Matz Ohlen

Director Transformer
Test Systems

Megger
Sweden

Frequency response analysis of power

transformers
Measuring and analyzing data as function of
frequency, variable frequency diagnostics
Impedance vs frequency FRA/SFRA (Sweep
Frequency Response Analysis)
Magnitude/phase vs frequency
Typical frequency range 20 Hz 2 MHz

Insulation characteristics vs frequency DFR/FDS

(Dielectric Frequency Response/Frequency Domain
Spectroscopy)
Capacitance and dissipation factor vs frequency
Typical frequency range a few mHz to 1 kHz

Transformer Diagnostics
Diagnostics is about collecting reliable
information to make the correct decision
Making the correct decisions improves reliability
and saves money
SFRA
TTR

FDS

Winding
Resistance

SFRA testing basics

Off-line test
The transformer is seen as a
complex impedance circuit
[Open] (magnetization
impedance) and [Short] (shortcircuit impedance) responses are
measured over a wide frequency
range and the results are presented
as magnitude response curves
(filter response)
Changes in the impedance can be
detected and compared over time,
between test objects or within test
objects
The method is unique in its ability to
detect a variety of winding faults,
core issues and other electrical
faults in one test

SFRA measurement circuitry

SFRA analysis tools

Visual/graphical analysis

Starting dB values for

The expected shape of star and delta configurations

Comparison of fingerprints from;

[Open] (excitation impedance/current)

[Short] (short-circuit impedance)

The same transformer

A sister transformer
Symmetric phases within the same transformer

New/missing resonance frequencies

Correlation analysis
DL/T 911 2004 standard
Customer/transformer specific

Typical response from a healthy

transformer
HV [short] identical
between phases
LV [open] as
expected for a Y tx

HV [open] as expected for

a Y tx. Double dip and
one response lower

Very low deviation

between phases for
all tests no winding
defects

Transformer with serious issues...

Large deviations
between phases for
LV [open] at low
frequencies
indicates changes in
the magnetic
circuit/core defects

Large deviations
between phases at mid
and high frequencies
indicates winding faults

SFRA standards and recommendations

DL/T 911-2004, Frequency Response Analysis on Winding
Deformation of Power Transformers, The Electric Power
Industry Standard of Peoples Republic of China, 2004
Cigre brochure 342 (2008), Mechanical Condition
Assessment of Transformer Windings Using Frequency
Response (FRA)
IEEE PC57.149/D7 (2009), Draft Trial-Use Guide for the
Application and Interpretation of Frequency Response
Analysis for Oil Immersed Transformers (Draft)
IEC 60076-18 Ed1.0 (2010), Power Transformers Pert 18.
Measurement of Frequency Response (Draft)
Internal standards by transformer manufacturers, e.g. ABB
FRA Standard

SFRA standards Key points

Standard

Dynamic range

Accuracy

Signal cable grounding

EPIS PRC DL/T 911

-100 to +20 dB

1 dB @ -80 dB

Wire, shortest length to

transformer core
grounding

CIGRE brochure 342

-100 to +20 dB
(measurement range)

1 dB @ -100 dB

Shortest braid principle

IEEE PC57.149/D7 (draft)

"Sufficient dynamic
range to
accommodate all
transformer test
objects"

"Calibrated to an
acceptable standard"

Grounded at both ends,

documented and
repeatable procedure

IEC 60076-18 (Draft)

-100 to +10 dB
min 6 dB S/N

ABB FRA Technical Standard

Better than
-100 to +40 dB

0.3 dB @ -40 dB
1 dB @ -80 dB
Shortest braid principle
Smoothing not allowed
1 dB @ -100 dB

Shortest braid principle

SFRA measurement Range - Why you

need at least -100 dB...
Westinghouse 40 MVA, Dyn1, 115/14 kV, HV [open]

Signal cable connection

Shortest braid principle

Source:IEC 60076-18 (draft)

SFRA Summary and conclusions

SFRA is an established methodology
for detecting electromechanical
changes in power transformers
Collecting reference curves on all
mission critical transformers is an
investment!
Ensure accuracy by selecting a highquality instrument
Ensure repeatability by following
international standards and practices

Insulation Testing
Dielectric Response Methods
7

5
FDS/DFR
4

HV Tan Delta
VLF

PDC
Polarization Index
"DC"

0
0,000001 0,00001

Frequency, Hz
0,0001

0,001

0,01

0,1

10

100

1000

10000

Dielectric Frequency Response

Measurements Tan delta from mHz to kHz
Hi

V
Lo
A

Ground

Measure at several frequencies

Use Ohms law:
Z ( ) =

U ( )
I ( )

Z ( )

C, tand, PF
( and )

CHL

CL

CH

Why perform dielectric frequency

response measurements...
Typical power factor values @ 20 C
"New"

"Old"

Warning/alert limit

Power transformers,
oil insulated

0.2-0.4%

0.3-0.5%

> 0.5%

Bushings

0.2-0.3%

0.3-0.5%

> 0.5%

IEEE 62-1995 states; The power factors recorded for routine overall tests on
older apparatus provide information regarding the general condition of the
ground and inter-winding insulation of transformers and reactors. While the
power factors for most older transformers will also be <0.5% (20C), power
factors between 0.5% and 1.0% (20C) may be acceptable; however, power
factors >1.0% (20C) should be investigated.

Dielectric Frequency Response

- Investigating high single number PF data
Dry transformer with old
oil (high conductivity)

Wet transformer with good oil

- Moisture +

What affects the response?

- Moisture +

- Oil Conductivity +

- Temperature +

DFR Moisture estimation (1-2-3)

Right click
Measured
DFR
response

Select
Send toMODS

DFR Moisture estimation (1-2-3)

Capacitor model
Oil
% Spacers
% Barriers

Measurement

Master
curve

DFR Moisture estimation (1-2-3)

2. Click
Auto match
1. Confirm
insulation
temperature

DFR Moisture estimation Result

Geometry
Moisture
Oil conductivity

Dielectric Frequency Response

- Investigating irregular shapes

CHL response

CH and CL responses

DFR analysis irregular responses

A measured irregular shape is not a
mishap It is information!
CH and CL has expected oil-paper response
CHL looks different with higher losses at
mid-frequencies
Contamination/conductive layer between
windings?
This particular transformer had a history
including an LTC replacement due to seriously
burned contacts...

Methods for dielectric response

measurements
DC (Polarization-Depolarization
Current measurements)

Strenghts

AC (Dielectric Frequency Response

measurements)

Shorter measurement time at

very low frequencies

Weaknesses

More sensitive to AC
interference
More sensitive to DC
interference
Limited frequency range (PDC
only)
Data conversion necessary
(combined PDC/DFR only)
Discharge before
measurement may be needed

Strenghts

Less sensitive to AC interference

Less sensitive to DC interference
Wide frequency range
No discharge necessarry

Weaknesses

Longer measurement time for very

low frequencies

Moisture assessment with dielectric

response methods takes a while
Available methods Measurement times

PDC Typically 0.5-3 hours

PDC+DFR approximately 15-25 minutes (2 mHz, with
and without discharge) to 2.5-4 hours (0.1 mHz with and
without discharge)
True AC DFR/FDS approximately 18 minutes (2 mHz) to
about 5.5 hours (0.1 mHz)

Availability Transformer off-line in field

Typically 1 day for complete diagnostic measurements

Measurement time (minutes) for DR

measurements
1000

100

2 mHz
1 mHz
0,1 Mhz
10

1
1st gen FDS

3rd gen FDS

PDC+FDS, no discharge PDC+FDS, with discharge

Typical DFR results for transfomers

with various moisture content

1.5% moisture

0.3% moisture
2.1% moisture

0.2% moisture

DFR results for a transfomer at

various temperatures
Temp

Moisture, % Oil conductivity, pS

21

2,4

10,4

27

2,3

13,8

34

2,4

22,8

49

2,3

39,3

DFR data acqusition is pending

insulation temperature
100,00

Frequency, mHz

10,00

eV=0,9
eV=0,7
eV=0,5
1,00

Corresponding data points

0,10
0

10

20

30

40

50

60

Insulation
Temperature

Ongoing project collecting measurement

results on various transformers

Old distribution transformers

New power transformers in factory
New power transformers in the field
Typical power transformers in various
conditions

Moisture assessment of transformers with

different low frequency limits
10,0

Moisture level, %

T1, 3C
T2, 7C
T3, 15C

1,0

T4, 15C
T5, 21C
T6, 23C
T7, 25C

Low frequency limit, mHz

0,1
0,1

10

Distribution transformer
T = 23C, f = 0.1-10mHz

Distribution transformer
T = 23C, f = 0.1mHz-10kHz

Distribution transformer
T = 23C, f0 = 0.1-10mHz
7

Auto
geometry

X (auto)
Y (auto)
Moisture

Oil, pS
2

1
Stop freq, mHz
0
0,1

10

Power transformer
T = 25C, f = 0.1mHz-1 kHz

Power transformer
T = 25C, f = 0.1mHz-1kHz

Power transformer, T = 25C, f0 =

0.1-10mHz
2

Auto
geometry

1,8
1,6
1,4
1,2

X (auto)

Y (auto)
Moisture, %

0,8

Oil cond, pS

0,6
0,4
0,2
Min freq, mHz

0
0,1

10

DR measurement frequency range

Conclusions so far...
Auto geometry estimation mode in MODS works
good
Limited value of measuring below 1-2 mHz at
normal temperatures (only a few results
collected so far from measurements at < 15C)
If geometry is (approximately) known, it may be
possible to reduce measurement time
Measurements at higher temperature can shorten
the measurement time

Summary and conclusions

Dielectric response measurement is an excellent
tool for insulation diagnostics

Moisture assessment using DFR measurements
and transformer insulation modeling is a
generally accepted standard diagnostic method

Transformer outage time is expensive and it is
necessary to minimize measurement time. DFR
measurements down to 1-2 mHz seem to be
sufficient for accurate moisture assessment at
normal temperatures

DFR is capable of identifying non-moisture issues
like contamination/sludge and/or conductive
layers

Questions and/or comments?

Additional material
Sweep Frequency Response Analysis

Application Examples

Time Based Comparison - Example

1-phase generator transformer, 400 kV

SFRA measurements before and after
scheduled maintenance
Transformer supposed to be in good
condition and ready to be put in
service

Time Based Comparison - Example

Obvious distorsion as by DL/T911-2004 standard (missing core ground)

Time Based Comparison After repair

Normal as by DL/T911-2004 standard (core grounding fixed)

Type Based Comparisons (twin-units)

Some parameters for identifying twin-units:

Manufacturer
Factory of production
Original customer/technical specifications
No refurbishments or repair
Same year of production or +/-1 year for large units
Re-order not later than 5 years after reference order
Unit is part of a series order (follow-up of ID numbers)
For multi-unit projects with new design: reference
transformer should preferably not be one of the first units
produced

Type Based Comparison - Example

Three 159 MVA, 144 KV single-phase
transformers manufactured 1960
Put out of service for maintenance/repair after
DGA indication of high temperatures
Identical units
SFRA testing and comparing the two
transformers came out OK indicating that
there are no electromechanical
changes/problems in the transformers
Short tests indicated high resistance in one
unit (confirmed by WRM)

Type Based Comparison 3x HV [open]

Type Based Comparison 3x HV [short]

3x HV [short] - details

Higher resistance on A-phase

Type Based Comparison 3x LV [open]

Design Based Comparisons

Power transformers are frequently designed in
multi-limb assembly. This kind of design can lead
to symmetric electrical circuits
Mechanical defects in transformer windings
usually generate non-symmetric displacements
Comparing FRA results of separately tested limbs
can be an appropriate method for mechanical
condition assessment
Pending transformer type and size, the frequency
range for design-based comparisons is typically
limited to about 1 MHz

Design Based Comparison - Example

40 MVA, 114/15 kV, manufactured 2006
Taken out of service to be used as spare
No known faults
No reference FRA measurements from
factory
SFRA testing, comparing symmetrical
phases came out OK
The results can be used as fingerprints
for future diagnostic tests

Designed Based Comparison HV [open]

Designed Based Comparison HV [short]

Designed Based Comparison LV [open]

Design Based Comparison

After Suspected Fault
Power transformer, 25MVA, 55/23kV,
manufactured 1985
By mistake, the transformer was
energized with grounded low voltage
side
After this the transformer was energized
again resulting in tripped CB
(Transformer protection worked!)
Decision was taken to do diagnostic test

Design Based Comparison

After Suspected Fault
10

100

1000

10000

100000

0
-10

Response (dBs)

-20
-30
-40
-50
-60
-70
-80
Frequency (Hz)

HV-0, LV open

A and C phase OK, large deviation on B-phase
(shorted turn?)

1000000

Design Based Comparison

After Suspected Fault
10

100

1000

10000

100000

Response (dBs)

-10

-20

-30

-40

-50

-60
Frequency (Hz)

HV-0 (LV shorted)

A and C phase OK, deviation on B-phase (winding
deformation?)

1000000

And how did the mid-leg look like?

Core limb
Insulation cylinder

LV winding

End