Sustainable Energy Technologies and Assessments 52 (2022) 102121

Contents lists available at ScienceDirect

Sustainable Energy Technologies and Assessments

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

Integration of hydrogen technology and energy management comparison

for DC-Microgrid including renewable energies and energy storage system
Oussama Hafsi a, *, Othmane Abdelkhalek a, Saad Mekhilef b, d, Mohamed Amine Soumeur a,
Mohamed Amine Hartani a, c, Abdeselem Chakar a
a
Laboratory of Smart Grids & Renewable Energies SGRE-L, Faculty of Technology, Department of Electrical Engineering, Tahri Mohamed University Bechar, B.P 417,
08000, Algeria
b
School of Science, Computing and Engineering Technologies, Swinburne University of Technology, Hawthorn, VIC 3122, Australia
c
Sustainable Development and Computer Science Laboratory SDCS-L (Ahmed Draia University - Adrar), Algeria
d
Power Electronics and Renewable Energy Research Laboratory PEARL, Department of Electrical Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala
Lumpur, Malaysia

A R T I C L E I N F O A B S T R A C T

Keywords: Microgrids-based distributed energy resources DERs have an electrical revolution, especially in the electrification
DC-Microgrid zones and clean energy. This paper proposes a comparative energy management study of an isolated direct
Renewable energies current micro-grid DCMG, which consists of two primary sources, a photovoltaic PV and wind turbine WT
DC/DC electronic converters
generator working at the maximum power point MPP to extract the maximum available, then an energy storage
Energy storage system
Energy management system and control
system ESS-based lead-acid batteries to assuring the DC-bus stability. A backup device based on Proton Exchange
Membrane Fuel Cell PEMFC to ensure the system’s reliability avoids the blackout scenario, adding to power
electronic converters for the flexible relationship of the system elements, and a community residential as hybrid
loads. This comparative study is based on classical techniques: classical PI CPI and state machine strategy SM;
intelligent methods such as fuzzy logic rule FLC and artificial neural network ANN; moreover, two optimization
methods equivalent consumption minimization strategy ECMS and external energy maximization strategy EEMS.
The full Matlab/Simulink model has been developed to check the performance of the proposed system and
energy management strategies for actual load and weather conditions. Simulation results based comparison show
that the optimization methods are enhanced the system performance, DC-bus stability and minimize the overall
system cost by reducing the hydrogen consumption.

Introduction many homes and industrial loads like LED, electronic devices, electric
vehicles, EV, data centers, electrolyzes …etc.) [4–6]. In addition, his
Due to the energy policy of the world, which aims to exploit clean integrated system can be connected without DC/AC conversions, which
and renewable energy and decrease carbon dioxide emission, the means more efficiency, moreover in DCMG no need for the synchroni­
microgrid MG-based distributed renewable generation is becoming an zation, frequency, the flow of reactive power, and harmonic current
optimal solution for future smart grids [1]. control [4,5,7]. DC MGs are mainly used for electrical systems, data
Microgrids are smaller local electricity networks that are capable of centers, communications stations, and plug-ins electric vehicles [8].
acting individually. Separately from the primary energy [2], it can be Among the sources of PV and WT are the Most-motivated electrifi­
divided into DCMG and current alternative ACMG. DCMG collects cation technical challenges Consumption by remote zones. Therefore,
distributed generator DG, ESS, and loads where the current sharing in­ different control algorithms are suggested to obtain maximum power in
direct form in the overall system can be in isolated form or an on-grid the changing conditions of the weather. The tracking of the maximum
structure [2,3]. Compared to ACMG, DCMG has The advantages of point power (MPPT) is used within an algorithm to optimize extraction
high performance, more compatible with renewable energy sources RES power by implementing it with DC/DC converter control to maximize
(i.e., PV and FC), ESS system (as batteries and supercapacitor SC), and the performance of the PV and wind systems [9]. However, depending

* Corresponding author.
E-mail addresses: oussama5bm@gmail.com, hafsi.oussama@univ-bechar.dz (O. Hafsi).

https://doi.org/10.1016/j.seta.2022.102121
Received 31 August 2021; Received in revised form 10 February 2022; Accepted 26 February 2022
Available online 10 March 2022
2213-1388/© 2022 Elsevier Ltd. All rights reserved.
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

on climate conditions, the output power of the PV and wind generators common coupling (PCC).
varies considerably. Due to battery capacity and scale limitations, this The diesel generator is a good solution for an autonomous system for
type of DC microgrid is insufficient for the long-term provision of re­ electrification. In [23,24] authors proposed a DCMG based on PV/WT/
quirements for energy [10,11]. DG/Battery and supercapacitor SC where the DG works as a backup
For this reason, it should be researched on a suitable source to ensure generator in off-grid mode. However, the DG has a lent response time, so
power continuously. Because of the diesel generator’s negative in­ the SC compensates the power balance in this temporary period.
fluences such as high continuation costs, carbon dioxide emissions, and In line with the world’s energy policy, the researchers are going to
Low scalability to meet changes in load requirements, the FC sources another suitable, clean, and reliable source [25], which can decrease the
have recently become an alternative to the diesel generators DG. Be­ CO2 emissions and ensure the power demand. In [26], Hing Han et al.
sides, it has many benefits that have been described, such as the rapid are achieving hierarchical energy management for an isolated DCMG
speed of Load-response, efficiency, the flexibility of fuel, modular output based PV/FC/Battery using two layer control, and the first is local
[12,13]. controllers where the second control is based on ECMS optimization
The utilization of proton exchange membrane fuel cell PEMFC in strategy. Moreover, Mohd Alam and al. [14] are developing a DCMG
Microgrids application and EV is more interesting. A vital issue linked based on a PV system as a primary source and FC-Electrolyzer/battery as
with autonomous Microgrids is the energy management system EMS. a hybrid energy storage, adding to coordinated EMS to save the FC and
This last should be designed to smooth the power exchanged between minimize the hydrogen consumption. In [27], a novel power manage­
sources, ESS, and loads to satisfy the load demand and the ESS con­ ment method based on adaptive fractional fuzzy sliding mode control
straints, add to some objectives represented in optimization issues like (AFFSMC) is developed for a hybrid energy system connected to the grid
minimizing the MG operating costs and increasing the MG equipment containing PV as the primary source and FC as auxiliary source super­
lifetime [14,15]. capacitor SC and battery based on hybrid storage systems. They con­
Fuel cells are being employed for both power generating and storage trasted their control technique to a traditional control method based on
applications. The choice of fuel cell technology and fuel and the heating the PI controller in a series of time-domain simulations. When Yuchen
and cooling of the stacks are the key differences in these fuel cell sys­ Pu and co-authors [28] suggested a hierarchical state machine energy
tems. Stationary fuel cells can be utilized as a significant power source. management control based on the lowest usage cost for a DC microgrid
They’re frequently used for power plants that aren’t connected to the with PV/FC/battery/electrolyzer. In [29] authors are studying a hybrid
grid and provide extra energy [16,17]. photovoltaic and wind turbine energy systems; an attempt is made to
When the primary power source is interrupted, backup power supply replace the conventional diesel-only backup system with a combination
systems offer electricity. A fuel cell/electrolyzer system is an excellent hydrogen-and-diesel backup system. They are making an optimization
alternative for backup power applications because it can generate study based on econo-environmental analysis by HOMER software.
hydrogen on demand. MPPT-droop dual-mode distributed control is employed at the device
To enhance the EMS performance of the DCMG and ensure the sys­ level, while an ECMS for distributed system net power between the
tem reliability and durability; many DCMG topologies and EMS are battery pack and the fuel cell is used at the system level. A Fuzzy Logic
developed and designed. Some papers are studying a DCMG with the Controller (FLC) for energy management strategy employing the Pro­
same topology PV/battery, in [7] the researcher suggested. Energy gram Logic Controller (PLC) unit is recommended in [30] for a PV/WT/
management between some DCMGs to avoid the backup unite. Mini­ hydro with battery storage and fuel cell HES to reduce peak demand and
mization capital cost and a stand-alone and connected MG-based PV/ minimize system cost.
battery mode have been studied. In this paper, we benefited from most of the previous studies then
The authors in [18] proposed decentralized control of DCMG based proposing through 1) - a flexible, reliable and sustainable isolated DCMG
on clusters of nanogrids. The design is made up of clusters of nanogrids system via suggesting a backup energy system based hydrogen chain
(households), each of which can work independently while also sharing production, 2) - an optimal energy management system via making a
resources with the rest of the community. For the coordinated power comparison study to achieve the following objectives:
sharing among the contributing nanogrids. An adaptive I-V droop
approach is used to coordinate power-sharing among the contributing - Achieve intelligent coordination control between the various Sour­
nanogrids, which depends on local measurements of the state of charge ces, ESS and loads.
and dc bus voltage. Authors in [19] studied an autonomous DCMG and - Extract the maximum power from the primary sources PV and WT.
energy management-based Cooperative Adaptive Droop where the sys­ - Minimize H2 consumption.
tem consists of PV/battery to give optimal voltage regulation. In [20] a - Enhance the voltage quality.
multi-agent supervisory control for isolated DC-Microgrid with Two - Improve the overall system efficiency.
power management methods equal power sharing and optimal power
dispatch, Both methods have the added benefit of restoring the average The following sections summarize this paper: While the second sec­
system voltage to its reference level. A novel power management tion describes the overall proposed system, the third section presents the
method (PMS) for power-sharing among hybrid energy storage battery/ modeling, sizing, and control of the DCMG parameters, in section 4, a
SC has been proposed and utilized to resolve the demand-generation detailed study of the EMS strategies has been delivered, the fifth section
difference and DC bus voltage regulation, as represented by the au­ shows the simulation, results and detailed discussion. Finally, the au­
thors in [17]. Furthermore, rerouting unwaged battery currents to SCs thors present a conclusion and perspectives.
for quick compensations in the proposed PMS extends battery life. In
[21]. Some topologies integrate the WT to the MG, especially in the DCMG conFiguration
regions with a good wind speed level to compensate for the absence of
solar energy at the beginning and end of the day; an optimization The proposed DCMG, as seen in (Fig. 1), consists of three renewable
strategy based on a Multi-Objective Genetic Algorithm (MOGA) is used energy sources, ESS, Electrolyzer (dump-load), and hybrid domestic
in [22]. The proposed methodology uses a techno-economic approach to load. PV and WT are primary sources working in MPPT mode and
establish the best system design by considering a variety of factors such generating 100 kW and 50 kW, respectively, in nominal conditions. The
as size, cost, and availability. The conclusion is the minimum system FC source works as a secondary source to ensure energy availability and
cost required to meet load requirements, which may also be used to avoid black-out. It generates 50 kW in nominal state and 60 kW as
monetize ancillary services such as voltage management that the smart maximum power. The battery has a significant role in the EMG. It is
DC microgrid-based PV/WT/Battery can give to the utility at the point of responsible for DC-Bus voltage stability, complementation the luck

2
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

energy, and absorbing the excess energy regarding the EMS policy; in
1
our case, it has a 50 kW. The electrolyzer is supposed to be a dc-load and Pm = ρπ R2 V 3 CP (β, λ) (4)
2
dump-load [4,31] to absorb the excess energy when the generated
power is is more than the load power and the battery SOC is more than where,
80 %. DCMG has AC and DC domestic loads. The following equation ρ is the air density (kg/m3), R is the radius of the wind turbine blade
represents the proposed isolated DCMG: (m), V is wind speed (m/s), CP (λ, β) is the power coefficient, which
PLoad = PPV + PWT + PFC + PElectrolyzer + PBattery (1) expresses the relationship between the tip speed ratio λ and the pitch
angle β.
The maximum power rate of charging/discharging the battery is 50 The equations for d-axis and q-axis currents are defined in [4,38] as:
kW. When the battery is depleted, much power comes from sources
(Pnet>0). In this study, the battery absorbs the extra energy until the disd Rs Lq 1
= − isd + ws isq + usd (5)
state of charge (SOC) reaches the higher limit ∈ [20 % – 80 %]. The dt Ld Ld Ld
surplus energy is then transferred into an electrolyzer to make hydrogen ( )
disq Rs Ld 1 1
fuel. The full parameters of the studied DCMG are represented in = − isq − ws isd + Φv + usq (6)
dt Lq Lq Lq Lq
(Table 1).
[ ( )]
Te = 1.5 P wm isq + isd isq Lsd − Lsq (7)
Modeling and control of the proposed DCMG
where:
PV system and MPPT control
isd, isq, usd and udq are the current and voltage of d and q axis,
respectively. ws is the angular frequency of the generator. Ld and Lq are
The PV system consists of PV arrays connected to DC-Bus via boost
the inductance of the generator. Φv is the magnetic flux linkage. Rs is the
dc/dc converter; this boost converter is controlled by MPPT algorithm
resistance of the stator, and P is the number of pole pairs.
type P&O to achieve the MPP by varying the duty cycle [32],
The WT is connected to the DC-bus via a buck-boost DC/DC con­
The output power of the PV system is [33,34]:
verter. Its dynamic mode is given by [4]
Gt (t)
PPV out (t) = PPV [1 + αt ((Tamb + (0.0256 × Gt )) − TC STC )] (2) diL vS (v + vS )
Rated
1000 = − − D (8)
dt L L
where: dvdc iL vdc iL
Ppv_out is the output power (W) of the PV module, Gt denotes the solar = − − D (9)
dt CS CS CS
irradiance (W/m2), Ppv rated denote the rated power (W) of the PV
module at standard test condition (STC), αt is the temperature coeffi­ where:
cient, defined by (− 3.7 × 10–3 (1/C◦ )), TC_STC represent the cell tem­ L and CS are the buck-boost parameters, and D is the duty cycle.
perature (C◦ ) at S.T.C., Tamb denotes the ambient temperature (C◦ ). The reference [39] is proposed a control method that only involves
The PV system is connected to DC-Bus through a boost converter, calculating the power output of the conversion chain of WT. This can be
which has a dynamic model: [35,26] accomplished without needing any speed or wind measurements or
⎧
diPV − (1 − u)vdc vPV another mechanical consideration from the DC link’s current and
⎪
⎪
⎨ = + voltage. The whole power operation is realized for a WT when the ratio
dt L L
(3) of the power variation to speed interpretation is equal to zero. The back-
⎪
⎪ dv (1 − u)i − v
⎩ dc = PV
+
dc
emf of a PMSG is, however, Relative to the rotational speed, and it can
dt C RC
therefore be seen that:
where: dP
IPV is the PV system current, Vdc is the dc bus voltage, C and L are the =0 (10)
dvdc
capacitance and inductance of the boost converter, respectively, R is the
resistance of the load, and u is the control input of the boost converter.
FC modeling and its control
To optimize the output power of the PV system, A MPPT algorithm
has been used. The authors in [36,37] give a detailed review of this
The proton exchange membrane fuel cell PEMFC is an electro­
technique. In our study. A P&O algorithm has been used for simplicity
chemical conversion system transforming the chemical energy of the
and good efficiency in our study.
fuel directly into electrical power [40], where the output voltage of a
single cell is given by [41,28]
WT system and MPPT control
Ucell = EOC − Uact − Eohmic (11)
In this paper, a permanent magnet synchronous generator PMSG ⎧ [ ( )]
based direct-drive WT has been proposed, the mechanical power ⎪
⎪ − 44.43 RT 1
⎪
⎪ E OC = K C E C + (T − 298) + In P H P 2
2 O2
⎪
captured by the rotor blades of the wind turbine [38] ⎪
⎪
⎪
zF zF
⎪
⎪
⎨ ( )
Table 1 U =
1
NA In
ifc (12)
⎪
⎪ act nom
Parameters of the proposed DCMG.
⎪
⎪ τs + 1 i0
⎪
⎪
⎪
⎪
Subsystems parameters ⎪
⎪
⎩
Uohmic = Rin ifc
PV 1Soltech 1STH-250-WH, 100 kW at STC
WT PMSG. Rated wind speed 12 m/s, R = 17 m, Cpopt = 0.48
Rated power 50 kW, Lq = Ld = 0.83 mH, Φv =0.28 wb, P = 8 where:
FC Rated power 50 Kw, I = 200 A, V = 250 V EOC is the t open-circuit voltage, Uact is the activation voltage drop,
Battery Lead-acid battery rate C = 1000 Ah, 240 V Uohmic is ohmic voltage drop, KC rated voltage constant, EC is the elec­
Load [min–max] = [30 kW-200 kW] tromotive forces under standard pressure, T is the operating tempera­
VDC 400 V
ture, z is transferring electron number, F is Faraday constant, R is the gas

3
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

constant, PH2 and PO2 are the gas pressures, ifc is the cell output current, where:
Rin is inner resistance of a stack, τs is the dynamic response time con­ Vin is nominal battery voltage. R is a load resistance, rL is the
stant, N is the number of cells. equivalent series resistor of the BDC inductor and P is the total power;
The reference current of FC is generated from EMS, this current the power generated, and load demand power. iL and Vout are the
should be compared with the actual FC current, and proportional- instantaneous values of inductor current and capacitor voltage or Vdc
integral-derivate PID regulates the error. Controller, the PWM signal respectively. Using a hysteresis control proposed in [31] to generate two
generated in controlled the boost converter, which seen its average complementary PWM signals, these signals control the switches of the
value precedent. bidirectional buck-boost DC/DC converter to regulate the charge/
discharge operations and keep the value of the DC-bus at 400 V.
III-4 battery modeling and its control

In the paper, a Lead Acid battery model can be calculated with Load profile estimation and renewable energy resources
constant resistance by a simple regulated voltage source. The voltage of
the battery’s output is [4,42]: The load estimation is a essential step in modeling standalone
microgrid systems. In our case, this paper is proposed a small commu­
Ub = E − Rb ib (13)
nity residential based on an isolated DC-Microgrid. The selected zone is
( ∫ ) allocated in the south of Algeria in Adrar region (Latitude / Longitude:
Q
E = E0 − K ∫ + Aexp − B Idt (14) 26◦ 58′ 0′′ N / 1◦ 4′ 60′′ E) [44].
Q− Idt
The village consists of homes, a health center, and an electrolysis
system. The summer represents the peak power in this desert region
where:
when the demand power arrives at 145 kWp and 171 kWh/d, in winter
Ub is the battery output voltage, E is open-circuit voltage, E0 is the
the demand power arrives to 65% from that power in almost all cases.
battery internal equivalent voltage, K and Q are the constant battery
The solar radiation, temperature and wind speed data have been
voltage, the polarization voltage and the battery capacity. Considering
∫ taken from the national society of electricity and renewable energy
the battery current Idt defines the actual battery charge. Note that A is
(SKTM), where Fig. 2 represents one day in summer (09 July 2017), the
the exponential zone amplitude, and B is the exponential zone time
natural weather characteristics are represented in the simulation and
constant inverse.
results section. In this section, we should be noted that the control of the
The battery connected to the DC-bus via buck-boost DC/DC con­
load and the electrolyzer isn’t concerned in this study.
verter bidirectional, the model dynamic of it is [35,43]:
⎧
⎪
⎪
diL Vin rL iL uVout Energy management system EMS
⎨ dt = L − L − L
(15)
⎪
⎪ dV ui V
⎩ out = L − out −
P The energy management system will enhance the dynamic response,
dt C RC CVout improve the performance and autonomy of the stand-alone DCMG, and
minimize fuel consumption. This study proposed a comparative analysis

Fig. 1. Block-diagram of the stand-alone DCMG.

4
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

Fig. 2. Average daily load profile for the proposed DC-Microgrid residential community in summer day.

of six 06 energy management techniques. Which are:

Table 2
State Machine Strategy decisions.
Classical PI strategy CPI state description

1 If SOC ≤ 30&ΔP ≥ 0→ P_FC* = P_FCmin

This strategy is based on the controlling of battery state of charge to
2 If SOC ≤ 30&ΔP < 0 & |ΔP| ≤ P Batt→ P_FC* = P_FCnom
obtain the battery power using the PI controller, which its subtraction 3 If SOC ≤ 30&ΔP < 0 & |ΔP| > P Batt&|ΔP| ≤ P Batt +P FCnom→ P_FC* =
from the required power to get the PEM fuel cell reference power, the P_FCnom
controller operation is based on the battery state of charge (SOC), if the 4 If SOC ≤ 30&ΔP < 0 & | ΔP| > P Batt +P FCnom→ P_FC* = P_FCmax
value of battery SOC less than the state of charge reference the battery is 5 If 90 > SOC > 30&ΔP ≥ 0→ P_FC* =
6 If 90 > SOC > 30 &ΔP < 0 & |ΔP| ≤ P Batt→ P_FC* = 0
giving the required power. The FC covers the missing energy, and 7 If 90 > SOC > 30 &ΔP < 0 & |ΔP| > P Batt&|ΔP| ≤ P Batt +P FCnom→
contrariwise the FC feed the DC-Microgrid by the needed power [6,35]; P_FC* =|ΔP| − P Batt
all this is shown in Fig. 3. 8 If 90 > SOC > 30 &ΔP < 0 & |ΔP| > P Batt +P FCnom→ P_FC* = P_FCmax
To have the best response time and stability while using the PI 9 If SOC ≥ 90 & ΔP ≥ 0→ P_FC* = 0
10 If SOC ≥ 90 & ΔP < 0 & |ΔP| ≤ P Batt→ P_FC* = 0
controller, the PI gains or parameters (Kp and Ki) are tuned for this
11 If SOC ≥ 90 & ΔP < 0 & |ΔP| > P Batt&|ΔP| ≤ P Batt +P FCnom→ P_FC*
method using the MATLAB control system toolbox. =|ΔP| − P Batt
12 If SOC ≥ 90 & ΔP < 0 & |ΔP| > P Batt +P FCnom→ P_FC* = P_FCmax
Where: ΔP = (P_PV þ P_WT)-P_Load, P_FCmin ¼ 1e3, P_FCnom ¼ 50e3,
State machine strategy SM P_FCmax ¼ 60e3, P_Batt ¼ 40e3.

The basic idea of state machine strategy executes an output based on

a flow chart or decision tree of stable conditions, depending on present Fuzzy logic control based rules FLC
input values and previous needs. This strategy is based on the same
approach proposed in [45]; the operation mode strategy usually works This paper presented energy management based fuzzy logic rules
on a specific operation or state of the available system based on the which ensure a smooth power profile of the islanded DCMG and mini­
hysteresis between different forms Fig. 3-a. Table 2 shows twelve states mize the fluctuations and power peaks [46].
12 the algorithm operation based to move between all these states to The fuzzy rules and membership functions are extracted based on the
obtain the output reference. The general schema of this technique is author’s expertise and the hybrid renewable system components limi­
presented in Fig. 4-b. tations [47]. Then ΔP subtracts between the load power and (PV & WT)

Fig. 3. Block diagram of EMS based on classical PI.

5
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

Fig. 4. (a) State machine strategy hysteresis. (b) Block diagram of SM.

powers, and the battery SOC considered inputs for the fuzzy logic sys­ represented x = [Pfc , α, Pbatt ] based on the following equality constraints:
tem. At the same time, the FC power reference is considered as an
PLoad = PPV + PWT + PFC + PBatt (17)
output.
Fig. 5 and Table 3 represent the membership functions based on the
(SOC − 0.5(SOCmax + SOCmin ))
trapezoidal and Mamdani type rule. α = 1 − 2μ (18)
SOCmax + SOCmin
Within the boundary conditions.
Artificial neural network ANN
⎧
⎨ Pfc min ⩽Pfc ⩽Pfc max
The proposed ANN contains three layers. The first is the input layer Pbatt min ⩽Pbatt ⩽Pbatt max (19)
⎩
with four neurons corresponding to the number of inputs: the load 0⩽α⩽100
power, the PV & WT powers, and the battery SOC; the second layer is the
hidden layer with ten nodes (Fig. 6-a). The third layer is the output layer where
with one neuron representing the output which is the FC reference ΔT The sampling time,Pload Load power, Pfc Fuel-cell power, Pbatt
power as shown in Fig. 6-b To achieve the minimum satisfying of the Battery power, SOCmax Maximum battery state of charge, SOCmin Mini­
consumption hydrogen, the trial error method is used in this study to mum battery state of charge, α Penalty coefficient, μ Constant (adjusted
choose the right number of the hidden layer [48,49]. between 0 and 1, Pbatt min Minimum battery power, Pbatt max Maximum
The same approach proposed in [48] is used in this study to build the battery power, Pfc max Maximum fuel cell power and Pfc min Minimum
NN optimization block. The weights and biases are applied and brought fuel cell power.
up to date, According to the Levenberg-Marquardt algorithm. The Fig. 7 is summarized the working principle of this strategy:
To train the ANN, FLC data is used to learn and teach the ANN and
used big data to cover the almost probability cases and give the optimal External energy maximization strategy EEMS
FC power reference in each case.
EEMS is a proposed strategy based on the same criteria as H2 con­
Equivalent consumption minimization strategy ECMS. sumption minimization strategies. The cost function consists of the fuel
cell hydrogen consumption and an equivalent battery H2 consumption.
The ECMS is an optimization method based on the cost function This work uses the same approach proposed in [45], where this pro­
implemented by many researchers. This technique minimizes the fuel posed technique aims to minimize hydrogen consumption by maxi­
cell’s fuel and the equivalent power demanded to preserve the battery mizing the battery energy demands within their operating constraints.
SOC [50,51]. In this work, the approach proposed in [45] uses the This method’s key benefit is that the cost function does not require
battery energy penalty coefficient to control the battery state of charge evaluating the corresponding battery energy that usually is empirically
SOC. Therefore, the optimization problem is defined as follows. calculated. Here, the cost function is simple and is supplied over a given
The objective function is imposed as follow: interval by external energy sources [52].
The objective function in this method is defined by:
F = [Pfc + αPbatt ]⋅ΔT (16)

So it’s must minimize this function to obtain an optimal solution

6
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

Fig 5. (a) Membership function. (b) Design of FLC.

problem function.
Table 3
With making the following inequality constraint on consideration:
Fuzzy Logic Rules.
Fuzzy logic rules Pbatt ΔT − (SOC − SOCmin )Vbattr Q⩽0 (22)

If SOC is Low & ΔP is h-negative then Pfc* is High Within the boundary conditions:
If SOC is Low & ΔP is negative then Pfc* is Medium {
If SOC is Low ΔP is medium then Pfc* is Low Pbattmin ⩽Pbatt ⩽Pbattmax
&
(23)
If SOC is Low & ΔP is positive then Pfc* is Low Vdcmin − Vdc ⩽ΔV⩽Pdcmax − Vdc
If SOC is Low & ΔP is h-positive then Pfc* is V-Low
If SOC is Medium ΔP is h-negative then Pfc* is Medium
&
where:
If SOC is Medium & ΔP is negative then Pfc* is Low
If SOC is Medium & ΔP is medium then Pfc* is V-Low
ΔV The voltage drop of the DC-bus, Q The rated battery capacity,
If SOC is Medium & ΔP is positive then Pfc* is V-Low Vbattr The rated battery voltage, Vdcmin The minimum dc bus voltage and
If SOC is Medium & ΔP is h-positive then Pfc* is V-Low Vdcmax The maximum dc bus voltage.
If SOC. High & ΔP is h-negative then Pfc* is Medium Fig. 8 describes this strategy’s general scheme, which there is the
If SOC. High ΔP is negative then Pfc* is Low
battery reference power. The fuel cell reference power is obtained by
&
If SOC. High & ΔP is medium then Pfc* is V-Low
If SOC. High & ΔP is positive then Pfc* is V-Low removing the battery reference power from the load power minus the PV
If SOC. High & ΔP is h-positive then Pfc* is V-Low and WT powers.

Simulation, results and discussion

1
F = − [Pbatt ΔT + Cr ⋅ΔV 2 ] (20)
2
This paper builds a PV/Wind/battery/hydrogen-based DCMG under
The proposed DCMG system hasn’t a supercapacitor, so Cr = 0, the Matlab/Simulink to verify the effectiveness of the proposed control and
objective function became EMS.
F = − Pbatt ΔT (21) As shown in Fig. 9, this system consists of DERs, load side, and power
electronic converters, representing the link between the system devices
To find an optimal solution, x = [Pbatt , ΔV] it must minimize the and hybrid energy backup system that are lead-acid batteries and

7
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

Fig. 6. (a) The ANN architecture. (b) Block diagram of the strategy ANN.

Fig. 7. Block diagram of the ECMS strategy.

hydrogen production/consumption chain. Moreover, each subsystem lifetime (avoid the compressive use), and maximizing the use of PV and
has local control system based on decentralized architecture and EMS WT. To test the six EMS performance on the stand-alone DCMG for 24
based on centralized supervision. hours profile, authors propose that a scale of 100 seconds is equivalent
Once again, we remember the goals noted in the beginning: mini­ to One day while the second 50 represents the half-day. Weather con­
mizing the use of FC (hydrogen consumption), increasing the battery ditions (irradiance, temperature, and wind speed) represent accurate

8
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

Fig. 8. EEMS architecture.

Fig. 9. Matlab/Simulink scheme of the proposed DCMG.

parameters for the selecting location, where the irradiation exceeds to the high wind speed, the WT system is achieved a high output power
1000 w/m2, afternoon the temperature exceeds to 45 ◦ C and sometimes [20–50 kW].
will be more than 50 ◦ C, concerning the wind speed, the south of Algeria Generally, all strategies have a good performance and cover the load
has a high wind speed potential [52,54], especially in this season as demand in any conditions, although the six strategies vary in sharing the
shown in Fig. 10. battery power, FC power, and the recovering power thanks to the elec­
PV, WT, and load power are common factors until testing the per­ trolyzer (dump-load). Fig. 12 is represented these essential factors.
formance of EMS, so until testing the performance of EMS as seen in Fig. 12 is illustrated clearly that the battery is charging at the
Fig. 11. Concerning the load, it has reached the peak power in the half beginning of the day and afternoon when the load power is minimum or
day by 145 KW, and another high demand power at evening, which is the generated power is more than demand power usually, the battery
equivalent to the second 80. the PV gave the maximum power of 87.2 state is charging to ensure the bus stability and balancing of the system.
kW on the half-day when the irradiation was maximum, but the tem­ In peak power cases such as the half-day (50 sec) or the lack of the
perature is very high, which influences the output power efficiency. Due generated power [80–100 sec], the battery device is discharging to keep

9
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

Fig. 10. Weather parameters of the selecting location.

Fig. 11. Power evolution of load side and DERs (PV/WT).

the reliability of the DCMG. When the first backup devices approximate which had an directly impact on the stability of the bus by the voltage
from 50 kW value, the second backup devices share the lack power [75 – change ΔV that is generated during managing the system, consequence
90 sec] to avoid the black-out case, the stress on the battery, and keep of this, the DC-bus curve is the best from the other in the overshoot (1.7
the durability of the system. In the second 63 sec the charging power is V) and the ripple voltage level as seen in Fig. 13.
more than 50 kW, so the buck converter switch of the dump-load is The battery state of charge evolution is shown in Fig. 14; the varia­
becoming ON. tion in SOC is not very different for strategies due to switching mode
FLC and ANN represent a sharing power for the FC more than the during the simulation process, although we can make some results
others about 45 kW as maximum values between 75 and 87 sec, and less because the battery SOC is a reflection of the power evolution of the
power battery consumption that is a disadvantage point, Conversely battery. CPI, SM, and EEMS represent a average charging/discharging
about ECMS and EEMS, they both represent a minimal utilization of the concerning the battery power; ECMS had more consumption and more
FC. Moreover, CPI and SM strategies lack recovering power concerning stress on the battery use. On the other hand, the decreasing of battery
the excess power, at the variance of FLC, ANN, ECMS, and EEMS which SOC in FLC and ANN strategies is less than the others due to more
are recovering a more surplus power, this power will be converting to hydrogen power use.
hydrogen. Fig. 15 illustrates the air and Hydrogen consumption via a fuel cell.
The battery is the first responsible of the DC-bus stability via the EEMS, CPI, and SM are giving an average consumption of air and H2
hysteresis local control, so the EMSs with CPI, SM, FLC, ANN and ECMS with an average value 93.17, 110.7, 116.9, 26.1, 31, and 32.75 (Ipm),
methods hadn’t a direct impact on the DC-bus, except for the EEMS respectively; this is meaning that more consumption of the battery

10
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

Fig. 12. The power balance of battery/FC/Electrolyzer for the six strategies.

power. FLC and ANN consumption of air and H2 represent more con­ comparing those techniques according to the battery state of charge, the
sumption. On the other hand, the ECMS defines the lowest air and hydrogen consumption, the overall efficiency, and the dynamic
hydrogen consumption and hydrogen 59.43 (Ipm) and 16.65 (Ipm), response.
respectively which causes an enhancing and minimization of the overall The obtained simulation results show that each strategy has advan­
cost. tages and disadvantages, where the ECMS has the lowest air and fuel
Table 4 is summarized the precedent results. The total efficiency is consumption with 59.43 and 16.65 Ipm, respectively. . The EEMS has
calculated by the following method: the demand side power (before the the best voltage performance, fastest response time, and minimal over­
converter) divided on the sources power plus the battery power. shoot thanks to the voltage change ΔV that directly impacts the stability
Concerning the recovering power, the optimization methods repre­ of the DC-bus. FLC has a smoothing power-sharing of the battery and
sent power value better than the intelligent methods (FLC and ANN), good overall performance, and ANN can be trained to give a perfect
when the classical methods are represented a less performance in saving arrangement. Although CPI and SM are traditional methods, they have
the excess energy. provided a satisfactory performance in achieving the desired goals.
In the future, to add to these techniques, extracting the maximum
Conclusion power, power management, and optimization issue, authors will opti­
mize the size of renewable energy sources according to the case study as
The comparative study of the most used energy management stra­ multi-objective optimization, adding to techno-economic-
tegies for a DCMG in this paper has been presented to expound the best environmental study to minimize the overall system cost.
technique for managing the DCMG overall system through DC-DC power
converters to facilitate this management, and those different techniques CRediT authorship contribution statement
can ease the dynamic behavior, can improve the DCMG efficiency and
minimize the fuel consumption. This study authorizes the prediction of Oussama Hafsi: Conceptualization, Methodology, Software, Writing
the optimal strategy for managing the overall system of DCMG and – original draft, Visualization. Othmane Abbelkhalek: Supervision.

11
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

Fig. 13. DC-bus voltage of the stand-alone DCMG.

Fig. 14. Battery state of charge of the DCMG.

12
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

Fig. 15. Air and hydrogen consumption from the FC in the DCMG system.

Table 4.C
omparison of different strategies based on difference parameters.
Strategy parameter CPI SM FLC ANN ECMS EEMS

Battery SOC (%) 50–49.91 50–49.92 50–49.94 50–49.94 50–49.90 50–49.92

Average H2 consumption (Ipm) 31 32.75 41.78 41.77 16.65 26.1
Average air consumption (Ipm) 110.7 116.9 149.1 149.1 59.43 93.17
Overall efficiency (%) 95.58 96.73 96.73 96.73 96.67 96.80
Electrolyzer power recove-ring (average value, Watt) 464.55 659.51 735.37 738.07 672.01 672.53

Saad Mekhilef: Supervision. Mohamed Amine Soumeur: Data cura­ [6] Han Y, Chen W, Li Q, Yang H, Zare F, Zheng Y. Two-level energy management
strategy for PV-Fuel cell-battery-based DC microgrid. Int J Hydrogen Energy 2019;
tion, Investigation, Visualization. Mohamed Amine Hartani: Investi­
44(35):19395–404. https://doi.org/10.1016/j.ijhydene.2018.04.013.
gation, Visualization, Methodology. Abdeselem Chakar: Investigation, [7] Sanjeev P, Padhy NP, Agarwal P. Autonomous power control and management
Visualization. between standalone DC microgrids. IEEE Trans Ind Informatics 2018;14(7):
2941–50. https://doi.org/10.1109/TII.2017.2773507.
[8] Xiao J, Wang P, Setyawan L, Xu Q. Multi-level energy management system for real-
time scheduling of DC microgrids with multiple slack terminals. IEEE Trans Energy
Declaration of Competing Interest Convers 2016;31(1):392–400. https://doi.org/10.1109/TEC.2015.2488639.
[9] Hartani MA, Hamouda M, Abdelkhalek O, Mekhilef S. Impacts assessment of
random solar irradiance and temperature on the cooperation of the energy
The authors declare that they have no known competing financial management with power control of an isolated cluster of DC-Microgrids. Sustain
interests or personal relationships that could have appeared to influence Energy Technol Assessm 2021;47:101484. https://doi.org/10.1016/j.
the work reported in this paper. seta.2021.101484.
[10] Sechilariu M, Wang BC, Locment F. Supervision control for optimal energy cost
management in DC microgrid: design and simulation. Int J Electr Power Energy
Acknowledgements Syst 2014;58:140–9. https://doi.org/10.1016/j.ijepes.2014.01.018.
[11] Liu J, Luo W, Yang X, Wu L. Robust model-based fault diagnosis for PEM fuel cell
air-feed system. IEEE Trans Ind Electron 2016;63(5):3261–70. https://doi.org/
Authors acknowledge the General Direction of Scientific Research 10.1109/TIE.2016.2535118.
and Technological Development, DGRSDT through the PRFU. Project [12] Han Y, Li Qi, Wang T, Chen W, Ma L. Multisource coordination energy
management strategy based on SOC consensus for a
(N◦ A01L07UN080120120180007). PEMFC–battery–supercapacitor hybrid tramway. IEEE Trans Veh Technol 2018;67
(1):296–305. https://doi.org/10.1109/TVT.2017.2747135.
[13] Ren H, Wu Q, Gao W, Zhou W. Optimal operation of a grid-connected hybrid PV/
References
fuel cell/battery energy system for residential applications. Energy 2016;113:
702–12. https://doi.org/10.1016/j.energy.2016.07.091.
[1] Arcos-Aviles D, Pascual J, Marroyo L, Sanchis P, Guinjoan F. Fuzzy logic-based [14] Alam M, Kumar K, Verma S, Dutta V. Renewable sources based DC microgrid using
energy management system design for residential grid-connected microgrids. IEEE hydrogen energy storage: modelling and experimental analysis. Sustain Energy
Trans Smart Grid 2018;9(2):530–43. https://doi.org/10.1109/TSG.2016.2555245. Technol Assessments 2020;42:100840. https://doi.org/10.1016/j.
[2] Zubieta LE. Power management and optimization concept for DC microgrids. In: seta.2020.100840.
IEEE First International Conference on DC Microgrids (ICDCM); 2015. p. 81–5. [15] Benhammou Aissa, et al. Exploitation of vehicle’s kinetic energy in power
10.1109/ICDCM.2015.7152014. management of tow -wheel drive electric vehicles based on ANFIS DTC-SVM
[3] Oussama H, Othmane A, Amine HM, Abdeselem C, Amine SM. Proposed energy comparative study. Int J Hydrogen Energy 2021;46(54). https://doi.org/10.1016/
management for a decentralized dc-microgrid based PV-WT-hess for an isolated j.ijhydene.2021.06.023.
community. Int J Power Electron Drive Syst 2020;11(4):2073–82. https://doi.org/ [16] Lindorfer J, Rosenfeld DC, Böhm H. Fuel cells: energy conversion technology.
10.11591/ijpeds.v11.i4.pp2073-2082. Elsevier Ltd; 2020.
[4] Madaci B, Chenni R, Kurt E, Hemsas KE. Design and control of a stand-alone hybrid [17] Singh P, Lather JS. Power management and control of a grid-independent DC
power system. Int J Hydrogen Energy 2016;41(29):12485–96. https://doi.org/ microgrid with hybrid energy storage system. Sustain Energy Technol Assessments
10.1016/j.ijhydene.2016.01.117. 2021;43:100924. https://doi.org/10.1016/j.seta.2020.100924.
[5] Han Y, Ning X, Yang P, Xu L. Review of power sharing, voltage restoration and [18] Nasir M, Jin Z, Khan HA, Zaffar NA, Vasquez JC, Guerrero JM. A decentralized
stabilization techniques in hierarchical controlled DC microgrids. IEEE Access control architecture applied to DC nanogrid clusters for rural electrification in
2019;7:149202–23. https://doi.org/10.1109/ACCESS.2019.2946706.

13
O. Hafsi et al. Sustainable Energy Technologies and Assessments 52 (2022) 102121

developing regions. IEEE Trans Power Electron 2019;34(2):1773–85. https://doi. [37] Podder AK, Roy NK, Pota HR. MPPT methods for solar PV systems: a critical review
org/10.1109/TPEL.2018.2828538. based on tracking nature. IET Renew Power Gener 2019;13(10):1615–32. https://
[19] Sahoo S, Mishra S, Jha S, Singh B. A Cooperative adaptive droop based energy doi.org/10.1049/iet-rpg.2018.5946.
management and optimal voltage regulation scheme for DC microgrids. IEEE Trans [38] Hu J, Shan Y, Xu Y, Guerrero JM. A coordinated control of hybrid ac/dc microgrids
Ind Electron 2020;67(4):2894–904. https://doi.org/10.1109/TIE.2019.2910037. with PV-wind-battery under variable generation and load conditions. Int J Electr
[20] Malysz P, Sirouspour S, Emadi A. An optimal energy storage control strategy for Power Energy Syst 2019;104:583–92. https://doi.org/10.1016/j.
grid-connected microgrids. IEEE Trans Smart Grid 2014;5(4):1785–96. https://doi. ijepes.2018.07.037.
org/10.1109/TSG.2014.2302396. [39] Zammit D, Spiteri Staines C, Micallef A, Apap M. MPPT with current control for a
[21] Eghtedarpour N. A synergetic control architecture for the integration of PMSG small wind turbine in a grid-connected DC microgrid. Green Energy Technol
photovoltaic generation and battery energy storage in DC microgrids. Sustain 2018;Part F10(January 2018):205–19. https://doi.org/10.1007/978-3-319-74944-
Energy, Grids Networks 2019;20:100250. https://doi.org/10.1016/j. 0_14.
segan.2019.100250. [40] Lajnef T, Abid S, Ammous A. Modeling, control, and simulation of a solar
[22] Shadmand MB, Balog RS. Multi-objective optimization and design of photovoltaic- hydrogen/fuel cell hybrid energy system for grid-connected applications. Adv
wind hybrid system for community smart DC microgrid. IEEE Trans Smart Grid Power Electron 2013;2013:1–9. https://doi.org/10.1155/2013/352765.
2014;5(5):2635–43. https://doi.org/10.1109/TSG.2014.2315043. [41] Dahale S, Das A, Pindoriya NM, Rajendran S. An overview of DC-DC converter
[23] Yin C, Wu H, Locment F, Sechilariu M. Energy management of DC microgrid based topologies and controls in DC microgrid. In: 2017 7th Int Conf Power Syst ICPS;
on photovoltaic combined with diesel generator and supercapacitor. Energy 2018. p. 410–5. 10.1109/ICPES.2017.8387329.
Convers Manag 2017;132:14–27. https://doi.org/10.1016/j. [42] Banguero E, Correcher A, Pérez-Navarro Á, Morant F, Aristizabal A. A review on
enconman.2016.11.018. battery charging and discharging control strategies: application to renewable
[24] Puchalapalli S, Tiwari SK, Singh B, Goel PK. A microgrid based on wind-driven energy systems. Energies 2018;11(4). https://doi.org/10.3390/en11041021.
DFIG, DG, and solar PV array for optimal fuel consumption. IEEE Trans Ind Appl [43] Agarwal A, Deekshitha K, Singh S, Fulwani D. Sliding mode control of a
2020;56(5):4689–99. https://doi.org/10.1109/TIA.2020.2999563. bidirectional DC/DC converter with constant power load, 2015 IEEE 1st Int. Conf.
[25] Valverde L, Bordons C, Rosa F. Integration of fuel cell technologies in renewable- Direct Curr. Microgrids, ICDCM 2015, no. c, pp. 287–292, 2015, doi: 10.1109/
energy-based microgrids optimizing operational costs and durability. IEEE Trans ICDCM.2015.7152056.
Ind Electron 2016;63(1):167–77. https://doi.org/10.1109/TIE.2015.2465355. [44] Aoulef Map. http://www.mapnall.com/it/Mappa-Aoulef_1135884.html (accessed
[26] Han Y, Zhang G, Li Q, You Z, Chen W, Liu H. Hierarchical energy management for Jan. 19, 2022).
PV/hydrogen/battery island DC microgrid. Int J Hydrogen Energy 2019;44(11): [45] Njoya Motapon S, Dessaint LA, Al-Haddad K. A comparative study of energy
5507–16. https://doi.org/10.1016/j.ijhydene.2018.08.135. management schemes for a fuel-cell hybrid emergency power system of more-
[27] Sedaghati R, Shakarami MR. A novel control strategy and power management of electric aircraft. IEEE Trans Ind Electron 2014;61(3):1320–34. https://doi.org/
hybrid PV/FC/SC/battery renewable power system-based grid-connected 10.1109/TIE.2013.2257152.
microgrid. Sustain Cities Soc 2019;44:830–43. https://doi.org/10.1016/j. [46] Zia MF, Elbouchikhi E, Benbouzid M. Microgrids energy management systems: a
scs.2018.11.014. critical review on methods, solutions, and prospects. Appl Energy 2018;222:
[28] Pu Y, Li Q, Chen W, Liu H. Hierarchical energy management control for islanding 1033–55. https://doi.org/10.1016/j.apenergy.2018.04.103.
DC microgrid with electric-hydrogen hybrid storage system. Int J Hydrogen Energy [47] Allahvirdizadeh Y, Mohamadian M, HaghiFam MR, Hamidi A. Optimization of a
2019;44(11):5153–61. https://doi.org/10.1016/j.ijhydene.2018.10.043. fuzzy based energy management strategy for a PV/WT/FC hybrid renewable
[29] Jahangir MH, Javanshir F, Kargarzadeh A. Economic analysis and optimal design system. Int J Renew Energy Res 2017;7(4):1686–99.
of hydrogen/diesel backup system to improve energy hubs providing the demands [48] Majed C, Karaki SH, Jabr R. Neural network technique for hybrid electric vehicle
of sport complexes. Int J Hydrogen Energy 2021;46(27):14109–29. https://doi. optimization. In: 2016 18th Mediterranean Electrotechnical Conference (MELECON);
org/10.1016/j.ijhydene.2021.01.187. 2016. p. 1–6. https://doi.org/10.1109/MELCON.2016.7495327.
[30] Maghami MR, Hassani R, Gomes C, Hizam H, Othman ML, Behmanesh M. Hybrid [49] Karaki SH, Dinnawi R, Jabr R, Chedid R, Panik F. Fuel cell hybrid electric vehicle
energy management with respect to a hydrogen energy system and demand sizing using ordinal optimization. In: SAE Int J Passeng Cars - Electron Electr Syst;
response. Int J Hydrogen Energy 2020;45(3):1499–509. https://doi.org/10.1016/ 2015. p. 60–9. 10.4271/2015-01-0155.
j.ijhydene.2019.10.223. [50] Rodatz P, Paganelli G, Sciarretta A, Guzzella L. Optimal power management of an
[31] Malla SG, Bhende CN. Voltage control of stand-alone wind and solar energy experimental fuel cell/supercapacitor-powered hybrid vehicle. Control Eng Pract
system. Int J Electr Power Energy Syst 2014;56:361–73. https://doi.org/10.1016/j. 2005;13(1):41–53. https://doi.org/10.1016/j.conengprac.2003.12.016.
ijepes.2013.11.030. [51] García P, Torreglosa JP, Fernández LM, Jurado F. Viability study of a FC-battery-SC
[32] Zhao Z, Cheng R, Yan B, Zhang J, Zhang Z, Zhang M, et al. A dynamic particles tramway controlled by equivalent consumption minimization strategy. Int J
MPPT method for photovoltaic systems under partial shading conditions. Energy Hydrogen Energy 2012;37(11):9368–82. https://doi.org/10.1016/j.
Convers Manag 2020;220:113070. https://doi.org/10.1016/j. ijhydene.2012.02.184.
enconman.2020.113070. [52] Motapon SN, Dessaint LA, Al-Haddad K. A robust H2-consumption-minimization-
[33] Bukar AL, Tan CW, Lau KY. Optimal sizing of an autonomous photovoltaic/wind/ based energy management strategy for a fuel cell hybrid emergency power system
battery/diesel generator microgrid using grasshopper optimization algorithm. Sol of more electric aircraft. IEEE Trans Ind Electron 2014;61(11):6148–56. https://
Energy 2019;188:685–96. https://doi.org/10.1016/j.solener.2019.06.050. doi.org/10.1109/TIE.2014.2308148.
[34] Ayop R, Isa NM, Tan CW. Components sizing of photovoltaic stand-alone system [53] Chabani A, Makhloufi S, Lachtar S. Overview and impact of the renewable energy
based on loss of power supply probability. Renew Sustain Energy Rev 2018;81: plants connected to the electrical network in southwest Algeria. EAI Endorsed
2731–43. https://doi.org/10.1016/j.rser.2017.06.079. Trans Energy Web 2018;March:169168. https://doi.org/10.4108/eai.29-3-
[35] Soumeur MA, Gasbaoui B, Abdelkhalek O, Ghouili J, Toumi T, Chakar A. 2021.169168.
Comparative study of energy management strategies for hybrid proton exchange [54] Zahraoui Y, Basir Khan MR, Alhamrouni I, Mekhilef S, Ahmed M. Current status,
membrane fuel cell four wheel drive electric vehicle. J Power Sources 2020;462: scenario, and prospective of renewable energy in Algeria: a review. Energies 2021;
228167. https://doi.org/10.1016/j.jpowsour.2020.228167. 14(9). https://doi.org/10.3390/en14092354.
[36] Verma D, Nema S, Shandilya AM, Dash SK. Maximum power point tracking (MPPT)
techniques: recapitulation in solar photovoltaic systems. Renew Sustain Energy
Rev 2016;54:1018–34. https://doi.org/10.1016/j.rser.2015.10.068.

14