Fault Location and Diagnosis in a Medium Voltage EPR

Power Cable
A. J. Reid, C. Zhou and D. M. Hepburn
School of Engineering and Built Environment
Glasgow Caledonian University
Glasgow, G4 0BA, UK

M. D. Judd and W. H. Siew

Institute for Energy and Environment
University of Strathclyde
Glasgow, G1 1XW, UK

P. Withers
Manchester X-ray Imaging Facility
Materials Science Centre
University of Manchester, M13 9PL, UK

ABSTRACT
This paper presents a case study on fault location, characterisation and diagnosis in
a length of shielded 11 kV medium voltage ethylene-propylene rubber (EPR) power
cable. The defect was identified on-site as a low resistance fault occurring between the
sheath and the core. A 43 m section was removed for further analysis. The fault
resistance was characterised and the location of the defect pinpointed to within a few
cm using a combination of time-difference-of-arrival location and infra-red imaging. A
combination of X-ray computed tomography, scanning electron microscopy and energy
dispersive X-ray spectroscopy were then applied to characterise any abnormalities in
the dielectric surrounding the breakdown region. A significant number of high density
contaminants were found to be embedded in the dielectric layer, having an average
diameter of the order of 100 μm, a maximum diameter of 310 μm and an average
density of 1 particle per 2.28 mm3. Scanning electron microscopy and energy-dispersive
X-ray spectroscopy were used to determine the geometry and elemental composition of
some initial contaminant samples. It was concluded that contamination of the EPR
layer, combined with an observed eccentricity of the cable’s core and sheath resulting
in a reduced insulation gap, may have led to an electric field concentration in the
region of the defect sufficient to initiate breakdown. Preventative strategies are
discussed for similar families of cables, including more stringent dielectric testing
requirements at the manufacturing stage and PD monitoring to detect incipient failure.
Index Terms — Power cable insulation, power system faults, condition monitoring,
dielectric breakdown.

1 INTRODUCTION preventative maintenance through the use of on-line partial

discharge (PD) monitoring can allow an incipient fault to be
POWER cables form an integral part of power generation,
detected and located before an eventual failure occurs [1]–[4]. If
transmission and distribution systems. With an increasing rate of
the initial defect is not detected in its incipient stages, it may
failure due to ageing and operational stresses, maintaining the
increase in severity, eventually leading to complete breakdown.
integrity of a cable network through effective monitoring and
Increased defect severity may be characterised by, for example, a
replacement is a subject of importance today. Understanding the
change in resistivity, a break in the conductors or by partial
characteristics, location strategies and underlying causes of a
bridging of the insulation. Various approaches to defect location
cable fault can be effective in informing monitoring and
[5] have been developed over the years with each method having
replacement schemes for such assets in the future. In many cases,
its own relative merits. The severity of such defects often
precludes the application of partial discharge monitoring. The comparatively low resistance of the shunt defect allowed
The aim of this case study was firstly to apply an effective Joule heating of the channel to be used as a means of
post-fault location technique in the lab to pinpoint the fault on a confirming or otherwise, TDOA location accuracy.
length of cable that has been taken out of service. Secondly, the
2.2 TIME-DIFFERENCE-OF-ARRIVAL
study aims to characterise the geometric detail of the region
MEASUREMENT
surrounding the fault through the application of X-ray and
microscopy techniques, allowing a hypothesis to be proposed for Figure 1 shows the experimental setup for TDOA location.
the initial cause of breakdown. It is suggested that the reported The TDOA location technique has been a subject of interest in
measurements on a medium voltage EPR cable, of the type overhead power line research [7]–[9]. For example, Bo et al
commonly used in power generation, will inform improved [10] have applied this method to overhead power lines and
future PD monitoring and testing strategies, leading to a reduced suggested it can be applied to underground power cables. In
failure rate in generation networks of this type. this case the cable was connected to the output of a 50 kV
transformer with the voltage applied between the core and
1.1 THE CABLE SAMPLE sheath. The supply voltage Vs was increased until transition
The cable was single phase EPR, rated at 11 kV. A length from a high to a low resistance occurred in the fault channel.
of 43 m was removed and taken out of service after the The resulting single fast transient current pulse was extracted
occurrence of a breakdown and transferred to the laboratory at each cable end by positioning two HFCTs the earth straps,
for further testing. The section under examination was as shown in Figure 1. Each HFCT had a sensitivity of 5 V/A
manufactured in 1983 to standard BS6622 [6] and has been in in the operational range 5 kHz - 10 MHz.
service from this date until the breakdown occurred in 2009.
The cable had a single core stranded compacted aluminium
inner conductor, stripable semiconducting screen, EPR
dielectric, wound copper tape outer conductor, PVC bedding,
aluminium wire armour and a PVC outer sheath with reduced
flame propagation characteristics. No PD monitoring was
previously applied on site (although PD may have initially
occurred before breakdown). During operation, the defect
Figure 1. Experimental setup for TDOA location of breakdown pulse from
increased in severity forming a partially-conductive channel cable fault.
through the insulation layer. This precluded PD-based location
techniques. The following sections will describe an Knowledge of the propagation velocity up of a transient
investigation of the fault characteristics, detailing the location electromagnetic wave in the EPR medium was necessary for
methods applied and the accuracy of these methods in this time-of-flight calculations. Due to dispersion effects, up varies
case. with frequency. It was therefore necessary to determine the
propagation velocity of transverse electromagnetic waves in
2 DEFECT LOCATION the coaxial transmission line formed by the cable in the
expected frequency range of electromagnetic radiation
2.1 INITIAL DIAGNOSIS emanating from a typical breakdown current pulse. To
Initial tests aimed to determine the defect type. These were determine up, the experimental setup shown in Figure 2 was
first carried out using an insulation resistance testing device to used.
investigate any discrepancies in the core, sheath and insulation
resistances. No abnormalities were found in the conductors,
but the resistance between outer and inner conductor was
found to vary between 1 - 10 MΩ at an applied voltage of
500 V rms, indicating a possible shunt resistance defect.
Results of fault resistance measurement were not repeatable,
varying by several MΩ on each occasion. The cable was next
tested at operational voltage. PD-free terminations at each end
Figure 2. Experimental setup for measurement of propagation velocity up.
were necessary in order to prevent spurious discharges from
interfering with subsequent time-difference-of-arrival Injection of repetitive pulses in the appropriate frequency
(TDOA) measurements. These were constructed using pre- range was achieved using a mercury wetted relay (MWR)
moulded stress cones. The application of a ramped ac voltage device. This allowed fast switching of high amplitude pulses
revealed that the fault underwent transition from a high to a in a nanosecond timescale, producing similar waveforms to
low resistance, manifested by a single transient waveform those observed as a result of cable breakdown. Both the initial
generating sufficient current to be extracted using two high injected pulse and the pulse reflected from the open circuit
frequency current transformers (HFCTs) at each cable end. end of the cable were captured using a non-linear attenuator
circuit based on PIN diodes, as shown in Figure 2. This novel 2.3 THERMAL LOCATION
approach protected the oscilloscope’s input voltage from the To verify the accuracy of the above results, thermal location
initial high amplitude injected pulse, while still allowing the was next applied. Energy dissipated by the fault current was
low amplitude reflected pulse to be measured with high sufficient to raise the surrounding region above ambient
sensitivity. The time difference between injected and reflected temperature. A thermal imaging camera (FLIR E45) with a
pulses was 581 ns, corresponding to twice the transit time of temperature resolution of 0.1oC was used pinpoint the location
the cable. Since the exact physical length of the cable could be to within a few cm.
measured, up could be calculated using the equation up = l /ΔT The cable was energised for 15 minutes at a maximum
where l is the physical length of the cable and T is transit time applied voltage of 35 V rms. This relatively low value was
along its length. In this case, up = 43.1/290.5 x 10-9 = 1.48 x due to the shunt fault forming a low resistance channel to
108 ms-1. earth, reducing the potential difference between core and
The measurement error in time difference was estimated to sheath. Thermal imaging revealed the location of the shunt
be ±4 ns, caused by limitation of the scope’s horizontal fault to be 1.8 m from the cable end. This falls within the
resolution combined with the error in estimating the starting measurement error of the previous location result, confirming
point of the reflected pulse’s rising edge due to dispersion the accuracy of the applied TDOA method. Figure 4 shows a
effects. thermal image of the energised cable overlayed on a
Variation in the conductivity of the fault was evidenced by photograph of the same planometric view of the cable layout.
a fluctuation in the voltage across the cable, with the peak of
the supply voltage occasionally clipping as the fault channel
underwent quasi-breakdown at high amplitudes. At this
transition point it was possible to observe low amplitude
emitted transients. An example is shown in Figure 3.

Figure 4. Fault location using infra red imaging. The location is 1.8 m from
the cable datum end. For clarity, this composite image consists of the
thermal image superimposed on a photograph of the experimental setup.
Figure 3. Measured traces at the respective HFCTs arising from breakdown
at the site of the cable fault. Δt = 261 ns. The cable had been placed in a spiral arrangement so that
the entire cable length was within the field of view. The site of
Fault location was then calculated using the equation: the fault was apparent on the inner coil of the spiral
arrangement as evidenced by increased temperature in the
region. Once thermal equilibrium was reached the peak cable
l − Δt ⋅ u p
d= (1) temperature was found to be 21.7oC (2.5oC above the ambient
2 level).
It was of interest to characterise parameters of the fault. For
where Δt is the time-difference-of-arrival between measured example, since the cable voltage, current and phase could be
signals at HFCTs 1 and 2. This was achieved by identifying measured, fault resistance could be evaluated in terms of its
similar features on the respective waveforms arriving at each variation with applied voltage. The resistance of the active
HFCT, such as the initial peaks, and estimating the time fault region is related to its reflection coefficient. Knowledge
difference between these features. Taking into account of this resistance is therefore useful in determining the
measurement errors in Δt and ΔT, the signals shown in Figure accuracy of reflection-based location methods. This will be
3 correspond to a fault location of d = 2.18 ± 0.56 m from discussed further in the next section.
HFCT1 (see Figure 1).
 3 DEFECT ANALYSIS applied voltages. Fault resistance fluctuated between 2 kΩ and
6.5 kΩ at the minimum applied voltage. The variation in fault
3.1 CHARACTERISATION OF RESISTANCE resistance was plotted against the voltage across the cable (test
The voltage across and current through the cable were object). This is shown in Figure 7.
measured as indicated in Figure 5, where, Ip is the primary
current, Is is the secondary current, Vs is the secondary voltage
and Ccable is the capacitance of the cable under test from inner Increasing primary
applied voltage
to outer conductor.

240V / 50kV

Figure 5. Experimental setup for measurement of fault resistance Rf of the

conductive channel between core and sheath.
Figure 7. Variation of fault resistance with the voltage across the cable.
The real part of the secondary current was calculated from
the phasor diagram in Figure 6 using the equation: Arrows shown in the figure indicate an increasing supply
voltage (and current) applied at the primary winding. In each
of the four tests, a corresponding increase occurs in the
I R =cos(δ ) ⋅ Is (2) voltage across the cable, until the fault reaches a transition
point where its resistance decreases significantly. In each of
the test cases, the defect resistance approaches a minimum of
around 200 Ω at the maximum applied primary voltage. This
minimum value was consistent with that initially obtained by
the site engineer, leading to decommissioning of the cable.

3.2 PHYSICAL INSPECTION

A short cable section of approximate length 200 mm was
removed at the location of the suspected fault. With the outer
Figure 6. Phase relationship between voltages and currents in the sheath and aluminium wire armour removed, visible damage
measurement set up of Figure 5.
was found at a location corresponding to 1.8 m from the cable
The fault resistance was calculated by dividing the end nearest HFCT1. This confirmed thermal and TDOA
secondary voltage by IR. Four different tests were carried out location accuracy to within a few cm. Carbonised tracking
to check repeatability. Due to the limitation in current which was evident around the perimeter of a 3 mm occlusion at the
could be drawn from the LV supply (around 16 A), the site of the breakdown. It was proposed that the internal
maximum voltage that could be applied to the cable was structure would be revealed in further detail by means of X-
around 35 V rms at the point where the defect resistance had ray computed tomography. This is described in the following
reached a steady-state value. Figure 7 shows the variation in section.
the fault resistance measurements for four repeated
experiments. 3.3 X-RAY COMPUTED TOMOGRAPHY
Due to the capacitive impedance of the cable, a phase The aim of the analysis was to capture internal geometric
difference existed between the secondary current and detail of the fault and surrounding medium. X-ray computed
secondary voltage, with the current leading the voltage by an tomography (or computed axial tomography) is ideally suited
angle between 0o and 90o, depending on the value of fault to this purpose. In this method, the sample is rotated axially,
resistance Rf. Calculation of the fault resistance required the with an X-ray image taken at each small increment in rotation
real part of the secondary current to be known. It was of the test object. On completion of the X-ray process digital
therefore necessary to measure the phase difference between geometric processing is applied to the acquired images in
Vs and Is. It was found that the capacitive impedance of the order to reconstruct the sample into a 3-dimensional digital
representation of the volume. This approach is highly
cable dominated, having an approximate value of 200 kΩ and
advantageous as it allows for non-destructive analysis of the
that the phase difference was small (~5o,), especially at higher
fault.
 The X-ray scanner used in this study was a Nikon custom acquired images: Acquisition time = 1.4 seconds; Applied
static bay with a 225 kV multi-metal anode source. The voltage = 110 kV; Applied X-ray current = 85 μA. The total
samples were positioned between the X-ray source and a 2k size of the raw images was around 20 GB. Since images were
by 2k Perkin Elmer 1621 XRD 16-bit amorphous silicon flat reconstructed at the maximum possible resolution, the
panel detector, allowing images to be captured in high detail. reconstructed volumes were around 50-100 GB is size and
Several factors had to be considered when determining the required intensive graphical and computational processing to
feasibility of X-ray imaging of the sample. Firstly, an plot. Fig. 8 shows a reconstructed orthographic slice on a
appropriate beam energy spectrum had to be chosen. Unlike plane normal to the cable’s longitudinal axis.
biological samples which are commonly analyzed using X-ray
computed tomography techniques, scanning of the cable is
challenging in that the beam must simultaneously have high
enough energy to penetrate through the conductive sections
and be detected by the receiver, whilst still allowing the subtle
internal features of the low density EPR layer to be captured
in high detail. After initial trials it was found that a
Molybdenium (Mo) anode source was most effective in
producing an appropriate photon energy spectrum to clearly
identify all internal features of the samples.
Secondly, it was desirable to achieve as fine a spatial
resolution as possible. In the present setup, X-ray resolution Figure 8. Reconstructed X-ray slice through defective cable section
was limited by the sample size due to the relative positions of showing breakdown channel and conducting particle contaminant.
the source, sample and detector array. The sample was placed
as close as possible to the X-ray source. The beam was
Several features are immediately evident. Firstly, the
emitted from the point source, passing through the sample,
breakdown channel is clearly visible, having a diameter of 2-3
and then detected by a flat panel array of around 0.5 m2. The mm. Tracking caused by the breakdown current has
closer the test object can be placed to the source, the greater extensively degraded the dielectric so that the channel is
the spatial resolution of the captured images on the detector. essentially an air gap at this stage. Secondly, the breakdown
An analogy here is that of a lamp projecting a shadow onto a has caused the inner semiconductor to detach from the inner
wall; the closer an object is placed to the lamp, the finer the conductor leaving a cavity. Thirdly, eccentricity is evident,
spatial resolution of the shadow that it casts. In the case of the with a 1 mm offset between the longitudinal central axes of
cable, which has a diameter of 30.5 mm with the outer the aluminium stranded core and the EPR insulation layer,
conductor removed, points on the sample could be positioned resulting in an increased insulation gap on one side of the
a maximum distance of 30.5 mm from the X-ray source. This cable and a reduced insulation gap on the other. Fourthly, the
led to a minimum achievable resolution of 18.3 μm. Test were presence of at least one high-density particle with an
initially carried out using the full cable sample, giving approximate diameter of 100 μm is evidently embedded
geometric details at the above resolution. To achieve finer within the insulation layer. The particle appears to have a
resolution, a small slice of the insulating layer was removed, similar density to that of the aluminium core. The appearance
allowing it to be rotated in front of the beam at a distance of of the particle on this arbitrary cross section suggests a
less than 15 mm. This allowed internal details of the EPR slice possible high contaminant density in this region.
to be viewed with a greatly improved minimum resolution of
Fig. 9 shows the results of volume rendering of the
9 μm.
reconstructed data. Rendering the high-density region as an
The third consideration in the feasibility of testing was isosurface, while separately rendering the volume of the EPR
tomographic reconstruction of the data. The geometry of the and semiconducting layers, allowed the effect of artificially
object plays an important role. Fortunately, the quasi-isotropic removing the insulation layer. This revealed the presence of
cable geometry lends itself ideally to volumetric multiple high-density contaminants. Several contaminants
reconstruction of a series of images around the cable’s central have been highlighted in Fig. 9 (c) for illustration purposes.
axis. This was because the reconstruction software was able to Damage to the inner core is evident around the breakdown
easily identify common highly contrasting features of region as is an increased contaminant density. Due to
consecutive images such as the outer edges of the aluminium sustained stressing, the fault has largely disintegrated the EPR
core and EPR layer. The data could therefore be reconstructed region surrounding it. Particle diameters are of the order of
with a high degree of accuracy. 100 μm.
Initially, the cable sample was scanned with the inner At this stage it was likely that the maximum resolution of
conductor, semicon and EPR layers intact. It was found that the measurement system was imposing a limit on the visible
the following settings achieved the optimal contrast on internal details of the insulation layer. As mentioned
previously the size of the sample imposes a lower limit on the The above results show a non-destructive method of
imaging resolution of the X-ray system and consequently, the obtaining the internal geometrical characteristics of the
smallest resolvable detail of the sample. A thin slice of EPR breakdown region and insulation layer, revealing details of
was therefore removed from the cable and analyzed high-density contaminants with a resolution of 9 μm. The
separately. This achieved an improved minimum resolution of following sections present the results of additional analysis
9 μm. Results are shown in Figs. 10 and 11. It can be seen using scanning electron microscopy and spectroscopy
from the 2D orthographic view of Fig. 11 that more techniques to reveal more detailed geometric detail and
contaminants are now visible, with diameters ranging from 40 elemental composition of the cable sample.
μm to 300 μm. Note that the bright regions near the edges of
the sample are not due to increased material density but are an
artifact of the geometric reconstruction algorithm.

(a)

Figure 10. Reconstructed 2D orthographic slice of a thin sample of

insulator near the defect region. Embedded high density particles appear as
light regions. Dimensions are in mm.

(b) (a)

(c) (b)

Figure 11. Reconstructed 3D geometry of a thin sample of insulator and

Figure 9. Reconstructed cable geometry from X-ray scans. (a) EPR layer is semiconductor removed from near the defect region. (a) Semiconducting and
fully visible, (b) EPR layer set to semi-transparent, (c) EPR layer hidden to EPR layers fully visible, (b) EPR and semiconducting layers set
show particle contamination and damage to inner conductor. semitransparent to reveal embedded high density particles. Dimensions in
mm.
Fig. 11 shows the reconstructed volume of the small
dielectric sample. Setting the low-density insulation region as
3.4 SCANNING ELECTRON MICROSCOPY
a semitransparent layer reveals the embedded contaminants.
The average particle density in this region is 1 particle per Computer aided X-ray tomography has revealed
2.28 mm3. geometrical detail down to a μm resolution. Detected
contaminants appear to have a density similar to that of
Aluminium. Investigation of whether the observed particles basis of apparent density as it appears in the X-rays.
may be the source of the breakdown requires further analysis. Computation of electric field enhancement using finite
Specifically, finite element modeling should allow the degree element analysis relies on knowledge of the electrical
to which the electric field enhancement in the region is likely conductivity and relative permittivity of the elements or
to induce breakdown. compounds in the cable sample. Knowledge of the elemental
In this case, an EDAX Evo 50 system was used, allowing composition is important in understanding the underlying
geometric details to be resolved at a sub-micron resolution. cause of breakdown since regions with varying permittivity or
With a maximum magnification of 1,000,000x, features down conductivity can greatly enhance the localized field.
to 2 nm could be resolved. The only disadvantage was the The EDAX Evo 50 SEM system used in the present
destructive nature of the method; it was necessary to investigation also facilitates Energy Dispersive X-ray (EDX)
physically cut the insulation material, since it is only possible spectroscopy, which allows the elemental composition of
to scan the surface of a material. This meant that the materials to be determined. Initial results have indicated that
probability of finding an anomalous particle on the surface of several elements were present within the small number of
a sample decreased significantly compared to the previous particles considered. Iron, Magnesium, Aluminum, Silicon,
technique. Oxygen, Calcium, Sulphur, Copper and Carbon were found to
The test procedure was as follows: Firstly a few, thin EPR be present in varying quantities from particle to particle. The
samples with a surface area of a few mm2 were cut from the differences in elemental composition between particles
EPR layer using a scalpel. For SEM microscopy to be indicates the material has not been synthesized and is some
effective the samples must be conducting. Charge buildup will sort of naturally occurring mined mineral ore. This suggests
manifest itself as high brightness regions, distorting the the presence of a filler material, perhaps Magnesium or
micrograph image. Clearly, EPR is an insulating material Aluminium Oxide, to bulk up the EPR at the manufacturing
particularly effective at storing charge. Prior to scanning, stage. Whether this is the case for the larger particles such as
therefore, it was necessary to gold-plate each sample to a those shown in the CT scans remains to be investigated, as
thickness of a few atoms. A standard Kimball filament was does the effect of electric field concentration on particles
used as the SEM electron source. With the chamber evacuated made of such compounds. This will be the subject of future
and the sample correctly positioned, it was found that a probe analysis.
current of 10 pA and an applied voltage of 15-20 kV produced
the best images. One of the initial results is shown in Fig. 12. 4 DISCUSSION
The embedded particle is clearly visible with more details of
The ultimate goal of the investigation is to diagnose the
the geometry now evident compared with the previous X-ray
initial cause of the observed breakdown. Analysis has
images. The particle shape is typical of those encountered in
suggested that high contaminant density, eccentricity of the
other SEM micrographs, having flat regions and several sharp
inner core, or a combination of both may have been
edges.
responsible. Previous work by the authors [11] has suggested
that the combination of field enhancement due to conducting
particle contamination and core eccentricity causing reduced
insulator thickness, has the potential to enhance the localised
field above breakdown strength for an aged insulator. It was
shown that a spherical conducting particle could increase the
electric field concentration to a maximum of 14.43 kV/mm in
the present cable geometry. Reported breakdown strengths of
EPR are in the region 30-40 kV/mm [12]–[15], with the
probability of breakdown increasing with temperature, ageing
time and moisture content. Breakdown strength for EPR has
been reported to reduce to around 20 kV/mm for ageing times
over 11000 hours and with a moisture content of around 1.4 %
and greater [12]. Results of the present study seem to confirm
this.
That the cable was operational for 26 years indicates that
inherent stress enhancement from particle content and
Figure 12. SEM micrograph of a sliced surface of EPR insulating material
from the cable sample showing an embedded high density particle.
geometric non-uniformity caused degradation of the materials.
Over the 26 year operating life, localised degradation likely
extended through the EPR until a breach was formed and
3.5 ENERGY DISPERSIVE X-RAY SPECTROSCOPY resulted in complete breakdown.
Based on results of the previous sections, the elemental The cable was manufactured according to British Standard
composition of the contaminants may only be assumed on the BS6622 [6]. Since tests for contaminants were not a
requirement, this was likely not considered at the time of ages. This could be tested through the implementation of
manufacture. BS6622 does, however, specify minimum monitoring strategies on remaining specimens in this family of
insulation thickness requirements. Examination of the cables. In addition, future work on finite element modelling
standard used at the time of manufacture reveals that the cable and testing of the breakdown strength of contaminated
does not meet these requirements. A ‘minimum average insulation material in light of the present results should
thickness’ of 3.4 mm and a ‘minimum thickness at a given contribute to cumulative research towards specifying
point’ 2.96 mm are specified. Since results show a minimum acceptable limits on the nature and levels of contaminants in
thickness at the breakdown location of 2.7 mm it must be extruded dielectric cable systems.
concluded that this was a factor leading to breakdown.
Other standards on medium voltage cable specification, 5 CONCLUSIONS
such as AEIC CS8-07 [16], do specify a maximum
This study has presented the results of an investigation into
permissible contaminant diameter and density per unit
a defective 11 kV shielded EPR power cable. Breakdown of
volume. Although the cable under inspection was not
the defective region was repeated by applying an ac voltage
manufactured according to this standard, if we compare the
between core and sheath. The fault was then successfully
stated maximum permissible particle diameter of 0.125 mm
pinpointed using time-difference-of-arrival techniques to
with the largest particle discovered through X-ray CT
isolate the origin of the breakdown pulse to within a few
scanning (0.310 mm as shown in Fig. 11 (a)), it is clear that
meters. The location of this shunt resistance fault was
the cable would also fail this initial test. Examination of the
confirmed to within a few cm using thermal imaging
small insulation sample of Fig. 11 (b) shows that the cable
techniques to identify the high temperature region produced
would also have exceeded the maximum permissible particle
by Joule heating of the low resistance fault channel when a
density requirement as stated in AEIC CS8-07 (1 particle >50
voltage was applied to the cable.
μm per cm3) if manufactured according to this standard rather
Subsequent X-ray analysis of the breakdown region and
than BS6622. Chan et al [4] correctly note that standard
concomitant through scanning electron microscopy of the
partial discharge testing of a new cable will only indicate the
surrounding EPR dielectric has suggested a possible cause of
presence of a contaminant if poor bonding occurs between the
breakdown to be degradation initiated by high field stressing
particle and the surrounding dielectric, resulting in a gas
due to particle contaminants in combination with an reduced
cavity. Although these particles may remain initially
insulation gap due to core eccentricity. X-ray computed
undetected, they can lead to breakdown in the long term
tomography has shown the presence of numerous high density
through electrical or water treeing. It may also be possible for
particles embedded in the dielectric layer. The maximum
stored charge on particles to lead to attraction of small
contaminant diameter in the region under examination was
reactive species which may exacerbate degradation.
310 μm and the average contaminant density was 1 particles
Some standards already specify the application of
per 2.28 mm3. The minimum insulation thickness in the
microscopy on new cable samples [16]. Although this
breakdown region was found to fall below the minimum
procedure is recommended as it would increase the detection
acceptable value specified in the BS6622 standard, and
rate of any manufacturing anomalies, the occurrence of non-
although no requirement for allowable contaminant levels was
uniform contaminant density or isolated contaminants means
specified in this standard, comparison with similar standards,
that the method of cutting small insulator sections at random
which do specify a minimum contaminant diameter, shows
is statistically unlikely to yield useful results. Unfortunately,
that the cable would have failed in this regard also.
only a single contaminant large enough to induce electric field
Scanning electron microscopy and energy dispersive X-ray
stressing sufficient to initiate PD or initial tree growth is
spectroscopy have shown the particle geometry in detail and
needed to initiate breakdown of an entire cable section. X-ray
suggested the elemental composition of the contaminants
methods such as those outlined in the present investigation are
consists of a metal-oxide ore filler material. In light of these
effective in detecting such contaminants, but application of the
results, it is suggested that operators of cables in the same
technique on a large scale at the manufacturing stage would be
family adopt a partial discharge monitoring strategy as this
uneconomical and labor intensive. To ensure extruded
may increase the probability of detecting the observed fault in
shielded cable reaches an operational lifespan more closely
the initial stages of degradation.
matching its designed value, it is therefore clear that, in
Future work will comprise of additional scans of the EPR
addition to stringent testing requirements, more rigorous
region. Further studies will be carried out to simulate electric
contaminant removal processes, such as those discussed in
field stressing due to particle contaminants using geometry
[17]–[19], should be employed at the manufacturing stage.
and elemental composition based in the above results.
Although strong bonding of contaminants to the surrounding
dielectric may preclude de novo PD monitoring as a detection
strategy, it is proposed that PD detection conducted at least on ACKNOWLEDGEMENT
a regular basis, if not continuously, may lead to successful This research was funded by the EPSRC through grants
detection of the effects of particle contamination as the cable EP/G029210/1 (University of Strathclyde) and EP/G028397/1
(Glasgow Caledonian University). The authors would like to [18] E. J. McMahon, “A tree growth inhibiting insulation for power cable,”
IEEE Transactions on Electrical Insulation, vol. EI-16, no. 4, pp. 304–
thank Dr. Vassilis Charissis for use of Glasgow Caledonian 318, August 1981.
University’s Virtual Reality and Simulation laboratory and Dr [19] B. Yoda, C. Ikeda, Y. Sekii, and M. Kanaoka, “Development of 500 kV
Mahesh Uttamlal for use of Glasgow Caledonian University’s cross-linked polyethylene insulated power cable,” IEEE Transaction on
Power Apparatus and Systems, vol. PAS-104, no. 1, pp. 32–38, January
Scanning Electron Microscopy facilities.
1985.

REFERENCES
[1] L. Renforth, R. Mackinlay and M. Michel, “MV cable diagnostics - Alistair J. Reid (M’11) graduated from the University of Strathclyde, UK, in
applying online PD testing and monitoring,” in Proc. MNC-CIRED Asia 2004 with a B.Eng. (Hons) degree in Electrical and Mechanical Engineering
Pacific Conference on MV Power Cable Technologies, September 2005. and received the Ph.D. degree in 2007 for research on partial discharge
[2] H. E. Orton, “Diagnostic testing of in-situ power cables: An overview,” monitoring. He conducted post-doctoral research within the Institute for
in Proc. Asia Pacific Transmission and Distribution Conference and Energy and Environment at the University of Strathclyde, from 2007-2011,
Exhibition. Yokohama, Japan: IEEE/PES, October 2002, pp. 1420– studying advanced radiometric techniques for partial discharge detection and
1425. diagnostics on underground power cables. He is presently a Research Fellow
[3] S. Boggs and J. Densley, “Fundamentals of partial discharge in the within the School of Engineering and Built Environment at Glasgow
context of field cable testing,” IEEE Electrical Insulation Magazine, vol. Caledonian University. Alistair is a Member of both the IET and IEEE.
16, no. 5, pp. 13–18, September/October 2000.
[4] J. C. Chan, P. Duffy, L. J. Hiivala, and J. Wasik, “Partial discharge VII: Martin D. Judd (M’02-SM’04) was born in Salford, England, in 1963. He
PD testing of solid dielectric cable,” IEEE Electrical Insulation graduated from the University of Hull in 1985 with a B.Sc. (Hons) degree in
Magazine, vol. 7, no. 5, pp. 9–16, Sep–Oct 1991. electronic engineering. His employment experience includes four years
[5] B. Clegg, “Underground Cable Fault Location”. McGraw Hill Book working for Marconi Electronic Devices Ltd followed by four years with EEV
Company Europe, 1993. Ltd, both in Lincoln, England. Martin received the Ph.D. degree from the
[6] British Standard BS6622: Specification for cables with extruded University of Strathclyde in 1996 for his research into the excitation of UHF
crosslinked polyethylene or ethylene propylene rubber insulation for signals by partial discharges in gas insulated switchgear. From 1999 to 2004
rated voltages from 3800/6600 V up to 19000/33000 V. London, UK: he held an EPSRC Advanced Research Fellowship concerned with
British Standards Institution (BSI), 1985. electrodynamics of electrical discharge phenomena. His fields of interest
[7] L. J. Lewis, “Traveling wave relations applicable to power-system fault include high frequency electromagnetics, generation and measurement of fast
locators,” Trans. American Institute of Electrical Engineers, vol. 70, no. transients, and partial discharges and energy harvesting. Dr. Judd is a Reader
2, pp. 1671–1680, July 1951. in the Institute for Energy and Environment at the University of Strathclyde.
[8] A. O. Ibe and B. J. Cory, “A travelling wave-based fault locator for two- He is a Chartered Engineer, a Member of the IET and Senior Member of the
and three-terminal networks,” IEEE Trans. Power Delivery, vol. 1, no. 2, IEEE.
pp. 283–288, April 1986.
[9] P. F. Gale, P. A. Crossley, B. Xu, Y. Ge, B. J. Cory, and J. R. G. Barker, Wah Hoon Siew received the B.Sc. (Hons.) degree in electronic and electrical
“Fault location based on traveling waves,” in Proc. 5th International engineering, the Ph.D. degree in electrical engineering in 1982; and the
Conference on Developments in Power System Protection, York, UK, Master of Business Administration degree from the University of Strathclyde,
Mar-Apr 1993, p. 54. Glasgow, U.K., in 1978, 1982, and 1985, respectively. Currently, he is a
[10] Z. Q. Bo, G. Weller, and M. A. Redfern, “Accurate fault location Reader in the Department of Electronic and Electrical Engineering at the
technique for distribution system using fault-generated high-frequency University of Strathclyde, Glasgow, U.K. His research interests include
transient voltage signals,” IEE Proc. Generation, Transmission and lightning protection to aircraft and grounded structures, electromagnetic
Distribution, vol. 146, no. 1, pp. 73–79, Jan 1999. compatibility in large installations due to switching and lightning, lightning,
[11] A. J. Reid, X.Hu, and M. D. Judd and W. H Siew, “Defect investigation EMP and ESD; and cable diagnostics. Dr. Siew is a Chartered Engineer and a
in medium-voltage EPR cable,” in Proc. IEEE International Symposium member of the Institution of Engineering and Technology. He is also a
on Electrical Insulation, San Juan, Puerto Rico, 10-13 June 2012. member of several CIGRE Working Groups under C4 and a member of the
[12] S. V. Nikolajevic, “The behavior of water in XLPE and EPR cables and Technical Panel for the IET Professional Network on electromagnetics.
its influence on the electric characteristics of insulation,” IEEE
Transactions on Power Delivery, vol. 14, no. 1, pp. 39–45, January Chengke Zhou (M’07) received the B.Sc. and M.Sc. degrees in electrical
1999. engineering from Hua Zhong University of Science and Technology, China in
[13] M. A. Dakka, A. Bulinski, and S. Bamji, “Correlation between DC 1983 and 1986, respectively, and the Ph.D. degree at the University of
polarization and failure characteristics of XLPE and EPR aged with ac Manchester U.K., in 1994. Since then, he worked in Glasgow Caledonian
voltage in a wet environment,” IEEE Transactions on Dielectrics and University, U.K., as a Lecturer, Senior Lecturer and in Heriot-Watt University
Electrical Insulation, vol. 12, no. 4, pp. 700–708, August 2005. as a Reader until 2007 when he went back to Glasgow Caledonian University
[14] J. C. Chan, M. D. Havtley, and L. J. Hiivala, “Performance as a Professor. He has published more than 100 papers in the area of partial
characteristics of XLPE versus EPR as insulation for high voltage discharge based condition monitoring of MV/HV plant and power system
cables,” IEEE Electrical Insulation Magazine, Vol. 9, No. 3, pp. 8-12, analysis.
May-June 1993.
[15] G. Bahder, M. Sosnowski, C. Katz, R. Eaton, and K. Klein, “Electrical Donald Hepburn (M’08) is a Senior Lecturer in the School of Engineering
breakdown characteristics and testing of high voltage XLPE and EPR and Built Environment at Glasgow Caledonian University (GCU). He has
insulated cables,” IEEE Transactions on Power Apparatus and Systems, been involved in insulation materials research and in partial discharge
vol. PAS-102, no. 7, pp. 2173–2185, July 1983. diagnostics for HV and MV system components for over 19 years. He is a
[16] “Specification for Extruded Dielectric Shielded Power Cables Rated 5 member of CIGRE Joint Working Group A2-D1.46 on transformer solid
through 46 kV”, 3rd ed. Association of Edison Illuminating Companies insulation diagnostics and of the Universities High Voltage network
(AEIC), 2007. (UHVnet) within the UK. He holds a B.A. (hons) degree from the Open
[17] S. Shinoda, K. Hikino, and M. Marumo, “275kV XLPE insulated University and a Ph.D. from GCU. Dr Hepburn is a Chartered Engineer, a
aluminum sheathed power cable for Okuyahagi No.2 power station,” Member of the IEEE, a Member of the IET and a Member of the Institute of
IEEE Transactions on Power Apparatus and Systems, vol. PAS-100, no. Physics.
3, pp. 1298–1306, March 1981.