Professional Documents
Culture Documents
Coordinated Power Management and Control of Standalone PV-Hybrid System With Modified IWO-Based MPPT
Coordinated Power Management and Control of Standalone PV-Hybrid System With Modified IWO-Based MPPT
Abstract—To augment the photovoltaic (PV) power generation Idc DC-link current of the microgrid (A).
conversion, a maximum power point tracking (MPPT) technique Ielz Output current of the electrolyzer (A).
plays a very significant role. This article introduces a hybrid Ipv Output current of the PV panel (A).
MPPT algorithm integrating modified invasive weed optimization Isof Measured or actual current of the fuel cell (A).
(MIWO) and perturb and observe (P&O) technique under the ∗
Isof Reference current of the fuel cell (A).
rapid weather change and partial shading scenarios for the efficient
extraction of the maximum power from the standalone PV-based
itermax Maximum no. of generations/iterations in MIWO.
hybrid system. MIWO handles the initial stages of MPPT followed Smax Maximum no. of the possible production of weeds.
by the application of the P&O algorithm at the final stages in view Smin Minimum no. of the possible production of weeds.
of acquiring the rapid global peak and maximal PV power. The σiter Standard deviation at the current generation in
studied microgrid comprises of the PV system, battery, electrolyzer, MIWO.
fuel cell, and load. A coordinated dc-voltage regulation and power σinitial Initial standard deviation in the MIWO algorithm.
management strategy between each subsystem of the hybrid mi- σfinal Final standard deviation in the MIWO algorithm.
crogrid is implemented to save the battery from the undesirable Xbest Best candidate or weed.
charging/discharging operation. Additionally, with the monitoring Xij Position of the parent weed.
of dc voltage, the dc/dc converter associated with the battery
and dc link plays as an MPPT circuit of the PV without the Xij+1 Updated position of the parent weed.
requirement of an extra dedicated circuit. Takagi–Sugeno (TS) Gmax Global maximum power point.
fuzzy controller is adopted for suppressing/mitigating the voltage k Iteration of the P&O-based technique.
oscillations of the microgrid during the variations in the solar ΔV Step change in voltage of the P&O-based technique
irradiance/temperature and power demand. The results clearly (V).
exhibit the superior performance of the proposed methodology Pb Battery power (kW).
compared with some of the existing techniques. Pel Electrolyzer power (kW).
Index Terms—Hybrid power generation, maximum power point Pf c Fuel cell power (kW).
(MPP), modified invasive weed optimization (MIWO), perturb and PL Load power (kW).
observe (P&O), photovoltaic (PV) system, voltage control. Ppv PV power (kW).
Q1 PWM signal for charging the battery.
Q2 PWM signal for discharging the battery.
NOMENCLATURE Se PWM signal of the electrolyzer converter.
Sf PWM signal of the fuel cell converter.
G Solar irradiance (W/m2 ).
vabc Instantaneous 3-Ф voltages at PCC (V).
H2 Hydrogen in the fuel cell.
Vb Output voltage of the battery system (V).
iabc Instantaneous 3-ф currents at PCC (A).
Vdc DC-link voltage (V).
Ibat Measured or actual current of the battery (A). ∗
∗ Vdc Reference dc-link voltage (V).
Ibat Reference current of the battery (A).
Ve Operating voltage of the electrolyzer (V).
Ve∗ Nominal operating voltage of the electrolyzer (V).
Manuscript received February 26, 2020; revised June 30, 2020 and August Vf Operating voltage of the fuel cell (V).
20, 2020; accepted August 23, 2020. Date of publication September 14, 2020;
date of current version August 26, 2021. (Corresponding author: Chittaranjan Vmpp Voltage at maximum power point (V).
Pradhan.) Vpv Output voltage of the PV panel (V).
Chittaranjan Pradhan and Terje Gjengedal are with the Department of Electri- Vref _P CC Nominal root-mean-square (RMS) voltage at PCC.
cal Engineering, UiT Norges Arctic University in Narvik, 8514 Narvik, Norway
(e-mail: chittaranjan.pradhan@uit.no; terje.gjengedal@uit.no).
Manoj Kumar Senapati and Paresh Kumar Nayak are with the Depart-
ment of Electrical Engineering, Indian Institute of Technology (Indian School
I. INTRODUCTION
of Mines) Dhanbad, Dhanbad 826004, India (e-mail: miitkgp@gmail.com; N RECENT years, there has been an increasing trend of
paresh@iitism.ac.in).
Siva Ganesh Malla is with the Confer Professional Group of Conferences Pvt.
Ltd., Vishakapatnam 530007, India (e-mail: mallasivaganesh@gmail.com).
I electricity generation worldwide using solar photovoltaic
(PV) systems due to numerous technical and environmental
Digital Object Identifier 10.1109/JSYST.2020.3020275 benefits [1], [2]. However, climatic dependence and the high
1937-9234 © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on November 27,2023 at 09:00:59 UTC from IEEE Xplore. Restrictions apply.
3586 IEEE SYSTEMS JOURNAL, VOL. 15, NO. 3, SEPTEMBER 2021
installation cost are the major challenges of presently existing is not tested with dump load and unbalanced load conditions. A
PV power systems. The nonlinear nature of the power–voltage local hierarchical control strategy for the battery energy storage
(P–V) characteristics of a PV system requires a maximum power of a standalone PV-battery system is recommended in [20]. In
point tracking (MPPT) technique for maximizing the power [21] and [22], a dc-link voltage control approach for a standalone
conversion efficiency. The maximum power point (MPP) in PV/wind-based system is projected. In [23], an adaptive power
a PV power-generation system is a unique point in the P–V management control strategy is highlighted for the PV/wind-
curve at a given solar irradiance and ambient temperature and it based dc-microgrid integration of the energy storage devices
varies with environmental experiences, such as solar irradiance, (i.e., FC, battery, and electrolyzer). A supervisory controller is
temperature, and partial shading. Since these parameters vary reported for the operation/power management among wind/PV
continuously, tracking the MPP effectively in the PV system is a renewable generation, energy storage, and load power in a
major challenge. In this perspective, different MPPT techniques hybrid distributed power system [24]. In [20]–[24], the partial
are reported in improving the power conversion MPPT effi- shading condition is not taken into consideration to analyze the
ciency of the PV system [3]–[10]. The hill-climbing method [3], performance of the controller. Furthermore, the authors have
incremental conductance [4], and perturb and observe (P&O) considered separate dc/dc converters for the solar PV module
[5] are the most popular conventional MPPT techniques. These for MPPT and the charge controller (i.e., charging/discharging
techniques exhibit good tracking capability under the uniform operation) in the battery system [20]–[24].
solar irradiance. However, because of the slow convergence, the The aforementioned study clearly shows that there are dif-
computational burden increases, and the requirement of more ferent hybrid MPPT algorithms/controllers available in the lit-
sensors makes these techniques less attractive. Furthermore, erature for the standalone PV systems consisting of multiple
these MPPT techniques are not suitable under varying environ- energy storage devices. However, still, there is a strong need to
mental scenarios [e.g., partial shading conditions (PSCs)] due design/implement a simple and cost-effective controller/MPPT
to the presence of multiple peaks in the power versus voltage algorithm for the standalone PV systems that operate with a
(P–V) characteristic curves of the PV [3], [4]. wide variety of power system contingencies and environmental
In the meantime, with the aim of overcoming the limitations effects. Further, an efficient power management control strategy
of conventional MPPT techniques, various artificial intelligence is also very crucial for ensuring a smoothing power balance
MPPT techniques have been proposed by the research works, among PV power, dynamic power demands, and the energy
especially in the highly intermittent environmental conditions storage devices (i.e., battery, FC, and electrolyzer) of standalone
[6]–[15]. These include fuzzy logic control (FLC) [6], artificial microgrid systems.
neural network, particle swarm optimization (PSO), firefly algo- To overcome the limitations of the available methodolo-
rithm, ant colony optimization, flower pollination algorithm, bat gies, the contributions of this article are summarized as
algorithm, Jaya algorithm, and grey wolf optimization (GWO), follows.
[7]–[12], etc. However, the abovementioned singly used soft 1) A maiden attempt has been made to introduce an efficient
computing techniques have enhanced multipeak global MPPT hybrid modified invasive weed optimization (MIWO)-
capability as compared with the conventional techniques [7], P&O MPPT algorithm for tracking the maximum power of
[12]–[14]. Hence, the researchers have suggested hybrid opti- the PV system. The proposed MPPT algorithm is validated
mization techniques by integrating two or more algorithms to by comparing it with the existing hybrid MPPT strategies
further upgrade the MPP search process with less computa- based on the PSO-P&O [13] and GWO-P&O [14].
tional time. The simulated annealing with PSO [7], GWO-FLC 2) The implementation of a single dc/dc converter (associated
[12], PSO-P&O [13], GWO-P&O [14], Jaya algorithm with between the dc bus and battery energy system (BES)] is
differential evolution (Jaya-DE) [15], and adaptive neurofuzzy employed for regulating the dc-link voltage and MPPT cir-
inference system-PSO [16] are such few examples. The overall cuit of the proposed standalone PV system. Furthermore,
performances (i.e., achieve the optimal solution, faster con- supervisory dc-link voltage control and power manage-
vergence, and efficiency) of the hybrid soft computing (i.e., a ment strategy for the PV-based standalone hybrid system
combination of two or more algorithms) based MPPT algorithms is designed.
are superior over the single soft computing algorithms. However, 3) The employment of the Takagi-Sugeno (TS)-fuzzy logic-
the convergence speed, the design complexity of the controller based robust controller to cater to the voltage stability
parameters, sensor requirements, and the implementation of of the standalone PV power system and its efficacy is
hardware/microcontroller costs of hybrid MPPT algorithms are equated with the conventional proportional plus integral
still not attractive. From this perspective, developing/designing (PI) controller.
a new and hybrid MPPT algorithm for augmenting the search 4) The validation of the proposed controller through OPAL-
performance in real-time control problems is always welcome. RT-based real-time hardware-in-the-loop (HIL) simula-
Meanwhile, to overcome the intermittent PV power problems tion platform under rapid meteorological changes of solar
and enhance the overall system performance (e.g., efficiency, irradiance/temperature, PSCs, and the change in load.
stability, and reliability), standalone PV systems are incorpo- The results clearly illustrate that the MIWO-P&O is demon-
rated with various energy storage devices [e.g., fuel cell (FC), strated to yield a better search performance to the global maxi-
battery, and electrolyzer, etc.] [17]–[20]. Furthermore, different mum power point (GMPP) than PSO-P&O [13] and GWO-P&O
control strategies are designed to keep the dc-link voltage in the [14] techniques. As well, it is investigated that the MIWO-P&O
permissible safety limits and ensure an optimum power balance improves the dynamics of the voltage/current and power profiles
between PV and energy storage devices [17]–[24]. A dynamic of the standalone PV system under a wide variety of operating
power management technique is presented for a standalone conditions as compared with the techniques reported in [13] and
hybrid microgrid comprising of solar PV, electrolyzer, FC, and [14]. The rest of this article is organized as follows. First, the
super-capacitor [16]. However, the performance of the microgrid configuration and modeling of the standalone dc microgrid are
Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on November 27,2023 at 09:00:59 UTC from IEEE Xplore. Restrictions apply.
PRADHAN et al.: COORDINATED POWER MANAGEMENT AND CONTROL OF STANDALONE PV-HYBRID SYSTEM 3587
electrolyzer cell [23], [24]. The electrochemical model of the convergence performance [30]. The Cauchy density function has
lead-acid battery is realized using a voltage-controlled source mainly two parameters, such as location and scale parameters.
with constant resistance [26], [27]. Due to the page constraints, The standard deviation is nothing but the scale parameters. The
the detailed modeling and the data/parameters of the PV, battery, MIWO has the ability to deal with the high-dimensional test
FC, and electrolyzer are not presented in this article. The design functions that help in solving complex search problems with
inductors L and capacitors C value of the microgrid (i.e., see a faster convergence toward the optimal solution. The newly
Fig. 1) converters/inverter are presented in the Appendix. generated child weeds are normally distributed over the search
space with the mean of the parent weed position and the varying
III. PROPOSED MIWO-P&O-BASED MPPT TECHNIQUE standard deviation, which is presented as follows [30]:
In the traditional P&O-based MPPT technique, two succes- (itermax − iter)n
sive samples of the power levels of the P–V curve are compared σiter = (σinitial − σfinal ) + σfinal (2)
iternmax
by perturbing the nominal voltage Vmpp in an erratic direction.
The direction of the perturbation is governed by the sign of the where iter and itermax represent the current (i.e., present) gener-
power variation for MPPT in the P–V curve. The corresponding ation/iteration and the maximum number of generations, respec-
voltage to the maximum power point (MPP) is expressed as tively. σiter , σinitial , and σfinal stand for the standard deviation at
follows [28], [29]: the current generation, the initial standard deviation, and the final
standard deviations, respectively. n is the nonlinear modulation
dPpv
Vmpp (k) = Vmpp (k − 1) + ΔV × sign (1) index. In this article, the values of itermax , σinitial , σfinal , and n
dVpv are selected as 100, 10e-6, 0.7, and 3, respectively.
where ΔV and k are the step voltage change and the number In the MIWO algorithm, maximum weeds Wmax are gen-
of iterations of the P&O-based technique, respectively. It is erated around the parent weeds. In this article, the value of
revealed that the P&O-based technique exhibits a better tracking Wmax is chosen as 15. The best weed can be used to move the
competency with faster convergence under the uniform solar remaining weeds to the best position. The position of each weed
insolation (i.e., solar irradiance) in the presence of a single peak, gets updated in such a way that all weeds slowly and steadily
as shown in Fig. 3(a) [4], [28], [29]. On the other hand, there gravitate toward the global maximum point in the search space.
are multiple peaks in which the highest point is denoted as the The updated value of the weed position (i.e., PV voltage) can
GMPP and all other points are local peaks during PSCs, as shown be obtained by using the previous position, and the difference
in Fig. 3(b). The conventional MPPT techniques, which assume between the previous position and the best position is as follows:
a single peak power point on the PV characteristic curve, cannot
guarantee convergence to GMPP; rather these methods mostly Xij+1 = Xij + mσiter × Cauchy(0, 1) × Xbest − Xij ;
get trapped into one of the local peaks due to their inability
to differentiate between the local and global peaks (GPs). In i = 1, 2 . . . . . . , Wmax
such situations, the conventional P&O algorithm fails to follow max
dPpv
the Vmpp corresponding to the GMPP (Gmax ) [13]–[15]. To m = ΔV × sign (3)
augment the GMPP searching ability (i.e., attain the optimal dVmpp
value and faster convergence with less oscillation) of the PV
system under PSCs, the MIWO algorithm is integrated with the where Xij is the ith weed position at the jth iteration. Xij+1 is
P&O technique, in this article. The location and magnitude of the update/new weed position at the jth iteration, Xbest is the
local and global MPPs are influenced stochastically by varying best weed found in the whole population. The value of ΔV is
shading pattern and the configurations of the PV array. Fig. 3(b) considered as 0.06 V. It is examined that MIWO improves the
clearly shows the presence of multiple power peaks with one search performance because the CDF and standard deviation
GMPP during PSCs and the magnitude/position of the MPP maintain better population-diversity characteristics in the search
varies with the change in the shading pattern of the PV. space [30].
The invasive weed optimization (IWO) is a simple and ef- While performing MPPT by employing the MIWO technique,
fective numerical stochastic optimization algorithm inspired by it takes more time for all the weeds to reach on a GMPP, and
colonizing weeds [30]. It is demonstrated that the IWO has a hence, in order to minimize the search process, the majority of
high capability in searching the global maxima/minima as com- the weeds (i.e., particles) when reaching mutually closer to each
pared with the existing recent evolutionary-based algorithms other, the P&O algorithm is instigated with its starting point as
and has a better adoption nature of the changing environmental the position of the best weed in the MIWO algorithm. The P&O
conditions [30]. First, initialize the randomly generated parent method is a promising algorithm, which converges to GMPP if
weeds/population in a search space. Second, each weed grows the starting point of the search is nearer to the GMPP. Hence, it
and produces its child weed. The generation of the number of is recommended to examine the P–V curve initially with MIWO
child weeds of each plant depends on its fitness value or ranking. and then execute the P&O method starting with the position
It is distributed from the maximum possible Smax production to of the best weed of MIWO. Thus, the MIWO-supported P&O
its minimum weeds Smin according to the Cauchy distribution (MIWO-P&O) technique can provide enhanced performance in
function (CDF) of random variables. In this article, the value the terms of GMPP Gmax with less oscillation under varying
of Smax and Smin is chosen as 5 and 1, respectively. Normally, environmental conditions. The flowchart of the MIWO-P&O-
the Gaussian distribution function (GDF) is employed in the based hybrid MPPT algorithm is exhibited in Fig. 4. In Fig. 4,
IWO algorithm to update the weeds position in the search space. “ε” is a tolerance factor, whose value is 0.005. The efficacy of
However, the CDF function can contribute to superior results in the MIWO-P&O algorithm over the PSO-P&O [13] and GWO-
place of GDF for producing a better optimal solution with a faster P&O [14] is presented in Section V.
Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on November 27,2023 at 09:00:59 UTC from IEEE Xplore. Restrictions apply.
PRADHAN et al.: COORDINATED POWER MANAGEMENT AND CONTROL OF STANDALONE PV-HYBRID SYSTEM 3589
Fig. 8. Fuzzy membership functions for (a) voltage/current error signal xi and
(b) the derivative of the voltage/current error signal xi .
Fig. 6. Flowchart of the coordinative power management algorithm. B. TS-Fuzzy Logic Control
Since the linguistic rule consequent of TS fuzzy has become
variable by means of its parameters with an infinite number
of gain variation characteristics, it can deal with the complex
control problems effectively. Consequently, it supervises the
system uncertainties and upgrades the stability of the power
system effectively during the meteorological changes and load
variations [34], [35]. On the other hand, it is examined that the
TS-fuzzy controller can contribute a better control solution than
the PI controller under nonlinear control problems and system
uncertainties for varying operating conditions [34], [35]. Thus,
with the purpose of enhancing the system dynamic performance,
the TS-fuzzy-based controller is employed in this article.
The deviations in voltage/current error xi and its derivative
Fig. 7. TS-fuzzy-based inverter controller. xi signal are taken as the input variables to the fuzzy control,
for designing the TS-fuzzy controller (as shown in Figs. 5 and
7). The input voltage/current error and its derivative signals are
the surplus power will be consumed by the electrolyzer through fuzzified by means of two linguistic memberships (MFs) values,
the switch Se . In order to control the dc-link voltage, the Q1 P and N for positive and negative, respectively, as described in
signal is integrated with the electrolyzer controller. Hence, the Fig. 8. The MFs of the two input linguist variables, P and N for
buck converter controller can operate the output voltage of the xi , and xi signals are expressed as (4) and (5), respectively.
⎧
electrolyzer Ve as well as the dc-link voltage at their nominal ⎪
values during the production of hydrogen. During that period, ⎨ 0, xi < −L1
the BES maintains the dc-link voltage through Q2 . However, in μP (xi ) = xi +L1
, −L 1 ≤ xi ≤ L1 and μN (xi )
⎪
⎩ 1,
2L1
the case of light load and maximum PV power, whenever the x i > L1
surplus power is more than the electrolyzer power rating, the ⎧
deloading operation of the PV system can be implemented for ⎪
⎨1, xi < −L1
power balancing [23]. = −x2L i +L1
, −L 1 ≤ x i ≤ L1 (4)
Furthermore, whenever the SoC of BES is less than 0.2 (i.e., ⎪
⎩0,
1
x i > L1
the battery is fully discharged as SoC is below the lower thresh- ⎧
old limit of 0.2 and the current flowing through it decreases), ⎪ xi < −L2
⎨0,
then FC will be switched ON (Sf ). It indicates that the load
−L2 ≤ xi ≤ L2 and μN (xi )
xi +L2
μP (xi ) = 2L2 ,
power is more than the power generation and BES is unable to ⎪
⎩1,
supply power. At this time, BES is switched OFF for increasing xi > L2
the lifespan of the battery. So, the reference current of the BES ⎧
is set zero (i.e., SoC is less than 0.2), as denoted in Fig. 5. The ⎪
⎨1, xi < −L2
−xi +L2
FC regulates the dc-link voltage at its nominal value so that = , −L2 ≤ xi ≤ L2 (5)
the battery current Ibat is maintained zero during this period. ⎪
⎩0, 2
2L
xi > L2
Hence, no power is supplied by the BES as the FC meets the
load demand. Similarly, according to the system events (e.g., The TS-fuzzy controller is represented by the following sim-
variations in the load power and PV power, etc.) when the plified four fuzzy rules.
SoC of the battery lies between 0.2 and 0.8, the BES will
1) Rule 1: If xi (κ) is N and xi (κ) is N, then Z1 = A1 xi (κ)
charge/discharge through the bidirectional dc/dc converter for a
Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on November 27,2023 at 09:00:59 UTC from IEEE Xplore. Restrictions apply.
PRADHAN et al.: COORDINATED POWER MANAGEMENT AND CONTROL OF STANDALONE PV-HYBRID SYSTEM 3591
value of the fuzzy constants is listed in the Appendix. Moreover, capacity is 18.9 kW (almost 30% more than the peak load of
the corresponding proportional Kp and integral Ki parameter the PNA). Hence, four PV strings are connected in parallel to
of the PI controller is reported in the Appendix. Based on the generate the net power of 18.92 kW, as shown in Fig. 2(b),
integral-square-error performance criteria, the coefficients of the where the installed capacity of 22-series-connected PV cells in
PI controllers are designed [35]. a single row is 4.732 kW. In order to decide the capacity of the
The output of the TS-fuzzy controller Y is obtained by using battery bank, the depth of discharge of the battery is considered
the generalized defuzzifier, which is evaluated as follows: as 60% [23]. This is decided that even when the PV power is
zero, it should cater to the energy requirement of 15 kW load
(F1 × Z1 + F2 × Z2 + F3 × Z3 + F4 × Z4 )
Y = (6) for approximately an hour. The installed capacity of the batter
(Z1 + Z2 + Z3 + Z4 ) energy system is evaluated as follows:
where F1 = min. {μP (xi ), μP (xi )}, F2 = min.{μP (xi ), 15 kW × 1 h
μN (xi )}, F3 = min. {μN (xi ), μP (xi )}, and F4 = min. Installed battery capacity = = 83.33 Ah (7)
300 V × 0.6
{μN (xi ), μN (xi )}.
The value of “Y” is adapted dynamically using the fuzzy-based where 300 V is the net output voltage Vb of the battery system.
controller, which results in improving the stability of the power Hence, 25 numbers of batteries are connected in series with
system during system events/contingencies. each having a 12 V rating to achieve the required 83.33 Ah and
300 V battery system.
Additionally, when there is no PV power, the FC supplies the
C. Unbalanced PCC Voltage Compensation Topology necessary power for power balancing at the peak load scenario.
In practice, the power systems unbalance in nature due to the Based on the peak load (PL,max = 14.6 kW), the FC capacity is
presence of the unbalance switching operation in each phase decided [23]. For the best utilization, the rating of FC is taken as
and faults (i.e., the current in each phase is not the same due 18 kW, which is assumed as 20% more than the peak load. As
to the unbalanced load) of a power system. The presence of far as the dump load (i.e., electrolyzer) is concerned, its rating
unbalanced loads has an unsustainable impact on the power depends on the maximum availability of the surplus power of
system voltage stability performance. Owing to the unbalanced the microgrid [23]. Since the electrolyzer is very costly, 60%
voltage drops across the LC filter results in making the PCC of the maximum available surplus power from the PV power
voltage unbalanced. The voltage unbalance factor (i.e., the ratio generation is considered in determining the electrolyzer capacity
of the negative sequence to the positive sequence of the funda- and expressed as follows:
mental voltage component) may not be within the permissible Installed capacity of the electrolyzer = 0.6 × (Maximum PV
limit (i.e., less than 1%) in each phase of the PCC during the power generation – Minimum load) = 0.6 × (18.9–5.85) kW =
unbalanced switching operation. So as to maintain a balanced 7.83 kW.
phase voltage and mitigate the load variations issues of the 3-Ф
power system, an appropriate inverter controller technique is V. RESULTS AND DISCUSSIONS
required to control individual phase voltages. To achieve this
goal, three individual TS-fuzzy controllers are used for each To investigate the performance of the proposed control
phase. Hence, the PWM-based modulation indexes (MIs) signals methodology, a PV-based standalone hybrid power system is
can generate/regulate for each phase independently, as shown in considered, as shown in Fig. 1. In this article, the detailed plant
Fig. 7. As a result, three different MI signals can generate under model (see Fig. 1) and the controller (see Figs. 4–8) are realized
the unbalanced load conditions to maintain the PCC voltage bal- on a real-time simulator (RTS), which can closely replicate
anced. However, in the case of balanced load conditions, equal the dynamics of physical systems and controller. The RTS is
MIs will be generated for each phase through the PWM-based a combination of advanced computer hardware and comprehen-
inverter controller. So, the presented controller maintains a stable sive software. It has a parallel-processing hardware architecture
and balanced 3-Ф supply at the load bus under both balanced and assembled in modular units. Each unit contains both processing
unbalanced load scenarios. The error signal and the derivatives and communication modules. With the aim of implementing the
of the error signal between the actual RMS phase voltage and the HIL setup, two OPAL-RT (manufactured by OPAL-RT Tech-
nominal PCC voltage (i.e., 230 V) are taken as the input signals nologies) units are considered with corresponding computers:
to the TS-fuzzy controller, as presented in Fig. 7. one used for the modeling of PV-based hybrid power system
(see Fig. 1) and the other one for the controller, as exhibited
in Fig. 9 [35], [36]. The analog signals are going from the
D. Unit Sizing of the Standalone Hybrid Power System plant (i.e., OPAL RT-1) to the controller (i.e., OPAL RT-2) and
The optimization of unit sizing is important in a renewable- the digital signals are coming from the controller to the plant.
based hybrid power generation system as it helps to minimize The communication between each OPAL-RT is done in actual
the operating and generation cost as well as maintain the power analog and digital signals. So, a HIL path is established between
system stability and reliability. In this article, based on the load both OPAL-RTs for supervising the real-time dynamics of the
profile of the Pacific Northwest area (PNA), the sizing/installed power system dynamics effectively [36]. The effectiveness of
capacity of the PV and energy storage systems, such as the BES, the suggested control approach is verified by considering the
FC, and electrolyzer of the microgrid, is designed [23]. For the following case studies.
analysis, the hourly load profile of the PNA is considered, where Case A. Performance Evaluation of MPPT Under Variation
the peak power demand PL,max and minimum demand PL,min of Solar Insolation (i.e., Solar Irradiance): In this case study,
are 14.6 and 5.85 kW, respectively [23]. the solar insolation is changed from 1000 to 900 W/m2 at
The installed PV module capacity is considered as 20%–30% the time t = 1.2 s, as signified in Fig. 10(a). A fixed load of
more of the peak load [19]. So, the estimated installation PV 14.6 kW is considered, in this case. For the above-mentioned
Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on November 27,2023 at 09:00:59 UTC from IEEE Xplore. Restrictions apply.
3592 IEEE SYSTEMS JOURNAL, VOL. 15, NO. 3, SEPTEMBER 2021
TABLE II
COMPARATIVE DYNAMIC DC VOLTAGE Vmpp PERFORMANCE
TABLE III
COMPARATIVE VOLTAGE DEVIATION AND EFFICIENCY ηM P P T
Fig. 9. HIL laboratory setup on the OPAL-RT platform. updates the reference dc-link voltage to operate the PV system
at MPP according to the climatological changes of the solar irra-
diance/temperature. The comparative reference dc-link voltage
Vmpp performance indexes in the terms of maximum voltage
deviation Vp , minimum voltage deviation Vd , and settling time
(i.e., faster convergence) for the above-mentioned MPPT tech-
niques are presented in Table II. From the results, it can be
illustrated that the proposed MIWO-P&O technique produces
a better dynamic MPPT voltage Vmpp response with a faster
converge as compared with the PSO-assisted P&O [13] and Grey
Wolf-assisted P&O [14].
Similarly, the comparative dynamic response of dc-link volt-
age Vdc of the microgrid is shown in Fig. 10(c), from which
it can be determined that the performance of the dc voltage
controller is satisfactory in both transient as well as steady-state
circumstances. The exact values of the percentage of actual
dc-link voltage deviations from its nominal value are mentioned
in Table III. In Fig. 10(c), the dc-link voltage is followed by the
Vmpp , which is generated by the presented MPPT algorithms.
The Vmpp signal is employed as the reference voltage signal to
the dc/dc converter (associated between the dc bus and BES)
control, as discussed in Sections II and III. The dc/dc converter
∗
helps to regulate the dc-link voltage corresponds to Vdc = Vmpp
during the system events. As a result, the PV system operates
at its MPP because of control of the dc-link voltage at Vmpp .
The comparative actual extracted output power Ppv,out of PV
for different MPPT techniques is presented in Fig. 10(d). The
efficacy of an MPPT algorithm is benchmarked using the MPPT
efficiency ηM P P T equation as follows [5]:
Ppv,out
%ηMPPT = × 100 (8)
PMPPT
Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on November 27,2023 at 09:00:59 UTC from IEEE Xplore. Restrictions apply.
PRADHAN et al.: COORDINATED POWER MANAGEMENT AND CONTROL OF STANDALONE PV-HYBRID SYSTEM 3593
Fig. 11. PV output power under the partial shading for one array.
TABLE IV
COMPARATIVE VOLTAGE DEVIATION AND EFFICIENCY ηM P P T
Fig. 14. (a) Power responses of different subsystems. (b) SoC of the battery
system.
Fig. 16. Instantaneous system parameters. (a) Unbalanced phase load currents.
(b) Balanced phase voltages. (c) Individual MI of each phase. (d) RMS phase
voltages at the PCC.
TABLE V
VALUE OF THE TS-FUZZY CONSTANTS AND PI-CONTROLLER GAINS
TABLE VI
DESIGN VALUE OF INDUCTORS L AND CAPACITORS C
Fig. 17. Single-line to ground (LG) fault. (a) DC-link voltage response.
(b) RMS voltage of Phase A.
Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on November 27,2023 at 09:00:59 UTC from IEEE Xplore. Restrictions apply.
3596 IEEE SYSTEMS JOURNAL, VOL. 15, NO. 3, SEPTEMBER 2021
[11] A. K. Podder, N. K. Roy, and H. Roy Pota, “MPPT methods for solar [35] C. Pradhan, C. N. Bhende, and A. K. Samanta, “Adaptive virtual inertia-
PV systems: A critical review based on tracking nature,” IET Renewable based frequency regulation in wind integrated power system,” Renewable
Power Gener., vol. 13, no. 10, pp. 1615–1632, Jul. 2019. Energy, vol. 115, pp. 558–574, Jan. 2018.
[12] A. M. Eltamaly and H. M. H. Farh, “Dynamic global maximum power [36] J. Bélanger, P. Venne, and J.-N. Paquin, “The what, where, and why of
point tracking of the PV systems under variant partial shading using hybrid real-time simulation,” in Proc. IEEE PES Gen. Meeting, Jul. 2010, vol. 1,
GWO-FLC,” Solar Energy, vol. 177, pp. 306–316, Jan. 2019. pp. 37–49.
[13] K. Sundareswaran, V. Kumar, and S. Palani, “Application of a combined
particle swarm optimization and perturb and observe method for MPPT in
PV systems under partial shading conditions,” Renewable Energy, vol. 75,
pp. 308–317, Mar. 2015.
[14] S. Mohanty, B. Subudhi, and P. K. Ray, “A grey wolf-assisted perturb & Chittaranjan Pradhan (Member, IEEE) received
observe MPPT algorithm for a photovoltaic power system,” IEEE Trans. the Ph.D. degree in electrical engineering from
Sustain. Energy, vol. 32, no. 1, pp. 340–347, Mar. 2017. the Indian Institute of Technology Bhubaneswar,
[15] N. Kumar, I. Hussain, B. Singh, and B. K. Panigrahi, “Rapid MPPT for Bhubaneswar, India, in 2018.
uniformly and partial shaded PV system by using JayaDE algorithm in He is currently a Postdoctoral Fellow with the
highly fluctuating atmospheric conditions,” IEEE Trans. Ind. Informat., Arctic Centre for Sustainable Energy, Faculty of Sci-
vol. 13, no. 5, pp. 2406–2416, Oct. 2017. ence and Technology, UiT The Arctic University of
[16] N. Priyadarshi, S. Padmanaban, J. B. Holm-Nielsen, F. Blaabjerg, and M. Norway, Tromsø, Norway. His research interest is
S. Bhaskar, “An experimental estimation of hybrid ANFIS–PSO-based focused on the areas of power electronics and con-
MPPT for PV grid integration under fluctuating sun irradiance,” IEEE trol, smart grids, electric vehicles, and clean energy
Syst. J., vol. 14, no. 1, pp. 1218–1229, Mar. 2020. technologies.
[17] F. Blaabjerg, Y. Yang, D. Yang, and X. Wang, “Distributed power- Dr. Pradhan was a recipient of the Certificate of Outstanding Contributing in
generation systems and protection,” Proc. IEEE, vol. 105, no. 7, Reviewing Award by the International Journal of Electrical Power and Energy
pp. 1311–1331, Jul. 2017. Systems, Elsevier, in May 2018, for his valuable and thorough feedback on
[18] J. Kumar, A. Agarwal, and V. Agarwal, “A review on overall control of dc manuscripts, and for his quick turnaround on reviews. He is a Reviewer Member
microgrids,” J. Energy Storage, vol. 21, pp. 113–138, Feb. 2019. of various international journals and conferences, including IEEE, IET, and
[19] A. Gupta, S. Doolla, and K. Chatterjee, “Hybrid ac–dc microgrid: Sys- Elsevier.
tematic evaluation of control strategies,” IEEE Trans. Smart Grid, vol. 9,
no. 4, pp. 3830–3843, Jul. 2018. Manoj Kumar Senapati (Member, IEEE) received
[20] R. K. Sharma and S. Mishra, “Dynamic power management and control the Ph.D. degree in electrical engineering from the In-
of a PV PEM fuel-cell-based standalone AC/DC microgrid using hybrid dian Institute of Technology (Indian School of Mines)
energy storage,” IEEE Trans. Ind. Appl., vol. 54, no. 1, pp. 526–538, Dhanbad, Dhanbad, India, in 2020.
Jan./Feb. 2018. He is currently an Assistant Professor with the
[21] Y. Yang, Y. Qin, S.-C. Tan, and S. Y. R. Hui, “Efficient improvement of Department of Electrical Engineering, Government
photovoltaic-battery systems in standalone DC microgrids using a local College of Engineering Keonjhar, Keonjhar, India.
hierarchical control for the battery system,” IEEE Trans. Power Electron., His current research interests include power elec-
vol. 34, no. 11, pp. 10796–10807, Nov. 2019. tronics and control, and soft computing techniques
[22] M. Trifkovic, M. Sheikhzadeh, K. Nigim, and P. Daoutidis, “Modeling to microgrid applications.
and control of a renewable hybrid energy system with hydrogen storage,”
IEEE Trans. Control Syst. Technol., vol. 22, no. 1, pp. 169–179, Jan. 2014.
Siva Ganesh Malla (Member, IEEE) was born in
[23] S. G. Malla and C. N. Bhende, “Voltage control of stand-alone wind
Anakapalli, Visakhapatnam, Andhra Pradesh, India,
and solar energy system,” Int. J. Elect. Power Energy Syst., vol. 56,
in 1986. He received the Ph.D. degree in electrical
pp. 361–373, Mar. 2014.
engineering from the Indian Institute of Technology
[24] M. K. Senapati, C. Pradhan, S. R. Samantaray, and P. K. Nayak, “Im-
Bhubaneswar, Bhubaneswar, India, in 2014.
proved power management control strategy for renewable energy-based
He is currently the Director of Confer Profes-
DC micro-grid with energy storage integration,” IET Gener. Transmiss.
sional Group of Conferences Pvt., Ltd, Visakhapat-
Distrib., vol. 13, no. 6, pp. 838–849, Mar. 2019.
nam, Andhra Pradesh, India. His research interests
[25] A. Safari and S. Mekhilef, “Simulation and hardware implementation of
are renewable energy sources, microgrid, power qual-
incremental conductance MPPT with direct control method using cuk con-
ity, electrical vehicles, power electronics converters,
verter,” IEEE Trans. Ind. Electron., vol. 58, no. 4, pp. 1154–61, Apr. 2011.
drives, optimization techniques, and FACTS.
[26] R. A. Jackey, “A simple, effective lead-acid battery modeling process for
electrical system component selection,” J. Passenger Cars-Electron. Elect.
Syst., pp. 219–227, 2008. Paresh Kumar Nayak (Member, IEEE) received the
[27] M. Ceraolo, “New dynamical models of lead-acid batteries,” IEEE Trans. Ph.D. degree in electrical engineering from the Indian
Power Syst., vol. 15, no. 4, pp. 1184–1190, Nov. 2000. Institute of Technology Kharagpur, Kharagpur, India,
[28] M. A. Danandeh and S. M. Mousavi G., “Comparative and comprehensive in 2014.
review of maximum power point tracking methods for PV cells,” Renew- He is currently an Assistant Professor with the De-
able Sustain. Energy Rev., vol. 82, no. 3, pp. 2743–2767, Feb. 2018. partment of Electrical Engineering, Indian Institute
[29] L. F. L. Villa, T.-P. Ho, J.-C. Crebier, and B. Raison, “A power electronics of Technology (Indian School of Mines) Dhanbad,
equalizer application for partially shaded photovoltaic modules,” IEEE Dhanbad, India. His major research interests include
Trans. Ind. Electron., vol. 60, no. 3, pp. 1179–1190, Mar. 2013. power system relaying and protection of microgrid.
[30] M. Misaghi and M. Yaghoobi, “Improved invasive weed optimization algo-
rithm (IWO) based on chaos theory for optimal design of PID controller,”
J. Comput. Des. Eng., vol. 6, no. 3, pp. 284–295, Jul. 2019. Terje Gjengedal (Senior Member, IEEE) was born
[31] T. Basso, “IEEE 1547 and 2030 Standards for distributed energy resources in Sandane, Norway, in 1958. He received the M.Sc.
interconnection and interoperability with the electricity grid,” Nat. Renew- and Ph.D. degrees in electrical engineering from the
able Energy Lab., Golden, CO, USA, Tech. Rep. NREL/TP-5D00-63157, Norwegian University of Science and Technology,
2014 Trondheim, Norway, in 1983 and 1987, respectively.
[32] C. Masetti, “Revision of European standard EN 50160 on power quality: Since 2009, he has been the R&D Director with
Reasons and solutions,” in Proc. 14th Int. Conf. Harmon. Qual. Power, Statnett. Besides, he was a Professor with the Norwe-
Bergamo, Italy, Sep. 2010, pp. 1–7. gian University of Science and Technology and the
[33] A. Anzalchi and A. Sarwat, “Overview of technical specifications for UiT The Arctic University of Norway, Narvik, Nor-
grid-connected photovoltaic systems,” Energy Convers. Manage., vol. 152, way. While being the Vice President of Statkraft, he
pp. 312–327, Nov. 2017. was working on the technology, system, and market
[34] C. Pradhan and C. N. Bhende, “Online load frequency control in wind integration of wind power and other renewable sources. His current research
integrated power systems using modified Jaya optimization,” Eng. Appl. interests include renewable energy, power system planning, and modeling.
Artif. Intell., vol. 77, pp. 212–228, Jan. 2019.
Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on November 27,2023 at 09:00:59 UTC from IEEE Xplore. Restrictions apply.