Renewable Energy Focus 49 (2024) 100550

Contents lists available at ScienceDirect

Renewable Energy Focus

journal homepage: www.elsevier.com/locate/ref

Optimal distributed generators allocation with various load models under

load growth using a meta-heuristic technique
Muhammad Zubair Iftikhar a, Kashif Imran a, *, Muhammad Imran Akbar b, Saim Ghafoor c
a
U.S.-Pakistan Center for Advanced Studies in Energy (USPCAS-E), National University of Sciences and Technology (NUST), H-12, Islamabad 44000, Pakistan
b
Multan Electric Power Company (MEPCO), Multan, Pakistan
c
Atlantic Technological University, Donegal, Ireland

A R T I C L E I N F O A B S T R A C T

Keywords: Distribution network planning and operation are facing several problems, including asset congestion, voltage
Distributed generators fluctuations, and system instability. The adequate planning and modeling of distributed generators and capacitor
Distribution network banks must quantify these problems. This article presents the optimal allocation of distributed generators in
Environmental emission
parallel with capacitor banks in distribution networks with single and multi-objectives using the Gazelle Opti­
Meta-heuristic algorithm
Multi-objectives framework
mization Algorithm (GOA) and Mountain Gazelle Optimization Algorithm (MGOA). The single objective
Voltage-dependent load model framework includes technical objectives like minimization of active power losses. The multi-objective framework
includes technical and non-technical objectives like simultaneously minimization of active power losses, voltage
stability, and voltage deviation, and minimization of polluting greenhouse gases and total electricity purchase
cost. Furthermore, these planning problems are investigated by three case studies on different nonlinear voltage-
dependent models at two different loading conditions from future planning perspectives. The effectiveness and
feasibility of the MGOA are evaluated on the IEEE standard 33 bus system. As a result, the MGOA demonstrates a
remarkable reduction in technical and non-technical objectives in all types of distributed generator placement.
Moreover, a comparative analysis of other existing research works validated the efficiency and feasibility of
established algorithms at each use case with different load models by improving all the objective functions of
network planning. In single-objective and multi-objective frameworks, the active power losses reduce to 94.42%
and 93.57% in the voltage-independent model, respectively. Meanwhile, the non-technical objectives are also
significantly improved for each load model, further validating the efficiency of the proposed algorithms.

Introduction system, the concept of optimal placing of distributed generators (DGs)

and capacitor banks (CBs) in radial distribution systems (RDS) using
In modern times, the proliferation of load demand at the consumer’s optimization algorithms at optimal size is getting the researcher’s
end is creating significant changes in the planning of the distribution attention as a better alternative solution [1–3]. DG placement improves
system, which creates a question for network planners: How do they the technical and non-technical benefits like reduction of APL,
supply power to consumers at minimum system loss and voltage drop? improvement of voltage deviation index (IVD), reduction of voltage
However, the load is increasing daily, a major problem for future stability (VS), reduction of the pollutant emission by DGs, and saving the
network planning. Installing and placing a new power generating station cost of commissioning the big power plant.
or upgrading the existing power distribution system is complicated and Distribution systems contain equipment like underground cable, a
costly. Some conventional generating stations generate electricity by distribution line, a primary and secondary distribution level trans­
burning oil, coal, and gases that pollute the environment through their former, and inductive and reactive load at the consumer’s side. The
byproduct emissions. In a power system, the major power consumption operation and power consumption of these equipment tend to cause
consumers are in the secondary distribution system, where active power power loss and poor voltage regulation in distribution networks (DN).
loss (APL) and voltage drop are due to increasing resistance/reactance Duong Quoc in [4] has categorized DGs into four types based on
(R/X) values along the system. For future planning of the distribution required power.

* Corresponding author.
E-mail address: [email protected] (K. Imran).

https://doi.org/10.1016/j.ref.2024.100550
Received 21 October 2023; Received in revised form 14 January 2024; Accepted 13 February 2024
Available online 23 February 2024
1755-0084/© 2024 Elsevier Ltd. All rights reserved.
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

1. Type 1: DGs that inject only Active power like Fuel Cell and Photo­ Algorithm (ICA), Genetic algorithm (GA) [6] and jellyfish search algo­
voltaic Cell. (Operate at unity power factor (UPF)). rithm [17]. Swarm-based algorithms are also deployed in different ar­
2. Type 2: DGs that inject only Reactive power like FACTS devices, D- ticles where Stud Krill herd Algorithm (SKHA) [11], Grasshopper
STATCOM, CB. (Operate at Zero power factor (ZPF)). Optimization Algorithm (GrOA), Cuckoo Search (CS) [31], Bat Algo­
3. Type 3: DGs that inject both Active and Reactive power like Con­ rithm (BA) [37], Ant Lion Optimization Algorithm (ALOA) [38], Particle
ventional Generator. (Operate at fixed or optimal power factor (FPF), Swarm optimization (PSO) [39] and Bacterial Foraging Optimization
(OPF)) Algorithm (BFOA) [40] are reported on same problem. Moreover,
4. Type 4: DGs that inject Active power but consume Reactive power Physics-based algorithms like Gravitational Search Algorithm (GSA)
like DFIG, Synchronous Condenser, and Wind Turbine. (Operate at [37] are reported in a few articles for optimal DG allocation problems.
UPF). Nowadays, Artificial intelligence(AI) based algorithms have
improved the efficiency of existing metaheuristic algorithms, making
Optimal allocation of DGs in parallel with CBs is also used to improve them hybrid algorithms by amalgamating two different algorithm types
voltage regulation, minimize power loss, and improve feeder loading of algorithms or integrating with AI algorithms. These improved and
[1,2,5–8]. CBs are mostly placed on buses with high inductive loads to hybrid algorithms are also reported in different articles, including
improve the voltage, reducing the grid’s reactive power demand. Type 3 Improved Grey Wolf Optimizer (IGWO), Improved grey wolf optimiza­
DGs inject active and reactive power into the system, which is also a tion with particle swarm optimization (I-GWOPSO) [3], Hybrid Grey
good solution to improve technical objectives. However, they are usu­ Wolf optimizer (H-GWO) [10], Stochastic fractal search algorithm
ally run by burning fossil fuels, polluting the environment. On the other (SFSA) [12], Modified Gradient-Based Algorithm (MGBA) [19], Non-
hand, DGs of type 1, in parallel with CBs, improve the network’s ob­ dominated Sorting Genetic Algorithm II (NSGA-II), Multi-objective
jectives without polluting the environment. Shuffled Bat algorithm (MOSBA) [23], Improved Elephant Herding
Optimization (IMOEHO) [28], Improved Harris Hawks Optimization
Literature review (IHHO) [29] and Constriction Factor Particle Swarm Optimization
(CFPSO) [36]. The findings of each metaheuristic algorithm are sum­
The optimal placement of DGs and CBs is considered a real and marized in Section 6 for each study.
complex problem from different objective function perspectives. In From a planning perspective, two load models are usually consid­
many pieces of literature [1,5,9–17] only DG placement with the single ered: linear voltage-independent (VID) and nonlinear voltage-
objective of APL minimization, voltage profile improvement, and system dependent (VD) models. Most studies are on the VID load model for
security enhancement or a combination of all these, is solved at different comparative analysis of proposed algorithms, where loads are assumed
load models with different meta-heuristic techniques. Additionally, re­ to have constant power (CP). From a planning viewpoint, the VD models
searchers simulate the experiment of placing DGs in parallel with CBs by are used in research studies, which are based on real-time distribution
using different optimization techniques with different objectives, as systems, includes four different types of existing load models residential
discussed in many research works [1,2,5–8,18,19]. In the context of load (RL), commercial load(CL), industrial load(IL), and a mixture of
planning, the single objective of minimization of APL [2,15,16,20–25], these three load model as mixed load (ML) [2,22,23,39,41]. However,
and multi-objective of the combination of APL, IVD, and VS [15,26–29] the different load growth models, including light load (LL), normal load
are discussed. Moreover, saving the environment from emission of (NL), and high load (HL) models, are also considered in further research
greenhouse gasses (GHG) which emit from power generation stations works for future planning of the system [5,11,31,33]. Nowadays,
[7,20,21], saving money for placing right DG type at the right place network planning engineers are more concerned about load growth for
[2,3,14,16,22,30,31] with a combination of minimization of APL, this future planning, which is why it has become an important research area
multi-objective task is also carried out in [1,32]. Furthermore, for from a planning perspective [42].
operational purposes, most of the studies like [1,3,5,9–13,16] examined In this article, comprehensive planning of the radial distribution
the benefits of placing and sizing the DGs type 1, which operates at unity system is carried out using a recently improved meta-heuristic optimi­
power factor. Although some articles like [2,4,14,18], and [24] exam­ zation algorithm that analyzed different load models with all possible
ined fixed power factor type 3 DG, only a few studies like [2,6–8] planning objectives of the network planner. As very few articles have
examined the benefits by placing both DG type 1 and CB simultaneously. been reported on the optimal allocation of DGs in parallel with CBs, this
The use of heuristic and meta-heuristic optimization algorithms to article has attempted to tackle the load growth of distribution systems by
solve real and complex optimization problems attracted the attention of satisfying the planning objectives without polluting the environment.
researchers. As explained in [9,30,33], analytical methods are used in Table 1 consists of a comprehensive overview of existing research
rare cases that have perceived the complexities of problem analysis in studies and proposed studies for the problem of DGs and CBs placement
the mathematical formulation of objective functions for DGs and CBs for planning engineers using different optimization techniques with
sizing and sitting. Analytical methods usually find exact solutions when different technical and non-technical objectives, which are not jointly
the problems are mathematically well-defined. However, they face examined in any single study.
limitations while dealing with complex and nonlinear problems. On the The highlights of research contributions are described as follows.
other hand, metaheuristic techniques are widely used for these problems
due to the high computational power of algorithms for finding the 1. Proposed a recently improved version of the Gazelle optimization
optimal solutions where the global optimum is challenging. Moreover, algorithm.
metaheuristic algorithms are more accessible to implement than 2. Proposed optimization problem includes technical and non-technical
analytical methods. Many researchers proposed novel and nature- objectives.
inspired algorithms that solved problems according to the rules of na­ 3. The single and multi-objective problems are analyzed by placing all
ture. These meta-heuristic algorithms are divided into different cate­ types of DGs.
gories according to the nature of inspiration [34,35]. They consist of 4. The effectiveness of the proposed algorithm is evaluated at IEEE 33
three major types: Evolutionary algorithms, Physics-based algorithms, bus systems.
and Swarm-based algorithms. Many of these metaheuristic techniques 5. Emphasized the benefits of placing DGs parallel with CBs at different
are proposed so far to solve the placement of DGs problem for radial load models.
distribution networks where the Evolutionary algorithm includes Water
Cycle Algorithm (WCA) [1], Harmony Search Algorithm (HSA), Particle The rest of this article is organized as follows. Section 3 describes the
Artificial Bee Colony algorithm (PABC) [5], Imperialist Competitive system modelling and problem formulation, which consists of DG

2
 M. Zubair Iftikhar et al.
Table 1
Summary of literature.
Year Method Ref. DG Type Load Model Load Type Objective Function

I II III I VID VD Normal High Active Voltage Voltage Environmental DG Annual

|| Constant Industrial Residential Commercial Mixed Load Load Power Deviation Stability Emission Cost Saving
III Power Load Load Load Load Losses Index Index

2014 GA, ICA [6] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓

2016 HAS- [5] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PABC
2016 MOSBAT [23] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
2016 GA [41] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
2017 KHA, [11] ✓ ✓ ✓ ✓ ✓ ✓ ✓
SKHA
2017 PSO [39] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
2017 ALO [38] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
3

2018 DE, PSO [7] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓

2018 WCA [1] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
2019 SA [8] ✓ ✓ ✓ ✓ ✓ ✓ ✓
2020 LSA [31] ✓ ✓ ✓ ✓ ✓ ✓ ✓
2021 SOS-NNA [2] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
2021 GA-SBO [32] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
2021 SMF [20] ✓ ✓ ✓ ✓ ✓ ✓ ✓
2021 QOGWO [22] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
2021 I-DBEA [15] ✓ ✓ ✓ ✓ ✓ ✓ ✓
2022 IGWOPSO [3] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
2022 MOFA [21] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
2023 GA, GWO [25] ✓ ✓ ✓ ✓ ✓ ✓ ✓
2023 WOA [16] ✓ ✓ ✓ ✓ ✓ ✓
2023 MGBA [19] ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
2024 GOA, P ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
MGOA

Renewable Energy Focus 49 (2024) 100550

M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

modeling, load modeling, and objective function. Section 4 describes the [( ( )β ) ( ( )β ) ( ( )β )

proposed algorithms, and Section 5 describes the implementation sce­ QLT = l
|Vi | o
ωo Qoi + ωI Qoi
|Vi | I
+ ωγ Qoi
|Vi | γ
narios, including case studies. Results are discussed in Section 6. Lastly, |Voi | |Voi | |Voi |
the conclusion is summarized. ( ( )βc ) ]
|Vi |
+ ωc Qoi
|Voi |
System modeling and problem formulation
(2)
The purpose of single and multi-objective functions, equality, The values of load exponent for active and reactive power of different
inequality constraints, load models, and DG modeling are introduced for load models are given in
optimal allocation of DGs and CBs in RDS. A DN comprises generating Table 2. The VD load model multipliers are defined in Table 3, where
units, time-varying loads, and interconnecting overhead or underground α and β are active and reactive power coefficients of voltage models. o,
conductors. Efficient planning of distribution networks requires accu­ I, γ, and c represent the power coefficient of constant, industrial, resi­
rate modeling of these components. Usually, load varies with time for dential, and commercial loads, ω define which load model is required
each type of load model, so placing DGs and CBs with varying loads is and its value is either 0 or 1, l represent the loading condition, and Voi ,Vi
inappropriate. Therefore, we simulate our problem at the maximum are voltage value of one and standard value of IEEE network at each bus,
load of RDS because if the peak load known as the worst case is fulfilled respectively.
by DGs, then other time-varying loads will easily be managed. This A DN usually consists of mixed loads, i.e., industrial, commercial,
approach is also essential for planning of future growth in load demand, and residential loads. The multipliers given in the last row of Table 3 are
where each load model is considered a balanced network with constant used to formulate a mixed load. Mixed load is the cumulative sum of the
active and reactive power [15]. active and reactive power of industrial, residential, and commercial
loads, respectively 40%, 45%, and 15% of the total load demand of the
active distribution network [2].
DG Modelling The results are also analyzed on different load growth factors for
future planning, and γ in the Eqs. (1) and (2) is used as a growth factor
Due to the operational characteristics, DGs are classified into four [11,13]. In this article, we examine the two loading conditions known as
types based on the connection method [4]. In this study, only three DG NL and HL for future planning, and their coefficients l are considered as
types are placed with four combinations. Since the mathematical 1 and 1.5, respectively [5,11,31,39]. The load modeling for this study is
modelling of type 3 and 4 DGs are same in MATLAB programming, only summarized in Fig. 1.
three types of DG are used for these combinations [7,2,3]. The first three
combinations have their own distinct nature due to the types of DGs. In
contrast, combination 4 is the amalgamation of type 1 and type 3 DGs, Objective Functions:
which aims to inject both active and reactive power into the network by
pollutant-free sources of energy. These combinations are modeled as The objective function of the proposed research is divided into single
negative loads, which are classified as follows. and multi-objective, including technical and non-technical objectives.
Technical objectives are the minimization of APL (kW) and Voltage
1. Combination 1 (C1): It includes the Type 1 DGs, which operate at deviation index (IVD) as well as the maximization of Voltage stability
unity Power Factor; they only inject Active power to Networks like index (IVS) [28,30]. In contrast, the non-technical objectives involve
PV. techno-economic considerations and environmental emissions concerns
2. Combination 2 (C2): It includes the Type 3 DGs, which operate at a [1,43]. The techno-economics objective is minimizing annual economic
fixed 0.85 lagging Power Factor; they inject both Active and Reactive loss while maximizing the annual savings after DG and CB placement
power to Networks like conventional generators (CG). and reducing the total cost of energy purchased. The Environmental
3. Combination 3 (C3): It includes the Type 2 DGs, which operate at Emission objective involves minimizing environmentally problematic
zero Power Factor; they only inject Reactive power to Networks like GHG emissions of DGs. These objectives are considered to improve the
D-FACTS devices, D-STATCOM, and CBs. distribution system efficiency. The mathematical equations of all ob­
4. Combination 4 (C4): It includes Type 1 DGs in Parallel with Type 2 jectives used in this study are presented in the following subsections.
DGs, like PV, parallel with the CBs.
Active power loss
The most important technical objective function is to minimize the
Load Modelling total active power loss to deliver power to end-users efficiently [1,3]. It
is calculated in RDS by the following Eq. (4) [3,16,25,44].
Most existing research articles on load modeling examined the VID
model, a constant load model that is mostly used for evaluating the ef­ ∑
tbr
APloss = |Ii |2 ∗ Ri (3)
ficiency of proposed optimization algorithms and to make a comparative i=1
analysis of existing studies. Nevertheless, the real scenarios-based load
models are usually VD models, which are mathematically formulated by f1 = min(APloss ) (4)
the following Eqs. (1) and (2) [2,24,25,30]. These load models will help
the planning engineers in modeling the integration of DG combinations Where tbr represents the total number of branches in RDN, i
from the perspective of different objectives, as explained in the up­
coming subsection. Table 2
[( ( )α ) ( ( )α ) ( ( )α ) Load models and their power exponents.
|Vi | o |Vi | I |Vi | γ
PLT = l ωo Poi + ωI Poi + ωγ Poi Load Model (LM) Power exponents
|Voi | |Voi | |Voi |
( ( )α ) ] αo βo αi βi αr βr αc βc
|Vi | c
+ ωc Poi Constant Power (CP) 0 0 0 0 0 0 0 0
|Voi |
Industrial (IL) 0 0 0.18 6.0 0 0 0 0
(1) Residential (RL) 0 0 0 0 0.92 4.04 0 0
Commercial (CL) 0 0 0 0 0 0 1.51 3.40

4
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Table 3 Where Vj is the value of the voltage at the weakest bus, APi , RPi
Load Model multipliers represents the active and reactive power of i the bus, and Rij , Xij denotes
LM ωo ωI ωγ ωc the resistance and reactance values between i and j bus.
CP 1 0 0 0
IL 0 0 0 1 Environmental emission
RL 0 1 0 0 The generating stations and DGs pollute the environment by pro­
CL 0 0 1 0 ducing Nitrogen oxides (NOx ), Sulphur dioxide (SO2 ), and Carbon di­
ML 0 0.40 0.45 0.15 oxide (CO2 ) gases by the combustion of fuels like coal, oil, and natural
gases [32]. The production of CO(Carbon Monoxide) due to incomplete
represents the current branch number, and Ii is the amount current combustion is problematic because it severely impacts the health of
passing through the current branch and Ri is the resistance of that human beings besides damaging the environment [20,21]. However,
branch, whereas APloss defines the total amount of APL in kW throughout hydrocarbon combustion mainly creates carbon dioxide while produc­
the RDN. ing 0.01% carbon monoxide [43]. Therefore, we only consider the
Carbon dioxide emissions in the following equation.
Voltage deviation index In this objective, we are reducing the gaseous emissions to decrease
Most types of equipment used at the consumer end are voltage sen­ the environmental impacts by Eq. (9) [1]:
sitive, so it is necessary to provide voltage within the permissible limits (
∑
TDG
)
and reduce its deviation [3,28]. The following Eq. (6) calculates the f4 = min GDGj + GGrid (9)
voltage deviation index: j

∑
tb
Where
IVD = (1 − VK )2 (5) ( )
k=1 GDGj = PGj ∗ CODG DG DG
2 + SO2 + NOx (10)

f2 = min(IVD) (6) ( )
GGrid = PGGrid ∗ COGrid
2 + SOGrid
2 + NOGrid
x (11)
Where tb represents the total number of buses in RDN, k represents
Where, PGj , PGGrid define the power injected by jth DG and grid,
the current bus number, and Vk is the voltage value at the bus is denoted
respectively.
by p.u.
Economic factor
Voltage stability index
The DGs and CBs must be placed at minimum cost, and the Eq. (12)
Voltage stability is also one of the most substantial security factors
can calculate it [1,7]:
for the distribution system. It is calculated on the weakest bus of the
( )
network, which usually has a minimum value of VSI, and the system may TCB ∑
∑ TDG
( )
collapse if it is below zero [25,30]. Hence, IVS must be greater than zero f5 = min CGrid + CCBi + CDGj (12)
for reliable distribution system operation. The voltage stability index is
i j=1

calculated in the radial distribution system by the following Eq. (8) Where:
[2,3,16,29]:
[⃒ ⃒ CDGj = α + β ∗ APDGj (13)
4 ( )2 ⃒ ⃒2 (
IVSi = ABS ⃒Vj ⃒ − 4 ∗ APi ∗ Xij − RPi ∗ Rij − 4 ∗ ⃒Vj ⃒ ∗ APi ∗ Rij − RPi ( )
)] captial cost($/kW) ∗ APDG ∗ Gγ
α= (14)
∗ Xij Lifetime(year) ∗ HA ∗ LFDG
(7)
β = O&Mcost($/kW.h) + fuel cost($/kW.h) (15)
( )
1
f3 = min (8)
IVSi

Fig. 1. Different load models of Distribution Networks

5
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

∑TCB
+ Cci ∗ |RPci | ) coefficients combined different objectives to form one objective function
i=1 (ei
CCB = (16) used by the optimization algorithm as a fitness function. However, ac­
Lifetime ∗ HA
cording to planning requirements, the distribution network planner can
AP
CGrid = CGrid ∗ APGrid (17) use other MCDM techniques like WPM, TOPSIS, and VIKER [21,22].
In SO, the coefficient of WSM is assigned one (1) to the APL objective,
Captial cost, Lifetime, O&Mcost and fuel cost of each DG are given in but in MO, the WSM coefficient is considered by the following equation.
Table 4, while APDG is the active power of each DG. APGrid is the active The value of each coefficient is assigned 1/3 (0.33) [3].
power produced by the grid and CAP Grid is active power cost at the grid,
which is considered as 0.044/kWh and Gγ is the annual benefit rate, α1 + α2 + α3 = 1
which is 0.05$/h [8]. HA is annual total hours that is 8760 and LFDG is
the power factor of DGs. Therefore, α and β denote fixed and variable Constraints
cost of generation of DG. ei and Cci are CBs installation and purchase
cost, which are equal to 1000 and 30000$/MVAR, respectively [1]. RPci While planning DN, the following equality and in-quality constraints
is the reactive power in MVAR of each CB. must be satisfied. These constraints are mathematically formulated as
follows:
Annual saving
Another objective of maximizing annual savings is using total annual Equality constraint
economic loss to reduce overall APL placing costs of DGs and CBs. Total The important key constraint is that the sum of generation at the grid
annual economic loss can be calculated using the Eq. (18) for a system and DG must equal the sum of Power demand, Active power loss, and
without DGs and CBs [16,22]. Reactive power loss in the network. Its mathematical formulation is as
follows [13,15,20,28]:
LEAwoDG = APwoDG
loss ∗ Cel ∗ HA (18)
∑
TDG

Where: Cel = cost of energy loss per kWh, taken as 0.05$ [16], and PGrid + PDGk = PLoad + APloss (22)
k=1
LEAwoDG define the value of annual economic loss without DGs.
The total annual economic loss can be calculated with the cost of DGs ∑
TDG ∑
tb
and CBs by the Eq. (19): QGrid + QDGk = QLoad + RPloss (23)
( ) k=1 i=1
TDG
∑ TDG
∑
wDG
LEAwDG = APloss ∗ Cel ∗ HA + CDGj + CCBi (19) The unbalanced three-phase power flow must be satisfied by Eqs.
j=1 i=1 (24) and (25).
Where, LEAwDG represent the value of annual economic loss with ∑
tb
( )
DGs. Pα = Vα Vβ Yα⋅β ⋅cos θα⋅β − δα + δβ (24)
The Eq. (20) calculates the overall annual savings. β

f6 = max(AnnualSavings) = max(LEAwoDG − LEAwDG ) (20) ∑

tb
( )
Qα = Vα Vβ Yα⋅β ⋅sin θα⋅β − δα + δβ (25)
β
Formulation of multi-objective function
In-equality constraint
All the above objectives have their effects on the placement of DG Voltage limits:
types. According to previous studies like in [15,26–29], we take the Consumer end equipment is usually voltage-sensitive, so voltage
multi-objective function as a combination of APL, IVD, and IVS for must be within the permissible limits. According to the standard, the
placement of DGs. In this article, we formulate the multi-objective allowable limit is ±5. The voltage equation at each bus must be satisfied
function, as described below. by the Eq. (26) [1,16]:
MO = min(α1 ∗ fAP + α2 ∗ fIVD + α3 ∗ fIVS ) (21) 0.95p⋅u⩽|Vi |⩽1.05p⋅u (26)
Where: DG Operating limits:
f1wDG f2wDG f3wDG DGs and CBs must be within the specified limit of 0 kW to 2000 kW.
fAP = woDG
f1
, fIVD = woDG
f2
, fIVS = woDG
f3 DG generation limits are formulated by Eqs. (27) and (28):

To investigate the emission of pollutant gases and economic aspects APmin max
DG ⩽APDG,t ⩽APDG (27)
of DG placement, type 1 DG with PV and type 2 DG with CB are placed
simultaneously, and customized characteristics are given in Table 4. RPmin max
DG ⩽RPDG,t ⩽RPDG (28)
In MO, different multi-criteria decision-making techniques (MCDM) Capacitor limits are as follows by Eq. (29):
are used [3]. Meanwhile, in this study, the weighted sum method (WSM)
is used in Eq. (23) for assigning equal weights to the objectives. The ∑
TCB ∑
tb
QCB < QLoad (29)
WSM technique is used for generating non-dominated solutions by its i i
quite simple and effective way of calculating the equal sum. The WSM

Table 4
DGs Characteristics [1]
DG Lifetime Capital Cost Fuel Cost O&M Cost Emission (lb/MWh)
Type (y) ($/kW) ($/kWh) ($/kWh) NOx SO2 CO2

Grid – – 0.044 – 5.01 11.3 2042

PV 20 3985 – 0.01207 – – –
CG 10 1224 0.0667 0.06481 0.279 0.93 1239.2
WT 20 1822 – 0.00952 – – –

6
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

DG Power Factor: LB and UB are the lower and upper bounds of the grazing field, σ is a
The power factor must also be within the limits of [0.7, 1] [3]: binary vector, and r is a random number.
The pseudo-code for GOA is given as follows.
min
p⋅ fDG,κ max
⩽p⋅ fDG,κ ⩽p⋅ fDG,κ (30)
Pseudo Code of GOA
Input: Set Search Agent as Gazelles, Dimension, Maximum Iteration
The following Eq. (31) calculates the power factor at the DG location:
Output: Global Optimal best fitness value of the objective function and its location
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ Start
p⋅ fDG = P2DG / P2DG + Q2DG (31) Step 1: Initialization alpha, beta, and delta position and Memory Wolves
Step 2: Set PSRs=0.34, Speed of gazelles=88kmph
Step 3: While iteration < Max Iteration (Start)
Proposed algorithm Evaluating the Top Gazelle Phase
Step 4: For i = 1 to Search agent
The Gazelle optimization algorithm and Mountain Gazelle optimi­ Step 5: Check boundaries and update the Top Gazelle fitness and
position at the random population.
zation algorithm are nature-inspired metaheuristic algorithms intro­
End (Search agent)
duced by Jeffrey O. Agushaka [34] and Benyamin Abdollahzadeh, Fitness value tracking
respectively [35]. Both algorithms are based on gazelle behavior, but Step 6: Prey old gazelle and evaluate the fitness value
GOA revolves around the survival of gazelles in predator-dominated Define the Elite gazelles.
fields, while MGOA models gazelles’ hierarchical and social life. These Step 7: If r < 0.5
Update the gazelles
algorithms are explained in the following subsections.
Step 8: Else
Step 9: If iteration > size(gazelle)/2
Update the gazelles by adding elite gazelles to the herd.
Gazelle optimization algorithm
Step 10: Else
Update the gazelles by using the predator population.
The GOA is based on the survival instinct of gazelles; when the
End
predators are spotted in the field, gazelles start to reach a safe spot. It End
contains two phases for reaching a safe zone: exploitation and exploring Updating top gazelle Phase
the whole field. Step 11: For i = 1 to Search agent
Step 12: Update fitness values and Top gazelles and Elite gazelles.
End
Exploitation Apply PSRs Phase
In this phase, gazelles are grazing in the field without fear of any Step 13: If rand () <PSRs
predator; on the other hand, the predator is stalking the movement of Update the gazelles using step sizes.
gazelles. Gazelle covers the random displacement with normal distri­ Step 14: Else
Update the gazelles by using the predator population.
bution probability of the field in surrounding areas. The mathematical
End
equation for this behavior is followed by the Eq. (32). End (While)
( ) Step 15: Return to global best fitness values and position of Top gazelle
gt+1 = gt + s• γ ∗ • γβ ∗• Et − γ β ∗• gt (32) End Stop

Where, gt+1 and gt are gazelles in the next and current iteration, t
denotes the number of iterations. While s denotes the current speed of
gazelles, and Et is the elite gazelle in the current iteration, and γ,γβ are Mountain gazelle optimization algorithm
uniform and random numbers for the displacement of gazelles in the
field, respectively. The mountain gazelle optimization algorithm is a model of wild
gazelles’ hierarchy and social life that move a very long distance each
Exploration day to search for food and can run very fast. The basic concept of ga­
This phase starts as a predator is spotted in the grazing field to chase zelles’ social life is used for modeling MGOA. It works on four main life
the gazelle. As gazelles feel in danger, they start flicking their tails and factors: Territorial solitary males, Maternity herds, Bachelor male herds,
stomping their feet, and this works on the Levy flight function, which and migration for food. In this proposed algorithm, exploration and
performs the random movement of gazelles in the field. The predator exploitation phases are carried out in parallel with these four main
starts running behind the gazelles at full speed but moves in each iter­ phases, which help to find the best solution in the minimum time. These
ation until they reach a safe zone. Gazelles move in one direction in four main factors are explained in the following subsections.
even-numbered iterations while moving in other directions in odd-
numbered iterations. As this phase starts, gazelles move first, and the Territorial solitary males
predator moves after the gazelles. This whole behavior of the gazelles When the newly born gazelles reach the age of adulthood and
after spotting the predator is modeled in the following Eq. (33). become strong enough, they start taking part in battles to take posses­
(
gt+1 = gt + S• μ• γ ∗ • γχ ∗• Et − γχ ∗• gt
)
(33) sion of females and over the territories. The newly young gazelles try
their best to gain possession of one of them. On the other hand, adult
Where S is the max speed of a gazelle that is 88 kmph, and μ is the gazelles try to protect their surrounding environment. The following Eq.
speed difference between a gazelle and a predator. The predator-chasing (36) models this behavior of male gazelles.
behavior is modeled by Eq. (34). ⃒ ⃒
TSM = Mg − ⃒γ t1 × Yg,H − γt2 × gt × F⃒ × RCV r (36)
gt+1 = gt + S• μ• CF ∗ • γ β ∗• (Et − γℓ ∗• gt ) (34)
Here,Mg is adult male gazelles, γt1, γ t2, ... are random numbers which
Where CF denotes the cumulative sum effect of the predator. The indicate the random places in gazelles territory, Yg,H is a newly young
exploration part also models the survival of gazelles, which is about male from the herd while gt is young gazelles in the current iteration, F is
66%, which means that predator success rates (PSRs) are only 34% in the exponential value of each iteration that try to converge the algo­
any attack. This success rate affects the whole model by Eq. (35). rithm in each iteration, And RCV is a random coefficient vector which
{ updates in each iteration and increases the searching capability.
gt + CF[LB + γ∗• (UB − ( LB) ]∗• σ ) if r⩽PSR
gt+1 = (35)
gt + [PSR(1 − r) + r ] gγ1 − gγ2 else

7
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Maternity herds “Backward forward sweeping” load flow analysis is used in this
This phase is vital when maternity herds give birth to new solid study, which calculates Active Power loss as 210.05 kW before the DG
gazelles. Here, young male gazelles try to take possession of females. placement in IEEE 33 bus system [1,3,16,28]. As the number of DGs
This behavior is modeled as follows by Eq. (37). increases, the cost of placement increases sufficiently. At the same time,
( ) ( ) active power losses do not decrease as much [25], so only three DG
MH = Yg,H + RCV + γ t3 × Mg − γt4 × grand × RCV (37)
placements are studied in each scenario.
Here grand is any random gazelle from all populations and Mg is an
adult male gazelle. Case Studies

Bachelor male herds Three case studies are proposed to analyze the effectiveness of the
In this phase, Male gazelles try to create their territory and start proposed metaheuristic algorithms at different loading conditions of
taking part in battles for females, creating violence in the environment. different load models.
This behavior of gazelles is mathematically formulated as follow by Eq. Case 1: SO of minimization APL with all four DG combinations.
(38). Case 2: MO of minimization APL, IVD, and IVS with all four different
( ) DG combinations.
BMH = (gt − D) + γt5 × Mg − γ t6 × Yg,H × RCV (38) Case 3: Environment Emission and Techno-economic aspects of the
Here, D denotes a random number and updates in each iteration. above two cases are examined.

Searching for food Implementation

Mountain wild gazelles cover long distances for food and always
migrate from one place to another. They do so because their running The implementation of the optimal allocation problem for this article
speed is around 88 kmph, and they have good jumping power, which is divided into different steps, which are described as follows:
helps them make it easy. This behavior is formulated by Eq. (39). Step 1: Simulate backward and forward sweep load flow methods for
calculating the Basis values of each objective.
SFF = (ub − lb) × γ t7 + lb (39) Step 2: Initializing the minimum and maximum values of DGs, DGs’
Where, ub and lb are lower and upper bound of grazing field. location, and their power factor values as lower and upper values of the
These four key factors work in parallel to generate a new population optimization algorithm.
of gazelles and are arranged in ascending order at the end of each iter­ Step 3: Define the objective function by the WSM technique at the
ation. Therefore, those with the best skills are kept as the best gazelles. end of the load flow method.
In contrast, the weak and older gazelles are removed from a population Step 4: Define the parameters of the proposed algorithm, like the
that converges towards the best solution after each iteration. The number of iterations and number of gazelles as search agents.
pseudo-code for this algorithm is shown as follows. Step 5: Assign the value of steps 2,3 and 4 to each proposed algorithm
and simulate the results according to cases 1 and 2.
Pseudo Code of MGOA
Input: Set Search Agent as Gazelles, Dimension, Maximum Iteration Step 6: Put these values in the equations of case 3 and simulate the
Output: Global Optimal best fitness value of the objective function and its location results of case 3.
Start Step 7: Compare the results with existing studies
Step 1: Initialization Random population of Gazelles and fitness value
The diagram in Fig. 3 illustrates the implementation steps of pro­
Step 2: Calculate the fitness of the population and Update the Best Gazelle
Start While iteration < Max Iteration (Start) posed problem.
Step 3: Calculate the vector of coefficients
Step 4: Update the location Results and discussion
Step 5: For each Gazelle
Check boundaries and calculate the cost function.
Proposed meta-heuristic algorithms GOA and MGOA are evaluated
End by performing load flow analysis at IEEE 33 bus system. The sizes of the
Step 6: Adding new gazelles to the herd
gazelles and the number of iterations is kept constant at 50 and 100,
End (While)
Step 7: Update the herd by sorting the herd in ascending order respectively, for a broader evaluation of results in both algorithms. The
Step 8: Return to global best fitness values and position of the herd as best gazelle simulated results are compared with other studies implemented on
END Stop MATLAB for the viability of the recently proposed algorithm with other
well-recognized algorithms. The problem of optimally placing DGs and
CBs is coded in MATLAB version of R2022b with an Intel i5 6th gener­
ation laptop with 16 GB of RAM.
Implementation scenarios The results of optimal sitting and sizing problems for all proposed
cases at IEEE 33 bus system are explained in detail in subsections.
In this article, the optimal sitting and sizing of DGs and CBs are
deployed on five different load models, like CP, RL, CL, IL, and ML with Case 1
load growth, by using two novel algorithms, which explained in previ­
ous section 4, and compared their results with well-established algo­ Normal loading
rithms that are reported in different previous research articles Table 5 summarizes the findings obtained by SO of loss minimization
[10,13,14,26–28]. The performance of the proposed algorithms is by the proposed MGOA by placing the DGs of all combinations at NL
evaluated on IEEE RDS: the IEEE 33 bus system. The technical details (normal load) conditions of the CP model. It also compared the results
like lines and load data of these benchmarks’ models are taken from the with other previously reported results. The most promising results of
website of MATPOWER [45]. each combination are bold, and the overall best results are bold and
The IEEE-33 Benchmark network consists of 33 buses with 32 underlined. The obtained results of APL minimization by GOA and
branches, including three lateral feeders with two looping lines oper­ MGOA are appreciably lower than the LSF [5], EA [9], HGWO [10],
ating at 12.66KV with constant active and reactive power of 3.715MW SKHA [11], SFSA [12], WCA [1], OCDE [13], DE [13], IRRO [14], I-
and 2.3 MVAR, respectively [25,45]. The single-line diagram (SLD) for DBEA [15], WOA [16], and SOS-NNA [2]. The APL of DN is minimized
the IEEE-33 network is depicted in Fig. 2. up to 70.6397 KW by MGOA and 70.6563 KW by GOA by placing DGs of

8
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Fig. 2. SLD of IEEE 33 Bus System

Fig. 3. Flowchart of proposed problem

combination 1, compared to APL of 210.0704 kW in the base case. Ac­ proposed algorithm. The results are compared with those of existing
cording to the authors’ observation, better APL minimization results methods reported in the literature review, as summarized in Table 5.
than MGOA and GOA have still not been reported in any literature. The obtained results also show that the proposed algorithm shows
In combination 2, where all DGs are assumed to operate at 0.85 promising results in comparison to GA [14], PSO [14], LSF [4], HAS-
lagging power factor, DGs are optimally sized and placed by the PABC [4], IMOEHO [29], QOGWO [20], I-DBEA [12]. The APL is

9
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Table 5
Comparative results of SO for VID at NL of IEEE 33 Bus system using all DG combinations.
Method Location DG Size APL R IVD IVS IVS− 1 Min V

(KW) (%) (p.u) (p.u) (p.u) (p.u) B

Base Case 210.0704 0.13277 0.66594 1.5016 0.90422 18

Combination 1 MW – – –
LSF[5] 6 3 28 1.369 0.791 0.820 99.79 52.68 – – – 0.9398 18
HAS-PABC[5] 30 24 14 1.068 1.073 0.755 72.81 65.49 – – – 0.9684 33
EA-OPF[9] 13 24 30 0.802 1.091 1.054 72.79 – – – – – –
EA[9] 13 24 30 0.798 1.099 1.050 72.787 – – – – – –
HGWO[10] 13 24 30 0.802 1.090 1.054 72.784 65.3525 – – – 0.9715 18
KHA[11] 24 14 30 0.915 0.750 1.142 73.2968 65.1084 – – – 0.9701 33
SKHA[11] 30 24 13 1.054 1.091 0.802 72.7853 65.3519 – – – 0.9687 33
SFSA[12] 13 24 30 0.8020 1.0920 1.0537 72.785 65.352 – – – – –
WCA[1] 14 24 29 0.8546 1.1017 1.181 71.052 66.177 – – – 0.973 33
OCDE[13] 13 24 30 0.8018 1.0915 1.0466 72.848 65.322 – – – 0.9686 33
DE[13] 17 19 30 1.0925 0.7714 1.0610 72.853 65.3197 – – – 0.9689 33
IRRO[14] 13 24 30 0.801 1.090 1.054 72.78 65.3544 – – – – –
IDBEA[15] 13 24 30 1.0980 1.0970 1.7150 94.8514 54.8477 – – – – –
WOA[16] 30 24 13 1.0536 1.0913 0.8017 72.787 65.3511 – – – 0.9687 18
SOS-NNA[2] 13 24 30 0.8018 1.0913 1.0536 72.7853 65.3519 – – – – –
GOA[P] 24 14 30 1.044 0.766 1.100 70.6563 66.3654 0.011933 0.89295 1.1199 0.97289 33
MGOA[P] 24 14 30 1.067 0.780 1.092 70.6397 66.3733 0.011541 0.89302 1.1198 0.97291 33
Combination 2 (MW)
GA/PSO 13 24 30 0.7226 1.051 1.196 14.56 – – – – – –
[6]
LSF[5] 6 3 28 1.1289 0.723 1.138 49.76 – – – – 0.9781 18
HAS-PABC[5] 12 30 25 0.862 1.159 0.816 15.91 – – – – 0.9944 18
IMOEHO 13 24 30 0.929 1.181 1.473 14.9 92.65 – – – – –
[28]
QOGWO 13 24 30 0.7569 1.0126 1.2116 15.352 – – – – 0.9921 8
[22]
IDBEA[15] 13 24 30 0.7491 1.0420 1.2395 14.570 – – – – – –
GOA[P] 13 30 24 0.726 1.206 1.000 14.4742 93.1098 0.00067379 0.96562 1.0356 0.99208 8
MGOA[P] 13 30 24 0.740 1.199 1.004 14.4628 93.1153 0.00059507 0.96712 1.034 0.99246 8
Combination 3 (MVAR)
BFOA[40] 18 30 33 0.349 0.821 0.277 144.04 – – – – 0.936 –
WCA[1] 14 24 30 0.3973 0.4511 1.000 136.7196 – – – – 0.951 18
GOA[P] 30 24 14 1.094 0.486 0.414 136.5576 34.9943 0.050816 0.7795 1.2829 0.94045 18
MGOA[P] 24 30 14 0.507 1.052 0.417 136.5045 35.0197 0.051823 0.77858 1.2844 0.94017 18
Combination 4 (MW/ MVAR)
CFPSO[36] 12 23 31 0.6737 0.6905 1.1359 26.84 87.27 – – – – –
3 11 29 0.577 0.205 1.120
NNA[2] 14 24 30 0.7381 1.1602 1.0588 15.1717 92.80 – – – – –
17 30 8 0.150 0.900 0.650
GOA[P] 12 25 30 0.894 0.663 0.872 14.5905 93.0544 0.0015907 0.94293 1.0605 0.98621 18
30 12 25 1.011 0.467 0.372
MGOA[P] 14 24 30 0.765 1.076 1.043 11.7249 94.4186 0.00064305 0.96367 1.0377 0.99158 8
24 30 13 0.516 1.009 0.370

reduced to 14.4628 KW, which is a remarkable reduction. minimized up to 14.5905 KW by GOA, which is also greater than
In combination 3, where DGs are operated at ZPF and assumed as CBs 14.4628 KW of type 2 DG. From Table 5, it is also observed that although
with a power rating in MVAR, they are optimally placed for APL the objective was only loss minimization, the stability of the network
reduction. Here, it can also be seen that the proposed algorithm shows and voltage deviation also improved and satisfied the allowable limits.
its significance by reducing the most promising loss compared to exist­ The convergence characteristics of the proposed GOA and MGOA at
ing literature review results at the same DGs. The APL is not reduced as each combination are shown in Fig. 4. Fig. 4 depicts the efficiency of the
much compared to combination one and two. However, compared to proposed MGOA algorithm that converged to the solution with greater
existing methods, it is appreciably lower at the proposed GOA and speed, given the same number of gazelles and iterations. Fig. 4(a) de­
MGOA. Results are also summarized in Table 5, where the signification picts the convergence characteristics of GOA and MGOA at combination
of proposed algorithms can be accessed by comparing them with BFOA 1, Fig. 4(b) for combination 2, Fig. 4(c) for combination 3, and Fig. 4(d)
[39] and WCA [9]. for combination 4. Moreover, Fig. 4 indicates that the most remarkable
The preferred combination for this article is combination 4, where results are achieved by combination 4, and it can also be observed that
DG of combination one and three are simultaneously placed for our GOA lost its stability when the number of DGs increased because GOA
problem of reduction of APL. In combination four, it can be observed has a low convergence rate compared to MGOA.
that MGOA substantially reduces losses compared to all other combi­ The results of the optimal solution of APL reduction by the proposed
nations, which is a superior outcome still not reported in any published MGOA algorithm at all combinations are illustrated in Fig. 5. These
literature. The results are also summarized in Table 5. On the other promising results show that system stability indices like voltage devia­
hand, the proposed GOA algorithm lost its significance because the tion index and voltage stability index are also improved with loss
number of DGs increased and did not give promising results, but the reduction. Clearly, the results in Fig. 5(d) show that the results are
proposed MGOA did. According to the author’s observation, the APL of significantly improved as compared to Fig. 5(a), (b), and (c) because the
DN is minimized up to 11.7249 KW by MGOA, which also enhances the power losses of Fig. 5(d) are minimal.
system’s stability, performance, and security. In comparison, the APL The results of SO for APL reduction for all other proposed loads

10
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Fig. 4. Convergence characteristics for IEEE 33-bus network for Active Power Loss at all combinations (a) for C1, (b) for C2, (c) for C3, and (d) for C4

models of IEEE 33 bus system are summarized in Table 6 using all DG efficiency every time in combination four due to an increase in the
combinations. The results are compared with well-established optimi­ number of DGs.
zation techniques of existing literature. In each load model, MGOA
provides the finest solution in each combination, while the most High Loading
remarkable results are from combination four because it is a good way to After investigating all proposed load models at normal loading
add active and reactive power in the RDS. The voltage profile of each conditions, the back-forward sweep method shows that the CP model
proposed load model is significantly improved at each combination has the highest APL of 210.0704 KW compared to all other models in the
concerning their base case, which is depicted in Figs. 6 and 7. Moreover, IEEE 33 Bus DN. So due to the highest losses in CP, the HL (high load)
the voltage profile of combinations 2 and 4 are near unity, which is our scenario is proposed for only the CP model for future planning and
desired voltage profile for reliable operation and planning of the dis­ optimally size and place all combinations of DGs on it. The number of
tribution system. While the voltage regulation of combination 2 is good gazelles and the number of iterations is kept the same, and the number of
so far because they exist in allowable limits, the voltage regulation of DGs and the rating of each DG for the optimization problem is also the
combination 3 could be better compared to other proceeding combi­ same. The base case of the HL of the CP model shows that the APL is
nations. It may be observed that combination 4 provides active and 515.5962KW, the IVD is 0.3273 p.u, and the IVS is equal to 0.5169 p.u.
reactive powers that also enhance the voltage regulation of the system Table 7 tabulated the results of the optimal allocation of all DG
by improving voltage stability and voltage deviation. GOA lost its combinations at CP with the HL scenario. Results show the superiority of

11
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Fig. 5. Optimal solutions obtained by Proposed MGOA at all combinations of DGs (a) for C1, (b) for C2, (c) for C3, and (d) for C4.

the proposed algorithm by evaluating the results with a well-established Additionally, Table 7 concludes the results of combination four,
optimization algorithm. In combination 1, the comparison shows that where DGs of UPF are placed with CBs. These results show the advan­
the proposed algorithms are efficient and superior when the load in­ tages and significance of using DGs in parallel with CBs for loss depre­
creases and provide the most promising results of APL as compared to ciation to improve the overall efficiency and stability of the active radial
LSF [4], HAS-PABC [4], KHA [7], and SKHA [7]. distribution network. Losses are minimized from 515.5962 KW to
After that, DGs from combination two operating at a fixed lagging 29.9836 KW with a reduction of 94.1847%, a remarkable achievement
power factor of 0.85 are placed for the problem of optimal allocation at using a combination of four. GOA also does not converge the solution as
CP using HL scenarios. Table 7 shows that the established MGOA algo­ MGOA does in combination 4 with increasing the number of DGs.
rithm proposed the best results for our loss minimization problem at the The voltage profile of the CP model at HP is significantly improved.
CP model at normal loading from the preceding DGs’ combination. At the same time, combinations 2 and 4 are near unity, which is
Furthermore, the effectiveness of the established algorithm can be appreciable good voltage regulation, as shown in Fig. 8, and the zoomed
conceptualized by comparing results with LSF [5] and HAS-PABC [5] by figure shows that the voltage using GOA is lower than the MGOA.
using combination 2 of DGs. Moreover, the DGs of combination 3 add
reactive power to the DN, which enhances the efficiency and stability of
the network but not as much as proceeding combinations of DGs.

12
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Table 6
Comparative results of SO for VD at NL of IEEE 33 Bus System using all DG combinations.
C Method Location DG Size (MW) APL R IVD IVS IVS− 1 Min V

(KW) (%) (p.u) (p.u) (p.u) (p.u) B

RL Base Case 158.7738 0.097615 0.70883 1.4108 0.91827 18

QOGWO[22] 14 24 30 0.7111 1.0524 0.9123 60.972 – – – – 0.9673 33
C1 GOA[P] 25 14 30 0.839 0.745 0.974 42.6564 73.1339 0.0073799 0.91141 1.0972 0.97774 33
MGOA[P] 14 24 30 0.714 1.042 0.997 42.0608 73.509 0.0071537 0.91556 1.0922 0.97885 33
C2 GOA[P] 30 24 13 1.026 0.950 0.674 11.6435 92.6666 0.00032925 0.97398 1.0267 0.99408 22
MGOA[P] 30 24 13 1.023 0.956 0.650 11.6008 92.6935 0.00044252 0.97205 1.0288 0.99359 8
C3 GOA[P] 30 14 5 0.776 0.316 0.365 118.7537 25.2057 0.044632 0.79482 1.2581 0.94489 18
MGOA[P] 14 24 30 0.304 0.449 0.771 118.7051 25.2364 0.045152 0.79302 1.261 0.94436 18
C4 GOA[P] 29 24 13 1.095 1.027 0.722 13.0174 91.8013 0.00051478 0.96947 1.0315 0.99293 18
10 30 2 0.280 0.697 1.160
MGOA[P] 24 30 13 1.040 0.960 0.733 9.9965 93.7039 0.00051193 0.97043 1.0305 0.99318 18
24 12 30 0.450 0.295 0.722
CL Base Case 152.3352 0.093905 0.71434 1.3999 0.92004 18
QOGWO[22] 13 24 30 0.7326 1.0433 0.8999 62.337 – – – – 0.9674 33
C1 GOA[P] 30 13 24 0.903 0.741 1.046 44.9005 70.5252 0.0073723 0.90973 1.0992 0.97728 33
MGOA[P] 13 30 24 0.705 0.947 1.018 44.8098 70.5848 0.0076576 0.91147 1.0971 0.97775 18
C2 GOA[P] 30 14 24 1.017 0.610 0.925 10.9749 92.7955 0.0004803 0.9691 1.0319 0.99283 8
MGOA[P] 13 24 30 0.629 0.946 1.010 10.9167 92.8337 0.0004486 0.97187 1.0289 0.99354 8
C3 GOA[P] 25 14 30 0.376 0.312 0.809 109.2243 28.3 0.04201 0.79999 1.25 0.94641 18
MGOA[P] 30 14 24 0.806 0.312 0.460 109.1672 28.3375 0.041721 0.80059 1.2491 0.94659 18
C4 GOA[P] 31 24 12 0.860 1.029 0.753 13.2 91.3349 0.0015052 0.94011 1.0637 0.98533 18
12 27 30 0.229 0.188 0.616
MGOA[P] 30 13 24 0.919 0.698 1.021 9.5344 93.7412 0.00049438 0.97004 1.0309 0.99307 8
24 13 30 0.465 0.278 0.773
IL Base Case 163.2323 0.098169 0.70732 1.4138 0.91777 18
QOGWO[22] 14 24 30 0.7202 1.0495 0.9014 56.796 – – – – 0.9674 33
C1 GOA[P] 30 13 24 1.020 0.795 1.001 34.6355 78.7815 0.0061001 0.9213 1.0854 0.98037 18
MGOA[P] 14 24 30 0.758 1.068 1.046 34.5571 78.8295 0.0058427 0.92369 1.0826 0.981 33
C2 GOA[P] 30 24 13 1.043 0.925 0.661 13.8138 91.5374 0.00036985 0.97366 1.0271 0.99399 8
MGOA[P] 13 30 24 0.668 1.015 0.961 13.7361 91.5849 0.00040081 0.97348 1.0272 0.99395 8
C3 GOA[P] 14 30 25 0.281 0.698 0.334 131.9666 19.1541 0.048751 0.78547 1.2731 0.9421 18
MGOA[P] 24 14 30 0.420 0.273 0.671 131.9228 19.1809 0.049712 0.78358 1.2762 0.94153 18
C4 GOA[P] 24 13 30 1.034 0.761 0.928 11.6664 92.8529 0.0017301 0.95022 1.0524 0.98796 18
30 10 25 0.587 0.179 0.235
MGOA[P] 30 13 24 1.016 0.779 1.063 10.5139 93.5589 0.00051978 0.97101 1.0299 0.99331 18
12 30 24 0.255 0.614 0.422
MIXED Load 159.4559 0.097365 0.70888 1.4107 0.91828 18
C1 GOA[P] 14 30 24 0.739 0.988 1.021 38.925 75.5889 0.0068388 0.91573 1.092 0.97889 33
MGOA[P] 30 14 24 1.013 0.729 1.051 38.8924 75.6093 0.0066178 0.91877 1.0884 0.9797 33
C2 GOA[P] 13 30 24 0.668 1.000 0.939 12.1785 92.3625 0.00043676 0.97289 1.0279 0.9938 8
MGOA[P] 30 24 13 1.017 0.957 0.655 12.1569 92.376 0.00042631 0.97259 1.0282 0.99372 8
C3 GOA[P] 14 30 24 0.293 0.739 0.445 123.0459 22.8339 0.046254 0.7906 1.2649 0.94363 18
MGOA[P] 24 14 30 0.438 0.291 0.731 123.042 22.8364 0.046659 0.78984 1.2661 0.9434 18
C4 GOA[P] 30 13 24 0.857 0.828 0.982 12.3348 92.2645 0.00099471 0.96117 1.0404 0.99079 18
12 30 4 0.249 0.569 0.360
MGOA[P] 14 24 30 0.722 1.053 0.990 11.0606 93.0636 0.00054924 0.96761 1.0335 0.99245 33
30 13 3 0.672 0.237 0.665

Case 2 Additionally, the DGs of C4 are analyzed for the same problem where
the well-established optimization algorithm NNA [2] provides the
In this case, the MO optimization problem is analyzed to find the lowest APL, which could be better in voltage regulation. On the other
optimal size and place of DGs of all combinations at all proposed load hand, our proposed MGOA has a slightly great loss compared to the
models, which minimize the APL and IVD and maximize the IVS in IEEE proceeding algorithm, but voltage regulation is improved sufficiently.
33 Bus by using the proposed GOA and MGOA. Firstly, proposed algo­ The results for all other proposed loads models of the IEEE 33 bus
rithms are deployed on the CP model at NL condition to solve the MO system are tabulated in Table 9 for minimization of MO, where it can be
problem using all DG combinations. A comparison of our simulation observed that the proposed algorithm is much more efficient in the MO
results and existing literature is tabulated in Table 8. problem and according to an author literature review from the MO
The results of combination one show that the minimum APL is ob­ problem is not analyzed still so far in any literature. MGOA provides the
tained by SFSA [12]. However, our proposed MGOA achieved maximum finest solution in each combination, while the most remarkable results
minimization of IVD, significant improvements in IVS, and good voltage are from combination 4. GOA also lost its efficiency every time in
regulation. The overall results of MGOA at combination 1 are remark­ combination four in the MO problem due to an increase in the number of
able compared to well-established SFSA [12] and NNA [2]. Moreover, DGs. The voltage profile of each proposed load model is significantly
DGs of C2 and C3 are analyzed for the problem of MO by the author by improved at each combination for their base case, while the voltage
using both proposed algorithms, and their simulation results are also profile of combination four lies near unity, which is required for reliable
summarized in Table 8, which still need to be studied in any existing distribution system planning.
literature. Results of MGOA again show its efficiency and superiority For future planning, the same problem of MO minimization is also
over GOA by providing the most promising solutions, which enhanced analyzed on the IEEE 33 bus system at HL, where all combinations of
the voltage regulation with a depreciation of APL and IVD. DGs are optimally placed and sized, and their results are tabulated in

13
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Fig. 6. Voltages profiles of IEEE 33 Bus system at VID all combinations of DGs at NL

Fig. 7. Voltages profiles of IEEE 33 Bus at VID for all combinations of DGs at NL (a) for RL, (b) for CL, (c) for IL, and (d) for ML.

Table 10. Results show the superiority of the proposed algorithm. In all operating at unity power factor like solar and DGs operating at 0 power
combinations, the comparison shows that the proposed algorithms are factor like CBs, are minimum as compared to all other combination.
efficient and superior to GOA when the load increases and provides the While the DGs of Combination 2 provide active and reactive power, they
most promising results in combination four, where GOA lost its use coal, oil, and fossil fuels, polluting our environment by producing
efficiency. CO gases. In comparison, Combination 4 is a green source of active and
reactive power, causing lower power losses and lower the emission of
CO gases.
Case 3 Additionally, the amount of electricity purchased from conventional
powerhouses is lower when the problem of SO and MO is analyzed. The
In this case, economics and environment objective functions are optimal placement of DGs from combination 4 is slightly higher than
accessed at case 1 and case 2 for optimal allocation of all DG combi­ that of combination two because the total cost of combination 4 is the
nations. From Table 11, Table 12, Table 13, and Table 14, it can be sum of two DG types: active power and reactive power. Moreover, the
observed that total electrical energy cost and total emissions produced annual savings increased by placing the DGs of combination 4, which
by generation sources are a combination 4 of DGs, which consist of DG

14
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Table 7
Comparative results of SO for VID at HL of IEEE 33-Bus system using all DG combinations.
Method Location DG Size (MW) APL R IVD IVS IVS− 1 Min V

(KW) (%) (p.u) (p.u) (p.u) (p.u) B

Base Case 515.5962 0.32731 0.5169 1.9346 0.84929 18

Combination 1 MW
LSF[5] 6 3 28 2.221 1.002 1.474 270.39 55.19 – – – 0.9012 18
HAS-PABC[5] 30 24 14 1.784 1.702 1.202 194.17 67.82 – – – 0.9494 33
KHA[11] 30 12 24 1.659 1.478 1.775 195.3122 – – – – 0.9496 33
SKHA[11] 30 24 13 1.656 1.737 1.427 194.716 – – – – 0.9491 33
GOA[P] 14 30 24 1.200 1.642 1.604 160.8251 68.8079 0.022641 0.85077 1.1754 0.96162 33
MGOA[P] 30 14 24 1.687 1.187 1.597 160.7687 68.8189 0.022018 0.85574 1.1686 0.96302 33
Combination 2 MW
LSF[5] 6 3 28 2.166 0.799 1.719 132.80 77.99 – – – 0.9285 18
HAS-PABC[5] 12 30 25 1.311 1.944 1.329 41.11 93.18 – – – 0.9792 18
GOA[P] 24 30 13 1.455 1.700 1.140 33.1113 93.5781 0.0018872 0.94674 1.0563 0.98759 8
MGOA[P] 13 24 30 1.141 1.533 1.700 33.0107 93.5976 0.0017257 0.94829 1.0545 0.988 8
Combination 3 MVAR
GOA[P] 25 14 30 0.636 0.714 1.652 319.5657 38.0202 0.10123 0.70778 1.4129 0.9185 18
MGOA[P] 14 30 25 0.704 1.634 0.613 319.5265 38.0278 0.10351 0.70469 1.4191 0.9175 18
Combination 4 MW/MVAR
GOA[P] 13 24 30 1.035 1.427 1.789 34.6233 93.8477 0.0024942 0.93414 1.0705 0.9843 8
23 17 30 0.619 0.501 1.368
MGOA[P] 30 13 24 1.549 1.195 1.614 29.9836 94.1847 0.0011932 0.95469 1.0475 0.98966 8
30 6 14 1.328 0.748 0.449

Fig. 8. Voltages profiles of all combinations of IEEE 33 Bus for CP at HL

Table 8
Comparative results of MO for VID at NL of IEEE 33 Bus system using all DG combinations.
C Method Location DG Size (MW) APL R IVD IVS IVS− 1 Min V

(KW) (%) (p.u) (p.u) (p.u) (p.u) Bus

Base Case NL 210.0704 0.13277 0.66594 1.5016 0.90422 18

C1 SFSA[12] 13 24 30 0.965 1.1337 1.3018 77.410 63.31 0.006232 0.9182 1.0891 – –
NNA[2] 16 24 44 1.123 1.453 1.139 95.799 54.59 0.003395 – 1.079 – –
GOA[P] 13 24 30 1.023 1.012 1.392 77.6622 63.0304 0.0024214 0.94495 1.0583 0.98673 33
MGOA[P] 13 30 24 1.026 1.419 1.109 78.8319 62.4736 0.0019664 0.95015 1.0525 0.98809 33
C2 GOA[P] 30 24 13 1.202 1.100 0.772 14.7263 92.9898 0.00033403 0.97346 1.0273 0.99408 22
MGOA[P] 13 30 24 0.769 1.229 1.026 14.6155 93.0426 0.00032067 0.97331 1.0274 0.99404 22
C3 GOA[P] 7 14 32 0.913 0.724 1.156 166.0732 20.944 0.014657 0.90059 1.1104 0.97105 30
MGOA[P] 7 30 14 0.667 1.427 0.731 164.4694 21.7075 0.015047 0.88522 1.1297 0.97078 33
C4 NNA [2] 30 14 24 1.082 0.770 1.177 12.3754 94.13 0. 000252 – 1.024 – –
11 30 24 0.500 1.000 0.550
GOA[P] 32 12 25 1.010 1.008 0.812 18.4277 91.2278 0.0005736 0.97838 1.0221 0.99318 18
30 5 10 1.054 0.010 0.488
MGOA[P] 24 32 13 1.127 0.918 0.865 13.5125 93.5676 0.00038907 0.98076 1.0196 0.99398 22
12 30 24 0.451 1.011 0.412

minimized most of the APL in all proposed load models and reduced the NL, where it can be analyzed that the emission of combination 4 is less
annual economic losses. than other combinations in each load model. From Fig. 10, the annual
Fig. 9 illustrates the effect of each combination on each load model at savings at CP model at NL can be analyzed that combination 4 gives the

15
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Table 9
Comparative results of MO for VD at NL of IEEE 33 Bus System using all DG combinations.
C Method Location DG Size (MW) APL R IVD IVS IVS− 1 Min V

(KW) (%) (p.u) (p.u) (p.u) (p.u) B

RL Base Case 158.7738 0.097615 0.70883 1.4108 0.91827 18

C1 GOA[P] 32 13 24 1.045 0.955 1.101 47.5031 70.0812 0.0018901 0.95856 1.0432 0.98789 29
MGOA[P] 30 13 24 1.235 0.923 1.083 47.0388 70.3737 0.0013416 0.95832 1.0435 0.99006 8
C2 GOA[P] 11 30 24 0.876 0.883 0.942 13.0664 91.7704 0.00046804 0.9733 1.0274 0.99391 33
MGOA[P] 13 24 30 0.671 0.976 1.049 11.7003 92.6308 0.00027069 0.97412 1.0266 0.99412 22
C3 GOA[P] 7 32 15 0.828 0.890 0.570 143.1793 9.8218 0.013405 0.90346 1.1069 0.97259 30
MGOA[P] 32 7 15 0.896 0.798 0.579 143.1266 9.855 0.013424 0.90336 1.107 0.97254 30
C4 GOA[P] 10 17 32 30 25 0.999 0.242 0.842 0.914 15.7317 90.0917 0.00032317 0.98177 1.0186 0.99427 22
2 0.716 1.079
MGOA[P] 13 12 25 3 30 30 0.773 0.853 1.006 11.671 92.6493 0.00029726 0.97402 1.0267 0.99409 22
0.312 0.684 0.709
CL Base Case 152.3352 0.093905 0.71434 1.3999 0.92004 18
C1 GOA[P] 13 30 24 0.882 1.215 1.068 50.0741 67.129 0.0014197 0.9575 1.0444 0.98985 8
MGOA[P] 24 13 30 1.062 0.886 1.208 49.994 67.1816 0.00144 0.95741 1.0445 0.98982 33
C2 GOA[P] 24 30 14 0.774 1.108 0.623 11.705 92.3163 0.00034223 0.97378 1.0269 0.99402 22
MGOA[P] 12 24 30 0.761 0.948 0.994 11.4897 92.4576 0.00028483 0.97437 1.0263 0.99417 22
C3 GOA[P] 7 15 32 0.753 0.589 0.912 131.8525 13.4459 0.012679 0.9057 1.1041 0.97311 30
MGOA[P] 14 30 25 0.715 1.287 0.380 130.291 14.4709 0.013602 0.89484 1.1175 0.97326 33
C4 GOA[P] 13 30 24 0.707 0.965 1.063 10.1331 93.3482 0.00025272 0.97419 1.0265 0.99413 22
30 12 24 0.683 0.409 0.363
MGOA[P] 24 30 13 1.043 0.943 0.724 9.656 93.6613 0.00027438 0.97425 1.0264 0.99414 22
30 13 24 0.787 0.299 0.474
IL Base Case 163.2323 0.098169 0.70732 1.4138 0.91777 18
C1 GOA[P] 32 13 24 1.071 0.957 1.147 38.9104 76.1626 0.0018726 0.96002 1.0416 0.9882 8
MGOA[P] 24 30 13 1.113 1.246 0.967 38.775 76.2455 0.0010989 0.96131 1.0402 0.99083 8
C2 GOA[P] 30 24 14 1.062 0.793 0.681 14.3312 91.2204 0.00040398 0.97359 1.0271 0.99397 22
MGOA[P] 24 30 13 0.983 1.041 0.685 13.8167 91.5355 0.00026879 0.97411 1.0266 0.9941 22
C3 GOA[P] 14 7 32 0.628 0.683 0.854 159.0047 2.5899 0.014146 0.90028 1.1108 0.97184 30
MGOA[P] 15 7 32 0.580 0.752 0.848 158.752 2.7447 0.014254 0.90035 1.1107 0.97188 30
C4 GOA[P] 13 8 30 30 24 2 0.830 0.982 1.127 12.8787 92.1102 0.00024521 0.97452 1.0261 0.99421 22
0.477 0.590 0.508
MGOA[P] 14 30 24 24 30 11 0.775 1.090 0.398 1.049 10.6432 93.4797 0.00027255 0.97415 1.0265 0.99412 22
0.628 0.296
MIXED Load 159.4559 0.097365 0.70888 1.4107 0.91828 18
C1 GOA[P] 30 13 24 1.234 0.940 1.146 43.8424 72.505 0.0011567 0.9602 1.0415 0.99055 33
MGOA[P] 13 30 24 0.936 1.235 1.093 43.5242 72.7046 0.001252 0.9594 1.0423 0.99034 33
C2 GOA[P] 25 30 13 0.736 1.096 0.669 12.9955 91.8501 0.00031751 0.97355 1.0272 0.99397 22
MGOA[P] 30 13 24 1.043 0.675 0.978 12.2493 92.3181 0.00026943 0.97413 1.0266 0.99411 22
C3 GOA[P] 7 32 15 0.710 0.905 0.597 148.7841 6.6927 0.013521 0.90384 1.1064 0.97264 30
MGOA[P] 15 32 7 0.578 0.874 0.779 148.1813 7.0707 0.013723 0.90224 1.1084 0.97231 30
C4 GOA[P] 13 6 25 8 30 33 0.842 0.857 0.937 15.2168 90.457 0.00044936 0.97326 1.0275 0.99389 22
0.084 0.439 0.495
MGOA[P] 14 12 24 30 30 24 0.747 1.076 1.016 10.2767 93.5552 0.00026942 0.97419 1.0265 0.99413 22
0.303 0.694 0.448

Table 10
Comparative results of MO for VID at HL of IEEE 33 Bus system using all DG combinations.
C Method Location DG Size (MW) APL R IVD IVS IVS− 1 Min V

(KW) (%) (p.u) (p.u) (p.u) (p.u) Bus

Base Case HL 515.5962 0 0.32731 0.5169 1.9346 0.84929 18

C1 GOA[P] 30 13 24 2.000 1.574 1.779 174.6612 66.1244 0.0040523 0.92184 1.0848 0.98105 33
MGOA[P] 24 13 32 1.740 1.670 1.851 182.4914 64.6058 0.0032258 0.94522 1.058 0.98324 8
C2 GOA[P] 30 13 24 1.700 1.215 1.700 33.963 93.4129 0.00094878 0.96028 1.0414 0.9911 33
MGOA[P] 13 24 30 1.236 1.608 1.700 33.782 93.4479 0.0009695 0.96008 1.0416 0.99105 33
C3 GOA[P] 30 32 14 1.318 0.971 1.381 372.0363 27.8435 0.028794 0.86806 1.152 0.96313 25
MGOA[P] 32 14 29 1.086 1.355 1.295 372.7779 27.6996 0.028467 0.86995 1.1495 0.96327 25
C4 GOA[P] 32 25 13 1.512 1.159 1.224 34.2661 93.3541 0.0011569 0.97126 1.0296 0.99069 22
30 11 25 1.612 0.736 0.515
MGOA[P] 24 30 13 1.672 1.585 1.214 30.1973 94.1432 0.00060649 0.95966 1.042 0.99094 22
7 14 30 0.769 0.443 1.353

maximum annual savings, slightly greater than the annual savings of Conclusion
combination 2.
From a future planning perspective, case 3 is also analyzed on HL od This work presents the results of applying two novel meta-heuristic
CP, where it shows that combination four at HL is also suitable for high- algorithms, the Gazelle Optimization Algorithm and the Mountain Ga­
loading models where C4 gives the lowest GHG emissions and saves zelle optimization algorithm, for solving the problem of optimal allo­
most of the money by minimizing the annual economic losses, results are cation of DGs and CBs to minimize APL and IVD and improve IVD. Using
tabulated in Table 15. a single and multi-objective framework by weighted sum method, the

16
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Table 11 loss. In contrast, combinations of DGs, like combination one operating at

Economic and Environmental objective for VID of SO problem of IEEE 33 Bus unity power factor consisting of PV DG, combination two consisting of
system at NL conventional generators operated at fixed lagging power factor, and
Constant Power combination three consisting of DG operating at zeros power factor like
Emission Cost Annual loss Annual Saving
CBs, were also analyzed. The effectiveness and superiority of the pro­
posed GOA and MGOA are evaluated on the IEEE 33 bus system. The
(lb/h) ($/h) ($) ($)
results were compared with well-established analytical and optimiza­
B 8037208.836 172.7031 92010.8222 0
C1 1733630.862 76.0683 30979.0059 61031.8163 tion algorithms in the existing literature. The comparison of results in­
C2 5261614.631 49.76 6349.8356 85660.9866 dicates the efficiency of the proposed combination using the proposed
C3 7886562.49 169.703 59789.1899 32221.6323 algorithm over the existing methods.
C4 1727370.94 75.4352 5173.827 86836.9952 The most of practical systems are unbalanced and three phase while

Table 12
Economic and Environmental objective for VD of SO problem of IEEE 33 Bus system at NL
Residential Load Commercial Load

Emission Cost Annual loss Annual Saving Emission Cost Annual loss Annual Saving

(lb/h) ($/h) ($) ($) (lb/h) ($/h) ($) ($)

B 7570281.95 162.6698 69542.9311 0 7337233.875 157.6621 66722.8363 0
C1 1694091.572 72.7622 18458.9774 51083.9537 1649796.795 70.7142 19661.9441 47060.8922
C2 5146664.949 54.0438 5094.7058 64448.2253 4960909.963 50.9987 4794.835 61928.0013
C3 7488252.691 161.0925 51993.0071 17549.924 7248836.393 155.9541 47815.439 18907.3973
C4 1669396.792 72.1463 4414.7591 65128.172 1643080.471 70.3317 4211.1002 62511.7361
Industrial Load Mixed Load
B 7868604.512 169.0801 71495.73 0 7668847.907 164.7878 69841.7015 0
C1 1724252.58 74.9819 15173.96 56321.7682 1702847.161 73.4786 17071.7752 52769.9263
C2 5428128.174 59.7693 6030.02 65465.7094 5244964.71 56.1561 5338.2662 64503.4353
C3 7804492.277 167.8696 57782.35 13713.3808 7594272.13 163.3634 53892.5553 15949.1462
C4 1703673.597 74.5136 4643.024 66852.7051 1703195.263 73.3074 4881.2403 64960.4612

this study is carried out on single phase and balanced IEEE networks.
Table 13 Moreover, optimal DG allocation needs to consider the uncertainty of
Economic and Environmental objective for VID of MO problem of IEEE 33 Bus
renewable-based DG which can also be investigated in future studies.
system at NL
Therefore, this study can be extended in future for planning and oper­
Constant Power ation of a practical DN. In future studies, the established efficiency of the
Emission Cost Annual loss Annual Saving proposed algorithm on Optimal Distributed Generators Allocation
(lb/h) ($/h) ($) ($)
problem can be implemented for complex DN of IEEE and EPRI such as
B 8037209 172.7031 92010.82 0 IEEE 8500 bus system, for the achievement of technical and non-
C1 491985.1 57.5104 34575.32 57435.5066 technical objectives from planning point of view.
C2 5194274 46.5708 6417.154 85593.6685
C3 7943836 171.0306 72037.92 19972.9017
CRediT authorship contribution statement
C4 1676129 74.6751 5957.123 86053.6995

Muhammad Zubair Iftikhar: Conceptualization, Methodology,

scope of research work covers five different load models of DN with two Software, Investigation, Writing – original draft. Kashif Imran: Meth­
different loading scenarios from current and future planning perspec­ odology, Supervision, Resources, Writing – review & editing. Muham­
tives. The significant features of the proposed combination 4 of DGs mad Imran Akbar: Visualization, Writing – review & editing. Saim
containing PV in parallel with CBs were demonstrated in the minimi­ Ghafoor: Supervision, Writing – review & editing.
zation of technical objectives, such as APL, with non-technical objec­
tives, such as GHG emissions, reduction of energy purchase costs, and
maximization of annual savings by the minimization of annual economic

Table 14
Economic and Environmental objective for VD of MO problem of IEEE 33 Bus system at NL
Residential Load Commercial Load

Emission Cost Annual loss Annual Saving Emission Cost Annual loss Annual Saving

(lb/h) ($/h) ($) ($) (lb/h) ($/h) ($) ($)

B 7570282 162.6698 69542.9311 0 7337234 157.6621 66722.8363 0
C1 705030.8 57.9544 20648.4389 48894.4922 665261.4 55.9772 21940.5864 44782.2499
C2 5092784 51.4449 5138.2682 64404.6629 4866822 46.4389 5045.8103 61677.026
C3 7538277 162.253 62693.7838 6849.1473 7292103 156.9756 57070.6105 9652.2258
C4 1879630 75.357 5146.6091 64396.322 1495915 68.1253 4265.1261 62457.7102
Industrial Load Mixed Load
B 7868605 7668848 7668848 7668848 7668848 169.0801 71495.7293 0
C1 803256.1 747859.5 747859.5 747859.5 747859.5 61.1877 17029.4823 54466.247
C2 5375821 5191063 5191063 5191063 5191063 57.2472 6065.1579 65430.5714
C3 7859411 7645750 7645750 7645750 7645750 169.1428 69526.7622 1968.9671
C4 1589271 1550071 1550071 1550071 1550071 72.7985 4700.5618 66795.1675

17
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Fig. 9. Comparison of GHG emissions by using all combinations of DGs at proposed load models

Fig. 10. Annual Economic Loss and Saving at IEEE 33 Bus system for VID at NL

Table 15
Economic and Environmental objectives for SO and MO problem of IEEE 33 Bus system at HL
Constant Power HL

SO MO

Emission Cost Annual loss Annual Saving Emission Cost Annual loss Annual Saving

(lb/h) ($/h) ($) ($) (lb/h) ($/h) ($) ($)

B 12466358.85 267.8764 225831.1519 0 12466358.85 267.8764 225831.1519 0
C1 2586407.262 114.6263 70475.7523 155355.3996 1013992.382 91.2721 80000.7255 145830.4264
C2 7948123.177 76.708 14481.2033 211349.9486 7811694.405 70.1199 14819.9577 211011.1942
C3 12064874.15 259.5976 139952.9668 85878.1851 12173912.04 262.0302 163277.1677 62553.9842
C4 2550489.993 112.6617 13190.6708 212640.4811 2318276.195 109.1689 13285.75 212545.4019

18
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

Declaration of competing interest [21] S. Anbuchandran, R. Rengaraj, A. Bhuvanesh, M. Karuppasamypandiyan, A Multi-

objective Optimum Distributed Generation Placement Using Firefly Algorithm,
J. Electr. Eng. Technol. 17 (2) (2022) 945–953, https://doi.org/10.1007/s42835-
The authors declare that they have no known competing financial 021-00946-8.
interests or personal relationships that could have appeared to influence [22] S. Kumar, K.K. Mandal, N. Chakraborty, Optimal placement of different types of DG
the work reported in this paper. units considering various load models using novel multi-objective quasi-
oppositional grey wolf optimizer, Soft Comput. 25 (6) (2021) 4845–4864, https://
doi.org/10.1007/s00500-020-05494-3.
Data availability [23] C. Yammani, S. Maheswarapu, S.K. Matam, Electrical Power and Energy Systems A
Multi-objective Shuffled Bat algorithm for optimal placement and sizing of multi
distributed generations with different load models, Int. J. Electr. Power Energy
Data will be made available on request. Syst. 79 (2016) 120–131, https://doi.org/10.1016/j.ijepes.2016.01.003.
[24] E.A. Al-Ammar, et al., ABC algorithm based optimal sizing and placement of DGs in
References distribution networks considering multiple objectives, Ain Shams Eng. J. 12 (1)
(2021) 697–708, https://doi.org/10.1016/j.asej.2020.05.002.
[25] T.E. Gümüş, S. Emiroglu, M.A. Yalcin, Optimal DG allocation and sizing in
[1] A.S. Abbas, Optimal Placement and Sizing of Distributed Generation and Capacitor
distribution systems with Thevenin based impedance stability index, Int. J. Electr.
Banks in Distribution Systems Using Water Cycle Algorithm 12 (4) (2018)
Power Energy Syst. 144 (January) (2022) 2023, https://doi.org/10.1016/j.
3629–3636.
ijepes.2022.108555.
[2] T.P. Nguyen, T.A. Nguyen, T.V.H. Phan, D.N. Vo, A comprehensive analysis for
[26] S. Sharma, S. Bhattacharjee, A. Bhattacharya, Quasi-Oppositional Swine Influenza
multi-objective distributed generations and capacitor banks placement in radial
Model Based Optimization with Quarantine for optimal allocation of DG in radial
distribution networks using hybrid neural network algorithm, Knowledge-Based
distribution network, Int. J. Electr. Power Energy Syst. 74 (2016) 348–373,
Syst. 231 (2021) 107387, https://doi.org/10.1016/j.knosys.2021.107387.
https://doi.org/10.1016/j.ijepes.2015.07.034.
[3] M.I. Akbar, S.A.A. Kazmi, O. Alrumayh, Z.A. Khan, A. Altamimi, M.M. Malik,
[27] K.H. Truong, P. Nallagownden, I. Elamvazuthi, D.N. Vo, A Quasi-Oppositional-
A Novel Hybrid Optimization-Based Algorithm for the Single and Multi-Objective
Chaotic Symbiotic Organisms Search algorithm for optimal allocation of DG in
Achievement with Optimal DG Allocations in Distribution Networks, IEEE Access
radial distribution networks, Appl. Soft Comput. J. 88 (2020) 106067, https://doi.
10 (2022) 25669–25687, https://doi.org/10.1109/ACCESS.2022.3155484.
org/10.1016/j.asoc.2020.106067.
[4] D.Q. Hung, N. Mithulananthan, R.C. Bansal, Analytical expressions for DG
[28] N.K. Meena, S. Parashar, A. Swarnkar, N. Gupta, K.R. Niazi, Improved Elephant
allocation in primary distribution networks, IEEE Trans. Energy Convers. 25 (3)
Herding Optimization for Multi-objective DER Accommodation in 14 (3) (2018)
(2010) 814–820, https://doi.org/10.1109/TEC.2010.2044414.
1029–1039.
[5] K. Muthukumar, S. Jayalalitha, Optimal placement and sizing of distributed
[29] A. Selim, S. Kamel, A.S. Alghamdi, F. Jurado, Optimal Placement of DGs in
generators and shunt capacitors for power loss minimization in radial distribution
Distribution System Using an Improved Harris Hawks Optimizer Based on Single-
networks using hybrid heuristic search optimization technique, Int. J. Electr. Power
And Multi-Objective Approaches, IEEE Access 8 (2020) 52815–52829, https://doi.
Energy Syst. 78 (2016) 299–319, https://doi.org/10.1016/j.ijepes.2015.11.019.
org/10.1109/ACCESS.2020.2980245.
[6] M.H. Moradi, A. Zeinalzadeh, Y. Mohammadi, M. Abedini, An efficient hybrid
[30] M. Kashyap, S. Kansal, R. Verma, Sizing and Allocation of DGs in A Passive
method for solving the optimal sitting and sizing problem of DG and shunt
Distribution Network Under Various Loading Scenarios, Electr. Power Syst. Res.,
capacitor banks simultaneously based on imperialist competitive algorithm and
209(March 2021) (2022), 108046. 10.1016/j.epsr.2022.108046.
genetic algorithm, Int. J. Electr. Power Energy Syst. 54 (2014) 101–111, https://
[31] M.C.V. Suresh, J.B. Edward, A hybrid algorithm based optimal placement of DG
doi.org/10.1016/j.ijepes.2013.06.023.
units for loss reduction in the distribution system, Appl. Soft Comput. J. 91 (2020)
[7] S. Sannigrahi, P. Acharjee, Maximization of System Benefits with the Optimal
106191, https://doi.org/10.1016/j.asoc.2020.106191.
Placement of Maximization System Benefits with the Load Optimal Placement of
[32] A.M. Hemeida, O.M. Bakry, A.A.A. Mohamed, E.A. Mahmoud, Genetic Algorithms
DG and of DSTATCOM Considering Variations DG and DSTATCOM Considering
and Satin Bowerbird Optimization for optimal allocation of distributed generators
Load Variations, Procedia Comput. Sci. 143 (2018) 694–701, https://doi.org/
in radial system, Appl. Soft Comput. 111 (2021) 107727, https://doi.org/10.1016/
10.1016/j.procs.2018.10.446.
j.asoc.2021.107727.
[8] S. Das, D. Das, A. Patra, Operation of distribution network with optimal placement
[33] M. Kashyap, S. Kansal, B.P. Singh, Optimal installation of multiple type DGs
and sizing of dispatchable DGs and shunt capacitors, Renew. Sustain. Energy Rev.
considering constant, ZIP load and load growth, Int. J. Ambient Energy 41 (14)
113 (January) (2019) 109219, https://doi.org/10.1016/j.rser.2019.06.026.
(2020) 1561–1569, https://doi.org/10.1080/01430750.2018.1517688.
[9] K. Mahmoud, S. Member, N. Yorino, A. Ahmed, Optimal Distributed Generation
[34] J. O. Agushaka, A. E. Ezugwu, and L. Abualigah, Gazelle optimization algorithm: a
Allocation in Distribution Systems for Loss Minimization 31 (2) (2016) 960–969.
novel nature-inspired metaheuristic optimizer, vol. 35, no. 5. Springer London,
[10] N. Mithulananthan, “Optimal Allocation of Distributed Generation Using Hybrid
2023. 10.1007/s00521-022-07854-6.
Grey Wolf Optimizer,” pp. 14807–14818, 2017.
[35] B. Abdollahzadeh, F.S. Gharehchopogh, N. Khodadadi, S. Mirjalili, Mountain
[11] S.A. Chithradevi, L. Lakshminarasimman, R. Balamurugan, Engineering Science
Gazelle Optimizer: A new Nature-inspired Metaheuristic Algorithm for Global
and Technology, an International Journal Stud Krill herd Algorithm for multiple
Optimization Problems, Adv. Eng. Softw. 174 (August) (2022) 103282, https://doi.
DG placement and sizing in a radial distribution system, Eng. Sci. Technol. an Int.
org/10.1016/j.advengsoft.2022.103282.
J. 20 (2) (2017) 748–759, https://doi.org/10.1016/j.jestch.2016.11.009.
[36] K. Balu, V. Mukherjee, Siting and Sizing of Distributed Generation and Shunt
[12] T.P. Nguyen, D.N. Vo, A novel stochastic fractal search algorithm for optimal
Capacitor Banks in Radial Distribution System Using Constriction Factor Particle
allocation of distributed generators in radial distribution systems, Appl. Soft
Swarm Optimization, Electr. Power Components Syst. 48 (6–7) (2020) 697–710,
Comput. J. 70 (2018) 773–796, https://doi.org/10.1016/j.asoc.2018.06.020.
https://doi.org/10.1080/15325008.2020.1797935.
[13] S. Kumar, K.K. Mandal, N. Chakraborty, Optimal DG placement by multi-objective
[37] U. Sultana, A.B. Khairuddin, A.S. Mokhtar, N. Zareen, B. Sultana, Grey wolf
opposition based chaotic differential evolution for techno-economic analysis, Appl.
optimizer based placement and sizing of multiple distributed generation in the
Soft Comput. J. 78 (2019) 70–83, https://doi.org/10.1016/j.asoc.2019.02.013.
distribution system, Energy 111 (2016) 525–536, https://doi.org/10.1016/j.
[14] S. Nagaballi, V.S. Kale, Pareto optimality and game theory approach for optimal
energy.2016.05.128.
deployment of DG in radial distribution system to improve techno-economic
[38] E.S. Ali, S.M.A. Elazim, A.Y. Abdelaziz, Ant Lion Optimization Algorithm for
benefits, Appl. Soft Comput. J. 92 (2020) 106234, https://doi.org/10.1016/j.
optimal location and sizing of renewable distributed generations, Renew. Energy
asoc.2020.106234.
101 (2017) 1311–1324, https://doi.org/10.1016/j.renene.2016.09.023.
[15] A. Ali, M. U. Keerio, and J. A. Laghari, “Optimal Site and Size of Distributed
[39] M. Kumar, P. Nallagownden, I. Elamvazuthi, Optimal placement and sizing of
Generation Allocation in Radial Distribution Network Using Multi-objective
distributed generators for voltage- dependent load model in radial distribution
Optimization,” J. Mod. Power Syst. Clean Energy, vol. 9, no. 2, pp. 404–415, 2021,
system, Reinf. Plast. 19–20 (June) (2017) 23–37, https://doi.org/10.1016/j.
10.35833/MPCE.2019.000055.
ref.2017.05.003.
[16] H. P. C, K. Subbaramaiah, P. Sujatha, Optimal DG unit placement in distribution
[40] A. Mohamed Imran, M. Kowsalya, Optimal Distributed Generation and capacitor
networks by multi-objective whale optimization algorithm & its techno-economic
placement in power distribution networks for power loss minimization, 2014 Int.
analysis, Electr. Power Syst. Res., vol. 214, no. PA, p. 108869, 2023, 10.1016/j.
Conf. Adv. Electr. Eng. ICAEE (2014, 2014,), https://doi.org/10.1109/
epsr.2022.108869.
ICAEE.2014.6838519.
[17] A. Selim, M.H. Hassan, S. Kamel, A.G. Hussien, Allocation of distributed generator
[41] B. Singh, V. Mukherjee, P. Tiwari, Genetic algorithm for impact assessment of
in power networks through an enhanced jellyfish search algorithm, Energy Rep. 10
optimally placed distributed generations with different load models from minimum
(November) (2023) 4761–4780, https://doi.org/10.1016/j.egyr.2023.11.032.
total MVA intake viewpoint of main substation, Renew. Sustain. Energy Rev. 57
[18] Multi-objective Adaptive Fuzzy Campus Placement based Optimization Algorithm
(2016) 1611–1636, https://doi.org/10.1016/j.rser.2015.12.204.
for optimal integration of DERs and DSTATCOMs, J. Energy Storage, 10.1016/j.
[42] S. Kamel, M. Khasanov, F. Jurado, A. Kurbanov, H.M. Zawbaa, M.A. Alathbah,
est.2023.109682.
Simultaneously Distributed Generation Allocation and Network Reconfiguration in
[19] A.M. El-Rifaie, et al., Modified Gradient-Based Algorithm for Distributed
Distribution Network Considering Different Loading Levels, IEEE Access 11
Generation and Capacitors Integration in Radial Distribution Networks, IEEE
(August) (2023) 105916–105934, https://doi.org/10.1109/
Access 11 (November) (2023) 120899–120917, https://doi.org/10.1109/
ACCESS.2023.3319456.
ACCESS.2023.3326758.
[20] Z. Tan, M. Zeng, L. Sun, Optimal Placement and Sizing of Distributed Generators
Based on Swarm Moth Flame Optimization, Front. Energy Res., 9(April) (2021),
10.3389/fenrg.2021.676305.

19
M. Zubair Iftikhar et al. Renewable Energy Focus 49 (2024) 100550

[43] A. Bayat, A. Bagheri, Optimal active and reactive power allocation in distribution Optimizer with linear population size reduction technique, Knowledge-Based Syst.
networks using a novel heuristic approach, Appl. Energy 233–234(October 2018) 264 (2023) 110297, https://doi.org/10.1016/j.knosys.2023.110297.
(2019) 71–85. 10.1016/j.apenergy.2018.10.030. [45] “GitHub - MATPOWER/matpower: MATPOWER – steady state power flow
[44] R. Ahmed, G.P. Rangaiah, S. Mahadzir, S. Mirjalili, M.H. Hassan, S. Kamel, simulation and optimization for MATLAB and Octave.” https://github.com/
Memory, evolutionary operator, and local search based improved Grey Wolf MATPOWER/matpower (accessed Feb. 13, 2023).

20