Distributed Power-Generation

Systems and Protection

This paper investigates the role of distributed power generation in power grid
resilience improvement and the associated protection issues.
By Frede Blaabjerg, Fellow IEEE, Yongheng Yang, Member IEEE, D ongsheng Yang, Member IEEE,
and X iongfei Wang , Member IEEE

ABSTRACT | Continuously expanding deployments of distributed gas), considerable worldwide attention has been paid to
power-generation systems (DPGSs) are transforming the making societies sustainable. Additionally, the exploita-
conventional centralized power grid into a mixed distributed tion and utilization of conventional fossil energy resources
electrical network. The modern power grid requires flexible pollute the natural environment and appear to affect the
energy utilization but presents challenges in the case of a high temperature on Earth. Thus, traditional centralized power
penetration degree of renewable energy, among which wind and generation using fossil fuels is considered unsustainable
solar photovoltaics are typical sources. The integration level of the in national long-term strategic plans. Consequently, many
DPGS into the grid plays a critical role in developing sustainable efforts globally have been directed toward developing
and resilient power systems, especially with highly intermittent more renewable energy sources, such as wind and solar
renewable energy resources. To address the challenging issues photovoltaics (PVs), solar thermal power, hydropower,
and, more importantly, to leverage the energy generation, bioenergy, and ocean power [1]–[4]. Typically, renewable
stringent demands from both utility operators and consumers energy sources are integrated in the form of distributed
have been imposed on the DPGS. Furthermore, as the core of power-generation systems (DPGS), as shown in Fig. 1. The
energy conversion, numerous power electronic converters power generation in Denmark has changed from central-
employing advanced control techniques have been developed for ized to decentralized with the widespread use of windfarms
the DPGS to consolidate the integration. In light of the above, this [5]. Fig. 2 depicts the evolution of worldwide renewable
paper reviews the power-conversion and control technologies energy capacity from 2000 to 2015, where hydropower
used for DPGSs. The impacts of the DPGS on the distributed grid ranks first with regard to total installed capacity, followed
are also examined, and more importantly, strategies for enhancing by wind and solar PV power. However, the most favorable
the connection and protection of the DPGS are discussed. sources are wind and solar PV power, as evidenced by the
growth rates for 2010–2015 shown in Fig. 3 [4].
KEYWORDS | Control; distributed power-generation systems The special requirements of hydropower (e.g., physical
(DPGSs); grid codes; grid resilience; photovoltaic (PV) power locations for river or lake resources) have slowed its utiliza-
systems; power conversion; power electronics; power grid tion and development. In contrast, wind and solar PV power
protection; wind power generation is easier to access, with less physical location dependency.
Hence, it has become dominant in DPGSs, as demonstrated
I. I N T RODUC T ION in Fig. 1. Furthermore, it has less impact on the environment
and a larger untapped capacity. As shown in Fig. 3, among
In the past decades, because of the foreseen exhaustion of the major renewable energy technologies, the worldwide
conventional fossil-based energies (e.g., coal, oil, and natural wind and PV power generation achieved the fastest growth
rates of 17% and 28%, respectively, in 2015 [4]. As a specific
Manuscript received March 9, 2017; revised March 22, 2017; accepted
April 10, 2017. Date of publication May 11, 2017; date of current version
example, in 2015, the wind power share of the total net gen-
June 16, 2017. This work was supported in part by the European Research Council (ERC) eration in Denmark was >50% [6]. Many other countries
under the European Union's Seventh Framework Program (FP/20072013)/ERC Grant
Agreement 321149-Harmony; and also by VILLUM FONDEN under the VILLUM
are catching up with a high growth rate [7]–[9]. Thus, these
Investigators GrantÐREPEPS. (Corresponding author: Frede Blaabjerg.) two renewables (wind and PV energy) will continue to be
The authors are with the Department of Energy Technology, Aalborg
University, Aalborg, DK-9220 Denmark (e-mail: fbl@[Link]; yoy@[Link];
the major resources of DPGSs [9]. Consequently, wind and
doy@[Link]; xwa@[Link]). PV-based DPGS are the focus of this study.
Digital Object Identifier: 10.1109/JPROC.2017.2696878
0018-9219 © 2017 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted,
but republication/redistribution requires IEEE permission. See [Link] for more information.
Vol. 105, No. 7, July 2017 | Proceedings of the IEEE 1311
Blaabjerg et al . : Distributed Power- Generation Systems and Protec tion

Fig. 1. Decentralized power generation in Denmark (left: centralized electric power infrastructure in 1985, right: decentralized electric
power infrastructure in 2015) [5]. As shown, more than 50% of the energy is covered by renewables.

Beyond clean energy generation for more sustainable capable of withstanding abnormal interruptions. With these
societies, the integration of massive DPGSs poses many chal- considerations, the transmission system operators (TSOs)
lenging issues to the distribution power grid and to the utility and/or distributed system operators (DSOs), together with
[10]–[14]. For instance, owing to the energy resource inter- other stakeholders, have issued stringent interconnection
mittency, the power injected into the distribution networks codes to guide the commissioning and operation of DPGSs
by the DPGS is always time-varying and fluctuating, which [15]–[17]. The relevant standards have helped to harmonize
may affect the network stability, especially at a high penetra- the way in which the TSOs/DSOs and other businesses have
tion degree of renewables under the current mixed energy worked to increase the penetration of DPGSs for a more eco-
infrastructure (i.e., conventional and decentralized genera- friendly society. In this case, these guidelines have been the
tion systems) [8]. Additionally, for operation in harsh envi- main design and planning benchmarks for DPGSs. On one
ronments (e.g., offshore wind DPGS), the DPGS should be hand, as the power electronics technology is the key to con-
necting distributed energy resources [18], the development
of the DPGS is driven by the fast advancement of power con-
verter technologies. On the other hand, the aforementioned

Fig. 2. Global accumulative capacity of renewable energy from Fig. 3. Global growth rate of installed capacity for renewable
2000Ð2015 based on the data available from IRENA [1], where technologies from the end of 2010 to 2015 and in 2015 based on
hydropower also includes pumped storage and mixed plants; the data from REN21 [4], where CSP represents the concentrated
marine energy covers tide, wave, and ocean energy. solar power.

1312 Proceedings of the IEEE | Vol. 105, No. 7, July 2017

 Blaabjerg et al. : Distributed Power- Generation Systems and Protec tion

challenges limit the focuses of DPGS research and develop-

ment to the reliability, affordability, scalability, flexibility,
stability, and efficiency of the technology [19], [20].
However, as previously mentioned, the modern DPGS is
mixed with highly penetrated renewable energy sources and is
vulnerable to severe weather/climate conditions. Therefore,
concerns regarding the resiliency have been raised, and the Fig. 4. Wind-turbine power system-based DPGS (in some cases, the
gearbox is removed).
responses of the DPGS to extreme weather conditions should
be addressed [19]–[23]; otherwise, power outages might
occur. For instance, an estimated 679 widespread power out- grid voltages. In general, to maximize the energy harvested, a
ages occurred in the United States between 2003 and 2012 maximum power pointing tracking (MPPT) scheme should be
[21], where the power interruption incidents were caused employed to control the wind turbine speed and/or the pitch
by extreme climate conditions. Accordingly, highly resilient angle. Clearly, the captured power fluctuates as the input
DPGSs are required. In contrast to system reliability, the
wind speed changes. Depending on the impedance of the dis-
resilience mainly involves the ability of the DPGS to
tributed grid, the power fluctuation can affect the stability of
1) anticipate potential events; the entire system, e.g., influence the voltage variations.
2) rapidly recover from the observed events; As shown in Fig. 4, a gearbox is typically adopted in the
3) adapt to prevent future events.
Wind DPGS. For multimegawatt (multi-MW) wind turbines,
Obviously, to minimize the impacts of disruptive events, the rotational speed of the turbine rotor is low. Hence, bulky
potential interruptions should be identified as early as possi- generators are needed to capture the wind energy, which may
ble by monitoring the entire system in real time. When a sud- incur a high installation cost. As a consequence of using a gear-
den incident is recognized, the DPGS should rapidly react to box, the mechanical power can be converted with a higher
the disturbances via operational and structural changes (e.g.,
speed and lower torque, reducing the size and weight of the
control). The time of response and recovery and the level of
electrical generator [33]. Currently, the use of power electronic
recovery are the main indicators for assessing the resilience
converters is inevitable. They provide controllability of the elec-
of DPGS. In the post-event period, the DPGS must learn
from these disruptions and adapt in order to prevent similar trical power and allow the implementation of advanced func-
future events. The aforementioned functions can be achieved tions (e.g., enhancing the resilience). Finally, a transformer is
through advanced monitoring and intelligent control systems used to boost up the voltage level so that a more efficient power
(at the power converter level and the entire system level). In transmission in the distributed grid is achieved.
literature, researchers have demonstrated schemes for enhanc- 1) Requirements for Wind DPGS: The power electronic
ing the system resilience [24]–[32]. This confirms that the converter is the core of the Wind DPGS. Its role is becom-
DPGS can offer great flexibility and numerous possibilities for ing increasingly critical with the fast growth of the capacity
enhancing the entire power grid and consolidate its resilience. of individual wind turbines (now close to 10 MW). Many
Herein, in light of the above, DPGS technologies are first
advanced functions can be realized through the control of
reviewed, mainly with regard to wind turbine power systems
the power electronic converter. This indicates that far more
and PV power systems, for which the necessity of a resilient
stringent requirements should be considered than ever
DPGS is presented. Then, resilience-related control schemes
before. Fig. 5 summarizes the demands of wind power sys-
are selectively discussed, and the main grid requirements are
introduced. The protection issues for a DPGS integrated into a tems at different levels.
power grid are explored in Section IV. Future challenges, par-
ticularly concerning the resilience of the distributed power grid,
are presented in Section V, along with concluding remarks.

II. CONFIGUR ATIONS OF T YPICA L DPGS

A. Wind-Power-Generation Technology (Wind DPGS)
A typical configuration of a wind power system connected
to distributed grids (Wind DPGS) is shown in Fig. 4. As
shown, the input of such a DPGS is wind, which is converted
into mechanical energy by the turbine. Then, the turbine Fig. 5. Demands for the wind-turbine power-system-based DPGS,
​​ in​​​ and ​​Qi​ n​​​are the active and reactive power exchanges
where P​
drives an electrical machine (generator) that is controlled by between the generator and the power electronic converter,
power electronic converters. An electrical power conversion respectively, and ​​P​o​​​ and ​​Qo​ ​​​are the active and reactive power
stage ensures that the output currents are in phase with the exchanges between the power converter and the grid, respectively.

Vol. 105, No. 7, July 2017 | Proceedings of the IEEE 1313

Blaabjerg et al . : Distributed Power- Generation Systems and Protec tion

The current (​​i​gen​​​) flowing in the generator rotor or sta-

tor should be regulated in order to control the electromag-
netic torque. This has two major purposes: 1) maximizing
the power extracted from the wind turbine; and 2) balanc-
ing the energy flow in the case of dynamics because of the
inertia mismatch between the mechanical and electrical Fig. 7. Wind-turbine DPGS based on the variable-speed generator
power conversions. For the grid side, the power converter technologies (AGs and SGs) with full-scale power electronics.
should be able to emulate the behaviors of conventional
power plants regardless of the wind speeds. That is, the
power electronic converter should maintain the frequency the DFIG are directly connected to the power grid. In con-
(Freq.) as well as the voltage amplitude of the distributed trast, the power electronic converters are the link between
grid (​​Vg​ ​​​). More importantly, under severe conditions, the the rotor windings and the distribution grid. Normally, the
Wind DPGS should withstand grid faults or even contrib- power processed by the converters accounts for 30% of the
ute to the grid voltage recovery [33]–[38]. This requirement capacity of the wind turbine [41]–[43]. The small capacity
in response to distribution grid faults indirectly reflects the of the power converters makes this DFIG concept attractive
need for high resilience for the Wind DPGS. Upon demand, from the viewpoint of cost saving. However, the DFIG sys-
the wind power system should also be able to exchange reac- tem uses slip rings, and in the case of grid faults, the power
tive power with the distributed grid. controllability is difficult to maintain [44]–[46]. That is, the
Because of the relatively large power capacity of an indi- DFIG system adopts partial-scale power electronics, and the
vidual wind turbine, failures of wind power conversion sys- controllability is limited under abnormal operation.
tems can occur in operation. Downtime of the entire DPGS To increase the controllability and power processing flex-
challenges the grid stability and incurs additional mainte- ibility, asynchronous generators (AGs) and synchronous gen-
nance costs. As a result, the reliability of the power elec- erators (SGs) with full-scale power electronics have gained an
tronic converters is important in the modern Wind DPGS. increasing market share, dominating the current wind power
Design for reliability has been introduced to such systems market [8]. A Wind DPGS with AG or SG is represented in Fig. 7,
[39], [40]. Additionally, at the same power level, the volt- which shows that the full-scale power electronics are the direct
age level of the wind generators may need to be increased to link between the distributed grid and the stator windings of
facilitate the power transmission; thus, step-up transform- the generators. Hence, the power generated by the wind tur-
ers are normally connected to the medium voltage, as pre- bine can be flexibly regulated. The generator can be a squirrel-
viously mentioned. Furthermore, the space of the nacelle cage induction generator (IG), a direct current (dc)-excited
and/or tower of wind turbines is limited, necessitating a SG (DCSG), or a permanent-magnet SG (PMSG). The elimi-
high power density and strong cooling for power converters. nation of slip rings, a simpler or even eliminated gearbox, full
Finally, the energy balancing is an important issue in the power and speed controllability, and better grid support ability
control and may result in extra costs for the entire DPGS. are the main advantages of this wind power concept in con-
trast to the DFIG-based wind systems. The main drawbacks
2) Wind DPGS Concepts: Wind power system designs
include the more stressed and expensive power electronic
include several constraints, such as the generator types, the
components and higher power losses in the power electronic
rating and topology of the power converters, and the speed
converter stage compared with the previous concept.
controllability [8], [33], [34]. As previously discussed, the
power electronic unit plays an important role in the Wind 3) General Control of Wind DPGS: Controlling a wind-
DPGS. Depending on the generator rating, the power elec- turbine DPGS involves both fast and slow dynamic control-
tronic converters vary significantly. Therefore, the configu- lers, as indicated in Fig. 5, because both the mechanical and
ration of the power conversion stage depends on the gen- electrical conversion subsystems should be controlled. The
erator type. The doubly fed induction generator (DFIG) has control functions can be categorized into three levels, as
dominated the market in the past decades. Fig. 6 shows a shown in Fig. 8. In general, the power flowing in and out
wind DPGS employing a DFIG, where the stator windings of of the generation system must be properly managed. The
power generated by the wind turbines should be controlled
using the mechanical systems (e.g., to adjust the pitch angle
of blades, yawing system, etc.). When the DSO has sent
out certain demands, the entire DPGS must satisfy these
demands through the control of the mechanical and electri-
cal systems. In addition, the currents injected into the dis-
tributed grid should be synchronized with the grid voltage.
As the input wind speed is not constant, the available
Fig. 6. Wind-turbine DPGS based on the DFIG technology with power also varies. Hence, more advanced functions should
partial-scale power electronics. be considered, such as the maximization of the generated

1314 Proceedings of the IEEE | Vol. 105, No. 7, July 2017

 Blaabjerg et al. : Distributed Power- Generation Systems and Protec tion

Fig. 8. General control structure for modern wind DPGS (​​i​gen​​​: generator current, ​​vd​ c​​​: dc-link voltage, Ω
​​ ​gen​​​: rotational speed of generator, ​​
θp​  itch​​​: pitch angle of rotor blade, i​
​​ g​​​: grid current, ​​vg​ ​​​: grid voltage, ​​Xf​ ilter​​​: filter impedance, PCC: point of common coupling).

power (MPPT) and the ride-through operation of the grid turbine wear out, the PV DPGS is more reliable than the
faults and grid support (injecting or absorbing reactive Wind DPGS. Nonetheless, both these DPGS technologies
power). For instance, the currents in the generators can share the same electrical conversion stages. That is, the
be controlled by adjusting the rotational speed of the wind power electronics are the key to the efficient and reliable
turbine to maximize the power production. In some cases, conversion of the solar energy, which is highly dependent
excessive power injection may drive the distributed grid on the environmental conditions (e.g., solar irradiance level
voltage level beyond the boundaries; thus, the wind-turbine and ambient temperature). Thus, in a similar way, many
DPGS should be able to limit the active power injection. advanced functions of the PV DPGS can be achieved through
Additionally, for operation under grid faults, the coordi- the control of the power electronic converters. For instance,
nated control of several subsystems in the wind turbine, the maximum power extraction from PV panels in response
such as the generator/grid side converters, braking chopper/ to extreme weather conditions, anti-islanding (AI), the ride-
crowbar, and pitch angle controller is necessary. through of distributed grid faults, etc., can be accomplished
Notably, various control functions can be achieved by properly controlling the PV converters.
through the control and adjustment of the entire wind DPGS 1) Requirements for PV DPGS: Distributed PV power-gen-
system. Thus, the resilience enhancement is also possible in eration systems are being rapidly developed. In some coun-
the control DPGS. The basic control functions, such as cur- tries, such as Germany, a large proportion of the electricity
rent regulation, dc-link stabilization, and grid synchronization generation is from distributed PV systems, and the propor-
must be rapidly performed to ensure stable and safe operation. tion continues increasing [50]–[57]. Although the continu-
ous deployment of PV DPGS to some extent resolves the high
B. PV Power-Generation Technology (PV DPGS) energy demands across the globe, the variability and nondis-
The fast development of PV cell technologies, the continu- patchability of PV DPGS (similar to the Wind DPGS) affect
ous cost reduction of PV modules, and advancements in power the stability and economical operation of distributed grids.
electronics have been the main driving forces for the inten-
sive deployment of PV DPGSs [47]–[50]. It is expected that
the cost of PV technology will continue declining, which will
make PV systems competitive among other renewable energy
systems. Hence, more PV DPGS will be seen in the future.
In contrast to the Wind DPGS, the DPGS with PVs as the
input does not involve mechanical conversion, as shown in
Fig. 9. Instead, the power generation is achieved by exploit- Fig. 9. PV power-system-based DPGS, where the block
ing the photovoltaic effect that converts solar energy to ªGeneratorº represents the PV panels that generate power via the
electrical energy. Because the mechanical parts of the wind photoelectric effect.

Vol. 105, No. 7, July 2017 | Proceedings of the IEEE 1315

Blaabjerg et al . : Distributed Power- Generation Systems and Protec tion

Fig. 10. Demands for PV power-system-based DPGS, where P​

​​ in​​​ is
the active power generated by PV panels, and ​​P​o​​​ and ​​Qo​ ​​​ are the
active and reactive power exchanges between the converter and
the grid, respectively.

To ensure the reliable, efficient, and less harmful trans-

fer of solar PV energy to the distributed grid, the PV DPGS
must comply with far stricter requirements than ever
before [51], [52], [58]. Fortunately, the control of PV power
Fig. 11. Connecting PVs to the ac grid (dc bus connections can
electronic converters can enable these functions by using be in series or in parallel): (a) module PV inverter for low-power
smart inverters [53]–[57]. In general, the demands of the PV applications; (b) string inverter for medium-power applications; and
DPGS can be categorized into three types, as shown in Fig. 10. (c) central inverters for commercial or utility-scale systems. DCÐDC
First, as previously mentioned, the power capacity of converters for the string inverter are optional. For high-voltage PV
systems (e.g., 750Ð1500-V dc), transformers are required.
the PV power-generation system is not as large as that of
the wind power system. Moreover, the power characteris-
tic of the PV DPGS is compatible with the behavior of the which otherwise may accelerate the degradation. Therefore,
distributed grid; thus, the requirements are easier to sat- proper temperature management may be needed for the PV
isfy than those of the Wind DPGS. For the PV side (i.e., DPGS. Additionally, to enhance the operation at a system
the power generator side), the current or voltage of the PV level (coordinated control), communication is essential.
panels should be controlled to capture as much energy as
possible. That is, the MPPT control should be performed 2) PV DPGS Concepts: Most of the aforementioned
for all PV DPGSs, regardless of the power rating. The power demands for the PV DPGS can be realized by the control
rating determines the configuration of PV power systems, of the power electronic systems, i.e., PV inverters. Hence,
as shown in Fig. 11. In some cases, dc–dc power converters an overview of the basic configurations for connecting PV
are required. Nonetheless, as the PV panels are degraded or panels to alternating current (ac) networks is presented
develop growing defects during operation, panel-level diag- in Fig. 11. As previously discussed, unlike the wind power
nosis and monitoring are also necessary for the PV DPGS. technology, the solar PV produces far less power per gen-
For the grid side (i.e., the distributed grid side), the require- erating unit (e.g., a single PV panel or string). Therefore,
ments are not as stringent as those for the wind DPGS; how- the PV DPGS normally consists of many panels or strings
ever, the power quality should normally be maintained at connected in parallel and/or series in order to increase the
a satisfactory level. The similarities between the PV DPGS output power within an acceptable range, as shown in Fig.
and the Wind DPGS for the distribution grid side include: 1) 11(b) and (c). In these two cases, the string/multistring
stabilizing the distribution grid voltage by providing ancil- inverters and center inverters are adopted as the interface
lary services; and 2) riding-through grid voltage faults. Both to the distributed grid.
are related to the grid resilience of the DPGS. The central inverter technology is the most widely
The power capacity per generating unit is low, but the adopted alternative for distributed power grids, as it is the
cost of energy is currently high; thus, there is great demand simplest way to collect dc power from PV panels with a low
for high-efficiency power conversion in order to achieve an construction cost. However, there are significant drawbacks
acceptable price per produced kWh for the PV DPGS. In for this configuration, including the following:
addition, as the power electronic converter is the core, simi- • high dc-link voltage (750–1500 V);
lar to the wind DPGS, the reliability of PV power converters • long dc cables (power losses);
is important, and also it is motivated by extending the total • losses due to a common MPPT and mismatch of
energy production (service time) and reducing the cost. panels;
Finally, owing to exposure or a smaller housing chamber, the • losses and reliability of the diodes;
PV power converters must be more temperature insensitive, • reliability of the DPGS depending on one inverter.

1316 Proceedings of the IEEE | Vol. 105, No. 7, July 2017

 Blaabjerg et al. : Distributed Power- Generation Systems and Protec tion

Fig. 12. General control structure of a PV power system connected to the distribution grid (ipv: PV output current, vpv: PV output voltage,
vdc: dc-link voltage, Si: solar irradiance level, Ta: ambient temperature, ig: grid current, vg: grid voltage, Xfilter: filter impedance).

Nevertheless, for a high-power and high-voltage PV lower physical inertia (no rotating components) than the
DPGS, multilevel power converters can be employed. In Wind DPGS, the control is simpler. However, in this case,
addition, several central inverters can be connected in par- the excessive active power from PV generators should be
allel to increase the power-generation flexibility. dispatched by 1) modifying the MPPT control; 2) activating
3) General Control of PV DPGS: According to the the dc chopper to absorb power; and 3) managing the power
demands shown in Fig. 10, and according to the previous exchange between the PV panels and the extra energy-stor-
discussions, the PV DPGS should be controlled to perform age systems. Notably, in these cases, the basic functions, such
these functions reliably and efficiently. Although the vari- as current regulation, dc-link voltage stabilization, grid syn-
ability of the PV inverter topologies and system configura- chronization, and AI, must be quickly performed. Regarding
tions increases the control difficulty, the general control resilience enhancement, the high-level coordinative control
objectives for a PV DPGS system are universal, including and operation among various DPGS may also be needed.
MPPT, grid synchronization, voltage/current control, active
power control, AI protection, system condition monitor-
III. R E SI L I E NC E-R EL E VA N T DE M A N DS
ing (e.g., PV panels), and ancillary services (especially for
A N D CON T ROL ST R AT EGI E S
resilience enhancement), as summarized in Fig. 12. With
the increasing PV capacity, the power flowing in and out of Renewable energy sources are variable, uncertain, and non-
the PV DPGS must be managed using other systems (e.g., dispatchable. Consequently, the DPGS based on wind and
energy storage systems) or even through itself; otherwise, PV resources may create severe issues, especially in response
the distributed grid voltage level and frequency may be vio- to extreme weather and in variable working conditions. This
lated. As previously mentioned, the entire Wind DPGS must property is referred to as the resilience of the DPGS. The
follow the set-point commands given by the DSO for system resilience of the DPGS characterizes the capacity to tolerate
stability concerns. This also applies for the PV DPGS. disruptions and the ability to recover from events [19]–[21].
That is, the more advanced features required for the To enhance the grid resilience, there are major three steps:
Wind DPGS in the past are now considered for the PV DPGS, 1) distribution grid planning; 2) enhanced operation; and
as the power capacity is drastically increasing in many areas. 3) emergency management. Fig. 13 shows possible opera-
For instance, delta power production control, frequency tional sequences of a typical DPGS. Many control demands
control through active power, voltage control through reac- at the converter level and/or system level are related to the
tive power, the ride-through operation of the distributed resilience of the distribution grid during the operation or
grid faults, and the provision of grid support in both nor- in contingent situations. In normal operation mode, the
mal and abnormal conditions to the grid have been adopted DPGS should be controlled at the converter level, while it
[57]–[60]. Typically, those features can be implemented in should also be coordinated among the entire distribution
the control loops of the power converters. Regarding the system. In the case of unexpected temporary grid incidents,
fault ride-through operation, because the PV DPGS has far the grid voltage level may decrease significantly. During this

Vol. 105, No. 7, July 2017 | Proceedings of the IEEE 1317

Blaabjerg et al . : Distributed Power- Generation Systems and Protec tion

Fig. 13. Operating time sequences for DPGS in connection with the resilience of the distribution grid, where the control (resilience-related
control functions) should be implemented at different levels, as indicated.

period, the distributed generators should remain connected vary without requiring actions from the DPGS. Accordingly,
in a short period (being the fault ride-through operation). how the corresponding active power control and reactive
Beyond this short period, the protection (e.g., islanding) power control are performed is a focus. For instance, the
schemes should be enabled. Both operation modes are operational windows related to the frequency and voltage
achieved at the converter level, as observed in Fig. 13. deviations are shown in Fig. 14, where the Wind DPGSs
Furthermore, when the grid is completely out of service, installed in Denmark and China should be able to oper-
the distributed generators may still power the critical load; ate within a range around the rated voltage and frequency
alternatively, together with the backup energy, they can [61]–[63]. In general, as shown in Fig. 14, the frequency and
help to restore the distribution grid. In this case, coordina- voltage deviation windows can be divided into the following
tion between the transmission system (system-level) or the three zones:
storage systems (converter-level) and the distribution sys- • continuous operation zones (normal operation);
tem may be necessary. As indicated in Fig. 13, at different • constrained operation zones (shaded areas);
stages, the required response time may vary (e.g., protection • immediate disconnection zones.
requires fast response).
Nonetheless, according to the discussions in Section II, When the frequency and voltage of the distributed grid are
it can be anticipated that the power electronic converters within 49–51 Hz (49.5–50.2 Hz for China) and 0.9–1.1 p.u.,
will be heavily involved in future DPGSs, making them respectively, the wind DPGS should remain connected in
power converter dominant. In light of this, and because the the normal operation mode. In addition, 100% power injec-
resilience is not specifically included in the design phase, tion is required. However, because of certain events, the
the resilience-related demands should be met by properly frequency and/or voltage can exceed the boundaries. In this
and intelligently controlling the power electronic con- case, the wind DPGS should perform power control to regu-
verters, which are also called smart inverters. As shown late the frequency and/or voltage, which will be discussed
in Fig. 13, system-level control and coordination are also later in this section. Nevertheless, as the power capacity of
important. In this section, the operational boundaries are the Wind DPGS is high, the frequency and voltage of the
reviewed, and then control strategies for regulating the fre- distributed grid are dictated by the power injected from the
quency and voltage of the DPGS are discussed. Finally, the Wind DPGS, as in the case with the conventional central
unintentional islanding issue and the restoration capabil- power plants.
ity of the DPGS are briefly explored. More specifically, the In most countries, for the PV DPGS, the capacity of a
response to abnormal conditions of the distributed grid is single PV DPGS is small compared with that of the Wind
covered in this part. DPGS, but it has recently increased drastically. With this
background, the DSO imposes basic requirements (i.e., grid
codes) on these systems in order to guarantee the quality
A. Tolerance of Frequency and Voltage Deviations of the generated power and ensure a stable connection.
For the DPGS, the grid requirements under normal For example, in IEEE Std. 929-2000 [64], the boundaries
operations include its tolerance to frequency and voltage of the grid voltage and frequency are specified as shown in
deviations, i.e., how much the frequency and voltage can Fig. 15. In normal operation, the PV DPGS should maximize

1318 Proceedings of the IEEE | Vol. 105, No. 7, July 2017

 Blaabjerg et al. : Distributed Power- Generation Systems and Protec tion

Fig. 16. Voltage profiles in the case of LVRT for wind (solid lines) and PV
DPGSs (dashed lines) in Spain (blue) and Denmark (red) [61], [62], [65].

Furthermore, owing to various eventualities (e.g., climate

related lightning strike and short circuit), the distributed grid
may enter a faulty condition. In this case, the distributed grid
voltage can be increased or decreased (i.e., voltage swells or
sags). As more renewables (wind and PV) have been con-
nected to the distributed grid, the grid code requires the DPGS
to ride through grid faults (low- and high-voltage faults) for
the stability concerns of the entire distributed network [61],
Fig. 14. Examples of frequency and voltage operational windows
[62], [65]. Fig. 16 exemplifies the low-voltage ride through
for wind DPGSs in (a) Denmark (power above 50 kW) and (b) China (LVRT) for DPGSs in Spain and Denmark. Generally, the
[61], [63]. When the frequency and voltage of the grid at the PCC are LVRT requirements can be divided into three zones. Taking
outside the indicated regions, disconnecting the wind DPGS within the requirement for the Wind DPGS in Denmark as an exam-
specified time durations is required. ple, in zone A, the DPGS should remain connected to the dis-
tribution grid and maintain the power production. When the
distributed grid voltage level is in zone B, the Wind DPGS
the output power, which is known as MPPT control. At the must also stay connected to the grid. Simultaneously, the
same time, the power quality should be maintained, e.g., DPGS must provide maximum voltage support by injecting
a total harmonic distortion (THD) level lower than 5%. reactive currents. The reactive-current support is for stabiliz-
However, when a PV DPGS with a higher power capacity ing the entire faulty distributed grid. In contrast, the wind
is connected to MV/HV distributed networks, the story DPGS is allowed to disconnect only when the grid voltage
may be revised. In this case, the power injection from the level sags into zone C (or the fault duration exceeds the limit),
PV DPGS can significantly affect the frequency and voltage as shown in Fig. 16. Similarly, in the case of high-voltage ride
profiles of the distributed grid. As a result, similar measures through (HVRT), in some countries, the DPGS should also
for the PV DPGS should be taken in order to regulate the remain connected and be able to provide reactive current
frequency and voltage. support upon demand. Nevertheless, when the grid fault is
cleared, the production of the DPGS should be resumed at a
limited rate.
As previously mentioned, reactive currents from the
DPGS are required in the case of distributed grid faults.
Fig. 17 shows examples of the required reactive current
delivery in Denmark and Spain in the case of fault ride-
through operation [61], [62], [65]. When the voltage level
exceeds the range (0.9–1.1 p.u.), the reactive-current supply
is prioritized to support the distributed grid voltage recovery.
During these periods, the active power production should
be maintained if possible, but this is not required, because it
may trigger the inverter protection. Additionally, in extreme
conditions, recurring faults may occur; in this case, the
Fig. 15. Example of the voltage and frequency windows for the PV
DPGS should also stay connected within certain defined peri-
DPGS in the IEEE Std 929-2000 [64], where the time to disconnect ods [61], [62]. Overall, the fault ride-through operation is a
the PV DPGS is indicated. scheme for tolerating voltage dips and rises in the distributed

Vol. 105, No. 7, July 2017 | Proceedings of the IEEE 1319

Blaabjerg et al . : Distributed Power- Generation Systems and Protec tion

Fig. 18. Frequency response and frequency regulation profiles for

the DPGS in Denmark [61], [62], where ​​P​avail​​​is the present available
Fig. 17. Reactive current ​​I​q​​​demands for DPGSs in Spain (blue) power, P​ ​​ min​​​is the minimum power, P​
​​ Delta​​​is the power difference, ​​
and Denmark (red) [61], [62], [65], where I​ ​​ q​​​is the required current f​set​​​is the frequency set by the TSOs, and f​ ​​ min​​​ and ​​fm
​ ax​​​ are the
during the grid fault and ​​V​PCC​​​is the voltage level at the PCC. minimum and maximum distributed grid frequencies, respectively.

grid and thus preventing the collapse of the entire system. overloading of the distributed grid. For example, it is neces-
Considerable research has focused on the fault ride-through sary to keep the active power constant during wind-speed
capability [8], [14], [41]–[46], [58], [66]–[68], which can changes (or solar-irradiance changes) or limit the ramp rate
be achieved through the control of the power electronic of the active power. Fig. 19 presents different constraint func-
converters of the DPGS. Notably, depending on the system tions required for the Wind and PV DPGSs, where the abso-
structure, in some cases, extra equipment may be required to lute production constraint, the delta production constraint,
assist the DPGS to ride-through grid faults. and the power gradient constraint are also included. Notably,
all these active power constraints can be realized through the
control of the power electronic interfaces in the DPGS.
B. Frequency and Voltage Regulation Another grid-stability index is the voltage level. The reac-
As renewable energy sources are variable and uncertain, tive power exchange between the distributed generators and
the injection of fluctuating power by the DPGS can affect the distributed grid can change the voltage profile at the point
the stability of the distributed grid. Either the grid frequency of common coupling (PCC). By controlling the reactive power,
or the voltage level may be of outside the boundaries, as pre- the voltage level is regulated, which is known as automatic volt-
viously discussed. In the case of frequency deviations, the age regulation (AVR). In general, the reactive power require-
DPGS should be able to automatically change the active ment is usually expressed in three different ways [61].
power production and must also perform frequency control • ​Q​control. In this control function, the reactive
in order to stabilize the distributed grid frequency. These power should be controlled independently of the
are known as frequency response and frequency regulation, active power at the PCC.
respectively, and are achieved through the frequency-active • Power factor control. The reactive power is con-
power droop relationship. trolled proportionally to the active power at the
Fig. 18 shows the frequency response and frequency point of connection, which results in a constant
control curves for the DPGS in Denmark. The DPGS should cos​φ​, where φ​ ​ is the power angle.
be able to reduce the active power production in the range
of 2%–12% of the nominal power [61], [62], according to
any critical frequency point (50 Hz ​≤ ​ fset
​ ​​  ≤​52 Hz; typi-
cally, the frequency is set as ​​f​set​​  =​50.2 Hz). Upon demand,
the DPGS must also enable the frequency-control functions,
as demonstrated in Fig. 18. That is, all the frequency points
and thus the droop curves should be implemented when
the frequency is measured. Clearly, an accurate frequency
measurement is necessary to ensure stable operation and
fast dynamics.
In addition, to further alleviate the impacts of the fluctu-
ating power generated by the DPGS, different active power
control constraints have been introduced in recent grid
codes. As stated in [61]and [62], these active power control Fig. 19. Different active power control functions for the DPGS to
functions are defined in order to prevent the instability or ensure grid frequency stability [61], [62].

1320 Proceedings of the IEEE | Vol. 105, No. 7, July 2017

 Blaabjerg et al. : Distributed Power- Generation Systems and Protec tion

Fig. 21. Non-detection zone to assess the AI control methods for

DPGS (UV�under voltage, OV�over voltage, UF�under frequency,
Fig. 20. Example of the reactive power control for voltage
OF�over frequency), where ​ΔP​and ​ΔQ​are the real and reactive
regulation [61], [62], where ​​Q​min​​​ and ​​Qm
​ ax​​​are the minimum and
power outputs of the grid, respectively.
maximum reactive power, respectively; ​​V​PCC​​​is the voltage level
at the PCC; and the superscripts ªminº and ªmaxº represent the
minimum and maximum values of V​ ​​ PCC​​​, respectively.
• Inverter-resident detection—relying exclusively on
software implementation inside the DPGS control
system, which does not require any hardware modi-
• Voltage control. This function controls the voltage
fication.
at the voltage reference point by changing the reac-
tive power generation. Fig. 20 shows an example As discussed in Section II, power electronic converters are
of the voltage control through the reactive power widely used in DPGSs. Hence, the inverter-resident detec-
adjustment of the DPGS. tion approach has gained popularity. Commonly, it can
be categorized into three groups: passive detection, active
The reactive power control and voltage control functions are
detection, and hybrid solutions (combination of passive
mutually exclusive, which means that only one of the three
aforementioned functions can be activated at a time. There and active detection). The reliability of the islanding detec-
are several additional power control functions defined for tion methods can be represented by the non-detection zone
the DPGS. All of these are designed to ensure the frequency (NDZ) defined in the power mismatch space, where the
and voltage stability of the DPGS. islanding is not detectable and there is potential for para-
sitic trips. Fig. 21 shows the NDZ. Different AI detection
methods are presented as follows.
C. Unintentional Islanding
1) Passive Islanding Detection Methods—Under/Over
Unintentional islanding can be one of the main technical Fre­quency (UF/OF) and Under/Over Voltage (UV/OV)
issues for Wind and PV DPGSs. Islanding operation occurs Island­ing Detection: Voltage and frequency monitoring of
when the power supply from the main distributed grid is the distributed grid is typically performed in order to trip the
interrupted—which can happen for several reasons—but inverter in case of UF/OF or UV/OV protections. The worst
the DPGS continues to supply power to the networks or the case for islanding detection is represented by the condition
local loads. Islanding operation results in 1) retripping the where the active power and reactive power are balanced,
line or damage to connected equipment due to the out-of- in which there is no change in the amplitude or frequency.
phase closure; and 2) safety hazards for distributed-system The passive methods have several advantages, such as a low
personnel that assume de-energized lines during the island- cost and a simple and straightforward structure, as only the
ing. Islanding occurs more easily in highly penetrated DPGS monitored voltage and frequency are used. This results in a
networks. To avoid these serious consequences, safety meas- large NDZ, which can be precisely determined in practice.
ures called AI requirements have been issued and embodied However, in most cases, passive islanding detection is con-
in the Wind and PV DPGSs [69]. The main approaches for sidered insufficient for AI protection in the DPGS.
islanding detection include the following.
•  Grid-resident detection—requiring either an 2) Active Islanding Detection Methods—Generation of
advanced communication system or an external Small Perturbations to Detect Islanding in PV Systems: The
switched capacitor at the PCC, which increases the active methods are based on the generation of small pertur-
entire system complexity and costs. bations at the output of the DPGS inverter. Small changes
•  External switch capacitor detection—based on in one of the power-system parameters (e.g., frequency,
the concept that an external capacitor being peri- phase, and harmonics) can be identified. If islanding occurs,
odically switched on in parallel with the grid pro- the small changes are amplified. Hence, compared with the
duces a zero-crossing delay proportional to the grid passive methods, the active islanding methods can quickly
impedance. detect islanding with a smaller NDZ. The most commonly

Vol. 105, No. 7, July 2017 | Proceedings of the IEEE 1321

Blaabjerg et al . : Distributed Power- Generation Systems and Protec tion

used techniques are a) frequency drift; b) voltage drift; and

c) grid-impedance estimation (output-power variation).
However, even small perturbations can push the distributed
grid voltage and frequency out of the nominal range [9]. In
the case of a weak network, the small changes can cause
the instability of the entire distributed grid. Hence, further
efforts should be directed toward designing proper active AI
methods for large DPGSs.
Fig. 22. Coupled relationship of grid restoration and recovery
3) Hybrid Islanding Detection Methods—Exploiting
between the DPGS and the transmission system.
the Strengths of the Passive and Active Methods: Here, the
hybrid islanding detection overcomes the limitations of the
active and passive techniques and exploits the advantages of emergencies in a far more quickly and efficiently. Moreover,
these techniques. In most cases, passive detection is used the DPGS can provide an initial source of power [74] dur-
to detect islanding first. If no clear disturbance is detected, ing system restoration in cases where the main generation
perturbations are injected (i.e., active methods are enabled). is unavailable or does not have black-start capability [75].
However, this may lead to a larger NDZ. Therefore, with the help of the DPGS, the entire grid res-
When the distributed grid is down, the distributed genera- toration can be started in a simultaneous bidirectional way,
tors (wind turbines or PV panels) may be required to power as shown in Fig. 22, which includes conventional top-down
critical loads. In this case, islanding and grid-connected oper- starting from the transmission level and bottom-up starting
ation should be seamlessly switched. More importantly, the from the distribution side. This bidirectional restoration
detection of the distributed grid failures and resynchroniza- procedure can significantly shorten the restoration time and
tion are essential to the operation transition. Overall, distrib- reduce the unserved electric energy during major grid fail-
uted generation systems can operate in the islanded mode. ures [76]. Specifically, the DPGS can play an essential role
To some extent, the employment of distributed generators in both aiding local critical load restoration and supporting
improves the grid resilience, as critical loads can still be pow- upstream grid black start.
ered on when the distributed grid is under disruption.
1) Aiding Local Critical Load Restoration: When DPGSs
are used to restore critical loads after an extreme event,
D. Extreme Climate Disaster and Restoration the major concern is to pick up more loads and reduce the
After a disaster, a resilient grid should be able to restore restoration time. Several technical topics have been stud-
service to critical loads as soon as possible, including hospi- ied regarding distribution system service restoration in the
tals, street lighting, water stations, and other infrastructures presence of a DPGS, including the development of optimal
that are associated with basic human needs [70]. However, energizing strategies and the enhancement of the restora-
a conventional restoration process usually is initiated at the tion ability and reconfiguration algorithms.
transmission level and proceeds toward the distribution To maximize the amount restored by the distributed gen-
level, as shown in Fig. 22. Thus, load restoration is per- erators, a multistage restoration procedure was proposed in
formed as the last step in the process. Because severe natu- [77]. The dynamic constraint of distributed generators and
ral disasters, such as hurricanes, floods, thunderstorms, and the limitation of the frequency deviation were considered.
blizzards, can impose a significant influence on the entire In [78], graphical theories were used in the DPGS-based load
power system and may even damage the large and central- restoration procedure to serve more loads and reduce the
ized power plants, bulk transmission lines, substations, and switching operations. In [79], the sequence of actions was
transformers [71], the conventional restoration strategies defined to coordinate multiple DPGSs for load restoration.
face many difficulties and take long time to complete the Because the centralized restoration strategies are highly
load restoration [72]. To cope with these challenges, new dependent on the communication infrastructure and are
techniques, such as distributed-generation decentralized prone to the single point of failure of the central control-
restoration strategies, may provide promising solutions for ler, several decentralized methods have been proposed for
enhancing the resilience of the grids. enhancing the restoration ability. In [80], a multiagent sys-
As previously mentioned, the DPGS can be isolated tem was proposed for load restoration to determine a fea-
from damaged portions of the main distribution grid in the sible restoration path. In [81], a distributed algorithm was
case of disruptions and sustain the power supply through developed for load restoration in the consideration of fault
the optimal management of multiple available distributed detection, location, and isolation, as well as a practical load
generation resources [73]. That is, the local generation, restoration procedure.
storage, and control of energy without the need for distant After an extreme event, multiple faults can damage the
generating units and long transmission lines can make the DPGS. Therefore, the reconfiguration of the DPGS can be
DPGS less vulnerable to disasters and allow it to respond to exploited to limit the fault propagation and allow the load to

1322 Proceedings of the IEEE | Vol. 105, No. 7, July 2017

 Blaabjerg et al. : Distributed Power- Generation Systems and Protec tion

Fig. 23. Examples of DPGS operations: (a) normal operation with frequency and voltage regulations; (b) islanded operation to power
critical load; and (c) restoration operation under distribution grid failures or disturbance. The green dashed line represents communication
links and the red lines with arrows represent distribution networks. The arrow indicates the direction of the power flow.

be served by numerous electrical islands. In [82], a strategy Furthermore, the DPGS should be able to handle large
for self-healing after natural disasters was proposed, which inrush load currents at the start of the re-energizing process
involved partitioning the distribution system into islanded [87]. In this case, the role of the energy-storage elements in
DPGSs, as demonstrated in Fig. 23. In [83], a graph-theo- the DPGS, such as ultracapacitors, batteries, or flywheels, is
retic restoration algorithm was designed to determine the very critical for maintaining the transient power balance of
optimal network configuration for a grid with distributed the entire system, as illustrated in Fig. 23.
generators. In summary, by integrating distributed generators
into distribution networks, the restoration capability can
2) Supporting Upstream Grid Black Start: DPGSs can be improved, along with the flexibility of the operation of
provide valuable energy to support the black start of the the entire DPGS. Fig. 23 further exemplifies the opera-
upstream grid. Compared with the local critical load resto- tion flexibility of a power grid with distributed genera-
ration, the requirements for the DPGSs to be served as the tors. According to Figs. 13 and 23, when there are dis-
initial black-start source are far higher in order to cope with ruptive events in the distribution and/or transmission
the challenges of supporting upstream grid restoration. networks, the entire power grid may be shut down to
When energizing the transmission lines, underground protect 1) the downstream equipment; 2) the distribu-
cables, and transformers, both sustained and transient over- tion network; and 3) the upstream power grid. In the
voltages may be induced by the capacitive charging current restoration period, the distributed generators, together
flowing through them [84], [85]. Therefore, the distributed with energy-storage elements and local combined heat
generators in the DPGS must have enough reactive power power plants, may contribute to the grid recovery. This
absorption capacity to absorb the reactive power during sys- depends on the capacity of the distribution network. For
tem restoration. Moreover, the DPGS must be able to with- PV systems, because there is no physical mass, the res-
stand a voltage increase that may result from sustained or toration contribution may be limited, as shown in Fig.
transient overvoltages. 23. Notably, if the distribution grid cannot be restored, it
In addition, large frequency deviations can occur when a can be changed to operate in the islanded mode in order
large load, such as a transmission line, is switched on, which to power critical loads in the system and prevent fault
can trip the protective relays and lead to a failure of the resto- propagation to adjacent networks. In this case, all the
ration [86]. Thus, the DPGS should have a fast dynamic char- distributed energy resources can be put into operation.
acteristic and be able to follow the load changes faster in order Notably, as shown in Fig. 23, communication and coordi-
to avoid large fluctuations in the frequency and voltages. nation are important.

Vol. 105, No. 7, July 2017 | Proceedings of the IEEE 1323

Blaabjerg et al . : Distributed Power- Generation Systems and Protec tion

I V. DP GS PRO T EC T ION
As the penetration of renewable-based DPGSs increases,
traditional SGs are being excluded from the grid. Thus, pro-
tection and coordination issues must be properly addressed.
Actually, the system protection has been defined in national
grid codes [61], [62], which state that the protective func-
tion is mainly to protect the distributed generator (e.g., wind
power plants) and the upward distribution grid from collapse.

Fig. 24. Comparison of the fault-current contributions from

A. Protection Issues in DPGS a conventional synchronous generator and an inverter-based
The major problem when integrating distributed gen- distributed generator.
erators into the distribution power grid is that the distri-
bution systems are designed as radial networks, delivering
the power in a unidirectional way: from substations to con- power electronic converters. To increase the fault-current
sumers [88]. The coordinated operation of circuit breakers detection capability, the inverters must be oversized, which
with overcurrent relays, reclosers, and fuses to protect this incurs additional costs. This means that a tradeoff between
unidirectional radial power grid from both temporary and the protection capability and the overall system cost should
permanent faults has been well established [89]. However, be made during the planning phase. Additionally, cost-effec-
with a high penetration of distributed generators, new mul- tive protection schemes should be developed.
tisource networks become active and are no longer radial, 2) Reduction in Reach of Impedance Relays: The reach
and the conventional protection is unsuitable for a high pen- of an impedance relay is the maximum fault distance that
etration level of distributed generators. The main impacts of triggers the relay in a certain impedance zone or in a cer-
the operation of distributed generators on the distribution tain time because of its configuration. This maximum
system protection are as follows. distance corresponds to a maximum fault impedance or a
1) Fault-Current Contribution: In the conventional net- minimum fault current that is detected [92]. In case of a
work based on SGs, phase–phase or phase–ground faults fault that occurs downstream of the bus where the DPGS is
normally result in an overcurrent. As exemplified in Fig. 24, connected to the utility network, the impedance measured
the fault current is significantly higher than the operational by an upstream relay is higher than the real fault imped-
or nominal (rated) current, which is the basic precondition ance. This is equivalent to an apparently increased fault
for traditional overcurrent-based protection techniques. distance, which is due to the increased voltage resulting
However, owing to the limited current rating of power elec- from an additional infeed at the common bus. As a con-
tronic devices, the fault current of the predesigned power sequence, the relay may be triggered with a faster grading
converter-interfaced DPGS is normally limited to a maxi- time response [93].
mum of about twice the nominal current [90]. Therefore, the 3) Auto Reclosure: For a temporary fault in the distrib-
fault current measured by the feeder protective relay, which uted grid, reclosers are intended to operate in a fast mode,
is located at the front end of the feeder, decreases drastically isolate the faulty feeder, and allow the fault to self-clear. To
compared with the case where no distributed generators are secure the proper operation of automatic reclosing and pre-
connected to the network. This may result in the delayed vent out-of-phase reclosure, distributed generators must be
operation of the relay in order to detect the faults. Because disconnected completely before the reclosure [94]. When a
of the fault-current contribution from the DPGS, the frac- distributed generation unit continues to operate after a sin-
tion of the fault currents measured by the overcurrent relays gle fault, two problems may arise if the utility reconnection
decreases. This reduction may cause the malfunction of the (i.e., automatic reclosure) is initiated after a short interrup-
overcurrent relays [91]. In the worst case, the fault is not tion. First, the fault may not have been cleared, as the arc
detected instantly. This can lead to high voltages, although was fed from the distributed generation unit. As a result, the
the fault currents are low. Moreover, if the fault remains instantaneous reclosure may not succeed. Second, because
undetected for a long period, it can spread throughout the of the active power unbalance, the frequency may change
entire distributed system and thus induce severe damage to in the islanded part of the distribution grid. In this case, an
equipment. Additionally, as indicated in Fig. 24, distributed attempt to reclose the switch will couple two asynchronously
generators employing power converters can react to grid dis- operating systems with active sources on both sides of the
turbances very fast. This may alleviate the impact of over- recloser, which results in the failure of the reclosure [95].
loading on the distributed-grid components [9]. To make the protection schemes work effectively and
Notably, the fault-current contribution from inverter- reliably in the presence of DPGSs, the following key techni-
based distributed generators depends on the sizing of the cal challenges should be well addressed.

1324 Proceedings of the IEEE | Vol. 105, No. 7, July 2017

 Blaabjerg et al. : Distributed Power- Generation Systems and Protec tion

• Unique fault characteristics of the distributed gen- transformations [97]. Using the dc values, the disturbance
erators—Because the short-circuit current levels of signal can be calculated as the deviation of the voltage signal
different distributed generators vary greatly depend- from a given reference. In the case of an asymmetrical fault,
ing on the oversizing and control, it is desirable to the dc components exhibit a ripple. Therefore, these com-
accurately model the transient characteristics of the ponents are first filtered out using notch filters. Then, they
current and voltage when a single fault or multiple are compared with the references. In [98], a fault-detection
faults occur in DPGSs. method based on the monitoring of the positive sequence
• Highly adaptive fault identification algorithms— component of the fundamental voltage was proposed. Using
Because the DPGS can be operated in different this method, both symmetrical and asymmetrical faults in the
modes (see Fig. 23) and its network can be recon- DPGS can be detected. In [99], a technique for differentiat-
figured dynamically, adaptive fault identification ing between three-phase, two-phase, and single-phase faults
algorithms that can effectively detect the change of was presented. This scheme is suitable for DPGSs with a high
the operational modes and network configurations penetration of distributed generators in islanded operation.
and accordingly adjust the protection scheme in an Numerous problems should be considered when imple-
adaptive manner are of great interest. menting the voltage-based scheme for DPGS protection. The
performance of the schemes can suffer owing to time delays
and filtering processes. Moreover, the detection time changes
B. DPGS Protection to Improve Resilience depending on the type of the fault and the magnitude of the
Protection for the DPGS is essential to enhance the grid voltage depth during the fault occurrence. Finally, the scheme
resilience by quickly identifying the fault and then isolating is highly dependent on the operational mode of the DPGS.
faulty components with little human intervention. In recent Improper protections can be triggered by voltage fluctuations
years, different protection schemes have been proposed for caused by nonfault events in islanded operation; the scheme
improving the reliable protection when the DPGS is operating is more robust in the grid-connected operational mode.
either in the islanded or grid-connected mode, aiming to pro-
tect the distributed generation sources and network within 2) Improved Overcurrent Protection: To protect the
power electronic device of the DPGS, the overcurrent
the DPGS, as well as the upstream grid network. This section
protection should consider the device rating. Any fault in
briefly reviews the available protection schemes, which are
the DPGS must be cleared without relying on high fault
categorized into six different types, as shown in Fig. 25.
currents.
1) Voltage-Based Protection: Fault detection in the case of Using symmetrical current components, an islanded
low-fault current networks can be achieved using voltage- DPGS can be protected against single line-to-ground and
source components. It is possible to calculate the values of line-to-line faults [100]. A symmetrical approach for protec-
voltage-source components for different types of faults [96]. tion was proposed in [101]. The protection scheme utilizes
The common practice is to monitor the output voltages of the a zero-sequence component to detect single-line-to-ground
distributed generation sources and then transform the three- faults; the negative-sequence current is used to identify the
phase ac voltages into dc quantities using the Park and Clarke line-to-line faults.
Furthermore, overcurrent protection schemes may ben-
efit from the communication in the DPGS [102]. A symmet-
rical component-based scheme was proposed, which relies
on communication and can locate both symmetrical and
asymmetrical faults in a timely manner. The communica-
tion is established only for exchanging status information
and not electrical measurements; thus, the required com-
munication bandwidth is reduced [103]. Another instan-
taneous overcurrent-based scheme was developed [104],
where an optical Ethernet cable was adopted. This scheme
offers instantaneous protection for local lines and remote
bus bars using two executive routines, regardless of where
the distributed generators are located.
The main problems with these protection schemes are
related to the high dependence on wide-area communica-
tion systems. Obviously, communication reliability prob-
lems can affect the protection performance. That is, in the
case of a failure in the communication system, the entire
Fig. 25. Categorization of the protection for DPGSs. protection scheme may be dysfunctional [105].

Vol. 105, No. 7, July 2017 | Proceedings of the IEEE 1325

Blaabjerg et al . : Distributed Power- Generation Systems and Protec tion

3) Differential Protection: Differential protection is overcurrent protection, the features of the distance protec-
based on comparing the currents entering and leaving the tion change very little, even in different operational modes.
protected zone. As long as the difference between these cur- Distance protection also has drawbacks. During faults,
rents exceeds a predefined fault value, the relays send a sig- the distributed generators located between the fault point
nal to the distribution generation source at the faulty zone, and the measurement point act as intermediate in-feeds,
and then the protection is enabled. Differential protection which affects the accuracy of the measurements and hence
has the fastest response time—approximately 5 ms—and the performance. Moreover, the fault resistance can affect
the fault value of the differential protection can easily be the measured impedance. Finally, current transients, har-
resolved. Moreover, it can be modified for both modes of monics, and decaying dc currents have significant effects on
operation, which makes it suitable for the protection of the accuracy of the measurements.
DPGSs [106].
5) Adaptive Protection: To fully utilize the DPGS for
In [101], differential protection was used together with
improving the grid resilience, the DPGS must usually
symmetrical-component calculations to detect faults and
operate reliably in both the grid-connected mode and the
determine fault types. In [107], differential relays were
islanded mode. Thus, the mode transition has a significant
used to protect a DPGS in both grid-connected and islanded
impact on the protection scheme, and the design of the
modes. By employing digital relays with communication, a
protection scheme is very challenging. Recently, adap-
differential protection scheme was developed in [108]. It
tive protection schemes have been presented as promising
addresses the problem of high impedance faults and is suit-
solutions for DPGS protection in the case of a high pene-
able for DPGSs with radial and meshed networks. In [109],
tration level. Such schemes allow the online adjustment of
the differential features were extracted from the fault cur-
both the relay settings and the characteristics using exter-
rent and voltages using the discrete Fourier transform, and a
nal signals [115].
decision-tree data-mining model was developed to make the
In [116], an adaptive fault current protection algo-
final decision. To reduce the cost of devices for differential
rithm was developed by analyzing the fault behavior of
protection, the optimal placements of protection zones and
a power converter-based DPGS. In this method, the set-
the protective devices for each zone were explored in [110].
tings of the instantaneous overcurrent protection scheme
A significant benefit of the differential protection prin-
are automatically adjusted for the new situation by com-
ciple for protecting the DPGS is that it can overcome the
paring the system impedance with the DPGS impedance.
problems of a low fault-current level and a reverse power
Another adaptive protection scheme using energy stor-
flow. However, protective devices must usually be installed
age and isolation transformers was proposed in [117].
at each line, and these devices rely on the communication
The protection scheme adaptively switches between
infrastructure to receive measurements. This incurs addi-
overcurrent protection in the grid-connected mode and
tional costs and introduces communication reliability prob-
voltage-based protection in the islanded mode. The mode
lems. Moreover, synchronized measurements are required,
transition is ensured by comparing the zero sequence
and unbalanced loads and transients may challenge the
impedance angles.
protection.
In [118], an algorithm based on numerical relays was
4) Distance Protection: Distance protection utilizes proposed to coordinate different relays in a specific micro-
impedance/admittance measurements to effectively detect grid. The overcurrent relay settings are calculated offline
faults and then perform trip actions. The protection scheme and then stored in the relays. The scheme has the ability
was first developed by Dewadasa for DPGS protection in both to detect faults with far smaller short-circuit levels in the
grid-connected and islanded operation modes [111], [112]. DPGS. In [119], an adaptive overcurrent protection strategy
In this scheme, faults are detected by employing a new type consisting of a real-time conventional block was introduced.
of admittance relay that has the characteristics of inverse An adaptive protection system that monitors and updates
time tripping. The main advantage of distance schemes is the setting of relays online according to the operating modes
that they are not affected by changes in the current levels, as of the DPGS was proposed in [120]. It employs communica-
they mainly depend on the measured impedance. tion links to collect data from intelligent electronic devices,
A new scheme with two procedures for main and backup and then the data are sent to a centralized controller for
protection was established using the extracted impedance real-time analysis. A new adaptive scheme based on a cen-
[113]. The first procedure is required to identify the fault tralized architecture was presented [121]. It performs offline
occurrence and provide a time reference for the exchanged fault calculations to determine the directional and nondi-
data. The latter operates in an automatic coordinated man- rectional overcurrent relay settings, which are then updated
ner with an inverse-time characteristic to provide backup periodically.
protection. In [114], the distance-protection approach Although adaptive relays provide flexibility, they have
for protecting the MV DPGS in both grid-connected and drawbacks. Replacing all the existing relays with adaptive
islanded modes is presented. Compared with the traditional ones is very expensive and requires the existing protection

1326 Proceedings of the IEEE | Vol. 105, No. 7, July 2017

 Blaabjerg et al. : Distributed Power- Generation Systems and Protec tion

Table 1 Comparison of Different Protection Schemes for DPGS

schemes currently used for distribution systems to be the power electronic technologies for wind and PV DPGSs,
upgraded. Additionally, adaptive relays usually need com- as the power electronics are the core of the energy con-
munication infrastructures for reliable and fast operation. version. More importantly, as the wind and PV energies
Moreover, prior-art knowledge of all possible DPGS config- are variable, uncertain, and nondispatchable, connecting
urations should be tuned according to the adaptive adjusting these renewables to the distributed grid may cause instabil-
rules, which makes adaptive relays very difficult to imple- ity. Therefore, stringent demands have been placed on the
ment in the case of large-scale DPGSs. DPGS. These were also reviewed in this article, and control
6) Fault-Current Compensation: Because the fault-cur- strategies were discussed. The investigation revealed that
rent levels of the DPGS differ between grid-connected and multiple control functions can be provided by the DPGS in
islanded operation, especially with inverter-interfaced dis- order to improve the reliability, performance, and resilience
tribution generation, it is challenging to design a protec- of the entire grid. The constraints can be implemented by
tion scheme that operates well in both modes. Hence, an properly controlling the power electronic converters of the
additional fault-current source (FCS) can be used to com- DPGS. This has become one important aspect for inverter-
pensate for the fault-current levels of different operational based DPGSs. However, it also introduces side effects. As
modes to the similar level, which allows the overcurrent the inverter-dominated DPGS does not have much physical
protection to function well with the conventional method inertia, the DPGS must be oversized in order to provide a
[122]. The synchronous condensers or the storage devices, satisfactory amount of fault currents, which increases the
such as flywheels, batteries, and ultracapacitors, can be total cost. Nonetheless, DPGS protection is challenging. In
used as FCSs for injecting high currents during faults [9]. this paper, the challenging issues regarding the DPGS were
The storage device-based FCS usually contains a storage summarized, and the state-of-the-art protection techniques
element, a power electronic converter, a triggering circuit, that can be applied to the DPGS were reviewed. Table 1 lists
and a charging module [123]. As soon as a fault is detected, the advantages and disadvantages of the protection schemes
the FCS is used to restore the system voltage, injecting as discussed for the DPGS.
much current as necessary. Once the fault is cleared, the
It can be concluded that the DPGS can increase the grid
FCS is switched off. The major problem with fault-current
resilience, as it can operate in both the grid-connected mode
compensation in the DPGS is that it requires significant
and the islanded mode. In the case of an islanded DPGS,
investments.
critical loads can be supplied upon demand when the main
grid is absent. Additionally, the DPGS can help to restore
V. CONCLUSION the transmission system after disruptions; in return, the
The technological developments in DPGSs were explored. DGPS benefits from the transmission grid when it must be
It was revealed that the DPGS-based wind and PV technolo- restored after failures. Communication and date processing
gies will be dominant in the future market and in future technologies may be critical for ensuring the reliable, effi-
power systems. This paper first provided an overview of cient, and resilient operation of distributed grids. 

Vol. 105, No. 7, July 2017 | Proceedings of the IEEE 1327

Blaabjerg et al . : Distributed Power- Generation Systems and Protec tion

R EFER ENCES [16] T. Basso, “IEEE 1547 and 2030 standards for [30] J. Kim, E. Muljadi, J. W. Park, and Y. C.
[1] International Renewable Energy Agency distributed energy resources interconnection Kang, “Adaptive hierarchical voltage control
(IRENA). (Mar. 2016). Renewable Energy and interoperability with the electricity of a DFIG-based wind power plant for a grid
Capacity Statistics, accessed on Mar. 6, 2017. grid,” Tech. Rep. NREL/TP-5D00-63157, fault,” IEEE Trans. Smart Grid, vol. 7, no. 6,
[Online]. Available: [Link] Dec. 2014. pp. 2980–2990, Nov. 2016.
DocumentDownloads/Publications/IRENA_ [17] H. Margossian, G. Deconinck, and J. Sachau, [31] R. Teixeira Pinto, M. Aragüés-Peñalba, O.
RE_Capacity_Statistics_2016.pdf “Distribution network protection considering Gomis-Bellmunt, and A. Sumper, “Optimal
[2] DAMVAD & Kariros Future. DK2050: Green grid code requirements for distributed Operation of DC Networks to Support Power
Growth in Denmark Towards 2050—Four generation,” IET Generat., Transm. Distrib., System Outage Management,” IEEE Trans.
Future Scenarios, Mar. 6, 2017. [Online]. vol. 9, no. 12, pp. 1377–1381, Sep. 2015. Smart Grid, vol. 7, no. 6, pp. 2953–2961,
Available: [Link] [18] F. Blaabjerg, K. Ma, and Y. Yang, “Power Nov. 2016.
Damvad_english_1007.pdf electronics—The key technology for [32] H. Farzin, M. Fotuhi-Firuzabad, and M.
[3] European Commission. (2016). Transforming renewable energy systems,” in Proc. EVER, Moeini-Aghtaie, “Enhancing power system
the European Energy System Through Innova- 2014, pp. 1–11. resilience through hierarchical outage
tion—Integrated Strategic Energy Technology [19] D. T. Ton and W. T. P. Wang, “A more management in multi-microgrids,” IEEE
(SET) Plan Progress in 2016. [Online]. Avail- resilient grid: The U.S. department of energy Trans. Smart Grid, vol. 7, no. 6,
able: [Link] joins with stakeholders in an R&D plan,” pp. 2869–2879, Nov. 2016.
files/documents/set-plan_progress_2016.pdf IEEE Power Energy Mag., vol. 13, no. 3, [33] F. Blaabjerg and K. Ma, “Future on power
[4] Renewables 2016: Global Status Report, pp. 26–34, Jun. 2015. electronics for wind turbine systems,” IEEE J.
accessed on Mar. 6, 2017. [Online]. Avail- [20] M. Panteli and P. Mancarella, “The grid: Emerg. Sel. Topics Power Electron., vol. 1,
able: [Link] Stronger, bigger, smarter?: Presenting a no. 3, pp. 139–152, Sep. 2013.
uploads/2016/10/REN21_GSR2016_FullRe- conceptual framework of power system [34] F. Blaabjerg, M. Liserre, and K. Ma, “Power
port_en_11.pdf resilience,” IEEE Power Energy Mag., vol. 13, electronics converters for wind turbine
[5] Danish Energy Agency. Overview Map of the no. 3, pp. 58–66, Jun. 2015. systems,” IEEE Trans. Ind. Appl., vol. 48,
Danish Power Infrastructure in 1985 and 2015, [21] “The president’s council of economic no. 2, pp. 708–719, Mar./Apr. 2012.
accessed on Mar. 6, 2017. [Online]. Avail- advisers and the U.S. department of energy’s [35] M. Tsili and S. Papathanassiou, “A review of
able: [Link] office of electricity delivery and energy grid code technical requirements for wind
tik/foer_efter_uk.pdf reliability, with assistance from the white farms,” IET Renew. Power Generat., vol. 3,
[6] [Link]. Electricity Generation, House office of science and technology no. 3, pp. 308–332, Sep. 2009.
accessed on Mar. 6, 2017. [Link] economic benefits of increasing electric grid [36] (Jan. 2003). Energinet—Wind Turbines
[Link]/EN/KLIMA-OG-MILJOE/Miljoerap- resilience to weather outages,” Tech. Rep., Connected to Grids With Voltages Below 100 Kv,
portering/Elproduktion-i-Danmark/Sider/ Aug. 2013. [Online]. Available: https:// accessed on Mar. 7, 2017. [Online]. Available:
[Link] [Link]/sites/prod/files/2013/08/f2/ [Link]
[7] F. Blaabjerg, D. M. Ionel, Y. Yang, and H. Grid%20Resiliency%20Report_FINAL.pdf
[37] Grid Code—High and Extra High Voltage,
Wang, “Renewable energy systems—Tech- [22] “2014 DOE resiliency electric distribution Tennet TSO GmbH, Dec. 2012.
nology overview and perspectives,” in Chap- grid R&D workshop report,” Dept. Energy,
ter 1 in Renewable Energy Devices and Systems Upton, New York, NY USA, Tech. Rep., [38] Requirements for Offshore Grid Connections in
with Simulations in MATLAB and ANSYS. Boca Jun. 2014. [Online]. Available: [Link] the Grid of TenneT TSO GmbH, Tennet TSO
Raton, FL, USA: CRC Press, 2017, gov/sites/prod/files/2014/07/ GmbH, Dec. 2012.
pp. 1–16, 2017. f17/2014ResilientGridWorkshop- [39] K. Ma, H. Wang, and F. Blaabjerg, “New
[8] F. Blaabjerg and K. Ma, “Wind energy sys- [Link] approaches to reliability assessment: Using
tems,” Proc. IEEE, to be published. [23] H. H. Willis and K. Loa, “Measuring the physics-of-failure for prediction and design
resilience of energy distribution systems,” in power electronics systems,” IEEE Power
[9] B. Kroposki et al., “Achieving a 100% renew- Electron. Mag., vol. 3, no. 4, pp. 28–41,
able grid: Operating electric power systems RAND Corp., Santa Monica, CA, USA, 2015.
[Online]. Available: [Link] Dec. 2016.
with extremely high levels of variable renew-
able energy,” IEEE Power Energy Mag., vol. 15, content/dam/rand/pubs/research_reports/ [40] H. Wang et al., “Transitioning to physics-of-
no. 2, pp. 61–73, Mar. 2017. RR800/RR883/RAND_RR883.pdf failure as a reliability driver in power
[24] R. K. Varma, S. A. Rahman, and T. electronics,” IEEE J. Emerg. Sel. Topics Power
[10] V. Knazkins, “Stability of power systems Electron., vol. 2, no. 1, pp. 97–114, Mar. 2014.
with large amounts of distributed Vanderheide, “New control of PV solar farm
generation,” Ph.D. Dissertation, TKH, as STATCOM (PV-STATCOM) for increasing [41] R. Pena, J. C. Clare, and G. M. Asher,
Stockholm, Sweden, 2004. grid power transmission limits during night “Doubly fed induction generator using back-
and day,” IEEE Trans. Power Delivery, vol. 30, to-back PWM converters and its application
[11] M. A. Chowdhury, A. H. M. Sayem, W. Shen,
no. 2, pp. 755–763, Apr. 2015. to variable speed wind-energy generation,”
and K. S. Islam, “Robust active disturbance
[25] P. Loevenbruck, “Residential storage to Electr. Power Appl., vol. 143, no. 3,
rejection controller design to improve low-
improve system resilience: Technical design pp. 231–241, 1996.
voltage ride-through capability of doubly fed
induction generator wind farms,” IET Renew. and cost estimation,” in Proc. CIRED [42] S. Müller, M. Deicke, and R. W. de Doncker,
Power Gen., vol. 9, no. 8, pp. 961–969, Workshop, 2016, pp. 1–4. “Doubly fed induction generator systems for
Nov. 2015. [26] X. Liu, M. Shahidehpour, Z. Li, X. Liu, Y. wind turbines,” IEEE Ind. Appl. Mag., vol. 8,
[12] Y. Hirase, K. Sugimoto, K. Sakimoto, and T. Cao, and Z. Bie, “Microgrids for enhancing no. 3, pp. 26–33, May/Jun. 2002.
Ise, “Analysis of resonance in microgrids and the power grid resilience in extreme [43] D. Xiang, L. Ran, P. J. Tavner, and S. Yang,
effects of system frequency stabilization conditions,” IEEE Trans. Smart Grid, vol. 8, “Control of a doubly fed induction generator
using a virtual synchronous generator,” IEEE no. 2, pp. 589–597, Mar. 2017. in a wind turbine during grid fault ride-
Trans. Emerg. Sel. Topics Power Electron., [27] Z. Wang and J. Wang, “Service restoration through,” IEEE Trans. Energy Convers.,
vol. 4, no. 4, pp. 1287–1298, Dec. 2016. based on AMI and networked MGs under vol. 21, no. 3, pp. 652–662, Sep. 2006.
[13] S. K. Mazumder and E. Pilo de la Fuente, extreme weather events,” IET Generat., [44] F. K. A. Lima, A. Luna, P. Rodriguez, E. H.
“Stability analysis of micropower network,” Transm., Distrib., vol. 11, no. 2, pp. 401–408, Watanabe, and F. Blaabjerg, “Rotor voltage
IEEE Trans. Emerg. Sel. Topics Power Electron., Jan. 2017. dynamics in the doubly fed induction
vol. 4, no. 4, pp. 1299–1309, Dec. 2016. [28] M. Panteli, D. N. Trakas, P. Mancarella, and generator during grid faults,” IEEE Trans.
[14] Y. Yang, P. Enjeti, F. Blaabjerg, and H. Wang, N. D. Hatziargyriou, “Boosting the power Power Electron., vol. 25, no. 1, pp. 118–130,
“Wide-scale adoption of photovoltaic energy: grid resilience to extreme weather events Jan. 2010.
Grid code modifications are explored in the using defensive islanding,” IEEE Trans. Smart [45] W. Chen, D. Xu, N. Zhu, M. Chen, and F.
distribution grid,” IEEE Ind. Appl. Mag., vol. Grid, vol. 7, no. 6, pp. 2913–2922, Nov. 2016. Blaabjerg, “Control of doubly-fed induction
21, no. 5, pp. 21–31, Sep./Oct. 2015. [29] T. Kang, S. Choi, A. S. Morsy, and P. N. generator to ride-through recurring grid
[15] T. N. Preda, K. Uhlen, and D. E. Nordgard, Enjeti, “Series voltage regulator for a faults,” IEEE Trans. Power Electron., vol. 31,
“An overview of the present grid codes for distribution transformer to compensate no. 7, pp. 4831–4846, Jul. 2016.
integration of distributed generation,” in voltage Sag/Swell,” IEEE Trans. Ind. Electron., [46] D. Santos-Martin, J. L. Rodriguez-Amenedo,
Proc. CIRED Workshop, 2012, pp. 1–4. to be published. and S. Arnaltes, “Providing ride-through

1328 Proceedings of the IEEE | Vol. 105, No. 7, July 2017

 Blaabjerg et al. : Distributed Power- Generation Systems and Protec tion

capability to a doubly fed induction generator Rep. (13/96336-43), Jul. 2016. [Online]. [78] J. Li, X.-Y. Ma, C.-C. Liu, and K. P.
under unbalanced voltage dips,” IEEE Trans. Available: [Link] Schneider, “Distribution system restoration
Power Electron., vol. 24, no. 7, pp. 1747–1757, [62] “TR 3.2.2 for PV power plants above 11 kW,” with microgrids using spanning tree search,”
Jul. 2009. [Link], Erritsø, Denmark, Tech. Rep. IEEE Trans. Power Syst., vol. 29, no. 6,
[47] K. O. Kovanen, “Photovoltaics and power (14/17997-39), Jul. 2016. [Online]. Available: pp. 3021–3029, Nov. 2014.
distribution,” Renew. Energy Focus, vol. 14, [Link] [79] F. O. Resende, N. J. Gil, and J. A. P. Lopes,
no. 3, pp. 20–21, May/Jun. 2013. [63] “Technical rule for connecting wind farm to “Service restoration on distribution systems
[48] Y. Xue, K. C. Divya, G. Griepentrog, M. power system,” National Standard using Multi-MicroGrids,” Eur. Trans. Electr.
Liviu, S. Suresh, and M. Manjrekar, Committee China, Tech. Rep. Power, vol. 21, no. 2, pp. 1327–1342,
“Towards next generation photovoltaic (GB/T-19963-2011), Jun. 2012. [Online]. Mar. 2011.
inverters,” in Proc. ECCE, Sep. 2011, Available: [Link] [80] J. M. Solanki, S. Khushalani, and N. N.
pp. 2467–2474. [64] IEEE 929-2000 IEEE Recommended Practice Schulz, “A multi-agent solution to distribution
[49] J. D. van Wyk and F. C. Lee, “On a future for for Utility Interface of Photovoltaic (PV) systems restoration,” IEEE Trans. Power Syst.,
power electronics,” IEEE J. Emerg. Sel. Topics Systems, IEEE Standards Association, 2000. vol. 22, no. 3, pp. 1026–1034, Aug. 2007.
Power Electron., vol. 1, no. 2, pp. 59–72, [65] K. Mentesidi, “The Spanish grid code,” [81] C. P. Nguyen and A. J. Flueck, “Agent based
Jun. 2013. presented at the EUROEUNMED Workshop, restoration with distributed energy storage
[50] M. Braun et al., “Is the distribution grid Grid Code Renew. Energies—Integr. Electr. support in smart grids,” IEEE Trans. Smart
ready to accept large-scale photovoltaic Grid, Jun. 2015. Grid, vol. 3, no. 2, pp. 1029–1038, Jun. 2012.
deployment? State of the art, progress, and [82] Z. Wang and J. Wang, “Self-healing resilient
[66] G. Ding et al., “Adaptive DC-link voltage
future prospects,” Progr. Photovolt., Res. Appl., distribution systems based on
control of two-stage photovoltaic inverter
vol. 20, no. 6, pp. 681–697, Sep. 2012. sectionalization into microgrids,” IEEE
during low voltage ride-through operation,”
[51] C. Whitaker, J. Newmiller, M. Ropp, and B. IEEE Trans. Power Electron., vol. 31, no. 6, Trans. Power Syst., vol. 30, no. 6,
Norris, “Renewable systems interconnection pp. 4182–4194, Jun. 2016. pp. 3139–3149, Nov. 2015.
study: Distributed photovoltaic systems [83] J. Li, X.-Y. Ma, C.-C. Liu, and K. P.
[67] M. S. El Moursi, W. Xiao, and J. L. Kirtley,
design and technology requirements,” Schneider, “Distribution System Restoration
Sandia National Laboratories, Livermore, “Fault ride through capability for grid
interfacing large scale PV power plants,” IET With Microgrids Using Spanning Tree
CA, USA, Tech. Rep. (SAND2008-0946 P), Search,” IEEE Trans. Power Syst., vol. 29,
Feb. 2008. Generat., Transmiss. Distrib., vol. 7, no. 9,
pp. 1027–1036, Sep. 2013. no. 6, pp. 3021–3029, Nov. 2014.
[52] R. Seguin, J. Woyak, D. Costyk, J. Hambrick, [84] J. D. Glover, M. S. Sarma, and T. J. Overbye,
and B. Mather, “High-penetration PV [68] H. M. Hasanien, “An adaptive control
strategy for low voltage ride through Power System Analysis and Design, 4th ed.
integration handbook for distribution New York, NY, USA: Cengage Learn., 2010.
engineers,” National Renewable Energy capability enhancement of grid-connected
Laboratory, Golden, CO, USA, Tech. Rep. photovoltaic power plants,” IEEE Trans. Power [85] Y. Tian, T. Lin, M. Zhang, and X. Xu, “A new
(NREL/TP-5D00-63114), Jan. 2016. Sys., vol. 31, no. 4, pp. 3230–3237, Jul. 2016. strategy of distribution system service
[69] G. Petrone, G. Spagnuolo, R. Teodorescu, M. restoration using distributed generation,” in
[53] B. Zhao, Z. Xu, C. Xu, C. Wang, and F. Lin, Proc. SUPERGEN, Apr. 2009, pp. 1–5.
“Network partition based zonal voltage Veerachary, and M. Vitelli, “Reliability issues
control for distribution networks with in photovoltaic power processing systems,” [86] Y. Xu, C. Liu, K. Schneider, F. Tuffner, and
distributed PV systems,” IEEE Trans. Smart IEEE Trans. Ind. Electron., vol. 55, no. 7, D. Ton, “Microgrids for Service Restoration
Grid, to be published. pp. 2569–2580, Jul. 2008. to Critical Load in a Resilient Distribution
[70] H. Gao, Y. Chen, Y. Xu, and C. C. Liu, System,” IEEE Trans. Smart Grid, to be
[54] Y. Hong, S. N. Pham, T. Yoo, K. Chae, K. H. published.
Baek, and Y. S. Kim, “Efficient maximum “Resilience-oriented critical load restoration
power point tracking for a distributed PV using microgrids in distribution systems,” [87] S. Thale and V. Agarwal, “A smart control
system under rapidly changing IEEE Trans. Smart Grid, vol. 7, no. 6, strategy for the black start of a microgrid
environmental conditions,” IEEE Trans. pp. 2837–2848, Nov. 2016. based on PV and other auxiliary sources
Power Electron., vol. 30, no. 8, pp. 4209– [71] Y. Wang, C. Chen, J. Wang, and R. Baldick, under islanded condition,” in Proc. PVSC,
4218, Aug. 2015. “Research on resilience of power systems 2011, pp. 2454–2459.
[55] M. J. E. Alam, K. M. Muttaqi, and D. under natural disasters—A review,” IEEE [88] M. Soshinkaya, W. H. J. Crijns-Graus, J. M.
Sutanto, “Mitigation of rooftop solar PV Trans. Power Syst., vol. 31, no. 2, pp. 1604–1613, Guerrero, and J. C. Vasquez, “Microgrids:
impacts and evening peak support by Mar. 2016. Experiences, barriers and success factors,”
managing available capacity of distributed [72] A. Gholami, F. Aminifar, and M. Renew. Sustain. Energy Rev., vol. 40,
energy storage systems,” IEEE Trans. Power Shahidehpour, “Front lines against the pp. 659–672, Dec. 2014.
Syst., vol. 28, no. 4, pp. 3874–3884, Nov. 2013. darkness: Enhancing the resilience of the [89] N. K. Choudhary, S. R. Mohanty, and R. K.
[56] T. Miyagawa et al., “Cooperative control of electricity grid through microgrid facilities,” Singh, “A review on microgrid protection,”
reactive power of distributed PV systems to IEEE Electrific. Mag., vol. 4, no. 1, pp. 18–24, in Proc. iEECON, 2014, pp. 1–4.
suppress voltage of distribution line along Mar. 2016. [90] H. AI-Nasseri, M. A. Redfern, and F. Li, “A
railroad track,” in Proc. EPE ECCE-Eur., 2015, [73] M. Panteli and P. Mancarella, “The grid: voltage based protection for micro-grids
pp. 1–7. Stronger, bigger, smarter?: Presenting a containing power electronic converters,” in
[57] J. Seuss, M. J. Reno, M. Lave, R. J. Broderick, conceptual framework of power system Proc. IEEE PES General Meeting, Jun. 2006,
and S. Grijalva, “Advanced inverter controls resilience,” IEEE Power Energy Mag., vol. 13, pp. 1–7.
to dispatch distributed PV systems,” in Proc. no. 3, pp. 58–66, May 2015. [91] D. Q. Hung and N. Mithulananthan,
PVSC, 2016, pp. 1387–1392. [74] T. Gözel and M. H. Hocaoglu, “An analytical “Multiple distributed generator placement in
[58] H. Kobayashi, “Fault ride through method for the sizing and siting of primary distribution networks for loss
requirements and measures of distributed distributed generators in radial systems,” reduction,” IEEE Trans. Ind. Electron., vol. 60,
PV systems in Japan,” in Proc. IEEE PES Electr. Power Syst. Res., vol. 79, no. 6, no. 4, pp. 1700–1708, Apr. 2013.
General Meeting, Jul. 2012, pp. 1–6. pp. 912–918, 2009. [92] M. A. Kashem and G. Ledwich, “Impact of
[59] A. Sangwongwanich, Y. Yang, and F. [75] J. D. Glover, M. S. Sarma, and T. J. Overbye, distributed generation on protection of single
Blaabjerg, “High-performance constant Power System Analysis and Design, 4th ed. wire earth return lines,” Electr. Power Syst.
power generation in grid-connected PV New York, NY, USA: Cengage Learn., 2010. Res., vol. 62, no. 1, pp. 67–80, 2002.
systems,” IEEE Trans. Power Electron., vol. 31, [76] C. L. Moreira, F. O. Resende, and J. A. P. [93] B. Hussain, S. M. Sharkh, S. Hussain, and
no. 3, pp. 1822–1825, Mar. 2016. Lopes, “Using low voltage microgrids for M. A. Abusara, “Integration of distributed
[60] A. Sangwongwanich, Y. Yang, and F. service restoration,” IEEE Trans. Power Syst., generation into the grid: Protection
Blaabjerg, “A sensorless power reserve vol. 22, no. 1, pp. 395–403, Feb. 2007. challenges and solutions,” in Proc. IET DPSP,
control strategy for two-stage grid-connected [77] T. T. H. Pham, Y. Besanger, and N. Hadjsaid, 2010, pp. 1–5.
PV systems,” IEEE Trans. Power Electron., to “New challenges in power system restoration [94] L. Kumpulainen and K. Kauhaniemi,
be published. with large scale of dispersed generation “Analysis of the impact of distributed
[61] “TR 3.2.5 for wind power plants above 11 insertion,” IEEE Trans. Power Syst., vol. 24, generation on automatic reclosing,” in Proc.
kW,” [Link], Erritsø, Denmark, Tech. no. 1, pp. 398–405, Feb. 2009. PSCE, 2004, pp. 10–13.

Vol. 105, No. 7, July 2017 | Proceedings of the IEEE 1329

Blaabjerg et al . : Distributed Power- Generation Systems and Protec tion

  [95] M. Geidl, “Protection of power systems [104] L. Bin, L. Yongli, B. Zhiqian, and A. [114] S. Voima and K. Kauhaniemi, “Using
with distributed generation-state of the Klimek, “Design of protection and control distance protection in smart grid
art,” Ph.D. dissertation, Power Syst. Lab., scheme for microgrid systems,” in Proc. environment,” in Proc. IEEE PES Innov.
Swiss Federal Inst. Technol. (ETH), UPEC, 2009, pp. 1–5. Smart Grid Technol., Oct. 2014, pp. 1–6.
Zürich, Switzerland, 2005. [105] G. Buigues, A. Dysco, V. Valverde, I. [115] M. R. Miveh, M. Gandomkar, S. Mirsaeidi,
  [96] M. Brucoli and T. C. Green, “Fault Zamora, and E. Fernandez, “Microgrid and M. R. Gharibdoost, “A review on
behaviour in islanded microgrids,” in Proc. protection: Technical challenges and protection challenges in microgrids,” in
CIRED, 2007, pp. 21–24. existing techniques,” Renew. Energy Power Proc. EPDC, 2012, pp. 1–5.
Quality J., vol. 6, no. 11, pp. 222–227, 2013. [116] Y. Han, X. Hu, and D. Zhang, “Study of
  [97] H. Al-Nasseri, M. A. Redfern, and F. Li, “A
voltage based protection for micro-grids [106] H. Lin, C. Liu, J. M. Guerrero, and J. C. V. adaptive fault current algorithm for
containing power electronic converters,” in Quintero, “Distance protection for microgrid dominated by inverter based
Proc. IEEE PES General Meeting, Jun. 2006, microgrids in distribution system,” in Proc. distributed generators,” in Proc. PEDG,
pp. 1–7. IECON, 2015, pp. 731–736. 2010, pp. 852–854.
  [98] C. Hou and X. Hu, “A study of voltage [107] H. H. Zeineldin, E. F. El-Saadany, and M. [117] K. Dang, X. He, D. Bi, and C. Feng, “An
detection based fault judgement method in M. A. Salama, “Distributed generation adaptive protection method for the inverter
micro-grid with inverter interfaced power micro-grid operation: Control and dominated microgrid,” in Proc. ICEMS,
source,” in Proc. Int. Conf. Electr. Eng., 2009, protection,” in Proc. PSC, 2006, pp. 105–111. 2011, pp. 1–5.
pp. 1–5. [108] E. Sortomme, S. S. Venkata, and J. Mitra, [118] M. Khederzadeh, “Adaptive setting of
“Microgrid protection using protective relays in microgrids in grid-
  [99] T. Loix, T. Wijnhoven, and G. Deconinck, communication assisted digital relays,”
“Protection of microgrids with a high connected and autonomous operation,” in
IEEE Trans. Power Del., vol. 25, no. 4, Proc. IET DPSP, 2012, pp. 1–4.
penetration of inverter-coupled energy pp. 2789–2796, Oct. 2010.
sources,” in Proc. IEEE PES/CIGRE Symp., [119] N. Schaefer, T. Degner, A. Shustov, T. Keil,
Jul. 2009, pp. 1–6. [109] S. Kar, S. R. Samantaray, and M. D. Zadeh,
“Data-mining model based intelligent and J. Jaeger, “Adaptive protection system
[100] A. R. Bergen, Power System Analysis. differential microgrid protection scheme,” for distribution networks with distributed
Englewood Cliffs, NJ, USA: Prentice-Hall, IEEE Syst. J., to be published. energy resources,” in Proc. IET DPSP, 2010,
1986. pp. 1–5.
[110] E. Sortomme, J. Ren, and S. S. Venkata, “A
[101] H. Nikkhajoei and R. H. Lasseter (Dec. differential zone protection scheme for [120] H. Laaksonen, D. Ishchenko, and A.
2006). Microgrid Fault Protection Based on microgrids,” in Proc. IEEE PES General Oudalov, “Adaptive protection and
Symmetrical and Differential Current Meeting, Jul. 2013, pp. 1–5. microgrid control design for Hailuoto
Components. [Online]. Available: http:// island,” IEEE Trans. Smart Grid, vol. 5,
[111] M. Dewadasa, R. Majumder, A. Ghosh, and no. 3, pp. 1486–1493, May 2014.
[Link]/2009publications/ G. Ledwich, “Control and Protection of a
CEC-500-2009-004/CEC-500-2009-004- Microgrid with Converter Interfaced Micro [121] A. Oudalov and A. Fidigatti, “Adaptive
[Link] Sources,” in Proc. ICPS, 2009, pp. 1–6. network protection in microgrids,” Int. J.
[102] S. M. Mirsaeidi, D. W. Said, M. H. [112] M. Dewadasa, “Protection for distributed Distrib. Energy Resour., vol. 5, no. 3,
Mustafa, M. Habibuddin, and K. Ghaffari, generation interfaced networks, Electrical pp. 201–226, 2009.
“Review and analysis of existing protection Engineering,” Ph.D. dissertation, Faculty [122] N. Jayawarna, C. Jones, M. Barnes, and N.
strategies for micro-grids,” J. Electr. Syst., Built Environ. Eng., Queensland Univ. Jenkins, “Operating microgrid energy
vol. 10, no. 1, pp. 1–10, 2014. Technol., Brisbane, QL, Australia, 2010. storage control during network faults,” in
[103] E. Sortomme, S. S. Venkata, and J. Mitra, [113] W. Huang, N. Tai, X. Zheng, C. Fan, X. Yang, Proc. IEEE Int. Conf. Syst. Syst. Eng. SoSE,
“Microgrid protection using and B. J. Kirby, “An impedance protection Apr. 2007, pp. 1–7.
communication assisted digital relays,” scheme for feeders of active distribution [123] F. van Overbeeke, “Fault current source to
IEEE Trans. Power Del., vol. 25, no. 4, networks,” IEEE Trans. Power Del., vol. 29, ensure the fault level in inverter-dominated
pp. 2789–2796, Oct. 2010. no. 4, pp. 1591–1602, Aug. 2014. networks,” in Proc. CIRED, 2009, pp. 1–4.

ABOUT THE AUTHORS

Frede Blaabjerg (Fellow, IEEE) received the Ph.D. Yongheng Yang (Member, IEEE) received the
degree in electrical engineering from Aalborg [Link]. degree in electrical engineering and auto-
University, Aalborg, Denmark, in 1995. mation from Northwestern Polytechnical Univer-
He was with ABB-Scandia, Randers, Den- sity, Shaanxi, China, in 2009 and the Ph.D. degree
mark, from 1987 to 1988. He became an Assis- in electrical engineering from Aalborg University,
tant Professor in 1992, Associate Professor in Aalborg, Denmark, in 2014.
1996, and Full Professor of power electronics and He was a postgraduate student with South-
drives in 1998 at Aalborg University. His current east University, Jiangsu, China, from 2009 to 2011.
research interests include power electronics and In 2013, he was a Visiting Scholar at Texas A&M
its applications such as in wind turbines, PV systems, reliability engineer- University, College Station, TX, USA. Since 2014, he has been with the
ing, power quality, and adjustable speed drives. Department of Energy Technology, Aalborg University, where currently
Prof. Blaabjerg has received 18 IEEE Prize Paper Awards, the IEEE he is an Assistant Professor. He has published more than 100 technical
PELS Distinguished Service Award in 2009, the EPE-PEMC Council Award papers and coauthored a book Periodic Control of Power Electronic Con-
in 2010, the IEEE William E. Newell Power Electronics Award 2014 as well verters (London, UK: IET). His research includes grid integration of renew-
as the Villum Kann Rasmussen Research Award 2014. He was the Editor- able energies, power electronic converter design, analysis and control,
in-Chief of the IEEE Transactions on Power Electronics from 2006 to 2012. and reliability in power electronics.
He was nominated by Thomson Reuters as one of the 250 most cited Dr. Yang is a Member of the IEEE Power Electronics Society (PELS)
researchers in engineering in the world in 2014, 2015, and 2016. In 2017, Students and Young Professionals Committee. He served as a Guest Asso-
he became Doctor Honoris Causa at the University of Politehnica, Tim- ciate Editor of IEEE Journal of Emerging and Selected Topics in Power Electronics
isoara, Romania. and a Guest Editor of Applied Sciences. He is an Associate Editor of CPSS
Transactions on Power Electronics and Applications.

1330 Proceedings of the IEEE | Vol. 105, No. 7, July 2017

 Blaabjerg et al. : Distributed Power- Generation Systems and Protec tion

Dongsheng Yang (Member, IEEE) was born in in the Department of Energy Technology. His research interests include
Jiangsu Province, China, in 1984. He received the modeling and control of grid-connected converters, harmonics analysis
B.S., M.S., and Ph.D. degrees in electrical engi- and control, passive and active filters, and stability of power-electronic-
neering from Nanjing University of Aeronautics based power systems.
and Astronautics (NUAA), Nanjing, China, in 2008, Dr. Wang serves as an Associate Editor for the IEEE Transactions on
2011, and 2016, respectively. Power Electronics, the IEEE Transactions on Industry Applications, and the IEEE
Since 2016, he has been with Aalborg Uni- Journal of Emerging and Selected Topics in Power Electronics. He received the
versity, Aalborg, Denmark, where he is currently second prize paper award and the outstanding reviewer award of the IEEE
a Postdoctoral Researcher in the Department Transactions on Power Electronics in 2014 and 2016, respectively, and the
of Energy Technology. His main research interests include design and best paper award at IEEE PEDG 2016.
control of grid-connected inverters, harmonic analysis and mitigation of
power-electronic-based power systems, and online impedance measure-
ment techniques.

Xiongfei Wang (Member, IEEE) received the B.S.

degree from Yanshan University, Qinhuangdao,
China, in 2006, the M.S. degree from Harbin Insti-
tute of Technology, Harbin, China, in 2008, both
in electrical engineering, and the Ph.D. degree in
energy technology from Aalborg University, Aal-
borg, Denmark, in 2013.
Since 2009, he has been with Aalborg Univer-
sity, where he is currently an Associate Professor

Vol. 105, No. 7, July 2017 | Proceedings of the IEEE 1331