Uploaded by mmehul

Advanced Partial Discharge Measuring System

advertisement
Advanced Partial Discharge Measuring System
for Simultaneous Testing of Cable Accessories
S. Meijer*, E, Gulski', P.D. Agoris', P.P. Seitz2, T.J.W.H. Hermans3 and L. Lamballais3
'Delft University of Technology, High-voltage Technology and Management
PO Box 5031, 2600 GA Delft, The Netherlands
2Seitz Instruments AG, CH-5443 Niederrohrdorf, Switzerland
3Prysmian Cables and Systems B.V., P.O. Box 495, 2600 AL Delft, The Netherlands
* E-mail: s.meijergtudelft.nl
signals, as is commonly used for on-site/on-line
measurements in e.g GIS [I]. The first step in making a
measurement is capturing the frequency spectrum.
Based on this spectrum, a certain centre frequency
which represents PD activity with the highest
signal-to-noise ratio is selected, see figure 1. Secondly,
a spectrum analyser can also analyse the coupler's
signals in the time domain, resulting in similar
phase-resolved PD patterns that are obtained with a
standardized measuring circuit. Such phase-resolved
PD patterns offers the possibility to recognize a certain
type of defect and to discriminate between insulation
defect and noise.
Abstract: In this paper measuring results of partial
discharges in power cable accessories are described.
For the purpose of measuring simultaneously partial
discharge activity in three or more accessories during
an acceptance test, a wireless partial discharge
detection system was developed. Therefore, each
power cable accessory is equipped with either an
internal or external sensor to decouple the partial
discharge signals. The system uses narrow-band
detection and is described in this paper. The computers
controlling the measuring units are connected via a
wireless network which allows communication
between the different units and the main computer
situated at the voltage source site. Moreover, effective
real time noise-suppression is possible by tuning the
system to a noise-free frequency band. Practical
examples of this system are shown based on several
measurements performed in the field.
50. 00
-
o 40.00
a 30.00
-
':5
INTRODUCTION
E
<J
Partial discharges (PD) may occur in power cables as a
result of insulation defects. The origin of discharging
defects can be related to poor workmanship (new or
repaired cables) or to service related insulation
degradation effects. In all these case, due to local field
enhancement till breakdown, discharge pulses are
produced at the defect site. It is known that each of
these PD pulses consists of energy frequencies up to
hundreds of MHz.
<
m
Ii
<
m
fM 4
f
-
1
-
:
t, A: )
01i T
10.00
u
0.00 _ f ~
~
0
100
F
A J
X,
4
~
J
W
m
X
Hi
;
L
L
_
-
-
+
Vl.
'
1X
9-4¢
4X@IV143
~~~~
200
300
400
500
Fre quency [M Hz]
Fig. 1.
Example of a signal-to-noise ratio frequency
spectrum as has been measured on-site. Here three
frequency ranges show additional activity which could
originate from partial discharges.
The basic parts of the VHF/UHF measuring setup are
shown in figure 2. It contains:
A sensor
1)
A 30 dB VHF/UHF pre-amplifier
2)
A coaxial cable
3)
A spectrum analyser (SA)
4)
A computer
5)
Different types of sensors can be used. In practice,
capacitive and inductive sensors are used, which can
be either internal or external sensors. Figure 2 shows
an internal inductive sensor, installed in the cable. To
have better sensitivity, the signals are amplified by a
30 dB pre-amplifier. Using a spectrum analyser, the
frequency spectrum is measured, which can contain
background noise signals and PD signals. Finally, the
Therefore, to achieve a high quality of the installation
of power cable accessories, partial discharge detection
during the on-site acceptance test is more and more
becoming an important issue. The presence of partial
discharge activity can be used to assess the condition
of the cable accessory and proper action can be taken.
To enable simultaneous monitoring of all accessories
during such a voltage test, the joints and terminations
are equipped with wireless VHF/UHF PD detectors.
VHF/UHF PARTIAL DISCHARGE DETECTION
As stated before, PD pulses consists of frequencies up
to hundreds of MHz. Consequently, a spectrum
analyser has been used to capture these high-frequency
1-4244-0189-5/06/$20.00 ©2006 IEEE.
20.00
-
687
Authorized licensed use limited to: Cornell University. Downloaded on April 16,2010 at 14:33:57 UTC from IEEE Xplore. Restrictions apply.
o
Phase 1
a) 100
-a
1u50
E
0
0
150
Phase 2
a) 100) t
-o
Fig. 2.
VHF/UHF PD detection system as installed in
the field consisting of (1) internal inductive sensor, (2) a
coaxial cable, (3) a VHF/UHF pre-amplifier, (4) spectrum
analyser and (5) PC.
0)
9u
E
50
0
data is stored on a PC, which contains software to
analyse and further interpret the measured data.
0
150
Phase 3
FIELD MEASUREMENT 1: 127kV CABLE
U) 100
Partial discharge measurements were performed during
the after-laying test of two 127 kV cable systems.
Measuring results as obtained at 130 kV (1.7UO) at the
cable terminations of one cable system are shown in
figure 4. PD measurements were performed using an
external PD sensor around the ground-wire of the cable
termination. The test voltage was applied during one
hour.
-a
0)
Cu
E
0
aL
Phase angle
Fig. 4.
Examples of phase resolved PD patterns as
obtained at the terminations of three phases of a 127 kV
cable system.
First of all, from the measuring results can be
concluded that the noise level was about 10-15 pC in
case of all three phases which corresponds quite well
to the results of the sensitivity check which was
performed before the actual measurements [2]. During
the sensitivity check, artificial pulses are injected into
the cable termination and the response is measured
using the PD detection system. An example is shown
in figure 3. From the obtained results can be concluded
that the response is different for each measuring
frequency. Moreover, the noise level that can be
reached at different measuring frequencies is different
as well. In the results shown above, the lowest
calibration impulse which can be detected is 10 pC for
most frequencies, except for 7 MHz (20 pC is still
detectable) and 20 MHz (50 pC is detectable). This
means that the transfer function for these two
frequencies is not so good and should not be used
during the measurements. Furthermore, a logarithmic
relation between the calibration impulses and the
response can be seen. Based on these relations it is
possible to estimate the actual noise level and the
actual PD level.
The presence of a insulation defect as indicated by the
PD pattern detected in phase 2 has not yet been
confirmed by visual or other inspections.
10000
1000
5;
a.
100
a)
10
en
50
FIELD MEASUREMENT 2: 150 kV CABLE
-.5 MHz
_
7
The acceptance test of two 2-phase circuits was
performed by applying a test voltage of 2.5UO=220 kV
during 10 minutes. The cable system was about 800
meters long and simultaneous measurements were
performed at the terminations at both sides. The
measuring results as were obtained at the far side (i.e.
not at the site of the voltage source) of one circuit, in
MHz
20 MHz
3
1
10
20
50
100
Calibration Impulse [pC]
200
30 MHz
50 MHz
75 MHz
500
Fig. 3.
Response of the VHF/UHF detection system to
calibration impulses at different measuring frequencies.
1-4244-0189-5/06/$20.00 ©2006 IEEE.
688
Authorized licensed use limited to: Cornell University. Downloaded on April 16,2010 at 14:33:57 UTC from IEEE Xplore. Restrictions apply.
Phase Red
CF=3 MHz
CF=123 MHz
CF=320 MHz
Phase Blue
Fig. 5.
Measuring results as obtained at one side of the cable system at 2.5 Uo.
both phases, are shown in figure 5. The following
observations can be done. First of all, the thyristor
pulses coming from the voltage source are still visible
at the far side, though strongly reduced and only at the
lower center frequencies of the SA. Secondly, at the
lowest center frequencies, corona activity in air is
visible. The patterns as were obtained at these 3 MHz
and 123 MHz for both phases look remarkably similar,
indicating system dependent effects.
Therefore, an advanced PD detection system was
developed, consisting of three VHF/UHF measuring
units as described before and as shown in figure 2 and
which were able to communicate to a main computer
via a wireless communication network. Having this
main computer next to the test engineering controlling
the voltage source improves the possibility to act if
required. Figure 6 shows the hardware of the wireless
system.
At 320 MHz, differences can be observed. No
disturbances from the thyristors can be seen and, as
expected, the corona activity is not clearly visible at
this high frequency as well. As a result, it can be
concluded that the spectrum analyzer, operating as
tunable filter, is a very effective means to reject noise
real-time. Nevertheless the patterns as measured for
both phases are different as well: in phase red a clear
PD pattern is visible, in phase blue, some activity can
be seen. However, due to the fact that the inception
and extinction voltages are around 2.5 U0, no actions
were required.
FIELD MEASUREMENT 3: 380 kV CABLE
Overview of one side of a 380 kV cable
Fig. 6.
system. The top inset shows the antenna which is used to
make the wireless communication with the main computer
and other measuring units possible. The lower inset shows
a detail of the internal inductive PD sensor.
The sensitivity check as described in [2] was
performed. Using a fast impulse generator which was
calibrated in the laboratory, artificial PD pulses were
Finally the after laying test of two 380 kV XLPE cable
circuits is described. A voltage of 1.7Uo was applied
during one hour. During this one hour of testing,
partial discharge activity had to be monitored for at the
accessories: two terminations and one joint. For that
purpose, each accessory was equipped with an internal
inductive PD sensor. Prior to the measurements, a
sensitivity check was performed to determine the
performance of each sensor and to obtain an estimation
of the relation between the detected uV-level and the
PD magnitude [2].
injected into each cable termination. The VHF/UHF
measuring unit was tuned at different center
frequencies and the response (or transfer function) was
measured in [tV's. As a result, 12 cable terminations
were investigated in this way. Based on these results,
the average transfer function and the spread in the
results were available, see figure 7.
Moreover, the measurements at each accessory had to
be performed simultaneously and of course any change
in PD magnitude or PD pattern (if present) had to be
reported immediately in order to take the required
actions. For that purpose, it would be beneficial to
have all data available at the voltage source site.
1-4244-0189-5/06/$20.00 ©2006 IEEE.
As can be concluded from figure 7, mainly two
frequency ranges show to be of interest: ranging from
60-200 MHz and 320-380 MHz. Although the
response in the first frequency range is higher
689
Authorized licensed use limited to: Cornell University. Downloaded on April 16,2010 at 14:33:57 UTC from IEEE Xplore. Restrictions apply.
0.w,^Cl~ 10At;-k.x>w 10Rtqs.l(4-@
visible, which proved the applicability of the system.
compared to the second, also the spread in the
amplitude is much higher. As a result, only the
frequency range between 160-200 MHz is of
importance and in this range, about 25 [tV represents 1
pC of PD magnitude. In the second range, a relation of
about 7 [tV per pC was found.
80.00
60.00
(D
40.00
E
20.00
<
0.00
0
200
100
300
500
400
No partial discharge activity in the accessories
detected during the test.
CONCLUSIONS
Based on the investigations described in this
contribution, the following can be concluded:
1)
Using a wireless communication network,
measuring data obtained at three different
cable accessories could simultaneously be
displayed and analyzed.
2)
Using a narrow-band VHF/UHF detection
system, a noise-level of about 10 pC was
reachable.
3)
By changing the center frequency of the
tunable filter, disturbances coming from
thyristor pulses and corona in air can be
effectively suppressed.
600
Frequency [MHz]
Fig. 7.
Average transfer function and its spread of 12
cable terminations.
ACKNOWLEDGMENT
Examples of measuring results are shown in figure 8. It
can be observed that a noise-floor of around 10 pC was
reached at all accessories. Large random disturbances
were observed, in particular at termination 2. However,
the fact that these were just single random disturbances
could easily be concluded from analysis of the
phase-resolved PD patterns. The measurements shown
in figure 7 were performed at a center frequency in the
second frequency range. During the rise of the test
voltage the measuring units were tuned at other
frequencies as well, to ensure that no incepting PD
source was missed. Especially at the lower center
frequencies, pulses originating from corona at the
terminations and originating from the thyristors were
Termination 1
70
60
2S
-4
E
30 3-
REFERENCES
[1]
o
0
Joint
70
70
60
50
-c-
c
40
Termination 2
60
-4
50
-S
--
40
0410
-
20
30
40
30
=
co
co
E
10~~~~~~~~~~~~~I0
50
1
E 2
020--
-20
1.01
0
60
10
30
20
40
50
0
60
-
X
T
X
1
co:
30
E 20
10
0 60
a50
U430
-
(D
70
60
50
)3
---
--E
E 20
30
40
50
60
L
-
.=40
30
20
Time [minutes]
70
0,
-
Time [minutes]
Time [minutes]
70
() 60
a. 50
B.F. Hampton, R.J. Meats, Diagnostic measurements
at UHF in gas insulated substations, IEE Proceedings,
Vol. 135, Pt. C, No. 2, pp. 137-144, 1988.
S. Meijer, P.D. Agoris, E. Gulski, P.P. Seitz, T.
Hermans, L. Lamballais, Simultaneous Condition
Assessment of Accessories of Power Cables using a
Wireless VHF/UHF PD Detection System, Conference
Proceedings CMD2006, Changwon, Korea, 2006.
[2]
20
0)2010
a.
E
The authors acknowledge TenneT TSO B.V. for their
support during the measurements.
40
40
were
-
20
-
I
10
0
60
120 180 240
Phase-angle
300
360
1-h
0
k
60
-
120
180
240
Phase-angle
300
360
0
60
120
180
240
Phase-angle
300
360
Fig. 8.
Measuring results simultaneously obtained at the three cable accessories during the acceptance test of one cable: the PD
magnitude versus time for one hour testing and examples of phase-resolved patterns obtained during several cycles of the test.
1-4244-0189-5/06/$20.00 C2006 IEEE.
690
Authorized licensed use limited to: Cornell University. Downloaded on April 16,2010 at 14:33:57 UTC from IEEE Xplore. Restrictions apply.
Download