IJEET
-1-
IJEET International Journal
of Electrical
Engineering in
Transportation
Volume 1
Number 1
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
-2-
IJEET
-3-
OBJECTIVES OF THE JOURNAL
The aim of the Journal is to strengthen the relationships which exist –– or must be developed
–– between the academic world and the industrial world. It is not the aim of the Journal to be
exclusively a publication vector for fundamental research work, but rather to be a tool to
promote research work having potentials for industrial applications in the short or medium
terms.
Topics covered by the Journal mainly concern but are not limited to:
* Electric vehicles, general
* Tests, Measurement and Simulation
* Vehicle applications, Transport systems
* Hybrid vehicles, general
* Drive Systems
* Energy Supply Infrastructure
* Electrical motors
* Energy and Environmental Impacts
* Energy management
* Standardizations
* Batteries
* Dynamic Control of vehicles
* Fuel Cell Systems
* MAGLEV Systems
* Other Energy Storage Systems
(Flywheel, Capacitor……)
* Embedded Systems
* Chargers and Charging Systems
* Auxiliary Vehicle Components
* Systems Diagnostic
* On Board Communication
The Journal will serve as a communication and liaison tool between public researchers and
industrialists. Practical experiments and reports on real industrial applications of research
results are particularly welcome. More fundamental research papers are of course also
accepted, as long as they include an application and present original results.
Papers published generally come from public or private research laboratories working in the
fields mentioned earlier. Columns of the Journal are also opened to industries which have
developed research activities in their own laboratories or in liaison with academic partners
or others engineering schools.
IJEET
The Journal also offers summaries of recently defended Ph.D. theses, announcements of
forthcoming events related to the topics of the Journal, examples of advanced applications
realized by students in industry as well as descriptions of research laboratories.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
-4-
INFORMATION FOR AUTHORS
IJEET is published three times the year. It contains original papers and review articles that
fit within the scope of the journal.
Submitting a manuscript
Prospective authors are invited to submit manuscripts written in correct English, complete
in their descriptions and not currently under consideration by another journal. All submitted
manuscripts will be peer-refereed, and may undergo modifications to attain the quality
expected in an international scientific journal. Papers requiring major revision will be
rejected.
Manuscripts should be submitted to the Editor only as electronic files (Word format –– Arial
police). They can be sent by email (to ijeet@utbm.fr) if their size is under 5Mo. Otherwise,
a CDROM containing the manuscript and the figures must be mailed to the Editor.
When preparing the manuscript the following should be noted:
x Formatting commands such as centering, justifying, indenting etc. must not
be used. Page design will be handled directly by the Publisher.
x The word processor's facilities must be used to indicate the text attributes:
bold, italic, subscript, superscript, strikeout.
x The hyphenation facilities of the word processor must not be used.
x When preparing tables, tabs not spaces must be used to align columns.
x When using symbols to denote special characters, a list of all codes used
must be supplied.
x An abstract of about 150 words must be included. The abstract should
indicate the subject and scope of the paper and summarize the author's
conclusions, so as to be a useful document in an information retrieval system.
x A short list of three to four keywords should be included.
HTU
UTH
References
References substantiating points made in the text or citing important and relevant work
should appear in a separate section at the end of the article. Citations in text are indicated
by bracketed Arabic numbers, numbered according to the order of citation. A referenced
article should contain (in this order) all authors' names, title of the article, name of the
publication, volume and number, month and year of publication, and inclusive page
numbers. A referenced book should list author name(s), title of the book, publisher, place
of publication, year of publication, and specific chapters or pages cited.
IJEET
Illustrations
All illustrations must be cited in text and have numbers and captions. The illustration
should remain readable at a two columns' width. Notice that colour illustrations will be
reproduced in black and white.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
-5-
IJEET COPYRIGHT POLICY
General Terms
The authors are supposed to transfer to the International Journal of Electrical Engineering in
Transportation (IJEET) all rights under copyright that may exist in and to the above paper, and any
revised or expanded derivative works submitted to IJEET by the undersigned based on the paper.
The paper must be original and never been published in another journal. In case the paper
incorporates text passages, figures, data or other material from the works of others, the authors must
have obtained any necessary permission.
Author Responsibilities
Authors are advised that it shall be acknowledged that statements and opinions given in work
published in IJEET are the expression of the authors. Responsibility for the content of published
papers rests upon the authors, not IJEET.
IJEET
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
-6-
SUBSCRIPTION 2005
Institutional Subscription
Individual Subscription
:
:
200 €€
75 €€
Payment must be sent to:
University of Technology of Belfort-Montbéliard
IJEET
90010 BELFORT CEDEX
France
Please contact the following email address for subscription: ijeet@utbm.fr
IJEET
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
-7-
Table of contents
Brushless DC Permananet Magnet Motor for Electric Bike and their Impulse System for
Battery Charging
9
S. Wiak, R. Nadolski, K. Ludwinek, Z. GawĊcki
P
Thermal analysis of Azipod permanent magnet propulsion motor
15
T. Jokinen, A. Arkkio, M. Negrea, I. Waltzer
Supercapacitors as an energy storage for fuel cell automotive hybrid electrical system
21
P. Thounthong, S. Raël, B. Davat
High-acceleration linear drives: Application to electromagnetic valves
27
C. Bernez, X. Mininger, H. Ben Ahmed, M. Gabsi, M. Lecrivain, E. Gimet, E.Sedda
Flywheel Electric Drive Characterization for Hybrid Vehicles
41
Y. Gao, S. E. Gay, M. Ehsani, R. F. Thelen, R. E. Hebner
Mobile System for Testing and Calibrating Vehicle Speed Sensors
D. Moga, M. Munteanu, T. Marita, C. Cret
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
47
IJEET
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
-8-
IJEET
-9-
Brushless DC Permananet Magnet Motor for Electric Bike
and their Impulse System for Battery Charging
S. Wiak1, R. Nadolski2, K. Ludwinek2, Z. GawĊcki2
P
P
P
P
P
P
P
1
P
- Institute of Mechatronics and Information Systems, Technical University of Lodz, Poland
ul. Stefanowskiego 18/22, 90-924 Lodz, Poland, e-mail: wiakslaw@p.lodz.pl
2
-Department of Electrical Machines, Technical University of Kielce
P
HTU
P
I.
UTH
P
In the literature two following types of motors have been
studied:
x radial flux machine
x axial flux machine.
INTRODUCTION
Axial flux machines proved to have optimal characteristics in
electrical traction for direct coupling to the vehicle wheels.
The recent development of integrated electric drives is
closely related to the evolution of smart motor technologies.
However, the development of new motor structures imposes
higher requirements than conventional one. For electric vehicles
application of electric motors driving the wheels is widely used. In
this case wheels are directly driven by the electric motor and the
gears are not necessary.
As it summarised in the literature [1-3] about 70% of the
total cost of DC motor drive depends on the motor and the rest
on power and signal electronics.
DC permanent magnet disc-type motor with radial structure
(radial field) is the subject of the investigations. The 3D motor
structure is proposed to be studied. The magnetic field
distributions and torque characteristics have given the
knowledge about the motor structure and material parameter
changes in order to increase the unit motor torque. The
investigations have been done for static and quasi-dynamic
states as well.
Utilisation into urban traffic relatively cheap light vehicles
electrically driven could cause reduction of the emission of local
fumes coming from combustion vehicles. Applying of the
unconventional sources of energy to electric vehicles (also
directly driven), like high efficient batteries are looked very
promisingly, and leads to the reduction of global emission of the
fumes. At present, the main difficulty of rapid introduction of
vehicles with electric driven is due to the limitations in technology
of battery production. Thanks to modern technology of electric
motors construction, e.g. brushless direct current motors with
permanent magnets (BLDCM with PM) [2, 3, 4, 5], as well as the
improvements of lead-acid batteries, contemporary electric
vehicles compete with conventional drives very well in respect of:
the driving speed, the maximum speed, obtained accelerations,
grade ability, etc.
The electromagnetic field analysis of the motor has been
carried out by means of finite element method.
The electromagnetic field, in our approach, is described by
joining different electromagnetic potentials, namely: reduced
potential for subspace with coils, scalar potential in the air and
ferromagnetic materials, magnetic vector potential with electric
potential in the area of eddy currents.
II.
As it is reported in the literature [1,2] the axial filed motors
could be applied in the low speed and high torque electrical
drives, exemplary for vehicles for disabled people, electric bikes
or electric scooter.
The proposed the direct drive of electric bicycle, while the
brushless PMDCM is mounted in hub of the ahead wheel of a
bicycle has been seen in Figure 1 and 2. Authors propose,
instead classical bike structure where second wheel is driven,
ahead wheel to be driven. This philosophy of wheel driving gives
the manufacturer to assemble such a bike according to the
customer demand. The battery supplying system consists of two
lead-acid accumulators. One part of the electronics control
system is mounted on the handlebar, but the other one is
mounted on the front guide fork.
In such a motors family induction motor and PM could take a
dominating role. Contactless AC servodrives are designed by use
of squirrel cage induction motors or permanent magnet
synchronous motors (PMSM) or BLDC motors.
Drive systems have largely remained DC powered with only a
few manufacturers applying induction motors with flux-vector
control techniques. The requirement for high efficiency and low
weight can be met with the use of rare earth permanent magnets
in brushless DC forms.
Big advantage of PMSM in the reference to induction motors
is in simplicity of their control. PMSM has also small eddy current
and iron core losses, thus their efficiency is higher. These motors
are designed for sine wave current supply or for trapezoidal wave
form currents. Therefore in the second case of supply is the
machine also called as comutatorless DC machine.
MOTOR APPLICATION
The motor structure with radial magnetic field is DC machine
with permanent magnets and reversed structure. Rotor is made
of Ne-Fe-B permanent magnets with demagnetisation curve of Br
= 1.25T and Hc = 1200 kA/m.
B
B
B
B
The motor study has been performed for motor keeping
constant velocity of 200 rpm. In order to carry out such an
analysis the mesh is being changed dynamically according to
changes of rotor movement. The mesh is generated following the
idea of so called ““slip surface”” at the air gap (see Fig. 2). The
torque characteristics are shown in Fig. 3.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 10 -
a)
The comparative study of motor characteristics should lead to
motor structure development in order to increase the average
torque, minimise the torque ripples as well, and to select the type
of motor for either motorbike or electric scooter.
Nm
30
9 A/mm2
P
25
20
6 A/mm2
P
15
10
5
0
b)
5
-5 0
10
15
10
15
20
Figure 3. Torque characteristics versus rotor position and
different current density of exciting current.
The effectiveness of the 3-D field model of DC motor with
permanent magnets has been verified by measurements done for
prototype machine. The radial-field motor structure has been
successfully applied to electric bike.
Figure 1. Electric bike: a) global view, b) ahead wheel with
mounted electric motor.
III.
IMPULSE CIRCUIT FOR BATTERY CHARGING
In order to extend the battery working time, leading to
increase the bike travelling distance, the impulse electric circuit of
transferring the energy coming from induced electromotive forces
in DC motor windings is proposed. The battery charging goes
during the so called generator braking.
a)
Brushless direct current motors with permanent magnets
thanks to presence of permanent magnets and outer turning
torque can run as generators. The rms. of induced electromotive
force for the voltage fundamental component in each winding is
described with the following simple equation:
b)
magnet
stator tooth
c n ) m z1ku1
E
(1)
slip surface
a)
where: z1, ku1 –– number of series turns and winding coefficient for
the one phase of the three-phase motor winding respectively, )m
–– magnitude of magnetic flux, n –– mechanical rotational speed, c
–– constant.
In direct current motors with permanent magnets the value of
induced emf depends practically only on rotational speed
(frequency) of a motor.
The measured curves of induced electromotive force in threephases of investigated BLDCM, with rated data as follows:
nominal power PN = 125W, nominal current IN = 8A, nominal
voltage UN = 24V, mechanical rated speed nN = 120 rpm, are
shown in Figure 4.
B
B
B
B
B
B
B
stator tooth
B
magnet
rotor core
Figure 2. DC motor structure with mesh: a) one pole pitch
geometry, b) slip surface at the air gap.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
B
B
B
B
B
B
B
B
IJEET
- 11 -
a)
uA, uB, uC - V
6
4
uA
uB uC n = 40 rpm
2
0
-2
-4
0,0
b)
0,1
0,2 0,3
t - s
0,4
0,5
The induced emf curves show that maximum instantaneous
values of these voltages, even at 200 rpm, are smaller than the
voltage on terminals of two accumulators in series connected.
Hence, frequently charging of the battery is possible to realise
with the help of frequency converter or other impulse system of
electric energy conversion.
In this case of the electric bicycle directly driven, the
frequency converter has been realised as the increasing voltage
AC/DC system. After suitable transformation the phase winding
currents (forced under the induced emf) can be put-upon to the
battery charging.
Battery charging is possible during the so-called generator
braking. In the frequency converter cooperating with electric drive
equipped with BLDC motor, the generator braking is realised
while the control system of the motor operation is switched off
(microcontroller Allegro MicroSystems Inc. type –– 3933 is
switched off). Impulse electric circuit of transferring energy is
switched on The scheme of the total system of the impulse
electric circuit transferring energy is shown in Fig. 5.
D1
D3
D7
D5
C1
Ac2
T1
C2
C3
Ac1
Tr1
D2
d)
D8
T2
D6
BLDCM
20
15
10
5
0
-5
-10
-15
0,00
uA uB uC n = 140 rpm
R2
C5
R3
1
P1
2
3
R1
4
12
15
SG 2525
uA, uB, uC - V
c)
D4
13
5
7
6
8
9
10
16
11
14
R4
R5
C2
C1
C3
R6
R7
R8
Vcc
C4
Fig. 5. Impulse electric circuit transferring energy
0,05
0,10
t - s
0,15
The frequency converter is used to transfer the alternating
current induced in each winding into direct current, while the
output voltage is higher then battery voltage.
The control unit of the impulse electric circuit transferring
energy to battery has been realised by use of specialized
integrated circuit SG 2525A, made by Silicon General.
Induced alternating electromotive forces (emf) in BLDCM
phase windings, shown in Fig. 4, are rectified by means of threephase diode rectifiers (Fig. 5). The electric energy through the
three-phase diode rectifiers is stored in electrolytic capacitor C1
of 4700PF. The energy stored in the capacitor is transferred into
system consisting of two switching MOSFET transistors (so
called keys), which switch cyclically with the frequency of 18 kHz.
Moreover, the proposed frequency converter has got the softstart system providing a smooth starting of the whole impulse
electric circuit of transferring energy to the battery.
Fig. 4. Measured curves of induced emf in three-phases of
investigated BLDCM for: a) n = 40 rpm, b) n = 80 rpm, c) n = 140
rpm, d) n = 200 rpm
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 12 -
IV.
LABORATORY TESTS OF IMPULSE CIRCUIT
FOR BATTERY CHARGING
The laboratory measurements have been done for the
model system shown in Figure 6.
x
x
x
Lecroy digital oscilloscope,
12 V baterry,
frequency converter.
The technique of measuring voltages and currents based on
Hall transducers application eliminate all problems connected
with signals separation, while measured.
In Figures 7, 8, and 9 voltage and current curves, in selected
points of the investigated impulse system of battery charging and
different values of rotational speed, are shown.
The battery voltage, charging battery current, phase current,
and phase voltage are shown in Figure 7; while rotational speed
varies from 0 to 76 rpm, and impulse width of transistors control
is being changed.
The battery voltage, charging battery current, phase current,
and phase voltage are shown in Figure 8; while rotational speed
varies from 76 (at the maximum value of charging current) to 172
rpm (while charging current is equal to 0A), and impulse width of
transistors control is being changed.
The switching voltages uDT1 and uDT2 of MOSFET transistors,
measured on the pins 11 and 14 of the integrated circuit
SG2525A, are shown in Figure 9. Measured curves of full voltage
duty-cycle of the switching MOSFET transistors and rotational
speed equals to 76 rpm are shown in Figure. 9a, while the same
voltage of low voltage duty-cycle of the switching MOSFET
transistors and rotational speed equals to 112 rpm are shown in
Figure 9b.
B
Fig. 6. Laboratory measurement set for investigation of the
frequency converter in state of generator breaking.
The model system is built of the following components:
x the brushless DC permanent magnet motor with rated data
PN = 125W, UN = 24V, nN=120rpm (built-in in hub of the
ahead wheel of a bicycle - shown in Fig. 1),
x PC with software and 12bit multifunction DAQ card, and a
set of Hall transducers measuring voltages and currents,
B
B
B
B
B
B
B
B
B
(a)
(c)
(b)
(d)
Fig. 7. Recorded curves a) battery voltage, b) charging current c) A-phase current, d) A-phase voltage
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 13 -
(a)
(c)
0,8
iak - A
0,6
0,4
0,2
0,0
0
2
4
6
8
10
12
14
t - s
(b)
(d)
Fig. 8. Recorded curves a) battery voltage, b) charging current c) A-phase current, d) A-phase voltage
20
15
uDT1, uDT2 - V
V.
uDT1
uDT2
10
5
0
-5
0,00000
0,00005
0,00010
0,00015
0,00020
t - s
(a)
20
uDT1
uDT2
uDT1, uDT2 - V
15
CONCLUSIONS
The proposed impulse system of battery charging leads to
increasing of the electric bike distance of travelling, driven by
BLDC motor. This system could be also successfully applied for
charging a set of batteries after simple improvements to be
introduced to the electric scheme.
The proposed solution, by use of the frequency converter,
after changing transformer turns ratio could be in simple way
adopted to charging of arbitrary selected numbers of
accumulators constituting the electric bike supplying battery.
Due to application of the pulse width modulator changes of
the duty-cycle of the switching MOSFET transistors with openloop could make a possibly smooth regulation of frequency
converter load factor. Such a solution allows the driver on
optimum choice between the travelling speed and the value of
battery charging current.
10
REFERENCES
5
0
-5
0,00000
0,00005
0,00010
t
0,00015
0,00020
- s
(b)
Fig. 9. Measured courses examples of voltage duty-cycle of the
switching MOSFET transistors a) full voltage duty-cycle, b) low
voltage duty-cycle
1. Zhang Z., Profumo F., Tenconi A,”” Wheels axial flux machines for
electric vehicle applications””, International Conference Electrical
Machines (ICEM’’94), Conference Proceedings, Paris, Sept. 1994.
2. Consoli A, ““Propulsion Drives for Light Electric Vehicles””,
International Conference Electrical Machines (ICEM’’94),
Conference Proceedings, Paris, Sept. 1994.
3. Wiak S., Welfle H., KomĊza K., Mendrela E., ““Electromagnetic Field
Analysis of Disk-type Induction Motor””, International Conference
Electrical Machines (ICEM 98), vol. 2/3, pp. 735-739, 2- 4
September 1998, Istanbul, Turkey.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 14 -
4. Nadolski R., GawĊcki Z., Staszak J. Ludwinek K., “Gearless drive of
light electric vehicles on the example of the bicycle driven with
brushless DC motor with three––phase winding””, 4th International
Workshop on Research and Education in Mechatronics 2003,
October 2003, University of Applied Sciences, Bochum, Germany.
5. Wiak, S., Nadolski, R.: Disc Type Motors for Light Electric Vehicles Comparative Study", The First Slovenian - Polish Joint Seminar on
Computational and Applied Electromagnetics, September 10 –– 12,
Maribor, Sáowenia 2001.
6. Wiak, S., Welfle, H., Nadolski, R.: Static and dynamic states
analysis of disc type motors for light electric vehicles. 15-th
International Conference on Electrical Machines ICEM'2002, 25-28
August 2002, Brugge - Belgium.
7. Wiak S., Nadolski R., Ludwinek K., GawĊcki Z.: DC Permananet
Magnet Motor for Electric Bike and their Impulse System for Battery
Charging. 16-th International Conference on Electrical Machines
ICEM'2004, 5-8 September 2004, Cracov - Poland.
P
P
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 15 -
Thermal analysis of Azipod permanent magnet propulsion motor
T. Jokinen1, A. Arkkio1, M. Negrea1, I. Waltzer2
P
P
P
P
P
P
P
P
1
- Helsinki University of Technology, Laboratory of Electromechanics
P.O. Box 3000, FIN-02015 Hut, Finland, e-mail: tapani.jokinen@hut.fi
2
–ABB Oy, Automation Technology Products Division
P.O. Box 186, FIN-00381 Helsinki, Finland, e-mail: ingmar.waltzer@fi.abb.com
P
P
HTU
P
UTH
P
HTU
UTH
Abstract
The Azipod electric propulsion system and its benefits are
introduced and the temperature rise during a sudden short circuit
in the motor's terminals is studied. The Azipod system has
excellent manoeuvrability, high efficiency, reduced fuel
consumption and it causes less vibration and noise than a
conventional propulsion system.
According to simulations, the one-phase short circuit proved to be
the most dangerous fault from thermal point of view. The most
critical component is the stator winding. There are only a couple of
minutes of time to do preventive actions before the stator winding
is permanently damaged.
Keywords: Electric ship propulsion, permanent magnet motor,
Azipod, thermal analysis
Fig. 2 Two Azipod propulsion units in a cruiser.
I. INTRODUCTøON
Electric propulsion system for ships has become increasingly
common after ABB developed the Azipod ® system in 1990.
Azipod is an azimuthing electric propulsion drive where the
electric propulsion motor is installed inside a submerged
azimuthing pod and coupled directly to the propeller (Fig. 1). The
speed of the motor is controllable giving a smooth torque over the
entire speed range including the zero speed.
Fig. 1 Azipod propulsion system
Typical power of the propulsion motor is in the range from 400
kW up to 20 MW. One ship is normally equipped with 1……3propulsion units (Fig. 2). In rigs, equipped with dynamic position
systems, up to 10 units may be used.
The ““Azipod”” system was originally developed for icebreakers
and ice-going vessels. The first installation was made in 1990
after some years of development and research work. The
performance and the properties of the Azipod system, compared
to conventional mechanical propulsion, has increased rapidly the
popularity of the system also in other types of vessels such as
cable layers, dredgers, shuttle tankers, chemical and product
tankers, support vessels, motor yachts, drill-ships and semisubmersible rigs. Especially in big cruise vessels, where the total
propulsion power is 40 …… 60 MW, the system is highly
appreciated by the ship-owners.
The range of the Azipod propulsion systems can from the
application point of view be divided into two parts, one for smaller
vessels and drives with unit power between 0,4 and 5 MW and
one for large ships, where the unit power is between 5 and 20
MW. The bigger units are typically used in cruise vessels and
tankers, while the smaller units in offshore support vessels, drillships, oilrigs, cable layers and small ferries.
For the propulsion market 400 to 5000 kW, ABB has developed a
modular high standardised concept ““Compact Azipod”” utilising
the permanent magnet motor technology. The permanent
magnets and the direct torque control have been the main factors
for improving the performance and extending the applicability of
““Compact Azipod””.
This paper deals with permanent magnet motors used in the ship
propulsion system. One problem in this application is the
temperature rise during a fault. If the ship is cruising at full speed
and a sudden short circuit occurs in the motor terminals, the
motor does not anymore provide a positive torque but the large
mass of inertia of the ship keeps the ship in motion. The propeller
turns the rotor at about 60––70% of the original rotation speed.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 16 -
II. BENEFøTS OF AZøPOD SYSTEM
The main advantages of the Azipod system are the following:
Excellent manoeuvrability, including low speeds, using
only the podded motors without any rudder.
Powerful and fast crash stop by reversing the propeller
rotation direction, which is typical not only for podded thrusters
but for all electric propulsion systems.
The podded configuration allows applying pulling
propellers, thus providing unprecedented uniformity of the ship
wake velocities in way of the propeller making possible to
achieve extremely good cavitation characteristics of propellers
and to reduce significantly propeller-induced fluctuations,
vibrations and noise. Low noise and low vibration is very
important in passenger ships.
High total efficiency and reduced fuel consumption.
Low emission because several diesel electric power
sources are installed in the ship and only so many diesel engines
are used as needed. The engines are running at their optimal
load with a good efficiency and low emission.
The ““Compact Azipod”” utilising the permanent magnet motor has
some extra benefits. Their origin is the absence of active
windings in the rotor. Most of the losses in a permanent magnet
motor are generated in the stator, whereas the losses in the
magnets and the rotor are small. This makes it possible to apply
the surface cooling directly to the seawater and keeping
simultaneously the power density high.
The stator core is shrunk to the motor housing, which is
surrounded by seawater, thus acting as an effective cooling
media for the motor. No external cooling air or water-coolers are
required like in bigger Azipod systems. This makes the design
user-friendly. It is possible to keep the diameter of the motor
relatively small. This has a premium effect on the hydrodynamic
properties of the pod and on the total propulsion efficiency.
A comparison of the fuel oil consumption for a supply vessel with
Compact Azipod propulsion and conventional propulsion is
shown in Fig. 3.
The Compact Azipod has been designed for underwater
dismounting and mounting for easy repair without any docking.
Before mounting, the Compact Azipod unit is equipped with a
watertight protection doom and pressurised.
8000
FOC, tons/year
The motor acts as a generator driving a large current in the
shorted stator winding. Eddy-currents are also induced in the
rotor and warm up the permanent magnets. If nothing is done,
the ship may continue its motion for 10––20 minutes, and the
stator winding and maybe the permanent magnets are overheated. The aim of the paper is to estimate for how long there is
time to take preventive actions before the windings or magnets
are permanently damaged.
Conventional
7000
6000
Azipod
5000
4000
30
40
50
60
70 Transit
70
60
50
40
30 Dyn. pos.
Rel. time in transit and dynamic position
Fig.3. Comparison of fuel oil consumption (FOC) with varying
operation profiles
III. METHOD OF TEMPERATURE ANALYSøS
The fast development of computers has made it possible to use
numerical methods of analysis for the thermal design of electrical
machines [1,2]. In the present paper, I-DEAS––TMG commercial
software is used to predict the temperature rise of the motor in
steady state and a lumped-parameter model for fault conditions.
Calculation of losses
The losses in the iron core are calculated with finite element
method using the computer program developed by Arkkio [3].
The magnetic field in the core of the motor is assumed twodimensional. The three-dimensional end-region fields are
modelled approximately by using constant end-winding
impedance in the circuit equations of the windings.
It is assumed that the current density in the stator conductors is
constant. This means that the skin effect in the stator winding is
neglected.
The equations for stator and rotor magnetic fields are written in
their own reference frames. The solutions of the two field
equations are matched with each other in the air gap. The rotor is
rotated at each time-step by an angle corresponding to the
mechanical angular frequency. The rotation is modelled by
changing the finite element mesh in the air gap.
The laminated iron core is treated as a non-conducting,
magnetically non-linear medium, and the non-linearity is
modelled by a single valued magnetisation curve. Thus, the
various loss components usually included under the title ““iron
losses”” are not taken into account when solving the magnetic
field. The iron losses are estimated afterwards from the time
variation of the magnetic field in the iron core.
The eddy current losses in permanent magnets are also
calculated with finite element method. It is assumed that the axial
total current flowing through each magnet is zero. This means
that the eddy currents in magnets do not close through the rotor
iron.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 17 -
II.1. Test motor
II.2. Thermal model
The test motor is a totally enclosed water-cooled permanent
magnet motor. The motor is not directly cooled through its frame
to the water as in a real Azipod system. The cooling water flows
in the ducts situated in the bottom of the stator slots (Fig. 4a). It is
assumed in the calculations that all the losses go out of the motor
through the cooling water ducts and nothing is going through the
frame to the environment. The permanent magnets on the
surface of the rotor yoke are covered and protected by laminated
pole shoes. Fig. 4b shows the cross-sectional geometry of the
machine.
Heat transfer.
The demanding task in the thermal analysis of an electrical
machine is to define the heat-transfer coefficients on the surfaces
of the solid bodies. The coefficient on the air-gap surfaces of the
stator and rotor is defined in terms of a dimensionless Nusselt
number Nu, the radial air-gap length lg and the thermal
conductivity of air kair
The motor was tested in the factory testing field. An inverter
feeds the test motor. The parameters of the test motor are:
Rated power = 820 kW
Rated voltage = 721 V
Rated current = 868 A
Number of pole pairs = 3
Frequency = 8,8 Hz
Winding connection star
B
B
B
hair gap = Nu·kair/lg.
B
B
B
B
B
(1)
B
The Nusselt number for the convective heat transfer between two
rotating smooth cylinders is given by [4]. In an actual machine,
however, there will be a greater heat transfer across the air gap
than described by the smooth cylinder equations, because there
are additional fluid disturbances caused by the winding slots.
Experimental results [5] suggest that the slotting will cause an
approximately 10% increase in the heat transfer. The Nusselt
numbers for small air-gap machines are thus obtained from the
modified expressions
Nu = 2.2 if Ta < 41,
U
Cooling water
Nu = 0.23 Ta0.63Pr0.27 if 41 Tad 100.
P
Winding
(2)
U
P
P
P
(3)
The dimensionless Taylor Ta and Prandtl Pr numbers are defined
from the air-gap dimensions and fluid properties [5]. The fluid
properties, which are temperature dependent, are taken at the
expected full load air-gap temperature. The critical value of 41 for
Taylor number refers to the change from a laminar flow, which is
normal for small air gap machines, to a turbulent flow.
In the end-cap region, a single heat-transfer coefficient is used to
model the heat transfer to and from all the surfaces in contact
with the circulating end-cap air. The heat-transfer coefficient [6]
h = 15.5(0.29v + 1) [W/Km2]
P
a.
(4)
P
where v is the velocity of the cooling air in m/s. It can be
estimated from the product of the rotor angular velocity Zr, the
radius rm of the rotor and the efficiency K with which the rotor
circulates the internal air [6]
B
B
v = rmZrK.
B
B
B
B
B
(5)
B
In the calculations, the value K = 0,5 was used.
Three-dimensional Thermal Model
The modelling of the motors is done using I-DEAS Master Series
6.0TM and TMG Thermal AnalysisTM3D heat transfer module is
used for the thermal analysis [7]. The program package solves
the thermal partial differential equations in the solid materials and
allows setting heat-transfer coefficients on the material
boundaries. The material properties may be isotropic or
orthotropic, and may vary with temperature.
P
b.
Fig. 4 The cross-section of the stator slot (a) and a quarter of the
test motor (b).
P
P
P
In our model, a sector of the motor comprising half of a pole pitch
contains 7100 nodes and 41000 elements respectively. Using IDEAS, the computation of one thermal field took about 1200 s.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 18 -
This was considered to be too long for the transient time-stepping
analysis and wherefore I-DEAS program was used only for
steady state analysis.
Lumped-Parameter Model for Fault Conditions
To overcome the very long computing time, a simplified lumpedparameter model was constructed to simulate the fault
conditions. Fig. 5 shows the thermal network. The parameters of
the network were extracted from the I-DEAS model, and the
thermal network was solved using a commercial circuit simulator.
stator
I3
winding
P sw
I4
P pm
cooling
water
permanent
magnets
CC3
sw
PR8p
sw-w
C2
RR2p
pm-a
R5p
R6p
R pm-rc
R1p
Temperature (qC)
R sc-sw
internal
air
R
rc-a
R3p
R R1
sc-a
CC1
rc
P sc
I1
C
C4sc
Fig. 6 shows a comparison between the simulated and measured
results. The good agreement between the predicted and
measured results is evident, confirming the validity of the adopted
thermal model.
C pm
RR9
a-w
R sw-a
PR7p
sc-w
The temperature-rise tests performed for the test motor allow the
validation of the thermal model. The temperature was measured
using 18 embedded thermo-couples and 10 PT-100 thermal
sensors, which were positioned in the stator winding or on its
surface, on the end-winding surfaces, on and below the
permanent magnets, and in air volumes in the end regions. An
infrared camera was also used to determine the temperatures of
such end-region parts to which it was difficult to fix
thermocouples or other sensors.
stator
core
PI2rc
150
140
130
120
110
100
90
80
70
60
50
40
30
20
10
0
M
ave
m
T stator winding
rotor
core
m = minimum
ave = average
M = maximum
Indices of the thermal resistances and capacitances:
m
ave
M
ave
T permanent magnet
Computed results
T air gap
Experimental results
rc = rotor core, sc = stator core, pm = perm. magnets
sw = stator winding, w = cooling water, a = internal air
Fig. 5 Lumped-parameter model for transient thermal analysis.
II.3. Simulations and experimental results
Using I-DEAS TMG, a steady state thermal model was set up for
the test motor. Due to the symmetry, a sector of the motor
comprising half of a pole pitch is modelled. Table I shows the
losses, i.e. the sources of the thermal field, computed for different
parts of the motor in the rated steady-state operation point and in
the fault conditions.
Table 1 Computed losses [kW]
Steady state and fault
conditions
Stator
winding
Stator
core
Rotor
core
Permanen
t magnets
Steady state
820 kW, f = 12 Hz
41.4
2.2
2.6
0.9
85.7
85.7
0.4
0.4
0.5
0.5
0.001
0.001
36.2
52.6
1.07
0.9
10.2
124.6
5.8
2.8
57.6
84.5
0.9
0.8
3.7
134.2
6.3
2.9
Three phase short circuit
- laminated rotor
- solid rotor
Two phase short circuit
- laminated rotor
- solid rotor
One phase short circuit
- laminated rotor
- solid rotor
Fig. 6 Comparison between the predicted and measured
temperatures.
The lumped-parameter thermal model was used to study the
transient temperature rise following a one-, two- or three-phase
short circuit in the terminals of the machine. It was assumed that
after the fault the propeller rotates the rotor at 67% of the rated
speed and that the frequency converter detects the fault
immediately and disconnects the machine from the voltage
supply. As the stator winding is star connected, the currents in
the healthy phases become zero. The electrical transient
following a sudden fault is very short compared to the thermal
time constants. For the thermal analysis, it was enough to
consider the operation of the faulted system in the steady state,
only.
The one- and two-phase short circuits cause a time-varying field
in the rotor. The losses induced in the rotor strongly depend on
whether the rotor yoke is of laminated or solid construction. To
consider this effect, both solid and laminated rotors were
analysed.
Figs. 7-9 show the transient temperatures obtained for the shortcircuit cases. The initial temperatures correspond to the operation
of the motor at the rated load. From the point of view of the
temperature rise, the one-phase short circuit (Fig. 7) is the most
dangerous fault. If this fault stays on, the temperature of the
faulted phase winding rises quickly (in two minutes) over the 200
qC limiting value. The maximum operating temperatures for
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 19 -
currently available permanent magnets are around 150 qC. A
one- or two-phase short circuit lasting for several minutes may
also cause temperature-rise problems for the permanent
magnets if the rotor yoke is made of solid steel.
200
Temperature (qC)
The three-phase short circuit seems to be the least dangerous of
the faults studied. It does not cause a time-varying field in the
rotor, the rotor losses are small, and it does not matter for the
thermal response whether the rotor yoke is made of solid or
laminated steel. One way to avoid the possible damages of the
one- or two-phase short circuits is to connect the stator winding
into a three-phase short circuit immediately when the frequency
converter detects a shorted winding. A properly designed
frequency converter can be used to connect the winding into a
three-phase short circuit. Another way is to turn the sort-circuit
current feeding pod transversely to the course of the ship. After
turning, the propeller is rotating very slowly and short circuit
current is close to zero.
250
150
100
50
0
0
1250
2500
Time (s)
3750
5000
0
1250
2500
Time (s)
3750
5000
250
200
Temperature (qC)
250
Temperature (qC)
200
150
0
50
0
1250
2500
Time (s)
3750
5000
250
200
Fig. 8 Transient temperatures in the permanent magnet motor
following a two-phase short circuit in the terminals of the
machine. Upper figure is for a motor having a solid rotor yoke; the
bottom figure is for a motor with laminated rotor. For the notation
used, see Fig. 7.
250
150
200
stator winding
(faulted phase)
100
50
0
0
1250
2500
Time (s)
3750
5000
Fig. 7 Transient temperatures in the PM motor after a one-phase
short circuit occurs in the machine terminals. Upper figure is for a
motor having a solid rotor yoke; the bottom figure is for a motor
with laminated rotor.
Temperature (qC)
Temperature (qC)
100
50
100
0
150
150
permanent magnets
internal air
stator core
rotor core
100
50
0
0
1250
2500
Time (s)
3750
5000
Fig. 9 Temperatures in the permanent magnet motor after a
three-phase short circuit in the terminals of the machine. For the
notation see Fig. 7.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 20 -
III. CONCLUSøONS
In the paper, the Azipod electric propulsion system and its
benefits are introduced. One problem in electric propulsion
systems is the temperature rise during a fault. When a fault
occurs in a ship propulsion drive, the permanent magnet motor
acts as a generator and transfers some of the kinetic energy of
the ship to the faulted sub-system. In the paper, the temperature
rise during a sudden short circuit in the motor terminals is
studied.
A commercial I-DEAS––TMG software was used for solving the
steady-state temperature field. The losses for the thermal model
were computed using time-stepping finite element analysis. The
temperatures obtained from the steady-state analysis show good
agreement with the measured results.
A simplified thermal network was constructed to study the
transient temperatures following one- two- or three-phase short
circuits in the terminals of the machine. The one-phase short
circuit proved to be the most dangerous fault from the thermal
point of view. There are only some minutes of time to switch off
the short circuit; otherwise, the whole phase winding may be
damaged due to the temperature-rise. In the Azipod system, the
short circuit proved not to be a problem because the time to turn
a pod transversely to the course of the ship is shorter than the
time in which the temperature of the stator winding has increased
to a too high level. The permanent magnets are not so critical as
the stator winding.
IV. REFERENCES
Vehkalahti, Finland, in 1955. He has worked with various research
projects dealing with modelling, design and measurement of electrical
machines. He received the MSc (Tech) and DSc (Tech) degrees from
Helsinki University of Technology in 1980 and 1988, respectively.
Tapani JOKINEN has been the Professor Emeritus since 2001. He was
the Professor of electrical engineering (Electromechanics) at Helsinki
University of Technology (HUT) from 1974-2001. Before that, he was the
associate professor in electrical machines at HUT, design engineer in
the AC machines development department of the company Strömberg
and an assistant at HUT. He was the head of the Department Electrical
Engineering from 1983-85 and the vice-rector of HUT from 1985-88. He
is doctor honoris causa of Tallinn Technical University, Estonia. His
special interest includes induction machines, high-speed electrical
machines, optimisation of electrical machines, creative problem solving,
and product development process. Tapani Jokinen was born in Kärkölä,
Finland, in 1937. He received the MSc (Tech) in 1962, the LicSc (Tech)
in 1967, and the DSc (Tech) in 1973 from Helsinki University of
Technology.
Marian NEGREA is a PhD student at Helsinki University of Technology,
Laboratory of Electromechanics. He was born in Bucharest, Romania on
28 July 1974. He graduated from University Polytechnica of Bucharest in
1998. In 1999, he obtained the MSc (Tech) degree from Helsinki
University of Technology. His research interests include fault diagnosis,
condition monitoring and thermal modelling of electrical machines.
Ingmar WALTZER has been Technology Manager and Quality Manager
at the ABB Electrical Machine factory in Helsinki since 2000. In the
1990ties he has worked as Business Development Manager and in the
1970ties and 1980ties as General Manager for the Strömberg Marine
Division and the Electrical Machine Factory. Waltzer was born in
Stockholm, Sweden, in 1940. He has worked with various projects in
traction, electrical propulsion drives, oil drilling systems and electrical
machines. He received the MSc (Tech) from Helsinki University of
Technology in 1963.
1. Shanel, M., Pickering, S.J., Lampard, D. ““Application of
computational fluid dynamics to the cooling of salient pole electrical
machines””. Proceedings of ICEM 2000, 28––30 August 2000, Espoo,
Finland, Vol. 1 pp. 338––342.
2. Driesen, J; Belmans, R.J.M; Hameyer, K. ““Finite-element modeling
of thermal contact resistances and insulation layers in electrical
machines””. IEEE Transactions on Industry Applications, Vol. 37, No.
1, January/February 2001, pp. 15-20.
3. Arkkio, A. ““Analysis of induction motors based on the numerical
solution of the magnetic field and circuit equations””. Acta
Polytechnica Scandinavica, Electrical Engineering Series no 59,
Helsinki
1987,
97
p.
(The
electronic
version:
http://lib.hut.fi/Diss/198X/isbn951226076X/)
4. Taylor, G.I. ““Distribution of velocity and temperature between
concentric cylinders””, Proc. Roy.Soc, 1935, 159, Pt.A, pp.546-578.
5. Gazley, C. ““Heat transfer characteristics of rotating and axial flow
between concentric cylinders””, Trans. ASME, January 1958, pp.7989.
6. Mellor, P.H; Roberts, D; Turner, D.R. ““Lumped parameter thermal
model for electrical machines of TEFC design””, IEE Proceedings-B,
Vol. 138, No. 5, September 1991, pp. 205 -218.
7. SDRC, ““Product Catalogue I-DEAS Appl. Products –– I-DEAS
TMG™™””, http://www.sdrc.com/pub/catalog/ideas /appl-prod/tmg.
HTU
UTH
BøOGRAPHøES
Antero ARKKIO has been the Professor of electrical engineering
(Electromechanics) at Helsinki University of Technology (HUT) since
2001. Before his appointment as Professor, he has been a senior
researcher and laboratory manager at HUT. Arkkio was born in
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 21 -
Supercapacitors as an energy storage
for fuel cell automotive hybrid electrical system
P. Thounthong, S. Raël, B. Davat
GREEN-INPL-CNRS (UMR 7037)
2, Avenue de la Forêt de Haye, 54516 Vandoeuvre-lès-Nancy, France, e-mail:
Phatiphat.Thounthong@ensem.inpl-nancy.fr
HTU
UTH
determines the feasibility of their use in a particular high power
application [7].
Abstract
The design, implementation and testing of a purely supercapacitors
energy storage system for automotive system having a fuel cell as
main source are presented. The system employs a supercapacitive
storage device, composed of six components (3500 F, 2.5 V, 400A)
associated in series. This device is connected to automotive 42 V
DC bus by a 2-quadrant DC-DC converter. The control structure of
the system is realised by means of analogical and digital control.
The experimental results show that supercapacitors are suitable as
energy storage device for fuel cell automotive electrical system.
Keywords: Automotive, Fuel Cell, Hybrid Electrical System,
Polymer Electrolyte Membrane, Supercapacitors.
This paper presents automotive hybrid system having fuel cell as
main source and supercapacitors as auxiliary source. It
especially details the control algorithm for supercapacitors
converter. The experimental results show that supercapacitors
technology is suitable for providing energy in automotive
electrical system.
iSuperC
iFC
2H 2
O2
I.- INTRODUCTøON
At the present time, automotive hybrid electrical system has been
developed for drastically cleaner and more economical vehicles.
Hybrid electrical cars, such as the Honda Insight and Toyota
Prius, were especially tested by U.S. Department of Energy
(DOE) and showed the fuel saving [1]. Manifestly, fuel cell has
been developed to become the main source in many
applications. The fuel cell transit bus, which has been designed
and developed by DOE, has been acknowledged as a zero
emission vehicle. Its only emission is in fact water vapour [2].
One of the main weak points of fuel cell is its slow dynamics [35]. In fact, the dynamics of fuel cell is limited by the hydrogen
delivery system, which contains pumps and valves, and in some
cases a reforming process. In particular, a step electrical load will
imply huge variation of the voltage of automotive 42 V DC
distribution bus, because the main source has slow dynamic
response. Moreover, the automotive system has problem when
starting electrical motor, which demands the DC bus high energy
in short time. To solve these problems, the system must have an
auxiliary source, to supply high transient energy. The new high
current supercapacitor technology has been developed for this
purpose [6]. Then the very fast power response of
supercapacitors can be used to complement the slower power
output of the fuel cell to produce the compatibility and
performance characteristics needed by hybrid automotive system
as shown in Fig. 1.
iL
2 H 2O
Fig. 1: Fuel cell and supercapacitors hybrid system
II.- HYBRøD SYSTEM STRUCTURE
II.1.- Fuel cell converter
The fuel cell converter, presented in Fig. 2, is a boost converter
used to adapt the low DC voltage delivered by our fuel cell, which
is around 12.5 V at rated power, to the standard automotive 42 V
DC bus. It is composed of a high frequency inductor L1, an
output filtering capacitor C1, a diode D1 and a main switch S1.
Switch S2 is a shutdown device for test bench security to prevent
the fuel cells stack from short circuit in case of accidental
destruction of S1, or of faulty operation of the regulator. Taking
into account the low voltage, we choose power MOSFETs for S1
and S2 [8].
Compared with batteries, supercapacitors have one or two orders
of magnitude higher specific powers, and much longer lifetime.
Because they are capable of millions cycles, they are virtually
free of maintenance. Their great rated currents enable fast
discharges and fast charges as well. Their quite low specific
energy, compared to batteries, is in most cases the factor that
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
iFC ( t )
iD ( t )
iS ( t )
vIN ( t )
Fig. 2: Fuel cell boost converter
iC ( t )
vBus ( t )
IJEET
- 22 -
II.2.- Two-quadrant supercapacitors converter
In previous work [9], one tried in hybrid system built with battery
as main source and supercapacitors as auxiliary source to control
currents in the different parts of the system (battery,
supercapacitors and load). One of the problems, which appear in
such a control, is the presence of dead time operation while the
system changes of operating mode (from steady state to a
sudden recovery state, for example). For new conception, the
hybrid system control presents vBus regulation through the power
delivered by the fuel cell and the supercapacitors [10], and the
current references are a consequence of the power demand.
The supercapacitors are connected to the DC Bus via a 2quadrant DC-to-DC converter, as shown in Fig. 3. L2 represents
the inductor used for energy transfer and filtering. The inductor
size is classically defined by switching frequency and current
ripple. Supercapacitors size is defined by DC bus energy
requirements deduced from hybrid power profile. The current
iSuperC, which flows across the storage device, can be positive or
negative, allowing energy to be transferred in both directions.
Finally, the converter is driven by means of complementary
pulses, applied on the gates of the two MOSFET S3 and S4.
B
B
B
B
More precisely, the DC bus voltage controller (PI controller)
generates a power reference, called PBusREF on Fig. 4. This signal
is limited in level and rate of change, to create fuel cell power
reference PFCREF, and then fuel cell current reference iFCREF1. The
difference between the two previous power references gives
supercapacitors power reference PSuperCREF, and then one of the
three supercapacitors current references, that is to say iSuperC3.
This signal defines supercapacitors modes of operation: normal
(charge from fuel cell) if iSuperC3 is zero, recovery (charge from DC
bus) if it is positive, and discharge if it is negative. The two other
supercapacitors current references, iSuperC1 and iSuperC2, are
generated by fuel cell current controller 2 (PI controller) and
supercapacitors voltage controller (P controller) respectively. The
hybrid control algorithm as explained hereafter does the choice
between these three references.
B
B
B
B
B
B
B
B
B
B
Fig. 3: Two-quadrant supercapacitor converter
B
B
II.3.- Hybrid system control principle
Because fuel cells are supplied with gas through pumps, valves
and compressors, they have large time constants (several
seconds). Consequently, they cannot correctly respond to fast
increasing or decreasing power loads, and may be damaged
because of repetitive step power loads. For this reason, the fuel
cell, in our hybrid system, is only operating in nearly steady state
conditions, and supercapacitors are functioning during transient
energy delivery (motoring) or transient energy recovery
(generating).
B
B
B
B
B
B
B
B
B
:4 2V
Rate of
DC Bus Voltage
Change
Controller
~p
B u sR E F
PI
B
B
B
B
B
y
limited
B
B
B
B
~
iF C R E F 1
~p
FCREF
-
B
B
I
Soft Start
B
B
Fig. 4: Hybrid system control structure
~
v B usR E F
B
B
B
B
B
B
B
Firstly, during normal operation, supercapacitors are charged by
the fuel cell up to the voltage level vSuperCNormal, which is within the
previously defined interval [vSuperCMin , vSuperCMax]. To meet this
target, fuel cell current controller 2 is supplied with fuel cell rated
current as reference, iFCRated, corresponding to fuel cell rated
power. Supercapacitors voltage controller is supplied with
vSuperCNormal as reference. The hybrid control algorithm leads to
select the minimum value among iSuperC1 and iSuperC2 if
supercapacitors voltage is less than vSuperCNormal (that is to say if
iSuperC2 is positive), zero otherwise. Note that during this
operation, charging current has to be limited in rate of change, in
order to avoid unstability due to a too fast increasing current
which would be seen as a peak load by the system. Note also
that each transition in the normal mode begins with the
initialisation of the integrator of fuel cell current controller 2.
The control principle of the hybrid structure is presented in Fig. 4.
The main point of this control principle is to regulate DC bus
voltage vBus with the following constraints: fuel cell electrical
power must be kept within an interval [PFCMin , PFCMax],
supercapacitive storage device voltage must be kept within an
interval [vSuperCMin , vSuperCMax], and fuel cell current slopes must be
limited to a maximum absolute value.
B
B
FC Current
Controller1
-
~
d
P ID
Fuel Cell
Processor
H
v~ B u s
PWM
S1
FC
Converter
~
iF C
-
~p
S u p erC R E F
~
v S u p erC M ea
~
iF C M e a
~
iF C R E F 2
Rate of
Change
~
iF C M e a
v~ F C M e a
~
-
FC Current
iS u p e r C 3
Controller2 ~
iS u p e r C 1
PI
Control Signal
~
v S u p erC R E F
y
SuperC Voltage Controller
~
iS u p e r C 2
-
P
v~ S u p e r C M e a
~
iS u p e r C R E F
Hybrid Control
Algorithm
limited
~
vFC
SuperC Current
Controller
S 3
1
-
0
S 4
~
v S u p erC
SuperC
Converter
~
iS u p e r C
~
v Su perC R E F
dSPACE
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
Analogue
B
IJEET
- 23 -
Secondly, when one of the two limitations (rate of change, and
level) on fuel cell power reference is working, a non-zero iSuperC3
signal is generated, which can be positive or negative, depending
on power condition at DC bus. Therefore, in the case discharge
mode, characterised by a transient fast increasing load, or by a
power load greater than PFCMax, the current reference iSuperC3
become negative in order to transfer the lacking energy to the DC
bus. Supercapacitors voltage controller is supplied with vSuperCMin
as reference, and the hybrid control algorithm leads to select the
maximum value among iSuperC3 and iSuperC2 if supercapacitors
voltage is greater than vSuperCMin, zero otherwise. In the case
recovery mode (transient fast decreasing load, or power load less
than PFCMin), the current reference iSuperC3 become positive.
Supercapacitors voltage controller is supplied with vSuperCMax as
reference, and the hybrid control algorithm leads to select the
minimum value among iSuperC3 and iSuperC2 if supercapacitors
voltage is less than vSuperCMax, zero otherwise. In the two cases,
the reference of fuel cell current controller 2 is set to zero, in
prevision of the next normal operation.
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
Fig. 6 presents the fuel cell response to a load disturbance
without the auxiliary source. These results show that the
regulation of the fuel cell converter works correctly, but also that it
generates a sharp transition of fuel cell current, which may be
dangerous for the fuel cell stack.
B
B
B
B
B
CH1: DC bus Voltage [10V/Div]
CH2: Fuel Cell Voltage [5V/Div]
CH3: Fuel Cell Current [10A/Div]
Time: 1s/Div
Finally, note that we have to use a drastically filtered signal (time
constant of several seconds) issued from iSuperC3 to define the
supercapacitors modes of operation.
B
B
Fig. 6: PEM fuel cell converter response to a load disturbance
III.- EXPERøMENTAL RESULTS AND DøSCUSSøON
III.1.- Fuel cell test bench
Fig. 5 shows the simplified diagram of the PEM fuel cell system
used for this research. Constructed by the Zentrum für
Sonnenenergie und Wasserstoff-Forschung (ZSW), Ulm,
Germany, the fuel cell stack is composed of 23 cells of 100 cm2,
as shown in Fig. 5. It is supplied in pure hydrogen (stored under
pressure in bottles) and in air from a compressor. The rated
output power of the system is 500 W, for a rated current of 40 A,
and approximately 12.5 V as output voltage.
P
Cathode
Fuel Cell Stack
Anode
Hydrogen
from bottle
Flow controller
Pressure
controller
Electric
heater
Excess
Humidifier
P
Air from
compressor
Flow controller
Pressure
controller
Excess
III.2.- Hybrid system test bench
Fig. 7 shows the hybrid system test bench. As storage device, we
use six SAFT supercapacitors (capacitance: 3500 F, rated
voltage: 2.5 V, rated current: 400 A, serie resistance: 0.8 m:)
connected in series. The specifications for elementary hybrid
experimentation are as follow: PFCRated = 500 W, PFCMin = 50 W,
PFCMax = 560 W, iFCRated = 40 A, vSuperCNormal = 13 V, vSuperCMin = 8
V, vSuperCMax = 15 V. To test the whole system, and safety
reasons, the fuel cell is replaced by an ideal 12.5 V power supply.
B
B
B
B
B
B
B
B
B
B
B
B
B
B
Fig. 8 shows experimental transient responses of the hybrid
system to a power step from 150 W to 441 W. One can observe
that the DC bus voltage is well regulated, and that main source
current smoothly increases from 13 A to 40 A with a slope of 4
A.s-1. Furthermore, during transient state, supercapacitors
transfer energy to the DC bus in order to compensate the lacking
energy, which is not supplied, by the main source.
P
P
Heat exchanger
Fig. 5: Simplified diagram and stack of our 500 W PEM fuel cell
system
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 24 -
SAFT Supercapacitors
FC Current Controller1
FC Converter
SuperC Current Controller
SuperC Current [A]
SuperC Converter
SuperC Voltage [V] Main Source Current [A] DC Bus Voltage [V]
iSuperC1 to iSuperC2 occurs at time t = 140 s, for a supercapacitors
voltage of nearly 13 V, because of the use of a high proportional
gain (3500) for supercapacitors voltage controller.
42.0
31.5
21.0
10.5
0.0
40.0
30.0
20.0
10.0
0.0
14.0
13.0
12.0
11.0
10.0
9.0
8.0
30.0
20.0
10.0
0.0
-10.0
0
20
40
60
80
100
time [s]
120
140
160
180
200
Fig. 9: Supercapacitors charge from 9 V to 13 V
Fig. 10 presents experimental transient responses of the hybrid
system to an excessive load. It shows that the supercapacitors
compensate the main source during both transient state and
steady state, because of fuel cell current slope limitation and fuel
cell power limitation respectively.
42.0
31.5
21.0
10.5
0.0
800
Load Power [W]
40.0
30.0
20.0
10.0
0.0
13.5
13.0
12.5
12.0
20.0
0.0
-20.0
-40.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0 6.5
time [s]
7.0
7.5
8.0
8.5
9.0
9.5
Fig. 8: Hybrid system responses to a power step from 150 W to
441 W
When the steady state is obtained (at the time t = 9.5s),
supercapacitors current becomes zero, because supercapacitors
voltage is greater than vSuperCNormal.
Fig. 9 presents hybrid system characteristics during normal
operating mode, through supercapacitors charge from 9 V to 13
V. The DC bus has a constant power load of about 120 W. It can
be observed that main source current slope is approximately 0.7
A.s-1, which is lower than the previous 4 A.s-1, necessary
condition for stability. Besides, the transition of iSuperCREF from
10.0
SuperC Current [A] Main Source Current [A] DC Bus Voltage [V]
SuperC Current [A]
SuperC Voltage [V] Main Source Current [A] DC Bus Voltage [V]
dSPACE
Interfacing Card
Fig. 7: Hybrid system test bench
600
400
200
0
42.0
31.5
21.0
10.5
0.0
50.0
40.0
30.0
20.0
10.0
0.0
20.0
0.0
-20.0
-40.0
-60.0
0
4
8
12
16
20
24
28
32
36
40
time [s]
Fig. 10: Hybrid system response when overloading
During the first interval, the current delivered by the main source
slowly increases (with a controlled slope) up to its maximum
value, the lacking energy being delivered by supercapacitors.
Then, the sudden decrease of the power load leads to a recovery
mode for supercapacitors, in order to allow a slow controlled
decrease of the main source current.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 25 4. T. A. Nergaard, J. F. Ferrell, L. G. Leslie, J. S. Lai. Design
considerations for a 48 V fuel cell to split single phase inverter
system with ultracapacitor energy storage, IEEE-PESC'02,
Queensland, June 2002.
5. J. Lee, J. Jo, S. Han. A 10KW SOFC-low voltage battery hybrid
power processing unit for residential use, the 2003 Fuel Cell
Seminar, pp. 33-40, Miami, November 2003.
6. M. Ortúzar, J. Dixon, J. Moreno. Design, construction and
performance of a buck-boost converter for an ultracapacitor-based
auxiliary energy system for electric vehicles, IEEE-IECON'03,
Roanoke, November 2003.
7. B. Destraz, P. Barrade, A. Rufer. Power assistance for diesel electric locomotives with supercapacitive energy storage, IEEEPESC'04, Aachen, June 2004.
8. P. Thounthong, S. Raël, B. Davat. Conception et réalisation d'un
convertisseur statique basse tension pour pile à combustible de type
PEM, EPF'04, Toulouse, September 2004.
9. M. Y. Ayad, S. Raël, B. Davat. Hybrid power source using
supercapacitors and batteries, EPE'03, Toulouse, September 2004.
10. W. Choi, P. Enjeti, J. W. Howze. Fuel cell powered UPS systems :
design considerations, IEEE-PESC'03, Acapulco, June 2003.
11. W. Friede, S. Raël, B. Davat. PEM fuel cell models for supply of an
electric load, ELECTRIMACS'02, Montréal, August 2002.
Fig. 11 corresponds to a sudden recovery of energy on the DC
bus. This energy is recovered by the supercapacitors while a
slow decreasing of the main source current is performed. In this
example, the main source never delivered less than 50 W in
order to maintain fuel cell converter operating in continuous
current mode.
SuperC Current [A] Main Source Current [A] DC Bus Voltage [V]
Load Power [W]
100
0
-100
-200
-300
-400
42.0
31.5
21.0
10.5
0.0
15.0
10.0
5.0
0.0
30.0
20.0
10.0
0.0
-10.0
0
4
8
12
16
20
24
28
32
36
40
time [s]
Fig. 11: Hybrid system response when recovering
IV.- CONCLUSøON
The major objective of this work is to propose a way of controlling
an automotive DC bus supplied by a hybrid source using
supercapacitors as auxiliary source, in association with a PEM
fuel cell as main source, knowing that this kind of electrical
source is not able to supply energy during fast transitions of load
because of current slope limitation, during peak loads because of
power limitation, and during recovery because of only positive
current.
The experimental results relative to our 500 W PEM fuel cell
confirm its slow dynamics. And results carried out by means of
our hybrid system test bench, which uses a storage device
composed of six SAFT 3500 F supercapacitors connected in
series, validate the control principle we use for the achievement
of a bi-directional hybrid power source.
V.- REFERENCES
1. K. J. Kelly, A. Rajagopalan. Benchmarking of OEM hybrid electric
vehicles at NREL, prepared for the DOE, Contract No. DE-AC36-99GO10337, August 2001, website : http://www.eere.energy.gov.
2. U.S. Department of Energy. ThunderPower bus evaluation at
SunLine Transit Agency, DOE/GO-102003-1786, November 2003,
website : http://www.eere.energy.gov.
3. R. Gopinath, S. Kim, J. H. Hahn, M. Webster, J. Burghardt, S.
Campbell, D. Becker, P. Enjeti, M. Yeary, J. Howze. Development
of a low cost fuel cell inverter system with DSP control, IEEEPESC'02, Queensland, June 2002.
HTU
HTU
UTH
UTH
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
- 26 -
IJEET
- 27 -
High-acceleration linear drives: Application to electromagnetic valves
C. Bernez1, X. Mininger1, H. Ben Ahmed1, M. Gabsi1,
M. Lecrivain1, E. Gimet2, E.Sedda2
P
P
P
P
P
P
P
P
P
P
P
P
P
P
1
P
- SATIE (UMR 8029 CNRS) - ENS Cachan – France
2
- PSA Peugeot Citroën – France
P
P
P
Abstract:
In this paper, the feasability of replacing the mechanical valve
actuating system with an electromagnetic one is studied. After a
mechanical study done in order to evaluate power and
consumption needed, several electromagnetic actuators are
examined to evaluate their viability for the specified application.
I. INTRODUCTøON
One of the present priorities in the automotive industry is to lower
exhaust emissions and vehicle fuel consumption. While only 12%
of the world's population own a car, ground transportation is still
responsible for one-fifth of all carbon dioxide emissions. Through
2020, a 60% growth in this value has been predicted by experts
(which would correspond to an additional 1.1 million vehicles!).
Against this backdrop, a large number of electrical systems are
being developed; these systems introduce adjustments above
and beyond the conventional mechanical systems they have
been intended to replace.
Over the past several years, electronic ignition systems have
gradually supplanted contact-breaker ignitions on gasolinepowered vehicles since they serve to facilitate the setting of
sparking advance at all engine speeds, in addition to improving
engine performances while reducing fuel consumption and air
pollution.
The world's auto manufacturers share the same objective today
as regards electromagnetic valves; for the time being, these
valves are actuated by means of a camshaft. It should be
recalled that with the four-stroke engine, one operating cycle
comprises two crankshaft rotations, during which both admission
and exhaust valves must be actuated. The camshaft + timing
chain assembly imposes a fixed set of valve opening and closing
angles (the camshaft rotates by 360° during an operating cycle,
whereas the crankshaft rotates by 720°). These angles cannot
therefore be set in accordance with a number of various
parameters, such as engine speed and external temperature,
even though the potential beneficial impact exerted by such
parameters is readily apparent. At present, separate driving
solutions are being considered for each valve. In this instance,
motion would no longer be obtained by a common device (the
camshaft), but rather by means of a valve-integrated
electromechanical device.
We will first lay out a global mechanical study of the valve
problem and then proceed by examining and detailing the various
electromagnetic structures capable of satisfying the set of
specifications.
Fig. 1: General view of the motor / Close-up on the valves
II. MECHANøCAL STUDY
II.1. Problem statement
The actuator must enable shifting a valve of mass M over a
distance 'X during a time 'T. The approach herein will search
for the order of magnitude of both the powers and mechanical
forces needed to be generated in order to satisfy the set of
specifications.
In this section of the article, we will neglect all types of frictions
and losses. The power values determined will thus always be
underestimated in comparison with the actual values to be
produced. Moreover, we will not consider the possibility of reinjecting energy from the system; in this manner, power
consumption would be reduced.
The motion being sought in this problem set-up is a very highacceleration controlled, alternative linear motion. Figure 2
illustrates such a type of motion. The various valve switching
times (transition from a closed position to an open position and
vice versa: Tf and To) as well as the lengths of time held in the
closed and open positions (Tpf and Tpo) depend on the rotational
speed of the thermal engine (engine speed). For example, Figure
3 indicates a few orders of magnitude for these time values. It is
to be recalled that the switching time is set regardless of engine
speed (3.5 ms); the time where the valve is held in the closed
position (Tpf) is generally longer than that associated with the
open position time (Tpo). This approach has already allowed
drawing an initial observation: the power applied is very closely
related to switching, with consumption of the actuation system
being basically imposed by the closed position time (Tpf).
We will begin by studying a simple analytical case, which serves
to determine the feasibility of an electrical actuator on its own or
the necessity of using a power storage supply.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 28 -
actuator
'x
v
M
Valve position
OFF
Fig. 4: Speed and acceleration profiles
stroke 'x
ON
Tf
Tpf
To
Tpo
time
Fig. 2: Actuator principle diagram and valve position during an
engine cycle
Numerical applications have been computed with this set of
­ M 158 g
values ° T 3ms
®
°'X 8mm
¯
Knowing that
Tpf [ms]
Tpo [ms]
Tf=To [ms]
100
80
60
40
20
0
4.5
3.5
2.5
1.5
0.5
-0.5
1000 2000 3000 4000 5000 6000
Engine speed [RPM]
Fig. 3: Valve cycle characteristic times
II.2. Simple case analytical study
'X
T
³
0
v(t ).dt , we can obtain the maximum
speed Vmax reached during the motion:
§
·
¨
¸
'
X
1
Vmax
(1)
T ¨ D1 D2 ¸
¨ 1 ¸
© 2 2 ¹
The maximum force derived during the motion is then:
§
·
¨
¸
'
M
.
X
1
Fmax M.Jmax
(2)
¨ D D ¸
2
D1.T ¨ 1 1 2 ¸
© 2 2 ¹
We are now in a position to deduce the maximum mechanical
power Pmax to be provided during shifting:
B
B
B
B
§
·
¸
'
M
.
X
1
¨
¸
Pmax M.J max.Vmax
D1.T3 ¨¨ 1 D1 D 2 ¸¸
© 2 2 ¹
2
2¨
Let's use the acceleration profile (and corresponding speed) as
shown in Figure 4. The acceleration J enables identifying the
other characteristic values of the system, such as the maximum
force attained during motion, the maximum power, and the
distance covered. Parameters D1.T and (1-D2).T represent the
end of the acceleration phase and the beginning of the
deceleration phase, respectively, with T being total travel time.
B
B
B
B
Remark: In this section, all curves have been given with
D1 = D2= D. We will consider the problem to be symmetrical with
respect to the time t = T/2. Moreover, all values are plotted using
S.I. units.
B
B
B
B
Fig. 5: Maximum force reached during the move
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
(3)
IJEET
- 29 'X ddistd 'X H
(6)
2
2
x The maximum power reached during motion must not
exceed a fixed value PmMax.
x The maximum force must be less than FmMax.
B
B
B
The sequential process used in genetic algorithms for minimising
Pd is shown in Figure 9.
When applying a deterministic method, the result always takes
the form of a local minimum that depends upon initial conditions.
The results presented have been obtained from a stochastic
method: the genetic algorithm, which yields (among other things)
a global minimum. Previous analytical results, taken as the
reference, will then be used to verify the numerical results. The
numerical constraints are: PmMax = 3,000 W, dist = 4 mm,
H = 0.1 mm, and FmMax = 1,500 N. The profiles presented below
(Figure 8) have been given for these specific values; two cases
are shown: the acceleration J(t) is subdivided into 20 and 50
samples, respectively. We obtain the following objective function
Pd and calculated constrained values:
B
B
B
Fig. 6: Mean and maximum powers necessary to ensure the
dynamic
B
At this point, the average dynamic power to be provided during
the motion can be calculated (let's recall herein that energy is
supplied for both acceleration and deceleration):
T/ 2
Pd 1 ³0 M.J(t).v(t).dt
(4)
T
The final expression for Pd can thus be deduced as follows:
B
Pd
2
M.Vmax
T
B
§
¨
M.'X 2 ¨
1
T3 ¨ 1 D1 D2
© 2 2
·
¸
¸
¸
¹
2
B
B
B
B
Pd [W]
443
444
B
20 samples
50 samples
B
Table 1:
Pmax [W] dist [mm] Fmax [N]
2972
4
1498
2999
4
1500
B
B
B
B
(5)
A couple of findings result from this approach, namely:
x Changing the acceleration profile (via D1 and D2)
makes it possible to reduce by approximately two-thirds
the average power over the displacement and by onehalf over Pmax (see Figure 6).
x The highest mechanical power reached during the
motion lies around 3 kW. This value proves to be
substantial, considering the imposed space limitation.
The force Fmax may exceed 1,500 N when typical
prototypes yield about half this amount within the same
volume.
The acceleration profile will thus be optimised in order to
minimise powers while imposing realistic constraints on the
forces provided by the actuator.
B
B
B
B
B
B
B
B
B
Fig. 7: Differences between 20- and 50-sample profiles
II.3. Optimisation of the acceleration profile
The objective herein is to minimise the average dynamic power
Pd by modifying the acceleration profile J. Each point forming the
(sampled) temporal acceleration profile constitutes a variable
used to minimise Pd.
Several constraints must be verified during this optimisation
process:
x The valve must move over a distance 'X/2 during time
'T/2. In numerical optimisation methods, this
constraint is expressed by relation (6) below.
Minimisation leads to deriving the lower distance
constraint. We then impose 'X/2 as the lower limit in
order to fulfil performance specifications (the variable
"dist" corresponds to the stroke distance), i.e.:
B
B
B
B
Fig. 8: Acceleration profiles for variable constraints
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 30 a more reasonable value, i.e. about 500 N, which leads
to the conclusion that the analytical study indicates a
580N limit (Figure 5) when the valve is propelled over
the entire half-period [0, T/2] (D1 = 0.5 in this case).
Optimisation start
Initialisation
B
Constraints :
H
The previous PmMax and H values are held constant while the
constraint relative to Fmax is lowered. The genetic algorithm
population still consists of ten items (acceleration profiles), each
represented by 20 samples. The optimised acceleration profile
has been given above (Figure 8) for three different FmMax
constraints. The other calculated characteristic values are as
follows (in all cases, the dist variable is equal to 4 mm):
FmMax
PmMax
B
B
B
Initial population
Random profiles
generation (n=0)
(which verify constraints)
{ J i (t i ) }
i
B
B
{1,...,N}
Table 2:
800
700
633
726
2470
2438
Fmax [N]
Pd [W]
Pmax [W]
New generation creation
based on the previous
(index n)
OBJECTIVE
=
Pd power minimisation
B
B
Pd
Pmax
Fmax
dist
600
977
2325
B
B
B
B
B
B
Constraints?
dist
Pmax<PmMax
Fmax<FmMax
NO
B
B
B
B
With the required forces, which are necessary to satisfy the set of
specifications, now evaluated, we can turn our attention to
examining the various feasible technical solutions.
III. LøNEAR DRøVES FEASøBøLøTY
YES
population(index n+1)
population(index n)
All
population items
lead to the same
acceleration profile?
YES
End
Printing results
Optimised profile: J
Evaluation:
Pd, Pmax, Fmax,dist
Fig. 9: Optimisation algorithm
Many observations can ultimately be derived:
x The profiles with 20 and 50 samples are both very
similar. We are thus not required to use 50 samples,
unless for example in the case of a real prototype
application.
x The average dissipated power Pd is only of relative
importance. If the device allows for energy recovery as
it brakes (from T/2 to T), Pd will equal the losses
generated as valves move a distance of 'X.
x It seems very difficult to reach a strength of 1,000 N
within the assigned actuator volume. We will therefore
gradually reduce the FmMax constraint in order to tend to
B
B
B
B
When forcing Fmax to decrease, the Pd power increases while
Pmax is reduced. This finding is interesting, yet variations are not
of the same order of magnitude (-6% for Pmax vs. +54% for Pd).
As Fmax tends to 500 N, the acceleration profile becomes
constant over the period [0, T/2], which indicates that the
optimisation strategy is no longer exerting an effect.
Evaluation
NO
B
As regards displacement of the plunger, which serves to actuate
the valve, several increasingly-complex solutions may be
envisaged. This series of solutions will not be presented in detail
herein and are intended for more in-depth study subsequently.
x The first solution is based on the use of 2
electromagnets, which tend to attract the plate in
accordance with their respective supply of coils. This
solution would be the easiest to implement yet does
seem unable to offer any freedom in positioning the
plate;
x The second one is quite the same as the first and also
uses 2 electromagnets, but the actuator has been
equipped with permanent magnets (polarised actuator).
This solution, while a bit more complex from a
mechanical perspective, could serve to limit actuator
power consumption if its dimensions have been well
optimised;
x The final solution would call for use of a linear actuator.
This structure is the most attractive since it may allow
for fully-variable lift control, which seems to be
impossible with the first two solutions.
For its advantages, the linear actuator is first envisaged. But what
kind of actuator has to be considered?
B
III.1. The all-electrical solution: Direct drive
B
In what follows, we will be focusing on the approximate intrinsic
performance of the direct drive based on electromagnetic linear
actuators.
B
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 31 -
III.1.1. Various types of actuators and compared performance
levels
H : Parameters reflecting the influence of parasite airgaps and
the drop in magnetic potential within the magnetic circuit.
On the basis of a few simple criteria, both operational and
functional in nature and established subsequent to an analysis
conducted on a large number of existing structures, it is now
possible to define a descriptive nomenclature of electromagnetic
actuators [5, 6].
The following figure depicts an example of the forcedisplacement characteristic, which incites several comments:
x The
force
generated
within
non-polarised
electromagnets uses the normal force component, as
opposed to classical multi-step actuators. As such, the
force generated in the first case is very high for small
airgap values. By setting the saturation induction of the
magnetic circuit at 1.4 T, the force density is equal to
approximately 80 N/cm², while the force density
generated within classical multi-step actuators rarely
exceeds 6 N/cm²;
x The force-displacement characteristic displays strong
non-linearity in both current and displacement. This
aspect raises the problem of how to control the core
dynamic, in particular for high acceleration levels;
x Lastly, this same characteristic exhibits a sizable
decrease with respect to core position. For large
airgaps, actuator efficiency thus proves to be highly
reduced.
Magnetic source of excitation
While the power source is necessarily of the produced current
type (supply), the excitation source may be generated by various
processes. In particular, the source produced by means of
supply (the classical case, accessible excitation winding) or by
rigid magnetisation (excitation by permanent magnets) can be
distinguished, as can that induced by the power source (e.g.
case of induction actuators). In the following series of figures,
three examples are given of linear-type actuators that fulfil the
criteria of this initial classification.
a)
b)
y
Snoy
c)
Fig. 10: Excitation source example: (a) permanent magnets, (b)
variable-reluctance only, (c) induced currents
F/Fo
1
Electromagnetic couplings
0.8
As regards the linear actuators, the type of power supply can be
discerned. This procedure consists of distinguishing the
conversion step W related to both the supply frequency of the
power winding and the total stroke of the mobile & .
0.6
nI/nIs=1.5
0.4
nI/nIs=1
Single-step actuators
In this first category, the two characteristic magnitudes defined
previously are very close to one another ( W |1 ). Actuators of
&
the mobile coil type or even of the electromagnet type are found
in this category.
In the case of variable-reluctance (non-polarised)
electromagnets, it is shown that the normalised attraction force
may be written in the following simplified form [14]:
2
2
F §¨ nI ·¸ . H
F0 © nIs ¹ yH 2
with:
F0 1 B2 .Snoy :
2P0 s
saturated magnetic circuit;
(7)
Maximum force obtained for a
0.2
nI/nIs=0.5
0
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Core position / stroke
Fig. 11: Example of single-step actuator: electromagnet with
plunger and it’’s normalised force
Multi-step actuator
This category features complete dissociation between power
supply frequency and the stroke (
W 1 ).
&
The vast majority of
both rotating and linear electromagnetic actuators lie within this
category.
Classification of such actuators may be performed on the basis of
electromagnetic coupling type, which reflects the type of
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 32 -
interaction in both the power and excitation sources that serve to
generate a force with a nonzero average value.
The following figure provides two main couplings: a ““split-winding
polar”” coupling and a ““global-winding split”” coupling.
l
Based on this simple formalist description, it is possible to
examine the compared performances of the previously-defined
couplings.
As a means of application, we will study the typical operating
case, i.e.: G.AL |50 A²/mm3, B f 0 | 0.7 T, ropt |10 mm,
P
Wp
Kf | 1.4, kfv | 0.22 (stepwise current supply). Results are shown
in the following figure for various configurations and couplings.
B
hb
R
B
B
B
It should be pointed out that for the single-airgap ("classic" case)
polar coupling, the volumic force is at best constant, at a value of
350 N/dm3. The split coupling exhibits better performance, i.e.
constant volumic force for a single-airgap architecture and an
increasing force with respect to volume for multi-airgap
architectures.
r
W
P
a)
P
Wp
F= 10
+
P
4
F= 10
100000
Ws
F= 10
7
F= 10
9
F= 10
11
F/V [N/dm3]
(4)
2
10000
W
(3)
1000
F= 10
(1)
0
100
(2)
10
b)
V [dm3]
Fig. 12: Electromagnetic couplings (tubular synchronous actuator
with permanent magnets): ““split-winding polar”” where W |1
1.E-02
Wp
1.E+02
1.E+04
1.E+06
1.E+08
1.E+10
Fig. 13: Volume force vs. volume for several achitectures: (1)
Single-airgap polar coupling with fixed radius, (2) Single-airgap
polar or split coupling with variable radius, (3) Single-airgap split
coupling with fixed radius, (4) multi-airgap split coupling
(a), ““ global-winding split”” where W 1 (b)
W
p
III.1.2. Performance: Analysis using similarity laws
It has been shown [3], by means of a qualitative analysis using
similarity laws based on a series of simple geometrical, magnetic
and thermal considerations, that both the force and volumic force
generated in electromagnetic actuators can be written in the
following form:
§W · 3
(8)
F|Kf .AL.G0,5.Bf0.¨¨ p ¸¸."r 2
© W ¹
F |K .A .G0,5.B .§¨ Wp ·¸. 1 with rvR
(9)
fv L
f0 ¨ W ¸
Va
©
¹ r
where:
K f and K fv : Coefficients dependent upon the type of power
1
1.E+00
Remarks:
The multi-airgap structure features a global-winding split
coupling, for which the number of airgap surfaces remains high.
These structures may appear, in particular, in either tubular form
(so-called "multi-rod" architecture) or planar form ("multi-plate"
architecture). This latter configuration has been illustrated in the
following Figure 14 [7].
U
supply and the relevant induction winding
production technologies
A L : Armature electric loading (in A/m)
G : Current density (in A/m²)
Bf0 : Maximum value of airgap excitation induction
" : Active actuator length
r : Airgap radius
Wp and W represent the pole pitch and electromechanical
conversion step, respectively.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
Fig. 14: Example of multi-airgap architecture
IJEET
- 33 Despite being qualitative in nature, this result clearly indicates the
superiority, in terms of both use of space and acceleration, of
direct actuation systems and especially the system featuring
multi-airgap architecture.
III.2. The all-electrical solution: Indirect drive
III.2.1. Operating principle
Various mechanical solutions are available for developing indirect
electromagnetic actuation. Such solutions often entail coupling a
rotating motor with the linear displacement charge by means of a
rotation/translation motion transformation system. Among such
intermediary systems, the following warrant mention: rod-crank,
pinion-rack and screw-nut (which is the most commonly used,
see Figure 15).
V [dm3]
M=0.16 kg
100000000
(a)
10000000
1000000
100000
10000
1000
100
10
(b)
1
0.1
1
10
100
(c)
1000
0.01
0.001
J [m/s²]
a)
acceleration
at fixed torque
2 M .J m
Torque
at fixed acceleration
Cmin 2J M.Jm
kt
M
Jm
b)
Fig. 15: Diagram of the operating principle of an indirect screwnut transmission actuation (a), Torque and acceleration vs.
transformation factor kt (b)
B
B
III.2.2. Performance comparison
Let's now take the case of a screw-nut system (with the inverse
of the screw step being denoted kt) and examine its dynamic
performance, which can be simply compared on the basis of the
acceleration capacities of the various solutions.
It can thus be demonstrated that the active volume of the rotating
motor (not including the transformation system), along with the
optimal transformation factor, vary as a function of both the
imposed acceleration and the actuated mass, in accordance with
the following law:
B
B
B
B
B
III.3. Application of the all-electrical solutions to
electromagnetic valves
Force profiles, which are necessary in order to propel the valve,
have been established, and the feasibility of a linear actuator
propelling the valve on its own will thus be studied.
In the present case, F is greater than 600 N. Since the
specifications impose dimensions, we can proceed with an
approximate determination of the shear stress VT:
F
VT F
|67N/cm2
(14)
Sa § 2 D2 ·
1
L
S¨ ¸
¨ 4 4 ¸
©
¹
B
B
B
U
D1
L
J max
Cm
Fig. 16: Actuator volume vs. maximum acceleration for
M = 0.16 kg (with transformation system volume not included and
kt=ktopt) for various configurations: a) classical indirect actuation;
b classical direct actuation; c) multi-airgap direct actuation
Sa
U
Va vJmax 3.M
3
2
(10)
k t _ opt vJ max 2.M
5
3
4
(11)
Fig. 17: Active surface Sa, top view
The volume force is obtained, knowing the third dimension:
F
t 10000 N / dm 3 . For purposes of illustration, with a
V
volume held on the order of 0.1 dm3, the maximum volumic force
is about 500 N/dm3, under the best case scenario according to
Figure 13!
The other characteristic value mentioned above is the specific
actuator acceleration (see Figure 16), which has been given for a
160-g mass. According to Figure 16, for a reasonable value
greater than 1,000 m/s2 (Figure 4), the required volume exceeds
1 dm3. With respect to this latter criterion, the linear actuator
alone cannot move the valve. At the present time, it would seem
very difficult, or even impossible, to produce a linear actuator
capable of moving the valve on its own; hence, in order to meet
P
For direct actuations (without any intermediary), we derive the
following:
: fixed-radius polar couplings
(12)
Va J max.M
6
Va J max 7 .M 7
6
: multi-airgap split couplings
(13)
P
P
P
For example, Figure 16 presents the evolution in active volume of
the actuator vs. the desired level of acceleration for an actuated
mass of 0.16 kg.
P
P
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
P
P
IJEET
- 34 in their vehicles (e.g. Porsche, Honda, Toyota, BMW). Only BMW
[4] has actually obtained a fully-variable valve displacement.
Lowering fuel consumption and exhaust emissions (by about
10%), while at the same time improving performance and engine
smoothness, are expected from this approach.
Electromagnet actuators are unable to control valve lift since their
strength profiles decrease too quickly with respect to
displacement (excessive airgap). A linear actuator would
therefore be required.
specifications, associating an energy supply reserve with an
actuator has been envisaged.
III.4. Hybrid solution:
electromagnets + spring
direct
linear
actuator
+
We will now study a spring + electromagnetic actuator assembly.
A flow chart for this solution set-up is given in Figure 18. Multiple
structures can be imagined. One of them is drawn on Figure 19
[1], as example.
The spring may be sized so as to shift the valve by itself, in which
case the actuator only generates the mechanical losses and
holds the valve in either the closed or open position. This solution
satisfies the specifications and is relatively simple to employ.
Why then should a linear actuator, which is a more complicated
system, be used? The answer lies in its ability to perform lift
control.
[
Air admission
Admission
valve
Throttle butterfly
Combustion
chamber
Spring
Linear
actuator
Fig. 20: Simplified diagram of an admission system
III.4.2. Dynamic study
This section focuses on the feasibility of associating a linear
actuator with springs.
Hypothesis: The acceleration displays a rectangular profile (like
in part II.2). The interval is limited to [0, T/2], with the remaining
interval [T/2, T] being symmetrical. Acceleration is constant from
t = 0 to t = D.T. The spring is characterised by both its stiffness Kr
and length x0, as follows:
(15)
Fres K r.(x(t)x 0)
U
Valve
Electromagnet
U
B
Fig. 18: Operating principle of linear actuation with a spring
B
B
B
We impose just one constraint on the linear actuator: force Fm is
limited to a maximum value FmMax. The corresponding
characteristic profiles are shown in Figure 21.
A mechanical limit stop is inserted at position x(E.T) within the
actuator. We set E<D in order to maintain a positive magnetic
force over the interval [0, T/2], which corresponds to the
acceleration phase.
B
B
B
Fig. 19: Example of a linear actuator architecture [1]
III.4.1. Lift control features
In current gasoline-powered engines, the valves can be either
open or closed, which yields two fixed positions. Valve
displacement depends on the geometrical profile of the camshaft.
The amount of air injected into the combustion chamber is
regulated by a throttle butterfly (Figure 20). Such technical
solutions result in power loss. Lift control allows removing this
throttle butterfly from the set-up. Engine output is then entirely
controlled by fully-variable intake valves, which regulate incoming
air volume. Many automobile manufacturers offer similar systems
Fig. 21: Characteristic profiles for an actuator with a spring
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
B
IJEET
- 35 -
Mathematical expressions of the profiles
We are confronted here with two strengths: Fres generated by the
spring, and Fm by the linear actuator. Let's write the motion
equation:
M.J(t) Fm(t)Fres(t)
(16)
B
B
B
­ k .t
®
¯Vmax
x(t )
t  >0,D .T @
2
0
'X
x
Linear
actuator
-FmMax
Spring
Upper electromagnet
Fig.22: Forces profiles
(18)
Introduction of constraints into the equations
Vmax
J
D.T max
(19)
And
x(D.T) 1 k.(D.T)2 1 Vmax.D.T
(20)
2
2
The temporal expression of the magnetic force can now be
calculated as:
1 2
­
°M .J max K r ( 2 k .t x0 ) , t  >0, E .T @
°
, t  @E .T ,D .T @
®M .J max
°0
, t  @D .T , T / 2@
°
¯
(21)
Study objectives
The linear actuator delivers significantly lower forces than the
spring within the volume assigned by these specifications. We
can expect a force of about 100 N in our volume, i.e. one-sixth of
the total force required to move the valve. We should thus allow a
considerable portion of the motion path [0, 'X/2] over which only
the linear actuator delivers a nonzero force, in order to position
the valve with the actuator. Moreover, the linear actuator alone
cannot ensure having the valve stay in the open or closed
position, since the equivalent spring delivers its maximum force
at these two extreme positions. A device, capable of developing
high forces over short lengths to hold the valve stuck in extreme
positions, would have to be added.
Two electromagnets will thereby be introduced into the device
(one up, the other down for locking in the closed and open
position, respectively). The force contours delivered by the
spring, along with the electromagnets and linear actuator, are
displayed below (Figure 22).
We now impose Fm < FmMax. It would seem a reasonable
approach to set E = D to ensure the relation below (see Figure
21):
(23)
FmMax M.J
B
B
B
B
B
Let '* be the distance over which a force is applied to the mass
in order to accelerate it:
'* x(D.T) 'X . D
(24)
2 1D
This function '*(D) is strictly increasing. D must be reduced so
as to allow the linear actuator to perform its action.
In order to respect the constraint on FmMax, we will introduce next
the following relation for x0:
§ M.Vmax
· V .D .T
(25)
x0 1 ¨
FmMax ¸ max
K r © D .T
2
¹
P
P
B
B
B
B
Results
Force profiles are provided in Figure 23 for Kr varying from
103 N/m to 150.103 N/m. A fixed point that is the consequence of
both the rectangular profile of acceleration and the FmMax
condition is present.
Why do several profiles satisfy the specifications? As spring
stiffness increases, the work generated by the linear actuator
rises as well.
The other curves represent the electromagnetic force for various
injected currents I.
These results reveal the impossibility of controlling valve position
using the previously-described system. The forces required to
ensure valve dynamics are too high in comparison with those
generated by the linear actuator within the restricted volume. In
order to position the valve, the actuator must either overcome the
spring force or compensate for the mechanical work delivered by
the spring. The linear actuator does not seem capable of
ensuring this function.
B
P
P
To reach this level of operations, we would expect to limit the
spring force to reasonable values in comparison with
electromagnetic force values. It would be even better for FEA (the
electromagnet force) to satisfy the following relation:
Fres ( x) FEA (x) , x  >0, x(D .T )@
FmMax
'X
­1 2
° k .t
®2
°̄Vmax .t x(D .T )
With:
v(D.T) Vmax k.D.TŸk
Lower electromagnet
B
t  >D .T , T / 2@
Fm (t )
Forces
B
B
We will now seek to determine an expression for Fm as follows:
d2x(t)
Fm(t) M 2 K r.(x(t)x0)
(17)
dt
The speed and position expressions are given below (as
deduced from acceleration):
v(t )
In this manner, the electromagnets could block (either open or
closed) the valve from all positions.
B
(22)
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
P
B
P
B
B
IJEET
- 36 Another condition must then be applied: the speed v is zero at
the start of the stroke, which then yields the following:
§ K r T ·º 'X
ª FmMax 'X º ª
(31)
. ¸¸» 0
«
».«1 cos¨¨
2 ¼ «¬
2
2
M
»
¬ Kr
¹¼
©
As FmMax changes, the spring must be adapted (by means of Kr)
to verify the specifications.
The characteristic profiles are shown below (Figure 24). In this
case, the changes for Fm and Fres are visible in comparison with
the previous section (III.4.2).
B
B
B
B
B
B
B
Fig. 24: Force, speed and acceleration profiles for constant Fm
B
Fig. 23: Forces for two duty-cycles D : 0.15 and 0.3
III.4.3. An alternative approach
In the previous section, the linear actuator was being partially
used. We will now change the force profile so that its use is
maximised. The actuator will now generate its maximum force
over the entire acceleration phase:
(26)
Fm(x) FmMax ,x>0,'X/2@
The previous expression is maintained for the spring force:
(27)
Fres K r.(x(t)x 0) , with here x 0 'X
2
The position x is characteristic in the motion equation:
2
M d 2x K r.x FmMax K r.x 0 , x>0,'X/2@
(28)
dt
The corresponding solution then becomes:
F
Kr
(29)
x(t) A.cos(Z.t) B.sin(Z.t) mMax x 0 , with Z
Kr
M
where A and B stem from the initial conditions, as follows:
F
­
Ÿ A mMax x0
° x(0) 0
Kr
°
'X § FmMax 'X ·§ § Z .T · ·
°
¸¨ cos¨
¨¨
¸ 1¸
®
2
2 ¸¹¨© © 2 ¹ ¸¹
© Kr
° x(T / 2) 'X / 2 Ÿ B
°
§ Z .T ·
sin¨
¸
°
© 2 ¹
¯
(30)
B
B
Even if the linear actuator were used with its maximum force
throughout the entire acceleration phase, the ratio between the
amount of spring work and total required work remains too low.
Moreover, if the valve could be stopped during its stroke, the
linear actuator would be unable to supply the energy needed to
re-shift the valve into either the open or closed position.
In conclusion, in order to enable valve lift control, a linear
actuator must be inserted into the valve device. We have
demonstrated that the FmMax force as well as the power generated
by this actuator within the allotted volume are too low to ensure
valve dynamics. Adding a spring solves the power problem but
not the force problem.
B
B
The main advantage offered by this structure now seems
impossible to achieve; the subsequent study will consider the two
simpler electromagnet-based solutions.
IV. PRECøSE ANALYSøS OF THE HYBRøD DøRECT DRøVE
IV.1. Unpolarised electromagnet
The aim of this part is to evaluate the opportunity to use an
electromagnetic actuator to move admission valves in IC
engines. Comparing to the mechanical solution, the
electromagnetic one allows independence between valves and
the camshaft, i.e. the valves can be moved at desired time. A
basic solution is an unpolarised structure (Figure 25 and Figure
26 represent the geometry chosen for the following study; several
others structures may be imagined).
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 37 -
x
to stop the plunger in an extreme position (called
locked positions), it is necessary to supply the
electromagnet corresponding, in order to create a force
opposing to the spring’’s force. When the supply is
turned off, the springs push on the plunger and move
the rod, and then the valve.
The profile of the currents injected in the upper (A) and lower (B)
coils is given by Figure 28.
Fig. 25: Geometry of the actuator
Fig. 26: Unpolarised electromagnet
The electromagnets are fixed, and the plunger attached to the
rod, which pushes on the valve, moves between the two extreme
positions. Depending on the plunger position, the springs are
more or less compressed, and the neutral position is in the
middle of the air gap.
This structure is called unpolarised because there is no
permanent magnet. The features of this solution are the
followings:
x without power supply, the permanent position of the
plunger is in the middle of the air gap;
x to start, a little period of oscillation of the plunger is
necessary: the springs start resonating thanks to the
electromagnets, and the plunger is then stuck to the
desired electromagnet. The oscillations are
indispensable because of the small force of the
electromagnet when the plunger is in the middle of the
air gap. The resonance frequency of the springs is
about 150 Hz;
Fig. 28: Normalised profiles for the unpolarised actuator
The study of this structure shows that it is able to develop the
effort to lock the plunger in the extreme positions, which is equal
to the one developed by the springs (Figure 29 is an example of
the effort developed by one electromagnet with a current of 15A
vs the air gap between the plunger and the electromagnet).
There are several drawbacks with this structure. The start, with
the resonance of the springs, needs a specific command.
Besides, the locked position has an important relative duration
(especially for low engine speed, see part II.1) and during this
time, the actuator drains current although there isn’’t any motion.
As another drawback is that this solution is too noisy, because of
the impact velocity, which is too important and hard to control.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 38 -
800.0
700.0
Effort (N)
600.0
500.0
400.0
300.0
200.0
100.0
0.0
0
1
2
3
Air gap (mm)
4
5
15A
Fig. 29: Magnetic efforts developped by a coil
IV.2. Polarised electromagnet
Fig. 31: Normalised profiles for the polarised actuator
IV.2.1. Presentation
Because of these drawbacks and in order to improve this
actuator, a new structure is described in this paper. The structure
is the same as the basic one, but with two permanent magnets
(see Figure 30, [8] and [9]).
Fig. 30: Polarised Actuator
That addition leads to new command strategies. The one
proposed is to lock the plunger in the extreme position only with
the force of the magnets. The plunger is then released by
negative current injection in the coil, in order to opposite its flux to
the magnets one, and to decrease the resultant magnetic force.
The potential energy stored in the springs moves the plunger.
The main advantage of this strategy is that there is no current
consumption in locked position, as no current is injected to
oppose to the springs force. The consumption, which highly
depends on Tpf as it has been shown in part II.1, may then be
decreased, particularly for low and medium engine speeds, which
are the most present in a typical operating cycle for a vehicle.
The profile of the currents injected in the upper (A) and lower (B)
coils is given by Figure 31.
The main drawback of the addition of the magnets is that they
add an air gap, which reduces the magnetic force developed by
the coils. On the other hand, this new air gap decreases the
global inductance, so the current variation in the coils may be
faster for the control. At first, it is then necessary to demonstrate
that this new structure is able to produce the forces needed for
the command strategy. Then it is interesting to verify that the
driving cycle and the dynamic performances are acceptable. An
analytical model is then developed with Matlab Simulink to
evaluate the actuator.
IV.2.2. Analytic model
Mathematical model
The first step needed is to have a mathematical model of the
actuator. The model is divided in two parts, which are
interdependent:
x an electric loop, with a usual electric circuit:
d)
U Ri (32)
dt
where ) corresponds to the flux resulting from the actions of the
coil and the magnets, U is the supply voltage, R is the coil
resistor and i is the current.
x a mechanical loop, which is the Newton’’s second law:
&
¦F
&
&
&
&
Fmag Fsp Ffr mg
&
Fmgt
³SP0[H(n.H) 2 (H.H)n]dS
&
mȖ
(33)
The forces are the magnetic one, from the coil and the magnets,
the springs one, the friction one and the weight force. The
magnetic force is rigorously expressed as a surface integral,
which comes from Maxwell’’s tensor:
& & &
1 & & &
&
(34)
&
where H is the magnetic field, n the unit vector perpendicular
to the portion dS, and S the external plunger’’s surface.
The springs force is the one presented in the previous parts.
Hydraulic frictions are equal to the plunger’’s speed multiplied by
a coefficient of friction, and static frictions are represented by a
hysteresis function depending on the plunger’’s position.
To evaluate the flux in the air gap, which is necessary to get the
electromagnet force, a lumped circuit from the actuator is
introduced.
Equivalent magnetic circuit
To be complete, the magnetic saturation and the flux leakage
have to be taken into account in the equivalent circuit. In this
study, iron losses are neglected. With the help of lines of the
magnetic flux, given by the FE method, several equivalent
magnetic circuits were created. The one used for the study is
presented with the following figures:
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 39 IV.2.3. Actuator control: dynamic of the motion
The Matlab model is able to give quickly the travel of the plunger,
depending on the current injected in the two coils of the optimized
structure.
x current injected only in one coil
The plunger is initialy stuck to the upper electromagnet, and a
negative current is injected in the upper coil (A) to release the
plunger.
Fig. 32: Flux lines
Fig. 35: Stroke with one coil
Fig. 33: Equivalent magnetic circuit
where Rf, Rk1 and Rk2 are the flux leakage reluctances, Rfe,
Rfm and Rp the reluctances in the iron, Ree and Rem the
reluctances in the air, Ra the reluctance of the magnet. Va is the
magnetomotive force.
Use of Matlab Simulink
The plunger does not do the entire stroke of 8mm: because of the
friction, it is not able to reach the lower electromagnet. The
plunger trajectory is then oscillating because of the springs.
x current injected in the two coils
The release current is the same as the previous one. But at an
adequate moment, a positive current is injected in the lower
electromagnet (B), and the plunger is then able to overcome the
friction, and the entire displacement is done.
The last two points are used to establish the model. It has to give
all the forces and the instantaneous position of the plunger in the
air gap. So, there is the electrical loop corresponding to the
electrical equation:
Fig. 36: Stroke with two coils
Fig. 34: Electrical loop
This loop is controlled by the electric out-voltage of a chopper,
which is the output of a closed loop control of the current in the
coil.
Moreover, there is the magnetic loop, which allows evaluating the
flux in the air gap due to the coupled action of the coil and the
magnets. In the first place, the reluctances are calculated
depending on the induction and the plunger’’s position, and the
flux M in the air gap is deducted. The magnetic force is then
calculated with (34). Another block, corresponding to (33), gives
the acceleration of the plunger, and by integration, its speed and
position.
A design process, which consists on modifying the magnets
dimensions in order to be able to use the command strategy
previously presented, is done with the analytical model [12].
In this exemple, it is shown that the entire travel can be done with
an acceptable time. The control has now to be studied more
precisely, by choosing the best currents to inject to get the
desired dynamic, with a controlled consumption. Figure 37 shows
as an exemple the influence of the release current on the travel
time. A trade-off has then to be made between the current
injected for the release and the dynamic.
To catch the plunger, a positive current is injected in the lower
electromagnet (B). This is first necessary to have the entire
displacement, as it has been shown, and also to limit the valve
bounce. Indeed, the electromagnetic force decreases faster than
the springs one when the air gap increases (Figure 23), and then
a bounce not controlled may allow the springs to get the plunger
back. The permanent magnets are only able to ensure the static
efforts.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 40 on direct linear actuator coupled to springs and electromagnets,
concerning the variable lifting of the valve, have then been given.
The use of structures {electromagnets and springs}, though they
also can't do lift control, has been considered under the terms of
their simple design. We have shown that those are viable, and
how permanent magnet addition decreases the electric
consumption of the actuator.
VI. REFERENCES
1.
Fig. 37: Influence of the release current
2.
3.
IV.2.4. Consumptions
The control is still to study: the parameters will be the amplitude
of the currents, the duration and the start of the current pulses, in
order to get the desired dynamic with a minimal power.
Nevertheless, the Figure 38 obtained with a simple model (and
the current profiles shown in Figure 28 and Figure 31) gives an
idea of the power consumption of the unpolarised and the
polarised electromagnet depending on the engine speed. The
consumption of the polarised actuator is always lower than the
polarised one, although the difference decreases with the engine
speed (as expected in IV.2.1).
4.
5.
6.
7.
8.
9.
10.
11.
12.
Fig. 38: Normalized power consumption for unpolarised and
polarised actuator
13.
14.
B. Lequesne: ““Permanent Magnet Linear Motors for short Strokes””,
IEEE transactions on industry applications, Jan/Feb 1996.
B. Lequesne: ““Design and Optimisation of Two-Spring Linear
Actuators””, ETEP, Nov/Dec 1999.
H Ben Ahmed, B. Multon, P.E. Cavarec : ““Actionneurs linéaires
directs et indirects : performances limites””, Journées
d’’Electrotechnique du club EEA « Avion et électricité », 18-19 mars
2004, Université de Cergy-Pontoise.
www.bmw.com. BMW (Car manufacturer official site (Europe)).
Jacek F. Gieras: ““Status of linear Motors in the United Sates””, 4th
Int. Symp. on Linear Drives for Industry Application (LDIA),
September 2003, Birmingham (UK).
P.E. Cavarec, H. Ben Ahmed, B. Multon: ““Actionneurs
électromagnétiques : classification topologiques””, Techniques de
l’’Ingénieur, D3 412.
P.E. Cavarec, H. Ben Ahmed, B. Multon: ““Force density
improvements from increasing ther number of airgap surfaces in
synchronous linear actuators””, Revue IEE proc. Elec. Power Appl.,
vol. 150, N° 1, January 2003, pp.:106-116.
H. Ben Ahmed, M. Gabsi, M. Lécrivain, C. Fageon, E. Sedda:
““Actionneur électromécanique de commande de soupape pour
moteur à combustion interne””, Brevet Français, déposé le 18
Février 2003 sous le n° 0301948, déposant PSA.
M. Lécrivain, M. Gabsi, H. Ben Ahmed, E. Sedda, C. Fageon:
““Actionneur électromécanique de commande de soupape pour
moteur à combustion interne et moteur à combustion interne muni
d’’un tel actionneur””, Brevet Français, déposé le 15 janvier 2004
sous le N° 04 50092, déposant groupe PSA.
H. Hattori, T. Izuo, M. Asano, T. Iida, S. Nitta: ““Electromagnetic
Actuating System””, U.S. Patent 6,334,413, Jan. 1, 2002.
G. Schmitz, F. Pischinger: ““Method for Controlling an
Electromagnetic Actuator Operating an Engine Valve””, U.S. Patent
5,868,108, Feb. 9, 1999.
X. Mininger, H. Ben Ahmed, M. Lécrivain, M. Gabsi, E. Sedda, C.
Fageon: ““Permanent Magnet Actuator for Admission Valve ””,
Electromotion, Volume 10, Number 3, July –– September 2003.
G. Lacroux: ““Les Aimants Permanents””, Tec & Doc, 1989.
M. Juffer: ““Circuits Magnétiques””, Techniques de l'Ingénieur, traité
Génie Electrique, 1995.
V. CONCLUSøON
After having explained the problems of the valves equipping the
thermal engines, and their application framework, we have
presented a panorama of the available electromechanical
solutions to replace the traditional mechanical structure
(camshaft). With additional possibilities of adjustment, these new
structures would lead to a reduction of the consumption and
pollution. In particular, variable lift of the valve would improve
these two parameters of more than 10%, according to some
world car manufacturers [4].
We initially have taken interest in the minimization of the
necessary mechanical power, by acting on the acceleration
profile. We have shown that for the dimensions allocated by the
specifications sheet, an only direct linear actuator does not
ensure sufficient dynamics. The limits of hybrid solution, based
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 41 -
Flywheel Electric Drive Characterization for Hybrid Vehicles
Y. Gao1, S. E. Gay1, M. Ehsani1,
R. F. Thelen2, R. E. Hebner2
P
P
P
P
P
P
P
P
P
P
1
P
- Advanced Vehicle Systems Research Program, Texas A&M University, USA, e-mail: Ehsani@ee.tamu.edu
P
HTU
UTH
2
P
- Center for Electromechanics, The University of Texas at Austin, USA
P
Abstract
Flywheels are envisioned as serious alternatives to electrochemical
batteries and ultracapacitors for energy storage in hybrid vehicles.
The combination of a flywheel with an electric motor drive is
referred to as a Flywheel Energy Storage System or FESS. This
paper is an evaluation of the characteristics of electric motor drives
for flywheels in hybrid vehicle applications. The first section
features an analysis of the energy and power requirements for
passenger vehicles over a standard acceleration profile and several
drive cycles. The second section consists of an analysis of the
parameters influencing the flywheel energy and power. The third
section contains an investigation of characteristics of the electric
motor drives for flywheels.
Keywords: flywheel, hybrid drivetrain, power sizing, energy
storage
bidirectional conversion between mechanical energy stored and
electrical energy used by the hybrid vehicle drive train. The FESS
operates in two quadrants since the angular velocity is always
positive. When the FESS is charging, the motor torque is positive
and accelerates the flywheel. When the FESS is discharging, the
electric motor drive acts as a generator, (i.e. negative torque)
thus extracting energy and decelerating the rotating mass. A
FESS unit of this type has been successfully demonstrated in a
transit bus [5-7] and is being developed for use in a hybrid
locomotive for high-speed passenger rail applications [8-11].
II.
VEHøCLE POWER AND ENERGY PARAMETERS
In a full hybrid drive train, a primary power source (internal
combustion engine or fuel cell) supplies the average power
demand while a secondary power source (energy storage
system) supplies the peaking power demand corresponding to
acceleration and regenerative braking [12]. Fig. 1 depicts this
operating principle.
I. INTRODUCTøON
Flywheels are envisioned as serious alternatives to
electrochemical batteries and ultracapacitors for energy storage
in hybrid vehicles. Flywheel energy storage systems offer
significant advantages compared to electrochemical batteries and
ultracapacitors in hybrid vehicle applications. They have high
specific energy, high specific power [1], long cycle life [2], high
energy efficiency [3], limited sensitivity to contamination, and
require minimal maintenance and minimal system overhead.
While it is possible to connect the flywheel to the hybrid drive
train by using a mechanical port, the preferred configuration uses
an electric motor drive for the coupling. The electric motor drive
directly interfaces with the dc bus of the hybrid vehicle power
system. Because, this configuration shares several similarities
with electrochemical batteries, it is sometimes referred to as a
““flywheel battery”” [4]. The single-port combination of a flywheel
with an electric motor drive is also referred to as a Flywheel
Energy Storage System or FESS.
The FESS is composed of two fundamental functional elements:
the kinetic energy storage element and the power transformation
element. Kinetic energy is stored in the momentum of inertia of
the rotating mass according to:
Wkin
1
2
J rmZ rm
2
(1)
Wkin is the kinetic energy stored, J rm is the moment of inertia
of the rotating mass, and
Z rm
is the rotational speed of the
mass. The electric motor and its power electronics perform the
Fig. 1: Decomposition of driving power demand into base and
peaking power demands
The peak power requirements are determined by considering two
driving patterns:
Maximum acceleration from 0 to a target speed,
typically significantly inferior to the vehicle’’s
maximum cruising speed.
Peaking, i.e. acceleration and regenerative
braking, during standard drive cycle such as FTP
75 Urban and Highway cycles.
For a given vehicle, target speed, and acceleration time, the
maximum acceleration power requirement is determined by the
torque vs. speed profile of the traction motor drive and its
transmission. Fig. 2 shows an idealized profile of traction effort
vs. vehicle speed.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 42 -
Fig. 3: Traction effort vs. speed profile for different X ratios
Fig. 2: Idealized traction effort vs. vehicle speed profile
When the vehicle accelerates from zero to target speed, the
traction effort profile goes through two regions:
- Constant effort region from zero to base speed Vb
-
Constant power region from base speed to target
speed V f
The traction motor drive power rating, Pa , needed to achieve
maximum acceleration from zero to target speed, V f , in t a
seconds with a vehicle mass, M v , is expressed as [13]:
Pa
Mv
V f2 Vb2
2 ˜ ta
(2)
If the ratio of the vehicle’’s maximum speed to the vehicle’’s base
speed is denoted as:
X
Vf
Vb
(3)
Then the traction motor drive power rating required can be
rewritten in terms of this ratio as:
Pa
Mv 2§
1 ·
V f ¨1 2 ¸
2 ˜ ta ©
X ¹
(4)
The power rating must be increased to account for the
efficiencies of the traction motor drive ( K m ) and the transmission
(K t ):
Pa
Mv
1 ·
§
V f2 ¨1 2 ¸
2 ˜ ta ˜ K m ˜Kt ©
X ¹
(5)
Fig. 3 shows the evolution of the traction effort vs. speed profile
for different X ratios for a notional automotive design: a 1000
kg vehicle having a maximum cruising speed of 160 km/h and
having capability to accelerate from 0 to 100km/h in 10 seconds.
A large X ratio is highly desirable since the maximum
acceleration requirement can be met with a minimum traction
motor drive power rating. Indeed, the required power rating with a
X ratio of 1.5 is 82.5kW while with a X ratio of 5 decreases
the power rating to 42.5kW. However, there is a diminishing
return for an increasing X ratio beyond X 5 . For a X
ratio tending towards infinity, the power rating tends towards
38.6kW for the vehicle considered. It is therefore preferable to
seek a traction motor drive with a X ratio of no more than 5. If
the aerodynamic drag losses are entirely supplied for by the
primary power source, the only energy losses affecting the
acceleration performance result from the efficiencies of the
transmission ( K m ) and traction motors (K t ). The energy
absorbed by the maximum acceleration test is given by:
Wa
2
1 M v ˜Vf
2 K m ˜Kt
(6)
Standard drive cycles provide additional information on the
acceleration power and kinetic energy requirements. The power
supplied by the energy storage device is equal to the dynamic
power augmented by the traction motor and transmission losses:
Pa (t )
M v ˜V ˜
dV
dt
K m ˜Kt
(7)
The kinetic energy requirements are derived by integration of the
power supplied by the energy storage device:
Wa (t )
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
t
³ P (t ) ˜ dt
0
a
(8)
IJEET
- 43 -
Fig. 4: Dynamic power and kinetic energy variations along
standard drive cycles
Fig. 4 shows the evolution of the dynamic power and kinetic
energy supplied by the energy storage in a 1000kg hybrid vehicle
driven along the FTP 75 Urban, FTP 75 Highway, US06, and
ECE-1 standard drive cycles. Important information is derived
from these graphs. The dynamic power demand for each cycle is
much smaller than that required by the maximum acceleration
test. Furthermore, the kinetic energy consumed in any 10-second
interval is much less than the kinetic energy consumed during the
maximum acceleration test. Consequently, if the FESS is
designed to meet the maximum acceleration test, it will be more
than capable of fulfilling the requirements of the standard drive
cycles.
III.
FESS POWER & ENERGY PARAMETERS
A FESS stores kinetic energy in a rotor spinning at high angular
velocities and uses an electric motor drive to interface with the
vehicle’’s DC bus to constitute a flywheel or mechanical battery
[14-16]. Fig. 5 shows the basic layout of a flywheel energy
storage system. The design parameters of a flywheel system are
its energy storage capacity and its power capability, which are
independently designed and optimized for the application Once
an energy storage capacity and power cycle capability have been
identified, the mass and volume of the system should be
minimized for optimal system density.
The amount of kinetic energy stored in a rotating flywheel is a
direct function of the moment of inertia of the rotor ( J f ) and the
square of its angular velocity ( Z f ), as per equation (1).
Traditionally, the strength of materials has limited the options for
increasing stored energy. The development of advanced
composites and magnetic bearings however, has made possible
to increase rotational velocity, thereby yielding compact energy
storage. The specific moment of inertia relates the flywheel’’s
moment of inertia to its mass and is proportional to the sum of the
inner and outer radii of the rotor:
2
2
J fs v Router
Rinner
Fig. 5: Flywheel Energy Storage System schematic
(9)
While the energy stored increases significantly with the outer
radius, the inner radius contributes equally to the energy stored
per unit mass. There is, however, a practical limit on the
combination of radius and rotational speed due to the mechanical
strength limitations of the rotor construction and materials. The
tangential speed of the outer rim of the rotor, called the tip speed,
Vt , is given by:
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
Vt
- 44 -
Router ˜ Z max
(10)
The mechanical stress on the rotor material is a direct function of
the tip speed, or inversely, the maximum working tip speed can
be expressed as:
Vtmx
k solid ˜ k hollow ˜
In (11),
Vw
Vw
U
(11)
is the working tensile strength limit, selected as a
portion of ultimate tensile strength, and
U is the mass density.
k solid and k hollow are constants that include the Poisson ratio
effect, Q , and are defined as follows [9]:
k solid
k hollow
8
3 X
1 , when Rinner
ª
Rinner
«2 Router
¬
0 Rinner 1
k hollow
(12)
0
(13)
§ 1 3X ·º
˜ ¨1 ¸»
3 X ¹¼
©
1 / 2
, when
(14)
In addition to the outer radius and speed, the rotor inertia, and
therefore the flywheel capacity, is determined by rotor length.
Rotor dynamics limit the maximum outer rotor length achievable.
The rotor has several critical (i.e. resonant) speeds and the
flywheel necessarily has a limited ability to avoid them or go
through them without failure. The stiffness of the rotor and shaft
designs, and the distance between bearing locations become
determining factors on practical rotor length. These geometric
constraints, along with the rotor material selection, allow a
relation to be drawn between physical size and the energy
stored. A simple test for flywheel size is to assume a rotor lengthto-diameter ratio of 1. Then, if a tip speed is selected according to
the construction desired (steel, composites, or other), a table of
radii yields volume, mass, and maximum energy storage
according to equation (15). Here, the material density ( U ) and
maximum tip speed, Vt max , are given to yield the energy storage
limit, Wlim , assuming a simple cylinder.
1
3
U ˜ S ˜ Router
˜ Vt 2max
2
(15)
Removing mass from the inner diameter or radius lightens the
system mass with only small reduction in energy stored, as
implied by equation (9), but the outer radius limitation on total
energy storage still holds.
The effective, or deliverable, flywheel energy is typically much
less than the stored energy because it is not desirable to drive
the flywheel all the way to zero speed. Consequently, the
effective flywheel energy is given by the difference between the
stored energy at maximum speed and the stored energy at
minimum practical speed:
1 ·
§
Wmax ¨1 2 ¸, n
© n ¹
2
Z max
2
Z min
(16)
Thus, the effective energy delivered by a flywheel is 75% for n=2,
89% for n=3, and so forth, in a pattern of diminishing return.
Consequently, the minimum flywheel operating speed will
generally be selected between 33% and 50% of its maximum
speed.
The flywheel itself does not generally impose a limit on the
power, or rate of energy exchange, to or from the FESS. The
power rating of the FESS is limited mechanically by the strength
of the shaft and coupling. The flywheel electric motor drive may
be used beyond its continuous operation rating. In this case, the
power rating is determined by the losses of the electric motor
drive and its cooling system. The short-term power rating may be
derived from the average power flow through the electric motor
drive over a sliding window in the standard drive cycles. The
power flow is thus expressed as:
1
t W w
d
W w ³t
Where W w is the duration of the window. The continuous power
Psd
Consequently, steel rotors have a practical operating limit in the
range 220-240 m/s, while the most advanced composite
structures can be operated at speeds ranging from 400 m/s to
1000 m/s.
Wlim
Weff
P ˜ dt
(17)
rating is determined solely by the torque-speed profile of the
electric motor drive. This profile is the same as that of the traction
motor drive (Fig. 2). In an effective FESS design, the rated power
should be available regardless of the state of charge of the
flywheel battery, i.e. the rotor speed. This yields optimal
acceleration and regenerative braking performance and matches
the traction power and energy requirements of the vehicle. The
electric motor drive constant power region range (base and
maximum speed) must be matched with the flywheel minimum
and maximum speed. The flywheel energy is thus available at
rated power at any speed beyond base speed. The requirement
for constant rated power availability from minimum-to-maximum
state of charge has an additional advantage. Constant power is
the best torque-speed profile to accelerate and decelerate a
mass with a minimum electric motor drive power rating.
It follows that if the flywheel’’s minimum speed is a third of its
maximum speed, then the electric motor drive must have an X
ratio of 3. The practical speed range of the flywheel and the
requirement for rated power therefore strongly influence the
technology and design of the electric motor drive. Fig. 7 shows
torque-speed profiles for a 45kW, 40,000rpm FESS electric motor
drive for different values of the X ratio. There appears to be a
drawback in having a large X ratio in that the low-speed torque
is proportionally much larger than for a low X ratio. A higher
torque requires a larger motor and a sturdier flywheel shaft.
However, it is necessary to consider that a low X ratio will only
yield rated power for a very limited range of speeds, which
dramatically reduces the performance of the FESS and its ability
to match the requirements of the hybrid drive train. This would
force the designer to oversize the electric motor thereby
increasing the weight of the FESS. Furthermore, the electric
motor and its drive would be grossly underutilized. It is therefore
much more beneficial to use a large X ratio.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 45 -
V.
Fig. 7: Variation of motor drive torque-speed profiles with the X
ratio
IV.
HøGH END MOTOR SPEED CHARACTERøSTøCS
FESS applications place a significant emphasis on the constant
power (or field-weakening) operation of the electric motor drives.
Hence, these motors are thus radically different from electric
motors for conventional applications and are very similar to
traction motors. Each technology is differently suited for the
application to FESS.
Permanent magnet brushless DC motors (BLDC) have an
inherently short constant power range, theoretically up to
X 2 but practically closer to X 1.5 . Furthermore, fieldweakening is achieved by using stator currents to counteract the
field of the permanent magnets on the rotor. The results is that
the power factor and the efficiency of the BLDC motor drive is
significantly lower in the constant power region than in the
constant torque region. The BLDC motor is therefore a bad
candidate for flywheel applications because it would yield a
limited practical energy range and a low efficiency over that
range.
The induction motor is the most mature and one of the most
rugged electric motor technology. While the breakdown torque
limits the constant power range, it is nevertheless possible to
achieve a X ratio up to 5 by using vector control [17]. However,
the large X ratio requires a more complicated machine and is
obtained at the expense of power factor and efficiency.
Switched Reluctance Motor (SRM) drive research at Texas A&M
University has shown that X ratios up to 8 are possible [18].
Furthermore, the SRM drive has a better efficiency in the
constant power region than BLDC or induction motor drives.
Finally, the large inertia of the flywheel rotor damps the torque
pulsation generated by the SRM operation thus resulting in a
quiet electric motor drive.
Homopolar synchronous induction motors offer a further
consideration for high-speed flywheel applications. This novel
motor design allows a solid rotor for high-speed and a
controllable field induced from a separate but stationary winding.
From the point of view of the terminals, the motor behaves as a
three-phase synchronous machine. A 30kW prototype has been
demonstrated with an efficiency of about 85% while operating at
constant power from 30,000 to 60,000 rpm [19].
CONCLUSøON
Matching the constant power range of the electric motor drive
with the practical energy range of the flywheel is of prime
importance in order to achieve an FESS with desired energy
storage capacity deliverable at rated power. An FESS designed
with a proper motor drive constant power range can be matched
optimally with the driving requirements of a hybrid drive train,
thereby resulting in minimized fuel consumption and optimized
vehicle system operation. Due to practical consideration on the
flywheel design, it is not desirable to seek a X ratio of more
than 4, which corresponds to an effective energy content of
approximately 94%. Switched reluctance motor drives appear to
have the best inherent qualities for FESS applications, although
additional simulations and experiments are required to validate
this choice.
VI. REFERENCES
1.
K. R. Davey, R. E. Hebner, ““A Fundamental Look at Energy
Storage Focusing Primarily on Flywheels and Superconducting
Energy Storage,”” Electric Energy Storage Applications and
Technologies (EESAT 2003), San Francisco, CA, Oct. 27-29,
2003.
2. M. M. Flynn, J. J. Zierer, R. C. Thompson, ““Performance Testing of
a Vehicular Flywheel Energy System,”” SAE 2005 World Congress,
April 11-14, 2005, in review.
3. R. E. Hebner, T. A. Aanstoos, ““Energy Storage for Sustainable
Systems; a White Paper on the Benefits and Challenges of Kinetic
Energy Storage,”” National Science Foundation Workshop on
Sustainable Energy, Georgia Tech, Atlanta, GA, Nov. 28-Dec. 1,
2000.
4. R. Hebner, J. Beno, A. Walls, "Flywheel Batteries Come Around
Again," IEEE Spectrum, April 2002, pp.46-51.
5. R.J. Hayes, J.P.Kajs, R.C.Thompson, and J.H.Beno, ““Design and
Testing of a Flywheel Battery for a Transit Bus,”” SAE Publication
No. 1999-01-1159, Society of Automotive Engineers, February
1999.
6. R.J. Hayes, D.A. Weeks, M.M. Flynn, J.H. Beno, A.M. Guenin, J.J.
Zierer, ““Design and Performance Testing of an Integrated Power
System with Flywheel Energy Storage,”” presented at SAE Future
Transportation Technology Conference, June 23-25, 2003, Hilton,
Costa Mesa, California and published in SAE Publication SP-1789.
7. L. Hawkins, B.T. Murphy, J.J. Zierer, R.J. Hayes, ““Shock and
Vibration Testing of an AMB Supported Energy Storage Flywheel,””
Presented at Eighth International Symposium on Magnetic
Bearings (ISMB-8), Mito, Japan on August 26-28, 2002 and
published in JSME’’s International Journal Series C, Vol 46, No 2
(June 2003).
8. J.D.Herbst, M.T.Caprio, and R.F. Thelen, ““Advanced Locomotive
Propulsion System (ALPS) Project Status 2003””, 2003 ASME
International Mechanical Engineering Congress and Exposition,
November 15-21, 2003, Washington, DC.
9. R.F. Thelen, J.D. Herbst, and M.T. Caprio, ““A 2 MW Flywheel for
Hybrid Locomotive Power””, IEEE Semiannual Vehicular
Technology Conference VTC2003-Fall, October 6-9, 2003,
Orlando, FL.
10. M.T. Caprio, R.F. Thelen, J.D. Herbst, 2 MW 130 kWh Flywheel
energy storage system, Electrical Energy Storage - Applications
and Technology (EESAT2003), October 27-29, 2003, San
Francisco, CA.
11. J. Herbst, M. Caprio, R. Thelen, ““Critical Design Factors in the
Devleopment of a Hybrid Electric Locomotive Propulsion System””
IEEE Vehicle Power and Propulsion Conference, October 6-8,
2004 Paris, France.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
12. M. Ehsani, Yimin Gao and K. Butler, "Application Of Electric
Peaking Hybrid (ELPH) Propulsion System To A Full Size
Passenger Car With Simulation Design Verification," IEEE
Transaction on vehicular Technology Vol.48, No.6, Nov. 1999.
13. Yimin Gao, H. Moghbelli, and M. Ehsani, etc. "Investigation of
High-Energy and High-Power Hybrid Energy Storage Systems for
Military Vehicle Application," SAE Future Transportation
Technology Conference, June 23-25, Costa Mesa, CA, Paper No.
2003-01-2287.
14. R.J. Hayes, D.A. Weeks, M.M. Flynn, J.J. Beno, A.M. Guenin, J.J.
Zierer, & T. Stifflemire, "Design and Performance Testing of an
Advanced Integrated Power System with Flywheel Energy
Storage," International Future Transportation Technology
Conference, June 23-25, 2003 Costa Mesa, CA, SAE paper
#03FTT-79.
15. R.J. Hayes, J.P. Kajs, R. C. Thompson, and J.H. Beno, "Design
and Testing of a Flywheel Battery for a Transit Bus," Society of
Automotive Engineers, February 1999, SAE publication #1999-011159.
16. Yimin Gao, and M. Ehsani, "A Mild Hybrid Drive Train for 42 V
Automotive Power System--Design, Control and Simulation," SAE
2002 World Congress, Detroit, MI., Paper No. 2002-02-1082.
17. Mehrdad Ehsani, Khwaja M. Rahman, Hamid A. Toliyat,
"Propulsion System Design of Electric and Hybrid Vehicles," IEEE
Transactions on Industrial Electronics, vol. 44, no. 1.
18. T. Kume, T. Iwakane, T. Yoshida, and I. Nagai, "A Wide Constant
Power Range Vector-Controlled Ac Motor Drive Using Winding
Changeover Technique," IEEE Trans. On Industry Applications,
Vol. 27, No. 5, pp. 934-939, Sept/Oct, 1991.
19. Perry Tsao; Senesky, M.; Sanders, S.R.; ““An integrated flywheel
energy storage system with homopolar inductor motor/generator
and high-frequency drive,”” Industry Applications, IEEE
Transactions on, Volume: 39 , Issue: 6 , Nov.-Dec. 2003 pp.1710––
1725.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
- 46 -
IJEET
- 47 -
Mobile System for Testing and Calibrating Vehicle Speed Sensors
D. Moga, M. Munteanu, T. Marita, C. Cret
Technical University of Cluj-Napoca, Romania
I. INTRODUCTøON
Usually vehicle speed is measured based on the number of
revolutions of the traction wheels in a given time unit. While this
method is popular and some robust transducers were developed,
its precision is affected by the variation of the tyre dimensions
due to pressure changing or weariness.
Optical or electromagnetic transducers are used to sense the
rotation of the wheel itself or the rotation of a mechanical part
rotating synchronously with the wheel. Electromagnetic
transducers can use Hall effect or proximity effects to sense the
movement. They are generally more robust than the others due
to the large operating temperature range and insensitivity to
automotive fluids or dust deposition. The inductive rotationalspeed sensors are benefiting from their non-contacting
(proximity) measurement principle, and thus they are wear-free.
The sensible element of the inductive transducers basically
consists in a core surrounded by a winding. In order to measure
the rotational speed of an axle, a rotating toothed pulse ring is
mounted on this I such a way that it is located directly opposite
and a narrow air gap is separating it from the transducer. The
transitions between the tooth space and tooth (leading tooth
edge) and at the transitions between tooth and tooth space
(trailing tooth edge) are responsible for modifying the magnetic
flux through the core.
One common setup for the proximity transducer is the one
sketched in figure 1:
The demodulator converts the change in the amplitude to a DC
signal. This DC signal is fed to a trigger stage. The operation of
the oscillator at frequencies of hundreds of kHz is convenient for
the usage of the transducer in the usually highly
electromagnetically polluted automotive environment because the
perturbation signals are usually within a much lower frequency
range, so they can be removed before demodulation.
Fig. 2: Typical processing stages of a proximity rotational speed
sensor
II. THE HARDWARE ARCHøTECTURE OF THE PROPOSED
SYSTEM
The system we proposed is a workbench for testing and
calibrating vehicle speed sensors. From a hardware point of view,
it is based on three basic blocks: an analog front-end, and USB
module and a mobile computer (laptop) as depicted in figure 3.
The analog front-end is implementing the processing stages
needed for obtaining a rectangular signal having a speed
proportional frequency: demodulation, amplification and
triggering. It is worth to note that all the processing is done in
such a manner that the in vehicle speed measurement system is
not affected at all (see figure 4).
The USB module is built around the Cypress CY7C68013
microcontroller. Its purpose is to real-time measure the period of
the rectangular signal fed to one of its port pin, to apply a filtering
algorithm and to transmit series of samples to the PC.
CY7C68013 is a low power microcontroller that contains a
standard 8051 core and a Smart USB Engine in the same chip.
Its internal architecture is described in the block diagram of figure
5.
Fig. 1: Proximity transducer facing toothed ring
For a large category of automotive transducers, the winding is fed
with a sinusoidal signal, so the altering of the inductance caused
by this varying flux is seen as an amplitude modulation of the
voltage across the transducer’’s coil. The switch operation is
achieved through a series of processing operation applied to the
voltage signal present across the inductor placed in the probe, as
presented in figure 2.
It can be observed that there is an internal RAM area that can be
accesses by both the 8051 core and the USB Engine. The core
has the ability to directly edit the data contents of the internal 8kbyte RAM and of the internal 512-byte scratch pad RAM via a
vendor-specific command. That means that the application
running on the PC has the ability to read or write these areas
through calls of the Windows Cypress USB driver with specific
control codes. On the other side, the 8051 core sees these areas
as belonging to the external data memory.
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
IJEET
- 48 -
Fig. 3: Wiring setup for extracting and processing desired signal from vehicle’’s speed measurement system
TP2
+8V
+8V
VCC_AN
2
C3
47u
-
R3
2K2
+
R6
4K7
5 +
6 -
U1A
TLC272
TP3
+5V
R8
1K
7
U1B
TLC272
8
5
+5V
P2
3K POT_mic
2 +
3 -
U4A
74HCT74
4
1
6
R9
2K2
U2
LM311
7
P1
10K POT_mic
4
3
2
1
+5V
2 +
2
3
U3
LM311
7
TTL Output
J2
OUTPUT
1
2
3
4
3 -
U7
TLP620
VIN
VOUT
+8V
VCC_AN
3
C10
100n
r5mm
+
+8V
VCC_AN
C11
1000u
cpol
C12
100n
r5mm
1
U6
LM7805C/TO220
IN
GND
U5
LM7808/TO220
2
J4
C9
JUMPER 100n
1
2 r5mm
1
GND
+12V CAR SUPPLY L1
1.2 mH
TP1
+12V
1
2
3
2
J3
ALIM
L2
1.2 mH
Fig. 4.Schematic diagram of the analog front-end
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
3
2
4
1
6
P3
3K POT_mic
1
TP5
TTL Out
3sip100
1
8
5
R10
2K2
JP1
Jumper Selector
14dip300
S
5
C1
6
1D
R
1
R1
56K
C5
1u
1
4
C1
47n
+
4
3
C8
1000u
R7
47K
OUT
+5V
3
C13
100n
r5mm
+
C14
1000u
cpol
TP4
Analog Out
1
D1
1N4148
R5
4M7
+
C7
1000u
4
Input
C6
47n
+
8
J1
INPUT
1
2
3
C4
4n7
R4
100R
8
R2
2K2
C2
2u2
Analog
IJEET
- 49 -
Fig. 5: Internal architecture of CY7C68013
III. THE SOFTWARE ARCHøTECTURE
The hardware organization of the CY7C68013 suggests the
following setup for a system that interfaces sensors to the PC
using this microcontroller:
A. Sensors with digital output can be connected to the I/O
ports of the chip, directly or optionally through
optocouplers
B. Sensors with analog output can be interfaced to the
I/O ports of the chip through an analog to digital
converter with parallel or serial output, or with I2C
compatible A/D converters
C. The application running on the 8051 core has to
program the chip in the adequate way for selecting the
appropriate alternate function for each of the pins used
D. Three specific areas can be defined in the internal 512byte scratch pad RAM:
1. Buffer1, in which the 8051 application saves
the results of the algorithms applied to the
sensors data
2. Buffer2, in which the PC application saves
the control setting for the algorithms applied
to the sensors data
3. Buffer3, in which the 8051 application saves
the status flags of the algorithms applied to
the sensors data
P
P
system a supplementary degree of flexibility. That is the ability to
transfer the content of the 8051 core code memory via the USB
engine. That makes the USB module fully programmable, turning
it into a valuable platform for experimenting real-time processing
algorithms.
An application running on the PC is controlling and
configuring the USB module via the USB Windows driver.
IV. EXPERøMENTS AND RESULTS
The whole system is managed through the graphical user
interface of the PC application. It offers the following facilities:
- Downloading of the processing code into the 8051
core;
- Starting and stopping of the 8051 core program;
- Real-time displaying of the speed values;
- Storing of data sequences and them simulating a trip
based on these values;
- Applying different filtering algorithms to the received
speed samples.
The PC application will acquire the measurement data asking
the driver to operate a transfer from the Buffer1 to the PC, will
program the acquisition parameters like sampling frequency,
samples averaging or active analogue channels by writing
control codes in the Buffer2, and will check the operational status
of the sensors and possible fault conditions by reading Buffer3.
While the above suggested approach is suitable for a simple yet
efficient measurement data transfer over the USB link, there is
another feature o this microcontroller that gives to the whole
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005
Fig. 6: Graphical User Interface
IJEET
- 50 -
Using this application, data sets containing the values of the
instantaneous frequency (or period) of the signal outputted by the
sensor and their associated correct values of the vehicle speed
can easily be constructed. These can serve as main data for
calibrating the vehicle speedometer. In figure 7 there is an
example of a sensor transfer curve constructed based on spline
fitting of 9 measurement points. As it can be seen, a linear fitting
can be an appropriate way for linearizing the sensor
characteristics.
Fig. 7: Transfer curve of a speed sensor
V. CONCLUSøONS AND FUTURE WORK
The system presented in this paper can be used as a flexible
workbench for studying and experimenting with speed sensors
appropriate for vehicle speed measurement. It offers tools for
supporting the calibration of these sensors while offering a
flexible platform for implementing and testing the embedded
software for electronic speedometers.
The abundance of serial busses offered by CY7C68013 (I2C, 2
UART’’s) offers further possibilities for interfacing the system with
other devices (optical barriers, radio modems, timers etc.) for
devising a complete calibration and telemetry system.
P
P
VI. BøBLøOGRAPHY
1.
HTU
Siemens Semiconductor Group, IC for Inductive Proximity
Switches
with
Short-Circuit
Protection,
http://www.infineon.com/cmc_upload/0/000/012/273/tca505.pdf
Philips Semiconductors, ““Semiconductor Sensors Data Handbook””,
Philips Electronics N.V. 1994, Netherlands
2.
UTH
International Journal of Electrical Engineering in Transportation, vol. 1, n°1, 2005