BASE PAPER

for the project of

“Embedded Based Emission Free
Vehicle with Solar Charging”

HARI PRASATH G – 730218106301

MOHAMED YUSUF K – 730218106008

SUGAIL MUSHRAF S – 730218106016

MOHAMED APRITH ZIHAN S – 730218106006

Guided by

Mr.Z. SHAHEEL HAMEED, M.E.,

HOD/ECE
REFERENCE
PAPER
for the project of
“Embedded Based Emission Free
Vehicle with Solar Charging”

HARI PRASATH G – 730218106301

MOHAMED YUSUF K – 730218106008

SUGAIL MUSHRAF S – 730218106016

MOHAMED APRITH ZIHAN S – 730218106006

Guided by

Mr.Z. SHAHEEL HAMEED, M.E.,

HOD/ECE
 IEEE International Conference on Power, Control, Signals and Instrumentation Engineering (ICPCSI-2017)

Next Generation Electric Bike

E-BIKE

 N. PAVAN KUMAR REDDY(Author) K.V.S.S VISHNU PRASANTH(Co-Author)

ELECTRICAL AND ELECTRONICS AUTOMOBILE ENGINEERING
HINDUSTAN UNIVERSITY HINDUSTAN UNIVERSITY
CHENNAI, INDIA CHENNAI, INDIA
[email protected] [email protected]

ABSTRACT High strength

The primary objective of the project is to design a feasible yet Low weight
highly adaptable E-bike. As the number of motor vehicles on Ease of manufacturing
the roads throughout the world increases at staggering rate Safety
each year, the dependence on oil-based fuel grows almost Low cost
unchecked. The increased use of non-renewable fossil fuels All the topics presented in this report has been supported well
brings with it environmental problems such as: the "greenhouse by Calculations, 3D CAD models and FEA data and charts.
effect", health problems for city delivers and concern over the Individual components have been separately discussed under
stability of fuel supply. To move away from this dependence on subtopics that require them.
oil, a vast amount of money is being spent on the development A. Materials used:
of electrical vehicles (EVs) that may be produced. This paper AISI Mild Steel 1020 is used as the material for the frame.
presents a study of electrical motorcycle design. The aim of this Thus, facilitates ease of manufacturing and procurement of the
study is to investigate how to design a simple, cost effective mater
model of electrical motorcycle with intelligent control system. B. Tube dimensions:
This can be implemented by removing the internal combustion Deciding the dimensions of the tubular frame to be used
engine, the exhaust system and other unnecessary components was crucial to the design of the frame as it would be affecting
from the motorcycle and replaced by an electrical motor, an the structural integrity and the weight.
intelligent controller, and a battery pack, cabling system and Mild steel tubes of 1inch O.D and 0.083-inch thickness used
monitoring instruments. for building the vehicle frame
Keywords--brushless motor, suspension, frame, lipeo4 C. Frame:
battery, intelligent controller It is the skeleton of the vehicle which acts as the back
bone of the vehicle and supports and unitizes the body parts of
I.INTRODUCTION the vehicle. It also provides good handling and safety. A vehicle
A single mild steel bar is bended into various forms such as an electric vehicle draws an advantage if it is light in
to give it a proper 2-wheeler frame, with minimal dimensions weight. The Frame of the vehicle must be designed such that it
used in motorsport and various other applications. must be able to accommodate the battery pack. The vehicle
The design tool used in this process is SOLIDWORKS 2015 draws its main structural rigidity from the frame and it’s
which consists of inbuilt tools to model the components to supporting members on which all components are mounted.
conduct extensive FEA on them. The help of ANSYS 16.0 was Thus, ensuring the optimal material is chosen for the frame.
also taken in the case of thermal analysis. S.No Specifications Dimensions(mm)
The different departments that were dealt with have been listed 1 Overall length 1500
below. 2 Overall height from ground 960
Chassis
Steering
Suspension 3 Wheel base 1219
Electrical powertrain 4 Seat width 270
II.DESIGN OVERVIEW 5 Seat height from ground 530
Table.1 specification of the frame
To design a vehicle according to the norms of the
competition the following objectives were taken into D. TORSIONAL ANALYSIS:
consideration in order to satisfy the competition criteria’s as The below figure shows
well as for the better performance of the vehicle on the track. the deformation of the vehicle frame when 3000N of
static load was added on the bottom when the top
Such ideologies are member is fixed.

978-1-5386-0814-2/17/$31.00 ©2017 IEEE

2280
 IEEE International Conference on Power, Control, Signals and Instrumentation Engineering (ICPCSI-2017)

Stainless
Mild
Material steel Cast Iron
Steel
SS304
Weight
Density 7840 8000 7200
(Kg/mm3)
Fig.1 side impact analysis of frame
Tensile
Displacement 161.8mm 440 490 430
Strength
Force applied 3000N (MPa)

F.O.S 15.8 Compressive

Strength 310 415 570
Table 2. side impact of the frame (MPa)
Poisson's
0.3 0.3 0.27
Ratio
Table 3. material composition

G. ASSEMBLY OF BIKE:
Vehicle chassis assembly was done using SolidWorks.
The following are some pictures showing different views
Fig.2 isometric analysis of frame showing electric motor and battery mounted .

E. REAR IMPACT ANALYSIS:

The below picture shows the deformation of
the vehicle frame when 3000N of static load was added
on rear end when front rake is fixed.

Fig.5 Isometric view.

III.STEERING
Fig.3 Rear impact analysis of frame
A. Trail(b):
F. Material Comparison: =(RW*Sin(AR)-OF)/Cos(AR)
Taking into consideration the options available mild steel = (266.7*Sin (21)-25.4)/Cos (21) =75.16mm
which is the blend of carbon, nickel, chromium and iron is where
AR = Rake angle
OF = Fork Offset
B. Mechanical Trail or True Trail(T):
=RW*Sin(AR)-OF
=266.7*Sin (21)-25.4
=70.17mm
Turning Radius(r)
Fig 4. Side view =B/μ*Cos(AR)
=1219/0.5*Cos (21)
chosen to build the frame of the electric motor-bike because of =2615mm
its higher tensile strength and compressive strengths when Where
compared with stainless steel and cast iron. B = Wheel base
μ = Steer angle
Wheel Flop Factor(f)
=b* Sin (ʋ) * cos (ʋ)
=75.16 * sin (72) * cos (72)

2281
 IEEE International Conference on Power, Control, Signals and Instrumentation Engineering (ICPCSI-2017)

=22.08 Motion Ratio = Wheel Displacement/Spring

Where Displacement
b = Trail = 0.31
ʋ = head angle Maximum ride height = 4.8 inches
Leaning Angle(α), Sprung Mass = 75 kg
=tan-1(v2/gr) Un sprung Mass = 15 kg
=tan-1(9.722/9.81*2.615)
Therefore,
=74.80
Preload Compression = Sprung mass /
Specification Value
Length of The Fork 27 Inches Total mass
Rake Angle 210 = 75/90
= 0.84
Sprung Weight (75+55) Kg
After Driver load (post load compression) = Weight of the
Un Sprung Weight 15kg driver/ Total weight+ Preload compression
Weight of Rider 55kg = 55 /90+0.84
Front Spring Rate 20 Kg/Inch = 0.61+0.84
=1.45
Rear Spring Rate 60kg/Inch
Table 4. specifications of steering
C. Cad Model of a Swing Arm in The Rear Suspension of a
IV.SUSPENSION Bike:

A. At the front
Mass transfer at front = 79.7kg (weight
transfer is 60:40)
C.G = 411m
Wheelbase =1219 mm Fig.7 swing arm

79.7*411/ 1219
V.BRAKES
Total load = Load transfer per mass + (Weight of the A. The brake pedal:
vehicle)/2 The brake pedal exists to multiply the force exerted by the
=26.87+45 driver’s foot. From elementary statics, the force increase will be
= 705.04 N equal to the driver’s applied force multiplied by the lever ratio
Percentage load = (71.87 /79.7) * 100
of the brake pedal assembly:
90.1 % Fbp= Fd× {L2÷ L1}
Load on damper = Total load * cos (90 – Where,
rake angle) Fbp = the force output of the brake pedal assembly
Fd = the force applied to the pedal pad by the driver
=71.87 * cos (90 – 21) L1 = the distance from the brake pedal arm pivot to the
=25.755 kg output rod clevis attachment
Therefore, force on front = 252.66 N L2 = the distance from the brake pedal arm pivot to the
Compression = Force on front in kg/ Front spring rate brake pedal pad
= 25.755 / 20 REAR,
= 1.28 inch Fd = 147.15 N
Total compression = twice the Compression L1 = 1
= 2 + 1.28 Fbp= 588.6 N
= 3.28 inch. FRONT,
Fd = 98.1
B. At the rear: L1 = 1
L2 = 6
Fbp = 588.6 N
Total Brake Pedal Force:
Fbp = 1176 N.
B. THE MASTER CYLINDER:
Assuming incompressible liquids and infinitely rigid
hydraulic vessels, the pressure generated by the master cylinder
will be equal to:
Fig.6 Damper Pmc= Fbp/Amc
Where,
Rear spring rate = 60 kg / inch Pmc = the hydraulic pressure generated by the

2282
 IEEE International Conference on Power, Control, Signals and Instrumentation Engineering (ICPCSI-2017)

master cylinder Rear,

Amc = the effective area of the master cylinder Ftire = 57.5926 ÷ 0.2159
hydraulic piston = 267.4965
Pmc = Fbp/Amc F = Ffront + Frear
=1176/506.45 = 1098.251 N
=2.3208 N/mm
F. Deacceleration developed:
Convention dictates that accelerations which oppose
C. Caliper: the direction of travel are called decelerations. In the case of a
It is the first functional responsibility of the caliper to vehicle experiencing a braking force, the deceleration of the
translate the hydraulic pressure fluid from the pipes and hoses vehicle will be equal to:
into a linear mechanical force. Once again assuming D = Ftotal ÷ mv
incompressible liquids and infinitely rigid hydraulic vessels, the Where,
one-sided linear mechanical force generated by the caliper will D= the deceleration of the vehicle.
be equal to: D = Total force / Mass of vehicle
Fcal= Pcal× Acal = 1098.215 / 150
Where, D = 8.45 m/s2 = 0.845g
Fcal= the one-sided linear mechanical force Brake efficiency,
generated by the caliper. = D /1 * 100
Acal= the effective area of the caliper hydraulic = 0.845 * 100
pistons found on one half of the caliper body. = 84 %
Clamping Force (CF) = Pmc x Area of caliper piston Stopping distance,
(Acp) x no. of pistons S = ut – ½ at2
Acp =506.45mm2 Where,
Front caliper, S = stopping distance, m
No. of pistons = 2 U = initial velocity, m/s2
CF = 2.3208*506.45*2 V = final velocity, m/s2
=2350.73 N T = distance of application of brake
Rear caliper, D = deceleration of vehicle
No. of pistons = 1 From the speed of 40 km/h U = 40 km/h
CF = 2.3208*506.45*1 = 40 / 3.6 m/s
=1175.36N S = ((40 / 3.6)2 ½) * 8.45 = 3600 / 219.02
S = 7m
D. Rotor:
While the rotor serves as the primary heat sink in the G. Force acting on the vehicle:
braking system, it is the functional responsibility of the rotor to Mass of the vehicle = 155kg
generate a retarding torque as a function of the Load distribution on the axle Front:
Tr= CF * Ffri * Reff Rear
Reff = D+d / 4 Static load distribution on axles determined via
Where, modelling
Tr= the torque generated by the rotor. = 60:40%
Reff= the effective radius of the rotor from the rotor. = 93:62
FRONT ROTOR Static Load on a tire Front = 186/2 = 93kg Rear = 124/2 = 62kg
Rotor diameter = 270 mm Load Transfer = (deceleration x height of c.g. x
Reff = 0.109 m mass of vehicle)/wheelbase
Tr = 179.36 Nm Deceleration= 0.845g
Rear rotor, Height of C. G= 0.52m
Rotor diameter = 200 mm Mass of vehicle=155kg
Reff= 0.070 mm Wheel base = 1.25m
Tr = 57.5926 Nm Load transfer = (0.845*0.52*150)/1.25
=52kg
E. Friction between tire and road: =30%( approx.).
The amount of slip generated will be a function
of the tire’s output characteristics, but the force reacted at the VI.ELECTRICAL POWERTRAIN
ground will be equal to:
Ftire = Tr÷ Rt Performance and driver comfort being the major objectives, has
Tire diameter = 21 inch focused on the very simple and lightweight powertrain design
Front, so as to reduce the losses encountered during power
Ftire= 179.36 ÷ 0.2159 transmission.
= 830.755 N.

2283
 IEEE International Conference on Power, Control, Signals and Instrumentation Engineering (ICPCSI-2017)

D. Voltage Regulator:
The twisty tracks of motocross demand driver focus on the An e-bike system has three power levels. The power
track. This brings in usage of properly selected type of power levels are oriented from a 48-V battery. 5 V specifies the power
layout. supply of microcontroller and other devices.
48 V can drive the power MOSFET directly
A. Lithium ion battery: 15 V specifies the power supply of MOSFET driver ICs in
Is arguably better for the environment, from a damage power bridge
easily by over charge E. Calculations:
Highest energy density to weight ratio The weight of the vehicle without the driver is 95kg,
Eliminates need for periodic care for a long life extra members and parts were casualties. By reducing the
Has no memory effect weight, we also reduce some unwanted
Achieve better cost-performance ratio for technologies Consider,
require a protection battery packs in series than for single cell. Weight of the vehicle = 155 kg (assumed with
driver)
B. Motor specifications: =155*9.81
=1520.5N
TRE=TM*n%*Gratio
TM=Motor torque
Tre=Required torque
n %= Efficiency
Gratio=Sprocket ratio
TRE=Wnormal *μ +CR*w
CR= coefficient of rolling resistance
Wnormal=weight of vehicle including
Driver
μ = coefficient of friction

Fig.8 BLDC motor

Name plate details:

Power - 1 KW
Speed - 3000rpm
Torque - 9Nm
Motor Type - BLDC
Brushless Direct Current (BLDC) motor is a type of
synchronous motor, where magnetic fields generated by both
stator and rotate have the same frequency. The BLDC motor
has a longer life because no brushes are needed. Apart from
that, it has a high starting torque, high no-load speed and small Fig.9 wiring diagram of electrical system
energy losses. The BLDC motor can be configured in 1-phase, Consider the weight distribution 60% front& 40% back.
2-phase, and 3-phase. Three-phase motors are the most popular TRE=Wnormal *μ +CR*w
among all the configurations and are widely used in e-bikes. CR= coefficient of rolling resistance
The structure of a BLDC motor is divided into two parts: Wnormal=weight of vehicle including Driver
1.Moving part called the rotor, represented by permanent μ = coefficient of friction
magnet
=1520.5*0.4+0.01*1520.5 = 623.42N
2. Fixed part called the stator, represented by phase windings of
magnetic circuit. Consider the weight distribution 60% in front and 40% in back.
Weight acting on Rear wheel
C. Frame mounted: =623.42*40/100
motor with dedicated chain drive: =249.3102N
mounting of a motor to the bike frame near the of the rear wheel, TRE=F* Perpendicular distance of tire (Radius)
often on one side. The motor has its own small chain that drives ` =249.3102*0.2159
a planetary gear on the rear wheel. This specialized chain is for =53.83Nm
the primary bike chain that remains in use with the bike. Then,
Gearing remains specific to the manufacturer or aftermarket TRE=TM * n% * Gratio
motor conversion kit. 53.83Nm = 9Nm *0.84 *GRE
GRE=7.12
=7(approx.)
By transmission ratio, we need
Z1number of teeth on pinion sprocket = 8

2284
 IEEE International Conference on Power, Control, Signals and Instrumentation Engineering (ICPCSI-2017)

Z2number of teeth on wheel sprocket = 56.

Acknowledgment
F. Function of controller: Authors would like to acknowledge. Dr. A. K. Pravathy with
the power electronics and renewable energy at the automobile
The electrical system comprises of battery, motor, Laboratory for the coil finite element analyses and
wires, rubber insulation and controller. Controller is a
device used to govern the speed of a motor. Raghavendran and Getizal for their assistance in vehicle
Used for connecting many electrical components. systems and machines aspects of this program at electronics
Monitors the amount of voltage required by the motor. Transportation Laboratory
It governs the speed with the amount of throttle given to it. Reference
The supply to head light is given through it.
[1] Electric Bicycles: A Guide to Design and Use (Morchin and Henryoman,
G. General specifications of controller: 2005)
[2] www.InstaSPIN-BLDC.com
Three external interrupts are used for getting position [3] Wikipedia
information from [4] https://www.scribd.com/document/101553375/Electric- Bike-System.
Hall sensors while one is used for brake mechanism.

Fig.10 micro controller circuit

One timer is used for counting the Hall

interrupts in a fixed time to get speed information. Three ADC
channels are used for detecting battery voltage, handlebar
voltage, and system feedback current. PWM output directly
controls the power bridge. Different duty cycle results in
different vehicle speeds the remaining IO can be used as
advanced function input pins or system status indicating pins.

VII. CONCLUSION
This paper lists the hardware design guidelines for
designing an electric bike and also lists the design criteria for
an effective selection of the main components of the electric
bike such as the motor selection, battery selection, controller
selection, material selection, brakes, suspension system.
Additional protection features for the controller are also listed
as well as miscellaneous mechanical component selections to
design an elegant electric bike.
There is a combination of positive and negative effect of that
factor on perception. Here most of the respondents are consider
the cost and the mileage while purchasing a new bike, so there
is ample potential to electric bike in two-wheeler sectors. But
their battery is one of the factor which is effected the electric
bike because its performance is not good.

2285
 Proc. IX National Conference with International Participation "Electronica 2018", May 17 - 18, 2018, Sofia, Bulgaria

Overview of the Ways to Design an Electric Bicycle

Vladimir Dimitrov
Department of Electronics, Faculty of Electronic Engineering and Technologies
Technical University of Sofia
8 Kliment Ohridski blvd., 1000 Sofia, Bulgaria
[email protected]

Abstract – This paper reviews possible approaches to the

design of an electric bicycle with an emphasis on three II. MAIN METHODS
different domains – electrical, mechanical and system level When choosing a concept for designing an EPAC, one
design. It reviews the available solutions to a wide range of needs to be aware of the different approaches that are
issues in each of the above mentioned domains. After possible. In order to be able to list them in some order the
grouping the solutions presented and classifying them on a approach taken in this paper is to separate them into three
domain basis the paper presents a classification on the ways to main domains – Electrical Engineering Domain,
design an electric bike, so that a designer can easily obtain the Mechanical Engineering Domain and System Level Design
required information in order to start their project. Domain. The main questions in each domain are [5], [6]:
Keywords – Electric bicycles, Electric Power-Assisted • Electrical Engineering Domain –What type of
Cycles, Light Hybrid Vehicles electric machine to use? What motor drive control
algorithm to implement? , What power converters need to
I. INTRODUCTION be designed? What will the primary energy storage be
The deployment of electrified road transport in the range (possibly a hybrid one)? How to charge the battery or what
of 5-10% is one of milestones for the year 2020 in Europe type of BMS system to implement?
[1]. However, the amount of electrified vehicles is not • Mechanical Domain – Where to place the electric
increasing as steadily as lawmakers would like them to- motor? What gears should we use in order to combine
only 1.4% of the new vehicles were electric ones in 2017 human and machine torque?
[2]. There is one kind of electric transport, where the • System Level Design – What kind of power
desired goal for electrification level is already reached. topology to use? Should it assist only (pedelec) or can it be
With 8% of the new bicycles being of the Electric Power- driven with pure electric power (ebike)? Should there be
Assisted Cycles (EPACs) type, the electrification of this any energy regeneration in the bike?
kind of transport is already in a very advanced state [3]. The paper reviews the possible solutions to the above
The EPAC are lighter, more compact and cheaper than posed questions.
electric cars and can be considered to be the forerunner of
the impending electrification of road transport, which some III. PROPOSED CLASSIFICATION
envisioned as a way to achieve sustainable transport not A. System Level Design
dependent on fossil fuels [4]. Being cheaper to design, Looking at the system level EPACs are a light hybrid
experiment and manufacture, this kind of transportation has electric vehicle that combines human and electric power.
stimulated the engineers and scientists to try different There are two main ways to design such a vehicle - in a
approaches to implement the overall system. This paper serial [7] or parallel way [5], [6], [8]. The parallel case
aims to summarize the possible approaches for designing shown in Figure 1 is the more popular way of designing
such a vehicle by first separating the design phase into such a vehicle. In this case, human torque is combined in
three different domains – electrical, mechanical and system the mechanical domain with the torque from the motor via
level design. For each of them the paper tries to summarize gears [9]. In the serial configuration, shown in Figure 2 the
the main problems and possible solutions that the designer human and electrical machine torque are coupled in the
must face in order to build the vehicle. Finally, on the basis electrical domain. The double conversion of human power
of the presented information a classification is proposed has the positive effect that optimal cadence can be achieved
that can be used as a general overview of possible ways to at all slopes, while lowering the overall efficiency. When
design an electric bicycle. implementing the serial hybrid approach, special care must
The paper is structured as follows: in Section 2 the three be taken to ensure a natural feeling during cycling,
main design domains – electrical, mechanical and system minimizing the index of perceived exertion. The problem
level are presented, alongside questions for the engineer of optimal control for serial ebikes is the main topic in
designing the vehicle. These issues need to be addressed [10].
during the design phase of the bicycle. The overview of the In this initial system level design phase one more
possible solutions in each different domain is the main question must be addressed – Should the system include
topic of Section 3. There, each approach is first discussed, regenerative braking? And should the bicycle be able to run
with its advantages and disadvantages, and emphasis is on pure electric energy, without pedaling input from the
given to commercially available solutions. In the final cyclist (ebike) or the cyclist needs to apply power at all
section the proposed classification is presented. Finally, times in order to operate it (pedelec, PAS)?
Section 4 concludes the paper.

978-1-5386-5801-7/18/$31.00 ©2018 IEEE

 Regarding the first question some theoretical are some commercially available fuel cell bikes from
considerations predict that around 35% of the energy can Pragma Industries in France and Linde H2 Bike in
be recovered, in the case of an ideal recovery system in a Germany.
certain hilly city test track [11]. Other studies argue that The brushed dc motor, while very popular in early
efficiency is less than 10% [12], while some real data implementations of EPACs [8], has been gradually
showed that it is around 16% [13]. As the added cost to replaced by the BLDC (mostly with trapezoidal back emf)
implement such a system is mostly a control algorithm that for almost all conversion kits or ebikes. The main reason
is a worthwhile investment from a design perspective [14]. for this is its optimal performance in terms of reliability,
Regarding the second posed question, it should be noted power density and naturally smaller dimensions [5]. The
that the law in some countries answers this question other types of ac motors, namely the induction machine,
unilaterally [3], [5]. Most of the sold EPACs in Europe are synchronous and switched reluctance motors have not yet
of the pedelec type, and it seems to be the more popular gained popularity. The induction machine, although
choice [5]. cheaper and more reliable, has the disadvantages of lower
Control System torque density and lower efficiency. Thus its application in
4 ECU
1
2
small electric vehicles, such as an ebike, is not very
popular. One example using IM is shown in [22]. The
3

Energy Source 3Ph Bridge 3Ph AC

Machine
potential volatility in the price of rare earth magnets has
Out1

Energy
Gears
spurred interest in the switched reluctance motor [23]. In
Storage
1
2
Inverter BLDC
C
Wheel
Drive
its classical form it offers better fault tolerance, cost and
R
S
1 easer sensorless control, while its disadvantages are lower
Rider Input
Pedals
BLDC Machine
Motor efficiency, torque density, noise and larger torque ripple in
Human
Planetary
Wheel Drive comparison to the brushless motor. A more detailed
Input

1
Gear Set
comparison for advanced various types of motors and their
applications for electric vehicles is shown in [24].
In order to correctly commutate the BLDC motor with
Fig. 1. Parallel Configuration of an electric bicycle trapezoidal back-emf the position of the rotor must be
known. This can be done by deducing it from the measured
back-emf or some other non-direct method (called
sensorless control) [25]. The other available method is to
measure it directly by a rotary sensor such as Hall sensors
or an encoder [26]. The latter approach is more common in
ebikes, with motors having adopted the integrated Hall
sensor approach. When using Hall sensors there are two
different switching control schemes - a two-phase 120°
conduction or a three-phase 180° conduction scheme. The
former achieves higher maximum torque, while the latter
has the advantage of better operation in the constant power
region. Both schemes can be improved by phase advancing
the pulses, compared to the feedback signal from the hall
Fig. 2. Serial Configuration of an electric bicycle
sensors at higher speeds [27]. The main reason to use an
B. Electrical Engineering Domain encoder as the feedback mechanism is the possibility to
In Figure 1 and 2 it is assumed that the energy source is implement more sophisticated control algorithm such as
a chemical battery and the drive is achieved with a Direct Torque Control [28], [29]. The use of such control
brushless dc motor – both of which could be changed with mechanisms allows for tighter control of the obtained
other alternatives and the choice lies in the electrical torque or speed command, less ripple and better drive
engineering domain. dynamics. Out of these reducing the ripple torque is the
The most common topology for the energy storage is a most important aspect as it directly relates to driving
chemical energy source. The VRLA battery is the cheaper comfort.
choice, popular in China [15]. The Li-Ion battery is C. Mechanical Engineering Domain
becoming the standard when obtaining conversion kits or When considering the questions in the mechanical
buying a new ebike in Europe [16]. There are some domain the main ones are: Where to put the motor? Should
experiments about adding alternative energy sources, such it have gears? How to couple the torque from the human
as a fuel cell [17], [18] or a super capacitor to either replace and motor in case of a parallel hybrid configuration?
the battery [19], or supplement it [20]. As the standard bike If only one motor is used its position – front or rear
is not expected to achieve spectacular accelerations, and wheel hub motor or center mount motor- effects the
the energy pulses that need to be absorbed when traction, weight distribution and ease of maintenance and
implementing regenerative braking are not large, the installation. For conversion kits the rear and front wheel
benefit of adding a high specific power device, as the hub motor placement is the most predominant type,
supercapacitor, seems counterintuitive [21]. On the other because they are cheaper, can easily be installed by using a
hand, using hydrogen can lower the range anxiety, offering larger range of “of the shelf” motors and have less strain on
greater ranges utilizing cheap fuel and rapid charging, the drivetrain. On the other hand, the central mounted
which is not possible without swapping the batteries. There motor with its weight distribution offers the best overall
performance as it also utilizes the existing gears (this can Rohloff Speedhub with 14). They are more expensive and
be seen as a weak point as it puts more strain on this heavy compared to the derailleur mechanism, but offer
drivetrain). Being more expensive it is offered from fewer very high gear ratios. Their main drawback is that they
manufactures like Bosch in Europe or Bafang in China. cannot achieve a finite number of transmissions, so still
Using a hub motor that is mounted on the front wheel cannot utilize the full potential of the electric motor or the
provides the advantage of better weight distribution, all- human cyclist.
wheel drive with the human powering the rear wheel and Finally, continuously variable transmissions are
finally easer installation, maintenance (especially when becoming interesting for use in bicycles [35], [36], [37].
changing a flat tire) and better compatibility with the The NuVinci series of CVT have more than twice the
current installed bike drivetrain. The main drawbacks of weight of a derailleur system, and also the maximum ratio
the front wheel motor are that traction can be a problem in range of 380% is noticeably lower than the SpeedHub or
hilly regions or when using a higher powered motor. Also the best derailleur systems. However, the possibility of
improper installation can be more dangerous. If the motor electronic control and continuous adjustment of the gear
is put on the rear wheel the traction is better, especially in ratio for torque control can lead to efficient operational
hilly regions and off-road, but the bike can become conditions for both the motor and human [38]. It should be
unbalanced, especially if the battery is also installed at the noted that the neatest way to implement an e- CVT is when
back. The installation and maintenance are worse compared the bicycle is based on a serial hybrid topology, because
to the front wheel drive [5]. Finally, there are some the system needs two motor/generators in order to control
experiments with dual motors for the front and rear wheel the output torque and bring about the constant cyclist input
that should also be mentioned [30]. They offer better speed to varying wheel speed [37].
performance as ATV compared to other alternative and D. Proposed Classification
there are some commercially sold such as Motorino Figure 3 shows the proposed classification of electric
MTgX2. bicycles by summing all the presented information and
When installing the motor on the front or rear wheel categorizing it according to the specific domain that it is
two main alternatives exist – should the motor directly applicable to.
drive the wheel, or should it be a gear motor? The latter Back EMF
Trapezoidal
Sinusoidal
BLDC motor
case leads to smaller motors that are more efficient, as they Electrical
Machine
Control
Type
Sensered
Hall Sensor
Encoder
work at higher rpm. This ultimately means less weight, but Type Sensorless

adds some reliability issues with the gears. The heavier Electric
DC motor
Fuel Cell
direct drive design, which is more popular in commercially Engineering Energy Storage
Type Battery
VRLA
Li-Ion
sold conversion kits, can give more torque. Also due to its Domain
Ultracapacitor LiFePO4

larger volume the motor can cool more efficiently. [16]. In

Europe for BLDC geared motors can be obtained from Front Wheel
Direct Drive
Geared
Heinzman, Sanyo and Panasonic, while Crystalyte makes Motor
Placement Rear Wheel
Direct Drive

direct drive motors. Electric Mechanical Mid Drive

Geared

The central mounting of the motor provides the best Bycicle/ Engineering
Pedelec Domain
overall performance, as it uses the available drivetrain to Gear Type
Derailleur type

optimally adjust the gear ratio. However, this means that Hub gear

CVT
the cyclist must exercise greater care to the proper gear Series Hybrid
Topology
change, if he needs to operate the ebike optimally. In this Parallel Hybrid

configuration the motor and the pedals do not have the System Level Regeneration
Yes

No
inherent uncoupling, as in the hub motor case [31]. Design

Automatic gear changing can be a valuable option [32]. Electric Energy Pedelec
assist
The type of gear system is the last important thing to Ebike

Fig. 3. Proposed Classification

take into consideration, with the options being the
The proposed classification allows the designer of an
derailleur type, hub gear or the newer continuously variable
electric bike to quickly evaluate his options and look for
transmission.
additional information before commencing the actual
The predominantly used derailleur system is very light,
design.
cheap, has excellent efficiency and can achieve very large
IV. CONCLUSION
gear ratios [33], [34]. As it is a very sensitive mechanism
The primary aim of this paper was to present the various
its main drawbacks are the required maintenance and
approaches that have been accumulated in the scientific
associated reliability issues, especially for offroad trips. An
literature to design an electric bicycle.
additional drawback is that gears cannot be changed while
In order to group them in some manner a new
stationary or a gear ratio cannot be skipped. Additional
classification is proposed that groups the various issues that
problems are the efficient use of the available gear ratios
the designer is faced in three broad categories – system
and the step size between various gears [35].
level domain design, electrical engineering domain design
An alternative approach is the gear hub approach with a
and mechanical engineering domain design. For each of the
single sprocket, which dramatically reduces maintenance
three domain the main issues are presented alongside the
and increases reliability, while at the same time having the
most popular ways to solve them. The advantages and
advantage of linear distance between adjacent gear ratios
disadvantages in choosing a particular solution are also
and on spot gear shifting. Commercially available are
discussed.
usually with 8 gear ratios (with the exception of such as
 The goal of the paper is to help a designer of an electric bicycle,” in ENERGYCON 2014 - IEEE International Energy
bike choose an architecture, and also be aware of the Conference, 2014, pp. 1434–1439.
[20] N. Hatwar, A. Bisen, H. Dodke, A. Junghare, and M.
various issues that he must address.
Khanapurkar, “Design approach for electric bikes using battery
and super capacitor for performance improvement,” in IEEE
ACKNOWLEDGMENT
Conference on Intelligent Transportation Systems, Proceedings,
The study presented in this paper has been done with ITSC, 2013, pp. 1959–1964.
the support of Project №181ПР0013-03 of Technical [21] W. C. Morchin, H. Oman, and Wiley-Interscience, Electric
University Sofia. Bicycles: A Guide to Design and Use. Wiley, 2006.
[22] K. Fujii, H. Kumai, E. Ishii, S. Higaki, S. Yamaguchi, and H.
REFERENCES A. Zhao, “Researches on small capacity vehicles driven by
[1] European Commission, “Electrification of the Transport induction machine,” in 6th International Workshop on
System: Studies and reports,” 2017. Advanced Motion Control. Proceedings (Cat. No.00TH8494),
[2] Confederation of the European Bicycle Industry, “EUROPEAN 2000, pp. 465–468.
BICYCLE MARKET 2017 edition Industry & Market Profile [23] J. Lin, N. Schofield, and A. Emadi, “External-Rotor 6-10
(2016 statistics),” 2017. Switched Reluctance Motor for an Electric Bicycle,” IEEE
[3] E. Fishman and C. Cherry, “E-bikes in the Mainstream: Trans. Transp. Electrif., vol. 1, no. 4, pp. 348–356, 2015.
Reviewing a Decade of Research,” Transp. Rev., vol. 36, no. 1, [24] E. Bostanci, M. Moallem, A. Parsapour, and B. Fahimi,
pp. 72–91, 2016. “Opportunities and Challenges of Switched Reluctance Motor
[4] A. Emadi, “Transportation 2.0,” IEEE Power Energy Mag., vol. Drives for Electric Propulsion: A Comparative Study,” IEEE
9, no. 4, pp. 18–29, 2011. Trans. Transp. Electrif., vol. 3, no. 1, pp. 58–75, 2017.
[5] A. Muetze and Y. C. Tan, “Electric bicycles - A performance [25] P. P. Acarnley and J. F. Watson, “Review of position-sensorless
evaluation,” IEEE Ind. Appl. Mag., vol. 13, no. 4, pp. 12–21, operation of brushless permanent-magnet machines,” IEEE
2007. Transactions on Industrial Electronics, vol. 53, no. 2. pp. 352–
[6] W. C. Morchin, “Battery-powered electric bicycles,” in 362, 2006.
Northcon/94 Conference Record, 1994, pp. 269–274. [26] V. Naveen Kumar, A. Syed, D. Kuruganti, A. Egoor, and S.
[7] M. Corno, F. Roselli, and S. M. Savaresi, “Bilateral Control of Vemuri, “Measurement of position (angle) information of
SeNZA - A Series Hybrid Electric Bicycle,” IEEE Trans. BLDC motor for commutation used for e-bike,” in Proceedings
Control Syst. Technol., vol. 25, no. 3, pp. 864–874, 2017. of the 2013 International Conference on Advanced Electronic
[8] D. J. Eichenberg, J. S. Kolacz, and P. F. Tavernelli, “Baseline Systems, ICAES 2013, 2013, pp. 316–318.
testing of the EV Global E-Bike SX,” 2001. [27] S. K. Safi, “Analysis and simulation of the high-speed torque
[9] A. Fuchs, “Principles of Human-Electric Hybrid Drives for performance of brushless DC motor drives,” IEE Proc. - Electr.
Human Powered Vehicles,” in 6th European Seminar on Power Appl., vol. 142, no. 3, p. 191, 1995.
Velomobile Design, 2009. [28] S. B. Ozturk, W. C. Alexander, and H. A. Toliyat, “Direct
[10] J. Guanetti, S. Formentin, M. Corno, and S. M. Savaresi, Torque Control of Four-Switch Brushless DC Motor With Non-
“Optimal energy management in series hybrid electric Sinusoidal Back EMF,” IEEE Trans. Power Electron., vol. 25,
bicycles,” Automatica, vol. 81, pp. 96–106, 2017. no. 2, pp. 263–271, 2010.
[11] O. Maier, M. Krause, S. Krauth, N. Langer, P. Pascher, and J. [29] Y. Zhou, D. Zhang, X. Chen, and Q. Lin, “Sensorless Direct
Wrede, “Potential benefit of regenerative braking on electric Torque Control for Saliency Permanent Magnet Brushless DC
bicycles,” in IEEE/ASME International Conference on Motors,” IEEE Trans. Energy Convers., vol. 31, no. 2, pp. 446–
Advanced Intelligent Mechatronics, AIM, 2016, vol. 2016– 454, 2016.
Septe, pp. 1417–1423. [30] M. Guarisco, F. Gao, and D. Paire, “Autonomy and User
[12] T. Lipowsky, J. Stegmaier, M. Lang, and P. Li, “Simulative Experience Enhancement Control of an Electrically Assisted
potential assessment of regenerative braking with electric Bicycle with Dual-Wheel Drive,” IEEE Trans. Ind. Appl., vol.
bicycles under consideration of realistic boundary conditions,” 53, no. 2, pp. 1476–1484, 2017.
in 2017 International Conference on Research and Education in [31] G. Cipriani, V. Di Dio, R. Miceli, G. R. Galluzzo, and M.
Mechatronics (REM), 2017, pp. 1–7. Russo, “Evaluation of performance and efficiency and type
[13] S. Saponara, R. Moras, R. Roncella, R. Saletti, and D. approval of an electrically assisted bicycle drive,” in 2013
Benedetti, “Performance measurements of energy storage International Conference on Renewable Energy Research and
systems and control strategies in real-world e-bikes,” in SAS Applications (ICRERA), 2013, pp. 1163–1168.
2016 - Sensors Applications Symposium, Proceedings, 2016. [32] P. Tandon, A. Awasthi, B. K. Mishra, P. Rathore, and R. K.
[14] C. C. Hua and S. J. Kao, “Design and implementation of a Shukla, “Design and simulation of an intelligent bicycle
regenerative braking system for electric bicycles based on transmission system,” IEEE/ASME Trans. Mechatronics, vol.
DSP,” in Proceedings of the 2011 6th IEEE Conference on 16, no. 3, pp. 509–517, 2011.
Industrial Electronics and Applications, ICIEA 2011, 2011, pp. [33] C. Kyle and F. Berto, “The mechanical efficiency of bicycle
703–707. derailleur and hub-gear transmissions,” Hum. power, no. 52, p.
[15] J. X. Weinert, A. F. Burke, and X. Wei, “Lead-acid and 8, 2001.
lithium-ion batteries for the Chinese electric bike market and [34] E. A. Casteel and M. Archibald, “A Study on the Efficiency of
implications on future technology advancement,” J. Power Bicycle Hub Gears,” no. 56420. p. V013T14A044, 2013.
Sources, vol. 172, no. 2, pp. 938–945, 2007. [35] D. Rockwood, N. Parks, and D. Garmire, “A continuously
[16] E. Starschich and A. Muetze, “Comparison of the Performances variable transmission for efficient urban transportation,”
of Different Geared Brushless-DC Motor Drives for Electric Sustain. Mater. Technol., vol. 1, pp. 36–41, 2014.
Bicycles,” in 2007 IEEE International Electric Machines & [36] M. Park and N. Park, “Development of new continuously
Drives Conference, 2007, vol. 1, pp. 140–147. variable transmission for electric bicycles,” SAE Tech. Pap.,
[17] J. J. Hwang, D. Y. Wang, N. C. Shih, D. Y. Lai, and C. K. no. 1, pp. 1–5, 2005.
Chen, “Development of fuel-cell-powered electric bicycle,” J. [37] P. a. Watterson, “An electric assist bicycle drive with automatic
Power Sources, vol. 133, no. 2, pp. 223–228, 2004. continuously variable transmission,” 2008 Int. Conf. Electr.
[18] M. Bertoluzzo and G. Buja, “Development of electric Mach. Syst., pp. 2992–2997, 2008.
propulsion systems for light electric vehicles,” IEEE Trans. Ind. [38] C. L. Chen and M. C. Tsai, “Torque control of power-assisted
Informatics, vol. 7, no. 3, pp. 428–435, 2011. bike via planetary gear mechanism,” in 2017 20th International
[19] K. Malan, M. Coutlakis, and J. Braid, “Design and Conference on Electrical Machines and Systems (ICEMS),
development of a prototype super-capacitor powered electric 2017, pp. 1–5.
 ENERGYCON 2014 • May 13-16, 2014 • Dubrovnik, Croatia

Design and Development of a Prototype

Super-Capacitor Powered Electric Bicycle
Keagan Malan #1 , Michael Coutlakis #2 , James Braid #3
#
School of Electrical and Information Engineering, University of the Witwatersrand, Johannesburg
Private Bag 3, Wits, 2050, Johannesburg, South Africa
1
[email protected]
2
[email protected]
3
[email protected]

Abstract—The design and development of a prototype super- The envisaged idea is that a fleet of E-bikes and associated
capacitor powered electric bicycle (E-bike) is presented. An charging stations, would be owned and maintained by the
existing general-purpose bicycle was retro-fitted with a brushed university, and be provided as a service to its members. The
DC motor driving the front wheel, a super-cap bank capable of
storing 74 kJ, and a bi-directional motor driver for acceleration communal E-bikes would be distributed around the campus,
and regenerative braking. It was envisaged that the designed typically at the prominent lecture venues and large parking
E-bike would fulfil the role of a campus commuter, whereby areas. At each point, a dedicated rapid-charger would be
members of the university’s community could utilise such E- available, to replenish the E-bike quickly, before its next use
bikes to traverse a large campus. Consequently, the range of the by another user.
E-bike was not as important as the rapid and frequent charge
cycling requirements, hence the benefits of using super-capacitors The focus of this paper is on the design, development and
as an energy storage medium was exploited. For the design of evaluation of a suitable prototype E-bike.
the E-bike, an 80 kg cyclist was assumed. However, the developed
II. BACKGROUND
prototype was tested with a 55 kg cyclist and demonstrated a sat-
isfactory range (pedalling-free) and top speed of approximately A. Existing E-bike Solutions
1100 m and 23 km/h respectively. Two weaknesses of the design Performance specifications of existing commercial E-bikes
that need future consideration are the unbalanced charging of the
series-connected super-capacitors and the inadequate response are summarised in Table I.
time of passive thermal fuses under short-circuit conditions of TABLE I
the super-cap bank. T YPICAL P ERFORMANCE OF C OMMERCIAL E- BIKES
Index Terms—Electric bicycle, E-bike, super-capacitor, regen-
erative braking. Performance Typical value and reference
Speed 15 to 29 km/h [3], 32 km/h [1]
Range 13 to 31 km [3], 16 to 80 km [1]
I. I NTRODUCTION
Motor power 150 to 500 W [3]
A. What is an E-bike? Charge time 2 to 6 hours [1]
Usable life-time ≤ 400 charge cycles [1]
An E-bike is a bicycle incorporating an electric motor for
additional propulsion, in combination with the user’s pedalling All the E-bikes encountered in the literature survey use
[1]. In the case of a pedal-assisted E-bike, the electric motor batteries to store between 50 Wh and 500 Wh (corresponding
augments the user’s efforts whilst in the case of a power-on- to 180 kJ and 1800 kJ respectively) of energy [1], [3]. The hill
demand E-bike, the electric motor only provides power when climbing ability is directly related to the motor’s output power:
the user requires, whether pedalling or not. Irrespectively, all typical performance up a 4 % slope is 16 km/h, with an 80 kg
E-bikes have pedals and a lack thereof would result in it being user onboard [1].
classified as an electric scooter or other electric vehicle [2]. Although E-bikes are more efficient than say petrol scooters,
In some countries legislation limits the motor size and or there are some associated environmental concerns, in particu-
the E-bike’s maximum speed; for example in the USA, these lar lead pollution [4]. In China 95 % of E-bikes use lead-acid
are limited to 750 W and 32 km/h respectively [1]. batteries [4]. Manufacturing and recycling of these batteries
results in around 1 and 2 % of the lead being “lost” in the
B. The Concept of a Campus Commuter processes respectively [4], resulting in lead pollution. For
A typical university campus is subject to a significant this reason, E-bikes are not entirely green. This problem is
amount of pedestrian traffic. Often, lecture venues are located worsened as the usable life-time of most batteries is limited to
relatively far from established parking areas. The time and a finite number of charge-discharge cycles, typically between
energy spent on foot, daily, to traverse a typical campus is not 300 [4] and 400 [1]. Consequently, during the life of the E-
insignificant; to improve the quality of campus life, a campus bike, the battery bank may need to be replaced up to five times
commuter in the form of an E-bike, is proposed. [4], thus compounding the problem.

978-1-4799-2449-3/14/$31.00 ©2014 IEEE 1434

 ENERGYCON 2014 • May 13-16, 2014 • Dubrovnik, Croatia

B. Batteries versus Super-Capacitors III. P ROPOSED E- BIKE S OLUTION

The recent demand for electric vehicles has resulted in the A. Conceptual Design
rapid improvement in battery technology [5]. Fundamentally For the campus commuter concept to be a success, the E-
however, the life and performance of batteries is limited by bike needs to fulfil certain fundamental requirements: it needs
their electro-chemical process used to store the energy [6]. to be simple to use, rugged and reliable, safe, and offer a
Typically, super-capacitors have energy densities pleasant user experience. Since it can be assumed that the
(≈10 Wh/kg) an order of magnitude lower than batteries, users would be familiar with riding a bicycle, an existing
however their power densities (≈10 kW/kg) are an order steel-tube chassis is chosen as a base from which to build up,
of magnitude higher [7]. Super-capacitors are thus suited naturally keeping the familiar pedals and brake levers. This
for rapid charge and discharge applications. In addition, the allows the user to pedal-assist where necessary and offers a
number of cycles in their useful life is far superior, typically high degree of familiarisation. For safety reasons, the E-bike’s
exceeding one million, compared to a mere one thousand for performance needs to be limited in terms of its top speed,
batteries [7]. With this in mind, [8] presents a short-range yet must be able to accelerate quickly enough to clear an
electric bus using super-capacitors as the primary energy intersection or cross a street. For the user’s benefit, an interface
source. displaying vital information (such as range remaining) is
Prevalent amongst current electric vehicle energy storage also required. Finally, to ensure that the campus E-bikes’
systems is the hybrid combination of batteries and super- availability are a maximum, the charging time needs to be
capacitors. In this configuration, the batteries are used as the kept to a minimum.
primary energy source whilst the super-capacitors are used Since the alternative solution to the campus commuter
to supplement sudden power requests [9], [10]; essentially concept would be a fleet of ordinary pedal-powered bicycles,
making the best use of both technologies. this E-bike must offer to the user, better performance and a
Recent research into super-capacitors utilising graphene more pleasant experience than its contemporary.
nanostructures has indicated the possibility of producing ca-
pacitors with increased energy densities, comparable to lead- B. Desired performance
acid batteries (45.8 Wh/kg), whilst still maintaining their sig- The desired performance specifications for the E-bike are
nificantly higher power densities [11]. shown in Table II. The range is as determined in section II-C,
the speed is based on the performance of the commercially
C. Typical Campus Route available E-bikes in Table I, and the acceleration is governed
At the authors’ resident university, a typical worst case by that of a typical cyclist as given in [12].
campus route from the main student parking area to the Although not directly related to the desired performance of
prominent lecture venues was profiled. The elevation-vs- the E-bike, a charging time (from completely depleted to fully-
distance data is plotted in Figure 1. The total distance of the charged) deemed to be acceptable by the would-be users, has
route is approximately 1100 m and the change in elevation is also been specified.
approximately 50 m. TABLE II
D ESIRED SPECIFICATIONS FOR PROPOSED E- BIKE .

Performance Specification
Range 1100 m up a 4.5 % slope
Speed 20 km/h
Acceleration ≥ 0.6 m/s2
Charge Time ≤ 1 minute

C. Assumptions Impacting the Design

In general, an 80 kg cyclist and a 25 kg E-bike (bicycle,
super-cap bank and motor) are assumed and the specifications
given in Table II must be satisfied or exceeded. The anticipated
coefficients of drag and rolling resistance at the E-bike’s top
speed are assumed to be similar to that of a typical cyclist, as
Fig. 1. Elevation profile for worst case campus route. given in [13].
The user is anticipated to stop three times en route; erring
From this profile, the steepest gradient en route is 6.0◦ on the conservative side, regenerative braking is neglected and
whilst the average gradient is 2.6◦ ; the latter corresponding the energy required to re-accelerate is assumed equal to the
to an average slope of 4.5 %. From this data, the energy and kinetic energy at travelling velocity. The E-bike is designed
power requirements of the E-bike are determined in section to operate in a full assist mode such that the user need only
IV-A. pedal to extend their range i.e. pedalling is not essential for

978-1-4799-2449-3/14/$31.00 ©2014 IEEE 1435

 ENERGYCON 2014 • May 13-16, 2014 • Dubrovnik, Croatia

operation. Therefore, the energy required to satisfy the range Using readily available 350 F, 2.7 V super-capacitors, a total
stipulated in Table II must be sourced from the stored electrical of 87 pieces would be required, comprising three parallel
energy. strings of 29 each. At a cost of roughly $ 10 per piece, it
was decided for demonstration purposes, to limit the super-
IV. D ESIGN OF P ROTOTYPE E- BIKE cap bank to 58 capacitors (i.e. two strings of 29 each) with a
A system level block diagram of the prototype E-bike is storage potential of 74 kJ.
shown in Figure 2.
C. User Controls and Interface
An acceleration signal from the user must be derived from
a twist-grip actuator on the right-side of the handlebar, much
like that of a motorcycle. When the right-side brake lever is
squeezed by the user, a signal to initiate regenerative braking
must be triggered, and braking at a rate proportional to the
amount of activation is to be undertaken.
The user interface should comprise a handlebar-mounted
unit that indicates the current level of charge, speed and
distance, and estimated energy / range remaining.
D. Motor Driver
A motor driver is required to allow bi-directional power
Fig. 2. System level block diagram of the E-bike. flow between the super-cap bank and the motor. Power flow
to the motor is controlled via the accelerator twist-grip whilst
To aid familiarisation and safety, a fundamental decision power flow from the motor is controlled via the regenerative
was made to keep the existing system (pedal-drive and brake lever. Although this task is usually simple in the case of
mechanical brakes) de-coupled from the retro-fitted system a DC machine fed from a constant voltage, it is slightly more
(electric propulsion and regenerative braking): the rear wheel complicated with the super-cap bank’s variable voltage.
remains driven by the pedals and is mechanically braked by E. Motor and Power Transmission
the left-side brake lever, whilst the front wheel is motorised
Since a 500 W DC machine (briefly mentioned earlier)
and regenerative braking is actuated via the right-side brake
that would satisfy the power requirements above was readily
lever.
available, it was used in the prototype E-bike. From the
A. Energy, Power and Acceleration nameplate, the machine produces 1.91 Nm of torque at its
rated speed of 2500 rpm, when supplied with 36 V and drawing
The potential energy (mechanical) required in section III-C
18 A. A pair of sprockets with a gear ratio of 15:1 would be
to overcome the height gain, the drag loss and that required for
required to reduce the speed and increase the torque to the
acceleration, are summed to 64 kJ. A typical 500 W permanent
specified values, when driving a 26 ” wheel.
magnet brushed DC motor has an efficiency of 78 % [14],
loosely assumed to remain constant across its operating range. F. Rapid Charger
The required motor driver would typically be 80 % efficient. A charger is required to replenish the super-cap bank from
This results in an energy requirement of 102 kJ (electrical). the electrical mains supply. A custom designed charger was
For a 25 kg E-bike mounted with an 80 kg user, to climb chosen in order to exploit the fast charging capability of
a 4.5 % slope at 20 km/h, a power of 250 W (mechanical) the super-capacitors. The rated charging current of the super-
is required. As a bench mark, a cyclist’s acceleration from capacitors is 20 A [15], therefore a constant current charging
standstill is typically between 0.6 and 0.8 m/s2 [12], requiring algorithm would be ideal.
a wheel-surface contact force of at least 63 N for the 105 kg
combination. V. I MPLEMENTATION OF P ROTOTYPE
The prototype E-bike is pictured in Figure 3 where the
B. Choice of Energy Storage
super-cap bank, user interface, and motor and power trans-
Exploiting their rapid charging ability and high number of mission are clearly visible.
useful cycles, super-capacitors were selected as the energy
source since a practically sized 6 kg bank could store the A. Super-cap Bank and Charge Balancing
required 102 kJ [15] to cover the desired range. In addition, The prototype E-bike’s super-cap bank comprises 58 ca-
super-capacitors are more readily available and perhaps safer pacitors, arranged in two parallel strings of 29 each. This
than other alternatives, such as flywheels. With ∼ 95 % of the configuration was not only chosen to avoid excessive current
super-capacitors’ stored energy being extractable, to satisfy levels at the lower voltage range, but also to avoid an excessive
the energy requirements of section IV-A, the required energy amount of series connections so as to limit the number of
estimate rises to 107 kJ (electrical). balancing circuits should these need to be added in the future.

978-1-4799-2449-3/14/$31.00 ©2014 IEEE 1436

 ENERGYCON 2014 • May 13-16, 2014 • Dubrovnik, Croatia

can buck or boost in either direction and thus easily allows

both acceleration and regenerative braking due to the bi-
directional nature. The ability to step up or step down in
either direction is extremely flexible. This capability allows
the super-cap bank to be used over a large voltage range that
extends above and below the motor’s rated voltage, and also
allows regenerative braking at any super-cap bank voltage and
and at any motor speed.

Fig. 3. Implemented prototype E-bike.

The resulting voltage range from 16 V to 78 V means that 96 %

of the stored energy can be extracted from the super-cap bank,
equal to 71 kJ. Fig. 4. Topology of bi-directional motor driver.
The two strings of super-capacitors are soldered onto two
PCBs and are vertically stacked to form a neat and compact In Figure 4, the MOSFET drive signals S1 to S4 are
super-cap bank. The complete bank is mounted inside a generated by bootstrap MOSFET driver ICs. MOSFETs M1
weather-proof box and mounted on a purpose-built aluminium and M4 require gate drivers with 100 % duty cycle capability
carrier, overhead the rear wheel. so simple floating fly-back supplies are used to power the
The tolerance in capacitance of each super-capacitor is bootstrap ICs. The motor current is sensed by a bi-directional
typically about 15 %. This impacts the charging strategy in Hall effect current sensor.
that, when charged by the same current in a series-connected As suggested in [17], the accelerator and brake inputs
string, some capacitors would be charged to higher voltages provide a torque reference signal. Since the DC motor’s torque
than others. Although various voltage balancing methods exist is directly proportional to its armature current, the current is
[16], the simplest method to prevent over-charging is to de-rate measured and controlled relative to the reference signals.
the maximum voltage of the bank. Subsequently reduced by A PIC16F1827 microcontroller is used to control the motor
a 10 % margin, the super-cap bank is charged to a maximum driver since it has four PWM generators, sufficient analogue
of 70 V. channels and is inexpensive. Closed loop control is necessary
To prevent catastrophic fault currents flowing should the between the accelerator or brake input and the PWM signals
super-cap bank be accidentally short-circuited, a 60 A thermal due to the wide variation of the super-cap bank voltage.
fuse was installed between the super-capacitor strings and the Initially, the microcontroller selects one of the four buck / boost
bank’s connection terminals. operating modes according to the brake state and the voltages
across the super-cap bank and motor. The microcontroller
B. Motor and Power Transmission implements a simple integration of error between the desired
The existing motor was mounted directly above the front and measured motor currents, which is used to determine the
wheel, supported by a steel sub-frame, and connected to the duty cycle. A duty cycle of more than 100 % in buck operation
wheel’s axle and the chassis’s front fork. The DC motor was or less than 0 % in boost operation is used to change between
supplied with an 11-tooth pinion pre-fitted; to achieve the the converter modes.
required 15:1 gear ratio, a corresponding 165-tooth sprocket A motor starting current of 40 A is allowed to provide
was laser cut and mounted flush against the front wheel’s hub. sufficient starting torque. Maximum current is limited to the
Finally, the motor and wheel were linked with an ASA25 type motor’s rated current of 18 A once the motor terminal voltage
chain. has reached 16 V.
1) Accelerator and Regenerative Brake: The twist-grip
C. Motor Driver Circuit accelerator is implemented by a single-turn potentiometer,
The permanent magnet brushed DC motor only requires axially connected to a knurled tube, that loosely fits over the
two-quadrant operation: both driving and braking in the existing handlebar. The grip’s travel is limited by a grub screw
forward direction. A bi-directional Buck-Boost Cascade as and a torsion spring returns it to its zero position.
described in [17] is very well suited to this application. The The regenerative brake makes use of the existing brake
structure of this converter is shown in Figure 4. This topology lever but replaces the cable with with a micro-switch and

978-1-4799-2449-3/14/$31.00 ©2014 IEEE 1437

 ENERGYCON 2014 • May 13-16, 2014 • Dubrovnik, Croatia

TABLE III
linear potentiometer. The micro-switch is used to detect and M EASURED PERFORMANCE FOR THE PROTOTYPE E- BIKE .
initiate braking, and signals the motor driver microcontroller
to undertake the regenerative braking operation (which takes Mass of user (kg) 55 65 100
preference over the accelerating operation); the amount of Distance travelled (km) 1.09 1.03 0.90
Economy (km/MJ) 20.1 19.1 16.4
braking being proportional to the linear potentiometer’s output.
Max. speed (km/h) 22.6 20.8 18.4
D. User Interface
The instrumentation consists of a 16 character, 4 line, al- Although not specifically tested with an 80 kg user, the
phanumeric LCD screen which displays the energy remaining results above show that the prototype E-bike satisfies the
in the super-cap bank, the speed of the E-bike and the distance requirements for range and speed as specified in Table II. As
travelled. All the computation and interfacing with the screen one would expect, the performance diminishes as the specified
is achieved with a PIC16F886 microcontroller. mass of the user is exceeded.
Using a potential divider, a voltage proportional to that The E-bike’s static force at standstill, with the motor draw-
of the super-cap bank is measured and the corresponding ing its starting current of 40 A, was measured to peak at 220 N,
amount of energy is computed. A small magnet attached to by means of a spring scale anchored to the rear of the bicycle.
one spoke of the front wheel provides a pulse-per-revolution For an 80 kg user, this equates to an acceleration of 2.1 m/s2 -
signal via a reed switch mounted on the front fork. From the more than three times the specified value - and improves the
period between the pulses, the angular velocity of the wheel E-bike’s hill climbing ability to a slope of 21.9 % after which
and hence speed of the E-bike is determined. Similarly, the it would stall.
distance travelled is computed from the number of pulses
counted since the last reset. Although not implemented, the B. Evaluation along Campus Route
range remaining can be easily determined from the three
The E-bike was evaluated along the route described in
functions above.
section II-C, by a 55 kg user. Although the capacity of the
E. Rapid Charger super-cap bank had been reduced from 107 kJ to 74 kJ during
the design stage, the E-bike was re-charged en route (the
A Two-Switch Forward Converter [18] was chosen to charge markers in Figure 1 show these points) and required a total
the super-cap bank. This topology was chosen as it is often of 126 kJ to reach the destination; significantly more than the
employed in arc-welding machines which perform a similar estimated 107 kJ and despite the significantly lighter user. For
purpose of controlling a high output current [18]. Compared an 80 kg user, an additional 12 kJ of potential energy alone
to other viable topologies, this one offers two advantages: it would be required, ignoring system inefficiencies.
requires two fewer switches than the Full-Bridge Converter
The large discrepancy can be attributed to two assumptions
and does not require the input capacitors as in the Half-Bridge
that were under-estimated: Firstly, the motor’s efficiency of
Converter.
78 % is a maximum when delivering rated power only; at
Essentially the charger is designed as a constant current lower speeds, such as when hill climbing, the lower back EMF
source that monitors the output voltage in order to prevent internal to the machine results in higher currents being drawn,
over-charging the super-cap bank. The charger is intended to and hence higher losses, both within the machine itself and
charge three parallel strings simultaneously, each at 20 A, and the motor driver circuitry. Secondly, the friction and windage
was therefore designed to output up to 60 A. losses in the motor and power transmission are amplified
With the intention of powering the charger from a domestic by the gear ratio and result in a significant braking torque;
AC outlet, the charger had a preset upper power limit that this would also make any pedal assistance by the user quite
was implemented to avoid tripping the mains supply. The demanding.
charger circuit is controlled with a dsPIC30F2011. Although
more expensive, it had far superior capabilities when compared C. Motor Driver Efficiency
to the low-end devices. The transformer and inductor of the
The bi-directional Buck-Boost Cascade proved to be very
charging circuit where hand-wound onto ETD59 cores, using
well suited as a DC motor driver. It can provide a starting
2 mm copper wire to minimise Ohmic losses.
current of 40 A and the use of the duty cycle ratio to change
Rated at 3 kW, the charger is capable of replenishing the
between buck and boost modes is simple and effective.
E-bike’s super-cap bank in under 30 seconds, well within the
To test the efficiency, the motor was replaced with a 2 Ω,
requirement of Table II.
600 W resistive load. The controller was programmed to give
VI. E VALUATION AND R ESULTS a constant output current, across a range of super-cap bank
voltages, such that the load power remained approximately
A. Measured Performance constant at 500 W. The results reveal an average efficiency of
The prototype E-bike was charged to 70 V and evaluated on about 92 %, especially near the transition from boost to buck,
a level, oval-shaped course, by users of different masses; the at circa 30 V. At lower voltages, the higher input current results
results are shown in Table III. in increased losses.

978-1-4799-2449-3/14/$31.00 ©2014 IEEE 1438

 ENERGYCON 2014 • May 13-16, 2014 • Dubrovnik, Croatia

D. Recovery of Energy during Regenerative Braking a super-cap bank of 58, 350 F capacitors, capable of storing
After driving the E-bike up the first segment (between the 74 kJ of energy, and a 500 W permanent magnet DC motor
first two markers) of the route in Figure 1, it was free-wheeled driving the front wheel. A bi-directional motor controller
back to the start, whilst braking, and 6.8 kJ of energy was allows for motoring and regenerative braking. In addition, a
recovered. This segment corresponds to a change in potential 3 kW rapid charger was also designed allowing the super-cap
energy of approximately 17 kJ, implying a regenerative brak- bank to be fully charged within 30 seconds; the benefit of
ing system efficiency of 40 % under these test conditions. using super-capacitors as the primary energy storage medium
The motor driver is symmetrical and so its high efficiency clearly being exploited. Despite under-estimating the system
(≈92 %) is expected to be the same irrespective of the di- inefficiencies, the results from the prototype E-bike appear
rection of current flow through it. Therefore, the significant promising, making the campus commuter concept technically
inefficiency of the regenerative braking must be due to losses feasible.
somewhere other than the motor driver; the high friction and R EFERENCES
windage losses, as described in section VI-B, are suspected. [1] A. Meutze and Y. Tan, “Electric bicycles - a performance evaluation,”
IEEE Industry Applications Magazine, vol. 13, no. 4, pp. 12–21, Aug.
E. Future Improvements to Prototype 2007.
[2] J. Weinert, A. Burke, and X. Wei, “Lead-acid and lithium-ion batteries
Improvements to the following components are deemed for the Chinese electric bike market and implications on future technol-
essential for the future success of such an E-bike: ogy advancement,” Journal of Power Sources, vol. 172, pp. 938–945,
Over-Current Protection: During an accidental short-circuit May 2007.
[3] H. Oman, W. Morchin, and F. Jamerson, “Electric-bicycle propulsion
of the super-cap bank, the time response of the thermal fuse power,” WESCON ’95, pp. 555–560, Nov. 1995.
proved inadequate. It is highly recommended that active over- [4] C. Cherry, J. Weinert, and Y. Xinmiao, “Comparative environmental
current protection be employed to allow for a faster response impacts of electric bikes in China,” Transportation Research part D,
vol. 14, pp. 281–290, 2009.
time under fault conditions. [5] J. Voelcker, “Lithium batteries take to the road,” IEEE Spectrum, vol. 44,
Charge Balancing: To prevent the possible over-charging no. 9.
of individual super-capacitors, the super-cap bank voltage was [6] A. Mirhoseini and F. Koushanfar, “Hypoenergy. hybrid supercapacitor-
battery power-supply optimization for energy efficiency,” Design, Au-
de-rated from 78 V to 70 V. As a result of this trade-off, the tomation and Test in Europe Conference and Exhibition, pp. 1–4, Mar.
total energy storage potential reduced from 74 kJ to 60 kJ. A 2011.
charge balancing circuit, such as those described in [16] would [7] C. Goh and A. Cruden, “Automated high current cycling test system
for supercapacitor characterisation,” International Symposium on Power
allow the charging of the super-cap bank to its full potential Electronics, Electrical Drives, Automation and Motion, pp. 748–753,
without running the risk of over-charging. Jun. 2012.
Regenerative Brake Lever: Whilst this worked effectively, [8] K. Zhiguo, C. Zhu, Y. Shiyan, and C. Shukang, “Study of bidirectional
dc-dc converter for power management in electric bus with supercapac-
the lack of tactile feedback that a typical cyclist is accustomed itors,” Vehicle Power and Propulsion Conference, IEEE VPPC ’06, pp.
to whilst squeezing the brake lever may result in the me- 1–5, 2006.
chanical failure of the internal potentiometer when squeezed [9] M. Bertoluzzo and G. Buja, “Propulsion systems for light electric
vehicles,” IEEE Transactions on Industrial Informatics, vol. 7, no. 3,
too hard. A spring-loaded lever with a mechanical end-stop pp. 428–435, Aug. 2011.
is recommended to provide the tactile feedback and prevent [10] E. Lomonova, A. Vandenput, J. Rubacek, B. d’Herripon, and G. Roovers,
over-extension respectively. “Development of an improved electrically assisted bicycle,” 37th IAS
Industry Applications Conference, vol. 1, pp. 384–389, Oct. 2002.
Rapid Charger Interface: The super-cap bank was charged [11] J. Lin, J. Zhong, D. Bao, J. Reiber-kyle, W. Wang, V. Vullev, M. Ozkan,
directly from the rapid charger via polarised connectors. For and C. Ozkan, “Electrochemical supercapacitor based on flexible pillar
the envisaged campus commuter concept, it is proposed that graphene nanostructures,” 69th Annual Device Research Conference, pp.
91–92, Jun. 2011.
a split core transformer be used, with one half of the core [12] W. Pein, “Bicyclist performance on a multiuse trail,” Transportation
located inside the E-bike’s chassis and the corresponding other Research Record 1578, vol. 1578, no. 970616, pp. 127–131, 1997.
half mounted on the bicycle stand at the charging station. [13] S. Brand and N. Ertugrul, “Electric assisted bicycles and measurement of
real-time performance characteristics for power management strategies,”
Tracking and Management of E-bike Fleet: With the ad- AUPEC Power Engineering Conference, pp. 1–6, Dec. 2007.
dition of a GPS and suitable communication module, the [14] (2013) Unite Motor, Traction motor (Brushed PMDC). [Online].
whereabouts of the campus’ fleet could be tracked for man- Available: http://www.unitemotor.com/en/cp036.htm
[15] (2013) Ioxus Inc., Series: Multi-Pin. [Online]. Available:
agement purposes. For security purposes, using the GPS data, http://www.ioxus.com/ultracapacitors/
the operational area of the E-bikes may be confined to campus [16] D. Linzen, S. Buller, E. Karden, and R. D. Doncker, “Analysis and
use only. evaluation of charge-balancing circuits on performance, reliability and
lifetime of supercapacitor systems,” IEEE Transactions on Industry
Applications, vol. 41, no. 5, pp. 1135–1141, September/October 2005.
VII. C ONCLUSION [17] F. Caricchi, F. Crescimbini, F. G. Capponi, and L. Solero, “Study of bi-
The concept of a campus commuter, comprising a fleet directional buck-boost converter topologies for application in electrical
vehicle motor drives,” Presented at the Applied Power Electronics
of E-bikes, has been presented. This paper focuses on the Conference and Exposition, vol. 1, pp. 287–293.
specification and design of such a power-on-demand E-bike. A [18] H. Niilo and T. Vaimann, “Two switch forward inverter for parallel-series
prototype E-bike that satisfied the 1100 m range and 20 km/h resonance alternating (PSA) converter for supplying electric welding
arc,” 10th International Symposium on Topical Problems in the Field of
top speed was developed and tested. The E-bike comprises Electrical and Power Engineering, pp. 140–144, Jan. 2011.

978-1-4799-2449-3/14/$31.00 ©2014 IEEE 1439

 Design, implementation and experimental
results of a wireless charger for E-bikes
Filippo Pellitteri**, Nicola Campagna*, Vincenzo Castiglia*,
Alfonso Damiano***, Rosario Miceli*, IEEE member,
*Department of Engineering,University of Palermo, Palermo, Italy
**School of Aerospace Engineering, Sapienza University of Rome, Rome,Italy
***Department of Electrical and Electronics Engineering, University of Cagliari- Italy-
Abstract— Based on the Inductive Power Transfer (IPT), the up-to-date idea of the active demand: in a smart
the wireless energy transmission is increasingly representing electrical network, the consumer is able to become
an attractive solution for vehicle battery charging. Due to its producer of energy. In the case of a surplus of energy
high smartness, the wireless solution may be considered an
stored in some devices of the grid, the power can flow
interesting battery charging method for electric bicycles, as
they represent light-weight and flexible means of towards other devices requiring energy. The wireless
transportation. According to the Vehicle-To-Grid (V2G) power transfer can represent a support to V2G, and
concept, the wireless power flow can occur in both the therefore be a Bi-Directional Inductive Power Transfer
alternative directions: from the grid to the battery or in the (BDIPT) [7]. Furthermore, the implementation of proper
opposite way. A Bi-Directional Inductive Power Transfer control strategies for the efficiency optimization and the
(BDIPT) system is therefore particularly convenient in the
investigation of different geometries of magnetic couplers
scenario of a multi-parking area. For the E-bike application,
a bicycle-to-grid or a bicycle-to-bicycle energy transfer is a have been widely investigated [8-9]. In this paper, based
viable solution by means of BDIPT. In this paper, a 100W on the magnetic coupler presented in [10-12], an
prototype of wireless battery charger for E-bikes is Unidirectional IPT-based system for E-bikes is presented,
proposed and described. along with the FPGA-based control system and the
technical information regarding the practical
Index Terms-- electric bicycles, inductive power transfer, implementation [13-16]. The simulation and experimental
wireless battery charging, bi-directional power converter
results are described as well. Previous experimental
I. INTRODUCTION results concerning the work and the physiological
compatibility of the prototype have been reported in [17-
The Electric Bicycles (E-bikes) are light and compact
21], as well as a deep investigation on the excellent
vehicles, representing a potentially consistent category of
results arising from FPGA applied to power converter
transportation means in the current and future scenarios
control [22]. The IPT system is also addressable to a bi-
of a smart and green urban mobility. The wireless
directional power flow, by using integrated and hybrid
solution may be considered an attractive method for E-
energy storage systems [23-24]
bike charging. E-bike wireless charging is based on the
Inductive Power Transfer (IPT), which is a widely used II. MODELLING OF THE UNIDIRECTIONAL INDUCTIVE
approach for consumer electronics [1]. In [2-3] an POWER TRANSFER (UIPT) SYSTEM
investigation about different coupling solutions is carried
Fig. 1 shows the circuit diagram of a typical UIPT
out, all of them consisting of a magnetic coupler made of
system. The DC source U1,dc is connected to the
a transmitter buried underground and a receiver installed
transmitter stage, composed by an high frequency DC/AC
inside the bicycle kickstand. Different cases of kickstand
NMOS Inverter and the primary coil of the magnetic
are investigated as well. The distance between transmitter
coupling structure. The capacitors C1 and C2 are used to
and receiver of the magnetic coupler is 2 cm. In [4] the
minimize the reactive power drawn from the system,
bicycle coil is placed on the side of the front basket and
creating a compensation network. There are different
the grid-connected coil is installed next to a wall, both in
compensation configurations depending on the type of
a vertical position. The wireless charging operation
connection between the coils and the capacitors: series-
occurs at a 5 cm distance between the coils. In [5] the
series (SS), series-parallel (SP), parallel-series (PS) and
receiver part is installed under the saddle of the bicycle,
parallel-parallel (PP). Each compensation topology has its
so that a mechanically movable coil is needed in the
own features.
transmitter side in order to perfectly align both the sides.
In [6] the receiver coil is put inside the bicycle kickstand.
The inductive coupling between the electrical grid and
the E-bike can be also exploited for a reverse power flow,
that is from the vehicle to the grid. By exploiting
reversible power electronics stages, the power is able to
flow from the grid to the rechargeable battery and from
the battery to the grid, according to the Vehicle-to-Grid
(V2G) idea. V2G is a wide-spread concept, belonging to Fig. 1. Circuit diagram of an UIPT.

978-1-7281-1356-2/19/$31.00 ©2019 IEEE 364

 The proposed UIPT is based on a SS compensation control algorithm.
because it offers some benefits for the wireless charger
application:
1) the resonant frequency can be chosen independently
of the magnetic coupling and the load;
2) the secondary current is independent from the load.
The receiver stage is composed by the secondary coil and Fig. 4.Block diagram of the closed loop control for the UIPT system.
its compensation network and an high frequency AC/DC
full-wave rectifier. The rectified voltage U2,dc is used to A. H-bridge design
recharge a battery.
By considering only the fundamental component of the The main components of an H-Bridge inverter are the
system, an equivalent circuit can be obtained. The power switches and the driver circuit. The IRFB4410Z
inverter is controlled through the Phase Shift technique power N-Channel MOSFETs [13] have been chosen for
and considering only fundamental component of the this design. They are rated for 100 V and 97 A, with a 9
output voltage, u1, it can be replaced by a sine-wave m on resistance, and are suited for high frequency
generator with frequency Ȧ0 and amplitude: applications. The maximum ratings are far above the
operating values of voltage and currents of the IPT
ସ system so that they will operate in safe conditions. To
‫ݑ‬ଵ ൌ ܷଵǡௗ௖ •‹ ߚ (1)
గ drive the 4 N-Channel MOSFET of the H-Bridge
In the receiver side, the AC/DC rectifier and the Inverter, the HIP4081A Driver [14] has been used. It is a
battery can be replaced by an equivalent AC resistance as high frequency, medium voltage IC Driver, capable of
shown in Fig. 2. In this way the equivalent circuit shown driving independently 4 N-Channel MOSFET in Bridges
in Fig. 3 is obtained. For this circuit the following set of configurations. It supports up to 80 V and has a peak
equations can be written: current capability of 2.5 A. For a proper operation of the
Driver, an external Bootstrap circuit with a capacitor, a
ଵ diode and a resistance is needed. To handle the high
‫ݑ‬തଵ ൌ ଓҧଵ ቀܴଵ ൅ ሺ௝ఠ஼ ሻ ൅ ݆߱‫ܮ‬ଵ ቁ െ ݆߱‫ܯ‬ଓҧଶ (2a)
భ frequency operation, ceramic capacitor and fast Schottky
ଵ
݆߱‫ܯ‬ଓҧଵ ൌ ଓҧଶ ቀܴଶ ൅ ሺ௝ఠ஼ ሻ ൅ ݆߱‫ܮ‬ଶ ቁ ൅ ‫ݑ‬തଶ (2b) diodes have to be used. The SB1100 [15] diodes have
మ
been chosen for this design.
From the set of equations 2 the primary and secondary The sizing of these component can be done as reported
current can be calculated. It can be found that, when the in [16], taking into account the total charge that has to be
system operates at the resonance frequency Ȧ0 the supplied to the circuit. The total charge QT can be
secondary current ଓҧଶ is: calculated considering the different contributions to the
bootstrap capacitor discharge:
‫ݑ‬തଵ (3)
ଓҧଶ ሺ߱ ൌ ߱଴ ሻ ൌ ݆ 
߱଴ ‫ܯ‬ ்ܳ ൌ ൫‫ܫ‬௅௄ǡீௌ ൅ ‫ܫ‬ொǡ஻ௌ ൅ ‫ܫ‬௅௄ǡ஻ௌ ൅ ‫ܫ‬௅௄ǡ஽௜௢ௗ௘ ൯‫ݐ‬௢௡ǡு ൅ ܳீ ൅ ܳ௅ௌ
(4)

where:
• QG is the maximum total gate charge of the
MOSFET;
• QLS is the charge required by the internal level
shifters;
• ILK,GS is the drain to source leakage current;
Fig. 2. Equivalent circuit of the rectifier stage. • IQ,BS is the bootstrap circuit quiescent current;
• ILK,BS is the bootstrap circuit leakage current;
• ILK,Diode is the diode reverse bias leakage current;
• ton,H is the maximum ON time of the high-side
MOSFET.
The maximum allowable voltage drop in the bootstrap
capacitor ¨VBS can be calculated as:
Fig. 3. Equivalent circuit of the UIPT system.
οܸ஻ௌ ൌ ܸ஽஽ െ ܸி െ ܸீௌǡ௠௜௡ െ ܸ஽ௌǡ௢௡ (5)

III. DESIGN OF THE UIPT SYSTEM where:

• VDD is the driver supply voltage;
The system was built entirely within the laboratory.
• VF is the forward voltage of the bootstrap diode;
The design process has covered the H-Bridges, the
• VGS,min is the minimum gate-source voltage;
coupled coil and the compensation network and the

365
 • VDS,on is the voltage across the MOSFET while in current is needed. The current signal must be fitted to the
on state. range of the ADC of the control board. The analog input
All these characteristics can be found in the can range from 0.4 to 2.9 V and the conditioning circuit
components datasheets. The bootstrap capacitor and was designed in order to take advantage of the full range.
resistor must be chosen according to the following: To sense the current, the LVR03R0100 resistor was used.
It has a 0.01 value with ±1% tolerance and 3 W power
ொ೅
‫ܥ‬௕௢௢௧ ൐ (6a) rating. Considering a 0 to 10 A range for the load current,
ο௏ಳೄ
௧೚೙ the voltage on the resistor will range from 0 to 0.1 V
ܴ௕௢௢௧ ൏ (6b)
ସ்ೄ respectively. The maximum dissipated power in the
resistance will be:
Due to high frequency operation, a snubber circuit is
ଶ
also required to minimize the ringing effect associated ܲ௦௘௡௦௘ ൌ ܴ௦௘௡௦௘ ‫ܫ‬ଶǡௗ௖ ൌ ͲǤͲͳ ή ͳͲଶ ൌ ͳܹ (9)
with the reverse recovery effect in power semiconductor.
The sizing of the snubber circuit must be carried out after that is below the rated power of the resistor.
the realization of the PCB because oscillations depend on To amplify and filter the voltage signal, an OP amp
the board parasitic capacities and inductances. The stage, shown in Fig. 5 , was used. The transfer function of
snubber capacitance will be chosen, knowing the this stage, in Laplace domain, is:
MOSFET output capacitance Coss, in the range
(0.5÷2)Coss. To choose the snubber resistance, an ‫ݒ‬௢ ܴ௕ ͳ (10)
ൌ
experimental test on the board is needed to measure the ‫ݒ‬௜ ܴ௔ ͳ ൅ ܴ௕ ‫ܥ‬
ringing frequency fring. The value of snubber resistance
Rsnub is calculated with the following expression:

ͳ (7)
ܴ௦௡௨௕ ൌ
Ͷߨ݂௥௜௡௚ ‫ܥ‬௢௦௦

With the future intention of making a bidirectional

system, the same H-bridge was connected at the
secondary side without sending any command and using
it as a passive rectifier. Fig. 5. Schematic of the conditioning circuit.

The OP amp has to amplify the voltage on the resistor

B. Coil and compensation network design to coincide with the maximum input voltage of the ADC,
The proposed structure consists of two identical flat equal to 2.5 V. This can be done by choosing Rb/Ra = 25.
winding coils made of 3 mm diameter copper wire. To To measure only the DC component of the signal and
minimize leakage flux, ferromagnetic material would be filter the high frequency component, a cutting frequency
needed but in order to lighten the weight of the wireless ft of 170 kHz has been chosen. To achieve this, the value
charger, taking into account that it has to be placed on a of the capacitor must be equal to:
bike, it was not used. In future development, a
ferromagnetic core could be inserted at least in the ͳ (11)
‫ܥ‬ൌ
transmitter side. The coil design was carried out taking ʹߨ݂௧ ܴ௕
into account magnetic simulation with femm and
experimental tests, reported in [10-12] and [17-19]. The The 0 voltage level has to be shifted to 0.4 V to
main characteristics of the coils are: self-inductance L1 = coincide with the minimum input voltage of the ADC.
L2 = 13 H; k = 0.377; air gap = 3 cm; resistance R1 = R2 This can be done by using an OP amp in level shifting
= 8.2 m; 9 turns; outer diameter = 15 cm; inner configuration, using a variable resistor in the positive
diameter = 9 cm. Experimental tests concerning the input to manually adjust the offset as shown in Fig. 5. To
physiological compatibility of the proposed magnetic realize the conditioning circuit the TL082 OP Amp, PET
coupler have been reported in [20-21]. The compensation capacitors, common resistors and a trimmer were used.
capacitance must be chosen according to the operating The values of the component are summarized in Table I.
frequency and the coil inductance with the following
TABLE I
formula: CONDITIONING CIRCUIT COMPONENTS VALUES
PARAMETER VALUE
ͳ (8)
‫ܥ‬ଵ ൌ ‫ܥ‬ଶ ൌ Resistor Ra 25 k
߱଴ଶ ‫ܮ‬ଵ Resistor Rb 1 k
Resistor Rc 1 k
Variable resistor Rt 0 – 250 k
C. Sensing and conditioning circuit Capacitor C 1 nF
For the closed-loop design the sensing of the load

366
D. Control algorithm composed by the DC source, the H-bridge transmitter and
The DC current I2,dc is proportional to the receiver coil receiver, the inductive coupling, the measurement and
current ଓҧଶ . Replacing (1) in (3) the receiver current visualization blocks, the conditioning circuit and the
results: closed loop control.

Ͷ
ܷଵǡௗ௖ •‹ ߚ
ȁଓҧଶ ȁ ൌ ߨ (12)
߱଴ ‫ܯ‬

As mentioned before, the Phase Shift ȕ can be used as

a control variable to regulate the load current against the
variations in supply voltage U1,dc or in mutual inductance Fig. 6. Schematic of the control algorithm implemented in the
M, if the inverter is controlled with a fixed frequency f0. FPGA.
In order to implement the control algorithm a Xilinx
Spartan®-3A/3AN FPGA Starter Kit board was used.
Experimental tests concerning the notable results
achievable through a FPGA-based control of power
converters have been shown in [22]. Some key features of
the board are:
• Xilinx 700K-gate XC3S700A FPGA;
• 4Mbit Xilinx Platform Flash configuration
PROM;
Fig. 7. Simulink schematic of the UIPT system.
• 50 MHz clock oscillator;
• High-speed differential or single ended I/O The first set of measurements has been used to verify
connectors; to design of the H-Bridges and the coils. The test bench is
• Two-input, SPI-based Analog-to-Digital composed by:
Converter (ADC) with programmable-gain pre- • DC Power Supply 0-30 V, 0-10 A, 300 W;
amplifier; • Transmitter side H-Bridge;
• Rotary-encoder, four slide switches and four • Coupling coils;
push buttons. • Receiver side H-Bridge;
The software used to implement the control algorithm • Electronic load 0-60 V, 0-5 A, 300 W;
was the ISE Design Suit 14.7. A schematic of the control • Oscilloscope Teledyne LeCroy 6Zi.
algorithm as implemented in the board is shown in Fig. 6. The system is analyzed under square wave operation.
The ADC SPI block manages the analog capture circuit. The transmitter and receiver AC voltages and the currents
It is composed by a LTC6912-1 programmable pre- waveforms have been acquired and compared to the
amplifier that scales the input analog signal and by a simulated one, as reported in Figg. 8. Fig. 8 (a) shows the
LTC1407-1 ADC. For this application, the pre-amplifier inverter output voltage and current, and it can be noticed
has a gain of -1 so the input voltage is not scaled. The that the simulated and experimental waveform overlap
reference voltage of the amplifier was 1.65 V and the almost perfectly. The inverter output current is almost
maximum range of the ADC is ±1.25 V centered around sinusoidal due to the filtering effect of the compensation
the reference voltage. The ADC output is a 14-bit, two’s network and coil impedance. Fig. 8 (b) shows the rectifier
complement digital signal. All of this information are input voltage and current and a good agreement between
used to properly design the signal conditioning system as simulation and experimental results can be observed. Fig.
described in the next section. The conditioned signal is 8 (c) shows the power at different stages of the system:
acquired and converter by the ADC. The acquired current input DC power P1,DC, inverter output power P1, rectifier
is then converted from SIGNED INTEGER to input power P2, output DC power P2,DC. In general, very
FLOATING POINT and it is compared to the reference similar results are achieved, apart from the output DC
current. The reference current can be modified using the power P2,DC. This difference is probably caused by not
rotary knob of the Spartan®-3A/3AN FPGA. The having considered some losses in the rectifier. The
resulting error is sent to a PI regulator in order to obtain efficiency of the simulated system, calculated as the ratio
the Phase Shift ȕ, which is used to generates the gate of the output DC power to the input DC power, is 85 %,
command for the switches of the H-Bridge. while the efficiency of the real system is 79%. This set of
measurements has been done in open loop, with a 24 V
IV. SIMULATION AND EXPERIMENTAL RESULTS
electronic load implementing the battery to be charged.
For the verification of the proposed system, several The second set of measurements has been used to
simulations in MatLab/Simulink environment and verify the correct operation of the closed loop control.
experimental tests have been carried out. Fig.7 shows the Fig. 9 shows the response of the system to a step change
Simulink schematic of the UIPT system. The schematic is of the reference current both in simulation (blue) and in

367
the real system (orange). It can be noticed that the
responses are slightly different, due to the parasitic
elements (inductance, capacitance) of the PCB that
cannot be considered in the simulation model. The step
response performance parameters are estimated and
compared for both cases in Table II. Figg. 10 show some
other responses to different perturbations to the system.
The yellow line is the output DC voltage while the
magenta line is the output DC current. The step response
performance parameters are similar to those highlighted
in Table II. Future developments of the work include a bi-
directional operation, thus using a storage system as
active load, such as an hybrid storage system based on the (a)
integration of batteries and supercapacitors [23-24].
TABLE II
STEP RESPONSE PERFORMANCE
PARAMETER SPIMULATION
ARAMETERS EXPERIMENTAL
Rise time (ms) 0.0521 0.2140
Settling time (ms) 5.2117 5.4185
Overshot (%) 8.9073 15.5313
Undershot (%) 0.1370 0.1215
Peak (A) 1.0891 1.1438
Peak time (ms) 1.9504 2.2511

V. CONCLUSIONS
(b)
In this paper, a 100W prototype of wireless battery
charger for E-bikes, based on Inductive Power Transfer
(IPT), has been proposed and described. Simulation and
experimental results have been thoroughly explained as
well. The control law has been successfully implemented
through FPGA. Future developments of this work include
the bi-directional operation of the system and the
integration of different types of storage elements, such as
batteries and supercapacitors.

ACKNOWLEDGMENT
This work was realized with the contribution of PON
R&I 2015-2020 “PROSIB”, CUP no:B66C18000290005, (c)
PRIN 2017 “Advanced power-trains and -systems for full Fig. 8. Experimental and simulated: (a) primary inverter output
electric aircrafts “, prot. no.: 2017MS9F49, SDES voltage and current; (b) rectifier input voltage and current; (c) power at
different stages of the system.
Laboratory – UNINETLAB, Laboratory of Electrical
APplications – LEAP - of University of Palermo.

REFERENCES
[1] V. Boscaino, F. Pellitteri, G. Capponi, R. La Rosa, “A
wireless battery charger architecture for consumer
electronics”, International Conference on Consumer
Electronics, ICCE 2012, Berlin, Germany, September 3-5,
pp. 84-88; 2012.
[2] H. Z. Z. Beh, G. A. Covic, and J. T. Boys, “Investigation
of Magnetic Couplers in Bicycle Kickstands for Wireless
Charging of Electric Bicycles,” Emerging and Selected
Topics in Power Electronics, IEEE Journal of, vol. 3, no. 1,
pp. 87–100, Mar. 2015.

Fig. 9. Simulation vs. experimental step response.

368
 [11] F. Pellitteri, V. Boscaino, A. O. Di Tommaso, R. Miceli,
G. Capponi, “Wireless Battery Charging: E-bike
Application”, International Conference on Renewable
Energy Research and Applications, ICRERA 2013,
Madrid, Spain, October 20-23, pp. 1-5; 2013.
[12] F. Pellitteri, V. Boscaino, A. O. Di Tommaso, R. Miceli,
G. Capponi, “Inductive Power Transfer for 100W Battery
Charging”, Industrial Electronic Society Conference,
IECON 2013, Vienna, Austria, November 10-13, pp. 1-6;
2013.
[13] MOSFET technical datasheet – [Online]. Available:
https://www.infineon.com/dgdl/irfb4410zpbf.pdf?fileId=55
(a)
46d462533600a40153561644141e3a
[14] MOSFET Driver technical datasheet – [Online]. Available:
https://www.renesas.com/eu/en/www/doc/datasheet/hip408
1a.pdf
[15] High voltage Schottky barrier rectifier - [Online].
Available:
https://www.diodes.com/assets/Datasheets/ds30116.pdf
[16] “AN486: High-Side Bootstrap Design Using ISODrivers in
Power Delivery Systems” – Application Note, [Online].
Available:
https://www.silabs.com/documents/public/application-
notes/AN486.pdf
[17] M. Caruso, V. Castiglia, A. O. Di Tommaso, R. Miceli, F.
(b) Pellitteri, L. Schirone, “Efficient contactless power transfer
Fig. 10. Step response to different perturbations: (a) reference system for EVs”, Universities Power Engineering
current step variation from 1 A to 1.5 A; (b) battery voltage step change Conference, UPEC 2016, Coimbra, Portugal, September 6-
from 12 V to 11.5 V. 9, pp. 1-6; 2017.
[18] F. Pellitteri, V. Boscaino, A. O. Di Tommaso, R. Miceli,
[3] H. Z. Z. Beh, G. A. Covic, and J. T. Boys, “Magnetic G. Capponi, “Experimental test on a Contactless Power
couplers in kickstands for wireless charging of electric Transfer system”, International Conference on Ecological
bicycles,” in IEEE Applied Power Electronics Conference Vehicles and Renewable Energies, EVER 2014, Monte-
and Exposition (APEC), 2014, pp. 1348-1355. Carlo, March 25-27, pp. 1-6; 2014.
[4] J.-I. Itoh, K. Noguchi, and K. Orikawa, “System design of [19] P. Livreri, V. Di Dio, R. Miceli, F. Pellitteri, G. Ricco
electric assisted bicycle using EDLCs and wireless Galluzzo, F. Viola, “Wireless battery charging for electric
charger,” in Power Electronics Conference (IPEC- bicycles”, International Conference on Clean Electrical
Hiroshima 2014 - ECCE-ASIA), 2014 International, 2014, Power, ICCEP 2017, Santa Margherita Ligure, Italy, June
pp. 2277– 2284. 27-29; 2017.
[5] RRC. (2014) RRC wireless power. [Online]. Available: [20] F. Pellitteri, G. Ala, M. Caruso, S. Ganci, R. Miceli,
http://www.rrcwirelesspower.com/fileadmin/website_wirel “Physiological compatibility of wireless chargers for
ess_power/Dokumente/News/Pressemittei electric bicycles”, International Conference on Renewable
lung_RRC_wireless_emobility_EN_021110.pdf. Energy Research and Applications, ICRERA 2015,
[6] SEW EURODRIVE (2013) Taking mobility to the next Palermo, Italy, November 22-25, pp. 1-6; 2015.
level. [Online]. Available: http://www.sew- [21] F. Pellitteri, M. Caruso, V. Castiglia, R. Miceli, C. Spataro,
eurodrive.be/marketing/17058414.pdf F. Viola, “Experimental Investigation on Magnetic Field
[7] F. Pellitteri, V. Boscaino, A. O. Di Tommaso, R. Miceli, Effects of IPT for Electric Bikes”, Electric Power
“Efficiency optimization in bi-directional inductive power Components and Systems,vol. 46, 2018, issue 2, pp. 125-
transfer systems”, International Conference on Electrical 34.
Systems for Aircraft, Railway, Ship Propulsion and Road [22] G. Ala, M. Caruso, R. Miceli, F. Pellitteri, G. Schettino, M.
Vehicles, ESARS 2015, Aachen, Germany, March 3-5, pp. Trapanese, F. Viola, “Experimental Investigation on the
1-6; 2015. Performances of a Multilevel Inverter Using FPGA-based
[8] F. Pellitteri, V. Boscaino, R. Miceli, U. K. Madawala, Control System”, Energies, vol. 12, 2019, issue 6, pp. 1-
“Power tracking with maximum efficiency for Wireless 17.
Charging of E-bikes”, International Electric Vehicle [23] F. Pellitteri, V. Castiglia, P. Livreri, R. Miceli, “Analysis
Conference, IEVC 2014, Florence, Italy, December 17-19, and design of bi-directional DC-DC converters for
pp. 1-7; 2015. ultracapacitors management in EVs”, International
[9] F. Pellitteri, A. O. Di Tommaso, R. Miceli, “Investigation Conference on Ecological Vehicles and Renewable
of inductive coupling solutions for E-bike wireless Energies, EVER 2018, Monte-Carlo, pp. 1-6; 2018.
charging”, Universities Power Engineering Conference, [24] P. Livreri, V. Castiglia, F. Pellitteri, R. Miceli, “Design of
UPEC 2015, Stoke-on-Trent, UK, September 1-4, pp. 1-6; a Battery/Ultracapacitor Energy Storage System for
2015. Electric Vehicle Applications”, IEEE 4th International
[10] F. Pellitteri, V. Boscaino, A. O. Di Tommaso, F. Genduso, Forum on Research and Technologies for Society and
R. Miceli, “E-bike battery charging: Methods and circuits”, Industry, RTSI 2018, Palermo, Italy, September 10.13, pp.
International Conference on Clean Electrical Power, 1-5; 2018.
ICCEP 2013, Alghero, Italy, June 11-13, pp. 1-8; 2013.

369
 Design Validation and Performance Evaluation of a
BLDC of Commercial Electric Bike and Its Performance
Comparision with Different Probable Designs
1 Aditya Raj, 2 Sourabh Paitandi, and 3 Mainak Sengupta
Dept. of Electrical Engineering
Indian Institute of Engineering Science and Technology, Shibpur, Howrah, W.B., India
Email: 1 [email protected], 2 [email protected], 3 [email protected]

Abstract—This paper presents the design validation and perfor-

mance evaluation of a commercially available 8 pole, 12 slot, 3 phase,
1KW Interior Permanent Magnet (IPM) BLDC machine, used in
popular commercial e-bikes. This machine has a fractional slot non-
overlapping (concentrated) winding. The slots per pole per phase
(SPP) is 0.5. The intricate design details of the machine have been
extracted otherwise and validated analytically. With the extracted
dimensional details (not available in datasheets), the machine has
also been simulated in 2D-FEM and the parameters and performance
indices have been duly calculated. Different performance indices such
as THD in induced emf, iron and magnet losses, cogging torque, etc.
have also been investigated in detail. The performance of the above
mentioned machine has been compared against designs with other (b) BLDC motor detached from
(a) BLDC motor fitted with its
possible slot-pole combinations with SPP=0.25 and 0.375, but of same ebike setup with an arrangement
controller and ebike pedal
nominal ratings. Appropriate inferences and conclusions are thereby of free run
drawn.
Index Terms—Brush-less, IPM, Non-overlapping concentrated wind-
Fig. 1: The BLDC motor investigated
ing, Finite element method, d and q-axis inductance

presence of lower order harmonics in the air gap flux distribution

I. I NTRODUCTION
is more, causing higher losses in the rotor magnets and back-iron
The permanent magnet electrical machines are gaining attention [12]. Hence, the PM machines have to be designed with minimum
due to their reduced losses and compact size that gives them a iron losses. For this, the best choices for number of poles are four,
decisive edge for use in automotive applications. For a given power six and eight [1], particularly when the rated speed is higher. Now,
output, the PM machines have higher energy efficiency, lesser BLDC machines with such concentrated winding have a range of
volume, higher power factor and easier controllability, than other SPP from 0.25 to 0.5.
rotating machines [1] [2]. Research on these machines also aims to This work was triggered when the authors were approached by
develop them to be produced at an affordable price. This can help a US-based popular manufacturer of electric bicycles. The IPM
to reduce the carbon footprints of the country on a large scale. BLDC machine investigated here in this paper is the one that
Brushless DC machine is such a machine and is well known in these is commercially used in those electric bicycles, and is shown in
applications due to its high efficiency, high reliability, long lifetime, Fig.1(b).
simple design, low noise and good torque-speed characteristics The BLDC studied here has non-overlapping concentrated wind-
[3] [4]. With the technological boon in power electronics, control ing with an SPP= 0.5 (3 phase, 8 pole,12 slot). The performance and
and automation, speed regulation techniques of such machines parameters of such concentrated winding machines majorly depends
are being improved for better performance and lesser cost [3] on the slot-pole combinations of the machine [14]. Hence this has
[6] [7] [8]. Non-overlapping windings are now being used for to be selected appropriately. Since the SPP is less than unity, the
these machines, instead of the conventional distributed double layer space available for slots is sufficient [9]. This paper deals with the
windings chiefly, due to cost savings and simpler manufacturing detailed investigation and parameter estimation of the 3 phase, 8
process [5] [10] [11]. These have fractional slots per pole per pole, 12 slot BLDC machine. Two other BLDC designs with same
phase (SPP), which lowers the alignment between the stator teeth nominal ratings and main dimensions with different slot numbers
and the magnet poles, and hence lowers the cogging torque too have been studied to investigate the effect ofslot-pole combinations
[12]. Additionally, due to shorter end-windings, these windings on performance of such machines.
are compact [5], and have lower stator copper losses [12] and
lower end-winding copper consumption [5]. This also allows for II. D IMENSIONAL DETAILS OF THE BLDC MACHINE
a lesser active length [13]. These machines also consume a lower As already mentioned, the machine was provided by a US-based
lamination material due to lower stator yoke depth. However, the seller of electric bicycles and the motor did not have any available
978-1-7281-4428-3/19/$31.00 © 2019 IEEE datasheet or design sheet. Also it was so compact that nothing
 A. Winding layout
The theoretical winding layout for an 8 pole 12 slot star con-
nected concentrated winding machine is determined using the Cros’
method [16]. For sake of continuity, it has been repeated here for
the chosen 8 pole 12 slot BLDC:-
1) The number of slots per pole per phase, i.e., SPP, can be
written as :
12 1 N
SP P = = = (1)
3∗8 2 D
2) A sequence of zeros and ones is determined from (1) [16].
(a) Image of the stator of disman- (b) Stator stamping of the BLDC
tled machine with one turn of a
For this combination, N = 1 one and D-N = 1 zero. The first
machine
phase dewound sequence is 10.
3) According to Cros’ [16], the highest winding factor can be
Fig. 2: Images of the BLDC machine under rewinding achieved when the sequence is as regular as possible. The
most regular combination for this case also happens to be
10.
could be measured. Hence it had to be completely dismantled 4) This regular sequence is repeated for, NNs = 12
1 = 12 times.
and rewound (to get the dimensions, winding details and even the Hence, the final sequence is:
SWG of the conductors) at the works of a small local manufacturer
10|10|10|....|10
of electric machines. A photo of the dismantled machine stator is | {z }
shown in Fig.2(a), and one stator stamping has been highlighted 12 times
in Fig.2(b). The complete specifications of the BLDC machine, 5) Above sequence is compared to the layout of the distributed
obtained after its re-winding, are listed in Table-I . Many of these winding with SPP = 1.
values had to be tediously measured from the dismantled machine. 10 10 10 10 10 10 ...
AC 0 BA0 CB 0 AC 0 BA0 CB 0 ...

TABLE I: General details of the BLDC machine where A’, B’, C’ are the return conductors of A, B, C
Parameter name Value
6) The start-phases of each-slot are follows:
Rated shaft output power (in KW) 1
Rated armature current (in A) 30 P hase : A B C A B C
Rated frequency (in Hz) 200
Rated speed (in rpm) 3000
Slot : 1 2 3 4 5 6
Rated torque (in Nm) 3.183
Inbuilt controller rating 48V DC-link, 30A All the steps mentioned above are concisely presented in Fig. 4,
Type of machine Inner rotor IPM BLDC where Sx (x = A, B, C) are the slot vectors involved with the
Number of pole pairs 4
Number of slots 12 corresponding phase and slot number for a double layer winding.
Slots per pole per phase (SPP) 0.5 The resulting winding pattern, as shown in Fig.3, completely agreed
Winding type Fractional slot whole-coiled concen-
trated winding
with that observed upon re-winding.
Coils per phase 4
Turns per coil 9
Conductor specifications 10 strands of SWG 25 wires (Bare B. Analytical calculation of induced emf
diameter=0.5mm and insulation
thickness=35µm)
Stator outer diameter (in mm) 97.8
The winding configuration, as validated, and the other dimen-
Stator inner diameter (in mm) 60 sions, as discussed earlier, have been used to simulate the machine
Stator stacking factor 0.95 in 2D-FEM. The flux density plots are obtained at a particular rotor
Rotor stacking factor 0.95
Rotor outer diameter (in mm) 58.18 position (as shown in Fig.-5 and Fig.-6). This is used for theoretical
Rotor inner diameter (in mm) 12.8 calculation of the fundamental component of the line induced emf
Stator core material M1924G
Rotor core material M1924G using the Faraday’s law, following,
Skew width (in no. of slots) 0
dψp
Bearing type 628RS ean = −
Stator active length (in mm) 48.1 dt
Rotor active length (in mm) 47.14
ψp = Tp ∗ φ = Tp ∗ Bm ∗ At ∗ sin(ω ∗ t) (2)
√
or, eab = 3 ∗ Tp ∗ Bm ∗ Wt ∗ L ∗ ω ∗ sin(ω ∗ t − π3 )
The 2D-FEM simulation of the machine has been done using
these details. The associated parameters like slot fill factor and pole where, ψp , flux linkage per phase; eab , line induced emf; Tp ,
embrace have been calculated for simulation purposes, and come turns/phase; Bm , maximum magnetic flux density through phase
out to be 36.54% and 0.81 respectively. turn (i.e., through a stator teeth); Wt , tooth or phase turn width;
L , stator active length; ω , electrical angular frequency.
III. D ESIGN VALIDATION The value obtained using (2) has been compared with the emf plots
directly obtained from 2D-FEM (shown in Fig.7), and, with the emf
The following features of the motor were practically observed waveform obtained experimentally from the BLDC machine (shown
and are theoretically validated: in Fig.8). The comparisions are listed in Table- II.
 Fig. 3: Winding layout of 8 pole 12 slot 3 phase (SPP q =0.5) IPM - BLDC motor

Fig. 6: 2D-FEM Analysis: Air gap flux density map for the BLDC
machine at 7.5omech

Fig. 4: Steps for cros’ method to determine winding layout [16]

Fig. 7: 2D FEM Analysis Line and phase induced emf

(5) on simplification, leads to (6)

Fig. 5: 2D-FEM Analysis: Flux density map for the BLDC machine
at 7.5omech eˆab = 2.1059 Eptm (6)
36.3
From the calculated value shown in Table II and (6), Eptm = 2.1059
C. Relation between peak values of trapezoidal emf and its funda- = 17.237 V
mental component
The fundamental peak is given by (3) D. Conductor dimensions
√ Z 2π
3 From (7) [17], the rated DC-link current= 29 A and rated phase
eˆab = ept sinθdθ (3) current = 23.7 A. Rotor losses have been neglected here. Assuming
π 0
a current density of 12 A/mm2 , the bare diameter of each conductor
From the odd symmetry,(3) reduces to (4) strand=0.5 mm (SWG-25). The values matched under tolerable
√ Z
2 3 π limits from practical observations during re-winding.
eˆab = |ept |sinθdθ (4)
π 0 Pair gap ≈ Pshaf t = 2 ∗ Eptm ∗ Idc
r
Putting the expression for ept , we get (5) 2 (7)
√ Z Iph = × Idc
π/6 6 ∗ E
3
Z 5π/6
2 3 ptm ∗ θ
eˆ
ab = ∗ sinθdθ + Eptm ∗ sinθdθ
π 0 π π/6
Z π (5) where Eptm - peak value of phase induced emf
6 ∗ Eptm ∗ (π − θ)

+ ∗ sinθdθ
5π/6 π
Fig. 8: Line induced emf from manual run as generator at 357 rpm,
(rms = 3.38V), (Avg. = 0.006137V)

TABLE II: Comparision of line induced emf(s) from FEM and (a) Ld contour (b) Lq contour
practical Fig. 10: 2D FEM simulation: Torque-speed map with inductance
Parameter From FEM From From manual contours
simulation theoretical run
calculation
Speed of run (in rpm) 3000 3000 357
Electrical frequency (in 200 200 23.8 C. d and q axis inductances (Ld , Lq )
Hz)
RMS value (in V) 24.73 25.67 3.73 The Ld and Lq of the machine at different operating conditions
Peak value (in V) 38 36.30 5.61 of torque and speed have been obtained from its 2D-FEM simu-
RMS/electrical frequency 0.124 0.128 0.157
(in Vs) lation and are shown in Fig.10(a) and Fig.10(b) respectively. The
inductances are seen to remain constant upto the rated speed for a
given torque output, and they tend to decrease at speeds higher than
IV. PARAMETER EXTRACTION rated (in the flux weakening zone). Hence the currents will follow
a reverse pattern, i.e., constant at a low value upto rated speed
The important parameters such as needed for the mathematical and reduce thereafter. At a given speed, the inductances decrease
modelling have been calculated theoretically and validated through and hence allow the currents to increase and hence to increase the
2D FEM simulation and practical experiment. torque output.
The values of the inductances at rated speed and torque have
A. Phase Resistance been analytically validated as follows:-
The self and mutual inductances of a synchronous machine can be
The phase resistance is theoretically calculated using (8). written as [18],
Lmts (2L + πτs )
R = ρCu × = ρCu × (8) 
Laa
 
1 1 cos2θr
 
Ll
A A  Lab  = 0 − 21 ◦ 
cos2(θr − 120 ) L1  (9)
◦
Lac 0 − 21 cos2(θr + 120 ) L2
where the symbols have their conventional meanings. The value
obtained has been matched against the values obtained from the
The d and q axis inductances are given by [18],
FEM simulation in Table- III.
 
   3 3
 Ll
TABLE III: Comparision of values of resistances (in Ω) obtained Ld 1 2 2 L1 
= 3 (10)
Lq 1 − 23
through various means 2 L2

From theoretical calcula- From 2D FEM simula- From practical determi-

tion tion nation From (10), L2 is negative for this machine because here Lq > Ld .
0.05 0.05 0.047 Hence, the following equations are derived (using Fig.9),
Laamax 1 1 −1 248.3855
     
 Laamin  1 1 1    174.3666 
 L
 Labmax  0 − 12 −1  l   −65.6856 
    
 = 0 L1 =  (11)
B. Inductances at no-load
L
 abmin   − 12 1 
 L −139.9572

2
L
acmax
 0 − 12 −1  −65.6673 
Lacmin 0 − 12 1 −139.9564
The profile of phase self inductance and inter-phase mutual
inductance under no-load, as a function of the position of the salient The units on RHS of the equations are in µH.
rotor is obtained using 2D-FEM simulation and shown in Fig.-9. Hence, L1 = 205.6428 µH, Ll = 5.73325 µH and L2 = -37.1358
µH. The values hence obtained for Ld and Lq have been compared
with those obtained from Fig.10(a) and Fig.10(b) respectively, in
Table IV
TABLE IV: Comparision of d and q axis inductances at rated speed
and torque for the IPM BLDC machine
Parameter Approx. value from Value from analyti-
Fig. 9: 2D FEM Analysis: Inductance waveforms at no load for the contour map (µH) cal calculation(µH)
Ld 251 258.5
BLDC Machine Lq 350 369.9
 V. P ERFORMANCE EVALUATION TABLE VI: Comparision of parameters in different slot-pole con-
figurations
The efficiency of the machine has been investigated at differ-
ent operating conditions of torque and speed using the 2D-FEM Parameters 12 slot/8pole 9 slot/8pole 6 slot/8pole
Turns per phase coil 9 11 17
simulation and is presented in Fig.11. At a given torque output, SPP 0.5 0.375 0.25
the efficiency is seen to increase upto rated speed and thereafter Winding factor 0.866 0.945 0.866
THD in induced emf per phase (%) 6.74 11.47 5.12
decrease in the flux weakening zone. A similar pattern is seen CT [15] 4 1 2
for efficiency at a given speed with varying torque output. The Torque ripple factor 0.181 0.058 0.224
Torque density (Nm/kg) 1.813 1.845 1.812
efficiency is seen to be maximum in a zone where speed is near Core loss at full load (W) 24.6 24.5 27.5
the rated and output torque is lesser than rated. Magnet loss (W) 0.9 5.35 20.54
Ld (µH) 223.8 253.2 520.7
Lq (µH) 334.2 361.9 615

for each of the design variants. The 9 slot design (SPP=0.375) is

seen to be rich in third harmonics and has the worst THD among all
the chosen designs, though it has lower losses and lower cogging
torque, that adds to lowest torque ripple among the considered
designs. The other two designs have lower triplen harmonics, and
have a better THD. The 6-slot design (SPP =0.25) has the best
THD but suffers from greater losses and cogging leading to more
torque ripple.
Fig. 11: 2D-FEM simulation: Efficiency map for the IPM BLDC The cogging torque for the three design variants have been
machine obtained from their 2D-FEM simulation and are compared in
Fig.15. The 8 pole 9 slot variant has the lowest cogging torque
The efficiency is analytically validated for the rated condition. (40 mNm), and the 8 pole 6 slot has the maximum value (520
Analytical validation is done using the loss data at rated speed mNm). The relative estimator for cogging torque as mentioned in
obtained from the 2D-FEM simulation, tabulated in Table V. Rated [15] is seen to hold good for 9 slot design, as it has the lowest CT
1000
efficiency = 1160.44 = 86.17% (From Fig.11, it is ≈ 87%), which among all designs. But for the other two design variants, the one
is very high for such a small motor. with a higher CT is seen to have slightly lower cogging torque.
This needs an in-depth validation.
TABLE V: Components of core losses of the BLDC machine
Loss component Average value Average value Total average
for stator (in for rotor (in value (in W) VII. C ONCLUSIONS
W) W)
Hysterisis loss 4.884 0.0242 4.9082
Eddy current loss 9.5218 0.1315 9.6533 The winding layout of the machine as proposed using Cros’
Excess loss 0.769 0.016 0.785
Total core loss 15.2378 0.1739 15.4117 method matched with the practical observation of the armature upon
Magnet loss N.A. 0.034 0.034 its de-winding. The calculated back emf constant and conductor
Copper loss (at rated current) 135 N.A. 135
Friction and windage loss (as- N.A. N.A. 10
dimensions were close to the practical observations under tolerable
sumed 1% of rated output) limits. Ld and Lq calculated using self and mutual inductance values
Total losses 150.2378 0.2079 160.4457
from FEM have a very good match with those directly given by
FEM maps. The motor loading tests are in progress to validate the
The practical estimation of efficiency is under progress and the proposed performance indices obtained from FEA. The mentioned
rewound motor is coupled with a brake-pulley loading arrangement design validations have been done upon specific approach by the
and an optical position sensor to run it in 120o conduction mode sellers of the commercial ebikes.
using the fabricated logic circuit. The setup operation with wave- On comparing the said machine with some other possible design
forms of line voltage (VRY ) and current (IR ) is shown in Fig.12. It variants with nominal ratings, it has satisfactory performance in-
must be mentioned that the integrated motor plus power converter dices, intermediate of the other two designs and hence appears to
and controller were all otherwise in a working condition in a nicely be a better design to be implemented in a hardware prototype for
compact but inaccessible contraption connected with pedals. the mentioned ratings.
VI. C OMPARISION OF THE BLDC MACHINE WITH OTHER
SLOT- POLE CONFIGURATIONS VIII. ACKNOWLEDGEMENTS
The performance indices of BLDC designs having different SPPs
(same no. of poles but different slot count) but similar voltage and The authors extend their special thanks to D. Snyder of Califor-
torque ratings have been compared and tabulated in Table VI. nia, USA, the staff of M/S GE motors Pvt. Ltd, Sheorapuli, and
The induced emf per phase at rated speed for the three design Mr. S K Mainuddin for their notable help in the entire process. The
variants have been obtained from their 2D-FEM simulation and are support received from the research colleagues, staff, particularly
compared in Fig.13. The rms values of different odd harmonic com- Mr. Netai Dutta, at the Advanced Power Electronics Lab, and the
ponents in the induced emf per phase have been shown in Fig.14 authorities of IIEST, Shibpur is also gratefully acknowledged.
 Fig. 12: Practical setup for efficiency estimation

Fig. 13: 2D-FEM simulation: Induced emf at rated speed

Fig. 15: 2D-FEM simulation: Cogging torques for the three BLDC
designs

trical, Electronics, and Optimization Techniques (ICEEOT), March 2016,

DOI:10.1109/ICEEOT.2016.7754942
[9] Florence Meier, ”Permanent-Magnet Synchronous Machines with Non-
Overlapping Concentrated Windings for Low-Speed Direct-Drive Applica-
tions”
[10] T. D. Strous, H. Polinder, Member, IEEE, J. A. Ferreira, Fellow,
IEEE,”Inductance calculations for PM machines with concentrated windings”,
Fig. 14: Odd harmonic spectrum of induced emf(s) IEEE International Electric Machines & Drives Conference (IEMDC), May,
2011, DOI:10.1109/IEMDC.2011.5994636
[11] Freddy Magnussen, Member, IEEE, and Heinz Lendenmann, Member, IEEE,
”Parasitic Effects in PM Machines With Concentrated Windings”, IEEE
R EFERENCES TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 43, NO. 5, SEPTEM-
[1] Chunting Chris Mi, Senior Member, IEEE, Gordon R. Slemon, Life Fellow, BER/OCTOBER 2007, DOI:10.1109/TIA.2007.904400
IEEE, and Richard Bonert, Member, IEEE, ”Minimization of Iron Losses of [12] Chun Tang, Wen L. Soong, Gene S. Liew and Nesimi Ertugrul, ”Ef-
Permanent Magnet Synchronous Machines.”, IEEE Transactions on Energy fect of Pole and Slot Number Changes on the Performance of a Surface
Conversion, vol. 20 , issue. 1 , Mar, 2005, DOI:10.1109/TEC.2004.832091 PM Machine”, International Conference on Electrical Machines, Sep, 2012,
[2] Anqing He, Chenxi Zhou, et.al. ”Evaluation of Fractional Slot Concentrated DOI:10.1109/ICElMach.2012.6349867
Winding Permanent Magnet Synchronous Machine for Electric Vehicle Appli- [13] Design and Optimization of Interior PM Machines with Distributed and
cation”, IEEE International Electric Machines & Drives Conference (IEMDC), Fractional-Slot Concentrated- Windings for Hybrid Electric Vehicles, IEEE
May, 2019, DOI: 10.1109/IEMDC.2019.8785223 Conference and Expo Transportation Electrification Asia-Pacific (ITEC Asia-
[3] Chang-liang Xia, Tianjin University, P.R. China, ”PERMANENT Pacific), Sep, 2014, DOI:10.1109/ITEC-AP.2014.6941169
MAGNET BRUSHLESS DC MOTOR DRIVES AND CONTROLS”, [14] F. Libert, J. Soulard, ”Investigation on Pole-Slot Combinations for Permanent-
ISBN:9781118188330 Magnet Machines with Concentrated Windings”, Jan, 2004
[4] Y.B. Adyapaka Apatya, Aries Subiantoro and Feri Yusivar, ”Design and [15] Z.Q. Zhu and D. Howe, “Influence of Design Parameters on Cogging Torque
Prototyping of 3-Phase BLDC Motor”, International Conference on Quality in Permanent Magnet Machines,” IEEE Trans. Energy Conv., vol. 15, no. 4,
in Research (QiR) : International Symposium on Electrical and Computer pp. 407- 412, Dec. 2000., DOI:10.1109/60.900501
Engineering, July, 2017, DOI:10.1109/QIR.2017.8168483 [16] J. Cros and P. Viarouge, ”Synthesis of high performance pm motors with
[5] Dieter Gerling,”Electrical Machines Mathematical Fundamentals of Machine concentrated windings”, IEEE Transactions on Energy Conversion, vol. 17,
Topologies”,pp-449, ISBN:9783662520321 no. 2, pp. 248-253, Jun. 2002
[6] Józef Gromba, ”Torque Control of BLDC Motor for Electric Bicy- [17] Bimal K. Bose, ”Modern Power Electronics and AC Drives”, ISBN:
cle”, International Symposium on Electrical Machines (SME), June 2018, 9780130167439
DOI:10.1109/ISEM.2018.8442987 [18] D. O’Kelly and S. Simmons., Introduction to generalized electrical machine
[7] Vladimir Dimitrov, ”Overview of the Ways to Design an Electric Bicycle”, theory, 3d ed. McGraw-Hill European electrical engineering series, 1968,
National Conference with International Participation (ELECTRONICA), May, ISBN:9780070940505
2018, DOI:10.1109/ELECTRONICA.2018.8439456 [19] Aditya Raj, B.Tech theses, ”Investigations on Permanent Magnet Machines for
[8] K. Latha Shenoy, M. Satyendra Kumar, ”Design Topology and Elec- Electric Vehicle Applications”
tromagnetic Field Analysis of Permanent Magnet Brushless DC Mo-
tor for Electric Scooter Application”, International Conference on Elec-
 EVS27
Barcelona, Spain, November 17-20, 2013

An E.Bike Design for the Fourth Generation Bike-Sharing

Services
Ricardo Meireles1, Jose Silva1, Alexandre Teixeira1, Bernardo Ribeiro1
1
CEIIA – Centro para a Excelência e Inovação na Indústria Automóvel
Rua Eng. Frederico Ulrich, 2650 (Tecmaia), 4470-605 Maia, Portugal

Abstract
Bicycle sharing systems have proven their value towards urban sustainable mobility. Appropriate design of
bikes for this application is fundamental for bike-sharing systems viability. A new e.Bike was designed for
the specific application of bike-sharing. This bike model has a significant number of new features that
improve the user experience through an adequate position of motor, as well as the addition of new
functions such as the locking of the bike and rider safety. At the same time the reduction of number of parts
and the use of new materials, contribute to the viability of this e.Bike model and the associated bike-
sharing system.

Keywords: Bicycle, Fleet, Intelligent, Mobility, Safety

frames and components, together with the

1 Introduction introduction of solid rubber tires. Also
implemented in the 2nd generation were specific
picked up and return locations.
Bicycle sharing systems have proven their value The 3rd generation added costumer tracking
towards urban sustainable mobility. Although it capability to discourage theft and misuse. The
looks a natural simple concept, implementation implemented systems included smart cards or
of a reliable and sustainable system has proven to mobile phone access, electronic lock, on-board
be difficult and at the same time challenging. A computers and telecommunication systems.
long journey has been done since the first system Bike –sharing future generation (4th) must address
implemented during the 60s in Amsterdam, with the efficiency of the system, both from the
the so called “white bikes” [1]. Since that time operator and costumer perspective. Also important
several innovating issues have been introduced in on future bike-sharing systems is the improvement
bike-sharing systems. These systems have of user friendly features. One of the greatly
evolved through several technological expected features in this category is the
generations that have been leading to the increase implementation of electric bicycles on sharing
of robustness, and recent inclusion of systems, allowing users to ride longer, on hilly
intelligence. The first generation included plain terrains with less effort. From the operator point of
bikes without dropping stations, which do not view, the 4th generation includes an IT
provide enough control, security or fleet (Information Technology) system to improve bike
management capability. distribution on the dropping stations and new
2nd generation brought the implementation of business models.
dedicated bicycle design feat to withstand In order to be used in 4th generation bike-sharing
intensive use and misuse. These include stronger systems, new bikes need to be developed in order

EVS27 International Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 1
to address all issues already included on the first
three generations and the specific requirements 3- Cost efficiency - Design should target a low
that allow the implementation of the 4th cost, low volume production that could be easily
generation, where electric powertrain is included, upgraded for medium volumes if needed. Thus,
as well as integration with smart management material selection should focus on cost-efficient
system, keeping the complete system affordable solutions that require low raw material cost, low
for the user as well as for the operator. cost for production equipment and tooling and
reduced costs for processing.
2 Requirements definition
The development of an e-bike for a bike-sharing 3 E.Bike development
system requires that two main points of concern
have to be considered. First is the need of the The e.Bike development was performed focusing
operator focus on a reliable, simple, efficient, on each requirement and defining technical
sustainable but profitable system. On the other solutions.
hand, are the features that both bicycle riders and
general citizens appreciate. Achieving the
balance between those two points is critical for 3.1 Motor position
the success of the system, mainly when the gold
is to design a bike system to be used in many After the study of three types of possible motor
different “hands”, and therefore needs to be layouts (front wheel, rear wheel and crank shaft),
robust, yet trendy and with enough enchantment the crank shaft motor position was the selected
to create the desire to be used. one. In an e.Bike is very usual to use, a front wheel
A deep analysis was performed in order to or a rear wheel motor, mainly due to the cost
understand the requirements applied to these factor, and simple installation and maintenance. In
bikes. These requirements were divided into 3 fact, this solution is the most appropriate when
different categories: considering bikes for private use.
1- Usability – The bike should include an electric
powertrain for riding assistance, which includes a
motor, a battery pack with enough energy storage
capacity to allow at least two consecutives uses
without running out of energy, for the average
riding times on such systems.
On the intelligent features, automatic movements
should be included in several functions in order
to reduce the human intervention and with this,
also reduce bike components.
On the safety point of view, new features should Figure 1: Crank shaft motor position
be included in order to improve the user safety
when riding the bike, considering specially the On a preliminary evaluation, the crank shaft motor
interaction with the cars and urban heavy position could be considered not the ideal solution
vehicles. due the fact that maintenance would be more
difficult to perform, because more components
2- Design – Considering the use of motor and need to be handled. However, for this same reason,
battery pack, weight must be reduced on other it is a good solution for an e.Bike sharing system,
components and the structure itself, in order to once the motor is considered a valuable component
keep the bike with the same performance level with a high risk to be vandalized or thieved.
and energy consumption. Lean design is a must, Having the motor enclosed inside body panels (as
so that few components are integrated on the explained latter on chapter 3.6) is and additional
bike. Other features need to be added, such as protection for this key component against
safe design to avoid vandalism, theft and misuse. vandalism actions. In the particular design of this
From the aesthetic point of view it should be e.Bike the motor, placed on the crank shaft, is
trendy and with flat surfaces for publicity protected inside robust body panels. But, the crank
exhibition.

EVS27 International Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 2
shaft motor position was also chosen because this 3.3 Locking and charging system
solution allows a good user friendly experience,
since the power is provided directed on the crank An essential system of the bike sharing is the
shaft, giving a comfortable control sensation to coupling on the dock. Electric bike sharing, not
the rider, similar to riding a conventional bike, only needs to have a robust mechanism to firmly
and thus more natural behaviour. The rider can lock the bike to the dock, but due to the fact that
have a better and more natural control on the the battery need to be charged, a system to connect
power transmitted to the wheel with this type of and perform charging simultaneously need to be
motor, also reducing the inertial and mechanical added.
lost associated with chain powered transmissions. Common ways to address this issue on current
The selected motor is compatible with rear hub electric bike sharing is either use detachable
gears, which makes it possible to be used batteries or use a connection cable for electric
combining human power and electric power charging. However, this is not a comfortable action
motor. for a rider using a shared bike. The rider expects to
have a simple drop-and-leave behaviour, when
leaving the e.Bike on the docking station, and this
3.2 Safety laser is more critical if we consider the metropolitan
user that in most of the time is in a rush to leave
Aiming towards the safety of the rider, a laser the bike and pick-up next public transportation.
line is added as an alert presence and bike Thus a simple and user friendly system must be
identification. This is a new trend among the built-in.
private bike market. This laser device with a very In order to address this issue, a system was
small size and low weight, is mounted on the developed that allows with a single user
back end of the rear dropout. This laser device is movement, to lock the bike in the docking station
protected under the body panels, and uses a laser and start charging. The system is composed by a
beam to project two red lines to each side of the coupling mechanism fitted on the bike side and by
bike. Each red line projected on the pavement is the respective nest on docking side. On the bike
parallel to the longitudinal direction, giving to frame the coupling mechanism was installed in the
the other users of the road, particularly bike head tube, for two main reasons - the first
automobile drivers, a sense of the space they related with ergonomics aspects – the rider can
need to allocate to the cyclist. dock the bike in a standing position even without
This allows other drivers to safely overcome the the need to off the bike, having a perfect eye
riders by respecting safety distance with visual control on the fitting mechanism – the second
reference. The laser red colour also helps to related with engineering aspects – the bike head
increase the feeling of alert to other drivers. tube was designed to have a thickness and
This is a low cost device that can save the life of geometry able to resist to vandalism torsion or
cyclist, especially during night use. unauthorized tentative to remove the bike from the
docking, being a robust place to install the locking
triggers. The locking and charging system was
specially designed to be easily used by a non-
trained person, hence both bike and docking
alignments angles were carefully selected in order
to compensate any terrain irregularity and allow
the correct orientation of coupling triggers.
On the bike side, there is a cap that protects the
electric plug from water and dust. This cap only
opens when the bike receives signal, trough the
intelligence box that is perfectly fitted on the dock,
preventing this way any risk of electric shock by
hiding the electric plugs. Only after receiving the
Figure 2: Safety laser line. signal that the bike is perfectly fit and in final
locked position, the system starts to charge the
battery, and create the communication link
between the bike and the rest of the system.

EVS27 International Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 3
This is one of the most critical mechanisms in all
system, and must withstand abusive load for
vandalism scenarios, but most of all, should be a 3.5 Structural Validation
user-friendly and intuitive system.
A bike-sharing system is by definition a bike that
must withstand very demanding use. In pursuit of
the highest safety level and to ensure the best
reliability, structural performance was evaluated
with Computer Aided Engineering (CAE)
simulations and Finite Element Analysis methods,
in order to comply with the safety requirements
and test methods defined in the European Standard
EN 14764 (City and trekking bicycles). A set of
simulations was made, considering steel and
aluminum components. Some examples of the
simulations performed were:
1) Steering assembly – Static strength and security
tests;
Figure 3: Locking and charging mechanism. 2) Frame and fork assembly – Impact test;
3) Frame - Fatigue test with pedaling forces;
4) Frame - Fatigue test with vertical force;
3.4 Bike Stem 5) Fork intended to use with hub- or disc-brakes -
strength test.
Towards anti-vandalism features a dedicated
stem without visible screws was designed. This
solution applies a new assembly approach to the
handle bar and stem. This cover is also used to
protect the electric and braking cables. Looking
forward for multifunction components that could
also be seamless integrated in the e.Bike all
critical fixtures are hidden. It includes an internal
cable path protection, and also integrates the
basket fixture without visible screws.
This new bike stem, due to its shape, also allows
to achieve a more ergonomic upright position
suitable for long and relaxing ride.
Figure 5: CAE simulation

The structural stability was also accessed by means

of buckling analysis and some additional tests
were performed to ensure the assembly integrity in
the event of vandalism acts.

3.6 Intelligence Box

An on-board computer provides real time

monitoring and control over the bicycle and
tracking during its use by combining sensorization,
calculation, and communication ability.
After evaluating current systems tracking ability
and typical vandalism misuse, one of the new
features proposed was the use of a 3 axis
Figure 4: Bike stem without visible screws
accelerometer. This will allow the operator, among

EVS27 International Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 4
other applications, to identify if the bicycle is clipping and screws placed on the bottom part of
being exposed to abusive use and as the the body. Panels are assembled in a specific order
monitoring is done in real time, the bike-sharing that assure the correct protection to dust and water.
system operator may decide to act by sending a
warning to the user or limiting bicycle
performance. On board decisions towards misuse 4 Conclusions
can also be programed in the computer allowing
for example the e.Bike motor to be turn off when
abusive use is detected. There are three key issues that must be addressed
when developing a bike sharing system, they are
the operator, the user and the citizens that lives and
uses the city. Designing a sustainable system
3.7 Body Panels requires two approaches, cost reduction, looking
for cost effective production processes, but also
To achieve stylish design at low volume requires add value design features, and both of
production and low cost, plastic body panels them can represent extra services for the operator
were designed, which also may be used as to charge or saving costs in maintenance.
publicity support during operation. A new e.Bike was designed for the specific
application of bike-sharing. This bike model has a
significant number of new features that improve
the user experience through an adequate position
of motor, as well as the addition of new functions
such as the locking of the bike. During the
development process new solutions were achieved
in terms of bike design that allows an improvement
in terms of components number reduction, system
security and monitoring. The use of new materials
such as DCPD allowed the introduction of body
panels at an adequate cost level giving the bike a
stylish shape and protecting critical bike
components.
Figure 6: Plastic body panels

Producing plastic parts for low volume series is Acknowledgments

always a challenge. DCPD-RIM was the selected
The work described here above was funded
technology for this case. In this technology the
through the EU QCA project.
tooling used is simple and based on a single core-
cavity structure allowing this way to have a cost
efficient production, and to achieve body panels
with good perceived quality and at a sustainable References
level of productions cost. [2]
Bike-share, can benefit from being a publicity [1] DeMaio, Paul, “Bike-Sharing: History, Impacts,
support that is in constant movement inside the Models of Provision, and Future”, Journal of Public
Transportation, Vol. 12, Nº4, pp. 41-56, 2009.
city area, and seen by large number of people. To
profit from this feature, flat body panels need to
[2] Teixeira, Alexandre, Ribeiro, Bernardo, Use of
be available on the bike to add publicity on. In DCPD/RIM on exterior panels for the automotive
this bike-sharing system the body panels were industry, RPD 2010 – Rapid Product Development,
designed with a style language that permits the Marinha Grande, Portugal, 2010.
use of publicity on the seat tube and down tube
panels, as they were designed with flat surfaces.
These panels also serve the purpose of protecting
the critical systems of the e.Bike, such as the
motor, battery and intelligence box.
The maintenance access to critical components,
is done through a non-visible system made by

EVS27 International Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 5
Authors
Ricardo Meireles, has been working
with CEIIA in product development
for 7 years on automotive components
and subsystems in both plastic and
metallic parts. Projects develop
include the NEXX SU bike Helmet
that uses a special design opening
mechanism, and a seat module for M1
vehicles with an hybrid structure of
injected plastic and metal stamp parts.
Regarding electric mobility has work
on Norwegian electric car Buddy M9
and the CEIIA MobiCar program.

Jose Silva - Licentiate in Mechanical

Engineering. He has 5 years’
experience in product development
participating and coordinating the
development of different systems and
modules for various OEM's, including
VW and McLaren. Alongside the
latest manufacturer, he attended as
team management regarding
subsystem design, packaging and
assembly.

Alexandre Teixeira, Head of Auto &

Mobility Unit. Polymer Engineer. Was
involved in more than 150 projects
related to the Aeronautic, Automobile,
Medical and Electronic on the last 10
year. Creator and coordinator of
projects on the area of technologies of
gas-assist injection systems, bio-
material injection (natural cork),
natural fibres. Presently he is
dedicated on the new mobility trends,
related with electric mobility systems,
in terms of new modules and systems
developments.

Bernardo Ribeiro holds a PhD in

Mechanical engineering (automotive)
and graduated with a Mechanical
Engineering Degree. Since 2003 he is
being working on automotive
powertrain R&D. He joined CEIIA in
2009 and at the moment he is the R&D
coordinator at CEIIA, with
responsibility of technological
intelligence, definition and supervising
of R&D projects.

EVS27 International Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 6
 International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering - (ICRIEECE)

Design and Development of BLDC controller and its

Implementation on E-Bike
Khan Farhan Ibne Faruque, Nuzhat Nawshin, Md. Fahim Bhuiyan, Mohammad Rejwan Uddin, Mahady Hasan and
Khosru M Salim

Fab Lab IUB

Independent University, Bangladesh
Dhaka, Bangladesh
[email protected]; [email protected]; [email protected]; [email protected]@gmail.com;
[email protected] and [email protected]

Abstract— In Bangladesh with this immense problem of traffic people uses it to reach the destination [2]. Bicycle is the only
the most compatible among all vehicles is bicycle. Though it is vehicle that is manufactured in Bangladesh and exported to
very user friendly, it requires massive amount of mechanical foreign countries. Despite all the pros of a bicycle, people
energy for pedaling for which the concept of incorporating would rather select a motorbike due to the strain in paddling a
electricity in bicycle came in mid-20th century. With the
bicycle (Fig.1). Due to this reason alone, the concept of E-bike
development of technology, the paper is focusing on developing
an electric bike with recent technologies and to make it budget took a revolution in many countries but this still didn’t make
friendly for Bangladesh. its strong ground in Bangladesh.
For hardware implementation, PROTEUS simulator is used to
design the BLDC controller by using Arduino IDE An electric bike (e-bike) is a cycle with an electric motor
microcontroller. A Hub motor which is very sleek in design and that supports pedaling [3]. A battery powers the motor. With
very light in weigh, is considered for the electric bicycle. Lithium the same amount of energy from the cyclist the e-bike has a
ion batteries was added for power backup because of its high higher speed compared to a conventional bicycle [4]. In other
energy density, low self-discharge rate and low maintenance. The words, the cyclist on the e-bike needs to give less energy to
mobility of the power supply was one of the main objectives so
reach the same speed as the cyclist on a conventional bicycle.
that it can be charged whenever it is needed.
This E-bike is a motorized vehicle which has no need for any
registration and it reduces all the hassle of traffic.
Keywords— Electric Bicycle, BLDC, Hub Motor, Arduino IDE,
Lithium Ion, Power backup, efficiency. For the E-bike, a BLDC motor is preferred over traditional
DC brushed motor [5]. They are more reliable and can be used
I. INTRODUCTION
in poor environmental conditions as the BLDC motor have no
In the urban area of Bangladesh, according to BRTA chance of sparking, unlike brushed motors, making them
(Bangladesh Road Transport Authority), over 332,000 better suited to environments with volatile chemicals and fuels
registered motorcycles were in the road in 2016 and now it has [6]. For sleek and clean appearance for the project, a BLDC
almost doubled [1]. The reason behind this massive increment HUB motor was used.
is because motorcycle overcomes the traffic easily; it is much
cheaper than four wheelers and consumes very less fuel.
Motorcycle has some drawbacks too, such as it needs
registration, it needs petrol to run and motorcycle causes many
deadly accidents too.

Fig.2. Functional block diagram of an E-bike

To develop an electrical bicycle, the paper will focusing on

developing a BLDC controller with power backup system.
The basic principle will be according to the block diagram of
Fig.2. The controller will work accordingly the signals
Fig.1.Conventional manual bicycle obtained from the BLDC motor. This controller will invert the
Though bicycle can overcome all these problems and is also DC power from battery to AC power that will run the motor.
1/6th of the price of any mediocre motorcycle but only few Li-ion battery will be used to build the battery bank.

978-1-5386-5995-3/18/$31.00 ©2018 IEEE

1461
 International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering - (ICRIEECE)

Fig.3. Proteus simulation circuit of BLDC controller

II. SIMULATION OF THE PROPOSED BLDC CONTROLLER

Simulation of the BLDC controller is carried out by using

Proteus Simulator (Fig.3). 1 kHz PWM signal is generated by
using microcontroller (Atmega 328p used in the control
circuit). The duty cycle of this PWM is controlled by a
reference input voltage into the micro controller. A voltage
divider circuit is then used to obtain the reference in the
Atmega 328p microcontroller to generate PWM signals and to
control its duty cycle and charging current. This driver initiate
the dead time to the switches. A driver (IR2101) is used to
drive the IRF540 N-Channel MOSFETs.
The code was done in Arduino IDE as Arduino pro mini
was used for this projects. The main challenges that were
Fig.5. Proteus simulated waveform showing gate signals with 50% duty cycle
faced was in developing the code, since almost everything
relied in this part. The main idea was to generate 6 different The BLDC control technique is done by using sensors like
PWM for different hall combination. In total of six hall sensors or optical sensors to detect position of the rotor
combination of hall sensors were input in 3 analog port of (Fig.4). But it can be driven also without any position
Arduino mini pro. The Arduino Pro mini has 6 PWM modules. detection by sensors [7]. The terminal voltages on the stator
PWM signals generated from those modules has a frequency windings of each phase are controlled by the power electronic
of 500HZ except D5 and D6 which had a frequency of 1kz. switches. The PWM signals generated by the microcontroller
This was one of the limitation faced since these has different to drive the switches by using reference signals from the hall
frequencies a precision was needed. sensors (Fig.5). Two phases are conducting at any instant. For
conduction of two phases at a time (Fig.6), electronic
switching is needed and by varying the switching speed, speed
of motor can be varied [8].

Fig.6. Waveform showing the output voltage (Vout) of 3 different phases

Fig.4. Hall sensors signals obtained from the BLDC motor

1462
 International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering - (ICRIEECE)

III. SYSTEM DEVELOPMENT AND TESTING B. Experimental testing of the controller

Almost all power-electronics products have become The following Fig.8 shows the experimental setup of the
microprocessor controlled in recent years. The main proposed electrical bicycle controller.
advantage of using microcontroller is that it reduces the
complex control hardware where the control functions can be
changed as many times as it is required without extra
hardware. Furthermore, it offers the possibility of improved
reliability and flexibility, different microcontroller is used.

A. BLDC controller development

After successful simulation, the controller was developed

after completing the PCB. The microcontroller has distinctive
PWM signal and is readily available. The micro controller
produces 6 distinctive PWM signal. A motor driver is used to
drive the motor, for this project a motor driver IR2101 was Fig.8. Experimental setup of Proposed BLDC controller
used. The power converter is a three phase power
semiconductor bridge. The main function of the power After the code and simulation was done successfully, the
converter is to transform power from the DC source to AC so entire circuit was tested before implementation. For the
the motor can convert electrical energy to mechanical energy. implementation purpose, a normal second hand bike was
The semiconductor bridge circuit was made out of 6 IRF540 bought for modification into E bike. Then the rear tyre was
N-Channel MOSFET (Fig.7). replaced with the hub integrated wheel. The controller was
The controller was one of the main objectives of this packed inside a box.
projects, the required specification for the motor controller to
work was 36 volt, 250 watt and a current drawing of 5 amp C. Battery bank design
max. To meet up the criteria and availability, a components
selection was done to get the optimized result possible. The E-Bike motor rating 250 W and the minimum
continuous operating time about 1 hour at full throttle requires:

Energy Requirement = Motor rating × Operation Time (1.1)

= 250W × 1hr
= 250Wh
So, a battery backup system to provide 250Wh of energy to
the load is requisite. Therefore, the minimum AH capacity for
the system is
= Battery Energy requirement / System Voltage (1.2)
= 250 / 36 = 6.94 Ah
So, for a 36V, 250W system, the current required is around
10 A. Therefore, 40 batteries each of 3.6V, 2.5A rating lithium
ion batteries are needed.10 batteries in series making one
branch, more 3 identical branches added in parallel making
3.6 V × 10 = 36 V; 2.5 A × 4 = 10 A
Since there was a budget and resource limitation of LI-ion
Batteries only 10 batteries was affordable for the purpose
making up only one branch and giving the run time of 22 min
Fig.7. Prototype of complete BLDC controller on full throttle theoretically. In the Fig.9 below shows the
A dc voltage of 36V obtained from the Battery Bank. The battery bank that was developed.
controller consists of 6 MOSFET IRF540N switches.
ATMEGA328P microcontroller is used to generate 6 PWM
signals. The duty cycle of this PWM signal is controlled by
the throttle input. A gate-driver module (IR2101) is used to
drive the MOSFETs. This driver provides amplifies gate
signals at the same duty cycle to drive one high side and one
low side MOSFET at a time. Bootstrap circuit is used to
complete the driver circuit and amplify the signal coming
from the microcontroller. Fig.9. 36V, 2.5Ah battery bank housing

1463
 International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering - (ICRIEECE)

IV. EXPERIMENTAL DATA ANALYSIS TABLE I

EXPERIMENTAL DATA OF THE DEVELOPED CONTROLLER WITHOUT ANY LOAD

After complete testing of the BLDC controller in Lab, the Duty Cycle Voltage Current
output voltage from the controller was obtained by using 10% duty cycle 40.3 V 0.5A
battery as input. The oscilloscope shows the PWM signals 50% duty cycle 40 V 0.7A – 0.8A
(Fig.10) and the corresponding output (Fig.11) obtained from
the controller. The PWM signals from the micro controller
100% duty cycle 38.7 V 0.96A – 1.1A
generated according to the hall sensor output signal. 100% duty cycle 38.7 V 2.0A
(sudden start)

In table 1, a full charged battery was at 41v, when the

throttle was at 10%, 50%, 100% the voltage dropped to 40.3v,
40v, 38.7v. At general start, the voltage drop was 38.7v and
the current draw increased from 1.1A to 2 A. On an average,
the current draw from battery was 0.5A to 1.1A

V. RESULTS AND DISCUSSIONS

After no load test, the controller and battery bank was

assembled in the cycle with the BLDC motor. The complete
electric bicycle shows in Fig.12.

Fig.10. Rectified input DC input voltage

Out of the 3 poles, 2 poles were detected at a time and

hence those poles were energised with the 6 binary
combinations which help turn the motor. With respect to the
signals, a 3 phase signal was generated with the controller and
an unfiltered raw signal at 70% duty cycle was generated as
shows in Fig.11.

Fig.12. Complete setup of electric bicycle

By completing full setup, the e-bike was tested with load.

The experiment was conducted with 3 different duty cycles:
10%, 50%, and 100%. For 100% there are 2 cases, soft start
and hard start. The experimental data was tabulated in table 2.
TABLE II
EXPERIMENTAL DATA OF THE DEVELOPED CONTROLLER WITH 120 KG LOAD

Time Duty Cycle Voltage Current

5:00 pm 10% duty cycle 40.3 V 0.5-1.4A
5:10 pm 50% duty cycle 39 V 1.8- 2.1A
5:25 pm 100% duty cycle 38V 2.0- 2.2A
5:30 pm 100% duty cycle 38V 5 – 6.6A
Fig.11. Output voltage before filtering
(sudden start)
Table 1 shows the experimental data obtained from the
controller after no load testing. The voltage of the battery bank In table 2, the voltage drop in 10%, 50% and 100% was
with corresponding current was recorded and showed in the similar but the current draw was slightly more than in no load
table. condition. With the full charged battery, the cycle run almost
half an hour on full throttle.

1464
 International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering - (ICRIEECE)

VI. CONCLUSIONS REFERENCES

In this paper, a detailed description of the design and [1] S. M. Labib, Hossain Mohiuddin, Shahadat Hossain Shaki “Transport
construction of a more efficient and lightweight BLDC motor Sustainability of Dhaka: A Measure of Ecological Footprint and Means
for Sustainable Transportation System” Journal of Bangladesh Institute
controller for electric easy bicycle has been presented. The of Planners ISSN 2075-9363 Vol. 6, December 2013.
controller and the electric bicycle has tested successfully in
the lab. The experimental data shows an excellent [2] https://bdnews24.com/business/2015/03/23/businessmen-expect-sharp-
performance of the developed e-bike with the Li-Ion battery rise-inbicycle-exports.
bank. With only 10 batteries, the cycle is capable to run [3] X. Zhou, G. Wang, S.Lukic, S. Bhattacharya, and A. Huang, “Multi-
almost 30 minutes. By adding more batteries, the run time can function bi-directional battery charger for plug-in hybrid electric
be enhanced. vehicle application,” in Proc. Energy Conversion Congr. and
Exposition, 2009, San Jose, CA, USA, 20-24 Sep. 2009, pp. 3930–
3936.
It can be concluded that the idea that is described through
this paper can potentially help people to cut their daily cost [4] Khosru M. Salim, Md. Jasim Uddin, M. Ishtiaque Rahman,
they spend for travelling, the traffic they face on daily basis Mohammad Rejwan Uddin “Design, Construction and Implementation
and reach their destination early. In contrast there are many of a Highly Efficient, Lightweight and Cost Effective Charger for
Electric Easy Bikes” ICDRET 2016.
rural areas in Bangladesh where the only vehicle available is
bicycle or there is no vehicle at all, there the electric bicycle [5] Mohammad Rejwan Uddin, Zaima Tasneem, Saila Ishrat Annie and
might cut down the daily labor one put to reach to work. With Khosru M Salim, “A High Capacity Synchronous Buck Converter for
further research, this project can further be developed and be Highly Efficient and Lightweight Charger of Electric Easy Bikes”
International Conference on Electrical, Computer and Communication
implemented in 3rd World country at a cheaper price. Engineering (ICECCE 2017), Chittagong, Bangladesh.

[6] Mohammad Iqbal, Rakibulhasan and Tanvir Ehsan “study on merits

and demerits of two transport systems: battery operated easy bike with
motor operated rickshaw at sylhet city in bangladesh” International
Conference On Mechanical Engineering And Renewable Energy 2015
(ICMERE2015) 26-29 November, 2015.

[7] Md.Zulkefa Rian1, Dr. A. N. M. Mizanur Rahman 2,* “Study on

Power Consumption and Social Aspects of Battery Operated Auto-
rickshaw” International Conference on Mechanical, Industrial and
Energy Engineering 2014.

[8] N.Z. Yahaya, M.K. Lee, K.M. Begam & M. Awan “The Analysis of
Dead Time on Switching Loss in High and Low Side MOSFETs of
ZVS Synchronous Buck Converter” ACEEE Int. J. on Control System
and Instrumentation, Vol. 02, No. 02, June 2011.

1465
 2016 IEEE 22nd International Symposium for Design and Technology in Electronic Packaging (SIITME)

E-bike electronic control unit

Florin Dumitrache, Marius Catalin Carp, and Gheorghe Pana
Electronics and Computers Department
Transilvania University of Brasov
Brasov, Romania
[email protected]

Abstract — The electric vehicles industry is continuously According to [1], the main components of a system with
evolving. One type of such electric vehicle is the electric bicycle BLDC motor are: (1) control logic, (2) power stage comprised
(e-bike). Electric bicycles, like other electric vehicles, use a BLDC of six switching devices (e.g. MOSFETs, IGBTs) and (3)
motor (Brushless Direct Current Motor). This paper presents a sensors used for the closed-loop feedback.
way of designing and implementing an electronic module for an
e-Bike. The paper shows how a low power, 8-bit microcontroller The performance of a BLDC motor is dictated mainly by
can be used to drive such a motor and also manage other useful the motor structure and the control logic that is been used. By
functions on an e-Bike. using different types of control logic, the torque ripple of the
BLDC motor can be minimized [2].
Keywords —BLDC; e-Bike; controller
II. BLOCK DIAGRAM
I. INTRODUCTION
The block diagram of the designed electric system is
Electric vehicles make use of BLDC motors as the presented in Fig. 2. The implementation of the proposed
propulsion method. Due to the fact that BLDC motors do not control unit is based on an 8-bit microcontroller.
have brushes, they present some advantages over the DC
brushed motors, from which we remember: (1) longer life span, The microcontroller receives information about the motor
(2) lower EMI (Electromagnetic interference) radiation, (3) position (rotor angle), via signals generated by three Hall-
noiseless operation, (4) grater torque to motor size ratio [1]. effect sensors contained within the motor.
Due to the geometry of the windings in the motor, the BEMF Using this data, the microcontroller uses a simple
(back electro-motive force) generated by the motor when in commutation table and switches the six power MOSFET
generator mode can be of two types: (1) trapezoidal and (2) transistors which drive the BLDC motor.
sinusoidal. The latter can be of interest if the driven motor does
not have Hall position sensors, and facilitates the calculation of Internet-based literature (images only) suggests that BLDC
the motor’s rotor absolute angle. controllers used in commercial e-Bikes contain linear power
supplies. This is a main issue regarding power efficiency, due
The internal structure of a BLDC motor is presented in to the significate power loss as heat in the internal power
Fig. 1. BLDC motors are 3-phase motors, and to properly drive supply.
such a motor, a special control circuit must be used. The
purpose of the control circuit is to energize the correct
winding(s) at the right moment. This is achieved by reading
information from certain rotor position sensors and generating
PWM (pulse width modulation) signals.

a) b)
Fig. 1. BLDC motor: a) simplified internal structure; b) motor windings Fig. 2. Block diagram of the designed electric system

Sponsor: S.C. PREH ROMANIA S.R.L.

978-1-5090-4445-0/16/$31.00 ©2016 IEEE 248 20-23 Oct 2016, Oradea, Romania

 2016 IEEE 22nd International Symposium for Design and Technology in Electronic Packaging (SIITME)
One of the improvements this design brings is the use of a The circuit requires the existence of a negative supply rail.
DC-DC step down converter, which greatly lowers the power This is achieved by using a 555 timer, U104, in a switched
consumption of the module and reduces the ambient capacitors configuration.
temperature in the case of the module. The latter is an
important issue, considering the fact that the electronics on an Other supply voltages are obtained using LDO (Low
e-Bike must be housed in a water-proof enclosure. dropout) regulators. The combination of a DC-DC pre-
regulator followed by linear regulators provide a low output
The user of the e-Bike receives relevant data (e.g. voltage ripple, while maintaining a good overall efficiency.
instantaneous speed, battery state of charge) from the motor
controller via Bluetooth protocol and can view the data on a In Fig. 3 b the logic circuitry of the controller is presented.
GUI (graphical user interface), i.e. on a smartphone. By using a The heart of this block is an AVR ATmega128 microcontroller,
Bluetooth transmitter instead of a graphical display, the power U201. This solution has been chosen as a good compromise
consumption is furthermore reduced. between processing power and low consumption.
In order to increase the EM radiation immunity of the
III. ELECTRONIC SCHEMATIC microcontroller, the voltage supply level of the microcontroller
The hardware implementation of the block diagram from is the maximum allowed, i.e. 5V.
Fig. 2 is presented in Fig. 3, without the power stage.
The interface between the microcontroller and the
The power supply is designed using a dedicated DC-DC Bluetooth module is made using a bidirectional level shifter,
step down IC (integrated circuit) – U101 in Fig. 3 a. By using comprised of Q201 and Q202 transistors. The level shifters are
an IC with integrated power MOSFETs, EM radiations are necessary because the microcontroller and the BT module are
reduced, due to the fact that there are no necessary external powered from different supply voltages (5V and 3.3V
traces on the PCB. respectively).

VBAT_36V +12V +12V U102 +5V U103 +3.3V

U101 TL780-05 LM1117-3.3
L101 TO263 TO252
13/14 1/2 3 1
VIN SW VI VO VI VO
47uH

GND
C101 C102 C103 SLF12575T C108 C109 C110 GND
10uF 10uF 470nF C105 10uF 10uF 10uF
1210 1210 1206 470nF 1210 1210 1210 R104 C111
C112 C113 C114 C115

2
1206 20k 6.8nF
3 0603 0603 470nF 470nF 10uF 10uF
CBOOT
1206 1206 1210 1210
R101 GND GND GND
2M2 GND 9
FB
0603
4 +5V
VCC
12
EN R105 GND
C106 1.8k
11 4.7uF 0603
SS/TRK
1206
7
RT
R106
GND 1k
R102 C104 R103 GND 0805
U104

8
120k 22nF 82k 5 -5V
BIAS
0603 0603 1206 7 4
C118 D101

VCC
DC R
C107
1uF 3
Q
1206 R107
GND GND GND 10k 10uF 1N4148
6 0805 1210 SOD80
SYNC
GND 2
TR CV
5 R108
PGOOD
8 10k C119
GND

6 1206 22uF
TH
10
AGND PGND
15/16 Q101 1210
C116 LM555 C117 BC847C
1

1nF SO8 10nF SOT23

GND LM46002-Q1 GND 1206 1206
HTSSOP16

GND GND GND GND GND

a)
+5V +5V +5V +3.3V +3.3V BT Antenna

R201 U203
16 33
21/52
Q201 3V3 RF_IO
1.5k U201 2N7002
6
1206 SOT23 RESET
20 25 R202 R203 31
VCC

RESET RESET PD0/SCL/INT0 PIO0

26 10k 10k 7 30
PD1/SDA/INT1 MISO PIO1
XTAL1 24 27 RXD1 8 29
XTAL1 PD2/RXD1/INT2 1206 1206 CS PIO2
XTAL2 23 28 TXD1 9 28
XTAL2 PD3/TXD1/INT3 CLK PIO3
X201 PD4/ICP1
29
HEADLIGHT
10
MOSI PIO4
27
51 30 26
CMD_FAR PA0/AD0 PD5/XCK1 BRAKE PIO5
50 31 +5V +3.3V 11 25
PA1 PA1/AD1 PD6/T1 SIG_L CTS PIO6
49 32 12 24
PA2 PA2/AD2 PD7/T2 SIG_R TX PIO7
48 13 1
16MHz PA3 PA3/AD3 RTS PIO8
C202 HC-49/US C203 PA4 47
PA4/AD4 PE0/RXD0/PDI
2
RxD0 Q202 14
RX PIO9
2
18pF 18pF 46 3 3
PA5 PA5/AD5 PE1/TXD0/PDO TxD0 2N7002 PIO10
1206 1206 45 4 23 15
PA6 PA6/AD6 PE2/XCK0/AIN0 HALL_A SOT23 USB_DN PIO11
PA7
44
PA7/AD7 PE3/OC3A/AIN1
5
PWMA R204 R205 22
UDB_DP
6 10k 10k
PE4/OC3B/INT4 PWMB
10 7 1206 1206 4 21
PB0/SS PE5/OC3C/INT5 PWMC AIO0 PCMCLK
GND GND 11 8 5 20
SCK PB1/SCK PE6/T3/INT6 HALL_B AIO1 PCMI
12 9 19
PB2/MOSI PE7/ICP3/INT7 HALL_C PCMS
13 17/32/34 18
PB3/MISO GND PCMO
14 61
PB4 PB4/OC0 PF0/ADC0 TEMP
+12V 15 60 BTM-112
U202 PB5
16
PB5/OC1A PF1/ADC1
59
ASENSE
GND
PB6 PB6/OC1B PF2/ADC2
17 58
PB7 PB7/OC2/OC1C PF3/ADC3 ADC3
2 6 +5V_ADC 57
IN OUT PF4/ADC4/TCK ADC4
35 56
PC0 PC0/A8 PF5/ADC5/TMS ADC5
+12V 36 55
PC1 PC1/A9 PF6/ADC6/TDO ADC6
3 37 54
TEMP PC2 PC2/A10 PF7/ADC7/TDI THROTTLE
38 VBAT
PC3 PC3/A11
39 33
4 5 U204:A PC4
40
PC4/A12 PG0/WR
34
GND TRIM PC5 PC5/A13 PG1/RD R207
8

LM358
41 43
SO8 PC6 PC6/A14 PG2/ALE
3 42 18 100k
PC7 PC7/A15 PG3/TOSC2
GND MAX6175 1 19 1206
TEMP PG4/TOSC1
SO8 2 62
AREF
GND

64
AVCC PEN
1 R212 R213 R214 R215 C210 D201 C208 R208
10k 10k 10k 10k 470nF 5V6 1uF 12k
4

R209 R210 ATMEGA128A 1206 1206 1206 1206 1206 SOD80 1206 1206
TQFP64 22/53/63
GND

paralel R=1k
10k 56k
1206 1206
GND GND GND GND GND GND GND GND GND GND

b)
Fig. 3. BLDC electronic control unit schematic: a) power supply; b) logic circuitry

978-1-5090-4445-0/16/$31.00 ©2016 IEEE 249 20-23 Oct 2016, Oradea, Romania

 2016 IEEE 22nd International Symposium for Design and Technology in Electronic Packaging (SIITME)
The BLDC motor is driven with six MOSFET power IV. SOFTWARE AND CONTROL LOGIC
transistors, M1 to M6 in Fig. 4 a, arranged in a 3-phase Due to the fact that the software used by commercial e-
H-bridge configuration. Bikes is confidential, the software has been designed based on
The bridge current is measured using a 4-point type the fundamentals of BLDC motors presented in the
measurement using a differential amplifier. The differential introduction of this paper and other studies [2][4].
amplifier is followed by a single diode precision rectifier [3]. The commutation scheme used in this design is described in
The use of this measurement circuitry comprised of U301 and Fig. 5. By analysing the diagram in Fig. 5, the system of
U302 makes possible the measurement of both the consumed equations (1) has been deduced, through which the control
power, when the motor acts as a load, and the recovered power, logic has been modelled as a combinational system. The inputs
when the motor acts as a generator. to this combinational system are the Hall position sensors
This novel method of current sensing can be further used to signals and a master PWM, and the outputs are the PWM
limit battery charge/discharge current, or calculate the battery signals.
state of charge more precisely.
𝐴𝐻𝑆 = 𝑃𝑊𝑀 ∙ (𝐻𝑎𝑙𝑙𝐴 ∙ ̅̅̅̅̅̅̅̅
𝐻𝑎𝑙𝑙𝐵 );
The MOSFET driver ICs receive the PWM signals from the
𝐵𝐻𝑆 = 𝑃𝑊𝑀 ∙ (𝐻𝑎𝑙𝑙𝐵 ∙ ̅̅̅̅̅̅̅̅
𝐻𝑎𝑙𝑙𝐶 );
microcontroller and ensure the proper control voltage levels
and timing for the power transistors. For the high side ̅̅̅̅̅̅̅̅);
𝐶𝐻𝑆 = 𝑃𝑊𝑀 ∙ (𝐻𝑎𝑙𝑙𝐶 ∙ 𝐻𝑎𝑙𝑙𝐴
(1)
MOSFET, the control voltage must be positively offseted with 𝐴𝐿𝑆 = 𝐻𝑎𝑙𝑙𝐵 ∙ ̅̅̅̅̅̅̅̅
𝐻𝑎𝑙𝑙𝐴;
a voltage equal to the voltage in the transistor’s source
connection. One of three MOSFET drivers is presented in Fig. 𝐵𝐿𝑆 = 𝐻𝑎𝑙𝑙𝐶 ∙ ̅̅̅̅̅̅̅̅
𝐻𝑎𝑙𝑙𝐵 ;
4 b. Using specialized MOSFET drivers rather than discrete { 𝐶𝐿𝑆 = 𝐻𝑎𝑙𝑙𝐴 ∙ ̅̅̅̅̅̅̅̅
𝐻𝑎𝑙𝑙𝐶 ;
level shifters with BJTs (bipolar transistors) assures fast, When choosing the commutation scheme the following
well-controlled rise and fall times of the power signals and factors have been taken into account: (1) simplicity of
thus, reducing power losses caused by the commutation of the implementation, (2) delivered performances motor speed
transistors. corresponding to an average speed of 20 to 25km/h and (3)
VBAT
amount of CPU load determined by the motor control function
in conjunction with the secondary functions (i.e. data
AH
M1
IPB037N06N BH
M3
IPB037N06N CH
M5
IPB037N06N
transmission, analogue signals processing, user commands
processing).
TO263 TO263 TO263

R301 R303 R305

10k C301 10k C302 10k C303
1206 2x10uF 1206 2x10uF 1206 2x10uF

The software flow diagram is presented in Fig. 6.

1210 1210 1210
A B C

M2 M4 M6
AL IPB037N06N BL IPB037N06N CL IPB037N06N
TO263 TO263 TO263

R302 R304 R306

10k 10k 10k
1206 1206 1206

+5V R307 R309

+5V
U301:B 100k 10k
1206 1206
8

OPA2320
GND
8

TSSOP8
3 RS301
8

R312 5 1 5m
3 7 2 SHUNT
ASENSE
1 33k 6 U301:A
2 1206 OPA2320
U302:A
4

TSSOP8 R308 R310

4

OPA2320
4

TSSOP8 100k 10k

RV301 1206 1206
GND
47k
R314 GND R313 -5V
1804-G

33k 33k
1206 1206
D313 R311 J301
10k
1206 1
A
2
B
1N4148 3
C
4

SOD80
6 GND BLDC Motor
7
5
U302:B
OPA2320
8

GND
TSSOP8

a)
Fig. 5. PWM commutation scheme

+12V

C401 C402 U401

1

10uF 470nF
1210 1206 R401 R403
VDD

5 3
PWMA INH DH AH
10R 6R2
2
1206 BST 1206
GND GND
C405 D403
100nF
1206
1N4148
4 SOD80
HS A

R402 6 8
R404
INL DL AL
GND

10R 6R2
1206 1206
MAX15013D
D404
7/9

SO8-EP

GND 1N4148
SOD80

b)
a) b) c)
Fig. 4. BLDC electronic control unit power circuitry: a) 3-phase H-Bridge Fig. 6. Electronic control unit software flow diagram: a) main program;
and current sensing circuit; b) MOSFET driver circuit b) motor control thread; c) data transmission thread

978-1-5090-4445-0/16/$31.00 ©2016 IEEE 250 20-23 Oct 2016, Oradea, Romania

 2016 IEEE 22nd International Symposium for Design and Technology in Electronic Packaging (SIITME)
The software presented in the flow diagram, in Fig. 6, is The result shown in Fig. 8 suggests that the proposed
structured in two threads, one for each main function fulfilled. commutation table will present good performances. This can be
The threads are governed by two separate timers, the motor deduced from the rotor speed graph, in which a speed variation
control thread having the highest priority. of around 0.5%
V. SOFTWARE SIMULATION VI. CONCLUSIONS
The system has been simulated using Matlab Simulink The entire system has been implemented and tested. The
simulation environment. The simulation took into account the results obtained are comparable with commercial e-Bikes, and
commutation table presented in Fig. 5 and described by (1). the proposed implementation presents a higher power-
The Simulink model is presented in Fig. 7. The results of the autonomy ratio.
simulation are (1) the stator current waveform, (2) rotor speed
and (3) electromagnetic torque ripple. These results are shown The results presented in Table I have been obtained using a
in the graphs in Fig. 8. 350W, 3-phase BLDC motor and a 36V / 5.5Ah Li-Ion battery
pack, mounted on a 13kg bicycle.
The use of Li-Ion cells has been taken into consideration
after studying the e-Bike market and observing that many
commercial e-Bikes use Lead-acid batteries. The designed e-
Bike is compared with a commercial e-Bike of about same cost
and complexity.

TABLE I. DESIGNED SYSTEM VS. COMMERCIAL SYSTEM

Criteria Designed system Commercial system

33 km/h (software
Maximum speed 25 km/h
limited at 25 km/h)
Motor power 350 W nominal 250 W nominal
Power-on-demand & Power-on-demand OR
Assist type
pedal-assist pedal-assist
Battery 36V Li-Ion / 5.5 Ah 36V Lead–acid / 12 Ah
Autonomy 20-22 km 40-45 km
Total mass 19-20 kg 27 kg

An overall conclusion is that the hardware improvements

Fig. 7. Simulink model of the proposed system (i.e. higher efficiency power supply, components with lower
power consumption, reduced switching losses in the power
stage), in conjunction with a simple yet effective software,
yield a low-cost and reliable product.

ACKNOWLEDGMENT
This work has been developed in the frame of a partnership
between Transilvania University of Brasov and the sponsor,
S.C. PREH ROMANIA S.R.L.

REFERENCES
[1] P. Yedamale, “Brushless DC (BLDC) motor fundamentals”, Application
Note AN885, Microchip Technology Inc., 2003.
[2] H.S. Chuang, Yu-Lung Ke, and Y.C. Chuang, “Analysis of
Commutation Torque Ripple Using Different PWM Modes in BLDC
Motors”, Conference Record 2009 IEEE Industrial & Commercial
Power Systems Technical Conference, pp. 1-6, May, 2009.
[3] D. Jones, M. Stitt, “Precision absolute value circuits”, Burr-Brown
Corporation, December, 1997
[4] W. A. Salah, D. Ishak, K. J. Hammadi, “Minimization of torque ripples
in BLDC motors due to phase commutation”, Przeglad
Fig. 8. Simulation results Elektrotechniczny, R. 87, pp 183-188, January 2011.

978-1-5090-4445-0/16/$31.00 ©2016 IEEE 251 20-23 Oct 2016, Oradea, Romania

 2nd IEEE Eurasia Conference on IOT, Communication and Engineering 2020

Design and Optimizated Simulation of Bicycle Frame Structure

Tian-Syung Lan1,2, Huo-Wang Zhang1, Jian-Song Gao3, Xuan-Jun Dai1*

1
College of Mechanical and Control Engineering,Guilin University of Technology, Guilin Guangxi, 541004,China
2
Department of Information Management,Yu Da University,Miaoli County,36143,Taiwan,Republic of China
3
College of Mechanical and Electrical Engineering, Handan University,Handan Hebei,056005,China
*
Corresponding Author: Email: [email protected]

Abstract

The frame is not only the main body of the bicycle but also
the bearing part of the bicycle. It bears the gravity of people
and the weights of the bicycle. It must have enough reliability,
durability, and lightness. According to the working conditions
of the frame, the strength is checked from static strength and
fatigue strength, and the length of each member of the bicycle Fig. 1 The creation of the basic structure of a frame.
frame and the angle between them are designed. The model is
created by using PRO/E three-dimensional design software. The lower half of the rear fork is created in the second step
The established three-dimensional model is imported into for the entire frame as shown in Fig.2(left).
ANSYS finite element software, and then the static and modal Then, the details of the part are processed, chamfered,
simulation of the frame structure is established. Finally, the rounded, and reinforced to obtain the final model as shown in
optimization of the frame is completed according to the Fig. 2.
analysis results.

Keywords: simulation, finite element analysis, design, frame

Introduction

The development of bicycles has been achieved rapidly[1].

Although many convenient means of transportation have
appeared in people's lives, bicycles as a traditional way of
Fig. 2 Frame rear fork creation.
transportation have always been close to life and can be seen
everywhere [2]. At present, the characteristics of bicycles can
C. Overall assembly model
be summarized as light, convenient, and environmentally
In the assembly, each part of coincidence, coplanar, parallel,
friendly. Moreover, it can effectively alleviate the problem of
distance is assembled according to the design requirements.
urban traffic congestion, especially in big cities [3,4].

Creation of frame 3D solid model

A. PRO/E modeling technology

PRO/E plays an important role in mechanical 3D graphics
software and focuses more on product design. Nowadays, the
software has been widely used in various fields such as
automotive, industrial design, mold production, and electronic
Fig. 3 frame assembly model
equipment [5,6].
Finite Element Analysis Based on ANSYS
B. Frame three-dimensional modeling process
Through the stretching and extension, the riser of the frame
A. Basic theory of finite element analysis
is obtained, the head tube is created by using the rotation and
The basic idea of the finite element method is to divide the
stretching instructions, and the upper tube is created by using
object into finite elements, which contributes to continuous
the protrusion. The first tube is connected to the riser, and then
geometric structure mesh and establishes finite nodes in each
the hybrid scan instruction is used. The lower tube and basic
element. For the analysis of each discrete element, the discrete
structure of the frame are completed as shown in Fig.1.
block is called the element, the node between the shape of the
To facilitate drawing, we create two reference planes, DTM1
control unit and the element is called the node, the internal
and DTM2. As the rear fork is a symmetrical structure, we only
force of the structure on the node is called the node force, and
need to use sketching, hybrid scanning, and materialization
the external force on the node such as concentrated force and
commands for half of the bicycle. The mirroring function
distributed force is called the node load [7-9].
allows completing the other half as shown in Fig.1(right).

ISBN: 978-1-7281-8060-1 301

Authorized licensed use limited to: Carleton University. Downloaded on May 31,2021 at 20:19:57 UTC from IEEE Xplore. Restrictions apply.
 2nd IEEE Eurasia Conference on IOT, Communication and Engineering 2020

B. ANSYS finite element analysis

The modules and steps of ANSYS finite element analysis are

shown in Table 1.

TABLE 1
ANSYS modules and steps
Logic Pretreatment Solver Post
Fig. 5 displacement cloud map.
module processing
Analysis Create a finite Enter the Enter the
step element model solver to deformation Under static loading conditions, the stress cloud is as follows.
solve map, contour
map, list, etc.

C. Static analysis of the frame

POR/E is used to draw a three-dimensional model of the
bicycle frame. After establishing the POR/E 3D model, the
ANSYS finite element software imports the frame model and
selects the static analysis module [10]. The material selected is
Fig. 6 stress cloud diagram.
aluminum alloy and its parameters are shown in Table 2.
Meshing adopts the tetrahedral free meshing. The unit size
From the above cloud map, the maximum stress of the frame
is set to 4mm, 21821 tetrahedral elements, and 6632 nodes as
is 51.35-295 MPa, so the design of the frame meets the strength
shown in Fig.4. According to the standard, the load boundary
requirements.
conditions acting on the frame are shown in Table 3.
E. Modal analysis of the frame
Under actual conditions, the load changes with time is called
dynamic load [11]. When the dynamic load is small, the effect
of external excitation on the frame is negligible. When the
dynamic load is large, the frame resonates if the external
excitation frequency is similar to or equal to a certain natural
frequency of the bicycle. The bicycle frame structure is
destroyed, so it is necessary to solve this phenomenon through
Fig. 4 Meshing and imposing restriction
modal analysis [12,13].
TABLE 2 The maximum natural frequency of its frame structure is as
Characteristics of aluminum alloy materials follows.
Density Young's Poisson's Yield Tensile Shear 𝑓𝑓𝑓𝑓 = 𝑉𝑉𝑉𝑉 ∗ 𝛺𝛺𝛺𝛺 (1)
modulus ratio Strength strength modulus 𝑓𝑓𝑓𝑓: The excitation frequency.
2700kg/ 71.0GPa 0.32 295MPa 370MPa 27GPa 𝑉𝑉𝑉𝑉 : The maximum speed of a mountain bike does not exceed
m3 60Km/h, so V takes 60.
TABLE 3 𝛺𝛺𝛺𝛺: The spatial frequency.
National standard technical requirements
Load Front tube Center axis Riser part F3 When the external excitation and the frame's natural
position bottom F1 part F2 frequency are similar, the vehicle body resonates. The speed at
Load 50 200 550 which the bicycle resonates [14] is defined as
（N）
direction Vertically Vertically Vertically
downward downward downward
𝑉𝑉𝑉𝑉 = 3.6𝐿𝐿𝐿𝐿𝑤𝑤𝑤𝑤 ∗ 𝑓𝑓𝑓𝑓 and (2)
1
𝐿𝐿𝐿𝐿𝑤𝑤𝑤𝑤 = , (3)
𝛺𝛺𝛺𝛺

To ensure the convergence of the solution and impose

𝐿𝐿𝐿𝐿𝑤𝑤𝑤𝑤 : The ground irregularity wavelength.
constraints on the frame, we apply the method of creating a
rigid domain in which the displacement of the bottom of the
The following table is the wavelength of China's uneven road.
front tube is UX, UY, UZ in three directions. The rear axle and
Combined with the above, the maximum frequency of the
the car The displacement of the frame contact constraints UX,
vehicle can be calculated by
UY, UZ is 0.
𝑉𝑉𝑉𝑉 60
D. Static strength analysis result
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚
𝑓𝑓𝑓𝑓𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 = 3.6𝐿𝐿𝐿𝐿 = = 52.08𝐻𝐻𝐻𝐻𝑧𝑧𝑧𝑧. (4)
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 3.6×0.32
Under static loading conditions, the displacement cloud map
is as follows. Modal analysis results are shown in Fig. 7 and Table 4.

302 ISBN: 978-1-7281-8060-1

Authorized licensed use limited to: Carleton University. Downloaded on May 31,2021 at 20:19:57 UTC from IEEE Xplore. Restrictions apply.
 2nd IEEE Eurasia Conference on IOT, Communication and Engineering 2020

TABLE 4
Road roughness wavelength
To reduce the stress concentration to a greater extent, the
Road surface Dirt road Cobbled Road Stone road Flat road
radius of the round should be made larger, and it can also
Roughness 0.77~2.5 0.32~6.3 0.74~5.6 1~6.3 effectively prevent the sudden situation and scratch the human
wavelength body.

Fig. 7 First and Second order modal Fig. 12 Rounded corners before and after change

TABLE 5
Modal analysis data
Mode diagram Natural Part
frequency /Hz
First order 349.95 Vertical fork
modal
Second order 369.93 Vertical fork
modal
Fig. 8 Third and Fourth order mode
Third order 706.19 Horizontal fork
modal
Fourth order 836.19 Vertical fork
modal
Fifth order 944.48 Vertical fork
modal
Sixth order 991.20 Vertical fork
modal
Seventh order 1163.39 Vertical fork
Fig. 9 Fifth and Sixth order modal modal
Eighth order 1244.24 Vertical fork
modal

Conclusions

This paper uses the relevant references to select and design

the various pipe fittings of the frame and determines the overall
size of the frame structure. The model was drawn by PRO/E
Fig. 10 Seventh and Eighth order modal 3D modeling software, and static strength and modal analysis
were performed by ANSYS finite element analysis. The main
The maximum displacement of the frame is 0.073 mm that conclusions of this paper is as following.
is far less than the deformation requirement of 4 mm and meets (1) The frame of aluminum alloy material has the advantages
the requirements of the design specifications under the of lightweight, high strength, elastic number, and so on, which
conditions of the frame. greatly contributes to the improvement of bicycle quality and
The natural frequency of each step is larger than 52.08 Hz, the improvement of comfort.
so the bicycle frame does not resonate, and the frame design (2) The main load-carrying capacity of the bicycle depends
meets the requirements. on the rigidity and strength of the bicycle frame structure, and
the main load-bearing portion is the connecting portion of the
F. Optimization suggestion riser pipe, the upper pipe, and the rear fork.
Since the human gravity directly acts on the riser and bears (3) Static strength analysis of bicycle by finite element
a large load in the actual production, the riser wall thickness method shows that the maximum stress of electric vehicle is
can be appropriately increased to increase its strength. 51.35-295 MPa and the maximum displacement is 0.073 mm
under the static load.
(4) The modal analysis results show that the minimum
natural frequency is 349.95 Hz, and resonance does not occur.

References

[1] Li Mengxi. Application of virtual simulation technology in

bicycle motion simulation and design optimization [D]. Tianjin
Fig. 11 Wall thickness before and after change.

ISBN: 978-1-7281-8060-1 303

Authorized licensed use limited to: Carleton University. Downloaded on May 31,2021 at 20:19:57 UTC from IEEE Xplore. Restrictions apply.
 2nd IEEE Eurasia Conference on IOT, Communication and Engineering 2020

University of Finance and Economics, 2017.

[2] Huang Wei. Research on custom bicycle design and simulation
detection method [D]. Tianjin University, 2017.
[3] Zhang Min. Design and analysis optimization of two-wheeled
lying bicycle [D]. Qingdao University, 2018.
[4] Xu Wen. Research and application of bicycle design method
considering the attitude of the rider [D]. Tianjin University,
2009.
[5] Wang Wei. Optimization design of rear suspension mechanism
of electric bicycle based on force ratio curve [D]. North
University of China, 2013.
[6] Zhang Wenda. Simulation study on static and dynamic
characteristics of an electric bicycle frame [D]. North
University of China, 2013.
[7] Guo Yunchun. Finite Element Analysis and Structural
Optimization of Bolted Magnesium Frame [D]. Chongqing
University, 2012..
[8] Wu Xiaofan. Research and application of bicycle design
method based on ergonomics [D]. Tianjin University, 2008.
[9] Zhang Li. Finite element modeling and structural optimization
of electric bicycle frame [D]. Tianjin University, 2008.
[10] Zhang Rong. Design and dynamic analysis of new mountain
bike [D]. Xi'an University of Technology, 2017.
[11] Pu Yibing. Design and Finite Element Analysis of Magnesium
Alloy Frame for Electric Bicycles [D]. Chongqing University,
2011.
[12] LifeMOD Manual, Biomechanics Research Group, Inc.2005.
[13] Depreitere B, Van Lierde C, Maene S, et al, Bicycle-related
head injury: a study of 86 cases, Accid Anal Prev 2004.
[14] Redfield R C, Brian S, Brian F,et al, Motion measurements in
the jumping of amountain bike, Journal of Instrumentation,
Systems and Automation Society,2004.

304 ISBN: 978-1-7281-8060-1

Authorized licensed use limited to: Carleton University. Downloaded on May 31,2021 at 20:19:57 UTC from IEEE Xplore. Restrictions apply.
 Environment Friendly booster bike

Ms.S.J.Lopes Mr. Ashutosh Gattelu Mr. Ashwin Ghosalkar Mr. Steon Gonsalves
Electronics and Electronics and Electronics and Electronics and
Telecommunication of engineering Telecommunication of Telecommunication of Telecommunication of
SFIT engineering engineering engineering
Mumbai, India Ex-Student, SFIT Ex-Student, SFIT Ex-Student, SFIT
[email protected] Mumbai, India Mumbai, India Mumbai, India

Abstract— in the above paper i presented an Environment This electric bike can be used for travelling off-road as well
Friendly E-bike/Booster Bike. For this E-bike a brushless Dc as on road, is small in size and compact so comfortable
motor (BLDC) [1] motor was used which acts as the main driving means for travelling. This can be used in large educational
unit. The DC power to BLDC motor could be supplied with the premises as well as company campuses. This E-Bike work on
help of Lead acid battery. The speed or velocity could be changed batteries, which are connected to motors [2], [3]. And the
with the help of a throttle, which works on the similar mechanism
power supply [4] to the motor is controlled by an on off
like in scooters such as Activa where by changing accelerator
speed could be varied. This E-Bike had three gears synchronized
switch. The batteries are connected in series and the power
with BLDC motor and those gears could be changed with the can be cut by the on off switch which is connected between
help of push button switch [1]. motor and battery. The material used in the bike is parts of an
This E-Bike has advanced features like anti-theft protection, an IR old bicycle so it is a low cost bike, basically consists of
sensor for accessing the E-BIKE; this is similar to accessing the battery. When the battery is fully charged it can travel up to
car by remote control. For that we used a micro controller 60kms; it has 3 gears for speed purpose. It will consist of an
8051.Also we have interfaced an LED display which provides us anti- theft system which will be controlled by a remote along
vital parameters like speed of the bike, distance covered, date and with an anti-theft alarm. The functions are controlled by a
time etc. In our E-Bike we also used turn signal indicators light on remote controller for switching ON and OFF and for security
a jacket for safety purpose. For that we used an Arduino
purpose. Along with this E-Bike we also used turn signal
controller which is a Lilly pad Arduino or an Arduino UNO. This
E-Bike attains maximum speed of 60kmph with power 1000w and indicators light on a jacket for safety purpose.
Ampere/hour rating of 15Ah.
The Table 1 gives review of different E-bikes available in the
market currently [5].
Keywords—BLDC, throttle, IR sensor, E-Bike/Booster Bike, Shaft,
Lilly pad LED TABLE I THE STUDY OF DIFFERENT E-BIKES.

I. INTRODUCTION Param
Our Hero Hero Hero
Implementation Electric Electric Electric
The energy from non-renewable energy sources such as fossil eters
(E-Bike) Zion Maxi Wave
fuels, coal, petroleum, and natural gas will either come to an Weight 35kg 90kg 87kg 96kg
end or will not be restored in our lifetimes. On the other side Top-
60kmph 25kmph 25kmph 25kmph
renewable energy generated from natural resources such as speed
Power 1000w 250w 100w 250w
sunlight, wind, rain, and tides and geothermal heat which are
Lithium Lithium Lithium
renewable naturally. Nowadays by considering the amount of Battery Lead acid 24v
ion 48v ion 48v ion 48v
pollution that vehicles create, and to support our PM’s Amper
15ah 20ah 28ah 20ah
“SWACHH BHARAT ABHIYAN” we are developing an e/hour
eco-friendly bike and for that we are making the use of natural Tra
renewable resource of energy which is never ending. In India ckin
g Yes No No No
environmental and noise pollution increasing daily and the
Syst
motor vehicles are one of the main reason for carbon emission em
and pollution in cities. We can’t avoid using vehicles but we Anti
can alternate the mode of use to electrical bike. And today -
with latest innovations electric vehicles have reached highest theft Yes No No No
levels of technological advancement. syst
em
They are cost effective and affordable to the middle class 35,990/- 34990/- 39,590/-
people comparing to other personal vehicles. The importance Cost 20,000/-Rs
Rs Rs Rs
of electric motor bikes is very high when comparing with the
traditional vehicle’s environmental and noise pollution. Try
an electric bike or scooter for your daily travelling and save
our environment from pollution.
 2. MANUFACTURING PROCESS AND TECHNOLOGY B. The Anti theft feature works as follows

When the cycle is locked and if a stranger tries to access the

A. Sketching cycle just by plugging in the supply, the tires will rotate. Then
This is the most productive and economical method to after one rotation the motor movement is detected by the
develop and predict any concept. We conceptually drew motor driver and the driver will send a signal to the controller.
sketch of E-BIKE on a paper and then according to the sketch
The wheel will be locked at one point of time since the motor
we decided the placement of batteries and motor on the cycle.
will not be getting power supply. After receiving the signal
Then other aspects like dimensions, space and feasibility of
from the driver the micro controller will send the signal to the
manufacturing
buzzer
B. Production
The first process was welding of the frame in which the
motor and the batteries were going to be placed. This work
had to be accurate else it would have created many problems
such as weight, space etc. After this the shaft of the wheel
was the next big step. This shaft and alignment of the wheel
was done with the help of a person with good mechanical
knowledge. After this remaining welding was done like the
frame of the personal transporter, handle etc. Along with this
process we started carpentry work. Wooden battery carriers
were made according to proper measurements of the batteries.
After this at the end we completed the controller code. For
adding features such as anti theft in micro controller 8051.

C. Testing
After assembling the entire cycle we carried out an initial
testing. Here we checked the supply voltage at the batteries.
We also ensured that wheel alignment was not changed due
to installation of the motor at the back wheel. Also we
checked the connectivity of all the wiring. Then we checked
the throttle [6] and motor throughput by increasing and
decreasing the throttle and then we finally checked the brakes
if they were able to handle the entire load of the cycle or not.

D. Painting
After all the testing we painted the cycle. Initially we applied
metal primer on the entire cycle then after two days when the Fig.. 1.Controller Block Diagram [7]
primer had dried up we painted the cycle using spray paint.
We used matt black color thus giving an attractive color to our
cycle.

3. Controller section [7]

A. Controller Description

We have used micro controller to include features like anti-

theft and support IR remote controlling. We took the hex
codes for corresponding commands via the remote and used
TSOP1437 IR sensor. Each button has a dedicated command
associated with it. Those hex codes excepted by using serial
screen of Arduino software.When the supply provided and
unlocking button is pressed the sensor used to receive the
signal and convert it into digital values which was inputted to
the controller .The controller then used to enable the port on
which motor driver circuit was connected.
C. Turn signal Bike jacket using Arduino controller [8] C. Bicycle Fork

While travelling on Bike safety is must. Turn signal indicator A bicycle fork is the part of a bicycle that holds the front
lights are implemented on jacket, these indicator lights helps wheel and allows the rider to steer and balance the bicycle. A
the vehicle behind the Bike to get an exact view that the bike fork typically consists of two blades, which are joined at the
rider is going to take a left or right. These lights are mostly top by a fork crown. The steering tube of the fork interfaces
helpful in traffic and in Mumbai the traffic is to an enormous with the frame via bearings called headset mounted in the head
level. Usually in traffic jams bike riders find their way from tube.
the narrowest portion of the road without turning on the signal
as they are useful when a whole radius turn is to be taken. The D.Bicycle Seat
other car drivers see the bike turn signal light when there is a
A bicycle saddle, often called a seat, is one the three contact
big right or left turn and during traffic jams they ignore the
points on an upright bicycle, the others being the pedals and
turn signal light as they are located at bottom portion. Thus we
the handlebars. It performs a similar role as a horse’s saddle,
are balancing this several issue which may cause accidents or
loss of life with the use of jacket, which helps the bike riders not bearing all the weight of the rider as the other contact
to easily indicate their switching positions from left to right points also take some of the load.
and there makeover, is a bit stylish because of use of LEDs. E. Drum Brake
For that we used an Arduino controller which is a lillypad
Arduino or an Arduino UNO. We used Arduino for providing The modern automobile drum brake was invented in 1902 by
delay of few micro seconds for the LEDs to blink and only one Louis Renault. A drum brake is a brake that uses friction
signals to be indicated during the particular indication with the caused by a set of shoes or pads that press against a rotating
help of switch. Here we used waterproof lillyadpad Arduino drum-shaped part called brake drum. Practically all cars use
and lillypad LEDs so the jacket can be easily washable. For disc brakes on the front wheels, and many use disc brakes on
jackets instead of using wires we used conducting threads with all wheels. However, drum brakes are still often used for
less amount of resistance because wires are not washable and handbrakes, as it has proven very difficult to design a disc
might get break. On conducting threads we applied insulating brake suitable for holding a parked car. Drum brakes allow
glue which will avoid short circuit if some conducting material simple incorporation of a parking brake.
gets stuck to the threads. We used two push button switches on
either hand of full sleeve jackets which controls total operation
of the indication. on switching the left button only left
indication light will blink similar on pressing right switch right
indicator light will blink on pressing both switch together both
indicator will blink which will indicate that biker is taking halt
or some issue with bike or he/she parking the bike.

II. MATERIAL USED FOR MAIN BODY PARTS [9]

As shown in Fig.2 some of the terms related to bicycle are
given below:
A. Bicycle Frame

A bicycle frame is the main component of a bicycle, onto

which wheels and other components are fitted. Frames are
required to be strong, stiff and light, which they do by
combining different materials and shapes.

B. Bicycle Handlebar

Bicycle handlebar or often bicycle refers to the steering

mechanism for bicycles; the equivalent of a steering wheel. Fig. 2. Layout of the bicycle
Besides steering, handlebars also often support a portion of the
rider’s weight, depending on their riding position, and provide
a convenient mounting place for brake levers, shift levers,
bells, etc
Above Table 2 gives the analysis of different parameters e.g. ACKNOWLEDGMENT
Distance covered, Maximum Speed and Charging time based
on their Ideal Result and Practically available result. The author wishes to thank Sidhesh Bhambid, Ashutosh
Gattelu, Ashwin Ghosalkar, and Steon Gonsalves for
TABLE II. RESULTS AND DISCUSSION
completing the project sincerely.

Sr
.
Parameters
Ideal Practical References
N Result Result
o.
1 Distance 50 Km 45 Km
[1] Muetze, Y.C. Tan, “Performance evaluation of electric bicycles”, in
2 Speed 55 Kmph 43 Kmph IEEE` Industry Applications conference, vol. 4, pp. 2865-2872, October
2005.
2 hour [2] [Online]Available:https://youtu.be/wZh5fBFdnVU
3 Charging Time 2 hours
15minutes [3] F.R Fuentes, G.J Estrada, “A novel four quadrant DC series motor control
drive for traction applications”, in IEEE Electric Vehicle Conference
(IEVC), pp. 1-4, March 2012.
[4] Supersonic batteries, [Online] Available: http:// www.sfsonicpower.com
[5] Hero motor corp,[Online] Available:www.Heromotorcorp.com
III.FUTURE WORK [6] E-bike school,[Online] Available:(www.ebikeschool.com/type-electric-
bicycle-throttle-best/)
[7] Micro Controller ,[Online] Available: (www.electronicsforu.com)
In this Advanced E-Bike we can have some further
improvements. As time passes by requirements of the people [8] [Online] Available:http://www.instructables.com/id/turn-signal-biking-
jacket/
go on increasing. The following are the future developments.
[9] Bicycle stock exchange ,[Online] Available:
[10] (https://bicycle.stackexchange.com/questions/244/terminology-index-a-
 Dynamometer for charging batteries list-of-bikes-part-names-and-cycling-concepts)
 Solar panels for charging batteries
 Battery level indicator
 High power PMDC motor
 Head light and tail light
 Long battery life for more efficiency
 More comfort in seating arrangement
 Light in weight material can be used

CONCLUSIONS
In the above mentioned E-bike, we used a BLDC motor as the
main driving unit. The reason for choosing BLDC motor was
simple i.e. its long life span compared to the Brush DC motor.
The brush DC motor wears out after continuous use for a long
run. The BLDC motor being brushless hardly has any chance
of wearing out, so spending 7-8K once at the beginning only,
for such reliable and durable motor does not matter much.
We have used lead acid batteries for supplying DC power to
BLDC motor, lead acid batteries and also come in compact
sizes, which made it easy for us to accommodate them on the
bicycle frame, so we didn’t face any difficulties in planning
for such batteries each of 12V is required and all batteries are
connected in series.
By using throttle i.e.an accelerator the speed of bike can be
increased or decreased which works similar mechanism like in
scooters such as Activa, also there is a provision of three gears
which synchronized with BLDC motor and to switch to proper
gear we have used a push button. In this E-Bike some added
advanced features like anti theft are provided, the micro
controller 8051 triggers an alarm as soon as it detects
unauthorized access. We also used IR sensor for accessing the
E-BIKE, and this is similar to accessing the car by remote
control. Also we have interface an LED display which will
provide vital parameters like speed of the bike, distance
covered, date and time etc.
 Modeling, Parameterization, and Benchmarking of a
Lithium Ion Electric Bicycle Battery
Weizhong Wang1, Pawel Malysz1, Khalid Khan2, Lucia Gauchia2 and Ali Emadi1
1
Electrical and Computer Engineering Department, McMaster University
Hamilton, Ontario, Canada
2
Electrical and Computer Engineering Department, Michigan Technological University
Houghton, Michigan, United States

Abstract—A lithium-ion battery from an electric bicycle Another common characterization test is operated in the
conversion kit is tested and modeled using electrochemical frequency domain using electrochemical impedance
impedance spectroscopy, and the hybrid pulse power spectroscopy (EIS) [7]. EIS can be performed in either a
characterization test (HPPC). Equivalent circuit model potentiostatic mode or galvanostatic mode at various SOC,
parameterizations are obtained from both time and frequency temperature, and age operating points. Typically the input is a
domain fitting and compared. Parameterization methods are
described and a novel quadratic programmed-based two stage
small AC voltage/current signal sweeping from 10 mHz to 1
parameter fitting algorithm is presented to process and generate kHz. The corresponding output current/voltage response is
model parameters. Experimental data is applied to the proposed measured and complex impedance is calculated. EIS data is
algorithm to assess fitting performance. The battery model is increasingly being used in state of health (SOH) assessment
validated by real-life riding cycles. Additional electric bicycle and estimation [8] [9] [10].
benchmarking tests are performed to assess real-world battery
performance under a variety of riding conditions and at different A variety of researchers present and apply their fitting results
assistance levels. The brief correlation between tiredness and of battery modeling using time-domain and/or frequency
assistance levels is investigated. domain experiments; usually exclusively either in time-domain
or frequency-domain. However, a systematic comparison and
Keywords—Battery management system; E-bike; lithium-ion, sensitivity analysis of model parameterization/fitting in both
battery; electrochemical impedance spectroscopy; pulse frequency domain and time domain, to the author’s knowledge,
characterization ; quadratic programming has not been adequately presented in the BMS design literature.
This paper focuses on this topic and uses an electric bicycle
I. INTRODUCTION application as a case study. The comparison between the two
Recently, electrified bicycles are drawing increased attention time-domain pulse-based characterization methods and EIS
from society these include Smart E-bikes and other ‘DIY’ E- frequency based fitting is explored in this paper. Equivalent
bikes. One of the core parts is the battery since it determines circuit model parameters fitted from the three methods will be
range and power capability. For lithium ion batteries a battery presented and compared. Matlab optimization fitting
management system (BMS) is essential to maximize algorithms are developed and employed using Fmincon,
performance and ensure safety [1] [2]. The design of the BMS nonlinear least squares, and quadratic programming (QP). A
heavily uses battery models in both offline development and novel fitting algorithm strategically using QP is highlighted in
online control and estimation [3]. One of the most popular and the paper to show faster fitting with improved capability to
practical approaches is the use of an equivalent circuit model produce globally optimal fitting results.
(ECM). The models generated from characterization based testing will
The ECM consists of electric circuit components, such as be validated using data generated from additional
resistors and capacitors, trying to accurately reflect the voltage benchmarking tests to assessment performance of the e-bicycle.
response under dynamic current in time domain and impedance These will include so-called user specific ‘ride-cycles’, range
response in frequency domain. Since the internal impedances tests, power output employing varying levels of rider effort,
are dependent on state of charge (SOC), aging and temperature, and a brief definition of least power assistance to eliminate
the parameters can be expressed in either empirical equations tiredness from riding.
[4] [5] or as lookup tables. The time domain model The main contributions of this paper are 1) time-domain and
parameterization is normally captured by two test procedures frequency domain experimental characterization and modeling
[6], Hybrid Power Pulse Characterization (HPPC) test of an electric bicycle battery, b) development and comparison
employing constant current pulses, and maximum energy pulse of time-domain and frequency domain model parameter fitting
based characterization test; the latter employs a combination of algorithms, and c) benchmarking results of an electric bicycle.
constant current and constant voltage pulses. Both test profiles The paper is organized as follows: section II outlines the
incorporate charging and discharging pulses to determine the experimental test plans and details. In section III, the details of
dynamic power capability and generate resistance parameters the characterization test procedures and the time/frequency
over the battery operating range.

978-1-5090-0737-0/16/$31.00 ©2016 IEEE

domain models are described. Section IV, presents the wirelessly. It can be utilized for other purposes, e.g. modeling
proposed fitting/parameterization algorithms. Benchmarking the E-bike and the friction,
and fitting results, discussion are given in Section V. This
The characterization tests are done on a new/fresh battery,
paper will be concluded in Section VI.
commonly referring to as at the beginning of life (BOL)
battery SOH. Only HPPC tests and EIS tests are performed.
II. EXPERIMENTAL METHODS Battery characterization experiments were done at the battery
testing lab at Michigan Technological University. The
experimental set-up is shown in Fig. 2. The time domain tests,
350 W high including capacity verification test and HPPC test, are
torque motor performed on the battery tester. The EIS test is done by the
Impedance analyzer. The battery is tested in an environmental
chamber in ambient temperatures ranging from -10 °C to 25
°C and across the full SOC range. In addition to battery only
testing, electric bicycle tests are performed to collect real-
48 V, 8.8 Ah world ride-cycle under different levels of rider effort including
high volume
zero rider effort, light rider effort, and heavy rider effort. It is
battery pack
(a) realized by different levels of battery/e-motor output levels,
i.e. 0%, 25%, 50%, 100% and 200% throttle levels. In each
case the range traveled, speed and profiles of battery SOC,
energy, and power are generated as used in the benchmarking
analysis. The electric bicycle with its battery is tested at 25 °C
to assess summer condition where typically people start to ride
bikes. An expanded detailed explanation of the experimental
test procedures and test case matrices are provided in the next
section.

(b)
Monitor of data
acquisition system

Original BionX
controller

(c)
Fig. 1. (a) The electricified bike (b) Opened battery pack (c) Data acquisitaion
system.
Fig. 2. Experimental setup for battery characterization.
The electric bicycle and its battery are depicted in Fig.1 (a)
and (b). A conventional bicycle is used with an electrifying
conversion kit that includes a rear wheel 350 W high torque III. TEST PROCEDURES AND MODELING
electric motor and a 48 V 8.8 Ah battery with BMS from
BionX. Unlike many conversion kits, the one employed has A. HPPC Test
regenerative braking for energy recapture. Additional custom The HPPC test is intended to determine dynamic power
voltage, current, and temperature sensing hardware and cycle capability as well as identify parameters. The procedure
analyst electronics from Grin Technology are installed for the applies 10-40 s constant current discharge and regenerative
purpose of collecting real-world data and benchmarking the (charging) pulses separated by rest periods, at each SOC level
electric bicycle, as shown in Fig. 1. The speed and cadence and temperature of interest. At the same time, voltage
sensor is purchased from Wahoo Fitness and installed on the response will be recorded [6]. Typical desired responses are
rear wheel to collect accurate real-life speed and cadence. It shown in Fig. 3 (a). Ideally, the voltage response is desired to
features with a wireless connection which enables data be within voltage limits and not be clipped. Some level of
exporting and location capturing. So that the speed profile can guesswork is used to determine the magnitude of the constant
be recorded for defining ‘riding cycles’ along with current and current pulses, particularly at the extreme operating points.
voltage from the battery pack. In addition to speed and Usually, HPPC employs conservative current pulses.
cadence, the grade can also be monitored and restored
Fig. 3 (b) illustrates the time domain battery ECM model and frequencies while real part maintains constant (value of Ohmic
how certain elements relate to measured voltage responses. resistance) [13]. This inductive behavior is essentially the self-
The blue line denotes the voltage response from a constant inductance of the conductive loop formed by the terminals,
current pulse. It can be derived that the battery ECM model connectors, and electrodes of the battery cell [15]. It can be
can be divided into two responses: instantaneous voltage modeled by a lumped constant inductance L.
drop/increase excited by ohmic resistance, and dynamic
voltage decay/growth triggered by RC pairs [11]. The number
of RC pairs can be chosen as many as desired to achieve
sufficient accuracy [12]. In this study, 2-RC architecture is
selected to balance the complexity of the circuit and accuracy.

Constant Current

1 Current
Voltage

0
HPPC

Minimum Voltage Fig. 4. Frequency Domain Equivalent Circuit Model.

Time
(a) 2) ZARC Elements
The semi-circles in Fig. 4 are interpreted to so-called ZARC
Ca Cb Cn
Ro elements. ZARC elements consist of a parallel pair of a charge
...
Ra Rb Rn transfer resistor and a constant phase element (CPE). ZARC is
beneficial to model the depression behavior, which is the
characteristic for Lithium-ion cells [16]. However, due to lack
of Laplace transformation, ZARC elements is not able to be
Voc
transferred in time domain without approximations [17].
A pure capacitor is a specialized CPE with depression factor Į
is 1, according to the definition in (1).
1
Z CPE = (1)
C ( jω )α
(b)

Fig. 3. (a) Diagrams of voltage response and current pulses for HPPC. (b)
where C indicates a generalized capacity. The depression
Time Domain Equivalent circuit model. factor Į with a range between 0-1 reflects the depression of
the semi-circle in EIS curve depicted in Fig. 3. The ZARC
elements can be described as follows:
B. EIS Test
Fig. 4 shows the typical shape of an EIS test at some operating 1
Z ZARC = (2)
point, the x-axis is the real part of the complex impedance and 1 R + C ( jω )α
the y-axis (typically inverted) the imaginary part of the
complex impedance. 3) Warburg
It is common to observe a linear slope with an angle of 45º at
1) Ohmic Resistance and Inductance
low frequencies in EIS curves, which is expressed as Warburg
The internal resistance always exists in batteries, which is impedance. Warburg impedance at the low frequency end is
caused by several factors in chemistry. The factors are, for used to characterize the diffusion process in the batteries [16].
example, the conductivity of the anodes and cathodes and the There are various expressions to describe Warburg Impedance
electrolyte [13]. Due to the thermal-sensitivity of these [13]. In this paper, the simplest and commonly-used one is
chemical components, the Ohmic resistance is typically highly applied, shown in (3). For more details, readers please refer to
dependent on temperature. Besides, it has been found the [13].
SOC, charging/discharging history and aging dramatically W
affect the changes of the resistance [14]. It has been captured ZW ∞ = (1 − j ) (3)
that the imaginary part continuously increases at high ω
where, j is the unit of imaginary numbers. ω refers to the efficient globally optimal QP solvers can be readily employed,
angular frequency and W denotes the parameter to fit. The for example MATLAB quadprog.
details of fitting algorithms for both time and frequency
domain will be explained in next section. min xT Qx + p T x
2
min ¦ k =1 (Vk − Vˆk ) = min(Cx − v )T (Cx - v ) (5)
n
C. Electric Bicycle Benchmarking
Since the electric bicycles are designed to assist people in their = min(xT (CT C)x − 2( v T C)x + v T v )
commuting and exploring trips, aspects such as how long the
battery lasts and how much energy is saved from riding the E- where v=[V1 V2 ..VN]T is a vector of N measured terminal
bike are two key criteria. The electrical bicycle features voltages, and Vˆt is modelled as
regenerative braking and different levels of electric powered
assistance from the motor. The ability of regenerating power Vk = VOC − Ro I k − Ra I a , k − ... − Rn I n , k = cTk x (6)
as well as assistance levels benefits prolonging the battery life. 
n RC
The assistance level range from 0% - 200%, this refers to the
different amounts of power driving the motor. Due to high where, ck T = ª¬1 − I k − I a,k ... − I n , k º¼ , and
uncertainties of various riding styles, the battery life will be
highly dependent on rider behavior. As a result, the range tests x = [Voc Ro Ra ... Rn ] . Voc refers to the open circuit
T

are only performed on electric only riding.

voltage, which is a function of SOC. Ik is the cell current at
In order to quantify the saved energy and tiredness from riding time step k. The current Ii,k through resistance on ith RC pair is
based on different assistance levels, the same trip is repeated derived as follows [18]
five times at five assistance levels. As a reference, the energy
−Δt −Δt
in the pure electric mode is approximated the same as the Ii,k = e τi
I i , k −1 + (1 − e τi
) I k −1 (7)
manpower consumes without assistance. This assumption will
be valid when the proportion of manpower is interested, as Note that IJi is pre-optimized in outer-loop.
shown in (4), since the efficiency on motor and transmission is
counterbalanced. In frequency domain, measured real part and imaginary part
are compared with fitted ones to minimize the difference, as
Et , human = Et ,batteryη shown in (8). x = [ X 1 X 2 ... X m ] and y = [Y1 Y2 ... Ym ] are
T T

Ei ,batteryη + Ei , human = Et , human the sets of all real part and imaginary part data at single SOC
Ei , human level. Xˆ k = cTk ș, Yˆk = dTk ș with linear regressed parameters
Percentage = (4)
Et , human vector ș = ª¬ Ro R1 R2
... Rn Wg º¼ . Vectors ck and dk
T

Et ,batteryη − Ei ,batteryη are functions of not only frequency but also time constants and
=
Et ,batteryη depression factors. A set of ck and dk over the whole frequency
C = [c1 c 2 ... c m ]
T
Et ,battery − Ei ,battery range forms the matrices
=
Et ,battery and D = [d1 d 2 ... d m ] . Where m stands for the number
T

where, Et and Ei,battery refer to total consumed energy and of frequency points.
consumed energy from battery at ith assistance level,
(( ) + (Yˆ − Y ) )
n 2 2
respectively. η is the efficiency of the motor and min ¦ Xˆ k − X k k k
k =1
transmission. The details and results will be shown in Section
V. At the same time, the correlation between the tiredness and (
= min ( Cș − x ) ( Cș − x ) + ( Dș − y )
T T
( Dș − y ) ) (8)
riding effort will be also established.
(
= min șT ( CT C − DT D ) ș − 2 ( xT C + y T D ) ș + xT x + y T y )
IV. BATTERY PARAMETERIZATION
According to Euler’s formula and polar form for fractional
This section describes the fitting algorithms developed and
powers of complex numbers, (re jφ )α = r α e jφα . The impedance
employed on the battery characterization data. A two-stage
optimization algorithm is employed where an inner-loop of ECM in Fig. 4 can be derived as (9).
optimization assumes that the IJi = RiCi time-constants are held Constraints in QP algorithm can be chosen to assign known
constant. Using such a simplification, the inner-stage fitting values, e.g. open circuit voltage based on SOC operating
can be formulated as a QP. The outer-loop optimization point, parameter smoothness, non-negativity and bounded
searches a reduced dimensional space to find optimal values of value constraints. The QP finds optimal values for a given
the time constant parameters. This approach is applicable to time-constant value.
both time-domain and frequency domain data and
characterization. It greatly speeds up the fitting routine since
Zˆ k = Xˆ k + jYˆk = c kθ + jd kθ significantly decrease the proportion of manpower, leading to
ª º
T
less tiredness.
« π π »
« 1 + τ 1ωkα1 cos( α1 ) 1 + τ nωkα n cos( α n )
2 2 1 » However, energy-wise speaking, these two consumed
c k = «1 ... »
« 1 + 2τ ω α1 cos( π α ) + τ 2ω 2α1 τ ω αn π
α 2 2α n ωk » approximately 1/10 total energy of the battery. To search for a
1 + 2 cos( n ) + τ n ωk
«  1 k 2
1 1 k n k
2 »
«¬ n ZARC elements »¼ balance between consumption and comfortability, Level 3
ª º
T with around 50% manpower is optimal in terms of energy and
« π
−τ 1ωkα1 sin( α1 )
π
−τ nωkα n sin( α n )
» tiredness. In principal there should be an exact threshold of
« 1 »
d k = «0 2 ... 2 » manpower optimizing the feeling and consumed energy.
« 1 + 2τ ω α1 cos( π α ) + τ 2ω 2α1 π
1 + 2τ nωkα n cos( α n ) + τ n2ωk2α n ωk » Speed
1 k 1 1 k
«  2 2 » 40
¬« n ZARC elements ¼»
30
(9)

Km /h
20

V. RESULTS AND DISCUSSION 10

1) E-bike Benchmarking 0
0 1 2 3 4 5 6 7 8
The range tests are performed in real-life riding with urban Time(mins)
road condition and constant load on the electric bicycle (200 Grade
10
lbs). In order to eliminate the uncertainties of the riding style,
this test is repeated three times to obtain the average result.

Degree
5
After every fully charging, the distance until the battery
automatically shut off was recorded. The three tests are 0

summarized in the following table. Note that the range test is

-5
based on pure electric mode without manpower. The average 0 1 2 3 4 5 6 7 8
range is around 30 km per full charge. Time (min)
(a)
TABLE 1 RIDING RANGE TEST
Test #1 Test #2 Test #3 Avg.
V

29.2 km 31.3 km 30.5 km 30.34 km

The second benchmarking test is to investigate a) the

relationship between riding effort and tiredness, b) the
A

consumed energy at each assistance level. The test is

performed in the same route, i.e. the distance and road
condition are approximately identical. Trip information at
W

200% assistance level, including mechanical and electrical

data, is recorded by sensors, as shown in Fig. 5. The grade is
an important factor for other research areas, e.g. systematically
Ah

modeling the E-bike and the friction. The energy and SOC
consumptions and proportion of manpower can be extracted
from Fig. 5 as well as other assistance levels. They are
%

summarized and compared in Table 2. The ‘Pure Assist’

column is assumed as pure manpower to be compared with,
explained in Section III. Additionally, tiredness is assessed (b)
from the feeling of testers. The higher percentage refers to Fig. 5. Collected riding cycle: (a) Speed and grade profiles, (b) Electrical
approaching nonassistant bikes which consume manpower. profiles.
TABLE 2 IMPORTANT INDICATORS AT EACH ASSISTANCE LEVEL 2) Fitting results
Pure
Assistance level Level 1 Level 2 Level 3 Level 4
Assist
The HPPC test is performed between upper limit and lower
Energy from battery limit of terminal voltage at every 5% SOC intervals. The EIS
7.50 10.87 15.89 26.55 33.94
(Wh) test is carried out at five SOC points (20%, 40%, 60%, 80%
Manpower (%) 77.9 67.97 53.18 21.77 0 and 100%) with no DC bias current. The both tests are
SOC consumption (%) 1.89 2.81 4 6.8 9.09 performed at -10, 0 and 20 °C.
Tiredness (%) 90 70 40 20 0
The proposed QP fitting algorithm is used to process both
It can be observed from Table 2 that higher assistance levels experimental results. The time domain fitting results are
offer more energy from battery/motor. The lower manpower shown in Table 3. The difference of measured terminal
engaged decrease the tiredness. ‘Pure Assist’ and Level 4 can voltage and fitted terminal voltage is analyzed. The mean error
at three temperatures is less than 3 mV ,which shows the
ability to accurately fit the battery dynamic response. The
detailed parameters are shown in Fig. 7. Only resistances are

Ohms
illustrated because the EIS ECM and HPPC ECM have the
same meanings on theses parameters. Others, e.g. CPE which
cannot be modeled in time domain, are excluded. That is, it is
comparable on resistances instead of CPE and Warburg
impedance.

Ohms
TABLE 3 RESULTS OF HPPC FITTING ALGORITHM

Temperature (°C) -10 0 20

Mean Error (mV) 2.952 2.403 2.053

Ohms
Max Error (mV) 49.10 41.76 11.85

Fig. 7. Comparison on internal resistances extrated from EIS and HPPC tests.

3) Model Validation
A 34-minute long riding cycle collected from the data
acquisition system is used to validate the proposed 2-RC
battery model with fitted parameters from HPPC test. The
pack voltage profile has been scaled down to module level
while ignoring cell-to-cell balancing and unbalanced
parameters in cells. The 20 °C parameters are applied to the
battery model since real-life riding is performed at ambient
(a)
temperature (around 25 °C). The modeled and measured
voltage responses are compared in Fig. 8. It can be seen that
the model can dynamically predict the real voltage response
with 25 mV maximum error and 2.73 mV mean error.

(b)
Fig. 6. Example of EIS fitting results, (a) fitted EIS and measured EIS, (b)
fitting errors.

The EIS curves are fitted by QP based algorithm. One of the

samples is shown in Fig. 6 to reflect the performance of the
fitting tool. Amplitude and phase angle errors are depicted as
well. They are more suitable to assess a Nyquist plot since the Fig. 8. Model validation using E-bike ride cycle data.
frequency information is included. It can be seen from Fig. 6
that the amplitude and phase angle errors are less than 0.05 dB VI. CONCLUSSIONS
and 0.1° (1% error), respectively.
Fig. 7 shows the parameterization results for both EIS and A novel QP based parameterization algorithm which can be
HPPC test on Ro, Ra, Rb to compare the consistency of the employed in both frequency and time domains is proposed. A
results from both tests. At 20 °C, Ro shows significant 48V battery pack from BionX E-bike conversion kit has been
agreement on both tests. For 0 and -10 °C, Ro from HPPC tests parameterized by HPPC and EIS tests. The experimental
is expected. The trend is that the Ohmic resistance grows with results are used to validate the proposed algorithm. Less than 3
the temperature decreases, which has been observed by many mV mean error on HPPC fitting and 1% mean error on EIS
researchers [19] [20] [21]. However, Ro from EIS test shows fitting were observed. The parameters from HPPC are input to
an insensitivity to temperature. The reason remains unknown. 2-RC ECM to verify the accuracy of fitting results and the
It will be investigated in a future study. battery model. The battery model is excited by a measured
current profile from a pre-defined and real-life riding cycle. It
has been proved that the model can precisely reflect the real Batteries,” ECS Trans., vol. 33, no. 39, pp. 41–53, 2011.
[10] D. a Howey, P. D. Mitcheson, S. Member, V. Yufit, G. J. Offer, and N.
battery’s dynamic response. Two benchmarking tests are P. Brandon, “Online Measurement of Battery Impedance Using Motor
performed on the electric bike system: a) the range tests in Controller Excitation,” IEEE Trans. Veh. Technol., vol. 63, no. 6, pp.
pure electric mode, b) the relationship between the manpower 2557–2566, 2014.
and tiredness has been approximately quantified. [11] R. Xiong, H. He, H. Guo, and Y. Ding, “Modeling for Lithium-Ion
Battery used in Electric Vehicles,” Procedia Eng., vol. 15, pp. 2869–
ACKNOWLEDGMENT 2874, 2011.
[12] H. He, R. Xiong, H. Guo, and S. Li, “Comparison study on the battery
This research was undertaken thanks to funding and support models used for the energy management of batteries in electric
from the Canada Excellence Research Chairs (CERC) vehicles,” Energy Convers. Manag., vol. 64, pp. 113–121, 2012.
[13] E. Karden, “Using low-frequency impedance spectroscopy for
program. characterization , monitoring , and modeling of Industrial batteries,” no.
May, 2001.
REFERENCES [14] G. Liu, M. Ouyang, L. Lu, L. Xu, and J. Li, “Online monitoring of
[1] A. Emadi, “Transportation 2.0,” IEEE Power Energy Mag., vol. 9, no. 4, lithium-ion battery aging effects by internal resistance estimation in
pp. 18–29, 2011. electric vehicles,” Control Conf. (CCC), …, pp. 6851–6855, 2012.
[2] B. Bilgin, P. Magne, P. Malysz, Y. Yang, V. Pantelic, M. Preindl, A. [15] B. Savova-Stoynov and Z. B. Stoynov, “Analysis of the inductance
Korobkine, W. Jiang, M. Lawford, and A. Emadi, “Making the Case for influence on the measured electrochemical impedance,” J. Appl.
Electrified Transportation,” IEEE Trans. Transp. Electrif., vol. 1, no. 1, Electrochem., vol. 17, no. 6, pp. 1150–1158, 1987.
pp. 4–17, 2015. [16] D. Andre, M. Meiler, K. Steiner, H. Walz, T. Soczka-Guth, and D. U.
[3] P. Malysz, L. Gauchia, and Y. Hong, Fundamentals of electric energy Sauer, “Characterization of high-power lithium-ion batteries by
storage systems. 2014. electrochemical impedance spectroscopy. II: Modelling,” J. Power
[4] M. Chen and G. A. Rincon-Mora, “Accurate Electrical Battery Model Sources, vol. 196, no. 12, pp. 5349–5356, 2011.
Capable of Predicting Runtime and I–V Performance,” IEEE Trans. [17] E. Barsoukov and J. R. Macdonald, Impedance Spectroscopy. 2005.
Energy Convers., vol. 21, no. 2, pp. 504–511, 2006. [18] R. Xiong, H. He, H. Guo, and Y. Ding, “Modeling for lithium-ion
[5] D. Fregosi, S. Bhattacharya, and S. Atcitty, “Empirical battery model battery used in electric vehicles,” Procedia Eng., vol. 15, no. 1, pp.
characterizing a utility-scale carbon-enhanced VRLA battery,” 2011 2869–2874, 2011.
IEEE Energy Convers. Congr. Expo., pp. 3541–3548, 2011. [19] D. Andre, M. Meiler, K. Steiner, C. Wimmer, T. Soczka-Guth, and D.
[6] The Idaho National Laboratory, “Battery Test Manual For Electric U. Sauer, “Characterization of high-power lithium-ion batteries by
Vehicles,” no. June, 2015. electrochemical impedance spectroscopy. I. Experimental investigation,”
[7] E.-H. Aglzim, M. Bin Jamaluddin, D. Chrenko, and a. Rouane, J. Power Sources, vol. 196, no. 12, pp. 5334–5341, 2011.
“Impedance Spectrometer Modelling in Matlab/Simulink for Measuring [20] J. P. Schmidt, S. Arnold, A. Loges, D. Werner, T. Wetzel, and E. Ivers-
the Complex Impedance of a Fuel Cell –EIS Method,” J. Clean Energy Tiffée, “Measurement of the internal cell temperature via impedance:
Technol., vol. 1, no. 4, pp. 255–259, 2013. Evaluation and application of a new method,” J. Power Sources, vol.
[8] D. I. Stroe, M. Swierczynski, A. I. Stan, V. Knap, R. Teodorescu, and S. 243, pp. 110–117, 2013.
J. Andreasen, “Diagnosis of Lithium-Ion Batteries State-of-Health based [21] L. W. J. Uang, “Improved Modeling of Lithium-Based Batteries using
on Electrochemical Impedance Spectroscopy Technique,” pp. 4576– Temperature-Dependent Resistance and Overpotential.”
4582, 2014.
[9] R. Mingant, J. Bernard, V. Sauvant-Moynot, D. a., M. S., H. J-L., and H.
F., “EIS Measurements for Determining the SoC and SoH of Li-Ion
 E-bike battery charging: methods and circuits
F. Pellitteri**, V. Boscaino**, A.O. Di Tommaso*, F. Genduso* and R. Miceli*
*Sustainable Development and Energy Saving Laboratory, Department of Energy, Information engineering and
Mathematical models, University of Palermo, viale delle scienze, building 9, 90128, Palermo, (Italy)
**Power Applications Laboratory, Department of Energy, Information engineering and Mathematical models,
University of Palermo, viale delle scienze, building 9, 90128, Palermo, (Italy)

Abstract— Today, academic and industrial research is Today, several leading firms are investing on efficient
focused on innovative battery charging methods to ensure energy harvesting solutions for handheld devices.
complete mobility of both handheld devices and electric Whereas in the past, wireless sensor networks and
vehicles. Wireless power transfer is actually the leading
strategy even if efficiency related issues are to be solved for
biomedical applications were the main application areas,
a successful marketing. In this paper, a wireless battery today multi-sourcing techniques are investigated to
charging station is proposed for electric assisted pedal bikes. provide efficient battery charging in handheld devices.
If compared with existing wireless solutions, the proposed Supplying the standby power by harvesters is one of the
system architecture improves power conversion efficiency of most attractive research topics.
the charging equipment. The simulation model of the whole In the past, main investments on power electronics
charging station is described in detail. The transmitter,
receiver and inductive coupling circuits are described and
research were focused on speeding up the transient
design criteria are given for further applications. The model response improving steady-state and dynamic load
is implemented in SPICE simulation environment and regulation keeping as a side effect the improvement of the
simulation results are presented to test the efficiency of the power conversion efficiency [20-21]. Nowadays, the
proposed topology. power conversion efficiency plays a key role in power
supply system design to accomplish with modern trends
Index Terms—wireless power transfer, wireless battery
in energy saving. In the literature, several innovative
charging, inductive power transfer, vehicular and wireless
technologies. solutions to efficiency related issues are proposed for a
wide variety of research areas, from front-end to point of
I. INTRODUCTION load converters [22-24].
Today, consumer and automotive markets call for Due to increasing oil price and global warming, the
high-performance devices. Ideally, an infinite source of electrical traction of transportation means represents a
energy is desired from the end-user. True mobility is the convenient solution in terms of cost and environmental
most attractive feature of both handheld devices and cleanness. Therefore, the use of electric power is
electric vehicles [1]. Academic and industrial research is particularly compliant with urban mobility, since the
actually focused on accomplishing true portability while distances covered by vehicles are relatively short inside
keeping high performances of devices. True portability cities. Among urban electric vehicles, bicycles are
while saving cost of energy from the electrical grid is the certainly the least noisy and polluting ones. Moreover,
key challenge to face with. In the last years, great efforts electric bikes are cheaper than the other electric vehicles.
have been made to both improve the power conversion In the electric bicycle, so-called “E-bike”, a part of the
efficiency of the whole charging station and integrating in classical energy source, that is muscle power, is replaced
the charging equipment low-emission renewable energy by electricity, providing the possibility of an electric
sources, thus reducing the amount and cost of energy assisted pedal [25-26].
drawn from the electrical grid [2-9]. In the near future, a rising number of electric vehicles
For years, powering handheld devices by renewable should move through the urban roads. In a similar
energy sources like fuel cells or micro photovoltaic scenario, new battery recharge stations will be necessary
generators have been the leading research topic [10-14]. in order to offer all vehicles the amount of electric energy
In the last few years, European and worldwide research which is needed for their traction. An electric vehicle
programs have been dominated by fuel cell and battery is commonly recharged through the traditional
photovoltaic based applications. Accurate modeling of “plug-in” method that is a wired connection between the
renewable energy sources is still an acknowledged open vehicle and a charging column which is linked to the
topic to shorten the overall design process [15-19]. electrical grid. Despite its simplicity, this charging
method turns out to be uncomfortable for the driver, who
has to plug the cable into the socket, and unsafe, due to
This publication was partially supported by the PON PON04a2_H the electrocution danger arising from old power cords and
"i-NEXT" italian research programme. This work was realized with the even more likely in wet conditions. Furthermore, the
contribution of SDES (Sustainable Development and Energy Savings)
Laboratory - UNINETLAB - University of Palermo. presence of wires could bother, especially when a lot of

978-1-4673-4430-2/13/$31.00 ©2013 IEEE 107

small vehicles are parked very close to each other, as in efficiently converts the AC signal into a 48V DC bus
parking lots. For this purpose, wireless battery charging is level; a battery charger, which controls the flow of charge
actually investigated for automotive applications also. from rectifier to battery; the E-bike LiFePO4 battery.
Nevertheless, the traditional energy sources still The following design procedure has been undertaken.
provide better performances in terms of vehicle autonomy According to the envisioned spatial constraints referring
and power conversion efficiency of the whole charging to an E-bike battery wireless recharge, the magnetic
equipment. Conventional fuels or grid-connected wired coupling structure has been selected and analyzed
charging still ensures higher performances. through magnetic simulation software. From the magnetic
In the literature, several innovative solutions for simulator analysis, inductive and resistive parameters
wireless power transfer which ensure complete mobility values have been obtained and consequently included in
are proposed [27-29]. Yet, if compared with their wired the circuit simulation model of the whole charging
counterparts, power conversion efficiency is relatively equipment. Based on the selected coil length, wire section
low. Even if wireless technology is already mature, the and material, skin effect losses versus operating
loss of a few percentage points of power conversion frequency have been modeled too.
efficiency is still the major hindrance to a successful In order to obtain the nominal 48V rectifier output
spread on the market. Since extremely low power levels voltage together with a high coupling efficiency value, a
are involved, handheld applications are less burdened by fixed and resonant operating frequency control has been
the loss of efficiency than automotive applications. implemented. Power load regulation is achieved by
Whereas prototypes of wireless battery chargers for varying the duty-cycle of the inverter square wave.
handheld devices are already available on the market,
wireless charging for electric vehicles is still not suited
for marketing.
Research is therefore focused on efficiency related
issues aiming at ensuring the highest mobility while
reducing consumption from the electrical grid.
In this paper a wireless solution of E-bike battery
recharge is proposed. Power transfer from grid to battery
is achieved through magnetic coupling: a coil, connected
to mains, transmits energy towards another coil, placed
upon the bike and linked to its rechargeable battery. Fig. 1. Proposed E-bike recharging system.
Cyclers should only park their bike and recharge will start
automatically, without needing to use bulky and Load-connected subsystem
dangerous power cords. In the proposed system simulation model, the battery
charger and the rechargeable battery are considered as the
actual load, so that the rectifier is the main power receiver
II. E-BIKE CHARGING SYSTEM
part which is widely described in this paper.
Fig. 1 shows a block diagram of the proposed In order to evaluate the system efficiency, the rectifier
recharging system, compliant with an E-bike battery. input resistance shall be evaluated. At the resonant
From the left to the right, the following parts are shown: frequency, the equivalent load resistance (Rr) of the
the grid-connected subsystem, also referred to as power resonant network is given by:
transmitter; the resonant network, including a magnetic
8
coupling structure; the load-connected subsystem, also Rr RL (1)
referenced as power receiver, which supplies the final S2
rechargeable battery. Nowadays LiFePO4 batteries are where RL is the actual load resistance as seen by the
typically employed for E-bike supply. The proposed rectifier.
recharge system complies with a 36V 10Ah LiFePO4 Considering a 48V DC bus level and a 2A load
battery, whose dimensions are 64*148*365mm. The current, RL value is equal to 24ȍ. Therefore, according to
battery voltage lies in the range 30Vy43.8V. The (1), Rr is given by (2):
designed system envisions an about 100W recharge.
8
The grid-connected subsystem includes an AC-DC Rr ˜ 24 # 19.4: (2)
stage, which converts the grid AC low frequency voltage S2
into a 72V DC voltage, and an inverter, which converts An active full-wave rectifier bridge has been preferred
the DC voltage into a high frequency AC signal. to the conventional four-diode configuration, in order to
Due to the magnetic coupling inside the resonant enhance power conversion efficiency. Indeed, since each
network, a wireless power transfer occurs from power MOSFET features a very low conduction resistance,
transmitter to power receiver, consisting of this AC signal power losses are less than using the traditional passive
flowing towards the load-connected subsystem. full-wave rectifier, even in the most efficient Schottky
The power receiver includes: an active rectifier, which diode case.

108
 The proposed rectifier, shown in Fig. 2, is A. Resonant network
implemented through an innovative “cross-coupled” The wireless power transfer from the grid-connected to
topology and a proper driving network, consisting of two the load-connected subsystem is achieved through
comparators (comp1-comp2) and two not logic gates magnetic coupling. Fig. 3 shows the SPICE model of the
(inv1-inv2). The cross-coupled configuration consists of resonant network, where self-inductances and equivalent
four active switches: two enhancement pMOS (M1-M2) resistances values are obtained from the proposed
and two enhancement nMOS (M3-M4). The rectifier AC magnetic analysis.
input signal, corresponding to the resonant network
output signal, is applied between M1 and M2 gate
terminals. The DC rectified output is sensed between M1-
M2 common source node and ground. M1 and M3 drain
terminals are connected, as well as M2 and M4 drain
terminals. M1 gate terminal is connected to M2-M4 drain
terminals; M2 gate terminal is connected to M1-M3 drain
terminals. All four gate MOSFETs don’t require any
external signal to be driven by: pMOS gates are driven by
the input AC signal polarities (-AC and +AC), while
nMOS ones are driven by two signals which are obtained
by processing the AC input signal through the driving Fig. 3. SPICE model of the resonant network.
network. The comparators convert the analog signals –
AC and +AC into two digital signals, B1 and B2: when – The magnetic coupling is represented by a mutual
AC (+AC) is positive to ground, B1 (B2) corresponds to inductance: L1 and L2 are primary and secondary self-
logic 1; when –AC (+AC) is negative to ground, B1 (B2) inductances, connected to the source and the load side
corresponds to logic 0. The not gates have not only a respectively; k is the coupling coefficient.
logic function, but also work as drivers, thus providing Primary and secondary self-inductances and their
the necessary gate currents to both the nMOS. Therefore, equivalent DC resistances have been acquired from
when the AC signal is positive, G3 and G4 correspond to magnetic simulation software femm 4.2.
logic 1 and 0, thus turning M3 on and M4 off; since M3 Two identical planar circular coils have been
is on, M2 gate terminal is tied to ground connections, introduced in the magnetic simulation file, as shown in
thus driving M2 into conduction. When the AC signal is Fig. 4. They represent power transmitter and power
negative, G3 and G4 correspond to 0 and 1, thus turning receiver coils and are distant 3cm from each other. Each
M3 off and M4 on; since M4 is on, M1 gate terminal is of them consists of 9 copper wire turns. Outer and inner
tied to ground connections, thus driving M1 into coil diameters are respectively 15cm and 8.8cm. Fig. 4
conduction. shows the axi-symmetric sight of the problem.
In brief, during the input AC signal positive half-wave, In order to build a more accurate mutual coupling
the peak voltage is rectified by M2 and M3 while M1 and electric model, skin effect losses have to be considered
M4 are off; during the negative half-wave, the peak too. Considering a 3mm wire diameter section, Fig. 5
voltage is rectified by M1 and M4 while M2 and M3 are shows the skin effect losses (Rac) vs. frequency (f) curve,
off. Subsequently, the same behavior of the traditional related to each coil, according to the following equation:
four-diode configuration is shown by the proposed active Ul
Rac (3)
rectifier, with the substantial benefit of an improved § 
r ·§ § 
r ··
efficiency. SG ¨ 1  e G ¸ ¨ 2r  G ¨ 1  e G ¸¸
¨ ¸¨ ¨ ¸¸
© ¹© © ¹¹
where ȡ is the resistivity of the conducting material, equal
to 1.68x10-8ȍÂm in the copper case; l is the wire length; r
is the radius of the round conductor; į is the skin depth,
which is given by:

U
G (4)
SP0 P r f
where —o is the permeability of free space, equal to
4ʌx10-7 HÂm-1; —r is the relative permeability of the
conducting material, which in the copper case is very
close to 1.

Fig. 2. Active rectifier network.

109
 and secondary side f1 and f2, according to (5) and (6):
1
f1 (5)
2S L1C1
1
f2 (6)
2S L2C2
As it’s been explained above, if the inverter MOSFETs
switching frequency is equal to the resonance frequency,
Fig. 4. FEM-based model of the magnetic structure. the double goal of a maximum power factor and a
maximum transfer power is obtained.
Therefore it’s convenient to choose both the resonance
frequencies equal to each other, in order to achieve both
the goals.
Provided that the system operates under the common
primary-secondary resonant frequency, the following
frequency-dependent efficiency relationship would be
useful to evaluate best-suited frequencies [32]:
RL
K (7)
§ R R  RL ·
RL  R2 ¨1  1 2 ¸
¨ 2S f M 2 ¸
© ¹
where R1 and R2 are the coils equivalent resistances,
including both DC and AC components, RL is the load
resistance seen by the resonant network, f is the operating
resonant frequency, M is the mutual inductance
coefficient, linked to the self-inductance and the coupling
coefficient according to (8).

M k L1L2 (8)

Fig. 5. Skin effect losses vs. operating frequency curve for the proposed As shown by Fig. 6, efficiency increases with
magnetic coupling structure. frequency. Therefore, an high frequency value is
desirable to achieve at least a 90% coupling efficiency.
The curve has been generated by MATLAB elaboration Nevertheless, too high frequency values are not
of equations (3) and (4), where the values have been fixed suggested for these power levels. A 100 kHz resonance
according to the proposed magnetic structure. frequency has been selected, thus bringing to a 92.1%
Two “compensation” capacitors are required in source coupling efficiency when working under the same
and load sides in order to respectively minimize the frequency.
amount of reactive power drawn from the supply and Fig. 6 refers to a fixed 3cm distance between
maximize the amount of transmitted power to the load- transmitter and receiver coils.
connected subsystem. Note that the operating frequency However, a higher efficiency is achievable if a shorter
corresponds to resonant frequencies of both primary and primary-secondary air gap is considered, so that the 3cm
secondary sides. The best capacitive compensation choice design choice could be seen as the “worst-case” in terms
for the proposed application proves to be the Series- of power transfer efficiency. Fig. 7 shows the coupling
Series (SS) topology, consisting in both the capacitors in efficiency vs. air gap corresponding diagram. The curve
series with the self-inductances, as shown in Fig. 3. The refers to a fixed operating frequency of 100kHz, which is
primary series compensation is preferred for its the resonant frequency. Now the mutual inductance M is
independence of the magnetic coupling. The secondary a variable, corresponding to air gap structures under test.
series compensation results in a voltage source feature, The figure highlights a maximum 98% efficiency value,
which is well-suited for the final DC bus level [30]. which represents the theoretical “best case”.
Therefore, both the primary and secondary capacitances
can be chosen independently of the load. Furthermore,
the SS topology results in maximum transfer efficiency
curve which is almost flat with increasing resonant
frequency [31].
Primary and secondary capacitance values C1 and C2
are related to the resonance frequencies of both primary

110
 common solutions, in the literature several innovative
solutions for current sensing are proposed to improve the
power conversion efficiency of a power conversion stage.
Since a resonant load is connected to the primary-side
inverter, a non-resistive network is not well-suited for the
specific application. Further improvement of the power
conversion efficiency could be achieved by soft switching
control techniques.
The signal across Rsense is then processed, through a
demodulation network and a microcontroller, in order to
properly generate the inverter MOSFETs gate drive
signals to comply with the required power level.
Fig. 6. Coupling efficiency vs. resonance frequency curve for the
proposed system at a 3cm air gap between the coils.

Fig. 8. Half-bridge inverter and sensing network.

III. COMMUNICATION LINK

Fig. 7. Coupling efficiency vs. air gap for the proposed system under a In order to properly fulfill the load power
fixed 100kHz operating resonance frequency. requirements, regulating the wireless transfer is
necessary. A transmitter side control could be an
A. Grid-connected subsystem appropriate technique to efficiently comply with this
The grid-connected subsystem includes the AC-DC goal. Therefore, a communication link between primary
stage, the inverter and the sensing network. The 72V DC and secondary side is needed, so that the load-connected
bus is converted to an AC signal flowing towards the subsystem could inform the grid-connected subsystem
load-connected subsystem by the inverter stage. about the required power amount. The proposed
Due to the choice of high switching frequency, communication channel is implemented through an
switching losses prevail on conduction ones. Therefore, a amplitude modulated signal which is sent from the power
half-bridge topology has been preferred to a four-MOS receiver back to the power transmitter. The amplitude
full-bridge in the proposed inverter circuit, thus reducing modulation is achieved by modulating the equivalent load
power losses. The half-bridge inverter, shown in Fig. 8, of the secondary-side resonant tank. For this purpose, an
generates a square wave between 72V DC supply and additional passive network is alternatively connected to
ground, which is the resonant network input signal. When or disconnected from the resonant tank, according to a
Mup is on and Mdn is off, the resonant primary side is fixed communication protocol. The modulation frequency
connected to the DC 72V supply; when Mup is off and must obviously be lower than the switching frequency in
Mdn is on, the resonant primary side is connected to order to produce an actual amplitude modulation signal.
ground. Therefore, by properly driving Mup and Mdn gates Then, the so-generated modulation signal is sent back to
the square wave duty-cycle is adjustable, thus allowing the transmitter side by the mean of the magnetic coupling
the control of transferred power. network. The modulated signal is monitored by the
The sensing network acquires the modulation signal primary-side sensing network and a proper demodulation
sent back from the load-connected subsystem in order to occurs by filtering the switching frequency harmonic. A
inform the grid-connected subsystem about the battery microcontroller decodes this demodulated signal and
power requirements, as it will be further discussed in the generates the proper inverter MOSFETs gates drive
next paragraph. Fig. 8 shows a simple resistor as the signals.
sensing network. Even if a resistive sense is the most

111
 IV. SIMULATION RESULTS been highlighted in comparison with the conventional
The simulation model of the power stage is full-wave diode rectifier solution. Regarding the resonant
implemented in SPICE-based simulation environment. network, a 91.6% coupling efficiency has been obtained.
Note that SPICE-based simulation environment is The power conversion efficiency of the overall
actually considered the best option for circuit simulation wireless charging system ߟெ஺ீƬோ௑ , including magnetic
of such a system. Thanks to the use of effective device coupling and receiver section, is given by (10):
models, the power conversion efficiency can be easily KMAG&RX KMAG ˜KRX 90.2% (10)
evaluated. The specific choice is further supported by
considerations on the whole charging equipment. where ȘMAG is the magnetic coupling efficiency and ȘRX is
Designers aim at modeling the whole system as closely as the power receiver efficiency.
possible to its effective behavior, including load and
sources.
Besides the power stage of both primary and
secondary sides, the complete wireless battery charging
station includes a digital section and a control subsystem.
MATLAB/Simulink simulation environment is
reasonably considered the best option for each further
subsystem. Note that to shorten the overall design
process, the whole charging station should be accurately
modeled and tested since the simulation stage. The
simulation of mixed analog-digital-power sections is
achieved by SPICE-MATLAB-ALDEC co-simulation
Fig. 9. Simulation results under 48V DC bus at the rectifier output. The
toolboxes, powerfully integrated in Simulink switching period is set at 10—s.
environment. Such a co-simulation procedure allows the
designer to test power conversion efficiency of the power V. CONCLUSIONS
stage, timing of the digital system, stability issues of the In this paper, an innovative recharge system for E-bike
control subsystem in an unique simulation setup. batteries has been proposed. Power transfer from the grid
Accurate analysis of measured data is key to reliability of to the load is achieved wirelessly, through a magnetic
measurements [33-36]. coupling structure. Although an air gap occurs between
Although the experimental prototype will include grid-connected and battery-side coils, the proposed
charger and battery, in simulation tests a constant 2A wireless solution allows an efficient recharge. Power
current independent source has been connected to the transmitter, magnetic coupling and power receiver have
rectifier output. An open loop configuration has been been accurately designed. Simulation results show
tested to evaluate the power transfer efficiency. excellent 98.5% power efficiency referred to the receiver
Tests have been carried out under the 100 kHz rectifier. Considering a 3cm air gap for the magnetic
resonant frequency. In order to obtain a 48V DC voltage structure, a 91.6% coupling efficiency is obtained. In the
value at the rectifier output bus, an 8% square wave duty- worst-case, the power conversion efficiency including
cycle is required. The duty-cycle-based control allows magnetic coupling and receiver section results in a 90.2%
regulating the rectifier output thus keeping constant the value.
operating frequency at the resonant values, with great
benefits in terms of coupling efficiency. Fig. 9 shows REFERENCES
steady-state waveforms on the secondary-side. As shown [1] A. Emadi, Y.J. Lee, and K. Rajashakara, “Power
in Fig.9, the DC bus level results equal to 48V with a electronics and motor drives in electric, hybrid electric and
ripple of about 150mV, that is only the 0.3%. plug-in hybrid electric vehicles” IEEE Trans. Power
By evaluating the overall power losses on each Electron., vol. 55, no. 6, pp. 2237-2245, June 2008.
component, the power conversion efficiency of each [2] R. Candela, V. Di Dio, E. Riva Sanseverino, and P.
Romano, “Reconfiguration techniques of partial shaded PV
section can be obtained. In the rectifier subsystem, the
systems for the maximization of electrical energy
total power loss is about 1.35W, corresponding to 1.4% production,” in 2007 International Conference on Clean
with respect to the 96W load power. The following Electrical Power, ICCEP ’07, 2007, pp. 716–719.
expression is used to calculate the power receiver [3] R. Inguanta, S. Piazza, C. Sunseri, A. Cino, V. Di Dio, D.
efficiency: La Cascia, R. Miceli, C. Rando, and G. Zizzo, “An
electrochemical route towards the fabrication of
PLOAD
K RX (9) nanostructured semiconductor solar cells,” in SPEEDAM
PLOAD  PRX 2010 - Symposium on Power Electronics, Electrical
Drives, Automation and Motion, 2010, pp. 1166–1171.
where PLOAD and PRX are respectively the load power
[4] V. Cecconi, V. Di Dio, A. O. Di Tommaso, S. Di
and the receiver power losses. Therefore, a 98.5% value Tommaso, D. La Cascia, and R. Miceli, “Active power
of the receiver efficiency is obtained. An increase of 5 maximizing for wind electrical energy generating systems
percentage points of the power conversion efficiency has moved by a modular multiple blade fixed pitch wind

112
 turbine,” SPEEDAM 2008 International Symposium on [18] V. Boscaino, G. Capponi, and F. Marino, “FPGA
Power Electronics, Electrical Drives, Automation and implementation of a fuel cell emulator”, International
Motion, 2008, pp. 1460–1465. Symposium on Power Electronics, Electrical Drives,
[5] G. Cipriani, V. Di Dio, and G. Ricco Galluzzo, “A Automation and Motion, SPEEDAM’ 10, Italy, 2010, pp.
photovoltaic system suitable for the battery pack charging 1297-1301. DOI: 10.1109/SPEEDAM.2010.5542090.
of an electrically power assisted velocipede,” in [19] N. Sisworahardjo, T. Yalcinoz, M. Y. El-Sharkh, and M.
SPEEDAM 2012 - 21st International Symposium on Alam, “Neural network model of 100 W portable PEM fuel
Power Electronics, Electrical Drives, Automation and cell and experimental verification”, Int. Journ. of Hydr.
Motion, 2012, pp. 1142–1147. Energy, vol. 35, pp.9104–9109, 2010.
[6] V. Di Dio, S. Favuzza, D. La Cascia, and R. Miceli, [20] F. Massaro, L. Dusonchet, S. Favuzza, G. Morello “Analisi
“Economical incentives and systems of certification for the economico-finanziaria degli investimenti sul fotovoltaico
production of electrical energy from renewable energy in Italia”. L’Energia Elettrica n. 03/09 maggio-giugno
resources,” in 2007 International Conference on Clean 2009, pp. 37-51, ISSN 0013-7308.
Electrical Power, ICCEP ’07, 2007, pp. 277–282. [21] Brenna, M; Falvo, MC; Foiadelli, F; Martirano, L;
[7] V. Di Dio, V. Franzitta, F. Muzio, G. Scaccianoce, and M. Massaro, F; Poli, D; Vaccaro, A “Challenges in Energy
Trapanese, “The use of sea waves for generation of Systems for the Smart- Cities of the Future” - 2nd IEEE
electrical energy and hydrogen,” in MTS/IEEE Biloxi - International Conference & Exhibition ENERGYCON
Marine Technology for Our Future: Global and Local 2012, pp. 818-825, Florence, Italy, 9-12 Sept. 2012
Challenges, OCEANS 2009, 2009. [22] V. Boscaino, P. Livreri, F. Marino, and M. Minieri,
[8] G. Cipriani, V. Di Dio, D. La Cascia, R. Miceli, and R. “Linear-non-linear Digital Control for DC-DC Converters
Rizzo: “A novel approach for parameters determination in with Fast Transient Response”, International Journak of
four lumped PV parametric model with operative range Power and Energy Systems, 2009, vol. 29, pp.38-47. DOI:
evaluations”, International Review of Electrical 10.2316/Journal.203.2009.1.203-4021.
Engineering (I.R.E.E.), vol. 8, Apr. 2013. [23] J.J. Yun, H.J. Choe, Y.H. Hwang, Y.K. Park, “Improve-
[9] V. Di Dio, R. Miceli, C. Rando, and G. Zizzo, “Dynamics ment of power-conversion efficiency of a DC-DC boost
photovoltaic generators: Technical aspects and economical converter using a passive snubber circuit” IEEE Trans. Ind.
valuation,” SPEEDAM 2010 International Symposium on Electron., vol.59, no.4, pp.1808-1814, April 2012.
Power Electronics, Electrical Drives, Automation and [24] A. Dannier, R. Rizzo: “An overview of Power Electronic
Motion, 2010, pp. 635–640. Transformer: Control strategies and topologies”,
[10] R.A. Dougal, S. Liu, R.E. White, “Power and Life Symposium on Power Electronics, Electrical Drives,
Extension of Battery–Ultracapacitor Hybrids” IEEE Trans. Automation and Motion SPEEDAM 2012, Sorrento, June
on Power Electron., 2002, vol. 25, n.1, pp.120-131, March 20-22, 2012, pp. 1552-1557.
2002. [25] V. Cecconi, V. D. Dio, A. O. D. Tommaso, D. La Cascia,
[11] V. Boscaino, G. Capponi, and F. Marino, “Experimental and R. Miceli, “Test bench realization and application of
test on a fuel cell-supercapacitor hybrid power supply for a specific working cycles for the characterization of
digital still camera” Proceedings of the IEEE International wheelchair electrical drives,” International Symposium on
Universities Power Engineering Conference, UPEC’09, Industrial Electronics, 2006, vol. 3, pp. 2188–2193.
Scotland, 2009, pp.1-5. [26] V. Di Dio, G. Cipriani, R. Miceli, and R. Rizzo: “Design
[12] Z. Jiang, R.A. Dougal, and R. Leonard, “A Novel Digital criteria of tubular linear induction motors and generators: a
Power Controller for Fuel Cell/Battery Hybrid Power prototype realization and its characterization”, Leonardo
Sources” in IEEE Int. App. Power Electron. Conf., APEC Electronic Journal of Practices and Technologies, vol. 12,
2005, 6-10 March 2005, pp. 467-473. 2013.
[13] V. Boscaino, M. Gaeta, G. Capponi, F. Marino, "Non- [27] R.F. Xue, K.W. Chang, and M. Je, “High-efficiency
linear digital control improving transient response: Design wireless power transfer for biomedical implants by optimal
and test on a multiphase VRM", Power Electronics resonant load transformation”, IEEE Trans. Circ. and Sys.
Electrical Drives Automation and Motion (SPEEDAM), I, vol. 60, no. 4, pp. 867-874, Apr. 2013.
June 14-16, 2010, pp.507-512. [28] F. Pellitteri, V. Boscaino, R. La Rosa, and G. Capponi,
[14] V. Boscaino, R. Collura, G. Capponi, and F. Marino, “A “Improving the efficiency of a Standard Compliant
fuel cell – battery hybrid power supply for portable Wireless Battery Charger”, in IEEE International
applications”. Symposium on Power Electronics, Electrical Universities Power Engineering Conference, UPEC 2012,
Drives, Automation and Motion SPEEDAM’10, Italy, UK, 2012, DOI: 10.1109/UPEC.2012.6398572, pp. 1-5.
2010, pp.580-585. [29] V. Boscaino, F. Pellitteri, G. Capponi, and R. La Rosa, “A
[15] V. Di Dio, D. La Cascia, R. Miceli, and R. Liga, Wireless battery charger Architecture for Consumer
“Integrated mathematical model of proton exchange Electronics”, IEEE International Conference on Consumer
membrane fuel cell stack (PEMFC) with automotive Electronics, ICCE 2012, Germany, 2012, pp.84-88. Doi:
synchronous electrical power drive”. IEEE International 10.1109/ICCE-Berlin.2012.6336483.
Conference on Electrical Machines, ICEM’08, pp.1-6. [30] S. Hasanzadeh, S. Vaez-Zadeh, “Resonance Based
[16] V. Boscaino, R. Miceli, G. Capponi, and D. Casadei, “Fuel Contactless Energy Transfer”, Power Electronics and
cell modelling and test: experimental validation of model Drive Systems Technology, PEDSTC 2012, vol. 1, pp.
accuracy,” in POWERENG 2013 - International 441 - 447, Feb. 2012.
Conference on Power Engineering, Energy and Electrical [31] S. Hasanzadeh, S. Vaez-Zadeh,nd A.H. Isfahani,
Drives, 2013, pp. 1–6. “Optimization of a Contactless Power Transfer System for
[17] M. Chun-Hua, “A novel three-dimensional, two-phase and Electric Vehicles”, IEEE Trans. on Vehicular Technology,
non-isothermal numerical model for proton exchange vol. 61, no. 8, pp. 3566 - 3573, 2012.
membrane fuel cell”, Journ. of Power Sources, vol. 195, [32] V. Boscaino, P. Livreri, F. Marino, and M. Minieri,
pp. 1880–1887, 2010. “Current-Sensing Technique for Current-Mode Controlled

113
 Voltage Regulator Modules”. Microelectronics Journal, [35] S. Nuccio, C. Spataro, G. Tinè: “Virtual instruments:
Elsevier, vol.39, no.12, Dec. 2008, pp.1852-1859. Uncertainty evaluation in the presence of unknown
[33] C. Spataro: “A/D Conversion Based Measurements: electromagnetic interferences”, AMUEM 2008 Internat.
Identification of the Parameters for the Uncertainty Workshop Advanced Methods for Uncertainty Estimation
Evaluation”, Proc. of AMUEM 2009 – International in Measurement, Trento, 22 July 2008, pp. 56-61.
Workshop on Advanced Methods for Uncertainty [36] S. Nuccio, C. Spataro, G. Tinè: “Electromagnetic
Estimation in Measurement – Bucarest, 07-08 July 2009. immunity of a portable PC-based measurement
[34] C. Spataro: “ADC Based Measurements: a Common Basis instrument”, IWADC 2003, Perugia, Italy, Sept. 2003.
for the Uncertainty Estimation”, Proc. of 17th IMEKO -
TC4 Symposium. Kosice, Slovakia. Sept. 08-10. 2010.

114
SICE-ICASE International Joint Conference 2006
Oct. 18-2 1, 2006 in Bexco, Busan, Korea
Regenerative Power Control for Electric Bicycle
Nitipong Somchaiwong' and Wirot Ponglangka2
Department of Electrical, Faculty of Engineering, Rajamangala University of Technology Lanna, Chiangrai Campus
Chiangrai, 57120, Thailand
(Tel: +66-0-5372-9600; E-mail: thaijoblessghotmail.com)
2Department of Electronic, Faculty of Engineering, Rajamangala University of Technology Lanna, Chiangrai Campus
Chiangrai, 57120, Thailand
(Tel: +66-0-5372-9600; E-mail: wirotepghotmail.com)
Abstract: This paper presents the design regenerative power control for electric bicycle. PM brushless dc motor use
on electric bicycle. In electrical bicycle system occurs due to the regenerative power regulating to prevent a DC-link
over-voltage condition. Regenerative power control circuit is absorbing regenerative power in the battery. Details of the
motor and the electronic converter are given. The regenerative power control for electric bicycle method is a sample
and a low-cost solution. Test results confirmed the feasibility of the proposed motor design.
Keywords: electric bicycle, regenerative power control, BLDCM, permanent magnet brushless dc motor.

1. INTRODUCTION
In this paper, we describe a studying and designed From fig. 1 is shown the original model of Electric
regenerative power control for an electric bicycle. This bicycle which motor inside the wheel installation. The
method uses to examine the energy remainder from a voltage of 36V is necessary for this motor.
motor while a motor was suspended to brake off. This
proposes capable to inspection the voltage that supplies
to motor and comparison with the reflected voltage from
the motor. Normally, a motor was performed as a load
of circuit and have the voltage equivalents to same
value. When we describing to this characteristic,
meanwhile a bicycle are running on a slope down of
route, caused to increasing a cycle speed of motor more
than a normally situation. In previous, is the main
reason to have an experiment of a regenerative power
control for electric bicycle. The experiments results
were succeed according to designed condition.

2. AN ELECTRIC BICYCLE THEORY

Fig. 2 The in wheel motor.
Electric bicycles were lead to more popular in used
on present-day. At the present moment, the energy price From fig. 2 is an in wheel motor type has voltage
from oils was more expensive. The used of motor type limiting on 36 V. and the motor speed is 300 rpm.
in this paper was in wheel of Electric bicycle type. The
useful of this motor type is use to perform an original
model in the studying and designing inspection and 3. BASIC OF IN WHEEL MOTOR DRIVING
regenerative power control circuit. The components of In wheel motor has a basic structure similar to
an Electric bicycle as shown in fig. 1 BLDCM (Brushless DC Motor). In generally, have a
coil is winding enclose to at stator and have permanent
magnetic located at the core of rotor as in figure 3.

Fig. 1 Original of An Electric bicycle.

89-950038-5-5 98560/06/$10 C 2006 ICASE

4362
 IF
Output > input = charger on
Output< input = off
End.
From the abovementioned, we can design the
regenerative power circuit accomplished as shown
in figure 5.

2 DI I

R1 Tr

I
Fig. 3 The Structure of In Wheel motor. Battery R4 IGBT Motor

From fig. 3 is shown the structure of In wheel motor

D,~~~
In
D2
~ 19
for the original model in this propose. IF

4. THE BASIC OF ENERGY CONTROL Fig. 5 A Regenerative Power Status Inspector Circuit.
CIRCUIT OFAN ELECTRIC BICYCLE
The operation of figure 5, if the voltage on "1"
In fig. 4 is shown a diagram of an Electric bicycle positions are greater than "2" positions, while a position
control circuit and divided to two parts. "1" as output but position "2" as the input. A result from
difference voltage is leading to control transistor on.
1. Regenerative power inspector circuit.
2. Battery charger circuit. The current is driving by transistor passing through R3
and R4 to ignite the gate of IGBT switches ON state. A
result from IGBT switches ON state, and causes the
voltage across on position "X" and D2 will prevent
over voltage of IGBT

4.2 BATTERY CHARGER CIRCUIT

The function of battery charger is that operate
continuing from regenerative power circuit by receive
voltage across "X" position entering to be input voltage
Fig. 4 Diagram of power control circuit of Electric of battery charger according to figure 6.
Bicycle.
RI
From fig. 4 this diagram is describing the principle of DI
power control circuit operation of an Electric bicycle.
The control circuit is dividing to two parts for
inspection the status of regenerative power and battery
charger control circuit. The principle operation of D2 2 , C3
control circuit is show in this detail.
x
4.1 REGENERATIVE POWER STATUS
INSPECTOR CIRCUIT TR2l TC2 R5
This circuit used to examine the regenerative power
status by checking the magnitude of voltage at the out
put of circuit. The method in order to inspection Fig. 6 Battery Charger Circuit.
accomplished by connecting a wire on a pole of motor
position comparison with input voltage by connected on An operation of battery charger of circuit in figure 6
a pole of battery position. accomplished with the voltage across input position or
IF magnitudes on the output voltage are greater "X" position of regenerative power circuit, there are the
than input voltages, a regenerative power control circuit current passing through R1 , R4 and R3 to activation
is operating by transmitting an electric signal to battery
charger circuits operating according to this procedure. the operation of Q2 and Q1 respectively. While Q1 is

4363
ON state, the output voltage usable to charge battery at motor state. The other hand, the red line is representing
mechanic energy supplied to motor state.
position (VX ) .
5.2 The experimental for regenerative power control
5. THE EXPERIMENT RESULTS Circuit
5.1 An experimental of relationship between voltage The experiment setting by initialization
supplied to motor and motor speed according modeling the velocity of motor speed are stable at
40 rpm with supplied the voltage to motor is 5 volts
Table 1 The relationship between voltage supplied to and supplied mechanical energy to motor for
motor and cycle speed.
regenerative power in experiment of Electric bicycle.
Voltage(V) speed(rpm)
5 40
10 90
15 130
20 180
25 220
30 270
35 310

From table 1 is the relationship between voltage

powers supplied with cycle speed occurred meanwhile
we supplying the power to motor.

Fig. 8 The situation Modeling for Regenerative State.

Table 2 The relationship between motor speed and
voltage while supplied mechanical energy to motor. Table 3 The results of modeling by supplied voltage
to motor 5 V.
speed(rpm) Voltage(V)
40 4.2
90 9.2
speed(rpm) Vx (V)
130 14 40 0
180 19 90 9.72
220 23 130 11.93
270 28 180 12.17
310 32 220 12.41
270 13.24
From table 2 is shown the relationship between cycle 310 15.24
speed and voltage while supplied mechanic energy to
motor and the results inform of graphic according to 18-
fig. 7 is the voltage versus motor cycle speed. 16
14-
12-
40
35;
30
6-
4-
25-
2
20 -
0
0 15-
40 90 130 180 220 270 310
10

5- Speed(V)
0-
40 90 130 180 220 270 310
Speed(rpm) Fig. 9 Graph illustrations for the Operating of
Regenerative Power Circuit.

Fig. 7 The relationship of voltage and cycle speed of From table 3 and fig. 9 is shown the characteristic
Motor. operating of inspector circuit while cycle speed of motor
more than 40rpm. The motor will generating output
From fig. 7 there are 2 state of the relationship voltage more than 5V, caused the regenerative power
between voltage and cycle speed according to the blue circuit operated and appearance the voltage across VX .
line is representing an electrical energy supplied to This voltage usable to charge battery.

4364
 6. SUMMARY
This paper proposes a studying and designed
regenerative power control for an electric bicycle by
generating a model situation of regenerative power. The
results of an experimental, we can storage an electrical
energy in real time and use as to original modeling in
order to improve in the practical application. The
propose method were derived completed with more
simply and save cost.
REFERENCES
[1] Tomy Sebastian and Shashi B. Dewan,
"Comparison of Dual - Converter Based Power
Supply Systems," IEEE Transations on Industry
Applications, Vol. 25, No. 2, March/April 1989.
[2] Dennis H. Braun, Tomas P. Gilmore and Walter A.
Maslowski, "Regenerative Converter for PWM AC
Drivers," IEEE Transations on Industry
Applications, Vol. 30, No. 5, September/October
1994.
[3] T.F. Chan, Lie-Tong Yan and Shao-Yuan Fang,
"In-Wheel Permanent-magnet Brushless dc Motor
Drive for an Electric Bicycle," IEEE Transations
on Energy Conversion, Vol. 17, No. 2 June 2002.
[4] Duane C. Hanselman, Brushless Permanent -
Magnet Motor Design. McGraw-Hill,Inc. 1994.
[5] T. Kenjo and S. Nagamori, Permanent-Magnet and
Brushless DC Motors. Oxford University Press,
New York, Oxford. 1985.
[6] R. Krishnan, Electric Motor Drives, Modeling,
Analysis, and Control. Prentice Hall, Inc. New
Jersey. 2001.

4365



Proceedings of the 36th Chinese Control Conference
July 26-28, 2017, Dalian, China

      

   


  
 
  


 



 

! 
 
"
#
$%"&
'
()* 





 

+,
 
#
$%"&



-  &

(**


-"
 
 "
.

" 

,! 
/0
 
"&",
&
-1(
!
,,!
-

"

, &
"

 

" ""
2
,
.
.
" 

!.
%

.
" &
0
"""

%  






"



3$

,! 


 "
!
"&",
.%
.


 
"
"

 
!.
%"

%&

.
"
",
!.
""
+


,!%
"
! ,
.








1 4-

""


.
!

" 



,! 

,0


" 

4 &
.
,&
,0

 
%  

 ,,
"



 /0
 
"&",5
" ""
2
,5
1 4-

""
",!  5
 ,,
"


-
,


"
"&",
"

!.

!.


     %"

!!


%  

6" ""
2
,
76 218
"
. &
"




",! 


!

%"
"

""
"

"

4 

"."

%
"


"&",

% "

 
!
9
.
"
 
&
 0
 ,



" &: ;


% !,

"&

%,
! 
"
,

,
""




!

%,
 
% "
"


!"%

.
 &
% "
"
,

"




" ""
2
,
 
%


%
"

" 


3 !
"/%


" &


 
% 
 
"
,,:;


4 
-
%
"


"&",
 0
 ,

+
"
!!

!! 


,



" ""
2
,

"


" ""
2
,
#$    
 
"
" 

,! ,:(;
-
,
!

-1(4 *(<-<
.
"
,
&
-
"

 = 
+>
#% &'  !    
!
 :?;
+
 "

.
 

,
-
,
!.
"!! &


 
% 
 


 "

!%
%

,
 
,
,
&
!
H(<
I
%  



 &

 
"""

 


.
,



 

  "
!.
"!! &


!.


""
",! 

1 
"""
"
!!"
"!! &

"!% &
!
H @
I
H@(
I

H((


 
!.

" :@;
I
%  :D;
-0



 ,!
"


 !  !" .


, 




"&",

" 



 ,!
.0
 "/ !
"&",:J;

6" ""
2
,
"

!! 
"
"



!

".

4 

#


,
" 



""
" 
 
"
!



1$

" 
""



 /,
""


,

!"
,

 
%
,0
G

G


G

0


, 

, 


!%



 ,!
.

!




 /,

" 


%


"



"
.



" 
"

""

31
.%

 
 

 ,!

*)I

%  

F
F(
F?
,0"
G


%



,


,


, 



",
,
G


"
! &

!%
 

 

%

,

"
:<;
#

G
,
.0

% 

"

,

",
,






"!
% 
!

  "
" 

!
# * (



 
 "/ !

 

"A


31
" 
0&





"A



"!:);

IH(<I
F
*
G
3C3
F@





























































@(

 -"
.0
"
"!!
&

- 
3B


#
3%



2
#
"



=&92* <? =&92* <?)
#
F<
@(
3%
C

4


?*D*D@1#* 
F
*

F( G
!B

"
#
$%"&
EF* )@ * C3C

F?
*(
EF* )*@ EF @ < G &
 
!B

"
#



$%"& .==&* )*D .==&* )* C2

10132


4 


!

/!"


6/!"
%

".


3.
"!! &

!%"
H
@I
!.
"!! &


4 
@
"

=,! 

=! 
0




%&

!!
%"
7"
"


"""
1 

%
""
! ,



"

1 

8
"
&!

!.
"!! &
 "
"
"/

,!
"
,


" 
"

".

!

-

!.
"/
-
9&
"
"

)M
6"


%

!

".

4 
(
.



""
"
!!
" 
*4
>
@*I
!

"/
!

""
G

G


: *; !"


-"

"
"

!

 &
.


 
"
"


 &

#  
"
-


"&",
!.
I
K
 6
K
F*6



! &

!%
 



","
"!
!

 
B"

F*6
""

,",
"

" 


.
!.
"

 

"A


 



 "
"
"

!

"
""&

,
#

"&",
"
!.



F G 31

"  "

"


/!"
 

I+C I $-
"!% &
#
"
,

!
 

"

F
G 3C3 I
K
 6
K

=
 
1 
6
,

3C3
"!

!
"!


,! 
+

"


"&",
!./C/



F %

&

%

, 

/"


""
F*6



=
 
2

I6#-
H @I 2*6  6
C2
@*I
*6 $ 6 C?**) *4
C2

6-
*?

(
I
+C
I6

D
)
<
F*6 ((
F6 I6
<J
(> **I
4 (
-
!.
"

!

66
C2
?
+C
 1
+F* (
I
@ F6 ((
*M
?6
**I
*(

(> **I
F*6
)*
I6
G F?6
J6
*M

I+C I $- *M **I
*( F6
3C3 ***
C2


6
6
F


G 4 @
2%


6
!"

3C3 F
*  #%) *"!     
F
+




,=,,
9



 /,



" ""
2
,


C2

"&
/!"
"

"&",
"

4 ?
+,!%
!.
"

!

",




"&",
#

+,!%
!.
"

!

.

""
 /,
""



!"
,0



4 
?

0&
"


""
G

G

"&
, 
 



, 
 


"9
#

"&",
"
!.



"
!
"
".


,
I+C
K
G
K
F
K
C2



J*
 "

"
&
 

"A


.
" ""

!
"


"
!
I+C

"
.

% 

%
 "/ !

K
G
K
F
K
F
K
C2
"

G
4"
 
G
 
4

"""
F6

F?6
.

1 


/
-


*
"

=

G
/
,

!"
!


" 

 



"
I $-
"
*
G

G

I+C

%
!

" 

".

+



"

!
,!
.
4 
(
"
" &

!%

!.



"0
,"

"
((N

  &
%
,!%
-

,0"

  &

""
,

.

! 
!


 
.
.0
"

 &
,!%
6

 ,

!!


 /9&


 
"


&

 
"!
 
!0

. 
%

1 

90 &
%


,!






.


O1 
! ,O


".
"""

 
!


, : ;
6"!



26*
!

."


#%#  (   
"!
%"
%  
4 
@
"."




!
6
 

",! 
""
F6

2%

"

1$

%
,

%  





% 

"

 


!"


#
""
" 
""

!"

!0
"
 




!.

&

".


31
%
" 

+


",!

"

,N
",! 
""

,

!.
9,"


 
% 
,! 

%  
" 
"
"

"


 

1 

.

% "

 

,! 

!%



!
9&
",

""

"

"
, 
"
".

4 
<




!
"

".

$"
+FL"
 
%
",! 

"
"
6/!"


 "

!
+F *(

"
!!

""

%
4 






,,/,


#
6

/!"
"!% &

"

" 

 /9&
" 

" 


+F *(
-0



""&


./ %
, 
"
! "
.
!
 

10133


,

"

&


"  

,! &

%
%


.
 

H((I
1 4-
",

""
"

F@( ".
"!

"&
0.



((M> P
6 $2
""
,0


,",
!.
".

C6 46) 46(( F?( 
6
M> P -IJ ?
? FD* +6 %"
1 4-
 ."
 

/"

"

( M D
) <
*
(
* *

""


!""
!



*
46(? F<* F@@ @M> P % !,

",
!""
  &


?
M> P
C2
46D
((
C2
% 


""
"

 &

.
, ,"
-

*

", 


""
1 4-

."

4 <

",! 


!""
"
.


" 


1 4-

!,"
 !

,!%

" 
9 &
#"

#%+,( &      
,
 


""


1 4-

#"


 
" 


/A
%"
.
,!
2
,!

!
 
 

&

1 

!

"!


""

"


.
 

 /
"/
, 

",! 

"
".

4 
D




 

-
"


"


,!


""
 ,






31
" 
&
,"
"

1$

!

#"
".

4 
)



%
!

#
6

/!"


%  
,! 
 

!
,!
,! 
-F#3
4 D
-,!


""
 ,



,
.0
-F#3
!
"
 


"

+ 

,!

,!
 "



 

""
!
%
4 !
 
!"

""


1 4-


,

",! 

-
, 


!



"
&

,!
,
1 4-


 

0
""


!
,!
""
%  

!""



,

H((I 
".

,

%



 "/ !
0

0&



"&",
"
.

F((
C2 *M 
 

1 4-



""
%  

F(?
C# FD M
C6 FJ M M
-F#3
".

4 
J


1 4-

""

$#
C F?J M (
-IJ ?
F( (M G ",! 

"
"

!"

=! 

  D*@* H((I
*
 F(
J ?) M
*?
F@
?

2) C2
((MNH>/ P
H((I C2 6#( < $<#
F@? @ M C2 # F?J (
F@
MNH>/ P 
M ( * 4
-IJ ?
(? * 4 Q+
?

C2

H((I C2
4 )
%
!

C2
F@* F@
MNH>/ P @MNH>/ P
),*-.    "! 
(@ *


 
.0"

!"

"
(/@


4 J
1 4-

""
",! 


,"


"



,
!.
"
4 
J
"."

#

C2
"!% &
,

@**

"

,


 



.



1 4-

0



D#

*#
#
"
,

 

 ."
 
" 


!%"

""
",!  


".

4
 
!.
!! "
0

31
!"


%


"!
!"



!.
""
! ,
" &


0



#/!"

6/!"


.



,

""
",! 

"
!
"

"



!
,!
,

#/!"
%  


"

 /!.
"



",


%  

" 
"
H(<
I


%  
 &

""L"
 
"
 0&

%
" 
,! 
.

6
!
,!
+
4 
J


"

+
"
!!


0
% 

&
"

 ,!

#
.


& 


31




""
!
.
1 4-
"


!

!
"

" 
"
.0
"


-"
,
"

"



 L"
,


"
"
F?J

F@





"



!.
",

!%" &
,

!%
"
"
%  


%  
!




0
% 

%
" 
"




 ,


 
 


B


,

1 


,! 
"


"


!"%


! 

"""

"
 


-
 %

"


9% 
""


" 


"=
"


,

!.



",
+



""



",
"

"" &

",
!! 
-
1 4-
.
"
.
"
,!
#


,



""
",! 
 

,!




",!  


"



" 
"

&
,!

10134



 

,!
 "


,"

,
&",
!.
"
&
,

#
6





.&
 

!
!""


/!"
%  
 

!
,!
,! 
"

1$
!
""

,!


" 
 "
%  

" 
%  
0
% 
"
,



 /,
,!"
-

",! 

.

"
-
1 4-
/
"9

.%,
"
".

4  *
"
,
&


!"
,




"!
"

&

"!
!

,!
.


"
"!
#
!
"

&
AA&
3+2
"


%
" 



!





% 
"
,!
.

%
" 





% 
&
3+
 ,
-
,",



%


 
"
".

4  

4  *

,! 
%,

#"
".

4 
*



!"
,
!


,
"

"



(
"!% &

0

/!"
,
#
!"
 &""



#/!"

 
1 4-

""
%  
" 
"

,! 
&

!
,!


.%,



#

!"


 
2
.%,
(


#
!"




 
2
.%,
4  
1",


%

# &""



 


 
0  "1


!
,!
%  
 ,!
&


,! 




% 


 


 


+
"
!!
" 

 /!,
 

%  
,! 

%  
 "
.

% 
 


.
"
"

-1(

%"


 
.
" 

4




 "
-
 &""

(

 



 /!.
 




.
1 4-


1 4-

%  
!



% 
+

""
",! 
"
,!%
.
,0"





"








!!
9!,
",! 

"&

,! ,
4 &


"
"
",! 
"

"
!! "
36

"
,


+-   &" "* ,",


%
!%


 



"

"
!

+
"
!!

 ,,
""
.



""!
"
" 

! 


 ,


 


,,
" 
"

" 
.
&
4F?
: ;


 
E
 !




! 
-
1 4-
!.
""
1( )
"
 

2" 



3



,!"
! 


 ,,
! 


2
3.
!! &
:;
1
R
 

%
!
1$


!"
 



"
 
-  &
* (
7*@8S) /)(

:;
2 



 
-
C.
-  &

".

4 

36
" 
!&"
,!



2

6" ""
2
1:;
 

1",
R
+",
* 7*<8S J/

:(;
2&
 1A

#!! 

-1(

-

3"
6" ""
2
1

&",:;

1," * ? 7*(8S?)/@ 

:?;





M

 
" ""



 /!
6" ""
2
1
6"


// 

60
14:;
3.
 "
+
-""


* <
( 7)8S?<<J/?<D(

:@;

#



4


#

+=
#!!


,!"
,,
3"


" ""

6" ""

1"
.
C/
60
14:;
+

-""

3.
 "
* @S / 

:<;


-
3
4

4

1 

, 
# &""



6" ""
2
1

&",:;
1

4 
36
" 
!&"
,!
 
R
#,
* ?
(*78S?J?/?JJ

:);
+
 =



M 
#-+C

/*   -F$-$F
6-C
3F+C-2
+F$+-
6 #F2

#C2
+C$ #-+C
6 66+C
+C
#
6F$ 

-
"&",
""


"!




1 - F:;

* )

 "/ !



,
"!

B"



10135


:D;


&
 

" ""

,:;

* <
: ;
-"&








".
" 



:J;
M,
1


0



 
1
%
, 

" ""
2
,:;

* <

" ""
2
,
"&",:;

* <

: *;
#=

-
-
2%

6" ""
2
1
0





2
1:;
#A 00
 .
* <

7@8S@D/<(

10136
 Study for developing the Energy Recovering Circuit for
Modern E-bike Controller

Emilian CEUCA*,Gh. Brezeanu**, V. Trifa*,

*Dept. of Electronics, University “1 DECEMBRIE 1918”, N. Iorga 11-13 street, Alba Iulia, ROMANIA,
[email protected],
**Dept. of Electronics, Politehnica University of Bucharest, Bucharest, ROMANIA,
[email protected]

Abstract: The aim of this paper is to present the driving cycle involves frequent accelerations and
implementation of an energy recovery system based on decelerations. In this case, because the power
BLDC motor deceleration, in order to be used in e-bike’s
propulsion systems. The energy which should be lost demand varies according to the driving cycle, the
during deceleration is now recovered, stored and then traction system is exposed to a variable speed
used in other applications. The energy recovery and operation. An optimal design of the control unit
reutilization leads to a considerable decrease in total will lead to minimize the power peaks upon
energy consumption necessary for traction mode. At the
same time combining with the super capacitors usage is
applying the torque command. Minimizing the
protecting the life expectancy of the battery, which power peaks conducts to an improved
normally supplies the propulsion system of an electric performance against “over and under” voltages
motor. because the DC link voltage changes upon a
sudden load variation. This paper investigates the
Keywords: e-bike controller, Block comutation, sin
commutation. methodology to reduce the current peaks during
accelerating and braking operations as well as the
1. Introduction behavior of a supercapacitor to reduce the energy
loses. The current peaks can be used by
Electric propulsion systems provide cost- recharging the supercapacitor until a voltage set
effective solutions and are of great interest for point value is achived.[1]
companies in order to gain competitive edge in Due to the time-variations of the system
the market. In this respect e-bikes are becoming parameters, the control unit has to take such
more and more attractive as they possess the variations into consideration. Therefore, an
advantage of high energy efficiency compared to adaptive control method is a smart and efficient
other energy forms. Electric vehicles like e-bike solution, in which the values of the controller
which is driven by electric motors use the motor parameters are modified during the systems real
as a generator when using regenerative braking time operation in order to achieve required
and its output is supplied to an electrical load; the performances. In figure 1 is designed the
transfer of energy to the load provides the schematic diagram of energy recovering with
braking effect. The most important function of BLDC Motor in the deceleration stage.[2]
the brake controller, however, may be deciding
whether the motor is currently capable of
handling the force necessary for stopping the
electric vehicle. In vehicles that use these types of
brakes, as much as any other piece of electronics
on board, the brake controller makes the entire
regenerative braking process possible.
One of the most challenging aspects with
regards the energy management system in e-
bicycles is the ability to optimally recover the
kinetic energy and store it in the battery pack.
The aim is to find a technical solution that
reduces the energy consumption, especially when
the e-bicycle is driven in the urban traffic areas as Figure 1. The Block diagram for the circuit of Energy
the largest part of energy is drawn because the Recovery. [2]

978-1-4799-8863-1/15/$31.00 © 2015 IEEE

245
 2. Strategy to Control the BLDC output power is constant, each coil must be
Motor supplied as long as the induced voltage constant.
Into the working cycle of the e-bike consists
by acceleration period, constant speed running
and the deceleration, shows us that the necessary
strength in the first stage is much greater than the
need to run at constant speed. In the braking stage
the kinetic energy gained by acceleration energy
is transformed into heat on the brake system and
is irreversibly lost. The question arises if, during
braking, the kinetic energy of the vehicle would
not be able to recover and store energy in the Figure 3. Waveforms of the voltages on the motor coils
batteries. Because energy can be extracted from Unlike brush DC motors, BLDC motors
these batteries and used again, we would say, that switching current through the windings defeating
we are dealing with the "regenerative" recovery. automatically mechanical switching must be done
Studies have shown a surprisingly large electronically. The control system must therefore
proportion of the kinetic energy dissipated during have information on the rotor position to supply
an urban cycle average to realize the function of correct windings. There are two ways to obtain
stopping or slowing down the vehicle. this information: read the rotor position using
There are several driver circuits topologies Hall effect sensors and estimate the rotor position
used in BLDC motor applications. The most based on the induced voltage zero crossing. The
common are type-phase buck converter and first method is easier to implement using only
three-phase Inverter Bridge. A typical three- digital circuits, but the second method involves
phase bridge inverter for driving a BLDC motor lower costs because eliminating the need for
is shown in Figure 2. The bridge consists of six sensors, engine construction is simpler.[3]
electronic switches. In practice using field effect This paper proposes a research to find the best
transistor technology built in metal-oxide- control strategy for recovering the energy during
semiconductor (MOSFET - Metal-Oxide-Semi- deceleration and use an amount of stored energy
conductor Field-Effect Transistor) or the isolated in the traction period. The simulated model will
bipolar transistors (IGBT - Insulated Gate Bipolar be compared with the experimental setup
Transistor). MOSFET switching losses are lower measurements obtained. We try to develop a
than IGBT transistors, but the internal resistance practical board for testing the control strategy
(and thus power dissipation) higher conduction. based on the thesPIC33EP256MC506 DSC
IGBT transistors are commonly used in microcontroller from Microchip. [3]
applications involving power or high voltage.
3. Circuit of the energy recovery
The e-bike simulation model is based on the
real equations of the boost circuit.

Figure 2. Schematic of Circuit for the Brushless motor

control

Typical waveforms for the voltage induced in

the stator of the three-phase BLDC motor shown
in figure 3, where it can be seen the EMF voltage Figure 4. The Boost circuit for every phase of the motor
induced in each winding of the motor is constant
over 120° and varies linearly with rotor angle In the figure 4 is shown the proposed boost
before and after the constant portion. Because the circuit for each phase of the motor. We calculate

246
the values of circuits for assuming that the output The practical approach is to determine the
voltage is greater than the battery voltage. The energy transferred. This energy is better to be
circuit is used when the Back-Electromotive- stored in supercapacitor because in the situation
Force voltage (BEMF) is lower than the battery of the battery will not be able to store this and
voltage. [3, 4] will produce thermal dissipation with destructive
The figure 5 showed the BEMF signals on effect on the battery.
each coil in the traction period. One of the BEMF Assuming a boost circuit with the source
is acquired with the small V/ div for the better voltage sinusoidal like in the figure 4 we develop
accuracy. a interface in Matlab for estimation of inductor
current and the voltage applied to the high side of
transistor. The interface values obtained are
shown in the figure 7.

Figure 5. The BEMF signals

All the inputs are with an attenuation of 10

times. This is the signal used in the circuit of
energy recovery in the first moment when the
brake is pushed. Is necessary to mention also the
situation when BEMF value is greater than the
battery voltage the solution is to block the
corresponding high side of the transistor and the
charging circuit will be resuming to the recovery
diode of the high side of the transistor. This will
be the case corresponding to a high speed of the
Figure 7. The wave form signals simulations used for
e-bike according to figure 6. In this case we recovering energy.
control the low side of the corresponding coil
with a PWM signal with small duty cycle. Then In figure 8 is shown the output voltage of
low side is turning off the energy stored in coil is boost circuit at lower speed. In this situation
transferred to battery through recovery diode boost circuit is driven to assure the voltage level
from the high side. for energy recovery.

Figure 6. Voltage on each coil in the recovery stage at high Figure 8. The wave form signals output of the boost circuit
speed at low speed

247
 The frequency at which the sectors are works utilizing a majority function for digitally
sequenced determines the speed of the motor; filtering the Back-Electromotive Force (BEMF).
driving faster the commutation allows achieving Each phase of the motor is filtered to determine
higher mechanical speed for rotor. Because of when to commutate the motor drive voltages.
this, detection of position using the BEMF at zero The figure 10 showed the design of the
in situation with very low speeds is not possible. controller for e-bikes produced “in house” at the
A lot of practical applications do not require “1 DECEMBRIE 1918” University of Alba Iulia.
positioning control or closed-loop operation at
low speeds. For these applications, a BEMF
sensing method is very appropriate. [5, 6]
At the maximum speed around 38 km/h
corresponding to 140 Hz used in simulated
results are shown in Figure 6 and the
corresponding waveform for power are shown in
the Figure 9.

Figure 10. The controller board proposed and developed in 1

DECEMBRIE 1918 University. (Eagle 3D)

The optimal design of the controller is

required for the power management in order to
reduce the energy losses during acceleration and
braking processes. Designers can expect
environmental demands to continue to drive the
need for advanced motor control techniques that
produce energy efficient solutions. However, the
implementation of advanced cost-effective motor
control algorithms is now a reality, thanks to the
new generation of Digital Signal Controllers
(DSCs).
These software-based motor control designs
enable rapid customization to address multiple
markets by changing only the control parameters.
References
Figure 9. The wave form signals simulations used for [1] A. Taut, O. Pop, E. Ceuca “System for Energy
recovering energy. (Matlab results) Recovering with BLDC Motor at Deceleration
In the figure 9, the first waveform is Momentum”, ISSE 2013, 36th International Spring
Seminar on Electronics Technology „Automotive
corresponding to BEMF voltage, the second Electronics“ May 8 – 12, 2013, Alba Iulia, Romania
represent the power transferred to battery or [2] E.Ceuca,A.Tulbure, M.Risteiu,”The Evaluation of
supercapacitor and the last is related to power Regenerative Braking Energy”. Proceedings-
SIITME2010 – 16th International Symposium for
losses on commutation process. The practical Design and Technology of Electronics Packages
aspect studied in our work is focused to minimize SIITME2010, 23-26 Sep 2010, Pitesti, Romania, pag.
65-70, ISBN: 978-1-4244-8122-4
the commutation’s losses. [2]
[3] AN1160, “Sensorless BLDC Control with Back-EMF
4. Conclusions Filtering Using a Majority Function”, Microchip
Technology Inc., 2012
This work is focused on the tests for develop [4] AN901, “Using the dsPIC30F for Sensorless BLDC
the motor control algorithm that is implemented Control”, Microchip Technology Inc., 2007
[5] AN992, “Sensorless BLDC Motor Control Using
using a dsPIC® Digital Signal Controller (DSC)
dsPIC30F2010”, Microchip Technology Inc., 2005
based on literature references and like extra [6] AN1017, “Sinusoidal Control of PMSM Motors with
features the circuit for energy recovery is added dsPIC30F DSC”, Microchip Technology Inc., 2005
with the additional supercapacitor. The algorithm

248
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 1, September 2021

Impact Factor: 5.731

Design and Analysis of Electric Bike

Praglbh Shivnay1 and Prof. R. S. Yadav2
Student, Department of Mechanical Engineering1
Guide, Department of Mechanical Engineering2
Marathwada Mitra Mandal's College of Engineering, Pune, Maharashtra

Abstract: The increased demand for electric bikes is the subject of this study. Our main focus is in the
automobile industry, where we are converting outdated bicycles to electric bicycles. The major goal of
this study is to present an accurate picture by linking the many energy sources that humankind has
access to humanity In order for humanity to progress in today's civilized environment, they must travel.
And in order to accomplish this, his journey should be as quick and painless as possible. The Electric
Bike, which is powered by a battery and so supplies voltage to the motor, is the subject of this study. This
study is concerned with the design and construction of an electric bike that runs on electricity as primary
energy. In the main system, there is a setting for a rechargeable battery. When compared to a traditional
car, the bike's electrical power can deliver better fuel economy, performance, and pollution reduction.

Keywords: Hybrid Electric bike, Electric Battery, electric bike, Lithium-Ion Battery

I. INTRODUCTION
The energy crisis is one of the biggest problems in today's globe, due to the rapidly dwindling resources of fuel,
diesel, and natural gas. Furthermore, environmental degradation is a factor in resource depletion, which is a worrying
warning. Our study offers a remedy to the aforementioned dangerous issues. The system that we invented is the Electric
Bike. This project has a number of advantages for team members as well as the broader public, and it raises awareness
of the necessity of using alternative forms of transportation. For a short trip, the electric bike, which is powered by a
battery, is the most frequent means of transportation.. Fuel cells and petrol-electric hybrids, both of which are under
development, might be added to increase the variety of e-bikes available and increase manufacturing, hence increasing
the efficiency of the electric drive system. Electric bikes have proven to be a viable means of reducing pollution to a
greater extent. .The electric bike will be powered by a battery, and the power will be supplied by the motor, which will
drive the other gear components. The main reason for using this E-bike is that it is user-friendly, economical, and
relatively inexpensive.

II. COMPONENTS OF E-BIKE

2.1 48V/20 AH Battery

A lithium-ion battery, sometimes known as a Li-ion battery, is a rechargeable battery type. Lithium-ion batteries are
widely used in portable electronics and electric vehicles, and their use in military and aerospace applications is
growing.

2.2 (24V, 800W) Motor

Copyright to IJARSCT DOI: 10.48175/IJARSCT-1941 165

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 1, September 2021

Impact Factor: 5.731

The motor is 800watt having capacity with maximum 3100rpm. Its specifications are as follows
Specifications:
 Rated Voltage: 48 Volt DC.
 Rated Power: 800W.
 Base motor RPM: 3100RPM.
 No Load Speed: 600 RPM.
 Rated Torque: 90Kg-cm.
 No load current: 2.8A
 Rated Current: 13.5A.
 Motor Weight: 7.80 kg.
 Gear Ratio: 1:6.

III. CHAIN DRIVE

A chain is a collection of interconnected links held together by steel pins. This configuration makes a chain more
durable, long-lasting, and effective at transmitting rotary motion from one gear to the next. The main advantage of
chain drive over traditional gear is that it can transmit rotary motion over a long distance using only two gears and a
chain, whereas traditional gear requires many gears to be arranged in a mesh to transmit motion. Two sprockets are
connected by a chain. The driver sprocket is one of the sprockets. The driven sprocket is the other type of sprocket. The
chain can carry motion and force from one sprocket to the next, and so from one shaft to the next. Power transmission
chains are chains that are used to convey motion and force from one sprocket to another.

IV. LED (LIGHT EMITTING DIODE)

The temperature of an LED diode has a significant impact on its performance. A forward voltage or current is applied
to the diode in a typical application condition, resulting in a light flux. The intensity of the light changes depending on
the current. The forward voltage-current relationship, on the other hand, is a function of the heat sink or board
temperature, and luminous flux fluctuates with diode temperature.
 Red - Battery Charging
 Orange - LED is on
 Yellow - Your Speed is less than 25 KMPH
 Green - ON
 Blue - Your Speed is more than 25 KMPH
 White - Blink Battery LOW
Copyright to IJARSCT DOI: 10.48175/IJARSCT-1941 166
www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 1, September 2021

Impact Factor: 5.731

V. DISC BRAKES

Drum brakes, as we all know, are insufficient in today's situations. They demand more effort from the rider,
necessitate frequent adjustments, and are less effective in general. Disc brakes are utilised to solve these issues. To stop
the car, a disc brake uses hydraulic pressure. It is extensively utilised in motorcycles, cars, trucks, buses, bicycles, and
other vehicles, but we will concentrate on motorcycles in this article.

VI. CONTROLLER

The controller, which comes in a number of forms, allows you to regulate the electric assistance on your electric bike
and is a crucial aspect of how they work. For convenience, the controller is mounted on the handlebar. Throttle-based
controllers use a basic throttle mechanism to operate. The throttle will be either a thumb-press or a twist-grip style. To
gain electric help with a throttle, simply pull back or press the throttle. Some electric bikes only require you to activate
the throttle, allowing you to ride without pedaling. Electric bikes are, for the most part, simple to operate, ride, and
maintain. In general, they require less maintenance than a regular bicycle.

VII. WORKING
The present study comprised of two separate procedures; firstly, the data for the operating conditions of the electric
two-wheelers were collected by conducting primary surveys among the vehicle owners with formatted questionnaires.
The daily travel data have been recorded from the available odometers, over a month. The energy consumption data
were also provided by the owners and were also estimated from the battery capacity. A daily pattern of utilization of
electric two-wheelers was thus obtained. Secondly, the real time performance of an electric two-wheeler was studied as
a part of the experiment, for obtaining the energy consumption data and the performance characteristics with the
present traffic conditions.

The experiment has been conducted based on a low speed electric two-wheeler available in the market. The data from
the electric two-wheeler study has been analyzed to obtain its pattern of energy consumption at different loading
conditions, its performance in present day traffic and to check the suitability with different traffic conditions. Working
Copyright to IJARSCT DOI: 10.48175/IJARSCT-1941 167
www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 1, September 2021

Impact Factor: 5.731

of E-bike is like having a power starter which starts the Controller. LED of Green gives signal as bike it been started.
Yellow LED also starts to blink as our speed is less then 25KMPH .Then we throttle the accelerator the controller unit
starts the motor and thus bike moves. LED of Blue color starts to blink as our will be more than 25 KMP

VIII. CALCULATIONS AND RESULTS

For Speed 60 km/hr
Motor Cycle weight = 111 Kg
Gross Weight = 111+70 =180 kg
Tyre Rim Diamete = 18×25.4 = 457.2 mm = 458 mm
Front Tyre = 2.75 ×18-42P
Rear Tyre = 3.00 ×18-52P

8.1 Requirements
1. Gross Weight Of Vehicle
2. Required Speed
3. Wheel Size
4. Efficiency Of Motor
5. Rolling Resistance
6. Coefficient Of Drag
7. Area Of Vehicle
8. Density Of Air
9. Forces
 Total Force(fq)
 Drag Force (fd)
 Acceleration Force (fa)
 Rolling Force (fr)
 Gradient Force (fg)

Copyright to IJARSCT DOI: 10.48175/IJARSCT-1941 168

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 1, September 2021

Impact Factor: 5.731

8.2 Dimensions of Vehicle

 Tyre Height = (2.75×25.4) × 0.82 = 57.277 mm = 58 mm
 Tyre Diameter = 457.2+57.277+52.277 = 571. 754 mm = 572 mm
 Tyre Radius = 572 / 2 = 286 mm
 Tyre Circumference = 2× 3.14×286 = 1796.99 mm
 Linear Wheel Travel = 1.80 m
 Speed Of Vehicle = (60×1000/3600) = 16.66 m/s
 RPM Required = Speed/ wheel travel =(16. 7×60)/1.80 = 556.67 rpm
 Mass (m) = 181 kg
 Gravity (g) = 9.81 m/s^2
 Velocity (v) = 16.7 m/s
 Coefficient Of Rolling Resistance = 0.02
 Air Density = 1.2 kg/m^3
 Air Drag = 0.82 _________ (For Electric Bike)
 Fr = m.g.Cr =181 × 9.81 ×0.0.0 = 35.512 N
 Fd = 1/2× Air density ×( d × Front Area × V^2) = 1/2 × 1.2 ×[ 0.8 × 1.2 × ( 16.7^2)] = 160.64 N
Ft=Fr+Fd+Fa+Fg
So, Neglecting acceleration & gradient forces for designing the 60kmph speed.No need to calculate high acceleration
for gradient no need to run high speed
Ft = Fr + Fd
= 35.512 + 160.64
Ft = 196.152 N
Power = Ft × V = 196.152 × 16.7 = 3275.7384
Watt = 3.27 KW
Motor Selection
P = ( 2 × 3.14 × 556.67 × T/ 60)
Where, P = 3275.7384
3275.7384 = ( 2 × 3.14 × 556.67 × T/60)
T = 56.19 Nm
For 3000 RPM
 BLDC Motor Standard RPM = 3000 RPM
 Required RPM at Wheel = 556.67 RPM
 Reduction Ratio = ( 3000/556.67) = 5.38
 Motor Selection for 1500 W
P = 2 × 3.14 × 1200 × T
Where, P = 1500
1500 = 2× 3.14 ×1200× T
T = 12 Nm
Torque on wheel at 1200 RPM = 12×6 = 72 Nm
Motor Selection For 1000 W
1000 =( 2× 3.14× N×T/60)
T = 7.9 Nm
Torque on wheel at 1000 RPM =7.9 × 6 = 47.4 Nm
Reduction Ratio = ( 1000/ 556.67 ) = 1.79

Copyright to IJARSCT DOI: 10.48175/IJARSCT-1941 169

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 1, September 2021

Impact Factor: 5.731

IX. ADVANTAGES
 Less Maintenance required
 System is very stable.
 No fuel required
 Health benefits
 Pollution free ride
 Low cost
 No Noise

X. DISADVANTAGES
 Running range is depends on the life of battery.
 Batteries can’t be recycle
 Battery has a rather short lifespan
 Maintenance and repairs are costly
 E-bikes tend to have low resale value

XI. CONCLUSION
With the increasing consumption of nonrenewable resources such as petroleum and diesel, we are moving toward
renewable sources such as solar, hydroelectric electricity, and batteries. There are different methods for conserving
energy. One such mode of transportation is the electric bike; it is also a new mode of transportation that provides us
with a convenient mode of transportation. It is also a new mode of transportation that provides us with a simple mode
of transportation for people of all ages. It is a low-cost mode of transportation that everybody may afford. The motor in
this bike is highly efficient, and the battery bank is light and fast. The electric bike's most important feature is that it
does not use fossil fuels, which saves billions of dollars in foreign currency. The second most important feature is that
it produces no pollution, is environmentally friendly, and operates quietly. The most viable solution for reducing
environmental pollution is to ride an on-board electric bike. If there is an emergency, it can be charged using an AC
converter

ACKNOWLEDGEMENT
We take this opportunity here to thank all those who had helped us in making this project a reality. First of all we
express our deep gratitude to our project guide PROF.R.S.YADAV for his valuable support, help & guidance from time
to time during the project work. We are also grateful to our Head of Department, Dr. P. S. Purandare and Principal Dr.
S. M. Deshpande for giving us this opportunity to present this project report. We are highly indebted to Mr. R. S.
Yadav of Mechanical Department for granting us this project and for his guidance which we were privileged to receive.
We convey our heartfelt gratitude to him for taking out time from his busy schedule and leading us through this project.
Last but not the least; we would like to thank our entire teaching staff who assisted us directly or indirectly throughout
the duration of this project.

REFERENCES
[1]. Daniels M W and Kumar P R (2005), “The Optimal Use of the Solar Power Automobile”, Control Systems
Magazine, IEEE, Vol. 19,
[2]. John Connors (2007), “Solar Vehicles and Benefits of the Technology”, ICCEP Paper
[3]. Mangu, R., Prayaga, K., Nadimpally, B. and Nicaise, S. (2010) Design, Development and Optimization of
Highly Efficient Solar Cars: Gato Del Sol I-IV. Proceedings of 2010 IEEE Green Technologies Conference,
Grapevine, 15- 16 April 2010, 1-6.

Copyright to IJARSCT DOI: 10.48175/IJARSCT-1941 170

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 1, September 2021

Impact Factor: 5.731

[4]. Husain, I. (2005) Electrical and Hybrid Vehicles Design Fundamentals. CRC Press, Boca Raton, London,
New York and Washington DC.
[5]. Miller, T.J.E. (1989) Brushless Permanent Magnet and Reluctance Motor Drive. Clarendon Press, Oxford.
[6]. Trembly, O., Dessaint, L.A. and Dekkiche, A.-I. (2007) A Generic Battery Model for the Dynamic Simulation
of Hybrid Electric Vehicles. 2007 IEEE Vehicle Power and Propulsion Conference, Arlington, 9-12
September 2007

Copyright to IJARSCT DOI: 10.48175/IJARSCT-1941 171

www.ijarsct.co.in
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/316995569

Design and fabrication of electric bike

Article · March 2017

CITATIONS READS

15 67,466

5 authors, including:

Shweta V. Matey Kunjan Shinde

Lokmanya Tilak College of Engineering Wright State University
3 PUBLICATIONS 17 CITATIONS 1 PUBLICATION 15 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Kunjan Shinde on 10 September 2018.

The user has requested enhancement of the downloaded file.

International Journal of Mechanical Engineering and Technology (IJMET)
Volume 8, Issue 3, March 2017, pp. 245–253 Article ID: IJMET_08_03_027
Available online at http://www.iaeme.com/IJMET/issues.asp?JType=IJMET&VType=8&IType=3
ISSN Print: 0976-6340 and ISSN Online: 0976-6359

© IAEME Publication Scopus Indexed

DESIGN AND FABRICATION OF ELECTRIC

BIKE
Shweta Matey
Assistant Professor,
Department of Mechanical Engineering, University of Mumbai,
Lokmanya Tilak College of Engineering, Koparkhairne, Navi Mumbai, India

Deep R Prajapati, Kunjan Shinde, Abhishek Mhaske, Aniket Prabhu

Undergraduate Student,
Department of Mechanical Engineering, University of Mumbai,
Lokmanya Tilak College of Engineering, Koparkhairne, Navi Mumbai, India

ABSTRACT
The main gist of this paper is to give the exact view by bridling the various sources
of energy available to mankind. In today’s modernized world travelling is very essential
for human beings in order to protract in this world. And to do so his travelling should
be done in minimum possible way and in jiffy. This paper details about the Electric Bike
which runs on the battery thereby providing voltage to the motor. This paper
compromises with design and fabrication of Electric Bike which makes use of Electric
energy as the primary source and solar energy if possible by attaching solar panels. It
also highlights on the design aspects of the bike. There is a provision for a charging the
battery by ejecting it from the main system. The electrical power generated which is
used to run the bike can give better fuel economy compared to conventional vehicle,
better performance and also causes less pollution.
Key words: Travelling, Electric Bike, Electric Energy, Solar Panels, Fuel Economy.
Cite this Article: Shweta Matey, Deep R Prajapati, Kunjan Shinde, Abhishek Mhaske
and Aniket Prabhu, Design and Fabrication of Electric Bike, International Journal of
Mechanical Engineering and Technology, 8(3), 2017, pp. 245–253.
http://www.iaeme.com/IJMET/issues.asp?JType=IJMET&VType=8&IType=3

http://www.iaeme.com/IJMET/index.asp 245 [email protected]

 Design and Fabrication of Electric Bike

Notations
P = Power
N = revolution per min
T = Torque
σs = Shear stress
σt = Tensile stress
σut = Ultimate Tensile Stress
fos = Factor of safety

1. INTRODUCTION
Energy crisis is one of the major concerns in today’s world due to fast depleting resources of
petrol, diesel and natural gas. In combination with this, environmental decay is an additional
factor which is contributing to the depletion of resources which is an alarming notification. Our
paper proposes the solution for this above perilous problems. The system which we innovated
is the Electric Bike. This project has various benefits both to the members of the team and also
external benefits thereby making awareness of using alternative modes of transport. The
Electric Bike which works on the battery that is powered by the motor is the general mode of
transport for a local trip. The solar panels can be alternative source for this by adding it to the
system. The Electric bike which will be running on battery, the power is supplied by the motor,
thereby supplying this power to drive the other gear components. The main purpose of using
this E-bike is that it is user friendly, economical and relatively cheap. The efficiency of this
system undeniable compared to conventional modes of transport. The following table shows
the specification of various electric bikes used in few countries:

Table 1 Specifications of E-Bike in various countries

Type of Speed limit Weight in Age required in
Country Watt
bike in km/hr kgs yrs
Australia Pedal 25 250 None None
Canada Hand 32 500 None Various
China P/H 30 200 20 None
Norway Pedal 25 250 None None
Israel Pedal 25 250 30 14
UK Hand 27 250 40 14
Taiwan Hand 25 200 None None
US Hand 25 750 None None
China P/H 30 500 None None

1.1. What is Electric Bike?

The Electric bike is a bike which is driven with the help of battery which is coupled to electric
motor.
Main principle: It works on the principle that the electromotive force of an A.C. motor which
receives electrical energy stored in D.C. battery is converted with the help of D.C. to A.C.
converter.
Working medium: Here for the motivation of prime mover the chemical reaction takes place
from which an energizing current is evolved which is responsible for the working. The working
medium is sulphuric acid which is separated into columns of H ions and negative SO4 ions
when mixed with water. If the poles of the cell are connected by a load, the flow of the electrons
is from negative to positive. A bivalent positive lead is produced from neutral lead when

http://www.iaeme.com/IJMET/index.asp 246 [email protected]

 Shweta Matey, Deep R Prajapati, Kunjan Shinde, Abhishek Mhaske and Aniket Prabhu

combined with bivalent negative of SO4 group to form lead sulphate. This results due to scarcity
of electrons at negative pole. Through the electron supply a bivalent positive lead is produced
at positive pole from quadrivalent positive lead. A combination of SO4 comes into existence
thereby ruling the combination of O2 which leads to formation of PbSO4. The atoms of oxygen
and hydrogen from electrolyte are released together to form water thereby decreasing the
density of battery acid.
Operation: In this a DC waveform which is obtained id made sinusoidal due to operational
transistorized D.C. to A.C. amplifying circuit by switching the electric energy in the form of
electric current which flows from battery to D.C. to A.C. converter circuit. By using amplifier
circuit the small A.C. current is amplified again. In order to drive the circuit through the
condenser, this amplified current is fed to the stator winding of the A.C. motor. The condenser
which is used acts as a storage of electric energy and delivers at the time of requirement. The
sprocket wheel installed on motor shaft is driven by the motive power of the electric energy.
The rear sprocket wheel is being rotated by the chain drive mechanism on which the other two
remaining sprocket wheels are installed. The wheel is driven by the rear wheel installed on the
rear sprocket. Thus the electric bike is mobilized by using electric power.

2. COMPONENTS OF E - BIKE
The Electric bike consists of following components viz, DC motor, Frame, Platform, Battery,
Drive etc. (Barve, Design and Development of Solar Hybrid Bicycle, 2016)
1. Dc motor: The motor is having 250 watt. Capacity with maximum 2100 rpm. Its
specifications are as follows:
Current Rating: 7.5amp
Voltage Rating: 48 Volts
Cooling: Air – cooled
Bearing: Single row ball

Figure 1 DC Motor.
2. Frame: The Frame is made up of M.S. along with some additional light weight components.
The frame is designed to sustain the weight of the person driving the unit, the weight of load to
be conveyed and also to hold the accessories like motor. Also it should be design to bear and
overcome the stresses which may arise able to due to different driving and braking torques and
impact loading across the obstacles. It is drilled and tapped enough to hold the support plates.
3. Platform: The Platform is designed with robust base so that it can hold the load along with
the weight of the driving person uniformly. It is fabricated from Mild Steel at a specific angle
in cross section and welded with a sheet of metal of specific thickness. The platform’s alignment
is kept horizontal irrespective whether it is loaded or unloaded and this is directly bolted and
welded to the frame.
4. Battery: The battery also acts as a condenser in a way that it stores the electric energy
produced by the generator due to electrochemical transformation and supply it on demand.

http://www.iaeme.com/IJMET/index.asp 247 [email protected]

 Design and Fabrication of Electric Bike

Battery is also known as an accumulator of electric charge. This happens usually while starting
the system.

Figure 2 Battery
5. Chain Drive: A Chain is an array of links held together with each other with the help of steel
pins. This type of arrangement makes a chain more enduring, long lasting and better way of
transmitting rotary motion from one gear to another.

Figure 3 Chain Drive

The major advantage of chain drive over traditional gear is that, the chain drive can transmit
rotary motion with the help of two gears and a chain over a distance whereas in traditional many
gears must be arranged in a mesh in order to transmit motion.
6. Braking System: For the braking system it is convenient to use braking system used in band
brake system which consist of spring loaded friction- shoe mechanism, which is driven with the
help of hand lever.

S
e

Figure 4 Braking System

7. Sprockets: The chain with engaging with the sprocket converts rotational power in to rotary
power and vice versa. The sprocket which looks like a gear may differ in three aspects::

http://www.iaeme.com/IJMET/index.asp 248 [email protected]

 Shweta Matey, Deep R Prajapati, Kunjan Shinde, Abhishek Mhaske and Aniket Prabhu

Figure 5 Various types of Sprockets

• Sprockets have many engaging teeth but gears have only one or two.
• The teeth of a gear touch and slip against each other but there is basically no slippage in case of
sprocket
• The shape of the teeth are different in gears and sprockets.
(Barve, Design and Development of Solar Hybrid Bicycle, March 2016)

3. DESIGN OF ELECTRIC BIKE

Here we have used permanent magnet self generating motor with 250 watt power and 2100rpm.
The motor runs on 48volts and 7.5amps power source. This motor can reach a peak current
during starting equal to 15 amps. (Barve, Design and Development of Solar Hybrid Bicycle,
March 2016)
P = 2 x 3.14 x N x T /60
250 = 2 x 3.14 x 2100 x T /60
T = 1.13 N m = 1136 N-mm
Reduction in chain drive
R chain = 66/11 = 6:1
Torque at wheel shaft = T x R chain = 1136 x 6 = 6820 N mm
Speed of wheel shaft = 2100 /6 = 350 rpm

DESIGNING OF SHAFT
BENDING:
The force which develops across a specific cross section of the shaft, it generates stress at that
point of cross section that are subjected to maximum loading. This internal or resisting moment
gives rise to the stress called as bending stresses.
Torsion: When the shaft which is twisted by the couple such that the axis of that shaft and the
axis of the couple harmonize, that shaft is subjected to pure torsion and the stresses generated
at the point of cross section is torsion or shear stresses.
Combined Bending and Torsion: In actual practice the shaft is subjected to combination of
the above two types of stresses i.e. bending and torsion. The bending stresses may occur due
any one of the following reasons:
1. Weight of belt
2. Pull of belts
3. Eccentric Mounting of shafts/gears
4. Misalignment of shafts/gears

http://www.iaeme.com/IJMET/index.asp 249 [email protected]

 Design and Fabrication of Electric Bike

On contrary, the torsional movement occurs due to direct or indirect twisting of the shaft.
Hence at any given point on cross-section of the shaft, the shaft is subjected to both bending
and torsional stresses simultaneously. Following stresses are taken in consideration while
designing the shaft:

Shaft design
T = 36000 N mm
T = 3.14 / 16 x σs x d3
Fs allowable = 80 N/mm2
6820 =3.14 x σs x d3/16
σs = 34.73 N/mm2
Material = C 45 (mild steel)
σ ut = 320 N/mm2 ------------ PSG design data book.
factor of safety = 2
σt = σb = σ ut/ fos
= 320/2
= 160 N/mm2
σs = 0.5 σt
= 0.5 x 160
= 80 N/mm2
σs is less then allowable so our shaft design is safe.

Design of Sprocket and Chain for Electric Bike

We know,
TRANSMISSION RATIO = Z2 / Z1 = 66/11 = 6
For the above transmission ratio number of teeth on pinion and the number of teeth sprocket
is in the range of 21 to 10, so we have to select number of teeth on pinion sprocket as 11 teeth.
So, Z1 = 11 teeth

SELECTION OF PITCH OF SPROCKET

The pitch is decided on the basis of RPM of sprocket.
RPM of pinion sprocket is variable in normal condition it is = 2100 rpm
For this rpm value we select pitch of sprocket as 6.35mm from table.
P = 6.35mm

CALCULATION OF MINIMUM CENTER DISTANCE BETWEEN SPROCKETS

THE TRANSMISSION RATIO = Z2 / Z1 = 66/11 = 6 which is less than 7
Dia. of small sprocket,
Periphery = π × dia. Of sprocket
11 × 6.25 = π × D
D = 11 × 6.25 / π
D = 21.8 mm
Dia. of sprocket,
Periphery = π × dia. Of sprocket

http://www.iaeme.com/IJMET/index.asp 250 [email protected]

 Shweta Matey, Deep R Prajapati, Kunjan Shinde, Abhishek Mhaske and Aniket Prabhu

66 × 6.25 = π × D
D = 66 × 6.25/ π
D = 131.3 mm
So from table, referred from PSG Design Data book
The minimum centre distance between the two sprocket = C’ + (80 to 150 mm)
Where C’ = Dc1 + Dc2
2
C’= 131.3 + 21.8
2
C’ = 76.5 mm
MINIMUM CENTER DISTANCE = 76.5 + (30 to 150 mm) MINIMUM CENTER
DISTANCE = 170 mm

CALCULATION OF VALUES OF CONSTANTS K1 K2 K3 K4 K5 K6 – (with

help of PSG Design Data)
Load factor K1 = 1.25 (Load with mild shock)

Distance regulation factor K2 =1.25(Fixed center distance)

Center distance of sprocket factor K3 =0.8
Factor for position of sprocket K4 = 1
Lubrication factor K5 = 1.5 (periodic)
Rating factor K6 = 1.0 (single shift)

CALCULATION OF VALUE OF FACTOR OF SAFETY

For pitch = 6.35 & speed of rotation of small sprocket = 2100 rpm
Factor of Safety for this drive = 8.55
Calculation of Allowable Bearing Stress:
For pitch = 6.35 & speed of rotation of small sprocket = 2100 rpm
Allowable Bearing stress in the system = 2.87kg / cm2
=2.87 * 981/100 =28 N /mm2

Calculating Maximum Tension on Chain

Maximum torque on shaft = Tmax = T2 = 6820 N-mm
Where,
T1 = Tension in tight side
T2 = Tension in slack side
O1, O2 = center distance between two shaft
Sin ∝ = R1 - R2
O1O2
Sin ∝ = 65.65 - 10.9
170
Sin ∝ = 0.33
∝ = 18.78

http://www.iaeme.com/IJMET/index.asp 251 [email protected]

 Design and Fabrication of Electric Bike

TO FIND θ
θ = (180 –2∝ ) X 3.14/180
θ = (180 –2*18.78) X 3.14/180
θ = 2.48 rad
According to this relation,
T1/T2 = eµθ
T1/T2 = e0.35 x 2.48
T1 = 2.38T2
We have,
T = (T1 – T2) X R
6820 = (2.38 T2 – T2) X 65.65
T2 = 75.27 N
T1 = 2.38 X 75.27
T1 = 179.16 N
So tension in tight side = 179.16 N
We know,
Stress = force / area x 2
Stress induced = 179.16/ (3.14 * 32 / 4) x 2
Stress induced = 12.67 N /mm2
As induced stress is less than allowable stress =28N /mm2design of sprocket is safe.

4. ADVANTAGES
• Easy to commute with low fatigue.
• Less maintenance cost.
• Normal Drag/Pedal is possible when power is not in use.
• Deployable batteries – can be taken inside house.
• Cost of the unit is very low.
• Easy to carry since it is portable.
• Less energy consumed.
• High efficiency can be obtained if inverter is used.
• If using solar panel, free utilization of energy can be done.
(T.Bhavani, April 2015)

5. DISADVANTAGES
• High intensity of wind load.
• High centre of gravity.
• Cannot tolerate drastic changes in environment.
• Needs Periodic Monitoring.
(T.Bhavani, April 2015)

http://www.iaeme.com/IJMET/index.asp 252 [email protected]

 Shweta Matey, Deep R Prajapati, Kunjan Shinde, Abhishek Mhaske and Aniket Prabhu

6. CONCLUSION
With the increasing consumption of natural resources of petrol, diesel it is necessary to shift
our way towards alternate resources like the Electric bike and others because it is necessary to
identify new way of transport. Electric bike is a modification of the existing cycle by using
electric energy and also solar energy if solar panels are provided, that would sum up to increase
in energy production. Since it is energy efficient, electric bike is cheaper and affordable to
anyone. It can be used for shorter distances by people of any age. It can be contrived throughout
the year. The most vital feature of the electric bike is that it does not consume fossil fuels
thereby saving crores of foreign currencies. The second most important feature is it is pollution
free, eco – friendly and noiseless in operation. For offsetting environmental pollution using of
on – board Electric Bike is the most viable solution. It can be charged with the help of AC
adapter if there is an emergency. The Operating cost per/km is very less and with the help of
solar panel it can lessen up more. Since it has fewer components it can be easily dismantled to
small components, thus requiring less maintenance.

REFERENCES
[1] Aikenhead, G. S. (2011). Bicycle Applications for On-Board Solar Power Generation. 9,10.
[2] Barve, D. S. (2016). Design and Development of Solar Hybrid Bicycle. International
Journal of Current Engineering and Technology, 377,378,379,380.
[3] Barve, D. S. (March 2016). Design and Development of Solar Hybrid Bicycle. International
Journal of Current Engineering and Technology, 378,379.
[4] Barve, D. S. (March 2016). Design and Development of Solar Hybrid Bicycle. International
Journal of Current Engineering and Technology, 380.
[5] FOGELBERG, F. (2014). Solar Powered Bike Sharing System. Goteberg, Sweden: Viktoria
Swedish ICT.
[6] FOGELBERG, F. (2014). Solar Powered Bike Sharing System with. Goteborg, sweden:
Viktoria Swedish ICT.
[7] GOODMAN, J. D. (2010, Jan 31). An Electric Boost for Bicyclists. The New York Times.
[8] Prof. Palak Desai, P. D. (June 2016). Design and Fabrication of Solar Tri Cycle.
International Journal of Engineering Sciences & Research, 664.
[9] Hameed Majeed Saber and Deepak Lal, Assessment of Solar Energy Distribution For
Installing Solar Panels Using Remote Sensing & GIS Techniques, International Journal of
Advanced Research in Engineering and Technology (IJARET) Volume 5, Issue 10, October
(2014), pp. 157-164.
[10] Srijan Manish, Jitendra Kumar Rajak, Vishnu Kant Tiwari and Rakesh, Quad Bike Design
And Simulation: A Pre-Manufacturing Methodology, International Journal of Advanced
Research in Engineering and Technology (IJARET)Volume 5, Issue 6, June (2014), pp. 68-
76
[11] T.Bhavani. (April 2015). Novel Design of Solar Electric Bicycle with Pedal. International
Journal & Magazine of Engineering, 108.

http://www.iaeme.com/IJMET/index.asp 253 [email protected]

View publication stats

 Available online at www.sciencedirect.com

ScienceDirect
Energy Procedia 101 (2016) 774 – 781

71st Conference of the Italian Thermal Machines Engineering Association, ATI2016, 14-16
September 2016, Turin, Italy

Design and Development of an Innovative E-Bike

C. Abagnalea, M. Cardoneb*, P. Iodicea, R. Marialtoc, S. Stranoa, M. Terzoa, G. Vorraro
a
Department of Industrial Engineering, University of Naples Federico II, Napoli Italy
b
Department of Chemical, Materials and Production Engineering, University of Naples Federico II, Napoli Italy
c
Istituto Motori, CNR, Napoli Italy

Abstract

A new model of power-assisted bicycle has been designed, set up and tested. The main innovative solutions for the pedelec
prototype are described in the present paper: the electric motor position; the new mechanical transmission; the low cost
measurement system of the driving torque; the special test rig. Differently from a common approach, in which the electric motor
is located on one of the three hubs of the bicycle, the idea of the pedelec prototype consists of an electrical motor in the central
position that, by means of a bevel gear, transmits the torque on the central hub. The other innovative solution is represented by
the motion transmission from the motor to the pedal shaft, achieved by two different gearboxes: the first one is a planetary
gearbox and the second one is a simple bevel gear. The pedelec prototype contains also a new low cost measurement system of
the driving torque based on a strain gauge load cell located on one side of the rear wheel, between the hub and the frame.
Moreover, a commercial cycling simulator has been suitably modified in order to properly install the different sensors for the
measurement of the performance of the pedelec. The test rig is able to reproduce an aforethought route or paths acquired during
road tests, to measure the performance of the e-bike in terms of instantaneous power and speed. The experimental test rig can
simulate the resistant torque of a predetermined track and it aims to test and to optimize the control strategy available on the
electronic control unit. The authors have also conducted an environmental analysis of the developed pedelec, in particular
comparing the e-bike with a thermal moped, in terms of environmental impact.
© 2016The
© 2016 TheAuthors.
Authors. Published
Published by Elsevier
by Elsevier Ltd. Ltd.
This is an open access article under the CC BY-NC-ND license
Peer-review under responsibility of the Scientific Committee of ATI 2016.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the Scientific Committee of ATI 2016.
Keywords: Electrically assisted bicycle, pedelec, pedal electric cycle

* Corresponding author. Prof. Massimo Cardone, Tel.: +39 081 7683675; fax: +39 081 2394165
E-mail address: [email protected]

1876-6102 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the Scientific Committee of ATI 2016.
doi:10.1016/j.egypro.2016.11.098
 C. Abagnale et al. / Energy Procedia 101 (2016) 774 – 781 775

1. Introduction

The large use of the travelling vehicles has increased the problems connected to the air quality and to the use of
petroleum [1, 2]. The human sensibility for the energetic and environmental problems is encouraging the research in
alternative solutions for the automotive field, as multiple-fueling, hybridization and electrification. At the same time,
particularly as concerns urban areas, new standards have imposed substantial modifications in the mobility. In this
context, a vehicle as the electrically assisted bike [3 – 5] can be considered a promising alternative vehicle for both
personal mobility and goods delivery, especially for small and medium distances: an assisted bike is able to move
with an average speed equal to the typical one of the town traffic but it requires energy for its mobility that is very
close to the necessary energy for the displacement of the transported people.
The electrically assisted bikes are normally powered by rechargeable battery, and their driving performance is
influenced by battery capacity, motor power, road types, operation weight, control, and, particularly, by the
management of the assisted power. A classification of the electrically assisted bikes can be based on two categories:
a first kind is represented by a pure electric bike [6 - 8], which integrates electric motor into bicycle frame or wheels,
and it is driven by motor force just using a handlebar throttle; a second kind is a power-assisted bicycle, or called
pedelec [9, 10] hereafter, which is a human–electric hybrid bicycle that supports the rider with electric power only
when the rider is pedaling. The pedelecs are characterized by a driving torque due to both an electric motor torque
and a rider one. Consequently, the management of the assistance torque is of particular interest in order to reach the
desired performances in terms of driveability and comfort.
The present paper deals with the activity carried out on a prototype of an innovative power-assisted bicycle [11-
19], designed at the Department of Industrial Engineering of the University of Naples Federico II: a pedelec
characterized by an innovative layout of the electrical assistance and a new low cost measurement system of the total
driving torque (rider torque + electrical motor torque).
The paper, that can be considered a synthesis of a wide activity including design, modelling, control and testing,
is organized as follows: Section 2 presents the new pedelec, with its main innovative solutions; Section 3 describes
the system modeling and control; the test rig suitably set up is detailed in Section 4; the results are illustrated in
Section 5. Finally, Section 6 contains the conclusions and the future developments.

2. The Pedelec Prototype

A prototype of an innovative power-assisted bicycle has been

adopted for the research activity. The new vehicle can be classified as
a power-assisted bicycle, or pedelec [9, 10]: it is a human–electric
hybrid bicycle that supports the rider with electric power only when
the rider is pedaling. The developed pedelec presents some innovative
solutions, resumed in three main aspects, as follows.

2.1. The central motor

Fig. 1 shows a photo of the pedelec prototype [11-13], where it is
possible to note the electric motor in the central position. In fact,
differently from a common approach, in which the electric motor is
located on one of the three hubs of the bicycle, a basic idea of the
pedelec prototype consists of an electrical motor that, by means of a
bevel gear, transmits the torque on the central hub. Fig. 1. Pedelec prototype

2.2. The new mechanical transmission

The motion transmission from the motor to the pedal shaft is achieved by two different gearboxes: the first one is
a planetary gearbox and the second one is a simple bevel gear. Fig. 2 shows a scheme of the second gearbox location.
Derailleur gears are adopted in the pedelec drivetrain. The gear ratio depends on the ratio of the number of teeth on
the rear sprocket to the number of teeth on the chain ring (equal to 39). The available gear ratios are listed in Table 1.
776 C. Abagnale et al. / Energy Procedia 101 (2016) 774 – 781

Table 1. Pedelec gear ratios

Number of teeth Gear Ratio
28 1 0.718
24 2 0.615
21 3 0.538
18 4 0.462
16 5 0.410
14 6 0.359

Fig. 2. Mechanical transmission between the motor shaft and the pedal shaft

2.3. The new measurement system

As concerns the measurements realized on the commercial pedelecs, they typically depend on the kind of control.
Particularly, as regards the feedback controller, the rider torque is often required. Indeed, in this kind of controlled
assisted bicycles, the contribution to the traction force of the vehicle by the electric motor can be made proportional
to the contribution of the rider. For instance, the motor could be controlled in order to double the effort made by the
cyclist. Moreover, the distribution between the motor power and the human power can further be controlled as
function of other informations (speed, different assistance modes selected by the user). With reference to the
measurements of the rider torque, two different devices are often employed. A first measurement system consists of
a torsiometer mounted on the pedal hub [20]. Instead, a second device is based on a Hall effect sensor useful to
measure the chain strength and located between the rear wheel hub and the frame [21].
The pedelec prototype contains a new low cost measurement system of the driving torque based on a strain gauge
load cell located on one side of the rear wheel, between the hub and the frame (Fig. 3). On the other side of the
wheel, a hinge constraint has been adopted in order to allow rotation in a plane parallel to the road surface. The
strain gauge load cell is linked to the frame and to the hub by means of three hinges (A, B and C in Figure 3), in
order to make the load applied to the load cell as close as possible to the chain strength.
The new measurement system, whose validation is the object of papers by the same authors [11-13], aims to
know the chain strength and consequently, known the radius of the central crown, the total torque applied to the
central hub and due to the contribution of both the electric motor and the rider.

Fig. 3. Detail of the measurement system

3. System Modelling and Controller Design

In a human–electric hybrid bicycle that supports the rider with electric power only when the rider is pedalling, the
management of the assistance torque (due to both an electric motor torque and a rider one) is of particular interest in
order to reach the desired performances in terms of driveability and comfort. The procedure followed in the pedelec
control, differently from the common approaches available in literature [22-28], is based on an optimal control of
the assistance torque. The system model is based on the longitudinal vehicle dynamics and on the electrical motor
one: the pedelec dynamics can be expressed as a combination of the bicycle longitudinal dynamics and the electric
motor dynamics (Fig. 4).
 C. Abagnale et al. / Energy Procedia 101 (2016) 774 – 781 777

The model has been derived and employed for the

synthesis of an optimal control of the assistance torque.
Moreover, an additional integral state has been adopted to
make zero the tracking error in steady state conditions.
The optimal pedelec torque control (OPTC) proposed
control method is based on a torque control designed via
an optimal approach to achieve multi-objective
performances regarding the external disturbance input,
control signal magnitude, and velocity tracking error. The
performance of the methodology has been evaluated
applying the proposed control to a model of the pedelec Fig. 4. Longitudinal components of forces acting on the pedelec
prototype, characterized by the measurement of the total
torque (rider torque and electric motor one) employed as feedback for the control. To this aim, a mathematical
model of the bicycle, equipped with the electric motor, has been developed, object of papers by the same authors
[14, 16, 17]. The equations of the model (a second order nonlinear system that constitutes the basis for the controller
synthesis) have been formulated representing the combined dynamic model of the whole system including the
bicycle and the electric motor. The full system dynamics has been used for the power-assistance controller design.
The OPTC consists in a pedelec torque control in which the input is the voltage to the electric motor and the
measurements are the total torque Tmis, the motor angular velocity and the motor current. The gradient resistance
accounts for a major part of the environmental disturbance, consequently, the nonlinear aerodynamic drag force is
usually negligible in low speed riding and mild head wind condition. For this reason, the linear model of the whole
system, constituted by the electric motor and the bike, has been obtained neglecting the aerodynamic resistance
force. This unmodelled phenomenon can be compensated thanks to the robustness of the controller with respect to
disturbances and model uncertainties. The scheme of the linearized model is reported in Fig. 5.

Fig. 5. Block diagram of the linearized pedelec model.

The state feedback OPTC design has been accomplished in order to minimize a weighted squared state error and
control effort. The controller strategy consists in finding the input voltage to the electric motor in order to follow a
target torque proportional to the actual rider one. To this aim, the total torque measurement from the novel torque
sensor must be elaborated in order to obtain the reference motor current.The controller effectiveness is influenced by
parameters that can vary, such as the total mass, the transmission ratio, etc. Also unmodelled phenomena could
affect the performance of the controller. The feedback action of the optimal control guarantees stability and
robustness in presence of limited variations of the controller parameters and unmodelled effects. This feature is
typical of the feedback controller. Moreover, the additional control action, based on the integral of the current error,
provides an improvement of the controller robustness. The feedback for the control is constituted by a new
measurement system of the driving torque that characterizes the proposed bicycle. The optimal controller, developed
starting from the linearized model, has been tested on the full nonlinear pedelec model.

3.1. Performance Evaluation

In order to evaluate the performances of the proposed approach, simulations have been carried out, modelling the
human controller as a proportional-integral action finalized to track a desired velocity.
The performances have been compared with the ones referred to a traditional assistance approach and the results
demonstrate that the proposed approach provides improvements in terms of riding comfort and energy employment.
778 C. Abagnale et al. / Energy Procedia 101 (2016) 774 – 781

The tracking performances and the disturbance rejection

capability have been tested in simulation, employing a rider
action modelled as a velocity controller based on a
proportional-integral action. In the simulation study, a rider,
that aims to follow a desired velocity, has been considered: the
chosen rider is a young man. So, the action provided by the
Fig. 6. Block diagram of the rider action. rider has been modelled as a torque dependent on the actual
pedelec velocity and the desired one, by means of a
proportional and an integral action. Fig. 6 shows a schematic diagram of the rider action. The rider coefficients Kpr
and Kir have been chosen in order to simulate the reaction time of the rider to the velocity response. A slope profile
for testing the controller has been adopted. The desired pedelec velocity, imposed by the rider, has been fixed to 5
km/h. Different test cases have been considered for the control verification:
x In the first case, the OPTC has been evaluated considering the nominal vehicle parameters.
x The influences of the total vehicle mass and the rear gear ratio on the OPTC performance have been
investigated in the second and the third case, respectively.
x In the fourth case, the comparison between the OPTC results with those obtained with a classical
pedelec control system has been performed.
The results highlight that the proposed approach provides reduced tracking errors and good robustness, moreover,
it performs better than classical pedelec assistance systems [19].

4. Test Rig and Measurement System

The experimental test rig has been realized starting from a commercial cycling simulator provided by ELITE srl
(Fig. 7). The commercial cycling simulator has been suitably modified in order to properly install the different
sensors for the measurement of the performance of the pedelec [18].

Figure 7. Detail of the cycling simulator.

Different paths can be simulated by means of the control software that manages the cycling simulator. The path
can be realized in different ways, assigning one of the following relationships:
x power – time;
x slope – time;
x slope – distance;
x GPS data.
The motor driver used to perform the experimental tests implements a sensorless control algorithm and it
determines when to commutate the motor drive voltages by sensing the BEMF voltage on an undriven motor
terminal during one of the drive phases. The driver power rating is 250W on 36V nominal voltage with maximum
current limited by hw at 7A and 20kHz PWM switching frequency. The control stage is based on a dsPIC DSC 16
bit microcontroller while the power stage is based on TI-DRV8332 integrated three phase motor drivers with an
advanced protection system. The available measurements on the test rig are:
 C. Abagnale et al. / Energy Procedia 101 (2016) 774 – 781 779

x the rear wheel angular velocity (by a magnetic pickup installed on the wheel with one pulse per revolution)
x the motor angular velocity (by three Hall effect sensors)
x the pedal shaft angular velocity (by a magnetic pickup installed on the pedal gear with five pulses per
revolution)
x the motor current (by Hall effect sensor)
x chain strength (by a strain gauge load cell).
The experimental data have been acquired by a DS1103 real-time board equipped with a 16-bit A/D and D/A
converter.

4.1. Experimental Tests

Preliminary experimental tests have involved the

numerical relationship between the torque applied to the
pedal shaft and the chain strength measurement. In this
way, it is possible to adopt the chain strength sensor as a
pedal shaft torque sensor.
The application of a known load Fp in correspondence
of the pedals (Fig. 8) determines a known value of the
pedal shaft torque: this value is correlated to the chain
strength measurement. The longitudinal tire-road force
generated by the pedal torque is balanced by the
constraint due to the load cell.
The experimental results are shown in Fig. 9, where
the measured chain strengths (called Fs) and the known Figure 8. Experimental setup for the estimation of the torque
torques applied to the pedal shaft are reported for all the acting on the pedal starting from the chain strength sensor.
rear gear ratios. The curves reported in Fig. 9 have been
implemented in the measurement system, in order to estimate the torque applied to the pedal shaft from the chain
strength measurement. In this way, it is possible to estimate the power at the central hub starting from the applied
torque and the pedal angular velocity.
Two experimental tests have been performed, in order to analyze the pedelec dynamics for a real urban riding. To
this aim, the test rig has been adopted to simulate two different on-road test tracks (EP1 and EP2), characterized by
the same origin and destination (and then the same cumulated climbing), but by different elevation profiles versus
distance. The total distance of the two trips amounts to 1.8 km for EP1 and to 2.2 km for EP2. The altitude profiles,
measured against distance, are reported for the two test tracks in Fig. 10.
For both experiments, all the available measurements have been acquired and some of them have been compared
with the ones provided by the training system software.

80 Test Tracks:
EP1
70 EP2
Gear Number: 60
Elevation (m)

50

40

30

20

10

0
0 0.5 1 1.5 2 2.5
Distance (km)

Figure 9. Results of the chain strenght sensor characterization Figure 10. Altitude profiles versus distance for itineraries
780 C. Abagnale et al. / Energy Procedia 101 (2016) 774 – 781

The authors [12-15] have evaluated the pedelec velocity acquired against the travelled distance for both the trips, the
instantaneous values of the gear ratios measured against the distance for the two itinerary, the pertinent profiles
against the distance travelled, the measured motor angular velocities for both experiments.

4.2. Energy Analysis

Fig. 11 shows the values of the electric power obtained by multiplying the input motor current for the battery
voltage (36 V). The values of the total power, instead, are calculated by the measurements of the chain strength and
the pedal shaft angular velocity. These results have been reached [12-15] starting from the evaluation of the motor
current values against the distance (values acquired by a hall effect sensor positioned downstream of the battery),
considering the chain strength versus distance and the overall power and the electric power acquired versus distance,
for the two itineraries.
In Fig. 11 and Fig. 12, the energies against the distance are obtained, for the two test tracks, by integration of the
instantaneous electric power and the total power on the overall distance. By observing these two figures, it is clear
that, while the electric energy for the two itineraries is very similar, the total energy required by the itinerary EP2 is
higher than the other one. This is due to the higher distance of the itinerary EP2 that affects mostly the power to
exceed the air strength and the power required to overcome the rolling resistances.
20 30
EP1 EP1
EP2 EP2
25
15
Electric energy (Wh)

Total energy (Wh) 20

10 15

10
5
5

0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
Distance (km) Distance (km)

Fig. 11. Electric energy versus distance for itineraries EP1 and EP2 Fig. 12. Total energy versus distance for itineraries EP1 and EP2

5. Conclusions

The present paper deals with a wide activity carried out on a prototype of an innovative power-assisted bicycle.
After the design and the modeling of the vehicle characterized by some innovative solutions, the control has
been implemented and testing on a suitable test rig has been carried out. The feedback for the control is constituted
by a new measurement system of the driving torque that characterizes the proposed bicycle. The results highlight
that the proposed approach provides reduced tracking errors and good robustness, moreover, it performs better than
classical pedelec assistance systems.
The preliminary tests have been useful for the appraisal of power and energy requests of a pedelec under real
driving conditions. The experimental test rig has simulated the resistant torque of these predetermined tracks in
order to test and to optimize the control strategy available on the electronic control unit. In these conditions, it has
been possible to measure the performance of the e-bike in terms of instantaneous power and speed versus distance,
by the installed sensors and data acquisition system.
The experimental results have allowed to compare the total energy and the electric one for both the tracks, so,
individualizing the optimum trip.
This study, then, has provided several results and guidelines that can assist for such improvements in the
performance of electric bicycles. Future developments will concern the design of several control strategies by means
of a hardware in the loop procedure.
 C. Abagnale et al. / Energy Procedia 101 (2016) 774 – 781 781

Acknowledgements
This research has been financially supported by MIUR (Ministero Istruzione Università e Ricerca) under the grant
named PON04a3_00408 "Bicicli e Tricicli elettrici a pedalata assistita di nuova generazione”.

References

[1] T. Onoda and T. Gueret, Fuel efficient road vehicle non-engine components: Potential saving and policy recommendations, Int. Energy
Agency Inf. Paper, 10 (2007), pp. 1–24.
[2] Iodice P, Senatore A. Appraisal of pollutant emissions and air quality state in a critical Italian region: Methods and results. Environmental
Progress and Sustainable Energy 2015;34:1497-1505
[3] H. Seki, K. Ishihara, and S. Tadakuma, Novel regenerative braking control of electric power-assisted wheelchair for safety downhill road
driving, IEEE Trans. Ind. Electron., 56 (5), (2009), pp. 1393–1400.
[4] C. C. Tsai, H. C. Huang, and S. C. Lin, Adaptive neural network control of a self-balancing two-wheeled scooter, IEEE Trans. Ind.
Electron., 57 (4) (2010), pp. 1420–1428.
[5] A. Emadi, Y. J. Lee, and K. Rajashekara, Power electronics and motor drives in electric, hybrid electric, and plug-in hybrid electric
vehicles, IEEE Trans. Ind. Electron., 55 (6) (2008), pp. 2237–2245.
[6] P. Fairley, China’s cyclists take charge: Electric bicycles are selling by the millions despite efforts to ban them, IEEE Spectr., 42 (6) (2005),
pp. 54–59.
[7] N. Somchaiwong and W. Ponglangka, Regenerative power control for electric bicycle, in Proc. IEEE Int. Joint Conf. SICE-ICASE, 2006,
pp. 4362–4365.
[8] M. J. Yang, H. L. Jhou, B. Y. Ma, and K. K. Shyu, A cost-effective method of electric brake with energy regeneration for electric vehicles,
IEEE Trans. Ind. Electron., 56 (6) (2009) pp. 2203–2212.
[9] D. Schneider, Easy rider: Convert your bicycle to a human–electric hybrid, IEEE Spectr., 46 (9) (2009), pp. 26–27.
[10] W. Du, D. Zhang, and X. Zhao, Research on battery to ride comfort of electric bicycle based on multi-body dynamics theory, in Proc. IEEE
Int. Conf. Autom. Logistics, Aug. 2009, pp. 1722–1726.
[11] C. Abagnale, M. Cardone, P. Iodice, S. Strano, M. Terzo, G. Vorraro, Analysis of a New Measurement System of the Chain Strength for
Electrically Assisted Bicycles, Proc. of the ASME 12th Biennial Conference on Engineering Systems Design and Analysis (ESDA2014),
ESDA2014-20364, Vol. 1, Copenhagen, Denmark, June 25–27, 2014.
[12] P. Iodice, C. Abagnale, M. Cardone, S. Strano, M. Terzo, G. Vorraro, Performance Evaluation and Environmental Analysis of an
Electrically Assisted Bicycle Under Real Driving Conditions, Proc. of the ASME 12th Biennial Conference on Engineering Systems Design
and Analysis (ESDA2014), ESDA2014-20438, Vol. 1, Copenhagen, Denmark, June 25–27, 2014.
[13] C. Abagnale, M. Cardone, P. Iodice, S. Strano, M. Terzo, G. Vorraro, Theoretical and Experimental Evaluation of a Chain Strength
Measurement System for Pedelecs. Engineering Letters , 22 (3) (2014), pp. 102 – 108.
[14] C. Abagnale, M. Cardone, P. Iodice, S. Strano, M. Terzo, and G. Vorraro, Derivation and Validation of a Mathematical Model for a Novel
Electric Bicycle, Lecture Notes in Engineering and Computer Science: Proceedings of The World Congress on Engineering 2015, 1-3 July,
2015, London, U.K., pp. 808-813.
[15] C. Abagnale, M. Cardone, P. Iodice, S. Strano, M. Terzo, G. Vorraro, Power requirements and environmental impact of a pedelec. A case
study based on real-life applications, Environmental Impact Assessment Review, 53 (7) (2015), pp. 1–7, doi:10.1016/j.eiar.2015.02.003.
[16] C. Abagnale, M. Cardone, P. Iodice, S. Strano, M. Terzo, G. Vorraro, Model-based control for an innovative power-assisted bicycle, Energy
Procedia, 81 (2015), pp. 606 – 617
[17] C. Abagnale, M. Cardone, P. Iodice, S. Strano, M. Terzo, G. Vorraro, A dynamic model for the performance and environmental analysis of
an innovative e-bike, Energy Procedia, 81 (2015), pp. 618 – 627
[18] C. Abagnale, M. Cardone, P. Iodice, S. Strano, M. Terzo, G. Vorraro, Design and Development of a Test Rig for E-bike Performance
Evaluation, Proc. of the 12th International Conference on Engines & Vehicles, Capri, Italy, September, 2015
[19] M. Cardone, S. Strano, M. Terzo, Optimal power-assistance system for a new pedelec model. Proc. IMechE, Part C: Journal of Mechanical
Engineering Science, DOI: 10.1177/0954406215604657, in press.
[20] X-CELL RT, http://www.thun.de/en/products/sensor-technology/.
[21] TMM sensor, http://www.idbike.com/tmm-powermanagement.htm.
[22] T. Tanaka, and T. Murakami, A Study on Straight-Line Tracking and Posture Control in Electric Bicycle, IEEE Trans. Industrial
Electronics, 56 (2009), pp. 159–168.
[23] C.L. Hwang, H.M. Wu and C.L. Shih, Fuzzy Sliding-Mode Underactuated Control for Autonomous Dynamic Balance of an Electrical
Bicycle, IEEE Trans. Control Systems Technology, 17 (2009), pp. 658–670.
[24] P.H. Chen, Intelligent application: ELEBIKE Fuzzy Control: Part 1, in Proceedings of the Seventh International Conference Machine
Learning and Cybernetics, Kunming, China, 2008, pp. 3581–3585.
[25] C.Y. Liang, W.H. Lin, and B. Chang, Applying Fuzzy Control to an Electric Bicycle, in Proceedings of First International Conference on
Innovative Computing, Information and Control, Beijing, China, 2006, pp.513-516.
[26] Q. Zheng and M. Tomizuka, A disturbance observer approach to detecting and rejecting narrow-band disturbances in hard disk drives, in
Proceedings of the AMC 2008, Trento, Italy,2008, pp. 254-259.
[27] M.T. White, M. Tomizuka, and C. Smith, Improved track following in magnetic disk drives using a disturbance observer, IEEE/ASME
Transactions on Mechatronics, 5 (2000), pp. 3 – 11.
[28] S.B. Chang, P.C. Chen, H.S. Chuang, C.C. Hsiao, Velocity control with disturbance observer for pedal-assisted electric bikes, Vehicle
System Dynamics, 50 (2012), pp. 1631 – 1651.
 International Journal of Innovative Technology and Exploring Engineering (IJITEE)
ISSN: 2278-3075, Volume-8, Issue- 6S4, April 2019

Design and Implementation of

Smart Electric Bike Eco-Friendly
Sunikshita Katoch, Rahul, Ranjit Kumar Bindal 

Abstract— In this paper, we are concerning about the growing

demand of energy all over the world, which motivate us to switch
over renewable resource of energy. There are many different
ways by which we can save energy in different sectors. Our main
focus is on automobile sector where we are converting old
petroleum bike to electric bike. In these electric bikes we use
electrical motor (BLDC motor) instead of combustion engine as
there is less pollution, low maintenance cost, reduces noise.
These bikes utilize chemical energy stored in the rechargeable
battery packs. This paper deals with the design and development
of electric bike which make use of electric energy as primary
source. There is a distribution for charging the battery emitting it
from the main system.
Index Terms: Hybrid electrical vehicle, BLDC motor, electric
bike, lithium-ion battery.

I. INTRODUCTION
The main reason to design the electric bike is to overcome Fig: 1 BLDC motor
the problem with the pollution and with the economy.
Battery
Future E bike is the best technical application as a solution
for the better world and upcoming generation. The E bike is A lithium ion battery are the rechargeable battery which is
a battery operated vehicle that is very economical with low used in many electric vehicles now a days in this battery the
maintenance cost and less pollution. E bikes are an attractive lithium ions moves from the negative electrode to the
alternative to both conventional bicycles and traditional positive electrode during discharge and back when charging.
automobiles, providing an environmentally friendly, fun, It is more efficient as it is less in weight, high speed, no
efficient and convenient way to travel. E-bikes are driven pollution, more reliable.
with the help of battery which is coupled with electric Frame
motor. It is the core structure on which other parts are assembled,
E- Bike is the plug-in electrical vehicles with two or three it supports the motor, provides base for the location of
wheels. The power on which this bike works is stored in a starring and supports the passenger or luggage, batteries also
rechargeable battery which drives the motor. Now a day’s attached to the frame.
these motorbikes are manufactured at a very large scale.
Typical parts used in E-bike are BLDC motor, battery,
controller, throttle, chain set. [1]

II. COMPONENTS OF E-BIKE

The e bike consists of following components
DC Motor
The motor is having rated power of 750w capacity with
max 500 rpm per min, their specifications are as follows:
Rated voltage – dc 48 v
Rated current -13.5 amp Fig: 2 frame of the e-bike
Protection level –IP33
Insulation class – f Breaking System
In this system, kinetic energy is converted into thermal
energy by friction. It is used in band brake system which
consists of spring loaded friction –shoe mechanism, which is
Revised Manuscript Received on April 12, 2019.
driven with the help of hand lever.
Sunikshita Katoch, Electrical Engineering Department, Chandigarh
University, Gharuan Mohali, India.
Rahul, Electrical Engineering Department, Chandigarh University,
Gharuan Mohali, India.
Ranjit Kumar Bindal, Electrical Engineering Department,
Chandigarh University, Gharuan Mohali, India.

Published By:
Retrieval Number: F11970486S419/19©BEIESP Blue Eyes Intelligence Engineering
DOI: 10.35940/ijitee.F1197.0486S419 965 & Sciences Publication
 Design And Implementation Of Smart Electric Bike Eco-Friendly

Chain set world in e-bicycle deals, trailed by the Netherlands and

It is the more common type of chain drive which is used Germany. E-bicycles can keep up speed with less exertion.
for transmission of mechanical power to long lasting E-bicycles are found to expand bike use. E-bicycles can
&better way of rotatory motion from one gear to another it possibly dislodge ordinary mechanized (inside burning)
is derived by a tooth head wheel called a sprocket it is modes, yet there are open inquiries regarding their job in
simple, reliable and efficient. uprooting customary bikes. E-bicycles have been appeared
to give medical advantages and a request of size less carbon
Sprocket dioxide than a vehicle venturing to every part of a similar
It is the wheel with teeth which holds the chain they are separation. Security issues have developed as an
used to transmit rotatory motion between to shafts. It is arrangement issue in a few locales and e-bicycle numbers
different from gear as it has many teeth & where as gear are currently moving toward levels in which satisfactory
consists of only one or two teeth. It is also different from wellbeing information can be gathered. Research on e-
pulley because the pulley works more smoothly [2]. bicycles is still in its earliest stages. As e-bicycle utilization
keeps on developing, so too will the requirement for further
research, so as to give the fundamental information to
illuminate strategy creators and industry.
C. SIMON WASHINGTON, NARELLE HAWORTH
(2014) [5] clarified that there are as of now in excess of 700
urban communities working bicycle share programs.
Indicated advantages of bicycle share incorporate adaptable
versatility, physical movement, emanations and fuel use.
Certain or express in the figuring of program benefits are
presumptions with respect to the methods of movement
Fig: 3 Sprocket of the bike supplanted by bicycle share ventures. An optional and
remarkable common and support. These two parts are then
III. LITRATURA REVIEW consolidated to gauge bicycle offer's general commitment to
A. JENNIFER DILL, GEOFFERY ROSE (2012) [3] changes in vehicle kilometers voyaged.
clarified that Electric bicycles are progressively regular in D. C.C.CHAN (2002) [6] talked about a reality where
China however are moderately uncommon in the United condition assurance and vitality preservation are developing
States. The meetings uncovered a few conceivable statistic concerns, the advancement of electric vehicles (EV) and half
markets for e-bicycles that could extend the bicycling and half electric vehicles (HEV) has taken on a quickened
populace: ladies, more established grown-ups, and individuals pace. The fantasy of having industrially practical EVs and
with physical impediments. Proprietors of e-bicycles noticed HEVs is turning into a reality. EVs and HEVs are step by
their capacity to travel longer separations and over slopes step accessible in the market. This paper will give a diagram
without any difficulty and to touch base at a goal, for of the present status of electric and half breed vehicles
example, work, less damp with sweat and less drained than a worldwide and their cutting edge, with accentuation on the
customary bike would permit. These highlights may beat a building logic and key innovations. The significance of the
portion of the regular hindrances to bicycling for all combination of innovations of vehicle, electric engine drive,
socioeconomics. The vast majority of the talked with e- hardware, vitality stockpiling, and controls and furthermore
bicycle proprietors utilized their e-bicycles to substitute for the significance of the reconciliation of society quality from
movement by either human-fueled bikes or customary engine government, industry, explore establishments, electric
vehicles. Thusly, the e-bicycle can address worries about power utilities, and transportation experts are tended to. The
medical issues identified with inertia, contamination, and test of EV commercialization is examined.
other open strategy issues to which private vehicles E. K.J.ASTROM, R.E.KLEIN (2005) [7] clarified that the
contribute. Further research is expected to decide if explicit elements of bikes is dissected from the point of view of
approaches are expected to expand reception of e-bicycles. control. Models of various multifaceted nature are exhibited,
The potential for strife between riders of e-bicycles and of beginning with straightforward ones and closure with
standard bicycles due to speed differentials is a worry. progressively practical models created from multimode
Regardless of whether speed differentials will represent a programming. Models that catch fundamental conduct, for
noteworthy issue will depend not just on the degree of example, self-adjustment just as models that show troubles
selection of e-bicycles however the qualities of the riders. with back wheel controlling are considered. Encounters
B. ELLIOTFISHMAN, CHRISTOPHER CHERRY (2016) utilizing bikes in charge instruction alongside
[4] talked about that E-bicycles speak to one of the quickest recommendations for the sake of entertainment and
developing sections of the vehicle showcase. More than 31 provocative examinations with demonstrated understudy
million e-bicycles were sold in 2012. Research has pursued fascination are introduced. At long last, bikes and clinical
this development and gives a combination of the most projects intended for kids with incapacities are depicted.
relevant subjects rising over the past on the expanding point
of e-bicycles. The center is transport as opposed to
recreational e-bicycle look into, just as the most basic
research holes requiring consideration. China drives the

Published By:
Blue Eyes Intelligence Engineering
Retrieval Number: F11970486S419/19©BEIESP 966 & Sciences Publication
DOI: 10.35940/ijitee.F1197.0486S419
 International Journal of Innovative Technology and Exploring Engineering (IJITEE)
ISSN: 2278-3075, Volume-8, Issue- 6S4, April 2019

IV. MATHEMATICAL CALCULATIONS & 3. J. Dill, G. Rose, “Electric bikes and Transportation policy-
RESULTS Insights from earlya dopters” SAGE Journals-Vol.2, Issue-1,
2012.
As the BLDC motor is being used with 750 watt power 4. E. Fishman, C. Cherry, “E-bikes in the mainstream-
and 2100 rpm with 48 volts and 13.5 amps. Motor can be Reviewing a Decade of research”- Vol 36 Issue 1, 2016.
reaching a peak current during starting equal to 15 amps[8- 5. S.Washington, N. Haworth, “Bike shares impact on car use-
9]. Evidence from the United states, Great Britain, and
P=2*3.14*N*T/60 Australia”- Vol. 31,PP. 13-20, 2014.
6. C. C. Chan “The state of the art of electric and hybrid
750=2*3.14*2100*T/60 vehicles “Proceeding to IEEE, Vol. 90 Issue 2, PP. 247-275,
T=3.41 N m=3412.19N-mm 2017.
Chain drive reduction=43/14=3.07:1 7. K.J. Astros, R.E. Klein, “Bicycle dynamics and control -
Wheel shaft torque =T*R chain=3412.1*3.07=10475.147 Adapted bicycles for education and research” IEEE control
N mm system magazine, Vol. 25 Issue 4, pp.26-47, 2017
Wheel shaft speed =2100/3.07=684 rpm 8. P. Zhang. , “Industrial Control Technology: A Handbook for
Engineers and Researchers”.
9. https://www.researchgate.net/publication/224299571_Electric
_bicycle_using_batteries_and_supercapacitors

AUTHORS PROFILE

Rahul Final year student B.tech of Electrical

Engineering From Chandigarh
University.(2015-2019)

Fig.4 Block diagram of the e bike

Sunikshita Katoch Final year student B.tech
V. CONCLUSIONS of Electrical Engineering From Chandigarh
University.(2015-2019).Diploma holder from
Electric vehicles are the future of our world with the Govt. Polytechnical. Collage, HAMIRPUR
increasing consumption of non renewable resources such as (HP).
petroleum, diesel which leads us to step our way towards the
renewable sources such as solar hydro electric power and Ranjit Kumar Bindal He received his
battery. There are alternative ways by which we can save B.Tech in Electrical Engineering from PTU,
energy. One of such way is electric bike; it is also the new Jalandhar, Punjab in 2002 and M.Tech in
way of transport which provides us easy way of transport to 2008 in Power Engineering from PTU,
provide of any age. It is cheap source of transport and Jalandhar.
affordable to anyone. The motor used in this bike has high
efficiency and the battery bank has less weight with high
speed. These bikes are environmental friendly, needs less
maintenance and can be also assembled to small component.

ACKNOWLEDGMENT
It Apart from the efforts of ours, the success of any
project depends largely on the encouragement and
guidelines of many others. We would like to show our
appreciation to the respected Dr. Inderpreet kaur (HOD EE)
for their support, we would also like to thank Dr. Ranjit
Kumar Bindal (Assistant professor) for giving us this
opportunity for making project based on next generation of
E- vehicles. Without their assistance and dedicated
involvement in every process, this project has never been
accomplished.
Last but not the least we appreciate and thank all those
people who contributed to this project directly or indirectly.

REFERENCES
1. S. Matey, A. Prabhu, “Design and Fabrication of Electric
Bike” International Journal of Mechanical Engineering and
Technology- Vol. 8 Issue 3- March 2017.
2. C.D. Ajudiya,M. M. trivedi, “Design and Development of E-
Bike –A Review” Iconic Research and Engineering journals-
Vol.1 Issue 5- Nov 2017 .

Published By:
Retrieval Number: F11970486S419/19©BEIESP Blue Eyes Intelligence Engineering
DOI: 10.35940/ijitee.F1197.0486S419 967 & Sciences Publication
 © April 2020 | IJIRT | Volume 6 Issue 11 | ISSN: 2349-6002

Electric bike with mileage enhancement

Mr. Ankit Giri1, Mr. Darshan Gaikar2, Mr. Shreyas Mhatre3, Mr. Aditya M4, Prof. Pranita Chavan5
1,2,3,4
UG Student, Department of Electrical Engineering, Pillai HOC college of Engineering and
Technology, Rasayani, Maharashtra, India
5
HOD, Department of Electrical Engineering, Pillai HOC college of Engineering and Technology

Abstract- This paper aims to convert the old bicycle into in the city, riding an electric bike is faster and
an eBike along with mileage enhacement capability. The cheaper than a car or public transport.
eBike has two motors, one as a motor and the other as a The basic design of the eBike and parts is similar to
generator. The main idea of this project is to produce
other bicycles and includes an additional electric
some voltage while working on the eBike slope and
motor It is powered by means of a rechargeable
stored energy generated in the battery, which will
further enhance the mileage of the eBike. A variety of
battery, which gives riders greater power and in the
batteries provide the power needed for motor operation. end a smoother, more cozy and much less biking
The project consists of two battery sets, the first battery experience. The Ebike is a better choice than
set is used to operate the motor and the second set is traditional bicycles and traditional vehicles because
fully discharged, which is automatically charged when they are environmentally friendly, require less
operating the slope eBike. maintenance, have less motor output noise and are
more efficient. The main goal of eBike is to
Index terms- Electric bike, mileage improvement, bldc
overcome and expand pollution. Instead of a
motor, battery, eBay controller, adxl345 sensor
traditional vehicle. These eBikes have a high
I.INTRODUCTION reputation abroad and the alternative to traditional
bicycles improves the user experience due to the
The idea of making electric motors has attracted eBike experience. An ebike combines the benefits of
cyclists since the late 1800s, and many American a regular bicycle with a motorbike. It is an ecological
inventors have attempted to combine electric motors means of transportation, with low maintenance cost,
with the usual mechanics of bicycles. Until the allowing you to move freely and easily without
technological breakthroughs of the 20th and 21st sweat. It is almost as environmental as bicycle
centuries, this idea was finally coming true. Light because it requires very little power to propagate
motors, batteries, rechargeable, high capacity, drive electric motors efficiently; Batteries need to be
trains smoothly, bicycle parts provide a way to enjoy disposed of properly. This is cheaper because the cost
the benefits of today's electric bikes and fitness and of recharging the battery from your main socket is
physical needs. A bicycle, electric bicycle, also very low. However you should buy a new battery
known as exercise or part of a daily e-bike trip, has every 2 or 3 years. Being able to combine electrical
an electric motor that can be used for propulsion. support with your own pedaling effort, increasing
There are many types of e-bikes available on the electric motor support when you are tired or on the
market, from small motor-only e-bikes to more road can help you regain your right size. There are
powerful e-bikes that are closer to the modern cycle also cases where Ebike is safer than a regular bicycle.
of operation than the human effort of pedal power.
Electric bikes are a new and better alternative to II. COMPONENTS OF EBIKE
urban transportation. These provide all the benefits of
a regular cycle: fun exercise, zero elimination, lack of The ebike has the following components
strength to eliminate the most serious shortcomings BLDC motor:-
of cycling. When you go downhill, consider hiking The bldc motor used the rating of 250w.This motor
up the hill and it's an e-bike experience. In most cases has been fitted into tyre of ebike.

IJIRT 149150 INTERNATIONAL JOURNAL OF INNOVATIVE RESEARCH IN TECHNOLOGY 91

 © April 2020 | IJIRT | Volume 6 Issue 11 | ISSN: 2349-6002

The specification of motor are as follow

The frame is core structure on which the components
as well as batteries are placed, it also provides the
base for the passengers

Arduino uno;-

FIGURE 1
Voltage: - 48v, 250w
Rpm:- 350rpm
Wheel size:16 inch
Design:- brushless FIGURE 4
Arduino Uno is an open source microcontroller board
Lead acid battery :- based on the Microchip ATmega328P
Microcontroller and developed by Arduino.cc. The
board consists of a set of digital and analog input /
output (I / O) pins that can interfere with various
expansion boards (shields) and other circuits. The
board can be programmed via a Type B USB cable
with 14 digital pins, 6 analog pins and the Arduino
IDE (Integrated Development Environment). It can
be powered by a USB cable or by an external 9-volt
battery, although it accepts voltages between 7 and
20 volts.
FIGURE 2
The batteries used are lead acid battery. the energy
ADXL345:-
stored in this batteries is used to run the bldc motor.
the specification of battery are as follow
Battery type:- lead acid
Rating:- 12v, 7amp hr

Frame of ebike :-

FIGURE 5
The ADXL345 is a small, thin, low power, 3-axis
accelerometer with high resolution (13-bit)
measurement at up to ±16g.The ADXL345 is used to
detect the slope when the ebike is in operation. when
FIGURE 3
the slope is detected the generation is started

IJIRT 149150 INTERNATIONAL JOURNAL OF INNOVATIVE RESEARCH IN TECHNOLOGY 92

 © April 2020 | IJIRT | Volume 6 Issue 11 | ISSN: 2349-6002

Voltage and current sensor module:- When the Ebike is being operated on the slope the
It is used to detect the voltage and current that is motor on the rear wheel is turned off this operation is
generated and it is displayed on lcd screen. performed by using the single channel relay and the
same instant the generator on the front tyre starts
III. WORKING OF THE SYSTEM OF EBIKE generating the energy, the energy being generated is
stored in the battery. The slope is being detected by
using the accelerometer sensor i.e ADXL345.When
the ebike comes at the end of slope it automatically
turns on the rear tyre and switches off the generator
on the front tyre

V. MATHEMATICAL CALCULATION &

RESULT
The Ebike works as normal electric bike on straight
road i.e motor operates through battery and bike is As the BLDC Motor being used is of rating 250w
running in normal condition. But when inclined or and 350 rpm with voltage and current rating of 48v
slope is detected the accelerometer sense the and 6 amperes.
downhill movement and the relay turns off the motor P=2*3.14*N*T/60
and the same time the generator which is present is 350=2*3.14*350*T/60
turned on due to which electricity generation is T=6.82Nm OF TORQUE IS GENERATED
started and automatically when the downhill road
comes to an end the bike automatically turns on the RESULTS OF VOLTAGE GENERATED ON
motor and the generator is turned off. The bike SLOPE
consist of two battery one which is fully charged and SLOPE IN METERS GENERATED VOLTAGE
is used by motor and second battery is completely 35M 13V
43M 16V
discharged which gets automatically charged during
50M 20V
the downhill movement by the generator. The
automatic switching process is done with use of
VI. FLOWCHART OF SLOPE DETECTION
Arduino, relay module and the accelerometer sensor
MECHANISM
i.e ADXL345 The Arduino is connected to
accelerometer and relay module as soon as
accelerometer sensor sense the change in road
condition it automatically operates the relay module
through Arduino so switching of motor and generator
is done.
IV. MODES OF OPERATIONS

A) Normal mode
When the Ebike is operated on the normal surface
road the Ebike operates as normal Ebike in which the
Ebike runs through the bldc motor and the speed of
bldc motor can be controlled using the throttle which
is connected to the controller of Ebike. This
controller is used to control the operation of ignition,
speed, braking, headlight

B) Regenerative mode

IJIRT 149150 INTERNATIONAL JOURNAL OF INNOVATIVE RESEARCH IN TECHNOLOGY 93

 © April 2020 | IJIRT | Volume 6 Issue 11 | ISSN: 2349-6002

[2] R.D. Belekar, Shweta Subramanian1, Pratik

Vinay Panvalkar2, Medha Desai3 „Alternator
Charging System for Electric Motorcycles‟
Volume: 04 Issue: 04 | Apr -2017
[3] Florin Dumitrache, Marius Catalin Carp, and
Gheorghe Pana Electronics and Computers
Department 2016 IEEE 22nd International
Symposium for Design and Technology in
Electronic.978-1-5090-4445-0/16/31.00 2016
IEEE 248 20-23 Oct 2016, Oradea, Romania, E-
bike electronic control unit.
[4] Abatan O. A, Adewale A.O, Alabi A.A,
Constant Electricity Generation From Self-
Charging Inverter, IJETAE ISSN 2250-2459,
ISO 9001:2008 Certified Journal, Volume 3,
Issue 12, December 2013.
[5] US 20010008191A1 Patent Application
Publication Pub. No: US 2001/0008191 A1
Smith et al. (43) Pub. Date: Jul.19, 2001
“ELECTRIC POWER GENERATIONSYSTEM
For ELECTRIC VEHICLES” Vincent A. Smith,
Vallejo, CA (US), Larry J. Kinchen, Vallejo, CA
(US), Susan W. Smith, Vallejo, CA (US)
VII. CONCLUSION [6] Faisal H. Khan, Leon M. Tolbert, “5 kW
Multilevel DC-DC Converter for Hybrid Electric
It reduces the pollution from cities. The city has been and Fuel Cell Automotive Applications”.
defined to develop a battery power vehicle for daily [7] Sreevalsan S Menon, Sooraj M S, Sanjay
use and to change ecofriendly methods for daily short Mohan, Rino Disney, Suneeth Sukumaran,
term basis within cities. In conventional work, the Design and analysis of kinetic energy recovery
concept of self-charging has been found in system in bicycles, IJIRSET,ISSN:2319- 8753,
conventional battery power vehicles, which was the Vol.2, Issue 8, August 2013.
goal of changing the concept of plug in charging.
This work started with study of technology, the use
of component and future of battery power electric
car. On the basis of application used in accelerometer
sensor ADXL345, relay, motor, motor controller as
were dynamo were selected. From the result we can
conclude that vehichle travels for long time.
Alternator produces 12 to 14v, which uses rotational
from wheel under fast motion. The result of fast
motion is the charging of batteries.

REFERENCES

[1] „DESIGN AND FABRICATION OF SELF

CHARGING ELECTRIC VEHICLE‟ M. Sathya
Prakash Department of Thermal Engineering,
Pannai College of Engineering &Technology
Sivagangai, India Vol 8. No.1 – Jan-March 2016

IJIRT 149150 INTERNATIONAL JOURNAL OF INNOVATIVE RESEARCH IN TECHNOLOGY 94

ISSN: 2455-2631 © April 2017 IJSDR | Volume 2, Issue 4

Electric Bicycle
1
Sanjeeb Kumar Pattanayak, 2Milan Tirkey, 3Pramod Lakra, 4Vivek Ranjan,
5
Soumya Ranjan Panda, 6Manas Ranjan Panda
1,2,3,4,5
B.Tech Mechanical Final year student, 6Assistant Professor
Gandhi Institute of Engineering and Technology
Gunupur

ABSTRACT:-Increasing demand for non-polluting mechanized transportation has revived the interest in the use of
electric power for personal transportation and also reduced reliance on automobiles. A low cost alternative to an
automobile is a bicycle. However, the use of bicycles has been limited to very short trips or as a recreational activity.This
report describes the design of an electric assisted bicycle that will extend the range of a typical rider. The rate of
improvements in technologies is at an exponential level despite that the electric bicycle is a concept that has been very
feasible for years but has not been fully explored. The human electric bicycle is designed to provide electromagnetic
propulsions to a bicycle therefore relieving the user of having to produce the energy required to run the bicycle.The
system design is based on mechanically coupling a dc motor as the primary power source to drive the bicycle and
electrically wiring the motor together with a dc rechargeable battery and efficient transmission from the source to the
motor.
Key words:-Pedelec, traction battery, electric vehicle (EV), hybrid electric vehicle (HEV), Powertrain

I. Introduction
An electric bicycle, also known as an e-bike or booster bike, is a bicycle with an integrated electric motor which can be used for
propulsion. There are a great variety of e-bikes available worldwide, from e-bikes that only have a small motor to assist the rider's
pedal-power (i.e., pedelecs) to somewhat more powerful e-bikes which tend closer to moped-style functionality: all, however,
retain the ability to be pedalled by the rider and are therefore not electric motorcycles. E-bikes use rechargeable batteries and the
lighter varieties can travel up to 25 to 32 km/h (16 to 20 mph), depending on the laws of the country in which they are sold, while
the more high-powered varieties can often do in excess of 45 km/h (28 mph). In some markets, such as Germany, they are gaining
in popularity and taking some market share away from conventional bicycles, while in others, such as China, they are
replacing fossil fuel-powered mopeds and small motorcycles. A key advantage of hybrid or plug-in electric vehicles
is regenerative braking due to their capability to recover energy normally lost during braking as electricity is stored in the on-
board battery

(Fig.1-A mountain bike styled e-bicycle)

II. LITERATURE REVIEW

In 1999,AVL Company proposed a hybrid system that used a 50 cccarburetted lean-burn two-stroke engine with a 0.75 kW
electric motor mounted on the engine crankshaft mainly to provide increased torque during acceleration.[1]

Su-Hau et al (2004) focused on the highly efficient energy usage of the battery energy and proposed an integrated management
system for electric motor.[2]

David and Sheng-Chung (2004) proposed new parallel-type hybrid-electric-power system comprises an engine’s energy
distribution and a torque-integrated mechanism (specifically including an engine, a motor/alternator, a CVT device, and PCM as
well as a 3-helical gear set).[3]

Wenguang et al (2005) presented an approach to control powertrain of series hybrid electric vehicles. A formulation of the
system equations and controller design procedure were proposed by them. They also proposed a new switching algorithm for the
power converter for motor torque and motor flux control.[4]

IJSDR1704079 International Journal of Scientific Development and Research (IJSDR) www.ijsdr.org 409
ISSN: 2455-2631 © April 2017 IJSDR | Volume 2, Issue 4

Daniel (2007) designed, developed and implemented a series hybrid electric vehicle. Though he proposed the architecture as
hybrid electric vehicle architecture, he showed that the vehicle runs well in the electric mode and left the hybrid conversion as
future expansion.[5]

Lukic et al (2007) tried to develop a driving cycle of the auto rickshaw in a typical large Indian city, in their case, Delhi. First,
they considered the existing driving cycles used in India are considered as candidates. Since these data were not applicable, GPS
data collected at various times of the day were applied to the analysis. They derived the new driving cycle from the gathered
information via GPS data as well as surveys of auto rickshaw drivers in India, which helped to get the entire picture for the
driving cycle.[6]

‘Investigation of an electric assisted bicycle and determination of performance characteristics’’ paper examines and realizes
an alternative design for a front wheel hub direct drive, which utilizes a three-phase brushless PM motor.[7]

III. What is an electric vehicle?

An electric vehicle (EV), also referred to as an electric drive vehicle, uses one or more electric motors or traction motors for
propulsion. An electric vehicle may be powered through a collector system by electricity from off-vehicle sources, or may be self-
contained with a battery or generator to convert fuel to electricity. EVs include road and rail vehicles, surface and underwater
vessels, electric aircraft and electric spacecraft.Electric vehicles are different from fossil fuel-powered vehicles in that they can
receive their power from a wide range of sources, including fossil fuels, nuclear power, and renewable sources such as tidal
power, solar power, and wind power or any combination of those. However it is generated, this energy is then transmitted to the
vehicle through use of overhead lines, wireless energy transfer such as inductive charging, or a direct connection through an
electrical cable.In an electric vehicle (EV), a battery or other energy storage device is used to store the electricity that powers the
motor. EV batteries must be replenished by plugging in the vehicle to a power source. Some electric vehicles have onboard
chargers; others plug into a charger located outside the vehicle. Both types, however, use electricity that comes from the power
grid. Although electricity production may contribute to air pollution, EVs are considered zero-emission vehicles because their
motors produce no exhaust or emissions.In a parallel hybrid motorized bicycle, such as the aforementioned 1897 invention by
Hosea W. Libbey, human and motor inputs are mechanically coupled either in the bottom bracket, the rear or the front wheel,
whereas in a (mechanical) series hybrid cycle, the human and motor inputs are coupled throughdifferential gearing. In an
(electronic) series hybrid cycle, human power is converted into electricity and is fed directly into the motor and mostly additional
electricity is supplied from a battery.The electricity may then be stored on board the vehicle using a battery, flywheel,
orsupercapacitors. Vehicles making use of engines working on the principle of combustion can usually only derive their energy
from a single or a few sources, usually non-renewable fossil fuels. A key advantage of hybrid or plug-in electric vehicles
is regenerative braking due to their capability to recover energy normally lost during braking as electricity is stored in the on-
board battery.

IV.Electricity sources
4.1.Connection to generator plants
 Direct connection to generation plants as is common among electric trains, trolley buses, and trolley trucks.
 Online Electric Vehicle collects power from electric power strips buried under the road surface through electromagnetic
induction
4.2. Onboard generators and hybrid EVs
 generated on-board using a diesel engine: diesel-electric locomotive
 generated on-board using a fuel cell: fuel cell vehicle
 generated on-board using nuclear energy: nuclear submarines and aircraft carriers
 renewable sources such as solar power: solar vehicle
It is also possible to have hybrid EVs that derive electricity from multiple sources. Such as:
 on-board Rechargeable Electricity Storage System (RESS) and a direct continuous connection to land-based generation
plants for purposes of on-highway recharging with unrestricted highway range
 on-board rechargeable electricity storage system and a fueled propulsion power source (internal combustion engine): plug-in
hybrid

V. Lithium-ion battery
Most electric vehicles use lithium ion batteries. Lithium ion batteries have higher energy density, longer life span and
higher power densitythan most other practical batteries. Complicating factors include safety, durability, thermal breakdown
and cost. Li-ion batteries should be used within safe temperature and voltage ranges in order to operate safely and

IJSDR1704079 International Journal of Scientific Development and Research (IJSDR) www.ijsdr.org 410
ISSN: 2455-2631 © April 2017 IJSDR | Volume 2, Issue 4

efficiently.Increasing the battery's lifespan decreases effective costs. One technique is to operate a subset of the battery cells at a
time and switching these subsets.
V.I. Electric motor
The power of a vehicle electric motor, as in other vehicles, is measured in kilowatts (kW). 100 kW is roughly equivalent to
134 horsepower, although electric motors can deliver their full torque over a wide RPM range, so the performance is not
equivalent, and far exceeds a 134 horsepower (100 kW) fuel-powered motor, which has a limited torque curve.Usually, direct
current (DC) electricity is fed into a DC/AC inverter where it is converted to alternating current (AC) electricity and this AC
electricity is connected to a 3-phase AC motor.
V.II. Hybrid EVs
A hybrid electric vehicle combines a conventional (usually fossil fuel-powered) powertrain with some form of electric propulsion.
As of April 2016, over 11 million hybrid electric vehicles have been sold worldwide since their inception in 1997. Japan is the
market leader with more than 5 million hybrids sold, followed by the United States with cumulative sales of over 4 million units
since 1999, and Europe with about 1.5 million hybrids delivered since 2000. Japan has the world's highest hybrid market
penetration. By 2013 the hybrid market share accounted for more than 30% of new standard passenger car sold, and about 20%
new passenger vehicle sales including kei cars.
V.III. On- and off-road EVs
EVs are on the road in many functions, including electric cars, electric trolleybuses, electric buses, battery electric buses, electric
trucks,electric bicycles, electric motorcycles and scooters, neighborhood electric vehicles, golf carts, milk floats,
and forklifts. Off-road vehiclesinclude electrified all-terrain vehicles and tractors.

V.IV. Energy and motors

In the systems above motion is provided by a rotary electric motor. However, it is possible to "unroll" the motor to drive directly
against a special matched track. In addition to the high-performance control systems needed, switching and curving of the tracks
becomes difficult with linear motors, which to date has restricted their operations to high-speed point to point services.
VI.Components
The type of battery, the type of traction motor and the motor controller design vary according to the size, power and proposed
application, which can be as small as a motorized shopping cart or wheelchair, through pedelecs, electric motorcycles and
scooters, neighborhood electric vehicles, industrial fork-lift trucks and including many hybrid vehicles.
VII.Energy sources
Although EVs have few direct emissions, all rely on energy created through electricity generation, and will usually emit pollution
and generate waste, unless it is generated by renewable source power plants. Since EVs use whatever electricity is delivered by
their electrical utility/grid operator, EVs can be made more or less efficient, polluting and expensive to run, by modifying the
electrical generating stations. This would be done by an electrical utility under a government energy policy, in a timescale
negotiated between utilities and government.
VIII. Issues with batteries
An electric vehicle battery (EVB) or traction battery is a battery used to power the propulsion of battery electric
vehicles (BEVs). Vehicle batteries are usually a secondary (rechargeable) battery. Traction batteries are used in forklifts,
electric Golf carts, riding floor scrubbers, electric motorcycles, full-size electric cars, trucks, and vans, and other electric vehicles.
IX. Efficiency
Because of the different methods of charging possible, the emissions produced have been quantified in different ways. Plug-in all-
electric and hybrid vehicles also have different consumption characteristics.
X.Charging
If a large proportion of private vehicles were to convert to grid electricity it would increase the demand for generation and
transmission, and consequent emissions. However, overall energy consumption and emissions would diminish because of the
higher efficiency of EVs over the entire cycle. A recharging system that avoids the need for a cable is Curb Connect, patented in
2012 by Dr. Gordon Dower. In this system, electrical contacts are fitted into curbs, such as angle parking spaces on city streets.
When a suitably authorized vehicle is parked so that its front end overhangs the curb, the curb contacts become energized and
charging occurs.
XI. Battery swapping
Instead of recharging EVs from electric socket, batteries could be mechanically replaced on special stations in a couple of minutes
(battery swapping).Batteries with greatest energy density such as metal-air fuel cells usually cannot be recharged in purely
electric way. Instead, some kind of metallurgical process is needed, such as aluminum smelting and similar.Silicon-air,

IJSDR1704079 International Journal of Scientific Development and Research (IJSDR) www.ijsdr.org 411
ISSN: 2455-2631 © April 2017 IJSDR | Volume 2, Issue 4

aluminum-air and other metal-air fuel cells look promising candidates for swap batteries. Any source of energy, renewable or
non-renewable, could be used to remake used metal-air fuel cells with relatively high efficiency.
XII.The Advantages of Electric vehicles
XIII.I Save Your Money On Gas
Driving a car that runs solely on gasoline can become an extremely expensive thing. Gas prices are unpredictable and have
reached some very high peaks in the past decade. Electricity is much cheaper than gas in the long run so driving an electric car
can same you a countless amount of money.
XIII.II. You’re Saving TheWorld
Gas powered vehicles emit a large amount of carbon emissions into the air that cause damage to the atmosphere and contribute to
environmental problems such as global warming. Electric vehicles have absolutely no emissions from them, which is a wonderful
thing for the environment because no further damage is being done.
XIII.III. Less Maintenance Is Required
Electric cars have motors that run solely on electricity, which makes them much different than a gasoline powered car. Gas cars
have engines and many intricate parts that have to remain sufficiently lubricated in order to run properly, and this work has to be
done pretty often. With electric cars, while some maintenance is needed of course, it is much cheaper and much less frequent than
it is with gas powered vehicles.
XIII.IV. Energy efficiency
Energy is not consumed while the vehicle is stationary, unlike internal combustion engines which consume fuel while idling.
However, looking at the well-to-wheel efficiency of EVs, their total emissions, while still lower, are closer to an efficient gasoline
or diesel in most countries where electricity generation relies on fossil fuels.
XIII.V. Future work of electric bicycle
The following materials will be needed for designing of an electric bicycle.
 a bicycle
 a cordless drill (any voltage will work but the bigger the better)
 lots of zip-ties (how many you will use depends on your type of bike and/or drill)
 a piece of small wood ( this will support the drills' battery)
 a bicycle hand brake ( this is for the throttle system)
 bicycle cable ( this also for the throttle system and goes into the hand brake)
Step 1: Adding the drill
First of all we have to take a drill and make such an arrangement that will help to run the bicycle. Here are some of the pictures
that will help us how to make the arrangement.

(Fig.2-A cycle and drill setup)

Step 2: Zip-tieing the drill down

Now begin to zip-tie the drill onto the bicycle frame. Make sure that the drills' chuck always stays on the side of the tire. Try to
keep the drill in line with the tire. Get the drill as snug AS POSSIBLE against the side of the tire. The more friction you have the
better the drill will grip the tire.

(Fig.3- drill tied to cycle)

Step 3: Adding the throttle system
Combine the hand brake and the cable and the slide the hand brake on to the handlebars of the bicycle. This will probably require
removing the handlebar grips. Finally, tighten down the hand brake.

IJSDR1704079 International Journal of Scientific Development and Research (IJSDR) www.ijsdr.org 412
ISSN: 2455-2631 © April 2017 IJSDR | Volume 2, Issue 4

(Fig.4 - Different views of throttle cable arrangement)

Step 4: Attaching the throttle cable

Beware this is the hardest part of the instructable. Take the bicycle cable and lead it down to the side of the drill where the trigger
is. Zip-tie the end of the black part of the cable to the SIDE of the drill. Wrap the metal part of the cable around the trigger and
back to the side of the drill. Zip-tie the black part of the cable and the end of the metal section of the cable together. Make sure
that the cable does not pull on the trigger. It should only pull on it when you pull on the brake handle. You can skip this step and
just pull the drills trigger with your hand. However you have to drive with one hand.
Step 5: You are done!!!!!
You are done!!!!!!! However this bicycle doesn't have a whole lot of economical value it only goes about as fast as a rather
drained barbie jeep. It is more just for fun. Or it can be used for riding around the block and laughing maniacally and scaring little
children all at the same time!! Or looking cool. Just remember to use a drill that you don't use a lot.
XIV. Conclusions
China has experienced an explosive growth of sales of non-assisted e-bikes including scooter type, with annual sales jumping
from 56,000 units in 1998 to over 21 million in 2008, and reaching an estimated fleet of 120 million e-bikes as of early 2010. This
boom was triggered by Chinese local governments' efforts to restrict motorcycles in city centers to avoid traffic disruption and
accidents.Users began replacing traditional bicycles and motorcycles and, in e-bike became an alternative to commuting by
car.Nevertheless, road safety concerns continue as around 2,500 e-bike related deaths were registered in 2007. China is the
world's leading manufacturer of e-bikes, with 22.2 million units produced in 2009. Production is concentrated in five regions,
Tianjin, Zhejiang, Jiangsu, Shandong, and Shanghai. China exported 370,000 e-bikes in 2009. The issues associated with electric
bicycles may be addressed by custom-designed drives that are most efficient over a given operating cycle. These include city
bicycles, hill bicycles, distance bicycles, and speedy bicycles.

REFERENCES

[1]. Sheu, K. B., Hsu, T. H., & Hsu, Y. Y.A novel parallel hybrid motorcycle transmission.In Materials science forum.Trans Tech
Publications.

[2]. Hsu, SuHau., Hsu, D. W., Fu, L. C., & Hsu, Y. P. (2004, June). Novel integrated management system design of electric
motorcycles. In American Control Conference, 2004.Proceedings of the 2004.IEEE.

[3].Huang, K. David.,&Tzeng, Sheng Chung. (2004). A new parallel-type hybrid electric-vehicle. Applied Energy, 79(1), 51-64.

[4].Yan, Wenguang.,Utkin, V., &Rizzoni, G. (2005, June). Power flow control for a series hybrid electric vehicle.In Proceedings
of the IEEE International Symposium on Industrial Electronics, 2005.ISIE 2005.IEEE.

[5].Quinn, C., Daniel Zimmerle, D., & Bradley, T. H. (2012). An evaluation of state-of-charge limitations and actuation signal
energy content on plug-in hybrid electric vehicle, vehicle-to-grid reliability, and economics. IEEE Transactions on Smart
Grid, 3(1), 483-491.

[6].Lukic, S., Mulhall, P., &Emadi, A. (2008). Energy autonomous solar/battery auto rickshaw. Journal of Asian Electric
Vehicles, 6(2), 1135-1143.

[7].S. I. Brand, N. Ertugrul, W. L. Soong, “Investigation of an Electric Assisted Bicycle and Determination of Performance.

IJSDR1704079 International Journal of Scientific Development and Research (IJSDR) www.ijsdr.org 413
Wireless Sensor Network based Management System
for Electric Bicycle-Sharing

M. Prist∗ , A. Freddi† , S. Longhi∗ , A. Monteriù∗ and P. Antonini‡

∗
Dipartimento di Ingegneria dell’Informazione, Università Politecnica delle Marche
Via Brecce Bianche, 60131 Ancona, Italy
email: {m.prist, s.longhi, a.monteriu}@univpm.it
†
SMARTEST Research Centre, Università degli Studi eCampus
Via Isimbardi 10, 22060 Novedrate (CO), Italy - email: [email protected]
‡
Picchio S.p.A.
Zona industriale Tronto, 64010 Ancarano (TE) - email: [email protected]

Abstract—Bicycle-sharing is an exciting new model of public- actually in use in several cities of Denmark, Norway and the
private transport provision that has quickly emerged in the past Netherlands, the BiciMAD system in use in Madrid [8], the
five years, and represents a possible way for enhancing sustain- Bewegen Technologies system in use in Birmingham Alabama
able mobility in smart cities of the future. This paper proposes [9], the e-bike Electrified S, the first intelligent and internet-
a wireless sensor network based management system for electric connected commuter bike, developed by VanMoof [10], just
bicycle-sharing of new design, where each bicycle and docking
to cite the most known.
station is equipped with a measurement and communication
device (i.e., a network mote), and is both sensor and actor of A bicycle-sharing system is generally based on a reduced
the wireless network. Each bicycle can keep track of its position, number of key elements, such as a management software,
battery level, and communicate with the docking station, which
a set of docking stations, a certain numbers of bicycles
manages the recharge cycle, and locks or releases the bicycle to
the user. Since all the information is shared through the network, equipped with a measurement and communication system, and
the management system can provide high level software services a hardware and firmware layer for the low-level management
as well (e.g., the booking service). This paper describes in detail [11]. The main drawback of the existing systems can be found
the design and the implementation of the proposed wireless sensor by analysing these key elements. For instance, the booking
network based management system for electric bicycle-sharing. service often requires the users to physically visit an office in
Keywords—Wireless Sensor Networks, WSN Simulators, Elec- order to complete the first authentication process and obtain
tric Bicycle-Sharing. their personal reservation card. The communication between
bicycle and docking station may happen continuously via
I. I NTRODUCTION GPRS/UMTS modules even when the bicycle is deposited,
thus reducing the battery life [12]. Other times, the com-
Sustainable mobility is an important and effective way for munication is based on consumer devices (such as tablets
contributing to the reduction, in the future, of both environ- or smartphones) which increase the cost of the system and
mental pollution and wasting of energy resources [1]. In this the chance of vandalism. Hence, the need to propose an
regard, the use of emission neutral vehicles, like bicycles, is innovative solution by introducing a simplified design based
very attractive to achieve sustainable mobility in the smart on a Wireless Sensor Network (WSN). The proposed WSN
city of the future, where entire zones should be accessible to is based on Argosd I motes [13], a custom version of the
public transport and pedestrians only [2], [3]. Shared electric well known and spread Sky mote, and it is designed by using
bicycles (often called e-bicycles or e-bikes) could provide an a state-of-the-art cross simulator, namely Cooja [14], [15],
even higher level of service compared to regular bicycle- [16], [17]. In detail, in the proposed architecture, each bicycle
sharing systems: e-bicycles reduce the effort required by the and docking station is equipped with a measurement and
rider, promoting greater travel distances and easier use over communication device (i.e., a wireless sensor network mote),
hilly terrain [4]. In order to increase the number of citizens which is, simultaneously, sensor and actor of the network. Each
which use bicycles as the principal means of transport, it is bicycle can keep track of its position, battery level and can
necessary to create bicycle-sharing systems which are rapid communicate these data to the docking station together with
to use, simple and relatively safe. Many of such systems have its IDentifier. The docking station knows, at any time instant,
been realized in several cities worldwide, sometimes they have the state of the bicycle (i.e., battery level and activities), it is
been extremely successful, while other times they experienced able to manage the recharge cycle, and locks or releases the
many difficulties and were only useful to collect the users bicycle to the user. The booking service can finally access the
opinion for a future realization. It is, however, clear that WSN information, acting at a higher level.
bicycle-sharing systems will be one key point for sustainable
mobility in smart cities in the near future [5], [6]. This paper describes the architecture of the WSN on which
the proposed management system is based. In detail, the
Currently, several bicycle-sharing stations are available as architecture of the electric bicycle-sharing system is described
commercial products, such as the GoBike system [7] which is in Section II. Section III details the hardware of the proposed
WSN. In Section IV, it is shown how the WSN has been tested

978-1-5090-2320-2/16/$31.00 2016
c IEEE
making use of a co-simulator. Simulation results are presented server: this can be done by using the App installed on the user’s
in Section V. Some remarks conclude the paper. smartphone, tablet, or simply by accessing the website. In case
of a successful transaction, the App provides a unique QR-
II. T HE P ROPOSED S YSTEM A RCHITECTURE code identifier. Once the user has reached the docking station
indicated during the reservation process, it is sufficient to click
The development of an electric bicycle-sharing system is on the “start” button of the tablet integrated within the docking
not trivial, since it is composed by different parts which must station, and show the previously provided QR-code in front of
be closely integrated in order to guarantee all the desired the tablet camera for identification. Upon image recognition,
working functionalities (see Fig. 1). Moreover, it is crucial the ID is extracted from the server and the system unlocks
that this complexity does not hinder the system usability: the the most charged available bicycle. Fig. 2 shows this two step
user should always be able to use the system in a simple and recognition process.
safe manner.

Fig. 1. The proposed system architecture.

The proposed sharing system is mainly divided into a Fig. 2. Use case.
software part and a hardware, part which are described in the
following subsections.
B. Hardware
A. Software From the hardware point of view, the system can be further
divided into three sections, namely the pickup station, the
The software of the sharing system is composed by the
electronics for the bicycle operation, and one section for data
following modules:
exchange among the different actors. In detail, the hardware
• a server for data storage, statistics and hosting; is composed by
• a booking web-based service for the management of • an assisted-cycling system (composed by the power
bicycles and docking stations/columns; system, a microcontroller with GPS and GSM system
for data acquisition and interfacing with the global
• an App for web service functionalities and for dis- server);
playing the bicycles position;
• a docking and recharging station (made of one or more
• an App for the management of battery charging dock-points, i.e. the stands on which the bicycles are
(within the docking station/column). physically placed, including the lock system and the
When the system is installed and runs for the first time, slave microcontroller, and a column which allocates
the docking station sends its ID to the server, together with the tablet for interfacing with the user);
the IDs of all the bicycles associated to it. Note that, in the • dedicated hardware and firmware components for in-
initial state, all bicycles are assumed not fully charged. After terfacing I/O systems among electric bicycles, the
this, the docking station periodically sends to the server data docking station (for recharging and releasing the bicy-
on its state, the charge status, the position of each associated cles), and the communication devices that implement
bicycle, etc. Afterwards, the docking station remains in stand- the WSN.
by until at least one bicycle becomes fully charged: in this
case, the docking state changes from “stand-by” to “active”, In this paper, particular attention is paid to the third section,
and the docking station waits for possible usage requests. namely the WSN interfacing between the station and the
bicycle. In detail, this work is devoted to explain how the
From the user-side, once the user is registered to the WSN can simplify and solve different problems that could
service, he/she can send a bicycle reservation request to the arise when a complex architecture, like that of the e-bicycle
sharing, is implemented in real scenarios.
 The communication between the dock-point and the electric
bicycle is wirelessly realized through a WSN. In this scenario,
a mote is associated with each bicycle and another mote is
associated with each dock-point of the docking station. Thus
for every bicycle-dock-point pair a direct connection exists (see
Fig. 3).
Although the developed architecture realizes a point to
point connection between each e-bike and the docking station,
the usage of a WSN allows to eliminate in the future all the
internal cable connections for data exchange (bus 485) which
currently exist.
Messages sent from the dock-point are in broadcast, since
it does not know in advance the ID of the bicycle which will
request the station. On the other hand, the packets sent by the
bicycle node are in unicast in order to reduce the probability of
a wrong pairing. The handshake between the bicycle and the
dock-point node is performed in the following three phases.
1) The first phase occurs when the bicycle is physically Fig. 3. The bicyle and the dock-point.
attached to the dock-point and the presence sensor
is activated. In this context, the state machine within
the bicycle changes from “NOT-CONNECTED” to
“CONNECTED” and the bicycle node enters the sensor network can thus not be separated by the rest of the
listening mode. The node associated with the dock- system.
point is activated at the moment in which the presence
sensor is activated. It then ask the master control The key actor of the proposed WSN is the mote Argosd I
unit to start the handshake protocol. Upon receiving [13]. Argos I is a Sky Mote based node, hardware designed
the consent, it sends a broadcast welcome message by UC Berkeley [18] and produced by Crossbow [19]. Our
to the bicycles (this mode is necessary because, as Argosd I has the same Sky microcontroller and radio chip,
mentioned previously, the address of the bicycle node but with a different architecture, and it was developed for
approaching the station is not known a priori). reducing the amount of components on board, the node size,
2) When the bicycle receives the welcome message, it and the power consumption, to increase the flexibility, and to
extracts the network address of the dock-point and reduce the current loss due to leakage that happens happens
sends an ACK message, waiting for the LOCK and when the chip is pulling electrical current, even when powered
STATUS command. down. The core element of Argosd I is a Texas InstrumentsTM
3) After the dock-point has received the ACK message MSP430 MCU [20] which has been widely used in wireless
from the bicycle, the control unit is informed that the sensors networks [21]. The main advantages of the MCU are
connection has been successfully established. It sub- the extremely low power during periods of Sleep Mode and the
sequently sends the LOCK command with the relative massive use in the WSN nodes. The MSP430 micro controller
STATUS request to the bicycle, which changes from has a 16-bit RISC CPU, connected via a data bus (MDB)
CONNECT to the LOCK state. From this moment and a 16-bit address bus (MAB) with memories (RAM and
on, any other command to the bicycle is sent only by Flash/ROM) and peripheral I/O. In addition, it has a 8 MBit
the associated station, all others being discarded. Serial Flash Memory Chip, the M25P80 by STMicroelectron-
ics, with advanced write protection mechanisms and access
The problem which may occur is when two or more via a high speed SPI-compatible bus. Moreover, an electronic
bicycles simultaneously request a station, so that the broadcast registration number with external power supply, the low-cost
welcome message is received by two or more nodes. This DS2411 silicon serial number by Maxim Integrated, has been
problem is solved with a two-steps approach, by sequentially used in order to provide an absolutely unique identity that
scanning all the bicycles first, and choosing the one associated can be determined with a minimal electronic interface and
to the mote with the maximum transmitting power. associated to the network address.
III. W IRELESS S ENSOR N ODE D ESIGN
Thanks to the flexibility of the developed node, it is
The wireless sensor network developed for the manage- possible to use it for both bicycles and stations. The nodes
ment, coordination and control of the electric bicycle-sharing installed on the bicycles manage the wireless communication,
system is composed, in the standard configuration, by 20 the transmission of status from the bicycle to the station,
sensor nodes, 10 of which are associated with the bicycles and the relay for the activation/deactivation of the the motor
10 with the docking and recharging stations. The 10 wireless electronics, the relay for power connection to the batteries,
nodes associated with the docking stations are connected over and the relay for the activation/deactivation of the GPS tracker.
a RS485 bus to the master control unit, which handles the The node of the workstation, instead, manages the contact
interface between the sensor network and the tablet which sensor, the battery charging, the electromechanical lock and
provides for the server connection. The architecture of the communication over the RS485 bus (see Fig.4)
 The Multi-path Ray-tracer Medium (MRM) represents one of
the five features that the Cooja makes available to simulate
the radio transmission medium. In this context it is used to
simulate the transmission power between bicycles and stations.

B. Simulation Scenario
The simulation focus, as shown in Figure 5, has been put
on the wireless sensor network architecture and not on all
hardware and software system functionalities. For this reason,
the RS485 bus communication structure used to excange
hooking and release commands, as well as bike lock and
battery charging, has been omitted. Moreover the simulation
has been limited to 8 nodes (rather than 20).

Fig. 4. Layout of the hardware elements connected by the WSN.

.

IV. W IRELESS S ENSOR N ETWORK S IMULATION

The wireless sensor network has been tested in a co-
simulation environment before actual implementation. This is
useful to debug the code, identify possible problems before
writing the firmware and modify the network topology before
the deployment. This paradigm is typical of the rapid proto-
typing procedure.

A. Cooja WSN Simulator

To this end, the Cooja cross simulator has been used,
since it allows both to emulate the programming code of the
real Argosd I nodes and to simulate some functionalities of
practical interest, such as the nodes transmission power, the
connections between the nodes, etc. The WSN is simulated
by using the Sky motes, which have the same characteristics Fig. 6. Advance Sky GUI to manage the Bicycle-station connection.
of Argosd (see Figure 5). Cooja, however, does not allow an
The digital sensor installed in the dock-point, used to
determine the bike presence, is virtually connected to the
Argosd I analog port of the associated dock-point. The contact
sensor activation, as well as the movement of the node, is
managed through the slider of the Advanced Sky GUI (see
Figures 6 and 7). In detail, four nodes represent four electric
bikes and the other four nodes are associated to the four dock-
points of the docking station. An Argosd I node with its ID
is associated to every bike and dock-points, and each pair
bike-dock-point realizes a point-to-point wireless network. The
dock-point node sends messages in broadcast mode, due to the
impossibility to know in advance what bike, with a specific
network address, requires to engage it. On the other hand, the
packets sent by the bike node are in unicast, in order to reduce
Fig. 5. Cooja Architecutre. the probability of a wrong pairing of a bike with another dock-
point.
easy way to move them within the simulated environment. The handshake between the bike and the dock-points can
To overcome this problem, we have developed the Advanced be decomposed in the following four steps.
Sky GUI software module: it is an extension of Contiki Cooja
network simulator for the Sky mote [13]. In addition to the 1) The first step occurs when the bike is physically
interfaces that are already present, like user button, reset hooked to the dock-point and the presence sensor is
button, LEDs and serial communication, it also implements activated. In this context, the state machine of the bike
an advanced interface to modify all of the 8 ADC ports, the system changes from NONCONNECTED to CON-
2 DAC ports and a joystick which moves the mote around NECTED and the wireless node starts listening the
the virtual environment in accordance with application needs. lock and battery charging commands. Afterwards, the
 is assumed to have four dock-points. In the first case, no
bicycles are connected to the docking station (see Fig. 8(a))
while two bicycles are simultaneously approaching the station.
In the second case, one bicycle is locked to a dock-point,
one dock-point is free, while two bicycles are simultaneously
approaching the remaining two free dock-points (see Fig. 8(b)).
In the last simulated scenario two bicycles are assumed to be
locked to two dock-points, another dock-point is considered
free, while one electric bicycle is approaching the remaining
free dock-point (see Fig. 8(c)). During the simulation with

Fig. 7. Advance Sky GUI to move the motes. (a) First scenario. (b) Second scenario.

WSN node associated to the dock-point is activated

when the bike presence sensor is activated, then it
requires the start of the handshake with the bike to the
master control unit, which sends a welcome broadcast
message to the bike.
2) When the bike receives a welcome message, it ex-
tracts the network address of the dock-point, sends
(c) Third scenario.
an ACK message and waits for the lock and the bike
state command. Fig. 8. The three simulated scenarios.
3) At last, after the dock-point has received the ACK
message from the bike, pairs with it, and commu-
Cooja of the first scenario, the advanced Sky GUI software
nicates the successful pairing to the control unit. In
module has been used to move the bicycle mote for reaching
addition to this, the lock command is sent to the bike,
the dock-point. Once the dock-point has been reached, the
which change its state from CONNECT to the LOCK.
“ADC7 input” has been fixed to its maximum value in order to
Any other command to the bike will only be sent from
show that the physical contact, between bicycle and dock-point,
the associated dock-point, all others will be discarded.
has occurred. Simultaneously on the GUI of the dock-point
When two or more bikes simultaneously engage the dock- mote, the ADC7 input slider is moved to is maximum value, in
points and the welcome messages are sent, there will be two order to notify that one bicycle has been correctly connected to
or more bikes to respond. It may occur that a dock-point is it. After the handshake between the two nodes, namely bicycle
paired with a bike that responds first but which is in a wrong and dock-point, the green LED indicates that the connection
location. To avoid this problem two steps has been followed. has been successfully established. The simulator does not allow
to simultaneously move two bicycles together, and also to
• The first step is the management, in a sequential way, simultaneously fix the ADC7 input of each bicycle and of each
of the pairing procedure, which means that dock-point dock-point. Thus, it is not possible to simulate the simultaneous
nodes can not start the handshake without the autho- coupling of two electric bicycles to the dock station. In order
rization of the master control unit and the procedure to overcome this limitation, the simulation has been stopped,
can be only done one by one and not simultaneously. the bicycle motes have been placed very close to the dock-
• The second step is to analyse the transmitted signal points, and the ADC7 inputs have been fixed to their maximum
power of the welcome message. The message with value; subsequently, the simulation has been started again.
the highest value is considered as sending from the From the data analysis, it is clear that the second scenario
correct bike. Then, its network address is stored in is quite similar to the first one, the presence of a bike already
the dock-points node and the pairing procedure starts. hooked to his position has not effects on the system status.
When a bike is paired with its dock-point, the wireless nodes
do not participate in the network join because they do not
V. R ESULTS
respond to messages from other nodes. Furthermore, according
We tested our WSN for the management of electric bicycle- to these considerations, the third scenario is similar to the
sharing in simulation, where an interesting and frequent event first, due to the fact that the presence or absence of a hooked
has been taken into account, namely the contemporary arrival bike does not affect the handshake procedure. We will then
of two e-bicycles to the docking station. This event is simu- limit our quantitative analysis to the first case only. In the first
lated in three different scenarios, where the docking station simulated scenario, the signal power has the maximum value
TABLE I. P OWER OF THE SIGNAL AMONG BICYCLES AND dock-points
IN THE FIRST SIMULATED SCENARIO .
07/05/2010 entitled “Promotion of interventions in favor of
local production systems, industrial and manufacturing districts
Signal power Dock 1 Dock 2 Dock 3 Dock 4 to strengthen the organizational system and to promote the
Bicycle 1 – – – – integration of productive chains, networks of enterprises and
Bicycle 2 – – – – business combination processes”.
Bicycle 3 -78.38dB -73.27dB -64.86dB -72.68dB
Bicycle 4 -81.70dB -78.45dB -72.80dB -65.32dB R EFERENCES
[1] O. B. Piramuthu and W. Zhou, “Bicycle sharing, social media, and envi-
ronmental sustainability,” in 2016 49th Hawaii International Conference
in correspondence of the bike nearest to the dock-point, as on System Sciences (HICSS), Jan 2016, pp. 2078–2083.
shown in Table I. In the first simulation scenario, Bicycle 3 [2] D. Paul, “Bike-sharing: History, impacts, models of provision, and
and 4 arrive at the same time to the docking station. When future,” Journal of Public Transportation, pp. 41–56, 2009.
these bicycles arrive simultaneously to the docking station, [3] P. Midgley, “Bicycle-sharing schemes: enhancing sustainable mobility
with respect to Dock-point 3, Bicycle 3 has a signal power in urban areas,” United Nations, Department of Economic and Social
Affairs, pp. 1–12, 2011.
equal to -64.86dB while Bicycle 4 has a signal power equal to
[4] G. Danciu, V. Racicovschi, M. Chefneux, and L. Kreindler, “Ecological
-72.80dB (see Table I). Thus, Bicycle 3 results closer to Dock- transportation system based on light electric vehicles,” in 7th Int. Symp.
point 3 with respect to Bicycle 4. Analogously, with respect to on Advanced Topics in Electrical Engineering, May 2011, pp. 1–4.
Dock-point 4, Bicycle 3 has a signal power equal to -72.80dB [5] C. Cherry, S. Worley, and D. Jordan, “Electric bike sharing—system
while Bicycle 4 has a signal power equal to -65.32dB. Thus, requirements and operational concepts,” in Bicycle Transportation, 90th
Bicycle 4 results closer to Dock-point 4 with respect to Bicycle Annual Meeting, Transportation Research Board, 2011.
3. Combining these information, one can state that Bicycle 3 [6] S. Ji, C. R. Cherry, L. D. Han, and D. A. Jordan, “Electric bike sharing:
is in front of Dock-point 3, while Bicycle 4 is on its right side; simulation of user demand and system availability,” Journal of Cleaner
Production, vol. 85, pp. 250–257, 2014.
Bicycle 4 is in front of Dock-point 4, while Bicycle 3 is on
its left side. From the simulation it is clear that the proposed [7] GoBike. [Online]. Available: http://gobike.com/.
solution is robust and prevents any erroneous coupling between [8] BiciMAD. [Online]. Available: http://www.bicimad.com/.
bicycle and dock-point. [9] Bewegen Technologies Inc. [Online]. Available: http://bewegen.com.
[10] VANMOOF [Online]. Available: http://http://vanmoof.com/.
VI. C ONCLUSIONS [11] R. Meireles, J. Silva, A. Teixeira, and B. Ribeiro, “An e.bike design
for the fourth generation bike-sharing services,” in Electric Vehicle
In general a bicycle-sharing system is based on a reduced Symposium and Exhibition (EVS27), 2013 World, Nov 2013, pp. 1–6.
number of key elements, such as a management software, [12] C. Kiefer and F. Behrendt, “Smart e-bike monitoring system: real-
a set of docking stations and bicycles equipped with mea- time open source and open hardware gps assistance and sensor data
for electrically-assisted bicycles,” IET Intelligent Transport Systems,
surement and communication systems and a hardware and vol. 10, no. 2, pp. 79–88, 2016.
firmware layer for the low-level management. In this paper a [13] L. Ciabattoni, A. Freddi, S. Longhi, A. Monteriù, L. Pepa, and
wireless sensor network based management system for a new M. Prist, “An open and modular hardware node for wireless sensor and
design of electric bicycle-sharing has been presented, where body area networks,” Journal of Sensors., vol. 2016, 2016. [Online].
all the information is shared through the network and the Available: http://dx.doi.org/10.1155/2016/2978073
management system can provide high level software services. [14] F. Osterlind, A. Dunkels, J. Eriksson, N. Finne, and T. Voigt, “Cross-
Furthermore, each bicycle can track positions, battery level level sensor network simulation with cooja,” in 31st IEEE Conference
on Local Computer Networks. IEEE, 2006, pp. 641–648.
and communicate them to the docking station, which manages
[15] J. Eriksson, F. Österlind, N. Finne, N. Tsiftes, A. Dunkels, T. Voigt,
the recharge cycle, and locks or releases the bicycle to the R. Sauter, and P. J. Marrón, “Cooja/mspsim: interoperability testing for
user. In order to enhance rapid prototyping procedure, the wireless sensor networks,” in Proc. of the 2nd International Conference
wireless sensor network has been tested in a COOJA co- on Simulation Tools and Techniques. ICST, 2009, p. 27.
simulation environment before real implementation. This is [16] M. Prist, A. Freddi, S. Longhi, and A. Monteriù, “An Integrated Simu-
useful to debug the code, identify possible problems before lation Module for Wireless Cyber-Physical System,” in Proceedings of
writing the firmware and modify the network topology before the 15th IEEE International Conference on Environment and Electrical
Engineering (EEEIC), Rome, Italy, Jun. 2015, pp. 1397–1402.
the deployment. In order to show the effectiveness of the
[17] M. Prist, S. Longhi, A. Monteriù, F. Giuggioloni, and A. Freddi,
proposed solution, three different simulation scenarios of the “An Integrated Simulation Environment for Wireless Sensor Networks,”
system and a solution has been tested. In these scenarios two in Proceedings of the IEEE International Symposium on a World of
simulated nodes have been used to reconstruct the problem that Wireless, Mobile and Multimedia Networks (WoWMoM), Boston, USA,
may occur when two bicycles approach the docking station Jun. 2015, pp. 1–3.
simultaneously. [18] Berkeley Wireless Embedded Systems Project,
(http://webs.cs.berkeley.edu/).
[19] Crossbow Technology Inc., (http://www.xbow.com/).
ACKNOWLEDGMENT
[20] Texas Instruments, “MSP430 16-Bit Ultra-Low Power MCUs.”
The work has been developed in cooperation with Picchio [21] E. Lattanzi and A. Bogliolo, “Virtualsense: A java-based open platform
S.p.A. and ROXOR S.r.l. and it has been supported by the for ultra-low-power wireless sensor nodes,” International Journal of
Italian Funding Program DGR n. 1495/2010 - art. 2 DM Distributed Sensor Networks, pp. 1–16, 2012.
348 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 1, NO. 4, DECEMBER 2015

External-Rotor 6–10 Switched Reluctance Motor

for an Electric Bicycle
Jianing Lin, Senior Member, IEEE, Nigel Schofield, Member, IEEE, and Ali Emadi, Fellow, IEEE

Abstract—As a cost-effective, healthy, and environmental However, there is still a significant challenge for E-bike
friendly personal mode of transportation, electric bicycles development with regard to the tradeoff between E-bike
(E-bikes) are gaining an increasing market share from conven- performance and price. The key difficulty for E-bike manu-
tional bicycles and automobiles. Considering the legal rules in
Ontario, Canada, a 500-W electric motor providing power assist facturers in Europe and North America is competition from
makes the E-bike more attractive to urban commuters. The simple the Chinese E-bike market. Designing a high-performance and
structure, high torque, and power density, as well as the potential low- cost E-bike is the main objective of many E-bike manu-
for low cost make the switched reluctance machine (SRM) a facturers, and involves the design of well-qualified powertrains,
strong candidate for E-bike traction. In this paper, a three-phase, including the electric propulsion machine, electronic controller,
external-rotor SRM with 6 stator poles and 10 rotor poles is
designed for a representative E-bike. The design of an external battery pack, and charger. Thus, the electric machine is an
rotor arrangement of the 6–10 SRM topology has not previously especially important component that influences overall E-bike
been reported, this brings the challenge of sizing the geometry performance and cost.
of this topology, but the solution offers a new contribution to For electric propulsion applications, electric machines with
published works. The external-rotor arrangement is chosen to high torque and power density, high efficiency, high reliability,
facilitate ease of integration into the wheel hub structure of a typi-
cal pedal bicycle. The increased rotor poles yield improved torque and robustness are required [5]. Moreover, reducing cost is of
ripple reduction than more conventional (i.e., 6–4 and 12–8) SRM significant importance to the development of electric vehicles,
design, which is an essential feature for low-speed rider comfort. particularly for the E-bike industry.
The final machine design is experimentally validated via a full A review of the E-bike market shows that the electric propul-
system prototype and dynamometer test facility. Results highlight sion machines are predominantly brushed dc and brushless dc
some limitation of the 2-D finite element analysis (FEA) study in
terms of the winding inductance calculation, more accurate 3-D (BLDC) machines. Lower efficiency and maintenance prob-
FEA model is implemented. lems limit brushed dc machines, thus powertrains employing
BLDC machine have a larger market share [3]. Table I lists
Index Terms—Electric bicycles (E-bikes), end winding, finite
element analysis (FEA), machine design, prototype machine, some popular 500 W BLDC powertrains in the North American
switched reluctance motor. E-bike market. Similarly, review of recent literature also indi-
cates great interest in BLDC over switched reluctance machine
I. I NTRODUCTION (SRM) and brushed dc [6]–[10]. However, the utilization of
high-energy rare earth permanent magnets in BLDC machines
O VER THE PAST decade, there has been an increasing
E-bike market growth, especially in China, slower in
Europe and North America, but with interest increasing. For
poses some uncertainty in the cost of these solutions in the
future given the recent volatility in the cost of key rare earth ele-
ments on the world markets [11]. Even though rare earth prices
example, it has been reported that the population of E-bikes in have reduced and stabilized since 2011, the potential for sup-
China exceeded 140 million in 2012 [1]. In the same year, about ply volatility is still a major concern and hence, the renewed
700 000 E-bikes were sold in Western Europe, nearly half of interest in machine topologies having no permanent magnets.
which were in Germany [2]. As an environmental friendly per- At the same time, the simple structure, high torque, and
sonal transportation, E-bikes also have the advantage of saving power density, as well as low cost of the SRM make it really
rider travel costs considerably. With one full charge, a common attractive for E-bike applications. There is already some pub-
commercial E-bike can travel a distance of about 20–30 km [3], lished evidence that SRMs can compete with PM motors. In
which will save about CAD$3.5 per charge or CAD$1500 per papers [16]–[18], an SRM is designed to compete interior per-
year, considering only gasoline costs as reported in [4]. manent magnet synchronous machines (IPMSM) employed in
Manuscript received May 05, 2015; revised October 07, 2015; accepted the Toyota Prius. Experimental results show that the SRM effi-
November 03, 2015. Date of publication November 20, 2015; date of current ciency, torque, and power-speed range are very competitive
version January 11, 2016. This work was supported in part by the Canada over the IPMSM.
Excellence Research Chairs (CERC) Program and in part by the Natural
Sciences and Engineering Research Council of Canada (NSERC) Discovery
SRMs have already been applied E-bikes, a 6–4 (stator
Grant Program. poles–rotor poles, the same below) and 12–8 SRM being two
The authors are with the McMaster Institute for Automotive Research and topologies that have been reported [19]. However, these motors
Technology, McMaster University, Hamilton, ON L8P 0A6, Canada (e-mail: are all at low power, and their performance and efficiency do
[email protected]; [email protected]; [email protected]).
Color versions of one or more of the figures in this paper are available online
not meet the suggested E-bike requirements for Canada.
at http://ieeexplore.ieee.org. As in-wheel construction is considered, an external-rotor
Digital Object Identifier 10.1109/TTE.2015.2502543 design is chosen, which has the advantages of lighter weight
2332-7782 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
LIN et al.: EXTERNAL-ROTOR 6–10 SWITCHED RELUCTANCE MOTOR FOR E-BIKE 349

TABLE I
C OMPARISON B ETWEEN C OMMERCIAL 500-W E-B IKE BLDC M OTORS
AND THE P ROTOTYPE M ACHINE

Fig. 1. Schematic of E-bike with integrated back wheel drive motor and
gearbox.

* Estimated from references.

on motor power alone [27], while the power and speed
requirement in the USA is 750 W and 32 km/h with
and ease of manufacturing [20], [21]. In-wheel external-
some variation across states [29], Europe has a much
rotor topology machines for EV traction have already been
lower maximum allowed motor power and E-bike speed
published. Among them, the three-phase 6–8 is one of the
of 250 W and 25 km/h, respectively [30].
most popular choices [22], [23]. In [21], a four-phase 8–6
2) For the gear box, its weight and size is usually propor-
external-rotor SRM and its driving system is analyzed and
tional to the gear ratio. But, as SRM has the potential
tested. A two-phase 6–6 SRM with a new sensorless control is
to increase power density at higher rpm, it leads to a
presented in [24]. To reduce torque ripple, there are also topolo-
tradeoff between gear box weigh/size and machine rpm.
gies with higher poles number, e.g., the three-phase 12–16 [25],
However, optimization of the system gear ratio was not
four-phase 16–18 [26], and four-phase 16–20 [20]. However, no
part of this study. An existing gear box design (3:1 reduc-
previous work has been made with three-phase 6–10 external-
tion) was chosen since this resulted in electrical operating
rotor SRM. Thus, in this paper, a 500 W, 500 rpm, external
frequencies of 50–80 Hz in the machine rotor and sta-
rotor, 6–10 SRM is designed and presented systematically. The
tor laminations, respectively, values which were suitable
SRM connected to a 3:1 reduction gear set that, along with the
in-terms of available lamination material.
other components in the powertrain, will make the E-bike road
3) A 48-V, 10-Ah Samsung LiMn battery was chosen as the
speed acceptable in Ontario, Canada [27].
E-bike dc power source since it is popular on 500 W
II. E-B IKE 6–10 SRM D ESIGN E-bike market. First, 48 V is rated voltage for the
machine. At the same time, its maximum discharge cur-
The goal of this paper is to provide a solution which can rent is higher than the rated peak phase current of the
fulfill the E-bike on-road operational requirements in Ontario, designed machine. This can guarantee the battery to oper-
Canada. Therefore, the objective is first specified before the ate safely. If a suitable capacitor placed parallel with dc
design procedure. battery, it could help to extend battery life.
The design process starts with mathematical analysis using 4) In-wheel structure has the potential of lighter weight and
standard design techniques [28]. Mathematical analysis pro- it is ease of manufacture [20], [21]. As the motor is placed
posed the initial design with the assistance of commercial in the middle of wheel, both the diameter and axial thick-
software (in this case speed PC-SRM). Then, 2-D FEA was ness is limited by the wheel structure. The outer diameter
applied to validate the mathematical method and also optimize of 200 mm and a thickness of 25 mm are suitable and also
the stator/rotor pole arcs. popular on the market.
5) Since the E-bike has a low operating speed, there is a strict
A. Design Objective Specification
requirement on machine generated torque ripple.
E-bike can be considered as a conventional pedal bicycle Fig. 1 illustrates an E-Bike schematic identifying the force
equipped with an electric machine, battery, and other auxil- components of E-bike kinematic equations [8]
iaries, such that human pedal and motor power are both used
to propel the rider. Thus, the powertrain has potential for both Fa = M a(t) (1)
parallel and serial traction effort. Fg = M g sin(θ) (2)
Along with the aforementioned requirements of high effi- 1
ciency, high torque and power density, and reasonable speed Fair = ρ M CD Af v(t)2 (3)
2
range, as will be determined, a number of other specifications Fr = M g Cr cos(θ) (4)
had to be considered at the design stage.
1) First of all is the legislation in Ontario, Canada, which is where Fa is the force to accelerate or retard the total mass, Fg
different from other countries. Legislation sets the max- is the gravitational force, Fair is the air friction force, and Fr is
imum electrical output power of E-bikes to be 500 W the wheel-to-road friction force with all the relevant parameters
and the maximum on-road speed not to exceed 32 km/h summarized in Table II.
350 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 1, NO. 4, DECEMBER 2015

TABLE II
E-B IKE K INETIC M ODEL PARAMETERS AND VALUES [8]

Fig. 3. Torque-speed envelope and operating points on the velocity cycle.

force components of (1)–(4), and accounting for the mechanical

power-train components
dwm Jw dww R
Td = Jm + + (Fa + Fg + Fair + Fr )
dt igear dt igear
(5)

where R is E-bike wheel radius, ww , wm , Jw , and Jm are the

wheel and motor angular velocity, wheel, and motor moment of
inertia, respectively, and igear is the gear ratio, 3:1.
Recharging via electrical braking is not considered, at least in
this paper. Also, when the total torque/power to yield the E-bike
load at certain speed is over maximum value allowed by local
legislation, the excess power is assumed to be provided by the
cyclist pedaling. Fig. 3 illustrates discrete load points calculated
at 0.5 s increments of the Portuguese velocity or duty cycle and
envelope of these points. The envelope shows a peak torque of
9.5 Nm at 500 rpm, which is used to specify the electric drive
torque-speed requirement. The Portuguese speed velocity cycle
is under 32 km/h and; with proper power coordination between
machine and human pedaling, the maximum motor power is
500 W, both of which meet the legislation in Ontario, Canada.

B. Phase and Poles Numbers

During the historical development of SRMs, different stator-
rotor pole and phase combinations have been studied, e.g.,
single-phase 8–8 SRM [31], two-phase 4–2 SRM [32], [33],
three-phase 6–4 SRM [34], three-phase 6–8 SRM [35], three-
Fig. 2. (a) Typical urban bike velocity cycle [8]. (b) Calculated E-Bike torque phase 12–8 SRM [36], and four-phase 8–6 SRM [37].
components for one velocity cycle. (c) Motor output power variation for one
As for phase number selection, it is known that the conven-
velocity cycle.
tional one-phase structure SRM has no self-starting ability and
the two-direction drive requirement necessitates the phase num-
ber to be more than two [38]. At the same time, SRM reliability
The kinematic equations in (1)–(4) are used to calculate the and power density will be improved when the phase number
required operating power and torque. Since an equivalent driv- increases and its torque ripple will also be reduced signifi-
ing cycle was not available for Canada, a European (Portuguese cantly. Therefore, it seems a higher phase number is preferred.
standard NP EN 1986-1) cycle was chosen as the reference However, the increased phases will need more power electronic
velocity profile for the machine design, as illustrated in Fig. 2(a) switches, which might increase the design cost and decrease
[8], from which the calculated total torque (Td ), SRM torque efficiency especially at high speed [39], [40]. Therefore, con-
(Tm ), human torque (Tp ), and braking torque (Tb ) required sidering all these factors, a three-phase structure is chosen here
for the E-bike propulsion are shown in Fig. 2(b). All these for the E-bike application.
torque values are referred to machine side. The correspond- As mentioned, the number of phases can influence torque rip-
ing SRM output power versus time requirements are shown ple. Torque ripple reduction can also be achieved by increasing
in Fig. 2(c). The total torque is calculated by summing the the number of poles. The excitation torque ripple frequency is
LIN et al.: EXTERNAL-ROTOR 6–10 SWITCHED RELUCTANCE MOTOR FOR E-BIKE 351

TABLE III TABLE IV

R ESTRICTIONS OF THE TO - BE -D ESIGNED SRM C OMPARISON B ETWEEN A NALYTIC AND 2-D FEA

ratio; Bs is the saturated flux density; As is the specific elec-

tric loading related to the system cooling capability; L is the
stack length, Nrpm is the speed in rpm, and D is the air gap
diameter. Iterating (7) with air gap diameter as well as the other
parameters is augmented by the commercial software, SPEED
PC-SRM.
determined from The ability to start the machine at any rotor position requires
a consideration of the machine pole arcs imposing some con-
w m Ps straints. One should avoid dead zones, which have no induc-
fripple = Pr (6)
2π 2 tance variation for any phase, leading to a zero torque region.
where Pr and Ps are the rotor and stator pole numbers, respec- That is to say, the following constraints (8) need to be satisfied,
tively. A higher rotor or stator pole number increases the torque which have been discussed in [37] and [45] in detail

ripple frequency proportionally, benefiting the average output
min(βs , βr ) ≥ P4π
r Ps
torque, though this comes at the expense of higher losses and (8)
hence lower efficiency [28]. βs + βr < P2πr
For conventional SRMs, it is usually to choose P r = P s ±
where βs is the stator pole arc and βr is the rotor pole arc. For
2, e.g., 6–4 and 6–8. These designs have the features of good
the three-phase, 6–10 external-rotor motor, this equates to
inductance ratio, high average torque, and low switching fre-
quencies. At the same time, it has also been proved that a higher min(βs , βr ) ≥ 12◦ , βs + βr < 36◦ (9)
number of rotor poles will offer a lower torque ripple [38]. In
[41]–[44], a novel internal-rotor SRM with 10 rotor poles and and βs < βr is preferred for this design because a smaller βs
6 stator poles was introduced. The new configuration has been provides a larger winding slot for this external rotor machine.
proved to have higher torque density and lower torque ripple The initial design choose βs = 12◦ and βr = 13◦ .
when compared with a conventional 6–4 or 6–8 SRM. Hence, Table IV compares the results between mathematical anal-
the proposed 6–10 structure is particularly suitable for a three- ysis and 2-D FEA, with Lu as the inductance at unaligned
phase external rotor machine since the external rotor provides position, Lau as the inductance at aligned position before sat-
more space for the rotor poles while increasing the machine air uration. It shows acceptable differences, though the analytical
gap diameter is an important feature for torque density. Thus, unaligned inductance has larger disagreement with 2-D FEA,
the 10-pole rotor and 6-pole stator structure is chosen for this which is mainly caused by inaccurate flux path estimation
low-speed E-bike application. during mathematical analysis.

C. Initial Machine Sizing D. Pole Arcs Optimization

Based on the analysis above, a 6–10 external-rotor SRM Since SRM usually operates with magnetic material sat-
capable of delivering 500 W at 500 rpm with an outer diameter uration to increase the energy conversion, the nonlinearity
of 220 mm and axial length of 25 mm is the design challenge. introduced will highly limit the mathematical analysis accuracy.
With a dc power supply of 48 V and maximum phase current of 2-D FEA is applied for the following optimization process.
35 A, the design objectives are to maximize torque production During the geometry optimization, a multidomain design
and efficiency, while at the same time, reducing torque ripple. methodology was followed, a flow diagram representation of
All the restrictions and objective are summarized in Table III. which is shown in Fig. 4. In this paper, the procedure only
To initialize the design, the air gap diameter is first derived focuses to propose optimum stator and rotor pole arcs, with the
from the output power equation as discussed by Krishnan [28] objective of high-average torque but acceptable torque ripple
and losses. All the other parameters stay as the value pro-
Pout = ke kd k1 k2 Bs As LNrpm D2 (7) posed by the proceeding mathematical analysis. Meanwhile, the
starting torque capability, current density, and slot fill factor
where ke is the efficiency and kd is the duty cycle for the limitations are set as the boundary constraints for this opti-
power electronic converter, both of which are set as 1 at the mization. Moreover, classical angular position control with
initial design stage; k1 = π 2 /120 is a constant coefficient and constant turn-ON/turn-OFF angles is applied during this geome-
k2 a variable dependent on unaligned and aligned inductance try optimization. Nonetheless, as optimization strategies may
352 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 1, NO. 4, DECEMBER 2015

Fig. 5. Designed 6–10 SRM.

TABLE V
S TRUCTURE PARAMETERS

Fig. 4. Multidomain design methodology diagram.

accidently lead to unexpected design results [46], traditional

tuning optimization is preferred and conducted in this paper.
For certain geometry, the electromagnetic characteristics can
be solved accurately and quickly based on the Maxwell’s equa-
tions in the FEA software-JMAG, when the phase current
and number of turns for each phase is already given. The 2-D FEA. It suggests that the necessary of considering end
torque/flux profiles, as look-up tables, are then modeled in winding for the short stack machine and an accurate 3-D FEA
MATLAB/Simulink for machine dynamic performance with model is to be built.
classic turn-ON and turn-OFF angular position control [40].
Therefore, the output torque together with torque ripple is
obtained. The flux density at each position is known from the A. Prototype Set-Up
FE analysis for each element; hence, the iron losses are obtained The 2-D final machine design is shown in Fig. 5. For ease
for each frequency component. At the same time, the copper of prototype manufacture, the winding is implemented with
losses are calculated in the software. The final geometry is 21 parallel strands of 0.51-mm diameter copper wire having
selected according to the balance of the output torque, torque a resistance of 84.22 mΩ/m, instead of the 3.26-mm diameter,
ripple, efficiency, and total machine mass. As the final geom- 2.06 mΩ/m one chosen at the design stage. Also two insulation
etry is decided, standard control iterated around performance sheets are placed on either side of the stator lamination stack,
to choose the best turn-ON/OFF angles. A formal optimiza- thus the thickness of stator had to be reduced to 21.8 mm instead
tion applied the procedure of which proposed an offline convex of 25 mm. Taking these manufacturing issues into considera-
optimizer for this machine is discussed in [47]. tion, the prototype machine’s structure parameters together with
At the last design step, the average losses over a whole initial and final solutions are as listed in Table V.
velocity cycle presented in Fig. 2(a), taken since this is more
representative of actual vehicle operation, are applied to a
detailed 3-D FE thermal analysis to predict the temperature B. Static Tests
variation and distribution. This analysis calculated the high- The flux-linkage profiles are first tested against the 2-D FEA
est temperature occurs in the stator coils but less than 80◦ , results via the static test bench illustrated in Fig. 6. The com-
when there reaches a balance between the heat generation and parison of results is presented in Fig. 7. As shown in Fig. 6, the
radiation, an acceptable upper limit for this application. stator is locked by the load cell arm which locates on the shaft.
As the connection between the shaft (stator) and bracket (rotor)
is a bearing, the reaction force between the rotor and stator are
III. R ESULTS AND E XPERIMENT VALIDATION all placed on the load cell from which torque can be calibrated.
MATLAB/Simulink and 2-D FEA are mainly used at the The rotor on the case is referenced to an angular measure-
design stage to predict machine static/dynamic performance. To ment; hence by incrementally rotating the measurement dial,
realize the design and verify the analysis results, a dynamome- the relative position of the stator and rotor can be accurately
ter test bench is designed and the system assembled. obtained.
However, the disagreement between experimental and 2-D During the static test, the flux-linkages are measured for dif-
FEA results leads the author to rethink about the accuracy of ferent currents at different rotor positions. The current range is
LIN et al.: EXTERNAL-ROTOR 6–10 SWITCHED RELUCTANCE MOTOR FOR E-BIKE 353

Fig. 9. Dynamometer test bench.

Fig. 6. Statics test bench setup.

Fig. 10. Flux-linkage-current curves for different turn-OFF angles.

shows much improved agreement especially at the unaligned

position and the aligned saturated region.
Fig. 7. Flux linkage profile comparison of the 2-D FEA and experimental test. From these comparisons, it is concluded that the end winding
influence for this short stack machine is much larger than con-
ventional machines. This influence is especially obvious at the
unaligned position. Since the end winding can introduce more
flux linkage for a given current, it will lead to larger inductance
at the unaligned position for the testing machine and 3-D model
than the 2-D model.
However, for the aligned saturated region, the most impor-
tant issue that caused mismatch between the 2-D model and
test/3-D model is the stator stack length change. For a saturated
flux density, a shorter stack will proportionally decrease the flux
linkage as Fig. 7 shows.
Fig. 8. Flux linkage comparison between 3-D FEA and experimental test.

C. Dynamic Tests
from 0 to 36 A, with an increment step of 6 A. The aligned Fig. 9 shows the dynamometer test bench. For the dynamic
and unaligned positions are checked by the output force, which load profile study, a no-load test is conducted first. During the
should be zero for both cases, and by exciting one stator phase no-load test, the pulley and belt system in Fig. 9 is discon-
to check the angular reference. nected. The test is at rated speed of 500 rpm and calculation
A 3-D FEA model is not practicable at the design stage, demonstrates the no-load mechanical losses are 3.3 W.
especially when the dimension of the end winding is not avail- As discussed, torque ripple is an issue for SRM design.
able. Hence 2-D FEA model is used to represent the designed However, because of the experimental limitation, the instan-
machine before the prototype, without considering the end taneous torque is not measurable, only steady average torque.
winding or the manufacturing issues. As can be seen in Fig. 7, Torque ripple evaluation is part of ongoing studies.
there is big difference between the predicted and measured For a constant turn-ON angle of 0 mechanical degree, which
results. For the unaligned rotor position, the test results show respects to the unaligned position, turn-OFF angles of 14◦ , 15◦ ,
a much higher slope than that obtained from 2-D FEA. At the and 16◦ were studied, as shown in Fig. 10. The flux-linkage-
aligned position, especially for the saturated region, the test current curves are derived from the integration of measured
flux-linkage values are lower than the 2-D FEA results. voltage minus the voltage drop due to phase resistance, which is
Having the final design, a 3-D FEA model with detailed end assumed to be constant during the short period of test. It shows
winding and accurate stator stack length was studied and results that these turn-OFF angles have little influence on the machine
compared with the test results in Fig. 8. As shown, the results output torque capability.
354 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 1, NO. 4, DECEMBER 2015

Fig. 11. Flux-linkage-current curves for different turn-ON angles.

Fig. 12. Experimental phase voltage and currents at 500 rpm with excitation
TABLE VI angle of (−3, 14).
P ERFORMANCE W ITH T URN - ON OF −3, AND T URN - OFF OF
14◦ AP C ONTROL

Fig. 13. Currents comparison between experiment and 3-D FEA model at
500 rpm with excitation angle of (−3, 14).

Considering the current increasing time, the flux-linkage-

current curve is tested with a 3◦ -earlier turn-ON angle, com-
pared with 0◦ turn-ON angle in Fig. 11, showing that an
advanced turn-ON angle will provide more output energy con-
version.
By calculating the area of the flux-linkage-current loop with
a turn-ON angle of −3 and turn-OFF angle of 14◦ , the average
output torque is estimated as a check of the torque measured
from the dynamometers. The phase resistance is measured dur-
ing the test (0.1 Ω), from which the copper losses can be
estimated in conjunction with the measured rms phase currents. Fig. 14. Experimental phase voltage and currents at 500 rpm with excitation
angle of (0, 16).
Mechanical no-load losses are measured leaving the remain-
ing power assumed to be iron losses. By summarizing all those
components of power, efficiency is calculated and compared
with analysis results as shown in Table VI.
In this comparison, the copper losses increase about 85%
for the prototype from the analytical design. However, this
is consistent with the reduced copper cross-section used to
manufacture the prototype due to problems with hand-winding
the stator coils. Because of the end winding issue and stator
stack length reduction, the torque capability of the prototype
is reduced compared with the analysis results at the design
stage. As a result of these two factors, the machine efficiency
decreases about 10%. These performance reductions are not Fig. 15. Currents comparison between experiment and 3-D FEA model at
500 rpm with excitation angle of (0, 16).
considered as detrimental to the design approach of machine
design since the manufacturing procedure could be significantly
improved by industry. the model is checked against the recorded phase current with a
Fig. 12 also shows the recorded experimental data of phase turn-ON angle of −3◦ and turn-OFF of 14◦ in Fig. 13. Figs. 14
current and phase voltage at 500 rpm. The phase is turn ON from and 15 show another example of recorded phase current and
−3◦ and turned OFF at 14◦ . A machine dynamic model is set-up phase voltage at 500 rpm, but with turn-ON angle of 0◦ and
in MATLAB/Simulink based on the static 3-D FEA data. Here, turned OFF of 16◦ .
LIN et al.: EXTERNAL-ROTOR 6–10 SWITCHED RELUCTANCE MOTOR FOR E-BIKE 355

The comparison in Figs. 13 and 15 both show good [5] K. T. Chau, C. C. Chan, and C. Liu, “Overview of permanent-magnet
agreement, so that it is concluded, as for this short-length brushless drives for electric and hybrid electric vehicles,” IEEE Trans.
Ind. Electron., vol. 58, no. 6, pp. 2246–2257, Jun. 2008.
machine, the 3-D-FEA-based model can more accurately pre- [6] E. Starschich and A. Muetze, “Comparison of the performances of dif-
dicts machine performance than the 2-D model. For con- ferent geared brushless-DC motor drives for electric bicycles,” in Proc.
ventional SRM, their axial length is usually larger than the IEEE Elect. Mach. Drives Conf., Antalya, Turkey, 2007, pp. 140–147.
[7] W. Chlebosz, G. Ombach, and J. Junak, “Comparison of permanent mag-
diameter. 2-D FEA is usually applied as the magnetomotive net brushless motor with outer and inner rotor used in e-bike,” in Proc.
force produced by the end winding is ignorable. However, for 19th Int. Conf. Elect. Mach., Rome, Italy, 2010, pp. 1–5.
this machine, the end winding will definitely increase the mag- [8] D. M. Sousa, P. J. C. Branco, and J. A. Dente, “Electric bicycle using
batteries and super-capacitors,” in Proc. Eur. Conf. Power Electron. Appl.,
netic flux compared with 2-D FEA results. The inductance Aalborg, Netherlands, Sep. 2007, pp. 1–10.
will be larger especially at unaligned position and aligned but [9] T. F. Chan, “In-wheel permanent-magnet brushless DC motor drive
unsaturated condition. Thus, the predicted output power/torque for an electric bicycle,” IEEE Trans. Energy Convers., vol. 17, no. 2,
pp. 229–233, Jun. 2002.
will be different. To achieve a good performance prediction, an [10] E. A. Lomonova, A. J. A. Vandenput, and J. Rubá, “Development of an
accurate end winding model is needed for short-stack machine. improved electrically assisted bicycle,” in Proc. IEEE Ind. Appl. Conf.,
Pittsburgh, PA, USA, Oct. 2002, pp. 384–389.
[11] M. Humphries, “Rare earth elements: The global supply chain,”
Congressional Research Service, Library of Congress, Washington, DC,
IV. C ONCLUSION CRS Report for Congress, 2013.
[12] Cyclone-USA. Cyclone 500 Watt Brushless Motor w/ Gearbox [Online].
In this paper, a novel SRM is designed for E-bike appli- Available: http://www.cyclone-usa.com/store.php?crn=200, Accessed
cation. The design is a three-phase, external-rotor structure on: Dec. 03, 2015.
with 10 rotor poles and 6 stator poles, a topology not previ- [13] Currie Technologies. Electric Bikes, Currie Electro-DriveConversion Kit
3700C with Plug + Play Battery Pack [Online]. Available: http://www.
ously reported in literature. It meets the E-bike’s requirement of currietech.com/dealers/wiki/images/7/7b/ConversionKit_3_700C_SS.
500 W output power at 500 rpm. With a reduction gearbox, the pdf, Accessed on: Dec. 03, 2015.
E-bike speed can be restrained under 32 km/h. The prototype [14] BionX International Corporation. BionX D500DV [Online]. Available:
http://ridebionx.com/products/ebike/ebike-systems/d-500-dv/#features,
machine has been built and test validated. Accessed on: Dec. 03, 2015.
There are two parts to the experimental test, the static test [15] TDCM Motor. 500 W+Rear Hub Motor [Online]. Available:
presents the flux-linkage profile for different currents and dif- http://www.tdcm-motor.com/products/electric-hub-motors/500w–
rear-hub-motor/, Accessed on: Dec. 03, 2015.
ferent rotor positions. However, the difference between the [16] Y. Takano et al., “Design and analysis of a switched reluctance motor for
test results and 2-D FEA shows a necessary to consider the next generation hybrid vehicle without PM materials,” in Proc. Int. Power
end winding for short-axial length machines. Decreasing of Electron. Conf., Sapporo, Japan, 2010, pp. 1801–1806.
[17] A. Chiba, Y. Takano, S. Member, and M. Takeno, “Torque density and
the stator lamination stack caused by manufacturing error also efficiency improvements of a switched reluctance motor without rare-
influenced the test machine performance. The dynamic test earth material for hybrid vehicles,” IEEE Trans. Ind. Appl., vol. 47, no. 3,
results show that the maximum output mechanical power at pp. 1240–1246, May/Jun. 2011.
[18] K. Kiyota, S. Member, and A. Chiba, “Design of switched reluctance
500 rpm is about 364 W, with an efficiency of 77%. The output motor competitive to 60 kW IPMSM in third generation hybrid electric
power is lower than the analysis results, which is mainly caused vehicle,” IEEE Trans. Ind. Appl., vol. 48, no. 6, pp. 2303–2309, Nov./Dec.
by the end winding and stator stack thickness reduction. 2012.
[19] H. Chen, “The switched reluctance motor drive for application in electric
Due to the experimental limitations, the prototype torque rip- bicycle,” in Proc. IEEE Int. Symp. Ind. Electron. (ISIE’01), Jun. 2001,
ple is not available currently, but will be studied in future work. vol. 2, pp. 1152–1156.
Further, the dynamometer has a noise profile that overshadows [20] S. Fujishiro, K. Ishikawa, S. Kikuchi, K. Nakamura, and O. Ichinokura,
“Design of outer-rotor-type multipolar SRM for EV,” J. Appl. Phys.,
that of the 6–10 SRM. Further work is ongoing to improve the vol. 99, no. 8, Apr. 2006. Art. no. 08R324
dynamometer test facility for future noise and vibration studies. [21] J. Lin, K. W. Cheng, Z. Zhang, and X. Xue, “Experimental investigation
This paper is modified based on a conference paper with of in-wheel switched reluctance motor drive system for future electric
vehicle,” in Proc. IEEE Power Electron. Syst. Appl., Hong Kong, 2009,
more than a 50% difference in material with a focus on pp. 1–6.
experimental verification. [22] H. Chen and J. J. Gu, “Switched reluctance motor drive with external
rotor for fan in air conditioner,” IEEE/ASME Trans. Mechatron., vol. 18,
no. 5, pp. 1448–1458, Oct. 2013.
ACKNOWLEDGMENT [23] M. A. Cinar and F. E. Kuyumcu, “Design and torque profile analysis of an
outer-rotor SR motor with different winding configurations,” PRZEGLD
The authors would like to thank Powersys Solutions for their ELEKTROTECHNICZNY (Electrical Review), ISSN 0033-2097, no. 2,
pp. 328–331, 2012.
support with the JMAG software. [24] J. Vazquez and P. Salmeron, “Active power filter control using neural
network technologies,” Proc. IEE—Elect. Power Appl., vol. 150, no. 5,
pp. 139–145, Mar. 2003.
R EFERENCES [25] W. Yaling, X. Yanliang, W. Yufang, and Z. Yun, “Outer-rotor switched
reluctance motor and its control system used in electric vehicles,” in Proc.
[1] T. Li, (2012, Nov. 01), New National Standards on E-Bikes to be Out Int. Conf. Elect. Mach. Syst., Aug. 2011, pp. 1–4.
Soon [Online]. Available: http://www.chinadaily.com.cn/business/2012- [26] A. Labak and N. C. Kar, “Outer rotor switched reluctance motor design
11/01/content_15866195.htm. Accessed on: Dec. 03, 2015. for in-wheel drive of electric bus applications,” in Proc. 20th Int. Conf.
[2] P. Walker, (2012, Oct. 12), On Your E-Bike: European Craze Powers Into Elect. Mach. (ICEM’12), 2012, pp. 418–423.
UK [Online]. Available: http://www.guardian.co.uk/lifeandstyle/2012/ [27] Insurance Corporation of British Columbia, (2002, Jun. 21), Motor
oct/12/e-bike-european-craze-uk, Accessed on: Dec. 03, 2015. Vehicle Act–MOTOR ASSISTED CYCLE REGULATION [Online].
[3] A. Muetze and Y. C. Tan, “Electric bicycles-a performance evaluation,” Available: http://www.bclaws.ca/civix/document/id/complete/statreg/
IEEE Ind. Appl. Mag., vol. 13, no. 4, pp. 12–21, Jul./Aug. 2007. 151_2002, Accessed on: Dec. 03, 2015.
[4] Electric Bikes, Cost of Operating Electric Bikes. eCo Wheelz, MI, [28] R. Krishnan, Switched Reluctance Motor Drives: Modeling, Simulation,
USA, [Online]. Available: http://www.eco-wheelz.com/articles/cost-of- Analysis, Design, and Application. Boca Raton, FL, USA: CRC Press,
operating-electric-bicycle.php, Accessed on: Dec. 03, 2015. 2001.
356 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 1, NO. 4, DECEMBER 2015

[29] E. Hicks, (2013, Apr. 23), Is My E-Bike Legal? USA Ebike Law [Online]. Nigel Schofield (M’06) received the B.Eng. (Hons.)
Available: https://www.electricbike.com/electric-bike-law/, Accessed on: and the Ph.D. degrees in electrical power engineering
Dec. 03, 2015. from the University of Sheffield, Sheffield, U.K., in
[30] Editorial Office Bike Europe, (2010, Aug. 05), EU Regulations for 1990 and 1997, respectively.
E-bikes & Pedelecs. [Online]. Available: http://www.bike-eu.com/ From 1993 to 1995, he served as a Senior
laws-regulations/artikel/2010/8/eu-regulations-for-e-bikes-pedelecs-part- Experimental Officer with the Department of
1-1018836, Accessed on: Dec. 03, 2015. Electronic and Electrical Engineering (EEE), Faculty
[31] S. Tadashi, P. C. Kjaer, C. Cossar, and T. J. E. Miller, “Fault tolerant of Engineering, University of Sheffield, before tak-
operation of single phase switched reluctance generator,” in Proc. 25th ing up the post of a Design Engineer in industry.
Annu. Appl. Power Electron. Conf. Expo. (APEC’97), Atlanta, GA, USA, From 1997 to 2001, he was a Postdoctoral Researcher
1997, pp. 553–558. with the Electrical Machines and Drives Research
[32] W. Pengov, J. R. Hendershot, and T. J. E. Miller, “A new low-noise two- Group, Department of EEE, University of Sheffield, where he then served
phase switched reluctance motor,” in Proc. Int. Conf. Elect. Mach. Drives, as a Lecturer from 2001 to 2004. On July 1, 2004, he was appointed to
San Antonio, TX, USA, 2005, pp. 1281–1284. a Mechatronics Lectureship with the School of Electrical and Electronic
[33] L. Z. Sun, G. Yang, and Q. Feng, “Study on the rotor levitation of one Engineering, University of Manchester, Manchester, U.K., where he became
high speed switched reluctance motor,” in Proc. 32nd IEEE Annu. Conf. a Senior Lecturer in 2009, lecturing across all undergraduate years and on
Ind. Electron., Paris, France, 2006, pp. 1322–1325. the M.Sc. program on power conversion. Since January 1, 2013, he has been
[34] F. Soares and P. J. Costa Branco, “Simulation of a 6/4 switched reluctance with the Department of Electrical and Computer Engineering, Faculty of
motor based on Matlab/ Simulink environment,” IEEE Trans. Aerosp. Engineering, McMaster University, Hamilton, ON, Canada, as a Full Professor
Electron. Syst., vol. 37, no. 3, pp. 989–1009, Jul. 2001. with Tenure. His research interests include electromagnetic powertrains for
[35] H. Chen and C. Zhang, “A three phase 6/8 structure switched reluctance all-electric and hybrid electric vehicles, the vehicular application of elec-
motor drive,” in Proc. Int. Conf. Power Syst. Technol., Perth, Australia, trochemical and electrostatic energy sources, hydrogen fuel cell systems,
2000, pp. 195–200. aerospace machines and actuators, and the industrial applications of electro-
[36] H. Chen and J. Gu, “Implementation of the three-phase switched reluc- magnetic devices and novel machines. He is a Member of the Institution of
tance machine system for motors and generators,” IEEE/ASME Trans. Engineering and Technology, Stevenage, U.K., and he is a Chartered Engineer
Mechatron., vol. 15, no. 3, pp. 421–432, Jun. 2010. in the U.K.
[37] A. K. Jain and N. Mohan, “Dynamic modeling, experimental characteri-
zation, and verification for SRM operation with simultaneous two-phase
excitation,” IEEE Trans. Ind. Electron., vol. 53, no. 4, pp. 1238–1249,
Aug. 2006. Ali Emadi (S’98–M’00–SM’03–F’13) received the
[38] T. J. E. Miller, “Switched reluctance motors and their control,” in B.S. and M.S. degrees in electrical engineering
Monographs in Electrical and Electronic Engineering, vol. 31, Oxford, with (highest distinction) from Sharif University of
U.K.: Clarendon, 1993, pp. 32–35. Technology, Tehran, Iran, in 1995 and 1997, respec-
[39] T. J. E. Miller, Electronic Control of Switched Reluctance Machines. tively, and the Ph.D. degree in electrical engineering
Oxford, U.K.: Newnes, 2001. from Texas A&M University, College Station, TX,
[40] A. Matveev, “Development of methods, algorithms and software for opti- USA, in 2000.
mal design of switched reluctance drives,” Ph. D. dissertation, Tech. Univ. He is the Canada Excellence Research Chair
Eindhoven,Eindhoven, Netherlands, Jan. 2006. in Hybrid Powertrain and Director of McMaster
[41] B. Bilgin, A. Emadi, and M. Krishnamurthy, “Design considerations for Institute for Automotive Research and Technology
switched reluctance machines with a higher number of rotor poles,” IEEE (MacAUTO), McMaster University, Hamilton, ON,
Trans. Ind. Electron., vol. 59, no. 10, pp. 3745–3756, Oct. 2012. Canada. Before joining McMaster University, he was the Harris Perlstein
[42] P. C. Desai, M. Krishnamurthy, N. Schofield, and A. Emadi, “Novel Endowed Chair Professor of Engineering and Director of the Electric Power
switched reluctance machine configuration with higher number of rotor and Power Electronics Center and Grainger Laboratories, Illinois Institute
poles than stator poles: Concept to implementation,” IEEE Trans. Ind. of Technology (IIT), Chicago, IL, USA, where he established research and
Electron., vol. 57, no. 2, pp. 649–659, Feb. 2010. teaching facilities as well as courses in power electronics, motor drives, and
[43] P. C. Desai, “The novel concept of switched reluctance machines with vehicular power systems. He was the Founder, Chairman, and President of
high number of rotor poles.” Ph. D. dissertation, Illinois Inst. Technol., Hybrid Electric Vehicle Technologies, Inc. (HEVT), IIT. He is the principal
Chicago, IL, USA, May 2009. author/coauthor of over 350 journal and conference papers as well as several
[44] P. C. Desai and A. Emadi, “Switched reluctance machine,” U.S. Patent 7 books including Vehicular Electric Power Systems (2003), Energy Efficient
230 360, Jun. 12, 2007. Electric Motors (2004), Uninterruptible Power Supplies and Active Filters
[45] P. J. Lawrenson, J. M. Stephenson, P. T. Blenkinsop, J. Corda, and (2004), Modern Electric, Hybrid Electric, and Fuel Cell Vehicles, (2nd ed,
N. N. Fulton, “Variable-speed switched reluctance motors,” IEE Electr. 2009), and Integrated Power Electronic Converters and Digital Control (2009).
Power Appl., vol. 127, no. 4, pp. 253–265, Jul. 1980. He is also the editor of the Handbook of Automotive Power Electronics and
[46] T. J. E. Miller, “Optimal design of switched reluctance motors,” IEEE Motor Drives (2005) and Advanced Electric Drive Vehicles (2014).
Trans. Ind. Electron., vol. 49, no. 1, pp. 15–27, Feb. 2002. Dr. Emadi was the Inaugural General Chair of the 2012 IEEE Transportation
[47] F. Peng, J. Lin, J. Ye, and A. Emadi, “Optimal calculation method for Electrification Conference and Expo (ITEC) and has chaired several IEEE
control of switched reluctance motor,” in Proc. IEEE Transp. Electrif. and SAE conferences in the areas of vehicle power and propulsion. He was
Conf. Expo. (ITEC’15), Dearborn, MI, USA, 2015, pp. 1–6. the Advisor for the Formula Hybrid Teams at IIT and McMaster University,
which won the GM Best Engineered Hybrid System Award at the 2010,
2013, and 2015 competitions. He is the founding Editor-in-Chief of the
IEEE T RANSACTIONS ON T RANSPORTATION E LECTRIFICATION. He was the
Jianing Lin (S’13) received the B.Eng. degree in
recipient of numerous awards and recognitions.
electrical engineering from Southeast University,
Nanjing, China, in 2011, and the M.A.Sc. degree in
mechanical engineering from McMaster University,
Hamilton, ON, Canada, in 2014. Currently, she is
pursuing the Ph.D. degree in electrical and computer
engineering at McMaster University.
She has experience in both electrical and mechani-
cal engineering. She was the recipient of the National
Scholarship in China, as an outstanding graduate
receiving several honors in the fields of electrical
engineering and business. In 2009, she began her research in electric machines
and power electronics. Her research interests include power electronics, electric
machine design and controller design for applications in electric vehicles, home
applications, low/high-speed low-power switched reluctance motor designs
and controllers. These designs are suitable for electric bicycles and home
applications.
2020 IEEE Region 10 Symposium (TENSYMP), 5-7 June 2020, Dhaka, Bangladesh

Implementation of Torque Sensor Technology in

Electric Bicycle with Solar Charging Kit

Snigdha Islam Rabib Ibne Hossain Sams Shafiul Amin

Department of Electrical and Department of Electrical and Department of Electrical and
Electronic Engineering Electronic Engineering Electronic Engineering,
Brac University Brac University Project Engineer,
Dhaka, Bangladesh Dhaka, Bangladesh Control & Applications Research
[email protected] [email protected] Centre, Brac University
Dhaka, Bangladesh
AKM Ferdous Dr. AKM Abdul Malek Azad [email protected]
Department of Electrical and Department of Electrical and
Electronic Engineering Electronic Engineering
Brac University Professor, Brac University
Dhaka, Bangladesh Dhaka, Bangladesh
[email protected] [email protected]

Abstract— Bicycles are one of Bangladesh's most efficient can circumvent the heavy traffic of the city, provide a more
methods of transport, despite the narrow roads and heavy environmentally friendly way to traverse the roads and to
traffic in the region. The paper discusses the torque sensor create a system that can work both on and off-grid.
technology being incorporated into the overall electrical system
of the bike. Integrating components of torque sensor technology Previously, V. Dimitrov designed an electrical bicycle
greatly reduces human effort by providing help from the battery mainly focused on electrical, mechanical, and system-level
bank to power a Brushless DC motor. It has also been ensured design. The electrical domain includes the following
that the e-bike has an option of charging off-grid via a solar questions- what types of the motor need to be used? How has
charging kit, where the batteries of the bike will be charged by the battery been charged? On the other hand, the Mechanical
PV panel using the battery swapping method. This paper domain comprises – Where is the torque imposed by the
describes field tests in three different stages conducted on the e- machine? Additionally, System Level Design subsumes –
bike, one with the pedal, one with the throttle only, and another Should it just assist with the pedal or can it be electrically
with varying loads along with the design of the solar charging driven? Moreover, this study illustrates the possible solution
kit. When evaluating the test results with the torque sensor
for the e-bike [1]. Rashid et al. implemented the vehicle's
pedal, it becomes apparent that the e-bike travels a longer
distance, decreases the power dissipation from the batteries and
overall electrical system with torque sensor technology. This
reduces the battery bank's energy consumption from the main study signifies the Design and Implementation of the
grid as it has an alternating charging option from solar energy. Electrical System incorporate with different components. The
entire system was developed after the integration of the torque
Keywords—E-bike, Torque Sensor, Solar Panel, Solar sensor module with the pedal [2]. F. R. Khan et al. worked on
Charging Kit, Renewable Energy Technology a torque sensor pedal with a solar battery charging station to
charge the batteries. Furthermore, this study demonstrates the
I. INTRODUCTION reduction of energy consumption from the national grid [3],
[4], [5]. R. Tarek et al. presented the torque sensor pedal on
In Bangladesh, traffic is enormous and it has become a
the vehicle to decrease battery bank overuse. Field tests were
daily challenge faced by all people from one point to the other.
also performed at various stages to ensure its overall viability
Carrying out basic day-to-day activities has become
[6].
increasingly difficult, and seeking a solution to this problem
has become a necessity. So, Bicycles are a highly efficient
II. PROJECT OVERVIEW
means of transport. But people will feel the exhaustion from
riding all the time to faraway places. Therefore, Electric The e-bike, as seen in Figure 1, has been designed to
Bicycle was introduced. An Electric Bicycle is a bicycle with work with full throttle-control and to work with torque- sensor
a motor system attached to it that assists the rider when the paddle simultaneously.
rider is no longer able to or desires not to pedal to get to his
destination. They have been widely expanded as they are The main objective of developing this bike has always
environmentally friendly and a cheap, efficient, and healthy been to produce something that is both energy-efficient and
way of travel. Even if the problems such as hills, fitness levels, cost-effective. Maintaining the integrity of the specifications,
health, or simple convenience interfere with enjoying a ride the e-bike consists of a standard bicycle that has been
on a bicycle, it's the e-bike that can take the scene and provide retrofitted with a torque sensor paddle, two 12V 30Ah Lead
a solution. It is a more convenient and easier form of the Acid batteries, a throttle, a power key, a motor controller, a
bicycle that throws out the disadvantages of a regular bike. brake system, a 250W BLDC motor, and our very own torque
Our goal with the torque sensor-based e-bike is to create a sensor control circuit. The e-bike also comes with a solar
more comfortable, easily accessible transport alternative that charging kit that consists of a PV panel and a charge controller

978-1-7281-7366-5/20/$31.00 ©2020 IEEE

718

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 15,2021 at 17:28:23 UTC from IEEE Xplore. Restrictions apply.
 In this capacity, our solar charging kit, which consists of a
400W solar panel and a solar charge controller, can be a viable
off-grid solution to those who want to use the electric bike but
live in remote areas where the grid hasn’t reached yet or the
cost is beyond their affordability. The use of this charging
method will ease the pressure on the main grid a well [7].

III. TORQUE SENSOR CIRCUIT

The torque sensor circuit consists primarily of three
components: a voltage regulator, a voltage divider, and an
amplifier circuit. Along with LM7805, the capacitors are used
to sift through any voltage and commotion when the voltage
is connected over the capacitor. Figure 4 displays the torque
sensor circuit [2], [3].
Fig 1: Overview of the E-Bike

which charges another group of two 12V 30Ah batteries that

can be used for swapping with the discharged ones of the e-
bike.
Figure 2 shows the system's overall signal flow diagram.

Fig. 4. The Torque Sensor Circuit

The LM358 amplifier is used to keep the voltage in the

circuit as low as it would be prudent to extend the circuit's
lifetime accordingly. Furthermore, this particular amplifier is
used mainly because of its maximum voltage capacity of 3.7V
which prevents the output voltage from crossing the threshold
voltage and thus prevents the system from suddenly stopping
Fig. 2. Signal Flow Diagram of the System [8]. The amplifier's gain can be altered through the
potentiometer as seen in Figure 4 which acts as the amplifier's
The entire system is powered by the 24V battery bank. The
feedback resistor Rf. This allows the rider to adjust the gain
torque sensor circuit only requires a 12V input to perform and
and to set it to suit his preferences. If the riders are old or
thus we used a buck converter to step down the voltage of the
extremely tired, they can increase the gain to get more
supply. The torque sensor circuit sends a 5V signal to power
assistance from the torque sensor while on the flip side if the
up the torque sensor module that collects and processes the
bike is low on battery and it would be better to conserve
torque sensor pedal signals.
energy, they can lower the gain for less battery consumption
Figure 3 shows the solar charging kit. Moreover, [2], [4]. After practical use of the bike, the voltage divider was
electricity is a valued commodity in Bangladesh. Thus, many installed, seeing that the voltage generated by the paddling
are now choosing to opt for a more readily available source of was very high. Thus, the resistors were incorporated in such a
energy like solar energy. way that the voltage generated by the torque sensor would be
only 0.6 times what it would have been without the divider.

IV. FIELD TEST

In the pursuit of finding out the overall feasibility of our
vehicle and to account for system battery and load profiles,
necessary tests were conducted. These tests were performed
in areas with minimal traffic and under various terrains to
improve the efficiency of test results. Tests were generally
carried out with a gain set at 50 percent unless otherwise
required. GPS tracking Android software known as Cycling –
Bike Tracker (Zeopoxa) was used for recording the total
distance traveled by the e-bike throughout the different stages
of the field test.
Fig. 3. Dedicated Solar Charging Kit

719

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 15,2021 at 17:28:23 UTC from IEEE Xplore. Restrictions apply.
 V. FIELD TEST ANALYSIS consumption we can safely say that when the system is kept at
full gain, there is more energy consumed from the battery than
A. Load Power with and without Torque Sensor when it is kept at the half gain and even more than when it is
For this test, we tested the entire system first with the at a quarter gain. The total energy consumed by the load was
torque sensor and without it. The battery voltage and current 31.5 KJ when the system was at FG condition while the total
along with the load voltage and current were recorded and the energy consumed in HG condition was 19.4 KJ and the total
results observed. From the area under the graphs we can see energy consumed for QG condition was 15.4 KJ.
that without the torque sensor, the load consumes 98.5 KJ of
energy while with the torque sensor it only consumes 51.9
KJ. This goes to show that incorporating the torque sensor
saves 47.3% more energy.

Fig. 8. Comparative Load Power Consumption at Different Gains

C. Effect of Riders of Different Weight Groups

Fig. 5. Load Power Consumption, without Torque Sensor So far, all the tests have been carried out with a rider
weighing 70 Kilograms. We had three weight riders- 60KG
(Lightweight Rider), 70 KG (Middleweight Rider), and 80 KG
(Heavyweight Rider) who took turns in riding the bike

Fig. 6. Load Power Consumption, with Torque Sensor

B. Effect of Changing Gain

For this test, we changed the gain of the amplifier to test
out the effect on the e-bike. The gain was changed to 25% or Fig. 9. Comparative Battery Power Analysis of Riders of Different Weight
Quarter Gain (QG), 50% or Half Gain (HG), and then 100% or Groups
Full Gain (FG). From an analysis of the power

Fig. 10. Comparative Load Power Analysis of Riders of Different Weight

Fig. 7. Comparative Battery Power Consumption at Different Gains Groups

720

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 15,2021 at 17:28:23 UTC from IEEE Xplore. Restrictions apply.
have recorded the results for this test. Surprisingly, Deep Cycle batteries for future work on the bike. It will
irrespective of the rider's weight, the power consumed by the discharge the batteries up to 20 percent SOC (State of Charge)
load seemed to vary somewhat insignificantly as the instead of the 50 percent SOC which we are currently using.
Lightweight Rider consumed 17.9 KJ of energy while the Thus, our system will be more cost-efficient and help
Middleweight and Heavyweight Riders consumed 18.1 KJ minimize the risk of damage.
and 19.5 KJ respectively.
Owing to the use of a torque sensor pedal, field test results VIII. CONCLUSION
and analysis indicate greater energy accumulation. Therefore,
it leads to a greater distance range compared to that without In a growing developing country such as Bangladesh, the
using the torque sensor pedal. e-bike will one day revolutionize the way people travel, and
provide people with a better, more environmentally
sustainable alternative. Introducing the torque sensor
VI. COMPARATIVE STUDY technology, it reaffirms the minimization of the energy
For the comparative e-bike analysis, we had to take into consumption from batteries by the load. Implementing the
account the e-bikes that are currently on the market which dedicated solar charging kit helps the user to have backup
have both pedal and throttle capabilities, i.e. the features are batteries which will be powered completely by PV panels.
similar to the working of our e-bike torque sensor as such, we Therefore, by adopting this renewable energy technology,
managed to find three bikes from which to collect information people living in remote areas, where electricity is not
from the Akij Eagle, the Avon E-lite and the DP Durbar accessible from the grid, can use this electric bike without
Electric Bicycle. The bikes were judged on five parameters: facing any battery charging issue. However, the
size, weight, distance, speed, and time to charge. The table Electromechanical Bike is doing enough to pave the way for
below indicates that the e-bike would be much more cost- Bangladeshi transportation to a better, greener tomorrow.
effective than the current bikes on the market, and is not only
lighter but also has a higher distance travel per price. It also REFERENCES
sits in the speed department comparably and it also takes a [1] V. Dimitrov, "Overview of the Ways to Design an Electric Bicycle,"
reasonable amount of time to charge. Having set out with the 2018 IX National Conference with International Participation
goal of creating a cost-effective alternative to Bangladesh's (ELECTRONICA), Sofia, 2018, pp. 1-4.
current modes of transport, this counts as a major step towards [2] Rashid, M., Al, J., Rahman, A., Siddique, A. R. M., & Azad, A. K. M.
that goal. (2017). A novel approach of improving battery performance and
longevity of the developed electrically assisted triwheeler vehicle by
implementing torque sensor technology. International Journal of
Specific- Our Akij Avon E- DP Vehicular Technology, 2017.
ations Proto- Eagle Bike Durbar [3] F. R. Khan, A. T. Aurony, A. Rahman, F. Munira, N. Halim and A.
type Electric Azad, "Torque sensor based electrically assisted hybrid rickshaw-van
with PV assistance and solar battery charging station," 2015
Bike International Conference on Advances in Electrical Engineering
(ICAEE), Dhaka, 2015, pp. 284-287.
Price BDT BDT BDT BDT
32,520 49,500 57,000 35,000 [4] Azad, A. (2016). Electrically assisted torque sensor-based wheelchair
with dedicated solar charge kit. IEEE Transportation Electrification
Weight 40Kg 70Kg 56Kg - Newsletter.
Distance 70 45-50 50 50 [5] J. Lee, J. Jiang and Y. Sun, "Design and simulation of control systems
Km/ Km/ Km/ Km/ for electric-assist bikes," 2016 IEEE 11th Conference on Industrial
Electronics and Applications (ICIEA), Hefei, 2016, pp. 1736-1740.
charge charge charge charge
[6] R. Tarek, A. Anjum, M. A. Hoque and A. Azad, "Solar electric
Speed 25.6 38 24 50 ambulance van unfolding medical emergencies of rural Bangladesh,"
Km/H Km/H Km/H Km/H 2016 IEEE Global Humanitarian Technology Conference (GHTC),
Seattle, WA, 2016, pp. 514-519.
Time to 5-6 4-8 6-8 3-6 [7] S. J. Chowdhury, R. Rahman and A. Azad, "Power conversion for
Charge hours hours hours hours environment friendly electrically assisted rickshaw using photovoltaic
Battery 24V 60V 48V 48V technology in Bangladesh," 2015 IEEE Transportation Electrification
Conference and Expo (ITEC), Dearborn, MI, 2015, pp. 1-6.
Capacity 30Ah 12Ah 12Ah 12Ah
[8] M. A. Hoque, P. Chakraborty, R. M. Huq, A. N. Bobby, N. T. Shuvo
and A. K. M. A. M. Azad, "Torque sensor based electrically assisted
hybrid tricycle: From design to development," 2017 IEEE Region 10
Humanitarian Technology Conference (R10-HTC), Dhaka, 2017, pp.
47-50.
Fig. 11. Table showing Comparative Study on e-bikes
[9] Chang, S. B., Chen, P. C., Chuang, H. S., & Hsiao, C. C. (2012).
VII. FUTURE IMPROVEMENT Velocity control with disturbance observer for pedal- assisted electric
bikes. Vehicle system dynamics, 50(11), 1631- 1651.
Our system uses human pedaling force to obtain real-time [10] Cheon, D. S., & Nam, K. H. (2017). Pedaling torque sensor- less power
information such as force or pace from the torque sensor. But, assist control of an electric bike via model-based impedance control.
the torque sensor is an expensive component. Since it is International journal of automotive technology, 18(2), 327-333.
directly attached to the pedal in a bike system, the risks of
damage are high. For these purposes, sensor-less control
methods based on a disturbance observer has so far been
researched [9]. Using this research, we want to introduce a
sensor-less power assist system for pedaling torque and
applied it to the electric bike in the future [10]. Besides, we
have considered replacing the Lead Acid batteries with Solar

721

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 15,2021 at 17:28:23 UTC from IEEE Xplore. Restrictions apply.
2018 IEEE Conference on Control Technology and Applications (CCTA)
Copenhagen, Denmark, August 21-24, 2018

An Anti-lock Braking System for bicycles

Matteo Corno1 , Luca D’Avico1 and Sergio M. Savaresi1

Abstract— This work presents an anti-lock braking system electronic devices for safety improvements so far. From a
(ABS) for bicycles equipped with hydraulic brakes. The systems control standpoint, there exist important differences between
employs an electrostatic hydraulic actuator capable of modu- powered vehicles (cars and motorcycles) and bicycles: first
lating the braking pressure when a loss of adherence is detected
while behaving passively during nominal conditions. This paper of all bicycle’s mass and inertias are smaller and, hence,
describes the control architecture and proposes three wheel faster dynamics come into play creating a more challenging
deceleration controllers: Bang-Bang (BB), Second Order Sliding control problem; second, the high centre of gravity of the
Mode (SOSM) and Proportional Integral (PI). The features and bicycle-driver-system and the moderate wheelbase increment
tuning procedure of each one are discussed and experimentally the criticality of the role played by the load transfers. From a
validated. Appropriate cost functions are adopted to quantita-
tively compare the presented ABS control logics on different structural viewpoint, constraints related to weights and vol-
road conditions proving the greater robustness of the PI in umes are more severe on bicycles than on cars/motorcycles.
terms of both performances and rider’s comfort level. The scientific community has already been devoted to
electric bicycles in a urban context [8] but most of the
I. INTRODUCTION literature focuses on solving sensing issues [9], [10], energy
In most of the metropolitan areas in Europe, bicycles and assistance management [11], [12]. To the best of authors’
are currently a growing market and, thanks to the cost knowledge, only three published works deal with braking
reduction of electronic components, electric bicycles are control for bicycles. In [13] a bicycle ABS prototype based
rapidly spreading as well. This choice is in fact plentiful of upon three different pressure phases increase-hold-decrease
advantages since it is healthy, eco-friendly, cost saving and has been developed; experimental results of a panic brake
traffic load reducing. Unfortunately, this trend has not been with the rear wheel on dry asphalt are reported. In [14] a
combined with an improvement of driving safety for cyclists. method to simulate both wheel lockup and nose-over motion
Bicycles belong to the category of the Vulnerable Road Users preserving the integrity of the human test driver in road tests
(VRU) and they have indeed the highest crash rate compared is described. In [1] a human-in-the-loop haptic based braking
to other road users [1], [2]. In terms of safety, the braking assistance system is presented: the rider is informed on the
manoeuvre is one of the most critical that is undertaken by proximity to the threshold safe deceleration by means of a
a two-wheeled vehicle and the major concern is on the front brake lever vibration; experimental tests are shown.
wheel lockup and the nose-over motion (stoppie) that are the This paper presents an ABS system for bicycles equipped
two main accident situations [2]. In the automotive context, with hydraulic brakes; the actuation and the control logic’s
the anti-lock braking system (ABS) deals with safety issues rationale behind the proposed control system are a novelty
during the braking manoeuvre. with respect to both the automotive field and the state of
ABS was first developed in the aeronautic field and in the art for bicycles. In this work, a continuous front wheel
the 1950s started to appear in the automotive industry. deceleration control is proposed and this goal is achieved
However, due to the cost reduction of micro-controllers, the through the position control of an electro-hydraulic actuator
commercial success of the ABS for road vehicles began only whose role is twofold: first, to hydraulically decouple the
in the 1970s [3], [4]. Nowadays, the ABS technology is braking circuit isolating the brake disk from the brake lever;
consolidated for both cars and motorcycles (mandatory in second, to modulate the available volume on the brake
Europe since 2004 and 2016 respectively). In the automo- disk side to regulate the brake disk’s pressure and, thus,
tive field, the state of the art of the ABS technology, is the front wheel deceleration. To achieve this objective, a
represented by two main categories: switching control logics classical cascade control architecture is adopted: an external
based upon wheel deceleration [3], [5] and wheel slip control wheel deceleration loop generates the reference position fed
[6], [7]. to an inner control loop to regulate the position of the
Bicycles have not been involved in this process yet because electro-hydraulic actuator. When the ABS is not active, the
of several reasons. From the economic point of view, the braking circuit behaves as a standard one. In this paper,
cost constraints are more relevant on bicycles compared three different wheel deceleration controllers are presented,
to cars/motorcycles. From a cultural perspective, bicycles experimental results on different road conditions are reported
have been seen as recreational vehicles with no need of and discussed as well.
This paper is structured as follows: in Section II, the
*This work was supported by blubrake http://blubrake.it/ system is presented and thoroughly described; in Section III,
1 M. Corno, L. D’Avico, and S. Savaresi are with Dipartimento
di Elettronica Informazione e Bioingegneria at Politecnico di Milano the control-oriented modelling is illustrated; in Section IV,
[email protected] the control design of three wheel deceleration controllers

978-1-5386-7698-1/18/$31.00 ©2018 IEEE 834

is tackled and experimental results are reported as well.
Finally, in Section V, the achieved control performances are FRONT
WHEEL
quantitatively analysed by means of suitable cost functions.1

II. SYSTEM DESCRIPTION

actuator BRAKE
The presented work has been developed on a mountain position DISK
bike with hydraulic brakes (see Figure 1) and the experi- BRAKE
LEVER
mental set-up is equipped as follows:
• Front and rear wheel encoders with 30 teeth (the pre-
sented ABS control logic relies on the measure of the
front one, while the rear one is only adopted to compare
different tests).
• A battery pack to provide power supply. Fig. 2. A schematic representation of the ABS system
• An Electronic Control Unit is devoted to process the
measured signals and to the control algorithms.
• An Electronic Braking Actuator (from now on EBA)

braking pad pressure [bar]

which is a modified Electrohydrostatic Actuator; it
weights 270g.
• An Inertial Measurement Unit (IMU) is available but
only for performance analysis (no information from this
DEAD- OPERATIVE
sensor is adopted for control purposes). ZONE ZONE

The EBA is connected to both sides of the braking circuit

(Figure 2) and it works as follows: when the system is not actuator position [mm]
active, the braking circuit behaves as a standard one: the
brake pressure is decided by the cyclist’s action on the brake Fig. 3. EBA: static map pressure-position
lever only. Instead, when the system is active, the actuator’s
piston is moved towards the brake lever and the two sides
of the braking circuit are separated: no fluid exchange can III. SYSTEM MODELING
take place. The quasi-static relationship between the position In this section the control-oriented modelling of the bicy-
and brake pad pressure is shown in Figure 3 (for confiden- cle dynamics of interest is addressed. As already explained
tiality reasons, the axes scales are omitted). Starting from in Section I, two categories of ABS algorithms can be
different initial pressures, the braking pad pressure is reduced distinguished: those based upon the wheel deceleration and
increasing the actuator position. The piston’s position is the the ones controlling the wheel slip. The envisioned control
control variable adopted in this work: it determines whether system has access to the front wheel velocity and actuator
the braking circuit is decoupled and the pressure reduction position measurements only. This prevents the development
on the brake disk contrasting the cyclist’s action on the brake of a wheel slip estimation algorithm. A wheel deceleration
lever. The actuator and its low level control is an evolution control is the only option. The wheel deceleration η is com-
of the work presented in [15], [16]. puted, through numerical differentiation, from the velocity of
the front wheel. The well known problems of wheel speed
estimation from encoder measurements has been solved
using the algorithm in [1], [17], [18] which reconstructs the
wheel velocity from discrete measurements also filtering the
periodic noise.
From a control-oriented modelling perspective, the control
variable is the actuator reference position xre f and the
controlled variable is the wheel deceleration η. Note that,
considering xre f as the input, the closed-loop dynamics of
the actuator position control are included in the modelling
Fig. 1. Bicycle equipped with the ABS system procedure presented here. Modelling the dynamics of interest
is a challenging task [19], [20]. The wheel dynamics between
The system controls the front wheel: from the stability the braking torque and the wheel deceleration are non-
point of view, a wheel lock event is more critical when the linear, depending on both road friction conditions and vehicle
front wheel is involved than the rear one. In any case, if speed; furthermore, another non linearity contribution comes
desired, this work can be applied to the rear wheel as well. from the relationship between the braking torque and the
corresponding braking pressure controlled by the actuator
1A patent protecting parts of this system have been filed. position due to the friction coefficient of the braking pad.

835
 4 them aims at exploiting different features of the braking
xref dynamics; the experimental validation is reported for each
2 one of them. Due to space constraints, tests performed on
[m/s2 ]

[mm]
0
high grip offroad (clay ground) only are presented here; in
Section V, suitable cost functions will show the achieved
Exp Data
-2
Identified
control results in all the tested road conditions.
0.5 1
time [s] First of all, the well-known Bang-Bang (BB) controller is
considered: it allows to deal with the presence of the non lin-
Fig. 4. Identification of the transfer function Gxre f η
earity discussed in Section III in a simple but quite effective
way; however, it does not take into account the limitation of
the control variable. When used close to its limits, the control
The main idea of the model is thus to capture the dominant
variable is rate-limited. This leads to the development of a
dynamics of the system and exploit this information in the
Second Order Sliding Mode controller which could better
control system’s design that will be addressed in Section
account for this limitation. Finally, a Proportional-Integral
IV. Future work will be devoted to a more accurate system
modulating control law (PI) is implemented to exploit the
identification that may improve control performances.
appealing feature of a modulating control law whose main
For the identification of the transfer function from xre f
drawback is a more delicate tuning phase of the control
to η, a step response test has been performed as follows:
parameters; for this purpose, the information derived from
the braking manoeuvre is started at 25km/h until a wheel
the identification procedure of the dominant system dynamics
deceleration threshold (3m/s2 ) is trespassed; the position
(see Section III) will be leveraged.
control loop is activated and a reference position step is
An appropriate test protocol is adopted for the evaluation
then imposed. Figure 4 shows the results of several tests
of the control performances:
performed with this very same test protocol. The collected
data are employed to identify Gxre f η with a gray-box iden- • the rider is asked to brake with only the front brake; in

tification approach. For this work’s purposes, a first order this way, the rear wheel velocity can be employed to
dynamics has been considerate accurate enough: assess the actual vehicle velocity;
• every braking manoeuvre begins as a panic brake (the
µ
Gxre f η (s) = , (1) rider brakes as hard as possible);
1 + s/(2π fP ) • a safe threshold η̄ has been chosen equal to 2.5m/s2
where µ and fP are the outputs of the identification as a suitable wheel deceleration condition to keep the
procedure and it is obtained µ = −0.50[m/s2 /mm] and cyclist secure from wheel lockup and nose-over motion;
fP = 0.87[Hz]. • when the condition η > η̄ = 2.5m/s2 becomes true,
the wheel deceleration controller is activated and it is
imposed η re f = η̄; adaptability of the reference value
IV. CONTROL DESIGN is subject of current investigation;
The control architecture is composed by two nested • the test is over and the ABS is turned off when the rear
loops: the inner one is devoted to the control of the speed drops below 7km/h.
actuator position while the external one controls the
wheel deceleration. Each of them has been designed and A. Wheel deceleration control: BB
implemented for this specific application. The inner loop is
an evolution of the one described in [15], [16] and reaches
a closed loop bandwidth of 25Hz.

Fig. 6. Deceleration control: BB block scheme

The main attractiveness of the Bang-Bang controller ( [21],

[22]) is its simplicity and its model-free nature. Figure 6
Fig. 5. ABS block scheme shows the adopted control scheme; the adopted controller
parameters are ēup = ēdown = 0[m/s2 ]. Note that this choice
Three wheel deceleration control approaches are of the switching thresholds means that there is no hysteresis
implemented: a Bang-Bang controller, a Second Order on the deceleration error and, thus, the control action’s delay
Sliding Mode controller and a PI controller. Each one of is minimized.

836
 Figure 7 plots the experimental results: the main goal is to rewrite equation (1) in state-space form and it becomes:
achieved since wheel lockups are avoided but quite evident
η̇ = Aη + Bxre f (2)
oscillations are present on both wheel speeds and deceler-
ation η. From the bottom plot, a significant tracking error where A = −2π fˆp and B = µ2π fˆp . The error between the
on the actuator position is also evident due to the saturation current wheel deceleration and the desired one is chosen as
limits that affect the position control loop. This aspect cannot the sliding variable, i.e.,
be taken into account with such a simple control structure: to
include this information in the control design phase, a more s = η − η re f . (3)
refined control law must be considered as described in IV-B. The first and second derivative of the sliding variable s are:
(
BB | OFFROAD (HIGH grip) ṡ = η̇ − η̇ re f
30 (4)
s̈ = η̈ − η̈ re f = Aη̇ + Bẋre f − η̈ re f .
20
In order to design an SOSM, y1 = s and y2 = ṡ are defined,
so that system (4) can be rewritten as
10
(
ẏ1 = y2
0 (5)
170.8 171 171.2 171.4 171.6 171.8 172 172.2 172.4 172.6 ẏ2 = ϕ + hṪ
10

where y1 and y2 are the auxiliary variables, Ṫ the auxiliary

5 control input and the other variables are defined as follows

0
ϕ = Aη̇ − η̈ re f
170.8 171 171.2 171.4 171.6 171.8 172 172.2 172.4 172.6
h = −B (6)
re f
Ṫ = −ẋ .
The studied control problem can be then reformulated by
stating that a suitable auxiliary control signal Ṫ must be
designed to steer y1 and y2 to zero in finite time given the
system (5) where y2 is unavailable for measurements and ϕ
170.8 171 171.2 171.4 171.6 171.8 172 172.2 172.4 172.6
and h satisfy the following conditions:

Fig. 7. Wheel deceleration control: BB, offroad (High Grip)

|ϕ| ≤ Φ (7)
0 < Γ1 ≤ h ≤ Γ2 (8)
From a physical point of view, constraint (7) means that,
B. Wheel deceleration control: SOSM given a constant actuator position reference xre f , the wheel
deceleration η is bounded. For what concerns the constraint
(8), it represents that (1) is defined by finite values of static
gain and dominant pole frequency.
Thus, the SOSM here described is of suboptimal type [21].
Under the hypothesis that the sM value can be detected, it
is possible to prove the convergence of the system trajectory
to the sliding manifold s = ṡ = 0 in finite time given the
following control law:
Fig. 8. Deceleration control: SOSM block scheme  
sM
Ṫ = −ẋre f = −α V sign s − (9)
2
The SOSM has three advantages: first, it provides a better
way to impose the rate of the control variable; second,
α∗ i f s − s2M  sM > 0
  
it has been proven to guarantee robustness [23]; third, it α = (10)
1 i f s − s2M sM ≤ 0
provides a continuous control action. This controller has
already been proved to guarantee robustness properties even where V is the control gain, α is the modulation factor and
in the presence of uncertainties in wheel control application sM = s(t) is the most recent value such that ṡ(t) = 0. This
[24]. In this work, the modelling presented in Section III statement holds if the following constraints are satisfied:
is leveraged to find a first tentative tuning of the controller  
3Γ1
parameters. α ∗ ∈ (0, 1] ∩ (11)
Γ2
The design of this controller is tackled considering the  
identified transfer function from the actuator reference posi- Φ 4Φ
V > max , (12)
tion xre f to the measured wheel deceleration η. It is possible α ∗ Γ1 3Γ1 − α ∗ Γ2

837
 The adopted tuning parameters of the SOSM controller A Proportional-Integral (PI) controller is now examined
are: α ∗ = 0.615[−],V = 65.5[mm/s]. The choice of these to achieve better results than the one obtained with the
parameters is defined by an admissible range of values given controllers presented above. Starting from the information
by (11), (12); the numerical values of Γ1 , Γ2 and Φ are de- withdrawn from the identified model of the wheel decel-
fined by (7), (8) whose definition is based upon the identified eration dynamics in Section III, a first tentative tuning of
dynamics discussed in Section III. The tuning parameters the PI controller can be defined; a refinement of the closed-
α and V define the rate of the control variable (see (9), loop performances is then performed experimentally with
(10)). Figure 9 shows the experimental validation of this a trial-and-error procedure. The saturation limits of the
controller and it is clear that the choice of these parameters control variable are taken into account by imposing suitable
is correct: a higher rate of the control variable would have bounds to the tuning parameters KP and TI . In Figure 10 the
been impossible to be tracked by the position control loop structure of the implemented PI controller with anti-windup
because of the saturation limits already discussed in Section compensation is shown.
IV-A; a lower rate of the control variable would not have
PI | OFFROAD (HIGH grip)
exploited all the available potential of the actuation system. 30

SOSM | OFFROAD (HIGH grip) 20

30

10
20

0
10 86 86.5 87 87.5 88
10

0
76.5 77 77.5 78 78.5
5
10

5 0

86 86.5 87 87.5 88

76.5 77 77.5 78 78.5

85.5 86 86.5 87 87.5 88 88.5

76.5 77 77.5 78 78.5 Fig. 11. Wheel deceleration control: PI, offroad (High Grip)

Fig. 9. Wheel deceleration control: SOSM, offroad (High Grip) Figure 11 shows the experimental results of the PI con-
troller: after the initial transient where the incipient wheel
This controller is able to control the average wheel de- lockup is promptly avoided, the wheel deceleration is re-
celeration around η re f . However, despite of the fact that duced and kept close to the safe reference value η re f up
the information on the saturation limits is now included to the deactivation of the ABS; the actuator position is
in the control design, significant oscillations of the control controlled within its saturation limits.
variable η are still present indicating the inadequacy of this V. COMPARISON AND DISCUSSION
controller to create a narrow limit cycle around the reference
The performances of the presented controllers are now
value. More accurate results are pursued by introduction of
quantitatively evaluated and compared following the same
a modulating controller as presented in Section IV-C.
test protocol presented in Section IV. Different road condi-
tions are considered: asphalt, high grip offroad (clay ground)
C. Wheel deceleration control: PI and low grip offroad (mixture of gravel and sand). The
adopted cost functions are:
v
u t of f
u
2 1 m
J per f ormance =t ∑ |η(t) − η re f |2
to f f − ton t=t
[
s3/2
] (13)
on
v
u to f f
1 u
Jcom f ort = t2
∑ |Ȧx |2 [m/s4 ] (14)
to f f − ton t=t on
Fig. 10. Deceleration control: PI block scheme

838
 where ton and to f f are the time instants when the ABS is [3] P. Wellstead and N. Pettit, “Analysis and redesign of an antilock brake
turned on and off respectively. In particular, (13) indicates the system controller,” IEE Proceedings-Control Theory and Applications,
vol. 144, no. 5, pp. 413–426, 1997.
wheel deceleration tracking performance; (14) quantifies the [4] H. Leiber and A. Czinczel, “Electronic control unit for passenger
rider’s comfort level by taking into account the amplitude of car antiskid,” in Vehicular Technology Conference, 1979. 29th IEEE,
the vehicle’s longitudinal oscillations. The optimal behaviour vol. 29. IEEE, 1979, pp. 65–69.
[5] I. Ait-Hammouda and W. Pasillas-Lépine, “Jumps and synchronization
is reached when both cost functions are close to zero. in anti-lock brake algorithms,” in AVEC 2008, 2008, pp. N–A.
Figure 12 shows the cost functions in all the tested road [6] I. Petersen, “Wheel slip control in abs brakes using gain scheduled
conditions; all the tests are performed with the same tuning optimal control with constraints,” Department of Engineering Cyber-
netics. Norwegian University of Science and Technology, Trondheim,
of the controller parameters presented in Section IV; each Norway, 2003.
one of the marker indicates a single braking manoeuvre. [7] J. Yu, “A robust adaptive wheel-slip controller for antilock brake
It is evident that, among the considered controllers, the PI system,” in Decision and Control, 1997., Proceedings of the 36th IEEE
Conference on, vol. 3. IEEE, 1997, pp. 2545–2546.
achieves the best results in terms of both cost functions: [8] G. Alli, S. Formentin, and S. M. Savaresi, “On the suitability of epacs
it is able not only to produce the more accurate tracking in urban use,” IFAC Proceedings Volumes, vol. 43, no. 18, pp. 277–
(lowest J per f ormance ) but also to contain the vehicle’s body 284, 2010.
[9] M. Corno, P. Spagnol, and S. M. Savaresi, “Road slope estimation in
oscillations produced by the control action reducing the bicycles without torque measurements,” IFAC Proceedings Volumes,
discomfort for the rider (lowest Jcom f ort ). In conclusion, vol. 47, no. 3, pp. 6295–6300, 2014.
both BB and SOSM produce uncomfortable oscillations and [10] G. Rallo, S. Formentin, M. Corno, and S. M. Savaresi, “Real-time ped-
aling rate estimation via wheel speed filtering,” IFAC-PapersOnLine,
achieve poor tracking results, while PI is able to provide a vol. 50, no. 1, pp. 6010–6015, 2017.
smoother transient with reduced chassis oscillations. [11] M. Corno, F. Roselli, and S. M. Savaresi, “Bilateral control of senzaa
series hybrid electric bicycle,” IEEE Transactions on Control Systems
ASPHALT OFFROAD OFFROAD
Technology, vol. 25, no. 3, pp. 864–874, 2017.
3000 3000 3000
(HIGH grip) (LOW grip) [12] D. Meyer, M. Steffan, and V. Senner, “Impact of electrical assistance
on physiological parameters during cycling,” Procedia Engineering,
2500 2500 2500 vol. 72, pp. 150–155, 2014.
[13] K. Enisz, I. Szalay, D. Fodor, K. Nagy, and R. Jakab, “Bicycle
2000 2000 2000 anti-lock braking system prototype development,” Acta Universitatis
Sapientiae Electr. Mech. Eng., vol. 4, pp. 45–57, 2012.
1500 1500 1500 [14] O. Maier, M. Pfeiffer, and J. Wrede, “Development of a braking
dynamics assistance system for electric bicycles: Design, implemen-
1000 1000 1000 tation, and evaluation of road tests,” IEEE/ASME Transactions on
Mechatronics, vol. 21, no. 3, pp. 1671–1679, 2016.
500 500 500 [15] F. Todeschini, S. Formentin, G. Panzani, M. Corno, S. M. Savaresi, and
0 2 4 6 0 2 4 6 0 2 4 6
L. Zaccarian, “Nonlinear pressure control for bbw systems via dead-
zone and antiwindup compensation,” IEEE Transactions on Control
Systems Technology, vol. 24, no. 4, pp. 1419–1431, 2016.
Fig. 12. Wheel deceleration control: cost functions [16] F. Todeschini, M. Corno, G. Panzani, S. Fiorenti, and S. M. Savaresi,
“Adaptive cascade control of a brake-by-wire actuator for sport mo-
torcycles,” IEEE/ASME Transactions on Mechatronics, vol. 20, no. 3,
pp. 1310–1319, 2015.
VI. CONCLUSIONS [17] N. Persson, “Event based sampling with application to spectral esti-
mation,” Ph.D. dissertation, Linkopings universitet, 2002.
In this paper an anti-lock braking system for bicycles [18] G. Rallo, S. Formentin, and S. M. Savaresi, “Online wheel speed
filtering for periodic disturbance reduction: a strategy for an advanced
equipped with hydraulic brakes is presented. The proposed bicycle application,” arXiv preprint arXiv:1610.01064, 2016.
control system aims to track a safe front wheel deceleration [19] S. M. Savaresi and M. Tanelli, Active braking control systems design
value in order to avoid critical situations such as wheel for vehicles. Springer Science & Business Media, 2010.
[20] S. M. Savaresi, M. Tanelli, and C. Cantoni, “Mixed slip-deceleration
lockup and nose-over (stoppie). Three different wheel decel- control in automotive braking systems,” Journal of dynamic systems,
eration controllers have been tested using an instrumented bi- measurement, and control, vol. 129, no. 1, pp. 20–31, 2007.
cycle on asphalt and offroad (high and low grip): Bang-Bang [21] G. Bartolini, A. Ferrara, A. Levant, and E. Usai, “On second order
sliding mode controllers,” in Variable structure systems, sliding mode
(BB), Second Order Sliding Mode (SOSM) and Proportional- and nonlinear control. Springer, 1999, pp. 329–350.
Integral (PI). Among the presented controllers, from the [22] G. Knowles, “Remarks on bang-bang control in a hilbert space,”
comparison of the experimental results, the PI controller is Journal of Optimization Theory and Applications, vol. 21, no. 1, pp.
51–57, 1977.
the most accurate in terms of wheel deceleration tracking [23] A. Levant, “Higher-order sliding modes, differentiation and output-
and it also guarantees the better comfort for the cyclist in all feedback control,” International journal of Control, vol. 76, no. 9-10,
the tested road conditions. pp. 924–941, 2003.
[24] M. Tanelli, C. Vecchio, M. Corno, A. Ferrara, and S. M. Savaresi,
“Traction control for ride-by-wire sport motorcycles: A second-order
R EFERENCES sliding mode approach,” IEEE Transactions on Industrial Electronics,
vol. 56, no. 9, pp. 3347–3356, 2009.
[1] M. Corno, L. D’Avico, G. Panzani, and S. M. Savaresi, “A haptic-
based, safety-oriented, braking assistance system for road bicycles,”
in Intelligent Vehicles Symposium (IV), 2017 IEEE. IEEE, 2017, pp.
1189–1194.
[2] O. Maier, M. Pfeiffer, C. Wehner, and J. Wrede, “Empirical survey on
bicycle accidents to estimate the potential benefits of braking dynamics
assistance systems,” in 4th Int. Cycling Safety Conf., Hannover,
Germany, 2015.

839