International Conference on Advances in Processing,

Testing and Application of Dielectric Materials

TEMPERATURE MONITORING OF METAL OXIDE SURGE ARRESTERS

Krystian Leonard CHRZAN*, Volker HINRICHSEN**

*Wrocław University of Technology, Poland
**Siemens AG, EV HBA 2, Berlin, Germany

Varistor overtemperature above the ambient is an important state parameter of metal oxide surge arresters. The temperature
monitoring using passive SAW sensors enables realisation of a surge counter function, an energy monitor, monitoring of
electrical ageing and pollution stress. For temperature measurements during pollution tests of metal oxide arresters the not so
advanced, TINY TALK sensors could be used. This method of temperature measurement was also applied in the field for
temperature control of arresters tested at the pollution station near Głogów, Poland. The preliminary results during the first year
of monitoring are presented and compared with results of similar measurements conducted in Germany close to the seacoast.

Key words: diagnostics, degradation, ageing, resistive current

1. THE ARRESTER TEMPERATURE - A limited operating time of a battery supply [1]. The
UNIVERSAL MONITORING PARAMETER measurement system comprising SAW sensors is
expensive. Therefore the TINYTALK sensors seems to
All parameters worth to be monitored on surge be a very cheap alternative in the lab and also in the
arresters – energy absorption due to impulse or field.
temporary overvoltages, power loss increase by
electrical ageing or moisture ingress – will finally affect a) Surge counter
the temperature of the MO column and therefore can be Criterion: sudden temperature
checked on the basis of a temperature measurement. increase above a certain
Different from other service parameters, e.g. the 3rd threshold, corresponding to e.g.
10% of rated energy absorption
harmonics content of the leakage current, which is often capability
used to detect possible electrical ageing, the temperature
is not only an indirect measure of the actual service b) Energy monitor
condition but the decisive operating parameter itself. Criterion: amplitude of a
The temperature of the MO column is the resulting sudden temperature increase
effect of all influences together. At any time,
independent from history and different origins the
remaining energy absorption capability or the decision c) Monitoring of electrical
on thermal stability after operation of the arrester are ageing
Criterion: average continuous
given by the actual MO temperature and its change with
overtemperature exceeds the
time. normal range
By evaluating the overtemperatures against ambient
temperature, following basic functions can be realised,
representing a complete overall monitoring concept: surge Fig. 1. Functions realised by temperature measurement
counter, energy monitor, monitoring of electrical ageing,
monitoring of pollution performance (fig.1)[1]. 2. TINYTALK SENSORS
Summarised, measurement of the arrester
temperature facilitates an overall arrester monitoring, The sensors TINYTAG, TINYVIEW and TINYTALK
comprising function, which have not been realised yet. can be widely used for measurement of different
System requirements such as measuring range, quantities like: humidity, temperature, voltage, current or
bandwidth or time resolution are low, and electrical vibrations. The measuring data (1800 or 3600 samples)
interference immunity is high. The main obstacle to its can be stored in an EPROM memory and transmitted to
realisation has been the difficulty to measure the computer via a serial link. During the measurement no
temperatures on high potential. For the first time this hardware connection to the environment is necessary
was offered by wireless passive surface acoustic wave while voltage is applied. Any influence of the test set up
(SAW) temperature sensors, which do not impose on the voltage distribution is thus avoided. The sampling
restrictions like galvanic or optical connections or rate can be changed from 0.5 second to 4.8 hours (the
 Krystian Leonard CHRZAN, Volker HINRICHSEN

whole measuring time from 15 minutes to 1 year). The was energised for a time lasting long enough to reach
data are stored even in the case of discharged battery steady state temperature distributions.
(typical lifetime of 4 years).
The sensor is mounted on printed board with the
dimensions of 34×52 mm. Logging is started and the data
retrieved (of-loaded) by means of the management
software (OTML) which is run on a WINDOWS based
host computer. The reading finishes after the memory is
full or after sampling a chosen data number. The logger
may be programmed to delay the start of the logging cycle
by up to 45 days.
TINYTALK loggers are designed to operate over
a temperature range of –40 °C to +75 °C (+125 °C) with
the accuracy about 0,5 °C. For the application in surge
arresters the sensor is integrated in aluminium tube of 60
mm diameter and 30 mm height which is usually used for
length adjusting of the metal oxide resistor stack. The
arrester current flows along the aluminium wall of the
tube (fig. 2). The thermal time-constant is around 150 Fig. 3. Varistor column with inserted temperature logger
seconds (about 30 times shorter than thermal time-
constant of the surge arrester). 4
Half ring φ 1200
Full ring φ 1500

3
Height ( m )

Fig. 2. TINYTALK temperature sensor integrated

in a Aluminium tube
0
3. TEMPERATURE DISTRIBUTION ALONG THE 0 2 4 6 8 10 12
ARRESTER Overtemperature ( K )
The temperature distribution gives more reliable Fig. 4. Measured temperature distribution along the arrester
information about the grading system than the (Ur = 336 kV) at applied maximum line-to-earth voltage
measurement of voltage distribution. Here, however, the of the system (Um = 420 kV) [2]
temperature distribution was experimentally verified
within the complete 400 kV arrester. For this purpose Figure 4 shows in comparison the results for the
seven temperature TINYTALK loggers were arranged arrester equipped with the standard grading ring of 1.2 m
within the MO resistor columns, three of them in the diameter and alternatively with the half ring of 1.5 m.
bottom unit and the other four in the top unit. diameter. The temperature distributions are basically in
For the measurement the arrester was erected directly agreement with the calculated voltage distribution. It can
on the floor (not in a height of the 2.3 m as in the be seen, however, that for the half ring the
disconnector), which is the worst case condition with overtemperature continuously increases with arrester
respect to operating temperature stress. In real service the height. The uniformity of the voltage distribution is partly
uniformity will be slightly better and the temperature forced by self grading due to the resistive effect of the
values somewhat lower. The applied voltage was the MO blocks, which starts for voltages above about 1.2
maximum line to earth voltage of the system. The arrester times their continuous operating voltage. Looking at the

184 Przegląd Elektrotechniczny – Special Edition

 TEMPERATURE MONITORING OF METAL OXIDE SURGE ARRESTERS

figures in more detail reveals, however, that the maximum

overtemperature stress of 9.5 K in case of the full ring of
1.2 m diameter is exceeded by just 1,5 K when using the 40,0

half ring of 1.5 m diameter. This difference is negligible,

Temperature °C
and as a final result the modifications of the grading ring 30,0

are approved by this investigation. The arrester can be

used without any restrictions. 20,0

4. TEMPERATURE MEASUREMENT OF MO
10,0
VARISTORS IN THE FIELD
2Fr 3Sa 4Su 5Mo 6Tu 7We 8Th 9Fr
Before application in the field the sensors were tested Time (starting 00-06-01)

to check their reliability under severe pollution conditions.

Fig. 6. Temperature inside the two unit 96 kV rated surge
For this purpose the arrester was non-uniformly polluted
arrester at the pollution test station Glogów in June 2000. Lower
(as described in [3]) and the internal discharges were line - the ambient temperature, middle line - the temperature
sustained for a period of 6 hours. Then the voltage was inside the bottom unit, upper line - the temperature inside the
switched off and the sensor resistivity to aggressive upper unit
products of partial discharges in a closed arrester was
tested for next 24 hours. The varistor temperature in the The top and middle trace on fig. 6 shows the
test exceeded 117 °C (fig. 5). But the sensors were not temperature in the top and bottom unit of 96 kV rated
damaged. The calibration procedure revealed their arrester. The maximum temperature is reached at noon,
excellent accuracy of 0,5 °C. Additionally, the test has the minimum in the night. The lower trace represents the
shown that the sensor and the battery can operate in the reference temperature in the arrester without voltage. The
temperature over 100 °C. It is more than the temperature temperature in the top unit is about 4 K higher than in the
limit of 75 °C given in technical data. bottom unit and about 8 K higher than the reference
temperature. On June 7. 2000 the voltage was switched
120 off. In the night three temperatures are equal. But at noon
the reference temperature is about 2 K lower than the
Temperature ( oC )

100 temperature in the both arrester units. This discrepancy is

caused by the different insolation of two-unit arrester and
80 the reference arrester (the reference arrester is situated in a
60 less insolated position).
Usually the temperature of arresters is only the
40 function of the ambient temperature. But on 30 March
2001 the apparent non-regularity was noticed. The
20 overtemperatures in two-unit arrester and in one unit
0 250
300 750
400 1750
500 arrester measured in March 2001 are shown in fig. 7.
Time ( minutes )

Fig. 5. Temperature increase during internal arcing test

(K)

10
The temperature measurement in the field was carried
out at the pollution station in the copper plant Glogów in
Overtemperature

Poland. For this purpose two TINYTALK loggers were

arranged within MO resistor column of two-unit arrester 5
with a rated voltage 96 kV with porcelain housing (one in
the top unit and one in the bottom unit). One logger was
mounted into a one-unit porcelain arrester with rated
voltage of 96 kV. The maximum continuous operating 0 Time ( days )
voltage of 77 kV was applied to the both arresters. The
fourth logger was installed into one-unit 96 kV rated 5 10 15 20 25
arrester that was not connected to the high potential. This
Fig. 7. The overtemperatures inside the two unit arrester and
sensor measures the reference temperature. The mean
inside the one unit arrester during March 2001 Bottom line -
value of the reference temperature equals the mean bottom unit, upper line - upper unit, circles at the middle -
ambient temperature. one unit arrester

Przegląd Elektrotechniczny – Special Edition 185

 Krystian Leonard CHRZAN, Volker HINRICHSEN

The temperature inside one unit arrester is mostly CONCLUSIONS

between the temperatures of both small units. Only on 30
March the temperature of one-unit arrester was higher 1. The TINYTALK sensors are resistant to internal
than in the upper unit of two-unit arrester. This situation is discharges and can therefore be used to measure the
showed also in the fig. 8. temperature inside MO arresters in the field.
2. The simple and cheap system has shown very good
performance so far and can be used for other
applications.
(K)

10
3. The temperature inside the arrester at the pollution
test station (pollution level III - heavy) changed
Overtemperature

mainly due to the changes of ambient temperature.

The recorded maximum temperature was 46 °C high.
5
The influence of pollution is rather small. Only one
times it was recorded the unusual temperature
increase probably due to formation of stable dry band
on the housing.
0 Time ( days )
26 27 28 29
Fig. 8. The overtemperatures inside the two unit arrester LITERATURE
and inside the one unit arrester at the end of March 2001.
Bottom line - bottom unit, upper line - upper unit, circles at the [1] Hinrichsen V., Scholl G., Schubert M., Ostertag T., On line
middle - one unit arrester monitoring of high-voltage metal-oxide surge arresters by
wireless passive surface acoustic wave (SAW) temperature
The reason of such different behaviour can be the sensors. Int. Symposium on HV Eng. London 1999, paper
formation of stable dry band on the housing of single unit 2.238.
arrester. In this case the temperature of varistor column [2] Hinrichsen V., Goehler R., Lipken H., Breilmann W.,
opposite to the dry band is higher than at the lower part of Economical overvoltage protection by metal-oxide surge
arresters integrated in high-voltage AIS disconnectors.
varistor column [4]. The phenomena of dry bands
CIGRE Session 2000, paper 33-104.
formation is a matter of statistics. Additionally, only one [3] Chrzan K., Koehler W., Feser K., Internal arcing test on
stable dry band on the housing can be formed under high voltage arresters. Int. Symposium on HV Engineering
specific weather conditions (the long time of high ISH, Graz 1995, paper 3220.
humidity, fog, drizzles). Therefore this phenomena was [4] Feser K., Koehler W., Qiu D., Chrzan K., Behaviour of zinc
observed only one time during 8 months lasting oxide surge arresters under pollution. IEEE Trans. on Power
registrations. Delivery, April 1991, pp. 688-694.

186 Przegląd Elektrotechniczny – Special Edition