Article

Planning Underground Power Distribution Networks to

Minimize Negative Visual Impact in Resilient Smart Cities
Francisco Pabón 1, * , Esteban Inga 2 and Miguel Campaña 3

1 Master’s Program in Electricity, Universidad Politécnica Salesiana, Quito EC170525, Ecuador

2 Master’s Program in ICT for Education, Smart Grid Research Group (GIREI), Universidad Politécnica
Salesiana, Quito EC170525, Ecuador
3 Master’s Program in Electricity, Instituto Superior Universitario Sucre, Quito EC170148, Ecuador
* Correspondence: [email protected]; Tel.: +59-302-396-2800

Abstract: This paper presents the application of heuristic methods in conjunction with graph theory in
the optimal routing and sizing of underground distribution networks in georeferenced (GIS) scenarios,
which are modeled and simulated in the advanced engineering tool CYMDIST. The tool allows the
deployment of underground networks to facilitate the design, planning, and implementation of
networks, taking into consideration distribution company regulations, thus allowing overview and
future planning in the growth of distribution systems. Further, this method is modeled in real
georeferenced scenarios, where the coverage of the electric service to all users connected to the
network is guaranteed according to population density and energy demand while minimizing the
number of distribution transformers used. The applied method considers the location of transformer
chambers, the capacity and coverage of the distribution transformers, and the voltage drops over
the line section, which should not exceed 5% of the nominal value as described in the ANSI C84.1
standard. Consequently, to verify the efficiency of the applied method, the limitations and restrictions
of the mathematical model are considered, as well as the characteristics of the georeferenced system
and a comparison with different research studies that address the subject presented here. In addition,
Citation: Pabón, F.; Inga, E.;
supply coverage is guaranteed to be 100%.
Campaña, M. Planning Underground
Power Distribution Networks to
Keywords: CYMDIST; heuristic techniques; graph theory; optimization; georeferenced
Minimize Negative Visual Impact in
Resilient Smart Cities. Electricity 2022,
3, 463–479. https://doi.org/
10.3390/electricity3030024
1. Introduction
Academic Editor: Hugo Morais
This paper presents a heuristic model based on graph theory for optimal deployment
Received: 3 August 2022 of power distribution grids [1]. In electrical distribution system planning, wire length is
Accepted: 8 September 2022 the parameter necessary to calculate the voltage loss and the related power losses [2,3].
Published: 16 September 2022 Moreover, simulation using the advanced engineering tool CYMDIST [4,5] verifies if the
Publisher’s Note: MDPI stays neutral
deployment achieved with the algorithm complies with the observation of technical criteria
with regard to jurisdictional claims in
related to the maximum voltage loss, which for distribution systems is 5%. Thus, the algo-
published maps and institutional affil- rithm developed in MATLAB creates a medium-voltage electrical network and sizes the
iations. elements of the network in medium- and low-voltages, which is then evaluated with the
specialized software for electrical distribution systems [6].
The objective is to provide an electrical distribution system with reliability and safety
through the optimal sizing of electrical equipment in underground distribution systems [7].
Copyright: © 2022 by the authors. Moreover, installing electrical equipment is considered for constructing the underground
Licensee MDPI, Basel, Switzerland. electrical distribution network [8]. The planning problem is considered NP-complete
This article is an open access article because of the number of variables included in formulation of the problem [9].
distributed under the terms and Heuristics are used to solve problems of higher complexity. Thus, the construction
conditions of the Creative Commons of heuristics to find near-optimal solutions is essential. Every optimization problem can
Attribution (CC BY) license (https://
be solved using an already-defined heuristic method, such as genetic algorithms (GAs),
creativecommons.org/licenses/by/
tabu search (TS), ant colony optimization (ACO), and particle swarm optimization (PSO),
4.0/).

Electricity 2022, 3, 463–479. https://doi.org/10.3390/electricity3030024 https://www.mdpi.com/journal/electricity

Electricity 2022, 3 464

among others. The main feature is that graph theory is used to support the algorithm by
assigning weights based on the topology of the electrical network [10,11].
The increase in the growth of electricity systems is directly related to the increase in
energy demand [12,13]. An unpredictable increase in demand makes current grid design,
expansion, and planning of the power distribution system complex [14,15]. Therefore,
the topology of a power distribution system is considered a complex network in continuous
growth [16–18].
The optimal sizing of electrical distribution networks must consider fundamental tech-
nical aspects such as voltage loss and the location of distribution transformers, since these
are parameters to determine the safety and quality of electrical service. Therefore, the ob-
servance of technical parameters allows the construction of optimal electrical distribution
networks with a view to possible expansions in georeferenced scenarios [14,19].
An electrical network must be designed so that the consumer farthest away from
the distribution transformer has adequate voltage levels to avoid damage to equipment
connected to the electrical network. Moreover, it is essential to verify that the user closest
to the distribution transformer does not have electrical overvoltage. These are restrictions
considered in this article, since these variables are directly related to the topology of
electrical networks [20,21].
Furthermore, the proposed heuristic ensures that demand is covered at the minimum
cost and also locates the electrical equipment of the distribution network, seeking to min-
imize the technical losses that occur in the different topology of electrical distribution
networks [22]. In each scenario, the number of users connected to the same transformer
station and the population density are varied to evaluate the model’s scalability and adapt-
ability to changing demand conditions [23,24].
The method proposed in this article efficiently analyzes critical distribution systems,
providing relevant information about the georeferenced scenario. The coordinate sys-
tem information contributes to the construction of the routing of the underground net-
work. The designed electrical network must ensure safety and reliability in energy dis-
patch, ensuring voltage reductions within the parameters established in applicable regula-
tions [25–28].
Optimal sizing of buried distribution networks ensures shorter reconnection time in
case of power system failure [7,22,29], reducing power supply replacement time [12,20,30].
Moreover, worldwide, there are plans for continuous improvement in the construction of
new metropolises, especially in developed and underdeveloped countries [17,31].
Therefore, it is necessary to implement new technologies that intervene in the effi-
cient and optimal solution of daily problems [32] while generating environments with a
significantly reduced visual impact [15]. Electric power distribution networks are mainly
of overhead construction; in addition to generating a very high visual impact, they are sus-
ceptible to many failures, generating more significant interruption to power supply [7,33].
Deploying buried networks minimizes visual impact and generates reliability for
the electricity system; however, their deployment is expensive [22]. Thus, to minimize
deployment costs, the development of optimization models for deploying distribution
networks that contribute to least-cost construction is required [25,30]. These help dis-
tribution companies in decision-making for upgrading and building power distribution
grids [20,34,34].
The work presented here opens possibilities for future jobs that incorporate evaluation
of contingencies and failures, and then the expansion of the electrical distribution network
is generated. In addition, it will be possible to integrate hybrid networks that integrate
overhead and underground topologies to evaluate scenarios that occur daily in electric
utilities and to assess more-efficient management mechanisms.
This article is organized as follows. Section 2 presents related work. Section 3 presents
the traditional formulation of the problem, and Section 4 discusses the results. Finally,
Section 5 presents the conclusions.
Electricity 2022, 3 465

2. Related Works
Conventional electrical systems may experience the need to open a circuit that is part
of the leading electrical network. This frequently happens in overhead electrical systems
due to the vulnerability to natural disasters and road accidents. As a result, the electrical
system is forced to suspend the supply of electrical power to specific sections of the network,
which generates costs for energy not supplied. On the other hand, constructing subway
electrical networks is of great interest to ensure continuity and service, guaranteeing energy
security. Figure 1 shows the requirements to be considered in subway electrical networks.

MOTIVATION Planning Underground Electrical Minimise the Cost

Distribution Networks of Resources

METHOD Sizing - Placement Industrial Network

Electrical
Distribution
Transformer
Electrical
Transformer
Room Commercial Network
Types
Consumers Underground
Underground Network
Network

Electric
Distribution Residential Network
Distribution
Subestation Transformer

ANALYSIS
Capacity - Coverage - NP-Complete
Flows Georeferenced
Minimun Spanning Tree Problem Relaxed

Figure 1. Underground electrical distribution network.

Undoubtedly, the dimensioning of electrical equipment in underground distribution

systems increases safety and continuity of service. In addition, this type of underground
construction allows the inclusion of generation systems with renewable resources, directly
increasing the resilience of the electrical distribution network.
For the deployment and optimal dimensioning of subway electrical distribution net-
works, it is necessary to implement heuristic models capable of observing restrictions such
as voltage drop and technical losses [35]. Consequently, the restrictions have a direct rela-
tionship with the cost variable, since the equipment required by the electrical network is to
be dimensioned with optimization criteria. The proposed methodology is based on graph
theory to analyze the deployment and simulation of electrical networks at minimum cost.
Another fundamental characteristic is that it considers the stratification of users according
to the energy consumed per time interval. Finally, the proposed mathematical model is
validated through simulation in CYMDIST, since this software receives georeferenced data
from MATLAB and interprets them to execute load flows of the optimal topology obtained
from the planning model.
Subway distribution networks offer benefits that improve the reliability index of the
electrical system since they improve the protection indexes against failures [36]. Then, these
are implemented in strategic sectors where an aerial network is not feasible due to security
problems in the energy supply and reduced general attractiveness due to the negative
visual impact it generates. The construction of a subway power grid involves significant
investments due to construction requirements, such as civil works, walls, and foundations
for the location of the necessary electrical equipment, such as transformer chambers [37,38].
Electricity 2022, 3 466

On the other hand, the power quality factor reduces maintenance costs by correctly
using the nominal capacity installed to transport electrical power from generation to the
final load (users) [39]. Consequently, it is essential to provide the necessary resources to
ensure that the deployment of an underground electrical distribution network guaran-
tees operational safety and continuity of electrical service. This is achieved with robust
protection systems and adequate insulation levels [40–42].
The proposed optimization algorithm’s efficiency directly depends on observance of
the constraints to reduce the cost of the objective function. An exhaustive search is executed,
and the machine time to solve the problem depends on each case study’s initial constraints.
In addition, the model allows cost reduction through the optimal sizing of the electrical
equipment of the distribution network, taking into consideration the optimal location of the
transformation chambers and the minimum required for several distribution transformers
to be located in a georeferenced area. Finally, the resulting topology of the electrical system
is validated by simulation with the specialized distribution network software CYMDIST.
Table 1 shows a summary of works related to deploying subway power distribution
grids and, in addition, the importance and novelty of this article.

Table 1. Summary of related works.

Parameters Considered Thematic

Author, Year Objectives
Voltage Capacity Cost Georeferenced Planning Resilience Graph Theory
Ciechanowicz, 2016 [43] Electrical network evaluation and modeling X X X - X X -
Li, 2016 [10] Reduces active power losses X - - - X - -
Sowmya, 2016 [44] Use of geographic data in power grids - - - X X - -
Roshanagh, 2016 [45] Planning new electrical network at minimum cost X X X - X - X
Xie, 2018 [46] Distribution network planning X X X - X - -
Pinzón, 2020 [47] Location of substations - X - X X - -
Cresta, 2021 [48] Management of electrical distribution networks X X - X - X X
Ayalew, 2022 [49] Expansion of the power grid with distributed generation X X X - X - -
Kostelac, 2022 [50] Planning and operational issues of microgrids X - X - X - -
Present work Planning and sizing of electrical distribution network X X X X X X X

Regarding works presented with traditional non-scalable models and pre-established

parameters, the present work innovates with a model capable of evaluating the reality of an
urban, suburban, and/or rural area where it is desired to deploy an underground network.
The simulation process of the topology achieved warns of the possibility of becoming
a suitable tool for consulting work evaluation of contingencies of previously deployed
networks that require expansion from the existing network. The current work does not
consider pre-existing overhead networks in the study area, but these will be incorporated
in future work by generating a layered solution and integrating them into the electrical
system to evaluate efficiency and reliability.

3. Problem Formulation
The design of electrical distribution networks is performed considering different types
of consumers, which are associated with the energy consumed over some time. This
consideration is necessary to decide if the existing electrical network can satisfy existing
and projected demand. End-users may be connected to the primary or secondary electrical
networks as long as the additional energy extracted from the distribution system guarantees
continuity of service and quality conditions for the end-user.
The topology of the distribution network can be radial or ring. The predominant
topology in distribution systems is radial, which implies having complex connection points
at low investment costs compared to ring systems. Thus, to guarantee energy security and
quality criteria, the timely selection of the electrical distribution network’s topology and
the electrical equipment’s optimal sizing are required.
The sizing of transformers is a fundamental part of the design, since the correct design
of this equipment avoids possible problems related to losses of energy not supplied due to
power outages. These operating conditions negatively impact energy security and raise
costs due to technical losses and unsupplied power.
Electricity 2022, 3 467

Furthermore, to select transformers in residential areas, it is essential to know the

manufacturer’s technical datasheet, the type of transformer, and the rated service power.
The standard power ratings for residential transformers are 50, 75, 100, 125, and 150 kVA.
Computation of design demand is classified according to six types of consumers, as
indicated in Table 2. This classification is directly related to the monthly consumption
level for each type of user. For the case study, the sizing of a subway network of type B is
considered, a typical characteristic of a residential system.

Table 2. Strata of electricity consumption.

Strata Rate of Consumption by Customer (kWh/Month/Customer)

E 0–100
D 101–150
C 151–250
B 251–350
A 351–500
A1 501–900

The optimal planning of a subway network depends on several criteria and informa-
tion, such as the type of user and the location of the service. This information facilitates the
categorization of the service; in the case of a residential load, it can be assumed that the
area will be urban or rural, while if the load is commercial or industrial, it is assumed that
the area is linked to work areas with high and medium impact loads for the system [51].
Consequently, to design the electrical network, it is essential to use the diversified
maximum demand (DMD). DMD refers to the maximum load projected to be present on
the distribution system. Equation (1) indicates the mathematical relationship to be used.
The variable M is related to the coincidence factor, and the variable N to the consumption
factor. The consumption factor is considered according to the stratum of each user and
represents the maximum projected amount of energy consumed by the customer per
month. The values of M and N are specified in the local rules and regulations of the
distribution companies.

DMD = ( M ∗ N ) (1)
A timely power supply study considering demand forecasting ensures electricity
supply, reducing unsupplied power rates. Undoubtedly, this allows technological growth
to guarantee energy supply and reduce investment costs. Consequently, demand growth
analysis is fundamental to designing a distribution system’s expansion and its connection
to the end-user.
The calculation of the design demand (DD) follows the expression of Equation (2).
MDD is the maximum design demand, PLD is the street lighting demand, RTL is the
resistive technical loss (RTL) demand with a value of 3.6% of the MDD value, and PF is the
power factor.

MDD + PLD + RTL

DD = (2)
PF
Planning for the construction of a subway electrical network considers, as a starting
point, the position of the electrical substation and the laying of the primary network
conductor. In radial electrical networks and for the case study, a voltage level of 13.8
kV is used. Another detail to consider in the present article is that the construction of
underground electrical networks is proposed, for which transformation chambers and
appropriate equipment must be considered to mobilize the construction of the underground
electrical network.
Electricity 2022, 3 468

Subway networks have several types of disconnectors and feeders that facilitate the
maintenance of this type of system that, due to their construction, have greater resilience
than overhead distribution lines. Adequate dimensioning of subway electrical distribution
network elements guarantees the system’s operation, avoiding overloads in the different
branches of circuits and transformers. In addition, load imbalances and technical losses
must be minimized to increase reliability and energy security.
It should be noted that subway networks are challenging to access for maintenance.
The sizing of system protections and correct identification of conductors facilitate the recon-
nection of service in case of failure. Consequently, the reliability of the installation increases.
In the design of a subway network, it is not enough to know the georeferenced location
or the types of loads in the system. It is necessary to size transformers, transformer cham-
bers, protection, electrical conductors, and accessories required to assemble the electrical
distribution network.
Another detail is that the voltage level at which the end-users are reached must be
verified. The maximum voltage levels allowed, depending on the electrical network, are
described in the literature. In addition, any subway electrical installation must be planned
and foresee expansion options for an eventual increase in demand [32]. Table 3 describes
the MDD according to the type of residential user for a subway network, providing the
relationship between the maximum diversified demand and the rated load (RL).

Table 3. Simulation model parameters.

Residential User MDD (kW) MDD (kVA) RL (kW) RL (kVA)

B 3.4 3.58 9.8 10.31
A 4.7 4.95 15.82 10.65

In order to plan an underground distribution network, it is essential to rely on maps

and geolocalized coordinates. Once this information is known, it is possible to foresee
optimal locations to reduce conductor laying distances. This contributes significantly to
reducing voltage drops and the increase of Joule losses.
For the correct application of subway network equipment, it is essential to find a
tool that calculates the location of such equipment and allows deciding on the variable
length and the maximum installed powers. Consequently, the objective is to locate the
transformer substations according to the capacity of the network and transformer units.
The capacities of the grid and transformation units are designed according to the demand
of users connected to the distribution grid.
In this article, OpenStreetMap is used to extract coordinate information in latitude and
longitude of the road structure and land distribution of the residential area. In other words,
the algorithm knows the topology of the zone or area to intervene in the planning process of
underground electrical networks. distribution transformer locations are determined based
on the centers of gravity of the loads, which are determined by the demand (Si ) of each
system user (i). According to the parameters related to longitude and latitude ( Loi , Lai ),
shown in Equations (3) and (4), it should be noted that transformers will be located along
the main streets below the sidewalks.

∑in=1 ( Loi ∗ Si )
Lodt = (3)
∑in=1 (Si )

∑in=1 ( Lai ∗ Si )
Ladt = (4)
∑in=1 (Si )
Electrical distribution systems represent complex structures due to their number
of connections according to topology (radial–ring); therefore, finding the optimal route
for subway distribution infrastructure requires combinatorial algorithms of NP-complete
complexity. The algorithm must verify constraints specific to the optimization problem,
Electricity 2022, 3 469

such as: minimizing investment and operation costs, reducing power lines, and meeting
quality-of-service parameters. An additional detail is that the proposed heuristic finds the
local optimum to the deployment problem of the electrical distribution network and relies
on graph theory through definition of the adjacency matrix. The adjacency matrix contains
the vertices and edges of the area to be intervened in during the planning process.
The heuristic model proposed in this paper helps find a close-to-optimal answer, where
the number and power of the transformers to be placed are determined. In addition, it
reduces the voltage drop from transformers to the farthest users of the distribution network.
The minimum path required for a radial topology is used to achieve the minimum spanning
tree (MST) algorithm. In an actual case study, the topology knowledge allows the designer
to evaluate the type of conductor used to construct the subway electrical network.
Graph theory is a combinatorial tool used to determine the relationships between
nodes and edges using geolocated coordinates of transformers, intersections, and end-users.
The applied model ensures that voltage drops are reduced by the optimal routing of the
underground power line, optimal transformer location, and optimal conductor selection
based on current and projected power demand. On the other hand, voltage sags in electrical
distribution systems represent a problem that affects the quality of electrical power supply
to the consumer. Voltage drops are directly related to the length and dimensioning of the
conductor, the impedance, and the service voltage level. Therefore, total voltage drop is
calculated by summing the voltage drops in each branch.
For the optimal sizing of distribution transformers, the mathematical model uses the
modified Prim algorithm, which requires a graph G = (V, E) of order n and size m based
on a set of vertices (V ) and edges ( E), taking into consideration the number of customers
connected to the same transformer, and complying with the regulation that voltage drops
must not exceed 5%. The distance calculation is made by applying the Haversine equation,
which relates the curvature and radius of the earth in terms of latitude and longitude
coordinates (Equation (5); results are given in km).
r    
4lat 4lot
D = 2 ∗ R ∗ arcsin sin2 2 + cos(lat1) ∗ cos(lat2) + sin2 2 (5)

where:
• D is the distance in km from a point i to a point j;
• lat1, lon2 are the coordinates of point 2;
• 4lat is the difference between the latitude coordinates of point 1 and point 2;
• 4lon is the difference between the longitude coordinates of point 1 and point 2;
• R is the earth’s radius, with a value of 6372.8 km.
Equation (6) represents the objective function and minimizes the total cost of the
distribution network. The capacity constraints of the distribution substation, the existence
of georeferenced electrical elements distributed in the area, the reduction of electrical
conductor length for the underground network, the sizing of the distribution transformer,
and the computation of voltage drops are represented in Equations (7)–(11), respectively.

OF = min∑ Cs + ∑ Capc + ∑ Cti + ∑ Vd (6)

Subject to:
n
Cs = ∑ (Cpi ∗ C f ) (7)
i =1

Capc = ∑ (z 6 m, ∀z, m<) (8)

z∈<

Rc = ∑ (z 6 r, ∀z, r<) (9)

z∈<
Electricity 2022, 3 470

α
Cti = ∑ (Capc ) (10)
i =1
n
2
Vd = ∗ ∑ ( Pi , j ∗ Lgi , j ) (11)
γ20 UnSc i,j=1

A summary of the variables used in this article is presented in Table 4.

Table 4. Variables related to the constraints.

Symbol Description Unit

Cs Restricts the capacity of the substation. R+
Cpi Partial capacity of residential user. kVA
n Number of nodes R+
Cf Multiplication factor projection future growth. %
Capc Limit of the capacity of each group. R+
m Vector length in the case study. R+
z Verification variable. R+
Rc Restricts the maximum allowable distance. Meters
r Maximum allowed distance. Meters
Cti Restricts the maximum load of the transformers. kVA
α Number of clusters. R+
Vd Voltage drop. V
γ2 0 Conductivity according to the type of conductor. Siemens
Un Applied voltage of the conductor. V
Sc Underground conductor section. mm2
Lg Length of the conductor from the transformer to the user. Meters
P Power losses. kW
i, j Source and target, respectively. Index

Figure 2 shows the flowchart of subway power grid deployment. The flowchart shows
the methodology used by the underground grid planning algorithm.
The algorithm starts with knowledge of the topology of the scenario for which the
underground grid deployment is being executed. Then, the revision boxes are located,
observing the regulations, which require a revision box every 40 m. Once the position of
the revision boxes is known, the algorithm to determine the minimum spanning tree of the
underground electrical network is executed. In addition, a check is performed to reduce
the cost of the minimum spanning tree; if the answer is yes, the resulting MST is selected as
the local answer.
The next step is to locate the distribution transformers. It is necessary to mention
that the applied model is scalable. Therefore, its application in different cases of study is
analyzed, determining the number of users connected to a distribution transformer; in
addition, the length of the conductor of the low voltage electrical network is determined.
Consequently, a multilevel multigraph is constructed and explained in detail in Figure 3.
Finally, the flowchart shows the need to lay out the distribution network observing that
the location of the distribution transformers complies with the regulatory standards in
the literature.
Electricity 2022, 3 471

Input
(Georeferenced case study)

Underground infrastructure
(Location of inspection boxes)

Electrical distribution network

(Routing the underground network)

no There is
a cost
reduced?

yes
Output
(Short test path tree)

Electrical distribution transformer

(Geographical allocation)

Output
(Latitude and longitude coordinates)

Electrical study
(Voltage drop calculation)

no Regulatory
standards
observed?

yes
Output
(solution)

Figure 2. Underground power grid deployment flowchart.

The heuristic model for the optimal deployment of subway networks has been applied
in an actual scenario, i.e., georeferenced. The case study is located in the East Dulwich
region in southeast London, England, in the London Borough of Southwark, located
at the coordinates of longitude −0.0764 to −0.0725 and latitude 51.4555 to 51.452 with
an approximate area of 84 square kilometers. The scenario comprises 715 loads with a
power demand of 4.9 MW between residential and commercial users. The power demand
estimation is based on each type of user’s average monthly energy consumption. The K-
medoids algorithm divides the scenario into six clusters, as shown in Figure 3. The location
of the load centers, number of users, and routes that mark the path to be followed by the
subway electrical conductor are verified. Using this information, the heuristic determines
how to connect all users with minimum cost and the shortest conductor length.
Electricity 2022, 3 472

(a) Scenario selection

(b) Spatial grid modeling

Power consumption
stratification

(c) Trajectory data mapping

Transformer
Substation
Well electrical
Buried electrical wire

(d) Electrical equipment

Figure 3. Layered evaluation of the proposed Methodology and Algorithm.

The heuristic is also developed using MATLAB software to achieve the stated ob-
jectives. The resulting electrical distribution system is then simulated with CYMDIST
software-Version 4.7 to obtain accurate load flow data and to validate the performance of
the heuristic.
The electrical equipment is selected based on technical requirements such as primary
and secondary network voltage and design demand. The distribution transformers are
oil-immersed type, and the voltage level of the medium-voltage network is 11 kV, while
the voltage level of the low-voltage network is 0.22 kV.
The simulation parameters are detailed in Table 5. The primary feeder follows the
typology proposed in the connectivity or adjacency matrix obtained as a near-optimal
solution. Additional detail is that the heuristic achieves 100% coverage for electricity
supply to users. Demand depends on a selected cluster and the number of users, so the
demand is identified for each cluster. The location of the substation is arbitrary.
On the other hand, the proposed method applies the Prim algorithm to determine the
optimal number of transformers to be placed, identifying the characteristics of the streets
of the scenario and a stretch from the transformer greater than 40 m in conjunction with the
calculation of the connectivity matrix, which is responsible for determining the shortest
distance between the houses, the transformers, and the substation. Table 6 illustrates initial
considerations for running the proposed optimization model.
Electricity 2022, 3 473

Table 5. Characteristics of the final scenario for analysis of the method.

Item Parameters Description

Final user 715
Residential users f.p. 0.95
Commercial users f.p. 0.85
Scenario
Total demand variable (4.2 MW)
Connections associated
15, 20, 25, 30
with each transformer
Primary feeder 1
Voltage level 11 kV
Medium-voltage network Type of installation Underground
Network settings Radial
Insulated power cord
Type of conductor
TCLPE 15 kV
Distribution transformer 11/0.22 kV immersed in oil
Voltage level 220 V
Low voltage network Network settings Radial
Insulated power cord
Type of conductor
TCLPE 2 kV

Table 6. Simulation model parameters.

Scenarios Transformer Demand (kW) Transformer Power without Tap (kVA)

Scenario A 124 150
Scenario B 106 112.5
Scenario C 88 112.5
Scenario D 71 75

4. Analysis of Results
For deploying the subway distribution network, conductor routing has been consid-
ered using the georeferenced scenario database with the osm file extension. The heuristic
model has been built in MATLAB software, while the power flow study has been analyzed
with CYMDIST software. Furthermore, it is noted that the heuristic has been evaluated
on a balanced power network. In order to verify the versatility of the proposed heuristic
and its easy adaptability, four case studies are carried out. The close-to-optimal solution
for each case study is shown in Table 7, where the length of the primary feeder conduc-
tor, the optimal number of transformers, the maximum length of the conductor in the
medium-voltage power network, and the percentage of voltage drop are shown. A detail
that can be appreciated with Table 7 is that as the distance increases, so does the voltage
drop; consequently, it can be inferred that the voltage drop is directly proportional to the
length. Furthermore, the number of transformers required in the underground distribution
network increases as the length of the topology increases. Topology length is defined as the
length of the road network.
Consequently, it is verified that the proposed heuristic model is scalable, i.e., it can
be applied to any case study. A maximum capacity of links in each transformer has been
determined for each scenario.
The heuristic algorithm can estimate the optimal number of transformers to be placed
in each case study to guarantee 100% coverage. The optimal number of transformers is
directly related to the design demand of each user. Knowing the optimum number of
transformers, the demand for street lighting, and the demand for technical losses allows for
determining the minimum power required at the power transformer to supply electricity
while observing safety and service continuity criteria.
Electricity 2022, 3 474

Table 7. Near-optimal solution for each case study.

Description Scenario A Scenario B Scenario C Scenario D

Length from source (m) 490 495 500 610
No. of transformers 35 40 49 60
Network length MV (km) 30 43 55 58
Voltage drop 2.1% 2.4% 2.7% 3%

Figure 4 shows the algorithm’s results applied to the four base scenarios, where each
base scenario must cover the power demand of the entire scenario, estimated at 5.4 MVA.
The model suggests that Case Study D requires the highest number of transformers, reach-
ing 60 with the sizing power of 79 kVA. On the other hand, Case Study A requires a
maximum number of 35 transformers of 137 kVA to cover demand. As explained in pre-
vious paragraphs, the solutions responded to local optimums since they depend on the
initial conditions of the problem to be solved. Another detail revealed in Figure 4 is that as
the length increases, it is necessary to place more transformers.

60
Scenario A
Scenario B
Scenario C
50
Scenario D

40
Number of Transformers

30

20

10

0
0 100 200 300 400 500 600 700
Number of Sources (Customers)

Figure 4. Number of transformers required related to the distance from the source to the load.

Figure 5 shows the maximum conductor length required for each case study. Length
is recorded from the main feeder and the total number of supplies for the 715 loads in the
scenario. The results show a minimum required length of 58 km of conductor for the MV
underground grid assembly for Case Study D, with a maximum capacity of 15 subscribers
per transformer. Scenario A has a minimum length of 30 km of underground MV network
with a maximum capacity of 30 subscribers per transformer.
As previously mentioned, the electrical distribution system has been modeled with
CYMDIST software. This allows different studies such as load flow, voltage drops, and power
profile of the electrical system, among others. Figure 6 illustrates the implementation of
the case study A distribution system. The blue lines indicate the underground electrical
network and its topology. The primary feeder, 35 transformers with delta-grounded star
connection, and the number of concentrated loads deployed in the area of interest can
be identified.
Electricity 2022, 3 475

60
Max-Cap 15
Max-Cap 20
Max-Cap 25
50
Max-Cap 30

Electrical conductor length (km)

40

30

20

10

0
0 100 200 300 400 500 600 700 800
Number of Sources (Customers)

Figure 5. Length of the conductor related to the distance from the source to the load.

S/E

Concentrated load Buried electrical wire Transformer

Figure 6. Medium- voltage network of scenario A.

An electrical system in optimal operation is established through reliability and power

supply quality indicators. These can increase due to problems in the system’s operation
due to different abnormal operating conditions; these conditions are considered voltage
drops due to technical losses in the MV electrical network.
It is essential to verify the maximum voltage drop from the transformer to the farthest
load, since compliance with the maximum acceptable 5% for voltage variation in the
transformer secondaries must be guaranteed. Figure 7 shows the voltage drop in each
study case. From the graph in Figure 7, it can be identified that the distance with the
highest voltage drop is 120 m from the main feeder; therefore, Scenario D shows a voltage
drop of approximately 4.1%, which is the highest voltage drop of the different case studies.
In addition, Case Study A shows a voltage drop of approximately 1.55%. Consequently,
Electricity 2022, 3 476

when analyzing the voltage drops in each case study. The voltage drops are within the
acceptable limits in the literature and do not exceed 5.5%.
200 250

Scenario A Scenario B
200

Voltage drop (V)

Voltage drop (V)

150

150

100

100

50
50

0 0
119.9985 119.999 119.9995 120 119.984119.986119.988 119.99 119.992 119.994119.996119.998 120
Distance from the S/S (m) Distance from the S/S (m)

400 500

350 Scenario C Scenario D

450
Voltage drop (V)

Voltage drop (V)

300

250
400

200

350
150

100
300
50

0 250
119.9988 119.999 119.9992 119.9994 119.9996 119.9998 120 119.966 119.967 119.968 119.969 119.97
Distance from the S/S (m) Distance from the S/S (m)

Figure 7. Stress drop in study scenarios.

Figure 8 shows the apparent power profile for each case study. The apparent power
profile is recorded from the power source to the user. It can be identified from Figure 8 that
as the distance increases, the apparent power profile decreases.
5 8

4.5 Scenario A Scenario B

7
Power profile (KVA)

Power profile (KVA)

4
6
3.5
5
3
4
2.5
3
2

1.5 2

1 1
500 550 600 650 700 750 800 850 0 10 20 30 40 50 60 70 80 90
Distance from the S/S (m) Distance from the S/S (m)

7 4

Scenario C Scenario D
6 3.5
Power profile (KVA)

Power profile (KVA)

5 3

4 2.5

3 2

2 1.5

1 1
550 600 650 700 750 800 185 190 195 200 205 210 215 220
Distance from the S/S (m) Distance from the S/S (m)

Figure 8. Power profile in (kVA) of the study scenarios.

Electricity 2022, 3 477

5. Conclusions
The model proposed in this paper shows that a close-to-optimal response is provided
for each case study. It demonstrates the scalability of the model and its easily adaptability
to any case of study. In addition, it is possible to verify the maximum conductor lengths
required in underground electrical networks. Knowing the power to be installed in dis-
tribution transformers, it is possible to plan the dimensioning of conductors, protection,
and civil works. Consequently, the proposed heuristic helps designers make decisions by
observing technical criteria with reduced costs.
The heuristic method has allowed the optimal deployment of the subway network,
considering reduced investment costs. The significant contribution of this article is that
it allows electric distribution companies to make expansion plans and project demand by
generating different case studies.
Electricity supply is susceptible to interruptions or failures, especially in overhead
networks, which causes economic losses. Thus, a robust algorithm for subway distribution
networks has been proposed to increase the security and reliability of the electricity supply
from the distribution system to the end-user.
The efficiency of the proposed mathematical model has been verified through a georef-
erenced system. By analyzing load flows for each case study, we verified that the electrical
network does not exceed 5% voltage drop from the secondary of the transformers to the
users farthest from the source. In future research, the proposed model will integrate
distributed generation.
The work presented opens possibilities for future works that incorporate evaluation of
contingencies and failures, and then the expansion of the electrical distribution network is
generated. In addition, it will be possible to incorporate hybrid networks that contemplate
overhead and underground topologies to evaluate scenarios that occur daily in electric
utilities and thus evaluate more-efficient management mechanisms.

Author Contributions: F.P.: Conceptualization, Methodology, Validation, Writing—review and

editing. E.I.: Conceptualization, Methodology, Software, Writing—original draft. F.P. and E.I.: Data
curation, Formal analysis. F.P.: Supervision. E.I.: Writing—review and editing. M.C.: Data curation,
Writing—review. All authors have read and agreed to the published version of the manuscript.
Funding: Universidad Politécnica Salesiana and GIREI supported this work as part of the Smart Grid
Research Group under the project Electric Vehicle Charging in Buildings and its Impact on the Sizing
and Planning of Electrical Distribution Networks.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Acknowledgments: This work was financially funded by the Smart Grid Research Group (GIREI)
of Universidad Politécnica Salesiana and Power Grids and Smart Cities (RECI), Salesian Institutions
of Higher Education (IUS).
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Zhang, B.; Lam, A.Y.; Domínguez-García, A.D.; Tse, D. An Optimal and Distributed Method for Voltage Regulation in Power
Distribution Systems. IEEE Trans. Power Syst. 2015, 30, 3226–3228. [CrossRef]
2. Villacres, F.; Inga, E. Planeación y dimensionamiento de redes eléctricas de distribución soterrada mediante un método
metaheurístico. Ing. Cienc. 2019, 15, 141–166. [CrossRef]
3. Herrera, E. Planeación óptima de redes de distribución eléctrica aérea usando métodos heurísticos y procesos de simulación.
Trab. Titulación 2018, 1, 37.
4. Tang, Y.; Mao, X.; Ayyanar, R. Distribution system modeling using CYMDIST for study of high penetration of distributed
solar photovoltaics. In Proceedings of the 2012 North American Power Symposium, NAPS 2012, Champaign, IL, USA,
9–11 September 2012. [CrossRef]
Electricity 2022, 3 478

5. Kumawat, M.; Gupta, N.; Jain, N.; Bansal, R.C. Swarm-Intelligence-Based Optimal Planning of Distributed Generators in
Distribution Network for Minimizing Energy Loss. Electr. Power Compon. Syst. 2017, 45, 589–600. [CrossRef]
6. Mortensen, L.K.; Shaker, H.R.; Veje, C.T. Relative fault vulnerability prediction for energy distribution networks. Appl. Energy
2022, 322, 119449. [CrossRef]
7. Haakana, J.; Kaipia, T.; Lassila, J.; Partanen, J. Reserve power arrangements in rural area underground cable networks. IEEE
Trans. Power Deliv. 2014, 29, 589–597. [CrossRef]
8. Hamouda, A.; Lakehal, N.; Zehar, K. Heuristic method for reactive energy management in distribution feeders. Energy Convers.
Manag. 2010, 51, 518–523. [CrossRef]
9. Sharafat, A.; Khan, M.S.; Latif, K.; Tanoli, W.A.; Park, W.; Seo, J. Bim-gis-based integrated framework for underground utility
management system for earthwork operations. Appl. Sci. 2021, 11, 5721. [CrossRef]
10. Li, H.; Mao, W.; Zhang, A.; Li, C. An improved distribution network reconfiguration method based on minimum spanning tree
algorithm and heuristic rules. Int. J. Electr. Power Energy Syst. 2016, 82, 466–473. [CrossRef]
11. Valenzuela, A.; Inga, E.; Simani, S. Planning of a resilient underground distribution network using georeferenced data. Energies
2019, 12, 644. [CrossRef]
12. Yagoub, M.M. Assessment of OpenStreetMap (OSM) Data: The Case of Abu Dhabi City, United Arab Emirates. J. Map Geogr. Libr.
2017, 13, 300–319. [CrossRef]
13. Nazari-heris, M.; Mohammadi-ivatloo, B.; Gharehpetian, G.B. A comprehensive review of heuristic optimization algorithms for
optimal combined heat and power dispatch from economic and environmental perspectives. Renew. Sustain. Energy Rev. 2017,
81, 2128–2143. [CrossRef]
14. Yang, N.C.; Chen, H.C. Decomposed Newton algorithm-based three-phase power-flow for unbalanced radial distribution
networks with distributed energy resources and electric vehicle demands. Int. J. Electr. Power Energy Syst. 2018, 96, 473–483.
[CrossRef]
15. Pavón, W.; Inga, E. Optimal Routing an Ungrounded Electrical Distribution System Based on Heuristic Method with Micro Grids
Integration. Sustainability 2019, 11, 1607. [CrossRef]
16. Kawamura, K.; Ametani, A.; Gudmundsdottir, U.S. Surge Analysis on a Long Underground Cable System. J. Int. Counc. Electr.
Eng. 2013, 3, 158–163. [CrossRef]
17. Salehi, J.; Haghifam, M.R. Long term distribution network planning considering urbanity uncertainties. Int. J. Electr. Power
Energy Syst. 2012, 42, 321–333. [CrossRef]
18. Ramesh, L.; Ravindiran, S.; Chowdhury, S.P.; Chowdhury, S.; Song, Y.H.; Goswami, P.K. Distribution system loss minimization
and planning using cymdist. In Proceedings of the 2007 42nd International Universities Power Engineering Conference, Brighton,
UK, 4–6 September 2007; Volume 1, pp. 316–321. [CrossRef]
19. Bosisio, A.; Berizzi, A.; Merlo, M.; Morotti, A.; Iannarelli, G. A GIS-Based Approach for Primary Substations Siting and Timing
Based on Voronoi Diagram and Particle Swarm Optimization Method. Appl. Sci. 2022, 12, 6008. [CrossRef]
20. Valenzuela, A.; Montalvo, I.; Inga, E. A decision-making tool for electric distribution network planning based on heuristics and
georeferenced data. Energies 2019, 12, 4065. [CrossRef]
21. Hamza, M.H.; Chmit, M. GIS-Based Planning and Web/3D Web GIS Applications for the Analysis and Management of MV/LV
Electrical Networks (A Case Study in Tunisia). Appl. Sci. 2022, 12, 2554. [CrossRef]
22. Rawat, T.; Niazi, K.R.; Gupta, N.; Sharma, S. Impact analysis of demand response on optimal allocation of wind and solar based
distributed generations in distribution system. Energy Sources Part B Econ. Plan. Policy 2020, 16, 75–90. [CrossRef]
23. Haghifam, M.R.; Olamaei, J.; Andervazh, M.R. Adaptive multi-objective distribution network reconfiguration using multi-
objective discrete particles swarm optimisation algorithm and graph theory. IET Gener. Transm. Distrib. 2013, 7, 1367–1382.
[CrossRef]
24. Kisuule, M.; Hernando-Gil, I.; Serugunda, J.; Namaganda-Kiyimba, J.; Ndawula, M.B. Stochastic planning and operational
constraint assessment of system-customer power supply risks in electricity distribution networks. Sustainability 2021, 13, 9579.
[CrossRef]
25. Mahmoud, I.; Kamel, S.; Abdel-Mawgoud, H.; Nasrat, L.; Jurado, F. Integration of DG and Capacitor in Radial Distribution
Networks Using an Efficient Hybrid Optimization Method. Electr. Power Compon. Syst. 2020, 48, 1102–1110. [CrossRef]
26. Mateo, C.; Prettico, G.; Gómez, T.; Cossent, R.; Gangale, F.; Frías, P.; Fulli, G. European representative electricity distribution
networks. Int. J. Electr. Power Energy Syst. 2018, 99, 273–280. [CrossRef]
27. Yu, D.C.; Liu, H.; Chiang, H.D. A Heuristic Meter Placement Method for Load Estimation. IEEE Power Eng. Rev. 2002, 22, 59.
[CrossRef]
28. Peralta, J.A.; de León, F.; Mahseredjian, J. Unbalanced multiphase load-flow using a positive-sequence load-flow program. IEEE
Trans. Power Syst. 2008, 23, 469–476. [CrossRef]
29. Gouda, O.E.; Osman, G.F.; Salem, W.A.; Arafa, S.H. Load cycling of underground distribution cables including thermal soil
resistivity variation with soil temperature and moisture content. IET Gener. Transm. Distrib. 2018, 12, 4125–4133. [CrossRef]
30. Krey, S.; Brato, S.; Ligges, U.; Götze, J.; Weihs, C. Clustering of electrical transmission systems based on network topology and
stability. J. Stat. Comput. Simul. 2015, 85, 47–61. [CrossRef]
31. Alvarez-herault, M.C.; Dib, J.P.; Ionescu, O.; Raison, B. Long-Term Planning of Electrical Distribution Grids: How Load. Energies
2022, 15, 6084. [CrossRef]
Electricity 2022, 3 479

32. Cunha, V.C.; Mantovani, J.R.S. Planning and Project of Medium Voltage Electric Power Distribution Systems. IEEE Lat. Am.
Trans. 2016, 14, 2298–2308. [CrossRef]
33. Adinolfi, G.; Ciavarella, R.; Graditi, G.; Ricca, A.; Valenti, M. A planning tool for reliability assessment of overhead distribution
lines in hybrid AC/DC grids. Sustainability 2021, 13, 6099. [CrossRef]
34. Park, B.; Myung, H. Resilient underground localization using magnetic field anomalies for drilling environment. IEEE Trans. Ind.
Electron. 2017, 65, 1377–1387. [CrossRef]
35. Garavaso, P.; Bignucolo, F.; Vivian, J.; Alessio, G.; Carli, M.D. Optimal Planning and Operation of a Residential Energy. Energies
2021, 14, 2045. [CrossRef]
36. Picard, J.L.; Aguado, I.; Cobos, N.G.; Fuster-roig, V. Electric Distribution System Planning Methodology Considering Distributed
Energy Resources: A Contribution towards Real Smart Grid Deployment. Energies 2021, 14, 1924. [CrossRef]
37. Liu, J.; Cheng, H.; Xu, Q.; Lan, Z.; Zeng, P.; Yao, L. Stochastic expansion planning of interconnected distribution networks with
renewable sources considering uncertainties and power transfer capability. J. Eng. 2017, 2017, 1600–1604. [CrossRef]
38. Chang, C.K.; Cheng, S.T.; Boyanapalli, B.K. Three-Phase Unbalance Improvement for Distribution Systems Based on the Particle
Swarm Current Injection Algorithm. Energies 2022, 15, 3460. [CrossRef]
39. Schwaeppe, H.; Böttcher, L.; Schumann, K.; Hein, L.; Hälsig, P.; Thams, S.; Lozano, P.B.; Moser, A. Analyzing Intersectoral Benefits
of District Heating in an Integrated Generation and Transmission Expansion Planning Model. Energies 2022, 15, 2314. [CrossRef]
40. Kornatka, M.; Gawlak, A. An Analysis of the Operation of Distribution Networks Using Kernel Density Estimators. Energies
2021, 14, 6984. [CrossRef]
41. Octon, P.; Pobedza, J.; Walczak, P.; Cisek, P.; Vallati, A. Experimental Validation of a Heat Transfer Model in Underground Power
Cable Systems. Energies 2020, 13, 1747.
42. Charan, T.; Mackey, C.; Irani, A.; Polly, B.; Ray, S.; Fleming, K.; Kontar, R.E.; Moore, N.; Elgindy, T.; Cutler, D.; et al. Integration of
Open-Source URBANopt and Dragonfly Energy Modeling Capabilities into Practitioner Workflows for District-Scale Planning
and Design. Energies 2021, 14, 5931. [CrossRef]
43. Ciechanowicz, D.; Pelzer, D.; Bartenschlager, B.; Knoll, A. A Modular Power System Planning and Power Flow Simulation
Framework for Generating and Evaluating Power Network Models. IEEE Trans. Power Syst. 2016, 32, 2214–2224. [CrossRef]
44. Sowmya, A.; Jitendra, A. Implementing Power Distribution System Using Geographic Information System. J. Comput. Sci. Inf.
Secur. 2016, 14, 221–224.
45. Roshanagh, G.; Ravadaneg, N. A Framework for Optimal Coordinated Primary- Secondary Planning of Distribution Systems
Considering MV Distributed Generation. IEEE Trans. Smart Grid 2016, 9, 1408–1415. [CrossRef]
46. Xie, S.; Hu, Z.; Zhou, D.; Li, Y.; Kong, S.; Lin, W.; Zheng, Y. Multi-objective active distribution networks expansion planning by
scenario-based stochastic programming considering uncertain and random weight of network. Appl. Energy 2018, 219, 207–225.
[CrossRef]
47. Pinzón, S.; Yánez, S.; Ruiz, M. Optimal Location of Transformers in Electrical Distribution Networks Using Geographic
Information Systems. Enfoque UTE 2020, 11, 84–95. [CrossRef]
48. Cresta, M.; Gatta, F.M.; Geri, A.; Maccioni, M. Resilience Assessment in Distribution Grids: A Complete Simulation Model.
Energies 2021, 14, 4303. [CrossRef]
49. Ayalew, M.; Khan, B.; Giday, I.; Mahela, P.; Khosravy, M. Integration of Renewable Based Distributed Generation for Distribution
Network Expansion Planning. Energies 2022, 15, 1378. [CrossRef]
50. Kostelac, M.; Herenčić, L.; Capuder, T. Planning and Operational Aspects of Individual and Clustered Multi-Energy Microgrid
Options. Energies 2022, 15, 1317. [CrossRef]
51. Li, J.; Ma, X.Y.; Liu, C.C.; Schneider, K.P. Distribution System Restoration with Microgrids Using Spanning Tree Search. IEEE
Trans. Power Syst. 2014, 29, 3021–3029. [CrossRef]