You are on page 1of 13

Nonlinear Controller Analysis of Fuel Cell–

Battery–Ultracapacitor-based Hybrid
Energy Storage Systems in Electric Vehicles

Hammad Armghan, Iftikhar Ahmad,


Naghmash Ali, Muhammad Faizan
Munir, Saud Khan & Ammar Armghan

Arabian Journal for Science and


Engineering

ISSN 2193-567X
Volume 43
Number 6

Arab J Sci Eng (2018) 43:3123-3133


DOI 10.1007/s13369-018-3137-y

1 23
Your article is protected by copyright and
all rights are held exclusively by King Fahd
University of Petroleum & Minerals. This e-
offprint is for personal use only and shall not
be self-archived in electronic repositories. If
you wish to self-archive your article, please
use the accepted manuscript version for
posting on your own website. You may
further deposit the accepted manuscript
version in any repository, provided it is only
made publicly available 12 months after
official publication or later and provided
acknowledgement is given to the original
source of publication and a link is inserted
to the published article on Springer's
website. The link must be accompanied by
the following text: "The final publication is
available at link.springer.com”.

1 23
Author's personal copy
Arabian Journal for Science and Engineering (2018) 43:3123–3133
https://doi.org/10.1007/s13369-018-3137-y

RESEARCH ARTICLE - ELECTRICAL ENGINEERING

Nonlinear Controller Analysis of Fuel


Cell–Battery–Ultracapacitor-based Hybrid Energy Storage Systems
in Electric Vehicles
Hammad Armghan2 · Iftikhar Ahmad1 · Naghmash Ali2 · Muhammad Faizan Munir1 · Saud Khan1 ·
Ammar Armghan3

Received: 8 September 2017 / Accepted: 5 February 2018 / Published online: 6 March 2018
© King Fahd University of Petroleum & Minerals 2018

Abstract
Rapidly depleting oil and natural gas resources, global warming issue, and depletion of fossil fuels are motivating the
development of alternative technology for vehicular systems. Thus, an increasing number of studies have been conducted on
fuel cell electric vehicles (FCEVs). This paper proposes a modeling and nonlinear control for hybrid energy storage system
(HESS) in FCEVs. HESS consists of fuel cell (FC) as the main source and battery and ultracapacitor (UC) as secondary
sources. Each source is connected to DC bus via DC–DC converter: FC is connected to DC bus via boost converter, while
battery and UC are connected to DC bus via buck–boost converter. Based on the nonlinear behavior of power sources and
converters, a dynamic model of the system is developed. A nonlinear control technique based on Lyapunov theory is applied
to meet the following requirements: (1) accurate DC bus voltage regulation and (2) rapid tracking of battery and UC current to
their desired reference values. Both mathematical analysis and simulations are performed to prove the asymptotic convergence
of the proposed controller. To verify the performance of the controller, simulations have been done on MATLAB/Simulink,
which show that the proposed controller ensures the stability of closed loop system and meet all the control objectives.

Keywords Power converters · Hybrid electric vehicles (HEV) · Fuel cell (FC) · Battery · Ultracapacitor (UC) · Lyapunov
redesign

1 Introduction
B Iftikhar Ahmad
iftikhar.rana@seecs.edu.pk The traditional vehicles have internal combustion engines
Hammad Armghan that run on fossil fuels. The resulting emission is a big factor
14mseeharmghan@seecs.edu.pk in the rapidly rising CO2 concentration in the atmosphere,
Naghmash Ali which results in greenhouse effect that is drastically chang-
naghmash@tuf.edu.pk ing the earth climate. On the other hand, the steep rise in
Muhammad Faizan Munir prices and depletion of oil resources are forcing us to look
14mseemmunir@seecs.edu.pk for alternatives [1–3].
Saud Khan Fuel cells are becoming a likeable candidate to replace fos-
14mseeskhan@seecs.edu.pk sil fuels in transportation systems. They have the potential to
Ammar Armghan improve fuel economy and can be more energy efficient than
aarmghan@ju.edu.sa internal combustion engines [4]. There are seven different
1 kinds of FCs available in the market, but proton-exchange
School of Electrical Engineering and Computer Science,
National University of Sciences and Technology, Islamabad, membrane fuel cells (PEMFCs) are most commonly used
Pakistan in vehicular applications. PEMFCs have many advantages
2 Department of Electrical Engineering, The University of over other FCs, such as low operating temperature, high effi-
Faisalabad, Faisalabad, Pakistan ciency, high durability, long cell life and zero emission [5,6].
3 Department of Electrical Engineering, Aljouf University, However, FC cannot be used alone due to its poor transient
Sakakah, Saudi Arabia response and inability to recover kinetic energy during brak-

123
Author's personal copy
3124 Arabian Journal for Science and Engineering (2018) 43:3123–3133

None of the present HESS fulfills all the requirements of EVs.


Figure 2 shows the proposed solution. It consists of PEMFC
used as main source and battery and UC as secondary sources
[14,15,20]. The components are explained below:

2.1 Fuel Cell

FC is our primary source and a single-switch DC–DC boost


converter is used to manage the operation constraints of FC
mentioned in [23] and it also regulates the low DC voltage
supplied by FC to the rated DC bus voltage [24]. The con-
Fig. 1 Hybrid electric vehicle schematic verter consists of a single insulated gate bipolar transistor
(IGBT) switch S1 , an inductor L 1 with its internal resistance
R1 and capacitor Cout to filter output voltage.
ing; therefore, a secondary power source is necessary [1,7,8].
For this reason hybridization of FC is done with other energy 2.2 Battery
storage units (ESUs) to capture regenerative braking energy
and to provide peak power. The ESUs usually consist of the The battery is connected to DC bus voltage via two-switch
battery module or UC module or a combination of both [9– buck–boost converter mainly because the flow of current can
11]. be negative or positive depending upon the demand and state
The comparative analysis performed in [12–14] and [15] of charge (SOC) [10]. The converter used for battery consists
shows that hybrid vehicles with FC as primary source and bat- of an inductor L 2 with its internal resistance R2 to filter the
tery and UC as ESUs could lead to more real-world solutions. current and two IGBT switches S2 and S3 that are driven by
Batteries have high energy density, but have limited life span input signals u 2 and u 3 applied to their gates.
and low power density, whereas UC has high power density
but low energy density and can discharge quickly. Therefore, 2.3 Ultracapacitor
utilizing both batteries and UCs seems to be a more practical
choice as it could not only provide high power density, but As we know that the current flowing through UC can be posi-
also increase the energy storage. tive or negative, so we used two-switch buck–boost converter
All ESUs have different voltage levels and dynamic char- [16]. The converter has an inductor L 3 with its internal resis-
acteristics. To ensure dynamic energy exchange between tance R3 and two IGBT switches S4 and S5 that are being
them, various converter topologies and control techniques operated by applying input signals u 4 and u 5 .
are presented [16–19]. The general structure of FC–battery–
UC-based hybrid energy storage system (HESS) is shown in
Fig. 1. Most of the control techniques used to study FC-based 3 Energy Management Strategy
HESS are linear [10,18,20–22] and can only guarantee sta-
bility for small region [16]. However, we know that DC–DC Different energy management strategies for HESS have been
converters and ESUs are nonlinear in nature and conventional proposed in the literature, for instance, [15,25–27] and [28].
linear controllers cannot satisfy the performance require- These strategies have the following advantages: (1) improv-
ments. In this paper a nonlinear control technique known ing the efficiency and reliability of the EVs by storing the
as Lyapunov redesign is applied to achieve the following extra energy recovered from regenerative braking and (2)
goals: (1) accurate DC voltage bus regulation under differ- efficiently delivering the energy under different load condi-
ent load conditions, (2) smooth tracking of UC and battery tions.
current to their respective reference values and (3) stabil- EVs’ load can be classified as constant and transient loads.
ity of the whole system. The performance of the developed Constant load does not have any peak demand such as air
controller is then tested by numerous numerical simulations conditioning or on-board electric load. These loads always
using MATLAB/Simulink. remain constant and are completely compatible with FC. The
transient load includes acceleration, braking and decelera-
tion. FC alone cannot respond to these loads because of cold
2 Electric Circuit Structure start-up and slow dynamic response; thus, a combination of
battery and UC will be used to solve this problem [20]. The
An ideal HESS for electric vehicles (EVs) should have high detailed energy management strategy can be expressed as
power density, longer lifetime, less cost and high efficiency. follows:

123
Author's personal copy
Arabian Journal for Science and Engineering (2018) 43:3123–3133 3125

4 Modeling of Power Sources

4.1 FC Modeling

PEMFC is a primary source in hybrid electric vehicles and is


used to supply steady-state load demands. The ideal potential
of single FC is 1.229 V. However, the actual potential is lower
than ideal potential because of voltage drops. There are three
types of major losses in FC [29]: (1) activation losses, (2)
ohmic losses and (3) concentration losses. At low current
density, the voltage drop is due to activation losses. Moreover,
concentration losses occur at high current density. A typical
V–I curve of FC is shown in Fig. 4 [29], where the voltage
drop can be observed as the current density changes. The
output voltage of FC can be written as:

Voutfc = E − Vact − Vcon − Vohmic , (1)

where Vact , Vcon and Vohmic are the activation, concentra-


tion and ohmic overvoltage, respectively, and E is the Nernst
Fig. 2 Fuel cell–battery–UC HESS instantaneous voltage.
An equivalent electric circuit model of FC based on charge
double-layer capacitor is shown in Fig. 5. The activation
and concentration losses are represented by equivalent series
resistances Ract and Rcon . The ohmic voltage drop is modeled
by equivalent series resistance Rohmic , and C represents the
double-layer equivalent capacitance. Equations to calculate
Ract , Rcon and Rohmic are mentioned in [30].

4.2 Battery Modeling

A lithium-ion battery pack is used as the secondary source


Fig. 3 Energy management strategy: Red arrow suggests power flow for both supplying high load together with FC and capturing
during low-load condition, green arrow suggests power flow during breaking energy. The dynamic model of battery is shown in
high-load condition, and blue arrow suggests power flow during tran- Fig. 6 [31]. It consists of a voltage-controlled source, a series
sient condition
resistance and two RC networks. Rs represents the instanta-
neous voltage drop in the battery. RC networks represent the
transient response in the battery internal impedance.
In low-load conditions, FC will continuously supply the The output voltage of battery can be expressed as:
required power. Since the efficiency of FC is optimum at
low-load demands, it can also be used to charge battery or
Vbattery = Voc − i battery Z eq , (2)
UC depending on SOC. In high-load conditions, FC cannot
be used alone because of fuel starvation phenomenon [29].
Hence, the combination of FC and battery will be used to where Voc is the open-circuit voltage, i battery is battery cur-
supply the required power. At transient conditions or fast rent, and Z eq is battery equivalent internal resistance. The
peak power, UC will take care of load as it has the ability battery open-circuit voltage is dependent on SOC and can be
to provide a large amount of power instantaneously. During expressed as [31]:
start-up condition, battery will provide the power because of
slow start-up response of FC and fast self-discharging char-
acteristics of the UC. All the power flow modes are shown Voc = −1.031 × e(−35×SOC) + 3.685 + 0.2156
in Fig. 3. × SOC − 0.1178 × SOC2 + 0.321 × SOC3 . (3)

123
Author's personal copy
3126 Arabian Journal for Science and Engineering (2018) 43:3123–3133

Fig. 6 Equivalent electrical circuit of battery

Fig. 4 V–I characteristic of PEMFC

Fig. 7 Equivalent electrical circuit of UC

5 Modeling of DC–DC Converters

5.1 Boost Converter (FC)


Fig. 5 Equivalent electrical model of FC
A unidirectional boost converter is used to connect FC to
DC voltage bus to prevent the current flow in reverse direc-
And SOC can be written as: tion which may cause damage to the FC. Figure 2 shows the
 simple configuration of boost converter which consists of a
i battery high-frequency inductor L 1 with nonlinearity R1 , an IGBT
SOC = SOCinit − dt, (4)
Cbattery switch, diode D1 and output filter capacitor C out .
There are two states of converter depending on the position
where SOCinit is the initial state of charge. of switch S1 . When the switch S1 is turned on by applying
signal at gate, the diode will not conduct and inductor cur-
4.3 UC Modeling rent will be equal to source current, and the same current
will flow from switch; at this time, the inductor will be stor-
The UC is used to supply power during transient load. Fig- ing the energy and load current will solely be provided by
ure 7 shows the dynamic model of UC. The model consists output capacitor Cout . When the switch S1 is off, diode will
of series equivalent resistance Rsc and capacitance Cuc rep- be forward-biased and the inductor current will be passing
resenting charging and discharging. Self-discharging losses through diode and charging the capacitor which was dis-
are represented by equivalent parallel resistance Rpc [32]. charged during previous state.
The voltage level of UC can be expressed as: Pulse width modulation (PWM) is used to control the
switch, and it can be one or zero. The mathematical model
 
− CuctRsc
of boost converter shown in Fig. 2 will be:
Vuc (t) = Vinit e . (5)
di fc vfc R1 vout
= − i fc − (1 − u 1 ) (7)
The amount of energy drawn from UC can be expressed as: dt L1 L1 L1
dv out i fc i1
  = (1 − u 1 ) − , (8)
1 dt Cout Cout
E uc = Cuc Vinit
2
− Vfinal
2
, (6)
2
where vfc is the voltage of FC; vout is the DC bus voltage;
where Vinit is the initial voltage before the discharging starts and i fc and i 1 are the inductor current and output current,
and Vfinal is final voltage after the discharging starts. respectively.

123
Author's personal copy
Arabian Journal for Science and Engineering (2018) 43:3123–3133 3127

5.2 Buck–Boost Converter (Battery) battery. The mathematical model will be similar as well,
therefore, by defining K as:
The converter that connects the battery to DC bus voltage
is a two-switch bidirectional buck–boost converter. Figure 2 
1 if i ucref > 0
shows the basic configuration of the buck–boost converter. K = (17)
0 if i ucref < 0.
It consists of a high-frequency inductor L 2 with nonlinearity
R2 and two IGBT switches S2 and S3 . The IGBT switches
And the virtual control is given by:
are controlled by applying gate signals u 2 and u 3 . When S2
is on and S3 is off, the converter operates as boost and battery
will discharge (i bat > 0), and when S3 is on and S2 is off, u 45 = [ K (1 − u 4 ) + (1 − K ) u 5 ]. (18)
the converter operates as buck and battery will be charged
(i bat < 0). One can define a variable H as follows: The global UC converter modeling can be expressed using
 following pair of differential equations:
1 if i batref > 0
H= (9)
0 if i batref < 0, di uc vuc R3 vout
= − i uc − u 45 (19)
dt L3 L3 L3
where i batref is the reference battery current. During boost
i 3 = u 45 i uc . (20)
operation, when S2 is on and S3 is off, the mathematical
model by using Kirchhoff’s voltage rule will be:
5.4 Global System Modeling
di bat vbat R2 vout
= − i bat − (1 − u 2 ) . (10)
dt L2 L2 L2 From Fig. 2 we can obtain the following equation by applying
Kirchhoff’s current law
And the output current will be:
i L = i1 + i2 + i3 . (21)
i 2 = (1 − u 2 )i bat . (11)

During buck operation when S3 is on and S2 is off, the Putting value of i 2 and i 3 from Eqs. (16) and (20)
mathematical model by using Kirchhoff’s voltage rule will
be: i 1 = i L − u 23 i bat − u 45 i uc . (22)
di bat vbat R2 vout
= − i bat − u 3 . (12) Putting the value of i 1 in Eq. (8) gives
dt L2 L2 L2

And the output current will be: dv out i fc iL i bat i uc


= (1 − u 1 ) − + u 23 + u 45 . (23)
dt Cout Cout Cout Cout
i 2 = u 3 i bat . (13)
The overall differential equations for the whole system
A global model can be obtained by defining virtual control will be:

u 23 = [ H (1 − u 2 ) + (1 − H ) u 3 ]. (14) di fc vfc R1 vout


= − i fc − (1 − u 1 ) (24)
dt L1 L1 L1
Hence, the battery converter model can be presented by di bat vbat R2 vout
the following pair of differential equations: = − i bat − u 23 (25)
dt L2 L2 L2
vbat vout di uc vuc R3 vout
di bat
= −
R2
i bat − u 23 (15) = − i uc − u 45 (26)
dt L2 L2 L2 dt L3 L3 L3
i 2 = u 23 i bat . dv out i fc iL i bat i uc
(16) = (1 − u 1 ) − + u 23 + u 45 .
dt Cout Cout Cout Cout
5.3 Buck–Boost Converter (UC) (27)

Since UC has both the discharging and charging properties, For control design it would be more appropriate to use
current will flow in both directions. Thus, the two-switch average control model, so by averaging the model over one
buck–boost converter can be used as the one we used with switching cycle, the overall system can be expressed as:

123
Author's personal copy
3128 Arabian Journal for Science and Engineering (2018) 43:3123–3133

dx 1 vfc R1 x4 ics of error e1 can be defined by taking derivative of Eq. (33)


= − x1 − (1 − β1 ) (28)
dt L1 L1 L1 as
dx 2 vbat R2 x4
= − x2 − β23 (29)
dt L2 L2 L2 e˙1 = x˙1 − I˙fcref . (37)
dx 3 vuc R3 x4
= − x3 − β45 (30)
dt L3 L3 L3
Putting the value of x˙1 from Eq. (28) gives
dx 4 x1 iL x2 x3
= (1 − β1 ) − + β23 + β45 , (31)
dt Cout Cout Cout Cout
vfc R1 x4
e˙1 = − x1 − (1 − β1 ) − I˙fcref . (38)
where x1 is the average FC current i fc , x2 is the average L1 L1 L1
battery current i bat , x3 is the average UC current i uc , x4 is
the average DC bus voltage v out , and β1 , β23 and β45 are the The error e1 must approach exponentially zero, so it
average values of control signal. enforces e˙1 to behave as follows:

e˙1 = −c1 e1 + e4 , (39)


6 Nonlinear Control and Stability Analysis
where c1 is a design parameter with value greater than 1
A Lyapunov-based nonlinear controller has been designed to
and e4 is the error between DC bus voltage and reference
achieve the following objectives:
voltage. By investigating Eqs. (38) and (39), the control law
is obtained as:
1. Accurate DC bus voltage regulation under varying load
2. Rapid tracking of battery current to its reference current   
L1 vfc R1
value β1 = 1 − c1 e1 − e4 + − x1 − I˙fcref .
3. Rapid tracking of UC current to its reference current value x4 L1 L1
4. Asymptotic stability of whole system to the required ref- (40)
erence value
In Eq. (40), e4 is a damping term and its dynamics will be
For perfect DC voltage bus regulation, v out must track studied later.
the reference voltage, but it cannot be done due to boost con- Similarly, the dynamics of error e2 and e3 must approach
verter’s non-minimum phase. So, to deal with this issue indi- zero for the tracking of battery and UC current to their ref-
rect voltage regulation approach has been used. This method erence values.
involves forcing the output voltage regulation through input
current regulation. Hence, i fc will be enforced to track its ref-
e˙2 = x˙2 − I˙batref (41)
erence value i fcref , which indirectly enforces v out to track its
reference value v outref . The FC reference current is related e˙3 = x˙3 − I˙ucref . (42)
to DC reference voltage by means of the following equation:
  Putting values of x˙2 and x˙3 from Eqs. (29) and (30) gives
Voutref i L − vuc i ucref − vbat i batref
Ifcref = γ , (32)
vfc vbat R2 x4
e˙2 = − x2 − β23 − I˙batref (43)
where γ is ideality factor and must be greater than 1 to con- L2 L2 L2
vuc R3 x4
sider all the losses including switching losses and the losses e˙3 = − x3 − β45 − I˙ucref . (44)
in the inductances ESR (R1 , R2 and R3 ). L3 L3 L3
To carry out our control objectives the following errors
are defined: The errors e2 and e3 must approach exponentially zero, so
it enforces e˙2 and e˙3 to behave as follows:
e1 = x1 − Ifcref (33)
e2 = x2 − Ibatref (34) e˙2 = −c2 e2 (45)
e3 = x3 − Iucref (35) e˙3 = −c3 e3 , (46)
e4 = x4 − voutref . (36)
where c2 and c3 are design parameters and their values must
For the DC voltage regulation and tracking of battery and be greater than 1. The control laws β23 and β45 can now
UC current, all the errors must converge to zero. The dynam- easily be obtained as follows:

123
Author's personal copy
Arabian Journal for Science and Engineering (2018) 43:3123–3133 3129
 
L2 vbat R2 Putting the value of x˙4 from Eq. (31) gives
β23 = c2 e2 + − x2 − I˙batref (47)
x4 L2 L2
  x1 iL x2
L3 vuc R3 e˙4 = (1 − β1 ) − + β23
β45 = c3 e3 + − x3 − I˙ucref . (48) Cout Cout Cout
x4 L3 L3
x3
+ β45 − v̇outref . (55)
The terms I˙fcref , I˙batref and I˙ucref present in control laws Cout
depict the change in reference currents which are essentially The error e4 must approach zero and behave as suggested in
bounded by current limiters. Eq. (52). Voutref will be generated according to the following
After obtaining all the control laws β1 , β23 and β45 , the control law:
next step is now to study the stability of the whole system.
The following Lyapunov function candidate is considered: x1 iL
v̇outref = c4 e4 + e1 + (1 − β1 ) −
Cout Cout
1 2 1 2 1 2 1 2 x2 x3
V = e1 + e2 + e3 + e4 . (49) + β23 + β45 (56)
2 2 2 2 Cout Cout
1 x1 iL
For the system to be globally asymptotically stable, the voutref = (c4 e4 + e1 + (1 − β1 ) −
s Cout Cout
derivative of V must be negative definite. Thus, taking the x2 x3
derivative of Eq. (49) gives + β23 + β45 ). (57)
Cout Cout
V̇ = e1 e˙1 + e2 e˙2 + e3 e˙3 + e4 e˙4 . (50)
7 Simulation Results
Putting values of e˙1 , e˙2 and e˙3 from Eqs. (39), (45) and
(46) The performance of the controller is tested under European
Extra Urban Driving Cycle (EUDC). EUDC represents the
V̇ = −c1 e1 2 − c2 e2 2 − c3 e3 2 + e4 (e1 + e˙4 ) . (51) driving condition in a city and known for high-speed driv-
ing mode. The full EUDC test lasts for 400 s with a vehicle
For Eq. (51) to be negative definite, the dynamics of e4 maximum speed of 90 km/h and average speed of 63 km/h
must behave as follows: [33].
Parameters of power sources used in the simulation are
e˙4 = −c4 e4 − e1 , (52) shown in Table 1. The parameters of the HESS are shown in
Table 2. Figure 8 shows the vehicle speed in km/h, and Fig. 9
where c4 > 0 is the design parameter. Now we can rewrite
the derivative V̇ of the Lyapunov function as
Table 1 Parameters of power sources
V̇ = −c1 e1 − c2 e2 − c3 e3 − c4 e4 .
2 2 2 2
(53) Component Specification

PEMFC 350 V, 250 A, 34 KW


This suggests that V̇ is negative definite and globally asymp-
UC module 205 Vdc, 2700 F
totically stable at the equilibrium. It should be noted that e4 is
Battery module 288 Vdc, 13.9 Ah, Li-ion
the error introduced between the desired and actual values of
DC link voltage, while e1 , e2 and e3 are errors between the ref-
erence and actual values of energy sources’ currents. Hence, Table 2 Parameters of the simulated system
the convergence of error e4 explicitly depends on the conver- Parameter Value
gence of errors e1 , e2 and e3 . The introduction of e4 enables
an indirect restrain on other errors. Moreover, if any one of Inductances L 1 , L 2 and L 3 3.3 mH
the errors e1 , e2 or e3 diverges, the term e4 will diverge, mak- Equivalent series resistances R1 , R2 and R3 20 mΩ
ing the whole Lyapunov function positive definite, resulting Output capacitor Cout 1.66 mF
in an indication of system instability. Under normal situa- Switching frequency 100 kHz
tions, the error e4 will be converging, hence behaving as an Car mass 1066 kg
additional damping term. Passenger weight (× 2) 70+70 Kg
The reference value of DC bus voltage is generated from Aerodynamic drag coefficient 0.19
the dynamics of error e4 Rolling resistance coefficient 0. 0048
Front area 1.8 m 2
e˙4 = x˙4 − v̇outref . (54)

123
Author's personal copy
3130 Arabian Journal for Science and Engineering (2018) 43:3123–3133

Fig. 8 Vehicle speed under EUDC

Fig. 10 Overall block diagram of the proposed controller

Fig. 9 Load current under EUDC

shows the load current that is calculated using the following


formula:
  Fig. 11 Block diagram of input signals u 2 and u 3
1 dvt
Pout = 0.5 pa vt AC x + MgCr + M
2
vt (58)
0.75 dt
 
1 dvt
i out = 0.5 pa vt2 AC x + MgCr + M vt ,
(0.75)(400) dt
(59)

where vt is the vehicle speed, M is the mass of the car, A is


the car frontal area, Cr is the rolling resistance coefficient, pa
is the air density, and C x is the aerodynamic drag coefficient.
It is assumed that inverter is operating at 75% efficiency and
DC bus voltage is regulated to 400 V. Fig. 12 Block diagram of input signals u 4 and u 5
The block diagram in Fig. 10 shows the HESS and the
applied nonlinear controller. Figure 11 shows the block dia-
gram that generates the input signals u 2 and u 3 from u 23 and For the system to be asymptotically stable, all errors must
i batref . Figure 12 shows the block diagram that generates the converge toward zero exponentially as shown in Fig. 13. Fig-
input signals u 4 and u 5 from u 45 and i ucref . ure 14 shows that the DC voltage is regulated to 400 V even
The constants c1 , c2 and c3 are the control design coef- when the load current is continuously varying. There is little
ficients. The numerical values selected for these parameters bit fluctuation in DC voltage as it jumps between 390 and
are c1 = 1000, c2 = 1000 and c3 = 100. The ideal- 410 V, but it is under acceptable range. Figure 15 shows that
ity factor γ is chosen to be 1.0014. It must be known that FC current is accurately tracked to its reference current value.
design parameters could only have positive values, but there Having UC and battery as secondary sources helps us to limit
is no any systematic method to find these values. The com- the FC current.
mon practice is to use trial-and-error method which consists During vehicle deceleration, the load current is negative
of randomly increasing the parameter values until we get the and can be used to charge UC or battery or both. It is more
desired response. appropriate to charge UC first because of its fast transient

123
Author's personal copy
Arabian Journal for Science and Engineering (2018) 43:3123–3133 3131

Fig. 15 FC current with its reference

Fig. 16 UC current with its reference

Fig. 13 Error functions with transient response

Fig. 17 Battery current with its reference

relatively small steady-state error. It can also be seen that FC,


UC and battery currents are tracking their reference values
accurately with very small transient state. Simulating output
Fig. 14 DC bus voltage with its reference
voltages of fuel cell, battery and ultracapacitor power sources
are shown in Figs. 19, 20 and 21, respectively. Finally, the
response. Figure 16 shows the charging of UC during such control signals u 1 , u 23 and u 45 are shown in Fig. 22.
an event. It can be seen that UC current is accurately tracking To evaluate the efficacy of the proposed controller, a 3-
its reference value. s load variation test is also performed on this system. The
Figure 17 shows the charging and discharging of the bat- performance of the proposed controller is compared with the
tery depending on load demand. A current limiter is used conventional PI controller. Under this test, vehicle is assumed
with battery to increase its lifetime. It is also observable that to be cruising at constant speed of 92.6 km/hr, while load
during high-load demand and acceleration, battery and FC torque and corresponding power requirements are varied to
are working simultaneously. replicate dynamic conditions. Figure 23 shows the compari-
Figure 18 presents the zoom-in window showing transient son of controller performances for tracking DC bus voltage
response and local convergence of DC bus voltage and FC, to 400 V. It can be observed that the proposed controller has
UC and battery currents. Note that the proposed controller provided us with the potential to quickly track the controlled
maintains the DC bus voltage close to reference value with a system while keeping dynamic stability intact.

123
Author's personal copy
3132 Arabian Journal for Science and Engineering (2018) 43:3123–3133

Fig. 18 Transient analysis of DC bus voltage and FC, UC and battery


currents

Fig. 19 Fuel cell source voltage

Fig. 22 Control signals u 1 , u 23 and u 45

Fig. 20 Battery source voltage

Fig. 23 Comparison with PI controller

Fig. 21 Ultracapacitor source voltage between sources and load, three power converters have been
used. A nonlinear control technique ‘Lyapunov redesign’
is applied to accomplish the control objectives. The perfor-
8 Conclusions mance of the proposed controller is validated under EUDC
by simulating it on MATLAB/Simulink. Simulation results
This paper mainly discusses the modeling and nonlinear con- indicated the fast tracking of FC, battery and UC currents to
trol of HESS used in FCEVs. The energy system proposed their reference values and tight DC bus voltage regulation.
consists of PEMFC, battery and UC. PEMFC is used as the The stability analysis of closed loop system is performed by
main power source, and battery and UC are used as secondary using Lyapunov stability theory. For further work a compar-
power sources. In order to achieve dynamic energy exchange ison can be made with other control methods for the better

123
Author's personal copy
Arabian Journal for Science and Engineering (2018) 43:3123–3133 3133

analysis of proposed controller. Moreover, a research can 17. Thounthong, P.; Pierfederici, S.; Martin, J.-P.; Hinaje, M.; Davat,
also be done on energy management strategies to find the B.: Modeling and control of fuel cell/supercapacitor hybrid source
based on differential flatness control. IEEE Trans. Veh. Technol.
reference values for FC, battery and UC currents. 59(6), 2700–2710 (2010)
18. Enang, W.; Bannister, C.: Modelling and control of hybrid electric
vehicles (a comprehensive review). Renew. Sustain. Energy Rev.
References 74, 1210–1239 (2017)
19. Song, Z.; Hou, J.; Hofmann, H.; Li, J.; Ouyang, M.: Sliding-mode
and Lyapunov function-based control for battery/supercapacitor
1. Chan, C.C.; Wong, Y.S.: Electric vehicles charge forward. IEEE hybrid energy storage system used in electric vehicles. Energy 122,
Power Energy Mag. 2(6), 24–33 (2004) 601–612 (2017)
2. Chan, C.C.: The state of the art of electric, hybrid, and fuel cell 20. Wong, J.H; Idris, N.R.N; Anwari, M.; Taufik, T.: A parallel energy-
vehicles. Proc. IEEE 95(4), 704–718 (2007) sharing control for fuel cell-battery-ultracapacitor hybrid vehicle.
3. Momoh, O.D.; Omoigui, M.O.: An overview of hybrid electric In: 2011 IEEE energy conversion congress and exposition, pp.
vehicle technology. In: 2009 IEEE vehicle power and propulsion 2923–2929. IEEE (2011)
conference, pp. 1286–1292. IEEE (2009) 21. Shuai, L.; Corzine, K.A.; Ferdowsi, M.: A new bat-
4. Mebarki, N.; Rekioua, T.; Mokrani, Z.; Rekioua, D.; Bacha, S.: Pem tery/ultracapacitor energy storage system design and its motor
fuel cell/battery storage system supplying electric vehicle. Int. J. drive integration for hybrid electric vehicles. IEEE Trans. Veh.
Hydrog. Energy 41(45), 20993–21005 (2016) Technol. 56(4), 1516–1523 (2007)
5. Yilmaz, M.; Krein, P.T.: Review of battery charger topologies, 22. Akar, F.; Tavlasoglu, Y.; Ugur, E.; Vural, B.; Aksoy, I.: A bidirec-
charging power levels, and infrastructure for plug-in electric and tional non-isolated multi input dc-dc converter for hybrid energy
hybrid vehicles. IEEE Trans. Power Electron. 28(5), 2151–2169 storage systems in electric vehicles. IEEE Trans. Veh. Technol.
(2013) 65(10), 7944–7955 (2015)
6. Marx, N.; Boulon, L.; Gustin, F.; Hissel, D.; Agbossou, K.: A 23. Khaligh, A.; Li, Z.: Battery, ultracapacitor, fuel cell, and hybrid
review of multi-stack and modular fuel cell systems: interests, energy storage systems for electric, hybrid electric, fuel cell, and
application areas and on-going research activities. Int. J. Hydrog. plug-in hybrid electric vehicles: State of the art. IEEE Trans. Veh.
Energy 39(23), 12101–12111 (2014) Technol. 59(6), 2806–2814 (2010)
7. Kisacikoglu, M.C.; Uzunoglu, M.; Alam, M.S.: Fuzzy logic control 24. Tahri, A.; El Fadil, H.; Giri, F.; Chaoui, F.-Z.: Nonlinear control and
of a fuel cell/battery/ultra-capacitor hybrid vehicular power system. observation of a boost converter associated with a fuel-cell source
In: 2007 IEEE vehicle power and propulsion conference, pp. 591– in presence of model uncertainty. IFAC Proc. Vol. 47(3), 575–580
596. IEEE (2007) (2014)
8. Payman, A.; Pierfederici, S.; Meibody-Tabar, F.: Energy control 25. Sabri, M.F.M.; Danapalasingam, K.A.; Rahmat, M.F.: A review
of supercapacitor/fuel cell hybrid power source. Energy Convers. on hybrid electric vehicles architecture and energy management
Manag. 49(6), 1637–1644 (2008) strategies. Renew. Sustain. Energy Rev. 53, 1433–1442 (2016)
9. Gao, W.: Performance comparison of a fuel cell-battery hybrid 26. Zhang, S.; Xiong, R.; Sun, F.: Model predictive control for power
powertrain and a fuel cell-ultracapacitor hybrid powertrain. IEEE management in a plug-in hybrid electric vehicle with a hybrid
Trans. Veh. Technol. 54(3), 846–855 (2005) energy storage system. Appl. Energy 185, 1654–1662 (2015)
10. Trilaksono, B.R.; Sasongko, A.; Rohman, A.S.; Dronkers, C.J.; 27. Olatomiwa, L.; Mekhilef, S.; Ismail, M.S.; Moghavvemi, M.:
Ortega, R. et al.: Model predictive control of hybrid fuel Energy management strategies in hybrid renewable energy sys-
cell/battery/supercapacitor power sources. In: 2012 international tems: A review. Renew. Sustain. Energy Rev. 62, 821–835 (2016)
conference on system engineering and technology (ICSET), pp. 28. Payman, A.; Pierfederici, S.; Meibody-Tabar, F.: Energy manage-
1–6. IEEE (2012) ment in a fuel cell/supercapacitor multisource/multiload electrical
11. Schaltz, E.; Khaligh, A.; Rasmussen, P.O.: Investigation of bat- hybrid system. IEEE Trans. Power Electron. 24(12), 2681–2691
tery/ultracapacitor energy storage rating for a fuel cell hybrid (2009)
electric vehicle. In: 2008 IEEE vehicle power and propulsion con- 29. Hoogers, G.: Fuel cell technology handbook. CRC Press, Boca
ference, pp. 1–6. IEEE (2008) Raton (2002)
12. Languang, L.; Han, X.; Li, J.; Hua, J.; Ouyang, M.: A review on the 30. Nehrir, M.H.; Wang, C.: Modeling and control of fuel cells: dis-
key issues for lithium-ion battery management in electric vehicles. tributed generation applications, vol. 41. Wiley, Hoboken (2009)
J. Power Source 226, 272–288 (2013) 31. Chen, M.; Rincon-Mora, G.A.: A Rincon-Mora. Accurate electrical
13. Schaltz, E.; Khaligh, A.; Rasmussen, P.O.: Influence of bat- battery model capable of predicting runtime and iv performance.
tery/ultracapacitor energy-storage sizing on battery lifetime in a IEEE Trans. Energy Convers. 21(2), 504–511 (2006)
fuel cell hybrid electric vehicle. IEEE Trans. Veh. Technol. 58(8), 32. Uzunoglu, M.; Alam, M.S.: Modeling and analysis of an fc/uc
3882–3891 (2009) hybrid vehicular power system using a novel-wavelet-based load
14. Bauman, J.; Kazerani, M.: A comparative study of fuel-cell-battery, sharing algorithm. IEEE Trans. Energy Convers. 23(1), 263–272
fuel-cell-ultracapacitor, and fuel-cell-battery-ultracapacitor vehi- (2008)
cles. IEEE Trans. Veh. Technol. 57(2), 760–769 (2008) 33. Larminie, J.; Lowry, J.: Electric vehicle technology explained.
15. Zandi, M.; Payman, A.; Martin, J.-P.; Pierfederici, S.; Wiley, Hoboken (2004)
Davat, B.; Meibody-Tabar, F.: Energy management of a fuel
cell/supercapacitor/battery power source for electric vehicular
applications. IEEE Trans. Veh. Technol. 60(2), 433–443 (2011)
16. El Fadil, H.; Giri, F.; Guerrero, J.M.; Tahri, A.: Modeling and
nonlinear control of a fuel cell/supercapacitor hybrid energy stor-
age system for electric vehicles. IEEE Trans. Veh. Technol. 63(7),
3011–3018 (2014)

123

You might also like