2016 International Conference on Condition Monitoring and Diagnosis - Xi'an - China

Detection of Bushing Insulation Defects by

Diagnostic Monitoring
L.V. Badicu'*, U. Broniecki', W. Koltunowicz', J. Suboci, M. Zenker2, A. Mrozik2
OMICRON Energy Solutions GmbH, Berlin, Germany
West Pomeranian University of Technology Szczecin
laurentiu-viorel.badicu@omicronenergy.com
The first two failure causes are directly related to the ageing
process of the insulation, which is influenced by high
temperatures, load cycles, transient over-voltages, oil leakage,
moisture ingress and cracks. The early identification of these
degradation factors can be done by continuously monitoring
the bushing's most important diagnosis factors: dissipation
factor (DF), capacitance (C) and partial discharge (PD) level
[3].

A bstract-Comparative diagnostic tests were performed in

the laboratory on oil-impregnated paper and resin-bonded paper
bushings. The bushings have undergone long-term AC tests and
insulation capacitance, dissipation factor and partial discharge
were continuously monitored. A 110 kV oil impregnated paper
bushing with artificially induced defects was tested. The effect of
insulation

contamination

and

oil

leakage

on

the

bushings'

insulation condition were investigated with on-line monitoring.

With concern to the resin bounded paper bushing, a new 52 kV
unit with design flaws was tested. After the tests, the bushing was

II.

opened and the capacitive layers were inspected individually. The

results are presented in the paper.

A.

A correlation between the monitored values and the bushing

This information would support the field engineers to set realistic
threshold limits of the monitored parameters and for asset
managers to establish appropriate condition-based maintenance
strategies.

Keywords-power transformers, bushing insulation, monitoring;

capacitance; dissipation/actor; partial discharges;
INTRODUCTION

It is known that bushings are responsible for

approximately 14% of major transformer failures regardless
the type of their insulation system - resin-bounded paper
(RBP), resin-impregnated paper (RIP) or oil-impregnated
paper (OIP) [1]. The failure rate is even higher, approx. 18%,
when the analysis is limited to transformers manufactured
after 1980 (Fig.1). Based on field experience, a bushing
manufacturer attributes oil leakages (80%), insulation
deterioration (13%) and mechanical damages (7%) as the main
root causes of failures [2].

The diagnosis of the bushing is done by analyzing the

trend of the DF and its rate of change. When the DF reaches a
pre-defined level of alarm [5], continuous operation is no
longer recommended. On the other hand, bushings with high
values of DF (above a pre-defined level) but with a stable
trend may stay in operation. Another approach derived from
practical experience recommends off-line investigation for
bushings whose trend of DF doubles the value over six months.

CT
0,29%
Tap Changer

The diagnosis of bushings based on their capacitance is

done by comparing the values from on-line monitoring with
the values from the latest off-line measurements. The
difference between them gives the capacitance variation L1C.
This quantity (L1C) has to remain within certain limits [6]
depending on the bushing voltage and number of grading
layers. If the number of the grading layers is not known, the
knowledge of the insulation dielectric strength helps when
estimating the number of the bushing layers and when
implicitly setting adequate threshold limits for capacitance
variation. According to [7], the dielectric strength should not
exceed 2 kV/mm for RBP, 3.6 kV/mm for RIP and 4.5 kV/mm
for OIP insulation.

30,

magnetic
circuit
2,63%

2,34%

Fig. 1. Failure location analysis of transformers with manufacturing period

after 1980 (342 failures, U 2: 100 kV) [I]

978-1-5090-3396-6/16/$31.00 2016 IEEE

Capacitance und dissipation/actor

Traditionally, the sum-of-currents method is used for

bushing insulation evaluation, but the changes of the system
voltage phase angle due to the daily load may lead to incorrect
bushing diagnosis. To increase the sensitivity of the
measurements, reference methods have become more and
more popular [3]. Depending on the source of the reference
signal, relative or absolute measurements can be performed.
With relative measurements, the bushing-to-bushing or dual
transformer comparison can be performed. For absolute
measurements, the reference signal is taken from a group of
voltage transformers (VTs) of the substation. The advantages
and disadvantages of absolute and relative measurements are
presented in [4].

condition can be done based on the performed measurements.

I.

TRANSFORMER BUSHING MONITORING

If continuous monitoring is not implemented, off-line C

and DF measurements can be performed in a wide frequency

64

distributed in the insulation and especially at the edges of the

grading layers. This is why even new REP bushings that show
a PD level around 100 pC may pass such a test. The
equipment used for the PD tests presented in this paper is a
synchronous three-channel system whose digital bandpass
filter (bandwidth from 9 kHz to 5 MHz) can be freely set for
measurements up to 35 MHz The PD results are visualized as
phase-resolved PD (PRPD) patterns which give the correlation
between PD pulse time occurrence and applied voltage phase.
Different types of the defects generate patterns of different
shapes. Both apparent charge magnitude and PD pulse
repetition rate need to be indicated in order to get complete
information about the defect type. An example of such a
PRPD pattern is presented in Fig. 3 - left. The PRPD pattern
presented here was obtained while performing PD
measurements on a 220 kV REP bushing affected by surface
discharge as a result of moisture ingress (Fig. 3 - right).

range to identify the presence of gaseous and solid by

products generated by different degradation mechanisms of
the bushings insulation. As an example, the results of such
measurements performed on OIP bushing with a damaged
measuring tap are shown in Fig. 2. The bushing was in
operation for three hours with the measuring tap ungrounded.
This led to a voltage potential of about 15 kV between the tap
and flange, which triggered electrical discharges resulting in
the bushing tap damage. The results of the DF measurements
performed on the defected bushing (curve 2) were compared
to the ones obtained on the healthy bushing from another
phase of the same transformer (curve 1). Curve 3 was obtained
after the defected bushing was subjected for 50 hours to the
operating voltage (tap grounded) in the lab.
10'

DF

10+-'-,..,-.....-..--...,......,...,,--.-
10
10
11
1
10'
Frequency [Hz]

Fig. 2. Off-line DF measurements in a wide frequency range on: healthy

bushing (1), ungrounded tap for 3h (2), DF measured after 50 h AC test with
the tap grounded (3) .

Fig. 3. Example of PRPD pattern of defective REP bushing.

In the defected bushing (Fig. 2, curve 4) two types of

relaxation processes were noted. At frequencies above 100 Hz,
the solid by-products increased the value of DF. The
relaxation at frequencies below 0.01 Hz is generated by the
presence of a small amount of gas.

III.

The set-up for the measurements carried out in the lab is

presented in Fig. 4. Two bushings of different insulation
systems and rated voltage were tested: a 110 kV OIP bushing
(porcelain cover removed) and a 52 kV REP bushing. A
voltage transformer of high accuracy class (0.2) was used,
which gave a highly stable reference signal for bushings DF
calculation.

Additionally, the bushing with damaged but grounded

measuring tap was left under the operation for about 50 hours
more and the results are also presented in Fig. 2. In the
frequency range between 0.01 and 1 Hz, the huge relaxation
process with time constant in the range between 0.9 and 1
second was observed [8]. This is because the electric field
forces spread the by-products within the bushings volume and
lower the resistivity between some insulation layers to 50-200
MQ or even to tens of kQs. In such case, the DF curve would
be close to curve 1. As measurements in a wide frequency
range can only be performed in off-line mode with limited
periodicity due to technical and economic reasons, continuous
on-line monitoring can be proposed but in order to detect the
DF and C changes at the power frequency at an early stage,
the sensitivity of DF and C measurements has to be high. To
ensure the reliability of the diagnosis, PD detection can be
proposed as additional complementary measurement.
B.

TEST SET-UP AND TESTS PERFORMED

Due to the high difference between their rated voltages, the

bushings under test were connected to the monitoring system
in sequence. The PD system calibration for a wide frequency
range was performed for each bushing separately before
initializing the monitoring sessions.
The bushings' behavior under relevant defect conditions
was investigated by carrying out on-line and off-line
measurements of the parameters indicated in Table I. The
results of the bushings C, DF and PD will be presented in the
paper.
TABLE!.
Bushing

Partial discharge monitoring

PD on bushings is measured based on a set-up that is

similar with what is used for factory acceptance tests [9]. In
general, OIP and RIP bushings that have passed the routine
test are PD free or the PD level does not exceed 10 pc. In case
of REP bushings, as the insulation is a laminate of resin and
paper, the bushing contains a considerable amount of air

65

SIMULATED DEFECTS AND PARAMETERS MEASURED

Tests

OIP

Oil dried and degassed

Oil
contamination
with
atmospheric air and transformer
sludge
Reduction of oil level by half

REP

Technological defects

Parameters

On-line
C, PD

Off-line
--

C, OF, PD

C, OF

by analyzing both the PD trend and the corresponding PRPD

patterns. However, by comparing the PRPDs from Stages 2
and 3, the presence of two different types of defects is obvious.
In conclusion, the oil leakage that may occur during OIP
bushings operation can be detected through continuous
monitoring of both C and PD.
Stage 2

Fig. 4.

Test set-up used for OIP and RBP bushings insulation evaluation.

IV.

RESULTS

2500 k,V
CODV

Oil impregnated paper bushing

The C and PD testing procedure was performed at the
voltage value of l.05 Un,l-Y3 (88 kV) - according to [lO]. The
PD testing was performed according to [9] and the measuring
sensitivity was around 2 pc.

A.

'S
1'1:1

30000 pF

20000 pF
I III

The porcelain cover of the 110 kV OIP bushing under test

was removed before the measurement and it was placed into a
plexi-glass enclosure filled-in with mineral oil. The enclosure
was filled with the same amount of oil that normally exists in
this type of bushing. The PD and C were monitored for this
bushing in the following stages:

Stage 3

10000pf

000 r
23 MlJ

Fig. 5.

Stage 1: Oil dried and degassed

Stage 2: Oil contamination with atmospheric air and sludge
Stage 3: Oil leakage by removing half of the oil.

.,,"""'

6 Apr

20 Apr

4 Mav

18 M.!IV

1.Jun

Test voltage and capacitance variation throughout the testing period

Stage

At Stage 1, the test voltage was gradually increased from

0.75 Un,l-Y3 to l.05 Un,l-Y3 (Fig. 5 - upper trend). The bushing
C value was constant, about 190 pF (Fig. 5 - lower trend), and
no PD signal was detected (Fig. 6) regardless of the test
voltage level.

Stage 2

Stage

l" 2.00 nC

U
1 aOne

The voltage at Stage 2 was constant and equal to 1.05

o,/1/-Y3 (88 kV) and the bushing capacitance remained stable
throughout the test. With concern to the PD measurements,
defect-specific PRPD patterns were detected, probably as a
result of discharges in the artificially introduced air bubbles
(Fig. 6 - middle PRPD). Therefore, it can be concluded that
for OIP bushings, the insulation contamination can be
evaluated based on PD analysis.

23 h4lf

Fig. 6.

In Stage 3, an oil leakage was simulated by removing half

of the oil from the bushing enclosure. The test voltage
remained constant (88 kV) for most of the stage duration. In
this case, an increase of the bushing capacitance from 190 pF
to 350 pF was recorded. The capacitance increase can be
explained based on the parallel capacitances introduced after
oil removal. The amount of paper left above the oil gets
contaminated with the moisture from the surrounding air. Thus
a higher capacitance was created having the (wet) paper as a
dielectric. The dielectric constant of the (wet) paper Cr is the
factor that primarily leads to the capacitance increase.

B.

PO trend and PRPO patterns throughout the OIP bushing testing

Resin bonded paper bushing

The C, DF and PD testing procedure was performed

starting with the voltage level of 1.05 Un,l-Y3 (32 kV) and
increased to the Um (52 kV), as indicated in [lO].
Measurements at the voltage level of l.15 Um were performed
as well (without exceeding the dry-power frequency withstand
voltage).
A 52 kV REP bushing with design flaws was investigated.
According to [10], the PD test is passed if the PD level does
not exceed the value of 300 pC while testing the bushing at
voltages between the l.05 and l.5 Um. The bushing being
presented here showed a PD level of 500 pC at Um

An increase of the PD level after reducing the bushing oil

level was noticed, as well (Fig. 6 - Stage 3). This can be seen

66

Power Factor qUmHz) I

During the laboratory tests, the test voltage took values

between l.0 Um and l.15 Um throughout the investigation. At
l.0 Um, both C and DF show stable values (Fig. 7 - upper
trends) while the PD level increases with a low slope (Fig. 7 lower trend). After increasing the test voltage to the value of
l.15 Un" a steeper increase of the DF and PD values can be
observed. The bushing C remains constant as no short circuit
of the grading layers occurred. The evolution of the PRPD
pattern characteristic to discharges in gas cavities of the
insulation is presented in Fig. 7 as well. It can be noticed that
the amplitude and their repetition rate of PD pulses
continuously increases.

3.11700
2.44 GO

1.2572

0,010&01

1.5929

0200

0010
0'"'

0020

Off-line measurements of C and DF were performed in a

wide frequency range on this RBP bushing as well (Fig. 8).
The results obtained here are very close to those obtained from
on-line monitoring at 1.0 0,/1' The voltage level applied for the
off-line tests was 100 V and therefore too low in order to
trigger the PD activity in the existing gas cavities (Fig 8). At
the next step, the bushing was opened for a detailed
investigation and several traces of PD activity were found Fig. 9.

OO:)liXI

Fig. 8. Off-line measurements of C and OF in a wide frequency range

It is commonly accepted that the capacitance is one of the

most important dielectric indicators of bushing insulation
condition. In this case however, the presence of the gas
cavities could not be identified as the capacitance value
remained constant throughout the test as no short circuit of a
grading layer occurred. Thus by measuring the PD level and
analyzing the PRPD patterns, incipient condition defects can
be identified before developing into serious failures.

Fig. 9. Evidences of PO activity in the cavities of RBP insulation

V.

CONCLUSIONS

In case of the OIP bushing, the presence of air bubbles and

oil leakage were found by PD monitoring. Typical PRPD
patterns of these defects were identified. The bushing
capacitance increased at a lowered oil level, but remained
constant when the oil was contaminated with atmospheric air
and transformer sludge.

1.15 Urn
15000kV

-: --------
4000kll

The design flaws of an RBP bushing tested at the 0,/1 were

detected through PD monitoring. At a test voltage of l.15 Um,
the presence of the defects were identified through DF
monitoring as well. The bushing capacitance value remained
constant as no short circuit of the grading layers occurred
during the tests. The presence of the simulated defects were
also detected through off-line C and DF measurements by
analyzing their values at the frequencies below 1 Hz.

IZ500pf
100 OOpf

0.15"

--------

j 050"

25.8&7 MO

0,008414

REFERENCES
025"

200nC

[I]
[2]
1"'"-"....,,,..---:---.

150ne

[3]

jU I 00 nC -

--""._---A.-'--"'-'

[4]
oooc

Z.Jul

"Jut

Jul

Jut

10 Jul

12Jul

Fig. 7. Capacitance, dissipation factor and PO variation for RBP busing

[5]
[6]

67

ClORE Technical Brochure 642 'Transformer Reliability Survey",

CLORE Working Oroup A2.37, December 2015.
L. Jonsson, H. Rudegard, "A manufacturer's view of bushing reliability,
testing and analysis" Proceedings of TechCon Asia-Pacific 2014, pp.
259-268, April 1-2, 2014, Sydney, Australia.
L.V. Badicu, W. Koltunowicz, A. Piccolo,A. McOuigan, C. Feely,
"Monitoring of a Distribution Transformer at Winchelsea Substation"
Proceedings of TechCon Asia-Pacific 2014, pp. 259-268, April 1-2,
2014, Sydney, Australia.
L.V. Badicu, U. Broniecki, W. Koltunowicz, S. KOrber, M. KrOger and
E. Voegel, "Prevention of transformer failure through continuous
monitoring", paper 274, in proceedings of the 19th International
Symposium on High Voltage Engineering (ISH) in Pilsen, 2015.
CLORE Techical Brochure 445 "Guide for Transformer Maintenance"
ClORE Working Oroup A2.34, February 2011.
T. Stirl, R. Skrzypek, S. Tenbohlen, R. Vilaithong, "Online Condition
Monitoring and Diagnosis for power Transformers their Bushings, Tap
Changer and Insulation System" Paper? Proceedings of the International

[7]

[8]

Conference on Condition Monitoring and Diagnosis CMD 2006,

Changwon, South Korea, April 2006.
H. M. Ryan, "High Voltage Engineering and Testing - 3'd Edition"

[9]

IEC 60270 (2000) "High-voltage test techniques - Partial

discharge measurements" International Electrotechnical
Commission, Publication 60270, 2000.

[10] ***IEC60137 -Insulated bushings for alternating voltages above 1000V,

Fifth edition, 2003-8.

printed by CPI Group Ltd, Croydon, UK, 2013, ISBN 978-1-84919-2637.

J. Buchacz, "Analysis of the state of the OlP bushings' insulation by
FDS method and oil tests", PhD thesis, Technical University of Opole,
2015.

68