Distributed Energy

Storage Devices
in Smart Grids
Edited by
Guido Carpinelli, Pasquale De Falco and Fabio Mottola
Printed Edition of the Special Issue Published in Energies

www.mdpi.com/journal/energies
Distributed Energy Storage Devices in
Smart Grids
Distributed Energy Storage Devices in
Smart Grids

Special Issue Editors

Guido Carpinelli
Pasquale De Falco
Fabio Mottola

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin
Special Issue Editors
Guido Carpinelli Pasquale De Falco
University of Napoli Federico II University of Napoli Parthenope
Italy Italy

Fabio Mottola
University of Napoli Federico II
Italy

Editorial Office
MDPI
St. Alban-Anlage 66
4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal
Energies (ISSN 1996-1073) (available at: https://www.mdpi.com/journal/energies/special issues/
distributed energy storage devices smart grids).

For citation purposes, cite each article independently as indicated on the article page online and as
indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. Journal Name Year, Article Number,
Page Range.

ISBN 978-3-03928-434-4 (Pbk)

ISBN 978-3-03928-435-1 (PDF)


c 2020 by the authors. Articles in this book are Open Access and distributed under the Creative
Commons Attribution (CC BY) license, which allows users to download, copy and build upon
published articles, as long as the author and publisher are properly credited, which ensures maximum
dissemination and a wider impact of our publications.
The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons
license CC BY-NC-ND.
Contents

About the Special Issue Editors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

Preface to ”Distributed Energy Storage Devices in Smart Grids” . . . . . . . . . . . . . . . . . . ix

Hyung Tae Kim, Young Gyu Jin and Yong Tae Yoon
An Economic Analysis of Load Leveling with Battery Energy Storage Systems (BESS) in an
Electricity Market Environment: The Korean Case
Reprinted from: Energies 2019, 12, 1608, doi:10.3390/en12091608 . . . . . . . . . . . . . . . . . . . 1

Mayank Jha, Frede Blaabjerg, Mohammed Ali Khan, Varaha Satya Bharath Kurukuru and
Ahteshamul Haque
Intelligent Control of Converter for Electric Vehicles Charging Station
Reprinted from: Energies 2019, 12, 2334, doi:10.3390/en12122334 . . . . . . . . . . . . . . . . . . . 17

Davide De Simone and Luigi Piegari

Integration of Stationary Batteries for Fast Charge EV Charging Stations
Reprinted from: Energies 2019, 12, 4638, doi:10.3390/en12244638 . . . . . . . . . . . . . . . . . . . 43

Andrea Mazza, Hamidreza Mirtaheri, Gianfranco Chicco, Angela Russo and

Maurizio Fantino
Location and Sizing of Battery Energy Storage Units in Low Voltage Distribution Networks
Reprinted from: Energies 2020, 13, 52, doi:10.3390/en13010052 . . . . . . . . . . . . . . . . . . . . 55

Fabio Mottola, Daniela Proto, Pietro Varilone and Paola Verde

Planning of Distributed Energy Storage Systems in μGrids Accounting for Voltage Dips
Reprinted from: Energies 2020, 13, 401, doi:10.3390/en13020401 . . . . . . . . . . . . . . . . . . . . 75

Nayeem Chowdhury, Fabrizio Pilo and Giuditta Pisano

Optimal Energy Storage System Positioning and Sizing with Robust Optimization
Reprinted from: Energies 2020, 13, 512, doi:10.3390/en13030512 . . . . . . . . . . . . . . . . . . . . 95

Abbas Marini, Luigi Piegari, S-Saeedallah Mortazavi and Mohammad-S Ghazizadeh

Coordinated Operation of Energy Storage Systems for Distributed Harmonic Compensation in
Microgrids
Reprinted from: Energies 2020, 13, 771, doi:10.3390/en13030771 . . . . . . . . . . . . . . . . . . . . 115

v
About the Special Issue Editors
Guido Carpinelli (Full Professor) is currently with the Department of Electrical Engineering and
Information Technology, University of Napoli Federico II, Italy. His research interests mainly
include power quality, power system analysis, and energy forecasting. He has served as Editor and
Technical Committee Member for several international journals and international conferences. He
has co-authored more than 200 papers in relevant journals and conference proceedings, three books
in Italian, three books in English, and one book in Chinese. He was Guest Editor of two Special
Issues in MDPI journals.

Pasquale De Falco (Post-doc researcher) received a Ph.D. in Information Technology and Electrical
Engineering from the University of Napoli Federico II, Italy in 2018. He is currently with the
Department of Engineering, University of Napoli Parthenope, Italy. His research interests mainly
include energy forecasting, energy data analytics, and probabilistic methods applied to power
systems. He serves as Editor of two MDPI journals (Electronics and Forecasting) and he was Guest
Editor of four Special Issues in MDPI journals.

Fabio Mottola (Assistant Professor) received a Ph.D. in Electrical Engineering from the University of
Napoli Federico II, Italy in 2008. He currently is with the Department of Electrical Engineering and
Information Technology, University of Napoli Federico II, Italy. His research interests mainly include
the planning and operation of electrical energy systems and lightning effects on power lines. He is a
Senior Member of the Institute of Electrical and Electronics Engineers.

vii
Preface to ”Distributed Energy Storage Devices in
Smart Grids”
Energy storage systems have been recognized as viable solutions for implementing the smart
grid paradigm, but have created challenges for load levelling, integrating renewable and intermittent
sources, voltage and frequency regulation, grid resiliency, improving power quality and reliability,
reducing energy import during peak demand periods, and so on. In particular, distributed energy
storage addresses a wide range of the above potential issues, and it is gaining attention from
customers, utilities, and regulators.
Distributed energy storage has considerable potential for reducing costs and improving the
quality of electric services. However, installation costs and lifespan are the main drawbacks to
the wide diffusion of this technology. In this context, a serious challenge is the adoption of new
techniques and strategies for the optimal planning, control, and management of grids that include
distributed-energy storage devices. Regulatory guidance and proactive policies are urgently needed
to ensure a smooth rollout of this technology.
This collection of recent contributions addresses the development of methodologies applied
to the integration of distributed energy storage devices in smart power systems. Several areas of
research, such as optimal siting and sizing of energy storage systems, adaption of energy storage
systems to load leveling and harmonic compensation, integration of electric vehicles and optimal
control systems, are investigated in the contributions collected in this book, including:

• An economic analysis of load leveling with battery energy storage systems in an electricity
market from the perspective of a utility company and/or a government agency (Kim, H.; Jin, Y.;
Yoon, Y. An Economic Analysis of Load Leveling with Battery Energy Storage Systems (BESS)
in an Electricity Market Environment: The Korean Case. Energies 2019, 12(9), 1608);

• A proposal of a multiport converter for integrating photovoltaics, electric vehicle charging

docks, and energy storage devices with the grid system (Jha, M.; Blaabjerg, F.; Khan, M.; Bharath
Kurukuru, V.; Haque, A. Intelligent Control of Converter for Electric Vehicles Charging Station.
Energies 2019, 12(12), 2334);

• The configuration and control of a converter able to integrate batteries in a fast electric vehicle
charging station by decoupling the grid power from vehicle power using several auxiliary
battery modules (De Simone, D.; Piegari, L. Integration of Stationary Batteries for Fast Charge
EV Charging Stations. Energies 2019, 12(24), 4638);

• A method that fully supports battery energy storage system location and sizing, considering the
time-variable generation and demand patterns and by considering both technical and economic
aspects (Mazza, A.; Mirtaheri, H.; Chicco, G.; Russo, A.; Fantino, M. Location and Sizing of
Battery Energy Storage Units in Low Voltage Distribution Networks. Energies 2020, 13(1), 52);

• A multi-step procedure for the optimal planning of the electrical energy storage systems in
the microgrids aimed at minimizing the total costs while considering the compensation of the
voltage dips (Mottola, F.; Proto, D.; Varilone, P.; Verde, P. Planning of Distributed Energy Storage
Systems in μGrids Accounting for Voltage Dips. Energies 2020, 13(2), 401).

ix
• A methodology for energy storage placement in the distribution networks, in which robust
optimization accommodates system uncertainty (Chowdhury, N.; Pilo, F.; Pisano, G. Optimal
Energy Storage System Positioning and Sizing with Robust Optimization. Energies 2020, 13(3),
512).

• An optimization model to operate energy storage systems as coordinated active harmonic

filters for distributed compensation (Marini, A.; Piegari, L.; Mortazavi, S.; Ghazizadeh, M.
Coordinated Operation of Energy Storage Systems for Distributed Harmonic Compensation
in Microgrids. Energies 2020, 13(3), 771).

This book is relevant for energy practitioners involved in smart grid planning, management,
and control. It also represents a good starting point for young researchers who want to discover
the potential and up-to-date areas of research related to the services provided by distributed energy
storage systems in smart grid context. Eventually, the contents of the book may be used by system
operators to address the planning of future installations of energy storage system to boost the
performance of networks.

Guido Carpinelli, Pasquale De Falco, Fabio Mottola

Special Issue Editors

x
 energies
Article
An Economic Analysis of Load Leveling with Battery
Energy Storage Systems (BESS) in an Electricity
Market Environment: The Korean Case
Hyung Tae Kim 1 , Young Gyu Jin 2, * and Yong Tae Yoon 1
1 Department of Electrical and Computer Engineering, Seoul National University, 1 Gwanak-ro, Gwanak-gu,
Seoul 08826, Korea; laputa11@snu.ac.kr (H.T.K.); ytyoon@snu.ac.kr (Y.T.Y.)
2 Department of Electrical Engineering, Jeju National University, 102 Jejudaehak-ro, Jeju-si, Jeju 63243, Korea
* Correspondence: ygjin93@jejunu.ac.kr; Tel.: +82-64-754-3677

Received: 22 March 2019; Accepted: 23 April 2019; Published: 27 April 2019

Abstract: The capacity of battery energy storage systems (BESS) is expected to increase for power
system applications. However, as the cost of BESS is high, economic feasibility must be considered
when using BESS in grid applications. Load leveling with BESS is one such application for which the
economic implications have been analyzed in the literature. However, these studies do not sufficiently
consider the fact that the leveled loads will lead to a change in electricity prices, thereby modifying
charging/discharging operations of BESS. Additionally, in a competitive electricity market, electricity
prices are not determined by the generator cost functions. Market participants’ strategic decisions
also affect prices. Therefore, we conducted an economic analysis of load leveling with BESS in an
electricity market from the perspective of a utility company and/or a government agency. In our
analysis of the Korean market, we examine whether the leveled loads necessarily lead to economic
benefits. Load leveling performance and the associated economic benefit are quantitatively analyzed
for varying sizes of BESS. Further, the policy implications related to using BESS are derived from the
analysis results.

Keywords: battery energy storage systems (BESS); economic analysis; load leveling; electricity market

1. Introduction
The urgent environmental need to reduce greenhouse gas emissions has led to the integration of
renewable energy sources (RESs) into the power grid. In 2018, the global cumulative installed capacity
of RESs such as wind power and solar photovoltaic (PV) generation amounted to 539,123 MW [1] and
402.5 GW [2], respectively. However, RESs have disadvantages in terms of variability, intermittency,
and uncertainty of electricity generation [3]. Therefore, energy storage systems (ESS), particularly
battery ESS (BESS), have recently been attracting significant interest in the electricity industry as an
effective means of mitigating the disadvantages of RESs [4,5].
The installed BESS capacity for power system applications is expected to increase consistently, not
only for utility-scale applications but also for distributed small-scale applications, whereby a large
number of BESS are expected to be connected to PV systems and behind the meter [6,7]. This trend of
increasing BESS usage is likely to be accelerated by the widespread adoption of electric vehicles (EVs),
supported by improvements in vehicle-to-grid and grid-to-vehicle technologies [8]. Furthermore, cost
reductions in BESS technology are widely expected to become a key determinant of their increased
installation. For instance, it is expected that the cost of lithium-ion BESS will decrease significantly,
namely by 54–61%, between 2016 and 2030 [6], or by 52% between 2018 and 2040 [9]. For the latter
period, it is estimated that USD 620 billion in investment in the vehicle and the electricity sectors will
ensue [9].

Energies 2019, 12, 1608; doi:10.3390/en12091608 1 www.mdpi.com/journal/energies

Energies 2019, 12, 1608

Despite the positive forecasts regarding the cost of BESS, currently, costs are still so high that
the economic feasibility of their installation is not always given [10]. For this reason, several studies
have dealt with the economic aspects of BESS [11–38]. For instance, studies examine the BESS size
and the associated charging/discharging strategy for optimal usage, either by minimizing total cost
or maximizing the net present value (NPV) of grid-connected renewable energy systems [11–15].
Further, for a given BESS capacity, studies examine the operating strategy (i.e., the charging/discharging
schedule) is determined to minimize the cost or maximize the net benefit under a dynamic pricing
scheme [16–18]. Other researchers consider how a bidding strategy for BESS in an electricity market
environment can also be developed to increase profit [19,20]. The economic aspects of BESS have
also been considered for EV charging station planning [21–23]. Finally, economic considerations are
important in the process of designing and operating a microgrid with ESS [24–26], where the methods
based on particle swarm optimization have been used to determine an optimal size and/or an optimal
operational strategy of ESS in a microgrid. Particular consideration is given to the integration of RESs
into the microgrid in [24–26].
Economic feasibility is also an important factor to be considered for the various grid applications
of BESS [5,27]. This study focuses on the application of load leveling and analyzes its economic benefits.
Peak shaving using the discharging operation of BESS (when electricity demand is particularly high) is
inevitably followed by the charging operation during off-peak periods. Therefore, the analysis of load
leveling with BESS in this study includes peak shaving. Previous studies dealing with the economic
aspects of load leveling with BESS obtained the following results: It was found that load leveling
with BESS achieves an increase in the cost savings of thermal units [28–30]. Further, research suggests
that BESS can be used to maximize the NPV of a renewable energy system and a substation [31,32].
Evidence also suggests that the leveled loads with BESS may provide economic benefits by reducing
the required capacity of other equipment, such as generators and transformers, and by reducing the
frequency with which investment in such network equipment is needed [33]. Finally, it has been shown
that an economic benefit can also be obtained by peak shaving or peak reduction with BESS under a
dynamic electricity pricing scheme [34–36]. Since the accurate forecasts of generation and load are
essential to improving the performance of peak shaving, a method for peak shaving combined with a
forecasting method based on the adaptive neuro-fuzzy inference system is presented in [37].
Customers of electricity providers, particularly the industrial customers, pay the electricity bill,
which is comprised of the consumed energy cost and the peak demand surcharge. Thus, the economic
benefit of load leveling or peak reduction with BESS is obvious to customers. Since the buying cost
of electricity is a major cost component of a utility company, the provider’s economic benefit with
BESS corresponds to the reduction in the purchasing cost. Further, this economic benefit is relevant
for the government agencies which are responsible for the policies of the electricity industry because
the government permits billing rates to customers considering the cost of a utility. It can induce a
utility to lower the bill if the cost can be reduced by load leveling with BESS. This study considers the
perspective of a utility company and/or government agency. Hence, the problem to address in this
study can be simply expressed as the following questions: Does load leveling with BESS necessarily
lead to the economic benefit of a utility company? If so, how does the economic benefit change as the
BESS size increases? If not, what is the reason for the results contrary to the previous studies?
Answering these questions fills a research gap as the insightfulness of previous studies is limited
by the fact that electricity prices are always regarded as given. However, if the number and/or size
of BESS is large, then the leveled loads with BESS will result in a change in electricity prices, which,
in turn, modifies the charging/discharging operation of BESS. Accounting for this reality induces
an iterative process that is challenging to resolve. If the electricity price follows a typical parabolic
cost function, the iterative process may be resolved by mathematically finding a converging point.
However, in a competitive electricity market, the strategic decisions of participants may distort the
electricity price pattern. Therefore, the change in electricity prices due to the leveled loads with BESS
should be estimated as a basis for performing any economic analysis. Consequently, in this study, we

2
Energies 2019, 12, 1608

derive the answers to the research questions by conducting an economic analysis of load leveling from
the perspective of a utility company, while considering changes in electricity prices, particularly thus
induced by large BESS in an electricity market environment. To illustrate the results, we consider the
situation of Korea’s electricity market and use the corresponding data as a case study.
The contributions of this paper are (1) the change in electricity prices due to the leveled loads are
explicitly considered in the economic analysis as the price function is defined with respect to loads;
(2) in defining the price function, a data-driven regression model is used and estimation accuracy is
achieved by data grouping; (3) it is shown as a main claim that it may be blind faith for load leveling
with BESS whereby a large cumulative capacity of BESS is always economically beneficial.
The remainder of the paper is organized as follows. To provide important background information,
Section 2 summarizes the structure of the Korean electricity market. Section 3 describes the optimization
formulation to perform load leveling and a method to estimate electricity prices with respect to the
load. Section 4 presents the necessary data for the analysis. Section 5 presents the results of the case
study, followed by a discussion in Section 6. Finally, concluding remarks are provided in Section 7.

2. Korean Electricity Market

The electric industry in Korea was re-structured in 2001, whereby the vertically integrated utility
company, the Korea Electric Power Corporation (KEPCO), was divided into an independent system
operator, the Korea Power Exchange (KPX); some generation companies (GENCOs), subsidiaries of
KEPCO; and the KEPCO itself playing correspondingly a reduced role. The Korean electricity market
was also defined as a result of the restructuring and is now run by the KPX.
The Korean electricity market is a cost-based pool. The GENCOs submit the amount of electricity
generation capacity and the cost function to the KPX. Then, the KPX determines the hourly generation
of each GENCO based on forecasted demand. The methods applied to this end, such as economic
dispatch or optimal power flow, must satisfy various conditions, such as balance of supply and demand,
reserve requirements, and maximum power flows of transmission lines. The KPX also sets the hourly
system marginal price (SMP). The SMP is usually set as the marginal generation cost of the most
expensive generator, or the marginal generator, and is used in the settlement process. The KEPCO
buys electricity through the market and pays the GENCOs according to the SMP. The KEPCO recovers
the cost of purchasing electricity by invoicing customers based on the billing rates approved by the
government agency.

3. Methodology

3.1. Load Leveling

The variation in original loads is mitigated or leveled by BESS charging/discharging operations.
The load leveling performance can be evaluated based on how close the leveled loads are to a reference
value. The mean-squared error (MSE) is selected in this study as a measure of closeness to the reference
value. Thus, the objective function to determine the optimal BESS operation for load leveling is the
MSE with respect to the reference value (Pre f ,h ):

2
1  
H
PL,h − Pre f ,h (1)
H
h=1

where PL,h is the original load at hour h and H is the number of hours corresponding to the time horizon
for load leveling. The reference, Pre f ,h , is defined for all h as the constant average of the original loads
during the entire time period:


H
Pre f ,h = 1
H PL,k , h = 1, 2, · · · H (2)
k =1

3
Energies 2019, 12, 1608

Alternatively, BESS can be operated for load leveling on a daily basis. In this case, the reference
value on a specific day d may be defined as:


24
Pre f ,h = 1
24 PL,k , k∈d (3)
k =1

For convenience, the references in Equations (2) and (3) are denoted as the yearly reference (Yref)
and the daily reference (Dref), respectively. After including BESS, PL,h in (1) is replaced with the leveled
L,h , defined as:
load, P
L,h = PL,h + PB,h
P (4)

where PB,h is the electricity used to charge the BESS (leading to a positive value) or discharged from
the BESS (resulting in a negative value) at hour h. Therefore, load leveling involves determining the
value of PB,h that minimizes Equation (1), subject to the BESS-related constraints described below.
The first constraint is related to the maximum hourly charging and discharging power, Pmax B
,
given as:
− Pmax
B ≤ PB,h ≤ Pmax
B (5)

In this study, the value of Pmax

B
is considered equal to BESS capacity. For normal operation, the
state of charge (SOC) of the BESS should be bounded:
upp
SOClow
B ≤ SOCB,h ≤ SOCB (6)
upp
where SOCB,h is the SOC at hour h and SOClowB
and SOCB are the lower and upper bounds of the
SOC, respectively. The value of SOCB,h is updated every hour according to the following rule:
⎧
⎪
⎪ SOCB,h−1 + ηB,ch PB,h /Pmax , if PB,h > 0 (charging)
⎪
⎨ B
SOCB,h =⎪
⎪ SOCB,h−1 , if PB,h = 0 (no operation) (7)
⎪
⎩ SOC
B,h−1 + PB,h / (ηB,dch PB ),
max if PB,h < 0 (discharging)

where ηB,ch and ηB,dch are the charging and discharging efficiencies, respectively. From Equation (7),
even if ηB,ch and ηB,dch are equal, the update equation is asymmetric because the effects of these values
are different for charging and discharging. However, the asymmetry for considering the loss of BESS
makes the solution of the optimization problem using the mixed integer quadratic programming
method extremely time-consuming. For example, when H is set to 168 h (one week), both ηB,ch and
ηB,dch are set to 95%, the BESS size is 50 GWh, the computation time to find the optimal solution is 110
s. By contrast, if the loss of BESS is neglected, that is, if ηB,ch and ηB,dch in Equation (7) are equal to one,
it can be reduced to a simpler equation:

SOCB,h = SOCB,h−1 + PB,h /Pmax

B (8)

As a result, for the same load leveling problem over 168 h, the computation time falls significantly
to just 0.2 s, which is a 99.82% decrease in computation time. The leveled loads with and without the
loss of BESS are shown in Figure 1 together with the original load. Figure 1 clarifies that the solutions
for the two cases are very similar in shape, and their mean absolute percentage errors (MAPEs) are
as small as 0.26%. This minor difference is not the primary concern of this study. Moreover, in
our analysis, H is equal to 8760 h (one year). This means that if the formulas that consider the loss
of BESS are applied, the computation time would be too large because the search space expands
exponentially as H increases. Consequently, Equation (8) without loss is used instead of Equation (7)
for the remainder of this study. This enhances computational efficiency without causing a noteworthy
loss of informativeness of results.

4
Energies 2019, 12, 1608

Figure 1. Leveled loads with and without the loss of battery energy storage systems (BESS) compared
with original loads.

3.2. Electricity Price as a Function of Load

Electricity prices should be represented as a function of load to estimate the change in price for the
leveled loads. Among the methods for identifying the relationship between input and output values,
we selected the polynomial regression method [38], in which the load corresponds to the input, and
the electricity price is the output. Then, the price function, ρh (PL,h ), can be represented as:

ρh (PL,h ) = β0 + β1 PL,h + β2 P2L,h + · · · + βm Pm

L,h
+ εh (9)

where βm ’s are the regression coefficients and εh is the error term. It should be noted that the original
loads, PL,h , are used in (9) because the price function is found using the original loads. Then, the
change in price after load leveling is estimated by substituting the leveled load, PL,h , into the price
function in (9).
The degree of the polynomial, m, is selected according to the adjusted R2 index, which indicates
the quality of the regression model. When a model fits the data well, the adjusted R2 index becomes
closer to 1 [38]. Therefore, the degree of the polynomial is chosen as the value where the adjusted R2
saturates at a value closer to 1 or its rate of increase becomes moderately small.

4. Case Description
The data on loads and electricity prices are taken from the Korean power system for the year
2018 [39,40]. The case study consists of three steps. The first step is to find the leveled loads using
BESS based on the formulation in Section 3.1. It is performed by varying the BESS size from 1 GWh
to 150 GWh by increments of 1 GWh, where the maximum size is approximately 1.7 times the peak
load of 87.03 GWh. The maximum and minimum SOC are set to 90% and 10%, respectively. The
two references of Yref and Dref are applied separately. The second step is to represent the electricity
prices as a function of loads using a polynomial regression. In order to determine a suitable degree of
polynomial, the adjusted R-squared values are calculated by varying the degree of the polynomial
(which entails varying the number of terms in the polynomial regression) from one to nine. Then, the
fitted price function of loads is derived as a polynomial function of the selected degree. The final step
is to perform the economic analysis by calculating the total cost of leveled loads based on the results
from the first and second steps.
For a more detailed economic analysis, another simulation, which is essentially the opposite of
the previous scenario, is performed to examine how the BESS operation for reducing the cost affects

5
Energies 2019, 12, 1608

the load leveling performance. In other words, BESS are operated with the goal of minimizing the
total cost, not optimizing load leveling. Then, it can be investigated to what extent the adjusted loads
are spread with respect to the reference, Yref or Dref, by calculating the MSE. In this second scenario,
the first step is to again vary the BESS size from 1 GWh to 150 GWh by increments of 1 GWh, thus
determining the charging/discharging schedule of BESS, which reduces the total cost for the original
electricity prices. Then, the adjusted loads and the corresponding total cost are calculated. The second
step is to calculate the MSE for the adjusted loads with respect to Yref and Dref.
The monthly and hourly load patterns are represented as box plots using Matlab in Figure 2a,b,
respectively. When generating Figure 2a, the hourly loads over the year are grouped by month. Thus,
the number of data points in each month is equal to 24 h times the number of days in the corresponding
month. Then, the monthly loads in each month are sorted in ascending order and represented as a box
plot, where the upper and lower sides of the blue box indicate the 75th percentile (the third quartile,
Q3) and 25th percentile (the first quartile, Q1), respectively. The red line in the middle of the blue box
indicates the median (50th percentile) of the sorted loads and the highest and lowest black lines indicate
the maximum and minimum values, respectively. When generating Figure 2b, the hourly loads over
the year are grouped by hour, such that there are 365 data for each hour group. Then, the hourly
loads for each hour are sorted in ascending order and represented as a box plot. Like Figure 2a, the
upper side, the red line, and the lower side of the blue box indicate the 25th percentile, 50th percentile
(median), and 75th percentile of the sorted loads. However, the hourly loads are distributed quite
widely, such that there are some points that are greater than Q3 + 1.5 times (Q3 − Q1) or less than
Q1 − 1.5 times (Q3 − Q1). These data points are referred to as whiskers. This type of points can be
interpreted as outliers in data analysis and are represented as red crosses in Figure 2b. If there are red
crosses outside the highest and lowest black lines, then the black line indicates Q3 + 1.5 times (Q3 −
Q1) or Q1 − 1.5 times (Q3 − Q1); otherwise, it indicates the maximum or minimum value of the sorted
hourly loads. The average hourly load of the analyzed year is equal to 61.25 GWh. Figure 2 shows
that summer (July and August) and winter (January, February, and December) are the peak seasons of
electricity consumption over the year. In addition, during a day, two peaks occur just before and just
after noon.

(a) (b)
Figure 2. Box plots for original loads: (a) Monthly loads; (b) Hourly loads.

The monthly and hourly patterns of original electricity prices are represented as the box plots
using Matlab in Figure 3a,b, respectively. The methods of grouping the price data for the year and the
meaning of the symbols of the box plot are the same as those for the loads in Figure 2a,b. Electricity
prices need to be represented as a function of load in this study, which requires a relationship to exist
between them. Figure 3b shows a similar pattern to Figure 2a, although peak and off-peak hours are
not as apparent as the loads. Therefore, it can be inferred that there is a positive relationship between
loads and electricity prices. However, the monthly variation of electricity prices in Figure 3a does not

6
Energies 2019, 12, 1608

show a clear pattern which was observed seen in Figure 2a. Therefore, we checked the correlation
coefficient between loads and electricity prices. The correlation coefficient for the entire dataset is 0.47,
indicating a moderately positive correlation. By contrast, Table 1 shows that the monthly correlation
coefficients are very high, indicating a highly positive relationship. Consequently, the polynomial
regression is performed on monthly data, so that a total of 12 price functions are found to estimate the
price change in response to the leveled loads. The specific results of the polynomial regression are
presented in Section 4.

(a) (b)

Figure 3. Box plots for original electricity prices: (a) Monthly prices; (b) Hourly prices.

Table 1. Correlation coefficients between monthly loads and electricity prices.

January February March April May June

0.84 0.79 0.81 0.92 0.72 0.85
July August September October November December
0.90 0.84 0.81 0.88 0.87 0.84

5. Simulation Results

5.1. Load Leveling

It is expected that the load will be leveled more and the MSE will decrease as the BESS size
increases. Further, the important result to be examined is not the absolute value of the MSE, but how
much the MSE is reduced when BESS size is increased. Therefore, the load leveling performance in
terms of the MSE is represented in Figure 4a as the percentage ratio to the MSE of the original loads
without BESS. Since two references of Yref and Dref are used in this study, two graphs of the MSE
ratio are shown in Figure 4a. Additionally, the original loads and the leveled loads with the BESS
sizes of 50 GWh and 150 GWh for the month of May are shown in Figure 5 to identify the changes
from load leveling more clearly. Another measure to examine the load leveling performance is the
maximum/minimum values of the leveled loads because they will be closer to a reference, that is,
the maximum will decrease and the minimum will increase, as the BESS size increases. Thus, the
maximum/minimum values of the leveled loads for Yref and Dref are given in Figure 4b.
Figure 4a shows that the marginal improvement in the load leveling performance by the unit
increase in BESS size gradually decreases as BESS size increases. This is because the load duration
of very high or very low loads is relatively short, so that a relatively small BESS can significantly
improve load leveling. However, as BESS size increases, the duration for which loads need to be
lowered or raised with respect to the reference value increases. Therefore, the degree of reduction in
MSE decreases. More specifically, the decrease in the degree of reduction is severe, especially for Yref,
because the load leveling is performed for the same constant reference over the entire time period.
Figure 4a clarifies this relationship by showing the slow decrease in the MSE for Yref, which makes a

7
Energies 2019, 12, 1608

significant difference for a large BESS. Specifically, the MSE for Dref is reduced to less than 10% for a
BESS size greater than 70 GWh. By contrast, the MSE for Yref is still equal to approximately 48%, even
for a BESS size of 150 GWh.
On the other hand, as Figure 4b shows, the maximum and minimum values of the leveled loads
are almost the same for the two references. Specifically, the maximum values of the leveled loads
for the two references are equal to each other in the beginning (hence why the lines in the graph
almost exactly coincide), but a difference in the minimum values becomes evident from the BESS size
of 30 GWh. The results can be clarified with Figure 5. Regardless of whether the load leveling is
performed during a year or a day, the goal is to lower the high loads and raise the low loads. Thus,
the overall shapes of Figure 5 for the two reference values are similar. However, the highs and lows
improve more for Yref than for Dref. This is because the number of days with high loads, such as
summer and winter weekdays, is greater than the number of days with low loads, such as holidays
and spring/autumn weekends. Therefore, lowering the maximum total loads for Yref is equivalent
to lowering the maximum of a day of high loads for Dref. However, after increasing the minimum
of the holidays and weekends of low loads, the minimum of the leveled days of low loads remains
unchanged. Consequently, as seen in Figure 4b, the minimum of the leveled load for Dref hardly
changes after some point, which corresponds to a BESS size of 30 GWh in this case study.

(a) (b)

Figure 4. Mean-squared error (MSE) and maximum/minimum values of the leveled loads with respect
to BESS size: (a) MSE for Yref and Dref; (b) Maximum and minimum values for Yref and Dref.

(a) (b)

Figure 5. Original and leveled loads with BESS of 50 GWh and 150 GWh in May: (a) Leveled loads for
Yref; (b) Leveled loads for Dref.

8
Energies 2019, 12, 1608

5.2. Representation of Electricity Prices

In Figure 6, electricity prices are represented as a scatter plot with respect to the load. Although
the electricity price appears to increase as the load increases, a relationship between them is difficult to
identify. Therefore, considering the monthly correlation coefficients in Table 1, we divide the data by
month and represent them as separate scatter plots in Figure 7. Contrary to Figure 6, the variation of
data points is small, so a relationship between electricity prices and loads becomes evident.

Figure 6. Electricity prices with respect to the loads over a year.

Figure 7. Electricity prices with respect to loads (blue dots) in each month and the fitted prices (solid
red lines) using the polynomial regression.

The results of the adjusted R-squared values are shown in Figure 8. By checking whether the
adjusted R-squared value is close enough to 1, the degrees of the polynomial of each month, from
January to December, are selected as 6, 3, 4, 4, 4, 4, 5, 3, 4, 4, 4, and 4, respectively. The fitted functions
are represented as the solid red lines on the original data points in Figure 7. The fitted graph of each

9
Energies 2019, 12, 1608

month in Figure 7 represents the original electricity prices well. Specifically, the MAPE of the fitted
electricity prices during a year is equal to 1.78%.

Figure 8. Adjusted R-squared values of the polynomial regression with a degree of polynomial from
one to nine.

5.3. Economic Analysis

First, the accuracy of the fitted electricity price function is examined in terms of the total cost.
The total cost for the original prices and loads is calculated as 50,776 billion KRW and the total cost for
the fitted prices and original loads is also equal to 50,776 billion KRW. This is evidence of the suitability
of the fitted price function. The estimated costs after load leveling with the BESS are shown in Figure 9
as a percentage of the total cost of 50,776 billion KRW for the original prices and loads without BESS.
Figure 9 shows that, regardless of whether the load leveling is performed for Yref or Dref, the total cost
monotonically increases as BESS size increases. Based on the expectation that an economic benefit
should accrue from load leveling due to the reduction in peaks, this result seems counterintuitive at
first. However, closer inspection of Figure 7 indicates that this should nevertheless be the natural result
because electricity prices increase fast in the range of small loads, but they increase slowly in the range
of high loads. As a result, the increased cost due to upward leveling of small loads is greater than the
decreased cost from downward leveling of high loads. Consequently, the total cost increases after load
leveling with BESS. In addition, Figure 7 shows that the total cost for Yref is greater than that for Dref.
This implies that more effort is required to flatten the loads over a longer time period, resulting in a
greater cost increase.
The results suggest that we need to rethink the purpose of load leveling. If the price function
is a quadratic one based on the cost of the physical operation of generators [41], the total cost will
monotonically decrease as BESS size increases, although the extent of the decrease will gradually
become smaller. Then, it should be a natural decision to use BESS for load leveling, and the problem
is essentially reduced to deciding on a suitable BESS size by comparing the cost decrease from the
leveled loads against the cost increase from the BESS. However, as Figure 9 indicates, the costs which
are physically incurred by generators is not the only factor to be considered in deciding on electricity
prices or bidding prices in a market environment. For example, under some bidding strategies, bid
prices for low loads could be set much higher than the actual generation costs and those for high loads

10
Energies 2019, 12, 1608

could be set at a level comparable to the actual generation cost. In this case, the electricity prices
shown in Figure 7 would result. Consequently, without other benefits, such as improved reliability
and stability due to the leveled load, one cannot simply assume that load leveling is a worthwhile grid
application of BESS from an economic standpoint, particularly in an electricity market environment.

Figure 9. Estimated cost of leveled loads with respect to BESS size.

5.4. BESS Operation for Reducing the Cost

The results of the second scenario, the cost minimization scenario, are shown in Figure 10. Like
Figure 9, the total cost and the MSE are represented in Figure 10 as the percentage of the corresponding
results for the original prices and loads without BESS. Figure 10 shows that the total cost decreases
linearly as the BESS size increases because BESS are operated to minimize the total cost. However, the
MSEs for Yref and Dref increase rapidly to tens of times the MSE without BESS, which is completely
against the purpose of load leveling. This is because, in the process of minimizing the cost, the loads
greater than the reference value are increased and the loads less than the reference value are decreased.
This means that the loads can be changed to deviate further from the reference value to reduce the cost.

Figure 10. Total cost from optimized BESS operation to reduce the cost and MSEs for Yref and Dref of
the resulting loads.

6. Discussion
The results of load leveling in Section 5.1 indicate that loads can be flattened more as BESS size
increases. However, it should be noted that the extent of load leveling performance is gradually

11
Energies 2019, 12, 1608

reduced, while the costs of BESS remain high. Therefore, if the total cost of electricity consumption
decreases due to the leveled loads, it may be necessary to compare the marginal reduction in the total
cost reduction with the marginal increase in the investment for the unit increase in BESS size. Then,
the optimal BESS size can be determined. However, as presented in Section 5.3, the total cost rather
monotonically increases as the BESS size increases and the load is correspondingly flattened more.
This suggests that the load leveling with BESS may not necessarily lead to economic benefits for a
utility company, particularly in a competitive electricity market environment. Subsequently, it becomes
meaningless to discuss how much the economic benefit is as the BESS size increases and which BESS
size is optimal for the load leveling application in terms of economic benefit. What is the reason for
these counterintuitive results? It can be inferred from Figure 7 that the complex strategic decisions of
the participants in an electricity market may result in a complex pattern of electricity prices, whereby
prices increase quickly in the range of small loads but increase slowly in the range of high loads. In
this case, it may be simply incorrect to state that the utility’s investment in BESS can be recovered after
some years through the reduced purchasing cost of electricity.
This conflict between the economic benefit and the load leveling application with BESS can be
verified more clearly by reviewing the results of the second scenario. In order to reduce the total cost,
BESS are charged during the periods of low electricity price and discharged during high price periods.
However, this means that operating BESS with the goal of reducing electricity costs actually widened
the variation of the resulting loads. Consequently, in a certain electricity market environment, the
economic operation of BESS to recover investment costs will increase the fluctuation of net loads. As a
result, the side costs of maintaining the reliability and stability of electricity networks may undermine
the utility company’s profit.
The results in this study lead to some guidelines for designing a policy related to the BESS for load
leveling. First, it should not simply be assumed that BESS with a high cumulative capacity is always
economically beneficial even if the investment cost in BESS is low enough. Second, load leveling
with the BESS may rather increase the total cost of electricity provision, particularly in a competitive
electricity market environment. In this case, the load leveling needs to be linked to other purposes than
economic benefit to justify the validity of using the BESS. Third, when a large number of distributed
BESS are integrated into the power system, the pricing mechanism of electricity should be elaborately
designed to prevent the synchronized operation of the BESS. This is necessary to ensure systems remain
sufficiently profitable and that the power system is reliable.

7. Conclusions
BESS, alongside renewable energy sources, are expected to become a major component of future
power systems. However, an analysis of their economic feasibility is essential because, at present, the
costs of investing in BESS are significant. Among the various grid applications of BESS, this study
focused on load leveling and analyzed its economic benefits in an electricity market environment. For
economic analysis, load leveling is formulated as a constrained optimization problem, which is then
solved using mixed integer quadratic programming. Additionally, the electricity prices are represented
as a function of the load using the polynomial regression method to estimate the adjusted prices for
the leveled loads.
The results of the economic analysis using Korean data show that the degree of reduction in the
MSEs of the leveled loads decreases as BESS size increases. The problem that MSE reduction decreases
with the increase in BESS size is more severe for the yearly reference because load leveling is performed
for the same constant reference over a longer time period. Further, the results show that, contrary to
the simple expectation that load leveling will result in decreased cost of electricity generation, the total
cost monotonically increases as the BESS size increases and accordingly, the loads are flattened more.
By contrast, if BESS are operated to minimize the cost, the variation of net loads severely increases,
which is the opposite to what load leveling is intended to achieve.

12
Energies 2019, 12, 1608

The results suggest some guidelines for designing policy related to BESS for load leveling. First, it
should not be assumed that a large cumulative capacity of BESS is always economically beneficial if its
price is low enough. Second, load leveling with BESS may actually increase the total cost of electricity
provision, particularly in a competitive electricity market environment. In this case, the justification for
pursuing load leveling needs to be based on some purpose other than economic benefit to ensure the
validity of using BESS. Finally, when a large number of distributed BESS are integrated into the power
system, the electricity pricing scheme should be elaborately designed to prevent the synchronized
operation of BESS and thereby, ensure the power system remains reliable.
Meaningful further research could consider the following topics: First, it is assumed that GENCOs’
bids do not change for utilities’ application of load leveling with BESS. That is, the price function
itself may change with leveled loads, which results in another iterative process. Thus, defining this
iterative mechanism becomes a research issue, which will be very difficult to be addressed because
it involves strategic, not just technical, decisions on behalf of GENCOs. Second, if the leveled loads
improve reliability and stability of power systems, a different kind of economic analysis can be done by
translating the improved reliability into monetary value. Third, it is assumed in this study that BESS
are operated only for load leveling. Thus, it may be necessary to consider that BESS are utilized partly
for load leveling and partly for another purpose, and to determine an optimal proportion between the
applications of BESS.

Author Contributions: H.T.K. developed the optimization model, set up the simulation environment, and
performed the simulations. Y.G.J. designed the study, performed the analysis, thoroughly revised the paper and
checked the overall logic of the work. Y.T.Y. provided insightful comments on the modeling and analysis.
Funding: This research was supported by Korea Electric Power Corporation (Grant number: R18XA03) and the
2019 scientific promotion program funded by Jeju National University.
Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations
Acronyms
RES Renewable energy source
PV Photovoltaic
ESS Energy storage systems
BESS Battery energy storage systems
EV Electric vehicle
NPV Net present value
KEPCO Korea Electric Power Corporation
KPX Korea Power Exchange
GENCO Generation company
SMP System marginal price
MSE Mean-squared error
SOC State of charge
MAPE Mean absolute percentage error
Indices
h Index of hour
d Index of day
Parameters
xPre f ,h Reference for load leveling
PL,h Original load at hour h
Pmax
B
BESS size
SOClowB
Lower bound of the SOC
upp
SOCB Upper bounds of the SOC
ηB,ch Charging efficiency of BESS
ηB,dch Discharging efficiency of BESS

13
Energies 2019, 12, 1608

H Number of hours
m Degree of polynomial in regression
Variables
L,h
P Leveled load at hour h
Electricity charged (positive) to or discharged
PB,h
(negative) from BESS at hour h
SOCB,h SOC of BESS at hour h
ρh Electricity price at hour h
βm Regression coefficients
εh Error term in regression

References
1. Global Wind Energy Council. Global Wind Report-Annual Market Update 2017. Available online:
http://gwec.net/publications/global-wind-report-2/ (accessed on 6 July 2018).
2. International Energy Agency. A Snapshot of Global PV (1992–2017). Available online: http://www.iea-pvps
.org/?id=266 (accessed on 6 July 2018).
3. Xie, L.; Carvalho, P.M.S.; Ferreira, L.A.F.M.; Liu, J.; Krigh, B.H.; Popli, N.; Ilic, M.D. Wind Integration in
Power Systems: Operational Challenges and Possible Solutions. Proc. IEEE 2011, 99, 214–232. [CrossRef]
4. Castillo, A.; Gayme, D.F. Grid-Scale Energy Storage Applications in Renewable Energy Integration: A Survey.
Energy Convers. Manag. 2014, 87, 885–894. [CrossRef]
5. Luo, X.; Wang, J.; Dooner, M. Overview of Current Development in Electrical Energy Storage Technologies
and the Application Potential in Power System Operation. Appl. Energy 2015, 137, 511–536. [CrossRef]
6. IRENA. Electricity Storage and Renewables: Costs and Markets to 2030. International Renewable Energy
Agency (IRENA), 2017. Available online: https://www.irena.org/publications/2017/Oct/Electricity-storage-an
d-renewables-costs-and-markets (accessed on 25 February 2019).
7. U.S. Energy Information Administration. Battery Storage Market Trends. U.S. Energy Information
Administration, 2018. Available online: https://www.eia.gov/analysis/studies/electricity/batterystorage/
(accessed on 25 February 2019).
8. Coignard, J.; Saxena, S.; Greenblatt, J.; Wang, D. Clean Vehicles as an Enabler for a Clean Electricity Grid.
Environ. Res. Lett. 2018, 13, 054031. [CrossRef]
9. Bloomberg NEF. Energy Storage is a $620 Billion Investment Opportunity to 2040. 2018.
Available online: https://about.bnef.com/blog/energy-storage-620-billion-investment-opportunity-2040/
(accessed on 19 February 2019).
10. Xia, S.; Chan, K.W.; Luo, X.; Bu, S.; Ding, Z.; Zhou, B. Optimal Sizing of Energy Storage System and Its
Cost-Benefit Analysis for Power Grid Planning with Intermittent Wind Generation. Renew. Energy 2018, 122,
472–486. [CrossRef]
11. Korpaas, M.; Holen, A.T.; Hildrum, R. Operation and Sizing of Energy Storage for Wind Power Plants in a
Market System. Int. J. Electr. Power Energy Syst. 2003, 25, 599–606. [CrossRef]
12. Ru, Y.; Kleissl, J.; Martinez, S. Storage Size Determination for Grid-Connected Photovoltaic Systems.
IEEE Trans. Sustain. Energy 2013, 4, 68–81. [CrossRef]
13. Harsha, P.; Dahleh, M. Optimal Management and Sizing of Energy Storage Under Dynamic Pricing for the
Efficient Integration of Renewable Energy. IEEE Trans. Power Syst. 2015, 30, 1164–1181. [CrossRef]
14. Khalilpour, R.; Vassallo, A. Planning and Operation Scheduling of PV-Battery Systems: A Novel Methodology.
Renew. Sustain. Energy Rev. 2016, 53, 194–208. [CrossRef]
15. Talent, O.; Du, H. Optimal Sizing and Energy Scheduling of Photovoltaic-Battery Systems under Different
Tariff Structures. Renew. Energy 2018, 129, 513–526. [CrossRef]
16. Telaretti, E.; Ippolito, M.; Dusonchet, L. A Simple Operating Strategy of Small-Scale Battery Energy Storages
for Energy Arbitrage under Dynamic Pricing Tariffs. Energies 2016, 9, 12. [CrossRef]
17. Hemmati, R. Technical and Economic Analysis of Home Energy Management System Incorporating
Small-Scale Wind Turbine and Battery Energy Storage System. J. Clean. Prod. 2017, 159, 106–118. [CrossRef]

14
Energies 2019, 12, 1608

18. Lujano-Rojas, J.M.; Dufo-López, R.; Bernal-Agustín, J.L.; Catalão, J.P.S. Optimizing Daily Operation of
Battery Energy Storage Systems Under Real-Time Pricing Schemes. IEEE Trans. Smart Grid 2017, 8, 316–330.
[CrossRef]
19. Faqiry, M.N.; Edmonds, L.; Zhang, H.; Khodaei, A.; Wu, H. Transactive-Market-Based Operation of
Distributed Electrical Energy Storage with Grid Constraints. Energies 2017, 10, 1891. [CrossRef]
20. Fleer, J.; Zurmühlen, S.; Meyer, J.; Badeda, J.; Stenzel, P.; Hake, J.-F.; Sauer, D.W. Price Development
and Bidding Strategies for Battery Energy Storage Systems on the Primary Control Reserve Market.
In Proceedings of the 11th International Renewable Energy Storage Conference, IRES 2017, Düsseldorf,
Germany, 14–16 March 2017.
21. Xiang, Y.; Liu, J.; Li, R.; Li, F.; Gu, C.; Tang, S. Economic Planning of Electric Vehicle Charging Stations
Considering Traffic Constraints and Load Profile Templates. Appl. Energy 2016, 178, 647–659. [CrossRef]
22. Baik, S.H.; Jin, Y.G.; Yoon, Y.T. Determining Equipment Capacity of Electric Vehicle Charging Station Operator
for Profit Maximization. Energies 2018, 11, 2301. [CrossRef]
23. Luo, L.; Gu, W.; Zhou, S.; Huang, H.; Gao, S.; Han, J.; Wu, Z.; Dou, X. Optimal Planning of Electric Vehicle
Charging Stations Comprising Multi-Types of Charging Facilities. Appl. Energy 2018, 226, 1087–1099.
[CrossRef]
24. Nikolovski, S.; Baghaee, H.R.; Mlakic, D. ANFIS-Based Peak Power Shaving/Curtailment in Microgrids
Including PV Units and BESSs. Energies 2018, 11, 2953. [CrossRef]
25. Kaviani, A.K.; Baghaee, H.R.; Riahy, G.H. Optimal Sizing of a Stand-Alone Wind/Photovoltaic Generation
Unit Using Particle Swarm Optimization. Simulation 2009, 85, 89–99. [CrossRef]
26. Baghaee, H.R.; Mirsalim, M.; Gharehpetian, G.B.; Talebi, H.A. Reliability/Cost-Based Multi-Objective Pareto
Optimal Design of Stand-Alone Wind/PV/FC Generation Microgrid System. Energy 2016, 115, 1022–1041.
[CrossRef]
27. Palizban, O.; Kauhaniemi, K. Energy Storage Systems in Modern Grids—Matrix of Technologies and
Applications. J. Energy Storage 2016, 6, 248–259. [CrossRef]
28. Lo, C.H.; Anderson, M.D. Economic Dispatch and Optimal Sizing of Battery Energy Storage Systems in
Utility Load-Leveling Operations. IEEE Trans. Energy Convers. 1999, 12, 824–829. [CrossRef]
29. Kerestes, R.J.; Reed, G.F.; Sparacino, A.R. Economic Analysis of Grid Level Energy Storage for the Application
of Load Leveling. In Proceedings of the 2012 IEEE Power and Energy Society General Meeting, San Diego,
CA, USA, 22–26 July 2012; pp. 1–9.
30. Hemmati, R.; Saboori, H. Short-Term Bulk Energy Storage System Scheduling for Load Leveling in Unit
Commitment: Modeling, Optimization, and Sensitivity Analysis. J. Adv. Res. 2016, 7, 360–372. [CrossRef]
[PubMed]
31. Dicorato, M.; Forte, G.; Pisani, M.; Trovato, M. Planning and Operating Combined Wind-Storage System in
Electricity Market. IEEE Trans. Sustain. Energy 2012, 3, 209–217. [CrossRef]
32. Cho, S.M.; Shin, H.S.; Kim, J.C. Modeling of Battery Energy Storage System at Substation for Load Leveling
and Its Economic Evaluation. Trans. Korean Instit. Electric. Eng. 2012, 61, 950–956. [CrossRef]
33. Dupont, G.; Baltus, P. Dimensioning and Grid Integration of Mega Battery Energy Storage System for
System Load Leveling. In Proceedings of the 2009 IEEE Bucharest PowerTech, Bucharest, Romania,
28 June–2 July 2009.
34. Zheng, M.; Meinrenken, C.; Lackner, K.S. Smart Households: Dispatch Strategies and Economic Analysis of
Distributed Energy Storage for Residential Peak Shaving. Appl. Energy 2015, 147, 246–257. [CrossRef]
35. Telaretti, E.; Dusonchet, L. Battery Storage Systems for Peak Load Shaving Applications: Part 1: Operating
Strategy and Modification of the Power Diagram. In Proceedings of the 2016 IEEE 16th International
Conference on Environment and Electrical Engineering (EEEIC), Florence, Italy, 7–10 June 2016.
36. Graditi, G.; Ippolito, M.G.; Telaretti, E.; Zizzo, G. Technical and Economical Assessment of Distributed
Electrochemical Storages for Load Shifting Applications: An Italian Case Study. Renew. Sustain. Energy Rev.
2016, 57, 515–523. [CrossRef]
37. Baghaee, H.R.; Mirsalim, M.; Gharehpetian, G.B. Multi-Objective Optimal Power Management and Sizing of
a Reliable Wind/PV Microgrid with Hydrogen Energy Storage Using MOPSO. J. Intell. Fuzzy Syst. 2017, 32,
1753–1773. [CrossRef]
38. James, G.; Witten, D.; Hastie, T.; Tibshirani, R. An Introduction to Statistical Learning with Applications in R,
7th ed.; Springer: New York, NY, USA, 2013.

15
Energies 2019, 12, 1608

39. Hourly Load Forecast of Korea Power eXchange. Available online: http://www.kpx.or.kr/www/contents.do?
key=223 (accessed on 18 January 2019).
40. System Marginal Price of Korea Power eXchange. Available online: http://www.kpx.or.kr/www/contents.do?
key=225 (accessed on 18 January 2019).
41. Bergen, A.R.; Vital, V. Power System Analysis, 2nd ed.; Prentice-Hall: Upper Saddle River, NJ, USA, 2000.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

16
 energies
Article
Intelligent Control of Converter for Electric Vehicles
Charging Station
Mayank Jha 1 , Frede Blaabjerg 2, * , Mohammed Ali Khan 1 , Varaha Satya Bharath Kurukuru 1
and Ahteshamul Haque 1
1 Department of Electrical Engineering, Jamia Millia Islamia, New Delhi 110025, India;
mayank.jha.595@gmail.com (M.J.); mak1791@gmail.com (M.A.K.); kvsb272@gmail.com (V.S.B.K.);
ahaque@jmi.ac.in (A.H.)
2 Department of Energy Technology, Aalborg University, Aalborg 9220, Denmark
* Correspondence: fbl@et.aau.dk

Received: 28 May 2019; Accepted: 14 June 2019; Published: 18 June 2019

Abstract: Electric vehicles (EVs) are envisaged to be the future transportation medium, and
demonstrate energy efficiency levels much higher than conventional gasoline or diesel-based vehicles.
However, the sustainability of EVs is only justified if the electricity used to charge these EVs is availed
from a sustainable source of energy and not from any fossil fuel or carbon generating source. In
this paper, the challenges of the EV charging stations are discussed while highlighting the growing
use of distributed generators in the modern electrical grid system. The benefits of the adoption of
photovoltaic (PV) sources along with battery storage devices are studied. A multiport converter
is proposed for integrating the PV, charging docks, and energy storage device (ESD) with the grid
system. In order to control the bidirectional flow between the generating sources and the loads, an
intelligent energy management system is proposed by adapting particle swarm optimization for
efficient switching between the sources. The proposed system is simulated using MATLAB/Simulink
environment, and the results depicted fast switching between the sources and less switching time
without obstructing the fast charging to the EVs.

Keywords: Grid Connected Photovoltaic Systems (GCPVS); Energy Storage Device (ESD); Electric
Vehicle (EV); Multiport Converter (MPC); Intelligent Energy Management System (iEMS)

1. Introduction
The current environmental challenges of reducing greenhouse gases and the potential shortage of
fossil fuels motivate widespread research on electric vehicle (EV) systems [1]. However, the research
on EVs is highly impacted by the consumer disposition for switching to EVs as an alternative for
conventional internal combustion engine vehicles. This willingness is the main factor in forecasting
future demand for EVs. In Reference [2], the authors depicted that the charging time is one of the main
challenges that the EV industry is facing. Generally, the EV charging levels are classified according to
their power charging rates [3]. Overnight charging takes place in level I, as the EVs are plugged to a
convenient power outlet (120 V) for slow charging (1.5–2.5 kW) over long hours. The main concern
of level-I is the long charging time, which renders this charging level unsuitable for long driving
cycles when more than one charging operation is needed. Moreover, from the electrical grid operation
point of view, the long charging hours at night overloads the distribution transformers as they are not
allowed to rest in a grid system with a high number of connected EVs [4]. Level-II charging requires
a 240 V outlet; thus, it is characteristically used as the prime charging means for public and private
facilities. This charging level is capable of supplying power in the range of 4–6.6 kW over a period
of 3–6 h to restock the depleted EV batteries. The time required is still the main drawback in this
charging level. Additionally, voltage sags and high-power losses in an electrical grid system with a

Energies 2019, 12, 2334; doi:10.3390/en12122334 17 www.mdpi.com/journal/energies

Energies 2019, 12, 2334

high penetration of level II charging are some of the challenges that are facing its widespread. Control
and coordination in level II would reduce the negative impacts of level-II charging [5]; however, this
requires an extensive communication system to be adopted.
In general, both levels-I and II require single-phase power sources with onboard vehicle chargers.
On the contrary, three-phase power systems are used with off-board chargers for level III fast charging
rates (50–75 kW). The use of fast charging stations significantly reduces the EV charging time for a
complete charging cycle. Additionally, widespread deployment of fast EV charging stations across the
urban and the residential areas would eliminate the EV range anxiety concern [6,7]. However, the
high-power charging rates are essential over a short interval of time for level-III charging impose a very
high demand on the utility grid [8,9]. The current grid infrastructure is not capable of supporting the
desired high charging rates of level-III. Thus, accomplishing fast charging rates while solely depending
on the electrical grid does require not only the improvement of the charging system, but also the
improvement of the electrical grid capacity. Additionally, drawing large amounts of current from
the electrical grid will increase the utility charges especially at the peak hours and consequently
will increase the system cost. The impact of an EV charging station load on electric grid systems is
thoroughly discussed in Reference [10].
A possible solutions to these challenges could be the installation of a distributed generator (DG)
near the fast charging station site, as it generates electrical supply that is projected towards on-site
use [11]. A few decades ago, a number of developments began to change the basics of operation of the
electrical grid industry leading to the rise of DGs. The ambitious targets of the higher deployment
rates of the DGs into the electrical pool is achievable, due to advancement of technologies, and the
enhancements, in the fields of power electronics and smart grids. Additionally, new regulations and
policies are continuously issued favoring the distributed generation and the net metering concepts.
However, the type of energy used in fueling the deployed DG sources on the demand side is a decisive
factor in the economic viability of the DGs concept in today’s electrical distribution market.
Regarding the distributed generation, renewable energy sources (RES) have a distinct advantage
in their ability to be deployed in residential and urban areas, due to their environmentally friendly
operation and minimal maintenance requirements. Consequently, photovoltaics (PV) are considered
as an effective solution, due to its sustainability and ubiquity. The major advantage with PV systems is,
they are effectively utilized for different power generation levels starting from low-power domestic
applications to mega power PV based power plants. The grid-connected PV installations are utilized
with deeper penetrating levels compared to the standalone PV installations. This is due to the
continuous reliance on the electric grid as a stable source/load that can compensate for the PV power
fluctuations. In Reference [12], research by George et al. simulated a solar energy-based charging
station by considering the decongestive knots in the capacity of network distribution systems.
In the proposed system, PV operates as grid-tied DG source. Solar energy is a preferable DG
source in EV charging applications for two reasons:

1. The PV panels are more effective than other renewables in populated and residential areas, due
to their noise-free operation and low maintenance requirements.
2. PV panels generate most of their energy during the highly priced grid tariff hours of the electrical
grid. Thus, the EV charging stations can offset the high-cost electricity with solar energy during
peak hours.

The PV power production is highly impacted by the ambient temperature and solar insolation
levels, which causes discontinuity of operation. Consequently, connecting the PV panels directly to the
load without any subsidiary systems leads to a negative impact on the performance of the connected
electrical loads. One of the applicable solutions for the aforementioned challenge is the design of a
hybrid RESs system using various RESs, which can relatively offset a portion of the local fluctuations in
every generation unit. However, this requires an appropriate selection of the most suitable generation
technologies, as well as a proper sizing of the (RES) [13]. The combination of PV sources with wind

18
Energies 2019, 12, 2334

energy is explored in Reference [14]. However, such a system requires either the insertion of storage
devices or a connection point to the electrical grid in order to support the necessary loads continuously.
The authors in Reference [15] presented the use of a PV source with fuel cells to meet the requirements
of a residential load. In doing so, a reserve capacity had to be maintained at the PV source to supply
the load changes as the fuel cells technology instills a slow dynamic response. This leads to a deviation
between the PV maximum power point (MPP) and the system operating point. A hybrid system,
composed of PV/fuel-cells/ultra-capacitors, is used in Reference [16] to accommodate these challenges.
In this study, the ultracapacitors are selected, due to their fast-dynamic response, which leads to their
ability to mitigate the rapid fluctuations of the PV power while continuously tracking the PV MPP.
However, this configuration fails to meet the load demands during extended hours of low insolation
level conditions (e.g., night hours or shady days).
Under such scenarios, storage devices can play a key role in the integrated RES. The literature
contains different examples of RES combined with energy storage systems. Various energy storage
technologies are studied in the literature to be connected with the wind generators, hydrogen cells,
and ultracapacitors [17–19] in order to provide steady output power to the electrical loads.
The authors in Reference [20] introduced the use of distributed batteries in a grid-tied PV power
plant to improve energy production. Achieving a constant power production from the PV power plant
is the main defined objective of inserting an energy storage system, as shown in Reference [21]. In
Reference [22], a storage battery is added to the grid-connected PV system to reduce the PV power
fluctuations in a defined range within a particular period of time while maximizing the revenue. This
is motivated by meeting the utilities regulations and restrictions on the PV power injected to the grid.
Therefore, from the literature, the advantages of integrating battery storage devices with PV
power in order to attain a stable power supply at a minimal operating cost is observed. In this research,
energy storage batteries are coupled with PV panels in a grid-tied system. The proposed hybrid system
is designed to provide means of fast charging for EVs. In this paper, energy storage devices, such
as batteries, are suggested to be combined with PV sources to sustain the continuous power supply
to the connected loads regardless of the power fluctuations in the PV sources [23]. Additionally, the
integration of the grid with hybrid PV-battery system allows for a higher degree of deregulation on the
demand side, which may result in achieving lower running costs for high performance.
The objective of this paper is to provide a grid-connected PV system with energy storage device
based fast charging solution for transportation infrastructure for high EV penetrations. An efficient
configuration of the proposed system using a multiport converter (MPC) and an optimal power flow
management tool are both desirable to supply the demanded high-power charging rates. Thus, new
converter topologies are presented in this paper, and the novel power flow management proposed could
result in a more efficient operation of the system. An overall image of the proposed grid-connected PV
system with energy storage devise for the EV charging station is shown in Figure 1. The presented
hybrid system is utilizing DG technology by the adoption of the PV/battery sources on the demand side.
These DG sources are connected to the electrical grid in a grid-tied system to provide fast charging
power rates to the EVs. Further sections of the paper are arranged as follows: Section 2 depicts the
charging scenarios available for EVs. Section 3 examines the various PV–EV charging architectures
and different converters available for integrating the PV, EV, and ESD with the grid. Section 4 depicts
the intelligent energy management and dynamic power flow between the various sources and loads in
a grid-connected PV–EV with ESD. Section 5 discusses the methodology for formulating the objective
function for power flow optimization using particle swarm optimization (PSO). Section 6 presents
the experimental analysis of the proposed system under different modes of operation, and the results
depict the efficiency of the proposed control and management techniques.

19
Energies 2019, 12, 2334


Figure 1. An overview of the proposed hybrid system.

2. EV Charging Scenario
As renewable energy is the future of power generation, it is necessary that it is considered for
electric vehicle charging as well. PV panel-based EV charging station is being established in many
countries. For understanding the solar-powered EV charging infrastructure [24], it is necessary to
realize the existing EV–PV system, which is in use already by the industries or are in the development
stage by a different academic institution.

EV Charging
EV charging can be performed by AC [3], as well as DC [25]. Power obtained from the grid is
converted into DC for recharging the battery. Range of power required for charging varies for HEV,
PHEV, and PEV.
Pcp = Vevh Ievh , (1)

where Pcp denotes Power of DC charge, Vevh voltage for electric vehicle charging and Ievh represents
current required for charging the electric vehicle.

tcp
Ecp = Pce dt, (2)
0

where Ecp is the energy delivered by the batter over tcp period.
When AC based charging is performed on EV, an AC/DC power converter is required for
connecting the battery with the 1- or 3-phase system. AC charger used globally can be categorized into
three types.
Type 1: It is a single-phase charging method generally adopted by the US (SAE J1772-2009) [26].
Three pin plugs are used for charging (phase, earth and neutral).
Type 2: It is a single, and three-phase charging method generally adopted by Europe
(VDE-AR-E-2623-2-2) [27]. Three-phase plugs contain five pins (Phase 1, Phase 2, Phase 3, neutral
and earth).
Type 3: It is a single, and three-phase charging method generally adopted by Alliance.
As per IEC 61851-1 standards [28], the charging can be classified into four modes of operation.
Both mode 1 and 2 derive from the standard power socket. In the case of mode 2, inbuild protection is

20
Energies 2019, 12, 2334

provided to the system. Mode 3 consists of electric vehicle supply equipment (EVSE), which ensures
protection and control functionality to be present in chargers. Mode 4 discuss the DC-based charging,
which is generally accounted when the power of the electric vehicle is more than 50 kW. DC-based
charger reduces the requirement of AC/DC converter on board hence reducing the space and weight
constraint of onboard chargers. The most commonly used DC charger is a combined charging system
(CCS) [29], Type 4 CHAdeMO [30] and Tesla dual charger [31]. The major benefit of using a DC-based
charging is that a bidirectional flow can be created between vehicle to home, building, load, grid or
any other power deficit system easily.
Economical and sustainability are two of the merit for using PV systems for electric vehicle
charging. The implementation of solar energy for charging results in higher efficiency regarding the
contacts of fuel usage and the life cycle of the battery [32]. As it can be determined from the discussion
above DC-based charging is mostly preferred for PV based EV charging. As both the EV and PV are
of DC nature, it’s easy to implement, and there is a possibility of smart charging, which indicates
the variation of charge concerning time. V2G can also be achieved easily by the implementation of
DC charging.
For charging the EVs from solar energy, various architecture can be implemented. EVSE based
charging in case of grid connection is implemented for charging EV directly for AC grid. Two of the
conventional power converters that can be implemented for the integration of PV, EV, and grid are (i)
Multi port-based converters for PV, EV, and grid integration. (ii) Separated power converters for EV,
PV, and grid with common point interlink. AC or DC interconnects are required between different
power converters. They help in interconnection the PV power amongst different EV and EV power to
the grid. From the power converter types mentioned four types of architecture could be deduced for
charging of electric vehicles.
Architecture 1: AC interconnection along with separate converter for PV and EV
The separate converter is used for PV panel and EV charging and discharging. PV panel consists
of a DC/AC converter with maximum power point tracking (MPPT), whereas the AC/DC converter is
implemented at the EV end. A grid of 50 Hz has a major role to play in this architecture as all the power
is passed through the grid. The major drawback is that DC power generated by PV can’t be directly
used for EV charging, creating an unnecessary requirement of DC to AC and AC to DC conversion in
the process.
Architecture 2: DC interconnection along with separate converter for PV and EV
In this architecture, both PV and EV requires a DC/DC converter. For EV charge control needed to
be present with converter, whereas, for PV, an MPPT controller is required. DC interconnection helps
in the utilization of DC power from the PV for battery charging directly. For grid interconnection, a
central inverter is used. The central inverter is vital as it draws power from PV and EV depending on
the demand side requirement. Separate construction of DC interconnection makes the architecture less
desirable as the existing AC base infrastructure is not utilized to its full extent.
Architecture 3: AC interconnection along with multiport converter for PV and EV
In this architecture, the PV–EV and grid are interconnected to a central DC link using multiport
converters. With the help of grid multiple, MPC relates to each other. Because of the interconnection
control of the system becomes easy along with low effective cost and high-power density. The previous
architecture used communication-based control, whereas in this architecture, control is provided
directly via MPC. One of the major disadvantages of this system is that DC for PV using one MPC
can’t be used to charge the EV connected to alternative MPC without AC conversion.
Architecture 4: DC interconnection along with multiport converter for PV and EV
This architecture consists of merit from architecture 2 and 3. MPC is used for interconnection of
PV and EV. DC interconnection is used to link different MPC. The grid is connected to the system with
the help of a high-power central inverter. All the architectures corresponding to charging of EV from
PV are depicted in Figure 2.

21
Energies 2019, 12, 2334

Figure 2. System architectures for electric vehicle (EV)—photovoltaic (PV) charging using multiport converters.

3. PV Assisted EV Charging
This section examines the design of charging infrastructure for an electric vehicle with the help of
PV panels. The proposed system can be adapted for residential areas and workplaces to charge electric
cars as per the convenience of the EV user. The major objective here is to maximize the use of PV energy
for EV charging by utilizing energy storage systems and minimize the energy exchange with the grid.

3.1. System Design Architectures for Solar-Powered Charging Stations

Messenger [33], and Stapleton [34] suggest that for any PV based system to operate effectively,
the entire system design should be done by considering the parametric and location constraints for
installation of PV systems, load requirements, and electric codes as per the location. Considering these
constraints, there are three major types of system design for charging stations: Off-grid PV system
with energy storage device (ESD), grid-connected PV (GCPV) system with ESD, and GCPV system
without ESD. Conventionally, an off-grid PV system with ESD type of architecture is used for EV
charging stations. In this architecture, energy from the PV module and, ESD (battery) together is
used for meeting the load power requirements, and any excess power is fed back into ESD. The major
drawback with this system is that effective utilization of this type of architecture can be only done if an
external control system is incorporated. Apart from the control aspect, the system is not reliable, due
to its dependency on solar irradiance [35]. In order to overcome these drawbacks, charging stations
associated with the GCPV System without ESD were implemented. In this type of architecture, the
PV power generation is mainly used for meeting the load power requirements and the excess power
other than charging EV load is injected into the grid. The grid acts as a storage unit in this type of
architecture. The only disadvantage of this system is, loss of energy storage capabilities with grid
failure could pose a difficulty to meet load requirements.
Considering all these concerns and constraints, the GCPV System with ESD is developed in this
research. The basic structure of this system is shown in Figure 1. The major advantage with this system

22
Energies 2019, 12, 2334

is, it is more reliable when compared to the conventional systems like this it will continue to satisfy
loads even when the grid fails. Also, it can be designed to have less reliance on the grid.
Apart from it, 100% utilization of time to use (TOU) pricing is possible in this type of system with
the presence of an additional controller to control the power injected into the grid from the DC side.

3.2. Power Converter Types for Grid Connected PV with ESD Assisted EV Charging
A grid-connected PV with energy storage device assisted EV charging system would involve the
PV system, AC grid, energy storage device, EV and a power electronic converter interface to combine
and link them. This converter must be capable of enabling the charging of EV from all the three utilities,
i.e., PV and AC grid and facilitate charging of EVs and ESD from both the sources. This operation will
guarantee that the grid can support the EV through G2V during the absence of PV and ESD and can
charge the ESD for emergency power requirements. In this section, an optimal multiport converter
topology that integrates EV, PV and the AC grid are proposed. The need to design such an optimal
topology is necessary to ensure that the converter has low cost, high power density, and high efficiency.
Conventionally, an optimal converter topology is chosen by considering several parameters, such as
converter volume, efficiency, the number of components, controllability, ripples, and opportunity for
efficiency improvement.

3.2.1. DC Link-Based Converter

Considering the above parameters, initially, the converter types are designed based on a DC-link.
These types of systems act as a high voltage energy buffer among the ports. Conventionally, there are
3 sub converters in a DC-link based multiport converter with multiple converter topologies. These
control strategies include a boost converter and MPPT algorithm integrated with the PV array for
maximum power, a buck-boost converter for controlling the charging needs of EV and ESD. The
complete setup is linked to the local grid with the help of a voltage source converter, which operates
bidirectionally and which is accountable for power stability with the grid. The major advantage with
such topology is, it’s simple power flow between PV and EV, and minimized DC/AC conversion losses.
This topology also improves the AC/DC conversion of VSC for charging the EVs during peak load
and when both PV and ESD were exhausted. Apart from the advantages mentioned, the topology
possesses some major drawbacks, due to its large DC link capacitors and multiple controllers for
each sub-converter.

3.2.2. Impedance-Network Based

Carli [35], Rasinab [36], and Rasin [37] discussed impedance network based multiport converter
topology. These converters deal with a variable DC-link voltage connected with a secluded DC-DC
converter for EV charging. These converters have major advantages, due to their low component
count, which improves the reliability of the system. The converter topology also has an advantage over
fewer control algorithms. The major drawback for such type of converters is that it is most intrinsically
modular and has very high control complexity, due to variable DC link voltage. The system architecture
of both DC link based multiport converter and impedance network-based converter is depicted in
Figure 3a,b.

23
Energies 2019, 12, 2334

Figure 3. Block representation of grid-connected PV–EV converters (a) DC-link based converter, (b)
Impedance-network based converter.

3.3. Multiport Converter

For this research, the DC link based multiport converters are used with modified control and
intelligent energy management system. There are two types of multiport converters, as shown in
Figure 4.

Figure 4. Block representation of multiport converters. (a) MPCA; (b) MPCB.

24
Energies 2019, 12, 2334

The major difference between both the possible MPC’s is the capability of isolating PV panels
from the grid. The Figure 4 depicts that, both the topologies adapt a central DC-link available within
the DC-DC converter. For efficient power flow management, the DC link voltage rating must be higher
than the peak voltage of the system. In Figure 4a, a non-isolated DC/DC converter is adapted for
achieving maximum power from the PV array. A high-frequency DC-DC conversion is adapted for
EV charging station by adding a high-frequency transformer between the conversion process as per
IEC 61727. For this research, the converter topology in Figure 4a is considered and tested with the
proposed energy management strategy.

4. Power Flow Management

In order to design the proposed system configuration, the flow of the power between the four
main elements in this system needs to be explored. The main elements are the connecting electrical
grid, the PV sources, the battery storage and the EVs charging load. The decision on the need for
a bi-directional power flow power electronic system, along with their sizing requirements, can be
decided based on power flow management. Consequently, the research attempts to solve the power
flow management problem by introducing their applicability on the studied application.
Conventionally, the power flow in a grid-connected PV/battery system is predefined by heuristic
rules that consider the load demand, the PV insolation levels and the off-peak utility hours [38].
However, a dynamic grid tariff complicates the solution of the proposed system further. In a dynamic
grid tariff system, the operation of the PV/battery system using the simplified heuristic rules will
provide running cost solutions that largely deviate from the minimal cost operation. Thus, the research
in this area has taken on an accelerated path.
In Reference [39], a Lagrangian relaxation technique is applied to determine the optimal hourly
battery charging or discharging current, where the objective is to maximize the contribution of the
hybrid system to the grid. The proposed technique in Reference [39] assumes there is no dispatch cost
associated with the PV/Battery output power. This leads to the negligence of the battery degradation
cost and its advisable operating conditions. Additionally, the formulation of the problem is limited to a
thermally based electrical grid system. A predefined rule-based model is presented in Reference [40]
where the battery energy storage is integrated into the renewable energy system in order to enable the
PV source to act as a dispatchable unit on an hourly basis. The objective of the battery storage utilization
in this paper leads to solutions that are not necessarily minimizing the running cost. Additionally, the
system is sensitive to solar power forecasts.
From the above literature, it is observed that, initially, the problem formulation should account
for the aging factor of the battery in order to extend the battery lifetime, and thus, increase the system
reliability. The desired power flow management topology has to accommodate non-linear functions.
This allows the generalization of the developed topology on different operating scenarios. An online
error compensation stage has to be included in the topology to allow the system to operate effectively
at mismatching conditions and forecasting deviations. Lastly, the online optimization stage should
be designed to operate with low computational time, which makes it easily integrated into real-time
controllers. Study regarding intelligent energy management strategy and dynamic power allocation is
required in order to overcome these drawbacks and achieve the proposed optimization.

4.1. Intelligent Energy Management Strategy

The EV–PV charger in its present structure will probably charge the EV from sun-based energy,
yet it does not possess any knowledge of its own. The energy price will remain low in early hours
of the day as per the forecasting; hence morning period is advantageous to charge EV from the grid.
While sunny afternoons provide advantage for charging from the solar. For the realization of control of
EV–PV system, smart charging algorithm necessary.

25
Energies 2019, 12, 2334

The situation of charging the vehicles at parking stations has been considered where every vehicle
has a predictable time of accessibility as load. The user defines the charge time window as input while
parking the vehicle, and the other preferences of the user could be pricing, type of charge, etc.
The other characterization of the vehicles is its capacity, chemistry, open circuit voltage level, state
of charge, temperature, etc. Depending upon the user preference and battery properties every vehicle
will assign the power to the system. Hence to conclude, energy utilization of the system is optimized
by considering customer preferences and load attributes using dynamic power management for loads.
The control techniques, advanced demand-side power management and clients’ preferences in utility
services are the main objectives for grid design. For that purpose, a given system has been designed to
show the elements of the smart grid. The basic structure of the complete system is shown in Figure 5.
This system accomplishes all the merits discussed and designed using energy management system,
power grid, charging and battery loads. The iEMS used to control every parking deck, and these
parking decks are made of multiple loads.

Figure 5. Architecture of the intelligent energy management system.

All the information related to the power available and the pricing as per the time period at the
grid side is automatically updated to the iEMS. This process helps in simultaneously monitoring the
multiple converters for making real-time decisions by iEMS.

4.2. Dynamic Power Allocation for Energy Management at EV Charging Station

The primary components of the system are EVs, ESD, energy management systems and the power
grid (utility). We consider vehicles clustered in parking decks, with each parking deck under the
control of an iEMS. Essentially, a cluster of vehicles and an energy storage device being controlled
by an iEMS is a manageable load for the grid. The iEMS acts as an interface between the vehicles
and the utility. Information about available power and pricing is regularly updated to the iEMS;
multiple chargers at a parking deck are then simultaneously supervised and controlled according to
the real-time inferences made by it.
At the time of plug-in, the customer will specify their preferences on the type of charge desired,
estimated time of availability, willingness to participate in V2G, the price they are willing to pay for
charging, etc. This information may be entered via an interface within the vehicle itself or an external
user GUI available at the parking deck or from an internet profile maintained by the user for that
vehicle [41]. The parking deck can be equipped with an intelligent charger at each vehicle parking
space, capable of acquiring the vehicle battery state and relaying it to the iEMS via a communication
medium; alternatively, each EV could be communication enabled, capable of sending its battery data
to the iEMS.

26
Energies 2019, 12, 2334

The electric load that can be sustained by the utility from a parking lot will change during the
course of the day as the system load varies. For vehicles willing to provide electricity back to the
grid, opportunities could also arise for V2G operation depending on the grid requirements. Thus, the
varying price of electricity can also be incorporated into the system information for better decision
making. Using the information from the utility and the vehicles, the iEMS will make a real-time
decision on power sharing to each vehicle and communicate this to the intelligent chargers.
Since the state of the system is continually changing with the arrival and departure of vehicles and
changing power requirements from the utility, the system states are sampled at regular time intervals
as the power allocation needs to be recomputed each time a vehicle plugs in/out, or there is a change in
power availability. The system is hybrid in nature; its event-based character arises from the plug-in,
plug-out activity and sampling time steps used by the iEMS for making its decisions. Moreover, the
process of vehicle charging is continuous with a non-linear power consumption curve, which varies
dynamically. The randomness of the preliminary states of charge, plug-in and out times and varying
power curves introduce dynamicity in the system, which makes the optimization for power allocation
a large scale, nonlinear, time-varying multi-objective problem with multiple constraints. In the ensuing
sections, we provide a framework for system modeling and operation, addressing system goals and
constraints along with optimization on a chosen objective.

4.2.1. System Modeling

The interactions in the system occur between the iEMS and the utility on the one hand and the
loads (vehicles) and the iEMS on the other. The iEMS acts as broker agent, taking information about
the power available from the utility on the one hand (can also be the transformer rating at the parking
deck), and the vehicle battery states and user preferences on the other hand. It makes a decision on
power allocation which is optimal in terms of satisfying both the parties. The system components can
be modeled on ‘Agents Based Approach’ [42], in which each entity has a set of attributes, states, and
functions and the entities interact with each other in order to achieve individual and system goals.

4.2.2. Optimization Objectives

A number of objectives are possible for the problem. Many objectives can be formulated around
user preferences. For example, the minimization of the time taken to charge the vehicle battery for all
the users, according to the price they are willing to pay. For users that do not mind compromising
on the time and would like to charge at the lowest prices, the objective could be to only charge when
the electricity cost is below a threshold. If the vehicle to the grid is also activated when the power
flow could be facilitated such that profit to each user is maximized [43]. Another objective could be to
minimize the overall power consumption (if desired by utility) while trying to guarantee a minimum
threshold SoC (say, 60%) for each vehicle. In this paper an optimization problem is formulated along
with the required constraints and the operating cost function is chosen as a combination of electricity
grid prices and the battery degradation cost. The proposed optimization procedure uses PSO, which
acts as a prediction layer for forecasting the system operation. This data is further processed for
switching between the different sources as per the load requirement and power available.

4.2.3. System Constraints

The primary constraint considered here is the power available from the utility. Type of charge
(slow, medium, fast), time of availability in the parking lot, the minimum desired state of charge at
plug-out, maximum price the user is willing to pay for charging, the maximum power that can be
absorbed by a vehicle battery, other battery requirements, etc. are other possible constraints. The user
could also specify the number of miles he/she plans to drive after plug out, in which case, the SoC for
achieving this will be guaranteed at plug-out. Another constraint for the iEMS could be in terms of
the layout of the parking deck, the sizing and capacity of the cables will place a limit on the power
that can be channeled to each vehicle(s). For a more robust system, we could consider the abrupt

27
Energies 2019, 12, 2334

plug-out by a user before the stated time, and in this case, the aim would be having some fairness in
the SoC distribution at every time step to have a reasonable SoC even before plug-out. Additional
constraints in the system could be in terms of the bandwidth availability of the communication channel
for sampling the states of the vehicles, which would limit the sampling time. System performance
with packet delays and drops could also be evaluated. The final goal of the iEMS is to be flexible in
terms of accommodating multiple objectives for different users. It also opens up the opportunity of
dividing the station into clusters, grouping vehicles/users with common objectives and assigning the
responsibility of optimizing on each cluster to a sub-iEMS thereby translating the problem into the
arena of distributed control. The decision on how much power to allocate to each sub-iEMS will be
made by the central iEMS or could also result from bidding actions by the sub-iEMSs.

4.3. Particle Swarm Optimization

PSO is an iterative stochastic optimization algorithm, which is derived from studying the pattern
in which a flock of birds or schools of fish travels [44]. Multi-dimension solution space is searched
by the algorithm by collective search containing different particles and the best solution found by
the other particles are communicated. The communication enables the system to take an informed a
decision about the movement of each particle to find the best solution available global. Weighting
factor and random variations are also considered in the algorithm to prevent any early convergence
whenever local minimum is present.
Following the analysis in Reference [45], the electric vehicle system constraints are adapted from
Reference [46]. Each constraint is defined by the user in the EV-iEMS. A given day is fragmented up
into different intervals as per the irradiance variation pattern.

5. Methodology
The initial configuration of the grid-connected PV system with an energy storage device, explored
in this paper is shown in Figure 1. The PV array, energy storage, and EV charging station (EV load) are
connected at the DC link through a high frequency DC/DC converter. An interconnection between
the DC link and the electric grid is achieved using a bi-directional AC/DC converter. The DC/DC
converter used at the energy storage is a bi-directional power electronic interfacing converter to allows
smooth operation during the battery charging and discharging modes. However, the uniqueness
of this study is the reliance on the proposed configuration on the power flow management results.
Thus, the shown configuration will be developed at the end of this section to reflect the findings of
the optimal power flow operation through a given period of operation. The battery power (Pb ) is
considered to be negative while it is charging and positive during discharging. All the charging slots,
and additional loads connected with the system are summed up to form the load power (PL ). Here,
the load power (PL ) is considered positive in all the cases as neither the battery storage, nor the EV
supply power back to the grid. The grid power (PG ) is negative while the distributed generation feeds
power into the grid, and it is considered positive when the grid is satisfying the loads.
Power flow management methodology is developed by considering all the operating modes
and assumptions. Initially, the forecasting data regarding the weather conditions in the area, power
generation both from PV and grid, grid tariff and the power consumption profiles of auxiliary loads
and the EV charging loads are considered for a particular time period. The process of determining
the forecasted data and the grid tariffs were thoroughly discussed in the literature and can be
applicable [47,48].
Once the forecasted data and the grid tariffs were obtained, the objective function is defined.
The defined objective function aims at optimal power scheduling between the power sources and
energy storage for minimizing the daily running cost of the system. In this paper PSO method, as
discussed in this section, is used to define the optimization problem. In this paper, the implementation
of PSO for power flow optimization is considered to be a predictive upper-level optimization stage,
where the process is based on early forecasted data. The system measurements considered for this

28
Energies 2019, 12, 2334

methodology are, PV output power (PPV ), grid power (PG ), load power (PL ), SOC of ESD and EVs,
and grid tariff (GT ).
The power balancing equation for GCPVS with ESD is given by Equation (3), and the state of
charge of the ESD and EVs is estimated using Equation (4):

Pb (t) = PL (t) − Ppv (t) − PG (t), (3)

Pb (t)Δt
SOC(t) = SOC(t − Δt) − , (4)
Q
where Δt is the time interval, and Q is the SOC of battery at the time of power exchange.
The system operation is constrained by the following limits.

SOCmin ≤ SOC(t) ≤ SOCmax , (5)

Pbmin ≤ Pb (t) ≤ Pbmax , (6)

PGmin ≤ PG (t) ≤ PGmax . (7)

The system constraints defined in Equations (5) and (6) are considered for improving the battery
lifetime, while Equation (7) defines the limits for utilizing power from the grid. At any given time (t),
the total cost of the system is expressed as:

CT (t) = CG (t) + CBD (t), (8)

where CT (t) is the total cost of the system, CG (t) is the operating cost of grid CBD (t) is the degradation
cost of the battery.
By considering the above equations, the system objective function for operating the system for a
given time period is given by Equation (9).
t = T
Min(CT ) = Min [CG (t) + CBD (t)], (9)
t=0

where the grid operating cost CG (t) is given in Equation (10):

CG (t) = GT (t)PG (t)Δt. (10)

However, the process of developing the degradation cost function for the energy storage device is
a complex process it involves various independent factors [49]. The three major aspects impacting
the operation, lifetime and health of the ESDs are battery temperature [50], average SOC [51], and the
depth of discharge (DOD) [52].
By considering the temperature impact, both the charging and discharging process affect the
temperatures of the ESD. This phenomenon directly affects the average power (Pavg) drawn from or to
the battery as given in Equation (11). In order to account for temperature variations in both the charging
and the discharging scenarios the absolute of average temperature is considered in the equation.

T = Tamb + Rth Pavg , (11)

where T is operating temperature of the battery, Tamb is the battery pack ambient temperature, RTh is
the thermal resistance of the battery pack.

29
Energies 2019, 12, 2334

The effect of battery temperature on the lifetime of the battery is calculated from the Arrhenius
equation [53].
bQ
L(T ) = aQe T , (12)

where a and b correspond to the curve fitting parameters.

The impact of temperature on the battery degradation cost is given by Equation (13)

tf
dt
Ctemp = Cbat , (13)
ti Ti L(T )

where Ti is the number of intervals in the total time period, and Cbat is the initial cost of the battery.
Apart from the temperature impact, average SOC of the battery also impacts the degradation
rate of the battery. A firm relation between the average SOC and battery degradation cost is given by
Equation (14):
m SOCavg − d
Csoc = Cbat , (14)
Q f ade n Ti
where m, d and n are constants of curve fitting, and Q f ade is the capacity fade at the end of the
battery lifetime.
Finally, a relation between the lifetime of the energy storage device and the DOD of the battery
are presented in References [54,55]. Equation (15) depicts the cost per kWh of the battery DOD.

Cbat
CDOD = , (15)
2 N DOD Q μ2

where N is the battery lifecycle for a particular DOD, and μ depicts the charging and discharging
efficiency of the battery.
Considering all the factors that impact the degradation of the battery, the cost function for
battery degradation is given by Equation (16) and the final objective function is achieved using (17)
after discretization, 
Cbd = max Ctemp , CDOD , CSOCavg , (16)


Ni
CTi , run = [CG (k) + Pb (k)ΔtCbd (k)]. (17)
k =1

The proposed methodology for optimizing the power flow contributes to the improvement of the
lifetime of the system components, especially the storage devices. The PSO implementation for the
optimal selection of different power sources is illustrated in Figure 6.

30
Energies 2019, 12, 2334

Figure 6. Implementation of particle swarm optimization (PSO) for the optimal selection of power source.

31
Energies 2019, 12, 2334

6. Experimental Work
A simulation system model, containing the realistic models for the electric network and the EV,
was developed to test the developed charging strategy and study its effects on the electric network. A
simplified illustration of the whole simulation system is presented in Figure 7.

Figure 7. Simulink implementation of grid-connected PV–EV (GCPE) charging architecture with energy
storage device (ESD) and intelligent energy management system (iEMS).

These models are developed in the MATLAB/Simulink environment. The main script takes
data on driving distances and solar irradiance as input parameters. It then calculates the amount of
power that should be supplied to the EVs and the electricity grid respectively each hour, in order for
the self-consumption to be maximized and grid impact to be reduced. The parameters used for the
simulation of the proposed system is given in Table 1.

Table 1. Simulation parameters of grid-connected PV–EV charging station with an energy storage unit.

Components Rating
Solar Array 20 kW
Boost Converter 5 kHz, 500 V
DC link Voltage 500 V
Grid Parameters 2500 MVA, 120 kV
High Frequency Transformer 25 kva, 5 kHz, 500:80:80, three winding transformer
Electric Vehicle Mahindra severity
Lithium Battery
EV Battery Nominal Voltage = 72 V
Rated Capacity = 200 Ah
Rated Power = 150 kW
Energy Storage Device
Power Conversion Efficiency = 90%

32
Energies 2019, 12, 2334

6.1. Electric Network Model

The used electric network model represents a typical grid-connected PV system with an energy
storage unit. The details for this synthetic electric network model are taken from MathWorks [56]. The
schematic of the developed electric network is presented in Figure 7. The developed system consists
of a 20 kW solar array connected to a 5 kHz boost converter with incremental conductance MPPT
algorithm. The PV network is connected to the grid via a three-phase voltage source converter (VSC).
The VSC can be operated both in inverter and rectifier mode in order to facilitate the power flow both
from the grid to charging stations and from PV to the grid. A 25-kVA high-frequency transformer is
used to provide isolation between GCPV and the EV charging, ESD unit.

6.2. EV Charging Model

The EVs can be modeled as systems composed of many dynamic dependencies. The key
components in the EV models are the charger and the battery. The EV model used in this research is
sourced from Reference [57]. It consists of blocks for the driving schedules, charger logic, converter
losses and the battery model as presented in Figure 8. The driving schedule block collects and passes
on the data regarding the EV ID, EV address and the EV demand DoDEV,n . The EV demand describes
the Depth of Discharge (DoD) of the nth EV. The EV ID specifies which of the EVs are connected and
the address specifies the household node for each EV. As soon as an EV is connected to the electric
network, the driving schedule block relays the details for the charger logic block and updates the
battery DoD information to the battery model block.

Figure 8. Architecture of EV charging model.

The charger logic block receives data from four sources. The first input comes from the battery
model block, which continuously updates the charger logic block with the current SoCEV,n . The second
input comes from the EV fleet aggregator who submits the EV schedules. These schedules are directly
used by the charger logic block. The third input receives the current voltages from the electric network
model. The fourth input comes from the driving schedule. The charger’s main function is to convert
the AC power into DC power, which is essential to charge the battery. The charger logic block processes
the received data and regulates the battery charging in a manner that the local network voltage limits,
set for each EV separately, are not violated. It also computes the appropriate charging power rates as
they are dependent on the batteries’ SoCEV,n levels, as explained in Reference [57].

6.3. Modes of Operation

The model is built on the idea of matching hourly charging powers and PV power generation,
meeting daily charging demands and achieving constant power exchange with the electricity grid.
Since the driving distance for each EV is randomly selected each day, the charging demand of the
car park is unique for each day of the simulated year. The level of self-sufficiency depends on the
amount of installed PV power. If the accumulated PV energy generation for one day equals the car
park demand, all PV generated power is supplied directly to the EVs. If the accumulated PV energy
generation falls short of the daily car park demand, the deficit is supplied from the electricity grid,
evenly distributed over all workday hours. If the daily PV energy generation exceeds the car park
demand; the excess power is supplied to the electricity grid as evenly distributed as possible. In this

33
Energies 2019, 12, 2334

case, where the PV system generates excess energy, the power exchange with the electricity grid is
different each hour, depending on whether the hourly PV energy generation is higher or lower than
the mean excess energy generation. The complete scenario of charging the EVS by considering all the
constraints is explained in nine different modes, as shown in Figure 9.

Figure 9. Modes of operation for GCPE charging architecture with an ESD and iEMS.

Mode 1: PV generation full, EV load Available and ESD has less charge.
In this scenario, the PV generation is happening under standard test conditions and considered
to generate maximum power. Since all the systems EV1, EV2 and ESD act as loads and are ready to
charge, and the power demanded is greater than the power generated from the PV. In this condition,
the grid supports the PV in satisfying the load.
Mode 2: PV generation partial, EV load Available and ESD has less charge.
In this scenario, the PV generation is considered to happen under varying irradiance conditions.
Here the power generated from the PV is low. Since all the systems EV1, EV2 and ESD act as loads and
are ready to charge, and the power demanded is greater than the power generated from the PV. In this
condition, the grid supports the PV in satisfying the load.
Mode 3: PV generation partial, EV load Available and ESD is fully charged.
In this scenario, the PV generation is considered to happen under varying irradiance conditions.
Here the power generated from the PV is low. Since both EV1 and EV2 act as loads and are ready to
charge, and the power demanded is greater than the power generated from the PV. In this condition,
both the grid and ESD supports the PV in satisfying the load depending upon the requirement of
the user.

34
Energies 2019, 12, 2334

Mode 4: No PV generation, EV load Available and ESD is fully charged.

In this scenario, the PV generation is zero. Since both the EVs are ready to charge, and the power
demanded is greater than the power generated from the PV and power available from the battery. In
this condition, the grid supports the ESD in satisfying the load and PV remains disconnected from
the system.
Mode 5: No PV generation, EV load Available and ESD has less charge.
In this scenario, the grid satisfies the changing needs of the EV station. The PV remains
disconnected from the charging dock, and once the charging needs of the EVs are satisfied, and the
load connected to the utility is less, then the grid charges the ESD.
Mode 6: PV generation full, Partial EV load Available and ESD has less charge.
In this scenario, the PV generation is happening under standard test conditions and considered to
generate maximum power. Since the ESD has less charge, and EV load available is less and ready to
charge, the PV satisfies the EV load first and then charges the ESD.
Mode 7: PV generation partial, Partial EV load Available and ESD has less charge.
In this scenario, the PV generation is considered to happen under varying irradiance conditions.
Here the power generated from the PV is low. Since the ESD has less charge, EV load available is less
and ready to charge, and the power demanded is greater than the power generated from the PV. In
this condition, PV satisfies the EV load first and then charges the ESD depending upon the generating
capability of the PV.
Mode 8: No PV generation, No EV load Available and ESD is fully charged.
In this scenario, the grid satisfies the utilities connected to it. The ESD remains connected with the
charging docks.
Mode 9: PV generation full, No EV load Available and ESD is fully charged.
In this scenario, the PV generation is happening under standard test conditions and considered to
generate maximum power. Since the EV load is not available and the ESD is fully charged, the PV
transfers its power to the grid.
By bearing these scenarios in mind, the model, considered above, is simulated to observe the
efficiency of the proposed charging and converter control strategy. The optimal power flow is achieved
by adapting the cost function achieved in Equation (17). The forecasting data to perform the experiment
is considered as follows:
The irradiance profile for performing the experiment is considered as depicted in Figure 10.

Figure 10. Irradiance variation for PV array.

From Figure 10, it is observed that there is variation in irradiance at multiple instances, this is
considered to replicate the behavior of PV system in real-time. By considering the real-time operation,
the PV systems depict five times the difference in energy yield for change in environmental conditions.
Due to this change, the ratings of the PV converter can be resized by 30%, which corresponds to a 3.2%
loss of energy [58]. Apart from the change in irradiance, the impact of using two stage conversion
for grid-connected PV system, the switching frequencies, the high frequency DC-DC converter at the
charging end and the impact of varying loads in the system also contribute for the power loss in the
system. A detailed explanation on power loss in grid-connected PV system and the impact of EV/ESD
charging and discharging in a grid-connected system were given in References [59–61]. Since the

35
Energies 2019, 12, 2334

system design considered for this research investigates the best design for grid-connected PV system
to achieve efficient charging of EV and ESD, and the power flow in the considered to be unidirectional,
the power loss calculation is neglected.
Apart from the irradiance profile, the major assumptions considered for this research are as follow:
The SOC of ESD is 10%and it is considered to be in charging mode and hence it acts as a load, and
the SOC of EV1 is 40%, and SOC of EV 2 is 30% and were connected with the GCPV. The process of
implementing PSO for developing the proposed methodology as per the forecasted data is given in the
following steps.
Initialization of PSO:
Objective: For charging the vehicle finding an optimal power source using Equation (17).
Topology: Swarm optimization star configuration is used to ensure that all particle communicates
with each other.
Definition of particles: Depending upon the power capacity of PV and ESD, particles are defined.
Fitness Function: The equation is used to maximize and minimize the power generated during the
charging and discharging process.
Search Space: (Constraints) The major constraints to be considered for this problem is that the
hours must be a positive real number, which is limited by arrival and departure time.
PSO parameters: The PSO parameters tested for this control are depicted in Table 2.

Table 2. Tested PSO parameters.

Parameters Value
No of particles in swarm 24
No of iterations 30
Parameters to be identified 2
Search Space Range [0 50; 0 10];
Swarm declaration Zero
Velocity Clamping [3; 1];

The best performance of the proposed switching strategy as per the power available is obtained
from the PSO implemented and the corresponding results are depicted in Figure 11.


(a) (b)

Figure 11. Best switching strategy as per the power available. (a) Best position for feeding the load; (b)
Best response of the algorithm.

36
Energies 2019, 12, 2334

Figure 11a shows the best position for feeding the load continuously as per the power available
from different sources. The initial position defines the power generation capacity of various sources at
the time of interaction with the load. Whereas, the current position defines the source feeding the load
irrespective of the availability of the source or the status of the load, and the best position is defined
based on the load power and generation capacity. Figure 11b depicts the best response of the PSO in
finding the optimal power flow path for satisfying the loads connected at the DC link.
Considering the optimal power flow path defined by the PSO for the forecasted data, the
experimental analysis is carried out as follows:
Initially, the PV generation is set to high by operating it under STC. During this period the PV
is able to charge the EVs with rated power without any support from the grid. After 0.6 s, the PV
generation is set to zero by varying the irradiance of the system. Here the grid comes into the scenario,
in order to support the EV charging. At 2.4 s, the PV generation is minimized to generate 20% of the
rated power by varying the irradiance of the system, as shown in Figure 10. In this case, the grid
continuously supports the charging process of the system. The PSO here monitors the generation and
load profile of both the grid and EV charging docks in order to provide fast switching between the
sources as depicted in Figure 12.

Figure 12. Availability of PV and grid for charging of EVs.

The voltage at the DC link is maintained constant around 500 V, as shown in Figure 13.

Figure 13. DC link voltage of grid-connected PV system.

A high-frequency transformer is used to provide isolation between charging docks, ESD and the
grid-connected PV system. Finally, as per the specification of the vehicle mentioned in the section,

37
Energies 2019, 12, 2334

the EVs were able to charge at rated voltage and current even with disturbances in PV generation, as
shown in Figure 14a,b.

(a) (b)

Figure 14. Battery characteristics while charging for electric vehicles (a) EV1 and (b) EV2.

From the results, it can be observed that the developed power flow optimization algorithm,
effectively identifies the availability of power generated by various sources and schedules them as per
the load requirement. The developed algorithm achieves fast switching between various sources in
order to satisfy the loads continuously without any interruption. Further, the developed algorithm
can also be adapted for improving the lifetime of the energy storage device, which is integrated with
the system.

7. Conclusions
In this paper, an optimized power flow algorithm for grid-connected photovoltaic system with an
energy storage device for charging of electric vehicles is realized. The following objectives are achieved
throughout the paper:

1. The need to adopt renewable energy systems for the charging of electric vehicles is studied.
2. The advantage of using an energy storage device with grid integrated PV systems for charging
EVs is depicted.
3. The major role of converters in achieving optimal and bidirectional power flow between various
sources and loads is realized, and a multiport converter is designed for achieving the objective.
4. The need for an intelligent-energy management system for operating the power converter as per
the availability of power from the sources and demand on the load side is achieved by developing
an optimal power flow algorithm.
5. The developed algorithm is tested with a simulation setup by observing the various modes of
operation on the system.

All results and conclusions are based on the assumption that the model created in MATLAB is
implemented as a system that adjusts EV charging. A grid-connected PV–EV charging system with an
energy storage device is developed in this paper. A multiport converter is designed to accommodate
efficient power flow between multiple sources and the loads. The developed converter is controlled
by adapting particle swarm optimization, which acts as an intelligent energy management system

38
Energies 2019, 12, 2334

for dynamic power management at the charging station. An objective function is formulated for
scheduling optimal power flow between various sources and loads in the system. The simulation
results depict that the new control strategy efficiently shifts between the sources within a stipulated
time and continuously supports the charging station.

Author Contributions: Conceptualization, methodology, software, validation, and formal analysis, M.J., V.S.B.K.
and M.A.K.; Investigation, A.H. and F.B.; Resources, and data curation, M.J.; writing—original draft preparation,
writing—review and editing, and visualization, V.S.B.K. and M.A.K.; supervision, A.H. and F.B.; project
administration, A.H. and F.B.
Funding: This research received no external funding.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Chan, C.C.; Wong, Y.S. Electric Vehicles Charge Forward. IEEE Power Energy Mag. 2004, 2, 24–33. [CrossRef]
2. Hidrue, M.K.; Parsons, G.R.; Kempton, W.; Gardner, M.P. Willingness to pay for electric vehicles and their
attributes. Resour. Energy Econ. 2011, 33, 686–705. [CrossRef]
3. Yilmaz, M.; Krein, P.T. Review of battery charger topologies, charging power levels, and infrastructure for
plug-in electric and hybrid vehicles. IEEE Trans. Power Electron. 2013, 28, 2151–2169. [CrossRef]
4. Fairley, P. Speed bumps ahead for electric-vehicle charging. IEEE Spectr. 2010, 47, 13–14. [CrossRef]
5. Clement-Nyns, K.; Haesen, E.; Driesen, J. The impact of Charging plug-in hybrid electric vehicles on a
residential distribution grid. IEEE Trans. Power Syst. 2010, 25, 371–380. [CrossRef]
6. Yunus, K.; la Parra, H.Z.D.; Reza, M. Distribution Grid Impact of Plug-In Electric Vehicles Charging at Fast
Charging Stations Using Stochastic Charging Model. In Proceedings of the 2011 14th European Conference
on Power Electronics and Applications, Birmingham, UK, 30 August–1 September 2011; pp. 1–11.
7. Botsford, C.; Szczepanek, A. Fast Charging vs. Slow Charging: Pros and cons for the New Age of Electric
Vehicles. In Proceedings of the EVS24 International Battery, Hybrid and Fuel Cell Electric Vehicle Symposium,
Stavanger, Norway, 13–16 May 2009.
8. Gnann, T.; Funke, S.; Jakobsson, N.; Plotz, P.; Sprei, F.; Bennehag, A. Fast charging infrastructure for electric
vehicles: Today’s situation and future needs. Transp. Res. Part D Transp. Environ. 2018, 62, 314–329.
[CrossRef]
9. Morrow, K.; Karner, D.; Francfort, J. Plug-in Hybrid Electric Vehicle Charging Infrastructure Review (U. S.
Department of Energy Vehicle Technologies Program—Advanced Vehicle Testing Activity); Idaho National Laboratory
(INL): Idaho Falls, ID, USA, 2008.
10. Deb, S.; Tammi, K.; Kalita, K.; Mahanta, P. Review of recent trends in charging infrastructure planning for
electric vehicles. Wiley Interdiscip. Rev. Energy Environ. 2018, 7, e306. [CrossRef]
11. Dickerman, L.; Harrison, J. A New Car, a New Grid. IEEE Power Energy Mag. 2010, 8, 55–61. [CrossRef]
12. Badea, G.; Felseghi, R.A.; Varlam, M.; Filote, C.; Culcer, M.; Iliescu, M.; Răboacă, M. Design and simulation
of romanian solar energy charging station for electric vehicles. Energies 2019, 12, 74. [CrossRef]
13. IEA PVPS. Trends 2015 in Photovoltaic Applications; IEA PVPS: Bangkok, Thailand, 2015.
14. Nehrir, M.H.; Wang, C.; Strunz, K.; Aki, H.; Ramakumar, R.; Bing, J.; Miao, Z.; Salameh, Z. A review of
hybrid renewable/alternative energy systems for electric power generation: Configurations, control, and
applications. IEEE Trans. Sustain. Energy 2011, 2, 392–403. [CrossRef]
15. Strunz, K.; Abbasi, E.; Huu, D.N. DC microgrid for wind and solar power integration. IEEE J. Emerg. Sel. Top.
Power Electron. 2014, 2, 115–126. [CrossRef]
16. Cingoz, F.; Elrayyah, A.; Sozer, Y. Optimized Resource Management for PV-Fuel-Cell-Based Microgrids
Using Load Characterizations. IEEE Trans. Ind. Appl. 2016, 52, 1723–1735. [CrossRef]
17. Varaha, K.; Bharath, S.; Khan, M.A. Applications of Artificial Intelligence Techniques in Engineering 698; Springer:
Singapore, 2019.
18. Kurukuru, V.S.B.; Khan, M.A.; Singh, R. Performance optimization of UPFC assisted hybrid power system.
In Proceedings of the 2018 IEEMA Engineer Infinite Conference (eTechNxT), New Delhi, India, 13–14 March
2018; pp. 1–6.

39
Energies 2019, 12, 2334

19. Eren, Y.; Erdinc, O.; Gorgun, H.; Uzunoglu, M.; Vural, B. A fuzzy logic based supervisory controller for an
FC/UC hybrid vehicular power system. Int. J. Hydrog. Energy 2009, 34, 8681–8694. [CrossRef]
20. Badawy, M.O.; Sozer, Y. Power Flow Management of a Grid Tied PV-Battery System for Electric Vehicles
Charging. IEEE Trans. Ind. Appl. 2017, 53, 1347–1357. [CrossRef]
21. Carbone, R. Grid-connected photovoltaic systems with energy storage. In Proceedings of the 2009
International Conference on Clean Electrical Power (ICCEP), Capri, Italy, 9–11 June 2009; pp. 760–767.
22. Beltran, H.; Pérez, E.; Aparicio, N.; Rodriguez, P. Daily solar energy estimation for minimizing energy storage
requirements in PV power plants. IEEE Trans. Sustain. Energy 2013, 4, 474–481. [CrossRef]
23. F. OECD/IEA. Medium-Term Market Report/Aut; IEA: Paris, France, 2014.
24. Deb, S.; Tammi, K.; Kalita, K.; Mahanta, P. Impact of electric vehicle charging station load on distribution
network. Energies 2018, 11, 178. [CrossRef]
25. Bauer, P.; Zhou, Y.; Doppler, J.; Stembridge, N. Charging of Electric Vehicles and Impact on the Grid. In
Proceedings of the 13th Mechatronika 2010, Trencianske Teplice, Slovakia, 2–4 June 2010; pp. 121–127.
26. SAE. SAE Electric Vehicle and Plug-in Hybrid Electric Vehicle Conductive Charge Coupler J1772; SAE: Warrendale,
PA, USA; Troy, MI, USA, 2010.
27. IEC 62196. Plugs, Socket-Outlets, Vehicle Connectors and Vehicle Inlets—Conductive Charging of Electric
Vehicles—Part 1, 2, 3; International Electrotechnical Commission: Geneva, Switzerland, 2014.
28. IEC 61851. Electric Vehicle Conductive Charging System—Part 1, 21, 23, 24; International Electrotechnical
Commission: Geneva, Switzerland, 2014.
29. Kubel, M. Design Guide for Combined Charging System; CharIN e.V: Berlin, Germany, 2015.
30. CHA Association. Technical Specifications of Quick Charger for the Electric Vehicle; CHA Association: Washington,
DC, USA, 2010.
31. Tesla EV Charging and Supercharging Technique. 2019. Available online: www.teslamotors.com (accessed
on 24 April 2019).
32. Messagie, M.; Boureima, F.-S.; Coosemans, T.; Macharis, C.; van Mierlo, J. A Range-Based Vehicle Life Cycle
Assessment Incorporating Variability in the Environmental Assessment of Different Vehicle Technologies
and Fuels. Energies 2014, 7, 1467–1482. [CrossRef]
33. Messenger, R.A.; Abtahi, A. Photovoltaic Systems Engineering; CRC Press: Boca Raton, FL, USA, 2017.
34. Stapleton, G.; Neill, S. Grid-Connected Solar Electric Systems: The Earthscan Expert Handbook for Planning, Design
and Installation; Routledge: Abingdon, UK, 2011.
35. Carli, G.; Williamson, S.S. Technical Considerations on Power Conversion for Electric and Plug-in Hybrid
Electric Vehicle Battery Charging in Photovoltaic Installations. IEEE Trans. Power Electron. 2013, 28, 5784–5792.
[CrossRef]
36. Rasin, Z.; Rahman, M.F.; Teknikal, U. Grid-connected Quasi-Z-Source PV Inverter for ElectricVehicle Charging
Station. In Proceedings of the International Conference on Renewable Energy Research and Applications,
Madrid, Spain, 20–23 October 2013; pp. 627–632.
37. Rasin, Z.; Ahsanullah, K.; Rahman, M.F. Design and Simulation of Quasi-Z-Source Grid-connected PV
Inverter with Bidirectional Power Flow for Battery Storage Management. In Proceedings of the IECON
2013-39th Annual Conference of the IEEE Industrial Electronics Society, Vienna, Austria, 10–13 November
2013; pp. 1589–1594.
38. Rydh, C.J.; Sandén, B.A. Energy analysis of batteries in photovoltaic systems. Part II: Energy return factors
and overall battery efficiencies. Energy Convers. Manag. 2005, 46, 1980–2000. [CrossRef]
39. Chiang, S.J.; Chang, K.T.; Yen, C.Y. Residential Photovoltaic Energy Storage System. IEEE Trans. Ind. Electron.
1998, 45, 385–394. [CrossRef]
40. Riffonneau, Y.; Bacha, S.; Barruel, F.; Ploix, S. Optimal power flow management for grid connected PV
systems with batteries. IEEE Trans. Sustain. Energy 2011, 2, 309–320. [CrossRef]
41. Bagula, A.; Castelli, L.; Zennaro, M. On the Design of Smart Parking Networks in the Smart Cities: An
Optimal Sensor Placement Model. Sensors 2015, 15, 15443–15467. [CrossRef] [PubMed]
42. Macal, C.M.; North, M.J. Tutorial on Agent-based Modeling and Simulation. In Proceedings of the 37th
Winter Simulation Conference, Orlando, FL, USA, 4–7 December 2005; pp. 2–15. [CrossRef]
43. Hutson, C.; Venayagamoorthy, G.K.; Corzine, K.A. Intelligent Scheduling of Hybrid and Electric Vehicle
Storage Capacity in a Parking Lot for Profit Maximization in Grid Power Transactions. In Proceedings of the
2008 IEEE Energy 2030 Conference, Atlanta, GA, USA, 17–18 November 2008; pp. 1–8.

40
Energies 2019, 12, 2334

44. Kennedy, J. Particle Swarm Optimization. In Encyclopedia of Machine Learning; Springer: Berlin, Germany,
2010; pp. 745–817.
45. Valle, Y.; Venayagamoorthy, G.K.; Mohagheghi, S.; Jean-Carlos Hernandez, R.G. Harley, Particle Swarm
Optimization: Basic Concepts, Variants and Applications in Power Systems. IEEE Trans. Evol. Comput. 2008,
12, 171–195. [CrossRef]
46. Mahindra Everito. Available online: http://www.mahindraverito.com/everito/discover-the-new-verito.aspx
(accessed on 5 May 2019).
47. Li, G.; Liu, C.C.; Mattson, C.; Lawarrée, J. Day-ahead electricity price forecasting in a grid environment. IEEE
Trans. Power Syst. 2007, 22, 266–274. [CrossRef]
48. Pindoriya, N.M.; Singh, S.N.; Singh, S.K. An adaptive wavelet neural network-based energy price forecasting
in electricity markets. IEEE Trans. Power Syst. 2008, 23, 1423–1432. [CrossRef]
49. Mazadi, M.; Rosehart, W.; Zareipour, H. Impact of Wind Generation on Electricity Markets: A
Chance-Constrained Nash Cournot model. Iran. J. Sci. Technol. Trans. Electr. Eng. 2012, 36, 51–66.
50. Hoke, A.; Brissette, A.; Smith, K.; Pratt, A.; Maksimovic, D. Accounting for lithium-ion battery degradation
in electric vehicle charging optimization. IEEE J. Emerg. Sel. Top. Power Electron. 2014, 2, 691–700. [CrossRef]
51. Motamedi, A.; Zareipour, H.; Rosehart, W.D. Electricity price and demand forecasting in smart grids.
IEEE Trans. Smart Grid 2012, 3, 664–674. [CrossRef]
52. Liu, S.; Wang, J.; Liu, Q.; Tang, J.; Liu, H.; Fang, Z. Deep-Discharging Li-Ion Battery State of Charge Estimation
Using a Partial Adaptive Forgetting Factors Least Square Method. IEEE Access 2019, 7, 47339–47352.
[CrossRef]
53. Yang, Y.; Hu, X.; Qing, D.; Chen, F. Arrhenius equation-based cell-health assessment: Application to thermal
energy management design of a HEV NiMH battery pack. Energies 2013, 6, 2709–2725. [CrossRef]
54. Markel, T.; Smith, K.; Pesaran, A.A. Improving Petroleum Displacement Potential of PHEVs Using Enhanced
Charging Scenarios; Elsevier: Amsterdam, The Netherlands, 2010.
55. Xu, B. Degradation-Limiting Optimization of Battery Energy Storage Systems Operation; Location Power Systems
Laboratory ETH Zurich: Zürich, Switzerland, 2013.
56. Detailed Model of a 100-kW Grid-Connected PV Array—MATLAB & Simulink—MathWorks France.
Available online: https://fr.mathworks.com/help/physmod/sps/examples/detailed-model-of-a-100-kw-grid-
connected-pv-array.html (accessed on 5 May 2019).
57. Ma, C.; Marten, F.; Töbermann, J.; Braun, M.; Iwes, F. Evaluation of Modeling and Simulation Complexity on
Studying the Impacts of Electrical Vehicles Fleets in Distribution Systems. In Proceedings of the 2014 Power
Systems Computation Conference, Wroclaw, Poland, 18–22 August 2014.
58. Mouli, G.R.C. Charging Electric Vehicles from Solar Energy: Power Converter, Charging Algorithm and
System Design. Ph.D. Thesis, Delft University of Technology, Delft, The Netherlands, 2018.
59. Mouli, G.R.C.; Bauer, P.; Zeman, M. System design for a solar powered electric vehicle charging station for
workplaces. Appl. Energy 2016, 168, 434–443. [CrossRef]
60. Apostolaki-Iosifidou, E.; Codani, P.; Kempton, W. Measurement of power loss during electric vehicle charging
and discharging. Energy 2017, 127, 730–742. [CrossRef]
61. Wu, T.F.; Chang, C.H.; Chang, Y.D.; Lee, K.Y. Power loss analysis of grid connection photovoltaic systems. In
Proceedings of the 2009 International Conference on Power Electronics and Drive Systems (PEDS), Taipei,
Taiwan, 2–5 November 2009; pp. 326–331.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

41
 energies
Article
Integration of Stationary Batteries for Fast Charge EV
Charging Stations
Davide De Simone * and Luigi Piegari
Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano, 20133 Milano, Italy;
luigi.piegari@polimi.it
* Correspondence: davide.desimone@polimi.it

Received: 4 November 2019; Accepted: 4 December 2019; Published: 6 December 2019

Abstract: One of the biggest issues preventing the spread of electric vehicles is the difficulty in
supporting distributed fast charging stations by actual distribution grids. Indeed, a significant
amount of power is required for fast charging, especially if multiple vehicles must be supplied
simultaneously. A possible solution to mitigate this problem is the installation of auxiliary batteries
in the charging station to support the grid during high peak power demands. Nevertheless, the
integration of high-voltage batteries with significant power is not a trivial task. This paper proposes the
configuration and control of a converter to integrate batteries in a fast charging station. The proposed
configuration makes it possible to decouple the grid power from the vehicle power using several
auxiliary battery modules. At the same time, the converter makes it possible to draw different
amounts of power from the battery modules, allowing the use of second life batteries performing in
different ways. This paper discusses the design, control, and operation of the converter. Moreover,
the effectiveness of the proposed control is shown by means of numerical results.

Keywords: battery electric vehicle; BEV; charging station; stationary storage; plug-in; EV; fast charge;
power quality; second life batteries

1. Introduction
Electric vehicles (EVs) are spreading in the private transportation sector. Although these vehicles
contribute less to city pollution, their spread is being hindered by limitations such as their high price,
short range, and long recharge time compared to the equivalent category of internal combustion
vehicles [1]. Battery technologies are being rapidly developed, with steady increases in their energy
and power densities and decreases in their prices [2].
Higher power density batteries can accept higher recharging power levels, allowing a fast charge.
Together with the spread of electric vehicles, the recharging infrastructure is undergoing significant
development in terms of charging points and installed power. A generic EV power profile during
the recharge phase is characterized by a high peak power demand that usually lasts 26–32 min [3].
Standard fast dc charging is performed at 50 kW dc, but multiple charging systems make it possible to
charge at higher power. By 2021, fast charging up to 350 kW might be possible [4]. The most common
charging standards and relative power ratings are summarized in Table 1 [4].
In urban areas, it is likely that many consumers will recharge their vehicles between 7:00 and 9:00,
at noon, and between 18:00 and 20:00 [3,5]. Grid operators might not be able to overcome future peak
power requirements without greatly oversizing the installed power. A possible approach to mitigate
this problem might be to add a battery electric storage system (BESS) to a fast charging station (FSS),
which will act as a buffer to reduce the peak power demand on the network without increasing the EV
charging time.

Energies 2019, 12, 4638; doi:10.3390/en12244638 43 www.mdpi.com/journal/energies

Energies 2019, 12, 4638

Table 1. Charging standards.

System kW Availability
United States
Combined Charging System (CCS) 50–350 European Union
Australia, Korea
China GB/T 237.5 China, India
Tesla Supercharger 135 Global
CHAdeMO 50–100 Global

Two possible approaches could be followed to interface a BESS with a charging station [6]: ac
coupling and dc coupling. In ac coupling, the storage system and charging converters are interconnected
through an ac bus. In dc structures, the interconnection is through a common dc link. Figure 1 shows
schematics of the two coupling methods.

(a) (b)
Figure 1. Charging station coupling methods: (a) ac coupling and (b) dc coupling.

At present, ac coupling is the most widespread because of the maturity of the ac technology
in terms of the converter design, protection devices, and standards. Although ac coupling is more
common, dc coupling provides a higher efficiency and lower cost because this configuration requires
fewer conversion stages.
Modular multilevel converters (MMCs) are widely used in dc/ac medium/high-voltage applications.
Their multilevel output waveform reduces the total harmonic content of the generated ac voltage and,
thus, the harmonics the of ac currents are quite limited also using a very small filter. Depending on the
number of levels and the grid impedance, a filter-less solution can be acceptable [7–10].
Double star chopper cells converters (DSCCs) replace the floating capacitors of traditional MMCs
with a battery. Recently, the use of this converter in an electric vehicle powertrain has been proposed [11].
Embedding DSCCs in EVs offers some major advantages. The storage system is embedded inside
the converter structure, which makes it possible to perform the battery management system (BMS)
operations without dedicated hardware [11]. Moreover, it is possible to balance the batteries within
each arm to exploit the load currents rather than moving energy from one cell to another, potentially
making the process more efficient. The control strategies available in the literature make it possible
to recharge the storage system [12] from any power source, i.e., ac three phase, ac single phase, and
dc. Finally, the modular structure ensures a highly redundant system [13]. The DSCC structure also
seems to be very promising for integrating a storage system in an EV charging station to support the
grid during a peak power demand. Indeed, batteries can be integrated, at the module level, directly
in the converter modules to ensure the correct power sharing between the modules. In this way, the
power converter also operates as a BMS controlling the power exchanged individually by each battery
module. Second life batteries can be used because each battery module is controlled separately from
the others, and the modularity of the solution makes it possible to substitute only one battery module
without the need to substitute all the others. With the proper choice of voltage levels, the dc bus of the
DSCC can be used inside the EV station for fast dc charging.

44
Energies 2019, 12, 4638

This paper proposes the use of a DSCC converter to integrate an auxiliary storage system with an
EV charging station. Moreover, a new control strategy to separately control the power exchanged by
this converter at its dc and ac ports is proposed. This control strategy was simulated in a dc coupled
charging station, and the results obtained show that is possible to meet the charging power demand
by drawing constant power from the three-phase grid. In this way, the power installation required
by the EV charging station is limited and lower than the power required by modern fast dc charging
systems. This mitigates a big problem with modern fast charging systems, facilitating the diffusion of
electric vehicles.

2. Materials and Methods

2.1. Converter Control

The investigated DSCC converter is composed of three phases, each one realized with two arms
embedding six modules (Figure 2). The common points of the three phases are used to obtain a virtual
dc bus. For each arm, the reference voltage is obtained as follows:
⎧ v∗ dc
⎪ ∗ ∗ ∗
⎨ v lower,k = 2 + v phase,k + v circ,k
⎪
⎪
⎪ ∗ (1)
⎩ v∗ v ∗ ∗
upper,k = 2 − v phase,k + v circ,k
dc

where v*lower,k and v*upper,k are, respectively, the phase k lower and upper reference arm voltages. v*dc is
the voltage reference for the dc bus of the converter, v*phase,k is the phase voltage reference, and v* circ,k is
the output of the circulating current controller for phase k used to balance the battery cells of the upper
and lower arms [13]. It is worth noting that the three virtual dc buses can have different reference
voltages:
v∗ dc,k = v∗ dc + 2 · v∗ circ,k (2)

This is done to make it possible to balance the state of charge (SOC) of the battery cells of the three
legs of the converter.
The modulation strategy must be able to control the dc bus voltage reference. In literature different
modulation techniques for this converter topology are available. The modulation techniques can be
divided in two categories: nearest level control technique (NLC) and pulse width modulation (PWM)
based ones [14]. NLC approximates the output voltage by turning on and off modules in order to
minimize the discretization error related to the voltages of the modules, while PWM techniques add a
switching signal to the output of NLC. As a consequence, NLC has a discretization error on the dc bus
voltage equal to one module voltage. Therefore, if the number of levels is not very high, it can be used
neither for a precise control of the dc bus voltage nor for proper balancing among the phases. This
is the reason why, in this paper a PWM based technique is used. Given the references for each arm
voltage, the control signals of the converter modules are obtained using phase disposition pulse width
modulation (PD-PWM) [15]. It is worth noting that the results of this paper can be obtained with any
PWM based technique. The choice of the PD-PWM is only based on its simplicity of implementation.
This modulation consists of comparing the references with multiple carriers, where the amplitude is
the voltage module, translated from the voltages of the other connected modules to avoid overlapping
among them, as shown qualitatively in Figure 3.
As the voltages of the battery modules are never perfectly matched, circulating currents might
flow among the phases. Given the arm currents, referring to the symbols of Figure 2, the circulating
currents (icirc,k ) and phase currents (iphase,k ) are defined as follows:
⎧  
⎪
⎨ icirc,k = 2 iupper,k + ilower,k
⎪ 1
⎪
⎪ (3)
⎩ iphase,k = iupper,k − ilower,k

45
Energies 2019, 12, 4638

where iupper,k and ilower,k are the phase k arm currents. It is worth noting that the dc component of
the circulating current is used to balance the batteries among the phases, while energy is exchanged
between the upper and lower arms using the second harmonic component of the same circulating
current. A dedicated circulating current controller is implemented to allow battery balancing among
the arms and phases [13]. The leg and arm balance is achieved following [16]. The logic scheme of the
circulating current controller is shown in Figure 4, where vph is the space vector of the phase voltages
generated by the converter; i*circ is the reference space vector of the circulating currents; and v*circ,a ,
v*circ,b , and v*circ,c are the dc bus references for each phase. In practice, the balancing is obtained with
three different actions:

1. Balancing among battery modules belonging to same arm is achieved by sorting the cells on
the basis of their SOC and using the most charged (discharged) when the current is discharging
(charging) them;
2. Balancing among battery modules belonging to different legs is achieved by changing the dc
reference voltage of each leg allowing the circulation of a dc current among the legs (this current
does not interest the load);
3. Balancing among upper and lower arms is achieved using an oscillating zero-sequence current
(equal on each phase) that takes more energy from the lower (upper) arm if this is the more charged.

Figure 2. Six modules per arm of the double star chopper cells converter (DSCC).

46
Energies 2019, 12, 4638

Figure 3. Phase disposition PWM (PD-PWM) qualitative modulation. Solid line: upper arm reference,
dashed line: lower arm reference.

Figure 4. Leg and arm balance controllers. SOCs are all the modules’ state of charge values. “Leg
unbalance” computes the mean SOC of each phase and compares the results. “Arm unbalance”
computes the imbalance between the upper and lower arms for each phase.

This balancing technique, anyway, is widely addressed in [16] and, therefore, it will not analyzed
in more details in this paper.
In order to control the dc power output of the converter, a dedicated controller acts on the average
dc bus voltage of the converter: reference dc current i*dc is computed given reference dc power P*dc
and dc bus voltage vdc . The difference between the reference and actual dc current, idc , is used for
the input of the dc current regulator acting on the converter virtual dc bus voltage reference v*dc , as
reported in Figure 5.

Figure 5. dc power regulator.

The ac power is controlled on a reference frame oriented on the grid voltage by means of a phase
locked loop (PLL). On this reference frame, the controls of the active and reactive power are decoupled.
Indeed, the direct axis oriented on the voltage space vector is indicated with “d” and the axis with a
90◦ lead is shown by “q”, where the quadrature component of the voltage is equal to zero, resulting in
the following [17,18]:
Pac = 32 vd id
(4)
Qac = 32 vd iq

47
Energies 2019, 12, 4638

Therefore, the control of the active power is obtained by dividing the active power reference P∗ac
by the direct component of the grid voltage vector, obtaining the direct current reference i∗d . Using a
multilevel converter, the filter impedance to connect to the ac grid is very low, and it is possible to
control the direct and quadrature components of the current by directly controlling, respectively, the
direct and quadrature components of the voltage produced by the converter.
The position of the voltage grid, vph , is then used to obtain the reference for the space vector of
the voltage to be used for the converter. The simplified control scheme is reported in Figure 6.

Figure 6. Active and reactive power regulator.

In [3] the charging station exchanges reactive power with the grid to provide grid services and
increasing the profitability of the installation. If no requirements is given for the reactive power, its
reference is usually set to zero to exchange power at unitary power factor minimizing the current and,
consequently, the losses. A zero reference for the reactive power is assumed for this paper.

2.2. Converter Sizing

Acting separately on the ac and dc power references, it is possible to simultaneously control the
dc and ac active power levels. The difference between the two is automatically stored or supplied by
the storage system. To simultaneously control the two power levels, the voltage levels of the converter
must be chosen so that the two controllers do not interact with each other. In particular, the lowest
voltage on the dc bus to regulate the power flows must be higher than the EV battery voltages and, at
the same time, it has to be higher than twice the highest voltage that is required on the ac side.
Allowing a voltage variation on the ac side of ±10% and considering an additional ±10% margin
to allow the control of the reactive power, the ac output voltage of the converter has to be able to reach
a value that is 20% higher than the rated ac voltage. As a consequence, to allow the full controllability
on the ac side, the minimum dc voltage vdc,min is as follows:

vdc,min = 2 · 1.2 · vph,rated (5)

where vph,rated is the peak of the rated grid voltage. If a ±10% margin is considered for the regulation of
the dc power and the balancing among the phases, the minimum module voltage vmodule,min can be
calculated as follows:
1.1vdc,min
vmodule,min = (6)
nmodules
where nmodules is the number of modules per arm. In [3], a charging station with a storage system
was dimensioned so that the batteries’ SOC did not go below 20% under normal working conditions.
The minimum number of cells in the stack is thus obtained by applying the following:
 
vmodule,min
nstack = (7)
vcell,min

where vcell,min is the minimum voltage achieved by one battery cell at the lowest SOC limit.
In this article, three vehicles of different categories are considered: the BMW i3 (EV 1 ), belonging
to the city car category; the Nissan Leaf e+ (EV 2 ), representing a medium size vehicle; and finally, the

48
Energies 2019, 12, 4638

Porsche Taycan (EV 3 ), which is a sports car. Their storage systems and charging power specifications
are summarized in Table 2 [4,19–21].

Table 2. Considered vehicles’ specifications.

Vehicle Usable Battery Capacity (kWh) Charging Power (kW)

EV 1 BMW i3 37.9 46
EV 2 Nissan Leaf e+ 56 100
EV 3 Porsche Taycan 83.7 350

The charging station is supposed to have one fast charging dc port for the vehicles. It draws power
from a 100 kW three-phase low-voltage grid. Each vehicle is supposed to arrive at the charging spot
with a 10% SOC. The end of charge is considered at a SOC of 90%. As shown in [3], the average time,
tswap necessary to connect the charging station to the next customer is 10 min. To grant the maximum
charging power to the vehicles, the stationary storage system is dimensioned to be able to recharge the
medium size and large EVs in a row. The energy that must be stored is given by the energy balance
between the network and the vehicles:
    
1 1 EEV,2 EEV,3 tswap
Estorage = · 0.8 · (EEV,2 + EEV,3 ) − Pnetwork · 0.8 · + + (8)
0.6 η PEV,2 PEV,3 60

where 0.6 represents the exploited energy of the stationary storage (20%–80%); η is the global efficiency
of the storage and converter, which is supposed to be 90%; EEV,k and PEV,k are, respectively, the battery
rated energy in Wh and the charging power in W of the k vehicle, while tswap is expressed in minutes.
The power exchange with the grid is managed in order to keep the stationary storage system SOC
below 80% and to not overcome the power limit of 100 kW. If the vehicle is being charged at less than
100 kW and the stationary storage reaches 80%, the network power is reduced to match the vehicle
power demand. The dc power controller instead is programmed to prevent the storage system from
reaching a SOC lower than 20%. If a vehicle whose charging power is higher than the grid power is
connected, the dc power will be the maximum allowed until the storage system reaches a minimum
SOC (SOCmin ) of 20%; then the dc power is limited to the power from the grid. To generate the dc
power reference, the stationary battery SOCmin is compared with the minimum allowed SOC of 20%.
The result is given as an input to a regulator whose output is saturated between the minimum and
maximum dc power levels of the installed DSCC. The result is that the maximum dc power Pdc,max is
compatible with the current SOCmin . The dc power required by the vehicle is then saturated between 0
and Pdc,max . The schematic of the minimum SOC regulator is reported in Figure 7.

Figure 7. Min SOC regulator.

The ac power reference is generated by comparing the average SOC with the maximum allowed
SOC of 80%. The result is given to a regulator whose output is saturated between 0 and the maximum
ac power, i.e., 100 kW (Figure 8).

49
Energies 2019, 12, 4638

Figure 8. Max SOC regulator.

2.3. Test Setup

The proposed control strategy was tested by simulating the continuous connection of the vehicles
one after the other with pauses of 10 min between subsequent connections (as shown in Figure 9b).

Figure 9. Simulation results: (a) grid power, (b) vehicle power, and (c) average SOC of stationary battery.

The initial average SOC of the charging station battery was considered to be 60%. To test the BMS
capabilities, all of the modules were initialized with random SOCs between 40% and 80%.

3. Results
The control strategy was validated using Simulink. Figure 9 shows that the ac and dc power levels
could be controlled separately by exploiting the DSCC storage system as an energy buffer, recharging
the vehicles at their full capabilities without overloading the grid.
Comparing Table 3 with Table 2, it is possible to state that the full charging power capabilities
were achieved even though the stationary storage system had a lower capacity than the highest-range
vehicle. Moreover, the stationary storage was sized to maximize its expected lifespan using only an
SOC range of 20%–80%.
In Figure 10, it is possible to analyze the effectiveness of the module balancing algorithm. Even
if the starting condition represents a very unbalanced situation, the BMS capabilities of the DSCC
topology allows the cells to be balanced without dedicated hardware in the first 40 min. From the

50
Energies 2019, 12, 4638

analysis of the figure, it is clear that, in the first 5 min, the balancing inside each arm is completed by
using the load current.

Table 3. Summary of the parameters used for the simulations.

Converter Parameters
Parameter Value Meas. Unit
vph,rated 311 V
vcell,min 3 V
vdc,min 828 V
nmodules 6
nstack 46
vmodule,min 138 V
Estorage 69 kWh

Figure 10. Modules’ SOCs.

Then, in a longer time of up to 40 min, the balancing among the arms is achieved. As the arm
balancing relies on the injection of circulating currents, the balancing power can be set to maximize the
efficiency or reduce the balancing time. It is worth noting that SOCs of all the modules is always kept
under the limit of 80% following the control rule presented in Section 3.

4. Discussion
At present, the main obstacles to the diffusion of EVs are their high costs, limited vehicle ranges,
and long recharging times. In the future, several efforts will be devoted to increase the vehicle range
while reducing the recharging time. For this reason, increasingly higher power levels will be required
for charging stations. Nevertheless, the widespread requests for high-power nodes represent a critical
scenario for actual electric transmission and distribution grids. For this reason, the option to add
a local storage system in proximity to a fast charging station has been under analysis in the recent
literature. In this paper, a new configuration to integrate battery modules in a fast charging station
was proposed and studied. In particular, a DSCC power converter was used to achieve the following
goals: (i) increasing the dc output voltage using lower voltage battery modules and (ii) balancing
the battery modules by separately controlling the power drawn by each module, allowing the use of
second life batteries. Moreover, a control strategy for the power converter was proposed in order to

51
Energies 2019, 12, 4638

decouple the control of the ac and dc power values exchanged by the converter. With the proposed
control, it is possible to exchange two different power values on the two ports of the converter, and
the power difference is automatically supplied or drawn by the stationary battery. In this way, it is
possible to increase the power available for the fast charging station, requiring only the average power
from the grid.
In this study, a power converter was modeled, and numerical results were obtained for a single
point fast charging station. A sequence of three different kinds of vehicles, one city car, one medium-size
car, and one sports car, were simulated, and the obtained numerical results showed the effectiveness
of the proposed converter and strategy. Moreover, the balancing capabilities of the power converter
were demonstrated.

Author Contributions: Conceptualization, D.D.S.; methodology, D.D.S. and L.P.; software, D.D.S.; validation,
D.D.S. and L.P.; formal analysis, D.D.S.; investigation, D.D.S. and L.P.; resources, D.D.S. and L.P.; data curation,
D.D.S.; writing—original draft preparation, D.D.S.; writing—review and editing, L.P.; visualization, D.D.S.;
supervision, L.P.; project administration, L.P.; funding acquisition, L.P.
Funding: This research received no external funding.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Hoimoja, H.; Rufer, A.; Dziechciaruk, G.; Vezzini, A. An ultrafast EV charging station demonstrator.
In Proceedings of the International Symposium on Power Electronics Power Electronics, Electrical Drives,
Automation and Motion, Sorrento, Italy, 20–22 June 2012; pp. 1390–1395.
2. Overview of Battery Cell Technologies, Marcel MEEUS. Energy Materials Industrial Research Initiative
(EMIRI). Available online: https://europa.eu/sinapse/webservices/dsp_export_attachement.cfm?CMTY_ID=
0C46BEEC-C689-9F80-54C7DD45358D29FB&OBJECT_ID=230DABFD-90AB-8F7D-083EF5BD909DD025&
DOC_ID=9C5B91FE-01BC-5F72-79D01E1939A9EE53&type=CMTY_CAL (accessed on 8 October 2019).
3. Richard, L.; Petit, M. Fast charging station with battery storage system for EV: Grid services and battery
degradation. In Proceedings of the 2018 IEEE International Energy Conference (ENERGYCON), Limassol,
Cyprus, 3–7 June 2018; pp. 1–6.
4. Voelcker, J. Porsche’s fast-charge power play: The new, all-electric Taycan will come with a mighty thirst.
This charging technology will slake it. IEEE Spectr. 2019, 56, 30–37. [CrossRef]
5. Jochem, P.; Landes, P.; Reuter-Oppermann, M.; Fichtner, W. Workload Patterns of Fast Charging Stations
Along the German Autobahn. World Electr. Veh. J. 2016, 8, 936–942. [CrossRef]
6. Srdic, S.; Lukic, S. Toward Extreme Fast Charging: Challenges and Opportunities in Directly Connecting to
Medium-Voltage Line. IEEE Electrif. Mag. 2019, 7, 22–31. [CrossRef]
7. Martinez-Rodrigo, F.; Ramirez, D.; Rey-Boue A., B.; de Pablo, S.; Herrero-de Lucas L., C. Modular Multilevel
Converters: Control and Applications. Energies 2017, 10, 1709. [CrossRef]
8. Van Hertem, D.; Gomis-Bellmunt, O.; Liang, J. HVDC Grids: For Offshore and Supergrid of the Future; John
Wiley & Sons: New York, NY, USA, 2016.
9. Li, J.; Bhattacharya, S.; Huang, A.Q. A New Nine-Level Active NPC (ANPC) Converter for Grid Connection
of Large Wind Turbines for Distributed Generation. IEEE Trans. Power Electron. 2011, 26, 961–972. [CrossRef]
10. Ruderman, A.; Reznikov, B. Time domain evaluation of filterless grid-connected multilevel PWM converter
voltage quality. In Proceedings of the 2010 IEEE International Symposium on Industrial Electronics, Bari,
Italy, 4–7 July 2010; pp. 2940–2945.
11. D’Arco, S.; Piegari, L.; Tricoli, P. A modular converter with embedded battery cell balancing for electric
vehicles. In Proceedings of the 2012 Electrical Systems for Aircraft, Railway and Ship Propulsion, Bologna,
Italy, 16–18 October 2012; pp. 1–6.
12. D’Arco, S.; Quraan, M.; Tricoli, P.; Piegari, L. Battery charging for electric vehicles with modular multilevel
traction drives. In Proceedings of the 7th IET International Conference on Power Electronics, Machines and
Drives (PEMD 2014), Manchester, UK, 8–10 April 2014; p. 2.
13. Quraan, M.; Yeo, T.; Tricoli, P. Design and Control of Modular Multilevel Converters for Battery Electric
Vehicles. IEEE Trans. Power Electron. 2015, 31, 507–517. [CrossRef]

52
Energies 2019, 12, 4638

14. De Simone, D.; Piegari, L.; D’Areo, S. Comparative Analysis of Modulation Techniques for Modular Multilevel
Converters in Traction Drives. In Proceedings of the 2018 International Symposium on Power Electronics,
Electrical Drives, Automation and Motion (SPEEDAM), Amalfi, Italy, 20–22 June 2018; pp. 593–600.
15. Ouerdani, I.; Bennani, A.; Ben, A.; Slama, B.; Montesinos Miracle, D. Phase Opposition Disposition PWM
Strategy and Capacitor Voltage Control for Modular Multilevel Converters. In Proceedings of the International
Conference on Recent Advances in Electrical Systems, Hammamet, Tunisia, 20–22 December 2016.
16. Brando, G.; Dannier, A.; Spina, I.; Tricoli, P. Integrated BMS-MMC Balancing Technique Highlighted by a
Novel Space-Vector Based Approach for BEVs Application. Energies 2017, 10, 1628. [CrossRef]
17. Zahoor, W.; Zaidi, S.H. Synchronization and dq current control of grid-connected voltage source inverter.
In Proceedings of the 17th IEEE International Multi Topic Conference 2014, Karachi, Pakistan, 8–10 December
2014; pp. 462–466.
18. Piegari, L.; Tricoli, P. A control algorithm of power converters in smart-grids for providing uninterruptible
ancillary services. In Proceedings of the 14th International Conference on Harmonics and Quality of Power -
ICHQP 2010, Bergamo, Italy, 26–29 September 2010; pp. 1–7.
19. Electric Vehicle Database. Available online: https://ev-database.org/car/1145/BMW-i3-120-Ah (accessed on
30 September 2019).
20. Electric Vehicle Database. Available online: https://ev-database.org/car/1144/Nissan-Leaf-eplus (accessed on
30 September 2019).
21. Electric Vehicle Database. Available online: https://ev-database.org/car/1116/Porsche-Taycan-Turbo-S
(accessed on 30 September 2019).

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

53
 energies
Article
Location and Sizing of Battery Energy Storage Units
in Low Voltage Distribution Networks
Andrea Mazza 1 , Hamidreza Mirtaheri 2 , Gianfranco Chicco 1, * , Angela Russo 1 and
Maurizio Fantino 2
1 Dipartimento Energia “Galileo Ferraris”, Politecnico di Torino, 10129 Turin, Italy;
andrea.mazza@polito.it (A.M.); angela.russo@polito.it (A.R.)
2 Links Foundation, 10138 Torino, Italy; hamidreza.mirtaheri@linksfoundation.com (H.M.);
maurizio.fantino@linksfoundation.com (M.F.)
* Correspondence: gianfranco.chicco@polito.it

Received: 12 November 2019; Accepted: 17 December 2019; Published: 20 December 2019

Abstract: Proper planning of the installation of Battery Energy Storage Systems (BESSs) in distribution
networks is needed to maximize the overall technical and economic benefits. The limited lifetime and
relatively high cost of BESSs require appropriate decisions on their installation and deployment, in
order to make the best investment. This paper proposes a comprehensive method to fully support the
BESS location and sizing in a low-voltage (LV) network, taking into account the characteristics of
the local generation and demand connected at the network nodes, and the time-variable generation
and demand patterns. The proposed procedure aims to improve the overall network conditions, by
considering both technical and economic aspects. An original approach is presented to consider both
the planning and scheduling of BESSs in an LV system. This approach combines the properties of
metaheuristics for BESS sizing and placement with a greedy algorithm to find viable BESS scheduling
in a relatively short time considering a specified time horizon, and the application of decision theory
concepts to obtain the final solution. The decision theory considers various scenarios with variable
energy prices, the diffusion of local renewable generation, evolution of the local demand with the
integration of electric vehicles, and a number of planning alternatives selected as the solutions with
top-ranked objective functions of the operational schedules in the given scenarios. The proposed
approach can be applied to energy communities where the local system operator only manages the
portion of the electrical grid of the community and is responsible for providing secure and affordable
electricity to its consumers.

Keywords: distribution system; batteries; storage; planning; scheduling; decision theory

1. Introduction
The progress of technologies concerning different types of batteries and their control systems,
together with the evolution of a regulatory framework in which energy storage is considered more
explicitly, are making Battery Energy Storage Systems (BESSs) progressively more cost-effective for
energy system applications. A BESS is specified by its power rating and energy capacity. Both of these
specifications impact the BESS investment cost and need to be defined separately [1]. Typical BESS
applications for power and energy systems include improvement of the quality of service, assistance
with primary and secondary frequency control to enhance network stability, the smoothing of power
fluctuations in the generation profiles for better integration of Renewable Energy Sources (RES), and
the promotion of higher users’ participation in demand management through time-shifting of the
energy usage [2]. Most BESS applications in electrical networks refer to times of 1–2 h and to BESS
sizes smaller than 50 kW. However, the number of applications for longer times (e.g., 2–5 h) and sizes

Energies 2020, 13, 52; doi:10.3390/en13010052 55 www.mdpi.com/journal/energies

Energies 2020, 13, 52

up to 500 kW is already significant, and applications for bigger sizes (also in the range 1–10 MW)
are increasing.
In the transmission system, the positive impacts of BESSs for providing a fast response to frequency
deviations [3], mitigating under-frequency transients [4], and being exploited for energy arbitrage [5]
have been observed. The integration of BESSs with appropriate control strategies has been shown
to be able to improve the frequency stability [6]. At the distribution system level, a BESS is mainly
used for enhancing the grid integration of RES by mitigating the effects of the uncertainty on load
and distributed generation [7,8], improving the distribution system reliability by avoiding operations
close to the line thermal limits and thus more exposed to the risk of protection trips [9], enhancing
the quality of the supply with relatively high-power and low-energy solutions [10], reducing the
need for grid expansion, shaving the power demand peaks through load shifting or load leveling [9],
optimizing the energy transaction costs [11], and integrating BESSs with the demand response in
microgrid applications [12]. In [13], a distinction is made between a centralized BESS (that participates
in reducing the demand deviation, avoiding the reverse power flow and correcting the power factor),
and a distributed BESS (that aims at annual energy loss reduction, reduction of the demand deviation,
and improvement of the voltage profile).
A key challenge is to determine the power rating, energy capacity, and location of BESSs in the
distribution network. A number of contributions have been published on the siting and sizing of
BESSs. Different objectives, such as peak shaving, voltage regulation, and reduction of the energy
not supplied, are combined in [14] in an optimal power flow-based approach. In [15], reliability
improvement, together with peak shaving, is considered as an objective. The review [16] presents
a categorization of the methods used to determine the BESS siting and sizing, in which four main
groups are identified (analytical methods, exhaustive search, mathematical programming, and heuristic
methods). Decision-making tools are also applied. For example, in [17], the optimal BESS sizing and
siting is identified in a microgrid with RES taking into account demand and generation uncertainty by
using decision theory criteria. In [18], hybrid energy storage systems, including BESSs, are addressed,
by determining the energy storage capacity in distribution systems through an assessment of the risk
tolerance of the investors. The results of BESS installations in terms of providing different services at
different voltage levels are reported in [19].
A specific aspect generally not highlighted in the reviews on BESS siting and sizing is the
distinction between applications for Medium Voltage (MV) and Low Voltage (LV) systems. The
literature contributions mainly refer to Medium Voltage (MV) distribution systems. However, the
formulation of a planning problem referring to the installation of BESSs in LV systems has various
differences with respect to what happens in MV systems. First of all, in MV systems, the study can
be conducted by assigning growth in the local generation and demand aggregated at the MV node
level. In this case, it is possible to mix up the contributions from different energy sources at the LV
level, without looking at the details of the individual sources. Additionally, the uncertainty that
characterizes the local generation is seen with respect to the aggregation of the generations, typically
taking into account possible correlations between the generation patterns due to external variables
(e.g., solar irradiance and temperature for photovoltaic systems). Then, it is possible to exploit the
smoothing effect due to the aggregation of a number of individual demands, define typical patterns for
the aggregated demand, and associate these typical patterns to predefined evolutions in time.
Conversely, for an LV system, the level of aggregation of demand and local generation is much
lower, and many more critical aspects appear. In particular, the local characteristics of the generation
and demand at each LV node have to be considered individually. The uncertainties of local generation
and demand increase as the smoothing effect of the aggregations is reduced. The setting up of scenarios
of development of the local generation cannot proceed with a generic aggregated effect at each LV node,
but has to take into account where there are different types of local generation and what reasonable
increment can be established for that type of generation. Some relevant differences between MV and
LV systems are summarized in Table 1.

56
Energies 2020, 13, 52

Table 1. Summary of relevant differences between Medium Voltage (MV) and Low Voltage (LV) systems.

Characteristics MV Grid LV Grid Consequences

LV grid cannot be reconfigured, thus the
proper network operation has to be
Structure Weakly meshed Radial
guaranteed thanks to the devices
connected to the grid
In the LV grid, it is advisable to apply
Load Balanced Unbalanced
three-phase load flow
In the LV grid, the voltage drop is
strictly correlated to the active power,
Branch R≈X R>>X and voltage control can thus be
effectively conducted by modifying the
net nodal active power
LV grid represents the last mile of the
grid and supplies the customer. A more
detailed representation of loads and
Load profile Aggregate Not aggregate
generation profiles is then necessary
and the aggregation impact is less
evident than in MV grids

To achieve the best BESS performance and to maximize the overall benefits, proper planning is
necessary. For instance, in [20], several configurations of BESSs are compared and the overall network
impact is evaluated and compared for different placements of BESSs in the network. In the relevant
literature, there are some contributions related to the optimal planning (sizing and placement) of
energy storage systems in LV distribution networks. The method applied in [21] aims to optimally
configure the energy storage systems to alleviate over- and under-voltage problems. The problem of
the optimal location is solved by a heuristic method based on voltage sensitivity analysis. Uncertainties
due to stochastic generation and demand are also considered in the optimal sizing and the worst-case
approach is applied to select the sizes.
Some contributions refer to LV distribution networks characterized by a high penetration of
photovoltaic generation, and consider the possibility of alleviating the negative impacts by the
installation of storage systems. In [22], a heuristic method is applied to determine the optimal location
and sizing of storage systems and the objective to be minimized is a cost function accounting for the
cost of storage systems and the cost due to voltage deviations. The optimal planning of BESSs proposed
in [23] aims at maximizing an objective function that includes both benefits and costs (i.e., energy
arbitrage, environmental emission, energy losses, transmission access fee, capital, and maintenance
costs of a BESS). Daily charge/discharge of the storage systems is also determined considering a proper
model of the BESS operation. In [24], the authors propose a procedure for the optimal placement and
sizing of distributed energy storage systems in low voltage distribution systems aimed at maximizing
the utilization of photovoltaic plants and minimizing the battery degradation. The multi-objective
optimization problem proposed by [25] is focused, from one side, on the minimization of energy losses
and, from the other side, on the minimization of costs associated with distributed generators and energy
storage systems. A distribution system with a high penetration of photovoltaics generators is considered
in [26]. A heuristic procedure for reducing the search space for the location of storage systems in a
low voltage microgrid is proposed in [27], where analytical considerations on the voltage sensitivity,
voltage unbalances, and line loading drive are included in the selection of candidate locations.
This paper proposes an overall procedure to address the BESS location and sizing in an LV
network, taking into account the characteristics of the local generation and demand connected at the
LV nodes and the time-variable generation and demand patterns. The proposed procedure aims to
improve the overall network conditions by considering both technical and economic aspects. This
condition aims to represent the conditions that could be found in the case of energy communities
where the local system operator only manages the electrical grid of the community and is responsible
for guaranteeing secure and affordable electricity for its consumers.

57
Energies 2020, 13, 52

An original algorithm is presented to consider both the planning and scheduling of BESSs in an LV
system. This algorithm combines the properties of metaheuristics (to explore the solution space as much
as possible), and a greedy algorithm able to find viable BESS scheduling in a relatively short period
of time, considering a specified time horizon. The final solution is obtained by using a multi-criteria
decision making (MCDM) approach based on the application of decision theory concepts [28] to a
number of selected planning alternatives evaluated for different weighted scenarios. Decision theory
is an appropriate tool for dealing with cases in which the uncertainty on possible future situations is
very large and is handled through scenario analysis.
The next sections of this paper are organized as follows. Section 2 describes the details of the
methodology used to address the planning problem. Section 3 presents the application to an LV
distribution system and discusses the results. The last section contains the concluding remarks.

2. Description of the Methodology

The BESS location and sizing is analysed as a planning problem seen from the point of view of
the electricity manager of an energy community, which is responsible for guaranteeing secure and
affordable electricity for its customers at a minimum cost, as well as for both the infrastructure and
quality of the service.

2.1. Data Resolution and Reference Period

The data used are assumed to have a constant time resolution Δt. A reference period of duration
Tref is assumed, in which the operation of the LV distribution system is analysed in detail by considering
the generation and demand patterns and a specific model for BESS scheduling. The planning problem
is set up for a time horizon multiple of Tref , namely, with an overall duration of TH = NH × Tref ,
where NH > 1 is an integer number. It is assumed that the BESSs that will be chosen by the proposed
procedure will be installed at the beginning of the time period of analysis. The planning time horizon
chosen is longer than the lifetime of the BESS, in order to include the replacement of the BESS during
the planning period.

2.2. Definition of the Scenarios

The electricity prices, diffusion of the local generation, and diffusion of electric vehicles have been
assumed to be uncertain data inputs that affect the solution of the planning problem. Several methods
can be applied to handle uncertain variables; in this proposal, several scenarios will be identified to
represent different instances.
Starting from the results determined for the reference period, a number of scenarios are constructed
to represent possible paths of evolution of selected quantities that affect the LV network operation. The
scenarios are defined by taking into account the long-term changes in time that may appear in the
following quantities:

(a) Electricity prices: the consumers or prosumers connected to the LV system are considered as price
takers, namely, they do not participate in the definition of energy prices in the wholesale electricity
market. MP trends of variation of the electricity prices are established by the user by considering
the final increase (or decrease) of the electricity price at the end of the planning time horizon;
(b) Diffusion of the local generation: the distributed generation (DG) connected to the LV network can
change at selected locations in different ways. For LV systems, it is likely that more prosumers
will install their local generation systems at locations where there is no local production. From
the point of view of the scenario definition, MDG trends of variation of the local generation are
considered, and each one is defined by assuming a rate of increase of the energy production from
local generation (not of the power installed);

58
Energies 2020, 13, 52

(c) Diffusion of electric vehicles: in the present situation, the diffusion of EVs is still relatively limited
in many jurisdictions. The number of EVs will increase in the future, and different hypotheses
about their number can be represented by MEV trends.

Under the hypotheses provided for the scenarios, the number of scenarios considered is equal to
M = MP × MDG × MEV .
The scenarios obtained are applied to calculate the objective function for the planning problem
(Section 2.4), and are weighted in order to be used in the decision-based approach illustrated in
Section 2.6.

2.3. Definition of the Sizing Alternatives

Due to the particular use of the BESS, which includes load leveling, the indication of [29] to use
the energy to power ratio equal to 2 is followed. In this way, data are described in terms of their energy
capacity, and the power rating is then directly linked to the energy capacity through the energy to
power ratio.
In the distribution network, there are K nodes, but it is assumed that a user-defined number KBESS
< K of nodes is taken into account for possible BESS location. A maximum value for the BESS energy
capacity Ck,max is assigned to each node k = 1, . . . , KBESS , depending on the characteristics of the
node. The final BESS energy capacity to be assigned to each node will be determined by the proposed
approach in the range from zero to Ck,max at the nodes k = 1, . . . , KBESS . To avoid the use of continuous
variables, this range is partitioned into a given number Λ of BESS energy capacity levels. Without loss
of generality, the number Λ is chosen as a constant for all the nodes, that is, it is independent of k. In
this way, the total number of alternative combinations of BESS sizes is S = ΛKBESS . Even for relatively
small numbers of nodes and BESS energy capacity levels, the number of combinations S can become so
high that even their enumeration becomes intractable with an exhaustive search process. For example,
if KBESS = 20 nodes and Λ = 5 BESS energy capacity levels, the result is S = 520 = 9.54 × 1013 . In this
situation, only parts of the combinations will be reached during the planning procedure, for instance,
that conducted by using a metaheuristic algorithm able to explore the solution space with a conceptual
direction of evolution towards the global optimum of the objective function employed in the definition
of the planning problem.

2.4. Definition of the Objective Functions

The solution of the power flow at each time step h = 1, . . . , H, together with the BESS operational
schedules, provide the data required to calculate the power flows and the LV network losses, and to
determine whether there is a reverse power flow with the power injected at the supply point. These
results are used as the contribution of the distribution system operation to the formulation of the
objective function for the planning problem.
The general objective function formulated for the planning problem is a penalized objective
function defined on the basis of the investment and operation and maintenance (O&M) costs, and of
penalty terms associated with violations of the voltage limits and with the presence of reverse power
flow to the MV distribution system. The expression of the objective function is constructed, starting
from a reference function fref and including some penalty terms to obtain the penalized objective
function fP :
fref = Δcinv + cO&M (1)

fP = fref × (1 + πV + πR ), (2)

where the addends have the following meaning:

• Δcinv : investment costs for BESS purchasing and installation, determined by using the cost per
kWh, depending on the BESS energy capacity, and the cost per kW applied to the inverter for grid
connection of the BESS (depending on the BESS power rating) [29];

59
Energies 2020, 13, 52

• cO&M : operation and maintenance costs, calculated by considering the costs of network losses
(evaluated using the electricity prices), the BESS aging applied as a reduction of the maximum BESS
energy capacity [29], and the maintenance costs considered as a percent of the investment costs;
• πV : penalty term associated with the violation of the upper voltage limit Vmax or the lower voltage
limit Vmin , with a penalty coefficient ρV assigned by the user in such a way that the penalty
is significantly higher than the terms of fref , without being excessive (otherwise, only feasible
solutions would be accepted during the optimization problem, going against the goal of the
metaheuristics to open the search space by also accepting penalized cases):


K 
H
πV = ρV max{Vkh − Vmax , Vmin − Vkh , 0}; (3)
k =1 h=1

• πR : penalty term associated with the total energy injected at the supply point in the cases of
reverse power flow from the low voltage to the medium voltage network that can impact the
voltage regulation, as well as the protection systems [30–32]:


H
πR = ρR max{−P0h , 0}Δt. (4)
h=1

2.5. Overall Calculation Procedure

The calculation procedure is composed of two main steps, as shown in Figure 1:

(1) Step A, essentially based on the exploitation of the features of a customized genetic algorithm, and
(2) Step B, where the calculation of potential scheduling for all batteries installed is suggested through
a greedy algorithm.

On the basis of the outputs of Step B, the objective function related to that particular set of BESSs
is evaluated.
First of all, all the input data are introduced: in particular, the code requires information about
the network data, number of nodes KBESS where the BESS can be installed, time horizon (through the
number of days Nd ) analysed, time discretization (i.e., number of time steps per day Nt ), and number of
chromosomes Nc for the genetic algorithm. Thanks to the above information, the procedure continues
with Step A (planning) and Step B (dispatching). After the calculation of the objective function at the
iteration nG for all the chromosomes, the convergence criterion is checked.

2.5.1. Step A

The initial population च଴ is created and the initial objective function values are collected in the
(0)
vector fobj . The population is composed of chromosomes with the coding shown in Table 2.

Table 2. Codification used in the proposed genetic algorithm.

Gene 1 Gene 2 Gene 3

Number of BESSs installed Nodes where BESSs are installed BESS energy capacities (kWh)

60
Energies 2020, 13, 52





 #

'%



(%
 
 




)%
 


  


*%
 







+%
 







 


 
 

$ 




%

  
 


 ! 
 

  

 ! 



-'

 
 &'

 &(



 
$  



 
 %

 -
 ,
'





-

,
'

 

   
  
 



   








  "



Figure 1. Scheme of the proposed procedure.

It is worth noting that according to the number of BESSs installed (indicated by gene 1), the number
of elements composing gene 2 and gene 3 will vary. This implies that a number of pre-defined rules are
needed to handle the genetic operators applied to the chromosomes, to make the creation of successive
generations possible.
1. Selection and Crossover
The selection process is based on the application of the biased roulette wheel.
Once the parents’ selection is made, the crossover may be applied according to the probability of
crossover pc .
The crossover follows some general rules:
• The number of batteries is inherited from the parent#1, and
• The crossover is only applied to the values contained in gene 2 and gene 3.

61
Energies 2020, 13, 52

Due to the particular codification used, some fixed rules have been used to apply crossover in the
presence of chromosomes with different characteristics. These rules are clarified with reference to the
chromosomes shown in Figure 2.

gene 1 gene 2 gene 3

chr#1 2 {10,18} {20,40}

chr#2 4 {5,10,12,14} {10,20,25,30}

Figure 2. Example of chromosomes.

Different cases do exist:

• If parent#1 = chr#1, the offspring will be that shown in Figure 3a (offspring#1), and
• If parent#1 = chr#2, the offspring will be that shown in Figure 3b (offspring#2). In this particular case,
the number of batteries imposed by parent#1 is higher than that available in parent#2. Therefore,
information regarding the remaining nodes and capacities is randomly picked up from a repository
containing all the nodes (and related capacities) referring to the current population (e.g., nodes
{15, 24} and capacities {5, 10} in offspring#2).

offspring#1 2 {5,10} {10,20}

(a) Codification of offspring#1

offspring#2 4 {10,18,15,24} {20,40,5,10}

(b) Codification of offspring#2

Figure 3. Example of offpring when the parents have different lengths.

2. Mutation
As the crossover operator, the mutation is only applied to the contents of gene 2 and gene 3. This
operator is only applied when the value of a random number extracted from a uniform distribution is
lower than the probability of mutation pm . If the mutation is allowed, the values to be substituted are
picked up from the repository containing all the nodes and relative capacities of the current population.

2.5.2. Step B
During Step B, the procedure calculates the operation setpoints of all the installed batteries.
These setpoints are required to calculate the objective function related to the configuration (number
of batteries, their positions, and their capacities) specified in every chromosome comprising the
population in the generation nG . The flow chart related to Step B is shown in Figure 4.

62
Energies 2020, 13, 52

From Step A

initialization day counter

N to main

Y
initialization node counter

Calculate the net load , the

mean net load and the
unbalance referring to node k
and day d

= +1
Rank the vector in descending order to
obtain and

A lgorithm 1

Figure 4. Flowchart related to Step B.

(n )
For every node k = 1, . . . ,K where the batteries are installed, the net load nd,kG ∈ R(Nt ,1) is
calculated as the difference between the load and local generation. With this information, it is possible
(n )
to calculate the mean value of the nodal net load nd,kG ∈ R(1,1) and thus an indicative value of unbalance
(n )
of the net load with respect to its mean value, i.e., ud,kG ∈ R(Nt ,1) , calculated as

(n ) (n ) (n )
ud,kG = nd,kG − nd,kG , k = 1, . . . , K. (5)

(n )
The vector ud,kG , k = 1, . . . ,K, provides information regarding the time steps during which the use
of the battery system can be useful to level the net load.
(n )
On the basis of the value of ud,kG , the daily time steps are reordered in a descending way, by
(n )
obtaining the unbalance ranked vector rd,kG ∈ R(Nt ,1) and the corresponding ranked index vector
(n )
νd,kG ∈ R(Nt ,1) : by knowing the time steps when the unbalance is higher, it is possible to apply the
scheduling algorithm by starting from the “most critical” time steps.
The pseudo-code of the scheduling algorithm is shown in Figure 5. Algorithm 1 tries to set the best
option for the current time instant, which is according to the order of solving, more critical with respect
to the next time instants, and pushes the battery constraint violations towards the lowest priority time
G (n )
steps. First of all, the set-points for a single battery (collected in the vector called pBESS ) are instructed,
(n )
starting from the value of unbalance ud,kG . The feasibility of that desired pattern depends on the
(−)
maximum exploitable storage level of the battery and its maximum charging power, called Pmax , as
(+) (n )
well as its maximum discharging power Pmax . According to the indices value collected in rd,kG , the

63
Energies 2020, 13, 52

(n )
algorithm starts from the most critical time step that is the first element of νd,kG and sets the set-point
at that step by respecting the battery’s power constraint, as shown at line 5.

Algorithm 1: Battery Scheduling

( )
1) , = [NaN, NaN, NaN, …, NaN]
2) for i=1 to T do
( )
3) j = , ,
( )
4) if , , > 0 then //Disharge
( ) ( ) (−)
5) , , = min( , , , )
6) = 0
7) for t = 1 to T do
ĸ -( [min )])
( ) (+)
8) , , +(  ( – ,
9) = max ( - , 0)
10) if > 0 then
( ) ( ) ( )
11) , , ĸ , , - , ,
12) exit t
13) end t

( )
14) if , , < 0 then //Charge
( ) ( ) (+)
15) , , = max( , , , )
16) = 0
17) for t = 1 to T do
ĸ -( [min )])
( ) (−)
18) , , +(1- ( - ,
19) = min ( – , 0)
20) if > 0 then
( ) ( ) ( )
21) , , ĸ , , + , ,
22) exit t
23) end t
24) end i

Figure 5. Pseudo-code of the battery scheduling.

Then, an auxiliary variable called I (that stands for integrator) is initialized at line 6 with the last
value of energy stored in the battery. Following this, the integration operation is executed to check
whether the energy constraint is respected or not. The parameter ω is defined as follows:
⎧
⎪
⎪ (n )
= NaN
⎨ 0, G
pBESS,t
ω=⎪
⎪ .
⎩ 1, else

Line 8 indicates that a temporary value for an integral operation equal to the maximum availability
state is considered at the time step t. The maximum availability state changes if either a charging or
discharging mode is considered: in the discharging mode, the state is the fully charged state, whereas, in
charging mode, the maximum availability state is the empty one.
Once the integration operator goes beyond the maximum and minimum state of charge (SoCmax
G (n )
and SoCmin ), a correction factor called C f resets pBESS,j and breaks integration execution. This operation
is similarly carried out for the charging mode with corresponding signs.

2.6. Selection of the Planning Alternatives

For each scenario, the ranking of the solutions is made on the basis of the objective function (from
the best solutions to the worst ones), and the Z top-ranked solutions are selected. The rationale for
this selection is that there is no guarantee that the global optimum will be reached from the execution

64
Energies 2020, 13, 52

of the metaheuristic, so, taking more than one solution from the ranking, enhances the possibility of
having good candidates to compare with the MCDM approach.
The number of planning alternatives is then defined as A = M × Z, that is, with the Z top-ranked
solutions for each one of the M scenarios analysed.
Since each planning alternative a = 1, . . . ,A exhibits a different performance according to the
scenario, fP (a, m) is the value of the objective function fP defined in (2), evaluated for the alternative a
when the scenario m occurs. The objective function values are then arranged into a matrix with A rows
(planning alternatives) and M columns (scenarios).
The MCDM approach is based on the application of decision theory criteria to the A planning
alternatives by considering the M scenarios. In the framework of the decision theory concepts, several
criteria can be applied to select the optimal planning alternative, taking into account that each scenario
m has a probability of occurrence pm .

2.6.1. Criterion of Minimum Expected Cost

The criterion of the minimum expected cost attempts to minimize the costs [33]. The optimal
planning alternative a∗ec is the one that minimizes the expected cost EC:
 
a∗ec = arg min EC(a) , (6)
a=1,...,A

where the expected cost EC(a) for each alternative a = 1, . . . ,A is determined as


M
EC(a) = pm fp (a, m). (7)
m=1

The assignment of the probabilities of occurrence is crucial; when the equal likelihood criterion is
adopted [34], each scenario has the same probability of occurrence.

2.6.2. Criterion of Minimax Weighted Regret

The criterion of minimax weighted regret attempts to minimize the regret corresponding to the
worst case [34]. For a given scenario m, it is possible to identify the best planning alternative as the one
corresponding to the lowest cost; if a planning alternative different from the optimal one is chosen,
a greater cost will be experienced and, therefore, the regret can be calculated. Let us consider the
scenario m, where the best planning alternative a∗m is

a∗m = arg min fp (a, m) (8)
a=1,...,A

and, when scenario m occurs and the alternative a different from a∗m is chosen, the regret can be
quantified as
R(a, m) = fp (a, m) − fp (a∗m , m). (9)

Considering the probability of occurrence of each scenario, the weighted regret Rw (a, m) is
determined as
Rw (a, m) = pm R(a, m). (10)

According to the criterion of minimax weighted regret, the optimal planning alternative a∗wr is the
one that minimizes the maximum weighted regret, that is,
 
a∗wr = arg min max (Rw (a, m)) . (11)
a=1,...,A m=1,...M

As in the case of the criterion of Section 2.6.1, the equal likelihood criterion can be adopted [33].

65
Energies 2020, 13, 52

2.6.3. “Optimist” and “Pessimist”Criterion

When the “optimist” criterion is applied [28,31], for each planning alternative, the best value
of the costs (i.e., the minimum value) over the possible scenarios is selected and, then, the selected
planning alternative a∗opt is the one that minimizes the cost corresponding to the best possible outcome
for each scenario, that is,  
a∗opt = arg min min fp (a, m) . (12)
a=1,...,A m=1,...M

Conversely, the “pessimist” criterion [28,31] attempts to minimize the worst outcome of the
planning alternatives. Therefore, the worst value of the costs (i.e., the maximum value) of each
alternative over the possible scenarios is selected and, then, the selected planning alternative a∗pes is the
one that minimizes the worst outcome, that is,
 
a∗pes = arg min max fp (a, m) . (13)
a=1,...,A m=1,...M

In addition, an “optimist”-“pessimist” criterion can be considered as a mixed approach. In this

case, both the worst and best outcome of each alternative are considered and these values are weighted
by a proper factor α ∈ [0, 1]:
 
a∗opt−pes = arg min α min fp (a, m) + (1 − α) max fp (a, m) . (14)
a=1,...,A m=1,...M m=1,...M

When applying the “optimist”, the “pessimist”, and the “optimist”-“pessimist” criteria, the
decision does not depend on the probabilities of the scenarios.

3. Case Study Application and Results

3.1. Network Data

The application presented in this paper is based on a rural LV network with 22 nodes, including
the slack node (Figure 6). The network is radial and is supplied by the MV system through an MV/LV
transformer (not represented in the figure). The network contains mainly residential and agricultural
customers. The number of nodes considered for BESS installation is KBESS = 5. The BESS data are
shown in Table 3. For discretization of the BESS energy capacity, the number of levels used is Λ = 47,
with BESS energy capacities considered to lie in the range 3 ÷ 49 kWh, and the energy to power ratio is
equal to 2, as indicated in Section 2.3 [29]. The total number of alternative combinations of BESS sizes
is S = ΛKBESS = 475 = 2.29 × 108 . Such a number is practically intractable with an exhaustive search
process. In this situation, the use of a metaheuristic algorithm to identify the solution of the planning
problem is justified.

Table 3. Battery Energy Storage System (BESS) capacity data.

Minimum Energy Maximum Energy

Discretization Steps Energy to Power Ratio
Capacity (kWh) Capacity (kWh)
3 49 47 2

66
Energies 2020, 13, 52

3 6

4 7

5 8

10 11

12

13

14 15

16 18 20 21

17 19

Figure 6. Rural LV network.

3.2. Input Data

Without loss of generality, the reference period Tref is assumed to be one year, and the planning
time horizon is assumed to be 15 years (i.e., NH = 15).
The batteries chosen are Lithium Ion (type Nickel Manganese Cobalt): the cost used in this paper
is the one expected in 2020 and equal to 167 $/kWh, whereas the inverter cost is equal to 50 $/kW. The
operation and maintenance cost is equal to 1% of the investment cost per year. The BESS lifetime has
been considered to be equal to 13.4 years [29]. The BESS degradation has been modeled as a reduction
of the energy capacity of the battery, equal to 2.4% per year [35].
For execution of the genetic algorithm, the probability of crossover is imposed as pc = 0.75, and
the probability of mutation is set to pm = 0.05.

3.2.1. Loads
Regarding the calculation of the load profiles, it takes into account the real contractual delivery
powers of each load. Two nodes are considered as commercial consumers, while the other ones
refer to residential customers. For the profiles, due to a lack of complete information from the actual
profiles, relevant (residential and commercial) profiles have been taken from an open-source dataset
(OpenEI) [36], normalized and multiplied by the nominal powers.

3.2.2. PV Generation
The considered PV production, in the simulation, is calculated based on yearly solar irradiance
and the nominal installed power [37]. Yearly solar irradiance data was obtained through collected data
from a third party weather service provider [38] for the specific location of the case study with the
coordinates of 46.4746◦ N, 11.2479◦ E.

3.2.3. EV Relevant Information

EVs are considered in some scenarios, through the load profile caused by their charging. Within
this analysis, the number of EVs that determines the EV charging profile is set to 10 at the beginning of

67
Energies 2020, 13, 52

the scenarios with EV and that number gradually increases by the rate of one EV per year. The nominal
power of the EV chargers is considered to be 3.6 kW, i.e., a standard power level of charging stations,
and charging events mainly occur during the night.

3.2.4. Energy Price Evolution

Energy price evolutions increase and decrease linearly, with the granularity of one year along
entire scenario horizons.

3.3. Definition of the Scenarios

The scenarios are defined by assuming the following entries:

(a) Electricity prices: MP = 2 trends of variation are considered, namely, with a 50% increase and 20%
reduction of the prices in the planning time horizon of 20 years. Linear variations of the prices
are assumed during the years;
(b) Diffusion of the local generation: MDG = 2 trends of variation of the local generation are considered,
with a 20% and 50% increase of the energy production from local generation in the planning
time horizon of 20 years. Linear variations of the local generation diffusion are assumed during
the years;
(c) Diffusion of charging points for electric vehicles: MEV = 2 trends are assumed, in which (i) no parking
lot and no battery charging stations will be built, and (ii) one parking lot considering 10 active
(i.e., with an EV connected for charging purposes) charging points at node 13 is assumed, with
the increase rate of the number of active charging points of 1 per year.

3.4. Calculation of the Objective Function

The objective function has been evaluated for each planning alternative and scenario. The best
three solutions for each one of the eight scenarios have been calculated for all the scenarios to provide
the starting matrix to carry out the calculations based on decision theory. The number of alternatives is
then 24. Table 4 shows the objective function values: it is worth noting that, with respect to scenario
2, the values of the objective function for alternatives 3, 8, and 12 are 0.5135, 0.5136, and 0.5140,
respectively. Scenarios 5–8 (associated with a large deployment of EVs) exhibit higher values of the
objective function. Table 5 reports the technical data of the alternatives, including the number of BESSs
installed, the nodes where each BESS is located, and the BESS energy capacity at each node. The
solutions include one to three BESSs. Looking at the BESS locations, it can be seen that, in three cases,
the slack node is chosen to install a BESS. In this way, the BESS acts to limit the reverse power flows. In
the alternative number 23, node 8 is selected two times, because the energy capacity of the BESS in one
installation reaches the limits, and the procedure did not find a better alternative to place the BESS in
another node.

68
Energies 2020, 13, 52

Table 4. Values of the objective function for each planning alternative and for each scenario (×1012 ).
Values in bold indicate the best solution for each scenario.

Scenarios
Alternative 1 2 3 4 5 6 7 8
1 0.373 0.600 2.44 3.39 5.98 8.31 16.7 23.2
2 0.371 0.515 1.78 2.47 5.84 8.10 16.3 22.6
3 0.371 0.514 1.78 2.47 5.83 8.10 16.3 22.6
4 0.381 0.529 2.07 2.88 5.15 7.15 16.5 22.9
5 0.351 0.543 1.78 2.47 5.46 7.58 16.2 22.5
6 0.393 0.545 2.09 2.90 5.55 7.71 16.9 23.5
7 0.412 0.572 1.38 1.92 5.65 7.84 14.2 19.8
8 0.370 0.514 1.39 1.93 5.82 8.08 14.6 20.3
9 0.431 0.598 1.39 1.92 6.00 8.33 13.5 18.7
10 0.366 0.601 1.49 2.07 5.97 8.29 14.8 20.6
11 0.366 0.628 1.45 2.01 5.97 8.28 14.3 19.9
12 0.350 0.514 1.78 2.10 5.97 8.29 15.0 20.8
13 0.406 0.564 2.20 3.06 5.22 7.24 17.0 23.6
14 0.433 0.601 1.73 2.41 5.18 8.31 15.7 21.8
15 0.432 0.600 1.74 2.42 5.35 8.31 15.8 22.0
16 0.402 0.559 2.19 3.04 5.17 7.17 16.9 23.4
17 0.404 0.561 2.12 2.94 5.19 7.21 16.1 22.3
18 0.402 0.559 2.20 3.06 5.49 7.62 17.0 23.6
19 0.378 0.525 1.75 2.43 5.13 7.12 16.5 23.0
20 0.404 0.561 1.28 1.78 5.24 7.28 14.7 20.4
21 0.431 0.598 1.91 2.65 5.96 8.28 16.5 22.9
22 0.348 0.567 1.46 2.03 5.84 8.11 14.9 20.6
23 0.351 0.600 1.80 2.50 5.96 8.28 15.8 21.9
24 0.366 0.602 2.20 3.05 5.99 8.31 17.0 23.6

Table 5. Technical data of the planning alternatives.

Alternative Number of BESSs Nodes Energy Capacity (kWh)

1 1 14 21
2 2 0, 18 23, 49
3 2 15, 18 39, 49
4 2 10, 11 47, 12
5 2 18, 20 36, 40
6 2 20, 13 28, 11
7 3 14, 17, 10 21, 42, 9
8 2 11, 18 3, 47
9 3 7, 11, 17 34, 36, 19
10 2 9, 17 28, 40
11 2 17, 16 38, 21
12 2 8, 18 24, 34
13 1 10 29
14 1 19 38
15 1 19 21
16 3 10, 1, 8 36, 6, 18
17 2 10, 7 32, 28
18 2 10, 12 15, 43
19 2 13, 10 42, 16
20 2 10, 17 28, 45
21 3 17, 8, 0 12, 49, 16
22 3 13, 17, 8 13, 41, 20
23 3 8, 17, 8 16, 23, 49
24 3 9, 8, 0 15, 49, 28

3.5. Decision Theory-Based Assessment of the Planning Alternatives and Scenarios

Several cases have been considered, with different values of probability of occurrence assigned
to each scenario (Table 6). In Case 1, all scenarios have the same probability of occurrence and, then,
the equal likelihood criterion is considered. Cases 2 and 3 consider that it is more probable that the

69
Energies 2020, 13, 52

scenarios with higher and lower price increases will occur, respectively. Cases 4 and 5 are focused on
the deployment of EVs. In particular, in Case 4, the scenarios with the large deployment of EVs are
weighted more, while in Case 5, the scenarios with no deployment of EVs are weighted more. In Case
6, the probabilities are higher for scenarios with a higher price increase and higher PV installation.
Finally, Case 7 weights the scenarios with a lower price increase and large deployment of EVs more.

Table 6. Values of scenario probabilities.

Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7

Scenario 1 0.125 0.05 0.20 0.05 0.20 0.05 0.15
Scenario 2 0.125 0.20 0.05 0.05 0.20 0.15 0.05
Scenario 3 0.125 0.05 0.20 0.05 0.20 0.05 0.15
Scenario 4 0.125 0.20 0.05 0.05 0.20 0.25 0.05
Scenario 5 0.125 0.05 0.20 0.20 0.05 0.05 0.25
Scenario 6 0.125 0.20 0.05 0.20 0.05 0.15 0.05
Scenario 7 0.125 0.05 0.20 0.20 0.05 0.05 0.25
Scenario 8 0.125 0.20 0.05 0.20 0.05 0.25 0.05

3.5.1. Application of the Criterion of the Minimum Expected Cost

Table 7 reports the expected costs for each alternative in all cases considered. From the values of
the expected costs, the selected planning alternative is Alternative 9 for cases 1–4 and 6–7. Alternative
9 is characterized by lower costs in scenarios 7 and 8 (generally associated with high values of costs);
as such, it is often the selected alternative. The criterion of the minimum expected cost provides
Alternative 20 as the preferred one in Case 5, when scenarios 1–4 are more weighted.

Table 7. Expected costs (×1012 ). Values in bold indicate the best solution for each case.

Alternative Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7

1 7.61 8.36 6.87 11.2 4.07 9.24 7.86
2 7.24 7.94 6.53 10.8 3.67 8.76 7.53
3 7.24 7.94 6.53 10.8 3.66 8.76 7.53
4 7.20 7.91 6.50 10.6 3.76 8.81 7.46
5 7.12 7.82 6.42 10.6 3.62 8.66 7.40
6 7.45 8.18 6.73 11.0 3.87 9.09 7.73
7 6.47 7.10 5.84 9.71 3.23 7.77 6.75
8 6.63 7.28 5.99 10.0 3.28 7.96 6.92
9 6.35 6.97 5.73 9.51 3.19 7.55 6.61
10 6.78 7.45 6.11 10.2 3.39 8.13 7.06
11 6.62 7.27 5.97 9.92 3.32 7.92 6.89
12 6.86 7.50 6.21 10.3 3.45 8.21 7.15
13 7.41 8.14 6.69 10.9 3.90 9.08 7.67
14 7.02 7.78 6.27 10.5 3.58 8.54 7.20
15 7.08 7.83 6.34 10.6 3.61 8.60 7.29
16 7.36 8.07 6.64 10.8 3.87 9.01 7.61
17 7.11 7.80 6.41 10.5 3.75 8.67 7.35
18 7.48 8.21 6.75 11.0 3.93 9.14 7.75
19 7.11 7.80 6.41 10.6 3.61 8.69 7.39
20 6.45 7.08 5.82 9.71 3.18 7.79 6.73
21 7.40 8.12 6.68 11.0 3.80 8.96 7.68
22 6.73 7.39 6.07 10.1 3.35 8.09 7.01
23 7.15 7.86 6.45 10.7 3.65 8.63 7.43
24 7.64 8.39 6.89 11.3 3.99 9.28 7.91

3.5.2. Application of the Minimax Weighted Regret Criterion

Table 8 reports the maximum weighted regrets for each alternative in all cases considered.

70
Energies 2020, 13, 52

Table 8. Maximum weighted regrets (×1012 ). Values in bold indicate the best solution for each case.

Alternative Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7

1 0.560 0.896 0.644 0.896 0.322 1.120 0.805
2 0.486 0.778 0.560 0.778 0.195 0.973 0.700
3 0.486 0.778 0.560 0.778 0.194 0.972 0.700
4 0.534 0.854 0.615 0.854 0.219 1.068 0.768
5 0.483 0.773 0.556 0.773 0.193 0.966 0.696
6 0.605 0.967 0.696 0.967 0.242 1.209 0.870
7 0.136 0.218 0.157 0.218 0.055 0.273 0.196
8 0.206 0.330 0.238 0.330 0.083 0.413 0.297
9 0.151 0.241 0.174 0.241 0.060 0.181 0.217
10 0.242 0.386 0.278 0.386 0.097 0.483 0.347
11 0.154 0.247 0.177 0.247 0.062 0.308 0.221
12 0.270 0.431 0.310 0.431 0.108 0.539 0.388
13 0.617 0.988 0.710 0.988 0.255 1.235 0.888
14 0.391 0.626 0.450 0.626 0.156 0.782 0.562
15 0.413 0.661 0.475 0.661 0.165 0.826 0.593
16 0.596 0.953 0.686 0.953 0.253 1.191 0.857
17 0.457 0.732 0.527 0.732 0.233 0.915 0.658
18 0.612 0.979 0.705 0.979 0.255 1.224 0.881
19 0.537 0.859 0.618 0.859 0.215 1.074 0.772
20 0.212 0.339 0.244 0.339 0.085 0.424 0.305
21 0.527 0.843 0.607 0.843 0.211 1.054 0.759
22 0.243 0.388 0.279 0.388 0.097 0.486 0.349
23 0.406 0.650 0.468 0.650 0.162 0.812 0.585
24 0.616 0.985 0.710 0.985 0.254 1.232 0.887

From the values of the maximum weighted regrets, the selected planning alternative is Alternative
7, for cases 1–5 and 7. The values of the objective function in scenarios with a large deployment of EVs
influence the results also for this criterion, as in 3.5.1.

3.5.3. Application of the “Optimist-Pessimist”Criterion

The “optimist-pessimist” criterion was applied, considering different values of the weighting
factor α ranging from 0 to 1. The case with α = 0 corresponds to the application of the “pessimist”
criterion, while the case with α = 1 is equal to the application of the “optimistic” criterion. The results
are reported in Table 9.

Table 9. Selected planning alternatives obtained by applying the “optimist-pessimist” criterion.

α Selected Planning Alternative

0.0 9
0.1 9
0.2 9
0.3 9
0.4 9
0.5 9
0.6 9
0.7 9
0.8 9
0.9 9
1.0 22

From these results, Alternative 9 is the selected alternative for all the values of the weighting
factor, with the exception of the case with α = 1. The “optimistic” criterion (α = 1) selects Alternative

71
Energies 2020, 13, 52

22, which is associated with the lowest value of the objective function; for the other values of the
weighting factor α, higher values of the objective function in scenarios 7–8 influence the results.

4. Conclusions
This paper has presented a novel procedure that combines planning and scheduling of the BESSs
installed in an LV grid. In this way, BESS siting and sizing is carried out with the support of a
specific assessment of the system operation. The proposed approach combines the properties of the
metaheuristics used to search for solutions in a wide space (for the creation of planning alternatives)
and the fast calculation of the greedy procedure that allows a viable solution to be found for BESS
scheduling. This approach makes it possible to overcome the limitation due to the intractable total
number of combinations of BESS sizes, simultaneously handling the non-trivial operational aspects
linked to variations in time of the power flows in the network and the BESS scheduling. Furthermore,
the uncertainty regarding future scenarios has been handled with a decision theory-based method.
Alternatives 7 and 9 have emerged as the most promising ones when using the decision theory
criteria. In particular, Alternative 9 has been preferred by the expected costs and the “optimist-pessimist”
criteria, in quite a robust way, over most of the cases with weighted scenarios. Alternative 7 has been
preferred by the minimax weighted regrets criterion, again over most of the cases with weighted
scenarios. In these two alternatives, three nodes are chosen for BESS installation (with node 17 in
common) and the BESS sizes in these nodes are intermediate with respect to the minimum and
maximum energy capacity limits used. For a problem of this kind, there can be no guarantee that
a globally optimal solution has been reached. For this reason, the creation of meaningful scenarios
taking into account the local conditions of loads and generation becomes a fundamental aspect that
can be successfully addressed in the case of relatively small electricity communities, where the energy
community manager may have an easier view on the decision variables involved in the system
operation and planning with respect to what occurs in larger systems. To further develop scenarios
of local loads and generation, future research will consider the deployment of other electrification
technologies, like heat pumps, that can significantly impact the demand; it will also consider the
inclusion of three-phase or single-phase PV systems with storage to take into account the trend of the
prosumers to install storage systems in their local plants with the aim of increasing self-sufficiency.

Author Contributions: All the authors contributed equally to this work. All authors have read and agreed to the
published version of the manuscript.
Funding: This research received no external funding.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Zidar, M.; Georgilakis, P.S.; Hatziargyriou, N.D.; Capuder, T.; Škrlec, D. Review of energy storage allocation
in power distribution networks: Applications, methods and future research. IET Gener. Transm. Distrib. 2016,
10, 645–652. [CrossRef]
2. Boicea, V.A. Energy Storage Technologies: The Past and the Present. Proc. IEEE 2014, 102, 1778–1794.
[CrossRef]
3. Chen, S.; Zhang, T.; Gooi, H.B.; Masiello, R.D.; Katzenstein, W. Penetration rate and effectiveness studies of
aggregated BESS for frequency regulation. IEEE Trans. Smart Grid 2016, 7, 167–177. [CrossRef]
4. Brogan, P.V.; Best, R.J.; Morrow, D.J.; McKinley, K.; Kubik, M.L. Effect of BESS response on frequency and
RoCof during underfrequency transients. IEEE Trans. Power Syst. 2019, 34, 575–583. [CrossRef]
5. Brivio, C.; Mandelli, S.; Merlo, M. Battery energy storage system for primary control reserve and energy
arbitrage. Sustain. Energy Grids Netw. 2016, 6, 152–165. [CrossRef]
6. Mosca, C.; Arrigo, F.; Mazza, A.; Bompard, E.; Carpaneto, E.; Chicco, G.; Cuccia, P. Mitigation of Frequency
Stability Issues in Low Inertia Power Systems using Synchronous Compensators and Battery Energy Storage
Systems. IET Gener. Transm. Distrib. 2019, 13, 3951–3959. [CrossRef]

72
Energies 2020, 13, 52

7. Zheng, Y.; Dong, Z.Y.; Luo, F.J.; Meng, K.; Qiu, J.; Wong, K.P. Optimal Allocation of Energy Storage System
for Risk Mitigation of DISCOs with High Renewable Penetrations. IEEE Trans. Power Syst. 2014, 29, 212–220.
[CrossRef]
8. Qiu, J.; Xu, Z.; Zheng, Y.; Wang, D.; Dong, Z.Y. Distributed generation and energy storage system planning
for a distribution system operator. IET Renew. Power Gener. 2018, 12, 1345–1353. [CrossRef]
9. Saboori, H.; Hemmati, R.; Sadegh Ghiasi, S.M.; Dehghan, S. Energy storage planning in electric power
distribution networks—A state-of-the-art review. Renew. Sustain. Energy Rev. 2017, 79, 1108–1121. [CrossRef]
10. Alegria, E.; Brown, T.; Minear, E.; Lasseter, R.H. CERTS Microgrid Demonstration with Large-Scale Energy
Storage and Renewable Generation. IEEE Trans. Smart Grid 2014, 5, 937–943. [CrossRef]
11. Zheng, Y.; Zhao, J.; Song, Y.; Luo, F.; Meng, K.; Qiu, J.; Hill, D.J. Optimal Operation of Battery Energy
Storage System Considering Distribution System Uncertainty. IEEE Trans. Sustain. Energy 2018, 9, 1051–1060.
[CrossRef]
12. Hussain, A.; Bui, V.-H.; Kim, H.M. Impact Analysis of Demand Response Intensity and Energy Storage Size
on Operation of Networked Microgrids. Energies 2017, 10, 882. [CrossRef]
13. Kumar, A.; Meena, N.K.; Singh, A.R.; Deng, Y.; Kumar, P. Strategic integration of battery energy storage
systems with the provision of distributed ancillary services in active distribution systems. Appl. Energy 2019,
253, 113503. [CrossRef]
14. Sedghi, M.; Ahmadian, A.; Aliakbar-Golkar, M. Optimal Storage Planning in Active Distribution Network
Considering Uncertainty of Wind Power Distributed Generation. IEEE Trans. Power Syst. 2016, 31, 304–316.
[CrossRef]
15. Ahmadian, A.; Sedghi, M.; Aliakbar-Golkar, M. Fuzzy Load Modeling of Plug-in Electric Vehicles for Optimal
Storage and DG Planning in Active Distribution Network. IEEE Trans. Veh. Technol. 2017, 66, 3622–3631.
[CrossRef]
16. Wong, L.A.; Ramachandaramurthy, V.K.; Taylor, P.; Ekanayake, J.B.; Padmanaban, S. Review on the optimal
placement, sizing and control of an energy storage system in the distribution network. J. Energy Storage 2019,
21, 489–504. [CrossRef]
17. Andreotti, A.; Carpinelli, G.; Mottola, F.; Proto, D.; Russo, A. Decision Theory Criteria for the Planning
of Distributed Energy Storage Systems in the Presence of Uncertainties. IEEE Access 2018, 6, 62136–62151.
[CrossRef]
18. Tang, Z.; Liu, J.; Liu, Y.; Huang, Y.; Jawad, S. Risk awareness enabled sizing approach for hybrid energy
storage system in distribution network. IET Gener. Transm. Distrib. 2019, 13, 3814–3822. [CrossRef]
19. Pavić, I.; Luburić, A.; Pandžić, H.; Capuder, T.; Andročec, I. Defining and Evaluating Use Cases for Battery
Energy Storage Investments: Case Study in Croatia. Energies 2019, 12, 376. [CrossRef]
20. Yunusov, T.; Frame, D.; Holderbaum, W.; Potter, B. The impact of location and type on the performance of
low-voltage network connected battery energy storage systems. Appl. Energy 2016, 165, 202–213. [CrossRef]
21. Giannitrapani, A.; Paoletti, S.; Vicino, A.; Zarrilli, D. Optimal Allocation of Energy Storage Systems for
Voltage Control in LV Distribution Networks. IEEE Trans. Smart Grid 2017, 8, 2859–2870. [CrossRef]
22. Crossland, A.F.; Jones, D.; Wade, N.S. Planning the location and rating of distributed energy storage in LV
networks using a genetic algorithm with simulated annealing. Int. J. Electr. Power Energy Syst. 2014, 59,
103–110. [CrossRef]
23. Jannesar, M.R.; Sedighi, A.; Savaghebi, M.; Guerrero, J.M. Optimal placement, sizing, and daily
charge/discharge of battery energy storage in low voltage distribution network with high photovoltaic
penetration. Appl. Energy 2018, 226, 957–966. [CrossRef]
24. Fortenbacher, P.; Ulbig, A.; Andersson, G. Optimal Placement and Sizing of Distributed Battery Storage in
Low Voltage Grids Using Receding Horizon Control Strategies. IEEE Trans. Power Syst. 2018, 33, 2383–2394.
[CrossRef]
25. Khalid Mehmood, K.; Khan, S.U.; Lee, S.; Haider, Z.M.; Rafique, M.K.; Kim, C. Optimal sizing and allocation
of battery energy storage systems with wind and solar power DGs in a distribution network for voltage
regulation considering the lifespan of batteries. IET Renew. Power Gener. 2017, 11, 1305–1315. [CrossRef]
26. Babacan, O.; Torre, W.; Kleissl, J. Optimal allocation of battery energy storage systems in distribution
networks considering high PV penetration. In Proceedings of the 2016 IEEE Power and Energy Society
General Meeting (PESGM), Boston, MA, USA, 17–21 July 2016.

73
Energies 2020, 13, 52

27. Carpinelli, G.; Mottola, F.; Proto, D.; Russo, A.; Varilone, P. A Hybrid Method for Optimal Siting and Sizing
of Battery Energy Storage Systems in Unbalanced Low Voltage Microgrids. Appl. Sci. 2018, 8, 455. [CrossRef]
28. French, S. Decision Theory, an Introduction to the Mathematics of Rationality; Ellis Horwood: Chichester, UK,
1989.
29. IRENA. Electricity Storage and Renewables: Costs and Markets to 2030; October 2017; ISBN 978-92-9260-038-9.
Available online: https://www.irena.org/publications/2017/Oct/Electricity-storage-and-renewables-costs-
and-markets (accessed on 17 December 2019).
30. Von Appen, J.; Braun, M.; Stetz, T.; Diwold, K.; Geibel, D. Time in the sun: The challenge of high PV
penetration in the German electric grid. IEEE Power Energy Mag. 2013, 11, 55–64. [CrossRef]
31. Mortazavi, H.; Mehrjerdi, H.; Saad, M.; Lefebvre, S.; Asber, D.; Lenoir, L. A monitoring technique for reversed
power flow detection with high PV penetration level. IEEE Trans. Smart Grid 2015, 6, 2221–2232. [CrossRef]
32. De Carne, G.; Buticchi, G.; Zou, Z.; Liserre, M. Reverse Power Flow Control in a ST-Fed Distribution Grid.
IEEE Trans. Smart Grid 2018, 9, 3811–3819. [CrossRef]
33. Anders, G.J. Probability Concepts in Electric Power Systems; John Wiley & Sons: New York, NY, USA, 1990.
34. Miranda, V.; Proenca, L.M. Probabilistic choice vs. risk analysis-conflicts and synthesis in power system
planning. IEEE Trans. Power Syst. 1998, 13, 1038–1043. [CrossRef]
35. Kubiak, P.; Cen, Z.; López, C.M.; Belharouak, I. Calendar aging of a 250 kW/500 kWh Li-ion battery deployed
for the grid storage application. J. Power Sources 2017, 372, 16–23. [CrossRef]
36. OpenEI. Available online: https://openei.org/datasets/files/961/pub/ (accessed on 17 December 2019).
37. Spertino, F.; Corona, F.; Di Leo, P. Limits of Advisability for Master–Slave Configuration of DC–AC Converters
in Photovoltaic Systems. IEEE J. Photovolt. 2012, 2, 547–554. [CrossRef]
38. World Weather Online. Available online: https://www.worldweatheronline.com/ (accessed on
17 December 2019).

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

74
 energies
Article
Planning of Distributed Energy Storage Systems in
μGrids Accounting for Voltage Dips
Fabio Mottola 1 , Daniela Proto 1 , Pietro Varilone 2 and Paola Verde 2, *
1 Department of Electrical Engineering and Information Technology, University of Naples Federico II,
I-80125 Naples, Italy; fabio.mottola@unina.it (F.M.); daniela.proto@unina.it (D.P.)
2 Department of Electrical Engineering and Information “Maurizio Scarano”, University of Cassino and
Southern Lazio, I-03043 Cassino, Italy; varilone@unicas.it
* Correspondence: verde@unicas.it

Received: 3 December 2019; Accepted: 10 January 2020; Published: 14 January 2020

Abstract: This paper deals with the optimal planning of the electrical energy storage systems in the
microgrids aimed at cost minimization. The optimization accounts for the compensation of the voltage
dips performed by the energy storage systems. A multi-step procedure, at the first step, identifies a
set of candidate buses where the installation of a storage device produces the maximum benefit in
terms of dip compensation; then, the life cycle costs in correspondence of different alternatives in
terms of size and location of the storage systems are evaluated by considering an optimized use of
the energy storage systems. The simulations on a medium voltage microgrid allowed validating the
effectiveness of the proposed procedure.

Keywords: electrical energy storage systems; voltage dips; power quality; microgrid planning

1. Introduction
The role of the systems for storing the electrical energy is gaining more and more importance in
the frame of modern power systems. This is thanks to the electrical energy storage systems’ (EESSs)
ability to provide a number of benefits across multiple levels [1–3]. Focusing on the distribution level
of electrical energy systems, the EESSs’ benefits are mainly related to the compensation action of the
intermittent effects of renewable power sources and to the support to the operation of the network by
providing services aimed at regulating voltage levels, at reducing losses, and deferring the investment
on the distribution system [4–6]. End-users can also benefit from EESSs through reduction of the cost
for the energy purchased as well as for the improvement of power quality (PQ) and reliability [7].
The convenience of the installation of the EESSs is still a critical issue and is strictly related to the
number of benefits that can be achieved contemporaneously from their installation [8]. In the case of
the microgrids (μGs), cost reduction and PQ improvement are matters of particular interest for both
grid operators and end-users.
Cost reduction refers to the ability of the μG owner to increase the share of power produced
from renewables inside the μG thus reducing the cost sustained for the imported power, as well as to
increase the system efficiency and reduce the network losses. PQ improvement refers to possibility of
reducing the level of the PQ disturbances at least up to the ranges admissible for the loads fed by the
μG. The typical characteristics of a μG make the objective of the PQ improvement more difficult to
attain than in the traditional distribution systems. The network structure, the lengths of the lines, the
installed powers and the levels of short circuit power jointly contribute to limiting the PQ robustness
of a μG, defined as its intrinsic capacity to maintain assigned disturbance levels when the external
conditions change [9,10]. The improvement of the performance of a μG in terms of voltage dips, one
of the most critical PQ disturbances, represents an attractive challenge for different reasons. First of
all, the voltage dips are the disturbances with serious consequences on processes and activities that

Energies 2020, 13, 401; doi:10.3390/en13020401 75 www.mdpi.com/journal/energies

Energies 2020, 13, 401

could evolve in higher sustained costs. Secondly, these costs are important not only for the industrial
users, for which are certainly the most critical [11], but also for commercial and service activities. Loss
of production, damages to equipment, halt of processes and data losses are some of the effects of
voltage dips, which unavoidably evolve into significant financial losses [12]. Further, the detrimental
effects of voltage dips can regard also residential customers, which can suffer for uncomfortable supply
service. Also, in the frame of smart grids, most of the new smart technologies are based on electronic
devices and control systems, which are particularly sensitive. Finally, it is worth emphasizing that the
importance of the improvement of the voltage dip performance is also proved by the fact that most of
the regulations on the quality of voltage by the national authorities in Europe started with actions on
voltage dips, even if only with wide measurement campaigns like in Italy [13].
In this context, where assuring high voltage dips performance is a primary issue, a possible
solution could be given by installing a number of EESSs in the μG, able to fast respond to the occurrence
of voltage dips. Clearly, the installation of EESSs is not without drawbacks, particularly those related to
the cost of this technology. Thus, EESSs need to be optimally sized by maximizing their benefits while
minimizing costs and need to be installed in those nodes considered strategical for the maximization
of their value in terms of dip compensation.
In the relevant literature, great effort has been paid to the optimal planning of EESSs in μGs with
the objective of the cost reduction. Optimal planning of EESSs in the distribution grids is proposed
in [6], based on the minimization of the investment and operation costs. In [14], a planning procedure
is proposed, which takes into account the minimization of the cost of energy imported from the
external grid while considering voltage support and minimization of network losses. The problem
of the optimal siting and sizing of EESSs in μGs is faced in [15] through the minimization of both
installation and operation costs, as well as power losses. Minimization of investment and operation
costs in distribution networks is considered in the planning tool proposed in [16], which accounts
for constraints on bus voltages. Costs of both power losses and EESS’ operation are included in the
analytical planning tool presented in [17]. In [18], the optimal allocation of EESSs in distribution
networks is performed on the basis of a tradeoff among investment and operation costs, technical
constraints and network losses. A planning tool is proposed in [19] aimed at the minimization of
installation and operation costs while accounting for the network technical constraints in the presence
of uncertainties. The mitigation of voltage dips thanks to the installation of the storage devices is
analyzed in [20], where a cost-benefit analysis is proposed. To the best of our knowledge, approaches
that include the benefits related to voltage dip compensation within the EESS planning has never been
proposed in the current technical literature.
In this paper, a minimum cost strategy for the planning of EESSs in a μG is proposed where
the cost is avoided thanks to the voltage dip compensation being taken into account. At this aim,
the optimal siting and sizing of EESSs in the μG is formulated in terms of a multi-period, non-linear,
constrained optimization problem. A multi-step approach is proposed based on the identification
of a limited set of candidate buses among the most exposed to the voltage dips. An optimization
tool is used to identify the total cost related to the possible design alternatives (size and location of
EESSs). The selection of the candidate buses allows optimizing EESSs’ installation while reducing
computational complexity. The reduced computational burden allows analyzing all the possible design
alternatives and choosing the optimal solution as the one corresponding to the maximum benefit.
Compared to the existing technical literature, the main contribution of this paper in terms of novelty
refers to the inclusion of the voltage dips’ cost in the planning of EESSs in μGs.
The paper is organized as follows. The problem formulation of the proposed planning method is
reported in Section 2. In Section 3, the method used to identify the set of candidate buses is described.
Section 4 reports the optimization tool used to derive the total cost of the planning solutions. Some
results related to the application of the proposed method to a case study based on an MV test system
are summarized and discussed in Section 5. Our conclusions are drawn in Section 5 together with
some comments on future research work.

76
Energies 2020, 13, 401

2. Problem Formulation
The planning problem refers to a three-phase, MV, balanced, μG including residential, industrial
and commercial loads. The μG is connected to an upstream grid through the Point of Common
Coupling (PCC). The distributed resources including Distributed Generation (DG) units and EESSs are
connected to the μG through power converters, and are supposed to be owned by the μG operator.
The contribution of both the DG units and EESSs is the provision of active power to supply
part of the loads and, reactive power to support the μG operation. The active power profile of the
EESSs (that is, their charging/discharging profile), and the reactive power of both DG units and EESSs
are determined on the basis of a minimum cost strategy. More specifically, the strategy is aimed at
minimizing the cost sustained by the μG operator for the energy imported from the upstream grid. The
size and location of the DG units are assigned whereas, the size and location of the EESSs are identified
according to the proposed planning problem based on the minimization, a total cost function. The
total cost function includes the installation cost of the EESSs, the operation cost of the μG, and the
benefit derived from the voltage dip compensation over the whole planning period.
The planning problem is developed in four steps, as shown in the flow chart in Figure 1. The
first step refers to the selection of the candidate nodes, chosen as those resulting among the most
vulnerable in terms of exposure to the voltage dips. For this reason, these nodes correspond to the sites
of the installation of the EES units with the preferred performance in terms of dip compensation. The
procedure for the selection is presented in Section 4.

Identification of candidate nodes (selected as those

most exposed to the voltage dips)

Definition of all the possible configurations in terms

of size and locations of the batteries (planning
alternatives)

For each planning alternative, the total cost is

evaluated (including the cost avoided trough dip
compensation).

Planning solution: alternative corresponding to the

maximum benefit.

Figure 1. Flow chart of the proposed planning procedure.

The second step refers to the identification of all the possible configurations in terms of size and
location for the EESS installation (limiting the installation only to the set of nodes derived from the
Step 1). For each of the possible configurations (planning alternatives), the total cost sustained by the
μG owner is evaluated according to the procedure detailed in the Section 5 (third step).
Finally, in the fourth step, the best planning alternative is selected as that corresponding to the
maximum benefit (BF) derived by the installation of the EESSs. This BF can be evaluated for each

77
Energies 2020, 13, 401

alternative by comparing the total cost sustained by the μG’s owner in absence of the EESSs (Cno
T
EESS )

and that resulting when the EESSs are installed (CEESS

T
):

BF(Ai ) = Cno
T
EESS
(Ai ) − CEESS
T (Ai ) i = 1 . . . NA (1)

where NA is the number of the design alternatives. The total cost evaluated when the EESSs are
installed includes the reduction of the cost item related to the voltage dips and of that related to the
grid operation thanks to the price arbitrage.

3. Identification of the Set of Buses Most Exposed to the Voltage Dips

A fault in a node can cause a voltage dip in one or more nodes of the electrical network in function
of several parameters that determine the electrical power system response to a short circuit. These
parameters are linked to the structural characteristics of the network and to the features of the installed
components. Examples of such “hardware characteristics” of the electrical system are the network
configuration, the installed generator powers, and the size, length and type of the lines.
The numbers of the voltage dips or their frequency in a time horizon, usually one year, is instead
linked to the frequency of the causes, the short circuits, which originated them.
The choice of the candidate nodes where the μG operator can place a compensating unit of the
voltage dips has been made according to the hardware characteristics of the network that constitutes
the μG. In particular, the set of candidate nodes are those nodes that mostly experience the voltage
dips whose residual voltage is below a critical value, Vcr , regardless the frequency of the faults which
originated these voltage dips.
This choice allows ascertaining the sensitiveness of the nodes directly to the effects of the short
circuits in any node of the network in terms of voltage dips linked to the characteristics of the network
and of the installed components. Consequently, for any possible configuration of the network (radial,
meshed, ring) the set of the candidate nodes is different for the same Vcr .
Inside a μG, the figure Vcr represents the limit value of vulnerability of the loads supplied by the
system, that is the minimum value below which the equipment experiences a trip, the most critical
effect of a voltage dip.
Theoretically, the value of Vcr should vary in function of the specific load fed at every node of
the μG. However, a more realistic scenario in a planning stage allows accounting for the vulnerability
limits of classes of equipment, as indicated by the International Standards IEC [21]. In the case of
equipment supplied by the μG are of Class II or Class III, Vcr corresponds to the limit value of 70% or
40% of the declared voltage, respectively, for the voltage dips lasting up to 200 ms.
Summarizing, for every possible configuration of the μG network, the set of candidate busses for
the installation of a compensating units is the set of busses which experience the largest number of
voltage dips whose residual voltage is below Vcr , regardless the frequency of the faults which caused
these voltage dips.
The Fault Position Method (FPM) is the most effective tool for the selection of the candidate bus
set in the given assumptions.
In fact, for a system with N nodes the FPM allows  obtaining
 the (NxN) matrix of the during fault
voltages in any node for short circuit in every node, V d f , by means of the following equation for a
three-phase fault in every node of the system:
    .   .  −1  
V d f = Ep f − ZSC diag ZSC diag Ep f (2)

78
Energies 2020, 13, 401

.   
where ZSC is the nodal short circuit impedance matrix, Ep f is the pre-fault voltage matrix and
  .  −1
diag ZSC is the matrix formed by the inverse of the diagonal elements of the short circuit impedance
matrix. If the pre-fault voltages are assumed to be 1 p.u., relation (2) can be written as:
  .   .  −1
V d f = [ones] − ZSC diag ZSC diag[ones] (3)
. 
where [ones] is a matrix full of ones and such that dimension is equal to the dimension of ZSC .
 
Any element V ik of V d f with magnitude less than 0.9 p.u. represents a voltage dip in the node
i caused by a three-phase short circuit in the node k. For different types of short circuits, similar
equations can be drawn [22].  
For every network configuration of the μG with N nodes, starting from the matrix V d f obtained
by the FPM, the set of candidate busses is chosen as constituted by the m nodes such that:

V i,k ≤ Vcr , i = 1, . . . , m and k = 1, . . . , N. (4)

As mentioned above, the value of Vcr in the equation (4) is chosen from the Standard (e.g., [23]).

4. Total Cost Evaluation

For each alternative, the total costs, CT , related to the inclusion of the EESSs in the μG is given by
the sum of the cost of installation, Cinst , and replacement, Crep , of the EESSs, the operation costs of the
μG, Cop , and the cost due to voltage dips.

CT (Ai ) = Cinst (Ai ) + Crep (Ai ) + Cop (Ai )+ Cvd (Ai ) (5)

All the cost items refer to the whole planning period and are detailed in the following sub-sections.
In (5), maintenance cost can be also added as a percentage of the installation cost. For ease of notation,
in the following equations, the reference to alternative Ai is omitted.

4.1. Installation Cost

The installation cost includes both the cost of the battery and the cost of the power converter
interfacing the storage device to the grid. The cost of the battery depends on the energy capacity,
whereas the cost of the power converter is related to the rated power. In case of EESSs, the rated energy
and power are linked through the nominal C-rate. Then, a unitary installation cost can be provided
once the nominal C-rate has been specified. In this case, the installation cost of the EESSs is given by
the product of the capacity unitary installation cost, ICEESS , and the battery size. Obviously, the former
depends on the considered battery technology and the latter refers to the specific design alternative.
The installation cost is then given by:


nb
Cinst = ICEESS Esize
EESS (6)
i
i=1

where nb is the number of installed EESSs and Esize

EESS
is the size of the ith EESS.
i

4.2. Replacement Cost

Based on the expected battery lifetime, which is given in terms of number of charging/discharging
cycles—and that depends on the battery’s stress factors, the replacement of the device could be
necessary during the considered planning period. The replacement cost is given by:

79
Energies 2020, 13, 401


nb
Crep = ri RCEESS Esize
EESS (7)
i
i=1

where RCEESS is the unitary cost for the battery replacement, and ri indicates the number of times the
ith battery needs to be replaced. To evaluate ri , the battery’s lifetime duration must be evaluated and
compared to the planning period. The battery’s lifetime, in years, (Lb ) can be evaluated based on the
lifecycle (i.e., life expressed in terms of number of charging/discharging cycles), Ncycles , and on the
number of cycles per day, ncycles,d :
Ncycles
Lb = (8)
365 ncycles,d
Ncycles represents the total number of cycles the battery can be used before the replacement; it strictly
depends on the battery technology and on the way stress factors act [24,25]. Its value is provided by the
battery manufacturer, and it is related to specified operation conditions, such as ambient temperature
and depth of discharge (DoD). The value of ncycles,d depends on the way the battery is operated, which
depends on the operation strategy.

4.3. Operation Cost

The operation cost of the μG refers to the cost of the energy imported from the upstream grid. It is
supposed that the μG is not able to sell energy to the upstream grid. The cost of the energy imported
implies the evaluation of the overall costs

- sustained by the μG’s owner to (i) supply the loads, (ii) charge the EESSs, (iii) compensate for the
power losses;
- avoided thanks to the EESS discharged energy.

The planning time horizon includes ny years, each represented by nd y typical days. Forecasted
daily profiles of the power requested by the loads, of the power delivered by the distributed generation,
and those of the energy price are known for each typical day of the first year. These profiles are
known with reference to all the nt time intervals—of duration Δt—in which each day is divided. The
corresponding profiles for the subsequent years are evaluated based on a specified yearly growth. The
operation cost for the whole planning period is then provided by the net present value of the sum of
the costs sustained at each typical day:
⎡ nt ⎤

ny
1 
nd y
⎢⎢  ⎥⎥
Cop = N y,d ⎢⎢⎢⎣ P1,k( y,d) Δt Prk( y,d) ⎥⎥⎥⎦ (9)
y=1 (1 + a) y−1 d=1 k =1

where a is the discount rate, N y,d is the number of days represented by the dth typical day in the
year y, P1,k( y,d) and Prk( y,d) are the power imported from the upstream network and the energy price,
respectively, at the kth time interval of the dth typical day of the year y.
In order to evaluate the power imported from the upstream grid, P1,k( y,d) , a minimum cost strategy
is formulated for the μG, on the basis of a two-stage procedure which allows decoupling the hourly
optimization algorithm so reducing the computational burden.
In the first stage, the hourly active power of the EESSs is evaluated. According to the cost
minimization approach, the value of the power of each EESS is derived by charging the battery during
low-price hours and discharging it during the high price hours, while constraints are imposed on the
rated power and energy capacity of the EESSs.
In the second stage an OPF is solved, at each time interval, with respect to the known variables
(i.e., specified values of loads’ active and reactive powers and active powers of DG units and EESSs)
and unknown variables (i.e., active and reactive power at the PCC, reactive powers of both DG units

80
Energies 2020, 13, 401

and EESSs.) Coherently with the cost minimization strategy, the OPF minimizes the power imported
at the PCC while satisfying the constraints on both the μG and the interfacing converters.
a. 1th Stage: EESS Profile Optimization
The power that can be charged or discharged by the EESSs at each time interval depends on the
stored energy available in the battery, on the rated power and on the price of energy corresponding to
that interval. A this aim, two sets of time intervals are identified in a day: the first set is related to the
highest price—where the EESSs can be discharged to reduce the costs of the imported power—Ωdch( y,d) ,
the second set is related to the lowest price—where the EESSs can be charged—Ωch( y,d) . Two different
iterative procedures are applied in Ωdch( y,d) and Ωch( y,d) (Figure 2).

Define set Ǎௗ௖௛ሺ೤ǡ೏ሻ of time Define set Ǎ௖௛ሺ೤ǡ೏ሻ of time

intervals for discharging intervals for charging

Sort the set Ǎௗ௖௛ሺ೤ǡ೏ሻ according to Sort the set Ǎ௖௛ሺ೤ǡ೏ሻ according to

decreasing price of energy decreasing price of energy

݇ൌͳ ݄ൌͳ

Evaluation of the maximum power that Evaluation of the maximum power that
can be discharged according to the SoC: can be charged according to the SoC:
௔௨௫ǡௗ௖௛ ͳ ௔௨௫ǡ௖௛ ͳ
ܲ௕ǡ௧ ൌ ߜ‫ܧ‬௕௥௧ௗ ܲ௕ǡ௧ ൌ െ ݁௕
భሺ೤ǡ೏ሻ ο‫ݐ‬ భሺ೤ǡ೏ሻ ο‫ݐ‬

Evaluation of the battery power: Evaluation of the battery power:

௔௨௫ǡௗ௖௛ ௔௨௫ǡ௖௛
ܲ௕ǡ௧ೖሺ೤ǡ೏ሻ ൌ ݉݅݊ሼܲ௕௥௧ௗ ǡ ܲ௕ǡ௧ ೖ ሺ೤ǡ೏ሻ
ሽ ܲ௕ǡ௧೓ሺ೤ǡ೏ሻ ൌ െ݉݅݊ሼܲ௕௥௧ௗ ǡ ܲ௕ǡ௧ ೓ ሺ೤ǡ೏ሻ
ሽ

Update the maximum power that can Update the maximum power that can

be discharged be charged
݇ ൌ݇൅ͳ ݄ ൌ݄൅ͳ
௧ೖ ௧೓
௔௨௫ǡௗ௖௛ ͳ ͳ ௔௨௫ǡ௖௛ ͳ
ܲ௕ǡ௧ ൌ ߜ‫ܧ‬௕௥௧ௗ െ ෍ ܲ௕ǡ௝ሺ೤ǡ೏ሻ ܲ௕ǡ௧ ൌ ݁௕ሺ೤ǡ೏ሻ െ ߟ௖௛ ෍ ܲ௕ǡ௝ሺ೤ǡ೏ሻ
ೖశభ ሺ೤ǡ೏ሻ ο‫ݐ‬ ߟௗ௖௛ ೓శభ ሺ೤ǡ೏ሻ ο‫ݐ‬
௝ୀ௧భ ௝ୀ௧భ

yes yes yes yes

௔௨௫ǡௗ௖௛ ഥ ௔௨௫ǡ௖௛
ܲ௕ǡ௧ ೖశభ
൐Ͳ ݇൏‫ܭ‬ ܲ௕ǡ௧ ൐Ͳ ഥ
݄൏‫ܪ‬
೓శభ

not not
not not
Power profile for the discharging: Power profile for the charging stage:
ܲ௕ǡ௞ሺ೤ǡ೏ሻ ‫  א ݇׊‬Ǎௗ௖௛ሺ೤ǡ೏ሻ ܲ௕ǡ௛ሺ೤ǡ೏ሻ ‫  א ݄׊‬Ǎ௖௛ሺ೤ǡ೏ሻ
Energy discharged: Energy charged:
ͳ
݁௕ሺ೤ǡ೏ሻ ൌ ෍ ܲ௕ǡ௞ሺ೤ǡ೏ሻ ο‫ݐ‬ ݁Ԣ௕ሺ೤ǡ೏ሻ ൌ െߟ௖௛ ෍ ܲ௕ǡ௛ሺ೤ǡ೏ሻ ο‫ݐ‬
ߟௗ௖௛
௞‫א‬Ǎ೏೎೓ሺ೤ǡ೏ሻ ௛‫א‬Ǎ೎೓ሺ೤ǡ೏ሻ

(a) (b)
Figure 2. Flow chart of the iterative procedure for the determination of the EESS profile during
(a) Ωdch( y,d) and (b) Ωch( y,d) .

81
Energies 2020, 13, 401

The iterative procedure that applies to Ωdch( y,d) allows obtaining the EESS’s profile during the
discharging stage (Figure 2a). The procedure is based on the identification of the time intervals
included in Ωdch( y,d) and on the identification of the set Ωdch which includes the time intervals of
( y,d)
Ωdch( y,d) sorted by decreasing price of energy. Then, at each time interval tk included in Ωdch , with
( y,d)
k = 1, . . . , K, the power discharged by the battery is identified as the minimum between the EESS rated
power (Prtd b
) and the power corresponding to the battery SoC (Paux,dch
b,t
):
k

! "
Pb,tk ( y,d) = min Prtd
b
, Paux,dch
b,t
(10)
k

tk ∈ Ω dch( y,d) y = 1, . . . , ny, d = 1, . . . , nd y

With reference to the first interval in which the battery can be discharged, k = 1, it is assumed that
the battery is fully charged, thus
1 rtd
Paux,dch = δE (11)
b,t1 Δt b
where Ertd
b
is the rated energy capacity of the battery, and δ is the admissible DoD. Once the battery
power is known through (10), the power corresponding to the SoC is updated:

1 
tk
1 rtd
Paux,dch = δEb − Pb, j( y,d) (12)
b,tk +1 Δt ηdch
j=t1

where ηdch is the discharging efficiency of the EESS. The procedure ends when all of the time intervals
in Ωdch are explored (i.e., k = K) or Paux,dch
b,t
≤ 0. Outputs of the procedure are the EESSs’ power
( y,d) k +1
profiles during the discharging stage (i.e., Pb,k( y,d) ∀ k ∈ Ωdch( y,d) ) and the total energy which can be
discharged by the battery:
1 
eb( y,d) = Pb,k( y,d) Δt (13)
ηdch
k∈Ωdch
( y,d)

The iterative procedure, which applies in Ωch( y,d) allows obtaining the EESS’s profile during the
charging stage (Figure 2b). The procedure is based on the identification of the time intervals included
in Ωch( y,d) and on the identification of the set Ωch which includes the time intervals of Ωch( y,d) sorted
( y,d)
by increasing price of energy. Then, at each time interval th included in Ωch , with h = 1, . . . , H, the
( y,d)
discharged power of the battery is identified as the minimum between the EESS rated power (Prtd
b
) and
the power corresponding to the battery SoC (Paux,dch
b,t
):
h

Pb,th ( y,d) = −min{Prtd

b
, Paux,ch
b,t
}
h
(14)
th ∈ Ω ch( y,d) y = 1, . . . , ny, d = 1, . . . , nd y

With reference to the first interval in which the battery can be charged, h = 1, it is assumed that
the battery is empty, thus
1
Paux,ch = − eb( y,d) (15)
b,t1 Δt
where eb( y,d) is given by (13). Once the battery power is known through (10), the power corresponding
to the SoC is updated:
1 
thh
Paux,dch = e b − η ch Pb, j( y,d) (16)
b,th+1 Δt ( y,d)
j=th1

82
Energies 2020, 13, 401

Where ηch is the charging efficiency of the EESS. The procedure ends when all of the time intervals
in Ωch are explored (i.e., h = H) or Paux,ch
b,t
≤ 0. Outputs of the procedure are the EESS’s power
( y,d) h+1
profile during the charging stage (i.e., Pb,h( y,d) ∀ h ∈ Ωch ) and the total energy, which can be charged by
the battery: 
eb( y,d) = −ηch Pb,h( y,d) Δt (17)
h∈Ωch
( y,d)

In order to make the strategy feasible, it is required that during each day, the energy charged and
discharged must be the same:
eb( y,d) = eb( y,d) (18)

In the case that (18) is not satisfied, the procedure for the discharging stage (Figure 2) must be
repeated by replacing δErtd
b
= eb( y,d) .
b. 2nd Stage: Optimal Power Flow
The Optimal Power Flow (OPF) in its general form is formulated as:

min fobj (x) (19)

ψυ (x) = 0 υ = 1, . . . , neq (20)

φγ (x) ≤ 0 γ = 1, . . . , nineq (21)

where the minimization of the objective function fobj is subject to equality and inequality constraints,
ψ and φ applied to the vector x of the optimization variables. Inputs of the OPF are the forecasted
values of the power load demand, the DG power production, the energy price and the EESS power.
Outputs are the reactive power of the DG units and of the EESSs, and the active and reactive power at
the PCC. All these quantities are required to be known at all the k time intervals of each typical day d
of year y. In order to minimize the operation costs defined in (9), the objective function is:
 
fobj (x) = P1,k( y,d) Δt Prk( y,d) k = 1, . . . , nt, y = 1, . . . , ny, d = 1, . . . , nd y (22)

The constraints to be satisfied in the OPF include the classical power flow equations:


n     
Pi,k( y,d) = Vi,k( y,d) V j,k( y,d) Gi,j cos δi, j,k( y,d) + Bi, j sin δi, j,k( y,d) (23)
j=1


n     
Qi,k( y,d) = Vi,k( y,d) V j,k( y,d) Gi,j sin δi, j,k( y,d) − Bi, j cos δi, j,k( y,d) (24)
j=1

V1,k( y,d) = Vspec (25)

δ1,k( y,d) = 0 (26)

k = 1, . . . , nt, y = 1, . . . , ny, d = 1, . . . , nd y

where n is the number of network buses and, with reference to the time interval k of the day d in
the year y, δi,j,k( y,d) is the difference between the phase angles at nodes i and j, Pi,k( y,d) and Qi,k( y,d) are
the active and reactive powers at bus i, Vi,k( y,d) is the ith bus voltage amplitude, Gi, j and Bi, j are the
(i, j)-terms of the matrices of the conductance and susceptance, respectively. Constraints (25) and (26)
refer to slack bus (i = 1), that is the PCC.
The active and reactive powers in (19) and (20) include the powers absorbed by the loads, the
power charged/discharged by the EESSs and the power produced by the DG units.

83
Energies 2020, 13, 401

The active and reactive power exchanged at the PCC must comply with the transformer rate, Srtd
tr ,
that is: # $
2  2 1/2
P1,k( y,d) + Q1,k( y,d) ≤ Srtd
tr (27)

k = 1, . . . , nt, y = 1, . . . , ny, d = 1, . . . , nd y

The same applies to the power flowing through the converters interfacing the EESSs and the DG
units, which cannot exceed the converter rated power, Srtd
DESS,i
, Srtd
DG,i
:

# 2  2 $1/2
Pb,i,k( y,d) + Qb,i,k( y,d) ≤ Srtd
DESS,i (28)

i ∈ ΩDESS
# 2  2 $1/2
Pdgi,k( y,d) + Qdg,i,k( y,d) ≤ Srtd
DG,i (29)

i ∈ ΩDG

k = 1, . . . , nt, y = 1, . . . , ny, d = 1, . . . , nd y

where Pb,i,k( y,d) has been previously derived in Step 1, Qb,i,k( y,d) is the EESS reactive power, and Pdgi,k( y,d)
and Qdg,i,k( y,d) are the DG active and reactive powers, ΩDESS and ΩDG are the sets of buses, where the
EESSs DG units are connected, respectively.
Further constrains refer to the limits imposed to the bus voltage and line current magnitudes:

Vmin ≤ Vi,k( y,d) ≤ Vmax (30)

k = 1, . . . , nt, y = 1, . . . , ny, d = 1, . . . , nd y

Il,k( y,d) ≤ Il,max (31)

l ∈ Ωl , k = 1, . . . , nt, y = 1, . . . , ny, d = 1, . . . , nd y

being Vmin and Vmax the minimum and maximum voltage magnitude values, Il,k( y,d) the current flowing
through the lth line of the set Ωl of network lines, which cannot exceed the maximum value, Il,max (i.e.,
line ampacity).
In this case, the 2nd stage does not converge due to constraint violation, it will be iteratively
applied by reducing the EESS contribution (i.e., by reducing Pb,i,k( y,d) , k = 1, . . . , nt), till convergence
is reached.

4.4. Cost of Voltage Dips

The cost of the voltage dips is strictly connected to the economic value the effects of the voltage
dips have on the equipment’s and the operating processes or the activities. The most critical effect of the
voltage dips is the trip of the device that is subject to the dip. The economic value of this detrimental
effect represents, in turn, the cost of the voltage dip, and depends on the function of the device into the
process or activity, the type of the process or activity, the linkage of the process in case stopped with
other processes of the same production line, in case of industrial manufacturing loads, or with other
operative functions, in case of loads different from manufacturing industries.
Generally, the cost of the voltage dips is the sum of three main components: direct, indirect and
hidden costs [23]. The direct costs relate to the interruptions of the specific device or equipment. They
include, for example, lost work, lost production, damaged equipment; the indirect costs include the
investment costs sustained by the end user to prevent or to solve the damages due the interruptions
caused by voltage dips. Finally, the hidden costs account for any second level effects that reflect on the
performance of the business, such as retaining customers, satisfying customers, and protecting the
company’s reputation. The hidden costs in some studies are included inside the indirect costs [26].

84
Energies 2020, 13, 401

The cost of the voltage dips can be estimated by means of direct methods or indirect methods.
The direct methods are analytical methods that require at least the availability of the voltage dips
measured at the load site, the deep knowledge of several characteristics of the devices or equipment
(dip susceptibility curve, operating mode, functional linkage of each device with other devices of the
same equipment, and of the same process, structure and topology of the electric lines feeding each
device and equipment), and the detection of the areas of the process exposed to the stoppage. They
are usually conducted as after-the-fact-case studies of the events of voltage dips occurred at a specific
site [26–28]. A direct method can also use the voltage dips expected at the site feeding the sensitive
loads obtained by means of the simulation of the electric system in short circuit conditions. The two
main methods available and widely used are the fault position method (FPM) and the critical distance
method (CDM). The most adequate method to use depends on several parameters, as reported in
the comparative study of [8]. Whichever of these methods obviously requires the knowledge of the
electrical system feeds the load sites.
The indirect methods estimate the cost of a disturbance without passing through the analytical
quantification of the effects of that disturbance on specific devices or equipment’s. They can use
macroeconomic analysis [11] or surveys of the customers asked to directly estimate their suffered
costs due to the supply disturbances or to provide qualitative economic figures of the electric supply
linked to that disturbance. In the case of the supply interruptions, for example, the National economic
regulation of continuity in Italy started with an extensive survey aimed to estimate two indices: the
willingness to pay (WTP) and the willingness to accept (WTA) [29]. WTP measures the additional
price of the electric service the users would pay for avoiding the interruptions; WTA instead measures
the amount of money they would accept for having to experience the outage. These indices allowed
fixing the incentive rate for the penalties and the rewards in the first regulatory scheme of the National
Authority of electricity ARERA, formerly AEEG [30].
The existence of the economic regulation of a disturbance could solve the problem of establishing
the costs for the loads, especially in the planning activities with long-term horizons, as in the case of
this study. In such cases, in fact, the specific details of the loads are not obtainable and would not be
reliable to assign them.
Unfortunately, no economic regulation of the voltage dips is still active [31]. However, some
preliminary studies [32,33] offer valuable figures resulting from measurements, survey and simulations,
which can guide in deriving the economic value to assign to the voltage dips.
In this paper, we referred to [32] where the authors analyzed the economic and technical data
collected for a number of medium-voltage (MV) industrial users that participated in the cost assessment
project and the monitoring campaign. In particular, the authors used an indirect method based on
a monitoring-and-survey approach to estimate the direct costs faced by a large variety of industrial
customers. As detailed in the paper, the costs were presented both as annual value per kW and as
event value for kW, with the latter independent from frequency. In this paper, we used the event value
per kW of the installed power of the load.
For the evaluation of the total costs of the voltage dips over a long-time horizon, the frequency of
occurrence of the voltage dips must be evaluated. The annual frequency can be estimated in average
from data measured on real systems. In this paper, we referred to the measured data on the MV Italian
National Grid (with reference to year 2016) at the MV busbars of the HV/MV substations (Table 1) [34].
Finally, the cost related to voltage dips over the planning period, Cvd , can be evaluated as:


nvd 
ny
1 
Cvd = y−1
nvdrv,y cvdrv,y Pi,y (32)
i=1 y=1 (1 + a) rvΩrv

where nvd is the number of busses sensitive to the voltage dips, n y is the planning time horizon (years),
a is the discount rate and, with reference to the voltage dip belonging to the residual voltage class, rv,
cvdrv,y is the cost suffered by the customer due to the single event of voltage dip at the year y, nvdrv,y is

85
Energies 2020, 13, 401

the number of the voltage dip events at year y, and Ωrv is the set of residual voltage classes which the
voltage dip event refers to (i.e., first column of Table 1). Pi,y is the power value to be considered for the
dip cost identification, with reference to the user connected to the ith bus of the μG at year y. In this
paper, we referred to the load nominal value.

Table 1. Annual Average Voltage Dip Number for MV substations connected to the Italian National
Grid (data taken from [34]).

Annual Average Voltage Dip Number

Residual Voltage [%] Duration of the Voltage Dips [ms]
20–200 200–500 0.5–1 × 103 1–5 × 103 5–60 × 103
80 ≤ u ≤ 90 33.93 4.35 0.93 0.34 0.05
70 ≤ u ≤ 80 12.91 3.01 0.38 0.21 0.07
40 ≤ u ≤ 70 17.07 3.95 0.31 0.11 0.03
5 ≤ u ≤ 40 5.22 1.39 0.12 0.02 0.00
1≤u≤5 0.27 0.05 0.07 0.03 0.10
Total 69.4 12.74 1.82 0.72 0.25

We assumed that the EEC units compensate only to voltage dips at the node where the EES is
installed. This assumption is very cautious, since the beneficial effects of the compensating action
can be wider, involving also further nodes electrically close to the installation node of the EES. With
structures of the network other from radial, the detection of the area, in which the voltage dips are
compensated by the EES installed in one node, requires the simulation of the system in short circuit
conditions with EES acting as voltage dip compensating unit. A FPM, applied in the presence of the
EES, giving as result the propagation of the voltage dips for short circuits in every node of the system,
could allow obtaining this result.

5. Numerical Applications
The planning procedure has been applied to an MV μG, which refers to the MV Cigrè benchmark
system [35]. The system under study is a 12.47 kV, three-phase, balanced, distribution network
(Figure 3). This network is constituted by 15 buses and it is connected to an upstream HV network by
means of two 115/12.47 kV transformers of 25 and 20 MVA, which connect the PCC (bus #1) to two
feeders: the first feeder is connected at the secondary side of the 25 MVA transformer (bus #2); the
second feeder is connected at the secondary side of the 20 MVA transformer (bus #13). The network
includes three switches: S1 (between buses #9 and #15), S2 (between buses #7 and #8) and S3 (between
buses #5 and #12).
The loads connected to the network are grouped in residential and commercial/industrial and are
listed in Table 2. The loads are assumed to have a yearly growth of 1%. The per unit power profiles
assumed for residential and industrial/commercial customers are reported in Figure 4. Two PV units
are connected to the buses #9 and #15 with rated powers of 5 MW and 2 MW, respectively. Figure 5
shows the per unit power profile of the PV generation at bus #9.
The assumed energy pricing tariff refers to the hourly values of the Italian market price. With
reference to the first year, the price values refer to a day of November 2019 [36] (Figure 6). A yearly
growth of 1% has also been assumed for the energy prices. Regarding the cost, which the customers
have to sustain due to the single voltage dip event, the value 2.9 €/kW [32] has been assumed with a
yearly growth of 1% for the industrial/commercial loads; for the residential loads, instead, the value
of zero was assumed. The number of events is evaluated according to the data reported in Table 1.
Particularly, the events related to the threshold voltage dip value, Vcr , of 70% has been assumed.

86
Energies 2020, 13, 401

HV network

2 13

3
14
4
15
5 9

6 S1
8
10

7
11
S2
12
S3

Figure 3. The microgrid under study.

Table 2. Load Input data.

Rated Power Rated Power

Bus # Load Type cos ϕ Bus # Load Type cos ϕ
(MVA) (MVA)
13.8 Residential 0.93 8 0.30 Comm./industrial 0.95
2
9.16 Comm./industrial 0.87 0.25 Residential 0.90
9
0.35 Residential 0.95 0.20 Comm./industrial 0.90
3
0.80 Comm./industrial 0.85 10 0.35 Residential 0.95
0.25 Residential 0.90 11 0.50 Residential 0.90
4
0.24 Comm./industrial 0.80 0.10 Residential 0.95
12
5 0.40 Residential 0.90 0.45 Comm./industrial 0.85
0.20 Residential 0.95 3.20 Residential 0.90
6 13
0.30 Comm./industrial 0.85 3.78 Comm./industrial 0.87
7 0.15 Residential 0.95 14 0.68 Comm./industrial 0.85
8 0.10 Residential 0.95 15 0.27 Comm./industrial 0.90

87
Energies 2020, 13, 401

Power (kW)

Figure 4. Load profile at bus #6.

2

1.8

1.6

1.4
Power (MW)

1.2

0.8

0.6

0.4

0.2

0
0 4 8 12 16 20 24
Hour

Figure 5. PV power profile at bus #15.

Market price (€/MWh)

Figure 6. Market energy price (data taken from [36]).

Regarding the EESSs to install, nickel manganese cobalt Li-ion battery technology has been
considered suitable for this application. The data of the various EESS’ features are reported in
Table 3 [37], which refer to values expected for 2020. Regarding the battery replacement cost, the value
of the cost expected for the 2025 has been considered. Note that the costs in Table 3 have been converted
in Euro (conversion rate of November 2019) before applying them in the proposed procedure.

88
Energies 2020, 13, 401

Table 3. EESS Data (data taken from [37]).

Parameter Value Parameter Value

Battery round trip efficiency 99 % Battery Cycle life 4800 cycles
Maximum Depth of Discharge 100 % Converter round trip efficiency 98 %
Battery installation cost 153 $/kWh Converter installation cost 59.6 $/kW
Battery replacement cost 110 $/kWh Maintenance cost 1.5%

The planning design alternatives refer to the installation of two EESSs whose size must be selected
among the rated power values 0, 5, 10 and 15 MW with a discharging nominal time of five hours.
In order to show the ability of the method to satisfy the technical constraints imposed on the
EESSs and on the μG, Figures 7 and 8 report the daily profiles of the active power of an EESS and of
the voltage values at the buses where an EESS and a PV unit are connected. The figures refer to the
first year of the planning period of the case study 1 and detail the profiles of the EESS connected at the
bus #6 and of the PV unit connected at the bus #15.

15

10

5
Power (kW)

-5

-10

-15
0 4 8 12 16 20 24
Hour

Figure 7. Active power of the EESS connected at the bus #6 (case study 1).

1.06
EESS
PV
1.04

1.02

0.98

0.96

0.94

0.92
0 4 8 12 16 20 24
Hour

Figure 8. Voltage values along a day of the first year of the planning period at the bus where EESS (bus
#6) and PV unit (bus #16) are connected (case study 1).

During the μG operation, the breakers can be operated open or closed depending on the operator
decisions aimed at avoiding undesired conditions or based on economic strategies. The states of the
breakers strongly affect the values that line currents and bus voltages can assume. In order to consider
the influence of the states of the breakers in terms of exposure of the buses to the voltage dips, the
proposed planning procedure has been applied with reference to the following case studies:

89
Energies 2020, 13, 401

- Case 1: S1 , S2 , S3 : open
- Case 2: S1 , S2 , and S3 : closed
- Case 3: S1 : closed, S2 , S3 : open
- Case 4: S1 , S3 : open, S2 : closed
- Case 5: S1 , S2 : open, S3 : closed
- Case 6: Case 1 (0.45), Case 2 (0.10), Case 3 (0.15), Case 4 (0.15), Case 5 (0.15)

Case 6 refers to the probability of occurrence of different network configurations (the probability
of occurrence of each configuration is indicated within the brackets).
The application of the procedure of Section 3 for the identification of candidate buses based
on voltage dip sensitiveness leads to the candidate buses reported in Table 4 that correspond to the
threshold voltage dip value of 70%. The results of the planning procedure corresponding to each
case study are reported in Table 5. In the table, the BF of each planning solution is reported together
with the reduction of the cost of voltage dips obtained thanks to the installation of the EESSs. For
comparative purposes, the planning results are reported for both the cases in which the reduction of
voltage dip cost is taken into account and those in which this cost reduction is neglected. Negative
values of BF are obtained in the cases where installation of EESSs is not convenient.

Table 4. Candidate buses (70% threshold values).

Case Study
Case 1 Case 2 Case 3 Case 4 Case 5 Case 6
3 6 8 3 3 8
4 7 9 4 4 9
5 8 10 5 5 10
6 9 11 6 6 11

Table 5. Siting and sizing of EESSs.

Case Accounting Voltage Dip EESSs’ Nodes EESSs’ Sizes (MW)

for Voltage BF (M$) Cost Reduction
Study EESS n. 1 EESS n. 2 EESS n. 1 EESS n. 2
Dip Costs (M$)
Yes 4.81 1.63 3 6 10 10
1
No 3.19 - 3 6 10 10
Yes −0.16 0.44 6 - 5 -
2
No −0.59 - 9 - 5 -
Yes −0.34 0.31 9 - 5 -
3
No −0.65 - 9 - 5 -
Yes 4.75 1.63 3 6 10 10
4
No 3.16 - 3 4 15 5
Yes 4.70 1.63 3 6 10 10
5
No 3.21 - 3 5 5 15
Yes 4.46 0.50 8 10 15 10
6
No 3.96 - 8 10 15 10

The results reported in Table 5 show that the convenience of installing EESSs in the μG basically
depends on the cost reduction obtained thanks to the price arbitrage. In fact, in the cases where
the price arbitrage does not allow installing EESSs (cases 2 and 3), the added benefits derived from
the voltage dip compensation is not sufficient to justify the adoption of the storage devices. This is
probably due to the fact that, being the candidate buses mainly characterized by residential loads, the

90
Energies 2020, 13, 401

benefit derived by dip compensation in terms of cost reduction is not significant (we remind that the
cost of voltage dip for the residential load was neglected). Moreover, in the cases where it is convenient
to install the EESSs (cases 1, 4, 5, and 6) the beneficial effect of dip compensation clearly appears only
when they are used in buses with large industrial loads. In fact, in cases 4 and 5, the installation of the
EESS n. 2 in bus #6 instead of bus #4 and in bus #6 instead of bus #5, respectively, allows obtaining
higher benefits.
In order to better prove the effect of the dip compensation on the total cost reduction, in Table 6, the
percentage values of the benefit related to the voltage dip cost reduction are reported, with reference
to the cases in which the planning procedure provides a solution. It is interesting to note that the
percentage benefit due to the voltage dip cost reduction falls within the range [33–35] % in cases 1, 4,
and 5. In case 6, the percentage value is lower (about 11%). This is mainly due to the fact that in case 6,
the solution associated to the largest total benefit corresponds to allocation buses (buses #8 and #10)
with voltage dip costs lower than those corresponding to the nodes selected in cases 1, 4, and 5.

Table 6. Benefit corresponding to the voltage dip cost reductions.

Percentage of Value of the Voltage Dip

Percentage Value of the Voltage Dip Cost
Case Study Cost Reduction on the Installation and
Reduction on the Total Benefit (%)
Replacement Cost (%)
1 33.89 6.55
4 34.32 6.55
5 34.68 6.55
6 11.21 1.61

In Table 6, the percentage values of the benefit corresponding to the voltage dip reduction
compared to the EESS’s installation and replacement costs are also reported. The value of about 6%
is obtained in the cases 1, 4 and 5. A lower value (about 2%), instead, is obtained in case 6. This is
due to two aspects: the first refers again to the reduced benefit in terms of voltage dip cost reduction
(compared to the other cases), the second refers to the increased size of storage, which is allocated in
this case study.
Figure 7 clearly shows that the maximum power that is charged/discharged by the EESS is always
lower than the rated power. Moreover, the comparison of Figures 6 and 7, reveals that the battery
absorbs power during the lowest price hours, and injects power to the grid during the highest price
hours. This is coherent with the considered minimum cost strategy. The SoC profile—here not reported
for the sake of conciseness—also satisfies the constraints on the rated battery capacity.
Figure 8 shows that the voltage at the critical buses, i.e., the buses where an EESS or a DG are
connected, falls within admissible ranges (0.9–1.1) p.u. The increased variation of the voltage in the
nodes where an EESS is connected which appears in the figure is due to the charge/discharge of the
EESS. In the node where a DG is connected, instead, a smooth profile appears; this is due to the power
rating of the generation system lower than that of the EESS and to the presence of the high load power
in the buses close to this node.
Still referring to the case study 1—first year of the planning period—Figure 9 shows the active
power imported at the PCC. Coherently with the minimum cost strategy, thanks to the use of the EESSs,
most of the requested power is shifted from the highest price hours to the lowest price hours.

91
Energies 2020, 13, 401

30

25

20

15

10

0
0 4 8 12 16 20 24
Hour

Figure 9. Active power imported from the whole μG along a day of the first year of the planning period.

6. Conclusions
In this paper, the planning of the electrical energy storage systems in the microgrids was considered.
A cost minimization approach was proposed including the benefits obtained through the voltage dip
compensation. Particularly, the cost reduction is obtained by shifting the load from the hours of high
energy price to those of low energy prices and by considering the voltage dip cost avoided thanks to
the compensation action. The motivation of this study is twofold. The first relies on the importance of
power quality aspects in the modern power systems characterized by widespread use of electronic
devices and control systems which are highly vulnerable to the voltage dips. The latter refers to
the benefits derived from the ability the storage systems have in providing multiple services, which
concur to sustain their high installation cost. The proposed method is based on a multi-step planning
tool which allows identifying the optimal location and size of the storage systems. The results of the
numerical simulations clearly showed that voltage dip compensation allowed obtaining non-negligible
economic benefits. Clearly, these benefits strictly depend on the cost assumed for the voltage dip
events, which can be specified only through a deep knowledge of the end-users. Future works will be
aimed at improving the procedure in order to better identify the areas of the network that achieve a
beneficial effect in terms of dip compensation from the installation of EESSs.

Author Contributions: F.M., D.P., P.V. (Pietro Varilone) and P.V. (Paola Verde) conceived and designed the
theoretical methodology and the numerical applications; F.M., D.P., P.V. (Pietro Varilone) and P.V. (Paola Verde)
performed the numerical applications and analyzed the results; F.M., D.P., P.V. (Pietro Varilone) and P.V. (Paola
Verde) wrote and revised the paper. All authors have read and agreed to the published version of the manuscript.
Funding: Financial support by Italian Ministry of University and Research (MIUR) through the Special Grant
“Dipartimenti di eccellenza” Lex. N. 232/31.12.2016 (G.U. n. 297/21.12.2016 S.O. n. 57).
Acknowledgments: The authors thank Guido Carpinelli (University of Naples Federico II) for his invaluable and
generous contribution in the continuous development of new ideas to deal with increasingly interesting problems
of modern power systems. The authors P. Varilone and P. Verde acknowledge the financial support by Italian
Ministry of University and Research through the Special Grant “Dipartimenti di eccellenza”.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Ai Wong, L.; Ramachandaramurthy, V.K.; Taylor, P.; Ekanayake, J.B.; Walker, S.L.; Padmanaban, S. Review on
the optimal placement, sizing and control of an energy storage system in the distribution network. J. Energy
Storage 2019, 21, 489–504. [CrossRef]
2. Hemmati, R.; Shafie-Khah, M.; Catalão, J.P.S. Three-Level Hybrid Energy Storage Planning Under Uncertainty.
IEEE Trans. Ind. Electron. 2019, 66, 2174–2184. [CrossRef]

92
Energies 2020, 13, 401

3. Zhang, N.; Li, R.; Jiang, Y. Cost-benefits analysis of battery storage system for industry consumers based
on different operation modes. In Proceedings of the 2018 IEEE 2nd International Electrical and Energy
Conference (CIEEC), Beijing, China, 4–6 November 2018; pp. 527–531. [CrossRef]
4. Saboori, H.; Hemmati, R.; Sadegh Ghiasi, S.M.; Dehghan, S. Energy storage planning in electric power
distribution networks—A state-of-the-art review. Renew. Sustain. Energy Rev. 2017, 79, 1108–1121. [CrossRef]
5. Carpinelli, G.; Celli, G.; Mocci, S.; Mottola, F.; Pilo, F.; Proto, D. Optimal Integration of Distributed Energy
Storage Devices in Smart Grids. IEEE Trans. Smart Grid 2013, 4, 985–995. [CrossRef]
6. Grover-Silva, E.; Girard, R.; Kariniotakis, G. Optimal sizing and placement of distribution grid connected
battery systems through an SOCP optimal power flow algorithm. Appl. Energy 2018, 219, 385–393. [CrossRef]
7. Carpinelli, G.; Mottola, F.; Proto, D. Probabilistic sizing of battery energy storage when time-of-use pricing is
applied. Electr. Power Syst. Res. 2016, 141, 73–83. [CrossRef]
8. Electricity Energy Storage Technology Options: A White Paper Primer on Applications Costs and Benefits.
2010. Available online: https://www.epri.com/ (accessed on 12 January 2020).
9. Carpinelli, G.; Di Perna, C.; Caramia, P.; Varilone, P.; Verde, P. Methods for Assessing the Robustness of
Electrical Power Systems Against Voltage Dips. IEEE Trans. Power Deliv. 2009, 24, 43–51. [CrossRef]
10. Caramia, P.; Varilone, P.; Verde, P.; Vitale, L. Tools for Assessing the Robustness of Electrical System against
Voltage Dips in terms of Amplitude, Duration and Frequency. In Proceedings of the International Conference
on Renewable Energies and Power Quality (ICREPQ’14), Cordoba, Spain, 8–10 April 2014.
11. Küfeoğlu, S.; Lehtonen, M. Macroeconomic Assessment of Voltage Sags. Sustainability 2016, 8, 1304.
[CrossRef]
12. Ding, Z.; Zhu, Y.; Chen, C. Economic loss assessment of voltage sags. In Proceedings of the CICED 2010
Proceedings, Nanjing, China, 13–16 September 2010; pp. 1–5.
13. 6th CEER Benchmarking Report on the Quality of Electricity and Gas Supply; CEER, 2016. Available online:
https://www.ceer.eu/1305/ (accessed on 12 January 2020).
14. Nick, M.; Cherkaoui, R.; Paolone, M. Optimal siting and sizing of distributed energy storage systems via
alternating direction method of multipliers. Int. J. Electr. Power Energy Syst. 2015, 72, 33–39. [CrossRef]
15. Nojavan, S.; Majidi, M.; Esfetanaj, N.N. An efficient cost-reliability optimization model for optimal siting
and sizing of energy storage system in a microgrid in the presence of responsible load management. Energy
2017, 139, 89–97. [CrossRef]
16. Lin, Z.; Hu, Z.; Zhang, H.; Song, Y. Optimal ESS allocation in distribution network using accelerated
generalised Benders decomposition. IET Gener. Transm. Distrib. 2019, 13, 2738–2746. [CrossRef]
17. Fantauzzi, M.; Lauria, D.; Mottola, F.; Scalfati, A. Sizing energy storage systems in DC networks: A general
methodology based upon power losses minimization. Appl. Energy 2017, 187, 862–872. [CrossRef]
18. Nick, M.; Cherkaoui, R.; Paolone, M. Optimal Allocation of Dispersed Energy Storage Systems in Active
Distribution Networks for Energy Balance and Grid Support. IEEE Trans. Power Syst. 2014, 29, 2300–2310.
[CrossRef]
19. Andreotti, A.; Carpinelli, G.; Mottola, F.; Proto, D.; Russo, A. Decision Theory Criteria for the Planning
of Distributed Energy Storage Systems in the Presence of Uncertainties. IEEE Access 2018, 6, 62136–62151.
[CrossRef]
20. Celli, G.; Pilo, F.; Pisano, G.; Soma, G.G. Including voltage dips mitigation in cost-benefit analysis of storages.
In Proceedings of the 2018 18th International Conference on Harmonics and Quality of Power (ICHQP),
Ljubljana, Slovenia, 13–16 May 2018; pp. 1–6. [CrossRef]
21. IEC 61000-4-11:2004. Testing and Measurement Techniques—Voltage Sags, Short Interruptions and Voltage
Variations Immunity Tests. 2004. Available online: https://webstore.iec.ch (accessed on 12 January 2020).
22. Carpinelli, G.; Caramia, P.; Di Perna, C.; Varilone, P.; Verde, P. Complete Matrix Formulation of Fault Position
Method for Voltage Dip Characterization. IET Gener. Transm. Distrib. 2007, 1, 56–64. [CrossRef]
23. IEEE Standard 1346. Recommended Practice for Evaluating Electric Power System Compatibility with Electronic
Process Equipment; The Institute of Electrical and Electronics Engineers, Inc.: New York, NY, USA, 1998.
24. Xu, B.; Oudalov, A.; Ulbig, A.; Andersson, G.; Kirschen, D.S. Modeling of Lithium-Ion Battery Degradation
for Cell Life Assessment. Trans. Smart Grid 2018, 9, 1131–1140. [CrossRef]
25. Han, X.; Lu, L.; Zheng, Y.; Feng, X.; Li, Z.; Li, J.; Ouyang, M. A review on the key issues of the lithium ion
battery degradation among the whole life cycle. ETransportation 2019, 1, 100005. [CrossRef]

93
Energies 2020, 13, 401

26. Fumagalli, E.; Lo Schiavo, L.; Delestre, F. Service Quality Regulation in Electricity Distribution and Retail;
Springer: Berlin/Heidelberg, Germany, 2007.
27. Di Fazio, A.R.; Duraccio, V.; Varilone, P.; Verde, P. Voltage sags in the automotive industry: Analysis and
solutions. Electr. Power Syst. Res. 2014, 110, 25–30. [CrossRef]
28. Weldemariam, L.; Cuk, V.; Cobben, J. Cost Estimation of Voltage Dips in Small Industries Based on Equipment
Sensitivity Analysis. Smart Grid Renew. Energy 2016, 7, 271–292. [CrossRef]
29. Bertazzi, A.; Fumagalli, E.; Lo Schiavo, L. The use of customer outage cost surveys in policy decision-making:
The Italian experience in regulating quality of electricity supply. In Proceedings of the CIRED 2005—18th
International Conference and Exhibition on Electricity Distribution, Turin, Italy, 6–9 June 2005; pp. 1–5.
30. Delibera n. 155/02, Testo Integrato delle Disposizioni Dell’autorità per L’energia Elettrica e il Gas in Materia di
Continuità del Servizio di Distribuzione Dell’energia Elettrica, G.U. Serie Generale n. 201 del 28 Agosto 2002.
Available online: https://www.arera.it/it/docs/02/155-02.htm (accessed on 12 January 2020). (In Italian)
31. Bollen, M.; Beyer, Y.; Styvactakis, E.; Trhulj, J.; Vailati, R.; Friedl, W. A European Benchmarking of voltage
quality regulation. In Proceedings of the 2012 IEEE 15th International Conference on Harmonics and Quality
of Power, Hong Kong, China, 17–20 June 2012; pp. 45–52.
32. Delfanti, M.; Fumagalli, E.; Garrone, P.; Grilli, L.; Lo Schiavo, L. Toward Voltage-Quality Regulation in Italy.
IEEE Trans. Power Deliv. 2010, 25, 1124–1132. [CrossRef]
33. Weldemariam, L.E.; Cuk, V.; Cobben, J.F.G. A proposal on voltage dip regulation for the Dutch MV distribution
networks. Int. Trans Electr. Energy Syst. 2019, 29, e2734. [CrossRef]
34. Relazione Annuale Sullo Stato dei Servizi e Sull’ Attività Svolta. AEEGSI Annual Report AEEGSI. 2018. Available
online: www.autorita.energia.it (accessed on 12 January 2020). (In Italian)
35. Benchmark Systems for Network Integration of Renewable and Distributed Energy Resources, Cigré Task Force C6.04,
Cigré Brochure 575. 2014. Available online: https://e-cigre.org/ (accessed on 12 January 2020).
36. Gestore Mercati Energetici. Esiti Mercato Elettrico, November 2019. Available online: https://www.
mercatoelettrico.org/it/ (accessed on 12 January 2020).
37. IRENA. Electricity Storage and Renewables: Costs and Markets to 2030, International Renewable Energy
Agency, Abu Dhabi. 2017. Available online: http://www.irena.org/publications/2017/Oct/Electricity-storage-
and-renewables-costs-and-markets (accessed on 12 January 2020).

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

94
 energies
Article
Optimal Energy Storage System Positioning and
Sizing with Robust Optimization
Nayeem Chowdhury, Fabrizio Pilo * and Giuditta Pisano
Department of Electrical and Electronic Engineering, University of Cagliari, 09123 Cagliari, Italy;
nayeem.chowdhury@enel.com (N.C.); giuditta.pisano@unica.it (G.P.)
* Correspondence: fabrizio.pilo@unica.it; Tel.: +39-320-437-2957

Received: 2 December 2019; Accepted: 17 January 2020; Published: 21 January 2020

Abstract: Energy storage systems can improve the uncertainty and variability related to renewable
energy sources such as wind and solar create in power systems. Aside from applications such as
frequency regulation, time-based arbitrage, or the provision of the reserve, where the placement
of storage devices is not particularly significant, distributed storage could also be used to improve
congestions in the distribution networks. In such cases, the optimal placement of this distributed
storage is vital for making a cost-effective investment. Furthermore, the now reached massive
spread of distributed renewable energy resources in distribution systems, intrinsically uncertain
and non-programmable, together with the new trends in the electric demand, often unpredictable,
require a paradigm change in grid planning for properly lead with the uncertainty sources and the
distribution system operators (DSO) should learn to support such change. This paper considers
the DSO perspective by proposing a methodology for energy storage placement in the distribution
networks in which robust optimization accommodates system uncertainty. The proposed method
calls for the use of a multi-period convex AC-optimal power flow (AC-OPF), ensuring a reliable
planning solution. Wind, photovoltaic (PV), and load uncertainties are modeled as symmetric and
bounded variables with the flexibility to modulate the robustness of the model. A case study based
on real distribution network information allows the illustration and discussion of the properties of the
model. An important observation is that the method enables the system operator to integrate energy
storage devices by fine-tuning the level of robustness it willing to consider, and that is incremental
with the level of protection. However, the algorithm grows more complex as the system robustness
increases and, thus, it requires higher computational effort.

Keywords: decision-making; distribution network planning; uncertainty; robust optimization; energy

storage system

1. Introduction
The share of renewable power generation in the global electricity generation is anticipated to
expand from today’s 23% to levels between 30%–45% by 2030 [1]. This technological alteration requires
a rethinking in the way power systems are planned, maximize the benefits from renewables affordably
and securely. Since renewable energy integration brings new challenges into the distribution network
planning an accurate planning model, which incorporates system uncertainty introduced by renewable
resources and loads, is necessary for making planning decision.
The technological development of large-scale electrochemical energy storage system (ESS) has
resulted in capital cost reductions and increased roundtrip efficiency enables them to become a feasible
option to deploy in the distribution network [2,3]. Storage applications such as energy arbitrage [4],
peak shaving [5], frequency regulation [6], voltage support [7], and congestion management [8]
have made it vital to integrate more ESS in the distribution network. Thus, optimal planning and

Energies 2020, 13, 512; doi:10.3390/en13030512 95 www.mdpi.com/journal/energies

Energies 2020, 13, 512

management of ESS are essential to identify ideal configurations. However, many of the optimization
algorithms proposed in recent literature do not adequately deal with uncertainties. For instance, the
real amount and position of distributed generation (DG) that is going to be connected to the system,
the mix of renewable energy sources (RES), the cost of ESS or the level of participation and the cost for
active demand [9].
Sizing of ESSs in distribution networks from DSO has been discussed in [10]. The number and
locations of the ESSs are assumed to be given. An AC-optimal power flow (AC-OPF) with semidefinite
programming (SDP) convex relaxation is adopted for network simulation. To consider uncertainties
in the model, a stochastic optimization approach has been considered. Two different problems have
been formulated respectively for the siting and sizing of ESS in distribution networks coupled with a
wind farm in [11]. However, the authors considered a linearized DC-optimal power flow (DC-OPF),
and the wind power forecast is assumed perfect. Apart from siting and operation, the authors of [12]
suggest the life cycle payment of storage. They present two models for a transmission-constrained
power network with storage. Both models use a DC-OPF framework. The first model selects optimal
siting and operation of the storage assuming a fixed group of different storage technologies. The
second model expands the DC-OPF framework to optimize the storage technology mix, new storage
capacity investments, and the network allocation of these resources. The authors of [13] provide a
mathematical model that simultaneously optimizes transmission switching operations, ESS siting
and sizing decisions and taking into account the limits on maximum allowable load shedding and
renewable energy curtailment amounts in the power system. The methodology proposed in [4], based
on a linearized DC-OPF, captured both the monetary and technical advantages of investment in storage
and adopted a sensitivity analysis to assess the impact of uncertain parameters. As opposed to an
analytical approach, the authors of [14] detail a heuristic approach for finding the optimal location(s)
and size of a multi-purpose ESS including transmission and distribution parts without considering the
uncertainty in the model. In the transmission storage part, a sensitivity analysis is performed using
complex-valued neural networks (CVNN) and time domain power flow (TDPF) to obtain the optimal
ESS location(s). In [15] a multi-criteria approach where a genetic algorithm (NSGA-II) has been used to
identify the optimal place, size, and scheduling of energy storage in the distribution network. The
authors created a full multi-objective (MO) optimization procedure able to identify the Pareto set of
design options with fixed network topology for a given medium voltage (MV) network. In addition
to that, the same authors of [15] have proposed a multi-criteria analysis approach selecting the best
planning alternative for energy storage integration in the distribution system [16]. However, heuristic
techniques often required a high computational burden and are not guaranteed to converge in global
optima [17].
Convex relaxation techniques have been developed to obtain an acceptable solution while
ensuring algorithmic efficiency. The two most commonly used relaxations for distribution network are
semi-definite program (SDP) and second-order cone programming (SOCP). Though both SOCP and
SDP have been proven exact under certain conditions [18,19]. In this paper, SOCP has been adopted
due to its higher algorithmic performances that imply fast convergence to global optima and to reduce
the heavy computation cost.
The current literature on energy storage study is divided into three classifications: (i) storage
sizing, (ii) storage operation, and (iii) storage siting. Less publications exist about the optimal location(s)
of the ESS than publications on optimal sizing likely due to the difficulty of finding optimal sites [20,21].
Storage siting is the least researched and most complicated of these three classifications. The optimal
operation studies of ESS consider that energy and power ratings of a storage unit are given, the purpose
of these studies is to identify operation strategies to optimize the exploitation of resources able of
contributing to network support at minimum cost. These studies typically do not address network
constraints. The optimal ESS size (i.e., energy and power ratings) depends on the state at which the
storage is optimally operated [22]. In turn, optimal storage siting depends on the amount of the storage
being considered and how it will be controlled. This problem becomes even more complicated when

96
Energies 2020, 13, 512

it considers distributed storage rather than a single storage unit. If this is the case, the amount of
storage located is typically undefined at first. ESSs are considered one of the solutions to manage the
DG downsides and to help incorporate RES in the distribution networks, which rely largely on the
flexibility of resources [23,24].
Capital intensity is the main barrier to the deployment of ESS [25]. Investment in the ESS, therefore,
requires a trade-off between long-term investment costs, short-term operating conditions, and the
benefits that these services will offer. It includes optimizing their network position (i.e., location)
and operating parameters (e.g., size and operating profile) simultaneously [26]. The objective of the
optimization is generally multiple since it spans from the reduction of CAPEX (Capital Expenditures)
for network upgrade to the reduction of energy loss costs as well as power quality costs [27]. RES
and electric vehicles (EVs) integration as well as the engagement of customers in flexibility programs
are opening new opportunities for ESS and making clear the need for optimal siting and sizing
methodologies [28].
ESS technologies can operate on different timescales, ranging from seconds to hours. The
services offered by ESS can be divided into power- and energy-related services, based on the timescale
of interest [29]. Transient stability and ancillary services, such as frequency regulation, spinning
reserve, and voltage control are power-related services. Back-up power provision, black-start,
uninterruptible power supply (UPS), standing reserve, and seasonal energy storage are typical
examples of energy-related services [28,30]. Both ESS owners and other system stakeholders can
benefit from the provision or the usage of these services.
The ESS optimal positioning and sizing problem aims at the maximization of the benefit-cost ratio
subject to the non-linear/non-convex network constraints that make the solution more cumbersome
and requires specific mathematical tool [31]. The research on the topic has been dramatically increasing
in the last two years since ESS are crucial for the energy transition towards the carbon free world, but
there is still room for new contributions, particularly on dealing with the uncertainties modeling.
For these reasons, the paper proposes an application of the Robust Optimization (RO) to solve
the ESS optimal location problem in distribution networks operated by a DSO. The objective of the
optimization problem is to use ESS for delivering power without any violation of technical limits
(e.g., maximum and minimum nodal voltage), minimizing the resort to RES or combined heat and
power (CHP) generation curtailment and load shaving. Indeed, the DSO can evaluate the ESS
installation as a non-network option to avoid the expensive and time consuming building or revamping
of networks, particularly in the current situation of limited markets of services offered by customers
or other producers. However, the convenience to do this strictly depends on the site, size, and
operation of the ESS, that in turn, depend on the state of the network, that is intrinsically uncertain.
The approach proposed in the paper deals with the uncertainties with an original implementation of
the Robust Optimization. The algorithm has been validated with an exemplary distribution network
representative of one class of the Italian distribution classes of networks produced by the project
ATLANTIDE (Archivio TeLemAtico per il riferimento Nazionale di reTI di Distribuzione Elettrica”
that means “Digital archive for the national electrical distribution reference networks”) [32].
The paper is organized as follows: Section 2 describes the detailed formulation of energy storage
placement problem. Section 3 discusses the uncertainty modeling approach. Section 4 describes the
solution methodology of a robust optimization problem. Sections 5 and 6 present the case study
with real data and conclusion, respectively. Finally, in the nomenclature of the symbols used in the
mathematical formulation has been reported.

2. Deterministic Formulation of Energy Storage Planning

The objective function (OF) of the deterministic model consists of minimizing the operational
extra-cost that should be sustained for complying with the technical constraints. Such cost includes
the penalty terms for RES (CRESc
n ) and biomass CHP generation curtailment (CCHPc n ), and the cost of
shaving the peak loads (CPLS
n ). Furthermore, since the goal of the paper is to evaluate the contribution

97
Energies 2020, 13, 512

of energy storages to the management of the network, even in uncertain conditions, the investment
cost CCAPEX_ESS
n to be sustained for the storage allocated in the network is added to the operational
cost, as in (1). ⎧ ⎫
⎪
⎪N  ⎪
⎪
⎨ CAPEX_ESS ⎬
minCtot = min⎪ ⎪
RESc
Cn + CnCHPc
+ Cn + Cn
PLS
⎪
⎪ (1)
⎩ ⎭
n=1

This minimization is subject to voltage and current limits, power flow equations, and storage
technical constraints. In the following, each cost term and constraints are detailed.

2.1. Penalty for RES Curtailment CRESc

n

To strongly penalize the generation curtailment of RES, the cost of curtailed energy due to network
constraint violations has been monetized as twice the price of energy paid in the wholesale market cEN
(here, 58 €/MWh, according to the average Italian energy selling price) [33], as in (2).


T
CRESc
n = 2·cEN ·PRESc
n (t) n = 1 · · · N (2)
t=1

where PRESc
n (t) is the energy curtailed at the time interval t by the RES generator connected to the n-th
bus of the network.
Since the increment of the network hosting capacity may be quantified via the possibly avoided
curtailment of RES production, the smaller this term, the better the storage allocation solution.

2.2. Penalty for Biomass CHP Curtailment CCHPc

n

This cost for biomass CHP curtailment is assumed proportional to the avoided cost for fuel
saving F [€/MWh], increased by 20%. The fuel cost F has been considered here equal to 80 €/MWh, by
assuming the average gas price ≈ 35 c€/m3 ≈ 36 €/MWht and by hypothesizing an efficiency for the
electric conversion about 45%. This assumption allows penalizing also the CHP curtailments, with
high cost, as in (3).
T
CCHPc
n = 1.2·F·PCHPc
n (t) n = 1 · · · N (3)
t=1

where PCHPc
n (t) is the energy curtailed at the time interval t by the biomass CHP connected to the n-th
bus of the network.

2.3. Peak Load Shaving Cost CPLS

n

Regarding the term referred to the active customers, in this paper only the cost of shaving the peak
loads has been considered, by assuming that it is not possible to fully control the customer demand but
only cut a quote of their consumption in some critical conditions. It is assumed, as the RES curtailment,
that this curtailed energy is paid at twice the energy price cEN to penalize load curtailment with the
higher cost, as renewable generation curtailment, according to (4).


T
CPLS
n = n (t) n = 1 · · · N
2·cEN ·PPLS (4)
t=1

n (t) is the energy curtailed at the time interval t to the customer connected to the n-th bus of
where PPLS
the network.

98
Energies 2020, 13, 512

2.4. Storage Investment Cost CCAPEX_ESS

n

The storage investment cost (SCn ) is a function of the size of the storage in terms of rated power
and energy as in (5).
SCn = cP ·Prated
n + cE ·Erated
n n = 1···N (5)

where cP and cE are the specific costs of the ESS adopted technology, reliant respectively on the power
rating Prated
n and the nominal capacity Erated
n of the n-th ESS located in the network (here cP = 200 €/kW
and cE = 400 €/kWh, according to the market cost of lithium-ion technology [15]).
To consider this cost in the objective function (1), only a daily quote of SCn is added to the
operational terms of (1), calculated as in (6).

KS
CCAPEX_ESS
n = ·SCn n = 1 · · · N (6)
365
where Ks is a capital recovery factor (here Ks = 0.1, for considering 10 years as ESS lifetime).
In this paper, it is assumed that the storages are DSO owned and managed for relieving
contingencies. Thus, the ESS OPEX (operational expenditures) is not considered in the optimization.
According to this point of view, it is supposed that the minimization of the network operational
cost, in terms of reduction of the curtailed power from RES and to loads, that would be necessary
to relieve contingencies, represents the only incomes that allow DSO to pay back EES CAPEX
(capital expenditures) and ESS OPEX. The depreciation of the ESSs is assumed negligible and not
added to the ESS cost term.

2.5. Load Balancing Constraints

The SOCP convex relaxation has been used in the proposed multi-temporal AC-OPF model.
Equations (7) and (8) are the nodal active and reactive power balance.
g
Pn (t) + PRES
n ( t ) − Pn
RESc (t) + PCHP (t) − PCHPc (t) − PD (t) + PPLS (t) − Pc (t) + Pd (t)
n

n
2 =  P (t)
n n n n
(7)
− Rmn ·Imn mn
mθn mθn
 
g
Qn (t) + QRES
n (t) − Qn
RESc
(t) + QCHP
n (t) − QCHPc
n (t) − QDn (t) − 2
Xmn ·Imn = Qmn (8)
mθn mθn
 
n (t); Qn (t)) and Pn
where (PRES RES CHP (t); QCHP (t) define the expected RES and CHP production in
n
terms of active and reactive powers, PDn (t) and QDn (t) are the active and reactive power delivered to
the load connected to the n-th node, Imn (t), Pmn (t), and Qmn (t) are respectively the current, the active
and the reactive power flowing in the branch from the m-th bus to the n-th one, Rmn and Xmn are the
resistance and reactance of the mn-th branch. Pcn (t) and Pdn (t) are the charging and discharging power
g g
of the storage at time t. Pn (t) and Qn (t) are the active and reactive power provided by the upstream
g g
connections (slack bus of the network). The values of Pn (t) and Qn (t) are zero except for the first node.

2.6. Network Constraints

The current magnitude quadratic term can be defined as the function of the corresponding active
and reactive power quadratic terms (Equations (9)–(11)).

2 P2mn + Q2mn
Imn ≥ 2
(9)
Vm

P2mn (t) + Q2mn (t) = S2l (t) (10)

imn (t)·vm (t) = S2l (t) (11)

99
Energies 2020, 13, 512

Equation (9) is relaxed ultimately by relaxing the magnitude of currents within each branch and
using a conic formation on the limitation of exchanged active power. For linearization purposes, the
quadratic terms of voltage and current magnitude have been replaced with the linear ones as in (12).

2
Imn = imn ; Vm
2
= vm (12)

The new variables (imn , vm ) successfully formulate the SOCP problem according to the
following constraint,
2
Vmin ≤ υm (t) ≤ Vmax
2
(13)

The Equation (13) provides the voltage limits of each bus.

Bus 1 is modelled as a swing bus with fixed complex voltage Vm (t).

2.7. Constraints for RES and Controllable Generator

Equations (14) and (15) impose the limits to the active and reactive power curtailment associated
with RES and CHP generators. In Equation (14), Pmin
n
RESc/CHPc
represents the lower bound of the active
power curtailment of RES and CHP generators. In this study, the lower bound value of curtailment
has been chosen as 0, which means the generators curtail all of their capacity. The upper bound,
PmaxRESc/CHPc
n , has been considered the capacity of the generators based on the expected values of each
time step.
Pmin
n
RESc/CHPc
≤ PRESc/CHPc
n (t) ≤ PmaxRESc/CHPc
n (14)

Qmin
n
RESc/CHPc
≤ QRESc/CHPc
n (t) ≤ QmaxRESc/CHPc
n (15)

Furthermore, the constraints about storages may be formulated as in (16)–(20).

 
Pd (t)
SOCn (t) = SOCn (t − 1) + Pcn (t)·ηc − n ·Δt (16)
ηd

0 ≤ Pcn (t) ≤ αcn ·Pc,max

n (t) (17)

0 ≤ Pdn (t) ≤ αdn ·Pd,max

n (t) (18)

SOCn,min ≤ SOCn (t) ≤ SOCn,max (19)

αcn (t) + αdn (t) ≤ 1 (20)

where αcn (t) αdn (t)

 [0 or 1] and  [0 or 1].
The state of charge (SoC) of ESSs is calculated by considering the initial SoC and the charging and
discharging efficiencies ηc and ηd (Equation (16)). To restrict the maximum charging and the depth of
discharging and for avoiding the simultaneous charging and discharging, the binary variables αcn and
αdn , of which only one can be different from zero, have been considered in Equations (17)–(20). Finally,
Equation (21) is added to force the SoC to be equal at the beginning and the end of the considered time
horizon T.
SOCn,0 = SOCn,T (21)

The multiplication of binary and integer variables during the estimation of the charging and
discharging power of the storage unit generates a quadratic term. A decomposition technique has
been used to linearize the relevant constraints by rewriting constraints in the form of (22) as in (23) and
(24) to avoid the bilinear terms.
x <= y · z · c (22)

x <= y · zmax · cmax (23)

x <= z · c (24)

100
Energies 2020, 13, 512

x and c are continuous, y binary, z integer. The continuous and integer variables are respectively
variable in [0, xmax ], [0, cmax ] and [0, zmax ].

3. Uncertainty Management
Uncertainties are mostly involved in decision-making problems. The uncertainty from electric
loads, wind, and solar power generation typically influence distribution planning in general and
storage allocation in particular. Several factors determine the evolution of each uncertainty. For
example, the consumers’ activities, energy savings and electricity providers’ rate policies influence the
electric load; the radiation of the sun and the velocity of air impact on the power output of PV and
wind [34].
A static robust optimization is used to consider the uncertainty in the optimal planning model.
To define the uncertainty set, an interval uncertainty model has been adopted with the flexibility to
regulate the robustness, called budget of uncertainty (Γi ). Static robust optimization devising seeks
for optimal solutions that optimize the objective function and encounter the problem requirements
for every possible revealing of the uncertainty in constraint coefficients. Hence, the variables are
independent of the uncertain parameters.
For a worst-case analysis, when considering the uncertainty, the following problem (25)–(27) is
dealt with:
min c·x (25)

Subject to

n 
aij x j + max aij ξij x j ≤ b j (26)
j=1 j∈Ji

l≤x≤u (27)

In the above optimization problem, Equations (25) and (26) represent the objective function and
inequality constraint, respectively. The uncertainty bound of the uncertain parameter x, that must
assume values between lower l and upper u bounds, as formulated by the Equation (27). b j is the value
of the right-hand side of i-th constraint.
For the i-th constraint, the auxiliary problem can be formulated as follows:

max aij ξij x j (28)
jJi

Subject to 
ξij ≤ Γi (29)
jJi

0 ≤ ξij ≤ 1 (30)

To make the model tractable, that means to convert the inner maximization problem to a
minimization problem, the dual of the above problem (28)–(30) needs to be formulated as follows:

min zi Γi + pij (31)
jJi

Subject to
zi + pij ≥ aij yi Vi , jJi (32)

xj ≤ yj (33)

zi , pij , y j ≥ 0 (34)

where zi , pij are the dual decision variables for constraints of the auxiliary problem.

101
Energies 2020, 13, 512

Incorporating model (31)–(34) into the original problem (25)–(27), the robust linear counterpart is
formulated as:
minc·x (35)

Subject to

n 
aij x j + zi Γi + pij ≤ bi (36)
j=1 jJi

lj ≤ xj ≤ uj (37)

zi + pij ≥ aij y j , Vi , jJi (38)

− yj ≤ xj ≤ yj (39)

zi , pij , y j ≥ 0 (40)

4. Robust Counterpart
Assume that all the decision variables should be considered before the revealing of the uncertainty
from solar power, wind generation, and electric loads. In the active power balance (7), uncertainties
PRES (t) and PDn (t) are modeled as symmetric and bounded variables P pv
(t), P 
w (t) and PD
n n (t). It
n n
should be mentioned here that PRESn (t) consists of the solar and wind generations. The uncertainty
takes values as in the following Equations (41)–(43).

pv pv  pv ˆlb ˆub

Pn (t) = Pn (t) + ΔPn (t) Ppv ≤ ΔPpv (t) ≤Ppv (41)


P
wind (t) = Pwind (t) + ΔPwind (t) Plbˆ
n n n wind
ˆ
≤ ΔPwind (t) ≤Pub
wind
(42)


PD  ˆlb ˆub
n (t) = PDn (t) + ΔPDn (t) PD ≤ ΔPD (t) ≤ PD (43)

In the robust model, the objective function (1) is identical to the deterministic model. The only
constraint that is affected by uncertainty is the electric power balance equation. The electric power
in the network should be met when the worst case of uncertainties occurs. For the power balance
equation, the worst case would occur at the maximum increase of the electric loads and the maximum
decrease in solar (PV) and wind power generation. Therefore, the robust formulation becomes as
in (44)–(47). ⎧ ⎫
⎪ ⎪
⎨ RESc
N
⎪ ⎪
CAPEX_ESS ⎬
min Ctot = min⎪
⎪ Cn + Cn CHPc
+ Cn + Cn
PLS
⎪
⎪ (44)
⎩ ⎭
n=1

Subject to
g
Pn (t) + PRES
n (t) −PRESc
n (t) + PCHPn (t) − PCHPc
n (t) − PDn (t) + PPLSn (t)
−Pcn (t) + Pdn (t) + max{Pub D
( t ) ∗ ξub (t) + Plb (t) ∗ ξlb (t)
D D D
(45)
−Pubpv (t) ∗ ξpv (t) − Ppv (t) ∗ ξpv (t) − Pwind (t) ∗ ξwind (t)
ub lb lb ub ub
 
−Plb wind
(t) ∗ ξlb
wind
(t)} − 2 =
Rmn ·Imn Pmn
mθn mθn

D (t) + ξD (t) + ξpv (t) + ξpv (t) + ξwind (t) + ξwind (t) ≤ Γ1 (t)
ξub lb ub lb ub lb
(46)

D (t), ξD (t), ξpv (t), ξpv (t), ξwind (t), ξwind (t) ≤ 1
ξub lb ub lb ub lb
(47)

where ξub
D
(t), ξlb
D
(t), ξub
pv (t), pv (t), ξwind (t), ξwind (t)
ξlb ub lb
are the scaled deviations from the random electric
loads, solar, and wind power generation, respectively. Γ1 (t) is the budget of the uncertainty of uncertain

102
Energies 2020, 13, 512

parameters at time t that lies between 0 to 1, where 0 being the deterministic case and 1 defined the
most robust case.
To make tractable the above problem, the following subproblem in Equations (48)–(50) need
+
to be formulated into the corresponding dual problem by introducing dual variables λ1 (t), ΠD (t),
− + − + −
ΠD (t), Πpv (t), Πpv (t), Πwind (t), Πwind (t) for constraints (49) and (50).
The subproblem can be formulated as in (48).

max{Pub
D
(t) ∗ ξub
D
(t) + Plb D
(t) ∗ ξlb
D
(t) − Pub
pv (t) ∗ ξpv (t) − Ppv (t) ∗ ξpv (t)
ub lb lb
(48)
−Pubwind
( t ) ∗ ξub (t) − Plb (t) ∗ ξlb (t)}
wind wind wind

Subject to
D (t) + ξD (t) + ξpv (t) + ξpv (t) + ξwind (t) + ξwind (t) ≤ Γ1 (t)
ξub lb ub lb ub lb
(49)

D (t), ξD (t), ξpv (t),

ξub pv (t), ξwind (t), ξwind (t)
lb ub
ξlb ub lb
≤1 (50)

The robust counterpart after applying the duality theory is formulated as in (51)–(55).

+ + +
min λ1 (t)Γ1 (t) + ΠD (t)+ Π−D (t) + Πpv (t) + Π−pv (t) + Πwind (t) + Π−wind (t) (51)

Subject to
+
λ1 ( t ) + Π D (t) ≥ Pˆub
D
(t), λ1 (t) + Π−D (t) ≥ Pˆlb
D
(t) (52)
+
λ1 (t) + Πpv (t) ≥ −Pˆub − ˆlb
pv (t), λ1 (t) + Πpv (t) ≥ −Ppv (t) (53)
+
λ1 (t) + Πwind ˆ (t), λ (t) + Π− (t) ≥ −Plbˆ (t)
(t) ≥ −Pub (54)
wind 1 wind wind

λ1 ( t ) , Π ± ± ±
D (t), Πpv (t), Πwind (t) ≥0 (55)

Finally, the tractable robust model can be formulated as the following (56) and (57).
⎧ ⎫
⎪ ⎪
⎨ RESc
N
⎪ CAPEXESS ⎪
⎬
minCtot = min⎪
⎪ Cn + Cn
CHPc
+ Cn + Cn
DR
⎪
⎪ (56)
⎩ ⎭
n=1

Subject to
g
Pn (t) + PRES
n ( t ) − Pn
RESc (t) + PCHP (t) − PCHPc (t) − PD (t) + PPLS (t)−
n n n n
+ + +
Pn (t) + Pdn (t) + λ1 (t)Γ1 (t) + ΠD
c (t) + Π−D (t) + Πpv (t) + Π−pv (t) + Πwind (t)+ (57)
−  
Πwind (t) − mθn Rmn ·Imn = mθn Pmn (t)
2

Moreover, the constraints (8)–(21) and (52)–(55) form the tractable problem.
The new model does not contain any uncertainty and is formulated as a mixed-integer second-order
conic programming (MISOCP) problem that can be solved efficiently using CPLEX that uses a branch
and cut algorithm to find the integer feasible solution.

5. Case Study
The procedure was applied to a test distribution network derived from the ATLANTIDE project [32].
The MV network, shown in Figure 1, representative of the industrial ambit, was constituted by 100
nodes, subdivided in seven feeders supplied by a primary substation equipped with a 25 MVA high
voltage/medium voltage (HV/MV) transformer. The total demand was about 30 MVA (372 GWh/year)
and the total installed DG capacity was 34 MW (27.2 GWh/year), as a mix of wind, PV and biomass
CHP generators.

103
Energies 2020, 13, 512

Figure 1. Test network (representative network derived from ATLANTIDE project [32]).

The mathematical formulation of the RO for an AC-OPF based energy storage planning tool
was programmed in General Algebraic Modeling System (GAMS) (GAMS Software GmbH, Frechen,
Germany) and solved using CPLEX 25.1.1 on a 2.30 GHz personal computer with 4 GB RAM. In this
experimental study, the worst case was considered when the load was high (ξD,t = 1) and wind and PV
generation was low (ξpv,t , ξw,t = −1).
For the sake of a comprehensive view, in the following, the results obtained by the application
of the described optimization to the network of Figure 1 in 12 typical days, differentiated between
working days, Saturdays, and holidays (Sundays included), and between seasons, have been reported.
The time horizon of 24 h of each typical day has been considered with a time step of 1 h. Three
scenarios have been considered: the certain one (solved by the deterministic OPF) and two uncertain
scenarios with different values of risk (Γ = 0.5 and Γ = 1), both solved with RO. Furthermore, for
highlighting the advantages provided by the storage systems the case of deterministic optimization
(certain) without storage has been added to the previously described cases.
All the buses of the test network were assumed candidates for storage placement. The available
ESS were considered of 1.0 MW/2 h storage capacity. The efficiencies for charging and discharging
were considered 90% each, which gives an overall roundtrip efficiency of 0.81. The initial state of
charge (SoC) has been considered 25% of its capacity.
In these typical days, some under-voltage conditions occur in the most distant nodes from the
HV/MV transformer and, thus, for solving these issues, it is necessary to resort to the load peak shaving.
Furthermore, some lines suffer for overloading depending on the non-coincidence of load demand and
DG production. ESSs prove to be useful for reducing the curtailment of the demand and production as
detailed in the next subsections.

5.1. Generation and Load Profiles

The generation and load profiles were simulated according to the ATLANTIDE load and generation
daily curves, that provide for different kinds of customers (i.e., industrial, residential, commercial,
and agricultural) and for several technologies of DG (i.e., wind turbine, PV, and CHP biomass-based)

104
Energies 2020, 13, 512

the hourly consumption/production for each typical day. An amount of 22 PV systems was assigned
to 20 nodes. The size of these systems is between 49–1048 kW. Node 8 had the biggest PV system,
whereas the lowest one was connected to node 78. Node 83 comprised two wind generators and two
CHP plants. Figure 1 depicts the nominal power of the PV, wind, and CHP of each node.
The load profiles indicated a peak load of 18.69 MW during the spring working day and 18.14 MW
during the summer working day with an average load of 13.79 MW and 13.38 MW, respectively. As an
example, the demand and production profiles and their balance at the HV/MV interface, during the
spring working day, are shown in Figure 2.

Figure 2. Load-production profiles of the whole network for the Spring working typical day.

5.2. Storage Placement

The optimization results of storage position for each typical day for the three considered cases
have been enumerated in Table 1. The ESS optimal positions may change from one typical day to
another even for the same case, but the results can be summarized by considering a given solution
valid for all the twelve typical days. In the following, it has been assumed that the placement in one
bus or a close one on two different typical days can be considered the same placement. For instance,
the bus 83 and the bus 84 in the deterministic case, that are the solutions for the TD8, and the TD5
respectively, can be considered as a unique optimal position around the bus 83.

Table 1. Storage placement for each typical day for the three considered cases.

Typical Days Deterministic Case Intermediate Case Robust Case

TD1 (Winter working day) 10, 32 10, 32, 77 10, 32, 48, 77
TD2 (Winter Saturday) - - -
TD3 (Winter holiday) - - -
TD4 (Spring working day) 10 10, 34 10, 34
TD5 (Spring Saturday) 84 83, 84 83, 84
TD6 (Spring holiday) 84 84 84
TD7 (Summer working day) - - -
TD8 (Summer Saturday) 83 12, 83 10, 32, 83
TD9 (Summer holiday) - - -
TD10 (Autumn working day) 12 12, 27 12, 27, 69
TD11 (Autumn Saturday) 84 84 69, 84
TD12 (Autumn holiday) 83, 85 83, 85 83, 85
Total number of ESS 4 5 6

105
Energies 2020, 13, 512

On the contrary, if two busses, even close, appeared in the solution of the same day they were
both considered necessary and two ESSs had to be placed on that nodes (e.g., the busses 83 and 85 in
the results of all the cases for the TD12 or the busses 83 and 84 in the results of the intermediate and
robust cases for the TD5). By applying these rules, the total number of ESSs that had to be placed in
the three cases are reported in the last row of Table 1. It is worthy of mentioning that the results were
substantially incremental: the intermediate case included the location of the deterministic case, and
the robust case (no risk) included, in turn, the intermediate one.
To analyze the impact of renewables and load uncertainty on the investment of the energy storage
in the distribution network, one of the worst-cases of RES (PV, wind or biomass based) and combination
of loads were considered. The worst-case scenario considered in this work was when the loads had
upper bound values, and the renewables had lower bound values. Three cases were considered by
varying the loads and renewables uncertainty bounds. In the first case, the budget of uncertainty
was zero (Γ = 0), i.e., the profiles of load and renewable generations were assumed following the
forecasted values. In the second case, the value of budget of uncertainty for both load and renewables
considered 0.5 (Γ = 0.5) that is between the zero (deterministic) and 1 (robust or worst case). In the
third case (Γ = 1), the considered worst-case scenario was evaluated. In this case, the uncertainty sets
of loads and renewables were considered broader to consider the possible extreme coordinates of the
uncertainty set.
The following figures compare the results of the studied cases (i.e., no control, deterministic OPF
no storage, deterministic OPF with storage, intermediate and robust). These results are related to the
most critical typical day, the winter working day (TD1). For the sake of clarity, the figures refer only to
the feeder F1 that is the longest feeder of the test network depicted in Figure 1 (i.e., the last bus is about
14.2 km far from the primary substation). Figure 3 shows the voltage profiles occurring at 9:00 am of the
winter working typical day, because, among other time intervals, this one was proved that experiments
the greatest load curtailment; Figure 4 shows the load curtailed during this typical day, in Figure 5 the
balances of DG production and curtailed demand, and, finally, Figure 6 the ESS charging/discharging
optimal profiles of one of the ESS optimal positioned in the feeder F1 (bus 10 of Figure 1). As it is
evident by the results, all the optimizations allow to solve the undervoltage conditions occurring in
the long feeder F1 (Figure 3); the more conservative the optimization (i.e., by moving from certain to
uncertain, intermediate and robust, optimization) the smaller the demand curtailed (Figure 4); in the
feeder F1 no generation curtailment results from the optimizations, thus the balance of production and
demand (curtailed) is closer to the original one (no control) in the robust case (Figure 5). It is worth
noticing that the voltage value at the sending end (the MV busbar of the primary substation) was lower
in the no control case than the other cases because the implemented model of the HV/MV transformer
is very simple and strongly suffers for the high demand, not curtailed in the control case. In future
works, the transformer model will be improved. These results, together with the ESS operation, are
discussed more in detail in the next subsections for each optimization case.

106
Energies 2020, 13, 512

Figure 3. Voltage profiles of feeder F1 for no control, no storage (certain deterministic optimal power
flow (OPF) without storage) and deterministic (certain deterministic OPF with storage), intermediate
and robust cases at 9:00 am of the winter working day.

Figure 4. Load curtailments experimented by the feeder F1 for the no storage (certain deterministic
OPF without storage), deterministic (certain deterministic OPF with storage), intermediate and robust
cases on the winter working day.

107
Energies 2020, 13, 512

Figure 5. Balance of distributed generation (DG) production and curtailed demand of the
feeder F1 for the no control, no storage (certain deterministic OPF without storage), deterministic
(certain deterministic OPF with storage), intermediate and robust cases on the winter working day.

Figure 6. Charging/discharging profiles of the energy storage system (ESS) optimally positioned in
bus 10 of the feeder F1 for the deterministic, intermediate and robust cases on the winter working day
and balances of powers (DG, loads, and ESS) in the same cases. The no control case has been added
for comparison.

108
Energies 2020, 13, 512

5.3. Deterministic Case (with and without Storage)

Due to the absence of uncertainty, in this case, the load and renewables profiles will remain the
same as the predicted values. It was witnessed that during the deterministic case with storage, at least
four storages need to cover their requirements. The ESS optimal positions can change from one typical
day to another, but, by summarizing the results in the 12 typical days (Table 1), they were located two
in the two lateral branches that start from the node 10 (feeder F1, positions are 10 or 12 and 32). The
third and fourth ESS had to be located around the node 83 (feeder F5, positions are two among 83, 84
and 85). It is essential to observe that node 83 is the node that had the highest number of renewables
and CHP connected; thus, it is noticeable to consider that as a privileged position for storages.
The highest load curtailment was experienced during the typical day of winter working day (TD1)
with the amount of 51.77 MWh/day for the case without storage and 30.46 MWh/day for the case
with storage.
By focusing on the feeder F1, as it is evident from Figure 3, the nodes of this feeder had
under-voltage issues in the no control case, and any optimization forces to resort load shedding
(Figure 4). By comparing these two certain cases, is it worth noticing that if the ESSs are not
available for the optimization (deterministic OPF no storage) much more demand had to be curtailed
(i.e., 41.62 MWh/day of the no storage case vs. 20.97 MWh/day in the case with storage).
The daily operation of the ESS was optimized as well as the optimal position. For instance, during
the winter working day, the daily operation profile of the ESS located around bus 10 is shown in
Figure 6, together with the balances of demand and production curves, with and without ESS. At the
beginning of the day, the ESS started to charge, keeping the final balance of demand, DG production
(minimal in the first hours of the day), and charging power for ESS so low to do not negatively impact
the network operation. At around 7:00 am, when the morning peak starts, the storage discharges for
reducing the power demand and keeping the voltage profile within the limit (Figure 3).

5.4. Intermediate Case

In this case, a narrow uncertainty bound is considered. The budget of uncertainty for the uncertain
parameters has been considered as Γ = 0.5. The optimization algorithm will look for a solution inside
the specified uncertainty bound. From Table 1, by considering the simulation results of the twelve
typical days, and assuming the most conservative hypotheses (i.e., the final result is the union of the
results obtained for each typical day), the intermediate case suggests at least five storage systems to be
installed: two in the feeder F1 and two in the feeder F5, as in the deterministic case, plus one ESS in the
feeder F4. The positions of the two storages in the feeder F1 and the two in the feeder F5 are more or
less the same of the deterministic case (F1 possible locations are the busses 10 or 12 for one lateral and
the busses 27, 32, or 34 for the other lateral, and two positions among the bus 83, 84 or 85 for the feeder
F5). In the feeder F4, the added storage system has to be installed around the node 77.
The load shedding, in this case, is more reduced and for the critical TD1 is equal to 15.89 MWh/day
(about 25% less than the deterministic case), as shown in Figure 4.
Furthermore, it can be observed in Figure 6 that the ESS located around bus 10 in the feeder F1 has
a similar trend of the same ESS in the deterministic case: it charges and discharges mostly in the same
hours for solving local contingencies. In particular, it charges when the load demand is low (at the first
hours of the day), and finally, at the end of the day for recovering their initial SoC; on the contrary, it
discharges in correspondence of the peaks of demand (7:00–21:00).

5.5. Robust Case

The third case can be considered the worst-case analysis. In this case, the budget of uncertainty for
uncertain parameters is equal to 1 (Γ = 1). This budget of uncertainty allows the algorithm to consider
the extreme points of the uncertainty set. Compared to the previous cases, the robust case provides
six storage systems to be installed in the network. The locations of storage for feeder F1, F4 and F5

109
Energies 2020, 13, 512

are like the deterministic and intermediate cases. However, the robust case suggests one more ESS in
the feeder F2 (bus 48). In Figure 6, the operation profile of the ESS located around bus 10, resulting
from the optimization for the feeder F1 is shown with the demand and production daily curves. The
behavior of the ESS is as the one in the other cases: the contribution to reducing the peaks at the cost of
a slight increase in demand when they charge. This increase does not alter the network operation and
does not produce any technical constraint violation, but it allows a further reduction of load shedding
(Figure 4). For the feeder 1 in the critical TD1, the demand is curtailed of 11.10 MWh/day (−30% than
the intermediate case).

5.6. Economic Analysis

In order to analyze the economic feasibility of the investments in storage systems, the comparison
between all the cases mentioned above, included the no storage one, has been considered. Table 2
summarizes the yearly operational costs (operational expenditures—OPEX) for the four considered
cases, the amount of load shedding and generation curtailment used for solving the contingencies, the
CAPEX for the ESS installation referred to one year only (among the ten years of the ESS life duration),
and in the last column, the total yearly cost is calculated as the summation of CAPEX and the OPEX.
In the no storage case, the yearly operational cost, of about 1480 k€, consists of penalty cost for load
shedding that accounts for 762.90 k€/year, and penalty cost of CHP curtailment worth 717.46 k€/year.
The peak shaving drastically decreases by using the ESS even in the deterministic case (the quantity
is about halved), and then it is significantly further reduced in the uncertain scenarios. The same
behavior can be observed for the generation curtailment of CHP. In the uncertain cases, compared
with the base case without storages, the resort to load shedding is much reduced (−44.8% in the
deterministic case becomes −60.5% in the intermediate case and −73.9% in the robust one) as well as
the generation curtailment (−22.8%, −43.9% and −66.3% in the deterministic, intermediate and robust
case respectively). The quantities related to the generation curtailment in Table 2 for these four cases
are referred only to the curtailment of CHPs.

Table 2. Daily operational cost of the test network and ESS CAPEX.

Load Generation Total Cost

OPEX CAPEX
Cases Shedding Curtailment CAPEX+OPEX
[k€/Year] [k€/Year]
[MWh/Year] [MWh/Year] [k€/Year]
No storage 1480.36 6602.12 7502.38 0 1480.36
Deterministic (Γ = 0) 974.47 3644.95 5791.67 400 1374.47
Intermediate (Γ = 0.5) 704.64 2605.34 4208.2 500 1204.64
Robust (Γ = 1) 437.80 1721.07 2525.22 600 1037.80

Consequently, a substantial reduction of the annual operational costs can be observed with the
ESS inclusion in the deterministic case and much more in the uncertain cases (−34.2%, −52.4%, and
−70.4%, in the deterministic, intermediate and robust case respectively). It is worth noticing that, in the
deterministic and uncertain cases, apart from the operational costs, an additional cost factor has to be
considered: the CAPEX for the ESS installation, split in ten years (the CAPEX of one 1.0 MW/2 h ESS is
assumed the same for each year). This negatively impacts on the final cost, much stronger with the
increment of budget or uncertainty, due to the growth of the investment costs for the increasing number
of storages. However, the reduction of OPEX not only covers such increase but, the final costs of all the
cases that use the storage systems for relieving the contingencies are smaller than the case without them
(no storage case). In particular, the percentage of total cost reduction is smaller than the one calculated
by considering the OPEX only (i.e., −7.2%, −18.6% and −29.9% for deterministic, intermediate and
robust cases, respectively), but the results prove the effectiveness of the optimization. In fact, these
results demonstrate that not only the ESS helps to reduce the operational cost, for relieving even the
worst-case and reducing even more the resort to load shedding and to the generation curtailment, but

110
Energies 2020, 13, 512

also that, with the assumed hypotheses, the ESS CAPEX can be amortized during the ten years of their
life duration.

6. Conclusions
This paper establishes the use of a SOCP convex relaxation of the power flow equations for
optimal placement of energy systems in an MV distribution network. The algorithm proposed in this
paper can be used to analyze the economic viability in comparison to investment and operational costs.
The application of robust optimization and having the flexibility to modulate the budget of uncertainty
helps to find a balance among the factors of economic efficiency and conservatism. The use of this kind
of flexibility also assisted in considering additional scenarios other than the worst-case scenarios that
most robust optimization problems account for. By considering the worst-case scenario only, such
problems do not provide an optimal solution. Rather, they offer only conservative solutions that could
be impractical. However, the analytical reformulation technique helped to find the robust equivalent
of the original problem that was solved with less computational encumbrance using CPLEX solver.
As planning includes a limited financial budget and resources, this study affords a comprehensive
approach, which is a consideration of different situations (budget of uncertainty). Furthermore, the use
of this innovative algorithm leads to understanding of the benefits of grid-connected storage devices
in distribution systems and the consideration of uncertainties into the planning phase.
In future works, the implementation of the HV/MV transformer model will be improved. Moreover,
the reactive power provision from storage will be considered in the future model. A term that takes
into account the depreciation of the ESSs due to their use will be included.

Author Contributions: Conceptualization, N.C., G.P., and F.P.; methodology, N.C.; software, N.C.; validation,
G.P. and N.C.; formal analysis, F.P., G.P., and N.C.; investigation, G.P. and N.C.; resources, N.C.; data curation,
N.C. and G.P.; writing—original draft preparation, N.C. and G.P.; writing—review and editing, N.C. and G.P.;
visualization, N.C.; supervision, G.P.; project administration, F.P.; funding acquisition, F.P. All authors have read
and agreed to the published version of the manuscript.
Funding: Nayeem Chowdhury has been funded from the European Union’s Horizon 2020 research and
innovation programme under Grant Agreement No 676042. The contribution of G. Pisano to this paper has been
conducted within the R&D project “Cagliari2020” partially funded by the Italian University and Research Ministry
(grant# MIUR_ PON04a2 _00381).
Conflicts of Interest: The authors declare no conflict of interest.

Nomenclature
Specific costs of energy storage in terms of energy
cE , cP
and rated power, respectively
cEN Energy price in the wholesale market
Storage investment cost (CAPEX, capital
CnCAPEX_ESS
expenditures) at node n
Cost of curtailing combined heat and power (CHP)
CCHPc
n
power generation at node n
CPLS
n Cost of peak load shaving at node n
CRESc
n Cost of renewable energy curtailment at node n
F Fuel cost for biomass CHP plant
Current flows in the branch from m-th to the n-th bus
Imn (t)
at the time interval t
KS Capital recovery factor
Active and reactive power demand of the loads at
PDn (t), QDn (t)
node n during the time interval t, respectively
Active and reactive power flows in the branch from
Pmn (t), Qmn (t)
m-th to the n-th bus at the time interval t, respectively
Charging and discharging power of storage at node n
Pcn (t), Pdn (t)
during the time interval t, respectively

111
Energies 2020, 13, 512

Maximum and minimum limits of charging and

Pc,max
n (t), Pd,max
n (t) discharging power of storage at node n during the
time interval t, respectively
Expected active and reactive power production of
PCHP
n (t), QCHP
n (t)
CHP at node n during the time interval t, respectively
Active and reactive power provided by the upstream
g g
Pn (t), Qn (t) connections at node n during the time interval t,
respectively
Power related to the peak load shaving at node n
n (t)
PPLS
during the time interval t
Amount of renewable power curtailment at node n
PRESc
n (t)
during the time interval t
Expected active and reactive power production of
n (t), Qn (t)
PRES RES renewables at node n during the time interval t,
respectively
Upper bound of active power curtailment of
PmaxRESc/CHPc
n
renewables and CHP at node n
Lower bound of active power curtailment of
Pmin
n
RESc/CHPc
renewables and CHP at node n
pv  Bounded variables of PV, wind and power demand of
n ( t ) , Pn
P wind (t), PD (t)
n loads at node n during time interval t, respectively
Deviation from expected power value of PV, wind
pv  
n ( t ) , Δ Pn
wind (t), ΔPD
ΔP n (t) and power demand of loads at node n during time
interval t, respectively
Lower bound of reactive power curtailment of
Qmin
n
RESc/CHPc
renewables and CHP at node n
Upper bound of reactive power curtailment of
Qmax
n
RESc/CHPc
renewables and CHP at node n
Rmn Resistance of the mn-th branch
Sl ( t ) Thermal capacity of the line at time interval t
State of the charge of storage unit at node n during
SOCn (t)
the time interval t
SCn The storage investment cost
Vmax , Vmin Maximum and minimum voltage limits, respectively
Xmn Reactance of the mn-th branch
Binary variables for charging and discharging of
αcn (t), αdn (t) storage at node n during the time interval t,
respectively
+ +
ΠD (t)Π−D (t), Πpv (t), Π−pv (t) Dual variables of load and PV at the time interval t
+
Πwind (t), Π−wind (t) Dual variables of wind at the time interval t
Charging and discharging efficiency of storage,
η c , ηd
respectively
Scaled deviations from the random electric loads and
ξub
D
(t), ξlb
D pv (t), ξpv (t)
(t), ξub lb
PV at the time interval t
Scaled deviations from the random wind power
ξub
wind
(t), ξlb
wind
(t)
generation at the time interval t
The budget of uncertainty of the uncertain parameter
Γi
i

References
1. International Renewable Energy Agency. REmap: Roadmap for A Renewable Energy Future, 2016 ed.; IRENA:
Abu Dhabi, UAE, 2016.
2. Dunn, B.; Kamath, H.; Tarascon, J.M. Electrical energy storage for the grid: A battery of choices. Science 2011,
334, 928–935. [CrossRef] [PubMed]

112
Energies 2020, 13, 512

3. Dvorkin, Y.; Lubin, M.; Backhaus, S.; Chertkov, M. Uncertainty sets for wind power generation. IEEE Trans.
Power Syst. 2015, 31, 3326–3327.
4. Pandžić, H.; Wang, Y.; Qiu, T.; Dvorkin, Y.; Kirschen, D.S. Near-optimal method for siting and sizing of
distributed storage in a transmission network. IEEE Trans. Power Syst. 2014, 30, 2288–2300.
5. Gayme, D.; Topcu, U. Optimal power flow with large-scale storage integration. IEEE Trans. Power Syst. 2012,
28, 709–717. [CrossRef]
6. Makarov, Y.V.; Du, P.; Kintner-Meyer, M.C.; Jin, C.; Illian, H.F. Sizing energy storage to accommodate high
penetration of variable energy resources. IEEE Trans. Sustain. Energy 2011, 3, 34–40. [CrossRef]
7. Grainger, B.M.; Reed, G.F.; Sparacino, A.R.; Lewis, P.T. Power electronics for grid-scale energy storage.
Proc. IEEE 2014, 102, 1000–1013. [CrossRef]
8. Vargas, L.S.; Bustos-Turu, G.; Larraín, F. Wind power curtailment and energy storage in transmission
congestion management considering power plants ramp rates. IEEE Trans. Power Syst. 2014, 30, 2498–2506.
9. Chowdhury, N.; Pilo, F.; Pisano, G.; Troncia, M. Optimal Location of Energy Storage Systems with Robust
Optimization. In Proceedings of the CIRED 2019-25th International Conference and Exhibition on Electricity
Distribution, Madrid, Spain, 3–6 June 2019.
10. Bucciarelli, M.; Paoletti, S.; Vicino, A. Optimal sizing of energy storage systems under uncertain demand and
generation. Appl. Energy 2018, 225, 611–621. [CrossRef]
11. Zhao, H.; Wu, Q.; Huang, S.; Guo, Q.; Sun, H.; Xue, Y. Optimal siting and sizing of Energy Storage System
for power systems with large-scale wind power integration. In Proceedings of the 2015 IEEE Eindhoven
PowerTech, Eindhoven, The Netherlands, 29 June–2 July 2015.
12. Wogrin, S.; Gayme, D.F. Optimizing storage siting, sizing, and technology portfolios in transmission-
constrained networks. IEEE Trans. Power Syst. 2014, 30, 3304–3313.
13. Peker, M.; Kocaman, A.S.; Kara, B.Y. Benefits of transmission switching and energy storage in power systems
with high renewable energy penetration. Appl. Energy 2018, 228, 1182–1197. [CrossRef]
14. Motalleb, M.; Reihani, E.; Ghorbani, R. Optimal placement and sizing of the storage supporting transmission
and distribution networks. Renew. Energy 2016, 94, 651–659. [CrossRef]
15. Celli, G.; Pilo, F.; Pisano, G.; Soma, G.G. Distribution energy storage investment prioritization with a real
coded multi-objective Genetic Algorithm. Electr. Power Syst. Res. 2018, 163, 154–163. [CrossRef]
16. Celli, G.; Chowdhury, N.; Pilo, F.; Soma, G.G.; Troncia, M.; Gianinoni, I.M. Multi-Criteria Analysis for
decision making applied to active distribution network planning. Electr. Power Syst. Res. 2018, 164, 103–111.
[CrossRef]
17. Paudyal, S.; Canizares, C.A.; Bhattacharya, K. Three-phase distribution OPF in smart grids: Optimality
versus computational burden. In Proceedings of the 2011 2nd IEEE PES International Conference and
Exhibition on Innovative Smart Grid Technologies, Manchester, UK, 5–7 December 2011.
18. Bose, S.; Low, S.H.; Teeraratkul, T.; Hassibi, B. Equivalent relaxations of optimal power flow. IEEE Trans.
Autom. Control 2015, 60, 729–742. [CrossRef]
19. Gan, L.; Li, N.; Topcu, U.; Low, S.H. Exact convex relaxation of optimal power flow in radial networks.
IEEE Trans. Autom. Control 2015, 60, 72–87. [CrossRef]
20. Papaefthymiou, S.V.; Papathanassiou, S.A. Optimum sizing of wind-pumped-storage hybrid power stations
in island systems. Renew. Energy 2014, 64, 187–196. [CrossRef]
21. Fossati, J.P.; Galarza, A.; Martín-Villate, A.; Fontan, L. A method for optimal sizing energy storage systems
for microgrids. Renew. Energy 2015, 77, 539–549. [CrossRef]
22. Balducci, P.J.; Alam, M.J.E.; Hardy, T.D.; Wu, D. Assigning value to energy storage systems at multiple points
in an electrical grid. Energy Environ. Sci. 2018, 11, 1926–1944. [CrossRef]
23. Fernández-Blanco, R.; Dvorkin, Y.; Xu, B.; Wang, Y.; Kirschen, D.S. Optimal energy storage siting and sizing:
A WECC case study. IEEE Trans. Sustain. Energy 2016, 8, 733–743. [CrossRef]
24. Jayasekara, N.; Masoum, M.A.S.; Wolfs, P.J. Optimal operation of distributed energy storage systems to
improve distribution network load and generation hosting capability. IEEE Trans. Sustain. Energy 2016, 7,
250–261. [CrossRef]
25. Babacan, O.; Torre, W.; Kleissl, J. Siting and sizing of distributed energy storage to mitigate voltage impact by
solar PV in distribution systems. Sol. Energy 2017, 146, 199–208. [CrossRef]
26. Hassan, A.; Dvorkin, Y. Energy storage siting and sizing in coordinated distribution and transmission
systems. IEEE Trans. Sustain. Energy. 2018, 9, 1692–1701. [CrossRef]

113
Energies 2020, 13, 512

27. Wang, H.; Lv, Q.; Yang, G.; Geng, H. Siting and sizing method of energy storage system of microgrid based
on power flow sensitivity analysis. J. Eng. 2017, 2017, 1974–1978. [CrossRef]
28. Erdinc, O.; Tascikaraoglu, A.; Paterakis, N.G.; Dursun, I.; Sinim, M.C.; Catalao, J.P. Comprehensive
optimization model for sizing and siting of DG units, EV charging stations and energy storage systems.
IEEE Trans. Smart Grid. 2017, 9, 3871–3882. [CrossRef]
29. Castillo, A.; Gayme, D.F. Grid-scale energy storage applications in renewable energy integration: A survey.
Energy Convers. Manag. 2014, 87, 885–894. [CrossRef]
30. Blanco, H.; Faaij, A. A review at the role of storage in energy systems with a focus on Power to Gas and
long-term storage. Renew. Sustain. Energy Rev. 2018, 81, 1049–1086. [CrossRef]
31. Lorente, J.L.; Liu, X.A.; Best, R.; Morrow, D.J. Energy storage allocation in power networks–A state-of-the-art
review. In Proceedings of the 2018 53rd International Universities Power Engineering Conference (UPEC),
Glasgow, UK, 4–7 September 2018.
32. Bracale, A.; Caldon, R.; Celli, G.; Coppo, M.; Dal Canto, D.; Langella, R.; Petretto, G.; Pilo, F.; Pisano, G.;
Proto, D. Analysis of the Italian distribution system evolution through reference networks. In Proceedings of
the 2012 3rd IEEE PES Innovative Smart Grid Technologies Europe, ISGT Europe, Berlin, Germany, 14–17
October 2012.
33. Italian National Energy Market Operator (Gestore mercati energetici). Available online: https://www.
mercatoelettrico.org/En/default.aspx (accessed on 20 January 2020).
34. Birge, J.R.; Louveaux, F. Introduction to Stochastic Programming; Springer Science & Business Media: Berlin,
Germany, 2011.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

114
 energies
Article
Coordinated Operation of Energy Storage Systems for
Distributed Harmonic Compensation in Microgrids
Abbas Marini 1 , Luigi Piegari 2 , S-Saeedallah Mortazavi 3 and Mohammad-S Ghazizadeh 1, *
1 Department of Electrical Engineering, Abbaspour College, Shahid Beheshti University, Tehran 1983969411,
Iran; a_marini@sbu.ac.ir
2 Department and Electronics, Information & Bioengineering, Politecnico di Milano, 20133 Milan, Italy;
luigi.piegari@polimi.it
3 Department of Engineering, Shahid Chamran University of Ahvaz; Ahvaz 6135783151, Iran;
mortazavi_s@scu.ac.ir
* Correspondence: ms_ghazizadeh@sbu.ac.ir; Tel.: +98-21-7393-2531

Received: 17 December 2019; Accepted: 7 February 2020; Published: 10 February 2020

Abstract: Energy storage systems (ESSs) bring various opportunities for a more reliable and flexible
operation of microgrids (MGs). Among them, energy arbitrage and ancillary services are the most
investigated application of ESSs. Furthermore, it has been shown that some other services could
also be provided by ESSs such as power quality (PQ) improvements. This issue could be more
challenging in MGs with widespread nonlinear loads injecting harmonic currents to the MG. In this
paper, the feasibility of ESSs to act as coordinated active harmonic filters (AHF) for distributed
compensation was investigated. An optimization model was proposed for the coordination of the
harmonic compensation activities of ESSs. The model takes into account the various technical and
systematic constraints to economically determine the required reference currents of various AHFs.
Simulation cases showed the performance of the proposed model for enhancing the harmonic filtering
capability of the MG, reduction in the compensation cost, and more flexibility of the distributed
harmonic compensation schemes. It was also shown that ESS activities in harmonic compensation do
not have much of an effect on the ESSs revenue from energy arbitrage. Hence, it could make ESSs
more justifiable for use in MGs.

Keywords: energy storage system; active harmonic filter; harmonics; power quality; optimization

1. Introduction
Energy storage systems (ESSs) have received special attention due to their great flexibility and
applicability in power systems. The smart charging and discharging behavior of ESSs could produce
many advantages for various power system applications. The energy arbitrage for the operation of
ESSs as consumers of low-price energy at off-peak periods and suppliers of high-price energy at peak
hours is the most favorable application of ESSs. Furthermore, they could provide some advantages
such as providing ancillary services and deferring investment costs in various sections of the power
and distribution systems (DSs) [1–4]. All of these applications are related to the supply and security
aspects of the energy. However, utilization of ESSs for the quality aspect of the energy for power
quality (PQ) improvements has not gained much attention until now. With respect to the major costs
that could be imposed by PQ related problems [5,6], it seems that investigations should be carried out
regarding possible the PQ advantages of ESSs. Since the penetration of nonlinear power electronic
(PE) based loads is increasing, it seems that ESSs could play an important role in enhancing PQ levels
affected by these new loads in future DSs.
An ESS converts electrical energy via a power interface that is usually based on a PE converter.
The general characteristics of ESSs are typically determined by their PE interface. Aside from the PE

Energies 2020, 13, 771; doi:10.3390/en13030771 115 www.mdpi.com/journal/energies

Energies 2020, 13, 771

interface, various management and control components are included in ESSs, which give them various
functionalities in the network. Flexibility and controllability of PE interfaces could be employed for
PQ improvement in DSs. In this paper, PQ improvement using ESSs is discussed.
Any deviation that occurs from a pure sinusoidal voltage or current can be seen as a PQ problem.
These can be listed as various issues such as harmonics, interruptions with different durations, and
voltage related problems [7]. Problems related to the poor PQ levels impose remarkable costs on
power systems [5,6]. Among these, harmonics are becoming more important due to an increase in
the number of nonlinear harmonic loads and PE converters connected to the network, especially in
DSs and microgrids (MGs). These loads could be employed in various applications from residential
and commercial loads to industrial ones. Hence, harmonics could be imposed on the MG in any
location at each voltage level. It has been reported that about 27% of all PQ problems are related to the
harmonic distortions in voltage and current [8,9]. Harmonics could increase losses in DSs up to 20%.
The losses are directly dependent on the number of nonlinear loads connected to the grid. A smart
MG with a voltage THD ranging from 1% to 8%, could lose its generation capacity by 4.7% to 42.2%,
respectively [5].
ESSs could play an important role in the mitigation of PQ related problems, especially regarding
harmonics [2,4,7,10]. The ability to control PE converters could enable ESSs to act as active harmonic
filters (AHFs). The change in the control algorithms of PE converters for taking additional services
is discussed in the literature. The battery storage system is the most promising ESS in this scope.
A smart battery controller was proposed in [11] for enhancing PQ and adjusting the steady-state
voltage and frequency of MG. In [12], the optimal planning of ESSs for the mitigation of PQ problems
was investigated to make ESSs more reasonable to be used in power systems. An investigation was
made in [13] on the PQ of wind turbines with and without ESSs. The result shows the positive
effects of ESSs as a solution for the mitigation of the harmonics of wind turbines. Battery energy
storage could simultaneously compensate active, reactive, and harmonic power in four quadrants.
A control approach for the compensation of voltage drop and harmonics was presented in [14]. In [15],
the possible utilization of battery inverters for mitigation of harmonics and unbalances was reported.
A harmonic and unbalance current controller was introduced based on the proportional-resonant
controller in [15].
It should be noted that conventional control methods do not consider all other AHFs to calculate the
compensation sharing of each AHF. In local controls such as current or voltage modes for eliminating
current or voltage harmonics, AHFs work independently to other filters. Some control methods
such as droop control provide some level of compensation sharing for a limited number of AHFs.
However, the droop control is affected by the harmonic impedance of the network and is disabled
to provide proper harmonic compensation sharing when AHFs become distributed. In this paper,
an optimization-based model was proposed for achieving a coordinated operation of AHFs. In this
approach, the MG operator tries to minimize the harmonic compensation cost with respect to the
network constraints, working limits of AHFs, ESS activities in energy and ancillary markets, and
harmonic standards. Hence, all constraints of AHF as well as the required harmonic compensation
actions could be considered in the proposed method as a coordinated operation of AHFs.
It has been shown that optimal harmonic compensation can be achieved by the coordinated
control of PE-based inverters [5,16]. This approach could avoid possible instabilities and ensure an
optimal compensation scheme. Hence, the coordinated control of ESSs was implemented in this paper
in order to show the capability of ESSs for distributed harmonic compensation. To this end, a new
optimization model named as ESS-AHF was proposed for the calculation of the harmonic compensation
sharing of each ESS. In this model, the MG operator tries to minimize harmonic compensation costs
limited to various operational and technical constraints. The results of energy and ancillary service are
used as setpoints for the ESSs to determine harmonic power truncations in each harmonic frequency.
It should be noted that the distributed compensation scheme has better applicability to cope with
the harmonics of the dispersed nonlinear loads of the MG. Furthermore, for reliable operation and

116
Energies 2020, 13, 771

unwanted stability problems, the coordinated operation of the AHFs should be ensured. Therefore,
the proposed ESS-AHF model could successfully meet the requirements for distributed compensation
and stability considerations.
The rest of this paper is organized as follows. In Section 2, active harmonic filters are introduced.
Various ESS applications and characteristics are investigated in Section 3. A harmonic model of ESSs
in harmonic compensation activities is also represented in this section. The proposed ESS-AHF model
is introduced in Section 4. In Section 5, the results of the implementation of the introduced models are
provided and analyzed. The paper then presents our concluding remarks in Section 6.

2. Active Harmonic Filters

Conventional mitigation methods of harmonics can be categorized as passive filters, static
compensators, and active power line conditioners (APLCs). Currently, passive filters are the most
prevalent technology for harmonic compensation. They consist of a single tune, second tune, and multi
tune topologies. The passive filters are economic solutions and there is a great body of literature
regarding their sizing and application for DSs [17]. Static compensators follow a principle similar to
that of passive filters. However, these solutions have some limitations. Insufficient flexibility is the
biggest disadvantage of passive solutions. Passive filters are usually designed for a specific harmonic
condition, and hence change in the harmonic distortion or working conditions could degrade the
applicability of the passive filters. Furthermore, resonance problems could affect the performance of
passive filters and could increase harmonic distortion in some points of the network. Due to higher
flexibility, there is an increasing approach in PE-based active harmonic filters (AHFs) such as APLCs.
They could be optimal choices to cope with PQ related problems. In addition to the conventional
APLCs, almost all energy resources with PE interfaces may be controlled to act, also, as harmonic
active filters. Among them, ESSs equipped with PE converters could also be accounted for a possible
AHF resource [18–21].
APLCs are considered as one of the most promising solutions for harmonic compensation activities.
These devices do not suffer from conventional limitations of passive filters such as limited tuning, and
especially resonance problems. They have the flexibility to work at a broad range of harmonics at any
working point. Hence, APLCs are going to be more employed in electric networks [22–24]. An APLC
is, in fact, a voltage or current source inverter that is controlled to produce harmonic compensation
currents or voltages at the output. The APLC injects a harmonic compensation voltage or current to
the network and eliminates harmonics in the source side. Unlike passive filters, APLC produces no
resonance in the network. Hence, they are more prevalent, despite their higher investment costs and
more complex control methods [22–24].
The schematics of two common types of APLCs called shunt and series topologies are represented
in Figure 1. In both types, the voltage and current measurements of the network are processed by the
APLC controller. The controller determines the required reference harmonic currents that should be
injected by the filter. In [25,26], common control strategies for producing the desired reference current
have been investigated. The calculated reference currents are then injected into the grid by a proper
switching method. The required power for producing compensation currents can be supplied by an
external source or taken by the grid using the same power converter.

117
Energies 2020, 13, 771

Figure 1. Schematics of the shunt and series active harmonic filters.

It has been shown that distributed compensation schemes could bring more flexibility and filtering
capability for harmonic compensation [5,16]. Since the harmonic compensation is provided all over
the grid, it is preferable to compensate current harmonics as close as possible to the disturbing source.
In this way, harmonic losses can be reduced in the network. Harmonic loads are usually positioned
at various locations with different nonlinear characteristics. Therefore, the distributed compensation
scheme could have better applicability to cope with the harmonics of the dispersed nonlinear loads.
Furthermore, for reliable operation of the MG, the coordinated operation of the AHFs should be
ensured. It has been highlighted that instability problems can arise in the case of uncoordinated
operation [5,16].
It should be noted that the penetration of nonlinear loads is increasing in MGs and compensation
methods solely based on APLCs could be expensive and insufficient. Investment and operation costs
are usually high for APLC devices [5]. Hence, other possible AHFs based on other PE converters
should be utilized for more economic and flexible solutions. There is a great body of literature about
the possible utilization of ESSs for PQ improvement [2,4,6,7,11,13,27]. However, to the best of the
author’s knowledge, the coordinated operation of ESSs for PQ problems has not been paid attention
thus far. In this paper, the possibility of storage systems for the coordinated mitigation of harmonics is
addressed. The coordinated operation of ESSs was achieved by an optimization problem including
various constraints that could guarantee the economic and reliable operation of PQ improvement
activities. In the next section, the possible utilization of ESSs for PQ improvement was investigated.

3. Energy Storage Systems for Power Quality Improvement

Energy storage systems have received special attention due to their great flexibility and applicability
to the various power system sections. In MGs, an ESS converts electrical energy via a power interface,
which is usually based on a PE converter. The general characteristics of ESSs are typically determined
by their PE interfaces. Aside from PE interfaces, various management and control components are
included in ESSs that enable them to have various functionalities in the network.
The main purpose of incorporating ESSs is to participate in energy markets. The term energy
arbitrage refers to the charging of ESSs in the off-peak periods and discharging in the peak periods.
The ESS revenues come from the difference between the off-peak and peak energy prices. Transmission
and distribution services could also benefit from ESSs. The extension of network lines could be
deferred using the flexibility of ESSs. This may avoid congestion problems in both the transmission
and distribution grids, which result in more reliable operations of the system as well as reduced prices
of energy [2]. Much work has been done on the integration of DERs in MGs. In MGs, the fluctuating
output power of DERs in both grid-connected and islanded modes can cause voltage instability.
This can also result in intermittent energy resources such as wind and solar farms. The ESSs could play
a role for energy buffering and could smooth fluctuations of intermittent DER by a smart charging and
discharging procedure. ESSs could then participate in voltage regulation and compensate reactive
power to enhance the stability of the system [2].

118
Energies 2020, 13, 771

One of the main applications of ESSs is in the area of ancillary services. Among various services,
frequency regulation is the main service that could be supplied due to the fast response speed of PE
interfaces. ESSs with high energy capacity could provide fast-spinning reserves. Furthermore, ESSs
could act as AHFs and compensate for various PQ problems. Integrating ESSs in various DERs and
flexible alternating current transmission system (FACTS) devices could enable this equipment for the
compensation of extra reactive and harmonic currents. They could also avoid the PQ problems related
to the application of DER in the MG.
ESSs vary in a broad range of technologies and categories. A common classification method is
based on the form of stored energy. Mechanical (such as pumped hydro energy storage, compressed
air energy storage (CAES) and flywheel), electrochemical (such as secondary batteries as the most
useful technology), chemical (such as fuel cell), thermal (such as cryogenic energy storage and
high-temperature thermal energy storage), and electrical (such as ultra-capacitors and superconducting
energy storage systems (SMES)) are the main categories. Review papers about various ESSs technologies
are available in [7,10]. In [10], the current practical implementations of ESSs all around the world
were investigated.
Typical applications of ESSs are usually specified based on their properties in terms of power
and energy capabilities. In [4], a comprehensive comparison was made among various ESSs in which
output powers could be represented based on output duration, which is shown in Figure 2. Various
battery types and mechanical energy storage systems such as pumped hydro and CAES with high
energy density could be used in applications such a load-leveling and emergency power. Some other
ESS types such as SMES and flywheels with conventional bearings have high power capacity. Hence,
they are suitable for some PQ problems such as instantaneous voltage drop, voltage flicker, and short
duration interruptions. Other ESSs such as flywheels with levitation bearing and super-capacitors
can be employed for high power requests for a short time [4]. However, the duration depends on the
requested power from the ESS.

Figure 2. Output power and duration of various ESSs (plot obtained with data taken from [4]).

The general connection of some ESSs suitable for PQ mitigation activities is represented in Figure 3.
The energy train may consist of a back-to-back converter for power conditioning between the MG
and the main energy source. For example, flywheels work with AC power with their permanent
magnet motors. Hence, they need an AC/DC converter for charging the DC link. Sometimes, a DC/DC

119
Energies 2020, 13, 771

converter can be employed before the DC link, as shown in Figure 1. After the DC link, a grid side
converter (GSC) is used to deliver the energy from the DC link to the PCC. In order to enable the ESSs
to participate in PQ mitigation activities, the control methods of the GSC should be modified.

Figure 3. General connection of the ESSs by a voltage source inverter to the PCC.

Application of ESSs for Harmonic Compensation

Aside from AHFs and passive filters, ESSs could play an important role in the mitigation of PQ
and harmonic problems [2,4,7,10]. The ability to control the PE converters of ESSs could enable them
to act as harmonic active filters. In the literature, there are some examples of the modification of the
control algorithms of PE converters for harmonic compensation. The battery storage system is currently
the most promising ESS due to economic considerations and mature technology [27]. A smart battery
controller was proposed in [11] for enhancing the PQ and adjusting steady-state voltage and frequency
of MG. An investigation was made in [13] into the PQ of wind turbines with and without ESSs and the
results showed the positive effects of ESSs as a solution for the mitigation of wind turbine harmonics.
Optimal harmonic compensation and the economic employment of ESSs could be achieved by
the coordinated control of PE-based inverters [5]. It seems that the optimal operation of ESSs could
efficiently consider various technical and systematic constraints in the calculation harmonic sharing of
each ESS. Hence, the coordinated control of ESSs was proposed in this paper for harmonic filtering
activities as an extension for various applications of ESSs. Therefore, the utilization of various ESSs
for harmonic compensation and the coordinated control of ESSs for a global distributed harmonic
compensation scheme were proposed and formulated. To this end, a harmonic model of ESSs is
required in the coordination algorithm. Based on Figure 3, the equivalent circuit model of ESSs in
harmonic mitigation can be represented as in Figure 4. The internal current source of the ESS is
controlled based on the internal parameters of the ESS such as the voltage of the DC link, control,
and switching method of the inverter. The GSC is controlled in the current control mode for harmonic
compensation [27].

Figure 4. Harmonic model of ESSs in harmonic compensation.

Internal impedance Zf shows the Thevenin impedance of the ESS seen in the output of the GSC.
The internal impedance depends on the internal parameters, the control blocks, and the topology of
the GSC. Some calculation methodologies of the GSC internal impedance as a VSI was investigated
in [28–30]. A simple model for the Zf can be represented as Equation (1), which includes two resistive
(Rfh ) and inductive parts (Xfh ). Both parts are considered as functions of various internal parameters
such as the voltage of the DC bus (V DC ), internal inductance (L), and also as gain factors of the

120
Energies 2020, 13, 771

internal control blocks (Gfh ) of the GSC. The current gain, the voltage gain, the gain of current control
compensator, the feedforward gain, and the modulator gain are internal control parameters affecting
Gfh , which should be considered in the calculation of Zfh [28].
( ) ( )
ZGSC
fh
= RGSC
fh
GGSC
fh
, V DC
f
, L f + jXGSC
fh
GGSC
fh
, V DC
f
, Lf (1)

Finally, it should be noted that the reactance of Xfh was employed for the output filter of the ESSs,
as shown in Figure 3. The proposed model in Figure 4 could be employed for the coordination of the
distributed ESSs for harmonic compensation. In the next section, the proposed model for harmonic
compensation with ESSs is introduced.

4. Coordination Model of ESSs for Harmonic Filtering

In this section, the proposed model for the coordination of ESSs for harmonic mitigation is
introduced. The model was developed based on the concept of the harmonic power flow [5,31] and
active power filter sizing and placement [32]. Since the main power transactions of ESSs are in the
fundamental frequency, the side operation for harmonic compensation could be implemented as a
sequential operation model. Hence, the harmonic spectrum approach could be employed to calculate
the harmonic filtering sharing of each ESS after running the day-ahead energy market [33]. In the
following, the optimization model for incorporating ESSs in the operation of the MG at the fundamental
frequency is first presented. After that, the proposed model for the coordination of ESSs working as
AHFs (named the ESS-AHF model) and related working constraints are introduced.

4.1. Operation in Fundamental Frequency

In the fundamental frequency, ESSs can be used in the daily operation of the MG for energy
arbitrage. This problem can be formulated as an operation problem including ESSs [34,35]. The goal
of the MG operator is to economically supply the load demand using energy purchased from the
main grid (costgrid ), the energy of the available DGs (costDG ), and eventually, the energy arbitrage of
ESSs (costESS ). Hence, the operation cost (OC) can be mathematically formulated as the minimization
problem shown in Equation (2).
!  "
grid
min OC = costt + costDG
t + costESS
t (2)
t

The objective function in Equation (2) consists of three main costs. The energy purchased from
the main grid can be calculated with respect to the price of energy and power imported from the main
grid to the MG as shown in Equation (3). Furthermore, DGs could also inject power to the MG whose
related costs can be calculated as per Equation (4) in each operation period. For ESSs, the costs should
be calculated with respect to the injected power in discharging mode and consumed power in charging
mode. This can be mathematically formulated as in Equation (5). In Equations (3)–(5), the terms of π
stand for prices of energy in the main grid for the injected power of DGs and prices for the charging
and discharging of ESSs, respectively.

grid grid MG
costt = πt pt (3)
 
costDG
t = πDG pDG
it , ∀i ∈ Ω
DG
(4)
i it
 
costESS
t = πdch dch cha cha
it pit − πit pit , ∀i ∈ Ω
ESS
(5)
i
There are some technical and systematic constraints for the minimization of the OC, which
should be taken into account. For ESSs, the working constraints are shown in Equations (6)–(10).
Three possible working modes of ESSs can be considered: charging, discharging, and idleness modes.

121
Energies 2020, 13, 771

These modes can be modeled as Equation (6) by using the binary variables of ycha (stands for charging
mode) and ydch (stands for discharging mode). If these two variables are simultaneously adjusted
to zero, then the ESS is idle in that period. The state of charge (SOC) of ESS (cet ) was calculated in
Equation (7) in each operation period. It was determined based on the SOC at the beginning of the
period in addition to power transactions in charging (pcha ) or discharging (pdch ) modes. In Equation (7),
charging and discharging efficiencies are also considered as ηcha and ηdch , respectively. Constraint (8)
is proposed for the calculation of the injection current (i) of the ESS to the connection bus in each mode
using the conjugate of bus voltage, vit , and the active power of ESS. Since the operation modes are
separated using binary variables in Equation (6), pcha and pdch do no occur simultaneously. Hence,
in Equation (8), either charging or discharging of the ESS could take place. Consequently, the injection
current could be positive or negative, with the injection current considered to be positive in charging
mode. In Equation (9), the working limits of the SOC are considered using Cmin and Cmax , respectively,
for the minimum and maximum limits. The maximum and minimum limits of active power are also
considered in Equation (10).
et + yet ≤ 1
ycha dch
(6)
1
ce,(t+1) = cet + ηcha cha
e pet − pdch
et (7)
ηdch
e

pcha
et pdch
et
iit = − , ∀i ∈ ΩESS (8)
v∗it v∗it

Cmin
e ≤ cet ≤ Cmax
e (9)

Pmin
e ≤ pcha dch
et , pet ≤ Pmax
e (10)

Power flow constraints are the main system constraints in the optimization model. Current
injections of loads, DGs, and nonlinear loads could be calculated in Equation (11) using the apparent
power (S) and the voltage of the connecting bus (vi ) at each operation period. The injection shown
in Equation (11) is valid for the set of load buses (ΩPQ ), the set of DG buses (ΩDG ), and finally the
set of nonlinear buses (ΩNL ). Without losing generality, it is assumed that the DGs work in constant
power mode. Anyway, other operation modes could simply be applied in the model [31]. Furthermore,
the efficient linear power flow constraints of [31] were employed for modeling the network. In this
power flow model, the MG was modeled with some incidence matrices using graph theory. Power
flow constraints are modeled in Equations (12) and (13) using A and B incidence matrices employed
for modeling the MG. In these equations, the matrices of the line currents, bus injections, and voltage
drops of buses are shown with [U], [I], and [ΔV], respectively. It has been shown in [31] that this
power flow model has no convergence problem in MGs and can easily handle meshed topologies and
unbalanced networks.  ∗
S 
iit = , ∀i ∈ ΩPQ , ΩDG , ΩNL (11)
vit
[U ] = [A][I ] (12)
T
[ΔV ] = [B] [Z][U ] (13)

Nevertheless, modeling the energy arbitrage mechanism of ESSs was not the main concern of this
paper. The output of the fundamental frequency optimization model is the SOC of each ESS and the
working points in the operation horizon were used for calculations in harmonic frequencies.

4.2. Coordinated Operation of ESSs as AHF

In this subsection, the proposed model for employing ESSs as AHFs (named the ESS-AHF model)
is introduced. The goal of the MG operator is to minimize the cost of procurement of PQ improvement
actions. Since the harmonics were only investigated in this paper, this goal could be reached using

122
Energies 2020, 13, 771

AHF resources, APLCs, and ESSs. Hence, the objective function can be mathematically shown as
Equation (14).
!     ( )"
min PQ cost = costAPLC
th
+ costESS
th
= πAPLC
fh
dAPLC
f th
+ πESS dESS
f h f th
(14)
th f th

The cost of PQ improvement actions includes two main costs: costs related to APLCs denoted by
the costAPLC and costs related to the ESSs working as AHFs, shown with costESS . The calculation of these
two cost terms is also represented in Equation (14) using the amount of the provided compensation
distortion power (d) and the offered compensation price (π) for each AHF (f ) at each period. In harmonic
frequencies, due to the increase in power losses of the AHF, some amount of distortion power is
imposed on the AHF. The payments are required for covering such costs for the owners of AHFs.
Hence, offered prices of harmonic compensation, currents should be capable to refund extra costs
imposed on the filter. On the other hand, since the required compensation actions are supplied in a
competitive mechanism, the offered prices should ensure sufficient participation of the ESS owner in
PQ improvement activities. Moreover, the offered prices of ESSs are generally lower than the prices
of APLCs, since the harmonic compensation action is taken as a side function from ESSs [5]. Hence,
some operation costs related to the investment cost can be decreased for ESSs. Anyway, the cost of all
AHFs can be calculated using the offered price of each AHF and the related provided compensation
distortion power as shown in Equation (14).
The distortion power generally includes the current distortion power [Marini, 2019 #445]. Hence,
the provided compensation distortion power could be shown as Equation (15) using the provided
harmonic compensation currents in each harmonic order, ifth . In this equation, the distortion power is
assumed to be limited by the maximum available power of the AHF, Dmax . For APLCs that do not
contribute at the fundamental frequency, this limitation is simply considered as the rating volt-ampere
of the filter. However, for AHFs based on ESSs, considerations should be made for power transactions
in the fundamental frequency. This limitation is set as the maximum value of available power of the
ESS (Pmax –Pet,1 ) for idle periods and allowable 10% variation in output power (Pet,1 ) for charging or
discharging periods. This constraint is mathematically shown as the maximum value of these two
values in Equation (16). Furthermore, with respect to the general model of AHF as shown in Figure 4,
working constraints could be formulated as Equations (17) and (18). In harmonic studies, harmonic
voltages and currents are superimposed to the fundamental voltage and current. Hence, voltage and
currents should be replaced with harmonic root mean square equivalents, as shown in Equations
(17) and (18). The maximum current of AHF is limited to the maximum value, Imax in Equation (17).
In Equation (18), the maximum allowable limit of AHF is considered as V max .
 ( )2
d2f t = V 2f t,1 i2f ht ≤ Dmax
ft
(15)
h>1

Dmax
ft
= max Pmax
e − Pet,1 , 10%(Pet,1 ) , ∀ f ∈ ESS (16)
 ( )2
I2f t,1 + i2f ht ≤ Imax
f
, ∀f (17)
h>1
 ( )2
V 2f t,1 + v2f ht ≤ V max
f
, ∀f (18)
h>1

The SOC of the ESS should also be taken into account in harmonic conditions. Although harmonic
power transactions have low values, their effects on the SOC of the ESS should be considered. Harmonic
active power (pfth ) can be calculated using the real value of the product of harmonic voltage and current
as shown in Equation (19). The SOC of the ESS should be modified with respect to the provided active
harmonic power. Since the PE converter of the ESS is generally a four-quadrant converter, harmonic

123
Energies 2020, 13, 771

active powers could be positive and negative. This could lead the ESS to be charged or discharged to
provide the harmonic compensation current. Variations in the SOC in harmonic condition (cfth ) with
respect to the provided active harming powers is mathematically shown in Equation (20). In Equation
(20), the SOC in the harmonic condition is supposed to be equal to the SOC for fundamental frequency
(Cft ) plus harmonic active powers, pfth . Harmonic reactive power is usually achieved by changing the
phase angle and modulation method of the converter. It can be considered as extra losses of active
power and is usually considered as a percentage of active harmonic power [36]. This modification could
be applied in Equation (20), which could be a subject for future research. Finally, the minimum and
maximum limitations of the SOC should also be met in harmonic activities, which is mathematically
shown in Equation (21). 
p f th = Re vith i∗ith , ∀i ∈ ΩESS (19)

c f th = C f t + p f th , ∀ f ∈ ESS (20)
h>1

Cmin
f
≤ c f th ≤ Cmax
f
, ∀ f ∈ ESS (21)

Harmonic standards are usually reported as total harmonic distortion (THD) and individual
harmonic distortion (IHD). Typical levels of harmonic standards for grid voltage are 5% and 3% for
THD and IHD, respectively [37]. The important goal of the MG operator is the economic satisfaction of
the desired harmonic levels across the entirety of the MG at each period. These constraints are shown
in Equations (22)–(23), respectively, for THD and IHD. In each operation period, the THD could be met
for each bus and the IHD should be met for each bus in each harmonic order.
 2 2
v2ith ≤ Vih,1 THDmax
i , ∀i, t (22)
h>1

2 2
v2ith ≤ Vit,1 IHDmax
i , ∀i, ∀t, ∀h (23)

The harmonic power flow constraints should also be considered in the model. Since the provided
harmonic current compensation is injected to the MG in different nodes, these constraints are essential
for calculations of the flow of harmonic powers. The constraints of Equations (24) and (25) are
the harmonic counterparts of fundamental power flow previously shown in Equations (12) and (13).
Furthermore, the harmonic spectrum approach is used for the harmonic modeling of nonlinear
loads [33]. It is required to determine the magnitude and phase angle of harmonic injections of
nonlinear loads in this approach. In Equation (26), the magnitude of harmonic injection (iiht ) is
determined based on the harmonic spectrum of nonlinear load (Ispec ) and current injection of the load
in the fundamental frequency (Iit,1 ) for all buses of the network with nonlinear load (ΩNL ). The phase
angle of harmonic injection (θith ) was also determined in Equation (27) using the phase spectrum of the
nonlinear load (θspec ) and the phase angle of its current injection in the fundamental frequency (θit,1 ).

[Uh ] = [Ah ][Ih ] (24)

[ΔVh ] = [Bh ]T [Zh ][Uh ] (25)

spec
iiht = Iiht × Iit,1 , i ∈ Ω NL
(26)
spec
θith = θith + hθit,1 + π, i ∈ Ω NL
(27)

The developed model in Equations (14)–(27) is the complete model proposed for the coordinated
operation of ESSs working as AHFs. It should be noted that the operation model of Equations (2)–(13)
and the ESS-AHF model of Equations (14)–(27) are related and the ESS-AHF model will be implemented
after the operation model. Hence, the ESS-AHF model could be accounted for as a sequential model,
as shown in Figure 5. The output nonlinear load current of the operation model is the setpoints for
the calculation of harmonic injections. These dependencies are mathematically shown in Equations

124
Energies 2020, 13, 771

(26) and (27). Harmonic injections are calculated in Equations (26) and (27) based on the harmonic
spectrum of the nonlinear load. Hence, these dependencies could affect the ESS-AHF model since
the harmonics will be related to harmonic injections of nonlinear loads. Furthermore, since ESSs are
employed as AHFs in addition to APLCs, the working conditions of ESSs could be affected by the
output of the operation model. The maximum available capacity of the distortion power for ESSs is set
as the maximum value of available power of the ESS and allowable 10% variation in output power for
charging or discharging periods, as shown in Equation (16). Hence, these setpoints could change the
ESS activities as AHFs.

Figure 5. The ESS-AHF model as a sequential model after the operation model.

The output of the ESS-AHF model is the calculated harmonic reference currents of the ESSs,
which should be injected into the MG. These reference currents are then transmitted to the ESSs by
the available infrastructures of the MG, and the AHF works in the current control mode. Various
technical constraints, harmonic conditions of the network, and the status of AHFs should be sent to
the MG control center for consideration in the optimization problem. All of these require a proper
communication infrastructure that could be available in future MGs. The availability of communication
links in AHF buses, its bandwidth, and time delay are important challenges of the communication
system that could affect the applicability of the proposed optimization method, which should be taken
into account in future studies. In the next section, the results of the implementation of the proposed
model will be represented.

5. Simulations
A modified version of the IEEE 33-bus test network was used to demonstrate the performance of
the ESS-AHF model. The test grid consists of 32 branches and five tie-switches, as shown in Figure 6.
Line and bus data were taken from [38] and are provided in Appendix A. In order to increase the effects
of harmonics, line impedances were increased by a factor of three. Data for the daily load demand
were taken from [39].
Four nonlinear loads were connected to the test network. The locations and active and reactive
powers of nonlinear loads are shown in Table 1. A general non-linear load model as a six-pulse diode
bridge rectifier was adopted for all nonlinear loads [33]. The represented harmonic loads are the main
sources of harmonic distortions in the MG. The harmonic spectrum of nonlinear loads is reported in
Table 2, where the data were taken from [33]. In the harmonic spectrum approach, the magnitude and
phase of the harmonic injections of nonlinear loads are used to calculate the flow of harmonic power in
the network individually at each harmonic frequency.
Five DERs were added to the test network in selected buses, as shown in Figure 6. A parking lot
of electric vehicles was considered at bus #12, a wind turbine was installed at bus #21, a photovoltaic
(PV) generator was installed at bus #24, a battery energy storage was connected at bus #28, and finally,
a PV array was connected at bus #31 equipped with a battery energy storage. The parking lot at bus
#12 included a battery ESS. Hence, three ESSs have been placed in the test network which could be
used as possible AHFs. Furthermore, two APLCs were connected to the network in buses #2 and #9,
as shown in Figure 6. Hence, the total number of AHFs was equal to five filters. Data of the DER and

125
Energies 2020, 13, 771

APLCs are presented in Table 3. The maximum and minimum of the SOC and maximum power of
each ESS are also reported in this table. In the last column, the offered price of distortion power is
shown based on $/kVARh. The offered prices for ESSs were lower than the APLCs since harmonic
filtering is considered as a side function from ESSs, which leads to a reduced distortion power cost.

Figure 6. Modified IEEE 33-bus test network.

Table 1. Nonlinear loads of the network.

Nonlinear Load Bus P (pu) Q (pu)

1 14 0.150 0.120
2 20 0.125 0.105
3 23 0.110 0.100
4 30 0.145 0.125

Table 2. Harmonic spectrum of nonlinear loads.

Harmonic Magnitude Phase (◦ )

5 0.35 180
7 0.43 180
11 0.05 0
13 0.08 0
17 0.04 180
19 0.04 180

The lower and upper bounds of voltage were taken as 0.95 pu and 1.05 pu, respectively.
The operation horizon for unit commitment problem is considered the next day, which is divided
into hourly periods. The proposed optimization models (in the form of a mixed-integer nonlinear
programming model) were implemented in GAMS software (GAMS Development Corporation,
Washington, D.C., USA) [40] in a personal computer equipped with a 2.93 GHz Intel processor and 8
GB memory. Without loss of generality, it was assumed that the MG works in grid-connected mode.
Hence, the voltage at the PCC is assumed to be sinusoidal.

126
Energies 2020, 13, 771

Table 3. Data of APLCs and DER of the network (values are in pu).

No ID Type Bus Smax Imax Cmin Cmax Pmax π ($/kVARh)

1 APLC #1 APLC 2 0.22 0.43 — — 0.32 860
2 APLC #2 APLC 9 0.18 0.35 — — 0.26 900
3 ESS #1 Electric vehicle + ESS 12 0.1 0.1 0.1 0.25 0.1 395
4 DG #1 Wind turbine 21 0.15 0.15 — — 0.15 —
5 DG #2 Photovoltaic 24 0.08 0.1 — — 0.08 —
6 ESS #2 Battery storage 28 0.15 0.25 0.15 0.50 0.15 380
7 ESS #3 Photovoltaic + ESS 31 0.1 0.35 0.1 0.22 0.1 380

Three different cases were considered to show the performance of the ESS-AHF model.

• Case I: ESSs are only used for energy arbitrage to reduce the cost of supplying energy to
the customers.
• Case II: Employing APLCs for harmonic compensation imposed by nonlinear loads.
• Case III: Employing ESSs as AHFs to satisfy harmonic standards.

The simulation results are reported in the following three subsections.

5.1. Case I: The Base Case

In the base case, the unit commitment model of 4.1 was implemented on the test network.
The various cost terms after running the model are summarized in Table 4. The total OC was equal to
2410 k$, 87% of which was due to the energy supplied by the main grid. The other 13% of the OC was
due to the energy produced or exchanged by DGs and ESSs. The cost of the energy supplied by DGs
was equal to 248 k$ and represented the revenue of DG owners for participating in the energy market.
In the revenue of DG #2, the PV at bus #24 was higher compared to other DERs for its proper location.
Moreover, smart charging and discharging of ESSs implied a revenue of 58 k$ from energy arbitrage.
This revenue is, of course, a cost for the MG operator. In particular, ESS #2, the battery storage at bus
#28, earnt about 50% of the total ESSs revenues.

Table 4. Operation cost for the test network in Case I.

Total OC (k$) Grid Cost (k$) DG Cost (k$) ESS Cost (k$)
248 58
2410 2104 DG #1 DG #2 ESS #1 ESS #2 ESS #3
86.4 161.6 17 28 13

The charging profile as well as the SOC of ESSs in the operation horizon is shown in Figure 7.
All ESSs were charged from 3 am to 6 am, which are the off-peak periods of the operation horizon.
Similarly, ESSs are discharged in peak periods from 5 pm to 7 pm. In Figure 7b, the SOCs of ESSs are
represented. Allowable limits in Table 4 for the SOC and maximum power are met by the optimization
model. Variation of the SOC for ESS #2 with more energy capacity at off-peak and peak periods can be
seen in this figure.
However, the harmonic conditions should also be considered for PQ monitoring of the MG.
In Figure 8, the voltage THD of the MG was analyzed. The voltage THD of all buses at 5 am (as the
lowest) and 6 pm (as the highest load demand), are shown in Figure 8a. Obviously, either for off-peak
periods or peak periods, the THD was above the standard level for nearly all system buses. All other
periods may have a THD trend between these two selected periods. Bus #17 had the maximum THD
compared to other buses. In Figure 8b, the voltage THD of this bus was represented in the operation
horizon. The standard THD level is never met and hence, compensation actions are necessary.

127
Energies 2020, 13, 771

Figure 7. Case I. (a) Charging profile and (b) the SOC of each ESS.

Figure 8. Case I. (a) Voltage THD for all buses at t = 5 and t = 18 and (b) maximum voltage THD of all
buses at each hour.

In order to mitigate the harmonics effects from the grid, two simulation cases are provided to
show the performance of the ESS-AHF model.

5.2. Case II: Harmonic Compensation Using APLCs

In Case II, only APLCs were employed for harmonic mitigation. APLCs were placed in the
network based on filtering requirements and economic conditions by optimization models. The voltage
THD of the system buses and the maximum THD after compensation of the APLCs are shown in
Figure 9 for this case. Although the harmonic pollution was improved when compared to Case I,
it was still above the standard level of 5%. This situation is worthwhile in peak periods, as shown in
Figure 9a, since harmonic injections of nonlinear loads are increased in these periods. The maximum
THD was for bus #32, which was about 11%. In Figure 9b, the maximum THD is shown for this bus.
For off-peak periods, APLCs could adjust the voltage THD at the standard level. However, most of the
time, the voltage THD is high, which could be the result of insufficient harmonic filtering capacity in
the MG.

128
Energies 2020, 13, 771

Figure 9. Case II. (a) Voltage THD for all buses at t = 5 and t = 18 and (b) maximum voltage THD of all
buses at each hour.

In Figure 10 the loading of each APLC is shown for Case II. In almost all periods (except off-peak
ones), the APLCs were loaded at their maximum rating. However, they could not satisfy the harmonic
standard in the MG. Hence, more harmonic filters are required to be employed for more filtering
capacity in the MG. This could be achieved using more APLCs. However, it seems that this is not
an economic solution. With respect to the high investment costs of the APLCs, it could significantly
increase the cost of PQ compensation actions.

Figure 10. The loaded capacity of the APLCs for harmonic filtering in Case II.

In the next case, the performance of ESSs for improving the harmonic situation of the MG as an
economical alternative to APLCs is presented.

5.3. Case III: Harmonic Compensation Using ESSs

In order to increase the harmonic filtering capability of the MG as well as reduce the costs of PQ
improvement actions, ESSs are also employed in this case for harmonic compensation. To this end, the
unloaded power capacity of ESSs is used for harmonic filtering activities. Even if the ESS does not
contribute to energy arbitrage, it can provide harmonic filtering capacity for the network. The results
of the implementation of the ESS-AHF model are represented in this subsection.
The voltage THD and the maximum THD at each hour are represented in Figure 11. As it could
be seen, the maximum THD in all periods of the operation horizon was adjusted at the standard level.
For either the off-peak periods or peak periods, the voltage THD was in the acceptable range. Hence,
using ESSs as AHFs has provided more harmonic filtering capacity for the MG. Furthermore, since the
ESSs are dispersed in the MG, the provided compensation currents could be considered as distributed
harmonic compensation actions. In this scheme, harmonic compensation currents are injected in the
grid, and hence harmonic propagation and the resulting losses are strongly reduced. In Figure 13b,

129
Energies 2020, 13, 771

the maximum THD of voltage buses is shown, which was adjusted to the 5% standard level. This
shows the effects of ESSs for the adjustment of harmonic standards for MGs.

Figure 11. Case III. (a) Voltage THD for all buses at t = 5 and t = 18 and (b) maximum voltage THD of
all buses at each hour.

The loadings of various AHFs are shown in Figure 12 for Case III. AHFs based on ESSs are almost
loaded at their maximum capacity. This is the result of their lower prices for compensation distortion
power as well as their location. Since the APLCs have expensive distortion power, their compensation
loadings are performed after ESSs and generally have lower values compared to ESSs. APLC #2 had
more compensation loading despite its higher cost. This shows the effects of filter location on its
provided harmonic compensation to the MG.

Figure 12. The loaded capacity of AHFs for harmonic filtering in Case III.

The harmonic active power and the SOC of ESSs after harmonic compensation activities are
reported in Figure 13. The provided harmonic active power was negative for all ESSs, which shows
that it worked in discharging mode for harmonic compensation. However, the harmonic injection
powers had low values compared to the fundamental active power previously shown in Figure 7.
This means that participating in harmonic compensation activities does not have a significant effect on
the SOC of ESSs, which can be seen Figure 13b. The trend of the SOC was close to the trend of that in
Figure 7 and all allowable limits were met in the harmonic conditions. Hence, the harmonic utilization
of ESSs could bring various advantages for the MG, and meanwhile, does not affect the ESS parameters
for energy arbitrage. The ESSs earn from both energy arbitrage and harmonic compensation.

130
Energies 2020, 13, 771

Figure 13. Case III. (a) Discharging profile and (b) the SOC of each ESS for harmonic compensation.

The required reference currents of the ESSs at each period are determined in the ESS-AHF model.
Fundamental harmonic and reference currents for ESS #2 at the peak period (i.e., 6 pm) are shown in
Figure 14. The bold curve is the required reference current calculated by the ESS-AHF model, which
should be injected by the ESS. The reference current should then be sent to the ESS by the available
infrastructures of the MG. The ESS is responsible to inject the required current by the MG operator
with a suitable control and switching method.

Figure 14. Fundamental harmonic and reference currents of ESS #2 in the peak period, (6 pm) in
Case III.

For economic comparisons, the total injections and compensation costs for Case III are presented
in Table 5. The total required distortion power was 8.95 pu and 6.29 for Case II and Case III, respectively.
Although the compensation distortion power in Case III was about 70% of that in Case II, it could
successfully meet the desired level of harmonics in the MG, as shown in Table 5. Furthermore,
the loading of compensation currents was more achieved by the ESSs in Case III as they were more
economic AHFs compared to the APLCs. This reduces the total compensation cost from 7868 k$ in
Case II to 3424 k$ in Case II. It showed a 44% decrease in the compensation cost. Hence, the MG
operator could keep the desired harmonic level with only 44% of the cost of the APLCs. This could
save remarkable costs for the MG operator, either in the operation of the APLCs or as requirements for
the installation of new APLCs.

131
Energies 2020, 13, 771

Table 5. Total harmonic compensation injection of AHF in Case II and Case III.

ID APLC #1 APLC #2 ESS #1 ESS #2 ESS #3 Total

Injection (pu) 4.63 4.32 0 0 0 8.95
Case II
Cost ($) 3980 3888 0 0 0 7868
Injection (pu) 0.16 1.79 1.74 1.29 1.31 6.29
Case III
Cost ($) 140 1607 689 491 497 3424

The simulations showed the performance of the ESS-AHF model in the coordination of distributed
ESSs for harmonic compensation in the MG. More filtering capacity and reduced compensation costs
are the main advantages of ESSs in harmonic compensation. Since harmonic compensation is utilized
as a side function, the provided distortion power has a lower price than APLCs. This implies that
compensation loading is first demanded by the ESSs. Furthermore, ESSs are dispersed in the MG,
and hence harmonic compensation will be distributed in the MG. The provided compensation scheme
will be accounted for by distributed harmonic compensation with more flexibility and applicability
for PQ improvements. The ESS owners could also benefit from both energy arbitrage and harmonic
compensation. This could decrease the investment costs of the ESSs and make them more justified for
use in the MG. The simulations showed that ESS compensation activities do not have much of an effect
on the SOC and ESS revenues from energy arbitrage. Finally, it should be noted that, unlike energy
and ancillary markets, PQ improvement actions are not security obligations of the network. Therefore,
any time that the working of ESSs in PQ improvement actions affects their application in ancillary
services and the security of the network, it could be temporarily ignored.

6. Conclusions
In this paper, the applicability of ESSs for harmonic filtering activities was investigated.
The common procedure for this was to modify the control methods of the PE converters of ESSs
to enable them for the injection of compensation currents. However, in this paper, the coordinated
operation of ESSs for a distributed harmonic compensation scheme was addressed. The proposed
coordination model employed conventional control algorithms for the injection of a reference current
that was determined in the proposed ESS-AHF optimization model. Hence, the provided compensation
scheme could be considered as a distribution compensation scheme. The simulation cases showed
the performance of the proposed model in enhancing the filtering capability of the MG, the reduction
in compensation cost, and the greater flexibility of the distributed compensation schemes. Hence,
requirements for new filters could be deferred and both the MG operator and the ESS owner could
benefit from reduced compensation costs.
ESSs can be used to perform several ancillary services where most of them need the energy stored
to be accomplished. What is demonstrated in this paper is that the ancillary service for harmonic power
filtering can be performed without significantly affecting the battery SOC. For this reason, this service
can be achieved for free, without compromising the others, and therefore the revenues from energy
arbitrage and other ancillary services can be maintained. This could make ESSs more economical for
use in future power systems. Investigations should be carried out for various technical and systematic
challenges of the proposed ESS-AHF model in future research.

Author Contributions: A.M. and L.P. conceived and designed the models; A.M. performed the experiments and
simulation; A.M., L.P., and S.-S.M. analyzed the data; M.-S.G. contributed reagents/materials/analysis tools; A.M.
and L.P. wrote the paper. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Conflicts of Interest: The authors declare no conflicts of interest. The founding sponsors had no role in the design
of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the
decision to publish the results.

132
Energies 2020, 13, 771

Nomenclature
Note that single parameters, single variables, and matrices are shown with capital letters, small letters, and capital
letters in brackets, respectively, in the paper. When each notation comes with a specific index and superscript,
it stands for the corresponding variable or parameter of related indices. Repetitive definitions can be ignored
using this method. For example, the letter ‘v’ always stands for voltage. When it comes in brackets, it stands
for the matrix of all voltage variables; when it comes individually with index i, it stands for a single variable of
voltage for bus i, and Vit,1 stands for the voltage parameter of bus i at period t for the fundamental frequency,
h = 1. The notations are as follows:

Abbreviations ΩPQ Set of all load buses

AHF Active harmonic filter ΩNL Set of all nonlinear buses
DER Distributed energy resource ΩNL Set of all nonlinear buses
DG Distributed generation ΩPQ Set of all load buses
DS Distribution system Superscripts
GSC Grid side converter cha Charge (for ESSs)
MG Microgrid dch Discharge (for ESSs)
OC Operation cost spec Harmonic spectrum
PCC Point of common coupling Notations
PE Power electronics A Branch current to bus injection incidence matrix
PQ Power quality B Branch current to bus injection incidence matrix
SOC State of charge (for ESSs) c SOC of ESS
THD Total harmonic distortion d Distortion power
VSI Voltage source inverter G Gain factor of GSC
Indices i Bus injection currents.
e ESSs of the network p Active power.
f Set of available AHF resources. q Reactive power.
h Harmonic order. S Apparent power
i Network buses. u Line currents
l Network branches v Bus voltages
t Operation horizon y Binary variable of ESS mode
ΩDG Set of all DG buses Z = R + jXImpedance = Resistance + Reactance
ΩESS Set of all ESS buses π Price of energy

Appendix A
Data of the 33-bus IEEE test network are listed in Tables A1 and A2, respectively.

Table A1. Bus data of 33-bus test network (pu).

No Pi Qi
1 0 0
2 0.022 0.013
3 0.02 0.009
4 0.027 0.018
5 0.013 0.007
6 0.013 0.004
7 0.044 0.022
8 0.044 0.022
9 0.013 0.004
10 0.013 0.004
11 0.01 0.007
12 0.013 0.008
13 0.013 0.008
14 0.027 0.018
15 0.013 0.002
16 0.013 0.004
17 0.013 0.004

133
Energies 2020, 13, 771

Table A1. Cont.

No Pi Qi
18 0.02 0.009
19 0.02 0.009
20 0.02 0.009
21 0.02 0.009
22 0.05 0.0225
23 0.02 0.011
24 0.013 0.004
25 0.093 0.044
26 0.013 0.006
27 0.093 0.044
28 0.013 0.004
29 0.027 0.016
30 0.044 0.133
31 0.033 0.016
32 0.047 0.022
33 0.013 0.009

Table A2. Line data of 33-bus network (pu).

No From To R X
1 1 2 0.0004 0.0002
2 2 3 0.002 0.001
3 3 4 0.0015 0.0008
4 4 5 0.0016 0.0008
5 5 6 0.0034 0.0029
6 6 7 0.0008 0.0026
7 7 8 0.0029 0.001
8 8 9 0.0042 0.003
9 9 10 0.0043 0.003
10 10 11 0.0008 0.0003
11 11 12 0.0015 0.0005
12 12 13 0.0061 0.0048
13 13 14 0.0022 0.0029
14 14 15 0.0024 0.0022
15 15 16 0.0031 0.0022
16 16 17 0.0053 0.0071
17 17 18 0.003 0.0024
18 2 19 0.0007 0.0006
19 19 20 0.0062 0.0056
20 20 21 0.0017 0.002
21 21 22 0.0029 0.0039
22 3 23 0.0019 0.0013
23 23 24 0.0037 0.0029
24 24 25 0.0037 0.0029
25 6 26 0.0008 0.0004
26 26 27 0.0012 0.0006
27 27 28 0.0044 0.0038
28 28 29 0.0033 0.0029
29 29 30 0.0021 0.0011
30 30 31 0.004 0.004
31 31 32 0.0013 0.0015
32 32 33 0.0014 0.0022

134
Energies 2020, 13, 771

References
1. Koohi-Kamali, S.; Tyagi, V.; Rahim, N.; Panwar, N.; Mokhlis, H. Emergence of energy storage technologies as
the solution for reliable operation of smart power systems: A review. Renew. Sustain. Energy Rev. 2013, 25,
135–165. [CrossRef]
2. Zakeri, B.; Syri, S. Electrical energy storage systems: A comparative life cycle cost analysis. Renew. Sustain.
Energy Rev. 2015, 42, 569–596. [CrossRef]
3. Luo, X.; Wang, J.; Dooner, M.; Clarke, J. Overview of current development in electrical energy storage
technologies and the application potential in power system operation. Appl. Energy 2015, 137, 511–536.
[CrossRef]
4. Kondoh, J.; Ishii, I.; Yamaguchi, H.; Murata, A.; Otani, K.; Sakuta, K.; Higuchi, N.; Sekine, S.; Kamimoto, M.
Electrical energy storage systems for energy networks. Energy Convers. Manag. 2000, 41, 1863–1874.
[CrossRef]
5. Marini, A.; Ghazizadeh, M.-S.; Mortazavi, S.S.; Piegari, L. A harmonic power market framework for
compensation management of DER based active power filters in microgrids. Int. J. Electr. Power Energy Syst.
2019, 113, 916–931. [CrossRef]
6. Beleiu, H.; Beleiu, I.; Pavel, S.; Darab, C. Management of Power Quality Issues from an Economic Point of
View. Sustainability 2018, 10, 2326. [CrossRef]
7. Das, C.K.; Bass, O.; Kothapalli, G.; Mahmoud, T.S.; Habibi, D. Overview of energy storage systems in
distribution networks: Placement, sizing, operation, and power quality. Renew. Sustain. Energy Rev. 2018, 91,
1205–1230. [CrossRef]
8. Buła, D.; Lewandowski, M. Comparison of frequency domain and time domain model of a distributed power
supplying system with active power filters (APFs). Appl. Math. Comput. 2014, 267, 771–779. [CrossRef]
9. Kalair, A.; Abas, N.; Kalair, A.; Saleem, Z.; Khan, N. Review of harmonic analysis, modeling and mitigation
techniques. Renew. Sustain. Energy Rev. 2017, 78, 1152–1187. [CrossRef]
10. Mohamad, F.; Teh, J.; Lai, C.-M.; Chen, L.-R. Development of energy storage systems for power network
reliability: A review. Energies 2018, 11, 2278. [CrossRef]
11. Alshehri, J.; Khalid, M.; Alzahrani, A. An Intelligent Battery Energy Storage-Based Controller for Power
Quality Improvement in Microgrids. Energies 2019, 12, 2112. [CrossRef]
12. Khani, H.; Farag, H.E. Optimal scheduling of energy storage to mitigate power quality issues in power
systems. In Proceedings of the 2017 IEEE Power & Energy Society General Meeting, Chicago, IL, USA, 16–20
July 2017; pp. 1–5.
13. Ramos, G.; Rios, M.; Gómez, D.; Palacios, H.; Posada, L. Power quality study of large-scale wind farm
with battery energy storage system. In Proceedings of the 2017 IEEE Industry Applications Society Annual
Meeting, Cincinnati, OH, USA, 1–5 October 2017; pp. 1–6.
14. Yang, D. Informatics in Control, Automation and Robotics; Springer: Berlin/Heidelberg, Germany, 2012; Volume 2.
15. Arulampalam, A.; Barnes, M.; Jenkins, N.; Ekanayake, J.B. Power quality and stability improvement of a
wind farm using STATCOM supported with hybrid battery energy storage. IEE Proc. Gener. Transm. Distrib.
2006, 153, 701–710. [CrossRef]
16. Marini, A.; Piegari, L.; Mortazavi, S.S.; Ghazizadeh, M.-S. Active Power Filter Commitment for Harmonic
Compensation in Microgrids. In Proceedings of the 45th Annual Conference of the IEEE Industrial Electronics
Society, IECON 2019, Lisbon, Portugal, 14–17 September 2019; pp. 7038–7044.
17. Das, J. Power System Harmonics and Passive Filter Designs; John Wiley & Sons: Hoboken, NJ, USA, 2015.
18. Abolhassani, M.T.; Enjeti, P.; Toliyat, H. Integrated doubly fed electric alternator/active filter (IDEA), a viable
power quality solution, for wind energy conversion systems. IEEE Trans. Energy Convers. 2008, 23, 642–650.
[CrossRef]
19. Abolhassani, M.T.; Enjeti, P.; Toliyat, H.A. Integrated doubly-fed electric alternator/active filter (IDEA), a
viable power quality solution, for wind energy conversion systems. In Proceedings of the Conference Record
of the 2004 IEEE Industry Applications Conference, 2004. 39th IAS Annual Meeting, Seattle, WA, USA, 3–7
October 2004; Volume 2033, pp. 2036–2043.
20. Ghatpande, O.A. Harmonic Compensation in A Grid Using Doubly Fed Induction Genertors. Master’s
Thesis, Missouri University of Science and Technology, Rolla, MO, USA, 2013.

135
Energies 2020, 13, 771

21. Naidu, N.S.; Singh, B. Doubly fed induction generator for wind energy conversion systems with integrated
active filter capabilities. IEEE Trans. Ind. Inform. 2015, 11, 923–933. [CrossRef]
22. Ying-Kwun, C. Determination of locations and sizes for active power line conditioners to reduce harmonics
in power systems. IEEE Trans. Power Deliv. 1996, 11, 1610–1617. [CrossRef]
23. Hong, Y.-Y.; Hsu, Y.-L.; Chen, Y.-T. Three-phase active power line conditioner planning. In Generation,
Transmission and Distribution, IEE Proceedings. IET 1998, 145, 281–287.
24. Alfonso, J.L.; Gonçalves, H.; Pinto, J. Active Power Conditioners to Mitigate Power Quality Problems in Industrial
Facilities; IntechOpen Publisher: London, UK, 17 April 2013; Open Access.
25. Griffo, A.; Carpinelli, G.; Lauria, D.; Russo, A. An optimal control strategy for power quality enhancement in
a competitive environment. Int. J. Electr. Power Energy Syst. 2007, 29, 514–525. [CrossRef]
26. Mortezaei, A.; Simoes, M.; Savaghebi, M.; Guerrero, J.; Al-Durra, A. Cooperative control of multi-master-slave
islanded microgrid with power quality enhancement based on conservative power theory. IEEE Trans. Smart
Grid 2016, 99, 1.
27. Wasiak, I.; Pawelek, R.; Mienski, R. Energy storage application in low-voltage microgrids for energy
management and power quality improvement. IET Gener. Transm. Distrib. 2013, 8, 463–472. [CrossRef]
28. Cespedes, M.; Sun, J. Impedance modeling and analysis of grid-connected voltage-source converters. IEEE
Trans. Power Electron. 2014, 29, 1254–1261. [CrossRef]
29. Cao, W.; Ma, Y.; Zhang, X.; Wang, F. Sequence impedance measurement of three-phase inverters using a
parallel structure. In Proceedings of the Applied Power Electronics Conference and Exposition (APEC),
Charlotte, NC, USA, 15–19 March 2015; pp. 3031–3038.
30. Cao, W. Impedance-Based Stability Analysis and Controller Design of Three-Phase Inverter-Based Ac
Systems. Ph.D. Thesis, University of Tennessee, Knoxville, TN, USA, 2017.
31. Marini, A.; Mortazavi, S.; Piegari, L.; Ghazizadeh, M.-S. An efficient graph-based power flow algorithm for
electrical distribution systems with a comprehensive modeling of distributed generations. Electr. Power Syst.
Res. 2019, 170, 229–243. [CrossRef]
32. Farhoodnea, M.; Mohamed, A.; Shareef, H.; Zayandehroodi, H. Optimum placement of active power
conditioner in distribution systems using improved discrete firefly algorithm for power quality enhancement.
Appl. Soft Comput. 2014, 23, 249–258. [CrossRef]
33. Grady, W.M.; Santoso, S. Understanding power system harmonics. IEEE Power Eng. Rev. 2001, 21, 8–11.
[CrossRef]
34. Marini, A.; Latify, M.A.; Ghazizadeh, M.S.; Salemnia, A. Long-term chronological load modeling in power
system studies with energy storage systems. Appl. Energy 2015, 156, 436–448. [CrossRef]
35. Marini, A.; Latify, M.-A.; Ghazizadeh, M.-S.; Salemnia, A. Joint Maintenance Scheduling of Generation Units
and Energy Storage Systems. Tabriz J. Electr. Eng. 2017, 46, 78.
36. Farahani, H.F.; Shayanfar, H.; Ghazizadeh, M. Modeling of stochastic behavior of plug-in hybrid electric
vehicle in a reactive power market. Electr. Eng. 2014, 96, 1–13. [CrossRef]
37. IEEE Recommended Practice for Monitoring Electric Power Quality; IEEE Std 1159–2009 (Revis. IEEE Std
1159–1995); IEEE: Piscataway, NJ, USA, 2009; pp. 1–81. [CrossRef]
38. Abbas Marini, L.P.S.S.; Mortazavi, M.S. Ghazizadeh. In A linear programming approach to distribution
Power Flow. In Proceedings of the 6th International Conference on Clean Electrical Power (ICCEP), Santa
Margherita, Italy, 27–29 June 2017.
39. Barrows, C.; Bloom, A.; Ehlen, A.; Ikaheimo, J.; Jorgenson, J.; Krishnamurthy, D.; Lau, J.; McBennett, B.;
O’Connell, M.; Preston, E. The IEEE reliability test system: A proposed 2019 Update. IEEE Trans. Power Syst.
2019, 99, 1. [CrossRef]
40. Rosenthal, E. GAMS-A user’s guide. In Proceedings of the GAMS Development Corporation, Washington,
DC, USA, 1–13 May 2008.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

136
 MDPI
St. Alban-Anlage 66
4052 Basel
Switzerland
Tel. +41 61 683 77 34
Fax +41 61 302 89 18
www.mdpi.com

Energies Editorial Office

E-mail: energies@mdpi.com
www.mdpi.com/journal/energies
MDPI
St. Alban-Anlage 66
4052 Basel
Switzerland
Tel: +41 61 683 77 34
Fax: +41 61 302 89 18
www.mdpi.com ISBN 978-3-03928-435-1