IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 35, NO.

4, AUGUST 2020 2011

Microgrid Protection Using Low-Cost

Communication Systems
Jerome Nsengiyaremye , Student Member, IEEE, Bikash C. Pal , Fellow, IEEE,
and Miroslav M. Begovic, Fellow, IEEE

Abstract—Power electronics interface of renewable energy to generation has changed, influencing operational characteristics
system is now the trend in both transmission and distribution seg- and requirements of the system. One important such requirement
ments of power network. Unlike synchronous generators, the fault is protection against fault. In many cases, there is a departure
feeding and control characteristic of these renewable generators
are different and mostly influenced by the topology, switching, and from traditional radial topology of feeders, which totally changes
control deployed in power electronics interface. So, the network the type of protection that is possible to use. Traditionally, due
protection design and operational requirements are now challenged to high cost of expanding the national grid to remote rural areas,
in the absence of large fault current. Although the differential microgrids have been assumed to be reasonably an economic
current principle still works, its implementation is limited by the alternative, and operate as isolated small grids. However, it has
significant cost associated to its communication system. This paper
proposes a differential line protection scheme based on local fault become necessary to meet energy requirements of cities from its
detection and comparing binary state outputs of relays at both generation from roof-top solar etc. with guaranteed reliability
ends of the line thus requiring a simple, flexible and low bandwidth when the national grid fails. With higher levels of quality of
communication system. The performance of the proposed scheme power supply and reliability demanded by consumers over time,
is assessed through simulation of an example system with several some distribution feeder automation strategies such as reclosers,
scenarios.
sectionalizers, whose purpose may be related to protection have
Index Terms—Distributed generation (DGs), fault-ride through, been adopted as practical solutions. This paper, however, focuses
inverter interfaced distributed energy resource (IIDER), inverter- on addressing the major issue of change of type of protection
only microgrid.
instead of tackling the entire problem together.
From strategic operational perspective, it is required to plan,
I. INTRODUCTION design and operate microgrid in a flexible way. Flexibility is
realized when microgrid is able to operate in grid-connected as
ITH the availability of cost effective solar and wind en-
W ergy conversion technologies, power electronics driven
microgrid for electrifying remote areas is now the preferred
well as in islanded mode. Power electronics offers the flexibility
but influences the fault behavior of the system, which renders the
standard network protections less effective. Inverter interfaced
option. It is also an option for more reliable energy solution
distributed energy resources (IIDERs) in grid distribution sys-
in the contexts of large cities. The energy and transport infras-
tems compromise conventional distribution system protection
tructures of many modern cities are planned as a single system
schemes such as overcurrent protection which is widely used
from energy point of view. Power electronics being an interface
in traditional radial feeders in the absence of local distributed
between renewable energy sources and network, the fault current
generation [1], [2]. The ineffectiveness of such schemes be-
flow behavior in the network is very dominated by its control.
comes more pronounced in IIDER-based microgrids operating
Microgrid is not a brand new concept though. It dates back
in islanded mode. This is because local generators can easily
to the beginning of power system industry in the end of 19th
deviate from grid operation requirements such as voltage and
century. However, it is only in the past two decades that it
frequency nominal values [3]. A flexible microgrid requires a
attracted significant research interest because the technology of
robust protection strategy. Recent research has dealt with these
challenges and new protection schemes have been proposed, and
Manuscript received July 22, 2019; revised October 17, 2019; accepted
December 7, 2019. Date of publication December 11, 2019; date of current extensively discussed [4].
version July 23, 2020. Paper no. TPWRD-00807-2019. (Corresponding author: Despite much research effort in microgrid protection, prac-
Jerome Nsengiyaremye.) tically transmission-based protection schemes are used due to
J. Nsengiyaremye is with the Electrical and Electronic Engineering, Imperial
College of Science, Technology and Medicine, London SW7 2AZ, U.K. (e-mail: lack of appropriate microgrid protection design standards and
j.nsengiyaremye17@[Link]). relays [5]. Unfortunately, in distribution microgrids, most of
B. C. Pal is with the Electrical and Electronic Engineering, Imperial Col- the schemes such as distance and overcurrent directional relays
lege of Science, Technology and Medicine, London SW7 2AZ, U.K. (e-mail:
[Link]@[Link]). become ineffective as well [1], [5]. For example, due to line
M. M. Begovic is with the Department ECE, Texas A&M University, College lengths that may not be sufficiently long to provide zoned pro-
Station, TX 77843-3128 USA (e-mail: begovic@[Link]). tection, distance relays are likely to fail. Furthermore, in systems
Color versions of one or more of the figures in this article are available online
at [Link] dominated by IIDERs, the predefined fault current references
Digital Object Identifier 10.1109/TPWRD.2019.2959247 make the fault current phase angles completely independent

0885-8977 © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See [Link] for more information.

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY TIRUCHIRAPALLI. Downloaded on April 14,2022 at 02:48:41 UTC from IEEE Xplore. Restrictions apply.
 2012 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 35, NO. 4, AUGUST 2020

of prefault voltage, which also challenges the performance of

distance relays [5]. Due to the complexity of inverter fault
current limiting function, especially when multiple inverters
are feeding into a fault via meshed microgrid, thus creating
potentially overlapping ranges of operating and fault currents,
the performance of conventional overcurrent directional relays
is affected. Typically, the overcurrent feature of these relays is Fig. 1. Voltage source control.
adversely affected by low fault currents generated by IIDERs
as they are restricted to a maximum value close to the rated These currents are assessed to extract the current directionality
current to protect the inverter components while the direction- information in the form of binary states. This information is
ality feature is affected by the independent characteristic of obtained using a novel simple and straightforward technique
current phase angle with respect to the voltage [5]. Furthermore, that detects the change in the current flow direction at each end
inverters being typically ungrounded, and most of inverters only of the line. Contrary to differential conventional protection, the
generating positive sequence currents even during unbalanced exchange of information between the line relays does not require
fault conditions, the directionality feature that is not based on data synchronisation.
positive sequence currents may not be reliable for the protection As most communication-assisted protection schemes in the
of microgrids dominated by IIDERs [1], [5]. literature do not comprehensively cover the communication as-
Differential protection principle typically meets the technical pect by routinely assuming the availability of systems that meet
requirements to protect microgrids due to its immunity to the the protection requirements, this paper provides a case study
non-sensitivity to bidirectional power flow, changing current of a low-cost communication system that can be used with the
levels, number of distributed energy resources (DERs) in the proposed protection scheme in distribution systems. Section II
microgrid, microgrid operation mode and weak infeed [4], [6], presents the IIDERs behaviour under fault conditions. The pro-
[7]. However, it has been demonstrated that much attention tection scheme principle details are discussed in Section III while
should be taken in choosing the differential parameter as tradi- Sections IV and V assess and discuss the performance of the
tional differential current relay may become ineffective in some proposed protection scheme under various scenarios. Section VI
situations when the IIDER penetration is high [8], [9]. This is due concludes the paper.
to the effect of the difference between the sequence components
of the fault currents on both sides of the line [8] as the system
II. IIDER BEHAVIOUR UNDER FAULT
becomes mainly influenced by inverter control.
Research studies recently undertaken based on data mining A. Plant Description
approach have identified appropriate parameters for microgrid Full-scale power converters are becoming indispensable in
protection [8], [10], [11]. It has been shown that positive and network integration of wind turbine and solar power. The con-
negative sequence current components are among top three figuration of the inverters adopted in this study is a three-phase
features for both fault detection and localisation [11], [12]. Of three-wire topology usually used in medium voltage distribution
these two features, the positive sequence current components are systems [18]. Each inverter is connected to the rest of the
preferable due to their presence in all types of faults [8], [10], microgrid through an LCL filter. The filter is formed of an LC
[13]–[16]. circuit and a step-up transformer. The microgrid side of the LC
While it is technically effective, the current differential pro- circuit is considered as the point of the common coupling (PCC)
tection requires the exchange of the measurements from both at which the IIDER is connected to the rest of the microgrid.
ends of the line, which requires high bandwidth, sophisticated It is at the PCC that the voltage and current are measured for
and significantly expensive communication systems. The cost control and protection of the inverter. This is illustrated by the
of such communication systems may hinder the adoption of black background side of Fig. 1. It could be possible to consider
conventional differential protection in distribution systems and measurements taken after the transformer, but high voltage level
microgrids, whose level of investment is relatively low. How- would challenge the cost-effectiveness of voltage measurement.
ever, it has been shown that the schemes with no communication For the simplicity of the control system, the generator is assumed
link between relays cannot effectively protect microgrids from to produce a (DC) voltage.
all types of faults [6], [17]. So a protection scheme that allows the
use of a low-cost communication system is considered viable,
and perhaps the only alternative. B. Inverter Behavior Under Faults
One way of reducing the communication cost requirement The behavior of inverters under fault in microgrid is deter-
is to minimize the size of the information to be exchanged mined by the injected current at the PCC. This current depends
between the relays without jeopardizing the accuracy of the on the inverter control system. Moreover, to protect electronic
fault detection and faulted line localization. So with minimal components of the inverter from damage, its fault current is
information size, a low bandwidth communication system can limited to a maximum value close to the rated current. The
be used as its cost is relatively low. limiting strategy of the current is practically achieved by pro-
This paper proposes such a line protection scheme using viding preset current references to the current controller when
positive sequence currents estimated at both ends of the line. the voltage loop controllers generated references exceed a set

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY TIRUCHIRAPALLI. Downloaded on April 14,2022 at 02:48:41 UTC from IEEE Xplore. Restrictions apply.
 NSENGIYAREMYE et al.: MICROGRID PROTECTION USING LOW-COST COMMUNICATION SYSTEMS 2013

Fig. 3. Study microgrid system.

Conventionally, current flow direction estimation has been

Fig. 2. Voltage and current signals measured at PCC and at the output of the based on the comparison of the voltage and current phase angles.
filter inductor in different fault conditions (three phase fault (3Φ), single phase However, with no voltage measurement devices installed at ev-
to ground fault (Φ-G), and phase to phase fault (Φ-Φ)). ery node in distribution systems, current-only based techniques
become necessary. Schemes based on current-only approach
threshold. The general structure of the IIDERs control is given proposed in the literature, so far, are only for fault direction
by the white background of Fig. 1. The signals generated by the estimation [13], [14], [20]–[23]. Research in [13], [20], [21]
VSC under different fault types measured at the filter inductor compares prefault and fault current phase angles to determine if
outputs and at the PCC are displayed in Fig. 2. a fault is upstream or downstream of the relay. However, this is
With the fault ride through conditions, the smart inverter will limited to radial systems where prefault power flow direction is
generate reactive power to the system during faults. So contrary known, and a separate fault detection device is required. On the
to synchronous generators in transmission systems where the other hand [14] applies directly the phase comparison approach
fault currents depend on the impedance of the passive elements which requires the exchange of the phase angle values between
of the network, the angle of the fault current in IIDERs will relays. In [22], [23], the fault direction is estimated at both end
mainly depend on the control system. However, this does not of the line to detect an internal fault. For the approach in [22],
mean that the traditional angular features used for protection the relay at each end of the line detects a fault on a fraction of
in transmission systems can be straightforwardly used in syn- the line based on an overcurrent principle. This scheme is only
chronous machine-based microgrids because of the challenges effective in grid connected microgrids with the grid providing
posed by short lines. significantly higher fault current than the DGs.
As multi-DER microgrids contain at least two DGs, power The study in [23] proposes an interesting approach based on
sharing strategy should be considered during fault with respect fault direction estimation technique proposed in [13], [20], [21]
to the system stability. It has been shown that among the most by considering the fault direction estimation at both ends of the
used techniques, droop-based and master-slave, the droop-based line. This study, however, has analysed and verified this approach
approach is more robust [19]. Consequently, due to its sensitiv- with extra high voltage transmission system data which may not
ity to fault, the master-slave technique is ideal for testing the necessarily be consistent with distribution systems.
effectiveness of any protection scheme. While the difference between the local prefault and fault
current phase angles could be used, as in [23], a clever, yet
III. PROTECTION SCHEME simple approach to detecting a change in current flow direction
keeping the prefault angle as the reference is to calculate the
A. Fault Detection cosine of the angle. When there is a change in direction, the
For the simplicity of fault detection analysis, a robust and cosine value changes the sign. This approach offers a new simple
reliable microgrid system shown in Fig. 3 is used. Typically, comparison reference, “zero”, as a constant and neutral number.
when a fault occurs at Point F on Line LBE one of the line relays, In three-phase systems, the positive sequence current is used.
3 or 4 will see a change in current flow direction with respect Therefore, for simplicity, the word “current” is subsequently
to the prefault current flow direction. So the change in current adopted in this study to refer to positive sequence current.
direction can be a useful indication of a fault on the line. It is The proposed approach can be equivalent to the principle of
important to note that this is true when the line considered does phase comparison protection scheme with the difference that, in
not present any taps. The fault detection under this study does this paper, the compared quantities are measured at one end
not cover tapped lines, which are commonly not used in network of the line at two successive instants and then compare the
like microgrids under consideration. Otherwise, an extension to transition patterns (events) from both ends rather than directly
the proposed scheme is possible and should be sought. comparing the two quantities measured at the two ends of the

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY TIRUCHIRAPALLI. Downloaded on April 14,2022 at 02:48:41 UTC from IEEE Xplore. Restrictions apply.
 2014 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 35, NO. 4, AUGUST 2020

directions of relays at both ends of the line must be the

same. This means that the current transformers supplying
measurements to the relays must be consistently polarized,
and both relays consistently installed. This convention
allows the definition of the same virtual reference (cos (0) =
1) for the cosine value sign for both relays. Therefore, the
relay can virtually decide a change of current direction or
a no-change condition.
The latter assumption above does not only solve the direction
change issue but also allows correct discrimination of fault from
load switching events. Under load switching events, both relays
register the same transition pattern (no change or change of
current direction), which is also the case for an external fault.
As previously mentioned, the exchange of information be-
tween the line relays requires a communication channel. Since
the adoption of communication-assisted protection schemes
may be hindered by the cost of the communication system,
the current direction transitions (the change or no change) are
directly transformed into binary states, “1” or “0” in order to
Fig. 4. Fault detection strategy: (a) principle (b) phasor estimation.
allow the use of simple and low-cost communication systems.
Moreover, this offers flexibility in data transmission.
line at the same instant. Moreover, the proposed principle is
built on full cycle Fourier technique recognised for its inherent
B. Low-Cost Communication System
immunity to harmonics.
The proposed protection scheme uses “the zero” as the ref- The overall cost of a communication system comprises the
erence for detection of the change in current flow direction by cost of the communication equipment (physical link and in-
computing the cosine value of the current phase angle. The relay terfaces), the installation cost (work and labor), the required
fault detection principle and the phasor computation algorithm installation time, and its reliability. The cost comparison analysis
are provided in Fig. 4. The phasor estimation is based on full cy- of existing communication systems has shown that unlicensed
cle discrete Fourier transform (DFT) due to its deterministic as- microwave radio offers practically the lowest cost option [25]. Its
pect and capability to reject harmonics. Non-Fourier algorithms low cost is due to the nonrequirement of right of way, low instal-
such as least squares-based algorithms, differential-equations lation cost, low cost of labor (no specialized expertise), and short
algorithms, Kalman filter algorithms and Walsh functions al- installation time [25]–[28]. The two most used spread-spectrum
gorithms suffer from harmonic effect and might not attain the techniques for unlicensed radio communication are frequency
required online speed computation [20], [24]. The update of hopping spread spectrum (FHSS) and direct-sequence spread
phasor estimates with time are computed using recursive full spectrum (DSSS). Both techniques can be used for protection
cycle technique. This technique offers computational simplicity applications. The FHSS offers more reliability at low frequency
and provides a stationary phase angle in steady state by default band, typically 902–928 MHz [24].
as one of the side benefits. To ensure low-cost communication system feasibility, the
When a fault occurs on a line, one of the ends of the line protection scheme should offer flexibility in reducing the mes-
experiences a change in current flow direction (i.e. the change sage size. On the other hand, the presence of other functions,
of its cosine value sign) while the other does not get affected for example the control signals for master-slave power sharing
much. With each line-end relay capturing the current direction strategy, which rely on communications may relieve some of the
transition, the fault on the line is detected by the relays by ex- pressure in that respect.
changing their current direction information. When both relays
register the same direction transition (no change or a change in
C. Practical Considerations of Low-Cost Communication
current direction) there is no fault detected on that line. On the
contrary, if one of the relays registers a change, a fault is detected System With the Proposed Protection Scheme
on the line. To analyze the communication requirements, the proposed
It may happen that there is no current flowing in the line protection scheme is classified as a teleprotection scheme.
before fault occurrence. In this case, the relay is not able to take Standard practice requires a certain level of reliability of such
a decision on the change of current direction. To tackle this issue, schemes to adhere to. The standard sets the security as the prob-
two assumptions are made: ability of unwanted command to be between 10−4 and 10−8 , and
r When the current is zero or very small, its phase angle is the dependability as the probability of missing command to be
set to zero i.e. the cosine is “one”. between 10−2 and 10−4 [27]. This means that one unwanted trip
r While it may not necessarily be the case when the current is requires a minimum of 108 noise bursts for a scheme designed
flowing in a line prior to fault occurrence, the monitoring with 10−8 security [29].

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY TIRUCHIRAPALLI. Downloaded on April 14,2022 at 02:48:41 UTC from IEEE Xplore. Restrictions apply.
 NSENGIYAREMYE et al.: MICROGRID PROTECTION USING LOW-COST COMMUNICATION SYSTEMS 2015

The proposed protection scheme is evaluated for the most TABLE I

COMPARISON OF COMMUNICATION SYSTEM
stringent requirements among teleprotection schemes i.e. 10−8 DELAYS OF DIFFERENT MESSAGE SIZES
security. To achieve such a level of security, it is important to
ensure that if a distinct message was corrupted, it should not
be received as a valid message. It is therefore necessary not to
send binary 1 and 0 as they are, because a corrupted binary 1
might become a binary 0 and vice-versa. To tackle this issue, two
approaches are usually used i.e. redundant bits, and duplicate
messages [29]. However, the former increases the bandwidth
requirement affecting the dependability as expressed in (1) while
the latter approach may require extra memory that might increase data word size is 8 bits. Along with this message size, two
the cost and time delay for the fault detection. others are compared in Table I in terms of their associated delays:
25-bit, and 16-bit messages.
Pmc = 2k(BER) (1) To obtain the total time required by the proposed scheme to
where Pmc is the probability of missing command, k the number output a decision, the relay delay (∼20 ms), and the transmission
of bits in a message, and BER, the bit error rate defined as interface delay (∼2 ms, typically between 1–5 ms) are added to
the ratio of erroneous bits to total bits transmitted on a digital the total communication time delay. This time is about 41.3 ms,
channel. 36.4 ms, and 36.7 ms for the 10-bit, 16-bit, and 25-bit message
Fortunately, the proposed protection scheme, being phasor- sizes respectively. So, the scheme requires about 2 cycles. This
based, has an inherent typical delay of one cycle (20 ms for 50 Hz is typically acceptable for protection applications as it is well
nominal frequency). Despite not usually desirable, this delay below the critical clearing time requirement for microgrids
can allow the relay to send the same message more than once reported in the literature (about 5 cycles) [19], [32]. This as-
in a cycle without a memory, thus creating a natural duplicate sessment has assumed a full duplex communication channel.
message approach. Furthermore, it is a standard practice that However, a half-duplex system can be used by automatically
a protection message is sent two to eight times per cycle [26], switching the channel in both direction [29]. This would farther
[27]. A combination of this natural duplicate message and the reduce the communication system cost but at the expense of
bit redundancy approaches offers an opportunity to achieve response time. This would add a delay of about a cycle bringing
the required security and dependability with a low bandwidth the total response time to about 3 cycles. This time should also
channel requirement and short transmission time. The total time be acceptable because circuit breakers can clear the fault in less
delay associated with radio communication system can then be than 2 cycles.
evaluated based on the message size, the number of messages It is important however, to highlight that the proposed protec-
exchanged per cycle, the relay processing time, the channel tion scheme may not be appropriate for special systems requiring
bandwidth capacity, and the allowed communication channel very high speed protection. Current differential protection can
latency [29]. be used, otherwise an extension to the proposed scheme based
While radios are characterized by high latency, technologies on fractional data window may be sought.
with variable latency are recommended for teleprotection. These Also, for any communication-assisted protection scheme, the
types allow for setting changes of some features that require failure of communication system leaves the system unprotected.
much processing time with no value addition to the scheme It is therefore important to have a back-up protection protection
application [27]. Such features are forward error correction (up in place to ensure the system protection in a such situation.
to 5 ms), message acknowledgement, retry or re-transmission,
data buffering (∼50 ms), and encryption (∼10 ms) [27]. This is IV. STUDY SYSTEM
because the relay sends data continuously allowing the receiving A medium voltage microgrid system, with loop configuration,
relay to get the right message even when some are corrupted. is used as a study system as shown in Fig. 3. While there are no
Practically radio latency of 2 ms is achieved [29]. Currently, impediments to using the same or similar technique applied to a
unlicensed radios exist with band-width as high as 38400 bps more traditional, radial grid with many laterals, the extensive
that can achieve the range of 15 to 30 km [26]. presence of tapped extensions and use of automation would
To evaluate the feasibility of the whole scheme (fault detection require more in-depth analysis. Nevertheless, this is a reasonable
and communication), a simple serial digital to digital bit stream configuration to first try the concept as it represents a reliable
communication protocol is considered. This protocol can use isolated grid, and a reasonably challenging case to evaluate the
interfaces such as Universal Asynchronous Receiver Transmitter effectiveness of the proposed protection scheme.
(UART) or Ethernet for message transmission [29], [30]. Taking
the UART interface as an example, the message size to be
A. Positive Sequence Equivalent Circuit
exchanged between relays can be assessed. This interface can
allow 1 to 23 bits per data word in addition to an overhead of The positive sequence equivalent circuit of the study system
two bits (typically one start bit and one stop bit) (message) [31]. with a three-phase fault on Line LBE is given in Fig. 5. The
This gives flexibility in choosing the appropriate word size for inverter interfaced distributed generator (IIDG) is represented
some applications. For many applications, the most common by the positive sequence current source [14], [15], [33].

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY TIRUCHIRAPALLI. Downloaded on April 14,2022 at 02:48:41 UTC from IEEE Xplore. Restrictions apply.
 2016 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 35, NO. 4, AUGUST 2020

TABLE II
STUDY MICROGRID PARAMETERS

Fig. 5. Positive sequence equivalent circuit with a fault on Line LBE .

When a three-phase fault occurs on line LBE , the positive

sequence fault current IF is given by the sum of faults currents
LBF and LEF from buses B and E respectively as expressed in
(2).

IF = IBF + IEF (2)

IBF = IF _IIDG1 + IF _IIDG3_CB

and

IEF = IF _IIDG2 + IF _IIDG3_CE

where IF _IIDG1 and IF _IIDG2 are positive sequence fault

currents generated by IIDG1, IIDG2 while IF _IIDG3_CB and
IF _IIDG3_CE are positive sequence fault current flowing re-
spectively through branches CBF and CEF from IIDG3.
The fault currents passing through the two branches are func-
The length of the lines in km: LAB = LCF = LED = 0.6, LBC = 3.0, LBE =
tion of the branch impedances and IIDGs currents as expressed 1.5, and LCE = 1.0.
in (3) and (4).
IF _IIDG3 (ZCE + ZEF ) factor can be estimated using the ratio of quadratic component
IF _IIDG3_CB = (3)
(ZCB + ZBF ) + (ZCE + ZEF ) iq to the direct current component id .
IF _IIDG3 (ZCB + ZBF ) From (5) it can be observed that positive sequence fault current
IF _IIDG3_CE = (4) angles depend mainly on the IIDGs fault currents i.e. their
(ZCB + ZBF ) + (ZCE + ZEF )
power factor. This offers simplicity in developing line protection
where ZCE , ZEF , ZCB , ZBF , ZCE , and ZEF represent line schemes for inverter-only microgrids. This is because it becomes
impedance of Lines LCE , LEF , LCB , LBF , LCE , and LEF possible to do the analysis with an assumption of two current
respectively. sources, one at each end of a line with prior knowledge of the
Recognising that the parameter characteristics of the lines current phase angle value range.
forming the loop are typically almost the same, since each line
should handle the same maximum current when one of the line V. SIMULATION SCENARIOS AND RESULTS
is isolated, and that the lines in the microgrid are short (the same
length assumption would make sense for a loop configuration in A. Simulation Scenarios
this case), the fault current IF could be estimated as follows To validate the effectiveness of the proposed scheme, a mi-
 IF _IIDG3   IF _IIDG3  crogrid study model was developed in Matlab/Simulink environ-
IF = IF _IIDG1 + + IF _IIDG2 + ment. The topology of the model given in Fig. 3 is used. The pa-
2 2
(5) rameters of the IIDGs, step-up transformers, and power lines are
The positive sequence current values generated by the in- provided in Table II. The IIDG1 adopts the Voltage-Frequency
verter for each type of fault are provided by the manufacturer control to form and maintain the voltage and frequency of the
as required by the standard [33]. Typically, the fault current system at reference values while the others are locally controlled
value is limited to a maximum value close to the rated inverter as current sources. To ensure appropriate power sharing among
current. IIDGs with low voltage fault ride-through capability IIDGs, the master-slave control is adopted. The IIDG1 operates
are required to generate reactive current to support the system as the master, and shares its current reference signals to other
voltage stability during fault [5]. This to equivalently say that IIDGs (the slaves). This strategy is chosen due to its simplicity
these IIDGS generate low power factor power during faults. With and highest sensitivity to faults. The control process is performed
inverters usually controlled in dq0 reference frame, the power in synchronous reference frame (dq0). To limit the fault current

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY TIRUCHIRAPALLI. Downloaded on April 14,2022 at 02:48:41 UTC from IEEE Xplore. Restrictions apply.
 NSENGIYAREMYE et al.: MICROGRID PROTECTION USING LOW-COST COMMUNICATION SYSTEMS 2017

generated by IIDGs and overvoltage under unbalanced faults,

the latched current limiting strategy, and the instantaneous sat-
uration limit are respectively adopted.
Three scenarios namely Reference Scenario, Scenario 1, and
Scenario 2 are assessed under this study.
1) Reference Scenario: This scenario assesses the effective-
ness of the protection scheme versus fault type, fault resistance,
and fault location. The existence of prefault currents in all
branches of the microgrid loop is assumed and ensured by the
following load conditions: Load 1 = 0.7 MW; Load 2 = 1.3
MW and Load 3 = 0.2 MW. The fault currents are limited to
1.2 times the value of the rated inverter current, an ideal value to
Fig. 6. Cosine values computed by relays for a three-phase fault on Line LBE .
challenge the proposed scheme and even more to conventional
scheme. The power factor is set to 0.56 (lagging) i.e. idref =
0.67 pu and iqref = −1.0 pu. These predefined fault current ref-
erences are assumed to be the same for all three IIDGs. Different
types of faults i.e. symmetrical (L-L-L-G) and unsymmetrical
faults (L-G and L-L) were simulated on Line LBE (between
relays 3 and 4) and Line LAB (between Relays 1 G and 1) at
time instant t = 0.1 s.
2) Scenario 1: This scenario evaluates the effectiveness of
the proposed protection scheme for different fault inception time
instants, and for different reactive current values generated by
IIDGs during fault. With the former, a fault on Line LBE at two
fault inception instants different from the Reference Scenarios
are considered; t = 0.11 s, and t = 0.116 s. On the other hand,
different power factors (PF) are considered in the following three
cases: Fig. 7. Illustration of current flow direction through the lines before and after
r The fault current references for all IIDGs are limited to a three-phase fault on Line LBE .
1.2 pu with a power factor of 0.85 (lagging) i.e. idref =
1.02 pu and iqref = −0.63 pu.
r The fault current references for all IIDGs are limited to covered in the previous scenarios. Initially, all loads at the nodes
are set to 0.5 MW. With such loading, there is no current flowing
1.2 pu with a power factor of 0.2 (lagging) i.e. idref = in the loop as the IIDGs can satisfy their respective local loads.
0.24 pu and iqref = −1.17 pu. To assess the performance of the scheme, at t = 0.1 s a three phase
r The fault current references for all IIDGs are limited to fault is simulated on Line LBE . For the load switching events,
1.2 pu with different power factors. The power factor of with the same loading condition, an additional load of 0.6 MW
IIDG1 (PF 1) = 0.85; PF 2 = 0.56, and PF 3 = 0.2. is connected at Bus B at t = 0.1 s. With this additional load kept
Other than the fault inception time instants, and fault current in the system, the 0.6 MW are switched off from the system at t
references, the parameters and conditions under Scenario 1 are = 0.15 s in order to verify the effectiveness of the scheme when
the same as in the Reference Scenario. a considerable load is switched off from the system.
3) Scenario 2: This scenario assesses the performance of Other than the loading parameters, other simulation parame-
the proposed protection scheme in case of load switching. This ters are the same as those of the Reference scenario.
scenario has been undertaken considering two cases:
r the performance of the scheme when there is current flow-
ing in a line prior to fault occurrence; and B. Simulation Results
r the performance of the scheme when there is no current The cosine values computed by each relay over time for a
flowing through at least one of the loop lines prior to fault three-phase solid fault are provided in Fig. 6. It is clearly that
occurrence. only Relay 4 registers a change in the cosine value sign just
Since the fault events are comprehensively covered in the after the fault inception time. With the rest of relays registering
previous scenarios when there is current flowing in the lines prior a no-change direction state, the fault is detected between Relays
to fault, the first case under this scenario will only consider load 3 and 4 as it is the only relay pair to register different cosine
switching event. The initial loading under this case is such that sign transitions. This case is reflected in Fig. 7 illustrating the
Load 1 = 0.5 MW, Load 2 = 0.8 MW, and Load 3 = 0.2 MW. current flow direction in the study system before and during the
At t = 0.1 s a an additional load of 0.5 MW is switched on at fault (note that due to line styles used to demarcate the cosine
Bus B. values computed by different relays in the figure, some values
The second case on the other hand, considers both load seem to be greater than one while they are actually not). A full
switching and fault events. This is because it is a special case not picture of the simulation results is provided in Table III. The

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY TIRUCHIRAPALLI. Downloaded on April 14,2022 at 02:48:41 UTC from IEEE Xplore. Restrictions apply.
 2018 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 35, NO. 4, AUGUST 2020

TABLE III
PERFORMANCE OF THE PROPOSED PROTECTION SCHEME

• Rf , PF, and t refer to fault resistance, power factor of fault reference currents of IIDGs, and fault inception time respectively.
• Different PFs refers to PF 1 = 0.85 for IIDG1, PF 2 = 0.56 for IIDG2, and PF 3 = 0.20 for IIDG3.

results demonstrate the effectiveness of the proposed protection they have different fault current references (i.e. different power
scheme for all simulated scenarios. factors) as shown in the table.
From the table, the first case (Rf = 0 Ω) under the Reference The results on the discrimination of faults from load switching
scenario shows that the scheme is effective for all types of fault. events are presented in the lower part of the table. In the previous
For any type of fault on line LBE , consistency in results is scenarios, results clearly show the effectiveness of the proposed
observed whereby only Relay 4 registers a change in the cosine protection approach in detecting faults when the current flows in
value sign (the angle changing from −159.4o to a value of 4th the faulted line prior to the fault occurrence. The discrimination
quadrant) as marked in red. With Relay 3 registering no change of fault from load switching event in the previous scenarios is
in cosine value sign, a fault is detected on the line protected by straight-forward as any change in current direction due to load
relay pair (3, 4). Each of the other relay pairs, (1 G, 1), (2, 8), switching event is seen by both relays protecting the line. This
(5, 5 G), (6, 7), and (9, 9 G), registers the same transition states, is clearly demonstrated with both Relays 3 and 4 registering
which indicates that there is no fault on their lines. Furthermore, the change. However, when there is no current flowing in the
the same consistency in results is obtained for the cases of faults faulted line prior to fault occurrence depending on the system
with large range of resistance. For example, Fig. 8 shows that loading, as explained in Section III-A, the relay pair monitoring
simulation results of a solid line to ground fault and the same directions should be the same to discriminate a fault from load
type of fault with a fault resistance of 100 Ω are consistently the switching.
same. The state of the system prior to fault or load switching event,
The results registered under all the cases of Scenario 1 are as provided in the lower part of Table III, shows that there is
also as consistent as those in Reference case. The consistency is no current flowing in Lines LBE and LBC due to initial system
clearly obtained for different fault inception time instants, when loading. When the load is switched into the system at Bus B at
IIDGs have the same fault currents references as well as when the time instant t = 0.1 s, the cosine values computed by Relays

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY TIRUCHIRAPALLI. Downloaded on April 14,2022 at 02:48:41 UTC from IEEE Xplore. Restrictions apply.
 NSENGIYAREMYE et al.: MICROGRID PROTECTION USING LOW-COST COMMUNICATION SYSTEMS 2019

sign of the cosine value is rather consistent. This shows that the
proposed scheme does not require high precision in the phasor
estimates to perform effectively. Furthermore, due to the two-
end event comparison approach, erroneous phasor estimates that
may be caused by IIDGs generating off-nominal frequencies or
unbalanced fault contribution during unsymmetrical faults do
not affect the effectiveness of the scheme. This is because the line
relay pair experience almost the same effects from the frequency
deviation or unbalance. This is shown in Fig. 9 presenting the
results from the simulation of a line to ground fault on Line LBE
with a wide range of off-nominal frequency (45 Hz, 49 Hz,
51 Hz, and 55 Hz) of IIDGs. From the figure, it is clear that
when the fault occurs, Relay 4 registers a change contrary to
its counterpart (Relay 3) in all cases. Other than the frequency,
other simulation parameters are the same as the ones in reference
scenario.

VI. CONCLUSION
In order to protect inverter-based microgrids, relays that can
Fig. 8. The cosine values computed by relays with a fault on Line LBE : overcome challenges brought by unconventional fault signatures
(a) solid line to ground fault (b) line to ground fault with a fault resistance of of IIDERs are required. These signatures are due to limited and
100 Ω (c) solid line to line fault (d) line to line fault with a fault resistance of
100 Ω. controlled inverter fault currents and it has been acknowledged
that differential relays can overcome their challenges among
other transmission system- and distribution system-based pro-
tection schemes. However, although technically appropriate,
differential relays are not economically viable due to their sig-
nificantly high communication costs. The proposed microgrid
protection relay was derived from an analysis of inverter-based
microgrids behavior and allows the use of a low-cost commu-
nication system. Its performance demonstrates the effectiveness
in all fault conditions and immunity to the inverter fault signa-
tures effect as verified by simulations on the microgrid study
system. The analysis in this paper has shown that the proposed
scheme offers an advantage of using a low-cost communication
system as it requires a low communication bandwidth system.
Moreover, the scheme offers inherent immunity to harmonics
and noise generated by power electronics, non linear loads,
switching transients, and errors in measurements.

REFERENCES
[1] “Microgrid protection systems,” Working Group C30, Subcommittee C of
the Power System Relaying and Control Committee, Jul. 2019.
[2] J. Shiles et al., “Microgrid protection: An overview of protection strategies
in North American microgrid projects,” in Proc. IEEE Power Energy Soc.
General Meeting, Chicago, IL, USA, 2017, pp. 1–5.
Fig. 9. The cosine values computed by relays with a line to ground fault on [3] S. M. Brahma, J. Trejo, and J. Stamp, “Insight into microgrid protection,”
Line LBE at t = 0.1 s with: (a) IIDGs frequency of 49 Hz, (b) IIDGs frequency in Proc. IEEE PES Innovative Smart Grid Technologies Eur., Istanbul,
45 Hz, (c) IIDGs frequency of 51 Hz, (d) IIDGs frequency 55 Hz. Turkey, 2014, pp. 1–6.
[4] CIGRE JWG B5/C6.26/CIRED, “Protection of distribution systems with
distributed energy resources,” Paris, France, 2015.
[5] A. Hooshyar and R. Iravani, “Microgrid protection,” Proc. IEEE, vol. 105,
2, 8, 6, and 7 change the sign. However, as they are in pairs i.e no. 7, pp. 1332–1353, Jul. 2017.
(2, 8) and (6, 7), there is no fault detected. The same situation [6] E. Sortomme, J. Ren, and S. S. Venkata, “A differential zone protection
happens when the load is later disconnected from the system. On scheme for microgrids,” in Proc. IEEE Power Energy Soc. General Meet-
ing, Vancouver, BC, Canada, 2013, pp. 1–5.
the other hand, when a three-phase fault occurs on Line LBE , [7] X. Liu, M. Shahidehpour, Z. Li, X. Liu, Y. Cao, and W. Tian, “Protection
Relay 3 registers a change as well. This clearly indicates a fault scheme for loop-based microgrids,” IEEE Trans. Smart Grid, vol. 8, no. 3,
on the as the remote relay sees no change. pp. 1340–1349, May 2017.
[8] S. Kar, S. R. Samantaray, and M. D. Zadeh, “Data-mining model based
The slight difference obtained in the angle values throughout intelligent differential microgrid protection scheme,” IEEE Syst. J., vol. 11,
the simulation results from relay pairs is not an issue as the no. 2, pp. 1161–1169, Jun. 2017.

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY TIRUCHIRAPALLI. Downloaded on April 14,2022 at 02:48:41 UTC from IEEE Xplore. Restrictions apply.
 2020 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 35, NO. 4, AUGUST 2020

[9] S. Chen, N. Tai, C. Fan, C. Fan, J. Liu, and S. Hong, “Sequence-component- [20] A. Ukil, B. Deck, and V. H. Shah, “Current-only directional overcurrent
based current differential protection for transmission lines connected with relay,” IEEE Sensors J., vol. 11, no. 6, pp. 1403–1404, Jun. 2011.
IIGs,” IET Gener., Transmiss. Distrib., vol. 12, no. 12, pp. 3086–3096, [21] A. Ukil, B. Deck, and V. H. Shah, “Fault direction parameter indicator
Oct. 7, 2018. device and related methods,” U.S. Patent 9 366 715 B2, Jun. 14, 2016.
[10] E. Casagrande, W. L. Woon, H. H. Zeineldin, and D. Svetinovic, “A [22] C. Zhou, G. Zou, J. Yang, and X. Lu, “Principle of pilot protection
differential sequence component protection scheme for microgrids with based on positive sequence fault component in distribution networks
inverter-based distributed generators,” IEEE Trans. Smart Grid, vol. 5, with inverter-interfaced distributed generators,” in Proc. IEEE PES GTD
no. 1, pp. 29–37, Jan. 2014. Grand Int. Conf. Expo. Asia (GTD Asia), Bangkok, Thailand, 2019,
[11] H. H. Zeineldin, N. H. Kanan, E. Casagrande, and W. L. Woon, “Data pp. 998–1003.
mining approach to fault detection for isolated inverter-based microgrids,” [23] Q. Li, H. Gao, D. Yu, and B. Xu, “Analysis and verification of a novel
IET Gener., Transmiss. Distrib., vol. 7, no. 7, pp. 745–754, Jul. 2013. current comparison pilot protection,” in Proc. IEEE Conf. Energy Internet
[12] M. Xu, G. Zou, C. Xu, W. Sun, and S. Mu, “Positive sequence differential Energy Syst. Integration (EI2), Beijing, China, 2017, pp. 1–5.
impedance protection for distribution network with IBDGs,” in Proc. [24] A. G. Phadke and J. S. Thorp, Synchronized Phasor Measurements and
IEEE Int. Conf. Power Syst. Technol., Wollongong, NSW, Australia, 2016, Their Applications. New York, NY, USA: Springer, 2008.
pp. 1–5. [25] S. V Achanta, B. Macleod, E. Sagen, H. Loehner, and S. E. Laboratories,
[13] A. K. Pradhan, A. Routray, and S. M. Gudipalli, “Fault direction estimation “Apply radios to improve the operation of electrical protection,” SEL J.
in radial distribution system using phase change in sequence current,” IEEE Rel. Power, vol. 1, no. 2, pp. 1–12, 2010.
Trans. Power Del., vol. 22, no. 4, pp. 2065–2071, Oct. 2007. [26] E. O. Schweitzer, D. Finney, and M. V. Mynam, “Applying radio commu-
[14] W. Li, J. He, D. Zhang, and Q. Zhang, “Directional pilot protection nication in distribution generation teleprotection schemes,” in Proc. 65th
based on fault current for distribution network with Distributed Generation Annu. Conf. Protective Relay Eng., 2012, pp. 310–320.
(DG),” J. Eng., vol. 2017, no. 13, pp. 1327–1331, 2017. [27] “Using spread spectrum radio communication for power system protection
[15] H. Gao, J. Li, and B. Xu, “Principle and implementation of cur- relaying applications,” IEEE PES PSRC Working Group H2 to the Power
rent differential protection in distribution networks with high penetra- System Relaying Committee, 2005.
tion of DGs,” IEEE Trans. Power Del., vol. 32, no. 1, pp. 565–574, [28] MEF, “Microwave technologies for carrier Ethernet services. Metro
Feb. 2017. Ethernet Forum, Los Angeles CA 90045 USA.,” Feb. 2011. [Online].
[16] J. Yang, C. Zhou, and G. Zou, “A protection scheme based on positive Available: [Link]
sequence fault component for active distribution networks,” in Proc. 2nd Technology_for_Carrier_Ethernet_Final_110318_000010_000.pdf.
IEEE Conf. Energy Internet Energy Syst. Integration, Beijing, China, 2018, [29] I. E. O. Schweitzer, K. Behrendt, and T. Lee, “Digital communications
pp. 1–5. for power system protection: Security, availability, and speed,” SEL J. Rel.
[17] S. Parhizi, H. Lotfi, A. Khodaei, and S. Bahramirad, “State of the art in Power, vol. 1, no. 1, 2010.
research on microgrids: A review,” IEEE Access, vol. 3, pp. 890–925, [30] D. P. Dickinson, “So many wireless technologies which is the right one
2015. for my application?,” 2015.
[18] H. R. Baghaee, M. Mirsalim, G. B. Gharehpetian, and H. A. Talebi, “A [31] Using the Universal Asynchronous Receiver Transmitter (UART) eTPU
new current limiting strategy and fault model to improve fault ride-through Function, Vernon Goler, Austin, TX, USA, 2004.
capability of inverter interfaced DERs in autonomous microgrids,” Sustain. [32] A. H. K. Alaboudy and H. H. Zeineldin, “Critical clearing time for isolating
Energy Technol. Assessment, vol. 24, pp. 71–81, Feb. 2017. microgrids with inverter and synchronous based distributed generation,”
[19] A. H. K. Alaboudy, H. H. Zeineldin, and J. Kirtley, “Microgrid stability in Proc. IEEE PES General Meeting, 2010, pp. 1–6.
characterization subsequent to fault-triggered islanding incidents,” IEEE [33] Short-circuit currents in three-phase a.c. systems—Part 0: Calculation of
Trans. Power Del., vol. 27, no. 2, pp. 658–669, Apr. 2012. currents, IEC Standard 60909-0, 2016.

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY TIRUCHIRAPALLI. Downloaded on April 14,2022 at 02:48:41 UTC from IEEE Xplore. Restrictions apply.