Together We Power

The World
SFRA Training
SFRA History
Initial research started in the 60s.
In the 80s Engineers mainly used HP
network analyzers
First generation of SFRA purpose built test
sets developed by Doble in 1990.
Now SFRA technique has gained world
wide acceptance with several
manufacturers producing dedicated
instruments for SFRA
SFRA History
Today
Dobles SFRA has become the de facto
SFRA standard in North America and
wordwide with > 700 Instruments sold
Doble has been cataloging results for ten
years
Doble has developed standardized SFRA
procedures for IEEE & CIGRE
Doble develops support network for
SFRA analysis and interpretation
National Grid FRA experience
started evaluating FRA in late 1980s
initially used an impulse technique (KEMA digital
LVI)
changed to swept frequency ( Ontario Hydro )
hundreds of tests
several examples of failures
standardised test procedure ( Euro Doble )
Method now used by many other utilities in
Europe, Far East, Australia, North America
Support: Integration
SFRA is one tool in the tool box
SFRA
Capacitance DC Resistance
Exciting
Currents
Leakage
Reactance
OC: Open Circuit SC: Open Circuit
Together We Power
The World
SFRA Training
Why Use SFRA?
SFRA is a test set for transformers, reactors, and
rotating machines
SFRA can tell you if anything is damaged or broken
inside your transformer without going inside.
SFRA is used alongside other electrical tests such
as the M4000 and oil analysis
Doble SFRA summary
Transformer Decisions
There are many decisions relating to transformers:
Is it safe to go back in service after a fault?
Has it been damaged in transit?
Is it deteriorating as it ages?
Will it fail unexpectedly?
Decision making for transformers is not easy
Quality data and engineers support good decisions
Transformer Data
Data may relate to different transformer areas:
Dielectric ability to carry voltage
Mechanical ability to carry current
Thermal ability to sustain power transfer
Field Testing provides the engineer with a
group of tools to assess the condition of a
transformer in each area
Engineer should choose the right tool for the job
Which Transformer Tests?
Tests depend on the decision you need to
make!
Motivation: SFRA When & Why?
Acceptance
Establish a Baseline
Assess Condition after Electrical Disturbance
Assess Condition after a Relocation
Asset Management - Ranking and Prioritizing
Reduce Catastrophic Failures
Winding technology hasnt changed much in 100
years
Available tools have changed a great deal
Why Do Things Move?
Mechanical Shock during Shipping
Loosening of Internal Structural
Components loss of clamping
pressure
Fault Duty
Mechanical Failure
Mechanical Failure
Failure Mode
The very large electromagnetic forces on windings
during fault conditions can cause winding movement
and even permanent winding and core deformation
This may result in insulation damage.
Turn to turn faults are the most likely.
Insulation damage usually results in a failure of the
transformer which is uneconomic to repair.
Design issues
Transformers should be designed to withstand through
faults.
However, transformers are rarely short-circuit tested
because of the costs involved.
So the short-circuit strengths of designs are not often
checked.
Many new transformers fail short circuit tests, so the short-
circuit strength of many designs must be suspect.
Design limitations
Design calculations dont take into account the effects of
twisting forces.
Designers cant design against tap to tap faults.
Tap windings are difficult to design for short-circuit
strength.
Specification
Impedance
Short-circuit current
Short-circuit forces
20 % 5 %
5 p.u. 20 p.u.
1 : 16
Detection
Dissolved gas analysis (DGA) will only indicate
a problem when the insulation has been damaged
(usually too late to repair).
Internal visual inspections often inconclusive
Winding failures can usually be diagnosed by
various electrical tests ( but not always ).
But latent damage is much harder to detect.
So your transformer may be critically damaged
without you being aware of it !
Service
The number and severity of short-circuit events
suffered are important.
Close-up lightning strikes are a common cause
of winding movement failures.
Tap-changer faults can cause tap winding
failures.
Faulty synchronisation can cause winding
damage and failure.
Ageing
As a transformer ages the insulation shrinks
and clamping pressure is lost.
- reduces strength.
Any minor winding deformation usually
results in mis-alignment of electromagnetic
centres - increases stresses during
subsequent faults.
Failure Mode
Hoop buckling of inner
winding
Conductor tipping
Conductor telescoping
Coil clamping failure
End insulation collapse
Spiral tightening
Lead displacement
SFRA 2009
Bushing failure.
This bushing
failed
catastrophically
This bushing
was hit by
porcelain and
failed
Neutral bushing lifted from
turret and is no longer vertical
SFRA 2009
Bushing.
But did the
windings move
because of the
pressure wave
in the oil?
SFRAshowed
that the
windings were
in good shape.
Transformer impact.
Or maybe we hit a
bridge?
Bridge Impact
Paint scrapes on delivery are suspicious?
How did the
scrapes get
there, and what
does the impact
recorder say?
Do we know if anything moved
inside the transformer?
t
Transport Issues
Transformer Arrives by Barge
Almost!
Symptoms - black box scenario
How can we tell if anything has moved?
SFRA as a transformer test
Sweep Frequency Response Analysis
A method to measure the frequency responseof the
passive elements (RLC) of a transformer.
The result is a transfer functionwhich produces a
fingerprint related to the mechanical geometryof
the transformer.
SFRA relates to Mechanical Integrity
SFRA What do you get?
Core
Tap windings
Main winding
Tap leads
Conclusions
SFRA is a means to provide data about
transformer mechanical integrity
Good results lead to good decisions
Doble support means you can extract
value from your SFRA measurements
Together We Power
The World
SFRA Training
Introduction:
RLC, dB & MHz
SFRA Theory and Practice
In a passive device there are three basic
components:
resistors
capacitors
inductors
They each have a different response to an AC
signal
Their response is closely related to their
geometry: both internal and in relation to other
components
FRA theory
What is a winding ?
Inductance
Series capacitance
Shunt capacitance
High frequency model
Winding
inductance
series capacitance ( turn to turn )
shunt capacitance ( turn to earth )
winding geometry determines values of L and
C
Winding frequency responses
windings have frequency dependent response
for transmitted signals
the detailed form of the frequency response
depends on winding geometry
a change in geometry will change the
frequency response
FRA theory
If we can measure changes in the frequency
response of a winding caused by winding
movement,
then we can detect winding movement
SFRA Theory and Practice
0
50
100
150
200
250
300
350
400
450
0.1 1 10 100 1000
Frequency, Hz
I
m
p
e
d
a
n
c
e
,

O
h
m
s
Resistance:
flat response v. frequency
Inductor:
Increased impedance with
increased frequency; dead
short at low frequency
Capacitor:
Reduced impedance with increased frequency; open circuit at low
frequency
Impedance of an ideal resistor, capacitor and inductor
SFRA Theory and Practice
dBs: as impedance increases, V
out
falls
Impedance, Z
V
out
V
in
Response in dBs =20 log
10
(V
out
/V
in
)
Each 20 dB drop means we are looking at
a tenth of the previous V
out
/V
in
SFRA Theory and Practice
Each 20 dB drop means we are looking at a
tenth of the previous V
out
/V
in
Response in dBs =20 log
10
(V
out
/V
in
)
0
+20
-20
-60
-40
dBs
V
out
= V
in
V
out
= 0.1 * V
in
V
out
= 0.01 * V
in
V
out
= 0.001 * V
in
V
out
= 10 * V
in
SFRA Theory and Practice
Need to reference each measurement to ground
I mpedance, Z
50 O Coax
Test
Measurement
L ead
V
out
V
in
50 O Coax
Reference
Measurement
Lead
50 O Coax
Signal L ead
V
signal
Means we get a consistent measurement
SFRA Theory and Practice
Response of a short circuit
Frequency, Hz
Response, dB
0
-50
-25 Response
0 dB across the frequency range
SFRA Theory and Practice
Response of ideal resistors - no inductance or
capacitance present
0
-25
500 O
Response
Frequency, Hz
Response, dB
-50
50 O Response
Flat across the frequency range
SFRA Theory and Practice
Response of an ideal inductor
0 dB down at
low frequency
means it looks
like a dead
short
Inductive
roll off
Larger inductances
start to roll off at lower
frequencies
SFRA Theory and Practice
Response of ideal capacitor
Low frequency
response is like
an open circuit
0 dB down at
high
frequency
like a dead
short
Capacitive
climb back
Knee point depends
on size of capacitor
Parallel RLC Circuit:
SFRA Theory and Practice
Dead short at
low frequency
Constant dBs
down v. frequency
Open circuit at
high frequency
Dead short at
high frequency
Open circuit at
low frequency
Impedance of the Parallel RLC Circuit:
SFRA Theory and Practice
0
1
2
3
4
5
6
0.1 1 10 100 1000
Frequency, Hz
I
m
p
e
d
a
n
c
e
,

O
h
m
s
Resonant frequency depends
only on L and C values
R affects size of
resonance peak
Response of the RLC Circuit:
SFRA Theory and Practice
Inductive roll off
Capacitive
climb back
Resonance
Response of Multiple Circuits:
Resonance ?
4. Understanding SFRA - Basic Principle
0
-20
-40
-60
-80
-100
10
0
10
10
Resonance ?
V
in
V
out
System 1
System 2
Resonance ?
We get a resonance for an inductor-capacitor
(LC) combination
Changing L or changing C gives a new
resonance
L and C are dependent on geometry
Changing R changes the size of the resonance
SFRA Theory and Practice
C
H
C
H
C
H
C
T
C
T
R
H
L
H
R
H
R
L
L
L
C
H
L
C
H
L
C
H
L
C
L
C
L
C
L
C
T
C
T
HV
Winding
LV
Winding
Inter
Winding
L
H
R
L
L
L
Real World Measurements
A real
transformer has
many
inductance and
capacitances
Each LC pair
gives a
resonance
SFRA Theory and Practice
Need to reference each measurement to ground
I mpedance, Z
50 O Coax
Test
Measurement
L ead
V
out
V
in
50 O Coax
Reference
Measurement
Lead
50 O Coax
Signal L ead
V
signal
Means we get a consistent measurement
SFRA is really many measurements
Some individual resonances
Hardware Test Leads
Three lead system
Signal & Reference
Leads
M5x00
Measurement
lead
Signal + Reference
Measure
Ground
Ground Signal
5. SFRA M5000 Series
M5300
M5200
M5400
5. M5000 Test Cable
At Test Set - Color Coded Leads
Yellow - Signal (Gives Vin)
Red- Reference (Measures Vin)
Black - Test (Measure Vout)
At Transformer
Red=> Vin
Black => Vout
Lead Groundsto Base of Bushings
Safety Ground- Twist Lock to Transformer
A note about Test Leads
60 ft /18 m
Cable Trunk 30 ft to the split
Shield Ground 12 ft
Application use if <= 362 kV
100 ft /30 m
Cable Trunk 55 ft to the split
Shield Ground 18 ft
Application use if > 362 kV
No longer using the white jumpers
Hardware Test Leads
IEC Definitions
Source lead
The lead connected to the voltage source of the measuring
equipment used to supply an input
voltage to the test object.
Reference lead (Vin)
The lead connected to the reference channel of the
measuring equipment used to measure
the input voltage to the test object (Vin).
Response lead (Vout)
The lead connected to the response channel of the
measuring equipment used to measure
the output voltage of the test object (Vout)
FRA test leads and equipment
High frequency co-axial cable
Impedance matched ( 50 ohms ) at equipment
(to avoid reflections in test leads)
Separate S and R leads for applying and
measuring signal at input terminal
Only then will layout of test leads have no
effect on measurement
Any length ( 18 m popular )
Applied and measured signals
Vi
Vo
Response = Vo/Vi
Frequency Response Analysis
Sweep Frequency Method
|
|
.
|

\
|
=
|
|
.
|

\
|
=

in
out
in
out
V
V
H
V
V
dB H
1
10
tan ) (
log 20 ) (
u
Most Useful Frequency Range
20 Hz - 2 MHz
Tutorial Session - Frequency Response Analysis
Achieve a Two Port Network
Hardware Range & Resolution
Signal Generation: Range &
Resolution
Two independent measurement channels
Oversample at up to 100 MS/s
20 V p-p
10 Hz-25 MHz
Log Scale at 1.02% of frequency
Conclusions
RLC components have different SFRA
responses
Even simple circuits may have complex
responses
Real transformers have many resonances
Together We Power
The World
SFRA Training
Test Procedures
SFRA Test Procedures
There is a basic set of recommended tests for any
transformer.
Doble software comes complete with templates
which have IEEE and CIGRE compliant tests.
Further tests may be performed for diagnostic
purposes:
per phase short circuit
interwinding tests
Reverse short circuit
etc
Setting up the transformer
Transformer should be disconnected
from system. Busbars removed if
possible. Line, neutral and any tertiary
line connections shall be disconnected
but tank earth, internal auxiliary
equipment and internal current
transformer connections shall remain
connected.
Transformer should be in normal
service condition (assembled, oil-
filled).
Setting up the transformer
Make connections as determined by
winding configuration. Phases not
under test are usually left floating.
In the case where two connections to
one corner of a delta winding are
brought out, the transformer shall be
measured with the delta closed but not
with the earth connected.
Setting up the transformer
If special connections have been specified
and are provided on the test object to enable a
frequency response measurement to be made
when it is arranged for transport then the
measurement must be made in both the fully
assembled (fluid filled) and transport
configurations (drained if required for
transport) before transport and subsequently
as specified by the purchaser.
Beware of different states of oil draining
Setting up the transformer
It is important that SFRA measurements
are always made in a consistent way and
that all details of the measurement
method are systematically recorded.
This will help to avoid false
discrepancies and ensure the
compatibility of frequency responses
during comparison.
Connections - look at each winding
separately:
Make measurements at extreme raise LTC
and nominal DETC
With previous results repeat those
procedures
HV - H1-H2, H2-H3, H3-H1
LV - X1-X0, X2-X0, X3-X0
Short Circuit - H1-H2, H2-H3, H3-H1
with X1-X2-X3-X1 shorted (all three phases, not X0)
Typical Measurements two winding unit
FRA test connections
FRA test connections
Measurements
5. SFRA Test Procedure
Open-Circuit (OC) Tests
Short-Circuit (SC) Tests
Supplementary Test: Inter-Winding Tests
LTC at extreme rise (16R or 1).
DETC at nominal position.
If possible to do additional tests - e.g. various tap
positions, short circuit tests, tertiary tests etc.
5. Typical Test Procedure
Delta-Wye Open Circuit (OC) Tests
Red Black
H1 - H3 HV Winding Tests
H2 - H1
H3 - H2
X1 - X0 LV Winding Tests
X2 - X0
X3 - X0
Supplementary
H1 - X1 Inter-Winding Tests
H2 - X2
H3 - X3
5. Typical Test Connections
Delta-Wye Short Circuit (SC) Tests
Red Black Short-Circuiting
H1 - H3 X1X2X3X1
H2 - H1 X1X2X3X1
H3 - H2 X1X2X3X1
Other Test Connections
Software
Nameplate
Test Template
SFRA Trace Shorted Leads What kind of trace?
SFRA Trace Open in Test Circuit
Open
Circuit
inside test
specimen
Open
Circuit
within test
leads
What kind of trace?
What would be the measurements ?
Single phase two winding unit?
Three phase three winding D-D-Y unit?
Check Latest Doble connection guide
Example from Manual
Measurements
S/C Connections
Example from Manual
Measurements
Measurements
Typical Results - Conclusions
Results should be expected form
Make sure connections are appropriate for
transformers
Check with previous results for consistency
Shorted cable measurement
Shorted cable open ground measurement
Open cable measurement
Ground check on transformer
Repeat measurements on good winding
Does it make sense????
Troubleshooting
Troubleshooting
If differences are observed when
comparing with a finger print result,
it is important to first verify the
measurement by repeating to ensure
that the differences are not caused by
bad measurement practice or by
making different measurement
connection.
Together We Power
The World
SFRA Training
Typical Results
Wye-wye : HVs
Note low frequency variations
Wye-wye: LVs
Less dB down than HV
Wye-wye HV Short Circuit
LV windings shorted
Another HV Wye
HV winding three phases one DETC positions
Another HV Wye
Detail three phases
Center phase
Two outer
phases
Another HV Wye
Detail three phases
Center phase
Two outer
phases
Delta-wye HVs
Typical response at low frequencies
Delta-Wye: LVs
Less dB down than HV
4. Simple Transformer Model
Which trace is highest
and lowest?
OC HV Trace
OC LV Trace
SC Trace
HV
LV
HV
V
i
V
o
V
i
V
o
V
i
V
o
LV
Open Circuit (OC)
Short Circuit (SC)
Lowest
Middle
Highest
Which Winding has higher
impedance?
4. Simple Transformer Model
Which trace is V
H
V
i
HV
V
H
LV
ZH > ZL:
VH: Lower response
or more attenuated
0
-20
-40
-60
-80
-100
10
0
10
10
V
H
V
L
V
i
V
L
Delta-Wye HV Short Circuit
LV windings shorted together
Autotransformer HV
HV LV and Tertiary compared
HV (series) winding
Tertiary winding
LV (common) winding
Measurement Sense: H1-H0 v. H0-H1
Effect is smaller when the two bushings are
similar e.g. H1-H3 v H3-H1
H0-H1
H1-H0
Variation with LTC position
16 Lower through 8 Lower for one winding
Variation with LTC position
Mid frequency detail
16 Lower
8 Lower
Variation with and without oil
LV winding
With oil
Lower
resonant
frequencies
Without oil
Higher
resonant
frequencies
Effect of Bad grounds
Original and Bad Red lead ground
Original
Bad red lead
ground
Effect of Bad grounds
Original and Bad Black lead ground
Original
Bad red lead
ground
Effect of Magnetization
Low frequency variation is severe but identifiable
Magnetization affects the
core response
Magnetization & Grounding variation
HV winding
High frequency variation
due to grounding
Magnetization affects the
core response
Magnetized center phase
Sister Units showing Magnetization
HV winding
Magnetization variation
Sister Units showing Magnetization
HV winding same unit, more magnetization
Magnetization variation
HV Wye winding: Open and Short
HV winding same unit, more magnetization
Open circuit wye winding
Very similar at high
frequencies
Short circuit wye winding
HV Delta winding: Open and Short
HV winding same unit, more magnetization
Open Circuit
Similar at high
frequencies
Short Circuit
Typical Results - Conclusions
Results vary between units depending on
size and type of unit
Results vary with magnetization &
grounding
Results vary with tap changer positions
Results vary with oil level
Know what to expect see next slide!!!
4. Simple Transformer Model
Which trace is V
CHL
V
i
HV
V
H
LV
CHL: High-to-Low or
Inter-Winding
0
-20
-40
-60
-80
-100
10
0
10
10
V
H
V
CHL
CHL
V
CHL
One Transformer: HV, LV, SC & IW
IW = Interwinding measurement
HV Open Circuit
H-X interwinding
Short Circuit
LV Open Circuit
Wye : HV open circuit
Note low frequency variations
2Highs and 1 Low
V and W shape
Experience: Certain Frequency Bands
Indicate Different Problem Conditions
Interpretation
400kHz to
2MHz:
Movement of
main and tap
winding leads
2kHz to 20kHz:
Bulk Winding
Movement Relative
to Each Other,
clamping structure
<2kHz: Core
Deformation, Open
Circuits, Shorted
Turns & Residual
Magnetism
20 kHz to 400kHz:
Deformation Within
the main and tap
windings
4. Simple Transformer Model
V
o
V
i
A
B
C
High Low High
=> Exciting Currents = 2 similar Highs & 1 Low
V shape = B Phase
W shape = A/C Phase
2. SFRA Logarithm Graph
100 Hz
1 MHz
2. What is SFRA? Linear Graph
Impossible to see 100 Hz
Interpretation
600 MVA GSU Transformer
SFRA results from factory and initial field
New SFRA results taken after the fire
Power of a Null Result
Fire: factory to field comparison
HV and LV results overlay almost
exactly up to 2 MHz
Short circuit results
to within 0.01 dB
Low frequency
variations are due to
core magnetization
Post fire conclusions
Results show no variation to 2 MHz
using different:
Doble test sets
test personnel
test leads
lead positions etc
Post fire conclusions
SFRA results show no indication of a
mechanical problem within the transformer
this is a NULL result
SFRA data was key in the decision to return
the unit to service
Full details in 2005 DobleConference Paper
Together We Power
The World
SFRA Training
Open and Short Circuit
Test Variation
SFRA Typical Results
This presentation gives typical results for open
and short circuit tests
It explains why the two tests are different and
the value in each test
A T-model of a transformer is used to make
things clearer
Only low frequencies are considered no
capacitive effects
Test Connections - SFRA Open Circuit
R small
R high
R small
HV Winding
LV Winding
Core
Signal & Reference
Test
M5100
Normal test on HVs the LVs float
Model is relevant for LOW FREQUENCIES
R small
R high
R small
HV Winding
LV Winding
Core
Signal & Reference
Test
M5100
Short circuit test on HVs - LVs
shorted
Model is relevant for LOW FREQUENCIES
LV Short
Test Connections - SFRA Short Circuit
Short Circuit v. Open Circuit
Open circuit responses dominated by core at
low frequency as with Exciting current
Different magnetic paths lead to different
responses typical outer v center phase
variation
Short circuit responses remove effect of core
at low frequencies as with leakage
Reactance
All three short circuit responses should be
identical
Short Circuit Test - Setup
Short secondaries together, use
proper size jumpers, close the delta,
X1-X2-X3-X1
Do not short multiple secondaries
together
Test each phase
Phase A
Phase B
Phase C
Short Circuit (SC) Analysis
Analyze the inductive roll-off region
This roll-off defines the primary
winding inductance model
Very similar to the LRT
SC Analysis
The SC test the phases should be no further than 0.1
dB from each other
Roll-off should be close to -20dB/decade
Analysis will require you to zoom in very close
Step 1:
Overlay the
three HVSC Test Results
Step 2:
Zoom in on the LF Region
that shows the Inductive
Roll-off
Step 3:
Identify the key artifacts
A) LF DC Section
B) Inductive Roll-off Knee
(Low-Pass Filter Cut-off
Frequency)
C) Inductive Roll-off Section
D)End of Inductive Roll-off
Step 4: Analyze LF DC Section
This is the few points that are at
the very beginning 20 Hz
There should be very close offset
between the three results.
If there is a fan-out at the
beginning this could indicate a
winding resistance problem.
Recommend a winding
resistance test
This example is good
Step 5: Locating the Knee point.
Starting at 20 Hz, look for
maximum deflection point
where the traces go into the
inductive roll-off section (next
slides explains this more)
The LP Filter cut-off frequency
by definition is the point where
the frequency response goes
from flat to -3db down
Step 5: Find the start and end points of the
Inductive Roll-off
The inductive roll-off is the ramped linear section
that drops at -20dB/Decade. On a logarithmic
plot it appears as a linear ramp. One of the
reasons logarithmic display is better than
linear.
Use the ruler-method if you need to find where the
ramp stays linear (straight line ramp down to
the right). Find the start and stop points where
is stays on a straight edge
Start
Stop
Inductive
Roll-off
Step 6: Choose a point to analyze
Select a point somewhere very close to the half
way points between Start and Stop and zoom
in very tight
Inductive
Roll-off
Zoom in here
Step 7: Calculate the offset
12.64-12.57=0.07dB =GOOD!
In general, most transformers will be below
0.1 db offset, but up to 0.2 dB is not
uncommon.
This analysis is sensitive to the same issues
as Leakage Reactance.
HV Wye winding: Open and Short
HV winding same unit, more magnetization
Open circuit wye winding
Very similar at high
frequencies
Short circuit wye winding
Good Short Circuit Results
Good Transformer Good Results Lo Frequency
Resistive section in
this case is not
horizontal. This is
common
All three have same
roll off
All three have very
similar resistance
element at 20 Hz
SC Analysis
If SC results do not match....
at 20 Hz -> Check Winding Resistance
Variation in results may occur near 20 Hz due to
variation in resistance element of shorting leads.
May indicate a need to check transformer with DC
resistance tests
Compare LRT results throughout roll off
Variation in inductive roll off is due to variation in
winding impedance
Center phase may be slightly different
We expect all three phases to be within 0.1 dB
Short Circuit Variation
Recommended is a three phase equivalent short
circuit test
Can do a per phase equivalent by repeating tests
and just shorting appropriate LV winding
Can do LV short circuit (with HVs shorted)
Per phase and LV measurements are useful in
diagnostic investigations
Short Circuit v. Open Circuit
At open circuit the load on the LV side is
infinite
At short circuit the load on the LV side is
zero
We should expect a result to lie between the
open and short circuit results!
Case Studies
1 Shorted Turn
2 Hoop Buckling
3 Axial Collapse
4 Clamping Failure
5 Bushing Failure
Case 1: Shorted Turn
Demonstrates need for good low
frequency resolution
Many electrical tests should show a
shorted turn
No baseline data required for
diagnosis
Case 1: Normal Response
Three phases
respond
differently at
low
frequencies
Three normal open circuit responses
Results as expected and acceptable
Case 1: Shorted Turn
Variation is clear indicator of shorted
turn on A phase
After an incident - one year later...
One phase has clear
inductive roll off associated
with short circuit
measurement
Case 1: Shorted Turn
Another transformer - no reference results
Low frequency responses clearly show
inductive roll off associated with short
circuit measurement
Results indicate shorted turn on one phase
Case 2: Hoop Buckling
Compressive Failure of Winding - also known as Hoop
Buckling
Need repeatability to within 1 dB
Need low frequency repeatability to within 0.3 dB
Expect increase in LC combination of winding bulk - seen
as left shift of resonance at mid range
Consequent increase in winding impedance - seen in short
circuit results
Case 2: Hoop Buckling
Example here of two 28 MVA sister
units
One known bad after internal
inspection
Other unit suspect based on
Dissolved Gas Analysis
Clear bulge
in winding
Case 2: Hoop Buckling
Inspection of bad transformer revealed hoop
buckling on TWO windings
Two bad phases have shift to lower
frequency at this frequency
Case 2: Hoop Buckling
The B phase is different -
which may be due to
design and construction
Open Circuit LV results
By itself, this is not conclusive
evidence as we have no
reference results
0.2 dB delta is significant here!
Low frequency
short circuit results
show consistent
increased
impedance on bad
phases
Case 2: Hoop Buckling
The variation was
consistent and repeatable:
attributable to variation
within the transformer
Results are anomalous
but symmetrical -
evidence is inconclusive
Case 2: Hoop Buckling
For the bad transformer we have picked
out two possible indicators of variation
for the two bad phases
For the suspect unit, we look at the
same areas
Small change in frequency and dB
Clear shift left at
same frequency
range
Case 2: Hoop Buckling
Open circuit results
show same effect on
ONE phase
In the known bad unit two phases had
a shift left and both had hoop buckling
Now we have just one phase shift left
Both units - displayed for comparison
Case 2: Hoop Buckling
This is good evidence that one
phase may have hoop buckling
0.2 dB delta is significant here!
Suspect unit shows same increased impedance
(more dBs down) for one phase - the same one
with the shift left on open circuit results
Case 2: Hoop Buckling
One phase appears to have symptoms of
hoop buckling
Case 2: Hoop Buckling
Diagnosis is that suspect unit has one bad phase through
hoop buckling
Transformer is bent, but not broken
Unit was returned to service for some months, under
advice that it was less likely to withstand a close in fault
Subsequent internal inspection revealed hoop buckling
on the suspect phase, as expected and predicted
Case 2: Hoop Buckling
Transformer returned to service before being rewound
Inspection showed hoop buckling
on suspect phase - as predicted
Case 2: Hoop Buckling
Need good resolution - especially below 1 kHz
for short circuit results
Need good repeatability over a range (less than
0.3 dB)
Use of results from similar unit as reference
was key for diagnosis
Case 3: Axial Shift of a Winding
Example uses reference results taken 7 years
previously
Need range up over 1 MHz
Need repeatability to within 1%
Need confidence in resultsto avoid a false negative
Failure mode is reduction in LC combination
Resonances should shift right at higher frequencies
as a consequence of LC reduction
Scattered blocking
under a collapsed
winding
Case 3: Axial Shift of a Winding
Case 3: Axial Shift of a Winding
-70
-60
-50
-40
-30
-20
-10
0
0 0.5 1
Frequency in MHz
A
m
p
l
i
t
u
d
e

i
n

d
B
22/11/94 8/11/01
Results from one GOOD phase taken in 1994 and 2001
Small dB variations
are acceptable: no
LC variation implied
No variation between traces implies
that THIS PHASE has little change
Case 3: Axial Shift of a Winding
-70
-60
-50
-40
-30
-20
-10
0
0 0.5 1
Frequency in MHz
A
m
p
l
i
t
u
d
e

i
n

d
B
22/11/94 8/11/01
Results from bad phase taken in 1994 and 2001
Clear and consistent shift to higher
frequencies of several resonances
Low frequency results, up to
over 200 kHz, are acceptable
Variation implies we have a
substantial problem
Case 3: Axial Shift of a Winding
-70
-60
-50
-40
-30
-20
-10
0
0 0.5 1
A
m
p
l
i
t
u
d
e

i
n

d
B
-70
-60
-50
-40
-30
-20
-10
0
0 0.5 1
Frequency in MHz
A
m
p
l
i
t
u
d
e

i
n

d
B
Good SFRA repeatability
allows confident
diagnosis of bad phase
Good SFRA repeatability
gives confidence in the
integrity of the good phase
Case 3: Axial Shift of a Winding
Bad phase clearly identified
SFRA results up to >1 MHz required
SFRA results taken with different test
sets by different test teams
Repeatability is key
Case 4: Clamping Failure
750 MVA transmission transformer
Close in fault caused center winding to
jump and break the clamping
Stress shield dropped on to winding,
altering capacitance of that winding
Case 4: Clamping Failure
Clamping beam broken
Stress shield collapsed on
to center phase
Case 4: Clamping Failure
-80
-70
-60
-50
-40
-30
-20
-10
0
0 5 10 15 20
Frequency in kHz
A
m
p
l
i
t
u
d
e

i
n

d
B
Low frequency results
show expected form
Three HV phase results shown together
Center phase shows shift
to lower frequency for
first resonances
Transformer has a problem further
investigation required
Case 4: Clamping Failure
Initial low frequency (< 2kHz) core
related resonance unaffected
Resonances usually associated with
bulk capacitance of windings to
ground are moved to lower
frequencies
Case 5: Bushing Failure
Transmission Transformer had a
bushing failure
No reference results available
Did the failure move the windings?
Sister unit available for reference
Minor variations in suspect unit at
high frequencies may relate to
some internal lead movement
Low frequency
variation as expected
between phases
Case 5: Bushing Failure
Variation is small - but needs further investigation
Six traces here - suspect
unit and sister unit
Case 5: Bushing Failure
Variation is still a concern - need to do a
phase-by-phase comparison of two sister units
There are two traces here - suspect
phase of suspect unit and same
phase of sister transformer
Case 5: Bushing Failure
Sister unit has exactly
the same variation - it
must be design related as
sister is known good
The two traces are
clearly very similar -
can just see one trace
behind the other
No variation between traces implies this
phase is OK, despite bushing failure
Suspect phase does show some
variation at high frequency
Sister unit shows same variation -
known to be a good transformer
Consequently - SFRA variation is
due to design and construction
Case 5: Bushing Failure
Sister Transformers may show variation between
corresponding phases on each transformer
If two phases are similar between the two units, it is
likely that the third phase is also similar
So we can use sister units as a reference, even if there
are variations in response between the units
Case 5: Bushing Failure
Experience: Certain Frequency Bands
Indicate Different Problem Conditions
Interpretation
400kHz to
2MHz:
Movement of
main and tap
winding leads
2kHz to 20kHz:
Bulk Winding
Movement Relative
to Each Other,
clamping structure
<2kHz: Core
Deformation, Open
Circuits, Shorted
Turns & Residual
Magnetism
20 kHz to 400kHz:
Deformation Within
the main and tap
windings
Together We Power
The World
SFRA Training
Using the Software
Using the Software
Communication protocol Ethernet
cable
Using the Software
Communication protocol CSMUAB (USB)
Cable
USB Drivers
Communication: Ethernet or USB over
Ethernet
Ethernet 192.168.1.0
USB over Ethernet 192.168.1.1
Windows will ask to install the USB driver
when the USB cable is plugged into the
M5X00 for the first time.
Point to location of USB drivers
Windows Vista
SFRA V5.1 will operate on Windows VISTA
The communication limitations still apply,
regarding profiles and settings
One issue has surfaced.
VISTA Workaround
Some people have had issues even when
logged in as an administrator.
The workaround so far has been to tell the user
to turn off the 'User Account Control' in the
Vista User Manager.
Turning this off has fixed every issue we have
had so far.
User Account Control
SFRA Software
The best way to become familiar with the
software is to use it
Topics on the following slides may help
direct discussion and experimentation
To try
Software installation
Connect to an M5200
Data management: settings, folders and files
Test philosophy:
Set up data files
Select particular files
Run test
Review results
To try
Choosing & Using templates
Select Apparatus & Run test
Reviewing, overlaying and comparing data
Analysis tab
Filenames & e-mailing files
Conclusions
Key points:
Collect data in an organized manner
Always make notes for tap positions
Backup for your test data regularly
Together We Power
The World
SFRA Training
Basic Analysis
3. Are Baseline Results Necessary?
Not always - but they are preferable
Phase-to-phase comparisons
Comparison with sister units
Comparison with units from the same
manufacturer
Results have a generally predictable shape
SFRA Analysis is Logical
Analyzing SFRA data is a logical and
methodical process
Examples are given here to show the
approach taken
SFRA Analysis : getting Started
Make sure all results are available:
Current SFRA results
Any previous SFRA results
Any results from similar units
Any electrical test or DGA data
Understand the context: why are these
tests being performed?
In general
Trace shapes:
HV
LV
Short Circuit
Do they look as expected? Any variance?
Analysis with benchmark
Have the same measurements been made?
LTC and DETC the same?
Oil, bushings etc
Sister Units
Real sister?
Example
Phase-by-phase
When it gets down to basics
Causes of variation
Check two useful documents:
Practical variability
Field and Factory variation
Integration
SFRA is one tool in the tool box
SFRA
Capacitance DC Resistance
Exciting
Currents
Leakage
Reactance
OC: Open Circuit SC: Short Circuit
Context
Link SFRA to other tests:
DC resistance
Leakage Reactance
Exciting Currents
Capacitance
Conclusions
Analysis takes experience and time
Using Cross-Correlation Coefficients to
Analyze Transformer SFRA Traces
The Problem
Analyzing SFRA Traces can be Challenging
No simple categorization available
Natural desire to apply limits to SFRA
results
Smart SFRA analysis routines
may have dubious outputs
The Reality
Human element needed - pictures
Every transformer is an individual
No perfect pass/fail criteria for any range
SFRA analysis is still prone to outlandish claims
from some quarters
Someone has to know what they are doing
The Stop-Gap
CCFs provide a method to:
Quantify SFRA Trace similarities
Communicate results
Apply limits to analysis
Dont rely on hidden or secret black box
analysis routines
Cross-correlation coefficients (CCF) are
commonly used in a variety of industries to
monitor signal integrity
What is a CCF really?
Take two sets of numbers
Work out a sameness value
Same different opposite
Math is simple, as shown here
Still keeping the math simple....
2
2
2
) ( * ) (
) )( (




=
Y Yi X Xi
Y Yi X Xi
CCF
In words, the CCF calculates the convoluted mean error
and normalizes to the individual mean over a particular
range of frequencies
Keeping the math simple....
CCF is a number between 1.0 and -
1.0.
For SFRA, its usually between 1.0
and 0.0
The closer the value is to 1.0, the
more similar the traces are.
The lower the value, the more the
two traces diverge.
CCF can also be expressed in Percent
Example CCFs
CCF
Good Match 0.95-1.0
Close Match 0.90-1.0
Poor Match <.89
No or very poor match <= 0.0
What you really need to know
Regions of Interest
Region Frequency Sub-Band Components
1 <2kHz Main Core; Bulk Winding Inductance
2 2kHz to 20 kHz Bulk Component Shunt Impedances
3 20 kHz to 400 kHz Main Windings
4 400 kHz to ~1 Mhz Main windings, Tap windings etc
Every transformer is different, these are just general
guidelines
SFRA Regions
Select your Band and Calculate
Selecting Limits
Depending on the type of comparison, the CCFs
limit should be adjusted. Starting Point!
Benchmark comparison:
Phase to Phase (A/C) or sister units:
Case 1: Benchmark Comparison
675 MVA GSU built in 2002
Suffered a fire on connected iso-phase bus
Bushing had been degraded during the fire
Expect some high frequency variation
The unit was known to be magnetized
Expect some low frequency variation
Case 1: Benchmark Comparison: H2-H0
Frequency Sub Band CCF
Region 1: 0 - 2 kHz 0.9879
Region 2: 2kHz 20 kHz 0.9964
Region 3: 20 kHz 400 kHz 0.9882
Region 4: 400 kHz 1 MHz 0.9988
Case 2: Phase-to-phase : Bent Unit
1960s vintage GE 50
MVA transformer
Tripped out of server
on protection
No reference SFRA
results available
Phase to phase
analysis required
Close up view of LV winding shift
Case 2: Bent Transformer: A/C Phase
comparison
Frequency Sub Band CCF
Region 1: 0 - 2 kHz 0.9831
Region 2: 2kHz 20 kHz 0.9868
Region 3: 20 kHz 400 kHz 0.8262
Region 4: 400 kHz 1 MHz 0.9567
Case 3: Two Large Sister Units
Two 370 MVA Alstom transformers
345/14 kV
Routine health assessment of the units was
conducted
SFRA was used to ascertain construction
consistency between the units
Determines if two units are in fact identical
Case 3: Sister Units
Frequency Sub Band CCF
Region 1: 0 - 2 kHz 0.9898
Region 2: 2kHz 20 kHz 0.9994
Region 3: 20 kHz 400 kHz 0.9914
Region 4: 400 kHz 1 MHz 0.9923
Considerations
Case studies did not change region limits
Not necessarily true in all cases
May change limits for different designs
Recommended limits are starting points
Can adjust limits; situation may require fine tuning
Still does not replace the trained SFRA user
Poor test technique can throw off results
Conclusion
CCF can help in the analysis of traces
by drawing attention to various
regions
CCF assign a Figure of Merit that
can be used to quantify SFRA traces
similarities
CCF can not diagnose the failure
mode, only the trained human eye and
experience can do that
Still does not replace the trained
SFRA user
7. Conclusions
SFRA is a useful and sensitive tool
Use in Conjunction with other Tools
SFRA is a reliable and repeatable means of
making FRA measurements
Use on relocation or after an incident
Use as part of factory QA
7. Notes...
Need for good grounds
Make sure transformer is in the same state
as last time
Use a reference library
SFRA Case Studies:
Dealing With Noise
Presented by Doble Engineering Company
239
Signal To Noise Ratio (SNR)
Overhead Lines
Ground Grids
Neighboring Transformers
Substation Equipment
The Test signal can
become swallowed by
neighboring sources of
interference
The Real World has
Real Sources of Line
interference
Doble Provides 20 Vpp
to prevent this problem.
SNR: Why its important
You may end up measuring the surrounding noise.
Results for a test set using low SNR
241
Doble SFRA Leads
Clip-1
Clip-2
Tests above 1 MHz are dominated by the test leads
(IEC & IEEE)
Every test connection is a chance to make a
mistake and mess up results; Doble keeps it simple
We all test in every possible environment so dont
complicate it!
Inside Transformers
Adverse Weather At Night
Done.
242
Advanced Analysis: Cross Correlation & Difference
Cross-Correlation Analysis
-Assess Condition by
Winding Region
Difference Analysis
- Compare offset variations
for simple quick analysis
243
Technical automated analysis
Diagnosis by difference and cross correlation is most advanced
available
Training and support mean we dont lose sight of the picture,
numbers can only tell you so much
Cross-correlation (CC) looks at curve shape will be the basis of
the Doble expert system
Some systems have a single CC value for the whole trace
244
Intermediate Frequency (IF) Bandwidth
Some Others
Doble follows Laboratory Best
Practices of always ensuring IF
Bandwidth is < 10 %
In simple Terms: We don t try to
make the results look prettier by
allowing other surrounding
signals to smooth the result out.
Doble
Frequency to be
measured
Nameplate Information
Auto Transformer with Tertiary
20 MVA
138/69/13.2 kV
Year of manufacture: 2011
Test configuration:
No Oil
Spark plug (transit) Bushings
The story unfolds
SFRA tests were performed on a transformer
after delivery to the site and then compared to
factory results
Something didnt quite agree..
H1-X1 Comparison to Factory
X2-H0X0 Comparison to Factory
Hey why are these different?
Factory
Results
Site Test
HVSC Test Results
Summary of initial SFRA analysis
Both HVOC and LVOC results were
satisfactory when compared to previous
Upon evaluation it was decided the HVSC
results did not compare to factory tests
An investigation followed and ultimately the
transformer was tested again
The Investigation showed .
That the shorting leads for the HVSC tests
were allowed to sag and come in contact with
the transformer tank.
The leads were tightened up with no sagging
and the short circuit tests were performed
again.
Retest Satisfactory - 9 tests shown
Factory results
and Retest after
fixing the leads
Shorting leads
resting on tank
Nameplate Information
Auto Transformer with Tertiary
100 MVA
220/132 kV
New unit
Test configuration:
Fully dressed
The story unfolds
SFRA tests were performed on a transformer
after delivery to the site and then compared to
factory results
Field results were different when compared to
previous (factory) tests
Again, something didnt quite agree..
Series Winding Factory vs. Field
Field
Factory
Common Winding Factory vs. Field
Field
Factory
Field retest after Demagnetization
But what is this here?
This is better
Lets get a closer look
Series Winding
Field Retest
Common winding field retest
This is better
This isn t good
Conclusion
Effects of magnetic viscosity were evident
in both the series and common windings
between tests.
Cause of difference in field retest
Tertiary was ungrounded for the test
SFRA Case Studies:
Field LV Anomaly
Presented by Doble Engineering Company
Nameplate Information
125 MVA
220 kV Wye Delta
Three winding transformer
In service unit, DGA starting at the end of
previous year showed rising trend
SFRA was performed
Initial SFRA analysis
SFRA results indicated problems in the LV
Even to the untrained eye one should be able
to pick out the differences
LVOC SFRA Test
Houston we have a problem!
Time to go inside
An internal investigation was
conducted and visible evidence of the
problem was found,
Pictures tell the story
Internal inspection
Internal Inspection
Internal Inspection
SFRA LVOC after repair
SFRA Example
Transformer Type: 3| Power Transformer
Manufacturer: Waukesha
Capacity: 30/40/50 MVA
Voltage: 138/27 kV
Tap Changer: 10% load tap changer; HV reconfiguration
tap changer, de-energized
Entered service J uly 2002
Removed from service two weeks later
Transformer
SFRA Example
Indications
Tank temperature recorded at 480 F (230 C)
Slight tank wall bulge
Tank wall discoloration
Strange noises while in service
DGA Acetylene from 0 to >300
Hydrogen from 10s to >500
FFAs from low to high
SFRA Example
Transformer Turns Ratio
Within 0.3% of nameplate and previous
results
SFRA Example
H1-H2 H2-H3 H3-H1
DETC LTC X0-X2 X0-X3 X0-X1
Pos Pos 7/22/02 8/5/02 % 7/22/02 8/5/02 % 7/22/02 8/5/02 %
C 1R 8.809 8.811 -0.02% 8.809 8.808 0.01% 8.809 8.808 0.01%
B 1R 9.033 9.035 -0.02% 9.033 9.032 0.01% 9.033 9.032 0.01%
A 1R 9.258 9.260 -0.02% 9.258 9.257 0.01% 9.258 9.257 0.01%
D 1R 8.575 8.574 0.01% 8.573 8.572 0.01% 8.573 8.572 0.01%
E 1R 8.336 8.340 -0.05% 8.336 8.338 -0.03% 8.336 8.338 -0.03%
C N 8.861 N/P N/A 8.861 N/P N/A 8.864 N/P N/A
C 1L 8.919 8.911 0.09% 8.919 8.918 0.01% 8.919 8.918 0.01%
C 2L 8.976 8.964 0.13% 8.976 8.975 0.01% 8.976 8.975 0.01%
C 3L 9.033 9.018 0.17% 9.033 9.031 0.02% 9.033 9.031 0.02%
C 4L 9.090 9.075 0.17% 9.090 9.090 0.00% 9.090 9.090 0.00%
C 5L 9.149 9.133 0.17% 9.149 9.149 0.00% 9.149 9.148 0.01%
C 6L 9.208 9.190 0.20% 9.211 9.208 0.03% 9.208 9.208 0.00%
C 7L 9.268 9.250 0.19% 9.268 9.268 0.00% 9.268 9.268 0.00%
C 8L 9.329 9.310 0.20% 9.329 9.310 0.20% 9.329 9.310 0.20%
C 9L 9.391 9.370 0.22% 9.391 9.391 0.00% 9.391 9.391 0.00%
C 10L 9.460 9.433 0.29% 9.453 9.453 0.00% 9.460 9.453 0.07%
C 11L 9.516 9.495 0.22% 9.523 9.518 0.05% 9.516 9.518 -0.02%
C 12L 9.580 9.559 0.22% 9.588 9.582 0.06% 9.580 9.582 -0.02%
C 13L 9.645 9.623 0.23% 9.645 9.647 -0.02% 9.645 9.647 -0.02%
C 14L 9.711 9.688 0.24% 9.711 9.713 -0.02% 9.711 9.713 -0.02%
C 15L 9.785 9.745 0.41% 9.782 9.780 0.02% 9.785 9.781 0.04%
C 16L 9.846 9.821 0.25% 9.846 9.849 -0.03% 9.849 9.849 0.00%
Power
Factor/Capacitance
CHL rose by a factor of ~3; slight
capacitance rise
Current (mA) Watts-Loss (W)
7/22/02 8/13/02 % 7/22/02 8/13/02 %
CH 9.946 9.974 0.28 0.218 0.254 16.51
CHL 23.540 23.650 0.47 0.436 1.239 184.17
CL 59.410 59.660 0.42 1.371 1.420 3.57
% PF (Corrected to 20C) Capacitance (pF)
7/22/02 8/13/02 % 7/22/02 8/13/02 %
CH 0.18 0.21 16.67 2638 2645 0.27
CHL 0.15 0.42 180.00 6245 6275 0.48
CL 0.18 0.19 5.56 15761 15825 0.41
Excitation Currents
Excitation Currents
4
5
6
7
8
9
10
N
2
L
4
L
6
L
8
L
1
0
L
1
2
L
1
4
L
1
6
L
LTC Posi ti on
m
A
H1-H2/mA
H2-H3/mA
H3-H1/mA
Leakage
Reactance
Nameplate 9.39%
Measured 3 phase equivalent 13.19%
H3-H1/O H1-H2/O H2-H3/O
Per Phase 596.60 174.00 173.30
Impedance 32.36% 9.15% 9.11%
SFRA Example
Discoloration and bulge
Fault pressure Relief
valve (did not operate)
Low Voltage
High Voltage
High Voltage Short Circuit
Black
specs...
Copper exposed on lead
Factory Inspection
Autotransformer transportation
damage found using SFRA
Case Study
Transformer Before Shipment
With oil and actual bushings > H X Open Circuit test
Transformer After Shipment
No oil and w/spark plug bushings > H X Open Circuit test
Transformer After Shipment
H2 X2 Open Circuit test before and after comparison
Transformer After Shipment
No oil and w/spark plug bushings > Tertiary Open Circuit
Case Study #1
As a result of the Swept Frequency Response
Tests a decision was made to perform an
internal inspection.
The following was observed:
Internal Inspection
Center winding clamping hardware loose
resulting in loss of pressure on the stack
Y1 lead laying on top of core
Loose boards were found inside transformer
Internal Inspection
Internal Inspection
Internal Inspection
Loose Clamping
Internal Inspection
Internal Inspection
SFRA Case Studies:
Possible Core Variation
Presented by Doble Engineering Company
Low Voltage Open Circuit Traces
Unit A
Unit B
Unit C
High Voltage Open Circuit Traces
Unit A
Unit B
Unit C
Before and after de-magnetizing
The suspect unit was de-magnetized at
the plant by energizing the LV winding
at 110% which is 15.2KV, 20 amps for
12 hours.
There was very little change.
Suspect Unit, Low Voltage
Suspect Unit, High Voltage
Conclusion
Upon further investigation it was
discovered that the core steels for all
units was of the same grade (M4),
however two of the transformers used
the steel from one manufacturer and the
third unit used the steel from a different
manufacturer.
This could have resulted in different
core characteristics (e.g., permeability).
SFRA Case Studies:
Magnetic Viscosity
Presented by Doble Engineering Company
Effect of Magnetization
Low frequency variation may be severe
Magnetization affects the
core response
Magnetic Viscosity
Drift can occur from demagnetized state to lowest energy
state depending on internal geometry (see Spring 2010 paper)
SFRA Case Studies:
Phase-to-Phase Find
Presented by Doble Engineering Company
Case Study
120 MVA, 245kV/144kV autotransformer
There was a fault and it was unclear if there
had been damage to the unit
SFRA Testing was done to assess the
condition of the unit
No historical data available
Nameplate data
SFRA HVOC results
Traces do not correlate
well in this area
Suggests some kind of
winding issue
SFRA HVSC results
Some variation
in magnitude &
resonance
Looking for less than 0.2 db phase separation;
Zoom in to check
HVSC inductive roll-off region
0.13 db acceptable
Course of action
Based on the SFRA results a recommendation was
made for internal inspection
It was discovered that the tertiary winding
connections had never been bolted together, and
theorized that the energy from the fault had pushed
the connection apart
There had been arcing across that opening
The missing Link
Internal Connection - Delta Link
Delta Link not made up
Conclusion
Although there wasnt any previous
data, a phase to phase comparison
was enough to support the indication
of a problem
* This might have been caught at the
factory if SFRA testing had been
performed at that time
SFRA Case Studies:
Field SFRA & LR Puzzle
Presented by Doble Engineering Company
Transformer details nameplate
and history
The failure LV bushing in pieces
Field test results good, bad and
inconsistent
Retest
Decision?
Puzzle outline
230/72 kV Federal Pioneer
50/66.7/83.3 MVA Autotransformer
Built 1994
Transformer suffered animal intrusion and tripped out
after tertiary fault in late 1990s
No testing performed
Unit successfully returned to service
Unit had also seen a number of short circuit faults
Transformer nameplate
Unit tripped out in July with operation of:
R and B phase elements
Gas accumulation/surge
Gas pressure relief
Site inspection shows X3 gaskets compromised, oil leaking from flange;
conservator isolated to limit spill
Inspection shows lower part of X3 bushing failed catastrophically
Porcelain shards throughout upper portion of core/coils
Arcing evident from lower part of X3 bushing to top of steel frame
2006 Incident
Transformer nameplate
ABB contacted regarding failure of 69kV 1200 A Type O+C
ABB suspect a known defect
Service advisories from August 1998 and April 1999
ABB offer 3 replacement bushings
Investigation
Transformer carefully cleaned and flushed with hot oil
X1 & X2 bushings removed, inspected, tested and re-installed
Routine electrical tests performed:
Bushing C1 & C2 & power factor measurements
Winding power factor & capacitance
HV single phase excitation
DC winding resistance
Turns ratio
Leakage reactance
SFRA
Investigation
Transformer
All as expected and all Good
Investigation Doble testing
All as expected and all Good
Investigation Doble tests on bushings
Some results are Investigate on HV C2
Investigation Bushing C2 tests
Surge arresters all as per
nameplate
Investigation IEx & Surge Arresters
Investigation LR results
Investigation LR results
Investigation SFRA results
Investigation SFRA results
Investigation SFRA results
Investigation SFRA results
X3-X0
X1-X0 &
X2-X0
Leakage reactance repeated with heavier shorting
cables; results do not change much.
SFRA performed all phases are consistent
Slight variation between phases not considered
unusual; may relate to animal incident
SFRA and LR are inconsistent!!!
Investigation further results
Much thought
Time passes
Time passes more people needed
You can never have too many supervisors
January following year
Repeat tests to review site
procedures and confirm all test
results
ALL Results confirmed
LR and SFRA still inconsistent
Repeat SFRA tests & LR tests
Dobles Lachman IEEE paper suggests perform LR tests from
LV side
Wouldnt necessarily explain the anomaly (X2 and X3 being
different in LR)
LR Results confirmed from LV side as being unbalanced
LR Results reconfirmed unbalanced with three phase supply
and clip on CTs (energize H measure T)
Variations on a theme clutching at
straws
Situation is difficult
LR indicates possible winding movement or anomaly
SFRA does not indicate significant winding movement
Other tests do not indicate a problem
What next? Internal inspection again? Energize and hope?
Summary
It was an LV bushing failure
What could be anomalous about the tertiary?
Clutching at straws
Everything else looks OK
Lets check the nameplate again
Action
Transformer nameplate desperation
Denoument
CT shorting block not shorted on tertiary
Affected high current LR
Marginal affect on low current SFRA
Everything is worth checking
Transformer successfully returned to service
Conclusion
Confidential Notice
Doble Engineering (Doble) hereby grants the recipient (you) the right to
retain this presentation and materials included within (the Presentation) for
private reference. No other rights, title, or interest, including, but not
limited to, the rights to copy, make use of, distribute, transmit, display or
perform in public (or to third parties), edit, translate, or reformat any
portion of the Presentation are hereby or otherwise granted and shall
remain expressly reserved by Doble. You acknowledge and agree that such
limited license is expressly conditioned upon your acceptance of the terms
herein. You further agree that, in the event of your breach, Doble will
suffer irreparable damage and injury for which there is no adequate remedy
at law. As such, Doble, in addition to any other rights and remedies
available, shall be entitled to seek injunction by a tribunal of competent
jurisdiction restricting you from committing or continuing any breach of
these terms.