IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 29, NO.

2, MARCH 2019 7700904

Partial Discharge Characteristics During Treeing

Process in Silicone Rubber at 20 and −100 °C
T. Han , Member, IEEE, J. G. Su, T. T. Ma, F. Y. Wang, Y. Q. Xing , Member, IEEE, and Y. Gao , Member, IEEE

Abstract—In the operation of high temperature superconduct-

ing cable, the temperature in current lead or connector with tradi-
tional cable varies from −196 °C to above 0 °C. As a main reason of
the insulation aging, electrical tree characteristics change with the
temperature. In this paper, partial discharge (PD) signals of elec-
trical tree are detected by a high-frequency current transformer.
The experimental temperature was set to be 20 °C and −100 °C. To
analyze the PD signal, the phase resolved partial discharge (PRPD)
is calculated. Besides, the multiscale permutation entropy (MSPE)
is also employed to characterize the nonlinear dynamic feature.
The result indicates that the low temperature suppresses the PD,
which will influence the PRPD. At −100 °C, PD starts occuring at
the phase from 90 °C to 135°, and then it begins to occur at 270°
after 10 min. However, PD occurs at 90 °C and 270° when the tem-
perature is 20 °C. The MSPE can characterize the different treeing
stage obviously. At both 20 °C and −100 °C, the stable MSPE value
in the first 20 min is close to each other and becomes scattered in Fig. 1. Schematic of experimental setup.
the next 40 min.
Index Terms—Electrical tree, current lead, high temperature tree and (multi-scale permutation entropy) MSPE has been im-
superconducting (HTS) cable, phase resolved partial discharge proved in other nonlinear system [12], [13].
(PRPD), multiscale permutation entropy (MSPE).
To research the electrical tree PD signals at different temper-
ature, the PD signals are detected by a HFCT and recorded by
I. INTRODUCTION a scopecorder. SiR samples were employed to initiate tree in-
side. The tree structure is recorded by a microscope. The result
IGH temperature superconducting (HTS) power cable has
H been successfully developed and operated [1], [2]. In the
cable terminal connected to the HTS power cable, the insula-
indicates that the PRPD and MSPE are related to the low tem-
perature and can characterize the PD signals of treeing process.
tion inside the traditional cables and joints may face a steep
temperature gradient [3]–[8]. It has been reported that electrical II. EXPERIMENTAL SETUP AND PROCEDURE
tree is the main reason which caused the insulation failure in A. Sample Preparation
cables [3]. And the partial discharge (PD) has been proved to be
an effective measurement to characterize electrical tree inside SiR sample was prepared as the following process. A mold
insulation [9]–[11]. As a nonlinear signal, the time-domain PD made of glass was employed, then the liquid SiR was injected
distribution is not enough to distinguish the tree structure or into the mold. After this, a needle with the curvature radius of
different treeing stage. However, the phase resolved partial dis- 3 µm was fixed in the mold. The distance from the needle tip
charge (PRPD) is sensitive enough to characterize the electrical to the edge of mold is set to be 2 mm. To degas the sample,
it was put into the vacuum tank for 120 min. At last, a 10-day
Manuscript received July 31, 2018; accepted December 6, 2018. Date of vulcanization was carried out at 20 °C. The sample was shown
publication January 1, 2019; date of current version January 28, 2019. This in Fig. 1. The dimension of the sample was 30 mm × 15 mm
work was supported by the Chinese National Natural Foundation under Grants with the thickness of 3 mm.
51707132 and 51677128, National Key Research and Development Program
2018YFB0904400. (Corresponding author: Yu Gao.)
T. Han, J. G. Su, T. T. Ma, F. Y. Wang, and Y. Gao are with the Key Laboratory B. Experimental Setup
of Smart Grid of Education Ministry, School of Electrical and Information
Engineering, Tianjin University, Tianjin 300072, China (e-mail:, hmgaoyu@ The AC voltage of 50 Hz was applied onto the needle elec-
tju.edu.cn). trode. The Root Mean Square (RMS) was set to be 10 kV. PD
Y. Q. Xing is with the State Key Laboratory of Reliability and Intelligence
of Electrical Equipment, and the Key Laboratory of Electromagnetic Field free treatment was carried out on the experiment circuits. During
and Electrical Apparatus Reliability, Hebei University of Technology, Tianjin the experiment, the temperature in Dewar tank was controlled
300130, China. by the controller shown in Fig. 1. Two groups of samples were
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. employed. One was tested at 20 °C, another was at −100 °C.
Digital Object Identifier 10.1109/TASC.2018.2890516 During the experiment, PD at −196 °C was also tested, while
1051-8223 © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
7700904 IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 29, NO. 2, MARCH 2019

Fig. 3. Time-domain PD signals at −100 °C.

Fig. 2. Time-domain PD signals at 20 °C.

no obvious PD was observed. A high-frequency current trans-

former (HFCT) was installed on the power cable to detect the
PD signal. Its bandwidth is from 1–10 MHz. A scopecorder
was employed to record the PD signals detected by HFCT. The
treeing pictures were recorded by a microscope and CCD. The
longest distance of tree from the needle tip is defined as tree
length.

III. RESULTS AND DISCUSSION

A. Time-Domain PD Signals at 20 and −100 °C
The time-domain PD signals were recorded. It was found that
at different growing process of the same tree, PD signal varies
obviously. The results are shown in Fig. 2 and 3. To find the
difference of PD in the whole treeing process, PD signals in
2 s at 2, 6, 10, 20, 30, 40, 50 and 60 min are employed to be
discussed here. The amplitude is normalized before analyzed.

B. PRPD at 20 and −100 °C

In the treeing process, PD is the main factor that affects the
tree growth. To analyze the PD signal distribution, the PRPD at
20 °C is calculated and shown in Fig. 4. Fig. 4. PRPD at different treeing time with temperature of 20 °C.
HAN et al.: PARTIAL DISCHARGE CHARACTERISTICS DURING TREEING PROCESS IN SILICONE RUBBER AT 20 AND −100 °C 7700904

Fig. 6. Tree structure of 20 and −100 °C at 60 min.

Fig. 5. PRPD at different treeing time with temperature of −100 °C.

With the treeing time, PD phase changes. At 2, 6 and 10 min,

PD occur at both the positive and negative peak of AC. After
20 min, there is a blank area in the partial signals between 90°
and 135°. In the tree channel, the PD at the needle tip is caused
by the AC voltage. While away from the tip, it can also be
triggered by the space charge accumulation on the surface of
the channels. It is indicated that with the growth of tree, much
more PD caused by space charge occurs. If the electric field
caused by space charge is high enough, the PD can still occur
even the AC voltage is decreasing. This discharge makes the
secondary PD phase at about 135° after 20 min.
When the temperature decreases to −100 °C, the PRPD is
similar in all the treeing time. Most of the PD occurs at the
positive cycle at this temperature. This is maybe related to the Fig. 7. MSPE at different treeing time with temperature of 20 °C.
different mobility of positive and negative charge. In the positive
half cycle, the electrons injected into the SiR in the negative occur if the electric field between the tip and space charge is
half cycle will be extracted and discharge occurs in the process. high enough.
However, the mobility of positive ion at −100 °C is maybe
much lower. Meanwhile, the mass of these ions is much higher
than electrons. These will lead to the smaller amount of injected C. MSPE at 20 and −100 °C
positive ions. As a result, there is few discharge observed in the Entropy is a measurement of complexity and regularity. It
negative half cycle in the first 6 min. has been employed to characterize the dynamic behavior of the
While at treeing time from 10 to 60 min, some PD appears nonlinear signals. According to [12], the MSPE can be employed
at the negative half cycle of AC, as shown in Fig. 5. The reason to analyze the nonlinear signals. Here it is employed to analyze
to cause this change may be also the space charge. As shown the PD signals shown in Fig. 2 and 3. Fig. 7 shows the MSPE
in Fig. 6, the tree size is small at −100 °C. The electric field at 20 °C. As shown in Fig. 7(a), the entropy of 2 and 6 min
is dominated by the AC voltage at the needle tip. With the increases first and then decreases to a stable value with the scale
growth of electrical tree, the distance from needle tip increases. increasing from 1 to 30. At 10 and 20 min, it increases with
The influence of space charge on electric field turns to be more the scale. It means that at 2 and 6 min, there are extreme points
obvious. The positive charges will be captured away from the which means the most complex dynamics characteristics at the
tip electrode. In the negative half cycle, the positive charges scale of 2 and 3. Besides, the stable MSPE in the first 10 min is
injected into the channel previously will be extracted. PD will much higher than that at 20 min. It is consistent with the growth
7700904 IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 29, NO. 2, MARCH 2019

at low temperature and the space charge distribution at different

temperature.

IV. CONCLUSION
Tree growth characteristics was analyzed by detecting the PD
signal. PRPD and MSPE are employed in this paper. It shows
that temperature and space charge distribution influence the PD
signals obviously. The following are the main conclusions.
(1) The PRPD at 20 and −100 °C changes with the treeing
time.
(2) Most of the PD at −100 °C occurs at the phase from 90
to 135°. It begins to occur at 270° after 10 min.
(3) MSPE can characterize the different treeing stage obvi-
ously. At both 20 and −100 °C, the stable MSPE value
in the first 20 min is close to each other and becomes
scattered in the next 40 min.

REFERENCES
[1] Y. B. Lin, L. Z. Lin, Z. Y. Gao, and H.M. Wen, “Development of HTS
transmission power cable,” IEEE Trans. Appl. Supercond., vol. 11, no. 1,
pp. 2371–2374, Jan. 2007.
[2] D. S. Kwag et al., “The electrical insulation characteristics for a HTS cable
termination,” IEEE Trans. Appl. Supercond., vol. 16, no. 2, pp. 1618–1621,
Jun. 2006.
[3] S. M. Baek, H. J. Kim, J. W. Cho, and H. S. Ryoo, “Cryogenic
electrical insulation characteristics of solid insulator for the HVDC
HTS cable,” IEEE Trans. Appl. Supercond., vol. 28, no. 4, Jun. 2018,
Art. no. 7700804.
[4] H. J. Kim, J. W. Cho, W. J. Kim, Y. S. Choi, and S. H. Kim, “The
Fig. 8. MSPE at different treeing time with temperature of −100 °C. basic dielectric charac-teristics of insulating materials for HTS DC cable
system,” IEEE Trans. Appl. Supercond., vol. 26, no. 3, Apr. 2016, Art.
no. 7701104.
process of electrical tree. In the first 10 min, the tree grows at [5] W. J. Kim, H. J. Kim, J. W. Cho, Y. S. Choi, and S. H. Kim, “Electrical
and mechanical characteristics of insulating materials for HTS DC cable
the highest speed, as a result, the PD is much more complex and cable joint,” IEEE Trans. Appl. Supercond., vol. 25, no. 3, Jun. 2015,
than that at 20 min. Art. no. 7701004.
There is a stagnation process in the last 20 min of the tree- [6] N. Hayakawa, K. Ishida, T. Matsuoka, and H. Kojima, “Electrical insu-
lation characteristics and mechanisms of low dielectric loss materials for
ing time. There is also a change of the MSPE in Fig. 7(b). In high-voltage HTS cables,” IEEE Trans. Appl. Supercond., vol. 27, no. 4,
the stagnation process, the stable MSPE is about 3.0 while it Jun. 2017, Art. no. 4800904.
is above 3.3 at 30 and 40 min. The entropy of scale 1 is cor- [7] A. Morandi, G. Mazzanti, and M. Marzinotto, “Feasibility of high voltage
DC superconducting cables with extruded warm dielectric,” in Proc. IEEE
responding to the disorder feature of time-domain signals. Its Conf. Elect. Insul. Dielectric Phenomena , 2014, pp. 796–799.
result is consistent with the signals in Fig. 2. [8] L. Ren, Y. J. Tang, J. Shi, and F. S. Jiao, “Design of a termination for
The MSPE at −100 °C is shown in Fig. 8. In the first 20 min, the HTS power cable,” IEEE Trans. Appl. Supercond., vol. 22, no. 3, Jun.
2012, Art. no. 5800504.
the stable value at each time is between 3.8 to 4.0. It is the result [9] X. R. Chen, Y. Xu, and X. L. Cao, “Nonlinear time series analysis of
of tiny tree at this low temperature. It means that in this tiny tree partial discharges in electrical trees of XLPE cable insulation samples,”
channel, the PD will be suppressed. Besides, the crystallization IEEE Trans. Dielec. Electri. Insul., vol. 22, no. 4, pp. 1455–1461, Aug.
2014.
temperature of SiR is around −60 °C [14]. It means in −100 °C, [10] M. D. Noskov, A. S. Malinovski, M. Sack, and A.J. Schwab, “Modelling
the crystallization in SiR will also suppress the tree growth. of partial discharge development in electrical tree channels,” IEEE Trans.
Within these tiny channels, space charge has a minor effect on Dielec. Electri. Insul., vol. 10, no. 3, pp. 425–434, Jun. 2003.
[11] M. C. Liu, Y. P. Liu, Y. D. Li, P. Zheng, and H. R. Rui, “Growth and
the electric field distribution near the needle tip. As a result, the partial discharge characteristics of electrical tree in XLPE under AC-DC
stable MSPE is near to each other in the first 20 min. composite voltage,” IEEE Trans. Dielec. Electri. Insul., vol. 24, no. 4,
However, this PD inside will still contribute to the tree growth. pp. 2282–2290, Aug. 2017.
[12] X. Chen, N. D. Jin, A. Zhao, Z. K. Gao, L. S. Zhai, and B. Sun, “The
In the next 40 min, the stable MSPE varies with the treeing time. experimental signals analysis for bubbly oil-in-water flow using multi-
This is caused by the more complex tree structure in this stage. scale weighted-permutation entropy,” Physica A-Statistical Mechanics Its
With the discharge caused by the space charges, the MSPE will Appl., vol. 417, no. 1, pp. 230–244, Jan. 2015.
[13] Z. Lei, J. Song, M. Tian, X. Cui, C. Li, and M. Wen, “Partial discharges
be higher, because this discharge is maybe not near the phases of of cavities in ethylene propylene rubber insulation,” IEEE Trans. Dielec.
90° and 270°. This is consistent with the result shown in Fig. 5. Electri. Insul., vol. 21, no. 4, pp. 1647–1659, Aug. 2017.
It can be found that, the dynamic characteristics of PD signals [14] T. Han, B. X. Du, Y. Yu, and X. Q. Zhang, “Effect of cryogenic tempera-
ture on tree characteristics in silicone rubber/sio2 nanocomposites under
in the treeing process at low temperature is different from that repetitive pulse voltage,” IEEE Trans. Appl. Supercond., vol. 26, no. 7,
at 20 °C. The difference is caused by the much smaller tree size Oct. 2016, Art. no. 0501404.