Aalborg Universitet
Review on Control of DC Microgrids and Multiple Microgrid Clusters
Meng, Lexuan; Shafiee, Qobad; Ferrari-Trecate, Giancarlo; Karimi, Houshang; Fulwani,
Deepak ; Lu, Xiaonan; Guerrero, Josep M.
Published in:
I E E E Journal of Emerging and Selected Topics in Power Electronics
DOI (link to publication from Publisher):
10.1109/JESTPE.2017.2690219
Publication date:
2017
Document Version
Early version, also known as pre-print
Link to publication from Aalborg University
Citation for published version (APA):
Meng, L., Shafiee, Q., Ferrari-Trecate, G., Karimi, H., Fulwani, D., Lu, X., & Guerrero, J. M. (2017). Review on
Control of DC Microgrids and Multiple Microgrid Clusters. I E E E Journal of Emerging and Selected Topics in
Power Electronics, 5(3), 928 - 948 . https://doi.org/10.1109/JESTPE.2017.2690219
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
- Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
- You may not further distribute the material or use it for any profit-making activity or commercial gain
- You may freely distribute the URL identifying the publication in the public portal -
Take down policy
If you believe that this document breaches copyright please contact us at vbn@aub.aau.dk providing details, and we will remove access to
the work immediately and investigate your claim.
Downloaded from vbn.aau.dk on: August 09, 2022
www.microgrids.et.aau.dk
Abstract-- This paper performs an extensive review on control
schemes and architectures applied to DC microgrids. It covers
multi-layer hierarchical control schemes, coordinated control
strategies, plug-and-play operations, stability and active damping
aspects as well as nonlinear control algorithms. Islanding
detection, protection and microgrid clusters control are also
briefly summarized. All the mentioned issues are discussed with
the goal of providing control design guidelines for DC
microgrids. The future research challenges, from the authors’
point of view, are also provided in the final concluding part.
Index Terms-- Microgrid, direct current, hierarchical control,
coordinated control, plug-and-play, nonlinear control, stability.
I. INTRODUCTION
INCE 19th Century, the invention of transformers and
poly-phase AC machines initiated the worldwide
establishment of a complete AC generation, transmission and
distribution grid. DC distribution systems, although
recognized as a natural and simple solution for utilizing
electric power at the beginning, were not widely applied
because of difficulties in voltage level conversion and long
distance transmission. Since the end of last century, the
development of semiconductor based power conversion
devices offers the possibility of flexible voltage/current
transformation and thus brings DC power back to the main
stage finding its applications, for instance, in home appliances,
data centers, and vehicle power systems [1]–[3].
Most recently, the revolutionary changes in the electric
power grid, including the penetration of renewable energy
sources (RES), the distributed allocation of generation and the
increasing participation of consumers, aim to establish a more
efficient and sustainable energy system, while facing
challenges on the organization, control and management
aspects. Active and independent distribution systems, named
Lexuan Meng is with the Department of Energy Technology, Aalborg
University, 9220 Aalborg, Denmark (e-mail: lme@et.aau.dk).
Qobad Shafiee is with the Department of Electrical & Computer
Engineering, University of Kurdistan, Sanandaj, Kurdistan, Iran (e-mail:
q.shafiee@uok.ac.ir).
Giancarlo Ferrari Trecate is with the Automatic Control Laboratory, École
Polytechnique Féd érale de Lausanne (EPFL), Switzerland (e-mail:
giancarlo.ferraritrecate@epfl.ch).
Houshang Karimi is with Department of Electrical Engineering,
Polytechnique Montreal, QC H3T 1J4, Canada (e-mail:
houshang.karimi@polymtl.ca).
Deepak Fulwani is with the Department of Electrical Engineering, Indian
Institute of Technology, Jodhpur, India (e-mail: df@iitj.ac.in).
Xiaonan Lu is with the Energy Systems Division, Argonne National
Laboratory, Lemont, IL 60439 USA (e-mail: xlu@anl.gov).
Josep M. Guerrero is with the Department of Energy Technology, Aalborg
University, 9220 Aalborg East, Denmark (Tel: +45 2037 8262; Fax: +45 9815
1411; e-mail: joz@et.aau.dk).
also microgrids (MGs) [1], are thus the key to achieve those
goals, realizing the autonomous operation of each regional
power system.
Certainly, the combination of DC distribution with the MG
concept becomes attractive, since (i) being RES, electric
vehicles (EV) and energy storage systems (ESS) naturally in
DC, efficiency is enhanced because of less number of power
conversion stages; (ii) the control and management of a DC
system is much simpler than in AC, which makes DC MGs
practically more feasible; (iii) most consumer electronic
appliances are in DC, such as computers, microwave-ovens,
modern lighting systems, and so on [2]–[6].
As a consequence, an increasing number of academic
research works and industrial demonstration projects on DC
MGs have been carried out, covering applications in RES
parks [2], DC homes [7], [8], ESSs [9], [10] and EV charging
stations [11], [12]. A whole picture of future employment of
DC MGs can be obtained based on these works, while a
number of key issues are also identified, including: (i)
planning and design of a DC MG realizing an optimal
combination of generation, storage and consumption; (ii)
control and management of a DC MG achieving economic and
autonomous operation; (iii) coordination of clusters of DC
MGs with proper regulation of power and energy exchange in
regional areas; (iv) grid policy-making, which enables the
overall system operation.
The objective of this paper is to provide an extensive
review on the control and management of DC MGs, as well as
the stability perspective which is closely coupled with control
algorithm. Similar to conventional power grids, power
converters interfaced DC MGs also require a multi-layer
control scheme, from the local control of distributed
generators (DGs) to system level optimization and
management. The common definition of hierarchical control is
recalled in Section II. Section III, IV and V discuss the control
algorithms applied in primary, secondary and tertiary levels
respectively. Section VI gives a summary on the coordinated
control schemes. Plug-and-play control and operation is
discussed in Section VII. Stability aspects and active damping
design are reviewed in Section VIII. Islanding, protection and
control of MG clusters are described in Section IX and X.
Section XI closes the paper.
II. MULTI-LEVEL CONTROL SCHEME OF DC MICROGRIDS
With the development and increasing utilization of power
electronic devices, the voltage/current regulation, power flow
control and other advanced control functions can be realized in
MGs by properly operating the interfacing power converters.
As widely accepted, MGs control and management is actually
Review on Control of DC Microgrids
Lexuan Meng, Member, IEEE, Qobad Shafiee, Senior Member, IEEE, Giancarlo Ferrari Trecate,
Senior Member, IEEE, Houshang Karimi, Senior Member, IEEE, Deepak Fulwani, Member, IEEE,
Xiaonan Lu, Member IEEE, and Josep M. Guerrero, Fellow, IEEE
S
2
multi-objective task which covers different technical areas,
time scales and physical levels. The domains of interest
include the above mentioned issues, for which a multi-level
control scheme [13], [14] has been proposed and widely
accepted as a standardized solution for efficient MGs
management. It comprises three principal control levels, as
shown in Fig. 1:
Primary control performs the control of local power,
voltage and current. It normally follows the set-points
given by upper level controllers and performs control
actions over interface power converters.
Secondary control appears on top of primary control. It
deals with issues in the system level, such as power
quality regulation, MG synchronization with external
grid for smooth reconnection, DG coordination, etc.
Tertiary control is issued with optimization, management
and overall system regulations.
Based on the same hierarchy shown in Fig. 1, the way of
implementing the control levels can be centralized,
decentralized, distributed or in a hierarchical fashion, as
shown in Fig. 2. It should be noted that the structures shown in
Fig. 2 are based on the control engineering definitions
summarized, for instance, in [15]–[17]. A central control unit
exists in centralized structure which collects and transmits
information to local DGs. Decentralized and distributed
structures (Fig. 2 (b) and (c)) do not require a central
controller. Decentralized control, as defined in [15], [16],
performs regulation based on local measurements, while in
comparison, distributed control is based on both local
measurement and neighboring communication [17]. The
hierarchical control structure distributes the control functions
into local controllers and upper level controllers so that the
complete system operates in a more efficient way. The choice
of the control structure can be different according to the MG
type (residential, commercial or military), and the legal and
physical features (location, ownership, size, topology, etc.).
Centralized control [18]–[35], as shown in Fig. 2 (a),
requires data collection from all the essential MG components.
Based on the gathered information, control and management
procedures can be executed in the controller to achieve proper
and efficient operation. The advantages of centralized control
include strong observability and controllability of the whole
system, as well as straightforward implementation. However,
it entails a single point of failure issue, and the central
controller breakdown will cause the loss of all the functions.
Other disadvantages are reduced flexibility and expandability,
as well as the necessity of considerable computational
resources. Therefore, centralized control is usually more
suitable for localized and small size MGs where the
information to be gathered is limited and centralized
optimization can be realized with low communication and
computation cost [18], [29], [33], [36].
Decentralized control in MGs, as shown in Fig. 2 (b), refers
to the control methods which do not require information from
other parts of the system. The controller regulates respective
unit with only local information. Decentralized schemes have
the advantage of not requiring real-time communication, even
Tertiary Level
Secondary Level
Primary Level
Power Sharing Control
MICROGRID
(System)
Synchronization Control
Power Quality Control Power Flow Control
Coordination
Voltage and Current Control
Decision-Making
ForecastingMicrogrid Supervision Main Grid Observation
Local Supervision
Upper Level Operators
Fig. 1. Hierarchical control scheme.
Communication
Unit 1
Controller
Unit 2
System
Unit 3 Unit 1 Unit 2
System
Unit 3
C1 C2 C3
Controllers
Unit 1 Unit 2
System
Unit 3
C1 C2 C3 Controllers
Communication
Unit 1 Unit 2
System
Unit 3
C1 C2 C3
Local
Controllers
Upper Level Controller
(a) (b)
(c) (d)
Fig. 2. Basic control structures: (a) centralized; (b) decentralized; (c)
distributed; (d) hierarchical.
though the lack of coordination between local regulators limits
the possibility of achieving global coordinated behaviors.
Droop control is a typical example of decentralized control
methods. It achieves power sharing between DGs without
communication, but the accuracy is limited by system
configuration as well as control and electrical parameters.
Recent progress in communication technologies [37]
(WiFi, Zigbee, etc.) and information exchange algorithms
[38]–[42] (P2P, gossip, consensus etc.) enable the possibility
of distributed control and management in practical
applications [43]–[45]. In that sense, functions provided by
centralized control scheme can also be realized in a distributed
way as shown in Fig. 2 (c). The controllers ‘talk’ with each
other through communication lines so that essential
information is shared among each local system in order to
facilitate a coordinated behavior of all the units. The main
challenge of a fully distributed control scheme is the
3
coordination among distributed units to fulfill either the
control or optimization objectives, which necessarily require
proper communication and information exchange schemes. In
recent years there had been a major trend to integrate
distributed algorithms into the control and management of
MGs. Consensus algorithms [46]–[48], as they offer a simple
and straightforward implementation, are widely applied. The
general purpose of consensus algorithms is to allow a set of
agents to reach an agreement on a quantity of interest by
exchanging information through a communication network.
While associated information is limited to only a few
quantities in case of secondary control, tertiary control may
need to exchange a number of different signals with
neighboring agents. The consensus algorithm either fetches
essential global information [49], [50] or can be well
integrated into control layers [51], [52] to help the local
control system perceive the ‘outer’ environment.
Actually as the modern energy systems are becoming more
complex and require higher intelligence, not all the functions
can be achieved in a distributed or decentralized manner,
especially when the system involves a complicated decision-
making process. A hierarchical control structure, as shown in
Fig. 2 (d), is thus widely used. Simple functions can be
implemented in the local controllers to guarantee a basic
operation of the system. Advanced control and management
functions can be implemented in the central controller.
Hierarchical control is thus becoming a standardized
configuration in MGs. The primary control, including basic
voltage/current regulation and power sharing, is usually
implemented in local controller. The secondary and tertiary
functions are conventionally realized in a centralized manner
as they require global information from all the essential units.
III. PRIMARY CONTROL
Primary control is the first layer in the hierarchical control
scheme shown in Fig. 1. It is responsible of local voltage and
current control to meet the operation and stability
requirements. Meanwhile, decentralized load power sharing
methods are also commonly implemented in this layer to
achieve proper source and load power management.
A. Active Current Sharing
In DC MGs or DC distribution systems, multiple power
electronic converters commonly coexist as the interfaces of
DERs. Hence, it is necessary to achieve proper load sharing
among them following their current or power ratings. This is
the similar concept proposed years ago for DC-based server
system with paralleled DC/DC converters.
Master-slave control is a common approach used for active
current sharing among multiple converters [53]. In this
scheme, one converter is selected as the master unit that
operates in voltage controlled mode to establish the DC bus
voltage, while the other converters are configured as slave
converters operating in current controlled mode. Hence,
multiple slave converters operate in DC-bus-feeding mode
while the voltage is stabilized by the master converter. Since
the output signal of the DC voltage controller in the master
converter is transferred to each of the slave converters, the
current sharing among slave converters can be achieved.
In order to enhance the resilience and reliability of DC
system, circular chain control (3C) is proposed, where circular
communication architecture is employed to enhance fault
isolation and detection [54]. The reference current in each
DC/DC converter is generated based on the measured output
current of the adjacent converter. Hence, a communication
loop is established. If a fault occurs, the related converter is
disconnected to isolate the fault and a new communication
loop with the rest of the converters is reorganized to maintain
proper load current sharing. It should be noted that high
bandwidth communication network is required in these control
strategies.
B. Droop Control and Virtual Impedance
As aforementioned, most of the current sharing methods
paralleled DC/DC converters are based on high bandwidth
communication network. Accordingly, they are mostly used in
centralized DC systems with relatively small scale, e.g., DC
server system, DC electrified aircraft, etc. However, in DC
MGs, since the DERs and loads are connected to the point of
common coupling (PCC) dispersedly, it can be unsuitable or
costly to use high bandwidth communication network
considering the data reliability and investment cost. Hence,
droop control as a decentralized method has drawn increasing
attention.
Droop control was also regarded as adaptive voltage
positioning (AVP) method in analog circuit design and the
control diagram is implemented as shown in Fig. 3 [55]. The
principle of droop control is to linearly reduce the DC voltage
reference with increasing output current. By involving the
adjustable voltage deviation, which is limited within the
acceptable range, the current sharing among multiple
converters can be achieved. In most of the cases, the current
sharing accuracy is enhanced by using larger droop
coefficient. However, the voltage deviation increases
accordingly. Hence, the common design criterion is to select
the largest droop coefficient while limiting the DC voltage
deviation at the maximum load condition:
**
dci dc i oi
v v r i
(1)
*
dc dc dc dcmax
v v v v
(2)
where v
dci
*
, i
oi
and r
i
are the reference DC voltage, output
current and droop coefficient of converter #i (i = 1, 2, 3, …),
respectively, v
dc
*
is the reference DC voltage, Δv
dc
is the DC
voltage deviation and Δv
dcmax
is its maximum value.
EA
+
-
CMP
+
-
PWM
Logic
Clock
V
s
R
i
++
V
ref
I
l
V
out
I
load
Fig. 3. Analog implementation of AVP current sharing method.
It is seen from (1) that the droop coefficient r
i
can be
regarded as a resistor since it represents the relationship
4
between DC voltage and current. Therefore, this droop
coefficient r
i
is also named as virtual resistance in droop-
controlled DC MGs. The interface converter with droop
control can be modeled by using Thévenin equivalent circuit,
as shown in Fig. 4. This virtual resistance allows additional
control flexibility of DC MGs.
Source
+
-
v
dci
i
oi
+
Current loop
PWM
Generator
Voltage loop
-
+
v
dci
-
+
Virtual
Impedance
r
i
i
oi
v
dc
*
Controller
(a)
Real
Impedance
Virtual
Impedance
i
oi
+
-
v
dci
v
s
+
-
(b)
Fig. 4. General control diagram and equivalent circuit model of droop-
controlled interface converter. (a) general control diagram; (b) thévenin
equivalent circuit
C. Non-linear Droop Control
The community has also proposed different nonlinear
control techniques at different levels. One of the techniques in
decentralized control is nonlinear droop. It has been an
established fact that the linear droop technique cannot ensure
low voltage regulation and proportional current sharing [100],
[101]. To achieve acceptable voltage regulation at full load
and to ensure proportional current sharing, nonlinear and
adaptive droop techniques are proposed in [102]–[106]. A
recent review on droop control techniques is reported in [107].
The generic droop can be given by the following equation:
0
()
j
ref j j j j
V V k i i
(3)
where k
j
is a positive function, α is a positive constant, V
refj
is
reference setting and i
j
is the current supplied by the j
th
source,
respectively. For constant values of k
j
, the above
characteristics represent the linear droop. Nonlinearity in
droop characteristic ensures that droop gain is high at full load
and has a low value at light loading conditions.
Fig. 5 shows improvement in current sharing with
nonlinear droop controller when two sources are considered.
There have been some proposals where shifting of droop
characteristic is done to ensure better regulation and current
sharing [100]. In [108], [109], an optimal control framework is
proposed for DC MGs. The proposed controllers require full
state information and therefore demand proper communication
among the sources. The same paper also proposes different
variants of optimal control which require less communication
and/or no communication. It has been proposed that droop
controller is a special case of the proposed optimal control
law. The droop control computes references for different
power converters which provide an interface for sources.
Fig. 5 . Nonlinear droop control for two sources
D. DC Bus Signaling
Besides droop control, DC bus signaling is another useful
distributed method for power management among sources and
loads [67], [68]. It is implemented by measuring the DC
voltage at the local coupling point. Multiple DC voltage
ranges are pre-defined to determine the operation modes.
Particularly, when the DC voltage falls into a certain range,
the corresponding operation mode is selected. Considering the
sources that are responsible of establishing DC bus voltage,
three operation modes are commonly employed, i.e., utility
dominating mode, storage dominating mode and generation
dominating mode, as shown in Fig. 6 (a), (b) and (c). In these
operation modes, utility grids, ESSs and DGs, e.g.,
photovoltaic (PV) panels, wind turbines (WT), etc., dominate
the DC MG and are responsible of establishing DC bus
voltage, respectively. Meanwhile, different operation modes
are selected depending on local DC bus voltage level, as
shown in Fig. 6 (d).
DC Bus
+ -
Energy
Storage
PV
Power
Wind
Power
DC
Load
Grid
Inverter
+
-
Z
grid
DC Bus
+ -
Energy
Storage
PV
Power
Wind
Power
DC
Load
Grid
Inverter
Z
es
+
-
(a) (b)
DC Bus
+ -
Energy
Storage
PV
Power
Wind
Power
DC
Load
Grid
Inverter
+
-
+
-
Z
pv
Z
wind
v
high
v
low
Mode I Mode II Mode III
Utility
Dominating
Storage
Dominating
Generation
Dominating
(c) (d)
Fig. 6. Different operation modes in DC bus signaling method. (a) Utility
dominating mode; (b) ESS dominating mode; (c) generation dominating
mode; (d) operation mode selection based on local DC bus voltage.
