You are on page 1of 47

Accepted Manuscript

Title: DG placement in radial distribution network by


symbiotic organism search algorithm for real power loss
minimization

Author: Bikash Das V. Mukherjee Debapriya Das

PII: S1568-4946(16)30474-4
DOI: http://dx.doi.org/doi:10.1016/j.asoc.2016.09.015
Reference: ASOC 3814

To appear in: Applied Soft Computing

Received date: 7-8-2015


Revised date: 7-9-2016
Accepted date: 8-9-2016

Please cite this article as: Bikash Das, V.Mukherjee, Debapriya Das, DG
placement in radial distribution network by symbiotic organism search
algorithm for real power loss minimization, Applied Soft Computing Journal
http://dx.doi.org/10.1016/j.asoc.2016.09.015

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. 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.
DG placement in radial distribution network by symbiotic organism
search algorithm for real power loss minimization

Bikash Das1, V. Mukherjee2 and Debapriya Das3

1
Department of Electrical Engineering, Govt. College of Engineering and Textile Technology, Berhampore, West
Bengal, India
2
Department of Electrical Engineering, Indian School of Mines, Dhanbad, Jharkhand, India

3
Department of Electrical Engineering, Indian Institute of Technology, Kharagpur, West Bengal, India

2
Corresponding author. Tel.: +91 0326 2235644; Fax: +91 0326 2296563
E-mail addresses: bcazdas@gmail.com (Bikash Das)1, vivek_agamani@yahoo.com (V. Mukherjee)2,
ddas@ee.iitkgp.ernet.in (Debapriya Das)3.
Graphical abstract

0.335
PSO
0.3105 TLBO
0.33
0.31 CS
0.325 ABC
0.3095
FF GSA
0.32 0.309 SFS
SOS
0.3085
0.315
200 400 600 800

0.31

0.305
0 2000 4000 6000 8000 10000 12000
NFFEs

Fig. Comparative convergence profile of fitness function value offered by different algorithms for 33-bus
distribution network.

Research Highlights
 Symbiotic organism search (SOS) algorithm is used to find optimum size of DGs.
 Results obtained by using SOS are compared with other existing methods.
 Effectiveness of the SOS is demonstrated through benchmark function.
 It is also tested on two power system examples.
 SOS gives near global optimal value.
 Minimum NFFEs is required for SOS.

Abstract
Incorporation of distributed generation (DG) in distribution network may reduce the network loss if DG
of appropriate size is placed at proper strategic location. The current article presents determination of
optimal size and location of DG in radial distribution network (RDN) for the reduction of network loss
considering deterministic load demand and DG generation using symbiotic organism search (SOS)
algorithm. SOS algorithm is a meta-heuristic technique, inspired by the symbiotic relationship between
different biological species. In this paper, optimal size and location of DG are obtained for two different
RDNs (such as, 33-bus and 69-bus distribution networks). The obtained results, using the proposed SOS,
are compared to the results offered by some other optimization algorithms like particle swarm
optimization, teaching-learning based optimization, cuckoo search, artificial bee colony, gravitational
search algorithm and stochastic fractal search. The comparison is done based on minimum loss of the
distribution network as well as based on the convergence mobility of the fitness function offered by each
of the comparative algorithms for both the networks under consideration. It is established that the
proposed SOS algorithm offers better result as compared to other optimization algorithms under
consideration. The results are also compared to the existing solution available in the literature.

Keywords: Artificial bee colony (ABC); cuckoo search (CS); distributed generation (DG); particle
swarm optimization (PSO); symbiotic organism search (SOS). Reviewer 2, Comments 1

1. Introduction
Electrical power demand is gradually increasing. Minimization of network power loss while satisfying
the power demand has become a major concern. To supply electrical power and to reduce the network
losses, distributed generation (DG) plays an important role. DG can be defined as small or medium
electrical power generating system installed nearer to the load. DGs may be conventional or non-
conventional source of energy [1]. Examples of DGs are solar cell, wind turbine, microturbine, fuel cell,
diesel generator etc [2].
DG is gaining more and more importance due to its various benefits offered to the customers and to
the utility houses including reduction of network losses, increase in the reliability of the network and
improvement of voltage profile of the network [2]. There are, mainly, three types of DGs. These are (a)
real power DG or unity power factor DG (UPF-DG), (b) reactive power DG and (c) real and reactive
power DG or lagging power factor DG.
Proper placement of DG in the distribution network is a very vital and challenging task. If the DGs
are not properly placed in the distribution network, it may lead to increase of electric power losses instead
of reducing the same. So, the placement of DG at proper strategic location is a very important task [3]. At
the same time, optimum sizing of DG plays a crucial role to reduce the network losses. For that reason,
finding the optimum location and obtaining the appropriate sizing of DG have become an interesting
topic to the research community.
Application of meta-heuristic techniques for solving real-life problems has become an interesting
topic for the researchers. Genetic algorithm (GA) is a nature inspired algorithm. GA is based on the
natural selection process of biological evolution to obtain the global optimum solution. Performance of
GA depends on its parameter selection [4]. Particle swarm optimization (PSO) (proposed by Kennedy
and Eberhart [5]) is also a nature inspired algorithm that is based on the movement of particles (swarms)
for finding the best position in a defined domain. PSO is being adopted by different researchers to solve a
variety of real-life problems [6, 7]. Like GA, performance of PSO also depends upon the parameters of
the algorithm. In the literature, application of different algorithms like Tabu search [8, 9], harmony search
[10, 11], ant colony [12, 13], bat algorithm [14], plant growth simulation algorithm [15], big bang-big
crunch (BBBC) optimization [16] etc have been reported by many researchers. Cuckoo search (CS),
which is based on the behavior of cuckoo bird, has been also applied to obtain global minimum value of
different real-life problems [17-19]. Another nature inspired algorithm (inspired by the honey bee and
named as artificial bee colony (ABC)) has been also applied by few researchers to solve different
problems [20]. ABC algorithm is based on the behavior of honey bee and their movement to collect
honey from the flowers. Teaching learning based optimization (TLBO) algorithm (proposed by Rao et al.
[21, 22]) is an optimization technique that is inspired by the impact of a good teacher on the performance
of the learners/students. TLBO is an efficient optimization technique [21, 22]. Some hybrid techniques
like fuzzy-GA [23], GA-PSO [24], bacterial foraging algorithm with fuzzy logic [25] have been also
adopted in the literature. Virus colony search [26], flower pollination algorithms [27], gravitational
search algorithm (GSA) [28], stochastic fractal search (SFS) [29] etc may be noted as a few among those
applied for solving different problems. Some other meta-heuristic algorithms are also reported in the
literature in this direction [30-36].
Different approaches have been adopted by many researchers to identify proper location and to
determine optimum size of the DG to reduce the network losses as well as to the improve voltage profile
of the network. Singh et al. have surveyed the impact of DG in the power system (see [37]). Some of the
researchers have tried to place DGs to improve voltage stability. Analytical approach has been taken by
some researchers to find out proper strategic location of DGs to minimize the network losses [38].
Acharya et al. [39] have proposed an expression to determine the size of DG and have also discussed the
analytical method to identify the location of the DGs and to reduce the network loss by using
approximate loss formula. Some of the researchers have concentrated on the classical optimization
methods for placing the DGs in the distribution network [40, 41]. With the objective to find out the
optimum size and location of DG, meta-heuristic techniques have been also adopted by many researchers.
As DG placement is a nonlinear optimization problem, application of different optimization techniques
has become an important topic for the researchers. Nara et al. [42] have applied Tabu search algorithm to
find out the optimum location of DG in the distribution network aiming at reduction of the network loss.
GA has been adopted by Borges and Falcao [43] to determine the size of the DG on account of loss
reduction. Use of ant colony for the aforementioned purpose may be observed in [44]. PSO [45] and rank
evolutionary PSO are used to optimize DG size (refer [46]). Different hybrid methodologies have been
also adopted by some researchers for the same purpose [47-49]. In the same direction, a combined GA
and PSO based methodology have been proposed by Moradi and Abediniin [50]. Some of the researchers
have also tried to optimize the size and the location of DG considering some other objective functions.
Singh and Goswami [51] have considered nodal pricing considering profit as an objective function. Some
of the researchers have considered voltage stability as objective function (like [52, 53]). A novel
methodology has been used in [54] to integrate DG to the network under reconfiguration scenario. Singh
and Parida [55] have applied a new approach to optimize DG size considering the effect of DG to the
utility and to the customer. BBBC algorithm has been applied by Sadighizadeh et al. in [56] for
optimizing the DG size. Initiations have been taken by few researchers to place DG in the distribution
network considering the pricing aspect. Some of the researchers have also considered voltage
improvement of the network as an objective function while placing the DGs in the distribution network
(see [43, 51, 57, 58]).
Cheng and Prayogo have proposed symbiotic organisms search (SOS) algorithm in [59]. It is a nature
inspired evolutionary optimization algorithm that is based on interaction between different individuals in
the nature. Basically, three phases (viz. mutualism phase, commensalism phase and parasitism phase)
may be noted in the SOS algorithm. All these symbiotic relationships help the individuals to increase
their fitness and survival stability on the long term basis. One major advantage of SOS is that it needs a
very few control variables and it also doesn’t have algorithm specific parameters in comparison to other
optimization algorithms [59]. Reviewer 2, Comments 1
The objective of the present paper is to optimize the location and size of the DGs in a RDN focusing
on the reduction of the network power losses considering deterministic load demand and DG generation.
SOS is employed to determine the optimum location and size of the DGs. In the present work, a study is
carried out for the placement of UPF-DG in two different RDNs such as 33-bus and 69-bus distribution
networks. The results, yielded by the proposed SOS algorithm, are compared to those offered by some
algorithms recently reported in the recent state-of-the-art literatures like adaptive fuzzy PSO [60], TLBO
[21, 22], CS [61], ABC [20], GSA [28] and SFS [29]. Obtained results, yielded by SOS, are compared to
the results offered by the aforementioned algorithms based on number of fitness function evaluations
(NFFEs) and computational time taken to converge. Emphasis is also given in the present paper to study
the improvement of voltage profile of the considered RDNs after the placement of UPF-DG. The network
loss, after placement of DGs for both the distribution networks while using SOS, is compared to the
existing solutions reported in the recent state-of-the-art literatures to verify the solutions obtained in this
research work.
The rest of this paper is organized as follows: In the next Section, an overview of SOS algorithm is
presented. Section 3 presents a comparative study of the performance of different algorithms, considered
in this research work, to solve ten standard mathematical benchmark optimization functions. Section 4
presents problem formulation. Section 5 focuses on the optimum sizing of DG using SOS algorithm. The
results, after placement of DGs in the distribution network by using the proposed SOS algorithm, are
compared to the existing results in Section 6. Finally, the conclusions of the present work are drawn in
Section 7. In this section, future research direction is also included.

2. SOS: An overview
The word ‘symbiosis’ came from the Greek word, meaning “living together”. Symbiotic relationship
describes the relationship of any two different organisms. In the ecosystem, most of the organisms are
dependent on some other organisms and they are interlinked with each other for their survival [59]. The
interaction between the organisms may be beneficial for both the organisms or may be beneficial for any
one of them while the other organism remained either unaffected, or may be benefited, or may be
harmed. The most common symbiotic relationships found in the nature are (a) mutualism, (b)
commensalism and (c) parasitism [59]. Each organism interacts with some other randomly selected
organism. The aforementioned three relationships are described in the next three sub-sections, in order.

2.1 Mutualism phase


Mutualism phase [59] describes the relationship between two randomly selected organisms in the
ecosystem, when both the selected organisms get benefit from the relationship. The common example of
this phase is the relationship between the bees and the flowers. When the bees fly amongst the flowers,
they gather nectar to collect honey and the bees get benefit. At the same time, pollination has been
occurred by the bees in this process. So, the flowers also get benefit from this process. Thus, both the
organisms, interlinked in this process, become benefited.
Let us consider 𝑋𝑖 is an organism (matched to the ith member of the ecosystem) that interacts with
another randomly selected organism 𝑋𝑗 in the ecosystem. Both the organisms are engaged in mutualism
relationship with the objective to increase their mutual survival advantage in the ecosystem. The new
solution, after the mutualism phase for 𝑋𝑖 𝑛𝑒𝑤 and 𝑋𝑗 𝑛𝑒𝑤 , may be represented by (1) and (2),
respectively,
𝑋𝑖𝑛𝑒𝑤 = 𝑋𝑖 + 𝑟𝑎𝑛𝑑(0,1) × (𝑋𝑏𝑒𝑠𝑡 − 𝑀𝑉𝑒𝑐𝑡𝑜𝑟 × 𝐵𝐹1 ) (1)
𝑋𝑗𝑛𝑒𝑤 = 𝑋𝑗 + 𝑟𝑎𝑛𝑑(0,1) × (𝑋𝑏𝑒𝑠𝑡 − 𝑀𝑉𝑒𝑐𝑡𝑜𝑟 × 𝐵𝐹2 ) (2)
where 𝑀𝑉𝑒𝑐𝑡𝑜𝑟 is the mutual vector that represents the mutual relationship in between two organisms.
𝑀𝑉𝑒𝑐𝑡𝑜𝑟 may be defined by (3).
𝑋𝑖 + 𝑋𝑗
𝑀𝑉𝑒𝑐𝑡𝑜𝑟 = (3)
2

In the ecosystem, organisms engaged in mutualism relationship may not be equally benefited. Any
one may get better benefit from the other. In (1) and (2), 𝐵𝐹1 and 𝐵𝐹2 are the benefit factors (that
represent the level of benefit of each organisms if they are partially or fully benefitted) and are, randomly,
chosen as either 1 or 2. Let, 𝑋𝑏𝑒𝑠𝑡 be the best solution in the ecosystem. The updated solution will be
accepted if the updated solution has better fitness value than the previous one.

2.2 Commensalism phase


Commensalism phase [59] describes the relationship between two randomly selected organisms engaged
in a relation through which one organism gets benefit but the other remains unaffected. Example of
commensalism relationship is the relation of remora fish with sharks. Remora fishes, usually, attach
themselves with sharks and eat foods of the leftovers. Thus, remora fish receives benefit whereas sharks
remain unaffected by the activity of the remora fish.
In SOS, Xi organism interacts with a randomly selected organism 𝑋𝑗 . From this interaction, 𝑋𝑖 gets
benefit whereas 𝑋𝑗 gets neither benefit nor suffer. It remains unchanged by the activity of 𝑋𝑖 . The
commensalism relationship between 𝑋𝑖 and 𝑋𝑗 can be represented by (4) [59].
𝑋𝑖𝑛𝑒𝑤 = 𝑋𝑖 + 𝑟𝑎𝑛𝑑(−1,1) × (𝑋𝑏𝑒𝑠𝑡 − 𝑋𝑗 ) (4)
2.3 Parasitism phase
In parasitism phase [59], two randomly selected organisms interact with each other. From this interaction,
one of the selected organisms gets benefit while the other one is affected by the activity of the benefited
organism. Parasitism relationship can be explained with the example of plasmodium parasite which uses
its relationship with anopheles mosquito to pass between human host. Inside human body, the parasites
reproduce while the host (human), as a result, suffers from malaria and may even die.
Let, 𝑋𝑖 organism plays the same role as the anopheles mosquito through the creation of an artificial
parasite (called as parasite vector). The parasite vector is created by duplicating organism of 𝑋𝑖 in the
search space and then modifying randomly selected dimension by randomly selected number. Let, 𝑋𝑗 be
the randomly selected organism which serves as the host (like human body) to the parasite vector.
Parasite vector always tries to replace 𝑋𝑗 in the ecosystem. After measurement of the fitness of the two
organisms (𝑋𝑗 and parasite vector), the organism with better fitness will survive in the ecosystem while
the other with less fitness function value will be eliminated [59].
The flowchart of SOS is shown in Fig. 1. Fig. 1

3. Comparison study of different algorithms applied to solve benchmark functions


To analyze the performance of SOS algorithm, it is applied to solve ten standard benchmark functions
[28, 29, 59], as given in Table 1. For the purpose of relative analysis of the result obtained by using SOS,
it is compared with the performance obtained by using PSO, TLBO, CS, GSA and SFS. The parameters
of the employed algorithms are tabulated in Table 2. The performance analysis is done based on twenty
independent trial runs for each algorithm for each test function. The analysis of the performance of
different algorithms is done based on the result evaluation as well as based on the convergence mobility.
Based on the performance analysis, the algorithms are ranked. For the analysis perspective, it is
considered that the algorithms converge when the difference between optimum result of the function and
the obtained results reach below 1×10-5 [59]. The results obtained below 1×10-5 are considered as equal to
zero. The results obtained using different algorithms for the minimization functions are tabulated in Table
3. The convergence mobility of the algorithms is shown in Fig. 2. From the obtained results, as presented
in Table 3 and Fig. 2, it may be seen that SOS performs better than the other algorithms in both the
aspects of obtaining optimum result as well as offering faster convergence characteristic. SOS takes
lesser NFFEs to converge as compared to the other considered algorithms. The SOS algorithm is ranked
as 1 for all the ten minimization functions considered in this experimentation.
Tables 1-3 Fig. 2
In addition to the benchmark functions test, separate experiments are performed to analyze the
performance of different phases of SOS. As mentioned in Section 2, SOS is consisting of three different
phases (named as mutualism phase, commensalism phase and parasitism phase). All the three phases
work together to form the complete SOS algorithm. But to analyze the performance of different phases of
SOS, the studied ten benchmark functions (f1 – f10) are considered. Table 4 shows the performance of
SOS in different phases. It may be noticed from Table 4 that performance of only mutualism phase
improves when commensalism phase is incorporated with the mutualism phase for all the benchmark
functions considered in this work. The performance of the total SOS algorithm (mutualism phase +
commensalism phase + parasitism phase) is better than the performance of mutualism phase along with
commensalism phase. The convergence mobility gets improved when different phases are incorporated to
form the complete SOS algorithm.
The study is also carried out to analyze the performance contribution of all the three phases of SOS.
The analysis is carried out on ten benchmark functions considered in this research work. The
performance in this regard is depicted in Fig. 3. It may be revealed that the contribution of mutualism
phase to converge any function is greater than the other phases. Commensalism phase also contributes
significantly in SOS algorithm for all the benchmark functions. From Fig. 3, it may be observed that the
contribution of parasitism phase is lesser as compared to the other two phases of the algorithm for the
cases considered in this work. But all the three phases are simultaneously required to form the SOS
algorithm and all the three phases have their individual important role towards the convergence of the
algorithm. Reviewer 2, Comment 2 Fig. 3

4. Problem formulation and selection of bus


4.1 Problem formulation
The main objective of this work is to optimize the size and the placement of DG to be placed in the RDN
to reduce the active power loss of the distribution network considering deterministic load demand and
DG generation. Improvement of voltage profile of the network is not considered while optimizing the DG
size. The total active power loss may be defined as the sum of the individual branch losses. Total active
power loss may be defined by (5)
𝑃𝑙𝑜𝑠𝑠 = ∑𝑁𝑏 2
𝑖=1 𝑟𝑒𝑎𝑙((𝐼𝑖 ) × 𝑍𝑖 ) (5)
where 𝑃𝑙𝑜𝑠𝑠 is the total active power loss, 𝑁𝑏 is the number of branch, 𝐼𝑖 is the line current magnitude and
𝑍𝑖 is the line impedance.
When DG is connected to the distribution network, DG injects current to the bus where it is
𝑘
connected. DG is considered to be connected to bus k and injects a current of value 𝐼𝐷𝐺 . The branch
current of the kth branch may be represented by (6)
𝐼𝑘𝑛𝑒𝑤 = 𝐼𝑘 − 𝐴𝐼𝐷𝐺
𝑘
(6)
where A = 1; if DG is connected to the bus and
= 0; if DG is not connected to the bus.
The active power loss, after placement of DG, may be represented by (7).
𝐷𝐺
𝑃𝑙𝑜𝑠𝑠 = ∑𝑁𝑏 𝑛𝑒𝑤 2
𝑖=1 𝑟𝑒𝑎𝑙((𝐼𝑖 ) × 𝑍𝑖 ) (7)
The reduction of active power loss may be defined by (8).
𝑟𝑒𝑑𝑢𝑐𝑒𝑑 𝐷𝐺
𝑃𝑙𝑜𝑠𝑠 = 𝑃𝑙𝑜𝑠𝑠 − 𝑃𝑙𝑜𝑠𝑠 (8)
The fitness function (FF), which requires to be minimized, is defined in (9)
𝐷𝐺
𝑃𝑙𝑜𝑠𝑠
𝐹𝐹 = (9)
𝑃𝑙𝑜𝑠𝑠
𝐷𝐺
where 𝑃𝑙𝑜𝑠𝑠 is the active power loss of the network after the placement of the DG and 𝑃𝑙𝑜𝑠𝑠 is the active
power loss of the network without placement of DG.
The optimization problem of the present work is subjected to some equality and inequality constrains,
as stated in the next two sub-sections.

4.1.1 Equality constrains


The equality constrains may be stated by (10).
∑ 𝑃𝑔𝑗 + ∑ 𝑃𝐷𝐺𝑗 = ∑ 𝑃𝐷𝑗 + 𝑃𝑙𝑜𝑠𝑠 (10)

4.1.2 Inequality constrains


The inequality constrains may be described by (11) - (12).
(a) Voltage constrains: It is defined in (11).

𝑉𝑗𝑚𝑖𝑛 ≤ 𝑉𝑗 ≤ 𝑉𝑗𝑚𝑎𝑥 j= 1, 2, ……..,N (11)

(b) Current constrains: It is defined in (12).


𝐼𝑖 ≤ 𝐼𝑖𝑚𝑎𝑥 i=1, 2,……… Nb (12)
Here, Pgj , PD j are the active power generation and active power demand of the jth bus, respectively;
PDG j is the injected active power by the DG connected to bus j and Ploss is the active power loss of the
network. In (11), Vj is the bus voltage of the jth bus of the network having N number of buses and 𝑉𝑗𝑚𝑖𝑛 ,
𝑉𝑗𝑚𝑎𝑥 are the minimum and the maximum limit of the bus voltage. In (12), Ii, represents the branch
current of the ith branch of the network consisting of Nb number of branches. 𝐼𝑖𝑚𝑎𝑥 is the maximum
permissible current flows through the ith branch.

4.2 Location of DG: a concept


For reduction of power loss of the network, choose of proper DG location is very important. In this work,
location of the DG is chosen by finding the most sensitive bus in the network. The study is carried out for
the placement of three DGs for both the networks under consideration. The location of DG is determined
by applying loss sensitivity analysis and choosing the most sensitive bus, one at a time. Then, after
placing DG at that chosen bus, next sensitive bus is determined. Once again, the process is repeated to
determine the next sensitive bus. In this way, three most loss sensitive buses are determined in the
network for the placement of DGs in the respective RDN to reduce the network losses. The procedure is
carried out to determine three most loss sensitive buses for both 33-bus as well as for 69-bus RDNs.
By applying the above mentioned procedure to both the networks under consideration, three most
sensitive buses are determined for the placement of DGs. These buses, for 33-bus distribution network,
are bus numbers 14, 24 and 30. Bus numbers 11, 18 and 61 are selected as the three sensitive buses for
the chosen 69-bus distribution network.

5. Optimum sizing of DG using SOS


Minimization of network loss can be done by optimizing the size of the DGs. For determining the
optimum size of the DGs, different methods have been adopted by different researchers. The present
work concentrates on the placement of active power DGs only. In this work, the optimum size of DGs is
determined by using the proposed SOS algorithm. The results obtained by using the SOS are compared to
the results yielded by using PSO, TLBO, CS, ABC, GSA and SFS for both 33-bus and 69-bus
distribution networks. The simulation is run on MATLAB 12b in a PC with Intel Core I5 (4th Gen)
processor having speed of 1.70 GHz with turbo boost up to 2.40 GHz and 8 GB of internal RAM. The
results are analyzed based on ten simulation runs for each case. The parameters of PSO and CS are
selected from parametric sensitivity analysis, considering FF of 33-bus distribution network as the test
function and the obtained results are tabulated in Tables 5 and 6, respectively. The parameters of PSO
(refer Table 5) are analyzed considering wmin = 0.4. It is found that PSO gives the best result for this study
at wmin = 0.4, wmax = 0.9 and C1 = C2 = 2. From Table 6, it may be noted that minimum FF for 33-bus
distribution network is achieved corresponding to Levy coefficient, λ = 0.5 and discovery rate of alien
egg (termed as Pa) as 0.25. For fair comparison of the results offered by each of the optimization
algorithms under consideration, the population size is considered to be same for each of the algorithms.
The population size of 30 is considered and the simulation is run for 12000 NFFEs for all the considered
algorithms. The parameters of the different adopted algorithms are tabulated in Table 7. The
performances are analyzed by using Newton-Raphson method of load flow analysis. Results of interest
are bold faced in the respective tables. Tables
Tables 4-6
5-7

5.1 DG size optimization for 33-bus distribution network


The first RDN under consideration is 33-bus distribution network with total active power load of 3715
kW [54]. The single-line diagram of this distribution network is shown in Fig. 4. Without any DG in the
network, the active power loss is 202.6771 kW. For placement of DGs, three most sensitive buses are
selected. These bus numbers are 14, 24 and 30. Table 8 shows the results obtained by using PSO, TLBO,
CS, ABC, GSA, SFS and the proposed SOS algorithm. From Table 8 it may be observed that all the
optimization algorithms considered in this work offer the same result except ABC. Fig. 5 shows the
comparative convergence of the fitness function to find out the optimum DG size for 33-bus distribution
network by using PSO, TLBO, CS, ABC, GSA, SFS and SOS. From Fig. 5, it may be noted that the
proposed SOS algorithm takes less NFFEs (1080 only) to converge. Fig. 6 shows the DG size
optimization results offered by SOS for this distribution network. Table 8
Figs. 4-6
From Table 8 and Fig. 7, it may be observed that the best as well as the average NFFEs taken to
converge by SOS are lesser as compared to the other adopted algorithms. The best and the average
NFFEs taken to converge by SOS are 1080 and 1224, respectively. Fig. 7
After optimizing the size of the DGs by using the proposed SOS algorithm and placing them at the
selected buses of the chosen distribution network, the active power loss is found to be reduced from
202.6771 kW (base case) to 71.4572 kW. The voltage profile of the network is also improved. From Fig.
8, a significant improvement in voltage profile may be noticed after the placement of DGs. Before the
placement of DGs in the 33-bus distribution network, the minimum voltage was at bus number 18 (i.e.
0.9131 p.u.). But after the placement of DGs, the minimum voltage is improved to 0.9687 p.u. (at bus
number 33). After placement of DG while adopting the proposed SOS algorithm, it is imperative to note
that the network loss is reduced and the voltage profile of the network is also improved at the same time.

Fig. 8

5.2 DG size optimization for 69-bus distribution network


The next RDN under consideration is 69-bus distribution network with active and reactive power load
demand of 3802.2 kW and 2694.6 kVAr [3], respectively. The total active power loss of the network
without any DG is 225.0028 kW. Fig. 9 portrays the single-line diagram of this distribution network.
Applying loss sensitivity analysis, three buses are selected for placing the DGs. These selected buses are
bus numbers 61, 18 and 11. The objective for placing the DGs at the respective buses is to reduce the
network active power loss. To determine the optimum DG size, all the seven considered optimization
techniques (viz. PSO, TLBO, CS, ABC, GSA, SFS and the proposed SOS) are applied. The results
offered by these algorithms are tabulated in Table 9. Table 9 Fig. 9
The adopted seven optimization algorithms offer the same minimum loss (i.e. 69.4286 kW) for this
distribution network, except ABC. The same for the ABC algorithm is 69.4322 kW. The comparative
convergence profile of the fitness function value, as offered by PSO, TLBO, CS, ABC, GSA, SFS and the
proposed SOS algorithm, is illustrated in Fig. 10. Comparing the nature of the convergence profile of the
fitness function value as depicted in Fig. 10, it may be noted that the proposed SOS algorithm takes less
number of NFFEs to converge to the near global minimum value. The optimization results of DG by
using SOS algorithm are shown in Fig. 11. From Fig. 11 and Table 9, it may be divulged that the size of
the DGs for bus numbers 61, 18 and 11 of the network under consideration are 1719.00 kW, 380.5 kW
and 526.70 kW, respectively. Figs. 10-11
The average NFFEs, taken by each of the adopted algorithm, are obtained for this distribution
network and are tabulated in Table 9. The best and the average NFFEs taken to converge for the 69-bus
distribution network are portrayed in the Figs. 12(a) and 12(b), respectively. From Figs. 12 (a) and 12 (b),
it may be clearly observed that the SOS algorithm takes lesser NFFEs to converge than the other
algorithms employed in this work. The best result obtained for 69-bus distribution network, is achieved
by using SOS, which takes only 960 NFFEs to converge.
Fig. 12
After the placement of DGs at the selected buses (bus numbers 61, 18 and 11) of the studied
distribution network, the network loss is found to be reduced from the base case of 225.0028 kW to
69.4286 kW while DG size is optimized by the proposed SOS algorithm. A significant improvement in
voltage profile (refer Fig. 13) may also be noticed after the placement of DGs in the network. From Fig.
13 it may be seen that before placement of the DGs, the minimum voltage was at bus number 65 (0.9092
p.u.) but after placement of DGs at the selected buses, the minimum voltage improves to 0.9790 p.u. at
bus number 65. After placement of DGs, optimized by the proposed SOS algorithm at the selected buses,
the active power loss of the network is reduced and the voltage profile of the network is also improved.
Fig. 13
Thus, while applying PSO, TLBO, CS, ABC, GSA, SFS and SOS to optimize the size of DGs for
both 33-bus and 69-bus distribution networks with an aim to reduce the network active power loss, it may
be noted that the proposed SOS gives optimum result within less number of NFFEs for both the
networks. The time taken by SOS is also lesser as compared to other algorithms (viz. PSO, TLBO, CS,
ABC, GSA and SFS).

6. Comparison of the results with existing results


The results obtained in this research work for both the 33- and 69-bus distribution networks are compared
with different existing results. Some of the researchers have tried to optimize the size and location for
placement of single DG to minimize the network loss. On the other hand, few researchers have tried to
place multiple DGs to minimize the network loss. The results, obtained in this work to optimize the
location and size of three DGs to be placed in both the networks under consideration using the proposed
SOS algorithm to minimize the network loss, are compared to the existing results and are tabulated in
Tables 10 and 11. Table 10 shows the comparison of the results obtained in this work with the existing
results for 33-bus distribution network while the same for 69-bus distribution network is included in
Table 11. From these two tables, it may be seen that the location and size of DGs, as obtained by using
the proposed SOS algorithm, gives minimum network loss as compared to the existing results for both
33- and 69-bus distribution networks. Tables 10-11

7. Conclusion and future direction Reviewer 2, Comments 1


In this work, placement issue of DG is discussed considering deterministic load demand and DG
generation. The optimum location and sizing of DG for 33-bus and 69-bus distribution networks are
determined taking into account the reduction of active power loss of the network as the prime
concern/objective. Apart from this, to analyze the performance of SOS, it is applied to solve ten standard
benchmark functions. From the performance comparison study of the SOS with other five algorithms like
PSO, TLBO, CS, GSA, SFS, it is found that the performance of SOS is better as compared to the other
five algorithms. The location and sizing of DGs are calculated by using seven optimization algorithms
like PSO, TLBO, CS, ABC, GSA, SFS and SOS. From the study, it is seen that DG size can be optimized
to reduce the network loss for both the networks under consideration. This study also reveals that PSO,
TLBO, CS, GSA, SFS and SOS converge faster than ABC. The convergence of SOS is better than other
algorithms, as used in this work. SOS may take little bit higher CPU time to converge as compared to
some other algorithms but it takes lesser NFFEs to converge than the other algorithms used for both the
studied cases of this work. The average NFFEs taken to converge is also lesser for SOS as compared to
the other algorithms adopted in this work such as PSO, TLBO, CS, ABC, GSA and SFS.
After optimizing the location and size of DGs and placing the same in the network, a significant
reduction in active power loss is noticed for both the considered networks. Improvement in voltage
profile for both the networks is also noticed after placement of the DGs in the network.
From the study, it may be concluded that SOS may be used to optimize the size and location of DGs
in RDN to reduce the network loss and it takes less number of NFFEs to converge to the desired
results.SOS takes higher computational time per iteration than the other algorithms used in this work,
despite of that, SOS takes lesser average computation time to converge than the algorithms like PSO,
TLBO, CS, ABC, GSA and SFS for both the cases studied in this work. It may also be concluded that
SOS takes lesser time to converge than PSO, TLBO, CS, ABC, GSA and SFS for both the cases
considered in this work.
The main focus of the present work is to apply SOS algorithm to optimize the DG allocation and
compare it with the results obtained by other algorithms. However, it is worth to mention that for
uncertainties of load and generation, the proposed SOS algorithm may also be applied by the future
researchers for DG allocation problem.

Acknowledgment
The authors would like to thank the editor-in-chief and the anonymous reviewers for their insightful
comments and suggestions for improving the quality of the paper.
References
[1] B. Das, D. Das, Dynamic performances of split-shaft microturbine generator (MTG) and diesel
generator as distributed energy resources, Int. J. Innovative Research Electr. Electron. Instrument.
Control Engin. 2 (10) (2014) 2031-2037.
[2] B. Das, D. Das, Dynamic performances of split-shaft microturbine generator (MTG) system in
stand-alone mode and when connected to a rural distribution network, Distributed Gener. Alternat.
Energy J. 29 (4) (2014) 25-48.
[3] J.A.M. Garcia, A.J.G. Mena, Optimal distributed generation location and sizing using a modified
teaching-learning based optimization algorithm, Int. J. Electr. Power Energy Syst. 50 (2013) 65-75.
[4] K.S. Swarup, Genetic algorithm for optimal capacitor allocation in radial distribution systems,
Proc. 6th WSEAS Int. Conf. Evolutionary Computing (2005) 152-159.
[5] J. Kennedy, R. Eberhart, Particle swarm optimization, Proc. IEEE Int. Conf. Neural Networks
(1995) 1942-1948.
[6] J. Kennedy, The particle swarm: social adaptation of knowledge, Proc. IEEE Int. Conf.
Evolutionary Computation (1997) 303-308.
[7] K. Prakash, M. Sydulu, Particle swarm optimization based capacitor placement on radial
distribution systems, IEEE Power Engineering Society General Meeting (2007) 1-5.
[8] F. Glover, Tabu search part 1, ORSA J. Computing 1 (3) (1989) 190-206.
[9] F. Glover, Tabu search part 2, ORSA J. Computing 2 (1) (1990) 4-32.
[10] K.S. Lee, Z.W. Geem, A new meta-heuristic algorithm for continuous engineering optimization:
harmony search theory and practice, Computer Methods in Appl. Mechanics Engin.194 (36-38)
(2005) 3902-3933.
[11] S. Das, A. Mukhopadhyay, A. Roy, A. Abraham, B.K. Panigrahi, Exploratory power of the
harmony search algorithm: analysis and improvements for global numerical optimization, IEEE
Trans. Systems Man Cybernetics Part B: Cybernetics 41 (1) (2011) 89-106.
[12] M. Dorigo, L.M. Gambardella, Ant colony system: a cooperative learning approach to the traveling
salesman problem, IEEE Trans. Evolutionary Computation 1(1) (1997) 53-66.
[13] X. Hu, J. Zhang, Y. Li, Orthogonal methods based ant colony search for solving continuous
optimization problems, J. Computer Sci. Tech. 23(1) (2008) 2-18.
[14] X.S. Yang, Bat algorithm for multi-objective optimization, Int. J. Bio-Inspired Comput. 3 (5)
(2011) 267-274.
[15] R.S. Rao, S.V.I. Narasimham, M. Ramalingaeaju, Optimal capacitor placement in a radial
distribution system using plant growth simulation algorithm, Int. J. Electr. Power Energy Syst. 33
(5) (2011) 1133-1139.
[16] O.K. Erol, I. Eksin, A new optimization method: big bang-big crunch, Advances Engin. Software
37 (2) (2006) 106-111.
[17] X.S. Yang, S. Deb, Cuckoo search via Lévy flights, World Congress Nature & Biologically
Inspired Computing (NaBIC) (2009) 210-214.
[18] X.S. Yang, S. Deb, Engineering optimization by cuckoo search, Int. J. Mathe. Modelling
Numerical Optimization 1 (4) (2010) 330-343.
[19] A.A. El-Fergany, A.Y. Abdelaziz, Capacitor allocations in radial distribution networks using
cuckoo search algorithm, IET Gener. Transm. Distrib. 8 (2) (2014) 223-232.
[20] F. Kang, J. Li, H. Li, Artificial bee colony algorithm and pattern search hybridized for global
optimization, Appl. Soft Comput. 13 (4) (2013) 1781-1791.
[21] R.V. Rao, V.J. Savsani, D.P. Vakharia, Teaching- learning based optimization: a novel method for
constrained mechanical design optimization problems, Comput. Aided Design 43 (3) (2011) 303-
315.
[22] R.V. Rao, V. Patel, Comparative performance of an elitist teaching-learning based optimization
algorithm for solving unconstrained optimization problems, Int. J. Industrial Engin. Computations
4 (1) (2013) 29-50.
[23] Y.T. Hsiao, C.H. Chen, C.C. Chien, Optimal capacitor placement in distribution systems using
combination fuzzy-GA method, Int. J. Electr. Power Energy Syst. 26 (7) (2004) 501-508.
[24] Y. Kao, E. Zahara, A hybrid genetic algorithm and particle swarm optimization for multimodel
functions, Appl. Soft Comput. 8 (2) (2008) 849-857.
[25] S.M. Tabatabaei, B. Vahidi, Bacterial foraging solution based fuzzy logic decision for optimal
capacitor allocation in radial distribution system, Electr. Power Syst. Res. 81 (4) (2011) 1045-
1050.
[26] M.D. Li, H. Zhao, X.W. Weng, T. Han, A novel nature-inspired algorithm for optimization: virus
colony search, Advances Engin. Software 92 (2016) 65-88.
[27] X.-S. Yang, Flower pollination algorithm for global optimization, Unconventional Comput.
Natural Comput. 7445 (2012) 240-249.
[28] Esmat Rashedi, Hossein Nezamabadi-pour, Saeid Saryazdi, GSA: A gravitational search
algorithm, Inform. Sci. 179 (2009) 2232-2248.
[29] Hamid Salimi, Stochastic fractal search: A powerful metaheuristic algorithm, Knowledge-Based
Systems 75 (2015) 1-18.
[30] Alireza Askarzadeh, A novel metaheuristic method for solving constained engineering
optimization problems: Crow search algorithm, Computers and structures 169 (2016) 1-12.
[31] Yacine Labbi, Djilani Ben Attous, Hossam A. Gabbar, Belkacem Mehdad, Aboelsood Zidan , A
novel rooted tree optimization algorithm for economic dispatch with valve-point effect, Int. J. Elec.
Power and energy system 79 (2016) 298-311.
[32] Ali Osman Topal, oguz Altum, A novel meta-heuristic algorithm: Dynamic virtual bats algorithm,
Information science 354 (2016) 222-235.
[33] Aparajita Mukherjee, V. Mukherjee, Chaotic Krill Hard algorithm for optimal reactive power
dispatch considering FACTS devices, Appl. Soft Comp. 44 (2016) 163-190.
[34] M. P. Saka, O. Hasancebi, Z. W. Geem, Metaheuristic in structural optimization and discussions
on harmony search algorithm, Swarm and Evolu. Comp. 28 (2016) 88-97.
[35] Seyadail Mirjalili, Andraw Lewis, The whale optimization algorithm, Adv. in Engg. Soft. 95
(2016) 51-67.
[36] J Alikhani Koupaei, S. M. M. Hosseini, F. M. Mablek Ghaini, A new optimization algorithm based
on chaotic maps and golden section search method, Engg. Appl. Artificial Intelligence, 50 (2016)
2001-2014.
[37] B. Singh, V. Mukherjee, P. Tiwari, A survey on impact assessment of DG and FACTS controllers
in power systems, Renew. Sustain. Energy Rev. 42 (2015) 846-882.
[38] C. Wang, M.H. Nehrir, Analytical approaches for optimal placement of distributed generation
sources in power system, IEEE Trans Power System 19 (4) (2004) 2068-2076.
[39] N. Acharya, P. Mahat, N. Mithulananthan, An analytical approach for DG allocation in primary
distribution network, Int. J. Electr. Power Energy Syst. 28 (10) (2006) 669-678.
[40] A. Keane, M. O’Malley, Optimal allocation of embedded generation on distribution networks,
IEEE Trans. Power Syst. 20 (3) (2005) 1640-1646.
[41] D. Gautam, N. Mithulananthan, Optimal DG placement in deregulated electricity market, Electr.
Power Syst. Res. 77 (12) (2007) 1627-1636.
[42] K.Nara, Y. Hayashi, K. Ikeda, T. Ashizawa, Application of tabu Search to optimal placement of
distributed generators, Proc. IEEE Power Engin. Society Winter Meeting. 2 (2001) 918-923.
[43] C.L.T. Borges, D.M. Falcao, Optimal distributed generation allocation for reliability, losses and
voltage improvement, Int. J. Electr. Power Energy Syst. 28 (6) (2006) 413-420.
[44] H. Falaghi, M.R. Haghifam, ACO based algorithm for distributed generation sources allocation
and sizing in distribution systems, Proc. IEEE Power Tech. (2007) 555-560.
[45] S. Devi, M. Geethanjali, Optimal location and sizing determination of distributed generation and
DSTATCOM using Particle Swarm Optimization algorithm, Int. J. Electr. Power Energy Syst. 62
(2014) 562-570.
[46] J.J. Jamian, M.W. Mustafa, H. Mokhlis, Optimal multiple distributed generation output through
rank evolutionary particle swarm optimization, Neurocomput. 152 (2015) 190-198.
[47] G. Celli , E. Ghiani, S. Mocci, F. Pilo, A multiobjective evolutionary algorithm for the sizing and
sitting of distributed generation, IEEE Trans. Power Syst. 20 (2) (2005) 750-757.
[48] M. Gandomkar, M. Vikilian, M.A. Ehsan, A genetic-based Tabu search algorithm for optimal DG
allocation in distribution systems, Electr. Power Compon. Syst. 33 (12) (2005) 1351-1362.
[49] M. Gomez-Gonzalez, A. Lopez, F. Jurado, Optimization of distributed generation systems using a
new discrete PSO and OPF, Electr. Power Syst. Res. 84 (1) (2012) 174-180.
[50] M.H. Moradi, M. Abedini, A combination of genetic algorithm and particle swarm optimization
for optimal DG location and sizing in distribution systems, Int. J. Electr. Power Energy Syst. 34 (1)
(2012) 66-74.
[51] R.K. Singh, S.K. Goswami, Optimum allocation of distributed generations based on nodal pricing
for profit, loss reduction, and voltage improvement including voltage rise issue, Int. J. Electr.
Power Energy Syst. 32 (6) (2010) 637-644.
[52] M. Ettchadi, H. Ghasemi, S. Vaez-Zedah, Voltage stability-based DG placement in distribution
network, IEEE Trans. Power Delivery 28 (1) (2013) 171-178.
[53] M. Nayeripour, E. Mahboubi-Moghaddam, J. Aghaei, A. Azizi-Vahad, Multi objective placement
and sizing of DG in distribution networks ensuring transient stability using hybrid evolutionary
algorithm, Renew. Sustain. Energy Rev. 25 (C) (2013) 759-767.
[54] A. M. Imran, M. Kowsalya, D.P. Kothari, A novel integration technique for optimal network
reconfiguration and distributed generation placement in power distribution networks, Int. J. Electr.
Power Energy Syst. 63 (2014) 461-472.
[55] A.K. Singh, S.K. Parida, Allocation of distributed generation using proposed DMSP approach
based on utility and consumers aspects under deregulated environment, Int. J. Electr. Power
Energy Syst. 68 (2015) 159-169.
[56] M. Sadighizadeh, M. Esmaili, M. Esmaili, Application of the hybrid Big Bang-Big Crunch
algorithm to optimal reconfiguration and distributed generation power allocation in distributed
systems, Energy 76 (2014) 920-930.
[57] A.J.G. Mena, J.A.M. Garcia, An efficient approach for the siting and sizing problem of Distributed
Generation, Int. J. Electr. Power Energy Syst. 69 (2015) 167-172.
[58] E. Karatepe, F. Ugrandi, T. Hiyama, Comparison of single and multiple-distributed generation
concepts in terms of power loss, voltage profile, and line flows under uncertainty scenarios,
Renew. Sustain. Energy Rev. 48 (2015) 317-327.
[59] M.Y. Cheng, D. Proyogo, Symbiotic organisms search: a new metaheuristic optimization
algorithm, Computers and Struct.139 (2014) 98-112.
[60] Y.T. Juang, S.L. Tung, H.C. Chiu, Adaptive fuzzy particle swarm optimization for global
optimization of multimodal functions, Inform. Sci. 181 (20) (2011) 4539-4549.
[61] Z. Moravej, A. Akhlaghi, A novel approach based on cuckoo search for DG allocation in
distribution network, Int. J. Electr. Power Energy Syst. 44 (1) (2013) 672-679.
[62] M.M. Aman, G.B. Jasmon, A.H.A. Bakar, H. Mokhlis, A new approach for optimum DG
placement and sizing based on voltage stability maximization and minimization of power losses,
Energy Convers. Management 70(2013) 202-210.
[63] T. Gozel, M.H. Hocaoglu, An analytical method for the sizing and siting of distributed generators
in radial systems, Electr. Power Syst. Res. 79 (6) (2009) 912-918.
[64] T. Gozel, U. Eminoglu, M.H. Hocaoglu, A tool for voltage stability and optimization (VS&OP) in
radial distribution systems using matlab graphical user interface (GUI), Simul. Modell. Practice
Theory 16 (5) (2008) 505-518.
[65] M.M. Aman, G.B. Jasmon, A.H.A. Bakar, H. Mokhlis, A new approach for optimum simultaneous
multi-DG distributed generation units placement and sizing based on maximization of system
loadability using HPSO (hybrid particle swarm optimization) algorithm, Energy 66 (2014) 202-
215.
[66] M.H. Moradi, S.M. R. Tousi, M. Abedini, Multi-objective PFDE algorithm for solving the optimal
siting and sizing problem of multiple DG sources, Int. J. Electr. Power Energy Syst. 56 (2014)
117-126.
[67] S. Devi, M. Geethanjali, Application of modified bacterial foraging optimization algorithm for
optimal placement and sizing of distributed generation, Expert Syst. Applications 41 (6) (2014)
2772-2781.
List of Figures

Initialize the ecosystem and termination criterion

Identify the best organism

Randomly select one organism X j , where X j  X i


Mutualism phase

Modify organisms X i and X j based on their mutual relationship

Calculate fitness values of modified organisms

Yes
Are the modified organisms fitter
than previous?
Keep the previous Accept the modified
organisms organisms

Randomly select one organism X j , where X j  X i


i=i+1
Commensalism

Modify organism X i with the help of X j and


phase

calculate the fitness value of modified organism

No Yes
Is the modified organism fitter
t=t+1 than the previous?

Keep the previous Accept the modified


organism organism

Randomly select one organism X j , where X j  X i


Parasitism phase

Create a parasite (Parasite_Vector) from organism


Xi

No Is Parasite_Vector Yes
fitter than X j ?
Keep X j and remove Replace X j with Parasite_Vector
Parasite_Vector

i = Np?
No
Yes
No
Is termination criterion fulfilled?

Yes
Display optimal
result
solutionsolution

Fig. 1. Flowchart of the SOS algorithm.


20
10

-5
0
10
10
log( f (x) )

1000 2000 3000


PSO
TLBO
-20
10 CS
SFS
SOS
GSA
-40
10
0 2 4 6 8 10
NFFEs x 10
4

(a) (b)
200
10
30
0

15 PSO 10
100
25 TLBO 10
CS
10 GSA 0.5 1 1.5 2
20 10
0
4
log ( f (x) )

SFS x 10

SOS
f (x)

15 -100
5 10
PSO
10 TLBO
10
-200 CS
200 400 600 800 1000 GSA
5 SFS
-300 SOS
10
0 0 2 4 6 8 10
0 2000 4000 6000 8000 10000 NFFEs 5
NFFEs x 10

(c) (d)

100 10
10 10
PSO
TLBO
0
0 10 CS
10
GSA
SFS
-10 SOS
log ( f (x))

10
log ( f(x))

-100
10

PSO -20
10
TLBO
-200
10 CS
GSA
-30
SFS 10
SOS
-300
10
-40
0 2 4 6 8 10 10
0 2 4 6 8 10
NFFEs 5
x 10 NFFEs x 10
5

(e) (f)
5
10
0
10
0
10

PSO
-100
10 -5 TLBO
log( f (x) )

10

log( f(x) )
CS
GSA
SFS
-10
10 SOS
-200
PSO
10 TLBO
CS -15
10
GSA
SFS
-300 SOS -20
10 10
0 2 4 6 8 0 2 4 6 8 10
NFFEs 5 NFFEs 5
x 10
x 10

(g) (h)

5
0
10
10

0
10
-50
PSO
10 TLBO
log( f (x) )

-5 CS
10
log( f(x) )

GSA
-100 PSO SFS
10 -10 SOS
TLBO 10
CS
GSA
-150 -15
10 SFS 10
SOS
-20
0 2 4 6 8 10 10
0 2 4 6 8 10
NFFEs x 10
5
NFFEs 5
x 10

(i) (j)
Fig. 2. Results of benchmark functions (a) 𝑓1 (𝑥), (b) 𝑓2 (𝑥) , (c) 𝑓3 (𝑥) , (d) 𝑓4 (𝑥), (e) 𝑓5 (𝑥), (f) 𝑓6 (𝑥) ,
(g) 𝑓7 (𝑥) , (h) 𝑓8 (𝑥),(i) 𝑓9 (𝑥) and (j) 𝑓10 (𝑥).

Reviewer 2, Comment 4
20
10 -0.65
Mutualism phase Mutualism phase
Commensalism phase -0.7 Commensalism phase
* Parasitism phase * Parasitism phase
-0.75
0 SOS
10 SOS
log (f (x))

-0.8
-1.02

f (x)
-0.85
-20 -0.9 -1.03
10
-0.95 -1.04
-1 2000 4000 6000
-40
10 -1.05
0 0.5 1 1.5 2 2.5 0 2 4 6 8 10
NFFEs x 10
4 NFFEs x 10
4

(a) (b)

100
70 10
Mutualism phase Mutualism phase
60 Commensalism phase Commensalism phase
0
* Parasitism phase 10 * Parasitism phase
50 SOS SOS
log ( f (x))

-100
40 10
f (x)

30
-200
10
20

10 -300
10
0 0 2 4 6 8 10
0 2 4 6 8 10 5
NFFEs 4 NFFEs x 10
x 10

(c) (d)
100 10
10 10
Mutualism phase Mutualism phase
Commensalism phase 0 Commensalism phase
0
* Parasitism phase 10
10 * Parasitism phase
SOS SOS
log ( f (x) )
log ( f (x) )

-10
-100
10
10
-20
10
-200
10
-30
10
-300
10 -40
0 2 4 6 8 10
10
0 0.5 1 1.5 2
NFFEs x 10
5
NFFEs 5
x 10

(e) (f)
100 5
10 10
Mutualism phase Mutualism phase
Commensalism phase 0
Commensalism phase
0 10 * Parasitism phase
10 * Parasitism phase
SOS SOS
log ( f (x))

-5

log ( f (x))
10
-100
10
-10
10
-200
10 -15
10
-300
10 -20
10
0 1 2 3 4 5 0 1 2 3 4 5 6
5 NFFEs x 10
4
NFFEs x 10

(g) (h)
50 5
10 10
Mutualism phase Mutualism phase
0 Commensalism phase Commensalism phase
10 * Parasitism phase 0
10 * Parasitism phase
SOS SOS
log ( f (x) )

-50
log ( f (x))
10
-5
10
-100
10

-10
-150 10
10

-200 -15
10 10
0 2 4 6 8 10 0 2 4 6 8 10
NFFEs x 10
5
NFFEs 5
x 10
(i) (j)
Fig. 3. Performance of different phases of SOS for benchmark functions (a) 𝑓1 (𝑥), (b) 𝑓2 (𝑥) , (c) 𝑓3 (𝑥) , (d) 𝑓4 (𝑥),
(e) 𝑓5 (𝑥), (f) 𝑓6 (𝑥) , (g) 𝑓7 (𝑥) , (h) 𝑓8 (𝑥),(i) 𝑓9 (𝑥) and (j) 𝑓10 (𝑥).
Reviewer 2, Comment 2

Fig. 4. Single-line diagram of 33-bus distribution network.


0.38
PSO
0.375 0.355 TLBO
CS
0.354 ABC
0.37
GSA
0.353
SFS
FF

0.365 SOS
0.352

0.36 500 1000 1500

0.355

0.35
0 2000 4000 6000 8000 10000 12000
NFFEs

Fig. 5. Comparative convergence profile of fitness function value offered by different algorithms for 33-
bus distribution network.
1300
Size of DG (kW) at bus no. 30
1200 Size of DG (kW) at bus no. 24
Size of DG (kW) at bus no. 14
Size of DG (kW)

1100

1000

900

800

700

600
0 2000 4000 6000 8000 10000 12000
NFFEs
Fig. 6.Profile of DG size optimization for 33-bus distribution network using SOS.
4500
3960
4000
3500
3000
NFFEs 2500
2000 1680 1680 1620
1500 1200 1170 1080
1000
500
0
PSO TLBO CS ABC GSA SFS SOS
Algorithms

(a)

10000
8901
9000
8000
7000
6000
NFFEs

5000
4000
3000
2052
2000 1404 1368 1534 1417 1224
1000
0
PSO TLBO CS ABC GSA SFS SOS
Algorithms

(b)
Fig. 7. NFFEs taken to converge by different algorithms for 33-bus distribution network (a) minimum
NFFEs and (b) average NFFEs.
Fig. 8. Comparative voltage profiles of 33-bus distribution network before and after placement of DGs.
Fig. 9. Single-line diagram of 69-bus distribution network.
0.335
PSO
0.3105 TLBO
0.33
0.31 CS
0.325 ABC
0.3095
GSA
FF

0.32 0.309 SFS


SOS
0.3085
0.315
200 400 600 800

0.31

0.305
0 2000 4000 6000 8000 10000 12000
NFFEs

Fig. 10. Comparative convergence profile of fitness function value offered by different algorithms for 69-
bus distribution network.
2000
Size of DG (kW)

1500 Size of DG (kW) at bus number 61


Size of DG (kW) at bus number 18
Size of DG (kW) at bus number 11
1000

500

0
0 2000 4000 6000 8000 10000 12000
NFFEs
Fig. 11.Profile of DG size optimization for 69-bus distribution network using SOS.
6000

5040
5000

NFFEs 4000

3000

2000
1260 1200 1260 1200 1200
960
1000

0
PSO TLBO CS ABC GSA SFS SOS
Algorithms

(a)

8000

7000 6768

6000

5000
NFFEs

4000

3000

2000 1692 1770


1434 1534 1487
1270
1000

0
PSO TLBO CS ABC GSA SFS SOS
Algorithms

(b)
Fig. 12. NFFEs taken to converge by different algorithms for 69-bus distribution network (a) minimum
NFFEs and (b) average NFFEs.
Fig. 13.Comparative voltage profiles of 69-bus distribution network before and after placement of DG.
List of Tables
Table 1 Benchmark functions
Function Name Test function Dimension Range Optimum
(D) value fitness
function
Beale 𝑓1 (𝑥) = (1.5 − 𝑥1 + 𝑥1 𝑥2 )2 + (2.25 − 𝑥1 + 𝑥1 𝑥2 2 )2 2 [-4.5, 0
+ (2.625 − 𝑥1 + 𝑥1 𝑥2 3 )2 4.5]

Six hump camel 1 2 [-5,5] -1.03163


𝑓2 (𝑥) = 4𝑥12 − 2.1𝑥14 + 𝑥16 + 𝑥1 𝑥2 − 4𝑥22 + 4𝑥24
back 3

Goldstein Price 𝑓3 (𝑥) = [1 +(𝑥1 + 𝑥2 + 1)2 (19 − 14𝑥1 + 3𝑥12 − 2 [-2,2] 3


14𝑥2 + 6𝑥1 𝑥2 + 3𝑥2 2 )] × [30 +(2𝑥1 − 3𝑥2 )2 (18 −
32𝑥1 + 12𝑥12 + 48𝑥2 − 36𝑥1 𝑥2 + 27𝑥2 2 )]

Sum Square 𝐷 30 [-10,10] 0


𝑓4 (𝑥) = ∑ 𝑖𝑥𝑖2
𝑖=1
Sphere 𝐷 30 [- 0
𝑓5 (𝑥) = ∑ 𝑥𝑖2 100,100]
𝑖=1
Step 𝐷 30 [-5.12, 0
𝑓6 (𝑥) = ∑(𝑥𝑖 + 0.5)2 5.12]
𝑖=1
Quartic 𝐷 30 [-1.28, 0
𝑓7 (𝑥) = ∑ 𝑖𝑥𝑖4 1.28]
𝑖=1
Gricwank 𝐷 30 [-600, 0
1 𝐷 𝑥𝑖
𝑓8 (𝑥) = [∑ 𝑥𝑖2 ] − [∏ cos ( )] + 1 600]
4000 𝑖=1 √𝑖
𝑖=1

Schwefel 2.22 𝐷 30 [-10, 10] 0


𝐷
𝑓9 (𝑥) = ∑|𝑥𝑖 | + ∏ |𝑥𝑖 |
𝑖=1
𝑖=1

Ackley 𝐷
30 [-32, 32] 0
𝑓10 (𝑥) = −20 exp −0.2√1⁄𝐷 ∑ 𝑥𝑖2
𝑖=1
( )
𝐷

− exp (1⁄𝐷 ∑ cos(2𝜋𝑥𝑖 )) + 20


𝑖=1
+𝑒
Table 2 Parameters of different algorithms
Algorithms Parameters Population Maximum
size NFFEs

PSO C1 = C2 = 2 If D = 2, Population =30 If D = 2,


wmax = 0.9, wmin= 0.4 Maximum NFFEs =10000
If D = 30, Population = If D = 30,
TLBO Teaching factor = either 1 or 2 100 Maximum NFFEs =
1000000
CS Levy co-efficient = 0.5
Discovery rate of alien eggs = 0.25
ABC -
GSA G0=100 and α = 20
SFS Maximum diffusion number is set to
1
SOS -
An entry “-” means not applicable
Table 3 Result of benchmark functions
Functions Attributes PSO TLBO CS GSA SFS SOS
𝑓1 (𝑥) Best FF 0 0 0 0 0 0
Worst FF 0 0 0.4524 0 0 0
Mean 0 0 0.05545 0 0 0
Std. Dev 0 0 0.12639 0 0 0
Lowest NFFEs to converge 2820 2760 2190 1980 1950 1710
Highest NFFEs to converge 35250 5100 -- 3840 3540 2790
Average NFFEs to converge 27245 3704 -- 3030 2865 2118
Rank 5 4 6 3 2 1

𝑓2 (𝑥) Best FF -1.0316 -1.0316 -1.0316 -1.0316 -1.0316 -1.0316


Worst FF -1.0316 -1.0316 -1.0316 -1.0316 -1.0316 -1.0316
Mean -1.0316 -1.0316 -1.0316 -1.0316 -1.0316 -1.0316
Std. Dev 0 0 0 0 0 0
Lowest NFFEs to converge 21210 1860 1470 1230 1170 1110
Highest NFFEs to converge 38550 4020 2070 2460 2460 2670
Average NFFEs to converge 31419 2430 1800 1965 1755 1578
Rank 6 5 3 4 2 1

𝑓3 (𝑥) Best FF 3 3 3 3 3 3
Worst FF 3 3 3 3 3 3
Mean 3 3 3 3 3 3
Std. Dev 0 0 0 0 0 0
Lowest NFFEs to converge 6090 1920 1770 1830 1620 1590
Highest NFFEs to converge 34620 3480 1950 2430 2520 1950
Average NFFEs to converge 27765 2430 1810 2190 2250 1674
Rank 6 5 2 3 4 1

𝑓4 (𝑥) Best FF 0 0 0 0 0 0
Worst FF 0 0 0 0 0 0
Mean 0 0 0 0 0 0
Std. Dev 0 0 0 0 0 0
Lowest NFFEs to converge 575800 204200 119100 25800 25000 24200
Highest NFFEs to converge 596500 221800 191100 34500 35700 26600
Average NFFEs to converge 585773 214000 158120 30330 31060 26200
Rank 6 5 4 2 3 1

𝑓5 (𝑥) Best FF 0 0 0 0 0 0
Worst FF 0 0 0 0 0 0
Mean 0 0 0 0 0 0
Std. Dev 0 0 0 0 0 0
Lowest NFFEs to converge 591300 122200 141100 33060 30180 28200
Highest NFFEs to converge 610800 225800 231300 41100 38100 34400
Average NFFEs to converge 599730 149140 169420 36900 33960 31450
Rank 6 4 5 3 2 1

𝑓6 (𝑥) Best FF 0 0 0 0 0 0
Worst FF 0 0 0 0 0 0
Mean 0 0 0 0 0 0
Std. Dev 0 0 0 0 0 0
Lowest NFFEs to converge 560700 324800 92700 30300 25300 24100
Highest NFFEs to converge 574600 344000 176300 47400 39600 27300
Average NFFEs to converge 568770 332420 121800 39120 32630 25780
Rank 6 5 4 3 2 1

𝑓7 (𝑥) Best FF 0 0 0 0 0 0
Worst FF 0 0 0 0 0 0
Mean 0 0 0 0 0 0
Std. Dev 0 0 0 0 0 0
Lowest NFFEs to converge 554300 93200 40300 20400 14400 10600
Highest NFFEs to converge 594700 104000 62900 31500 24900 13600
Average NFFEs to converge 573640 98300 53980 26850 21450 11860
Rank 6 5 4 3 2 1
Continued to next page
Table 3 Continued

Functions Attributes PSO TLBO CS GSA SFS SOS


𝑓8 (𝑥) Best FF 0 14.1718 0.0271 0 0 0
Worst FF 0.0221 75.7591 1.7147 0 0 0
Mean 0.00641 35.7482 0.5067 0 0 0
Std. Dev 0.005847 17.1344 0.59146 0 0 0
Lowest NFFEs to converge 582000 -- -- 40200 52200 29800
Highest NFFEs to converge -- -- -- 51900 58200 39800
Average NFFEs to converge -- -- -- 47850 55800 34360
Rank 4 6 5 2 3 1

𝑓9 (𝑥) Best FF 0 0 0 0 0 0
Worst FF 0 0 0.0787 0 0 0
Mean 0 0 0.08142 0 0 0
Std. Dev 0 0 0.03581 0 0 0
Lowest NFFEs to converge 583800 565700 372500 57600 41000 40200
Highest NFFEs to converge 635500 637500 -- 62400 56400 42600
Average NFFEs to converge 599260 600940 -- 59100 52300 41480
Rank 4 5 6 3 2 1

𝑓10 (𝑥) Best FF 0 0 5.0581 0 0 0


Worst FF 0 0 14.0587 0 0 0
Mean 0 0 10.6110 0 0 0
Std. Dev 0 0 2.90370 0 0 0
Lowest NFFEs to converge 642000 401100 -- 114000 108000 75300
Highest NFFEs to converge 698900 417900 -- 122100 115200 110500
Average NFFEs to converge 661690 407880 -- 117050 108900 87500
Rank 5 4 6 3 2 1
Table 4 Performance of different phases of SOS in benchmark functions test
Functions Attributes Mutualism Phase Mutualism + Mutualism +
Commensalism Phase Commensalism +
Parasitism Phase
𝑓1 (𝑥) Best FF 0 0 0
Worst FF 0.7621 0 0
Mean 0.07621 0 0
Std. Dev 0.72299 0 0
Lowest NFFEs to converge 2190 1830 1710
Highest NFFEs to converge --- 2670 2790
Average NFFEs to converge --- 2178 2118

𝑓2 (𝑥) Best FF -1.0316 -1.0316 -1.0316


Worst FF -1.0316 -1.0316 -1.0316
Mean -1.0316 -1.0316 -1.0316
Std. Dev 0 0 0
Lowest NFFEs to converge 1830 1470 1110
Highest NFFEs to converge 2790 2190 2670
Average NFFEs to converge 2142 1734 1578

𝑓3 (𝑥) Best FF 3 3 3
Worst FF 3 3 3
Mean 3 3 3
Std. Dev 0 0 0
Lowest NFFEs to converge 2550 1830 1590
Highest NFFEs to converge 9510 7830 1950
Average NFFEs to converge 6042 3138 1674

𝑓4 (𝑥) Best FF 0 0 0
Worst FF 0 0 0
Mean 0 0 0
Std. Dev 0 0 0
Lowest NFFEs to converge 28200 26600 24200
Highest NFFEs to converge 31400 27400 26600
Average NFFEs to converge 29240 29240 26200

𝑓5 (𝑥) Best FF 0 0 0
Worst FF 0 0 0
Mean 0 0 0
Std. Dev 0 0 0
Lowest NFFEs to converge 32200 29800 28200
Highest NFFEs to converge 41800 37000 34400
Average NFFEs to converge 37320 32360 31450

𝑓6 (𝑥) Best FF 0 0 0
Worst FF 0 0 0
Mean 0 0 0
Std. Dev 0 0 0
Lowest NFFEs to converge 27700 25500 24100
Highest NFFEs to converge 37700 28500 27300
Average NFFEs to converge 32660 27440 25780

𝑓7 (𝑥) Best FF 0 0 0
Worst FF 0 0 0
Mean 0 0 0
Std. Dev 0 0 0
Lowest NFFEs to converge 12200 11400 10600
Highest NFFEs to converge 14600 13000 13600
Average NFFEs to converge 13720 12040 11860
Continued to next page
Table 4 Continued

Functions Attributes Mutualism Mutualism + Mutualism +


Phase Commensalism Commensalism
Phase + Parasitism
Phase
𝑓8 (𝑥) Best FF 0 0 0
Worst FF 0 0 0
Mean 0 0 0
Std. Dev 0 0 0
Lowest NFFEs to converge 35400 31400 29800
Highest NFFEs to converge 45000 42600 39800
Average NFFEs to converge 42210 36600 34360

𝑓9 (𝑥) Best FF 0 0 0
Worst FF 0 0 0
Mean 0 0 0
Std. Dev 0 0 0
Lowest NFFEs to converge 42600 41800 40200
Highest NFFEs to converge 44200 43400 42600
Average NFFEs to converge 43320 42680 41480

𝑓10 (𝑥) Best FF 0 0 0


Worst FF 0 0 0
Mean 0 0 0
Std. Dev 0 0 0
Lowest NFFEs to converge 75700 75700 75300
Highest NFFEs to converge 110900 110900 110500
Average NFFEs to converge 91200 94700 87500

Reviewer 2, Comment 2
Table 5 Parametric sensitivity analysis of PSO for 33-bus distribution network

Population wmin wmax Value of FF

C1 = C2 = C1 = C2= 1 C1 = C2 = C1 = C2 = 2 C1 = C2 =
0.5 1.5 2.5
20 0.4 0.1 0.352723 0.352842 0.352724 0.352724 0.352787
0.2 0.353071 0.352963 0.352958 0.352739 0.352863
0.3 0.353004 0.352897 0.352990 0.352812 0.352751
0.4 0.352779 0.352674 0.352669 0.352678 0.352681
0.5 0.352784 0.352789 0.352683 0.352788 0.352790
0.6 0.352675 0.352678 0.352687 0.352685 0.352685
0.7 0.352578 0.352575 0.352583 0.352572 0.352579
0.8 0.352589 0.352584 0.352574 0.352576 0.352578
0.9 0.352584 0.352582 0.352590 0.352570 0.352590
1.0 0.352589 0.352588 0.352587 0.352576 0.352598

30 0.4 0.1 0.353577 0.353271 0.352874 0.353571 0.353574


0.2 0.354583 0.355571 0.354575 0.353573 0.354576
0.3 0.355583 0.356104 0.354551 0.353572 0.353612
0.4 0.355529 0.354527 0.353412 0.353168 0.353237
0.5 0.352674 0.352653 0.352670 0.352668 0.352891
0.6 0.352573 0.352570 0.352569 0.352569 0.352569
0.7 0.352575 0.352576 0.352578 0.352568 0.352569
0.8 0.352570 0.352568 0.352569 0.352567 0.352568
0.9 0.352569 0.352569 0.352568 0.352566 0.352568
1.0 0.352578 0.352580 0.352567 0.352567 0.352568

40 0.4 0.1 0.352776 0.352743 0.352754 0.352709 0.352711


0.2 0.352784 0.352741 0.352670 0.352659 0.352742
0.3 0.352791 0.352705 0.352675 0.352669 0.352669
0.4 0.352690 0.352682 0.352631 0.352608 0.352649
0.5 0.352672 0.352669 0.352668 0.352631 0.352668
0.6 0.352572 0.352575 0.352571 0.352567 0.352570
0.7 0.352589 0.352584 0.352572 0.352568 0.352569
0.8 0.352571 0.352570 0.352569 0.352567 0.352569
0.9 0.352586 0.352589 0.352576 0.352567 0.352570
1.0 0.352590 0.352574 0.352568 0.352567 0.352572

50 0.4 0.1 0.352698 0.352687 0.352851 0.352786 0.352621


0.2 0.352672 0.352734 0.352721 0.352601 0.352663
0.3 0.352641 0.352678 0.352690 0.352736 0.352616
0.4 0.352671 0.352761 0.352735 0.352682 0.352658
0.5 0.352590 0.352589 0.352576 0.352576 0.352579
0.6 0.352610 0.352590 0.352576 0.352571 0.352587
0.7 0.352587 0.352584 0.352583 0.352575 0.352576
0.8 0.352576 0.352573 0.352569 0.352568 0.352574
0.9 0.352575 0.352572 0.352569 0.352567 0.352569
1.0 0.352576 0.352573 0.352568 0.352567 0.352569
Table 6 Parametric sensitivity analysis of CS for 33-bus distribution network

Population Levy Value of FF


coefficient (λ) Pa = 0.25 Pa = 0.50 Pa = 0.75
20 0.1 0.352910 0.352937 0.352874
0.2 0.352821 0.352856 0.352863
0.3 0.352793 0.352797 0.352832
0.4 0.352768 0.352783 0.352793
0.5 0.352605 0.352681 0.352738
0.6 0.352636 0.352678 0.352691
0.7 0.352709 0.352731 0.352757
0.8 0.352753 0.352785 0.352796
0.9 0.352783 0.352811 0.352839

30 0.1 0.352721 0.352702 0.352718


0.2 0.352735 0.352787 0.352730
0.3 0.352698 0.352723 0.352718
0.4 0.352595 0.352623 0.352669
0.5 0.352566 0.352589 0.352597
0.6 0.352572 0.352580 0.352597
0.7 0.352579 0.352586 0.352582
0.8 0.352575 0.352583 0.352595
0.9 0.352629 0.352653 0.352673

40 0.1 0.352830 0.352823 0.352862


0.2 0.352773 0.352809 0.352828
0.3 0.352721 0.352745 0.352782
0.4 0.352638 0.352683 0.352719
0.5 0.352593 0.352598 0.352638
0.6 0.352587 0.352621 0.352657
0.7 0.352599 0.352603 0.352611
0.8 0.352663 0.352698 0.352634
0.9 0.352694 0.352671 0.352710

50 0.1 0.352706 0.352727 0.352732


0.2 0.352687 0.352677 0.352693
0.3 0.352602 0.352621 0.352684
0.4 0.352582 0.352593 0.352637
0.5 0.352569 0.352583 0.352610
0.6 0.352568 0.352569 0.352589
0.7 0.352568 0.352581 0.352597
0.8 0.352567 0.352598 0.352621
0.9 0.352609 0.352652 0.352683
Table 7 Parameter settings of the employed algorithms
Algorithms Parameters Population Maximum
size NFFEs

PSO C1 = C2 = 2, wmax = 0.9, wmin= 0.4 30 12000

TLBO Teaching factor = either 1 or 2 30 12000

CS Levy co-efficient, λ = 0.5 30 12000


Discovery rate of alien eggs, Pa = 0.25

ABC - 30 12000

GSA G0=100 and α = 20 30 12000


SFS Maximum diffusion number is set to 1 30 12000
SOS - 30 12000
An entry “-” means not applicable.
Table 8 Results of DG sizing for 33-bus distribution network
Methods Size of DG (kW) at bus number Value of FF Minimum loss CPU time to converge(sec) NFFEs taken to converge
Used (kW)
30 24 14 Minimum Maximum Average Minimum Maximum Average
PSO 1071.42 1099.44 753.98 0.3526 71.4572 21.9046 41.3821 29.1597 1680 2580 2052
TLBO 1071.42 1099.44 753.98 0.3526 71.4572 14.0952 25.2408 18.6120 1680 1920 1404
CS 1071.42 1099.44 753.98 0.3526 71.4572 13.5955 22.7680 16.8150 1620 1860 1368
ABC 1065.12 1100.64 763.11 0.3526 71.4612 33.7458 137.4396 68.5500 3960 11070 8901
GSA 1071.42 1099.44 753.98 0.3526 71.4572 14.0236 26.6537 17.9934 1200 2040 1534
SFS 1071.42 1099.44 753.98 0.3526 71.4572 15.2364 26.7625 19.2736 1170 1830 1417
SOS 1071.42 1099.44 753.98 0.3526 71.4572 13.1406 20.3496 15.4719 1080 1560 1224

43
Table 9 Results of DG sizing for 69-bus distribution network
Methods Size of DG (kW) at bus number Value of FF Minimum loss CPU time to converge (sec) NFFEs taken to converge
Used (kW)
61 18 11 Minimum Maximum Average Minimum Maximum Average
PSO 1719.00 380.50 526.70 0.3086 69.4286 105.7501 150.6839 119.6717 1260 2070 1692
TLBO 1719.00 380.50 526.70 0.3086 69.4286 55.7826 95.7304 91.5096 1200 1860 1434
CS 1719.00 380.50 526.70 0.3086 69.4286 42.8362 283.0023 92.3834 1260 5400 1770
ABC 1714.80 372.30 556.50 0.3086 69.4322 185.0120 324.4362 222.3758 5040 7830 6768
GSA 1719.12 380.50 526.70 0.3086 69.4286 51.4534 153.8775 106.3478 1200 1740 1534
SFS 1719.12 380.50 526.70 0.3086 69.4286 49.5382 147.2673 92.6374 1200 1710 1487
SOS 1719.12 380.50 526.70 0.3086 69.4286 47.7873 107.9805 87.5948 960 1680 1270

44
Table 10 Comparison of results for 33-bus distribution network
References Methods used Number of Selected Size of Active
DG placed bus/buses DG/DGs power loss
(MW) (kW)
Base case - - - - 202.6771

Ref. [46] RPSO 2 30 1.4830 116.71


32 0.3836
RPSO 3 6 1.2274 76.91
14 0.6068
31 0.6870

Ref. [62] PSO 1 7 2.8951 114.89

Ref. [63] Analytical method 1 6 2.49078 111.17

Ref. [64] Grid search 1 6 2.60050 111.03

Ref. [65] HPSO 1 8 3623.90 131.85


HPSO 2 16 1313.90 87.65
22 2212.30
HPSO 3 15 1364.1 84.16
29 444.0
31 1973.0

Ref. [66] GA 3 11 1.50 106.3


29 0.4228
30 1.0714
PSO 3 8 1.1768 105.3
13 0.9816
32 0.8297
GA/PSO 3 11 0.925 76.91
16 0.863
32 1.20

SOS [Proposed] SOS 3 14 0.75398 71.4572


24 1.09944
30 1.07142
An entry “-” means not applicable.

45
Table 11 Comparison of results for 69-bus distribution network
References Method used Number of Selected Size of Active
DG placed bus/buses DG/DGs power loss
(MW) (kW)
Base case - - - - 225.0028

Ref. [46] RPSO 1 61 1.873 83.22


Ref. [62] PSO 1 61 2.0264 84.04

Ref. [63] Analytical method 1 61 1.80782 83.37

Ref. [64] Grid search 1 61 1.86303 83.22

Ref. [65] HPSO 1 61 3684.7 87.13


HPSO 2 48 0.5476 86.68
61 3.6851
HPSO 3 46 0.1529 87.0
61 3.6525
63 0.0322

Ref. [66] GA 3 21 0.9297 89.0


62 1.0752
64 0.9848
PSO 3 17 0.9925 83.20
61 1.1998
63 0.7956
GA/PSO 3 21 0.9105 81.10
61 1.1926
63 0.8849

Ref. [67] MBFO 1 61 1.8792 83.40


BFO 1 61 2.0258 84.20

SOS [Proposed] SOS 3 11 0.5267 69.4286


18 0.3805
61 1.7190
An entry “-” means not applicable.

46

You might also like