Renewable and Sustainable Energy Reviews 67 (2017) 988–997

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

journal homepage: www.elsevier.com/locate/rser

A review on issues and approaches for microgrid protection

Belwin J. Brearley n, R. Raja Prabu
EEE Department, B.S. Abdur Rahman University, Chennai, India

art ic l e i nf o a b s t r a c t

Article history: In a traditional power grid, power generation is done at various potential locations and transmitted into
Received 30 September 2015 power grid and then distributed to the customer premises. The traditional grid incurs heavy investment,
Received in revised form limited reliability, increased emissions of green house gases and increased transmission line losses. It has
31 March 2016
made the utility to opt for connecting numerous renewable based micro sources near the customer
Accepted 9 September 2016
premises, as per their requirement and providing intelligent control for the grid. As a comprehensive
solution, microgrids are suggested by the researchers which would provide reliable, quality and efficient
Keywords: supply to its customers. The connection of microgrid in the existing distribution network makes the
Distributed generation radial network more complicated. It also causes the magnitude of fault current to change dynamically
Fault currents
depending on the modes of operation (grid connected or islanded mode), type of distributed generator,
Microgrid control
status of distributed generators and number of distributed generators. The conventional protection
Renewable energy
Fault current limiter schemes are designed for radial power flow with centralized power generation, which makes the ex-
Inverter isting protection scheme to fail with the microgrid. Hence, an attempt is made to relook into the basic
Communication standards concepts and significance of microgrid, issues faced by the microgrid in the context of protection and
Power distribution protection various protection strategies.
Energy storage & 2016 Elsevier Ltd. All rights reserved.
Over current protection

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 989
2. Microgrid control. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 989
2.1. Centralized control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 989
2.2. Decentralized control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 990
3. Challenges in microgrid protection systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 990
3.1. Dynamics in fault current magnitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 990
3.2. LOM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 990
3.3. Unnecessary disconnection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 991
3.4. Blinding of protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 991
4. Microgrid protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 991
4.1. Current limiter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 992
4.2. Centralized protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 992
4.2.1. Adaptive protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 992
4.3. Protection based on variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 994
4.3.1. Current samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 994
4.3.2. Wavelet packet transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 994
4.3.3. Node voltages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 994
4.3.4. Traveling wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 994
4.3.5. Local variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 994
4.4. Distance protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 995

n
Corresponding author.

http://dx.doi.org/10.1016/j.rser.2016.09.047
1364-0321/& 2016 Elsevier Ltd. All rights reserved.
 B.J. Brearley, R.R. Prabu / Renewable and Sustainable Energy Reviews 67 (2017) 988–997 989

4.5. Multi agent protection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 995

5. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 996
6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 996
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 996

1. Introduction i.Bidirectional power flow in the distribution network

ii.Dynamic characteristics of micro sources
Nowadays, the demand for electrical energy is continuously iii.Limitation of fault current during islanded mode
growing. In order to meet the increasing demand several power iv. Topological changes in the network due to the intermittent
plants have to be constructed. But the traditional power plants nature of the micro sources
have lot of draw backs such as carbonic emissions, low efficiency, v. Variation in fault current with respect to mode of operation
high transmission losses, higher construction & fuel cost, more vi. Type of DG (inverter fed or direct fed DG)
construction time and reduced reliability [1]. Hence, researcher’s vii. Number of DG used.
thought of connecting several small generators called micro
sources/ Distributed Generators (DGs) (o500 kW) such as gas Hence, proper protection scheme is very much essential for the
turbines, solar power plants, wind, fuel cell, microturbine, small microgrid to ensure safe operation both during grid connected and
hydro electric power plant, Combined Heat and Power [2] as islanded mode.
clusters, near the Low Voltage (LV) side to meet the required de- Microgrid pilots have been successfully implemented in various
mand, thus evolving a new concept known as “Microgrid”. developed countries such as Japan (Akagi – includes solar cell and
Microgrid refers to cluster of micro sources along with the wind turbine, Aichi – includes photo voltaic, fuel cell and com-
storage unit, which operates as a single controllable unit to meet bined heat and power sources), USA (CERTS - the Consortium for
the local power demands. As per the utility, the microgrid is a Electric Reliability Technology Solutions, U.S. Army microgrid –
controlled cell of power system network which responds to the includes combined heat and power sources), Korea (KERI -Korean
transmission network. As per the customer, it is a system which is Energy Research Institute – includes photo voltaic simulator, fuel
cell, diesel generator and wind turbine simulator), Spain (Malaga
specially designed to provide reliable, efficient and quality power
Smartcity), Finland (Hailuoto – includes diesel generator and wind
supply with the help of local micro source controller, system op-
power generator) and Germany (Mannheim field test).
timizer and distributed protection unit [3].
The need for microgrid control and the strategies of control are
The power generation from micro sources may be either a Di-
discussed in Section 2. The various issues faced in microgrid pro-
rect Current (DC) or Alternating Current (AC) supply, depending on
tection are presented in detail in Section 3. The possible protection
the type of micro sources used. In case of AC power, the generated
strategies suggested for the microgrid are looked in Section 4.
power is rectified to DC. The DC power is then integrated to the
Sections 5 and 6 provide the discussion and conclusion respectively.
grid through Voltage Source Inverter (VSI) controlled by Pulse
Width Modulation (PWM) technique. With the aid of PWM the
output voltage and angle of the VSI can be controlled. The output
2. Microgrid control
of the VSI is interfaced to the grid through an inductor at the Point
of Common Coupling (PCC). The voltage & phase angle at VSI, re-
The variables to be controlled for a microgrid are voltage, fre-
actance offered by inductor and the voltage & phase angle at PCC
quency, active power and reactive power, which mainly depends on
decides the real and reactive power flow from the micro source to
the mode of operation. During grid connected mode, the utility grid
the utility grid. maintains voltage and frequency of microgrid. Hence, the main
In a conventional power system network, when a new load is responsibility of the microgrid controller is to balance the real and
suddenly added then inertia of the system compensates the initial reactive power. During islanded mode, in addition to real and re-
energy balance. But in case of micro sources, since the inertia is active power balance, the microgrid controller should also take care
very low, an efficient battery storage system is essential to com- of voltage and frequency. The power balance is either carried out by
pensate the same. Local Controller (LC) alone or with the help of set points given from
The two modes of operation of microgrid are grid connected central controller through communication devices.
mode and islanded mode [4]. During grid connected mode, the These control strategies are carried out with the help of either
microgrid receives power from both the utility and from the micro centralized or decentralized supervisory control that includes (i) Market
source. Under grid connected mode, major portion of the real Operator (MO) (ii) Microgrid Central Controller (MCC) (iii) LC associated
power required for the load is met by the DGs connected to the with (a) DG units (LCDG) (b) load (LCload) (c) storage unit [10–12].
microgrid and the remaining few portion and the variation in the LCload is responsible to maintain reliable supply to its customers
real power demand is met by the grid [3,5]. During islanded mode, whereas and LCDG look after generation requirement for micro-
load/ generation shedding is carried out to maintain power balance. grid. Based on the technical and economical aspects set points for
The critical loads are made to receive quality power at all time and LCload and LCDG are provided by the MCC in case of Centralized
the remaining loads are made to undergo load shedding [3,6]. control. The Distribution Network Operator (DNO) is used for
Protection plays a vital role for a reliable power system net- places where more than one microgrid exists. It is responsible for
work. The ultimate objective behind the microgrid is to provide the market functions of each specific area.
reliable supply to its customers. Hence for a fault in the utility,
microgrid should get isolated [7,8] and for a fault inside the mi- 2.1. Centralized control
crogrid, it should isolate the minimum faulty part. The traditional
protection schemes designed for radial flow with high fault cur- In centralized control, the MCC updates the look up table with
rent will not operate faithfully to an microgrid [9] for the following recent information for every predefined interval of time with re-
reasons. spect to
990 B.J. Brearley, R.R. Prabu / Renewable and Sustainable Energy Reviews 67 (2017) 988–997

i. DG - status of DG (ON/OFF), type of DG, amount of generation. During islanded mode of operation voltage and frequency of
ii. Load - demand the microgrid is controlled by making one of the inverter fed DG as
iii. Network parameters such as current, voltage at each relay the master unit, which through its control strategy maintains the
point based on the control requirement voltage and frequency. The remaining DGs operate for power
iv. PCC- mode of operation (grid connected/islanded mode) balance with real and reactive power droop. The command for
load shedding is dictated to the LC by MCC.
The central controller involves two major modules namely The drawback in a fully decentralized control system is that
Energy Management Module (EMM) and Protection Coordination since each unit is controlled by its LC based on local information,
Module (PCM). EMM is responsible to provide the set points for the controller is not aware of system-wide variables and other
active, reactive power, voltage, frequency to each LC through local controllers’ actions. In order to overcome this drawback it is better
communication network based on IEC 61850 standard. During grid to have a hybrid controller, in which the decisions taken by the LC
connected mode the central controller. is rechecked by the MCC considering the system wide variables. If
the MCC detects any lack in control, then it immediately sends the
i. Monitor system parameters (above discussed parameters from modified control commands to the LC.
DG, Load, bus whichever parameters are essential for the spe-
cified control)
ii. Schedules economic generation of active and reactive power 3. Challenges in microgrid protection systems
control of the DGs and also demand side management.
iii. Ensure synchronized operation of microgrid with the utility To ensure reliable, safe operation of the power system network
network. proper protective devices with better selectivity, fast operation,
simplicity, flexibility, different setting opportunities and low price
The functions of the central controller in the islanded mode are has to be chosen.
As discussed earlier, the traditional protection schemes de-
i. Control commands for (a) inverter connected to master DG with signed for radial flow with high fault current for distribution
voltage and frequency control (b) inverter connected to other network will not operate faithfully to an microgrid because of
DGs with real and reactive power balance of microgrid.
bidirectional power flow, dynamic characteristics of DGs, inter-
ii. Provide commands to load controllers to undergo load shedding
mittent nature of the DG and variation in fault current. The major
based on the load criteria, microgrid energy reserves.
challenges faced in the protection [15] of microgrid are

The PCM take care of protection coordination of the microgrid

i. Dynamics in fault current magnitude
in case of fault in microgrid and utility grid. In general, the cen-
ii. Loss of Mains (LOM)
tralized protection scheme applies adaptive algorithms for mi-
iii. Unnecessary disconnections (lack of selectivity)
crogrid protection in order to incorporate the changes in fault
iv. Blinding of protection
current due to the dynamic change in the microgrid network.
The implementation of fully centralized control of microgrid
3.1. Dynamics in fault current magnitude
requires extensive communication between the central controller
and controlled units which is especially infeasible for large geo-
The connection of DG in the LV network changes the fault level
graphic areas. A small fault in the communication system affects
to a considerable level mainly because of two major modes of
the entire designed control mechanism to fail [13].
operation namely grid connected mode and islanded mode. In grid
connected mode, the fault current will be very high because both
2.2. Decentralized control
the utility and the DGs present inside the microgrid feed the fault.
In remote areas, wherever distances between the DGs are long, But during islanded mode the fault current is very low, since the
the implementation of centralized control needs more commu- only source in the microgrid is low capacity DGs.
nication link which warrants more cost. In such cases, decen- Moreover the fault current contributed by DG varies with re-
tralized microgrid control can be better implemented. The de- spect to the type of DG. Fault current contributed by synchronous
centralized control is also preferred in microgrid with more type DG is 5 times the rated current and that by the inverter fed
number of DGs which in turn is governed by different owners. DG is 1.5 times the rated current [16]. Also the renewable source
In decentralized control, more emphasis is given to LC so as to based DGs are highly intermittent in nature, hence such type of DG
maximize autonomy of the DGs and loads. Similar to centralized contribute to fault only if it is in ON condition. Thus the magnitude
control, here the autonomy is achieved using a hierarchical of the fault current keep on changing depending on the mode of
structure which includes DNO, MCC and LC. The main objective of operation, type of DGs and the number of DGs. Hence, it is difficult
LC is to (i) maximize the production in order to satisfy the demand to predict the fault current accurately [17].
(ii) provide the maximum possible export to the distribution
network (iii) improve the performance of microgrid. Here, each DG 3.2. LOM
is individually controlled for power sharing with inverter control.
During grid connected mode, voltage and frequency of the mi- Loss of mains refers to the disconnection of microgrid from the
crogrid is maintained by the utility. The DG is designed with fre- utility source, but it remains connected with part of the load in
quency droop characteristics for real power sharing and with utility. This is caused due to (a) fault in utility grid (b) problem in
voltage droop for reactive power sharing (Fig. 1) [9,14] as circuit breaker connecting with utility source. During such unin-
f¼fmax  m (P  Pmax) and V ¼Vmax n (Q-Qmax) respectively, tentional islanding, since a part of the network is electrified and
where fmax and Vmax are the maximum frequency and voltage of the islanding is undetected, hence the fault attending person is in
DG unit; m and n are the coefficients of frequency droop and danger [18]. Also this process results in uncontrolled voltage and
voltage droop characteristics; Pmax and Qmax refers to maximum frequency, risks to plants available in the islanded system, un-
real and reactive power of DG unit; P and Q refers to instantaneous synchronized reclosures which ultimately results in the damage of
real and reactive power of DG unit. customer equipment and measuring devices.
 B.J. Brearley, R.R. Prabu / Renewable and Sustainable Energy Reviews 67 (2017) 988–997 991

Fig. 1. Real and reactive power droop characteristics.

Vth
Ik =
3 Zth (1)

where Vth is the pre-fault voltage at the faulted point and Zth is the
Thevenin's impedance..
Let Zg, Zs, ZL represent the impedance of DG, utility source and
transmission line respectively, then the thevenin’s equivalent cir-
cuit of the network can be represented as Fig. 4. Thevenin’s im-
pedance is calculated as
(Zs + l. ZL ). Zg
Zth = + (1 − l)ZL
(Zs + l. ZL + Zg ) (2)

Thus the thevenin’s impedance offered at the faulted point is

increased due to the additional impedance offered by DG.
The fault current contributed from the grid is
.Zg
Igrid = . IK
(Zs + l. ZL ) + Zg (3)

Fig. 2. Unnecessary disconnection. The contribution of the fault current by the utility is nonlinear
with location and size of DG. Hence in an microgrid when fault
3.3. Unnecessary disconnection occurs at the lower end of the feeder, the grid impedance will be as
large as DG impedance and as a result the short circuit current
When a DG is located close to the substation in a feeder and if it remains well below the pickup current of the feeder relay in the LV
contributes for a fault occurring nearer to substation in a neigh- network which makes the relay to undetect fault and consequently
boring feeder (Fig. 2), then in such case the major portion of the this leads to malfunctioning of the entire protection system [19].
fault current is contributed by the DG. This makes the current to In order to overcome the above said issues, proper protection
exceed the pickup value of the relay available in the healthy feeder strategy has to be evolved for the microgrid to with stand the above
(Relay1) and hence it get tripped before operation of the relay said challenges and to operate efficiently. The subsequent section
corresponding to the faulted feeder (Relay2) [19]. reviews the various protection schemes available for microgrid.

3.4. Blinding of protection

4. Microgrid protection
In a microgrid when fault occurs at the lower end of the feeder
both the utility and the DG contribute to the fault current. Now, A proper protection scheme for the microgrid is to be designed,
the Thevenin's impedance offered at the faulted point is increased so that in the event of any fault, a minimum portion of the system
compared to the traditional network due to the additional im- is isolated, without affecting the rest of the system. This can be
pedance offered by DG. In Fig. 3, a small DG is connected at a done by the combination of a primary and back up protective
distance d. For a fault located at a distance l, the maximum fault devices. The primary protective device is responsible to act for any
current in each phase is calculated as fault within its zone. In case, if the primary protective device fails

Fig. 3. Blinding of protection. Fig. 4. Thevenin's equivalent circuit.

992 B.J. Brearley, R.R. Prabu / Renewable and Sustainable Energy Reviews 67 (2017) 988–997

Fig. 5. Classification for microgrid protection.

to operate, then the backup device gets into operation. The con- series inverter, thereby the current is being limited appropriately.
ventional system is radial and hence the protection scheme is The insertion of series resistance causes more active power loss;
designed with over current relay and fuses which are very simple. hence a compromise has to be carried out for the selection of R/L
As discussed earlier the introduction of DGs in the existing system ratio. The RL feed forward algorithm functions well, but it circulates
makes the power system network more complicated. Also the fault more active power between the shunt and series inverter. Hence, a
current is unpredictable, since it changes with the mode of op- flux charge model feedback algorithm is used to reduce the active
eration, type of DG and number of DG. Hence, designing an effi- power circulation and to insert a virtual inductance through the
cient protection scheme to cope up with the above said issues is a series inverter using proper control algorithm.
challenging task. The various protection schemes addressed is The fault current limiters are to be optimally located so as to
tabulated (Fig. 5) and the same is discussed. minimize the total co-ordination time involved in all the relays,
the total operating time of the relay, irrespective of the mode of
4.1. Current limiter operation [22].

The Fault Current Limiters (FCL) are placed near the PCC to limit
4.2. Centralized protection
the fault current, contributed by the utility grid to the microgrid
and by the DGs in the microgrid towards the utility. The FCL is kept
In centralized protection, a microgrid has one central unit. The
in minimum position under normal condition (to avoid unwanted
LV microgrid is connected to the MV network using a delta/star
voltage drops and losses) and in maximum position under faulty
transformer. The neutral in star helps to avoid earthing problems
condition [20,21].
during islanding. Four Protection Devices (PD1, PD2, PD3 and PD4)
The fault current limiters can be broadly classified into three
are generally used (Fig. 6). PD1 (voltage and frequency relay) is lo-
types namely, super conducting FCL, solid state FCL, electro-
cated at the PCC, PD2 (directional over current relay) in the feeder,
magnetic FCL. In superconducting type FCL, the current limiting
PD3, PD3a (non directional over current relay) in service connec-
behavior depends on the non linear response of the super-
tions, PD3b (over current protection with fuses) and PD4 (voltage
conducting materials to temperature, current and magnetic fields.
and frequency relay) in DGs. A Microgrid Management System
Hence under faulted condition with heavy fault, any one of these
parameters exceeds the predefined value [21], in such cases cur- (MMS) is used to monitor the status of the microgrid and to set the
rent limiting is done. Under normal condition the current is below rating of the protective devices accordingly. The protective devices
the threshold, hence, the entire current is given to the protective PD1 and PD2 must be fast enough during islanded operation.
devices. In solid state FCL the current limiting behavior is decided The communication between the protective devices, is to be done
based on the ON/OFF of the semiconductor devices. In electro- using the standards IEC 61850 [24]. The operating curve for the
magnetic FCL, the current limiting behavior is decided based on protecting devices for both normal and islanded operation is being
the variation of the magnetic field. set. The MMS sends the status of the microgrid (normal or islanded)
A hybrid superconducting fault current controller [22] is a to the protecting devices. Based on the status of the microgrid, the
combination of super conducting FCL, solid state FCL, electro- protecting devices compare the measured parameter with the op-
magnetic FCL. This is one of the efficient methods suggested to erating curves and correspondingly issues trip signal [25].
limit and control the fault current in the microgrid.
During voltage sag, more line current flows through the line 4.2.1. Adaptive protection
impedance interconnecting the micro and utility grid [23] which It is to be noted that, the introduction of the DGs through the
in turn causes more line current to flow towards the load which microgrid in the existing power system network changes the fault
may damage the power electronic switches in the inverter. Hence current magnitude and fault level. Hence, the power rating of the
to protect the entire circuitry [16] the voltage is set to pre-fault protection devices also has to be revised (increased appropriately).
voltage using shunt inverters with appropriate control loops so as Also the status of the DG, mode of operation and other network
to limit the fault current. changes is to be updated instantaneously to central controller [25–
Later along with the shunt inverter, a series inverter is inserted 28] to devise proper protection scheme. Such type of protection is
and two current limiting algorithms are proposed to limit the cur- adaptive Microgrid protection. Adaptive microgrid protection is
rent during voltage sag [23]. The current limiting algorithms are RL suggested based on over current, differential and on sequence
feed forward current limiting algorithm and flux charge model components. Adaptive protection is done effectively by a commu-
feedback algorithm. During voltage sag at PCC, the voltage at the nication scheme [29,30] working with the help of Transmission
microgrid end is lifted by inserting impedance in series through the Control Protocol/Internet Protocol (TCP/IP) based ethernet network.
 B.J. Brearley, R.R. Prabu / Renewable and Sustainable Energy Reviews 67 (2017) 988–997 993

Fig. 6. Centralized protection.

4.2.1.1. Adaptive over current protection. In adaptive over current communication for instantaneous data transfer between terminals of
protection, Microgrid Central Processing Unit (MCPU) forms all the protected element. Differential protection works on the principle
possible meaningful configurations of the microgrid network of Kirchoff’s Current Law (KCL). The time stamped current values is
along with the DG status and store it in an structure called event received from the either end relays of the feeder through commu-
table. Then, for each configuration, fault current seen by each relay nication link. The current at two ends is compared [28], if the current
for all possible fault locations is stored in another structure called difference exceeds the threshold limit then the trip signal will be
fault current table. For each set of configurations, the relay settings provided. For proper protection, the restraint and operating slopes are
with hierarchy of operation of the relay for any particular fault properly designed to avoid the mismatch caused due to the inclusion
with its time delay is also stored in another structure called action of current transformer [34]. The over current protection is kept as the
table [16,26,31,32]. backup protection to improve the reliability [26,27].
For a fixed interval of time the status of the DG and the grid But, the dynamics of microgrid (transition between grid con-
configuration (grid connected or islanded) and current in each relay nected mode and islanded mode) sometimes makes the differ-
location is updated in MCPU periodically. With the help of the event ential scheme to fail. Therefore to make the differential scheme
table, fault current table and action table, appropriate signal will be effective even under transient conditions, advanced differential
given to the appropriate relays in the specific time by the MCPU. protection scheme, based on wide area monitory and high fre-
Because of some reasons if the relay fails to operate, then the relay quency current samples can be applied [35–38].
immediately upstream to the corresponding relay (it should operate
after some back up time based on the action table). 4.2.1.3. Adaptive protection based on sequence components. In the
Similarly, if a fault occurs in the grid, the DGs will contribute adaptive over current and differential protection, all the relays are
the fault current to the grid. Under such situation, the upstream devised with communication assistance which will be costlier.
relay which connects the microgrid to the main grid will operate Hence, to reduce the cost [39] microgrid is devised with a Mi-
to isolate the microgrid from the main grid. If it fails, the sub- crogrid Protection Commander (MPC), several Communication
sequent downstream relay operates which is just the reverse of assisted Microgrid Protection Relays (CMPR) and several Microgrid
the previous case (fault inside microgrid) [33]. Hence, the hier- Protection Relays (MPR) as shown in Fig. 7.
archy of operation of the relay will be reversed which is also stored Each MPR/CMPR has five modules namely the (i) directional
in the MCPU in separate table. In order to distinguish the fault module (ii) islanded protection module (iii) grid connected pro-
inside the microgrid and outside (main grid) the microgrid the tection module (iv) interfacing module (v) trip module. The di-
MCPU determines the direction of current in the PCC. If the di- rectional module determines the power flow direction and also
rection is forward it assumes the fault to be inside the microgrid calculates fault ratio (indicates change in sequence components
and follows the corresponding hierarchy table and if the current from normal condition to fault condition). The grid connected and
direction is reversed then it considers that the fault is in the utility islanded protection module is designed in such a way so as to offer
and it follows the relay sequence operation accordingly. protection in the specified modes respectively. The interfacing
module has a neutral voltage displacement and synchronism
4.2.1.2. Adaptive differential protection. The Differential protection check function. The neutral voltage displacement function is used
relays are generally used to protect equipment such as distributed to detect fault on the grid side to activate islanded operation. The
generators, transformers and short transmission lines. The differential synchronism check function is used to ensure proper reconnection
protection operates well for an internal fault. But, it requires a reliable of the small grid back to the main grid.
994 B.J. Brearley, R.R. Prabu / Renewable and Sustainable Energy Reviews 67 (2017) 988–997

Fig. 7. Adaptive protection based on sequence components.

The microgrid network is divided in several smaller areas and a difference in spectral energy is set as 1 and in case of islanded
CMPR is placed in each area to look after the protection in that mode the threshold is set as  0.3 [41,42].
particular zone. Several MPR are also installed within this zone.
The difference between CMPR and MPR is that the CMPR can 4.3.2. Wavelet packet transform
communicate with the MPC, but the MPR don’t have such com- Wavelet packet transform is used to extract, parameterize and
munication technology. In addition to this fuses, reclosures and localize all frequency contents in a processed signal. It decouples
sectionalizes are also available to offer better protection [29,40]. signal into time and frequency localized signals that parameterize
The CMPR sends the fault detection signal (whether it has iden- the frequency contents of the processed signal. Since Power elec-
tified any fault or not) and the direction of the fault to the MPC. The tronic controllers are used, the three phase quantity is converted
MPC receives such information’s from all the CMPRs within its net- into dq frame and then WPT is applied to the dq signal [43].
work and then judges the region in which the fault has occurred Minimum Description Length (MDL) data criteria is used for se-
based on the received information. The MPC again waits for a while lecting optimal wavelet basis function and is found that Doube-
to receive the recent information regarding the direction. Based on chies db4 wavelet function as optimal.
the recent information, it further verifies the faulted part and finally The high frequency sub bands WPT coefficient of dq axis
sends the tripping signal to the corresponding CMPR and wait signal component contains signatures which can facilitate detecting and
for all other CMPRs. In case, if the corresponding CMPR fails to op- classifying transient disturbances as fault or non fault. For doing so
erate, then after some time delay the circuit breaker issues a failure current samples, ia, ib, ic are collected and converted into dq
tripping signal to its neighboring circuit breakers. After a pre-calcu- component. Xdq ¼ Id2 þIq2, if idq≠0, then it identifies as some fault
lated time delay, the neighboring circuit breaker gets tripped off, has occurred and vice versa [43,44].
thereby it provides both primary and back up protection.
Once a fault is reported in the small microgrid by the MPC, the 4.3.3. Node voltages
corresponding sub network gets isolated from the remaining mi- In this method, the node voltages are set as state variables. If
crogrid network. If fault further persist, then the sub-network
the residue between the measured voltage value and the state
waits for a while whether the MPR inside its network gets acti-
estimated voltage value is very high then it implies the presence of
vated. If the MPR is able to locate and isolate the fault then the
fault, if not, the line is said to be healthy. The state variables are
remaining area of the sub-network remains working autono-
estimated for a sub-area. Hence communication required in this
mously. If not the entire sub-network will be kept in off condition
protection scheme is less [45].
till the clearance of fault. After the clearance of fault the sub-
network will be synchronized back to the microgrid network.
4.3.4. Traveling wave
The traveling waves are extracted by Mathematical Morphol-
4.3. Protection based on variables ogy Filters (MMF) which develops apparent feature of dilation and
erosion signals. Two wave fronts are extracted (one at first half of
The Microgrid protection can also be done based on various the line and the other at the second half of the line) and its time
parameters such as samples of current, voltage, angles, traveling and polarity are stored. If fault occurs in first half then product of
wave, Wavelet Packet Transform (WPT) and Total Harmonic Dis- the polarities of the two wave fronts has the positive value and if it
tortion (THD) which is discussed in detail below. lies in the second half then it will be negative [4].

4.3.1. Current samples 4.3.5. Local variables

In this method, current samples are collected from either ends The local variables such as rms voltage, rms current, voltage
of the feeder and the spectral energy of both the current samples THD, current THD, power angle, symmetrical voltage component,
are calculated. If the difference in current spectral energy exceeds symmetrical current component for all the three phases (21 vari-
the threshold limit, then trip signal will be issued. During grid ables) are processed using feature extraction [46,47]. Feature se-
connected mode for LG (Line-Ground), LLG (Line-Line-Ground) lection is done in each relay in such a way to eliminate the un-
and LLLG (three phase fault) fault the threshold value for wanted attributes. Based on the information gain the feature
 B.J. Brearley, R.R. Prabu / Renewable and Sustainable Energy Reviews 67 (2017) 988–997 995

ranking is done. It is found that the current samples have much

impact towards the fault rather than the voltage and power angles.
The above said 21 variables calculated in each relay is send to
all the other relays through Ethernet communication channel.
Then each relay manipulates differentiation between the local and
remote variables and stores in the variable Z (which has n values if
there are n relays). Finally the statistical classifier present in each
relay takes decision whether fault has occurred or not based on
the calculated parameter Z. Fig. 8. Distance protection.
Only the lines with fault have more deviation between the local
and remote variable. Hence, the statistical classifier easily predicts Zset2 = Kr2 × (Zline + KbZ set1) (10)
the fault and then finally the faulted area can be isolated.

4.4. Distance protection Zset 3 = Kr 3 × (Zline + KbZ set2 ) (11)

Distance protection which offer high selectivity is applied where Kb is the branch coefficient which is calculated as
majorly in transmission lines. The impedance is calculated at each i3 i + i2
relay point using the formula Kb = = 1
i1 i1 (12)
V
Zm = m The new relay settings make the protection scheme to work
im (4) effectively for grid connected and islanded mode with good se-
where Vm is the measured voltage and im is the measured current lectivity and sensitivity [48,49].
at the relay point. Under normal condition, the measured im- The dynamics in relay setting due to the topological changes
pedance includes the load impedance and hence the value is high. occurring in the network based on status of DG and circuit
But, in case if there is any fault in the line then in such cases the breakers is incorporated for distance protection with the help of a
measured impedance will be equal to the line impedance alone, central controller. The relay setting for all possible configurations
(depending on status of DG, CB) is calculated offline and stored in
which is very low. Hence, by comparing the measured impedance
an action table. Similar to the adaptive over current protection
with set value, the occurrence of fault and whether the occurred
here also whenever a change is felt in the network, immediately
fault is within the particular protection zone is identified.
the central controller updates the relay setting by referring with
Usually for a region three zones of protection is offered. Zone
the action table and informs the same with the appropriate relays.
1 protection is instantaneous protection which covers 80% of the
zone. Zone 2 protection covers the area uncovered by zone 1. Zone
4.5. Multi agent protection
3 offers back up protection for the adjacent line. Both zone2 and
zone 3 operates with definite time, time of operation of zone 3 is
A multi agent is a group of intelligent hardware and software
greater than 2. The set value is chosen in such a way that
agent distributed in the network and work together to achieve a
Zset1 oZset2 oZset3. The set value is calculated as
global goal, here its efficient protection of microgrid.
Zset1 = Kr1 × Zline (5) The architecture for multi agent scheme designed for adaptive
over current protection involves three layers namely equipment layer,
substation layer and system layer (Fig. 9). The over current relays
Zset2 = Kr2 × (Zline + Z set1) (6) located in various parts of microgrid network forms the equipment
layer. The equipment layer is the lowest layer in the multi agent
network which involves measurement agent, protector agent, mobile
Zset 3 = Kr 3 × (Zline + Z set2 ) (7) agent and performer agent. The measurement agent monitors the
local parameters (current and voltage) and transfers it to the protector
where Kr1, Kr2, Kr3 are the reliability coefficient of each zone, agent. From the received local variables, protector agent checks
usually set as a value between 0.8 and 0.85 whether fault has occurred or not (by comparing the monitored local
The distance protection if applied with microgrid, (Fig. 8) for a variables with the set value). If it detects fault then it sends in-
fault at k, the DG connected at Bus 3 also contributes to the fault formation to the performer agent and the performer agent trips sui-
current. The DG impedance will also be get added along with the table Circuit Breakers so as to isolate the faulty part. Mobile agents
line impedance. Hence the impedance seen from the line23 is high exchange the information of the equipment to the substation layer.
compared to impedance at bus 2. The impedance measured by The substation layer communicates with the equipment layer and
relay R 23 is transfers the information to the system layer. The substation layer
Vm i Z + i3Zk involves region agent and management agent. The region agents
Zm = = 1 line23 transmit the information obtained from mobile agent to system layer
im i1 (8)
and to management agent. The management agent calculates the
The measured impedance becomes larger than the set value as settings of the relays incorporating the topological changes occurred
per Eq. (8) and the protection scheme lacks in selectivity. Thus the in the microgrid network and then transfers it to protector agent.
measuring impedance in the upstream/downstream network gets The highest layer forms the system layer. The system layer
disturbed with the inclusion of DG with considerable capacity. To which is the control center for microgrid network, monitors the
reduce the impact of the infeed current by DG in the protection entire network by communicating with substation layer. It updates
scheme the set value of the relays are recalculated as itself with recent information of the network which includes local
variables such as status of CBs, status of DG, current, voltage. Based
Zset1 = Kr1 × Zline (9) on these variables it run the load flow and also calculates the
996 B.J. Brearley, R.R. Prabu / Renewable and Sustainable Energy Reviews 67 (2017) 988–997

topology is also incorporated for relay setting calculations.

References

[1] BobLasseter. “Microgrids” Role of distributed generation in reinforcing the

critical electric power infrastructure. IEEE WM Panel; 2001.
[2] Kroposki B, et al. Making microgrids work. Power Energy Mag, IEEE
2008;6:40–53.
[3] LasseterRH. “Microgrid”. In: Proceedings of the IEEE power and energy society
winter meeting, vol. 1. New York; 2002. p. 305–8.
[4] GiriVenkataramanan Mahesh Illindala. Microgrid and sensitive loads. In:
Power engineering society winter meeting, vol. 1. IEEE; 2002. p. 27–31.
[5] XinyaoDyskoAdam. M. Burt published in IEEE transaction on smart grid. Tra-
velling wave based protection scheme for inverter dominated microgrid using
mathematical morphology, vol. 5, 5, September; 2014.
[6] Majumder Ritwik, Dewadasa Manjula, Ghosh Arindam, Ldwich Gerard, Zare
Firuz. Control and protection of a microgrid connected to utility through back
to back converters. Electr Power Syst Res 2011;81:1424–35.
[7] Katiraei F, et al. Microgrids management. Power Energy Mag IEEE 2008;6:54–
65.
[8] Katiraei F, Iravani MR. Micro-grid autonomous operation during and sub-
sequent to islanding process. IEEE Trans Power Deliv 2005;20:248–57.
Fig. 9. Architecture for multi agent scheme. [9] Shahnia Farhad, Ruwan, chandrasena PS, Rajakaruna Sumedha, Ghosh Ar-
indam. Primary control level of parallel distributed energy resources con-
verters in system of multiple interconnected autonomous microgrid within
self healing networks. IET Gener Transm Distrib 2013;8(2):203–22.
setting parameters for relays and communicates it with the sub- [10] Pedrasa MA, Spooner T. A survey of techniques used to control microgrid
generation and storage during island operation. In: Australian universities
station layer. The system layer has a system agent and evaluation
power engineering conference; 2006.
agent. The information received from region agent is sort out by [11] Zamora Ramon, Srivastava Anurag K. Controls for microgrids with storage:
system agent and only the essential data’s are passed to the eva- review, challenges, and research needs. Renew Sustain Energy Rev 2010:2009–
18.
luation agent. The evaluation agent rechecks the protection sys-
[12] De Brabandere K, Vanthournout K, Driesen J, Belmans R. Control of microgrids.
tem. If it detects any lack in protection then it immediately sends In: Proceedings of IEEE power engineering society general meeting; 2007.
the modified setting values to the management agent thereby a [13] OlivaresDEMehrizi- SaniAEtemadiAHPalma-Behnke, et al. Trends in microgrid
smooth adaptive protection is carried out with the help of multi control. In: IEEE transaction on smart grid, vol. 5(4). p. 1905–19.
[14] Zamani Mohammed Amin. Protection and control of active distribution net-
agent scheme [50–52]. work and microgrids. A Ph.D. thesis submitted to University of western On-
tario, London; Dec 2012.
[15] IhamäkiJukka. Integration of microgrids into electricity distribution networks.
Master's Thesis submitted in Lappeenranta University of Technology during
5. Discussion 2012; 2012.
[16] Selim Taha, Ustun. Design and development of a communication assisted
microgrid protection system. Ph.D. thesis submitted in School of Engineering
The major components present in the microgrid for which the
and Science, Faculty of Health, Engineering and Science, Victoria University;
protection is to be provided are DG, line, loads and PCC. The trans- 2013.
former present in the PCC should be designed with differential [17] BarkerPP, De MelloRW. Determining the impact of distributed generation on
power systems. I. Radial distribution systems. In: Power engineering society
protection relay and the PCC with over current, under voltage, over
summer meeting, 2000, vol. 3. IEEE; 2000. p. 1645–56.
voltage, under frequency, over frequency relays. For the microgrid [18] David M LavertyRobert J BestMorrowD John. Loss of mains protection system
line the differential protection is well suited as the primary protec- by application of phasor measurement unit technology with experimentally
assesed threshold settings. IET Gener Transm Distrib 9(2);146–53.
tion and over current as back up protection. Over current protection
[19] PapasopilotopoulosVAKleftakisVA, et al. Hardware in the loop simulation for
with inverse time characteristics is best suited protection for loads. protection blinding and sympathetic tripping in distribution grids with high
However, by reviewing the various protection strategies ad- penetration of distributed generation. In: IET conference. p. 1–6.
dressed for microgrid it is suggested that adaptive protection is the [20] Najy WKA, Zeineldin HH, Woon WL. Optimal protection coordination for
microgrids with grid-connected and islanded capability. IEEE Trans Ind Elec-
best protection scheme because this is the only scheme which tron 2013;60(4):1668–77.
incorporate the dynamic changes in the status of DG and CBs for [21] Ghanbari T, Farjah E. Unidirectional fault current limiter: an efficient interface
the relay settings and also works faithfully for all operating con- between the microgrid and main network. IEEE Trans Power Syst
2013;28:1591–8.
ditions. Thus for microgrid loads adaptive over current protection [22] Ebrahimpour M, Vahidi B, Hosseinian SH. A hybrid superconducting fault
and for distribution lines adaptive differential protection forms the current controller for DG networks and microgrids. IEEE Trans Appl Super-
best protection scheme. The cost for implementing adaptive pro- cond 2013;23(5).
[23] Vilathgamuwa DM, Loh Poh Chiang, Li Yunwei. Protection of microgrids dur-
tection scheme with advanced digital relays and sensors is very ing utility voltage sags. IEEE Trans Ind Electron 2006;53(5):1427–36.
high, but if the cost is compared with the energy price and if its [24] Blair SM, Coffele F, Booth CD, Burt GM. An open platform for rapid-prototyping
calculated for long duration then it seems to be a better choice. protection and control schemes with IEC 61850. IEEE Trans Power Deliv
2013;28(2):1103–11.
[25] LaaksonenHannu Jaakko. Protection principles for future microgrids. In: IEEE
transactions on power electronics, vol. 25 (12); 2010.
[26] Ustun TS, Ozansoy C, Ustun A. Fault current coefficient and time delay as-
6. Conclusion
signment for microgrid protection system with central protection unit. IEEE
Trans Power Syst 2013;28(2):598–606.
Increasing power demand can be met by addition of micro- [27] Ustun TS, Ozansoy C, Zayegh A. Modeling of a centralized microgrid protection
grids. The insertion of DG in the distribution network makes ex- system and distributed energy resources according to IEC 61850-7-420. IEEE
Trans Power Syst 2012;27(3):1560–7.
isting protection algorithm ineffective. The reason for the failure of [28] Ustun TS, Ozansoy C, Zayegh A. Differential protection of microgrids with
the existing algorithm and the various protection schemes sug- central protection unit support. In: IEEE TENCON spring conference; 2013. p.
gested for different microgrid environments are discussed in de- 15–9.
[29] Zamani MA, Yazdani A, Sidhu TS. A communication-assisted protection
tail. Adaptive protection is the best available protection for mi- strategy for inverter-based medium-voltage microgrids. IEEE Trans Smart Grid
crogrid, since dynamic changes in the distribution network 2012;3:2088–99.
 B.J. Brearley, R.R. Prabu / Renewable and Sustainable Energy Reviews 67 (2017) 988–997 997

[30] BrahmaSMGirgisAA, Development of adaptive protection scheme for dis- microgrid protection. IET Gener Transm Distrib 8(2):2014;310–20.
tribution systems with high penetration of distributed generation. IEEE Trans [42] Kar Susmita, Samantaray SR. Combined S transform and data mining based
Power Deliv 19;53-63. intelligent microgrid protection scheme. IEEE Int Conf 2014.
[31] Oudalov Alexandre, Fidigatti Antonio. Adaptive network protection in micro- [43] ASaleh S, et al. Implementing and testing D-q WPT based digital protection
grids. Int J Distrib Energy Resour 2009;4(3):201–25. for microgrid systems. IEEE Trans Ind Appl 2014;50(3).
[32] Amir HEtemadi, Iravani Reza. Overcurrent and overload protection of directly [44] Saleh SA. Signature coordinated digital multi relay protection for microgrid
voltage controlled distributed resources in a microgrid. IEEE Trans Ind Elec- systems. (September). IEEE Trans Power Electron 2014;29(9).
tron 2013;60(0):12. [45] JanssenPierre. Monitoring, protection and fault location in power distribution
[33] Laaksonen Hannu, Ishchenko Dmitry. Adaptive protection and microgrid networks using system-wide measurements. Ph.D. thesis submitted in Eco-
control design for Hailuoto Island. IEEE Trans Smart Grid 2014;5(3). lePolytechnique; 2013.
[34] Ustun Taha Selim, Zayegh Aladin, Ozansoy Cagil. Differential protection of [46] Casagrande E, Woon WL, Zeineldin HH. A differential sequence component
microgrids with central protection unit support. IEEE 2013. protection scheme for microgrids with inverter-based distributed generators.
[35] Nthontho MPChowdhury SPWinberg SChowdhury S. Protection of domestic IEEE Trans Smart Grid 2014;5(1):29–37.
solar photovoltaic based microgrid. In: 11th international conference on de- [47] Salehi V, Mohamed A, Mazloomzadeh A, Mohammed OA. Laboratory-based
velopments in power system protection; 2012. p. 1–6. smart power system, Part II: Control, monitoring, and protection. IEEE Trans
[36] SorotommeE. A differential zone protection scheme for microgrids. In: IEEE Smart Grid 2012;3(3):1405–17.
international conference; 2013. [48] Lin Hengwei, Guerrero Josep M, Quintero Juan Carlos Vasquez, Liu Chengxi.
[37] Haron Ahmad Razani, Mohamed Azah, Shareef Hussain. Co-ordination of over Distance protection for microgrids in distribution system. In: 41st annual
current, directional and differential relay for the protection of microgrid sys- conference of IEEE industrial electronics society. IECON 2015:000731–6.
tem. Sci Direct, Procedia Technol 2013;11:366–73. [49] Lin Hengwei, Guerrero Josep M, Quintero Juan Carlos Vasquez, Liu Chengxi.
[38] Mohammad E, Mohammad Liang Che, Shahidehpour Khodayar. Adaptive Adaptive Distance Protection for Microgrids. In: 41st annual conference of
protection system for microgrid. IEEE Electrif Mag 2014. IEEE industrial electronics society. IECON 2015:000725–30.
[39] Zamani MA, Sidhu TS, Yazdani A. A protection strategy and microprocessor- [50] Colson CM, Nehrir MH. A review of challenges to real-time power manage-
based relay for low-voltage microgrids. IEEE Trans Power Deliv 2011;26 ment of microgrids. In: Power & energy society general meeting, 2009. PES ‘09
(3):1873–83. IEEE 2009:1–8.
[40] Zamani MA, Sidhu TS, Yazdani A. Investigations into the control and protection [51] Coury DV, Throp JS, Hopkinson KM, Birman KP. Agent technology applied to
of an existing distribution network to operate as a microgrid: a case study. adaptive relay setting for multi-terminal lines. IEEE PES SM, Seattle 2000.
IEEE Trans Ind Electron 2014;61(4):1904–15. [52] kulasekera Asitha Lakruwan. Multi agent based control and protection for an
[41] Kar SSamantaray SR. Time-frequency transform-based differential scheme for inverter based microgrid.Sri Lanka: University of Moratuwa; 2012.