See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/355073267

Optimal Capacity Estimation and Allocation of Distributed Generation Units

with Suitable Placement of Electric Vehicle Charging Stations

Conference Paper · August 2021

DOI: 10.1109/TENSYMP52854.2021.9550958

CITATIONS READS

10 51

4 authors:

Pavitra Sharma Akhilesh Kumar Mishra

Birla Institute of Technology and Science Pilani IILM University Greater Noida
17 PUBLICATIONS 97 CITATIONS 25 PUBLICATIONS 185 CITATIONS

SEE PROFILE SEE PROFILE

Puneet Mishra Hitesh Mathur

Birla Institute of Technology and Science Pilani Birla Institute of Technology and Science Pilani
60 PUBLICATIONS 845 CITATIONS 33 PUBLICATIONS 205 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Pavitra Sharma on 12 March 2024.

The user has requested enhancement of the downloaded file.

 Optimal Capacity Estimation and Allocation of
Distributed Generation Units with Suitable
Placement of Electric Vehicle Charging Stations
Pavitra Sharma1, Akhilesh Kumar Mishra2, Puneet Mishra3, Hitesh Dutt Mathur4
Department of Electrical & Electronics Engineering
Birla Institute of Technology & Science, Pilani
Pilani (Rajasthan), India
[email protected], [email protected], [email protected],
[email protected]
2021 IEEE Region 10 Symposium (TENSYMP) | 978-1-6654-0026-8/21/$31.00 ©2021 IEEE | DOI: 10.1109/TENSYMP52854.2021.9550958

Abstract—The optimal capacity estimation and allocations of [4][5][6][7][8]. In [4] the authors proposed an analytical
Distributed Generation (DG) units along with appropriate technique to calculate the optimal location and size of
placement of Electric Vehicle charging stations (EVCS) makes Distributed Energy Resources (DERs). In [5] authors utilize
a substantial contribution in curtailing power losses and an artificial bee colony algorithm to estimate the optimum
improving the voltage stability of a system. In this regard, this location, power factor, and size of a DG unit. The study in [6]
paper formulates a multi-objective function to minimize the was focused on optimal allocation and sitting of photovoltaic
power losses and voltage deviation of buses in the distribution PV cells in radial distribution systems based on the Ant Lion
network. The optimization problem is solved using three optimizer algorithm. The multi-objective function is
different types of optimization algorithms, namely Particle
formulated for the optimization problem but the authors did
Swarm Optimization (PSO), Grey Wolf Optimization (GWO),
and Hybrid Particle Swarm Grey Wolf Optimization
not present any information regarding the efficiency of the
(HPSOGWO). To simulate the practical situation, voltage- algorithm compared with other optimization algorithms. In
dependent load and various electric vehicle (EV) charging [7] optimal allocation of DG units is done using a hybrid
patterns based on the location of EVCS are considered in this particle swarm optimization algorithm (HPSO). To place DG
study. Solar photovoltaic, wind turbine, and diesel generator- at a suitable position, loss sensitivity analysis is performed
based DGs are taken into account in this study. The proposed and for optimum sizing HPSO algorithm is used and results
algorithm is tested on an IEEE 33 bus network considering obtained are compared with other optimization algorithms.
different scenarios. The results obtained show that HPSOGWO The work in [8] applied a hybrid method based on loss
provides the most optimal solution among all the considered sensitivity factors and Moth-Flame optimization to determine
algorithms, with the least power loss and voltage deviation for the optimal placement and size of DG units. The loss
all scenarios. sensitivity factors are used to estimate the candidate bus for
DG allocation and Moth-Flame optimization is used to
Keywords—Optimal allocation, Capacity estimation, determine optimal size and placement of solar and wind-
Electric Vehicle charging stations, Distributed Generation, based DG units. Further, the authors of [9] aim to cut-down
Hybrid Particle Swarm Grey Wolf Optimization the total investment value by optimal placement of wind-
I. INTRODUCTION based DG and battery storage system (BESS). The cost of
investment for BESS and DG is taken to be proportional to
The small-scale power generation systems, connected their size. It is reported in [10], Hybrid Grey Wolf
close to consumer loads are known as Distributed Generation Optimization (HGWO) is suitable to solve the discrete, non-
unit (DG). In recent years, the integration of DGs to the convex problems. Also, there is considerable reduction in loss
conventional grid has increased rapidly due to its advantages and improvement in voltage profile. The results are compared
in technical, environmental, and economic aspects. with various metaheuristic algorithms and shows that the
Photovoltaic, Wind turbines, biomass, microturbines, fuel HGWO algorithm outshines all the other algorithms. The
cells are the distributed energy resources that come under authors in [11] utilize grasshopper optimization (GOA)
Distributed Generation. It is mentioned in [1], that in the technique to determine the optimal place and size of DG
upcoming days DG would contribute around 20% of total units. The multi-objective function is formulated to minimize
power generation. Along with DG, electric vehicles (EVs) the active power losses and to improve voltage profile of the
are also getting popular as a potential alternative to fossil- system. The voltage dependent electric load demand, and
fuel-driven transportation. The penetration of EVs in the seasonal variations of wind and solar based DGs are
network increases the system’s total load which is a challenge considered but didn’t take into account impact of electric
for the current distribution system [2]. It is observed that vehicle charging load demand and optimal placement of
power loss and voltage stability of an electrical network is electric vehicle charging stations. A genetic algorithm-based
dependent on DG’s sizing, EVs charging load, location of optimization problem is formulated in [12] to determine the
DGs, and EVs charging stations (EVCS) in a network. optimal site and size of electric vehicle parking lots. The
Therefore, if the sizing of DGs and their integration with optimization problem considers the distribution system’s
EVCS are planned optimally and strategically in a network reliability and power losses along with investment cost.
then they are always committed to reduce the network’s However, this article didn’t consider the renewable DG units
power losses, enhancement of voltage stability margin, and various electric vehicle charging patterns based on
improvement of voltage profile, improvement of power location of parking lots in the system. Further, an approach
quality of supplied power and system reliability [3]. for simultaneous optimal allocation of renewable energy
In the past few years, many algorithms have been sources and EVCS in smart grids is proposed in [13]. A multi-
formulated for optimal placement and capacity of DG for objective optimization problem is formulated to reduce
minimizing overall power losses in the distribution network power losses, charging and demand supplying costs, and

Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE. Downloaded on October 14,2022 at 11:53:52 UTC from IEEE Xplore. Restrictions apply.
 voltage fluctuations. A hybrid genetic algorithm and particle speeds are 12.5 ⁄A0( , 3.5 ⁄A0( , 25 ⁄A0( respectively.
swarm optimization are used to solve this multi-objective The WTG (Bonus 150/30) having a specified rated power of
optimization problem but authors didn’t discuss the impact of 150 is considered in this study [16].
different electric vehicle charging patterns.
C. Diesel Generator (DEG) as DG
By the conducted literature survey, we remark that the
The diesel generator having a constant power factor of 0.9
researchers did not focus on optimal capacity estimation and
lagging is considered in this paper.
allocations of DG units together with appropriate placement
of EVCS simultaneously, and considering different charging D. Electric Load Modelling
patterns and voltage-dependent load. To investigate these In a practical distribution system, the load is continuously
issues, this paper formulates a multi-objective function with changing. Therefore, this study considers voltage-dependent
a purpose to determine the optimal capacity and location of load i.e., residential, commercial, and industrial load models
multiple DGs along with EVCS, to reduce power losses and to simulate the practical scenario. In load flow problems, the
voltage fluctuations in the system. The considered test system here considered exponential-based static load modeling is
includes voltage-dependent load models, renewable energy- more appropriate with respect to dynamic load modeling
based DG units, and different electric vehicle charging [17]. In the static load model, load behavior is represented as
patterns. an algebraic function of voltage magnitude [17]. In the case
The paper is organized as follows. Section II discusses the of exponential load, the mathematical relation between the
distributed generation units and load modelling. Electric load parameters and voltage magnitude is shown in (3) & (4)
[18].
EF
vehicle charging load modelling is detailed in Section III. The
= C0 D@ ) = 1,2, … . . IJ (3)
objective function and operational constraints are formulated
C,.1 )
EL
KC,.1 ) = KC0 D@ ) = 1,2, … . . IJ (4)
in Section IV. The results for the considered case study are
presented in Section V. Lastly, Section VI concludes the

where, N and NO are active and reactive power exponents

paper.
II. DISTRIBUTED GENERATION UNITS AND LOAD MODELLING
commercial, and industrial respectively. P> @ and KP> @
which vary according to different load classes i.e., residential,

refers to real and reactive power load at ) Q bus, while PR and

The DG units are usually modeled as a constant power

KPR are the values of active and reactive power load at initial
factor model. Power electronic and synchronous generator-

working conditions, respectively. D@ is the magnitude of the

based DGs are referred to as controllable DGs [14]. In this
study, all DGs are modelled as constant power factor model
voltage at the ) Q load bus. Table I shows the exponent values
having a power factor of 0.9 lagging. The details of modelling
of various DG units and electric load are discussed below.
corresponding to different load types. In this paper, a winter
A. Solar Photovoltaic (PV) as DG day is considered and load type on buses is mentioned in
The power output of a photovoltaic array is a function of Table II.
solar insolation and its relation is shown in (1) [15]. III. ELECTRIC VEHICLE CHARGING LOAD MODELLING

= ∗ ∗ (1)
The stochastic behavior of EVs charging load is a result
of several factors, such as the number of vehicles, battery
capacity, time at which it is plugged in or plugged out,

PV array surface in / ,
where stands for perpendicular irradiance falling on the charging speeds, charging patterns, and daily distance
represents the converter traveled by an EV [19]. Moreover, various types of vehicles
efficiency of PV arrays which is assumed to be 95%. like private, public, buses behave differently which increases
Moreover, is the rated value of the power of each PV uncertainty in EV charging load. In this paper, EVs charging
array and is the solar irradiance at standard test load is modelled using few parameters such as the number of
vehicles, daily distance traveled by an EV, time at which it is
/
condition (STC), which are considered to be 5 and 1000
plugged in i.e., arrival time or plugged out, i.e., departure
respectively.
time. The Monte Carlo simulation technique is used to

) is
B. Wind turbine Generator (WTG) as DG estimate parameters such as daily distance, arrival, and
departure time of EVs from their respective probability
The power output of a wind turbine generator (
density functions.
a function of wind speed and is expressed by (2) [15].

0; ≤ , & ≥
It is assumed that EVs considered in this study are private

⎧
& (), (,,
−
vehicles. It is also assumed; the charging of EVs will start as

⎪
3
∗2 4 ; < ≤ (2)
& ()
soon as they reach their residence/workplace until the battery

= - −
-./01 () & -
is fully charged. These EVs are charged according to the

⎨
()
− i.e., residential (STU), commercial (TTU), and
behaviors of three different charging station/parking nodes
⎪ +9
(,
: ∗ ( & − - ); - < & < (,
-./01
industrial (VTU) charging stations. At STU, the EV charging
⎩ − -
-./01
(,
where ; < is the active power output of whereas, for TTU it begins at 8:00 am when the owner arrives
starts at 4:00 pm after the owner arrives at the residence,
; < ; <
. Also,
=>
kinds of working shifts therefore for VTU, at morning 6:00 am
and represents the active power output of at the workplace. Usually, industrial employees have three

?, , =@ and => are measured, rated, cut-in, and cut-out

at cut-out and rated wind speed respectively. Furthermore,

wind speeds in ⁄A0( respectively. The specifications of the

first shift’s EV charging starts, similarly, the second shift and
the third shift start at 2:00 pm and 10:00 pm respectively. The
considered , including rated, cut-in, and cut-out wind

Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE. Downloaded on October 14,2022 at 11:53:52 UTC from IEEE Xplore. Restrictions apply.
 duration of charging ( ,E ) for each N Q
EV can be obtained
from (5)[20].
WE YZ
= ∀N ∈ I (5)
XRR
100
,E YZ YZ
,E

YZ
where XRR
ℎ/100 ; YZ represents the charging rate
refers to the energy consumption per hundred

; YZ is the charging efficiency of EVs. I

kilometers in
of EVs in

( E ,E ) of EVs can be further calculated using (6) [20].

represents the total no. of EVs. The departure/end time

E ,E = _ ,E + =,E − 1 ∀N ∈ I (6) Fig. 1 Charging load profile of 65 number of EVs at residential (STU),
where _ ,E represents the time at which EV charging commercial (TTU) and industrial (VTU) charging station

YZ_
(/) can be
begins. However, by accumulating the charging power of
each period, the total charging load P> Hence, the Backward-Forward sweep power flow
calculated. It is noted that EV charging periods are algorithm is mainly used for distribution networks for fast
independent of each other, therefore, using (7) the daily load and accurate results [21]. Consider a radial distribution

a line having a total impedance o _ .

network shown in Fig. 2. Buses r & s are connected through
bc
profile of a large number of EV charging can be calculated.

YZ_
P> (/) = a YZ
,E (/) (7)
EdX
where YZ ,E is the charging power of N
Q
vehicle.

at residential (STU), commercial (TTU), and industrial (VTU)

Figure 1 shows the daily charging load profile of 65 EVs

charging stations. Table III shows the parameters of electric

vehicles considered in this work. The peak load of each EV Fig. 2 Radial distribution network

o = S _ + pq (8)
charging load pattern is considered for determining optimal
locations of four EVCS. _ _

TABLE I. THE EXPONENT VALUES CORRESPONDING TO DIFFERENT Active power loss for the line between r & s can be written
LOAD TYPES[18] as:
+ K_
s C( = S _9 : (9)
_
fgh fih fgj fij fgk fik
Residential Commercial Industrial
_)
D
Load Type

Day 0.72 2.96 1.25 3.5 0.18 0.6

Summer The total active power loss (TAPL) can be computed by:
Night 0.92 4.04 0.99 3.95 0.18 0.6
Day 1.04 4.19 1.5 3.15 0.18 0.6 buv

s C = a s C( (10)
Winter
Night 1.3 4.38 1.51 3.4 0.18 0.6
Constant Power 0 0 0 0 0 0 _)
_dX
where Iw is the total number of branches present in the
network, - = 1: Iw and Iw is the number of buses.
TABLE II. DIFFERENT LOAD TYPES ON VARIOUS BUSES

Load Type Bus No.

Constant Power 2,3,4,5,6,7
An important function of the voltage deviation (VD) is to
determine the weak buses in a network. It is a measure of
Residential Load 8,9,10,11,12,13,14,15,16,17,18,19,20,21,22
voltage stability margin in the power system network to
Industrial Load 23,24,25
maintain voltage within the permissible limits after the
Commercial Load 26,27,28,29,30,31,32,33 occurrence of disturbance. Total voltage deviation (TVD) is
bu
given by:

DW = a|1 − D | (11)
TABLE III. PARAMETERS OF EV
PHEV Parameters Value
YZ dX
where, D is the voltage at bus r, - = 1: Iw .
XRR
TEYZ (kWh)
(kWh/100km) 12
24
UlTEm n /UlTEm@E (%) 100/20
In this paper, two objective functions (OF) are formulated

,E / ,E
index. i.e., Minimize lz1 = s z and Minimize lz2 =
YZ YZ,m n
to minimize active power loss & minimize voltage deviation

TVD.
(kW) 3
YZ 85%
This multi-objective problem is solved using the weighted
IV. OBJECTIVE FUNCTION FORMULATION AND OPERATIONAL sum approach and the overall objective function (OOF) is
CONSTRAINTS
llz = &X ∗ s C + & ∗ DW (12)
defined in (12).

where &X and & are the weights of lz1 & lz2
In order to formulate the objective function, power flow

respectively. The values considered for &X and & are 0.7 and
calculation is an initial step. Conventional power flow
algorithms such as Gauss-Seidel, Newton Raphson are
incompetent for distribution network as it has low X/R ratio.

Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE. Downloaded on October 14,2022 at 11:53:52 UTC from IEEE Xplore. Restrictions apply.
 0.3 respectively. The objective function in (12) is minimized Start

with respect to the following constraints.

Input Load
and Line
Power Balance Constraints
YZ_
+ + = + (13)
Data

P> P> P>__ <_ @ Initialize optimization parameters

@ , P>__ and
such as population size, iteration

< are the active power of the

count (i=1), lower and upper limits
where of capacities of DG and locations of
DGs and EVCS
electrical grid, active losses of the network, and power output
of all DGs respectively.
KP> + KP>__ = K +K (14)
Update the system data using values

<_ @
obtained from optimization
algorithm

where K @ , KP>__ and K <_ are the reactive power of the Execute power flow and evaluate
initial fitness using multi-objective
function
electrical grid, reactive losses of the network, and power
output of all DGs respectively.
Add penalty factor
Constraints
Voltage limits of the bus No to objective

D@m@E < D@ (/) < D@m ) = 1,2, … . . Iw (15)

Satisfied

n
function

where, D@m@E and D@m n are the minimum and maximum

Yes

Update DG parameters and EVCS

location based on updated

D@ is the voltage magnitude at bus ) in pu.

optimization parameters
bus voltage limits, having values 0.95 and 1.05 respectively.
Execute power flow and update
new fitness using multi-objective
Power limits of DERs function

m@E
<,m < <,m < m n
<,m ∀ ∈• (16)
Km@E
<K <,m < K <,m
∀ ∈•
m n
(17)
Check for maximum no. of
No i= i+1

<,m
iterations

m@E m n
where, Y€,m
and Y€,m are
the minimum and Yes

Q
maximum limits on the active & reactive power output of the
Print optimal results
DER, respectively.
Fig. 3 Flow chart of the proposed method
A. Algorithm for capacity estimation of DG, allocation of
DG and EVCS Start Replace with the current
particle

In this paper, 3 optimization techniques are adopted to Create initial population

randomly Yes
estimate the optimal capacity and placement of DG along Set values:α,A,C
Run PSO
with suitable placement of EVCS for minimizing both Evaluate the fitness of all
Update values : α,A,C
wolves
objective functions. The flow chart of the proposed method is Evaluate the fitness
of all particles
No

shown in Fig. 3. In order to set the lower and upper limits of Memorize the best values of For current Calculate the values : Update the position of grey
particle Xα,Xβ,Xδ
control parameters for optimization certain pre-assumptions particles and swarm
If rand(0,1)<prob
wolves

are made:
Check for Check for small number of
Update the velocity No
maximum iterations
• For the scenarios in which DGs are integrated into the no. of
No and position of
particles
iterations Export mean of the positions
test system, the total load of the system is fulfilled Yes
of three best wolves

only by DGs which means that the power taken from Yes
Stop
the grid is zero. Further, the sum of capacities of all
DGs present in the network is less than or equal to the
total connected load on the system to restrict the over- Fig. 4 Flowchart of HPSOGWO algorithm
sizing of DGs.
V. CASE STUDY AND RESULTS
• DGs are permitted to be located at any bus except the
grid-connected bus i.e., bus number 1. The IEEE 33-bus radial distribution system is
considered. During the analysis, the base voltage and base
• The total four EV charging stations are considered in MVA considered are 12.66 kV and 100 MVA respectively.

stations (STU 1 & STU 2) are assigned to locate for

this system for optimal location. Two charging The test system consists of 33 buses and 32 branches as
shown in the single-line diagram depicted in Fig. 5. It is
residential buses. However, for a commercial and assumed that the initial total real and reactive power loads on

(TTU & VTU) is allocated to each group.

industrial group of buses, one charging station the system (apart from EV charging load) are 3715 kW and
2300 kVAr, respectively. In this study, five scenarios are
considered i.e., a test system without DG and EVCS; without
1) Hybrid Particle Swarm Grey Wolf Optimzation DG but with EVCS; with a PV as DG and EVCS; with a PV
(HPSOGWO) : This algorithm is developed by Şenel, F.A., and WTG as DG and EVCS and a test system with all the
Gökçe, F., Yüksel, A.S. et al. [22]. It utilizes the GWO three DGs, PV, WTG, DEG, and EVCS. In scenario 2,
algorithm to support the PSO algorithm in order to reduce the optimum allocations of EVCS are determined using all three
possibility of falling PSO into a local minimum [23]. The optimization algorithms. Moreover, for scenarios 3,4, and 5
flowchart of the HPSOGWO algorithm is shown in Fig. 4. the optimum capacity and allocations of DGs and allocations
The optimal results obtained from HPSOGWO are discussed of EVCS are obtained using all three optimization algorithms.
in Section V. These results are summarized in Table IV.

Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE. Downloaded on October 14,2022 at 11:53:52 UTC from IEEE Xplore. Restrictions apply.
 TABLE IV. RESULTS OBTAINED FROM VARIOUS OPTIMIZATION TECHNIQUES IN ALL THE CONSIDERED SCENARIOS
Different Optimal Bus Min value of
Optimal EVCS Bus No. Maximum Voltage
Different Cases Optimization No. locations of Capacity of DG TAPL
locations Deviation
Techniques DG (kW)
Base Scenario 1 –
Without DG and N.A. N.A. N.A. N.A. 165.7162 0.0839 at 18th bus

STUXat 8, STU at 20
EVCS

TTU at 26, VTU at 24

PSO N.A. N.A. 202.22 0.0913 at 18th bus
Scenario 2 –
STUXat 8, STU at 22
TTU at 26, VTU at 23
Without DG but with GWO N.A. N.A. 203.75 0.0913 at 18th bus
STUXat 8, STU at 19
EVCS

TTU at 26, VTU at 23

HPSOGWO N.A. N.A. 201.07 0.0913 at 18th bus
STUXat 8, STU at 21
TTU at 26, VTU at 24
PSO 26 (PV) 4.52 MW 123.11 0.0301 at 26th bus
STUXat 15, STU at 22
TTU at 26, VTU at 25
Scenario 3 –
With PV as DG and GWO 26 (PV) 4.6 MW 135.40 0.0313 at 26th bus
STUXat 8, STU at 19
EVCS

TTU at 26, VTU at 23

HPSOGWO 26 (PV) 4.52 MW 119.7 0.03 at 26th bus
STUXat 8, STU at 20
TTU at 29, VTU at 24
24 (PV) 1.6 MW (PV)
PSO 66.52 0.0377 at 18th bus
STUXat 18, STU at 22
26 (WTG) 2.82 MW (WTG)
Scenario 4 –
TTU at 33, VTU at 25
24 (PV) 2 MW (PV)
With PV and WTG GWO 90.75 0.0423 at 18th bus
STUXat 8, STU at 19
as DG and EVCS 26 (WTG) 3 MW (WTG)

TTU at 26, VTU at 24

24 (PV) 1.6 MW (PV)
HPSOGWO 61.69 0.0377 at 18th bus
26 (WTG) 2.81 MW (WTG)
STUXat 13, STU at 22
13 (PV) 1.05 MW (PV)
TTU at 30, VTU at 23
PSO 30 (WTG) 1.57 MW (WTG) 32.45 0.0125 at 13th bus
24 (DEG) 1.8 MW(DEG)

STUXat 18, STU at 22

Scenario 5 – 33 (PV) 1.5 MW (PV)
TTU at 33, VTU at 25
With PV, WTG, and GWO 13 (WTG) 1.2 MW (WTG) 47.93 0.0238 at 33rd bus
DEG as DG and
EVCS 24 (DEG) 1.8 MW(DEG)

STUXat 11, STU at 19

30 (PV) 1.43 MW (PV)
TTU at 30, VTU at 24
HPSOGWO 11(WTG) 1.2 MW (WTG) 27.98 0.0111 at 11th bus
24 (DEG) 1.8 MW(DEG)

All three optimization algorithms are iterated for 200 optimal solutions obtained by HPSOGWO. It means, 320 PV
iterations with 70 population sizes for every scenario. arrays and 19 WTGs are required in this scenario. The TAPL
Further, Fig.6, 7 and 8 shows the convergence characteristics and maximum VD values observed in this case are 61.69 kW
for scenario 3, 4 and 5 respectively considering all the three and 0.0377 on the 18th bus respectively. However, in scenario
optimization algorithms. In base scenario 1, the value of 5 the optimal locations and rated capacities of PV, WTG, and
TAPL is lower than scenario 2 as in this case EV charging DEG are 30th bus with 1.43MW, 11th bus with 1.2MW, 24th
load is also considered. As observed in Table IV, for bus with 1.8MW respectively. This implies that the test
scenarios in which DG gets integrated into the test system, system requires approximately 286 PV arrays, 8 WTGs and a
HPSOGWO is providing an optimal solution with the lowest diesel generator of 1800kW to fulfill the total load demand.
TAPL and voltage deviation as compared to PSO and GWO. The TAPL and maximum VD value in this scenario come out
In scenario 3, the optimal location and capacity of DG are 26th to be lowest among all scenarios i.e., 27.98 kW and 0.0111
bus and 4.52 MW respectively. It indicates that in this case on the 11th bus. It is evident that as the number of DGs is
approximately 904 PV arrays are required to fulfill the total increasing in the test system the value of TAPL is decreasing.
load demand. The value of TAPL and maximum VD Fig.9 shows the voltage profile in all the scenarios. It is
observed at these optimal solutions is 119.7 kW and 0.03 on significant from Fig.9 that the most stable voltage profile is
the 26th bus respectively. observed for Scenerio 5 with minmum voltage deviation at
each bus.

Fig. 5. IEEE 33 bus modified test system

Similarly, in scenario 4, 1.6MW rated PV allocated on the

Fig. 6 Convergence characteristics for Scenario 3 considering HPSOGWO,
24rd bus and 2.81 MW rated WTG on the 26th bus are the
GWO, and PSO.

Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE. Downloaded on October 14,2022 at 11:53:52 UTC from IEEE Xplore. Restrictions apply.
 considered scenarios. It is also noticeable from the results that
as the number of DGs is increasing in the test system the total
power loss is decreasing and voltage profile is improved.
ACKNOWLEDGMENT
This work is supported by the Department of Science and
Technology, Govt. of India, New Delhi under the “Internet of
things (IoT) Research of Interdisciplinary Cyber-Physical
Systems Programme” with a sanction letter
DST/ICPS/CLUSTER/IOT/2018/ GENERAL.
REFERENCES
[1] P. Prakash and D. K. Khatod, “Optimal sizing and siting techniques for
Fig. 7 Convergence characteristics for Scenario 4 considering HPSOGWO, distributed generation in distribution systems: A review,” Renewable
GWO, and PSO. and Sustainable Energy Reviews, vol. 57. Elsevier Ltd, pp. 111–130,
May 01, 2016, doi: 10.1016/j.rser.2015.12.099.
[2] O. Hafez and K. Bhattacharya, “Optimal design of electric vehicle
charging stations considering various energy resources,” Renew.
Energy, vol. 107, pp. 576–589, 2017, doi:
https://doi.org/10.1016/j.renene.2017.01.066.
[3] I. Diahovchenko, M. Kolcun, Z. Čonka, V. Savkiv, and R.
Mykhailyshyn, “Progress and Challenges in Smart Grids: Distributed
Generation, Smart Metering, Energy Storage and Smart Loads,” Iran.
J. Sci. Technol. - Trans. Electr. Eng., vol. 44, no. 4, pp. 1319–1333,
2020, doi: 10.1007/s40998-020-00322-8.
[4] N. Acharya, P. Mahat, and N. Mithulananthan, “An analytical approach
for DG allocation in primary distribution network,” Int. J. Electr.
Power Energy Syst., vol. 28, no. 10, pp. 669–678, 2006, doi:
10.1016/j.ijepes.2006.02.013.
[5] R. Palanisamy and S. K. Muthusamy, “Optimal Siting and Sizing of
Multiple Distributed Generation Units in Radial Distribution System
Using Ant Lion Optimization Algorithm,” J. Electr. Eng. Technol., vol.
Fig. 8 Convergence characteristics for Scenario 5 considering HPSOGWO, 16, no. 1, pp. 79–89, 2021, doi: 10.1007/s42835-020-00569-5.
GWO, and PSO. [6] A. Ali, A. Youssef, T. George, and S. Kamel, “Optimal DG allocation
in distribution systems using Ant lion optimizer,” 2018 Int. Conf.
Innov. Trends Comput. Eng., pp. 324–331, 2018.
[7] M. A. Tolba, V. N. Tulsky, and A. A. Zaki Diab, “Optimal allocation
and sizing of multiple distributed generators in distribution networks
using a novel hybrid particle swarm optimization algorithm,” in 2017
IEEE Conference of Russian Young Researchers in Electrical and
Electronic Engineering (EIConRus), 2017, pp. 1606–1612, doi:
10.1109/EIConRus.2017.7910880.
[8] H. Abdel-mawgoud, S. Kamel, M. Ebeed, and M. M. Aly, “An efficient
hybrid approach for optimal allocation of DG in radial distribution
networks,” in 2018 International Conference on Innovative Trends in
Computer Engineering (ITCE), 2018, pp. 311–316, doi:
10.1109/ITCE.2018.8316643.
[9] J. Qiu, Z. Xu, Y. Zheng, D. Wang, and Z. Y. Dong, “Distributed
generation and energy storage system planning for a distribution
system operator,” IET Renew. Power Gener., vol. 12, no. 12, pp. 1345–
Fig. 9 Voltage profile of buses for all scenarios
1353, 2018.
[10] R. Sanjay, T. Jayabarathi, T. Raghunathan, V. Ramesh, and N.
VI. CONCLUSION Mithulananthan, “Optimal allocation of distributed generation using
This paper proposed a method to estimate the optimal hybrid grey wolf optimizer,” Ieee Access, vol. 5, pp. 14807–14818,
2017.
capacity and placement of multiple DGs along with EVCS
[11] K. S. Rani, B. K. Saw, P. Achargee, and A. K. Bohre, “Optimal Sizing
optimal allocation in a practical distribution system by using and Placement of Renewable DGs using GOA Considering Seasonal
various meta-heuristics optimization algorithms. The multi- Variation of Load and DGs,” in 2020 International Conference on
objective function is employed to lower the power losses and Computational Intelligence for Smart Power System and Sustainable
improve voltage stability. The proposed method is tested on Energy (CISPSSE), 2020, pp. 1–6.
a modified IEEE 33 bus network having DGs based on solar [12] M. Moradijoz, M. Parsa Moghaddam, M. R. Haghifam, and E.
photovoltaic (PV), wind turbine (WTG), and diesel generator Alishahi, “A multi-objective optimization problem for allocating
parking lots in a distribution network,” Int. J. Electr. Power Energy
(DEG) with voltage-dependent load models such as Syst., vol. 46, pp. 115–122, 2013, doi:
residential, commercial, and industrial load types. In this https://doi.org/10.1016/j.ijepes.2012.10.041.
study, various electric vehicle charging load patterns are [13] M. R. Mozafar, M. H. Moradi, and M. H. Amini, “A simultaneous
formulated using Monte Carlo simulations. Further, the approach for optimal allocation of renewable energy sources and
optimal allocation of three types of Electric vehicle charging electric vehicle charging stations in smart grids based on improved GA-
stations (EVCS), namely, Residential Charging Station PSO algorithm,” Sustain. Cities Soc., vol. 32, pp. 627–637, 2017, doi:
https://doi.org/10.1016/j.scs.2017.05.007.
(RCS), Industrial Charging Station (ICS), and Commercial
[14] S. M. Moghaddas-Tafreshi and E. Mashhour, “Distributed generation
Charging Station (CCS) is determined. The results show that modeling for power flow studies and a three-phase unbalanced power
HPSOGWO outperforms all the other algorithms for all the flow solution for radial distribution systems considering distributed

Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE. Downloaded on October 14,2022 at 11:53:52 UTC from IEEE Xplore. Restrictions apply.
 generation,” Electr. Power Syst. Res., vol. 79, no. 4, pp. 680–686, 2009, 2019, no. 18, pp. 4819–4823, 2019, doi: 10.1049/joe.2018.9335.
doi: 10.1016/j.epsr.2008.10.003. [20] H. Liu, Y. Ji, H. Zhuang, and H. Wu, “Multi-objective dynamic
[15] H. R. Baghaee, M. Mirsalim, G. B. Gharehpetian, and H. A. Talebi, economic dispatch of microgrid systems including vehicle-to-grid,”
“Reliability/cost-based multi-objective Pareto optimal design of stand- Energies, vol. 8, no. 5, pp. 4476–4495, 2015, doi: 10.3390/en8054476.
alone wind/PV/FC generation microgrid system,” Energy, vol. 115, no. [21] U. Eminoglu and M. H. Hocaoglu, “Distribution systems
October 2017, pp. 1022–1041, 2016, doi: forward/backward sweep-based power flow algorithms: A review and
10.1016/j.energy.2016.09.007. comparison study,” Electr. Power Components Syst., vol. 37, no. 1, pp.
[16] “124-an-bonus-150-30 @ en.wind-turbine-models.com.” [Online]. 91–110, 2009, doi: 10.1080/15325000802322046.
Available: https://en.wind-turbine-models.com/turbines/124-an- [22] F. A. Şenel, F. Gökçe, A. S. Yüksel, and T. Yiğit, “A novel hybrid
bonus-150-30. PSO–GWO algorithm for optimization problems,” Eng. Comput., vol.
[17] K. Qian, C. Zhou, M. Allan, and Y. Yuan, “Effect of load models on 35, no. 4, pp. 1359–1373, 2019, doi: 10.1007/s00366-018-0668-5.
assessment of energy losses in distributed generation planning,” Int. J. [23] A. K. Mishra, P. Mishra, and H. D. Mathur, “Load Frequency Control
Electr. Power Energy Syst., vol. 33, no. 6, pp. 1243–1250, 2011, doi: of a Nonlinear Power System via Demand Response Control Strategy
10.1016/j.ijepes.2011.04.003. Based Fractional Order Fuzzy Controller,” in 2020 21st National
[18] N. Rostami and M. O. Sadegh, “The Effect of Load Modeling on Load Power Systems Conference (NPSC), 2020, pp. 1–6, doi:
Flow Results in Distribution Systems,” vol. 6, no. 1, pp. 16–27, 2018, 10.1109/NPSC49263.2020.9331919.
doi: 10.12691/ajeee-6-1-3.
[19] T. Rawat and K. R. Niazi, “Impact of EV charging/discharging
strategies on the optimal operation of islanded microgrid,” J. Eng., vol.

Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE. Downloaded on October 14,2022 at 11:53:52 UTC from IEEE Xplore. Restrictions apply.
View publication stats