energies
Article
Partial Discharge Measurement under an Oscillating
Switching Impulse: A Potential Supplement to the
Conventional Insulation Examination in the Field
Ming Ren *, Ming Dong *, Chongxing Zhang and Jierui Zhou
State Key Laboratory of Electrical Insulation for Power Equipment, Xi’an Jiaotong University,
Xi’an 710049, China; zhangcx111@126.com (C.Z.); ronaldo@stu.xjtu.edu.cn (J.Z.)
* Correspondence: renming@mail.xjtu.edu.cn (M.R.); dongming@mail.xjtu.edu.cn (M.D.)
Academic Editor: Paul Stewart
Received: 14 June 2016; Accepted: 1 August 2016; Published: 9 August 2016
Abstract: Partial discharge (PD) detection under oscillating switching impulse (OSI) voltage was
performed on three types of insulation defects, including a protrusion on a conductor, a particle
on an insulator surface, and a void in an insulator, which are three kinds of the common potential
insulation hazards in gas insulated power apparatus. Experiment indicated that the PD sequences
under OSI were composed of various combinations of the single pulse, the multiple pulses, and the
reverse polarity pulse. The difference between the PD inception voltage (PDIV) and the breakdown
voltage (BDV) under OSI voltage was greater than that under alternating current (AC) voltage in some
cases, which can provide a more sufficient margin below the BDV for PD diagnosis. The OSI voltage
also showed a better performance for exciting PDs with detectable magnitudes from small-scale
defects, of which the AC voltage was incapable under our test conditions. The different PD activities
with different interfaces under an impulse and a slowly varying voltage were speculated to be
associated with the gradient of the background electric field and the space-charge mobility.
Keywords: gas insulated switchgear; power apparatus; partial discharge; impulse voltage; SF6 ;
insulation diagnosis
1. Introduction
Partial discharge (PD) detection has been widely used in the quality control of high-voltage
power apparatus serving a power system, by which the overall insulation status and local condition
are expected to be examined during routine tests, handing over tests and even field tests. An AC
withstand test combined with PD detection has become a compulsory test during delivery, especially
for an ultrahigh-voltage gas-insulated switchgear or a transmission line. The insulation defects are
caused by manufacturing errors at the factory as well as transportation and assembly in the field. Thus,
a short-time AC resonant withstand test needs to be implemented after assembly and repair, and a PD
test is also applied in an effort to determine any unknown damage. Aside from these tests, some online
diagnostic techniques, particularly the ultrahigh frequency (UHF) method and acoustical wave (AE)
method [1], have been developed to avoid insulation failure during operation. Admittedly, these offline
and online methods have served irreplaceable roles in insulation diagnosis for many years. However,
practical experience and studies have shown that the AC PD test is probably not a universal solution
for finding all types of insulation defects, especially for conductive protrusions and particles on the
insulator surface. This problem is supposed to be subject to diffusion-limited charge accumulation,
which confines the discharge intensity to an undetectable level [2,3]. Some hidden defects are hardly
exposed by the AC voltage field test and can continuously deteriorate the insulation and shorten
the service life of the equipment prior to failure. An increasing number of failures have occurred
Energies 2016, 9, 623; doi:10.3390/en9080623
www.mdpi.com/journal/energies
Energies 2016, 9, 623
2 of 14
Energies
2016,
9, 623
14 an
prior
to the
end
of the designed service life. In this case, the impulse voltage test is expected 2toof be
effective supplement to an insulation quality examination in the field because of its high-field effect
test is expected to be an effective supplement to an insulation quality examination in the field because
and limited destructiveness. To address the inconvenience due to the low delivery efficiency of the
of its high-field effect and limited destructiveness. To address the inconvenience due to the low
impulse generator, the IEC 60060-3 standard recommends oscillating impulse voltages as alternatives
delivery efficiency of the impulse generator, the IEC 60060-3 standard recommends oscillating
to aperiodic
impulses,
are more
feasible inimpulses,
a field test
[4]. Figure
1 shows
oscillating
impulse voltages
aswhich
alternatives
to aperiodic
which
are more
feasibleaninongoing
a field test
[4].
impulse
of a gas-insulated
switchgear
in the
Figurewithstand
1 shows antest
ongoing
oscillating impulse
withstand
test field.
of a gas-insulated switchgear in the field.
Figure 1. On-going oscillating impulse withstand test implemented on gas insulated switchgear in
Figure 1. On-going oscillating impulse withstand test implemented on gas insulated switchgear in
the field.
the field.
Thus, PD detection could be simultaneously performed with withstand tests to exclude the
Thus, PD
detection
simultaneously
performed
withstand
excludethe
the gap
critical
critical
defects
arisingcould
frombetransportation
and
assembly with
[5]. In
the pasttests
few to
decades,
breakdown
stressed
SF6 gas under
impulse
conditions
studied
[6–8].
Relevant in
defects
arising in
from
transportation
and
assembly
[5]. In has
the been
past extensively
few decades,
the gap
breakdown
theories
on the
streamer
and leader
mechanisms
have been
developed
by experiments
and on
stressed
SF6 based
gas under
impulse
conditions
has
been extensively
studied
[6–8]. Relevant
theories[9]based
[10,11].
aims of most
these
previousby
studies
were the
of the power
the simulations
streamer and
leaderThe
mechanisms
haveof
been
developed
experiments
[9]design
and simulations
[10,11].
and of
thethese
failure
mechanisms
of the
SF6 the
insulation.
PDspower
under equipment
impulse conditions
Theequipment
aims of most
previous
studies
were
designThe
of the
and thewere
failure
generally
thought
to
be
phenomena
occurring
in
the
initial
stage
of
breakdown
and
seldom
studied
mechanisms of the SF6 insulation. The PDs under impulse conditions were generally thought to be
in terms of insulation diagnosis. Further, we have previously investigated the PD characteristics and
phenomena
occurring in the initial stage of breakdown and seldom studied in terms of insulation
impact factors under lightning and switching impulses for various artificial defects in SF6 gas [12,13].
diagnosis. Further, we have previously investigated the PD characteristics and impact factors under
On the basis of previous studies, this paper summarizes the PDs under impulse voltages from
lightning and switching impulses for various artificial defects in SF6 gas [12,13].
the viewpoint of insulation diagnosis in practice. The PD behaviors under an oscillating switching
On the basis of previous studies, this paper summarizes the PDs under impulse voltages from
impulse (OSI) are described by exemplifying the typical PD sequences and PD pulses caused by
the different
viewpoint
of insulation
diagnosis in
The PD behaviors
an oscillating
switching
defects.
The effectiveness
of practice.
various voltages
including under
the switching
impulse
(SI),
impulse
(OSI)
are
described
by
exemplifying
the
typical
PD
sequences
and
PD
pulses
caused
oscillating switching impulse (OSI), and AC voltages on the PD diagnosis is quantitatively compared. by
different
defects.for
Thethe
effectiveness
of activities
various voltages
including
oscillating
The reasons
different PD
under an
impulse the
andswitching
a slowly impulse
varying (SI),
voltage
are
switching
impulse
(OSI),
AC voltages
the PD diagnosis
is quantitatively
compared.effect.
The reasons
discussed,
and they
areand
speculated
to be on
associated
with the field-dependent
stabilization
for the different PD activities under an impulse and a slowly varying voltage are discussed, and they
Experimental
are 2.
speculated
to beSetup
associated with the field-dependent stabilization effect.
2.1. Partial Discharge
2. Experimental
Setup(PD) Measurement under Impulse Conditions
Under surge voltage conditions, one difficult problem in PD detection is the separation of PD
signals with a magnitude on the order of approximately a few milliamperes from the great capacitive
displacement
on the
order of approximately
few amperes.
example,
65separation
kV oscillating
Under surgeflow
voltage
conditions,
one difficult aproblem
in PDFor
detection
is athe
of PD
2.1. Partial Discharge (PD) Measurement under Impulse Conditions
signals with a magnitude on the order of approximately a few milliamperes from the great capacitive
Energies 2016, 9, 623
3 of 14
Energies 2016, 9, 623
3 of 14
displacement flow on the order of approximately a few amperes. For example, a 65 kV oscillating
lightning
produce
a
lightning impulse
impulse (OLI)
(OLI)voltage
voltageapplied
appliedtotoaatest
testobject
objectwith
witha acapacitance
capacitanceofof7575pFpFcould
could
produce
displacement
current
above
2 A,
asasshown
a displacement
current
above
2 A,
shownininFigure
Figure2.2.
Figure
Figure 2.
2. PD
PD signals
signals superimposed
superimposed on
on the
the current
current flow
flow under
under an
an impulse.
impulse.
To address this problem, a capacitor could be employed as the balance bridge in parallel with
To address this problem, a capacitor could be employed as the balance bridge in parallel with the test
the test object [14]. This method is effective for the capacitive specimen under an AC voltage but
object [14]. This method is effective for the capacitive specimen under an AC voltage but unsatisfactory
unsatisfactory under impulse conditions because the equivalent circuit is no longer a
under impulse conditions because the equivalent circuit is no longer a lumped-parameter circuit but
lumped-parameter circuit but distributed. Moreover, it is also difficult to obtain the equivalent
distributed. Moreover, it is also difficult to obtain the equivalent parameters for each specific condition,
parameters for each specific condition, especially in a field test. Another approach is to filter the surge
especially in a field test. Another approach is to filter the surge flow that occupies the relatively low
flow that occupies the relatively low frequency range by introducing a second-order RC filtering
frequency range by introducing a second-order RC filtering circuit at the front of the preamplifier, as
circuit at the front of the preamplifier, as shown in Figure 3. However, this approach gives rise to a
shown in Figure 3. However, this approach gives rise to a serious decrease in the sensitivity for PD
serious decrease in the sensitivity for PD detection due to the large ratio of the surge amplitude in
detection due to the large ratio of the surge amplitude in the low-frequency range to the PD magnitude
the low-frequency range to the PD magnitude at a high frequency. In this study, a feasible PD
at a high frequency. In this study, a feasible PD detection circuit was applied under impulse conditions,
detection circuit was applied under impulse conditions, as shown in Figure 3. Signal coupling is
as shown in Figure 3. Signal coupling is realized by two current transformers (CTs), both of which were
realized by two current transformers (CTs), both of which were placed around the ground pillars.
placed around the ground pillars. One wideband CT (CT1, 20 kHz–110 MHz, 5 V/A, IPC-CM-500)
One wideband CT (CT1, 20 kHz–110 MHz, 5 V/A, IPC-CM-500) responds to the total transient signal
responds to the total transient signal over a wide frequency range covering the displacement-current
over a wide frequency range covering the displacement-current and PD signals. CT2 was a
and PD signals. CT2 was a homemade CT (3 kHz–1.7 MHz, 5 V/A), which only responds to the
homemade CT (3 kHz–1.7 MHz, 5 V/A), which only responds to the displacement-current signal in
displacement-current signal in the low-frequency range. By subtracting the output of CT2 from that
the low-frequency range. By subtracting the output of CT2 from that of CT1 with a differential
of CT1 with a differential amplifier, most of the displacement component in the total current signal
amplifier, most of the displacement component in the total current signal could be canceled. Finally,
could be canceled. Finally, the output of the differential module was processed by a high-pass filter to
the output of the differential module was processed by a high-pass filter to separate the PDs from the
separate the PDs from the residual displacement current. A CT could be employed for impulse PD
residual displacement current. A CT could be employed for impulse PD detection because it can
detection because it can quantify the severity of the PD and can be conveniently installed on power
quantify the severity of the PD and can be conveniently installed on power equipment without
equipment without contact. The sensitivities of the measurement systems for the impulse PD and AC
contact. The sensitivities of the measurement systems for the impulse PD and AC PD can be verified
PD can be verified by calibration, and their minimum detectable levels are 5 pC and 2.0 pC, respectively.
by calibration, and their minimum detectable levels are 5 pC and 2.0 pC, respectively. The simplified
The simplified circuit of the oscillating impulse generator used in the experiment is shown in the
circuit of the oscillating impulse generator used in the experiment is shown in the dotted box in
dotted box in Figure 4, by which OSI voltage (33/1000 µs, 8 kHz) can be produced. The waveform of
Figure 4, by which OSI voltage (33/1000 μs, 8 kHz) can be produced. The waveform of the OSI voltage
the OSI voltage is shown in Figure 5.
is shown in Figure 5.
Energies 2016, 9, 623
Energies 2016, 9, 623
Energies 2016, 9, 623
Energies 2016, 9, 623
4 of 14
4 of 14
4 of 14
4 of 14
Figure
preposingsignal
signal process circuit
circuit for PD
PD measurement.
ofof
C1,
C2,
C3,C3,
Figure
3. 3.
AA
measurement.The
Theloop
loopconsisting
consisting
C1,
C2,
Figure
3.
Apreposing
preposing signalprocess
process circuit for
for PD measurement.
The
loop
consisting
of C1,
C2,
C3,
R1,
and
R2
is
employed
as
a
second-order
RC
filter.
R3
and
R4
are
used
for
gain
control.
R1,
and
R2
is
asasaasecond-order
RC
filter.
R3
areThe
used
forgain
gaincontrol.
control.
Figure
3.R2
Aisemployed
preposing
process circuit
for
PD measurement.
loop
consisting
of C1, C2, C3,
R1,
and
employedsignal
second-order
RC
filter.
R3 and
and R4
R4 are
used
for
R1, and R2 is employed as a second-order RC filter. R3 and R4 are used for gain control.
Figure 4. Measurement circuits for impulse PD detection. Rs: the current-limiting resistor, 1 MΩ; C0:
Figure 4. Measurement circuits for impulse PD detection. Rs: the current-limiting resistor, 1 MΩ; C0:
Figure
4. sample,
Measurement
for coupling
impulse capacitor
PDdetection.
detection.
Rdetection,
1 MΩ;
the
test
60–160 circuits
pF; Cp: the
for RPD
1.6 nF; C1: the resistor,
DC charging
s : the current-limiting
Figure
Measurement
for impulse
s: the current-limiting resistor, 1 MΩ; C0:
the
test 4.sample,
60–160 circuits
pF; Cp: the
couplingPD
capacitor
for PD
detection, 1.6 nF; C1: the DC charging
C0capacitor,
:the
thetest
testsample,
sample,
pF;
C
:
the
coupling
capacitor
for
PD
detection,
1.6
nF;
C
:
the
DC
charging
0.0111 60–160
µ60–160
F; Rf: the
wave-front-adjusting
resistor,
3.0
kΩ;
R
t
:
the
wave-tail-adjusting
resistor,
1 DC charging
pF; wave-front-adjusting
Cp: pthe coupling capacitor
for 3.0
PD kΩ;
detection,
nF; C1: the
capacitor, 0.0111 µ F; Rf: the
resistor,
Rt: the 1.6
wave-tail-adjusting
resistor,
capacitor,
µF;
RR
wave-front-adjusting
resistor,3.0
3.0kΩ;
kΩ;RRt: tthe
: the
wave-tail-adjusting
resistor,
300
kΩ; L0.0111
f: the adjusting
inductor,
85.3 mH.
f : f:the
capacitor,
0.0111
µ
F;
the
wave-front-adjusting
resistor,
wave-tail-adjusting
resistor,
300 kΩ; Lf: the adjusting inductor, 85.3 mH.
300300
kΩ;
LfL: fthe
kΩ;
: theadjusting
adjustinginductor,
inductor, 85.3
85.3 mH.
mH.
Figure 5. The waveform of the OSI voltage. T1: front time; T2: time-to-half-value; f: oscillating frequency.
Figure 5. The waveform of the OSI voltage. T1: front time; T2: time-to-half-value; f: oscillating frequency.
Figure
Thewaveform
waveformof
ofthe
the OSI
OSI voltage.
voltage. TT1: :front
time; T2: time-to-half-value; f: oscillating frequency.
Figure
5.5.The
1 front time; T2 : time-to-half-value; f : oscillating frequency.
2.2. Artificial Defect Models and Experimental Setup
2.2. Artificial Defect Models and Experimental Setup
2.2. Artificial Defect Models and Experimental Setup
2.2. Artificial
Models aand
Experimentaltest
Setup
In ourDefect
experiment,
stainless-steel
chamber that could withstand a gas pressure up to
In our experiment, a stainless-steel test chamber that could withstand a gas pressure up to
0.7 MPa
andexperiment,
applied voltages
up to 150 kV
Hz AC that
peak)could
was used
to build
a SFpressure
6 gas insulated
In our
a stainless-steel
test(50
chamber
withstand
a gas
up to
ourand
experiment,
a stainless-steel
testkV
chamber
could was
withstand
gas pressure
upinsulated
to 0.7 MPa
0.7In
MPa
applied voltages
up to 150
(50 Hz that
AC peak)
used toa build
a SF6 gas
0.7 MPaas
and
applied
voltages
to specific
150 kV (50
Hz AC system
peak) was
to build
SF6 gas insulated
system,
shown
in Figure
6a.up
Any
insulation
mayused
contain
somea insulation
defects,
system,
as shown
inup
Figure
6a.kV
Any
may
contain
insulationsystem,
defects,as
and
applied
voltages
to 150
(50specific
Hz ACinsulation
peak) wassystem
used to
build
a SF6some
gas insulated
system,arise
as shown
in Figure
Any specific insulation
system
may contain
some insulation
defects,
which
from the
design,6a.
manufacturing,
assembling,
and service
stress history.
According
to the
whichinarise
from
design,
manufacturing,
assembling,
and service
history.defects,
According
to the
shown
Figure
6a.the
Any
specific
insulation system
may contain
somestress
insulation
which
arise
which arise
design,
manufacturing,
assembling,
and
service stress
history.
to the
nature
of thefrom
two the
or three
major
dielectric boundaries,
the
insulation
defects
couldAccording
be classified
and
nature
of
the
two
or
three
major
dielectric
boundaries,
the
insulation
defects
could
be
classified
and
from
the design,
manufacturing,
assembling,
andcan
service
stress
history.
According
the
natureand
of
the
nature
of the
or three major
boundaries,
the
insulation
defects
could to
beConsidering
classified
simplified
intotwo
small-scale
defect dielectric
models which
be used
in experimental
studies.
the
simplified
into
small-scale
defect
modelsthe
which
can be used
in experimental
studies.and
Considering
the
two
or
three
major
dielectric
boundaries,
insulation
defects
could
be
classified
simplified
into
simplified into
models
which can
used
in three
experimental
studies.including
Considering
the
contribution
of small-scale
the electricdefect
field to
the excitation
ofbe
the
PDs,
defect models
a metal
contribution
of the
electric
field can
to the
excitation
of the PDs, three
defect
models including
a metal
small-scale
defect
models
which
be
used
in
experimental
studies.
Considering
the
contribution
contribution
of
the
electric
field
to
the
excitation
of
the
PDs,
three
defect
models
including
a
metal
protrusion (see Figure 6b), a metal particle (a metal needle was employed as the particle) on an
protrusion
(see
Figure
6b),
a metal of
particle
(a metal
needle
was
employed
as the
particle)
on an
ofinsulator
the electric
field
to the
excitation
the PDs,
three
defect
models
a were
metal
protrusion
protrusion
(see
Figure
6b),
a6c),
metal
(a metal
was
employed
as 6d)
the
particle)
on an
surface
(see
Figure
andparticle
a gaseous
void
inneedle
an insulator
(seeincluding
Figure
developed
insulator
surface
(see
Figure
6c),
and
a
gaseous
void
in
an
insulator
(see
Figure
6d)
were
developed
(see
Figure
6b),
a metal
particle
(a and
metal
needle was
employed
thetest,
particle)
anwere
insulator
surface
insulator
surface
(see Figure
a gaseous
void
in an insulator
(see
Figure
6d)
developed
to
simulate
the
defects
with 6c),
different
dielectric
interfaces.
In as
our
all theon
defect
models
were
to
simulate
the
defects
with
different
dielectric
interfaces.
In
our
test,
all
the
defect
models
were
(see
a gaseous
void
in an
insulator
(see Figure
6d)test,
were
to
simulate
to Figure
simulate
theand
defects
with
different
dielectric
interfaces.
our
alldeveloped
the defect
models
werethe
placed
in a6c),
pair
of
parallel
plate
copper
electrodes
which
canInprovide
a uniform
background
electric
placed in a pair of parallel plate copper electrodes which can provide a uniform background electric
defects
with
different
dielectric
our test,
all
theprovide
defectofmodels
were
placed
inelectric
a pair
placedThe
in aprotrusion
pair of parallel
plate
copper
which
can
astainless
uniform
background
field.
defects
(asinterfaces.
shown electrodes
in In
Figure
6b,c)
were
made
steel
(Crl8Ni9Ti).
Toof
field. The protrusion defects (as shown in Figure 6b,c) were made of stainless steel (Crl8Ni9Ti). To
field.
The
protrusion
defects
(as
shown
in
Figure
6b,c)
were
made
of
stainless
steel
(Crl8Ni9Ti).
To
parallel
plate
copper
electrodes
which
can
provide
a
uniform
background
electric
field.
The
protrusion
make the void defect model (see Figure 6d), two epoxy resin plates, each with an artificial pitting at
make the void defect model (see Figure 6d), two epoxy resin plates, each with an artificial pitting at
make
the shown
void
model
(see
Figure
6d), two
epoxyof
resin
plates,
each
artificial
pitting
at
defects
(as
in Figure
were
made
stainless
steel
(Crl8Ni9Ti).
Toanmake
the Resin
void defect
the
center,
weredefect
spliced
face6b,c)
to
face
by using
aofmixture
E51
epoxy
resinwith
(Deyuan
Epoxy
Co.,
the center, were spliced face to face by using a mixture of E51 epoxy resin (Deyuan Epoxy Resin Co.,
the center, were spliced face to face by using a mixture of E51 epoxy resin (Deyuan Epoxy Resin Co.,
Energies 2016, 9, 623
5 of 14
model (see Figure 6d), two epoxy resin plates, each with an artificial pitting at the center, were spliced
Energies 2016, 9, 623
of 14
face
to face by using a mixture of E51 epoxy resin (Deyuan Epoxy Resin Co., Ltd., Feicheng,5China)
and 10% DDS curing agent (Jianxin Imp & Exp Co., Ltd., Cangzhou, China). To ensure a good contact
Ltd., Feicheng, China) and 10% DDS curing agent (Jianxin Imp & Exp Co., Ltd., Cangzhou, China).
between the electrode and the epoxy resin without gas gap, the electrode was also spliced to the epoxy
To ensure a good contact between the electrode and the epoxy resin without gas gap, the electrode
resin
plate which has a void inside. After splicing, the defect samples were kept in a vacuum oven at
was also spliced to the epoxy resin plate which has a void inside. After splicing, the defect samples
˝
55 were
C forkept
over
h for drying.
dielectric
constant
the relative
E51 anddielectric
DDS mixture
wasof4.1–4.3,
in24
a vacuum
oven The
at 55relative
°C for over
24 h for
drying.ofThe
constant
the
which
is
consistent
with
that
of
the
epoxy
resin
block.
The
scale
of
the
artificial
defect
was
E51 and DDS mixture was 4.1–4.3, which is consistent with that of the epoxy resin block. Thecontrolled
scale of
bythe
theartificial
parameters
listed
Table 1. by the parameters listed in Table 1.
defect
was in
controlled
Figure
6 test chamber and PD defect models. (a) Test chamber with PD sensors; (b) Metal
Figure
6. 6.SFSF
6 test chamber and PD defect models. (a) Test chamber with PD sensors; (b) Metal
protrusion; (c) Metal particle on insulator surface; (d) Gaseous void in insulator. The plate electrodes
protrusion; (c) Metal particle on insulator surface; (d) Gaseous void in insulator. The plate electrodes
are made of copper, and the protrusion and needle are made of steel. D is the gap distance between
are made of copper, and the protrusion and needle are made of steel. D is the gap distance between the
the plate electrodes. θ is the inclination angle of the insulator. L is the length of the protrusion or the
plate electrodes. θ is the inclination angle of the insulator. L is the length of the protrusion or the needle
needle on the surface. a and b are the radius and the height of the gaseous void in the insulator,
on the surface. a and b are the radius and the height of the gaseous void in the insulator, respectively.
respectively. The permittivity of epoxy is 4.1.
The permittivity of epoxy is 4.1.
Table 1. Scale parameters of the three defects investigated.
Table 1. Scale parameters of the three defects investigated.
Gap Between the Plate Electrodes
Defect Scale Parameters
Defects
(Background
Field)
(Locally
Enhanced
Field)
Gap Between the Plate Electrodes
Defect Scale
Parameters
Defects
Protrusion
D = 30 mm Field)
r = 0.7 mm,
L = 1, 2,Field)
5 10 mm
(Background
(Locally
Enhanced
Surface particle
D = 35 mm, θ = 60°
r = 0.7 mm, L = 1, 2, 5, 10 mm
Protrusion
D = 30 mm
r = 0.7 mm, L = 1, 2, 5 10 mm
x{=a/b}
2 mm
˝
Surface
DD==3520mm,
r = 0.7 mm,
L = 1,=2,0.5,
5, 10
Void particle
mmθ = 60
2b} = 2, 15, 45, 340 mm3
V{=πax{=a/b}
= 0.5, 2
Void
D = 20 mm
V{=πa2 b} = 2, 15, 45, 340 mm3
The overall configuration of the PD detection system is shown in Figure 7. The PD signals are
recorded by a digital oscilloscope (DSO, LeCroy 64Xs-B, Teledyne LeCroy, Chestnut Ridge, NY,
The overall configuration of the PD detection system is shown in Figure 7. The PD signals are
USA), sampling rate 10 Gs/s; analog bandwidth 600 MHz). To avoid electromagnetic (EM)
recorded by a digital oscilloscope (DSO, LeCroy 64Xs-B, Teledyne LeCroy, Chestnut Ridge, NY, USA),
interference, metal armored double-shielded measurement cables were used for signal transmission.
sampling rate 10 Gs/s; analog bandwidth 600 MHz). To avoid electromagnetic (EM) interference,
All recording apparatus were placed in an EM-shielded room. The voltage divider used in the test
metal armored double-shielded measurement cables were used for signal transmission. All recording
had a response time less than 100ns and a measurement range up to 150 kV. The coupling capacitance
apparatus
were placed in an EM-shielded room. The voltage divider used in the test had a response
value in parallel with the test object was 1000 pF. A photomultiplier tube (PMT, R7600U, Hamamatsu,
time
less
than
100nsover
andaawavelength
measurement
range
up to 150
kV.
coupling
capacitance
value
in parallel
Japan), response
range
of 300–850
nm)
forThe
optical
detection
was placed
behind
the
with
the
test
object
was
1000
pF.
A
photomultiplier
tube
(PMT,
R7600U,
Hamamatsu,
Japan),
response
observation window of the chamber and employed to assist with the PD current measurement. The
over
a wavelength
300–850were
nm) constructed
for optical detection
wasfused
placed
behind
observation
optical
windows range
of the ofchamber
of synthetic
silica
glassthe
(above
85%
window
of
the
chamber
and
employed
to
assist
with
the
PD
current
measurement.
The optical
transmittance from 175 nm onward and 90% from 220 nm onward).
windows of the chamber were constructed of synthetic fused silica glass (above 85% transmittance
from 175 nm onward and 90% from 220 nm onward).
Energies 2016, 9, 623
6 of 14
Energies 2016, 9, 623
6 of 14
Figure 7. Schematic of the overall impulse PD detection system.
Figure 7. Schematic of the overall impulse PD detection system.
In our test, the PD inception voltages (PDIVs) under OSI and AC voltages were determined by
In ourvoltage
test, the
PD inception
voltagesof(PDIVs)
OSI andby
ACthevoltages
were
step-up
method.
The amplitude
the OSI under
was controlled
output of
the determined
charging
by step-up
voltage
method.
The
amplitude
of
the
OSI
was
controlled
by
the
output
of
transformer of the Marx generator. The same test procedure was repeated on the same the
test charging
object
transformer
generator.
test
procedure
the same
test
until
until that of
thethe
PDMarx
current
pulses orThe
lightsame
pulses
were
detectedwas
for 5repeated
times foron
every
10 tests
at object
a certain
thatvoltage
the PDwhich
currentwas
pulses
or lightaspulses
were
detected PDIV
for 5 times
every
10The
tests
at a method
certain voltage
identified
the fifty
probability
(abbr.for
PDIV
50%).
same
was
which
was to
identified
as BDVs
the fifty
probability
PDIV
(abbr. PDIV
).
The
same
method
was
applied
applied
determine
under
AC and OSI
voltages.
To investigate
the
PD
activity
under
the
50%
equivalent BDVs
voltage
levelsAC
butand
different
voltage types,
the predetermined
PDIVsunder
were the
used
as the
to determine
under
OSI voltages.
To investigate
the PD activity
equivalent
benchmarks
of different
the applied
voltage
level
1.1 times PDIV
(abbr.
1.2 benchmarks
times PDIV of
voltage
levels but
voltage
types,
the(e.g.,
predetermined
PDIVs
were1.1PDIV),
used as the
PDIV)
and solevel
on).(e.g.,
Considering
influence
the residual
spacePDIV
charge,
time and
interval
the(1.2
applied
voltage
1.1 timesthe
PDIV
(abbr. of
1.1PDIV),
1.2 times
(1.2the
PDIV)
so on).
between
two
tests
was
kept
above
5
min.
Considering the influence of the residual space charge, the time interval between two tests was kept
above 5 min.
3. Results
3. Results
3.1. PDs under OSI Voltage
3.1. PDsThe
under
OSI Voltage
differences
in the PDs caused by various defects are reflected in their PD sequences in terms
ofThe
thedifferences
pulse magnitude
and
time by
interval.
various
PD sequences
complicated
in the PDs
caused
variousThe
defects
are reflected
in their involve
PD sequences
in terms
mechanisms
ranging
from
the
generation
of
an
initiatory
electron
to
streamer
corona
and
leaders
in
of the pulse magnitude and time interval. The various PD sequences involve complicated mechanisms
the same or different channels, which have been discussed in our previous work [12]. Overall, the
ranging from the generation of an initiatory electron to streamer corona and leaders in the same or
three different types of PD sequences occurring in the different interfaces are described in general
different channels, which have been discussed in our previous work [12]. Overall, the three different
as follows:
types of PD sequences occurring in the different interfaces are described in general as follows:
i. PDs at the metal protrusion in the gas: As described in our previous paper [12], the various
i. PDs at the metal protrusion in the gas: As described in our previous paper [12], the various
types of PD sequences could be detected under impulse voltages including a single pulse, a single
types
of PD
sequences
could be smaller
detected
undercompact
impulseand
voltages
including
a single
pulse,
single
pulse
followed
by successive
pulses,
incompact
multiple
pulses,
and aeven
pulse
followed
by
successive
smaller
pulses,
compact
and
incompact
multiple
pulses,
and
even
reverse-polarity pulses. The single pulse (see Figure 8a(i)) and the compact (see Figure 8a(i)) and
reverse-polarity
pulses.
The
single
pulse
(see
Figure
8a(i))
and
the
compact
(see
Figure
8a(i))
and
incompact multiple pulses (Figure 8a(ii)) are the most common types under an OSI voltage.
incompact
multiple
pulses
(Figure
8a(ii))
are the
under
OSI voltage. region
ii. PDs
from the
triple
junction
between
the most
metal,common
gas, and types
insulator:
Theantriple-junction
ii.
PDs
from
the
triple
junction
between
the
metal,
gas,
and
insulator:
Theaccelerate
triple-junction
(the interface where the insulator, electrode, and gas are in close proximity) can
the
region
(the interface
where
the insulator,
electrode,
and gas and
are in
close proximity)
accelerate
generation
of initial
electrons
via enhanced
field emission
electron
release by can
ion impact
andthe
surface photon
emission.
Therefore,
in general,
a metal
particle
an insulator
surface
could
reduce
generation
of initial
electrons
via enhanced
field
emission
andonelectron
release
by ion
impact
and
the electrical
of Therefore,
the dielectric
the gas and
solid. As shown
Figurereduce
8b,
surface
photon strength
emission.
in interface
general, between
a metal particle
on the
an insulator
surfaceincould
if a relatively
impulse
voltage
is applied
(33 kV,the
seegas
Figure
PD As
is large
in magnitude
theeven
electrical
strengthlow
of the
dielectric
interface
between
and8b),
thethe
solid.
shown
in Figure 8b,
at
the
rising
edges
of
the
oscillating
periods.
These
single
or
multiple
pulses
have
features
to
even if a relatively low impulse voltage is applied (33 kV, see Figure 8b), the PD is large insimilar
magnitude
those
occurring
the absence
of an
insulator
surface pulses
but are have
more features
active. similar to
at the
rising
edgesat
ofthe
theprotrusion
oscillatinginperiods.
These
single
or multiple
those occurring at the protrusion in the absence of an insulator surface but are more active.
Energies 2016, 9, 623
7 of 14
Energies 2016, 9, 623
7 of 14
iii. PDs in the gaseous voids in an insulator: The lower dielectric constant of the gaseous
iii. PDs in the gaseous voids in an insulator: The lower dielectric constant of the gaseous voids
voids results in a locally enhanced field that causes PDs to occur successively inside the void.
results in a locally enhanced field that causes PDs to occur successively inside the void. Under an
Under
an impulse
rate of change
in the applied
fieldtime
withistime
is greater
than
impulse
voltage, voltage,
the rate the
of change
in the applied
electricelectric
field with
greater
than the
the
dissipation
rate
of
the
PD
residual
charges
in
the
void.
In
this
case,
the
direction
of
the
total
dissipation rate of the PD residual charges in the void. In this case, the direction of the total electric
electric
field
is
changed
when
the
background
electric
field
decreases
to
a
certain
value,
resulting
field is changed when the background electric field decreases to a certain value, resulting inin
reverse-polarity
PDs,
most
of which
consist
of successive
pulses
withwith
a high
repetition
rate. The
reverse-polarity
PDs,
most
of which
consist
of successive
pulses
a high
repetition
rate.typical
The
PDtypical
sequences
detected
for
a
void
in
an
insulator
under
an
OSI
are
shown
in
Figure
8c.
PD sequences detected for a void in an insulator under an OSI are shown in Figure 8c.
(a)
(b)
(c)
Figure
PDcurrent
currentand
andlight
lightpulses
pulses(sequences)
(sequences) excited
excited by
6-insulated system
Figure
8.8.PD
by the
thethree
threedefects
defectsininaaSF
SF
6 -insulated system
under OSIs. (a) a protrusion fixed on a ground plate, 42 kV OSI, negative high voltage (HV); (b) metal
under OSIs. (a) a protrusion fixed on a ground plate, 42 kV OSI, negative high voltage (HV); (b) metal
particle on an insulator surface in contact with the HV plate, 38 kV OSI, positive HV; (c) cylindrical
particle on an insulator surface in contact with the HV plate, 38 kV OSI, positive HV; (c) cylindrical
void inside an insulator filled with air, 33 kV OSI, positive HV.
void inside an insulator filled with air, 33 kV OSI, positive HV.
3.2. Efficiencies of the OSI, SI, and AC Voltages for Exciting PDs
3.2. Efficiencies of the OSI, SI, and AC Voltages for Exciting PDs
3.2.1. PD Inception and Breakdown under Various Applied Voltages
3.2.1. PD Inception and Breakdown under Various Applied Voltages
Potential insulation defects are expected to be eliminated by manufacturing and a field test
Potential insulation defects are expected to be eliminated by manufacturing and a field test before
before the power equipment is put into operation. Owing to corona stabilization, some potential
the power equipment is put into operation. Owing to corona stabilization, some potential defects
defects are not sensitive to the power frequency AC. With an increase in the applied voltage, the
are not sensitive to the power frequency AC. With an increase in the applied voltage, the insulation
Energies 2016, 9, 623
8 of 14
will experience the process from PD inception to complete flashover, which makes PD detection
possible. For manufacturers and power operation departments, they hope the defects can be detected
at a relatively low applied voltage below breakdown voltage. Therefore, a sufficient difference between
the inception voltage and the breakdown voltage (BDV) is the premise of an effective PD diagnosis.
For example, to find the detectable PDs under applied voltage of 0.7 times BDV is certainly better
than under 0.9 times BDV. However, discharges excited by defects are driven by complex mechanisms
in different ways. The time lag of the initiatory electron, the probability of avalanche formation,
Energies 2016, 9, 623
8 of 14
and the effect of space charge stabilization are involved. Although some studies and the reports of
insulation
willon
experience
the process
from PD(CIGRE)
inception to
complete flashover,
which
makes
PD
International
Council
Large Electric
Systems
workgroups
[15–17]
have
summarized
the
detection possible. For manufacturers and power operation departments, they hope the defects can
availabilities
of different types of voltages for examining the insulation status, the effectiveness of
be detected at a relatively low applied voltage below breakdown voltage. Therefore, a sufficient
these voltages
has not
beenthe
quantitatively
investigated
through
experiments.
difference
between
inception voltage
and the breakdown
voltage
(BDV) is the premise of an
To quantify
effectiveness
of thetoAC,
and OSIPDs
voltages
for exciting
PDs,
the detectable
effectivethe
PD diagnosis.
For example,
find SI,
the detectable
under applied
voltage of
0.7 times
is certainly
under 0.9 times
BDV.various
However, applied
dischargesvoltages.
excited by defects
driven
scale of theBDV
defects
wasbetter
also than
investigated
under
Then,arethe
PD inception
by complex mechanisms in different ways. The time lag of the initiatory electron, the probability of
voltage (PDIV) and the PD excitation efficiency ξ, which is defined as the ratio of the BDV to the PDIV
avalanche formation, and the effect of space charge stabilization are involved. Although some studies
(Equation (1)),
were
analyzed
for different
a specific
voltage type,
greater
and the
reports
of International
Council cases.
on LargeFor
Electric
Systemsapplied
(CIGRE) workgroups
[15–17]
have efficiency
summarized
the
availabilities
of
different
types
of
voltages
for
examining
the
insulation
status,
the
ξ means a lower applied voltage under which the detectable PDs can be exited from a defect.
By this
effectiveness of these voltages has not been quantitatively investigated through experiments.
parameter, the capabilities of the different applied voltage on exciting detectable PDs can be evaluated
To quantify the effectiveness of the AC, SI, and OSI voltages for exciting PDs, the detectable scale
and compared
of thequantitatively.
defects was also investigated under various applied voltages. Then, the PD inception voltage
(PDIV) and the PD excitation efficiency ξ, which is defined as the ratio of the BDV to the PDIV
BDV50% applied voltage type, greater efficiency
(Equation (1)), were analyzed for differentξcases.
“ For a specific
PDIV
ξ means a lower applied voltage under which the
detectable
50% PDs can be exited from a defect. By this
parameter, the capabilities of the different applied voltage on exciting detectable PDs can be
BDVevaluated
voltage at fifty percent probability and PDIV50% is partial
50% is breakdown
and compared quantitatively.
(1)
where
discharge
inception voltage at fifty percent probability.
BDV50%
Figure 9 shows the PDIVs and the valuesξ=of ξ for
different defects under AC and(1)positive OSI
PDIV50%
voltages as a function of the gas pressure. For the two types of defects exposed to a SF6 gas atmosphere,
BDV50% as
is breakdown
voltage at increased
fifty percentin
probability
andofPDIV
50% is partial discharge
their PDIVswhere
increased
the gas pressure
the range
0.05–0.4
MPa. As predicted, for
inception voltage at fifty percent probability.
a protrusion, the
PDIVs
of
both
the
positive
and
negative
points
under
an
OSI
voltage are greater than
Figure 9 shows the PDIVs and the values of ξ for different defects under AC and positive OSI
those undervoltages
an ACasvoltage
over
rangeFor
of gas
pressures
This
attributed
to the
a function
of the
the entire
gas pressure.
the two
types ofinvestigated.
defects exposed
to aisSF
6 gas
theirprior
PDIVsto
increased
as discharge.
the gas pressure
in the
of 0.05–0.4 MPa.
As
presence ofatmosphere,
the time lag
the first
Forincreased
a particle
onrange
an insulator,
the PDIVs
under
predicted, for a protrusion, the PDIVs of both the positive and negative points under an OSI voltage
an OSI voltage
are greater than those under an AC voltage, except for a low gas pressure. The PD
are greater than those under an AC voltage over the entire range of gas pressures investigated. This
excitation efficiency
a negative
voltage
is greater
that of
AC voltage
for gas pressures
is attributedfor
to the
presence ofOSI
the time
lag prior
to the firstthan
discharge.
Foran
a particle
on an insulator,
less than 0.2
but
lower
forvoltage
gas pressures
than
0.2anMPa.
For most
investigated,
the
theMPa
PDIVs
under
an OSI
are greater greater
than those
under
AC voltage,
exceptcases
for a low
gas
pressure.
The
PD
excitation
efficiency
for
a
negative
OSI
voltage
is
greater
than
that
of
an
AC
voltage
efficiency of the positive OSI voltage is greater than that of the AC voltage. In general, the OSI voltage
for gas pressures less than 0.2 MPa but lower for gas pressures greater than 0.2 MPa. For most cases
provides PD detection with a more sufficient margin below the BDV compared to the AC voltage,
investigated, the efficiency of the positive OSI voltage is greater than that of the AC voltage. In
especially for
extreme
locally
defects
such
protrusion
and a particle
an insulator
general,
the OSI
voltageenhanced
provides PD
detection
withas
aa
more
sufficient margin
below the on
BDV
compared
to
the
AC
voltage,
especially
for
extreme
locally
enhanced
defects
such
as
a
protrusion
and
surface. Moreover, in most cases, ξ exhibits an increasing trend as the gas pressure increases. It is
a particle
on an insulator
Moreover,
in most
ξ exhibits an increasing
as the gas which is
inferred that
this upward
trendsurface.
in ξ is
attributed
tocases,
the intensifying
coronatrend
stabilization,
pressure increases. It is inferred that this upward trend in ξ is attributed to the intensifying corona
an intrinsicstabilization,
feature of which
SF6 inis the
gas pressure
from
one to several
bars.
an intrinsic
feature ofrange
SF6 in the
gas pressure
range from
one to several bars.
(a)
(b)
Figure 9. Cont.
Energies 2016, 9, 623
9 of 14
Energies 2016, 9, 623
9 of 14
(c)
(d)
Figure 9. Values of the PDIV and the ratio of the BDV to the PDIV for different defects under AC and
positiveof
OSI
voltages
pressure
for BDV
(a), (b) to
a metal
on a conductor
Figure 9. Values
the
PDIVversus
and the
thegas
ratio
of the
the protrusion
PDIV forfixed
different
defectsunder
under AC and
positive and negative voltages and (c), (d) a metal particle on an insulator surface in contact with the
positive OSI voltages versus the gas pressure for (a), (b) a metal protrusion fixed on a conductor under
electrode under positive and negative voltages.
positive and negative voltages and (c), (d) a metal particle on an insulator surface in contact with the
3.2.2.
Detectable
Scales
of negative
the Three Defects
under Various Applied Voltages
electrode
under
positive
and
voltages.
Electrical discharges in an insulation system could be characterized by the order of magnitude,
the detectability,
relevance
for aging
and failure.
TheApplied
different types
of discharges could be
3.2.2. Detectable
Scales ofand
thetheThree
Defects
under
Various
Voltages
classified as noncritical and critical for insulation. Some noncritical discharges such as surface
Electrical
discharges
inTownsend
an insulation
system
could
be characterized
by characterized
the order ofbymagnitude,
emission,
glows, and
avalanches
that exist
during
normal operation are
very lowand
and quasi
continuous currents.
Critical
streamers, leaders,
anddischarges
sparks
the detectability,
the relevance
for aging
anddischarges
failure. such
Theasdifferent
types of
could
are caused by a defect-induced locally enhanced electric field and relevant to the aging and failure of
be classified
as
noncritical
and
critical
for
insulation.
Some
noncritical
discharges
such
as
surface
dielectric insulation. Although most types of discharges have a pulsed nature, not all of them can be
emission, glows,
Townsend
that exist during
are characterized
by
detectedand
by the
sensitivity avalanches
limited PD measurement
owing tonormal
the low operation
intensities. Therefore,
a
streamer
discharge is currents.
of great concern
in PDdischarges
diagnosis because
induces
a detectable
pulsedand sparks
very low and
quasitype
continuous
Critical
suchitas
streamers,
leaders,
and is the root of more intensity types such as leaders, sparks, and breakdown. The intensity
are caused current
by a defect-induced
locally enhanced electric field and relevant to the aging and failure of
and the probability of the occurrence of this type of discharge depend upon the electric field for
dielectric insulation.
Although
most
types
of discharges
have adetectable
pulsed scales
nature,
not
alltypes
of them can be
excitation and the geometry and scale
of the
defect. The minimum
of the
three
detected byofthe
sensitivity
owing toexperimentally
the low intensities.
Therefore, a streamer
defects
under SI,limited
OSI, andPD
AC measurement
voltages were investigated
and are quantitatively
summarized
in thisconcern
section. in PD diagnosis because it induces a detectable pulsed current and
type discharge
is of great
For a void in an insulator, the critical field of the streamer discharge inception voltage is
is the root determined
of more intensity
types such as leaders, sparks, and breakdown. The intensity and the
by its volume and shape. The shape of the void can be described by the ratio of the radius
probability(a)
oftothe
occurrence
of this
type
of discharge
depend
upon
electric
for excitation
the length (b) (denoted
by x).
According
to Pederson’s
Model [18],
thethe
critical
corona field
inception
voltage, U
inc, can
be estimated
by
and the geometry
and
scale
of the defect.
The minimum detectable scales of the three types of defects
Uinc  ( E / experimentally
p)cr  p  d  F
under SI, OSI, and AC voltages were investigated
and are quantitatively(2)summarized
in this section.
where the function F depends on the product of gas pressure (p) and defect radius (r) for a void and
For a isvoid
givenin
by an insulator, the critical field of the streamer discharge inception voltage is
1
determined by its volume and shape. The
shape
can be described by the
F  [1
 B / (2 of
p  rthe
)n ]  fvoid
(3) ratio of the
radius (a) to
the length (b) (denoted by x). According to Pederson’s Model [18], the critical corona
where B = (Kcr/C)1/β·(E/p)cr−1 and f = 3εr/(2εr + 1). For an air-filled void, Kcr = 9, C = 4.15 × 10−4, n = 1/β,
inception voltage,
Uinc
, can
β = 2, and
(E/p)
cr = be
25 estimated
V·(Pa·m)−1. f by
is the dimensionless field enhancement factor defined in
Equation (4). In this work, f was determined by a finite element method (FEM).
Uinc “
¨d¨F
UpE{pq
0 l
void  f
cr ¨Ep
(4)
(2)
where ΔUvoid is the potential difference along the void; E0 is the background electric field; l is the
where the function
F depends
length of the
defect, i.e., on
2b. the product of gas pressure (p) and defect radius (r) for a void and is
given by
To investigate the influence of void shape on electric field distribution, FEM numerical
n different
calculations were performed on the
with
F voids
“ r1 in
` dielectric
B{p2p ¨ rq
s ¨ f ´1 ratio x but the same volume. It
(3)
indicates that, the dimensionless field enhancement factor (f) defined in Equation (4) decreases
where B = (Kcr /C)1/β ¨(E/p)cr ´1 and f = 3εr /(2εr + 1). For an air-filled void, Kcr = 9, C = 4.15 ˆ 10´4 ,
n = 1/β, β = 2, and (E/p)cr = 25 V¨(Pa¨m)´1 . f is the dimensionless field enhancement factor defined in
Equation (4). In this work, f was determined by a finite element method (FEM).
∆Uvoid “ f ¨ E0 ¨ l
(4)
where ∆Uvoid is the potential difference along the void; E0 is the background electric field; l is the
length of the defect, i.e., 2b.
Energies 2016, 9, 623
10 of 14
To investigate the influence of void shape on electric field distribution, FEM numerical calculations
were
Energiesperformed
2016, 9, 623 on the voids in dielectric with different ratio x but the same volume. It indicates
10 of 14
that, the dimensionless field enhancement factor (f ) defined in Equation (4) decreases approximately
approximately
linearly
with
the
increase
of x. Based
on Equation
(4), the
critical
linearly with the
increase
of x.
Based
on Equation
(3) and
Equation(3)
(4),and
the Equation
critical corona
inception
corona
voltage (Utoincincrease
) is indicated
theratio
increase
of the ratio
(x). Figure
10
voltage inception
(Uinc ) is indicated
with to
theincrease
increasewith
of the
(x). Figure
10 shows
the field
shows
the fieldfactor
enhancement
factor
thestreamer
void andinception
the streamer
inception
versus
(the
ratio
enhancement
of the void
andofthe
voltage
versusvoltage
x (the ratio
of xthe
radius
of the
radius
(a) to
theItlength
It can be
concluded
thatvoid
the prolate
void in the
insulator
is more
(a)
to the
length
(b)).
can be(b)).
concluded
that
the prolate
in the insulator
is more
critical
than
critical
thanvoid,
the oblate
void, even
they
havevolume
the same
volumethe
because
former
can
induce
the oblate
even though
theythough
have the
same
because
formerthe
can
induce
a greater
aLaplace
greaterfield.
Laplace
is noted
that this inception
voltage
can beofthought
of as a necessary
but not
It isfield.
notedItthat
this inception
voltage can
be thought
as a necessary
but not sufficient
sufficient
for PD occurrence
the
for theelectron
initiatory
electron
also be
condition condition
for PD occurrence
because thebecause
time lag
fortime
the lag
initiatory
should
alsoshould
be considered.
considered.
this
study,oftwo
of voidwith
defect
= 0.5 andto2investigate
are employed
to
In this study,Intwo
groups
voidgroups
defect models
x =models
0.5 andwith
2 are xemployed
the PD
investigate
the PD
intensity
undervoltages.
different applied voltages.
intensity under
different
applied
Figure
Figure 10.
10. The
The influence
influence of
of the
the void
void shape
shape on
on the
the electric
electric field
field in
in the
the void
void and
and the
the streamer
streamer inception
inception
3. The relative permittivities of the
3
voltage.
The
volume
of
the
void
in
the
calculation
is
1.57
mm
voltage. The volume of the void in the calculation is 1.57 mm . The relative permittivities of the
insulator
and the
the gas
gas in
in the
thevoid
voidare
are4.1
4.1and
and1.0,
1.0,respectively.
respectively.The
Thesurface
surface
resistivity
inner
face
insulator and
resistivity
of of
thethe
inner
face
of
−20 Ω.
´20
of
the
void
is
10
the void is 10
Ω.
Gaseous voids were created at the center of the gap length between the upper and lower plate
Gaseous voids were created at the center of the gap length between the upper and lower plate
electrodes to ensure a symmetrical field distribution. The polarity effect could not be exerted on the
electrodes to ensure a symmetrical field distribution. The polarity effect could not be exerted on the PD
PD events. Thus, a conclusion can be drawn by applying one polarity impulse to the sample. The
events. Thus, a conclusion can be drawn by applying one polarity impulse to the sample. The average
average PD magnitudes as a function of the void volume with prolate (x = 0.5) and oblate (x = 2)
PD magnitudes as a function of the void volume with prolate (x = 0.5) and oblate (x = 2) shapes are
shapes are shown in Figure 11a,b, respectively. Compared with the prolate voids, the oblate voids
shown in Figure 11a,b, respectively. Compared with the prolate voids, the oblate voids produced PDs
produced PDs with a higher intensity, which were reflected in the pulse magnitudes (pC). The AC
with a higher intensity, which were reflected in the pulse magnitudes (pC). The AC voltage was not
voltage was not able to excite PDs above the detectable level (5 pC) for a small prolate void with a3
able to excite PDs above the detectable level (5 pC) for a small prolate void with a volume of 1.5 mm .
volume of 1.5 mm3. In contrast, the PD under an OSI voltage has the greater average magnitude, even
In contrast, the PD under an OSI voltage
has the greater average magnitude, even for a void with
for a void with a volume
of 1.5 mm3; it is greater than 30 pC. The magnitudes of the PDs under an SI
a volume of 1.5 mm3 ; it is greater than 30 pC. The magnitudes of the PDs under an SI voltage were
voltage were between those under AC and OSI voltages for most cases, except for the oblate void
between those under AC and
OSI voltages for most cases, except for the oblate void with a volume
with a volume
of 340 mm3, of which the PD magnitude is greater than that of the OSI voltage. In
3
of 340 mm , of which the PD magnitude is greater than that of the OSI voltage. In general, the PD
general, the PD intensity shows an obvious increase by two orders of magnitude as the void volume
intensity shows an obvious increase
by two orders of magnitude as the void volume increases from 1.5
increases from
1.5 to 340 mm3.
to 340 mm3 .
It is noted that a tiny void in an actual insulator is not analogous to the electrode spacing in real
It is noted that a tiny void in an actual insulator is not analogous to the electrode spacing in real
equipment. The signal induced in the coupling circuit is immeasurably low and thus obscured by
equipment. The signal induced in the coupling circuit is immeasurably low and thus obscured by
noise. In reality, a detectable void in real scale equipment is larger than expected.
noise. In reality, a detectable void in real scale equipment is larger than expected.
Energies 2016, 9, 623
Energies 2016, 9, 623
11 of 14
11 of 14
Energies 2016, 9, 623
11 of 14
(a) x = 0.5
(b) x = 2
Figure 11.
11. Average
Average(a)
PDx magnitude
versus the
the volume
volume of
of the
the void
void in
in the
the(b)
insulator.
The OSI,
OSI, SI,
SI, and
and
Figure
PD
versus
insulator.
The
=magnitude
0.5
x=2
50
Hz
AC
voltages
applied
to
the
samples
are
equal
to
1.6
times
the
PDIV.
50 Hz AC voltages applied to the samples are equal to 1.6 times the PDIV.
Figure 11. Average PD magnitude versus the volume of the void in the insulator. The OSI, SI, and
50 Hz AC voltages applied to the samples are equal to 1.6 times the PDIV.
With respect to the protrusion defect in a uniform field, its corona inception and PD charge are
With respect to the protrusion defect in a uniform field, its corona inception and PD charge are
sensitive to the protrusion length [19]. With this consideration, the influence of the protrusion scale
With
respect
to the protrusion
a uniform
field, itsthe
corona
inception
and
PD chargescale
are on
sensitive
to the
protrusion
length [19].defect
Withinthis
consideration,
influence
of the
protrusion
onsensitive
the PD behaviors
is investigated
byWith
changing
the length of
the
protrusion
fixed
to onescale
of the
to
the
protrusion
length
[19].
this
consideration,
the
influence
of
the
protrusion
the PD behaviors is investigated by changing the length of the protrusion fixed to one of the parallel
parallel
plates
from
1
to
10
mm.
The
radius
of
the
protrusion
was
1.4
mm,
and
this
value
merely
on the
PD1 behaviors
investigated
thewas
length
the and
protrusion
fixed
to oneaffects
of the the
plates
from
to 10 mm.isThe
radius ofby
thechanging
protrusion
1.4of
mm,
this value
merely
affects
the plates
time lag
for1PD
inception
and
has aoflimited
effect onwas
PD1.4
propagation.
Thus,
themerely
shape of
parallel
from
to
10
mm.
The
radius
the
protrusion
mm,
and
this
value
time lag for PD inception and has a limited effect on PD propagation. Thus, the shape of the protrusion
theaffects
protrusion
tip
does
not
have
an
effect
on
the
PD
charge.
the time lag for PD inception and has a limited effect on PD propagation. Thus, the shape of
tip does not have an effect on the PD charge.
12 shows
thenot
average
PD
magnitudes
a function of the protrusion length under OSI,
theFigure
protrusion
tip does
have an
effect
on the PDascharge.
Figure 12 shows the average PD magnitudes as a function of the protrusion length under OSI,
SI, and Figure
AC voltages.
Asthe
expected,
intense PDs
for of
longer
protrusions
stressed
either
12 shows
average more
PD magnitudes
as aoccur
function
the protrusion
length
underbyOSI,
SI, and AC voltages. As expected, more intense PDs occur for longer protrusions stressed by either
positive
negative
An obvious
polarity
effect for
is observed
when comparing
the
results
SI, andor
AC
voltages.impulses.
As expected,
more intense
PDs occur
longer protrusions
stressed by
either
positive or negative impulses. An obvious polarity effect is observed when comparing the results
positive
or negative
impulses.
An obvious
polarity
is observed
when
comparing
theSIresults
under
positive
and negative
impulse
voltages.
The PDeffect
magnitudes
under
positive
OSI and
voltages
under
positive
and
negative
impulse
voltages.
The
PDmagnitudes
magnitudesunder
underpositive
positive
OSI
and
SI voltages
under
positive
and
negative
impulse
voltages.
The
PD
OSI
and
SI
voltages
are similar and slightly greater than those under an AC voltage (see Figure 12a). The significant
areare
similar
and
slightly
greater
than
under
an AC
AC voltage
voltage(see
(seeFigure
Figure 12a).The
The significant
similar
and
slightly
greater
than those
those
under
an
differences
between
the PD
magnitudes
under
impulses
and an AC
voltage12a).
could be significant
found at the
differences
between
the
PD
magnitudes
under
impulses
and
an
AC
voltage
could
be
found
at the
differences
between
the
PD
magnitudes
under
impulses
and
an
AC
voltage
could
be
found
negative point of the protrusion, as shown in Figure 12b. Moreover, the OSI voltage hasata the
better
negative
point
of
the
protrusion,
as
shown
in
Figure
12b.
Moreover,
the
OSI
voltage
has
a
better
negative point
of the protrusion,
as shown
Figure
12b. Moreover,
the from
OSI voltage
has a better
performance
regarding
the excitation
of PDs in
with
detectable
magnitudes
small protrusions.
performance
regarding
magnitudesfrom
fromsmall
small
protrusions.
performance
regardingthe
theexcitation
excitationof
ofPDs
PDs with
with detectable
detectable magnitudes
protrusions.
Positivepoint
point
(a)(a)Positive
(b)
Negative
point
(b)
Negative
point
Figure
AveragePD
PDmagnitude
magnitude versus
versus the
the protrusion
SI,SI,
and
50 50
HzHz
ACAC
voltages
Figure
12.12.
Average
protrusionlength.
length.The
TheOSI,
OSI,
and
voltages
Figure
12.
Average
PD magnitude
versus the
protrusion
length.
The
OSI,
SI, and
50 Hz
AC voltages
applied
to
the
samples
are
equal
to
0.7
times
the
BDV.
The
radius
of
the
protrusion
is
0.8
mm.
applied to
to the
the samples
samples are
are equal
equal to
to 0.7
0.7 times
times the
the BDV.
BDV.The
Theradius
radiusof
ofthe
theprotrusion
protrusionisis0.8
0.8mm.
mm.
applied
general, in the presence of a metal particle on an insulator surface, PDs are initiated from the
In In
general,
in the presence of a metal particle on an insulator surface, PDs are initiated from the
In general,
of a metal
particlewhere
on anthe
insulator
surface,
PDs are
initiated
from the
triple
junctionin
ofthe
the presence
gas, insulator,
and particle
local field
is enhanced
and
the initiatory
triple junction of the gas, insulator, and particle where the local field is enhanced and the initiatory
triple
junction
of theThe
gas,
particle
the local
field is on
enhanced
and the
initiatory
electrons
emerge.
PDinsulator,
activity isand
partly
relatedwhere
to the field
distribution
the insulator
surface.
In
electrons emerge. The PD activity is partly related to the field distribution on the insulator surface. In
electrons
emerge.
PD with
activity
is partly
related
the field as
distribution
insulator
this study,
metal The
needles
different
lengths
weretoemployed
the particleon
andthe
were
fixed onsurface.
the
this study, metal needles with different lengths were employed as the particle and were fixed on the
insulator
contactwith
withdifferent
the platelengths
electrode.
Figure
13 shows
theparticle
dependence
of thefixed
average
In this
study,surface
metal in
needles
were
employed
as the
and were
on the
insulator
surfaceon
in the
contact
with the
plateAselectrode.
Figure 13 shows
the dependence
of observed
the average
PD
magnitude
protrusion
length.
with
the
protrusion,
the
polarity
effect
was
also
insulator surface in contact with the plate electrode. Figure 13 shows the dependence of the average
PD magnitude
on the
protrusion
length. As with
the protrusion,
the polarity
effect was also
observed
the particle
the
insulator surface.
PD magnitudes
for various
protrusion
under
an
PDfor
magnitude
onon
the
protrusion
length. The
As with
the protrusion,
the polarity
effectlengths
was also
observed
forOSI
the voltage
particlewere
on the
insulator
surface.
The
PD
magnitudes
for
various
protrusion
lengths
under
much greater than those under SI and AC voltages. In the test, all of the types
of an
OSI voltage were much greater than those under SI and AC voltages. In the test, all of the types of
Energies 2016, 9, 623
12 of 14
Energies
2016, 9, 623
12 of 14
for
the particle
on the insulator surface. The PD magnitudes for various protrusion lengths under
an OSI voltage were much greater than those under SI and AC voltages. In the test, all of the types of
voltages
investigated
were
incapable
exposing
protrusions
with
a length
less
than
2 mm,which
whichisis
voltages
investigated
were
incapable
ofof
exposing
protrusions
with
a length
less
than
2 mm,
stilla ahazardous
hazardous
actual
situation.
still
inin
anan
actual
situation.
(a) Particle in contact with the positive plate
(b) Particle in contact with the negative plate
Figure
Average
PD
magnitude
versus
the
length
theneedle-like
needle-likeparticle
particleononananinsulator
insulatorsurface.
surface.
Figure
13.13.Average
PD
magnitude
versus
the
length
ofofthe
The
OSI,
SI,
and
50
Hz
AC
voltages
applied
to
the
samples
are
equal
to
1.6
times
the
PDIV.
The
The OSI, SI, and 50 Hz AC voltages applied to the samples are equal to 1.6 times the PDIV. The diameter
diameter
of
the
protrusion
is
1.4
mm.
of the protrusion is 1.4 mm.
Discussion
4.4.Discussion
Once
the
discharges
occur,
initial
Laplace
electric
is distorted
byresidual
the residual
charges
Once
the
discharges
occur,
thethe
initial
Laplace
electric
fieldfield
is distorted
by the
charges
and
and
is
subject
to
a
Poisson
distribution.
The
charge
caused
by
discharges
can
accumulate
in
space,
is subject to a Poisson distribution. The charge caused by discharges can accumulate in space, resulting
a charge
that may
electric
field
in theof
vicinity
of the protrusion
tip. In
inresulting
a chargeincloud
that cloud
may reduce
thereduce
electricthe
field
in the
vicinity
the protrusion
tip. In some
somethis
cases,
this stabilization
the development
of abelow
discharge
below alevel;
detectable
cases,
stabilization
effect caneffect
retardcan
theretard
development
of a discharge
a detectable
thus,
level;
thus,
an
increased
background
electric
field
is
needed
to
initiate
high-intensity
discharges
such
an increased background electric field is needed to initiate high-intensity discharges such as a streamer
as
a
streamer
or
leader.
In
the
case
of
an
impulse
voltage
stress,
the
growth
rate
of
the
field
can
or leader. In the case of an impulse voltage stress, the growth rate of the field can be greater than thebe
greater of
than
mobility
of thus,
the space
charge; thus,effect
the stabilization
effect
cancan
be be
limited.
This
can be
mobility
thethe
space
charge;
the stabilization
can be limited.
This
used to
explain
used
to
explain
why
the
PD
magnitude
under
impulse
voltages
is
greater
than
that
under
AC
or
other
why the PD magnitude under impulse voltages is greater than that under AC or other stable voltages.
stable
Thevoltages.
interaction between the electric field and the discharge is simply diagrammed in Figure 14.
The interaction
betweenregion
the electric
and the discharge
simply
diagrammed
in Figure
14.
V cr is defined
as the boundary
wherefield
the background
electricisfield
Eb exceeds
the critical
value
V
cr is defined as the boundary region where the background electric field Eb exceeds the critical value
Ecr for effective ionization (for SF6 , (E/p)cr ¨p = 88.5 kV/cm); vc is defined as the boundary of the space
Ecr fordiffusion
effective ionization
(for
SF6, (E/p)crtime
·p = 88.5
kV/cm);
vc in
is defined
the(tboundary
of the space
charge
region. The
dissipation
of the
charge
the gas as
zone
sw ) depends on the
charge
diffusion
region.
The
dissipation
time
of
the
charge
in
the
gas
zone
(t
sw) depends on the electric
electric field distribution in front of the discharge channel [10]. If the applied background electric field
field distribution in front of the discharge channel [10]. If the applied background electric field slowly
slowly increases, the charge diffusion region vc moves in front of the critical field region V cr , and the
increases, the charge diffusion region vc moves in front of the critical field region Vcr, and the
propagation of the discharge stops before it extends to the critical length for streamer (lcr ). On the
propagation of the discharge stops before it extends to the critical length for streamer (lcr). On the
contrary, if the applied background electric field rapidly increases, vc < V cr , and an intensive discharge
contrary, if the applied background electric field rapidly increases, vc < Vcr, and an intensive discharge
can develop. For a void in an insulator, the first PD is initiated without residual charge in the void
can develop. For a void in an insulator, the first PD is initiated without residual charge in the void
when Eb is sufficiently high (Eb > Estr ), and an initiatory electron is generated. After that, a reverse
when Eb is sufficiently high (Eb > Estr), and an initiatory electron is generated. After that, a reverse
field (Er ) is developed inside the void and limits the total field strength. The subsequent PDs start once
field (Er) is developed inside the void and limits the total field strength. The subsequent PDs start
the total field strength (Eb – Er ) is recovered above a critical level by increasing the background field.
once the total field strength (Eb – Er) is recovered above a critical level by increasing the
background field.
Energies 2016, 9, 623
Energies 2016, 9, 623
13 of 14
13 of 14
Figure 14.
14. Comparison
ComparisonofofPDPD
propagation
affected
by space
the space
a varying
slowly varying
Figure
propagation
affected
by the
chargecharge
under under
a slowly
voltage
voltage
and an impulse
and
an impulse
voltage. voltage.
5. Conclusions
5.
Conclusions
PD detection
detection under
under impulse
impulse voltage
voltage conditions,
conditions, has
been investigated
investigated based
based on
on three
three artificial
artificial
PD
has been
test
models
in
terms
of
the
measurement
method,
test
object,
and
PD
behavior.
The
general
PD
test models in terms of the measurement method, test object, and PD behavior. The general PD
sequence features
features were
were briefly
briefly summarized
summarized by
by exemplifying
exemplifying the
the typical
typical PD
PD pulses
pulses of
sequence
of different
different defects
defects
under
an
OSI
voltage.
The
single
pulse,
multiple
pulses,
and
the
reverse
polarity
pulse
under an OSI voltage. The single pulse, multiple pulses, and the reverse polarity pulse with
with different
different
mechanisms were
were considered
considered to
to be
be the
the main
main components
components of
of the
the PD
PD sequences.
sequences. Although
Although the
the PDIVs
PDIVs
mechanisms
under
an
OSI
voltage
were
greater
than
those
under
an
AC
voltage
over
entire
range
of
gas
pressures
under an OSI voltage were greater than those under an AC voltage over entire range of gas pressures
investigated, the
the PD
i.e., the
the ratio
ratio of
of the
the BDV
BDV to
to the
under an
an OSI
OSI voltage
voltage
investigated,
PD excitation
excitation efficiency,
efficiency, i.e.,
the PDIV,
PDIV, under
is greater
greater than
than that
that under
under an
an AC
AC voltage
voltage for
for most
most cases,
cases, which
which means
means PD
PD detection
detection under
under an
an OSI
OSI
is
voltage
has
a
more
sufficient
margin
below
the
BDV.
It
was
also
found
that
the
OSI
voltage
has
good
voltage has a more sufficient margin below the BDV. It was also found that the OSI voltage has good
performance for
defect
with
more
detectable
magnitudes.
In
performance
for the
the excitation
excitationof
ofPDs
PDsfrom
froma asmall-scale
small-scale
defect
with
more
detectable
magnitudes.
contrast,
anan
AC
voltage
is is
incapable
good
In
contrast,
AC
voltage
incapableofoffinding
findingsome
somedefects
defectseven
evenwith
with hazardous
hazardous scales.
scales. The
The good
detectability
imparted
by
the
impulse
voltages
accounts
for
the
weakened
corona
stabilization
under
detectability imparted by the impulse voltages accounts for the weakened corona stabilization under
an impulse.
impulse. The
an impulse
impulse voltage
voltage stress,
the growth
growth rate
rate of
of the
the
an
The authors
authors believe
believe that,
that, in
in the
the case
case of
of an
stress, the
field
can
be
greater
than
the
mobility
of
the
space
charge;
thus,
the
stabilization
effect
can
be
limited.
field can be greater than the mobility of the space charge; thus, the stabilization effect can be limited.
Acknowledgments: The authors would like to thank the National Natural Science Foundation of China (Grant
Acknowledgments: The authors would like to thank the National Natural Science Foundation of China (Grant
No. 51507130),
China Postdoctoral
Postdoctoral Science
Science Foundation
Foundation (Grant
(Grant No.
No. 2014M560777),
2014M560777), Special
Special China
China Postdoctoral
Postdoctoral
No.
51507130), China
Science Foundation (Grant No. 2016-11-141158), Shaanxi International Cooperation and Exchanges Foundation
(Grant
(Grant No.
No. 2016KW-072)
2016KW-072) and
and Fundamental
Fundamental Research
Research Funds
Funds for
for the
the Central
Central Universities.
Universities.
Author
conceived and
and designed
performed the
Author Contributions:
Contributions: M.R.
MR and
and M.D.
MD conceived
designed the
the experiments;
experiments;C.Z.
CZ and
and J.Z.
JZ performed
the
experiments and relevant simulations; M.R. and C.Z. analyzed the data; M.R. wrote the paper.
experiments and relevant simulations; MR and CZ analyzed the data; MR wrote the paper.
Conflicts of Interest: The authors declare no conflict of interest.
Conflicts of Interest: The authors declare no conflict of interest.
References
References
1.
1.
2.
CIGRE Joint Task Force. Partial Discharge Detection System for GIS: Sensitivity Verification for the UHF
CIGRE Joint
Task
Force. Partial
Discharge
Detection
System for GIS: Sensitivity Verification for the UHF
Method
and the
Acoustic
Method.
Electra 1999,
183, 75–87.
Method and the Acoustic Method. Electra 1999, 183, 75–87.
Takahashi, T.; Yamada, T.; Hayakawa, N.; Okubo, H. Space charge behavior in SF6 gas and sequential
generation of PD pulses. IEEE Trans. Dielectr. Electr. Insul. 2000, 7, 141–145.
Energies 2016, 9, 623
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
14 of 14
Takahashi, T.; Yamada, T.; Hayakawa, N.; Okubo, H. Space charge behavior in SF6 gas and sequential
generation of PD pulses. IEEE Trans. Dielectr. Electr. Insul. 2000, 7, 141–145. [CrossRef]
Hinterholzer, T.; Boeck, W. Breakdown in SF6 influenced by corona-stabilization. Electr. Insul. Dielectr. Phenom.
2000, 1, 413–416.
High-Voltage Test Techniques-Part 3: Definitions and Requirements for On-site Testing; International Electrotechnical
Commission (IEC): Geneva, Switzerland, 2006.
Meijer, S.; Gulski, E.; Smit, J.J.; Kanters, A.J.L. Risk assessment of fixed defects in GIS under different voltage
wave shapes. In Proceedings of 18th International Conference and Exhibition on Electricity Distribution,
Turin, Italy, 6–9 June 2005.
Okabe, S.; Yuasa, S.; Kaneko, S.; Ueta, G. Evaluation of breakdown characteristics of gas insulated switchgears
for non-standard lightning impulse waveforms method-method for converting non-standard lightning
impulses waveforms into standard impulse waveforms. IEEE Trans. Dielectr. Electr. Insul. 2009, 16, 42–51.
[CrossRef]
Anis, H.; Srivastava, K.D. Breakdown of rod-plane gaps in SF6 under positive switching impulses. IEEE Trans.
Power App. Syst. 1982, 101, 537–546. [CrossRef]
Niemeyer, L.; Ullrich, L.; Wiegart, N. The mechanism of leader breakdown in electronegative gases.
IEEE Trans. Electr. Insul. 1989, 24, 309–324. [CrossRef]
Hayakawa, N.; Yoshitake, Y.; Koshino, N.; Ueda, T.; Okubo, H. Impulse partial discharge characteristics and
their mechanisms under non-uniform electric field in N2 /SF6 gas mixture. IEEE Trans. Dielectr. Electr. Insul.
2005, 12, 1035–1042. [CrossRef]
Wiegart, N.; Niemeyer, L.; Pinnekamp, F.; Boeck, W.; Kindersberger, J.; Morrow, R.; Zaengl, W.; Zwicky, M.;
Gallimberti, I.; Boggs, S.A. Inhomogeneous field breakdown in GIS-the prediction of breakdown probabilities
and voltages. III. Discharge development in SF6 and computer model of breakdown. IEEE Trans. Power Deliv.
1988, 3, 939–946. [CrossRef]
Niemeyer, L. A generalized approach to partial discharge in modeling. IEEE Trans. Dielectr. Electr. Insul.
1995, 2, 510–528. [CrossRef]
Ren, M.; Dong, M.; Xiao, Z.; Qiu, A. Partial discharge pulses at artificial protrusions in SF6 under negative
IEC 60060-3 standard impulses. IEEE Trans. Dielectr. Electr. Insul. 2014, 21, 273–283. [CrossRef]
Ren, M.; Dong, M.; Xiao, Z.; Qiu, A. Partial discharges in SF6 filled void under standard aperidic and
oscillating switching impulses. IEEE Trans. Dielectr. Electr. Insul. 2014, 21, 262–272. [CrossRef]
Densley, R.J.; Salvage, B. Partial Discharges in Gaseous Cavities in Solid Dielectrics under Impulse Voltage
Conditions. IEEE Trans. Electr. Insul. 1971, EI-6, 54–62. [CrossRef]
CIGRE Working Group. Insulation coordination of GIS: Return of experience, on site tests and diagnostic
techniques. Electra 1998, 176, 67–95.
Haller, R.; Hauschild, W.; Mosch, W. High-voltage field testing of GIS from a physical view-point.
In Proceedings of the Second International Conference on Properties and Applications of Properties and
Applications of Dielectric Materials, Beijing, China, 12–16 September 1988.
Kornhuber, S.; Boltze, M.; Haller, R.; Mraz, P.; Pihera, J. PD behavior of basic test arrangements under
different measurement conditions. In Proceedings of the International Conference on Condition Monitoring
and Diagnosis (CMD), Bali, Indonesia, 23–27 September 2012.
Crichton, G.C.; Karlsson, P.W.; Pedersen, A. Partial Discharges in Ellipsoidal and Spheroidal Voids.
IEEE Trans. Electr. Insul. 1989, 24, 335–342. [CrossRef]
Seeger, M.; Niemeyer, L.; Bujotzek, M. Partial discharges and breakdown at protrusions in uniform
background fields in SF6 . J. Phys. D: Appl. Phys. 2008, 41, 185204. [CrossRef]
© 2016 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/).