Frequency Response Analysis (FRA) for

Transformer Winding Deformation and

Movement:
From Measurement to Diagnosis
Presenters
• Dr Dahlina Sofian
Dr Dahlina Sofian graduated from the University of Manchester with a PhD in
Electrical Power Engineering specialising in “Transformer FRA Interpretation for
Detection of Winding Movement”. She had over 10 years’ experience with GE Power
Transformers where she was heavily involved in product standardisation, design
optimisation and diagnostic tests for large power transformers product line. She was
also part of the working group for IEC 60076-18 Measurement of frequency response.
She joined GE Grid Integration (HVDC) in 2018 and was involved in HVDC solution
standardisation and the development of next generation submodules for modular
multilevel converters. Dahlina has recently joined Scottish Power Renewables as Dr Dahlina Sofian
Senior Electrical Equipment Engineer, with continued interest in power transformers
design optimisation and diagnostics.

• Rafal Zaleski
Rafal Zaleski graduated from the Silesian University of Technology in 2002 with a
Master of Science degree. Following education, he started gaining his first practical
experience in Poland with Energo-Complex company. There, he was responsible for
transformer condition assessment through electrical testing and analysis on on-line
condition tests results. He joined Doble Engineering Company in 2009 as Senior Test
Engineer where he initially continued with transformer condition assessment quickly
expanding his skills across other substation apparatus. Nowadays he holds Principal
Engineer position within Doble and he is responsible for consultancy services related
to purchase of new-built transformers, eg. factory progress inspections, FAT
witnessing etc. He has been an active expert member for two CIGRE working groups
on FRA measurement and interpretations and has ample experiences in field tests Rafal Zaleski
using FRA technique.

2
Table of contents
• Part 1 (Dr Dahlina Sofian)
• Introduction
• Understanding the frequency response
Dr Dahlina Sofian

• Part 2 (Rafal Zaleski)

• Factors influencing the measurement
• Importance of good reference traces
• FRA examples Rafal Zaleski

3
 Dr Dahlina Sofian

Part 1
Introduction
Understanding the Frequency Response

4
Introduction
• Frequency Response Analysis (FRA) is  Comparative test – relies on
commonly used to assess the mechanical compared responses being
integrity of transformer active part measured in the same way and
in the same transformer
• Normally performed during the following: condition.
Tap winding with partial axial
• Factory fingerprint before transport to site collapse and inter-turn short-circuit
0

• Short circuit test

10 100 1000 10000 100000 1000000

-10

-20

• Site fingerprint/diagnostic after transportation prior to -30

commissioning

dB
Magnitude,dB
-40

Amplitude,
-50

• Relocation and installation -60

-70

• Routine diagnostic protocol -80

-90 Before fault

After fault
• After transformer alarm or trip -100
Frequency, Hz

• After through fault, lightning, seismic event

5
International Collaborations (2008 – 2020)
Cigre WG A2.26 IEEE WG C57.149 IEC PT 60076-18 Cigre A2.53

Further collaborations needed to share case studies towards effective and reliable diagnosis

6
Cigre WG A2.26

• Outlined various measurement

techniques and influencing factors that
could affect the interpretation.

• A chapter dedicated to FRA interpretation

includes some examples of frequency
response measurements and
diagnostics.

• Highlights on measurement
configurations which were later
implemented in IEEE and IEC standard
documents.

Highlighted measurement techniques and sensitivities of FRA in detecting problems

7
IEEE WG C57.149

• Presented measurement configurations

similar to those standardized by IEC but
with few differences recommended
configurations for a new set of
measurements.

• In IEEE, the end-to-end short-circuit

measurements are recommended in addition
to the open-circuit and new measurements be
carried out only with the load tap-changer
in extreme position, whereas IEC
recommends measurements with and without
the tap winding in circuit.

Further elaboration of the measurement technique and diagnostic case studies

8
IEC PT 60076-18

• First international standard on the

measurement of frequency response for power
transformers.

• Specifies the measurement method (connection

and configuration), the measuring equipment
and the measurement records.

Unified way of making FRA measurement with best practice and emphasis on repeatability

9
 Cigre A2.53

• Further emphasis on understanding of

frequency response and the factors
influencing the measurements
• Collection FRA case examples

• Reviewed literature of quantitative

assessment and assessed numerical
indices for FRA interpretation

Deep dive on understanding of frequency response and objective interpretation techniques

10
Back to Basic
IEC 60076-18 (2012): Measurement of frequency response
End-to-End End-to-End Short Circuit

Capacitive Inter-Winding Inductive Inter-Winding

Frequency response measurement circuit

Amplitude

The amplitude of the frequency response is the scalar ratio between the response signal
(Vout) and the reference voltage (Vin) (presented in dB) as a function of the frequency.

11
 Inductive Inter-Winding

Back to Basic
Typical frequency responses

End-to-End
LV winding response of generator transformer

Capacitive Inter-Winding

Series winding response of auto-transformers

Blue : Short Circuit, Red : Open Circuit

Each transformer has unique end-to-

end frequency response. However,
the frequency response follows
specific trend/pattern depending on
transformer design and construction.

12
 Back to Basic
What does each end-to-end frequency response mean?
0
Interaction
-10 Core between Winding structure Setup
windings • Frequency response from low to high frequencies
-20 is influenced by different components within power
-30
transformer
• Dependent on the interaction of electrical
Amplitude, dB

-40 parameters (inductance, capacitance and mutual

-50
couplings)
• When deformation/movement occurs, these
-60 parameters change and hence frequency
-70 response also change consequently
• “Knowing what and where to look” is the key for
-80 FRA interpretation strategy
-90 1 2 3 4 5 6 7
10 10 10 10 10 10 10
Frequency, Hz

13
Understanding of Frequency Response
Simple representation

R
Series LC
C
Resonance (maxima)

R L Parallel LC
Anti-resonance (minima)

Resonances and anti-resonances

superimposed along frequency response

14
Understanding of Frequency Response – Model Evolution
TB 812 (2020): Advances in the interpretation of transformer Frequency Response Analysis (FRA) R-L-C-M equivalent circuit
transformer modelling
LV HV LV HV
CgL/N CHL/N CgH/N

Mcore+Mair
Single Core Tank
phase two Lcore+Lair

Single winding NCsL NCsH

winding with transformer CHL/N
core and CgL/N CgH/N
Single mutual
winding with induction Influence of non-tested winding at mid-frequencies range
core

Single air-
core
winding
More complex transformer model with the following:
• Three-phase power transformers with star and
delta connected windings
• Additional windings (tertiary and tap windings)

n-stage lumped element LC ladder network model

Space coefficient 𝛼 = √𝐶𝑔 𝐶𝑠

15
Single air-core winding

16
Single winding with core

17
Single winding with core and mutual inductive

18
Single phase two winding transformer

19
 Identification of Unique Frequency Response Features
“U shape” Feature
Sensitivity studies on transformer model to replicate “U shape” feature

Magnitude (dB)

Magnitude (dB)
E

• ‘U shape’
between 1 kHz

Magnitude (dB)
Magnitude (dB)

50Ω
to 10 kHz V1 V3

• Delta connected
0.5CgL 0.5CHL 0.5CgH
HV windings
• Due high
CsL CsH
inductance/low Core LV HV Tank
series
capacitance

Z.D. Wang, B.Z. Cheng, P.A. Crossley, D. M. Sofian, J. Sanchez, “Fundamental Understanding of
0.5CgL V2 0.5CHL V4 0.5CgH
Frequency Response Analysis ‘U shape’ through Transformer Modelling”, CIGRE COLLOQUIUM
50Ω SCA2 /SCB2/ SCD1 India, November 22, 2019.

20
FRA as Comparative Test
Comparison with previous fingerprint (time based)
The best comparison as differences indicate a problem in a winding, but fingerprint is
often unavailable
Comparison with sister units (type-based)
Can be difficult as need to identify whether “normal differences” or a problem in a winding
Comparison with different phases (construction based)
Can be difficult as need to identify whether “normal differences” or a problem in a winding
“Normal differences” : Differences caused by factors other than problems in a
transformer winding, such as manufacturing tolerances, different lengths of
internal leads to LTC, winding inter-connections and slight variations in similar
design.

Knowing what to look for and where to look at

Changes in overall shape and numbers of peaks (maxima) and troughs (minima). Also,
shifts in resonant frequencies
• Requires knowledge of transformer structure and high-frequency performance of
transformers
• Differences may signify problems with transformer or problems with the
measurement/normal differences
21
Part 1 Conclusions

• Further collaborations needed to share case studies towards effective and reliable diagnosis

• It is important to ensure the measurements are being made in the same way and the
transformer is in the same condition for correct diagnosis – repeatability and consistent
recording

• Understanding of frequency response is important as changes may be due to genuine

problems or may be also due to other factors which will be highlighted further in Part 2 –
“Knowing what to look for and where to look at”

22
 Rafal Zaleski

Part 2
Factors influencing the measurement
Importance of good reference traces
FRA examples

23
 Factors Influencing the Measurement
It is important to ensure the measurements are being made in the same way and the transformer is in the
same condition for correct diagnosis – repeatability and consistent recording

Dr Dahlina Sofian

24
Effect of core magnetization

Low frequency variation is severe - but identifiable

25
Effect of tap position

Tap position is one of the most obvious factors influencing result

26
Nominal tap position (N)

(N) from (N-1) is not the same as (N) from (N+1)

27
Effect of state of other winding

Same phase LV traces at tap:

1, 6, 12 (nominal) and 13
In order from left

Impact of HV tap position on LV test

28
Effect of state of other winding

HV in 120kV

HV in 60kV

Impact of HV series/parallel switch on LV test

29
Effect of state of other winding

Delta closed

Delta opened

Impact of LV delta state on HV test

30
Effect of internally grounded delta winding

Impact of internally earthed delta winding on responses of other windings

31
Experiment with delta winding

Impact of state of delta winding on responses of other windings

32
Effect of insulating liquid
Effect of
transportation
bushings

With oil

Without oil

Measurement in fully assembled condition and in transportation condition

33
Effect of long GIS busbar

Measurement made through earth switch

Measurement made directly to bushing terminals

34
Effect of missing core earth

35
Effect of poor grounding

Can you spot difference between the two grounding braids?

36
Effect of poor grounding

…one was not correctly earthed

37
Importance of Good Reference Results

38
Importance of good reference results

The SFRA results not always give easy to interpret results eg. completely
aligned responses of three phases. Although in some cases experience
plays important role and allows to recognize certain situations,
sometimes additional reference results are extremely helpful in
interpretation.

Such reference can be usually taken from:

• Previous results
• Sister unit
It is important to obtain the same test setup as was used for reference
results in order to use them.

39
Importance of good reference results

Suspicious difference was measured at LV side. Sister transformer was available

to obtain reference but it was connected to HV cables.

40
Importance of good reference results

Small distribution
transformer

41
FRA Examples

42
Shorted turns

Let’s start with something easy.

750MVA, 380/230/13.8kV autotransformer YNay0d11, failed in service

43
Movement of Single Turns

So is this a good transformer? It tripped from service due to flashover across

contacts of A phase OLTC selector.

44
Movement of Single Turns

So is this a good transformer? It tripped from service due to flashover across

contacts of A phase OLTC selector.

45
Movement of Single Turns

So is this a good transformer? It tripped from service due to flashover across

contacts of A phase OLTC selector.

46
Movement of Single Turns

So is this a good transformer? It tripped from service due to flashover across

contacts of A phase OLTC selector.

47
Movement of Single Turns

Luckily it was possible to see damage through inspection hole.

Scrapping confirmed slight winding deformation on A phase

48
Axial Collapse
Initial symptoms
 35 year old 400/132 kV 240 MVA auto-transformer
 10:53 a.m. Monday 5th November 2001
 Buchholz alarm
 Transformer switched out for investigation
 Analysis of Buchholz gas and main tank oil indicated serious fault
 No abnormal system events
 Transformer was operating at 80% loading

Hardly non-conclusive electrical test results, but…

49
Axial Collapse

50
Axial Collapse

Previous results

51
Axial Collapse

Sister unit failed in service

52
Axial Collapse

53
Buckling Deformation

Transformer details:
 375/750MVA ONAN/ODAF
 400/275/13kV autotransformer
 No taps
 Built 1966
 Deterioration in DGA results
 Removed from service for investigation

54
Buckling Deformation
20

One of gas increases 18

coincided with CT failure

16

120 14

12

10
100

80
4

60 0
18/02/1982 17/02/1985 17/02/1988 16/02/1991 15/02/1994 14/02/1997 14/02/2000 13/02/2003 12/02/2006 11/02/2009

C2H4 C2H2

40

20

0
18/02/1982 17/02/1985 17/02/1988 16/02/1991 15/02/1994 14/02/1997 14/02/2000 13/02/2003 12/02/2006 11/02/2009

H2 CH4 C2H4 C2H6 C2H2

DGA showed three potentially damaging events

55
Buckling Deformation
T4673
A phase B phase C phase

Main windi ngs 4,405 pF 3,634 pF 4,294 pF

to earth, CH 0.51 % 0.56 % 0.52 % Sister
Main windi ngs 8,786 pF 8,711 pF 8,525 pF transformer
to tertiary, CHT 0.55 % 0.56 % 0.56 % Suspect
Tertiary wi nding 17,924 pF transformer
to earth, C T 0.52 %

T4971
A phase B phase C phase

Main windings 4,246 pF 3,368 pF 4,273 pF

to earth, CH 0.75 % 0.51 % 0.62 %

Main windings 8,277 pF 5,775 pF 9,030 pF

to tertiary, CHT 0.42 % 1.91 % 0.54 %

Tertiary winding 20,668 pF

to earth, CT 0.46 %

Significant difference in capacitance and power factor were noted for phase B

56
Buckling Deformation

Also a significant difference in frequency response was noted for phase B

57
Buckling Deformation

58
Tap Winding Movement
 275/33kV 120MVA YNd1 transformer feeding steelworks suffered flashover
between taps while on ‘hot standby’
 Barrier board between tap-changer and main tank broken
 SFRA tests indicated movement for tap winding:
 Difference for Blue phase at Tap 1 (taps ‘all-in’), but not at Tap 10 ( ‘all-out’)
 Compared responses with sister reference transformer
 Note faulty transformer was without oil, while reference was oil filled
 Internal inspection confirmed conductor tilting for Blue phase tap winding
 Special SFRA Tests showed possibility of internal resonant over-voltage
between taps involved in flashover for switching surge at HV line terminal

59
Tap 10 HV SFRA for faulty transformer

60
Tap 10 HV SFRA for reference transformer

61
Tap 1 HV SFRA for faulty transformer

62
Tap 1 HV SFRA for reference transformer

63
Broken tap-changer barrier board

64
Flashover damage between Blue phase tap leads

65
Internal View

66
Part 2 Conclusions

67
Part 2 Conclusions

 Reference results if available, can greatly support interpretation

of results. This can help either to confirm winding damage or to
clear doubts regarding suspicious traces
 It is important to be aware of number of different factors which
can affect the measurement, eg. tap position, state of stabilizing
winding. Results obtained in different conditions may not make
good reference.
 Little difference can still mean dangerous deformation but at the
same time, large differences can be caused by factors other than
winding damage. This makes interpretation more difficult.

68
Questions

69