Mechanical Systems and Signal Processing 30 (2012) 306–322

Contents lists available at SciVerse ScienceDirect

Mechanical Systems and Signal Processing

journal homepage: www.elsevier.com/locate/ymssp

Nonlinear effects caused by coupling misalignment in rotors

equipped with journal bearings
Paolo Pennacchi n, Andrea Vania, Steven Chatterton
Department of Mechanical Engineering, Politecnico di Milano, Via La Masa, 1, I-20156 Milano, Italy

a r t i c l e in f o abstract

Article history: Misalignment is one of the most common sources of trouble of rotating machinery
Received 10 March 2011 when rigid couplings connect the shafts. Ideal alignment of the shafts is difficult to be
Received in revised form obtained and rotors may present angular and/or parallel misalignment (defined also as
23 November 2011
radial misalignment or offset). During a complete shaft revolution, a periodical change
Accepted 24 November 2011
of the bearings load occurs in hyperstatic shaft-lines, if coupling misalignment between
Available online 16 December 2011
the shafts is excessive. If the rotating machine is equipped with fluid-film journal
Keywords: bearings, the change of the loads on the bearing causes also the variation of their
Coupling misalignment instantaneous dynamic characteristics, i.e. damping and stiffness, and the complete
Hyperstatic rotors
system cannot be considered any longer as linear.
Journal bearings
Despite misalignment is often observed in the practice, there are relatively few
Rotordynamics
Nonlinear effects studies about this phenomenon in literature and their results are sometimes conflicting.
The authors aim at modeling accurately this phenomenon, for the first time in this
paper, and giving pertinent diagnostic information. The proposed method is suitable for
every type of shaft-line supported by journal bearings. A finite element model is used
for the hyperstatic shaft-line, while bearing characteristics are calculated by integrating
Reynolds equation as a function of the instantaneous load acting on the bearings,
caused also by the coupling misalignment. The results obtained by applying the
proposed method are shown by means of the simulation, in the time domain, of the
dynamical response of a hyperstatic shaft-line. Nonlinear effects are highlighted and the
spectral components of the system response are analyzed, in order to give diagnostic
information about the signature of this type of fault.
& 2011 Elsevier Ltd. All rights reserved.

1. Introduction

In her comprehensive book about Rotordynamics, Muszyńska [1] observed that rotor misalignment could be considered
as the second most common malfunction after unbalance. She remarked also that this interesting topic was not object of
much attention by researchers.
The authors of this paper share her position and want to contribute by presenting a paper aimed at explaining the
reason of the presence of superharmonic components, i.e. of nonlinear behavior, in rotor vibration spectrum as a
consequence of rigid coupling misalignment, owing to wrong assembly or imperfect flange machining, of an hyperstatic
shaft-line equipped with journal bearings. The method is suitable to be applied to every type of shaft-line.

n
Corresponding author. Tel.: þ 39 02 2399 8440; fax: þ39 02 2399 8492.
E-mail address: [email protected] (P. Pennacchi).

0888-3270/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ymssp.2011.11.020
 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322 307

Nomenclature ru radius of the unbalance mass

U entraining velocity of the oil
Symbols
x vector of generalized displacements of the
system (rotor þfoundation)
[C] damping matrix of the complete system xðf Þ vector of generalized displacements of the
(rotor þ bearings þ foundation) foundation
½Cðf Þ  damping matrix of the foundation xðrÞ vector of generalized displacements of all the
½CðrÞ  internal damping matrix of the rotor rotor nodes
½CðrrÞ , ½Cðrf Þ , ½Cðf rÞ , ½Cðf f Þ  damping coupling matrices xcðrÞ vector of the generalized displacement of the
that are due to oil-film forces constrained nodes
cðbÞ
rsi damping coefficient of ith bearing between r xfðrÞ vector of the generalized displacement of the
and s direction free nodes
e Euler’s number xjðrÞ vector of generalized displacements of jth
Fðbf
bi
Þ
dynamic forces of the oil-film of the ith bearing rotor node
on the foundation xðrÞ vertical displacement of jth rotor node
j
FðbrÞ
b
dynamic forces of the oil-film of the ith bearing
i
on the journal yjðrÞ horizontal displacement of jth rotor node
FC equivalent force caused by the coupling W weight force
misalignment W relative velocity
Fu unbalance force Da magnitude of the angular misalignment
½GðrÞ  gyroscopic matrix Dr magnitude of the radial misalignment
g gravitational acceleration Dx generalized displacements imposed by
h oil-film thickness in the bearing coupling misalignment on the shaft
Ic set of the indexes of the rotor nodes correspond- DxjC generalized displacements imposed by
ing to a journal bearing coupling misalignment on the d.o.f.s corre-
i index sponding to the jCth coupling node
i imaginary unit z coordinate along the axial direction of the
j index journal bearing
jC index of the coupling node Z oil viscosity
[K] stiffness matrix of the complete system y angular position of the rotor
(rotor þ bearings þ foundation) WðrÞ
xj rotation about horizontal direction of jth
½Kðf Þ  stiffness matrix of the foundation rotor node
½KðrÞ  stiffness matrix of the rotor WðrÞ
yj rotation about vertical direction of jth
½KðrÞ
cc  part of the rotor stiffness matrix correspond- rotor node
ing to the d.o.f.s of the constrained nodes in x coordinate along the sliding direction of the
the rows and in the columns journal bearing
½KðrÞ
cf
 part of the rotor stiffness matrix correspond- ja phase of the angular misalignment with
ing to the d.o.f.s of the constrained nodes in respect to rotor phase reference
the rows and of the free nodes in the columns jr phase of the angular misalignment with
½KðrÞ
fc
 part of the rotor stiffness matrix correspond- respect to rotor phase reference
ing to the d.o.f.s of the free nodes in the rows ju phase of the unbalance with respect to rotor
and of the constrained nodes in the columns phase reference
½KðrÞ
ff
 part of the rotor stiffness matrix correspond- O rotational speed
ing to the d.o.f.s of the free nodes in the rows
and the columns Subscripts
½KðrrÞ , ½Kðrf Þ , ½Kðf rÞ , ½Kðf f Þ  stiffness coupling matrices
that are due to oil-film forces C coupling
ðbÞ
krsi stiffness coefficient of ith bearing between r c constrained
and s direction f free
½M mass matrix of the complete system (rotorþ i index
foundation) j index
½Mðf Þ  mass matrix of the foundation r radial misalignment
½MðrÞ  mass matrix of the rotor a angular misalignment
mu unbalance mass
nb number of the journal bearings Superscripts
nC number of rotor couplings
nr number of rotor nodes (b) bearing
p oil pressure in the journal bearing (f) foundation
R vector of the static reactions on the bearings (r) rotor
308 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322

However, before dealing with this topic, it is necessary to observe, as it was also done by Lees [2], that the term
‘‘misalignment’’ is used to indicate several situations corresponding to different physical processes. Actually, various kinds
of rotor misalignment were analyzed in literature.
A first group of papers considered the misalignment of the journal alone with respect to the bearing, without dealing
with the complete dynamics of the shaft-line; some pioneering studies were presented by Boyd [3,4] and another example
was given by Bou-Saı̈d and Nicolas [5]. A similar case was analyzed by Prabhu [6] that highlighted the effects of relative
misalignment, between the journal and the bearing, on film thickness, friction and damping and found out superharmonics
vibration response in the shaft. Bouyer and Fillon [7] observed experimentally that journal misalignment with respect to
the bearing reduces hydrodynamic effects that could not counteract the relative misalignment, between the journal and
the bearing, caused by an external applied torque.
Anyhow a shaft cannot be ‘‘simply’’ misaligned by itself: the causes of this type of malfunction are discussed by Bently
[8], which also mentioned that misalignment produces additional loading on the bearings, but he did not present any
mathematical model on the matter. On the contrary, several other papers considered both theoretical aspects and direct
dynamic simulation of misalignment in Rotordynamics.
A group of papers considered misalignment between rotors joined by flexible couplings. Xu and Marangoni [9,10]
studied the misalignment of a flexible coupling, highlighting the similarities with universal joint operation and the
presence of 2X component in the vibration spectrum. In that study, the shaft-line was supported by ball bearings and
the cause of the 2X component was ascribed to the variable stiffness of the flexible coupling as a consequence of the
misalignment. Often papers in literature report that misalignment causes 2X components, but they neither consider the
type of coupling nor the type of the bearings. Also Sekhar and Prabhu [11] considered flexible coupling, used Gibbons’
theoretical model [12] and performed some simulations in which 2X components were evident. Lee and Lee [13] used a
test rig, equipped with ball bearings and flexible coupling, to verify the results of their theoretical model and observed the
results on the orbit shape. Hu et al. [14] designed a test rig with flexible rotors equipped with three or more journal
bearings suitable to study the misalignment. However, in this case, the focus is on ‘‘lateral misalignment’’ of the supports,
rather than on coupling misalignment. Al-Hussain and Redmond [15] presented a theoretical model of two coupled Jeffcott
rotors supported by rigid bearings. Radial misalignment was simulated and only 1X component resulted in the lateral
vibration steady-state spectra. They observed also excitation of torsional vibrations. Al-Hussain [16] proposed a further
model about angular misalignment affecting the flexible joint connecting two Jeffcott rotors installed on journal bearings.
In this case only stability conditions were analyzed. A combination of theoretical analysis and experimental tests was
presented by Patel and Darpe [17,18], which considered two identical Jeffcott rotors supported by ball bearings and
connected by a pin-bush flexible coupling.
Another group of paper dealt with the diagnosis of misalignment using vibration measurements. Prabhakar et al. [19]
presented a preliminary analysis about the use of continuous wavelet transform of the rotor vibration to detect flexible
coupling misalignment. No experimental evidence was given as a support of the proposed method. Peng et al. [20,21] used
improved wavelet analysis of vibration measurement in order to extract fault features in rotor system, including also
coupling misalignment. Model based methods were also used: Sinha et al. [22] employed model parameter estimation of
combined unbalance and misalignment of a couple of test rigs, performing sensitivity analysis as well. Saavedra and
Ramı́rez [23,24] considered a test rig with a flexible coupling. Pennacchi and Vania [25] used misalignment equivalent
excitation in a model based approach to identify the misalignment of a flexible joint of a gas turbine unit.
The last group of papers considered the effects of rigid coupling misalignment: Bachschmid and Pennacchi [26]
presented the successful identification of angular and radial misalignment in the shaft-line of a test-rig. Lees [2] focused
his interest on rigid couplings, introducing a simple model that considers orientation and different tightening of the
coupling bolts and neglects the presence of journal bearings. Nonlinear system response resulted as a consequence of
coupling between torsional and lateral vibrations. Tsai and Huang [27] used a transfer matrix to model radial
misalignment and found results similar to [15], even if they gave different motivations to the presence of 1X component
only. More recently, Bahaloo et al. [28] studied speed transients of a misaligned rotor with a rather simplified model. On
the contrary, the model presented hereafter by the authors is suitable to study the behavior of real rotating machinery,
considering all the possible features.
It is evident, from the analysis of the current literature, that a complete analysis of the dynamic effect of rigid coupling
misalignment on a real shaft-line, i.e. a hyperstatic shaft-line with several bearings and couplings like those used in turbo
machines, is lacking. In this paper, the authors propose a complete and original method to simulate the behavior of real
shaft line, supported by several oil-film bearings, with rigid coupling misalignment. Nonlinear effects are highlighted and
the spectral components of system response are analyzed, in order to give pertinent diagnostic information.

2. Modeling of rigid coupling misalignment in real shaft-lines

Usually, turbo machines are composed of some shafts connected by couplings, often of rigid type, and are supported by
journal bearings. The resulting shaft-line is hyperstatic, like that represented in Fig. 1. The shaft-line, which represents a
steam power unit, is composed of three different shafts, connected by two rigid couplings. In particular, the first shaft is
the HP-IP (high pressure–intermediate pressure) turbine, the second is the LP (low pressure) turbine and the last one is the
 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322 309

Rigid Rigid
coupling coupling
HP-IP turbine LP turbine Generator

Brg.#1 Brg.#2 Brg.#3 Brg.#4 Brg.#5 Brg.#6 Brg.#7

Fig. 1. Example of hyperstatic shaft-line composed of three shafts. The vertical pennants indicate the position of vibration measuring planes.

Fig. 2. Degrees of freedom of the nodes of a general rotor element.

generator. However, the model proposed is applicable to other types of machines, with different number of couplings,
shafts and bearings.
The shaft-line is modeled in a standard way by means of a finite beam model and rigid disks, considering only the
lateral vibrations, and 4 degrees of freedom (d.o.f.s) – two translational and two rotational – are considered per each node
(see Fig. 2). Axial and torsional vibrations will be neglected. Considering the general jth element of the shaft-line, the
generalized displacement vector xjðrÞ of the jth rotor node is ordered as follows:
n ðrÞ oT
xjðrÞ ¼ xj WxðrÞj yjðrÞ WyðrÞj ð1Þ

Two subsequent nodes, the jth and the (j þ1)th, define the jth element of the shaft-line. Index j is the main index used
to order shaft nodes and elements. If the shaft-line has nr nodes, thus nr 1 elements, the vector xðrÞ of the generalized
displacements of all the shaft-line nodes is composed by all the ordered vectors xðrÞ
j , as shown in Eq. (2):
n oT n ðrÞ ðrÞ ðrÞ ðrÞ ðrÞ ðrÞ ðrÞ ðrÞ
o T
xðrÞ ¼ x1ðrÞT    xnðrÞT ¼ x1 Wx1 y1 Wy1    xnr Wxnr ynr Wynr
r
ð2Þ

The mass matrix of the rotor ½MðrÞ , which takes also into account the secondary effect of the rotatory inertia, the
internal damping matrix ½CðrÞ , the stiffness matrix ½KðrÞ , which takes also into account the shear effect, and the gyroscopic
matrix ½GðrÞ , all of order ð4nr  4nr Þ, can be defined by means of standard Lagrange’s methods, considering beam elements
and rigid disks, as shown in [29,30]. Damping and gyroscopic matrices will be used in the following for the dynamic
simulations.
The shaft-line is supported on nb journal bearings. They are located in correspondence of some nodes of the shaft, which
are labeled by the indexes belonging to the set:
n o
Ic ¼ jBrg:#1    jBrg:#nb ð3Þ

Shaft nodes whose indexes belong to Ic are defined as constrained nodes and the vector of their generalized
displacements is indicated as xcðrÞ , while the remaining nodes are the free nodes and the corresponding vector is xðrÞ
f
:
ðrÞ
xcðrÞ ¼ xj2I c

xfðrÞ ¼ xj=
ðrÞ
2I c ð4Þ

During the machine installation, the shaft-line is statically aligned, using several methods like those described in [31],
in order to nullify static bending moments and shear forces on the rigid coupling flanges. This is realized by displacing the
310 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322

supports in vertical directions with respect to the geodesic line, so that elements of xcðrÞ are generally not all null. The static
centerline is a catenary.
While the term angular misalignment is self explanatory, the concept of offset is often used in technical literature. Offset
is defined in [31] as the ‘‘distance between two parallel shafts not coaxial’’ and it is intended fixed in the space. Offset
occurs for instance when there is a wrong static alignment between the two rotors to be coupled and there is an offset
between the static centerlines. In this case, static bending moments are not null on the flanges and static reaction forces on
the bearings are different from those of the case of ideal static alignment, but are not changing owing to the rotation of
the shaft.
In the case considered in the paper, on the contrary, the two static centerlines of the two rotors are ideally aligned, i.e.
the joint centerline is a catenary, but the flanges are wrongly machined (e.g. wrong radially distributed bolt holes) and/or
mounted, so that a rotating offset, i.e. a radial rotating misalignment, is imposed.
If rigid couplings were ideally aligned, without any radial or angular misalignment on their flanges, the static reaction
forces on the bearings could be calculated, in order to determine the oil-film dynamic characteristics.
Otherwise, like described in this paper, it is necessary to take into account the effect of rigid coupling misalignment on
the static centerline and, as a consequence, on the reaction forces of the bearing, considering that these forces are changing
owing to the rotation of the shaft, i.e. to the orientation of the misalignment with respect to the phase reference.

2.1. Effect of rigid coupling misalignment on the shaft-line

Let us consider Fig. 3 in which a close up of a rigid coupling of the machine is shown. For the sake of simplicity, only a
coupling is considered, being the model presented easily generalizable to nC couplings.
In the general case, the coupling faces are connected in correspondence of the jCth shaft node and both radial and
angular misalignment may occur as a consequence of a wrong mounting or imperfect machining. However, not only the
magnitudes of these misalignments have to be considered, but also the relative phase with respect to the phase reference
and the fact that the shaft is rotating with rotational speed O.
Hence, both types of misalignments are conveniently represented by means of rotating vectors, using for simplicity a
complex notation [32]:

angular misalignment : Daeija eiOt

radial misalignment : Dreijr eiOt ð5Þ

Fig. 3. Rigid coupling misalignment: reference of angular and radial misalignment.

 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322 311

Therefore, the effect of the rigid coupling misalignment is to impose generalized displacementsDxj ðyÞ, which are
function of the angular position y ¼ Ot of the shaft, on the d.o.f.s corresponding to the jCth coupling node.
2 3 2 3
1 0 1 0
( ) ( )
6 0 i 7 Dreijr 6 0 i 7 Dreijr
DxjC ðyÞ ¼ 6
6
7
7 eiO t
¼
6
6
7
7 eiy ð6Þ
4 i 0 5 Daeija 4 i 0 5 Daeija
0 1 0 1

Vector DxðyÞ representing all the displacements that are due to misalignment, has the same size of xðrÞ , i.e. ð4nr  1Þ,
because the displacements of the d.o.f.s, which are not corresponding to the coupling nodes, are set equal to 0.

2.2. Calculation of the static reactions on the bearings

The static reactions R(y), i.e. the load, on the bearings can be calculated by imposing the static equilibrium of the
free-body shaft-line and considering the imposed shaft alignment conditions, i.e. the displacements xcðrÞ , that is:

RðyÞ ¼ ½KðrÞ ðxðrÞ þ DxðyÞÞ þ W ¼ ½KðrÞ xðrÞ þ½KðrÞ DxðyÞ þW ¼ ½KðrÞ xðrÞ þ FC ðyÞ þ W ð7Þ

where W is the weight force vector that can be calculated as

 T
W ¼ ½MðrÞ  g 0 0 0    g 0 0 0 ð8Þ

and FC ðyÞ is the equivalent force due to the coupling misalignment. The only elements of vector FC ðyÞ that are different from
zero are those corresponding to the d.o.f.s of jCth node and they can be calculated by considering the stiffness submatrix
corresponding to the jCth node:
2 3
1 0
( )
6 0 i 7 Dreijr
6 7
FC 9j ¼ j ¼ ½KðrÞ 9j ¼ j 6 7 eiy ð9Þ
C C4 i 0 5 Daeija
0 1

After these considerations, Eq. (7) can be rewritten by reordering the d.o.f.s of the nodes and grouping the free and the
constrained ones:
2 ðrÞ 38 9 ( ) 
Kf f KðrÞ < xðrÞ =  ( )
fc f Wf FC f ðyÞ 0
4 5 þ þ ¼ ð10Þ
ðrÞ RðyÞ
Kcf KðrÞ
cc
: xcðrÞ ; Wc 0

since, obviously, FC c ðyÞ ¼ 0, because couplings are not in the same position of bearings.
The static free displacements as function of the angular position y of the shaft are obtained as

xfðrÞ ðyÞ ¼ ½KðrÞ

ff
1 ð½KðrÞ
fc
xðrÞ
c þ Wf þFC ðyÞÞ ð11Þ

and the reactions on the bearings as

ðrÞ ðrÞ ðrÞ ðrÞ
RðyÞ ¼ ½Kcf xf þ½Kcc xc þWc ð12Þ

Notice that, due to the presence of the coupling misalignment, the reactions of Eq. (12) have generally both the vertical
and the horizontal components and that they are 1X periodical.

2.3. Calculation of the oil-film dynamic characteristics

It is commonly agreed that the static reactions on the bearings are used as the input loads to calculate the oil-film
dynamic characteristics, because the static loads are largely predominant on the dynamical ones, due, for instance, to
unbalances. However, the loads are not constant and depend on the angular position y of the shaft, see Eq. (12), in the case
considered in the paper, as a consequence of the presence of the coupling misalignment. Hence, also the dynamic
characteristics of the oil-film in the bearings have to be calculated as a function of y. For the sake of simplicity, only
constant rotational speeds will be taken into account. This notwithstanding, the proposed method could be easily
extended also to speed transients.
Therefore, starting from the actual type and geometry of the nb bearings, Reynolds equation in the isoviscous form
[33,34] is used:
! !
3 3
@ h @p @ h @p @h
þ ¼ 6U Z þ 12W ð13Þ
@x Z @x @z Z @z @x
312 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322

where x and z are the coordinates along the sliding and the axial directions, respectively, Z the viscosity, h the oil-film
thickness, p the pressure, U the entraining velocity and W the relative velocity. Eq. (13) is then expressed in non-dimensional
form. The effects of small displacements and squeeze velocities are also introduced.
The integration of the Reynolds equation is performed by using finite differences. The journal static equilibrium
position is found by balancing the pressure distribution with the reactions R(y) in an iterative way. Then, the non-
dimensional stiffness and damping coefficients are obtained for each static equilibrium position, using standard methods
ðbÞ ðbÞ
described in [33,34]. Also the corresponding dimensional coefficients, krsi ðyÞ and crs i
ðyÞ, are calculated by means of those,
considering the loads R(y).

3. Simulation of shaft-line dynamical behavior

In order to perform the simulation of the dynamical behavior of the shaft-line at the operating speed, a dynamical
model has to be set-up. The matrices of the coupled shafts, i.e. the complete shaft-line, have been already introduced and
now the effects owing to the foundation dynamics and to the oil-film characteristics are accounted for.
Different methods can be used to model the foundation. For the sake of brevity, only pedestals, i.e. lumped 2 d.o.f.s
systems, will be considered. A discussion about other methods (modal or rigid) is reported in [35]. In a similar manner to
the shaft-line, also the d.o.f.s of the foundation, represented by its horizontal and vertical displacements and connected by
means of the nb bearings to the rotor, can be ordered in a vector:
n oT
xðf Þ ¼ xðf1 Þ yðf1 Þ  xðfnbÞ yðfnbÞ ð14Þ

Therefore, the complete vector of the generalized displacements of the system is

 T
x ¼ xðrÞT xðf ÞT ð15Þ

The structure of ½Mðf Þ , ½Cðf Þ  and ½Kðf Þ  is not relevant in this paper and depends on how the supporting structure is
implemented, see [35].
The dynamic forces of the oil-film of the ith bearing on the journal located at the jth rotor node, as a consequence of
dynamic displacements of the rotor d.o.f.s only, are
8 9 8 ðrÞ 9
2 > xjðrÞ >
3>
> >
>
2 3>> x_ j >>
ðbÞ
kxxi ðyÞ 0
ðbÞ
kxyi ðyÞ 0 >> >
>
ðbÞ
cxx ðyÞ 0 cðbÞ
xyi ðyÞ 0 > > ðrÞ >
> >
>
6 7>
> ðrÞ >
>
< Wxj = 6
i
7>< W xj >
> _ >
=
6 0 0 0 077 6 0 0 0 077
FðbrÞ
bi
ðyÞ ¼ 6
6 ðbÞ
6
7 yðrÞ >6 ðbÞ
ðbÞ ðrÞ ðbÞ ðrÞ
7> y_ ðrÞ > ¼ ½Ki ðyÞxj ½Ci ðyÞx_ j ð16Þ
0 5>
ðbÞ
4 kyxi ðyÞ 0 kyyi ðyÞ >
> j > > 4 cyxi ðyÞ 0 cðbÞ
yyi ðyÞ 0 5>> j >>
> ðrÞ >
> > > ðrÞ >
> >
0 0 0 0 >>W >
: >
; 0 0 0 0 > : W_ y >
> >
;
yj
j

while those that are due to the displacements of the supporting structure only, modeled by means of pedestals and
corresponding to the ith bearing, are
2 ðbÞ ðbÞ
38 ðf Þ 9 2 38 ðf Þ 9
kxxi ðyÞ kxyi ðyÞ < xj = ðbÞ
cxx ðyÞ cðbÞ
xyi ðyÞ
< x_ j = _ ðbÞ _ðbÞ
i
Fb ðyÞ ¼ 4 ðbÞ
ðbf Þ 5 4 ðbÞ 5 ¼ ½K i ðyÞxðfj Þ ½C i ðyÞx_ ðfj Þ ð17Þ
kyxi ðyÞ kyyi ðyÞ : yj ; cyxi ðyÞ cyyi ðyÞ : y_ ðfj Þ ;
i ðbÞ ðf Þ ðbÞ

The actual bearing force between the rotor and the foundation is given by the difference between Eqs. (16) and (17).
In this way, the coupling effect of the oil-film forces is taken into account by the relative displacements of the nodes
of the rotor and of the foundation in correspondence of the bearings. Thus, the fully assembled system of equation is
built up.
This requires the definition of the stiffness coupling matrices ½KðrrÞ , ½Kðrf Þ , ½Kðf rÞ , ½Kðf f Þ  and the corresponding damping
matrices ½CðrrÞ , ½Cðrf Þ , ½Cðf rÞ , ½Cðf f Þ , which are sparse and of order ð4nr  4nr Þ, ð4nr  2nb Þ, ð2nb  4nr Þ and ð2nb  2nb Þ,
respectively. The structure for the stiffness matrices is

½KðrrÞ ðyÞ ¼ diagð    ½KðbÞ

i
ðyÞ Þ ð18Þ

2 3
     
6 ðbÞ
kxxi ðyÞ 0
ðbÞ
kxyi ðyÞ 0 7
6 7
½Kðf rÞ ðyÞ ¼ 6
6 ðbÞ ðbÞ
7
7 ð19Þ
4 kyxi ðyÞ 0 kyyi ðyÞ 0 5
     
 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322 313

2 3
   
6 ðbÞ
kxxi ðyÞ
ðbÞ
kxyi ðyÞ 7
6 7
6 7
6 0 0 7
½K ðyÞ ¼ 6
ðrf Þ
6
7 ð20Þ
7
ðbÞ ðbÞ
6 kyxi ðyÞ kyyi ðyÞ 7
6 7
4 0 0 5
   

_ ðbÞ
½Kðf f Þ ðyÞ ¼ diagð    ½K i ðyÞ Þ ð21Þ

Damping coupling matrices have similar structure. The fully assembled system of equations, without external
excitation, results:
½Mx€ þ ½CðyÞx_ þ ½KðyÞðx þ DxðyÞÞ ¼ 0 ð22Þ

with
" #
½MðrÞ  0
½M ¼ ð23Þ
0 ½Mðf Þ 

" #
½CðrÞ  þ O½GðrÞ  þ ½CðrrÞ ðyÞ ½Cðrf Þ ðyÞ
½CðyÞ ¼ ð24Þ
½Cðf rÞ ðyÞ ½Cðf Þ  þ ½Cðf f Þ ðyÞ

" #
½KðrÞ  þ ½KðrrÞ ðyÞ ½Kðrf Þ ðyÞ
½KðyÞ ¼ ð25Þ
½Kðf rÞ ðyÞ ½Kðf Þ  þ ½Kðf f Þ ðyÞ

The remaining external forcing systems acting on the shaft-line are the weight W and the residual unavoidable
unbalance distribution, which will be taken into account by the equivalent unbalance at the juth node:
h iT h iT
Fu ðtÞ ¼ 0 ^ 1 0 i 0 ^ 0 mu r u O2 eiju eiOt ¼ 0 ^ 1 0 i 0 ^ 0 mu r u O2 eiju eiy ¼ Fu ðyÞ ð26Þ

By considering all the external forcing and Eq. (7), the fully assembled system of equations is nonlinear, because many
of its terms depend on the angular position y ¼ Ot:
½Mx€ þ ½CðyÞx_ þ ½KðyÞx ¼ FC ðyÞWþ Fu ðyÞ ð27Þ

where the over-bars indicate that the corresponding vectors are padded with zeros in positions corresponding to
foundation d.o.f.s.
The nonlinear system of equations in (27) is integrated in the time domain using Newmark’s implicit method, in which
all the quantities depending on y are evaluated for each integration step.

3.1. Simulation results

The machine, the model of which has been used for the simulations with the described method, is a steam turbo-
generator unit of about 320 MVA, already sketched in Fig. 1 and described in detail in [36]. Node number is 175 and
bearings #1, #2 and #7 are of tilting pad type, while the others of 2-lobes type.
Fig. 4 shows the static reaction calculated in case of ideal alignment of the shaft-line, i.e. when the centerline follows a
catenary and DxðyÞ ¼ 0. Obviously, static reactions have only the vertical component, while the reaction on bearing #7 is
very small and can be barely seen in the figure.
Different combinations of radial and angular misalignment conditions of the rigid coupling between HP-IP and LP
turbines have been analyzed. The first step of the proposed method is the calculation of the loads R(y): Figs. 5 and 6
illustrate the effect of radial and angular misalignment for two cases, respectively. The vectors of the static reactions are
plotted vs. the angular rotation over a complete revolution. It is easy to observe that static reactions do not have only
vertical component and that loads R(y) change remarkably on the bearings closer to the misaligned rigid coupling, i.e.
bearings #2 and #3. Some effects can be appreciated also on bearings #1 and #4, while practically there are not changes on
generator bearings. Therefore, without loss of generality, only the misalignment of the rigid coupling between HP-IP and
LP turbines will be considered in the following, whereas the dynamical response on generator bearings will be neglected.
The simulation of the dynamic response of the shaft-line is performed at the operating speed of 3000 rpm, considering
also the presence of an unbalance placed about in the middle of the LP turbine (0.3 kgm with phase 01).
An example is shown in Fig. 7, where only the steady-state shaft orbits in the vibration measuring plane close to
bearings from #1 to #4 are shown. These measuring planes are actually external to the journal bearings, and their exact
positions along the shaft line are indicated by the ‘‘pentagonal pennants’’ in Fig. 1. In this case, radial misalignment of
100 mm at 901 and angular misalignment of 15 mrad at 01 have been applied. Orbits in generator bearings are less affected
314 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322

Fig. 4. Static reactions in case of ideal alignment of the rigid couplings.

Fig. 5. Static reactions in case of radial misalignment of 50 mm at 01 of the coupling between HP-IP and LP turbine.

by the misalignment and are not shown for the same reasons explained before. Similarly, the orbits in the other nodes of
the shaft-line are not displayed both for the sake of brevity and because they are never measured in the practice.
The case considered would be already critical in the reality and these vibration amplitudes would be dangerous for the
actual rotating machine, because the bearing clearance would be exceeded. Orbits, in the measuring planes close to
bearings from #1 to #3, show rather clearly the presence of superharmonics components (owing to the deformed shape
with respect to the elliptical one) and the presence of nonlinear effects. These considerations are clearer from Figs. 8–11,
where the full spectra of the vibration, corresponding to the orbits in the measuring planes close to bearings from #1 to #4,
are considered.
Even though a logarithmic scale is used for the amplitudes, it is easy to observe that at least 72X and 73X components have
remarkable amplitudes (i.e. measurable, the horizontal dashed lines in the figures indicate the amplitude of 1 mm) in the bearings
 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322 315

Fig. 6. Static reactions in case of angular misalignment of 5 mrad at 01 of the coupling between HP-IP and LP turbine.

Fig. 7. Shaft-line orbits in bearings from #1 to #4 in case of radial misalignment of 100 mm at 901 and angular misalignment of 15 mrad at 01.

from #1 to #3. Bearing #4 is rather ‘‘far’’ from the misaligned coupling and does not ‘‘feel’’ so much the effect of the misalignment:
higher harmonic components are much smaller than the 71X component.
It is also interesting to consider the effect of the increase of the radial or of the angular misalignment, ceteris paribus.
Fig. 12 shows the orbits in the measuring plane close to bearing #1 for increasing values of radial misalignment from 50 to
200 mm. It is possible to note how the dimension of the orbits is increasing, i.e. vibration amplitudes along measuring
316 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322

Node 4 − Meas. plane close to bearing #1 − Full spectrum

10−3
Mean value
−1X
10−4
−2X
1X
2X
−3X
10−5

Amplitude [m]
−4X 3X
10−6

10−7

10−8

10−9

10−10
−300−250−200−150−100 −50 0 50 100 150 200 250 300
Frequency [Hz]

Fig. 8. Full spectrum of shaft-line orbit close to bearing #1 in case of radial misalignment of 100 mm at 901 and angular misalignment of 15 mrad at 01.

Node 61 − Meas. plane close to bearing #2 − Full spectrum

10−3
−1X
1X
10−4 −2X
2X
−3X
10−5
Amplitude [m]

10−6 3X 5X

10−7

10−8

10−9

10−10
−300 −250 −200 −150 −100 −50 0 50 100 150 200 250 300
Frequency [Hz]

Fig. 9. Full spectrum of shaft-line orbit close to bearing #2 in case of radial misalignment of 100 mm at 901 and angular misalignment of 15 mrad at 01.

Node 78 − Meas. plane close to bearing #3 − Full spectrum

10−2

Mean value
10−3
1X
−1X

10−4 −2X
2X
Amplitude [m]

−3X
10−5
3X
4X 5X
10−6 −4X

10−7

10−8

10−9
−300 −250 −200 −150 −100 −50 0 50 100 150 200 250 300
Frequency [Hz]

Fig. 10. Full spectrum of shaft-line orbit close to bearing #3 in case of radial misalignment of 100 mm at 901 and angular misalignment of 15 mrad at 01.
 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322 317

Node 122 − Meas. plane close to bearing #4 − Full spectrum

10−3
−1X Mean
value 1X

10−4

−2X

Amplitude [m]
10−5 2X

−3X

10−6 4X

10−7

10−8
−300 −250 −200 −150 −100 −50 0 50 100 150 200 250 300
Frequency [Hz]

Fig. 11. Full spectrum of shaft-line orbit close to bearing #4 in case of radial misalignment of 100 mm at 901 and angular misalignment of 15 mrad at 01.

Node 4 − Meas. plane close to bearing #1

−250

Δr=50μm@0°
−260
Δr=100μm@0°

Δr=150μm@0°
−270

−280
[μm]

−290

−300

−310
Δr=200μm@0°
−320
−40 −30 −20 −10 0 10 20 30
[μm]

Fig. 12. Comparison of orbits in the measuring plane close to bearing #1 for different radial misalignments.

directions are increasing. Also orbit shapes are deviating from the elliptical shape, clearly indicating the presence of
superharmonic components. Similar results can be drawn by considering Fig. 13, which shows the orbits in the same
measuring plane for increasing values of angular misalignment from 5 to 35 mrad.

3.2. Effect of radial misalignment

As it has been shown in the previous section, the model is able to consider simultaneously both radial and angular
misalignment, with arbitrary combination of amplitudes and phases, and to explain the arising of nonlinear effects. However,
because of the infinite possible combinations of radial and angular misalignments, a comprehensive analysis is rather awkward.
Thus, the effects of radial and angular misalignments are analyzed separately, starting from the former one.
In order to evaluate the ‘‘degree’’ of nonlinearity, the ratios between the superharmonic component amplitudes and the
synchronous one are considered. Moreover, the harmonic components along the two principal directions, vertical and
horizontal, are considered, because these are the directions of the measuring probes on actual machines. In this way, it is
easier to give pertinent diagnostic information.
318 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322

Node 4 − Meas. plane close to bearing #1

−50

−100

−150
Δ=5mrad
−200
Δ=10mrad
−250
[μm] Δ=15mrad
−300
Δ=20mrad
−350
Δ=25mrad
−400
Δ=30mrad
−450
Δ=35mrad
−500

−550
−250 −200 −150 −100 −50 0 50 100 150 200 250
[μm]

Fig. 13. Comparison of the orbits in the measuring plane close to bearing #1 for different angular misalignments.

Vertical vibration − Radial misalignment

80
Brg. #1
60 Brg. #2
2X/1X [%]

Brg. #3
40 Brg. #4

20

0
40 60 80 100 120 140 160 180 200 220
Radial misalignment [μm]

50
Brg. #1
40 Brg. #2
3X/1X [%]

Brg. #3
30
Brg. #4
20

10

0
40 60 80 100 120 140 160 180 200 220
Radial misalignment [μm]

15
Brg. #1
Brg. #2
4X/1X [%]

10 Brg. #3
Brg. #4

0
40 60 80 100 120 140 160 180 200 220
Radial misalignment [μm]

Fig. 14. Vertical vibration in bearings from #1 to #4: 2X/1X, 3X/1X and 4X/1X amplitude ratios for increasing magnitudes of radial misalignment at 01.

Figs. 14 and 15 show 2X/1X, 3X/1X and 4X/1X amplitude ratios, respectively, for the vertical and the horizontal
directions and for the vibration measuring planes close to bearings from #1 to #4. Radial misalignment magnitudes are in
the range from 50 to 210 mm with phase of 01, i.e. in phase with the unbalance.
 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322 319

Horizontal vibration − Radial misalignment − Phase 0°

150
Brg. #1
Brg. #2

2X/1X [%]
100 Brg. #3
Brg. #4

50

0
40 60 80 100 120 140 160 180 200 220
Radial misalignment [μm]

40
Brg. #1
30 Brg. #2
3X/1X [%]

Brg. #3
20 Brg. #4

10

0
40 60 80 100 120 140 160 180 200 220
Radial misalignment [μm]

20
Brg. #1
15 Brg. #2
4X/1X [%]

Brg. #3
10 Brg. #4

0
40 60 80 100 120 140 160 180 200 220
Radial misalignment [μm]

Fig. 15. Horizontal vibration in bearings from #1 to #4: 2X/1X, 3X/1X and 4X/1X amplitude ratios for increasing magnitudes of radial misalignment at 01.

Vertical vibration − Radial misalignment − Phase 90°

150
Brg. #1
Brg. #2
2X/1X [%]

100 Brg. #3
Brg. #4

50

0
50 100 150
Radial misalignment [μm]

20
Brg. #1
15 Brg. #2
3X/1X [%]

Brg. #3
10 Brg. #4

0
50 100 150
Radial misalignment [μm]

2
Brg. #1
1.5 Brg. #2
4X/1X [%]

Brg. #3
1 Brg. #4

0.5

0
50 100 150
Radial misalignment [μm]

Fig. 16. Vertical vibration in bearings from #1 to #4: 2X/1X, 3X/1X and 4X/1X amplitude ratios for increasing magnitudes of radial misalignment at 901.
320 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322

Horizontal vibration − Radial misalignment − Phase 90°

100
Brg. #1
80 Brg. #2

2X/1X [%]
Brg. #3
60
Brg. #4
40

20

0
50 100 150
Radial misalignment [μm]

25
Brg. #1
20 Brg. #2
3X/1X [%]

Brg. #3
15
Brg. #4
10

0
50 100 150
Radial misalignment [μm]

2
Brg. #1
1.5 Brg. #2
4X/1X [%]

Brg. #3
1 Brg. #4

0.5

0
50 100 150
Radial misalignment [μm]

Fig. 17. Horizontal vibration in bearings from #1 to #4: 2X/1X, 3X/1X and 4X/1X amplitude ratios for increasing magnitudes of radial misalignment at 901.

Vertical vibration − Angular misalignment − Phase 0°

800
Brg. #1
600 Brg. #2
2X/1X [%]

Brg. #3
400 Brg. #4

200

0
5 10 15 20 25 30 35
Angular misalignment [mrad]

300
Brg. #1
Brg. #2
200
3X/1X [%]

Brg. #3
Brg. #4

100

0
5 10 15 20 25 30 35
Angular misalignment [mrad]

80
Brg. #1
60 Brg. #2
4X/1X [%]

Brg. #3
40 Brg. #4

20

0
5 10 15 20 25 30 35
Angular misalignment [mrad]

Fig. 18. Vertical vibration in bearings from #1 to #4: 2X/1X, 3X/1X and 4X/1X amplitude ratios for increasing magnitudes of angular misalignment at 01.
 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322 321

Horizontal vibration − Angular misalignment − Phase 0°

400
Brg. #1
300 Brg. #2

2X/1X [%]
Brg. #3
200 Brg. #4

100

0
5 10 15 20 25 30 35
Angular misalignment [mrad]

250
Brg. #1
200 Brg. #2
3X/1X [%]

Brg. #3
150
Brg. #4
100

50

0
5 10 15 20 25 30 35
Angular misalignment [mrad]

40
Brg. #1
30 Brg. #2
4X/1X [%]

Brg. #3
20 Brg. #4

10

0
5 10 15 20 25 30 35
Angular misalignment [mrad]

Fig. 19. Horizontal vibration in bearings from #1 to #4: 2X/1X, 3X/1X and 4X/1X amplitude ratios for increasing magnitudes of angular misalignment at 01.

With the exception of the 2X/1X amplitude ratio for the vertical direction in bearing #1, the ratios have more than
linear increasing trends as a function of radial misalignment magnitude. The contribution of 4X components becomes
significant only for high radial misalignment magnitude. 2X horizontal component becomes almost equal to 1X one for
bearing #1 when the misalignment is 210 mm.
For the considered machine model, if the radial misalignment is imposed in quadrature to the unbalance, the increasing
of the ratios becomes quicker than in the case of 01 phase, as shown in Figs. 16 and 17 for the 2X/1X ratio and the vertical
direction only. In this case, 2X amplitude is about equal to 1X amplitude in horizontal direction and about one and half in
vertical direction, when the radial misalignment is 100 mm. Similar results, in terms of quick increasing when the
misalignment phase is 901, are obtained for 3X/1X and 4X/1X ratios.

3.3. Effect of angular misalignment

The effect of increasing magnitude of angular misalignment from 5 to 35 mrad with phase of 01 is shown in Figs. 18 and
19. Also in this case, nonlinear effects are evident. Even if the direct vibrations, corresponding to the maximum value of the
considered angular misalignment, are comparable to those of the maximum value of radial misalignment, the ratio
between the superharmonics components and the 1X component is greater.

4. Conclusions

Several studies in literature deal with rotor misalignment, but it looks like that this term is used to indicate different
physical processes. In the paper, the authors have dealt with rigid coupling misalignment of a hyperstatic shaft-line
equipped with journal bearings. A general method, suitable for every kind of rotating machine, has been presented in
detail. The model allows considering both radial and angular misalignment, separately or combined.
In order to show the effects that are caused by rigid coupling misalignment, some simulations are presented by
considering a shaft-line reproducing a real turbine unit. The results of the simulations highlighted the presence of
nonlinear effects in the system response: superharmonic components are the most remarkable effects of rigid coupling
misalignment and are due to the variable loading on the bearings. The ratios between higher order harmonic components
322 P. Pennacchi et al. / Mechanical Systems and Signal Processing 30 (2012) 306–322

and synchronous vibrations are analyzed. These results allow giving diagnostic information about the signature of this
type of fault.

References

[1] A. Muszyńska, Rotordynamics, Taylor & Francis Group—CRC Press Book, New York, USA, 2005.
[2] A.W. Lees, Misalignment in rigidly coupled rotors, J. Sound Vib. 305 (1–2) (2007) 261–271.
[3] J. Boyd, Bearing misalignment—Part 1, Lubrication Eng. 25 (8) (1969) 326–327.
[4] J. Boyd, Bearing misalignment—Part 2, Lubrication Eng. 25 (9) (1969) 366.
[5] B. Bou-Saı̈d, D. Nicolas, Effect of misalignment on static and dynamic characteristics of hybrid bearing, Tribol. Trans. 35 (2) (1992) 325–331.
[6] B.S. Prabhu, An experimental investigation on the misalignment effects in journal bearings, Tribol. Int. 40 (2) (1997) 235–242.
[7] J. Bouyer, M. Fillon, An experimental analysis of misalignment effect on hydrodynamic plain journal bearing performances, Trans. ASME—J. Tribol.
124 (2) (2002) 313–319.
[8] D.E. Bently, Fundamentals of Rotating Machinery Diagnostics, Bently Pressurized Bearings Press, Minden, NV, USA, 2002.
[9] M. Xu, R.D. Marangoni, Vibration analysis of a motor-flexible coupling-rotor system subject to misalignment and unbalance, Part I: Theoretical
model and analysis, J. Sound Vib. 176 (5) (1994) 663–679.
[10] M. Xu, R.D. Marangoni, Vibration analysis of a motor-flexible coupling-rotor system subject to misalignment and unbalance, Part II: Experimental
validation, J. Sound Vib. 176 (5) (1994) 681–691.
[11] A.S. Sekhar, B.S. Prabhu, Effects of coupling misalignment on vibrations of rotating machinery, J. Sound Vib. 185 (4) (1995) 655–671.
[12] C.B. Gibbons (1979) Coupling misalignment forces. In: Proc. of 5th Turbomachinery Symposium, Texas A&M University, College Station, USA,
pp. 111–116.
[13] Y.-S. Lee, C.-W. Lee, Modelling and vibration analysis of misaligned rotor-ball bearing systems, J. Sound Vib. 224 (1) (1999) 17–32.
[14] W. Hu, H. Miah, N.S. Feng, E.J. Hahn, A rig for testing lateral misalignment effects in a flexible rotor supported on three or more hydrodynamic
journal bearings, Tribol. Int. 33 (3–4) (2000) 197–204.
[15] K.M. Al-Hussain, I. Redmond, Dynamic response of two rotors connected by rigid mechanical coupling with parallel misalignment, J. Sound Vib.
249 (3) (2002) 483–498.
[16] K.M. Al-Hussain, Dynamic stability of two rigid rotors connected by a flexible coupling with angular misalignment, J. Sound Vib. 266 (2) (2003)
217–234.
[17] T.H. Patel, A.K. Darpe, Experimental investigations on vibration response of misaligned rotors, Mech. Syst. Signal Process. 23 (7) (2009) 2236–2252.
[18] T.H. Patel, A.K. Darpe, Vibration response of misaligned rotors, J. Sound Vib. 325 (3) (2009) 609–628.
[19] S. Prabhakar, A.S. Sekhar, A.R. Mohanty, Vibration analysis of a misaligned rotor-coupling-bearing system passing through the critical speed, IMechE
Proc. Inst. Mech. Eng. Part C—J. Mech. Eng. Sci. 215 (2001) 1417–1428 (part C).
[20] Z. Peng, F. Chu, Y. He, Vibration signal analysis and feature extraction based on reassigned wavelet scalogram, J. Sound Vib. 253 (5) (2002)
1087–1100.
[21] Z. Peng, Y. He, Z. Chen, F. Chu, Identification of the shaft orbit for rotating machines using wavelet modulus maxima, Mech. Syst. Signal Process.
16 (4) (2002) 623–635.
[22] J.K. Sinha, A.W. Lees, M.I. Friswell, Estimating unbalance and misalignment of a flexible rotating machine from a single run-down, J. Sound Vib.
272 (2004) 967–989.
[23] P.N. Saavedra, D.E. Ramı́rez, Vibration analysis of rotors for the identification of shaft misalignment—Part 1: theoretical analysis, IMechE Proc. Inst.
Mech. Eng. Part C—J. Mech. Eng. Sci. 218 (9) (2004) 971–985.
[24] P.N. Saavedra, D.E. Ramı́rez, Vibration analysis of rotors for the identification of shaft misalignment—Part 2: experimental validation, IMechE Proc.
Inst. Mech. Eng. Part C—J. Mech. Eng. Sci. 218 (9) (2004) 987–999.
[25] P. Pennacchi, A. Vania, Diagnosis and model based identification of a coupling misalignment, Shock Vib. 12 (4) (2005) 293–308.
[26] N. Bachschmid, P. Pennacchi, Accuracy of fault detection in real rotating machinery using model based diagnostic techniques, JSME Int. J. Ser. C
46 (3) (2003) 1026–1034.
[27] C.-Y. Tsai, S.-C. Huang, Transfer matrix for rotor coupler with parallel misalignment, J. Mech. Sci. Technol. 23 (5) (2009) 1383–1395.
[28] Bahaloo H., Ebrahimi A., Samadi M. (2009) Misalignment modeling in rotating systems. In: Proceedings of ASME Turbo Expo 2009: Power for Land,
Sea and Air, GT2009, June 8–12, 2009, Orlando, FL, USA, paper GT2009-60121, pp. 1–7.
[29] M. Lalanne, G. Ferraris, Rotordynamics Predictions in Engineering, John Wiley & Sons Inc., Chichester, England, 1998.
[30] D.W. Childs, Turbomachinery Rotordynamics, John Wiley & Sons Inc, Chichester, England, 1993.
[31] V. Wowk, Machine Vibration: Alignment, Mc Graw-Hill, New York, 2000.
[32] N. Bachschmid, P. Pennacchi, A. Vania, Identification of multiple faults in rotor systems, J. Sound Vib. 254 (2) (2002) 327–366.
[33] T. Someya, Journal-Bearing Databook, Springer-Verlag, Berlin, Germany, 1989.
[34] G.W. Stachowiak, A.W. Batchelor, Engineering Tribology, Butterworth Heinemann, Burington, USA, 2005.
[35] P. Pennacchi, N. Bachschmid, A. Vania, G.A. Zanetta, L. Gregori, Use of modal representation for the supporting structure in model based fault
identification of large rotating machinery: Part 1—theoretical remarks, Mech. Syst. Signal Process. 20 (3) (2006) 662–681.
[36] P. Pennacchi, N. Bachschmid, A. Vania, G.A. Zanetta, L. Gregori, Use of modal representation for the supporting structure in model based fault
identification of large rotating machinery: Part 2—Application to a real machine, Mech. Syst. Signal Process. 20 (3) (2006) 682–701.