Received November 5, 2021, accepted November 22, 2021, date of publication November 29, 2021,

date of current version December 9, 2021.

Digital Object Identifier 10.1109/ACCESS.2021.3131161

A Bio-Inspired Heuristic Algorithm for Solving

Optimal Power Flow Problem in
Hybrid Power System
MANZOOR AHMAD1 , NADEEM JAVAID 1,2 , (Senior Member, IEEE),
IFTIKHAR AZIM NIAZ1 , (Senior Member, IEEE),
AHMAD ALMOGREN 3 , (Senior Member, IEEE),
AND AYMAN RADWAN 4 , (Senior Member, IEEE)
1 Department of Computer Science, COMSATS University Islamabad, Islamabad 44000, Pakistan
2 School of Computer Science, University of Technology Sydney, Ultimo, NSW 2007, Australia
3 Department of Computer Science, College of Computer and Information Sciences, King Saud University, Riyadh 11633, Saudi Arabia
4 Instituto de Telecomunicacoes, Universidade de Aveiro, 3810-193 Aveiro, Portugal

Corresponding authors: Nadeem Javaid ([email protected]) and Ahmad Almogren ([email protected])

This work was supported by King Saud University, Riyadh, Saudi Arabia, through Researchers Supporting under Project RSP-2021/184.
The work of Ayman Radwan was supported in part by the FCT / MEC through Programa Operacional Regional do Centro, and in part by
the European Union through the European Social Fund (ESF) under Investigador FCT under Grant 5G-AHEAD
IF/FCT-IF/01393/2015/CP1310/CT0002.

ABSTRACT In recent studies, emphasis has been placed on optimal power flow (OPF) problems in
traditional thermal, wind, and solar energy sources-based hybrid power systems. Various metaheuristic
algorithms have been proposed to find optimal solutions to the OPF problems in the hybrid power system.
The OPF, due to the quadratic nature of its primary objective function, is a nonlinear, nonconvex, and
quadratic optimization problem. In this study, we have proposed a bio-inspired bird swarm algorithm (BSA)
to find an optimal solution to the OPF problem in the hybrid power system because it performs well in
the case of optimizing the well-known Rastrigin quadratic benchmark function. In this study, uncertainty of
utility load demand and stochastic electricity output from renewable energy resources (RESs) including wind
and solar are incorporated into the hybrid power system for achieving accuracy in operations and planning
of the system. We have used a modified IEEE-30 bus test system to verify and measure the performance of
BSA and a comparison is made with well-known evolutionary metaheuristic algorithms. The proposed BSA
consistently achieves more accurate and stable results than other metaheuristic algorithms. Simulation-based
optimization results have shown the superiority of BSA approach to solve the OPF problems by satisfying all
constraints and minimum power generation cost 863.121 $/h is achieved in case study 1. Simulation-based
experiment results have indicated that by imposing the carbon tax (ton/h) the power generation from RESs
was increased. In case study 2, the proposed BSA approach has also outperformed and minimum electricity
cost 890.728 $/h is achieved as compared to other algorithms.

INDEX TERMS Deterministic optimal power flow, uncertainty of utility load demand, bio-inspired bird
swarm algorithm, stochastic solar and wind power.

I. INTRODUCTION power through electricity generation units or power plants.

A. BACKGROUND AND MOTIVATION The transmission system carries the electricity through trans-
Generally, an electric power grid or power system consists of mission lines from electricity generation plants to utilities
power generation plants, electricity transmission and distribu- or load centers. The electricity distribution system feeds
tion systems. The power generation plants generate electrical the electricity through distribution lines to nearby homes,
agricultural units, industries, and commercial buildings. The
The associate editor coordinating the review of this manuscript and traditional electric power grids are responsible for produc-
approving it for publication was Ahmed F. Zobaa . ing electricity and carrying it to residential, industrial, and

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
VOLUME 9, 2021 159809
 M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

commercial consumers through electricity transmission and by private parties in a power system. The ISO purchases
distribution lines. Two authorities, 1) independent system scheduled renewable electricity from private parties in order
operator (ISO) and 2) electric utility, are responsible to con- to cater the growing consumers’ load demand. Wind power
trol operations and planning of the power system in a country. generation depends upon stochastic wind speed at different
The ISO is an independent authority established by the gov- times of day. Similarly solar PV power generation depends
ernment to ensure reliability of electricity generation and the upon uncertain solar irradiance during the day time. Due to
transmission system in the electrical power grid. An electric the fluctuant and intermittent solar and wind power output,
utility is an authority that engages in feeding the electricity the available power from solar PV units and windfarms may
through distribution lines to consumers by balancing the be more or less than wind-scheduled power at different times
demand and supply of the electrical load. of day. In an overestimation scenario, the ISO is required to
In the electrical power grid operations and planning, the have a spinning reserve based on utility load demand, when
control always resides on the generation side and the power power supplied by solar PV units and windfarms operators
generation plants adjust their electricity generation accord- is less than wind-scheduled power. The ISO has to increase
ing to the changes in electricity demand from consumers. the reserve cost associated with reserve electricity generation
Sometimes power generation plants produce surplus electric- units to balance the supply and demand in this scenario.
ity, which is transmitted to the nearby area by transmission An underestimation scenario may arise when actual renew-
lines or stored [1]. Therefore, it is of practical importance able energy received from RESs is greater than scheduled
to balance load demand and electricity supply in the power power. In that case, the surplus power output from RESs is
system. For this purpose, many techniques have been applied wasted and ISO bears a penalty cost if it is not stored or
in the research literature. On the generation side, to address transmitted to a nearby area [8], [9]. Incorporating stochastic
optimal power flow (OPF) problems in the power system power generation from wind and solar into the system raises
is considered as a technique for finding stable and secure the complexity of power system operations and planning. The
operating points of electricity generation plants and their utility load demand is also uncertain in nature due to variation
optimal scheduling on an hourly basis [2]–[5]. in consumers’ load demand that directly affects spinning
In 1962, Carpentier first introduced economic dispatch reserve cost in the power system. Moreover, considering the
problem extension as the OPF problem in traditional thermal uncertainty of utility load demand has significant importance
energy sources-based power systems [6]. The OPF is one to achieve accuracy in the operations and planning of the sys-
of the well-known and well-studied research areas in the tem. Therefore, an effective technique is required to reduce
power system. It can be defined as: ‘‘To find out the stable the overall electricity generation cost.
and secure operating points (levels) for electricity generation
plants in order to meet load demand of utility in power system, B. LITERATURE REVIEW
generally with attention to minimize electricity generation In the research literature, various studies have been
cost’’ [6]. In traditional thermal energy sources-based power documented using two types of optimization algorithms
systems, the OPF is a nonconvex, nonlinear, and quadratic for solving the OPF problems in the power system. These
problem due to the quadratic nature of its primary objective optimization algorithm types are traditional mathemati-
function to reduce electricity generation cost. The primary cal algorithms or methods and metaheuristic algorithms.
objective function of OPF problem has been modeled as Numerous mathematical optimization methods including lin-
quadratic curve and its various forms such as valve-point ear programming [10], linear/quadratic programming [11],
loading effect quadratic curve, piecewise quadratic curve, sequential linear programming [12], newton method [13],
and prohibited operating zones quadratic curve for the tra- generalized benders decomposition (GBD) [14], nonlinear
ditional thermal energy source [6], [7]. Researchers have programming [15]–[17], mixed integer nonlinear program-
also proposed various techniques for solving the OPF prob- ming (MINLP), [18], interior point method [19], [20], and
lems considering other objectives, in addition to the primary simplified gradient method [21] have been applied to solve
electricity generation cost minimization objective. These the OPF problems. In these traditional methods, nonlinear
objectives include minimizing voltage deviation, power loss objective function and constraints are converted into linear
in transmission lines, and emission pollution and enhancing form before solving the OPF problem because the mathe-
voltage stability index [6]–[9]. matical method cannot handle the nonlinear properties of the
In the last decade, integration of environment friendly problem [22]. This convergence in constraints and objective
and clean electricity output from renewable energy functions may affect the accuracy of operations and planning
sources (RESs) including wind and solar into thermal power of the power system.
systems have become necessary due to the rising demand for The OPF problem in thermal energy sources-based power
electricity and global warming issues. Therefore, the power systems widely has been studied by researchers using meta-
systems are striving towards a sustainable system future due heuristic algorithms. In the last decade, numerous studies
to rapidly growing integration of RESs in power systems. have been documented based on metaheuristic algorithms
On the electricity generation side, the RESs such as solar such as binary backtracking search algorithm (BBSA) [6],
photovoltaic (PV) units and windfarms are being owned adaptive group search optimization (AGSO) [23], improved

159810 VOLUME 9, 2021

M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

colliding bodies optimization (ICBO) [25], differential search TABLE 1. Abbreviations.

algorithm (DSA) [24], moth swarm algorithm (MSA) [26],
stud krill herd (SKH) [27], [28], differential evolution
(DE) [29], hybrid of genetic algorithm (GA) and PSO [30],
GA based on multi-parent crossover (GA-MPC) [31],
improved social spider optimization (ISSO) [32], modified
grasshopper optimization (MGO) [33], improved moth flame
optimization (IMFO) [34], multi-objective EA based decom-
position (MOEA/D) [35], modified pigeon-inspired opti-
mization (MPIO) [36], and adaptive moth flame optimization
(AMFO) [37] to find optimal solutions to the OPF problems
in traditional power systems. The abbreviation of different
terms and methods are specified in Table 1.
The OPF problem primary objective – electricity genera-
tion cost minimization, is considered in all of the aforemen-
tioned studies [6], [23]–[37]. Moreover, other objectives such
as reducing power loss in transmission lines, emission pollu-
tion, etc. are considered in some studies. In these studies, the
performance of proposed metaheuristic algorithms has been
measured on one or more IEEE-30, IEEE-57 and IEEE-118,
bus test systems. The power systems are rapidly growing
with RESs integration due to increasing electricity demand
and global warming issue due to traditional thermal energy
sources.
In the literature, some studies have been documented
[38]–[42] to find optimal solutions to the OPF problems in
traditional thermal and wind energy sources-based hybrid
power systems with a focus on minimizing the overall
cost of electricity generation. In study [38], gbest guided
ABC (GABC) has been utilized to find optimal solutions
to the OPF problems in the thermal and wind energy
sources-based hybrid power systems. In which objectives
were to minimize electricity generation cost and emission
pollution. In study [39], modified bacteria foraging algo-
rithm (MBFA) is employed for solving the OPF problem in
the traditional thermal and wind energy sources-based hybrid
power system. A doubly-fed induction generator model is
utilized to justify the OPF problem inequality constraints
and problem is formulated with various objective functions
in above study. In study [40], authors applied ant colony
optimization (ACO) and MBFA for solving the OPF prob-
lems in the traditional thermal and wind energy sources- In the last decade, some studies also have been doc-
based hybrid power systems. In study [41], authors utilized umented [43]–[53] to find optimal solution to the OPF
multi-objective glowworm swarm optimization (GWSO) to problems in the thermal, wind, and solar energy sources-
solve the OPF problems in the thermal and wind energy based hybrid power systems. In recent studies [43]–[53],
sources-based hybrid power systems. In all of the aforemen- different metaheuristic approaches including grey wolf opti-
tioned studies, [38]–[41], the authors have utilized a modified mizer (GWO) [43], fuzzy membership function based PSO
IEEE-30 bus test system to verify and measure performance (FMF-PSO) [44], improved adaptive DE (IADE) [45], mod-
of the applied approaches. In study [42], the authors have ified imperialist competitive algorithm based on sequen-
adopted self-adaptive evolutionary programming (EP) for tial quadratic programming (MICA-SQP) [46], modified
solving the OPF problems in the traditional thermal and wind JAYA [47], hybrid of phasor PSO and GSA [48], barna-
energy sources-based hybrid power systems. In all of the cles mating optimization (BMO) [49], PSO [50], Hybrid of
aforementioned studies [38]–[42], the authors have applied DE and PSO [51], MBFA [52], and sunflower optimization
well-known Weibull probability density function (PDF) for (SFO) [53] have been proposed for solving the OPF prob-
modeling uncertainty of stochastic wind speed to incorporate lems in hybrid power systems. In studies [43]–[53], mostly
wind power output into hybrid power systems. Lognormal PDF and Weibull PDF have been applied for

VOLUME 9, 2021 159811

 M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

TABLE 2. Summary of studies to optimal solutions to the OPF problems in hybrid power systems.

modeling uncertainty of the stochastic solar irradiance and optimal solutions to the nonlinear and quadratic OPF prob-
wind speed, respectively. lems because DE has a premature convergence property [54]
Table 2 represents the summary of all of the aforemen- and PSO is incapable of searching neighborhood existing
tioned studies [38]–[53] have been conducted on thermal and solutions in nonlinear quadratic optimization problems [55].
wind or thermal, wind, and solar energy sources-based hybrid
power systems. As specified in Table 2, OPF problem primary
objective – quadratic fuel cost ξq (electricity generation cost) D. CONTRIBUTION AND PAPER ORGANIZATION
minimization is considered in all studies. Moreover, other In study [56], authors have proposed a new bio-inspired bird
objectives such as reducing power loss in transmission lines swarm algorithm (BSA). In which, it has been observed that
Ploss is considered in 9 studies and emission pollution Ep is the BSA has good diversity and can flexibly regulate its four
considered in 10 studies. It is also observed from Table 2, different search strategies such as foraging, vigilance, pro-
in most studies the uncertainty of power output from RESs ducer and scrounger to explore the search space. Moreover,
was incorporated for solving the OPF problems in hybrid BSA can improve its convergence speed without affecting
power systems. the stability and accuracy of optimal solutions by making
better balancing among exploration and exploitation of search
space. In fact, under suitable interpretations, DE and PSO
C. PROBLEM STATEMENT mutation operators are distinct forms of the proposed BSA
In study [8], authors have proposed success history-based approach. In which, the bird’s social behaviour such as the
adaptation differential evolution (SHADE) to find optimal scrounger formula is similar to the DE mutation operator and
solution to the OPF problems in hybrid power system. The the foraging formula is similar to the PSO. Moreover, the
OPF problem objectives - to reduce electricity generation BSA has prominent distinguishing features, in addition to the
cost and emission pollution are considered. In study [9], the merits of the DE and PSO.
author has proposed a fuzzy logic technique based on PSO In study [56], optimization results have proved the supe-
finding optimal solution to the multi-objective OPF problems riority of the BSA as compared to DE and PSO to optimize
in hybrid power systems, by considering objectives to reduce the Rastrigin function F9 , which is a well-known quadratic
active power output cost and power loss in transmission lines. benchmark function for performance evaluation of optimiza-
In both studies [8], [9], uncertainty of stochastic solar irra- tion algorithms. The primary objective function of the OPF
diance and wind speed are incorporated into the power system problem is to reduce power generation cost, which follows a
to solve the OPF problems. However, the utility load demand quadratic nature function. Inspired by the study [56], we have
uncertainty has been ignored in these studies [8], [9]. On the proposed BSA to find an optimal solution to the OPF prob-
other hand, SHADE and PSO may be inefficient to find lem in the hybrid power system. This research study is an

159812 VOLUME 9, 2021

M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

extension of [57] our conference paper already published.

It have following knowledge contribution in academic
research;
• A method based on BSA is proposed for finding an
optimal solution to the OPF problem in the hybrid power
system by incorporating uncertainty of the utility load
demand and stochastic power output from RESs.
We have used a modified IEEE-30 bus test system to
measure and evaluate the performance BSA for finding an
optimal solution tothe OPF problem in the hybrid power
system. Simulation based optimization results have shown
the superiority of BSA approach to solve the OPF problems
(see Table 10 and Figure 8).
The organization of paper is as follows: The uncertainty
modeling of utility load demand, stochastic solar irradiance,
and wind speed are described in section II. The section III
consists of OPF problem formulation, objective function,
and different constraints. In section IV, we have explained
the proposed metaheuristic method. The simulation-based
experiment results for the proposed BSA approach and other
algorithms are specified in V and concluding remarks are
written in section VI.

II. UNCERTAINTY MODELLING OF WIND SPEED, SOLAR

IRRADIANCE, AND UTILITY LOAD DEMAND
The most significant aspect of uncertainty modeling is to use
an appropriate PDF for predicting the values of uncertain or
random variables. We have used the Lognormal PDF and
FIGURE 1. Weibull fitting of wind speed.
Weibull PDF for modeling uncertainty of solar irradiance
and stochastic wind speed by adopting same strategy pro-
posed in study [8]. We have utilized the same formula and
approach of Gaussian (normal) PDF presented in studies where, scale factor c and shape factor k are parameters and
[59], [60] for modeling uncertainty of utility load demand. its mean (Mwbl ) can be calculated as:
The Hong’s point estimate method (PEM) proposed in Mwbl = c × 0(1 + k −1 ), (2)
study [61] has been used for calculating the utility load
demand on load buses. where, 0() represents the gamma function and it can be
In this section, first, we have described the model for formulated as:
Z ∞
handling the uncertainty of wind speed to incorporate wind
0(x) = e−1 t x−1 dt. (3)
power output. Secondly, we have explained the model for 0
incorporating stochastic solar power in the power system.
For the performance evaluation of BSA, we have modified
Lastly, we have described Gaussian PDF for handling uncer-
standard IEEE-30 bus test system, in which two traditional
tainty of utility load demand.
thermal energy sources at 5 and 11 generator buses are
replaced with wind energy sources (windfarms). The total
A. UNCERTAINTY MODELING OF STOCHASTIC WIND number of wind turbines (WTs) in each windfarm and values
SPEED of c and k of Weibull PDF parameters are given in Table3.
In the research literature, the well-known Weibull PDF Frequency distributions and Weibull fitting of wind speed
has been mostly applied for modeling the wind speed of windfarms energy sources attached at buses 5 and 11 are
v (m/s) [38]–[40], [42] to incorporate the stochastic nature determined after execution of 8000 Monte Carlo simulation
wind electricity generation into the power system. The uncer- scenarios and plotted in Figure 1.
tainty modeling stochastic wind speed using Weibull PDF can
be defined as: TABLE 3. Windfarms and Weibull PDF parameters [8].

 (k−1)  
v k k
fv (v) = × × e−(v/c) for 0 < v < ∞,
c c
(1)

VOLUME 9, 2021 159813

 M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

In a windfarm, the actual power output of WT depends

upon wind speed v (m/s) it meets. In both windfarms,
each WT has a 3 MW rated power output. The cumulative
rated power generations of windfarms connected at generator
buses 5 and 11 are 75 MW from 25 WTs and 60 MW
from 20 WTs, respectively. The WT electricity generation can
be formulated as follows [8]:
P , for vr < v ≤ vout ,

 wr


v − vin

Pw (v) = Pwr × , for vin ≤ v ≤ vr , (4)
 vr − vin
0, for v < vin and v > vout ,



where, vin is cut-in, vout is cut-out and vr is cut-in rated wind

speed met to WT and its rated power output is represented as
Pwr . We have considered vr = 16 m/s, vout = 25 m/s, and
vin = 3 m/s for wind power output calculation purpose.
The histograms in Figure 2 indicate wind power output
based on wind speed Weibull distribution plotted in Figure 1,
from windfarms connected at bus 5 and bus 11. It is observed
from WT power output Eq. 4 that WT provides rated power
output Pwr when wind speed is in range [vr vout ]. The wind
power output is calculated as [8]:

k
k
− vcr − vout
fw (Pw ) P =P
 =e −e c . (5)
w wr

The WT provides wind power in continuous form when wind

speed is in range [vin vr ] and wind power output for this region
is measured as [8]:
k(vr − vin )  Pw k−1 FIGURE 2. Windfarm power output.
fw (Pw ) = k × vin + (vr − vin )
c × Pwr Pwr
 vin + PPw (vr −vin ) k TABLE 4. Solar PV units and Lognormal PDF parameters [8].
wr
×e− c . (6)
The electricity generation is zero when wind speed is not in
range [vin vout ].
v

k
k
− cin − vout
fw (Pw ) P =0 = 1 − e
 +e c , (7)
w
energy sources connected at 8 and 13 buses with two solar
B. UNCERTAINTY MODELING OF SOLAR IRRADIANCE PV units. The lognormal PDF fitting and solar irradiance
frequency distributions for solar energy sources connected
The probabilistic model for solar irradiance I (W /m2 ) fol-
at 8 and 13 buses are achieved after executing simulation
lows lognormal PDF and it is generally utilized for han-
of 8000 Monte Carlo scenarios, which are plotted in Figure 3.
dling the stochastic solar power generation in hybrid power
The solar PV unit available or actual solar power generation
systems [58]. The uncertainty of solar irradiance I (W /m2 )
subject to solar irradiance I (W /m2 ) and solar PV unit elec-
mathematically can be written as [58]:
 −(lnI −µ)2 tricity generation modeled as [8]:
1 
fI (I ) = √ × e 2σ 2 for I > 0, (8) I

, for I ≥ Ic ,
I σ 2π Ppvr ×


Istd
Ppv (I ) = 2 (10)
where µ in mean of solar irradiance and σ is standard devi- I
, for 0 < I < Ic ,


Ppvr ×


ation of solar irradiance. Lognormal PDF mean (Mlgn ) is Istd × Ic
defined as: where Istd and Ic indicate the solar irradiance and certain solar
σ2


Mlgn = e µ+ 2 .
irradiance point in a standard environment, respectively. The
(9)
solar PV unit rated power output is represented as Ppvr .
Assumed values for mean (µ) of solar irradiance and standard In this study, we assumed Istd = 800 W /m2 and
deviation (σ ) have been specified in Table 4. In a modified Ic = 120 W /m2 . The rated power outputs related to solar
IEEE-30 bus test system, we have replaced two thermal PV units that are connected at generator bus 8 and 13 are

159814 VOLUME 9, 2021

M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

FIGURE 3. Lognormal fitting of solar irradiance. FIGURE 4. Power output from solar PV units.

Ppvr = 35 MW and Ppvr = 50 MW, respectively. The his-

tograms in Figure 4 represent the stochastic solar power
output based on solar irradiance I (W /m2 ) distribution shown
in Figure 3, from solar PV units.

C. UTILITY LOAD DEMAND UNCERTAINTY MODELING

The utility load demand is also stochastic due to variation in
consumers’ load demand that directly affects spinning reserve
cost in a power system. Therefore, modeling the utility load
demand uncertainty has a significant impact on solving the
OPF problem and achieving accuracy in planning and opera-
tions of the power system. In this study, utility active (real)
load is considered as a random variable and power factor
is considered as constant. According to the constant power
factor, the change in reactive power of each load bus or PQ
bus depends upon its active load in the power system. We have
used a Gaussian PDF [59], [60] to model the uncertainty of
utility load demand on each load bus in transmission system.
The prediction of load demand on a load bus follows FIGURE 5. Modified IEEE-30 bus test system.
Gaussian PDF as:
 (Pd,i −µi )2
1 −
2σi2
f (Pd,i ) = √ ×e , (11) We have utilized modified IEEE-30 bus system for perfor-
(2π)σi
mance evaluation of our proposed method, in which four
where, Pd,i is active load demand, σi is standard deviation thermal energy sources at 5, 8, 11, and 13 generator buses
and µi is mean value of active load at ith load bus. are replaced with RESs due to emission pollution and global

VOLUME 9, 2021 159815

 M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

FIGURE 6. Gaussian distribution of utility load demand on load or PQ buses.

warming issues. The down-arrow (↓) symbol represents load load of each load bus, and 10% of the mean value µ has
or PQ bus shown in Figure 5. The active (real) load demand been considered as standard deviations σ of active load on
mean value µ has been considered equal to the base active each load bus for modeling uncertainty of utility active load

159816 VOLUME 9, 2021

M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

demand across the modified IEEE-30 bus test system. Plotted

histograms in Figure 6 based on Gaussian PDF and represent
the distribution of active load demands on load or PQ buses,
which are achieved after execution of 8000 Monte-Carlo
simulation scenarios.
In the research literature, many approximation methods
based on the analytical approach have been documented [62]
for handling the uncertainty in power systems. The com-
mon uncertain source method, the discretization method,
the truncated Taylor series expansion method, and the point
estimate method are examples of these methods. Some of
the analytical approximation methods follow the uncertain or
random variable’s PDF. In 1998, H.P. Hong [61] developed an
efficient PEM to measure the moments of Z = h(x), where Z
represents an uncertain or random quantity and it is a function FIGURE 7. Utility load demnd on load buses.

of n uncertain variables. The PEM is a simple to use method

for measuring the moments Z and does not require derivatives In this research work, we have utilized Hong’s PEM [61]
of h(x) or any iteration as compared to other approximation by taking Gaussian PDFs of active load demands on modified
methods such as the discretization method and Taylor series IEEE-30 bus test system load or PQ buses as input from
expansion method. The PEM can be utilized directly with a plotted histograms in Figure 6, to calculate active load on
deterministic computer program. Based on the above facts, each bus. The deterministic load on each load bus obtained
we used Hong’s PEM for calculating the approximate load from using particular schemes of Hong’s PEM such as 2m
on load buses. and 2m+1 is specified in Table 5 and plotted in Figure 7. The
Hong’s PEM concentrates upon statistical information deterministic active load demand on each load bus based on
obtained through initial few moments of a random variable both 2m and 2m+1 schemes is similar. Therefore, we have
on m concentrations for every variable. In which m × n points used the simplest 2m scheme of Hong’s PEM for calculating
concentration matching is considered to achieve first non- load demand on each load bus to find an optimal solution to
crossed moments (m × n) and crossed second-order moments the OPF problem in the hybrid power system.
of each uncertain variable. The simplest case of Hong’s PEM
is the 2m scheme, in which skewness of an uncertain vari- III. THE OPF PROBLEM FORMULATION
able’s PDF and correlation between the uncertain variables The OPF in the traditional thermal energy sources-based
are considered for predicting the value of random quantity. power system is a quadratic nature nonconvex and nonlinear
Another particular case of Hong’s PEM is the 2m +1 scheme, problem, in which stable and secure settings of operating
in addition to the skewness of the uncertain variable’s PDF points in electricity generation plants are obtained for mini-
and correlation between the uncertain variables, kurtosis of mizing certain objectives. The OPF problem objective written
the uncertain variable’s PDF is also considered in this scheme as [45]:
to calculate the random quantity. More detail of Hong’s PEM
is available in [61]–[63].
Minimize: f (x, u)
s.t.: g(x, u) = 0
TABLE 5. Deterministic utility load on load buses.
h(x, u) ≤ 0, (12)

where, u is a set of control variables and x is a set of state

variables. The function f (x, u) represents the objective of
the OPF problem. The function g(x, u) represents equality
constraints and h(x, u) represents inequality constraints.
The power flow in system is controlled by control or
independent variables, while state variables described power
system state. The control variables consist of all bus genera-
tors active power excluding slack (swing) bus active power,
all generators or energy sources voltage magnitudes, shunt
compensator at selected buses, and transformer tap in power
system network. The state variables consist of generators’
reactive power, swing bus active power output, line load-
ing of transmission lines, and voltages magnitude at load
buses.

VOLUME 9, 2021 159817

 M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

A. CONSTRAINTS respectively. On load buses and generator buses voltage

Balancing both active (real) power and reactive power follow magnitude boundary limits are defined by constraint (21)
equality constraints, while security constraints and equip- and (22), respectively. NG is number of generator buses.
ment’s operating limits of transmission lines and load buses In each inequality constraints max and min superscripts show
follow inequality constraints. The description of both types boundary limits of the corresponding parameter. NL is num-
of constraints is provided herein. ber of load buses.

1) EQUALITY CONSTRAINTS B. OBJECTIVE FUNCTION

In a power system, active (real) power generation from energy To reduce electricity generation cost from traditional ther-
sources must be equal to active (real) load demands and power mal energy sources and RESs and including emission cost
loss in transmission lines. Similarly, reactive power output (e.g., carbon tax) is our objective for finding an optimal
from all energy sources also must be equal to demand and loss solution to the OPF problem in the hybrid power system. The
of reactive power. In the power system, equality constraints minimization of electricity output cost objective is stated as
can be written as according to study [45]: follows:
XNB 
PGi − PDi = Vi Vj Gij cos(θij ) + Bij sin(θij ) N
XWF
j=1 Minimize ξ = ξT (PTG ) + ξw,j (Pw,j )
∀ i ∈ NB (13) j=1
XNB 
QGi − QDi = Vi Vj Gij sin(θij ) − Bij cos(θij ) NPV
j=1 X
+ ξpv,k (Ppv,k ) + ξE . (23)
∀ i ∈ NB (14)
k=1
where, the term PGi indicates the active power and term
QGi represents reactive power generation from energy source The first term ξT (PTG ) represents power generation cost of
attached at ith bus. The term PDi represents the load demand thermal energy sources, which based on valve-point loading
PNWF
of active power and term QDi is demand of reactive power at effects quadratic curve. The second term j=1 ξw,j (Pw,j )
ith load bus. The term Vi represents ith bus voltage magnitude, represents the cost of power output from windfarms. In objec-
while term Vj represents jth bus voltage magnitude. The term tive function, the third term Nk=1 ξpv,k (Ppv,k ) is cost of solar
P PV
θij = θi − θj shows difference of voltage angles θi and θj at power output from solar PV units. The last term ξE is the
ith and jth buses. The term Gij is transfer conductance and Bij cost of emission (e.g., carbon tax). The detailed modeling of
is transfer susceptance in ith and jth buses. traditional thermal and RESs power generation cost functions
and emission cost are provided herein.
2) INEQUALITY CONSTRAINTS
In power systems, load buses and transmission lines security 1) THERMAL POWER COST CURVE
constraints, secure and stable equipment’s physical settings, The traditional thermal energy source required fossil fuel to
and equipment’s operating lower and upper limits are con- produce electricity. The power generation cost of fossil fuel
sidered as inequality constraints. These are mathematically based energy sources can be calculated by regular quadratic
described as follows [45]: fuel curve, valve-point effects quadratic fuel curve, and piece-
Pmin max
i = 1, . . . , NTG , (15) wise quadratic fuel curve [38]. In some practical cases,
TGi ≤ PTGi ≤ PTGi
thermal power is generated from traditional thermal energy
Pmin max
ws,j ≤ Pws,j ≤ Pws,j j = 1, . . . , NWF , (16) sources using different fossil fuels like natural gases,coal and
Pmin
pvs,k ≤ Ppvs,k ≤ Pmax
pvs,k k = 1, . . . , NPV , (17) oil. The power output cost from these types of energy sources
Qmin ≤ max
QTGi ≤ QTGi i = 1, . . . , NTG , (18) is calculated using the piecewise quadratic fuel cost curve.
TGi
In this study, we assumed that the same fossil fuel based
Qmin
ws,j ≤ Qws,j ≤ Qmax
ws,j j = 1, . . . , NWF , (19) traditional thermal energy sources are used for power genera-
Qmin
pvs,k ≤ Qpvs,k ≤ Qmax
pvs,k k = 1, . . . , NPV , (20) tion purposes. Therefore, to calculate the power cost related to
thermal energy source, we used two forms of fuel cost curve;
min
VGi ≤ max
VGi ≤ VGi i = 1, . . . , NG, (21)
1) quadratic fuel curve and 2) valve-point effects quadratic
VLmin
p
≤ VLp ≤ VLmax
p
p = 1, . . . , NL. (22) fuel curve. The generated power (MW) from thermal energy
The active (real) power output boundary limits of traditional source followed a quadratic relationship with fossil fuel cost
thermal energy source and RESs such as windfarms and solar ($/h) as [45]:
PV units are represented in constraints (15)-(17) and NTG , NTG
NWF and NPV are total number of traditional thermal energy
X
ξT0 (PTG ) = ai + bi PTGi + ci P2TGi , (24)
sources, windfarms, solar PV units, respectively. The reactive i=1
power output boundary limits of all energy sources or gen-
erators including traditional thermal energy sources, wind- where, PTGi is ith thermal energy source power output. ai , bi ,
farms, and solar PV units are defined by constraints (18)-(20), and ci are ith thermal energy source cost coefficients.

159818 VOLUME 9, 2021

M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

In a traditional thermal energy source-based power system, where, Pwav,j , Kwr,j , and fw (Pw,j ) represent available wind
the objective function is modeled using valve-point loading power, coefficient of windfarm reserve cost, and wind power
effects quadratic fuel cost curve for more precise and realistic for jth windfarm, respectively.
measuring of thermal power cost. Because traditional thermal In a scenario, when wind power supplied by windfarms
energy source’s steam turbines have multi-valve, in such operator is greater than the wind-scheduled power, if it not
case a variation befall in fuel cost curve. In such case, the possible to reduce power generation from thermal energy
fossil fuel cost curve of thermal energy source is measured sources, the windfarms surplus electricity is dumped and
using valve-point loading effects quadratic fuel cost curve as ISO bears penalty cost. The penalty cost λwp,j related to
follows [45]: jth windfarm can be formulated as [39]:
Z Pwr,j
NTG
λwp,j (Pwav,j − Pws,j ) = Kwp,j


X Pw,j − Pws,j
ξT (PTG ) = ai + bi PTGi + ci P2TGi Pws,j
i=1
×fw Pw,j dPw,j , (28)


TGi − PTGi ) ,
+ di × sin ei × (Pmin

(25)
where, Kwp,j is penalty cost coefficient of jth windfarm.
where, coefficients di and ei represent the valve-point loading The cost related to any windfarm power is calculated by
effect of ith TG. Pmin
TGi represents the minimum power output of adding penalty cost, reserve cost, and its wind-scheduled
ith TG during operation. For this study, cost coefficient values power cost. The cost coefficients and wind-scheduled power
of fuel cost curve and valve-point loading effects related to are specified in Table 7. The total wind power cost ξw,j (Pw,j )
TG1 and TG2 are specified in Table 6. for jth windfarm can be calculated by adding wind-scheduled
power cost and both penalty and reserve costs as [39]:
TABLE 6. Thermal power emission and cost coefficients [6].
ξw,j (Pw,j ) = λws,j (Pws,j ) + λwr,j (Pws,j − Pwav,j )
+λwp,j (Pwav,j − Pws,j ). (29)

TABLE 7. Cost coefficients and wind-scheduled power.

2) WIND POWER COST FUNCTION

We have assumed that private parties hold RESs such as solar 3) SOLAR POWER COST FUNCTION
PV units and windfarms and ISO purchases scheduled power
Solar power cost is also directly proportional to solar-
from private parties according to the signed agreement. The
scheduled power. The solar-scheduled power cost λpvs,k is a
wind power cost follows directly proportional relationship to
function of scheduled power provided from k th solar PV unit,
wind-scheduled power and can be formulated as:
as follows [42]:
λws,j (Pws,j ) = gj Pws,j , (26) λpvs,k (Ppvs,k ) = hk Ppvs,k , (30)
where, λws,j is a function to calculate power cost of wind- where, scheduled power cost coefficient hk is related to
scheduled power Pws,j received from jth windfarm and k th solar PV unit and Ppvs,k is delivered solar-scheduled
gj is cost coefficient of wind-scheduled power related to power from k th solar PV unit.
jth windfarm. In Figure 4, the distributions of power generation from
The distributions of windfarms power output are shown in solar PV units are plotted. Similar to the windfarms
Figure 2. The available power from windfarms can be less power output behaviour, in the underestimation scenario,
or more than wind-scheduled power because of fluctuant and the available power from solar PV units can be more than
stochastic wind power output. In an overestimation scenario, solar-scheduled power and in an overestimation scenario, the
when wind power supplied by windfarms operators is less available solar power can be less than solar-scheduled power.
than the wind-scheduled power, the ISO is required to have In such a case, the ISO requires a spinning reserve energy
a spinning reserve based on utility load demand. The wind source. According to the concept presented in the study [42],
power reserve cost λwr,j for jth windfarm can be calculated we have modeled solar reserve cost λpvr,k of k th solar PV
as [39]: unit power output, for an overestimation scenario. It can be
Z Pws,j formulated as [42]:
λwr,j (Pws,j − Pwav,j ) = Kwr,j


Pws,j − Pw,j
0 λpvr,k (Ppvs,k − Ppva,k ) = Kpvr,k × fpv (Ppva,k < Ppvs,k )
×fw Pw,j dPw,j , (27) × Ppvs,k − E(Ppva,k , < Ppvs,k ) , (31)



VOLUME 9, 2021 159819

 M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

where, Kpvr,k represents coefficient of reserve cost of Algorithm 1: Pseudocode of Proposed Method
k th solar PV unit, Ppva,k is actual available power received 1 Input: d: Dimension of problem or vector of control
by k th solar PV unit, and fpv (Ppva,k < Ppvs,k ) represents the variables Xmax = [xmax 1 , x2 , . . . , xd ] and
max max
calculation of overestimation scenario related to k th solar PV Xmin = [xmin , xmin , . . . , xmin
1 2 d ], N: Population size or birds
unit. E(Ppva,k < Ppvs,k ) represents the prediction (expecta- (uniformly distribution ∈ [Xmax , Xmin ])
tion) of power output from solar PV unit k th below solar- 2 M: Generations, P ∈ [0.8, 1]: The foraging probability,
scheduled power Ppvs,k . FQ=10: Frequency of bird’s flight, FL ∈ [0.5, 0.9]:
For underestimation scenario, the penalty cost λpvp,k Followed coefficient, C = S= 1.5, a1 = a2 =1:
related to k th solar PV unit power output can be formulated Constant parameters and ε: Smallest constant value
as [42]: 3 Define the itr=0 (iteration variables), related
λpvp,k (Ppva,k − Ppvs,k ) = Kpvp,k × fpv (Ppva,k > Ppvs,k ) parameters and fitness value for evaluation of
objective function, and finding an optimal solution
× E(Ppva,k > Ppvs,k ) − Ppvs,k , (32)

4 while (itr < M ) do
where, Kpvp,k represents the penalty cost coefficient for 5 if (itr mod FQ 6= 0) then
k th solar PV unit. fpv (Ppva,k > Ppvs,k ) represents the calcula- 6 for i=1 to N do
tion of underestimation scenarios related to k th solar PV unit. 7 if P > rand(0, 1) then
E(Ppva,k > Ppvs,k ) represents the prediction (expectation) of 8 Individual bird in swarm switch into
power output from solar PV unit k th above solar-scheduled foraging behaviour to explore food
power Ppvs,k . patches (Eq. 36)
Similar to windfarm power cost, the cost related to a solar 9 else
PV unit power is calculated by adding penalty cost, reserve 10 Otherwise bird switch int vigilance
cost, and solar-scheduled power cost. The cost coefficients behaviour(Eq. 37)
and solar-scheduled power related to solar PV units are spec- 11 end
ified in Table 8. The total solar power cost ξpv,k (Ppv,k ) for 12 end
k th solar PV unit by adding solar-scheduled power cost and 13 else
both penalty and reserve cost can be measured as [42]: 14 Based on their food reserves birds often
switched into producer birds and scrounger
ξpv,k (Ppv,k ) = λpvs,k (Ppvs,k ) + λpvr,k (Ppvs,k − Ppva,k ) birds, after arrived on new site.
+λpvp,k (Ppva,k − Ppvs,k ). (33) 15 for i=1 to N do
16 if ith bird is a producer then
TABLE 8. Cost coefficients and solar-scheduled power.
17 ith bird explore the search space for
food patches (Eq. 40)
18 else
19 Scrounging for hunting food patches
(Eq. 41)
20 end
4) EMISSION COST 21 end
The combustion of fossil fuels in traditional thermal sources 22 end
of energy is the core cause of greenhouse/harmful gases 23 The evaluation of new solutions vector or
including SOx , COx , and NOx emission into the environment. population
With growing global environmental concerns, to regulate 24 Greedy approach is applied by comparing the new
the power system for accounting the minimum emissions is solutions vector (population) and previous
necessary. The total emission E (ton/h) into the environment solutions vector
by a traditional thermal energy source can be formulated as 25 Search for global optimal
follows [6]: 26 itr= itr+1.
XNTG n o 27 end
αi +βi PTGi +γi P2TGi × 0.01+ωi × e(µi PTGi ) ,


E= 28 Output: An optimal solution in the new solutions
i=1
(34) vector (population)

where, emission coefficient for ith TG are αi , βi , γi , ωi , and µi has been applied to curb greenhouse gases and encourage
and values of these coefficient related to TG1 and TG2 are investment in clean forms of energy, [59]. Carbon tax (Ctax )
listed in Table 6. imposed on emitted greenhouse gases and emission cost
In the recent decade, due to global environmental issues, ($/ton) can be calculated as:
many countries are imposing a carbon tax to minimize carbon
emission into the environment. Therefore, carbon tax widely Emission cost, ξE = Ctax E. (35)

159820 VOLUME 9, 2021

M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

IV. PROPOSED APPROACH TO SOLVE THE OPF PROBLEM of bird’s behaviours such as vigilance behaviour, foraging
Various nature-inspired metaheuristic algorithms have been behaviour, and flight behaviour are briefly described here.
developed as a substitute to the mathematical methods
for solving optimization problems, in research literature. A. FORAGING BEHAVIOUR
Population-based BSA [56] is a new stochastic swarm intel- Stochastic decision (Rule 1) is taken according to the prob-
ligence algorithm. To address the optimization problems, ability P of bird foraging food. Individual birds in swarms
intelligence of bird swarms extracted from bird’s social switched into foraging behaviours if probability P is greater
behaviours has been utilized. The birds in the swarm improve than randomly selected constant value from a uniform normal
their fitness through social behaviours and interactions with distribution (0,1), otherwise the bird has vigilance behaviour.
other birds in the swarm. The working model of BSA is The best recorded experience or memory about searching
based on three types of bird behaviours such as foraging, vigi- food items can be utilized to explore food patches, and social
lance, and flight behaviour. The BSA can flexibly regulate its behaviour and information are shared immediately (Rule 2).
four different search strategies such as foraging, vigilance, It can be mathematically modeled as [56]:
producer, and scrounger to explore the search space. Based
t+1 t t



on these facts, the BSA can improve its convergence speed xi,j = xi,j + pi,j − xi,j × C × rand 0, 1
through better balancing between exploitation and explo- t
+ gj − xi,j


× S × rand 0, 1 , (36)


ration of search space without affecting the stability and accu-
racy of the optimal solution. Therefore, the proposed method where, C represents cognitive and S represents social acceler-
based on BSA may provide a more stable and accurate ated positive coefficients. pi,j represents the past best position
solution for the OPF problems in the hybrid power system. (local optimal) of ith bird, while gj is the swarm shared
A bird’s social behaviours and interactions with other birds best (global optimal) past position. The function rand(0,1)
in the swarm can be made understand based on well-defined represents uniform distribution of numbers in (0, 1). The term
rules as follows: t (i ∈ {1, 2, . . . , N }) represents N virtual birds’ position
xi,j
Rule 1: Individual birds in a swarm may switch into two at time t, having vigilance or foraging behaviour and term
types of behaviours; 1) foraging behaviour and 2) vigi- (j ∈ {1, 2, . . . , D}) represents dimensions of available search
lance behaviour, on the basis of the random or stochastic space in which birds take flight.
decision.
Rule 2: In a swarm, the individual bird may update B. VIGILANCE BEHAVIOUR
or improve fitness through social behaviour and by According to Rule 3, birds would not travel directly towards
promptly recording self and swarm’s best memory or the swarm’s center. However, individual birds may struggle to
previous experience to explore the food patches in a spe- travel towards the swarm’s center and birds’ movement may
cific area during foraging behaviour. The best-recorded be affected by competition with each other. Individual bird’s
experience or memory about searching food items can movement or vigilance behavior modeled as [56]:
be utilized to explore food patches, and social behaviour
and information are shared immediately. t+1 t t



xi,j = xi,j + A1 meanj − xi,j × rand 0, 1
Rule 3: On the basis of the bird’s vigilance behaviour, t
× rand − 1, 1 ,



+A2 pk,j − xi,j (37)
an individual bird wishes travel to the swarm’s center.  
The competition between birds’ movement towards the pFiti
A1 = a1 × exp − ×N , (38)
center of the swarm may affect the individual bird’s sumFit + ε
  
struggle to reach the center of the swarm. The proba- pFiti − pFitk N × pFitk
A2 = a2 × exp ,
bility of a bird near a swarm’s center based on birds’ |pFitk − pFiti | + ε sumFit + ε
food reserves and a bird having greater food reserves (39)
than other birds will be at the swarm’s center.
Rule 4: The birds in a swarm have flight behaviour due where, positive constants a1 and a2 are within range [0,2].
to foraging behaviour or any other reason. During the meanj is the average position of jth bird’s swarm and k(k 6 = i)
flight behaviour the birds in the swarm can be often represents a random nonnegative integer (k ∈ {1, 2, . . . , N }).
switched again into two types of birds; 1) producer birds The sumFit is best fitness values sum of birds swarm and pFiti
and scrounger birds on the basis of their food reserves. is ith bird best fitness value. The smallest positive constant ε
Birds that have food reserved between lowest and high- is used for avoiding zero division error.
est are randomly switched into scrounger and producer. Individual birds travel towards the swarm’s center because
Rule 5: After arrival at a new place, birds divide into of indirect and direct effects. The swarm average fitness value
producers and scroungers. The producers search food is measured in the form of indirect effect and induced by
items or patches and randomly followed by scroungers environments. The direct effect is made by specific interfer-
to search food patches. ence and A2 is used to simulate it. If k th bird (k 6= i) best
Precise pseudo code of proposed method based on above fitness value is better as compared to ith bird best fitness
defined rules of BSA is given in Algorithm 1. Three types value, in that case A2 > a2. It indicates the k th bird may

VOLUME 9, 2021 159821

 M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

suffer a smaller interference as compared to ith bird. There-

fore, k th bird would quickly move towards the swarm center
as compared to ith bird. However, there are some random
movements of birds towards the center of the swarm. In the
case of minimizing optimization problems, the bird’s smallest
fitness value is considered a better fitness value.

C. FLIGHT BEHAVIOUR
The birds in a swarm have flight behaviour due to foraging
behaviour or any other reason. During the flight behaviour
the birds in the swarm can be often switched again into two
types of birds; 1) producer birds and scrounger birds on the
basis of their food reserves (Rule4). The producer behaviours
and scrounger behaviour can be written as, respectively [56]:
t+1 t t
,


xi,j = xi,j + randn 0, 1 × xi,j (40)
xi,j = xi,j + xk,j − xi,j × FL × rand 0, 1 , (41)
t+1 t t t

where, k(k 6 = i), represents a nonnegative integer (k ∈

{1, 2, . . . , N }). FL is the following coefficient for hunt-
ing food – bird’s scrounger behaviour (Rule 5). The term
randn(0, 1) is random number normal distribution.

FIGURE 9. Load or PQ buses voltage magnitude profiles.

TABLE 9. The modified IEEE-30 bus test system.

including harmony search algorithm (HSA), DE, SHADE,

PSO, and ABC. For a fair comparison, we used the same
number of iterations and parameter settings in BSA approach
and other algorithms. For experimental purposes, generator
buses voltage magnitude lower and upper limits have been
kept [0.95, 1.1] p.u. based on 100 MVA. The voltage mag-
nitudes boundary limits of load or PQ buses have been kept
[0.95 1.05]p.u.. The modified IEEE-30 bus test system has
FIGURE 8. Optimal power generation cost convergence. 435.0 MW power generation capacity has been utilized for
performance evaluation and its further detailed is available in
V. SIMULATION RESULTS AND CASE STUDIES Table 9.
The simulation-based optimization results of the BSA are We have implemented the proposed method and other algo-
measured and a comparison is made with other algorithms rithms in MATLAB R2017a and used the MATPOWER6.0

159822 VOLUME 9, 2021

M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

TABLE 10. Simulation results for electricity generation cost case studies.

package for load flow calculation. The execution of dependent on the value of scale parameter c of Weibull PDF
simulation-based experiments have been performed on and the lowest wind power cost achieved at an intermediate
Microsoft Windows 10 64-bits with Intel Core(TM) i7-5500U value. Similarly, the electricity generation cost of an indi-
CPU @2.40 GHz and RAM @8.00 GB. We have conducted vidual solar PV unit is also calculated by adding penalty
two case studies for performance evaluation of the proposed cost, reserve cost, and solar-scheduled power cost related to
BSA method. Initially, we have solved the OPF problems solar PV unit. Solar-scheduled power cost related to solar
in a hybrid power system by considering the objective to PV units also follows a direct relationship to solar-scheduled
reduce electricity generation cost. In second case, carbon tax power. It is observed that solar power output cost did not
(e.g., emission cost) is included in power generation cost min- monotonically increase with the values of lognormal PDF
imization objective function to reduce emissions pollution. parameters such as standard deviation σ and mean µ for
solar irradiance. Therefore, serious attention is required to
choose the suitable value of solar-scheduled power associated
A. CASE STUDY 1: GENERATION COST MINIMIZATION with solar PV units. If the mean µ value is kept low, then
The simulation-based experiments have been conducted for it is suggested to choose a smaller value for solar-scheduled
performance evaluation of the BSA-based proposed method power.
and other metaheuristic algorithms to solve the OPF problems
in a hybrid power system. The OPF problem objective to TABLE 11. Case study 1: Trade-off.
reduce the electricity output cost is written in Eq. 23, exclud-
ing last term (emission cost or carbon tax). In Eq. 23, the
first term ξT (PTG ), represents the power output cost related
to thermal energy sources. The second term N j=1 ξw,j (Pw,j )
P WF
is power output cost related to windfarms and the third term
PNPV
k=1 ξpv,k (Ppv,k ) is power output cost of solar PV units.
The electricity generation cost of an individual windfarm is
calculated by adding three types of wind related cost such as In Table 10, optimal values of objective function, parame-
1) wind-scheduled power cost, 2) reserve cost, and 3) penalty ters, control and state variables obtained from BSA, and other
cost. Wind-scheduled power cost of an individual windfarm evolutionary algorithms are given, where minimum power
follows a direct relationship to wind-scheduled power and output cost is represented in boldface. A minimum value of
a high spinning reserve is required when wind-scheduled power generation cost 863.121 $/h as specified in Table 10
power is kept high. In such a case, overall wind power was obtained in BSA approach and it has outperformed as
cost increases while at a lower rate penalty cost decreases. compared to other metaheuristic algorithms. The optimal
The wind speed and wind electricity generation are highly power generation cost 864.082 $/h, 864.344 $/h, 864.454 $/h,

VOLUME 9, 2021 159823

 M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

865.339 $/h, and 866.977 $/h are obtained using PSO, ABC, utility load demand uncertainty and stochastic nature power
DE, SHADE, and HSA, respectively. The optimal power generation from RESs. The power generation cost for ther-
generation cost convergence of the proposed method based mal energy sources is measured using a valve-point loading
on BSA and other metaheuristic algorithms are graphically effects quadratic fuel curve. The Gaussian PDF, Lognormal
plotted in Figure 8a. All load or PQ buses voltage magnitude PDF, and Weibull PDF have been used for modeling uncer-
profiles obtained during performance evaluation of BSA and tainty of utility load demand, stochastic solar irradiance, and
other algorithms are plotted in Figure 9a. For case study 1, wind speed, respectively. The simulation-based optimization
the trade-off between optimal power generation cost and results have shown the superiority of the BSA to solve the
convergence time of the BSA-based proposed method and OPF problems by satisfying all constraints and minimum
other algorithms are presented in Table 11. power generation cost 863.121 $/h is achieved in case study 1.
It has been observed from optimization results that the gener-
B. CASE STUDY 2: GENERATION PLUS EMISSION COST ation of the clean form of energy from RESs has increased and
MINIMIZATION emission pollution has decreased in the hybrid power system
In case study 2, the minimum electricity generation cost by imposing a carbon tax. In case study 2, the proposed BSA
expressed in Eq. 23 including emission cost (e.g., carbon tax) approach has also outperformed and minimum electricity cost
is an objective function for performance evaluation of the 890.728 $/h is achieved as compared to other algorithms. The
BSA-based proposed method. In this study, a carbon tax rate comparative evaluation and simulation-based optimization
Ctax ($20/ton) is imposed on released NOx , SOx , and COx results confirmed the superiority of BSA approach over other
from fossil fuel-based energy sources. It is observed that the metaheuristic algorithms. The optimization performance of
generation of the clean form of energy from solar and wind BSA in terms of accuracy, stability, and efficiency have
have increased and emission (ton/h) pollution decreased in made it attractive for application to real-time optimization
the power system by imposing a carbon tax. problems. The simulation results have encouraged for further
study. In future, the application of proposed BSA approach
TABLE 12. Case study 2: Trade-off. can be extended in a large-scale traditional thermal energy
sources-based power system to solve the other optimization
problems such as unit commitment, chance-constrained OPF,
Transient stability constrained OPF.

REFERENCES
[1] S. Rahim, N. Javaid, R. D. Khan, N. Nawaz, and M. Iqbal, ‘‘A convex
optimization based decentralized real-time energy management model
with the optimal integration of microgrid in smart grid,’’ J. Cleaner Prod.,
For case study 2, optimum values of objective function vol. 236, Nov. 2019, Art. no. 117688.
related parameters, control variables, and state variables have [2] H. Xiao and M. Cao, ‘‘Balancing the demand and supply of a power grid
been specified in Table 10 by applying the proposed BSA system via reliability modeling and maintenance optimization,’’ Energy,
vol. 210, Nov. 2020, Art. no. 118470.
method and other algorithms. Simulation results indicate that [3] L. Gan, P. Jiang, B. Lev, and X. Zhou, ‘‘Balancing of supply and demand
the BSA outperforms and provides minimum power gener- of renewable energy power system: A review and bibliometric analysis,’’
ation cost 890.728 $/h in case study 2 along with satisfying Sustain. Futures, vol. 2, Mar. 2020, Art. no. 100013.
[4] I. Colak, S. Sagiroglu, G. Fulli, M. Yesilbudak, and C.-F. Covrig, ‘‘A survey
constraints. The optimal power generation cost 892.397 $/h, on the critical issues in smart grid technologies,’’ Renew. Sustain. Energy
892.779 $/h, 892.865 $/h, 894.076 $/h, and 895.441 $/h Rev., vol. 54, pp. 396–405, Feb. 2016.
are obtained from DE, ABC, PSO, SHADE, and HSA, [5] A. F. Attia, R. A. El Sehiemy, and H. M. Hasanien, ‘‘Optimal power
respectively. For case study 2, the convergence of optimal flow solution in power systems using a novel sine-cosine algorithm,’’ Int.
J. Electr. Power Energy Syst., vol. 99, pp. 331–343, Jun. 2018.
power generation costs using the proposed BSA method and [6] A. E. Chaib, H. R. E. H. Bouchekara, R. Mehasni, and M. A. Abido,
other algorithms is plotted in Figure 8b. All load or PQ buses ‘‘Optimal power flow with emission and non-smooth cost functions using
voltage magnitude profiles are plotted in Figure 9b. The trade- backtracking search optimization algorithm,’’ Int. J. Elect. Power Energy
Syst., vol. 81, pp. 64–77, Oct. 2016.
off between optimal power generation cost and convergence [7] S. Frank and S. Rebennack, ‘‘An introduction to optimal power flow:
time of proposed method and other algorithms is represented Theory, formulation, and examples,’’ IIE Trans., vol. 48, no. 12,
in Table 12. The simulation-based experiment results indicate pp. 1172–1197, Aug. 2016.
[8] P. P. Biswas, P. N. Suganthan, and G. A. J. Amaratunga, ‘‘Optimal power
that power generation from RESs increased when the carbon flow solutions incorporating stochastic wind and solar power,’’ Energy
tax (ton/h) was imposed on carbon emission from fossil fuel- Convers. Manage., vol. 148, pp. 1194–1207, Sep. 2017.
based energy sources. [9] S. S. Reddy, ‘‘Multi-objective optimal power flow for a thermal-wind-solar
power system,’’ J. Green Eng., vol. 7, no. 4, pp. 451–476, Apr. 2018.
[10] K. Zehar and S. Sayah, ‘‘Optimal power flow with environmental con-
VI. CONCLUSION straint using a fast successive linear programming algorithm: Application
We have proposed a new bio-inspired bird swarm algorithm to the Algerian power system,’’ Energy Convers. Manage., vol. 49, no. 11,
for finding optimal solutions to the OPF problems in the tra- pp. 3362–3366, Nov. 2008.
[11] P. Fortenbacher and T. Demiray, ‘‘Linear/quadratic programming-based
ditional thermal, wind, and solar energy sources-based hybrid optimal power flow using linear power flow and absolute loss approxima-
power system, in this study. In which, we have incorporated tions,’’ Int. J. Electr. Power Energy Syst., vol. 107, pp. 680–689, May 2019.

159824 VOLUME 9, 2021

M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

[12] S. Mhanna and P. Mancarella, ‘‘An exact sequential linear programming [34] M. A. Taher, S. Kamel, F. Jurado, and M. Ebeed, ‘‘An improved moth-
algorithm for the optimal power flow problem,’’ IEEE Trans. Power Syst., flame optimization algorithm for solving optimal power flow problem,’’
early access, Jul. 14, 2021, doi: 10.1109/TPWRS.2021.3097066. Int. Trans. Elect. Energy Syst., vol. 29, no. 3, p. e2743, Mar. 2019.
[13] H. Ambriz-Perez, E. Acha, and C. R. Fuerte-Esquivel, ‘‘Advanced SVC [35] P. P. Biswas, P. N. Suganthan, R. Mallipeddi, and G. A. J. Amaratunga,
models for Newton-Raphson load flow and Newton optimal power flow ‘‘Multi-objective optimal power flow solutions using a constraint han-
studies,’’ IEEE Trans. Power Syst., vol. 15, no. 1, pp. 129–136, Feb. 2000. dling technique of evolutionary algorithms,’’ Soft Comput., vol. 24, no. 4,
[14] B. Liu, J. Li, H. Ma, and Y. Liu, ‘‘Generalized benders decomposition pp. 2999–3023, Feb. 2020.
based dynamic optimal power flow considering discrete and continuous [36] G. Chen, J. Qian, Z. Zhang, and S. Li, ‘‘Application of modified pigeon-
decision variables,’’ IEEE Access, vol. 8, pp. 194260–194268, 2020. inspired optimization algorithm and constraint-objective sorting rule on
[15] R. S. Salgado and E. L. Rangel, ‘‘Optimal power flow solutions multi-objective optimal power flow problem,’’ Appl. Soft Comput., vol. 92,
through multi-objective programming,’’ Energy, vol. 42, no. 1, pp. 35–45, Jul. 2020, Art. no. 106321.
Jun. 2012. [37] H. Buch and I. N. Trivedi, ‘‘An efficient adaptive moth flame optimization
[16] V. Vasylius, A. Jonaitis, S. Gudžius, and V. Kopustinskas, ‘‘Multi-period algorithm for solving large-scale optimal power flow problem with POZ,
optimal power flow for identification of critical elements in a country multifuel and valve-point loading effect,’’ Iranian J. Sci. Technol., Trans.
scale high voltage power grid,’’ Rel. Eng. Syst. Saf., vol. 216, Dec. 2021, Electr. Eng., vol. 43, no. 4, pp. 1031–1051, Dec. 2019.
Art. no. 107959. [38] R. Roy and H. T. Jadhov, ‘‘Optimal power flow solution of power system
[17] S. Tu, A. Wachter, and E. Wei, ‘‘A two-stage decomposition approach incorporating stochastic wind power using Gbest guided artificial bee
for AC optimal power flow,’’ IEEE Trans. Power Syst., vol. 36, no. 1, colony algorithm,’’ Int. J. Elect. Power Energy Syst., vol. 64, pp. 562–578,
pp. 303–312, Jan. 2021. Jan. 2015.
[18] M. Pourakbari-Kasmaei, M. Lehtonen, M. Fotuhi-Firuzabad, [39] A. Panda and M. Tripathy, ‘‘Optimal power flow solution of wind inte-
M. Marzband, and J. R. S. Mantovani, ‘‘Optimal power flow problem grated power system using modified bacteria foraging algorithm,’’ Int.
considering multiple-fuel options and disjoint operating zones: A solver- J. Elect. Power Energy Syst., vol. 54, pp. 306–314, Jan. 2014.
friendly MINLP model,’’ Int. J. Electr. Power Energy Syst., vol. 113, [40] A. Panda and M. Tripathy, ‘‘Security constrained optimal power flow solu-
pp. 45–55, Dec. 2019. tion of wind-thermal generation system using modified bacteria foraging
[19] F. Capitanescu and L. Wehenkel, ‘‘Experiments with the interior-point algorithm,’’ Energy, vol. 93, pp. 816–827, Dec. 2015.
method for solving large scale optimal power flow problems,’’ Electr. [41] S. R. Salkuti, ‘‘Optimal power flow using multi-objective glowworm
Power Syst. Res., vol. 95, pp. 276–283, Feb. 2013. swarm optimization algorithm in a wind energy integrated power system,’’
[20] H. Wei, H. Sasaki, J. Kubokawa, and R. Yokoyama, ‘‘An interior point Int. J. Green Energy, vol. 16, no. 15, pp. 1547–1561, Dec. 2019.
nonlinear programming for optimal power flow problems with a novel [42] L. Shi, C. Wang, L. Yao, Y. Ni, and M. Bazargan, ‘‘Optimal power
data structure,’’ IEEE Trans. Power Syst., vol. 13, no. 3, pp. 870–877, flow solution incorporating wind power,’’ IEEE Syst. J., vol. 6, no. 2,
Aug. 1998. pp. 233–241, Jun. 2012.
[21] E. P. de Carvalho, A. dos Santos, and T. F. Ma, ‘‘Reduced gradient [43] I. U. Khan, N. Javaid, K. A. A. Gamage, C. J. Taylor, S. Baig, and
method combined with augmented Lagrangian and barrier for the optimal X. Ma, ‘‘Heuristic algorithm based optimal power flow model incor-
power flow problem,’’ Appl. Math. Comput., vol. 200, no. 2, pp. 529–536, porating stochastic renewable energy sources,’’ IEEE Access, vol. 8,
Jul. 2008. pp. 148622–148643, 2020.
[44] M. A. Ilyas, G. Abbas, T. Alquthami, M. Awais, and M. B. Rasheed,
[22] X. Yuan, B. Zhang, P. Wang, J. Liang, Y. Yuan, Y. Huang, and X. Lei,
‘‘Multi-objective optimal power flow with integration of renewable
‘‘Multi-objective optimal power flow based on improved strength Pareto
energy sources using fuzzy membership function,’’ IEEE Access, vol. 8,
evolutionary algorithm,’’ Energy, vol. 122, pp. 70–82, Mar. 2017.
pp. 143185–143200, 2020.
[23] N. Daryani, M. T. Hagh, and S. Teimourzadeh, ‘‘Adaptive group search
[45] S. Li, W. Gong, L. Wang, X. Yan, and C. Hu, ‘‘Optimal power flow
optimization algorithm for multi-objective optimal power flow problem,’’
by means of improved adaptive differential evolution,’’ Energy, vol. 198,
Appl. Soft Comput., vol. 38, pp. 1012–1024, Jan. 2016.
May 2020, Art. no. 117314.
[24] K. Abaci and V. Yamacli, ‘‘Differential search algorithm for solving multi-
[46] J. Ben Hmida, T. Chambers, and J. Lee, ‘‘Solving constrained optimal
objective optimal power flow problem,’’ Int. J. Electr. Power Energy Syst.,
power flow with renewables using hybrid modified imperialist competitive
vol. 79, pp. 1–10, Jul. 2016.
algorithm and sequential quadratic programming,’’ Electr. Power Syst.
[25] H. R. E. H. Bouchekara, A. E. Chaib, M. A. Abido, and R. A. El-Sehiemy, Res., vol. 177, Dec. 2019, Art. no. 105989.
‘‘Optimal power flow using an improved colliding bodies optimization [47] E. E. Elattar and S. ElSayed, ‘‘Modified JAYA algorithm for optimal
algorithm,’’ Appl. Soft Comput., vol. 42, pp. 119–131, May 2016. power flow incorporating renewable energy sources considering the cost,
[26] A.-A. A. Mohamed, Y. S. Mohamed, A. A. M. El-Gaafary, and emission, power loss and voltage profile improvement,’’ Energy, vol. 178,
A. M. Hemeida, ‘‘Optimal power flow using moth swarm algorithm,’’ pp. 598–609, Jul. 2019.
Electr. Power Syst. Res., vol. 142, pp. 190–206, Jan. 2017. [48] Z. Ullah, S. Wang, J. Radosavljević, and J. Lai, ‘‘A solution to the optimal
[27] H. Pulluri, R. Naresh, and V. Sharma, ‘‘Application of stud krill herd power flow problem considering WT and PV generation,’’ IEEE Access,
algorithm for solution of optimal power flow problems,’’ Int. Trans. Electr. vol. 7, pp. 46763–46772, 2019.
Energy Syst., vol. 27, no. 6, p. e2316, Jun. 2017. [49] M. H. Sulaiman and Z. Mustaffa, ‘‘Optimal power flow incorporat-
[28] H. Pulluri, R. Naresh, and V. Sharma, ‘‘A solution network based on stud ing stochastic wind and solar generation by metaheuristic optimizers,’’
krill herd algorithm for optimal power flow problems,’’ Soft Comput., Microsyst. Technol., vol. 27, pp. 3263–3277, Oct. 2020.
vol. 22, no. 1, pp. 159–176, Jan. 2018. [50] T. Samakpong, W. Ongsakul, and N. Madhu Manjiparambil, ‘‘Opti-
[29] P. P. Biswas, P. N. Suganthan, R. Mallipeddi, and G. A. J. Amaratunga, mal power flow incorporating renewable uncertainty related opportunity
‘‘Optimal power flow solutions using differential evolution algorithm inte- costs,’’ Comput. Intell., to be published.
grated with effective constraint handling techniques,’’ Eng. Appl. Artif. [51] S. Duman, S. Rivera, J. Li, and L. Wu, ‘‘Optimal power flow of power sys-
Intell., vol. 68, pp. 81–100, Feb. 2018. tems with controllable wind-photovoltaic energy systems via differential
[30] A. Gacem and D. Benattous, ‘‘Hybrid genetic algorithm and particle swarm evolutionary particle swarm optimization,’’ Int. Trans. Elect. Energy Syst.,
for optimal power flow with non-smooth fuel cost functions,’’ Int. J. Syst. vol. 30, no. 4, Apr. 2020, Art. no. e12270.
Assurance Eng. Manage., vol. 8, no. S1, pp. 146–153, Jan. 2017. [52] A. Panda, U. Mishra, M.-L. Tseng, and M. H. Ali, ‘‘Hybrid power sys-
[31] H. R. E. H. Bouchekara, A. E. Chaib, and M. A. Abido, ‘‘Optimal power tems with emission minimization: Multi-objective optimal operation,’’
flow using GA with a new multi-parent crossover considering: Prohibited J. Cleaner Prod., vol. 268, Sep. 2020, Art. no. 121418.
zones, valve-point effect, multi-fuels and emission,’’ Electr. Eng., vol. 100, [53] M. A. M. Shaheen, H. M. Hasanien, S. F. Mekhamer, and
no. 1, pp. 151–165, Mar. 2018. H. E. A. Talaat, ‘‘Optimal power flow of power systems including
[32] T. T. Nguyen, ‘‘A high performance social spider optimization algorithm distributed generation units using sunflower optimization algorithm,’’
for optimal power flow solution with single objective optimization,’’ IEEE Access, vol. 7, pp. 109289–109300, 2019.
Energy, vol. 171, pp. 218–240, Mar. 2019. [54] A. M. Shaheen, R. A. El-Sehiemy, and S. M. Farrag, ‘‘Solving multi-
[33] M. A. Taher, S. Kamel, F. Jurado, and M. Ebeed, ‘‘Modified grasshopper objective optimal power flow problem via forced initialised differential
optimization framework for optimal power flow solution,’’ Elect. Eng., evolution algorithm,’’ IET Gener., Transmiss. Distrib., vol. 10, no. 7,
vol. 101, no. 1, pp. 121–148, Apr. 2019. pp. 1634–1647, May 2016.

VOLUME 9, 2021 159825

 M. Ahmad et al.: Bio-Inspired Heuristic Algorithm for Solving OPF Problem in Hybrid Power System

[55] A. R. Bhowmik and A. K. Chakraborty, ‘‘Solution of optimal power IFTIKHAR AZIM NIAZ (Senior Member, IEEE)
flow using non dominated sorting multi objective opposition based grav- is currently a Professional Engineer and an Assis-
itational search algorithm,’’ Int. J. Elect. Power Energy Syst., vol. 64, tant Professor, and the Ph.D. Coordinator with the
pp. 1237–1250, Jan. 2015. Department of Computer Science. He has been
[56] X.-B. Meng, X. Z. Gao, L. Lu, Y. Liu, and H. Zhang, ‘‘A new bio-inspired coordinating with different government and pri-
optimisation algorithm: Bird swarm algorithm,’’ J. Exp. Theor. Artif. Intell., vate sectors and organizations, improving the aca-
vol. 28, pp. 673–687, Jun. 2015. demic standard of students. He has been the Ph.D.
[57] M. Ahmad, N. Javaid, I. A. Niaz, S. Shafiq, O. U. Rehman, and
Supervisor of the HEC, since 2007. Because of
H. M. Hussain, ‘‘Application of bird swarm algorithm for solution of
his professional experience, expertise, social and
optimal power flow problems,’’ in Proc. Conf. Complex, Intell., Softw.
Intensive Syst., Matsue, Japan, Jul. 2018, pp. 280–291. communication skills, he has been on the expert
[58] T. P. Chang, ‘‘Investigation on frequency distribution of global radiation panel of the PEC Accreditation Team, since 2009. He is also a member of
using different probability density functions,’’ Int. J. Appl. Sci. Eng., vol. 8, various national and international professional bodies and clubs that is one
no. 2, pp. 99–107, Sep. 2010. of the necessary conditions for managing projects. His research interests
[59] F. Yao, Z. Y. Dong, K. Meng, Z. Xu, H. H.-C. Iu, and K. P. Wong, include wireless sensor networks, energy optimization in smart grids, and
‘‘Quantum-inspired particle swarm optimization for power system oper- software engineering.
ations considering wind power uncertainty and carbon tax in Australia,’’
IEEE Trans. Ind. Informat., vol. 8, no. 4, pp. 880–888, Nov. 2012.
[60] M. Ebeed and S. H. Aleem, ‘‘Overview of uncertainties in modern
power systems: Uncertainty models and methods,’’ Uncertainties
in Modern Power Systems. New York, NY, USA: Academic,
2021, pp. 1–34. Accessed: Oct. 28, 2021. [Online]. Available:
https://www.sciencedirect.com/book/9780128204917/
[61] H. P. Hong, ‘‘An efficient point estimate method for probabilistic analysis,’’
Rel. Eng. Syst. Saf., vol. 59, no. 3, pp. 261–267, 1998.
[62] M. Aien, A. Hajebrahimi, and M. Fotuhi-Firuzabad, ‘‘A comprehensive
review on uncertainty modeling techniques in power system studies,’’
Renew. Sustain. Energy Rev., vol. 57, pp. 1077–1089, May 2016.
[63] X. Li, J. Cao, and D. Du, ‘‘Probabilistic optimal power flow for power sys- AHMAD ALMOGREN (Senior Member, IEEE)
tems considering wind uncertainty and load correlation,’’ Neurocomputing, received the Ph.D. degree in computer science
vol. 148, pp. 240–247, Jan. 2015. from Southern Methodist University, Dallas, TX,
[64] S. Shargh, B. K. Ghazani, B. Mohammadi-Ivatloo, H. Seyedi, and USA, in 2002. He is currently a Professor with the
M. Abapour, ‘‘Probabilistic multi-objective optimal power flow consider- Computer Science Department, College of Com-
ing correlated wind power and load uncertainties,’’ Renew. Energy, vol. 94,
puter and Information Sciences (CCIS), King Saud
pp. 10–21, Aug. 2016.
University (KSU), Riyadh, Saudi Arabia, where he
is also the Director of the Cyber Security Chair,
MANZOOR AHMAD received the master’s CCIS. Previously, he worked as the Vice Dean of
degree in computer science from Quaid-i-Azam the Development and Quality at CCIS. He also
University, Islamabad, Pakistan, in 2002, and worked as the Dean of the College of Computer and Information Sci-
the master’s degree in computer science from ences and the Head of the Academic Accreditation Council, Al Yamamah
Mälardalens högskola Eskilstuna Västerås, Swe- University. His research interests include mobile-pervasive computing and
den, in 2010. He is currently pursuing the Ph.D. cyber security. He served as the General Chair for the IEEE Smart World
degree in computer science with COMSATS Uni- Symposium and a Technical Program Committee Member of numerous
versity Islamabad, Islamabad, under supervision international conferences/workshops, such as IEEE CCNC, ACM BodyNets,
of Dr. Iftikhar Azim Niaz and co-supervision of and IEEE HPCC.
Dr. Nadeem Javaid. Since 2003, he has been asso-
ciated with COMSATS University Islamabad. He has three research publica-
tions in well reputed international conferences. His research interests include
artificial intelligence and optimal power flow.

NADEEM JAVAID (Senior Member, IEEE)

received the bachelor’s degree in computer sci-
ence from Gomal University, Dera Ismail Khan,
Pakistan, in 1995, the master’s degree in elec-
tronics from Quaid-i-Azam University, Islamabad,
Pakistan, in 1999, and the Ph.D. degree from the
University of Paris-Est, France, in 2010. He is AYMAN RADWAN (Senior Member, IEEE)
currently an Associate Professor and the Found- received the Ph.D. degree from Queen’s Univer-
ing Director of the Communications Over Sen- sity, Kingston, ON, Canada, in 2009. He is cur-
sors (ComSens) Research Laboratory, Department rently a Senior Research Engineer (Investigador
of Computer Science, COMSATS University Islamabad, Islamabad. He is Auxiliar) with the Instituto de Telecomunicações,
also working as a Visiting Professor with the School of Computer Sci- University of Aveiro, Aveiro, Portugal. He is
ence, University of Technology Sydney, Sydney, Australia. He has super- mainly specialized in coordination and manage-
vised 137 master and 24 Ph.D. theses. He has authored over 900 articles ment of EU funded projects. He participated in the
in technical journals and international conferences. His research interests coordination of multiple EU projects. He is also
include energy optimization in smart grids and in wireless sensor networks the Project Coordinator of the CELTIC+ Project
using data analytics and blockchain. He was a recipient of the Best University ‘‘MUSCLES,’’ and participating in the coordination of ITN-SECRET. He has
Teacher Award from the Higher Education Commission of Pakistan, in 2016, also been the Technical Manager of the FP7-C2POWER Project and the
and the Research Productivity Award from the Pakistan Council for Science Coordinator of the CELTIC+ ‘‘Green-T’’ Project. His current research inter-
and Technology, in 2017. He is also an Associate Editor of IEEE ACCESS and ests include the Internet of Things, 5G, and green communications.
an Editor of Sustainable Cities and Society journals.

159826 VOLUME 9, 2021