CHAPTER 1

INTRODUCTION TO FREQUENCY RESPONSE ANALYSIS (FRA)

1.1 Introduction:

In this chapter, various aspects of Frequency Response Analysis (FRA) like,

principle of FRA, types, various transfer function approaches and relevant circuit diagram,
factors influencing FRA, advantages and applications of using FRA approach etc., are
elaborated.

1.2 Principle of FRA:

The principle of FRA is based on the concept that a transformer can be modelled as a
complex RLC network, that responds to different frequencies in different ways. Also due to
this R, L and C network being unique for a transformer, its frequency response is unique.
This unique response can be used for condition monitoring of transformers by comparing
the frequency response obtained at any time, with the response obtained under healthy
conditions. In FRA, the transfer function approach is used for analysis purpose. The
comparison of transformer can be on the following basis:

 Time based comparison ( for understanding ageing effects)

 Type based comparison (for Fault detection, etc.)
 Design based comparison (for phase-phase comparison, etc.)

Fig 1.1 Complex RLC network of a Transformer winding and its sample Frequency Response

1
 1.3 Why FRA?

Conventional techniques like impulse tests, ratio test, SC test, IR test, DGA, Partial
discharge test, etc. are used to assess the condition of a transformer. These techniques have
certain drawbacks like problems in repeatability, influence due to loading effects, source side
and environmental influences, identification of location of fault, difficulties in interpreting
test results through waveforms alone etc. These drawbacks are, to an extent compensated in
Frequency Response Analysis (FRA).

1.4 Types of FRA:

FRA is classified based on the ways in which different frequency signals are injected
into the transformer: either as discrete frequencies (DFRA) or as several frequency
components combined and then injected into the transformer (IFRA & SFRA). The
classification of FRA is shown below:
Types of FRA

Sweep FRA (SFRA)

Lightning impulse FRA
(LIFRA)
Discrete Frequency Online (superimposed
FRA (DFRA) test)
Switching impulse FRA
(SIFRA)
Impulse based FRA
(IFRA)
Lightning impulse FRA
(LIFRA)
Offline
Switching impulse FRA
(SIFRA)

1.5 Types of Transfer Functions:

For FRA of transformers, two different transfer function approaches [3] can be carried out
namely: self-admittance transfer function and transfer voltage transfer function (TVTF) [6].
Both transfer functions can be used for any type of FRA, but depending on the type of
impulse used for testing, the self-admittance transfer function can preferably be open circuit
self-admittance (OCSA) for SIFRA and short circuit self-admittance (SCSA) for LIFRA.

2
 1.5.1 Circuit setup for different Transfer Functions [6]:

For self-admittance transfer function, only the winding under test (LV or HV) is used to
obtain the transfer function. For transfer voltage transfer function, both LV and HV winding
are used to obtain the transfer function. The circuit diagrams showing these setups are shown
below:

Fig 1.2 Self-admittance Transfer Function Fig 1.3 Transfer Voltage Transfer Function (TVTF)

For self-admittance transfer function, the supply voltage to winding is taken as input
and the current through the resistance (Rs) is the response. The ratio of response to the
input in the frequency domain yields the admittance transfer function for different frequency

signals. Y(s) = ( Rs ) .
V Rs

Vin

For transfer voltage transfer function, the supply voltage to winding is the input and
the voltage transferred across the other winding is the response. The ratio of response and
input in the frequency domain yields the transfer voltage transfer function for different
V2
frequency signals. Y(s) = .
V1

If required, the magnitude scale can be converted to dB scale using the formula:

Y(s) in dB = 20*log10[Y(s)]. The FRA plot is drawn with frequency in the x-axis and the
magnitude of transfer function in the y-axis.

3
 1.6 Factors Influencing FRA:

 Various factors like the energy of feeding source, test lead effects, grounding, noise
and data collection methods may influence the FRA test results.
 To have good repeatability and reliability, we have to ensure that the transformer and
test setup are consistent from one measurement to another (tap settings, external
connections, etc.). Documentation of setup in detail must be made for every
measurement.
 As the SNR (Signal to Noise Ratio) becomes low at high frequency region (>5MHz),
interpretation in this region can be avoided for reliable diagnosis.

1.7 Advantages of FRA:

Frequency Response Analysis has the following advantages over some of the
conventional techniques:

 It is a Non-Destructive (NDT) type of test where dismantling of the transformer is not

necessary.

 It can be performed at any voltage level since it uses characteristic of the transformer
like transfer function approach (self-admittance or transfer voltage transfer function).
 It can be used to determine the type, location and severity of the fault.
 It can detect even minor electric shorts (even 2% winding short) effectively.

 The transfer function of the transformer doesn’t change even if there is a slight change
in the wave shape of impulse, since it is the ratio of response to the input in frequency
domain.

 Repeatability, reliability and reduced time consumption are additional features of

FRA.

 Diagnosis can be done in both offline and online.

4
1.8 Applications of FRA:

The following are the areas where Frequency Response Analysis can be used:

 Detecting electrical faults like: short and open circuits, partial discharges, etc.
 Detecting mechanical faults like: axial and radial displacements, deformations,
dislocations, etc.
 Detecting magnetic core faults.
 Comparing different phases of the same transformer to check for symmetry.

1.9 Tabulation of Transformers tested:

In this project an attempt is made to diagnose faults in the following transformers:

TYPE OF
TRANSFORMER TYPE OF FRA FAULT DESCRIPTION ABOUT
RATING PERFORMED ARTIFICIALLY FAULT
CREATED
No fault, shorts between
tappings like 0%-12.5%,
3 phase, 315kVA,
LIFRA(SCSA) 12.5%-37.5%, 37.5%-
11kV/433V, Dyn11,
SIFRA( phase 62.5%, 62.5%-87.5%,
ONAN DT ELECTRICAL
comparison) 87.5%-100%, 0%-37.5%,
(100%-87.5%-62.5%-
DFRA (OCSA) top12.5% & bottom 12.5%,
37.5%-12.5%-0%) in 1U
bottom 12.5% & 62.5%-
37.5%.
No fault, 440-415 short,
1 phase, 3kVA,
440-415 open, 415- 380
440/240V, dry type LIFRA(TVTF)
short, 230-240 short on LV
transformer DFRA(OCSA, ELECTRICAL
side.
(440-415-380-0/ TVTF)
1) 440 V signal,0 V Gnd
240-230-0)V
2) 440 V Gnd, 0 V signal
1 phase, 1kVA, No fault, 0-30, 30-60, 0-60,
LIFRA(TVTF,
240/240V, dry type 60-120, 200-240 (tapping
SCSA)
transformer ELECTRICAL short)
DFRA(OCSA,
(240-200-120-60-30-0/ 1) 240 V signal, 0 V Gnd
TVTF)
240)V 2) 240 V Gnd, 0 V signal
1 phase, 3kVA,
250/750V disc type Discs connected
LIFRA(TVTF) MECHANICAL
transformer (6 numbers 1) 1-2-3-4
DFRA (OCSA, DISPLACEMEN
of discs in HV side, 250- 2) 1-2-3-5
TVTF) T OF DISCS
200-150-100-50-0V in 3) 1-2-3-6
LV side)
Table 1.1

CHAPTER-2

5
 REVIEW OF RELATED LITERATURE

2.1 Chronological survey of Research activities in FRA:

 1960: Low Voltage Impulse Method. First proposed by W. Lech & L. Tyminski in
Poland for detecting transformer winding deformation.

 1966: Results Published; “Detecting Transformer Winding Damage - The Low

Voltage Impulse Method”, Lech & Tyminsk, The Electric Review, ERA, UK

 1976: “Frequency Domain Analysis of Responses From L.V.I. Testing of Power

Transformers”, A.G. Richenbacher, 43rd Doble Conference

 1978: “Transformer Diagnostic Testing by Frequency Response Analysis”, E.P. Dick

& C.C. Erven, Ontario Hydro, IEEE Transactions of Power Delivery.

 1978: FRA Test developed at Ontario Hydro, Canada

 1980’s: Further research carried out by Central Electricity Generating Board in UK

 1988 - 1990’s : Proving trials by European utilities, the technology cascades

internationally via CIGRE, EuroDoble and many other conferences and technical
meetings

 2004: First SFRA Standard, ”Frequency Response Analysis on Winding Deformation

of Power Transformers”, DL/T 911-2004, is published by The Electric Power
Industry Standard of People’s Republic of China2007: IEEE PC57.149™/D6 Draft
Trial-Use Guide for the Application and Interpretation of Frequency Response
Analysis for Oil Immersed Transformers is prepared

 2007 &2008: CIGRE WG A2.26-Task Force 3-Interpretations on FRA & CIGRE

Report 342, ”Mechanical-Condition Assessment of Transformer Windings Using
Frequency Response Analysis (FRA)” are published

 2008-till date-various international publications , IFRA & SFRA test kit

developments for some specific fault identification by various industries and
worldwide standardization activities

2.2 International scenario of ongoing FRA related research works:

6
 Analysis of various factors influencing FRA-Voltage, wave shape, SNR, EMI etc.,

 HF Equivalent circuit of transformers based on FRA test results.

 Comparison FRA with other conventional and non-conventional diagnostic

methods of transformers.

 FRA- feasibilities- on line &off line.

 FRA- for fault identification and localization.

 Sensitivity analysis using analytical tools like De.coeff, OWA, mean, S.D,
correlation and regression, variance, tolerance band, weighing method, etc.

 FRA using SI, LI & Sweep Frequency- different possible test configurations and
their selectivity

 Modelling based approach on FRA & SFRA-PSPICE, MATLAB etc.,

 FRA- time based, type based and construction based approaches

 Identification of reliable set of dominant frequencies for FRA.

 Transfer Function approach with polar plots.

 Transfer Function approach- sensitivity for layer, disc and interleaved disc
windings.(of Distribution transformer, Power transformer , EHV transformer with
different methods of winding connections)

 Uses of amplitude and phase plots- combined approaches.

 Role of current shunts & HF C.T, digitizers, neutral grounding, cables –on
sensitivity and reproducibility.

 Vibration sweep frequency analysis

 Amplitude SFRA with F,G & DSO-linear sweeping and envelope acquisition.

CHAPTER 3

7
 EXPERIMENTAL INVESTIGATIONS ON ELECTRICAL FAULTS

3.1 Introduction:

This chapter deals with the Experimental investigations for electrical fault diagnosis on single
phase dry type transformers (one 1 kVA and two 3 kVA) and a three phase oil filled
distribution transformer (315 kVA).Electrical faults simulated are short circuit faults at
various tappings of a winding.

3.2 Experimental setup for DFRA:

8
 Fig.3.1 General Circuit for DFRA (Electrical)

[10]
Fig 3.1 shows a general layout of a DFRA setup . It has an Arbitrary Function Generator
(AFG) to generate signal of specified frequency which is given as input to the transformer
under test through matching impedance (50Ω) since the internal impedance of AFG is 50Ω so
that most of the applied voltage will appear across the input winding of the transformer
without reflections.

9
3.2.1 Open Circuit Self Admittance Transfer Function:

Fig.3.2 OCSA for 1ph, 1kVA Transformer Fig.3.3 OCSA for 1ph, 3kVA Transformer

Fig 3.2 and Fig.3.3 show experimental arrangements for a single phase 1kVA , 240/240V
transformer and a single phase 3kVA, 240/440V transformer respectively to obtain their
Open Circuit Self Admittance(OCSA) transfer functions using DFRA.The input signal which
is supplied by an Arbitrary Function Generator is given to one winding under test as shown
above and the response, current through that winding is obtained by measuring voltage across
the measuring resistor connected in series with the input winding. The other winding is left
open.

Fig.3.4 OCSA for 3ph, 315kVA Transformer Photo3.1

Fig.3.4 shows circuit diagram and the Photo 3.1 shows the experimental arrangement for a 3
phase 315kVA, 11000/433V transformer to obtain its Open Circuit Self Admittance(OCSA)
transfer function.The input signal is given to the winding in between 1U terminal and 1V
terminal( which is the winding under test) and the terminals 1V and IW are shorted and

10
grounded through a resistor.The current through the winding(which is the response) is
measured using a series resistance similar to that of single phase transformers.

3.2.1.1 Procedure:

Various faults are simulated in these transformers as shown in the Photo 3.2, which are listed
in the Table 3.1 and the measurements are done as follows.

 The voltage across the winding under test (V) is measured by connecting channel1 of
Digital Signal Oscilloscope(DSO) across that winding.
 The current through the winding(I) is measured by connecting channel 2 of DSO
across the measuring resistor(Rs). Dividing the voltage across this resistor by the
ohmic value of Rs will give the current value(I).
 Then dB magnitude of the OCSA transfer function is calculated by using formula
20log10(I/V).
 The transfer function graph is plotted by taking the Frequency in X-axis and the dB
magnitude of the OCSA transfer function in Y-axis.

Photo 3.2 Photo 3.3

11
3.2.2 Transfer Voltage Transfer Function:

Fig.3.5 TVTF for 1ph, 1kVA Transformer Fig.3.6 TVTF for 1ph, 3kVA Transformer

Fig.3.5 and Fig.3.6 show experimental arrangements for a single phase 1kVA 240/240V
transformer and a single phase 3kVA 240/440V transformer respectively to obtain their
Transfer Voltage Transfer Function(TVTF).The input signal which is supplied by an
Arbitrary Function Generator is given to one of the windings as shown above and the
response is the voltage transferred to the other winding.

3.2.2.1 Procedure:

Various faults are simulated in these transformers as shown in the Photos 3.4 and 3.5 which
are listed in the Table 3.1 and the measurements are done as follows.

 The voltage across the excited winding (V 1) is measured by connecting channel1 of

DSO across that winding.
 The voltage across the other winding (V2) is measured by connecting channel2 of
DSO across that winding.
 Then dB magnitude of the Transfer Voltage Transfer Function is calculated using the
formula 20log10(V2/V1).
 The transfer function graph is plotted by taking the Frequency in X-axis and the dB
magnitude of the Transfer voltage transfer function in Y-axis.

12
 Photo 3.4 Photo 3.5

3.2.3 Tabular column:

TRANSFORMER FAULT GROUPING OF FAULTS IN OCSA PLOT GROUPING OF FAULTS IN TVTF PLOT
WITH RATINGS SIMULATED

3kVA, 1ph, Electrical 1. No fault, 230-240V short 1. HV: 415-440V short, LV: 240V (case A & case B)
240/440V, dry-type (case A & case B) 2. HV: 415V, 440V(open), LV: 240V (case A & case B)
transformer 3. HV: 440V, LV: 230V(230-240V short)
(case A & case B)
4. HV: 440V(380-415V short), LV: 240V
(case A & case B)
5. HV: 415-440V short, LV: 240V (case A); HV:
415V(440V open), LV: 240V (case A)
6. HV: 415-440V short, LV: 240V (case B); HV:
415V(440V open), LV: 240V (case B)

1kVA, 1ph, Electrical 1. No fault, 0-30V (case A), 200-240V 1. No fault, 0-30V (case A), 200-240V (case B)
240/240V, dry-type (case B) 2. No fault, 0-30V (case B), 200-240V (case A)
transformer 2. No fault, 0-30V (case B), 200-240V 3. 0-30V (case A), 0-30V (case B)
(case A) 4. 30-60V (case A), 30-60V (case B)
3. 0-30V (case A), 0-30V (case B) 5. 0-60V (case A), 0-60V (case B)
4. 30-60V (case A), 30-60V (case B) 6. 200-240V (case A), 200-240V (case B)
5. 0-60V (case A), 0-60V (case B)
6. 200-240V (case A), 200-240V (case B)
7. 60-120V (case A), 60-120V
(case B)

315kVA, 3ph, Electrical 1. No fault, 100%-87.5% short, 12.5%-0% 1. NIL

11k/433V, short
Dyn11,ONAN,Distr 2. No fault, 87.5%-62.5%, 62.5%-37.5%,
ibution Transformer 37.5%-12.5% short

Table 3.1

13
Fig 3.7 HV terminal Signal applied LV terminal GND Fig 3.8 HV terminal GND LV terminal Signal applied

3.3 Experimental setup for LIFRA:

Fig.3.9 General Lightning Impulse setup

[5]
Fig.3.7 shows the general LIFRA setup in which 1.2/50 µs standard lightning impulse
voltage is generated and applied to one end of the transformer winding under test. The other
end of the winding under test is grounded through a series resistance of appreciable value.
The other windings of the transformer which is not under test should be shorted and earthed.

14
3.3.1 Short Circuit Self Admittance Transfer Function:

Fig.3.10 SCSA for 1ph, 1kVA Transformer

Fig.3.8 shows the experimental arrangement for a single phase 1kVA, 240/240V dry type
transformer to obtain its SCSA transfer function using LIFRA. The impulse signal is applied
to the winding under test and the response is the current through the winding which is
measured using a resistance (Rs) in series with the winding under test as shown in the fig. The
non-impulse winding is shorted and earthed.

Photo 3.6 Photo 3.7

Fig.3.11 SCSA for 3ph, 315kVA Transformer

15
Fig 3.9 shows the experimental setup a three phase 315kVA 11k/433V oil filled Distribution
transformer to obtain its SCSA transfer function using LIFRA. The impulse voltage signal is
applied to the 1U terminal of the winding under test and the response, impulse current is
measured as a voltage across the series resistance in series with the winding under test. The
LV windings, which are not under test, are shorted and earthed along with the neutral.

Photo 3.8 Photo 3.9

3.3.1.1 Procedure:

Various faults are simulated in these transformers as shown in the Photo 3.8 and 3.9 which
are listed in the Table 3.1 and the measurements are done as follows.

 The voltage waveform(in time domain) across the winding under test (V) is captured
by connecting channel1 of DSO to the impulsed end of the winding with respect to
ground.
 The waveform of the current through the winding(in time domain) is captured by
connecting channel 2 of DSO across the measuring resistor(Rs) .
 Frequency domain transformation of these time domain signals( obtained for both the
voltage(V) and current(I))are then carried out using Fast Fourier Transform(FFT)[7].
 The transfer function is then calculated by using the formula
SCSA Transfer function=I in frequency domain/V in frequency domain
 Then dB magnitude of the OCSA transfer function is calculated by using formula
SCSA transfer function in dB = 20log10(I/V).
 The transfer function graph is plotted by taking the Frequency in X-axis and the dB
magnitude of the OCSA transfer function in Y-axis.

16
3.3.2 Transfer Voltage Transfer Function:

Fig.3.12 TVTF for 1ph, 1kVA Transformer

Fig.3.11TVTF for 1ph, 3kVA Transformer

Fig.3.10 and Fig.3.11 show the experimental arrangements for a single phase 1kVA,
240/240V dry type transformer and single phase 3kVA 240/440 V dry type transformer
respectively to obtain their Transfer Voltage Transfer Function using LIFRA. The impulse
signal is applied to one end of the winding under test, with the other end earthed through
resistance (Rs). The response is the voltage transferred to the other winding (non-impulsed
winding).

17
 Photo 3.10 Photo 3.11

3.3.2.1 Procedure:

Various faults are simulated in these transformers as shown in Photo 3.10 & 3.11 which are
listed in the Table 3.1 and the measurements are done as follows.

 The voltage waveform (in time domain) across the impulsed winding (V 1) is captured
by connecting channel1 of DSO to the impulsed end of the winding with respect to
ground.
 The waveform of the (capacitively/inductively) transferred voltage in the other
winding, V2 (in time domain) is captured by connecting channel 2 of DSO across the
other winding whose one end is grounded.
 Frequency domain transformation of these time domain signals (obtained for both the
voltages (V1and V2)) are then carried out using Fast Fourier Transform(FFT).
 The transfer function is then calculated by using the formula
TVTF = V2 in frequency domain/V1 in frequency domain
 Then dB magnitude of the OCSA transfer function is calculated by using formula
TVTF in dB = 20log10(V2/V1).
 The transfer function graph is plotted by taking the Frequency in X-axis and the dB
magnitude of the TVTF in Y-axis.

18
 CHAPTER 4

EXPERIMENTAL INVESTIGATIONS ON MECHANICAL FAULTS

4.1 Introduction:

This chapter deals with the Experimental investigations for mechanical fault
diagnosis on single phase, 3kVA, 750/250V dry type transformer, with 6 discs on the HV
side. Mechanical faults simulated are, axial displacements and dislocations.

4.2 Experimental setup for DFRA:

Fig 4.1General circuit for DFRA (Mechanical)

Fig 4.1 shows a layout of a DFRA setup [8]. It has an Arbitrary Function Generator (AFG) to
generate signal of specified frequency which is given as input to the transformer under test
through matching impedance R1(50Ω) since, the internal impedance of AFG is 50Ω, so that
most of the applied voltage will appear across the input winding of the transformer without
reflections.

19
4.2.1 Open Circuit Self Admittance Transfer Function:

Fig 4.2 OCSA with No fault condition

Fig 4.3 OCSA with fault (one disc distance) Fig 4.4 OCSA with fault (two disc distance)

Fig 4.2, 4.3, 4.4 show experimental arrangements for a single phase 3Kva , 750/250V dry
type transformer respectively to obtain its Open Circuit Self Admittance(OCSA) transfer
function using DFRA. Fig 3.2 shows that first 4 discs of HV side of the transformer are
connected, Fig 3.3, first 3 discs and disc 5 are connected, which implies that disc 4 is
th
displaced by one disc distance(equivalent to dislocation of the 4 disc by one disc distance)
and in Fig 3.4, first 3 discs and disc 6 are connected, which implies that disc 4 is displaced by
th
two disc distance(equivalent to dislocation of the 4 disc by two disc distance). The test
signal which is supplied by an Arbitrary Function Generator is given to one end of the
winding under test through matching impedance R1(50Ω) as shown above and the response,
current through that winding is obtained by measuring voltage across the measuring resistor
connected in series with the other end of the same winding. The non tested winding (LV) is
left open.

20
4.2.1.1 Procedure:

Various faults are simulated in the transformer which are listed in the Table 4.1 and the
measurements are done as follows.

 The voltage across the winding under test (V) is measured by connecting channel1 of
Digital Signal Oscilloscope(DSO) across that winding.
 The current through the winding(I) is measured by connecting channel 2 of DSO
across the measuring resistor(Rs). Dividing the voltage across this resistor by the
ohmic value of Rs will give the current value(I).
 Then Db magnitude of the OCSA transfer function is calculated by using formula
20log10(I/V).
 The transfer function graph is plotted by taking the Frequency in X-axis and the Db
magnitude of the OCSA transfer function in Y-axis.

4.2.2 Transfer Voltage Transfer Function:

Fig 4.5 TVTF with No fault condition

Fig 4.6 TVTF with fault (one disc distance) Fig 4.7 TVTF with fault (two disc distance)

21
Fig.4.4, 4.5 and 4.6 show experimental arrangements for a single phase 3Kva, 750/250V
transformer respectively to obtain its Transfer Voltage Transfer Function(TVTF). The input
signal which is supplied by an Arbitrary Function Generator is given to one of the windings
through matching impedance R1 (50Ω) as shown above and the response is the voltage
transferred to the other winding.

4.2.2.1 Procedure:

Various faults are simulated in the transformer which are listed in the Table 4.1 and the
measurements are done as follows.

 The voltage across the excited winding (V 1) is measured by connecting channel1 of

DSO across that winding.
 The voltage across the other winding (V2) is measured by connecting channel2 of
DSO across that winding.
 Then Db magnitude of the Transfer Voltage Transfer Function is calculated using the
formula 20log10(V2/V1).
 The transfer function graph is plotted by taking the Frequency in X-axis and the Db
magnitude of the Transfer voltage transfer function in Y-axis.

4.2.3 Tabular column:

TRANSFORMER FAULT SIMULATED GROUPING OF GROUPING OF

WITH RATINGS FAULTS IN OCSA FAULTS IN TVTF
PLOT PLOT

3Kva, 1ph, 250/750V, Mechanical 1. discs connected: 1. discs connected:

dry-type ( HV 6 discs) 1-2-3-4, 1-2-3-5, 1-2-3-6 1-2-3-4, 1-2-3-5,
transformer 1-2-3-6
2. discs connected:
1-2-3, 4-5-6

Table 4.1

22
4.3 Experimental setup for LIFRA:

Fig 4.8 General Lightning Impulse setup (Mechanical)

Photo 4.1 Photo 4.2

Fig.4.6 shows the general LIFRA setup[2] in which 1.2/50 µs standard lightning impulse
voltage is generated and applied to one end of the transformer winding under test. The other
end of the winding under test is grounded through a series resistance (Rs) of low ohmic
value. The other windings of the transformer which is not under test should be shorted and
earthed.

23
4.3.1 Transfer Voltage Transfer Function:

Fig 4.9 TVTF for 1ph, 3kVA Transformer

Fig 4.7 and Photo 4.1 show the experimental arrangement for a single phase 3kVA,
750/250V dry type transformer and its Photographic view respectively, to obtain its Transfer
Voltage Transfer Function using LIFRA. The impulse signal is applied to one end of the
winding under test, with the other end earthed through resistance (Rs). The response is the
voltage transferred to the other winding (non-impulsed winding).

4.3.1.1 Procedure:

Various faults are simulated in this transformer as shown in Photo 4.2 which are listed in the
Table 4.1. The measurements are done as follows.

 The voltage waveform (in time domain) across the impulsed winding (V 1) is captured
by connecting channel1 of DSO to the impulsed end of the winding with respect to
ground.
 The waveform of the (capacitively/inductively) transferred voltage in the other
winding, V2 (in time domain) is captured by connecting channel 2 of DSO across the
other winding whose one end is grounded.
 Frequency domain transformation of these time domain signals (obtained for both the
voltages (V1and V2)) are then carried out using Fast Fourier Transform(FFT).
 The transfer function is then calculated by using the formula
TVTF = V2 in frequency domain/V1 in frequency domain
 Then dB magnitude of the OCSA transfer function is calculated by using formula
TVTF in dB = 20log10(V2/V1).
 The transfer function graph is plotted by taking the Frequency in X-axis and the dB
magnitude of the TVTF in Y-axis.

24
 CHAPTER 5

EXPERIMENTAL INVESTIGATIONS WITH

ONLINE FRA

5.1 Introduction:

This chapter deals with online frequency response investigations conducted on a

3kVA, 240/440V single phase transformer. The input to the transformer winding under test is
the composite signal comprising of impulse signal superimposed over power frequency signal
and its response is the composite current through the winding. The experiments were
performed using both lightning impulse voltage superimposed on 50 Hz signal and switching
impulse voltage superimposed on 50Hz. As the investigations are mainly intended to
understand the feasibility of FRA as an on line diagnostic technique and, to understand the
difficulties associated with, importance is not given to the magnitude of the composite signal
and, only signals of low magnitudes are engaged.

5.2 Block diagram & Experimental setup:

5.2.1 Superimposition of LI over Power Frequency:

Fig 5.1 Lightning Impulse signal over Power frequency

25
5.2.2 Superimposition of SI over Power Frequency:

Fig 5.2 Switching Impulse signal over Power frequency

5.3 Circuit description:

Fig 5.1 and 5.2 show the experimental arrangements for Online FRA[9] on a single
phase, 3kVA, 240/440V dry type transformer. A 1kVA, 240/240V, isolation transformer is used as
source for power frequency (50Hz) signal. A Resistor (2400Ω) is placed between this source and the
transformer under test, which damps/reduces the effect of impulse signal over the a.c source. This
provision is introduced to prevent the impulse signal from affecting the power frequency source,
though it drops some amount of a.c voltage available from a.c source. A capacitive divider
arrangement (comprising of CD1&CD2) is connected between the Impulse voltage source and the
transformer under test, to get an impulse of low magnitude. This gets superimposed over the already
existing power frequency voltage across the transformer winding under test. This capacitive voltage
divider also helps in reducing the effect of power frequency signal over the impulse generation set
up.

Online FRA tests can be grouped in the following two ways:

 Impulse signal superimposed directly over the alternating voltage appearing across
the winding under test(Direct a.c method)
 Impulse signal superimposed over the induced alternating voltage appearing across
the winding under test.(induced a.c method)

In Direct AC case, Impulse signal is directly applied at the Input terminal along with Power
frequency signal. i.e., both the impulse and power frequency signal is applied at same point in
the transformer.

26
In Induced AC case, Impulse signal is applied at the secondary side of the transformer while
the Power frequency signal is at primary. So impulse is superimposed over the induced AC
signal available at the secondary.

The above circuit diagrams show both the direct and induced AC superimposed test circuits.
The Impulse signal to be superimposed is made available across capacitor CD2. In Direct AC
method, when the switch S closes through terminal 2, both the signals (Power frequency and
Impulse) will be applied to the transformer winding at the same terminal. In Induced AC
method, when the switch S closes through terminal 3, the Power frequency signal will be
applied at the primary of the transformer and the impulse gets superimposed over the induced
AC signal available at the secondary.

Photo 5.1 Photo 5.2

Photo 5.3 Photo 5.4

27
Photo 5.1 and 5.2 shows the setup for superim2position of lightning impulse and power
frequency signal and Photo 5.3 and 5.4 shows the setup for superimposition of switching
impulse and power frequency signal.

5.4 Procedure:

Various faults are simulated in the transformer and the measurements are done as follows.

 The composite voltage applied across the winding under test (V), is measured by
connecting the same to channel1 of Digital Signal Oscilloscope(DSO) .
 The current through the winding(I), is measured by connecting channel 2 of DSO
across the measuring resistor(Rs). Dividing the voltage across this resistor by the
ohmic value of Rs will give the current value(I).
 Then dB magnitude of the OCSA transfer function is calculated by using formula
20log10(I/V).

28
 CHAPTER 6

EXPERIMENTAL INVESTIGATION ON PHASE-TO-PHASE COMPARISON

6.1 Introduction:

In this chapter, an Experimental investigation is done a three phase oil filled (ONAN)
distribution transformer (315kVA). In the experimental work the three windings, on the HV
side of the transformer are tested one by one individually and the OCSA transfer function
plots were developed to check for symmetry between phases. Checking for the phase
symmetry in transformer is required to ensure a three phase balanced power supply from the
transformer and this is called as the design based comparison within a transformer.

6.2 Experimental setup for phase comparison:

6.2.1 LIFRA based phase comparison:

Fig. 6.1 LIFRA phase comparison of 3ph, 315kVA Transformer

Fig. 6.1 show the experimental setup for performing phase comparison in the three phase
transformer, using LIFRA[1] with one winding under test, at a time. The setup remains same
as that of the LIFRA for electrical faults.

For testing the winding between (1U & 1V) , Impulse voltage signal (lightning
impulse) is injected on one of the HV terminal, say 1U, with the other two terminals (1V,
1W) shorted and earthed through a resistance of low ohmic value. The LV windings are also
shorted and earthed during test. A similar test setup is arranged for the testing of other two
HV windings individually, i.e. for testing winding between (1V& 1W), 1V energized, with
1U, 1W shorted and earthed through the resistance (Rs) and for testing the third winding
which is in between (1W &1U), 1W energized, with 1U, 1V shorted and earthed through the

29
resistance (Rs). For all these configurations, the LV side is always completely shorted and
earthed.

Photo 6.1 Photo 6.2

6.2.2 SIFRA based phase comparison

Fig. 6.2 SIFRA phase comparison of 3ph, 315kVA Transformer

Fig. 6.2 shows the experimental setup for performing phase comparison on the transformer
using SIFRA with one winding tested at a time. The setup remains same as that of the SIFRA
for electrical faults. For testing the winding between (1U & 1V) , Impulse voltage signal
(switching impulse) is injected on one of the HV terminal, say 1U, with the other two
terminals (1V, 1W) shorted and earthed through a resistance of low ohmic value. The LV
winding terminals 2u, 2v and 2w are kept open during test and the other terminal of the LV
windings 2n is earthed. A similar test setup is arranged for the testing of other two HV
windings individually, i.e. for testing winding between (1V& 1W), 1V energized, with 1U,

30
1W shorted and earthed through the resistance (Rs) and for testing the third winding which is
in between (1W &1U), 1W energized, with 1U, 1V shorted and earthed through the
resistance (Rs).For all these configurations, the LV side terminals 2u, 2v and 2w are always
kept open and the terminal 2n is connected to earth.

Photo 6.3 Photo 6.4

6.3 Self admittance transfer function:

6.3.1 LIFRA based short circuit self admittance transfer function:

Fig. 6.3 LIFRA based SCSA for 3ph, 315kVA Transformer

Fig.6.3 show the circuits used for LIFRA based phase comparison on a 3-ph, 315kVA,
11000/433V, Dyn11, ONAN, distribution transformer. The standard lightning impulse is
generated using a Marx Impulse generator setup and applied to the winding under test, with
the terminal connections of the windings as shown in the diagrams. The impulse applied to
the phase under test is taken as input (V) and the response is the current (I), through the tested
phase, which is measured in terms of voltage across the resistance (Rs). The ratio of I and V

31
(each in frequency domain) gives the short circuit self-admittance transfer function of the
transformer. Similarly, tests are performed on other phases as well and the corresponding
transfer function for each phase is obtained for phase comparison.

6.3.2 SIFRA based open circuit self admittance transfer function:

Fig. 6.4 SIFRA based OCSA for 3ph, 315kVA Transformer

Fig.6.4 shows the circuits used for SIFRA based phase comparison on a 3-ph, 315kVA,
11000/433V, Dyn11, ONAN, distribution transformer. The standard switching impulse is
generated using a Marx Impulse generator setup and applied on to the phase under test with
the terminal connections as shown in the circuit diagrams. The impulse applied to the phase
under test is taken as input (V) and the response is the current (I), through the tested phase,
which is measured in terms of voltage across the resistance (Rs). The ratio of I and V (each
in frequency domain) gives the open circuit self-admittance transfer function of the phase
of the transformer under test.

Similarly, tests are performed on other phases as well and the corresponding transfer function
for each phase is obtained for phase comparison.

32
 Photo 6.5 Photo 6.6

6.4 Procedure:

The three phases are tested individually as shown in Photo 6.4. The procedure for obtaining
the Transfer function (SCSA TF for LIFRA & OCSA TF for SIFRA) of the phases one by
one, with both lightning and switching impulses, remain the same, as explained below:

 The voltage waveform(in time domain) across the winding under test (V) is captured
by connecting channel1 of DSO to the impulsed end of the winding with respect to
ground.
 The waveform of the current through the winding (in time domain) is captured by
connecting channel 2 of DSO across the measuring resistor(Rs) .
 Frequency domain transformation of these time domain signals (obtained for both the
voltage(V) and current(I)) are then carried out using Fast Fourier Transform(FFT).
 The transfer function is then calculated by using the formula
Transfer function=I in frequency domain/V in frequency domain
 Then dB magnitude of the transfer function is calculated by using formula
Transfer function in dB = 20log10(I/V).
 The transfer function graph is plotted by taking the Frequency in X-axis and the dB
magnitude of the transfer function in Y-axis.

33
 CHAPTER 7

RESULT ANALYSIS

7.1 Introduction:

This chapter deals with the results obtained from the various experimental
investigations explained in the earlier chapters and the inferences got from the analysis on the
same. For this purpose, the transfer function plots under no fault condition (reference or Ideal
patterns) and under various artificially created faulty conditions (actual patterns) on different
tested transformers were first drawn. Comparisons were made between the reference and
response patterns to identify the faults, their severity and their location in the transformer.

7.2 Analysis on electrical faults:

7.2.1 Fault diagnosis through DFRA approach:

The tabular column pertaining to test results on different transformers for different
transfer functions obtained through the procedure explained in chapter 3 are given for ready
reference.

OCSA TRANSFER FUNCTION TABULATION for 3-ph, 315kVA, 11000/433V, Dyn 11,
ONAN Distribution Transformer, (electrical faults).

REFER APPENDIX S.NO 4

34
Transfer function Plots depicting Fault in transformer by Change in Frequency Response

Frequency (Hz)
0
0 5000000 10000000 15000000 20000000 25000000 30000000
-10
OCSA MAGNITUDE in dB

-20

-30
NO FAULT
-40
100%-
87.5%
-50 SHORT

-60 12.5%-0%
SHORT

-70

-80

Frequency (Hz)

0
0 5000000 10000000 15000000 20000000 25000000 30000000
-10
OCSA MAGNITUDE in dB

-20

-30 no fault

-40 87.5%-62.5%
winding short

-50 62.5%-37.5%
winding short
-60 37.5%-12.5%
winding short
-70

-80

Frequency (Hz)
0
0 5000000 10000000 15000000 20000000 25000000 30000000
-10
OCSA MAGNITUDE in dB

-20

-30

-40 100% - 87.5%

winding short
-50

62.5% - 37.5%
-60 winding short

-70

-80

35
From the plots shown , it is observed that, when there is an electrical fault in a transformer,
there are changes in the FRA plot for faulty case when compared to the reference FRA plot
(obtained under no fault condition) in the following aspects:

 Magnifications or demagnifications in the magnitude of transfer function at certain

dominant frequencies (poles and anti-poles).
 Shift of certain dominant frequencies, to either side in the FRA plot for faulty case
when compared to the reference FRA plot.
 Appearance of new dominant frequencies or absence of already present dominant
frequency pole/anti-pole.

OCSA TRANSFER FUNCTION TABULATION for 1ph, 1kVA, 240/240V, dry-type

transformer (electrical faults).

REFER APPENDIX S.NO 2

36
Transfer function Plots depicting Fault in transformer by Change in Frequency Response

Frequency (Hz)

0
0 500000 1000000 1500000 2000000 2500000
OCSA MAGNITUDE in dB

-10

-20 NO FAULT

-30
0V-30V
TAPPING
-40 SHORT
CASE(A)

-50 200V-240V
TAPPINGSHORT
CASE(B)
-60

-70

Frequency (Hz)

0
0 500000 1000000 1500000 2000000 2500000
OCSA MAGNITUDE in dB

-10

-20
0V- 30V
-30 TAPPING
SHORT
CASE(A)
-40

0V- 30V
-50 TAPPING
SHORT
CASE(B)
-60

-70

Frequency (Hz)
0
0 500000 1000000 1500000 2000000 2500000
OCSA MAGNITUDE in dB

-10

-20

-30 30V- 60V

TAPPING
SHORT
-40 CASE(A)

30V- 60V
-50 TAPPING
SHORT
CASE(B)
-60

-70

Frequency (Hz)
0
0 500000 1000000 1500000 2000000 2500000
OCSA MAGNITUDE in dB

-10

-20
30 -60
SHORT
-30

-40 0 -60
SHORT
-50
200 -240
-60 SHORT

-70

37
TVTF TRANSFER FUNCTION TABULATION for 1ph, 1kVA, 240/240V, dry-type
transformer (electrical faults). REFER APPENDIX S.NO 6

Transfer function Plots depicting Fault in transformer by Change in Frequency Response

Frequency (Hz)

20
TVTF MAGNITUDE in dB

NOFAULT

10

0V-30VTAPPING
0 SHORTCASE(A)
0 5000000 10000000 15000000 20000000 25000000

-10
200V-240VTAPPING
SHORTCASE(B)

-20

-30

-40

Frequency (Hz)

20
TVTF MAGNITUDE in dB

10
30V- 60V
TAPPING
SHORT
0 CASE(A)
0 5000000 10000000 15000000 20000000 25000000

-10

30V- 60V
-20 TAPPING
SHORT
CASE(B)
-30

-40

Frequency (Hz)

15
TVTF MAGNITUDE in dB

10

0 0V- 60V
0 5000000 10000000 15000000 20000000 25000000 TAPPING
-5 SHORT
CASE(A)
-10

-15
0V- 60V
-20 TAPPING
SHORT
-25 CASE(B)

-30

-35

Frequency (Hz)

20
TVTF MAGNITUDE in dB

10 0 -30
SHORT

0
0 5000000 10000000 15000000 20000000 25000000
-10 0 -60
SHORT

-20

200 -
-30 240
SHORT
-40

-50

38
From the above plots it is observed that, when there is a fault in a transformer, there is change
in the FRA plot. Also there are changes in the plot, when there is change in the fault location.
These changes in the FRA plot can be as follows:

 Magnifications or demagnifications in the magnitude of transfer function at certain

dominant poles and anti-poles.
 Shift in the frequency of dominant poles and anti-poles to either side of the present
location.
 Appearance of new dominant frequencies or absence of already present dominant
frequency pole/anti-pole.

The two transfer function plots (OCSA TF & TVTF) drawn for the same transformer
(for the same artificially created faults), are compared. It is observed that the OCSA transfer
function is easy to interpret, when compared to TV Transfer function, due to fewer spikes in
the transfer function plot.

OCSA TRANSFER FUNCTION TABULATION for 1ph, 3kVA, 240/440V, dry-type

transformer (electrical faults).

REFER APPENDIX S.NO 3

39
Transfer function Plots depicting Fault in transformer by Change in Frequency Response

Frequency (Hz)
0
0 500000 1000000 1500000 2000000 2500000

-10
OCSA MAGNITUDE in dB

-20
NOFAULT
-30
230V-240V
-40 TAPPINGSHORT
(0VTAPPING
SIGNALAP-
PLIED)
-50
230V-240VTAP-
PINGSHORT
-60 (240VTAPPING
SIGNALAPPLIED)

-70

-80

Frequency (Hz)
0
0 500000 1000000 1500000 2000000 2500000
OCSSA MAGNITUDE in dB

-10

-20

-30
230V-240V
TAPPINGSHORT
-40 CASE(B)

-50 230V-240V
TAPPINGSHORT
CASE(A)

-60

-70

It can be observed that there is change in FRA plot, even for small short between 230V-240V
tapping (4.167% of total winding) at different locations. This change shows that FRA plots
can also be used to identify minor faults along with their locations.

TV TRANSFER FUNCTION TABULATION for 1ph, 3kVA, 240/440V, dry-type

transformer (electrical faults).

REFER APPENDIX S.NO 7

40
Transfer function Plots depicting Fault in transformer by Change in Frequency Response
Frequency (Hz)
TVTF MAGNITUDE in dB
20

10

0
0 5 0 00 0 0 0 1 0 00 0 0 00 1 5 00 0 0 00 2 0 00 0 0 00 2 5 00 0 0 00

-1 0

HV415V-440V
-2 0 TAPPINGSHORT
LV240VCASE(A)

-3 0
HV415V-440V
TAPPINGSHORT
-4 0 LV240VCASE(B)

-5 0

Frequency (Hz)
TVTF MAGNITUDE in dB

10

0 HV440VLV230V
0 5 0 00 0 0 0 1 0 00 0 0 00 1 5 00 0 0 00 2 0 00 0 0 00 2 5 00 0 0 00 WITH230V-240V
TAPPINGSHORT
-5
CASE(A)

-1 0

-1 5
HV440VLV230V
-2 0 WITH230V-240V
TAPPINGSHORT
CASE(B)
-2 5

-3 0

-3 5

Frequency (Hz)
TVTF MAGNITUDE in dB

15

10

0 HV440VWITH
0 5000000 10000000 15000000 20000000 2 5 0 0 0 0 0 0 380V-415V
TAPPINGSHORT
-5 LV240VCASE(A)

-1 0

-1 5
HV440VWITH
-2 0 380V-415V
TAPPINGSHORT
LV240VCASE(B)
-2 5

-3 0

-3 5

Frequency (Hz)
TVTF MAGNITUDE in dB

20

10

0
0 5000000 10000000 15000000 20000000 25000000 HV415-440
SHORTLV240
CASE(B)
-1 0

-2 0

HV415WITH
-3 0 440OPENLV
240CASE(B)

-4 0

-5 0

It can be observed that the fault in the non-signalled winding can also be identified by using
TVTF which may not be possible using OCSA transfer function.

41
7.2.2 Fault diagnosis through IFRA approach:

The transfer function plots pertaining to test results on different transformers obtained
through the procedure explained in chapter 3 are shown below. For LIFRA, the non-tested
winding is generally shorted and earthed while for SIFRA, the non-tested winding is left open
in accordance to “IEC 60076 - IMPULSE TESTING STANDARD FOR
TRANSFORMERS” [1].

SCSA TRANSFER FUNCTION PLOTS for 3-ph, 315kVA, 11000/433V, Dyn 11, ONAN
Distribution Transformer, (electrical faults).
SCSA MAGNITUDE IN dB

1 FREQUENCY IN Hz

0.5

0
0 500000 1000000 1500000 2000000 2500000
-0.5

-1
NO FAULT
-1.5
37.5%-12.5%
-2 WINDINGSHORT

-2.5
87.5%-62.5%
WINDINGSHORT
-3

-3.5
SCSA MAGNITUDE IN dB

1 FREQUENCY IN Hz

0.5

0
0 500000 1000000 1500000 2000000 2500000
-0.5

-1
Bottom12.5%
-1.5 and 62.5%-
37.5%winding
-2 short

-2.5 TOP 12.5%

and BOT-
-3 TOM12.5%
windingshort
-3.5

1 FREQUENCY IN Hz
SCSA MAGNITUDE IN dB

0.5

0
0 500000 1000000 1500000 2000000 2500000
-0.5

-1
0%-12.5%
-1.5 winding
short
-2
100%-
-2.5 87.5%
winding
short
-3

-3.5

42
 From the plots shown , it is observed that, when there is an electrical fault in a
transformer, there are changes in the FRA plot for faulty case when compared to the
reference FRA plot (obtained under no fault condition) in the following aspects:

 Magnifications or demagnifications in the magnitude of transfer function at certain

dominant frequencies (poles and anti-poles).
 Shift of certain dominant frequencies, to either side in the FRA plot for faulty case
when compared to the reference FRA plot.
 Appearance of new dominant frequencies or absence of already present dominant
frequency pole/anti-pole.

Depending on the sampling rate of the impulse signal in time domain and the number
of samples taken to perform FFT of the time domain signal, the plot may vary. Hence it is
essential to maintain the same sampling rate and number of samples throughout the
testing. Too large the number of samples, more difficult is the interpretation.

43
TV TRANSFER FUNCTION PLOTS for 1ph, 1kVA, 240/240V, dry-type transformer
(electrical faults).

10
TVTF MAGNITUDE IN dB

FREQUENCY IN Hz
5

0
0 500000 1000000 1500000 2000000 2500000
-5

-10 NO FAULT

-15
0-30 V
-20 SHORT
(CASEA)
-25
200-240V
-30 SHORT
(CASEB)
-35

-40

20
TVTF MAGNITUDE IN dB

FREQUENCY IN Hz

10

0
0 500000 1000000 1500000 2000000 2500000
NO FAULT

-10
SUPPLYAT
220V,240
-20 OPEN(CASEA)

-30 SUPPLYAT
220V,240
OPEN(CASEB)

-40
TVTF MAGNITUDE IN dB

10
FREQUENCY IN Hz
5

0
0 500000 1000000 1500000 2000000 2500000
-5

-10
0-60V
SHORT
-15 (CASE A)

-20 0-60V
SHORT
(CASE B)
-25

-30

44
SCSA TRANSFER FUNCTION PLOTS for 1ph, 1kVA, 240/240V, dry-type transformer
(electrical faults).

FREQUENCY IN Hz
0
0 500000 1000000 1500000 2000000 2500000

-10
SCSA MAGNITUDE IN DB

-20

-30

NO FAULT
-40

0V-60V
TAPPING
-50 SHORT

-60

FREQUENCY IN Hz
0
0 500000 1000000 1500000 2000000 2500000

-10
SCSA MAGNITUDE IN DB

-20

-30 NO FAULT

-40
30V-60V
TAPPING
SHORT
-50

-60

45
 FREQUENCY IN Hz
0
0 500000 1000000 1500000 2000000 2500000

-10
SCSA MAGNITUDE IN DB

-20

NO FAULT
-30

-40
200V-240V
TAPPING
SHORT
-50

-60

It can be seen that SCSA transfer function is more easily interpretable for electrical
faults compared to TVTF for LIFRA. Also it depends on sampling rate and number of
samples for FFT, as explained for previous transformer.

SCSA TRANSFER FUNCTION PLOTS for 1ph, 3kVA, 240/440V, dry-type transformer
(electrical faults).

46
 TVTF MAGNITUDE IN dB
25
FREQUENCY IN Hz
20

15

10

0
0 500000 1000000 1500000 2000000 NO FAULT
2500000
-5

-10
LV440
WITH440
-15 OPENLV
240(CASE
-20 B)

-25

10

0
0 500000 1000000 1500000 2000000 2500000
-10

-20

HV440 LV230
-30 WITH230-240
SHORT(CASEA)
-40

-50 HV440 LV230

WITH230-240
SHORT(CASEB)
-60

-70

30
FREQUENCY IN Hz
TVTF MAGNITUDE IN dB

20

HV440 WITH
10 380-415
SHORTLV
240(CASEA)
0
0 500000 1000000 1500000 2000000 2500000

-10
HV440 WITH
380-415
-20 SHORTLV
240(CASEB)

-30

-40

It can be observed that in TVTF, there is possibility to check for fault in the non-
impulsed winding as well. Since the secondary voltage is also to be measured (for which the
secondary winding should be kept open), care should be taken so that the lightning impulse is
not dangerously high, which may damage the transformer or the DSO attached to the
transformer for measurement of input and its response.

7.3 Analysis of mechanical faults:

47
7.3.1 Fault diagnosis through DFRA approach:

The tabular column pertaining to each of the mechanical fault artificially created as
explained in chapter 4 is given below for reference. The test frequencies are carefully
selected such that, there is predominant change in the transfer functions at this frequencies
(frequency selection is through careful observation of the input and its response in time
domain. More points may be taken in case necessary).

OCSA TRANSFER FUNCTION TABULATION for 1ph, 3kVA, 250/750V, dry-type

transformer; HV: disc winding (with 6 discs) & LV: layer winding, (mechanical faults).

REFER APPENDIX S.NO 1

Transfer function Plots depicting Fault in transformer by Change in Frequency Response

Frequency (Hz)
-60
0 500000 1000000 1500000 2000000 2500000

-62
OCSA MAGNITUDE in dB

-64

-66

1-2-3-4
-68

1-2-3-5
-70

1-2-3-6
-72

-74

-76

TV TRANSFER FUNCTION TABULATION for 1ph, 3kVA, 250/750V, dry-type

transformer HV: disc winding (6 discs) and LV: layer winding, (mechanical faults).

REFER APPENDIX S.NO 5

Transfer function Plots depicting Fault in transformer by Change in Frequency Response

48
 Frequency (Hz)
0
0 5000000 10000000 15000000 20000000 25000000
TVTF MAGNITUDE in dB

-10

-20 1-2-3-4

1-2-3-5
-30
1-2-3-6

-40

Frequency (Hz)
0
0 5000000 10000000 15000000 20000000 25000000
-5
TVTF MAGNITUDE in dB

-10

-15

-20

-25 1-2-3
DISC

-30
4-5-6
-35 DISC

-40

-45

Photo 7.1 Discrete Input frequencies and their Responses (snapshot view)

49
 It can be observed from the plots that, when there is a local displacement
(dislocation), there is both change in magnitude and shift in dominant peaks in the actual
plots (obtained under faulty condition).

But, when telescoping (axial displacement of total winding) occurs, there is change in
magnitude of the dominant poles in the plots. Also, it can be seen that TVTF plots are more
easily interpretable for mechanical faults, when compared to OCSA transfer function plots,
due to easily observable changes in the TVTF plots.

7.3.2 Fault diagnosis through IFRA approach:

The transfer function plots pertaining to test results on different transformers obtained
through the procedure explained in chapter 4 is shown below. For LIFRA, the non-tested
winding is generally shorted and earthed while for SIFRA, the non-tested winding is left open
in accordance to “IEC 60076 - IMPULSE TESTING STANDARD FOR
TRANSFORMERS” [1].

OCSA TRANSFER FUNCTION TABULATION for 1ph, 3kVA, 250/750V, dry-type

transformer; HV: disc winding (with 6 discs) & LV: layer winding, (mechanical faults).

Transfer function Plots depicting Fault in transformer by Change in Frequency Response

10
FREQUENCY IN Hz

0
0 500000 1000000 1500000 2000000 2500000
TVTF MAGNITUDE IN dB

-10
1-2-3-4 DISCS
-20
1-2-3-5 DISCS

-30 1-2-3-6 DISCS

HV440 WITH
-40 380-415 SHORT
LV240(CASEA)

-50 HV440 WITH

380-415 SHORT
LV240(CASEB)
-60

50
 It can be observed that DFRA method is more interpretable compared to LIFRA
mainly due to large difference in the number of points considered for drawing the plot. Also
it can be seen that not only electrical faults but mechanical faults like winding movements
which are difficult to identify without dismantling the transformer can also be identified
through FRA, either through IFRA or DFRA.

7.4 Analysis on phase symmetry:

7.4.1 Phase-to-Phase comparison through LIFRA approach:

As explained in chapter 6 (sec 6.2.1), the phase to phase comparison is made on

315kVA, 11000/433V, Dyn11, ONAN, Distribution Transformer by plotting short circuit
self-admittance transfer function (SCSA) obtained through Fast Fourier Transformation
(FFT) using lightning impulse. The plot for each phase individually tested is also shown
below:

SCSA TRANSFER FUNCTION PLOTS for 3-ph, 315kVA, 11000/433V, Dyn 11, ONAN
Distribution Transformer (phase comparison)

Frequency (Hz)
0.2

0
0 500000 1000000 1500000 2000000 2500000
-0.2
SCSA MAGNITUDE in dB

-0.4

-0.6 1U PHASE

-0.8
1V PHASE
-1
1W
-1.2 PHASE

-1.4

-1.6

51
 Photo 7.2 Photo 7.3

7.4.2 Phase-to-Phase comparison through SIFRA approach:

As explained in chapter 6 (sec 6.2.1), open circuit self-admittance transfer function

(SCSA) plots for the three phases (individual) obtained through Fast Fourier
Transformation (FFT) using switching impulse. The plot for each phase individually tested
is shown below.

OCSA TRANSFER FUNCTION PLOT for 3-ph, 315kVA, 11000/433V, Dyn 11, ONAN
Distribution Transformer (phase comparison).

52
 Frequency (Hz)
0
0 500000 1000000 1500000 2000000 2500000

-0.01
OCSA MAGNITUDE in dB

-0.02 1U PHASE W
OPEN V GND

-0.03 1V PHASE U
OPEN W GND

-0.04 1V PHASE W
OPEN V GND

-0.05 1W PHASE U
OPEN V GND

-0.06

Photo 7.4 Photo 7.5

53
 From the transfer function plots shown (LIFRA & SIFRA), it can be observed that for
a major range of frequency bands, there is not much change in the frequency response
between the phase windings. This shows that the transformer phases are symmetric since
each phase must have similar RLC ladder network to show similar, (unique) frequency
response. The minor differences in the plot between phases are due to special tappings
available only in one of the phases (between 1U and 1V terminals) and design based
constraints in the 3-phase transformer.

In LIFRA based phase comparison, only one of the phases on the tested side (i.e. HV
side) is bypassed by shorting the two terminals of that phase, but not both the phases other
than the phase that is under test. This is because since HV is delta connected, shorting two of
the phases also bypass the third phase, thereby shorting the entire HV side.

In SIFRA based phase comparison, no phase is bypassed on the tested side (i.e. HV
side) so as to test all the three phases simultaneously. The plot corresponding to signal
applied at every terminal (1U, 1V, 1W) remains similar. The minor variations are due to
presence of special tappings in one of the phases and its location corresponding to the
impulsed terminal.

54
 CONCLUSION

From the experimental investigations carried out on various dry and oil-filled
transformers, for the electrical and mechanical fault diagnosis through various transfer
function approaches the following conclusions are made.

 FRA can be effectively used as a Non-Destructive diagnostic tool for identifying,

measuring and locating electrical and mechanical faults in transformers.
 OCSA Transfer function, SCSA Transfer function as well as Transfer Voltage
Transfer function with various frequency response approaches like DFRA & IFRA
are capable of identifying the mechanical and electrical faults though their
sensitivities differ in locating the same.
 The difference between ideal plot and actual plot is observable in terms of
1. Change in the magnitude of some dominant frequencies.
2. Shift in some dominant frequencies.
3. Appearances of new dominant frequencies/absence of some of the
existing dominant frequencies.
 For detecting electrical faults, it is observed that the OCSA transfer function is easy to
interpret, when compared to TV Transfer function, due to fewer spikes in the transfer
function plot.
 TVTF plots are more easily interpretable for mechanical faults, when compared to
OCSA transfer function plots, due to easily observable changes in the TVTF plots.
 Both the SCSA & OCSA Transfer functions can be used for comparing phase
symmetry in three phase transformers through LIFRA & SIFRA approaches
correspondingly.
 Online FRA can be used as effective non destructive tool for fault diagnosis in
transformers provided proper isolation/potential division is ensured between the super
positioned signals.

55
FUTURE SCOPE:

SWEEP FREQUENCY (SFRA) response approach can be tried out on the

transformers for electrical and mechanical fault diagnosis which can probably reduce
further, the time required for fault diagnosis.

Fault diagnosis can also be tried out on all the phases simultaneously with the help
of signal sources of adequate capacity and a four channel DSO for measurements
purposes.

Three dimensional diagnostic approaches (time-frequency-TF magnitude) can be

attempted for easier interpretation on location & severity of faults.

56
 BIBLIOGRAPHY

[1] IEC 60076 part-III &IV Impulse Voltage Test on Power Transformers.
[2] CIGRE WG A2.26: Report Mechanical Condition Assessment of Transformer
Windings Using FRA.
[3] IEEE PC57.149™/D3 Draft Trial-Use Guide for the Application and Interpretation of
Frequency Response Analysis for Oil Immersed Transformers.
[4] Riszard Malewski and Bertand Poulin, “Impulse Testing of Power Transformers using
the Transfer Function Method”. IEEE Transactions on Power Delivery, Vol. 3, No. 2,
Pg 476-489, April 1988.
[5] M.Wang, A.J.Wandermaar and K.D. Srivastava, “Condition Monitoring of
Transformers in Service by the Low Voltage Impulse Test Method”, High Voltage
Engineering Symposium, Conference Publication No 467, 22-27, Pg 1.45.S3-1.48.S3,
August 1999.
[6] J.A.S.B Jayasinghe, Z.D.Wang et al. “Winding Movement in Power Transformers: A
Comparison of FRA Measurement Connection Methods”, IEEE Transactions on
Dielectrics and Electrical Insulation, Vol.13,No.6, Pg 1342-1349, December 2006.
[7] Simon A. Ryder , Stefan Tenbohlen, “A Comparison of the Swept Frequency and
Impulse response methods for making Frequency Response Analysis Measurements”,
ALSTOM RESEARCH CENTRE.
[8] Essam Al-Ammar, George G.Karady and H.Jin Sim, “ Novel Technique to Improve
the Fault Detection Sensitivity in Transformer Impulse Test”, IEEE Transactions on
Power Delivery, Vol. 23, No. 2, Pg 717-725, April 2008.
[9] L Satish and A Saravanakumar, “Identification of Terminal Connection and System
function for Sensitive Frequency Response Measurement on Transformers”, IEEE
Transactions on Power Delivery, Vol. 23,No.2, Pg 742-750,April 2008.
[10] T.Leibfried, K.Feser, “Monitoring of Power Transformers using the Transfer Function
Method”, IEEE Transactions on Power Delivery, Vol. 14, No. 4, Pg 1333-1341,
October 1999.

57
APPENDIX

58