Mechanical Systems and Signal Processing 36 (2013) 618629

Contents lists available at SciVerse ScienceDirect

Mechanical Systems and Signal Processing

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

Active vibrations control of journal bearings with the use

of piezoactuators
b
a

Jir Tuma a,n, Jir Simek

, Jaromr Skuta
, Jaroslav Los a
a
b

Faculty of Mechanical Engineering, VSBTechnical University of Ostrava, 17. listopadu 15, CZ 708 33, Ostrava, Czech Republic
207, CZ 190 00, Praha 9, Czech Republic
TECHLAB Ltd., Prague, Sokolovska

a r t i c l e in f o

abstract

Article history:
Received 19 April 2011
Received in revised form
28 October 2012
Accepted 28 November 2012
Available online 22 January 2013

Rotor instability is one of the most serious problems of high-speed rotors supported by
sliding bearings. With constantly increasing parameters, new machines problems with
rotor instability are encountered more and more often. Even though there are many
solutions based on passive improvement of the bearing geometry to enlarge the
operational speed range of the journal bearing, the paper deals with a working prototype
of a system for the active vibration control of journal bearings with the use of
piezoactuators. The controllable journal bearing is a part of a test rig, which consists of
a rotor driven by an inductive motor up to 23,000 rpm. The actively controlled journal
bearing consists of a movable bushing, which is actuated by two piezoactuators. The
journal vibration is measured by a pair of proximity probes. The control system enables
run-up, coast-down and steady-state rotation. A real-time simulator dSpace encloses the
control loop. Force produced by piezoactuators and acting at the bushing is controlled
according to error signals derived from the proximity probe output signals. As it was
proved by experiments the active vibration control extends considerably the range of the
operational speed.
& 2012 Elsevier Ltd. All rights reserved.

Keywords:
Active vibration control
Journal bearings
Hydrodynamic bearings
Piezoactuators
Rotor instability
Fluid-induced vibration

1. Introduction
Both the VSBTechnical University, Faculty of Mechanical Engineering, and TECHLAB Ltd., Prague, are focused on longterm research in the eld of rotor dynamics. One of the most serious problems is instability of high-speed rotors due to the
journal bearing oil lm. To study possibilities of affecting rotor behavior by controlled movement of bearing bushings, a
test rig was designed, manufactured and assembled. Even though there are many solutions based on passive improvements of the bearing geometry to enlarge the operational speed range of the journal bearing, such as a lemon bore,
pressure dam, tilting pad, parametric stiffness excitation [1], etc., the approach to preventing the journal bearing
instability, presented in the paper, is based on the use of the active vibration control (AVC). In the introduction it should be
emphasized that research of active vibration control was aimed at rigid rotors and the standard design of sliding bearings,
where the journal displacement is measured at the closest position to the bearing bushing. Some publications address the
vibration of exible rotors, where there is large displacement of the rotor. Such a large displacement is easily measurable
by eddy-current sensors [2,3]. Problems with the measurement of displacements with the use of this sensor type are
discussed in this article.

Corresponding author. Tel.: 420 59 699 3482; fax: 420 59 699 6129.

E-mail addresses: jiri.tuma@vsb.cz (J. Tuma), j.simek@techlab.cz (J. Simek),

jaromir.skuta@vsb.cz (J. Skuta),
jaroslav.los.st@vsb.cz (J. Los).

0888-3270/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ymssp.2012.11.010

ma et al. / Mechanical Systems and Signal Processing 36 (2013) 618629

J. Tu

619

Many authors pay attention to the active control of sliding bearings with the use of active magnetic bearings (AMB) [4]
and giant magnetostrictive material (GMM) [5]. Although the authors of the paper on using GMM state that experiments
can be carried out up to 1700 rpm, they publish only measurements at 350 rpm. The instability due to the oil lm is a
problem of high-speed rotors. With utmost probability instability could not arise at such low rotational speed as 350 rpm.
Therefore, the active vibration control was not aimed at eliminating instability of sliding bearings, but only at positioning
the shaft axis. Favorite means for rotor control are magnetorheological liquids, but the delayed response of the liquid
viscosity to the magnetic eld change does not allow using this method for closed loop control of high-speed rotors [6].
Piezoactuators as a tool to control of rotating machines have been intensively investigated in the literature since the end of
1980s. The advantage of journal bearings with piezoactuators is that the bearing bushing mounting stiffness remains
unchanged in the case of an accidental loss of electric power supply. One of the rst original contributions dated from the
beginning of the 1990s [7]. These papers did not study the effect of the oil lm on the onset of instability and its
suppression using the active vibration control. Worth mentioning are papers [8,9] dealing with the problem of the rotor
instability.
There are two promising ways how to control the rotor movement, either by magnetic bearings or by piezoactuators,
which are added to the system. However, AMB are still very expensive and cases, where it is possible to install AMB to
already designed rotor, are very rare. As already mentioned, the ABM requires retainer bearings in case of power failure,
which complicates machine design. On the other hand, installation of piezoactuator into bearing housing is relatively easy.
The non-rotating loose bushing can be inserted into any bearing housing. The only problem is to nd piezoactuators for
reasonable prize.
The topic of this paper is focused at research of methods suppressing sliding journal bearing instability by the active
vibration control with the use of piezoactuators [10,11]. Stacked piezoactuators are modern actuators acting with great
force and in a wide frequency range to several kHz. The only disadvantage of the piezoactuators is a small travel range.
As it was mentioned before the active vibration control system of journal bearings with the use of the piezoactuators
which is described in this paper differs from the other attempts to test such a system by this that our bearing design is
much closer to the industrial design than academic laboratory models [1,2,12]. The rotor is rigid and the position of the
journal inside the bearing bushing is measured in close distance to it. Closest to our solution is to use giant
magnetostrictive material [5]. Unfortunately, the authors have not noticed a positive effect on the operating rotational
speed range.
The research work was granted by the Czech Science Foundation as a part of the research project No. 101/07/1345
Active control of journal bearings aimed at suppressing the rotor instability. The control system adds an electronic
feedback to the rotor-bearing system actuating the position of a movable bushing. The current passive damper changes
into an active component of the system with controllable properties. The laboratory test facilities, including the journal
bearing equipped with the movable bushing, were designed by TECHLAB Ltd., Prague. The research group of Technical
University of Ostrava developed the control system, which is based on piezoactuators, and put the system into operation
[13,14].

2. Test rig
The sketch of a controllable journal bearing arrangement, which is implemented for the active vibration control, is
shown in Fig. 1 while some details can be seen on photos in Fig. 2. The test rig consists of a rigid shaft of 30 mm diameter
supported in two cylindrical hydro-dynamic journal bearings of the length of 30 mm. An input for the oil inlet is in the
horizontal plane of symmetry of the bushing. Shafts with various radial clearances are available for testing but results of

Fig. 1. Arrangement of the controllable journal bearing.

620

ma et al. / Mechanical Systems and Signal Processing 36 (2013) 618629

J. Tu

Fig. 2. Test rig for active vibration control research.

the experiments are presented only for radial clearance of 45 mm. The bearing span is 200 mm. Bearing bushings are
supported in rubber O rings, which ensure sealing of oil inlet and at the same time enable movement of bushings within
the clearance in bearing casing. Bearing bushings can be excited by means of piezoactuators oriented in vertical and
horizontal directions and fastened to the rig frames. The preloaded open-loop LVPZT piezoactuators are of the P-844.60
type, the product of the Physik Instrumente Company. The piezoactuator require a low voltage amplier with the 100 V
peak value at the output. The piezoactuator travel range is up to 90 mm, the pushing force is up to 3000 N and the pulling
force is up to 700 N. The resonant frequency of piezoactuators is of order thousands of hertz while the frequency range for
transferring a control signal with no distortion is required to be of order hundreds of hertz, which is half the frequency of
the rotor rotation speed.
The test shaft is driven by high-frequency motor through an elastic membrane coupling, constituting two joints, so that
the shaft is decoupled from motor and free to move. Concerning a lubricant, it was initially used the hydraulic oil of the VG
32 grade and then bearing special oil for high-speed grinder spindle bearing of the OL-P03 grade.
Shaft movement is measured by two pairs of proximity probes. These sensors are working on eddy-current and
& Kjr Company, but the
electrical capacitive principles. The eddy current sensors IN-085 are a product of the Bruel
sensors from Bently Nevada Rotor Kit were alternatively tested as well. After some problems with the measurement errors
of the eddy-current sensors the capacitive sensors of the capaNCDT CS05 type supplied by the Micro Epsilon Company,
were installed. It is possible to put one or two discs on the shaft, thus increasing bearing load and rotor mass. However,
lowest stability limit should be achieved with the minimum bearing load, i.e. with hollow shaft without discs. The test rig
was designed for speeds up to 23,000 rpm. This paper describes tests in which the rotor instability with AVC ON and AVC
OFF due to the oil lm begins at much lower speeds, as reported in the section dealing with lubrication. The speed range of
the rotor during tests depends on combination of the journal clearance and the oil viscosity.
3. Journal bearing model
There are many ways how to model journal bearings [1517], but this paper prefers a lumped parameter model, which
is based on the concept developed by Muszynska [18] with the support of Bently Rotor Dynamics Research Corporation
[19]. The reason for using this concept was that it offers an effective way to understand the rotor instability problem and to
create a model of a journal vibration active control by manipulating the bushing position by actuators, which are a part of
the closed loop system composed of proximity probes and a controller. Another approach can be based on the lubricant
ow prediction using a FE method for Reynolds equation solution. This more sophisticated method does not allow
analyzing behavior of the vibration active control to design and tune the controller.
3.1. Lumped parameter model of the journal bearing
Methods for controller design are based on the lumped parameter model which enables to derive a transfer function
between the controlled variables and the control variable of the closed loop control systems. Partial differential equations
like Reynolds equations or the NavierStokes equations do not produce a transfer function. The transfer function is a tool
for tuning the controller aimed to the stability of the closed loop and a suitable response to disturbance.
Let the rotor angular velocity be designated by O of the unit in radians per a second. It is assumed that the bushing is a
movable part in two perpendicular directions while the rotor is rotating. This mathematical model proposes to use
complex variables as position vectors to describe motion of the rotor and bushing in the plane, which is perpendicular to
the rotor axis. The real part of the position vector r is a horizontal coordinate x(t) while the imaginary part is a vertical
coordinate y(t) of the journal centre. The coordinate system is tied to stationary bearing housing with a cylindrical bore,
inside of which is inserted movable bearing bushing. The origin (0, 0) of the coordinate system in the complex plane is
situated in the centre of the mentioned cylindrical bore as it is shown in Fig. 3. The positions of the journal centre and
bushing mean the intersection of both movable component axes with the complex plane. The position of the journal centre
in the complex plane is designated by the mentioned position vector r, while the position of the bushing centre is

ma et al. / Mechanical Systems and Signal Processing 36 (2013) 618629

J. Tu

621

Fig. 3. Coordinate system in the complex plane.

designated by the position vector u (see Fig. 3). The coordinates of the end-point of this vector are
0,0

cylindricalborecentre

r xt jyt journal rotor centre

u ux t juy t bushingcentre
where j is an imaginary unit.
The internal spring, damping and tangential forces are acting on the rotor. The external forces refer to forces that are
applied to the rotor, such as unbalance, impacts and preloads in the form of constant radial forces. All these external forces
are considered as an input for the mathematical model. The uid pressure wedge is the actual source of the uid lm
stiffness in a journal bearing and maintains the rotor in equilibrium. As it is stated in [18,20] these bearing forces can be
modeled by a spring and damper system, which is rotating at the angular velocity lO, where lis a dimensionless
parameter, which is slightly less than 0.5. This model is an approximation, but as will be shown later the model
satisfactorily explains the onset of instability.
The parameter l is denominated as the uid averaged circumferential velocity ratio. It is assumed that the rotating
journal drags the uid in a space between two cylinders into motion and acts as a pump. It is easy to understand that the
uid circular velocity is varying across the gap as a consequence of uid viscosity. The validity of the assumption can be
veried experimentally. It is known that an oscillation (an onset of instability) of the rotor starts when the rotor rotational
speed exceeds a certain value and stops when the speed decreases under the other one. It can be shown by experiments,
that when the rotor system is excited by a non-synchronous perturbation force with respect to the rotor rotational speed
the resonance appears at the frequency, which is approximately equal to l O.
Fluid forces acting on the rotor in coordinates rotating at the same angular frequency as the spring and damper system
are determined by the position of the journal centre relating to the bushing centre and therefore they are given by the
formula
Frot K rrot urot D r_ rot u_ rot

where scalar parameters, K and D, are specifying proportionality of stiffness and damping to the relative position of the
journal centre displacement vector rrot urot and velocity vector r_ rot u_ rot , respectively. Eq. (1) represents relationship
between complex variables and replaces two real equations in fact.
To create the model of the rotor system, the uid forces have to be expressed in the stationary coordinate system, in
which the rotor centerline displacement and velocity vectors are designated by r and r_ , respectively. Conversion of the
complex rotating vector rrot to the stationary coordinate system can be done by multiplication of this vector by a factor
exp j lO t , which is the same as multiplying the vector in the stationary coordinates by the factor exp j l O t . The
relationship between the mentioned vectors in rotating and stationary coordinates are given by the formulas:
F K ru D r_ u_ jDlO ru

where the complex term j Dl O r has the meaning of the force acting tangentially to the direction of the vector r. As the
rotor angular velocity increases, this force can become very large and can cause rotor instability.
As was mentioned, the rotor is under the inuence of the external forces, for instance produced by unbalance or simply
by gravity. To obtain general solution this external perturbation force, resulting from unbalance, is assumed to be rotating
at the angular velocity o, which is considered to be completely independent of the rotor angular velocity O. The unbalance
force, which is produced by unbalance mass m located at a radius r u , acts in the radial direction and has a phase d at time
t0
FP mr u o2 ejot d

ma et al. / Mechanical Systems and Signal Processing 36 (2013) 618629

J. Tu

622

The equation of motion for a rigid rotor operating in a small, localized region in the journal bearing, is as follows:
M r D r_ u_ KjDlO ru FP

where M is the total rotor mass. The trajectory of the rotor centerline is called an orbit. As in the case of Eq. (1) the complex
Eq. (4) can be replaced by two real equations. The complex variables are used to simplify not only writing mathematical
formulas but for easy creation of the simulation model in Matlab-Simulink [21] for testing a control system.
Model parameters for simulation are related to the above-described test rig. The mass M of the rotor is 1.6 kg and the
experiments result in the value of l 0.475. The stiffness K and damping D of the oil lm depend on the distance between
the journal axis and the bushing axis. If the position of the journal axis is close to the bushing axis then both parameters
are almost constant. When the journal is approaching the bushing wall then both parameters grow to innity. In the
simulation calculations the oil lm stiffness is set in such a way to match the onset of instability during the experiment.
The damping parameter is adjusted to the value which ensures that the journal orbits of the simulation and experiment
are similar. Typical values for the dimensions of the test rig is K 4000 N/m and D 1000 Ns/m. The value of the product m
ru 0.00001 kg m corresponds to the ISO balancing class between G 1 and G 2.5 at 2500 rpm.
3.2. Equation of motion as a servomechanism
For the stability analysis of the journal bearing it is assumed that the bushing is not moving u 0. According to the
model (2), the rotor and uid wedge bearing systems can be demonstrated as a servomechanism working in the closed
loop, which is shown in Fig. 4. According to the direction of the acting force at the shaft the direct and quadrature dynamic
stiffness is introduced. To obtain the Laplace transform of the motion equation, the imaginary variable jo is replaced by a
complex variable s
K Direct s K Ds Ms2 ,

K Q uadrature s jlOD

and the equation of motion (4) takes the form

r FPerturbation K Quadrature s r=K Direct s

The transfer function 1=K Direct s (dynamic compliance) is stable due to the positive value of coefcients. The feedback
path in the closed-loop system acts as a positive feedback and introduces instability for the closed-loop system. The gain of
the positive feedback depends on the rotor angular velocity O. The closed-loop system is stable for the low rotor rotational
speed. But there is a margin for the stable behavior. If the gain of the positive feedback crosses over some limit value then
the system becomes unstable.
The stability of the closed-loop dynamic system depends on the open-loop frequency transfer function
G0 jo

K Q uadrature jo
l O D



K Direct jo
o Dj KMo2

As is known, the closed-loop dynamic system is stable according to the Nyquist stability criterion if, and only if, the
locus of the function G0 jo in the complex plane does not enclose the point (  1,0) as o is varied from zero to innity [22],
see Fig. 5. Enclosing the point (  1,0) is interpreted as passing to the left of the mentioned point. The locus of the function
G0 jo for three different values of the rotor angular velocity O is shown in Nyquist diagram in Fig. 5, which is plotted as an
illustrating example for K/D 100 rad/s. All the contour plots are of the same shape, but they are differing only in a scale
and correspond to the stable, steady-state and unstable vibration. When the steady-state vibration occurs, the stability
margin is achieved. In this case the locus of the G0 jo function meets the point (  1,0), therefore
G0 joCRIT 1

The complex equation (8) represents two real equations. Solution of the imaginary part of the Eq. (8) results in formula
for an angular frequency, at which a system can oscillate without damping. This frequency is designated by oCRIT . Solution

Fig. 4. Shaft and uid wedge bearing systems as a servomechanism.

ma et al. / Mechanical Systems and Signal Processing 36 (2013) 618629

J. Tu

623

Fig. 5. Nyquist diagram showing stable, margin and unstable locus.

Fig. 6. Closed control loop.

of the real part of the Eq. (8) results in the formula for mechanical resonance

o2CRIT K=M and oCRIT lO

It can be concluded that the frequency of the rotor subharmonic oscillation is the same, as the uid average
circumferential angular velocity. The measurement shows that the value of the parameter l is equal to approximately
0.475. The stability margin corresponds to the mechanical resonances of the rigid rotor mass supported by the oil wedge as
a spring. It can be noted that the frequency oCRIT is not equal to the rotor critical speed as a result of the rotor bending
when the vibration is excited by the rotor unbalance.
If the system was linear, then the unstable rotor vibration would spiral out to innity when the rotor angular frequency
crosses some threshold. The rotor angular frequency threshold is inversely proportional to the ratio l.
p
OCRIT K=M=l
10
As can be experimentally veried, the frequency spectrum of the uid-induced vibration contains the single dominating
component, which would be a solution of the second order linear differential equation without damping. The journal
lateral vibration is limited by the inner bearing surface. The stiffness and damping coefcients are non-linear functions of
the eccentricity ratio, especially when the rotor is approaching the journal wall. If the magnitude of vibration is increasing,
the oil-lm stiffness and damping increases as well. A new balance forms a steady-state limit cycle of the rotor orbital
motion.
The maximal angular velocity of journal stable behavior is given by the theoretical formula (10). This formula allows
calculating the stiffness of an oil lm for the known mass of the rotor, the lambda factor and known angular velocity of the
onset of instability. Experiments have shown that the limit angular velocity for the journal stable rotation depends on the
radial clearance of the journal inside a bushing and the viscosity of the lubricant.
4. Closed control loop
Active vibration control of journal bearings uses the bushing position as the control variable u and the shaft position as
a controlled variable r. The control variable is an output of a controller. The controller transforms an error signal computed
as a difference of a set point and actual position of the shaft. As is evident from the block diagram in Fig. 6, the controller is
of the proportional type with the gain K P .
Provided that the perturbation force is zero FP 0 the equation of motion is as follows:
M r D r_ KjDlO r D u_ KjDlO u

11

The Laplace transfer function relating the displacement of the bushing to the displacement of the shaft is given by
Ds KjDl O
M s2 Ds KjDl O
jo D KjDl O
Go jo K P
jo D KjDl OMo2

Go s K P

12

624

ma et al. / Mechanical Systems and Signal Processing 36 (2013) 618629

J. Tu

Fig. 7. Force acting at the bushing vs. bushing displacement in horizontal direction (axis X).

Fig. 8. Dependence of piezoactuator force on bushing displacement.

For the stability margin the open-loop frequency transfer function G0 o is equal to 1. The frequency of the steadystate vibration at the stability margin is given by o l O and K P o2 M=K1. If the feedback gain K P is positive then the
maximal rotational speed OMAX for the rotor stable behavior is greater than the critical rotational speed without any
control which is given by Eq. (10). Increasing of the margin for the rotational speed is given by the formula
p
OMAX OCRIT K P 1 :
13
The control system does not stabilize the behavior of the journal bearing directly by changing the position of the
bearing bushing, but indirectly by changing force that acts on this bushing. Except of the controller gain, the displacement
of the bushing depends on stiffness of its connection with the bearing body through rubber seal rings as it is shown in
Fig. 7. The dependence of force acting to the bushing on the bushing displacement is shown in Fig. 8. The working point of
the electromechanical system results from the voltage which is supplied to the piezoactuator. The gain K P in Fig. 6 results
not only from the setting up of the controller, but from the property of the bushing clamping as well. Properties of the
piezoactuator of the P-844.60 type (catalogue values) and measured stiffness of clamping (5.5  106 N/m) gives the
bushing travel range which is equivalent to the control variable range (see right part of diagram in Fig. 7). The
piezoactuators are energized by voltage ranging from 0 to 100 V which corresponds to the amplier input voltage ranging
from 0 to 12 V. The range of the rotor stable rotational speed is limited by the travel range of piezoactuators and
measurement errors of the proximity probes.

5. Operation without active vibration control

When putting the test rig into operation, we met these problems:

 choice of lubricating oil determining bearing friction looses,

 measurement accuracy of shaft position,
 mounting of piezoactuators to avoid torsional loading and enable adjusting position at the accuracy of micrometers.

ma et al. / Mechanical Systems and Signal Processing 36 (2013) 618629

J. Tu

625

5.1. Lubrication
To reach the maximum motor speed higher than 6000 rpm it was necessary to increase bearing clearance to 90 mm
(radial clearance 45 mm) with simultaneous decrease of the calculated rotor stability limit. Hydraulic oil of the VG 32 grade
(kinematic viscosity of up to 32 mm2/s at 40 1C) used as a lubricant was then substituted by bearing special oil of the
OL-P03 type (VG 10 grade, kinematic viscosity 2.54 mm2/s at 40 1C). The hydraulic oil enables to reach the maximum
rotational speed of 16,000 rpm and the instability onset at the same value of speed. The bearing special oil enables to reach
the motor maximum nominal rotational speed 23,600 rpm, while the instability onset starts at 4300 rpm. All tests were
undertaken at ambient temperature about 20 1C. Lubricating oil was not preheated during tests.
5.2. Measurement of shaft position
The choice of sensor type for small displacements is crucial to the success of experiments. The rst experiment with the
use of proximity probes, which are based on the eddy-current principle, showed the shape of the rotor centerline
trajectory (orbit) considerably differing from a circle or an ellipse. To explain this phenomenon, the reasons were looked
for in the uniformity of motor rotation, the misalignment of the motor and rotor axes, the oil pump and the interference of
the proximity probe output signals. To achieve perfect decoupling of the test shaft from driving motor, another exible
coupling was installed between original coupling and the shaft. The rotation uniformity is not as smooth as in case of using
a DC motor. According to the catalogue data for eddy-current sensors the interference between signals causes an error
below 1 mm in measurements. Inspection of the shaft non-circularity showed, that the deviation from the circle is less than
1 mm as well. Finally it was proved for the eddy-current sensors that the main source of the proximity probe periodical
error is non-homogeneity of the rotor material magnetic and electric conductivity properties of the shaft surface. The error
signal is repeated synchronously with rotor rotation. The same material is passing-by below the tip of the second
proximity probe after one quarter of the rotor revolution; therefore a phase shift of the period quarter may be discovered.
The peak-to-peak value of the regular periodic error reaches 11 mm. The spectrum of the measurement periodic error is
composed from low harmonics of the rotor rotational frequency.
The dependence of the proximity probe regular error on the rotation angle can be approximated by a sum of
trigonometric functions differing in the number of waves. The reduction of the proximity probe error by subtraction of the
error signal requires a tacho pulses measurement and prediction of the rotational angle.
As mentioned earlier, it is preferable to use a proximity sensor based on the measurement of the electrical capacitance
than to compensate predictable measurement errors. The output signal of such a capacitive sensor does not contain any
low harmonics of the shaft rotational frequency. But the output signal of the electric capacity sensor contains an error part
with frequency that is greater than 1 kHz. The amplitude of this error signal at the mentioned frequency is greater than
one micrometer. The error signal is ltered out by an analog RC lter with a cut-off frequency of 200 Hz. The effect of this
low-pass lter is shown in Fig. 9.
Comparison of the signals measured by the sensor of the capacitive type and the sensor based on the eddy-current
principle is shown in Fig. 10. The sensors are oriented in the horizontal direction (X) and the shaft rotational speed is
increased up to 7000 RPM until the onset of instability starts up. The sampling frequency was set to 2048 Hz. On the left
panel of this gure is a record of the output signal of the capacitive type sensor. The record on the right panel corresponds
to the output of the sensor based on the eddy-current principle. The shaft does not vibrate before the onset of instability,
therefore the output signal of the capacitance type sensor is without periodic oscillation of the shaft position while this
part of the record for the eddy-current sensor is corrupted by regular oscillation.
The frequency spectra of the displacement signal in the horizontal direction during the shaft run-up for both the
proximity probe type are shown in Fig. 11. The shaft rotational speed begins at 0 rpm and reaches the speed, at which
starts the instability. To compare the background noise level, the vertical scale ranges from  80 to  20 dB (reference
1 mm). As was mentioned, the output signal of the sensor based on the eddy-current contains low harmonics of the
rotational frequency. In contrast to the eddy-current sensors the noise level of the sensors based on the capacitive
principle is considerably less.
The measurements of the shaft displacement during run up proceeded with extremely low viscous oil VG10 without
preheating. It is technically impossible to increase the rotor rotational speed smoothly from 0 rpm. Rotor starts at the

Fig. 9. Frequency spectrum of the ltered and unltered output signal of the sensor based on electrical capacity (capaNCDT CS05).

ma et al. / Mechanical Systems and Signal Processing 36 (2013) 618629

J. Tu

626

Fig. 10. Measurement of the shaft horizontal displacement using the sensor based on electrical capacity (capaNCDT CS05) and the sensor based on the
eddy current (IN-085).

Fig. 11. Multispectra of the shaft horizontal displacement using the sensor capaNCDT CS05 and the sensor IN-085 (dB reference 1 mm).

speed of 230 rpm and then the speed is continuously increased up to the onset of instability. The journal movement begins
at the bottom of the bearing sleeve and with increasing speed it moves up in direction of rotation. At the level of the sleeve
centre the journal starts to move towards the bearing centre. With innite speed or zero load the journal centre
coincidences with bearing centre, which is generally unstable position in circular bearing. The instability onset is at about
4300 rpm.

5.3. Mounting of piezoactuators

Choice of the piezoactuator type was veried by measurement of the dependence of acting force on the open-loop
piezoactuator travel. As is shown in Fig. 8, force of 500 N is sufcient to overcome exibility of the sealing O rings.
Flexible tip was used to attach the piezoactuator to the bushing rod and the frame structure for compensation of
misalignment and possible bending load.
The test rig instrumentation allows active vibration control only in the journal bearing at the opposite side to the
driving motor. Before beginning of operational tests, the initial position of the piezoactuators has to be adjusted in the
middle position of the operating travel range. This position corresponds to half the output voltage of the controller, the full
range of which is equal to 12 V. A screw at a piezoactuator holder is tightened in this position. The range of the shaft
displacement for the full scale of the controller output voltage is shown in Fig. 7 for the horizontal (Axis X) direction of the
shaft displacement.

6. Active vibration control

The test rig was designed for speeds up to 23,000 rpm. This paper is focused to prove safely and without doubt the
stabilization effect of the active vibration control system. The speed range of the rotor during tests results from the journal
clearance and the oil viscosity at 20 1C which were at disposal.

ma et al. / Mechanical Systems and Signal Processing 36 (2013) 618629

J. Tu

627

As was mentioned earlier, the signal from the proximity probes is connected to the dSpace signal processor. The output
of the signal processor is connected to the input of the amplier that powers the piezoactuator. The electronic feedback
(see Fig. 12) in the below presented experiments was of the proportional controller type. Although improved dynamic
properties of the control loop require adding a derivative component, the noisy signal produced by the proximity probes is
the reason, for which the derivative feedback was not used [23]. Even if the sensors based on the electrical capacity
principle have a smaller error than the eddy current ones, the active vibration control has been tested with sensors based
on eddy currents.
The shaft rotational speed during the tests under active control (ON) and without active control (OFF) was increasing at
the ramp rate of 7000 rpm per a minute for all measurements. The onset of instability starts at 4300 rpm for the oil of the
VG 10 grade. Because the piezoactuator travel range cannot cover the change of the shaft position from the very beginning
up to the level of the bushing centre position, the active vibration control is switched ON when the shaft lifts up into the
stabilized position, which corresponds approximately to 3000 rpm. Due to the measurement error the controller output
voltage starts to oscillate with a limited magnitude. As is clear from Fig. 13, if the active control is switched ON during the
rotor run-up, the onset of instability is changed to 7300 rpm. This increasing of the limit speed corresponds to the
controller gain K P  2. The result of measurements at half the open-loop gain (50%) is shown in the middle part of Fig. 13.
The onset of instability occurs at about 6200 rpm. The active vibration control is immediately switched OFF after starting
the unstable vibration with frequency, which is 0.475 multiple (factor l) of the shaft rotation frequency. This phenomenon
is called whirl due to the oil lm and is different from the rst natural frequency of the shaft, which is equal to a constant
value. The output of the signal processor is saturated on the full voltage range from 0 to 12 V.

Fig. 12. Active vibration control system.

Fig. 13. Time history of the rotor rotation speed when the active vibration control is ON and OFF.

628

ma et al. / Mechanical Systems and Signal Processing 36 (2013) 618629

J. Tu

As is demonstrated in Fig. 13, the active vibration control signicantly extends the range of operating rotational speed.
With active control ON, onset of instability is increased by about 3000 rpm in comparison with the operating range
without the active vibration control. The electronic feedback is clearly seen as a complementary way to the traditional
journal bearing design modications and other tools, which prevent instability or shift the rotor instability onset to higher
rotational speed.
Experiments show that a movement of the bushing is transferred to the movement of the bearing journal. The oil
wedge, which is created by pumping the oil below the journal, is stiff enough to be able to convey a motion of the bushing
to the motion of the bearing journal. The operation of the control loop is the best proof that the movement of the bushing
is transferred to the movement of the bearing journal. Closed loop control shows that the journal vibration in a wide speed
range is of low amplitude without knocking the journal at the bushing wall. The other important result of experiments is
that the journal bearing behavior may be described by a linear model of the journal bearing for small uctuations around
the steady-state position.
It seems that parametric dampers whose stiffness alternates periodically in time are the future for vibration damping of
rotors. Research in this area is being conducted at the TU Darmstadt, Germany [24,25], and Vienna University of
Technology, Austria [1]. The laboratory test rig which was built by the authors of this article may help in research of this
type of vibration damping for instability of the rotors due to the oil lm.

7. Conclusion
The lumped parameter model of the journal motion inside the hydrodynamic bearing is based on the concept
developed by Muszynska. According to the mentioned approach the bearing forces are modeled by a spring and damper
system. This system is rotating at the angular velocity, which is a constant fraction of the rotor rotational speed. The
equation of motion contains the complex vector and parameters. Mathematical model allows predicting the behavior of
the rotor, which conrms its suitability for the analysis of the control system.
The test rig for experimental investigation of affecting behavior of the rotor supported in sliding bearings by external
excitation was put into operation. The rotational speed of the rotor is up to 23,000 rpm. The test rig consists of a rigid shaft
of 30 mm diameter supported in two cylindrical hydrodynamic journal bearings with the span of 200 mm. Standard
behavior of the rotor was achieved with low viscosity oil, with which the oil lm had insufcient load capacity to shift
journal centre into unstable position at the bearing centre. The proposed goal of the project was achieved through
controlled movement of only one of two bearing bushings. The bearing bushing was being moved by means of two
piezoactuators oriented in vertical and horizontal directions. The shaft rotational velocity, at which occurs the onset of
instability, was substantially increased. It seems, that there is a large potential for further improvements, which could lead
to active control of high-speed rotor behavior in real operating conditions.

Acknowledgment
This research has been supported by the Czech Grant Agency project No. P101/12/2520 Active vibration damping of
rotor with the use of parametric excitation of journal bearings.
References
[1] H. Ecker, A. Tondl, Increasing the stability threshold of a rotor by open loop control of the bearing mount stiffness, in: Proceedings of the ISCORMA-3,
Cleveland, Ohio, 2005.
[2] F. Lebo, S. Rinderknecht, Modellbasierte Regelung eines Skalierten Elastischen Flugtriebwerkrotors mit Piezostapelaktoren, in: Proceedings of the
SIRM 20119. Internationale Tagung Schwingungen in rotierenden Maschinen, Darmstadt, Deutschland, 2123, Februar 2011 Paper ID-27.
[3] F. Dohnal, R. Markert, T. Hilsdorf, Enhancement of external damping of a exible rotor in active magnetic bearings by time-periodic stiffness
variation, in: Proceedings of the SIRM 20119. Internationale Tagung Schwingungen in rotierenden Maschinen, Darmstadt, Deutschland, 2123.
Februar 2011 Paper ID-08.

[4] S. Furst,
H. Ulbrich, An active support system for rotors with oil-lm bearings, Proc. IMechE Ser. C (1988) 6168. paper 261/88.
[5] H.Y. Lau, K.P. Liu, W. Wang, P.L. Wong, Feasibility of using giant magnetostrictive material (GMM) based actuators in active control of journal
bearing system, in: Proceedings of the World Congress on Engineering 2009 vol II, WCE 2009, July 13, 2009, London, UK.
[6] J. Zapomel, P. Ferfecki, Application of controllable magnetorheological dampers for reducing lateral vibration of rotors excited by a base plate
oscillationa computational investigation, in: Proceedings of the SIRM 20119, Internationale Tagung Schwingungen in rotierenden Maschinen,
Darmstadt, Deutschland, 21 23, Februar 2011 Paper ID-51.
[7] A.B. Palazzolo, R.R. Lin, R.M. Alexande, A.F. Kascak, G. Montague, Test and theory of piezoactuatorsactive vibration control of rotating machinery,
ASME Trans. J. Vib. Accoust. 113 (2) (1991) 167175.
[8] C. Carmignani, P. Forte, E. Rustighi, Active control of rotor vibrations by means of piezoelectric actuators, in: Proceedings of the DETC2001 18th
Biennial Conference on Mech Vibration and Noise, Pittsburgh, Pennsylvania, 2001.
[9] B.-H. Rho, K.-W. Kim, The effect of active control on stability characteristics of hydrodynamic journal bearings with an axial groove, Proc. Inst. Mech.
Eng. Part C: J. Mech. Eng. Sci., vol. 216 (2002).
[10] S.J. Kim, J.D. Jones, Inuence of piezo-actuator thickness on the active vibration control of a cantilever beam, J. Intell. Mater. Syst. Struct. 6 (1995) 61.
[11] T. Ushijima, S. Kumakawa, Active engine mount with piezo-actuator for vibration control, SAE International Congress & Exposition, 1993.

[12] A. Alizadeh, C. Ehmann, U. Schonhoff,

R. Nordmann, Active bearing of rotors utilizing robust controlled piezo actuators, in: 19th Biennial Conference
on Mechanical Vibration and Noise, Parts A, B, and C. 5, 2003.

ma et al. / Mechanical Systems and Signal Processing 36 (2013) 618629

J. Tu

629

[13] J. Tuma, J. Skuta,

R. Klecka, J. Los, J. Simek,
A Laboratory test stand for active control of journal bearings, in: Proceedings of the Colloquium Dynamics
of Machines 2010, Inst. of Thermomechanics, Prague, February 23, 2010, pp. 95100.

[14] J. Simek,
J. Tuma, J. Skuta,
R. Klecka, Unorthodox behavior of a rigid rotor supported in sliding bearings, in: Proceedings of theColloquium Dynamics
of Machines 2010, Inst. of Thermomechanics, Prague, February 23, 2010, pp. 8590.
[15] J.W. Lund, Review of the concept of dynamic coefcients for uid lm journal bearings, Trans. ASME J. Tribol. 109 (1987) 3741. January.
[16] J.W. Lund, Spring and damping coefcients for the tilting pad journal bearings, Trans. ASLE 7 (1964) 342352.

Turbomaschinen, Dissertation, Fakultat

fur
Maschinenbau, Technischen
[17] L. Malcher, Die Federungs- und Damfungseigenschaften
von Gleitlagern fur
Hochschule Karlsruhe, 1975.
[18] A. Muszynska, Whirl and whiprotor/bearing stability problems, J. Sound Vib. 110 (3) (1986) 443462.
[19] D.E. Bently, A. Muszynska, Fluid-generated instabilities of rotors, Orbit 10 (I) (1989). April.
[20] A. Tondl, Quenching of Self-Excited Vibrations, Academia, Prague, 1991.

[21] J. Tuma, J. Simek,

A. Vtecek, Simulation study of a rotor system response to kinematic perturbation, Acta Mechanica Slovaca, 3/2008.
[22] R. Burns, Advanced Control Engineering, Butterworth Heinemann, Oxford, 2001.

[23] A. Vtecek, J. Tuma, M. Vteckova, Stability of rigid rotor in journal bearing. Transactions of the VSBTechnical
University of Ostrava. Mechanical
Series No. 2. 2008, LIV, paper 1638, pp. 159164.
[24] A.A.G. Siqueira, R. Nicoletti, N. Nicklas, et al., Linear parameter varying control design for rotating systems supported by journal bearings. J. Sound
Vib. 331 (10) (2012) 22202232.
[25] F. Dohnal, Experimental studies on damping by parametric excitation using electromagnets. Proc. Inst. Mech. Eng. Part CJ. Mech. Eng. Sci. 226 (C8)
(2012) 20152027.