Deformation Measurement System for Buckling
Tests on Continuously Transposed Cable using a
High-Speed Camera
D. Geißler1* and T. Leibfried2
1,2
Institute of Electric Power Systems and High Voltage Technology
Karlsruhe Institute of Technology (KIT), Karlsruhe, Federal Republic of Germany
*
Email: daniel.geissler@kit.edu
Abstract— An optical deformation measurement system for
buckling tests on Continuously Transposed Cable (CTC) using
high-speed camera recordings and marker tracking is proposed.
The aim of the buckling test setup is to analyze the mechanical
strength of CTCs with respect to radially acting electromagnetic
forces as they appear in power transformers. Therefore a CTC
winding arrangement is fed by a short-circuit current while the
deformation is recorded by the camera system. The verification
of the new deformation measurement system is done by an
independent system based on acceleration sensors. Especially
operationally aged power transformers can be affected by high
short circuit currents, resulting in serious damages on windings.
Thus, the influence of ageing phenomena on the mechanical
strength of CTCs like degradation of the paper insulation shall
be investigated with the proposed system. Experimental results
are proposed for a factory-new CTC and the influence of paper
insulation on its stiffness is evaluated from the marker tracking
results during the short circuit tests.
I. INTRODUCTION
Short circuit events in power transformers can lead to
serious damage on windings or in the worst case, to a total
loss of the transformer. Especially, high radial
electromagnetic forces, as a result of short circuit currents
applying to the low voltage winding, may lead to the so called
radial buckling: Parts of the winding collapse and buckle
between the radial supports under the influence of the radially
inwards directed force [1]. It is a matter of common
knowledge, that the life of a power transformer is mainly
based on the aging of the cellulose insulation [2]. Especially
operationally aged transformers with degraded cellulose often
have a reduced short circuit strength. This may lead to
irreversible deformation of windings if high fault currents
occur. The cellulosic insulation around the conductors is
assumed to influence the mechanical strength of especially
Continuously Transposed Cables (CTCs) with their low
bending stiffness.
In the past there have been many investigations and
experiments concerning the short-circuit strength of power
transformers. In [3] the authors used acceleration sensors to
monitor axial and radial displacements of a winding during a
short-circuit event. In [4] strain gauges were used to measure
the coil displacements by converting the displacements into
strain in an acrylic resin plate. The authors of [5] designed a
special test set-up to represent a winding section of a real
978-1-4799-5183-3/14/$31.00 ' 2014 IEEE 111
transformer and to allow easy replacement of the test
windings as well as gather reproducible tests. Criteria for
identifying the initiation of the buckling process were derived
from Impedance Measurements and Frequency Response
Analysis (FRA).
The deformation measurement system for CTC buckling
tests proposed in this paper can be used to investigate the
mechanical strength of CTCs. Furthermore, the influence of
the paper insulation on the CTC stiffness can be evaluated.
The buckling tests are done with two single-layered CTC
windings, each having three turns, arranged around an iron
core. The inner winding is connected in series to the outer
winding by opposite winding direction. Thus, current flow in
inner and outer winding is directed in opposite. This leads to
an inwards directed radial force on the inner and an outwards
directed force on the outer winding. Especially the
appearance of radial buckling between spacers caused by the
inwards directed radial force is of interest. The necessary high
testing short circuit currents of up to 30 kA (peak) are
delivered by a high current transformer.
During the short circuit tests the CTC deformations are
recorded with a high-speed camera. The radial deflection of
the inner CTC winding between two spacers is especially of
interest. Therefore, optical tracking markers are placed on the
interesting CTC segments. With motion tracking algorithms,
the marker movement can be correlated to the testing current
which is recorded synchronously. A validation for the
measurement system is proposed by the use of calibrated
acceleration sensors placed near to the tracking marker
locations.
The last part of this paper shows some measurement results
from a factory-new CTC where the paper insulation thickness
is varied between original thickness and no paper. The
differences in the recorded radial deflections enable some
conclusions concerning the influence of the paper insulation
on the CTC bending stiffness.
II. FUNDAMENTALS
A. Structure of CTCs
Continuously Transposed Cables (CTC) are stranded
multiple conductors which are often utilized in power
transformers. They offer the advantage to reduce eddy current
and loop voltage losses and reach low winding costs.
Furthermore, the winding space factor is increased as a result
of the thin enamel based insulation. The whole CTC is
usually wrapped by several layers of insulating paper.
A CTC consists of rectangular shaped enamel coated
copper strands, which are arranged in two stacks (Fig. 1A).
Usually a number of 11 to 81 strands are used in total. Strand
thickness in radial direction is much lower than its height
(Fig. 1B). The reason for this is the reduction of loop
voltages as a result of radial varying magnetic stray fields.
Loop voltages can also arise between the strands as they are
connected in parallel at the winding endings. To compensate
the corresponding losses the single strands are transposed
with each other: After a fixed distance, called transposition-
pitch, the two radial outer- and innermost lying strands
change their stack. Due to these regular changes in radial
position each strand is exposed to the same average stray field
along one winding turn and the induction of loop voltages is
avoided. For electrical insulation between turns the CTC is
wrapped by several layers of paper. Mainly the small strand
thickness is responsible for the reduced bending stiffness as
long as no epoxy bonded CTCs are considered. That is why
the buckling test setup focuses on radial buckling effects.
A) B)
axial winding direction
Enamelled strands
hS
tS
Paper
Transposition
insulation
radial winding direction
Fig. 1: CTC with 21 strands and paper insulation partially removed (A),
CTC cross-section (B)
B. CTC test setup
In order to realize testing conditions similar to the short
circuit conditions in a power transformer, the CTC test setup
is derived from the mechanical structure of transformer
windings. A detailed description of short-circuit forces in
power transformers can be found in [6].
A top view for the test setup is shown in Fig. 2. Two CTC
based windings are wound in three turns around a core. The
inner winding is supported against the core by the use of axial
spacers. Both windings are connected in series so that an
opposite current direction between inner and outer winding is
enforced. This leads to a radially inwards directed
electromagnetic force acting on the inner CTC turns. The
outer turns are pressed radially outwards. As a result of the
windings´ series connection, only one feeding current source
is needed. Furthermore there is no need for a closed iron core.
This enables an easy exchange of windings. In addition, the
sheeted core can be built up as hollow cylinder to cut down
weight and material consumption. All winding endings have
to be clamped properly to prevent slippage of CTC in
circumference direction as a result of radial forces. For better
clarity the destined CTC clamping supports are not part of
Fig. 2. Their purpose is to add tension to the windings and
112
offer electrical connection to the CTC endings. More details
concerning the test setup can be found in [7].
CAD
Photography:
iron core
inner winding
outer winding
axial spacers
1m
current series
direction connection
Fig. 2: CAD model and photography of the CTC test setup
C. Measurement system using acceleration sensors
For radial deflection measurements, a piezoelectric
acceleration sensor is placed in the middle of a CTC segment
between two axial spacers, Fig. 3. Through double integration
of the acceleration time signal a(t) using signal processing
according to Fig. 4, the radial deflection signal s(t) can be
calculated. The signal processing steps between the
integration steps are necessary to remove offset signals or low
frequent drifts that lead to enormous errors after passing the
integration. The double integration of a constant offset in
acceleration data for example leads to a huge parabolic error
in distance. Thus, for example in [3], where acceleration
sensors were used for coil displacement measurements
without adequate signal processing resulting in displacement
time series with parabolic shapes. The disadvantage of the
signal processing procedure according to Fig. 4 is that all
information concerning the CTC stationary deformation after
the short circuit is lost, as it is based on the low frequent
acceleration data. Thus, only the transient movement can be
calculated. The measurement system proposed in the
following chapters, however, can record the persistent
deformation in addition. This is important to investigate the
point of plastic deformation and the beginning of buckling.
A) B)
core acceleration
sensor
g
tape
Fig. 3: Positioning of acceleration sensor on the inner CTC winding centered
between to axial sticks (A), Cross-section of mounting with tape (B)
 a(t) + detrend + + s(t) directly onto the CTC´s paper insulation or for the covered
- (linear) ʃ -
detrend
(linear) ʃ - turns onto the mounting plates. These plates have to be glued
v(t) properly to the CTC´s outer insulation.
lower
mean mean
envelope Before a short circuit test, the camera has to be positioned
properly so that the objective is perpendicular to the plane in
Fig. 4: Signal processing path for calculating the displacement signal s(t) out which the markers are located. Furthermore, the camera has
of acceleration signal a(t)
to be centered with respect to the markers that shall be
III. OPTICAL TRACKING SYSTEM recorded. To receive precise tracking results out of the
recorded videos, the camera has to be calibrated for each test
A. Principle functionality setup. Therefore a ruler is placed into the region of interest
A segment between two spacers as region of interest (ROI) (ROI). Fig. 6 shows the ruler placement inside the second
is observed from above using a high-speed camera (Fig. 5). In marker plane for the middle CTC turns. As only radial
order to dissolve the 100 Hz periodic deflection with a movement is expected, a calibration into this direction is
sufficient time resolution the minimum camera frame rate is sufficient. The markers mainly move along a radial oriented
chosen to 1000 fps. Motion tracking markers are placed on line. If perspective distortion would be a problem, a more
the CTC´s middle between two spacers. Here, the radial elaborate 2D calibration or grid calibration has to be chosen.
deflection will be at its maximum. For calibration purposes a With ROI-length and the camera resolution
ruler is placed near to the markers in radial direction. To (pixel) in the corresponding direction the length-resolution
avoid perspective distortions the ruler has to be placed into can be calculated according to (1).
the same plane together with the markers. For observing the
middle or lower turns that are covered by the upper CTC the lrROI  LROI wROI (1)
markers can be placed on special mounting plates. To
compensate residual movements from low frequent camera or Typical values for the test setup and camera system
test setup fluctuations, a so called reference marker is fixed to proposed in this paper lead to a length resolution of about
the iron core. Its purpose is to compensate the distorted 97.7 μm/pixel. This value states that the tracking marker has
deflection signals from the tracking markers on the CTCs. to move by at least 97.7 μm until the movement can be
recognized by the camera sensor. The real length resolution is
much higher because the CCD sensor pixels can resolve
HS camera different brightness levels and not only black or white. Thus,
can be seen as the lower limit. None the less, has
to be maximized by the camera arrangement in order to gain a
LROI high sensitivity for the marker tracking. The camera
Mref M1 M2 orientation and ROI adjustment according to Fig. 7, for
example, leads to a better resolution compared to Fig. 6. In
addition, cropping the ROI to a small stripe enables higher
M3 M4 frame rates for many high-speed camera models as fewer
CCD sensor lines have to be read-out.

marker
mounting Mref
Mx: tracking
plate marker
position
M1
inner CTC outer CTC
iron core winding winding M3

Fig. 5: Cross-sectional view through test setup (not to scale) with positioning y M2
of high-speed camera and mounting of markers M1 - M4 and reference M4
marker Mref. Region of interest (ROI) is marked by LROI x

B. Marker positioning and Calibration

For motion tracking a high contrast marker type, shown in Fig. 6: Example for tracking marker positioning with calibration ruler placed
the right bottom edge of Fig. 5, is chosen. This kind of inside the marker plane defined by M3 and M4
marker allows automatic detection through the motion
tracking software and reliable tracking for the implemented
picture correlation algorithms. The markers are stuck either

113
 y M4 M2 M3 M1 Mref
x

Fig. 7: Optimized ROI selection for a better length-resolution compared to

the ROI shown in Fig. 6.

C. Radial deflection calculation with compensation Fig. 9: Trajectory plot from x- and y-deflection with best-fit ellipse. Error in
For a marker with the label M, the motion tracking camera justification angle is deviated from major axis orientation
software estimates the absolute coordinates and over
recording time. In a first step these coordinates are converted This calculation shall be retraced with the sample
into relative coordinates and using the measurement shown in Fig. 10. For the sake of clarity only
two of the markers shown in Fig. 6 are considered (marker
corresponding marker origin as reference point:
M3 and the reference marker Mref fixed to the iron-core). The
xM  xM  xM 0 , yM  yM  yM 0 (2) red lines correspond to the relative coordinates of M3, while
the blue lines correspond to the reference marker´s relative
The compensation of residual reference marker movements coordinates. The compensated coordinate-vectors and
is realized by subtracting the reference marker´s relative are drawn in black. Especially, the x-component of Mref
coordinates and . This leads to the compensated,
shows a significant reference marker movement in the form
relative coordinate-vectors and according to (3). of a low frequent oscillation that is also overlaid to the x-
xM,c  xM  xref , yM,c  yM  yref (3) component of M3. After compensation, this residual
component is successfully eliminated and does not affect the
The radial marker movement has to be calculated from radial movement signal.
both x- and y-component. Only if the camera´s y-axis would
be perfectly aligned into radial direction, the x-component of
marker movement could be neglected. In order to correct
small deviations in camera justification the angle α between
the camera´s y-axis and the radial movement direction of the
marker is estimated from the trajectory plot of and
(Fig. 8). The shape of the trajectory usually is an ellipse
for the movement data captured with the test setup of this
work, Fig. 9: Both direction signals have the same frequency
but a small phase shift due to residual marker movements in
winding direction. For this purpose the major axis of the best
fit ellipse is calculated, to extract the angle α. The radial
component can then be calculated as shown in (4) with
being the main direction of marker movement. This
compensation method has not significant influence for
camera setups as they can be seen in Fig. 7 where α usually
lies between 2° and 5°. It is more relevant for ROIs as shown
in Fig. 13 with a parallel observation of 2 or more CTC
segments.
rM,c =  yM,c cos  α  (4)

dy
x-direction

α
dx
dr
Fig. 10: Short circuit test with a duration of 230 ms. Tracking results for one
CTC Trajectory of marker and a reference marker divided into x- and y-direction (upper and
marker movement y-direction middle plot) and compensated radial marker movement (lower plot).

Fig. 8: Schematic o ma k t aj to y o a am a misalignm nt by α

114
D. Verification of accuracy
The accuracy of marker tracking during the short-circuit
tests was verified by the use of an acceleration sensor based
measurement system presented in [7]. In order to receive the
transient, 100 Hz periodic displacement components, the
signal processing according to Fig. 4 is implemented. The
removal of mean and linear drift parts from acceleration and
velocity however eliminates all information concerning
persistent CTC deformation at the end of the short-circuit
tests.
For marker tracking verification, a calibrated acceleration
sensor is placed near to marker M3 as shown in Fig. 7. A
second reference acceleration sensor placed on the iron core
is used similarly to the marker tracking method. The
comparison of radial displacement from tracking the marker
M3 and the transient displacement calculated from
acceleration data is visualized for two different short circuit
currents in Fig. 11 and Fig. 12. The first test T1, with a steady
state peak current of 10.7 kA, causes a radial deflection that is
near to the worst-case length-resolution which was Fig. 11: Short circuit test with steady state testing current of 10.7 kA: Radial
deflection estimated from acceleration sensor measurement (black), radial
estimated with formula (1) to 97.7 μm/pixel. Furthermore no
deflection estimated from marker M3 by the tracking algorithm (red).
significant persistent deflection can be derived from the
tracking results. The second test T2, with a steady state peak
current of 18.4 kA, results in a much higher transient
deflection. Again, the tracking shows no persistent radial
deflection which was requested for better comparability to the
results from the acceleration sensor measurements.
To compare the measurement systems, the relative error err
between the deflection from acceleration sensor a and
deflection from the marker tracking is defined according
to (5) with the assumption for a being the real deflection.
rM  racc
err  (5)
racc

The relative error for both short-circuit tests T1 and T2 is

calculated in Table 1. It is distinguished between the error
ss of steady-state deflection and the error k of peak
deflection. Both values lay around 10 % for both of the tests.
This is an acceptable result, especially with regard to the
advantages the high-speed camera tracking system offers:
 Estimation of persistent deflection Fig. 12: Short circuit test with steady state testing current of 18.4 kA: Radial
 Tracking of multiple markers from one recording deflection estimated from acceleration sensor measurement (black), radial
deflection estimated from marker M3 by the tracking algorithm (red).
 Immunity against high magnetic fields
IV. APPLICATION OF MEASUREMENT SYSTEM
Table 1: Relative error between acceleration sensor measurements and
tracking measurement for the two short circuit test T1 and T2 A. CTC test setup configuration and testing method
As an application for the measurement system, an
ss a ss k a k ss k
investigation of insulation paper influence on the radial CTC
(mm) (mm) (mm) (mm) (%) (%) stiffness shall be presented. Therefore two segments are
T1 0.122 0.138 0.190 0.209 -11.6 -9.0 observed by the high-speed camera in parallel (Fig. 13). All
CTCs inside the left segment S4 are in their origin state, while
T2 0.609 0.678 1.072 1.226 -10.2 -12.6 the inner CTC turns inside the right segment S5 were stripped
of the insulating paper. As the middle turn is exposed to the
highest radial electromagnetic force, the markers M4 and M5

115
are tracked in addition to the reference markers Mref,4 and
Mref,5.
From tracking results, the influence of paper insulation on
the CTC stiffness shall be evaluated. Furthermore, the CTC´s
resistance against radial buckling can be compared. The
specifications of the CTC under test and some dimensions of
the winding are listed in Table 2.
As perspective distortions are not insignificant, two
calibration rulers are placed near to the marker line in the
middle of each segment. Furthermore, each segment has its
own reference marker. The parallel observation of two
segments leads to a worst-case length resolution of about
= 120 mm / 512 pixel = 234 μm/pixel.
Short-circuit tests are started at low testing currents so that
in the beginning both CTC segments only deform elastically.
With every test, the short-circuit current is increased until a
distinct permanent deformation occurs. During each test the
marker movement is tracked by the high-speed camera and
the current with time is recorded.
As the CTC test setup is approximately cylindrically
symmetric except for the winding endings a 2D rotation
symmetric FEA based magnetic fields simulation can be used
to calculate the electromagnetic force over time for each turn
from current over time. This enables a correlation between
CTC deflection and the mechanical line load acting on the
conductor. The detailed procedure for the FEA simulation can
be found in [7].
Mref,4 Mref,5
M4
M5
y S5
S4
x
Fig. 13: Test setup for short-circuit tests on to CTC segments: With paper
insulation (left) and without paper insulation at the inner turns (right).
B. Measurement results
From the time courses of radial marker movement the
radial peak deflection i k during the first half wave as well
as the persistent deformation i n at the end of each test are
evaluated. Furthermore the radial peak line load is
calculated from the peak current during the first half wave
using the FEA based 2D model of the test setup.
Fig. 14 shows the radial deflection of M5 for two example
short circuit tests: One without and the other with persistent
deformation at the end of the test.
In Fig. 15 both peak and persistent radial deflections for the
short-circuit tests are plotted over . For better visualization
the absolute values with respect to the marker position in the
beginning of the tests are visualized. For the first two testing
currents resulting in a line load of 6.67 kN/m and 7.56 kN/m
both segments deform linear elastic, as no persistent
deflection is notable. These points can be used to compare the
CTC stiffness for loads stressing the copper beyond the yield
point. Fig. 16 shows the quotients from the relative radial
peak deflections. The first two values state an increase of
CTC stiffness by the paper insulation by a factor of 2.2. For
increasing loads, the factor is varying as the copper Young´s
modulus is reduced as result of plastic stain. As soon as a
significant persistent deformation occurs, the CTC inside the
corresponding segment begins to buckle and deforms
plastically.
The exact load for a beginning of plastic deformation
however, cannot be identified reliably out of the absolute
persistent deflections (Fig. 15) because of the large gap
between 7.56 kN/m and 8.78 kN/m. Thus, the highest line
load for each segment without persistent deflection is
compared. This results in an increase of critical line load from
7.56 kN/m to 9.25 kN/m by 22.4 % as effect of paper
insulation. The plastically deformed segments after the short
circuit tests are shown in Fig. 17 and illustrate the influence
of the paper insulation impressively. Above 11.6 kN/m
marker tracking for the segment without paper insulation had
to be stopped because the radial deflection was too high so
that the marker mounting plate loosened from the CTC.

Table 2: Specifications of CTC under test

core outer diameter 990 mm

number of axial spacers 24
number of single strands 21
strand height 8.30 mm
strand radial thickness 1.40 mm

copper yield strength Rp 0.2 7 - MPa

Fig. 14: Time course of radial movement of marker M5 for test no. 2 and 4
insulation thickness per side 0.9 mm for the CTC segment without paper insulation

116
 12 V. CONCLUSION
S4: r_peak S5: r_peak
10 S4: r_persistent S5: r_persistent A high-speed camera based tracking system for
deformation measurements of CTCs during short-circuit tests
has been developed. The use of optical markers and motion
radial deflection (mm)

8
tracking software has been successfully verified by another
6 measurement system based on acceleration sensors. The
deviation between both systems is about 10 %, even for small
4 deflections of 100 μm.
The tracking system has the advantage to recognize
2 persistent deflections which are an important indicator for
plastic CTC deformation and the beginning of buckling. The
0 practical application of the measurement system has been
6 7 8 9 10 11 12 13 shown with an investigation of the influence of the paper
radial peak line load (kN/m)
insulation on the CTC stiffness. The parallel observation of
Fig. 15: Measurement results for segment 4 & 5: Peak and persistent radial two CTC segments - one with and the other without paper
deflections (absolute values with respect to the original marker positions) covering - leads to an increase of linear stiffness by a factor
5
of 2.2 as result of paper insulation. The increase in critical
buckling line load has been estimated to 22.4 % if the paper
rM5,pk / rM4,pk
4
covering is available.
ratio (1)

3
REFERENCES
2

1
[1] K. Hiraishi, Y. Hori, and S. Shida, Mechanical strength of transformer
windings under short-circuit conditions, IEEE Trans. Power App. Syst.,
0 vol. PAS-90, no. 5, pp. 2381–2390, Sep. 1971.
[2] B. Gorgan, P. V. Notingher, Calculation of the Remaining Lifetime of
6 7 8 9 10 11 12 13
Power Transformers Paper Insulation, Optimization of Electrical and
radial peak line load (kN/m)
Electronic Equipment (OPTIM), 2012.
Fig. 16: Quotient from radial peak deflection of segment 5 (with paper) and [3] Y. H. Oh, K. D. Song, Displacement measurement of high-voltage
segment 4 (without paper) winding for design verification of short-circuit strength of transformer,
IEEE PES Transmission and Distribution Conf. and Exposition, 2003.
[4] K. Kurita, T. Kuriyama, K. Hiraishi, S. Kusumoto, S. Shida, Y. Hori,
Mechanical Strength of Transformer Windings under Short-Circuit
Conditions, IEEE Trans. Power App. Syst., vol. PAS-88, issue: 3,
March 1969.
[5] M.P. Saravolac, P.A. Vertigen, C.A. Sumner, W.H. Siew, Design
Verification Criteria for evaluating the Short Circuit Withstand
Capability of Transformer Inner Windings, Cigre 2000, Paper No. 12-
208.
[6] S. V. Kulkarni, S. A. Khaparde, Transformer Engineering - Design and
Practice, Marcel Dekker Inc., 2004.
[7] D. Geißler, T. Leibfried, Test setup and results concerning the
mechanical strength of continuously transposed conductors for power
transformers, presented at the 18th Int. Symp. High Voltage Eng.,
Seoul, Korea, Aug. 25-30, 2013, paper PC-06.
Fig. 17: Test setup after short-circuit tests: Segment S5 without paper
insulation (right) shows significantly stronger buckling than segment S 4 with
paper insulation.

117