Hybrid Energy Storage Systems and Battery Management

for Electric Vehicles

Sangyoung Park, Younghyun Kim and Naehyuck Chang

Seoul National University, Korea

{sypark, yhkim, naehyuck}@elpl.snu.ac.kr
ABSTRACT
Electric vehicles (EV) are considered as a strong alternative of in-
ternal combustion engine vehicles expecting lower carbon emis-
sion. However, their actual benets are not yet clearly veried
while the energy efciency can be improved in many ways. The
carbon emission benets from EV is largely diminished if we
charge EV with electricity from petroleum power plants due to
power loss during generation, transmission, conversion and charg-
ing. On the other hand, regenerative braking is direct power con-
version from the wheel to battery and one of the most important
processes that can enhance energy efciency of EV. Power loss dur-
ing regenerative braking can be reduced by hybrid energy storage
system (HESS) such that supercapacitors accept high power as bat-
teries have small rate capability.
Conventional charge management does not systematically ex-
change charge between the supercapacitor and battery. However,
asymmetry in acceleration and deceleration as well as battery charg-
ing and discharging capability make the supercapacitor state of
charge (SoC) management override the efciency optimization.
Unlike previous works, we show how charge migration during idle
and cruise/stopping time can be benecial in terms of energy ef-
ciency and cruise range. Systematic charge migration decouples
SoC management and charging efciency optimization giving a
higher degree of freedom to charging efciency optimization. We
demonstrate the proposed charge migration between the superca-
pacitor and battery improves energy efciency by 19.4%.
Categories and Subject Descriptors
C.3 [Special-Purpose and Application-Based Systems]: Real-
time and embedded systems

This work is supported in part by the Mid-Career Researcher Pro-

gram through NRF grant funded by the MEST (No. 20120005640).
The ICT and ISRC at Seoul National University provides research
facilities for this study.

Corresponding author.
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for prot or commercial advantage and that copies
bear this notice and the full citation on the rst page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior specic
permission and/or a fee.
DAC 13, May 29 - June 07, 2013, Austin, TX, USA.
Copyright 2013 ACM 978-1-4503-2071-9/13/05 ...$15.00.
General Terms
Algorithm, Design, Performance
Keywords
Electric vehicle, Battery-supercapacitor hybrid, Regenerative brak-
ing, Charging/discharging asymmetry
1. INTRODUCTION
Electric vehicles (EV) are rapidly gaining popularity as demand
for cleaner means of transportation increases. Most countries ac-
tively promote deployment of EV. Governments offer subsidies and
tax credits to EV manufacturers and customers to give a boost to
EV market. For example, the US Government provides federal tax
credits to EV consumers according to battery capacity of the ve-
hicle such that Chevrolet Volt and Tesla vehicles are eligible for
one-time $7,500 tax credit.
However, the actual benet from EV is not yet clearly conrmed
in terms of the entire life cycle of EV and electricity generation
for EV charging. A recent analysis points out that the maximum
annual prot considering the real power grid electricity price and
battery degradation is only $10 to $120 per EV, which is denitely
not sufcient to attract the customers due to the higher EV price
comparing with internal combustion engine vehicle [1]. Average
power plant boiler and turbine efciency is at around 33%, power
transmission efciency is at around 93%, battery charger shows
at around 70% efciency, and the battery charging efciency is at
around 90% [2, 3]. The overall efciency is at around 23%, which
is not meaningfully higher than the efciency of internal combus-
tion engines, which is known as 20% or higher.
Braking energy is often above 30% of traction energy and goes
up to 80% in heavy city trafc [4]. Regenerative braking is effec-
tive battery recharging with the energy directly coming from the
wheels to the battery unlike the long lossy energy trip from the
power plants for plugin recharging. Thus, efcient harvesting of re-
generative braking energy is the key to maximize the annual prot
of EV ownership. A brief calculation shows that increasing the re-
generative braking efciency by 10% is equivalent to 3% to 8% of
improvement in the gas mileage.
This paper focuses on three important factors in regenerative
braking. The rst one is limited rate capability of the battery [5].
Batteries in EV are subject to peak current, which is often 10 times
higher than the average current. FreedomCAR and Vehicle Tech-
nologies (FCVT) have dened requirements of an energy storage
system (ESS) for EV and HEV (hybrid EV) such as total energy,
power capacity, total life time, and so on [6]. The second factor is
the maximum power transfer during the regenerative braking. The
traction motor is a non-ideal power source with appreciable internal
impedance that requires the maximum power transfer control.
This paper aims at a higher energy efciency during a sequence
of driving cycles consisting of braking, stopping/cruising, and
acceleration. We enhance the regenerative braking energy har-
vesting efciency applying the maximum power transfer tracking
(MPTT) [7, 8]. The previous regenerative braking architecture
commonly connect the battery bank directly to the DC bus. This
makes the traction motor voltage output clamped to the battery ter-
minal voltage, which does not allow the maximum power point
tracking (MPPT) or MPTT
1
. We introduce a regenerative braking
architecture that allows arbitrary voltage control on the DC bus for
MPTT.
We perform the near optimal charge replacement [9] for accel-
eration and compensate the limited rate capability of the battery.
The arbitrary voltage control capability on the DC bus is also ben-
ecial for the optimal acceleration. Most of all, we decouple the
acceleration and deceleration optimization, which was coupled by
the state of charge (SoC) management of the supercapacitor in the
previous works. We apply charge migration [10] from the superca-
pacitor to battery while stopping or cruising, and the charge migra-
tion manages the supercapacitor SoC without sacricing the energy
loss during acceleration and regenerative braking. As for the third
factor, we specially focus on the battery asymmetry in charging and
discharging [11] and tailer the charge migration.
The proposed method more efciently enhances effective gas
mileage of EV with accurate prediction of the vehicle route and
trafc conditions. Manual driving nowadays is largely relied on a
GPS navigator, a semi-autonomous driving such as adaptive cruise
control utilizes the trafc condition via onboard radars, and au-
tonomous driving even more utilizes computerized trafc informa-
tion. Therefore, it is not surprising to have trafc and driving infor-
mation and predict the near future driving patterns available.
The proposed method allows for inherent asymmetry in ESS
charging/discharging behaviors during acceleration and decelera-
tion of the EV. We make use of the idle and cruise periods of the
EV to perform proactive charge migration. Our results show that
the proposed technique increases energy efciency by 19.4%.
2. BACKGROUND
2.1 Hybrid Energy Storage System for EV
Hybrid ESS (HESS) is an energy storage composed of multi-
ple heterogeneous energy storage devices. The key components
are ESS banks, a charge transfer interconnect (CTI), converters
for the power sources and load devices, and a microprocessor-
based charge management policy controller. An ESS bank typi-
cally consists of homogenous ESS elements organized into a two-
dimensional array to meet the power/energy capacity and the volt-
age rating. There is also a bidirectional charger (or two unidirec-
tional chargers connected in opposite directions) that controls the
charge and discharge current of the ESS array (or equivalently, the
current owing into and out of the ESS bank.) Because the SoC,
terminal output voltage, and power rating of different ESS arrays
may not be compatible with each other, direct connection among
ESS arrays is generally not feasible. The primary function of CTI
is to provide charge transfer paths among storage banks, power
sources and load devices. There are various ways to implement
CTI including a single DC bus wire, a segmented DC bus, multiple
DC buses, or a more complicated interconnect network such as a
1
MPPT maximizes power output from the power source. MPTT
considers power converter efciency and maximizes the power go-
ing into the battery.
mesh network. Converters for the power sources and load devices
can be any of chargers, DC-DC converters, AC-DC rectiers, and
DC-AC inverters as appropriate. These components are not differ-
ent from the components in a homogeneous ESS system. Charge
management policy controller is a microprocessor-based controller
in charge of the CTI current owing from or to each ESS bank and
the CTI voltage according to the elaborated charge management
policies.
A combination of a battery and an energy storage element with
a higher power capacity can be a good complementary setup both
for efciency and cost in ESS for regenerative braking. Among
them, battery-supercapacitor HESS is considered a promising solu-
tion to mitigate rate capability problem of batteries while meeting
other ESS constraints. Adding appropriate amount of supercapaci-
tor could increase the overall energy efciency and thus the cruise
range. Despite its benets, supercapacitor is still expensive and
cause severe volumetric overhead in EV. Therefore, the key issue in
HESS is determination of the supercapacitor capacitance and SoC
management.
Previous works mostly focused on supercapacitor SoCso that su-
percapacitors are not fully charged during regenerative barking and
not fully depleted during acceleration. Unfortunately, optimization
of the HESS charging efciency is not accordance with the ideal su-
percapacitor SoC management [8]. In other words, previous works
perform SoC management of the supercapacitor without system-
atic migration charge between the battery and the supercapacitor.
Therefore, SoC management of the supercapacitor generally over-
ride the efciency optimization.
A heuristic approach can maintain the supercapacitor SoC in-
versely proportional to the vehicle speed [12], and use of machine
learning to avoid modeling complexity of the entire regenerative
braking process [13]. Such complicated problems are often solved
by intuitive methods based on expert rules. This type of heuristic
approach enhance load balancing among the fuel cell, battery, and
supercapacitor [14]. Cost of power converter is additional overhead
for the SoC management of supercapacitor, and sometimes, size of
the power converter has a higher priority over energy efciency in
the supercapacitor SoC management [15]. Such SoC maintenance
signicantly restricts the efcient charging and discharging during
acceleration and braking [8]. Charge migration gives a greater free-
dom in the SoC management of the supercapacitor. Charge migra-
tion in HESS have been intensively studied recently: denition and
optimization of HESS from the view point of computer-aided de-
sign and design automation [16], various optimization objectives
such as energy efciency, battery lifetime, peak power minimiza-
tion for charge allocation (charging HESS) [17], replacement (dis-
charging HESS) [9], and migration (internal transfer) [10], which
are named after computer memory management.
2.2 MPTT for Regenerative Braking
Regenerative braking is the key feature in EV to enhance en-
ergy efciency as it allows reuse of kinetic energy during braking.
It is widely adopted in commercial EV and HEV including Toy-
ota Prius, Honda Insight, Tesla Roadster, and Chevrolet Volt. Fig-
ure 1(a) shows a typical regenerative braking system. Regenerative
brakes in EV involve using an electric motor as an electric gener-
ator and stores energy in energy storage for later use. Not all of
the kinetic energy is recovered using regenerative braking due to
the following reasons. Regenerative brakes alone cannot make the
vehicle to a complete stop, and it only works on wheels with an
electric motor, whereas braking force is often required from other
wheels. Figure 1(b) shows a typical braking pattern. Regenerative
braking force plus hydraulic braking force should match the driver
Wheel
Wheel
Motor Driver ESS
Hydraulic
control
Wheel
Wheel
Pedal
sensor
B
r
a
k
i
n
g

f
o
r
c
e
Time
Hydraulic
braking force
Regenerative
braking force
(a) A regenerative braking system.
(b) A typical regenerative braking pattern.
Figure 1: Regenerative braking for EV.
demand from pedal sensor. In this paper, we do not control the
portion of regenerative braking force and hydraulic braking force,
and thus we assume there is a well-dened control algorithm that
maximizes regenerative braking energy while meeting the drivers
demands. Thus, hydraulic brakes should always be used together
with regenerative brakes, so some portion of kinetic energy will still
be dissipated as heat. However, the portion of regenerative braking
force to the hydraulic braking force is not very controllable be-
cause it should satisfy the drivers demands and match pressure on
the pedal sensor.
2.3 Battery Rate Capability
Although numerous types of devices such as supercapacitors
and ywheels are considered for ESS, batteries are the primary
energy storage in EV. ESS management algorithms should care-
fully count on the battery characteristics to achieve high energy
efciency. A dominant phenomenon that explicitly dictates bat-
tery charging/discharging efciency is the rate capability. The rate
capability is modeled in the Peukerts formula, which is an em-
pirical equation to evaluate the relationship between the charg-
ing/discharging efciency by the battery current.
C
p
= I
k
t, (1)
where C
p
is the battery capacity at a nominal discharge current in
Ampere-hour, I is the battery current relative to the nominal battery
current, and t is time in hours. Figure 2(a) illustrates the rate capa-
bility of a Li-ion battery by the discharging current ranging from1C
to 6C. A higher C-rating current signicantly decreases the amount
usable energy extracted from the battery [18]. We observe the same
phenomena for the charging operation.
Rate capability of the same battery is not the same for the char-
ing and discharging. We see distinct asymmetry in charging and
discharging as shown in Figure 2(b). Peukerts constant is 1.485
while for discharging it is 1.034 in the case of 1.9 Ah 18650 lithium
ion cells [11]. However, such asymmetry is not well investigated
and considered in the previous EV ESS management papers.
2.4 EV Energy Efciency
The energy efciency of the EV power train depends on ef-
ciency of its consisting components starting from the traction force
on the wheels to axle, motor, inverter, charger, and nally, to ESS
10 20 30 40 50 60 70
4
6
8
10
Discharge time (min)
B
a
t
t
e
r
y

v
o
l
t
a
g
e

(
V
)
76.3 min
10.7 kJ
33.2 min
10.0 kJ
1C
2C
4C
6C
14.6 min
8.2 kJ
5.3 min
4.3 kJ
(a) Rate capability [18].
(b) Peukert plot of a 1.9 Ah 18650 Li-ion cell [5].
Rate (h
-1
)
C
a
p
a
c
i
t
y

(
A
h
)
Charge
Discharge
0 1 2 3 4 5
0.5
1
1.5
2
k =1.034
k =1.485
Figure 2: Battery characteristics.
banks. Speed and traction force of the vehicle is converted to motor
torque and rpm by the following steps. Traction force is equal to
total running resistance which is described as,
R
tot
= R
R
+R
A
+R
G
+R
I
+R
B
, (2)
where R
R
, R
A
, R
G
, R
I
, and R
B
are rolling resistance, aerodynamic
resistance, gradient resistance, inertia resistance, and brake force
provided by hydraulic brakes, respectively. Simple models exists
for calculating the resistance values using vehicle mass, drag coef-
cients, drag area, vehicle speed, and so on [19].
R
R
=C
rr
W, (3)
R
A
=
1
2
C
d
Av
2
, (4)
R
G
=Wsin, (5)
R
I
= ma, (6)
where C
rr
, W, C
d
, A, v, , m, and a is the rolling resistance co-
efcient, vehicle weight, air density, drag coefcient, car frontal
area, vehicle speed, gradient angle, vehicle mass, and vehicle ac-
celeration, respectively. We calculate the motor torque and angular
velocity from the traction force, wheel size and axle ratio infor-
mation. Many EV have single speed constant ratio transmission,
which makes the calculation the easier.
R
tot
R
B
=
w
d
w
/2, (7)

m
=
w
/G, (8)

m
=
v
d
w
2 =
2v
d
w
G, (9)
where F
traction
,
w
,
m
, G,
m
, d
w
are traction force, wheel torque,
motor torque, axle ratio, motor angular velocity, and wheel diame-
ter.
The scope of our optimization is focused on electrical compo-
nents. A motor driver is typically insulated-gate bipolar transis-
tor (IGBT)-based inverters, and battery and supercapacitor charg-
ers are transistor based converters, which have their own efciency
map. Moreover, ESS is subject to the devices cycle efciency as
will be shown in the next section. We should consider all the losses
in the regenerative braking process and systematically minimize it
to achieve the MPTT [8]. The energy efciency of the regenerative
braking process depends on the motor generation efciency, motor
driver efciency, ESS charger efciency, and the cycle efciency of
the ESS components. Three-phase brushless DC motors (BLDC)
are generally used for EV. The electrical model of the BLDC motor
we used is given as,
V
k
= Ri
k
+(LM)
di
k
dt
+E
k
, (10)
E
k
= K
k

m
F(
e
+
k
), (11)
T
k
= K
t
i
k
F(
e
), (12)
where V
k
, E
k
, T
k
, i
k
are the voltage, back-EMF voltage, torque, and
current of k-th phase, R, L are the resistance, inductance of each
phase, M is the mutual inductance, K
t
and K
e
are the torque con-
stant and back-EMF constant,
m
and
e
are the angular speed and
angle of the rotor, F(
e
) is the back-EMF reference as function of
rotor,
k
is the phase difference between phases [20]. We calculate
the relationship between torque, angular speed, and motor input
voltage and current.
Conversion efciency of inverters and chargers in EV power
train is not constant. We describe them as a function of V
in
, V
out
,
I
out
, which are input voltage, output voltage and output current with
reasonable accuracy [21]:
= f (V
in
,V
out
, I
out
). (13)
3. CHARGE MANAGEMENT IN EV
Charge management such as charge allocation, replacement, and
migration requires careful determination of sources, destinations,
amount of current, CTI voltage, and so on, in order to maximize
the energy efciency, which is dened as

trans f er
=
Total energy transferred to the destination
Total energy extracted from the sources
, (14)
where the sources are power sources or discharging ESS banks, and
the destinations or load devices or charging ESS banks. According
to the observation from recent related works [9, 10, 17, 22], charg-
ing efciency is strongly dependent on the type of the bank, the
magnitudes of the charging currents, SoCs of the EES banks, volt-
age and current characteristics of the external power source, and
so on. Excessive mismatch between the input voltage level and the
EES bank terminal voltage results in unnecessarily large power loss
in the chargers. Severe mismatch between the input current and the
destination EES bank charging current results in a high IR loss and
rate capacity effect. The destination EES banks must be compatible
with the input power source in terms of the energy capacity as well.
The voltage on the CTI signicantly change the efciency of
the charges, which should be carefully determined by the input
source voltage and the destination bank voltage. The optimal charg-
ing/discharging current and the CTI voltage changes over time as
charge allocation progresses. We continue to monitor, calculate
the optimal setup and control the charge allocation process accord-
ingly [9, 10, 17, 22]. Elaborated charge management policies im-
proves efciency up to 30% comparing with conventional homoge-
neous ESS systems [9].
Motor
driver
IGBTs
Battery
bank
Supercap
bank
Charger

2
Charger

1
Charge
migration path
DC bus
Traction
motor
Figure 3: EV HESS topology and charge migration.
A typical battery-supercapacitor HESS topology for an EV is
shown in Figure 3. The battery and supercapacitor banks are con-
nected to the shared DC bus via chargers. Some previous works
consider direct parallel connection of battery and supercapacitor.
However, the topology poses a lot of stress on the battery bank,
and offers no freedom of control for systematic optimization, so
exclude it from discussion.
Charge migration is also necessary in the HESS for an EV.
Charge migration can be benecial under certain circumstances of
EV. For example, in cold start up situation, supercapacitor would
be empty and sudden acceleration results in drawing large cur-
rent from the battery bank. After driving is over, electrical energy
remains in the supercapacitor, which is susceptible to high self-
discharge rate of supercapacitors. For both cases, it is better to
migrate charge from one to another for energy efciency.
However, cold start and after-drive conditions are not the only
scenarios that migration can take effect. A recent study found out
that there exists optimal distribution of charge current during the
regenerative braking phase [8]. Unlike common belief, utilizing
supercapacitor as much as possible is not the optimal because large
supercapacitor SoC uctuation leads to degradation in charger ef-
ciency. We made an observation from Section 2.3 that there is
signicant asymmetry in battery charging and discharging. The
optimal discharge current of both banks during acceleration dif-
fers from the optimal charging current during regenerative braking,
which implies imbalance in ratio of net power between the bat-
tery bank and supercapacitor bank. Migration during vehicle idle
and cruise times can resolve this imbalance and enhance the overall
driving energy efciency.
4. CHARGEMANAGEMENTEFFICIENCY
ENHANCEMENT
4.1 Charge Management Efciency Enhance-
ment Problem
We formulate EV energy efciency enhancement problem as op-
timization for given driven prole. Enhancement of the energy ef-
ciency is equivalent to increasing the SoC remaining in the HESS
at the end of given driving prole is executed. We make a discrete
time approach and divide the driving prole into N equal time slots.
The dene E
HESS
as the optimization objective dened as
E
HESS
[N] = E
bat
[N] +E
cap
[N], (15)
where E
bat
[N], E
cap
[N], N are energy remaining in battery bank at
Time Slot N, energy remaining in the supercapacitor at Time Slot
N.
We dene efciency of the two chargers and motor driver as a
function of battery voltage (V
bat
), battery current (I
bat
), superca-
pacitor voltage (V
cap
), supercapacitor current (I
cap
), DC bus volt-
age (V
bus
), motor RMS voltage per phase (V
k
), and motor RMS
current per phase (i
k
). We assume that the hydraulic braking force
prole during braking is given so that the torque and angular veloc-
ity of the motor and thus the voltage and current of the motor can
GPS
navigator
Map
data
Vehicle
dynamics
Speed
profile
Charge
allocation
policy
Charge
allocation
policy
Motor V/I
profile
Supercap/
battery/
converter
models
Charge
migration
policy
Vehicle
/motor
models
I
bat
(t)
I
cap
(t)
I
bat
(t)
I
cap
(t)
I
bat
(t)
I
cap
(t)
(t)
V
k
(t), i
k
(t)
SOC
cap
SOC
bat
SOC
cap
SOC
bat
Figure 4: EV HESS management framework.
be obtained from the vehicle speed prole using vehicle dynamics
and the torque-current model of the motor in Section 2.4.
4.2 EV HESS Management Policy
We deliberately take advantage of charge migration in EV HESS
charge management. Figure 4 demonstrates the framework of the
proposed EVHESS management. First, we independently optimize
the charge allocation and charge replacement during regenerative
braking and vehicle acceleration. Such independent optimization
is feasible thanks to charge migration. We perform charge alloca-
tion considering the motor the power source and nd the optimal
prole of the DC bus voltage, supercapacitor bank current, and bat-
tery bank current. The ESS in an EV is subject to strict weight and
volume constraints. Therefore, it is not recommended to install an
excessively large supercapacitor in an EV. We assume that the ca-
pacity of the supercapacitor is designed to accommodate average
energy capacity for only one time acceleration or deceleration. We
thus adjust the SoC of the supercapacitor to its minimum level be-
fore the regenerative braking and apply charge allocation algorithm
presented in [17]. The algorithm considers most non-ideal charac-
teristics of the power converter and battery and derives the optimal
supercapacitor and battery charge currents. Likewise, we perform
charge replacement considering the motor as the load device and
nd the optimal prole of the DC bus voltage, supercapacitor bank
current, and battery bank current using the algorithm in [9]. We
maintain the SoC of the supercapacitor to be the minimum limit at
the end of acceleration to maximize its capacity utilization
2
. The
optimal charge allocation and replacement of EV are not symmet-
rical even though we apply a symmetric deceleration and accelera-
tion proles. The nal supercapacitor and battery SoC at the end of
charge allocation is generally different from the initial supercapac-
itor and battery SoC at the beginning of charge replacement. The
major reasons that cause such asymmetry include battery asymme-
try discussed in Section 2.3, existence of hydraulic brake force, and
so on.
However, the proposed method is able to provide superior energy
efciency because charge migration lls the gap between charge
replacement during acceleration and charge allocation during de-
celeration supercapacitor SoC requirement. The charge migration
algorithm tries to evenly distribute the migration current as much as
possible over time while considering the efciency of the convert-
ers. We set the initial conditions of the charge migration problem
as the nal status of the charge allocation problem, and the terminal
conditions as the initial status of the charge replacement problem,
and apply charge migration algorithm in [10].
2
We cannot fully empty the supercapacitor because the power con-
verter requires a certain minimum input voltage.
(a) Driving profile extracted from NEDC [21].
(b) Motor torque and RPM.
(c) Motor RMS voltage and current.
S
p
e
e
d

(
k
m
/
h
)
20
40
60
10 20 30 40 50
Time (s)
Time (s)
Time (s)
M
o
t
o
r

t
o
r
q
u
e

(
N

m
)
M
o
t
o
r

R
P
M
R
M
S

v
o
l
t
a
g
e

(
V
)
R
M
S

c
u
r
r
e
n
t

(
A
)
10 20 30 40 50
10 20 30 40 50
-40
0
40
-1000
0
0
400
800
-200
0
200
400
Motor torque
Motor RPM
RMS voltage
RMS current
-400
20
-20
60
1000
2000
Figure 5: Driving cycle and motor trace.
The framework solves charge management problem ahead of
time to nd input conditions for the charge migration policy. The
underlying assumption here is that we know the vehicle behavior in
the near future using GPS navigator and semi-autonomous driving
features discussed in Section 1.
5. EXPERIMENTS
We validate the proposed approach by simulation through a com-
mercial EV on standard driving cycles. The target vehicle is a 5-
door hatchback full-EV Nissan Leaf. Nissan Leaf is equipped with
an 80 kW, 280 Nm electric motor, a 24 kWh Li-ion battery, weighs
1521 kg, has an axle ratio of 7.94:1, and drag coefcient of 0.28.
Other parameters required to calculate the electrical outputs of the
motor from the driving prole are extracted from other cars of sim-
ilar size. The size of the supercapacitor is 15 F, which is similar
to the capacity used in previous works on battery-supercapacitor
HESS for EV [13]. It consists of 200 series connection of 3000 F
supercapacitors to exhibit maximum voltage of 500 V. The driving
prole we used for simulation is shown in Figure 5(a). It is a part of
the driving cycle, ECE-15 UDC (Urban Driving Cycles), in stan-
dard driving prole NEDC (New European Driving Cycle) [23].
Figures 5(b) and 5(c) show the motor torque, motor RPM, RMS
voltage, and the RMS current of the electric motor calculated from
the equations in Section 2.4. The baseline policy for comparison
is balanced charging/discharging during deceleration and accelera-
tion without active migration. Figure 6 shows the experimental re-
sults for the proposed and baseline policies. The ESS receives the
regenerative braking energy from 0 s to 10 s. The charging prole
of the ESS controlled by MPTT technique during the period is the
same both for the proposed and baseline. However, the proposed
policy performs gradual migration from the battery to supercapac-
itor during 10 s to 30 s to prepare for acceleration in 30 s to 50 s as
opposed to the baseline policy. The acceleration of the EV is much
supported by supercapacitor discharge power, which benecial for
the overall energy efciency. The proposed policy consumes 19.4%
less energy (164.3 kJ) than the baseline (201.3 kJ) for the same pro-
le.
(a) Current profile of the proposed technique.
(b) Voltage profile of the proposed technique.
(c) Current profile of the baseline.
C
u
r
r
e
n
t

(
A
)
0
200
400
10 20 30 40
Time (s)
Time (s)
Time (s)
V
o
l
t
a
g
e

(
V
)
C
u
r
r
e
n
t

(
A
)
0
100
200
300
-400
200
0
400
(d) Voltage profile of the baseline.
V
o
l
t
a
g
e

(
V
)
10 20 30 40
10 20 30 40
Time (s)
10 20 30 40
-200
0
100
200
300
-200
Supercapacitor current
Battery current
Supercapacitor current
Battery current
Figure 6: Experimental results (C = 15 F).
6. CONCLUSIONS
Electric Vehicles (EV) still should be invested more to make EV
more commercially competitive. Energy efciency enhancement
is one of the most demanding requirements to make EV commer-
cially competitive. There are many ways to enhance EV energy
efciency, especially for electricity generation, transmission and
conversion, which are major power loss. However, they are not
only related to EV development but nation-wide infrastructure ren-
ovation.
This paper introduces systematic enhancement of regenerative
braking efciency for hybrid energy storage systems (HESS) in
EV. Energy efciency enhancement of the regenerative braking
gives signicant impact on carbon emission because it is energy
harvesting directly from the wheels to EV HESS unlike plugin
charging from the grid electricity coming from fuel through power
plant, transformers, transmission lines, and distribution lines. The
proposed method decouples ESS charging efciency optimization
from the supercapacitor state of charge (SoC) management, which
generally sacrices the charging efciency in previous works due to
the limited supercapacitor capacitance. We characterize the asym-
metric factors during acceleration and regenerative braking includ-
ing the battery rate capacity asymmetry in charging and discharg-
ing. We show that systematic charge migration between the su-
percapacitor and battery among idle and cruise times enhances the
energy efciency signicantly by balancing the asymmetry. Exper-
imental results show that the proposed approach achieves 19.4%
energy efciency improvement. Such efciency gain comes from
more active and efcient use of a precious energy storage element,
the supercapacitor, through the proposed method.
7. REFERENCES
[1] S. B. Peterson, J. Whitacre, and J. Apt, The economics of using
plug-in hybrid electric vehicle battery packs for grid storage, J. of
Power Sources, 2010.
[2] D. K. Bellman, Power plant efciency outlook, in Working
Document of the NPC Global Oil & Gas Study, 2007.
[3] US Energy Information Adminstration, State electricity proles, in
http://www.eia.gov/electricity/state/, 2013.
[4] H. Yeo, D. Kim, S. Hwang, and H. Kim, Regenerative braking
algorithm for a HEV with CVT ratio control during deceleration, in
Proc. of SAE CVT Congress, 2004.
[5] D. Doerffel and S. A. Sharkh, A critical review of using the peukert
equation for determining the remaining capacity of lead-acid and
lithium-ion batteries, J. of Power Sources, 2006.
[6] S. Lukic, J. Cao, R. Bansal, F. Rodriguez, and A. Emadi, Energy
storage systems for automotive applications, IEEE T. on Industrial
Electronics, 2008.
[7] Y. Kim, Y. Wang, N. Chang, and M. Pedram, Maximum power
transfer tracking for a photovoltaic-supercapacitor energy system,
in Proc. of ISLPED, 2010.
[8] S. Chakraborty, M. Lukasiewycz, C. Buckl, S. Fahmy, N. Chang,
S. Park, Y. Kim, P. Leteinturier, and H. Adlkofer, Embedded
systems and software challenges in electric vehicles, in Proc. of
DATE, 2012.
[9] Q. Xie, Y. Wang, Y. Kim, D. Shin, N. Chang, and M. Pedram,
Charge replacement in hybrid electrical energy storage systems, in
Proc. of ASP-DAC, 2012.
[10] Y. Wang, Y. Kim, Q. Xie, N. Chang, and M. Pedram, Charge
migration efciency optimization in hybrid electrical energy storage
(HEES) systems, in Proc. of ISLPED, 2011.
[11] M. Dubarry, C. Truchot, M. Cugnet, B. Y. Liaw, K. Gering,
S. Sazhin, D. Jamison, and C. Michelbacher, Evaluation of
commercial lithium-ion cells based on composite positive electrode
for plug-in hybrid electric vehicle applications. part I: Initial
characterizations, J. of Power Sources, 2011.
[12] R. Carter and A. Cruden, Strategies for control of a
battery/supercapacitor system in an electric vehicle, in Proc. of
SPEEDAM, 2008.
[13] M. Ortuzar, J. Moreno, and J. Dixon, Ultracapacitor-based auxiliary
energy system for an electric vehicle: Implementation and
evaluation, IEEE T. on Industrial Electronics, 2007.
[14] E. Schaltz, A. Khaligh, and P. Rasmussen, Inuence of
battery/ultracapacitor energy-storage sizing on battery lifetime in a
fuel cell hybrid electric vehicle, IEEE T. on Vehicular Technology,
2009.
[15] J. Cao and A. Emadi, A new battery/ultracapacitor hybrid energy
storage system for electric, hybrid, and plug-in hybrid electric
vehicles, IEEE T. on Power Electronics, 2012.
[16] M. Pedram, N. Chang, Y. Kim, and Y. Wang, Hybrid electrical
energy storage systems, in Proc. of ISLPED, 2010.
[17] Q. Xie, Y. Wang, Y. Kim, N. Chang, and M. Pedram, Charge
allocation for hybrid electrical energy storage systems, in Proc. of
CODES+ISSS, 2011.
[18] D. Shin, Y. Wang, Y. Kim, J. Seo, N. Chang, and M. Pedram,
Battery-supercapacitor hybrid system for high-rate pulsed load
applications, in Proc. of DATE, 2011.
[19] J. Y. Wong, Theory of ground vehicles. Wiley-Interscience, 2001.
[20] A. Tashakori, M. Ektesabi, and N. Hosseinzadeh, Modeling of bldc
motor with ideal back-emf for automotive applications, in in Proc.
WCE, 2011.
[21] Y. Choi, N. Chang, and T. Kim, DC-DC converter-aware power
management for low-power embedded systems, IEEE T. on CAD,
2007.
[22] Y. Wang, Q. Xie, M. Pedram, Y. Kim, N. Chang, and M. Poncino,
Multiple-source and multiple-destination charge migration in
hybrid electrical energy storage systems, in Proc. of DATE, 2012.
[23] NEDC, European Council Directive, 70/220/EEC with amendments.