IX International Symposium on
Lightning Protection
26th-30th November 2007 – Foz do Iguaçu, Brazil
IN-SERVICE TESTING AND DIAGNOSIS OF
GAPLESS METAL OXIDE SURGE ARRESTERS
Vegard Larsen
Kjetil Lien
Doble TransiNor As
Doble TransiNor As
Vegard.larsen@doble.no
Kjetil.lien@doble.no
Sorgenfriveien 9, N-7037 Trondheim, Norway
Abstract - This paper describes testing strategy and risk
assessment for on-line in-service diagnosis of gapless metal
oxide surge arresters (MOSA) mounted on insulated base.
Different case studies are presented to show field experience
with procedures for testing, diagnosis and risk assessment
based on different levels of available information of the
surge arrester and its past history.
Risk assessment is based on comparison with baseline
readings performed when the arresters were commissioned,
on trend analysis over time and/or on comparison with
maximum recommended resistive leakage currents specified
by arrester manufacturers. This implies that the system
gives the necessary information about the arrester
conditions in order to utilize their lifetime and to take them
out of service before they fail.
1 INTRODUCTION
The surge arrester is known to be a rather in-expensive
component in the power system. The destiny of the surge
arrester is normally to be specified, purchased, installed
and forgotten. For this reason the surge arrester is often
overlooked when condition assessment of a substation is
considered. The fact is that the arrester is one of the key
components for protection of crucial high-voltage
apparatus like power transformers, cables etc. against
overvoltages. In cases of arrester failures, increased safety
risk for the maintenance staff and severe damages to
neighbouring apparatus may be the consequences. Aged
surge arresters may offer reduced overvoltage protection
especially for (old) equipment with degradation of the
insulation system. In addition, experience has shown that
is far more economic to replace the surge arrester before
it fails than to deal with unplanned outage and
interruption of supply.
Failure statistics from the insurers in the utility market
(Marsh and HSB) show a substantial increase of
significant transformer events over the last 15 years. HSB
reports that about 45% of the failures are due to electrical
disturbance out of them 16% caused by lightning.[4] This
increasing number of failures raise the question whether
the apparatous has an adequate protection.
2 IN-SERVICE DEGRADATION OF MOSA
In service the surge arresters will be exposed to a
combination of stress originating both from the network
and the local environment.
These stress, separately or together in different
combinations, may cause ageing or damage to the
MOSA-blocks. The main categories of degradations of
the surge arresters are:
•
•
Degradation of the insulation properties
Degradation of the protective characteristics
There is a whole range of mechanisms that can cause
degradation of MOSA or in worst case failure of the
MOSA:
•
•
•
•
•
Sealing defects leading to ingress of moisture
Discharges due to surface contamination
Overloading due to temporary and transient
overvoltages
Long term ageing during normal service voltage, e.g.
incorrect arrester specification corresponding to
actual system voltage and overvoltage stress
Internal partial discharge
Overloading typically occur after fault situations with
high temporary overvoltages in the network. If the rated
voltage of the arrester has been selected too low, this will
further increase the risk that the arrester could be
overloaded even for the temporary overvoltage that it
should be dimensioned to withstand.
One consequence of the degradation of the protective
characteristic of the MOSA is an increase with time of the
resistive component of the continuous leakage current
flowing through the arrester will increase with time.
Increase in the resistive leakage current will cause an
increase in the power losses and hence increased
temperature of the ZnO-blocks. The resistive leakage
current may, at a certain instant, exceed a critical limit
where the accumulated energy in the ZnO-blocks exceeds
the energy capability of the arrester (the energy that can
be dissipated to the surroundings). The arrester will then
get thermally unstable (“thermal runaway”) and fail.
3 FAILURE MODES OF SURGE ARRESTERS
An arrester failure may appear in different ways:
•
•
•
An arrester with porcelain housing may in worst case
explode and cause severe damages to the
surroundings. Such a failed arrester is shown in
Figure 1. In case of arresters with polymer housing,
the housing may burst open, but the risk for objects
being scattered is more limited.
The arrester can be causing an earth fault due to
internal flashovers etc. Such arresters can be difficult
to locate.
Aged or overloaded arresters may show reduced
protection against overvoltages, i.e. during severe
transient overvoltages, for instance due to multiple
lightning strokes or high-energy temporary
overvoltages, the arrester can fail before it actually
has suppressed the overvoltages. Thus, the apparatus
that the arrester is set to protect may be subject to
overvoltages that can cause damage to it.
Off-line measurements provide a testing environment
with good control of parameters that affect a reliable and
repeatable measurement. This approach requires to
deenergize the arrester and either use a separate/portable
voltage source or take the arrester to a laboratory. Online measurements are usually done on a temporary basis
using portable instruments or permanent installed devices.
Solutions for on-line continuous measurements are also
available. On-line methods have the advantage of
providing data to condition assessment without taking the
arrester out of service. Off-line tests are employed as
well to retest surge arresters that are taken out of service.
This combined approach could be used to verify
conclusions from on line measurements in cases of doubts
[8].
A detailed overview of the different applied
methods/indicators can be found in Ref [1, 2]. The most
used methods in service are:
• Visual inspection
• Surge counters
• Temperature measurements - thermo vision
• Leakage current measurements
Visual inspection is a common and valuable approach for
locating external abnormalities on the surge arrester.
Experienced service crew can detect external
deterioration of the end fitting seal areas, damage of the
arrester housing, surface contamination etc. However, this
method gives little or no information about the internal
life of the arrester and should be combined with other
methods for obtaining a reliable and more complete
condition assessment of the surge arrester.
Surge counter with/without mA- meters for
measurements of the total leakage current are frequently
installed on MOSA. These indicators are of no practical
use for diagnosis of condition of the arresters [1].
Fig.1 - MOSA with porcelain housing that failed
catastrophically in service.
4 METHODS FOR MONITORING MOSA
DEGRADATION
Several different methods/indicators are used by the
power industry today for in-service monitoring, diagnosis
and assessment of the MOSA. These methods vary
greatly both in handling complexity and the level of
information provided.
The two main approaches are on-line measurements and
off-line measurements respectively.
Temperature measurements with infrared thermo vision
are a very frequent used multi purpose maintenance
technique. This technique is employed for surge arresters
as well. Experience shows that thermal imaging can be
used to track surge arrester degradation. [2] The
measurements are considered only to be indicative with
regards to the condition of the arrester because of the
sensitivity to detect increased block temperature on the
housing surface of the arrester.
Leakage current measurements are the most used
diagnostic method for assessing the condition of MOSA.
A range of different on-line and off-line methods for
measurements of leakage current are frequently in use. In
the field this current is normally measured only at the
earthed end of the arrester. The arrester must be equipped
with insulated base and earth leads separated from earth
potential to measure the leakage current. For on site in-
service measurement the method with indirect
determination of the resistive leakage current component
by means of third harmonic analysis with compensation
for harmonics in the voltage (THRC) is providing the best
available information quality with respect to diagnostic
efficiency [1,3].
5 ELECTRIC PROPERTIES OF MOSA
DETERMINATION OF THE THRC
5.1 Current-voltage characteristic
In normal service the arrester is carrying a continuous,
but small leakage current, typically in the range of 0.2-3
mA. The leakage current is dominated by a capacitive
current component, while the resistive component may be
in the range of 5-20 % of the capacitive component.
Furthermore, the resistive component is temperature and
voltage dependent as seen from the typical currentvoltage characteristic in Figure 2 [1]. Thus, the ZnOelements of the MOSA can be represented by the electric
equivalent circuit shown in Figure 3, where the equivalent
resistance is nonlinear. The typical operating voltage U
(phase-to-ground) for a MOSA is in the range of 50-80%
of its rated voltage Ur*. The definitions may vary
depending on ANSI/IEEE C62.11 or IEC 99-4 is used.
1,2
The current-voltage characteristic shown in Figure 2 is
representative for a MOSA when it is stressed by a pure,
sinusoidal voltage (fundamental frequency component
only). The total leakage current flowing through the ZnOelements can be divided into different components:
• Fundamental capacitive leakage current component
• Fundamental resistive leakage current component
• 3rd harmonic resistive current component due to the
nonlinear resistance of the ZnO-elements said to be
generated by the arrester itself.
The resistive components (1st and 3rd) will at a specific
voltage and temperature reflect the operating point at the
current-voltage characteristic of the arrester and therefore
change with it due to ageing. Both these components can
be used as a measure of the arrester condition. For field
measurements in three-phase configurations, however, the
only practical solution is to determine the 3rd harmonic
component of the resistive current. The leakage current
for the same arrester can vary within a wide range due to:
• Harmonic content of the system voltage
• Actual temperature of ZnO elements caused by both
ambient condition and discharges
• Operating voltage
5.3 Effects of harmonic in the operating voltage
1,0
U / Ur
5.2 Leakage current
0,8
0,6
AC resistive, +20 °C
0,4
AC resistive, +40 °C
0,2
0,0
0,01
AC capacitive
0,10
1,00
10,00
100,00
Current - mA
Fig.2 - Typical current voltage characteristic for a
MOSA.
It
Ic
0.2-3 mA
Ir
U
10-600µA
Fig. 3 - Equivalent electric circuit for a MOSA.
Presence of harmonics in the operating voltage can
generate a 3rd harmonic capacitive component in addition
to the 3rd harmonic resistive component. These two
components can not be separated if only the total 3rd
harmonic leakage current is measured.
This 3rd harmonic capacitive leakage current component
may be of the same size or higher than the 3rd harmonic
resistive component generated by the arrester itself. The
evaluation error in this case may be large. For instance, if
the third harmonic content in the voltage is 0.5% or even
1%, the evaluation errors in the third harmonic
component will be in the ranges of ±50 % and ±100%
respectively [1]. Furthermore, since the harmonic content
varies with the load and thereby with time, it will not be
possible to tell if an apparent increase in the resistive
leakage current is due to true ageing/increase in the
resistive leakage current or simply due to varying
harmonic content in the operating voltage, which is of no
relevance. Measurements in the transmission grid (300420 kV) have shown that the 3rd harmonic content
typically is in the range of 0.2-1%. [5].
A method for compensation (THRC) of the effect of
harmonics in the operating voltage is widely used since
many years [3].
5.4 Effects of actual temperature and operating
voltage on the resistive leakage current
The influence of the block temperature and the operating
voltage can be significant. For this reason it is
recommended to measure both the operating voltage and
the ambient temperature. Ambient temperature
measurements can be used to estimate the block
temperature by keeping in mind that the time constant for
temperature changes of the blocks are in the range of a
few hours. By doing so, the measured values of the
resistive leakage current can be recalculated and referred
to the so-called standard reference conditions, i.e. to an
ambient temperature of 20oC and an operating voltage 0.7
times the rate voltage. In this way, measurements
performed at different temperatures and/or operating
voltages can be compared directly.
Table 1 below illustrates variations of the leakage current
with temperature and operating voltage for the same
condition of specific MOSA for 420 kV systems. For
instance, if two measurements have been performed at
ambient temperatures of 0oC and 40oC and the same
voltage respectively, the actual measured values may
deviate by more than 100% relative to each other, even
o
though the normalized values referred to 20 C should be
the same as long as the arrester condition is unchanged.
Table 1 : Influence of temperature and operating voltage
on resistive leakage current.
Temperature [oC]
0
0
0
20
20
20
40
40
40
Operating
voltage [kV]
380 400 420 380 400 420 380 400 420
Measured
resistive leakage
current [µA]
31
39
47
47
48
70
67
82
99
Measured value
normalized to
o
20 C and
U/Ur=0.7 [µA]
46
46
46
46
46
46
46
46
46
6 RISK ASSESSMENT AND TESTING STRATEGY
For metal oxide surge arresters, the best practice of risk
assessment is based on the trend and level of the resistive
leakage current at standard reference conditions. If the
resistive leakage current exceeds a certain threshold
value, use the following steps in the final
evaluation/judgement:
I.)
If the resistive leakage current is unrealistically
high, i.e. in the mA-range and many times higher
than MOSAs of the same type, check that the
arrester base and the earth lead are properly
insulated from the arrester pedestal. If the arrester
base is non-insulated, circulating currents will be
induced in the earth system and cause incorrect
measurement of the leakage current.
II.)
III.)
IV.)
Consider to re-test the MOSA in one or two days
to confirm the high reading. If the high reading is
confirmed, proceed with step III or IV. The reason
for re-testing the MOSA is: The MOSA could
have been subject to a transient overvoltage
causing a temporary higher current for several
hours due to the transient energy absorbed.
Monitor the MOSA continuously to follow the
development in the resistive leakage current. If the
resistive leakage current increases from the already
advanced level, proceed with step IV.
Contact the arrester manufacturer and consider
replacing the arrester due to the high resistive
leakage current.
The threshold value for the resistive leakage current will
vary from arrester type to arrester type. Threshold values
may be established in different ways:
• Some manufacturers provide data as so-called
“maximum recommended levels” for the resistive
leakage current for each arrester type. When
maximum recommended values are given, the
corrected values for resistive leakage current can be
compared directly to the maximum recommended
levels. Maximum recommended levels may be in the
range of 100-500 µA depending on type.
• If maximum recommended resistive leakage current
values are not available from manufactures, risk
assessment or threshold values for the arrester type
can be established based on experience as follows:
o Measure the resistive leakage current just after
commissioning. Use this as a baseline reading for
the arrester. If the leakage current later increases
by a factor larger than 3-4 times the baseline value,
this indicates severe ageing. Go to step II) to IV).
o Make an individual comparison of all three
arresters in a three-phase configuration of the same
type. If one arrester shows consistently and
significantly higher levels than the other arresters,
this might indicates aging of this arrester.
o Compare the resistive leakage currents in all
arresters of the same type in the grid: First, if one
or a few arresters show significantly higher levels
than the other arresters of the same type, this may
indicate ageing and thereby requires closer followup. Second, if a number of arresters show low
values at the same level, these may be used as
good/acceptable levels for this arrester type. Third,
if one or some arresters have been operating in
service for just a couple of years, the measured
values are expected to be close to baseline
readings for the arresters.
In general, it is recommended to conduct testing of the
MOSAs after special fault situations and after periods
with rough climatic/pollution conditions in the grid.
III.)
IV.)
V.)
VI.)
7 CASE STUDIES
Below are four case studies presented to show risk
assessment for surge arresters based on THRC leakage
current measurements. Further experiences can be found
in references [6, 7, 8, 9].
7.1 Testing of 420 kV MOSAs at transmission utility
Single measurements were performed for 24 MOSAs of
three different brands at a transmission utility. The results
are shown in Figures 4 to 6. The following is concluded:
Figure 4: Seven arresters showed low resistive leakage
currents, i.e. around 20% of the maximum recommended
level. The condition of these arresters is good. One
arrester showed approximately 90%, i.e. several times the
values of the two neighboring phases as well as the main
arrester population. This unit should be monitored closely
to check for further increase of current, either with
frequent measurements or by continuous monitoring. The
four remaining arresters showed values of 45%, 50%,
60% and 70%. Especially the unit showing 70% could be
tested more frequently, for instance every six months.
Figure 5: Four out of six units showed readings around
20%, i.e. the condition is good. The two last arresters
showed readings around 50-55%, i.e. the condition is
satisfactory. New measurements should be performed in
1-2 years, depending on the age of the arresters.
Figure 6: One arrester showed a resistive leakage current
significantly higher than the 5 other arresters. No
maximum recommended level was available for this type
of arresters, but by looking at the 5 mentioned units, it
seems reasonable to assume that the resistive leakage
current should not exceed 700 µA (ref. the
Resistive leakage current in percent
of max. recommended (100%)
II.)
Classify all arresters (name of substation, bay/line
and phase, nameplate data (manufacturer, type
designation, year/date of commissioning etc.),
historical data/failure rates, importance etc.)
Establish threshold levels/maximum recommended
levels for the resistive leakage current for each
arrester type.
Define action limits (good condition, re-test/monitor
continuously, replace)
Define measurement regularity (normal, frequent,
monitor continuously, after special fault situations)
Define verification actions after replacement
(laboratory test, dissection/inspection).
Evaluate measurements, action limits, regularity of
measurements and verification tests to possibly
improve the testing strategy.
120
100
80
Bay 1
Bay 2
60
Bay 3
Bay 4
40
20
0
1
2
3
4
5
6
7
8
9
10
11 12
Arrester number
Fig. 4 - Twelve arresters of type A in two substations.
Operating voltage: 415 kV, ambient temperture: 18 oC.
Resistive leakage
current in percent of
max. recommended
I.)
recommendation in Section 6). If the insulated base and
the insulation of the arrester earth wire are checked and
found satisfactory and there is no indication of temporary
heating due to transients, the arrester should be replaced
as soon as possible. The five other arresters are in
good/satisfactory condition.
120
100
80
Bay 7
60
Bay 8
40
20
0
1
2
3
4
5
6
Arrester num ber
Fig. 5 – Six arresters of type B in one substation.
Operating voltage: 415 kV, ambient temperature: 27 oC.
Resistive leakage current
(uA)
The following testing strategy is recommended for
MOSAs but can be modified based on local experience:
3000
2500
2000
Bay 5
1500
Bay 6
1000
500
0
1
2
3
4
5
6
Arrester num ber
Fig. 6 - Six arresters of type C in the same substation.
Operating voltage: 410 kV, ambient temperature: 18 oC.
7.2 Testing of 145 kV MOSAs at chemical factory
The factory had six 145 kV MOSAs installed at a
switching station. All six arresters were of the same make
and type and commissioned in 1984, i.e. probably first
generation MOSAs. The factory is located in a costal area
and the arresters may therefore be exposed to external
pollution, which might cause accelerated ageing of the
ZnO-blocks. The factory owner had no information about
the condition of the MOSA, except that the surge
counters had not been operating since 1989 and showed
just a few counts. The factory owner still had concerns
about the condition of the arresters since a failure with an
outage would cause high production losses.
Measurements of the resistive leakage current were
performed in 2002. The measurements showed the
following results in per cent referred to the maximum
recommended leakage current:
• 2 units showed around 130%.
• 3 units showed around 90-95%.
• 1 unit showed 70%.
Two arresters showed resistive leakage currents
significantly above the maximum recommended level,
while three other arresters showed values close to this.
Due to the potential consequences in case of a failure, the
factory owner decided to replace all six arresters.
In 2004, the six replacement units (same make, but
different type from the original MOSAs) were tested after
two years in service. All six units showed resistive
leakage currents in the size of 35-40% of the maximum
recommended level, i.e. good condition. New
measurements are scheduled for the autumn of 2007.
7.3 Testing of remaining 300 kV MOSAs after failure
A transmission utility experienced a catastrophic
failure/explosion of a 300 kV MOSA after approximately
9 years in service. The two remaining arresters were then
tested with the following results:
• One unit showed 545%
• The second unit showed 60%.
Hence one of the remaining arresters in service was
severely aged and was immediately taken out of service
to prevent a second arrester failure. This arrester was sent
for laboratory check. The reason for the aging appeared to
be the coating of the ZnO-blocks, which caused internal
partial discharges and a partly “conductive” surface.
7.4 Testing of 110 kV MOSAs
Fig. 7 - Measurements on 18 surge arresters.
During early 2007, measurements were performed on 18
MOSAs of the same type in one 110 kV substation. Two
of the arresters had significantly higher readings (230%
and 400% respectively) than all other as shown in Figure
7. The utility contacted the manufacturer, which took the
arresters out of service for laboratory testing. This test
showed ingress of moisture that caused internal heating
and increase of the resistive leakage current.
8 CONCLUSIONS
Leakage current measurements based on THRC have
proven to be a reliable and efficient system for assessing
the condition of gapless metal oxide surge arresters in
service in accordance with the recommendation from IEC
[1]. By implementing a testing strategy for the arresters in
the grid, it will on one hand be possible to utilize the
lifetime of the arrester and on the other hand to replace
bad or aged arresters before they fail. This will contribute
to increase the reliability in the power supply, reduce the
failure and outage costs and increase the safety for the
personnel.
9 REFERENCES
[1] IEC International Standard 60099-5 “Surge arresters - Part
5: Selection and application recommendations”, Edition 1.1,
2000-03.
[2] “Techniques for Evaluating Substation Surge Arresters: A
Field Guide for Technique Selection”, EPRI, Palto Alto CA:
2002.1001784.
[3] Lundquist, J. et al: “New method for measurement of the
resistive leakage currents of metal-oxide surge arresters in
service”. IEEE Transactions on Power Delivery, Vol. 5, No.
4, November 1990.
[4] Bartley,W: “An analysis of Transformer failures“
http://www.hsb.com/thelocomotive/story
[5] Schei, A: ”Content of 3rd harmonic voltage in transmission
and distribution systems. Error in arrester leakage current
measurements”, presented Session panel 1 CIGRE, Paris
2000 together with paper P1-05.
[6] Schei, A. et al: “Resistive leakage current measurements on
metal oxide surge arresters in service – measuring
equipment and results from measurements in 145 kV and
300 kV stations”. CIGRÉ symposium, Berlin, 1993, paper
no. 140-01.
[7] Leemans, P. and Moulaert, G. G.: “Experience with
leakage-current testing of 380 kv MOV surge arresters in
the field, utilizing an LCM portable instrument”. Doble
Client Conference, 1994.
[8] Tyagi, R. and K., Sodha, N. S., Jain, S. M.: “Condition
monitoring of surge arresters through third harmonic
resistive leakage current measurement.” Doble Client
Conference, 2001.
[9] Akbar, M. and Ahmad M.: “Failure study of metal oxide
surge arrester”, Electric Power Systems Research 50, 1999,
pp 79 -82.