THE DURABILITY AND PERFORMANCE OF POLYMER HOUSED METAL OXIDE SURGE ARRESTERS
UNDER IMPULSE CURRENT STRESSES
K Lahti, K Kannus and K Nousiainen
Tampere University of Technology, Finland
Unexpectedly, many failures of modern MOA types
have occurred in some laboratory tests where arresters
are stressed with either multipulses or series of impulse
current stresses. In this paper the effects of internal
humidity content on the internal flashover failure
tendency were studied. The test results are discussed
and compared to the results of a dry impulse current
stress series conducted earlier on arresters of the same
types at Tampere University of Technology
Test procedures and failure cases
A total of eight different types of commercially
available, new, distribution class, gapless polymer
housed metal oxide surge arresters were used in the
tests. In addition to this two older specimens, previously
used in networks, were also tested. The arresters were
from seven manufacturers from both Europe and North
America. The total number of test specimens was 20.
The procedures of the cyclic test series are described in
Fig. 1 (total duration of the test was 100 weeks). Every
week the MOAs were subjected to a +50°C water bath
for 6 days. Every week accurate internal leakage current
(e.g. LC) measurements were performed to check the
water content penetrated into the MOAs. Between every
humidity stress period an impulse current stress series
was applied to the MOAs.
Internal flashover occurred during the test in 12 (out of
20) of the arresters tested. Three of the other eight
specimens failed (short circuited) during the test due to
heavy moisture ingress, in three of the arresters some
moisture ingress was measured but the test could be
completed and in two of the arresters no humidity
+ 50oC water
immersion, 6 days
LC (dc)
measurements
+ impulse
current stress
+ 50oC water immersion, 6 days
Fig. 1. One cycle of the test series
TABLE 1 −Flashover failure cases
Failure at
Arrester
Power
losses
before
failure at
Uc [W]
Even very stringent environmental factors may stress
metal oxide surge arresters (MOAs) during their fairly
long desired life time (20 – 50a). For example, a large
number of impulses, ambient humidity and rain will
cause remarkable stresses on MOAs especially in
regions of high lightning activity. Detailed data on the
effects of different stresses on MOA performance are
difficult to obtain from field experiences because these
experiences are often based on failure cases and failed
arresters are normally burned with power arcs to such a
condition that only general assumptions of the initial
reason of the failure can be made. Either exhaustive
laboratory tests or instrumentation of arresters in field
service is needed to gather thorough information.
PD + LC (ac+dc)
measurements +
impulse current stress
Arrester
structure
SUMMARY
week impulse no
4.3
I
376
38
A1
10.3
I
951
96
B1
1.0, 0.9, 1.0
III
2, 4, 50
C1 − C3 1, 1, 5
>100, >200
F1, F2 46, 46 ~460, ~460 III
0.7, 0.7, 1.4
II
I1 − I3 2, 2, 2 20, 12, 14
19.6
I*
244
25
A3o
3.8
II*
145
15
E3o
*Older model of the type A or E arrester.
ingress was measured. No internal flashovers occurred
in these eight specimens. Some details of the flashover
incidences are given in Table 1.
Conclusions
The internal moisture contents caused by the ca. two
year test series in stringent conditions had clear effects
on the internal flashover tendency in only one arrester
type (F) which has considerable internal air spaces.
Arrester types with tight glass fibre reinforced epoxy
tube around the MO discs did not exhibit this tendency
even in cases of very high internal power losses. The
types with housing moulded directly on the arrester
inner parts had only moderate internal power loss
increases and no clear effects of this to failure cases
could be shown.
Surprisingly, many internal flashover failures occurred
during the test series with only moderate impulse
current stresses (12 out of 20 arresters failed). Almost
all the failure cases were flashovers over one or more
MO discs and failures occurred for most of the arrester
types tested. The main reason for the failures must be
related to the voltage stress caused by longer duration of
the test impulses (2.5/70µs) compared to typical
standard current impulse (8/20µs).
LA DURABILITE ET LA PERFORMANCE DES PARAFOUDRES A OXYDE DE ZINC A ENVELOPPE
SYNTHETIQUE SOUS CHOCS DE SURTENSIONS
K Lahti, K Kannus and K Nousiainen
Université de Technologie de Tampere, Finlande
De façon inattendue, beaucoup de cas de défaillance des
types modernes de POZ ont eu lieu pendant les tests en
laboratoire lorsque les parafoudres ont été soumis à des
multipulsions ou aux chocs de surtension. Dans ce
document, l’impact du contenu de l’humidité intérieure
sur la tendance de défaillance des arcs est étudié. Les
résultats des tests sont énoncés et comparés aux résultats
d’une série d’impulsions de surtension à sec conduite
sur des parafoudres du même type antérieurement à
l’Université de Technologie de Tampere.
Procédures de tests et cas de défaillance
Au total, huit différents types de parafoudres à oxyde de
zinc sans trou, disponibles dans le commerce, neufs et
de classe de distribution ont été utilisés dans les tests.
De plus, on a testé deux spécimens plus âgés
précédemment utilisés dans le réseau. Les parafoudres
provenaient de sept fabricants d‘Europe et d’Amérique
du Nord. Le nombre total de spécimens de test était 20.
Les procédures des séries de tests cycliques sont durée
totale 100 semaines. Chaque semaine, les POZ étaient
immergés dans une eau à +50ºC pendant six jours.
Chaque semaine, le courant continu (CC) était mesuré
pour vérifier le contenu de l’eau pénétrée dans les POZ.
Entre chaque période d’agression d’humidité, des séries
d’agressions de courants continus ont été appliquées sur
les POZ.
A1
B1
C1 − C3
F1, F2
I1 − I3
A3o
E3o
Défaillance à
semaine
Nombre
d’impulsions
38
96
1, 1, 5
46, 46
2, 2, 2
25
15
376
951
2, 4, 50
~460, ~460
20, 12, 14
244
145
I
I
III
III
II
I
II
Baisse de
tension
avant
défaillance
à Uc [W]
TABLEAU 1 – Cas de défaillance d’arcs
Structure de
parafoudre
Même les conditions environmentales très rigoureuses
peuvent avoir un impact sur les parafoudres à oxyde de
zinc (POZ) pendant la durée d’utilisation souhaitée (20
– 50 ans). Par example, le nombre élevé d’impulsions
de choc, l’humidité ambiante et la pluie exercent des
agressions importantes sur les POZ, surtout dans les
régions d’intenses activités de foudres. Il est difficile
d’obtenir, à partir des expériences de terrain, des
données précises sur l’impact de différentes aggressions
sur les POZ, parce que ces expériences sont souvent
basées sur les cas de défaillance et parce que les
parafoudres défaillis sont d’habitude brûlés avec les arcs
de telle sorte qu’il n’est possible d’avancer que des
hypothèses générales sur la raison originale de
défaillance. Pour obtenir des informations approfondies,
on a besoin de faire soit des tests minutieux en
laboratoire soit des études de mécanismes des
parafoudres sur le terrain.
L’arc interne a eu lieu dans 12 (des 20) parafoudres
testés. Trois des huit autres spécimens ont défailli
(court-circuités) pendant le test à cause de la formation
importante d’humidité, dans trois des parafoudres
certaine formation d’humidité était mesurée mais le test
pouvait être poursuivi jusqu’à la fin, et dans deux des
parafoudres aucune humidité n’était mesurée. Il n’y a
pas eu d’arcs internes dans ces huit spécimens. Certains
détails sur les incidences d’arcs sont donnés dans le
Tableau 1.
Parafoudre
RÉSUMÉ
4.3
10.3
1.0, 0.9, 1.0
>100, >200
0.7, 0.7, 1.4
19.6
3.8
Conclusions
La formation d’humidité intérieure causée par les séries
de tests pendant près de deux ans, dans des conditions
rigoureuses, a eu un impact net sur la tendance des arcs
internes seulement dans un type de parafoudres (F), qui
a des espaces d’air internes importants. Les types de
parafoudres munis d’un solide tube époxy renforcé de
fibre de verre autour des disques d’OZ n’ont pas
présenté cette tendance même en cas de baisses de
courant internes très élevées. Les types ayant
l’enveloppe attachée directement sur les parties internes
de parafoudres avaient seulement des augmentations
modérées de baisses de courant internes, et il n’était pas
possible de démontrer l’impact clair de cela sur les cas
de défaillance.
Étonnamment, beaucoup de défaillances d’arcs internes
ont eu lieu pendant les séries de tests lorsque les chocs
de surtensions étaient modérés (12 des 20 parafoudres
ont défailli). Presque toutes les défaillances étaient des
arcs sur une ou plusieurs disques d’OZ, et les
défaillances sont arrivées à la plupart des types de
parafoudres testés. La raison principale de défaillance
doit être liée aux chocs de tension causés par la durée
plus longue d’impulsions de test (2.5/70µs) comparées à
des impulsions typiques de courant standard (8/20µs).
THE DURABILITY AND PERFORMANCE OF POLYMER HOUSED METAL OXIDE SURGE ARRESTERS
UNDER IMPULSE CURRENT STRESSES
K Lahti, K Kannus and K Nousiainen
Tampere University of Technology, Finland
ARRESTERS TESTED
A total of eight different types of commercially
available, new, distribution class, gapless polymer
housed metal oxide surge arresters were used in the
tests. In addition to this, two older specimens previously
used in networks, were also used. The arresters were
from seven manufacturers in both Europe and North
America. The total number of test specimens was 20.
The eight commercial arrester types cover a wide range
of polymer arrester technology available today. Some
properties of the arresters tested are listed in Table 1.
The exact housing material recipes are not known to the
authors. ‘Moulded housing’ in the table refers to a
housing moulded directly onto the arrester body. The
The internal gas space was evaluated by dismantling the
arresters and checking the interfacial area between
housing and inner structure visually. The types with
‘moulded housings’ could not be checked because the
housings could not be dismantled without destroying
them.
20
20
18
18
17
19.5
24.4
19.5
Uc
EPDM
EPM
X
X
X
X
X
X
X
5
4
6
4
4
4
3
3
Test
specimens
Nos.
Silicone
Silicone
EPDM
EPDM
EPDM
EPM
Silicone
Silicone
Thickness of
housing
between sheds
[mm]
A
B
C
D
E
F
H
I
Moulded
housing
Internal gas
spaces or voids
TABLE 1 -ARRESTER TYPES TESTED
Uc [kV]
Unexpectedly many failures of modern MOA types
have happened in some laboratory tests where arresters
are stressed with either multipulses [1] or series of
impulse current stresses [2]. The typical failure mode in
these tests was an internal flashover over one or more
metal oxide discs in an MOA. In this paper the effects
of internal humidity content on the internal flashover
failure tendency are reported. The polymer housed
arresters tested were exposed to a humidity stress in
such a way that internal humidity content in the
arresters gradually increased. At the same time impulse
current stress series were regularly applied to the
specimens. The test results are discussed and compared
to the results obtained in a similar impulse current stress
series [2] conducted earlier on arresters of the same
types at Tampere University of Technology. In that test
series only impulse current stresses were applied on
arresters without any other stresses.
The two older specimens, A3o and E3o, were 1991
models of the arrester types A and E. The data given in
Table 1 are also valid for these arresters with the
exceptions that the Uc –value of specimen E3o is 18 kV
and A3o has an internal gas space. Nominal discharge
current (In) of all the arresters was 10 kA.
Housing
material
Metal oxide surge arresters (MOAs) can be exposed
even to very stringent environmental conditions during
their fairly long desired life time (20 – 50a). For
example, a large number of impulses, ambient humidity
and rain will cause remarkable stresses on MOAs
especially in regions of high lightning activity. Detailed
data of the effects of different stresses on MOA
performance are difficult to obtain from field
experiences because these experiences are often based
on occasional failure cases. Failed arresters are normally
burned with power arcs in such a way that only general
assumptions of the initial reason of the failure can be
made. Either exhaustive laboratory tests or
instrumentation of arresters in field service are needed
to gather thorough information.
arrester types marked in the ‘internal gas space’ column
have a mechanical structure allowing an internal air
space. The arresters of type F have a considerable
internal gas space (total 5-10 cm3 ), while the arresters
of types E and I have smaller voids inside (some mm3 or
less). In type D arresters the interface between the
housing and the arrester body is sealed using an elastic
sealing paste.
Arrester
type
INTRODUCTION
A1, A2, A3o
B1, B2
C1, C2, C3
D1, D2
E1, E2, E3o
F1, F2
H1, H2
I1, I2, I3
Max. continuous operating voltage according to
IEC 60099-4
Ethylene-propylene diene monomer
Ethylene-propylene monomer
Mechanical Structure of the Specimens
Moisture can penetrate polymer housed arresters by
diffusion through polymer material or by capillary
effect because of leaks/cracks in the end cap sealings or
housing. Interfaces between different materials are the
potential areas where moisture can form a conducting
layer along the arrester. The area into which the
humidity will first form a conducting layer inside an
arrester depends on the mechanical structure of the
arrester. In this paper we report internal flashovers in
arresters and the probability of moisture in the flashover
path is of great interest. The arrester types are divided
PD + LC (ac+dc)
measurements +
impulse current stress
+ 50oC water
immersion, 6 days
LC (dc)
measurements
+ impulse
current stress
TEST ARRANGEMENTS AND PROCEDURES
The total testing time was ca. 2 years and arresters
which did not fail during the test gathered thus totally
1000 impulses. As an exception the arresters of type H
were taken later on the test series and they were tested
for ca. 20 months (800 impulses).
TABLE 2 – ARRESTER TYPES ACCORDING TO
INTERNAL STRUCTURE.
Fig. 1 One cycle of the test series
RESULTS AND DISCUSSION
General Results and Failure Cases
Internal flashover occurred during the test in 12 (out of
20) of the arresters tested. Three of the other eight
specimens failed (short circuited) during the test due to
heavy moisture ingress (D1, D2, E2), in three of the
arresters some moisture ingress was measured but the
test could be carried out to the end (A2, B2, E1) and in
two of the arresters no humidity ingress was measured
(H1, H2). No internal flashovers occurred in these eight
specimens. Some details of the flashover incidences are
given in Table 3. AC power loss and residual voltage
curves for all the arresters (except arresters of type C
and I) are given in Figs 2 – 8. These parameters give an
indication of the internal humidity content inside an
MOA and the impulse current behaviour of an MOA.
The arresters of type C and I failed during the first two
test weeks and power loss or residual voltage trends
were thus not gathered before the failures. Considerable
partial discharge activity was measured in MOAs D1,
D2, B1, E2 and E3o during the test and some activity
also in MOAs A2, B2 and A3o [3].
TABLE 3 −Flashover failure cases
Failure at
Arrester
week impulse no
4.3
A1
38
376
I
10.3
B1
96
951
I
1.0, 0.9, 1.0
2, 4, 50
III
C1 − C3 1, 1, 5
>100, >200
F1, F2 46, 46 ~460, ~460 III
0.7,
0.7, 1.4
2,
2,
2
20,
12,
14
II
I1 − I3
19.6
244
I*
25
A3o
3.8
II*
15
145
E3o
*Older model of the type A or E arrester.
MOA
types
I
A, B
II
D, E, I
III
C, F
IV
H
Housing moulded directly on MO disc
surfaces, mechanical support strips inside
housing material
Tight glass fibre reinforced epoxy tube on
the surface of MO discs, separately
manufactured housing
Untight glass fibre reinforced epoxy
support (net) on the surface of MO discs
Housing moulded directly on MO disc
surfaces, glass fibre reinforced epoxy tube
with openings inside housing material
100
AC losses, A1
AC losses, A2
Residual voltage, A1
Residual voltage, A2
45
AC power loss at Uc (W)
Group
50
Description
Power
losses
before
failure at
Uc [W]
Between every humidity stress period an impulse stress
series was applied to the MOAs. The impulse current
test procedure was the following: The MOA was
stressed with five positive current impulses followed by
five negative current impulses. The time interval
between the current impulses was on average 1.5
minutes and a total of 10 current impulses were applied
to an arrester in approximately 20 minutes. The peak
value of the current impulses varied from 1.7 to 2.7 kA
depending on the type of the MOA tested. The wave
shape of the current impulses with MOAs was
approximately 2.5 / 70 µs defined according to IEC
60060-1. The front steepness of the voltage curve
during the current impulses with MOAs (measured from
the residual voltage recordings) was 400 − 480 kV/µs.
The energy stress of one impulse to an arrester was
approximately 10 kJ.
+ 50oC water immersion, 6 days
Group
(Table 2)
In this test the arresters are subjected to a humidity
stress due to which internal moisture in MOAs
gradually increases. The content of the test series is
described in Fig. 1. Every week the MOAs were
subjected to a +50°C water bath for 6 days in order to
cause a relatively fast moisture diffusion inside the test
specimens. Every second week accurate AC power loss
(Leakage Current, e.g. LC) measurements were
performed (at Uc voltage) to check the water content
penetrated into the MOAs. Internal DC leakage current,
partial discharge and electro magnetic radiation
measurements were also performed. These results are
presented in [3]. The DC leakage current measurements
were performed both before and after the impulse
current stressings to find out possible changes caused by
that stress. Residual voltages were measured each week
to check the impulse current performance of the MOAs.
40
35
90
80
70
30
60
25
50
20
40
15
30
10
20
5
10
0
Residual voltage (kV)
into groups according to their internal structure in Table
2.
0
0
20
40
60
80
100
Test time (weeks)
Fig. 2. Residual voltage and AC power loss
measurements for MOAs A1 and A2 during the test.
30
60
25
50
20
40
15
30
10
20
5
10
0
0
0
20
40
60
Test time (weeks)
80
35
50
90
45
80
40
60
25
50
20
40
15
30
10
20
5
0
20
40
60
Test time (weeks)
80
40
35
70
60
20
10
0
0
0
0
70
50
20
40
15
30
10
20
5
10
0
100
Fig. 5. Residual voltage and AC power loss
measurements for MOAs E1 and E2 during the test.
20
40
60
Test time (weeks)
80
100
Fig. 8. Residual voltage and AC power loss
measurements for MOAs A3o and E3o during the test.
MOA A1 failed at the 376th impulse. An internal
flashover occurred over one (lowest) MO disc (Fig. 9A).
Power losses rose slightly during the test from the initial
value of 2.1W to 4.3W before the failure and some
internal humidity was thus present inside the MOA
when the flashover happened. MOA B1 failed at 951st
impulse. An internal flashover happened over the whole
length of the arrester along the surface of a glass fibre
reinforced epoxy strip. The strip was inside silicone
housing material. Power losses rose during the test from
the initial value of 1.2W to 10.3W. A change of power
losses occurred few weeks before the final failure
(Fig. 3) and a prefailure may thus have occurred then.
Some moisture was, at least, present inside the MOA
during the failure.
100
90
40
80
AC losses, F1
AC losses, F2
Residual voltage, F1
Residual voltage, F2
35
30
25
70
60
50
20
40
15
30
10
20
5
10
0
Residual voltage (kV)
45
AC power loss at Uc (W)
30
80
5
100
50
35
10
80
80
90
30
60
40
60
Test time (weeks)
100
AC losses, A3o
AC losses, E3o
Residual voltage, A3o
Residual voltage, E3o
10
25
20
0
100
40
30
0
80
15
90
0
40
60
Test time (weeks)
50
Residual voltage (kV)
AC power loss at Uc (W)
45
20
20
100
AC losses, E1
AC losses, E2
Residual voltage, E1
Residual voltage, E2
20
25
Fig. 4. Residual voltage and AC power loss
measurements for MOAs D1 and D2 during the test.
50
30
Fig. 7. Residual voltage and AC power loss
measurements for MOAs H1 and H2 during the test.
100
70
40
5
0
30
0
10
50
10
Residual voltage (kV)
AC power loss at Uc (W)
40
60
AC losses, H1
AC losses, H2
Residual voltage, H1
Residual voltage, H2
15
100
AC losses, D1
AC losses, D2
Residual voltage, D1
Residual voltage, D2
45
70
0
Fig. 3. Residual voltage and AC power loss
measurements for MOAs B1 and B2 during the test.
50
80
20
Residual voltage (kV)
70
25
Residual voltage (kV)
80
AC power loss at Uc (W)
35
90
90
AC power loss at Uc (W)
AC power loss at Uc (W)
40
100
30
100
AC losses, B1
AC losses, B2
Residual voltage, B1
Residual voltage, B2
45
Residual voltage (kV)
50
0
0
20
40
60
Test time (weeks)
80
100
Fig. 6. Residual voltage and AC power loss
measurements for MOAs F1 and F2 during the test.
Arresters of types C and I failed very soon after the test
was started (Table 3). In all the cases the failures were
internal flashovers over one or more MO discs. The
power losses were at initial level before the failures and
failure processes were thus not affected by internal
humidity.
Arresters F1 and F2 degraded gradually during the test.
The final internal flashovers occurred after ca. 460
impulses. Partial degradation started much earlier as can
be seen in the residual voltage curves in Fig. 6.
Flashovers over MO discs have thus happened much
earlier than after the 460 impulses. A very high internal
leakage current caused by internal humidity flowed
inside the arresters, which very obviously affected the
flashover events. A flashover over one MO disc in
MOA F1 can be seen in fig. 9B.
A
B
Fig. 9. Flashover failure over one MO disc in MOA A1
(A) and in MOA F1 (B).
MOA A3o failed at 244th impulse when an internal
flashover occurred over several MO discs in the hole
inside of the MO discs where a mechanical support bar
is assembled. Power losses of the arrester increased
during the last few weeks before the failure. The final
losses were 19.6W while the initial level was 2.8W.
MOA E3o failed at 145th impulse. In this case an
internal flashover also occurred between MO discs and
glass fibre reinforced epoxy tube around the discs. The
power losses increased during the test from the initial
level of 0.5W to 3.8W.
Internal Humidity vs. Flashover Failures
Similar impulse current stress series without any
moisture stress have been conducted earlier at Tampere
University of Technology [2] for most of the arrester
types studied in this test. In that test series exactly
similar impulse current stresses (5+5 impulses of ~2kA,
2.5/70µs in 20 minutes) were applied to the test
specimens once a day or less frequently. No other
stresses were applied to those arresters. Nine arresters
out of 48 specimens studied failed during the test series.
Failure cases of arrester types similar to the types in this
test series are indicated in Table 4. All the failure cases
(except one) were internal flashovers over the MO
disc(s) of an arrester. In one case (type A, failure at 14th
impulse) a current channel was found in one MO disc
after the failure.
The possible effects of internal moisture on internal
flashover failure probability are discussed in the
following based on the results of the two test series.
TABLE 4 –Failures due to impulse current stress
without any other stresses conducted [2].
MOA type / type No of specimens / Failed at
index at [2]
No of failure cases impulse
A / 300* + 350
18 / 2
9th, 14th
C / 900
2/1
9th
D / 100
6/0
E / 500
6/2
121st, 527th
I / 950
2/2
51st, 93rd
A3o / 390
2/1
882nd
*Same as type A but Uc=22kV, type 350 same as type A
Type A: 1 out of 2 MOAs failed in these tests with
moisture stress and 2 of 18 in the dry test series. The
failure of MOA A1 was an internal flashover failure
over one MO disc similar to the other failure case in the
‘dry test series’. The internal power losses due to
humidity ingress were moderate (max total losses
~10W). The arrester A1 failed already when the internal
power losses were ca. 5W while the other test specimen,
A2, stood up the whole test series without a failure with
power losses over 10W during the last half of the testing
period (ca. 1 year and 500 impulses). The mechanical
structure of type A is such that the conductivity formed
by internal moisture is most likely located in the
interface between the housing material and MO discs,
which is also the area where the flashovers occurred. In
any case, according to these results the moderate
amounts of internal moisture did not notable affect the
flashover probability of this type.
Older type A: The arrester A3o failed in the ‘wet test’
and one MOA out of two failed in the ‘dry test series’.
Both failure cases were totally similar internal
flashovers but A3o failed after a much lower number of
impulses than the MOA in ‘dry series’. Some moisture
was inside MOA A3o during the failure and this may
have had a slight effect on the failure if it was on the
area where the arc burned.
Type B: 1 out of 2 MOAs of this type failed in these
tests. This type was not studied in the ‘dry test series’.
Internal power losses for this type were at a moderate
level during the test. The one failure case was different
from all the other cases in these tests. Humidity may
have partially caused this failure because humidity most
likely first penetrates the interface between the housing
and mechanical support strip where the flashover also
occurred. Clear evidence of this, however, cannot be
evinced.
Types C and I: Internal flashovers occurred in all the
arresters of these types with only one exception (in the
‘dry test series’). The failures were all similar flashovers
over MO discs in both test series and occurred typically
almost immediately after the impulse stress was started.
No remarkable increases of internal power losses were
measured for these MOAs. The flashover cases were
thus typical for these MOA types and these failure cases
were not affected by any internal moisture in the
arresters.
Type D: No flashovers occurred in either of the test
series. Power losses caused by internal moisture reached
very high levels in MOAs D1 and D2 but the moisture
did not cause any problems for the impulse current
behaviour of the arresters. In any case, the conducting
layer formed by the moisture was most likely at the
interface between housing and the tight glass fibre
reinforced epoxy tube and not between the tube and the
MO discs where the flashovers are most likely.
Type E: Two failures out of six specimens occurred in
the ‘dry test series’. No flashovers occurred in the two
MOAs in the tests with moisture stressing but the older
arrester E3o had a flashover. These three cases were
similar flashovers over MO discs. Not even very high
internal moisture content (E2) caused problems in the
impulse current behaviour of the arrester. The structure
of this type is similar to the type D above and also in
this case the humidity was most probably not in the area
where flashovers typically occur.
even high conductivities caused by internal humidity
will typically cause such effects when MO discs are
covered by tight glass fibre epoxy tube and the moisture
remains outside the tube (types D and E). In structures
with housing moulded directly on the inner parts of the
MOA very high internal conductivity is not typically
reached and internal humidity will thus not cause
flashover problems.
Type F: Internal flashovers occurred in these arresters
when high internal humidity content was measured. The
flashover events were very probably caused by the
moisture but power losses of 50W or more were needed
(see Fig. 6) to start the problems. Even much lower
power losses would fail an arrester immediately under
service AC stress. The internal structure of the type has
considerable internal air space and humidity penetrating
this space would be in contact with the MO disc
surfaces.
CONCLUSIONS
Type H: No flashovers occurred for this type and nor
did any measurable humidity penetrate inside the
arrester structure during the stringent humidity stress.
Thus the effect of internal moisture cannot be evaluated
but the service behaviour of this type under these
stresses appears excellent.
The number of test specimens was limited and only
major effects of moisture could thus be seen in the
results. According to the results the internal moisture
seemed to have a clear effect on the internal flashover
probability only in case of type F arresters, when a
considerable amount of moisture was present on the
surface of the MO discs where the flashover probability
is highest. The amount of moisture needed for those
immediate reactions was so high that the MOAs would
have failed much earlier under normal service AC
stress.
Such a high number of internal flashover failures in
both tests is in any case astonishing and must be related
to the long impulse (2.5/70µs). Peak values (~2kA) and
energy (~10kJ) of the impulses should be easily handled
with these In=10kA MOAs. The parameters of these
impulses are also quite normal compared to real
lightning surges [4]. The test with longer impulse
actually acts as a voltage stress for the whole MOA
structure. This kind of stress is not included in the IEC
60099-4 standard and it should be considered whether
voltage stress using long current impulse should be
included there. Similar flashover failures have also
occurred with multipulse stresses [1] which introduce
also a long voltage stress as the impulses used in our
tests.
Arrester Internal Structure vs. Flashover Failures
Arresters with internal air space(s) between the internal
parts and housing are problematic with respect to seal
integrity [5]. When the air space also reaches the
surfaces of the MO discs possible internal moisture may
also cause internal flashover problems (type F). Not
The internal moisture contents caused by the ca. twoyear test series in stringent conditions had clear effects
on the internal flashover tendency only in one arrester
type. High internal power losses due to the humidity
was, anyhow, needed to cause the effects. Arrester types
with tight glass fibre reinforced epoxy tube around the
MO discs did not show this tendency even in case of
very high internal power losses. The types with housing
moulded directly on the arrester inner parts had only
moderate internal power loss increases and no clear
effects of this on failure cases could be shown.
Surprisingly, many internal flashover failures occurred
during the test series with only moderate impulse
current stresses (12 out of 20 arresters failed). Almost
all the failure cases were flashovers over one or more
MO discs and failures occurred for most of the arrester
types tested. The main reason for the failures must be
related to the voltage stress caused by the longer
duration of the test impulses (2.5/70µs) compared to
typical standard current impulse (8/20µs).
REFERENCES
[1] Darveniza M., Tumma L. R., Richter B. and Roby
D.A.,1997, ”Multipulse lightning currents and
metal-oxide arresters”, IEEE Transactions on Power
Delivery, 12, 1168-1175.
[2] Kannus K., Lahti K., Nousiainen K., 1999, ”Effects
of impulse current stresses on the durability and
protection performance of metal oxide surge
arresters”, ISH’99, vol 2, London, UK, 4 p.
[3] Lahti K., Pakonen P., Kannus K., Nousiainen K.,
1999, ”Possibilities to reveal internal moisture in
polymeric surge arresters by means of PD and EMR
measurements”, IEEE Power Tech’99, Budapest,
Hungary, 6 p.
[4] Anderson R. B., Eriksson A. J., 1980, ”Lightning
parameters for engineering application”, Electra,
No 69, 65 – 102.
[5] Lahti, K., Richter, B., Kannus, K., Nousiainen, K.,
1999, ”Internal degradation of polymer housed metal
oxide surge arresters in very humid ambient
conditions”. ISH’99, vol. 2, London, UK, 4 p.