Charge accumulation effects on time transition of partial discharge activity at GIS spacer defects

Abstract

The partial discharge (PD) measurements are considered as the most important tool for condition monitoring of Gas Insulated Switchgears (GISs). However, if spacer surface is involved in PD activity, charge accumulation process greatly affects the time transition of PD characteristics. This paper investigates this effect considering the defect types of delamination at the electrode/spacer interface and metallic particles adhered to the spacer surface as the most serious defects that can lead to major failure. Two different electrode configurations were built to simulate the two defect types. The PD characteristics were measured and analyzed from immediately after voltage application up to several minutes where steady state characteristics have been obtained. As a result, it was found that the generation rate of PD pulses changed considerably with time for both defect types. This tendency of PD characteristics has been discussed considering the effect of charge accumulation. Comparing these effects for negative and positive PD enabled to clarify the similarities and differences in the charge accumulation effect for both defect types which can contribute to the defect type identification in GIS. It was found that the effects of charge accumulation on the electric field strength for the delamination defect case as well as on the ionization volume for the metallic particle case were the responsible parameters for the different PD tendencies.

Full text

Page 1

Charge Accumulation Effects onTime Transition of Partial
Discharge Activity atGIS Spacer Defects
Diaa-Eldin A. Mansour, Kanako Nishizawa
Department of Electrical Engineering and Computer Science
Nagoya University
Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan
Hiroki Kojima, Naoki Hayakawa, Fumihiro Endo
EcoTopia Science Institute
Nagoya University
Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan
and Hitoshi Okubo
Department of Electrical Engineering and Computer Science
Nagoya University
Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan
ABSTRACT
The partial discharge (PD) measurements are considered as the most important tool for
condition monitoring of Gas Insulated Switchgears (GISs). However, if spacer surface is
involved in PD activity, charge accumulation process greatly affects the time transition
of PD characteristics. This paper investigates this effect considering the defect types of
delamination at the electrode/spacer interface and metallic particles adhered to the
spacer surface as the most serious defects that can lead to major failure. Two different
electrode configurations were built to simulate the two defect types. The PD
characteristics were measured and analyzed from immediately after voltage application
up to several minutes where steady state characteristics have been obtained. As a result,
it was found that the generation rate of PD pulses changed considerably with time for
both defect types. This tendency of PD characteristics has been discussed considering
the effect of charge accumulation. Comparing these effects for negative and positive PD
enabled to clarify the similarities and differences in the charge accumulation effect for
both defect types which can contribute to the defect type identification in GIS. It was
found that the effects of charge accumulation on the electric field strength for the
delamination defect case as well as on the ionization volume for the metallic particle
case were the responsible parameters for the different PD tendencies.
Index Terms — Partial discharges, charge accumulation, spacer, gas insulated
switchgear, delamination gap, metallic particle.
1 INTRODUCTION
INTEREST in insulation condition monitoring and
diagnosis in gas insulated switchgears (GISs) aims principally
to achieve high reliability of such equipment [1]. Partial
discharge (PD) measurement is usually used for this purpose
[2]. In particular, PD measurements have been investigated for
identification of defect type and size in GIS [3, 4]. Generally,
PD inception and propagation characteristics are changed by
the local electric field at defects. If a dielectric surface is
involved in the PD activity, changes in PD characteristics may
occur even over a short time scale as a result of charge
accumulation on the dielectric surface from previous PD
events [5]. In GIS, this phenomenon occurs if the defect type is
related to a solid insulating spacer. It is expected that
understanding the effect of charge accumulation on PD activity
will be helpful in PD diagnosis of an actual GIS.
Different defect types that can be related to solid spacers in
GIS are voids, delamination at the interface, and particles on
surfaces [6]. The role of charge accumulation in PD activity
initiated at voids has been extensively investigated by many
researchers [7, 8]. However, less attention has been paid for
the other two types especially for a delamination defect due to
the experimental difficulties in building micro gap structure
Manuscript received on 23 June 2009, in final form 5 September 2009.

Page 2

required to simulate actual delamination gap lengths in a GIS
spacer which are usually below 100 m.
From this point of view, in this paper, the effect of charge
accumulation on PD activity for electrode/epoxy delamination
defect is explored and compared to that for metallic particles
adhered to the spacer surface. This effect is observed from the
time transition of PD generation rate over 10 minutes from just
after voltage application. To simulate the two defects, different
electrode configurations were built and set in a GIS model.
The tendency in the generation rate for positive and negative
PD is investigated for both defects to enable extracting the
certain features of partial discharges that can be affected by
accumulated charges. Finally, to clarify the
accumulation process, surface potential measurements are
carried out after voltage application and then, the modification
in local electric field is analyzed and discussed for both defect
types.
charge
2 EXPERIMENTAL ARRANGEMENT
2.1 ELECTRODE CONFIGURATION
Two electrode configurations were built to simulate the
different defect types related to solid insulating spacers as
shown in Figures 1 and 2. Both electrode setups were installed
in a pressurized chamber filled with SF6 gas where
experiments were conducted.
Figure 1 shows the electrode system setup to simulate the
delamination defect at the electrode/epoxy interface in SF6 gas.
The electrode setup consisted of molded type high voltage
electrode and grounded plane electrode. The high voltage
electrode was molded to prevent discharges at the electrode
edge. The high voltage electrode diameter is 60 mm and the
molding insulation diameter is
permittivity of 3.7. The grounded plane electrode was made of
SUS304 with average surface roughness of 7.4 m. An
alumina filled rectangular epoxy plate (100 mm × 100 mm × 5
mm thickness) with relative permittivity of 6.0
sandwiched between high voltage and grounded electrodes. A
stack of thin dielectric films with each 25 m thickness was
used to adjust the gap between the epoxy plate and grounded
electrode. This made it possible to change the gap length from
direct contact to very small gap length, 50 m, in order to
simulate delamination at the electrode/epoxy interface in a GIS
spacer. The dielectric film was set away from the discharge
area to avoid its interaction with PDs. The gap was kept open
to be filled with the same gas and pressure inside the chamber.
Actually, when a delamination suddenly occurs, the gas
pressure in the delamination gap could be below or equal to
atmospheric pressure and then the SF6 gas would permeates
gradually from the GIS tank into the delamination gap. For this
reason, in this study, SF6 gas at 0.1 MPa is considered for the
time transition of partial discharge activity at the initial stage
of delamination occurrence.
For metallic particles adhered to the spacer surface,
electrode system setup is shown in Figure 2 with enlarged
shape of the tip of a metallic particle. An epoxy plate filled
90 mm withrelative
was
with alumina was placed in the parallel-plane electrodes of a
60 mm gap length. The relative permittivity is 6.0. A square-
cut metallic particle was fixed at the triple junction between
the epoxy plate, the grounded electrode, and SF6 gas. The
particle material was made of aluminum. Different particle
sizes were used with diameters of 0.25 mm and 0.45 mm, and
length L 3 mm. The pressure of SF6 gas was 0.4 MPa in this
case. For both electrode setups, ac high voltage with 60 Hz
was applied.
Guard epoxy
insulation
Epoxy plate
Grounded plane
electrode
Dielectric film
(Gap adjusting)
Small gap
(50 m)
 60 mm
 90 mm
To PD detecting circuit
Figure 1. Electrode system setup to simulate delamination at the
electrode/epoxy interface in SF6 gas (Delamination defect).
60 mm
Particle
Epoxy plate
(50  50  5 mm thickness )
Diameter: 0.25, 0.45 mm
Length: 3 mm
ac HV
SF6gas
Particle
tip
To PD detecting
circuit
Figure 2. Electrode system setup for metallic particles adhered to the spacer
surface in SF6 gas (Metallic particle defect).
Shielding room
Gas tank
ac HV
Detecting
circuit
(50 
SF6gas
Electrode
setup
Oscilloscope
(20GS/s, 4GHz)
Time
PD current
PD-CPWA
Computer
Figure 3. PD measurement system.

Page 3

2.2 PARTIAL DISCHARGE MEASUREMENT SYSTEM
After the electrode setup was installed in a pressurized
chamber filled with SF6 gas, the PD measurement system was
established as shown in Figure 3. PD current pulses were
detected with 50 Ω resistor. The detected PD pulses were fed
into a large bandwidth digital oscilloscope (20 GS/s, 4 GHz).
All PD current pulses were then analyzed sequentially by PD-
Current Pulse Waveform Analysis (PD-CPWA) developed in
[9]. By PD-CPWA, the transition of PD current and number of
PD pulses over the time could be obtained.
2.3 VOLTAGE APPLICATION
In order to measure the PD inception voltage (PDIV), the
target ac high voltage with frequency 60 Hz has been increased
gradually until PD occurs. However, for investigating temporal
change of PD characteristics, in this paper, the target ac
voltage is applied suddenly to prevent PD occurrence and
charge accumulation during voltage increasing process. Figure
4 shows the conventional and sudden voltage application
method with typical ac waveform for sudden voltage
application.
For the delamination defect, two different voltages were
applied to change the local electric field at the delamination
gap. These applied voltages (Va) are 1.0 and 1.3 times PDIV
which is 14.5 kVrms. In case of a metallic particle defect, the
applied voltage Va was set at 120 kVrms to give the same
operating electric field of actual GIS (2.0 kVrms/mm).
Therefore, to obtain different local electric fields at the particle
tip, different particle sizes were used as mentioned in section
2.1. Table 1 shows PDIV of various particle sizes and the
corresponding ratio of applied voltage Va to PDIV. As a
particle became thick, PDIV increased.
3 TEMPORAL CHANGE OF PARTIAL
DISCHARGE CHARACTERISTICS
3.1 PARTIAL DISCHARGE ACTIVITY JUST AFTER
VOLTAGE APPLICATION
Figure 5 shows the condition of PD generation immediately
after sudden voltage application for delamination defect and
metallic particles defect. The voltage polarity is expressed in
terms of the polarity of the grounded electrode for the
delamination defect and in terms of the particle tip for the
metallic particle defect. For both defects, the PD polarity is
defined according to the PD type which is negative streamer at
positive applied voltage and positive streamer at negative
applied voltage. The voltage is applied suddenly at time t = 0.
For the delamination defect in Figure 5a, negative PD
pulses appeared constantly at all cycles, but positive PD did
not, except that an initial large positive PD pulse during
transients immediately after switching on. At about t = 0.1 s,
positive PD pulses started to appear but the number of their
PD pulses and the magnitude of PD current are much smaller
than negative ones. Positive PD pulses increased gradually and
it takes more than 2 s until it appears at most cycles. Similar
tendencies are obtained for other applied voltages.
For the metallic particle defect in Figure 5b, negative PD
pulses appeared at most cycles. At about t = 0.1 s after the
voltage application, positive PD pulses appeared at most
cycles. At about t = 0.3 s after the voltage application, positive
PD pulses appeared at nearly all cycles. It is evident that PD
activities change greatly within the first few seconds after the
voltage application. This change takes shorter duration
compared to the case of delamination defect.
3.2 TIME TRANSITION OF PARTIAL DISCHARGE
CHARACTERISTICS
Figure 6 shows the time dependence of number of PD
pulses expressed in pulses per second (pps) when different
applied voltages (1.0×PDIV and 1.3×PDIV) were applied on
the electrode setup of delamination defect. In these figures, the
number of PD pulses is obtained using measured data with
fixed time duration. The time duration of measured data is
chosen 2 s and 0.6 s for 1.0×PDIV and 1.3×PDIV,
respectively.
For the number of PD pulses at 1.0×PDIV shown in Figure
6a, the number of negative PD pulses was about 400 pps
directly after voltage application. This number increased to a
maximum value of 470 pps before reducing to a very few
number of PD pulses after about 5 minutes. The positive PD
pulses at 1.0×PDIV didn’t appear at all. For the case of
1.3×PDIV shown in Figure 6b, negative and positive PD
Target voltage
PDIV
Target voltage
(a) Conventional voltage application(b) Sudden voltage application
(c) Typical waveform of sudden voltage application
TimeTime
Voltage [a.u.]
Time [20 ms/div]
Figure 4. Voltage application method
Table 1. PDIV and its fraction of Va for various particle sizes on an epoxy
plate surface (Va = 120 kVrms)
Particle length
[mm]
Particle diameter
[mm]
PDIV
[kVrms]
Va/PDIV
0.25 mm60 2.0
3
0.45 mm100 1.2

Page 4

pulses appeared. The number of negative PD pulses increased
from about 2400 to about 3200 pps before reducing to a steady
state value 2500 pps after about 5 minutes. But the number of
positive PD pulses continued to increase drastically after
voltage application until reached a steady state value of about
2700 pps after the same mentioned time (5 minutes). As
evident from these figures, the time transition of PD generation
rate exhibits different tendencies for negative and positive PD
pulses and the higher the over voltage ratio against PDIV, the
larger the generation rate of positive PD pulses to the negative
PD pulses. Similar tendencies in time transition of PD
generation rate are obtained for higher applied voltages [10].
For the metallic particle defect, the electric field
distribution is first calculated around the tip of each size of
metallic particles as shown in Figure 7. The electric field
calculation was performed under applied voltage of 120 kVrms.
The electric field of the tip of metallic particle with a diameter
(b) Applied voltage = 1.3PDIV(a) Applied voltage = 1.0PDIV
500
400
300
200
100
0
PD generation rate [pps]
10864
Time [min]
20
Negative PD
Positive PD
4000
3000
2000
1000
0
PD generation rate [pps]
10864
Time [min]
20
Negative PD
Positive PD
Figure 6. Time transition of PD generation rate for delamination defect
(a) Delamination defect case (Va= 1.3PDIV)
(b) Metallic particle defect case ( = 0.45 mm, L = 3 mm, 1.2PDIV)
0.250.200.150.100.050.00
-100
0
100
PD current [mA]
0.500.450.400.350.300.25
Time [sec]
-100
0
100
PD current [mA]
Negative PD
Positive PD
-10
0
10
PD current [mA]
0.250.200.150.10 0.050
-10
0
10
PD current [mA]
0.500.450.400.350.300.25
Time [sec]
Figure 5. Temporal change of PD pulses just after voltage application

Page 5

of 0.25 mm and length of 3 mm is stronger than that with a
diameter of 0.45 mm and length of 3 mm.
The time dependence of PD generation rate corresponding
to different particle sizes is shown in Figure 8. The number of
PD pulses at all particle sizes increases immediately after
voltage application, then decreases and becomes almost
constant. In the case of Figure 8a, with a diameter of 0.45 mm
and length of 3 mm, the period until number of PD pulses
tends to decrease is longer than that in the case of Figure 8b,
with a diameter of 0.25 mm and length of 3 mm. Also it is
important to note that when the particle becomes thinner,
where the electric field becomes stronger, the generation rate
of positive PD pulses to the negative PD pulses increases
exhibiting similar tendency to the delamination defect case.
However, in inverse to the delamination defect case, the time
transition of PD generation rate exhibits similar tendencies for
negative and positive PD pulses. These tendencies are also
confirmed for longer particles [11].
Generally speaking for the obtained results, the negative
and positive PD pulses have approximately similar tendency
for the metallic particle case and different tendency for the
delamination case. This is considered to be resulted from the
initial electron generation mechanisms which represent with
the charge accumulation the main factors affecting PD activity,
as will be further detailed in the next sections.
4 CHARGE ACCUMULATION PROCESS
4.1 DISTRIBUTION OF ACCUMULATED SURFACE
CHARGES
Obtaining the distribution of accumulated surface charges is
the first step to analyze the effect of charge accumulation on
the local electric field. The electric charge of an epoxy
samples was removed by cleaning them with ethanol or by
using new samples. The charge free initial state has been
verified by measuring the surface potential of a sample.
For both the delamination defect and the metallic particle
defect, surface potential measurements were performed after
the epoxy surfaces were exposed to PD activity for 10 minutes
as shown in Figure 9. For the delamination defect, the charge
distribution is uniform around the surface below high voltage
electrode as shown in Figure 9a. This implies that PD occurs
all over the epoxy surface. However, for the metallic particle
defect, the charging area expands only around the particle tip
with decreasing the charge density as shown in Figure 9b. For
both defects, the accumulated charges develop with PD
activity. For the delamination defect, this is proved from the
change in PD inception electric field until a steady state
characteristics have been obtained when a balance between
charge accumulation and charge neutralization occurs [12].
For the metallic particle defect, the development of
accumulated charges is confirmed from the surface potential
measurements after 1 minute and 10 minutes of voltage
application [11].
4.2 ANALYSIS OF CHARGE ACCUMULATION EFFECT
ON THE ELECTRIC FIELD
PD activity is affected by the electric field at the PD
location, i.e. by accumulated charges on the epoxy surface.
Hence, it is important to investigate the effect of accumulated
surface charges on the electric field. The electric field at the
PD location is composed of two components. The first
component Ea is generated by the externally applied voltage.
The second component Eq is associated with the surface
charges accumulated by PD activity which generates local dc
field. To obtain the resultant electric field dependence on
accumulated charge density, a simplified calculation is carried
out based on the charge distribution obtained in Figure 9.
For the delamination defect, the applied electric field is
uniform and also the charge distribution on the epoxy surface
is uniform. Thus, the equations of uniform electric field
distribution and the equations of boundary conditions at
(a)  = 0.45 mm, L = 3 mm (1.2PDIV)
(b)  = 0.25 mm, L = 3 mm (2.0PDIV)
500
400
300
200
100
0
PD generation rate [pps]
10864
Time [min]
20
Negative PD
Positive PD
200
150
100
50
0
PD generation rate [pps]
1086420
Time [min]
Negative PD
Positive PD
Figure 8. Time transition of PD generation rate for metallic particle defect.
300
250
200
150
100
50
0
Electric field strength [kVrms/mm]
20151050
10-3 Distance from particle tip [m]
 = 0.25mm
 = 0.45mm
Figure 7. Electric field distribution around particle tip (120 kVrms)