Journals & Magazines >IEEE Access >Volume: 8

Stability and Control Aspects of Microgrid Architectures–A Comprehensive Review

Stability and control aspect of hybrid AC/DC microgrid; hierarchical control architecture, load dynamics, influence of feeder characteristics, coordinated control strateg...

Abstract:

Self-governing small regions of power systems, known as “microgrids”, are enabling the integration of small-scale renewable energy sources (RESs) while improving the reli...Show More

Metadata

Abstract:

Self-governing small regions of power systems, known as “microgrids”, are enabling the integration of small-scale renewable energy sources (RESs) while improving the reliability and energy efficiency of the electricity network. Microgrids can be primarily classified into three types based on their voltage characteristics and system architecture; 1) AC microgrids, 2) DC microgrids, and 3) Hybrid AC/DC microgrids. This paper presents a comprehensive review of stability, control, power management and fault ride-through (FRT) strategies for the AC, DC, and hybrid AC/DC microgrids. This paper also classifies microgrids in terms of their intended application and summarises the operation requirements stipulated in standards (e.g., IEEE Std. 1547-2018). The control strategies for each microgrid architecture are reviewed in terms of their operating principle and performance. In terms of the hybrid AC/DC microgrids, specific control aspects, such as mode transition and coordinated control between multiple interlinking converters (ILCs) and energy storage system (ESS) are analysed. A case study is also presented on the dynamic performance of a hybrid AC/DC microgrid under different control strategies and dynamic loads. Hybrid AC/DC microgrids shown to have more advantages in terms of economy and efficiency compared with the other microgrid architectures. This review shows that hierarchical control schemes, such as primary, secondary, and tertiary control are very popular among all three microgrid types. It is shown that the hybrid AC/DC microgrids require more complex control strategies for power management and control compared to AC or DC microgrids due to their dependency on the ILC controls and the operation mode of the hybrid AC/DC microgrid. Case study illustrated the significant effects of microgrid feeder characteristics on the dynamic performance of the hybrid AC/DC microgrid. It is also revealed that any transient conditions either in the AC or DC microgrids could pro...

Stability and control aspect of hybrid AC/DC microgrid; hierarchical control architecture, load dynamics, influence of feeder characteristics, coordinated control strateg...

Published in: IEEE Access ( Volume: 8)

Page(s): 144730 - 144766

Date of Publication: 07 August 2020

Electronic ISSN: 2169-3536

DOI: 10.1109/ACCESS.2020.3014977

Contents

CCBY - IEEE is not the copyright holder of this material. Please follow the instructions via https://creativecommons.org/licenses/by/4.0/ to obtain full-text articles and stipulations in the API documentation.

Nomenclature

AbbreviationExpansion

Acronyms:
CC:	Central control
CCVSI:	Current-controlled voltage source inverter
CPL:	Constant power load
DBS:	Data bus signaling
DCL:	Digital communication link
DER:	Distributed energy resources
DFIG:	Doubly-fed induction generator
DSTATCOM:	Distribution static-synchronous compensator
EMS:	Energy management system
ESS:	Energy storage system
ILC:	Inter-linking converter
IM:	Induction motor
LC:	Local control
LFO:	Low-frequency oscillation
LVRT:	Low voltage ride-through
MGCC:	Microgrid central controller
MPPT:	Maximum power point tracking
PCC:	Point of common coupling
PEC:	Power electronic converter
PF:	Power factor
PI:	Proportional-integral
PLL:	Phase-locked loop
PLS:	Power line signaling
PMSG:	Permanent magnet synchronous generator
PV:	Photovoltaic
RES:	Renewable energy sources
SoC:	State of charge
VCVSI:	Voltage-controlled voltage source inverter
VI:	Virtual impedance
VPD:	Voltage active power
VQD:	Frequency reactive power
VR:	Virtual resistor
VSC:	Voltage source converter

SECTION I.

Introduction

Power electronic converter (PEC) interfaced distributed energy resources (DERs), such as solar photovoltaics (PVs), wind generators, micro-turbines and energy storage systems (ESSs) have paved the way for self-sustainable entity called microgrids [1]–[4]. AC microgrid was the main microgrid architecture for powering the remote-area power supply systems and distribution networks [4], [5]. Subsequently, the DC microgrid concept was emerged due to the fact that majority of the DERs generate DC power and increased utilisation of modern DC loads in recent years. The major advantages of the DC microgrid are efficiency, no reactive power requirement, and elimination of the need for AC-DC or DC-AC conversion stages [6]–[9]. However, since the existing distribution networks are operated at AC, and AC based DERs and loads are still prominent in power networks, the AC microgrid are still dominant. Also AC microgrid can be directly connected to the existing distribution networks without or with minimum modifications. Consequently, by merging a DC microgrid with an AC microgrid through a bidirectional interlinking converter (ILC), it has enabled to extract the benefits of both microgrids, and this architecture has been popular in the recent decade [10]–[12]. The concept of merging DC and AC microgrids through a bi-directional single or multiple parallel operating ILCs is known as the hybrid AC/DC microgrid, where both AC or DC type DERs, loads can easily be integrated into microgrids and ensures smooth power transfer between two sub-grids.

The hybrid AC/DC microgrid can be operated either in grid connected mode or in autonomous mode while the power transfer between them is controlled by the ILC. The hybrid AC/DC microgrid autonomous operation mode is far more critical compared to the grid-connected operation mode. The key challenge in autonomous operation mode is generation-demand management [13]. Furthermore, the intermittent behavior of DERs make this issue more severe. The ILC in grid-connected mode are responsible for DC microgrid voltage support and power transfer between the hybrid AC/DC microgrid and the distribution network. Extensive research studies have been carried out in the last decade on hybrid AC/DC microgrid power-sharing. In AC microgrid power-sharing realization, active power-frequency and reactive power-frequency droop characteristics are commonly employed while in DC microgrid current-voltage and active power-voltage droop characteristics are commonly employed. Therefore, in hybrid AC/DC microgrids, control strategies of AC and DC sub-grids should be coordinated with the ILC, which ensure power quality and economic benefit to its users.

The microgrid control techniques can be primarily classified into three types, 1) primary control, 2) secondary control, and 3) tertiary control. The primary control maintains stable voltage/frequency and it does not require communication links as it is operating locally. The primary control is also known as local control (LC). The voltage and frequency deviation after the LC actions are compensated by the secondary controller. The secondary control is responsible for power quality and energy management of microgrids. Centralized control and decentralized control are the two sub-division of secondary control. In the decentralized type of secondary control, all DERs are working in parallel to balance energy management using the locally obtained information from the neighboring DERs. In centralized control, there is a central controller, where optimal control decision is achieved by obtaining relevant information from the microgrid network and loads. The tertiary control operates after the secondary control and it helps to improve the power quality by energy management between multiple microgrids and the power grid. The tertiary control is implemented by coordination between microgrids and the power grids which ensures technical and economic benefits to its users. These control aspects are imperative for any type of microgrid architecture, hence it is important to shed lights on these control schemes in the context of microgrid architectures.

This paper presents a critical review of power management and control studies in AC, DC, and hybrid AC/DC microgrids, with special emphasis on primary control, secondary control, hierarchical control, and coordination control between ILC and ESS. This paper is structured as follows: Section II presents the definition and classification of microgrids, Section III discusses IEEE standards for interconnection of DERs, Section IV delineates the microgrid research focus areas, Section V presents the control and stability aspects of AC microgrids, control techniques and stability studies of DC microgrids are discussed in Section VI, control and stability aspects of hybrid AC/DC microgrids are presented in Section VII, in Section VIII fault ride-through of microgrids are reviewed, in Section IX hybrid AC/DC microgrid dynamic performance under different control strategies are analysed, and finally, conclusions of the review are summarized in Section X.

SECTION II.

Microgrid Definition and Classification

Various definitions have been proposed for microgrids over the past decade by researchers and various institutions considering the characteristics associated with different microgrid applications [14]–[17]. For example, in [14] microgrid is defined as “the concept of roaming DERs and various loads in the existing power system, such as solar-PV, wind turbines, micro-turbines, and storage devices which can be operated either in grid-connected mode or in stand-alone mode”. Researchers in [15] defined the microgrid as “ a small-scale power system that consists of DERs, controllers, and load”. The department of energy of the United States defined the microgrid as “a group of interconnected DERs and loads with precisely described electrical boundary that acts as a single controllable body w.r.t. the power grid and it can be operated either in grid connected mode or islanded mode” [16], [17]. Essentially, all these definitions have recognised the pivotal role of DERs in microgrids and microgrid’s capability to operate as a self-sustained entity independent from the main grid.

A. Microgrid Classification

Microgrids can be primarily classified into three categories based on the system architecture and voltage characteristics, 1) AC microgrid, 2) DC microgrid, and 3) Hybrid AC/DC microgrid [18]–[20]. In addition, microgrids can be classified based on the application area, such as, 1) Utility microgrids, 2) Institutional microgrids (e.g., consortium for electric reliability technology solutions (CERTS) microgrid), 3) Commercial and industrial microgrids, 4) Transportation microgrids, and 5) Remote-area microgrids (e.g., King Island Microgrid) [21], [22]. The microgrid classification is illustrated in Fig. 1.

FIGURE 1.

Classification of microgrids.

1) AC Microgrids

The AC microgrid is the most conventional microgrid type. Different forms of DERs such as wind-turbine generators, micro-turbines, solar-PV, and fuel cells are integrated into the power networks via PECs [23]. As the conventional power networks are operating on AC, AC microgrid offers minimum modifications to integrate to the existing utility grid. AC microgrids are connected to the medium voltage and low voltage distribution networks which could enhance power flow in distribution networks and decrease transmission line power losses. But, they introduce new issues, such as system stability, power quality, synchronization of DERs and reactive power shortage, these issues could be resolved by applying the advanced control techniques [24], [25].

2) DC Microgrids

Owing to technological developments in PECs, a large number of DC loads and power converters have been used for different types of applications [26]. Moreover, DC-based DERs and different kinds of ESSs create a new opportunity for DC microgrids. On average, around 30% of the generated AC power passes through a PEC before it is utilized [27]. Limited number of power conversion stages and no reactive current circulation are the main advantages of DC microgrids [28], [29].

3) Hybrid AC/DC Microgrids

The AC and DC microgrids are combined to form the AC/DC hybrid microgrids; hence, they will offer the benefits of both microgrids such as increased reliability, efficiency and economic operation. The hybrid AC/DC microgrid is facilitating direct integration of AC and DC-based DERs, ESSs, and loads with the existing distribution system [30], [31]. DC appliances are being used in a wide range of domestic and commercial applications, and they require AC to DC conversion. Hybrid AC/DC microgrids reduce the number of required power converters and therefore lower the power losses. Moreover, the amount of power losses in the inverter is small compared to the rectifier [32], [33]. However, the network structure of hybrid AC/DC microgrids is complex. Therefore, the coordinated control of individual AC and DC microgrids, effect of intermittent nature of DERs, reactive power compensation, and ILC control strategies must be thoroughly investigated.

SECTION III.

IEEE Standards for Interconnection of DERs

Microgrids are mostly covered by the distribution network and DER standards. High penetration of DERs in the low voltage or medium voltage distribution networks reduce the dependency and control capability of centralized synchronous generators, which drastically change the operating characteristics of the power network. Smaller capacity DER systems (known as microgrids) connected to the utility grid pose no special concern to the conventional power system engineers. This small capacity microgrids are treated as negative loads in the conventional power system while the system voltage and frequency are regulated by the synchronous generator [34]. However, integration of large number of microgrids to the utility grid is a matter of concern for power system operators to maintain system stability and reliability [35]. IEEE published IEEE Std. 1547–2018 (revision of IEEE Std 1547-2003) for DERs interconnection and interoperability with existing power systems [36]–[38]. IEEE Std.1547-2018 defines two main categories of DERs (e.g., Category A and Category B) based on performance requirements, such as voltage and frequency regulation, three sub-categories of DERs (e.g., Category I, Category II and Category III) based on disturbance ride-through (low-voltage ride-through). Since a microgrid can be considered as a group of DERs, the requirements stipulated in the IEEE Std. 1547–2018 also applicable to microgrids.

Also, this standard helps to effectively integrate DERs into the existing electric power system and discussed seven different operation aspects, such as 1) reactive power capability, 2) voltage and reactive power control, 3) abnormal operating performance, 4) voltage disturbance ride-through requirements, 5) frequency tripping requirements, 6) islanding and protection aspects, and 7) power quality aspects. Fig. 2 shows a DER interconnected to the existing electric power system which describes the scope of the standard. Except the power quality and protection requirements, other aspects are succinctly discussed in the subsequent sub-sections [37].

FIGURE 2.

Illustration of a DER connected to the utility grid as per IEEE Std. 1547-2018.

A. Reactive Power Capability of the DER

The DER should have the capability of either injecting or absorbing reactive power for its output active power levels equal to or higher than the steady-state active power capability. Fig. 3 illustrates the reactive power injection and absorption capability of both the category A and B DERs respectively. When the DER active power output ( $P_{r}$ ) operating range is greater than 5% and less than 20% of the rated value, it can dispatch reactive power ( $S_{r}$ ) up to the minimum value as shown in Fig. 3. Similarly, for the DER with the active power output operating range greater than 20% of the rated value, it can absorb or inject reactive power up to the maximum capability as specified in Fig. 3. The DERs should be capable of injecting/absorbing reactive power while it is dispatching active power up to the maximum capability limit [37].

FIGURE 3.

Reactive power capability curve of the DER, (a) Category A, (b) Category B [37].

B. Voltage and Reactive Power Control

The DER shall participate in voltage regulation by changing the reactive power dispatch. Despite the DER’s participation in voltage regulation, it should not exceed the minimum reactive power capability limits specified in the previous subsection. The DER’s voltage and reactive power control functions are classified into four modes, 1) constant power factor (PF), 2) voltage-reactive power, 3) active power-reactive power, and 4) constant reactive power [37]. These modes are described into the following sub-subsections.

1) Constant Power Factor Mode

In this mode, the DER is operating at a constant PF where the power system operator has specified the PF requirements, and it should not exceed the reactive power capability limits specified in the previous subsection. The power system operator can adjust PF settings either remotely or locally and the DER is allowed to maintain constant PF for up to 10 s.

2) Voltage-Reactive Power Mode

In this mode, the DER controls the reactive power dispatch as a function of voltage (a piece-wise linear function of voltage-reactive power characteristics). This mode voltage-reactive power characteristic is illustrated in Fig. 4, and the voltage-reactive power parameters are configured by the default values given in Table 1 if not specified by the power system. The power system operator can adjust voltage-reactive power characteristics either locally or remotely and the voltage reference ( $V_{ref}$ ) is autonomously adjusted by the DER. Therefore, the voltage-reactive power characteristics need to be changed based on the $V_{ref}$ . The DER can autonomously adjust its reactive power capability to a lower value. Moreover, the DER can reduce its active power dispatch to meet this requirement; however, improper settings of those parameters may cause instability.

TABLE 1 Voltage-Reactive Power Parameters [37]

FIGURE 4.

Voltage-reactive power characteristics of category A and B DERs [37].

3) Active Power-Reactive Power Mode

In this mode, the DER controls the reactive power dispatch as a function of active power output (a piece-wise linear function of active-reactive power characteristics). This mode active-reactive power characteristic is illustrated in Fig. 5, and active-reactive power parameters are configured by the default values. The left-hand side of this characteristic curve corresponding to DER absorbing active power, whereas right-hand side corresponding to DER injecting active power. The power system operator can adjust active-reactive power characteristics either locally or remotely and the response time of the DER should not exceed 10 s [37].

FIGURE 5.

Active-reactive power characteristics of category A and B DERs.

4) Constant Reactive Power Mode

In this mode, the DER dispatches a fixed amount of reactive power, where the power system operator has specified the required reactive power requirements. It should not exceed the minimum capability limits specified in the previous subsection. The power system operator can adjust reactive power settings either locally or remotely and the DER is allowed to maintain constant reactive power for up to 10 s.

C. Performance Under Abnormal Operating Conditions

Many abnormal operating conditions may arise on the power system, where the DER is connected, and it is mandatory to respond appropriately. The DER’s response contributes to the stability of the power system and can avoid possible damages to the DER itself. When the DER’s actual output voltage is greater than the over-voltage threshold or less than the under-voltage threshold, it should terminate to energize the power system and must trip within the predefined clearing time. The under-voltage and under-voltage tripping thresholds and the clearing times must be specified over a range of allowable settings given in Table 2 for category I, II, and III DERs respectively. The power system operator can adjust voltage and clearing time set-points either locally or remotely. The power system operator specifies the maximum clearing times and the over-voltage/under-voltage thresholds within a range of allowable settings [37].

TABLE 2 Abnormal Operating Performance of the DER [37]

D. Voltage Disturbance Ride-Through Requirements

This subsection describes the DER performance during voltage disturbance. The voltage disturbance ride-through requirements are not applicable if the frequency is outside the specified ride-through range. The DER should have the capability to provide specified voltage disturbance ride-through while the DER is within the capability limits. The voltage disturbance ride-through thresholds and the clearing times must be specified over a range of allowable settings given in Table 3 for category I, II, and III DERs respectively. The DER current exchange cessation with power system is not more than the maximum specified time and with no delay. It does not necessarily imply DERs disconnection, isolation, or a trip but may include momentary cessation or trip. When the voltage disturbance is within the mandatory operation region, the DER functionalities are to, 1) maintain synchronism with power system, 2) continue to exchange current with the power system, and 3) neither trip nor cease to energize. In the permissive operation region, the DER maintains synchronism with the power system and it will not trip. Moreover, when the voltage disturbance is within the continuous operation region of any duration, it will not cause the DER to cease to energize or trip from the power system and the DER will continue to deliver power and remain in operation [37].

TABLE 3 LVRT Capability Requirements [37]

E. Frequency Tripping Requirements

This subsection exemplifies the DER frequency tripping requirements during disturbance. The frequency ride-through requirements such as under frequency thresholds, over frequency thresholds and critical clearing time during abnormal conditions for category I, II, and III DERs are illustrated in Table 4. It is considered that the nominal system frequency is 60 Hz. The power system operator specifies the maximum clearing times and the over frequency/under frequency thresholds within a range of allowable settings. For the two under frequency trip functions (under frequency-1 and under frequency-2), under frequency-2 is used as the limiting requirement. Similarly, for the two over frequency trip functions (over frequency-1 and over frequency-2), over frequency-2 is used as the limiting requirement. Therefore, there are three distinct operating regions in the frequency tripping requirements.

$f > 62.0$ ; No ride-through requirements
$61.2 < f\leq 61.8$ ; Mandatory operation
$58.8 \leq f\leq 61.2$ ; Continuous operation
$57.0 \leq f\leq 58.8$ ; Mandatory operation
$f < 57.0$ ; No ride-through requirements

TABLE 4 Frequency Tripping Requirements [37]

F. Islanding Requirements

A portion of an area power system is solely energized by one or more power systems through the PCC, while a portion of the area power system is electrically separated from the rest of the power system and DER is connected. When the DERs solely energize a portion of the power system, and it is electrically separated from the rest of the power system, the DER energizing the island is called “islanding”. Islanding is categorized into two types, (1) unintentional islanding, (2) intentional islanding. Since microgrids are designed to operate either in grid-connected or islanding modes, the requirements set in the IEEE Std. 1547–2018 are also valid for microgrids.

1) Unintentional Islanding

For unintentional islanding, the DER energizes a portion of the power system through the PCC, the DER detects the island (cease to energize the power system), and subsequently DER shall trip within 2.0 s. The power system operator and the DER operator may mutually be extended the clearing time from 2.0 s to up to 5.0 s. DER ride-through requirements may be terminated by the unintentional islanding specifications detection.

2) Intentional Islanding

An intentional island may include any portion of the area power system is known as intentional power system island. For intentional islanding, the DER must be designed and operated in coordination with the power system operator. Intentional islanding is categorized into two types, (1) scheduled islanding, (2) unscheduled islanding.

a: Scheduled Intentional Island

Scheduled intentional island is formed through the power system operator or the DER operator dispatch means (such as automatic generation control action or energy management system) or mutual action. Scheduled intentional island ensures enhanced reliability, economic output power dispatch decisions, and preemptive power system operator action to island ahead of abnormal weather conditions.

b: Unscheduled Intentional Island

Unscheduled intentional island is autonomously formed following detection of abnormal conditions at the power system, and the relay will initiate the control action to isolate the intentional island from the power system. Unscheduled intentional island may switch to intentional island and disconnected from the power system if any of the following conditions are satisfied:

Violation of any of the voltage disturbance ride-through and frequency disturbance ride-through requirements.
If unintentional islanding conditions are met.
Violation of any of the trip conditions for category I, II, and III DERs are met.

SECTION IV.

Microgrid Research - Focus Areas

Microgrids consist of various intermittent renewable resources and the weather conditions determine their output power; the key challenge is to ensure stable operating conditions. To cope up with the output power fluctuations of renewable resources, ESS is implemented in the microgrid, and it needs to be adjusted instantly [39], [40]. The control systems should be designed considering all these issues to make sure that the microgrids are operating in stable condition. The key research areas are exemplified in the subsequent sub-sections.

A. Voltage and Frequency Stability

A microgrid is considered as an inverter dominated small-scale power grid, where DERs and ESSs are connected by PECs called inverters and controlled via hierarchical schemes, with the ability to operate either in autonomous mode or grid-connected mode. The voltage and frequency deviations in microgrids arise due to the mismatch between the generation and the load demand. The grid forming inverters should maintain voltage and frequency within permissible limits. To maintain stable voltage and frequency in autonomous mode is more challenging than in grid-connected mode due to the absence of inertia and reactive power support.

B. Distributed Energy Storage

The intermittent behavior of DERs results in variable output power which leads to voltage instability and potential costs in microgrid which can be solved by deploying an ESS. The major challenge for the ESS is to obtain a dynamic balance between generation and load demand while maintaining voltage and frequency within acceptable range. The ESS controller manages the power balance based on the dispatchable energy of each storage system [41]–[44]. The effective operation of microgrid depends on the optimum management of ESS, which remains as a key challenge for microgrids [45], [46].

C. Reactive Power Compensation

One of the most important issues for AC microgrids is reactive power compensation, which is directly related to the voltage instability and power quality problems [47]. The reactive power demand is mainly originating from the induction machine loads connected to the microgrid. The power quality in a microgrid can be improved by maintaining generation and load balance, while keeping microgrid voltage within acceptable limits. If the reactive power flow through the microgrid is effectively controlled and compensated, power quality problems can be solved. Moreover, the reactive power issue in microgrid during grid forming mode is even more critical due to the absence of the power grid [48]–[50].

D. Distributed Coordinated Control

Aforementioned, concept of hybrid AC/DC microgrid is gaining enormous popularity in the power industry for integrating various AC and DC based DERs, ESSs and loads with higher energy efficiency and better compatibility [51]–[53]. The ILC is an effective means to connect both the AC and the DC microgrids to form hybrid AC/DC microgrids. The ILC facilitates the transfer of real power into DC sub-grid from AC sub-grids and regulates DC bus voltage, while ILC transfers real and reactive power into AC sub-grid from DC sub-grid and regulates voltage and frequency. Voltage source converter (VSC) based DERs operate as a static generator [54]. Moreover, to transfer a large amount of power and to avoid overstressing a single ILC, researchers proposed multiple ILCs between the AC and the DC sub-grids. Furthermore, both the AC and the DC sub-grids are equipped with storage systems due to the intermittent behavior of renewable sources. Therefore, a coordinated control strategy is required for the safe and stable operation of the entire microgrid. The control can be centralized or distributed with or without communication links [52], [55], [56].

E. Harmonics Mitigation

Renewable sources such as solar-PV, wind, micro-turbines, or even fuel-cell are connected to the microgrid via VSC. The high frequency switching ripple introduces harmonics, which affects the power quality as well as dynamic stability margin of the microgrid [57], [58]. The LCL filter is used to interconnect an inverter to the microgrid to eliminate harmonics and improve power quality. The high performance and low cost are the main reasons for widespread use of LCL filters in the microgrid. The LCL filter performance is influenced by line impedance which may lead to resonance and instability issues [59]–[61].

F. Protection Issues in Microgrids

A sophisticated protection system is required for a microgrid because of the bi-directional power flow between microgrid and main power grid. In addition, the microgrid protection system should resolve grid forming faults and grid following faults [62], [63]. The former protection scheme should isolate the smallest part of microgrid and the latter isolate microgrid from the utility grid. Moreover, mode of operation and nature of fault determine the amount of fault current and direction of flow of fault current [64]. The following features make the microgrid protection schemes distinct from the conventional protection systems [65], [66]

Fault detection failure in the grid forming mode
Tripping of adjacent feeders due to fault current contribution of DER
False operation of sectionalizers due to the presence of voltage from the distribution lines from the DERs
Overrating of equipment due to higher fault current level, and
Auto reclosing failure (when a fault occurs, the breaker/ isolator should operate to isolate the faulty parts, and finally, the recloser should operate to reconnect the healthy parts of the microgrid networks.)

SECTION V.

Control and Stability of AC Microgrids

Robust control techniques are required for a stable and economic operation of the AC microgrid. This section discusses the following control techniques and stability issues of AC microgrids:

Voltage and frequency control
Coordination between DERs and power sharing aspects
Microgrid synchronization
Power flow control between multiple microgrids
Smooth mode switching
Loss reduction
Economic efficiency
Blackout mitigation

A. Hierarchical Control Schemes

To fulfill the above mentioned control aspects, three different levels of control strategies, namely (1) primary, (2) secondary, and (3) tertiary controls are applied in AC microgrids, and is known as the hierarchical control. The AC microgrid hierarchical control structure is shown in Fig. 6. Objectives of this three-level hierarchical control scheme are described in brief in subsequent sections.

FIGURE 6.

The conceptual framework of the hierarchical control schemes.

1) Primary Control

The prime functions of the primary control are: (1) to maintain the voltage and the frequency within acceptable limits, (2) active and reactive power sharing between DERs operating in parallel, and 3) to ensure plug and play operations. The primary control should be faster compared to secondary and tertiary control schemes. During the mode transition from the grid-connected to islanded mode, primary control should maintain stable voltage and frequency. Otherwise, power generation and load demand mismatch could cause voltage and frequency fluctuations, which could lead to microgrid instability. The primary control scheme is consisted of voltage and current control loops of DERs, where VSC based DERs can be operated either as a voltage controlled voltage source inverter (VCVSI) or a current controlled voltage source inverter (CCVSI) [67]–[70]. Most commonly, the droop control and virtual impedance based methods are used as primary control strategies for power sharing among DERs in AC microgrids.

2) Secondary Control

Although primary control maintains stable voltage and frequency, still there is a steady-state voltage and frequency deviations in AC microgrids. Secondary control is a controller which regulates the steady-state voltage and frequency deviations. Secondary control is usually slower than the primary control [71], [72]. As optimization strategies are used in secondary control it requires bi-directional communication links. Secondary control schemes can be divided into centralized and decentralized control schemes. Usually, centralized control is used in small AC microgrids while decentralised schemes are used in large microgrids.

3) Tertiary Control

The steady-state voltage and frequency deviations caused by the primary control are regulated by the secondary control, whereas the power flow between the utility grid and the microgrid are managed by the tertiary control, and it facilitates economic operation of the microgrid. When the microgrid is connected to the main grid, tertiary control manages the power flow between the utility grid and the microgrid by adjusting the power reference of DERs. As the last control level, it is slow compared to primary and secondary control schemes [69]. Different optimization algorithms, e.g., grossing algorithm, game theory, particle swarm optimisation, are employed to ensure economic operation. When all DERs are operating at the same marginal cost, then the condition for optimal economic operation can be achieved [73], [74].

B. Communication Based Control Schemes

Renewable energy resources and energy storage systems within the microgrid should be coordinated properly to supply various loads. Therefore, the parallel operating DERs and battery storage systems are connected via the communication links [75] to coordinate their operation. Besides, the high capital investment of communication links, communication based control schemes can maintain proper voltage regulation and power sharing. The subsequent subsections will critically review the communication based control strategies of the AC microgrid.

1) Centralised Control Schemes

In centralised control schemes a central controller is employed in microgrids to determine the control actions of the DERs. An in-depth analysis of the microgrid central controller (MGCC) is given in [76], [77]. The MGCC initiates mode switching action (grid-connected to islanded mode or islanded to grid-connected mode) by observing the PCC voltage and the frequency. Besides, it monitors the power generated by individual DERs and optimizes the DERs power generation. Furthermore, the MGCC forecast the load and the generation plan within the microgrid based on the available previous load profile data and the system conditions [78]. The MGCC strategy needs the same synchronization signals and current sharing modules. The PEC interfaced DER has a phase locked loop (PLL) which maintains consistent phase among the synchronization signal, the frequency, and the output voltage. The current sharing module measures the load current at the PCC, which generates reference current of each DER by dividing total load current by the number of parallel operating DERs, where the reference load current is proportional to the capacity of individual DER. As the DERs are connected in parallel and controlled by the synchronization signal, there is a negligible frequency and phase difference. Central limit control is proposed in [79], [80], where all the DERs have the same configuration and tracks the same reference current to achieve equal current sharing. Also, multistage centralised control strategy in the presence of high penetration of electric vehicle is proposed in [81], [82]. As the microgrid is controlled centrally, it can achieve proper current sharing in both the steady-state and transient conditions. Due to centralized control, it is difficult to expand the microgrid and it cannot tolerate a single point of failure.

2) Distributed Control Schemes

This method has no central controller and each converter has individual control unit [83]–[86]. The point of common coupling (PCC) or common bus is used as the current sharing bus to generate and share reference current among parallel operating DERs. Same like the centralized control, an additional current loop is required to generate reference current for the current sharing. In [84], a distributed AC microgrid control is proposed to restore voltage, frequency, and reactive power sharing, whereas in [85], a two-layer distributed control of AC microgrid is proposed. In [87], a robust decentralized controller for active and reactive power sharing for an islanded AC microgrid is presented where robustness of the system is ensured by a feedback linearization approach. In decentralized control, adjacent DERs exchange information among them rather than all the DERs available in the microgrid. Therefore, it requires a lower communication bandwidth. It can tolerate a single point of failure. As the adjacent DERs exchange information between them, it degrades the microgrids extenderbility and redundancy.

3) AC Master Slave Control

In this method of control, several DERs act as master DERs and regulates bus voltage and determines the current reference for the slave DERs [88]–[90]. Slave DERs are continuously tracking current references defined by the master DERs for current sharing. Since the slave DERs communicate with the master DERs, there is no need of the PLL unit for synchronization. The major drawback of this control is in case of a failure of the master; the entire microgrid system will fail [88], [89]. Many approaches have been proposed to solve this problem. Random selection of master by the rotating priority window is proposed in [91], whereas auto master-slave control is developed in [89], where the highest capacity DER unit acts as a master unit and is driving the power to the bus and generates a reference for the other DERs. In [88], utility interface at PCC is being used as a master DER. In summary, this control method offers excellent power sharing capability. In case of a failure of the master DER, it switches to am improved control strategy, i.e., a normal DER acts as a master DER. Like other communication based control, it adds expandability and redundancy of the system. A comparative analysis of these three control methods are summarized in Table 5.

TABLE 5 Comparison of Communication Based Control Methods in AC Microgrids

4) Droop Based Control Techniques

This section critically reviews control aspects, problems, and prospects of droop control. The droop control method, which is independent of inter-unit communication, is most commonly used between parallel operating converters (also know as LC). It can be implemented easily and it enables plug and play operation. A DER with voltage phasor $(\textit E\angle \delta)$ (shown in Fig. 7) is connected to the AC bus with voltage $(V_{o}\angle o)$ . The feeder impedance is $\mathit {R_{1}+jX_{1}}=\textit Z\angle \theta$ . The output power of the DER injected to the AC bus is calculated as [92]; $\begin{equation*} S=P+jQ=V_{o}*I^{*} =\frac {V_{o}*E\angle (\theta -\delta)}{Z}-\frac {V^{2}_{o}\angle \theta }{Z}\tag{1}\end{equation*}$ View Source where, $S$ , $P$ , and $Q$ are the complex power, real, and reactive power respectively. If the line is inductive i.e. $\theta =90^{0}$ , the real and reactive power can be calculated as [93]; $\begin{align*} P=&\frac {V_{o}*E}{Z}cos(\theta -\delta)-\frac {V^{2}_{o}}{Z}cos\theta \\ Q=&\frac {V_{o}*E}{Z}sin(\theta -\delta)-\frac {V^{2}_{o}}{Z}sin\theta \tag{2}\end{align*}$ View Source

FIGURE 7.

A DER connected to the microgrid.

a: Conventional Droop Control

Droop control enables DERs (static generators) to behave similar to the synchronous generator’s droop used in a conventional power system. Most commonly used conventional droop are active power-frequency droop ( $\textit P-\textit f$ ) and reactive power-voltage droop ( $\textit Q-\textit v$ ). The simplified circuit of a VSC based DER with $\textit P-\textit f$ and $\textit Q-\textit v$ droop connected to AC bus is shown in Fig. 8. The VSC consists of power sharing control loop, voltage control loop and inner current control loop. Power angle is related to active power, while voltage is related to reactive power. Therefore, the $\textit P/\textit Q$ droop control can be expressed as [94]; $\begin{align*} \omega=&\omega _{ref}-m_{p}*P \\ E=&E_{ref}-n_{Q}*Q\tag{3}\end{align*}$ View Source where $m_{P}$ and $n_{Q}$ are active and reactive droop slope. $\omega _{ref}$ and $E_{ref}$ are rated angular speed and rated voltage respectively. The droop coefficients have great influence on system stability and may be adjusted heuristically or by tuning algorithms [95], [96]. The $m_{P}$ and $n_{Q}$ can be formulated as; $\begin{align*} m_{P}=&\frac {\omega _{max}-\omega _{min}}{P_{max}-P_{min}} \\ n_{Q}=&\frac {E_{max}-E_{min}}{Q_{max}-Q_{min}}\tag{4}\end{align*}$ View Source

FIGURE 8.

A DER with ${P} - {f}$ and ${Q} - {v}$ droop control.

Here, subscripts $\textit {max}$ and $\textit {min}$ represent the maximum and minimum quantity of the respective variables. The conventional droop control slopes with inductive feeder-based microgrid are illustrated in Fig. 9. The major disadvantages of conventional droop are:

It assumes that feeder is highly inductive. But low or medium voltage feeders are mostly resistive [97].
As voltage is not a global variable, reactive power control is a challenge which results in circulating reactive current [98].
For non-linear loads, only fundamental components of the voltage and current are considered. The droop control needs to be modified to reduce total harmonic distortion (THD) [99], [100].

FIGURE 9.

${P} - {f}$ and ${Q} - {v}$ droop characteristics.

b: Voltage Active Power (VPD) and Frequency Reactive Power (FQD) Droop Control

Low voltage distribution feeders are mainly resistive. Therefore, conventional droop control does not work under these circumstances. For a purely resistive feeder, phase angle ( $\theta$ ) is small and cos( $\theta$ )≈ 1. Based on these assumptions (Eqn. 2) real and reactive power can be rewritten as [93]; $\begin{align*} P=&\frac {V_{o}*E-V^{2}_{o}}{Z} \\ Q=&-\frac {V_{o}*E}{Z}\delta\tag{5}\end{align*}$ View Source

Therefore, modified voltage-real power ( $\textit P-\textit v$ ) and frequency-reactive power ( $\textit Q-\textit f$ ) droop schemes are characterized by [101]; $\begin{align*} E=&E_{ref}-m * P \\ \omega=&\omega _{ref} +n * Q\tag{6}\end{align*}$ View Source where, $m$ and $n$ are active and reactive droop coefficients respectively. Graphical representation of VPD/FQD droop slopes are shown in Fig. 10. It provides improved performance in low-voltage resistive line based AC microgrid. However, its performance depends on microgrid parameters.

FIGURE 10.

${P} - {v}$ and ${Q} - {f}$ droop characteristics.

c: Angle Droop Control

Angle droop control is proposed in [102]–[104]. Similar to the conventional $\textit P-\textit f$ droop and $\textit Q-\textit v$ droop control, for the angle droop, the voltage angle ( $\delta$ ) is related to active power ( $\textit P$ ) instead of frequency ( $\textit f$ ). The $\delta$ of a DER can be calculated from (9); $\begin{equation*} \delta =\int \omega dt\tag{7}\end{equation*}$ View Source

The angle droop control is utilized to balance the power requirements between DER’s. The equations of voltage magnitude and phase angle are; $\begin{align*} \delta=&\delta _{ref}-m_{P}(P-P_{ref}) \\ E=&E_{ref}-n_{Q}(Q-Q_{ref})\tag{8}\end{align*}$ View Source

Here, $\delta _{ref}$ is reference phase-angle of the DER. Angle droop method ensures proper load sharing between DERs. Therefore, frequency deviation is small. However, if the power converters are not properly synchronized, the processing delays of the digital processor clock makes a slight frequency difference among the synchronising converters which may cause phase synchronization problems and ultimately microgrid becomes unstable [105].

d: Improved Droop Control

In [106], it is reported that the virtual impedance method is similar to the angle droop control and the frequency droop control with an additional power derivative feedback is similar to the virtual impedance method. The large droop co-efficient introduced by the improved droop control, ensures improved active and reactive power sharing which means large virtual impedance mitigates the effect of feeder impedance mismatch. By simultaneously adopting frequency droop and virtual impedance droop, the modified droop can be formed. The $\textit Q- \textit v$ and $\textit P-\textit f$ droop control can be expresses as follows; $\begin{align*} E_{ref}=&E^{*}-D_{Q}Q \\ \omega _{ref}=&\omega ^{*} -D_{P}P\tag{9}\end{align*}$ View Source where, $D_{Q}$ and $D_{p}$ are $\textit Q- \textit v$ and $\textit P-\textit f$ droop gain respectively. $E_{ref}$ , and $\omega _{ref}$ are reference voltage and angular frequency, $\omega ^{*}$ and $E^{*}$ are no load angular frequency and voltage respectively. Angle droop control can be formulated as; $\begin{align*} \delta _{o}=&\delta _{ref}-m_{p}p \\ E_{o}=&E_{ref}-n_{Q}Q\tag{10}\end{align*}$ View Source

Here, $n_{Q}$ and $m_{p}$ are $\textit Q-\textit v$ and $\textit P-\delta$ droop co-efficient respectively. After differentiating the above equation, it can be written as follows; $\begin{equation*} \omega _{o}=\omega _{ref}-m_{p}\frac {dP}{dt}\tag{11}\end{equation*}$ View Source

Therefore, frequency droop and virtual impedance droop are combined to form the modified droop as; $\begin{align*} \omega _{o}=&\omega _{*}-D_{P}*P-m_{p}\frac {dP}{dt} \\ E_{o}=&E^{*}-(D_{Q}+n_{Q})Q\tag{12}\end{align*}$ View Source This improved droop control method can achieve improved transient response and is more robust for a wide range of system parameter variations. The only drawback is it requires a high bandwidth for the controller.

e: Virtual Frame Transformation Method

A new reference frame (orthogonal linear transformation) is used to transform real-reactive power to a new reference frame [107]–[109], which can decouple the active and reactive power. In this approach, the real/reactive power is independent of feeder impedance. The expression of real and reactive power is given by; $\begin{align*} \left [{ \begin{array}{c} P^{*} \\ Q^{*} \end{array} }\right] = \begin{bmatrix} sin\theta &\quad -cos\theta \\ cos\theta &\quad sin\theta \\ \end{bmatrix} \left [{ \begin{array}{c} P \\ Q \end{array} }\right]\tag{13}\end{align*}$ View Source

This transformed real and reactive power ( $P^{*}$ and $Q^{*}$ ) are used in the basic droop characteristics equations. Similarly, the expression of voltage and frequency in this reference frame are [110]; $\begin{align*} \left [{ \begin{array}{c} \omega ^{*} \\ Q^{*} \end{array} }\right] = \begin{bmatrix} sin\theta &\quad cos\theta \\ -cos\theta &\quad sin\theta \\ \end{bmatrix} \left [{ \begin{array}{c} \omega \\ E \end{array} }\right]\tag{14}\end{align*}$ View Source

Conventional droop equation is used to calculate $\textit E$ and $\omega$ . This approach decouples real and reactive power, which improves system stability. However, this approach does not consider non-linear load’s negative impedance impact. Moreover, the frame transformation angles should be same, otherwise transformed voltage and frequency will be inaccurate.

f: Virtual Impedance Scheme

The virtual impedance control strategy is proposed in [111]–[113]. In order to decouple real and reactive power, a supplementary control loop based on virtual output impedance is incorporated to adjust the output impedance of the VCVSI. The control architecture is illustrated in Fig. 11. The reference voltage equation can be expressed as; $\begin{equation*} V_{ref}=V_{0}^{*}-Z_{v}(s)*I_{o}\tag{15}\end{equation*}$ View Source

FIGURE 11.

Virtual output impedance based control scheme.

Here, $Z_{v}(s)$ denotes the virtual output impedance. Output impedance, in general, is considered as purely inductive. Therefore, $Z_{v}(s)=s*L_{v}$ can be obtained by drooping the output voltage proportional to the derivative of the output current. The non-linear loads result in high harmonics. These harmonic effects can be eliminated by using a high-pass filter instead of $s*L{v}$ as follows [114]; $\begin{equation*} V_{ref}=V^{*}_{o}-L_{v} I_{o} \frac {s}{s+\omega _{c}}\tag{16}\end{equation*}$ View Source where, $\omega _{c}$ is the high-pass filter cutoff frequency. Properly adjusted time varying virtual output impedance can eliminate the current spike when the DER is connected to the microgrid. The time variant virtual output impedance can be formulated as [115]; $\begin{equation*} Z_{v}(t)=Z_{f}-(Z_{f}-Z_{i})e^{-t/T}\tag{17}\end{equation*}$ View Source where, subscripts $\textit i$ and $\textit f$ represent initial and final values, whereas $\textit T$ denotes the start-up process time constant. This method can handle non-linear loads. However, poor voltage regulation, undesirable voltage, and frequency deviations may occur in the DER due to inaccurate selection of the time constant.

g: Adaptive Droop Control Method

Compensation based adaptive droop control method for parallel operating DERs in AC microgrid is proposed in [97]. This approach adds two additional terms with the reactive power droop. Voltage is dropped across the feeder when DERs supply power to critical loads. One additional term compensates feeder voltage drop while other term improves reactive power sharing and hence improves the stability. A simple microgrid with two DERs supplying a load is shown in Fig. 12. The voltage equation for DER $i$ can be written as; $\begin{equation*} V_{i}=E^{*}_{i}-n_{Qi}Q_{i}-\frac {r_{i}P_{i}}{E_{i}}-\frac {x_{i}Q_{i}}{E_{i}}\tag{18}\end{equation*}$ View Source Here, $\textit r_{1}$ , $\textit n_{Q1}$ and $\textit x_{1}$ are the feeder resistance, reactive droop co-efficient and reactance of the DER-1, whereas $\textit r_{2}$ , and $\textit x_{2}$ are feeder resistance and reactance of DER-2 respectively. Here, the last two terms represent the voltage drop for feeder resistance and reactance respectively. By incorporating this two terms in the conventional $\textit {Q-v}$ droop equation of (3), the reactive power droop can be modified as; $\begin{equation*} E_{i}=E^{*}_{i,ref}-n_{Qi}Q_{i}+\frac {r_{i}P_{i}}{E^{*}_{i}}+\frac {x_{i}Q_{i}}{E^{*}_{i}}\tag{19}\end{equation*}$ View Source This method improves voltage regulation by considering voltage drop in reactive power droop control loop. However, voltage droop still depends on the active power control. In [97], this problem was solved by considering voltage droop as a non-linear function of both the active and reactive power as follows; $\begin{align*} E_{i}=&E^{*}_{i,ref}-n_{i}(P_{i},Q_{i})*Q_{i}+\frac {r_{i}P_{i}}{E^{*}_{i}}+\frac {x_{i}Q_{i}}{E^{*}_{i}} \\ n_{i}(P_{i},Q_{i})=&n_{Qi}+m_{Qi}Q^{2}_{i}+m_{Pi}P^{2}_{i}\tag{20}\end{align*}$ View Source Here, $m_{Qi}$ , $m_{Pi}$ , and $D_{Qi}$ are droop coefficients. The adaptive droop control method can improve reactive power sharing and stability margin. This method modifies the droop by considering feeder resistance and reactance. Therefore, small errors may cause instability of the entire microgrid. A concise overview of these droop control based strategies are summarized in Table 6.

TABLE 6 Review of Droop Control Based Control Methods

FIGURE 12.

Two DERs supplying a load.

C. Stability Studies of AC Microgrid

Small-signal stability analysis is an important technique to identify the dynamics of the AC microgrid. AC microgrids are usually comprised of VSC based DERs with active/reactive power controllers for parallel operation, dynamic loads (motor loads), passive loads, and resistive or inductive lines [130]–[132]. Extensive research studies have been carried out on small-signal analysis, voltage, and transient stability of AC microgrid to identify the dynamics of each parameter, such as controller dynamics, load dynamics, and system dynamics. In [133], performance of a microgrid controlled by a distributed controller was evaluated by small-signal stability analysis which synthesizes primary control loop, secondary control loop, and communication latency. It is reported that a microgrid with low latency communication link can eliminate some of the non-crucial communication edges to improve the rate of convergence. An AC microgrid with synchronized control under multiple parallel operating circumstances is evaluated in [134] by small-signal stability analysis. This research demonstrated that a doubly-fed induction generator (DFIG) with synchronized control strategy may enhance microgrid frequency support and provide power reserve capability. In [128], small-signal analysis of inverter dominated hybrid AC/DC microgrid is developed where the induction motor (IM) is used as a dynamic load. It was found that the droop gain of the VSC with dynamic load is small compared to the droop gain with static loads. Table 7 summarizes the small-signal analysis methods used for AC microgrid.

TABLE 7 Microgrid Modelling Methods

SECTION VI.

Control and Stability of DC Microgrids

For stable operation of DC microgrids, different control strategies such as centralized, decentralized, and distributed control strategies are used. The following control aspects are listed below based on literature:

Power sharing
Load sharing
Stable voltage
Power flow control between the adjacent DC microgrids
Smooth mode switching
Power loss reduction
Economic operation
Reducing blackouts

A. Various Control Structures

DC microgrid consists of converter based parallel DERs and battery storage systems. Each converter has voltage, current, and droop control as a local controller. Solar-PV operates on maximum power point tracking (MPPT) as a source dependant controller, and state of charge controller for energy storage devices. These local controllers are coordinated centrally for proper energy management via communication links, which are called centralized control. From the communication point of view, DC microgrid control strategy can be divided into three categories, 1) decentralized, 2) centralized, and 3) distributed control. Fig. 13 illustrates three basic control structures of the DC microgrid.

FIGURE 13.

Basic control structures of a DC microgrid; (a) Distributed control, (b) Decentralized control, (c) Centralized control.

1) Decentralized Control

The decentralized control method is based on LC. Data bus signaling (DBS), adaptive control of droop coefficient, and power line signaling (PLS) are commonly used in decentralized control. In DBS based decentralized control, DC bus voltage variation is taken as input for coordination among multiple DERs, ESS, and the utility grid. The utility dominating mode, storage dominating mode, and generation dominating mode are three basic operating modes of the DC microgrid and the DC bus voltage determines the operating mode. Therefore, the reliability and effectiveness of the decentralized control depend on the voltage measurement of the common DC bus [135], [136]. The principle of the DBS is the coordinated operation among solar-PV, wind, and ESS, which are achieved by the current-voltage (I-V) droop characteristics [137]. The grid interfacing inverters and ESS are controlled by poly-line droop curves with multiple segments [138]. The state of charge (SoC) controller of the ESS controls the output power of the solar-PV, wind, and utility grid. Virtual resistor (VR) with frequency shape is proposed in [139], where a super capacitor is used in ESS to cope with high-frequency power ripples and low-frequency response. As DBS relies on the local DC bus voltage and does not need communication infrastructure, it can be implemented easily. Therefore, the reliability and effectiveness of the decentralized control depend on the voltage measurement of the common DC bus.

The adjustment of droop coefficient adaptively without considering the change of operating mode is an extension of the conventional droop control method. It can adjust SoC among multiple ESSs to maintain accurate charging and discharging of ESSs. In [140]–[142] droop coefficient adjustment are used to balance SoC of multiple ESSs in both charging and discharging processes. Droop coefficient adjustment based on fuzzy logic is used in [143] to balance SoC of the ESS, whereas fuzzy logic based VR is employed to balance SoC of multiple ESSs in [144]. Adaptive droop adjustment needs expertise, and improper setting of droop curve may lead to the unstable operation of DC microgrid [145], [146]. PLS is also used in the decentralized method, which is a communication based method, and it is complex to implement than other two methods [147], [148].

2) Centralized Control

The centralized control of a DC microgrid has a central control (CC) system with digital communication networks which connects sources and loads. Distributed sources in small DC microgrids are controlled by CC with a high bandwidth communication using master/slave control [149], [150], while in large DC microgrids hierarchical control is used [69]. The hierarchical control method comprises LC of inverters and digital communication links (DCLs) based coordinated control. A campus microgrid system with hierarchical control is proposed in [151], while in [152] a hierarchical control of a stand-alone DC microgrid is employed for stable and economic operation. The ESS charging and discharging control with a different SoC is investigated for DC bus voltage stabilization, while DC microgrid is critically loaded. The researchers in [141] proposed an adaptive droop control as a primary controller for SoC balancing, while supervisory control is employed for mode transition and coordinated control of SoC of multiple ESSs as a higher level control. Since considerable independence between different control levels are introduced by the hierarchical control, DC microgrid continues its operation even after the failure of centralized controller. The CC provides the best performance, but it cannot tolerate a single point of communication failure and requires a costly redundant communication link [153], [154].

3) Distributed Control

In distributed control, there is no central control unit and all LCs communicate among themselves through dedicated DCLs. The distributed control can tolerate some point of communication failure compared to CC. Consensus algorithm has been used in microgrid applications [155], [156]. Basically, the consensus algorithm can work with low bandwidth communication techniques that ensure stable DC bus voltage and enhance output current sharing by proportional-integral (PI) controller [157]. The researchers in [158] proposed a dynamic consensus algorithm, which has DC bus voltage measurement system with noise resiliency, while neighbouring DERs voltage are also observed to calculate local DC set points. In [159], a linear circuit model based distributed control is proposed for droop control strategy of the DC microgrid, while [160] proposed an agent based approach for distributed control of DC microgrid. A unified distributed control is proposed in [161], and it does not require DBS. Energy management and DC bus voltage regulation are obtained by distributed control of ESSs in a DC microgrid. Distributed control can tolerate some point of communication failure and have better information awareness compared to centralized control. However, the stability margin and high complexity on analytical performance are the major shortcomings of distributed control [162]. Table 8 summarizes basic properties of these control schemes.

TABLE 8 Comparative Analysis of DC Microgrid Control Techniques

B. Control Techniques Used in DC Microgrids

This subsection critically reviews general control methods employed for individual DERs and storage systems in DC microgrids. As mentioned in the previous section, DC microgrid control architecture consists of LC and coordinated control. The LC of the DERs and the storage systems include voltage, current, and droop control. The proper LC of the individual DERs and the coordinated control between different components in the DC microgrid ensure proper power-sharing and stable operation. Some of these control techniques are presented bellow.

1) Droop Control

Droop control method is commonly employed in DC microgrids for DC bus voltage regulation and proper power sharing. Droop control methods in DC microgrid are classified into two categories, 1) power-voltage (P-V) droop or voltage droop, 2) current-voltage (I-V) droop or current droop [163], [164]. In current droop the DERs current reference is generated from the measured DC bus voltage, whereas in voltage droop the DERs voltage reference is generated from the load current. For both of these droop control methods, the output current/power dispatch of each DER is regulated by the DC bus voltage and the droop characteristics. Improved power sharing performance can be achieved with a large droop gain which can cause poor voltage regulation.

2) Inverse Droop

Inverse droop is proposed in [165] for DC-DC converter based parallel operating DERs in DC microgrids. The problem associated with the droop is the DERs output reference voltage decrease with the increase of the loads in the DC microgrids. Since the droop coefficient of the output current feedback is positive (negative for conventional droop), the reverse droop will linearly increase the reference output voltage with the load increase. Although the inverse droop does not use input voltage in the control loop, this droop control method increases the output reference with the increase of the load. This approach can be implemented without the central controller which ensures reliability and flexibility.

3) Adaptive Droop

Droop control method is commonly employed in DC microgrids for DC bus voltage regulation and proper power sharing. Improved power sharing performance can be achieved with large droop gain which can cause poor voltage regulation. Problems associated with the droop-based control method are the voltage deviations and the current sharing inaccuracy. To solve these problems researchers have proposed adaptive droop, where the gain of the adaptive PI controller is adjusted autonomously to improve current sharing errors and handle the non-linearity of the microgrid dynamics [166]–[168].

4) Virtual Impedance Method

The output impedance mismatch of the interfacing converter and to enhance active damping researcher have proposed [114], [176] virtual impedance method. Constant power loads (CPLs) have introduced negative incremental impedance and hence instability issues that are eliminated by a virtual impedance based method. This control approach is utilized to match the output impedance matching of the interfacing converter at harmonic frequencies and resonance damping. Virtual impedance in DC microgrid is related to active power sharing, while in AC microgrid virtual impedance is related to active and reactive power [174], [181]. Researcher in [182] proposed multi-agents stabilization to eliminate the negative impedance impact of CPL which will improve the fault-tolerant capability of the DC microgrid.

5) Master-Slave Control

The recent technological advancement in communication technologies (Zigbee, WiFi, etc.) and exchange of information (consensus, gossip, p2p, etc.) have paved the way for distributed control and power management in DC microgrid applications [183], [184]. The fast communication infrastructure based master-slave control for DC microgrid has been proposed in [185], [186], where some DERs operate as a master, and some other DERs operate as a slave. The master DERs regulate the DC bus voltage and acts as VSCs, whereas slave DERs act as current source converters. The reliability of the master-salve control will be challenged if there is a single point of communication infrastructure failure [187]. Key characteristics of the aforementioned control methods are presented in Table 9.

TABLE 9 Control Methods Used in DC Microgrid

C. Stability Studies of DC Microgrids

Stable operation of the DC microgrid is a major challenge where instability may occur due to abnormal weather conditions of intermittent DERs, control and line parameters, and load dynamics. The stability study of the DC microgrid is carried out to test the reliability and quality of the microgrid. Load dynamics influence the stability of the microgrid. Generally, impedance (or passive) loads and active loads (CPLs) are being used in the power system. Passive load introduces positive impedance which improves damping, while CPL loads introduce negative impedance which results in poor damping of the microgrid. Basic stability studies in DC microgrid are summarized in Table 10.

TABLE 10 Stability Studies in DC Microgrids

SECTION VII.

Control Techniques of Hybrid AC/DC Microgrids

The AC microgrid is the most commonly used microgrid configuration as the existing power networks are operating at AC, while the DC microgrid is getting the attention due to advantages such as increasing use of DC-based loads and sources, no synchronization problem, and no reactive power requirements. Therefore, hybrid AC/DC microgrid is the optimal architecture as it combines the advantages of both the DC and the AC microgrids [31], [194]–[196].