THE DIAGNOSTIC TESTING OF HIGH VOLTAGE SILICON CARBIDE SURGE ARRESTERS
J. D. F. McDonald, M. Darveniza, T. K. Saha
Department of Computer Science and Electrical Engineering
The University of Queensland
Email: mcdonajd@csee.uq.edu.au, matt@csee.uq.edu.au, saha@csee.uq.edu.au
Abstract
Reliable operation of electrical overstress protection devices is critical to maintain proper function
of power transmission systems. This paper details the methodology employed and results obtained
in a diagnostic testing procedure used to determine the condition of a number of silicon carbide
surge arresters removed from service on high voltage networks. The arresters were all shown to be
in satisfactory condition, highlighting the comparative durability of high voltage arresters. The
investigation also showed that both partial discharge and radio interference voltage measurements
provide a more sensitive measure of arrester degradation than traditional diagnostic procedures.
Keywords: diagnostic testing, surge arrester, silicon carbide, partial discharge
1.
INTRODUCTION
A fundamental constraint on the reliability of an
electrical power transmission system is the
effectiveness of its protective network. The role of
the protective network is to safeguard system
components from the effects of electrical overstress.
It is therefore of great importance that effective
techniques are developed to accurately assess the
condition of the components used to provide this
protection.
The first truly effective overstress protection for high
voltage networks was provided by the silicon carbide
(SiC) surge arrester. Developed in the 1930’s, it was
used almost exclusively on both transmission and
distribution systems until the development of the
metal oxide surge arrester in the 1970’s. Since then
the silicon carbide arrester has been all but replaced on
distribution systems but it is estimated that there are
tens of thousand of these arresters still in service today
[1] on sub-transmission and transmission systems.
to fail explosively also presents a hazard to both
adjacent equipment and personnel.
The goal of this preliminary investigation then was the
development of a diagnostic testing procedure that
could determine the condition of high voltage (rated
33 kV and above) silicon carbide surge arresters. The
testing programme should be able to be completed
quickly, and preferably in field without need the to
remove the arrester from service.
To facilitate this several silicon carbide arresters
removed from service were subjected to an exhaustive
diagnostic testing programme. Although a number of
investigations of this nature have been performed on
distribution level arresters [1,2] the results of this
study are intended for high voltage arresters. From
the results obtained the effectiveness of the specific
testing procedures could be assessed and the most
effective methods selected for further development
and application.
2.
Although the rugged construction of high voltage
arresters makes them resistant to degradation, the
majority of silicon carbide surge arresters still
operating have been in service at least 10 years and in
some cases as long as 30 years [1] so its is likely that a
number are somewhat degraded. Darveniza et al [1]
suggested that as many as 75% of distribution level
silicon carbide surge arrester in service for over 13
years exhibit some form of degradation. Not only do
these degraded arresters present a threat to system
reliability, but the tendency of silicon carbide arresters
EXPERIMENTAL DESIGN
The basic construction of surge arresters governs their
response to diagnostic test procedures. A thorough
understanding of the construction and unique features
of high voltage silicon carbide surge arresters will
facilitate appropriate diagnostic test selection and
allow more effective interpretation of the results
obtained.
2.1
Arrester construction
Silicon carbide arresters consist of one or more silicon
carbide non-linear resistive blocks connected in series
with multiple spark gaps [3]. These components are
placed in a porcelain casing preventing moisture
ingress and providing strength to the entire structure.
Figure 1 Silicon Carbide Surge Arrester
The lightning impulse spark-over voltage test, power
frequency spark-over voltage test and power
frequency withstand test were used to test spark gap
performance when the arresters were subjected to
either high frequency impulses or sustained power
frequency overvoltages.
A non-standard current impulse test was included to
check the integrity of the silicon carbide blocks
themselves, with comparison of the waveforms
produced by the arresters under test of more use than
actual quantitative measurements.
The inclusion of both AC/DC leakage current
measurements served a twofold purpose. The DC
leakage current measurements, used to complement
that traditional 5kV insulation test, provided an
assessment of the condition of the gap grading
resistors. Together with the AC leakage current
measurements these tests also provided a gross
assessment of overall arrester condition.
High voltage gapped silicon carbide surge arresters are
usually constructed by connecting a number of low
voltage arresters in series, increasing the voltage
rating of the total arrester. The applied voltages
however are often unevenly distributed across the unit
arresters of a multi-unit arrester, with the line-end
arresters more severely stressed, leading to spark-over
at voltages lower than expected. This can be
prevented by placing large resistors in parallel with
the spark gaps. The flow of power frequency current
through these “gap grading” resistors will produce the
desired voltages across the spark gaps.
2.2
Test Selection
A testing programme was selected that would
adequately assess not only the condition of the whole
arresters but also the condition of their major
constituents including the:
•
spark gaps
•
non-linear resistive blocks
•
gap grading resistors
Rather than developing entirely new tests with
unproven performance and no comparative results, the
tests selected were chosen after a review of
appropriate literature.
The majority of testing
procedures used are detailed in [3], but references
such as [4], [5] also provided a number of alternative
testing techniques.
The testing regime completed in this investigation
consisted of the following tests.
The partial discharge and radio interference voltage
measurements required the arresters to be energized at
rated voltage and provided an assessment of arrester
condition in a standard operational situation.
Finally, a visual inspection of the arresters’ internal
components was completed to check for the presence
of degradation, which should have been detected
during the proceeding testing process.
2.3
Failure Criteria
As the results of this investigation were highly
dependent upon the standards used to assess
satisfactory arrester behaviour, three main criteria
were used to evaluate arrester performance. Arresters
were considered “failed” if their measured behaviour
did not:
a)
meet the required performance measures
outlined in the relevant industry standards [3]
b) satisfy the performance levels guaranteed in
manufacturers data [6]
c)
show consistency with test results obtained
on similar arresters, a technique used in [1]
When using non-standard testing procedures, this last
criterion was the only method available for classifying
arrester performance. Any arrester that could not meet
all these criteria when applicable was considered to
have “failed”.
3.
3.1
EXPERIMENTAL RESULTS
3.2
Test Samples
The test sample consisted of six XAA station class
arresters manufactured in 1978 by ASEA. Described
as self-supporting, the arresters had 110mm diameter
active parts and their general construction is shown in
figure 2.
Figure 2 XAA station class arrester
Impulse Testing
The function of an arrester is to limit the amount that
the voltage across the device it is protecting rises
when subjected to voltage overstress. Thus superior
arrester performance is characterised by consistent
impulse spark-over at lower impulse voltage levels.
Arresters A-C were tested in direct accordance with
clause 5.2.2 of [3]. In all cases the arresters performed
satisfactorily, sparking–over on each of five
consecutive applications of a 1.2/50µs voltage impulse
with prospective peak of 106kV.
Rather than adhering directly to standard procedure
when testing arresters D-F, experimental procedure
was modified slightly to determine the minimum
standard lightning impulse spark-over voltage. This
was defined as the lowest voltage at which the arrester
a)
sparked over on 5 consecutive impulses or
b) sparked over on 4 out 5 consecutive impulses then
sparked over on 10 consecutive impulses.
This measurement offers a more effective comparison
with the maximum standard lightning impulse sparkover voltage defined in [3] and the 100% guaranteed
impulse spark-over voltage level detailed in [6].
The total test sample consisted of
•
3 arresters rated at 36kV taken from an
electricity utility.
•
3 multi-unit arresters, each made up of 3
unit arresters with total rating of 121 kV,
supplied by a transmission company.
The investigation focused predominantly on these
arresters whose parameters are detailed in the
following table.
All impulse tests on arresters D-F were performed
with a grading ring attached to the top of the arrester
as required for normal operation. The grading ring
significantly affects the value of impulse spark-over
voltage and measurements taken without the ring in
place would be invalid.
Figure 3 Impulse Spark-over of arresters D-F
Lightning Impulse Sparkover of
121 kV Arresters
350
300
Table 1 Arrester Designations
Arrester
A
Rating
[kV]
36
B
36
C
36
D
121
E
121
F
121
Class
10 kA
50-60 Hz
10 kA
50-60 Hz
10 kA
50-60 Hz
10 kA
50 Hz
10 kA
50 Hz
10 kA
50 Hz
Date of
assembly
1978
250
200
150
100
1978
1978
1978
1978
50
0
Arrester D
Arrester E
Measured minimum impulse sparkover voltage
Guaranteed 100% impulse sparkover voltage
Maximum impulse sparkover voltage
1978
Arrester F
As can be seen in figure 3, the measured minimum
spark-over voltage for each arrester is lower than both
the level guaranteed by the manufacturer and that
required by the relevant standards [3], demonstrating
satisfactory arrester behaviour.
Table 2 AC/DC leakage current measurements
5 kV insulation test
Arrester
3.3
Power Frequency Spark-over Testing
Experimental procedure used in this test followed that
outlined in clause 5.12 of AS1307.1 [3] irrespective of
the test object used.
Figure 4 Power Frequency Spark-over Voltage
Power Frequency Sparkover
Voltage
A
B
C
D
E
F
DC leakage
current
[µA]
12.1
11.3
9.5
0.25
0.25
0.25
0.42
0.46
0.54
20
20
20.2
851
848
827
677
669
695
250
3.4.1
200
To determine whether the performance of an arrester
in this test was satisfactory a minimum acceptable
value of internal resistance was required. Although
figures as high as 10GΩ have been recommended in
the past, the figure of merit used in recent
investigations [1] was 2 GΩ, which suggests that
arresters A-C are all in satisfactory condition.
150
100
50
0
A
B
C
D
E
F
Arrester
Measured sparkover voltage
Guaranteed minimum sparkover voltage
Standard minimum sparkover voltage
Leakage Current Measurements
This test is a commonly used and often effective
measure of the integrity of an arrester’s internal
components. It is particularly useful for determining
the condition of arresters that do not contain
resistively graded gaps, as the only connection
between spark gap electrodes is insulation. A low
value of resistance would indicate significant internal
corrosion.
5 kV Insulation Test
The measured resistance of arresters D-F however was
much lower than expected, indicating either severe
internal corrosion, or that the gap-grading resistors
dominate the measure of DC resistance. Given the
consistency of measurements between the arresters, it
was felt the grading resistors were dominating the
arresters’ internal impedance.
This was later
confirmed through visual inspection.
3.4.2
Direct comparison was possible with measured values
and the minimum power frequency voltage specified
by the relevant standards and the performance level
guaranteed by the manufacturers. The measured
performance of each of the arresters exceeded both
these requirements; highlighting that under normal
operating conditions the arrester should remain an
open circuit. Thus these arresters all demonstrated
satisfactory experimental behaviour.
3.4
DC
resistance
[GΩ]
AC leakage
current at
rated
voltage
[µA]
AC Leakage current test
AC leakage current measurements are more
commonly used as a diagnostic tool for metal oxide
surge arresters where an accurate measure of the
resistive component of leakage current is required. As
it was expected that the total leakage current through a
silicon carbide surge arrester would be fairly small, in
the order of 600µA [2], the magnitude of the total
leakage current, both capacitive and resistive, was
measured. The experimental set-up used is shown in
figure 5.
Rather than testing all arresters at a common voltage
level, in a similar fashion to the 5 kV insulation test,
the AC leakage current was measured when the
arresters were energized at their respective rated
voltage levels.
Figure 5 AC leakage current measurement
Figure 6 Partial Discharge Measurements
Although the measured AC leakage current was larger
than 600µA, the consistency between measurements
from arresters of the same type would suggest that the
arresters were all in satisfactory condition.
The ERA Discharge detector superimposes the
measured discharges on an ellipse representing the
50Hz time base. From the two traces shown below, it
is possible to compare both the magnitude and
frequency of the discharges. In this case far more and
much larger discharges were produced by arrester A,
suggesting that it is more degraded than arrester C.
3.5
Current Impulse Test
This test assessed the integrity of the silicon carbide
blocks themselves. The arresters were subjected to
1.2/50µs voltage impulses with prospective peaks well
above the measured minimum lightning impulse
spark-over voltage and the waveforms of both the
current through and voltage across the arresters were
measured. A sudden drop in voltage coincident with a
sharp rise in current would indicate a block failure.
Again all arresters were found to perform
satisfactorily with no block flashovers detected.
3.6
Figure 7 Partial Discharges at rated voltage
Partial Discharge/Radio Interference Tests
Partial discharges within an insulating material are
both symptomatic of and an accelerator of degradation
of its internal condition. For the majority of their
service life silicon carbide surge arresters behave as
insulators making the partial discharge test an
applicable measure of arrester condition.
Partial discharges can be measured using a number of
different techniques. In this study however, direct
measurement of the current impulses produced or
traditional partial discharge testing was used along
with radio interference voltage (RIV) measurement,
which measures the noise produced by the discharges
in the radio frequency (150kHz – 30 MHz) range.
In this study the partial discharges were measured by
an ERA Discharge detector model 3, configured for
straight detection, as shown in figure 6. The device
under test is represented as Ca, while the discharge
detector is designated as Zm and Ck is a 2000pF
blocking capacitor. The experimental technique used
was similar to that in [4], including pre-stressing of
the arresters.
Radio interference voltage measurements were taken
in a similar manner, using a PO Engineering Dept
D1061 Radio Interference Meter. The meter was
placed in series with the blocking capacitor rather than
in series with the test object as specified in [5].
Circuit noise at high voltage operation prevented RIV
measurements from being taken from arresters D-F.
Table 3 Partial discharge / RIV measurements
Arrester
A
B
C
D
E
F
Discharge level at rated voltage
Partial Discharge
RIV Test
Test [pC]
[V x 10-6]
142
891
64
200
40
5
3.2
2.5
12.6
-
The magnitude of discharges measured when the
arresters were energized with rated voltage was used
as the measure of comparison because this is
indicative of normal arrester behaviour. Although it is
difficult to specify the exact level of discharges
demonstrating significant arrester degradation, the
comparative results provide a more sensitive measure
of arrester condition. For example, it would seem that
arrester A has suffered the most internal degradation,
and that the 121kV arresters were in better condition
than the 33kV arresters.
3.7
Visual Inspection
A complete visual inspection of internal arrester
components supported the satisfactory test results
obtained. The arresters appear to have remained well
sealed and all internal components are relatively free
from degradation. Arresters A and B are the only
arresters with noticeable internal degradation and that
is confined to rusting on the steel end caps of some
silicon carbide blocks as shown below. The presence
this degradation is consistent with the results obtained
from partial discharge tests and RIV tests.
Figure 8 Arrester component degradation
Although the majority of tests produced very
consistent results within groups of similar arresters,
the results of the partial discharge and radio
interference voltage measurements suggest that these
tests are far more sensitive than other more standard
diagnostic techniques. These tests detected internal
arrester degradation even before it appeared to affect
arrester performance. The results were confirmed by
visual inspection of the arresters that found
conspicuous degradation in only arresters A and B,
both of which displayed heightened partial discharge
and radio interference measurements.
It would therefore seem that the partial discharge and
radio interference voltage tests are sensitive and
accurate measures of the internal condition of high
voltage silicon carbide surge arresters. A true measure
of the performance of these tests however will be
obtained only when the next stage of the investigation,
in which further testing will be conducted using a
more extensive sample containing both failed and
satisfactory arresters, has been completed.
5.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the great
assistance and helpful advice provided by S. Wright in
the completion of the required experimental work.
6.
4.
[1]
M. Darveniza, D. R. Mercer, and R. M.
Watson, “An assessment of the reliability of
in-service gapped silicon-carbide distribution
surge arresters,” IEEE-Transactions-onPower-Delivery, vol. Vol.11, pp. p.1789-97,
1996.
[2]
J. H. Shaw and N. V. Holmgren, “A surgearrester tester developed for EPRI,”
Transmission and Distribution, vol. 37, pp.
50-2, 1985.
[3]
Australian Standard AS 1307.1 - 1986, Surge
Arresters (Diverters) Part 1 - Silicon Carbide
Type for A.C. Systems. Sydney, 1986.
[4]
Australian Standard AS 1018 - 1985, Partial
Discharge Measurements. Sydney, 1985.
[5]
Australian Standard AS 2650 - 1986, High
Voltage A.C. Switchgear and Control Gear Common Requirements. Sydney, 1986.
[6]
“TYPE XAA Surge Arresters, Catalogue
FMK No.7: 10 E (Cat E 594),” . Sweden:
ASEA, 1965.
CONCLUSIONS
The results of this investigation show that all arresters
appear to be in satisfactory condition. In all standard
tests, the performance of the arresters exceeded that
required by the relevant Australian standards, the
manufacturers performance standards where they
existed, and in all but the partial discharge and radio
interference voltage tests, results obtained were
consistent within the groups of common arresters.
Even the results obtained in the final two tests showed
that arrester performance did not exceed the standard
criteria for acceptable discharge magnitudes defined in
[4, 5].
The limited internal degradation after 20 years of
service suggest that high voltage arresters may be far
more durable than originally expected, with the
majority of arresters still able to perform acceptably in
operation.
REFERENCES