You are on page 1of 11

Electrical Power and Energy Systems 92 (2017) 156–166

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

journal homepage:

A protection scheme for microgrid with multiple distributed generations

using superimposed reactive energy
Syed Basit Ali Bukhari ⇑, Muhammad Saeed Uz Zaman, Raza Haider, Yun-Sik Oh, Chul-Hwan Kim
College of Information and Communication Engineering, Sungkyunkwan University, Suwon, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: With the growing integration of distributed generation, distribution networks have evolved toward the
Received 1 December 2016 concept of microgrids. Microgrids can be operated in either the grid-connected mode to achieve peak
Received in revised form 1 April 2017 shaving and power loss reduction or the islanded mode to increase the reliability and continuity of sup-
Accepted 3 May 2017
ply. These two modes of operation cause a challenge in microgrid protection, because the magnitude of
Available online 12 May 2017
fault current decreases significantly during the transition of a microgrid from the grid-connected mode to
the islanded mode. This paper proposes a protection scheme for the microgrid based on superimposed
reactive energy. The proposed scheme uses the Hilbert transform to calculate the superimposed reactive
Microgrid protection scheme
Superimposed reactive energy
energy (SRE). The sequence components of superimposed current are adopted to detect fault incidents in
Fault detection the microgrid. The faulty phase and section are recognised by using the directional characteristics of SRE
Fault location along with a threshold value. Moreover, a relay structure, which enables the proposed protection scheme,
High-impedance fault is designed. The significant feature of the proposed protection scheme is that it has the ability to protect
the looped and radial microgrids against solid and high-impedance faults. To verify the efficacy of the
proposed approach, extensive simulations have been carried out using the MATLAB/SIMULINK software
package. The results show that the proposed scheme successfully identifies and isolates various types of
fault in a microgrid and performs well with different fault resistances and fault locations.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction nical obstacles, which need to be explored. Microgrid protection

and its entities is a major obstacle [8].
In conventional power systems, power is generated centrally The main problem related to microgrid protection is the protec-
and then transmitted to the customer end by using long transmis- tion during the islanded mode, when the microgrid works indepen-
sion and distribution networks. In this case, power flows are unidi- dently of the main grid. In this case, the fault currents are small due
rectional in the distribution networks. Nowadays, mounting load to the limited current-carrying capacity of power electronics
demand, developments in renewable energy technology and devices. The fault current is merely 2–3 p.u. of rated current in case
increasing concerns on global warming have led to a new trend of inverter based DGs [9]. Thus, the traditional overcurrent protec-
of electricity production at the distribution level. These technolo- tion, which relies on the assumption of high fault current, is insuf-
gies are usually called distributed generation (DG) [1,2]. The intro- ficient for the protection of islanded microgrids. However, in the
duction of DGs at the distribution level decreases the power grid-connected mode, the main grid also contributes to fault cur-
transfer burden on the transmission and distribution networks rent therefore, the magnitude of the fault current is significantly
[3]. These new technologies have enabled the smaller distribution larger. Although it is possible to use overcurrent relays for the pro-
networks called microgrids [4,5]. tection of grid-connected microgrid, the existence of DGs alters the
A microgrid is a small part of a power system which consists of direction and magnitude of fault current and hence compromise
parallel DGs, energy storage devices and electrical/heat loads. It the relay coordination.
can work in the grid-connected as well as the islanded mode, for A protection scheme for microgrid must ensure safe operation of
providing uninterrupted service to customers, and for improving microgrid in both modes of operation and should take into account
the reliability, operational optimality and power quality of the sys- (1) two-way power flow in distribution networks (2) presence of
tem [6,7]. Microgrids offer various benefits, but present some tech- looped feeders and (3) reduced fault current magnitude in the
islanded mode. As, traditional protection schemes mainly based
⇑ Corresponding author. on overcurrent relays are ineffective in protecting microgrids.
E-mail address: (S.B.A. Bukhari). Hence, there is a need for alternate schemes of microgrid protection.
0142-0615/Ó 2017 Elsevier Ltd. All rights reserved.
S.B.A. Bukhari et al. / Electrical Power and Energy Systems 92 (2017) 156–166 157

Various schemes for microgrid protection have been developed connected mode) or consider radial configuration of microgrids
and reported in the literature [10,11]. In [12], the authors used a only. Some schemes, which protect the microgrids in both modes
central controller along with a digital relay to update the precalcu- of operation, are not suitable for HIFs. The schemes developed for
lated relay settings according to the operating mode of the micro- HIFs, require additional devices in microgrids; therefore, they
grid. This scheme was upgraded by Coffele [13] who suggested an involve very high cost.
online method to calculate the settings of adaptive overcurrent This paper proposes a novel scheme for the protection of micro-
relays. The developed scheme had the capability to automatically grids based on superimposed reactive energy (SRE). In this paper,
adjust the settings of all overcurrent relays in response to the sta- SRE is defined as the integral of superimposed reactive power dur-
tus of DGs, operational mode of the microgrid and active networks ing a specified period of time. A new fault detection ratio based on
management. However, relatively sophisticated fault calculations sequence components of superimposed current is developed for
were required for microgrids operating in different modes. More- detection of fault incidents in microgrids. The proposed scheme
over, the scheme was not suitable for high-impedance faults uses the Hilbert transform to calculate the SRE. The directions of
(HIF). In a research by Mirsaeidi et al. [14], a central processing SRE flow at both ends of each feeder are employed to identify
unit-based protection scheme for microgrids was proposed. Pre- the faulty line in a microgrid. This paper uses a threshold on SRE
fault and post fault positive sequence impedances of distribution of each phase to classify the faults. To validate the effectiveness
lines at both ends were calculated and compared to locate the of the proposed scheme, extensive simulations are carried out
faulty section. However, the suggested scheme required complete using MATLAB/SIMULINK software. The main contributions of this
upgrading of the protection devices currently used in the distribu- paper are as follows:
tion networks; moreover, the presented scheme was not able to
protect the microgrid against HIFs. In [15], Oureilidi and Konstanti-  The proposed scheme has the ability to protect the looped and
nos suggested the switching of converter control from droop con- radial microgrid against solid and HIFs.
trol to current control during faults in microgrid. The faulty section  The scheme is valid for both islanded and grid-connected
was identified by measuring the fault current provided by each microgrid and there is no need to modify the setting of relays
converter. The scheme was only applicable to inverter-dominated when the microgrid changes its mode.
microgrid; moreover, it required some external energy source in  The proposed protection scheme also offers backup protection
case of HIF. A travelling wave scheme based on mathematical mor- against faults in case of failure of main protection.
phology filter was presented in [16]. Two different logics were
The rest of the paper is structured as follows. Section 2 illus-
developed for microgrids with different configuration and topolo-
trates the basic principle of the superimposed components, Hilbert
gies. The developed method was independent of fault resistance.
transform and SRE characteristics, which are the basis of the sug-
The scheme required accurate signal synchronisation and very high
gested protection scheme. Section 3 explains the proposed scheme
sampling frequency, which is impractical due to lack of cost effec-
in details. Exhaustive simulations are performed to test the effi-
tive DSP hardware. In [17], the wavelet packet transform (WPT)
ciency of the proposed protection in Section 4. Finally, the paper
was suggested to detect and clear the faults in microgrids. The
is concluded in Section 5.
scheme used a half-band digital high-pass filter to realize the
WPT. The authors did not consider HIFs and looped structure of
microgrids. A voltage-based protection strategy for microgrids 2. Basic principle
was proposed in [18,19]. The abc-dq0 transformation of voltage
was used to detect the fault incidents in microgrids. The scheme 2.1. Superimposed components
was limited to islanded microgrids and solid faults. Moreover,
the scheme did not consider the microgrid with looped configura- Subsequent to short-circuit fault on any point in a microgrid,
tion. The authors in [20] suggested the deployment of energy stor- the variables like current, voltage, power etc. may face significant
age devices in the islanded mode of microgrids to equalise the fault changes with respect to their corresponding prefault values. The
current magnitude in islanded and grid-connected modes. How- change in the variables due to a fault is known as the superim-
ever, the installation of such devices with high short-circuit capac- posed component/fault-imposed component. The superimposed
ity was not economical. Zamani et al. [21] developed a protection components appear due to the fault in the network. They keep fault
strategy by using negative and zero-sequence components to pro- signature and are independent of the values under normal-running
tect the microgrid against various types of faults. The scheme also conditions. Hence, any electrical variable during a fault comprises
had a backup protection in case of communication failure. As the two components: (1) a normal operating component and (2) a
scheme used zero sequence current as a directional element, it superimposed component. Thus, any arbitrary variable xðtÞ during
required a grounding transformer inside the microgrid. In [22], a a fault can be expressed as
differential protection scheme based on a statistical classifier has xf ðtÞ ¼ xf i ðtÞ þ xpre ðtÞ ð1Þ
been developed for microgrids. Various features from voltage and
current signals were obtained at both ends of a line in microgrids. where xf ðtÞ, xpre ðtÞ and xfi ðtÞ are the fault, prefault and superimposed
A learning-from-data feature selection approach was used to select components of variable xðtÞ respectively.
most useful subset of features among the ones obtained initially at xf i ðtÞ ¼ xf ðtÞ  xpre ðtÞ ð2Þ
each relay point. Finally, a differential operation is applied to
selected features to detect and locate the faults in microgrids. Eq. (2) shows that the superimposed component can be com-
The scheme was implemented on islanded microgrids against solid puted by subtracting the normal or prefault component from the
faults only. A protection scheme based on wavelet and data fault component. Therefore, the superimposed component does
mining-based approach was suggested by Mishra et al. [23]. The not appear under normal operation but exists if a fault occurs
statistical features of current waveform extracted by a wavelet within the microgrid. Superimposed components can be calculated
transform were used to develop the data-mining models. The fault by using delta filters [24]. The simplest delta filter subtracts the
classification was conceded based on these models. The scheme integral multiple delayed version of the input signal from the input
was applicable to inverter-interfaced DGs only. signal itself.
The main limitations of above-mentioned schemes are that they xf i ðtÞ ¼ xðtÞ  xðt  nTÞ ð3Þ
either protect the microgrids in one mode (islanded or grid-
158 S.B.A. Bukhari et al. / Electrical Power and Energy Systems 92 (2017) 156–166

where T is the time period of the signal and n is an integer (n = 6 in The Hilbert transform-based instantaneous reactive power
this paper). In discrete form, the superimposed components can be using superimposed current and voltage can be computed as pro-
obtained by subtracting the present sample of an input variable posed in [25,26].
from its corresponding sample in the previous cycle.
^fi ðtÞ  ifi ðtÞ
qðtÞ ¼ v ð10Þ
xfi ½kTs  ¼ x½kTs   x½kTs  nT ð4Þ
where v^ fi ðtÞ represents the Hilbert transform of superimposed volt-
where T s is the sampling period, N is the number of samples per age signal and ifi ðtÞ represents the superimposed current; qðtÞ rep-
cycle and k is the sampling step (k ¼ 0; 1; 2; 3 . . .). resents the instantaneous reactive power. The SRE during time
period t can be obtained by integrating Eq. (10) over t as
2.2. The Hilbert transform Z Z
t t
Eq ðtÞ ¼ qðsÞds ¼ ^fi ðsÞ  ifi ðsÞds
v ð11Þ
This scheme uses the Hilbert transform to calculate the SRE, 0 0
which is the convolution of function f ðtÞ with the Hilbert transfer
The Hilbert transform-based SRE calculation method is advan-
function HðtÞ [25]. The Hilbert transfer function in the time domain
tageous over traditional calculation methods because it uses orig-
can be represented as
inal sampled current and voltage signals. Hence, computing the
HðtÞ ¼ 1=pt ; 1 < t < þ1 ð5Þ discrete Fourier transform of current and voltage signals for all har-
monics components is not required. Moreover, the proposed
Because HðtÞ is not an integrable function, the Cauchy integral is
scheme continuously obtains the SRE in real time, increasing the
used to define the Hilbert transform.
operating speed of the protection scheme, whereas conventional
^fðtÞ ¼ fðtÞ  1=pt ¼ 1 P
fðsÞ computation methods need a complete cycle of the measured
ds ð6Þ
p 1 ts data [27].

where P is the principle value of the integral.

2.3. Characteristics of SRE during faults
The frequency response of the Hilbert transfer function can be
expressed as
8 To understand the characteristics of SRE during faults, a micro-
< þj; for x < 0
> grid having three buses and a load connected to bus B as depicted
H ðxÞ ¼ jsgn ðxÞ ¼ 0; for x ¼ 0 ð7Þ in Fig. 1a is considered. In order to develop an appropriate scheme
j; for x > 0 to protect microgrids with various configurations, microgrid
should be divided into small sections and each section (microgrid
As negative frequency does not exist in real-world applications, line) is protected independent of the other sections. To accomplish
the Hilbert transform can be considered as an all-pass filter except this, the downstream and upstream of each microgrid line can be
for DC components. It induces a 90° phase shift for all positive replaced by their equivalent circuits. During faults, the inverter
frequency components current is limited by its control system to 2–3 p.u. of the normal
The Hilbert transform can be represented in discrete form as load current and inverter behaves like a constant current source.
[26] Therefore, an inverter during faults can be represented by a con-
^f½n ¼ f½n  H½n stant current source along with an impedance. Ref. [28] well
describes the equivalent circuit representation of inverters during
 faults. The other system components and load can also be replaced
0; for ev en v alues of n
H½n ¼ ð9Þ by their equivalent impedances [29]. The equivalent circuit repre-
2=np; for odd v alues of n sentation of a microgrid line is adopted from [30]. Fig. 1b shows

CB-2U CB-2D CB-1D Main Grid
Line-2 Line-1

Transformer MGPR-2U MGPR-2D MGPR-1U MGPR-1D

DER VSC Converter

(a) Load

Upstream Upstream B C Downstream Downstream

Equivalent Equivalent Equivalent Equivalent
source Impedance MGPR-1U MGPR-1D Impedance source

Upstream Equivalent F2 F1 Downstream
Circuit (b) Equivalent Circuit

Fig. 1. Microgrid test system. (a) Single-line diagram. (b) Upstream and downstream equivalent circuits of Line-1.
S.B.A. Bukhari et al. / Electrical Power and Energy Systems 92 (2017) 156–166 159

the downstream and upstream equivalent circuits of Line-1 of the The SRE at bus B for one cycle can be computed as
microgrid shown in Fig. 1a. Faults F1 and F2 represent the forward Z 2p=x
and reverse faults in the microgrid with respect to the relay MGPR- EBq ¼ ^Bfi ðtÞ  iBfi ðtÞ dt
v ð16Þ
1U respectively. 0

2.3.1. SRE flowing through Bus B during fault F1 Xu p

EBq ¼  V2 ð17Þ
As explained earlier, that the sum of the superimposed and pre- Xu þ Xa m x
fault quantities gives the total variation at all point in the network. According to Eq. (17), the sign of SRE flowing through bus B for
In addition, the superimposed network is a network with all volt- fault F1 is negative. Thus, if a relay computes the value of SRE as
age sources set at zero. The superimposed voltage at fault point is less than zero, the fault is considered as a forward fault with
therefore zero before the initiation of fault transient. Thus, the respect to that relay.
voltage of the fictitious superimposed source at fault point is equal
in magnitude and opposite in sign to the prefault voltage at the 2.3.2. SRE flowing through bus B during fault F2
fault point. The superimposed quantities and their network repre- For the fault at point F2 (reverse fault for relay at bus B), the
sentation are well explained in [24,31]. subsequent superimposed network is shown in Fig. 3.
The change in relay point current and voltage due to a fault can
be obtained by zeroing all the pre-fault voltage sources and repre- diBfi ðtÞ
ðLu2 þ Lab þ Ld Þ ¼ Dvfi ¼ Vm sinðxt þ hÞ ð18Þ
senting all the load and system components by their equivalent dt
impedances. Hence, for the fault at point F1 (forward fault for relay where Lu2 is the equivalent inductance between bus B and fault F2,
at bus B), the superimposed network of the microgrid shown in whereas Lab and Ld represents equivalent inductances of Line-1 and
Fig. 1 is drawn in Fig. 2. For easiness, only the inductive parts of the downstream circuit respectively.
impedances are reflected. Moving the time origin to the fault initi-
ation point and assuming the prefault voltage at fault point equal iBfi ðtÞ ¼ ðcosðxt þ hÞ  coshÞ ð19Þ
toV m sinðxt þ hÞ, the superimposed current and voltage at bus B Xu2 þ Xab þ Xd
can be computed as
Xd þ Xab
diBfi ðtÞ vBfi ðtÞ ¼  Vm sinðxt þ hÞ ð20Þ
ðLu þ La Þ ¼ Dvfi ¼ Vm sinðxt þ hÞ ð12Þ Xu2 þ Xab þ Xd
where iBfi ðtÞ is the superimposed current at bus B; Lu and La repre- The Hilbert transform of voltage signal v
^ Bfi ðtÞ can be computed
sent the upstream equivalent inductance and line inductance by shifting its angle by 90°.
between bus B and fault F1 respectively. Xd þ Xab p
^Bfi ðtÞ ¼ 
v sinðxt þ h  Þ ð21Þ
Vm Xu2 þ Xab þ Xd 2
iBfi ðtÞ ¼ ðcosh  cosðxt þ hÞÞ ð13Þ
Xu þ Xa The SRE at bus B for one cycle will be

vBfi ðtÞ ¼ Lu

diBfi ðtÞ
Vm sinðxt þ hÞ ð14Þ
Xd þ Xab p
EBq ¼ V2 ð22Þ
dt Xu þ Xa Xu2 þ Xab þ Xd m x
The Hilbert transform of voltage signal v
^ Bfi ðtÞ can be computed According to Eq. (22), for the fault at point F2, which is a reverse
by shifting its angle by 90° as given below: fault for the relay at point B, the value of SRE is positive.
Xu  p
^Bfi ðtÞ ¼ 
v Vm sin xt þ h  ð15Þ 2.3.3. Threshold setting
Xu þ Xa 2
A threshold value for SRE is essential to avoid nuisance tripping
during disturbances. Eqs. (17) and (22) imply that the value of SRE
will be different for different fault locations. Hence, selecting a
B i iCfi C threshold for SRE is challenging. To address this problem, this
Lu Ld paper defines SRE coefficients, which will be used for fault direc-
Lb tion determination and fault phase classification. The SRE coeffi-
cients can be obtained for each phase as
vBfi vCfi
Eqa Eqb Eqc
Kqa ¼ ; Kqb ¼ ; Kqc ¼ ; ð23Þ
if Eqmax Eqmax Eqmax
vfi =- vfpre
Eqmax ¼ MaxðjEqa j; jEqb j; jEqc jÞ ð24Þ
Fig. 2. Superimposed network of the microgrid for fault F1. where Eqa , Eqb ,Eqc and K qa , K qb ,K qc are the SREs and SRE coefficients
for phase A, B and C respectively. During the fault, the maximum
absolute value of SRE coefficients is equal to 1. Hence, a forward
fault is detected if the value of any coefficient at a specific relay
L u1 L u2 i Bfi i Cfi point is 1. When the SRE coefficient for any phase is greater than
L ab Ld
a threshold, the phase is considered as a faulty phase. In this study,
the threshold is assumed to be 0.2.
v Bfi v Cfi
if 3. Proposed protection scheme

v fi = - v fpre This paper proposes a protection scheme for microgrids using

SRE and sequence components of superimposed current, thus pro-
Fig. 3. Superimposed network of the microgrid for fault F2. tecting the microgrid against various types of faults in both
160 S.B.A. Bukhari et al. / Electrical Power and Energy Systems 92 (2017) 156–166

islanded and grid-connected modes. The proposed scheme uses disturbances and to avoid nuisance tripping a threshold RFDðthÞ is
microprocessor-based relay to detect fault incidents and to isolate used. When the value of RFD exceeds this threshold for a definite
the faulty sections in a distinctive way. A relay structure hereafter period of time, the occurrence of fault is recognised and a proper
referred as the microgrid protection relay (MGPR) is developed, action will be taken based on the proposed protection scheme
which enables the proposed scheme. The current and voltage sig- (described in the next section). In this paper, the threshold is
nals at each end of the line are processed locally through the MGPR assumed to be 2.5.
installed at that end. After detection of a fault in the microgrid, the
faulted section is recognised by a suitable logic, which is based on 3.2. Communication-aided fault location
communication between various MGPRs. Consequently, based on
the type of fault, appropriate trip signals are sent to the related cir- In this study, we use the directional characteristics of SRE to
cuit breakers. The next subsections explain the proposed scheme in identify the fault location. As explained in Section 2.3, SRE will
detail. be negative for forward faults and positive for reverse faults. Thus,
we can discriminate between reverse and forward faults. When a
3.1. Fault detection strategy fault incident is detected by the relay, the directional element
waits for a short prespecified time to obtain the stable fault direc-
As stated earlier, microgrids can operate in both grid-connected tion. This time delay is necessary to prevent the fluctuations in SRE
or islanded mode and the magnitude of fault current decreases sig- direction due to the frequent energy exchange between capacitive
nificantly during microgrid transition from being grid-connected and inductive elements at the start of the fault transient process. In
to islanded mode. Thus, detecting a fault in the islanded mode is this study, the time delay is set to two power frequency cycles. This
difficult. This paper proposes a new scheme that is based on the time is adequate for a stable direction outcome.
sequence components of current. However, these components A fault in one section can be a forward fault for many relays;
are also present under normal operating conditions because most hence, to avoid nuisance tripping and to improve the selectivity
of the distribution systems are inherently unbalanced in nature, of the scheme, the faulty line is identified by communicating infor-
and may cause the mal-operation of the protection scheme. The mation among different MGPRs. After receiving information from
problem becomes more complex in the case of HIFs or even for other MGPRs, an MGPR installed at a line determines through a
faults in the islanded mode. To address the problem, in this study, suitable logic whether the fault is in its zone or not. The logic
we use sequence components of superimposed current. The ratio mechanism is explained in the following section.
of sequence components of superimposed current and prefault
current can be defined as the fault ratio because this ratio only 3.3. Main and backup protection
appears during a fault and can be expressed as
In the proposed scheme, a communication link is developed
ji1fi j þ ji2fi j
Fr ¼ ; ð25Þ between three adjacent MGPRs. Each MGPR sends two signals to
ji1pre j þ ji2pre j
other MGPRs: (1) a fault detection signal SDet which shows whether
where ji1fi j and ji2fi j are the rms values of positive and negative or not the MGPR has detected a fault incident in its region and (2) A
sequence components of superimposed current, whereas ji1pre j and fault direction signalSDir which gives the direction of fault from the
ji2pre j are the rms values of positive and negative sequence compo- MGPR prospective. The signals SDet and SDir are the binary codes as
nents of current before the occurrence of fault. Ideally, this ratio depicted in Table 1.
should be equal to zero under normal operation because superim- In the meantime, each MGPR also receives the signals sent by
posed components appear only during faults. However, practically other two adjacent MGPRs and determines through an appropriate
F r will not be zero under normal condition due to the load variation logic mechanism the zone in which a fault has occurred. Each
and inherently unbalance nature of distribution network. When a MGPR is designed with two protection zones: (1) main protection
fault hits the microgrid during the grid-connected mode, the value zone and (2) backup protection zone. To understand the concept, a
of F r changes significantly. Therefore, the system imbalance is not of part of the microgrid considered in Fig. 1a is redrawn in Fig. 4 with
much importance and F r can be used to detect faults in the grid- protection zones of MGPR-1D.
connected mode. However, during the islanded mode and HIFs
the F r may not be necessarily larger so, the system imbalance can Table 1
cause nuisance tripping. Thus, one cannot ignore the system imbal- Binary codes of fault detection and fault direction signals.
ance during the islanded mode and HIFs. The system imbalance is
Signal name/signal value 1 0
defined by a ratio known as ‘‘system imbalance ratio” and can be
SDet Fault detected No fault
computed as proposed in [21].
SDir Forward fault (FF) Reverse fault (RF)
ji0pre j þ ji2pre j
Rn ¼ ; ð26Þ
ji1pre j
During normal operation, the system imbalance ratio is usually
Zone 2 Zone 1
less than 20% [21,32]. Hence, a new ratio, which also takes into C
account the prefault system imbalance ratio, can be expressed as
RFD ¼ ; ð27Þ Line-2 Line-1
The fault ratio F r cannot exceeds the system imbalance ratio MGPR-2U MGPR-1U MGPR-1D
under normal operation. Therefore, RFD can retain a maximum
value 1 under normal operation. However, during faults RFD will
be much greater than one. Therefore, the value of RFD is less sensi- Load
tive to system imbalance and can efficiently be used to detect fault
incidents. To further desensitise the protection scheme from load Fig. 4. Protection zones of MGPR-1D.
S.B.A. Bukhari et al. / Electrical Power and Energy Systems 92 (2017) 156–166 161

Zone 1: It is the main protection zone of MGPR-1D covering must be greater than the operating times of the primary protec-
Line-1. The relay operates immediately when a fault occurs in tion relay and the circuit breaker associated with primary
this zone. protection.
Zone 2: MGPR-1D operates as a backup protection for this zone.
Therefore, for any fault in this zone, the relay operates after a The fault zone identification logic for MGPR-1D is shown in Fig. 5.
prespecified time delay. The time delay is necessary and allows It is clear from Fig. 5 that MGPR-1D not only receives the signals (SDet
the main protection of zone 2 to operate first. This time delay andSDir ) from MGPR-1U to recognise faults in main protection zone

Line-2 Line-1
S Det(1D) S mz(trip)
S Dir(1D)
Load S Det(1U)
FF Trip
Communication Link S Dir(1U) Main Protection Unit

TD Backup
S bz(trip)
Communication Link S Det(2U).

S Dir(2U)
Backup Protection Unit

Fig. 5. Fault zone identification logic for MGPR-1D.

Pre fault Pos., System Rn

Neg. and Zero Imbalance
Seq. Est. Ratio Est. Fault
iabc i012(pre) Detection Yes=1 SDet(1D)
ratio RFD No=0
Super Impd. ifi(abc) Pos., Neg. and
Comp Est. Zero Seq. Est. ifi(012)
Fault Ratio Est. Fault Detection Unit

Eqa SRE Coefficients Est. Kqa

vabc Super SRE Est. Is
HT. Eqb Kqa, Kqb,
Impd. for each (Kqa=-1, or Yes =1
Est. Eqc and Kqc TD SDir(1D)
Comp Est. Phase Kqb Kqb=-1, or No=0
ifi(abc) Kqc
|Eqb|, |Eqc|).
Fault Direction Unit

SDet(1D) Smz(trip)
Communication Port

FF Trip
Main Protection Unit
Sbz(trip) Fault Zone Identification
SDir(2U) Backup Protection Unit Unit
FF Trip
Yes=1 Phase A trip
Kqa |Kqa|≥0.2
Yes=1 Phase B trip
Kqb |Kqb|≥0.2
Yes=1 Phase C trip
Kqc |Kqc|≥0.2
Fault Classification Unit Tripping Unit

Fig. 6. Simplified schematic diagram of MGPR-1D for a microgrid shown in Fig. 1a.
162 S.B.A. Bukhari et al. / Electrical Power and Energy Systems 92 (2017) 156–166

Main Grid


MGPR : Microgrid Protection Relay.

CB : Circuit Breaker.
Tr. : Transformer.
CB-12 : DG protection.


Photovoltaic Park

CB-21 300-kW Synchronous

2 MGPR-21
CB-25 CB-23 CB-27


2km 0.5km
0.4kV Line
CB-52 CB-72
MGPR-52 CB-32 MGPR-72
24.9kV F3 7
5 MGPR-32
CB-56 CB-78
24.9kV CB-34 Load-3



Photovoltaic Park



MGPR-89 Load-4


4 CB-98 F5
CB-46 MGPR-98
MGPR-46 9
Load-2 CB-910
MGPR-910 Load-5


Line F1
CB-65 CB-64 MGRR-109
MGPR-65 10

640-KW Wind Farm


Fig. 7. Single-line diagram of the study microgrid.

but also receives the signals from MGPR-2U to operate as a backup munication can be used as a communication medium for the pro-
protection for Line-2. If MGPRs located at both ends of Line-1 detect posed scheme. With the progression in wireless communication,
a fault as a forward fault then the main protection zone (Line-1) is standard technologies are offered for local area metropolitan area
considered as faulty and a trip signal SmzðtripÞ will be issued. On the and wide area. Wireless communication has many advantages over
other hand, if MGPR-1D detects a fault as a forward fault whereas wire communication, including fast development, high mobility and
MGPR-1U identifies it as a reverse fault then the fault zone identifi- low cost. Wireless signal transmission requires 0.1 ms for a distance
cation unit of MGPR-1D check the status of the fault at MGPR-2U by less than 30 km, which is acceptable for most of the distribution sys-
its received signals. If MGPR-2U recognises the fault as forward fault tem applications. Synchronised phasor measurements are required
then the backup protection zone is regarded as faulty and after a pre- when the distance is greater than 30 km. In this study, communica-
specified time delay, backup protection unit generates a trip signal tion delay is taken as 3 ms to simulate a more severe condition.
SbzðtripÞ . Finally, the trip signals (SmzðtripÞ and SbzðtripÞ ) from main and
backup protection units are combined using OR logic operation to 3.4. Microgrid protection relay (MGPR)
obtained the forward fault (FF) trip signal.
The proposed protection scheme can be implemented with the To implement the proposed protection scheme, a relay
communication facilities of a smart grid. Wireline or wireless com- structure is designed. Fig. 6 shows the schematic diagram of
S.B.A. Bukhari et al. / Electrical Power and Energy Systems 92 (2017) 156–166 163

MGPR-1D for a microgrid presented in Fig.1a. Each MGPR has five farm. The photovoltaic parks and the wind farm are connected
units: ‘‘fault detection unit”, ‘‘fault direction unit”, ‘‘fault zone identi- through power electronics converters.
fication unit”, ‘‘fault classification unit” and ‘‘tripping unit”. A droop control strategy is used for all DGs so that the microgrid
can work in both modes without controller mode switching and is
3.4.1. Fault detection unit adopted from [33]. A synchronisation algorithm is also integrated
This unit generates fault detection signal SDet to be sent to other in the control strategy of each DG for resynchronisation of the
relays. The fault detection unit benefits from superimposed com- microgrid with the main grid. The microgrid is interfaced to the
ponents and the proposed fault detection ratio, as described in Sec- main grid through a 69/24.9 kV Dyn transformer and each DG
tion 3.1. This unit is supposed to detect all types of faults including source is connected through a 0.4/24.9 kV transformer.
HIFs in both grid connected and islanded modes of operation.
4.2. Case studies
3.4.2. Direction determination unit
This unit is designed to identify the right direction of the fault. A complete fault analysis for all locations in the study microgrid
The output of this unit SDir is sent to the other relays to discrimi- is executed. The fault types considered in this study include single-
nate the faulty line. SDir is also used in the fault zone identification line to-ground (SLG) faults, line-to-line faults (LL), double-line-to-
unit (to be explained next) to identify a faulted zone. The direction ground faults (DLG), three-phase faults and HIFs. Moreover, the sit-
determination unit takes advantage of SRE described in Section 2.3. uations of circuit breaker and MPGR failures are also simulated to
The SRE identifies the correct direction not only for solid faults but validate the effectiveness of backup protection. The tests are per-
also for HIFs. formed under two different scenarios: grid-connected mode and
islanded mode. In addition, the looped case is developed by simu-
3.4.3. Fault zone identification unit lating the fault on Line-46 of the study microgrid. Due to the lim-
This unit is responsible for operating the relay when a fault itation of space, only few simulation results are included in this
occurs in main or backup protection zones. The unit receives paper.
SDet and SDir signals sent by other relays and determines through
an appropriate logic mechanism (described in Section 3.3) the zone
4.2.1. Results for solid faults
in which a fault has occurred. A time delay is provided for the
Let us assume that a solid fault impacts the Line-46 of the study
backup protection operation to allow the main protection of the
microgrid in the islanded mode of operation. The purpose of this
faulty zone to operate first. After that time delay, the relay will
case study is to verify the efficacy of the proposed scheme when
send a signal to the circuit breaker to open the fault, if the fault
the microgrid operates in the islanded mode. The current, fault
is still present. The backup trip signal is sent after a time delay of
detection ratio and SRE coefficients for this case at MGPR-46 are
0.3 s.
shown in Fig. 8. It can be depicted from the figure that the value
of SRE coefficient for faulty phase is 1. The SRE coefficients are
3.4.4. Fault classification unit obtained after 2 cycles delay. This delay allows the directional ele-
To enhance the proficiency of the proposed protection scheme, ment to obtain the stable fault direction. The results for grid con-
single-phase tripping capability is also considered. After the iden- nected mode when a three phase fault impacts the Line-89 of
tification of the faulted zone by the fault zone identification unit,
the fault classification unit classifies the fault by comparing the
absolute value of SRE coefficients for each phase with the prespec-
ified value, i.e., 0.2. If the value of the SRE coefficient for any phase
is greater than this threshold, the phase is considered as a faulty
phase; otherwise, it is a healthy phase.

3.4.5. Tripping unit

The last phase of decision making in the MGPR is accomplished
in the tripping unit. It receives signals from the fault zone identifi-
cation unit and the fault classification unit, and determines
whether or not a trip signal should be sent to the relevant circuit

4. Simulation results

To validate the efficacy of the proposed protection scheme and

the MGPR, extensive simulations have been performed by using
the MATLAB/SIMULINK software package. The details of the simu-
lated microgrid, which, hereinafter, is stated as ‘‘study microgrid”,
and the simulation case studies, are presented in the following

4.1. Study microgrid

Fig. 7 illustrates the single-line diagram of the study microgrid.

The basic structure of this system and its parameters are obtained
from [14,32]. The operating voltage of the microgrid is 24.9 kV. The
system contains two photovoltaic parks of capacity 500 kW, a syn- Fig. 8. Simulation results at MGPR-46 for SLG fault at Line 46 of the study microgrid
chronous machine of capacity 300 kW and a 640 kW-capacity wind in islanded mode.
164 S.B.A. Bukhari et al. / Electrical Power and Energy Systems 92 (2017) 156–166

Fig. 9. Simulation results at MGPR-89 for three-phase fault at Line 89 of the study
microgrid in grid connected mode. Fig. 10. Simulation results at MGPR-46 for DLG HIF at Line 46 of the study
microgrid in islanded mode.

study microgrid are shown in Fig. 9. These results are obtained at

MGPR-89 of study microgrid. tection relay and in the case of primary protection failure; the
Table 2 reports the operating relays, the operating times of the backup relay sends a trip signal after 0.3 s. The main and backup
main and backup relays, fault detection ratio and SRE coefficients protections are designed in such a manner that the smallest part
for selected scenarios at various locations in the study microgrid. of the microgrid is isolated in response to the fault. The proposed
It can be observed from Table 2 that the proposed fault detection protection scheme clears all types of faults in the islanded mode
ratio RFD is above the threshold and hence detects fault incidents within 73 ms. The faulty section in the grid-connected mode is iso-
successfully. Moreover, the faults are first cleared by the main pro- lated rapidly as compared to the islanded mode. This is because the

Table 2
Simulation results of various types of solid faults at different locations in the microgrid working in the islanded mode and the grid connected mode.

F.T F.L Main protection operation Back up operation

Grid connected mode (GCM) Islanded mode (IM) GCM IM
Op. MGPR RFD SRE coefficients Toper (ms) RFD SRE coefficients Toper (ms) Op. MGPR Toper (ms) Toper (ms)
Kqa Kqb Kqc Kqa Kqb Kqc
AG F1 46 170.9 1 0.04 0.02 68.6 24.01 1 0.03 0.04 72.2 34 368.6 372.2
64 187.7 1 0.01 0.02 68.5 14.72 1 0.03 0.02 72.5 56 368.5 372.5
F3 23 205.1 1 0.04 0.05 68.0 26.50 1 0.04 0.03 72.0 12 368.0 372.0
32 22.2 1 0.01 0.04 72.3 12.50 1 0.05 0.06 72.6 43 372.3 372.6
BG F5 89 317.3 0.05 1 0.02 68.0 21.67 0.01 1 0.02 72.3 78 368.0 372.3
98 54.90 0.01 1 0.01 70.6 24.40 0.03 1 0.02 71.3 109 370.6 371.3
BCG F4 56 297.8 0.01 0.64 1 68.1 37.10 0.04 0.90 1 71.2 25 368.1 371.2
65 58.25 0.01 0.61 1 70.5 11.20 0.01 0.88 1 72.7 46 370.5 372.7
ABG F2 34 156.8 0.52 1 0.09 68.9 26.91 0.81 1 0.03 72.0 23 368.9 372.0
43 46.09 0.51 1 0.01 70.8 10.54 0.82 1 0.03 72.8 64 370.8 372.8
ACG F5 89 515.0 1 0.02 0.63 67.5 37.42 0.94 0.04 1 71.2 78 367.5 371.2
98 99.40 1 0.04 0.94 69.5 34.45 0.93 0.03 1 71.5 109 369.5 371.5
BC F1 46 325.8 0.03 1 0.955 67.9 34.75 0.04 0.95 1 71.5 34 367.9 371.5
64 359.0 0.04 1 0.95 67.7 20.13 0.03 0.95 1 72.4 56 367.7 372.4
AB F3 23 410.0 0.98 1 0.03 67.5 38.45 0.95 1 0.01 71.1 12 367.5 371.1
32 41.84 0.97 1 0.01 70.9 10.50 0.94 1 0.02 72.8 43 370.9 372.8
ABC F4 56 315.4 0.84 1 0.98 68.0 47.80 0.9 0.97 1 70.8 25 368.0 370.8
65 67.80 0.82 1 0.850 70.0 11.49 0.91 0.99 1 72.6 46 370.0 372.6
ABCG F1 46 378.9 0.823 1 0.99 67.6 38.50 0.86 0.96 1 71.1 34 367.6 371.1
64 344.4 0.82 1 0.93 67.7 24.00 0.86 0.95 1 72.2 56 367.7 372.2
F3 23 425.8 0.85 1 0.94 67.5 11.50 0.90 0.96 1 72.6 12 367.5 372.6
32 45.35 0.87 1 0.95 70.7 42.40 0.89 0.94 1 70.9 43 370.7 370.9

F.T = Fault Type. F.L = Fault Location. Toper = Relay Operating Time Op. = Operating.
S.B.A. Bukhari et al. / Electrical Power and Energy Systems 92 (2017) 156–166 165

Table 4
Effect of reactive power compensation devices on proposed scheme.

F.T F.L Op. MGPR RFD SRE coefficients Toper (ms)

Kqa Kqb Kqc
AG F4 56 173.4 1 0.07 0.09 68.6
65 36.7 1 0.05 0.06 72.0
BG F5 89 288.5 0.01 1 0.08 68.1
98 56.9 0.01 1 0.01 70.5
BC F2 34 188.1 0.05 1 0.95 68.5
43 53.08 0.07 1 0.93 70.7
ACG F5 89 523.2 1 0.03 0.59 67.5
98 103.5 1 0.07 0.98 69.5
ABG F1 46 365.5 0.48 1 0.09 67.7
64 359.0 0.48 1 0.07 67.7
ABC F4 56 352.4 0.82 1 0.83 67.7
65 74.35 0.82 1 0.84 70.0
ABCG F1 46 370.9 0.81 1 0.93 67.6
64 338.7 0.81 1 0.93 67.9

metry in fault currents. It can be depicted form Fig. 10 that the pro-
posed scheme successfully detects and locates the HIFs in the
study microgrid.
Fig. 11. Simulation results at MGPR-89 for three-phase HIF at Line-89 of the study A three-phase HIF is also simulated at Line-89 of the study
microgrid in islanded mode. microgrid to test the performance of proposed scheme in radial
configuration. The simulation results obtained at MGPR-89 are
shown in Fig. 11. It can be observed from Fig. 11 that the fault
fault current magnitude in the grid-connected mode is much
detection ratio and SRE coefficients are above the threshold and
higher than the islanded mode.
hence the proposed scheme has the ability to protect the radial
microgrid against HIFs.
4.2.2. Results for high impedance faults The results for selected scenarios in both grid-connected and
HIFs are difficult to detect in the microgrid particularly when islanded modes are given in Table 3. It can be observed from Table 3
HIFs occur in the islanded microgrid. Many utilities and research- that the proposed scheme successfully detects the HIFs but is
ers use 40 O resistance as the maximum reasonable resistance relatively slower as compared to solid faults in islanded and
for HIFs [21]. However, in this study, various cases with resistance grid-connected modes. Nevertheless, the speed of operation of
up to 200 O are considered to simulate more severe conditions. The the proposed scheme is still less than 0.1 s, which is acceptable
method used for modelling HIFs is adapted from [34]. for distribution systems protection.
To check the performance of proposed scheme for HIF in looped
configuration of microgrid, a DLG HIF is simulated at Line-46 of the 4.2.3. Effect of reactive power compensation
study microgrid in the islanded mode. The simulation results The proposed scheme uses the directional characteristics of SRE
obtained at MGPR-46 are depicted in Fig. 10. The variation in SRE for fault location and fault classification. The proposed scheme is
coefficient are because the HIFs introduce nonlinearity and asym- also tested on a system with reactive power compensation. To

Table 3
Simulation results of various types of high impedance faults at different locations in the microgrid working in the islanded mode and the grid connected mode.

F.L F.T Rf (O) Op. MGPRs Grid connected mode Islanded mode
RFD SRE coefficients Toper (ms) RFD SRE coefficients Toper (ms)
Kqa Kqb Kqc Kqa Kqb Kqc
F1 AG 50 46 35.85 1 0.07 0.04 71.4 13.30 1 0.01 0.05 72.6
64 39.38 1 0.08 0.03 71.1 9.25 1 0.01 0.02 72.8
100 46 19.55 1 0.09 0.08 72.4 8.78 1 0.01 0.01 72.9
64 20.49 1 0.09 0.09 72.3 6.40 1 0.01 0.01 73.0
200 46 11.23 1 0.02 0.08 72.7 5.18 1 0.02 0.02 73.1
64 11.23 1 0.03 0.06 72.8 4.05 1 0.02 0.01 73.2
F5 BCG 50 89 124.6 0.04 0.70 1 69.4 20.04 0.05 0.88 1 72.3
98 23.42 0.02 0.91 1 72.3 23.05 0.02 0.82 1 72.3
100 89 69.79 0.02 0.75 1 68.4 12.40 0.06 0.87 1 72.6
98 13.10 0.05 0.80 1 72.6 13.10 0.05 0.88 1 72.5
200 89 36.65 0.06 0.75 1 71.4 7.02 0.04 0.85 1 73.0
98 08.02 0.04 1 0.82 72.9 8.02 0.05 0.82 1 73.0
F3 ABCG 50 23 41.67 0.83 0.93 1 70.9 22.90 0.8 0.91 1 72.3
32 10.25 0.85 0.94 1 72.8 10.40 0.82 0.93 1 72.8
100 23 26.45 0.84 1 0.93 72.0 14.88 0.78 0.90 1 70.8
32 07.25 0.86 1 0.94 72.9 5.98 0.81 0.92 1 73.1
200 23 15.45 0.85 1 0.93 72.5 8.65 0.80 0.92 1 72.9
32 04.23 0.82 1 0.93 73.2 3.18 0.81 0.92 1 73.3
166 S.B.A. Bukhari et al. / Electrical Power and Energy Systems 92 (2017) 156–166

simulate this effect, three capacitor banks each of 500 kvar are generators. IEEE Trans Power Deliv 2007;22:634–41.
placed at buses 4, 6 and 7. The synchronous generator is discon-
[10] Mirsaeidi S, Mat Said D, Wazir Mustafa M, Hafiz Habibuddin M, Ghaffari K. An
nected from bus 7 in this case. The results are shown in Table 4. analytical literature review of the available techniques for the protection of
It can be observed from the table that the proposed scheme pro- micro-grids. Int J Electr Power Energy Syst 2014;58:300–6.
vides a stable SRE direction in the presence of reactive power com- 10.1016/j.ijepes.2014.01.032.
[11] Hosseini SA, Abyaneh HA, Sadeghi SHH, Razavi F, Nasiri A. An overview of
pensation devices. microgrid protection methods and the factors involved. Renew Sustain Energy
Rev 2016;64:174–86.
[12] Oudalov A, Fidigatti A. Adaptive network protection in microgrids. Int J Distrib
5. Conclusion Energy Resour 2009;5:201–26.
[13] Coffele F, Booth C, Dyśko A. An adaptive overcurrent protection scheme for
distribution networks. IEEE Trans Power Deliv 2015;30:561–8. http://dx.doi.
Microgrids have been developed to meet the power quality and org/10.1109/TPWRD.2013.2294879.
reliability requirements of customers. However, the development [14] Mirsaeidi S, Mat Said D, Mustafa MW, Hafiz Habibuddin M. A protection
of microgrids has been complemented with major challenges. strategy for micro-grids based on positive-sequence component. IET Renew
Power Gener 2015;9:600–9.
One major problem related to microgrids is the need to design
[15] Oureilidis KO, Demoulias CS. A fault clearing method in converter-dominated
an effective protection strategy which can work in both the grid- microgrids with conventional protection means. IEEE Trans Power Electron
connected and the islanded modes of operation. This paper pro- 2016;31:4628–40.
posed a protection scheme for microgrids based on SRE. The SRE [16] Li X, Dyśko A, Burt GM. Traveling wave-based protection scheme for inverter-
dominated microgrid using mathematical morphology. IEEE Trans Smart Grid
was computed by the Hilbert transform. A new ratio based on 2014;5:2211–8.
the sequence components of superimposed current and prefault [17] Saleh SA, Ahshan R, Abu-Khaizaran MS, Alsayid B, Rahman MA. Implementing
current was developed to detect both solid and high-impedance and testing–wpt-based digital protection for microgrid systems. IEEE Trans Ind
Appl 2014;50:2173–85.
faults in the microgrid. The directional characteristics of SRE were [18] Tumilty RM, Brucoli M, Burt GM, Green TC. Approaches to network protection
adopted to identify the faulty section. The proposed scheme does for inverter dominated electrical distribution systems. In: Power electron
not need adaptive relay setting for islanded and grid-connected mach drives 2006 3rd iet int conf on, IET; 2006. p. 622–6.
modes of operation. Finally, to validate the efficacy of the sug- [19] Al-Nasseri H, Redfern MA, Li F. A voltage based protection for micro-grids
gested protection scheme, several simulations were performed. containing power electronic converters. In: 2006 IEEE power eng soc gen
Simulation results indicated that the suggested scheme can effi- meet. IEEE; 2006. p. 1–7.
[20] Jayawarna N, Jones C, Barnes M, Jenkins N. Operating microgrid energy storage
ciently protect the microgrid against solid and high-impedance control during network faults. In: 2007 IEEE int conf syst syst eng. IEEE; 2007.
faults in both radial and looped configurations. p. 1–7.
[21] Zamani MA, Yazdani A, Sidhu TS. A communication-assisted protection
strategy for inverter-based medium-voltage microgrids. IEEE Trans Smart
Acknowledgement Grid 2012;3:2088–99.
[22] Casagrande E, Woon WL, Zeineldin HH, Svetinovic D. A differential sequence
component protection scheme for microgrids with inverter-based distributed
This work was supported by the National Research Foundation
generators. IEEE Trans Smart Grid 2014;5:29–37.
of Korea (NRF) grant funded by the Korea government (MSIP) (No. TSG.2013.2251017.
2015R1A2A1A10052459). [23] Mishra DP, Samantaray SR, Joos G. A combined wavelet and data-mining based
intelligent protection scheme for microgrid. IEEE Trans Smart Grid
References [24] Benmouy G, Roberts J. Superimposed quantities: their true nature and
application in relays. WA USA SEL USA: Schweitzer Engineering Laboratories,
[1] Lin X, Zhang R, Tong N, Li X, Li M, Yang D. Regional protection scheme designed Inc. Pullman; 1999.
for low-voltage micro-grids. Int J Electr Power Energy Syst 2015;64:526–35. [25] Poularikas AD. Transforms and applications handbook. CRC Press; 2010. [26] Cui T, Dong X, Bo Z, Juszczyk A. Hilbert-transform-based transient/intermittent
[2] Mathur A, Das B, Pant V. Fault analysis of unbalanced radial and meshed earth fault detection in noneffectively grounded distribution systems. IEEE
distribution system with inverter based distributed generation. Int J Electr Trans Power Deliv 2011;26:143–51.
Power Energy Syst 2017;85:164–77. TPWRD.2010.2068578.
ijepes.2016.09.003. [27] Luo S, Dong X, Shi S, Wang B. A directional protection scheme for HVDC
[3] Bayat A, Bagheri A, Noroozian R. Optimal siting and sizing of distributed transmission lines based on reactive energy. IEEE Trans Power Deliv
generation accompanied by reconfiguration of distribution networks for 2016;31:559–67.
maximum loss reduction by using a new UVDA-based heuristic method. Int J [28] Orozco-Henao C, Bretas AS, Leborgne RC, Herrera A, Martinez S. Fault location
Electr Power Energy Syst 2016;77:360–71. in distribution network with inverter-interfaced distributed energy resources
ijepes.2015.11.039. in limiting current. In: Harmon qual power ICHQP 2016 17th int conf on, IEEE;
[4] Ma J, Wang X, Zhang Y, Yang Q, Phadke AG. A novel adaptive current protection 2016. p. 231–6.
scheme for distribution systems with distributed generation. Int J Electr Power [29] Saadat H. Power system analysis. WCB/McGraw-Hill; 1999.
Energy Syst 2012;43:1460–6. [30] Mirsaeidi S, Said DM, Mustafa MW, Habibuddin MH, Ghaffari K. Fault location
[5] Tan X, Li Q, Wang H. Advances and trends of energy storage technology in and isolation in micro-grids using a digital central protection unit. Renew
microgrid. Int J Electr Power Energy Syst 2013;44:179–91. Sustain Energy Rev 2016;56:1–17.
10.1016/j.ijepes.2012.07.015. rser.2015.10.162.
[6] Gu W, Wu Z, Bo R, Liu W, Zhou G, Chen W, et al. Modeling, planning and [31] Johns AT, Salman SK. Digital protection for power systems. IET; 1997.
optimal energy management of combined cooling, heating and power [32] Mirsaeidi S, Said DM, Mustafa MW, Habibuddin MH, Ghaffari K. A protection
microgrid: A review. Int J Electr Power Energy Syst 2014;54:26–37. http:// strategy for micro-grids based on positive-sequence impedance. Distrib Gener Altern Energy J 2016;31:7–32.
[7] Bakar AHA, Ooi B, Govindasamy P, Tan C, Illias HA, Mokhlis H. Directional 21563306.2016.11744002.
overcurrent and earth-fault protections for a biomass microgrid system in [33] Zamani MA, Yazdani A, Sidhu TS. A control strategy for enhanced operation of
Malaysia. Int J Electr Power Energy Syst 2014;55:581–91. inverter-based microgrids under transient disturbances and network faults.
10.1016/j.ijepes.2013.10.004. IEEE Trans Power Deliv 2012;27:1737–47.
[8] Laaksonen HJ. Protection principles for future microgrids. IEEE Trans Power TPWRD.2012.2205713.
Electron 2010;25:2910–8. [34] Sortomme E, Venkata SS, Mitra J. Microgrid protection using communication-
[9] Nimpitiwan N, Heydt GT, Ayyanar R, Suryanarayanan S. Fault current assisted digital relays. IEEE Trans Power Deliv 2010;25:2789–96. http://dx.doi.
contribution from synchronous machine and inverter based distributed org/10.1109/TPWRD.2009.2035810.