Available online at www.sciencedirect.

com
ScienceDirect

Energy Reports 9 (2023) 455–462

www.elsevier.com/locate/egyr

2022 3rd International Conference on Power, Energy and Electrical Engineering (PEEE 2022)
18–20 November 2022

Planning of fast charging stations with consideration of EV user,

distribution network and station operation
Madathodika Asna ∗, Hussain Shareef, Achikkulath Prasanthi
Department of Electrical & Communication Engineering, United Arab Emirates University, Al Ain, United Arab Emirates
Received 16 December 2022; accepted 16 January 2023
Available online xxxx

Abstract
Despite the benefits of electric vehicles (EV) as a sustainable alternative to fossil fuel-powered vehicles, the lack of suitable
fast charging stations (FCS) continues to be a barrier to widespread EV adoption. Different factors affecting EV user, distribution
network and CS operation must be taken into account when selecting CS locations and capacity. The purpose of this study is
to design a FCS location and capacity planning framework that considers EV user travel time and waiting time, distribution
network power loss, and station utilization aspects. Besides, a prior utilization rate-based queuing algorithm is proposed to
determine the CS capacity, which simultaneously reduce user waiting time and affirms adequate station utilization. Then binary
atom search optimization algorithm is employed to solve the FCS location and capacity problem. The proposed model is tested
on an integrated transportation network and distribution network and the numerical results are provided and analysed from the
perspectives of the distribution network, EV user, and CS operation.
© 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
Peer-review under responsibility of the scientific committee of the 3rd International Conference on Power, Energy and Electrical Engineering, PEEE,
2022.

Keywords: Fast charging station; Location and capacity; Station utilization; Grid safety; Atom search optimization

1. Introduction
With growing energy crisis and environmental challenges related to transport sector, the world is currently
focusing on clean transportations such as electric vehicles (EVs). Compared to traditional energy vehicles, EVs
have many advantages, including zero exhaust emissions and zero air pollution. However, successful deployment of
EVs require adequate charging stations (CS) facilities. Planning of CS locations are crucial. Improper allocation of
CSs into the distribution network will cause many impacts on the power grid such as increased power losses,
voltage instability, demand–supply imbalance, etc. [1]. Additionally, insufficient station capacity causes higher
station congestion and longer waiting time for charging. To alleviate the above issues brought on by CSs, there
is a need for careful planning on where to build the CSs and how to distribute the chargers efficiently.
∗ Corresponding author.
E-mail address: [email protected] (M. Asna).

https://doi.org/10.1016/j.egyr.2023.01.063
2352-4847/© 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.
org/licenses/by/4.0/).
Peer-review under responsibility of the scientific committee of the 3rd International Conference on Power, Energy and Electrical Engineering,
PEEE, 2022.
M. Asna, H. Shareef and A. Prasanthi Energy Reports 9 (2023) 455–462

Planning of CS location and capacity have been widely considered in Refs. [2–6]. Ref. [2] studied the
optimization of FCS locations with the goal of minimizing the station investment cost and power loss cost using
hybrid grey wolf optimization-particle swarm optimization (PSO) algorithm. Although the authors of Ref. [2]
optimized CS locations, the study did not take station capacity planning into account. CS capacity is also crucial for
the power loss and the station congestion analysis. Furthermore, authors in Ref. [3] presented a grid safety driven
CS planning model to aid the power quality parameters, namely power loss and voltage profile utilizing hybrid
genetic algorithm-PSO algorithm. However, the authors neglected the EV user aspects and the station congestion
issue. A distribution and transportation network coordinated CS location and capacity model is investigated in
Ref. [4]. The suggested model used binary lightning search algorithm to solve the CS problem while accounting
for the transportation costs, station installation cost and substation loss cost. However, they did not consider the
station utilization and waiting time aspects during station capacity optimization. The influence of travel distance
and waiting time on electric ride hailing fleet is thoroughly investigated in Ref. [5]. However, the authors in Ref. [5]
did not analyse the effect of station utilization on congestion management. Ref. [6] formulated the CS placement
model considering multi objective criteria, such as travel cost, station cost and power loss.
In the previous studies, only one or two elements affecting the CS planning has been examined, they generally
ignored the combinations of user travel time-waiting time factors, grid safety-waiting time factors, etc. However, it
is important to develop an integrated CS planning model that takes these multiple factors into account. In addition,
the majority of existing studies address the station congestion problem solely take waiting time into account but
ignore the station utilization aspect. Based on this, this paper develops a fast CS location and capacity model
with four important factors into account: travel time, waiting time, grid safety and station utilization which aims
at benefiting three different entities of EV user, grid operator and station operator. Moreover, an optimal station
capacity estimation method based on station utilization is suggested, which helps to keep the station busy with EVs
without causing congestion. Recently developed binary atom search optimization algorithm (BASO) is applied to
find the best CS locations and capacity. The effectiveness of the suggested planning scheme is validated by testing
on a real road network and distribution network.

2. Methodology of fast charging station location and capacity planning

This section provides a detailed description of adopted methodology for planning the FCS, depicted in Fig. 1.
The planning design is composed of three major steps; specifically, user travel time calculation, station capacity
estimation and power flow analysis. Input data set from multiple sources is the basis of location planning, specifically
real road network data which includes EV locations and CS locations coordinates as well as distribution network
loading data. This work is an extension of the study described in [7], where the detailed mathematical model is
presented.

Fig. 1. Schematic diagram of FCS location and capacity planning model.

456
M. Asna, H. Shareef and A. Prasanthi Energy Reports 9 (2023) 455–462

2.1. Travel time calculation

EV range anxiety is an important issue that obstructs EV uptake; hence, the CS accessibility must be taken into
account when planning locations. Therefore, we estimate the EVs’ travel times to the station. The Google map API
service is utilized to determine the travel time from EV location to CS locations, which provides accurate travelling
time between each pair of EV location-CS location, which are defined by latitude and longitude coordinates, using
real road traffic data.
The time taken by an EV i to all FCSs can be expressed as
Tit = t1 t2 t3 · · · t j · · · t N cs
[ ]
(1)
where ti is the travel time to CS j and N cs is the total number of CSs.
Based on the above travel time information, EV user chose the nearest CS. The minimum travel time for an EV
i to the nearest FCS can be obtained as
t t
Tmin i = min (Ti ) (2)
The obtained travel time data is further is used to estimate the number of EVs that goes to each CS.

2.2. Capacity planning for FCS

Once the number of EVs going to each CS is computed, the next step is to determine adequate CS capacity,
taking user waiting time and CS capacity utilization factors into account. User waiting time here specifically refers
to time spend by an EV in the queue before starting the charging process and CS capacity utilization refers to
proportion of time the chargers are occupied with EV. Capacity utilization is of utmost importance to CS owners as
it reflects the productivity and profitability of CS. For instance, if a station is operating at 50% capacity, then, on
average, half of its chargers are often idle. To avoid such idle conditions, a minimum value for the utilization rate
must be established while planning the CS capacity. Taking this into account and using the concept of M/M/c multi
server queuing theory, we estimate the charging units to be placed at each station. The flow diagram of capacity
planning is described in Fig. 2. Initially, a prior station utilization value is defined and using Eq. (3) the station
capacity (NiC ) at a CS j is calculated.
N ev
j ×T
ch
N cj = (3)
η

Fig. 2. Flow chart of station capacity estimation.

Here N ev
j denotes the number of EVs at CS j, T
ch
refers to average charging time taken by EVs and η is the
predefined station utilization rate. T ch depends on the EV State of Charge (SoC) and in this study we assume that
457
M. Asna, H. Shareef and A. Prasanthi Energy Reports 9 (2023) 455–462

the SoC of EVs arriving at the station is a random value that varies between 20% and 50%. Then using the piecewise
linear SoC-time curve, the charging time is calculated. (For more details, see Ref. [7]).
Further, to ensure user satisfaction, average waiting time is restricted within the tolerable limit. The mathematical
formula for calculating the waiting time, T jw are provided below.
N cj +1
(N ev ch
j ×T ) × P jo
T jw = (4)
N ev c c ev ch 2
j × (N j − 1)!(N j − N j × T )

where
N cj −1 N cj
∑ (N ev ch n
j ×T ) (N ev ch
j ×T ) 1
P jo =[ + × ] (5)
n=0
n! N cj ! (1 − η)
where P jo gives the probability that there are no EVs waiting to be charged at the station.
If the number of chargers found using Eq. (3) cannot satisfy the waiting time constraint, the charger capacity
is incremented until the waiting time is set within the specified limits. In this way an optimal CS capacity can be
achieved that balances utilization rate and waiting time.

2.3. Power flow analysis

The operation of FCSs negatively alter the power flow at the distribution network, due to their high-power
consumption. In this study, real power losses and voltage profile degradation associated with EV fast charging are
taken into consideration while placing CSs. The power loss associated with any FCS has direct relationship with
the CS capacity, the number of EVs arriving at the station and also the bus to which the station is connected.
The power consumed by any FCS j is calculated as follows:
P jcs = N ev c
j × Nj × P
ch
(6)
where P ch represents the rated charger power.
Then using Eq. (7), the total active load at bus m to which the CS j is connected can be obtained. Here I Pmb
represents initial load at bus m before connecting the FCS.
T Pmb = P jcs + I Pmb (7)
Thereafter, the power loss in the distribution network is computed using the backward and forward sweep power
flow method.
N b
∑
Ploss = gm,n [Vm2 + Vn2 − 2Vm Vn cos(θm − θn )] (8)
m,n=1
m̸=n

where N b is the number of buses in the distribution network, gm,n is the conductance of line m − n, V is the
magnitude of bus voltage and θ is the angle of bus voltage.
In addition to power loss calculation, magnitude of voltage at each bus is monitored to avoid voltage collapse
due to EV charging.

3. Formulation of location and capacity optimization problem

In order to find the best FCS locations, the planning model described above is transformed into a discrete
optimization problem which comprise of decision variables, objective functions and constraints. The decision
variable in this case, is a binary vector having {0,1} values that represents the FCS location, as given in Eq. (9).
According to Eq. (10), a ‘1’ means FCS at that location has been selected while a ‘0’ means it has been discarded.

[ ]
L = l1 l2 l3 · · · li · · · l N cs (9)
458
M. Asna, H. Shareef and A. Prasanthi Energy Reports 9 (2023) 455–462

where
{
1, i f FC S is selected
li = (10)
0, otherwise
As mentioned in the preceding section, the primary considerations in this CS planning are EV user satisfaction,
CS operation and distribution grid impact. For the purpose of calculating CS capacity, EV user satisfaction and
CS operation were taken into account. The capacity allocation plan will directly affect the optimal solution of the
location optimization problem; therefore, the objective function is defined with the remaining factor distribution
grid impact as its top priority. Specifically, to minimize the total power loss caused by FCSs.
The objective function is expressed as follows:
cs
min Ploss = T Ploss − I Ploss (11)
cs
where is the additional power loss when FCSs are connected to the grid, T Ploss is the total power loss when
Ploss
FCSs are present and I Ploss is the initial power loss when no CS are connected to the grid.
Different constraints of the optimization problem are as follows.
(1) Bus Voltage Constraint
The bus voltage of the distribution network shall meet its upper and lower limit constraints:
Vi min ≤ Vi ≤ Vi max ∀i ∈ N b (12)
(2) Waiting time constraint
The charging wait time should not exceed the tolerable waiting time (T jwmax ).
T jw < T jwmax j = 1, 2, . . . . . . N cs (13)
(3) Charging station constraint
Number of FCS to be installed in the study area shall meet its upper and lower limit constraints. This indicates
that at least one CS should be chosen as the solution.
0 < N cs ≤ Nmax
cs
(14)

4. Solution technique of location and capacity optimization problem

The recently developed BASO is utilized to solve the FCS location and capacity problem. BASO is a physics
inspired metaheuristic optimization algorithm that imitates the idea of molecular dynamics and atomic motion [8].
The atomic motion in BASO is governed by three different forces: the interatomic potential, the interaction force,
and the geometry-constrained force. Subsequently, the atoms undergo exploring and exploiting the entire search
space by interacting with each other. The position and velocity vectors are used to depict the motion of the atoms.
The updating process of atom velocity and position are formulated as
vid (t + 1) = randid × vid (t) + aid (t) (15)
⎧ d
⎪0, i f S(vi (t + 1)) ≤ r1
⎪
⎨
xid (t + 1) = 1, i f r1 < S(vid (t + 1)) ≤ r2 (16)
⎪
⎪
⎩ d
xbest (t), otherwise
where vid (t + 1) and xid (t + 1) is the velocity and position during (t + 1)th iteration, similarly vid (t) and xid (t) are
velocity and position during tth iteration. aid is the acceleration of ith atom.
1
S(vid (t + 1)) = d
(17)
1 − e−2vi (t)
1
r1 = rand
3
1 (18)
r2 = + r1
3
459
M. Asna, H. Shareef and A. Prasanthi Energy Reports 9 (2023) 455–462

The main advantage of BASO is that it effectively makes use of the dynamic motion of atoms and keeps a good
balance between exploration and exploitation with a minimal set of control parameters.
Thus, to determine the optimal FCS location and capacity, BASO is employed to minimize the objective function
given in (11), which is subject to the constraints in (12)–(14).

5. Numerical analysis
The proposed location and capacity model is validated on the real road network and distribution network as
depicted in Figs. 3 and 4 respectively. The locations of current gas station are taken as potential sites for charging
station locations and EVs are distributed across various regions based on the population density. Every hour, 100
EVs are predicted to recharge. The distribution network’s bus, line and load data can be found in [1].

Fig. 3. Test distribution network.

Fig. 4. Test road network.

Note that the input parameters related to CS location problem are chosen according to [7]; additionally, algorithm-
specific parameters of BASO are derived from [8]. For the CS capacity planning, the initial utilization rate was
set to 75% and the maximum waiting time as 20 min. The simulation verification is performed using MATLAB
software and it has been the integrated with Google map API web service to retrieve the travel time information.
An average execution time of 7 min is taken to solve the location and capacity optimization problem using BASO.
Table 1 displays the optimal CS locations, station capacity, and power loss associated with FCS planning model.
The optimization yielded 4 best CS locations with 6,5, 10 and 20 chargers at each of the four locations. Selection
of the best solutions depends upon user’s satisfaction, CS operational efficiency and impact on distribution network.
460
M. Asna, H. Shareef and A. Prasanthi Energy Reports 9 (2023) 455–462

Table 1. Optimal location, capacity and objective function value.

cs
CS location No. Station capacity Power loss Ploss (kW) Power loss percentage
FCS1, FCS10, FCS11, FCS12 6,5,10,20 56.17 31.84

Fig. 5. EV travel time to charging station.

Fig. 6. Results of waiting time and station utilization..

Travel time and waiting time for charging indicates the user satisfaction. Fig. 5 illustrates the travel time of EVs
to CSs. 73% of EVs arrive at CSs in less than 15 min and only 3% take longer than 30 min, which supports that
majority of EV users can reach the station quickly. Fig. 6 displays the average wait time for EVs, the number of
EVs at each charging station, and the utilization rate as a percentage. The majority vehicles wait less than 10 min,
with the highest wait being 12 min. The number of EVs waiting at the FCS 12 is large, however, there is very little
waiting time and a high CS utilization rate at this station. It is clear that the proposed CS capacity method can
achieve optimal distribution of chargers that maintains adequate station utilization and shorter waiting time.
Voltage profile variation and power loss indicates the distribution network impact. Fig. 7 illustrates the total load
demand at each bus following the placement of CSs, with bus no. 23 experiencing the highest load due to FCS12.

Fig. 7. Load demand at distribution network.

461
M. Asna, H. Shareef and A. Prasanthi Energy Reports 9 (2023) 455–462

Fig. 8. Voltage profile of distribution network.

Additionally, as seen in Table 1, the power loss has increased by 56.17 kW after locating the FCSs. The voltage
profile of all the buses (Fig. 8) are between 0.9 (Vmin ) and 1.05 (Vmax ), which is acceptable. Thus, the FCS location
model was able to allot CSs while causing less damage to distribution network.
From the above analysis, conclusion can be drawn that the proposed the FCS location and capacity model can
have more positive effects in terms of EV user happiness, enhanced CSs operation, and grid safety.

6. Conclusion
This work develops a planning model for fast CSs, aiming to obtain ideal station location and charger units to
be placed at real road network and distribution network. The problem formulation considers EV user travel time
and waiting time factors, station utilization factors and grid safety factors. The solution methodology is based on
a binary atom search optimization algorithm. The simulation results demonstrates that the suggested location and
capacity planning model is effective. Besides, the station capacity planning results shows that the proposed method
could solve the imbalance between station utilization and waiting time.

Declaration of competing interest

This work was supported by the Roadway, Transportation and Traffic Safety Research Center (RTTSRC) at UAE
University [fund code 31R224-RTTSC(1)-2019].

Data availability
Data will be made available on request.

References
[1] Deb S, Tammi K, Kalita K, Mahanta P. Impact of electric vehicle charging station load on distribution network. Energies 2018;11(1):1–25.
http://dx.doi.org/10.3390/en11010178.
[2] Ahmad F, Iqbal A, Ashraf I, Marzband M, Khan I. Placement of electric vehicle fast charging stations in distribution network
considering power loss, land cost, and electric vehicle population. Energy Sources, A Recover Util Environ Eff 2022;44(1):1693–709.
http://dx.doi.org/10.1080/15567036.2022.2055233.
[3] Awasthi A, Venkitusamy K, Padmanaban S, Selvamuthukumaran R, Blaabjerg F, Singh AK. Optimal planning of electric vehicle charging
station at the distribution system using hybrid optimization algorithm. Energy 2017;133:70–8. http://dx.doi.org/10.1016/j.energy.2017.05.
094.
[4] Mainul Islam M, Shareef H, Mohamed A. Optimal location and sizing of fast charging stations for electric vehicles by incorporating
traffic and power networks. IET Intell Transp Syst 2018;12(8):947–57. http://dx.doi.org/10.1049/iet-its.2018.5136.
[5] Li J, Liu Z, Wang X. Public charging station location determination for electric ride-hailing vehicles based on an improved genetic
algorithm. Sustain Cities Soc 2021;74(March):103181. http://dx.doi.org/10.1016/j.scs.2021.103181.
[6] Asna M, Shareef H, Muhammed MA, Ismail L, Prasanthi A. Multi-objective quantum atom search optimization algorithm for electric
vehicle charging station planning. Int J Energy Res 2022;(July):1–24. http://dx.doi.org/10.1002/er.8399.
[7] Asna M, Shareef H, Prashanthi A, Mohklis H, Errouissi R, Wahyudie A. Analysis of an optimal planning model for electric vehicle
fast-charging stations in al ain city, United Arab Emirates. IEEE Access 2021;9:73678–94. http://dx.doi.org/10.1109/access.2021.3081020.
[8] Too J, Rahim Abdullah A. Binary atom search optimisation approaches for feature selection. Conn Sci 2020;32(4):406–30. http:
//dx.doi.org/10.1080/09540091.2020.1741515.

462