25-02-2013, 03:21 PM
Fiber-Optics-Based Fault Detection in Power Systems
Fiber-Optics-Based Fault.pdf (Size: 760.3 KB / Downloads: 57)
Abstract
A fiber-optics-based sensing network applicable for
fault detection in power system is presented. The proposed scheme
is secure and immune from interferences. At each monitoring location,
passive rugged fiber-Bragg-grating-based sensors are deployed.
They use fast and compact magnetostrictive transducers
instead of current or potential transformers to translate currentinduced
magnetic field into optical signal. These sensors can be
compensated for temperature drift and easily be integrated into
an optical sensing network. A broadband light source at a substation
scans the change in reflected optical power at a unique frequency
band that corresponds to the surge in magnetic field associated
with an increased fault current at a certain location. A
unique feature of this real-time scheme is that it only requires current
information for fault detections in both radial and networked
systems with various pole structures and line configurations. It can
easily coordinate with other protective devices and is free from
any time-current coordination curves. The proposed scheme has
been extensively tested by simulations. They confirm that the proposed
scheme is able to detect the faults irrespective of the type
and location. It also performs well in presence of harmonics, high
impedance, and sensors malfunctions, as well as sensor noise.
INTRODUCTION
CONVENTIONALLY, a sequence of relays and circuit
breakers (CBs) are deployed along power systems.
When a fault occurs, the corresponding current transformers
(CTs) and/or potential transformers (PTs) send fault current/
voltage information to relays at the substation. Based on
the current/voltage information and pre-set rules (a family of
time-current tripping curves), the relays will then command
appropriate breakers to trip. If the primary relay fails to clear
the fault, the backup relay designated by the rules will open the
next responsible breakers.
Besides the difficulty of coordinating different protective devices
in a large system, the saturation and hysteresis incurred
in the CT response during a large surge in fault current is another
major challenge faced by the conventional fault detection
method. Most protective devices make operating decisions
based on root mean square (rms) values of fault currents. If the
CT signal is distorted by saturation, its rms value will be much
lower than the actual fault current. In reality, this can interfere
the coordination among relays.
OPTICAL SENSING NETWORK
The proposed sensing network uses magnetostriction based
fiber-optics current sensors (FOCSs) [8]–[10] to transform magnetic
field from a current into mechanical strain on a FBG. Particularly,
giant magnetostrictive materials, such as Terfenol-D
[11] that has high sensitivity and fast response ( 0.1 ms) are
considered [12]. As these passive FOCSs do not require any
power source and can modulate optical carriers directly; they
afford rapid signal transmission. Since the FBG in each FOCS
reflects power at a unique wavelength band, fault signals from
various FOCSs can co-exist in a single strand of optical fiber
cable, that is, many sensors can be addressed by a single optical
source and monitored by detectors through a single connection.
Such a network is crucial to maintain the robustness and extreme
reliability of the communication grid. Furthermore, optical networks
are inherently immune to electromagnetic interference
(EMI). They are superior to other wire-line systems based on
metallic cables, such as broadband over power line systems that
have intractable EMI problems.
Networked System
Conventionally, the protection of a network system not only
requires different types of relays, but also large number of relays
because of the preset fault current direction. In the proposed
method, since a sensor can monitor the fault current in
both directions, scanners combined with the processors in the
control center or substations can replace a majority of relays in
the system.
A networked system should be equipped with sensors at both
the terminals of a line as the power flow is bidirectional. Each
sensor is assigned a pre-determined current flow direction. For
example, three sensors at assume the current flows from
bus 1 to bus 2. If any one sensor at senses a current flowing
in the pre-set direction, then the phase information of that sensor
will indicate a value close to the reference phase. If the current
flows in the opposite direction, then the phase information will
show a big difference with the reference phase. A networked
system can be divided into several zones. Each zone is equipped
with the sensing network and breakers. A zone is bounded by at
least two sensor locations. For instance, a single line connecting
two buses can constitute a zone. The zone can extend to other
buses and generators in which case the larger zone including
all the inner sensors is bounded by other external sensors. The
innermost zone acts as primary protection whereas the encircling
bigger zone acts as a backup protection. The objective of the zone
is not only to clear the fault by detecting the culpable zone, but
also to minimize the outage area.
FIBER BASED FAULT DETECTION SCHEME
When there is a fault, the surge in magnetic field induced
by the fault current will lead to an increase of reflected optical
power from a sensor at a certain frequency band (unique
to the fault location) when the sensing network is scanned by a
broadband light source at a substation. The average reflectivity
shown in Fig. 2 serves as the magnitude response of the FOCS
for H. The phase angle of H at certain location can be calculated
by finding the zero crossings of the real-time magnetic
field intensities at that location. To generate the phase information,
sensors must be biased with an appropriate DC magnetic
field. The phase information represents the direction of the current,
a valuable indicator of the fault location. The processor will
analyze both the H magnitude and phase information provided
by FOCSs to determine the fault location and its type. However,
this method cannot distinguish between the double-line
and double-line-to-ground faults.
High Impedance
Simulations under impedance fault also validate the proposed
method with impedances up to 100 which are by far the
most common ones in most places. The phase information will
be very instrumental in locating the high impedance fault even
though magnitude surges are typically very small. Figs. 15 and
16 show simulation results for a SLG fault between buses 1 and
2 with impedance of 100 . The surges in sensor readings at
are insufficient to indicate any fault decisively. However,
a closer look at the very small phase information at and
can easily conclude that the fault is between buses 1 and
2. Thus, the phase information is very handy in determining the
fault type and location in case of high-impedance faults.
Harmonics Studies
The effect of harmonics on fault detection is also investigated.
Particularly, Figs. 17 and 18 show sensor readings under a DLG
fault between buses 1 and 2 with 20% of 3rd harmonics, 10%
of 5th harmonics, and 5% of dc components. The difference of
magnitude results in Figs. 17 and 11 is insignificant while the
difference of phase results in Figs. 18 and 12 are very small.
Hence, the scheme does not require any modifications or adjustments
to determine the type and location of the fault.
CONCLUSIONS
A fiber-based sensing network is investigated. In the sensing
network, magnetostrictive-based FOCSs with FBGs are passive
and provide a fast and effective method for real-time
monitoring of line currents various locations simultaneously.
For protection purpose, they can replace conventional PTs
and CTs that are susceptible to saturation and hysteresis.
The fiber-based sensing network enables the development of
new fault detection and location schemes and flexible coordinating
strategies among protective devices without relying
on any time-current coordination curves. Simulation results
show that these schemes performance superbly for both radial
and networked power systems under various conditions
including different types of pole structures and line configurations.
For radial power systems, they usually require magnitude
readings from sensors. For networked power system,
these schemes need additional phase reading.
Fiber-Optics-Based Fault.pdf (Size: 760.3 KB / Downloads: 57)
Abstract
A fiber-optics-based sensing network applicable for
fault detection in power system is presented. The proposed scheme
is secure and immune from interferences. At each monitoring location,
passive rugged fiber-Bragg-grating-based sensors are deployed.
They use fast and compact magnetostrictive transducers
instead of current or potential transformers to translate currentinduced
magnetic field into optical signal. These sensors can be
compensated for temperature drift and easily be integrated into
an optical sensing network. A broadband light source at a substation
scans the change in reflected optical power at a unique frequency
band that corresponds to the surge in magnetic field associated
with an increased fault current at a certain location. A
unique feature of this real-time scheme is that it only requires current
information for fault detections in both radial and networked
systems with various pole structures and line configurations. It can
easily coordinate with other protective devices and is free from
any time-current coordination curves. The proposed scheme has
been extensively tested by simulations. They confirm that the proposed
scheme is able to detect the faults irrespective of the type
and location. It also performs well in presence of harmonics, high
impedance, and sensors malfunctions, as well as sensor noise.
INTRODUCTION
CONVENTIONALLY, a sequence of relays and circuit
breakers (CBs) are deployed along power systems.
When a fault occurs, the corresponding current transformers
(CTs) and/or potential transformers (PTs) send fault current/
voltage information to relays at the substation. Based on
the current/voltage information and pre-set rules (a family of
time-current tripping curves), the relays will then command
appropriate breakers to trip. If the primary relay fails to clear
the fault, the backup relay designated by the rules will open the
next responsible breakers.
Besides the difficulty of coordinating different protective devices
in a large system, the saturation and hysteresis incurred
in the CT response during a large surge in fault current is another
major challenge faced by the conventional fault detection
method. Most protective devices make operating decisions
based on root mean square (rms) values of fault currents. If the
CT signal is distorted by saturation, its rms value will be much
lower than the actual fault current. In reality, this can interfere
the coordination among relays.
OPTICAL SENSING NETWORK
The proposed sensing network uses magnetostriction based
fiber-optics current sensors (FOCSs) [8]–[10] to transform magnetic
field from a current into mechanical strain on a FBG. Particularly,
giant magnetostrictive materials, such as Terfenol-D
[11] that has high sensitivity and fast response ( 0.1 ms) are
considered [12]. As these passive FOCSs do not require any
power source and can modulate optical carriers directly; they
afford rapid signal transmission. Since the FBG in each FOCS
reflects power at a unique wavelength band, fault signals from
various FOCSs can co-exist in a single strand of optical fiber
cable, that is, many sensors can be addressed by a single optical
source and monitored by detectors through a single connection.
Such a network is crucial to maintain the robustness and extreme
reliability of the communication grid. Furthermore, optical networks
are inherently immune to electromagnetic interference
(EMI). They are superior to other wire-line systems based on
metallic cables, such as broadband over power line systems that
have intractable EMI problems.
Networked System
Conventionally, the protection of a network system not only
requires different types of relays, but also large number of relays
because of the preset fault current direction. In the proposed
method, since a sensor can monitor the fault current in
both directions, scanners combined with the processors in the
control center or substations can replace a majority of relays in
the system.
A networked system should be equipped with sensors at both
the terminals of a line as the power flow is bidirectional. Each
sensor is assigned a pre-determined current flow direction. For
example, three sensors at assume the current flows from
bus 1 to bus 2. If any one sensor at senses a current flowing
in the pre-set direction, then the phase information of that sensor
will indicate a value close to the reference phase. If the current
flows in the opposite direction, then the phase information will
show a big difference with the reference phase. A networked
system can be divided into several zones. Each zone is equipped
with the sensing network and breakers. A zone is bounded by at
least two sensor locations. For instance, a single line connecting
two buses can constitute a zone. The zone can extend to other
buses and generators in which case the larger zone including
all the inner sensors is bounded by other external sensors. The
innermost zone acts as primary protection whereas the encircling
bigger zone acts as a backup protection. The objective of the zone
is not only to clear the fault by detecting the culpable zone, but
also to minimize the outage area.
FIBER BASED FAULT DETECTION SCHEME
When there is a fault, the surge in magnetic field induced
by the fault current will lead to an increase of reflected optical
power from a sensor at a certain frequency band (unique
to the fault location) when the sensing network is scanned by a
broadband light source at a substation. The average reflectivity
shown in Fig. 2 serves as the magnitude response of the FOCS
for H. The phase angle of H at certain location can be calculated
by finding the zero crossings of the real-time magnetic
field intensities at that location. To generate the phase information,
sensors must be biased with an appropriate DC magnetic
field. The phase information represents the direction of the current,
a valuable indicator of the fault location. The processor will
analyze both the H magnitude and phase information provided
by FOCSs to determine the fault location and its type. However,
this method cannot distinguish between the double-line
and double-line-to-ground faults.
High Impedance
Simulations under impedance fault also validate the proposed
method with impedances up to 100 which are by far the
most common ones in most places. The phase information will
be very instrumental in locating the high impedance fault even
though magnitude surges are typically very small. Figs. 15 and
16 show simulation results for a SLG fault between buses 1 and
2 with impedance of 100 . The surges in sensor readings at
are insufficient to indicate any fault decisively. However,
a closer look at the very small phase information at and
can easily conclude that the fault is between buses 1 and
2. Thus, the phase information is very handy in determining the
fault type and location in case of high-impedance faults.
Harmonics Studies
The effect of harmonics on fault detection is also investigated.
Particularly, Figs. 17 and 18 show sensor readings under a DLG
fault between buses 1 and 2 with 20% of 3rd harmonics, 10%
of 5th harmonics, and 5% of dc components. The difference of
magnitude results in Figs. 17 and 11 is insignificant while the
difference of phase results in Figs. 18 and 12 are very small.
Hence, the scheme does not require any modifications or adjustments
to determine the type and location of the fault.
CONCLUSIONS
A fiber-based sensing network is investigated. In the sensing
network, magnetostrictive-based FOCSs with FBGs are passive
and provide a fast and effective method for real-time
monitoring of line currents various locations simultaneously.
For protection purpose, they can replace conventional PTs
and CTs that are susceptible to saturation and hysteresis.
The fiber-based sensing network enables the development of
new fault detection and location schemes and flexible coordinating
strategies among protective devices without relying
on any time-current coordination curves. Simulation results
show that these schemes performance superbly for both radial
and networked power systems under various conditions
including different types of pole structures and line configurations.
For radial power systems, they usually require magnitude
readings from sensors. For networked power system,
these schemes need additional phase reading.