Diagnostic Field Testing and Condition Assessment of Power Transformers in Service
Diagnostic Field Testing and Condition Assessment of Power Transformers in Service
ABSTRACT:
In this era of reformation, liberalization and unbundling of electricity markets, asset management
in power sector has assumed greater prominence. Power transformers are key components in any
transmission & distribution network and loss of a transformer can have an enormous impact on
reliability and availability of power supply and on cost. As society is more and more dependent
on electricity for development, the utilities are under pressure to meet the ever-growing demands
for reliable power supply. Economic factors are the main consideration and in order to minimize
capital expenditure on new equipment, it is a common policy among utilities to maximize the use
of existing networks by operating at their design capability. This can be achieved by according
importance to the maintenance practice. A survey of the literature indicates that there are more
failures of transformer due to poor maintenance, improper operation, severe weather conditions
and manufacturing and design defects than due to insulation ageing. The utilities shall have a
systematic O & M practice that includes diagnostic tests for condition assessment and health
checkup of the equipment. The objective of the condition monitoring tests is to detect the first
symptoms of incipient faults, ageing development or other problems and monitor their evolution
to enable the operator to take appropriate action to avoid major failure. The paper reviews the
results of various diagnostic tests including dielectric response methods for condition assessment
of power transformers.
INTRODUCTION
Condition monitoring of power transformers has been a continuous process and has seen many
improvement over the years. Although several diagnostic tests such as dielectric loss angle, IR/PI,
DGA, furan analysis are available, interpretation of data still appears to be a challenging task [1,
2, 3, and 4]. Interpretation of data requires care and experience. In recent times dielectric
response methods have been introduced for detection and determination of moisture content and
ageing of pressboard/paper insulation system of power transformers [5, 6, and 7]. Moisture in the
transformer affects the dielectric strength and the rate at which the insulation ages and in some
cases, there is also threat of bubble evolution above a certain temperature when the load is
suddenly increased [8]. Presently, the water content of the cellulose of a transformer in service is
determined indirectly by measuring the moisture content in the oil sample. The moisture
distributes unequally between the oil and pressboard, the greater part residing within the solid
insulation. As the water concentration in the oil is highly temperature dependent, the
measurement of moisture in oil is not a reliable indicator of dryness of the cellulose [7]. In
addition to the conventional measurement of power frequency loss angle, recent attention has
focussed on measuring various dielectric response parameters, which characterize some known
polarization phenomena. Many testing agencies in Europe have adopted these test methods and
data are being generated to study the efficacy of the test in assessing the status of the transformer
insulation system.
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
CPRI has been carrying out variety of condition monitoring tests on power transformers in
service for over 20 years. These tests include insulation resistance, dielectric loss angle,
capacitance, partial discharge (PD), winding resistance, TTR. The dielectric response methods
such as recovery voltage measurement and dielectric spectroscopy methods have also been
adopted by CPRI for comprehensive diagnosis of health status of power transformers. These new
techniques have been found to be effective in monitoring moisture dynamics in paper-oil
insulation system. These diagnostic tests become significant and valid only when they are
sensitive to the changes in the electrical properties of the insulation system. Since each parameter
can be related to certain information on the status of the insulation, it is essential to carryout
several tests to get a real feel of the insulation condition.
Mechanical stresses between conductors, leads and windings due to over currents or fault currents
mainly caused by system short circuits. Possible variations of supporting parts of the winding or
core may also cause deformation of the windings or of the cleats or leads. There can be collapse
of the windings also. The deformation distortions in the windings cause changes in the
geometrical distance of the windings, which in turn cause changes in the winding inductances and
internal capacitance.
Thermal stresses due to local overheating overload currents and leakage flux when loading above
nameplate rating or due to malfunction of the cooling system. Local hotspots, loose joints etc can
give rise to high thermal stresses. Severe partial discharge resulting in arcing is another
possibility.
Dielectric stresses due to system overvoltages, transient impulse conditions or internal resonance
within a winding. These stresses cause deterioration of physical and chemical properties of the
transformer insulation. The situation is more complicated due to the fact that dielectric failure is
often and final stage consequent to the mechanical and/or thermal stresses especially if moisture
and oil deterioration have already placed the transformer in a hazardous condition.
DIAGNOSTIC TESTS
For many years, the insulation system of power transformers has been and still is constituted by a
combination of mineral oil and cellulosic paper and pressboard. These materials are subjected to
deterioration under the operating conditions of the equipment. The deterioration of the
transformer insulation is primarily a function of the temperature and time and it is also influenced
by other factors such as moisture and oxygen content.
Following are the prominent diagnostic tests considered for condition assessment of transformers.
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
PI Insulation condition
Less than 1.0 Dangerous
1.0-1.1 Poor
1.1-1.25 Questionable
1.25-2.0 Fair
More than 2.0 Good
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
Measurement of partial discharges in power transformers for detection and localisation of internal
defects is still a subject of intense discussion. Presently, CPRI is employing High Frequency
Current Transformers (HFCTs) for sensing and measurement of partial discharges in power
transformers. The HFCTs can be incorporated either in the ground leads of the test taps of the
HV bushings or in neutral lead of the transformer. The measuring system is calibrated in
accordance with IEC-60270 by injecting the calibrating pulses at the HV terminals of the HV
bushings and sensing them at the bushing test taps and at the neutral. However external electrical
interference limits the sensitivity of measurement.
The RVM method is a diagnostic tool used for evaluation of paper-oil insulation system of the
transformers. The principle of the measurement may be explained as follows:
1. A DC voltage of 2 kV charges the insulation under test for a predetermined time of tc sec.
2. The test object is short circuited for pre-selected time of td.
3. Open the short circuit and allow the residual polarization to build up recovery voltage
across the test object.
4. Measure the maximum recovery voltage and its rise time. Under this condition the test object
discharges through its own DC resistance. These parameters are strictly related to the
polarization processes and their intensities.
5. Changing tc and td in a time range of 20msec to 10,000sec while tc/td = 2, constant, a series
of values Vr and tr are btained.
Plotting Vr as a function of tc a plot is obtained. This plot is called polarization spectrum. The
polarization spectrum is the characteristic of the insulation under test under specified conditions.
The polarization spectrum is a smooth curve exhibiting a global peak or one maximum value. The
corresponding time constant is called the dominant time constant. The dominant time constant is
a function of moisture content in the insulation. In some cases, the polarization spectrum exhibits
local maxima or characteristic inflexions which are related to polarization processes of fairly
different time constants and considerable intensity. These local maxima indicate presence of non
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
water ageing byproducts or non-uniform distribution of humidity in the bulk of the paper
insulation. The RVM method allows monitoring of ageing process and a reference measurement
taken at the time of commissioning can be helpful. Fig (2) shows polarization spectra obtained on
220kV class transformers.
The state and condition of the transformer insulation can be ascertained directly from polarization
curves. The displacement of dominant peak towards lower time constant is an indication of
degradation of the transformer insulation. A small time constant indicates high moisture content
or ageing status of the insulation. Non-uniform distribution of moisture and/or ageing byproducts
causes local maxima in the polarization spectrum.
CIGRE task force 15.01.09 proposes an improved interpretation of the RVM data [7].
Accordingly, the moisture content in the solid insulation is no longer estimated from the position
of the dominant peak in the polarization spectrum. Instead, the polarization spectrum is
examined for evidence of any subsidiary maxima away from the dominant time constant. The
dominant time constant corresponds to the oil peak and nay sub-dominant maximum would likely
stem from polarization phenomenon in the solid insulation. The moisture content is then
estimated from this corresponding time constant using the published calibration curves.
The increase in moisture content in the transformer insulation can also occur due to oil leakage or
small repair with temporal and partial discharge of oil or due to defects in the breathing
system/filter etc. As the moisture accelerates the degradation processes in the cellulosic paper,
there is a strong need for monitoring its level in the paper.
IEEE Std 62-1995 furnishes guidelines for assessment of moisture content in the solid insulation
system of transformers as detailed below [10]:
In the recent IEEE Std C57.106-2002, the permissible moisture level in the solid insulation
system of transformers is derived from values of water content in oil, assuming thermal stability
and moisture equilibrium between paper and oil [11,12].
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
Dielectric spectroscopy
Dielectric spectroscopy is another method used for insulation diagnosis of power transformers
and medium voltage power cables. The dielectric spectroscopy is nothing but the measurement of
dielectric loss, capacitance and permittivity of the insulation as a function of frequency. This
frequency domain analysis of dielectric losses and capacitance is used to detect moisture content
and ageing byproducts in the insulation. Fig.3 shows typical dielectric spectroscopy patterns
obtained on 220kV & 11 kV class transformers. The main features of the frequency response of
the paper-oil insulation system are
a) The curve exhibits a minimum loss (tan delta minimum) usually in the frequency range 1-
100Hz at room temperature. As ageing progresses or moisture content increases the tan delta
minimum tends to shift towards right.
b) The tan delta-frequency characteristic also exhibits a peak at low frequencies and low
moisture contents
c) Appearance of low frequency dispersion at low frequencies. The low frequency dispersion
refers to an increase in both permittivity and loss factor with decreasing frequency and the
phenomenon is typical for paper-oil insulation system. Most difficulties arise in
distinguishing the low frequency dispersion and the increase of loss due to DC conductivity.
As ageing progresses and moisture content increases, steepness of the curve increases
towards lower frequencies.
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
In most cases, however, a displacement of a winding after an external short circuit does not
immediately lead to a transformer failure, but there is a high risk that a mechanical damage in the
turn or coil insulation due to abrasion or crushing of the aged, brittle paper may eventually cause
an insulation breakdown at the next over-voltage stress. Therefore, simple non-intrusive offline
methods for detection of winding movement and failures are of high importance, because opening
a transformer and visual inspection is time consuming and expensive.
CASE STUDIES
Table – 1.1
Sl. Test Parameter
No.
1. Tan 0.31 – 5.54
2. Moisture in paper / pressboard (%) < 2.0 – 5.65
3. Water level in oil (ppm) 14 – 53
4. BDV of Oil (kV) 29 – 73
5. Tan of Oil at 90 C 0.0039 – 0.4
12
6. Oil resistivity ( cm) x 10 0.31 – 5.54
7. H2 (ppm) 0 – 994
8. C2H2 (ppm) ND
9. Furan content (ppb) 0 – 2162
Analysis of the data showed that Tan values form four distinct groups as shown in Table –2.
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
Table –1.2
Group Tan (%) No. of
transformers
A 1.0 7
B 1.0 < Tan 2.0 9
C 2.0 < Tan < 4.0 7
D 4.0 3
The Tan values obtained on group A & B transformers indicated low/normal dielectric losses.
The oil test results also indicated normal condition of the transformers. The group C transformers
exhibited high dielectric losses. The respective polarisation spectra also revealed higher moisture
level in the paper/pressboard insulation used therein (3.7% – 4.6%). Insulation condition of these
transformers was not healthy. It was recommended to dehydrate the transformers to extend their
life.
The group D transformers were found to be in highly deteriorated condition. The Tan values
indicated extremely high dielectric losses. The polarisation spectra revealed higher level of
moisture (4.19% – 5.65%) in the solid insulation system of the transformers. The oil test results
also confirmed the unhealthy status of the transformers. The BDV (29 kV) & resistivity were low
and moisture level (53 ppm) and dissipation factor (0.4) were high. However, the furan contents
in the oil samples were low. On the basis of this data, it was recommended to withdraw the
transformers from service immediately and subject them to overhauling and dehydration.
Table-2.1
Test scheme Tan PI Moisture content Moisture content
(%) (%) (%)
RVM Spectroscopy
HV v/s LV grounded 0.808 1.26 3.59 3.5
HV v/s LV(un- 0.788 1.21 3.27 3.4
grounded)
LV v/s HV grounded 0.812 1.56 3.18 3.5
As it can be seen, the tan values lie in the normal acceptable range for a 25 years old
transformer. The dielectric losses are low. Fig.4 shows the polarization spectra obtained on the
three insulation sections of the transformer.
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
The polarization spectra are smooth curves exhibiting global dominant peaks. From the curves it
can be inferred that the moisture is uniformly distributed in the insulation. The estimated moisture
levels corresponding to the global peaks are furnished in Table-3. These values are high for an
in-service transformer.
The dielectric spectroscopy patterns obtained on the transformer are depicted in Fig-5.
The slope of the characteristics increases towards lower frequencies and the tan delta minimum
appears at the higher frequency end. These characteristics indicate high moisture level in the
insulation. The estimated moisture levels from these characteristics were comparable to the
values obtained from the RVM test. The oil test results also indicated high moisture content
(41ppm) and low BDV (32kV). The DGA results were normal indicating normal internal
condition of the transformer. The results of furan analysis indicated healthy condition of the
Paper/Pressboard insulation of the transformer. Based on this analysis, it was recommended for
filtration and hot circulation of the oil.
These 20 years old auto transformers were not commissioned and stored in the yard. In order to
commission these auto transformers, the utility wanted to assess the effect of long storage time on
their insulation system. CPRI conducted the IR/PI, tan delta, RVM, dielectric spectroscopy, oil
analysis, DGA, furan analysis and DP test to assess health status of their insulation system. For
conducting DP test two paper samples were extracted from the transformers. The diagnostic test
data presented two categories of transformers. The first category of transformers exhibited
moisture absorption throughout the insulation structure, whereas the second category exhibited
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
non-uniform distribution of moisture in the insulation. Results of the PI, Tan and RVM tests
obtained on the representative unit of the first category are presented in Table-3.1.
Table – 3.1
Insulation Moisture
Resistance Polarizatio Tan level-
Insulation section (%)
60sec (G) n Index RVM
(%)
HV vs Tertiary
0.508 1.05 6.677 4.05
grounded
Tertiary vs HV
0.234 1.11 6.349 4.05
grounded
HV vs Tertiary
0.487 1.18 5.336 4.05
ungrounded
As it is evident in the Table-2, the IR and PI values were low and the tan delta values were high
indicating high dielectric losses in the transformer insulation. The polarisation spectra shown in
Figures-6, indicated high moisture level (4.05%) uniformly distributed throughout the structure of
transformer insulation.
The Dielectric spectroscopy patterns shown in Figures-7 indicated very high dielectric losses in
the entire transformer insulation. The steep slopes towards lower frequencies indicate high DC
conductivity of the oil. The appearance of tan delta minimum point at the high frequency end
indicates high moisture content in the solid insulation system of the transformer. The estimated
moisture levels in the three insulation sections of the transformer (4.7%, 5.0% and 4.9%) were
also quite high for an EHV class transformer.
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
The oil test results indicated very low BDV (25kV) and high moisture content (56ppm).
DGA indicated normal internal condition of the transformer. No furan content was detected.
The DP values obtained on the two paper samples of the transformer were 913 and 893. These
results indicated healthy condition of the paper insulation. On the basis of these results it was
recommended for dehydration of the transformer.
Table-3.2 presents the data obtained on the representative unit of second category of
transformers.
Table – 3.2
Insulation Moisture
Insulation section
Resistance Polarisatio Tan level-
60sec n Index (%) RVM
(G) (%)
HV vs Tertiary
7.46 2.10 2.368 < 2.0
grounded
HV vs Tertiary
20.3 2.47 0.224 < 2.0
ungrounded
Tertiary vs HV
11.5 2.90 0.314 2.63
grounded
As it can be seen the table-3, the HV winding insulation with respect to tertiary grounded exhibits
high tan delta values indicating high dielectric losses in this insulation section. The Tan delta
values obtained on the other two sections were in the normal range.
The RVM patterns obtained on the transformer are shown in Figures- 8. The spectrum in Fig.
8(a) indicates the homogeneous condition of HV winding insulation with respect to ground. The
initial section of the spectrum in Fig. 8(b) indicates increase in DC conductivity of the oil in the
barrier insulation system. The tertiary winding insulation with HV winding grounded exhibits
non-uniform distribution of moisture as depicted in Fig.8(c). The estimated average moisture
level in this part of the transformer insulation was 2.63%..
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
The Dielectric spectroscopy pattern shown in Fig-9 (a) indicates higher dielectric losses in the
Paper / Pressboard insulation of the HV winding insulation section with respect to ground
(Tertiary grounded). The estimated average moisture level in this insulation section was 2.7%.
The insulation section between HV winding and Tertiary winding shows {Fig .9(b)} low
dielectric losses. The tan delta varies from 0.377% at 1000Hz to 4.35% at 0.1Hz and the
estimated average moisture level was 1.0%. The tan delta-frequency characteristic of the tertiary
winding insulation with respect to ground (HV grounded) shown in Fig.9(c) shows tan delta
variation from 1.176% at 1000Hz to 9.33% at 0.1Hz. The estimated average moisture level was
2.6%.
The oil test results showed low BDV (30kV) and high moisture content (45ppm). The DGA
results indicated normal internal condition of the transformer. No furan content was detected in
the oil. No symptoms of paper ageing in the oil. The DP values obtained for the two paper
samples extracted from the two locations in the HV winding of the transformer were 888 and 909.
These values indicated healthy condition of the paper insulation. The mechanical strength of the
paper insulation was good.
This 18 years old transformer was operating at a partial load of 110 MVA instead of 165 MVA.
It was reported by the site engineers that any increase in the load would immediately trigger the
Buckholtz relay. A variety of diagnostic tests including the physicochemical analysis of the oil
were carried out to detect and identify the defective component in the transformer. The IR, Tan ,
RVM and PD test results are summarized in Table-4.1.
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
As it is evident in Table-4.1, the Tan of the two sections of the HV winding insulation i.e., with
respect to ground and with respect to LV winding, were extremely high. The dielectric losses in
the HV winding insulation were abnormally high. In contrast to this, the dielectric losses in the
LV winding were relatively lower. The partial discharge magnitude at the operating voltage of
220kV (line to line) was also extremely high. As the Tan values of the HV winding insulation
indicated abnormal deterioration, it was suspected that intense partial discharge activity may be
taking place in the HV winding part of the transformer.
Table – 4.1
PD magnitude
IR Tan Relative moisture
Test scheme (pC)
(M ) (%) content in paper (%)
at 220kV
HV versus LV
223 5.939 4.14
grounded
LV versus HV 106,000
175 1.501 4.09
grounded
HV versus LV 320 6.158 4.55
Fig.10 presents the polarization spectra obtained on the HV and LV windings of the transformer.
The average moisture level in the HV winding insulation with respect to ground was 4.14%. The
polarization spectrum obtained for the barrier insulation exhibited local maxima as shown in
Fig.10(b). These characteristic inflexions indicate presence of substantial levels of contamination
or degradation byproducts of insulation ageing.
The estimated moisture content in the paper was also high (4.55%). Although, the moisture level
in LV winding insulation was high, it did not exhibit heterogeneity. The DGA test results
indicated high levels of methane, ethane and ethylene gas concentrations. The high concentration
of ethylene may be attributed to severe hot spots (300C - 700C) in the transformer. These
results indicated high risk of thermal breakdown in the transformer. Therefore it was
recommended to withdraw the transformer from service immediately for thorough inspection and
overhauling.
However, two months later the site engineers reported that the transformer failed before any
decision could be taken. On physical inspection of the failed transformer, the R – phase HV
winding was found to be damaged extensively. The transformer was repaired at site by replacing
the R – phase limb with a new one. Vacuum drying was carried out before filling the transformer
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
with the oil. Results of the RVM & PI tests carried out after the repair and subsequent
dehydration are presented in Table – 4.2.
Table – 4.2
IR Relative moisture
Test scheme PI
(M ) content in paper (%)
HV versus LV
721 1.44 1.94
grounded
LV versus HV
778 1.49 1.60
grounded
HV versus LV 968 1.70 2.09
These data lie in the normal range expected for a healthy transformer in service. Subsequently,
the transformer was reinstated into service.
Conclusions:
1. It is essential to perform several condition monitoring tests, since each test can give only
limited information on the insulation condition of the power transformers.
2. The field experience gained so far, shows that Dielectric response methods can be used as
effective diagnostic tools for reliable assessment of insulation system of power transformers.
REFERENCES
1. “Dielectric diagnosis of Electrical equipment for AC applications and its effects on insulation
coordination”. - State of the art report. Report of WG. 33/15.08, CIGRE 1990.
2. E.Hitronniemi, H. Nordman et al, “Experiences of on and off-line condition monitoring of
power transformers in service”. CIGRE 1992.
3. G. Breen, “Essential requirements to maintain transformers in service”. CIGRE ’92.
4. G. Csepes, I. Kispal, R. Brooks et al “Correlation between electrical and chemical testing
techniques for assessing degradation of oil-paper insulation” CIGRE 1998.
5. “Polarisation Spectrum analysis for diagnosing of insulation systems” Information Tettex
Instruments AG, Zurich, Switzerland.
6. Bognor et al. “diagnostic tets of high voltage oil-paper insulating systems using DC
dielectrometrics” CIGRE 1990.
7. “CIGRE Task Force 15.01.09. Dielectric response methods” Electra No.202, June 2002.
8. George K. Frimong, Mark Perkins and Asim Fazlagic, “Estimation of Moisture in Celluslose
and Oil Quality of Transformer Insulation using Dielectric Response Measurements”, 2001
Doble Client Conference.
9. Dielectric Spectroscopy IDA 200 - Manual.
10. “IEEE Guide for Diagnostic Field Testing of Electric Power Apparatus – Part 1 Oil filled
Transformers and Reactors”. IEEE Std 62-1995.
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009
SESSION II-1
11. “IEEE Guide for Acceptance and Maintenance of insulating oil equipment”. IEEE Std C57
106-2002.
12. Brian Sparlings, Jacques Aubin, “Assessing water content in insulating paper of power
transformers”. Int. Conf. On Large Power Transformers- Modern trends in application,
installation, operation & maintenance, Oct. 2006, New Delhi, India.
International Conference on Condition Monitoring & Diagnostic Engineering Management of Power Station / Substation Equipment - 2009