IEEE Transactionson Dielectrics and Electrical Insulation Vol. 6 No.

1, February 1999 95

Analysis of Water Trees in Underground HV

Cables using the KFUPM ]Micro-PIXE Facility
M. Ahmed, M. A. AI-Ohali
Center for Applied Physical Sciences,King Fahd University of P?troleum& Minerals, Dhahran, Saudi Arabia

M. A. Garwaii
Physics Department, King Fahd University of Petroleun & Minerals, Dhahran, Saudi Arabia

K. AI-Soufi and S. Narasimhan

Metrology, Standard and Materials Division, King Fahd University of Petroleum& Minerals, Dhahran, Saudi Arabia

ABSTRACT
Scanning with the micro-PIXE technique was employed to analyze water trees in the XLPE in-
sulation of a field-aged underground H V cable. X-ray spectra of bow tie and vented water trees,
the inner and outer semiconductive compounds, and an insulation spot free from any water tree
were acquired. Simultaneously, two-dimensional elemental distribution profiles across the wa-
ter trees were also measured. Various trace element impurities were identified in the analyzed
spots and their possible sources are suggested. Differences in elemental distribution profiles
in the scanned areas were observed and have been disciissed on the basis of the mechanism of
incorporation of these elements into the insulation. This study demonstrates the effectiveness
of the micro-PIxE facility available in this laboratory in analyzing water trees in underground
power cables.

1 INTRODUCTION lem encountered by the utility sector in Saudi Arabia [16]. The average
life of ,I typical HV cable in this country is considerably shorter com-

U NDERGROUND HV cables in certain locations tend to deteriorate pared to the 15 to 20 yr life in industrialized countries. The reason for
prematurely resulting in costly replacements. One of the major the short life can be attributed to the severe environmental conditions
causes of degradation has been attributed to the growth of water trees characteristic of the arid regions and coastal areas near the sea. Water
in polyethylene insulation of the cables. Although much literature on trees ar: a common occurrence in field-aged cables in Saudi Arabia [16].
the phenomenon of water trees exists in journals dealing with electri- An uncerstanding of the role of water trees in cable breakdown can be
cal insulation [l-lo], very few studies [ll-151 have been reported deal- useful io the cable industries in Saudi Arabia to develop better quality
ing with trace element microscopy of the water trees to understand the cables with improved life span. It can also help local utility companies
mechanism of their growth and propagation. The water trees are bush- to draw up more stringent site selection criteria for laying underground
like hydrophilic structures of microscopic dimensions. Water-soluble HV cables.
impurities around the cable penetrate the insulation through the wa-
ter trees, thereby affecting its dielectric and mechanical properties. The Water tree structures have typical dimensions of -500 pm,and can
common contaminants in water trees are mostly inorganic salts of Na, contain impurity elements in concentrations as low as in the parts per
Mg, Al, Si, S, C1, K and Ca, although some heavy elements were also million range [ll].The micro-FIXE technique [17] is capable of measur-
detected (111. The presence of impurities in water trees can be affected ing trace element concentrations at ppm levels and their distribution
by several factors including soil and environmental conditions, and the profiles with p m spatial resolution. The technique therefore has a great
type and power ratings of the cables. Distribution profiles of the con- potenti: 1 for non-destructive analysis of water trees in underground HV
taminants across water trees can provide valuable information on how cables.
these elements are incorporated and the way they affect cable perfor-
The :;canning proton microprobe facility in this laboratory [MI, in-
mance. Determination of the impurity elements and their concentration
stalled on the 3 MV Tandetron accelerator ca. 5 years ago, has been ap-
profiles across the water trees can therefore be important in understand-
plied successfully in a wide variety of fields, including geology [19],
ing the mechanism of cable breakdown in a particular locality.
petrochlrmical catalysts [ZO], metallurgy [21], and high temperature su-
Premature breakdown of underground HV cables is a frequent prob- perconcuctors [22].

1070-9878/99/ $3.00 01999 IEEE

 96 Ahmed et al.: Water Trees in Underground HV Cables

Typical resolution of the microbeam for 2.5 MeV protons is -4 pm at sections having water trees were coated with a thin layer of carbon to
a current of 50 to 100 pA and the maximum scanning range is ~ 0 . mm
5 make the analyzed surface electrically conducting. Topography of the
in each of the X and Y directions. A secondary electron detection sys- water trees was photographed using a scanning electron microscope
tem is available to position the microbeam quickly on any region of in- (SEM).
terest on the sample. From the scanned area of the sample, an average Target
elemental composition spectrum as well as multiple elemental distribu-
tion maps in color can be produced simultaneously.
2 EXPERIMENTAL METHOD
2.1 SAMPLE PREPARATION
Samples of field-aged underground HV cables were obtained from a
Saudi utility company in the Eastern Province. The cable had a voltage
rating of 35 kV, a single core copper conductor with a cross sectional
Amplifier1
area of 500 mm', and an XLPE insulator of thickness 1 cm. The cable
was imported from abroad and unfortunately no further data such as the
extrusion process, polymerization and crosslinking procedures, treat- 0

ments and additives, etc. which might have been useful in interpreting x-ray energy
the results could be obtained. The particular cable section from which
the sample was prepared was in service in the Jubail industrial area of Elemental ID Data Elemental
the Eastern Province close to the Arabian Gulf and located near a gas Spectrum Acquisition Distribution Map
station. According to the supplier of the cable sample, mechanical dam- System
age was observed at a cable junction box which allowed water to ingress Figure 2. A schematic of the micro-PIXE set up used for analyzing the
into the cable and travel laterally along the core, producing the observed HV cable sample.
water trees. Because of this lateral seeping along the core, there was no
appreciable bloating of the PVC jacket. However, there was a charac':er- 2.2 MICRO-PIXE MEASUREMENTS
istic gasoline smell in the cable sample collected. A layout of the micro-PIxE set up used for the measurements on wa-
Outer ter trees in the samples is shown schematically in Figure 2. A 2.5 MeV
Semicon proton beam of -4 p m resolution was used to scan bow-tie and vented
Lay$ water trees. X-ray elemental composition spectra averaged over each
0.5 mm2 scanned area, covering the water trees, were acquired. Simul-

-
taneously, two-dimensional concentration distribution maps for several
elements present in the trees were obtained. In addition, spot measure-
ments on the inner and outer semiconductive compound layers, and on a
region of the XLPE insulation free from any water tree, were also carried
out to acquire elemental composition spectra. The quantitative analysis
Inner
software GUPIX [23] was used to analyze the X-ray spectra in order to
Semic
Layer estimate relative elemental concentrations.
Outer Semicon

PVC Jacket
Inner Semicon Vented Tree
power cable
Figure 1. studied,
A schematic
showing
of the
different
cross section
components.
of the 35 kV underground

A cross sectional view of the cable is shown schematicallv in Figure 1.

. I "

The co]lected cable length was cut into small pieces. The conductor core Figure 3. A schematic of a thin section of the XLPE insulation from the
cable under study showing the analyzed regions (0).
and all layers except the inner and outer semiconductive compounds
surrounding the X ~ P insulation,
E were removed. Each small piece Icon-
3 RESULTS AND DISCUSSIONS
taining the insulation and the semiconductive layers was microtoined
into thin sections (-100 p m thick). The thin seciions were first exam- A schematic representation of a thin section of the XLPE insulation
ined under an optical microscope for the presence of water trees. The from the HV cable under study is shown in Figure 3. Several bow-tie
IEEE Transactions on Dielectrics and Electrical Insulation Vol.6 No.1, February 1999 97

and vented water trees could be seen under an optical microscope. SEM
photo-micrographs of a bow-tie water tree and a vented water tree in the
thin section which were analyzed by micro-PIXE technique are shown in
Figure 4.
polymerization of ethylene which forms the base of the cable insulation
[HI. The presence of S is most likely due to the antioxidants used in the
manufacture of the resin [ll],as it-was observed throughout the insu-
lation, both inside and outside the water trees. The origin of Pb is most
probably the leaded gasoline from the nearby gas station. A leak in the
gas station reservoir might have contaminated the soil around the cable.
As mentioned earlier, there was a faint characteristic gasoline smell in
the collected samples.

1
100000

t s
10000

co"c'"\
100
Y Si 1

Ph

~~

'0 100 200 300 400 500 600 700 800 900 1000
Channel Number

Figure 6. Micro-FIXE elemental spectrum from a water tree-free spot in

the XLPE insulation.
A similar elemental spectrum was obtained from a spot in the XLPE
insulation away from any water tree (Figure 6), however with a much
reduced Pb/S ratio. It indicates that water trees tend to attract and con-
centrate trace element impurities such as Pb.

10000

Figure 4. SEM photo-micrographs of a vented water tree (a) and a bow-

1000
tie water tree (b) analyzed by the micro-P1XE technique.

100000

10000

Counts
Ph
Channel Number

Figure 7. Micro-PIXE elemental spectrum averaged over the scanned

10
area of the vented water tree.

1
Channel Number
Figure 5. Micro-PIXE elemental spectrum averaged over the scanned
area of the bow-tie water tree.
Figure 5 shows the micro-PIXE elemental composition spectrum av-
eraged over the scanned area on the bow-tie water tree. The spectrum
shows the presence of Si, S and Pb as trace element impurities. The
source of Si is probably the silica-based catalyst normally used in the
Figure 7 shows a PIXE spectrum averaged over the scanned area on
the vented water tree. This particular vented tree grew from the con-
ductor side of the cable. The spectrum shows the presence of Si, s, Cu
and Pb. The major difference between this spectrum and the spectrum
from the bow-tie tree (Figure 5 ) is the presence of a large Cu peak in the
vented tree, while the bow-tie tree did not show the presence of any Cu.
The source of Cu in the vented tree is most likely the result of diffusion
from the copper conductor. The origins of Si, S and Pb have been dis-
cussed earlier in connection with the bow-tie tree.
98 Ahmed et al.: Water Trees in Underground HV Cables

Table 1. Relative concentrations of elements in the HV cable sample at

various locations.
oooot
t
1000 :micon semicon
0.05 2.0

0.02 0.22
0.20
Mn 0.03

I
Fe 0.12
Ni 0.20
cu 0.98 0.33 -
Zn
Channel Number LPb 1 1.4 2.1 I 2.6 1.0 -

Figure 8. Micro-PIXE elemental spectrum from a spot in the inner semi-

conductive layer.
The micro-PIXE elemental spectrum from a spot in the inner semicm-
ductive compound is shown in Figure 8. In addition to Si, S, Cu and Pb, a
trace of Ca could be detected which probably arises from an impurity in
the semiconductive compound itself. The relatively smaller Cu/S ra:io
here compared to that in the vented water tree is somewhat surprising
if the presence of Cu in the XLPE insulation is the result of diffusion of
Cu from the conductor. It is possible that the vented tree, because of its Figure IO. Micro-PIXE elementaldistribution maps from a 0.5"' cov-
ering the bow-tie water tree. Darker shades indicate higher concentrations.
a f h t y for water, might have accumulated Cu from a water-soluble ccp-
per salt in the soil. This possibility then implies that the water solutde bow-tie tree are shown in Figure 10 for S and Pb. Darker shades in these
impurities entered the cable through the conductor channel. maps indicate higher concentrations. The spatial correlation between
these two elements can be seen in the maps. Although the sources of
these elements are thought to be different (S arising from the antioxi-
looooolt 1
dants used in the manufacture of the XLPE resin, while leakage of leaded
gasoline from the nearby gas station is the possible source of Pb), the di-
1000

1 sis7
rect correlation observed in their distributions suggests that these two
elements have a chemical affinity for each other, forming a PbS com-
pound.

L-.---.---L .. L... . I
'0 100 200 300 400 500 600 700 800 900 1000
Channel Numbei Figure 11. Micro-PIXE elemental distribution maps from a 0.5 mm2 area
covering the vented water tree. Darker shades indicate higher concentra-
Figure 9. Micro-PIXE elemental spectrum from a spot in the outer semi- tions.
conductive layer.
The elemental distribution maps from a 0.5 mm2 area covering the
No Cu could be detected in the micro-PIXE spectrum from a spot in vented water tree are shown in Figure 11 for S, Cu and Pb. Again one
the outer semiconductive compound as shown in Figure 9. This is ex- can observe a positive correlation between the distributions of S and Pb,
pected, because this layer is physically separated from the conductor by similar to their distributions in the bow-tie tree. However, Cu appears to
the insulation. On the other hand, the outer semicon spectrum shows be mostly concentrated around the lower part of the tree near the copper
several additional trace element impurities such as V, Mn, Fe, Ni arid conductor core. This is expected on the basis of diffusion of the metal
Zn. These metallic impurities may have originated from the adjacent from the conductor core, as this analyzed vented tree grew from the con-
copper wire mesh. ductor side.
The relative concentrations of elements in the analyzed areas in the The present study demonstrates the effectiveness of the micro-PIXE
bow-tie water tree, the vented water tree, the XLPE insulation, and the facility available in this laboratory in analyzing microscopic water trees
inner and outer semiconductive compounds are given in Table 1. that develop in some underground HV cables. Results obtained show
The elemental distribution maps from a 0.5 mm2 area covering the that measurements of trace element impurities and their distribution
IEEE Transactions on Dielectrics and Electrical Insulation
profiles across water trees can provide valuable data in understanding
the role of the water trees in premature degradation of these cables.
ACKNOWLEDGMENT
The work was carried out using the microbeam facility at the Re-
search Institute, King Fahd University of Petroleum & Minerals, Dhahran.
Thanks are due to Saudi Consolidated Electric company (SCECO),
Jubail, for supplying the cable The assistance of Madhusood-
hanan Pillai, Tandetron Accelerator Operator is acknowledged.
REFERENCES
M. T. Shaw and S.H. Shaw, "Water Treeing in Solid Dielectrics", IEEE Transactions
on Electrical Insulation, Vol. 19, pp. 419452,1984.
W. Golz, "Water-tree Growth in Low Density Polyethylene", Colloid and Polymer
Science, Vol. 263, pp. 286-292,1985.
E. E Steennis, "Water Treeing in Polymer Cable Insulation", Kema Scientific & Tech-
nical Reports, Vol. 8 (3), pp. 149-208,1990.
R. Ross and J. J. Smit, "Composition and Growth of Water Trees in XLPE", IEEE
Transactions on Electrical Insulation, Vol. 27, pp. 519-531,1992.
J. D. Cross and J. Y. Koo, "Some Observations on the Structure of Water Trees", IEEE
Transactions on Electrical Insulation, Vol. 19, pp. 303-306,1984.
J. C. Filippini and C. T. Meyer, "Water Treeing Test using the Water Needle Method",
IEEE Transactions on Electrical Insulation, Vol. 23, pp. 275278,1988.
N. Shimizu, K. Uchida and S. Rasikawan, "Electrical Tree and Deteriorated Region in
Polyethylene", IEEE Transactionson Electrical Insulation, Vol. 27, pp. 515518,1992.
S. S.Bamji, A. T. Bulinski, Y.Chen and R. J. Densley, "Threshold Voltage for Electri-
cal Tree Inception in Underground HV transmission cables", IEEE Transactions on [22]M. Faiz, M. Ahmed and M. A. AI-Ohali, "Micro-PIXE analysis of a single crystal of
Electrical Insulation, Vol. 27, pp. 402404,1992.
L. A. Dissado, S. V. Wolfe, J. C. Filippini, C. T. Meyer and J. C. Fothergill, "An Analy- [23] J. A. Maxwell, J. L. Campbell and W. J. Teesdale, "The Guelph PIXE software pack-
sis of Field-Dependent Water Tree Growth Models", IEEE Transactions on Electrical
. I 345-356,1988.
Insulation, Vol. 23, DD.
[lo] R. Patsch, "Composition and Growth of Water Trees in XLPE", IEEE Transactionson Manuscript was received on 20 May 1998, in revised form 9 October 1998.
Electrical Insulation, Vol. 26, pp. 532-542, 1992.
Vol. 6 No. 1, February 1999 99
[ill l? F. Hinrichsen, G. Kajrys, A. J. Houdayer, A. Jeremie, A. Belhadfa, J. l? Crine, and
J. L. Campbell, "Micro-PEE Analysis of Impurity Distributions in "Trees" Grown in
HV cables", Nuclear Instruments and Methods, Vol. B 45, pp. 532-535, 1990.
[12] A. Belhadfa, A. J. Houdayer, !I E Hinrichsen, G. Kajrys, J. St-Pierre, G. Kennedy, J.
l? Crine and N. Burns, "Impurities in semiconductive compounds used as HV cable
shields", IEEE Transactions on Electrical Insulation, Vol. 24, PP. 709-712,1989.
A. 1. Houdayer, Private m " i c a t i o n , 1992.
[14] A. Belhadfa, A. Houdayer, l? F. Hinrichsen and G. Kajrys, "Migration of impuri-
ties from semicon shield thorough PE insulation under various experimental condi-
tions", IEEE Symposium on Electrical Insulation, Toronto, Canada, June 34,1990.
(151 I? F. Hinrichsen, A. Houdayer, A. Belhadfa, J.-I? Crine, S.Pelissou and M. Cholewa,
"A localized trace element analysis of water trees in XLPE cable insulation by Micro-
PIXE and EDX", IEEE Transactions on Electrical Insulation, Vol. 23, pp. 971-978,
1988.
[16] Personal communication with Saudi Consolidated ElectricCompany (SCECO),East-
ern Region, Saudi Arabia.
[17] S.A. E. Johansson and J. L. Campbell, P I X E A Novel Tecliniqiiefor Elemental Analysis,
Wiley, 1988.
[18] M. Ahmed, J. Nickel, A. B. Hallak, R. E. Abdel-Aal, A. Coban, H. A. Al-Juwair, and
M. A. Aldaous, "The KFUPM scanning nuclear microprobe facility", Nucl. Inst. &
Meth., Vol. B 82, pp. 584-588,1993.
[19] N. Khandaker, M. Ahmed and M. Ganvan, "Study of volcanic sediments by micro
beam-PIXE technique", Nucl. Instr. & Meth, Vol. B 109, pp. 587-591,1996.
[20]M. Ahmed, A. Rahman, J. Nickel and M. A. Ganvan, "Micro-PIXE measurement of
nickel distribution is supported nickel oxide catalyst", Nucl. Instr. & Meth, Vol. B
103, pp. 233-236,1995.
[21] A. N. Shuaib, J. Nickel and M. Ahmed 1995, "Wear studies of tungsten carbide cut-
ting tools using th Micro-PLXE technique", Nucl. Instr. & Meth, Vol. B 103, pp. 68-72,
1995.
Y1 Ba2 Cu3 07-8 superconductor", Nucl. Instr. & Meth, Vol. B 114, pp. 138-142,1996.
age", Nucl. Instr. & Meth, Vol. B 43, p. 218, 1988.