Radiated Electromagnetic Field Signature of Faulty

and Polluted Porcelain Insulators

Ehsan Azordegan , Behzad Kordi

David R. Swatek

Electrical & Computer Engineering Department

University of Manitoba
Winnipeg, Canada
[email protected]

Insulation Engineering & Testing Department

Manitoba Hydro
Winnipeg, Canada

Abstract Porcelain cap and pin insulators are by far the most
popular suspension insulators in high-voltage distribution
networks all around the world. Inspection and condition
monitoring of HV insulators is also a very hot research topic
because of the critical and vital role that they play in
distribution systems.
A new condition assessment method based on electromagnetic
radiations from porcelain insulators is presented. In a lab
environment, a 45 KV transformer is connected to a string of
two porcelain insulators. Electromagnetic radiations from the
insulators are captured by different receivers. A polluted
insulator and a cracked insulator were studied as a faulty
insulator on the string. Pollution is artificially added to the
surface of the insulator and another insulator was intentionally
cracked for the tests. The electromagnetic radiated signature of
a polluted insulator and a cracked insulator was captured and
analyzed and a comparison between them is also presented.

I.

Recording, processing and classifying the electric field

distribution around the insulators can be used as a low cost
and easy-to-use maintenance and inspection tool.
Partial discharges are known to be indications of the presence
of fault or pollution on insulators. In some research work [5],
using a Rogowski current sampler on the ground wire, partial
discharges are recorded and their relation to insulators
resistance is studied. In this paper, the characteristics of these
random discharges are investigated by analyzing remotely
captured electromagnetic radiations from a polluted and
cracked insulator on a 45 KV line. An electric field probe, a
parallel plate dipole and a 1-12 GHz Horn antenna were used
as receivers in the lab environment to capture the timedomain electromagnetic field radiation waveforms. The
captured electromagnetic fields are stored using a 7104A
Agilent digital oscilloscope for further analysis.

INTRODUCTION

II.

Insulators have a remarkable role in high voltage

transmission and distribution networks. Suspension Insulators
should handle both the mechanical and electrical stress on
overhead lines. When one insulator in a string of insulators
becomes faulty, it puts more voltage across other insulators in
that string, increasing the electric field stress beyond normal
operating levels [1]. This voltage produces higher electric
field stress inside insulators than what they are designed for.
Punctures within the cement, pollution and cracks on the
porcelain shell not only reduce mechanical strength of the
insulator, but also cause undesired partial discharges. Partial
discharges are localized breakdowns inside insulation
systems due to high electric stress. The characteristics of the
defects, i.e. voids, fractures and flaws, inside the insulators
can affect the statistical behavior of the breakdown
mechanism. Partial discharge activity erodes the insulating
material over time and may result in completed breakdown,
i.e. flashover [2-4]. Flashovers are threatening to anyone
standing close to the transmission line structure and also
cause damage to all other equipment in the vicinity which
increases the chance of blackouts. Studying the flashover and
corona characteristics of outdoor insulators can be very
helpful in avoiding utility equipment failure and blackouts.

978-1-4244-8286-3/10/$26.00 2010 IEEE

MEASUREMENT SETUP

Fig. 1 shows the measurement setup in a high-voltage lab that

has been used for generation of partial discharges and
recording of radiated electromagnetic fields. A 45-KV AC
transformer is used as the high voltage source and is
connected to a string of two cracked/polluted porcelain
insulators. A 7104A Agilent digital oscilloscope is used for
digitization and storage of the captured electromagnetic
radiations by the horn antenna and electric field probe. The
oscilloscope has a bandwidth of 1 GHz and a maximum
sampling rate of 4 Giga samples per second. A variac is used
to control the level of HV source. To synchronize the
captured signals with the AC source, all the recording
instruments are connected to the oscilloscope by the same
coaxial cables and they are all located approximately 2 meters
away from the insulators. The oscilloscope is set on the
external trigger mode and captures the radiations when the
amplitude of radiations exceeds a certain limit. To study the
phase dependency of these discharges, we capture the 60 Hz
signal from the HV line on the second channel of the
oscilloscope. A parallel plate dipole is used as an electric
field sensor to capture the 60 Hz signal. Characteristics of the
probes will be discussed in the next section.

449

na
Ho
rn

An
ten

HV Line

tP
Do

e
rob

45 KV Transformer

7104A
Oscilloscope

Variac

Figure 1. Test setup

III.

CHARACTERISTICS OF THE PROBES

Since the time-domain voltage waveforms recorded in our

measurements are representatives of the incident
electromagnetic fields, the time-domain characteristics of the
probes are great importance in this work. A review of such
characteristics is given below.
A. D -Dot Electric Field Probe
D-Dot sensors are high frequency, free space sensors that
can measure the time derivative of the electric displacement
field. Fig.2 shows PRODYNs ACD-30 passive D-Dot sensor
which was used in this study and has a bandwidth of 1 GHz.
The output voltage of the probe is given by [6],

V o = R Aeq

D cos ( )
t

(1)

Figure 3. Spectogram of noise signal

When terminated by a high impedance (the oscilloscope

input) the voltage at the terminals of the probe is directly
proportional to the incident electric field.
IV.

TEST RESULTS

where, R is sensors load characteristic impedance (50 ohms),

Aeq is the sensor equivalent area given by the manufacturer,

A. Source-off ( Noise Measurement)

Before starting the measurements, we turned off the HV
source and captured all the existing signals in the air. This
signal was treated as noise signal in this study. To distinguish
between RF noise in the air and partial discharge radiations,
the characteristics of this signal should be studied.

B. Parallel Plate Dipole

Parallel plate dipoles are one of the simplest sensors that
are used to record the 60 Hz electric field. This sensor consists
of two parallel conductive plates separated by 1.7 millimeters.
The ground conductor of an SMA connecter is soldered to the
center of the lower plate when the inner wire is soldered to the
upper conductor.

In order to capture a broad frequency range, we used two

different receivers. The D Dot probe captures mostly below
1 GHz radiations while the horn antenna can capture up to 12
GHz. To analyze the recorded data, a time-frequency analysis
was used. Fig.3 shows the spectrogram of the recorded noise
signal. Spectrogram of the noise signal shows very powerful
components at 885 MHz, 597 MHz and 100 MHz. They
mostly come from cell phone base stations and FM radio
stations. As it is shown in Fig.3, the horn antenna can hardly
detect the FM 100 MHz signals.

D is the Magnitude of electric displacement vector and is

the angle between the normal vector of sensor ground plane
and the incident electric field, E.

B. Polluted insulator
When one of the insulators on the string is contaminated,
receivers capture electromagnetic radiations due to partial
discharge activities. Fig.4 shows a string of two insulators that
was used in the lab. In all cases we studied, the insulator
attached to the HV line was the polluted one and the one
connected to the ground was a sound insulator.
Figure 2. ACD-30 D-Dot probe

450

Figure 6. A cracked insulator on the string

probe and channel two is connected to the PPD.

V.

DISCUSSION

Figure 4. A polluted insulator on the string

To imitate the real pollution on the insulators, we sprayed

water on the shells which were artificially polluted before each
test. Fig.5 shows one sample of the collected data on the
oscilloscope where channel one is connected to the D-Dot
probe and channel two is connected to the PPD. The red
sinusoidal waveform on channel one is generated from the
PPD data via a MATLab code.

In this study we had a limit on the oscilloscope where it

could not store more than 4 million points per one triggered
screen. For a two channel configuration, the scope can sample
up to 2 Giga samples per second. At this point there was a
trade off between the resolution and the time window. 40
Mega samples per second is the maximum sampling rate the
scope can operate at for capturing a 25 millisecond time
window. Smaller time windows can be captured with higher
resolutions.

C. Cracked Insulator
We intentionally cracked the surface of one the insulators
in order to study the effect of cracks on the porcelain shell.
Fig.6 shows the cracked insulator that was used in this test. It
should be noted that the length, depth and shape of the cracks
on porcelain shells will change the characteristics of partial
discharge radiations.

Different analyses can be done on the recorded data to

develop different reference graphs. Frequency domain tools,
time domain tools or even time frequency tools such as
wavelets can be applied to the recorded data. In this study we
calculated the discharge amplitude for each phase angle at
each cycle and then it was averaged over the total number of
cycles. These averaged data were plotted over one 60 Hz
cycle. The graph is called reference phase graph as we use it to
compare the phase dependency of partial discharge activities
for different cases [7].

Fig.7 shows one sample of the collected data on the

oscilloscope where channel one is connected to the D-Dot

Figure 7. A sample of recorded data for a cracked insulator

Figure 5. A sample of recorded data for a polluted insulator

451

VI.

Electromagnetic radiation signatures of a polluted and a

cracked insulator on a string of two insulators were studied. A
45 KV high voltage line was connected to a faulty porcelain
string and radiations were captured using a D-Dot electric
field probe and a PPD. The received signal on 7104A Agilent
scope was then saved on an external hard drive for further
analysis. Using MATLAB, the amplitude of discharge
radiation for each phase angle was calculated and averaged
over the total number of cycles. Reference phase graphs were
developed using the calculated data. Radiation signatures from
a polluted and a cracked insulator were then compared using
those graphs. It was shown that electromagnetic radiations are
stronger at the end of positive cycle for a cracked insulator and
they are stronger at the end of negative cycle for a polluted
insulator. Applying other time domain or frequency domain
tools on the recorded data can develop other reference graphs
which can be used to categorize the radiation signatures.

Figure 8. Reference phase graph for a polluted insulator

ACKNOWLEDGMENT

Figure 9. Reference phase graph for a cracked insulator

Fig.8 shows the reference phase graph for a polluted

insulator under test and Fig.9 shows the reference phase graph
for the cracked insulator under test. In both cases, partial
discharge radiations mostly happen at the end of positive and
negative half cycles. In the case of polluted insulator,
electromagnetic radiations at the end of negative cycles are
stronger than those at the end of the positive cycle. However,
when there is a cracked insulator on the string, more
electromagnetic radiations can be seen at the end of the
positive cycle comparing to the negative cycle. To compare
the amplitude of radiated discharge at the end of each half
cycle, we calculated the area under the graph for each cycle.
Table 1 shows a summary of this comparison.

The authors would like to acknowledge the financial and

technical support provided by Manitoba Hydro and the
Natural Sciences and Engineering Research Council of
Canada (NSERC) during this research.
REFERENCES
[1]

[2]

[3]
Table 1. Summary of reference phase graphs

Cracked Insulator
Positive
Cycle
Total Area
Under

Polluted Insulator

Negative
Cycle

Positive
Cycle

[4]

Negative
Cycle
[5]

17.71

8.42

10.99

24.09

110~180

280~340

100~170

265~345

[mV*Deg]
90%
Phase Range
Peak

[6]

146

314

144

CONCLUSION

[7]

315

452

G. H. Vaillancourt, J. P. Bellerive, M. St-Jean, C. Jean, New Live Line

Tester for Porcelain Insulators on High-Voltage Power Line, IEEE
Transactions on Power Delivery, Vol. 9, January 1994, pp. 208-219.
Power Systems Research Center, Evaluation of Critical Components of
Non-Ceramic Insulators (NCI) In-Service: Role of Defective Interfaces,
Final Project Report, PSERC Publications 04-32, Arizona, August
2004.
A. Rawat and R. Gorur, Microstructure based evaluation of field aged
and new porcelain suspension insulators, Dielectrics and Electrical
Insulation, IEEE Transactions on, vol. 16, no. 1, pp. 107 -115, february
2009.
B. Almgren, U. Astrom, and D. Wu, A survey of the flashover
performance of hvdc converter station insulators, Power System
Technology, 1998. Proceedings. POWERCON '98. 1998 International
Conference on, vol. 1, pp. 516-519 vol.1, Aug 1998.
Y. Cheng, C. Li, and X. Huang, Study of corona discharge pattern on
high voltage transmission lines for inspecting faulty porcelain
insulators, Power Delivery, IEEE Transactions on, vol. 23, no. 2, pp.
945-952, April 2008.
< http://www.prodyntech.com/home/page/surface type >, Prodyn
Technologies website, June 2010.
H. Edin, Partial discharges studied with vairable frequency of the
applied voltage, Ph.D. dissertation, 2001.