REVISTA INGENIO
Optimal georeferenced deployment of charging stations for electric vehicles in
distribution networks using a trajectory-based heuristic model
Despliegue óptimo georreferenciado de estaciones de carga para vehículos eléctricos en redes de
distribución usando un modelo heurístico basado en trayectoria
Miguel Andaluz | Universidad Politécnica Salesiana, Quito, Ecuador
https://doi.org/10.29166/ingenio.v6i1.4303 pISSN 2588-0829
2023 Universidad Central del Ecuador eISSN 2697-3243
CC BY-NC 4.0 —Licencia Creative Commons Reconocimiento-NoComercial 4.0 Internacional ng.revista.ingenio@uce.edu.ec
, (), -, . -
e progressive increase in the consumption of fossil fuels and the constant eorts to reduce CO2 emis-
sions bring together the search for alternatives and the transport sector being one of the most dependent
on fossil fuels and the cause of approximately 80% of the air pollution, the electric vehicle emerges as an
alternative in mobility. at is why this article proposes a methodology for the optimal location of elec-
tric vehicle charging stations, given in a georeferenced distribution network scenario using a heuristic
for the insertion of electric vehicles, taking into account energy consumption, travel and autonomy. de-
veloped based on real data, reducing the minimum location of charging stations. Evaluated in the distri-
bution network of Santo Domingo-Ecuador, in a way that guarantees a technical and economic balance.
El progresivo aumento del consumo de combustibles fósiles y el constante esfuerzo por reducir las emi-
siones de CO2 se unen para la búsqueda de alternativas, y siendo el sector del transporte uno de los más
dependientes de los combustibles fósiles y causante de aproximadamente el 80% de la contaminación
atmosférica, el vehículo eléctrico surge como una alternativa en movilidad. Es por ello por lo que este ar-
tículo propone una metodología para la ubicación óptima de estaciones de carga de vehículos eléctricos,
en el escenario de una red de distribución georreferenciada, utilizando una heurística para la inserción
de vehículos eléctricos, teniendo en cuenta el consumo de energía, los viajes y la autonomía, desarrollada
con base en datos reales, reduciendo la ubicación mínima de las estaciones de carga; considerando la red
de distribución de Santo Domingo-Ecuador, de manera que garantice un equilibrio técnico y económico.
1. introduction
e need for means of transportation for the develop-
ment of our occupations has been present throughout
the history of the human race. According to the latest
Ecuadorian Energy Balance of 2017, the transport sector
represented 52.29% (45,098 kBEP; 73,427.76 GWh) of
the total national energy consumption [1].
Looking for mechanisms that provide exibility in the
consumption of sources that come from fossil energy, thus
facilitating the migration to other primary energy sour
-
ces. ese would allow the development of the same ac-
tivities but with a minimum environmental impact and
reduced polluting emissions to the planet [2].
e agencies in charge of energy planning must con-
sider scenarios from the point of view of supply and de-
mand where electric mobility systems are representative,
as well as mechanisms and technical inputs not only at
Received: 26/10/2023
Accepted: 19/12/2023
Optimal deployment, heuristic model,
energy consumption, charging stations.
Despliegue óptimo, modelo heurístico,
consumo de energía, estaciones de carga.
5
Optimal georeferenced deployment of charging stations for electric vehicles in distribution networks using
a trajectory-based heuristic model
the engineering level but also in the regulatory framework
that allow the technological transition without producing
disadvantages in the operation of the electric power su-
pply system’s [3].
As electric vehicles increase their market share, it’s
going to get some attention from power companies. Its
inclusion in power systems represents a large increase in
load demand, causing many problems of power quality
degradation, increased energy losses. However, a problem
may occur between the network operators and the owners
of the charging stations since it may be the case that the-
re are dierences because the owners of the charging sta-
tions look for the commercial place where they can charge
the electric vehicles, but at lower cost. On the other hand,
the electricity network operators estimate that the char-
ging stations are located in such a way that they allow a
predetermined number of vehicles to be fed, impacting
the electricity network as little as possible [4].
Various solution methods worldwide have been pro-
posed to locate charging stations. For example, genetic
algorithms and voronoi diagrams have been incorpora-
ted. ese algorithms do not consider very important fac-
tors such as: load prole, consumption, autonomy, and
geographical considerations. at is why we start, for the
optimization process, from candidate sites which can be
conventional service stations, bus stops, shopping cen-
ters, parking lots, parks, etc. Consequently, the proposed
model does not start from scenarios where candidate si-
tes are considered, as would happen with voronoi when
segmenting the area of analysis but starts from a study
area. at is, he knows the study area based on its carto-
graphic reality [5].
Regarding the prole of charge and consumption of
Electric Vehicles (), the historical information of the re-
cords of electric taxis that operate in the city of Loja was
considered, as well as a model developed by the authors
that takes into account the process of charging of EV ba-
tteries modied in a novel model that represents the elec-
tric vehicle battery charging system based on its state of
charge and its current variability and charging time.
e general problem lies in optimally locating and si-
zing the charging stations along a georeferenced distribu-
tion network of 34 nodes, so that the proposed heuristic
starts from candidate sites in the network, of which they
can be public places, that is, it is an iterative method that
knows the study area since this information is extracted
from Open Street Maps (), as well as the use of -
soware to implement graph theory that will allow
nding the nodes and topology that is part of the solution
set. To later evaluate the voltage proles and load losses
simulated in Cymdyst [6].
2. method
2.1. ENERGY CONSUMPTION OF ELECTRIC VEHICLES
e prerequisite for the planning of charging stations is
to create the conditions for an adequate consumption
of electrical energy. On the other hand, electric vehicles
have zero emission characteristics; Low engine noise and
higher propulsion eciency [7], [8], [9], [10], [11].
From the point of view of transport systems, whether
public or conventional, a huge proportion of energy con-
sumption is due to the inecient movement of trac. e
exible energy consumption estimation model is based on
the evaluation of consumption based on data from other
vehicles on the road network, which have the possibility
of being accurate thanks to the dierent vehicle models
and energy eciency [12], [13].
e cost for energy consumption per 100 km of an
electric transport is up to three times less than the cost
of a conventional vehicle that uses fossil fuel, this taking
into account that in Ecuador there are lower rates, both
for gasoline and electricity [14].
When analyzing the real cost of electricity in the
country and the international price of gasoline, the EV is
still lower than that of a thermal combustion vehicle, the-
refore, the electric vehicle is more protable and ecient
even with the fuel subsidy that exists in the country. is
advantage is also visible in Europe [14].
2.2. ELECTRIC DISTRIBUTION NETWORK IN ELEC
TRIC VEHICLES
Within the exponential growth of EVs in moderate
portions, it should not cause too many inconveniences,
however, its wide adoption will probably create an im-
pact on the operation and management of electrical dis-
tribution networks, such as congestion, voltage problems
and load imbalances between phases [15].
Depending on the autonomy of the Electric Vehicle,
the excessive charge of the batteries of said cars will have
an impact on the distribution system, which would in-
crease the load demand, introducing disturbances in the
Interconnected Electric System, which imposes an in-
crease in the generation and make probable reinforce-
ments with the penetration of renewable energy in order
to maintain the balance between what is generated and
consumed [16] (see Table 1).
Approximately, the battery charging speed depends on
the output of the charging station and the technical speci-
cations of the electric car. e peak daily load curve during
a day in the worst case would have a higher consumption
6
Andaluz M.
in the midday and aernoon hours, and a lower consump-
tion in the early morning [17] (see Figure 1).
To guarantee the continuity of the electricity supply and
stabilize the demand curve, which in fact changes accor-
ding to the time and type of daily charge of an electric
vehicle, strategies and procedures are considered where it
does not aect the electrical system and carry out a mas-
sive integration of electric vehicles in a planned way [18].
2.3. MOST REPRESENTATIVE CHARACTERISTICS OF
THE FLEET SYSTEM TO DETERMINE CONSUMPTION
To carry out the cost comparison, the most used com-
bustion vehicle in Ecuador was taken into account, the
model is the Chevy Aveo, and the Nissan Leaf model as
an electric vehicle, for which the initial cost of the elec-
tric vehicle vs. the combustion vehicle, the electric vehicle
has an increase in cost with 85% compared to the cost of
the conventional vehicle.
e costs for energy consumption were determined
based on the technical specications provided by the ma-
nufacturer of the electric vehicle. Several brands and mo-
dels of electric vehicles are expected to soon circulate on
the roads of Ecuador.
It is established that vehicle users generally log less
than 50 km per day, with a performance index for elec-
tric vehicles of 8 km/kWh (0.122 kWh/km) under ideal
conditions of trac and geography, it is concluded that
the energy demanded by the EV of the network would be
0.144 kWh for each kilometer traveled.
2.4. ALGORITHM
e algorithm will be responsible for determining the
optimal location of charging stations by extracting the
characteristics of electric vehicles, through the network
of 34 georeferenced nodes. is route may be useful for
the study of any real scenario of an electrical system de-
pending on the demand scenarios determined by Cymd-
yst (see Table 3).
Consequently, in [33] the heuristic model is explai-
ned in a standard way to solve the programming problem
Table 1
Charge mode data
Charging mode
Mode1 Mode2 Mode3 Mode4
Corrent (A) 16A 32A 64A Hasta 400A
Type of load slow slow Accelerated charging Fast charging
Power (kw) 3,8-11 7,7-22 14,8-43 40-120
Specic Connector No No Yes Yes
Figure 1
Model of charging stations. Electric vehicles in distribution networks
To determine the characteristics of the vehicle eet, a
comparison was made of both the conventional and elec-
tric vehicles, taking into account the route, autonomy
and consumption. For this, the costs of various models
of electric vehicles that are used in the United States wi-
thout taxes and without subsidies are shown, but these
low-end vehicles have already been inserted in Ecuador
(see Table 2).
7
Optimal georeferenced deployment of charging stations for electric vehicles in distribution networks using
a trajectory-based heuristic model
for which the kmeans algorithm will be used to generate
cluster, through the distribution network model of 34 no-
des generated in Cymdyst will be distributed in scenario
where you will get the power, voltage and consumption
at which the charging stations act taking into considera-
tion, public places [19].
3. Results and discussion
Once the model to be used is proposed, a result will be
obtained, which is developed in two scenarios that are
based on a base case study where it will be the starting
point to analyze the dierent behaviors of the network
when the charging stations come into operation. and the
impact on the elements to future case studies of load to
the distribution network.
3.1. ANALYSIS OF OPTIMAL LOCATION OF CHAR
GING STATIONS IN THE DISTRIBUTION NETWORK
One of the objectives of this article is to nd the opti-
mal location for charging stations, to evaluate in a geo-
referenced distribution network taking into account the
characteristics of electric vehicles, inserted in Ecuador
both in their consumption and autonomy compared to
conventional vehicles. based on satisfying user demand.
In the rst instance, it is necessary to extract the coordi-
nates of the area to be studied, through the Open Street
Maps that helps the georeferencing of the scenario, the-
refore the longitude and latitude given below were obtai-
ned as data (see Table 4).
Next, the results obtained from the optimization are
presented to nd the strategic points of charging stations
for the correct functioning of the network, since strate-
gic points of access to the public in the georeferenced ne-
twork were taken into account, such as parks, centres
Table 2
sales prices in Ecuador
Vehicle type Model Sales price in the usa without tax [usd]
Chevy Bolt 37.495
Ford Focus Electric 29.120
Nissan Leaf 30.680
EV Fiat 500e 31.800
BYD e5 34.990
Volkswagen e- Golf 28.995
Table 3
Pseudocode of the solution algorithm
Algorithm placement of charging stations
Step 1: Georeferencing and scenario generation
Step 2: Get the coordinates of the area.
Step 3: Declaration of variables
Xij, Zij, λ
Step 4: Read OSM le
Openstreetmap.
Step 5: Minimum enabling distance.
For k longitud (Xij)
[v]=BVE (λ, Xij)
end for
Step 6: Writing Purpose Function.
Step 7: Candidate sites for the study area
Step 8 End
8
Andaluz M.
commercial, gas stations, the partition of the scenario
arises from the distribution network, therefore the mini-
mum distance from the location of the electric vehicle is
taken depending on the characteristics towards the char-
ging station (see Figure 2).
Figure 2 shows the strategic points to place the char-
ging stations for electric vehicles, the algorithm suggests
a number of 8 candidate sites for the charging stations,
which represents a minimum connectivity route where it
will be analyzed in the network of distribution, the analy-
sis was carried out with fast charging feeders, taking into
consideration the characteristics of electric vehicles, as
well as the technical factors analyzed in chapter 3. In the
comparative study of [20], the number of 5 stations is su-
ggested for 11 buses, taking into account that the load is
1 hour. For a strong network like the one studied, each
station has 3 fast charging points of 25 minutes, which
means that 58 vehicles could be charged in approxima-
tely one hour.
3.2. FLOW SCENARIO WITHOUT LOADING STATIONS
Once the simulation has been carried out in the radial
distribution network, which has 147 bars, one of which
is oscillation, with an operating voltage of 13.60 kV, the
analysis was carried out in 34 main nodes where the
other bars are load bars and are located related to die-
rent types whether residential, industrial and commer-
cial are shown in gure 3 (see Figure 3).
Figure 4 shows the load ow with maximum demand,
without the insertion of charging stations, where the be-
havior of the bars given in (pu) can be seen, where it goes
from a voltage limit to 1 (pu) in the 34 main nodes, so it
is observed that none exceeds the operating limit, as well
as no node has a low voltage prole, for which the ow
runs the network is optimal (see Figure 4).
Next, in table 5, the result of the load ow is obser-
ved, as well as the losses in lines and cable (see Table 5).
For the case study, it can be seen that the ow is op-
timal and there is no overload or overvoltage in any bus
and distribution lines, therefore, the losses are evaluated
and these are minimal. is is essential and it can be men-
tioned that the vehicle management problem is stochas-
tic demand and peak hours [21], a situation that would
be resolved at this point.
3.3. FLOW SCENARIO WITH INSERTION OF CHAR
GING STATIONS IN THE DISTRIBUTION NETWORK
rough the load ow at maximum demand with fully
discharged electric vehicles, the voltage behavior in the
network is analyzed since each selected node of the dis-
tribution network will have 3 charging stations each with
a power of 150Kw since it has with a fast charge that be-
nets the use of electric vehicles.
Figure 5 shows the voltage in (pu) of each bar
analyzed once the charging stations have been inser-
ted, where the network behaves eciently, although the
Table 4
Limits of the real scenario to study
Length Latitude
-79,19 -0,254
-79,14 -0,242
Figure 2
Optimal location and load points located using the Heuristic model
Figure 3
Georeferenced network to study in Cymdyst
9
Optimal georeferenced deployment of charging stations for electric vehicles in distribution networks using
a trajectory-based heuristic model
voltage prole is still within the prescribed limits, and it
already operates with an overvoltage of 9.06% in the Ne-
twork feeder (see Figure 5 and Table 6).
Aer analyzing the chargeability represented in -
gure 6, the lines with the highest index in the distribu-
tion network with approximately 35% are those that are
close to the source, as they move away the percentage
remains stable except for the nodes where they are con-
nected. the charging stations, at node 17 where it increa-
ses by 24% and goes constantly until reaching the furthest
node where there is also a 13% chargeability in the lines
and conductors where there is not a considerable percen-
tage, therefore it is not necessary to increase lines in this
section of the network. One of the most widely used indi-
ces in the operation and planning of distribution systems
was proposed by Gallego. In which it is used to nd a set
of candidate nodes to locate capacitors on a distribution
network which allows dening certain indices based on
the impact on the technical losses that they generate on
the system. e philosophy of this indicator is to nd the
nodes that will have the least impact on technical losses
by installing EV charging stations there. is is achieved
by making use of the exchange ratio between the active
power losses of the system and the reactive power injec-
tions in the nodes. Depending on the lesser impact ge
-
nerated to the technical losses of the network, the set of
possible spaces for the Charging Stations is previously
available depending on the needs of each of the agents in-
volved; that is, the electrical network and the mass trans-
port network (see Figure 6).
When analysing the losses in the lines, it is observed
that the greatest number is at the source due to the exten-
sive distance to the charging stations, therefore, it is whe
-
re there is the greatest number of losses in the nodes of
the distribution network, where they are connected. the
charging stations, as well as in the conductors, the losses
are minimal, therefore the voltage drop at maximum load
demand is 0.254%, thus being acceptable and optimal for
the operation of charging stations in the electrical distri-
bution network (see Figure 7).
e transformer loads of the two simulated cases in
Figure 7 are shown in a rst base case in the absence of
EV charging stations. In the simulation it can be seen that
the maximum load of the feeder transformer is approxi-
mately 60% in the base case. However, it increased to 70%
when charging stations were integrated into the distribu-
tion network. e same happened with the transformer
at node 17, where its load percentage increased from 57%
in the base case to 69% in the case of inserting charging
stations. It is also clear that there is no overload on any
transformer. Furthermore, when comparing the results
of the base case with the insertion of charging stations, it
Table 5
Load ow results without insertion
Simulation Results
kW kVAr k VA
Total production 449,34 97,76 459,85
Total Charges 448,64 96,13 458,82
Losses Lines 0,43 1,26 1,34
Cable Losses 0 0 0
Total Losses 0,7 3,96 4,03
Figure 4
Voltage at each Node in (pu)
Figure 5
Voltage in (pu) with insertion of VE in the network
10
Andaluz M.
can be seen that the maximum load of the transformers
is almost identical in both cases. is is because the ad
-
ditional loads resulting from EV charging shied from
peak hours.
Finally, it is estimated that in the 5-year projection
distribution network, it would necessarily be necessary to
increase transformers and distribution lines to meet the
demand in existing loads, as well as in that of electric ve-
hicles (see Figure 8).
e worst case is examined in which, in 2030, 80% of
the entire car eet is electric in the distribution network
are shown in Figure 8. In this case, the chargeability of
the transformer exceeds the maximum capacity threshold
of 150% and, since it will work above 100% of its capa
-
city, even applying smart techniques applied, or placing
power with solar panels, it should be considered to repla-
ce it with one. of larger size, likewise with the conductors,
their caliber would have to be increased as new distribu-
tion lines, likewise in the worst case there is an overload
in the nodes where the charging stations are located with
200% abnormal conditions. e voltage prole is critical
as it almost reaches the lowest level.
4. Conclusions
In relation to the modelling and charging behavior of
electric vehicles, it should be taken into account that this
developed model can be applied to any locality in real
scenarios with the consideration of the characteristics
and distance traveled, only of the national automotive
mobility system.
rough an adequate study of the characteristics of
electric vehicles, it is concluded that the optimal positions
of each charging station can be displayed, depending on
the user’s requirement, where they could recharge in 25
minutes, reducing the range of anxiety. of drivers allowing
it to obtain considerable autonomy around the city throu
-
gh the distribution network.
rough the study at the capacity level of the distri-
bution network system, and knowing both the load of the
transformers, where the increase in energy is noted from
108.4% in the base case to 121.8% in the charging station
scenario, as well Like the total losses of the simulation in
the load station insertion scenario, the impact of VE is
still evident, but even during the peak period, in the base
case it is highly tolerable in fact, the increase with respect
to the base case is around 15%, compared to the scenario
with charging stations. e voltage prole is very simi-
lar in the rst two cases, so currently this level of electric
vehicle diusion is acceptable without any intervention,
therefore, although no changes were made to the compo-
nents of the electrical system, the income is acceptable for
electric vehicle charging stations.
As a result of each operation scenario in the georefe-
renced distribution network and the application of the op-
timization model as well as the analysis of the power ow,
they are favorable especially for the operation of electric
Figure 6
Chargeability in the lines of the distribution network
Figure 7
Transformer Chargeability
Table 6
Load ow results with insertion
Simulation results
kW k VA r k VA
Total production 3634 287,65 3645,37
Total Charges 3598,55 96,14 3599,83
Losses Lines 22,29 62,49 66,35
Cable Losses 0,03 0 0,03
Total Losses 35,45 3.96 4.03
11
Optimal georeferenced deployment of charging stations for electric vehicles in distribution networks using
a trajectory-based heuristic model
vehicles, especially in public transport in the study sec-
tor, specically in a cooperative of taxis that benet from
said charging stations.
In the simulated worst-case scenario to the year 2030,
the installation of charging stations on this network in
the future is likely to create problems, particularly with
respect to transformer overloading and expected volta-
ge limits, as transformers start to operate at 150% of its
capacity, for which it will necessarily have to be replaced
since the useful life operating above 100% decreases dras-
tically, for which investments are required to change said
components, and guarantee the full availability of the high
power charging stations.
referencias
[1] Ministerio de Energía y Recursos Naturales no Renova-
bles, «Balance Energético Nacional 2017», 2019.
[2] A. Corral, F. Izurieta, V. Guayanlema, J. P. Díaz, H. Acu-
rio, and P. Quintana, «Assessment opportunities for
energy eciency in ecuadorian road transport. Opor-
tunidades de evaluación de eciencia energética en el
transporte terrestre en Ecuador», vol. 84, .o 200, pp.
309-315, 2017.
[3] J. Clairand, «Smart charging for electric vehicle aggrega-
tors considering users’», vol. 3536, .
o
, pp. 1-12, 2018,
: 10.1109/ACCESS.2018.2872725
[4] M. M. Vazifeh, H. Zhang, P. Santi, and C. Ratti, «Optimi-
zing the deployment of electric vehicle charging stations
using pervasive mobility data», Transp. Res. Part A, vol.
121, .o january 2017, pp. 75-91, 2019, : 10.1016/j.
tra.2019.01.002.
[5] Y. Fan, D. Mengnuo, G. Pei, Z. Zhang, L. Yingying, and
W. Xiao, «e inuence on distribution network plan-
ning and investment strategy of power grid corp under
electric power system reform», in 2018 China Internatio-
nal Conference on Electricity Distribution (), 2018,
pp. 2426-2430, : 10.1109/CICED.2018.8592014.
[6] J. A. Sanguesa, V. Torres-Sanz, P. Garrido, F. J. Martínez
and J. M. Márquez-Barja, «Smart cities a review on elec-
tric vehicles: technologies and challenges», pp. 372-404,
2021.
[7] E. García, «Demand response systems for integrating
energy storage batteries for residential users», 2016.
[8] W. Kempton, «Electric vehicles as a new power electric
utilities source for» vol. 2, .o 3, pp. 157-175, 1997.
[9] K. Schneider, C. Gerkensmeyer, R. Fletcher, and S. Mem-
ber, «Impact assessment of plug-in hybrid vehicles on
Pacic northwest distribution systems» pp. 1-6, 2008.
[10] D. C. Published et al., «e beginners guide to electric
vehicles ()», .o august, 2014.
[11] K. Hu, J. Wu, and T. Schwanen, «Dierences in ener-
gy consumption in electric vehicles: an exploratory re-
al-world study in Beijing», 2017.
[12] F. Badin, F. Le Berr, H. Briki, M. Petit, S. Magand, and
E. Condemine, «Evaluation of s energy consumption
inuencing factors», vol. 6, pp. 112-123, 2013.
[13] X. Qi, G. Wu, K. Boriboonsomsin, and M. J. Barth,
«Data-driven decomposition analysis and estimation
of link-level electric vehicle energy consumption under
real-world trac conditions», Transp. Res. Part D, .o
april, pp. 0-1, 2017, : 10.1016/j.trd.2017.08.008.
[14] B. W. B. Auquilla, «Estudio de estrategias para la inser-
ción de vehículos eléctricos tipo taxi en la ciudad de
Quito», 2019.
[15] G. Celli, F. Pilo, G. Monni, and G. G. Soma, «Optimal
multi-objective allocation of fast charging stations»,
2018.
[16] P. Frías and C. Mateo, «Evaluación del impacto de la in-
tegración del coche eléctrico en las redes de distribución
de energía eléctrica», pp. 56-61.
Figure 8
Transformer chargeability worst case
12
Andaluz M.
[17] M. R. Rodríguez, «Impacto de un parque vehicular eléc-
trico en los requerimientos de reconguración de la red
de distribución de la ciudad de Santiago», 2013.
[18] M. Campaña and E. Inga, «Despliegue óptimo geo-
rreferenciado de estaciones de carga vehicular pública
considerando capacidad de ujo y distancias máximas
habilitantes (Georeferenced optimal deployment for pu-
blic vehicle charge station considering restrictions of ca
-
pacity ow and maximal enablers distance)», vol. 15, .o
2, pp. 68-78, 2019.
[19] M. Campaña, «Optimal Allocation of public charging
stations based on trac density in smart cities», 2019.
[20] L. A. M. Maldonado, «Análisis técnico para la imple-
mentación de estaciones de carga rápida para vehículos
eléctricos en la provincia de Galápagos», 2020.
[21] T. Carlos and H. Rojas, «Estaciones de carga de vehícu-
los eléctricos y su inuencia en el funcionamiento de la
red de distribución», 2016.
