You are on page 1of 11

Applied Energy 228 (2018) 2050–2060

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Coordinated control of three- and single-phase inverters coexisting in low- T


voltage microgrids

Danilo I. Brandaoa, , Lucas S. de Araújoa, Tommaso Caldognettob, José A. Pomilioc
a
Graduate Program in Electrical Engineering, Federal University of Minas Gerais, Av. Antonio Carlos, 6627, Belo Horizonte, MG 31270901, Brazil
b
University of Padova, Via Gradenigo, 6/b, Padova, PD 35131, Italy
c
University of Campinas, Av. Albert Einstein, 400, Campinas, SP 13083852, Brazil

H I GH L IG H T S G R A P H I C A L A B S T R A C T

• Coordinated control of three- and ar-


bitrarily connected single-phase en-
ergy sources.
• microgrid's
Accurate power flow regulation at the
point-of-common-cou-
pling.
• Unbalance compensation by single-
and three-phase inverter-interfaced
resources.
• Active and reactive power sharing
proportional to local power avail-
abilities.
• Compliance with constraints and
limits of distributed energy resources.

A R T I C LE I N FO A B S T R A C T

Keywords: This paper proposes the third generation of the Power-Based Control, that is, an approach to effectively co-
Distributed power generation ordinate the operation of single- and three-phase inverter-interfaced distributed energy resources that can be
Hierarchical control arbitrarily connected among the phases of three-phase four-wire low-voltage microgrids. The aim of the ap-
Microgrid proach is to precisely regulate the power exchanged with the main grid by fairly exploiting the available dis-
Power quality
tributed resources while respecting their own local constraints. This allows to achieve demand-response, un-
Single-phase inverter
Unbalance compensation
balance compensation, and improved voltage profiles, which is valuable for limiting stress conditions to the
distribution infrastructure. The technique is based on a master/slave microgrid architecture where the dis-
tributed inverters act as slave units driven by a centralized master controller. This latter employs the Power-
Based Control in order to steer the contribution of the inverters on the basis of the microgrid power status. In
particular, active, reactive, and unbalance power terms are processed by the master controller and the corre-
sponding microgrid's power needs shared among the distributed energy resources, to achieve the compensation
target at the point-of-common-coupling with the main grid. The strategy proposed herein drives the three-phase
inverters to operate balanced, which avoids unwanted voltage fluctuations at the DC-side of the inverters; in-
stead, single-phase inverters, which can be connected arbitrarily among the phases of the distribution grid, are
exploited to compensate any residual unbalances (i.e., negative- and zero-components). The control technique is
analyzed and demonstrated by computer simulations considering a real urban meshed distribution network.
Steady-state and dynamic results and performances are reported and discussed considering typical demand and
generation profiles.


Corresponding author.
E-mail address: dibrandao@ufmg.br (D.I. Brandao).

https://doi.org/10.1016/j.apenergy.2018.07.082
Received 30 March 2018; Received in revised form 28 June 2018; Accepted 14 July 2018
Available online 23 July 2018
0306-2619/ © 2018 Elsevier Ltd. All rights reserved.
D.I. Brandao et al. Applied Energy 228 (2018) 2050–2060

1. Introduction technique is based on the virtual resistance. The authors of [16] pro-
pose an accurate power sharing among DERs employing one-way
The need of increasing power generation in a sustainable way has communication and transient disturbance terms. This method adjusts
brought to a significant penetration of distributed generators (DGs) the virtual output inductance of converters making the total real output
based on low-carbon technologies into the power distribution system. inductance plus the virtual one to be equal for all DERs. However, the
Government policies have stimulated the evolution of the power system control is performed sequentially with slowly varying compensation
toward small and dispersed generators, typically renewable, that are factors that produces transients in the active power contributions and
interfaced to the grid by means of electronic power converters (i.e., slow dynamics. In [17] it is presented a distributed hierarchical control
inverters) [1]. The proliferation of single- and three-phase grid-tied architecture for voltage and frequency stabilization and reactive power
inverters opens the possibility of using such devices to implement in sharing in an islanded MG. It uses a modular procedure for designing
low-voltage (LV) grids new useful services, also known as ancillary plug-and-play local controller based only on the knowledge of power
services, that go beyond the simple active power injection [2]. line parameters, and convex optimization. The authors of [6] propose a
Although many electricity distribution codes [3,4] require DGs to dispatch power scheduling MG using a central controller to plan the tie-
operate with a high power-factor, the need of power systems that are line power exchange communication between the MG and the dis-
smarter, that is, more suitable to integrate distributed generation, re- tribution system operator (DSO). The physical communication infra-
newables, and to make the most from the deployed technologies, calls structure may be based on fiber optics, and such approach performs
for additional ancillary services, typically aiming at improving the power curtailment and ensures constant power injected into the grid.
power quality for the supplied loads. New rules [5] already expect the Finally, the cited works deal only with three-phase converters and in
smart inverters to be equipped with communication means to allow most cases they need to operate unbalanced to perform compensation,
grid operators to remotely steer converters operation. Thus, the control which causes voltage fluctuations on their DC buses. Such DC voltage
strategy of the power converters should be enhanced, to also support oscillation may require retrofitting the existing three-phase inverters,
local voltage control, flat power flow to the grid [6], reactive power which are commonly designed with small capacitor. The exception is
compensation [7], current harmonic mitigation [8], unbalance com- [18] that addresses minimization of unbalance voltage in LV MG with
pensation [9–18], and a cooperative operation of the distributed energy single-phase DERs through multiobjective approach used for the sche-
resources (DERs) [19]. duling of loads and energy resources. Although, it does not discuss
In view of these changes in the legacy grid, the microgrid (MG) neither the MG structure, nor the coordination of single- and three-
paradigm has been adopted to more easily ensure a reliable and safe phase DERs.
network operation and accommodate DERs with ancillary service cap- A subject that is rarely investigated in the literature is the control of
abilities. In MGs, the sharing of active and reactive power, negative- single- and three-phase DERs coexisting in three-phase four-wire dis-
sequence, and harmonics among DERs, as well as the coordination of tribution networks. In general, single-phase converters can be con-
ancillary services, may be implemented in a centralized, decentralized, nected arbitrarily, namely, connected line-to-line or line-to-neutral.
or distributed fashion [20]. In [9] a solution using the active-power/ Thus, the following classification of DERs is therefore possible: (1)
frequency and reactive-power/voltage droop techniques attenuates the three-phase DERs that operate balanced, these contribute to balanced
negative-sequence voltage component on the basis of the DERs im- active and reactive power injections, but do not take part in unbalance
balance in sharing the load current, showing though to be highly af- compensation; (2) line-to-line DERs, connected between only two
fected by line impedances values. In [10] a strategy for voltage un- phases, that contribute to active, reactive, and negative-sequence
balance compensation by a three-phase converter is reported, but compensation, but do not take part in zero-sequence compensation; and
focusing on an individual, single inverter case. A strategy using robust (3) line-to-neutral DERs, connected between a phase and the neutral
control compensating the unbalance of local loads is considered in [11]. conductor, that contribute to active, reactive, and unbalance (i.e., ne-
In [12] it is proposed to divide the three-phase generators into three gative- and zero-sequence) compensation.
single-phase generators, so the unbalance could be properly compen- Though the usual structure of LV networks is three-phase four-wire,
sated by controlling the bus voltage magnitude, but this method is only the compresence of single-phase loads and inverters connected line-to-
applicable to local loads, due to the influence of different line im- neutral and line-to-line is not so usual around the world, but it is a
pedances. In [13] a reactive and harmonic power sharing method common practice in some regions of South America, especially in Brazil.
considering current and voltage controlled three-phase converters in a In this context, the general approach presented herein shows an im-
MG is presented, without though considering unbalanced compensation portant application value.
and without taking into account the different actual ratings of the Fig. 1(a) shows a three-phase circuit of a real urban meshed system
converters for the sharing of the compensation effort, which may with arbitrarily connected DERs and non-uniform line impedances.
overstress inverters' components during heavy load operation. In these Referring to this kind of system configuration, this paper proposes the
works communication is not necessary, which is an advantage obtained Power-Based Control (PBC) to achieve an accurate coordinated control
at the cost of inaccurate power sharing, due to the influence of un- of three-phase inverters coexisting with arbitrarily connected single-
known parameters (e.g., impedances interconnecting the sources). phase inverters to allow: (i) accurate power flow regulation at the
Considering communication-based control techniques, it is crucial point-of-common-coupling (PCC); (ii) unbalance compensation by ex-
to avoid the need of high performance (e.g., wide bandwidth) com- ploiting all sort of DERs; (iii) proportional active and reactive power
munication links, which may increase system costs and lower relia- sharing among DERs; and (iv) compliance with the local constraints of
bility. In [14] a MG with three DERs is shown to succeed in sharing the DERs.
reactive, unbalance, and harmonic demand, however it may broadcast a The PBC [21] was originally developed for controlling dispersed
low-frequency signal to DERs, and there exists a trade-off between low- inverters in a smart single-phase microgrid. This control technique
pass filter cut-off frequencies, communication bandwidth, and transient presents a low implementation complexity and minimal requirements
overcurrent stresses. In [15] a secondary control calculates the level of in terms of communication and knowledge of the controlled network as
unbalance and sends a control signal to the DERs, used to adjust their main merits. This represents an alternative to other approaches, like the
voltage references to achieve global unbalance sharing; despite the use one discussed in [22], based on optimum control, which typically make
of communication, the control is unable to completely remove the in- use of more detailed network and converters models. The first gen-
fluence of line impedances, notably, for reactive power control. More- eration of PBC (PBC-I), here referred to as phase-independent strategy,
over, in order to enhance the voltage quality in a certain node, it defines two scalar coefficients αP and αQ to share proportionally to the
heavily distorts the voltage waveform across DERs, because such DERs power availabilities the active and reactive power contributions

2051
D.I. Brandao et al. Applied Energy 228 (2018) 2050–2060

Fig. 1. Considered (a) meshed low-voltage microgrid with arbitrary single-phase DERs coexisting with three-phase DERs and loads, and (b) its control organization.

to support the entire microgrid. It can be easily extended to three-phase supervise the microgrid’s voltage at PCC. During grid-connected op-
networks, but it does not provide unbalance compensation. The second eration the UI contributes in compensating harmonic currents at PCC,
generation of PBC (PBC-II) described in [23], here referred to as phase- whereas during islanded operation the UI, acting as a voltage forming
dependent strategy, defines six scalar phase coefficients αPm and αQm, device, defines the voltage for the islanded MG [24]. The operation of
where m stands for phases a, b, and c. It shares proportionally the active the EGs and the UI is coordinated by means of a master/slave algorithm,
and reactive power among the DERs that are connected at the same m- named PBC. The MC may be embedded in the UI, while the EGs perform
th phase and provides load unbalance compensation at the PCC. as slave agents. A low bit-rate communication infrastructure is assumed
However, it drives also the three-phase inverters for unbalanced op- for information exchange among the agents. Agents can be added or
eration, which may be undesirable due to the high AC power compo- removed according to their will in participating to the microgrid con-
nent absorbed at the DC-side of the inverter. Then, this paper proposes trol and the availability of a suitable communication link. In case a
the third generation of PBC (PBC-III), named 3Φ-balanced/phase-de- resource experiences communication failures, it may be excluded from
pendent strategy, defining eight scalar coefficients αPm, αQm, αP3Φ, and the control and start working as a traditional non-dispatchable DER,
αQ3Φ. The merits of the resulting control include (i) dispatchable grid namely, according to its local needs.
power flow, (ii) proportional sharing of the active and reactive power This architecture accommodates three aspects of MG control, herein
among the DERs connected at the same m-th phase, (iii) balanced op- referred to as primary, secondary, and tertiary level. Primary level is
eration for three-phase inverters, (iv) load unbalance compensation at relevant to the behavior of individual grid-tied converters, and it is
PCC, and (v) operation complying with DER’s constraints. To the best of usually performed autonomously by each EG; it comprises the basic
the authors’ knowledge, no coordinated control with those features was control and specific functions of a DER, such as: current/voltage con-
previously presented in the literature considering coexisting arbitrary trol, synchronization, maximum power extraction, frequency/voltage
single-phase and three-phase distributed units. regulation, islanding detection, fault ride through capability. Secondary
The rest of the paper is organized as follows. Section 2 describes the level concerns the coordination of the set of resources that form the MG,
considered MG structure and control architecture. Section 3 recalls the it is performed by the EGs (i.e., slave units) on the basis of the signals
necessary unbalance compensation concepts to introduce the nomen- provided by the MC. Finally, the tertiary control concerns the interac-
clature adopted herein. Section 4 presents the new third generation of tion among the MG and the main grid (e.g., DSO), and involves the
PBC technique, whose performance is evaluated in Section 5, under interaction of all the agents. The secondary and tertiary levels are im-
dynamic time-domain simulation; and phasor-domain simulation re- plemented by the PBC, which attains power flow regulation at the
sults are adopted for comparing with the previous variants of PBC. microgrid’s PCC by fairly exploiting all the DERs that are available
Section 6 concludes the paper. within the MG, respecting their own constraints. It is underlined that
the tertiary level provides only reference commands of active and re-
2. Microgrid structure and control organization active power at the PCC and differs from other hierarchical control
strategies that use power control loops [25–26], which may be critical
The considered MG structure and its control organization is shown in terms of system stability and practical implementation, because
in Fig. 1. Two classes of components constitute the architecture: the utilities are inclined to avoid drastic changes in their up and running
Energy Gateways (EGs) and the Utility Interface (UI). The role of EGs is to central DSO, due to reliability and economic concerns.
optimally manage the energy resources (e.g., renewable sources, sto- A practical implementation should include local controller endowed
rage devices) that may be available at consumers’ premises and to in- with communication module, power converter that may be im-
terface these resources to the grid. They are of one of the three types plemented within a digital microprocessor platform, and a central
defined in the previous section and their power flow can be remotely controller that can be integrated into a supervisory control and data
adjusted on the basis of signals coming from a Master Controller (MC). acquisition (SCADA) system. Regarding centralized MGs and low-rate
The UI, instead, interfaces the whole LV MG to the utility; its role is to communication, guidelines for practical implementation are provided

2052
D.I. Brandao et al. Applied Energy 228 (2018) 2050–2060

balanced. Three-phase EGs contribute proportionally to their power


availability to active power injection and reactive power compensation,
while single-phase EGs contribute to active and reactive powers and
unbalance compensation. The harmonics are compensated by the UI
converter at the PCC.
The control is performed on the basis of the status of the EGs, which
is collected by the MC at the beginning of each control cycle, k. Let us
Fig. 2. Equivalent circuit of the system in Fig. 1(a), with the required polarity indicate with k a generic control period, with m a generic phase of the
scheme for line-to-neutral and line-to-line connected inverters. three-phase system, and with 3ϕ a three-phase quantity for three-phase
devices. The status of the j-th EG (j = 1, 2, …, J) is a set of power
quantities representing the availability of the EG in terms of exchanging
in [27] and in the IEEE Standard 2030.2 [28], communication re- power; it comprises: the actual output power [PGj (k ), QGj (k ) ] of the EG,
max min
quirements are discussed in the standard IEC 61850. This latter can be the maximum [PGj (k ) ] and the minimum [PGj (k ) ] power that can be
implemented over TCP/IP networks using the existing infrastructure. generated (e.g., if local storage is present), and the power rating
[ AGj (k ) ] of the EG’s inverter. Once the MC has gathered the required
3. Fundamentals of unbalance compensation for linear loads information, the status of the whole MG is computed in terms of the
power quantities introduced in the following. The computed quantities
Let us consider a generic MG structure with three-phase and arbi- are:
trarily connected single-phase loads and DERs, as shown in Fig. 1(a). Its
equivalent circuit can be represented as shown in Fig. 2. To properly • the total active and reactive power delivered by the three- and
indicate how resources are connected, with EGmnNj we refer to a generic single-phase EGs along cycle k:
EG connected at the j-th node (Nj) of the distribution network among J
phases m and n, namely, the phase a, b, and c, or neutral conductor. PG3ϕt (k ) = ∑ PG3ϕj (k )
According to this definition, EGbcN3, for example, indicates an EG j=1 (1.a)
connected at the 3-rd network node (N3) and between phase b and J
phase c; EGcN8 indicates an EG connected at the 8-th node (N8) between PGmt (k ) = ∑ PGmj (k )
phase c and neutral conductor. The measured quantities in an EG are j=1 (1.b)
the current through itself and the voltage across its point of connection
J
with the polarity indicated in Fig. 2. Clearly, this circuit may present
negative- and zero-sequence components as well.
QG3ϕt (k ) = ∑ QG3ϕj (k )
j=1 (1.c)
The unbalance compensation problem was originally investigated
J
by prof. Charles P. Steinmetz around 1917; prof. Steinmetz showed
that, in steady-state, a generic ungrounded unbalanced linear load can QGmt (k ) = ∑ QGmj (k )
j=1 (1.d)
be compensated by eliminating the negative-sequence components,
which can be done through appropriate reactive impedances (capaci- and, in the same way, the total minimum [PGmin min
3ϕt (k ) , PGmt (k ) ] and
tive and inductive) connected between the phases. This principle stands maximum [PGmax 3ϕt (k ) , P max
Gmt (k ) ] active power, the total maximum
on the 30° phase shift caused by the phase-to-line current transforma- [QGmax max
3ϕt (k ) , QGmt (k ) ] reactive power, and the total power rating
tion. However, for three-phase four-wire systems, the unbalance com- [ AG3ϕt (k ), AGmt (k ) ] of the three- and single-phase inverters. The max-
pensation corresponds to eliminate the negative- and zero-sequence imum available capacity for reactive power compensation is defined as:
components. Thus, under sinusoidal and symmetrical supply: (i) the
QGmax
3ϕt (k ) = AG3ϕt (k )2−PG3ϕt (k )2 , (2.a)
negative-sequence currents absorbed by three-phase loads, either con-
nected line-to-line or line-to-neutral, can be compensated by reactive max
QGmt (k ) = AGmt (k )2−PGmt (k )2 , (2.b)
elements connected line-to-line (Steinmetz theorem); clearly, the total
active power balance is zero. (ii) The zero-sequence currents absorbed
by three-phase loads connected line-to-neutral can be compensated
only by injecting a compensating current in the neutral wire. This re-
• the total capacity of active and reactive power that is available to
contribute to the MG power needs:
quires active power transfer from one phase to the other, but the total M
power balance is again zero. max
PGt (k ) = PGmax
3ϕt (k ) + ∑ max
PGmt (k ),
These well-known statements imply that in a generic three-phase m=1 (3.a)
four-wire MG the only devices capable of reducing zero-sequence M
components are the line-to-neutral DERs, and those three-phase four- max
(k ) = QGmax ∑ max
QGt 3ϕt (k ) + QGmt (k ),
wire inverters that are properly designed to operate unbalanced. The m=1 (3.b)
next Section describes the third generation of PBC able to efficiently

• from Kirchhoff’s law, the total active and reactive phase power ab-
manage this scenario.
sorbed within the MG during cycle k:
4. Third generation PBC
PLmt (k ) = PGridm (k ) + PUIm (k ) + PGmt (k ), (4.a)
A key feature of PBC is of being a model-free approach. It makes DERs
QLmt (k ) = QGridm (k ) + QUIm (k ) + QGmt (k ), (4.b)
contribute to microgrid’s power needs in proportion of their availability
of generating power under both grid-connected and islanded modes of where PGridm and QGridm are the active and reactive phase power mea-
operation, without the need of knowing any detail about the controlled sured at the grid side of the PCC, and PUIm and QUIm are the active and
system. The PBC-III described herein includes three-phase EGs coex- reactive phase power delivered by the UI.
isting with arbitrarily connected single-phase EGs (i.e., connected line-
to-line and line-to-neutral). This strategy aims at achieving proportional • the total three-phase active and reactive power absorbed within the
sharing of active and reactive power among single-phase inverters MG during cycle k:
connected at the same m-phase, while three-phase inverters operate

2053
D.I. Brandao et al. Applied Energy 228 (2018) 2050–2060

M
Table 1
PLt (k ) = PG3ϕt (k ) + ∑ PLmt (k ), 3Φ and phase scaling coefficients – implemented in the master controller.
m=1 (5.a)
Power condition Coefficients
M
QLt (k ) = QG3ϕt (k ) + ∑ QLmt (k ). PG∗3ϕt (k + 1) < PGmin
3ϕt (k )
αP3ϕ = −1
m=1 (5.b) ∗ min αPm = −1
PGmt (k + 1) < PGmt (k )

Note that the distribution power loss through the power line im- PGmin ∗ max
3ϕt (k ) ⩽ PG3ϕt (k + 1) ⩽ PG3ϕt (k )
PG 3ϕt (k + 1)
αP3ϕ = max (l)
PGt
pedances is considered in PLmt(k) and PLt(k), as well as any other in- ∗ (k + 1)
min ∗ max PGmt
verter not participating in the PBC. The errors introduced by measuring PGmt (k ) ⩽ PGmt (k + 1) ⩽ PGmt (k ) αPm = max (k )
PGmt
instruments are compensated by the integrative action created by the PG∗3ϕt (k + 1) > PGmax αP3ϕ = 1
3ϕt (k )
feedback of the power quantities sent through the communication link ∗ max αPm = 1
PGmt (k + 1) > PGmt (k )
to the MC. Besides, the system stability is guaranteed as long as the QG∗3ϕt (k + 1) ⩽ QGmax ∗
3ϕt (k )
QG 3ϕt (k + 1)
processing of the PBC algorithm in the MC is slower than the inherent αQ3ϕ = max (k )
QGt
delay of communication infrastructure, condition that can be easily ∗
QGmt max
(k + 1) ⩽ QGmt (k ) ∗ (k + 1)
QGmt
αQm = max (k )
achieved. QGmt

• the references for the total three-phase active [P ∗


G3ϕt (k + 1) ] and
Table 2
reactive [QG∗3ϕt (k + 1) ] power to be provided by the three-phase EGs
in the next control cycle k + 1: EG’s power references – implemented in each local controller.

M Scaling coefficients EG’s power references


PG∗3ϕt (k + 1) = PLt (k )− ∑ ∗
PPCCm (k + 1),
(6.a) αP = −1 ∗
PGj min
(k + 1) = PGj
m=1
−1 < αP < 1 ∗ max
PGj (k + 1) = αP ·PGj
M ∗ max
αP = 1 PGj (k + 1) = PGj
QG∗3ϕt (k + 1) = QLt (k )− ∑ ∗
QPCCm (k + 1), −1 ⩽ αQ ⩽ 1 ∗ max
QGj (k + 1) = αQ·QGj
m=1 (6.b)

• the references for the total active [P∗Gmt (k + 1)] and reactive max
QGj (k ) = AGj (k )2−PGj (k )2 . (9)
[Q∗Gmt (k + 1)] phase power to be provided by the single-phase EGs
in the next control cycle k + 1: In Table 2, P∗Gj (k + 1) and Q∗Gj (k + 1) are the active and reactive
∗ ∗
power references, respectively, for EGj in the next control cycle. The
PGmt (k + 1) = PLmt (k )−PPCCm (k + 1), (7.a) sign of the α coefficients indicates an operation of delivering (if posi-
∗ ∗ tive) or storing (if negative) active power (αP), or the generation of
QGmt (k + 1) = QLmt (k )−QPCCm (k + 1), (7.b)
capacitive or inductive reactive power (αQ). Besides, Table 2 holds for
where P∗PCCm
(k + 1) and Q∗PCCm
(k + 1) are, respectively, the active and the control of the power flow of DERs, in both the grid-connected and
reactive references of the phase power flow at the PCC. Once these islanded operation. In this way, the inverters prioritize active power

quantities are set, PG∗3ϕt , QG∗3ϕt , PGmt ∗
, and QGmt can be computed. These over reactive power injection, and three-phase inverters operate ba-
references are selected by the MC to regulate the power flow at the PCC lanced, contributing with balanced active and reactive power injection
among the different phases of the MG, according to the energy state of only. The line-to-line single-phase DERs contribute to the microgrid’s
the UI and the EGs. Note that they are estimated, for the next control active power needs and compensate reactive power and negative-se-
cycle, on the basis of the quantities measured during the last control quence currents; finally, the line-to-neutral DERs contribute to active
cycle. and reactive power, and negative- and zero-sequence compensation.
Considering the polarities in Fig. 1(a), the exchanged powers at the
terminals of the UI are: 5. Simulations results
∗ ∗
PUIm (k + 1) = PPCCm (k + 1)−PGridm (k + 1), (8.a)
To evaluate the MG dynamics and stability when the PBC-III is
QUIm (k + 1) = ∗
QPCCm (k + ∗
1)−QGridm (k + 1). applied to a meshed LV network, the electrical circuit of Fig. 1(a) was
(8.b)
firstly implemented in PSIM software, which allows to accurately re-
Grid power references PGridm and QGridm guarantee the balanced produce also fast microgrid’s transients. Secondly, the circuit of
condition at the grid side, and they are set on the basis of long-term Fig. 1(a) was implemented in Matlab/Simulink using phasor simula-
energy management strategies or set to zero in islanded mode. tion, in order to analyze the effectiveness of the control strategy during
a period of 24 h and compare the obtained results with the other gen-
• finally, the 3Φ scaling coefficients α P 3ϕ and αQ3ϕ , and the phase scaling erations of PBC. The power line impedance values and the load power
coefficients αPm and αQm (all ranging in the interval [−1, 1]) are values are listed in Tables 3 and 4, respectively. The LV microgrid is
computed and broadcasted to all the EGs. αP3ϕ and αQ3ϕ are sent to connected to the medium voltage feeder through a delta-wye trans-
three-phase devices, while αPm and αQm are broadcasted to the former (N0-N1), whose parameters are reported in Table 5.
single-phase devices connected to the corresponding phase (the For what concerns the structure of the considered LV microgrid in
same applies for both the connections: line-to-neutral, EGmNj, and Fig. 1(a), it consists of twenty-six nodes, twenty-three distinct loads, six
line-to-line, EGmnNj). The active power is controlled by the variable single-phase EGs–of which three are connected line-to-neutral (EGaN15,
αP, while the reactive power is controlled by the variable αQ. These EGbN18, EGcN9) and three are connected line-to-line (EGabN21, EGbcN3,
coefficients calculated for different operating conditions are re- EGcaN19), –and one three-phase EG (EG3ϕN31) that must operate ba-
ported in Table 1. lanced due to the small capacitor at its DC side. All the EGs are con-
trolled as current sources synchronized with the fundamental grid
Finally, given αP and αQ, the j-th EG (EGj) defines its local active and voltage. The power parameters of all the EGs are listed in Table 6.
reactive power injection according to Table 2. The maximum deliver- The following results concern (i) the islanded and grid-connected
able reactive power of the j-th EG is defined as: modes considering several typical disturbances; (ii) unbalance

2054
D.I. Brandao et al. Applied Energy 228 (2018) 2050–2060

Table 3 Table 6
Power line impedance values. Parameters of the distributed EGs.
Branch Length [m] R [Ω] L [μH] Parameter EGj (N15, N21, N9, N19, N18, N3, N31)

N1-N2 35.85 0.032 31.10 Power rating [kVA] (5.0, 9.0, 7.0, 10.0, 6.0, 5.0, 20.0)
N2-N4 23.03 0.0206 19.98 Max. power capacity [kW] (4.5, 8.0, 1.5, 8.5, 5.0, 2.5, 15.0)
N4-N3 6.68 0.0037 5.46 Min. power capacity [kW] (0, 0, 0, 0, 0, 0, 0)
N4-N8 14.95 0.0084 12.21 Power capacity [kW] (4.5, 8.0, 1.5, 8.5, 5.0, 2.5, 15.0)
N8-N11 30.67 0.0172 25.06
N11-N17 33.64 0.0188 27.48
N17-N20 22.34 0.0125 18.25
described PBC generation strategies in terms of unbalance compensa-
N20-N18 22.95 0.0129 18.75
N18-N16 33.62 0.0188 27.47
tion performance and distribution power loss.
N20-N21 12.96 0.0073 10.59
N21-N23 36.77 0.0206 30.04
N23-N28 39.42 0.0221 32.21 5.1. Short time-span simulation
N28-N29 9.95 0.0056 8.13
N29-N31 39.61 0.0354 34.36
N28-N25 22.61 0.0202 19.61
To evaluate the PBC under several conditions, a simulation com-
N25-N19 31.93 0.0285 27.70 prising different operating modes and disturbances is considered. For
N19-N14 33.62 0.03 29.16 these short time-span simulations, let us consider constant EGs power
N14-N12 14.12 0.0126 12.25 generation, as specified in the bottom row of Table 6. The results are
N12-N13 13.76 0.0123 11.94
shown in Figs. 3 and 4.
N13-N15 38.64 0.0345 33.52
N12-N10 6.67 0.006 5.79 The simulation starts with a meshed network structure (i.e., S1 is
N12-N9 26.16 0.0234 22.69 closed) and the islanded operation mode (i.e., CB2 is open). At 0.53 s,
N9-N6 42.31 0.0378 36.70 the grid voltage is restored (i.e., CB1 is closed) and the MG is properly
N6-N5 31.17 0.0278 27.04 connected to the mains at 0.59 s (i.e., CB2 is closed). At this last in-
N5-N4 0.16 0.0001 0.14
terval, most of the αP coefficients are saturated, which indicates that the
EGs are injecting their maximum power, while their surplus power
capacity is being used to distributedly compensate the reactive power
Table 4
Load active and reactive power values. (see αQ in Fig. 4). Note from Fig. 4 that both the αP and αQ coefficients
are updated once per fundamental cycle, resulting in a slow transmis-
bar Pa [W] Qa [VAr] Pb [W] Qb [VAr] Pc [W] Qc [VAr]
sion rate of 16.67 ms, which can be easily accommodated by modern
N1 560.6 237.6 280.3 118.8 280.3 118.8 telecommunication means.
N2 0 0 93.4 39.6 0 0 In grid-connected mode the microgrid’s power needs can be satisfied
N3 0 0 1027.8 435.6 1027.8 435.6 by exploiting connection with the utility; in this situation the EGs may
N4 560.6 237.6 280.3 118.8 280.3 118.8
cease to operate providing their maximum power, and they can provide
N8 186.9 79.2 747.5 316.8 747.5 316.8
N11 1401.6 594 747.5 316.8 467.2 198
reactive and unbalance compensation as ancillary services. Observe in
N17 186.9 79.2 467.2 198.0 373.8 158.4 Fig. 3 that the grid (iGm) and UI (iUIm) currents are balanced and in-
N20 0 0 1121.3 475.2 280.3 118.8 phase with the PCC voltages, indicating a full compensation of the
N18 747.5 316.8 934.4 396 280.3 118.8 negative-sequence and of the reactive power terms, in steady-state
N16 93.4 39.6 373.8 158.4 373.8 158.4
operation. The achieved zero-sequence compensation can be appre-
N21 280.3 118.8 373.8 158.4 186.9 79.2
N23 2055.7 871.2 1401.6 594 1495.1 633.6 ciated by also noticing the reduction in the UI neutral current (black
N28 0 0 0 0 0 0 curve), which is an effect of the contributions from the line-to-neutral
N29 841 356.4 841 356.4 373.8 158.4 single-phase EGs.
N31 186.9 79.2 560.6 237.6 467.2 198.0 Finally, at 0.82 s, CB1 is suddenly open and the MG transits to the
N25 654.1 277.2 560.6 237.6 560.6 237.6
N19 560.64 237.6 934.4 396 1308.2 554.4
stand-alone mode again. The observed voltage sag transition across the
N14 373.8 158.4 280.3 118.8 841 356.4 disconnection corresponds to the islanding detection interval. Despite
N12 0 0 0 0 0 0 of that, the MG operates properly with the UI defining the voltage
N13 280.3 118.8 373.8 158.4 654.1 277.2 magnitude and frequency for all the distributed units. At 0.94 s, the MG
N15 560.6 237.6 280.3 118.8 93.4 39.6
changes to radial topology (i.e., S1 is open). Observe that, differently
N10 467.2 198 467.2 198 467.2 198.0
N9 467.2 198 280.3 118.8 467.2 198.0 from the meshed case, the coefficient αPa saturates in the radial con-
N6 467.2 198 467.2 198 280.3 118.8 figuration of the network, due to the increased voltage profile at the
N5 0 0 373.8 158.4 373.8 158.4 microgrid’s nodes.
To show the proportional sharing of active and reactive power
among the three- and single-phase inverters arbitrarily connected to the
Table 5 LV MG, Fig. 5 shows the converter’s output power during the grid-
Parameters of the step-down transformer. connected operation from 0.7 to 0.8 s. The proportional power con-
S = 75 kVA tribution is among the single-phase EGs (1Φ EG) connected at the same
VH = 13.8 kV VL = 0.22 kV m-phase, and the three-phase EGs (3Φ EG) in comparison to the sum of
RH = 0.0076 p.u. RL = 0.0076 p.u.
all single-phase ones (∑EG1Φ). Then, the pairs of EGs connected at the
LH = 0.0157 p.u. LL = 0.0157 p.u.
Rm = 227.27 p.u., Lm = 32.59 p.u.
same m-phase, N15 and N21 (phase a), N9 and N19 (phase c), N18 and N3
(phase b), and the ∑EG1Φ and EG3ΦN31 are grouped and analyzed.
Fig. 5(a) quantifies the active power contribution in proportion to their
compensation considering coexisting three- and single-phase EGs arbi- maximum power capacity (Pmax). Recall that each EG includes energy
trarily connected among phases; (iii) transition modes dynamics; (iv) storage, therefore, when the total power produced by the primary
grid power flow control; (v) overall voltage profile enhancement source cannot be delivered at its output to the grid, the overproduced
throughout the microgrid’s nodes; and (vi) comparison of the three power can be stored locally. Whereas, Fig. 5(b) analyzes the reactive
power injection of EGs that is proportional to their actual maximum

2055
D.I. Brandao et al. Applied Energy 228 (2018) 2050–2060

Fig. 3. Simulation results of microgrid comprising different operating modes. From top to bottom: grid and PCC voltages, UI and grid currents.

power.

5.2. Long time-span simulation

To evaluate the effectiveness of PBC and to compare its three gen-


erations along a period lasting a full day, a long time-span simulation is
reported considering the meshed microgrid configuration. The profiles
of PV power generation and active and reactive load power obey the
typical shape represented in Fig. 6, which shows that the corresponding
load and EG power values change according to their power capacity
(see Tables 4 and 6, respectively).
Figs. 7–12 show the results comparing the MG without the PBC, and
then considering the three generations of PBC. Figs. 7 and 8 correlate
the scaling coefficients of the corresponding PBC generation with the
operation of the three-phase inverters. Figs. 7, 9, and 10 correlate the
Fig. 4. Scaling coefficients related to Fig. 3. scaling coefficients of the corresponding PBC generation with the
ability to perform unbalance compensation. Fig. 11 shows the voltage
profile throughout the MG considering the PCC (i.e., N2) and the vol-
deliverable reactive power (Qmax) calculated by (9). Observe that the
tage across the distributed units (i.e., N15, N21, N9, N19, N18, N3, N31).
three-phase EG operates balanced and contributes in proportion to the
Finally, Fig. 12 quantifies the voltage unbalance and the distribution
sum of all single-phase EGs (∑EG1Φ), for both the active and reactive
power loss.

Fig. 5. Proportional output power of EGs related to Fig. 3 in steady state [0.7–0.8 s].

2056
D.I. Brandao et al. Applied Energy 228 (2018) 2050–2060

connected at the same m-phase to contribute proportionally with their


power capacity. By issuing the coefficients αP3Φ and αQ3Φ to the three-
phase converters, they operate balanced, as shown in Fig. 8; by issuing
the coefficients αPm and αQm to the single-phase inverters, unbalanced
compensation is achieved. The balanced operation of the three-phase
inverters, without loss of performance in terms of power quality, is
achieved at the expense of performing the unbalance compensation
exploiting only the single-phase inverters; nevertheless, three-phase
inverters properly designed for this application can contribute to un-
balance compensation as done in the PBC-II.
Fig. 11 shows the voltage profile throughout the MG demonstrating
the power quality improvement as a consequence of a proper co-
ordinated control of distributed units in terms of power flow control
and distributed compensation. Such contribution is noticeable during
heavy power production by EGs and regular load demand, in which the
scaling coefficients are not-saturated (interval 9–13 h) and then they
can perform ancillary services and simultaneously comply with the
power flow requested at the grid (P∗Gridm and Q∗Gridm). When the αP
Fig. 6. From top to bottom: typical PV power generation profile, and typical coefficients are saturated (e.g., due to physical limitations) which
active and reactive load demand profile. means heavy MG load demand and tight EGs power capacity, the vol-
tage profile enhancement is lessened. Note that for all the PBC gen-
The MG with no PBC corresponds to the case where all the EGs erations, the grid power flow is properly regulated, the three-phase
inject their maximum power capacity, and do not perform reactive power (i.e., P3Φ = Pa + Pb + Pc) has to be considered for the 1st gen-
power control. It is equivalent of considering always αP = 1 and αQ = 0 eration. The reactive power is fully compensated by the EGs since the
that results in a non-dispatchable MG. The PBC-I based on αP and αQ αQ coefficients are not saturated, as visible in Fig. 7.
drives all the EGs to contribute proportionally to their corresponding Finally, Fig. 12 quantifies the voltage unbalance (kd) calculated by
power capacity, with the three-phase inverters operating balanced. On (10) and the distribution loss calculated by (11). These equations were
the other hand, it does not perform unbalance compensation as can be applied to the long time-span simulation and their values are updated
seen by noticing the unbalanced power flow through the PCC in Fig. 9 once a control cycle period. Note that these equations are not part of
and, consequently, the imbalanced voltages at the PCC in Fig. 10. PBC strategies, used herein for the sake of comparison.
The PBC-II based on phase scaling coefficients αPm and αQm drives all max min
3·(V pcc −V pcc )
the EGs connected at the same m-phase to contribute proportionally to kd (k ) = ·100[%]
their power capacity. Then, applying αPm and αQm to a three-phase Vpcc _ a + Vpcc _ b + Vpcc _ c (10)
converter, it is driven as three single-phase inverters, which produces
N
an unbalanced operation for these systems (see Fig. 8). The unbalanced
ploss (k ) = ∑ Rn ·In2
operation of three-phase converters may lead to intolerably high DC n=1 (11)
voltage fluctuations at the DC-side, an issue that can be possibly over-
max min
come by retrofitting them for MG applications. Besides, the 2nd gen- where V pcc and V pcc are the maximum and minimum values of abc
eration of PBC shows good performance in terms of unbalance com- voltages at the PCC, and Rn is the n-th network’s power line resistance
pensation, as shown in Figs. 9–12. while In is the current flowing through Rn.
The PBC-III based on 3Φ and phase scaling coefficients αP3Φ, αQ3Φ, From Fig. 12, in terms of voltage unbalance, the strategies are
αPm and αQm drives all the three-phase EGs and single-phase EGs ranked as: PBC-I, no-PBC, PBC-III and PBC-II, such that PBC-I presents
the worst unbalance factor while the PBC-II the best one. The PBC-III is

P 1st generation PBC Q


1 1

0.5 0.5

0 0
Pa, Pb, Pc 2 generation PBC
nd
Qa, Qb, Qc
1 1

0.5 0.5

0 0
Pa, Pb, Pc, P3 3rd generation PBC Qa, Qb, Qc, Q3
1 1

0.5 0.5

0 0
0 6 12 18 24 0 6 12 18 24
Time [h] Time [h]
Fig. 7. The scaling coefficients of the three generations of PBC.

2057
D.I. Brandao et al. Applied Energy 228 (2018) 2050–2060

PG3 [kW] 1st generation PBC QG3 [kvar]


3 3
2 2
1 1
0 0
PG3 [kW] 2nd generation PBC QG3 [kvar]
3 3
2 2
1 1
0 0
PG3 [kW] 3rd generation PBC QG3 [kvar]
3 3
2 2
1 1
0 0
0 6 12 18 24 0 6 12 18 24
Time [h] Time [h]
Fig. 8. Output power of the three-phase converter under the three generations of PBC.

inferior to the PBC-II because in that the unbalance compensation is kind of PBC control proposed herein features an effective exploitation
performed only by the single-phase inverters. In terms of distribution of distributed three-phase inverters and of single-phase inverters arbi-
loss, all PBC generations shows quite similar results, all of them are trarily connected to a generic three-phase four-wires distribution net-
superior to the no-PBC because this does not compensate neither re- work. In particular, in the PCB-III (i) three-phase inverters operate
active power nor unbalances. Thus, the PBC contributes to reduce the balanced, contributing to the active and reactive power needs of the
distribution power loss. network where they are installed, (ii) single-phase inverters connected
between phases contribute to active and reactive power needs and
negative-sequence compensation, (iii) single-phase inverters connected
6. Conclusions between a phase and the neutral conductor contribute to active and
reactive power needs, and negative and zero-sequence compensation,
This paper proposes the third generation of the Power-Based Control without impairing voltage quality. This assignment of the compensation
(i.e., PCB-III), which is a model-free control approach to accurately effort among the available sources is convenient for the three phase
regulate the power that a microgrid exchanges with the upstream grid, converters to not experience severe DC-side power fluctuations, en-
by sharing the control effort among distributed energy resources in a hancing, in this way, reliability and reducing hardware costs.
way that is proportional to the local power availabilities. The specific

Ppcc_a, Ppcc_b, Ppcc_c [kW] no-PBC


20
15 P*Grid not dispatchable
10
5
0
-5
Ppcc_a, Ppcc_b, Ppcc_c [kW] PBC-I
20
15 P*Grid3 =15kW P*Grid3 =7.5kW
10
5
0
Ppcc_a, Ppcc_b, Ppcc_c [kW] PBC-II
20
15 P*Gridm=5kW P*Gridm=2.5kW
10
5
0
Ppcc_a, Ppcc_b, Ppcc_c [kW] PBC-III
20
15 P*Gridm=5kW P*Gridm=2.5kW
10
5
0
0 6 12 18 24
Fig. 9. Values of the PCC active power under the three generations of PBC.

2058
D.I. Brandao et al. Applied Energy 228 (2018) 2050–2060

Vpcc_a, Vpcc_b, Vpcc_c [V] no-PBC


180
178
176 P*Grid and Q*Grid not dispatchable
174
Vpcc_a, Vpcc_b, Vpcc_c [V] PBC-I
180
178 P*Grid3 =15kW P Grid3 =7.5kW
*

176 Q Grid=0var
*

Vpcc_a, Vpcc_b, Vpcc_c [V] PBC-II


180
178 P*Gridm=5kW P*Gridm=2.5kW
176 Q*Grid=0var
Vpcc_a, Vpcc_b, Vpcc_c [V] PBC-III
180
178 P*Gridm=5kW P*Gridm=2.5kW
176 Q*Grid=0var
0 6 12 18 24
Time [h]
Fig. 10. Magnitudes of the PCC voltages under the three generations of PBC.

The proposed approach is compared with the outer two previous performance in terms of unbalance compensation and consequently of
generations. The strategies PCB-I, PCB-II, and PCB-III are herein eval- voltage profile enhancement, but it would require three-phase inverters
uated and compared in simulation by referring to a distribution net- with capacitors of generous size at the DC-side, to properly absorb the
work with realistic topology and parameters. The obtained results show power fluctuations due to the unbalance operation; the 3rd generation
that the proposed control scheme is effective and stable under several of the PBC is a proposal that give a solution to this last issue, and
conditions, including the islanded and the grid-connected modes and showed to be capable of managing the coordination of arbitrarily
the transitions between these two. With respect to the other genera- connected single-phase distributed energy resources coexisting with
tions, the 1st generation of PBC (i.e., PCB-I) does not perform unbalance three-phase inverters operating balanced.
compensation; the 2nd generation (i.e., PCB-II) shown the best

190
180
no-PBC

170
160
150
Vpcc_a, VEGs_a, [V] Vpcc_b, VEGs_b, [V] Vpcc_c, VEGs_c, [V]
190
180
PBC-I

170
160
150
Vpcc_a, VEGs_a, [V] Vpcc_b, VEGs_b, [V] Vpcc_c, VEGs_c, [V]
190
180
PBC-II

170
160
150
Vpcc_a, VEGs_a, [V] Vpcc_b, VEGs_b, [V] Vpcc_c, VEGs_c, [V]
190
180
PBC-III

170
160
150
0 6 12 18 24 0 6 12 18 24 0 6 12 18 24
T i m e [h ] Time [h] T im e [h ]
Fig. 11. Magnitudes of the voltages across distributed units under the three generations of PBC.

2059
D.I. Brandao et al. Applied Energy 228 (2018) 2050–2060

0.7
0.6
0.5

kd [%]
0.4
0.3
0.2
0.1
0
5
4 no PBC
ploss [kW]

3
PBC-I
PBC-II
2 PBC-III
1
0
0 6 12 18 24
Time [h]
Fig. 12. Voltage unbalance factor and distribution power loss of no-PBC and the three generations of PBC.

Acknowledgement current and voltage controlled DG interface inverters for reactive power sharing and
harmonics compensation in islanded microgrids. IEEE Trans Power Electron
2018;PP(99):1–12.
This research was financially supported by CAPES, CNPq (grant [14] Prodanović M, Green TC. High-quality power generation through distributed con-
420850/2016-3 and 302257/2015-2) and FAPESP (grant 2016/08645- trol of a power park microgrid. IEEE Trans Ind Electron 2006;53(5):1471–82.
9). [15] Savaghebi M, Jalilian A, Vasquez JC, Guerrero JM. Secondary control scheme for
voltage unbalance compensation in an islanded droop-controlled microgrid. IEEE
Trans Smart Grid 2012;3(2):797–807.
References [16] He J, Li YW, Blaabjerg F. An enhanced islanding microgrid reactive power, im-
balance power, and harmonic power sharing scheme. IEEE Trans Power Electron
2015;30(6):3389–401.
[1] Kroposki B, et al. Achieving a 100% renewable grid: operating electric power sys-
[17] Riverso S, Tucci M, Vasquez JC, Guerrero JM, Ferrari-Trecate G. Stabilizing plug-
tems with extremely high levels of variable renewable energy. IEEE Power Energ
and-play regulators and secondary coordinated control for AC islanded microgrids
Mag 2017;15(2):61–73.
with bus-connected topology. Appl Energy 2018;210:914–24.
[2] Yang Y, Enjeti P, Blaabjerg F, Wang H. Wide-scale adoption of photovoltaic energy:
[18] Carpinelli G, Mottola F, Proto D, Varilone P. Minimizing unbalances in low-voltage
grid code modifications are explored in the distribution grid. IEEE Ind Appl Mag
microgrids: optimal scheduling of distributed resources. Appl Energy April
2015;21(5):21–31.
2017;191:170–82.
[3] Distributed Micro and Minigeneration connection under electric power compensa-
[19] Han H, Hou X, Yang J, Wu J, Su M, Guerrero JM. Review of power sharing control
tion system, GED-15303, Jan. 2016, in Portuguese.
strategies for islanding operation of ac microgrids. IEEE Trans Smart Grid
[4] Reference technical rules for the connection of active and passive users to the LV
2016;7(1):200–15.
electrical utilities, Italian Standard CEI 0-21, July 2016, in Italian.
[20] Stadler M, et al. Value streams in microgrids: a literature review. Appl Energy Jan.
[5] Pacific Gas and Electric Company, Electric Rule No. 21, Generating Facility
2016;162:980–9.
Interconnections, Section H.2.i, 2015.
[21] Caldognetto T, Buso S, Tenti P, Brandao DI. Power-based control of low-voltage
[6] Zhao-xia X, Grurrero JM, Shuang J, Sera D, Chaltz E, Vásquez JC. Flat tie-line power
microgrids. IEEE J Emerg Selec Top Power Electron 2015;3(4):1056–66.
scheduling control of grid-connected hybrid microgrids. Appl Energy Jan.
[22] Caldognetto T, Tenti P. On microgrid evolution to local area energy network (E-
2018;210:786–99.
LAN). IEEE Trans Smart Grid 2017;PP(99). 1-1.
[7] Yang Y, Wang H, Blaabjerg F. Reactive power injection strategies for single-phase
[23] Brandao DI, Caldognetto T, Marafão FP, Simões MG, Pomilio JA, Tenti P.
photovoltaic systems considering grid requirements. IEEE Trans Ind Appl
Centralized control of distributed single-phase inverters arbitrarily connected to
2014;50(6):4065–76.
three-phase four-wire microgrids. IEEE Trans Smart Grid 2017;8(1):437–46.
[8] Sreekumar P, Khadkikar V. A new virtual harmonic impedance scheme for harmonic
[24] Serban I. A control strategy for microgrids: seamless transfer based on a leading
power sharing in an islanded microgrid. IEEE Trans Power Del 2016;31(3):936–45.
inverter with supercapacitor energy storage system. Appl Energy
[9] Cheng PT, Chen CA, Lee TL, Kuo SY. A cooperative imbalance compensation method
2018;221:490–507.
for distributed-generation interface converters. IEEE Trans Ind Appl
[25] Guerrero JM, Vasquez JC, Matas J, Vicuna LGd, Castilla M. Hierarchical control of
2009;45(2):805–15.
droop-controlled AC and DC microgrids—a general approach toward standardiza-
[10] Liu X, Deng Y, Liu Q, He X, Tao Y. Voltage unbalance and harmonics compensation
tion. IEEE Trans Ind Electron 2011;58(1):158–72.
for islanded microgrid inverters. IET Power Electron 2014;7(5):1055–63.
[26] Savaghebi M, Jalilian A, Vasquez JC, Guerrero JM. Secondary control for voltage
[11] Rezaei MM, Soltani J. A robust control strategy for a grid-connected multi-bus
quality enhancement in microgrids. IEEE Trans Smart Grid 2012;3(4):1893–902.
microgrid under unbalanced load conditions. Int J Electr Power Energy Syst
[27] Chuang A, McGranaghan M, Grady M. Master controller requirements specification
2015;71:68–76.
for perfect power systems. Galvin Electricity Initiative 2007.
[12] Kamel RM. New inverter control for balancing standalone micro-grid phase vol-
[28] IEEE approved draft guide for the interoperability of energy storage systems in-
tages: a review on MG power quality improvement. Renew Sustain Energy Rev
tegrated with the electric power infrastructure. IEEE P2030.2/D9.0, Dec. 2014.
2016;63:520–32.
[13] Mousavi SYM, Jalilian A, Savaghebi M, Guerrero JM. Autonomous control of

2060

You might also like