Received 12 July 2024, accepted 15 August 2024, date of publication 19 August 2024, date of current version 30 August 2024.

Digital Object Identifier 10.1109/ACCESS.2024.3446032

Metaheuristic Algorithm Based Energy

Management System for Electric
Vehicle Charging Station
JENNIE ANGELA JOSE SHIRLEY , (Student Member, IEEE), R. P. POOJA , AND
MADDIKARA JAYA BHARATA REDDY , (Senior Member, IEEE)
National Institute of Technology Tiruchirappalli, Tiruchirappalli 620015, India
Corresponding author: Maddikara Jaya Bharata Reddy (jayabharat_res@yahoo.co.in)

ABSTRACT Electric Vehicles (EVs) are at the forefront of the transition to sustainability. Their almost zero
tailpipe emissions and contribution to noise pollution make them suitable for minimizing the current levels of
environmental pollution. One aspect that hinders the widespread adoption of EVs is the limited availability
of Electric Vehicle Charging Stations (EVCS) in remote areas, making the charging costs extremely high.
The goal of reducing EV charging cost can be achieved by integrating renewables into a conventional
power system. This study proposes a method for optimizing the charging cost by utilizing renewable energy
resources for an EVCS in a modified IEEE 33 bus system. This study involves the incorporation of three
renewable energy sources: solar, wind and biogas. Energy management between the various sources and
the EVCS was achieved through a metaheuristic algorithm-based Energy Management System (MAEMS).
The proposed MAEMS was used to develop a dynamic pricing scheme. A techno-economic analysis of the
system was conducted. The analysis was performed on the MATLAB-Simulink platform, considering the
standards of the Indian EVCS system.

INDEX TERMS Electric vehicle charging stations, energy management system, metaheuristic algorithms,
renewable energy sources.

I. INTRODUCTION distributed Electric Vehicle Charging Stations (EVCS), with

The increasing dependence on fossil fuels to meet the proper planning of the location and capacity [4].
ever-growing demand for electricity has led to long-lasting Furthermore, the charging duration and cost must be
deteriorating effects on the environment [1]. This has subse- minimized for EVs to be a better selling commodity [5]. This
quently resulted in poor quality of life and chronic illnesses in necessitates an intelligent energy management system (EMS)
children and adults alike [2]. Under such circumstances, the that would help in the optimal allocation of resources to the
introduction of renewable, eco-friendly, and non-exhaustible components of the power system and in presenting a more
energy sources has proven to be a great choice. economical pricing scheme for EV users.
With regard to growing concerns about fossil fuels, the EMS for EVCS has been extensively studied world-
transition to Electric Vehicles (EVs) over their fossil fueled wide [6]. Resource allocation and cost optimization are
counterparts is preferred [3]. Vehicular electrification enables problems that have gained recognition. These problems need
an effective transition to a sustainable and zero-carbon emis- to be addressed to enable the safe, reliable, and economical
sion environment. Similar to the case of fossil fueled Internal operation of power systems with EVCS and renewables.
Combustion Engine (ICE) vehicles, EVs require evenly An EMS for a residential EVCS with a photovoltaic (PV)
power source and energy storage system (ESS) was proposed
The associate editor coordinating the review of this manuscript and in [7]. Real-time coordination between the PV power station,
approving it for publication was Chandan Kumar . grid and the ESS was implemented, thereby enabling the
2024 The Authors. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.
116354 For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/ VOLUME 12, 2024
J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

EVCS to function as a stand-alone system. This study did not [1] A modified IEEE 33 bus system with a wind farm,
consider the different types of charging available to the EVs. a solar PV system and biogas plant with EVCS was
The optimal configuration and techno-economic assess- developed.
ment for a solar-powered EVCS were analyzed for four [2] Alternating Current (AC) and Direct Current (DC)
Indian cities with varied solar conditions in [8]. The charging were considered in the EVCS, which was
environmental benefits of the proposed system are also developed according to Indian standards.
evaluated. [3] MAEMS is implemented using three metaheuristic
An EMS for public EVCS integrated with a community algorithms: Artificial Bee Colony (ABC), Social Group
microgrid was proposed in [9], based on the switching Optimization (SGO), and Particle Swarm Optimization
mechanism from one trading market to another. The EMS (PSO). The results were compared with those obtained
minimizes the charging costs and satisfies the community using the non-linear solver.
load. However, the environmental footprint of the proposed [4] The dynamic pricing scheme obtained was compared
model has not been assessed. with the conventional time-of-day grid tariff.
A dynamic EMS for an integrated solar and energy storage [5] Techno-economic assessment of the developed model
charging station, based on a real-world situation in Taiwan is made.
was developed in [10]. The proposed EMS optimized the Figure 1 presents the components of this study. The following
economic efficiency of the EVCS. However, this study section provides an overview of the EMS and metaheuristics.
considers only the EVCS for fast charging.
A biogas–solar powered EVCS was proposed for a
typical location in Karnataka in [11]. Techno-economic
and environmental assessments were also conducted. It was
observed that dependency on the grid reduced with improved
environmental benefits in terms of carbon dioxide emissions.
A grid connected PV battery system was proposed and
analyzed using HOMER to check both the economic and
technical feasibility of the model in [12]. A system with
reduced charging costs was developed. Different patterns and
trends in EV charging were not considered in this analysis.
The authors of [13] proposed a fuzzy-logic based approach
to develop a method for charging priority identification,
considering parameters such as the battery state of charge
(SOC), charging cost, peak power demand, and waiting time.
This model aims to benefit both the utility and EV users.
A hybrid biogas and solar grid-connected EVCS system
with residential loads in Bangladesh was analyzed using
MATLAB-Simulink in [14] and optimized using a fuzzy
logic-based algorithm. A high level of expertise is required to
develop a fuzzy-based model for complex power distribution
systems.
A hybrid generation system comprising of two or more FIGURE 1. Components of the Study.

unreliable and intermittent energy sources can provide better

system reliability. Wind and solar power have complementary
energy generation profiles; thus, the installation of a hybrid II. THEORETICAL BACKGROUND
solar-wind energy system would ensure a high efficiency The availability of limited energy resources necessitates the
and stable power supply [15], [16]. A techno-economic proper management and utilization of resources. Ensuring
assessment of a hybrid solar-wind system, single solar PV a balance between power generation and consumption, that
system, and single wind system was developed in [17]. It was is, energy management, would help maintain a reliable
concluded that the hybrid system is more economical and has power system [18]. An Energy Management System (EMS)
a better system performance than either of the single systems. enables the reliable, efficient, and economic operation of a
The existing literature does not mention wind, solar power system by considering the generated power, status
and biogas system-based EVCS. Furthermore, the use of of various components, forecasted load, weather, and cost
metaheuristic methods for optimization would make the of energy generation. An EMS can thus be enlisted to
system more robust and reliable. This study proposes an arrive at solutions for single or multi-objective functions
metaheuristic algorithm-based Energy Management System such as monitoring real and reactive power, minimizing
(MAEMS) for power systems with an EVCS. The major operational costs, minimizing losses, and energy balancing
contributions of this study are as follows: in transmission systems [19].

VOLUME 12, 2024 116355

 J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

The Supervisory Control and Data Acquisition (SCADA) The metaheuristic model is a well-known and higher-level
system collects real time data from equipment in remote method used to find the best solution to complex optimization
locations, and controls them based on specified constraints. problems. It falls under the category of stochastic optimiza-
Intelligent Electronic Devices (IEDs) are present in the field. tion. Often, this method is preferred over other conventional
They can be sensors or actuators. The sensors obtain data iterative techniques, optimization algorithms, and simple
from the equipment, whereas the actuators implement the heuristics because of its practical efficacy and reduced
control mechanism decided by the central controller. Remote computational efforts. Different algorithms in metaheuristics
Terminal Units (RTUs) collect data from sensors and send are designed based on ideas that exist in nature. Some of the
them to the central controller, that is, the SCADA master unit. commonly used models in population–based metaheuristics
Programmable Logic Controllers (PLCs) collect decisions are the Genetic Algorithm (GA), Artificial Bee Colony
from the central controller and send them to the actuators to (ABC), Particle Swarm Optimization (PSO), Gravitational
control the respective equipment [20], [21]. RTUs measure Search Algorithm (GSA), and Spotted Hyena Optimizer
data once every 4-10s [22]. (SHO). These techniques have found applications in a wide
range of industries, from artificial intelligence and the
Internet of Things to engineering design because of their
proven efficiency and robustness [25].
In this study, a MAEMS was designed to minimize
the charging cost for consumers and maximize renewable
energy utilization. MAEMS optimizes the problem based on
the real-time situation of the various parameters involved.
The following section elucidates the proposed energy-
management algorithm for optimization.

III. MAEMS ALGORITHM

A MAEMS was developed for a modified IEEE 33 bus
system with EVCS. The MAEMS achieves its objectives by
considering the generated power, power demand, charging
time of the EV, and the base tariff. All the resources
considered are assumed to be owned and operated by the
utility. The utility takes care of the relationship between the
consumer and the supply side.
The following constraints are observed during the design
FIGURE 2. Supervisory control and data acquisition system. of the system:
• The voltage (V) range lies within the specified lower Vll
Figure 2 shows a typical SCADA system. Different sensors (0.95 pu) and upper Vul (1.05 pu) limits. The rated line-
collect real-time information, that is sent to the RTUs to-line voltage of the system was 415 V.
connected to them. Through communication networks, the
information reaches the SCADA master unit, where decisions Vll < V < Vul (1)
are made to control and monitor the system. This helps to
prevent and mitigate faults in the power system, thereby • The frequency (f) was always maintained within the
ensuring its safe operation. The developed MAEMS works fixed lower fll (49.5 Hz) and upper ful (50.5 Hz) limits.
using an SCADA system, with voltage and current sensors The rated frequency of the system is 50 Hz.
available at each node.
For EMS, in the SCADA control center, we require fll < f < ful (2)
a dedicated algorithm that can manage the system with
• The line limits are always followed, that is, the current
the given constraint for a reduced cost. Mathematical
flowing through the transmission lines (Iline ) should be
programming is a classical optimization technique in which
less than or equal to the rated current (Irated ).
one or more objective functions are optimized based on a set
of constraints. Mathematical programming methods include Iline ≤ Irated (3)
non-linear, linear, second-order conic and mixed-integer
programming. Although they can reach a global optimum, MAEMS aims to minimize the charging cost (Cc ) and
they lack efficiency owing to the high computational effort maximize renewable energy utilization (URES ). The EV user
required when compared to metaheuristics. Furthermore, benefits from the developed dynamic pricing scheme. The
the optimization process is time consuming and cannot objective functions can be defined as
be adopted for complex systems with a large number of
variables [23], [24]. min(Cc ) and max(URES ) (4)

116356 VOLUME 12, 2024

J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

First, the power demand from the EVCS (Pd,EVCS ) is B. MINIMIZATION OF CHARGING COST
calculated as The charging costs in the developed dynamic pricing scheme
Xn depend on the power demand and generation. The power
Pd,EVCS = Pi (5) obtained from two renewable sources, namely, the solar PV
i=1
system and the wind farm was checked. Their per unit costs
where Pi denotes the power demand of the ith EVCS were calculated based on their respective generations. The
connected to the system. pseudocode for developing a dynamic pricing scheme is
The total power demand in the system (Pd,total ) is given by presented in Algorithm 1.
Xm
Pd,total = Pd,EVCS + Pload,j + Ptr.losses (6) Algorithm 1 Dynamic Pricing Scheme
j=1 1 Input:

Total Power Demand Pd,t
where Pload,j denotes the power demand of each load con-
Renewable Power Generated (Pen,ren)
nected to the system and Ptr.losses represents the transmission
Cost per kWh (Solar PV System - S1 , S2 , S3 ;
losses in the system.
Wind farm −W1 , W2 , W3 )
The SOC of the EV battery was constantly monitored to
Generated Power (Solar PV system Psolar , Wind farm
ensure that the battery did not enter unhealthy cycles of over-
Pwind )
charging or deep discharging. The power demand of an EV
Lower limits −Psolar,1 , Pwind, 1
depends on the SOC of the battery, and a battery with a higher
Upper limits −Psolar,u , Pwind,u
SOC has a lower power demand.
2 Output: Optimised charging cost and dynamic pricing
scheme
20% ≤ SOC ≤ 80% (7)
3 Initialization: Set the initial charging cost as zero
4 Compute total power demand in the system for a time
The charging duration is an important parameter for
instant ‘t’ according to equation (6)
optimizing the cost, as it is another indicator of the power
5 Calculate the total renewable power generated at time
demand from the perspective of the EVCS.
‘t’ according to equation (8)
The renewable power generated (Pgen,ren ) is evaluated as
6 If Psolar ≤ Psolar,1 , do
the sum of the generations from the solar PV system, wind
7 PV Cost = S1
farm, and the biogas plant, and can be expressed as
8 elseif Psolar,1 , < Psolar < Psolar,u do
Xn 9 PV Cost = S2
Pgen,ren = (PPV,t + Pwind,t + Pbiogas,t ) (8) 10 else do
t=1
11 PV Cost = S3
where, PPV,t , Pwind,t and Pbiogas,t indicate the power generated 12 end (If loop)
by the solar PV system, the wind farm and the biogas plant at 13 If Pwind ≤ Pwind,1 do
time ‘t’, respectively. 14 Wind Cost = W1
15 elseif Pwind,1 < Pwind < Pwind,u do
A. MAXIMIZATION OF RENEWABLE ENERGY UTILIZATION 16 Wind Cost = W2
The power generated from renewables, together with the 17 else do
power from the grid, was used to provide power to the system. 18 Wind Cost = W3
When the power generated by renewables exceeds the power 19 end (If loop)
demand, no power is drawn from the grid. In contrast, when 20 If Pgen,ren ≥ Pd,t do
the generated renewable power is less than the total power 21 Dynamic Price calculation based on per unit
demand, a deficit is provided by the grid. The power drawn cost of renewable sources
from the grid (Pgrid ) is given by 22 Else do
( 23 Dynamic Price calculation based on per unit
0, if Pgen,ren = Pd,total cost of renewable sources and grid tariff
Pgrid = (9)
Pd,total − Pgen,ren , if Pgen,ren ̸ = Pd,total 24 end (If loop)
25 end
If the power generated from the renewable power sources
exceeds the power demand, the excess is fed back to the grid. The generation from renewables is compared with the
The Renewable Energy Utilization (%RE) was calculated as a demand. If the power obtained from renewable energy
percentage of the power contribution of the renewable sources sources is greater than or equal to the demand, then the
when compared to the total power obtained. per-unit charging cost is determined from the costs of the
renewable sources. However, if the renewable sources are
Pgen,ren unable to meet the power demand, the grid tariff and the
%RE = ×100 (10)
Pgen,ren + Pgrid respective per -unit cost of the renewables at that instance of

VOLUME 12, 2024 116357

 J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

FIGURE 3. Metaheuristic algorithms for minimizing charging cost in modified IEEE 33 bus system.

time are considered. The dynamic pricing scheme provides a electrical parameters are also considered. The per-unit costs
constant per kilowatt hour (kWh) cost every hour. of the solar PV system and wind farm vary depending on
Three metaheuristic algorithms, namely, Artificial Bee the individual power generation, as depicted in Algorithm 1.
Colony (ABC), Social Group Optimization (SGO), and These costs are determined using metaheuristic algorithms,
Particle Swarm Optimization (PSO) are used to determine as shown in Figure 3.
the dynamic pricing scheme. These optimization techniques
have better computational efficiency than non-linear pro- 1) PARTICLE SWARM OPTIMIZATION (PSO)
gramming, which has also been analyzed. The voltage, PSO is a stochastic optimization method based on the swarm
frequency, power and current data from the different IEDs behavior of herds, fish, insects, and birds. In this technique,
were used in the SCADA master unit as the input for each potential solution is a particle, with the position and
the MAEMS algorithms. The constraints for each of these velocity as attributes. The particles in the swarm change

116358 VOLUME 12, 2024

J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

their positions based on their learning experience. Thus, the Each employed bee ‘xn ’ in the swarm ‘Xn ’ generates a new
relative position of the particles is updated from one iteration solution (vm
n ) in the vicinity of its present position as given by
to the next and gradually converges to the optimum value.
vm m m m m


Each particle has a personal best position (pBest), and each n = xn + ∅n · xn − xk (15)
swarm has a global best position (gBest) [26], [27]. where ‘xk ’ is a random solution, ‘m’ is a random dimension
The velocity of particle ‘n’ during the iteration ‘m’ (vm
n ) is index and ‘∅’ is a random number in the range [−1, 1].
given by After the solution set ‘Vn ’ is obtained, a greedy selection is
  performed. If the fitness of ‘Vn ’ is better than that of ‘Xn ’,
vm
n = wvn
m−1
+ c1 r1 pBestm−1n − xm−1
n ‘Xn ’ is replaced by ‘Vn ’.
  If a food source is abandoned, it is replaced with ‘Xn ’ given
+ c2 r2 gBestm−1 − xm−1 n (11) by
where ‘w’ is the inertia weight used to balance the global and n = lbm + rand(0, 1) · (ubm − lbm )
xm (16)
local exploration, ‘c1 ’ and ‘c2 ’are acceleration coefficients where rand (0,1) is a random number between 0 and 1, and
and ‘r1 ’ and ‘r2 ’ are random vectors evenly distributed in the lb, ub are the lower and upper bounds of dimension ‘m’.
range [0, 1]. MAEMS was implemented on a modified IEEE 33 bus
The position of particle ‘n’ during the iteration ‘m’ (xm n) system, as described in the following section.
can be given by
IV. TEST BED SYSTEM
xm m−1
n = xn + vm
n (12)
This study was conducted using a modified IEEE 33 bus
system. The EVCS and renewable sources are appended
2) SOCIAL GROUP OPTIMIZATION (SGO)
to the system. The MAEMS was developed using the
SGO is based on the group solving abilities of humans, which
MATLAB Simulink platform. The voltage and current
are superior to those of individual problem solving. In the
sensors connected to each node gather data from the loads
improving phase, each person’s knowledge is honed with the
and generators and send them to the SCADA master unit via
influence of the best person (gBest) in the group [28]. The
the RTUs. The MAEMS in the SCADA master unit optimizes
updating for person ‘n’ for the dimension ‘m’ (Xnewm n ) is this problem.
given as

Xnewm m m A. MODIFIED IEEE 33 BUS SYSTEM

n = c · Xoldn + (r · (gBest−Xoldn )) (13)
An IEEE 33 bus system is used as the base [30]. The
where ‘c’ is the self-introspection parameter that lies between system voltage, frequency and line parameters were altered to
0 and 1 and ‘r’ is a random number between 0 and 1. ‘Xnew’ match the Indian standards [31], [32]. Renewable sources are
is accepted if it gives a better fitness than the older value incorporated into the system. The grid at node 1 is considered
‘Xold’. to be a swing bus rated at a 415 V line-to-line and 50 Hz
In the acquiring phase, a person in the group (Xn ) interacts frequency. A graphical representation of the modified system
and learns from the best person and from others who have is shown in Figure 4.
better knowledge (Xr ), such that A wind farm rated at 30 kW was located on bus 11.
The output power of the wind farm is proportional to its
Xnewm m m m



n = c · Xoldn + r1 · Xn − Xr wind speed. The nominal and maximum wind speeds were
+ (r2 ·(gBest−Xm n )) (14) considered to be 13.5 m/s and 15 m/s, respectively. A solar PV
system rated at 35 kW was made available at bus 18. The rated
where ‘r1 ’ and ‘r2 ’ are random number sequences between output power was obtained at an irradiance of 1000 W/m2 and
0 and 1, respectively. ‘Xnew’ is accepted if it gives a temperature of 25◦ C. The irradiance profile varied throughout
better fitness than the older value ‘Xold’. In this phase, the day, with the maximum value being obtained at noon. The
a person gains knowledge from members who are more irradiance fell to zero before sunrise and after sunset.
knowledgeable through random interactions. A biogas plant rated at 45 kW was placed on bus 25. The
output of the biogas plant was considered constant throughout
3) ARTIFICIAL BEE COLONY (ABC) ALGORITHM the 24 hour period in this study. Three loads were part of
The ABC algorithm is based on the foraging behavior of the system. Representative of all possible loads in a practical
bees. There are three types of honeybees, employed bees, system, they are positioned at nodes 3, 6 and 33 and are
onlooker bees, and scout bees. The employed bees search for rated at 12 kW, 1 kW and 7 kW respectively. The peripherals
food around the source stored in their memory and provide attached to the respective nodes of the system are presented
information about the sources to the onlooker bees. Onlooker in Table 1.
bees evaluate and select suitable food sources. Scout bees Three EVCSs were considered at nodes 2, 19, and 20.
arise from a few employed bees who abandon their food The locations are selected based on [33], which proposes
sources and search for new ones [29]. the optimal locations for the placement of the EVCS in an

VOLUME 12, 2024 116359

 J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

FIGURE 4. Graphical representation of modified IEEE 33 bus system with implementation of MAEMS.

TABLE 1. Peripherals in the modified ieee 33 bus system. EVCSs at nodes 2 and 20 provide for AC charging, whereas
those at node 19 provides consumers with both AC and DC
charging.
EVs that may be charged via AC charging are equipped
with an onboard charger, which converts the AC power from
the grid to DC. However, the DC fast charger, houses the
converter within it enabling the EV batteries to be directly
charged by it. AC charging is more economical than DC
charging. EVCSs are designed for specific power ratings
and the time to full charge depends on the SOC of the
connected EV battery and the voltage rating of the battery.
The maximum current that can be drawn from the AC EVCS
IEEE 33 bus system based on the voltage sensitivity factor is 15 A and that from DC EVCS is 200 A, based on Indian
(VSF), voltage stability index (VSI), and ampacity limit. The standards [32], as specified in Table 2.

116360 VOLUME 12, 2024

J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

TABLE 2. Specifications of evcs in india. MATLAB Simulink platform. The simulation was performed
for a duration of 24 hours, that is, a day, and the variations in
the connected EVs and output of the sources were considered
accordingly. The results are discussed below.

A. SOURCES
A solar PV system, wind farm and biogas plant were present
in the system as mentioned in Section IV. The power
generated from each renewable energy source is shown in
Figure 5, and the total renewable power generation is shown
in Figure 6 (Step 5, Algorithm 1).
B. PROPOSED TARIFF STRUCTURE
A dynamic pricing scheme is proposed based on the
availability of renewable sources. The higher the renewable
generation, the lower is the price specified for the respective
source. Because the biogas plant has the same generation
throughout the day considered, its per-unit cost, in rupees
(Rs.) is fixed, and is considered as Rs. 6.875. The per-
unit costs of the solar PV system (S1 , S2 and S3 ) and
wind farm (W1 ,W2 and W3 ) in Algorithm 1 were optimized
using metaheuristic algorithms. The lower and upper bounds
assumed for the cost of renewable sources are presented in
Table 3, as per Indian standards [34].

TABLE 3. Bounds for optimization.

FIGURE 5. Power obtained from a) Wind Farm b) Solar PV System c)

Biogas Plant.

The Time of Day (TOD) tariff, as given in [35] is

considered the grid tariff, as shown in Table 4.

TABLE 4. Time of day grid tariff.

FIGURE 6. Total power obtained from renewable sources.

V. RESULTS AND DISCUSSION From Figure 5, it can be observed that the wind profile
The proposed MAEMS algorithm for the EVCS was is dynamic and varies with time. The generated solar power
implemented in the modified IEEE 33 bus system in the depended on the available solar irradiance. Here, solar

VOLUME 12, 2024 116361

 J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

irradiance and hence, solar power, is absent from 0 to 6 hours The grid meets the demand that cannot be satisfied by
and 17 to 24 hours. The generated solar power reached its renewable power generation. The excess power generated
peak value around 11 hours. The output from the biogas plant was fed back to the grid, as indicated by the negative values.
was constant throughout the considered 24 hour period. Thus, The power across the point of common coupling, that is,
the total renewable power generated is dynamic in nature, the point of connection between the grid and the rest of the
with a maximum at noon. system is depicted in Figure 9.

B. POWER DEMAND
Figure 7 shows the number of EVs connected to the system,
which is used for the proposed work.

FIGURE 9. Power across the point of common coupling.

FIGURE 7. Total number of EVs connected to the system.

C. RENEWABLE UTILIZATION
The percentage of renewable energy utilization throughout
the day is shown in Figure 10. This indicates the percentage
of the power contribution of the renewable sources when
compared to the total power obtained. An average of 68.94%
of renewable power was used throughout the day.

FIGURE 8. Total power demand.

The power demanded by the loads connected to nodes 3, 6,

and 33 was 20 kW and remained constant throughout the day.
The total power demand over the course of the day is shown in
Figure 8 (step 4, Algorithm 1). The total power demand is the
sum of the demand from the EVCS, other connected loads
and the transmission losses in the system. The total power
demand fluctuated, with the highest values observed between
6 to 12 hours and 18 to 22 hours. FIGURE 10. Renewable utilization.

116362 VOLUME 12, 2024

J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

D. DYNAMIC PRICING SCHEME Consider the time interval of 0 to 6 hours. The power
The cost per kWh to be paid by EV owners if the grid tariff is demand during this time was between 35 kW and 66.6 kW.
applied to the EVCS is shown in Figure 11. This corresponds On the other hand, the power generated by the renewables lies
to the TOD tariff shown in Table 4 of Section IV. In this case, between 49.39 kW and 61.22 kW. Because the solar power is
the consumer pays an average of Rs. 6.579 per kWh. zero at this time, a per-unit cost of S1 , as shown in Table 3,
is considered for the solar PV system. However, the wind
profile varied between 4 kW and 15 kW. Thus the per-unit
cost of the wind system may be W1 or W2 , depending on the
instantaneous power profile. The consumer pays an average
of Rs. 4.444 per kWh over the 24 hour period, when non-
linear programming is used.

2) DYNAMIC PRICING SCHEME – PSO

To perform optimization, 50 swarms were considered. The
minimum and maximum inertia weights were assumed to be
0.3 and 0.8. The acceleration factors were assumed to be 2 and
2.01, respectively. The per kWh cost determined using PSO
is shown in Figure 13.

FIGURE 11. Time of day grid tariff.

The dynamic pricing scheme obtained using non-linear

programming and the metaheuristic algorithms, according to
the steps of Algorithm 1, is presented below.

1) DYNAMIC PRICING SCHEME – NON-LINEAR

PROGRAMMING
The per-kWh cost determined using non-linear programming,
based on the interior point algorithm, is shown in Figure 12.
The upper and lower bounds specified in Table 3 were used
as constraints for this method.
FIGURE 13. Dynamic pricing scheme – PSO.

Consider the time interval of 6 to 12 hours. The power

demand during this time lies between 66.6 kW and 109.2 kW.
On the other hand, the power generated by the renewables lies
between 49.39 kW and 107.7 kW. Because the solar power
and the wind power generations vary over the complete range
specified in Table 3, the per-unit costs of the systems would
change depending on the instantaneous power profile. The
grid tariff is included in the calculation of the dynamic pricing
scheme when the demand is not met by the renewable energy
generation. The consumer pays an average of Rs. 4.439 per
kWh over the 24 hour period, when PSO is used.

3) DYNAMIC PRICING SCHEME – SGO

To perform the optimization, a group of 50 people was
considered. The self-introspection factor was assumed to be
0.7. The per-kWh cost determined using the SGO is shown in
FIGURE 12. Dynamic pricing scheme – non-linear programming. Figure 14.

VOLUME 12, 2024 116363

 J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

FIGURE 14. Dynamic pricing scheme – SGO. FIGURE 15. Dynamic pricing scheme – ABC.

TABLE 5. Dynamic pricing scheme.

Consider the time interval of 12 to 18 hours. The power
demand during this time lies between 33.9 kW and 86.9 kW.
In contrast, the power generated by the renewables lies
between 54.05 kW and 99.67 kW. Because the solar power
and the wind power generations vary over the complete range
specified in Table 3, the per-unit costs of the systems would
change depending on the power profile at a particular instant
of time. The grid tariff is included in the calculation of the
dynamic pricing scheme when the demand is not met by the
renewable energy generation. The consumer pays an average
of Rs. 4.423 per kWh over the 24 hour period, when SGO is
used.
case of the proposed dynamic pricing scheme. The average
4) DYNAMIC PRICING SCHEME – ABC ALGORITHM per unit costs obtained from the non-linear programming,
To perform the optimization, 100 food sources and 50 honey- PSO, SGO and ABC algorithms were Rs. 4.444, 4.439,
bees were considered. The per kWh cost determined using 4.423 and 4.413, respectively. The ABC algorithm proved to
the ABC algorithm is shown in Figure 15. In the duration be the most beneficial in terms of cost.
between 18 and 24 hours, the power demand lies between The computational efficiency of the algorithms was tested
42.3 kW and 105.9 kW. However, the power generated by using an Apple M2 chip with 8 core GPU and 8 GB
the renewables was between 45 and 75 kW. Because the solar RAM. In terms of computational efficiency, the metaheuristic
power is zero in this duration, S1 is considered as the per- algorithms performed better than non-linear programming
unit cost for the solar system. However, the wind profile (17.54 seconds). Among the metaheuristic algorithms,
varied over the complete range specified in Table 3. Thus the SGO and ABC algorithms required similar times of
the per-unit cost for the wind system may be W1 , W2 or 11.78 seconds and 11.66 seconds, respectively. They were
W3 , depending on the instantaneous power profile. The grid quicker than PSO, that took 16.29 seconds for the same initial
tariff is included in the calculation of the dynamic pricing conditions.
scheme when the demand is not met by the renewable energy Hence, the ABC algorithm is the best method for this
generation. The consumer pays an average of Rs. 4.413 per application in terms of cost as well as computational
kWh over the 24 hour period, when the ABC algorithm is time. If the generated power and demand are changed, the
used. algorithm continues to provide the best possible cost.
A comparison of the results obtained using the different
algorithms is presented in Table 5. If consumers are charged E. PAYBACK PERIOD
using the time-of-day grid tariff, they pay an average of The time required to repay the cost of the original investment
Rs.6.579 per kWh. In contrast, they pay lesser per kWh in the on a project is called the payback period [36]. A lower

116364 VOLUME 12, 2024

J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

payback period indicated a more feasible project. The TABLE 8. Main differences between the proposed work and existing
research.
payback period (PBP) may be given by
Investment Cost
PBP = years (17)
Annual Revenue

TABLE 6. Payback period for the renewable sources.

The payback period was evaluated for each renewable

source based on their ratings, as listed in Table 6. The
payback periods obtained for the solar PV system, wind farm
and biogas plant were 6.4 years, 3.3 years and 0.66 year
respectively.

F. GREENHOUSE GAS (GHG) EMISSIONS

The GHG emissions from each renewable source were
calculated based on their respective ratings, as shown in
Table 7. The solar PV system did not release GHGs
during operation. The wind farm contributes approximately
726 kilograms (kgs) of GHGs annually. The biogas plant
contributes approximately 6159.4 kgs of GHGs per annum.
For the same amount of energy obtained from renewables,
a coal power plant emits 138902.8 kgs of GHGs.

TABLE 7. Estimated ghg emissions for the renewable sources.

G. THE MAIN DIFFERENCES BETWEEN THE PROPOSED

AND EXISTING WORK
The proposed work is compared with the existing research in
Table 8. The developed system is superior in the following
respects:
[1] The modelled EVCS considers both AC and DC
charging. [4] Techno economic assessment of the developed system
[2] The EMS was implemented in SCADA using meta- is made.
heuristic algorithms, thereby making the system more
robust and reliable. VI. CONCLUSION
[3] Three renewable sources were considered for the With growing concerns about climate change and increased
system: solar, wind and biogas. pollution, the transition to EVs and the integration of

VOLUME 12, 2024 116365

 J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

renewable sources in the power system has proven to be an [11] H. N., S. Hampannavar, D. B., O. M. Longe, S. Mansani, and
effective method to curb the ill effects on the environment. V. Komanapalli, ‘‘Assessment of microgrid integrated biogas–photovoltaic
powered electric vehicle charging station (EVCS) for sustainable future,’’
This study deals with the development of a metaheuristic Energy Rep., vol. 9, pp. 139–143, Nov. 2023.
algorithm-based Energy Management System for an IEEE [12] M. M. Rahman, E. A. Al-Ammar, H. S. Das, and W. Ko, ‘‘Optimal design
33 bus system incorporated with an EVCS. Three renewable of grid connected PV battery system for probabilistic EVCS load,’’ in Proc.
Adv. Sci. Eng. Technol. Int. Conf. (ASET), Feb. 2020, pp. 1–6.
sources, namely, a solar PV system, a wind farm and a [13] S. Hussain, M. A. Ahmed, K.-B. Lee, and Y.-C. Kim, ‘‘Fuzzy logic weight
biogas plant were considered. A dynamic pricing scheme was based charging scheme for optimal distribution of charging power among
developed to ensure that consumers pay a lower average price electric vehicles in a parking lot,’’ Energies, vol. 13, no. 12, p. 3119,
Jun. 2020.
for EV charging, through the MAEMS in the SCADA master [14] A. K. Karmaker, Md. A. Hossain, H. R. Pota, A. Onen, and J. Jung,
unit. Furthermore, the payback period and GHG emissions ‘‘Energy management system for hybrid renewable energy-based electric
associated with the proposed system were estimated. vehicle charging station,’’ IEEE Access, vol. 11, pp. 27793–27805, 2023.
[15] P. Roy, J. He, T. Zhao, and Y. V. Singh, ‘‘Recent advances of wind-solar
The ABC algorithm proved to be the most beneficial hybrid renewable energy systems for power generation: A review,’’ IEEE
method, with the lowest average per-unit charging cost and Open J. Ind. Electron. Soc., vol. 3, pp. 81–104, 2022.
the highest computational efficiency. The proposed scheme [16] M. Z. Malik, K. Zehra, I. Ali, Ubedullah, M. Ismail, A. Hussain, V. Kumar,
M. Abid, and M. H. Baloch, ‘‘Solar-wind hybrid energy generation
using the ABC algorithm is 32.92% more cost efficient system,’’ in Proc. IEEE 23rd Int. Multitopic Conf. (INMIC), Nov. 2020,
than the conventional case. The average renewable energy pp. 1–6.
utilization of the proposed model was approximately 68.94%. [17] A. N. Celik, ‘‘Optimisation and techno-economic analysis of autonomous
photovoltaic-wind hybrid energy systems in comparison to single photo-
The modelled system saw an approximately 95% reduction voltaic and wind systems,’’ Energy Convers. Manage., vol. 43, no. 18,
in GHG emissions, when compared to its coal-powered pp. 2453–2468, Dec. 2002.
counterpart. [18] M. S. Alam and S. A. Arefifar, ‘‘Energy management in power distribution
systems: Review, classification, limitations and challenges,’’ IEEE Access,
Thus, the developed MAEMS achieves minimization of vol. 7, pp. 92979–93001, 2019.
charging costs and maximization of renewable energy uti- [19] Y. Zahraoui, I. Alhamrouni, S. Mekhilef, M. R. B. Khan,
lization. The reduced GHG emissions and favorable payback M. Seyedmahmoudian, A. Stojcevski, and B. Horan, ‘‘Energy
management system in microgrids: A comprehensive review,’’
periods highlight the benefits of incorporating renewable Sustainability, vol. 13, no. 19, p. 10492, Sep. 2021.
sources into power systems. [20] R. K. Chauhan, M. L. Dewal, and K. Chauhan, ‘‘Intelligent SCADA
system,’’ Int. J. Power Syst. Optim. Control, vol. 2, no. 1, pp. 143–149,
Jan. 2010.
REFERENCES [21] A. J. Calderón Godoy and I. González Pérez, ‘‘Integration of sensor and
actuator networks and the SCADA system to promote the migration of the
[1] A. G. Olabi and M. A. Abdelkareem, ‘‘Renewable energy and climate legacy flexible manufacturing system towards the Industry 4.0 concept,’’
change,’’ Renew. Sustain. Energy Rev., vol. 158, no. 1, pp. 1–24, 2022. J. Sensor Actuator Netw., vol. 7, no. 2, p. 23, May 2018.
[2] F. Perera and K. Nadeau, ‘‘Climate change, fossil-fuel pollution, and [22] D. K. Mohanta and J. B. Reddy, Synchronized Phasor Measurements for
children’s health,’’ New England J. Med., vol. 386, no. 24, pp. 2303–2314, Smart Grids, document 137836, 2017.
Jun. 2022. [23] Y. D. Shahakar and D. A. Shahakar, ‘‘Introduction and classification of
[3] M. Muratori, M. Alexander, D. Arent, M. Bazilian, P. Cazzola, E. M. Dede, classical and advanced optimization methods for electric power system,’’
J. Farrell, C. Gearhart, D. Greene, A. Jenn, M. Keyser, T. Lipman, in Proc. Int. Conf. Innov. Trends Adv. Eng. Technol., Dec. 2019, pp. 53–59.
S. Narumanchi, A. Pesaran, R. Sioshansi, E. Suomalainen, G. Tal, [24] C. L. B. Silveira, A. Tabares, L. T. Faria, and J. F. Franco, ‘‘Mathematical
K. Walkowicz, and J. Ward, ‘‘The rise of electric vehicles—2020 status optimization versus metaheuristic techniques: A performance comparison
and future expectations,’’ Prog. Energy, vol. 3, no. 2, Mar. 2021, for reconfiguration of distribution systems,’’ Electr. Power Syst. Res.,
Art. no. 022002. vol. 196, Jul. 2021, Art. no. 107272.
[25] W. Wong and C. I. Ming, ‘‘A review on metaheuristic algorithms: Recent
[4] M. H. Ghodusinejad, Y. Noorollahi, and R. Zahedi, ‘‘Optimal site selection trends, benchmarking and applications,’’ in Proc. 7th Int. Conf. Smart
and sizing of solar EV charge stations,’’ J. Energy Storage, vol. 56, Comput. Commun. (ICSCC), Jun. 2019, pp. 1–5.
Dec. 2022, Art. no. 105904.
[26] D. Wang, D. Tan, and L. Liu, ‘‘Particle swarm optimization algorithm: An
[5] G. Krishna, ‘‘Understanding and identifying barriers to electric vehicle overview,’’ Soft Comput., vol. 22, no. 2, pp. 387–408, Jan. 2017.
adoption through thematic analysis,’’ Transp. Res. Interdiscipl. Perspect., [27] A. G. Gad, ‘‘Particle swarm optimization algorithm and its applications:
vol. 10, Jun. 2021, Art. no. 100364. A systematic review,’’ Arch. Comput. Methods Eng., vol. 29, no. 5,
[6] A. Golder and S. S. Williamson, ‘‘Energy management systems for electric pp. 2531–2561, Apr. 2022.
vehicle charging stations: A review,’’ in Proc. 48th Annu. Conf. IEEE Ind. [28] S. Satapathy and A. Naik, ‘‘Social group optimization (SGO): A new
Electron. Soc., Oct. 2022, pp. 1–6. population evolutionary optimization technique,’’ Complex Intell. Syst.,
[7] K. Kouka, A. Masmoudi, A. Abdelkafi, and L. Krichen, ‘‘Dynamic energy vol. 2, no. 3, pp. 173–203, Aug. 2016.
management of an electric vehicle charging station using photovoltaic [29] Y. Xu, P. Fan, and L. Yuan, ‘‘A simple and efficient artificial bee colony
power,’’ Sustain. Energy, Grids Netw., vol. 24, Dec. 2020, Art. no. 100402. algorithm,’’ Math. Problems Eng., vol. 2013, pp. 1–9, Aug. 2013.
[30] K. R. Guerriche and T. Bouktir, ‘‘Maximum loading point in distribution
[8] H. N., S. Hampannavar, D. B., and S. M., ‘‘Analysis of microgrid
system with renewable resources penetration,’’ in Proc. Int. Renew. Sustain.
integrated photovoltaic (PV) powered electric vehicle charging stations
Energy Conf. (IRSEC), Ouarzazate, Morocco, Oct. 2014, pp. 481–486.
(EVCS) under different solar irradiation conditions in india: A way towards
[31] A. Gautam, R. Shukla, K. Kishore, P. Jain, R. K. Porwal, and N. Nallarasan,
sustainable development and growth,’’ Energy Rep., vol. 7, pp. 8534–8547,
‘‘Analyses of Indian power system frequency,’’ in Proc. IEEE Int. Conf.
Nov. 2021.
Power Syst. Technol. (POWERCON), Sep. 2020, pp. 1–6.
[9] F. Ahmad, M. S. Alam, S. M. Shariff, and M. Krishnamurthy, ‘‘A cost- [32] (2021). Handbook of Electric Vehicle Charging Infrastructure Implemen-
efficient approach to EV charging station integrated community microgrid: tation. [Online]. Available: https://www.niti.gov.in/sites/default/files/2021
A case study of Indian power market,’’ IEEE Trans. Transport. Electrific., -08/HandbookforEVChargingInfrastructureImplementation081221.pdf
vol. 5, no. 1, pp. 200–214, Mar. 2019. [33] S. K. Sahu, S. Kumar, D. Ghosh, and S. Poddar, ‘‘Modelling and analysis
[10] K.-Y. Wu, T.-C. Tai, B.-H. Li, and C.-C. Kuo, ‘‘Dynamic energy of distribution system performance with integration of electric vehicle
management strategy of a solar-and-energy storage-integrated smart charging station in real-time environment,’’ in Proc. IEEE 18th India
charging station,’’ Appl. Sci., vol. 14, no. 3, p. 1188, Jan. 2024. Council Int. Conf. (INDICON), Guwahati, India, Dec. 2021, pp. 1–6.

116366 VOLUME 12, 2024

J. A. J. Shirley et al.: Metaheuristic Algorithm Based Energy Management System for EVCS

[34] A. K. Singh and A. H. Idrisi, ‘‘Evolution of renewable energy in India: R. P. POOJA is currently pursuing the B.Tech.
Wind and solar,’’ J. Inst. Eng. (India), Ser. C, vol. 101, no. 2, pp. 415–427, degree in electrical and electronics engineering
Nov. 2019. with the National Institute of Technology (NIT)
[35] (2019). Report on Time of Day (TOD) Tariff For Tiruchirappalli, Tiruchirappalli, India.
Tangedco. Accessed: Jul. 12, 2024. [Online]. Available: She underwent an internship at leading multina-
http://www.tnerc.gov.in/PressRelease/files/PR-230820220439Eng.pdf tional consultancies, including Petrofac and Tech-
[36] M. Gjelaj, C. Træholt, S. Hashemi, and P. B. Andersen, ‘‘Cost-benefit nip Energies, India, in 2023. She has published one
analysis of a novel DC fast-charging station with a local battery storage
journal article in a well-recognized high-quality
for EVs,’’ in Proc. 52nd Int. Universities Power Eng. Conf. (UPEC),
journal. Her research interests include energy
Aug. 2017, pp. 1–6.
management systems, machine learning, electric
vehicle technologies, embedded systems, power system technologies, green
energies, and power electronics.

MADDIKARA JAYA BHARATA REDDY (Senior

Member, IEEE) was born in India, in 1980.
He received the B.Tech. degree in electrical
and electronics engineering from Acharya Nagar-
juna University, Guntur, India, in 2002, and the
M.E. degree in electrical engineering and the
JENNIE ANGELA JOSE SHIRLEY (Student Ph.D. degree from the Birla Institute of Tech-
Member, IEEE) was born in India, in 2003. She is nology (BIT), Ranchi, India, in 2004 and 2008,
currently pursuing the B.Tech. degree in electrical respectively.
and electronics engineering with the National He has 20 years of experience in teaching and
Institute of Technology (NIT) Tiruchirappalli, research. He is currently a Professor with the Department of Electrical
Tiruchirappalli, India. and Electronics Engineering, National Institute of Technology (NIT)
She interned at Jyoti Sohar Switchgear LLC, Tiruchirappalli, Tiruchirappalli, India. His current research interests include
Oman, and Voltamp Power Oman, in 2022. Her smart grids, substation automation, wide-area protection, digital relaying,
research interests include smart grids, wide-area soft computing applications in power systems, and power system protection.
monitoring systems, electric vehicles, and high- Dr. Reddy is an Editor of the Electric Power Components and Systems
voltage technology. (Taylor & Francis Publications) and a Subject Editor of IET Generation,
Ms. Jose Shirley was awarded the DAAD WISE Scholarship, in 2023, for Transmission & Distribution.
a research internship at the University of Stuttgart, Germany.

VOLUME 12, 2024 116367