Professional Documents
Culture Documents
HVPD Technical Document For On Line PD Testing of MV HV Cables May 2009
HVPD Technical Document For On Line PD Testing of MV HV Cables May 2009
Cable Box
Insulated Gland
Cable Earth Strap
There can be some restrictions in attaching the HFCT sensor to a suitable point in some
cases (particularly with older Paper Insulated Lead Covered cables with compound-filled
(solid insulation) cable boxes). In these instances the cable earth and the switchgear
earths are bonded together. Figure 2 below shows an example of a PILC terminations
with insulated glands where it is possible to make On-line PD Measurements (Fig 2a)
and two examples where it is not due to solidly bonded earths (Fig 2b).
TEV SENSOR
HFCT SENSOR
Insulated Glands to
between cable and
switchgear (required)
Whilst access to insulated earth straps can be an issue with some older cable
installations (as shown in Fig 2b) it should be noted that it is possible to modify the earth
connections on cable installations relatively simply and quickly to allow for on-line PD
testing to be carried out (this is particularly so with XLPE cable terminations when the
earth strap can be looped out of the termination box).
Figure 4: Attachment of 3x HFCT Sensors connected around each phase of an 11kV XLPE Cable
after the switchgear has been de-energised and made safe
Technical Discussion
Earthing through the earth strap is important for the collection of on-line PD data as the
power cables can be viewed as essentially (very big!) co-axial cables. When a PD event
occurs within the bulk of the insulation (between the conductor and the earth screen) a
PD pulse is generated on both the conductor (PD conductor current = i+) and the earth
screen PD conductor current = i-). These PD pulse signals are of the same
magnitude but have opposite polarity. As a result of this, if a split-core, High
Frequency Current Transformer (HFCT) sensor is placed around the whole cable
(containing both the conductor and the earth screen), the net current from the PD pulses
is zero. Thus, in order to measure the PD signal one of either the conductor PD current
(i+) or the earth screen PD current (i-) has to be intercepted separately. This is why the
earth straps play such an important part in PD detection, as they allow the earth (screen)
currents to be measured separately from the conductor currents. This is illustrated in
Figure 5 below.
PD conductor current = i+
PD earth current = i-
i+
i+
iOK
i(insulated gland)
iOK
i+ + i - = 0
i+
i(insulated gland)
iOK
i+
i(insulated gland)
i-
i-
No Signal
(plumbed gland)
Figure 5: Options for attaching Split-Core HFCT Sensor to Earth Strap or Core
3-Phase Belted Cable PD Testing
In the case of 6.6kV to 11kV belted (3x cores, 1x earth screen around all 3 cores)
cables (see Figure 6 below) the construction of the cables does not have each core
screened separately. Hence each phase can see the voltage of the other phase (i.e. the
insulation between phases is common). For the belted case, if there are phase/phase
PD events, then these events will not produce any signal on the earth strap. The normal
and inverted pulses lie on two (or more) conductors depending on the position of the
discharging void.
Figure 6:
Left: 11kV PILC Belted cable, Right: 33kV PILC 3-core cable
The construction of belted cables has large implication for carrying out PD testing Online. If earth straps are the only detection method available, then only phase/ground
events are detectable for belted cables, and phase/phase events are not recorded. It is
clear that for belted cables measured in this way, at a stroke, this rules out a whole class
of PD events from the measurements. In order to include the phase/phase events in the
measurements, some way of measuring the currents in each conductor must be
proposed. Whilst this is very difficult in conventional, PILC cable compound filled boxes,
for dry type, XLPE terminations, this is generally much easier, as access to the cores
below an earth point is usually possible (depending on the plant owners Safety Rules).
In such a case the best option is to get access to the individual cores of the cables by
attaching an HFCT sensor around the core of each phase. With this sensor connection
both phase-to-phase and phase-to-earth PD activity can be detected.
PD Reader - This is the analysis software which utilises the unique Event Recogniser
software module to analyse and characterise the collected pulses into cable PD,
switchgear / local PD and noise. A screenshot of the PD Reader Segment Analysis
page is shown below in Figure 7.
Figure 7: PDGold Software Event Recogniser Page showing Cable PD and Noise
case where the terminations/sealing ends are not correctly constructed, are difficult to
assess, as such occasions are uncommon. However, in general, at transmission
voltages, internal PD activity even at very low levels (e.g. a few tens of pico Coulombs)
will probably be fatal in service with the only question being the time to failure. Hence it
is essential to ensure that these installations are tested to be PD free on commissioning
and then tested further at regular intervals in the service life.
Historically, PD tests have been found to be difficult to achieve on site, mainly due to the
low levels of PD which would be significant and also due to extraneous interference and
noise. As XLPE single core cables are all PD tested in the factory before dispatch they
themselves are very unlikely to be the source of any PD activity. With XLPE systems,
the difficult components to ensure PD free operation are most commonly the joints and
terminations/sealing ends which are made up on site. These accessories are thus the
focus for the On-line PD Testing of HV cable systems as described herein.
The testing method which has been devised by HVPD is based on our accumulated
knowledge of testing HV cables and employs the use of three types of PD sensors as
described below.
In order to carry out On-line PD measurements it is necessary to achieve suitable and
safe access to the HV plant to be tested for the connection of the PD sensors. A variety
of sensors are suitable for on-line PD testing of High Voltage Cables equipment, these
are the Transient Earth Voltage (TEV), the High Frequency Current Transformer
(HFCT), and RF Antennae Sensors or Probes.
The TEV and HFCT are seen as most advantageous as they allow direct coupling onto
the equipment under test for detection of internal discharges. The RF Antennae and
probes are used in conjunction with the TEV/HFCT sensors for detecting and eliminating
external signals that can confuse measurements.
The split-core HFCT is clipped around the cable earth strap at the bottom of the sealing
end. The HFCT sensor (bandwidth >50MHz) is capable of picking-up both local
discharge in the sealing end (as per the TEV sensor) and also lower frequency PD
pulses coming from down the cable. Some photographs of portable sensor installations
at cable terminations are shown below in Figure 10.
What do the Partial Discharge Pulses look like from an HV Cable Test?
The shape of a partial discharge pulses will depend on the origin of the PD pulse. If, for
example, the PD signal is from within the termination then the PD pulse will be of high
frequency (>4MHz) and will be typically oscillatory in nature. Figure 12 shows an
example of a typical, internal PD pulse as measured by the HFCT and TEV sensors on
an HV cable termination. There is good correlation between the pulses, with the higher
frequencies being detected by the TEV sensor (green plot) and the lower frequencies by
the HFCT (red plot). The distinct shape and correlation of the pulses from the two
sensors indicates internal partial discharge activity in the HV termination.
0.1
Chan 1
Chan 2
0.2
0
-0.1
-0.2
-0.3
0
Time (uSec)
Chan 1
Chan 2
Curs 1
Curs 2
For PD sites along the length of a cable (the PD sites will typically be at cable joints
along the cable length as the cable sections themselves very rarely discharge) the
frequency and shape of the PD pulses will be significantly different to those seen at the
cable terminations. As these PD pulses are launched onto a transmission line of known
characteristics this means that the mono polar nature of the PD is largely preserved as
the PD pulse passes along the cable. Typical frequencies for these type of pulses are
from a hundred kHz (from PD sites a long way down the cable) up to 4MHz (for PD sites
nearby). An example of a PD pulse measured on an HV cable using the HFCT sensor
connected around the earth strap is given below in Figure 13. It can be noted that the
cable pulse in Figure 13 is mono polar and of a very different shape to the oscillatory
pulse from a cable sealing end as shown above in Figure 12.
Segment Waveforms
Main segment
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
-0.01
-0.02
65
70
75
Time (uSec)
80
85
Ch arg e current * dt
Where the current is measured with a calibrated High Frequency Current Transformer as
per the HVPD method, this calculation can be applied to give the charge in Coulombs
directly from a measurement of the PD pulse waveform in milliVolts. If output of the
HFCT is Vhf, and the transfer impedance of the HFCT is Zhf, then the charge of a PD
pulse measured at the cable termination (in pCs) is:-
Ch arg e
1
Z hf
hf
* dt
This is a measurement of the area under the monopolar cable PD pulse shown in Figure
11 on the previous page.
These PD pulses can occur more than once per power cycle, depending on how active
the discharging site is. Figure 14 overleaf shows the PD activity across a 50Hz power
cycle (20ms long). It can be noted that the PD pulses are larger in the first half of the
cycle compared to the second half. This can be typical if the pulses originate from
metalwork in contact with either the high voltage or earth. These pulse patterns across
the power cycle have been used for many years in laboratory testing to tell something
about the origins of the PD source. For example, if the PDs are all from a corona
discharge, then all the PDs would be the same height, and they would all start at the
peak of the 50Hz power cycle.
Measurement of the PD values such as those in Figure 14 can be made simply by
measuring the peaks of the pulses and counting the number of pulses above a
threshold. This simple method is referred to peak and count measurements and it has
been used for a number of years now to characterise PD activity in medium voltage
cable and switchgear installations which have low noise levels. However for HV cables
and HV cable sealing ends in outdoor switchyards this method is not sufficiently robust
due to the effect of noise (RF & switching noise) on the measurements, as described
overleaf.
0.06
Chan 1
0.04
0.02
0
-0.02
-0.04
-0.06
-0.08
0
10
12
Time (mSec)
14
16
18
RF Interference
RF interference is a particular problem when the PD measurements are made in outdoor
switching yards where the pylons and overhead connections act as antennae for radio
signals in the air. In these instances any PD activity tends to be mixed together with a
lot of noise pulses and thus the traditional peak and count measurements are not
sufficient.
Pulses are close to the sensor, and produce the largest amplitude signal at the
start of the pulse.
PD signals which have travelled some distance, often have a peak amplitude in
their middle
PD sites which originate in cables often have a monopolar pulse shapes.
Signals which have travelled down cables lose some of their HF components and
have longer risetimes and lower frequency content.
PDs which originate on the surfaces of insulators often exhibit slower pulses than
internal PDs.
Pulse shapes can often be fixed by the geometry of the HV plant which causes
them. Hence large outdoor sealing ends represent different circuits than small
components in confined enclosures, and thus produce different pulse types.
Direct
Reflected
PD
Measurement
Remote
Direct
Reflected
T
L
Figure 16: PD Pulse Trains as seen from the Measurement End
With reference to Figure 16, the time difference between the first two pulses (the direct
pulse and the reflected pulse) T, locates the site of the PD event. It can also be noted
from Figure 16 that the two pulses will continue to travel up and down the cable, until
they become too small to be seen above the noise level. During this time, the pulses are
reflected at exactly a Cable Return Time = L away from the previous arrival at the
measurement end. This gives rise to sets of pulses of diminishing size, each spaced at
the cable return time, L. If L is the return time of the cable length (this can be easily
measured with the HVPD LongshotTM Unit with PDMap Cable Mapping Software) then
the location of the PD event is:
The concept of operation is that if a pulse is received by the Signal HFCT connected to
the Transponder Trigger Unit which exceeds the Transponders adjustable trigger level,
then the Trigger Unit will emit a signal to the Pulse Generator Unit which outputs a large,
100V pulse (into 50 Ohms) to a Pulse-Injection HFCT which sends a large pulse back
down the cable via a pulse injection HFCT. This process essentially converts the SingleEnded location system into a Double-Ended location system.
The Test Set-Up for a Double-Ended PD Mapping Test (with the Portable Transponder)
is shown below in Figure 18.
SUBSTATION A
SUBSTATION
Earth Straps
Detection
HFCTs
Pulse Injection
HFCT
DISCHARGE TRIGGER UNIT
6
POWER
50 Ohms
OUT
TRIGGERED
ON
2
HVPD
Longshot
Unit
10
Min
IN
INPUT GAIN
INPUT
DISCHARGE TRIGGER UNIT
6
4
Max
HV Cable up to 5km
IN
PULSE OUT
TRIGGERED
ON
2
INPUT
POWER
50 Ohms
OUT
10
Min
Max
INPUT GAIN
PULSE OUT
Portable
Transponder
Without transponder
With transponder
Figure 19: PD pulses as seen by a detection HFCT with and without the used
of the transponder T = time difference between direct and reflected pulses
Insulation
Condition
Assessment
Discharge within
acceptable limits
Some Concern
Monitoring
recommended
Some Concern,
regular monitoring
recommended
Major Concern,
Locate PD and then
repair or replace
Figure A1:
Colour
Code
PD in XLPE
PD in PILC
0pC 250pC
0pC 2500pC
250pC 350pC
2500pC 5000pC
350pC- 500pC
5000pC 7000pC
> 500pC
>7000pC
Discussion
The indicative figures for PD Levels vs. Condition given in Figure A1 show that for Paper
Insulated Lead Sheathed Cables (PILC), discharge below 3000pC is acceptable (green),
between 3000pC and 10000pC is a cause for concern (amber) and anything above
10000pC (red) could be considered to be potentially on a trend to failure and the
problem should be investigated further (possibly with cable mapping testing and then
repair).
The indicative figures are much less for XLPE cables at less than 250pC (Green),
between 250 and 500pC (Amber) and above 500pC (Red - large cause for concern).
For mixed cables (a mixture of XLPE, PILC, EPR or other cable types) it is necessary to
find where the source of the discharge is (position on the cable) before making any
diagnostic decision. To achieve this it is necessary to apply Cable Mapping Technology
(refer to HVPD for further information on HVPDs On-Line Cable PD Mapping
Technology).
The respective PD Levels vs Condition for PILC and XLPE Joints and Terminations
given at the bottom of Figure A1 are slightly higher than for the cables as accessories
have more insulation in them than the cables and are generally more resistant
degradation from PD activity.
Appendix 2:
Introduction
The UK Electricity Distribution Companys Grid/Primary (132kV/33kV) substation had an
HVPDOSM-F64 Partial Discharge Permanent Monitor installed in July 2002. Fourteen
switchgear panels and their feeder cables were being continuously monitored due to
their critical location in the system, supplying power for the Manchester Commonwealth
Games complex.
On 29th April 2003 the HVPD monitor sent out a warning alarm to the Customer and
HVPD that there were excessive PD activity levels in the Queens Park/Eastlands circuit.
A site visit was made by Dr Lee Renforth and Dr Ross Mackinlay of HVPD who used the
HVPD LongshotTM PD Spot Tester. High levels of discharge in the circuit in question
were confirmed with this testing and the customer was alerted. Further to discussions
between the customer and engineers from HVPD, tests were carried out using the
PDMap software and Portable Transponder on 23rd May 2003 in order to locate the
source of the discharge activity.
From the panel where the discharges were detected, a cable tee led to two transformers
at Queens Park and Eastlands substations. Using the HVPD On-Line Cable Mapping
technology, two discharge sites were detected in the Queens Park cable at
approximately 1564m and 1681m from Stuart St substation. The discharges were
approximately 6,000pC and 9,000pC in magnitude respectively and both were on the red
phase.
The joints and the cable between them were replaced on 26th June 2003. Further PD
tests were carried out at Stuart St. on 4th July, showing that the discharge activity had
been eliminated. This Case Study briefly presents the use of the HVPD Longshot unit
and PDMap software used to locate the partial discharge sites.
Results
The mapping results are shown in Figure A3. The measured magnitudes are large at
around 6,000pC and 9,000pC peak value. It should be noted that the discharges are
attenuated by their travel along the cable due to the cables surge impedance.
PD Map of Circuit
9,000
All
8,000
All Phases PD
7,000
6,000
5,000
4,000
3,000
2,000
1,000
0
0
10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105
Location (% along cable)
The results showed two sites of PD activity on the feeder. The PD Map waveforms for
these sites are shown in Figures A4 and A5 overleaf.
The map shows locations for these positions are as follows:First Site at 58.4% from Stuart St (1564m from Stuart St)
Second Site at 62.8% from Stuart St (1681m from Stuart St)
30
Voltage (mV)
20
10
0
-10
-20
-30
10
15
20
25
Time (uSec)
30
35
40
45
50
30
Voltage (mV)
20
10
0
-10
-20
-30
10
15
20
25
Time (uSec)
30
35
40
45
50
These results were compared to the cable route map and locations corresponded to
trifurcating joints at 1575m (SJ8699/27) from Stuart St and a further trifurcating joint at
1695m from Stuart St (SJ8699/18).
The locations using this technique will result in measurements that are about 1%
accurate over the cable length. The measurements in this case are better than 1%, as
data was obtained from both ends of the circuit.
Although the measurements were made on the earth straps of the cables at Stuart St,
the largest signals were obtained on the red phase.