Energy management and Control of a fuel

cell/supercapacitor multi-source system for electric

vehicles
Imad Eddine Aiteur, Cristina Vlad, Emmanuel Godoy

To cite this version:

Imad Eddine Aiteur, Cristina Vlad, Emmanuel Godoy. Energy management and Control of a fuel
cell/supercapacitor multi-source system for electric vehicles. 19th International Conference on Sys-
tem Theory, Control and Computing (ICSTCC), Oct 2015, Cheile Gradistei, Romania. �10.1109/ic-
stcc.2015.7321392�. �hal-01257838v2�

HAL Id: hal-01257838

https://centralesupelec.hal.science/hal-01257838v2
Submitted on 8 Apr 2020

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 Energy management and control of a fuel cell/supercapacitor
multi-source system for electric vehicles
Imad Eddine Aiteur, Cristina Vlad and Emmanuel Godoy
Laboratoire des Signaux et Systèmes (L2S, UMR CNRS 8506)
CentraleSupélec-CNRS-Université Paris-Sud
Automatic Control Department
3, rue Joliot Curie
91192, Gif-sur-Yvette cedex, France
E-mail: {imad-eddine.aiteur, cristina.vlad, emmanuel.godoy}@centralesupelec.fr

Abstract—This article presents a control strategy of a multi- electric motor and to maximize the global system efficiency
source system dedicated to automotive applications, aiming to by minimizing the hydrogen mass consumption.
maximize the global system efficiency. A parallel-architecture of In the literature, different control strategies have been con-
the multi-source system is considered, composed of a fuel cell
as a primary power source, a supercapacitor as a secondary sidered for power supply systems of electric/hybrid vehicles
storage element, which provides peak power in fast transients, powertrains. Passivity-based control has been employed to reg-
and power converters for the power sources interconnection to ulate the DC bus and the secondary storage element voltages of
the DC bus. The power management is realized as a two-level different multi-source systems (FC/SC [3], [4] or FC/battery
control structure. The first level contains inner control loops for [5] configurations). Lyapunov-based control [6] and sliding
fuel cell/supercapacitor currents and a DC bus voltage control
loop, designed using PI controllers. The higher level consists of an mode control [7] have also been applied to power supply
energy management supervision system based on the equivalent systems for electric vehicles, using one or two additional
consumption minimization strategy. The fuel cell power demand storage elements (FC/SC, FC/SC/battery). In addition to this,
is computed by minimizing the hydrogen mass consumption with cascade control using PI-controllers has been considered, with
respect to system physical constraints. The performances of the inner loops for current regulation of power electronic devices
proposed control structure are evaluated in simulation, in terms
of fuel economy, using an urban driving cycle. and outer loops for voltage regulation of the secondary source
Index terms— energy management, supervision system, (SC or battery) [8] and of the DC bus. Also, decentralized
ECMS, real-time multi-level control strategy, multi-source model predictive control [9], [10], and singular perturbation
system, fuel cell, supercapacitor, automotive applications. control approach [11] have been used for a FC/SC supply
system of an electric vehicle.
I. INTRODUCTION The purpose of the supervision level is to manage the
According to the latest studies, the carbon dioxide (CO2 ) power split between the FC and the additional storage/supply
emissions have a major contribution to the greenhouse effect elements while reducing the fuel consumption and maximizing
due to fossil energies combustion, mostly in the transport field, the system efficiency. Different energy management strate-
where an important quantity of greenhouse gas emissions is gies (ruled-based [12] or optimization-based [13]) have been
produced [1]. For environment preservation, it is necessary to applied and validated using a static model of the fuel cell.
explore ecological means to produce energy such as renew- For optimization-based strategies, the solution is obtained by
able or electric energy sources. Therefore, in the automotive minimizing an objective function subject to system physical
field, multi-source systems have been considered for electric limits, integrated as constraints. Thus, the supervision level of
vehicles powertrains. an electric vehicle multi-source power system has to compute
The use of a fuel cell (FC) as main power source provides the fuel cell power which minimizes the hydrogen mass
an interesting solution, although its slow dynamics demands to consumption and satisfies the physical constraints: positive FC
be assisted by a secondary supply/storage element, such as a power, limited variation of the FC power, maximum capacity
supercapacitor (SC) or a battery, which is able to produce peak of the secondary storage element.
power demand during fast transients and to recover energy The optimization problem can be solved on-line or off-
during the braking phase. Generally, several architectures have line based on the knowledge of the driving cycle. Off-line
been investigated as it concerns multiple sources interconnec- optimization methods (dynamic programming [14]-[15], Pon-
tion to the DC bus: series, cascade or parallel architecture [2]. triagyn’s minimum principle [16]) use the assumption of full
In this article, the control strategy is elaborated for a parallel knowledge of the driving cycle. However, the solution opti-
FC/SC multi-source system with both sources connected to mality depends on the employed method and its complexity
the DC bus through DC/DC power converters. The proposed can be an issue for real-time implementation as it requires a
control approach has the following objectives: to stabilize the high-capacity storage. The advantage of on-line methods [17],
multi-source power supply system, to regulate the DC bus [18] is the reduced computational time, whereas the solution
voltage in order to guarantee a stable supply voltage for an obtained using the instantaneous power load is sub-optimal.
Equivalent consumption minimization strategy (ECMS) [19], A. Energy management model
[20] and model predictive control [21] are among the on-line The fuel cell is an electrochemical device that produces
approaches that have been used for power management of electrical energy, water and heat as a result of the redox
electric or hybrid vehicle powertrain. reaction between hydrogen and oxygen [24]. Generally, the
This article proposes a control structure that stabilizes the FC is part of a fuel cell system composed by: stack, air
DC bus voltage and uses the information delivered by an compressor, fuel tank and humidifier [25]. In automotive appli-
energy management supervision system to regulate the FC cations, the stack is usually a polymer electrolyte membrane
power, which ensures the fuel consumption minimization over fuel cell (PEMFC), mainly for its convenient characteristics
the whole driving cycle. Thus, the model of the FC/SC multi- (high power density, small volume).
source system is described in Section II. Section III presents The fuel cell static model is given by the static output
the two-level control structure and the ECMS optimization voltage (vf c ) - current (if c ) characteristics, presented in Fig.
problem formulation. In Section IV, simulation results are pre- 2. This characteristics is obtained by interpolation of different
sented in order to evaluate the performances of the proposed measured values of vf c and if c and by approximation with a
control structure and the improvement of the hydrogen mass fifth order polynomial:
consumption due to the use of the supervision level. Finally,
Section V presents conclusions and perspectives of this work. vf c = a5 i5f c + a4 i4f c + +a3 i3f c + a2 i2f c + a1 if c + a0 (1)
−8 −5
where: a5 = −9.01 · 10 , a4 = 1.98 · 10 , a3 = −0.0016,
II. P OWER SUPPLY SYSTEM MODELING
a2 = 0.0617, a1 = −1.17, a0 = 41.32.
A parallel architecture has been considered for the FC/SC Based on the previous static characteristics and on the
multi-source system, which presents several advantages com- hydrogen flow expression:
pared to the series or cascade architectures: more effective Ncell MH2
power distribution, high efficiency and DC bus voltage regu- φH2 = if c (Pf c ) (2)
ne−1 F
lation [2]. The fuel cell is connected to the DC bus through
a step-up (boost) DC/DC converter, that transforms the power with Pf c - the power produced by the FC [kW], Ncell - the
supplied by the FC into DC bus power [22]. A bi-directional cells number, MH2 - the hydrogen molar mass [gram · mol−1 ],
DC/DC power converter is used to interconnect the superca- F - Faraday constant [C · mol−1 ], if c - the FC current [A],
pacitor, which assists the SC charge/discharge processes when the instantaneous hydrogen flow values have been obtained
energy recovery or peak power is required [23]. The multi- for a FC power in the range of [0, 2.5] kW. Fig. 3 shows the
source system architecture is illustrated in Fig. 1. static φH2 − Pf c characteristics obtained using a second order
Thereafter, two models are presented: the static model of polynomial approximation:
the fuel cell and the supercapacitor energy dynamics, used to φH2 = b2 Pf2c + b1 Pf c + b0 (3)
elaborate the energy management strategy of the supervision
system, and the system non-linear model employed for control where: b2 = 0.0018, b1 = 0.0134, b0 = 0.0001. The hydrogen
design. flow expression given in equation (3) is used in Section III
In this study, the proposed two-level control structure is to define an optimization problem, solved by the supervision
evaluated using a static model of the fuel cell. Indeed, for system at each sampling time for fuel consumption reduction.
practical implementation of this control structure, the energy Another objective of the energy management strategy is to
management strategy and the control design should consider assure a desired SC charge level at the end of the driving
the dynamic model of the fuel cell. cycle.

Pfc,ref
Lfc ifc vbus
Pload
Cbus
v fc [v]

vFC FC a1

Step-up DC/DC (hb)

Psc,ref ifc [A]

Lsc isc
Fig. 2. Polynomial approximation of the static vf c − if c characteristics of
vSC Csc a2 the fuel cell.

Bi-directionnel DC/DC (hrp) Therefore, the scaled SC state of energy is defined as:
SoE(t) = EEsc,max
sc (t)
, where Esc (t) and Esc,max are the instan-
Fig. 1. Fuel cell/supercapacitor architecture taneous and the maximum energy stored in the SC respectively.
In addition to this, the multi-source system and the load are
Hydrogen Flow [gram.s-1]
with: M - the vehicle mass, V - the vehicle velocity, S - the
frontal surface, Cx - the drag coefficient, α - the road slope,
ρair - the air density, g - the gravity acceleration, Cr - the
vehicle rolling resisting coefficient.
In the literature, there exist different types of driving cycles
which can be divided on standard profile or real-data driving
cycles [1]. In this study, simulations of the proposed control
Pfc [kW] strategy are done using Artemis cycle (real-world urban driv-
ing cycle) [27], which uses real measurements collected from
vehicles in real driving conditions.
Fig. 3. Hydrogen flow φH2 polynomial approximation.
III. E NERGY MANAGEMENT CONTROL STRATEGY
The proposed two-level control structure is presented in Fig.
interconnected to the DC bus via a capacity Cbus [F]. Thus, 4. The high level includes an energy management supervi-
the system power balance equation is the following: sion system. This system computes at each sampling time,
Pload (t) + PCbus (t) = ηb Pf c (t) + ηrp Psc (t) (4) the fuel cell and supercapacitor optimal powers (Pf c,optimal ,
Psc,optimal ) using an optimization strategy that minimizes the
with: ηb , ηrp - the efficiencies of DC-DC step-up and bi- hydrogen mass consumption, with the aim of maximizing
directional power converters respectively, Pload - the power the system global performance. The supervisor takes into
demand, Psc - the SC power, PCbus - the Cbus power. consideration the DC bus power variations during transients
The dynamics of the SC state of energy is defined in and the evolution of SC state of energy (SoE) that must
function of the power demand, FC and DC bus powers as reach the desired reference value (SoEref ) at the end of the
follows: driving cycle. The reference FC current value (if c,ref ) is then
dSoE Psc (t) Pload (t) + PCbus (t) − ηb Pf c (t) determined at each instant by using equation (2) and the poly-
(t) = − =−
dt Esc,max ηrp Esc,max nomial approximation of the hydrogen flow φH2 introduced in
(5) equation (3). At low level, the two converters are controlled
by fast inner current loops using classic PI controllers and
B. Control model
PWM (pulse wave modulation) generators. The inner loops are
Let us consider the state-space vector x(t) = designed to regulate the FC and SC currents (if c and isc ) to

0
if c (t) isc (t) vsc (t)
vbus(t) and the input vector their reference values if c,ref and isc,ref respectively. The bi-
u = α1 α2 vf c , with if c , isc - the fuel cell and directional converter has to guarantee that the DC bus voltage
supercapacitor currents, vf c , vsc , vbus - the FC, SC and DC (vbus ) is stabilized at the reference value (Vbus,ref ). DC bus
bus voltages, α1 , α2 - the duty cycles of the step-up and voltage regulation is done using a PI controller that generates
bidirectional power converters respectively. the reference SC current value (isc,ref ).
The fuel cell/supercapacitor power system is represented by
A. Equivalent consumption management strategy (ECMS)
the following 4th order non-linear model:
 di The energy management strategy is employed to ensure an
fc 1
dt = Lf c (vf c − (1 − α1 )vbus ) optimal power split between the fuel cell and the supercapac-


 disc
 1
dt = Lsc (vsc − (1 − α2 )vbus ) (6)
itor, by minimizing the fuel consumption while assuring the
dvsc
= − C1sc isc same charge level of the SC, at the beginning and at the end
 dvdtbus


Pload of the driving cycle. The approach adopted in this study is the
 1
dt = Cbus ((1 − α1 )if c + (1 − α2 )isc − vbus
equivalent consumption minimization strategy (ECMS), based
where Lf c , Lsc are the step-up and bi-directional converters on the concept of equivalent fuel consumption [28]. Therefore,
inductances and Csc , Cbus are the SC and DC bus capacities. the energy stored in the supercapacitor is transformed into fuel
The duty cycles α1 , α2 represent the control inputs to be consumption using a conversion positive factor λ [gram · J−1 ].
computed while vf c is determined at each instant using the Besides, ECMS represents an on-line optimization strategy,
polynomial approximation of the static vf c −if c characteristics that relies only on the instantaneous power demand. For this
of the fuel cell given in equation (1) and represented in Fig. reason, this energy management strategy can easily be applied
2. to real-time applications due to its low computation time.
C. Vehicle model Using the power balance equation (4) and considering:
The power demand is determined using the vehicle dynam- 0 PCbus (t)
Psc = Psc (t) − (8)
ics [26], which depends on the velocity variation, the losses ηrp
due to rolling and to aerodynamic frictions, and gravity:
where PCbus (t) = Cbus vbus dvdtbus
, the cost function to be
dV minimized is represented by the total fuel consumption γeq (t):
Pload = V [0.5ρair V 2 SCx x+M (g sin(α)+ +gCr cos(α))]
dt 0
(7) γeq (t) = φH2 (t) + λPsc (t) (9)
Pload To satisfy the imposed damping factors ξif c , ξisc and the
SoEref
PC Energy management strategy (ECMS) response time of each loop, two PI-controllers are designed
bus
Equation (1)
with HP Ii = ki (1 + τ1i s ), i = if c , isc . These controllers are
Pfc,optimal
vfc used for FC and SC current regulation respectively [11].
ifc,ref
+-
a1 Hence, the closed loops transfer functions are given as
PI ifc
follows:
FC/SC 1 + τif c s
isc,ref System Hi f c = ξif c
(14)
vbus,ref
PI isc*
a2 1 + 2 ωi s + ( ωi1 s)2
+- +- PI fc fc

vbus isc 1 + τisc s

Hisc = ξ
(15)
1+ 2 ωiisc s + ( ωi1 s)2
sc sc

voltage loop current loops 2ξif c ωi2 Lf c 2ξisc

fc
with τif c = ωif c , kif c = τif c Vbus , τisc = ωisc , kisc =
Fig. 4. Energy management control strategy. ωi2sc Lsc
τisc Vbus and ωif c , ωisc are the desired natural frequencies
for each loop.
By replacing the hydrogen flow expression (equation (3))
in the cost function equation (9), the following optimization C. Voltage loop
problem must be solved in order to compute the instantaneous Considering that the inner current loops are much faster than
0
power Psc (t): the voltage loop, the DC bus voltage regulation can be done
0 0 0 using isc,ref as control signal. Therefore, it can be considered
min
0
β2 Psc2 (t) + β1 Psc (t) + β0 + λ(k)Psc (t) (10) that the SC current has already reached its steady date value
Psc (t)
∗
2
(isc,ref ). By defining the virtual control variable Isc = (1 −
ηrp ηrp b2
with: β2 = b2 , β1 = − ηb (2 ηb Pload + b1 ) and β0 = α2 )Isc,ref [3], the DC bus voltage dynamics is expressed as:
ηb2
b2 2 b1
P
ηb2 load
+ ηb Pload
+ b0 .
Once the solution of the optimization problem is found, the 1 ∗ Pload
Vbus (s) = ((1 − α1 )If c (s) + Isc (s) − ) (16)
instantaneous optimal values Psc,optimal (t) and Pf c,optimal (t) Cbus s Vbus
are determined using equations (8) and (4). A PI-controller, HP I,vbus = kvbus (1 + τv 1 s ), is designed to
Then, the supervision system verifies a posteriori that the bus
regulate Vbus to its reference value Vbus,ref . Then, the closed
optimal power values respect the system physical limits. First loop transfer function has the following form:
of all, powers supplied by the fuel cell and the supercapac-
itor are bounded inside the intervals [Pf c,min , Pf c,max ] and Vbus 1 + τvbus s
Hvbus = = ξ
(17)
[Psc,min , Psc,max ] respectively, where Pf c,min (≥ 0), Pf c,max , Vbus,ref 1+ 2 ωvvbus s + ( ωv1 s)2
bus bus
Psc,min , Psc,max are the minimum and maximum values of
2ξvbus
each power sources. Secondly, the scaled SC state of energy where τvbus = , kvbus =
ωvbus Cbus τvbus ωv2bus ,
ξvbus is the
(SoE) is limited by its minimum and maximum values: damping factor and ωvbus represents the closed-loop natural
SoEmin ≤ SoE(k) ≤ SoEmax (11) frequency.

The optimization problem (10) is solved using a con- IV. SIMULATION RESULTS
version factor defined as a proportional controller λ(t) = The two-level control structure is applied to a FC/SC multi-
s0 + s1 (SoEref − SoE(t))) [29], where s0 and s1 represent source system of an electric vehicle, and tested in simulation
equivalence factors. s0 and s1 are chosen in order to guarantee using the Artemis driving cycle shown in Fig 5. Vehicle
the same SC state of energy at the beginning and at the end parameters are shown in Table I, while the multi-source system
of the driving cycle (SoE(Tcycle ) = SoE(T0 )). Their values and the DC bus parameters are given in Table II.
influence the SC charge and discharge processes over the
driving cycle and their tuning has to be adapted to driving TABLE I
conditions (road slope, acceleration and braking phases). V EHICLE PARAMETERS

B. Current loops M [kg] S [m2 ] ρair [kg · m−3 ] Cx Cr α[◦ ]

700 1.8 1.225 0.3 0.01 0
Using the dynamic equations given in (6), the fuel cell and
supercapacitor currents are:
The parameters of the PI controllers are tuned such that
Vbus(s) Vbus (s) − Vf c (s) the current control loops are at least 10 times faster than
If c (s) = α1 + (12)
Lf c s Lf c s the DC voltage control loop. Thus, for DC bus voltage
Vbus (s) Vbus (s) − Vsc (s) regulation, the currents have already reached their steady
Isc (s) = α2 + (13) state values. The following closed-loop time responses are
Lsc s Lsc s
Velocity [km.h ]
DC bus voltage evolution is illustrated in Fig. 9. The
-1

reference value of 50 V is maintained over the whole cycle,

with voltage peaks during load variations. In Fig. 10, the power
supplied by the FC/SC multi-source system is compared to the
load power. The power demand is delivered in 20 ms with a
maximum overshoot of 9%.
5
Pload [kW]

-5

SoE
Time [s]

Fig. 5. Artemis driving cycle

TABLE II
M ULTI - SOURCE SYSTEM AND DC BUS PARAMETERS Time [s]
Multi-source system DC bus Fig. 6. Supercapacitor state of energy (SoE)
Lf c = 2 · 10−4 [H] ηb = 0.95 Cbus = 1.2 · 10−3 [F]
Lsc = 1 · 10−4 [H] ηrp = 1 vbus,ref = 50 [V]
Csc = 130 [F] SoEmax = 1
Pf c,max = 3 [kW] SoEmin = 0.2
Pf c,min = 0 [kW] Esc,max = 20 [kJ]
11.4

11
ifc [A]

10.6

used: tres,if c = tres,isc = 0.5 ms, tres,vbus = 10 ms and

47 47.0014
the damping factors are chosen such that the control loops
are critically damped (ξi = 0.7, i = if c , isc , vbus ). There-
fore, the PI-controllers parameters that ensure the previous
specifications are: kif c = 5.83 · 10−2 , kisc = 2.92 · 10−2 ,
kvbus = 1.03 · 10−4 , τif c = 1.4 · 10−4 , τisc = 1.4 · 10−4 , Time [s]
τvbus = 0.0025.
ECMS algorithm is applied using a sampling time of 1 Fig. 7. Fuel cell current (solid line), fuel cell reference current (dash-dot
line).
s. The equivalent factors s0 and s1 are chosen in order to
guarantee that the SC state of energy (SoE) stays inside the
interval [SoEmin , SoEmax ] and that its values at the beginning
and at the end of the cycle are equal. Fig. 6 illustrates the
SoE evolution using s0 , s1 ∈ [1.4, 1.7] · 10−5 , values that
have been tuned by multiple tests. A 2.5% error between the
isc [A]

SoE initial and final values is observed. On the other hand, the
fuel consumption over the driving cycle has been computed in
two cases: for the considered FC/SC multi-source system and
when only the fuel cell is connected to the DC bus, in order to
show the supervision level contribution. The hydrogen mass
consumption mH2 = 14.1 g has been obtained, showing a
fuel economy of 21%, compared to the use of only one power
source. Time [s]
Tacking into account the fuel cell slow dynamics, it is
Fig. 8. Supercapacitor current (solid line), supercapacitor reference current
important to evaluate the system performance on the beginning (dash-dot line).
of the driving cycle. Fig. 7 and Fig. 8 show the FC and
the SC current regulation responses over the time intervals
t ∈ [20, 120] s. The reference current if c,ref is computed at V. CONCLUSIONS
each sampling step by the supervision system and isc,ref is A two-level control structure has been elaborated for electric
determined by the DC voltage control loop. The simulation vehicles, which assures the DC bus voltage regulation while
results show that the specified closed-loop response time is optimizing the system global efficiency by minimizing the
satisfied without steady state-error. hydrogen mass consumption. The simulation results illustrate
 [6] H. El Fadil, F. Giri, and J. Guerrero. Lyapunov based control of
50.3 hybrid energy storage system in electric vehicles. In American Control
Conference (ACC), 2012, 2012.
50.1
[7] M. Becherif, M. Y. Ayad, A. Henni, M. Wack, A. Aboubou, A. Allag,
50
and M. Sebai. Passivity-based control and sliding mode control applied
vbus [V]

55 55.01
to electric vehicles based on fuel cells, supercapacitors and batteries on
the dc link. Energy Management, pages 107–132, 2010.
[8] P. Thounthong, V. Chunkag, P. Sethakul, and B. Davat. Comparative
study of fuel-cell vehicle hybridization with battery or supercapacitor
storage device. Vehicular Technology, 58:3892 – 3904, 2009.
[9] A.Vahidi and W. Greenwell. A decentralized model predictive control
approach to power management of a fuel cell-ultracapacitor hybrid. In
American Control Conference, 2007.
[10] W. Greenwell and A. Vahidi. Predictive control of voltage and current
Time [s] in a fuel cell - ultracapacitor hybrid. Industrial Electronics, 57:1954 –
1963, 2010.
Fig. 9. DC bus voltage (solid line), DC bus reference voltage (dash-dot line). [11] M. Ghanes, M. Hilairet, J. P. Barbot, and O.Bethoux. Singular pertur-
bation control for coordination of converters in a fuel cell system. In
ELECTRIMACS, 2011.
2.45 [12] L. N. Degliuomini, D. Zumoffen, M. Basualdo, D. Feroldi, and J. Rierac.
Adaptive predictive robust control for fuel cells hybrid vehicles. In
2.4
Vehicle Power and Propulsion Conference, 2010.
2.35 [13] D. Gao, Z. Jin, and Q. Lu. Energy management strategy based on fuzzy
Power [kW]

38 38.02 38.04 logic for a fuelcell hybrid bus. Journal of Power Sources, 185:311–317,
2008.
[14] A. Ravey. Conception et gestion de l’énergie des architectures pour
véhicules hybrides électriques. PhD thesis, University of Technology of
Belfort-Montbéliard, 2013.
[15] L. V. Perez, G. R. Bossio, D. Moitre, and G. O. Garcia. Optimization
of power management in an hybrid electric vehicle using dynamic
programming. Mathematics and Computers in Simulation, 73:244–254,
2006.
[16] N. Kim, S. Cha, and H. Peng. Optimal control of hybrid electric vehicles
Time [s] based on pontryagin’s minimum principle. Control Systems Technology,
19:1279 – 1287, 2010.
Fig. 10. Power supplied by the multi-source system (solid line), power [17] P. Garcı́a, J. P. Torreglosa, L. M. Fernández, and F. Jurado. Control
demand (dash-dot line) strategies for high-power electric vehicles powered by hydrogen fuel
cell, battery and supercapacitor. Expert Systems with Applications,
40:4791–4804, 2013.
[18] A. Sciarretta, M. Back, and L. Guzzella. Optimal control of parallel
good performances of the control loops and a significant hybrid electric vehicles. Control Systems Technology, 12:352–363, 2004.
fuel economy due to supervision system based on equivalent [19] A. Chasse and A. Sciarretta. Supervisory control of hybrid powertrains:
An experimental benchmark of offline optimization and online energy
consumption strategy. management. Control Engineering Practice, 19:1253–1265, 2011.
As perspectives of this work, the equivalent consump- [20] J. P. Torreglosa, F. Jurado, P. Garcı́a, and L. M. Fernández. Hybrid fuel
tion minimization strategy could be improved in order to cell and battery tramway control based on an equivalent consumption
minimization strategy. Control Engineering Practice, 19:1182–1194,
simplify the equivalence factors tuning for different driving 2011.
cycles. Also, the supervision level could consider the fuel [21] S. Kermani, S. Delprat, T. M. Guerra, R. Trigui, and B. Jeanneret.
cell dynamics by using a dynamic model of the FC. Besides Predictive energy management for hybrid vehicle. Control Engineering
Practice, 20:408 – 420, 2012.
this, a different on-line optimization-based method could be [22] P. Thounthong, S. Rael, and B. Davat. Control strategy of fuel
employed, such as model predictive control. Finally, the en- cell/supercapacitors hybrid power sources for electric vehicle. Power
ergy management control system could be implemented and Sources, 158:806 – 814, 2006.
[23] L.M. Fernandez, P. Garcia, C.A. Garcia, J.P. Torreglosa, and F. Jurado.
evaluated on an embedded system. Comparison of control schemes for a fuel cell hybrid tramway integrat-
ing two dc/dc converters. International Journal of Hydrogen Energy,
R EFERENCES 35:5731 – 5744, 2010.
[1] G. Rousseau. Véhicule hybride et commande optimale. PhD thesis, [24] G. Fontes. Modélisation et caractérisation de la pile PEM pour l’étude
École nationale supérieure des Mines de Paris, 2008. des interactions avec les convertisseurs statiques. PhD thesis, Institut
[2] M. Cacciato, F. Caricchi, F. Capponi, and E. Santini. A critical National Polytechnique de Toulouse, 2005.
evaluation and design of bi-directional dc/dc converters for super- [25] J. T. Pukrushpan. Modeling and control of fuel cell systems and fuel
capacitors interfacing in fuel cell applications. In Industry Applications processors. PhD thesis, University of Michigan, 2003.
Conference, 2004. [26] R. Mosdale. Transport électrique routier - véhicules électriques à pile à
[3] M. Hilairet, M. Ghanes, O. Béthoux, V. Tanasa, J.P. Barbot, and combustible. Technical report, Techniques de l’ingénieur, 2003.
D. Normand-Cyrot. A passivity-based controller for coordination of [27] M. Andre. The artemis european driving cycles for measuring car
converters in a fuel cell system. Control Engineering Practice, 21:1097 pollutant emissions. Science of the Total Environment, 334-335:73–84,
– 1109, 2013. 2004.
[4] M. Becherif, M.Y. Ayad, and A. Miraoui. Modeling and passivity-based [28] P. Garcı́a, J. P. Torreglosa, L. M. Fernández, and F. Jurado. Viability
control of hybrid sources: Fuel cell and supercapacitors. In Industry study of a fc-battery-sc tramway controlled by equivalent consumption
Applications Conference, 41st IAS Annual Meeting, 2006. minimization strategy. International Journal of Hydrogen Energy,
[5] A. Tofighi and M. Kalantar. Adaptive passivity-based control of pem fuel 37:9368 – 9382, 2012.
cell/battery hybrid power source for stand-alone applications. Advances [29] T. Hofman and M. Steinbuch. Rule-based energy management strategies
in Electrical and Computer Engineering, 4:164–171, 2010. for hybrid vehicles. Electric and Hybrid Vehicles, 1:71–94, 2007.