. As it can be seen

in Figure D.1.a, one interesting arrangement is with two discs on the shaft.
The discs can be placed in different positions along the shaft. The distances
and can be equal or not. Bently Nevada recommends , = 25% in
relation with the effective shaft length. In Figure D.1.b the disc is placed
at the midspan, something interesting in case someone aims to work and
detect phenomena of Jeffcott rotor.
In Figure D.1.c there are three disc which according to Bently Nevada,
a good suggestion is to use distance , = 25% and = 50% in relation
with the effective shaft length, or 12% and 50% respectively. The second
configuration is more preferable in case the bearings are not very stiff. In
the present and all forth coming options an unbalance mass can be placed in
order to develop higher excitations in x and y direction. The unbalance mass
can be easily placed/threaded on the discs, since each disc brings peripheral
holes with thread as it is shown in Figure D.1.c.
The last configuration Figure D.1.d, a disc is placed at any position for
further treatment. The distance can be chosen from the executor of the
task.

D.2

Preload condition

a.

b.
Preloadsprings

Nylonrod

Contactsurface
Preloadsprings

Figure D.2: a)Preload frame, b)Preload rod

To develop preload condition can be either applied with a preload frame,
Figure D.2.a, or a nylon rod manually, Figure D.2.b. The second is obvious
from the figure that is used manually to force the shaft during demonstrating
a task. The preload frame is usually used with the oil whirl/whip option.
As can be seen from the figure the preload frame has four springs that can
apply controlled forces in different directions. The application of a preload
43

force is interesting task in showing how the shaft position and the orbit
shape changes.

D.3

Rub condition

This task can be carried out by using a screw made out of copper which
can be also called rub screw. The main purpose is to develop rub(friction)
by adjusting the rub screw at the Probe Mount until it has surface contact
with the shaft.

D.4

Oil whirl/whip

Figure D.3: Journal-Bearing option

10.5

25

Oil whirl/whip option is task that can be achieved by a special shaft

which is accommodated
in a plastic bearing. The bearing is supplied con399
300 oil. Running a task by using the setting up of Figure D.4,
tinuously with
the executor can observe different phenomena. Oil whirl which is a self excited vibration because of the oil forces that are growing within bearing. Oil
whip which takes place while the oil whirl approaches the first critical speed.
The whip will be maintained at the first25.6natural frequency while the shafts
rotational speed increases. In addition, this option gives the chance to accomplish observations like, oil distribution, oil wedge and so forth. Further
25.40
information about
330 oil whirl/whip look [11].
357

D.5

Perturbator

25

399 seen from Figure D.1.c, is a task that can execute

Perturbator as can be
Preloadframe
300
nonsynchronous rotation in both directions. In the figure the Perturbator
Preloadframe
(1)
disc(middle disc)
is rotating
independently by another source.
This option
k
can be arranged in different configurations and combined similarly with all
(4) observes when
(2)
the other option mentioned above. This is a task that
the
system becomes unstable(backward and25.6
forward frequencies) with the main
k
system running
at a stable velocity.
Inboardbearing
10.5

frame

bear

(3)

331.5
357

44

Discsplacednexttoeachother
Nogapbetween

25.40

Appendix E

Perturbation orbits
The following orbits were captured during the experimental task with perturbator disc running in asynchronous rotational speed. The experiments
were accomplished for two configuration, one at the midspan(Jeffcott Rotor)
and one at the 2/3 of the span. The experiments were executed for rotational
speeds: constant speed 1000rpm and perturbator speed 200rpm3000rpm.

45

E.1JeffcottRotorforward
Constantspeed:

Perturbatorspeed:
1000rpm

y(mm) 1000rpm
0.3

Constantspeed:

0.3

0.2

0.2

0.1

0.1

0.1

0.1

0.2

0.2
0.2

0.1

Constantspeed:

0.2

0.3

x(mm)

Perturbatorspeed:
1490rpm

y(mm) 1000rpm
0.3

0.1

0.3

0.1

0.1

0.1

0.2

0.2

Constantspeed:

0.2

0.3

x(mm)

Perturbatorspeed:
1790rpm

y(mm) 1000rpm
0.3

0.1

0.3

0.1

0.1

0.1

0.2

0.2

Constantspeed:

0.2

0.3

x(mm)

Perturbatorspeed:
1990rpm

y(mm) 1000rpm
0.3

0.1

0.2

0.1

0.1

0.1

0.1

0.2

0.2
0.1

0.1

0.2

0.3

0.1

x(mm)

46

0.1

0.3

0.1

0.2

0.1

x(mm)

0.3

x(mm)

Perturbatorspeed:
2990rpm

0.2

0.2

0.1

x(mm)

Perturbatorspeed:
1930rpm

Constantspeed:

0.3

0.3

y(mm) 1000rpm

0.2

0.2

0.2

0.2

0.1

Perturbatorspeed:
1660rpm

Constantspeed:

0.1

y(mm) 1000rpm
0.2

0.1

0.2

0.2

0.2

0.1

Constantspeed:

0.1

y(mm) 1000rpm
0.2

0.1

0.2

0.2

0.2

Perturbatorspeed:
1300rpm

y(mm) 1000rpm

0.1

0.2

0.3

x(mm)

E.2JeffcottRotorbackward
Constantspeed:

Perturbatorspeed:
1000rpm

y(mm) 1000rpm
0.3

Constantspeed:

0.3

0.2

0.2

0.1

0.1

0.1

0.1

0.2

0.2
0.2

0.1

Constantspeed:

0.2

0.3

x(mm)

Perturbatorspeed:
1742rpm

y(mm) 1000rpm
0.3

0.1

0.3

0.1

0.1

0.1

0.1

0.2

0.2

Constantspeed:

0.2

0.3

x(mm)

Perturbatorspeed:
1990rpm

y(mm) 1000rpm
0.3

0.1

0.3

0.1

0.1

0.1

0.1

0.2

0.2

Constantspeed:

0.2

0.3

x(mm)

Perturbatorspeed:
2485rpm

y(mm) 1000rpm
0.3

0.1

0.2

0.1

0.1

0.1

0.1

0.2

0.2
0

0.1

0.2

0.3

x(mm)

47

x(mm)

0.1

0.1

0.2

0.3

x(mm)

Perturbatorspeed:
2318rpm

0.1

0.1

0.2

0.3

x(mm)

Perturbatorspeed:
2980rpm

0.2

0.3

Constantspeed:

0.3

0.2

Perturbatorspeed:
1921rpm

y(mm) 1000rpm

0.1

0.1

0.2

0.2

0.2

Constantspeed:

0.2

0.1

y(mm) 1000rpm

0.1

0.2

0.2

0.2

Constantspeed:

0.2

y(mm) 1000rpm

0.1

0.2

0.2

0.2

Perturbatorspeed:
1326rpm

y(mm) 1000rpm

0.1

0.1

0.2

0.3

x(mm)

E.32/3ofthespanforward
Constantspeed:

Perturbatorspeed:
1000rpm

y(mm) 1000rpm
0.3

Constantspeed:

0.3

0.2

0.2

0.1

0.1

0.1

0.1

0.2

0.2
0.2

0.1

Constantspeed:

0.2

0.3

x(mm)

Perturbatorspeed:
1493rpm

y(mm) 1000rpm
0.3

0.1

0.3

0.1

0.1

0.1

0.2

0.2

Constantspeed:

0.2

0.3

x(mm)

Perturbatorspeed:
1987rpm

y(mm) 1000rpm
0.3

0.1

0.3

0.1

0.1

0.1

0.2

0.2

Constantspeed:

0.2

0.3

x(mm)

Perturbatorspeed:
2485rpm

y(mm) 1000rpm
0.3

0.1

0.2

0.1

0.1

0.1

0.1

0.2

0.2
0.1

0.1

0.2

0.3

0.1

x(mm)

48

0.1

0.3

0.1

0.2

0.1

x(mm)

0.3

x(mm)

Perturbatorspeed:
2975rpm

0.2

0.2

0.1

x(mm)

Perturbatorspeed:
2127rpm

Constantspeed:

0.3

0.3

y(mm) 1000rpm

0.2

0.2

0.2

0.2

0.1

Perturbatorspeed:
1656rpm

Constantspeed:

0.1

y(mm) 1000rpm
0.2

0.1

0.2

0.2

0.2

0.1

Constantspeed:

0.1

y(mm) 1000rpm
0.2

0.1

0.2

0.2

0.2

Perturbatorspeed:
1329rpm

y(mm) 1000rpm

0.1

0.2

0.3

x(mm)

E.42/3ofthespanbackward
Constantspeed:

Perturbatorspeed:
1000rpm

y(mm) 1000rpm
0.3

Constantspeed:

0.3

0.2

0.2

0.1

0.1

0.1

0.1

0.2

0.2
0.2

0.1

Constantspeed:

0.2

0.3

x(mm)

Perturbatorspeed:
1590rpm

y(mm) 1000rpm
0.3

0.1

0.3

0.1

0.1

0.1

0.1

0.2

0.2
0.2

0.3

x(mm)

Perturbatorspeed:
2120rpm

Constantspeed:

y(mm) 1000rpm
0.3

0.1

0.3

0.1

0.1

0.1

0.1

0.2

0.2

Constantspeed:

0.2

0.3

x(mm)

Perturbatorspeed:
2477rpm

y(mm) 1000rpm
0.3

0.1

0.2

0.1

0.1

0.1

0.1

0.2

0.2
0

0.1

0.2

0.3

x(mm)

49

x(mm)

0.1

0.1

0.2

0.3

x(mm)

Perturbatorspeed:
2315rpm

0.1

0.1

0.2

0.3

x(mm)

Perturbatorspeed:
2980rpm

0.2

0.3

Constantspeed:

0.3

0.2

Perturbatorspeed:
1987rpm

y(mm) 1000rpm

0.1

0.1

0.2

0.2

0.2

Constantspeed:

0.2

0.1

y(mm) 1000rpm

0.1

0.2

0.2

0.2

Constantspeed:

0.2

y(mm) 1000rpm

0.1

0.2

0.2

0.2

Perturbatorspeed:
1245rpm

y(mm) 1000rpm

0.1

0.1

0.2

0.3

x(mm)

Bibliography
[1] Kari Tammi. Active vibration control of rotor in desktop test environment. PhD thesis, VTT, Technical Research Centre of Finland, April
2003.
[2] Xi Wu Jim Meagher and Chris Lencioni. Response of a warped flecible
rotor with a fluid bearing. International Journal of Rotating Machinery,
2008(147653):9, June 2007.
[3] Tomas Lukas Petr Ferfecki, Jan Ondrouch. Investigation of vibration
of a rotor system supported by absolutely rigid bearings with a shaft
containing a notch. Technical report, Technical university of Ostrava,
2006.
[4] Agnes Muszynska. Oil whip of a rotor supported in a poorly lubricated
bearing. Technical report, Bently Nevada, 1998.
[5] Enayet Halim. Experiment design for simulating different faults in a rotor kit. Technical report, Chemical and Materials Engineering Department University of Alberta,Edmonton, AB, Canada, T6G2G6, JUNE
2005.
[6] M. Paterno P. Forte and E. Rustighi. A magnetorheological fluid
damper for rotor applications. International Journal of Rotating Machinery, 2004:8, January 2003.
[7] Jan-Olov Aidanp
a
a. M7010T dynamics in mechanical systems, fem in
structural dynamics-lucture notes. LTU.
[8] Ronald D.Ziemian William McGuire, Richard H.Gallagher. Matrix
structural analysis. John Wiley & Sons, Inc, 605 Third Avenue, New
York, NY 10158-0012, 2nd edition, 2000.
[9] ANDREW DIMAROGONAS. Vibration for Engineers. A Pearson
Education Company, PRENTICE HALL, Upper Saddle River, New
Jersey 07458, 2nd edition, 1996.
[10] Jan-Olov Aidanp
a
a. M7010T dynamics in mechanical systems, rotor
dynamics-lucture notes. LTU.
50

[11] Bently Nevada Asset Condition Monitoring, 1631 Bently Parkway

South Minden, Nevada USA 89423, USA. Operation Manual Rotor
Kit Oil Whirl/Whip Option Model RK4, 11 2007.
[12] Agnieszka(Agnes) Muszyska. Rotordynamics. Taylor & Francis Group,
6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 334872742, 1st edition, 2005.
[13] Cornelius Scheffer and Paresh Girdhar. Machinery Vibration Analysis
& Predictive Maintenance. Elsevier Linacre House, Jordan Hill, Oxford
OX2 8DP 200 Wheeler Road, Burlington, MA 01803, JUL-2004.
[14] John J. McKetta. Encyclopedia of Chemical Processing and Desgn.
MARCEL DEKKER, INC. 270 Madison Avenue, New York, New York,
10016, 1996.

51