Page 0670

EVS24
Stavanger, Norway, May 13-16, 2009

Design Methodology of Energy Storage Systems for a

Small Electric Vehicle
João P. Trovão1,3, Paulo G. Pereirinha1,3,4, Humberto M. Jorge2,3
1
IPC-ISEC, Polytechnic Institute of Coimbra, Rua Pedro Nunes, Coimbra, Portugal, {jtrovao, ppereiri}@isec.pt
2
DEEC – FCTUC, University of Coimbra - Pólo II, P-3030-290 Coimbra, Portugal, hjorge@deec.uc.pt
3
INESC-Coimbra, Institute for Systems and Computers Engineering at Coimbra, Portugal
4
APVE, Portuguese Electric Vehicle Association Portuguese Branch of AVERE, Lisboa, Portugal

Abstract
With the current state of technological development, the future of Electric Vehicles (EVs) seems to go
through the hybridization of various Energy Storage Systems (ESSs). This strategy seeks to benefit from
the best qualities of each available energy source, and is especially useful in urban driving. In this work, the
need for multiple energy sources hybridization is addressed. A methodology to optimize the sizing of the
ESSs for an electric vehicle taking as example the ISEC-VEIL project, using different driving cycles,
maximum speed, a specified acceleration, energy regeneration and gradeability requests are presented. It is
also studied the possibility of using a backup system based on solar energy, that may be considered in the
design, or as an extra to cope with unforeseen routines and to minimize the recharge of ESSs. Some
simulation results of multiple energy sources hybridization are presented, considering different ESSs and
different scenarios for the small presented EV, in order to verify the proposed designs.

Keywords: Neighborhood Electric vehicle (NEV), optimization, energy storages, battery electric vehicle (BEV), cost
reduction.

1 Introduction also increases pollution emissions, especially

Greenhouse Gas emissions, which must be prevented
The energy supply problem, with economic, for the sustainability of the planet and for life quality.
ecologic and geopolitical aspects, is at the heart of Concerning urban pollution, the emissions of ICE
today’s political and scientific agenda. In 2005, the vehicles are one of its major sources, especially in
transports sector accounted for 60.3% of world oil medium and large cities. The high incidence of
consumption, against 45.4% in 1973 [1][2]. This respiratory problems, allergies, asthmas, and some
increasingly fuel consumption leads to forecast oil cancers is an increasing problem, leading to public
shortage and price rise, confirmed by the July 2008 health concerns, as air pollution contributes
crude oil spot prices, very close to US$150. Even if definitively to mortality and morbidity [3][4].
the present world economic crisis releases, for the Due to its very high efficiency, no local emissions,
moment, some of the price problem, there are also silent driving, and its ability to recover breaking
very important energy dependence and security energy, electric machines are the key components to
concerns as the crude oil come mainly from middle sustainable mobility and world’s future, both in pure
east and/or instable countries. EVs or in HEVs. A transition step while looking
The mass utilization of Internal Combustion forward the ideal solution of Zero Emissions
Engine (ICE) vehicles in the transportation sector

EVS24 International Battery Hybrid and Fuel Cell Electric Vehicle Symposium 1
 Page 0671

Vehicles, are the Low Emissions Vehicles, as the For the “full electric” EV the solutions pass by
HEVs, specially the Plug-In HEVs (PHEVs) [5][6]. significant progresses in battery technology and by
using different energy sources with optimized
2 Energy Storage Systems for management of energy flow. [9] [10]
EVs 2.3 Multiple Energy Sources
Hybridization
2.1 The Energy Storage Issue in EVs
As mentioned before, none of the available energy
To allow EV to become the effective sustainable
sources can easily fulfil alone all the demand of
transportation solution, a great effort has to be done
EVs to enable them to compete with gasoline
in R&D to overcome the major technical issue in
powered vehicles. In essence, these energy sources
EVs: the energy storage.
have a common problem: they have either HSE or
Typically, an EV stores energy in batteries (Lead-
HSP, but not both. A HSE energy source is
Acid, NiMH or Li-Ion, for instance) that are bulky,
favourable for long driving range, whereas a HSP
heavy and expensive. The specific energy, in
energy source is desirable for high acceleration
Wh/kg, of gasoline is about 12500 Wh/kg (which
rate and high hill climbing capability. The concept
only 2000-3000 can be considered useful energy,
of using and coordinate multiple energy sources to
due to the very low efficiency of the ICE) against
power the EV is typically denominated
typically 35 in good lead-acid batteries or 60 in
hybridization. Hence, the specific advantages of
NiMH, which gives an idea of the volume and
the various EV energy sources can be fully
weight necessary to store the energy needed to do
utilized, leading to optimized fuel economy while
the same work. Li-ion batteries have higher specific
energy, usually from 80 to 120 Wh/kg, but they still
satisfying the expected driving range and
quite expensive and have some safety issues that
maintaining other EV performances. [10][11]
have to be carefully addressed. Due to this problem,
A lot of work has been done to investigate
with current batteries technologies it is very methodologies to sizing and control strategies for
difficult to make a general purpose EV that fuel-cell-battery [13], fuel-cell-supercapacitor [11],
effectively competes with ICE cars. For massive fuel-cell-battery-supercapacitor [10][13], battery-
deployment of EV its driving range problem must supercapacitor [12] vehicles. These studies add a
be solved. significant knowledge to the field but do not
provide a detailed comparison of a battery-
supercapacitor-PV array, principally of the daily
2.2 Main Available Energy Storage
energy evolution, in function of the test cycle. As
Systems the fuel-cell vehicle is an expensive choice, it is
At the present and in the foreseeable future, the necessary to analyze other solution for EV
viable EVs energy sources seem to be batteries, hybridization.
fuel cells, supercapacitors (SCs) and ultrahigh- The present study aims at developing a
speed flywheels. methodology to help the designers to optimise the
Batteries are the most mature source for EV sizing of the power components in a small urban
application. But they offer either high specific EV. For that, it used different driving cycles,
energy (HSE) or (relatively) high specific power specified acceleration performance and maximum
(HSP). Fuel cells are comparatively less mature speed requests.
and expensive for EV application. They can offer
exceptionally HSE, but with very low specific 3 Veil Prototype: Hybridization of
power. In spite of some quite expensive
prototypes, such low specific power poses Energy Storage Systems
serious problems to their application to EVs that At the Electrical Engineering Department of the
desire a high acceleration rate or high hill Engineering Institute of Coimbra (DEE-ISEC) the
climbing capability. Also, they are incapable of author’s team has started the on going VEIL
accepting the high peaks of regenerative energy project to convert a small vehicle, initially with an
during EV braking or downhill driving and, ICE, into an electric vehicle (Figure 1) [7][8][15].
worse, their overall energy efficiency is very low For VEIL project prototype, it was considered to
(about 25% from “wind to wheel”). SCs have be viable the hybridization of three energy sources:
low specific energy for stand alone application. a HSE storage system – Batteries –, a HSP system
However, they can offer exceptionally HSP (with – SCs – and photovoltaic panels, PV. Figure 2
low specific energy). Flywheels are shows this hybridization configuration. [7] [8]
technologically immature for EV application.

EVS24 International Battery Hybrid and Fuel Cell Electric Vehicle Symposium 2
 Page 0672

using the generated energy. The solar panels array

works as an energy back-up, as, whenever solar
energy exists, the panels generate energy that is
used by the system. In the tractive mode, this
assists the stored energy and when the vehicle is
parked, it provides a charge of the storage systems.
The power generated and stored in the energy
system (Pst) at any time should be at least equal to
EV power demand (Pdem).
Pst ≥ Pdem (1)
Pst is composed by the batteries power (PBat), the
SC power (PSC), the PV array power (Ppv) and the
Figure 1: VEIL project in preparation at ISEC campus. regenerative break power (Preg):
Pst = PBat + PSC + Ppv + Preg (2)
4 Design Methodology On the other hand, the PV array has not a response
The considered configuration of the VEIL capability for a high power request and should be
powertrain consists of a battery bank, a seen as a backup system in order to recharge the
supercapacitor bank, a photovoltaic panels array, batteries for a prolonged use, for daily journey and
a DC/DC converter, an inverter, an AC motor, for extended parking periods, and for covering its
and a transmission. self discharge. With this consideration, (2) can be
The multiple power converter regulates the VFD simplified to (3), and the PV array energy
DC-link voltage and adapts the different voltage complement is analyzed separately of the other
level of the considered Energy Storage Systems ESSs.
(ESSs). The VFD inverter converts the regulated Pst = PBat + PSC + Preg (3)
DC voltage to an AC voltage to drive the AC
The proposed methodology is based on the
motor. The transmission is a gearbox that
combination of ESSs with the lowest cost and
increases the motor torque using a fixed gear
weight, minimizing the number of storage units in
reduction.
series and parallel, maximizing autonomy and with
When the EV demands high power, the batteries
a relatively better performance.
and SCs provides power to the vehicle’s wheels
For the considered hybridization, the three energy
through the DC/DC converter, the inverter, AC
devices can be divided into two types: a classic
motor, and the transmission. On the other hand,
type (batteries and SCs), because, traditionally, the
when the EV demands low power, the batteries
EVs use only batteries or more recently batteries
provides power to the wheels through the DC/DC
and SCs, and a special type, as it is not common
converter, the inverter, the AC motor, and the
the EVs to use photovoltaic panels. Therefore, the
transmission and charges the SCs through the
design methodology needs to divide into a classic
reversible DC/DC converter. When the vehicle
energy storage design and a backup energy supply
brakes, the AC motor converts the kinetic energy
design.
of the vehicle into electricity and charges
principally the SCs and residually the batteries
through the inverter and the DC/DC converter 4.1 Classic Energy Storage Design
The design is based on the estimated power
demand required by the EV powertrain, so that the
EV makes a particular speed profile trip,
responding to requests of drive cycles, for
maximum speed, acceleration, breaking and
gradeability.
The layout of the batteries bank and that of the SCs
bank is shown in Figure 3. The battery and
supercapacitor units are connected in series to
form a branch and multiple branches are paralleled
to form a battery bank and SC bank. Therefore, the
total cost (CTc) of the ESSs is given by the sum of
the unit cost (CBat and CSC) of the number of the
Figure 2: EV project power scheme. series units (XBat and XSC) and parallel branches

EVS24 International Battery Hybrid and Fuel Cell Electric Vehicle Symposium 3
 Page 0673

KBat 1 KBat 2 KBat 3 K Bat n KSC 1 KSC2 KSC 3 KSCn Constraints on the number of branches paralleled
XBat 1 X SC1 are discussed on the next subsections, because
X Bat 2 XSC 2
KBat, and KSC variation ranges depend on the EV
X Bat 3 X SC3
desired performance (autonomy and dynamic
response).

X Bat i XSCi 4.1.1 Drive Cycles

Given the need for energy to propel an automobile,
Batteries Banks Supercapacitors Banks since there are different types of
Figure 3: Layout of the Batteries and Supercapacitors engines/powertrain it was necessary for test
Banks. procedures to compare several engines/powertrain
with each other. These test procedures are called
(KBat and KSC) of the batteries and SCs, as driving cycles. A driving cycle is a standardised
presented in (4). The costs CTc, CBat, and CSC are driving pattern, described by means of a speed-
in European Currency (Euro, €). time table. Therefore, the speed and the
CTc = C Bat ⋅ X Bati ⋅ K Batn + CSC ⋅ X SCi ⋅ K SCn (4) acceleration are known for each point of time and
The unit cost and specific characteristics of the the required mechanical power as a function of
samples battery units and SCs cells used in this time can be determined for each vehicle.
work are presented in Table I. The presented Several driving cycles are used and the typically
information is for commercial devices, now the world-wide can be divided into three groups:
available on the market. European Driving Cycles; US Driving Cycles and
Japanese Driving Cycles. For this work it was
Table 1: Batteries and Supercapacitors Unit considered the official’s European Driving Cycles,
Characteristics
like ECE 15 and EUDC.
Battery A Battery C Battery B The ECE 15 driving cycle represents a typical big
Reference SC
Li-ion NiMH Li-ion
Type TS-LFP40AHA TS-LFP90AHA
VH Module
PC2500
city urban driving. It is characterised by low
VH F 10S
Manufacturer Thunder Sky Thunder Sky Saft Maxwell
vehicle speed (max. 50 km/h), perfectly adapted to
Weight (kg) 1.5 3.2 3 0.725 the chosen vehicle type that is limited to 45 km/h,
Volume (L) 1 2.17 1.7 0.6
Voltage (V) 3.2* 3.2* 12* 2.7**
by law. The EUDC cycle describes a suburban
Capacity/Capacitance 40Ah @ 0.3C 90Ah @ 0.3C 13.5 Ah @ 2C 2700 F route. At the end of the cycle the vehicle
Specific Power [W] 384 864 360 ---
Cost (€) 78.5 135.8 137.5 24.0
accelerates to highway-speed. Both speed and
* Nominal cell voltage; ** Maximum cell voltage acceleration are higher than in ECE 15. The
EUDCL is a suburban cycle for low-powered
From (4), one can see that CBat, CSC, XBat, XSC, vehicles. It is similar to the EUDC but the
KBat, and KSC should be reduced to minimize CTc maximum speed is 90 km/h. And, finally, the
cost function. However, there are constraints on NEDC is a combined cycle consisting of four ECE
reduction of all cost function coefficients. 15 cycles followed by an EUDC or EUDCL cycle.
For constrains on CBat and CSC, it is assumed that The NEDC is also called the ECE cycle. For the
the cost of each battery and SC units are fixed VEIL project the maximum speed of the suburban
and independent of the other coefficients. The cycle part was readjusted for a 50 km/h maximum
constraints on the numbers of unit series, XBat and speed. The considered NEDC is presented in the
XSC, are function of the powertrain topology. For Figure 4.
the current VEIL project, it is considered for all
energy storages a nominal DC-link voltage of 96 100
NEDC Cycle

V with a limited voltage variation (±10%).

[km /h]

50

For these considerations and for the selected 0

0 200 400 600 800 1000 1200

ESSs, the variation of the constraints XBat and

4 Powers
x 10

4 Electric Power Demand (Pe)

Ef f ectiv e Regenarativ e Power

XSC, in order to hold the DC-link voltage, are

[W ]

presented in Table 2. 0 200 400 600 800 1000 1200

Energies

Table 2: Variation of the Constraints XBat and XSC for 400 Regenerativ e Energy
Energy Demand
[W .h]

96 V ± 10%
200

0
0 200 400 600 800 1000 1200
Time [s]
DC-link Voltage 1st approach
Reference Xmin Xmax
variation [V] XBat and XSC Constraints Figure 4: NEDC for low-speed vehicle and NEDC
Battery A 25 35 87.5 < Vdc < 105.0 XBat = 30 applied to the VEIL speed versus time, and resultant
Battery B 25 35 87.5 < Vdc < 105.0 XBat = 30 powers and energies.
Battery C 7 9 90.0 < Vdc < 105.0 XBat = 8
SC 32 42 86.4 < Vdc < 105.0 XSC = 37

EVS24 International Battery Hybrid and Fuel Cell Electric Vehicle Symposium 4
 Page 0674

After modelling the vehicle, taking into account 4.1.2 Maximum EV Speed Request
the mechanical parts, including body and For the case of the vehicle cruising at its maximum
transmission units, and the dynamic and speed, the power demand is given by:
aerodynamic vehicle characteristics, with the
1 2 
model used [8][14] applying the chosen driving Pvmax = (μ RR ⋅ m ⋅ g ) ⋅ vmax +  ⋅ ρ ⋅ CD ⋅ AF ⋅ vmax  ⋅ vmax (5)
2 
profile, results were obtained for power and
where Pv is the mechanical EV power demandmax
energy demand, and regenerative power and
(W) for the maximum speed drive, μRR is the
energy, respectively. These results are presented
coefficient of rolling resistance, m is the
in Figure 4.
considered EV mass (kg), g is the gravity
Analyzing the power graph shown in Figure 4,
acceleration (9.81 m/s2), ρ represents the air
we can easily separate the EV demands of high
density (1.204 kg/m3 at 20ºC), CD is the drag
power, low power and the effective regenerative
coefficient, AF is the frontal projection area (m2)
power1. The cases of the high and low EV power
and vmax is the maximum vehicle speed (m/s). In
demands and regenerative power will be
this project and with the aforementioned
analyzed in the next subsections. For the driving
restrictions the maximum speed considered is 50
cycle analysis, the main interest is the energy
km/h on the flat road. In this paper, the EV project
required to perform a chosen driven cycle. In
vehicle specifications are presented in Table 4, and
other words, this analysis permits to obtain
they were used for this design study.
information about the need of energy available in
the main energy storage that will define the basic Table 4: EV Project Vehicle Specifications
autonomy of the EV, usually the battery banks. Parameter Value
Although the test cycles have different types of Vehicle mass (kg) 500*
requests (e.g. acceleration, maximum speed and Rolling Resistance Coefficient 0.015
braking), as the energy used by SCs is renewable, Aerodynamic Drag Coefficient 0.51
Front Area (m2) 2.4
then in terms of autonomy it is only necessary to Wheels radius (m) 0.26
assess the number of batteries to use. Therefore, Gearbox transmission ratio 10
subtracting the effective regenerative energy of * With the mass of the typical ESSs and 1 passenger.
the energy demand (Figure 4), we obtain the
Computed equation (5), the mechanical EV power
required energy to move the sample vehicle for
demand for the maximum speed drive 50 km/h was
the chosen cycle. So, for the modified NEDC
3 kW. Therefore, if it was considered 70% of
(limiting the maximum speed @ 50 km/h), the
efficiency of the VEIL full power chain, the
sample vehicle travel a distance of 9.31 km, with
electrical power demand is 4.3 kW. Simulation
an average speed of 27.94 km/h and require
results of the case study are presented in Figure 5.
about 1000 Wh.
With the information presented in Figure 4, we 50 km/h Cte.

can evaluate the minimum number of KBat for the 40

[km/h]

20

sample vehicle to achieve one modified NEDC 0

Cycle. The results for each considered battery 0 20 40 60 80 100

Powers
120 140 160 180 200

type are presented in Table 3. 5000

4000

3000
[W]

Table 3: Evaluation of the Constraint KBat for NEDC 2000

1000 Electric Power Demand (Pe)

Power Demand (Pu)

Cycle 0
0 20 40 60 80 100 120 140 160 180 200

Energy Demand
300
Total Modified Autonomy
200
Reference XBat KBat energy NEDC [km]
[W.h]

100
[kWh] Cycle [#x]
0
Battery A 30 1 3.840 3.766 x 35.0 0 20 40 60 80 100
Time [s]
120 140 160 180 200

Battery B 30 1 8.640 8.474 x 78.8

Battery C 8 1 1.450 1.343 x 12.5 Figure 5: Mechanical, Electrical Power and Energy
demands to the VEIL at the constant speed 50 km/h.
As to assure the nominal DC-link voltage of When the vehicle cruises at the maximum vehicle
96 V, there is a minimal number of batteries in speed, are the battery packs alone that provide
series, XBat (Table 2), the number of modified power to EV due to the limited energy capacity of
NEDC cycles possible is bigger than one, for all the other energy systems.
cases. In the previous analysis, this was based on the
batteries energy density; now, for the maximum
1
speed, the analysis is made on the specific power
Effective Regenerative Power is the power that for the present
work can be recovered from the SCs (about 65% of the Total
of each battery types. The results of this analysis
Regenerative Power available on the wheels). are presented in Table 5.

EVS24 International Battery Hybrid and Fuel Cell Electric Vehicle Symposium 5
 Page 0675

Table 5: Evaluation of the Constraint KBat @ 50 km/h requirement. In this case, we need a 45 kW for 8 s,
Specific which is 360 kJ. The system will normally operate
Reference XBat KBat Power
Autonomy
[h]
Autonomy
[km]
at 96 V ± 10% and, we assume for this application,
[kW] it will be operating at or near its nominal voltage
Battery A 30 1 11.52 0.893 44.7
Battery B 30 1 25.92 2.009 100.0
most of time. Dividing the nominal voltage by the
Battery C 8 2 5.76 0.674 33.7 nominal cell voltage to get the required number of
The XBat are the presented in Table 2. cells in series, like in 4.1, and XSC = 37 cells. For
other hand, the SC current average (ia) is given by:
4.1.3 Acceleration Request
P P  (7)
ia =  SC + SC  2
For the case of the acceleration request, during  Vmin Vmax 
initial phase of each travel, the power demanded where PSC is the requested power to SCs, and Vmin
by the EV from the powertrain is given by: and Vmax are, namely, the minimum and maximum

1 dv(t )  SC voltages system. Thus, ia was calculated to be
Pa =  μ RR ⋅ m ⋅ g + ⋅ ρ ⋅ C D ⋅ AF ⋅ v 2 (t ) + M ⋅  ⋅ v(t ) (6)
 2 dt  474.3 A. The cell resistance for the chosen SC,
where Pa is the EV power demand in (W), and Rcell, is 1 mΩ, and the RC time constant is the
v(t) is the vehicle speed in function of the time product of its capacitance value and resistance
(m/s). value. For this example, time constant of one SC is
The initial acceleration performance for the EV 2.7 s, therefore, the total stack resistance (Rtotal) is
project prototype is defined as accelerating the SC time constant (2.7 s) divided by the total
vehicle from standstill to 50 km/h in 8 s, like to capacitance of the SC system (Ctotal). Having all
the NEDC cycle, when the vehicle start at the variables defined to compute the voltage drop
standstill to vmax (see Figure 4 and 6). (dV) during the discharge of the capacitors (8),
The power demanded by the EV project to
dt 
dV = ia ⋅  + Rtotal  (8)
achieve the aforementioned acceleration  total
C 
performance was calculated using (6) and is we will rearrange (8) and solve for Ctotal, resulting
shown in Figure 6. The mechanical EV power Ctotal = 264.32 F. Now, with the total value of SC
demand for the standstill to maximum speed system capacitance and the number of series cells
drive 50 km/h was 34.54 kW. Therefore, if it is required (XSC = 37 cells) and the expression (9), we
considered 70% of efficiency of the VEIL power can calculate the number of parallel (KSC).
chain, the electrical power demand is 49.35 kW, K
as can be seen in Figure 6. Ctotal = Ccell ⋅ SC (9)
X SC

50
Severest Acceleration in ECE 15 Drive Cycle The value of KSC was calculated to be 3.62, and it
40
is assumed KSC = 4.
[km/h]

30

20
10
For other value of the SC power request, namely,
0
120 if lower SC power is required, decreasing KSC and
4 Powers

4
x 10

Electric Pow er Demand (P)

Pow er Demand (Pa)
increasing the supply of power by batteries, we
have other configurations of KBat and KSC, such as
[W]

0
those presented in Table 6.
116 117 118 119 120 121 122 123 124 125 126

Table 6: Evaluation of the Constraints KBat and KSC for

Figure 6: Mechanical and Electrical power demand of acceleration request
the vehicle during the considered severest initial
acceleration. Combination Reference XBat KBat Specific Power [kW]
A Battery A 30 1 11.52
During initial acceleration, both the battery and B Battery B 30 1 25.92
SC in the EV powertrain have to supply high C Battery C 8 2 5.76
D Minimum battery power required 4.30
power to the vehicle. It is evident that if fewer
XSC KSC Capacitance [F]
branches in the battery banks (smaller Kbat) are A SC 37 3 218.92
used, more branches in the SC bank (larger KSC) B SC 37 2 145.95
have to be used to meet the vehicle power C SC 37 4 291.90
demand and maintain stable DC-link voltage and D SC 37 4 291.90
vice versa.
For this analysis and for maximum SC power Analyzing the values of KSC shown in Table 6, it is
request, let us assume that the batteries provide clear that if the batteries specific power is greater,
the power required travelling at maximum speed, the number of parallels of SCs (KSC) decreases.
cf. 4.1.2, and we need only that the SC give the Owner, this is not linear: even a rise of 4.5 times
difference to the acceleration EV power the specific power of batteries (case B), the
number of necessary parallel SC is only halved.

EVS24 International Battery Hybrid and Fuel Cell Electric Vehicle Symposium 6
 Page 0676

4.1.4 Energy Regeneration request that its speed is low, the term with greater impact
In this case, when the EV brake, it is possible to on the power demand is the third term, which is
use regenerative energy because the Pdem linked to the inclination of the road.
becomes zero and the motor torque becomes Using (10), the defined gradeability and
negative and thus enables the energy generation. specifications of the vehicle, presented in Table 4,
In this case, the most demanding braking NEDC the power demanded by the vehicle during grading
cycle define the minimum value of the SC was calculated to be about 4 kW. It should be
capacity necessary for the sample vehicle noticed that around 0.7 kW is from the rolling
recovered the all braking energy. The resistance force and the aerodynamical drag force,
representation of the power produced during and the net power demand for the grading is 3.35
braking and the effective regenerative power that kW. Considering again 70% of the efficiency of
can actually be stored (considering an energy the VEIL power chain, the electrical power
return path with efficiency of 65%) are shown in demand is 5.75 kW for the considered gradeability.
Figure 7. Applying the expression (6), suitably The graphs of power demanded by the EV project
adapted to the case of a deceleration of 50 km/h to achieve the aforementioned climbing
to standstill in 13 s, the total power that can be performance are shown in Figure 8.
stored is 10.7 kW, as presented in Figure 7. During grading, only the battery provides power to
the vehicle. This is because the supercapacitor has
Severest Braking in Midified NEDC
60
limited energy capacity and cannot be used to
power the vehicle during long periods of grading.
40
[km/h]

20

0
1175 1180 1185 1190 1195
Therefore, the power supplied by the battery banks
0
4
x 10 Regenerative Powers is about 5.75 kW, and the power supplied by the
-0.5
supercapacitor bank is zero. Since the
[W]

-1

-1.5 Total Regenerative Pow er

Ef f ective Regenerative Pow er
supercapacitor does not provide power to the
-2
1175 1180
[Time [s]
1185 1190 1195
vehicle during grading, gradeability does not
Figure 7: Total and Effective regenerative power of constrain KSC, but constrains KBat.
the vehicle during the considered severest breaking 23.5 km/h @ 10% Climbing

phase. 24.5

24

23.5

Applying the same approach as in the previous 23

section, expressions (7), (8) and (9), we can to 22.5

0 20 40 60 80 100 120 140 160 180 200

Powers

calculate the value of the SC total capacity 6000

required in this phase, Ctotal = 91 F, giving a KSC 4000

[W]

2000

to 1.34, then rounded to the higher integer, 0

Electric Pow er Demand (Pe)
Pow er Demand (Pu)

KSC = 2. This value is the lower limit of the 0 20 40 60 80 100

Energy Demand
120 140 160 180 200

400

restriction KSC, if we want to store all of the 300

regenerative energy in the more demanding

[W.h]

200

100

braking NEDC modified cycle. 0

0 20 40 60 80 100 120 140 160 180 200
Time [s]

Figure 8 – Power demand to achieve the defined

4.1.5 Gradeability Request
climbing.
The vehicle power demand during grading (Pg)
is given as: Again, the discussion of constraint KBat is based on

1  the value of the specific power of batteries pre-
Pg =  μ RR ⋅ m ⋅ g + ⋅ ρ ⋅ C D ⋅ AF ⋅ v g + M ⋅ g ⋅ sin (θ ) ⋅ v g (10) selected. The results of this analysis are presented
 2 
where θ is the climbing angle and g is the in Table 7.
maximum request vehicle speed at grading Table 7: Evaluation of the Constraint KBat @ considered
periods. gradeability
In this paper, gradeability for the VEIL project Specific Autonomy Autonomy
vehicle is defined as the ability to grade a road Reference Power XBat KBat [h] [km]
with a climbing angle 6º, which is about 10% [kW]
Battery A 30 1 11.52 0.67 15.70
inclination at a constant speed of 23.5 km/h. Battery B 30 1 25.92 1.5 35.25
The first two terms of (10) do not depend on the Battery C 8 2 5.76 0.5 11.75
inclination, but they are function of the minimum The XBat are the presented in Table 2.
power need to put a vehicle in motion. The first
term is related to the rolling resistance force and 4.1.6 Synopsis of the constraints KBat and KSC
the second is related to the aerodynamic drag In the previous sections, it were studied the
force. Considered, when the vehicle is climbing, constraints KBat and KSC for different sample

EVS24 International Battery Hybrid and Fuel Cell Electric Vehicle Symposium 7
 Page 0677

vehicle operations. The minimum values of the commercial PV, the constraint Xpv was calculated
constraints KBat and KSC resulting of these to be 5. But, the number of PV to make the array is
analyses are presented in Table 8. function of the dimension of the VEIL rooftop and
Table 8: Constraints KBat and KSC
the hood (the hinged cover over the motor vehicles
that allows access to the motor compartment for
Constraints maintenance and repair), and of the considered PV
Battery SC
KBat KSC
commercial model. The usable space on the
Battery A 1 32 rooftop of the considered vehicle is defined by a
Battery B 1 2 surface about 1300 x 1100 mm, and the hood space
Battery C 2 43 defined by 550 x 1100 mm. With this dimension it
is possible to implant 5 selected PVs, 4 in the
To store all of regenerative energy, the minimum rooftop and 1 in the hood, as it is presented in
is KSC = 2, possible only for the case of type B Figure 9.
battery. In the other cases, KSC must be greater,
depending on the value of specific battery power
of A and C type; in case C, the need to place a
minimum of KSC = 4, is setting the value of Kbat
in the minimum amount required for a cruising
speed of 50 km/h. Somehow, what will define the
balance between these constraints is the cost and
the weight of the ESSs, as will be analyzed
further on.

4.2 Backup Energy Supply (PV array)

Design
The backup energy system, photovoltaic panels a)
or cells, also can be connected in series to form a

branch and multiple branches are paralleled to
form a PV array. Therefore, the total cost (CTb) of 







the PV array system is given by the unit cost
(CPV) of the number of the series units (Xpv) and
parallel branches (Kpv) of the PVs, as presented 




in (11). The costs CTb, and CPV are in European
Currency (Euro, €).
CTb = C PV ⋅ X pvi ⋅ K pvn (11) b)
The sample photovoltaic panels used in this work Figure 9: a) View from the top of VEIL project; b)
are modules (BP MSX 30) designed for Schema of the PV array implementation.
applications requiring a combination of light As there is not usable space in the VEIL vehicle to
weight, compactness, and ruggedness. Its unit get more PV panels, we find that the Xpv and Kpv
cost and specific characteristics are presented in may have the values, 5 and 1, respectively. Thus,
Table 9. The presented information is for a the total cost of the PV array will have only two
commercial device, now available in the values, zero cost is without the support of
electrical market. renewable energy or a fixed cost of 1200 €. The
same approach can be made for the weight (0 or 15
Table 9: Characteristics of PV panels kg).
Warranted
Voltage Dimensions Mass Cost
For the solar energy, the study uses the average
PV Array Power hourly statistics for direct normal solar radiation
[V] L x W [mm] [Kg] [€]
[W]
[Wh/m²] – for the last 30 years at the project
BP MSX 30 27 16.8-21 616 x 495 3 240
location, Coimbra (see Figure 10). As an example,
for a typical day of two different months,
The constraint Xpv is limited by the DC-link November and August, with the minimum and
voltage and for its considered value and maximum solar radiation, respectively, and using
the efficiency PV array model, it can be computed
2
For KBat considered. If KBat increases, KSC can decrease to a the global generated energy by the panels, as
minimum of 2. shown in Figure 11, and considering, somehow,
3
For KBat considered. If KBat increases, KSC can decrease to a the mounting position of the PV panels
minimum of 2.

EVS24 International Battery Hybrid and Fuel Cell Electric Vehicle Symposium 8
 Page 0678

40.24 km, followed by 8 h parked and again 4.5

800
NEDCs. The total journey distance is about 80.48
700 km. For this approach it is needed to include an
additional restriction (12), where the total energy
Norm al So la r Radiation [W h/m ²]

600

500

400
needed (WTotalmin) for to accomplish the considered
300 number of considered cycle test, is given by:
200
WTotal ≥ N º Cycle ⋅ W NEDC (12)
100

0 where WTotal is the sum of the total energy of the

20

15 Out
Nov
Dec battery banks, the total regenerative energy in all
10
May
Jun
Jul
Aug
Sep
studied cycle tests and the minimum energy
5 Apr

0
Jan
Fev
Mar
recovered by the PV array, and it is given by (13):
WTotal min = K bat ⋅ Wbat / k +
Hour of Day
Moth of Year

Figure 10: Average hourly normal solar radiation.

bat
(13)
+ N º Cycle ⋅ W reg / cycle + K PV ⋅ W PV / k PV
In Figure 11, we can see that the variation of the For this example the minimum energy needed in
energy accumulated during the typical day of the each NEDC cycle is about 1000 Wh.
considered two months, is positioned between Applying the constraint (13) together with the
the 900 Wh and 1350 Wh a day. With these restrictions listed previously, cf. 4.1.6, we see that
results, the advantage of this topology PV array minimizing the cost of storage systems for energy,
investment is the recovering of at least 900 Wh a depending on the different types of batteries, get
day, when the car is at direct sunlight. Next it the results shown in Table 10.
needs to review the weight of the increased
number of branches in parallel to batteries versus Table 10: Evaluation of the Total Costs for 9 NEDCs
investing in this backup system. Total
Weight
Expected
Reference XBat KBat XSC KSC XPV KPV Cost Autonomy
[kg]
Daily Average Energy [€] [km]
1400
Battery A 30 2 37 3 5 0 7374 170.48 81.3
August's Daily Average Energy
November's Daily Average Energy Battery B 30 1 37 2 5 0 5850 149.65 90.1
1200
Battery C 8 6 37 3 5 0 9264 224.48 90.6

1000 From Table 10, it can be seen that the most viable
option, both for cost and weight to the presented
800 scenario, is to use the battery type B, because it
Energy [W.h]

will minimize the use of SC. So with a single bank

600
of batteries of the type B and 2 banks of SCs, we
can make the outward journey and return. The
400
other options are more expensive and heavier, and
200
the solution with the same chemical in the solution
B is better than the solution with the NiMH
0
batteries. The only solution which could consider
0 5 10 15 20
Hour of Day the use of PV is a solution with batteries of type C,
it can reduce the number of banks of batteries to
Figure 11: Daily average energy for typical months.
use up 5 and the PV array, although this brings an
increase in initial cost of investment, plus a
4.3 Cost Reduction by Constraints minimization of the cost of energy purchased for
KBat and KSC loads of storage systems for energy.
To analyze the cost reduction with the previous The same approach can be made for other
study of constrains, for example, we can use a scenarios, including a scenario corresponding to a
hypothetic scenario, where there were considered typical routine for mobility in big European cities,
three different time periods: a first displacement with low average speed and very frequent stops
in the morning, starting at 7:30 am and taking 1.5 and goes. To simulate this behaviour it was used
h (to get to work, for instance), a second period the ECE 15 cycle, presented in Figure 4 (first
where the car is parked outdoor and lasting 8 h, 200s). The travel in the morning is constituted by a
and a third period equal to the first one, sequence of 27 ECE 15 cycles, corresponding to
corresponding to the return back home, from nearly 27.35 km. The same distance has to be
17:00 to 18:30 pm. To fulfil the 1.5 h travel in travelled in the evening to make the way back.
the morning and in the evening, based on NEDC This scenario consists then in 27 ECE 15 cycles
cycle, we need to use 4.5 NEDCs, totalizing during 1.5 h, followed by a period of 8 h parked

EVS24 International Battery Hybrid and Fuel Cell Electric Vehicle Symposium 9
 Page 0679

outdoor, and then again 27 ECE cycles. The total presented, can be made comparative simulations of
journey distance is 54.7 km (2 times 27.35 km). the various options proposed. These simulations
The results are presented in the Table 11. are made using the VEIL dynamical model, with
Table 11: Evaluation of the Total Costs for 27 ECEs
the different weights (Tables 10, 11 and 12) of
considered energy storages4, and the three
Total Expected
Reference XBat KBat XSC KSC XPV KPV Cost
Weight
Autonomy
presented different scenarios for typical driving
[€]
[kg]
[km] cycles, were computed using Matlab/Simulink®,
Battery A 30 2 37 3 5 0 7374 170.48 61.6 provided with the SimPowerSystems library [14].
Battery B 30 1 37 2 5 0 5850 149.65 68.6
Battery C 8 5 37 3 5 0 8134 200.48 58.5 The results, namely the daily average energy
evolution, for the three considered scenario are
The results presented in Table 11, also notes that, presented in Figure 12, 13 and 14.
for the proposed scenario, the more feasible For Figure 12, it can conclude that the presented
option is to use batteries type B. Also, in this solutions based on the B and C-type batteries,
scenario, the use of solar energy is not a viable achieve the journey request, but the solution with
solution in terms of investment, compared to batteries A, not. It appears that the development of
other considered solutions. We conclude that the solutions B and C is very similar. It is noted that in
same configuration based on 1 bank of batteries the case of A-type batteries, taking into account
type B and 2 banks of SCS, allows both the first the weight of the designed systems, the vehicle has
and the second considered scenario. not the energy needed to do the full journey trip.
Finally, a third scenario corresponds to an extra However, if it is inserted the backup system (PV
urban utilization at the VEIL full speed that for Array), recovering the solar energy, when the
this kind of vehicle is limited to 45 km/h. vehicle is parked, this allows to do his way back.
Nevertheless, the study was done considering the
Daily Average Energy for NEDC Modified Cycle Scenario
slightly higher speed of 50 km/h, as in Figure 8. 9000

In this case, 1.5 h allows to cover a 73.59 km

8000
distance between home and work in the morning
and the same distance in the evening, to return 7000

back home (almost 149.18 km, in total). The

results are presented in the Table 12. 6000

Table 12: Evaluation of the Total Costs @ 50 km/h for 5000

Energy [W.h]

2 periods of 1.5 h
4000

Total Expected
Weight
Reference XBat KBat XSC KSC XPV KPV Cost Autonomy 3000
[kg] August's PV Energy
[k€] [km] Bat. A + SC + PV
Battery A 30 4 37 2 5 0 11.2 233.65 178.8 2000
Bat. B + SC + PV
Bat. C + SC + PV
Battery B 30 2 37 2 5 0 9.92 245.65 201.2 Bat. A + SC
Battery C 8 10 37 2 5 0 12.78 293.65 168.8 Bat. B + SC
1000 Bat. C + SC
For this scenario, as expected, it is needed greater
energy to perform the path-way plus the way 0

back. However, it is concluded that the best 0 5 10 15 20

Hour of Day
solution is still based on the batteries type B,
there is only need to insert one more bank to Figure 12: Daily Average Energy for NEDC Modified
increase the autonomy and achieve the desired Cycle Scenario.
trip. This gives us clues to the possibility, in the For ECE 15 Cycle Scenario (see Figure 13),
future, to exist modular systems of energy analyzing the curves of the daily average energies,
storages configurable by the user's request. they show that the SCs are very important in this
Therefore, the results of the proposed scenario, because the circuit has a higher number
configurations were based on 500 kg weight of of acceleration and braking. All designed
the VEIL Thus there is interest on verifying by situations reached the objectives of the scenarios
simulation if the proposed design and the outlined. The situation based on the B-type battery
influence of the weight of the storage systems shows that it is the option that offer more
does not significantly affect the performance of autonomy and allow without problems making
the vehicle. these first two scenarios.

5 Simulation
Based on the results for the different designs 4
It was considered a 3 kg weight increase for each necessary SCs
obtained through the methodology previously and PV array DC/DC converters.

EVS24 International Battery Hybrid and Fuel Cell Electric Vehicle Symposium 10
 Page 0680

Daily Average Energy for ECE 15 Cycle Scenario

9000 during the night. It is also important to compensate
the batteries self discharged when parked outside,
8000
without the possibility to plug to an electric power
7000
source, for long periods.

6000
6 Conclusions
This work presents a complete methodology to
Energy [W.h]

5000

optimize the sizing of the ESSs for an EV. As an

4000
example, it was used the ISEC EV VEIL project
3000
[7][15] from three considered driving cycles,
specified maximum speed, acceleration
2000 August's PV Energy
Bat. A + SC + PV
performance, regenerative energy and gradeability
Bat. B + SC + PV
Bat. C + SC + PV
requests. Some simulation results of multiple
1000
Bat. A + SC
Bat. B + SC
energy sources hybridization were presented,
0
Bat. C + SC
considering different ESSs and different scenarios
0 5 10 15 20
Hour of Day for the small presented EV. The emphasis of this
Figure 13: Daily Average Energy for ECE 15 Cycle work is the comparative study on the impact of
Scenario. different ESSs utilization, namely different
combinations of the considered sources, versus the
When the vehicle cruises at the maximum different scenarios for daily use. It should be
vehicle speed during the stipulated 2 periods of clearly stated that the presented results depend
1.5 hours, it can be conclude that, unless an strongly on the ESSs price, weight, energy and
increase of energy necessary to carry out these power, and may change as these characteristics
trips, the behaviour of the 3 cases studied is modify.
identical to previous scenarios, showing again, The complete methodology is fast, objective, and
the option with the B type batteries, is the most quite accurate, since the simulations obtained for
indicated. However, it is seen that in the case of the proposed designs are within the minimum
C-type batteries, it can not make the specified request results. It was also studied the possibility
journey back but that the help of the PV array is of using a backup system based on solar energy,
enough to end this trip. that may be considered in the design, or as an extra
Daily Average Energy @ 50 km/h Cte. Cycle Scenario
to cope with unforeseen routines and to minimize
the recharge of ESSs through the power network.
16000 The methodology can be improved by inclusion of
techniques for multi-objective optimization,
14000
minimizing the initial cost, the cost of the ESSs
12000
recharges, and weight, maximizing the autonomy
and life cycle of ESSs. This type of study is
10000 important for the correct sizing of the ESSs,
Energy [W.h]

depending on the intended uses of an EV, in order

8000
to maximize the autonomy and the performance
6000
with a minimum initial cost and utilization costs.
August's PV Energy
4000 Bat. A + SC + PV
Bat. B + SC + PV
Bat. C + SC + PV
References
2000 Bat. A + SC
Bat. B + SC
[1] IEA, World Energy Outlook 2004, OCDE,
Bat. C + SC
pp.58, 2004.
0
0 5 10 15 20 [2] IEA, Key World Energy Statistics, 2006
hour of Day
edition.
Figure 14: Daily Average Energy @ 50 km/h Cte.
[3] N. Künzli, R. Kaiser, S. Medina, et al.,
Cycle Scenario.
Public-health impact of outdoor and traffic-
From all presented results, it is shown that the related air pollution: a European assessment,
use of a backup system based on solar energy, The Lancet, vol. 356, Issue 9232, pp. 795-
can be a good means to recharge the batteries, 801, Sept. 2000.
allowing in some cases to deal with unexpected [4] L. Int Panis, R. Torfs, Health effects of traffic
routes and especially minimize the need of full related air pollution, European Ele-Drive
recharge of the ESSs through the power network

EVS24 International Battery Hybrid and Fuel Cell Electric Vehicle Symposium 11
 Page 0681

World Electric Vehicle Journal Vol. 3 - ISSN 2032-6653 - © 2009 AVERE

Transportation Conference (EET-2007), [15] Paulo G. Pereirinha, João P. Trovão, L.
Conf. Proc., 30th May - 1st Jun. 2007. Marques, M. Silva, J. Silvestre, F. Santos,
Advances in the Electric Vehicle Project-
[5] P.G. Pereirinha, J.C. Quadrado, J. Esteves,
VEIL Used as a Modular Platform for
Sustainable Mobility: Part II – Some
Research and Education, EVS24
Possible Solutions Using Electric and
International Battery, Hybrid and Fuel Cell
Hybrid Vehicles, 1st Inter. Conf. on
Electric Vehicle Symposium, Stavanger,
Electrical Engineering (CEE´05), Conf.
Norway, 13-16 May 2009.
Proc., Oct. 2005.
[6] R. Gremban, PHEVs: the Technical Side, Authors
European Ele-Drive Transportation
Conference (EET-2007), Conf. Proc., João Pedro F. Trovão received the
May/Jun. 2007. Electrical Engineering degree in
1999 and the MSc degree in
[7] P. G. Pereirinha, J. Trovão et al, The Electrical Engineering – Power
Electric Vehicle VEIL Project: A Modular System in 2004, from the Coimbra
Platform for Research and Education, University, Portugal. Currently he is
European Ele-Drive Transportation teaching at the Electrical Engineering
Conference (EET-2007), Conf. Proc., Department of the Engineering
May/Jun. 2007. Institute of Coimbra (DEE-ISEC)
[8] Paulo G. Pereirinha, João P. Trovão, and he is preparing his Ph.D. degree.
Comparative study of multiple energy His teaching and research interests
sources utilization in a small electric cover the areas of electric drives,
vehicle, European Ele-Drive Transportation renewable energy, power quality and
Conference (EET-2008), Conf. Proc., Mar. rotating electrical machines.
11-13, 2008.
Paulo G. Pereirinha obtained his PhD
[9] K.T. Chau, Y.S. Wong, Hybridization of in Electrical Engineering from the
energy sources in electric vehicles, Energy Coimbra University, Portugal. Full-
Convers Mgmt 42 – 9, pp. 1059-1069, time Assistant from 1995, Eq. Adjunct
2001. Professor from 1997 and Coordinator
Professor since 2009 at the Electrical
[10] R. Schupbach , J. Balda , M. Zolot and B. Engineering Department of the
Kramer, Design methodology of a combined Engineering Institute of Coimbra
battery–ultracapacitor energy storage unit (DEE-ISEC). His classes and research
for vehicle power management, Proc. IEEE interest includes electrical machines,
Power Electron. Spec. Conf. (PESC’03), electric vehicles, finite elements and
Acapulco, Mexico, 2003, p. 88. renewable energies. He is a founding
member of the Portuguese Electric
[11] Wu, Y.; Gao, H., Optimization of Fuel Cell Vehicle Association and member of its
and Supercapacitor for Fuel-Cell Electric administration board. Member of the
Vehicles, IEEE Trans. on Veh. Technol., Portuguese Electrotec. Normalisation
Vol. 55, No. 6, Nov. 2006, pp. 1748-1755. Technical Committee, CTE 69 –
[12] R. Schupbach and J. Balda The role of Electrical Systems for Electrical Road
Vehicles.
ultracapacitors in an energy storage unit
for vehicle power management, Proc. IEEE
Veh. Technol. Conf., Orlando, FL, 2003, p. H. Jorge – received his Electrical
3236. Engineering degree in 1985 and his
PhD degree in 1999, both from the
[13] W. Gao, Performance comparison of a fuel University of Coimbra. He is
cell-battery hybrid powertrain and a fuel Auxiliary Professor at the
cell–ultracapacitor hybrid powertrain, Department of Electrical Engineering
IEEE Trans. Veh. Technol., vol. 54, pp. of University of Coimbra. His
846, May 2005. teaching and research interests
[14] João P. Trovão, Paulo G. Pereirinha, include power systems, load
Fernando J. T. E. Ferreira, Comparative research, load forecast, load profile,
Study of Different Electric Machines in the power quality, power distribution,
Powertrain of a Small Electric Vehicle, 18th energy efficiency and energy storage.
International Conference on Electrical He is an IEEE member since 1992.
Machines (ICEM’08), Vilamoura, Portugal,
6-9 Sept. 2008.

EVS24 International Battery Hybrid and Fuel Cell Electric Vehicle Symposium 12