Journal Pre-proof

A hybrid algorithm based optimal placement of DG units for loss

reduction in the distribution system

M.C.V. Suresh, J. Belwin Edward

PII: S1568-4946(20)30131-9
DOI: https://doi.org/10.1016/j.asoc.2020.106191
Reference: ASOC 106191

To appear in: Applied Soft Computing Journal

Received date : 27 May 2019

Revised date : 24 December 2019
Accepted date : 18 February 2020

Please cite this article as: M.C.V. Suresh and J.B. Edward, A hybrid algorithm based optimal
placement of DG units for loss reduction in the distribution system, Applied Soft Computing
Journal (2020), doi: https://doi.org/10.1016/j.asoc.2020.106191.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the
addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive
version of record. This version will undergo additional copyediting, typesetting and review before it
is published in its final form, but we are providing this version to give early visibility of the article.
Please note that, during the production process, errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.

© 2020 Published by Elsevier B.V.

 Journal Pre-proof
*Declaration of Interest Statement

Declaration of Interest Statement

Corresponding author

of
M C V Suresh
Research Scholar,
Department of School of Electrical Engineering,

pro
Vellore Institute of Technology (VIT), Vellore, Tamilnadu, India
Email: [email protected]

Co Author

J. Belwin Edward
Associate Professor,
Vellore Institute of Technology (VIT),
re-
Vellore, Tamilnadu, India

07-May-2019
lP
Dear Editors in chief,

The authors declare that they have no conflict of interest.

rna

Thank you,

Sincerely,

M C V Suresh
Jou
 Journal Pre-proof
*Highlights (for review)

Highlights

 Hybrid algorithm based optimal position and sizing of DG units for loss reduction.

of
 Hybrid algorithm is joined execution of both GOA and CS algorithm.

 GOA is executed to decide optimal answer for the optimization issue.

pro
 GOA optimization behavior is upgraded by utilizing CS algorithm.

 Perfect position of DG is settled with respect to voltage profile and power loss.

re-
lP
rna
Jou
 Journal Pre-proof
*Manuscript
Click here to view linked References

A Hybrid Algorithm Based Optimal Placement of DG Units for Loss

Reduction in the Distribution System

of
M C V Suresh1*, J. Belwin Edward2
1*

pro
Department of School of Electrical Engineering, Vellore Institute of Technology (VIT), Vellore,

Tamilnadu, India
*
Email: [email protected]
2
Associate Professor, Vellore Institute of Technology (VIT), Vellore, Tamilnadu, India

Abstract re-
Distributed generation (DG) has been utilized in some electric power networks. Power loss

reduction, environmental friendliness, voltage improvement, postponement of system upgrading,

lP
and increasing reliability are some advantages of DG-unit application. This paper uses a hybrid

technique to optimize the position and size of DG units to reduce losses in the distribution

system. The hybrid technique is the joined execution of both the Grasshopper Optimization
rna

Algorithm (GOA) and Cuckoo Search (CS) technique. Here, the GOA optimization behavior is

upgraded by utilizing the CS technique. Here, the perfect position of the DG unit is settled with

respect to the power loss, line power flow and voltage profile using the proposed system. For

improving the dynamic execution, the limit of DG is directed by the proposed technique with
Jou

respect to the cost work. The motivation behind the proposed system is to produce optimal

capacity to lessen the aggregate power loss and enhance the voltage profiles of power

distribution networks. The proposed hybrid technique is executed in MATLAB/Simulink

working platform and the dynamic dependability execution is tested and considered with IEEE

33-bus distribution networks and IEEE 69-bus system. The stability by diminishing loss of the
 Journal Pre-proof

distribution system is investigated by executed different load state of the system. The execution

of the proposed system is analyzed and compared with different existing techniques.

of
Keywords: Cuckoo Search (CS), Distribution Networks, Distribution Generation, Grasshopper

Optimization Algorithm (GOA), Loss Reduction.

pro
Nomenclature
Abbreviations Symbols
DG Distribution Generation Vi V j Voltage magnitudes
Grasshopper Optimization c ij conductance between the
GOA
Algorithm nodes
Mixed Integer Non-Linear upper and lower limit of
MINLP Vi max Vi min
Programming re- the i  th bus
specified voltage
CS Cuckoo Search Vi sp
magnitude
ORCS One Rank Cuckoo Search Algorithm N number of buses
active and reactive
GA Genetic Algorithm PDGi QDGi
generation outputs
active and reactive loads
lP
VRs Voltage Regulators Pdi Qdi
at the node
real and imaginary parts of
Comprehensive Teaching Learning- Dij M ij
CTLBO the nodal admittance
Based Optimization
matrix
min
PDGi P max Q min upper and lower limit of
, DGi , DGi
rna

SKHA Stud Krill Herd Algorithm max

the active and reactive
Q DGi generating unit of DG
apparent power limit
LSF Loss Sensitivity Factor Aijmax
between the bus i and j
N
MOALO Multi-Objective Ant Lion Optimizer P
i 1
Li
total active power load of
DN
GRP Gray Relation Projection N obj number of objectives
Jou

Complex Affine Arithmetic Based

CATHPF Three-Phase Harmonic Power Flow y s yc control vector
Technique
commitment of every DG to Loc DGi allocation position of
CDGHV
harmonic voltage i  th DG
maximal allowable
SSO Salp Swarm Optimization N  DG
injection number of DG
mn random behavior of the
LSA Lightning Search Algorithm PDGi particles
f 1 ( x ) , f 2 ( x) , line losses function,
voltage deviation function,
f 3 ( x) cost function of DG
 Journal Pre-proof

1. Introduction

As of late, in the distribution system, the DG units are expanded because of the natural worries

of
of fuel cost vulnerabilities, the progression of power markets and advances in innovation [1]. In

the planning of distribution systems, an enormous amount of effort is offered by this model [2,

pro
3]. The energy injection is the main role of the DG units; be that as it may, deliberately set and

worked DG units can yield a few different advantages to utilities [4, 5]. Here, the DG units are

for the most part utilized for loss reduction while voltage and load capacity improvement,

reliability quality change and network redesign deferral is different advantages [6, 7]. The DG
re-
units also assume an essential role in providing auxiliary authorities with, for example, reactive

power support, loss compensation, frequency control, spinning reserve and other rapid reaction

benefits under power industry deregulation [8, 9]. Then again, the power streams, extreme power
lP
misfortunes, and resulting feeder over-burdens are happened due to the ill-conceived and

shamefully worked DG units. Consequently, the DG unit’s optimal placement is vital to stay

away from these issues [10].

rna

DG devices can be set for grid reinforcement, power loss reduction and peak operating costs,

tension profiles and load factors improvement, system redesign or disposal, and system integrity,

reliability and efficiency improvement in the power frameworks [11-13]. Here, the system losses
Jou

and the cost of the system are expanded due to the nonoptimal position of the DG units [14]. The

vast majority of researchers have developed new methodologies to reduce power losses on the

assumption that firms should allocate DG sources at peak load [15]. However, these advances are

not reasonable to address the problems that occurred during the season of optimal allocation and

sizing of DG units [16]. In addition, the stability and reliability of the framework support is

essential due to the position of multiple DG units in the distribution network [17]. The numerous
 Journal Pre-proof

DG allocations in the distribution network cause numerous issues and these issues are resolved

by the optimization techniques such as Genetic Algorithm (GA), Differential Evolution (DE),

of
Evolutionary Programming (EP), Mixed Integer Non-Linear Programming (MINLP), Particle

Swarm Optimization (PSO) technique [18]. Regardless of the way that the above systems are

pro
utilized for reducing the power losses, the complexity of the algorithm is exceptionally high

because of the expanded number of samples required [19]. To conquer these difficulties,

advanced innovative hybrid schemes are required for optimal sizing and siting of DG units [20].

This research proposes a hybrid GOA and AI technique to optimize the distribution network
re-
allocation and sizing of DG units. Here, the AI technique used the Cuckoo Search (CS)

algorithm for enhancing the searching behavior of the GOA. In view of the bus value, line limit,

DG power limit and the corresponding constraints, the optimal placement and sizing of DG is
lP
resolved. The rest of the section of the paper is portrayed in the section underneath. Section 2

incorporates the recent research works about the DG units in the distribution networks. Section 3

depicts the problem formulation of DG location and sizing. Section 4 shows the proposed
rna

technique. The experimental results and conclusions are included in Section 5 and 6.

2. Recent Research Works: A Brief Review

Various research works have beforehand existed in the literature which depended on the optimal
Jou

location and sizing of distributed generation in the distribution networks utilizing different

procedures and different viewpoints. Portions of the works are explored here.

MCV Suresh et al. [21] have introduced a novel approach for the optimal locations and sizing

of DG units in distribution networks. The author used the fuzzy and One Rank Cuckoo Search

Algorithm (ORCS) to reduce total power losses and enhance voltage profiles of power

distribution networks. The optimum DG locations and optimum DG unit sizes were located using
 Journal Pre-proof

these hybrid techniques. Mehdi Attar et al. [22] have represented a Genetic Algorithm (GA)

which has been utilized as a tool to decide the number, location and rated power of Voltage

of
Regulators (VRs). The GA approach was used to remove voltage violation and reduce power

losses under all load conditions. For the optimal allocation of DGs in radial distribution systems,

pro
a Comprehensive Teaching Learning-Based Optimization (CTLBO) technique was clarified by

Imran Ahmad Quadri et al. [23]. These strategies are used for reducing the annual energy loss

and cost investment funds in the distribution systems. To tackle the issue of DG units' allocation

and sizing in the distribution networks, a productive evolutionary technique in view of the Pareto
re-
optimality was received by Imen Ben Hamida et al. [24]. The author has used a fuzzy set theory

to select the best trade-off solution among Pareto's set.

For the solution of optimal placement and sizing of Distributed Generation (DG) in the radial
lP
distribution system using Stud Krill Herd Algorithm (SKHA) was presented by S.A. ChithraDevi

et al. [25]. The principle objective was to limit the line losses considering in the view of different

constraints like voltage limit, DG real power generation limit, power balance constraint and DG
rna

location constraint. For the optimal distributed generation (DG) planning, a two-stage

optimization technique was displayed by Yang Li et al. [26] and their strategic thinking about the

integration of energy storage. In their approach, Loss Sensitivity Factor (LSF) approach was
Jou

used to decide the installation locations and the initial capacity of DGs. To boost the investment

benefits the author used the Multi-Objective Ant Lion Optimizer (MOALO) which is utilized to

get the Pareto-optimal solutions, and afterward the 'best' compromise solution. These solutions

are calculated by Gray Relation Projection (GRP) technique. To follow the harmonic

contributions of individual DGs in conveyance frameworks, A Complex Affine Arithmetic

Based Three-Phase Harmonic Power Flow technique (CATHPF) was proposed by S. Wang et al.
 Journal Pre-proof

[27]. The commitment of every DG to harmonic voltage (CDGHV) was computed in view of

CATHPF.

of
2.1. Background of the Research

An important contributing factor in the distribution networks is the optimal placement and sizing

pro
of DG units. This can be achieved by optimizing DG sizing and sitting, which helps to reduce

total power losses and improve power distribution network voltage profiles. However, the

distribution network disturbances such as voltage violation, total power loss, line losses and

harmonic distortions are some of the most dominating problems in DGs. Under transient status,
re-
where loads are reallocated at irregular and unmanageable rates, it can mark the time current

curve of protective devices such as fuses, breakers and relays that cause other offline power

sources to swing. However there are many approaches have been implemented for the
lP
installation and sizing of DGs in the distribution networks such as Fuzzy and one rank cuckoo

search Algorithm (ORCS), Genetic algorithm (GA), comprehensive teaching learning-based

optimization (CTLBO), fuzzy set theory, Stud Krill herd Algorithm (SKHA) and so on. Fuzzy
rna

and ORCS have been working to find the optimum DG locations and optimum DG unit sizes.

But iteration changes in the distribution of probabilities make it difficult to analyze theoretically.

GA has the capability to determine the number, location and rated power of voltage regulators in
Jou

the distribution networks. But it exhibits limitations as there is no guarantee that global maxima

will be found, and the system also leads to several complexities. SKHA can find the optimal

location and sizing of DG but it leads to the complexity of technique. Although the above

techniques are used for sizing and location of DGs, the complexity of the technique is very high

due to increased number of samples required. To overcome these challenges, optimal placement

of DGs using advanced technology is required. The optimal placement and sizing of DG units is
 Journal Pre-proof

performed by the utilization of Grasshopper Optimization Algorithm (GOA) and Cuckoo Search

(CS) algorithm. By employing the hybrid technique, the system gives optimal solution with

of
reduced power loss. The proposed system gives less complexity of the technique and reduced

computational time. The mathematical formulation of the DG allocation and sizing is depicted

pro
in the below section.

3. Mathematical Formulation of Distribution Generation (DG) Allocation and Sizing

The optimum position of the DGs in the distribution networks is a critical contributing factor. In

that network, total power losses are reduced and power profile changes are finished by using the
re-
engagement factor that can be accomplished by optimal DG sizing and placement. Be that as it

may, the distribution network is caused by a few issues, for example, voltage violation, total

power loss, line losses and harmonic distortions et cetera. For optimal allocation and sizing of
lP
separation networks, there are numerous innovations on display. The proposed technique

introduced an optimal technique capability for the separation scheme by reducing line loss,

reducing voltage deviation and reducing separation network costs as well.

rna

3.1. The Objective Functions

(a) Line Losses Minimization: The minimization of line losses is the first objective function

after the DG injection into the distribution network. This function can be achieved as follows,

c
Jou

f1  min ij (Vi
2
 V j2  2ViV j cosij ) (1)
( i , j )b

where the set of branches present in the network is denoted as b , the nodes present in the

branches are denoted as (i, j )  b . In the nodes (i, j ) , the voltage magnitudes are represented as Vi

and V j severally. The conductance between the nodes is denoted as c ij , the difference between

the nodal phase angles  i and  j is expressed as  ij .

(b) Minimization of Voltage Deviation: By the second objective function, the voltage deviation
 Journal Pre-proof

between nodal voltage and specified voltage magnitude is reduced. The nodal voltage magnitude

here assumes a noteworthy part for assessing system security and power quality (PQ). In order to

of
achieve a perfect voltage level in the distribution power system, minimizing voltage deviation

can help to calculate the following [28],

pro
2
N
 Vi  Vi sp 
f 2  min  
 max  V min
i  0  Vi i



(2)

where Vi is denote as voltage magnitude at i  th bus, Vi sp is denote as specified voltage

magnitude Vi max and Vi min is represent as upper and lower limit of i  th bus. The number of
re-
buses is denoted as N . In order to find the difference between the node i and the specified

voltage, the exponent of the equation (2) is set to 2.

(c) Minimization of Cost: The cost of the real and the reactive power generated by DG can be
lP
written as follows,
2
N
 Vi  Vi sp 
f 3  min  
 max  V min
i  0  Vi i



(3)
rna

where

C( PDG )   * PDG   * PPDG  

2
$ / MWh (4)

In the above equation, the cost coefficients are considered as follows,

  0,   20,   0.25
Jou

The cost of the DG's reactive power is calculated on the basis of the maximum complex power

generated by DG and can be calculated as follows,


C (QDG )  Cost ( Sg max)  Cost  Sg max 2

 Qg 2 * k (5)

Pg max
Sg max  (6)
cos 

here the power factor is expressed as Pg max  1.1* pg . The power factor for carrying the
 Journal Pre-proof

analysis is considered to be the unity and lagging power factor (0.9). The k value is deemed to be

0.1, where k=0.05-0.1.

of
3.2. Constraints

In this section, the distribution network contains three types of constraints that include equality

pro
and inequality in power flows and DG capacity.

(a) Equality Constraints: The power flow equation's equality constraints are given as follows

[28],

N
PDGi  Pdi  Vi V j ( Dij cos ij  M ij sin  ij )
re- (7)
j 1

N
QDGi  Qdi  Vi V j ( Dij sin  ij  M ij cos ij ) (8)
j 1
lP
where the active and reactive generation outputs are represented as PDGi and QDGi , the active and

reactive loads at the node is denoted as Pdi and Qdi severally. The Dij and M ij is represents as

real and imaginary parts of the nodal admittance matrix.

rna

(b) Inequality Constraints: The power flow equation's inequality constraints are given as

follows [29],

min
PDGi  PDGi  PDGi
max
(9)
Jou

min
QDGi  QDGi  QDGi
max
(10)

The load bus voltage constraint is expressed as follows,

Vi min  Vi  Vi max (11)

The thermal limit is expressed as follows,

Aij  Vi 2 Dij  ViV j ( Dij cos ij  M ij sin  ij )  Aijmax (12)

min max min max

where PDGi , PDGi , QDGi and Q DGi is represent as the upper and lower limit of active and reactive
 Journal Pre-proof

10

generating unit of DG respectively. In the circuit, the apparent power limit between the bus i and

j is denoted as Aijmax .

of
(c) The capacity of DG Constraints: To ensure system reliability, the penetration of each DG

has a limit to a system of distribution. In view of the 25% intrusion factor and the maximum

pro
injection of DG into the distribution network, the maximum DG of total active electrical load

should be less than 25% and is calculated as follows[29],

N DG N

P
i 1
DGi  0.25 PLi
i 1
(13)
re-
where the total active power load of DN is expressed as  PLi .
N

i 1

3.3. Overview Formulation of DG Constraints

lP
The problem of multi objective optimization (MOO) can be formulated and solved on the basis

of aggregating goals and constraints and can be calculated as follows,


min f1 ( y s , y c ), f 2 ( y s , y c ), , fN obj ( y s , y c )  (14)
rna

Subject to

hi ( y s , yc )  0, i  1,, p (15)

g i ( y s , yc )  0, i  1,, q (16)
Jou

where the number of objectives are expressed as N obj , the state and the control vector is

represented as y s and y c respectively. In the injected DG units, the independent adjustable

variables are denoted as y c . The injecting position and DG’s active power output is the two

important variables of the each installed DG which has the prespecified power factor. Here, y c

can be illustrated as follows,

 Journal Pre-proof

11


ycT  LocDG1 , PDG1 ,, LocDGi , , PDGi ,, LocDGN  DG , PDGN  DG  (17)

where the maximum permissible injected number of DG is denoted as N  DG , the allotment

of
location of i  th DG is denoted as Loc DGi . Here, the state vector y s to the system's voltage

quantity and phase angle is calculated and given below:

pro
y sT  V1 ,1 ,,V N , N  (18)

4. The Proposed Technique for Optimal Allocation and Sizing of DG

In the proposed work, there are two parts are considered; the optimal location of DG is the first
re-
part and the optimal sizing of DG is the second part. Here, measuring solutions for optimal

sitting and separation generation are designed as a multi-purpose control optimization complex.

By optimizing these two parts, the proposed technique introduces a hybrid technique which is the
lP
joined execution of both the Grasshopper Optimization Algorithm (GOA) and CS algorithm. If

output is a bus number then DGs is to be installed. The answer obtained from the GOA is

updated by using the CS algorithm.

rna

4.1. Working Procedure of GOA and CS Technique

This section describes grasshopper operation to optimize DG's location and capacity based on

bus values, line limits, DG power limits and constraints [3, 28-29]. The GOA here is a meta-

heuristic population-based technique that works in view of grasshoppers’ swarming behavior in

Jou

nature [30]. The following technique steps are described to optimize the location and size of DG

units in the distribution networks.

Steps of the Technique

Step 1: Initialization

Generate and initialize the bus values, line limits, DG power limits and the corresponding

generation constraints which are acts as the input of the GOA.

 Journal Pre-proof

12

Step 2: Random Generation

The random generation gives the values as randomly and the uniform range is [0, 1].

of
 PDG
11 12
PDG 1n 
 PDG
 21 i 22
i
2n 
i

P PDG  PDG 
Ri   DGi i i
(19)
    
 m1 

pro
 PDGi m2
PDG i
mn
 PDG 
i 

mn
where PDGi
is the random behavior of the particles.

Step 3: Fitness Function

re-
The position of the grasshoppers is used to find the fitness of the population and the multi-

objective function is

J  min  f 1 , f 2 , f 3  (20)
lP
where f 1 is the line losses function, f 2 is the voltage deviation function, f 3 is the cost function

of DG.

min
PDGi  PDGi  PDGi
max min
; QDGi  QDGi  QDGi
max
rna

Vi min  Vi  Vi max

Aij  Vi 2 Dij  ViV j ( Dij cos  ij  M ij sin  ij )  Aijmax

N DG N

 PDGi  0.25 PLi

Jou

i 1 i 1

th th
where PDGi and QDGi is the i inequality constraint and Vi is the i load bus voltage

constraints, Aij is the thermal limits constraints, total active power of DN is represented as PLi .

Step 4: Social Interaction

The social interaction is introduced from the motivating force of locust swarming models. Here,

the social interaction of the grasshopper is derived as follows

 Journal Pre-proof

13

S i   m1 u v ,ij u v ,ij
N
(21)
j 1
j 1

of
where m1 is a function that defines the strength of social forces and the function m1 shows the

th
social interaction i.e., attraction and repulsion of the grasshopper, u v ,ij is a unit vector of i

pro
grasshopper to the j th grasshopper.

Step 5: Gravity Force

The space between two grasshoppers with the help of function m1 is categorized to the comfort
re-
zone, attraction and repulsion region. If the distance becomes greater than 10, the function value

returns to zero. In order to overcome this problem, the distance of grasshoppers is kept in the

interval of [1, 4].

lP
M 2   m2 e v (22)

where m2 is the gravitational constant and ev represents a unit vector towards the center of earth.
rna

Step 6: Wind Advection

The nymph’s (grasshopper young ones) movement is highly correlated with the wind direction.

However, we do not consider the gravity force but assume the wind direction is always towards a

targeted value (T). The wind advection solution can be calculated as follows,
Jou

M 3   m3 e d (23)

where m3 is the constant drift and e d represents the unity vector in the direction of wind.

Step 7: Exploration and Exploitation

The process of optimization is divided into two phases: exploration and exploitation. Exploration

refers to a significant change in highly randomized behavioral solutions. Greater solutions

 Journal Pre-proof

14

changes lead to more exploration. In exploitation, solutions generally face on a smaller scale and

tend the solutions to search locally. Proper balancing of the two phases resulted in finding the

of
optimum global for a given problem of optimization. With the following equation, the parameter

p is derived to reduce exploration and increase exploitation in proportion to the number of

pro
iteration.

pmax  pmin
p  pmax  l (24)
N

where N is the maximum number of iterations l is the current iteration, p max and p min the values

are 1 and 0.00001.

re-
Step 8: Updating the position using CS: Lévy Flight

Lévy flights are random walks whose directions are random and their step lengths are derived
lP
from the Lévy distribution. These Lévy flights are performed by animals and insects and it is

characterized by series of straight flights followed by sudden turns. Compared to normal random

walks, Lévy flights are more efficient in exploring large scale search areas. That is mainly due to
rna

Lévy flights variances increases much faster than that of the normal random walk. Lévy flights

can reduce the number of optimization algorithms iterations by about four orders compared to

normal random walk. Here, update the DG units’ optimum position and size based on the (p)

value. Update the T (target solution) after the values are updated if a better solution is found m=
Jou

m+1. If the particle failed to reach good solutions, the solution for generating the random number

using the AI technique used the CS flight search equation (25) for the better solution [31].

X mt 1  xmt    Levy( ) (25)

where   0 the level of scale, is related to the level of interest,  means the entrance-wise
multiplication. While in researching the Levy flight equation is distributed the following
 Journal Pre-proof

15

probability distribution.

Levy ( )  t  , 1    3 (26)

of
Step 9: Find the fitness probability rate using pa  [0,1] , whereas the best solution can be

determined by the equation (20).

pro
Step 10: Remember the best solution ever made.

Step 11: For checking the iteration range the conditions are:

 Iteration value is increased by m=m+1, when iteration does not reach the maximum


value. re-
The process is terminated when the iteration reach maximum value.
lP
rna
Jou

Figure 1: Optimal Placement and Sizing of Distribution DG Units: Hybrid Technique

 Journal Pre-proof

16

The system provides the optimum location and size of the DG units in the distribution

network at minimum cost once the process is completed. The proposed adaptive technique is

of
explained briefly in figure 1.

5. Experimental Results and Discussions

pro
In this section, the optimal allocation and sizing of the IEEE 33-bus distribution networks [31-

35] and the IEEE 69-bus system with DG are considered and tested to determine the

performance and effectiveness of the proposed technique. In the MATLAB/Simulink platform,

the technique's environmental configurations are implemented and the dynamic stability
re-
performance is tested using standard IEEE benchmark systems. Here, the different load condition

of the system was executed by reducing the loss of the distribution system that can be analyzed.

This compares the execution of the proposed technique to existing works such as the Lightning
lP
Search Algorithm (LSA) and the Salp Swarm Optimization (SSO) algorithm.

5.1. IEEE 33-Bus Case

Figure 2 shows the IEEE 33-bus system that consolidates the DG units. Here, the hybrid GOA
rna

and AI technique is used to solve the problems that occurred during the allocation and sizing of

DG units in the distribution networks. Here the IEEE 33-bus distribution network has the three

types of scenarios with a DG unit. The voltage profile of the proposed technique for different
Jou

types of loads is depicted in scenario 1. Scenario 2 illustrates the analysis of power loss for

various load types. Scenario 3 shows the fitness analysis for different types of loads. Here, the

normal voltage and power loss of the three types of the load is illustrated in Figure 3. As seen

from the fig 3 (a), it is observed that the voltage in the IEEE 33 bus system gives normal state of

voltage with the per unit value of 1 to 0.902. By using the proposed technique, the Fig 3 (b)

shows minimum power loss in the IEEE 33 bus system.

 Journal Pre-proof

17

of
pro
Figure 2: IEEE 33-Bus Distribution Network
re-
lP
rna

Figure 3: DGs voltage and power loss of IEEE 33-bus system (a) Normal Voltage (b) Normal

Power Loss

Figure 4 shows the loading conditions for type 1, type 2, and type 3 of the proposed
Jou

technique and the existing technique. It illustrates the voltage profile analysis of the proposed

and existing IEEE 33 bus distribution network techniques such as SSA and LSA. Here, the IEEE

33 bus system's voltage profile analysis is carried out using three types of loads such as load

1,2,3 are light load, nominal load and heavy load. The subplot 4(a) shows the load-1 voltage,

figure 4(b) shows the load-1 voltage comparison results, figure 4(c) shows the load-2 voltage,

figure 4(d) shows the load-2 voltage comparison results, figure 4(e) shows the load-3 voltage,
 Journal Pre-proof

18

figure 4(f) shows the load-3 comparison results.

of
pro
re-
lP
rna
Jou

Figure 4: Voltage Profile Analysis and Comparison Results of IEEE-33 Bus System Using

Proposed technique for Different Types of Load Levels (a) Load 1 Voltage (b) Load 1 Voltage

Comparison (c) Load 2 Voltage (d) Load 2 Voltage Comparison(e) Load 3 Voltage (f) Load 3

Voltage Comparison
 Journal Pre-proof

19

In the voltage profile analysis, the bus numbers are indicated in the x-axis and the voltage

level (p.u) is indicated in the y-axis. The voltage of the load 1, 2 and 3 is compared with the

of
previously developed techniques such as SSA and LSA and also the normal and the loaded level

voltage. This comparison result shows a better voltage level for the proposed technique. The

pro
voltage level is slightly higher than the normal and the loaded voltage in figure 4(a), 4(c) and

4(e). In the comparison results of the figure 4(b), 4(d), and 4(f) illustrates, the voltage level of the

proposed GOA-CS technique is almost equal to the SSA technique and slightly increased when

compared with the other techniques.

re-
Figure 5 shows the power loss profile analysis of the IEEE 33 bus distribution network's

proposed technique. In this way, the power loss achieved by the imaginative procedure for all the

load conditions is surveyed with different strategies, for example, the LSA and SSA. Three types
lP
of loads such as load 1 (light), load 2 (nominal) and load 3 (heavy) are used to implement the

power loss profile analysis of the IEEE 33 bus system. The figure 5(a) illustrates the load 1

power loss, figure 5(b) illustrates the comparison results of the load 1 power loss, figure 5(c)
rna

illustrates the load 2 power loss, figure 5(d) illustrates the comparison results of the load 2 power

loss, figure 5(e) illustrates the load 3 power loss, figure 5(f) illustrate the comparison results of

the load 3. In the power loss analysis, the bus numbers are indicated in the x-axis and the power
Jou

loss (KW) is indicated in the y-axis. Here, the proposed technique is capable of significantly

reducing distribution network power loss with DGs. The power loss of the load 1, 2 and 3 is

compared with the previously developed techniques such as SSA and LSA and also the normal

and the loaded level power loss. These comparison result shows that the proposed technique very

effective to reduce the power loss.

 Journal Pre-proof

20

of
pro
re-
lP
rna

Figure 5: Power Loss Profile Analysis and Comparison Results of IEEE-33 Bus System Using
Jou

Proposed technique for Different Types of Load Levels (a) Load 1 Power Loss (b) Load 1 Power

Loss Comparison (c) Load 2 Power Loss (d) Load 2 Power Loss Comparison(e) Load 3 Power

Loss (f) Load 3 Power Loss Comparison

The fitness analysis and the comparison result of the proposed technique illustrated in Figure

6. Here, figure 6(a), 6(c) and 6(e) illustrate the fitness level of the load 1, load 2 and load 3. The

comparison results of the load 1, 2 and 3 fitness are shown in the figure 6(b), 6(d) and 6(f). In the
 Journal Pre-proof

21

graph x-axis is denoted as iteration ranges and y-axis is denoted as fitness level. In proposed

technique the fitness function is given based on the voltage deviation, line losses and the cost of

of
the system. If these three functions are increased the proposed technique gives better result.

Here, the proposed method has minimum fitness value than the existing technique is shown in

pro
figure 6(b), 6(d) and 6(f) such as SSA and LSA, which means the proposed technique has

minimum line losses, minimum voltage deviation, and minimum cost. Figure 6 (a) shows the

fitness graph of load 1 using proposed technique. In figure 6 (a), the graph shows that the

proposed technique converges at the iteration count of 30 with the fitness value of 129. Figure 6
re-
(b) depicts the load 1 fitness comparison of proposed and existing techniques. The fitness of LSA

is converged at the iteration count of 42 with fitness value of 168. The fitness of SSA is

converged at the iteration count of 39 with fitness value of 165. By comparing the proposed
lP
technique the existing techniques takes too much of iterations to get the solution. But the

proposed technique gives better solution than the existing techniques. Figure 6 (c) shows the

fitness graph of load 2 using proposed technique. From the Figure 6 (c), the graph showing the
rna

proposed technique converges at the iteration count of 38 with fitness value of 162. Figure 6 (d)

depicts the load 2 fitness comparison of proposed and existing techniques. The fitness of LSA is

converged at the iteration count of 46 with fitness value of 178. The fitness of SSA is converged
Jou

at the iteration count of 44 with fitness value of 169. By comparing the proposed technique the

existing techniques takes too much of iterations to get the solution. But the proposed technique

gives better solution than the existing techniques. Figure 6 (e) shows the fitness graph of load 3

using proposed technique. From the Figure 6 (e), the graph showing the proposed technique

converges at the iteration count of 44 with the fitness value of 168. Figure 6 (f) depicts the load 3

fitness comparison of proposed and existing techniques. The fitness of LSA is converged at the
 Journal Pre-proof

22

iteration count of 59 with fitness value of 181. The fitness of SSA is converged at the iteration

count of 52 with fitness value of 178. By comparing the proposed technique the existing

of
techniques takes too much of iterations to get the solution. But the proposed technique gives

better solution than the existing techniques.

pro
re-
lP
rna
Jou

Figure 6: Fitness Analysis and Comparison Results of IEEE-33 Bus System Using Proposed

technique for Different Types of Load Levels (a) Load 1 Fitness (b) Load 1 Fitness Comparison

(c) Load 2 Fitness (d) Load 2 Fitness Comparison(e) Load 3 Fitness (f) Load 3 Fitness

Comparison
 Journal Pre-proof

23

Table 1: Parameter Configuration of the IEEE-33 Bus System Using Proposed and the Existing

Technique

of
Scenario Parameters Solution Techniques

pro
Proposed SSA LSA
(GOA-CS)

Normal Loss 210 210 210

Load Loss 214.3143 214.3143 214.3143
After Dg Added 123.6232 134.2687 138.1508
Loss
DG Size
DG Located Bus
No
re- 1716.7755
2
1041.2204
15
954.4456
23

Case 1: 0.5 Cost($) 34.3361 20.8253 19.0892

Load Minimum Bus 0.9574 0.9462 0.9326
Voltage
lP
Normal Loss 210 210 210
Load Loss 219.9908 219.9908 219.9908
After Dg Added 139.5915 143.0741 148.6011
Loss
DG Size 926.99 853.8882 757.097
rna

DG Located Bus 24 16 13
Case 2: 1 Load No
Cost($) 18.540 17.078 15.142
Minimum Bus 0.9006 0.9062 0.8993
Voltage

Normal Loss 210 210 210

Jou

Load Loss 219.9908 219.9908 219.9908

After Dg Added 139.5915 143.0741 148.6011
Loss
DG Size 926.9943 853.8882 757.0975
DG Located Bus 24 16 13
Case 3: 1.5 No
Load Cost($) 18.5404 17.0785 15.1426
Minimum Bus 0.9006 0.9062 0.8993
Voltage

Table 1 illustrates the parameter configuration of the proposed and the existing technique. It
 Journal Pre-proof

24

involves the normal loss, load loss, after DG added loss, DG size, DG located bus number, cost

of the system, minimum bus rate and the voltage level of the proposed technique and the existing

of
technique namely SSA and LSA. It involves 3 cases; case 1 involves the details of the 0.5 load,

case 2 depicts the details of the load 1 and the details about the load 1.5 are included in the case

pro
3.

5.2. Experiment on the IEEE 69-Bus Case

Figure 7, which combines the DG units, shows the IEEE 69-bus system. Here, the crossover

technique of GOA and AI is connected to take care of the problems that have occurred in the
re-
season of the portion of DG units and to estimate the distribution networks. Here, there are three

kinds of situations in the IEEE 69-bus distribution network with DG unit. The voltage profile of

the proposed technique for various sorts of loads is delineated in situation 1. The situation 2
lP
represents the power loss analysis for various kinds of loads. The fitness analysis for various

sorts of loads is delineated in the situation 3.

rna
Jou

Figure 7: IEEE 69-Bus Distribution Network

 Journal Pre-proof

25

The normal voltage of the IEEE-69 bus system with DG and power loss is shown in Figure 8.

As seen from the Figure 8(a), it is observed that the voltage in the IEEE 33 bus system gives

of
normal state of voltage with the per unit value of 1 to 0.975. By using the proposed technique,

the Figure 8 (b) shows minimum power loss in the IEEE 69 bus system.

pro
re-
lP

Figure 8: IEEE-69 Bus system with DG has voltage and power loss (a) Normal Voltage (b)

Normal Power Loss

rna

Figure 9 describes the IEEE 69-bus system's distinctive loading conditions using the

proposed technique and current technique. Figure 9 has the proposed and current tension profile

strategies, such as the IEEE 69 bus distribution network SSA and LSA. Here, IEEE 69 bus

system voltage profile analysis is performed using three types of loads, e.g. load 1, 2, 3 are light
Jou

nominal and heavy. The subplot (a), (c) and (e) delineates the load 1, 2, and load 3 voltage;

Figure (b), (d) and (f) represents the correlation results of the load 1, 2 and load 3 voltage. In the

voltage profile analysis, the bus numbers are shown in the x-axis and the voltage level (p.u) is

demonstrated in the y-axis. The voltage of the load 1, 2 and 3 is contrasted with the already

created strategies, for example, SSA and LSA with the normal and the loaded level voltage. This

correlation result demonstrates that the proposed technique has better voltage level.
 Journal Pre-proof

26

of
pro
re-
lP
rna
Jou

Figure 9: Voltage Profile Analysis and Comparison Results of IEEE-69 Bus System Using

Proposed technique for Different Types of Load Levels (a) Load 1 Voltage (b) Load 1 Voltage

Comparison (c) Load 2 Voltage (d) Load 2 Voltage Comparison(e) Load 3 Voltage (f) Load 3

Voltage Comparison
 Journal Pre-proof

27

of
pro
re-
lP
rna
Jou

Figure 10: Power Loss Profile Analysis and Comparison Results of IEEE-69 Bus System Using

Proposed technique for Different Types of Load Levels (a) Load 1 Power Loss (b) Load 1 Power

Loss Comparison (c) Load 2 Power Loss (d) Load 2 Power Loss Comparison(e) Load 3 Power

Loss (f) Load 3 Power Loss Comparison

Figure 10 delineates the Power Loss Profile Analysis of the proposed technique for the IEEE

69 bus distribution network. Along these lines, with various systems, for instance, the LSA and
 Journal Pre-proof

28

SSA, the power loss achieved by the inventive technique for all the load conditions are studied.

Here, the power loss profile analysis of the IEEE 33 bus system is actualized by utilizing three

of
kinds of loads, for example, load 1, 2, 3 are light nominal and heavy. The figure (a), (c) and (e)

outlines the load 1, 2 and load 3 power loss, figure (b), (d) and (f) represents the examination

pro
results of the load 1, 2 and load 3 power loss. In the power loss investigation, the bus numbers

are shown in the x-axis and the power loss (KW) is demonstrated in the y-axis. Here, the

proposed technique is dexterous to broadly diminishing the power loss of the distribution

networks with DGs. The power loss of the load 1, 2 and 3 is compared with existing techniques
re-
and furthermore the normal and the loaded level power loss. This comparison result

demonstrates that the proposed technique is extremely successful to decrease the power loss.

The fitness analysis and the comparison result of the IEEE-69 bus system utilizing the
lP
proposed technique outlined in Figure 11. Here, Figure 11 (a), (c) and (e) delineate the fitness

level of the load 1, load 2 and load 3. Figure 11 (b), (d) and (f) represents the examination results

of the load 1, 2 and 3 fitness. In the graph x-axis is denote as iteration ranges and y-axis is
rna

represent as fitness levels. Figure 11 (a) shows the fitness graph of load 1 using proposed

technique. From the Figure 11 (a), the graph showing the proposed technique converges at the

iteration count of 38 with the fitness value of 161. Figure 11 (b) depicts the load 1 fitness
Jou

comparison of proposed and existing techniques. The fitness of LSA is converged at the iteration

count of 44 with fitness value of 184. The fitness of SSA is converged at the iteration count of 41

with fitness value of 168. By comparing the proposed technique the existing techniques takes too

much of iterations to get the solution. But the proposed technique gives better solution than the

existing techniques. Figure 11 (c) shows the fitness graph of load 2 using proposed technique.

From the Figure 11 (c), the graph showing the proposed technique converges at the iteration
 Journal Pre-proof

29

count of 38 with fitness value of 162.

of
pro
re-
lP
rna
Jou

Figure 11: Fitness Analysis and Comparison Results of IEEE-69 Bus System Using Proposed

technique for Different Types of Load Levels (a) Load 1 Fitness (b) Load 1 Fitness Comparison

(c) Load 2 Fitness (d) Load 2 Fitness Comparison(e) Load 3 Fitness (f) Load 3 Fitness

Comparison
 Journal Pre-proof

30

Figure 11 (d) depicts the load 2 fitness comparison of proposed and existing techniques. The

fitness of LSA is converged at the iteration count of 44 with fitness value of 178. The fitness of

of
SSA is converged at the iteration count of 43 with fitness value of 169. By comparing the

proposed technique the existing techniques takes too much of iterations to get the solution. But

pro
the proposed technique gives better solution than the existing techniques. Figure 11 (e) shows the

fitness graph of load 3 using proposed technique. From the Figure 11 (e), the graph showing the

proposed technique converges at the iteration count of 44 with the fitness value of 168. Figure 11

(f) depicts the load 3 fitness comparison of proposed and existing techniques. The fitness of LSA
re-
is converged at the iteration count of 59 with fitness value of 181. The fitness of SSA is

converged at the iteration count of 54 with fitness value of 178. By comparing the proposed

technique the existing techniques takes too much of iterations to get the solution. But the
lP
proposed technique gives better solution than the existing techniques.

Table 2 outlines the parameter configuration IEEE-69 bus system utilizing of the proposed
rna

and the current technique. It includes the normal loss, load loss after DG added loss, DG size,

DG located bus number, cost of the system, minimum bus rate and the voltage level of the 69-

bus system using proposed technique and the current technique, in particular, SSA and LSA. It

includes 3 cases; case 1 includes the subtle elements of the 0.5 load, case 2 portrays the points of
Jou

interest of the load 1 and the insights about the load 1.5 are incorporated into the case 3. Table 3

depicts the statistical analysis such as mean, median and standard deviation (SD) of proposed

and existing techniques. The mean of SSA obtains 1.2145, LSA achieves 1.1538, GOA gives

1.0643 and the proposed technique acquires 0.9681. The median of SSA obtains 0.9723, LSA

gives 0.9416, GOA attains 0.9302 and the proposed technique acquires 0.9062. The SD of SSA

obtains 0.3253, LSA acquires 0.2858, GOA attains 0.2065 and the proposed technique achieves
 Journal Pre-proof

31

0.1099.

Table 2: Parameter Configuration of the IEEE-69 Bus System Using Proposed and the Existing

of
techniques

Scenario Parameters Solution Techniques

pro
Proposed SSA LSA
(GOA-CS)

Normal Loss 225.1789 225.1789 225.1789

Load Loss 228.0480 228.0480 228.0480
After Dg Added
Loss
DG Size
re-141.4399

1930.7119
154.5098

1840.8385
160.6709

1804.9624
DG Located Bus 17 54 3
No
Case 1: 0.5 Cost($) 38.6143 36.8172 36.0991
lP
Load Minimum Bus 0.9226 0.9221 0.9205
Voltage

Normal Loss 225.1789 225.1789 225.1789

Load Loss 230.1268 230.1268 230.1268
After Dg Added 147.6647 163.5727 183.4039
rna

Loss
DG Size 1990.7119 1870.1965 1765.7547
DG Located Bus 6 31 11
Case 2: 1 Load No
Cost($) 39.8144 37.4046 35.3156
Minimum Bus 0.9236 0.9223 0.9204
Voltage
Jou

Normal Loss 225.1789 225.1789 225.1789

Load Loss 234.3862 234.3862 234.3862
After Dg Added 151.6664 171.2928 186.0802
Loss
DG Size 1890.6615 1790.7862 1748.5033
DG Located Bus 12 26 32
Case 3: 1.5 No
Load Cost($) 37.8135 35.8167 34.9708
Minimum Bus 0.9221 0.9210 0.9204
Voltage
 Journal Pre-proof

32

Table 3: Performance analysis of IEEE 33 bus system

Objective function value

Solution Technique

of
f1 f2 f3

GA [36] 0.1063 0.0407 1.0537

pro
PSO [37] 0.1053 0.0335 1.0804

GA-PSO [37] 0.1034 0.0124 1.0517

CS 0.1012 0.0118 1.0508

LSA 0.1010 0.0111 1.0447

SSA
re-
0.1007 0.0011 1.0321

GOA 0.1001 0.0008 1.0117

Proposed 0.0089 0.0005 1.0010

lP

Table 4: Performance analysis of IEEE 69 bus system

Objective function value

rna

Solution Technique
f1 f2 f3

GA [36] 0.089 0.012 1.0303

PSO [37] 0.0832 0.0049 1.0335

Jou

GA-PSO [37] 0.0811 0.0031 1.0237

CS 0.0752 0.0029 1.0232

LSA 0.0667 0.0019 1.0177

SSA 0.0561 0.0017 1.0088

GOA 0.0452 0.0015 1.0065

Proposed 0.0013 0.0009 1.0048

 Journal Pre-proof

33

Table 5: Elapsed time with 100 number of trails

Solution Techniques Elapsed Time (sec)

of
GA 31.956

PSO 24.647

pro
GA-PSO 22.426

CS 18.736

LSA 17.695

SSA 16.789

GOA
re- 11.352

Proposed 10.769
lP
Table 6: Performance comparison of proposed with existing method for 50 number of trails

Performance 50 trails

Measures Accuracy Specificity Recall Precision

rna

GA 1.2097 1.0211 0.894 0.725

PSO 1.675 1.669 0.75 1.305

GA-PSO 2.852 2.777 2.553 2.996

Jou

CS 3.452 2.718 2.002 2.550

LSA 3.500 3.778 3.21 2.942

SSA 3.617 3.899 3.726 3.8801

GOA 4.361 4.1237 4.811 4.521

Proposed 5.726 5.982 5.54 4.68

 Journal Pre-proof

34

Table 7: Performance comparison of proposed with existing method for 100 number of trails

Performance 100 trails

of
Measures Accuracy Specificity Recall Precision

GA 0.65 0.54 0.57 0.56

pro
PSO 0.72 0.64 0.65 0.65

GA-PSO 0.8 0.74 0.83 0.71

CS 0.91 0.84 0.83 0.96

LSA 0.96 0.93 0.87 0.99

SSA 1.04
re- 0.99 0.97 0.93

GOA 3.789 5.126 4.269 5.780

Proposed 5.112 6.127 5.62 7.235

lP

Table 3 depicts the performance analysis of proposed and existing technique in the IEEE 33

bus system. The result of GA, PSO and GA-PSO are taken from the reference [36, 37]. The
rna

remaining techniques are executed through the MATLAB/Simulink working platform. The

objective value of each technique is depicted in the Table 3. The proposed technique objective

value is more optimal than the existing technique by clearly watching the Table 3. Table 4
Jou

depicts the performance analysis of proposed and existing technique in the IEEE 69 bus system.

Table 5 explains the elpased time with 100 number of trails. The proposed technique achieves

the optimal solution with less elapsed time of 9.489 sec. The performance obtained from this

hybrid technique is compared with the existing approaches to analyze which technique has better

performance. The performances are analyzed subject to the classes true positive (TP), true

negative (TN), false positive (FP) and false negative (FN). The correctly labelled positive signals
 Journal Pre-proof

35

are known as TP. The correctly labelled negative signals are known as TN. The incorrectly

labelled negative signals are known as FP. The incorrectly labelled positive signals are known as

of
FN. In order to evaluate the optimal placement of DG units results, the performance measures

are computed as follows,

pro
TP  TN
Accuracy  (27)
TP  TN  FP  FN

TP
Pr ecision  (28)
TP  FP

TP
Re call (or ) specificit y 
re- (29)
TP  FN

TN
Sensitivit y  (30)
TN  FP

In the above equation, if the sensitivity value is positive, (i.e.) true positive fraction, the fault
lP
occurs in the line signal of the transmission power system. If the specificity value is negative,

(i.e.) true negative fraction that means the transmission line signal is in a healthy condition.
rna

Table 6 and 7 illustrates the accuracy, sensitivity, recall and specificity of the proposed method

and existing techniques for 50 and 100 number of trails. By observing the Table 6 and 7, the

analyzed result describes that the proposed method performs very well than the existing

techniques in terms of accuracy, sensitivity, recall and specificity.

Jou

6. Conclusion

A combined technique for solving location and capacity issues for DG units in the distribution

system is presented in this research. The location and capacity of DG is determined by GOA and

CS techniques using this technique. The searching behaviour of the GOA is improved by the CS

levy flight algorithm in the proposed technique. The combined technique for the 33 and 69 bus

systems is implemented to minimize losses and increase voltage stability. To determine the
 Journal Pre-proof

36

effectiveness of this research, the results obtained from the proposed technique were compared

with the results obtained from the two other existing techniques. Also, the proposed technique is

of
able to find the distribution network's better optimized solution. The combined technique is

shown the higher capacity in finding the optimum solutions in this research. The future scope of

pro
the research work is the assessment of impact on power loss, by considering both leading and

lagging power factor while sizing of DGs. Testing these methods on larger bus systems. Though

the power loss reduction was very good in loss sensitivity approach still a better method can be

devised for sizing of DG.

Funding Information
re-
This research did not receive any specific grant from funding agencies in the public, commercial,

or not-for-profit sectors.
lP
References

[1] T. Gözel and M. Hocaoglu, "An analytical method for the sizing and sitting of distributed

generators in radial systems", Electric Power Systems Research, vol. 79, no. 6, pp. 912-918,
rna

2009.

[2] P. Georgilakis and N. Hatziargyriou, "Optimal Distributed Generation Placement in Power

Distribution Networks: Models, Methods, and Future Research", IEEE Transactions on

Jou

Power Systems, vol. 28, no. 3, pp. 3420-3428, 2013.

[3] N. Acharya, P. Mahat and N. Mithulananthan, "An analytical approach for DG allocation in

primary distribution network", International Journal of Electrical Power & Energy Systems,

vol. 28, no. 10, pp. 669-678, 2006.

[4] D. Hung and N. Mithulananthan, "Multiple Distributed Generator Placement in Primary

Distribution Networks for Loss Reduction", IEEE Transactions on Industrial Electronics,

 Journal Pre-proof

37

vol. 60, no. 4, pp. 1700-1708, 2013.

[5] R. Rao, K. Ravindra, K. Satish and S. Narasimham, "Power Loss Minimization in

of
Distribution System Using Network Reconfiguration in the Presence of Distributed

Generation", IEEE Transactions on Power Systems, vol. 28, no. 1, pp. 317-325, 2013.

pro
[6] D. Hung, N. Mithulananthan and R. Bansal, "Analytical Expressions for DG Allocation in

Primary Distribution Networks", IEEE Transactions on Energy Conversion, vol. 25, no. 3,

pp. 814-820, 2010.

[7] D. Popović, J. Greatbanks, M. Begović and A. Pregelj, "Placement of distributed generators

re-
and reclosers for distribution network security and reliability", International Journal of

Electrical Power & Energy Systems, vol. 27, no. 5-6, pp. 398-408, 2005.

[8] A. Mohamed Imran and M. Kowsalya, "Optimal size and sitting of multiple distributed
lP
generators in distribution system using bacterial foraging optimization", Swarm and

Evolutionary Computation, vol. 15, pp. 58-65, 2014.

[9] S. H, Lee and J. W. Park, "Selection of Optimal Location and Size of Multiple Distributed
rna

Generations by Using Kalman Filter s", IEEE Transactions on Power Systems, vol. 24, no.

3, pp. 1393-1400, 2009.

[10] Z. Liu, F. Wen and G. Ledwich, "Optimal Sitting and Sizing of Distributed Generators in
Jou

Distribution Systems Considering Uncertainties", IEEE Transactions on Power Delivery,

vol. 26, no. 4, pp. 2541-2551, 2011.

[11] V. Pandi, H. Zeineldin and W. Xiao, "Determining Optimal Location and Size of

Distributed Generation Resources Considering Harmonic and Protection Coordination

Limits", IEEE Transactions on Power Systems, vol. 28, no. 2, pp. 1245-1254, 2013.

[12] S. Ghosh, S. Ghoshal and S. Ghosh, "Optimal sizing and placement of distributed
 Journal Pre-proof

38

generation in a network system", International Journal of Electrical Power & Energy

Systems, vol. 32, no. 8, pp. 849-856, 2010.

of
[13] R. Al Abri, E. El-Saadany and Y. Atwa, "Optimal Placement and Sizing Method to

Improve the Voltage Stability Margin in a Distribution System Using Distributed

pro
Generation", IEEE Transactions on Power Systems, vol. 28, no. 1, pp. 326-334, 2013.

[14] D. Khatod, V. Pant and J. Sharma, "Evolutionary programming based optimal placement of

renewable distributed generators", IEEE Transactions on Power Systems, vol. 28, no. 2, pp.

683-695, 2013.
re-
[15] A. El-Zonkoly, "Optimal placement of multi-distributed generation units including different

load models using particle swarm optimization", IET Generation, Transmission &

Distribution, vol. 5, no. 7, p. 760, 2011.

lP
[16] L. Ochoa and G. Harrison, "Minimizing Energy Losses: Optimal Accommodation and

Smart Operation of Renewable Distributed Generation", IEEE Transactions on Power

Systems, vol. 26, no. 1, pp. 198-205, 2011.

rna

[17] K. Nekooei, M. Farsangi, H. Nezamabadi-Pour and K. Lee, "An Improved Multi-Objective

Harmony Search for Optimal Placement of DGs in Distribution Systems", IEEE

Transactions on Smart Grid, vol. 4, no. 1, pp. 557-567, 2013.

Jou

[18] D. Singh, D. Singh and K. Verma, "Multi-objective Optimization for DG Planning With

Load Models", IEEE Transactions on Power Systems, vol. 24, no. 1, pp. 427-436, 2009.

[19] G. Celli, E. Ghiani, S. Mocci and F. Pilo, "A Multi-objective Evolutionary Algorithm for

the Sizing and Sitting of Distributed Generation", IEEE Transactions on Power Systems,

vol. 20, no. 2, pp. 750-757, 2005.

[20] R. Singh and S. Goswami, "Optimum allocation of distributed generations based on nodal
 Journal Pre-proof

39

pricing for profit, loss reduction, and voltage improvement including voltage rise

issue", International Journal of Electrical Power & Energy Systems, vol. 32, no. 6, pp. 637-

of
644, 2010.

[21] M. Suresh and B. J., "Optimal Placement of DG Units for Loss Reduction in Distribution

pro
Systems Using One Rank Cuckoo Search Algorithm", International Journal of Grid and

Distributed Computing, vol. 11, no. 1, pp. 37-44, 2018.

[22] M. Attar, O. Homaee, H. Falaghi and P. Siano, "A novel strategy for optimal placement of

locally controlled voltage regulators in traditional distribution systems", International

re-
Journal of Electrical Power & Energy Systems, vol. 96, pp. 11-22, 2018.

[23] I. Quadri, S. Bhowmick and D. Joshi, "A comprehensive technique for optimal allocation

of distributed energy resources in radial distribution systems", Applied Energy, vol. 211, pp.
lP
1245-1260, 2018.

[24] I. Ben Hamida, S. Salah, F. Msahli and M. Mimouni, "Optimal network reconfiguration

and renewable DG integration considering time sequence variation in load and

rna

DGs", Renewable Energy, vol. 121, pp. 66-80, 2018.

[25] S. ChithraDevi, L. Lakshminarasimman and R. Balamurugan, "Stud Krill herd Algorithm

for multiple DG placement and sizing in a radial distribution system", Engineering Science
Jou

and Technology, an International Journal, vol. 20, no. 2, pp. 748-759, 2017.

[26] Y. Li, B. Feng, G. Li, J. Qi, D. Zhao and Y. Mu, "Optimal distributed generation planning

in active distribution networks considering integration of energy storage", Applied Energy,

vol. 210, pp. 1073-1081, 2018.

[27] S. Wang, X. Liu, K. Wang, L. Wu and Y. Zhang, "Tracing harmonic contributions of

multiple distributed generations in distribution systems with uncertainty", International

 Journal Pre-proof

40

Journal of Electrical Power & Energy Systems, vol. 95, pp. 585-591, 2018.

[28] S. Das, D. Das and A. Patra, "Reconfiguration of distribution networks with optimal

of
placement of distributed generations in the presence of remote voltage controlled bus",

Renewable and Sustainable Energy Reviews, vol. 73, pp. 772-781, 2017.

pro
[29] A Mohamed Imran, M Kowsalya and D P Kothari, “A novel integration technique for

optimal network reconfiguration and distributed generation placement in power distribution

networks”, International Journal of Electrical Power & Energy Systems, Vol.63, pp.461-

472, 2014
re-
[30] S. Saremi, S. Mirjalili and A. Lewis, "Grasshopper Optimization Algorithm: Theory and

application", Advances in Engineering Software, vol. 105, pp. 30-47, 2017.

[31] J. Tra and D. Chandran, "Cuckoo and Levy Flights Algorithm Applied to Unit-Commitment
lP
Problem", International Journal of Advanced Research in Electrical, Electronics and

Instrumentation Engineering, vol. 03, no. 12, pp. 13670-13677, 2014.

[32] M. Suresh and E. Belwin, "Optimal DG placement for benefit maximization in distribution
rna

networks by using Dragonfly algorithm", Renewables: Wind, Water, and Solar, vol. 5, no. 1,

2018.

[33] M. Suresh and B. J., "Optimal Placement of DG Units for Loss Reduction in Distribution
Jou

Systems Using One Rank Cuckoo Search Algorithm", International Journal of Grid and

Distributed Computing, vol. 11, no. 1, pp. 37-44, 2018.

[34] MCV Suresh and Belwin Edward J, "Optimal Placement Of Distributed Generation In

Distribution Systems By Using Shuffled Frog Leaping Algorithm", ARPN Journal of

Engineering and Applied Sciences, Vol. 12, No. 3, pp.863-868, 2017

 Journal Pre-proof

41

[35] S. A. Taher and M. H. Karimi, "Optimal reconfiguration and DG allocation in balanced and

unbalanced distribution systems", Ain Shams Engineering Journal, Vol.5, No.3, pp.735–

of
749, 2014

[36] M. Moradi, M. Abedinie and H. Bagheri Tolabi, "Optimal multi-distributed generation

pro
location and capacity by Genetic Algorithms", 2010 Conference Proceedings IPEC,

Singapore, Singapore, pp. 614-618, 2010

[37] M. Moradi and M. Abedini, "A combination of genetic algorithm and particle swarm

optimization for optimal DG location and sizing in distribution systems", International

re-
Journal of Electrical Power & Energy Systems, Vol. 34, No. 1, pp. 66-74, 2012.
lP
rna
Jou
 Journal Pre-proof
*Credit Author Statement

A Hybrid Algorithm Based Optimal Placement of DG Units for Loss

Reduction in the Distribution System

of
Credit Author Statement

pro
M C V Suresh: Conceptualization, Methodology, Writing - Original Draft, Writing - Review &
Editing, Visualization
J. Belwin Edward: Supervision

re-
lP
rna
Jou