IET Generation, Transmission & Distribution

Research Article

Assessment of PMU-based wide-area angle ISSN 1751-8687

Received on 7th January 2019

criterion for fault detection in microgrid

Revised 23rd July 2019
Accepted on 30th July 2019
E-First on 9th September 2019
doi: 10.1049/iet-gtd.2019.0027
www.ietdl.org

Nikhil Kumar Sharma1 , Subhransu Ranjan Samantaray1

1School of Electrical Sciences, Indian Institute of Technology, Bhubaneswar, India
E-mail: [email protected]

Abstract: The penetration of distributed generation in the power distribution system has increased significantly in recent years,
bringing serious protection challenges. Hence, it is mandated to develop effective and reliable fault detection scheme for
present-day microgrid environment. This study presents a phasor measurement unit-based (PMU) wide-area phase angle
criterion for building a protection system for the microgrid. The absolute value of the rate of change of voltage phase angle
difference between the point of common coupling and the bus closest to the fault point is utilised for fault detection in grid
connected as well as in the islanded modes of operation. C37.118.1 complied PMUs are employed at the appropriate locations
for accurate and faithful measurement of voltage phase angle and the rate of angle difference (RAD) for detecting faulted
conditions. PMUs are connected to phasor data concentrator, which receives the time synchronised phasor data and acts as a
centralised processing unit. The proposed protection index has been extensively validated on MATLAB/Simulink environment as
well as the real-time digital simulator with external PMUs. The test results show that the proposed RAD is a potential candidate
for providing effective fault detection measure for microgrid including large variations in operating conditions.

1 Introduction distance-based fault detection scheme for microgrid protection.

This scheme was proposed to mitigate the issue of intermittent
The demand for clean and economical energy sources is increasing nature of DGs in the protection system. However, this scheme is
exponentially due to the rise in environmental pollution level by restricted only to the microgrids with inverter-based DGs. Ma et al.
the conventional energy generating sources, depletion of the [13] have proposed scheme based on steady-state fault current
conventional sources and increase in the energy demand over the deviation for primary and backup protection. The performance of
years. Hence, renewable and green energy sources such as solar the proposed scheme is not satisfactory for high resistance fault.
panels, wind turbines, fuel cell, and biomass are becoming more Sadeghkhani et al. [14] have proposed a fault detection technique
popular to address the solution of these problems [1]. The rapid based on transient monitoring function derived from the inverter
integration of these sources to the existing power network can current data. The accuracy of fault detection in this scheme is
make the system more sustainable; however, it creates serious dependent on the length of the window.
challenges for protective relaying. The conventional protection To alleviate the aforementioned problems in microgrid
schemes are becoming unreliable considering the present microgrid protection, wide-area measurement can be a possible platform for
environment and wide variations in operating conditions [2–4]. building reliable measure. The trend of using the phasor
The non-directional overcurrent relays used in the distribution measurement units (PMUs) in the microgrid for fault detection and
system may fail to detect the fault, as power flow in the protection purpose is increasing nowadays. In such an attempt,
conventional grid is unidirectional, whereas it is bidirectional in Pignati et al. [15] have proposed a real-time fault detection scheme
case of microgrids with the integration of renewables [5]. A by computation of PMU-based state estimator. The test bed used
microgrid consists of several distributed generators, which include was a real 10 kV, three-phase distribution system of Netherlands
induction generator, synchronous generator, and inverter-based with DG integration having 18 buses and PMUs were installed in
distributed generations (DGs). In case of abnormality, the fault all the buses. This scheme used the real-time data obtained from
current contribution of the DGs depends on the type of DG PMU to detect and locate the fault, which has the advantage of low
technology. The synchronous generator-based DGs contribute latency, high refresh rate, and accuracy but has a high response
much higher fault current than inverter-based DGs, as later restricts time. In a similar attempt, PMU data were used for fault detection
the fault current to lower values to protect the power electronics in [16]. A rule-based scheme for microgrid protection using PMU
components [6, 7]. The microgrid has two modes of operations: measurement was illustrated in [17]. A sequence components based
grid-connected mode and islanding mode. The fault current level in differential scheme for fault detection is suggested in [18], which is
islanding mode is much lower than the fault current level in grid- only effective for the microgrid integrated with inverter-based
connected mode. Lower fault current level creates problems in DGs. The installation cost of the PMU is considered as one of the
detection of faulted conditions and affects the relay coordination major hindrances in employing it to the distribution grid. Meier et
[8]. al. [19, 20] have introduced the concept of micro-PMU for
Furthermore, the network topology and operating conditions of building wide-area-based protection system for distribution level
the microgrid may change to meet the energy demand. The fault microgrid. The micro-PMUs have all the features of conventional
current also varies significantly with the change in the topology of PMUs, and due to low cost it can find wider application in
the microgrid. The relay used for the protection should be capable distribution systems. Thus, the proposed method extends the PMU
of detecting faults in these variations in topology. The conventional application in building an effective and reliable protection scheme
fault detection schemes based on the overcurrent principle [9], for the microgrid.
harmonic distortion of the current [10], and distance-based relaying The methodology proposed in this paper detects the fault in the
scheme [11] have also been proposed for microgrids. Most of these microgrid by monitoring the absolute value of rate of angle
schemes are useful only for radial topology and for unidirectional difference (RAD) of phase voltage between point of common
flow of power. Arroudi and Joós [12] have proposed a modified coupling (PCC) and bus nearest to the faulted line using IEEE

IET Gener. Transm. Distrib., 2019, Vol. 13 Iss. 19, pp. 4301-4310 4301
© The Institution of Engineering and Technology 2019
 17518695, 2019, 19, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-gtd.2019.0027 by Bangladesh Hinari NPL, Wiley Online Library on [16/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
denoted by VF and VS, respectively. The circuit shown in Fig. 1 is a
voltage divider model, and hence the pre-fault voltage VFP can be
calculated as

Z2 + ZL
V FP = V (1)
Z1 + Z2 + ZL S

The fault impedance and fault voltage are included in the circuit
after the occurrence of the fault. Therefore, the post-fault voltage
VF can be calculated as

ZF ∥ (Z2 + ZL)
VF = V
Z1 + ZF ∥ (Z2 + ZL) S
Fig. 1 Model of three-phase fault in a microgrid for the analysis of phase
jump Here, Z2 + ZL can be represented as Z2L. Hence, the above equation
can be rewritten as
C37.118.1 complied PMUs [21, 22]. Further assessment is carried
out using different PMU locations for the same fault situations and VF =
ZF ∥ (Z2L)
V (2)
behaviour of the relaying scheme for different types of faults. Z1 + ZF ∥ (Z2L) S
Owing to recent developments of low-cost micro-PMUs, the
proposed paper finds easy implementation by placing micro-PMUs The above equation can be simplified further as
at the selected bus of the microgrid. The proposed scheme attempts
to assess the performance of the RAD using fewer numbers of ZFZ2L
PMUs across the microgrid. Therefore, six PMUs are deployed in VF = V (3)
Z1Z2L + ZF Z1 + Z2L S
the present 15-bus microgrid including PCC and critical load
centres to monitor the complete microgrid. It is observed that with Applying (1) in (3), the fault voltage can be represented as
the PMUs placed at strategic locations, the performance of the
proposed relaying scheme is highly promising. However, the G − jB
optimal placement of PMUs can be one of the future prospective VF = V
G + Gf − jB FP (4)
measures for developing the protection scheme. The proposed
scheme has been extensively tested on the microgrid for wide If V FP is considered as reference phasor having magnitude 1.0 pu
variations in operating conditions including grid connected as well
and angle 0°, phasor diagram for various fault conditions can be
as the islanded mode, for all types of faults, fault location, and
drawn based on unit circles. The conductance is denoted by G, the
varied fault resistance and found to be highly effective. The
susceptance is denoted by B for corresponding impedances, and the
proposed relaying scheme is examined for its insensitivity toward
fault conductance is denoted by Gf. In general, the combined
critical no-fault cases and the impact of noise in the signal is also
analysed. impedance of the line and the load can be considered higher than
This paper is organised as follows: Section 2 deals with the the impedance of source, i.e. Z1≪Z2L, and thus (3) can be further
insight of the proposed fault detection scheme. Section 3 provides simplified as follows:
details about the studied microgrid. Section 4 describes the result
and analysis and Section 5 presents the validation of the proposed VF =
ZF
V (5)
scheme on the real-time digital simulator (RTDS) platform with Z1 + ZF S
external PMUs. The concluding remarks about this research work
are given in Section 6. The source voltage can be assumed as 1 pu. The impedances Z1
and ZF are given as
2 Proposed fault detection scheme
Z1 = R1 + JX1
A fault condition in an electric power system creates a drop in the
system voltage. The voltage is a complex quantity, and hence in the
event of a fault, the change in the voltage is not limited to the ZF = RF + JXF
reduction in the magnitude. It also changes the phase angle, which
is known as phase angle jump [23, 24]. To monitor the operating Equation (5) can be represented in vector form as
condition of the system and to identify the faults, six PMUs are
installed at various locations. These PMUs monitor the voltage V¯ F =
ZF
1∠0° (6)
signal continuously and the desired parameters are obtained from Z1 + ZF
the post-processing of the signals.
The phase angle jump of phase voltage is given as
2.1 Angle-based fault detection scheme
del Φ = arg V F
The phase angle jump is defined as the sudden increase in the
voltage phase angle if the system is subjected to a fault. To analyse
XF X1 + XF
the concept of the phase angle jump, a model of three-phase-to- del Φ = tan−1 − tan−1 (7)
ground fault is shown in Fig. 1. During the fault, phase angle jump RF R1 + RF
occurs either due to the difference in the reactance to the resistance
ratio between the source and the feeder or due to the transformation It is observed from (7) that in the event of a fault, there exists phase
of voltage sags to lower-voltage levels [23]. angle difference between the buses connecting the faulty line.
It is assumed that a three-phase fault has occurred at line joining Hence, the rate of phase angle difference is used as the key
buses B1 and B2 as shown in Fig. 1. The equivalent impedance of indicator for fault detection in this research work. The absolute
source and line in front of the fault point F is denoted by Z1. The RAD can be calculated from the equation below as:
impedance of the line joining the fault point to the load is denoted
by Z2 and ZL represents the load impedance, which is assumed to d del Φ
RAD = (8)
dt
be constant. The voltage across the fault and the source voltage is

4302 IET Gener. Transm. Distrib., 2019, Vol. 13 Iss. 19, pp. 4301-4310
© The Institution of Engineering and Technology 2019
 17518695, 2019, 19, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-gtd.2019.0027 by Bangladesh Hinari NPL, Wiley Online Library on [16/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

Fig. 2 Analysis and modeling for the proposed fault detection scheme
(a) Variation of phase voltage with fault resistance for a three-phase fault, (b) Schematic diagram of the proposed scheme (c) Class P PMU adaptive algorithm [22], (d) Studied
microgrid test bed [25]

Equations (5) and (7) can be used to plot the variation of fault can be observed that the magnitude of the phase voltages, as well
voltage with the fault resistance in case of different fault as their angle, changes according to the corresponding semi-circles.
conditions. The variation of fault voltage in three-phase fault is A similar figure can be plotted for all types of phases and ground
shown in Fig. 2a [24]. Since the fault is balanced in nature, all faults.
phase voltages are equal in magnitude and displaced by 120° with The absolute value of the RAD between phase voltages of two
each other. The fault resistance Rf = ∞ is marked as points p, q, and buses is considered as the indicator for fault detection. The
r in Fig. 2a. Point O is the coordinate for zero fault resistance. It threshold for the proposed RAD-based relay is selected after

IET Gener. Transm. Distrib., 2019, Vol. 13 Iss. 19, pp. 4301-4310 4303
© The Institution of Engineering and Technology 2019
 17518695, 2019, 19, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-gtd.2019.0027 by Bangladesh Hinari NPL, Wiley Online Library on [16/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 3 Absolute value of phase angle difference and RAD
(a) Phase angle difference, (b) RAD in grid-connected mode during three-phase fault at the mid-point of line DL9 with the measurements at PCC and B6, (c) Phase angle difference,
(d) RAD in grid-connected mode during three-phase fault at the mid-point of line DL5 with the measurements at PCC and B11

extensive test cases including faulty situations under wide the circuit is bidirectional, the circuit breakers are installed on both
variations in operating conditions, fault types, fault resistances, and ends of the lines to isolate the faulty section.
faults at different locations of line. The threshold of 0.2 rad/s is set The details of utility, transformer ratings, DGs voltage level,
considering the worst-case scenario of a three-phase fault created at power ratings, and load ratings are specified in [25]. The Simulink
the mid-point of DL9 in grid-connected mode with fault resistance implementations of the DG are detailed models, and the loads
of 100 Ω and the PMU measurements were taken between the considered are constant impedance loads. To monitor the health of
buses PCC and B9, which is located away from the faulted line. the entire system, six numbers of PMUs are installed at buses PCC,
The relay is set to issue the tripping command instantaneously after B3, B6, B9, B11, and B14. The PMUs monitor the voltage and
RAD crosses the threshold. current phasors continuously. This complete setup is modelled and
The proposed scheme implementation is shown in Fig. 2b. Two simulated using SimPower system blocks with the power
PMUs are kept at the two buses of the line segment. The absolute component in the Simulink platform.
value of the phase voltage angle difference and the RAD of the
phase voltage is calculated from the information retrieved by IEEE 4 Result and analysis
C37.118.1 complied PMUs. The RAD between the two buses is
compared with a set threshold value. If RAD exceeds the set value, As mentioned earlier, six PMUs are placed at buses including PCC,
a tripping command is issued to the circuit breaker to isolate the B3, B6, B9, B11, and B14, which monitor the complete topology
faulty line. of the microgrid. The total DG capacity is 17.16 MW and the total
load connected is 14.15 MW and 2.34 MVAR. The proposed paper
2.2 PMU description uses point-to-point PMU phasor information to detect the fault
inception.
The proposed paper uses IEEE C37.118.1 complied extended A three-phase-to-ground Three-phase to ground fault (LLLG)
Kalman filter (EKF) PMU algorithm for wide-area monitoring, fault is created at the mid-point of the line DL9 at 0.1 s in the grid-
which has been extensively assessed and characterised in [22]. The connected mode of operation and fault resistance is fixed to 0.001
performance of the EKF-based PMU is best suited for fundamental Ω. The absolute value of phase angle difference and RAD between
phasor estimation of the fault transient as found and discussed in PCC and bus B6 is shown in Figs. 3a and b. It is observed from the
[22]. The C37.118.1 standard defines two classes of PMUs based results that the absolute phase angle jump reaches 0.75 rad
on performance: class P and class M. Class P PMU is used for this (maximum) and RAD reaches 0.25 rad/s (maximum) following the
paper, as it has a fast response time, which is essential for fault inception. Similar fault analysis is shown in Figs. 3c and d,
protection application. The schematic representation of the when the fault is occurring at the mid-point of line DL5, and the
adaptive class P PMU algorithm is shown in Fig. 2c. RAD is computed from the PMU data of PCC and B11.
In the practical application, the signals will always be polluted. The change in voltage phase angle and RAD are found to be 1
Hence, it is necessary to filter the harmonics and DC offset to rad and 0.59 rad/s, respectively, after one cycle of fault inception. It
extract the fundamental component of the signal. The filter is observed that the RAD is the index, which grows rapidly
parameters should be adaptable to any change in the frequency of following the fault inception and a threshold can be set to issue the
the harmonic components. An anti-alias filter is used in the model tripping signal.
of PMU, for filtering out the higher-order frequency component. To analyse the effect of the fault in islanding mode of operation,
This filter is set to 3 dB cut off at 720 Hz, and the sampling rate is the main circuit breaker is opened. Figs. 4a and b show the
fixed to 60 frames/s. This filter can track as high as 11th harmonic, simulation results when the fault is occurring at the mid-point of
which is sufficient for the protection requirements. the line DL9 and measurements are taken at PCC and B6. Figs. 4c
and d shows the results for the three-phase fault on DL5 and
3 Studied microgrid test bed measurements considered are between buses PCC and B11. It is
observed that RADs in Figs. 4b and d are high even in the
The medium-voltage microgrid simulated and studied for the islanding mode of operation. The maximum magnitude of absolute
present research work is shown in Fig. 2d. The voltage level of this phase angle difference and RAD are 1.75 rad and 0.7 rad/s,
system is 12.47 kV at an operating frequency of 60 Hz. In this grid, respectively.
the utility is connected to PCC through a circuit breaker to create Similar observations are made for the three-phase fault at the
the grid-connected or islanding modes of operation. The grid mid-point of line DL5 in islanding mode. It can be further observed
capacity is fixed to 100 MW. This microgrid consists of three DGs from the results that following fault inception, the angle difference
(wind farm, solar farm, and hydro generator), five transformers, and RAD between the PCC and the bus close to the faulted line
and several loads connected to the buses. Since the power flow in section increase rapidly. The absolute value of the RAD, which is

4304 IET Gener. Transm. Distrib., 2019, Vol. 13 Iss. 19, pp. 4301-4310
© The Institution of Engineering and Technology 2019
 17518695, 2019, 19, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-gtd.2019.0027 by Bangladesh Hinari NPL, Wiley Online Library on [16/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 4 Absolute value of phase angle difference and RAD
(a) Phase angle difference, (b) RAD in islanding mode during three-phase fault at the mid-point of line DL9 with the measurements at PCC and B6, (c) Phase angle difference, (d)
RAD in islanding mode during three-phase fault at the mid-point of line DL5 with the measurements at PCC and B11

Table 1 Impact of PMU location on response time occurring at the mid-point of different lines keeping fault resistance
Faulted line PMU location Operating time, ms fixed to 0.001 Ω. In many cases, the RAD reaches the threshold as
Grid Island fast as 21 ms (∼1.5 cycles). Table 2 lists similar kinds of
measurements for all types of symmetrical and asymmetrical faults.
DL2 PCC-B3 24 23
It is observed that the response time is well within the permissible
PCC-B6 33 25 limit as required by the protection relay. Hence, the proposed
PCC-B9 25 24 scheme has the capability to detect all types of faults in the
PCC-B11 25 24 microgrid. In all cases, it is found that even though the PMU
B3-B14 22 23 measurements are made between buses away from the faulted line,
DL9 PCC-B3 35 25 the fault can be detected within the desired response time.
PCC-B6 23 21 Figs. 5a and b illustrate the RAD values for different PMUs in
grid-connected and islanded modes of operations of microgrid,
PCC-B9 40 21
respectively, in case of a three-phase fault at the mid-point of line
PCC-B11 40 22 DL9 with fault resistance of 0.001 Ω and fault inception at 0.1 s. It
B3-B6 60 27 is observed that the RAD values between the PCC and the PMUs
DL5 PCC-B6 31 21 located at the buses B3, B6, B9, and B11 cross the threshold with
PCC-B9 22 21 the different response time. Hence, PMUs located other than the
PCC-B11 21 21 faulty sections can also be used for fault detection with delayed
B9-B11 27 32
response time varying between 23 and 40 ms. It is further observed
that PMU located on bus B14 is unable to identify this fault
B3-B11 25 30
situation (for both grid-connected and islanding modes) as the
DL13 PCC-B14 22 20 section is completely out of the fault loop.
B3-B14 22 18 To assess the performance of the proposed RAD-based relaying
B11-B14 23 18 scheme for different types of faults, all types of lines and ground
faults are impressed at the mid-point of line DL3 and PMU
measurements are taken between PCC and B3. The fault inception
Table 2 Impact of types of faults on response time time is 0.1 s and fault resistance considered is 0.001 Ω. It is
Fault type Faulted line PMU location Operating time, ms observed from Figs. 5c and d that the RAD crosses the threshold of
Grid Island 0.2 rad/s for all types of faults for both islanded and grid-connected
modes of operations with a variation in the response time between
LLLG DL2 PCC-B3 24 23 23 and 32 ms. Fault-type identification is possible as the RAD for
DL5 PCC-B9 22 21 the respective faulty phase will be above the threshold indicating
DL13 PCC-B6 22 20 the involvement of fault. By this way, the phase involved with the
LLG DL9 PCC-B3 29 25 fault process can be detected providing fault type.
DL9 PCC-B6 26 25 The fault current magnitude reduces to lower value in case of
DL13 B11-B14 20 21 high resistance faults. Hence, detecting the high resistance fault is a
challenging issue in microgrids. The proposed protection scheme is
LL DL2 PCC-B9 19 18
tested for three-phase fault at mid-point of line DL5 with
DL5 PCC-B6 18 17 measurements from the PMUs at PCC and B11 with wide variation
DL13 B3-B14 20 18 in fault resistance in grid-connected mode and the islanded mode
LG DL2 PCC-B11 30 27 of operation. The results are shown in Figs. 6a and b. It is observed
DL2 B3-B14 32 38 that the RAD-based relay is able to cross the threshold even for
DL5 PCC-B11 25 24 faults with fault resistance of 100 Ω with a response time of 29 ms
in grid-connected mode and 27 ms in islanding mode of operation.
This shows the robustness of the proposed scheme for identifying
considered as the primary criterion for fault detection, jumps up to faults with wide variation in fault resistance, as it is based on angle
0.85 rad/s (48.70°/s) in some conditions. criteria instead of the magnitude of fault current.
As the speed of the relay (response time) and accuracy are The proposed scheme uses the IEEE C37.118.1 complied PMUs
prime requirements for reliable fault detection, the analysis is with EKFs for removing noise and polluting signals from the
carried out for different fault cases with wide variations in voltage and current phasors. The susceptibility of the proposed
operating conditions. Table 1 shows variation in response time with scheme for noise interference is tested by adding Gaussian noise of
the location of PMUs when a three-phase-to-ground fault is 2.5% standard deviation [26] (10 dB signal-to-noise ratio) to the

IET Gener. Transm. Distrib., 2019, Vol. 13 Iss. 19, pp. 4301-4310 4305
© The Institution of Engineering and Technology 2019
 17518695, 2019, 19, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-gtd.2019.0027 by Bangladesh Hinari NPL, Wiley Online Library on [16/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 5 Impact of measurements from different PMUs and fault types on RAD
(a) Variation in RAD for different PMUs during three-phase fault at the midpoint of line DL9 in grid-connected mode, (b) in islanding mode and (c) RAD values for different faults
at the midpoint of line DL3 and measurement at PCC and B3 in grid-connected mode, (d) in islanding mode

Fig. 6 Impact of fault resistance and noise on RAD

(a) Variation in RAD with fault resistance during three-phase fault at the midpoint of line DL5 with measurement at PCC and B11 in grid-connected mode, (b) in islanding mode, (c)
Phase A voltage signal of bus B11 with and without noise and (d) impact of noise on RAD during three-phase fault at the midpoint of line DL5 with measurement at PCC and B11

voltage and current signals of buses PCC and B11. The phase A performance comparison among the schemes for a three-phase-to-
voltage signal of bus B11 with and without noise is shown in ground fault at the mid-point of line DL3 with variations in fault
Fig. 6c. The LLLG fault is simulated at the mid-point of the line resistance. It is observed that the existing methods fail in case of
DL5 with fault resistance of 0.001 Ω, and RAD is computed by the high resistance faults, whereas the proposed scheme succeeds. This
voltage signals of buses PCC and B11. It is observed from Fig. 6d happens as in the case of high resistance faults, and the fault
that the PMUs deployed for measurement are capable of filtering current may fall below the set threshold. Thus, the existing
the superimposed noise signals, and hence the impact of noise in schemes fail due to either not meeting their standard threshold (for
the proposed scheme is negligible. the differential scheme) or response time (overcurrent relay), which
For comparative assessment, the proposed scheme is compared are primary indicators for reliable and efficient microgrid
with the widely used overcurrent and differential scheme. The protection.
overcurrent relay setting was kept at 120% [4] of the rated current, An extensive comparison of the proposed scheme with the other
and the differential relay was set to operate when the difference in existing schemes is shown in Table 4. It is found that the proposed
the current exceeds 10% of rated value [27]. Table 3 shows the methodology outperforms the existing schemes ([10, 13–15]) in

4306 IET Gener. Transm. Distrib., 2019, Vol. 13 Iss. 19, pp. 4301-4310
© The Institution of Engineering and Technology 2019
 17518695, 2019, 19, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-gtd.2019.0027 by Bangladesh Hinari NPL, Wiley Online Library on [16/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 3 Comparison of the proposed methodology with existing overcurrent and differential current-based schemes (LLLG
fault at DL3)
Fault R, Ω Mode of operation Proposed scheme Overcurrent scheme [4] Differential scheme [27]
0.001 grid yes yes yes
0.001 island yes yes yes
1 grid yes yes yes
1 island yes yes yes
20 grid yes yes yes
20 island yes no no
50 grid yes no no
50 island yes no no
100 grid yes no no
100 island yes no no
Yes – relay operates and no – relay does not operate subject to set threshold and desired response time.

Table 4 Extended comparison of the proposed methodology with other existing schemes
Comparison aspect Proposed Petit et al. [10] Ma et al. [13] Sadeghkhani et al. [14] Pignati et al. [15]
scheme
test bed medium-voltage medium-voltage MG medium-voltage MG low-voltage MG (0.4 kV) medium-voltage
Microgrid (MG) (20 kV) (10 kV) MG (10 kV)
(12.47 kV)
parameter for fault RAD symmetrical overcurrent scheme inverter current-based voltage-based
detection component of the monitoring function weighted least-
current square estimator
application of yes no no no yes
synchrophasors
protection against high yes yes no no yes
resistance fault
protection against far-end yes no yes no yes
fault
grid-connected mode yes yes yes yes yes
islanding mode yes no no yes no
observations same relay setting suitable for backup performance is not response time and high response time
for all operating protection satisfactory for high accuracy depends on the
conditions resistance fault window length

case of critical fault condition including high resistance fault and from 10 to 35% of the grid capacity by varying the generation level
remote end faults. Furthermore, it is found that the proposed of DG-1 (hydro plant) only and the result is shown in Fig. 7b.
scheme is also able to perform for all types of faulted conditions at Similarly, the fault study is carried out for line DL9 with
both the islanded and grid-connected modes, whereas the existing measurements from PCC and B6 with variations in DG-2 (wind
methods fail to detect the fault condition in the islanding mode of turbine) penetration. From the results shown in Fig. 7c, it is
operation and with different types of DGs. The comparison is made observed that with increase in DG penetration, the RAD value
with a restriction on the response time of 30 ms for effective increases and for all the levels of penetration, the relay is capable
relaying decision. of detecting the fault. Therefore, it can be summarised that the DG
To test the performance of the proposed relaying scheme for control strategy does not affect the performance of the proposed
certain cases, which may malfunction the relay operation, some scheme.
critical no-fault tests such as capacitor switching, load switching,
and motor starting are performed. The change in the voltage and 5 Validation on RTDS
the current magnitude under these conditions are very high, which
may result in the false operation of the protection schemes. A RTDS is a powerful tool to simulate and study the power system
capacitor bank of rating 500 kVar and an induction motor of rating contingencies. It represents the behaviour of a real power system
0.1 MVA is connected at one end of the line DL5 as a load to being modelled [28]. To test the effectiveness of the proposed
simulate the effect of capacitor bank switching and motor starting, scheme, validation is carried out on the RTDS hardware installed at
respectively. The load switching is studied by increasing the load to the School of Electrical Sciences, Indian Institute of Technology,
150%. The mentioned non-faulty critical cases were initiated at Bhubaneswar. Fig. 7d shows the medium-voltage microgrid
0.1 s and cleared at 0.15 s. The variation in RAD computed from modelled using RSCAD, which runs on the hardware of RTDS.
the data of PMUs placed on buses PCC and B11 is shown in The voltage rating and the frequency considered are 12.47 kV and
Fig. 7a. It is observed that the change in the RAD is small and far 60 Hz, respectively. Circuit breakers are installed at both ends of
below the threshold in all these situations. Hence, it can be the lines to isolate the faults. The internal C37.118 complied
concluded that the proposed scheme remains insensitive to these GTNET PMUs, available in the RTDS, are used to retrieve the
no-fault conditions. phasor information. The list of various components with their
The microgrid shown in Fig. 2d includes both the synchronous- specifications are shown in Table 5.
based and inverter-based distributed energy resources, which work Extensive tests are conducted to validate the proposed RAD-
on different DG control strategies. The impacts of different control based fault detection scheme. Fig. 8a shows the RAD between
strategies on the proposed scheme are validated by types of DGs buses PCC and B2 when three-phase-to-ground fault occurs on the
and penetration level separately. An LLLG fault is created at the line DL2 with fault resistance 0.001 Ω. The fault was created at the
mid-point of line DL13, and the RAD is measured from the PMUs instant t = 0.1 s at grid-connected mode. It is observed that the
placed at PCC and bus B13. The total DG penetration is varied RAD crosses a set threshold value of 0.2 rad/s within 30 ms from

IET Gener. Transm. Distrib., 2019, Vol. 13 Iss. 19, pp. 4301-4310 4307
© The Institution of Engineering and Technology 2019
 17518695, 2019, 19, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-gtd.2019.0027 by Bangladesh Hinari NPL, Wiley Online Library on [16/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 7 Variation in RAD with no-fault cases and DG penetration
(a) Impact of critical no-fault cases on RAD-based relaying scheme, (b) Impact of variation in DG-1 (hydro) penetration, (c) Impact of variation in DG-2 (wind) penetration, (d)
Microgrid test system on the RTDS platform

Table 5 System components with specification

Component Specification
grid 5000 MW, 115 kV, 60 Hz
solar plant 1.74 MW, 230 V
wind generator 2 MW, 230 kV
transformer (TR1) 50 MVA, 115/12.47 kV
transformer (TR2) 100 MVA, 12.47/0.23 kV
transformer (TR3) 1 MVA, 12.47/0.23 kV
loads L1, L2, L3, and L4 0.6 MW, 0.2 MVAR
lines (DL1, DL2, DL3, and DL4) length 5 km, ground resistance 50 Ω, positive sequence resistance 0.1854 Ω/km, inductive reactance 0.3766
Ω/km, and capacitive reactance 0.546 MΩ/km

the occurrence of the fault. A similar observation was made for from the PMUs are sent to phasor data concentrator (enhanced
fault in islanding mode of operation, as shown in Fig. 8b, with a phasor data concentrator) which is retrieved by MATLAB in the
response time of 24 ms. Table 6 shows the impacts of PMU workstation. Three PMUs of the external PMU panel were placed
locations, variations in fault resistance, fault type, and fault at the buses PCC, B2, and B4.
location on the RAD, and it is observed that the response time is A line–ground fault with fault resistance of 0.001 Ω is created
well within the desired range. at the mid-point of line DL2, and RAD is found out between PCC
The test results obtained from both Simulink and RTDS and B2. It is observed from Fig. 8c that the fault is created at
platform show that the proposed RAD-based index can reliably 17:51:40:000 Indian standard time and the relay crosses the
identify the fault situations with wide variations in the operating threshold value of 11.50°/s (∼0.2 rad/s) in 28 ms. A similar fault
condition and faults with a response time of two cycles (maximum) condition is created at the mid-point of line DL4, and RAD is
following fault inception. The communication between the PMUs found out between PCC and B4 shown in Fig. 8d. The response
can be established by using IEC 61850 point-to-point time of the relay is found to be 29 ms for the same threshold. Thus,
communication protocol [29]. This is to be noted that the proposed with a hardware interface, the proposed RAD scheme is able to
scheme does not consider the communication latency, which may identify the faulty situation.
delay the response time by half a cycle (maximum) [30] The proposed scheme can be further enhanced to centralised
considering the line segment, not the accuracy. protection system equipped with PMU-based wide-area
To validate the proposed scheme in the real-world scenario, a measurements. In this system, all the measurements can be
software in the loop was established in the laboratory using RTDS retrieved at the PDC (PDC: the central monitoring unit), and
and external PMUs. The RTDS is connected to the external PMU protection function such as RAD can be derived for faulty
panel (ERL ZCP2225) with the help of Ethernet cables, which situations to issue the tripping signal for the faulted condition in
consists of three PMUs (Tesla 4000 power set recorder). The data any part of the microgrid. This may further add a delay in the

4308 IET Gener. Transm. Distrib., 2019, Vol. 13 Iss. 19, pp. 4301-4310
© The Institution of Engineering and Technology 2019
 17518695, 2019, 19, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-gtd.2019.0027 by Bangladesh Hinari NPL, Wiley Online Library on [16/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 8 Validation in RTDS platform with measurements from GTNET PMUs and external PMUs
(a) Variation in RAD for the three-phase fault at the midpoint of line DL2 with measurement by GTNET PMUs placed at PCC and B2 in grid-connected mode, (b) islanding mode,
(c) RAD during line to ground fault in grid-connected mode with measurement by external PMUs placed at PCC and B2, (d) RAD during three-phase fault in islanding mode with
measurement by external PMUs placed at PCC and B4

Table 6 Variation in operating speed

Line PMU location Fault type Fault distance (% of actual length) Operating time, ms
Grid Islanding
DL2 PCC-B2 LLLG 50 30 25
PCC-B1 LG 50 40 35
PCC-B3 LL 90 40 35
PCC-B4 LLL 10 21 19
PCC-B2 LLL 20 20 20
PCC-B1 LLG 30 23 18
DL4 PCC-B3 LG 60 20 21
PCC-B4 LLLG 80 57 50
PCC-B2 LL 40 33 30
PCC-B1 LLG 50 40 38
PCC-B3 LLL 70 20 18

response time; however, it will be within the desired range as the concerning accuracy and speed. The performance accuracy of both
communication latency will be minimal for microgrid topology. simulation and RTDS with external PMUs indicate that the
With the development of low-cost micro-PMUs, the wide-area- proposed wide-area-based RAD is a potential candidate for
based centralised protection scheme will be the comprehensive one identifying faults in the microgrid. Furthermore, the proposed
for the future microgrid. scheme provides a framework for centralised protection systems
for the microgrid.
6 Conclusion
The proposed research work focuses on the assessment of wide-
7 Acknowledgments
area voltage phase angle criterion for fault detection in a microgrid. This work was supported by the Central Power Research Institute,
It deploys few C37.118.1 complied PMUs at different buses to Bangalore, Government of India, through RSOP/20157/DG/
monitor the status of the grid. This paper shows that RAD between 3/15122015.
PCC and the bus close to the faulted line segment can be
effectively used to detect the faults in the microgrid. The proposed 8 References
RAD is found to be effective, fast and reliable under different
operating conditions (grid connected and islanded), fault types, [1] Stadler, M., Siddiqui, A., Marnay, C., et al.: ‘Control of greenhouse gas
emissions by optimal DER technology investment and energy management in
fault distance, and fault resistance. While comparing with existing zero-net-energy buildings’, Eur. Trans. Electr. Power, 2011, 21, (2), pp. 1291–
schemes, the RAD-based scheme provides improved performance 1309

IET Gener. Transm. Distrib., 2019, Vol. 13 Iss. 19, pp. 4301-4310 4309
© The Institution of Engineering and Technology 2019
 17518695, 2019, 19, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-gtd.2019.0027 by Bangladesh Hinari NPL, Wiley Online Library on [16/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[2] Parhizi, S., Khodaei, A., Bahramirad, S.: ‘State of the art in research on [17] Liang, X., Wallace, S., Nguyen, D.: ‘Rule-based data-driven analytics for
microgrids: a review’, IEEE Access, 2015, 3, (1), pp. 890–925 wide-area fault detection using synchrophasor data’, IEEE Trans. Ind. Appl.,
[3] Zamani, M., Sidhu, T., Yazdani, A.: ‘Investigations into the control and 2017, 53, (3), pp. 1789–1798
protection of an existing distribution network to operate as a microgrid: a case [18] Casagrande, E., Woon, W., Zeineldin, H., et al.: ‘A differential sequence
study’, IEEE Trans. Ind. Electron., 2014, 61, (4), pp. 1904–1915 component protection scheme for microgrids with inverter-based distributed
[4] Hooshyar, A., Iravani, R.: ‘Microgrid protection’, Proc. IEEE, 2017, 105, (7), generators’, IEEE Trans. Smart Grid, 2014, 5, (1), pp. 29–37
pp. 1332–1353 [19] Meier, A., Culler, D., McEachern, A., et al.: ‘Micro-synchrophasors for
[5] Anderson, P.M.: ‘Power system protection’ (IEEE Press, New York, 1999), distribution systems’. ISGT, Washington, D.C., 2014, pp. 1–5
pp. 201–240, 249–257 [20] Meier, A., Stewart, E., McEachern, A., et al.: ‘Precision micro-
[6] Massoud, A.M., Ahmed, S., Finney, S.J., et al.: ‘Inverter-based versus synchrophasors for distribution systems: a summary of applications’, IEEE
synchronous-based distributed generation; fault current limitation and Trans. Smart Grid, 2017, 8, (6), pp. 2926–2936
protection issue’. IEEE Energy Conversion Congress and Exposition, Atlanta, [21] ‘IEEE standard for synchrophasor measurements for power systems –
GA, 2010, pp. 58–63 amendment 1: modification of selected performance requirements’, IEEE Std.
[7] Nimpitiwan, N., Thomas, G., Ayyanar, R., et al.: ‘Fault current contribution C37.118.1a-2014, pp. 1–25
from synchronous machine and inverter based distributed generators’, IEEE [22] Kamwa, I., Samantaray, S., Joos, G.: ‘Compliance analysis of PMU
Trans. Power Deliv., 2007, 22, (1), pp. 634–641 algorithms and devices for wide-area stabilizing control of large power
[8] Najy, W., Zeineldin, H., Woon, W.: ‘Optimal protection coordination for systems’, IEEE Trans. Power Syst., 2013, 28, (2), pp. 1766–1778
micogrids with grid connected and islanded capability’, IEE. Trans. Ind. [23] Bollen, M., Styvaktakis, E.: ‘Tutorial on voltage sag analysis’. ICHQP,
Electron., 2013, 60, (4), pp. 1668–1677 Orlando, FL, USA, October 2000, pp. 193–194
[9] Mahat, P., Chen, Z., Jensen, B., et al.: ‘A simple adaptive overcurrent [24] Kim, S.D., Morcos, M.M., Gomez, J.C., et al.: ‘Voltage-sag magnitude and
protection of distribution systems with distributed generation’, IEEE Trans. phase jump due to short circuits in distribution systems with variable fault
Smart Grid, 2011, 2, (3), pp. 428–437 resistance’, Electr. Power Compon. Syst., 2006, 33, (5), pp. 493–512
[10] Petit, M., Pivert, X., Garcia, L.: ‘Directional relays without voltage sensors [25] Samantaray, S., Kamwa, I., Joos, G.: ‘Phasor measurement unit based wide-
for distribution networks with distributed generation: use of symmetrical area monitoring and information sharing between micro-grids’, IET Gener.
components’, Electr. Power Syst. Res., 2010, 80, (10), pp. 1222–1228 Transm. Distrib., 2017, 11, (5), pp. 1293–1302
[11] Chilvers, I., Jenkins, N., Crossley, P.: ‘Distance relaying of 11 kV circuits to [26] Mohanty, R., Balaji, U.S.M., Pradhan, A.K.: ‘An accurate noniterative fault-
increase the installed capacity of distributed generation’, IEE Proc. Gener. location technique for low-voltage DC microgrid’, IEEE Trans. Power Deliv.,
Transm. Distrib., 2005, 152, (1), pp. 40–46 2016, 31, (2), pp. 475–481
[12] Arroudi, K., Joós, G.: ‘Performance of interconnection protection based on [27] Mohammed, M.: ‘High-speed differential busbar protection using wavelet-
distance relaying for wind power distributed generation’, IEEE Trans. Power packet transform’, IEE Proc. Gener. Transm. Distrib., 2005, 152, (6), pp.
Deliv., 2018, 33, (2), pp. 620–629 927–933
[13] Ma, J., Wang, X., Zhang, Y., et al.: ‘A novel adaptive current protection [28] Faruque, M., Strasser, T., Lauss, G., et al.: ‘Real-time simulation technologies
scheme for distribution systems with distributed generation’, Int. J. Electr. for power systems design, testing, and analysis’, IEEE Power Energy Technol.
Power Energy Syst., 2012, 43, (1), pp. 1460–1466 Syst. J., 2015, 2, (2), pp. 63–73
[14] Sadeghkhani, I., Golshan, M., Saini, A., et al.: ‘Transient monitoring [29] Ali, N.H., Ali, B.M., Othman, M.L., et al.: ‘Performance of communication
function-based fault detection for inverter-interfaced microgrids’, IEEE Trans. networks for integrity protection systems based on travelling wave with IEC
Smart Grid, 2018, 9, (3), pp. 2097–2107 61850’, Int. J. Electr. Power Energy Syst., 2018, 95, pp. 664–675
[15] Pignati, M., Zanni, L., Romano, P., et al.: ‘Fault detection and faulted line [30] Michel, G., Pleinka, G., Adamiak, G., et al.: ‘Digital communications for
identification in active distribution networks using synchrophasor-based real- relay protection’, IEEE Power Syst. Relaying Committee Working Group H9,
time state estimation’, IEEE Trans. Power Deliv., 2017, 32, (1), pp. 381–392 Tech. Rep., 2002, pp. 1–11
[16] Ge, Y., Flueck, A., Kim, D., et al.: ‘Power system real-time event detection
and associated data archival reduction based on synchrophasors’, IEEE Trans.
Smart Grid, 2015, 6, (4), pp. 2088–2097

4310 IET Gener. Transm. Distrib., 2019, Vol. 13 Iss. 19, pp. 4301-4310
© The Institution of Engineering and Technology 2019