Sei sulla pagina 1di 9

IEEE Transactions on Power Delivery, Vol. 9, No.

2, A p d 1994
EFFECTS OF VOLTAGE SURGES ON EXTRUDED DIELECTRIC CABLE LIFE
PROJECT UPDATE

611

by
Richard A. Hartlein, Member
Georgia Power Company
AtlanG, Georgia

V. S. Harper, Member
Georgia Power Company
.
Atlanta, Georgia

KEYWORDS
Cable, Lightning, Water Tree,Impulse, Thumper
ABSTRACT
Electric utility engineers have commented [I], [2] that
extruded distribution cables frequently fail during or shortly
after a thunder storm. These engineers also comment that
failures often reoccur on cable circuits where previous
failures were located with a thumper. Linemen at Georgia
Power often make similar comments.
To investigate this observation, crosslinked W P E ) and tree
retardant crosslinked (TRXLPE) cable designs were
subjected to accelerated water treeing tests. Samples were
subjected to simulated lightning surges or simulated
thumping surges. Crosslinked cables removed after 15 years
of service operation were also subjected to these surges.

The results show that, in some cases, lightning surges do


reduce extruded distribution cable life. Also, high level
thumping surges appear to reduce cable life once cables are
well aged.

program were published in [3] and [4]. Those results


provide data on cables aged in the laboratory and subjected
to 40 kV, 70 kV and 120 kV lightning surges. The data also
include laboratory aged cables which were subjected to a
25 kV thumper surge to simulate surges used to locate field
failures. Aging times of up to 600 days were reported.
To gather additional information on the effects of surges on
aged cable, a second phase was added to the test program.
In this phase, aging was continued on selected cables beyond
600 days, crosslinked polyethylene cables aged in service
(XLPEF) were added to the test program and a 25 kV
lightning surge magnitude was also introduced.
To minimize time and cost, only XLPE, TRXLPE and
XLPEF cables were included in the second phase. These
compounds were selected because they represent the majority
of cable insulations used by electric utilities.
This paper presents the results gathered to date on the cables
tested in phase two. This new data provides greater insight
into the effects of lightning and thumping surges on aged
cables. A brief review of the test program is also included.
TEST CABLES

INTRODUCTION
The Electric Power Research Institute (EPRI) sponsored the
work to investigate the effects of voltage surges on extruded
dielectric cable life under Project RP2284-01. Since voltage
surges may affect insulation materials differently, five
commonly used insulation types were initially chosen for this
test program.
These include high molecular weight
polyethylene (HMWPE), tree retardant high molecular
weight polyethylene (TRHMWPE) and ethylene propylene
rubber (EPR) as well as XLPE and TRXLPE.
Preliminary results and a detailed description of the test
A paper recommended and approved
93 SM 357-4 PVRD
by the IEEE Insulated Conductors Committee of the IEEE
Power Engineering Society for presentation at the IEEE/
PES 1993 Summer Meeting, Vancouver, B.C., Canada. July
18-22, 1993. Manuscript submitted January 4, 1993;
made available for printing April 22, 1993.

PRINTED IN USA

Harry Ng, Member


Electric Power Research
Institute
Palo Alto, California

Fifteen kV cables were used because they are very common


and 15 kV cables are easily managed in accelerated
laboratory tests. Since the ac voltage and surge voltage
stress distribution varies with insulation thickness, cables
with different insulation thicknesses were evaluated. Table
1 outlines the cables utilized for this program.
The XLPE and TRXLPE cables were made by one
manufacturer specifically for use in this test program. This
was done to minimize the variations in cable quality that can
occur between manufacturers. They were manufactured to
the AEIC CS5-82 specification for crosslinked polyethylene
insulated power cable using a triple extrusion, steam curing
process. The conductor shields were extruded, conventional,
semiconducting XLPE with a nominal thickness of 15 mils.
The nominal insulation thickness was either 175 mils or 220
mils. The insulation shields were extruded, semiconducting
XLPE with a nominal thickness of 30 mils. Extra smooth
or extra clean shields were not available at the time these
cables were manufactured.
Since the XLPEF cable was obtained from the field, the

0885-8977/94/$04.00 Q 1993 IEEE

612
manufacturing process is unknown and only one insulation
thickness was available. Given the age of the cable, it was
very liketp made using a two pass, steam cure process.

superimposed on the negative peak of the 60 Hz ac aging


voltage.

All of the cables had a 1/0 AWG, Class B, 3/4 hard,


stranded aluminum conductor. The metallic shields were
concentric, bare copper wires. Conductor moisture blocking
material and overall jackets were not used.
Table 1
Test Cables

Insulation Material

Insulation Thickness

--------T

TIME

XLPE
XLPEF'
TRXLPE

175mils and 220 mils


175 mils
175 mils and 220 mils

I This cable obtained from the field after 15 years of service


at 7.6 kV to ground on the Tampa Electric System in Tampa
Florida.

IMPULSE WAVE SELECTION AND APPLICATION


Lightning Surge Selection
A detailed discussion of the process used to select and apply
the impulse waves used in this test program was presented
earlier [4]. Lightning surge magnitudes were based on the
surge levels which can be imposed on a 15 kV cable system.

The 25 kV, 40 kV and 70 kV impulses are well defined


square waves. The 120 kV impulse is a square wave that
starts with a high frequency decaying waveform on the front
of a 70 kV square wave. The oscillating nature of the 120
kV impulse is typical of a field cable with an open circuit
end. A 200 p s pulse width was chosen for all lightning
surges to represent the long pulse durations which can occur
in service. A 200 ns risetime was chosen for all the
lightning surges to represent the very fast rise wavefronts
which are also known to occur in service. A plot of the
70 kV waveform is shown in Figure 1.

50 MlCROSECONOYDIVlSION

Figure 1. 70 kV x 200 p lightning surge applied to


cables during aging test on the negative peak
of the 8.7 kV, rms,60 Hz ac voltage.
(25 kV and 40 kV surges differ only in
magnitude)

In all lightning surge applications, the voltage impulse was


referenced to ground. Therefore, when the surge occurred,
the applied voltage rapidly changed from -12.3 kV, the
negative peak of the ac sine wave, to the positive absolute
value of the applied surge. Thus, cables subjected to the
25 kV surge experienced an immediate change in voltage
from -12.3 kV to +25 kV, a total voltage change of
37.3 kV. Cables subjected to the 40 kV surge experienced
a voltage change of 52.3 kV. Cables subjected to the
70 kV surge experienced a voltage change of 82.3 kV.
Cables subjected to the 120 kV surge experienced a voltage
change of 132.3 kV followed by subsequent oscillations
associated with the 120 kV surge.
A termination imperlance equal to the characteristic
impedance of the test cable was used to minimize reflections
and, therefore, control the wave shape during the 25 kV,
40 kV and 70 kV surges. This termination was removed to
produce the 120 kV oscillating surge.
Specialized equipment was built to provide the required
lightning surges. Both the surge magnitude and the surge
width were variable and controllable.

Lightnine Surne ADdication


Fault Location rrhumoer) Surge Selection
Lightning surges most often occur in the field while the ac
system voltage is present. Therefore, it was considered
mandatory that the lightning impulses be superimposed on
the normal 60 Hz ac waveform. This was accomplished by
discharging a surge capacitor through gas filled gaps and
metal oxide arrester blocks onto the energized 60 Hz aging
test bus. To make this test as realistic as possible, the ac
test voltage was reduced from 3 X rated voltage to ground
to 1 X rated voltage to ground during the surge application.
To simulate the reverse polarity stress conditions which can
occur in service, the fast-rise, positive polarity surges were

Underground cable faults are frequently located using a


voltage surge generated by a capacitive discharge device
commonly called a thumper. This impulse is injected into
the faulted cable and causes an arc at the cable fault. Utility
workers then locate the fault by using a variety of acoustic
and electromagnetic detection devices. The characteristics
of the impulse varies depending on the thumping equipment
used, the circuit configuration and worker techniques. To
simulate worst case conditions, a relatively large magnitude
thumping surge was used with many repetitions.

613
Thumuer Surne Auulication

r
I

To apply the thumping surge, a thumper was connected to


each cable rack individually using a dummy cable length
which contained a simulated fault to ground. A Biddle
Model No. 652025 thumper with a 12 pf capacitor was used
to apply the thumping surge. A typical waveform at the
open end of the test cable is shown in
Figure 2.

Figure 2. Typical waveform as seen at the open (far)


end of the cable subjected to a 25 kV
thumping surge.

applied, the voltage was lowered to 1X the operating voltage


to ground (8.6 kV). This was also done to represent field
conditions.
Aging of Cables Removed From the Field
The XLPEF cables aged for 15 years in service at 7.6 kV.
An examination of the insulation revealed numerous, large
bowtie and vented water trees. The vented trees were as
long as 170 mils and the bowtie trees were as long as 60
mils. A C breakdown tests were conducted according to
AEIC CS5-82 starting at 100 V/mil.
The resulting
breakdown strength of five, 30 foot long samples was 220,
220,220,220 and 260 V/mil. Since cables with this type of
watertreeing and dielectric strength are considered to be well
aged, the aging voltage used in the laboratory was the same
magnitude as the service voltage (7.6 kV).
Initially, the XLPEF cables were aged in conduits filled with
tap water but with no water in the conductor. The conductor
of this cable was sealed immediately after receipt at the
laboratory. This was done to preserve field conditions as
much as possible. However, since no failures occurred after
aging in the laboratory for 200 days, deionized water was
added to the conductors. Deionized water was used to avoid
adding ions which may not have been present in service.
Aging Temuerature

ACCELERATED AGING PROCEDURE

Aeinn of New Cables


The cable aging test was designed to simulate and accelerate
field aging and was pattemed after the AEIC CS5-82, B.5
accelerated water treeing test which was in force when this
project was initiated. In accordance with the AEIC
accelerated water treeing test procedure, the XLPE and
TRXLPE cables were thermally preconditioned. This is
done to reduce the high concentration of volatiles contained
in the newly manufactured cable insulations.

Normally, the AEIC accelerated water treeing test is


performed with sufficient current to achieve a 90C
conductor temperature in air. However, in phase one, the
75C rated HMWPE cables were connected in series with
the 90C rated XLPE, TRXLPE and EPR cables. To
prevent overheating of the HMWPE cable and to maintain
the same aging conditions on all cables, a maximum
conductor temperature of 75 "C was chosen for all test
cables in phase one. For consistency, this temperature was
also used in the second phase of the test program.

This temperature was achieved with 260 amperes in the


conductor using a lZhour-on, 12-hour-off load cycle.

Preconditioning was conducted after the cables were placed


in the aging conduits but before they were filled with water.
AEIC specifies a preconditioning conductor temperature of
130 "C for 10 days using current in the conductor.
However, 110 "C was used in this test to avoid the
possibility of overheating the insulation. All temperature
tolerances were f 5 " C . After preconditioning, the
conductors and conduits were filled with tap water.

Aging With ImDulses


The lightning surges were applied an average of 2.3 times a
week. They were always applied at the end of the
12-hour-on period while the cables were still warm.
Approximately 5 to 10 minutes before the lightning impulse
was applied, the aging voltage was reduced from 26 kV to
8.7 kV.

To accelerate aging, 3X rated voltage to ground (26 kV, 60


Hz ac) was applied to both the XLPE and TRXLPE
175-mil-wall and 220-mil-wall cables. Thus, the aging stress
was 149 V/mil on the thin wall cables and 118 V/mil on the
thick wall cables. This procedure was followed to represent
the different stresses that thick and thin wall cables
experience in the field. When the lightning surge was

Thumping surges were applied in a manner that represents


field conditions. Cables used for the thumping test condition
were removed from the aging test and connected to the
thumper. Nine hundred thumping impulses were applied to
each test cable rack individually. The cables were then
placed back in the aging test. Thumping was performed

614
every 60 days after the first 120 days of aging.

Aging Failures

Control samples were exposed to the same aging conditions


as the surged samples but with no surges applied. All
samples were e n e r g i d , deenergized and filled with water
at the same time. Every effort was taken to ensure that the
only difference between the control and surged samples was
the application of the voltage surges.

To simplify the discussion, an abbreviation is used to


describe the test conditions. The insulation material and
wall thicknesses in mils is followed by the surge or control
test variable. For example, the XLPE, 175-mil-wall cable
subjected to a 70 kV surge is referred to as XLPE, 175/70.

Each surge test condition was applied to a separate set of


cable samples. No cables were subjected to both thumping
and lightning impulses. This would have been an interesting
condition to evaluate because it represents the kind of surge
combinations that can occur in service. However, individual
surge effects were considered the most important variables
to investigate first.

As mentioned earlier, at the start of this project ac


breakdown tests were considered an important diagnostic
tool for evaluating surge effects. Therefore, during the first
360 days of aging in the initial tests, numerous samples were
removed from the test and subjected to an ac breakdown
test. By the end of 360 days, almost no samples remained.
Thus, the 175/control and 175/70 surged samples of XLPE
and TRXLPE cables were unavailable for further aging
beyond 360 days.

Redicates
Two, 180-foot-long coils of cable were subjected to each test
condition. Each coil was a continuous length which
contained six, 30-foot-long samples. Thus there were 12
samples or replicates for each test condition.

Other conditions examined for these cables as well as the


XLPEF test cables started the test program after the data
showed that aging time is more important than breakdown
strength. Therefore, data are available for longer than 360.
Also, since unaged TRXLPE cable was still available, the
TRXLPE 175/control condition was repeated.

Summarv of Test Variables Investigated


Table 2 shows a matrix of the test conditions examined in
phase 2. A Cy) indicates that a test covering this condition
was performed. An (N) indicates that a test to evaluate this
condition was not performed due to time or space
limitations.
Table 2
Matrix showing all of the test conditions
reported in this paper.

To compare the time to failure data for all cables, line


graphs are employed which show when each failure
occurred. The data presented in this manner provide a
visual method of observing early failures or clusters of
failures. An "X" at the end of a line means either the test
condition was discontinued to perform breakdown tests or
most of the samples failed at the time the "X"is shown. No
"X" at the end of the line means the cable is still under test.
Except for the XLPEF cable, failures during the surge
application are rare. The failures as a function of time are
shown in Figures 3-5.

175 mil wall thickness


surge condition
CNTRL 25kV 4OkV 70kV 12OkV
XLPE
TRXLPE
XLPEF

Y
Y
Y

N
N
Y

Y
N
N

Y
Y
Y

Y
Y
N

220 mil wall thickness


surge condition
CNTRL 25kV 40kV 70kV 120kV
XLPE
TRXLPE
XLPEF

Y
N
N

N
N
N

N
N
N

Y
N
N

_ .............................

*....................... *..x ................

-!?5,M!! 4!??.Pu!?e.A
175 HII 70kv Pulse

.....................

175 MI1 ConlMl

_ .........................................
175 MI1 l2UV

They included ac breakdown tests, impulse breakdown tests,


time to failure and visual analysis. After a thorough
investigation, time to failure analysis was the only method
which revealed differences between control and surges
samples. Therefore, only time to failure data is presented
for the cables tested in phase two.

PUIS0

....y....A.A*.A ...A . y . . . A n........X .............


.........y .......* . n M M k x ......................
.....................................................................................
_ ....................................................y....... ....A .......*.x ...........................
220 MI1 7WV Pula8
_....
......................
.........
.....y..X .....
y

_ .....................................................
175 MI1 Thumper

RESULTS

In phase one, four traditional methods of evaluating the


integrity of an extruded power cable were used to determine
the effects of surges on each cable tested in the project.

A..A&.A. ..A
.........
.......A...)(......................................................
.

.......................................

N
N
N

.......... .... ....A

u . X

...........................

500

600

100

200

300

400

Oays of Aging

700

Figure 3. XLPE failures as a function of time in the aging


test. Each failure is represented by a triangle.

615

I75 Mil Cantmi

..............................

.........

to have a significant effect on the cable until approximately


450 days of aging. At that point, the cable failed rapidly
until all the samples were consumed.

..............A........
. & ...............................

&...&

175 Mil 70kv Pulse


.........................................
K............................................................................
175 Mli IZWv Pulse

........................................... ..A..
.......A..
..&

.75 Mil Thumper


............................................................

...........A..
..
..A.
.A
yX..
. ..............

...........

.....................................

.......................................................................................................................

200

400

600

Days of Aging

800

1000

Figure4. TRXLPE failures as a function of time in the


aging test. Each failure is represented by a
triangle.

175 Mil Cnnlmi

.........................................................................................................................
175 MI1 25 kv Pulse

...............................................

f................. 2..............*......*..& ................

175 Mil 70 kv Pulse


-. ......................................................

.i&.
....2......ai.
................................

Mil Thumper
-.175
...............................................................................

....................................
A..

For the 220 wall XLPE cables, the only apparent difference
between the control condition and the 70 kV lightning surge
condition is that failures occurred earlier on the samples
Interestingly, the
subjected to the lightning surge.
observation made for the 175 wall XLPE cable subjected to
the thumping surge also applies to the 220 wall XLPE cables
subjected to the thumping surge.
The control and 120 kV lightning surge TRXLPE cables
appear to experience similar failure rates. However, as
before, the samples subjected to the thumping surge fail
rapidly after they have aged for 450 days.
The effect of the 25 kV and 70 kV lightning surges on the
XLPEF cable is very pronounced.
Multiple failures
occurred on the samples subjected to lightning surges while
no control samples failed. Once again, there are no clear
differences between the 25 kV and 70 kV surge levels.
Interestingly, the thumper has not had a significant effect on
the cable life.
Unlike the XLPE and TRXLPE cables, the XLPEF cables
often failed when the lightning surge was applied. This
phenomena will be discussed in more detail later.
Statistical Analysis of Time to Failure Data

Figure 5 . XLPEF failures as a function of time in the


aging test. Each failure is represented by a
triangle.
Discussion of Time to Failure Data
Examination of the time to failure data begins to provide
interesting insight on surge effects. For the 175 wall XLPE
cables, it is difficult to compare the control condition to the
surge conditions because the control test was terminated so
early in the test program. Therefore, it is unclear how
strongly the lightning surges influenced cable life. There is
also no clear difference between the 40 kV and 70 kV
lightning surge conditions. However, comparing the 40 kV
and 70 kV lightning surge conditions to the thumper test
condition is useful.
The samples subjected to the lightning surges failed fairly
regularly throughout the test after the first 150 days of
aging. On the other hand, once the samples subjected to the
thumping surge started to experience a few failures, they
failed very rapidly. In fact, the thumper failure points on
the graph were artificially spread out to distinguish
individual failures. Thus, the thumping surge did not appear
~

To provide a more in depth analysis of the time to failure


data, Weibull probability statistics were employed. The
estimated failure rates for the XLPE, TRXLPE and XLPEF
cables using the Weibull model are presented in Figures 6 9. The test conditions are noted on the graphs following the
insulation material description. The abbreviations used are
similar to those employed for the aging failure graphs except
that a "0" represents the control condition.
The characteristic life for each test condition is the point at
which 63.2% of the samples have already failed. For
convenience, the 63.2% level is shown with a dotted line on
each graph. If fewer than three failures occurred for a given
test condition, a failure model could not be calculated.

. --

TIME, DAYS

Figure 6. Weibull life model for 175 wall XLPE cable.

616

Discussion of Statistical Model


I

/ /

F6r the XLPE and TRXLPE cable designs, there are slight
differences in characteristic life between test conditions.
However, in all cases, the confidence intervals overlap,

-50

/ 1 ,

,'k

TIME. DAYS

Figure 7. Weibull life model for 220 wall XLPE cable.

Figure 10. Weibull life model parameters. 90%


confidence intervals on characteristic life.
The calculated characteristic life values are
shown as a "*'

TIME, DAYS

Figure 8. Weibull life model for TRXLPE cable.

Figure 11. Weibull life model parameters. 90%


confidence intervals on shape factor. The
calculated shape factor values are shown as a

TIME, DAYS

"*U.

Figure 9. Weibull life model for XLPEF cable.


making distinctions between the characteristic lives of each
sample statistically inconclusive.

Confidence Intervals

To gain further insight from the Weibull statistical mdel,


the values for characteristic life (indicated as
with
corresponding 90 % confidence intervals are shown for each
test condition in Figure 10. The values for the Weibull,
shape parameter (also indicated as
with corresponding
90% confidence intervals are shown in Figure 11. %e
shape parameter gives an indication of bow the probability
distribution function changes with aging time.
I*")

I*")

h e XLPEF cable design does show a noticeable difference


in characteristic life between samples subjected to the 25 kV
lightning surge and those subjected to the 70 kV lightning
surge. Although there is a slight overlap in the confidence
intervals of the two test conditions, the amount is very
small. Thus the observation that the two test conditions
differ is statistically valid, indicating that lowering the surge

617
level from 70 kV to 25 kV increases cable life.
There is a distinct difference in the slope for the 220 wall
XLPE cable between the control and the 70 kV lightning
surge condition. This verifies the observation made on the
time to failure data that the samples subjected to the 70 kV
surge failed earlier than the control samples.
Another interesting observation for the XLPE and TRXLPE
Weibull curves is that the slope for the thumper test
condition is consistently higher than the slope for other test
conditions.
Although there is some overlap in the
confidence intervals, this trend indicates that the samples
subjected to the thumper surge very likely experience a wear
out phase. That is, after a period of occasional failures,
they start to fail rapidly, as if the cable reached the end of
its reliable service life.
FAILURE MECHANISMS
Several possible reasons for differences between control and
surged sample failure rates as well as differences in lightning
surge and thumping surge failure rates were discussed in [4].
However, failures of the XLPEF cables during the lightning
surge application is a newly observed phenomena. Almost
all of the other cable designs failed sometime after the surge
was applied. The different failure mode for the XLPEF
cable can be explained by the degree of water treeing in the
insulation.
The XLPEF cables age at 1X r a t d voltage which allows the
dielectric to experience greater deterioration before they fail
than the other cables which were aged at 3X rated voltage.
Thus the ratio of the surging voltage to aging voltage (degree
of deterioration) is larger for the cables operated at 1X than
for the cables operated at 3X rated voltage.
This phenomena is demonstrated by the very large bowtie
and vented water trees which formed while the XLPEF cable
was in service. Many of the trees found in the insulation of
this cable extend through as much as 97 9% of the insulation
wall. They continued to grow slowly in the aging test where
the applied voltage magnitude is 1X the operating voltage.
However, they easily become a failure path in the presence
of a large magnitude lightning surge.
The XLPE and TRXLPE cables started the test new, with no
water trees present. Since the aging voltage on these cables
was 3X the operating voltage, failures occurred in the aging
test long before the water trees reached the length observed
in the XLPEF cables. Since extremely long water trees are
not present in these cables, they are not nearly as susceptible
to failure during voltage surges.

CONCLUSIONS
Phase 2
The results of EPRI Project RP2284-01 have provided some
interesting insight into the effects of voltage impulses on
extruded cables aged in the laboratory. Several conclusions
drawn during the first phase of the project are presented in
[3] and [4]. Only new conclusions developed during
phase 2 are stated in this paper. They apply only to the
cables evaluated in the test program described in this paper.
They may or may not be applicable to cables operating in
service. Obviously, many questions remain unanswered.

1. Lightning Voltage Impulses Can Reduce Cable Life.


This observation was made in phase one and reinforced
in phase 2. The 175 mil and 220 mil XLPE cables, the
175 mil TRXLPE cables and the 175 mil XLPEF cables
which were surged often failed more than the same
cables that were aged without the voltage impulses.
2. Thumping Voltage Surges Can Reduce Cable Life.
This is a new observation. After 400 days of aging in
the laboratory, the XLPE and TRXLPE cables subjected
to a thumping surge failed more rapidly than the same
cables subjected the control or lightning surge condition.
This implies that once cables have been in service long
enough to develop water tree growth similar to that
obtained after 400 days of aging under the conditions
described in this paper, they may fail prematurely if
subjected to thumping surges.
3. Lighting Surge Magnitude Does Not Strongly Affect
Cable Failure Rates.
Only the XLPEF cables demonstrated a change in failure
rate as a function of surge magnitude. However, the
effect was not observed in the characteristic life, but in
the slope of the Weibull model curve.
REFERENCES
1. Minutes of the 78"' Meeting of the Institute of Electrical
and Electronics
Engineers Insulated Conductors
Committee Meeting, April 21-23, 1986. pp. V-C.
2. Minutes of the 80'' Meeting of the Institute of Electrical
and Electronics Engineers Insulated Conductors
Committee Meeting, April 27-29, 1987. pp. V-E2.
3. R. A. Hartlein, V.
Voltage Surges on
IEEE Transactions
Volume 4, Number

S. Harper and H. Ng. "Effects of


Extruded Dielectric Cable Life,"
on Power Delivery, April, 1989
2, pp. 829-841.

618

4. Electric Power Research Institute, "Effects of Voltage

Surges on Solid-Dielectric Cable Life",


6902, Interim Report, September, 1990.

Report EL-

ANOWLEDGEMENTS
The authors would like to thank the Electric Power Research
Institute for funding this research project. It addresses an
area that has concerned electric utility engineers for many
years.
The authors would also like to thank the following Georgia
Power Research Center personnel for their hard work and
dedication to accomplish the project goals.

- Mr.

Larry Coffeen who designed and constructed the


lightning surge test equipment.

- Mr.

Boyd Pettitt who processed the large volume of data


generated and developed many of the graphs and
maintained the surge test equipment.

In addition, the authors would like to thank Dr. Russell G.


Heikes of the Georgia Institute of Technology for providing
the statistical data analysis and the project advisor, Mr. P.
Pate1 for his suggestions and guidance.
BIOGRAPHIES
Richard A. Hartlein (M '80) was born in Atlanta, Georgia
on March 20, 1952. He received a Bachelor and Master of
Mechanical Engineering degrees in 1976 and 1982 from the
Georgia Institute of Technology in Atlanta, GA. He has
been employed at the Georgia Power Research Center since
1970 where he is currently manager of the mechanical
section. He has conducted test programs on extruded, thin
wall, 230 kV transmission cable, water tree resistance of
extruded cable designs, and power cable ampacity inside
riser shields. He has also managed EPRI projects on the
short-circuit characteristics of cable metallic shields.
He is a member of the IEEE Insulated Conductors
Committee. He is chairman of ICC Task Group 10-27
which writes the IEEE 404 splice testing standard. He is
also a past chairman of the Cable Engineering Section of the
Association of Edison IlluminatingCompanies and chairman
of the task group which writes the S I C CS5 specification
for XLPE insulated distribution cables.

V. S. H a m r (M '74) was born in Marietta, Georgia on


October 3, 1943. He received a Bachelor of Electrical
Engineering degree in 1966 from the Georgia Institute of
Technology in Atlanta, Georgia. He has been employed
with the Georgia Power Company since 1966 and is
presently a Research Manager. Previous fields of work
include substation control, power line relaying, fault current
calculations and project engineering for customer
substations.

Mr. Harper is a member of the IEEE Power Engineering


Society, Insulated Conductors Committee and has served on
several EPRI task forces and committees.
H a m W. Hg (SM '74) received his B.S. degree in
Electrical Engineering in 1971 from the University of
Arizona, Tucson, Arizona. From 1971 to 1983 he was with
Tucson Electric Power Co. in Tucson, Arizona in various
distribution engineering positions, starting as a Distribution
Engineer. In 1976, he was appointed Supervisor of
Distribution Engineering.
In 1983 he joined the Electric Power Research Institute in
Palo Alto, California where he is presently Manager,
Distribution System Design and Operations, in the Electrical
Systems Division. His work has been in a broad area
related to the distribution of power by electric utilities.
Some of his responsibilities include distribution cables,
electrical fault location, wood pole management, distribution
design and analysis software, energy conservation and
amorphous steel core distribution transformers. He has
authored many papers related to power distribution.
Mr. Ng is a member of the IEEE Power Engineering
Society, the IEEE Insulated Conductors Committee, and the
American Wood Preservers Association. He serves on the
Wood Pole Working Group of the Distribution Committee of
IEEE, the research committee of the American Wood
Preservers Association and as advisor to the Forest Products
Laboratory, Mississippi State University. He is also a
Registered Professional Engineer (Electrical) in the states of
Arizona and California.

619
Diecussion
MARC AUDET, KILBORN INC. TORONTO, CANADA :
The authors have
reeented data on impulee
testing of XLPE RND EPR insulation under
laboratory conditione, and thie data seems
to indicate little difference in erformance
between the control samplee a n 8 the test
samples, although the teet eamples were
eub ected
to
impulee waves t o
simulate
ligitnin surges. It appeare from thie that
any furtxer teeting of similar cables under
laboratory
conditions with
bare
eam 1,s
immersed in water i s not liable to result in
much useful information.
The teat data presented by the authore on1
reinforces
reviouely
ubliehed data thaz
XLPE
E%R
ineulatea cables and cablee
ineulatnedd with similar compounds fail when
subjected to moieture, generally from water
treeing.
Various papers have been presented, which
provide information or teat data concerning
the performance of such cablee of varioue
CO f i U atione and compounds.
Some of thie
in?or#afion wae preeented as etatietical
data on cablee installed in the field. The
data indicates that all similar compounde,
whether TRXLPE or EPR provide good life if
the cablee are sealed against moieture at
the
shield.
and
at
the
conductor
termination.
It appears that future efforte in regard to
cab1
m r v ent s uld be dire ed tow
w'llg
provfdtn8 %ez?er cakne conf iguraelone
filled
conductors,
sealed
condinuoue
ehielding tapes or eimilar barriers againet
moisture
if long cable life is t o be
realized.
Some really useful information could have
been obtained if the teat eamplee had been
made up in such a manner ae to reflect
actual
field
conditions.
with
amroved
terminations
etrees
-relief
debices
rounded shielde, and impervious jackete:
8uch teete would yield really useful data
for
cable
suppliere,
s ecifiere,
and
installere. Perhape the facifitiee available
to the authore did not permit thie manner of
testing.
It would have been intereeting to know where
the test cablee failed. whether they f led
at or near the termina'tione due to vo?tage
streae, or somewhere in the middle of the
cables,
or
randomly.
Also,
were
the
compounde teated in any ray after failure,
to eee if any chemical change had occurred?

surge had a very noticeable effect on cable performance. In


Figure 3, the 175 mil and 220 mil cables clearly had more
failures earlier in the test when the 40 kV and 70 kV surges
were applied. In the case of the field aged XLPE cables, the
surged samples failed and the samples which were not
surged experienced no failures.

In many cases, Weibull statistics do show that the


characteristic life of a specific cable design may be similar
for control and surged conditions (Figures 6 - 9). However,
the slope of the Weibull curves in Figures 6 and 7 is very
different for control samples and surged samples. This is
direct evidence that there is indeed a difference between
control and surged samples.
This is demonstrated again in Figure 11 where the
confidence intervals are plotted for the Weibull shape factor.
For example, the there is a significant difference in the
calculated shape factor and the corresponding confidence
interval for the XLPE 220 control samples and the XLPE
220 samples subjected to the 70 kV lightning surge.

Test Cables

Since the scope of our work was to evaluate the effect of


lightning surges on aged underground cable, we used a very
common procedure of aging cables in an accelerated water
treeing test. They had unfilled, stranded conductors and no
jackets to allow for maximum accelerated aging. Water
impervious jackets and/or blocked conductors would have
significantly increased the time to achieve significant aging.
Other tests are currently underway in our laboratory to judge
the effects of jackets and moisture barriers on cable aging.
These results will be presented sometime in the future.
It is important to note that there is a considerable amount of
cable in service which is very similar to the cable in our
test. The results obtained in this test program should be
applicable to those cables.

Manuscript received August 9, 1993.

Test Procedure
RICHARD A. HARTLEIN, V. S. HARPER AND HARRY
NG: The authors would like to thank the discussor for the
opportunity to clarify and provide comments on several
aspects of the paper.

Suree Effects
Although there were some test conditions where the surge
appeared to have no effect, there were others where the

All test samples were constructed with standard terminations


as used in the field. All cable neutrals were grounded. All
reported test sample failures occurred in the cable within the
water filled tube. Termination failures were repaired and
not considered cable failures.
These procedures are
commonly used throughout the cable industry in accelerated
aging tests on extruded dielectric cables.
Manuscript received October 5, 1993.

Potrebbero piacerti anche