Test Procedure and Acceptance Criteria for PD

Commissioning Testing of Transmission Class Cables

M. Fenger
Kinectrics Inc
800 Kipling Avenue, KL 204
Toronto, Ontario M8Z 6C4 - CANADA

Abstract: Over the last decade, partial discharge testing has current itself, i.e. the higher the actual partial discharge current
gained acceptance as a valid diagnostic tool for condition the higher the induced currents, and partly on the radial
assessment of cable insulation. It is well known and understood proximity of the partial discharge location relative to the cable
that the results obtained from a partial discharge test depend not conductor, i.e. the closer to the cable conductor, the higher the
only on the conditions under which the test was performed but
induced current on the conductor [2, 3]. The relationship
also on the test equipment itself including the type of sensor used
and its location. The issues related to actual discharge, induced between the induced charge on the high voltage electrode and
charge and measured apparent charge are also well understood. the actual discharge it self may be evaluated by the λ-
Yet, for commissioning testing of solid dielectric, test function [2, 3]. The rise-time of the induced partial discharge
specifications often reference magnitude levels in pC of apparent current is similar to the frequency of the actual partial
charge as the only pass/fail acceptance criterion. The practice is discharge current itself whereas the fall time of the induced
based on factory acceptance testing of individual cable current is dependent on the impedance of the measuring
components. This paper outlines the difference between PD system itself [2]. In this case, the impedance of the cable
testing performed on individual components in the factory and
system is defined not just by the series impedance of a PD
PD testing performed on installed systems in the field with
respect to magnitude calibration of PD pulses. The paper further sensor and monitor but also in the impedance of the part of the
proposes a test procedure as well as a framework for acceptance cable system a given PD pulse must travel through prior to
criteria for after-laying PD commissioning testing of transmission being detected. The rise-time of the partial discharge current
class cable and provides test experience gained through 5 years of itself depends on path and velocity of the partial discharge
testing of HV and EHV cable systems. (avalanche) and, consequently, the frequency depends
primarily on (1) the strength of the electrical field in the void
I. INTRODUCTION (the higher the strength of the electrical field the higher the
velocity of the avalanche itself, the faster the rise time of the
Partial discharges occur in the bulk of high voltage insulation PD current and the higher the frequency of the PD current)
materials where local electrical field conditions are and (2) the size of the void relative to the direction of the
sufficiently high to sustain PD activity. In the case of electrical field (the longer the void, the longer the duration of
extruded cables (EPR or XLPE cables) partial discharges the PD pulse, the longer the rise time of the PD current and the
typically occur in cavities at the conductor shield, cavities in lower the frequency of the PD current). Also, local conditions
the insulation due to shrinkage or gas-formation, near defects in the void such as gas pressure, temperature, void surface
in the insulation shield, near loosely bound solid particles in conductivity has an impact on the partial discharge behavior.
the insulation, at protrusions, at splinters or fibers or near
contaminants in the insulation shield. In cable joints or In addition, as the induced PD currents propagate through the
terminations, partial discharges typically occur along dielectric cable towards the cable ends, they are subjected to primarily
interfaces, along stress interfaces, in cavities near the attenuation and, to a lesser extent, dispersion [10]. In
conductor or insulation shield due to, for instance, addition, it should be noted that the magnitude of high
misalignment during installation or thermal movement as a frequency currents propagating along the coaxial cable line
result of normal operation. Finally, partial discharges may further deteriorates as a result of impedance mismatch due to
also occur within the cable insulation itself around joints and cross bonding.
mechanically degraded spots and or impurities resulting in the
formation of electrical trees. It can thus be intuitively seen that for shorter cable runs
induced currents as a result of partial discharge activity may
II. ACTUAL, INDUCED & APPARENT CHARGE be readily detected via a terminal measurement, i.e. via a
capacitive or inductive sensors connected to the conductor or
Partial discharges are a high frequency phenomenon. shield at the end of a cable. For longer cable runs, attenuation
Fundamentally, whenever a partial discharge occurs internal to will prevent the measurement of inducted currents related to
a cable section or a cable joint, a charge – and consequently a PD activity occurring from the opposite cable and thus a
high frequency current - is induced on the cable conductor distributed PD measurement must be performed. A distributed
(high voltage electrode) and the cable shield (grounded PD measurement refers to the scenario where sensors are
electrode). The magnitude of a measured partial discharge connected to joints and splices throughout the length of the
signal depends partly on the magnitude of the partial discharge

978-1-4244-6301-5/10/$26.00 @2010 IEEE

cable. Often, external clamp-on radio frequency current alternating voltage increases, the dipole response is limited
transformers are placed around the bonding lead connecting a and the dielectric constant diminishes. Thus, at higher
joint to the link box (cross bonding box). Also, in some frequencies, the change in dielectric constant, referred to as
instances, joints may be equipped with integrated partial dielectric dispersion, should be considered.
discharge sensors.
Qinj = Vinj ⋅ Cinj ( f ) (3)
Furthermore, it was previously concluded that a framework for
performing a meaningful sensitivity assessment on a field
installed transmission class cable would fulfill the following With respect to (2): If the impedance of the cable under test is
criteria [11]: significantly higher than the impedance of the injection
capacitor, nearly full charge transfer is achieved. If not, the
1. Does not attempt to correlate discharge pulses equivalent circuit acts as a voltage divider and the charge
measured in mV or mA to charge associated with the division needs to be considered accordingly.
partial discharge current itself.
2. Does not rely on averaging to increase the signal-to- In addition, it should be understood that an IEC 270 style
noise ratio. calibration relates only to the apparent charge on the cable
3. Does not rely on narrow band-pass measurements to terminal. Due to attenuation of high frequency pulses
increase the signal-to-noise ratio. traveling through coaxial conductors, the apparent charge on
4. Does inject pulses that have rise times and the sheath or conductor at the discharge spot higher than the
magnitudes similar to the currents induced by measured apparent charge on the terminal [10, 11].
typically partial discharge pulses.
Lastly, an installed sectionalized transmission class cable
constitutes a distributed capacitive load and not a lumped
III. THE MEASUREMENT OF APPARENT CHARGE capacitive load. For a lumped capacitance, it can be assumed
that the difference in apparent charge as measured on the
It is well understood that IEC270 applies to a narrow-band, terminal and apparent charge at the location of the PD source
low frequency partial discharge measurement performed on a is minimal as only attenuation along of a shorter length of
HV component acting as a lumped capacitance [10]. Voltage cable needs to be considered. For a distributed capacitance,
signals may be injected via a coupling capacitor thus the difference between apparent charge as measured on the
providing a relationship between actual, detected pulse terminal and apparent charge at the location of the PD source
magnitude as measured in mV and apparent charge as will be further amplified due as a result of signal losses due to
measured in pC. When injecting a voltage of magnitude Vinj stray capacitances, grounding etc. Thus, for distributed
through a coupling capacitor of value Cinj, the charge injected, capacitive systems, such as field installed cables, to ensure
Qinj, into the system can be calculated via the well known optimal sensitivity, PD measurements should be performed at
capacitor formula each accessory. Thus, the apparent charge may be measured
at or very close to the PD source location itself. Should a
Qinj = Vinj ⋅ Cinj system allowing for measurement and storage of individual
(1)
signal waveforms the apparent charge of a PD signal may be
evaluated by integrating up the appropriate area underneath
with the calibration constant K being calculated as the PD signal
t2
1
K=
Qinj
=
V inj
C inj = V inj C inj
1
(2)
Q PD = ∫R PD
V PD (t ) (4)
V meas V meas V meas t1

This approach involves two assumptions:

1. The duration of the injected pulses is such full t1 t2

polarization is achieved (for the injection capacitor)
2. All of the charge on the capacitor is transferred to the
terminal of the cable under test.

With respect to (1): If a capacitor is driven with a time- QP

varying voltage that changes rapidly enough, then the
polarization of the dielectric cannot follow the signal. As an
example of the origin of this mechanism, the internal
microscopic dipoles contributing to the dielectric constant
cannot move instantly, and so as frequency of an applied Figure 1: Example evaluation of apparent charge
Where RPD is the restive input impedance of the PD monitor
(typically 50 Ohms) and t1 and t2 define the beginning and end Rated
No of Failure PD But no
time of the peak associated with the apparent PD pulse itself. Voltage
Accessories Rate failure
[kV]
To preserve the waveform of a PD signal thus preserving 115 18 5.56% 0.00%
information on the type and nature of a PD source, modern 132/138 254 0.79% 0.00%
partial discharge systems make use of broadband acquisition
220/230 1,239 1.78% 0.81%
technology and not narrow band technology. Consequently,
IEC270 does not apply for field measurements performed on 345/400 972 0.35% 0.31%
installed transmission class cables with broad band partial Table 1: Failure & PD Occurrence Rates
discharge measurement systems.
For this study, only 9 individual phases (18 terminations, 156
Previously, it was shown that any a assessment on any system joints) were not subjected to PD testing. For the remaining
Provide only for a normalization of the sensor and partial phases, PD testing was performed during the AC HiPot on 24
discharge monitor as defined by individual phases immediately following a successful AC
HiPot and at a reduced voltage level of 1.2U0. For the
H PD.Sensor ( s ) ⋅ H PD.Monitor ( s ) as loaded into the test remaining phases, partial discharge testing was performed on
impedance and does not provide any meaningful information all accessories during the AC HiPot.
on the sensitivity of the PD measurement itself [12].
Furthermore, it was shown that for broadband PD For 132/138kV and 220/230kV class cables, AC HiPot testing
measurements, the frequency response of the system should be was performed at 1.7U0 whereas for the 345kV/400kV class
characterized for the entire frequency domain for which the cables AC HiPot testing was performed at 1.25U0 and 1.2U0
PD sensor and PD instrument is able to detect signals. In respectively – where 1U0 corresponds to rated line-to-ground
other words, the injection pulses used for the sensitivity voltage. As can be seen, the failure rates for the 132/138kV
assessment should match the frequency response of the sensor, and 220/230kV class cables are somewhat comparable. The
the frequency response of the partial discharge monitor used failure rates for the 345kV/400kV class cables are
and match the frequency content of partial discharge pulses significantly lower. This is likely related to the comparatively
from known partial discharge sources [12]. lower test voltage level for test circuits. The relatively high
failure rates for the 115kV class cables are very likely not
Therefore, taking these factors into account PD magnitudes representative as the cable population here is relatively small.
are not diagnostic of nature and a given discharge level as
evaluated in mV or pC does not correlate directly to the health Due to the issues outlined in Sections II and III, the PD tests
of the insulation system under test. As a consequence, from a on all cable circuits were performed as distributed PD
scientific point of view, an acceptance criterion for cable measurements using broadband PD monitors having a
system should not be based on a PD level alone but on a frequency span from DC to 400MHz. Furthermore, when PD
number of additional criteria – see Section IV. testing was performed, the testing followed the decision
schedule outlined in Figure 2. In the absence of clearly
IV TESTING PROCEDURE & ACCEPTANCE CRITERIA defined PD pass/fail criteria, when PD is detected, the decision
process with respect to component replacement involves
For the 15 years or so, a combination of AC HiPot & Partial discussion between the various stakeholders (end client,
Discharge Testing has gained ground as an after-laying installer, equipment manufacture, test service provider and
acceptance test for XLPE cable systems. Cables are subjected third party consultants). Therefore, it is important general
to an AC HiPot withstand in according to IEC 60840 and IEC guidelines have been agreed to prior to testing. Due to the
62067. Often, during the AC HiPot test, a partial discharge issues related to PD magnitudes an evaluation of the severity
test is performed. For transmission class cable systems, due to of a PD source is not determined on the measured PD level
the issues related to attenuation, reflection at joints and signal alone. As can be seen from Figure 2, when a PD source is
loss due EM coupling, a distributed PD test is often detected, the Partial Discharge Inception Voltage (PDIV) and
performed. Partial Discharge Extinction Voltage (PDEV) are measured.
In addition, phase plots of the source are obtained so the PD
Table 1 provides a quantative summary of AC HiPot & PD magnitude and intensity can be quantified. Furthermore, the
tests performed on a XLPE cable population rated 115kV and AC HiPot is allowed to continue and thus the dynamic
above containing approximately 1,570 km of XLPE cable behavior of the PD source can be assessed and in most cases
rated 115kV, 267 individual phases, 2,588 joints and 534 quantified.
terminations is given.
 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.
250k V AC HiPot
NO Internal MAYBE
END OF ACT ION Additional Testing
YES NO PD
Cont. Lower AC
Remaining YES
Voltage by
AC HiPot 500v/sec +
END PROC. B

Repeat Meas Evaluate PDIV, PDEV

START Evaluate PD Stability
Adjacent PDEV
Joints PDIV Evaluate Pattern

PD Meas. • Notify Client of Kinectrics Deliver to Client)

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

PROC. C1

+
Figure 2: PD Decision Tree
Though no standard for PD pass/fail criteria exist the guiding 22 minutes into the AC HiPot and was hereafter present. The
principle for the tests performed on the cable systems as part PDEV was measured to approximately 220kV. A standard
of this study has been that the accessories should be PD free – calibration as per IEC 60270 was performed on this
to the sensitivity level of the PD test – at the voltage test level. termination yielding a calibration constant of 0.81pC/mV thus
However, for the 220/230kV cable systems, in three out of ten yielding an apparent charge of 42pC and 51pC respectively.
instances were PD activity was detected in joint or However, evaluating the apparent charge based on the
termination, the accessory were not replaced but conditionally waveform captured yields a discharge level of approximately -
accepted by the end client. In all of these cases, the PDIV was 1,100pC and +1,500pC. Similar differences have been
higher than 1.4U0 and the PDEV was higher than 1.2U0. observed on 220/230kV cables systems. The significant
Furthermore, in these instance, follow up maintenance testing difference relates to the issues outlined in Section III.
was mandated by the end client 3-5 years after
commissioning. In all other instances where PD was detected, Lastly, of the cable systems involved with this study, two
the accessory affected was replaced or rebuilt and a full AC failures have occurred within 5 years of passing an AC HiPot
HiPot & PD retest was performed. PD test. One failure was related to third party external
damage occurring during operation. The fault of the second
An example of a PD source detected from a 345kV failure was not positively identified but evidence pointed
termination during an AC HiPot test is given in Figure 3. The towards thermo-mechanical movement of a stress cone within
AC HiPot test level was set to 250kV as per IEC 62067. As a joint.
can be seen negative and positive polarity discharge pulses
having magnitudes of approximately -52mV and +62mV can
be found centered at 45° and 225° phase angle with reference
to the phase-to-ground test voltage. The frequency content of
the 1st moment for the signal activity measured was
approximately 2MHz. The PD source occurred approximately
 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
Conduction and Breakdown in Solid Dielectrics, pp. 119 – 122, 1998,
Segment Wav ef orm

150 1998
100 Ch 3
150
Segment Waveform
[8] S. Boggs & R. J. Densley, ”Fundamentals of partial discharge in the
50
context of field cable testing” IEEE Electrical Insulation Magazine”,
V olt s (mV)

100 Ch 3
0
50
Vol. 16, No. 5, pp. 13 – 18, Sept.-Oct. 2000
Volts (mV)
-50
0

[9] IEC 60270, ”High Voltage Test Techniques- Partial Discharge

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

-150
Measurements” December 2008
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Time uSec
[10] N. Oussalah, Y. Zebboudj & S. A Boggs, “Partial Discharge Pulse
Propagation in Shielded Power Cable and Implications for Detection
Figure 3: Example of PD detected on a 345kV Cable Sensitivity”, IEEE Electrical Magazine, Vol 23. Issue 6, pp.. 5 – 10,
Termination Nov/Dec 2007.
[11] M. Fenger, H. G. Sedding, “Sensitivity Assessment for PD
Measurements on Transmission Class Cables”, Proceedings of the 2007
JiCable Conference, A7.1, 2007.
V. CONCLUSIONS [12] M. Fenger, “Sensitivity Assessment for PD Measurements on
Transmission Class Cables”, Proceedings of the 2008 ISEI, 2008.
Previously, the framework for PD Sensitivity Assessment
Procedure for broadband PD measurements performed on
transmission class cable systems was outlined [12]. The
procedure has been used on a population of cable system for
the past 5 years. In the absence of a standard for PD testing of
field installed HV & EHV Cable systems a PD measurement
procedure as outlined in Figure 1 has been in use. To date, PD
the procedure has been effective in avoiding in-service
failures.

Activities for tracking the service performance of the cable

systems being part of this study is on-going and it is expected
an update will be produced in 5 years which will then involve
10 years of operational experience a population of cable
systems subjected to a combination of AC HiPot & PD
acceptance testing.

VI. ACKNOWLEDGMENT

The author would like to acknowledge Silvano Rizzetto and

Mark Credland of Kinectrics for their help in acquiring the
data presented.

VII. REFERENCES
[1] L.A. Dissado & J.C. Fothergill, “Electrical Degradation and Breakdown
in Polymers”, IEE Publications, 1
[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