THIRD GENERATION OF FLYWHEELS: A PROMISING SUBSTITUTE TO

BATTERIES
Richard M. Stephan Rubens de Andrade Jr. Guilherme G. Sotelo
LASUP/UFRJ P.O. Box 68504 21945-970 Rio de Janeiro BRAZIL
rms@ufrj.br randrade@dee.ufrj.br sotelo@coe.ufrj.br

Abstract – A review of flywheels as energy storage II. FLYWHEELS’ IMPORTANCE

systems is organized in three generations. The
performance is compared with that of batteries and other This section presents the comparison between flywheels
energy storage systems in terms of power and energy and several techniques to storage energy.
density. It is shown that a new generation, using magnetic A. Energy Density
bearings and high speeds (>60000rpm) can substitute
batteries in the near future. A laboratory prototype Energy density means the ratio between the stored energy
illustrates the main parts of a 3rd Generation flywheel. and the volume or weight of the equipment. For a flywheel,
the stored energy is given according to the equation:
Keywords - Energy Storage, Flywheels, Batteries, 1
Magnetic Bearings, Superconductors, SMES, Ef = ⋅ J ⋅ω 2 , (1)
2
Supercapacitors..
where ω is the rotating speed of a disk and J its moment of
I. INTRODUCTION inertia.
The energy per volume that can be stored is amazing. For
Nowadays, an increasing number of industrial processes instance, a cylinder of radius R, mass density ρ, and volume
and commercial applications ask for a continuous and high V0 has inertia:
quality electric energy supply. In the eventual failure of 1
energy or even voltage variations, a backup system must J= ⋅ ρ ⋅ V0 ⋅ R 2 . (2)
guarantee the power delivery. Usually, during the first
2
Using (1) and (2), the stored energy per volume is:
seconds or minutes, this requirement has been fulfilled by
batteries. A supervisory control coordinates the switch Ef 1 1
between the energy sources. These systems are called UPS = ⋅ ρ ⋅ R2 ⋅ω 2 = ⋅ ρ ⋅ v2 , (3)
V0 4 4
(Uninterruptible Power Supply) or DVR (Dynamic Voltage
Restorer) and the market value reaches 1010 US$ per year. where v is the tangential velocity. An up to date estimation
There are political, technical and economical trends to can be obtained taking v=0.9km/s and ρ=7000 kg/m3:
substitute batteries, which are environmental aggressive. Ef
Among the alternatives, the most promising [1-2] are: = 1.42 × 109 J/m3 . (4)
V0
- flywheels, that store energy in rotating masses,
- supercapacitors, that store energy in electric fields, As comparison, a capacitor (C), at a voltage (V), stores:
- SMES (Superconducting Magnetic Energy Storage), 1
Ec = C ⋅ V 2 . (5)
that store energy in magnetic fields. 2
This paper shows the advantages of flywheels in The capacitance of parallel plates of area A, at a distance
comparison with the others. d, is given by:
The drawbacks presented by flywheels are gradually ε⋅A
overcome [3-4] due to: C= , (6)
d
- new composite materials based on carbon and fiber
where ε is the permittivity of the dielectric between the
glass, that allow tangential speeds of 1km/s,
- vacuum systems, that reduce air friction, plates. Therefore, using (5) and (6) with Vo = A⋅ d:
2
- magnetic bearings, that reduce contact friction, Ec 1 V 
- power electronics, that allow power conditioning, = ⋅ε ⋅  . (7)
- micro electronics, that allow cheap and sophisticated V0 2 d 
control systems. (V/d) is the breakdown strength of the dielectric. A good
The evolution points to a new generation of flywheels, value is 200 MV/m. Considering ε = 8.85x10-11 F/m, ten
which can integrally substitute batteries. The effort of the times the permittivity of free space, the stored energy per
Laboratory of Applied Superconductivity LASUP/UFRJ to volume is:
follow this tendency is also described. Ec
= 1.76 × 106 J/m3 . (8)
V0
Manuscript received on 16/01/2008. Revised on 07/05/2008 and 09/07/2008.
Accepted by recommendation of the Editor F.ernando L. M. Antunes.

Eletrônica de Potência, vol. 13, no. 3, Agosto de 2008 171

 Nowadays, supercapacitors, made of a porous structure TABLE II
that awards very large contact areas and extremely thin Energy Density of Different Types of Batteries*
layers, can present values 40 times higher [5], reaching J/m3 J/kg Wh/kg
Ec Lead-Acid 0.36 x 109 0.15 x 106 41
= 0.7 × 108 J/m3 . (9) Nickel Cadmium 0.50 x 109 0.14 x 106 39
V0
NiMH 1.08 x 109 0.34 x 106 95
Nevertheless, this is one order of magnitude lower than Alkaline 1.15 x 109 0.40 x 106 110
that of flywheels. Lithium-ion 0.83 x 109 0.46 x 106 128
On the other hand, the energy stored in magnetic fields per *http://www.allaboutbatteries.com
volume is given by:
response rate of the storage device. In this respect, direct
Em 1 B2 storage devices, where energy is already stored in electrical
= ⋅B⋅H = , (10)
V0 2 2⋅µ (supercapacitors) or magnetic form (SMES), have higher
where µ is the material permeability. response rates in comparison with indirect storage devices,
For ferromagnetic circuits, usual values of B are 1T and where an intermediate medium is necessary, for instance
µ= 4π ×10-4 H/m, thousand times the value of free space. chemical (batteries) or mechanical (flywheels, water dam)
This gives: [7]. Figure 1 presents a qualitative view of Power Density x
Energy Density of the considered storage devices.
Em
= 0.40 × 103 J/m3 . (11)
V0
Using SMES devices [6], this value can be improved. In
extreme cases, fields of 3.3 T can be obtained in free space,
which would lead to:
Em
= 0.40 × 107 J/m3 . (12)
V0
Again, it is much lower than that of flywheels.
Large amounts of energy are stored in water dams.
EW
= ρ ⋅ g ⋅h , (13)
V0
where g is the acceleration of gravity and h the reservoir
height. With ρ = 103 kg/m3 and g=10m/s2, equation (13)
shows that heights as large as 105m (100 km!) would be
necessary to reach values comparable to that of flywheels.
The calculations presented above are summarized in Table
I and explains the interest of the scientific community to Fig. 1 Power Density x Energy Density.
investigate the storage of energy in rotating masses.
As can be concluded, there exists an overlapping area
TABLE I between flywheels and batteries regarding power and energy
Order of magnitude of the Stored Energy/Volume densities. Other aspects will be considered in the next
Energy storage Energy density Order of section.
type expression magnitude (J/m3)
Flywheel ¼ ρ v2 109 III. STATE OF THE ART
2
Supercapacitor ½ ε (V/d) 108 The idea of using rotating masses to avoid velocity
SMES B2 / (2 ) 107 variations of motors and generators is an old one. It is based
Water Dam ρgh 106 on the fact that energy does not change abruptly with the
application of limited forces. This kind of flywheel can be
For batteries, the energy stored depends on the materials classified as the 1st Generation and the objective is just to
employed and there is no simple deduction for the energy maintain a fixed rotational speed with low variations.
density. Typical values of energy density of representative With the introduction of power electronics rectifier and
types of batteries are shown in TABLE II. All batteries inverter circuits and control methods, it was possible to
present the same energy density order of magnitude (109) in generate a fixed voltage and frequency from a flywheel at
J/m3, the value obtained for flywheels. different speeds. Figure 2 illustrates this 2nd Generation of
B. Power Density flywheels. The energy flow, indicated by arrows, can also be
reversed to charge the flywheel. The rotating mass can now
Besides the energy density, the power density vary its speed and even so the rectifier/inverter circuit is able
characterizes the energy storage technology. Power gives the to produce a fixed frequency and voltage. The objective is to
rate at which energy can be transferred. It depends not only store energy to fill up grid failures. This application is know
on the power rating of the conversion unit, but also on the as FESS (Flywheel Energy Storage System).

172 Eletrônica de Potência, vol. 13, no. 3, Agosto de 2008

 TABLE III
BESS x FESS
BESS FESS
Energy conversion Chemical Electrical Mechanical Electrical
losses ~35% ~10%
Idle losses
0.05% 1 to 3 %

Recharging /
15 to 50 2.5 to 90
Discharging time
Life cycle
5 years 25 years

This table shows that FESS’s are advantageous in case of

Fig. 2. Flywheels at the 2nd Generation.
frequent use, since the higher idle losses are compensated for
An initial comparison of this system with BESS (Battery with the higher energy conversion efficiency. Moreover, the
Energy Storage System) is depicted in Table III [6]. recharge of flywheels can be quicker than that of batteries.
Losses during the energy conversion process refer to the Table IV shows a comparison between commercially
transformation from chemical to electrical energy for BESS available FESS and BESS highlighting the power and energy
and from mechanical to electrical for FESS. density. The differential advantage of FESS lies in its power
Idle losses are mainly due to leakage currents, in the case density and that of BESS in its energy density. In other
of batteries, and friction and eddy currents, in the case of words, FESS application niche is for energy quality, when
flywheels. the requirement is of short duration (~10 seconds), whereas
The ratio between recharging and discharging time is BESS are appropriate for power failure of longer duration.
important when considering operational cycles. Table V clarifies this point showing the values of a
The life cycle gives the expectation of use and 5 years is flywheel for 2 hours. In this case, both power and energy
the best valuation for BESS. density present no advantage in comparison with batteries.

TABLE IV
Power and Energy Density of BESS and FESS
Available Energy Available Power Duration Energy density Power density
Producer
(kWh) (kW) (seconds) (Wh/kg) (W/kg)
Acumetricsa
0.4 140 10 1 340
b
Active Power
0.8 240 12.5 0.7 190
Trinity Flywheelc
0.4 100 15 3 730
Optimal Energy
33.7 45 5
Systemsd
Lead-Acid
- - - 41 150
batteries
a
http://www.acumetrics.com
b
http://www.activepower.com
c
http://www.afstrinity.com
d
http://www.optimalenergysystems.com

TABLE V
FESS for long periods or time
Available Energy Available Power Duration Energy density Power density
Producer
(kWh) (kW) (seconds) (Wh/kg) (W/kg)
Beacon Powerd
2 1 7200 5.5 2.8
d
http://www.beaconpower.com

TABLE VI
BESS x FESS of the 3rd Generation
Available Energy Available Power Duration Energy density Power density
Producer
(kWh) (kW) (seconds) (Wh/kg) (W/kg)
Boeing
8.3 100 300 56 670
Lead-Acid
- - - 41 150
batteries

Eletrônica de Potência, vol. 13, no. 3, Agosto de 2008 173

 Ef 1 2
IV. FUTURE PERSPECTIVES: THE 3RD GENERATION = v = 56Wh / kg , (14)
ρ ⋅V0 4
The limits imposed to flywheels can be overcome which is exactly the same value.
increasing the rotating speed. As the stored energy depends
on the velocity squared (Equation 1), gains in velocity
advantageously increase the energy without compromising
the weight (mass).
High speeds (~60000rpm) are possible with rotating
masses of composite materials of high strength, like carbon
or glass fiber, vacuum chambers and magnetic bearings to
avoid friction losses [6,8,9]. Research groups all over the
world pursue this goal, including the Laboratory of Applied
Superconductivity/UFRJ. Some results are presented in [10].
Figure 3 presents a photograph of the laboratory prototype.

2 Fig. 4. Flywheel drawing showing the connection of the

3 flywheel mechanical system with the switched reluctance drive, the
grid controller and the grid.
4
Since the pioneer work of Alessandro Volta in 1800, an
improvement of batteries have been observed, parallel to the
described development of flywheels. The Lead Acid
batteries, developed at the second half of the XIX century,
continue to be the cheapest one. The Nickel Cadmium cell
5
dates the threshold XIX-XX century and the Alkaline
batteries are from the middle of last century. New kinds of
batteries like Lithium-ion with energy density as high as 128
Fig. 3. Flywheel prototype photograph showing: the encoder(1), SR
Wh/kg and power density of 720 W/kg are still better than
drive(2), flywheel and SMB rotor(3), SMB stator(4) and vacuum
chamber enclosure(5). the 3rd generation numbers presented above (see Table II).
Nevertheless, considering that the life cycle of FESS is at
Its main components are an axial Superconducting least five times greater than the lifetime cycle of BESS, the
Magnetic Bearing (SMB), a Switched Reluctance (SR) drive, higher initial costs of FESS can be diluted along the years.
which operates as motor/generator, a Vacuum Chamber, a Moreover, the prices of the 3rd Generation of flywheels tend
Liquid Nitrogen (LN2) feeder and a rotating mass (the SMB to decrease with the production growth and technological
rotor plus the SR rotor). The encoder is necessary for the improvements and a promising substitute to batteries can be
drive control. The power electronic rectifier/inverter circuits foreseen.
are not shown. The control is embedded in a DSP
development system. V. SMB: THE 3RD GENERATION PILLAR
More details can be seen in Figure 4. This drawing shows
the radial positioning given by Permanent Magnet Bearings The development of new FESS depends strongly on the
(PMB), the Superconducting Magnetic Thrust Bearing bearing used in the system. The application of active
(SMB) and the power electronics circuits that control the magnetic bearing in high-speed rotors is well known in the
motor/generator and make the connection to the grid. The literature [12], but the use of SMB for FESS is still under
SMB is an assembly of Nd-Fe-B magnets and investigation [13-14]. The advantages of SMB are:
superconducting blocks of YBa2Cu3O7-y (YBCO), which are - no need of control systems,
cooled with liquid nitrogen. - relatively high magnetic pressure,
Recently, Boeing announced its brand new flywheel [11] - very-low energy losses during operation (no contact),
also based on superconductor bearings, which data are - low maintenance cost.
compared with BESS in Table VI. Energy and power density The superconductors used in the SMB are based on
of this 3rd Generation of flywheels are both higher than that YBCO blocks and need to be cooled below 92K. This can be
of Lead-Acid batteries. done with the cheap Liquid Nitrogen. Due to the high non-
It is illustrative to compare the energy density of this linearity involved in the superconductor levitation, the
flywheel (56Wh/kg) with the estimation presented in Eq.(4). investigation of better topologies for superconductor bearings
From (3), the ratio Stored Energy/Mass is simply ¼ v2. For has been extensively researched by our group [15-16]. The
v=0.9km/s and remembering that 3600 J = 3600 Ws = 1 Wh,
it follows:

174 Eletrônica de Potência, vol. 13, no. 3, Agosto de 2008

main goal and top contribution of these works are the VI. CONCLUSION
increase of levitation force and stiffness.
A rotating SMB may be a thrust (axial) or a journal A 3rd Generation of flywheels is on the verge of
(radial) bearing. Both of them have influence on the axial applications. New materials allow speeds of the order of
and radial directions. There are three principal topologies 60000 rpm. Magnetic bearings based on superconductors or
that can be used in the SMB bearings: Flux Shaper (FS), active feedback loops can be implemented. The stored
Axially Magnetized Ring (AMR) and Halbach array. The energy and the power capacity are comparable to that offered
SMB is composed of a magnetic rotor and YBCO by traditional lead-acid batteries.
superconductors in the stator. Figure 5 shows a cut view This 3rd Generation breaks old technological concepts,
comparing three magnetic rotors topologies of thrust bearing offering to developing countries an opportunity to jump
and the respective flux lines. The Halbach configuration may engineering gaps.
present higher levitation force than the other configurations This paper highlighted the advantages of flywheels and
[15], but it is extremely hard to be built. In the authors’ illustrated the main components of the 3rd Generation with
opinion, the increase in the levitation force does not the help of a laboratory prototype.
compensate for the work necessary to construct this rotor.
ACKNOWLEDGEMENT

To CNPq and FAPERJ for the financial support. To

A.C.Ferreira, L.G.B.Rolim, J.L.da Silva Neto, W.Suemitsu,
R.Nicolsky and O.J.Machado for their contribution to the
LASUP/FESS Project.
Fig. 5. Three topologies of magnetic rotors for thrust SMB.
REFERENCES
One FS and two AMR prototypes of rotating SMB with
~75mm of diameter were developed, and the magnetic rotors [1] J.M.Carrasco, L.G.Franquelo, J.T.Bialasiewicz,
are presented in Figure 6. The stator base is composed of an E.Galvan, R.C.Portillo Guisado, M.A.M.Prats, J.I.Leon
YBCO disk having 75mm in diameter and thickness of and N.Moreno-Alfonso, “Power Electronics Systems for
16mm. Grid Integration of Renewable Energy Sources: A
Survey”, IEEE Trans. On Industrial Electronics, Vol.53,
no.4, pp.1002-1016, 2006.
[2] P.F.Ribeiro, B.K.Johnson, M.L.Crow, A.Arsoy and
Y.Liu, “Energy Storage Systems for Advanced Power
Applications”, Proceedings of the IEEE, Vol. 89, no.12,
pp.1744-1756, 2001.
[3] R.Hebner, J.Beno and A.Walls, “Flywheel Batteries
Come Around Again”, IEEE Spectrum, pp.46-50, April,
Fig. 6. Magnetic rotors built for tests. 2002.
The levitation force measurement of the prototypes shown [4] R.G.Lawrence, K.L.Craven and G.D.Nichols, “Flywheel
above, for refrigeration in the absence of the rotor magnetic UPS”, IEEE Industry Applications Magazin, Vol.9, no.3,
field, is shown in Figure 7. Masses up to 25kg can be pp.44-50, 2003.
levitated above a circle having just 75mm of diameter. This [5] A.A.Ferreira and J.A.Pomilio, “Estado da Arte Sobre a
result illustrates a magnetic pressure of up to 60kPa Aplicação de Supercapacitores em Eletrônica de
presented by the SMB, without friction loss, independently Potência”, Eletrônica de Potência – Vol. 10, nº 2,
of the angular velocity. November, pp.25-32, 2005.
Experimental results of energy transfer with this prototype [6] R.Silberglitt, E.Ettedgui and A.Hove, Strengthening the
are reported elsewhere [10] and [18]. Grid: Effect of HTS Power Technologies on Reliability,
Power Transfer Capacity and Energy Use, RAND
300
(http://www.rand.org), 2002.
AMR1
250 [7] T.Sels, C.Dragu, T. Van Craenenbroeck, R. Belmans,
AMR2
FS “New Energy Storage Devices for an Improved Load
200 Managing on Distribution Level”, IEEE Porto Power
Levitation Force (N)

Tech Conference, Porto, 2001.

150
[8] A.M.Wolsky, “The Status and Prospects for Flywheels
100 and SMES that Incorporate HTS”, Physica C, pp.1495-
1499, 2002.
50
[9] J.R.Hull, “Applications of Bulk High-Temperature
0
Superconductors”, Proceedings of the IEEE, Vol.92,
No.10, pp.1655-1667, 2004.
-50
0 5 10 15
[10] R.deAndradeJr., G.G.Sotelo, A.C.Ferreira, L.G.B.Rolim,
Gap (mm) J.L.daSilva Neto, R.M.Stephan, W.I.Suemitsu, and
Fig. 7. Levitation force measurements of SMB prototypes. R.Nicolsky, “Flywheel Energy Storage System

Eletrônica de Potência, vol. 13, no. 3, Agosto de 2008 175

 Description and Tests”, IEEE Trans. on Applied Rio de Janeiro (UFRJ) in 1980, and the Dr.-Ing. degree in
Superconductivity, Vol. 17, nº. 2, pp. 2154-2157, electrical engineering from Ruhr Universität Bochum,
June2007. Germany, in 1985. He has an MBA degree (2005) from the
[11] Superconductor Week, Vol.20, 2006. Center for Scientific Enterprise, London (CSEL), on
[12] G.Schweitzer, H.Bleuler and A.Traxler, Active Magnetic Technology Enterprise Development.
Bearings: basics, properties and applications of active During 1977, he worked as an engineer at Furnas Centrais
magnetic bearings, Zurich: VDF, 1ª ed., 1994. Elétricas, Rio de Janeiro. Since 1978, he has been with the
[13] S.Nagaya, N.Kashima, M.Minami, H.Kawashima, and Department of Electrical Engineering, UFRJ. He spent a
S.Unisuga, “Study on high temperature superconducting sabbatical leave at CEPEL, the Research Center of
magnetic bearing for 10 kWh flywheel energy storage ELETROBRAS in 1993. His main interests are in the fields
system”, IEEE Trans. on Applied Superconductivity, vol. of applications of superconductivity, control of electrical
11, pp. 1649-1652, march, 2001. drives and power electronics.
[14] A.C.Day, J.R.Hull, M.Strasik, P.E.Johnson, Dr. Stephan is member of SOBRAEP and IEEE.
K.E.McCrary, J.Edwards, J.A.Mittleider, J.R.Schindler,
Rubens de Andrade Jr. received.the B.Sc., M.Sc and
R.A.Hawkins, M.L.Yoder, “Temperature and Frequency
D.Sc. degree in physics from Universidade Estadual de
Effects in a High-Performance Superconducting
Campinas (UNICAMP), in 1985, 1989 and 1995
Bearing”, IEEE Trans. on Applied Superconductivity,
respectively. Since 1999, he has been with the Department of
Vol. 13, nº 2, pp. 2179-2184, June, 2003.
Electrical Engineering, UFRJ. He has worked with selective
[15] G.G.Sotelo, A.C.Ferreira, and R.deAndradeJr., “Halbach
surfaces for solar heaters, electrochemical alloy deposition,
Array Superconducting Magnetic Bearing for a Flywheel
vortex dynamics of type II superconductors, HTS preparation
Energy storage system”, IEEE Trans. on Applied
and characterization (Hg-1212) and vortex dynamics of Hg
Superconductivity, vol. 15, pp. 2253-2256, June, 2005.
based superconductors (Hg-1212 and Hg-1223). At moment,
[16] G.G.Sotelo, A.C.Ferreira, and R.deAndradeJr.,
his main interests is in the applications of superconducting
“Magnetic bearing sets for a flywheel system”, IEEE
materials in power electrical systems and transportation, also
Trans. on Applied Superconductivity, Vol. 17, nº. 2, pp.
he has interest in the simulation of superconducting devices.
2150-2153, June, 2007.
[17] M.R.Ribeiro, M.El-Mann, G.G.Sotelo, L.G.B.Rolim, Guilherme G. Sotelo received the B.Sc. degree in physics
J.L.daSilva Neto and Richard M. Stephan, “Sistema from the Federal University of Rio de Janeiro (UFRJ), Rio de
Armazenador de Energia Cinética – Simulações e Janeiro, in 2001, the M.Sc. and D.Sc. degrees were obtained
testes”, submitted to the XVII Brazilian Conference on in electrical engineering from the UFRJ in 2003 and 2007,
Automatic Control (CBA’2008). respectively. Actually he is Post-Doc in the Laboratory for
Applied Superconductivity of Rio de Janeiro. His main
BIOGRAPHIES
interests are in the fields of applications of
Richard M. Stephan received the B.Sc. degree in superconductivity, phenomenological models for
electrical engineering from the Instituto Militar de superconductors and electromagnetic simulations with finite
Engenharia (IME), Rio de Janeiro, in 1976, the M.Sc. degree element method.
in electrical engineering from the Universidade Federal do

176 Eletrônica de Potência, vol. 13, no. 3, Agosto de 2008