IEEE Transactions on Power Delivery, Vol 14, No. 3, July 1999
940
Watts Loss of Polymer Housed Surge Arresters
in a Simulated Florida Coastal Climate
David R. Miller, Member, IEEE, J.J. Woodworth, Member, IEEE and C.W. Daley, Member, EEE
Cooper Power Systems
Olean, NY 14760-0388
Abstract: Polymer housed surge arresters, evaluated under
accelerated conditions of a Florida coastal climate, are
shown to exhibit an increasing watts loss characteristic with
service life corresponding to aging of the polymer housings.
Surge arrester housings made of materials other than
silicone rubber are shown to age more rapidly and therefore,
show significantly higher watts loss with service life.
Lifetime energy costs of polymer housed surge arresters,
will need to factor in the watts loss of the housing due to wet
surface contamination.
Keywords: Polymer housed surge arresters, watts loss,
accelerated aging, silicone housings, multistress aging,
surge arrester aging, coastal climate, Florida cycle.
I. INTRODUCTION
Surge arrester use of polymer housings was introduced in
the mid to late 1980’s. Since their introduction, the
technology has advanced rapidly with the use of new
materials and constructions. The early metal oxide surge
arresters exhibited increasing watts loss with normal system
conditions due to penetration of moisture into the metal
oxide blocks thereby promoting oxidation [13. This form of
aging no longer takes place in more recently manufactured
metal oxide surge arresters. However, watts loss of surge
arresters is dependent not only on the internal watts loss of
the metal oxide valve blocks but also on the external watts
loss due to wet surface contamination. As some housing
materials are known to age more rapidly than others [2],
watts loss can be expected to increase with normal service
aging on some polymer housed surge arresters.
Until very recently, the industry has largely concentrated on
evaluating surge arrester performance under various system
disturbances.
Now that acceptable surge arrester
performance under these conditions has been demonstrated,
today’s emphasis on testing is on evaluating polymer
housing performance under various climatic conditions
[2,3,4,51.
PE-201-PWRDo-o5-1998 A paper recommended and approved by the
IEEE Surge Protective Devices Committee of the IEEE Power
Engineering Society for publication in the IEEE Transactions on Power
Delivery. Manuscript submitted December 12,1997; made available for
printing May 18, 1998.
These aging tests are performed by combining various
climatic stresses in order to have a synergistic effect on both
the polymer housing and on the arrester construction that
duplicates normal service aging, but at a much faster rate.
Such life tests are intended on reproducing the combined
effects of voltage, ultraviolet radiation, moisture,
contamination, and thermal cycling (multistress aging
testing) as seen in service.
In an earlier publication [2], it was shown that multistress
aging testing is an important design test as polymer housed
surge arresters having improper design of the insulation
system fail under test conditions accelerating normal service
aging. It was also shown that surge arrester housings made
of materials other than silicone rubber exhibited
significantlyhigher leakage current under wet contamination
which manifests itself by increased watts loss and faster
aging of the polymer housing. Aging, is an important
consideration of polymer housed surge arresters and had has
been the fundamentalreason for multistress testing.
For many years, the utility industry has evaluated the power
loss of distribution transformers. A parallel economic
analysis can be applied to arresters. Very simply, watts loss
savings convert to net dollar savings particularly when
thousands of arresters are installed on a distribution system.
In the analogy, the continuous surge arrester watts loss
during dry conditions is analogous to the “no load” loss of
the transformer as this loss is always present. The surge
arrester watts loss due to wet surface contamination, is
analogous to “load loss” of the transformer. As some
utilities are beginning to evaluate the long term operating
costs of polymer housed surge arresters, the average watts
loss under the operating climatic conditions needs to be
determined.
The present paper focuses on the watts loss aspect of various
arrester designs under accelerated conditions of a simulated
Florida coastal climate.
11. SIMULATION OF THE CLIMATIC CYCLE
In a research program co-sponsored by the Electric Power
Research Institute (EPRI) and Florida Power and Light
Company (FPL), nonceramic insulators were evaluated in
an accelerated multistress aging chamber that was designed
0885-8977/99/$10.00 0 1998 IEEE
941
to simulate the weather pattern of the coastal Florida climate
[6]. As this climate cycle forms the basis for the present
work, as summary of the simulation in [6] is presented in
this section.
The simulation of the coastal Florida climate stems from an
analysis of the transmission line outages, which identified
two distinct causes of outages; lightning and contamination.
From May through September, lightning was identified as
the main reason for the outages. This period is characterized
by clean rain, ultraviolet and warm temperatures. From
October through April, contamination was identified as the
main reason for the outages. This period is characterized by
some clean rain, ocean salt contamination, ultraviolet and
cooler temperatures. Hence, a summer period and a winter
period was established for the simulation of the coastal
climate.
The technical basis for the parameters for the summer and
winter periods was established through an analysis of the
weather records and measurements of the solar radiation.
This established the average temperatures of 41-45OC for the
summer period and 31-41OC for the winter period, and a
solar irradiance level of 1.0 mW/cm2at 350 nm wavelength
for both periods. Clean rain having a conductivity of 50-70
uS/cm was selected and a salt fog salinity of 2.5 kg/m3
(4000 uS/cm) was chosen based on tests that established a
salinity level that produced a visual discharge activity
similar to that observed in service. A sequence of summer
and winter cycles in which one calendar year in service was
assumed to be represented by 10 laboratory days of the
summer cycle followed by 11 laboratory days of the winter
cycle. These two climatic cycles illustrated in Fig. 1, and
repeated over a 24 hour period, form the so called “Florida
cycle” giving an acceleration factor for aging of 365/21, or
about 17.
In support of the proper simulation of the coastal climatic
cycle, the erosion found on nonceramic insulators in the
multistress chamber was found to correlate well with the
erosion found on nonceramic insulators that had service
experience.
111. DESCRIPTION OF THE TEST CHAMBER
The accelerated aging chamber is constructed from
polycarbonate sheet on an aluminum h e , 1.2m wide x
1.2m high x 2.4m long, and can be used for simultaneously
aging 16, lOkV rated polymer housed surge arresters. The
chamber is capable of subjecting these surge arresters to the
stresses of ultraviolet radiation (type UV-A fluorescent
lamps), salt water spray (IEC 507 nozzles), clean water
spray (demineralized), and controlled temperature
under MCOV voltage (8.40 kV rms). A data acqu
system is utilized for monitoring leakage current and
voltage, performing real time analysis, calculation of watts
loss, and controlling aging cycle parameters as described
above for coastal Florida cycle. A 15 kV, 25 kVA
distribution transformer used in reverse served as the test
supply and the voltage supply was regulated by computer.
Each surge arrester was fused so that flashover or failure
automatically disconnected it from the test supply.
IV. SURGE ARRESTERS EVALUATED
Table I describes the polymer housed surge arresters
evaluated in this test. The surge arresters were heavy and
normal duty, 10 kV rated and the applied voltage was 8.4
kV rms throughout the test.
The housing materials included silicone rubber (SR),
ethylene-propylene rubber (EPDM), and co-polymers of
EPDM and SR (EPDWSR). Three surge arresters of each
housing (two with EPDM) were evaluated in the test.
Summer Cycle
TABLE I
CHARACTERISTICS OF SURGE ARRESTERS EVALUATED
Sample
Number
Housing
Type
Creepage
Distance
Strike
Distance
Watts Loss
at
203
203
0.19
Time (hours)
Winter Cycle
1
2
3
4
5
Time (hours)
Fig. 1. Summer and Winter Cycles of the “Florida Cycle”.
.
0.17
942
It is evident that watts loss increases significantly for the
EPDWSR housed surge arrester during the wet periods, i.e.,
during rain in the summer and during fog and rain in the
winter cycles. No increase in watts loss is seen during the
wet periods for the SR housed surge arrester. Clearly, this
result is due to the loss of hydrophobicity of the EPDWSR
housed surge arrester which takes place at a very early stage
of the test. The SR housed surge arrester exhibits good
retention of hydrophobicity preventing watts loss on the
polymer housing. Watts loss on surge arrester housings
gives rise to the degradation effects of tracking and erosion
which is commonly referred to as aging.
Surge arresters A6, A7 and A8 were "gapped" types with
grading resistors all others were "gap-less'' types.
V.RESULTS OF TESTS
Fig's. 2, 3,4and 5 show comparative watts loss results near
the beginning of the accelerated test for portions of the
summer and winter cycles for the EPDWSR and SR housed
surge arresters A9 and A6, respectively.
I
o ~ , , , , , , , , . , , , , , , , ,
0
1
2
3
5
4
7
6
8
, , ,
9101112
Time (h)
o ~ , ~ , , , , , , , , , , , l , , , l , l , l ,
0 1 2 3 4 5 6 7 8 9 1 0 1 1 '
Time Ih)
121
\
I
I
2.5
I 1 121 2.5
bog 15 I HeaWV I Fog 1.5I HeaWV I Fog I HeatNV
a
a
a
1
2.5
I 1
121
1s
Cycle
Fig. 2. Watts loss for EPDWSR housed surge arrester A9
during an early part of the accelerated test for a portion of
the summer cycle.
Fig. 4. Watts loss for EPDWSR housed surge arrester A9
during an early part of the accelerated test for a portion of
the winter cycle.
51.4
U) 1
.2
3 1.0
0.2
0 . 0 , .~ ' ,
0
2
0.2
.
~
'
4
I
.
)
.
6
I
Time (h)
.
I
8
,
I
.
I
10
.
I
,
I
.
I
'
I
12
1
Fig. 3. Watts loss for SR housed, gapped, surge arrester A6
during and early part of the accelerated test for a portion of
the summer cycle.
.
I
.
2
121
/Fog 1.51
Cycle
.
2
I
.
4
2.5
Heat
IUV
I
'
I
.
I
6
Time (h)
121
I 1
IFog 1.51
a
2.5
Heat
RIV
Cycle
.
I
.
~
8
.
~
.
10
121
I 1
I Fogl.cl
d
~
12
2.5
Heat
NV
1
Fig. 5. Watts loss for SR housed, gapped, surge arrester A6
during an early part of the accelerated test for a portion of
the winter cycle.
.
I
'
943
Although not illustrated in this paper, EPDM housed surge
arresters, A4 and A5, show similar watts loss as the
EPDWSR housed surge arresters.
lerated aging tests continued for 184 laboratory
days. Using the acceleration factor of 365/21, or 17.38 for
this test, this gives a service life equivalent to 1841365 x
17.38 or 8.76 years. The results of these tests, extrapolated
to 10 years, are shown in Fig’s. 6, 7, 8 and 9 for surge
arresters A9, A6, A4 and A2,respectively.
q
18
-L
16
14
2 12
-B 10
58
0
2
o
h
I
6
18
iG
.
I
I
164
A
4
2
0
Years
Fig. 8. External watts loss for EPDM housed surge arrester
A4 over a 10 year simulated service life in a Florida coastal
climate.
0
2
6
4
8
10
Years
Fig. 6. External watts loss for EPDWSR housed surge
arrester A9 over a 10 year simulated service life in a Florida
coastal climate.
1
*i
-1
5 6
0
2
4
Years
6
8
10
Fig. 9. External watts loss for SR housed surge arrester A2
over a 10 year simulated service life in a Florida coastal
climate.
Years
Fig. 7. External watts loss for SR housed, gapped, surge
arrester A6 over a 10 year simulated service life in a Florida
coastal climate.
Shown plotted in Fig’s. 6 through 9 is the average daily
watts loss over the simulated service life in a Florida coastal
climate. It is evident that both the EPDM and EPDWSR
housed surge arresters show significantly greater watts loss
over their silicone rubber housed counterpart which is
attributed to aging of the polymer housings. The silicone
housed surge arrester shows no loss of hydrophobicity over
the first four years of service as evidenced by a watts loss
that corresponds to the internal watts loss of the metal oxide
valve blocks.
944
VI. DISCUSSION
Over the course of these tests, it became evident that
polymer housed surge arresters exhibit an increasing watts
loss characteristic with service life that is due to aging of the
housings. Aging of polymer housings is related to a loss of
hydrophobicity and roughening of the housing by dry band
arcing during rain and fog, and because of ultraviolet
radiation during the dry periods. Non-silicone housed surge
arresters show significant aging, as manifested by .higher
watts loss than silicone housed surge arresters.
It also became quite clear that the lifetime energy costs,
which often only consider the internal or watts loss of the
metal oxide valve blocks, needs to factor in the watts loss of
the housing. In this simulation of the Florida coastal climate,
every 17.4 years of service life has one year of rain and fog
in which the watts loss for the non-silicone housed surge
arresters is at least 10 times higher than the internal watts
loss of the metal oxide valve blocks.
In this simulation of the Florida coastal climate, the dry
periods have been intentionally made short in order to obtain
the largest possible acceleration factor so that the laboratory
test time is reduced to a reasonable level. However, in doing
so, hydrophobic materials such as silicone, lose their
hydrophobicity in a shorter time than in actual service. For
materials such as EPDM and mixtures of EPDM and
silicone, in which hydrophobicity is lost very quickly, these
short dry periods do not have a significant effect on the
outcome of the accelerated test. It is thought that the up-turn
in watts loss, which occurs after about four years of service
life, as seen in Fig’s. 7 and 9 for the silicone housed surge
arresters, occurs much too soon. In fact, it is thought that in
actual service, with considerably longer dry periods between
fog and rain, and in which hydrophobicity is never
completely lost, the up-turn in watts loss will not occur.
4. Lifetime energy costs of polymer housed surge arresters,
which often only consider the internal or watts loss of the
metal oxide blocks, needs to factor in the external watts loss
of the housing due to wet surface contamination.
5. Non-silicone housed surge arresters show an external
watts loss at least 10 times higher than the internal watts loss
due to wet surface contamination.
VIII. REFERENCES
1. W.McDermid, “Testing of Zinc Oxide Arresters”, Doble
paper 9-201, 1986.
2. J.J. Kester, D.R. Miller, S.J. Bema and B.T. Steinbrecher,
“Multistress Aging tests of Polymer Housed Surge
Arresters”, IEEE paper PE-327-PWRD-O-01-1997.
3. V. Chaudhry, G.S. Gorur, M. Dyer and R.S. Thallam,
“Electrical Performance of Polymer Housed Zinc Oxide
Arresters Under Contaminated Conditions”, IEEE paper
9QSM295-6 PWRD.
A. Bargigia, M. de Nigris, A. Pigini and A. Sironi,
“Definition of Testing Procedures to Check the Performance
of ZnO Surge Arresters in Different Environmental
Conditions”, CIGRE paper 33-206, 1992.
4.
G.P. Fini, G. Marrone, A. Porrino, “Results of
Accelerated Ageing Tests on Components of Electric System
made with Polymeric Materials”, CIGRE paper 15-07,1988.
5.
6. H.M. Schneider, W.W. Guidi, J.T. Burnham, R.S. Gorur
and J.F. Hall, “Accelerated Aging and flashover Tests on
138 kV Nonceramic Line Post Insulators”, IEEE Trans
PWRD,Vol8, NO 1, Jan 1993, pp 325-336.
IX.BIOGRAPHIES
VII. CONCLUSIONS
The following can be concluded from the Florida coastal
climate simulation on polymer housed surge arresters:
1. Watts loss of polymer housed surge arresters is dependent
not only on the internal watts loss of the metal oxide valve
blocks but also on the external watts loss of the polymer
housing due to wet surface contamination.
EPDM and EPDWSR housed surge arresters show
significantly higher external watts loss than silicone housed
surge arresters due to wet surface contamination.
2.
Aging of polymer housed arresters is manifested by
increasing watts loss with service life.
3.
David R. Miller received the BS and
MS degrees in Ceramic Engineering
from the New York State College of
Ceramics at Alfred University in 1973
and 1977 respectively. In 1977 he
joined Cooper Power Systems where
he is currently a Senior Staff
Engineer. Mr. Miller has co-authored
several technical papers and is coholder of two patents. Active research
and development interests
include surge arrester, varistors and insulating materials. Mr.
Miller is a member of the IEEE, American Ceramic Society
and the National Institute of Ceramic Engineers.
945
Jonathan J. Woodworth received his
EE in 1972 from Ohio Inst. Tech.,
and an M B A from St. Bonaventure U.
in 1995. He is presently the Arrester
Product
Marketing Manager at
Cooper Power Systems in Olean,
NY. He has been involved in the
arrestef product line since 1979.
Currently, he is the chair of IEEE
WG writing C62.22, “Application
Guide on Arresters”. He also chairs
IEC TC37 WG 9 writing the test standard for Gapped
Metal Oxide Arresters. He is also a member of the US
National committee to IEC TC37 and Chairman of the
NEMA High Voltage Arrester Section of the Power
Equipment Division. He holds several world wide arrester
patents and he possesses the world’s largest and most
prestigious handshake collection.
Charles W. Daley received his BS
degree in Electrical Engineering fkom
Rensselaer Polytechnic Institute in
1984. He was hired by Cooper Power
Systems in 1994 as Senior Product
Engineer to work on Lightning
Arrester products. Prior to 1994, he
was an Electrical Component
Engineer for Fusion Systems
Corporation. His experience includes
HighVoltageEngineering and data/
instnunentation management.
e
946
Discussion
Tiebin Zhao (Hubbell Power Systemsmhe Ohio Brass
Company, Ohio): I would like to make the following
comments before the posing questions to the authors.
The authors state that aging is an important consideration
of polymer housed surge arresters and is the fundamental
reason for multi-stress testing. However, the concept of
aging needs to be clarified. Aging of polymeric materials is
generally understood to be any significant change in the
material, such as change in chemical composition, electrical
properties, mechanical properties, and structure of the
insulation. Aging should be quantifiable by diagnostic
techniques that are appropriate for the aging phenomena
being evaluated. Aging of polymeric materials depends on
the physical and chemical properties of the materials, the
aging test stress levels and duration of the applied stress.
Different diagnostic techniques used could demonstrate
different aging levels of the materials. An understanding of 1
the materials, the diagnostic techniques and the aging tests
can lead to technically sound results. A few diagnostic
techniques which have been commonly used are discussed in
[ I , 21.
Aging of polymeric arrester (or insulator) housing
materials is not closely related to the magnitude of leakage
currents on the housings. This has been demonstrated by
many researchers [3]. As an example, one can often measure
much higher leakage currents or watts loss on porcelain
housed arresters than on polymer housed arresters, yet one
would not claim that porcelain housings age more rapidly
than polymer. On the other hand, although leakage currents
on silicone rubber housings are typically quite low, aging of
the housings can still occur because of the presence of corona
produced by water drops [4,51. The presence or level of
visible corona cannot be used as a dependable indication of
aging of the housing materials. By our experience, very low
levels of surface discharges on some housing materials can
cause high levels of erosion, while high levels of corona may
have little or no effect on other housing materials.
Monitoring of leakage currents over test samples has been
utilized in many studies as a gauge of physical and chemical
changes of polymeric materials. However, with the exception
of gross changes in the rate of accumulation, the data does
not appear to reflect notable compound degradation. As a
matter of fact, the leakage current is a function of test
parameters, such as the voltage stress level, fog water
conductivity level, water flow rate, as well as a function of the
material properties, such as hydrophobicity, material
degradation, and salt deposits on the surface of the samples
(the leakage current flow along the surface of housings is
largely affected by the deposits). There is no general
agreement as to whether the leakage currents as determined by
the counts of current pulses, cumulative charges or watts losses
are closely related to the degradation of the materials.
Watts loss due to leakage currents over the external
housings of the polymer arresters is nowhere near the level of
economic significance as the authors imply. For a typical
13.8 kV system, an analysis shows that the watts loss of the
polymer housed arresters at a very worst case portrayed in the
paper is only a very small portion of the conductor losses. As
an example, calculations indicated that at a load current level
of 400 A, the conductors (0.1663 Wmile) can account for up
to 80 kW/3Q, mile, while the arrester housings, assuming 24
arresters/3@mile, at worst only account for 216 W/3@ mile,
or less than 0.3 % of the conductor losses. These losses are
further reduced since the housings only conduct significant
currents when they are wet.
It is interesting to see in Figs. 7, 8 and 9 that the watts loss
of all three arresters reached a maximum value at about 5 to 6
“years”, although the polymer housing materials of the three
arresters are basically different. It could be assumed that the
test parameters or the test sequence had been changed during
this period of time.
Based on the comments above, I would like the authors to
respond to the following questions:
1. How do the authors defme the aging of the polymeric
materials? What physical or chemical indications of
polymer aging were evidenced in the tests performed?
2. How do the authors substantiate the relationship between
the aging of polymer housings and the leakage current
(watts loss) measurements?
3. Do the authors have any data which would indicate how
closely the test measurements of external housing watts
loss compare with what would occur in field service?
4. Could the authors give an explanation why the watts loss
reached a maximum value at almost the same period of
time for the three arresters in Figs. 7 to 9?
References
[11 Reference 6 of the paper.
[2] T. Zhao and R. A. Bernstorf, “Ageing Tests of Polymeric
Housing Materials for Non-Ceramic Insulators,’’ IEEE
Electrical Insulation Magazine, Vol. 14, No. 2, MarcWApril
1998, pp. 26 - 33.
[3] R. S . Gorur, E. A. Cherney, and R. Hackam, “A
Comparative Study of Polymer Insulating Materials Under
Salt-Fog Conditions,” IEEE Transactions on Electrical
Insulation, Vol. EI-2 1, No. 2, April 1986, pp. 175 182.
[4] A. J. Phillips, D. J. Childs, and H. M. Schneider, “Water
Drop Corona Effects on Full-scale 500 kV Non-Ceramic
Insulators,” IEEE Transactions on Power Delivery, PE-235PWRD-O-01- 1998.
[SI A. J. Phillips, D. J. Childs, and H. M. Schneider, “Aging
of Non-Ceramic Insulators due to Corona from Water
Drops,” IEEE Transactions on Power Delivery, PE-236PWRD-O- 11- 1998.
-
~
David R. Miller (Cooper Power Systems): The authors
wish to thank Mr. Zhao for his opinions and questions.
With reference to the term “aging” of polymer housed
surge arresters, there appears to be some confusion as to its
meaning. When dealing with organic materials, and only
materials and not devices, material scientists would
certainly agree with Mr. Zhao’s clear definition as a
“change in chemical composition, electrical and
mechanical properties”. However, when dealing with
devices, such as a polymer housed surge arrester, power
947
system personnel are not so concerned as to the changes in
chemical composition as to how the device will continue to
perform. Therefore, in response to the first two questions,
aging of polymer housed surge arresters is correctly
defined by increased watts loss, either internally or
externally or both, external flashover, or internal flashover.
For this reason, a detailed examination of the housings or
physical or chemical tests of the housing materials were not
part of the scope of this investigation.
In response to the third question, although we do not
have any field data on watts loss on surge arresters, we are
of the opinion that hydrophobicity is never totally lost in
service on SR arresters.
Regarding the final question, we cannot explain the
anomaly in the watts loss results which shows a maximum
value at almost the same period of time for the three surge
arresters in Figs. 7 through 9. Additional tests are in
progress and when these results are published at a later
date, we will attempt to answer this question.
Manuscript received March 29, 1999.
3