Renewable Energy 172 (2021) 1087e1102

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Innovative power-sharing model for buildings and energy

communities
Gianfranco Di Lorenzo a, Sara Rotondo a, Rodolfo Araneo a, *, Giovanni Petrone b,
Luigi Martirano a
a
Department of Astronautical, Electrical and Energy Engineering, University of Rome ”La Sapienza“, Rome, Via Eudossiana, 18, 00184 Italy
b
Department of Information and Electrical Engineering and Applied Mathematics (DIEM) - University of Salerno, Salerno, Via Giovanni Paolo II 132, 84084,
Italy

a r t i c l e i n f o a b s t r a c t

Article history: The paper proposes an innovative power-sharing model, i.e., a power-system architecture for aggregation
Received 19 January 2021 of users able to share the power produced by common generators and energy services. The model is
Received in revised form suitable for both multi-tenant buildings and groups of multiple buildings and it is applicable for both
2 March 2021
existing and new buildings. It is scalable for larger systems and suitable for an easier integration with
Accepted 12 March 2021
Available online 25 March 2021
storage systems. The novel principle of the model is that the energy produced by common generators can
be shared among the end-users in a unidirectional way, so that each user remains passive towards the
distributor, except a single active user that assumes the role of balance node. This key feature allows for
Keywords:
Energy communities
easily implementing the model in all the residential and tertiary multi-units buildings in full compliance
Nearly zero energy buildings with national regulations, with the adoption of power sharing contracts as well. This paper discusses the
Power sharing model feasibility of the model through a dynamic Matlab/Simulink model, which is used to show its effec-
Power sharing contracts tiveness in several case studies. The significance of this work consists of approaching the energy sharing
Renewable energy sources in buildings with a completely new strategy, based on an innovative system architecture that can be
Demand response programs effectively implemented.
© 2021 Elsevier Ltd. All rights reserved.

1. Introduction net-zero energy is a transversal concept applicable to buildings and

Energy demand of commercial, tertiary, and residential build-
ings corresponds to a significant part of the whole energy demand.
In EU this energy is estimated to be about 40% of the total energy
consumption. Consequently, new international and national energy
and sustainable policies have introduced the use of new technol-
ogies in various sectors (e.g, internet of things [1], internet of en-
ergy [2], blockchains [3]) to promote the rational use of energy as a
new construction and operation rule [4]. The approach is based on
the smart-grid [5], where renewable storage systems and loads
work in synergy thanks to DSM techniques [6] with the ultimate
goal of reaching the target of NZEBs, i.e., buildings with a very low
energy demand (mostly covered with renewable sources) [7]. The
more widely to districts: for instance, in Ref. [8] Kneifel et al. pro-
pose a multivariate regression model for the management of net-
zero energy residential test facilities; in Ref. [9], Heendeniya et al.
make a review on multi-energy system co-planning models for the
optimal sizing and scheduling of resources in nearly zero-energy
districts. In this framework, a significant role is played by the
transition of the heat generation from gas-fired heaters/furnaces to
electrically supplied heat pumps, significantly increasing the elec-
trical load demand of the building CSs. In Ref. [10], the resis-
torecapacitor model is used to develop a thermal energy
consumption prediction model of smart buildings that are finally
modelled as flexible resources of active distribution networks.

* Corresponding author.
E-mail addresses: gianfranco.dilorenzo@uniroma1.it (G. Di Lorenzo), sara.rotondo@uniroma1.it (S. Rotondo), rodolfo.araneo@uniroma1.it (R. Araneo), gpetrone@unisa.it
(G. Petrone), luigi.martirano@uniroma1.it (L. Martirano).

https://doi.org/10.1016/j.renene.2021.03.063
0960-1481/© 2021 Elsevier Ltd. All rights reserved.
G. Di Lorenzo, S. Rotondo, R. Araneo et al.
Nomenclature
BU
CPSS
CSs
DSM
DSO
EC
IC
MPPT
MUB
NG-POD
NZEB
PLL
POC
PSL
PSM
PV
PWM
TBS
U User
VSC
Nowadays, one of the most exciting challenges is the aggrega-
tion of users in ECs [11], wich allows power and energy usage to be
optimized. The key point of this strategy is to include renewable
generation in the TBSs [12] to reduce the impact of the users in
terms of grid-power demand. The underlying principle to promote
the creation of virtuous systems is the active role that users must
play to obtain important economic advantages [13].
The present paper focuses on two important energy poles:

a MUB with energy aggregation capability;

a collection of neighboring buildings willing to share energy
resources as EC.
The majority of the buildings in Europe is composed of multiple
units. According to the final use of the units, it is possible to
distinguish between:

Multi unit residential building, where the building is formed
only by welling units.

Multi unit residential and commercial building, where the
building is formed by welling units and units for commercial
and tertiary activities.
The access to the single units can be external or internal from a
common area. The cluster of the units forms the condominium,
whose community association is responsible for the management
of the common areas of the building and CSs (e.g., lighting, central
heating, elevators, etc.). In condominiums, the CSs are supplied by a
point of connection to the DSO, which is independent of that for the
private users.
According to current European standards, each building of new
construction must respect the NZEB model. Single-unit NZEBs
require the adoption of building energy management systems,
specific DSM models, and the operation in microgrid configuration
[14]. The NZEB model applied to a MUB increases the complexity of
the whole system, if compared to the system of a single dwelling.
The application of the NZEB model requires the design and oper-
ation of the technical systems having in mind the overall building
and not just the individual units [15]. In particular, small-scale grids
should serve cluster of units, which introduces several legislative
and economic problems. A new solution to improve the small-scale
grid approach is investigated in this paper: a MUB, although legally
recognized as a group of independent units, from a technical point
1088
Renewable Energy 172 (2021) 1087e1102
Balance user
Common power-sharing system
Common services
Demand side management
Distributor system operator
Energy community
Incremental conductance
Maximum power point tracking
Multi-unit building
Natural gas point of delivery
Nearly zero-energy building
Phase-locked loop
Point of connection
Power-sharing link
Power sharing model
Photovoltaic
Pulse width modulation
Technical building system
Voltage-source converter
of view must be considered as a single large system.
Recently, the European Directive 2018/2001, also referred to as
revised Renewable Energy Directive (RED-II), introduced renewable
ECs as new forms of final users. They consist of legal entities formed
by aggregation of users, municipalities, etc., which are willing to
share the energy produced by a common renewable generating
system with the only goal of increasing the social welfare of the
community. The introduction of ECs in the actual electricity
network configuration may allow a greater exploitation of the
produced renewable energy [16], less energy drawn from the grid,
and, therefore, less costs for users. ECs act as MUBs (when the
cluster of units corresponds to individual buildings) and the con-
dominium becomes an aggregator, which has the task to manage
the energy community and the exchange of energy [17]. In both EC
and MUB cases, the members maintain their right of final users,
even if they form part of aggregation entities [18].
A huge amount of literature examines the concept of
community-based cooperative energy consumption, with partic-
ular reference to grids. In Ref. [19], Bera et al. propose to orchestrate
a community-based game among customers which considers the
dynamic nature of the composition of the community with the aim
to alleviate energy consumption cost to costumers. In Ref. [20], Fina
et al. develop a model to estimate the cost-optimal large-scale
economic potential of shared rooftop PV systems based on neigh-
bourhood ECs by allocating buildings in characteristic settlement
patterns and then in ECs by an upscaling approach. The energy
sharing concept is further extended in Ref. [21], where the author
proposes a peer-to-peer approach for the energy exchange in
buildings. Additionally, the community-cooperative concept has
been extended in the framework of a transactive energy market to
multi-interconnected microgrids [22] and to microgrid clusters
[23]. The cooperative energy consumption is strictly related to the
optimal operation of grids, in terms of management of distributed
energy resources, that is investigated in Ref. [24] with holistic
methods and with an analytical target cascading algorithm in
Ref. [25]. Finally, in Ref. [26], the authors apply the energy man-
agement system to buildings as well, through a multi-agent-based
approach.
In the well-assessed context of cooperative energy consump-
tion, following the implementation idea initially presented in
Ref. [27], we introduce a novel approach, defined as PSM, which
fosters the optimized integration of renewables and energy sour-
ces. PSM can be properly tailored both for multi-tenant buildings
and for groups of multiple buildings, and can be applied to existing
G. Di Lorenzo, S. Rotondo, R. Araneo et al.
as well as to new buildings. In addition, the PSM is scalable for
larger smart grids and allows for an easy integration of generation
and storage systems. The proposed PSM is based on an innovative
under-lying idea with respect to the aforementioned existing
literature: the energy produced by common renewable generators
can be shared among the connected end-users only in a unidirec-
tional way and each user cannot receive more than its current
power demand; in this way, each user remains passive towards the
DSO. To deliver the excess produced power to the power grid, the
balance, or slack, node is included into the system architecture and
its role is played by the building CSs, which acts as an active node;
this node is called BU. This key feature allows for an easy imple-
mentation of the PSM in all existing and new MUBs, in full
compliance with national regulations, with the support of appro-
priate power-sharing contracts. At this stage, we don’t design any
DSM program to encourage end-users to make changes in their
energy consumption profiles, but we deliberately investigate the
effectiveness of a simple PSM that shares energy democratically,
following the natural variations of the end-users’ profiles in a dy-
namic and responsive way. Of course, the proposed architecture can
host load scheduling/shifting algorithms for maximizing the on-
site photovoltaic energy consumption and users’ profits as well.
In this way, it is possible to install and integrate generation and
storage systems serving the building and connected to all the
tenants/users. The architecture of the system comprises a dc bus
backbone (defined as a PSL) that connects the common generators
to the users, which have an independent point of connection with
the DSO. The produced energy is delivered through the PSL to the
users by means of unidirectional inverters. Therefore, the whole
system acts as a system with multiple generators (one for each
user) without power exchange among users. The only active user to
the power grid is the balance node of the building CSs, equipped
with a bi-directional converter.
This paper describes the fundamental mechanisms within the
new proposed PSM and its policy impacts, to guide regulators,
decision-makers, and the research community through the key
steps involved in its design. To investigate the feasibility of the PSM,
a dynamic Matlab/Simulink model has been developed and used
with the only goal of illustrating the fundamental details for the
implementation of the novel EC model and of demonstrating its
powerful capabilities. A prototype is under development in the
electrical engineering laboratory of La Sapienza University and
experimental results will be presented in future works.
The paper is organized as follows: in Section 2, after the
description of the existing regulatory framework, the novel PSM is
presented and illustrated, for existing and new MUBs. In Section 3,
the Matlab/Simulink model is illustrated and the relevant results
are shown in Section 4. Finally, in Section 5 some economic con-
siderations are provided and in Section 6, conclusions are drawn.
2. Power-sharing model
2.1. Regulatory barriers
Nowadays, the critical aspect for the creation of MUBs and ECs is
the national regulatory that does not encourage the aggregation of
users in common power systems. First, the electricity network
national regulatory code does not include a configuration that
recognizes a system made by single generation and several con-
sumers, but only single generation and single consumer (Simple
Production and Consumption Systems). In the particular case of
MUBs, the generation system can be installed in the building, but it
can feed only the CSs, making the resulting configuration uneco-
nomical. As a consequence, in the current legislation scenario each
unit must be provided with a POC, and a system consisting of
1089
Renewable Energy 172 (2021) 1087e1102
multiple units connected to the power grid with a single POC
cannot be realized. In addition, using the existing distribution
network to share the energy produced in MUB systems may create
problems related to the payment of the general charges included in
the electric bill, which at the moment are not considered in the case
of self-consumed energy. However, single states are starting to
recognize forms of aggregation, to be in reception of the European
Directives.
A recent example comes from Italy, where Law n. 8 was
approved on February 28th, 2020, recognizing ECs at national level.
Even though a law recognizing the REDII Directive is still missing,
Law n. 8 opens opportunities to experimental cases, where energy
can be shared through the existing distribution network, paying the
electric charges both on the energy provided by the power grid and
the energy shared in the building or in the community. In any case,
since legislative scenarios are starting to include these new forms of
final users, it is important to understand how they should work and
how to optimize the energy exchange.
2.2. Novel PSM model
The proposed novel PSM is based on the capacity of the users to
consume electrical energy that is locally generated and managed by
a CPSS based on a dc bus, avoiding exchange of power among the
units themselves. In the proposed architecture, the users remain
passive towards the DSO and the role of active user is assumed only
by the BU for which the direction of the energy in the POC can be
reversed. It should be noted that no regulatory limits are broken,
making this the a preferred solution for ECs as introduced by the
EU. As discussed in Ref. [28], to connect a RES generator to the grid
it is necessary to become an active network users. This can be
achieved only by interacting with the DSO because it is necessary to
comply with several technical conditions required for accessing
distribution networks (the access should be non-discriminatory).
The main requirements concern limitation of active power injec-
ted that could generate abnormal over-frequency, reactive power
compensation, and resilience to voltage dips, by means of the so
called Low Voltage Fault Ride Through curve. In the proposed
model, the common generators and storage systems act as a mul-
tiple generator with a variable power. The energy locally generated
can flow only from the common generator to the users and never
from user to user. Therefore, for each user and at any time, the
locally generated and absorbed energy can never exceed the load.
At the POC with the distributor, the energy direction is always from
the network to the user and can never be reversed, so that the user
is always passive.
Since users have usually independent POCs to the power grid,
the PSM can easily be implemented in special power-sharing con-
tracts in full compliance with national regulatory systems. The
energy is delivered to the users through the PSL by means of sec-
ondary connection points (sharing points of connection) with a
unidirectional inverter managed by a local control. The main rule of
PSM is that the users can only consume the power/energy coming
from the CPSS and they can never deliver energy to the CPSS (Fig. 1).
The rule of control is simple: instant by instant, the power is
split among all the users and the power (possibly) not consumed by
the users is entirely delivered into the power grid through the BU.
The scheme reported in Fig. 1 explains the operation of the
system. A common photovoltaic generator installed on the roof of
the building is connected to the CPSS through a dc/dc converter and
energizes a dc bus for the power sharing with all the users. The
photovoltaic dc/dc converter is controlled by means of a MPPT al-
gorithm to optimize the PV power production in every environ-
mental condition. Each user (U1-UN) has its own POC. The user CS
for the building (i.e., lighting, elevator, parks, etc.) has an
G. Di Lorenzo, S. Rotondo, R. Araneo et al.
Fig. 1. Scheme of power-sharing model (PSM).
independent POC with the grid (POC-CS). The users are connected
to the PSL through a special dc/ac inverter (IN-Un) controlled by the
CPSS, which allows power flow only from PSL to the users. The
inverter installed on the CSs of the building (IN-UCS) has been
identified as the balance node of the system and is the only active
user for the DSO. The CPSS control system also prevents the users
(U1-UN) from delivering energy to the public electricity grid; the
units, therefore, are not active components of the system.
To allow the DSO to have access to the data needed for electric
charges, each user is equipped with an electric meter at the
inverter. If the CPSS is owned by users, they can stipulate a power-
sharing with the contributing community. If the CPSS is owned by a
third party, the users can stipulate a power purchase contract with
such party. The model, herein designed for a MUB, can be applied
contract in the presence of at least two users (one passive user and
one BU) and extended to multi-buildings (with several active BUs).
The reasons that motivate the users to adopt the proposed
1090
Renewable Energy 172 (2021) 1087e1102
scheme are several: (i) the users have the possibility to share a
common generator installed over a common building area without
the necessity to construct and exercise the own generator, that
should be constructed over a private area that often is not available
especially in existing condominium; (ii) the exploitation of the
renewable energy source is improved thanks to multiple users with
different load profiles that interact reducing the installed peak
power and the unbalance; (iii) the system architecture based on a
backbone dc link allows for the integration of any distributed
resource with high level of efficiency [29].
In the following sections, the applicability of the proposed PSM
for existing and new buildings is discussed.
2.3. MUB for existing buildings
For existing buildings, in the case of heating systems, the au-
thors suggest the installation of a heat pump generation supplied
G. Di Lorenzo, S. Rotondo, R. Araneo et al.
by electrical energy coupled with the existing gas-fired heater. The
heat pump is supplied by the building CSs, connected to the grid
trough the POC-CSs. The heating system thus becomes hybrid since
it is composed of a central thermal storage (i.e., a boiler) fed by both
the gas-fired heater and the heat pump.
The terminals of the heating systems, i.e., to traditional radia-
tors, will remain unchanged. The domestic hot water production
will be supported by electric heat pumps that substitute the
traditional gas boilers. In this configuration, cooking ranges will be
the only gas operated components. Fig. 2 shows the complete en-
ergy building architecture, where points of connection for charger
stations of electric vehicles (POC-EV1 and POC-EV2) are also
considered. The most innovative part of this proposed model is the
implementation of a PV in power-sharing mode.
The dc PSL (see Fig. 1) allows the connection of the PV generator
to several users and to the building CSs switchboard (e.g., elevators,
stairs lights, auxiliaries, etc.). The switchboard constitutes the
swing node of the system.
Each unit is supplied by the PSL through its own dc/ac inverter
characterized by a limited and unidirectional power and controlled
by a distributed management system that defines the amount of
Fig. 2. Proposed model for existing MUBs.
1091
Renewable Energy 172 (2021) 1087e1102
available power based on the actual system consumption. The PV
system is sized based on the heating production devices and the
electrical users’ consumption. When the thermal demand is low,
the power in excess generated by the PV panels is shared by the
users according to their need or injected into the power grid
through the slack node.
2.4. MUBs’ model for new buildings
The model proposed for new buildings and NZEB constructions
is based on the usage of only electrical energy, not considering any
type of gas consumption and connection with the gas connection,
as shown in Fig. 3. In the previous case, the ranges were still gas-
operated, whereas in this case they are replaced by the inductive
types. The heating system, previously composed of existing natural
gas boilers and new heat pumps, becomes fully electric with the
adoption of heat pumps. It would be also interesting the imple-
mentation of ground source heat pumps, where the relatively
constant temperature of the earth could be exploited.
In this configuration, the CSs of the building are connected to the
building main electric panel, including the charging stations of
G. Di Lorenzo, S. Rotondo, R. Araneo et al.
Fig. 3. Proposed model for new MUBs.
electrical vehicles. In addition, greater impulse is given to the
power-sharing configuration, since the PV system can be oversized
to supply the ground source heat pumps and the private units of the
building.
2.5. Control strategy
The power generated by the PV source is shared among the N
users and the in excess power is fed into the power grid through the
BU. The power PPV produced by the common generators is split, at
each tk of the sampling-time of measured variables, according to
the following rules:

the PPV power is divided among the users according to the
nominal power of the inverter Pn;i and, for each user i, the
assigned power (APi ) is compared with the actual load demand
Li ;

if the load demand Li is completely satisfied by the dc-link (APi >
¼ Li ), the power-sharing strategy is stopped for this user until a
new control action is needed; the delta power (APi  Li ) will be
at disposal of the other users;
1092
Renewable Energy 172 (2021) 1087e1102

if there is a surplus of power with coming from the satisfied
loads, this is assigned to users whose load demand Li is not yet
satisfied, always according to the nominal power of the inverters
of the users that continue to participate to the power-sharing
strategy;

the power that is not assigned to the users at the end of the
power-sharing control loop is reversed into the grid through the
BU.
The control system has been implemented in a Matlab code in a
vectorial manner and demonstrated to be computationally effec-
tive, reliable, and robust. Hence, the code doesn’t require large
computational resource and it can be easily implemented on
commercial Programmable Logic Controllers.
The strategy can be achieved with a two-levels control system:
general and local. The first level operates at the CPPS level (i.e.,
building level), while the second level works at the converter level.
As mentioned above, the maximum power coming from the PV
generator is transferred to the dc bus and constitutes the total
available power PPV at the PSL. This available power is monitored by
the CPSS with a sampling rate of the order of 1 min to perform the
control strategy.
G. Di Lorenzo, S. Rotondo, R. Araneo et al.
The controller consists of a main programmable logic controller
with inputs and outputs. In this scheme, the constraints are the
inverter nominal power of users Pn;i . The input data are:

the load demand Li of each user;

the power generated by the PV source PPV .
The outputs are:

the power assigned to each inverter APi ;

the power sent to the balance user PBU.
The flow-chart of the control system is reported in Fig. 4.
It is worth noting that, in real applications, a suitable commu-
nication protocol must be designed for exchanging data among
devices in order to achieve an efficient and reliable control strategy.
In Ref. [30], a low-cost communication system based on the IEC
61850 standard with a server-client TCP/IP protocol has been
developed for small-scale grids. It represents a valid solution for
collecting data related to the users power consumption, PV power
production, and inverters configuration.
3. MATLAB/SIMULINK model
The main purpose of the proposed PSM is to maintain the
supply-demand balance by providing the required flexibility
necessary to follow intermittent and non-dispatchable renewable-
energy resources via the implementation of energy-management
Fig. 4. Flow chart of the control strategy.
1093
Renewable Energy 172 (2021) 1087e1102
programs. In addition, such a scheme can have an active role in
mitigating issues related to financial feasibility and installation
problems (e.g., building rooftop terrace usage) of PV systems for
small residential consumers [31].
A possible implementation of the PSM shown in Fig. 1 is detailed
in this section, through a dynamic Matlab/Simulink model that is
developed to assess its feasibility and to show the real advantages
offered by the novel PSM.
The consumers are equipped with individual VSCs [32] that are
parallel connected to a constant-voltage dc bus fed by a PV source
through a centralized dc/dc converter, on which the MPPT strategy
is implemented [33]. The PSL architecture is completed with an
inverter connected to the building CSs that acts as the slack node of
the power system.
The implemented Matlab/Simulink model is shown in Fig. 5 and
consists of:

a PV system installed on the rooftop of a construction. The PV
array consists of 10 parallel strings composed of 5 Sunpower
crystalline modules connected in series with nominal power of
305 Wp;

a dc-dc step-up converter that boosts the PV array output
voltage of the dc bus at 500 V. The converter is controlled by an
incremental conductance algorithm to track the maximum po-
wer of the PV system under variable temperature or irradiation
conditions;

a dc bus connecting all the users. It is worth noting that the PSL
architecture shown in this paper can be extended to different dc
G. Di Lorenzo, S. Rotondo, R. Araneo et al.
Fig. 5. Matlab/Simulink model.
bus solutions (e.g., bipolar dc bus, Ring dc-bus) depending on
the system power ratings, the small-scale grid extension, the
safety, and the fault protection constraints [34];

four VSCs: the first operates as the balance node, while the
second, the third, and the fourth supply power to variable three-
phase loads that represent the consumers. Each consumer is
also connected to the power grid through a 10 kW three-phase
transformer. The VSCs maintain the dc bus voltage constant
thanks to a voltage external control loop. Single phase users can
also be considered together with their points of connection to
the DSO and the relevant single-phase unidirectional inverter
connected to the PSL.
Converters are controlled by a central aggregator which man-
ages the energy sharing among the nodes while the MPP tracker
maximizes the PV power output. The VSC of the i-th consumer
handles its power output PVSC;i to feed the variable load of the final
user Pl;i by minimizing the power exchange with the grid Pg;i :
Pl;i  PVSC;i ¼ Pg;i y0 : (1)
If the power needed to satisfy all the consumers is smaller than
the PV power output, the excess of power is injected into the power
grid by the balance node.
3.1. MPPT controller
The PV array efficiency is mainly affected by the efficiency of
three components: (i) PV module; (ii) inverter; (iii) MPPT algorithm.
The voltage-current characteristic of a PV array is non-linear and
1094
Renewable Energy 172 (2021) 1087e1102
has a single maximum power point (MPP) if temperature and
irradiance are the same for all modules (uniform conditions) [35].
The MPPT algorithms are essential to draw the maximum power
from the PV array. These techniques are deployed in the boost
converter by changing the duty cycle D. In this analysis the algo-
rithm used for the MPP tracker is the IC. This technique can extract
the maximum power from the PV array also in non-uniform
working conditions.
The IC algorithm is based on the condition that the power-
voltage characteristic has a zero slope at the maximum power
point. By differentiating the power P ¼ VI with respect to the
voltage, we obtain:
dP dðVIÞ dI DI
¼ ¼IþV yI þ V : (2)
dV dV dV DV
When DDVI ¼ VI the algorithm finds the MPP.
The IC algorithm is used together with an integral regulator to
minimize the error signal e and the time required to find the global
maximum. The error is given by the sum of the instantaneous
conductance I and the incremental conductance dI , i.e.,
V dV
I dI
e¼ þ : (3)
V dV
The instantaneous errors e are cumulated by the integral regu-
lator which multiplies them by the integral gains and finally adds
them to the controller output. In Ref. [36] guidelines are provided to
select the optimal parameters of the IC with integral regulator.
Moreover, a low-pass filter and a band-pass filter are added to the
G. Di Lorenzo, S. Rotondo, R. Araneo et al. Renewable Energy 172 (2021) 1087e1102

proposed MPPT algorithm in order to minimize the errors in the I Table 2

V
Main data of the VSC model.
and in the DDVI calculations, respectively [37]. The filters help to
Block Name Sub-Block Name Parameter Name Parameter Value
reduce the noise in the PV voltage and PV current measurements.
The main data of the Matlab/Simulink MPPT model are reported VSC control Voltage regulator Kp 7
in Table 1. Ki 800
Current regulator Kp 0.3
Ki 20
3.2. Voltage source converter Choke impedance R 2 mU
L 250 mH
Output filters Series connected R 1.8850 mU
The 500 V dc bus voltage is converted into a 260 Vrms ac voltage L 250 mH
through a three-level voltage source converter (VSC). The technique Parallel connected P 100 W
used to independently control active and reactive grid currents is Q 10 kvar
the vector current control [38]. The control system of voltage and
current exploits two loops: (i) the dc bus voltage is regulated to the
500 V nominal value by an external control loop and (ii) the active overall system to possible faults. The main data of the Matlab/
Id and reactive Iq current components that feed the grid are Simulink VSC model are reported in Table 2.
managed through an internal control loop.
In this study, the VSC works with a unity power factor. The 4. Results and discussion
current component Id allows for regulating the active power flow
based on the load required by the end-user, whereas the reactive To show the behavior of the proposed architecture, three con-
current component Iq is considered to be zero. A necessary feature sumers’ nodes and one balance node are considered, as shown in
of the power grid side converter control is the grid synchronization, Fig. 5.
obtained through a synchronization algorithm. This technique ex- First, we assess the behavior of the proposed control strategy for
ploits a PLL [39] that can detect the phase angle of the grid voltage the energy sharing shown in Fig. 4. For the example shown in
to obtain a unity power factor. Fig. 6(a) we consider a PV system with a delivering power output
The sample time used by voltage and current controllers and by equal to 8 kW, which feeds three consumers’ loads (User-1, User-2,
the PLL synchronization system is 100 ms, while VSCs and pulse and User-3) by means of inverters with a rated power of 7.5 kW,
generators use a sample time of 1 ms. The active and reactive power 5 kW, and 2.5 kW, respectively. We assume that the users power
flow can be separately managed by a Park transform that allows the consumption ranges between 0 and 4 kW for User-1 (black line in
synchronization with the power grid voltage through the trans- Fig. 6), while is fixed to 4 kW and 2 kW for User-2 and User-3,
formation from a stationary reference frame abc to a rotating respectively.
reference frame dq. The synchronization with the grid voltage im- Initially, when the energy demand of User-1 equals 1 kW, all the
plies that the q voltage component Vq is equal to zero, while the users energy demands are satisfied (since their summation is equal
d voltage component Vd is equal to the grid voltage peak. Before to 7 kW) and 1 kW of excess power is supplied to the slack node.
using the Park transform, the grid voltage phase needs to be esti- When the demand of User-1 increases to 2 kW, there is no excess
mated. The tracked phase angle is the output of a PLL model which power for the balance node. Next, when the demand of User-1
uses the three-phase voltages evaluated on the power grid side as further increases to 3 kW, the total demand is equal to 9 kW so
inputs. that the PV production is no longer sufficient to satisfy all the loads.
The currents Id;ref and Iq;ref represent the reference current Hence, the 8 kW power output of the PV field is divided by the PSM
components while Id and Iq are the grid current components. In the control strategy among the users by considering the following
internal current control loop, a PI-regulator is present, whose input steps:
is the difference between the reference and measured components
and whose output is the signal for the PWM.
according to the rated power of the inverters, User-1 has 4 kW
The external voltage control loop keeps the dc bus voltage 8  7:5), User-2 has 2.66 kW (i.e., 8  5), and User-3 has
(i.e., 15 15
constant. The dc bus voltage is compared with the reference bus 8  2:5);
1.33 kW (i.e., 15
voltage and the difference d is the input of a Pi-regulator:

since the assigned power to User 1 (4 kW) is greater than its
power demand (3 kW), the exceeding power (1 kW) is shared

if d > 0, Id is increased to enhance the active power flow injected
between User-2 and User-3 according to the rated power of their
into the grid;
1  5) and 0.33 kW for
inverters: 0.66 kW more for User-2 (i.e., 7:5

if d < 0, Id is reduced to decrease the active power flow injected
1  2:5);
User-3 (i.e., 7:5
into the grid.

the PV power is then assigned to the users as follows: 3 kW to
The PI-regulator output is the reference current Id;ref . In the User-1, 3.33 kW to User-2 and 1.66 kW to User-3.
proposed architecture, all the converters are equipped with a
voltage regulator that partecipates to maintain stable the dc bus With this reasoning in mind, it is straightforward to see that
voltage. This design choice allows to increase the resiliency of the when the energy demand of User-1 increases to 4 kW, the energy

Table 1
Main data of the MPPT algorithm.

Block Name Sub-Block Name Parameter Name Parameter Value

MPPT controller PWM generator fc 5 kHz

Duty cycle regulator Kp 5
R
dD ¼ Kp e þ Ki e dt Ki 1
Sampling time for MPPT TMPPT 200 ms

1095
G. Di Lorenzo, S. Rotondo, R. Araneo et al. Renewable Energy 172 (2021) 1087e1102

Fig. 6. Behavior of the control strategy in two scenarios (nominal power of inverters: User-1 7.5 kW; User-2 5 kW; User-3 2.5 kW): (a) PV output equal to 8 kW; (b) PV output equal
to 20 kW.

sharing is 4 kW to User-1, 2.66 kW to User-2, and 1.33 kW to User-3. In Fig. 6(b), we investigate more deeply the control strategy,
Finally, we can observe that when the energy demands of User-1, assuming a power output of the PV array equal to 20 kW and a more
User-2, and User-3 are all satisfied, the excess power is again sup- variable load of User-1. We observe again that all the loads de-
plied to the balance node. mands are satisfied except when the energy demand of User-1

1096
G. Di Lorenzo, S. Rotondo, R. Araneo et al.
equals 8 kW. In this case, since the load demand is greater than the
nominal power of the inverter (which is 7.5 kW), only 7.5 kW is
delivered to User-1 and the remaining power is fed to the balance
node.
In Fig. 7, we test the dynamic response of a user node in the
Matlab/Simulink model: for the sake of clarity and simplicity, only
one user is considered in addition to the balance user. In Fig. 7(a),
soft linear variations of the reference load profile are considered, in
the scenario of a PV output equal to 15.25 kWp (10 strings) and a
consumer’s inverter nominal power equal to 10 kW. After an initial
transient, the consumer inverter output follows the consumption of
the load to minimize the power exchange with the power grid.
Power is required from the grid only when the load demand is
higher than the inverter rated power, in which case it cannot
obviously satisfy the load demand. In this proposed architecture,
the balance node only intervenes to provide the excess power flow
to the grid. In Fig. 7(b), the system is stressed by sharp variations of
Fig. 7. Dynamic behavior of the Matlab/Simulink model with one consumer node and the balance node under a variable load profile with soft (a) and sharp (b) variations.
1097
Renewable Energy 172 (2021) 1087e1102
the load profile. We observe a good stability of the control system of
the inverter: the system reaches the steady-state under bounded
input applied to it and is able to get the desired response without
any intolerable variation, i.e., it is able to follow the energy demand
of the load in the range of its rated power. The power grid ex-
changes energy only during the transitions, for short time intervals
and when the energy demand of the load exceeds the rated power
of the inverter.
The system dynamic response is tested also in the presence of
irradiance and temperature variations. In particular, in Fig. 8(a), the
PV system is assumed working at 25 C; the irradiance is reduced by
steps of 250 W=m2 from 1000 W=m2 to 250 W=m2 and, finally,
increased abruptly back to 1000 W=m2 and decreased to 200
W=m2 . Consequently, the PV array output power decreases from
15.25 kW to 11.38 kW, next to 7.22 kW and then to 3.66 kW.
Correspondingly to the abrupt irradiance increasing, it rises up back
to 15.25 kW and, finally, falls down to 2.83 kW. The response of the
G. Di Lorenzo, S. Rotondo, R. Araneo et al.
Fig. 8. Time trend of the power output (black line) of the boost converter controlled by the MPPT, under variable irradiance G (a) (red line) (T ¼ 25 + C) and temperature T (b) (red
line) (G ¼ 1000 W=m2 ).
system is excellent: the MPPT is able to follow the irradiance var-
iations maintaining stability and accuracy. The effect of the MPPT is
visible in the small spikes of the PV power output and in the ringing
effect when the irradiance increases abruptly from 250 W= m2 to
1000 W=m2 . In Fig. 8(b), the MPPT is tested under temperature
variations as well; the irradiance is assumed equal to 1000 W= m2 .
Again, it is possible to observe the ability of the MPPT to maximize
the PV power output, in response to the effects of the temperature
variations. In general, it is worth noting the effectiveness of the
incremental-conductance method.
Postponing an accurate resilience assessment of the proposed
PSM to subsequent studies, we observe that using the general
resilience time-dependent quantification proposed in Ref. [40] as
equation (1) and recalled in the more recent paper [41], the average
residence of the system (intended as adaptability and recovering
ability of the system) against abrupt irradiance and temperature
variations, or even outage of an end-user inverter, is above 88%. The
main vulnerability of the PSM lays in the fact the system is unable
to cope with the boost converter or the BU inverter outage, since
they are fundamental components. It should be observed that
1098
Renewable Energy 172 (2021) 1087e1102
when the supervisor is down, all the generated energy is fed into
the grid through the BU node, and no energy is delivered to the
end-users, since the priority is to avoid that any end-user may
appear as active to the DSO. Improved self-healing schemes with
protection coordination studies will be investigated in further
research.
Finally, in Fig. 9, the behavior of the full Matlab/Simulink model
is reported with three consumers’ nodes and one balance node. In
this scenario, we assume a nominal power for the three users’ in-
verters equal to 10 kW. Under a sharing strategy, each user has an
initial available power of 5.08 kW. When a user has a lower power
demand, the excess power will always be shared among the
remaining two users. The energy demand is assumed variable for
User-1 (dashed black line) and fixed for User-2 (8 kW, dashed blue
line) and for User-3 (4 kW, dashed red line). In Fig. 9, we have also
reported the time trend of the power requested by the users and
the effective delivered power (continuous lines) calculated through
the Matlab/Simulink model (the efficiency of the VSC is 0.98%). The
main numerical results are reported in Table 3; for the sake of
clarity,figures have been rounded without considering the
G. Di Lorenzo, S. Rotondo, R. Araneo et al. Renewable Energy 172 (2021) 1087e1102

Fig. 9. Dynamic behavior of the Matlab/Simulink model controlled by the proposed democratic power sharing control strategy with three consumers’ nodes (User-1, User-2 and
User-3) plus a balance node under a variable load of User-1 (dashed line: load power demanded by the user; solid line: effective power delivered by the common generator to the
user).

Table 3 Table 5
Summary of the results in Fig. 9. Cost of PV battery energy storage taken from Ref. [43].

Time Requested power Delivered power Storage capacity Battery Battery þ inverter/charger

[s] [kW] [kW] 3 kWh 1140 $=kWh 1920 $=kWh

8 kWh 1060 $=kWh 1470 $=kWh
U-1 U-2 U-3 Total U-1 U-2 U-3 Total 18 kWh 870 $=kWh 1159 $=kWh
1 1 8 4 13 1 8 4 13
2 2 8 4 14 2 8 4 14
3 4 8 4 16 4 7.25 4 15.25
4 6 8 4 18 5.625 5.625 4 15.25 costs. The scaling up from single residential PV systems to com-
5 8 8 4 20 5.625 5.625 4 15.25
mercial PV systems reduces the End-user PV cost of more than 30%,
6 5 8 4 17 5 6.25 4 15.25
Comments: whereas more than 20% of cost saving may be obtained by scaling
1: all Users’ load demand satisfied up a battery energy storage from 3 kWh to 18 kWh, as shown in
2: all Users’ load demand satisfied Table 5. The cost of the common power-sharing system (including
3: surplus power of User-1 and User-3 given to User-2 PSL and control systems) is estimated at approximately 25% of the
4: surplus power of User-3 shared between User-1 and User-2
5: surplus power of User-3 shared between User-1 and User-2
final installation costs. These figures show how the cost savings
6: surplus power of User-1 and User-3 given to User-2 largely compensate the cost of using a more complex power-
sharing architecture, like the one proposed in this paper.
To better grasp the economic benefits behind the proposed PSM,
efficiency of the inverter stage. The results confirm the robustness we study the energy community with the three users, named User-
of the proposed architecture and the effectiveness of the control 1, -2 and -3 having the load profiles derived from measured data
strategy that shares the energy produced by the common PV array over the two years 2018 and 2019. The users live in a building with a
among the consumers,based upon their instantaneous load re- common PV field on the roof with a rated power of 22 kWp. In this
quirements and the nominal power of the installed inverters. analysis we neglect the energy costs for the common services.
In the basic scenario (denoted as BS, see inset in Fig. 11), the PV is
connected to the grid through the POC of the common services
5. Economic considerations (POC-CS) and the three users have their own separate POCs (POC-1,
POC-2, POC-3), with rated power Pn;i equal to 3 kW, 4.5 kW and
The goal of this paper is to present a novel PSM and to assess its 6 kW, respectively.
robustness and performance as a whole dynamic system. Although Without the adoption of the PSM, all the energy (EPV ) produced
out of the scope of the present paper, some preliminary economic by the PV generator would be fed into the grid through the com-
considerations seemed in order. Table 4 reports the end-user costs mon service (CS) and sold at the hourly single national price c2 with
of the most common PV plants in the U.S.. The cost accounts for all
a total revenue of RBS for the energy community:
system hardware (storage not included) and project-development

Table 4
End-user PV costs taken from Ref. [42]. The costs in the brackets include the inverter costs.

Sector Description PV size Cost [$=W]

Residential rooftop-mounting 3e10 kW 2.70 (3.11)

Commercial rooftop-mounted, ballasted racking 10 kW - 2 MW 1.83 (2.10)
Utility-Scale ground-mounted, fixed-tilt ¿ 2 MW 1.06 (1.44)

1099
G. Di Lorenzo, S. Rotondo, R. Araneo et al. Renewable Energy 172 (2021) 1087e1102

Fig. 10. Time trend of the measured load consumptions of three users and of the generation a common PV field.

energy EiBS by the providers with a fixed tariff equal to c1 ¼ 0:159

V/kWh, given by the flat energy tariff (0,069 V/kWh), metering &
transmission costs (0,00861 V/kWh), general system costs
(0,054343 V/kWh), taxes (0,0125 V/kWh), and VAT (10%). Thereby,
the total bill (BBS BS
i ), the revenue (Ri ) and the effective energy
expense (EiBS ) for the single user are given by:

8
>
> BBS BS
i ¼ c1 Ei User energy bill
>
>
>
< Pn;i BS
UserBS
i ¼ RBS
i ¼ R Revenue for selling PV energy
>
> Pn
>
>
>
:
Fig. 11. Time trend of the cumulative energy produced by the PV generator, shared EiBS ¼ BBS BS
i  Ri Effective energy expense
among the three users and fed into the grid through the balance node.
(5)
In the power sharing scenario (denoted as PSM), each user is
connected to the PV generator system through the dc-link with
8
< BS
Esold ¼ EPV Energy sold at balance node
inverter having nominal power Pn;i equal to the rated power of the
CS ¼ POCs. The POC-CS acts as balance node, i.e., POC-CS]POC-BU. The
: RBS ¼ c ðhÞE ðhÞ Revenue for selling PV energy
2 PV PSM shares energy among the users who receive trough their own
(4) inverter the amount of renewable energy Esh;i ; the in excess energy
is fed into the grid through the balance node and is sold at the
where h is the hour index over the year.The users purchase the hourly price c2 :

Table 6
Energy and economic savings with PSM.

Basic scenario Power sharing scenario Savings

User
POC Energy produced Energy purcased Bill Energy sold Revenues Balance Energy purcased Bill Energy Sold Total revenue Balance

Pn;i EPV EiBS BBS BS

Esold RBS BABS EiPSM BPSM PSM
Esold RPSM BAPSM Si S%i
i i i i i i

[kW] [MWh] [MWh] [V] [MWh] [V] [V] [MWh] [V] [MWh] [V] [V] [V] [%]

U1 3 8.57 1362 726 636 1.01 161 325 164 799 125.5
U2 4.5 14.88 2364 1089 1275 5.16 820 486 334 942 73.8
U3 6 35.08 5575 1452 4123 20.58 3270 649 2621 1501 36.4
CS/BU 22 58.64 58.64 3267 26.85 1460

1100
G. Di Lorenzo, S. Rotondo, R. Araneo et al. Renewable Energy 172 (2021) 1087e1102

8 PSM X exchange of power among the users and preventing them from
< Esold ¼ EPV  Esh;i Energy sold at balance node supplying power into the grid, thus complying with any regu-
BU ¼ i lation framework.
: PSM PSM
R ¼ c2 ðhÞEsold ðhÞ Revenue for selling PV energy
The control strategy has been deeply investigated and a Matlab/
(6)
Simulink model has been developed to assess the feasibility of the
The total bill (BPSM ), the revenue (RPSM ) and the effective energy proposed architecture. The results are encouraging for further steps
i i
expense (EiPSM ) for each user are computed as before. With resect to into this ongoing research. The next planned steps will focus on the
integration of battery energy systems and electric vehicles, to
the basic scenario the user asks for a reduced amount of energy to
implement grid-to-vehicle and vehicle-to-grid advanced services.
the provider that is paid with a reduced bill BPSM .
i Further research directions will consider supercaps to increase the
8 PSM stability of dc bus, the investigation of more advanced converter
>
> Ei ¼ EiBS  Esh;i Reduced purchased energy schemes and the possible evolution towards dc distribution end
>
>
>
>
>
> PSM
¼ c1 EiPSM
users’ networks.
>
< Bi User energy bill
UserPSM ¼ Declaration of competing interest
i >
> Pn;i PSM
>
> RPSM ¼ R Revenue for selling PV energy
>
>
i
Pn
>
>
>
: The authors declare that they have no known competing
EiPSM ¼ BPSM
i  RPSM
i Effective energy expense financial interests or personal relationships that could have
(7) appeared to influence the work reported in this paper.

To prove the benefit of the PSM architecture, the previous References

equations are used to calculate the savings for each user, computed
as the difference between the effective energy cost in the two
scenarios, i.e., Si ¼ EiBS  EiPSM and S%i ¼ ESBSi 100 for a real case.
i
[1] L.-C. Hwang, C.-S. Chen, T.-T. Ku, W.-C. Shyu, A bridge between the smart grid
and the internet of things: theoretical and practical roles of lora, Int. J. Electr.
Power Energy Syst. 113 (2019) 971e981.

The measured time trend of the three users’ loads and the PV
output are reported in Fig. 10 (black line); the load consumption of
the three users after the application of the PSM is reported as well
(red lines). In Fig. 11, we have shown the cumulative time trend of
the energies shared among the three users, the in excess energy fed
into the grid trough the balance node, and the energy produced by
the common PV generator. Table 6 summarizes the most relevant
results: the last two columns show, in absolute and percentage
values, the savings for each user, computed as the difference be-
tween the balances in the two scenarios, i.e., Si ¼ BABS PSM and
i  BAi
Si
S%i ¼ BABS
100.
i
6. Conclusions
European energy policies promote the constitution of ECs at the
level of buildings and clusters of buildings as an optimal way to-
wards a clean energy transition. In this new scenario, users play an
active and direct role. The ability to integrate common generators
into the building systems and share the produced energy
respecting national regulatory systems is a key-point of these ECs.
In this paper an innovative power-sharing model has been
proposed; it is suitable for multi-tenant/user buildings and can be
extended to group of buildings. The power-sharing model is
directly applicable for both existing and new buildings, regardless
of the adopted national regulatory system.
The power-sharing model relies on three distinguishing
features:

the system architecture is based on a common dc bus that is
used to deliver the energy produced by the generator to
different consumers, with the presence of a balance node used
to feed into the grid the in excess produced power;

the control strategy is based on an equal sharing of the available
output energy of the common generator, based on the instan-
taneous demands of the users and according to the nominal
power of the installed inverters;

the system architecture supports only a unidirectional power
flow from the common generators to the users, avoiding any
[2] F. AlFaris, A. Juaidi, F. Manzano-Agugliaro, Intelligent homes’ technologies to
optimize the energy performance for the net zero energy home, Energy Build.
153 (2017) 262e274.
[3] G. van Leeuwen, T. AlSkaif, M. Gibescu, W. van Sark, An integrated blockchain-
based energy management platform with bilateral trading for microgrid
communities, Appl. Energy 263 (2020) 114613.
[4] J. Lowitzsch, C.E. Hoicka, F.J. van Tulder, Renewable energy communities un-
der the 2019 European Clean Energy package - governance model for the
energy clusters of the future? Renew. Sustain. Energy Rev. 122 (2020) 109489.
[5] R. Pal, C. Chelmis, M. Frincu, V. Prasanna, Match for the prosumer smart grid
the algorithmics of real-time power balance, IEEE Trans. Parallel Distr. Syst. 27
(12) (2016) 3532e3546.
[6] M. Child, C. Kemfert, D. Bogdanov, C. Breyer, Flexible electricity generation,
grid exchange and storage for the transition to a 100% renewable energy
system in europe, Renew. Energy 139 (2019) 80e101.
[7] M. Manganelli, G. Greco, L. Martirano, Design of a new architecture and
simulation model for building automation toward nearly zero energy build-
ings, IEEE Trans. Ind. Appl. 55 (2019) 6999e7007.
[8] J. Kneifel, D. Webb, Predicting energy performance of a net-zero energy
building: a statistical approach, Appl. Energy 178 (2016) 468e483.
[9] C.B. Heendeniya, A. Sumper, U. Eicker, The multi-energy system co-planning
of nearly zero-energy districts - status-quo and future research potential,
Appl. Energy 267 (2020) 114953.
[10] Z. Li, S. Su, Y. Zhao, X. Jin, H. Chen, Y. Li, R. Zhang, Energy management strategy
of active distribution network with integrated distributed wind power and
smart buildings, IET Renew. Power Gener. 14 (12) (2020) 2255e2267.
[11] N. Good, P. Mancarella, Flexibility in multi-energy communities with electrical
and thermal storage: a stochastic, robust approach for multi-service demand
response, IEEE Trans. Smart Grid 10 (2019) 503e513.
[12] L. Martirano, G. Parise, G. Greco, M. Manganelli, F. Massarella, M. Cianfrini,
L. Parise, P. di Laura Frattura, E. Habib, Aggregation of users in a residential/
commercial building managed by a building energy management system
(BEMS), IEEE Trans. Ind. Appl. 55 (2019) 26e34.
[13] R. Aghamolaei, M.H. Shamsi, J. O’Donnell, Feasibility analysis of community-
based pv systems for residential districts: a comparison of on-site central-
ized and distributed pv installations, Renew. Energy 157 (2020) 793e808.
[14] M.R. Narimani, Maigha, J.-Y. Joo, M. Crow, Multi-objective dynamic economic
dispatch with demand side management of residential loads and electric
vehicles, Energies 10 (5) (2017), 1996e1073.
[15] A. Rosato, M. Panella, R. Araneo, A. Andreotti, A neural network-based pre-
diction system of distributed generation for the management of microgrids,
IEEE Trans. Ind. Appl. 55 (6) (2019) 7092e7102.
[16] A. Rosato, M. Panella, R. Araneo, A distributed algorithm for the cooperative
prediction of power production in PV plants, IEEE Trans. Energy Convers. 34
(1) (2019) 497e508.
[17] K. Bruninx, H. Pandzic, H.L. Cadre, E. Delarue, On the interaction between
aggregators, electricity markets and residential demand response providers,
IEEE Trans. Power Syst. 35 (2020) 840e853.
[18] R. Moura, M.C. Brito, Prosumer aggregation policies, country experience and
business models, Energy Pol. 132 (2019) 820e830.
[19] S. Bera, S. Misra, D. Chatterjee, C2C: community-based cooperative energy
consumption in smart grid, IEEE Trans. Smart Grid 9 (2018) 4262e4269.

1101
G. Di Lorenzo, S. Rotondo, R. Araneo et al.
[20] B. Fina, H. Auer, W. Friedl, Cost-optimal economic potential of shared rooftop
pv in energy communities: evidence from Austria, Renew. Energy 152 (2020)
217e228.
[21] H. Mehrjerdi, Peer-to-peer home energy management incorporating
hydrogen storage system and solar generating units, Renew. Energy 156
(2020) 183e192.
[22] M. Daneshvar, B. Mohammadi-Ivatloo, K. Zare, S. Asadi, A. Anvari-
Moghaddam, A novel operational model for interconnected microgrids
participation in transactive energy market: a hybrid IGDT/Stochastic
approach, IEEE Trans. Ind. Informat.
[23] M. Daneshvar, B. Mohammadi-Ivatloo, S. Asadi, A. Anvari-Moghaddam,
M. Rasouli, M. Abapour, G.B. Gharehpetian, Chance-constrained models for
transactive energy management of interconnected microgrid clusters, J. Clean.
Prod. 271 (2020) 122177.
[24] K.P. Detroja, Optimal autonomous microgrid operation: a holistic view, Appl.
Energy 173 (2016) 320e330.
[25] H. Gao, S. Xu, Y. Liu, L. Wang, Y. Xiang, J. Liu, Decentralized optimal operation
model for cooperative microgrids considering renewable energy un-
certainties, Appl. Energy 262 (2020) 114579.
[26] A. Anvari-Moghaddam, A. Rahimi-Kian, M.S. Mirian, J.M. Guerrero, A multi-
agent based energy management solution for integrated buildings and
microgrid system, Appl. Energy 203 (2017) 41e56.
[27] L. Martirano, S. Rotondo, M. Kermani, F. Massarella, R. Gravina, Power sharing
model for energy communities of buildings, IEEE Trans. Ind. Appl. 57 (1)
(2021) 170e178.
[28] L. Lo Schiavo, M. Delfanti, E. Fumagalli, V. Olivieri, Changing the regulation for
regulating the change: innovation-driven regulatory developments for smart
grids, smart metering and e-mobility in italy 57 506 e 517.
[29] K. Siraj, H.A. Khan, Dc distribution for residential power networks-a frame-
work to analyze the impact of voltage levels on energy efficiency, Energy Rep.
6 (2020) 944e951.
[30] R.A.G. Burbano, M.L.O. Gutierrez, J.A. Restrepo, F.G. Guerrero, LED design for a
small-scale microgrid using IEC 61850, IEEE Trans. Ind. Appl. 55 (6) (2019)
7113e7121.
[31] R. Araneo, U. Grasselli, S. Celozzi, Assessment of a practical model to estimate
the cell temperature of a photovoltaic module, Int. J. Energy Environ. Eng. 5
(2014) 1e15.
1102
Renewable Energy 172 (2021) 1087e1102
[32] J.I. Leon, S. Kouro, L.G. Franquelo, J. Rodriguez, B. Wu, The essential role and
the continuous evolution of modulation techniques for voltage-source in-
verters in the past, present, and future power electronics, IEEE Trans. Ind.
Electron. 63 (2016) 2688e2701.
[33] T. Kawabata, S. Higashino, Parallel operation of voltage source inverters, IEEE
Trans. Ind. Appl. 24 (2) (1988) 281e287.
[34] M. Salehi, S.A. Taher, I. Sadeghkhani, M. Shahidehpour, A poverty severity
index-based protection strategy for ring-bus low-voltage dc microgrids, IEEE
Trans. Smart Grid 10 (6) (2019) 6860e6869.
[35] G. Petrone, C.A. Ramos-Paja, G. Spagnuolo, Photovoltaic Sources Modeling,
John Wiley & Sons, Chichester, West Sussex, United Kingdom, 2017, https://
doi.org/10.1002/9781118755877, 978-1-118-67903-6.
[36] U. Chauhan, A. Rani, V. singh, B. Kumar, A modified incremental conductance
maximum power point technique for standalone pv system, in: 2020 7th
International Conference on Signal Processing and Integrated Networks
(SPIN), 2020, pp. 61e64.
[37] B. Ashok Kumar, M. Srinivasa Venkatesh, G. Mohan Muralikrishna, Optimi-
zation of photovoltaic power using pid mppt controller based on incremental
conductance algorithm, in: C. Kamalakannan, L.P. Suresh, S.S. Dash,
B.K. Panigrahi (Eds.), Power Electronics and Renewable Energy Systems,
Springer India, New Delhi, 2015, pp. 803e809.
[38] A. Egea-Alvarez, S. Fekriasl, F. Hassan, O. Gomis-Bellmunt, Advanced vector
control for voltage source converters connected to weak grids, IEEE Trans.
Power Syst. 30 (2015) 3072e3081.
[39] M. Davari, Y.A.-R.I. Mohamed, Robust vector control of a very weak-grid-
connected voltage-source converter considering the phase-locked loop dy-
namics, IEEE Power Electron. Lett. 32 (2017) 977e994.
[40] M. Ouyang, L. Duen ~ as-Osorio, Multi-dimensional hurricane resilience assess-
ment of electric power systems, Struct. Saf. 48 (2014) 15e24.
[41] L. Das, S. Munikoti, B. Natarajan, B. Srinivasan, Measuring smart grid resil-
ience: methods, challenges and opportunities, Renew. Sustain. Energy Rev.
130 (2020) 109918.
[42] R. Fu, D. Feldman, R. Margolis, U.S. Solar Photovoltaic System Cost Benchmark:
Q1 2018, National Renewable Energy Laboratory, 2019. https://www.nrel.gov/
docs/fy19osti/72399.pdf.
[43] webpage, Battery storage price index e may 2020, Solar Choice, https://www.
solarchoice.net.au/blog/battery-storage-price-index-may2020/, 2020.