t's a snapshot of the page taken as our search engine crawled the Web.
The web site itself may have changed. You can check the current page or check for previous versions at the Internet Archive. Yahoo! is not affiliated with the authors of this page or responsible for its content.
A Microprocessor-based Digital Feeder Monitor with High-Impedance Fault Detection GER-3796
A MICROPROCESSOR-BASED
DIGITAL FEEDER MONITOR
WITH HIGH-IMPEDANCE FAULT DETECTION
R. Patterson
W. Tyska
GE Protection and Control
Malvern, PA
B. Don Russell
B. Michael Aucoin
Department of Electrical Engineering
Texas A&M University
College Station, Texas
Presented to:
Forty-Seventh Annual Conference for Protective
Relay Engineers
Texas A&M University
College Station, Texas
March 21-23, 1994
- 1 -
A Microprocessor-Based Digital Feeder Monitor
with High-Impedance Fault Detection
R. Patterson
W. Tyska
GE Protection and Control
Malvern, PA
B. Don Russell
B. Michael Aucoin
Department of Electrical Engineering
Texas A&M University
College Station, TX
Introduction
The high impedance fault detection technology developed at Texas A&M
University after more than a decade of research, funded in large part by
the Electric Power Research Institute, has been incorporated into a
comprehensive monitoring device f or overhead distribution feeders. This
digital feeder monitor (DFM) uses a high waveform sampling rate for the
ac current and voltage inputs in conjunction with a high-performance
reduced instruction set (RISC) microprocessor to obtain the frequency
response required for arcing fault detection and power quality
measurements. Expert system techniques are employed to assure security
while maintaining dependability. The DFM is intended to be applied at a
distribution substation to monitor one feeder. The DFM is packaged in a
non-drawout case which fits the panel cutout for a GE IAC overcurrent
relay to facilitate retrofits at the majority of sites where
electromechanical overcurrent relays already exist.
High impedance Faults
To understand the performance of the DFM, it is necessary to ,define the
high impedance faults targeted by this device. A high impedance fault is
characterized by having an impedance sufficiently high such that it is
not detected by conventional phase or ground overcurrent protection. A
downed conductor fault occurs when the conductor is no longer intact on
pole top insulators, but instead is broken and in contact with earth or
a grounded based-Digital-Feeder-Monitor-with-High-Impedance-Fault-/' class='doin' >object. An arcing fault is any high impedance fault which
exhibits arcing.
Combinations of these types are possible. An example is an arcing, high
impedance, downed conductor fault. The intent of the DFM is to detect
high impedance faults which arc, and to differentiate those which are
downed conductors from those which are not. Electrical signatures are
used to identify the presence of arcing. If the arcing begins with a
loss of load or with an overcurrent disturbance (as might occur when a
conductor falls across another phase or neutral wire and then falls to
ground) , the DFM assumes that a conductor is down. If neither of these
conditions initiates the arcing, the DFM assumes that the conductor is
still intact. In the interest of system security, - 2 -
the DFM considers loss of load or an overcurrent disturbance to indicate
a downed conductor if and only if one of these starts the arcing, and
not if one these occurs after the initiation of arcing. The reason for
this is that, following a recloser operation, power system load levels
will often change sufficiently such that the DFM cannot distinguish
between a recloser operation and a loss of load due to a broken
conductor.
Algorithms Associated with High Impedance Fault Detection
An algorithm is simply a set of rules for solving a based-Digital-Feeder-Monitor-with-High-Impedance-Fault-/' target='blank' class='doin' >problem. For a
microprocessor-based device, an algorithm is implemented by the software
code run by the microprocessor. In the DFM, the detection of a downed
conductor or arcing condition is accomplished through the execution of
the following algorithms:
Energy Algorithm
Randomness Algorithm
Expert Arc Detector Algorithm
Load Event Detector Algorithm
Load Analysis Algorithm
Load Extraction Algorithm
Are Burst Pattern Analysis Algorithm
Spectral Analysis Algorithm
Arcing Suspected Identifier Algorithm
Energy Algorithm
Arcing causes bursts of energy to register throughout the frequency
spectrum, and they are readily detected at non-fundamental and non-
harmonic frequencies. This characteristic of arcing faults is
represented in Figure 1. The Energy Algorithm monitors a specific set of
non-fundamental frequency component energies of phase and neutral
current. After establishing an average value for a given component
energy, the algorithm indicates arcing if it detects a sudden, sustained
increase in the value of that component. The DFM runs the Energy
Algorithm on each of the following parameters for each phase current and
for the neutral: (1) even harmonics, (2) odd harmonics, and (3) non-
harmonics. on a 60-Hz system, the non-harmonic component consists of a
sum of the 30, 90, 150, ... , 750-Hz components, while on a 50-Hz
system, it consists of a sum of the 25, 75, 125,
625-Hz components. If the Energy Algorithm detects a sudden, sustained
increase in one of these component energies, it reports this to the
Expert Arc Detector Algorithm, resets itself, and continues to monitor
for another sudden increase.
Randomness Algorithm
The Randomness Algorithm identifies another characteristic of these
faults, that of having energy magnitudes which vary considerably from
one half-cycle to the next, as shown in Figure 2. The Randomness
Algorithm monitors the same set of component energies as the Energy
Algorithm. based-Digital-Feeder-Monitor-with-High-Impedance-Fault-/' target='blank' class='doin' >However,, rather than checking for a sudden, sustained
increase in the value of the monitored component energy, it looks for a
sudden increase in a component
- 4 -
followed by highly erratic behavior. This type of highly random behavior
is indicative of many arcing faults. Just as with the Energy Algorithm,
if the Randomness Algorithm detects a suspicious event in one of its
monitored components, it reports this to the Expert Arc Detector
Algorithm, resets itself, and continues to monitor for another
suspicious event.
Expert Arc Detector Algorithm
The purpose of the Expert Arc Detector Algorithm is to assimilate the
outputs of the basic arc detection algorithms into one belief-in-arching
confidence level per phase. Note that there are actually 24 independent
basic arc detection algorithms, since both the Energy Algorithm and the
Randomness Algorithm are run for the even harmonics, odd harmonics, and
non-harmonics for each phase current and for the neutral. The
assimilation performed by the Expert Arc Detector Algorithm, then, is
accomplished by counting the number of belief-in-arcing indications
determined by any one of the twenty-four algorithms over a short period
of time. Also taken into account is the number of different basic
algorithms that indicate a belief in arcing. Various weights are
assigned to each of the parameters to reflect the significance of the
information in each parameter. These weights were derived from the
analysis of hours of data from over 300 staged faults and other events.
The Expert Arc Detector Algorithm's belief-in-arcing confidence level
for each phase increases as the number of basic algorithms that indicate
a belief in arcing increases. It also increases with increasing numbers
of indications from any one basic algorithm. These confidence level
increases occur because multiple, consecutive indications and multiple,
independent indications are more characteristic of the presence of
arcing than a single algorithm giving a single indication.
Load Event Detector Algorithm
The Load Event Detector Algorithm examines, on a per-phase basis, one
reading of RMS values per two-cycle interval for each phase current and
the neutral. It then sets flags for each phase current and for the
neutral based on the following events: (1) an overcurrent condition, (2)
a precipitous loss of load, (3) a high rate-of-change, (4) a significant
three-phase event, and (5) a breaker open condition. These flags are
examined by the Load Analysis Algorithm. Their states contribute to that
algorithm's differentiation between arcing downed conductors and arcing
intact conductors, and inhibit the Expert Arc Detector Algorithm from
indicating the need for an arcing alarm for a limited time following an
overcurrent or breaker open condition.
Load Analysis Algorithm
The purpose of the Load Analysis Algorithm is to differentiate between
arcing downed conductors and arcing intact conductors by looking for a
precipitous loss of load and/or an overcurrent disturbance at the
beginning of an arcing episode. A typical downed conductor pattern
recognized by the algorithm is shown in Figure 3. The presence of arcing
on the system is determined
- 6 -
based on the output of the E