Sweep Frequency Response Analysis SFRA Training

Sweep Frequency Response Analysis
SFRA Training
2014 International Conference of Doble Clients
April 6 - 11, 2014
©2014 Doble Engineering Company. All Rights Reserved

DISCUSSION OUTLINE
I. Basic idea
II. Objective
III. Test setup and methodology
IV. Basic features of SFRA trace
V. Data analysis
- factors to consider
- acceptance criteria
©2014 Doble Engineering Company. All Rights Reserved 2
WHAT IS SFRA?
• SFRA is a tool that can give us an indication of winding

movement or electrical short circuits in transformers
• SFRA can tell you if something is damaged or broken inside

the transformer without climbing inside it.
• SFRA is a means to provide data about the transformer

mechanical integrity
TEST SETUP
The most useful measurement is the response of an individual winding. This is

accomplished by injecting a signal of known frequency into one end of the winding and
measuring the response at the other end, followed by a sweep through other frequencies.
Open circuit test
Vin Delta-Wye
Vout H1-H3 All other bushings
H2-H1 float
Response H3-H2
Source X1-X0
Reference X2-X0 Short circuit test
Ground X3-X0
H1-H3 X1X2X3X1
H2-H1 shorted, others
H3-H2 float

SFRA TEST DEFINITIONS
• Open-Circuit (OC) Tests - commonly referred to as HVOC,

LVOC, TVOC – ex. H1- H2, all other terminals floating
• Short-Circuit (SC or HVSC) tests - same connections as the

HVOC but with the secondary shorted (neutral floating)

ex. High Voltage winding - open circuit tests on a transformer

OBJECTIVE – WHEN AND WHY TO DO THIS TEST
Performed at the factory for:
• providing benchmark for future measurements

• complying with customer specification
• post short-circuit test evaluation
• quality assurance

OBJECTIVE – WHEN AND WHY TO DO THIS TEST
Performed in the field:

• diagnostic after transportation prior to commissioning
• after relocation and installation or as part of benchmarking (!)
• as part of a routine diagnostic protocol
• after transformer alarm (e.g., sudden pressure, on-line
monitoring)
• after significant change in off-line diagnostic data (e.g., C, X L)
• after through fault, lightning, seismic event 8
SFRA measurements have shown the following:
• Radial Deformation - Hoop Buckling

• Winding that have telescoped
• Turn-to-Turn short circuits in the winding
• Open circuited winding
• High resistance connections
• Bulk and local movement

SFRA is just one member of the diagnostic team
Power Factor Exciting

and currents
Capacitance
Leakage
PD
Reactance
DGA/Oil
Turns
Screen
ratio

A bit of FRA HISTORY
• 1960: W. Lech & L. Tyminski, Poland, propose method for

detecting transformer winding deformation
• 1976: First Doble Paper: A.G. Richenbacker “Frequency Domain
Analysis of Responses from L.V.I. Testing of Power Transformers”
• 1978: E.P. Dick & C.C. Ervin, Canada, first IEEE paper
“Transformer Diagnostic Testing by FRA”
• 1980’s: European FRA Field Trials
• 1990’s: FRA Application Development

DOBLE & FRA STANDARDS
• 1999 - EuroDoble writes FRA Specification

• 2000: FRA testing commercialized
• 2008 CIGRE publishes Brochure #342 on FRA

• 2012 IEC FRA Standard 60076-18 published
• 2013 IEEE FRA Guide C57-149 is published
• Doble Engineering is involved in writing these standards

WHY IS IT CALLED S-FRA?
• Early research used a technique called the impulse method

• Field trials showed this to be difficult to perform in the field and obtain
consistent repeatable results
• The industry has since settled on the Sweep method
• As stated earlier this is done by injecting a signal of known frequency
into one end of the winding and measuring the response at the other
end, followed by a sweep through other frequencies

SYSTEM ANALYSIS
SFRA
RLC networks, dB’s & Hertz

BASIC IDEA
Frequency Response Measurement
Big Grey Box

Input
Output

OBJECTIVE
Frequency response analysis (sweep method) looks at how well a transformer winding transmits a
low-voltage signal at each frequency by evaluating a ratio of the transmitted voltage (output) to the
applied voltage (input) in dB
Applied signal Vin

Measurement of
applied signal Vin Measurement of
transmitted signal Vout
The signal transmission depends on winding impedance, which, at each frequency, represents a
different path for the signal. Since its capacitive and inductive elements are related to the transformer
construction, changes in the response indicate changes in geometry.
SYSTEM ANALYSIS
The impedance is a distributed network of electrical components which are

passive in nature and can be modeled as resistors, inductors and capacitors.
• In a passive device there are three basic components:
– resistors
– capacitors
– inductors
• They each have a different response to an AC signal

Transformer equivalent circuit model
HV Ths
CH CH CH
Winding CT CT
RH LH RH LH
Inter
CHL CHL CHL
Winding
RL LL RL LL
CL CT CL CT CL
LV
Winding

BASIC IDEA
Frequency Response
Input Output
Measure the Frequency Response of passive elements; Resistors, Inductors

and Capacitors (RLC) for the Apparatus and plot the results.
Result = called the Transfer Function, Bode plot, footprint

Definition – Transfer Function
A transfer function is an expression of the ratio of

the output of a system with respect to its input (i.e.
the gain of the system)
• expressed relative to frequency in dB
Response in dB = 20log10(Vout/Vin)

SYSTEM ANALYSIS - RLC NETWORKS
• R, L, C components exist in many forms inside a transformer
• L and C components contribute to an individual resonant

frequency
• L and C components have strong geometric dependence and

are affected by construction and placement details.
RLC NETWORK- BASIC IDEA
Response of ideal resistors - no inductance or capacitance present
Response, dB 50  Response
-25
R
Vin Vout 500 
Response
Z=R
-50
100 1010 Frequency, Hz
Flat across the frequency range
Response of an Inductor L:
0
L
Vin Vout -20
-40
Z=2fL
-60
-80
Impedance increases with frequency
-100
- dead short at low frequency
100 1010
Response of the Capacitor C:
C 0
Vin Vout
-20
-40
1
Z= -60
2fC
-80
-100
Reduced impedance with increased frequency
- open circuit at low frequency 100 1010
Response of the parallel RLC Circuit:
0
Vin Vout -20 Resonance
-40
-60 Inductive Capacitive
-80 Response Response
-100
100 1010
BASIC FEATURES OF AN SFRA TRACE
Resonant frequency (peak or valley)

depends only on the L and C values
Im pe d a nc e , O h m s
5
Resistance (R)
4
affects size of
3
2 resonance peak
1
0
0. 1 1 10 100 1000
Fre que ncy, Hz

SUMMARY - RLC NETWORKS FIX THIS
• We get a resonance (peak or valley) for every LC combination.
• Changing L or C gives us a new resonance, since L and C are

dependent on geometry any change will result in a new picture
• Changing the resistance (R) only affects the size of

the resonance

What’s in the picture?
• A typical SFRA trace is a magnitude plot of the transfer

function of a piece of power distribution equipment
• The X axis (horizontal) is frequency (Hz) typically displayed

in a logarithmic scale
• The Y axis (vertical) is in decibels (dB) and displays the ratio

of the amplitude of output voltage to input voltage
High Voltage Winding Open Circuit traces
Attenuation or gain in Decibels (dB)
Frequency in Hertz

Definitions - LOG SCALE and dB’s
The X axis is Frequency, the measurement unit is Hertz (Hz) and is plotted in a
logarithmic scale.
The increments are commonly referred to as decades, example: 1, 10, 100, 1k, 10k,
etc.
The Y axis is decibels (dB) and is a logarithmic unit of measurement that expresses
the magnitude of a physical quantity.

Using a Logarithm Scale
10? = Y
Log base 10 of Y is the power that 10 should be raised to, in
order to get Y: log10Y = X
log1010 = 1 101 = 10
log10100 = 2 102 = 100
log100.01 = -2 10-2 = 1/102 = 0.01
log101 = 0 100 = 1
BASIC FEATURES OF AN SFRA TRACE (cont.)
Vout = 10Vin 20log10(10Vin/Vin) = 20(1) = 20 dB

Vout = Vin 20log10(Vin/Vin) = 20(0) = 0 dB
Vout = 0.1Vin 20log10(0.1Vin/Vin) = 20(-1) = -20 dB
Each drop of 20 dB corresponds

to 0.1Vout_previous
WHY USE THE DECIBEL? - dB
The bel (symbol B) was used mostly in telecommunication, analog electronics,

and acoustics.
Invented by engineers at the Bell Telephone Labs it was used to quantify the
reduction in audio level over a 1 mile length of standard telephone cable
The bel is too large, so the decibel (dB), is more commonly used.
Presentation of data on a logarithmic scale can be helpful when the data covers
a large range of values
Using a log scale reduces this to a more manageable range
TRANSFORMER TEST SET UP
• Transformer should be disconnected from system
• Transformer should be in normal service condition (assembled, oil-

filled) - make notes if special conditions exist, empty, transit
bushings, dual windings, etc. This will save on confusion later
• The transformer tank shall be grounded
• Make connections as determined by winding configuration. Phases

not under test are usually left floating.
TEST SETUP
All core(s) and tank grounds, and terminal(s) of a buried winding grounded during
normal operation should be grounded during testing.
Open circuit test

Vin Delta-Wye
Vout H1-H3 All other bushings
H2-H1 float
Response H3-H2
Source X1-X0
Reference X2-X0 Short circuit test
Ground X3-X0
H1-H3 X1X2X3X1
H2-H1 shorted, others
H3-H2 float

Doble SFRA TEST – CONNECTIONS
Red Lead: Source & Reference (Vi) Three Lead Test
Green clip:
Ground
Black Lead: Response (Vo)

SFRA Three Lead Measurement and Grounding
• We need to reference each measurement to ground

V signal V in V out
Im pedance, Z
50  C oax
50  C oax 50  C oax
Signal L ead T est
R eference
M easurem ent M easurem ent
L ead L ead
Which means we get a consistent measurement by minimizing signal

reflection and reducing the influence of the coax leads
SFRA DATA ANALYSIS
Experience tells us:
Certain Frequency Bands can Indicate Different

potential Problem Conditions
Think of the Orchestra..........

TRANSFORMER CONSTRUCTION

IEC STANDARD 60076-18 FRA Characteristics
Region Frequency Band Components

1 0 – 2 kHz Main Core
2 2 kHz to 20 kHz Interaction between windings
3 20kHz to 1 MHz Main Winding structure
4 >1 MHz Coil Leads, Grounding, bushing influence
Every transformer is different, these are general guidelines

SFRA SUGGESTED REGIONS OF INFLUENCE
<10kHz: Open circuits, 150 - 400kHz:

shorted turns, Main windings
residual magnetism
10kHz to 150kHz: 400kHz - 1MHz

Tap windings and
Interaction between
windings leads, grounding
48
DATA ANALYSIS
• Know what to expect

• Typical Results
• Here are some examples

Wye Delta HV Winding Open Circuit tests - HVOC
For outer phases: two resonant

points in low-frequency range
reflect two different flux paths in
the core; thus, two magnetizing
reactances, each having its own
resonance.
Typical center
phase variation Center phase: one resonant point in low-frequency range
reflects two equal paths of the magnetic flux; therefore,
two equal magnetizing reactances have a resonance at
the same frequency.

LV WINDING OPEN CIRCUIT TESTS - LVOC
Middle phase differs from the outer

phases due to a different magnetic
flux core path.

HV Short Circuit test – HVSC ( LV windings shorted)

High Voltage Short Circuit Tests - HVSC
Short-circuit tests are used to look at the HV winding without

the influence of the core.
The transformer core’s reluctance circuit influences the low

frequencies. To analyze the without the influence of the core,
the secondary windings are short circuited.
This test is is similar to the leakage reactance test.

DELTA-WYE HV OPEN CIRCUIT TESTS

Delta-Wye: LV Open Circuit Tests- LVOC

AUTOTRANSFORMER HV OPEN CIRCUIT TESTS

AUTOTRANSFORMER LV OPEN CIRCUIT TESTS

Autotransformer Delta tertiary open circuit tests

LVOC TESTS - VARIATION WITH LTC POSITION
Low Frequency area - all tests done on the same phase
16 Lower
8 Lower

HVOC TESTS - VARIATION WITH DETC POSITION
Low Frequency area - all tests done on the same phase
Position A Position E

LVOC test - unit tested with and without oil - same phase
Transformer
filled with oil Without oil
Lower resonant Higher
frequencies resonant
frequencies
HVOC same phase - effect of poor measurement connection
Original
Next date
poor ground
HVOC test same phase - effect of core magnetization
Low frequency variation is severe - but identifiable
Magnetization affects the

core response in the low
frequency area

DATA ANALYSIS: FACTORS TO CONSIDER (cont.)
The low-frequency range of the trace (e.g., < 2 kHz) is influenced by the magnetic state
of the core. Between two test dates, this state can be changed by application and then
removal of ac/dc excitation and by a subsequent relaxation of the core.
Factory: 10/7/2009
Field: 12/16/2009
LVOC: Impact of magnetic

state of the core
OC Analysis - effects of residual magnetism
• If core is magnetized, it should only affect the Low Frequency results
• This does not negate test results, but the impact needs to be
considered
• If necessary, Demagnetize and retest the transformer
• Problem may be confirmed by comparing the results to the to

Exciting Current tests
HVOC PHASE COMPARISON FACTORY TO FIELD
Two different
Factory Test
bushing types
Full of oil
Field Test no oil
In the factory they used standard bushings, in the field they had transit
bushings, this would be evident in the higher frequencies
DATA ANALYSIS Change in HV can affect the LV response
Impact of HV series/par switch on

LVOC test
HV in 120 kV
HV in 60 kV

DATA ANALYSIS: Change in HV can affect the LV response
Impact of HV switch
on LVOC test
HV in 161 kV
HV in 138 kV
HV in 115 kV
30MVA, 161138115-13.2GrdY kV

DATA ANALYSIS: Effect of buried tertiary ground position
X1-X0
 open &  closed &
ungrounded ungrounded
Impact of open/closed
ungrounded buried 
H1-H0
7.5MVA, 115GrdY-4.16GrdY kV with buried 

DATA ANALYSIS Effect of buried tertiary ground position
 closed &
X1-X0 grounded
 closed &
ungrounded
Impact of gnd/ ungnd buried 
H1-H0
 closed &
grounded
7.5MVA, 115GrdY-4.16GrdY kV with buried 

HVSC Test – Effect of poor shorting leads on the LV winding

DATA ANALYSIS: Main Core Ground Connection
Unit: 11.2 MVA, 69GrdY-13.2GrdY kV
Ungrounded
Grounded
HVOC: Impact of ungrounded core on HVOC test.

CL capacitance increased by 1000 pF.
DATA ANALYSIS: HVOC: Impact of Grounded X0
Unit: 11.2 MVA, 69GrdY-13.2GrdY kV
Grounded
Ungrounded

DATA ANALYSIS: LVOC: Impact of LV LTC position
bridging
non-bridging
Unit: 12 MVA, 45-13.8GrdY kV

DATA ANALYSIS: FACTORS TO CONSIDER (cont.)
A few more notes:
• It is essential to consider the factors (that may influence the results) not only for
the analyzed data, but also for the data used as the reference, be it data obtained
in the field or at the factory.
• Connected bus and measuring cable extensions can impact the data.
• It is important to consider the operational/ repair/maintenance history of the unit,
e.g., general repairs affecting the inner geometry, bushing replacement, changing
position of internal links (series versus parallel), past through faults, etc.
• These changes require a new benchmark.

DATA ANALYSIS: ACCEPTANCE CRITERIA
Two strategies can be considered:
1. Comparison with reference data:

a) Results of previous tests on the same unit (e.g., factory or initial field
data) – best.
b) Results of tests on a sister unit – beware! not all twins are identical.
c) Results of tests on other phases – frequently, phase traces of open-
circuit test differ, especially on the middle phase.
2. Identification of typical characteristics

DATA ANALYSIS: Comparison to Previous data
1a) Ideally, when all factors (other than defects) are eliminated, the comparison should result in a
near perfect overlays. However, small variations between successive measurements should be
expected. Stray impedances at contact points and induced voltages from the environment will be
different. The “feel” for the extent of these variations comes with experience in observing the data.
Difference in
Difference in core
grounding Region of interest
circuit
magnetic state
All six phase traces should

H1-H3 look like one line
Two
Open-circuit test
traces Short-circuit test
1a) A Change of the magnitude at the resonance points, without changing the frequency or the
shape of the trace, is a resistive change associated with changes in conductor or test cable
connections or changes in temperature. They are typically, not seen as a problem.
15 MVA, 69-13.2GrdY kV

1a) However Once the data is confirmed as credible, analysis, for the most part,
relies on several key indicators:
• Did the picture change?
• Is there an appearance of a new resonance?
• Disappearance of an existing resonance
• Shift in a resonant frequency (significant above low-frequency range).

1a) Example: hoop buckling Shift in resonant
frequency in one
phase

1a) The impact of a shorted turn on the open-circuit test is similar to one created by the short-
circuited winding in the short-circuit test, i.e., it eliminates the influence of the magnetizing
impedance thus making the open-circuit test trace resemble the short-circuit test trace.
Open-circuit test trace in the

presence of a shorted turn
Normal open-circuit test trace

DATA ANALYSIS: Comparison to similar units
1b) Example: 18 LVOC on 6 sister units
Unit: 30 MVA, Auto, 115GrdY-69GrdY-13.2 kV

DATA ANALYSIS: Comparison to similar units
1b) Example: 18 short-circuit tests on 6 sister units

Unit: 30 MVA, Auto, 115GrdY-69GrdY-13.2 kV
0.1 dB

DATA ANALYSIS: Comparison of phases in the same unit
1c) Example: Three HVOC traces

Unit: 15 MVA, 67-34.5GrdY kV

DATA ANALYSIS: 2) Typical Characteristics HVOC-LVOC
LVOC
HVOC
HV side, due to a higher

impedance, has a higher (-) dB
For outer phases: two resonant than LV side.
points in low-frequency range
Middle differs from the outer phases
reflect two different flux paths in
due to a different magnetic flux core
the core; thus, two magnetizing
path.
reactances, each having its own
resonance.
DATA ANALYSIS: 2) Typical Characteristics HVSC
Review of much of empirical data shows that deviation between short-circuit test traces of each phase in the range <10 kHz is typically
< 0.2 dB
HVSC Acceptance Criteria
In the range <8-10 kHz, all three

phases are very close, reflecting
the similarities between phase
leakage channels, i.e., the space
within and between the windings.

DATA ANALYSIS 2) Typical Characteristics HVSC
A review of empirical data shows that deviation between short-circuit

test traces of each phase in the range <10 kHz is typically < 0.2 dB.

TYPICAL RESULTS - CONCLUSIONS
• Results can vary between units – depends on size and type of unit
• Results can be different with magnetization & grounding effects
• Variations with tap changer positions
• Change based on the oil level
• Bushing configuration will change response
• When a diagnostically significant change is detected, a confirmation
should be looked for in data obtained via other measuring techniques: XL ,
Iex and loss, C, Rdc.
DATA ANALYSIS: FACTORS TO CONSIDER
If there is a change in test data,

trust no one: the tester is guilty
until proven innocent!

Sweep Frequency Response Analysis SFRA Training

Uploaded by

Copyright:

Available Formats

Sweep Frequency Response Analysis SFRA Training

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sweep Frequency Response Analysis SFRA Training

Uploaded by

Copyright:

Available Formats

What is SFRA and what does it test for?

What is SFRA and what does it test for?

What are the different types of SFRA tests?

What are the different types of SFRA tests?

Sweep Frequency Response Analysis

©2014 Doble Engineering Company. All Rights Reserved

• SFRA is a tool that can give us an indication of winding

• SFRA can tell you if something is damaged or broken inside

• SFRA is a means to provide data about the transformer

The most useful measurement is the response of an individual winding. This is

©2014 Doble Engineering Company. All Rights Reserved 4

• Open-Circuit (OC) Tests - commonly referred to as HVOC,

• Short-Circuit (SC or HVSC) tests - same connections as the

©2014 Doble Engineering Company. All Rights Reserved 5

©2014 Doble Engineering Company. All Rights Reserved 6

Performed at the factory for:

• providing benchmark for future measurements

©2014 Doble Engineering Company. All Rights Reserved 7

Performed in the field:

• Radial Deformation - Hoop Buckling

©2014 Doble Engineering Company. All Rights Reserved 9

Power Factor Exciting

©2014 Doble Engineering Company. All Rights Reserved 10

• 1960: W. Lech & L. Tyminski, Poland, propose method for

©2014 Doble Engineering Company. All Rights Reserved 11

• 1999 - EuroDoble writes FRA Specification

• 2008 CIGRE publishes Brochure #342 on FRA

• Doble Engineering is involved in writing these standards

• Early research used a technique called the impulse method

©2014 Doble Engineering Company. All Rights Reserved 13

©2014 Doble Engineering Company. All Rights Reserved 14

Frequency Response Measurement

Big Grey Box

©2014 Doble Engineering Company. All Rights Reserved 15

Applied signal Vin

The impedance is a distributed network of electrical components which are

• In a passive device there are three basic components:

• They each have a different response to an AC signal

©2014 Doble Engineering Company. All Rights Reserved 18

Measure the Frequency Response of passive elements; Resistors, Inductors

©2014 Doble Engineering Company. All Rights Reserved 19

A transfer function is an expression of the ratio of

©2014 Doble Engineering Company. All Rights Reserved 22

• R, L, C components exist in many forms inside a transformer

• L and C components contribute to an individual resonant

• L and C components have strong geometric dependence and

Response of ideal resistors - no inductance or capacitance present

Response of the Capacitor C:

Response of the parallel RLC Circuit:

Resonant frequency (peak or valley)

©2014 Doble Engineering Company. All Rights Reserved 32

• We get a resonance (peak or valley) for every LC combination.

• Changing L or C gives us a new resonance, since L and C are

• Changing the resistance (R) only affects the size of

©2014 Doble Engineering Company. All Rights Reserved 33

• A typical SFRA trace is a magnitude plot of the transfer

• The X axis (horizontal) is frequency (Hz) typically displayed

• The Y axis (vertical) is in decibels (dB) and displays the ratio

Attenuation or gain in Decibels (dB)