nanomaterials

Article
Research on Creeping Flashover Characteristics of
Nanofluid-Impregnated Pressboard Modified Based
on Fe3O4 Nanoparticles under Lightning
Impulse Voltages
Bingliang Shan 1 , Meng Huang 1, *, Yupeng Ying 1 , Mingkang Niu 1 , Qian Sun 2 , Yuzhen Lv 2 ,
Chengrong Li 1 , Bo Qi 1 and Zhaoliang Xing 3
1 State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources,
North China Electric Power University, Beijing 102206, China; shanbingliang521@163.com (B.S.);
18811360731@163.com (Y.Y.); niumingkang@163.com (M.N.); lcr@ncepu.edu.cn (C.L.); lqicb@163.com (B.Q.)
2 School of Energy, Power and Mechanical Engineering, North China Electric Power University,
Beijing 102206, China; 15652912576@163.com (Q.S.); yuzhenlv@163.com (Y.L.)
3 State Key Laboratory of Advanced Power Transmission Technology, Global Energy Interconnection Research
Institute Co. Ltd., Beijing 102209, China; 15811444029@163.com
* Correspondence: hm2016@ncepu.edu.cn




Received: 21 February 2019; Accepted: 23 March 2019; Published: 3 April 2019


Abstract: Creeping flashover of mineral-oil-impregnated pressboard under impulse stress is a

common insulating failure in oil-immersed transformers, arousing increasing attention. Recent
studies have shown that the breakdown strength of transformer oil under positive lightning impulse
voltage can be significantly improved through nanoparticles-based modification, and Fe3 O4 has
shown the best improvement in breakdown strength compared to other nanoparticles that have been
used. This paper presents the creeping flashover characteristics of pure oil-impregnated pressboard
(OIP) and nanofluid-impregnated pressboard (NIP) based on Fe3 O4 nanoparticles under positive
and negative lightning impulse voltages, respectively. It was found that NIP possessed higher
resistance to creeping flashover than OIP. The relative permittivities of oil and oil-impregnated
pressboard before and after nanoparticles-based modification were measured, and the results revealed
that the addition of nanoparticles led to a better match in relative permittivity between oil and
oil-impregnated pressboard, and a more uniform electric field distribution. Furthermore, the shallow
trap density in NIP was obviously increased compared to that of OIP through the thermally stimulated
depolarization current (TSDC), which promoted the dissipation of surface charges and weakened the
distortion of the electric field. Therefore, the creeping flashover characteristics of oil-impregnated
pressboard were greatly improved with Fe3 O4 nanoparticles.

Keywords: Fe3 O4 nanoparticles; creeping flashover; nanofluid-impregnated pressboard; shallow traps

1. Introduction
Oil/pressboard composite insulation systems have been widely applied in large power
transformers due to their excellent dielectric properties and relatively high mechanical strength [1].
The interface between oil and pressboard is usually regarded as the insulating weak point, where
surface charges easily accumulate under external stress, leading to distortion of the local electric field
and insulation failure [2,3]. Significant research efforts have been devoted to the creeping flashover
characteristics of oil-impregnated pressboard (OIP), and theoretical research on the mechanism process
has seen rapid growth in the past decades [4,5]. The previous literature has demonstrated that the

Nanomaterials 2019, 9, 524; doi:10.3390/nano9040524 www.mdpi.com/journal/nanomaterials

Nanomaterials 2019, 9, 524 2 of 12

generation and development of creeping flashover is closely related to local electric stress, dielectric
strength, and moisture content of the insulating materials [4–7]. Thus, effective measures must be taken
to weaken the distortion of local electric fields and improve the dielectric performance of materials so
as to restrain the occurrence of creeping flashover in oil-impregnated pressboard and assure the safe
operation of transformers.
Recent studies have shown that the dielectric performance of mineral oil can be improved through
the modification of nanoparticles, which has gained enormous attention [8–18]. It has been illustrated
that the AC breakdown strength of pure oil can be improved by the addition of Fe3 O4 [11], TiO2 [12,13],
ZnO [13], and SiO2 [14] nanoparticles. The analysis of Lee et al.’s study shows that the AC breakdown
strength of nanofluid modified with Fe3 O4 nanoparticles is two times higher than that of pure oil and
that the modification effect is related to the external field, even though the dielectric loss of Fe3 O4
nanofluid has been found to be larger than that of mineral oil [11]. Kopčanský et al. confirmed that
the DC breakdown voltage of transformer oil could be increased by using Fe3 O4 nanoparticles [15].
However, when Kudelcik et al. subjected Fe3 O4 nanofluid to an external magnetic field of 20 mT,
the improvement of DC breakdown strength was reduced [16]. Segal et al. found that the positive
impulse breakdown voltage of nanofluid using Fe3 O4 nanoparticles was increased by up to 50% when
compared to that of pure oil [17]. The lightning impulse breakdown strengths of pure oil and various
nanofluids have also been studied by Zhou et al. and their results showed that the insulating property
of pure oil could be improved with the addition of nanoparticles [9]. They concluded that Fe3 O4
nanoparticles, with unique electrical and physical properties, showed great potential in improving the
dielectric performance of transformer oil compared to the potential of other nanoparticles. Meanwhile,
the dielectric strength of vegetable-oil-based nanofluid could also be obviously enhanced with Fe3 O4
nanoparticles [18].
Since the pioneering work of Segal et al., it was reported for the first time that Fe3 O4 nanoparticles
could move freely in and out of nanofluid-impregnated pressboard (NIP) without binding to
cellulose [19]. This phenomenon has provided a new idea for improving the dielectric properties
of OIP by immersing the pressboard in the transformer-oil-based nanofluid modified using Fe3 O4
nanoparticles, since these nanoparticles would not be blocked out of the pressboard. To date, no work
on the creeping flashover characteristics of NIP based on Fe3 O4 nanoparticles under lightning impulse
voltages has been reported, which deserves further study.
The present work aims to investigate the effects of Fe3 O4 nanoparticles on the creeping flashover
behaviors of oil-impregnated pressboard under lightning impulse stress. Dielectric permittivity testing
and electrostatic probe techniques were used to measure both the relative permittivities and space
charge behaviors of oil/pressboard before and after modification by Fe3 O4 nanoparticles. Combined
with the results of thermally stimulated depolarization current (TSDC), the mechanism of creeping
flashover characteristics modification is proposed.

2. Experiment

2.1. Samples Preparation

The transformer oil (KI 25X) was strictly filtered so as to meet the requirement of pure oil proposed
by the Conseil International des Grands Réseaux Électriques (CIGRE) working group 12.17 [20].
Fe3 O4 nanoparticles were prepared by the chemical co-precipitation method using iron powder and
FeCl3 ·6H2 O as sources of iron, and modified by oleic acid to prevent the aggregation of nanoparticles.
The nanoparticles obtained possessed an average diameter of 20 nm and there was no agglomeration,
as shown in Figure 1. The nanofluid was prepared by evenly dispersing Fe3 O4 nanoparticles into
filtered oil by the ultrasonic method with a concentration of 0.2 g/L. The power of the ultrasonic
dispersion device was set to 50 W/L and the ultrasonic dispersion operation lasted no less than half
an hour. Pressboard samples with a size of 85 mm × 60 mm × 2 mm were dried in a ventilated oven
at 105 ◦ C for 48 h. Then, pressboard samples and oil samples were treated under vacuum below 1 kPa
Nanomaterials 2019, 9, 524 3 of 12
Nanomaterials 2019,
Nanomaterials 2019, 9,
9, xx FOR
FOR PEER
PEER REVIEW
REVIEW 33 of
of 12
12

at 80 ◦ C for
vacuum
vacuum 48 h11to
below
below getat
kPa
kPa rid
at 80of°C
80 °Cmoisture
for 48 hhand
for 48 air,rid
to get
to get and
rid ofthe
of moisture
moisture
moisture and
andcontent ofthe
air, and
air, and oil samples
the moisture
moisture gained
content
content ofwas
of the
the
10
oil ppm. After
oil samples that,
samples gained the
gained was treated
was 10
10 ppm. pressboards
ppm. After
After that,were
that, the immersed
the treated into the
treated pressboards dry
pressboards wereoil and nanofluid
were immersed
immersed into respectively,
into the
the dry
dry oil
oil
then
and placed
and nanofluid under
nanofluid vacuum below
respectively,
respectively, then 1placed
then kPa forunder
placed 48 h. vacuum
under Finally,
vacuumpressboard
below 11 kPa
below samples
kPa for 48
for 48with a moisture
h. Finally,
h. Finally, content
pressboard
pressboard
of 0.5% were
samples
samples withobtained.
with aa moisture
moisture content
content ofof 0.5%
0.5% were
were obtained.
obtained.

Figure1.
Figure 1.TEM
1. TEMof
TEM ofFe
of Fe33O
Fe O44 nanoparticles.
nanoparticles.
Figure 3 O4 nanoparticles.

2.2. Impulse Creeping

2.2. Impulse
2.2. Creeping Flashover
Flashover Test
Creeping Flashover Test
Test
The creepingflashover
The creeping
creeping flashover
flashover test test
test was was performed
was performed
performed in an
in in an insulating
an insulating
insulating container120
container measuring
container measuring measuring
120 mm ×× 80
mm 80
120
mm mm× 50× 80
mm, mmas × 50 mm,
schematically as schematically
illustrated inillustrated
Figure 2. in
A Figure 2.
needle-plane A
mm × 50 mm, as schematically illustrated in Figure 2. A needle-plane electrode system, where theneedle-plane
electrode electrode
system, system,
where the
where the steel needle (tip ± 5 µm) ◦ to horizontal and connected to
steel needle
steel needle (tip radius
(tip radius of of 35radius
35 μm)ofwas
±± 55 μm) 35 placed
was placed was
at 30°
at toplaced
30° to at 30and
horizontal
horizontal and connected to
connected to the
the high-voltage
high-voltage
the high-voltage
generator
generator opposing
opposinggenerator
aa copper
copper opposing
plate a copperconnected
plate electrode
electrode plate electrode
connected the connected
to the
to ground, was
ground, to the
was ground,
located
located in the
in was
the located
container.
container.
in the container.
The oil-impregnated
The The oil-impregnated
oil-impregnated pressboard
pressboard was pressboard
was placed was
placed horizontally
horizontally inplaced horizontally
in tight
tight contact
contact with in tight
with two contact
two electrodes with
electrodes andandtwo
the
the
electrodes
distance and theneedle
distance between
between distance
needle tip between
tip and the
and needle
the plane
plane tip and
could
could the plane
be adjusted
be adjusted could
from
from be
20 to
20 adjusted
to 40
40 mm. In
mm. Infrom 20 topure
addition,
addition, 40 mm.
pure oil
oil
In
was
wasaddition,
selectedpure
selected as oildielectric
as the
the was selected
dielectric liquidas the
liquid for dielectric
for the OIP
the OIP testliquid
test andfor
and the OIP as
nanofluid
nanofluid test and
as the
the nanofluid
dielectric
dielectric as the
liquid
liquid fordielectric
for the NIP
the NIP
liquid
test. for the NIP test.
test.

Figure 2.
Figure 2. Experimental
Experimental setup
setup for
for creeping
creeping flashover
flashover tests.
tests.

During
During the
the test,
test, an
an impulse
impulse voltage
voltage generator
generatorwas
generator wasused
was usedto
used totoprovide
provide1.2/50
provide 1.2/50 µs
1.2/50 μs standard lightning
μs standard lightning
impulse waveforms.
impulse waveforms. For For the validity of the test, the step-by-step
For the validity of the test, the step-by-step method method was used in
in the testing
the testing
process.
process. At
At the beginningof
the beginning
beginning oftests,
of tests,aaaselected
tests, selectedproper
selected propervalue
proper valueof
value ofofimpulse
impulse
impulse voltage
voltage
voltage was
was
was applied
applied
applied 3 times.
33 times.
times. If
If
If
thethe creeping
creeping flashover
flashover did
did not
not happen,
happen, the
the voltage
voltage was
was increased
increased by
by 5
5 kV
kV
the creeping flashover did not happen, the voltage was increased by 5 kV and applied to theand applied to the samples
samples
another
another three
three times.
times. The
The previous steps were
previous steps were repeated
repeated until
until creeping
creeping flashover
flashover occurred.
occurred. TheThe time
time
interval
interval between experiments
between experiments
experiments was was at least 22 min,
at least and six
min, and six effective
effective creeping
creeping flashover
flashover trials
trials were
were
successfully
successfully conducted.
conducted.

2.3. Relative
2.3. Relative Permittivity
Permittivity Test
Test
Nanomaterials 2019, 9, 524 4 of 12
Nanomaterials 2019, 9, x FOR PEER REVIEW 4 of 12
Nanomaterials 2019, 9, x FOR PEER REVIEW 4 of 12
2.3. Relative Permittivity
The relative Test
permittivities of oil samples and pressboard samples were tested using a Jiantong
The
The relative permittivities of
dielectric relative
impedance permittivities
instrument of oil samples
samples and
oil(Baoding, pressboard
China)
and according
pressboard samplesto IEC
samples were tested
tested using
were 60250 aa Jiantong
standards
using and
Jiantong
dielectric
Novocontrol impedance
dielectric impedance 40instrument
Conceptinstrument broadband (Baoding,
dielectric
(Baoding, China) China)
spectrometer
accordingaccording to IEC
to(Montabaur,
IEC 60250 60250 and
Germany),
standards standards
respectively.
Novocontrol and
Novocontrol
Each sample Concept
was tested 40 broadband
five times and dielectric
then the spectrometer
average value (Montabaur,
was
Concept 40 broadband dielectric spectrometer (Montabaur, Germany), respectively. Each sample wasrecorded. Germany), respectively.
Each
testedsample wasand
five times testedthen five
thetimes
averageandvaluethen the wasaverage
recorded.value was recorded.
2.4. Measurement of Trap Distribution Characteristics
2.4. Measurement
Measurement of Trap Distribution CharacteristicsCharacteristics
Trap distribution characteristics of NIP and OIP were measured through TSDC, with the
Trap distribution
schematic distribution
diagram characteristics
is characteristics
shown in Figure of
of NIP3. NIP
First,and
and OIP OIP
thewere were measured
measured
test sample wasthrough
heatedthrough
TSDC,
to 313 TSDC,
with with
the schematic
K before the
being
schematic
polarized diagram
diagram isunder
shown a in is shown
Figure
negative involtage
3. First,
DC Figure
the testof3.4sample
First,
kV for the
30test
was sample
heated
min. was
to 313
Secondly, heated
Kthe
before to 313
being
temperature Kofbefore
polarized beinga
under
the sample
polarized
negative
was quickly under
DC a negative
voltage
lowered of
to 4263kVK DC
for
and voltage
30 min.the
then of 4polarization
kV forthe
Secondly, 30temperature
min. Secondly,
voltage of cut
was the
the temperature
sample
off. Afterwas of the
quickly
that, the two sample
lowered
sides
was
to the
of quickly
263 K andlowered
sample then
werethe topolarization
shorted263 K forand then
2 min inthe
voltage polarization
was to
order cutreleasevoltage
off. After was
the that,
free thecuttwo
charges.off.sides
Afterofthat,
Finally, the the
sampletwo sides
sample were
was
of the sample
shorted
heated atfor
a rate were
2 min
of 2 in shorted
K/minorder forrelease
to
until it2reached
min the in 353
order to release
freeK charges.
when the free
theFinally, thecharges.
depolarization sample Finally,
currentwaswas the sample
heated at a rate
simultaneously was
of
heated at a
2 K/min until
measured rate of 2 K/min
it reached6514
by a Keithley until it reached
353 electrometer 353
K when the depolarization K when the depolarization
(Tektronix Inc.,current was OR,
Beaverton, current
simultaneously
USA). was simultaneously
measured by a
measured
Keithley 6514 by aelectrometer
Keithley 6514 electrometer
(Tektronix Inc., (Tektronix
Beaverton, Inc.,
OR, Beaverton,
USA). OR, USA).

Figure 3. Schematic of thermally stimulated depolarization current (TSDC) measurement.

Figure 3.
Figure Schematic of
3. Schematic of thermally
thermally stimulated
stimulated depolarization current (TSDC) measurement.
2.5. Measurement of Surface Charge Density
2.5. Measurement of Surface Charge Density
2.5. Measurement of Surface Charge Density
The surface charge densities of OIP and NIP were measured by an electrostatic probe and the
The surface charge densities of OIP and NIP were measured by an electrostatic probe and the
Theissurface
diagram charge densities
schematically illustratedofinOIP and4,NIP
Figure were
where themeasured by an
gap distance electrostatic
between needleprobe
tip andand the
plate
diagram is schematically illustrated in Figure 4, where the gap distance between needle tip and plate
diagram is
electrode wasschematically
set to 30 mm. illustrated
The test in Figurewas
system 4, where
based the gaptheory
on the distance
of between
capacitive needle
probe,tipwhich
and plate
has
electrode was set to 30 mm. The test system was based on the theory of capacitive probe, which has
electrode
been widelywas set to
used in 30 mm. The
research on test system wascharge
the deposited basedat onthe
thesurface
theory of
of insulators.
capacitive probe, which has
The polarization
been widely used in research on the deposited charge at the surface of insulators. The polarization
been widely
voltage of –10used
kV wasin research
applied on the needle
to the deposited chargefor
electrode at 30
themin
surface of insulators.
to generate The polarization
space charges, and the
voltage of –10 kV was applied to the needle electrode for 30 min to generate space charges, and the
voltage of –10
electrostatic kV was
probe was applied
used to to the needle
measure electrode
the surface for 30accumulation
charge min to generate space charges,
characteristics and the
of samples
electrostatic probe was used to measure the surface charge accumulation characteristics of samples at
electrostatic
at probe was
5 min intervals. Powerused to then
was measureshutthe
offsurface
and thecharge
decayaccumulation
features of thecharacteristics of samples
surface charges were
5 min intervals. Power was then shut off and the decay features of the surface charges were measured
at 5 min intervals.
measured as well. Power was then shut off and the decay features of the surface charges were
as well.
measured as well.
U0
Shielding system Ce U0
Shielding system Ce
Electrostatic probe
High voltage Electrostatic probe
High voltage Cx
Cx

Oil-impregnated Cd
pressboard
Oil-impregnated Cd
pressboard
Figure
Figure4.4.Schematic
Schematicof
ofTSDC
TSDCmeasurement.
measurement.
Figure 4. Schematic of TSDC measurement.
Nanomaterials 2019, 9, 524 5 of 12
Nanomaterials 2019, 9, x FOR PEER REVIEW 5 of 12
Nanomaterials 2019, 9, x FOR PEER REVIEW 5 of 12

3.3.Results
Resultsand
andDiscussion
Discussion
3. Results and Discussion
3.1.Lightning
3.1. LightningImpulse
ImpulseCreeping
CreepingFlashover
FlashoverCharacteristics
Characteristics
3.1. Lightning Impulse Creeping Flashover Characteristics
Thecreeping
The creepingflashover
flashovervoltages
voltagesofofOIP OIPand
andNIPNIPunder
underpositive
positivelightning
lightningimpulse
impulsevoltages
voltagesareare
The creeping flashover voltages of OIP and NIP under positive lightning impulse voltages are
recordedininFigure
recorded Figure5.5.ItItappeared
appearedthat thatthe
theaverage
averagecreeping
creepingflashover
flashovervoltages
voltagesofofthese
thesetwo
twosamples
samples
recorded in Figure 5. It appeared that the average creeping flashover voltages of these two samples
increasedwith
increased withthe
theexpansion
expansionofofthe theelectrode
electrodegapgapdistance.
distance.When
Whencompared,
compared,the theaverage
averageflashover
flashover
increased with the expansion of the electrode gap distance. When compared, the average flashover
voltageofofNIP
voltage NIPwas
wasapparently
apparentlyhigher
higherthanthanthat
thatofofthe
theOIP
OIPunder
undervarious
variouselectrode
electrodegap gapdistances.
distances.
voltage of NIP was apparently higher than that of the OIP under various electrode gap distances.
Moreover,the
Moreover, theimprovement
improvement of of creeping flashover voltage ofofOIP could bebe
increased to 1.19
Moreover, the improvement of average creepingflashover
average creeping flashover voltage
voltage of OIP OIP could
could increased
be increased to 1.19to
times
1.19times when
timeswhen modified
when modifiedwith Fe
with O nanoparticles and the electrode gap distance was set to 30 mm.
Fe3Fe
O43 O 4 nanoparticles
and and the electrode gap distance was
to set to 30 mm.
3 4
modified with nanoparticles the electrode gap distance was set 30 mm.

Figure 5. Creeping
Figure Creeping flashover voltage
flashover of theofoil-impregnated
voltage pressboard
the oil-impregnated (OIP) and (OIP)
pressboard the nanofluid-
and the
Figure5.5. Creeping flashover voltage of the oil-impregnated pressboard (OIP) and the nanofluid-
impregnated pressboard
nanofluid-impregnated (NIP)
pressboardunder positive
(NIP) under lightning
positive impulse
lightningvoltages.
impulse voltages.
impregnated pressboard (NIP) under positive lightning impulse voltages.

TheThe Weibull
Weibull distributionisiscommonly
distribution commonlyused used for
for analyzing
analyzing the
the breakdown
breakdowndata dataofofelectrical
electrical
The Weibull distribution is commonly used for analyzing the breakdown data of electrical
insulation
insulation [21].[21].
The The
WeibullWeibull distribution
distribution fittingfitting
curvescurves of the creeping
of the creeping flashoverflashover voltages
voltages of of
pressboard
insulation
pressboard [21]. The underWeibull distribution fitting curves ofwerethe creeping andflashover voltages
in of
samples undersamples
positive lightningpositive lightning
impulse impulse
voltages were voltages
calculated andcalculated
are presentedare presented
in Figure 6. Here,
pressboard samples under positive lightning impulse voltages were calculated and are presented
with in
it isFigure 6. Here,
visually it is visually
presented presented
that the breakdownthat the breakdown
probability ofprobability
OIP and NIPof OIP and NIPwith
increased increased
the applied
Figure 6. Here,
the applied it is visually presented that the breakdown probability of OIP and NIP increased with
voltages undervoltages
various under various
electrode gap electrode
distances.gap distances. Meanwhile,
Meanwhile, NIP possessedNIP apossessed a lower
lower breakdown
the applied voltages
breakdown probability under varioustoelectrode
compared OIP when gap
thedistances.
same Meanwhile,applied
NIP possessed a lower
probability compared to OIP when the same voltages werevoltages
applied were
under various under various
electrode gap
breakdown
electrode gap distances, which is beneficial in improving the dielectric performance of materials.various
probability compared to OIP when the same voltages were applied under
distances, which is beneficial in improving the dielectric performance of materials.
electrode gap distances, which is beneficial in improving the dielectric performance of materials.

Figure 6. The
Figure Weibull
6. The Weibulldistribution
distributionfitting
fittingcurves
curvesof
of the
the creeping
creeping flashover voltagesunder
flashover voltages underpositive
positive
lightning impulse
lightning voltages.
impulse voltages.
Figure 6. The Weibull distribution fitting curves of the creeping flashover voltages under positive
The average
lightning valuevoltages.
impulse (AVG) and SD of chop time for samples are shown in Table 1. It is of great
significance to note that the average chop time of oil-impregnated pressboard could be apparently
 Nanomaterials 2019, 9, x FOR PEER REVIEW 6 of 12

The2019,
Nanomaterials average
9, 524 value (AVG) and SD of chop time for samples are shown in Table 1. It is of great
6 of 12
significance to note that the average chop time of oil-impregnated pressboard could be apparently
prolonged by the addition of nanoparticles. For example, the chop time of oil-impregnated
prolonged by the
pressboard addition
under of nanoparticles.
positive lightning impulseFor example, the chop
voltages was time of
prolonged byoil-impregnated
72.8% when the pressboard
electrode
under positive lightning impulse voltages was prolonged by 72.8% when
gap distance was 20 mm, and the improvement in chop time was still relatively the electrode
largegap distance
when the
waselectrode
20 mm, gap
and distance
the improvement in chop
was increased to 30time was
or 40 mm.stillThese
relatively
resultslarge when
are in the electrode
accordance gap
with our
distance waswork
previous increased
on thetostreamer
30 or 40 characteristics
mm. These resultsof Fe3O are in accordance
4 nanofluid with
[18,22], our previous
indicating that Fework
3O4

on the streamer can

nanoparticles characteristics of Fe3 O
effectively restrain 4 nanofluid
the [18,22], indicating
process of creeping flashover atthat
the Fe 3 O4 nanoparticles
oil/pressboard can
interface,
effectively restrain the
thus extending the development
process of creeping
time andflashover at the
reducing the oil/pressboard interface,
average propagation thus
speed of extending
the positivethe
streamer. time and reducing the average propagation speed of the positive streamer.
development

Table
Table 1. Chop
1. Chop time
time ofof OIP
OIP andNIP
and NIPunder
underpositive
positivelightning
lightning impulse
impulse voltages.
voltages.
OIP NIP NIP IncreaseRate
Rate
Electrode Gap Distance (mm) OIP
ElectrodeGap Increase
Distance (mm) AVG (μs) SD (μs) AVG (μs) SD (μs) (%)
AVG (µs) SD (µs) AVG (µs) SD (µs) (%)
20 12.5 1.5 21.6 0.7 72.8
20 30 12.5 1.5
17.3 1.0 21.6 28.8 0.7 0.7 72.8
66.5
30 17.3 1.0 28.8 0.7 66.5
40 22.1 1.4 36.0 1.5 62.9
40 22.1 1.4 36.0 1.5 62.9

Since the flashover always occurred along the surface of the insulating container instead of along
Since
the the flashover
pressboard samplesalways
underoccurred
negativealong the surface
lightning impulseofvoltages,
the insulating
owing container instead of
to the insufficient
along the pressboard samples under negative lightning impulse voltages, owing to the insufficient
insulation distance between the container and the electrodes when the electrode gap distance was 40
insulation
mm, thedistance
negativebetween
flashoverthe container
measures wereand the electrodes
carried out underwhen the electrode
the electrode gap distance
gap distance of 20 mmwas
and 30
40 mm, themm. The flashover
negative creeping flashover
measuresvoltages and the
were carried out corresponding Weibull
under the electrode gapdistribution
distance offitting
20 mm
andcurves
30 mm.ofThe
OIPcreeping
and NIPflashover
under positive lightning
voltages and the impulse voltages
corresponding are shown
Weibull in Figures
distribution 7 and
fitting 8,
curves
respectively.
of OIP and NIP under positive lightning impulse voltages are shown in Figures 7 and 8, respectively.

Nanomaterials 2019, 9, x FOR PEER REVIEW 7 of 12

Figure 7. Creeping flashover voltages of OIP and NIP under negative lightning impulse voltages.
Figure 7. Creeping flashover voltages of OIP and NIP under negative lightning impulse voltages.

Figure The
8. 8.
Figure TheWeibull
Weibulldistribution
distributionfitting
fitting curves
curves of
of the creeping flashover
the creeping flashovervoltages
voltagesunder
under negative
negative
lightning impulse voltages.
lightning impulse voltages.

The average creeping flashover voltage increased with the expansion of the electrode gap
distance under applied voltages, which is similar to the results under positive lightning impulse
voltages. The creeping flashover performance of OIP was still apparently improved by modification
with Fe3O4 nanoparticles. Nevertheless, it is worth noting that the creeping flashover voltage of OIP
Nanomaterials 2019, 9, 524 7 of 12

The average creeping flashover voltage increased with the expansion of the electrode gap distance
under applied voltages, which is similar to the results under positive lightning impulse voltages.
The creeping flashover performance of OIP was still apparently improved by modification with Fe3 O4
nanoparticles. Nevertheless, it is worth noting that the creeping flashover voltage of OIP under
relatively low breakdown probabilities (e.g., 5%) was almost the same or a little higher than that of
NIP owing to the wider dispersion of the breakdown voltages of NIP.
The average chop time of OIP and NIP under negative lightning impulse voltage is demonstrated
in Table 2. It can be seen that the differences in chop time under the same electrode gap distance was
relatively small, distinct from the results under positive lightning impulse voltage. These differences
may arise from the different development processes of positive streamers and negative streamers [10].

Table 2. Chop time of OIP and NIP under negative lightning impulse voltages.

Electrode Gap OIP NIP Increase Rate

Distance (mm) AVG (µs) SD (µs) AVG (µs) SD (µs) (%)
20 25.0 2.8 27.1 4.6 8.4
30 30.1 4.6 30.3 5.6 0.6

From the experimental results above, it can be reasonably concluded that NIP possessed higher
resistance to creeping flashover than OIP under lightning impulse voltages. Simultaneously, the
average chop time under positive lightning impulse voltages could also be prolonged, which is of
great significance for restraining the propagation of the positive streamer along the surface and for
enhancing the dielectric performance of oil/pressboard composite insulation systems in transformers.

3.2. Effect of Change in Relative Permittivities

The creeping discharge at the interface between oil and pressboard is tightly related to the electric
field distribution, wherein the stronger the local tangential electric field at the interface, the more
easily the creeping flashover occurs [23]. Since the lightning impulse voltage is mainly composed of
high-frequency alternating components [24], the process of analyzing electric field distribution along
the interface between oil and pressboard is similar to that under AC stress.
The structure of the oil/pressboard interface is shown in Figure 9. Owing to the production
process of pressboard, there were many grooves on the surface of the pressboard filled with insulating
Nanomaterials
oil [25]. The2019, 9, x FOR electric
tangential PEER REVIEW 8 of as
field distribution was calculated and analyzed from current density 12
follows [26]:
dD dE
𝐽 =J =
𝜎𝐸σE++ dt==𝜎𝐸 σE++𝜀ε 0𝜀ε r
dt = 𝐸(𝜎 + 𝜀 𝜀 𝜔),
= E ( σ + ε 0 ε r ω ), (1)
(1)

where J 𝐽is is
where thethe
tangential current
tangential density,
current σ is conductivity,
density, E is the tangential
𝜎 is conductivity, electric fieldelectric
𝐸 is the tangential strength, D
field
is electric displacement vector, = 8.85 × 10 −12 F/m is vacuum permittivity, ε is relative permittivity,
strength, 𝐷 is electric displacement
ε 0 vector, 𝜀 = 8.85 × 10 F/m is vacuum r permittivity, 𝜀 is
and is the angular frequency.
relative permittivity, and 𝜔 is the angular frequency.
ω

Figure 9.
Figure 9. The
The structure
structure schematic
schematic of
of the
the oil/pressboard
oil/pressboard interface.
interface.

The typical conductivity of insulating pressboard is about 10−15 S/m and the conductivity of pure
oil is approximately 10−12 S/m, while the conductivity of nanofluid based on Fe3O4 nanoparticles is
almost 10–500 times that of pure oil [26]. From Equation (1) and the parameters mentioned above, it
can be deduced that 𝜎 ≪ 𝜀 𝜀 𝜔 under an AC electric field. In other words, the tangential electric
Nanomaterials 2019, 9, 524 8 of 12

The typical conductivity of insulating pressboard is about 10−15 S/m and the conductivity of pure
oil is approximately 10−12 S/m, while the conductivity of nanofluid based on Fe3 O4 nanoparticles is
almost 10–500 times that of pure oil [26]. From Equation (1) and the parameters mentioned above,
it can be deduced that σ  ε 0 ε r ω under an AC electric field. In other words, the tangential electric
field distribution is mainly inversely proportional to the relative permittivity in this condition.
The relative permittivities of oil and oil-impregnated pressboard were measured, as shown
in Table 3. Since the relative permittivity of pure oil is almost half that of OIP, the electric field
strength in oil would be higher under applied lightning impulse voltages. Meanwhile, it has been
reported the breakdown strength of OIP is usually greater than that of pure oil [23,25] and as a result,
creeping flashover usually starts from pure oil. With the addition of Fe3 O4 nanoparticles, the relative
permittivity of pure oil-based nanofluid was enhanced from 2.2 to 3.0, while that of NIP increased by
just 0.2. The improvement in relative permittivity was probably due to the additional polarization
generated by Fe3 O4 nanoparticles [27].

Table 3. The relative permittivities of oil samples and pressboard samples.

Relative Permittivity Permittivity Ratio of Oil to

Sample
Oil (εoil ) Oil-Impregnated Pressboard (εboard ) Oil-Impregnated Pressboard

OIP 2.2 4.2 0.52

NIP 3.0 4.4 0.68

For a deeper analysis of the modification effect, based on the change in relative permittivities,
on creeping discharge in the nanofluid/pressboard insulation system, the permittivity ratio of oil
to oil-impregnated pressboard (ε oil /ε board ) was also defined and used to measure the electric field
distribution along the oil/pressboard interface region, as shown in Table 3. A better match in relative
permittivity between pressboard and nanofluid was observed since the permittivity ratio was increased
from 0.52 to 0.68, which helped to reduce the electric field strength in oil and enhanced that in NIP
instead, leading to a more uniform electric field distribution at the interface and preventing the
occurrence of creeping flashover. Moreover, previous literature has proven that the dielectric strength
of Fe3 O4 nanofluid is much higher than that of mineral oil [15,17,28], which also contributed to
improving the dielectric performance of nanofluid/pressboard composite insulation systems. Thus,
the creeping flashover voltage of oil-impregnated pressboard can apparently be improved due to the
higher dielectric strength of Fe3 O4 nanofluid and the more reasonable electric field distribution.

3.3. Effect of Shallow Trap Characteristics

It can be deduced that the creeping flashover characteristics of oil-impregnated pressboard can
also be affected by the transport characteristics of charge carriers based on the creeping discharge
mechanisms of a dielectric solid [29]. Therefore, the surface charge accumulation and dissipation
characteristics of OIP and NIP were tested, as shown in Figure 10.
From Figure 10a, it can be seen that the surface charge density of OIP reached a saturated state
after applying a voltage for about 600 s while that of NIP increased for 300 s before reaching a steady
value. The maximum surface charge density of OIP (about 8.54 pC/mm2 ) was considerably higher
than that of NIP (about 0.34 pC/mm2 ) under the same applied voltage, indicating that more space
charge carriers had accumulated on the surface of the OIP. While there was a continuous decline of
charge carriers on the surface of OIP after the external electric field was removed, the surface charge
density was still high (about 2.71 pC/mm2 ) after 30 min. On the contrary, NIP possessed a rapid
surface charge decay rate, which was beneficial to effectively prevent surface charge accumulation.
than that
value. Theofmaximum
NIP (aboutsurface
0.34 pC/mm
charge )density
2 under the same applied voltage, indicating that more space
of OIP (about 8.54 pC/mm2) was considerably higher
charge carriers had accumulated on the
than that of NIP (about 0.34 pC/mm ) under the
2 surface of same
the OIP. Whilevoltage,
applied there was a continuous
indicating decline
that more of
space
charge
charge carriers
carriers on hadthe surface of OIP
accumulated on after the external
the surface of the electric fieldthere
OIP. While was removed, the surface
was a continuous charge
decline of
density was still high (about 2.71 pC/mm 2) after 30 min. On the contrary, NIP possessed a rapid
charge carriers on the surface of OIP after the external electric field was removed, the surface charge
surface
densitycharge
Nanomaterials
was 2019, decay
highrate,
9, 524
still which
(about was
2.71 beneficial
pC/mm to effectively
2) after 30 min. On prevent surface NIP
the contrary, charge accumulation.
possessed a 9rapid
of 12
surface charge decay rate, which was beneficial to effectively prevent surface charge accumulation.

(a) (b)
Figure 10. (a) Charge(a)
accumulation characteristics and (b) charge dissipation(b)characteristics of OIP
and NIP samples.
Figure10.
Figure 10.(a)
(a)Charge
Chargeaccumulation
accumulationcharacteristics
characteristicsand
and(b)
(b)charge
chargedissipation
dissipationcharacteristics
characteristicsof
ofOIP
OIP
and NIP samples.
and NIP samples.
It has been reported that the charge transporting process is closely related to the trap
characteristics
ItIthas
hasbeen
beenof materials
reported that
reported [30,31].
the The charge
thatcharge
the thermally
transportingstimulated
process iscurrent
transporting closelymethod
process related is the
to
is closely anrelated
effective totechnique
trap characteristics
the trap
to
of study trap[30,31].
materials
characteristics parameters as [30,31].
well as stimulated
The thermally
of materials storage andcurrent
The thermally transport characteristics
method
stimulated of space
is an effective
current method charges
technique
is an toinstudy
effective materials,
trap
technique
and has been
parameters
to study trap widely
well as used
as parameters asinwell
storage recent
andas years [9,23,32,33].
transport
storage In order
characteristics
and transport to clarify
of space charges
characteristics thespace
of inrelationship
materials,
chargesandbetween the
has been
in materials,
transport
widely
and hasused process
been of surface
inwidely
recent years
used chargesyears
andIn
in[9,23,32,33].
recent trap characteristics
order to clarify
[9,23,32,33]. theof
In order pressboard
torelationship
clarify samples, the
between
the relationship TSDC of oil-
transport
between the
impregnated
process
transport pressboard
of surface
process charges before and
andcharges
of surface after modification
trap characteristics were measured,
of pressboard
and trap characteristics ofsamples, and
pressboard the
TSDC results are shown
of oil-impregnated
samples, TSDC of oil- in
Figure
pressboard11.
impregnated before and after
pressboard modification
before and afterwere measured,
modification andmeasured,
were the results and
are shown in Figure
the results 11. in
are shown
Figure 11.

Figure 11. The

Figure 11. The test
test curves
curves of
of TSDC
TSDC for
for OIP
OIP and
and NIP.
NIP.

Figure
From Figure 11, it can be 11. seen
clearly The test
thatcurves of TSDC
the peak valuefor
ofOIP
the and
NIPNIP.
TSDC curve was 84 pA, while
that of OIP was just around 52 pA. To make a quantitative analysis of the trap characteristics of NIP
and OIP, trap parameters such as the quantity of trapped charges and trap energy level were also
calculated according to the TSDC curves, and are summarized in Table 4.

Table 4. The trap parameters of NIP and OIP.

Samples Quantity of Trapped Charges (nC) Trap Energy Level (eV)

OIP 101.04 0.49
NIP 159.64 0.48

In essence, traps are the localized states which can convert free charges into bound charges, and
they can be divided into deep traps and shallow traps according to their energy level [32]. It is clear
from Table 4 that both NIP and OIP possessed shallow traps, since the trap energy levels in NIP
and OIP were 0.48 and 0.49 eV, respectively. Moreover, the quantity of trapped charges in NIP was
Nanomaterials 2019, 9, 524 10 of 12

1.58 times greater than that in OIP, which suggested that the density of shallow traps in NIP was
significantly increased by the modification with Fe3 O4 nanoparticles, thus promoting the dissipation of
space charges. For shallow traps, the potential barriers overcome by space charges during the jumping
process are small, which can inhibit the accumulation of space charges in materials, thus leading to the
decrease of space charge density [33].
From the above results, one conclusion that can be reasonably drawn is that, compared to that
of OIP, the rapid surface charge decay rate in NIP is mainly attributable to its increased shallow
traps density resulting from the addition of Fe3 O4 nanoparticles, which is consistent with our
previous studies [7,22]. In particular, the creeping flashover usually starts from the weakest link
of oil-impregnated pressboard. The accumulation of space charges usually occurs at the interface of
oil/pressboard due to the mismatches in the dielectric parameters of oil and pressboard when under
an external electric field. For these reasons, the distortion of the electric field along the interface would
be aggravated by these charges and the local electric field would increase dramatically, promoting the
creeping flashover more easily. With the addition of Fe3 O4 nanoparticles, plenty of shallow traps were
produced in the NIP, which can promote the dissipation of surface charges, leading to the suppression
of local electric field distortion caused by charge accumulation. Hence, the maximum intensity of the
local electric field was effectively reduced, which contributed to improving the creeping flashover
voltage of NIP under lightning impulse voltages.

4. Conclusions
This paper investigated the effects of Fe3 O4 nanoparticles on the creeping flashover strength of
oil-impregnated pressboard under lightning impulse stress, revealing the modification mechanism
based on Fe3 O4 nanoparticles in details. It can be convincingly concluded that, under applied voltages,
NIP possesses higher resistance to creeping discharge than OIP at various electrode gap distances.
The modification mechanism of the creeping flashover based on Fe3 O4 nanoparticles is clarified
as follows: with the addition of nanoparticles, a better match in relative permittivity between
nanofluid and NIP would be obtained and the breakdown strength of the oil sample would be
improved, thus promoting a more uniform and more substantial electric field distribution at the
interface and improving the dielectric performance of nanofluid/pressboard composite insulation
systems. Meanwhile, the density of shallow traps in NIP was also significantly increased due to the
addition of Fe3 O4 nanoparticles, which can promote the migration of surface charges and restrain the
accumulation process, leading to the weakening of the distortion of the local electric field caused by
charge accumulation and enhancement of the electric field distribution uniformity along the interface.
Hence, the creeping flashover performance of oil-impregnated pressboard was apparently improved.

Author Contributions: B.S. and M.H. conceived and designed the experiments, and wrote the paper. Y.Y. and
Q.S. prepared the materials. B.S., M.N., and Z.X. performed the experiments and analyzed the data. Y.L., C.L.,
and B.Q. gave some good advice during the experiments and discussion.
Funding: This research was funded by the National Natural Science Foundation of China (51337003, 51472084)
and State Key Laboratory of Advanced Power Transmission Technology (GEIRI-SKL-2017-009).
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Liao, R.; Cheng, L.; Yang, L.; Zhang, Y.; Wu, W.; Chao, T. The insulation properties of oil-impregnated
insulation paper reinforced with nano-TiO2 . J. Nanomater. 2013, 1, 373959. [CrossRef]
2. Qi, B.; Zhao, X.; Li, C.; Wu, H. Transient electric field characteristics in oil-pressboard composite insulation
under voltage polarity reversal. IEEE Trans. Dielectr. Electr. Insul. 2015, 22, 2148–2155. [CrossRef]
3. Huang, B.; Hao, M.; Hao, J.; Fu, J.; Wang, Q.; Chen, G. Space charge characteristics in oil and oil-impregnated
pressboard and electric field distortion after polarity reversal. IEEE Trans. Dielectr. Electr. Insul. 2015, 23,
881–891. [CrossRef]
Nanomaterials 2019, 9, 524 11 of 12

4. Dai, J.; Wang, Z.; Jarman, P. Creepage discharge on insulation barriers in aged power transformers. IEEE Trans.
Dielectr. Electr. Insul. 2010, 17, 1327–1335. [CrossRef]
5. Qi, B.; Wei, Z.; Li, C. Creepage discharge of oil-pressboard insulation in ac-dc composite field: Phenomenon
and characteristics. IEEE Trans. Dielectr. Electr. Insul. 2016, 23, 237–245. [CrossRef]
6. Ueno, H.; Watabe, K.; Tada, K.; Nakayama, M. Negative creeping discharge characteristics of a gas/solid
composite insulation system under pulse voltages. Appl. Phys. 1998, 37, 6595–6600. [CrossRef]
7. Lv, Y.; You, Z.; Li, C.; Yang, G.; Bo, Q. Creeping discharge characteristics of nanofluid-impregnated
pressboards under ac stress. IEEE Trans. Plasma Sci. 2016, 44, 2589–2593. [CrossRef]
8. Primo, V.A.; Garcia, B.; Albarracin, R. Improvement of transformer liquid insulation using nanodielectric
fluids: A review. IEEE Electr. Insul. Mag. 2018, 34, 13–26. [CrossRef]
9. Zhou, Y.; Sui, S.; Li, J.; Ouyang, Z.; Lv, Y.; Li, C. The effects of shallow traps on the positive streamer
electrodynamics in transformer oil based nanofluids. J. Phys. D Appl. Phys. 2018, 51, 105304. [CrossRef]
10. Sima, W.; Cao, X.; Yang, Q.; Song, H.; Shi, J. Preparation of three transformer oil-based nanofluids and
comparison of their impulse breakdown characteristics. Nanosci. Nanotechnol. Lett. 2014, 6, 250–256.
[CrossRef]
11. Lee, J.C.; Seo, H.S.; Kim, Y.J. The increased dielectric breakdown voltage of transformer oil-based nanofluids
by an external magnetic field. Int. J. Therm. Sci. 2012, 62, 29–33. [CrossRef]
12. Du, Y.F.; Lv, Y.Z.; Li, C.R.; Zhong, Y.X.; Chen, T.M.; Zhang, S.N.; Zhou, Y.; Chen, Z.Q. Effect of water
adsorption at nanoparticle-oil interface on charge transport in high humidity transformer-oil based nanofluid.
Colloid Surf. A Physicochem. Eng. Asp. 2012, 415, 153–158. [CrossRef]
13. Hanai, M.; Hosomi, S.; Kojima, H.; Hayakawa, N.; Okubo, H. Dependence of TiO2 and ZnO nanoparticle
concentration on electrical insulation characteristics of insulating oil. Electrical Insulation & Dielectric
Phenomena. In Proceedings of the 2013 Annual Report Conference on Electrical Insulation and Dielectric
Phenomena, Shenzhen, China, 20–23 October 2013; pp. 781–783.
14. Jin, H.; Andritsch, T.; Morshuis, P.H.F.; Smit, J.J. AC breakdown voltage and viscosity of mineral oil based
SiO2 nanofluids. In Proceedings of the 2012 Annual Report Conference on Electrical Insulation and Dielectric
Phenomena, Montreal, QC, Canada, 14–17 October 2012; pp. 902–905.
15. Kopčanský, P.; Tomčo, L.; Marton, K.; Koneracká, M.; Potočová, I.; Timko, M. The experimental study of the
dc dielectric breakdown strength in magnetic fluids. J. Magn. Magn. Mater. 2004, 272, 2377–2378. [CrossRef]
16. Kudelcik, J.; Bury, P.; Kopcansky, P.; Timko, M. Dielectric breakdown in mineral oil ITO 100 based magnetic
fluid. Phys. Procedia 2010, 9, 78–81. [CrossRef]
17. Segal, V.; Hjortsberg, A.; Rabinovich, A.; Nattrass, D.; Raj, K. AC (60 Hz) and impulse breakdown strength of
a colloidal fluid based on transformer oil and magnetite nanoparticles. IEEE Int. Sympos. Electr. Insul. 1998,
2, 619–622.
18. Lv, Y.; Wang, J.; Yi, K.; Wang, W.; Li, C. Effect of oleic acid surface modification on dispersion stability and
breakdown strength of vegetable oil-based Fe3 O4 nanofluids. Integr. Ferroelectr. 2015, 163, 8. [CrossRef]
19. Segal, V.; Rabinovich, A.; Nattrass, D.; Raj, K.; Nunes, A. Experimental study of magnetic colloidal fluids
behavior in power transformers. J. Magn. Magn. Mater. 2000, 513–515. [CrossRef]
20. CIGRE. Effect of Particles on Transformer Dielectric Strength. Working Group 17 of Study Committee 12; CIGRE:
Paris, France, 2000.
21. Fabiani, D.; Simoni, L. Discussion on application of the weibull distribution to electrical breakdown of
insulating materials. IEEE Trans. Dielectr. Electr. Insul. 2015, 12, 11–16. [CrossRef]
22. Lv, Y.; Wang, Q.; Zhou, Y.; Li, C.; Ge, Y.; Qi, B. Effect of Fe3 O4 nanoparticles on positive streamer propagation
in transformer oil. AIP Adv. 2016, 6, 035110. [CrossRef]
23. Li, J.; Si, W.; Yao, X.; Li, Y. Partial discharge characteristics over differently aged oil/pressboard interfaces.
IEEE Trans. Dielectr. Electr. Insul. 2009, 16, 1640–1647. [CrossRef]
24. Sun, P.; Sima, W.; Zhang, D.; Jiang, X.; Zhang, H.; Yin, Z. Failure characteristics and mechanism of
nano-modified oil-impregnated paper subjected to repeated impulse voltage. Nanomaterials 2018, 8, 504.
[CrossRef]
25. Li, Y.; Zhang, Q.; Wang, T.; Li, J.; Guo, C.; Ni, H. Degradation characteristics of oil-immersed pressboard
samples induced by partial discharges under dc voltage. IEEE Trans. Dielectr. Electr. Insul. 2017, 24,
1110–1117. [CrossRef]
Nanomaterials 2019, 9, 524 12 of 12

26. Cho, S.; Lee, Y.; Kim, Y. DC field distribution in HVDC transformer considering the effects of space charge
and temperature due to presence of oil immersed pressboard. IEEE Trans. Dielectr. Electr. Insul. 2014, 21,
866–872. [CrossRef]
27. Du, B.; Li, X.; Li, J. Thermal conductivity and dielectric characteristics of transformer oil filled with BN and
Fe3 O4 nanoparticles. IEEE Trans. Dielectr. Electr. Insul. 2015, 22, 2530–2536. [CrossRef]
28. Lv, Y.; Rafiq, M.; Li, C. Study of Dielectric Breakdown Performance of Transformer Oil Based Magnetic
Nanofluids. Energies 2017, 10, 1025.
29. Huang, M.; Wang, L.; Ge, Y.; Lv, Y.; Qi, B.; Li, C. Creeping flashover characteristics improvement of
nanofluid/pressboard system with TiO2 nanoparticles. AIP Adv. 2018, 8, 035205. [CrossRef]
30. Lv, Y.; Ge, Y.; Sun, Z.; Sun, Q.; Huang, M.; Li, C. Effect of nanoparticle morphology on pre-breakdown and
breakdown properties of insulating oil-based nanofluids. Nanomaterials 2018, 8, 476. [CrossRef]
31. Lv, Y.; Zhou, Y.; Li, C.; Ma, K.; Wang, Q.; Wang, W.; Zhang, S.; Jin, Z. Nanoparticle effects on creeping
flashover characteristics of oil/pressboard interface. IEEE Trans. Dielectr. Electr. Insul. 2013, 21, 556–562.
[CrossRef]
32. Tian, F.; Bu, W.; Shi, L.; Yang, C.; Yi, W.; Lei, Q. Theory of modified thermally stimulated current and direct
determination of trap level distribution. J. Electrost. 2011, 69, 7–10. [CrossRef]
33. Lei, Q.; Tian, F.; Yang, C.; He, L.; Yi, W. Modified isothermal discharge current theory and its application in
the determination of trap level distribution in polyimide films. J. Electrost. 2010, 68, 243–248.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).