Sizing and Energy Management Strategy for

Hybrid FC/Battery Electric Vehicle

B. Bendjedia1,2,4, H. Alloui3, N. Rizoug4, M. Boukhnifer4, F. Bouchafaa2, M.E. Benbouzid5
1
Laboratory of Instrumentation, USTH Bab-Ezzouar University, Algeria
2
LACoSERE Laboratory, University of Laghouat, Algeria
3
Control Machines Laboratory, Polytechnic Militairy School, Algiers, Algeria
4
ESTACA, Rue Georges Charpak, 53061 Laval Cedex 9 – France
5
LBMS IUT de Brest, Rue de Kergoat, 29238 Brest Cedex 03– France
Abstract - This paper focalizes on sizing of hybrids pared with the case of batteries only. Also, the cost and
sources composed with Fuel cells FCs and battery pack. volume of the Hybrid Source are lower by the same
Also an experimental validation of energy management of way.
FC/Battery electric vehicle is tested. The system is com- Recent works on the FC [4], [5] leads to the slow
posed of a fuel cell system as the main source and batteries output power response because of the slow internal
as an assisting one. This last one is connected to a bidirec- electrochemical and thermodynamic characteristics.
tional DC/DC converter, and a DC/DC boost converter is Consequently, a storage system is required to satisfy
associated to the fuel cell stack. To increase the power
power demand and to compensate the difference be-
efficiency and to achieve the best performances of the
hybrid source, an online EM strategy is used to share the tween the FC output and the load [1], [6], [7].The lithi-
power between the main and the auxiliary source by de- um-ion battery has high energy density, high thermal
termining the power profile of each one. This strategy is performance and is recyclable. Several technologies
based on frequency separation; it takes into account the exist when the difference is in the composition of the
slow dynamics of FC, fuel consumption and the batteries positive electrode.
limits. Also, it is needed to make these results as a refer-
ence to be compared with other strategies which are cur- In our case, an oxidized cobalt material composes the
rently under development in our laboratory. positive electrode, and the negative electrode is prepared
In the objective to verify the efficiency of the proposed of carbon material. The electrolyte is a lithium salt in an
approach, both simulation and experimental results leads organic solvent. The promising characteristics of the Li-
to confirm its efficiency, the robustness and stability re- ion batteries include long battery life of 1000 cycles ,
grading dynamic performances during power demand, and high specific power of 300 W/kg, high specific energy
regenerative braking, fuel consumption. of 100 Wh/kg and low memory effect [8]. These bril-
liant characteristics give the Li-iON battery a high pos-
Keywords - Electric vehicle, EMS, Frequency separation,
Fuel cell.
sibility of replacing NiMH or others as next generation
batteries for vehicles. Li-ion batteries have twice energy
I. INTRODUCTION density of NiMH batteries, which are low expensive [8].
For these several raisons, this type is selected to be hy-
Reducing pollution is one of the actual challenges, bridized with the fuel cell stack in recent works [8],
when in the recent years; renewable energies have been [10].
fortified by international institutions to meet energy To control the hybrid storage system, an energy man-
saving policies. Electric Vehicles are used to reduce agement strategy (EMS) is required, to distribute the
considerably pollutants and consumption of fossil ener- demanded power between the two sources and to
gy. achieve the best performances as optimal fuel consump-
The existing battery-electric vehicles(BEV) have a lim- tion and respect sources operating limits. These perfor-
ited driving ranges (300Km). Hydrogen fuel-cell (FC) mances can enhance greatly the hybrid source lifetime.
vehicle technology is getting wide thought as a strategy Until this day, no specific technic or strategy has been
to decrease lubricates consumption and reduces destruc- capable to impose itself as the best solution in each
tive emissions [1], [2], [3]. Also, they have significantly actual situation and a considerable work has thus to be
longer driving ranges and lower refueling times than done in order to compare and evaluate different strate-
comparable BEVs, and it is also possible for them to use gies [9], [10].
the least amount of petroleum (well-to-wheels) per mile,
depending on the type of hydrogen used Several energy management strategies have been uti-
lized in this context like (not exhaustive) [10]:
In a fuel cell hybrid vehicle (FCHV), the electric power-
train is powered by a fuel cell system (FCS) and an - State machine control;
energy storage system (ESS). This energy storage sys- - Rule-based fuzzy logic;
tem can be either batteries or supercapacitors. The hy- - EMS based on classical PI controllers;
brid architecture allows defining power management - Optimal control;
strategies and offers opportunities in the powertrain - Frequency decoupling strategy.
sizing. The benefit of hybridization in term of sizing that
the hybrid source weight has a gain of 3 to 5 times com-

978-1-5090-3474-1/16/$31.00 ©2016 IEEE 2111

The last one guarantees that the fuel cell system pro- TABLE III
KOKAM CELLS SPECIFICATIONS [13]
vide low frequency demands, whereas the secondary
KOK_13HP KOK_40HE
energy sources deal with high frequency demands. This
C_elb= 13Ah C_elb=40Ah
is achieved through the low-pass filter, wavelet. This
U_elb= 3.7 V U_elb=3.7 V
strategy can improves greatly the lifetime of the fuel-cell I_elb_Dis = 104 A I_elb_Dis =40 A
system as the applied dynamic stress on the fuel system I_elb_C= 39A I_elb_C=40A
when it is prevented [10]. W_elb= 0.325Kg W_elb=0.885Kg
V_elb= 0,174 L V_elb=0.441 L
In this work, we have used the frequency decoupling Cost= 30.36 Euros Cost = 76.33 Euros
strategy to manage FC/batteries hybrid power source for
the EV. In the proposed system, the Fuel Cell presents Table 1: Vehicle characteristics[13]
the main source and the battery form the secondary one. Parameters Value
Vehicle mass(Kg) 860
The constraint imposed relates to the fuel cell power Frontal area (m2) 2.75
limitation and its dynamic response, state of charge of
Air density (Kg/m3) 1.2
the batteries. It‟s noted that in this strategy, the power
Penetration coefficient 0.3
flow of the system in different operating modes is man-
aged with the same control algorithm (without any algo- Rolling resistant coefficient 0.008
rithm commutation). 2500

Volume [L] & weight [Kg]

Vehicle Mass(Kg)
To validate the Energy Management performances, 2000 ESS Weight (Kg)
ESS Volume (L)
dynamic models of the sources and converters are de-
1500
veloped and used as simulation tools. The simulation
results are presented, which prove the efficiency of the 1000
proposed models and strategy. An experimental valida- 500
tion is proposed to confirm the proposed approaches and
the obtained results from the simulation. 0
1 2
Battery Only Hybrid Source
II. THE HYBRID SOURCE SIZING
100
In the objective to demonstrate the combination utili-
ty, two different technologies such high energy, high 80
power „Li-ion cells‟ and Fuel cell system are used for
Cost [K£]

60
sizing. Through this study, we can justify the hybridiza-
tion of fuel cell system with batteries by showing the 40
different gains (masses, volumes, costs, autonomy...).
20
This study also shows that the batteries power type is
needed in the hybridization, when the high energy bat- 0
1 2
teries can‟t supply the vehicle for long ranges without Battery Only Hybrid Source
hybridization. Figure 1: Sizing results of hybrid source and battery only
The Energy Management Strategy gives us the maxi-
mum and minimum power and energy for each compo- Figures (1) presents the size of the hybrid source com-
nent. According to these energy curves, three parameters posed by FC and KOKAM_HP_13 Li-ion technology
can influence the sizing of the ESS: the maximum con- and the mono_source which consists of 40HE_KOKAM
sumed power (Pbat_cons, PFC_cons), the recovered power Figure 1 shows that the VEHICLE (RED) is very im-
(Pbat_rec), and the maximum energy (Ebat_cons, EFC_cons). portant in case of supplying only with batteries but for
The parameters used in the fuel cell sizing system are: the hybrid source, it is shown that there is a gain of 35%
- Weight density; storage density and volumetric density; when the battery of power KOKAM HP can contribute
- The mass of hydrogen consumption from the needed as an auxiliary source. on the contrary, the volume of the
energy consumption. ESS is almost at the same range for the two categories.
 Vehicle parameters The weight of HP Kokam associated to the fuel cell
increases the benefits of hybridization regarding the
The estimation of the needed power and energy for weight (blue) when the batteries source is heavier by
the proposed profile, a dynamic model of the vehicle is 200%.
simulated using Matlab-Simulink software. In the other hand the cos is reduced in the case of hybrid
source to the half, these results are expected for this
To size the ESS of the vehicle in real conditions, the
cycles range regarding the limitation of batteries as
urban and road Artemis cycles (Figure.1), with an aver-
primary source (energy), and the dynamic limitation of
age slope of 2.5% are chosen. This mixed cycle is re-
fuel cell systems.
peated 30 times to accomplish the autonomy of about
700km.

2112
III. FC/BATTERY SYSTEM DESCRIPTION AND MODELING 0.5

concentration loss [V]

0.4
As shown in Figure 2, two DC/DC converters are used 0.3
to link the sources with DC bus link. A boost converter 0.2

is connected to the fuel cell system, another bidirection- 0.1

al DC/DC converter is associated to the reversible 0

0 0.2 0.4 0.6 0.8 1 1.2
source. It operates in Boost mode when the secondary Current Density (A/cm2)
(B)
source provides current to the load and in Buck mode 0.35

when current resulting from braking and/or from the fuel 0.3

ohmic losses (V)

cell is received to the stock. 0.25

0.2

0.15

0.1

0.05

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Current Density (A/cm2)
(c)
1.4

1.2

Cell voltage [V]

1

0.8

0.6

0.4

0.2

Figure. 2 Hybrid fuel cell/battery system scheme 0

0.2 0.4 0.6 0.8 1 1.2 1.4
Current density [A/Cm2]

- FUEL CELL MODEL (D)

Figure. 3 (A)-Activation losses, (B)-Concentration losses, (C)-Ohmic
Fuel cell systems are widely used as a static-state losses, (D) Cell Voltage
models which are classically used for component sizing, R(Ω/cm2) α I0 K α1
developing management strategies [11]. In our study
several components as compressor, heat exchanger and 0.19 0.5 10-6.912 1.1 0.085 1.4
manifold are considered in steady state to simplify the
Table. I Fuel cell model parameters
energy management task.
- BATTERY MODEL
The fuel cell voltage is a function of reactant partial
It is mentioned on [12] that the battery internal re-
pressure, fuel cell temperature, cathode pressure, fuel
sistance is supposed constant during the different modes
cell temperature, and membrane humidity. The fuel cell
as charge and discharge cycles and does not vary with
voltage is calculated by deducting the fuel cell losses
the state of charge or others. In this work, we have used
from the fuel cell open circuit voltage E, that is:
the same battery model when we introduce a simple
Vcell= ENenst - Vcons- Vact - Vhom (1) modification on the resistance which will be in relation
with : at least with the state of charge of the device.
The voltage battery is termed by: [12]:
[ ]

( )
where:
: the battery voltage (V)
: battery open voltage (V)
( ) (5) 0: battery resistance (Ω)
: battery current (A)
The figure 3 presents the cell voltage versus the K : polarization constant (V/Ah)
current density j. The fuel cell losses are attributed to Q : capacity of the battery (Ah)
three categories: the activation loss, Ohmic loss and :∫ :is the actual charge of battery (Ah)
concentration loss. Fuel cells are connected in series to A : exponential zone amplitude (V)
form the FC stack, then the total voltage of the stack is B : exponential zone time constant inverse
Vst=Vcell*Nss and the stack power is Pst=Vst*Ist. Nss is The variable resistance expression is validated exper-
the number of series cells. imentally:
R= Rinitial*(5.128 - 0.265*Soc + 67*10-4*Soc2 - 7.9*10-5 *Soc3
0.22
activation losses [V]

0.21

0.2 + 4.18*10-7*Soc4 - 7.51*10-10*Soc5)

0.19
(7)
- Rinitial is resistance fully charged
0.18

0.17

0.16
0 0.2 0.4 0.6 0.8 1 1.2 1.4
- Soc: State Of Charge
Current Density (A/cm2)

(A)

2113
The Figure 4 shows the simulation results of charging The behavior of both converters is described by the
and discharging characteristics curves (voltage, internal following equations system:
resistance and SOC) under various current profiles.
(8)
Q(A/h) Vbat(v) R(Ω) K(Ω) A(v) B(A/h) (9)

2.3 3.366 0.01 0.0076 0.26 26.5484 ( )

Table.2 Li-Ion Battery parameters [12] If the battery converter operates in "Buck" mode,the
2.5 model of both converters will be defined by the follow-
2
ing equations:
Battery current [A]

1.5

0.5

-0.5

-1

-1.5

-2
0 500 1000 1500 2000 2500 3000 3500
Time [S]
(A) ( )
0.09
Internal resistance[Ohm]

0.08

0.07
where:
0.06

0.05
- VFC: FC Voltage,
0.04

0.03
- iF: FC current,
0.02 - eB : Battery voltage ,
0.01
0 500 1000 1500 2000
Time [S]
2500 3000 3500
- iB : Battery current,
(B) - rB : Battery Resistance,
100
- VDC: DC bus voltage,
State Of Cahrge [%]

95

90 - D1,2,3: Control signals of TFC, B

- RL,LL: Resistance and inductance Load.
85

80

75

70 VI. FREQUENCY SEPARATION ENERGY MANAGEMENT

STRATEGY:
65

60
0 500 1000 1500 2000 2500 3000 3500
Time[S]
(C) The objective of this strategy is to define the mission
6
of each source based on a frequency vision. Its principle
Battery voltage [V]

5
is illustrated in Figure. 8, a low pass filter is used to
4

3
filter the required power to propel the EV (PV). The part
2 made of low frequencies will be provided by the fuel
1 cell (PFC), and the harmonic high frequency portion is
0
0 500 1000 1500
Time[S]
2000 2500 3000 3500 returned to the battery pack (Pbat). So the cutoff frequen-
(D) cy value that determines the mission of each source.
Figure. 4 Dynamic caracterestics of a 2.3Ah, 3.3V Li-Ion battery This value is related to the characteristics of the storage
(A)-Battery current, (B)-Internal resistance, (C)-State Of Charge, (D)
battery Cell Voltage
system and is expressed as a function of the specific
A critical test for the dynamic battery model is to switch power (dP) and energy density (ED).
between tow charge and discharge modes in order to It is mentioned in recent works that it is crucial to im-
verify the model‟s continuity. Figure 4 shows four dis- pose to the battery a zero energy balance during a cycle
charge-charge cycles for the Li-Ion battery. [10]. This is reasoned by, that this is benefic to the stor-
This figures show that the battery current affect di- age system life time and the fuel consumption.
rectly the State Of Charge (SOC), resistance and the
voltage battery characteristics. For this reason, it‟s nec-
essary to take into account this point in the energy man-
agement strategy and in the global system control to
avoid eventual battery damaging.
The modeled sources present dissimilar behaviors and
dynamic characteristics, for this reason each source is
dedicated for a special running mode of the vehicle -
(starting, braking, and constants speeds). +
- MODEL OF DC/DC GLOBAL SYSTEM
The Modeling of the DC/DC converters "Boost and
Figure. 5 EMS basic principle scheme
Buck-Boost" is a crucial task to simulate the system.

2114
VII. SIMULATION RESULTS 120
Speed achieved by the vehicle
Speed requedted by the vehicle

In this work, the vehicle speed profile used is a New 100

European Driving Cycle (NEDC) cycle. The vehicle and 80

Speed(kph)
others simulation parameters are summarized on Ta- 60

bles.3 and 4 which define the different parameters used 40

in simulation. 20

Vehicle model parameters Value 0

Vehicle total mass M 1400 kg
0 200 400 600
Time(s)
800 1000 1200

Frontal area S 2 Figure. 6 NEDC profile with the measured vehicle speed.
2m 4
x 10
Drag coefficient Cd 0.335 4
Vehicle power
Coefficient of rolling friction fr 0.009 3 FC power
Battery power
Air density ρ 1.2 kg.m3 2

Gravity g 9.8 m.s 2 1

Powers(w)
TABLE 3. VEHICLE MODEL SPECIFICATIONS. 0

Parameter Value -1

Motor Traction power 30 kW -2

DC bus voltage 400 V -3

Fuel Cell power 40 kW -4

0 200 400 600
Time(s)
800 1000 1200

Battery power 30 kW Figure. 7 Power response of the sources

TABLE 4. THE POWERTRAIN COMPONENTS PARAMETERS.
450

In the objective of the implementation, a control 400

scheme endorsed in [8] is used, when a PI regulators are 350

300
used to control the DC bus (VDC) voltage, FC and batter-
VDC(v)
Tension (V)

250
ies currents. the PWM strategy is used to send the con- 200

trol pulses to the IGBT of the boost chopper and bidi- 150

rectional converter. 100

According to figure 6, it is shown that the speed ref- 50

0
erence is well followed, this is can be explained by that, 0 200 400 600
Temps (s)
800 1000 1200

the sum of the sources powers (Fuel Cell and provided Figure. 8 DC bus Voltage
by the pack of the batteries or absorbed) ensures the Two DC motors are used to emulate the propulsion
required traction power with respecting the constraints and the regenerative phases of the electric vehicle. These
imposed by source‟s sizes. motors are controlled by a separate control loop and
From the power profile figure, it is clear that the sum calculator (PCI card), this is to not perturb the hybrid
of the powers provided by the FC and/or recovered by source operation.
the battery ensures the required vehicle traction power Because we have a problem of hydrogen availability
with respecting the constraints imposed on power in our laboratory, an FC emulator is used in this work.
sources. It‟s a Buck chopper controlled with a voltage loop, its
In the power flow figure, it‟s remarkable that there are task is to reproduce the same FC characteristic (voltage-
some phases, the FC supplies the load fully at steady current). The chopper output current is measured and
states. In other, the load is supplied by both FC and the used in the FC model to obtain the voltage reference
batteries, because the primary power source (green) is using Matlab/Simulink and adapted PCI Card. The
lower than then load power (blue). In this time the bat- chopper control signal (pulses) is obtained by regulation
tery current (red) (Figure 7), becomes positive to ensure of fc voltage voltage.
the required power to the load. For the second source, Lead-Acid batteries
From the DC bus regulation figure, it‟s notable that (12V/150Ah) are available in the laboratory and are
the Vdc is controlled at the desired voltage (400V), and used in the test bench as shown in the figure 9.
it‟s affected slowly at each load variation (Figure 8), we From the figure 11, it is clear that FC supplies the
can improve this problem by using robust controller for load totally in the steady state phases, and sometimes
DC bus control out loop. provides the batteries, when its power exceeds the re-
VIII. EXPERIMENTAL RESULTS quired load power.
To confirm the obtained results from numerical simu- The storage system (batteries) contributes in the sys-
lation, an experimental test bench is used to evaluate the tem stability by recovering the difference of power at
performance of the energy management strategy in a each transient period (acceleration and decelerating).
Hardwar In the Loop approach as close to the real oper-
ating condition.

2115
10
bridization‟) weight is more important compared with
2 the hybrid source. The HP technology can enhance
1
3 greatly the hybrid ESS weight and the cost in the same
5
time.
4 6
The separation frequency EMS is used, when the op-
timal cutoff frequency is chosen based on studying its
effect on the dynamic and other performances of the
global system. Its effectiveness and robustness is con-
firmed firstly, by the simulation results. As a result, the
batteries pack can reduce powerfully the imposed power
stress on the main source and assist it correctly using the
7
proposed EMS. An FC emulator is presented and used,
the validation of the proposed approaches is achieved
8
successfully and the simulation results are confirmed.
X. REFERENCES
9 [1] P.Thounthong, S.Pierfederici, J-P.Martin, M.Hinaje and B.Davat
“ Modeling and Control of Fuel Cell/Super capacitor Hybrid
Source Based on Differential Flatness Control „, IEEE Trans.
Veh. Techno, vol. 59, no. 6, july 2010
Figure. 9 The experimental test bench [2] B. Blunier, M. Pucci, G. Cirrincione, M. Cirrincione, and A.
- 1- PCI6052E card, 2- FC emulator, 3- DC vehicle mo- Miraoui,“A scroll compressor with a high-performance sensor-
less induction motordrive for the air management of a PEMFC
tor, 4- DC Load emulator motor, 5- DC/DC load emula- system for automotive applications”, IEEE Trans. Veh. Technol.,
tor reversible chopper, 6- DC/DC boost and bidirection- vol. 57, no. 6, pp. 3413–3427,Novomber. 2008.
al converters, 7-RL load 8- Lead-Acid batteries [3] P. Thounthong, V. Chunkag, P. Sethakul, B. Davat, and M.
12V/150Ah, 9- speed sensor (dynamo), 10- measure- Hinaje, “Comparative study of fuel-cell vehicle hybridization
with battery or supercapacitor storage device,” IEEE Trans. Veh.
ment devices. Technol., vol. 58, no. 8, pp. 3892–3904, October. 2009.
[4] N. Yousfi-Steiner, P. Moçotéguy, D. Candusso, and D. Hissel,
“A review on polymer electrolyte membrane fuel cell catalyst
degradation and starvation issues: Causes, consequences and di-
agnostic for mitigation,” J. Power Sources, vol. 194, no. 1, pp.
130–145, October. 2009.
[5] P. Thounthong, B. Davat, S. Raël, and P. Sethakul, “Fuel starva-
tion: Analysis of a PEM fuel cell system,” IEEE Ind. Appl. Mag.,
vol. 15, no. 4, pp. 52–59, July/August. 2009.
[6] D. Feroldi,M. Serra, and J. Riera, “Design and analysis of fuel-
cell hybrid systems oriented to automotive applications,” IEEE
Trans. Veh. Technol., vol. 58, no. 9, pp. 4720–4729, Nov. 2009.
[7] J. Bernard, S. Delprat, F. N. Büchi, and T. M. Guerra, “Fuel-cell
Figure. 10 NEDC profile hybrid powertrain: Toward minimization of hydrogen consump-
(bottom :UDC and top :EUDC) tion,” IEEE Trans. Veh. Technol., vol. 58, no. 7, pp. 3168–3176,
September. 2009.
[8] H. Alloui, M. Becherif, K. Marouani, "Modelling and Frequency
Separation Energy Management of Fuel Cell-Battery Hybrid
Sources System for Hybrid Electric Vehicle", 21st IEEE Medi-
terranean Conference on Control & Automation (MED), June
2013.
[9] S. Caux, W. Hankache, M. Fadel, D. Hissel,”On-line fuzzy
energy management for hybrid fuel cell systems”, International
journal of hydrogen energy , International Journal of Hydrogen
Energy vol. 35 (5) p. 2134-2143, (2010).
[10] Souleman Njoya Motapon, , Louis-A. Dessaint, and Kamal Al-
Haddad, A Comparative Study of Energy Management Schemes
for a Fuel-Cell Hybrid Emergency Power System of More-
Electric Aircraft IEEE transactions on industrial electronics, vol.
Figure. 11 Experimental results of demanded current (yellow), 61, no. 3, march 2014
[11] Spiegel, Colleen. “PEM fuel cell modeling and simulation using
batteries pack current (violet), Fuel Cell current (green) and D MATLAB”. Academic press, 2011.
motor mechanical speed (bleu) [12] Olivier Tremblay, Louis-A. Dessaint,”Experimental Validation
of a Battery Dynamic Model for EV Applications”, AVERE,
IX. CONCLUSION World Electric Vehicle Journal Vol. 3 - ISSN 2032-6653, 2009
In this paper, a frequency splitting strategy is used to [13] A. . Hammani, R. . Sadoun, N. . Rizoug, P. . Bartholomeüs, B. .
Barbedette, and P. . Le Moigne, “Influence of the management
manage the powers in the objective of a hybrid source strategies on the sizing of hybrid supply composed with battery
sizing (composed with fuel cell and battery) which en- and supercapacitor,” in 2012 1st International Conference on Re-
sure the supplying of the electric vehicle. The obtained newable Energies and Vehicular Technology, REVET 2012,
results confirm that the HE batteries („only with no hy- 2012, pp. 1–7

2116

Powered by TCPDF (www.tcpdf.org)