2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

Partial Discharge Characteristics of XLPE Cable

Joint and Interfacial Phenomena with Artificial
Defects
Yanuar Z. Arief*, Hussein Ahmad*, and Masayuki Hikita**
* Institute of High Voltage and High Current, University of Technology Malaysia, Skudai,
81310 Johor Bahru, Malaysia. Email: yzarief@fke.utm.my
** Department of Electrical Engineering, Kyushu Institute of Technology, 1-1 Sensui-cho,

804-8550 Kitakyushu-shi, Japan. Email: hikita@ele.kyutech.jp

Abstract—This paper deals with partial discharge characteristics).

(PD) causing degradation of XLPE cable joint and
interfacial phenomena of XLPE/EPR interface due to
artificial defects such as metal particles, needle-like
void and fiber soaked with water. We measured aging
time dependence of phase-resolved PD (φ-q-n) patterns
and PD statistic parameters when such kinds of
defects were placed at XLPE/EPR interface. As a
result, φ-q-n patterns were found to depend upon the
defect type varying with aging time. In order to grasp
the degradation mechanisms, we compared a practical
XLPE cable joint with a model sample to simulate the
real one. Moreover, electron spectroscopy for chemical
analysis (ESCA) was also carried out to interpret the
degradation mechanism.
KeywordsüPartial Discharge; XLPE/EPR interface; Phase-
resolved PD Pattern; Electron Spectroscopy for Chemical
Analysis
I. INTRODUCTION
In order to improve long-term reliability of XLPE
power cables, it is essential to understand the mechanisms
of degradation and breakdown of solid insulation
materials such as XLPE and its interfacial phenomena.
The insulation performance of polymers usually
decreases quickly with high voltage application, if partial
discharges (PD) occur in voids or defects. Hence, PD
detection and diagnosis are one of the most important
means to test HV cables. In the recent years, PD tests
have been widely conducted after XLPE cables were
completed, and the importance of PD test has been
already recognized [1]. Moreover, we have investigated
physical phenomena and degradation processes by
measuring PD of XLPE cable joint [2-6]. However,
identification of defects and degradation processes of
XLPE power cable joints particularly in the interface,
have not been fully understood yet [7-9]. From this point
of view, we have been trying to identify the types of
defects and elucidate the degradation processes of XLPE
cable joint by measuring phase-resolved PD (φ-q-n)
patterns and PD statistic parameters, etc (PD
In this paper, first we will show the PD characteristics
of simulated XLPE/EPR interface of XLPE cable joint
containing three kinds of artificial defects namely metal
particle, air void, and wet fiber, respectively. Then we
investigated aging time dependence of PD characteristics
up to breakdown when a needle-like void defect was
placed in practical XLPE cable joint and compared the
result with model samples. Moreover, to interpret the
degradation mechanism, we used electron spectroscopy
for chemical analysis (ESCA).
II. EXPERIMENTAL SETUP
Figures 1 and 2 illustrate the specimens simulating the
XLPE cable joint which contain three kinds of defects at
the EPR/XLPE insulation interface and the experimental
circuit, respectively. We piled EPR insulating rubber in
which a semiconducting rubber (SCR) as a column high-
voltage electrode was molded in the center, and XLPE
(insulation) sheet used as a “sandwich” sample to
simulate the EPR/XLPE interface of XLPE power cable
joint. Table 1 shows the kinds of defects which were
artificially placed at the EPR/ XLPE interface of the
specimens. The whole set of the specimens with the
electrode system was then immersed in a silicone oil to
prevent unwanted discharges which might occur around
the HV electrode.
We measured PD patterns (i.e., φ-q-n characteristics)
by a computer-aided PD measuring system. An
impedance matching circuit with an ultra-wide frequency
response up to 800 MHz was adopted to measure phase-
dependence of PD current pulse waveforms and to
prevent reflection and distortion of PD current pulses.
The phase angle of 360 at one cycle of the applied ac
voltage with a frequency of 60 Hz was divided into 12
sections in order to investigate the phase-dependence of
PD current pulses, which is known as the so-called phase-
angle gate control technique [10].
Figure 3 illustrates the experimental setup of practical
XLPE cable joint for measuring PD characteristics by the
so-called foil electrode method [11], in which aluminum

1-4244-2405-4/08/$20.00 ©2008 IEEE 1518

 2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

foils are rapped onto the cable sheath at the cable joint to TABLE I.
form coupling capacitance (500 pF).We used 6.6 kV THE ARTIFIAL DEFECTS INTRODUCED AT THE INTERFACE
XLPE power cable for experiments.
Figure 4 illustrates a needle-like void defect (depth: 3 Size
Silicone oil
mm) placed at the EPR/XLPE interface of XLPE cable Defect Description
(mm)
with without
joint (insulation thickness is set as 19 mm from the
Metal Copper wire l=2; φ=0.6 CW-SO CW-NSO
conductor). We measured phase-resolved PD (φ-q-n) particle (CW)
patterns with a computer-aided PD measuring system Air void Air void l=5;depth=0.5 AV-SO AV-NSO
[12]. We used a tuning amplifier to attain high S/N ratio (AV)
by selecting a tuning frequency (fc=10 MHz). In Wet Cotton fiber l=2; φ=0.5 FW-SO FW-NSO
fiber soaked with
addition, we injected the amount of charge for calibration water (FW)
by the direct method with injecting charge from the cable
edge. Pb tape SCR

EPR
XLPE
70 70
SCR

φ =18

φ =19
9 1
[mm]
12
Semi-conducting 18 Insulating
rubber 0.5 mm
rubber 2 mm Diameter of a needle: 1 mm
SCR
observation
Air 25 mm EPR
4 mm 3 mm φ =9.5 mm
XLPE 5 mm
Defect SCR
2 mm
Radius of curvature: 5 μm Conductor 4.5 mm
Needle-like void [mm]
XLPE sheet
dimension:
Needle width = 0.5 mm insulating rubber
Depth = 0.5 mm Figure 4. Artificial needle void defect at EPR/XLPE interface
Radius of curvature = 5 μm of XLPE cable joint (SCR = semiconducting rubber)
XLPE sheet Defect
III. EXPERIMENTAL RESULTS
Figure 1. Model sample for simulating XLPE cable joint

Applied voltage A. Simulated XLPE/EPR Interface

Specimen
Rk (1MΩ) Electrode PD 1) PD inception voltage and breakdown voltage
Silicone
High oil Oscilloscope
voltage Detection Figure 5 shows PD inception voltages (PDIV) Vi and
impedance
Ck PC
breakdown voltages (BDV) VB for three kinds of the
(500pF)
Differential
artificial defects at the EPR / XLPE interfaces of the
Switch amplifier
Voltage
divider Matching PD measuring model samples with or without silicone oil at the interface.
circuit Current system As can be seen from the figure, Vi of CW-SO (metal
particle defect with silicone oil applied) is 17 kVrms,
whereas Vi of CW-NSO (metal particle defect without
Oscilloscope
silicone oil applied) is much lower, about 7 kVrms. In
Figure 2. Experimental setup of PD measuring system other words, by applying silicone oil at the interface
between EPR and XLPE insulation, Vi for CW defect was
Artificial defect increased by about 10 kVrms. On the other hand, Vi of
Direct calibration AV-NSO (air-void defect without silicone oil) is 10
Rk=1 MΩ Foil electrodes
(500 pF) kVrms, while no PD was observed with silicone oil at the
Joint interface up to about 30 kVrms. Almost no effects on
AC High voltage
DC supply PDIV and BDV of the samples containing FW (fiber
Preamplifier Cut core soaked with water) defect (water conductivity: σ =
C distributor 42 μS/mm) at the artificial defects were seen, regardless
of silicone oil application at the interface.
Attenuator
PD measuring㩷system 2) Waveform of single PD current pulses
Tuning amplifier
Oscilloscope PC
Waveforms of PD single current pulses for the samples
with the three kinds of defects at the voltage of 1.1 times
of PDIV are recorded in [4] . Note that these waveforms
Figure 3. Experimental setup of XLPE cable joint for PD were measured by the ultra-wide frequency band
measurement using foil electrode method technique (frequency bandwidth: 750 MHz) together with

1519
 2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

a digital oscilloscope (Sampling rate: 10 GSa/s). It is
apparent that most measured PD current pulses have very
sharp rise times tr (about 1 ns), although a few PD pulses
have rise times larger than 2 ns. The amplitude of PD
current pulse for CW-SO is about 0.25 mA which
corresponds to an apparent discharge quantity below 1 pC.
For CW-NSO, the PD current peak value is about 8.5
mA and its corresponding apparent discharge quantity is
We measured ageing time dependence of φ-q-n patterns
and PD statistic parameters with the configuration as
shown in Figure 4 at PD inception voltage (PDIV: applied
voltage Va = 11.1 kVrms).
40

Applied voltage [kVrms]

quite high, about 72 pC. On the other hand, for AV-NSO, 30
current peak values are 5.1 mA and 48.5 mA measured in
the phase angle range of 330o∼360o and 0o∼30o, which 20
correspond to 34 pC and 203 pC, respectively. No PD
pulse was detected by this technique for AV defects with 10
silicone oil. Nevertheless, no change was observed for the
peak waveforms of PD current pulses for the FW defects, 0
Copper Void Fiber Copper Void Fiber soaked
CW AV FW CW AV FW
irrespective of silicone oil placed at the interface.
Without silicone oil With silicone oil

3) Time Evolution of PD Patterns and current PD inception voltage CW: Copper wire
AV: Air void
Breakdown voltage FW: Fiber soaked with water
waveforms
Figure 5. PDIV (Vi)and BDV (VB) of the model sample of XLPE
Figures 6 shows the time evolution of φ-q-n cable joint with the defects.
characteristics and rise time tr of PD pulses for CW-NSO.
The voltage application was 10 kVrms. At the beginning
of the test after applying HV ac voltage, the measured φ- 200 200
q-n pattern looks like a hill with maximum discharge
quantities qm smaller than 100 pC, but qm gradually Charge [pC]
increases to approach 200 pC after the uninterrupted PD 0 0
ageing experiment up to 120 min. Further experimental
results revealed that qm of PD then decreased gradually to
be less than 100 pC after 180 min or tests, and no PD
3 3
pulses could be finally measured by both the computer
tr [ns]

and digital oscilloscope techniques for 5h continuous

0 0
aging experiments. The results mean that the PD 0 180 360 0 180 360
mechanism may be changed. It is also noted that tr (1) 0 min. (2) 60 min.
200 100
fluctuates in the range of 0.8∼3.5 ns for these samples.
For AV-NSO, obvious changes of φ-q-n
Charge [pC]

characteristics were not observed for 5h aging test and qm 0 0

was less than 5 pC. It was also found that tr became
slightly longer after 30 min from 1̚2 ns to 2̚3 ns
similar to that of copper wires, and then became shorter
3 3
to 1̚2 ns after 1h test [4].
tr [ns]

While, φ-q-n characteristics for FW-NSO which 0 0

emerges from zero cross point of the applied voltage at 0 180 360 0 180 360
the beginning of the test with qm larger than 122 pC. (3) 120 min. (4) 180 min.

After 30-min test, qm increases to be higher than 180 pC, Figure 6. Evolution of φ-q-n characteristics and tr of PD for the
and reaches highest value (about 300 pC) after 90 min (a)copper wire defect without silicone oil at 10 kVrms
test. qm diminishes to be smaller than 80 pC after 150-
min ageing test and finally disappeared after 3h [4].
Treeing penetrating into the XLPE insulation is
clearly seen for FW-NSO defect after 3h ageing test as
shown in Figure 7, in which the location of the treeing
and the detailed tree structures are illustrated. This is a
typical electrical treeing pattern since it is permanent
deterioration of XLPE insulation under the combined
actions of PD and the FW defect at the XLPE/EPR 㪌㫄㫄
XLPE
interface of the model sample.
B. Ageing Time Test Figure 7. Treeing deterioration when a FW defect was placed at
XLPE/EPR interface at 5 kVrms
1) Practical Cable Joint

1520
 2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

Figures 8 (a), (b) and (c) show φ-q-n patterns for the appear intermittently. qm and n increase abruptly leading
XLPE cable joint with a needle-like void defect at to breakdown at 55min.
different ageing times, i.e. just after voltage application, As a result, breakdown of the model sample takes
at abrupt increase in charge and just before breakdown place after such a sudden increase of qm and n together
with keeping voltage at PDIV. It can be seen that PD with much occurrence of PD with the magnitude as much
occurred in the phase angle region between the zerocross as 2-4 pC. We observed the surface of the EPR/XLPE
and the peak of Va till breakdown. Note that the phase interface with an optical microscope when q increased
angle region where PDs appear remains unchanged. abruptly.
Moreover, Figures 10 (d) and (e) show the relationship of Figures 11 (a) and (b) show photographs of XLPE
maximum charge qm vs ageing time t, and repetition rate sheets around the needle-like void before and after the
n vs ageing time t, respectively. sudden increase of qm and n, respectively. It is found in
As can be seen in the figure, after 15 min, the PD Figure 11 (a) that no distinct degradation is observed
magnitude decreases below the detection level of PD around the void, while tree initiating from the needle-like
measuring system (in this case, 17.5 pC). After 104 min void toward the high voltage side can be observed for the
of ageing time, PD pattern abruptly changed, i.e. as abrupt increase in qm (Figure 11 (b)).
shown in Figures 8 (d) and (e), qm and n abruptly
increase by about three times as much as those at the
initiation of voltage application.
Figure 9 shows time sequential characteristics of
charge q at 104 min (Fig. 8(b)). It is obvious from the
figure that q increases abruptly from about 50 to 300 pC.
Moreover, it can be seen from the magnified figure that
PD occurred in the phase angle region between the
zerocross and the peak of Va. We considered that the
sudden increase of q (qm) exhibits a sign of tree
(d.) qm-t characteristics
initiation.

2) Model Sample
We measured ageing time dependence of φ-q-n patterns
and PD statistic parameters till breakdown for the model
sample with a void defect at PDIV (Va=11.5 kVrms).
Figures 10 (a), (b), and (c) show φ-q-n patterns from
the beginning of voltage application up to the time at just
before breakdown. Experimental results reveal that the
phase angle region of PD occurrence is the almost same
as that for XLPE cable joint with the void defect.
(e.) n-t characteristics
Figures 10 (d) and (e) show qm vs t, and n vs t
Figure 8. Aging time dependence of PD characteristics for XLPE
characteristics, respectively. As can be seen from the cable joint with a needle-like void defect (a) to (c) φ-q-n
figure, PD with the magnitude of about 100 pC seems to characteristics (Va=11.1 kvrms (PDIV), fc= 10 MHz,
1 set data of 300 cycles)

Figure 9. q-t characteristics for XLPE cable joint with a needle-like void defect at aging time when q suddenly increases
(Time sequential characteristics of Fig. 8(b))

1521
 2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

Figure 10. Aging time dependent of PD characteristics for model
sample with a needle-void defect, (a) to (c)φ-q-n patterns, (d)
and (e) qm-t and n-tracteristics (Va=11.5 kVrms (PDIV),
1 set data of 300 cycles)
2 mm
defect
HV
observation
XLPE sheet
Observation area of XLPE sheet with a needle void defect
(a) Non PD increase abruptly (b) PD increase abruptly
Figure 11. Photograph of EPR/XLPE interface for model sample
using optical microscope
C. Chemical Analysis of XLPE Model Sample
Typical electron spectroscopy for chemical analysis
(ESCA) spectra of O1s, and C1s peaks for the virgin and
aged XLPE around Cu wire defect  (CW) or in the
treeing region are recorded in [4]. The CW specimen was
aged at 8 kV for 30h. It is found that O1s peak heights and
areas of the aged XLPE samples increase due to PD and
treeing degradation, compared to the virgin one.
However, the heights and areas of the C1s peaks of
degraded XLPE are reduced apparently as recorded in
[4]. Actually, the asymmetrical ESCA curves can be
1522
further separated into symmetrical ones with different
binding energies, giving detailed information of the
molecular bonding structures of the materials.
Based on the ESCA analysis results of the XLPE
model samples as shown in Figure 12, we found that the
areas of the separated symmetrical C-O peaks of the
asymmetrical C1s binding energy peaks increased
obviously whereas those of C-C peaks decreased, which
implies that severe XLPE insulation ageing occurred and
scission of C-C bonds or main molecular chains of the
XLPE dielectric material.
100
Area ratio of binding energy [%]
C-O
80
60
40 C-C
20
0
Virgin Around Treeing
XLPE Cu wire Region
Figure 12. Area ratio variation of the C-O and C-C peaks obtained from
the ESCA spectra of the virgin and aged XLPE materials of the CW
and FW samples due to PD
IV. DISCUSSION
Experimental results for both XLPE cable joint and
model sample with the needle-like void allow us to
recognize an agreement between these results on PD
characteristics such as φ-q-n patterns. It can be said that
the model sample simulates XLPE cable joint very well
in terms of PD characteristics. Moreover, we can also say
that the sudden increase of qm and n exhibits a sign of
tree initiation and its evolution development.
It is obvious that Vi and VB increased by placing
silicone oil at the interfaces to eliminate small air void or
layer. PD for CW-SO is considered to be caused by the
local electrical field enhancement due to existence of the
artificial conductive defect. The result that Vi and VB for
FW defect were not changed by applying silicone oil
suggests that different mechanisms control these
processes. In other words, serious effects of water and
ionic impurities must be taken into account including
treeing at the interface. It has been confirmed that water
treeing might be initiated at the interface of the simulated
cable joint sample and finally grows into electrical tree
due to these effects and PD as shown in Figure 7.
There are some PD pulses with larger discharge
quantities (e.g., q higher than 150 pC), while a lot of
smaller PD pulses (e.g., q lower than 100 pC) also exist
(Fig. 6). It is proposed that smaller pulse discharges may
be governed by Townsend process, while larger PD
pulses may be caused by the streamer discharge
mechanism. It is reasonable to image that larger PD
should produce more charges at the interface, which
 2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia
would influence  the local electric stress around the
defects so that PD characteristics varied more
considerably.
From the above results, it is interesting that the
maximum discharge quantity qm of PD first increased and
then decreased in short-time ageing test for CW and FW
defects. PD pulses disappeared after 3-hr test for both
defects and could not be measured by the electrical pulse
method, which may imply that swarming pulsive micro
discharge (SPMD) or glow discharges have taken place
[15]. It was reported that PD was closely related to
sufficient supply of primary electrons, since such electrons
would reduce the statistical time lag of discharges [16]. It
is then assumed that SPMD for both defects may be
originated from the reduction of over voltage across the
defects at the interfaces owing to the existence of abundant
free electrons generated by preceding PDs. Furthermore, it
was also reported that gas pressure inside a void in the
epoxy sample first increased and then gradually decreased
to low pressure below 60 kPa, which affect the PD
behavior [17]. On the other hand, PD activities also create
highly reactive species like ozone and ions of oxygen.
Hence rapid pressure variation is probably due to chemical
reactions of PD with dielectrics.
The effect of water for FW-NSO defect exposed to
discharges promotes the clustering of degradation
products, thus leading to the formation of nodules and
treeing on it. Oxidative reactions with the polymers (XLPE
/EPR) have occurred during PD to form conductive acid
droplets and higher surface conductivity. These effects
may lead to a transition of PD mechanism from spark type
pulse discharge to pseudo glow and, finally to a true glow
discharge. The latter two forms of discharges were
observed to prevail during the PD exposure duration [18].
Usually, internal PD in polymer insulating materials is
much easier to be initiated and developed under low gas
pressures and high surface conductivity. This may allow
the surface charges at the interface to redistribution more
easily so that it changes the local electric field distribution
and also reduces the dielectric strength. It has been
confirmed that SPMD is still harmful to electrical
insulation materials [19, 20], so that it is imperative to
measure SPMD by other techniques in the future research.
Chemical analysis results (ESCA) of the degraded
XLPE samples reveal the increase of O1s peak but
decrease of C1s peak. It means that scission of C-C bonds
or main molecular chains and oxidation of the XLPE
insulation proceeded. Oxidation reactions generally
involve a free radical chain reaction. Some of the main
steps in this reaction are recorded in [22-26]:
Radicals formed during initiation react with oxygen,
leading to chain reactions. The decomposition of
hydroperoxides by heat or UV light causes formation of
alkoxy and hydroxy radicals leading to chain branching as
evidenced by both FT-IR and XPS results.
VI. CONCLUSIONS
Partial discharge characteristics of XLPE cable joint
1523
and interfacial phenomena of XLPE/EPR interface due to
artificial defects such as metal particle, needle-like void
and fiber soaked with water was investigated. The
following conclusions are obtained.
By applying silicone oil at the simulated model
interface, PDIV and breakdown voltage were much
higher for the copper wire and air void defects, whereas
little effect was observed for the fiber soaked with water
defects as water and ionic impurities dominated the PD
and breakdown process.
The experimental results for both XLPE cable joint
and model sample with the needle-like void allow us to
recognize an agreement between these results on PD
characteristics such as φ-q-n patterns. It can be said that
the model sample simulates XLPE cable joint very well
in terms of PD characteristics. We also found that the
sudden increase of qm and n exhibits a sign of tree
initiation and its evolution development.
PD characteristics of the CW specimen changed from
pulse-like PD into swarming pulsive micro discharge
(SPMD) or glow discharges during the ageing test, while
electrical treeing deterioration in the FW sample seems
responsible for the fast local breakdown of XLPE.
Among three kinds of the artificial defects, the FW defect
seems the most detrimental one to the insulation integrity
and life of the samples. Moreover, after 3h of ageing
experiments, PD pulses of the samples with CW and AV
defect disappeared finally (could not be detected by PD
measuring system.
ACKNOWLEDGMENT
The authors would like to thank to Fujikura Ltd for
providing XLPE cable joints, model sample materials and
their additional accessories for this research work. Many
thank also to Mr. Takehiko Mizuno of Viscas Co. for
valuable discuss during completing this project.
REFERENCES
[1] K. Uchida, T. Hirata, I. Tanaka, T. Takahasi, H. Takehana,
“Partial Discharge and Breakdown Characteristics of various
defects in XLPE cable joints, IEEJ, vol. 155-B, No.10, pp. 1185-
1192, 1995.
[2] S. Tokunaga, T. Tsurusaki, Y.Z. Arief, S. Ohtsuka, T. Mizuno,
and M. Hikita, “Partial discharge characteristics till breakdown for
XLPE cable joint with an artificial defect”, Proc. of. The 7th
International Conference on Properties an Applications of
Dielectric Materials (ICPADM), S23-2, pp. 1206-1209, Nagoya,
Japan, 2003.
[3] S. Tokunaga, , Yanuar Z. Arief, T. Tsurusaki, T. Mizuno, S.
Ohtsuka and M. Hikita, “Partial discharge characteristics and
electric field calculation for XLPE cable and joint of EPR/XLPE
interface with void and metal particle defects”, Joint Conference
of ACED and Korea-Japan Symposium on Electrical Discharge
and High Voltage Engineering, Vol. 2, pp. 503-506, Seoul, Korea,
2002.
[4] H. Yoshida, Z. Thian, M. Hikita, H. Miyata, “Study on the time
evolution of partial discharge characteristics and interfacial
phenomena of simulated XLPE cable joint”, International
Symposium on Electrical Insulation and Material (ISEIM), D2-3,
pp.501-504, 1998.
[5] S. Tokunaga, T. Tsurusaki, Y.Z. Arief, S. Ohtsuka, T. Mizuno and
M. Hikita, “Partial discharge characteristics for XLPE cable joint
of EPR/XLPE interface with artificial defects”, The Papers of
Technical Meeting on Dielectrics and Electrical Insulation, IEEJ,
DEI-02-105, Jpn, pp. 43-48, 2002.