th
Draft, submitted to Jicable 99, '5 International Conference on Insulated Power Cables', Versailles, France
On-site partial discharge diagnosis and monitoring on HV power cables
Diagnostic et surveillance permanente de câbles haute tension, par mesures de
décharges partielles
Detlev W. Gross, Power Diagnostix Systems GmbH
Brüsseler Ring 95, 52064 Aachen, Germany
Email: gross@pd-systems.com
Résumé
L'utilisation croissante des matériaux à isolation
polymère dans les câbles électriques de 400kV et
plus, requiert des systèmes de mesures adéquats,
afin de dévoiler des défauts naissants et éviter des
claquages prématurés. Cet article décrit les mé-
thodes et techniques utilisées pour l'enregistrement,
l'analyse et la surveillance de l'activité de décharges
partielles sur de tels câbles en opération normale.
Les matériaux à isolation polymère tels que XLPE,
EPR et PE de première génération étant largement
utilisés dans les câbles de distribution, un nombre
considérable de vieux câbles en service nécessite
ainsi une technique de diagnostique fiable, pour
supporter la décision de remplacer un circuit de
câbles, ou encore de prolonger sa durée de vie utile.
Les méthodes pour la mesure de décharges
partielles in-situ, ainsi que les techniques de
couplage sont discutées en détail.
Introduction
Partial discharge activity is the most prominent
indicator of insulation defects and ongoing
degradation of high voltage insulation systems.
With transmission class cables on-site partial
discharge measurements on preferably embedded
sensors validate the proper installation of the pre-
fabricated and factory tested accessories.
Sets of modular instruments offer the assessment of
the partial discharge signals of all accessories of a
cable system with the first application of rated
voltage. The full information concerning all
accessories is available within minutes and a bad
joint, for instance, is clearly identified. Thus, the
costly installation of two joints is avoided in case, as
the faulty component is identified prior to breakdown.
Subsequently, these instruments can be used to
remotely monitor the cable system during the first
load cycles.
With distribution class cables the main concern
besides the verification of a new installation is to
identify weak cables and accessories of aged cable
systems. Therefore, mainly provisional and external
sensors are used to capture signals of partial
discharge activity. Compact and portable instruments
Abstract
The increased application of polymeric insulation
materials with power cables in the voltage range of
400kV and above requires enhanced measures to
unveil incipient failure and to avoid early breakdown.
The contribution describes methods and techniques
to register, analyze, and monitor partial discharge
activities with such cables in a field environment.
As polymeric insulation material such as XLPE, EPR,
and PE from the early days is widely applied with the
distribution class cables, a large population of aged
cables demands reliable diagnostic techniques to
support the decision regarding the replacement or
the continuation of the service of such cables. Field
applicable partial discharge measurement methods
and coupling techniques are discussed in detail.
offer convenient acquisition of the partial discharge
pattern. Further, coupling capacitor based partial
discharge fault location either at line frequency or
with the cable energized by a low frequency high
voltage source at 0.1Hz, for instance, is applicable.
Partial Discharge Sensors
In general, several physical principles are suitable to
capture partial discharge impulses occurring within a
cable or accessory [1-4].
In order to offer effective coupling as well as relia-
bility, the design of partial discharge sensors must
follow basic guidelines. A sensor must provide:
• No influence on the insulation properties of the
accessory to be monitored, i.e. no installation of
additional HV components weakening the relia-
bility of the overall system.
• High coupling efficiency to the partial discharge
signals.
• Noise rejection or immunity as integral property
of the sensor.
• Low cost of the sensor and effortless installation
or integration.
With pre-fabricated accessories capacitive sensors
fulfill these guidelines at best in case they are
integral part of the accessory. Alternatively, such
sensors can be integrated during the commissioning
of the cable system.
Fig. 1 shows an example of a capacitive sensor
utilizing the stress-cone (1) of a termination or a pre-
3 1
1 2
2
S S
4 4
Fig. 1, Stress-cone as capacitive sensor
molded joint [3, 8, and 9]. As shown with Fig. 1, a
metal mesh (2) is wrapped around the overlap of the
stress-cone (1) and the outer semi-conductive layer
of the cable (shown hatched). Subsequently, this
metal mesh is connected to the inner conductor of a
coaxial signal cable (3), while the signal cable's
shield is connected to the neutral of the HV cable.
With this example the stress-cone and the metal
mesh form a capacitive divider between HV
conductor (4) and the ground potential, i.e. the joint
box. The resistance of the cylindrical part of the outer
semi-conductive layer (S) loads this divider.
The ratio of the capacitive divider determines the
basic sensitivity. Depending on the frequency range
used a sensitivity of 1pC and below is easily
achievable with such sensor.
The high frequency properties of this type of sensor
differ with respect to the origin of the partial
discharge signals. For discharges occurring under-
neath the stress-cone a frequency response from DC
up to 100MHz and above is effective. For discharges
occurring with remote accessories both the high-
pass properties of the sensor and the dispersion of
the cable apply. The sensitivity against such signal
[dB]
0 transformer (Fig. 9) can be used to effectively couple
-20
-40
0.01 0.1 10 [MHz]
1
Fig.2, Frequency response with remote PD source
traveling along the cable follows in principle the
graph shown with Fig. 2, whereas the cable proper-
ties determine the upper cut-off and the sensor
design determines the lower cut-off.
Another example of an integrated sensor, shown
with Fig. 3, illustrates the utilization of the com-
pression sleeve of a composite type joint. With this
3
Fig. 3, Sensor connection with composite joint
type of accessory either that compression sleeve (2)
is isolated from the springs (not shown), or it
contains an isolated contact ring to avoid short-
circuiting the high frequency signals.
Both examples, as shown with Fig. 1 and 3, do have
in common, that there is no influence on the in-
sulation properties of the accessory, as the sensor
installation is hidden behind the power frequency
ground. The cylindrical resistor of semi-conductive
material (S) carries the capacitive current of the
stress-cone.
Generally, this embedded capacitive coupling
technique has been applied successfully to various
types of accessories in the range of 100kV rated
voltage and above. External capacitive sensors can
be used in case the embedded sensors are not
available. However, as this type of external foil
sensor is not shielded normally, additional measures
are required to obtain sufficient noise reduction.
With measurements on distribution class cables
external foil sensors and other provisional
installations are applicable in most cases, only.
Accessories of distribution class cables with
polymeric insulation are mostly not tightly shielded.
Thus, external foil sensors offer a good sensitivity,
even in case not used differentially.
Additionally, a clamp-on high frequency current
to partial discharge activities within a medium
voltage joint. Typically, these types of joints offer a
strong external high frequency field due to the not
evenly laid and often externally connected neutral.
Finally, piezoelectric sensors to capture the
ultrasonic signals of partial discharge activities are
applicable mainly with distribution class cables.
However, with transmission class cables the
dimensions of the accessories and the ultrasonic
properties of the materials used with the accessories
may isolate or obstruct ultrasonic partial discharge
signals.
 Pattern Acquisition System
The core data acquisition procedure is based on an
event-recording concept, the so-called phase
resolving partial discharge analyzer (PRPDA). The
PRPDA acquires three-dimensional distribution func-
tions with the coordinates pulse amplitude, phase
angle with respect to the high voltage signal, and
number of events. These distribution functions are
displayed as "color pictures" or fingerprints wherein
color (or gray-scale) represents the pulse repetition
rate. The partial discharges are recorded by a
commercial PRPDA system known as ICMsystem.
The basic partial discharge acquisition system con-
sists of:
• The signal coupler or sensor which is part of the
test object, usually two couplers can be combined
for gating or real time pulse rejection purposes.
• Signal conditioning modules, which adapt the in-
strument to the sensor and eventually transmit
the preprocessed signal to the data acquisition
system.
• Data Acquisition System or Partial Discharge
Pattern Recorder
• Flexible interfacing through fiber optics, modems,
or IEEE488 bus allows using the system for
various applications and locations.
• Graphical user-interface (GUI) to display and
analyze the data.
This paper presents results of measurements on
cables, only. However, similar techniques are
presently applied with testing and monitoring on
rotating machines, power transformers, and gas-
insulated switchgear (GIS), for instance [6, 7, 9, 10,
and 12].
Multiplexer Multiplexer
Analog Analog
Processing Processing
Pattern Pattern
Acquisition Acquisition
FO cable
Control & Control &
RAS RAS
Fig. 4, Configuration and actual view of the multi-channel multi-location pattern acquisition system
Multi-Point Acquisition
The degradation process with polymeric insulation
material in the voltage range of 400kV is irreversible
and often fast progressing once started. Therefore,
the identification of faulty components with the first
application of rated voltage is a main objective.
Avoiding an early breakdown simplifies the repair or
replacement of the defect component and further
avoids potential triggering of treeing with other
components due to the partly excessive over-
voltages caused by a breakdown.
Thus, commissioning measurements with polymeric
cables in the voltage range of 400kV demand an
acquisition system to collect, transmit, and display
the data of all accessories as fast as possible.
In order to comply with these requirements the basic
pattern acquisition system as described previously
was enhanced accordingly.
Figure 4 shows the configuration and an actual view
of such a multi-channel and multi-location acquisition
system. The multiplexer allows connecting to eight
sensors installed with each site, i.e. joint chamber or
manhole. Each input channel is isolated by means of
a high frequency isolation transformer according to
the requirements of the cross-bonding joints.
Further processing of the analog signals is made in a
frequency band of 2 to 20MHz. The wide-band
signal-conditioning amplifier used processes the
signal envelope [9] found with that frequency band.
Subsequently, the envelope's peak amplitude is
acquired. Along with the phase position of the
captured event, the partial discharge pattern is
generated. Control and communication circuits allow
to chain up to eight instruments, whereas each
instrument acts as fiber optic repeater. Thus, only
two multi-mode fibers are required to include another
instrument (Fig. 4).
 Fig. 5, Graphical user interface of the multi-channel pattern recorder.

Graphical User Interface

A control computer equipped with an optical interface

connected to the optically chained instruments
serves to display the acquired data. The graphical
user interface scans through the activated channels
and displays the acquired partial discharge pattern.
Small pattern displays for up to eight channels in
parallel are placed at the left and right hand side of
the central pattern display, which follows the scan
process or optionally remains on a specific channel
of interest. With the example given with Fig. 5
channel 3 shows up internal partial discharges of a
defect joint, while for the other channels the noise
floor is displayed, only. The software allows further Fig. 6, External discharges (due to hard rain)
controlling the setups of the instruments connected.
Additionally, a modem gateway offers remote access
of the instruments installed on site.
The acquired partial discharge pattern is directly
determined by the properties of the discharging site.
It reflects, for instance, the availability of the initial
electron to start the avalanche, the geometry of the
defect, the properties of the surfaces contributing to
the discharge process, and the properties of the gas
atmosphere involved [3, 5, 10, and 11].
As an example, Fig. 6 shows rain-induced dis-
charges with an outdoor termination nicely centered,
while with Fig. 7 both phase shift and orientation of
the pattern clearly identifies an ongoing electrical
treeing with polymeric material (PE). Fig. 7, Internal discharges (treeing)
 PD Fault Location
The location of partial discharges based on the
acquisition of the pulse train of the reflected traveling
wave requires a broadband coupling to the high
frequency partial discharge signals. Preferably, a
coupling capacitor provides this coupling. Alterna-
tively, a sensor of sufficient capacity embedded
within a termination reduces pickup of external
disturbances. Further, this method requires a straight
cable section, i.e., disconnected from the grid, as the
reflection pattern becomes very puzzling in case of a
T-stub included, for instance. The reflection factor of
r = –0.33 introduced by such a T-stub causes multi-
ple reflections and thus heavily hampers the inter-
pretation.
In order to provide a reflection factor of close to r =
+1, the impedance terminating the cable regarding
the frequency band of interest must be large
compared with the cable's high frequency im-
pedance mostly found in the range of 20O to 30O.
For PD fault location the cable is powered preferably
by a low frequency high voltage source or a resonant
test set offering a variable test voltage. Alternatively,
the grid can energize the cable. However, a power
inductor with suitable high frequency properties is
required to provide the appropriate reflection factor
and to isolate the cable section under test from the
disturbances coming from the grid.
Fig. 8, ICMcompact, panel of the fault location program
The compact partial discharge detector captures the
partial discharge signal with an embedded
100Msample analog to digital converter. Subse-
quently, a digital signal processor applies correlation
techniques and further numeric treatment of the
signals acquired in order to reduce the influence of
radio interference and to compensate the effects of
dispersion that the pulse undergoes while traveling
along the cable, for example.
The graphical user interface to control that instru-
ment (Fig. 8) offers with the left-hand side DSO-like
display the continuously refreshing pulse train
captured and the trigger control, while the right-hand
side display accumulates the partial discharge
mapping data.
With the latter display each dot represents a valid
location result. With our example (Fig. 8) the
accumulation (A) represents the location of the PD
fault with respect to the echoes of the partial dis-
charge traveling the full cable accumulating at (B).
Subsequently, the distribution of these results is
used to automatically calculate the fault locations. As
every valid location result contributes to that display,
multi-fault conditions are shown clearly, as well.
Additionally, the control of a low frequency source is
offered with this program, if used together with a
suitable source.
 Fine Location
The results obtained by reflection based fault
location are validated with a fine location using near-
field probes, i.e. antennas, high frequency clamp-on
current transformers, and piezoelectric ultrasonic
sensors.
The battery-operated version of the ICMcompact
(Fig. 9) simplifies such on-site inspection of sub-
station equipment and cable accessories, such as
medium-voltage cable joints. Independent operation
of this portable instrument is provided by a wireless
transmission of the synchronization signal (300m,
approx.) and a battery lifetime of more than 3 hours
(NiMH).
The instrument offers a vivid B/W pattern and
oscilloscope-like display refreshing ten times per
second. Modular pre-amplifiers adapt the instrument
to the different sensors and pickups while keeping it
portable. Capturing of screenshots is possible via the
serial interface of the unit.
Interpretation strategies concerning the phase–
resolved partial discharge pattern do not change too
much as phase position and relative amplitude is
invariant, regardless of the type of sensor utilized
and the physical appearance of the partial discharge
impulses used [4].
Moving the clamp-on CT (Fig. 9) over a heat
shrinkable cable joint, for example, while observing
the display, pinpoints the discharge source without
sophisticated knowledge. However, detailed identi-
fication and differentiation of the partial sources and
properties requires experience in general.
Measures to acoustically capture the partial
discharge activity offer this easy try-and-locate
approach as well. With lead sheathed cables this
ultrasonic pickup is the only possible choice for fine
location. Rough evaluation is made with head-
phones, while the phase-resolved analysis of the
acoustic signals is available with the pattern display
of the instrument.
Fig. 9, Battery-operated unit with clamp-on CT
References
[1] Toya A. et al, 'Development of a partial
discharge automatic monitoring system for
EHV XLPE insulated cable lines', Proc. of the
4th Intl. Conf. on Insulated Power Cables,
Versailles, France, 1995.
[2] Pultrum E. et al, 'HF partial discharge
detection of HV extruded cable accessories',
Proc. of the 4th Intl. Conf. on Insulated Power
Cables, Versailles, France, 1995.
[3] D. Gross and R. Comanns, 'TE-Messung und
Fehlerdiagnose an Kabeln und Garnituren',
ETG Fachtagung 'Teilentladungserfassung an
elektrischen Isoliersystemen', Esslingen,
Germany, 1995.
[4] Bernhard Fruth and Detlev Gross, 'Partial
Discharge Signal Generation, Transmission
and Acquisition'. IEE Proc.-Sci. Meas.
Technol., Vol. 142, No. 1, January 1995.
[5] B. A. Fruth, D. W. Gross, 'Modelling of
Streamer Discharges between Insulating and
Conducting Surfaces', Conference
Proceedings of the 1995 5th Intl. Conf. on
Conduction and Breakdown in Solid
Dielectrics, Leicester, England, 1995.
[6] B. A. Fruth, D. W. Gross, 'Tools and Sensors
for flexible On-Line Diagnosis of High Voltage
Insulation Systems', Proc. of the IEEE
International Symposium on Electric Power
Engineering, Stockholm Powertech, Sweden,
1995.
[7] E. Robles, B. Fruth, and D. Gross, 'A High
Sensitivity Technique to Detect Incipient
Failures on In-Service High Voltage Bushings',
Proc. 8th Int. Symp. High Voltage Engineering,
Graz, Austria, 1995.
[8] C.G. Henningsen, K. Polster, Bernhard Fruth
and Detlev Gross. 'Experience with an On-
Line Monitoring System for 400kV XLPE
Cables'. Proc. 1996 IEEE Transmission and
Distribution Conference, Los Angeles,
September 15-20, 1996.
[9] B. Fruth and D. Gross, 'Partial Discharge Data
Acquisition System for Substation Monitoring',
EPRI Substation Equipment Diagnostics
Conference V, New Orleans, 1997.
[10] D.W. Gross and B.A. Fruth, 'Distortion of
Phase resolved Partial Discharge Pattern due
to Harmonics and Saturation', Annual Report
Conf. on El. Insulation and Dielectric
Phenomena, Atlanta, 1998.
[11] D.W. Gross and B.A. Fruth, 'Characteristics of
Phase Resolved Partial Discharge Pattern in
Spherical Voids', Annual Report Conf. on El.
Insulation and Dielectric Phenomena, Atlanta,
1998.
[12] U. Sundermann, B. Fruth and D. Gross, 'On-
Line Partial Discharge Testing of Power
Transformers', EPRI Substation Equipment
Diagnostics Conference VI, New Orleans,
1998.