Ain Shams Engineering Journal 12 (2021) 621–630

Contents lists available at ScienceDirect

Ain Shams Engineering Journal

journal homepage: www.sciencedirect.com

Electrical Engineering

A novel hybrid GWO-PSO optimization technique for optimal reactive

power dispatch problem solution
Mohamed A.M. Shaheen a,⇑, Hany M. Hasanien b, Abdulaziz Alkuhayli c
a
Electrical Engineering Department, Future University in Egypt, Cairo 11835, Egypt
b
Electrical Engineering Department, Faculty of Engineering and Technology, Future University in Egypt, Cairo 11835, Egypt
c
Electrical Engineering Department, College of Engineering, King Saud University, Riyadh, Saudi Arabia

a r t i c l e i n f o a b s t r a c t

Article history: This paper provides an application of the hybrid Grey Wolf Optimization and Particle Swarm
Received 2 June 2020 Optimization (GWO-PSO) method to reach a solution to the optimal reactive power dispatch (ORPD)
Revised 6 July 2020 problem in the scope of electric power networks. PSO is a swarm based meta-heuristic optimization algo-
Accepted 11 July 2020
rithm whose target is to seek the best solution to a problem by moving particles in a specific exploration
Available online 6 August 2020
field. On the other hand, GWO is a meta-heuristic optimization technique which is inspired by grey
wolves. In this article, GWO is hybridized with a PSO method to improve the progress of the GWO.
Keywords:
There are two objectives minimized in this research study to improve the electric power network perfor-
Improvement of load buses voltage
GWO
mance. They are: 1) power losses in the transmission systems, and 2) the deviation of voltages at the load
Minimization of power losses buses. The problem of ORPD has many restrictions on the networks which must be considered during the
ORPD solution. The hybrid GWO-PSO is proven as an effective optimization technique when seeking the global
PSO best solution to an optimization problem. The success of the introduced hybrid technique is verified uti-
lizing more than one standard IEEE test system. A valuation to the introduced technique is performed by
comparing it with other optimization techniques stated through the literature. The simulation results
confirm that the usage of the hybrid GWO-PSO techniques causes an observable improvement in a wide
scale of the electric power networks behavior.
Ó 2020 The Authors. Published by Elsevier B.V. on behalf of Faculty of Engineering, Ain Shams University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-
nd/4.0/).

1. Introduction voltages, tapping settings of transformers, and the location of

shunt capacitors. When the solution of the optimization problem
Dealing with the optimal reactive power dispatch (ORPD) prob- -ORPD- is reached, the constraints keep within their maximum
lem is important when talking about the field of electric power and minimum bounds [1–6].
networks. Many objectives are to be optimized in this problem. The ORPD problem is considered as a non-linear optimization
For example, the minimization of the power losses, improving one. In the literature, it was resolved by several conventional
the voltage magnitude at the PQ buses. The voltage can be methods [7–12]. Linear programming (LP) [7], interior point
enhanced by minimizing the voltage deviation. Generally, the method [8], quadratic programming (QP) [9], dynamic program-
objectives are reached by targeting the best tuning of specified ming (DP), and Newton-Raphson [10] are all examples of the con-
control parameters. These control parameters can be the generator ventional methods.
The conventional techniques suffered from many weak points.
For example, they consume several numerical iterations which
⇑ Corresponding author. lead to long computational time to come to the end of simulations
E-mail addresses: mohamed.shaheen@fue.edu.eg (M.A.M. Shaheen), hanyhasa- without approximations. Another example is that the assumptions
nien@ieee.org (H.M. Hasanien), aalkuhayli@ksu.edu.sa (A. Alkuhayli). conventional methods should be replaced by other ones which can
Peer review under responsibility of Ain Shams University. overcome their weak points. In the past years, there was an ongo-
ing progress in the field of optimization methods. As a result, there
are now numerous optimization methods that can be tested on
various applications in the scope of electric power networks. Dif-
Production and hosting by Elsevier ferent techniques have been developed to search for the optimal

https://doi.org/10.1016/j.asej.2020.07.011
2090-4479/Ó 2020 The Authors. Published by Elsevier B.V. on behalf of Faculty of Engineering, Ain Shams University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
622 M.A.M. Shaheen et al. / Ain Shams Engineering Journal 12 (2021) 621–630

solution of the ORPD. Hybrid PSO and imperialist competitive algo- erator bus voltage, transformer tap settings, and shunt capacitor
rithms [4], wind driven optimization algorithm [5], genetic algo- output reactive powers [37]. Accordingly, the dependent variables
rithm [11], SFO [13,14], HFPSO [15], gravitational based search stated in (4) such as the active generated power at the slack bus,
optimizer [13,16], brain storm optimization algorithm (BSOA) reactive power outputs at the PV buses and voltages at the PQ
[17], TSA [18], Salp swarm optimization [19–21], Sine-Cosine algo- buses. The ORPD is explained in [34–40] as:
rithm [22], Coyote optimization algorithm [23], Seeker optimiza-
tion algorithm [24], bacteria foraging optimization (BFO) [25], 2.1. Objectives
cuckoo search algorithm (CSA) [26–28], Grasshopper optimization
algorithm [29], and firefly algorithm [30], are all examples of the There are two goals for the research. They are: 1) real power
recently used optimization techniques. Moreover in [31], the losses reduction in the network under study (case 1), and 2)
genetic algorithm (GA) is used to enhance the voltage stability. improving the voltage profile. A mathematical representation of
The dynamic PSO is also used to minimize the power losses in the first fitness function can be written as in (5):
[32]. Improved DEA method with the insertion of penalty factor X X
to reduce the power losses and to improve of voltage profiles at minPL ¼ Ploss
k ¼ Gk ðV2i þ V2j  2Vi Vj coshk Þ ð5Þ
k2Nl k2Nl
load buses is proposed in [33].
The hybrid GWO-PSO optimization technique is a newly devel- where, V i and V j are the magnitudes of the buses i and j voltages
oped technique which is introduced in [34]. It can handle both the respectively, Gk represents mutual conductance between bus i and
unconstrained as well as the constrained optimization problems. j,hk represents the difference in voltage angles between bus i and
This method targets the optimal solution by keeping the better bus j.
solutions and terminating the furthest ones. Meanwhile, the second fitness function targets the minimum
In this article, a novel technique known as hybrid GWO-PSO voltage deviation. So, the mathematical representation of the sec-
algorithm is inspected to optimize the control parameters stated ond fitness function (F2) is written as shown in (2):
in the problem of ORPD. The results of the fitness functions, their
convergence, and the control parameter selection are inspected. X
N Load

Despite the being of other optimization techniques, the hybrid min F2 ¼ VD ¼ jVi  Vref j ð6Þ
i¼1
GWO-PSO optimization one confirmed its simplicity, effectiveness,
and easiness as a tool to be used in the implementation and it also where, Vref is the defined reference for the voltage of the buses, and
showed competitive performance when comparing it with other NLoad represents the number of load buses in the tested network.
optimization techniques. So, this article focuses on finding the best
control parameters to reduce the active power losses of the system. 2.2. System limits
Also, the article introduces how to enhance the voltage profiles at
PQ buses using the aforementioned optimization method. This 2.2.1. The equality constraints
research is applied to more than one test system. These tests are To minimize the fitness function, there are equality constraints
done on various systems to prove the competence of the intro- to be considered. These constraints are stated as shown in (7) and
duced hybrid optimization method. Single objectives cases are (8):
considered in this study. XNB
Pgi  Pdi  Vi j¼1
Vj ðGk coshk þ Bk sinhk Þ ¼ 0 ð7Þ
2. Problem formulation
XNB
Q gi þ Q ci  Q di  Vi j¼1
Vj ðGk sinhk  Bk coshk Þ ¼ 0; iNpq ð8Þ
The ORPD, in general representation, aims at optimizing a non-
linear fitness function while keeping many constraints of operation where, Bk represents the susceptance between bus i and bus j, P gi
within their specified maximum and minimum limits [35]. In a and Q gi represent the active and reactive generated powers at bus
Mathematical form, it can be written as shown in (1) and (2): i, P di , Q di are the real and reactive load at bus i;Q ci is the reactive
MinFðx; uÞ ð1Þ power of the VAR supply inserted at bus i, NB is the number of
buses. Npq is the number of the PQ buses.
Subjectto : gðx; uÞ ¼ 0;
ð2Þ 2.2.2. The inequality constraints
hðx; uÞ  0
The inequality constraints, also, must be kept within their limits
where, F represents the fitness function, u and x represents the and they are explained in (9)–(14):
decision parameters and the dependent variables.
The decision parameters are defined to be the generators’ volt- Vmin
i  Vi  Vmax
i ; iNB ð9Þ
ages, transformers’ tap settings, and the reactive power injected
m  Tm  Tm ; mNt
Tmin max
into the network by the shunt capacitors. The control parameters ð10Þ
can be stated as follows:
gi  Q gi  Q gi ; iNpv
Q min max
ð11Þ
uT ¼ ½VG1    VGNGen ; Q C1    Q CNCom ; T1    TNTr  ð3Þ

ci ; iNc
Also, the dependent variables can be stated as follows: Q min  Q ci  Q max ð12Þ
ci

x ¼ ½VL1    VLNLB ; Q G1    Q GNGen ; Sl1    SlNline 

T
ð4Þ
Pmin
s  Ps  Pmax
s ð13Þ
The ORPD fitness function targets the most minimization in the
real power losses, while keeping the equality and inequality con- Sk  Smax ; kNl ð14Þ
k
straints of the electric power network within their limits [36]. To
Solve the problem of ORPD, this means to reach the fine tunings where, T m represents the change in tapping of transformer m, N pv
of the controlled parameters stated in (3). The parameters are gen- represents the number of PV buses, N t represents the number of
 M.A.M. Shaheen et al. / Ain Shams Engineering Journal 12 (2021) 621–630 623

the on-load tap changing transformers. Nc is the number of capaci-

tors, Nl is the number of branches in the power network, Ps is the
real power at slack bus, Pmin
s and P max
s are the lower and upper
bounds of the power at the slack bus, and Sl is the apparent power
through a transmission line l.

3. Proposed algorithms

3.1. Particle swarm optimization technique

The PSO is classified as a swarm-based metaheuristic optimiza-

tion technique [41]. In PSO, the exploration strategy for finding the
global optimization is a population-based one. Its motivation
comes from the birds’ behavior [42]. The particles are taken, then
moved around through the exploration field in order to determine
the optimal population satisfying the problem best solution. Parti-
cles are generated in a multidimensional exploration field, and
every particle uses its own experience as well as its neighboring
particle experience to adjust its position accordingly. Also, particles
are guided by the optimal position reached by themselves and
their neighboring particles [41]. The strong points of the PSO are
that it is easy to perform and there is no need for many parameters’
adjustments. Change in the velocity item can be set according to
the following rules [42]:
if x > pbestx; v x  randxa; otherwise, v x ¼ v x þ randxa.
if y > pbesty; v y  randxa; otherwise, v y ¼ v y þ randxa.
Where pbest is the best position reached ever. a is velocity
adjusting constant, rand denotes a random number in [0,1]. The Fig. 2. Flowchart of the GA.
velocity must also update according to the following rules [42]:
if x > gbestx; v x ¼ v x  randxb; otherwise, v x ¼ v x þ randxb.
if y > gbesty; v y ¼ v y  randxb; otherwise, v y ¼ v y þ randxb.
where gbest is the best location of the total swarm so far. And b 3.2. Grey wolf optimization technique
is the velocity adjusting constant
The GWO was initially introduced by Mirjalili et al., [42]. Its
inspiration comes from the behavior and the hunting strategy of
the grey wolves in nature [43]. In a group, the grey wolves follow
a strong leadership hierarchy. The leaders of the group are called
alpha (a) wolves. The next category of the grey wolves is consid-
ered to be of the secondary ones. They help the alpha ones. They
are known as the beta (b) wolves. Moreover, delta (d) wolves have
priority level lower than alpha and beta wolves. Their mission is to
submit to alpha wolves and beta ones, but control the omega
wolves. The least priority wolves are the omegas (x), which must
follow the leading grey wolves [44]. The GWO technique is
explained in the mathematical representations as follows [43]:

3.2.1. Social hierarchy levels

In the mathematical representation of the hierarchy of the grey
wolves, alpha wolf is known as the most appropriate solution. Con-
sequently, beta wolf is known as the second most appropriate solu-
tion and then, the next appropriate solution is called delta. The
other populations which represent the furthest solutions are
known to be the omegas (x). Through the GWO technique, the

Table 1
Fitness function solution of case
1 for the 14-bus system.

Method Ploss (MW)

GWO-PSO 13.2716
PSO 13.2718
HHO 13.2787
SFO 13.2786
Fig. 1. Flowchart of hybrid GWO-PSO algorithm.
624 M.A.M. Shaheen et al. / Ain Shams Engineering Journal 12 (2021) 621–630

Table 2
Control variables of ORPD for the IEEE 14-bus system in Case 1.

Method
GWO-PSO SFO HHO PSO
Control variables V1 1.0534 1.0171 1.0599 1.008
V2 1.06 0.9909 1.0545 1.06
V3 0.94 0.952 1.06 0.9411
V6 0.9525 1.0099 1.0175 0.9941
V8 0.9561 1.0279 1.0517 1.0155
Q9 0.2009 5.2264 0.5153 0.2
Q14 6.7453 6.4784 6.0592 3.3283
T4-7 0.9972 0.998 1.0599 0.9972
T4-9 0.9623 0.9867 0.9 0.963
T5-6 0.9882 0.9788 0.9807 0.9882

hunting process is directed by alpha, beta, and delta. The omegas 3.2.4. Exploring a victim and attacking
should only take the same steps and obey the wolves of higher ‘A’ is a value between [2a, 2a]. If |A| < 1, the wolves attack their
priorities. victim. Attacking the victim is the exploitation skill and finding a
victim is the exploration skill. The haphazard values of ‘A’ guide
3.2.2. Surrounding the victim: the population to go away from the victim. If |A| > 1, the wolves
The grey wolves encircle the victim while hunting. The sur- diverge from their victim.
rounding strategy, is mathematically represented as in (15) and
(16): 3.3. Elementary procedures of the hybrid GWO-PSO technique

! ! ! !   Initialization of the search agents and defining the solution area.
D ¼  C : X P ðtÞ  X ðt Þ ð15Þ
 Run the GWO technique.
! ! !!  Generation of the lowest values for all agents.
X ðt þ 1Þ ¼ X P ðt Þ  A : D ð16Þ  Pass these agents to the PSO technique as initial points.
! ! !  Return the modified positions back to the GWO.
where A and C represent coefficient vectors. X P is the location of  Repeat these steps until the stopping criteria is reached.
!
the victim. X is the location of the wolf and ‘t’ indicates the current
! ! The flowchart of the introduced optimization technique is
iteration. The determination of the vectors A and C can be
expressed as in (17) and (18): shown in Fig. 1.

! !! ! 3.4. Genetic algorithm optimization technique

A ¼ 2: a : r 1 : a ð17Þ

! Genetic algorithm uses the natural evolution and survival of the

!
C ¼ 2: r 2 ð18Þ fittest method with the genetic reproduction for quick search and
! optimization. The most important aspect of GA is that, it can do the
where a values decrease linearly from 2 to 0 through iterations and
! ! search to find the best among the possible answers in the interval
r 1 , r 2 are random vectors between [0, 1]. that was chosen by the user. It can find the optimal answer after
several iterative calculations, according to the survival of the fittest
3.2.3. Hunting process: law. Search process consists of artificial mutation, crossover and
The hunting process is often directed by the alphas, betas and selection operations [46]. The flowchart of GA is shown in Fig. 2
deltas as they gain more experience in determining the position [47].
of the victim. The remaining exploration agents have to follow
the optimal agent’s location and modify their locations accord-
4. Simulation results
ingly. The adjustment of the agents’ locations is expressed as
shown below [45]:
4.1. Systems tested in this study
8! ! ! ! 9
> 
>
> D a ¼  C 1: X a  X  >
>
>
> > There are three standard test systems tested in this research.
<! ! ! ! =
 These systems are the IEEE 14-bus, 30-bus and 118-bus test
D b ¼  C 2: X b  X  ð19Þ
>
> ! ! >
>
>
:!
>  ! >
>
D d ¼  C 3: X d  X  ;
Table 3
Fitness function solution of
8! ! ! !  9
>  case 2 for the 14-bus system.
>
> X 1 ¼  X a  A 1 :ð D a Þ >
>
>
> ! >
<!
 ! !  = Method VD (p.u)
X 2 ¼  X b  A 2 :ð D b Þ ð20Þ
>
> ! >
> GWO-PSO 0.2133
>
:!
>  ! !  > >
X ¼  X  A :ð D Þ ;
PSO 0.2144
3 d 3 db GA 0.2133
HHO 0.2133
! ! ! SFO 0.2133
! X1 þ X2 þ X3 BSO4 0.3
X ðt þ 1Þ ¼ ð21Þ MGBTLBO 0.248
3
 M.A.M. Shaheen et al. / Ain Shams Engineering Journal 12 (2021) 621–630 625

Table 4
Control variables of ORPD for the IEEE 14-bus system in Case 2.

Method
GWO-PSO SFO HHO GA PSO
Control variables V1 1.06 0.94 0.94 0.94 1.06
V2 0.982 0.94 0.94 0.94 1.027
V3 1.0331 1.06 0.94 0.94 1.06
V6 1.0167 0.94 0.94 1.06 0.94
V8 1.0225 1.06 0.94 1.06 1.0098
Q9 0.2 0.2 0.2 0.2 0.2
Q14 0.05 0.05 0.05 0.05 0.05
T4-7 1.1 1.1 1.1 1.1 1.1
T4-9 1.1 1.1 1.1 1.1 1.1
T5-6 0.9 0.9 0.9 0.9 0.9

systems. The main target from selecting these test systems is the
investigation of the ability of the introduced hybrid GWO-PSO
optimization technique to compete other existing optimization
techniques when dealing with the ORPD. The research study begins
with the standard IEEE 14-bus test system. It includes five PV
buses, eleven loads, twenty transmission lines, three tap changers,
and two inserted VAR sources which are shunt capacitors. Regard-
ing the next tested system, it is the IEEE 30- bus standard test sys-
tem. It includes six PV buses, twenty-one loads, forty-one
transmission lines, four tap changers, and three shunt capacitors.
Meanwhile the last standard system tested during this study is
the IEEE 118-bus test system. It includes fifty-four PV buses,
ninety-nine loads, one hundred eighty-six transmission lines, nine
tap changers, and finally, fourteen shunt capacitors. The introduced
hybrid GWO-PSO optimization technique is tested on the above
mentioned test networks for understanding the ORPD Problem.
Every test system is investigated from two aspects, named here
as case1 and case 2. The first case is to minimize reactive power
loss, and the second one is to minimize the voltage deviation. Pen-
alty functions are added with highly valued integers as penalty fac- Fig. 4. Convergence curves for Case 2 of IEEE 14-bus test system.

tors to the fitness functions to enforce the operational constraints

such as bus voltage constraints and power flow limit constraints of
the transmission lines to be kept within their acceptable physical
and two shunt capacitors at buses 9 and 14. This data regarding
limits.
the test system can be found in MATPOWER toolbox [48]. Table 1
summarizes the values of the fitness function for the IEEE 14-
4.2. Results of test systems bus system in case 1 using the hybrid GWO-PSO optimization
technique compared with other optimization techniques. Also,
4.2.1. Results of the IEEE 14-Bus System: Table 2 obtains the values of the adjusted control variables of
The 14-bus system includes five generating units at buses 1, 2, the 14-bus system in the first case when tuned by the hybrid
3, 6, and 8; twenty branches; three load tap changing transformers, GWO-PSO. These adjusted control variables are compared with
those obtained by other ones stated in literature such as PSO,
MGBTLBO [49], HHO [50] and BSO. Moreover, in case 2 which
is the case of voltage deviation minimization, when comparing
GWO-PSO optimization technique with other well-established
optimization techniques for handling the ORPD problem, it
reached minimum voltage deviation, and this is proved and

Table 5
Fitness function solution of
case 1 for the 30-bus system.

method Ploss (MW)

PSOGWO 5.09037
PSO 5.1041
GA 5.0977
HHO 5.1435
SFO 5.1072
FA 5.64
ABC 5.79
BFOA 10.01
Fig. 3. Convergence curves for Case 1 of IEEE 14-bus test system.
626 M.A.M. Shaheen et al. / Ain Shams Engineering Journal 12 (2021) 621–630

Table 6
Control variables of ORPD of the IEEE 30-bus system in Case 1.

Method
GWO-PSO SFO HHO GA PSO
Control variables V1 0.9615 0.9983 0.986 1.0228 0.94
V2 1.0020 0.9798 0.94 1.0181 1.06
V5 0.9437 0.9887 0.94 0.9859 1.06
V8 0.9623 1.0199 0.9649 0.9858 0.9873
V11 0.9476 0.9801 0.94 0.9859 1.06
V13 1.0464 1.0038 0.94 0.9815 0.9904
Q10 0.4032 2.7144 4.7898 2.0124 0
Q12 5.8091 12.2275 0 6.1953 0
Q15 3.4241 2.8415 2.7553 3.4021 2.5088
Q17 2.7263 2.0519 0 1.6718 3.0836
Q20 0.1675 3.0805 0.2897 2.4060 2.0175
Q21 7.2091 3.7349 4.4506 4.0786 0
Q23 3.4740 0.5728 2.7215 2.9779 1.3776
Q24 1.1119 5.9756 0 1.0617 0
Q29 1.5752 1.0771 1.4104 3.1220 1.1811
T11 0.9746 0.9916 0.9526 0.9996 0.9762
T12 1.0105 0.9939 1.0723 0.9715 0.9888
T15 0.9776 0.9961 1.0017 0.9794 0.9693
T36 0.9392 0.9396 0.9504 0.9477 0.9399

Table 7
Fitness function solution of
the IEEE 14-bus test system is efficient and competitive. The
case 2 for the 30-bus system. value of power losses in case 1 besides the value of voltage devi-
ation obtained by the GWO-PSO optimization technique are the
Method VD (p.u)
best through all over the other studied methods.
PSOGWO 0.27800
PSO 0.2816
GA 0.3732
SFO 0.3983 4.2.2. Results of the IEEE 30-Bus System:
GSA 0.4672 The 30-bus system includes six generators, forty-one branches,
four load tap changing transformers, and nine shunt capacitors.
This data of the test system can be found in [48]. At 100 MVA base,
the active power demand is 2.834p.u [51]. In this test system, there
are nineteen control variables. They are: generating units’ voltages
summarized in Tables 3 and 4. This comparative illustration of at buses 1, 2, 5, 8, 11 and 13; transformers between the buses 6–9,
the results assures that the hybrid GWO-PSO can get the best 6–10, 4–12, and 28–27; in addition to the reactive power from the
solution for the objectives to be optimized. The convergence shunt connected capacitors at buses 10, 12, 15, 17, 20, 21, 23, 24
curves of the introduced technique and the PSO algorithm for and 29. Table 5 summarizes the values of the fitness function for
case 1 and case 2 through 2000 iterations applied to the 14- the IEEE 30-bus system in case 1 using the hybrid GWO-PSO opti-
bus system are shown in Figs. 3 and 4. It can be observed that mization technique compared with other optimization techniques.
the performance of the introduced technique when applied to Also, Table 6 obtains the values of the adjusted control variables of

Table 8
Control variables of ORPD of the IEEE 30-bus system in Case 2.

Method
GWO-PSO SFO GA PSO
Control variables V1 1.0379 1.0131 0.94 1.06
V2 1.0167 1.0117 1.06 0.94
V5 1.0577 0.9932 1.06 0.9948
V8 0.9675 0.94 1.06 1.0443
V11 1.0324 0.9876 0.9784 0.9415
V13 0.9936 0.94 0.9557 1.06
Q10 1.7404 0 0 0
Q12 0.4043 0 7.3591 0
Q15 1.3995 0 6 0
Q17 1.3873 2.7875 0 0
Q20 1.3738 2.3701 1.1450 0
Q21 1.7700 10 0 0
Q23 3.4364 2.4401 0.8939 0
Q24 0.7805 0 0 0
Q29 2.9774 6 3.7102 1.5779
T11 1.0093 1.0084 1.0509 1.1
T12 1.1 1.1 0.9553 0.9
T15 1.0999 1.0422 1.1 1.1
T36 0.9900 1.0211 1.0029 0.9893
 M.A.M. Shaheen et al. / Ain Shams Engineering Journal 12 (2021) 621–630 627

ABC, BFOA, HHO and PSO. Meanwhile, in case 2, when compar-

ing GWO-PSO optimization technique with other well-
established optimization techniques for handling the ORPD
problem, it reached minimum voltage deviation, and this is
proved and summarized in Table 7 and 8. This comparative
illustration of the results assures that the hybrid GWO-PSO
can get the best solution for the objective functions. To guaran-
tee the dependability of the introduced GWO-PSO optimization
technique for handling the ORPD problem, several trials were
performed for both case 1 and case 2. The convergence curves
of the introduced algorithm and the PSO algorithm for case 1
and case 2 through 2000 iterations applied to the 30-bus system
are shown in Figs. 5 and 6. It can be observed that the solutions
for the IEEE 30-bus test system fitness functions are efficient
and competitive. The value of power losses in case 1 besides
the value of voltage deviation obtained by the GWO-PSO opti-
mization technique are the best through all over the other stud-
ied methods.
Fig. 5. Convergence curves for Case 1 of IEEE 30-bus test system.
4.2.3. Results of the IEEE 118-Bus System:
The third system tested in this study is the 118-bus system.
It’s considered as a large-scale network. The data of this sys-
tem are stated in [40]. The active load demand is 4242 MW.
In this test system, there are seventy-seven control variables.
The introduced hybrid GWO-PSO optimization technique is
selected to get the minimum active power losses and to get
the minimum voltage deviation. Table 9 summarizes the val-
ues of the fitness function for the IEEE 118-bus system in case
1 using the hybrid GWO-PSO optimization technique com-
pared with other optimization techniques. Also, Table 10
obtains the values of the adjusted control variables of the
118-bus system in the first case when tuned by the hybrid
GWO-PSO. These adjusted control variables are compared with
those obtained by other ones stated in literature such as FA,
ABC, and PSO. Meanwhile, in case 2, when comparing GWO-
PSO optimization technique with other well-established opti-
mization techniques for handling the ORPD problem, it
reached minimum voltage deviation, and this is proved and
Fig. 6. Convergence curves for Case 2 of IEEE 30-bus test system. summarized in Table 10 and 11. This comparative illustration
of the results assures that the hybrid GWO-PSO can get the
best solution for the objective functions. The robustness of
Table 9 the GWO-PSO optimization method is verified by performing
Fitness function solution of many runs to assure the best selection of the control parame-
case 1 for the 118-bus system. ters. The convergence curves of the introduced algorithm and
method Ploss (MW) the PSO algorithm for each fitness function through a maxi-
mum number of iterations, which is set 2000 iteration applied
PSOGWO 131.8458
PSO 131.8975 to the 118-bus system, are shown in Figs. 7 and 8. The plot
HHO 132.0394 clarifies that the results of the IEEE 118-bus test system fitness
ABC 136.99 functions are efficient and competitive. The value of power
FA 135.42 losses in case 1 besides the value of voltage deviation obtained
HFA 134.24
by the GWO-PSO optimization technique are the best through
all over the other studied methods. (Table 12) obtains the val-
ues of the adjusted control variables.
Table 10
Fitness function solution of case
5. Conclusions
2 for the 118-bus system.

method VD (p.u) This article has introduced a novel application of the hybrid
PSOGWO 1.26654835 GWO-PSO optimization method to solve the ORPD problem in
PSO 1.2738 the field of electric power networks. The introduced optimiza-
HHO 1.3592
tion method is applied to three standard test networks. They
are IEEE 14-, 30-, and 118-bus test networks. The fitness of
the introduced optimization method is verified in case of mini-
the 30-bus system in the first case when tuned by the hybrid mization of the real power loss and the voltage level. A compar-
GWO-PSO. These adjusted control variables are compared with ison between the GWO-PSO and other famous optimization
those obtained by other ones stated in literature such as FA, algorithms is also included in this article. The application of
628 M.A.M. Shaheen et al. / Ain Shams Engineering Journal 12 (2021) 621–630

Table 11
Control variables of ORPD of the IEEE 118-bus system in Case 1.

Method Method
GWO-PSO HHO PSO GWO-PSO HHO PSO
Control variables V1 0.9735 0.94 1.06 V87 1.0583 0.94 0.9642
V4 0.9697 0.94 0.94 V89 0.9594 0.94 1.0026
V6 0.9584 0.94 0.9514 V90 1.0181 0.94 0.9905
V8 0.9817 0.94 0.9820 V91 0.94 0.94 1.0015
V10 1.0412 0.94 0.9522 V92 1.0203 0.94 1.06
V12 0.9539 0.94 1.0564 V99 0.9575 0.94 0.9938
V15 0.9491 0.94 0.9560 V100 1.0336 0.94 1.0264
V18 1.0030 0.94 0.94 V103 1.06 0.94 0.94
V19 1.0109 0.94 0.94 V104 1.0246 0.94 1.06
V24 0.9408 0.94 0.94 V105 0.9754 0.94 1.06
V25 1.0252 0.94 0.9884 V107 0.94 0.94 1.0086
V26 1.0168 0.94 0.9930 V110 0.9441 0.94 0.9914
V27 1.06 0.94 0.9670 V111 0.9896 0.94 0.9430
V31 0.9472 0.94 0.94 V112 1.0313 0.94 0.9741
V32 0.9573 0.94 1.0030 V113 0.94 0.94 0.9615
V34 1.0431 0.94 0.94 V116 0.9941 0.94 0.94
V36 0.94 0.94 1.0078 QC5 0.5050 0.9986 20
V40 0.9899 0.94 0.9739 QC34 18.6293 1.0144 9.4939
V42 0.9414 0.94 0.9450 QC37 19.5920 1.0061 20
V46 1.0071 0.94 1.0392 QC44 11.8643 1.0152 9.0616
V49 1.0521 0.94 0.9806 QC45 19.9199 0.9995 20
V54 0.9690 0.94 1.06 QC46 0.8703 1.0148 19.9984
V55 0.9609 0.94 0.9736 QC48 2.0485 1.0083 3.2423
V56 0.9503 0.94 0.9715 QC74 4.3115 0.9945 17.1856
V59 0.9675 0.94 1.0600 QC79 5.5856 1.0017 3.9148
V61 0.94 0.94 0.9482 QC82 6.8463 0.9931 19.9982
V62 0.9492 0.94 1.06 QC83 0.6722 1.0020 20
V65 1.06 0.94 0.9953 QC105 8.7964 1.0153 18.9893
V66 0.9664 0.94 0.94 QC107 8.7415 1.0141 20
V69 0.9763 0.94 1.0330 QC110 10.5917 1.0010 20
V70 1.0310 0.94 0.9952 T8 0.9952 0.9 1.0059
V72 0.9469 0.94 0.9498 T32 1.1 0.9 1.1
V73 0.9401 0.94 0.94 T36 1.0100 0.9 1.0110
V74 1.0311 0.94 0.94 T51 1.0056 0.9 1.0110
V76 0.9436 0.94 0.94 T93 1.0004 0.9 1.0001
V77 0.9825 0.94 0.9935 T95 1.0153 0.9 1.0155
V80 0.9489 0.94 0.94 T102 1.0003 0.9 1.1
V85 1.0364 0.94 0.9887 T107 0.9212 0.9 0.9243
T127 0.9593 0.9 0.9589

Fig. 8. Convergence curves for Case 2 of IEEE 118-bus test system.

Fig. 7. convergence curves for Case 1 of IEEE 118-bus test system.

the ability to get a better acceptable solution when handling the

the GWO-PSO to solve the ORPD problem has resulted in a ORPD problem for different test networks. From the simulation
reduction in the power losses by (0.06 – 49)% for the systems results obtained, it’s recommended for the future scope to apply
under study. Also, the voltage deviation reduced by (7–40)% as the GWO-PSO technique to solve more problems of wide scale
a result of applying the GWO-PSO on the different test systems. power networks and with inclusion of renewable energy
As seen in the simulation results, the technique introduced has sources and microgrids.
 M.A.M. Shaheen et al. / Ain Shams Engineering Journal 12 (2021) 621–630 629

Table 12
control variables of ORPD of IEEE 118-bus system in Case 2.

Method Method
GWO-PSO HHO PSO GWO-PSO HHO PSO
Control variables V1 0.9400 0.94 0.9658 V87 1.0562 0.94 1.0594
V4 0.9514 0.94 0.9875 V89 1.0273 0.94 0.9476
V6 0.9526 0.94 1.0490 V90 1.0527 0.94 1.0533
V8 1.0575 0.94 1.0592 V91 1.06 0.94 0.9400
V10 0.9830 0.94 1.0084 V92 1.0599 0.94 0.9704
V12 1.0166 0.94 1.0336 V99 0.9770 0.94 0.9818
V15 0.9580 0.94 0.9730 V100 0.9400 0.94 0.9811
V18 0.9400 0.94 1.0159 V103 1.01611 0.94 0.9611
V19 0.9776 0.94 1.0024 V104 0.9594 0.94 1.0403
V24 1.0458 0.94 1.0092 V105 0.9400 0.94 0.9713
V25 0.9400 0.94 1.0549 V107 1.01358 0.94 1.0599
V26 1.0589 0.94 0.9400 V110 1.0600 0.94 0.9983
V27 0.96975 0.94 0.9924 V111 1.0112 0.94 0.9806
V31 0.9412 0.94 0.953 V112 1.0175 0.94 1.0008
V32 0.9993 0.94 1.0600 V113 1.02404 0.94 1.0379
V34 0.9444 0.94 1.0323 V116 0.9817 0.94 0.9903
V36 1.04346 0.94 0.9965 QC5 14.4362 4.1622 20
V40 0.9580 0.94 0.9726 QC34 1.0805 1.9006 2.6951
V42 1.0386 0.94 1.0600 QC37 19.9290 1.3771 5.0154
V46 1.0600 0.94 1.0236 QC44 5.8244 0.8092 3.2958
V49 0.9877 0.94 0.9578 QC45 15.0217 1.7274 6.9563
V54 1.0600 0.94 1.0229 QC46 6.8349 0.9128 17.9187
V55 0.9400 0.94 0.9924 QC48 19.9776 2.5031 19.7170
V56 1.0600 0.94 0.9400 QC74 13.4657 0.7128 17.9834
V59 0.9400 0.94 1.0306 QC79 19.6471 4.7328 13.9687
V61 0.96494 0.94 1.0508 QC82 0.1907 0.9426 8.1431
V62 1.0600 0.94 0.9812 QC83 18.8268 1.0836 0.5242
V65 0.94241 0.94 1.0417 QC105 16.6707 0.9987 5.1734
V66 0.9624 0.94 0.981 QC107 17.5799 0.9977 18.2017
V69 0.9507 0.94 1.0552 QC110 19.8260 1.0799 5.8121
V70 1.0600 0.94 1.0199 T8 0.9959 0.9 0.9840
V72 1.0579 0.94 0.9939 T32 1.1 0.9 1.0284
V73 0.9467 0.94 0.9863 T36 0.9739 0.9 0.9738
V74 1.0509 0.94 0.9731 T51 1.0028 0.9 1.0037
V76 1.0188 0.94 1.0083 T93 1.0166 0.9 1.0187
V77 1.0182 0.94 1.0477 T95 0.9998 0.9 0.9956
V80 0.9405 0.94 0.9905 T102 1.1000 0.9 1.039
V85 1.0600 0.94 1.0011 T107 0.9004 0.9 1.0998
T127 0.9463 0.9 0.9259

References [13] Shaheen MAM, Hasanien HM, Mekhamer SF, Talaat HEA. Optimal Power Flow
of Power Systems Including Distributed Generation Units Using Sunflower
Optimization Algorithm. IEEE Access 2019;7:109289–300.
[1] Shaheen AM, Farrag SM, El-Sehiemy RA. MOPF solution methodology. IET
[14] Qais Mohammed H, Hasanien Hany M, Alghuwainem Saad. Identification of
Gener., Transmiss. Distrib. 2017;11(2):570–81.
electrical parameters for three-diode photovoltaic model using analytical and
[2] Sakr WS, El-Sehiemy RA, Azmy AM. ‘‘Adaptive differential evolution algorithm
sunflower optimization algorithm. Appl Energy September 2019;250:109–17.
for efficient reactive power management”. Appl. Soft Comput.
[15] M. A. M. Shaheen, S. F. Mekhamer, H. M. Hasanien and H. E. A. Talaat, ‘‘Optimal
2017;53:336–51.
Power Flow of Power Systems Using Hybrid Firefly and Particle Swarm
[3] H. Singh and L. Srivastava, Modified differential evolution algorithm for multi-
Optimization Technique,” 2019 21st International Middle East Power Systems
objective VAR management, Int. J. Electr. Power Energy Syst., vol. 55, pp.
Conference (MEPCON), Cairo, Egypt, 2019, pp. 232-237.
731_740, Feb. 2014.
[16] Duman S, Sonmez Y, Guvenc U, Yorukeren N. Optimal reactive power dispatch
[4] Mehdinejad M. Solution of optimal reactive power dispatch of power systems
using a gravitational search algorithm. IET Gener. Trans- miss. Distrib. Jun.
using hybrid particle swarm optimization and imperialist competitive
2012;6(6):563–76.
algorithms. Int. J. Electr. Power Energy Syst. 2016;83:104–16.
[17] K. Lenin, B. R. Reddy, and M. S. Kalavathi, ‘‘Brainstorm optimization algorithm
[5] A. M. Shaheen, R. A. El-Sehiemy, and S. M. Farrag, A novel framework for power
for solving optimal reactive power dispatch problem,” Int. J. Res. Electron.
loss minimization by modified wind driven optimization algorithm, Proc. Int.
Commun. Technol., vol. 1, no. 3, 2014.
Conf. Innov. Trends Comput. Eng. (ITCE), 2018, pp. 344_349.
[18] El-Fergany Attia A, Hasanien Hany M. Tree-Seed Algorithm for Solving Optimal
[6] Polprasert J, Ongsakul W, Dieu VN. Optimal reactive power dispatch using
Power Flow Problem in Large-Scale Power Systems Incorporating Validations
improved pseudo-gradient search particle swarm optimization. Electr. Power
and Comparisons. Appl. Soft Comput. J. 2017;64:307–16.
Compon. Syst. 2016;44(5):518–32.
[19] El-Fergany Attia A, Hasanien Hany M. Salp swarm optimizer to solve optimal
[7] Lee KY, Yang FF. Optimal reactive power planning using evolutionary
power flow comprising voltage stability analysis. Neural Comput. Appl. 2019.
algorithms: A comparative study for evolutionary programming,
[20] Hany M.Hasaniena, Attia A.El-Fergany, ‘‘Salp swarm algorithm-based optimal
evolutionary strategy, genetic algorithm, and linear programming. IEEE
load frequency control of hybrid renewable power systems with
Trans. Power Syst. Feb. 1998;13(1):101–8.
communication delay and excitation cross-coupling effect,” Electr. Power
[8] Zhu JZ, Xiong XF. Optimal reactive power control using modified interior point
Syst. Res., vol. 176, ISSN 0378-7796, Nov. 2019.
method. Electr. Power Syst. Res. 2003;66(2):187–92.
[21] Qais Mohammed H, Hasanien Hany M, Alghuwainem Saad. Enhanced salp
[9] Quintana VH, Santos-Nieto M. Reactive-power dispatch by successive
swarm algorithm: Application to variable speed wind generators. Eng. Appl.
quadratic programming. IEEE Trans. Energy Convers. Sep. 1989;4(3):425–35.
Artif. Intell. April 2019;80:82–96.
[10] Jan R-M, Chen N. Application of the fast Newton-Raphson economic dispatch
[22] Abdel-Fattah Attia, Ragab A. El Sehiemy, Hany M.Hasanien, ‘‘Optimal power
and reactive power/voltage dispatch by sensitivity factors to optimal power
flow solution in power systems using a novel Sine-Cosine algorithm,” Int. J.
flow. IEEE Trans. Energy Convers. Jun. 1995;10(2):293–301.
Electr. Power Energy Syst. vol. 99, pp. 331-343, July 2018.
[11] Wu QH, Cao YJ, Wen JY. Optimal reactive power dispatch using an adaptive
[23] Qais Mohammed H, Hasanien Hany M, Saad Alghuwainem, Nouhb Adnan S.
genetic algorithm. Int. J. Electr. Power Energy Syst. 1998;20(8):563–9.
Coyote optimization algorithm for parameters extraction of three-diode
[12] El-Ela AA. Optimal reactive power dispatch using ant colony optimization
photovoltaic models of photovoltaic modules. Energy 2019;187:15.
algorithm. Electr. Eng. 2011;93(2):103–16.
630 M.A.M. Shaheen et al. / Ain Shams Engineering Journal 12 (2021) 621–630

[24] Dai C, Chen W, Zhu Y, Zhang X. Seeker optimization algorithm for optimal [37] Qais Mohammed H, Hasanien Hany M, Alghuwainem Saad. Transient search
reactive power dispatch. IEEE Trans. Power Syst. 2009;24(3):1218–31. optimization for electrical parameters estimation of photovoltaic module based
[25] Amjady N, Fatemi H, Zareipour H. Solution of optimal power flow subject to on datasheet values. Energy Convers. Manage. June 2020;214(112904):1–9.
security constraints by a new improved bacterial foraging method. IEEE Trans. [38] R. A. El Sehiemy, A. A. Abou El ELA, and A. M. Shaheen, ‘‘A multi-objective
Power Syst. Aug. 2012;27(3):1311–23. fuzzy-based procedure for reactive power-based preventive emergency
[26] Sulaiman MH. Loss minimisation by optimal reactive power dispatch using strategy,” Int. J. Eng. Res. Africa, vol. 13, p. 91, Dec. 2014.
cuckoo search algorithm. Proc. 3rd IET Int. Conf. Clean Energy Technol. (CEAT), [39] Ghasemi M, Ghanbarian MM, Ghavidel S, Rahmani S, Moghaddam EM.
2014. Modified teaching learning algorithm and double differential evolution
[27] Raha SB, Som T, Mandal KK, Chakraborty N. Cuckoo search algorithm based algorithm for optimal reactive power dispatch problem: A comparative
optimal reactive power dispatch. In: Proc. Int. Conf. Control, Instrum., Energy study. Inf. Sci. 2014;278:231–49.
Commun. (CIEC). p. 412–6. [40] Rajan A, Malakar T. Optimal reactive power dispatch using hybrid Nelder-
[28] Kalaam Rahila N, Muyeen SM, Al-Durra Ahmed, Hany M. Hasanien, and Khaled Mead simplex based firefly algorithm. Int. J. Elect. Power Energy Syst. Mar.
Al-Wahedi ‘‘Optimisation of controller parameters for grid-tied photovoltaic 2015;66:9–24.
system at faulty network using artificial neural network-based cuckoo search [41] Chopra Nitish et al. Hybrid GWO-PSO Algorithm for Solving Convex Economic
algorithm”. IET Renew. Power Gener. 2017;11(12):1517–26. Load Dispatch Problem. Int. J. Res. Adv. Technol. 2016.
[29] Omnia S. Elazab, Hany M. Hasanien, Ibrahim Alsaidan, Almoataz Y. Abdelaziz, [42] Hasanien Hany M. Particle Swarm Design Optimization of Transverse Flux
and S.M.Muyeen, ‘‘Parameter Estimation of Three Diode Photovoltaic Model Linear Motor for Weight Reduction and Improvement of Thrust Force. IEEE
Using Grasshopper Optimization Algorithm,” Energies, vol. 13, no. 2, 20 Trans. Ind. Electron. 2011;58(9):4048–56.
January 2020. [43] Mirjalili Seyedali, Mirjalili Seyed Mohammad, Lewis Andrew. Grey wolf
[30] Liang R-H, Wang J-C, Chen Y-T, Tseng W-T. An enhanced firefly algorithm to optimization. Adv. Eng. Soft. 2014;69:46–61.
multi-objective optimal active/reactive power dispatch with uncertainties [44] Mahmoud Hassan Y, Hasanien Hany M, Besheer Ahmed H, Abdelaziz Almoataz
consideration. Int. J. Elect. Power Energy Syst. Jan. 2015;64:1088–97. Y. Hybrid cuckoo search and grey wolf optimiser-based optimal control
[31] Devaraj D, Roselyn JP. Genetic algorithm based reactive power dispatch for strategy for performance enhancement of HVDC-based offshore wind farms.
voltage stability improvement. Int. J. Electr. Power Energy Syst. 2010;32 IET Gener. Transm. Distrib. 2020;14(10):1902–11.
(10):1151–6. [45] El-Fergany AA, Hasanien HM. Single and multi-objective optimal power flow
[32] Badar AQ, Umre BS, Junghare AS. Reactive power control using dynamic using grey wolf optimizer and differential evolution algorithms. Electr. Power
particle swarm Optimization for real power loss minimization. Int. J. Electr. Compono. Syst. 2015;43(13):1548–59.
Power Energy Syst. 2012;41(1):133–6. [46] Bakhshi Hamid, Dehghani Afshin, Jafaripanah Sahar. Using the Genetic
[33] Sakr WS. Efficient reactive power management via enhanced differential Algorithm based on the Riedel Equation to Predict the Vapor Pressure of
evolution algorithm with adaptive penalty factor. Int. J. Power Eng. Energy Organic Compounds. Int. J. Eng. 2018;31(6):863–9.
2015;6(2):542–50. [47] Hermawanto, Denny. ‘‘Genetic algorithm for solving simple mathematical
[34] Lenin K, Reddy BR, Suryakalavathi M. Hybrid Tabu search simulated annealing equality problem.” arXiv preprint arXiv:1308.4675 (2013).
method to solve optimal reactive power problem. Int. J. Electr. Power Energy [48] MATPOWER. [Online]. Available: http://www.pserc.cornell.edu// matpower/.
Syst. 2016;82:87–91. [49] Ghasemi M. Solving optimal reactive power dispatch problem using a novel
[35] Farzana Islam, Hany M. Hasanien, Ahmed Al-Durra, and S.M.Muyeen, ‘‘A New teaching-learning-based optimization algorithm. Eng. Appl. Artif. Intell.
Control Strategy for Smoothing of Wind Farm Output using Short-Term Ahead 2015;39:100–8.
Wind Speed Prediction and Flywheel Energy Storage System”, in the [50] Qais Mohammed H, Hasanien Hany M, Alghuwainem Saad. Parameters
proceeding of American Control Conference, ACC, 27-29June 2012, Montreal, extraction of three-diode photovoltaic model using computation and Harris
Canada. Hawks optimization. Energy 2020;195.
[36] Gafar MG, El-Sehiemy RA, Hasanien HM. A Novel Hybrid Fuzzy-JAYA [51] Bhattacharya A, Chattopadhaya PK. Solution of optimal reactive power flow
Optimization Algorithm for Efficient ORPD Solution. IEEE Access using biogeography-based optimization. Int. J. Electr. Elec- tron. Eng. 2010;4
2019;7:182078–88. (8):568–76.