You are on page 1of 4

2014 Annual Report Conference on Electrical Insulation and Dielectric Phenomena

The Influence of Dielectric Dissipation Factor on


Transformer Frequency Response Analysis
Mehdi Bagheri, B. T. Phung, and Trevor Blackburn
University of New South Wales, Sydney, Australia
m.bagheri@unsw.edu.au

Abstract— Frequency Response Analysis (FRA) is regarded as frequency response spectrum, the HV/LV windings frequency
a robust technique to recognize transformer winding deformation. response spectra for “wet” and “dry” type test objects for
It has been implemented and used since the last decade. However, various temperatures are measured. In the meantime, DDF
its interpretation and measurement sensitivity are still under measurement is performed for each and every stage and the
development. Recent studies reveal that apart from winding
results are recorded. Deviations in FRA resonance frequencies
deformation, FRA data can also be influenced by moisture
variation in transformer paper insulation. Technically speaking, under different DDF values are discussed. For DDF
moisture variation will certainly influence the Dielectric measurements, to have precise understanding about the
Dissipation Factor (DDF) in transformer. Therefore, it is of interest moisture migration between the oil and paper insulation, the
to investigate the correlation between DDF value and FRA moisture content of the oil is measured and paper humidity can
spectrum. A model transformer was fabricated for this study and be obtained. Indeed, for each DDF value the deviation in FRA
the experiment was concentrated on the temperature changes and resonant frequencies and moisture contents in paper and oil
paper moisture variation and eventually DDF alteration on the insulations are highlighted and discussed in detail. This study
FRA spectrum deviation. DDF was measured for the “wet” and shows that the changes in dielectric dissipation factor in turn
“dry” model transformer at different temperatures. In parallel, the
can influence the FRA resonance points in the mid- and high-
FRA spectrum was also recorded. The correlation between the
transformer DDF value and FRA spectrum is highlighted and frequency bands.
discussed.
II. DIELECTRIC DISSIPATION FACTOR (DDF)
Keywords: Frequency Response Analysis, Dielectric Dissipation
Factor, Moisture Variation, Transformer, Winding Deformation. A. Single Frequency

I. INTRODUCTION Dielectric Dissipation Factor (DDF) provides an indication of


the quality of insulation. The purpose of the measurement is to
Frequency Response Analysis (FRA) has been used as determine the capacitances and the tanδ between the windings
diagnostic method for transformer winding deformation and the earthed parts and between the different windings of the
recognition since the last decade [1]-[3]. Researchers now
transformer. Insulations between windings usually can be
recognize FRA for different abilities in transformer defect
recognition due to its good sensitivity on any changes in represented with parallel or series resistances and capacitances.
transformer active parts or even insulation system [4]-[5]. The DDF can be measured by applying a voltage across this
Transformer temperature variation and moisture migration from capacitance and measuring the ampere and watts of loss and
the paper insulation and their impacts on frequency response then calculating tanδ. The capacitance values are required while
spectrum are under investigation. Indeed, each and every planning transformer overvoltage protection and calculating the
insulation parameter seems to be able to change the FRA over-voltages affecting the transformer. The tanδ is essentially
spectrum [6]. However, this influence may be insignificant in the power factor (for small δ) and is thus representative of the
some cases. power loss and hence the insulation quality. The tanδ value is
One of these insulation parameters indicating the insulation useful for evaluation the dryness of the insulation or aging and
quality is the Dielectric Dissipation Factor (DDF). For any oil contamination. Other method to recognize the insulation
impregnated paper insulation, the DDF can be influenced by humidity in transformer is the transformer oil titration process
moisture, temperature and degree of polymerization (DP) which is known as Karl-Fischer Titration (KFT). Recognition of
variation, while for oil insulation only its contamination also paper humidity using oil humidity will be taken into account in
affects the DDF value [7]-[8]. In fact, any changes in real and the KFT.
imaginary parts of the complex permittivity will significantly
B. Sweep Frequency
impact DDF value. Having this information, the current study
has specifically focused on the influence of DDF changes on In order to facilitate the measuring of humidity in the paper
frequency response data. To this end, a model transformer with without using oil samples, three electrical methods are
concentric continuous disc type HV and LV windings was commonly considered worldwide. They are the Polarization and
fabricated and used as the test object. The test object tank was Depolarization Current (PDC), Frequency Domain
manufactured with plexiglass; hence its interior compartment Spectroscopy (FDS), and Recovery Voltage Measurement
was quite visible. In order to examine the DDF impacts on (RVM) [9]. Among all, FDS is commonly used in industry over

978-1-4799-7525-9/14/$31.00 @2014 IEEE 634


the past few years. In fact, instead of studying the polarization which generally causes the FRA spectrum oscillations to appear
process in time domain, the study of dielectric response can be more at very high frequencies.
conducted in frequency domain when an AC sinusoidal voltage To run the experiment, initially the transformer drain valve
U(ω) is applied [10]. The complex impedance and dielectric was opened and the test object was deliberately left exposed to
dissipation factor (tanδ) are measured. In this method, the the laboratory ambient for two weeks so that the paper
imaginary and real parts of the capacity of an insulation system insulation was saturated with moisture. The average readings of
are estimated over a frequency range and humidity amount ambient temperature and relative humidity were 23 ºC and 26
corresponding to this capacity is determined using diagram %, respectively. Hence, the initial moisture content for paper
comparison method. Indeed, this is in turn the tanδ insulations was 4.1 %, calculated using air relative humidity
measurement as a function of swept frequency. method. Afterwards, the glassy tank was filled with dry
transformer oil (< 5ppm, at 70 ºC) then left until the paper is
III. EXPERIMENTAL STUDY fully impregnated by oil, and eventually both of them reached
In order to study the impacts of DDF on the FRA spectrum, a equilibrium.
model transformer with concentric continuous disc type The moisture content of the oil was measured using KFT
windings was fabricated and used as the test object. The tank method at 23 ºC and moisture content of the paper insulation
was manufactured using plexiglass material. Line and neutral was derived through MIT oil-paper equilibrium curves [12],
leads of the windings were brought out from the tank through accordingly. These values were 11 ppm and 4 %, respectively.
appropriate HV and LV bushings. The test object is shown in Derived moisture contents of the paper insulation through air
Figure 1. The FRA setup was then prepared based on test relative humidity and KFT methods led to similar results (4.1 %
object HV and LV winding terminals as illustrated in Figure 2 and 4 %). Next, the test object was heated up to 30 ºC to be
(end-to-end measurement [11]). ready for the initial stage of FRA measurement. FRA
measurements were performed over the range of 30 to 90 ºC in
10 ºC increments but the spectra were recorded for HV winding
at 30, 50, 70 and 90 ºC. The frequency response trace for HV
winding was recorded over the range 20 Hz – 20 MHz while its
terminals were left open circuit (end-to-end measurement). FRA
spectra of HV windings for 30 ºC and 90 ºC (the lowest and
highest temperatures) are shown in Figure 3 over the frequency
range of 5 kHz – 20 MHz. The frequency response magnitude
over the frequency range 20 Hz – 5 kHz was constant at 0 dB
Figure 1. Manufactured glassy model transformer. and thus is not shown in Figure 3.
Discussion
According to Figure 3, when examining from low
frequencies to 800 kHz all of the traces are perfectly matched.
Moving from 800 kHz to higher frequencies, the discrepancy
between the spectra becomes obvious.
As the test object temperature changed from 30 to 90 ºC, the
local resonances in FRA spectra have slightly shifted to lower
frequencies. Detailed analysis of Figure 3 shows that the
oscillations trend of the spectra seems to be similar, while the
Figure 2. Implemented FRA measurement setup on glassy model transformer.
local resonant frequencies are shifted toward lower frequencies
as the temperature increased and equilibrium was reached. In
A. Study on “Wet” Model Test Object the meantime, some of the resonant magnitudes are reduced.
This deviation comes certainly through changes in the
The test object was placed inside the oven and wiring windings’ inductance, total capacitance, and resistance or
connections for FRA measurement were brought out through a insulation conductance.
bushing mounted on a small opening on the oven top. The oven From a mathematical point of view, local resonances and
was equipped with a sensitive thermostat and a digitally anti-resonances in FRA spectra can be generated due to the
controlled heater to govern the internal temperature. The FRA interaction between inductive and capacitive reactances.
test setup was then kept unchanged for the entire experiment. Obviously each and every local resonance (ith resonance) can
FRA measurement was performed on a winding by injecting a take its own inductance and capacitance values (Li and Ci). The
swept sinusoidal signal Vin of 20 volts (peak-peak) at the line- resonant frequencies in FRA spectrum would be changed if the
lead, and detecting the response Vout at the neutral-lead. The inductance or total capacitance is changed. Measurement results
measurement was conducted for the frequency band 20 Hz- 20 on inductance and resistance variations for low, mid and high
MHz. The reason to extend the upper frequency band limit frequencies show that these parameters are not altered
throughout measurement from the typically used 2 MHz to 20 significantly due to temperature changes and moisture
MHz is the absence of an iron core in the model transformer migration; therefore, deviation of resonant points in FRA

978-1-4799-7525-9/14/$31.00 @2014 IEEE 635


spectrum seems to be coming through changes in the total
capacitance. In fact, increasing the temperature will cause the 0.25
oil insulation showing a tendency to absorb the water. Hence,
the moisture migrates from paper into the oil insulation. 0.2
Decreasing the humidity in paper and its increase in the oil
causes the winding capacitance to change.

DDF
0.15
0
Deviated Frequency Band
-10 0.1

-20
0.05
-30
Magnitude [dB]

-40 0
20 30 40 50 60 70 80 90 100
Temperature ºC
-50
Figure 4. DDF variation spectra for HV-LV windings, “wet” model test object.
-60

-70 Frequency Band: 5 kHz - 20 MHz In fact, moving to the left hand side on Bode diagram for
-80
HV winding spectrum (at 30 °C )
HV winding spectrum (at 90 °C ) FRA data potentially determines higher value for DDF. These
10
4
10
5
10
6
10
7 resonant frequencies will take opposite side (higher frequencies)
Frequency [Hz]
when DDF takes lower values. This investigation was further
studied through the “dry” model transformer in the next
subsection.
B. Study on “Dry” Model Test Object
To study the impact of the DDF on FRA trace for dry
transformer winding, the oil was drained and the model
transformer was heated up to 90 °C in the electric oven and then
vacuumed (less than 750 mTorr) to remove the moisture content
from the paper insulation. Figure 5 shows the vacuum process.
After that, the vacuum was broken through a three-way valve
and then dry transformer oil was injected into the container once
more. The moisture content of the oil was measured after 11
Figure 3. FRA spectra for HV winding due to moisture migration from paper
into the oil insulation at 30°C and 90°C, “wet” model test object.
days through KFT method and gave a result of 3 ppm at 23 °C.
The test object was heated up to 30 °C same as before to
Study on frequency response spectra when the temperature prepare for frequency response measurement. The FRA spectra
decreases from 90 to 30 ºC led to similar results. Although for HV and LV windings were then measured at 30 and 90 °C
moisture absorption for paper insulation is different with over the frequency range of 20 Hz – 20 MHz (end-to-end
desorption, the equilibria should be the same and hence similar measurement), similar to the last experiment. Also, DDF
spectra were observed in the reverse procedure. measurement at power frequency (50 Hz) was performed as in
To have precise investigation about the DDF and its relation the last experiment. Figure 6 shows the frequency response
to FRA spectra deviation, this parameter was measured for each spectra just for 30 and 90 °C.
and every temperature on the “wet” model transformer and the
results are provided in Table I and plotted in Figure 4. Table I
also shows the moisture content in oil and paper insulations at
different temperatures for the “wet” model transformer. The
maximum applied voltage for DDF measurement was 5 kV to
avoid any kind of undesirable flash-over in the test object. On
examining DDF measurement results, it is obvious that for the
“wet” model transformer as the DDF experiences higher values,
the local resonances in FRA spectrum shift to lower
Figure 5. Vacuum process of the model transformer.
frequencies.
According to Table II, DDF values for the model transformer
Table I. DDF at 5 kV, moisture content in oil and paper insulations for “wet”
type test object. are significantly reduced due to the transformer dry-out process.
In addition as far as the temperature increases, the FRA
Temp. 30 °C 40 °C 50 °C 60 °C 70 °C 80 °C 90 °C
resonant points in Figure 6 have shifted to lower frequencies. In
DDF 0.0203 0.0319 0.0484 0.0721 0.1119 0.1858 0.2849
Paper the meantime, the DDF value has increased as provided in
3.6 3.0 2.6 2.2 2.0 1.8 1.7 Table II and Figure 7. This is the same as the previous case for
MC%
Oil
18 21 25 35 52 63 95 the “wet” model transformer. Indeed, a similar relationship
MC[ppm] between the FRA resonant points and DDF in the “wet” model

978-1-4799-7525-9/14/$31.00 @2014 IEEE 636


transformer has occurred for the “dry” one. However, the IV. CONCLUSION
deviation values are quite different in both FRA spectrum and The study presented in this article explored the relation between
DDF data. It should be also highlighted that for both Tables I DDF and FRA spectrum. It was shown that increasing the DDF
and II, the DDF values have increased significantly after 70 °C. value in a model transformer will cause the local resonant
0 frequencies in FRA spectrum move slightly to lower
-10
Deviated Frequency Band
frequencies, while deacreasing in DDF value will tend to shift
-20
these resonant points to higher frequencies. Furthermore, the
humidity contents in oil and paper insulations for each and
-30
every temperature were measured and moisture migration trends
Magnitude [dB]

-40 were illustrated. This in turn reveals that FRA resonant point is
-50 potentially initiated by winding capacitance, and of course the
latter is a function of paper humidity or even oil humidity. It is
-60
worth noting that the measurement of the DDF was performed
-70
at a single frequency (50 Hz). On the other hand, the Frequency
-80
Frequency Band:9 kHz- 20 MHz

HV winding spectrum (at 30 ºC), After dry-out


y-o
Domain Spectroscopy (FDS) can provide measurement of DDF
-90 4
HV winding spectrum (at 90 ºC), After dry-out
dr
5 6 7
value over a range of frequencies. Since this study revealed a
10 10
Frequency [Hz]
10 10
trend in FRA spectrum deviation due to DDF value alteration, it
0 appears that FRA and FDS data can be expected to be quite
-10 correlated and thus will be investigated in future work.
-20
REFERENCES
-30
[1] M. Bagheri, B.T. Phung, T. Blackburn, Transformer frequency response
Magnitude [dB]

-40 analysis: mathematical and practical approach to interpret mid-frequency


-50 oscillations”, IEEE Trans. Dielectr. Electr. Insul., Vol. 20, No. 6, pp.
1962-1970, 2013.
-60 [2] M. Bagheri, Mohammad S. Naderi, T. Blackburn, B.T. Phung,
-70
“Frequency response analysis and short circuit impedance measurement
in detection of winding deformation within power transformers”, IEEE
-80 Electr. Insul. Mag., vol. 29, issue 3, pp. 33-40, 2013.
[3] M. Bagheri, Mohammad S. Naderi, T. Blackburn, “Advanced transformer
-90
10
6
10
7
winding deformation diagnosis: moving from off-line to on-line”, IEEE
Frequency [Hz]
Trans. Dielectr. Electr. Insul., Vol. 19, No. 6, pp.1860-1870, 2012.
Figure 6. FRA spectra for HV winding due to moisture migration from paper [4] Mehdi Bagheri, B. T. Phung, T. Blackburn, “Influence of moisture
into the oil insulation at 30°C and 90°C, “dry” model test object. content variation on frequency response analysis of transformer winding”,
IEEE Int’l Electr. Insul. Conf. (EIC’14), Philadelphia, U.S.A., pp. 333-
337, June 2014.
Table II. DDF at 5 kV, moisture content in oil and paper insulations for “dry” [5] M. Bagheri, B. T. Phung, T. Blackburn, “Influence of temperature and
type test object. moisture content on frequency response analysis of transformer winding,”
IEEE Trans. Dielectr. Electr. Insul., Vol. 21, No. 3, pp. 1393-1404, 2014.
Temp. 30 °C 40 °C 50 °C 60 °C 70 °C 80 °C 90 °C
[6] M. Bagheri, B.T. Phung, T. Blackburn, Ali Naderian “Influence of
DDF 0.0028 0.0029 0.0044 0.0069 0.0088 0.0147 0.0210 temperature on frequency response analysis of transformer winding”,
Paper IEEE Int’l Electr. Insul. Conf. (EIC’13), pp. 40-44, 2013.
1.1 1.0 0.9 0.8 0.6 0.5 0.4
MC% [7] M. Bagheri, Mohammad S. Naderi, T. Blackburn, B.T. Phung, “Dean-
Oil Stark vs FDS and KFT methods in moisture content recognition of
3 3 5 8 10 16 21
MC[ppm] transformers”, IEEE Int’l Conf. Power Energy (PECON’12), Kota
Kinabalu, Malaysia, 2012, pp. 712-717.
0.022
[8] M. Bagheri, Mohammad S. Naderi, T. Blackburn, T. Phung, “FRA vs.
0.02 short circuit impedance measurement in detection of mechanical defects
within large power transformer”, IEEE Int’l. Sympos. Electr. Insul.
0.018
(ISEI’12), Puerto Rico, USA, 2012, pp. 301-305.
0.016 [9] M. Bagheri, Mohammad S. Naderi, T. Blackburn, B. T. Phung, “Dean-
0.014
Stark vs FDS and KFT methods in moisture content recognition of
transformers”, IEEE Int’l Conf. Power Energy (PECON’12), Kota
DDF

0.012
Kinabalu, Malaysia, pp. 712-717, 2012.
0.01 [10] A. Bouaïchz, I. Fofana, M. Farzaneh, A. Setayeshmehr, H. Borsi, E.
Gockenbach, A. Beroual, Ngnui Thomas Aka, “Dielectric spectroscopy
0.008
techniques as quality control tool: A feasibility study,” IEEE Electr. Insul.
0.006 Mag., vol. 25, issue 1, pp. 6-14, 2009.
0.004 [11] IEC Standard 60076-18; Measurement of frequency response, Edition
1.0, 2012-07.
0.002
20 30 40 50 60 70 80 90 100 [12] Y. Du, M. Zahn, B. C. Lesieutre, A. V. Mamishev,S. R. Lindgren,
Temperature ºC
“Moisture equilibrium in transformer paper-oil systems”, IEEE Electr.
Figure 7. DDF variation spectra for HV-LV windings, “wet” model test object. Insul. Mag., Vol. 15, No. 1, pp. 11-20, 1999.

978-1-4799-7525-9/14/$31.00 @2014 IEEE 637

You might also like