IEEE Transactions on Dielectrics and Electrical Insulation Vol. 22, No.

5; October 2015 2841

Electrical Treeing Behavior at High Temperature

in XLPE Cable Insulation Samples
Xiangrong Chen1,2 , Yang Xu2 , Xiaolong Cao2 and S. M. Gubanski1
1
High Voltage Engineering, Department of Materials and Manufacturing Technology
Chalmers University of Technology
SE 412 96 Göteborg, Sweden
2
State Key Laboratory of Electrical Insulation and Power Equipment
Xi’an Jiaotong University
Xi’an 710049, China

ABSTRACT
This paper reports on investigations aiming to understand the behavior of electrical
trees formed in XLPE cable insulation at elevated temperatures. Samples cut from
a real high voltage cable equipped with an embedded electrode system and
composed of a needle separated by 2 mm from counter electrode were used. The
temperature was varied between 50 and 90 C and the range of voltage changed
between 9 and 18 kV. The electrical treeing process was followed by means of
microscopic observation from the moment of initiation until the leading branches
connected the counter electrode and partial discharge (PD) activity was monitored
simultaneously. The shape and fractal dimension of the resulting trees were
analyzed. It was found that temperature and voltage level had pronounced effect on
the process of electrical tree formation. At lower voltages different type of trees
could be formed, whereas at higher voltages branch trees dominated. Increase of
test temperature reduced the time for tree initiation, whereas the tree growth time
was less affected, except for the lowest voltage level at which the slowest tree
development was observed at the highest test temperature. It is postulated that the
impeding effect of elevated temperature on tree growth is due to a change of tree
channel sidewall conductivity, whereas the observed reduction of tree initiation
time with the increase of temperature and voltage level is attributed to lowering of
the threshold energy for damage at higher temperature as well as increasing of the
injection current at higher voltage level.
Index Terms - Electrical treeing, high temperature, XLPE cable insulation,
partial discharge, initiation time, growth time.

1 INTRODUCTION of its conductor can be as high as 90 C, the usual operating

temperature is maintained at 50-60 C [13]. However, the
ELECTRICAL treeing is known to be one of the routes
growing demand for upgrading and uprating of the existing
to cause failure in polymeric cable insulation [1] by initiating
cable systems [14], the knowledge on XLPE insulation
material degradation from points of electric stress
performance at higher temperatures ( 50 C) becomes very
enhancements, e.g. from voids, protrusions [2] and water trees
important.
[3, 4]. Initiation and growth of electrical trees is accompanied
The aim of the work presented here was therefore to
by partial discharge (PD) activity within the developing
elucidate the influence of elevated temperature on electrical
branches [5-10].
treeing characteristics in sectioned XLPE cable insulation
With the development of industrialization in China, the
over a range of ac voltage (50 Hz) level. The trees were
demand for high/extra high voltage XLPE-insulated cable
networks is growing at a fast rate. Therefore, the generated at 50, 70 and 90 C and the voltage ranged from 9
degradation of cable insulation caused by electrical treeing to 18 kV. The treeing process was recorded by a real-time
has been of a great concern [11, 12]. Despite that for an microscope detection system and PD measurements were
extruded XLPE power cable the rated loading temperature performed simultaneously throughout the tree growth
process. Fractal dimension analysis [15] was used to
Manuscript received on 27 June 2014, in final form 18 February 2015, quantitatively analyze two-dimensionally projected tree
accepted 19 February 2015. images.

DOI 10.1109/TDEI.2015.004784
2842 X. Chen et al.: Electrical Treeing Behavior at High Temperature in XLPE Cable Insulation Samples

2 EXPERIMENTAL
2.1 SAMPLE PREPARATION
Semi-circular insulation sections, 5 mm wide, were cut
from a new commercially manufactured high voltage extruded
XLPE cable by mechanically removing the outer semi-
conducting layer and the conductor, while leaving the inner
semi-conducting layer with a thickness of 0.5 mm and the
insulation with a thickness of 15 mm. A steel needle with a tip
radius of 51 μm was thereafter slowly inserted at a
temperature of 125 C into each of the XPLE insulation
sample, providing the separation between its tip and the semi-
conducting layer of 2±0.2 mm, as illustrated in Figure 1. After
the insertion, each sample was annealed at the elevated Figure 3. View of temperature controlled test cell and the real-time microscope
observation system.
temperature for approximately 6 min for reducing the
mechanical stress build up around the needle and then cooled
the recommendations of IEC 60270 [17]. However, for
down to room temperature. Only samples without the
avoiding false recordings, the threshold of PD detection was
presence of mechanical stress and void were used in the
set during the experiments to 10 pC. Sampling of PD pulse
experiments. More detailed information on the sample
series was performed by recording the activity during 500
preparation procedure can be found in [16].
power frequency cycles (10s).
A schematic view of the used microscope based real-time
digital imaging system and the temperature control system are
shown in Figure 3. The detailed information about the
microscopic observation system is described in [16]. The
temperature control system was based on an oil filled bath, in
which the sample was kept in a specially manufactured test
cell allowing for microscopic observations. This arrangement
allowed for optical observations of the treeing process at
elevated temperatures while simultaneously measuring the PD
activity.

3 RESULTS
The experiments were carried out at the elevated
Figure 1. Schematic view of test sample with inserted needle electrode.
temperatures of 50, 70 and 90 C. Each of the temperature
levels was stabilized for 30 minutes before voltage application.
In order to obtain statistically significant results, 10 samples
2.2 EXPERIMENTAL SETUP were tested at 9, 11, 13, 15 and 18 kV rms of the applied
Figure 2 shows the experimental setup, consisting of a voltage. Both the tree initiation and its propagation were
variable autotransformer T1, a power frequency 50 Hz step-up recorded during the time of voltage application. The length of
transformer T2, a 10 kΩ protection resistor R, a 1:1000 (C1:C2) tree (L) was measured as the furthest tree extent from the
capacitive voltage divider and a 2000 pF shunt capacitance Ck. needle electrode. The fractal dimension of electrical trees was
A RC detection impedance Z was used in series with the estimated by software developed by the box-counting method
sample Cx, the latter being placed in a temperature control [15, 18]. The detailed information of the calculation process
unit. The detection impedance was connected via a coaxial can be found in [16, 19].
cable to an amplifier (A) and a PD detection system with a
sensitivity of 2 pC, which was developed in accordance with 3.1 APPEARANCE OF ELECTRICAL TREES AT 50 OC
Figure 4 shows a set of electrical trees typically appearing at
the temperature of 50 C. Shapes of the trees grown at 9 kV
can be characterized as branch trees (Figure 4a) and ‘branch-
pine’ trees (Figure 4b), as defined in [12], accounting
respectively for about 40% and 60% in the total tree number
(10 samples). The electrical tree shape at 11 and 13 kV are
branch trees (Figures 4c and 4d), evolving gradually at higher
voltages into bush trees. At 15 kV branch tree (Figure 4e) and
bush tree (Figure 4f) are found, accounting respectively for
about 70% and 30%, while at 18 kV all the trees are of bush
Figure 2. Experimental setup including PD detection system. type (Figure 4g).
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 22, No. 5; October 2015 2843

(a) region A (b) region B (c) region C

(a) branch tree (b) branch-pine tree Figure 6. Propagation of the branch-pine tree at 9 kV.

In contrast, three distinct regions can be distinguished in the

growth of the branch-pine trees (Figure 4b). Typical tree images
during the tree propagation are shown in Figure 6. In region A,
the tree extends initially as a branch structure (first 2 min).
Thereafter, in region B, it grows stepwise with leading branch
ceasing at each step, meanwhile dense micro-channels appear.
(c) branch tree (d) branch tree The PD magnitude increases with the tree growth at first.
However, after about 12 min, the PD magnitude starts to decrease
and a transfer of the tree structure occurs. At about 20 min a pine
structure develops from the main branch channel outwards,
meanwhile the PD magnitude declines to below 10 pC. The PD
magnitude becomes again larger than 10 pC after 24 min. In
region C, the branch-pine tree grows rapidly towards the counter
electrode. The average growth rate in region C is 157 μm/min.
(e) branch tree (f) bush tree
Figure 7 shows the time evolution of the tree growth of the
typical branch trees at 11, 13, and 15 kV and PD
characteristics of the tree at 11 kV. It can be seen that the tree
growth time decreases with an increase of the stressing
voltage. The branch tree (Figure 4c) at 11 kV grows rapidly to
the counter electrode with an average rate of 436 µm/min. The
PD magnitude increases substantially with the tree growth.
(g) bush tree The discharge activity during the tree growth is intensive with
Figure 4. Influence of voltage level on appearance of electrical trees at 50 C. a wing-like pattern, as shown in Figure 8a. The average
growth rates of branch trees at 13 and 15 kV are 735 and 1053
The variations of the tree length with time and the µm/min respectively. After the tree initiation, the PD activity
magnitude of largest PDs detected (Qmax) of the trees grown at of the branch trees at 15 kV is more intensive than the activity
9 kV are shown in Figure 5, where 9 kVaL, 9 kVaQmax, 9 at 13 kV (note the scale difference in Figures 8a, 8b and 8c).
kVbL and 9 kVbQmax correspond to the length and PD activity
of the trees depicted in Figures 4a and 4b, the similar
corresponding expressions are also defined in following
figures. For the branch tree (Figure 4a), the growth proceeds
rapidly after the initiation until it reaches the counter
electrode. The PD magnitude increases consistently with the
tree growth, which proceeds at a rate of 246 µm/min.

Figure 7. Evolution of electrical tree length with propagation time at 11, 13

and 15 kV and related PD activity of the tree at 11 kV and temperature of 50
C.

(a) 4 min (b) after tree initiation (c) after tree initiation
Figure 5. Evolution of electrical tree length with propagation time and related
PD activity at 9 kV and temperature of 50 C. Figure 8. PD patterns of branch trees at 11 kV (a), 13 kV (b) and 15 kV (c).
 2844 X. Chen et al.: Electrical Treeing Behavior at High Temperature in XLPE Cable Insulation Samples

The growth of bush trees at 15 and 18 kV and the associated in which the time to breakdown (TTB) appears as a sum of the
PD characteristics of the tree at 15 kV are shown in Figure 9. initiation time (the time from the voltage application to the
The three regions A, B and C are clearly defined by the growth appearance of the first tree channel) and the growth time (the
curve of the bush tree at 15 kV. Typical images illustrating the time after tree initiation to the leading channel connecting the
tree development are shown in Figure 10. In region A, the counter electrode). The presented results indicate that the tree
initial tree grows rapidly with a branch structure in the first 30 s. shape can be diverse, with initiation time, TTB, growth rate
Thereafter, the densely packed structure starts to appear. The and fractal dimension gradually changing with stressing
tree growth rate decelerates slightly with the growth of the voltage. The bush trees at 18 kV have the shortest initiation
densely packed structure, while the PD magnitude increases time and TTB. The branch-pine trees at 9 kV have the longest
rapidly. The average growth rate during this time is 632 μm/min. initiation time and TTB. The initiation time and TTB of
In region B, the densely packed structure grows continuously branch tree at 11, 13 and 15 kV decrease continuously with
towards the counter electrode with an average growth rate of 89 increasing the voltage level. The bush trees at 15 kV have
μm/min. The PD activity in region B is intensive with a wing- much longer initiation time and TTB than the bush trees at 18
like pattern (see Figure 11a). In region C, a branch develops in kV. As compared to room temperature results presented in
the direction of the counter electrode from the boundary of the [16], the TTB of electrical tree at 50 C decreased significantly.
densely packed structure. The bush trees at 18 kV also initiate
with a branch structure, turning into the densely packed Table 1. Statistical results of electrical tree characteristics at 50 C
structure after 15 s. Thereafter rapidly growing branches extend Fractal
Initiat Average
from the boundary of the densely packed structure. The average Voltage Electrical -ion TTB growth
dimension
growth rate of the bush tree at 18 kV is 1367 μm/min. The PD of
(kV) tree shape time (min) rate
projected
pattern (Figure 11b) shows that the discharge activity is (min) (μm/min)
tree
intensive after the tree initiation. Branch tree 3.2 21.3 221 1.4
The types of the electrical trees and the growth 9 Branch-
24 72 43 1.62
pine tree
characteristics observed at 50 C are summarized in Table 1, 11 18.6 24.7 386 1.4
Branch tree
13 5 9.4 661 1.62
Branch tree 3.5 5.3 1095 1.63
15
Bush tree 21 29.6 416 1.79
18 Bush tree 2.8 4.5 1627 1.73

O
3.2 APPEARANCE OF ELECTRICAL TREES AT 70 C
The electrical trees at 70 C are predominantly branch trees,
as shown in Figure 12 for experiments performed at different
voltage levels. Similar result was as earlier found by Densley

Figure 9. Evolution of electrical tree length with propagation time of bush

trees at 15 and 18 kV and related PD activity of the tree at 15 kV and
temperature of 50 C.

(a) branch tree (b) branch tree

(a) region A (b) region B (c) region C

Figure 10. Propagation of the bush tree at 15 kV.

(c) branch tree (d) branch tree

(a) 9 min (b) after tree initiation

(e) branch tree (f) branch tree after breakdown
Figure 11. PD patterns of bush trees at 15 kV (a) and 18 kV (b). Figure 12. Influence of voltage level on appearance of electrical trees at 70 C.
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 22, No. 5; October 2015 2845

[20]. Evolution of the tree growth at 9 and 11 kV together

with the related PD characteristics are shown in Figure
13.Three regions can be distinguished for the tree grown at 9
kV. Typical tree images during the tree propagation are
shown in Figure 14. In region A, the tree grows rapidly with a
branch structure in the first 2 min period. The average growth
rate during this time is 303 μm/min. In region B, the
propagation of the leading channel is firstly stopped for 12
min. Some micro-channels extend gradually around the main
channels near the needle electrode. The tree channels become
dense during this time. The PD magnitude remains between
70 pC and 100 pC during the first 14 min of the tree growth.
The PD magnitude starts to increase apparently after 14 min. Figure 15. Evolution of electrical tree length with propagation time at 13, 15
Meanwhile, the new branch channels formed from the dense and 18 kV and temperature of 70 C.
region start to grow continuously towards the counter
electrode. In region C, the leading branch grows rapidly to
connect the counter electrode. The average growth rate in
region C is 130 μm/min. At 11 kV the three regions cannot be
recognized and the tree grows rapidly towards the counter
electrode after being initiated. The average growth rate is 560
μm/min and the PD magnitude increases substantially. (a) 13 kV (b) 15 kV (c) 18 kV

Figure 16. PD patterns of branch trees at 13 kV, 15 kV and 18 kV.

Table 2 shows a summary of the results obtained at 70 C

for the test performed at different voltage levels. They show
that initiation time and TTB decrease consistently with an
increase of the voltage, whereas the growth rate accelerates.
The fractal dimension of the trees, which is a measure of the
density of damage within a tree structure [15, 21] shows an
increase between 11 kV and 18 kV.
Table 2. Statistical results of electrical tree characteristics at 70 C
Fractal
Initiat Average
dimension
Voltage Electrical -ion TTB growth
of
(kV) tree shape time (min) rate
Figure 13. Evolution of electrical tree length with propagation time and projected
(min) (μm/min)
related PD activity at 9 and 11 kV and temperature of 70 C. tree
9 18.6 33.7 166 1.47
11 6.2 10.3 563 1.44
13 Branch tree 3.4 5.5 1200 1.46
15 2.3 3.2 2799 1.49
18 0.8 1.3 5565 1.56

3.3 APPEARANCE OF ELECTRICAL TREES AT 90 OC

Figure 17 shows a typical set of electrical trees grown at 90
(a) region A (b) region B (c) region C
C. The electrical tree shape at 9 kV is characterized by a
Figure 14. Propagation of the branch tree at 9 kV. mixed structure of bush and branch structure (referred further
to as bush-branch tree). It is narrow and dense, much different
Figure 15 shows the development of the tree length with time from the trees observed at 50 and 70 C. The electrical trees at
of the representative branch trees at 13, 15 and 18 kV. The higher voltages show a branch-like structure with an
equivalent PD patterns are shown in Figure 16. The results ‘umbrella’ shape. The evolution of electrical tree length and
show that the time of tree growth decreases significantly with the respective PD characteristics of typical trees grown at 9
the increase of the applied voltage and the PD activity during and 11 kV are shown in Figure 18. Three distinct growth
the tree propagation becomes more intensive. The branch tree regions of the bush-branch tree at 9 kV can again be seen here.
grows explosively at 18 kV (see Figure 12e) in the direction Typical tree images during the tree propagation are shown in
of the counter electrode. In the case illustrated in the Figure Figure 19. In region A, an initially filamentary tree grows
10f, breakdown took place 15 s after the tree initiation. The during 1 min in the direction perpendicular to the needle
PD pattern (Figure 16c) shows that the discharge activity electrode. Thereafter, a few channels start to extend in the
prior to the breakdown becomes very insensitive and takes field direction near the needle electrode. The discharge activity
place continuously, independently of the voltage phase. is low in this region and the average growth rate during this
 2846 X. Chen et al.: Electrical Treeing Behavior at High Temperature in XLPE Cable Insulation Samples

continually during this time. The discharge activity becomes

intensive when the leading channels meet the counter electrode.
The average growth rate in region C is 37 μm/min. The branch
tree at 11 kV grows first in the perpendicular direction also.
Thereafter, it starts to extend rapidly in the direction of the
field towards the counter electrode, but some of the tree
channels still grow perpendicularly, as shown in Figure14b.
(a) bush-branch tree (b) branch tree The average growth rate is 446 μm/min. The PD magnitude
increases consistently with the tree growth (see Figure 18).
The growth for the representative branch trees at 13, 15 and
18 kV and PD characteristics of the tree at 13 kV are shown in
Figure 20. It can be seen that the tree growth is accelerated
significantly with the increase of the voltage. After the trees
initiation, all the trees grow firstly perpendicular to the needle-
(c) branch tree (d) branch tree
plane direction. Thereafter, the trees develop along the field
with some channels extending perpendicularly. The PD
patterns of the branch trees at 13, 15 and 18 kV after the tree
initiation are shown in Figure 21.

(e) branch tree

Figure 17. Influence of voltage level on appearance of electrical trees at 90 C.

Figure 20. Evolution of electrical tree length with propagation time at 13, 15
and 18 kV and related PD activity of the tree at 13 kV and temperature of 90
C.

Figure 18. Evolution of electrical tree length with propagation time and
related PD activity at 9 and 11 kV and temperature of 90 C.

(a) 13 kV (b) 15 kV (c) 18 kV

Figure 21. PD patterns of branch trees at 13, 15 and 18 kV.

The results obtained at 90 C are summarized in Table 3 and

show that the initiation time of the electrical tree decreases
significantly between 9 kV and 18 kV. A monotonic behavior
(a) region A (b) region B (c) region C
of TTB is also evident. However, the fractal dimension of the
Figure 19. Propagation of the bush-branch tree at 9 kV.
Table 3. Statistical results of electrical tree characteristics at 90 C
time is 84 μm/min. This stage is followed by a slow Fractal
propagation region B, where the leading branch of the tree Initiat Average
dimension
Voltage Electrical -ion TTB growth
grows slowly towards the counter electrode with a rate of 4.5 (kV) tree shape time (min) rate
of
μm/min. Meanwhile, some micro-channels extend gradually projected
(min) (μm/min)
tree
to the sides of the main channels to form a densely packed Bush-
structure. The PD magnitude increases slowly from 40 pC up 9 9.4 117 25 1.58
branch tree
to 70 pC. In region C, some branch channels develop from the 11 0.6 5.1 564 1.43
edge of the densely packed structure and the tree growth 13
Branch tree
0.66 2.8 1510 1.4
accelerates. Finally, the leading branches extend rapidly to 15 0.5 2.1 2460 1.49
18 0.14 0.5 7677 1.4
connect the counter electrode. The PD magnitude increases
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 22, No. 5; October 2015 2847

electrical trees grown at 9 kV has the highest value of 1.58.

The fractal dimension of the electrical tree at high voltage
(above 9 kV) is in the range of 1.4-1.49.

4 DISCUSSION
4.1 ELECTRICAL TREE SHAPE
The earlier presented results show that the type of electrical
tree types change apparently with temperature, though the
branch type remains dominant. This is for example a very
different result as compared to the observations made on (a) branch-pine tree at 50 C.
vulcanized silicone rubber, in which dense bush trees
dominated at similar temperature range (60 and 90 C) [22].
The tree shapes appearing at low voltage (9 kV) in our
experiments show apparent differences with those growing at
high voltage levels. It was also found that the PD behavior
differs between the voltage levels. At the same time, the
branch-pine trees, sometimes also called monkey puzzle [23],
grown at 50 C demonstrated similar characteristics as those
reported for experiments done at room temperature [16],
where after a period of low PD activity, the pine branch
structure appeared co-incidental with PD magnitudes
dropping below the threshold level of 10 pC. Despite that the (b) branch tree at 50 C.
leading branches ceased to grow in length, the pine branch
structure still develops. However the time of tree evolution at
elevated temperature is much shorter. The hypothesis
formulated in [16], that the formation of branch-pine tree is
associated with generation of conducting branches near to the
needle electrode, can be checked by examining the tree
morphology under the reflected light, as suggested in [24]. In
such experiments non-conducting branches appear as white
channels whereas conducting ones are dark, the latter being a
consequence of conducting carbon deposit on the branch
walls.
The reflected light observation of the branch-pine tree (c) branch tree at 70 C.

grown at 50 C (Figure 4b) is shown in Figure 22a, where the

main channel of the electrical tree appear darker, suggests that
carbonized degradation byproducts condense on its sidewalls.
The transformation of the channel sidewalls conductivity
occurs most probably during the development of the pine
structure. As suggested in [16], when the tree channel
conductivity is sufficiently high, the field at the needle
electrode can easily be extended to the channel tip located at
the closest distance from the needle electrode, activating this
way the formation of pine structure during the tree growth.
Conversely, in the case of branch trees grown at 9 kV and 50 (d) bush-branch tree at 90 C.
Figure 22. Reflected light observation of the electrical trees of high
C, the PD magnitude increases continuously with the tree temperature in XLPE cable insulation at low voltage (9 kV).
growth. Figure 22b shows that the branch channels appear
white in the reflected light observation, which indicates that Figure 22c also indicates for a possibility of changing its
conducting products do not accumulate within the tree conductivity. The color of the main channel appeared darker
channels and the condition for the formation of pine structure than its leading part, indicating a change in the wall composition.
is missing. However, the ongoing PD activity during the tree formation
As in the case of the branch tree (Figure 12a) grown at low suggests that the conductivity change in the main channel is
voltage at 70 C, the PD activity remained at the level insufficient to suppress the discharges. Therefore, the PD
between 70 pC and 100 pC until the leading channel stopped activity in the tree structure facilitates a growth of newly formed
to extend in length, the reflected light observation shown in side branches, which transformed into a leading channel.
2848 X. Chen et al.: Electrical Treeing Behavior at High Temperature in XLPE Cable Insulation Samples

In contrast, the bush-branch tree grown at low voltage at 90

C exhibits a long period of low PD activity after the tree
initiation (Figure 18). During this time the tree extends slowly
with in parallel ongoing formation of the densely packed
structure near the needle electrode and only when the leading
branch comes close to the counter electrode, the PD
magnitude substantially increases. Figure 22d shows the
reflected light observation of the bush-branch tree from
Figure 17a. It can be seen that the main channel is surrounded
by the densely packed structure, which appears white. Some
sections of the main channel that can be seen near the needle
electrode are however dark, while the color of the leading
branches is still white. This suggests that a change of the wall
composition of the main channel near the needle electrode had
also occurred. However, the long period of low PD activity Figure 23. Electrical tree initiation time with different applied voltages at high
shows that the discharges in the tree structure were temperatures (the bars represent the range of the tree initiation time).
suppressed, being a result of an increase in channel wall
conductivity and a formation of the densely packed region kV/mm [27]. According to the expression proposed by Mason
(bush structure) near the needle electrode. [28], the maximum electric field strength at the needle tip is
The dominated branch type tree at high voltage level and between 690 kV/mm and 1150 kV/mm, which are much
high temperature appears due to the combined effect of both higher than the threshold value provided above. Moreover, a
the factors. Dissado and Fothergill suggest in [1] that after the melting of lamella crystals becomes more significant when
tree initiation, the discharge at the needle vicinity produces increasing the experiment temperature, generating this way an
sufficient cations, which accumulate on the sidewalls at the increased content of amorphous phase and free volume [26].
branch tips. The back-avalanching of electrons deep trapped This, in turn, may enhance injection of hot electrons thus
in the material allow the tree to grow with the formation of promoting the tree initiation at higher temperature.
new branches. The increase of temperature is also likely to As reported in [29], the theoretical expression of the voltage
enhance accumulation of charges at the branch tips as well the dependence of the tree initiation time (tI) can be expressed as,
extent of degradation produced at the channel tip by a given
discharge, since the penetration of the material by a charged B 3/2 C
ln (tI )  + ln ( ) (1)
pulse increases with temperature [1]. V A
One possible explanation of the appearance of bush type
trees at 50 C is that the local field at the tree boundary is where A, B and C are constant,  is the metal work function,
more dependent upon the space charge and its arrangement, and V is the applied voltage. Figure 24 shows the relationship
rather than on the field intensification at the needle tip. When between the tree initiation time and reciprocal voltage at
the sidewall charge is sufficiently strong during the early different temperatures. The slopes of the regression lines are
formation of the tree, it can cause dense branching from the 43.6, 53.4 and 65.7 for the temperature of 50 C, 70 C and 90
channel sides, producing this way the bush type structure [1]. C, respectively. The threshold energy for damage (Gth) is
In addition, a possible explanation of the tree growth given by [1],
perpendicular to the needle-plane direction at 90 C could Gth  A exp (-B 3/2 V01) (2)
eventually result from inhomogeneity of material strength,
which tends to introduce weak paths in the polymer that are where V0 is the threshold voltage for tree initiation. Assuming
followed by tree channels [25]. that the threshold voltage is independent of the temperature, it
can be seen that the higher temperature would have lower
4.2 TREE INITIATION AND GROWTH TIME
Figure 23 shows summarized plot of electrical tree
initiation time as a function of temperature in the studied
XLPE material tested at the different voltage levels. It should
be noticed that the initiation time reduces consistently at
higher temperatures. As reported in [26], the mechanism of
electrical tree initiation under ac stress in time scale over 1 s
involves charge carrier injection and extraction, where some
of the injected electrons may be accelerated in the free
volume to form ‘hot electrons’. When colliding with polymer
molecules bond scission takes place and free radicals are
formed that cause material deterioration as well as buildup of
tree channel. The estimated field strength for generation of the Figure 24. Relationship showing ln (tI) versus reciprocal voltage (1/V) for
high mobility carriers in XLPE has been estimated at ~290 electrical trees at high temperature.
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 22, No. 5; October 2015 2849

threshold energy for damage as the slopes increase with the

increasing of temperature. Moreover, the injected current due
to the field emission of the metal electrode increases at higher
stressing voltage [29]. Therefore, more charges are injected
into the dielectric at higher voltage, together with the
decreased threshold energy for damage at higher temperature;
which may stimulate the tree initiation.
By assuming that the tree initiation time (tI) is inversely
related to the injection current (J), Arrhenius plots of the
ln(1/tI) for the tree initiation process at different voltage levels
are shown in Figure 25. It can be observed that the activation
energy of the trees at 9 kV (0.23 eV) is significantly lower
than the activation energy of the trees initiated at higher
voltage levels (0.49 – 0.86 eV). One should notice here that Figure 26. Electrical tree growth time with different applied voltages at high
different types of trees are formed in these two regimes, bush- temperatures (the bars represent the range of the tree growth time).
branch trees at 90 C and 9 kV and branch trees at higher
voltage levels, indicating that the injection process at higher low voltage displays another interesting feature. The bush-
voltage levels becomes more dependent on temperature. branch tree observed at 90 C has the longest tree growth time,
whereas the tree growth times of the branch-pine tree at 50 C
Furthermore, according to the theory of filamentary
and the branch tree at 70 C become respectively medium and
electromechanical breakdown [30], the filamentary
short. This effect seems to be of great practical importance, as
electromechanical breakdown field (Efem) can be given as
may indicate that the mechanism of electrical tree growth may
follows,
be slowed down at elevated temperatures. A more detail
1/4
 16 G Y  explanation of the involved mechanism still requires further
Efem   2 2  (3) investigations. However, a possible way of arguing could be
 0 r r  by considering the formation of conductive byproducts on the
where the toughness G is a constant, Y is the Young’s tree channel sidewalls.
modulus, 0 is the permittivity of free space , r is the relative On the other hand no much difference between the tree
permittivity of the dielectric and r is the radius of a tubular growth times is seen at the higher voltage levels. Figure 27
crack. The change of the relative permittivity remains shows an example of branch tree (Figure 12c) observed in
insignificant at temperatures increasing up to 90 C [31]. reflected light, in which evidences for formation of the
Assuming that the radius r of the filament is constant, the conductive byproducts at the channel sidewalls are missing.
relationship of Efem  Y1/4 can be obtained. The Young’s This enhances significantly the PD activity associated with the
modulus of XLPE material decreases with increasing of the tree formation and yields its rapid development. Moreover, as
temperature [32]. Therefore, a decrease in the breakdown suggested in [33], the charge deposited by the PD activity at
strength of XLPE cable insulation is expected, which may high temperature can be effectively removed by the increased
also facilitate the tree initiation at higher temperatures. bulk conductivity [34] leading to electric field enhancement
The summarized results of the electrical tree growth time as further away from the tree tips inside to the material. Thus, the
a function of the applied voltage are shown in Figure 26. A negative effect of the space charge on the tree development is
significant decrease of this parameter is observed, similarly to eliminated and the growth is enhanced in the direction of the
the data presented in [20]. However, the tree growth time at counter electrode, yielding faster growing trees with low
fractal dimension at higher voltage levels and high temperature.

Figure 25. Arrhenius plot of the reciprocal of tree initiation time ln (1/tI) for Figure 27. Reflected light observation of a branch tree formed at 13 kV and
trees formed at different voltage levels. temperature of 70 C.
2850 X. Chen et al.: Electrical Treeing Behavior at High Temperature in XLPE Cable Insulation Samples

Furthermore, the pressure of the gases developed by the PD [4] Y. Hayashi, Y. Suzuoki, T. Kato, F. Komori, and H. Mashima,
“Temperature dependence of electrical tree inception from water
activity will also favor the tree development at high
tree degradation”, IEEE Conf. Electr. Insul. Dielectr. Phenomena
temperature [1], since the gas pressure inside the tree channel (CEIDP), pp. 498-501, 2009.
is reduced by the increase of its diffusion and absorption rate [5] S. J. Dodd, “A deterministic model for the growth of non-
in material bulk. conducting electrical tree structures”, J. Phys. D: Appl. Phys., Vol.
36, pp.129-141, 2003.
[6] R. Sarathi, A. Nandini and T. Tanaka, “Understanding Electrical
5 CONCLUSIONS treeing phenomena in XLPE cable insulation under Harmonic AC
voltages adopting UHF technique”, IEEE Trans. Dielectr. Electr.
It is found that temperature and voltage level have Insul., Vol. 19, pp. 903-909, 2012.
pronounced effect on the process of electrical tree formation [7] J. D. Wu, T. Iizuka, K. Monden and T. Tanaka, “Characteristics of
in XLPE cable insulation material. At lower voltages different initial trees of 30 to 60 um length in Epoxy/Silica nanocomposite”,
IEEE Trans. Dielectr. Electr. Insul., Vol. 19, pp. 312-320, 2012.
type of trees can be formed, whereas at higher voltages [8] G. Chen and C. H. Tham, “Electrical treeing characteristics in
branch trees dominate. Increase of test temperature reduces XLPE power cable insulation in frequency range between 20 and
the time for tree initiation, whereas the tree growth time is less 500 Hz”, IEEE Trans. Dielec. Electr. Insul., Vol. 16, pp.179-188,
affected, except for the lowest voltage level at which the 2009.
[9] S. Alapati and M. J. Thomas, “Electrical treeing and the associated
slowest tree development is observed at highest temperature. PD characteristics in LDPE nanocomposites”, IEEE Trans. Dielec.
The impeding effect of elevated temperature on tree growth Electr. Insul., Vol. 19, pp. 697-704, 2012.
at low voltage level needs further investigations, but the [10] E. M. Jarvid, A. B. Johansson, J. H. M. Blennow, M. R. Andersson
indicated change of conductivity of tree channel sidewalls is and S. M. Gubanski, “Evaluation of the performance of several
object types for electrical treeing experiments”, IEEE Trans.
considered here as the main reason for the observed behavior. Dielec. Electr. Insul., Vol. 20, pp.1712-1719, 2013.
It seems that conducting decomposition byproducts are easier [11] A. Ishibashi, T. Kawai, S. Nakagawa, H. Muto, S. Katakai, K.
formed at lower temperatures and voltages, damping this way Hirotsu and T. Nakatsuka, “A study of treeing phenomena in the
PD activity within the tree channels. On the other hand, the development of insulation for 500kV XLPE cables”, IEEE Trans.
Dielec. Electr. Insul., Vol. 5, pp.695-706, 1998.
branch trees formed at high temperature and high voltage [12] A. S. Xie, X. Q. Zheng, S. T. Li and G. Chen, “The characteristics
have predominantly non-conducting branches and intensive of electrical trees in the inner and outer layers of different voltage
PD activities can develop during their growth. The space rating XLPE cable insulation”, J. Phys. D: Appl. Phys., Vol. 42, pp.
charge accumulated around the tree by the PD activity may 1251-1260, 2009.
[13] N. Shimizu, Y. Shibata, K. Ito, K. Imai and M. Nawata, “Electrical
also influence the tree shape leading to lower dimensional tree tree at high temperature in XLPE and effect of oxygen”, IEEE
and its rapid growth at high temperature. As a result, Conf. Electr. Insul. Dielectr. Phenomena (CEIDP), pp. 329-332,
measurements of PD intensity may provide a potential means 2000.
of early on-line detection of pre-breakdown phenomena in [14] Y. Liu, X. L. Cao, Y. Xu, J. Y. Liu, Z. Y. Zhang and Y. J. Fei,
“Insulation state analysis and upgrading feasibility study of
cable insulation. existing AC 10kV XLPE distribution cables in Jiangsu Grid of
The observed reduction of tree initiation time with the China”, Int’l. Council on Large Electric Systems (CIGRE), paper
increase of temperature and voltage level may be attributed to SC B1-112, pp. 1-11, 2010.
lowering of the threshold energy for damage at higher [15] K. Kudo, “Fractal analysis of electrical trees”, IEEE Trans. Dielec.
Electr. Insul., Vol. 5, No.5, pp.713-727, 1998.
temperature as well as increasing of the injection current at [16] X. R. Chen, Y. Xu, X. L. Cao, S. J. Dodd and L. A. Dissado,
higher voltage level. “Effect of tree channel conductivity on electrical tree shape and
breakdown in XLPE cable insulation samples”, IEEE Trans.
Dielectr. Electr. Insul., Vol. 18, pp. 847-860, 2011.
ACKNOWLEDGMENT [17] IEC 60270, Partial Discharge Measurement, 3rd ed., 2000.
[18] J. Feder, Fractals, Plenum Press: New York, 1988.
The authors would like to acknowledge the financial
[19] X. R. Chen, Y. Xu, J. Xu, W. Shi, W. H. Yang, Y. Liu and X. L.
support from the National Science Foundation of China under Cao, “Propagation and partial discharge characteristics of
Projects No.50877057. They thank the helpful discussions electrical trees in 110 kV XLPE cable insulation at power
with Profs. L. A. Dissado and S. J. Dodd of University of frequency applied voltage”, High Voltage Eng., Vol. 36, No.10,
Leicester, UK. Xiangrong Chen thanks Mr. Meng Wang and pp. 2436-2443, 2010 (In Chinese).
[20] R. J. Densley, “An investigation into the growth of electrical trees
Ms. Libin Hu for their help in the experimental work. in XLPE cable insulation”, IEEE Trans. Electr. Insul., Vol.14, pp.
Xiangrong Chen also thanks his family for great support and 148-158, 1979.
encouragement as well as Chalmers Area of Advance in [21] L.A. Dissado, “Understanding electrical trees in solids: from
Energy for providing resources to prepare this manuscript. experiment to theory”, IEEE Trans. Dielec. Electr. Insul., Vol. 9,
No. 4, pp. 483-497,2002.
[22] B. X. Du, Z. L. Ma, Y. Gao and T. Han, “Effect of ambient
temperature on electrical treeing characteristics in silicone rubber”,
REFERENCES IEEE Trans. Dielec. Electr. Insul., Vol. 18, No. 2, pp. 401-407,
[1] L. A. Dissado and J. C. Fothergill, Electrical degradation and 2011.
breakdown in polymers. Peter Peregrinus for IEE: London, UK, 1992. [23] G. Chen and C. H. Tham, “Electrical treeing characteristics in
[2] J. Densley, “Ageing mechanisms and diagnostics for power cables-an XLPE power cable insulation in frequency range between 20 and
overview”. IEEE Electr. Insul. Mag., Vol. 17, No. 1, pp. 14-22, 2001. 500 Hz”, IEEE Trans. Dielec. Electr. Insul., Vol. 16, No. 1, pp.
[3] S. Hvidsten, S. Kvande, A. Ryen and P. B. Larsen, “Severe degradation 179-188, 2009.
of the conductor screen of service and laboratory aged medium voltage [24] A. S. Vaughan, I. L. Hosier, S. J. Dodd and S. J. Sutton, “On the
XLPE insulated cables”, IEEE Trans. Dielectr. Electr. Insul., Vol. 16, structure and chemistry of electrical trees in polyethylene”, J.
pp. 155-161, 2009. Phys. D: Appl. Phys., Vol. 39, No. 5, pp. 962-978, 2006.
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 22, No. 5; October 2015 2851
[25] L. A. Dissado and P. J. J. Sweeney, “Physical model for Yang Xu (M’05) was born in Shaanxi, China in
breakdown structures in solid dielectrics”, Phys. Rev. B, Vol. 48, 1969. He received the B.S., M.S. and Ph.D. degrees
No. 22, pp. 16261-16268, 1993. in electrical engineering from XJTU, Xi’an, China, in
[26] N. Shimizu and C. Laurent, “Electrical tree initiation”, IEEE 1991, 1994 and 2005, respectively. He has been an
Trans. Dielec. Electr. Insul., Vol. 5, No. 5, pp. 651-659, 1998. associate professor of XJTU. His research interests
[27] S. Boggs and J. B. Kuang, “High field effects in solid dielectrics”, lie in on-line monitoring of power equipments. He is
IEEE Electr. Insul. Mag., Vol. 14, No. 6, pp. 5-12, 1998. a member of the National Standardization Technical
[28] J. H. Mason, “Breakdown of solid dielectrics in divergent fields”, Committee of insulating materials, China, and also a
Proc. IEE-Part C: Monographs, Vol.102, No.2, pp. 254-263, 1955. member of the CIGRE SC B1 WG-28.
[29] T. Tanaka and A. Greenwood, “Effects of charge injection and
extraction on tree initiation in polyethylene”, IEEE Trans. Power
App. Syst., Vol. 97, No. 5, pp. 1749-1759, 1978.
[30] J. C. Fothergill, “Filamentary electromechanical breakdown”, Xiaolong Cao (SM’98) was born in Shaanxi, China
IEEE Trans. Electr. Insul., Vol. 26, No. 6, pp. 1124-1129, 1991. in 1945. He received the B.S., M.S. and Ph.D.
[31] Y. Mecheri, L. Boukezzi, A. Boubakeur and M. Lallouani, degrees in electrical engineering from XJTU, Xi’an,
“Dielectric and mechannical behavior of cross-linked China, in 1968, 1983 and 1986, respectively. He
polyethylene under thermal ageing”, IEEE Conf. Electr. Insul. became a Professor of XJTU in 1996. His research
Dielectr. Phenomena (CEIDP), pp. 560-563, 2000. interests include power cable, electrical insulation
[32] X. G. Qi and S. Boggs, “Thermal and mechanical properties of test, nano-dielectrics. He is the Vice Chairman of the
EPR and XLPE cable compounds”, IEEE Electr. Insul. Mag., Vol. National Standardization Technical Committee of
22, No. 3, pp. 19-24, 2006. insulating materials, China, and also a Member of
[33] N. M. Chalashkanov, S. J. Dodd, J. C. Fothergill and L. A. the CIGRE SC B1.
Dissado, “On the relationship between bulk charge transport and
electric breakdown mechanisms in epoxy resins”, 22 nd Nordic
Insulation Sympos., Tampere, Finland, pp. 29-32, 2011.
Stanislaw M. Gubanski (M'89-SM'90-F'01)
[34] J. C. Fothergill, S. J. Dodd, L. A. Dissado, T. Liu and U. H.
received the M.Sc. (high voltage engineering) and
Nilsson, “The measurement of very low conductivity and
Ph.D. degrees (material science) from the Technical
dielectric loss in XLPE cables: a possible method to detect
University of Wroclaw, Poland, in 1973 and 1976,
degradation due to thermal aging”, IEEE Trans. Dielec. Electr.
respectively. He was a Research Fellow at the
Insul., Vol. 18, pp. 1544-1553, 2011.
University College of North Wales Bangor, U.K
from 1976 to 1977, and a senior lecturer at the
Xiangrong Chen (S’08-M’11) was born in Hunan, Technical University of Wroclaw, Wroclaw,
China in 1982. He received the M.S. and Ph.D. Poland, from 1977 to 1988. Afterwards he was
degrees in electrical engineering from Xi’an Jiaotong associate professor (1989-1996) at the Royal
University, China, in 2008 and 2011, respectively. Institute of Technology, Stockholm, Sweden. Currently, he is Professor in
From 2012 he is an Assistant Professor in High High Voltage Engineering at the Department of Materials and Manufacturing
Voltage Engineering at the Department of Technology, Chalmers University of Technology, involved in the work of
Manufacturing and Materials Technology, Chalmers IEEE and CIGRE. He was Chair of the Nominations Committee and Vice
University of Technology, Gothenburg, Sweden. His President (Technical & Administrative) of IEEE Dielectrics and Electrical
research interests include insulating materials for Insulation Society. Now he is a Senior Associate Editor of the IEEE
HVDC applications. Transactions on Dielectrics and Electrical Insulation.