Journal of Power Sources 395 (2018) 262–270

Contents lists available at ScienceDirect

Journal of Power Sources

journal homepage: www.elsevier.com/locate/jpowsour

A novel equivalent consumption minimization strategy for hybrid electric T

vehicle powered by fuel cell, battery and supercapacitor
Huan Lia,b,∗, Alexandre Raveya,b, Abdoul N'Diayea,b, Abdesslem Djerdira,b
a
FEMTO-ST, CNRS, Univ. Bourgogne Franche-Comte, UTBM, France
b
FCLAB, CNRS, Univ. Bourgogne Franche-Comte, France

H I GH L IG H T S

• Novel SECMS strategy for FCHEV with three power sources.

• Solving two degrees of freedom energy management problem.
• Experimental validation of designed SECMS strategy through test bench.

A R T I C LE I N FO A B S T R A C T

Keywords: The aim of this paper is to present a sequential quadratic programming (SQP) based equivalent consumption
Fuel cell hybrid electric vehicles (FCHEVs) minimum strategy (ECMS) (SECMS) for fuel cell hybrid electric vehicle (FCHEV) powered by fuel cell, battery
Equivalent consumption minimum strategy and supercapacitor. In order to decrease hydrogen consumption and increase the durability of power sources,
(ECMS) fuel cell is chosen as the main power source and supplies steady current, battery is designed as the main energy
Three power sources
buffer and the replacement of fuel cell failure and supercapacitor is operated to supply peak power. Low energy
Experimental test bench
density of supercpacitor lets its equivalent hydrogen consumption be taken as zero for many ECMS researches.
This simplification leads to suboptimal fuel economy and complex of control system. SECMS considers hydrogen
consumption of three power sources into objective function to solve this problem. A rule based control strategy
(RBCS) and an hybrid ECMS operating mode control strategy (OMCS) (HEOS) are also designed to compare with
SECMS. An experimental test bench is built to validate the comparative study of three strategies. The results
show that compared with RBCS and HEOS, hydrogen consumption of SECMS decreases of 2.16% and 1,47%
respectively and it also has the most smooth fuel cell current, which means a lowest fuel cell degradation.

1. Introduction
Traditional gasoline and diesel vehicles have lead to many problems
such as global warming, environment pollution and exhaustion of
petroleum energy. Electric vehicles including pure electric vehicles,
hybrid electric vehicles, and plug in hybrid electric vehicle are thought
to be the best way to solve these problems [1]. Compared to traditional
internal combustion engine, fuel cell has high efficiency and zero pol-
lution emission, which is ideal energy source for electric vehicles [2].
Gas supply of fuel cell stack lags behind the load variation, which
leads to difficulties to track the dynamic response of current specially
related to transportation application. Consequently, at least one kinds
of energy storage sources (ESS) should be added as the power sources to
FCHEV. Batteries have high energy density, which are the most widely
used ESS in transportation systems. But batteries also have some
shortcomings like low power density, long charging time, high cost,
short lifetime and seriously affected by temperature. In contrast, su-
percapacitor has high power density, very high lifetime and are not
affected by temperature, which makes it suitable as device for power
pulse. However low energy density, voltage balancing needed and high
self-discharge are the main barrier for supercapacitor to be widely used
in the hybrid electric vehicles. One of the most promised solution
proposed for FCHEV supplying is the topology with fuel cell, battery
and supercpacitor. This topology of power train allows the main com-
ponents to give play to their advantages: fuel cell as main steady power
source, battery as energy buffer and supercapacitor as device for power
pulse. In order to achieve this hybridization and reach the above goal,
an energy management strategy (EMS) is necessary.
In the literature, the EMSs can be classified into rule based control
strategies (RBCS) [3] and optimization based control strategies [4].

∗
Corresponding author. FEMTO-ST, CNRS, Univ. Bourgogne Franche-Comte, UTBM, France.
E-mail address: huan.li@utbm.fr (H. Li).

https://doi.org/10.1016/j.jpowsour.2018.05.078
Received 20 February 2018; Received in revised form 22 May 2018; Accepted 23 May 2018
0378-7753/ © 2018 Elsevier B.V. All rights reserved.
H. Li et al.
Nomenclature
baSOC Battery SOC value
F Faraday constant
IFC Fuel cell current
mBA Battery equivalent hydrogen consumption
mFC Fuel cell hydrogen consumption
mSC Supercapacitor equivalent hydrogen consumption
Ncell The number of fuel cell stack current
RBCS uses direct rules or fuzzy rules to split power demand among
different power sources, making it simple to design and allowing real
time control. State machine control strategy [5], stiffness coefficient
model control strategy [6], operation mode control strategy (OMCS) [7]
and fuzzy logic control strategy [8] are kinds of RBCS which are widely
used. In these strategies, the rules are designed in accordance with
engineering experiences, consequently, optimal power split is difficult
to reach [9,10]. To consume less hydrogen, increase driving distance or
extend the lifetime of fuel cell and ESSs, optimization based control
strategies are used to find the optimal result. It can be divided into
global optimization strategies and local optimization strategies [11].
Dynamic programming and genetic algorithm are the most effective
strategies to solve global optimization problem. Prior knowledge about
drive condition and long calculation time limit their application on the
real time vehicle control. Pontryagins minimum principle [12] and
ECMS transform global optimization problem into instantaneous ones
which instantaneously calculate the optimization objective function to
split power among power sources [13].
Up to now, few papers focus on building EMS which takes into ac-
count three power sources. State machine control strategy of [5] and
fuzzy logic strategy of [14–16] are used to control the power split
among three power sources. But they belong to RBCS which lead to
suboptimal results. There are lesser researches evaluating the fuel
economy potential of supercapacitor and battery combination for op-
timization strategy [17]. Solves this problem with dynamical pro-
gramming but the proposed method cannot be used in real time. Two-
level control structure, where first level calculates the optimal results
between fuel cell and battery and the second one lets supercapacitor
improve the battery performance, are widely used for three power
sources like [18]. ECMSs in Refs. [19–21] are designed as two level
architecture and take equivalent hydrogen consumption of super-
capacitor as zero which not only counter to the aim of minimizing
whole hydrogen consumption but also increase the complication of
Fig. 1. Powertrain architecture.
263
Journal of Power Sources 395 (2018) 262–270
PBA Battery output power
PFC Fuel cell power
PSC Supercapacitor output power
R Ideal gas constant
Sfc Fuel cell on/off state
SOCbamax Maximum battery SOC value
SOCbamin Minimum battery SOC value
T The temperature of fuel cell stack
VFC Fuel cell voltage
EMS due to the need of an additional EMS to calculate supercapacitor
power demand. Thus, SECMS strategy is proposed to consider energy
cost of all three power sources into the objective function to solve this
problem. At the same time, a RBCS and HEOS strategies are also
compared to demonstrate the superiority of SECMS in minimizing the
hydrogen consumption and prolonging fuel cell lifetime.
This paper is organized as following: the first section is introduction,
section two gives the vehicle architecture and the model of the power
train including fuel cell, battery, supercapacitor and DC/DC converters.
In the third section, SECMS, RBCS and HEOS are explained. In the
fourth part, the used validating test bench is described and experiment
results are compared for different control strategies. Finally, conclu-
sions are drawn.
2. Vehicle models
2.1. Power train architecture
The series architecture is chosen for FCHEV, as shown in Fig. 1.
Proton Exchange Membrane Fuel Cell (PEMFC) is the main energy
source to supply steady state power and is connected to the DC bus via a
unidirectional DC/DC power converter. Lead acid battery as main en-
ergy storage source is directly connected to DC bus to hold the bus
voltage. Supercapacitor as peak power supply is connect to the DC bus
through a bidirectional DC/DC power converter.
The longitudinal dynamics of a road vehicle can be described in the
following equation (1), through which the required power Pcycle at wheel
to drive the vehicle can be calculated [22]:
d
Pcycle (t ) = v ⎛m v (t ) v (t ) + Fa (t ) + Fr (t ) + Fg (t ) ⎞
⎝ dt ⎠ (1)
where Pcycle is the power demand from drive cycle, Fa is the aerodynamic
friction, Fr the rolling friction, and Fg the force caused by gravity when
H. Li et al. Journal of Power Sources 395 (2018) 262–270

driving on slope road: Fuel cell theoretical efficiency is defined as the ration between
1 power generated and the power of hydrogen supplied as equation (11)
Fa = ρ A Cx v 2 [25].
2 (2)
VFC IFC
Fr = m v Cr g cos(α ) (3) PFC −ΔHLHV IFC VFC
ηLHV = = =
PH2 2F 1.254 (11)
Fg = m v g sin(α ) (4)
where ηLHV is fuel cell theoretical efficiency, ΔHLHV is the lower heating
Variable ρ is the air density, m v the vehicle mass, A front surface of the value of hydrogen.
vehicle, g gravity force, Cx the drag coefficient, Cr the aerodynamic drag Some auxiliary systems are needed to make sure the normal op-
coefficient, v the speed of the vehicle and α the angle defining the slope eration of fuel cell system, such as electrical control border, cooling fan
of the road. and air compressor. As the result, the efficiency of fuel cell system is
The power demand on the DC bus from the fuel cell, battery and decided by fuel cell theoretical efficiency and real auxiliary efficiency,
supercapacitor is given as equation (5) (6) which can be calculated as equation (12)
Pcycle
Pdemand = VFC ⎛ PFC − PAUX ⎞
ηDC / AC ∗ηmotor (5) ηFCS = ηLHV *ηaux = ⎜ ⎟

1.254 ⎝ PFC ⎠ (12)

Pdemand = PFC ∗ηDCFC + PSC ∗ηDCSC + PBA (6)
where Pdemand is power demand, ηDC / AC is converter efficiency of DC/AC
connected to motor, ηmotor is motor efficiency, ηDCFC is unidirectional
DC/DC converter efficiency connected to fuel cell, ηDCSC is bidirectional
DC/DC converter efficiency connected to supercapacitor.
2.2. Fuel cell model
PEMFC as the main power source for the FCHEV transforms the
where ηFCS is fuel cell system efficiency, ηLHV is fuel cell theoretical
efficiency, ηaux fuel cell real auxiliary efficiency, PAUX fuel cell power
consumed by auxiliary system.
According to the study of [26], the compressor power is up to 93.5%
of the total auxiliary power, so a precise compressor model to calculate
its power variation along with fuel cell current is built and the other
auxiliary power is set as a constant value. The power consumed by the
air compressor is shown as equation (13)
γ−1

chemical energy into electrical energy through the reaction between CP Tair ⎛ ⎛ Pout ⎞ γ ⎞
Pcp = − 1⎟ Fcp
ηmec ηmot ⎜⎜ ⎝ Pin ⎠
⎜ ⎟

hydrogen and oxygen [23]. A steady state model is built for the fuel cell ⎟
⎝ ⎠ (13)
stack.
The fuel cell stack output voltage is expressed as following equation where Pcp is air compressor power, CP is heat capacity of air, Tair inlet air
(7) [24]. temperature, ηmec is compressor mechanical efficiency, ηmot is the effi-
ciency of compressor motor, Pin is input air pressure, Pout is output air
Vstack = Ncell ∗ (Ecell − Vact − Vohm) (7) pressure, γ is ratio of the specific heat of air. Fcp is the compressor air
where Ecell is the electromotive potential, Vact is the cell activation flow rate which can be defined according to the fuel cell current as
losses, Vohm is the cell ohmic losses. function (14)
The cell electromotive voltage Ecell can be obtained from equation Ncell ∗IFC
(8) Fcp = S∗Mair
4XO2 ∗F (14)
RT
Ecell = 1.229 − 0.85e−3 (T − Tc ) + ln ( PO2 PH2) where S is the stoichiometric ratio, Mair is the number of air moles, XO2
2F (8)
oxygen molar fraction.
where Tc is the temperature correction offset, PO2 is the oxygen pressure It should be noticed the DC/DC converter connected to fuel cell
at the interface of cathode catalyst layer, PH2 is hydrogen pressure at the stack affects the output power of fuel cell stack on DC bus. So, its ef-
interface of anode catalyst layer. ficiency is also included into fuel cell system. According to equation
The activation losses Vact can be described using the Tafel equation (12), the fuel cell system efficiency can be calculated and shown in
(9) Fig. 2.
It can be observed that the maximum efficiency point 42.83% occurs
RT ⎛ IFC ⎞
Vact = ln ⎜ ⎟ at fuel cell current 9.5A . From current 4.5A to 20A in red color, the fuel
2αF ⎝ I0 S ⎠ (9) cell system efficiency is above 40% which is defined as a high efficiency
where α is charge transfer coefficient, I0 is the exchange current density, zone. To reduce the final hydrogen consumption, fuel cell should be
S is the catalyst layer section area. operate to seek maximum efficiency point within this zone.
The cell ohmic losses Vohm can be obtained by computing the The hydrogen consumption rate can be defined by fuel cell current
membrane resistance expression as equation (10) as the following equation (15) [27]:

i δmem t MH2 Ncell

Vohm = Rmem i = ∫0 γ (T , λ (z )) dz
(10)
mH 2 = ∫0 2F
IFC (t ) dt
(15)
s
Rmem is the internal resistance of the cell, δmem is the membrane thick-
Table 1
ness and γ (T , λ (z )) is the membrane local resistance.
Fuel cell parameters.
According to the above described function (7) (8) (9) (10) and fuel
cell parameters as Table 1, the polarization curve of the fuel cell stack Company Ballard NEXA PEMFC
can be defined. Cell number 47
Rated power (W) 1200
2.3. Fuel cell efficiency model Operating voltage range (V) [22,50]
Maximum current (A) 46
Air supply Air blower + filter
In order to minimize fuel cell hydrogen consumption, fuel cell stack
Cooling Air fan cooled
should be operated to seek maximum efficiency point at high efficiency Fuel supply 99.99% dry H2 @1.2 bar
region, so a precise fuel cell efficiency model is needed.

264
H. Li et al. Journal of Power Sources 395 (2018) 262–270

2.6. DC/DC converter model

Fuel cell stack is connected to DC bus through a 1-quadrant DC/DC

boost converter and supercapacitor is connected to DC bus through a 2-
quadrant DC/DC buck/boost converter with boost operation for dis-
charging mode and buck operation for charging mode. Each converter
consists of two IGBT transistors which are controlled by two com-
plementary pulse width modulated (PWM) signals. Two DC/DC con-
verters have same architecture, which are shown in Fig. 1. Different
from buck/boost converter for supercapacitor, the S1 IGBT transistors of
fuel cell 1-quadrant DC/DC boost converter is always set in the state of
off.
The relationship between input power and output power of two
converters is described in the following equation (19)
Pin
Iout = ηconv
Uout (19)
Fig. 2. Fuel cell system efficiency curve along with stack current.
where Pin represents input power, Uout is output voltage and ηconv is DC/
DC converter efficiency.
where m H2 represents the hydrogen mass rate, MH2 is the hydrogen
molar mass.
3. Energy management strategy

2.4. Battery model The hierarchical control strategy proposed in this paper consists of
two level control loop: the high level control loop corresponding to EMS
Lead acid battery is used in the test bench, which is connected to the and the low level control loop corresponding to DC/DC converter
DC bus to hold the bus voltage. The battery state of charge (SOC) can be control, which is shown in Fig. 3.
calculated according to battery current as equation (16) [28] [29]:
ηb t 3.1. Low level control loop
SOC (t ) = SOCinit −
Cnom
∫0 Ib (t ) dt
(16)
Once the reference currents of fuel cell and supercapacitor are de-
Cnom represents the batteries nominal capacity, SOCinit is initial battery fined by high level control loop, two classical PI controllers are applied
SOC, ηb is charge and discharge efficiency. to adjust duty cycles of PWM signals to control the real fuel cell and
supercapacitor output currents to track reference currents. The low
2.5. Supercapacitor model level control loop is denoted in red zone in Fig. 3.

Supercapacitor is modeled as a capacitor and an equivalent re-
sistance [30] as shown in Fig. 1. Capacitor represents supercapacitor
performance at discharge and charge state and resistance represents the
supercapacitor ohmic losses. The open circuit voltage of supercapacitor
has line relationship with its SOC, so the SOC can be described as
equation (17). The supercapacitor current can be calculated through
following equation (18).
Voc = SOC∗ (Vmax − Vmin) + Vmin
Voc − Voc2 − 4RSC PSC
I=
2RSC
where Vmax is supercapacitor maximum voltage, Vmin output minimum
voltage, Voc capacitor voltage, RSC equivalent resistance.
3.2. High level control loop
Three EMSs are designed for the high level control loop. A novel
SECMS is designed to operate fuel cell to seek for the maximum effi-
ciency point at high efficiency zone. Meanwhile fuel cell dynamical
current change rate is limited to increase the lifetime of fuel cell. The
supercapacitor is operated to supply peak power to decrease maximum
(17) transient current of battery to increase its lifetime. SOC values of bat-
tery and supercapacitor are kept within a reasonable range and the
terminal SOC of battery is close to the initial one. A RBCS is designed as
a benchmark against SECMS. In order to prove that neglecting super-
(18)
capacitor equivalent hydrogen consumption at ECMS objective function
can not reach the optimal result for three power sources power train, a
HEOS is also designed.

Fig. 3. Hierarchical control architecture.

265
H. Li et al. Journal of Power Sources 395 (2018) 262–270

3.2.1. SQP based equivalent consumption minimization strategy
All energy needed by vehicle is supplied indirectly by the fuel cell
system at last. In order to minimize hydrogen consumption, the in-
stantaneously consumed electrical energy from battery and super-
capacitor can be equivalent to the chemical energy from fuel cell. The
instantaneous hydrogen consumption composes of direct hydrogen
consumption from fuel cell system and indirect equivalent hydrogen
consumption from battery and supercapacitor, as shown in equation
(20),
m w (t ) = mfc (t ) + mBA (t ) + mSC (t )
= mfc (t ) + λba Pba (t ) + λsc Psc (t ) (20)
Supercapacitor penalty coefficient Ksc is composed of SOC coeffi-
cient Seff and peak power coefficient Speak . Seff is familiar to Kba to re-
strict supercapacitor SOC value at reasonable range. Speak is used to let
supercapacitor supply peak power firstly. In order to avoid the frequent
on/off cycles of fuel cell and frequent charge/discharge cycles of su-
percapacitor due to the large amplitude changes of supercapacitor SOC
in short time, supercapacitor SOC is equivalent to battery SOC to define
Seff . K SC , Seff and Speak can be defined as equations (26)–(28) respec-
tively:
K SC = Seff ∗Speak (26)
ax + b − Sopt 2

where m w (t ) is the whole hydrogen consumption. λba and λsc are battery Seff =
⎪(
⎧ 1− 2S −S
max min ) Smin ≤ x ≤ Smax
⎨ 20
and supercapacitor equivalent factor for equivalent hydrogen con-
sumption.
( ax + b − Sopt
⎪ 1 − 2 Smax − Smin
⎩
) x< Smin, x > Smax
(27)
In order to keep fuel cell working at high efficiency zone, battery
1 0 ≤ Iload ≤ 30
SOC of end cycle same to start value, supercapacitor supplying peak Speak = ⎧
⎨ − 0.01∗Iload + 1 Iload < 0, Iload > 30 (28)
power, the relevant penalty coefficients are added into equation (20), ⎩
and the objective function is defined as equation (21), where x is the instantaneous supercapacitor SOC, Sopt is optimal SOC,
fw (t ) = K eff mfc (t ) + Kba mBA (t ) + Ksc mSC (t ) Smax the maximum SOC, Smin the minimum SOC, Iload is load current
demand on the DC bus, a and b are the transform coefficients from
= K eff mfc (t ) + Kba λba Pba (t ) + Ksc λsc Psc (t ) (21)
supercapacitor SOC to equivalent battery SOC and their value are
where Kba and Ksc are penalty coefficients which limit battery and su- decided by battery minimum SOC and maximum SOC.
percapacitor SOC range and variation between instantaneous SOC and As described in the above content, all the factors of the objective
initial SOC, K eff is the fuel cell efficiency penalty coefficient that urges function fw can be decided. Some other constrains should be added to
fuel cell to operate at maximum efficiency at high efficiency zone [31]. make sure the normal operation of power sources. In case fuel cell
According to equation (15), fuel cell hydrogen consumption mfc (t ) works at low efficiency zone with at very low current and avoids fre-
can be computed directly. The equivalent hydrogen consumption of quent on/off cycle, minimum current is set as 4.5A , when fuel cell
battery is calculated through equation (22): current is lower than that value, fuel cell is shut off. In case of de-
gradation caused by large transient current, dynamical change rate of
maverage
mBA (t ) = Pba (t ) ∗ ∗ηba fuel cell is limited to [-1,1] ( A/ s ). The current range of supercapacitor is
Paverage (22) also limited to make sure its normal operation.
for supercapacitor as equation (23): The ECMS strategy is transformed into a non-linear programming
problem, which has fuel cell reference current and supercapacitor re-
maverage
mSC (t ) = Psc (t ) ∗ ∗ηsc ference current as variables, equivalent hydrogen consumption as
Paverage (23) weighting function and subjects to several constrains and inequality
where ηba ηsc is battery and supercapacitor equivalent transform effi- constraints. SQP is one of the most successful methods for the numerical
ciency, maverage is fuel cell average hydrogen consumption, Paverage is fuel solution of constrained nonlinear optimization problems [32]. There-
cell average power. fore, SQP is programmed in C language to solve ECMS optimization
Fuel cell efficiency penalty coefficient K eff is defined as equation problem in real time.
(24):
3.2.2. Rule based control strategy
2
⎧ 1 − 2∗ η − ηopt
K eff =
⎪ ( ηmax − ηmin ) η ≥ 0.4 A RBCS is designed as the benchmark against SECMS. The RBCS is
divided into two parts: load following control strategy (LFCS) to decide
⎨ 4
( η − ηopt
⎪ 1 − 2∗ ηmax − ηmin
⎩
) η < 0.4
(24)
what current the fuel cell should be operated according to power de-
mand by drive cycle and battery SOC, and, OMCS to calculate super-
capacitor current according to supercapacitor SOC and the difference
where η is the instantaneous efficiency, ηopt is optimal efficiency
between fuel cell supplied power and load power. LFCS is a real-time
(0.4283), ηmax the maximum efficiency (0.4283), ηmin the minimum
rule based strategy. The main idea of LFCS is that fuel cell as the main
efficiency (0.4), which define the zone described in section 2.3. When
power source is operated at high efficiency zone same as SECMS, and its
fuel cell system efficiency is below than 0.4, a large penalty value K eff is
power changes follow the load power demand on DC bus. Frequent fuel
calculated to shut fuel cell stack down or operate fuel cell stack to meet
cell on/off cycles degrade fuel cell more seriously, so it is operated at on
power demand by drive cycle when battery and supercpacitor SOC are
state until battery SOC is above maximum value or power demand
lower than limited SOC range.
value is very low. The conditions that set fuel cell on or off are listed as
Regarding battery SOC penalty coefficient Kba , it is defined as
following, where Sfc = 0 means fuel cell is off, Sfc = 1 means fuel cell is
equation (25):
on, Id current demand on the DC bus, tofftime is the fuel cell minimum off
4 time that means since the fuel cell was last on, the restart should not be
Kba =
⎪ (
⎧ 1 − 2 ∗ (u − Bint )
Bmax − Bmin ) Bmin ≤ u ≤ Bmax
less than this time. tontime is similar to tofftime means the fuel cell
⎨ 20
( 2 ∗ (u − Bint )
⎪ 1 − Bmax − Bmin
⎩
) u< Bmin , u > Bmax
(25)
minimum on time.

where u is the instantaneous battery SOC, Bint is battery initial SOC,
Bmax the maximum SOC, Bmin the minimum SOC. Kba operates the
battery SOC to return back to its initial SOC. When battery SOC reaches
Bmin or Bmax , high Kba value is defined as the penalty factor to avoid the
battery continues to discharge and charge respectively.
Condition 1: if the fuel cell is off and baSOC < SOCbamin , fuel cell is
turned on immediately, which is not limited by tofftime . The current of
fuel cell is set as Ifcmax .
Condition 2: if the fuel cell is on and baSOC > SOCbamax , fuel cell is
turned off, which is not limited by tontime . The current of fuel cell is 0.

266
H. Li et al.
Condition 3: if the fuel cell is on and SOCbamin < baSOC < SOCbaint ,
the fuel cell current is adjusted according to load power.
Condition 4: if the fuel cell was previously off, and the average of
the last 5 s of Id is larger than fuel cell maximum efficiency point Iopt
and the time since the fuel cell was last on is larger than tofftime , the
fuel cell is started.
Condition 5: if the fuel cell was previously on, and the average of the
last 5 s of Id is less than fuel cell maximum efficiency current Iopt and
the time since the fuel cell was last off is larger than fuel cell
minimum on time tofftime , the fuel cell is off.
Condition 6: if Id on the bus is greater than ESS maximum current on
the bus Iessmax , the fuel cell stays on or is turned on.
Fuel cell current Ifc is decided by current demand on DC bus Id and
battery SOC value as equation (29). Fuel cell current is limited to high
efficiency zone (4.5 A, 20 A). Beside meeting Id , fuel cell also tries to
charge the battery to its SOC initial value. The charge current is cal-
culated according to SOC value.
⎧ Imin Ich + Id < Imin
Ifc = Ich (baSOC ) + Id Imin ≤ Ich + Id ≤ Imax
⎨
⎩ Imax Ich + Id > Imax (29)
when fuel cell reference current is decided by equation (29), the dif-
ference current Idi between Id and Ifc is supplied by battery and su-
percpacitor, which is decided by OMCS. The flow chart of OMCS is
shown in Fig. 4.
The main idea of OMCS is when Idi < 0 and supercapacitor SOC
SOCsc is smaller than initial supercapacitor SOC SOCint , super-
capacitor firstly be charged at maximum charge current I SCmin . When
supercapcitor discharge, its current I SC > 0 , when it is charged, its
current I SC < 0 . when I di > I bamin and SOCsc in the range of
minimum SOC and maximum SOC, the supercapacitor supplies max-
imum discharge current. Otherwise supercapacitor current is zero.
Battery current is passively decided by difference between current de-
mand by drive cycle, fuel cell current and supercapacitor current on DC
bus.
3.2.3. Hybrid ECMS OMCS strategy
Due to low energy density and the role of supercapacitor as peak
power supplier, many researches on ECMS simplify the equivalent hy-
drogen consumption of supercapcitor into zero. On the based of equa-
tion (21), a new objective function is defined as (30),
Fig. 4. Flow chart of OMCS.
267
Journal of Power Sources 395 (2018) 262–270
m w (t ) = mfc (t ) + mBA (t )
= mfc (t ) + λba Pba (t ) (30)
Trough the simplified ECMS, the fuel cell current is calculated. The
supercapacitor current is then calculated by other methods like PI based
control strategy and filtration based control strategy. To prove that the
simplification of ECMS can not reach optimal resolution, a HEOS
strategy is designed. The first part of HEOS is simplified ECMS to cal-
culate fuel cell current and the second part is OMCS to calculate su-
percapacitor current.
4. Experiment implementation
4.1. Test bench description
To compare the designed SECMS, RBCS and HEOS to each other, a
test bench is developed. It includes PEMFC, battery, supercapacitor,
DC/DC converters, measurement instruments and sensors, power
supply, power load, MicroAutoBox and PC. The architecture of test
bench is shown in Fig. 5.
The fuel cell stack is the main power source and its parameters is
shown in Table 1. Four lead acid batteries connected in series are used
as the main energy storage source in the test bench. While in realistic
operation, lithium ion batteries are widely used in electric vehicles.
Since this test bench relies on a reference power determined by a ve-
hicle model emulated by an active load, only the dynamic power re-
sponse can be different when switching technologies between lead-acid
and Li-ion, which because the test bench does not take into account
facts such as power density or weight of the batteries. Besides, the peak
power from experimental driving cycle is not big enough to clearly see a
difference using lead acid or Li-ion in the experiment. Moreover, if a
difference appears, it would have impact on all the three compared
strategies in the same way, which would not lead to a difference of
results among the three of them. The capacity and rated voltage for
single battery is 90 A h and 12 V respectively. Further more, the battery
could be managed to replace the fuel cell as the main power source in
the case of running out of hydrogen consumption or damage of fuel cell
stack. Four MAXWELL supercapacitors are operated to supply peak
power, with two in series and two in parallel. Each supercapacitor rated
capacitance is 58 F and each rated voltage is 16 V. So supercapacitor
voltage range is limited to 15 V–30 V.
A boost DC/DC converter for fuel cell and a buck/boost DC/DC
H. Li et al. Journal of Power Sources 395 (2018) 262–270

Fig. 5. Test bench architecture.

Fig. 6. Experimental results of battery SOC and supercapacitor SOC.

Fig. 7. Experimental results of fuel cell, battery and supercapacitor current for three control strategies.

converter for supercapacitor are packed into one box in parallel ar-
chitecture. The DC bus voltage is decided by battery voltage. Power
supply and power load are programmed to supply negative power and
positive power of drive cycle respectively, which are both connected to
the DC BUS directly. Some measurements instruments and sensors are
used to measure current and voltage of relevant power sources.
MicroAutoBox Π from dSPACE is used as the control unit, which
gathers all control signals needed by EMS and outputs 20 KHz PWM
signals to DC/DC converters. Human machine interface (HMI) in the PC
is designed using the ControlDesk 4.2 program, which can monitor the
predefined variable's instantaneous value, tune control parameters of
whole control system and record all real time variable's value in order
to analyze the final experiment results.

268
H. Li et al.
Table 2
Experiment results.
EMSs SECMS RBCS HEOS
Initial supercapacitor and battery SOC 0.8 0.8 0.8
Final supercapacitor SOC 0.5865 0.8304 0.8199
Final battery SOC 0.7915 0.7946 0.7906
Hydrogen consumption (L) 194.07 210.35 194.77
Equivalent hydrogen consumption (L) 225.06 230.02 228.42
Fig. 8. Compare results of fuel cell reference current for SECMS and HEOS.
4.2. Experiment results
The WVUCITY drive cycle, New York Bus drive cycle and LA92
drive cycle are connected together to test the performance of three
designed EMSs under constrains condition. The whole drive cycle time
is 3443s and the distance is 22.11 km. Considering the possibility of fuel
cell failure or out of hydrogen in the tank, battery SOC range is set from
0.75 to 0.85, its initial SOC is set 0.8. Regarding supercapacitor, its
minimum, maximum and initial SOC is set 0.1, 0.9 and 0.8 respectively.
From the experimental test of the emulated driving cycle, the
downsized load power profile of drive cycle, the battery SOC, super-
capacitor SOC, fuel cell current, battery current and supercapacitor
current for three control strategies are shown in Fig. 6 and Fig. 7. The
experiment comparison results are shown as Table 2.
From Figs. 6 and 7, It can be observed that regarding to SCEMS, the
experimental results of fuel cell current and battery current are similar
to simulated results. The final simulation result of battery SOC is 0.792
similar to the experiment value 0.7915 and this difference is negligible.
Supercapcitor current is sensitive to current requirement of drive cycle,
fuel cell current and battery SOC, therefore small sensor noise and
process calculation noise lead to the little difference between experi-
mental supercapacitor current and simulated one of SECMS. Con-
sidering low energy density of supercapacitor and its role as the sup-
plier of peak power, this difference can be neglected.
As shown in Fig. 6, battery SOC and supercapacitor SOC of three
control strategies are in the limited range (0.75, 0.85) and (0.1, 0.9)
respectively. According to Table 2, the final battery SOC of three con-
trol strategies are almost equal to initial value. The difference between
initial battery SOC and final battery SOC of HEOS is the largest among
three control strategies. The hydrogen consumption of Table 2 re-
present fuel cell real hydrogen consumption at the end of drive cycle.
The final battery SOC, supercapacitor SOC and hydrogen consumption
at the end of drive cycle are different for three control strategies. In
order to make a fair comparison of hydrogen consumption, the final
battery and supercapacitor SOC variation should also be considered
into the equivalent hydrogen consumption. Because three battery final
SOC are all less than the initial value, so when the drive cycle is over,
fuel cell is operated at maximum efficiency point 9.5 A to charge bat-
tery until its SOC value increase to initial SOC value. Similarity for
supercapacitor, it's also charged for SEMS or discharged for RBCS and
HEOS till the SOC of supercpacitor back to 0.8. The sum of hydrogen
269
Journal of Power Sources 395 (2018) 262–270
consumption to charge battery SOC back to initial SOC and hydrogen
consumption for the end of drive cycle is defined as the equivalent
hydrogen consumption. It can be observed that equivalent hydrogen
consumption of SECMS is the least, RBCS is the most. Taking RBCS as
the basis, the decrements of SECMS and HEOS are 2.16% and 0.69%.
Compared to HEOS, ECMS decreases 1.47% .
As shown in Fig. 7 (a), all fuel cells of three strategies are operated
at high efficiency zone [0.4 0.428]. Regarding to SECMS and HEOS,
fuel cell currents are around maximum efficiency point. The fuel cell of
SECMS and HEOS start at 454.8s and 462.9s respectively and they keep
working until end of drive cycle. The current of RBCS changes along
with current demand at DC bus. It can be noted that the fuel cell de-
gradation of RBCS is the highest due to its high dynamical change of
current reference. In Fig. 7 (c), supercapacitor only supplies peak power
conforming to original design objective. It also can be observed that
Supercapacitor charge/discharge at different time and hold different
period of time for three strategies. Battery passively supplies other DC
bus current besides fuel cell current and supercapacitor current on the
DC bus as Fig. 7(b). In Fig. 8, the fuel cell reference currents of SECMS
and HEOS are shown. It can be observed that HEOS has more current
spikes than SECMS. The current spikes of HEOS are supplied by su-
percapacitor of SECMS to make sure the steady of fuel cell current. In
the whole, SECMS consumes the least hydrogen, has most steady cur-
rent change and less on/off cycles. HEOS has similar fuel cell current as
SECMS but with more spike current change and consumes more hy-
drogen. RBCS has highest hydrogen consumption. The fuel cell dyna-
mical current changes much and frequent on/off cycles are included
which mean that it has the most degradation for overall drive cycle.
5. Conclusion
A SECMS strategy is proposed for FCHEV supplied by three power
sources: fuel cell, battery and supercapacitor. Fuel cell is operated to
seek for the maximum efficiency point in the defined high efficiency
zone, while the battery assumes as the main energy storage source to
buffer energy demand by vehicle and the supercpacitor dedicates to
provide the peak power. This work originally takes into account the
hydrogen consumption of all three components in the adopted objective
function of energy management strategy. In order to prove the super-
iority of the new approach, the RBCS and HEOS have also been im-
plemented. The WVUCITY, New York Bus and LA92 drive cycles have
been emulated on experimental test bench with the three above control
strategies. The experiment results show that the proposed SECMS has
the least hydrogen consumption and it offers a longest durability of fuel
cell.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://dx.
doi.org/10.1016/j.jpowsour.2018.05.078.
References
[1] H. Li, A. Ravey, A. N'Diaye, A. Djerdir, Equivalent consumption minimization
strategy for hybrid electric vehicle powered by fuel cell, battery and supercapacitor,
IECON 2016-42nd Annual Conference of the IEEE Industrial Electronics Society,
2016, pp. 4401–4406.
[2] W. Zhang, J. Li, L. Xu, M. Ouyang, Optimization for a fuel cell/battery/capacity
tram with equivalent consumption minimization strategy, Energy Convers. Manag.
134 (2017) 59–69.
[3] J. Peng, H. He, R. Xiong, Rule based energy management strategy for a seriespar-
allel plug-in hybrid electric bus optimized by dynamic programming, Appl. Energy
185 (2017) 1633–1643.
[4] C. Manzie, O. Grondin, A. Sciarretta, G. Zito, Ecms controller robustness in flex-fuel
hybrid vehicles, J. Dyn. Syst. Meas. Contr. 136 (6) (2014) 064504.
[5] Q. Li, H. Yang, Y. Han, M. Li, W. Chen, A state machine strategy based on droop
control for an energy management system of pemfc-battery-supercapacitor hybrid
tramway, Int. J. Hydrogen Energy 41 (36) (2016) 16148–16159.
[6] H. Yun, S. Liu, Y. Zhao, J. Xie, C. Liu, Z. Hou, K. Wang, Energy management for fuel
H. Li et al.
cell hybrid vehicles based on a stiffness coefficient model, Int. J. Hydrogen Energy
40 (1) (2015) 633–641.
[7] P. Garca, J.P. Torreglosa, L.M. Fernndez, F. Jurado, Control strategies for high-
power electric vehicles powered by hydrogen fuel cell, battery and supercapacitor,
Expert Syst. Appl. 40 (12) (2013) 4791–4804.
[8] J.-J. Hwang, J.-S. Hu, C.-H. Lin, Design of a range extension strategy for power
decentralized fuel cell/battery electric vehicles, Int. J. Hydrogen Energy 40 (35)
(2015) 11704–11712.
[9] L. Xu, M. Ouyang, J. Li, F. Yang, L. Lu, J. Hua, Application of pontryagin's minimal
principle to the energy management strategy of plugin fuel cell electric vehicles, Int.
J. Hydrogen Energy 38 (24) (2013) 10104–10115.
[10] A. Z. nejhad, B. Asaei, A fuzzy - Genetic algorithm approach for finding a new hev
control strategy idea, in: 2010 1st Power Electronic & Drive Systems & Technologies
Conference (PEDSTC), pp. 224–229.
[11] P. Zhang, F. Yan, C. Du, A comprehensive analysis of energy management strategies
for hybrid electric vehicles based on bibliometrics, Renew. Sustain. Energy Rev. 48
(2015) 88–104.
[12] C. Zheng, S.W. Cha, Y.-i. Park, W.S. Lim, G. Xu, Pmp-based power management
strategy of fuel cell hybrid vehicles considering multi-objective optimization, Int. J.
Precis. Eng. Manuf. 14 (5) (2013) 845–853.
[13] A. Panday, H.O. Bansal, A review of optimal energy management strategies for
hybrid electric vehicle, Int. J. Vehic. Tech. 2014 (2014) 1–19.
[14] Q. Li, W. Chen, Y. Li, S. Liu, J. Huang, Energy management strategy for fuel cell/
battery/ultracapacitor hybrid vehicle based on fuzzy logic, Int. J. Electr. Power
Energy Syst. 43 (1) (2012) 514–525.
[15] J. Solano Martnez, R.I. John, D. Hissel, M.-C. Pra, A survey-based type-2 fuzzy logic
system for energy management in hybrid electrical vehicles, Inf. Sci. 190 (2012)
192–207.
[16] H. Hemi, J. Ghouili, A. Cheriti, A real time fuzzy logic power management strategy
for a fuel cell vehicle, Energy Convers. Manag. 80 (2014) 63–70.
[17] M. Ansarey, M. Shariat Panahi, H. Ziarati, M. Mahjoob, Optimal energy manage-
ment in a dual-storage fuel-cell hybrid vehicle using multi-dimensional dynamic
programming, J. Power Sources 250 (2014) 359–371.
[18] Z. Yu, D. Zinger, A. Bose, An innovative optimal power allocation strategy for fuel
cell, battery and supercapacitor hybrid electric vehicle, J. Power Sources 196 (4)
(2011) 2351–2359.
[19] P. Garca, J.P. Torreglosa, L.M. Fernndez, F. Jurado, Viability study of a fc-battery-sc
tramway controlled by equivalent consumption minimization strategy, Int. J.
270
Journal of Power Sources 395 (2018) 262–270
Hydrogen Energy 37 (11) (2012) 9368–9382.
[20] W. Zhang, J. Li, L. Xu, M. Ouyang, Optimization for a fuel cell/battery/capacity
tram with equivalent consumption minimization strategy, Energy Convers. Manag.
134 (Supplement C) (2017) 59–69.
[21] P. Garca, J.P. Torreglosa, L.M. Fernndez, F. Jurado, Control strategies for high-
power electric vehicles powered by hydrogen fuel cell, battery and supercapacitor,
Expert Syst. Appl. 40 (12) (2013) 4791–4804.
[22] L. Guzzella, A. Sciarretta, Vehicle Propulsion Systems, Introdution to Modeling and
Optimization, first ed., Springer, 2005.
[23] A. Tani, M.B. Camara, B. Dakyo, Energy management based on frequency approach
for hybrid electric vehicle applications: fuel-cell/lithium-battery and ultra-
capacitors, IEEE Trans. Veh. Technol. 61 (8) (2012) 3375–3386.
[24] F. Gao, B. Blunier, A. Miraoui, A.E. Moudni, A multiphysic dynamic 1-d model of a
proton-exchange-membrane fuel-cell stack for real-time simulation, IEEE Trans.
Ind. Electron. 57 (6) (2010) 1853–1864.
[25] J. Larminie, A. Dicks, M.S. McDonald, Fuel Cell Systems Explained vol. 2, J. Wiley
Chichester, UK, 2003.
[26] C.H. Zheng, C.E. Oh, Y.I. Park, S.W. Cha, Fuel economy evaluation of fuel cell
hybrid vehicles based on equivalent fuel consumption, Int. J. Hydrogen Energy 37
(2) (2012) 1790–1796.
[27] C. Zheng, C. Oh, Y. Park, S. Cha, Fuel economy evaluation of fuel cell hybrid ve-
hicles based on equivalent fuel consumption, Int. J. Hydrogen Energy 37 (2) (2012)
1790–1796 10th International Conference on Clean Energy 2010.
[28] A. Ravey, A. Mohammadi, D. Bouquain, Control strategy of fuel cell electric vehicle
including degradation process, IECON 2015-41st Annual Conference of the IEEE
Industrial Electronics Society, 2015, pp. 003508–003513.
[29] A. Nguyen, J. Lauber, M. Dambrine, Optimal control based algorithms for energy
management of automotive power systems with battery/supercapacitor storage
devices, Energy Convers. Manag. 87 (2014) 410–420.
[30] H. Li, A. Ravey, A. N'Diaye, A. Djerdir, Equivalent consumption minimization
strategy for fuel cell hybrid electric vehicle considering fuel cell degradation, 2017
IEEE Transportation Electrification Conference and Expo (ITEC), 2017, pp.
540–544.
[31] S. Onori, L. Serrao, G. Rizzoni, Hybrid Electric Vehicles: Energy Management
Strategies, (2015).
[32] X.Q. Xing, K.W. Lum, H.J. Poh, Y.L. Wu, Geometry optimization for proton-ex-
change membrane fuel cells with sequential quadratic programming method, J.
Power Sources 186 (1) (2009) 10–21.