You are on page 1of 6

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 17, NO.

1, JANUARY 2002

155

The Impact of Inrush Currents on the Mechanical


Stress of High Voltage Power Transformer Coils
Michael Steurer, Member, IEEE, and Klaus Frhlich, Senior Member, IEEE

AbstractFrom failure experience on power transformers, it


was very often suspected that inrush currents, occurring when energizing unloaded transformers, were reason for damage. In this
paper, it was investigated how mechanical forces within the transformer coils build up under inrush compared to those occurring
at short circuit. Two-dimensional and three-dimensional computer
modeling for a real 268 MVA, 525/17.75 kV three-legged step up
transformer was employed. The results show that inrush current
peaks of 70% of the rated short circuit current cause local forces
in the same order of magnitude as those at short circuit. The resulting force summed up over the high voltage coil is even three
times higher. Although inrush currents normally are smaller, the
forces can have similar amplitudes as those at short circuit however with longer exposure time. Therefore, care has to be taken to
avoid such high inrush currents. Today controlled switching offers
an elegant and practical solution.
Index TermsControlled switching, electromagnetic forces, finite element method, inrush currents, power transformer, short
circuit currents.

NOMENCLATURE
SC
IR
LV
HV
TC1, TC2
FEM
TOI
MMF,

Short circuit.
Inrush.
Low voltage (winding or coil).
High voltage (winding or coil).
Tap changer (winding or coil).
Finite-element method.
Theory of images (method).
Magnetomotive force (in ampere turns).
I. INTRODUCTION

OWER transformers in electric systems are one of the most


cost effective apparatuses. Breakdown causes high costs
for repair or replacement and may often lead to very high costs
due to temporary loss of the power delivery capability. Therefore, power transformers are highly protected against various
causalities of failure. One major reason for internal faults is attrition of the winding and conductor insulation due to vibrations
initiated by the electromechanical forces at service current and
over currents. Accordingly, the coils of transformers are mechanically supported and prepressed by bandages, wedges, and
Manuscript received September 22, 1999.
M. Steurer was with the High Voltage Lab, Swiss Federal Institute of Technology, Zurich CH-8092, Switzerland. He is now with the Center for Advanced
Power Systems, Florida State University, Tallahassee, FL 32310 USA (e-mail:
steurer@caps.fsu.edu).
K. Frhlich is with the High Voltage Lab, Swiss Federal Institute of Technology, Zurich CH-8092, Switzerland (e-mail: froehlich@eeh.ee.ethz.ch).
Publisher Item Identifier S 0885-8977(02)00589-7.

heavy bolts. The dimensioning criteria for these support structures are usually the forces caused by the highest possible current peak which normally occurs under short circuit (SC) conditions. Therefore, type testing of transformers ask for SC tests
only, as it is assumed that this is the worst case with regard to
the maximum forces.
In the past, it was discussed very often in several panels to
what extend IR currents are a jeopardy for the transformer windings. As it is well known, the value of these IR currents occurring when energizing unloaded transformers can be in the same
order of magnitude as a fault current [1], thus producing significant electromagnetic forces. The difference to SC is that the IR
may last for several tens of seconds, whereas a fault normally is
cleared within tens of milliseconds. In addition, an IR situation
is much more frequently as this can be considered as normal
service operation.
Recent cases of insulation failures in power transformers
which were frequently energized under no load condition
support the suspicion of many people that IR currents have
a jeopardous effect. It is assumed that due to the unbalanced
MMF and high saturation of the transformer core local (axial)
forces under severe IR conditions, especially on the TC coils,
could exceed the values taken as dimensioning basis. So, it
seems likely possible that conductor and winding insulation is
damaged due to mechanical overstress at IR.
Although it is requested by many utilities that high IR currents shall indeed be avoided from a systems operation point of
view, possible damaging effects to the transformer itself have
rarely been investigated. To the authors knowledge, only one
paper deals with forces on transformer windings at IR conditions where it is shown that the windings exhibit fairly large
axial forces, exceeding those built up under short circuit conditions [2].
Out-of-phase synchronization may also cause unbalanced
MMF conditions in step up transformers. Force calculations
show that the axial forces in the windings can be two to ten
times higher than under SC conditions [3]. However, the
magnetic field pattern caused by such erroneous operations are
different to the ones at IR conditions, because when energizing
unloaded transformers the secondary (mostly the low voltage)
side is not excited at all.
A first rough estimation of the distribution of the radial magnetic field component along transformer coils was presented in
[4]. The results show that the local radial magnetic field at the
ends of tap changer coils to be up to 80% higher at IR compared
to SC conditions, assuming the IR current to be as high as the
SC current.

08858977/02$17.00 2002 IEEE

156

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 17, NO. 1, JANUARY 2002

coil. Although in reality, the coils are rather complex in their mechanical structure, they are treated as solid cylindrical conductors here, so pressure and tensile stress are both possible inside
the coils. This is to be independent of any particular winding
support structure when calculating the forces for the purpose of
comparison.
Following these approximations the specific field calculation
method employed shall be explained along with the comparison
of 2-D and 3-D calculations. In addition, details of the method
used for calculating the axial and radial forces will be discussed
in the following sections.
A. Magnetic Field Calculation With FEM

Fig. 1. (a) Principal sketch of a transformers cross section with relevant force
directions (only coils on the middle leg shown); (b) cross section of 2-D FEM
model.

In order to learn more about how the IR current produces


mechanical forces to the transformer windings, the situation of
a three-legged step up transformer (data see the Appendix) was
modeled by means of a suitable computer code. Two- and three
dimensional (2-D, 3-D) calculations were carried out. The goal
was to calculate the forces in qualitative comparison to those
which occur at SC. The result should also give information on
the technical benefit if the IR current is avoided by controlled
switching, which seems to be the most effective approach for
the time being [5], [6].
Starting with general statements on force calculations the following chapter will explain reasons for choosing 2-D FEM and
how the mechanical stress on the coils was calculated from the
local magnetic stray field density.

As stated above, force calculation can be reduced to field


calculations when the current density is impressed. So manufactures of power transformers usually employ the method
of theory of images (TOI) to calculate the mechanical stress
of windings and coils for dimensioning purpose. This method
is based on the assumption, that the non-saturated iron cores
and yokes have infinite magnetic permeability and, therefore,
field to be perpendicular at their boundaries. An
force the
appropriate 2-D mapping of the actual 3-D geometry and the
exploitation of the SC conditions (balanced MMFs) leads to
fairly fast and accurate calculation tools. However, it is not
proven until now whether this method provide the electrodynamic forces correctly under IR conditions where unbalanced
MMFs highly saturate the iron and, therefore, reduce its
magnetic permeability significantly. Furthermore, the magnetic
field of unbalanced MMFs decreases only linear with distance
whereas fields from quasi-dipole (balanced) MMFs decrease
with the square of the distance.
All these considerations gave reason not to choose any transformer design tool employing the TOI but the commercially
available program MAXWELL from ANSOFT Inc. based on
the finite element method (FEM) [7].

II. SETUP FOR FORCE CALCULATIONS


For better understanding of the following, Fig. 1(a) shows
the 3-D model of the three-legged transformer with HV, LV,
and TC coils on the middle leg. The relevant forces and their
direction are indicated. Throughout this work, axial (index )
always means in the -direction (coils axis) and radial (index
) means perpendicular to the coil. Furthermore, in Fig. 1(a), the
cut planes and used for comparing 2-D with 3-D models
are shown.
According to electrodynamics the local force density in the
coils volume is equal to the outer vector product of the local
current density in the coil and the local magnetic (stray) flux
density as given by
(1)
Since only the maximum forces caused by the peak of the IR
and SC current are of interest and the field is in phase with
the exciting MMF (at the power frequency of 50/60 Hz), a dc solution of the problem is sufficient. That means is independent
and must be impressed. Therefore, calculating the local
of
force density and the total force on coils can be reduced to the
calculation of the flux density in each point of the respective

B. Comparison of 2-D and 3-D Model


To avoid unnecessary complexity of the model, the actual
3-D geometry of a three-legged power transformer as shown in
Fig. 1(a) was simplified into a (1-phase) 2-D equivalent axial
symmetric geometry [Fig. 1(b)]. The yokes in this 2-D model
are shaped in a special way to ensure correct representation of
the non-axial symmetric yokes (see the Appendix).
To verify the results gained with this 2-D representation a 3-D
FEM model has also been set up [geometry see Fig. 1(a)]. Both
models (2-D and 3-D) include the non linear magnetizing curve
of the iron.
The 2-D model considers only excitation of the coils on the
middle leg which is actually a condition not possible for a three
phase transformer with the HV side star connected. Therefore,
in the 3-D model, the coils on adjacent legs were excited in
opposite direction as it occurs as a rule for IR [8]. The results
show that field patterns at IR are similar, however with higher
magnitude because of the higher MMF.
With only the coils on the middle leg excited, comparison of
field values in the coils of the 2-D model shows high conformity with the 3-D results in the cut plane under the yoke
[Fig. 1(a)] for both IR and SC conditions. In the cut plane

STEURER AND FRHLICH: THE IMPACT OF INRUSH CURRENTS ON THE MECHANICAL STRESS OF HIGH VOLTAGE POWER TRANSFORMER COILS

157

1) Axial Forces: To derive the local force distribution,


has to be integrated along the coil. This was performed numeriequidiscally by summing up the values calculated at
tant points per line as shown in
(6)
Fig. 2. HV coil with the lines 1, 2, 3 along which the
and cross section dA for force calculation.

B field was calculated


where

perpendicular to the yoke values of


differ not more than
20% to those in the plane at IR conditions.
From this, it was concluded that a model that considers only
the coils on the middle leg is adequate for the purpose of qualitative comparison of forces at IR and SC for a three-legged transformer. This aspect and the high conformity between 2-D and
3-D results were taken as legitimization, to use the 2-D model
for the force and stress calculations presented here.
In the following, the strategy used for analyzing the field
to calculate the axial and radial mechanical stress is introduced.
C. Field Density and Force Distribution in Coils
It is plausible that the magnetic field can not be assumed to be
constant over a coils radial thickness. Therefore, the field was
calculated along three lines in discrete distance across the coils
(see lines
, 2, 3 in Fig. 2).
width
However, the program MAXWELL does not provide plotting
of the local (axial) pressure and tensile stress distribution along
a coils height. Therefore, the local force has been calculated as
follows.
It is assumed that
(2)
and, therefore

MMF in coil;
coils height
axial distance between points, correin Fig. 2
sponds to
radial component of on line
Note that although the employed software only allows equidistant points, this is not necessary for numerical integration. Fiwill give the total force distribunally, the sum of all three
tion in axial direction to
(7)
The external axial force on the coil, which is the resulting value
at the coils end (integration over the coils height), must be
in the
absorbed by axial suspensions. The maximum of
coil itself is usually much higher than the latter and gives the
dimensioning basis for the local stress, especially in regard to
buckling of windings.
2) Radial Pressure: In contradiction to the axial forces the
ones in radial direction shall not be summed up along the coils
height. In fact, they are better described by radial pressure stress
which must be supported either by the suspension or by the conductors themself in azimutal direction (tensile stress).
distribution along one of the
With the knowledge of the
, 2, 3 in Fig. 2) the local pressure on an cylindrical
lines (
stripe (with the radius and the height ) can be derived from
(5) to

(3)
(8)
That means the current density is azimutal only and constant
over the coils cross section (due to its subdivision into single
turns all carrying the same current). According to (1), that means
there are only axial and radial components of the forces.
Equations (4) and (5) show the axial and radial component of
the force along each of the three lines on an infinitesimal thick
) building
winding ( ) of one third of the coils thickness (
in Fig. 2.
the cross section
(4)
(5)
where
radial position of the line (
, 2, 3)
current density (constant) in the coil
coils thickness
radial and axial field component.
From (4) and (5) the axial and radial mechanical stress, local
and on the entire coil can be derived as it is explained in the
following two chapters.

The three radial pressures can be summed up as


(9)
the axial compowith good accuracy because of
nent of is nearly linear dependent of . This allows the need
for integration in radial direction to be reduced to summing up
represents the radial mechanical
the three values. Finally,
stress distribution along a coils height.
III. RESULTS
As the highest possible IR current peak
is usually smaller,
but in the same order of magnitude as the highest possible SC
, calculations varying the ratio
from
current peak
0.25 to 1 have been carried out. The following calculations presented in Figs. 37 show that equivalence in force occurs at a
of about 0.7. For the individual excitations of the
ratio
coils see the Appendix.
Fig. 3 shows the distribution of the radial component of the
magnetic stray field along the transformers upper TC2 coil.

158

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 17, NO. 1, JANUARY 2002

= 2 of (upper) TC2 coil.

Fig. 3.

2-D and 3-D values of B along line k

Fig. 4.

Resulting axial force distribution along upper TC2 coil.

Fig. 5. Axial force distribution along the HV coil.

Since the radial component


is nearly constant over the coils
(see Fig. 2) are
thickness, only the values along line
values at IR
plotted. It can clearly be seen in Fig. 3 that
conditions are higher than the SC values, especially at the coils
). Furthermore, Fig. 3 shows good
upper end (for

Fig. 6.

Radial pressure on the (lower) TC1 coil (proportional to B ).

Fig. 7. Ratio of axial forces and radial pressure at IR over SC against the ratio
IR current over SC current.

agreement between the 2-D (solid and broken lines) and 3-D
model (dotted lines; the discontinuities are due to a rather coarse
FEM mesh in 3-D). Fig. 4 shows the axial force distribution
at the coils end
along the TC2 coil, with the resulting force
to be approximately 20% higher at IR than at SC, although the
values (Fig. 3) are locally more than twice as
corresponding
. As the value
high. This is due to the assumption
at the coils lower end has a positive value the coil is subject to
an upward force.
In the same manner, Fig. 5 shows the axial force distribution
along the HV coil. In contrary to the TC coils, the force mainly
is compensated within the coil with an inner peak of axial presapproximately 30% more at IR. However, the resure stress
summed up over the coil is three times higher
sulting force
at IR than at SC and even in the opposite direction.
, the radial pressure stress
For the chosen ratio
, which is proportional to , is smaller at IR compared to SC
in the HV coil, but approximately the same in the TC coils as
shown for TC1 in Fig. 6.
According to [1] and unpublished information of transformer
may vary from 0.15 to 0.6.
manufacturer, the ratio
Therefore, in Fig. 7, the ratio of peak ( ) and resulting ( )

STEURER AND FRHLICH: THE IMPACT OF INRUSH CURRENTS ON THE MECHANICAL STRESS OF HIGH VOLTAGE POWER TRANSFORMER COILS

axial forces as well as the radial pressure peak ( ) at IR over


. Note that for the HV coil
SC are plotted against
differ from
, whereas for the TC coils (TC2)
(see also Fig. 4). As the ratio for the investigated transformer
, the maximum of the axial pressure stress in
is
the HV coil at IR is approximately 40% of that at SC (point A).
do not apply for this parAlthough values of
ticular transformer, the presented dependencies of
and
seem to be realistic estimations for other transformers even if values may vary with geometry. Following the
line of the peak force ( ) in the HV coil, the indicated limit
, where forces at IR are equal to the ones at SC,
(point B). For the axial and radial
is reached at
stress of the TC coils (TC2), the relations are about the same.
is higher at IR
Only the resulting axial force on the HV coil
range.
nearly in the entire

159

good concordance. The calculated ratio of forces at IR over SC


proofs that these forces are higher at IR than at SC for IR current
peaks of more than 60% of the rated SC current. Concluding
from the results the following statements are appropriate:
IR currents of the maximum possible value cause axial
forces on the (energizing) HV and TC windings of a typical power transformer that are approximately as high as
under rated SC conditions.
To avoid serious risk on damage of insulation within transformers who are subject to frequent unloaded energizations, countermeasures should be taken, e.g., controlled
switching strategies [5] might be applied in the future.
APPENDIX
A. Transformer Data
Three-phase three-legged step up transformer with

IV. DISCUSSION
It stands to reason that many energization operations with
even smaller IR currents which produce forces in the order of
SC forces might cause damage on the windings because the mechanical stress appears more frequently and with a much longer
duration (as stated in the Introduction). So windings and coils
may be damaged seriously, although the results of this work
show that forces in the coils of a typical power transformer are
about the same or a little smaller at the highest possible IR current compared to those at the rated SC current. It seems likely
possible that these damages are mostly in the form of reduction
of insulation capability (attrition of winding and conductor insulation material) and, therefore, may cause insulation failures
a certain time span after the occurrence of high IR currents.
This work also encourages for further research work on calculation of mechanical forces in transformers at excitation conditions caused by switching transients. Although results will be
similar for the comparison of the overall mechanical stress at IR
and SC conditions, state-of-the-art high sophisticated 3-D modeling and transient analysis should be employed to investigate
on the interaction between the phases and windings on different
legs.
Concerning the method of calculation of electrodynamic
forces in transformer windings under unbalanced excitation
conditions (such as IR), it seems advisable to pay extra caution
when using programs based on the theory of images. It
is likely possible that these programs lead to wrong results
because of certain simplifications and assumptions made there
for the field calculation which are not fulfilled at unbalanced
MMF conditions with high saturation of the iron.
V. CONCLUSION
The results of this work, based on magnetic field calculations
on a 2-D FEM model of a real 268 MVA three-legged step up
transformer show clear evidence that the axial electrodynamic
forces in the windings calculated with the maximum possible IR
current are in the same order of magnitude as with the rated short
circuit current. Magnetic field values in the coils region have
been double checked with results from a 3-D FEM model with

MVA
kV
kV
The HV and LV coils are shifted 0.5% of the coils height against
each other (to represent the worst case of mechanical tolerance).
Table I shows the partitioning of the magnetic excitations
at the coils for the filed calculations at IR and SC. With
A
as the nominal exciting MMF of the HV coil and
, the individual MMFs for SC and
the SC ratio
IR were calculated as follows:
(10)
(11)

B. Shape of the Yokes in 2-D Model


For mapping the actual 3-D geometry of the yokes to an
equivalent 2-D representation, several approaches are possible
and will be discussed now. As shown in Fig. 8(a), which is a
sketch of the lower part of core and yoke from Fig. 1(b), the
first approach could simply be a cap of the initial height of
the 3-D-yoke (A). This would obviously lead to a cross section
, increasing linear with the radius, which can not
be accepted. The next step would be to decrease the height with
, so the cross section stays constant (B). Because the cross
is much to high compared with the real cross
section at
) one would rather reduce
section of the 3-D yoke (
the initial height of the 2-D representation to
(12)
for constant cross section (C). But
and shape it again with
changes the geometry of the magnetic
as the reduced height
flux lines significant in the area close to the middle leg it seemed
at
and reduce it
more adequate to start with the height

160

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 17, NO. 1, JANUARY 2002

REFERENCES

TABLE I
PARTITIONING OF MMFS USED FOR FIELD CALCULATIONS

[1] W. Schmidt, Vergleich der groesstwerte des Kurzschluss-und Einschaltstromes von Einphasentransformatoren, ETZ-A, vol. 79, no. 21,
pp. 801806, 1958.
[2] A. I. Lure and A. B. Vasilev, Calculation of magnetic field and electrodynamic strength of transformers under magnetising current inrush,
Elec. Technol., pp. 2335, 1992.
[3] C. M. Arturi, Force calculation in transformer windings under unbalanced mmfs by a nonlinear finite-element code, IEEE Trans. Magn.,
vol. 28, pp. 13631366, Mar. 1992.
[4] J. Brunke and M. Steurer, Contribution to question 1.4 on report 13.110
cigre 1998, in Proc. CIGRE Conf., Paris, France, Sept. 1998.
[5] J. Brunke, Elimination of transient inrush currents when energizing unloaded power transformers, Ph.D. dissertation, ETH, Zurich, 1998.
[6] V. S. Chuprikov, V. A. Kuzmenko, A. I. Lure, and A. N. Panibrates,
Reducing the switching current of transformers, Russ. Elec. Eng., vol.
68, no. 2, pp. 3340, 1997.
[7] MAXWELL Program Documentation, 1997.
[8] R. Kuechler, Die Transformatoren. Berlin, Germany: Springer-Verlag,
1956.

Fig. 8. (a) Four approaches for the 2-D representation of the yokes; (b)
reduction of equivalent height by ratio of arc sections (top view on upper yoke
with middle leg underneath).

stronger with than in (B). The shape (D) finally chosen for
2-D representation follows:
(13)
is proportional to the ratio of the cirwhere the height
) over the corresponding arc of radius
cumference (
covered by the real yoke ( ) as shown in Fig. 8(b).
This approach seemed the most reasonable because the comparison of significant field values showed good concordance
between the 2-D and 3-D models. The outer leg is modeled as a
cylinder with a wall thickness of
(14)
derived from the demand to have the same cross section for the
magnetic flux as the middle leg.
ACKNOWLEDGMENT
The authors wish to express their sincere thanks to SIEMENS
AG/Nrnberg for providing the data of the transformer, Prof.
H. Brechna from ETH Zurich, and Dr. E. Schmidt from
TU-Vienna, for the numerous and stimulating discussions.

Michael Steurer (M02) received a Master of


electrical engineering from the Vienna University of
Technology, Vienna, Austria, and the Ph.D. degree
in technical science from the Swiss Federal Institute
of Technology, Zurich, Switerland, in 1995 and
2001, respectively.
He is currently with the Center of Advanced Power
Systems, Florida State University, Tallahassee, FL,
where he is working in the field of power systems for
all-electric ships.

Klaus Frhlich (SM88) was born in 1945 in


Salzburg, Austria. He received a Master of electrical
engineering and a Ph.D. in technical science from
the Vienna University of Technology, Austria.
After 11 years in Switchgear and High Voltage
Technology with BBC (later ABB) in Switzerland,
he became a Full Professor at the Vienna University
of Technology, in 1990. Since 1997, he has been
a Full Professor of high voltage technology at the
Swiss Federal Institute of Technology, Zurich,
Switzerland.
Dr. Frhlich is a member of CIGRE Study
Committee 13 and the convenor of CIGRE Working Group 13.07 (Controlled
Switching).

You might also like