Protection of Meshed Microgrids with

Communication Overlay
Anish Prasai†, Yi Du†, Andrew Paquette†, Edward Buck‡, Ronald Harley†, and Deepak
Divan†
†Georgia Institute of Technology ‡Eaton Corporation
Department of Electrical and Computer Engineering Innovation Center
777 Atlantic Dr. NW 4201 N. 27th St
Atlanta, GA 30332 Milwaukee, WI 53216-1807

Abstract -- The higher reliability of meshed grids, whether in [8]. However, protection in a meshed microgrid is a
power or communication, when compared with radial grids is nontrivial issue where standard protection schemes are not
well established. However, while meshed grids provide higher applicable. This is due to the fact that the concept of
reliability, protection of meshed grids is challenging. In smaller upstream and downstream does not exist and power flow can
systems such as microgrids, line impedances can be small and reverse its direction through a line segment dynamically,
many branches may experience approximately the same level of
fault current. Locating faults quickly before damage to the
depending on the source and load profile. In such a scenario,
equipment and system occurs becomes a non-trivial task and if a fault were to occur anywhere in the network, identifying
cannot be achieved without communication. This paper and isolating the fault becomes difficult without
discusses a comprehensive scheme for protecting meshed communication.
microgrids with multiple levels of protection for redundancy. Using differential relays with a communication overlay
Power line carrier (PLC) technology is proposed as cost effective through fiberoptics or Cat-5 cables has been proposed for a
and robust means of communication, obviating the need for microgrid with a ring bus architecture which can be extended
additional cabling. to a meshed network as well, but such a system requires
Index Terms—Microgrid, meshed, protection, Power line
significant investment in communication infrastructure [2],
carrier, differential protection [9]-[10]. This paper provides a set of innovative approaches
based on existing schemes for protection of meshed
I. INTRODUCTION microgrids with the use of a low cost communication
protocol, power line carrier (PLC), which is already widely
Redundancy is critical in realizing high levels of reliability,
used by the utilities [11]. The focus will be on the network
and similar to distribution networks, where each distribution
feeder may supply multiple network transformers with the portion of the meshed microgrids, with brief discussions of
secondaries of the network transformers connected together, a load and source protection provided to establish their
microgrid must have redundancy built-in as well at the compliance with the proposed network protection schemes.
system level. Thus, should a fault occur at any one or more
lines, there are other alternate paths through which power can II. UNDERLYING CONTROL OF SOURCES AND SYSTEM
flow to the system loads. An approach used in distribution In this work, the individual sources are operated based on
systems has been to network or “mesh” the system; Con frequency and voltage droop to provide the real and reactive
Edison distributes approximately 86% of their electricity power sharing, respectively, as shown in Fig. 5a and 5b. This
through underground meshed low voltage network with over enables the control and dynamic sharing of output power
80,000 miles of cable in New York City alone [1]. However, based on only local measurements when additional sources
this approach has not been adequately applied to microgrids. and loads are connected and/or disconnected from the
The closest thing to a “meshed” network that has been microgrid. Various implementations of droop control for
proposed for microgrids has been a ring bus architecture with
generators and inverters have been proposed in the literature
N-1 level of redundancy [2]. Moreover, by meshing the
[12]-[15].
system more heavily, and providing redundant paths for
power flow, reliability can greatly be increased, not only This work employs a supervisory controller (SC) at the
during post-fault contingencies, but also when undergoing system level overlaid above the autonomous droop control of
repairs or expansions of the network where the majority, if the individual sources. In conjunction with system wide
not all, of the loads can remain online. communication, the SC obtains the measured states of the
While there are publications on the topic of protection of system and pushes the operating state to its optimal operating
microgrids, these works typically target radial networks [3]- point in steady-state by modifying the droop settings and/or

978-1-4244-5287-3/10/$26.00 ©2010 IEEE 64

turning selected sources on/off based on load demand. modes of operation, transitions from one mode to another are
simplified. Here, the microgrid is assumed to have adequate
generation capacity to supply all the loads during the islanded
mode of operation.
With a stiff grid during grid-connected mode, the droop
settings of the sources in the microgrid are updated by the SC
to realize the desired power sharing. A fast static switch is
typically used to make the grid connection so if either the
utility grid or the microgrid encounters a fault or becomes
unreliable, the two systems are isolated [3]. Various
(a) publications address the protection requirements and strategy
for protecting the utility grid – microgrid interface [3]-[8].

III. PROTECTION SCHEME

Protection of a meshed microgrid is divided into four
different zones, similar to a radial grid: source, load, bus and
network. Source protection isolates a faulted source from the
rest of the microgrid, load protection isolates a load side fault
from the rest of the network, bus protection detects and
(b) isolates a fault when it occurs on the busses, and network
Fig. 1. a) Frequency droop to control real power, and b) voltage droop to protection detects and isolates a fault in the lines
control reactive power. interconnecting the busses. In some cases, the load can also
act as a source, whether it’s a regenerative type load, or a
The supervisory controller does not attempt to control fleet of EV/PHEVs. Without knowing what type of load will
individual components or sources on an instantaneous basis be connected to the microgrid, the load protection needs to
or manage the transient behavior but only attempts to take into account bi-directional power flows. Fig. 2 depicts
optimize the steady-state operation from energy efficiency the architecture of an example meshed microgrid considered
and system performance perspectives. Thus, should either in this paper with the different zones of protection indicated
the communication system or the supervisory controller fail, by the colored lines and circuit breakers. Bus protection
the microgrid can continue operating reliably, although less circuit breakers are not explicitly highlighted as all the
than optimally. breakers connected to a given bus are involved in isolating
The two primary modes of operation for a microgrid is the faulted bus. The directionality information of each relay
grid-connected and islanded mode of operation. By operating associated with the circuit breakers are given by the arrow
next to each circuit breaker. For bus and network protection,

Fig. 2. Architecture of the microgrid with four different zones of protection where all the breakers in a given bus is involved in bus protection.
the sources with the voltage and frequency droop during both this directionality only defines the polarity of the current,

65
whereas for load and source relays, it implies the direction of ends of each line so the PLC signals do not propagate to the
power flow as well. When power flows or currents are in the busses and subsequently onto the adjacent lines. For the
direction indicated by the arrow, the value is considered to be cases where communication with other busses or the
positive, otherwise, they are negative. supervisory controller is necessary, each bus has a network
While much of the protection schemes discussed in this router that will route the data packets through specific lines to
paper have been applied in the utility grid in some form or the intended bus or device on the network. In essence, the
another, this paper addresses unique approaches to re- communication infrastructure itself is a meshed network
applying these protection schemes to a meshed microgrid where each bus is a node, as exemplified by Fig. 4b. Similar
network. to the power network, the meshed communication network
As communication is an integral part of the network offers higher reliability through redundancy, such that when
protection in this paper, a brief discussion of the power line one or more nodes fail, the entire system is not compromised.
carrier (PLC) technology proposed for the meshed network
protection is critical to understanding the design of the TCP Througput for each individual stream of a simultaneous
50MByte transmission using the 14Mbps PLC devices
protection system. 5

4.5

A. Power Line Carrier (PLC)-Based Communication 4

TCP Throguhput (Mbps)

3.5

Various communication mediums are employed by the 3

utilities for protection and control including leased phone

2.5

lines, pilot wires, fiber optics, microwave, point to point 1.5

radio, and PLC. Availability, reliability, cost, and bandwidth 1

0.5
are all considerations that go into choosing the appropriate 0

communication technology. PLC provides a state-of-the-art 1 2 3 4

Number of simultaneous traffic streams

6

communication standard that is secure, uses existing Fig. 3. Laboratory test of data throughput of a 14 Mbps PLC.
infrastructure with no additional cabling, and offers high
reliability with low cost.
PLC has been used for communication and protection in
substations and transmission lines since the late 1920’s. In
transmission and distribution applications, as well as in some
low voltage utility applications such as automated meter
reading, frequencies of 30 kHz to 500 kHz are typically used
[11]. In broadband power line and in-home local area power
line network applications, frequencies well above 1 MHz are
used with data rates up to 200 Mbps [16].
(a)
Digital PLC modulation techniques such as orthogonal
frequency division multiplexing (OFDM) allow good
bandwidth utilization, ability to cope with interference and
multipath effects, and provide high data rates in excess of 100
kbps [17]. Full differential current comparison typically
requires a data rate in the range of 56 kbps to 115 kbps [18],
thus making digital PLC quite suitable for the protection
scheme proposed in this paper. Based on laboratory testing,
even under harsh environmental conditions (SNR < 0 dB), the (b)
PLC protocol exhibits high throughput with relatively low Fig. 4. a) Simplified diagram of the communication architecture using PLC,
and b) meshed communication network of the example microgrid that
errors when transmitting over 300 ft of cable. Fig. 3 shows provides high reliability through redundancy.
that for a 14Mbps PLC, throughput for a single stream of data
was 4.5 Mbps, which is more than enough bandwidth for the B. Network Protection
protection scheme proposed in this paper. The network fault is defined in this paper as the fault that
The PLC signal is typically coupled onto the line using occurs on a line connecting two busses. These lines are given
coupling capacitors, a line tuner for impedance matching, and in blue in Fig. 2. In a meshed network, power can flow in
a transceiver. The carrier signal is prevented from entering any direction, and the concept of ‘upstream’ or ‘downstream’
other parts of the network by a signal trap in the form of an from a fault does not exist, rendering identification of the
LC filter connected in series with the line. The line trap also fault location a non-trivial task. For example, in Fig. 2, if a
prevents the carrier signal from being shorted by an external network fault occurs on Line 2 connecting Bus 3 and 6, from
fault. While the LC filter is a popular configuration of the the perspective of Bus 3, this fault may as well have occurred
line trap, other circuit configuration may also be employed. on the line connecting Bus 6 and 5. Similarly, from the
The proposed communication architecture for the meshed
perspective of Bus 6, the fault could also have occurred on
microgrid is depicted by Fig. 4a, where PLC communication
the line connecting Bus 3 and 1.
is restricted to the lines/cables only, and traps are set at the

66
 For network protection, unit protection is needed which of the faulted branch. However, while communication is
reliably determines if the fault is on the line being protected. vital for the nominal operation of the system, the
Traditional network protection schemes include distance communication itself can fail. Therefore, suitable backup
relays or phase relays, or communications enabled schemes protection must be designed in case of communication
such as directional comparison, directional blocking and failure.
permissive schemes or differential current protection. Small
TABLE I. FAULT CURRENT FOR LINE-TO-LINE FAULT AT LINE 1.
line impedances, and lines which may not be primarily Current Mag. Pos. Seq. Neg. Seq.
inductive make phase and distance relays unsuitable for Relay Location (kA) (kA) (kA)
microgrid protection. Overcurrent schemes which may rely Bus 1, Relay 1 0.86 0.63 0.42
on phase current, or negative- or zero-sequence current are Bus 1, Relay 2 1.21 0.69 0.69
Bus 5, Relay 2 0.37 0.18 0.19
also ineffective in achieving the unit protection necessary for
Bus 6, Relay 2 1.11 0.58 0.58
network protection. Coordination between relays using time Bus 6, Relay 3 0.85 0.60 0.39
delays based on breaker location or distance to fault is not
applicable due to the meshed architecture and small line TABLE II. FAULT CURRENT FOR LINE-TO-LINE FAULT AT LINE 2.
impedances. The only information individual busses are able Current Mag. Pos. Seq. Neg. Seq.
Relay Location (kA) (kA) (kA)
to collect based on local measurements, and without
Bus 1, Relay 1 0.85 0.62 0.39
communication with other relays, is the magnitude and Bus 1, Relay 2 1.09 0.58 0.57
direction of the instantaneous fault current. This information Bus 5, Relay 2 0.35 0.13 0.25
alone is inadequate for network protection. Thus, traditional Bus 6, Relay 2 1.21 0.69 0.69
protection schemes proposed for radial microgrids cannot be Bus 6, Relay 3 0.86 0.61 0.40
effectively utilized. For instance, applying the traditional
TABLE III. FAULT CURRENT FOR LINE-TO-LINE FAULT AT LINE 1 WITH THE
overcurrent based protection would have the undesirable GENERATOR AT BUS 1 SWITCHED OFF.
effect of opening all the breakers in a meshed network that Current Mag. Pos. Seq. Neg. Seq.
see the fault current and potentially causing a system-wide Relay Location (kA) (kA) (kA)
blackout. Bus 1, Relay 1 0 0 0
Bus 1, Relay 2 0.97 0.52 0.55
The system shown in Fig. 2 is simulated in PSCAD with a Bus 5, Relay 2 0.32 0.16 0.15
line-to-line fault at two different locations: Line 1 and 2. Bus 6, Relay 2 0.95 0.55 0.53
Detailed diesel gen-set, and average inverter models are used Bus 6, Relay 3 1.07 0.81 0.48
in the simulation. The wind turbines are modeled as PM
machines with back-to-back inverters. In this example, the Three different levels of protection, from primary to
microgrid has already been operating in an islanded mode of tertiary, are proposed for protecting the meshed network,
operation when the fault occurs. The fault current levels at where should one level fail to isolate the fault in a timely
select locations are tabulated in Table I and II for the line-to- manner, the next level is automatically executed. The
line fault on Line 1 and 2, respectively. Fault currents are primary protection describes a protection scheme with a fully
shown at various places throughout the microgrid to show operational communication system in all the lines, the
that fault current levels are very similar throughout the secondary scheme is triggered if communication over a given
network. Table I and II show that current magnitude and and line is lost before or during a fault, and the tertiary, or the
direction are not sufficient to identify the fault location. backup protection, proposes an approach if both the primary
Further, as the fossil fuel gen-sets are switched on/off by and secondary protection fail to isolate the fault in a timely
the supervisory controller to maximize fuel efficiency and manner.
track the load profile of the system [19], or power output of 1) Primary Protection: Differential Protection with
renewable resources vary based on weather conditions, the Communication
fault current levels can vary significantly over a day. This In a meshed microgrid with distributed sources, when a
characteristic is exhibited by Table III where the generator fault occurs on a line, the fault is fed from both ends of the
connected to Bus 1 is off when a line-to-line fault occurs on line. For example, the fault in Fig. 5 that occurred on Line 2
Line 1. The differences in fault currents observed between will be fed by the sources connected to Busses 3 and 6,
Table I and III has the potential to negatively impact the among others. The breakers on Busses 3 and 6 should trip to
response of the relays with static trip settings. This issue is isolate this fault. However, Busses 1, 4, and 5 may also
further exacerbated if only inverter based sources were online experience the fault current and under traditional protection
when the fault occured, where the fault currents are limited to schemes, may inadvertently trip as well. Thus, the protection
2-3 pu. For this reason, overcurrent relays with static settings scheme must ensure that the breakers in these three busses do
are not suitable. not trip and unnecessarily reduce the available transfer
The answer to the question on how to automatically capacity (ATC) of the system, or disconnect other sources
identify the faulted section of the line and isolate just that and critical loads.
section without bringing down the other lines is vital towards To adequately address this problem, differential current
reliably operating a meshed microgrid. Communication protection is used to identify whether a fault has occurred on
between busses is the key towards coordinating the location that line. Normally, the current entering the line equals the

67
current exiting the line, and the differential current is zero.
But when there is a fault on the line, the currents do not sum
to zero, which can then be used to initiate appropriate
protection process. This is the same approach used for bus
and transformer protection in most substations [20], and the
same idea has been previously discussed for protection in a
ring-bus microgrid architecture [2]. Lines can extend over
relatively long distances, and a cost-effective and reliable
communication system is required to quickly transmit
(a)
information to adjacent busses. Power line carrier (PLC)
based communication can be used on the existing power
lines, obviating the need for additional cabling, and reducing
0.2
the capital required for a new communication infrastructure.

VBus4(kV)
Communicating actual current magnitude and direction 0
ensures that the line is protected from both low and high -0.2
impedance faults of all types (SLG, LL, LLG, etc). In larger 1.99 2 2.01 2.02 2.03 2.04
systems extending over relatively long distances, it may also 1
become necessary to pass GPS-synchronized time stamps

IB4-brk3 (kA)
such that the current comparisons are synchronized. 0
In a very noisy communication environment, passing
relatively large packets of three-phase current magnitude and -1
1.99 2 2.01 2.02 2.03 2.04
directional data may take too long. In such a scenario, only
the direction and GPS-synchronized time stamps are passed.
I Line-diff (kA)
4 Line 4
However, while this scheme will provide protection from low 2 Line 2,3,6
impedance faults, high impedance faults may be harder to
detect without the magnitude comparisons as well. 0

Once the fault has been identified, the breakers at the two 1.99 2 2.01 2.02 2.03 2.04
ends of the line are tripped. Reclosers may be used to try to 1
ILine2 (kA)

restore the line, but if the fault persists, the breakers remain
open until manually closed. The PLC communication 0
remains active with the breakers open, as long as the signal is -1
not shorted or corrupted by the fault, since the PLC modules 1.99 2 2.01 2.02 2.03 2.04
are placed on the line side of the signal traps, as shown in Time(s)
Fig. 4. However, once the faulted branch has been isolated,
further communication is no longer required for the (b)
Fig. 5. a) A line-to-line fault occurs on Line 4, with b) the simulation results
protection of that particular line. If the communication is still showing the corresponding three-phase voltages on Bus 4, currents through
operational over that line, only the data being routed from Bus 4/Breaker 3, differential currents through each lines, and current through
other busses is passed through the faulted line. Line 2.
In the case of a breaker failure, after a brief time delay, the
other breakers connected to the same bus will close in order 2) Secondary Protection: Differential Protection with
to isolate the faulted line. Loss of Channel
A line-to-line fault on Line 4 is simulated in PSCAD as If a fault occurs in the meshed network and a
depicted in Fig. 5a. Three-phase voltages and currents communication on the faulted line fails due to impedance
through Breaker 3 on Bus 4, current through Line 2, and mismatch and other issues, then the line is immediately
differential currents of the faulted line and other lines are tripped, effectively assuming that the fault has caused the
plotted in Fig. 5b. The fault occurs at time t = 2 s. The communication failure and thus the fault has occurred on that
differential currents of Line 2, 3 and 6 remain close to zero, particular line. The health of the communication link over a
while the differential current of Line 4 jumps sharply. When particular line is continuously being monitored at all times,
the current exceeds the predefined threshold value on Line 4, and when emergence of fault coincides with the loss of
a trip signal is sent to both Breaker 1 on Bus 6 and Breaker 3 communication, then the line fault and communication loss
on Bus 4. The fault is isolated after the arc is extinguished. are highly correlated and appropriate protective action is
During the fault in this example, Line 2 sustains the fault sequenced.
current until the fault is cleared. Signal traps are used at the two ends of every line to
prevent fault on a line from shorting out or corrupting the
PLC signal on the other unfaulted lines. These traps also
ensure that the PLC signals do not collide with the signals in
the other lines, improving throughput, and reducing noise.

68
 If a communication failure has occurred in a line, prior to normally used due to its high sensitivity, stability and speed
the fault, then determining whether a fault has occurred in [20].
this particular line becomes more challenging. Because the The basic configuration for a differentially connected
grid is meshed, the data can be transmitted to the adjacent overcurrent bus protection scheme is shown in Fig. 6. The
busses through alternate paths, around the line with the output of each CT is summed by connecting the secondaries
communication failure. This is achieved through in parallel. Under normal conditions, and during external
communication routers, similar to the ones used in directing faults, the currents sum to zero. For an internal fault, the
internet traffic, installed on the busses which intercepts the currents do not sum to zero and the differential current flows
signal before the signal trap, and relays the packets of through the overcurrent relay.
information onto the appropriate line, as depicted in Fig. 4a. If CTs saturate due to large fault currents, incorrect
As speed of communication is a critical element of this differential current is induced. The simplest way to address
protection scheme, in relatively large systems, or in a noisy CT saturation is to react to faults within the first ¼ cycle or
environment, only the direction and GPS-synchronized time less before the CTs actually saturate. However, this may not
stamp is routed. However, with just a directional component be economically feasible due to requirement for fast
and without a magnitude value, this scheme is only able to responding and expensive sensors and solid state switches.
detect and isolate low impedance faults. One approach to addressing this issue is through the
The SC is able to protect against high impedance faults utilization of high impedance differential relaying. With this
based on the data it is continuously receiving on the operating relay, the ‘false’ differential current is forced to flow through
states of each busses over a slower time period. Once the SC the saturated CTs instead of activating the relay. However,
has identified that a high impedance fault exists in a the preferred approach to addressing CT saturation is the
particular branch, it will send the trip signal to the breakers of percentage restrained differential relaying, where the
that branch. differential current is compared to the total current, and
3) Tertiary/Backup Protection activated only when this ratio exceeds a threshold.
In certain cases, the data may take undesirably long time to Algorithms for CT saturation detection are used to improve
reach the remote bus due to the circuitous route, and in large the performance. While the high impedance differential
microgrids, this time delay could be excessive. In other cases, relaying requires dedicated CTs with matched ratios, the
there may be a momentary communication failure over the percentage restrained differential relaying can tolerate lower
entire microgrid. If the data does not successfully reach the quality CTs with mismatched ratios, and CTs can be shared
remote bus, backup protection is responsible for clearing the with other circuits or relays.
fault.
In such a scenario, after certain time has passed, and fault Bus 4
Protected
Zone
has not been cleared, then all the lines that experience the
relatively high current due to low impedance fault will trip. 1 2 3
The states of the relays prior to the system-wide trip are
stored and the circuitry in the busses will pass the data onto Overcurrent
Relay
the supervisory controller once the breakers are tripped. The
supervisory controller, upon collecting the data from each of
the busses and subsequently identifying where the fault has Fig. 6. Basic configuration for differential current bus protection, where
occurred, will trigger all the breakers that had tripped to the current in each branch is summed by connecting CT secondaries in
reclose in a sequential and coordinated fashion, restoring parallel.
service to everything except the faulted line.
Backup protection is coordinated by using time delays to Differential bus protection is simulated in PSCAD where a
give sufficient time for the primary protection to act. Time three phase fault at Bus 4 occurs, as shown in Fig. 7. The
delayed overcurrent relays may be used. voltage and differential current at Bus 4, and Bus 6, and the
In the case of system wide communication failure lasting current through Breaker 1 on Bus 4 are shown in Fig. 8.
for longer periods of time where the SC is not able to restore When the differential current in Bus 4 exceeds the threshold
service remotely, fault direction indicators on each relays are value in a half cycle, a trip signal is sent to Breaker 1, 2 and 3
used to allow utility personnel to identify the faulted line and of Bus 4. The differential current in Bus 6 remains zero. The
take appropriate actions manually. However, due to the fault is isolated after the fault current goes through zero
redundant nature of the communication architecture, where crossing. The voltage on Bus 6 subsequently recovers to
independent communication nodes exist at each bus, system- nominal levels once the fault is isolated.
wide communication failure is not very probable. D. Protection of Uni- and Bi-Directional Loads
C. Bus Protection The busses connected to bi-directional loads such as
Bus protection has to be highly sensitive to internal faults regenerative type loads, and energy storage in the form of
that induce even the smallest current, and be completely EV/PHEVs, experience power flow in both forward and
impervious to external faults with current many times greater reverse directions. The chosen protection scheme must be
than the nominal levels. Differential current protection is able to take this into account, and protect the load and the
network without any communication. Communication-less

69
 protection is desired here as enforcing a standardized than the peak load current. However, in a microgrid with
communication protocol on every load is not practically switching generators and varying output power of renewable
feasible. This is not to say that communication should not be sources, the short circuit current of the system can vary
in place to provide flexible charging to the EV/PHEVs. continuously. For this reason, the trip settings will be
However, the protection is not dependent on communication. modified by the SC remotely through communication for
each of the relays based on the available generation capacity
of the system. Load unbalances are taken into account when
zero and negative sequence components are employed.
Detection of high impedance fault without a differential
current measurement schemes is more difficult. Commercial
products such as GE’s Digital Feeder Monitor (DFM), and
Nordon Technologies’ High Impedance Fault Analysis
System (HIFAS) both use harmonic analysis and advanced
algorithm to detect presence of high impedance fault. These
commercial products may be utilized at the load feeders for
detecting the high impedance fault.
For load centers with local sources and energy storage,
Fig. 7. A three-phase fault occurs on Bus 4. power can flow in both directions. If these sources operate in
a current/power mode without voltage mode, upon
disconnecting from the microgrid, they shut down and no
further action is needed. If they can transition to islanded
0.2 mode, then upon detecting fault or instability on the
(kV)

0
microgrid, they isolate themselves and operate in an islanded
Bus4

mode.
V

-0.2
1.99 2 2.01 2.02 2.03 2.04

1
(kA)
B4-brk1

-1
I

1.99 2 2.01 2.02 2.03 2.04

4
(kA)

Bus 4 Fig. 9. Fault has occurred on the line connecting the microgrid to an
Bus 6
office building.
Bus-diff

0
2) Load Fault
I

1.99 2 2.01 2.02 2.03 2.04

If a fault occurs within the load center itself, then
0.2 protection in the form of fuses and low-voltage breakers are
(kV)

0
utilized. The trip setting of these elements are typically set
Bus6

based on load ratings and are designed to trip quickly so the

V

-0.2
upstream breaker, like Breaker 2 in Fig. 9, interfacing the
1.99 2 2.01 2.02 2.03 2.04
Time(s) microgrid, does not trip first and completely disconnect all
the loads connected to that branch.
Fig. 8. Voltages and currents of Bus 4 and Bus 6 during a three phase to ground E. Protection of Sources
fault of Bus 4.
Protection of a source, such as a diesel or natural gas gen-
1) Line Fault set or even renewable resource like wind turbines and PV
For a low impedance fault that occurs on the line between arrays have been discussed by various papers [3]-[8]. These
the bus and the load, the bus will feed the fault. In Fig. 9, sources must be able to sustain the fault currents for longer
Bus 4 feeds the fault that occurred on the line supplying an periods of time without tripping in order to give the
office building. Here, the traditional protection schemes microgrid’s protection system enough time to identify the
utilized in radial networks are applicable based on relays that faulted branch and initiate appropriate protection measures.
sense non-directional overcurrent, negative and zero- In the case of power electronic-interfaced sources like wind
sequence overcurrent, directional earth-fault current, turbines, PV arrays, legacy generators that are not compliant
over/under voltage and frequency, zero-sequence with droop based controls and cannot be interfaced directly,
overvoltage, etc [3]-[8]. The trip settings for current based and in certain cases larger energy storage systems that are
relays are based on the peak load current drawn by that rated to support the microgrid during contingencies, there is a
particular branch. This value can typically be 2-3X higher lower limit on the maximum fault current they can sustain –
typically 2-3 pu. Thus, there must be an adequate amount of

70
short-circuit capacity available in the system to ensure that REFERENCES
the fault currents are sustained for long enough periods of [1] “Overview of Con Edison System and LIC Network”, Con Edison, Jan
time to identify and isolate the fault without dropping critical 22, 2010, <http://www.coned.com/messages/LICReport/Overview.pdf>.
loads. [2] E. Sortomme, S. S. Venkata, and J. Mitra, "Microgrid Protection Using
Communication-Assisted Digital Relays," IEEE Transactions on Power
In the case when a fault occurs on the microgrid, and there Delivery, Accepted for Publication in Future Issue.
is not enough short-circuit capacity on the system, then all the [3] H. Nikkhajoei, and R.H. Lasseter, “Microgrid Protection,” in Conf.
breakers will trip, and the network protection scheme Proceed. of IEEE Power Engineering Society General Meeting 2007, pp. 1-6,
discussed in Sec. III.B.3 under Backup Protection will be June 2007.
[4] M. Dewadasa, R. Majumder, A. Ghost, and G. Ledwich, “Control and
employed to identify and isolate the location of the fault protection of a microgrid with converter interfaced micro sources,” in Conf.
offline, and systematically have the SC bring-up the system Proceed of International Conference on Power Systems, 2009, pp. 1-6, Dec.
again. However, this method should be used as the last resort 2009.
as it entails interruption of power to most or all of the loads. [5] T. Loix, T. Wijnhoven, and G. Deconinck, “Protection of microgrids
with a high penetration of inverter-coupled energy sources,” in Conf.
In Fig. 10, a fault has occurred on the line between the Proceed. of CIGRE/IEEE PES Joint Symposium on Integration of Wide-
generator and Bus 1. However, from the perspective of the Scale Renewable Resources into the Power Delivery System, pp. 1-6, July
generator, this fault could just as well have occurred past Bus 2009.
1 in an entirely different line. So, the generator needs to [6] D. Salomonsson, L. Soder, and A. Sannino, “Protection of Low-
Voltage DC Microgrids,” IEEE Trans. on Power Delivery, vol. 24, iss. 3, pp.
sustain the fault for a pre-determined amount of time, time 1045-1053, April 2009.
enough to identify the faulted branch or component. If the [7] S. Conti, L. Raffa, and U. Vagliasindi, “Innovative solutions for
fault is still not cleared by the end of the selected period, the protection schemes in autonomous MV micro-grids,” in Conf. Proceed. of
generator assumes the fault is located on the line interfacing it International Conference on Clean Electric Power, 2009, pp. 647-654, June
2009.
with the microgrid, and subsequently trips. [8] B. Li, Y. Li, Z. Bo, and A. Klimek, “Design of protection and control
The circuit breaker, Breaker 1 in Fig. 10, is also scheme for microgrid systems,” in Conf. Proceed. of the 44th International
configured to trip as soon as the associated relay detects Universities Power Engineering Conference, 2009, pp. 1-5, Sept. 2009.
reversal in current/power flow. [9] J. J. Meeuwsen, W. L. Kling, and S. P. J. Rombouts, "The influence of
protective relay schemes on the reliability indices of load points in meshed
operated MV networks," in Electricity Distribution. Part 1: Contributions.
CIRED. 14th International Conference and Exhibition on (IEE Conf. Publ.
No. 438), pp. 14/1-14/5 vol.4, 1997.
[10] Schweitzer Engineering Laboratories, Inc., “SEL-387L Line Current
Differential Relay,” Accessed Jan 20, 2010, <https://www.selinc.com/SEL-
387L>.
[11] "IEEE Guide for Power-Line Carrier Applications," IEEE Std 643-
2004 (Revision of IEEE Std 643-1980) , vol., no., pp.0_1-134, 2005
[12] K. De Brabandere, B. Bolsens, J. Van Den Keybus, A. Woyte, J.
Drisen, and R. Belmans, “A Voltage and Frequency Droop Control Method
Fig. 10. A fault has occurred on the line connecting a generator to a bus. for Parallel Inverters,” in IEEE Transactions on Power Electronics, vol. 22,
iss. 4, pp 1107-1115, July 2007.
[13] P. Piagi, and R.H. Lasseter, “Autonomous control of microgrids,” in
IV. CONCLUSION Conf. Proceed. of IEEE Power Engineering Society General Meeting 2006,
Oct 2006.
This paper proposes a multi-level approach to provide the [14] S. Krishnamurthy, T.M. Jahns, and R.H. Lasseter, “The Operation of
most effective form of network protection of a meshed Diesel Gensets in a CERTS Microgrid,” in Conf. Proceed. of 2008 IEEE
microgrid, while ensuring a high level of reliability and Power and Energy Society General Meeting – Conversion and Delivery of
power quality by quickly and automatically identifying Electrical Energy in the 21st Centruy, pp. 1-8, July 2008.
[15] H. Nikkhajoei, and R.H. Lasseter, “Distributed Generation Interface to
faulted points in the system, and effectively isolating them. the CERTS Microgrid,” in IEEE Transactions on Power Delivery, vol. 24,
The paper has clearly shown that the traditional iss. 3, pp. 1598-1608, July 2009.
communication-less protection schemes are not applicable in [16] Atheros Communications. "White Paper - Powerline Communications
a meshed microgrid where fault at one location is Performance Testing," Accessed June 20, 2010,
http://www.atheros.com/pt/plc/downloads/whitepaper_PLCPerformanceTesti
indistinguishable from another. Further, the variable short- ng.pdf
circuit current levels of the system due to changing [17] PRIME Alliance. “MAC Spec White Paper,” Accessed June 27, 2010,
generation capacity makes traditional overcurrent relays with <http://www.prime-alliance.org/portals/0/specs/
static settings ineffective. Protection based on differential MAC_Spec_white_paper_1_0_080721.pdf.
[18] R. Moxley and K. Fodero, "High-speed distribution protection made
current measurement and comparison is proposed that utilizes easy: communications-assisted protection schemes for distribution
the power line itself as the communication medium. While applications," in Protective Relay Engineers, 2005 58th Annual Conference
the primary protection scheme is dependent on for, pp. 18-26, 2005.
communication, due to the redundant nature of the meshed [19] A. Prasai, A. Paquette, Y. Du, R. Harley, and D. Divan, “Minimizing
Emissions in Microgrids While Meeting Reliability and Power Quality
network, the communication architecture is highly reliable. Objectives,” in Conf. Proceed. of International Power Electronics
However, backup protection scheme is also proposed in the Conference 2010, June 21-24 2010, unpublished.
rare event that the communication does fail. [20] "IEEE Guide for Protective Relay Applications to Power System
Busses," IEEE Std C37.234-2009, pp. C1-115, 2009.

71