Experiences with Commissioning Testing of HV &

EHV Cable Systems: The Influence of Voltage Level

and Duration for Identifying Life Limiting Defects
M. Fenger
Kinectrics Inc.
800 Kipling Avenue, Unit #2, KL 204
Toronto, ON – M8Z 6C4
CANADA
[email protected]

Abstract- Over the last decade and a half, thousands of kilometres Partial discharges are a high frequency phenomenon.
of installed XLPE HV and EHV cable systems have been subjected Fundamentally, whenever a partial discharge occurs internal to
to after-laying commissioning testing prior to energization. The a cable section or a cable joint, a charge – and consequently a
commissioning test usually consists of a combination of AC HiPot high frequency current - is induced on the cable conductor
& Partial Discharge (PD) testing. PD testing, in particular, has
(high voltage electrode) and the cable shield (grounded
gained acceptance as a valid diagnostic tool for condition
assessment of cable insulation. It is well known and understood electrode). The magnitude of a measured partial discharge
that the results obtained from a partial discharge test depend not signal depends partly on the magnitude of the partial discharge
only on the conditions under which the test was performed but also current itself, i.e. the higher the actual partial discharge current
on the test equipment itself including the type of sensor used and its the higher the induced currents, and partly on the radial
location. Relying on experienced gained with AC Withstand and proximity of the partial discharge location relative to the cable
PD commissioning testing of more than 2,100 km of HV & EHV conductor, i.e. the closer to the cable conductor, the higher the
cable systems the paper examines the influence of voltage source induced current on the conductor [2, 3]. The relationship
and withstand duration of the withstand test for successfully between the induced charge on the high voltage electrode and
identifying life limiting defects in newly laid HV & EHV Cable
the actual discharge itself may be evaluated by the lambda-
systems both in terms of breakdown and also in terms of detection
of partial discharge sources from the same defects. The paper function [2, 3]. The rise-time of the induced partial discharge
further provides statistical summary of tests performed on more current is similar to the frequency of the actual partial
than 2100 km of HV and EHV cable systems including failure rates discharge current itself whereas the fall time of the induced
of accessories and PD occurrence rates. The paper also provides current is dependent on the impedance of the measuring
several case studies of PD detected in cable accessories during AC system itself [2]. In this case, the impedance of the cable
HiPot commissioning testing. system is defined not just by the series impedance of a PD
sensor and monitor but also in the impedance of the part of the
I. INTRODUCTION cable system a given PD pulse must travel through prior to
being detected. The rise-time of the partial discharge current
Partial discharges occur in the bulk of high voltage itself depends on path and velocity of the partial discharge
insulation materials where local electrical field conditions are (avalanche) and, consequently, the frequency depends
sufficiently high to sustain PD activity. In the case of primarily on (1) the strength of the electrical field in the void
extruded cables (EPR or XLPE cables) partial discharges (the higher the strength of the electrical field the higher the
typically occur in cavities at the conductor shield, cavities in velocity of the avalanche itself, the faster the rise time of the
the insulation due to shrinkage or gas-formation, near defects PD current and the higher the frequency of the PD current)
in the insulation shield, near loosely bound solid particles in and (2) the size of the void relative to the direction of the
the insulation, at protrusions, at splinters or fibers or near electrical field (the longer the void, the longer the duration of
contaminants in the insulation shield. In cable joints or the PD pulse, the longer the rise time of the PD current and the
terminations, partial discharges typically occur along dielectric lower the frequency of the PD current). Also, local conditions
interfaces, along stress interfaces, in cavities near the in the void such as gas pressure, temperature, void surface
conductor or insulation shield due to, for instance, conductivity has an impact on the partial discharge behavior.
misalignment during installation or thermal movement as a
result of normal operation. Finally, partial discharges may In addition, as the induced PD currents propagate through the
also occur within the cable insulation itself around cable towards the cable ends, they are subjected to primarily
mechanically degraded spots and or impurities resulting in the attenuation and, to a lesser extent, dispersion [10]. In
formation of electrical trees. addition, it should be noted that the magnitude of high
frequency currents propagating along the coaxial cable line
II. ACTUAL, INDUCED AND APPARENT CHARGE

978-1-4673-0487-0/12/$31.00 ©2012 IEEE 242

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on January 21,2021 at 18:01:34 UTC from IEEE Xplore. Restrictions apply.
further deteriorates as a result of impedance mismatch due to
joints and cross bonding. Q inj V inj 1
K= = C inj = V inj C inj (2)
It can thus be intuitively seen that for shorter cable runs V meas V meas V meas
induced currents as a result of partial discharge activity may
be readily detected via a terminal measurement, i.e. via a This approach involves three major assumptions:
capacitive or inductive sensors connected to the conductor or
shield at the end of a cable. For longer cable runs, attenuation A. The test objective is a pure, lumped capacitance
will prevent the measurement of induced currents related to B. The duration of the injected pulses is such full
PD activity occurring from the far end of the cable and thus a polarization is achieved (for the injection capacitor)
distributed PD measurement must be performed. A distributed C. All of the charge on the capacitor is transferred to the
PD measurement refers to the scenario where sensors are terminal of the cable under test.
connected to joints and splices throughout the length of the
cable. Often, external clamp-on radio frequency current With respect to (A): No cable objective is purely capacitive.
transformers are placed around the bonding lead connecting a Even for shorter lengths of cable systems, for instance
joint to the link box (cross-bonding box). Also, in some subjected to type tests in the laboratory, the inductive
instances, joints may be equipped with integrated partial component plays a factor with respect to HF pulse
discharge sensors. Lastly, it was previously concluded that a propagation.
framework for performing a meaningful sensitivity assessment
on a field installed transmission class cable would fulfill the With respect to (B): If a capacitor is driven with a time-
following criteria [11]: varying voltage that changes rapidly enough, then the
polarization of the dielectric cannot follow the signal. As an
1. Does not attempt to correlate discharge pulses measured example of the origin of this mechanism, the internal
in mV or mA to charge associated with the partial microscopic dipoles contributing to the dielectric constant
discharge current itself. cannot move instantly, and so as frequency of an applied
2. Does not rely on averaging to increase the signal-to-noise alternating voltage increases, the dipole response is limited
ratio. and the dielectric constant diminishes. Thus, at higher
3. Does not rely on narrow band-pass measurements to frequencies, the change in dielectric constant, referred to as
increase the signal-to-noise ratio. dielectric dispersion, should be considered.
4. Does inject pulses that have rise times and magnitudes
similar to the currents induced by typically partial With respect to (C): If the impedance of the cable under test is
discharge pulses. significantly higher than the impedance of the injection
capacitor, nearly full charge transfer is achieved. If not, the
In response, it is suggested that a sensitivity assessment be equivalent circuit acts as a voltage divider and the charge
based on so-called “sensitivity by construction” via laboratory division needs to be considered accordingly.
experiments involving injecting known, fast partial discharge
like signals into a cable assembly using matched impedance With respect to (4): PD is a high frequency phenomenon. It is
and measuring the response at a joint and a termination with a well understood that the typical waveform injected on the
sensor similar in design and manufacture to that used in the terminal of a cable assembly is slower and of higher
field for the same joint and termination design [13]. magnitude than a typical partial discharge pulse.

III. THE MEASUREMENT OF APPARENT CHARGE Lastly, In addition, it should be understood that an IEC 60270
style calibration relates only to the apparent charge on the
It is well understood that IEC 60270 applies to a narrow-band, cable terminal for a pure, lumped capacitive test object. Due
low frequency partial discharge measurement performed on a to attenuation of high frequency pulses traveling through
HV component acting as a lumped capacitance [10]. Voltage coaxial conductors, the apparent charge on the sheath or
signals may be injected via a coupling capacitor thus conductor at the discharge spot will be higher than the
providing a relationship between actual, detected pulse measured apparent charge on the terminal [10, 11].
magnitude as measured in mV and apparent charge as
measured in pC. When injecting a voltage of magnitude Vinj Lastly, an installed sectionalized transmission class cable
through a coupling capacitor of value Cinj, the charge injected, constitutes a distributed capacitive load and not a lumped
Qinj, into the system can be calculated via the well known capacitive load. For a lumped capacitance, it can be assumed
capacitor formula that the difference in apparent charge as measured on the
terminal and apparent charge at the location of the PD source
Qinj = Vinj ⋅ C inj (1) is minimal. For a distributed capacitance, the difference
between apparent charge as measured on the terminal and
apparent charge at the location of the PD source will be further
with the calibration constant K being calculated as
amplified due as a result of signal losses due to stray

243

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on January 21,2021 at 18:01:34 UTC from IEEE Xplore. Restrictions apply.
 Where RPD is the restive input impedance of the PD monitor
(typically 50 Ohms) and t1 and t2 define the beginning and end
time of the peak associated with the apparent PD pulse itself.
t1 t2
To preserve the waveform of a PD signal thus preserving
information on the type and nature of a PD source, modern
partial discharge systems make use of broadband acquisition
QP technology and not narrow band technology. Consequently,
IEC 60270 does not apply for field measurements performed
on installed transmission class cables with broad band partial
discharge measurement systems.

Figure 1: Example evaluation of apparent charge Previously, it was shown that any a assessment on any system
provide only for a normalization of the sensor and partial
capacitances, grounding etc. Thus, for distributed capacitive discharge monitor as defined by
systems, such as field installed cables, to ensure optimal H PD.Sensor ( s ) ⋅ H PD.Monitor ( s ) as loaded into the test
sensitivity, PD measurements should be performed at each impedance and does not provide any meaningful information
accessory. Thus, the apparent charge may be measured at or on the sensitivity of the PD measurement itself [12].
very close to the PD source location itself. Should a system Furthermore, it was shown that for broadband PD
allowing for measurement and storage of individual signal measurements, the frequency response of the system should be
waveforms the apparent charge of a PD signal may be characterized for the entire frequency domain for which the
evaluated by integrating up the appropriate area underneath PD sensor and PD instrument is able to detect signals. In
the PD signal - See Figure 1 other words, the injection pulses used for the sensitivity
t2
1 assessment should match the frequency response of the sensor,
Q PD = ∫R
t1 PD
VPD (t ) (4) the frequency response of the partial discharge monitor used
and match the frequency content of partial discharge pulses
from known partial discharge sources [12].

START Evaluate PD Results END

+ AC HiPot +
250kV (On-Site)
END

YES • Notify CLIENT of

PD
+ PD Activity, PD
NO Activity Level, PD Stability
PROC. C2
YES PROC. A
Increase
Voltage to Cont.
250kV AC HiPot
NO Internal MAYBE
END OF ACTION Additional Testing
YES NO PD
Cont. Lower AC
Remaining YES
Voltage by
AC HiPot 500v/sec +
END PROC. B

Repeat START Meas Evaluate PDIV, PDEV

Adjacent PDEV Evaluate PD Stability
Joints PDIV Evaluate Pattern

PD Meas. • Notify Client of Kinectrics

@ PDEV- o PDIV & PDEV Deliver to Client)
Memo
5kV o PD Level Report
PD Meas.
@ PDEV-
5kV Increase
Voltage to DECISION:
CLIENT
PDIV +
5kV

PROC. C1

Figure 2: PD Decision Tree

244

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on January 21,2021 at 18:01:34 UTC from IEEE Xplore. Restrictions apply.
 Rat. PD But No PD But No
AC HiPot Only AC HiPot & PD No of Failures
Voltage Failure Failure
Failures
[kV] Phases Term. Joints Phases Term. Joints Term. Joint Phases1 Joints Term. Joint
66/69/72 0 0 0 54 108 0 0 1 0 0.0% n/a 0.9% 0.0%
110/115 12 12 69 25 62 48 1 0 0 2.7% 0.9% 0.0% 0.0%
132/138/161 67 134 333 44 76 161 2 1 1 1.8% 0.4% 1.3% 0.6%
220/230 15 30 210 153 270 1,338 14 13 11 8.3% 0.9% 4.8% 0.8%
345/400 12 24 60 36 112 894 1 4 0 2.1% 0.1% 3.6% 0.0%
Table 1: Failure Rates & PD Occurrence Rates for HV & EHV Cable Systems. 1 Phases refers to terminated and
jointed cable conductors.

between cable systems subjected to AC HiPot testing only and

Therefore, taking these factors into account PD magnitudes cable systems subjected to a combination of AC HiPot & PD
are not diagnostic of nature and a given discharge level as testing.
evaluated in mV or pC does not correlate directly to the health
of the insulation system under test. As a consequence, from a Table 2 shows the average test level for each category of cable
scientific point of view, an acceptance criterion for cable ratings. As well, Table 2 shows the distribution of phases,
system should not be based on a PD level alone but on a terminations and joints for each cable rating category. As can
number of additional criteria – see Section IV. be seen, 132/138 kV and 220/230 kV constitute more than 50
% of the accessories tested and, thus, the failure statistics and
IV. TESTING PROCEDURE AND ACCEPTANCE PD occurrence rates obtained on these two categories of cable
CRITERIA systems are likely the most representative.

For the 15 years or so, a combination of AC HiPot & Partial As can be seen from Table 1 and 2 a number of observations
Discharge Testing has gained ground as an after-laying can be extracted:
acceptance test for XLPE cable systems. Cables are subjected
to an AC HiPot withstand in according to IEC 60840 and IEC (1) No failures of terminations have been observed during any
62067. Often, during the AC HiPot test, a partial discharge of the commissioning tests performed.
test is performed. For transmission class cable systems, due to
the issues related to attenuation, reflection at joints and signal (2) The PD occurrence rates for joints is approximately similar
loss due EM coupling, a distributed PD test is often for cable systems rated 132 kV to 230 kV and the failure rate
performed. is within the same range. This suggests AC HiPot testing alone
only detects about half of all life limiting defects in newly laid
Table 1 provides a quantitative summary of AC HiPot & PD cable systems and therefore addition of partial discharge
tests performed on a XLPE cable population rated 115 kV and monitoring technically adds to the validity of an after laying
above containing approximately 2,100 km of XLPE cable commissioning testing.
rated 69 kV and above, 418 individual phases, 3,113 joints and
828 terminations is given in Table 1. All tests constitute (3) The failure and PD occurrence rates for 345/400 kV cable
commissioning tests performed on newly laid cable. systems is significantly lower than for cable systems rated 230
Maintenance tests on field aged cable systems are not included kV and below. Also, the test levels for cable systems rated
in this survey. For this study, only partial discharge tests 230 kV and below are significantly higher than those for
performed at over-voltage are included. For some cable 345/400 kV cable systems. This suggests the test levels used
systems, partial discharge testing was performed after a for 345kV/400 kV cable systems are not high enough for
successful AC HiPot test at rated line-to-ground voltage (U0). identifying life limiting defects in said cable systems.
For this study, these cases are treated as a withstand-test only.
As can be seen from Table 1, the database distinguishes (4) The number of accessories for 66-72kV and 110/115kV
class cable systems is
likely not sufficient to Rat. Voltage Workmanship Issue
Rat. Voltage Avg. Test Level % Population
form any statistical [kV] Phases Accessories
[kV] [U0] Phases Term. Joints trends. 66/69/72 1.9% 0.9%
66/69/72 1.73 13% 13% 0% Assuming the 110/115 2.7% 0.5%
110/115 1.90 9% 9% 4% individual components 132/138/161 3.6% 0.6%
132/138/161 1.60 27% 25% 16% delivered prior to 220/230 22.6% 2.1%
220/230 1.69 installation have
40% 36% 50% 345/400 10.4% 0.5%
passed factory quality
345/400 1.11 11% 16% 31%
control and assuming Table 3: AC HiPot Test Level vs.
Table 2: AC HiPot Test Level vs. Cable Rating the factory quality Cable Rating

245

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on January 21,2021 at 18:01:34 UTC from IEEE Xplore. Restrictions apply.
 1 2 3 4 5 7 8 9 10 11 12 13 14 15 16
Cable Rating [kV] 220.0 220.0 220.0 230.0 230.0 138.0 138.0 400.0 230.0 220.0 132.0 225.0 225.0 225.0 220.0
Rated Volt. U0 [kV] 127.2 127.2 127.2 132.9 132.9 79.8 79.8 231.2 132.9 127.2 76.3 130.1 130.1 130.1 127.2
Test Level [kV] 216.0 216.0 216.0 180.0 180.0 128.0 136.0 260.0 226.0 216.0 130.0 221.0 221.0 221.0 221.0
Test Level [U0] 1.7 1.7 1.7 1.4 1.4 1.6 1.7 1.1 1.7 1.7 1.7 1.7 1.7 1.7 1.7
Test Current [Amp] 109.2 109.5 n/a n/a n/a 41.5 33.2 140.5 126.0 n/a 53.8 n/a 110.6 76.4 84.8
Test Freq. [Hz] 38.87 38.77 n/a 84.10 84.10 30.32 40.30 23.81 35.24 n/a 23.75 n/a 39.26 39.31 43.15
Failure Volt. [kV] 216 216 216 180 180 128 136 260 226 171 103 171 221 151.2 186.8
Failure Volt. [U0] 1.7 1.7 1.7 1.4 1.4 1.6 1.7 1.1 1.7 1.3 1.3 1.3 1.7 1.2 1.5
Time to Failure [sec] 70 367 n/a 1,260 8 1,260 1,851 n/a 38 n/a 9 n/a 3,480 n/a n/a
No of Cycles1 2,721 14,229 n/a 105,966 673 38,203 74,595 n/a 1,339 n/a 214 n/a 136,625 n/a n/a

Failure ID [Cable,
Joint Joint Joint Joint Joint Cable Joint Cable Joint Joint Joint Joint Joint Joint Joint
Joint, Termination]

Table 4: Failure Statistics. (1 No of cycles at AC HiPot Voltage).

control has been effective in identifying life limiting defects it therefore vary from test to test depending on the capacitive
can be assumed the failures and PD occurrence rates load of the cable under test. A more accurate measure of time-
experienced during on-site testing constitute installation to-failure may thus be number-of-cycles to failure. As can be
related defects. Table 3 provides a statistical break-down of seen from Figure 4, the 50 percentile for cycles to failure is
installation related defects per accessory (terminations plus 26,216 cycles and therefore 50 % of the failures experienced
joints) for the various classes of cable systems. As can be occur within the first 26,216 cycles. This cumulative
seen, the number of accessories subjected to installation distribution function for cycles-to-failure thus indicates the
related defects range from 0.5% to 2.1 %. It should be importance of test duration: To effectively identify
considered that 220/230 kV class cable systems constitute the installation related life limiting defects a large number of
majority of the database and, thus, the statistical breakdown cycles, at over-voltage, should be applied to the cable system.
for this cable class is likely the most representable. Based on Furthermore, when comparing the curve shapes for the
the data provided in Table 4, the Cumulative Probability cumulative probability density functions in Figure 4 it can be
Density Functions for Time-to-Failure and Cycles-to-Failure seen the shape of the density function for cycles-to-failure
may be calculated and plotted (see Figure 4). As can be seen, shows linear proportionality between percentage failures and
the 50-percentile for Time-to-Failure is 219 seconds and thus number of cycles after the 40th percentile and shows no
50% of the failures experienced occur within the first 5 indications of an asymptotic value towards the 100th percentile
minutes of the AC HiPot tests. However, the AC HiPot tests as the number of cycles goes towards infinity. As the majority
included in this study are performed primarily with fixed of data consists of tests performed on long cable systems, at
inductance, variable frequency test sets and the test frequency thus at lower frequencies, the data presented here may suggest
that the set time duration of 60 minutes may need to be
10,000 1,000,000 adjusted upwards when testing at lower frequencies (higher
loads) in order to effectively identify all life limiting defects in
an HV and EHV cable system.
100,000
1,000
Table 4 provides statistics on the individual failures
10,000
experienced. As can be seen from Table 4, all failures
Time to Failure [Sec]

Cycles to Failure

experienced thus far occurred at over-voltage (note, one data

100 1,000 point is missing as data was unavailable). This suggests soak
testing, i.e. a 24 hour test at rated-line-to-ground voltage is
100 inadequate for detecting life limiting defects in transmission
10 class cable accessories.
Sec to Failure 10

Cycles to Failure
Of the 17 failures experienced, 10 occurred at the withstand
1 1
level whereas 7 occurred while ramping up to the withstand
20 30 40 50 60 70 80 90
voltage. For the failures which occurred at withstand voltage,
Test Frequency the number of cycles to failure varies from 214 cycles to
136,625 cycles. When plotted against test frequency it can be
Figure 3: Time-to Failure and Cycles-to-Failure vs. seen there is no correlation between time-to-failure or cycles-
Frequency

246

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on January 21,2021 at 18:01:34 UTC from IEEE Xplore. Restrictions apply.
to-failure relative to the test frequency – see Figure 3. This is magnitudes, an evaluation of the severity of a PD source is not
expected as the time-to-failure will depend on the type and determined on the measured PD level alone. As can be seen
nature of the specific defect as well as its location within the from Figure 2, when a PD source is detected, the Partial
accessory relative to the cable conductor and outer sheath. Discharge Inception Voltage (PDIV) and Partial Discharge
Extinction Voltage (PDEV) are measured. In addition, phase
For the PD tests performed at over-voltage, these fall in to two plots of the source are obtained so the PD magnitude and
categories: (A) Joint-Hopping and (B) Continuous Monitoring. intensity can be quantified. Furthermore, the AC HiPot is
Joint-Hopping refers to the case where PD testing is allowed to continue and thus the dynamic behavior of the PD
performed periodically at over-voltage using mobile PD test source can be assessed and in most cases quantified.
crews. Continuous Monitoring refers to the case where all
accessories are monitored continuously during the ramp-up to Though no standard for PD pass/fail criteria exists, the guiding
the AC HiPot Test, during the AC HiPot Test and during the principle for the tests performed on the cable systems as part
Ramp Down for the AC HiPot Test. of this study has been that the accessories should be PD free –
to the sensitivity level of the PD test – at the voltage test level.
When PD testing was performed, the testing followed the However, for the 220/230 kV cable systems, in three out of ten
decision schedule outlined in Figure 2. In the absence of instances where PD activity was detected in the joint or
clearly defined PD pass/fail criteria, when PD is detected, the termination, the accessory was not replaced but conditionally
decision process with respect to component replacement accepted by the end client. In all of these cases, the PDIV was
involves discussion between the various stakeholders (end higher than 1.4U0 and the PDEV was higher than 1.2U0.
client, installer, equipment manufacture, test service provider Furthermore, in these instances, follow-up maintenance testing
and third party consultants). Due to the issues related to PD was mandated by the end client 3-5 years after
commissioning. In all other instances where PD was detected,
the accessory affected was replaced or rebuilt and a full AC
1.1
Percentile: Time-to-Failure [Sec] HiPot & PD retest was performed.
1

0.9 It should be noted that Joint Hopping limits PD measurements

0.8
to detection of PD sources which are present at the time of
measurement only. Some PD sources are periodically present.
0.7
Thus, the PD occurrence rate for Joint Hopping may be
0.6 understated. Furthermore, in response to client specific test
0.5 specs, in some cases, PD measurements are performed after
0.4
successful AC HiPot testing at an over-voltage of 1.2U0
following preconditioning the cable system at 1.4U0. In this
0.3
case, PD sources having PDIVs above 1.4U0 will not be
0.2 detected. It is thus possible the PD occurrence rate may be
0.1 understated in for these cases. Considering the PD occurrence
Time-to-Failure [sec] rates derived from this study consists primarily of PD data
0
0 300 600 900 1,200 1,500 1,800 2,100 2,400 2,700 3,000 3,300 3,600 3,900
obtained via Joint Hopping, it is likely that the PD occurrence
rates are understated. Continuous PD monitoring at all
1.1
Percentile: Cycles-to-Failure accessories naturally provides for the most effective PD test.
1

0.9 As the majority of PD testing for the database studied

0.8
presently consists of data obtained with Joint Hopping a
statistical analysis of the PD presently does not provide for
0.7
significant insight. However, this will change within the next
0.6 number of years as the majority of cable systems tested is now
0.5 tested using daisy-chained PD systems allowing for
continuous PD monitoring of all accessories during the AC
0.4
HiPot. An update to the database containing statistically
0.3 significant trends on PD acquired on HV and EHV accessories
0.2 data is expected in two years time. However, in the following,
0.1 examples of PD data acquired during commissioning testing
Cycles-to-Failure are provided.
0
0 20,000 40,000 60,000 80,000 100,000 120,000 140,000 160,000

Figure 4: Cumulative Distribution Functions for Time-to-

Failure (top) and Cycles to Failure (bottom)

247

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on January 21,2021 at 18:01:34 UTC from IEEE Xplore. Restrictions apply.
 Segment Wav ef orm
150
Ch 3 Segment Waveform
100
150
50
V olts (mV)

100 Ch 3
0
50

Volts (mV)
-50
0
-100
-50
-150
-100
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time uSec -150

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Time uSec

Figure 5: Example of PD detected on a 345 kV Cable

Termination

An example of a PD source detected from a 345 kV

termination during an AC HiPot test is given in Figure 5. The
AC HiPot test level was set to 250 kV as per IEC 62067. As Figure 6: Example of PD Measurements on a 220 kV
can be seen negative and positive polarity discharge pulses XLPE joint at 153 kV (1.2U0)
having magnitudes of approximately -52 mV and +62 mV can
be found centered at 45° and 225° phase angle with reference
to the phase-to-ground test voltage. The frequency content of pC and +4.2 pC respectively thus passing the end-clients 5 pC
the 1st moment for the signal activity measured was acceptance level and the joint was left in place. The joint has
been in service for approximately 24 months and no in-service
approximately 2 MHz. The PD source occurred
approximately 22 minutes into the AC HiPot and was failure has been observed.
thereafter present. The PDEV was measured to be
approximately 220 kV. A standard calibration as per IEC Lastly, of the cable systems involved with this study, four
60270 was performed on this termination yielding a failures have occurred within 5 years of passing an AC HiPot
calibration constant of 0.81 pC/mV thus yielding an apparent PD test. Three failures have occurred on 220/230 kV rated
systems. Of these failures, two failures were related to third
charge of 42 pC and 51 pC respectively. However, evaluating
the apparent charge based on the waveform captured yields a party external damage occurring during operation whereas the
discharge level of approximately -1,100 pC and +1,500 pC. root cause of the third failure was not positively identified but
Similar differences have been observed on 220/230 kV cables evidence pointed towards thermo-mechanical movement of a
systems. The significant difference relates to the issues stress cone within a joint. One in-service failure has occurred
on a 400 kV rated cable system operated at 345kV line
outlined in Section III.
voltage. The failure occurred within a cable section itself.
Another example of PD detected in a 220 kV XLPE joint is The cause of the failure has not been disclosed though third
provided in Figure 6. In this particular case, partial discharge party damage was ruled out.
measurements were performed following a successful AC
HiPot at 216 kV (1.7U0) for 60 minutes. Partial discharge IV. CONCLUSIONS
measurements were performed at 153 kV (1.2U0) following
conditioning of the cable system at 178 kV (1.4U0) for 15 Experiences with commissioning testing of approximately
seconds. 2,100 km of HV & EHV cable systems rated 66 kV to 400 kV
has been provided. The database covers a total of 418
As can be seen from Figure 6, negative and positive polarity individual phases, 3,113 joints and 828 terminations. A
quantitative representation of the database was discussed and
PD pulses can be found centered around 45° and 225° degree
the following can be concluded:
phase angle with reference to the phase to ground voltage.
The PDIV and PDEV were measured to 153 kV and 145 kV
• No failures of terminations have been observed during any
respectively. Thus, PD activity may occur intermittently
during on-line conditions in the event of a transient voltage of the commissioning tests performed.
wave being applied to the system. A standard calibration as
per IEC 60270 was performed as per the end-client’s test • The PD occurrence rates for joints is similar and the joint
protocol on this joint. The results yielded a calibration failure rates is in the same range for cable systems rated
constant of 0.19 pC/mV thus yielding an apparent charge of -3 132 kV to 230 kV. This suggests AC HiPot testing alone
only detects about half of all life limiting defects in newly

248

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on January 21,2021 at 18:01:34 UTC from IEEE Xplore. Restrictions apply.
 laid cable systems and therefore addition of partial Conduction and Breakdown in Solid Dielectrics, pp. 119 – 122, 1998,
1998
discharge monitoring technically adds to the validity of an
[8] S. Boggs & R. J. Densley, ”Fundamentals of partial discharge in the
after laying commissioning testing. context of field cable testing” IEEE Electrical Insulation Magazine”,
Vol. 16, No. 5, pp. 13 – 18, Sept.-Oct. 2000
• The failure and PD occurrence rates for 345/400 kV cable [9] IEC 60270, ”High Voltage Test Techniques- Partial Discharge
Measurements” December 2008
systems are significantly lower than for cable systems rated [10] N. Oussalah, Y. Zebboudj & S. A Boggs, “Partial Discharge Pulse
230 kV and below. Also, the test levels for cable systems Propagation in Shielded Power Cable and Implications for Detection
rated 230 kV and below are significantly higher than those Sensitivity”, IEEE Electrical Magazine, Vol 23. Issue 6, pp.. 5 – 10,
for 345/400 kV cable systems. This suggests the test levels Nov/Dec 2007.
[11] M. Fenger, H. G. Sedding, “Sensitivity Assessment for PD
used for 345 kV/400 kV cable systems are not high enough Measurements on Transmission Class Cables”, Proceedings of the
for identifying life limiting defects in said cable systems. 2007 JiCable Conference, A7.1, 2007.
It is suggested the test level for 345 kV/400 kV cable [12] M. Fenger, “Sensitivity Assessment for PD Measurements on
systems be a minimum of 1.7U0. Transmission Class Cables”, Proceedings of the 2008 ISEI, 2008.
[13] M. Fenger, J. Levine “Sensitivity Assessment for HV & EHV Field
Partial Discharge Measurements via Laboratory Testing.”, Proceedings
• The number of accessories for 66-72 kV and 110/115kV of the 2012 ISEI, 2012.
class cable systems is likely not sufficient to form any
statistical trends for these two voltage classes of cables.

• The data presented here may suggest that the set time
duration of 60 minutes may need to be adjusted upwards
when testing at lower frequencies (higher loads) in order to
effectively identify all life limiting defects in an HV and
EHV cable system.

• The PD occurrence rates referenced in this study are likely

understated as the majority of PD data relates to data
acquired using the Joint Hopping methodology

• Continuous PD monitoring of all accessories using a daisy-

chained PD system provides for a more reliable PD test.

• Though no standard exists for Partial Discharge

Commissioning Testing including pass/fail PD levels the
guiding criterion is that the cable system should be free of
partial discharge activity at the AC HiPot test level.

REFERENCES
[1] L.A. Dissado & J.C. Fothergill, “Electrical Degradation and
Breakdown in Polymers”, IEE Publications, Peter Peregrinus Ltd.
London, United Kingdom
[2] A. Pedersen, G. C. Crichton and I. W. McAllister, “The Theory and
Measurement of Partial Discharge Transients”, IEEE Trans on
Dielectrics and Electrical Insulation, Vol 26, No. 3, pp. 487-497, 1991
[3] Pedersen, G.C. Crichton and I. W. McAllister, “PD Related Field
Enhancement in The Bulk Medium”, Gaseous Dielectrics VII, Plenum
Press, New York, 1994.
[4] A. Pedersen, G. C. Crichton and I. W. McAllister, “PD-Related
Stresses In The Bulk Dielectric And Their Evaluation”, Annual Record
– CEIDP, pp.474-480, 1993
[5] A. Contin, A. Cavallini, G.C. Montanari, F. Puletti, “A novel technique
for the identification of defects in stator bar insulation systems by
partial discharge measurements”, IEEE ISEI, pp. 501-505, Anaheim,
USA, April 2000.
[6] F. Puletti, F. Ombello, M. Albertini, A. Cavallini, G.C. Montanari,
“Improved diagnostic tools for cable accessories by digital PD
detection systems", Conf. On High Voltage Plant Life Extension, pp.
2.2.1-2.2.9, Linkebeek, Belgique, November 2000.
[7] Z. Nadolny, J. M. Braun, & R. J. Densley, “Investigation of partial
discharge pulse shapes occurring at interfaces in model joints”,
Proceedings of the 1998 IEEE 6th International Conference on

249

Authorized licensed use limited to: UNIVERSIDADE DE SAO PAULO. Downloaded on January 21,2021 at 18:01:34 UTC from IEEE Xplore. Restrictions apply.