Received January 3, 2020, accepted January 14, 2020, date of publication January 21, 2020, date of current version

February 14, 2020.

Digital Object Identifier 10.1109/ACCESS.2020.2968462

A Novel Crow Search Algorithm Auto-Drive PSO

for Optimal Allocation and Sizing of Renewable
Distributed Generation
HASSAN M. H. FARH 1 , ABDULLAH M. AL-SHAALAN 1, ALI MOHAMED ELTAMALY 2,3 ,

AND ABDULLRAHMAN A. Al-SHAMMA’A 1,4

1 Electrical
Engineering Department, College of Engineering, King Saud University, Riyadh 11421, Saudi Arabia
2 SustainableEnergy Technologies Center, King Saud University, Riyadh 11421, Saudi Arabia
3 Electrical
Engineering Department, Mansoura University, Mansoura 35516, Egypt
4 Department of Mechatronics Engineering, College of Engineering, Taiz University, Taizz 6803, Yemen

Corresponding author: Hassan M. H. Farh (hfarh1 @ ksu.edu.sa)

This work was supported by the Deanship of Scientific Research at King Saud University, Riyadh, through the initiative of DSR Graduate
Students Research Support.

ABSTRACT The penetration of renewable distributed generations (RDGs) into traditional distribution
systems (TDSs) remedies many of its deficiencies and shortcomings. Also, it provides mutual technical,
economic and environmental benefits for both electricity companies and their customers as well. With a 25%
load increase for the standard IEEE 30-bus system, buses 19, 26 and 30 have the lowest voltage magnitudes
among all buses. Therefore, these weak buses are selected initially to allocate RDGs. Three cases, namely,
one RDG allocated, two RDGs allocated and three RDGs allocated, of RDGs insertion are covered. A novel
crow search algorithm auto-drive particle swarm optimization (CSA-PSO) technique is proposed for the first
time in this study to specify the optimal allocation, sizing, and number of RDGs based on the total cost and
power losses minimization objectives. A new reduction percent formula is used to estimate the reduction
in total cost and the total power losses. These will help us to discern between the best cases based on total
cost minimization and those based on total power losses minimization to pick up the best among all best
cases. In brief, RDGs allocated on buses 19 and 30 is the best among all cases based on total cost reduction
and total power losses reduction. Therefore, buses 19 and 30 are recommended to allocate a wind farm and
a solar photovoltaic, respectively based on technical and economic issues. Finally, the simulation findings
revealed the superiority of the CSA-PSO algorithm in solving the optimal power flow problem with RDGs
compared to the state-of-the-art metaheuristic techniques.

INDEX TERMS Crow search algorithm, particle swarm optimization, renewable distributed generation,
sizing, solar, allocation, photovoltaic, wind farm.

I. INTRODUCTION in distribution lines and are not appropriate for addressing

A. MOTIVATION AND INCITEMENT environmental, technical, and economic issues in the electric-
In recent decades, distributed generation (DG) has earned a ity market. The voltage instability problem may expose the
great attention from utilities with the dramatic increase in security of power system to a malfunction that may lead to a
electric demand, deregulated power systems, and congested partial or total severe outages occurrences causing blackouts
transmission power networks in the traditional distribution as have been experienced and witnessed globally in some
systems (TDSs) that are operating in central, radial and countries [1]–[3].
unidirectional manners. These TDSs are subject to voltage The insertion of DGs, especially renewable distributed
instability, high power losses as a result of the high R/X ratios generations (RDGs), into the TDSs, become compulsory to
overcome the deficiencies and shortcomings of TDSs. They
The associate editor coordinating the review of this manuscript and have many benefits over the TDSs, including technical, eco-
approving it for publication was F. R. Islam . nomic and environmental benefits to the electric utility and its

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see http://creativecommons.org/licenses/by/4.0/
VOLUME 8, 2020 27807
 H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

customers. Technical benefits include reducing transmission may be single or multi-objective functions depending on
power losses, improving voltage profile, enhancing stability, the planning process, and they may include technical or
and reliability, alleviating transmission and distribution con- economic issues under some equality, and inequality con-
gestion, and increasing the overall power system efficiency. straints. Some researches [1]–[25],[27]–[32] have focused
Whereas, economical benefits include fuel savings, reduction on technical issues like power losses (active and reactive),
of electricity price, and reduction of the generation, transmis- loadability, voltage deviations, voltage stability and emis-
sion, and distribution costs because the DG units are allocated sion of generating units while they overlooked the economic
closer to the load centers. Also, operational and maintenance issues. Whereas, other researches [21], [33]–[36] focused on
costs as well as the reserve requirements and costs will be economic objectives like fuel cost, active, and reactive power
decreased [4]–[7]. In addition, environmental benefits, such costs and optimal DGs investment cost while they neglected
as greenhouse gas emissions reduction, pollution-free envi- the technical issues.
ronment creation, and CO2 emissions reduction will be accu-
mulated [8]–[10]. These multiple benefits cannot be achieved C. CONTRIBUTION AND PAPER ORGANIZATION
without the appropriate allocation and sizing of the DGs On the basis of the previous literature review, this paper
in power system networks. Their improper allocation and is counted as the first study that proposes a novel crow
sizing may increase the power losses that could lead to the search algorithm auto-drive PSO (CSA-PSO) to determine
unsatisfactory performance of the electricity infrastructure the optimal allocation, sizing, and number of RDGs based
(low quality, instability, and unreliability) [11], [12]. on both total cost minimization and power losses minimiza-
tion. The proposed CSA-PSO was applied to the standard
B. LITERATURE REVIEW IEEE 30-bus test system for solving the OPF problem with
Numerous optimization techniques are developed, adopted RDGs. The objective of hybridization is not only improve
and utilized to solve the optimal power flow (OPF) problem the performance of the traditional PSO but also to do two
related to the RDGs. These optimization techniques can be sub-roles or two-sub problems at the same time. In addition,
categorized into analytical, conventional, and soft comput- hybrid CSA-PSO integrates the advantages of the CSA and
ing optimization techniques. Each optimization technique PSO in order to follow the OPF global solution in the search-
has its own pros and cons. Although analytical techniques ing area. Each metaheuristic algorithm has its own merits
are non-iterative, accurate, efficient, easy in implementation, and task. CSA performs a function or a role (outer loop)
and consume less computational time, they are not appro- while PSO performs a different role (inner loop). CSA is the
priate for multiple DGs and multiple objectives alone. In outer metaheuristic algorithm used to automatically specify
[7], [13]–[17], an analytical formula was used to specify the optimal sizing of the RDGs (global searching of optimal
optimal allocation and sizing of a single DG and single RDGs allocation and sizing). Whereas, PSO is the inner
objective (power losses minimization) in a power distribution metaheuristic algorithm that used for the optimal power flow
system. In these studies, four different formulas were used of the power distribution system after RDGs insertion (local
to calculate the total active power losses in a power distribu- searching of OPF solution including RDGs). Therefore, it can
tion system. These formulas are called exact, branch current, be said that the proposed CSA is auto driving PSO for opti-
branch power flow, and voltage formula [7], [13]–[17]. Con- mal allocation and sizing of the RDGs. For validation and
ventional techniques include linear programming (LP), non- substantiation purposes, the results obtained with a hybrid
linear programming (NLP), dynamic programming (DP), CSA-PSO were compared with three other metaheuristic
sequential quadratic programming (SQP), ordinal optimiza- techniques, which are the phasor particle swarm optimiza-
tion (OO), optimal power flow (OPF), and continuous power tion, and a gravitational search algorithm (PPSOGSA) [34],
flow (CPF) [18]. On the other hand, soft computing tech- TLBO, and PSO [37] algorithms. Detailed comparisons and
niques are efficient in finding global solutions, and can verifications are introduced with and without RDGs pene-
be divided into artificial intelligence (AI) and bio-inspired tration based on two different objective functions; total cost
(BI) optimization techniques. The optimization techniques minimization and power losses minimization. Three cases of
based-AI include differential evolution (DE) [19], fuzzy logic RDGs insertion are covered; one RDG allocated (Case#1),
control (FLC), genetic algorithm (GA) [20]–[22] and artifi- two RDGs allocated (Case#2) and three RDGs allocated
cial neural network (ANN). Whereas, the other ones based (Case#3). A new reduction percent formula is used to evaluate
BI include particle swarm optimization (PSO) [19]–[23], the total cost reduction and the total power losses reduction.
[24], cuckoo search optimization (CSO), teach learning- These will help in discerning between all the best cases
based optimization (TLBO) [25], ant colony (AC), simu- based on the total cost minimization and those based on total
lated annealing (SA), artificial bee colony (ABC), intelligent power losses minimization in order to pick up the eventual
water drop (IWD), and the shuffled frog leaping algorithm optimal allocation, sizing, and numbers of RDGs. The rest
(SFLA) [18]–[26]. of this paper is organized as follow: Section 2 introduces
Optimal siting and sizing of the DGs depend on the objec- the OPF mathematical problem formulation with RDGs.
tive functions which are determined by the designers and Section 3 presents the improved IEEE 30-bus power distribu-
planners and need to be achieved. The objective functions tion system. In Section 4, wind farm and solar photovoltaic

27808 VOLUME 8, 2020

H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

generation systems are introduced and discussed. Section (FC−RDGs ), and the penalty of the inequality constraints.
5 the crow search auto-drive PSO algorithm for OPF solution FC−Tot can be expressed as follow:
with incorporating RDGs is explained. In the next sections,
the results are presented and discussed followed by the con- FC−Tot = FC−Th + FC−RDGs + Penalty (6)
clusions.
where
II. OPF MATHEMATICAL FORMULATION WITH RDGS
FC−Th is a quadratic function of the active power generated
The OPF is a non-linear optimization problem that is solved and can be expressed as follows [35], [36]:
through identifying the control (independent) variables based NG
X
on the minimization of a predetermined objective function FC−Th (PGi ) = ai P2Gi + bi PGi + ci ($/h) (7)
subject to some equality and inequality constraints. In gen- i=1
eral, the OPF mathematical problem can be expressed by
Eq.(1) [35], [36]. where, ai , bi , ci are the cost coefficients of the ith thermal
generation unit that generate PGi .
Fobj = Min F(x, y) (1) FC−RDGs equals the summation of both the cost of wind
Related to: (FC−W ) and photovoltaic (PV); (FC−PV ); power systems as
follow:
g(x, y) = 0 (2)
h(x, y) ≤ 0 (3) FC−RDGs = FC−W + FC−PV (8)

where FObj is the objective function required to be minimized; The direct cost function of the wind power (FC−W ) is
g(x, y) is the equality constraints; h(x, y) is the inequality a function of the scheduled/estimated power which can be
constraints; x and y are the vectors of control (independent) formulated as follows [35], [36]:
and dependent variables, respectively. The control variables
control the power flow in the network and can be formulated FC−W (PWs−j ) = gj PWs−j (9)
in a vector as follow:
where, gi is cost coefficie is the scheduled/estimated power
x = [P2 . . . PNG ; V1 . . . VNG ; TS 1 . . . TS NT ; Q1 . . . QNC ] (4) from the wind farm.
Similarly, the direct cost function of the solar PV genera-
where PNG is the generated power at all generation buses
tion system (FC−S ) is a function of the scheduled/estimated
except the slack bus, VNG is the voltage at generation buses,
power which can be formulated as follows [35], [36]:
TSNT is the transformers tap-setting, and QNC is the shunt
VAR compensators. NG, NT, and NC are the total number FC−PV (PPVs−k ) = gk PPVs−k (10)
of generators, transformers and shunt VAR compensations,
respectively. where, gk is the cost coefficient of the PV generation sys-
The state (dependent) variables describe the power system tem k, and PPVs−k is the scheduled/estimated power from the
state and can be expressed in a vector as follow: PV generation system.
The control or independent variables are self-constrained
y = [P1 ; VL1 . . . V LNL ; Q1 . . . QNG ; Sl1 . . . S lNTL ] (5)
while the independent variables related to the power system
where P1 is the generated power at the slack bus, VL is state are embedded in the objective function. The penalty for
the voltage at load buses, QNG is the reactive power of all the previous state variables mentioned in equation (6) can be
generation units, and Sl is the loading of transmission lines. expressed as follows [38]:
NTL and NL are the transmission line numbers and load bus NL
numbers, respectively. X
Penalty = λP (PG1 − PLim
G1 ) + λV
2
(VLi − VLiLim )2
i=1
A. OBJECTIVE FUNCTIONS NG NTL
In this work, two different objectives are taken into consid-
X X
+ λQ Gi ) + λS
(QGi − QLim 2
(Sli − Slimax )2
eration, namely, the total cost minimization and transmission i=1 i=1
losses minimization. The details of these objectives are intro- (11)
duced below.
where λP , λV , λQ , and λS are the penalty factors of the slack
1) TOTAL COST MINIMIZATION bus generated power, reactive power of all generation buses,
The essential objective function considered for optimal sizing voltage of load buses and transmission lines loadings respec-
of RDGs (wind and photovoltaic) in addition to the OPF tively. The penalty equals 0 if the solutions are acceptable,
problem is the total generation cost. The total generation cost and the dependent variables are between the lower and upper
(FC−Tot ) includes three main parts, which are the fuel cost of limits. Conversely, it has a value if the solutions are located
the thermal generation units (FC−Th ), the cost of the RDGs outside the predetermined range.

VOLUME 8, 2020 27809

 H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

2) TRANSMISSION POWER LOSSES MINIMIZATION

The transmission power losses (PLoss ) is one of the criti-
cal objectives that must be minimized in the OPF problem.
PLoss in a power distribution system can be expressed as
follows [23]–[35], [36]:
NBr
X
PLoss = gij [Vi2 + Vj2 − 2 Vi Vj cos(δi − δj )] (12)
ij = 1
i 6= 1

where Vi is the bus i voltage; Vj is the bus j voltage; gij is

the conductance between i and j buses; δi is the bus i voltage
angle; δj is the bus j voltage angle; NBr is the number of
branches; PLoss is the objective function, which needs to be
minimized.

B. OPF EQUALITY AND INEQUALITY CONSTRAINTS

The OPF problem includes equality and inequality power
flowOPF constraints, which must be satisfied during this opti-
mization problem. Equality constraints represent the power
FIGURE 1. Improved IEEE 30-bus test system inherited WT and PV as a
flow equations. Both active and reactive power are necessary RDGs sources.
to be equal with the power demand and power losses in the
power system networks as follow [35], [36]–[39], [40]: where equations (15), (16), and (17) are the generators
NB
X inequality constraints; Eq. (18) is the transformers inequality
PGi −PDi −Vi Vj [Gij cos(δij ) + Bij sin (δi j )] = 0 ∀i ∈ NB constraints; equation (19) is the shunt compensators inequal-
j=1 ity constraints; and equations (20) and (21) are the security
(13) inequality constraints.
NB
X
QGi −QDi −Vi Vj [Gij sin(δij ) − Bij cos (δi j )] = 0 ∀i ∈ NB III. IMPROVED IEEE 30-BUS TEST SYSTEM WITH RDGS
j=1 In this research, the IEEE 30-bus power distribution system
(14) is improved by inserting the RDGs; wind turbine and solar
photovoltaic generation units; on the weak buses (bus19 and
where bus30), as shown in Fig.1. The data for the improved IEEE-
δij = δi −δj is the voltage angles difference between i and 30 bus power distribution system used in this research are
j buses, summarized in Table 1. Three weak buses (bus19, bus26 and
NB is the buses number, bus30) that have the lowest voltage are selected for investi-
PDi and QDi are active and reactive power demands on bus gation. Three cases of RDGs insertion (one RDG allocated,
i, respectively. two RDGs allocated, and three RDGs allocated) are covered
PGi and QGi are active and reactive power generated on bus and discussed. The reason behind this is to determine the opti-
i from any generation source. mal RDGs allocation, sizes, and numbers based on the total
Gij and Bij are admittance matrix elements (conductance cost and power losses minimization objectives. This topic
and susceptance); Yij = Gij + jBij . will be discussed in detail in the simulation results section.
The inequality power flow constraints reflect the upper A novel CSA-PSO algorithm is proposed in this study to
and lower limits of the power system instruments in order to solve the OPF problem including RDGs. For the validation
guarantee the security of the power system. These inequality and substantiation purposes, the new proposed CSA-PSO is
power flow constraints are introduced in the following equa- compared with the state-of-the-art metaheuristic techniques
tions: PPSOGSA [34], TLBO, and PSO [37] algorithms.
min
VGi max
≤ VGi ≤ VGi , i = 1, ...., NG (15) A new formula reduction percent is proposed to calcu-
late the reduction percent in total cost and power losses,
Pmin
Gi ≤ Gi ,
PGi ≤ Pmax i = 1, ...., NG (16) respectively. These will help us to discern between the best
Qmin
Gi ≤ QGi ≤ QGi , i = 1, ...., NG
max
(17) cases (best one RDG allocated, best two RDGs allocated, and
TSimin ≤ TSi ≤ TSimax , i = 1, ...., NT (18) best three RDGs allocated) to pick up the best-of-the-best
RDGs cases based on both technical and economic factors.
Qmin
Ci ≤ Ci ,
QCi ≤ Qmax i = 1, ...., NC (19)
Reduction percentage refers to the minimization amount per-
VLimin ≤ VLi ≤ VLimax , i = 1, ...., NL (20) cent of the objective function (total generation cost or total
Sli ≤ Slimax , i = 1, ...., NTL (21) power losses). It equals the difference between the objective

27810 VOLUME 8, 2020

H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

TABLE 1. The summarized data of the improved IEEE-30 bus power where u is the wind speed; uin , ur , uout and Pr are the WT
system used in this study.
parameters.
A solar photovoltaic generation system can be used as a
RDG at the weak buses to cover the load increase and add
some technical, economical and environmental benefits to the
power system. It includes a PV array connected to the load bus
through a DC-DC converter and a DC-AC inverter. The PV
array contains some parallel strings. Each string has a series
of PV modules. The output power generated from the PV
array can be increased through tracking the maximum power
under uniform and partial shading conditions [44]–[47]. Also,
it depends on location characteristics, such as temperature
and irradiance. Also, it is a function of the irradiance (G) of
the location and the rated power of the PV modules (Pr ) given
by Eq. (26) [48]:
function value without RDGs; (FObj )Without−RDGs ; and the
G2
  
objective function value with RDGs; FObj )With−RDGs divided 0 < G < RC
Pr


into the base case (FObj )Without−RDGs ; as follows: PPV (G) = − RC
 GSTC  (26)
(FObj )Without−RDGs −(FObj )With−RDGs  G
Pr
 RC≤G
Reduction% = × 100 GSTC
(FObj )Without−RDGs
(22) where GSTC is the irradiance at standard test conditions
(STC) (1000W.m−2 ) Rc is a certain irradiance, and Pr is the
IV. WIND FARM AND SOLAR PHOTOVOLTAIC rated power of the PV generation unit.
GENERATION SYSTEMS
Each wind turbine (WT) has its parameters i.e. cut-in speed V. PROPOSED CROW SEARCH ALGORITHM AUTO DRIVE
(uin ), rated speed (ur ), cut-out speed (uout ) and rated or max- PSO TECHNIQUE
imum power (Pr ). Also, each location has its wind speed A crow search algorithm (CSA) represents one of the
characteristics. The wind speed characteristics in any location metaheuristic algorithms that were proposed by Askarzadeh
are the Weibull scale parameter (c) and the Weibull shape in 2016 [49]. CSA emulates the social behavior of the crow
parameter (k). Matching between the WT characteristics and birds. These intelligent birds live in groups and have small
the location characteristics produce an increase in the wind body size but large brains. Also, they can remember the
energy generated and a reduction in the energy cost. Eltamaly hiding places of their food and reassemble it even after a long
and Farh [41] proposed a criterion to select the appropriate time. Crows follow the other crows in order to observe their
WTs for each location based on which WT and location have food’s hiding places and steal it once they leave. If a crow
a maximum capacity factor (CF) and minimum energy cost feels that another one is following it, it moves to another place
[41]. The wind speed (u) distribution likes approximately the far away from the food’s hiding place in order to deceive the
Weibull distribution as follow [41], [42]: following birds. The CSA principles are summarized below:
   
k  u k−1 u k - Crows live in a group.
f (u) = exp − (k > 0, u > 0, c > 1)
c c c - Crows memorize their food’s hiding places.
(23) - Crows follow other ones to steal their foods from hiding
places.
f (u) is integrated to estimate the cumulative distribution func- - Crows protect their caches from thievery based on
tion as follows: awareness probability.
Zu    
u k The crows in the group (flock) are the candidate solutions
F(u) = f (u) du = 1 − exp − (24)
c in the population where the population contains N candidate
0 solutions. The position of each crow, i at iteration t (candidate
The output power generated from the WT generator depends solution) adjusted in between lower and upper values, can be
on both the wind speed (u) of the site and the WT parameters represented by a vector xit according to Eq.(27) [50].
as shown in Eq.(25) [37]–[43].
xit = xi1 , xi2 , xi3 , ... ... . . . , xid
 t t t t

(27)
! uin < u < uout


 0

 2
u − uin 2
PW (u) = Pr uin ≤ u ≤ ur (25) where xit is the position of each crow i at iteration t with
u 2 − u2


 r in the problem dimension d where xit ∈ [L, U] randomly, (i =
ur < u ≤ uout 1, . . . .., N ).

Pr

VOLUME 8, 2020 27811

 H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

The whole candidate solutions (X ) of the population size N

with dimension d at iteration t can be represented by Eq. (30)
 t t t

x11 x12 · · x1d
 xt t
x22 · · x2d t 
X = 21  (28)
 · · · · · 
t t
xN 1 xN 2 · · xNd t

Based on the smart behavior of the crow discussed-above, a

metaheuristic CSA is hybridized with the PSO to solve the
OPF problem inserting RDGs into the IEEE 30-bus distri-
bution system. CSA auto-drive PSO algorithm is developed
and applied for the first time in this paper. CSA is the outer
metaheuristic algorithm used to automatically specify the
optimal sizing of the RDGs (wind and PV). PSO is the inner
metaheuristic algorithm that used for the optimal power flow
of the power distribution system after RDGs insertion. The
hybrid CSA-PSO is used to specify the optimal sizing of
the RDGs based on the total cost minimization of the power
system taking into consideration the equality and inequality
power flow constraints. The flowchart of the proposed CSA
auto-drive PSO is presented in Fig. 2. Also, the logic of the
hybrid CSA-PSO is summarized in the following steps:
Step 1: Initialization of CSA-PSO parameters.
Adjust the initial values of flight length (fl), awareness
probability (AP), population size (N ) and the iterations num-
ber (tmax ). In this study, The CSA parameters are adjusted
to be fl = 2, AP = 0.1, N = 25 and tmax = 100. Whereas,
the PSO initialization parameters are acceleration coefficients
(C1 , and C2 ) are 1.89 and 2.123 respectively, inertia weight
(ω) = −0.162, N = 40 and tmax = 100.
Step 2: Position and memory of crow’s initialization.
Generate the position of crows (initial solutions) randomly;
each crows’ vector represents a candidate solution as pre-
sented in the matrix shown in Eq. (28).
FIGURE 2. Flowchart of the CSA auto-drive PSO for optimal sizing of the
The initial memory for each initial solution in the popula- RDGs.
tion is assigned after estimating its fitness function (objective
function) and is represented by a vector Mit as defined in Step 5: Position and memory update of crow i.
Eq.(29). Crow i updates its position and follows crow j in order to
know its food’s hiding place. Two major states may occur
Mit = Mi1t , Mi2t , Mi3t , . . . . . . . . . , Midt
 
(29) during updating the position as follows:
The memory M of all population (crows) at iteration t for State 1 (Pursuit): Crow j does not observe that crow i is
dimension d is initialized as shown in Eq. (30) following it. Therefore, crowi will discover the food’s hiding
 t t t
 place of crow j and the crow i will update its position based
M11 M12 · · M1d on Eq.(31) [50], [51].
 Mt M22t · · M2d t 
M = 21 (30) xit+1 = xit + r1 ∗ fl ∗ (mtj − xit )

 · · · · ·  (31)
MNt 1 MNt 2 · · MNd t
where ri is a random number in between [0, 1], and is the mtj
Step 3: Optimal power flow solution based on PSO. memory of crow j.
Solve optimal power flow problem using PSO and return State 2 (Evasion): Crow j knows that crow i is following it.
the cost of each solution (crows’ vector) to estimate the Thus, crowj will move to random position in order to mislead
total generation costs, including the RDGs cost (objective crow i.
function) using Eq. (6). Therefore, the position of crow i at iteration t+1 can be
Step 4: Objective function evaluation. updated for the two states as shown in Eq.(32) [50], [51].
Each initial solution (crows’ vector) in the population N (
t+1 x t + r1 ∗ fl ∗ (mtj − xit ) if r1 ≥ AP
is evaluated through calculating its fitness using the fitness xi = i (32)
function (objective function). Move to random position if r1 < AP

27812 VOLUME 8, 2020

H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

where xit+1 is the new position of crow i, and xit is the previous
position of the crow i.
After the position of crow i is updated, repeat step 3 (PSO
routine for optimal power flow solution) and step 4 (objective
function evaluated). Also, the memory is updated where each
crow (solution) updates its memory according to the fitness
value (objective function). If the objective function value of
the new crow’s position is better than the current memory’s
value, then it updates its memory; otherwise, the memory
will not be changed. The memory is updated based on the
following conditions:
(
t+1 x t+1 if fObj (xit+1 ) > fObj (mti )
mi = i t (33)
mi Otherwise FIGURE 3. Voltage profile at all buses for different cases; without and
with RDGs in addition to 25% load increasing.

where fObj (xit+1 ) is the current value of objective function;

fObj (mti ) is the memorized value.
Step 6: Termination criterion.
Step 5 is repeated until the iterations number (tmax ) is
ended. Therefore, the optimal sizing of the RDGs (best solu-
tion) is the best memorized position related to the best objec-
tive function value.

VI. SIMULATION RESULTS AND DISCUSSIONS

The proposed CSA auto-drive PSO was applied to the IEEE
30-bus test distribution system for solving the OPF problem
with and without RDGs. This process was carried out by
using MATLAB code V2016. To evaluate the performance of
the proposed CSA-PSO algorithm, the results obtained with FIGURE 4. Total cost (FC −Tot ) minimization without and with one RDG
hybrid CSA-PSO are compared with three other metaheuris- inserted at the weak bus 19 or 26 or 30.

tic techniques, namely, PPSOGSA [34], TLBO and PSO

[37] algorithms. These detailed comparisons were achieved PPSOGSA, TLBO, and PSO algorithms for validation pur-
and introduced with and without inserted RDGs based on poses. Table 2 shows the best simulation results for the OPF
two different objective functions; total cost minimization and solution with one RDG inserted at the weak bus 19 or 26
power losses minimization. or 30. Also, it provides comparisons of the proposed
With 25% load increase; the three weak buses that have the CSA-PSO to the PPSOGSA, TLBO and PSO to solve this
lowest voltage magnitude are bus 19, bus 26 and bus 30 as OPF problem based on the two objective functions; total
shown in Fig. 3. These weak buses are selected initially for generation cost and total power losses minimization. Some
RDGs allocation. Three cases of RDGs insertion (one RDG observations resulting from the simulation results presented
allocated, two RDGs allocated, and three RDGs allocated) in Table 2 and Figures 4 and 5 are summarized as follows:
are discussed. OPF solution with these three cases of inserted - Insertion of RDG on bus 19 is preferred to minimize the
RDGs are performed using the new proposed CSA-PSO and total generation cost, where it achieved the minimum
compared with PPSOGSA, TLBO and PSO based on total total cost of 694.22 $/h, compared with the total gen-
cost and the transmission power losses minimization objec- eration cost with RDG on bus 26, or 30. Figure 4 proves
tives. The reason behind this plan is to specify the optimal that RDGs penetration will reduce the total generation
allocation, sizing, and numbers of RDGs based on these cost compared with the one without RDGs. In addition,
two different objective functions. The detailed description, penetration of RDG on bus 19 into the test system
comparison, discussion and analysis of these three cases of makes the total generation cost the lowest compared
RDGs insertion are discussed below. with the penetration of RDG on bus 26 or bus30. Also,
Case #1: One RDG inserted at weak bus 19 or 26 or 30. it enhances the voltage stability of all buses as witnessed
This case represents the OPF solution for the IEEE 30-bus in Fig. 3.
test distribution system with one RDGs inserted at weak - Insertion of RDG on bus 30 is preferred to mini-
buses 19 or 26 or 30; separately. Optimal sizing of each mize the transmission power losses, where it achieved
RDG on these weak buses has been determined based on the the minimum power losses of 2.2525 MW compared
total cost and power losses minimization using the proposed with the power losses with RDG on bus 19 or 26.
CSA-PSO, and then compared with the three metaheuristic Figure 5 exhibits that RDGs penetration results in a

VOLUME 8, 2020 27813

 H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

TABLE 2. Best simulation results of the OPF solution with one RDG inserted at the weak bus 19 or 26 or 30.

FIGURE 5. Power losses minimization without and with one RDG inserted
at the weak bus 19 or 26 or 30.
FIGURE 6. Total cost (FC −Tot ) minimization without and with two RDGs
inserted on buses (19 and 26) or buses (19 and 30) or buses (26 and 30).
significant reduction in the total power losses compared
with that without RDGs. In addition, penetration of RDG
on bus 30 into the test system has the lowest power losses
compared with penetration of RDG on bus 19 or 26.
- The proposed CSA-PSO has superior performance com-
pared to PPSOGSA, TLBO, and PSO based on the two
objective functions (total cost and power losses mini-
mization), as shown in Fig. 4 and Fig. 5.
- In brief, the penetration of RDG on bus 19 is preferred
in order to minimize the total generation cost while
penetration of RDG on bus 30 is preferred to mini-
mize the power losses. Also, the proposed CSA-PSO
outperformed the other three metaheuristic PPSOGSA,
TLBO and PSO algorithms to solve the OPF problem
with and without RDGs. Finally, the optimal control FIGURE 7. Power losses minimization without and with two RDGs
variables of the test power system with RDG on bus inserted on buses (19 and 26) or buses (19 and 30) or buses (26 and 30).
19 are introduced in Table 3.
Case#2: Two DGs inserted at weak buses (19 and at the weak buses 19 and 26 or 19 and 30 or 26 and 30.
26) or (19 and 30) or (26 and 30) Optimal sizing of each two RDGs has been determined
- This case represents the OPF solution for the IEEE based on total cost and total power losses minimization
30-bus test distribution system with two RDGs inserted using the new proposed CSA-PSO and compared to the

27814 VOLUME 8, 2020

H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

TABLE 3. Optimum control variables of the test power system with RDG on bus 19.

TABLE 4. Best simulation results of two different objectives for two RDGs inserted at weak buses (19 and 26) or (19 and 30) or (26 and 30).

three metaheuristic PPSOGSA, TLBO and PSO algo- the previous case (one RDG insertion) and the base case
rithms for the validation purposes. Table 4 introduces (without RDGs). In addition, penetration of RDGs on
the best simulation results of the OPF solution with buses 19 and 30 in the power system has the lowest total
two RDGs inserted at weak buses 19 and 26 or 19 and cost compared with the penetration of RDGs on buses
30 or 26 and 30. Also, it provides comparisons of the 19 and 26 or buses 26 and 30. Also, it enhances the
new proposed CSA-PSO with the PPSOGSA, TLBO voltage stability at all buses, especially on weak buses
and PSO to solve this OPF problem based on the two 19, 26 and 30 as shown in Fig. 3.
objective functions (total generation cost and total power - Insertion of RDG on buses 19 and 30 is preferred
losses minimization). The optimal control variables of to minimize the total power losses, where it achieved
the test power system with RDGs on buses 19 and 30 are the minimum power losses of 2.14425 MW com-
presented in Table 5. Some observations are summarized pared with the total power losses with RDGs on buses
from the simulation results introduced in Table 4 and 19 and 26 or on buses 26 and 30. Figure 7 proves
Figures 6 & 7 as follows: that the RDGs penetration has a significant reduc-
- Insertion of RDGs on buses 19 and 30 is preferred to tion in the total power losses compared with base
minimize the total generation cost, where it achieved case (without RDGs). In addition, penetration of RDGs
the minimum total cost of 617.09 $/h compared with on buses 19 and 30 into the test power system has
the total generation cost with RDGs on buses 19 and the lowest power losses compared with the pene-
26 or buses 26 and 30. Figure 6 proves that the RDGs tration of RDGs on buses 19 and 26 or on buses
penetration would reduce the total cost compared with 26 and 30.

VOLUME 8, 2020 27815

 H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

TABLE 5. Optimum control variables of the test power system with RDGs on buses 19 and 30.

TABLE 6. Statistical results of opf with two rdgs allocated on buses

19 and 30 using different metaheuristic techniques.

buses 19, 26 and 30. Optimal sizing of these three RDGs

was determined based on the total cost and total power losses
FIGURE 8. Convergence characteristic of the proposed CSA-PSO algorithm minimization using the proposed CSA-PSO and compared
compared to other algorithms.
with the three metaheuristic PPSOGSA, TLBO, and PSO
algorithms for the validation purposes. Table 7 shows the
best simulation results of the OPF solution with three RDGs
- The simulation findings prove the superiority of the new
inserted at the weak buses 19, 26 and 30. Also, it provides
proposed CSA-PSO algorithm to solve the OPF prob-
comparisons of the proposed CSA-PSO with the PPSOGSA,
lem with and without RDGs compared with PPSOGSA,
TLBO, and PSO to solve the OPF problem based on the
TLBO and PSO for both the two objective functions
two objective functions (total generation cost and total power
(total cost and transmission power losses minimiza-
losses minimization). Finally, the optimal control variables of
tion), as shown in Fig. 6 and Fig. 7. Figure 8 shows
the test power system with RDGs on buses 19, 26, and 30 are
the convergence characteristic of the proposed CSA-
introduced in Table 8. Some observations are summarized
PSO algorithm compared with PPSOGSA, TLBO and
from the simulation results presented in Table 7 as follows:
PSO. This figure proves that CSA-PSO has the superior
performance in terms of convergence time and track- - Both the total cost and the total power losses started
ing the global solution followed by TLBO, PPSOGSA, to increase with the three RDGs inserted at the weak
and PSO. In addition, the statistical results given buses 19, 26, and 30 (618.698 $/h and 2.56471 MW),
in Table 6 show that the proposed CSA-PSO has the best respectively compared with the previous case (Case#2).
and more stable solutions compared to other metaheuris- - The proposed CSA-PSO outperformed the TLBO, fol-
tic techniques. lowed by the PPSOGSA and PSO algorithm to solve
- In brief, the penetration of RDGs on buses 19 and 30 is the OPF problem based on both the total cost and power
preferred to minimize both the total cost and the total losses minimization objectives.
power losses. In Addition, the new proposed CSA-PSO Figure 9 gives us an indication of the relation between
has superior performance in solving the OPF problem the total power losses versus the RDGs sizes of the best
with and without RDGs compared with the other three simulation results based on total power losses minimization.
metaheuristic PPSOGSA, TLBO, and PSO algorithms. Based on total power losses minimization as an objective,
Case#3: Three RDGs placed at weak buses 19, RDG 30 has the lowest power losses for the case of one RDG
26 and 30 insertion while RDGs on buses 19 and 30 has the lowest
This case represents the OPF solution for the IEEE 30-bus power losses for the case of two RDGs insertions. On the
test distribution system with three RDGs inserted at weak contrary, the power losses for three RDGs insertion case on

27816 VOLUME 8, 2020

H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

TABLE 7. Best simulation results of two different objectives for three TABLE 8. Optimum control variables of the test power system with RDGs
RDGs inserted at the weak buses (19, 26 and 30). on buses 19, 26, and 30.

FIGURE 10. Total cost reduction % for the best buses to allocate RDGs
based on the objective function of total cost minimization.

buses 19 and 30 have superior performance compared with

the other cases in terms of total cost minimization and total
power losses minimization. In the case of total cost minimiza-
tion, the optimal sizes of RDGs 19 and 30 are 43.408MW
and 36.287MW, respectively, in which their optimal sizes are
43.464MW and 16.567MW in the case of total power losses
minimization.
A new reduction percent formula shown in Eq. (22) can
be used to calculate the total cost reduction and the total
FIGURE 9. Power losses versus the RDGs sizes for the best RDGs power losses reduction. These will help us to differentiate
insertion based on total power losses minimization. between the weak buses to select the best RDGs penetration
case. Figure 10 shows the total cost reduction for RDG19,
RDGs (19 and 30), and RDGs (19, 26, and 30), which have
buses 19, 26, and 30 started to increase. As shown in Figure 9, the best results among the three cases based on the objective
as the RDG size increases, the total power losses decreases function of total cost minimization. The best one among the
until it reaches a minimum value (2.14425 MW) at the RDGs cases of RDG19, RDGs (19, and 30), and RDGs (19, 26 and
size (60.031 MW) for the two RDGs on weak buses 19 and 30), is that which has the highest total cost reduction percent.
30 using the proposed CSA-PSO algorithm. Following this This will guide and assist us to select to choose the optimal
minimum value, with increasing the RDGs size (three RDGs numbers of RDGs based on economic benefits. As shown
on buses 19, 26, and 30 with the total size of 71.047 MW), in Figure 10, RDGs on 19, and 30 have the highest total cost
the total power losses increases to 2.56471MW as shown reduction (22.9%) compared with RDG 19 and RDGs (19,
in Figure 9. Therefore, the optimal size and number of RDGs 26, and 30); 13.26% and 22.699%; respectively. Therefore,
can be determined easily. As a result, the two RDGs at weak RDGs 19 and 30 represent the best case of all cases in term

VOLUME 8, 2020 27817

 H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

allocated, RDGs allocated on weak buses 19 and 30 are pre-

ferred based on both total cost and power losses minimization
objectives. RDGs allocated on weak buses 19 and 30 reduced
both the total cost to lowest value (617.09 $/h) and total power
losses also to 2.14425 MW compared with the previous case
(Case#1) and other two RDGs insertion in Case#2. Both the
total generation cost and the total power losses started to
increase with the three RDGs insertion at the weak buses
19, 26, and 30; 618.698 $/h and 2.56471 MW; respectively,
compared with the previous case (Case#2). Therefore, RDGs
on buses 19 and 30 are preferred to minimize both the total
generation cost and the total power losses. On the other hand,
a new reduction percent formula is used to evaluate the total
cost reduction and the total power losses reduction. These cal-
FIGURE 11. Total power losses reduction % based on the objective
function of total power losses minimization.
culations will help us to differentiate between the best cases
based on total cost minimization and the best cases based on
total power losses minimization so that the best RDGs case
of total cost reduction. On the other hand, Figure 11 shows among all best cases can be identified. The results revealed
the total power losses reduction percent for RDG30, RDGs that RDGs allocated on buses 19 and 30 have the highest total
(19, and 30), and RDGs (19, 26, and 30) which have the best cost reduction (22.9%) compared with RDG 19 and RDGs
results among the three cases based on the objective function (19, 26 and 30) with 13.26% and 22.699%, respectively. Also,
of total power losses minimization. The best one from these they have the highest total power losses reduction, 76.15%,
cases, RDG30, RDGs (19, and 30) and RDGs (19, 26, and compared with RDG 30 and RDGs (19, 26 and 30) with
30), is that which has the highest total power losses reduction 74.95% and 71.48%, respectively. Therefore, buses 19 and
percent. This will help us to choose the optimal numbers of 30 are preferred to allocate RDGs where they will achieve
RDGs in terms of technical benefits. As shown in Figure 11, both technical and economic benefits for power system net-
RDGs 19 and 30 have the highest total power losses reduction work. Finally, the simulation results revealed the superiority
percent (76.15%) compared with RDG 30 and RDGs (19, of CSA-PSO algorithm in solving the OPF problem with
26, and 30), 74.95% and 71.48%, respectively. Therefore, and without RDGs compared with PPSOGSA, TLBO and
RDGs on buses 19 and 30 represent the best case of all cases PSO for both the two objective functions (total cost and total
in term of total power losses reduction. In brief, it can be transmission power losses minimization).
concluded that, the insertion of RDGs on buses 19 and 30 has
the highest total cost reduction and the highest total power ACKNOWLEDGMENT
losses reduction. Therefore, RDGs allocated on buses 19 and The authors would like to thank Deanship of scientific
30 are recommended where they will collect both technical research for funding and supporting this research through
and economic benefits for the power system network. the initiative of DSR Graduate Students Research Support at
King Saud University.
VII. CONCLUSION
REFERENCES
The insertion of RDGs into the power system network pro-
[1] B. Poornazaryan, P. Karimyan, G. Gharehpetian, and M. Abedi, ‘‘Optimal
vides many technical and economic benefits such as voltage allocation and sizing of DG units considering voltage stability, losses and
stability, power losses reduction, reliability improvement, load variations,’’ Int. J. Electr. Power Energy Syst., vol. 79, pp. 42–52,
total cost reduction, and power system performance effi- Jul. 2016.
[2] H. Chen, J. Chen, D. Shi, and X. Duan, ‘‘Power flow study and voltage
ciency. This study proposed a novel CSA-PSO algorithm to stability analysis for distribution systems with distributed generation,’’ in
solve the OPF problem with RDGs allocation at weak buses Proc. IEEE Power Eng. Soc. Gen. Meeting, Jun. 2006, p. 8.
[3] R. S. Al Abri, E. F. El-Saadany, and Y. M. Atwa, ‘‘Optimal placement
based on total cost generation minimization and total power and sizing method to improve the voltage stability margin in a distribution
losses minimization. Using the standard IEEE 30-bus test sys- system using distributed generation,’’ IEEE Trans. Power Syst., vol. 28,
tem three weak buses (bus19, 26, and 30) that have the lowest no. 1, pp. 326–334, Feb. 2013.
[4] R. S. Rao, K. Ravindra, K. Satish, and S. V. L. Narasimham, ‘‘Power
voltage in the case of a 25% load increase are selected to loss minimization in distribution system using network reconfiguration in
allocate RDGs initially on it. Three cases of RDGs insertion the presence of distributed generation,’’ IEEE Trans. Power Syst., vol. 28,
are covered; one RDG allocated (Case#1), two RDGs allo- no. 1, pp. 317–325, Feb. 2013.
[5] R. Viral and D. Khatod, ‘‘Optimal planning of distributed generation
cated (Case#2) and three RDGs allocated (Case#3). For one systems in distribution system: A review,’’ Renew. Sustain. Energy Rev.,
RDG allocated, RDG allocated on bus 19 is preferred based vol. 16, no. 7, pp. 5146–5165, Sep. 2012.
[6] D. Q. Hung and N. Mithulananthan, ‘‘Multiple distributed generator place-
on total cost minimization objective where the power system ment in primary distribution networks for loss reduction,’’ IEEE Trans. Ind.
has the lowest total generation cost by 694.22 $/h. Whereas, Electron., vol. 60, no. 4, pp. 1700–1708, Apr. 2013.
RDG allocated on bus 30 is preferred based on the total power [7] D. Q. Hung, N. Mithulananthan, and R. Bansal, ‘‘Analytical
strategies for renewable distributed generation integration considering
losses minimization objective where the power system has energy loss minimization,’’ Appl. Energy, vol. 105, pp. 75–85,
the lowest total power losses by 2.2525 MW. For two RDGs May 2013.
27818 VOLUME 8, 2020
H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

[8] M. Zangiabadi, R. Feuillet, H. Lesani, N. Hadj-Said, and J. T. Kvaløy, [30] A. Zeinalzadeh, Y. Mohammadi, and M. H. Moradi, ‘‘Optimal multi
‘‘Assessing the performance and benefits of customer distributed objective placement and sizing of multiple DGs and shunt capacitor banks
generation developers under uncertainties,’’ Energy, vol. 36, no. 3, simultaneously considering load uncertainty via MOPSO approach,’’ Int.
pp. 1703–1712, Mar. 2011. J. Electr. Power Energy Syst., vol. 67, pp. 336–349, May 2015.
[9] A. Soroudi and M. Ehsan, ‘‘A distribution network expansion planning [31] M. Nayeripour, E. Mahboubi-Moghaddam, J. Aghaei, and A. Azizi-Vahed,
model considering distributed generation options and techo-economical ‘‘Multi-objective placement and sizing of DGs in distribution networks
issues,’’ Energy, vol. 35, no. 8, pp. 3364–3374, Aug. 2010. ensuring transient stability using hybrid evolutionary algorithm,’’ Renew.
[10] S. Bracco, G. Dentici, and S. Siri, ‘‘Economic and environmental opti- Sustain. Energy Rev., vol. 25, pp. 759–767, Sep. 2013.
mization model for the design and the operation of a combined heat and [32] H. Abdel-mawgoud, S. Kamel, M. Ebeed, and M. M. Aly, ‘‘An efficient
power distributed generation system in an urban area,’’ Energy, vol. 55, hybrid approach for optimal allocation of DG in radial distribution net-
pp. 1014–1024, Jun. 2013. works,’’ in Proc. Int. Conf. Innov. Trends Comput. Eng. (ITCE), Feb. 2018,
[11] G. Celli, E. Ghiani, S. Mocci, and F. Pilo, ‘‘A multiobjective evolutionary pp. 311–316.
algorithm for the sizing and siting of distributed generation,’’ IEEE Trans. [33] A. J. G. Mena and J. A. M. García, ‘‘An efficient approach for the siting
Power Syst., vol. 20, no. 2, pp. 750–757, May 2005. and sizing problem of distributed generation,’’ Int. J. Electr. Power Energy
[12] P. Kayal and C. Chanda, ‘‘Placement of wind and solar based DGs in Syst., vol. 69, pp. 167–172, Jul. 2015.
distribution system for power loss minimization and voltage stability [34] Z. Ullah, S. Wang, J. Radosavljevic, and J. Lai, ‘‘A solution to the optimal
improvement,’’ Int. J. Electr. Power Energy Syst., vol. 53, pp. 795–809, power flow problem considering WT and PV generation,’’ IEEE Access,
Dec. 2013. vol. 7, pp. 46763–46772, 2019.
[13] N. Acharya, P. Mahat, and N. Mithulananthan, ‘‘An analytical approach [35] P. P. Biswas, P. Suganthan, and G. A. Amaratunga, ‘‘Optimal power flow
for DG allocation in primary distribution network,’’ Int. J. Electr. Power solutions incorporating stochastic wind and solar power,’’ Energy Convers.
Energy Syst., vol. 28, no. 10, pp. 669–678, Dec. 2006. Manage., vol. 148, pp. 1194–1207, Sep. 2017.
[14] D. Q. Hung, N. Mithulananthan, and R. C. Bansal, ‘‘Analytical expressions [36] E. E. Elattar, ‘‘Optimal power flow of a power system incorporating
for DG allocation in primary distribution networks,’’ IEEE Trans. Energy stochastic wind power based on modified moth swarm algorithm,’’ IEEE
Convers., vol. 25, no. 3, pp. 814–820, Sep. 2010. Access, vol. 7, pp. 89581–89593, 2019.
[15] T. Gözel and M. H. Hocaoglu, ‘‘An analytical method for the sizing and [37] A. M. Eltamaly and M. S. Al-Saud, ‘‘Nested multi-objective PSO for
siting of distributed generators in radial systems,’’ Electric Power Syst. optimal allocation and sizing of renewable energy distributed generation,’’
Res., vol. 79, no. 6, pp. 912–918, Jun. 2009. J. Renew. Sustain. Energy, vol. 10, no. 3, May 2018, Art. no. 035302.
[16] R. Viral and D. Khatod, ‘‘An analytical approach for sizing and siting of [38] H. R. El-Hana Bouchekara, M. A. Abido, and A. E. Chaib, ‘‘Opti-
DGs in balanced radial distribution networks for loss minimization,’’ Int. mal power flow using an improved electromagnetism-like mechanism
J. Electr. Power Energy Syst., vol. 67, pp. 191–201, May 2015. method,’’ Electr. Power Compon. Syst., vol. 44, no. 4, pp. 434–449,
[17] T. Rambabu and P. V. Prasad, ‘‘Optimal placement and sizing of DG based Feb. 2016.
on power stability index in radial distribution system,’’ in Proc. Int. Conf. [39] S. Kamel, A. Ramadan, M. Ebeed, J. Yu, K. Xie, and T. Wu, ‘‘Assessment
Smart Electr. Grid (ISEG), Sep. 2014, pp. 1–5. integration of wind-based DG and DSTATCOM in Egyptian distribution
[18] P. Prakash and D. K. Khatod, ‘‘Optimal sizing and siting techniques for grid considering load demand uncertainty,’’ in Proc. IEEE Innov. Smart
distributed generation in distribution systems: A review,’’ Renew. Sustain. Grid Technol.- Asia (ISGT Asia), May 2019, pp. 1288–1293.
Energy Rev., vol. 57, pp. 111–130, May 2016. [40] A. Ramadan, M. Ebeed, and S. Kamel, ‘‘Performance assessment of
[19] H. Manafi, N. Ghadimi, M. Ojaroudi, and P. Farhadi, ‘‘Optimal placement a realistic egyptian distribution network including PV penetration with
of distributed generations in radial distribution systems using various DSTATCOM,’’ in Proc. Int. Conf. Innov. Trends Comput. Eng. (ITCE),
PSO and DE algorithms,’’ Elektronika ir Elektrotechnika, vol. 19, no. 10, Feb. 2019, pp. 426–431.
pp. 53–57, 2013. [41] A. M. Eltamaly and H. M. Farh, ‘‘Wind energy assessment for five loca-
[20] F. Ugranlı and E. Karatepe, ‘‘Optimal wind turbine sizing to minimize tions in Saudi Arabia,’’ J. Renew. Sustain. Energy, vol. 4, no. 2, Mar. 2012,
energy loss,’’ Int. J. Electr. Power Energy Syst., vol. 53, pp. 656–663, Art. no. 022702.
Dec. 2013. [42] G. L. Johnson, Wind Energy Systems. Englewood Cliffs, NJ, USA:
[21] M. Rahmani-andebili,‘‘Simultaneous placement of DG and capacitor Prentice-Hall, 1985, pp. 147–149.
in distribution network,’’ Electr. Power Syst. Res., vol. 131, pp. 1–10, [43] A. M. Eltamaly, M. A. Mohamed, and A. I. Alolah, ‘‘A novel smart
Feb. 2016. grid theory for optimal sizing of hybrid renewable energy systems,’’ Sol.
[22] S. R. Gampa and D. Das, ‘‘Optimum placement and sizing of DGs consid- Energy, vol. 124, pp. 26–38, Feb. 2016.
ering average hourly variations of load,’’ Int. J. Electr. Power Energy Syst., [44] H. M. Farh, A. M. Eltamaly, and M. S. Al-Saud, ‘‘Interleaved boost
vol. 66, pp. 25–40, Mar. 2015. converter for global maximum power extraction from the photovoltaic
[23] A. S. Bouhouras, K. I. Sgouras, P. A. Gkaidatzis, and D. P. Labridis, system under partial shading,’’ IET Renew. Power Gener., vol. 13, no. 8,
‘‘Optimal active and reactive nodal power requirements towards loss pp. 1232–1238, Jun. 2019.
minimization under reverse power flow constraint defining DG type,’’ [45] A. M. Eltamaly, H. M. Farh, and M. S. Al-Saud, ‘‘Grade point average
Int. J. Electr. Power Energy Syst., vol. 78, pp. 445–454, assessment for metaheuristic GMPP techniques of partial shaded PV sys-
Jun. 2016. tems,’’ IET Renew. Power Gener., vol. 13, no. 8, pp. 1215–1231, Jun. 2019.
[24] A. El-Zonkoly, ‘‘Optimal placement of multi-distributed generation units [46] A. M. Eltamaly and H. M. Farh, ‘‘Dynamic global maximum power point
including different load models using particle swarm optimisation,’’ IET tracking of the PV systems under variant partial shading using hybrid
Gener. Transm. Distrib., vol. 5, no. 7, p. 760, 2011. GWO-FLC,’’ Sol. Energy, vol. 177, pp. 306–316, Jan. 2019.
[25] B. Mohanty and S. Tripathy, ‘‘A teaching learning based optimization [47] H. M. Farh and A. M. Eltamaly, ‘‘Maximum power extraction from the
technique for optimal location and size of DG in distribution network,’’ photovoltaic system under partial shading conditions,’’ in Modern Maxi-
J. Electr. Syst. Inf. Technol., vol. 3, no. 1, pp. 33–44, May 2016. mum Power Point Tracking Techniques for Photovoltaic Energy Systems.
[26] B. Singh, V. Mukherjee, and P. Tiwari, ‘‘A survey on impact assessment Cham, Switzerland: Springer, 2020, pp. 107–129.
of DG and FACTS controllers in power systems,’’ Renew. Sustain. Energy [48] S. Surender Reddy, P. R. Bijwe, and A. R. Abhyankar, ‘‘Real-time
Rev., vol. 42, pp. 846–882, Feb. 2015. economic dispatch considering renewable power generation variability
[27] M. Aman, G. Jasmon, A. Bakar, and H. Mokhlis, ‘‘A new approach and uncertainty over scheduling period,’’ IEEE Syst. J., vol. 9, no. 4,
for optimum simultaneous multi-DG distributed generation units place- pp. 1440–1451, Dec. 2015.
ment and sizing based on maximization of system loadability using [49] A. Askarzadeh, ‘‘A novel metaheuristic method for solving constrained
HPSO (hybrid particle swarm optimization) algorithm,’’ Energy, vol. 66, engineering optimization problems: Crow search algorithm,’’ Comput.
pp. 202–215, Mar. 2014. Struct., vol. 169, pp. 1–12, Jun. 2016.
[28] M. Moradi and M. Abedini, ‘‘A novel method for optimal DG units [50] A. Meddeb, N. Amor, M. Abbes, and S. Chebbi, ‘‘A novel approach based
capacity and location in Microgrids,’’ Int. J. Electr. Power Energy Syst., on crow search algorithm for solving reactive power dispatch problem,’’
vol. 75, pp. 236–244, Feb. 2016. Energies, vol. 11, no. 12, p. 3321, Nov. 2018.
[29] C. L. Borges and D. M. Falcao, ‘‘Optimal distributed generation allocation [51] A. Y. Abdelaziz and A. Fathy, ‘‘A novel approach based on crow search
for reliability, losses, and voltage improvement,’’ Int. J. Electr. Power algorithm for optimal selection of conductor size in radial distribution net-
Energy Syst., vol. 28, no. 6, pp. 413–420, 2006. works,’’ Eng. Sci. Technol., Int. J., vol. 20, no. 2, pp. 391–402, Apr. 2017.

VOLUME 8, 2020 27819

 H. M. H. Farh et al.: Novel CSA Auto-Drive PSO for Optimal Allocation and Sizing of RDG

HASSAN M. H. FARH was born in Egypt, in Jan- ALI MOHAMED ELTAMALY received the B.Sc.
uary 25, 1984. He received the B.S. degree (Hons.) and M.Sc. degrees in electrical engineering from
from Zagazig University, in 2006, and the M.Sc. Al-Minia University, Egypt, in 1992 and 1996,
degree from King Saud University, Riyadh, Saudi respectively, and the Ph.D. degree in electri-
Arabia, in 2013. He is currently a Researcher and cal engineering from Texas A&M University,
teaches many electrical courses with the Electri- in 2000. He is currently a Full Professor with
cal Engineering Department, College of Engineer- King Saud University, Riyadh, Saudi Arabia, and
ing, King Saud University. His electrical courses Mansoura University, Egypt. He is Chair Professor
include power electronics, electric drives, electri- of Saudi Electricity Company. He is also the Chair
cal circuits, power system operation and control, of Power System Reliability and Security, King
electrical machines, power system protection, and numerical methods in Saud University. He has published 20 books and book chapters. He has
electrical engineering. He was a Research Assistant with the College of authored or coauthored more than 150 refereed journal and conference
Engineering Research Center, King Saud University, from April 2013 to papers. His current research interests include renewable energy, smart grids,
August 2015. From February 2009 to April 2013, he was a Teaching Assis- power electronics, motor drives, power quality, artificial intelligence, evolu-
tant with the Electrical Engineering Department, College of Engineering, tionary and heuristic optimization techniques, and distributed generation. He
King Saud University. He has published more than 20 articles in high- held a number of patents in USA patent office. He has supervised a number of
impact journals and international conferences papers. In addition, he has M.S. and Ph.D. theses. He was involved in a number of national/international
authored or coauthored many book chapters. He has reviewed many arti- technical projects. He has received the Distinguish Professor Award for
cles for many refereed journals in his research areas. His current research Scientific Excellence, and the Egyptian Supreme Council of Universities,
interests include renewable energy (wind and solar photovoltaic), distributed Egypt, in June 2017. He has awarded many prizes in different universities in
generation, power quality, power electronics, and smart control technologies Egypt and Saudi Arabia. He is participating as an editor and an associate
(artificial intelligence and metaheuristic optimization techniques). He was editor of many international journals. He has chaired many international
involved in many technical projects and studies related to electrical power conference sessions.
quality and renewable energy applications. He has received the Excellence
in Scientific Research Award from the College of Engineering, King Saud
University, in 2013.

ABDULLAH M. AL-SHAALAN is currently pur-

suing the Ph.D. degree in electrical engineer-
ing from the University of Manchester, U.K.,
in July 1984. He has been a Professor with the
Electrical Engineering Department, College of
engineering, King Saud University, Riyadh, Saudi
Arabia, since 1996. He joined the College of Engi-
neering, King Saud University. He has supervised
many M.Sc. and Ph.D. dissertations and theses.
He has participated in many local, regional, and
international technical and scientific committees, seminars, conferences, and ABDULLRAHMAN A. Al-SHAMMA’A was born
workshops. He offered consultancy services to many governmental entities in Sana’a, Yemen, in 1984. He received the bach-
and private sectors in his specialization areas. He is an active member of elor’s degree in electrical engineering–electrical
many scientific societies and academic affiliations. He has authored and power and machines from Sana’a University,
translated books in various fields of electrical engineering. He is also writing Sana’a, in 2008, and the M.Sc. and Ph.D. degrees
articles for public awareness in the safety of electrical installations and in electrical engineering from King Saud Univer-
energy conservation. His specialization and research interests include power sity, Riyadh, Saudi Arabia, in 2013 and 2019,
system planning, reliability evaluation, electrical safety, environmental pro- respectively. He joined King Saud University,
tection, load management, renewable energy sources, and energy conserva- in 2019, where he is currently a Researcher with
tion. He has obtained a lot of prizes, trophies, and certificates of appreciation the Department of Electrical Engineering. He has
for his vast contributions and wide activities aimed to develop and improve also published many research articles in refereed journals in addition to
the performance of several electric utilities and sectors. He is currently a many international conference papers. His research interest includes the
Supervisor of the Zamil Industrial Group Chair for Electricity Conservation design, controlling, and optimization of renewable energy systems as well
with the College of Engineering, King Saud University. He has been a part- as multilevel power electronics converters for micro-grids and electric drive
time Consultant of Saudi Arabian Standards Organization (SASO), since applications. In the field of multilevel converters, he received two patents for
1995. He has conducted many funded projects for local and foreign agencies. simple, efficient power converters, including applications to grid-connected
He is currently an Expert Representative of the International Electrotechnical photovoltaic plants.
Commission (IEC).

27820 VOLUME 8, 2020