Measurement Considerations using
SFRA for Condition Assessment of
Power Transformers
Mats Karlstrom, Peter Werelius, Matz Öhlen,
Pax Diagnostics,
Lars Adeen and Eddie Brynjebo
E.ON-ES, Sweden
Weidmann, New Orleans, Sept 2008
Agenda
• Introduction
• Measurement example
• The consequence of bad connections
– Main conductor
– Grounding
• Instrument performance
• Understanding external parameters
affecting response from core
Weidmann, New Orleans, Sept 2008
SFRA testing basics
• Off-line test
• The transformer is a complex RLC circuit
• Impedance/response is measured (20Hz–
2MHz), and plotted as function of
frequency
• Measurements can be compared over
time, on the same transformer, between
“sister transformers” or between phases
• Potential problems are visible as changes
in amplitude and/or frequency shifts
• The method is unique in its ability to detect
winding deformations
Weidmann, New Orleans, Sept 2008
Test results – always comparisons
•
•
Specific transformer
elements are visible in
different parts of the curve
Low frequencies
– Core problems and
shorted/open windings
•
Medium frequencies
– Winding deformations
•
Taps and
connections
High frequencies
– Tap connections and other
winding connection
problems or changes
Weidmann, New Orleans, Sept 2008
Winding
deformations
Core + windings
Test results – always comparisons
Core NOT grounded
Core grounded
Reproducibility is of utmost importance!
Avoiding responses from external sources are essential in
order to optimize investment in time and instrumentation
Weidmann, New Orleans, Sept 2008
Example of reproducible results
• 105 MVA, Single phase Generator Stepup (GSU) transformer
• SFRA measurements before and after a
severe short-circuit in the generator using
– Two different test instruments (Same make
and model)
– Tests performed by two different persons
Weidmann, New Orleans, Sept 2008
Before (Test 1) and after (Test 2) GSU stressed
by short circuit currents, logarithmic scale
Weidmann, New Orleans, Sept 2008
Before (Test 1) and after (Test 2) GSU stressed
by short circuit currents, linear scale
Weidmann, New Orleans, Sept 2008
105 MVA, Single phase GSU
• Measurements “before” and “after” were
almost identical.
• Conclusion:
– Good correlation. Analysis established no
problems in transformer
– Transformer could be safely put back to
service
Weidmann, New Orleans, Sept 2008
Important parameters for reproducibility
• Good connections
• Good grounding practice
• High instrument dynamic range/low noise floor
through complete frequency range
• Understanding core property influence in lower
frequencies when performing “open” - circuit
SFRA measurements
Weidmann, New Orleans, Sept 2008
Bad connection
• Bad connections affected the curve in
higher frequencies
Weidmann, New Orleans, Sept 2008
Good connection
• After proper connections were made
Weidmann, New Orleans, Sept 2008
Typical effects of grounding practice
C. Homagk et al, ”Circuit design for reproducible on-site measurements of
transfer function on large power transformers using the SFRA method”, ISH2007
Weidmann, New Orleans, Sept 2008
Ground connection to ensure
repeatability at high frequencies
Good grounding practice;
use shortest braid from cable
shield to bushing flange.
Weidmann, New Orleans, Sept 2008
Poor grounding practice
Example cable set
Weidmann, New Orleans, Sept 2008
Instrument performance
• Smaller transformers often have very high
attenuation at first resonance
• Noise problem not necessarily originates
from substation static. Instrument noise
often the culprit
• Test with open cables (Nothing connected)
to determine the maximum dynamic range
Weidmann, New Orleans, Sept 2008
Inherent noise level – ”Noise floor”
”Open”/noise floor measurements
Two different make and models
Weidmann, New Orleans, Sept 2008
Effect of instrument performance
-70dB
-80dB
-100dB
Weidmann, New Orleans, Sept 2008
Dynamic Range – Comparison
H1 – H2 (open & short) measurements
Two different make and model
Weidmann, New Orleans, Sept 2008
Influence of core in “open” measurements
• Try to minimize the effect, however, some
differences are still to be expected and
must be accepted.
• Preferably:
– perform SFRA measurements prior winding
resistance measurements (or demagnetize
the core prior to SFRA measurements)
– Use same applied voltage level in all SFRA
measurements
Weidmann, New Orleans, Sept 2008
Measurement before and after
measurements of winding resistance
After
Before
After
Before
Weidmann, New Orleans, Sept 2008
Measurement after WRM measurements
and after demagnetization
Directly after WRM
After demag
Weidmann, New Orleans, Sept 2008
Influence of applied voltage level
(HV winding, 30MVA 31.5kV/0.525kV/0.525kV)
2.83 V peak-to peak
10V peak-to peak
Influence of applied voltage
is more pronounced on LV
windings
Weidmann, New Orleans, Sept 2008
Influence of applied voltage level
1 V peak-to peak
10V peak-to peak
Weidmann, New Orleans, Sept 2008
Other considerations
• SFRA (Sweep frequency response analysis)
provides good detail data in all frequencies
Black = Imported Impulse measurement
(Time domain converted to Frequency Domain)
Red = SFRA Measurement
Weidmann, New Orleans, Sept
Deviations Low Frequency = Method
2008Deviation High Frequency = Cable practice
Zoom View
Impulse instrument sample rate limts
frequency resolution to 2kHz.
Weidmann, New Orleans, Sept 2008
Field Verification Unit
Field verification unit with known frequency
response is recommended in various
standards to verify cables and instrument
Weidmann, New Orleans, Sept 2008
Summary
• The basis of SFRA measurements is comparison and
reproducibility is of utmost importance
• Small deviations in the low frequency range of open
measurements must be accepted since it depends on
core status that in many cases are difficult to control.
• To ensure high repeatability the following is important
– Use of a high quality, high accuracy instrument with inputs and
output impedance matched to the coaxial cables (e.g. 50 Ohm)
– Preferably use same applied voltage in all SFRA measurements
– Make sure to get good connection and connect the shields of
coaxial cables to flange of bushing using shortest braid
technique.
– Make good documentation, e.g. make photographs of
connections.
Weidmann, New Orleans, Sept 2008
Thank You!
• Questions or comments?
– Mats Karlstrom
Mats.karlstrom@paxdiagnostics.com
– WebSite: www.paxdiagnostics.com
Weidmann, New Orleans, Sept 2008