ABCM Symposium Series in Mechatronics - Vol.

6 Part I - International Congress

Copyright © 2014 by ABCM Section VII - Vibration & Magnet Bearings

MAGNETIC BEARING WITH UNIAXIAL CONTROL USING MAGNETIC,

ELECTRODYNAMIC AND ELECTROMAGNETIC LEVITATION

Rogério Issamu Yamamoto

Oswaldo Horikawa
ESCOLA POLITÉCNICA DA UNIVERSIDADE DE SÃO PAULO, SÃO PAULO - SP - Brazil
[Link]@[Link]
ohorikaw@[Link]

Isaias da Silva
UNIVERSIDADE FEDERAL DE SÃO PAULO, DIADEMA - SP - BRASIL
[Link]@[Link]

Abstract. Magnetic bearings use magnetic force to sustain loads without any contact between the rotor and the
bearing, therewith there is no friction or wear. Due to these advantages, these bearings are applied in wide range of
applications, from flywheel energy storage systems to artificial hearts. In the first, the magnetic bearing reduces
energy loss by friction, raising the performance, and in the second, the magnetic bearing minimizes the damage to
blood cells. Many types of magnetic bearings are known, each one based on different levitation techniques, e.g. based
on the use of superconductive materials. However, the superconductivity is obtained only at temperatures around 100
K, imposing limitations for practical applications at room temperature. A more promising technique is the
electrodynamic levitation at room temperature. The repulsion force for the levitation is generated by the relative
movement between a magnetic field and a conductor. The inconvenient of this technique is that high heat rates are
dissipated in the conductor. In order to reduce the heat generation a well-designed electrodynamic suspension must be
accomplished. Another levitation technology is based on electromagnetic forces, i.e. the levitation force is obtained by
electromagnetic coils. Based on readings of position sensors the current to the coils is regulated keeping the object in a
fixed position. The main disadvantage of the electromagnetic levitation is its construction complexity because it
requires a sensor, a controller, a power amplifier and an actuator for each degree off freedom of the system. In this
context, focusing applications like flywheel energy storage system in which the rotor starts from zero and reaches
50.000 rpm order speeds or even higher, this work presents a magnetic bearing that combines three levitation
techniques. The first one is based on permanent magnets. This provides stable levitation forces in the radial direction
of the rotor. However this technique does not assure stability of the rotor position along its axial direction. To solve
this problem, a second technique, the electrodynamic levitation at room temperature based on permanents magnets is
applied which supplies damping for the bearing system, contributes to the rotor stability along its axial direction and
reduces the electromagnetic levitation energy consumption and the third is an active electromagnetic bearing which
levitate the rotor at low speed and assist the electrodynamic bearing in axial stability. The proposed hybrid bearing is
simpler than a bearing whose rotor is sustained only by electromagnetic bearings in 5 degrees of freedom, therefore
reducing the energy consumption when using only electromagnetic bearings and can work at room temperature
without a refrigeration system. This work presents the principle of the new hybrid bearing, the development steps of
this bearing besides showing both computer simulated numerical and experimental results.

Keywords: Magnetic Bearing; Flywheel; high speed rotors; electrodynamic levitation

1. INTRODUCTION

Magnetic bearings are devices which use the magnetism to sustain a rotor in space, so there is no contact between
the rotor and the bearing. Such bearings has been applied on system like flywheels or artificial hearts in which it is
necessary or even imperative the reduction or elimination of the friction, at the first case the objective is to increase the
device efficiency and in the second case the bearing must avoid the damage caused on blood cells (HAMLER, et al.,
2004) e (HORIKAWA, et al., 2008).
Magnetic bearings can be classified in two large groups, i.e. passive and active bearings. Passive bearings do not
require controllers or actuators to stabilize a rotor. On the other hand active type bearings need controllers, sensors and
actuators to keep the rotor stability (SCHWEITZER, 2009).
Among the passive bearings there are the following types: permanent magnet bearings, superconductor bearings and
electrodynamic bearings, and among the active bearings the best known is the active electromagnetic bearing type.
Passive bearings are only based on static and constant magnetic field, i.e. there isn’t any control system to control
the magnetic force, therefore its feasibility to simultaneously stabilize the rotor in all degree of freedom are limited and
they are generally used for very specific applications, e.g. when the required bearing stiffness and damping are low.
Passive magnetic bearings can use different concepts to generate the levitation force which sustain the rotor weight,

815
 ABCM Symposium Series in Mechatronics - Vol. 6 Part I - International Congress
Copyright © 2014 by ABCM Section VII - Vibration & Magnet Bearings

among this concepts three types can be highlighted: permanent magnet, superconductor and electrodynamic levitation
(SCHWEITZER, 2009).
Permanent magnet bearings only use permanent magnets to generate levitation force, the concept applied in these
type of bearings is the simple repulsion or attraction forces, which appear when two or more magnets are placed in the
field of each other. These kinds of bearings present low cost and a long life cycle, basically they can be made by two
magnetic rings, generally made of NdFeB alloy, due the magnetic characteristics of this compound, and they can be
arranged in several ways, like those present in Fig. 1 (YONNET, 1978).

Figure 1: Four basic arrangements of passive magnetic bearings – (a) and (b) attraction type (c) and (d) repulsion type
(YONNET, 1978)

However, according to Earnshaw’s theorem (EARSHAW, 1848), the stability in all degrees of freedom cannot be
achieved using only permanents magnets, nonetheless these bearings can be applied to control some degrees of freedom
of a body or used as auxiliary bearings reducing passive loads on the main bearing (MOSER, et al., 2006) and
(YONNET, 1978).
Superconductors bearings use special materials called superconductors which present a unique feature called
superconductivity, i.e. such materials virtually have no electrical resistance and their relative magnetic permeability
becomes null when they are subjected to temperatures around 100 K, called critical temperature, TC. According to the
Meissner-Oschenfeld effect these materials, when submitted to an external magnetic field and when the temperature is
below the TC, tend to eliminate the field inside the material, as shown in Fig. 2, so generating a repulsion force,
(SCHWEITZER, 2009) and (CALLISTER, 2002).

Figure 2: (a) superconductor material above the critical temperature and (b) below the critical temperature (T c)
(RODGERS, et al., 2004)

The main disadvantage of these bearings is the condition to the Meissner-Oschsenfeld effect works, i.e. the material
must be below the critical temperature (TC) so that the material becomes superconductor. Therefore, the application of
these types of bearings requires the use of an efficient cooling system which must be linked to the superconductor
material, (HAMLER, et al., 2004) and (NICOLSHY, et al., 2000).
Electrodynamic bearings include a class of devices based on the principle of electrodynamic repulsion, which in turn
is based on the Lenz Law, i.e. according to this law "an electric current induced in a conductor always appears to
oppose the variation of the phenomenon that caused it". In fact this means that any variable magnetic flux will be
reflected when focused on a conductor surface generating repulsion and drag forces which are normal and tangential to
the conductive surface, respectively (FURLANI, 2001). Figure 3 shows an example of a simple electrodynamic
levitation system that uses an aluminum disc and a permanent magnet piece. When the disc rotates with an angular
speed that causes a tangential velocity to be higher than a certain critical velocity, the magnetic fields interaction, i.e.
the magnet field and the disc induced field, generates two forces, a normal force (F1) and a tangential force (F2). The
normal force is responsible for the levitation and the tangential force called the drag force causes heat dissipation in the
disc.

816
 ABCM Symposium Series in Mechatronics - Vol. 6 Part I - International Congress
Copyright © 2014 by ABCM Section VII - Vibration & Magnet Bearings

F1 Magnet

N
F2
S

Metallic
Disc
ω
y

R x

Figure 3: An electromagnetic levitation simple scheme.

The levitation and the drag force are dependent mainly on the frequency of the magnetic flux, i.e. on high
frequencies levitation force dominates and in low frequencies the drag force predominates, as can be seen in Fig 4
(STEPHAN, et al., 2009).

Figure 4: Levitation and drag forces characteristics in an electrodynamic suspension (AMANTI, et al., 2008)

The electrodynamic bearings have relatively simple construction architecture when compared to electromagnetic
bearings, however they have both low stiffness and damping compared to the active bearings and their stability is
ensured only at high speeds (AMANTI, et al., 2008).
Electromagnetic bearings use the force produced by electromagnetic fields to actively control the dynamics of an
electromagnet, so the control system acts on the electromagnet current generating an electromagnetic force that keeps
the rotor in a specific position. Figure 5 illustrates a radial electromagnetic bearing schema, in this figure is indicated
the key components of an active electromagnetic bearing (SCHWEITZER, 2009). Referring to the figure can be noted
that to control one degree of freedom it is necessary a sensor to measure the displacement of the rotor and a
microprocessor, which has the control logic. The input signal of the microprocessor is provided by the sensor and the
output signal feeds a power amplifier which is used like an interface between the microprocessor and the electromagnet,
i.e. converts electric current to electromagnetic force applied to levitate the rotor. The control logic ensures the stabilit y
and damping for the rotor. These characteristics can be adjusted by the controller parameters (SCHWEITZER, 2009).

Sensor

Electromagnet

Rotor Micro-
processor

Amplifier

Figure 5: Radial electromagnetic bearing scheme (SCHWEITZER, 2009)

817
 ABCM Symposium Series in Mechatronics - Vol. 6 Part I - International Congress
Copyright © 2014 by ABCM Section VII - Vibration & Magnet Bearings

2. PROPOSED HYBRID TYPE BEARING

The proposed bearing is a magnetic bearing of hybrid type; therefore it is a magnetic bearing which uses more than
one levitation concept. The rotor levitation along its radial direction is achieved by two passive radial permanent
magnet bearings placed at the rotor’s end and that provide, in the adopted magnets arrangement, high positive radial
stiffness, however the axial stiffness is negative. So, in order to partially overcome this negative axial stiffness and also
to introduce some radial damping to the rotor an electrodynamic bearing is used. However, as stated before,
electrodynamic bearings only work when the relative speed between the magnetic field and the conductor is higher than
a critical speed. So, in order to provide the rotor stabilization at low speeds, allowing the application of the
electrodynamic bearing, an axial type electromagnetic bearing is used, comprising the hybrid type proposed bearing.
Figure 6 depicts the proposed hybrid magnetic bearing scheme.

Figure 6. Proposed hybrid magnetic bearing scheme

2.1 RADIAL PASSIVE MAGNET BEARING

The adopted radial passive magnet bearing is depicted in Fig. 7. This configuration was chosen after computer
simulation of some magnet ring arrangements using finite element method. This radial bearing has a negative axial
stiffness, ka, and a positive radial stiffness, kr, and the stiffness are related according the following equation (YONNET,
1978)

ka  2kr (1)

kr > 0

ka < 0
Ø15

Ø20

Ø40
Ø8

NdFeB Permanent S N S
magnets
10 10

Figure 7: Magnetic ring arrangements for the passive radial bearings used through this work (dimensions in mm)

818
 ABCM Symposium Series in Mechatronics - Vol. 6 Part I - International Congress
Copyright © 2014 by ABCM Section VII - Vibration & Magnet Bearings

Using the finite element method and considering the magnets dimensions and material shown in Fig. 7, some
numerical results were obtained. Figure 8 shows the magnetic forces as function of the inner magnets displacement
along the radial and axial directions. Through these figures one can see that the axial and the radial stiffness are
and , respectively. As predicted in Eq.1, radial stiffness is twice in magnitude the
axial stiffness. In the proposed bearing two pair of ring type magnets are used, one at each rotor extremities as shown in
Fig. 6. Thus, the rotor total stiffness are and .

50
Magnetic radial force [N]

40

30

20

10

0 Radial stiffness kr ~ 20 N/mm

-10
0 0.5 1 1.5 2
Radial displacement [mm]
a) Magnetic radial force versus radial displacement
0
Axial stiffness ka ~ -40 N/mm
Magnetic axial force [N]

-20

-40

-60

-80
0 0.5 1 1.5 2
Axial displacement [mm]
b) Magnetic axial force versus axial displacement

Figure 8 – Radial (a) and axial (b) forces generated by the passive radial magnetic bearing

2.2 ELECTRODYNAMIC BEARING

The electrodynamic bearing is used to improve positive axial and radial stiffness as well as damping as the rotor
angular speed increases. The main problem associated with these bearings is the high heat generation as consequence of
the eddy currents induced in the bearing conductor part. There are many works about electromagnetic levitation that
propose solutions to the heat losses (FILATOV, et al., 2002), (FILATOV, et al., 2002) e (LEMBKE, 2003). Among
these works, the so called homopolar bearing architecture has the simplest configuration and presents the lowest heat
generation. This is due to the uniform distribution of the magnetic field around the conductor, i.e. no eddy current is
induced in the conductor when it is rotating without radial or axial oscillations relatively to the uniform magnetic field
center. Therefore, the forces and damping are generated only when the rotor displaces from the field center. So,
following the homopolar bearing concept, a radial and axial electrodynamic bearing was designed. Figure 9 shows four
coaxial magnet rings working in attraction mode assembled in two steel discs. An aluminum disc fixed to a rotor is
placed between the magnets. This disc is the electrodynamic suspension conductor.
The steel discs are used in the electrodynamic bearing to close the magnetic circuit, increasing the magnetic flux that
cross through the aluminum disc. So, when the rotor’s angular speed is higher than a critical angular speed and, when it
is displaced away from its centered position, an electric current is induced in the aluminum disc, generating axial, radial
and tangential forces and also some damping. The axial force contributes to compensate the axial negative stiffness
effect, the radial force and the damping increase the passive radial stiffness and reduce rotor’s radial oscillations and the
tangential or drag force causes a torque that tries to reduce rotor’s angular speed. Figures 10, 11 and 12 depict the
electrodynamic radial and axial forces, the drag torque and the solid losses, respectively, obtained by computer
simulation using the configuration shown in Fig. 8 and considering a rotor radial and axial displacement of 0.2 mm and
a rotor angular speed of 2500rpm. Figure 10 shows that the electrodynamic stiffness for these operating conditions are
approximately , and for axial direction and for radial directions 1 and 2, respectively.
The reason that the radial stiffness values are different from each other is because the rotor radial displacement was
applied only along radial direction 1. Figure 11 depicts the drag torque that acts on the bearing conductor disc for the
mentioned operating conditions. Observing this figure, one can see that the drag torque is very small, this agrees with

819
 ABCM Symposium Series in Mechatronics - Vol. 6 Part I - International Congress
Copyright © 2014 by ABCM Section VII - Vibration & Magnet Bearings

the results presented in Fig. 12 which shows that the solid losses in the conductor disc is about 180mW. These losses
are considered satisfactory for the present operating conditions and bearing dimensions.

Figure 9: Radial and axial electrodynamic bearing scheme

3
Force [N]

1 Radial force 1
Radial force 2
Axial force
0
0 5 10 15 20
Time [s]

Figure 10: Electrodynamic bearing forces as function of time (rotor’s angular speed of 2500 rpm and rotor axial and
radial displacement of 0.2 mm)

-3
x 10
2
Drag torque [Nm]

-1

-2
0 5 10 15 20
Time [s]

Figure 11: Electrodynamic bearing drag torque as function of time (rotor’s angular speed of 2500 rpm and rotor axial
and radial displacement of 0.2 mm)

0.2
Solid losses [W]

0.15

0.1

0.05

0
0 5 10 15 20
Time [s]

820
 ABCM Symposium Series in Mechatronics - Vol. 6 Part I - International Congress
Copyright © 2014 by ABCM Section VII - Vibration & Magnet Bearings

Figure 12: Electrodynamic bearing solid losses as function of time (rotor’s angular speed of 2500 rpm and rotor axial
and radial displacement of 0.2 mm)

2.3 AXIAL ELECTROMAGNETIC BEARING

The active electromagnetic bearing is responsible to compensate for the negative axial stiffness of the passive
magnet bearing, therefore allowing the hybrid bearing operating at low and high rotation speeds. This bearing is formed
by a gap sensor, a controller, an amplifier and an electromagnet actuator, as illustrated in Fig. 13. The actuator shown in
Fig. 13 were simulated and the results presented in Fig 14. The bearing is at its nominal operating point when the gap
between the actuator face and the steel disc fixed to the rotor’s ends is 0.3 mm. This value was chosen based on the
value of the passive axial stiffness for the whole bearing . To overcome this negative stiffness, the
actuator force/current and also the force/displacement factors must be positive and higher in modulus than this value.
Adopting the values depicted in Fig. 14, the designed actuator is capable of stabilize the rotor axially.

Figure 13: Axial electromagnetic actuator

Electromagnetic force [N]

150
Force/current factor ki ~ 70 N/A at 0,3 mm gap

100

50

0
0.5 1 1.5 2
Current [A]
a) Electromagnetic force versus current determined for 0.3 mm gap
100
Force/displacement factor for 0,5 A kd~85 N/mm
Electromagnetic force [N]

80 Force/displacement factor for 0,75 A kd~155 N/mm

Force/displacement factor for 1 A kd~170 N/mm
60

40

20

-20
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Gap [mm]
b) Electromagnetic forces versus gap

Figure 14: Electromagnetic forces generated by the axial electromagnetic actuator

2.4 BEARING CONTROL SYSTEM

As stated before, the rotor is unstable in the axial direction, i.e. the magnetic axial stiffness ka is negative. In order to
stabilize the rotor along the axial direction, a closed loop control system must be implemented. Figure 15 shows the
closed loop system block diagram. Considering the bearing dynamic model shown in Fig. 16, the transfer function that

821
 ABCM Symposium Series in Mechatronics - Vol. 6 Part I - International Congress
Copyright © 2014 by ABCM Section VII - Vibration & Magnet Bearings

describes the bearing and the rotor axial dynamics are derived. Using Newton second law, assuming that the rotor axial
displacement is enough small, and assuming null initial conditions, the axially controlled magnetic bearing transfer
function is given by the following equation:

ki
X ( s) m
G( s)   (2)
I ( s) ( 2 k a  kd )
s2 
m
Where , , , , and are the rotor axial displacement from its nominal position, the instantaneous
current in the electromagnetic actuator, the force/current factor, the magnetic bearing passive axial stiffness, the
force/displacement factor and the rotor mass, respectively. Eq. (2) denominator is the characteristic equation of
and is used to determine both the system fundamental frequency and its stability. So, finding the roots of Eq. (2) yields:

( 2 ka  kd )
s (3)
m

This equation shows that the system has one root located in the right side of the complex plane, i.e. the system is open
(2 ka  kd )
loop unstable. Also de system fundamental frequency is n  , . A classical digital PID
m
controller algorithm is used to stabilize the rotor axially.

Rotor axial reference Computer Rotor real axial

position Xr(s) position X(s)
Electromagnetic
PID K Actuators Rotor
m
Controller
mp
Amplifier
K [V/V]

Sensor
ks=5000 [V/m]

Figure 15: Block diagram for the axially controlled magnetic bearing

ki
FMag (Δx) = ka Δx y

Rotor FElectromag (Δi) = ki Δi x

m z
ω c
FElectromag (Δx) = kd Δx

Axial Electromagnetic
kr actuator
Radial Magnetic
Bearing

c - Damping provided by the controller

Figure 16: Axially controlled magnetic bearing dynamic diagram (half bearing only)

822
 ABCM Symposium Series in Mechatronics - Vol. 6 Part I - International Congress
Copyright © 2014 by ABCM Section VII - Vibration & Magnet Bearings

3. EXPERIMENTAL RESULTS

In order to demonstrate the effectiveness of the proposed hybrid magnetic bearing, a prototype was built, Fig. 17.
The prototype main parts were made of aluminum alloy. The rotor was made of stainless steel and the electromagnetic
actuator core and the rotor ends discs of SAE 1020 steel. All permanent magnets are NdFeB alloy. The rotor was
accelerated using two air nozzles that applies compressed air against grooves machined in its central position.

Passive permanent
magnet bearing

Air actuator
Electromagnetic bearing
Electromagnetic bearing

Figure 17: The prototype without the electrodynamic bearing

The first experimental analyses were done without the electrodynamic bearing, i.e. only with the radial passive
magnetic bearings and the axial electromagnetic bearings. Afterwards, the electrodynamic bearing was incorporated to
the prototype. Figure 18 presents the prototype with the three types of bearings. The rotor axial and radial displacements
were measured by noncontact gap sensors and its angular speed by a noncontact capacitive sensor which reads two teeth
machined in the steel disc used as the electromagnetic actuator target. Due to practical reasons, two rotor angular speeds
were adopted for the experiments: 1500 and 1900 rpm.

Passive permanent
magnet bearing

Electromagnetic bearing Electrodynamic Electromagnetic bearing

bearing

Figure 18: The prototype with three bearings

Figure 19 shows the rotor radial displacements at 1500 and 1900 rpm with and without the electrodynamic bearing.
Observing this figure, one can see that the rotor radial oscillations are reduced by almost 50% and 28% when the rotor
angular speeds are 1500 and 1900 rpm, respectively. Figure 20 depicts the rotor axial displacements. Also here, the
amplitude of oscillations is attenuated when electrodynamic bearing is used. These results demonstrate the effectiveness
of the electrodynamic bearing to improve both the bearing passive stiffness and the rotor damping.

823
 ABCM Symposium Series in Mechatronics - Vol. 6 Part I - International Congress
Copyright © 2014 by ABCM Section VII - Vibration & Magnet Bearings

Figure 19: Rotor radial displacements at 1500 rpm and 1900 rpm with and without electrodynamic bearing
Rotor axial displacement [mm]

0.1

0.05

-0.05
At 1500rpm without electrodynamic bearing
At 1500rpm with electrodynamic bearing
-0.1 At 1900rpm without electrodynamic bearing
At 1900rpm with electrodynamic bearing
-0.15
0 0.02 0.04 0.06 0.08 0.1
Time[s]

Figure 20: Rotor axial displacements at 1500 rpm and 1900 rpm with and without electrodynamic bearing

4. CONCLUSION

This work presented a new concept of hybrid magnetic bearing which can be applied in several mechatronic
systems, e.g. electromechanical batteries called flywheels, artificial hearts pumps, turbomolecular pumps etc. The
proposed bearing architecture was based on three magnetic bearings concept: i) passive magnet bearing, ii)
electromagnetic active bearing and iii) electrodynamic bearing. The radial passive magnetic bearing assures a high
radial stiffness. The negative axial stiffness of the passive bearing is compensated by the active axial electromagnetic
bearing. Finally, the electrodynamic bearing contributes to increase the rotor passive axial and radial stiffness and also
improving radial damping. Numerical simulations provided several data, which assisted the development of the
proposed bearing and the experiments demonstrated, as expected, that the electrodynamic bearing improves the rotor
radial stability and also reduces its radial and axial oscillations. Despite these improvements efforts will be continued to
achieve more intensive reduction in the rotor oscillation. Also, the dynamic behavior of the rotor as well as the axial
direction control analyses using the proposed hybrid magnetic bearing under higher rotor angular speeds will be
analyzed in a future work.

5. ACKNOWLEDGMENT

This study was supported by “Fundação de Amparo à Pesquisa do Estado de São Paulo, FAPESP, Proc.
2011/11638-0” and it is developed at Federal University of São Paulo, UNIFESP and at the University of São Paulo,
USP.

6. REFERENCES

AMANTI N., LÉPINE X. and TONOLI A. Modeling of Electrodynamic Bearing [Journal] Journa of Vibration and
Acoustics. - 2008. - Vol. 130.

824
 ABCM Symposium Series in Mechatronics - Vol. 6 Part I - International Congress
Copyright © 2014 by ABCM Section VII - Vibration & Magnet Bearings

CALLISTER W. D. Jr. Propriedades Magnéticas: Supercondutividade [Book Section] Ciências e Engenharia dos
Materiais. - Rio de Janeiro : LTC, 2002.
EARSHAW SAMUEL. On the Nature of the Molecular Forces which Regulate the Constitution of the Lumiferous
Ether [Journal] Transactions of the Cambridge Philosophical Society. - 1848. - Vol. 7. - p. 97.
FILATOV A. V., MASLEN E. H. and GILLIES G. T. Stability of an Electrodynamic Suspension [Journal] Journal
of Applied Physics. - 2002. - pp. 3345-3353.
FILATOV A. V., MASLEN E.H. and GILLIES G. T. A method of noncontact suspension of rotating bodies using
electromagnetic forces [Journal] // Journal of Applied Physics. - 2002. - Vol. 91. - pp. 2355-2371.
FURLANI E. P. Permanent Magnet and Electromechanical Devices [Book]. - San Diego : Academic Press, 2001.
HAMLER A [et al.] Passive magnetic bearing [Journal] // Journal of Magnetism and Magnetic Materials. - 2004. -
Vol. 272. - pp. 2379-2380.
HORIKAWA O. [et al.] Magnetic Suspension of the Rotor of a Ventricular Assist Device of Mixed Flow Type.
[Journal] // Artificial Organs. - 2008. - pp. 32-334.
LEMBKE T. Induction Bearings - A Homopolar Concept for High Speed Machines [Journal] Dep. of Electrical
Machines and Power Electronics, Royal University of Technology. - 2003.
MOSER R., SANDTNER J. and BLEULER H. Optimization of Repulsive Passive Magnetic Bearings [Journal]
IEEE Transactions on Magnetics. - 2006. - 8 : Vol. 42. - pp. 2038-2042.
NICOLSHY R. [et al.] Development of hybrid bearing system with thrust superconducting magnetic bearing and
radial active electromagnetic bearing [Journal] Physica C. - 2000. - pp. 2509-2512.
PAULA A. G. and VIANNA D. M. Levitação Eletrodinâmica - O Ensino de Física, Baseado no Enfoque CTS, Na
discussão para Melhoria da Qualidade do Nosso Ar [Conference]. - São Luís : SNEF, 2007.
RODGERS V. G. J. and SOMANTRI L. Induced Magnetization, Fighting its Creator [Online]. - 2004. - january 27,
2011. - [Link]
SCHWEITZER G. Introduction and Survey: 1.4 Classification of Magnetic bearings [Book Section] Magnetic
Bearings: Theory, Design, and application to Rotating Machinery / ed. Schweitzer Gerhard and Maslen Eric H.. -
Zürich/Florianópolis e Charlottesville : Springer, 2009.
STEPHAN R. M. and DAVID E. G. Implantação da Tecnologia MagLev-Cobra no Campus da UFRJ [Conference].
- São Paulo : CONINFRA 2009 - 3º Transportation Infrastructure Conference., 2009.
YONNET J. P. Passive magnetic bearing with permanent magnets [Journal] IEEE Transactions on Magnetics. -
September 1978. - 5 : Vols. MAG-14. - pp. 803-805.

7. RESPONSIBILITY NOTICE

The authors are the only responsible for the printed material included in this paper.

825