Proceedings of the 9th International Conference on 1-11

Properties and Applications of Dielectric Materials

July 19-23,2009, Harbin, China

Space charge characteristics of multi-layer oil-paper insulation under different DC voltages

Chao TANG 1, 2*, G CHEN2, M FU3 , Ruijin LIA0 1
1 The State Key Laboratory of Power Transmission Equipment & System Security and New Technology
Chongqing University, CHINA
2 School of Electronics and Computer Science
University of Southampton, UK
3 AVERA T&D Technology Centre, Stafford, UK
*E-mail: tangchao_1981@163.com

Abstract: In this paper, space charge in oil-paper
insulation system has been investigated using the
pulsed electroacoustic (PEA) technique. A series of
measurements were carried out when the insulation
system was subjected to different applied voltages.
Charge dynamics in the insulation system during the
volts-on, volts-off and decay have been analyzed. It has
been found that homocharge injection occurred both at
the anode and the cathode. Positive charges are
observed to accumulate in the layers, which indicate
that the oil-paper layer interfaces act as a barrier for
positive charges. The decay tests showed that after
30min, about a quarter of space charges remained in
the sample, about 90% charges disappeared after 2
hours. Finally, total charge variation in these tests were
analyzed.
Keywords: Space Charge, Oil-paper, DC Voltage,
Multi-layer
1 INTRODUCTION
The oil-paper insulation has been used in power
transformers , power cables and HVDC equipments for
quite a long time because of its low cost and desirable
physical and electrical properties. But they degrade
under a combined stress of thermal, electrical,
mechanical, and chemical stresses during routine
operations, which will effect the lifetime of power
equipments significantly. A lot of research works about
the physical and electrical characteristics of oil-paper
insulation have been conducted over the years.
However, only a few studies about space charge in
oil-paper insulation are reported.
Therefore, a better understanding of charge dynamics,
in accordance with the specific variation of the charge
distribution in multi layer insulation system becomes
increasingly important and need a careful investigation.
Thus, in this paper, based on the PEA technique, a
series of measurements were carried out when the
paper insulation system was subjected to different
applied voltages. Charge dynamics in the insulation
system during the volts-on and volts-off have been
analyzed and the influence of temperature on charge
dynamics discussed.
2 EXPERIMENTS
In this study, the measurements were carried out upon
oil immersed insulation paper. The insulation paper
which is made by Croylek Ltd, has a thickness of ~
6011m for single layer. The parameters of the paper
meet the international standard IEC 554.2. The
insulation oil used in this experiment was Nytro lOX
transformer oil, which was provided by Nynas Oil
Company.
Before PEA test, several pretreatment steps were
carried out upon samples. Firstly, these new insulation
paper were cut into round shape with a diameter of
~2cm. Then the paper samples were kept in a vacuum
oven at 393K (120°C) for 20min, in order to make the
moisture content less than 0.2%. After that, the paper
samples were impregnated by fresh degassed insulation
oil in sealed oven. As thin paper was used, the sample
can be fully impregnated after 1 day.
In practical application, oil-paper is usually used as
multilayer insulation. Therefore, in this experiment, the
PEA tests are performs on three layers oil-paper
samples (~12011m after oil immersed and being
pressed by electrodes, in Figure I). The samples were
stressed at three different DC voltage levels (4kV, 6kV,
8kV) at the same temperature of 20°C. Each time, an
electric stressing time of I hour was tested. In addition,
space charge evolution after the removal of the applied
electric field was also measured.
30 60 90 120 150 180 210 2-10 2,-0
Time (min)
Figure 2. Voltage profile for space charge measurements
The space charge measurements were taken at various
times during the periods of both volts-on and volts-off
(short-circuit condition) using the PEA technique. In
the PEA technique, acoustic pressure waves are
generated due to the interaction of pulsed electric field
and charge layer. Detection of acoustic pressure waves
allows one to determine charge distribution across the
sample. The principle of the PEA can be seen in many
literatures [I , 2]. The voltage profile in the present
study is shown in Figure 2.

978-1-4244-4368-0/09/$25.00 ©2009 IEEE 914

 The charge injection deepens into the bulk with the
3 RESULTS AND DISCUSSION increase of the applied voltage. Under 4kY,
homocharges of each polarity enter no more than one
3.1 Charge distribution with volts-on layer. Under 6kY, homocharges of each polarity enter
more than one layer, it is possible that a small quantity
Obviously from the space charge distributions showed neutralized in the middle layer. Under SkY, some
in Figure 4, the cathode peak is sharp and evident; in homocharges reach the layer near to the opposite
contrast, the anode peak is wide and flat, this is electrode.
because of attenuation and scattering of acoustic wave Along with charge injection, electrode peaks offset
caused by the oil-paper sample. Together with volts off towards the inner sample direction, especially the
figures , the positions of electrodes were confirmed and anode peak. There is less negative charge observed in
marked with black lines. the region adjacent to the cathode. There may be two
reasons for this phenomenon. Firstly, compared to
10
positive charge, the amount of negative charge

~
~

.iO'
5

·5
\ 1

f
~'0'
.../
151
- 20 1 2 1
injection didn't increase with the applied voltage,
which indicates that negative charge may travel faster
~
., · 10 and with less blocked, and it reaches the anode quickly
A
., · 15 or is neutralized on the way. Secondly, the PEA
~ l' -2:1
technique only shows the net charge, the injected
..=" · 25 Omin 5 min 10min 3 0 min 60 rnnl-
U
.so charge may be present together with the positive
. " charge which is blocked at the interfaces .
Sam ple thickne ss (11m )
Theoretically, in a PEA figure, the integral area of
(a) 4kV positive peaks is basically the same as that of negative
20
peaks. However, because of the signal scattering in
~

'a
10
_ ::--- oil-paper sample and the contribution of pulse, the test
§ 0 results are some differences from that derived from
'\ 1 r ~1 151 20 1 2<1
.~-. · 10
\ I theory.
~ -20
A
· 30
(iI
,f , 3.2 Charge distribution with volts-off
..a ·40 o min 5 min 10 min 3Jmin 50 min I-
u I
· 50
Normally, Charges can be roughly classified as fast and
· 60
Sa mple thickn ess (11m) slow charges [4, 5]. 'Slow' charges are those which are
(b) 6kV essentially trapped in deep traps while 'fast' charges
20 are those which escape from the traps very shortly after
~ 10

1: ~ the de voltage is removed. Charges trapped in the

Q. · 10
0
\ ,r ~1 151 201 2 1 insulating material are often stable, i.e. slow charge.
.~.
I
-2:1 The volts-on measurement includes contributions from
~ . )3
A both fast and slow charges while the volts-off'
., . OJ

~' · 00 measurement involves only slow charge .

..= · 00 o min :5 min
U
· 70
I
·00
Sample thickness (um)
(c) 8kV
Figure 4. Space charge dynamics at different de electr ic fields
(volts on)
As shown in Fig. 4 (a), (b) and (c), the peak value at
electrodes increases with the applied voltage from 4kY
to SkY. Under each applied voltage, because of
homocharges injection at both electrodes, quantities of
electric charges on both electrodes decreases as the
stressing time increases. The positive charges
accumulate in the vicinity of the anode, while the
negative charges adjac ent to the cathode. Charge
injected quickly after voltage was switched on, an
approximate equilibrium between positive and negative
charge injection was observed after 30min. No big
differ ence noticed after that time. However, it may take
a relatively long period (e.g. up to 10, OOOs), to reach
the stable regim e, i.e. the equilibrium between charge
injection, extraction and transport [3].
10 min 30 min 60 mn After volts-on tests , slow charge distributions for
I three-layer samples under volts-off condition have
been taken as shown in Figure 5. The charge
distributions in the bulk hardly changes, while the
induced charges form on both electrodes which
confirmed the homocharge injection happened there
under volts-on condition. What's more, the induced
charge density increases gradually with the increase in
the applied voltage (Fig. 5 a, b, c), from a maximum of
21 Clm 3 (4kY) to 26 Clm 3 (6kY) and then 33 Clm 3
(SkY), which reflects the increase of charge injection.
Fig. 5 also shows that positive charges are dominant in
the bulk after the removal of the applied voltage. It is
noteworthy that only positive charges appears at the
sample layer interfaces, which means the oil-paper
layer interfaces act as a barrier for positive charges
only.

915
 30
25 . - 1- 30 se c 1 min 2 min Smin 10 min 20 min 30 mn
'a 25

~ 20
8.£. ,5
~
20

10
•I 5min 10 min :::0 m in OOmin l -

.~.
= 10
15
f:
,"
I
.
,
I
A
'"01' / 1
I
I
.~:T'
A'"
g',. 5
il +: .
I

e
E 0
, v I '01 I I, ,,' 2 ,
e~
0
J \ j~~ 1
.5
, T 1 ~/ I
I
101
: .- 'T 21 '
· 10 ·5
Sample thi ckn ess (run) Sa mple thi ckness (IUn)

(a) 4kV (b) 6kY

3)
30 _ 1- :zo seC- 1 mn 2 min Sm in 10 rrin 20 min 30 min

.. 'a 25
I
5 min 10 m il :::0 min 00 min 1-
~
1
.
20

.~.
1
"1 ,1 I
I 15
'I, I

) \
I

I
~/
.4-
1\ .1
,
A

e
i!
'"0"
<;
10

0
H/' I
.J . , :
,
,
~.
.l.
I

-10
s V' I
,
'0 '
Sample thickn ess bun)
I ~

. -5

·10
1 \f 101

Sample rhieknoss (jun]

I
'Y- .---
I
2

(b) 6kV (c) 8kY

Figure 6. Space charge decay after de electric stress ing
~35
",
;.~
3J
.&
5m in 10 min 3J min OO minl -

~25 3.4 total charge

.~. 20
I
~ 15
The total absolute amount of charge accumulated in the
A
...'"
bt'
10
5
)
J ~ . samples can be calculated based on the charge-density
" \ ....I- ../ ' \. I distribution given in Fig. 4-6.
..::
U
0
s V , ' 01 '10.,
Q(T) = rlp(x,t)~dx
-5

-10
J. (2)
Sample thi cknes s (um)
(c) 8kV Where p(x,t) is the charge density, S is the
Figure 5. Space charge dynamics at differ ent de electric fields electrode area, and d is the thickness of the sample.
(volt s oft)
As we know, a trend of movement and combination
exists between positive and negative charge. So the
3.3 Charge distribution with decay total charge is always changing during volts-on and
volts-off tests. In this paper, the total charge at the time
After 1 hour de stressing, space charge distributions immediately after power on/off in the volts-on/volts-o ff
after the removal of the applied voltage are shown in test was calculated.
Figure 6. Compared to volts-on tests, charges are more The relationship between the total amount of charge
easily injected than they are transported away from the accumulated and the time under different applied
electrodes. The charge decay speed becomes much voltages is shown in Fig. 7. Total charge in volts-on
more slowly with the time increase. After 30min, about condition is the total injection charges including both
a quarter of space charges remains in the sample. fast moving charge and slow moving charge. At first,
Further tests indicate that about 90% charges the higher the applied voltage is, the more charges
disappeared after 2 hours decay. injected. Secondly, a tendency of decline, then increase,
-1- 30s ec - 1min 2 min 5 min 10 tri o 2J min 30 min
35
finally to stable can be observed in Figure 7 (a). This
~3) , can be explained as follows. At the very beginning,
'd
-s- 25 .,
o~ 20
J observable homocharge injected from both electrodes ,
~. I and then fast moving charge moved quickly into the
' ;;' 15

~ 10 ~ J. center of samples and neutralized, which made the total

J : 1 absolute amount of charge show a little decrease. After
...g'., 5 J .A~ ~ I
~
,-, 0
,V , 30min electrical stressing, it reached an approximate
10 1 251
-5 ~ r-"'-
TI equilibrium between positive and negative charge.
· 10
Sample thickne ss (um)
(a) 4kV

916
 3,5
__ ·UX _ 6k\" siv presented. Space charge dynamics in volts-on, volts-off
I-
and decay are analyzed. The results are summarized as

1.5
.. . .- . . follows:
( I) Homocharge injection is observed both at the
anode and the cathode, and the injection speed is very
0.5
o
~
fast. The higher the applied voltage, the more the
o 10 co _~ O -to ;0 60 70 amount of injection charge and the deeper the charge
volt s on lim .. (min)
injected into the sample layers.
(a) Volts on
(2) Positive charges are found to accumulate at the
1_ -tk\+_ - 6k\ + siv
layer interfaces, which indicate that the oil-paper layer
interfaces act as a barrier for positive charges. This will
-.- - . . . ~
effect the distribution of electrical field and deteriorate
. the electrical behavior of oil-paper insulation.
(3) The decay tests showed that after 30min, about a
o quarter of space charges remained in the sample, about
o 10 zo .~ O -to ;0 70
volt s off tun e (min )
90% charges disappeared after 2 hours.
(b) Volts off In our opinion, both the electric and thermal stresses
1.8
have significant impact to space charge behavior,
1_ -4k\ · ____6k\ · 8kV I
_ 1.6 further studies of space charge in oil-paper with
~ 1.-t variable applied voltages and temperatures are under
~ l.~ -..::.
;t. 1 <, performing which will be published later.
] 0.8 -~
. ~
.; 0.6 ACKNOWLEDGEMENT
~ OA One of the authors (Chao Tang) wishes to thank the
O. ~

o 'Chinese Scholarship Council Funding for Joint

o 10 15 ~O 30 35 Training PhD Student' for the financial support
Dec ay nm e (min )
provided.
(c) Decay
Figure 7. Total charge under different applied voltages vs time. REFERENCES

Roughly, traps in electrical materials can be divided
into two types, deep traps and shallow traps. When the
charges moved into the deep traps, more energy is
required for them to get out, which slow down the
velocity of charge movement and finally formed slow
moving charge. After power off, the volts-off tests
were taken at once. The fast moving charges moved
out immediately, only slow moving charges remain in
the sample. In Figure 7 (b), the amount of slow moving
charges in sample is stable except a little fluctuant at
8kV. In addition, under different applied voltages, total
charge at volts-off stabled in different horizontal. That
may be explained as that, when increasing the applied
voltage, the amount of injected charges in the material
increases as well. As a result, more and more charges
are trapped, the distortion of electric field becomes
more serious, and the electrical performance gets
worse.
It can be seen in Figure 7 (c), the higher the applied
voltage is, the faster the charge decayed. As we know,
at the early time when the power is off, more charges
exist in higher voltage tested samples, which generated
higher induction electrical field and made the charges
moving faster and neutralized quickly.
[I] G. Chen, M. Fu, X. Z. Liu, L. S. Zhong . Ac aging and
space-charge characteristics in low-density polyethylene polymeric
insulation, Journal of Applied Physics, 2005, Vol. 97, No.3, pp. 1-7.
[2] G. Chen, Y. Tanaka, T. Takada, L. Zhong. Effect of polyethylene
interface on space charge formation . IEEE Transactions on
Dielectrics and Electrical Insulation , 2005, Vol. II , No. I, pp.
113-12I .
[3]ltC. Ciobanu , W. PfeitTer, C. Schreiner. Charge packet evolution
in paper-oil insulation and derived technological considerations.
Proceedings of the 7th International Confer ence on Properties and
Applications of Dielectric Materials , 2003, Vol. 2, pp. 695-698 .
[4] G. Chen, Y. L. Chong, M. Fu. Calibration of the pulsed
electroacoustic technique in the presence of trapped charge.
Measurement Science and Technology, 2006, Vol. 17, No.7, pp.
1974-1980 .
[51 Mingli Fu, George Chen, Len A. Dissado, John C. Fothergill.
Influence of thermal treatment and residues on space charge
accumulation in XLPE for DC power cable application, IEEE
Transactions on Dielectrics and Electrical Insulation, 2007, Vol. 14,
No. I, pp. 53-64.

4 CONCLUSIONS

In this paper, PEA measurements on multi-layer

oil-impregnated insulation paper samples were

917