See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/3113114

Determination of Complex Permeability of Silicon Steel for Use in High-

Frequency Modeling of Power Transformers

Article  in  IEEE Transactions on Magnetics · May 2008

DOI: 10.1109/TMAG.2007.914857 · Source: IEEE Xplore

CITATIONS READS
48 7,500

4 authors, including:

Tadeusz Daszczyński Yuriy Serdyuk

Warsaw University of Technology Chalmers University of Technology
15 PUBLICATIONS   63 CITATIONS    108 PUBLICATIONS   2,187 CITATIONS   

SEE PROFILE SEE PROFILE

Stanislaw M. Gubanski
Chalmers University of Technology
333 PUBLICATIONS   8,063 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

SFRA measurments on physical model of power transformer View project

WATERTREE – Predicting failure of dynamic subsea cables by insulation breakdown View project

All content following this page was uploaded by Tadeusz Daszczyński on 26 January 2016.

The user has requested enhancement of the downloaded file.

438 IEEE TRANSACTIONS ON MAGNETICS, VOL. 44, NO. 4, APRIL 2008

Determination of Complex Permeability of Silicon Steel for Use

in High-Frequency Modeling of Power Transformers
K. G. Nilanga B. Abeywickrama1 , Tadeusz Daszczyński2 , Yuriy V. Serdyuk1 , and
Stanislaw M. Gubanski1 , Fellow, IEEE
Division of High Voltage Engineering, Chalmers University of Technology, SE 412 96 Göteborg, Sweden
Warsaw University of Technology, 00-611 Warsaw, Poland

Information about frequency dependence of complex permeability of silicon steel is a vital input parameter in calculations of trans-
former winding inductance used for modeling high-frequency behavior (100 Hz–1 MHz). We present two ways of determining small
signal complex permeability spectra in frequency domain and compare and discuss the results. The first method is based on an optimiza-
tion procedure, in which inductance of a winding is measured and calculated by analytical formulas and finite-element modeling. The
second method makes use of a single sheet tester. We show that the magnitude of effective permeability of the silicon steel laminations
remains significant up to about 100 kHz. We also report on the effect of magnetic viscosity on complex permeability.
Index Terms—Complex permeability, magnetic viscosity, power transformers, silicon steel, single sheet tester.

I. INTRODUCTION shown that the influence of the core remains significant even
at frequencies above 100 kHz [4]. The purpose of the work

S ILICON steel, otherwise called electrical steel, is the

most popular soft magnetic material in the electric
power industry as the core material of electrical machines.
described in this paper has therefore been twofold, i.e., 1) to
estimate the small signal local permeability of silicon steel
laminations indirectly by using an optimization procedure
Grain-oriented silicon steel is mainly used in manufacturing and thus obtaining the complex permeability spectra and 2) to
of transformer cores, which provides the required magnetic experimentally determine the complex permeability spectra and
anisotropy and lowest losses when magnetized in the rolling the effect of magnetic viscosity on these spectra. In the study,
direction [1]. Eddy-current losses are further reduced at power a distribution type transformer (T1: 50 KVA, 11/0.4 kV, Yd11)
frequency (50/60 Hz) by using laminations of silicon steel. was used for both the modeling and the measurements. A single
The laminations are normally insulated with a thin organic or sheet tester [5] based on impedance measurements (performed
inorganic surface coating for nonoriented silicon steel and by by means of an impedance analyzer HP4292A) was built for
phosphate coating for grain-oriented silicon steel that prevents measuring the complex permeability of steel laminations taken
flow of inter-laminar currents. The laminations are typically out from the transformer. Samples of two other silicon steel
produced in strips 0.35–0.8 mm thick for nonoriented grades grades originating from different manufacturers were also
and 0.23–0.35 mm thick for oriented grades [2]. investigated.
Despite the fact that transformers work at power frequency,
there are situations where the windings are exposed to high-fre-
quency excitations, e.g., transient over voltages containing II. EFFECTIVE COMPLEX PERMEABILITY OF SILICON
high-frequency components and during frequency response STEEL LAMINATIONS
measurements. To analyze these phenomena in power trans-
formers, high-frequency transformer models have been devel-
oped and lumped circuit modeling is one of the most frequently A. Complex Permeability
used approaches. In this approach, windings are split into Magnetic induction and related phenomena inside a piece
finite sections that can be represented by lumped capacitances, of steel lamination subjected to an external magnetic field can
conductances, inductances, and resistances. Permeability of the be derived from Maxwell’s equations [6]. It is assumed in this
laminated core and its losses are the required input parameters derivation that displacement current density is negligible com-
for calculating inductance and effective resistance (including pared to the conduction current in the studied frequency range
core losses and winding resistance) of the windings. However, as well as the effect of electric field inside the insulation layer
at higher frequencies, the effective complex permeability of on the lamination. Fig. 1 shows a lamination sample of size
the core becomes frequency dependent due to repellence of and thickness . is the length of the sample in
magnetic flux out of the laminated core by the counteracting the perpendicular direction to -plane. If eddy-current loops
magnetic flux produced by increasing eddy currents. are assumed to be large enough along the -direction, the field
The influence of the core has usually been neglected above problem becomes one-dimensional and can be reduced to a
10 kHz [3] when modeling transformer windings for use in single equation for -component of the magnetic field
frequency response analysis (FRA). However, the authors have that depends only on :

Digital Object Identifier 10.1109/TMAG.2007.914857 (1)

0018-9464/$25.00 © 2008 IEEE

ABEYWICKRAMA et al.: DETERMINATION OF COMPLEX PERMEABILITY OF SILICON STEEL 439

Fig. 1. Rectangular lamination sample placed on xy -plane.

Fig. 2. Relative complex permeability of single lamination in z and y directions
^ =
( 0
 j and  ^ = 0
 j ) according to (4) and (5).
For time-harmonic case ( -angular frequency) with boundary
condition , solution of (1) can be repre-
sented as magnetic material. The magnetomotive force (MMF) that drives
the amount of flux along the length is . Division
(2) of MMF by the corresponding flux, as defined in (3), yields the
reluctance of the lamination sample in -direction as
Here, is the propagation constant defined as ,
which is directly related to the skin depth (6)
by . is the local relative permeability of the
lamination sample in the -direction and denotes the conduc-
This reluctance is a complex number that depends on frequency.
tivity of silicon steel. Averaged magnetic flux density in -direc-
If one can determine the frequency-dependent reluctance (6) of
tion can be evaluated in terms of total magnetic flux
the lamination sample experimentally, the complex relative per-
through the cross section as
meability can be calculated from

(7)
(3)

Thus, effective relative complex permeability in -direction of III. ESTIMATION OF SMALL SIGNAL LOCAL
a single lamination can be expressed as PERMEABILITY
For calculating the effective permeabilities of both the bulk of
(4) the core [see (4)] and the mitered joint regions of a transformer
core [4], the local permeability and conductivity of sil-
icon steel are mandatory input parameters. These data are often
Similarly, the effective relative complex permeability in -di- unavailable, unless the manufacturers provide the information.
rection can be derived as It is practically not possible to remove a lamination sample
from a stacked core of a transformer for performing measure-
(5) ments of . Instead, one can estimate the small signal local
permeability indirectly by using an optimization algorithm, in
Here, in calculating , should be used instead of . If we which inductance of a winding at a low frequency ( 500 Hz)
assume the following typical parameter values for 3% silicon is numerically calculated and compared with a measured
steel: S/m, , , the value. The reason for using this lower frequency inductance
resulting effective permeabilities as a functions of frequency are is to avoid the effect of winding capacitance, which together
shown in Fig. 2. with the core inductance produces the first resonance of the
measured impedance [7]. A flow diagram of the developed
B. Reluctance of a Fragment of Steel Lamination optimization algorithm is illustrated in Fig. 3. First, it makes
Reluctance is a measure of the opposition to the magnetic use of realistically assumed initial values of the required pa-
flux driven by a magnetomotive force and it is analogous to the rameters for calculating the winding inductance (calculation of
concept of resistance in an electric circuit. Reluctance is depen- permeability of the bulk lamination and the joints that is then
dent on geometrical parameters as well as on permeability of the used in a 3-D model of the transformer core to calculate the
440 IEEE TRANSACTIONS ON MAGNETICS, VOL. 44, NO. 4, APRIL 2008

Fig. 4. Single sheet tester.

IV. COMPLEX PERMEABILITY MEASUREMENTS

If a lamination sample is available, there is always a possi-
bility to measure complex relative permeability. Not only at
a single frequency, but also measurements of the frequency re-
sponse of the complex permeability in a desired direction with
respect to the rolling direction are possible. Properties of silicon
steel are usually determined by means of Epstein frame method
or single sheet tester as recommended by IEC-Standards (IEC
404-2 [11] and IEC 404-3 [12]). These types of measurements
focus on specific power loss (W/kg) and permeability at power
frequency (50 or 60 Hz). For the purpose of this paper Epstein
frame method is not suitable, as it determines properties of a
stack of lamination with joints. Although the single sheet tester
is ideal for measuring characteristics of a single sheet of silicon
steel at power frequency, the double yoke system made up of
grain-oriented silicon steel or nickel iron alloy, as recommended
by IEC 404-3 to provide low reluctance path, makes it unsuit-
able for determining the high-frequency characteristics of the
test specimen. Therefore, a high-permeability material with suf-
Fig. 3. Flowchart showing the basic steps of the algorithm used for estimating ficiently flat high-frequency characteristics (permeability and
the small signal local permeability  of electrical steel. losses) should be used instead for the yokes.

A. Single Sheet Tester

winding inductance) and then compares it with the measured Measurements of complex permeability were carried out
one. Depending on the sign of the error in this comparison, an using the single sheet tester (see Fig. 4) proposed by Roger
increment of is chosen and the calculation continues until the et al. [5]. The tested lamination strip is placed in between two
difference reaches a predefined error level. The iteration ends U-type high permeability ferrite cores (MAGNETICS Ferrites,
at this point and it yields the values of calculated inductance 2500, 50% roll-off frequency 1.2 MHz [13]) as shown in
and . Fig. 4. This arrangement provides very low reluctance for the
Estimation of the complex permeabilities of bulk of the core rest of the magnetic path around the test specimen. Size of the
and core joints , and calculation of inductance of lamination strip (30 30 cm ) was selected so that the large
a winding (see Fig. 3) were performed by means of 2-D and 3-D enhancement of the magnetic field at the edges of the strip does
finite-element models developed by the authors [4], [8], [9]. The not affect the tested area between yokes. The tested area (shaded
whole algorithm was implemented as a script in a finite element region in Fig. 4, 2.5 5 cm ) should also be sufficiently large in
software COMSOL Multiphysics [10]. order to avoid the possible influence of material heterogeneity
The optimization procedure was applied to transformer T1. [14]. A clamping structure was made up of plexiglass in order
The measured inductance value of T1 was 25 mH at to hold two yokes tightly against the test sample, leaving no air
500 Hz. The algorithm started with initial conditions gaps between the sample surface and the yoke face. This is a
, and error tolerance , very important issue when calculating the reluctance of the test
and after 10 iterations it estimated that the local permeability specimen due to the fact that even a small air gap could result
in the -direction is 448, which appeared a bit higher than the in a significant error in the measurements [15]. Two measuring
measured value (438). coils (20 turns each connected in series) were wound on each
ABEYWICKRAMA et al.: DETERMINATION OF COMPLEX PERMEABILITY OF SILICON STEEL 441

Fig. 5. Equivalent magnetic circuit of single sheet tester in Fig. 4.

on the U-core (see Fig. 4) and their impedance at different

frequencies (100 Hz–1 MHz) was recorded by the impedance
analyzer.

B. Interpretation of Measurements
The equivalent magnetic circuit of the experimental setup Fig. 6. Measured relative complex permeability of lamination sample 30H102
in the rolling direction and transverse direction.
(Fig. 4) is depicted in Fig. 5. Two coils connected in series pro-
duce the same amount of flux. In Fig. 5, is the reluctance
of the ferrite yoke, represents the reluctance of the gap be-
tween the yokes and the lamination strip, which appears due to • Impedance measurement without lamination strip and two
the insulation layer on the lamination strip. stands for the coils connected in series with opposite directions com-
leakage reluctance accounting for leakage flux outside lamina- pared with previous measurement force the flux into the
tion strip and is the reluctance of the tested area of the lami- ferrite yokes. Hence, the permeance of yokes can be
nation strip. Flux density in the yokes during the measurements calculated from , H.
was kept very small so that remained constant. The other • Permeance of small insulation gap is approximately
two reluctance values ( and ) were also not dependent on calculated using geometrical parameters H.
the flux density [5]. When , , , and are known, one can calculate the
Equivalent permeance of the whole magnetic circuit (Fig. 5) complex reluctance of the lamination sample . Then, the rel-
can be simplified by placing before and expressed as ative complex permeability can be calculated by means of (7).
Here, the length of the magnetic path in the test sample is
(8) defined as the inner distance between the yoke faces as shown
in Fig. 4 as recommended in [12]. One should notice here that
thus, the reluctance of the lamination sample becomes the dc and ac resistances of the wire are also included in all the
impedance measurements. These are calculated analytically and
(9) deducted from the measured impedance (10).
Relative complex permeabilities of lamination strips taken
Knowing the permeance of a magnetic circuit, number of turns out from T1 and of two other silicon steel samples (30H102 and
of the coil and neglecting the copper wire resistance, the 30R122) were determined by means of the aforementioned pro-
inductance of the measuring coil can be expressed as [16]. cedure. Figs. 6–8 illustrate the obtained results. Two samples of
Hence, the impedance of the two series-connected coils is given 30H102 and 30R122 steel exhibit different local permeability
by values in the rolling direction; 504 for 30H102 and 472
for 30R122. Consequently, the loss peak appears at lower fre-
(10) quency for 30H102 (5.46 kHz) compared to 30R122 (6.9 kHz)
(the higher the local permeability the lower the frequency of the
Therefore, corresponding permeance values can be determined loss peak). The complex permeability in the transverse direc-
by the following measurements. tion showed anisotropy factor , which could be
• Impedance measurement with the lamination strip explained by crystallographic anisotropy of silicon steel [17],
provides the equivalent impedance of the test [18]. The lamination sample taken out from T1, which is a very
setup. Then, it is possible to calculate from old transformer, exhibited lower local permeability (438) com-
, which is frequency dependent. pared to new samples.
• Impedance measurement without lamination strip provides It can clearly be noticed from Figs. 6–8 that at 10 kHz, above
the leakage impedance of the test setup, as the equal which the influence of the magnetic core is often neglected in
and oppositely directed magnetic flux in the yokes cancel modeling transformers for FRA purposes, the permeability falls
each other and only the leakage flux remains. Thus, can down to 45% of its low-frequency value. Therefore, the neg-
be calculated from , H. ligence of the core influence above 10 kHz is not justified at
442 IEEE TRANSACTIONS ON MAGNETICS, VOL. 44, NO. 4, APRIL 2008

Fig. 7. Measured relative complex permeability of lamination sample 30R122 Fig. 9. Change in real part of the complex permeability with time after sudden
in the rolling direction and transverse direction. removal of dc magnetization from a steel sample of T1.

measured real part also becomes lower compared with the ana-
lytical prediction at higher frequencies. This could result from
the fact that the analytical expression used is derived by consid-
ering only the classical eddy-current approach.

V. EFFECT OF MAGNETIC VISCOSITY ON COMPLEX

PERMEABILITY SPECTRA
It was found in authors’ previous study [19] that the low-fre-
quency response of winding impedance is affected by magnetic
viscosity, yielding relaxation in permeability over a long pe-
riod of time. To investigate this phenomenon, another lamina-
tion sample was taken out from the core of the transformer T1
and tested as described in the previous section. An additional
coil was wound on the steel sample and supplied with a dc
current so that the amount of MMF produced by the coil was
enough to drive the lamination sample into saturation. Then,
Fig. 8. Measured and analytically calculated relative complex permeability of the dc current was suddenly removed and the measurements
lamination sample of T1 in the rolling direction. were conducted over a one-day period. Figs. 9 and 10 depict the
time varying real and imaginary parts of the relative complex
permeability in the rolling direction. The real part jumped to a
all, as the silicon steel further supports magnetic flux well up to higher value and started relaxing towards a steady state. Fig. 11
100 kHz . shows the impedance spectrum of the low voltage winding on an
The measurements of complex permeability provide a possi- outer limb of transformer T1, which was measured after sudden
bility to evaluate actual conductivity of steel by applying opti- removal of the dc magnetization from the core [19]. One can
mization algorithm that minimizes the error between the mea- readily see the analogy between the variations of the real part of
sured and calculated [using (5)] permeability spectra, where the permeability (Fig. 9) and changes in measured impedance.
is treated as unknown parameter [5]. Such an algorithm was The imaginary part of the permeability, which is directly related
implemented in Matlab and is based on least-square nonlinear to losses in the steel laminations, is also decreasing with time
curve fitting technique with 200 frequency points ranging from and its peak is shifting towards higher frequencies (due to de-
100 Hz to 1 MHz. The procedure was applied to data obtained creasing [4]). If one takes a closer look at the variation of the
for a steel sample taken from transformer T1 and its conduc- first peak of impedance, the increase in magnitude of the peak
tivity was found to be 4 10 Sm . This value was further used (less damping) with time (see Fig. 11) is a direct consequence
for calculating permeability spectra and the results are shown in of the decreasing imaginary part of the permeability (Fig. 10).
Fig. 8. The calculated complex permeability spectra were ob- It is known that the characteristics of viscosity can be de-
tained by means of (5) with parameters (estimated in scribed using a logarithmic function [20], [21]. Fig. 12 shows
Section III). One can observe that, at low frequencies, the mea- the decay of permeability at 800 Hz fitted with a function in
sured real part of the permeability appears lower because the the form of . It is seen that the
estimated (448) is higher than the measured one (438). The logarithmic function can well approximate only the initial part
ABEYWICKRAMA et al.: DETERMINATION OF COMPLEX PERMEABILITY OF SILICON STEEL 443

Fig. 10. Change in imaginary part of the complex permeability with time after Fig. 12. Change in imaginary part of the complex permeability at 800 Hz with
sudden removal of dc magnetization from a steel sample of T1. time after sudden removal of dc magnetization. Solid lines represent fitted log-
arithmic approximation.

high-frequency modeling of power transformers. Magnetic

viscosity measurements confirm that the frequency response
measurements ( 10 kHz) on transformers can be affected
by viscous nature of complex permeability. This effect may
extend in time for about one day (see Fig. 12) after a change in
magnetization.

ACKNOWLEDGMENT
This work was supported by SIDA-SAREC of Sweden and
ABB/SECRC.

REFERENCES
[1] S. Tumanski, “Investigation of the anisotropic behaviour of SiFe steel,”
J. Magn. Magn. Mater., vol. 254, pp. 50–53, 2003.
[2] A. J. Moses, “Electrical steels: Past, present and future developments,”
Inst. Elect. Eng. Proc., vol. 37, no. 5, pp. 223–245, Sep. 1990.
[3] E. Rahimpour, J. Christian, K. Feser, and H. Mohseni, “Transfer func-
Fig. 11. Variation of impedance (LV winding on an outer limb) with time after
tion method to diagnose axial displacement and radial deformation of
sudden removal of dc magnetization to the core of T1.
transformer windings,” IEEE Trans. Power Del., vol. 18, no. 2, pp.
493–505, Apr. 2003.
[4] K. G. N. B. Abeywickrama, A. D. Podoltsev, Y. V. Serdyuk, and S. M.
of the viscosity effect. However, it cannot be used to describe Gubanski, “Computation of parameters of power transformer windings
permeability change when approaching the steady-state value at for use in frequency response analysis,” IEEE Trans. Magn., vol. 43, no.
5, pp. 1983–1990, May 2007.
. Moreover, it is not possible to evaluate single effec- [5] D. Roger, E. Napieralska-Juszczak, and A. Henneton, “High frequency
tive relaxation time from Fig. 12 due to complex and distributed extension of non-linear models of laminated cores,” Int. J. Comput.
nature of relaxation processes [21], [22]. Math. Elect. Electron. Eng., vol. 25, no. 1, pp. 140–156, 2006.
[6] J. Lammeraner and M. Stafl, Eddy Current. London, U.K.: ILIFFE
Books, 1966.
VI. CONCLUSION [7] K. G. N. B. Abeywickrama, Y. V. Serdyuk, and S. M. Gubanski, “Ex-
Anisotropic complex permeability spectra (in the frequency ploring possibilities for characterization of power transformer insula-
tion by frequency response analysis (FRA),” IEEE Trans. Power Del.,
range 100 Hz–1 MHz) of silicon steel laminations, used for vol. 21, no. 3, pp. 1375–1382, Jul. 2006.
power transformers, have been measured by means of a single [8] A. D. Podoltsev, K. G. N. B. Abeywickrama, Y. V. Serdyuk, and S.
sheet tester. Furthermore, it has been used to investigate the vis- M. Gubanski, “Multi-scale computations of parameters of power trans-
former windings at high frequencies. Part I: Small-scale level,” IEEE
cous nature of the permeability after sudden removal of mag- Trans. Magn., vol. 43, no. 11, pp. 3991–3998, Nov. 2007.
netization. An optimization algorithm has also been developed [9] A. D. Podoltsev, K. G. N. B. Abeywickrama, Y. V. Serdyuk, and S.
to determine the complex permeability when no steel sample is M. Gubanski, “Multi-scale computations of parameters of power trans-
former windings at high frequencies. Part II: Large-scale level,” IEEE
available for measurements. Trans. Magn., vol. 43, no. 12, pp. 4076–4082, Dec. 2007.
It was found from the measurements performed by the [10] COMSOL Multiphysics 3.2, Comsol AB. Stockholm, Sweden [On-
single sheet tester that silicon steel laminations can support line]. Available: www.comsol.com
[11] Magnetic Materials Part-2: Method of Measurements of the Magnetic
magnetic flux well up to 100 kHz. Therefore, the influence Properties of Electrical Steel Sheet and Strip by Means of An Enstein
of ferromagnetic core cannot be neglected above 10 kHz in Frame, IEC Standard 60404-2, 1996.
444 IEEE TRANSACTIONS ON MAGNETICS, VOL. 44, NO. 4, APRIL 2008

[12] Magnetic Materials Part-3: Method of Measurements of the Magnetic Tadeusz Daszczyński received the M.Sc. degree in electrical engineering from
Properties of Electrical Steel Sheet and Strip by Means of a Single the Warsaw University of Technology, Warsaw, Poland, in 2007. He was a vis-
Sheet Tester, IEC Standard 60404-3, 1992. iting student at Chalmers University of Technology (CTH), Göteborg, Sweden,
[13] MAGNETICS Ferrite Core Material Catalog [Online]. Available: during academic year 2006–07, where he wrote his master thesis on high-fre-
http://www.mag-inc.com/pdf/2006_Ferrite_Catalog/2006_Materials. quency properties of silicon steel laminations.
pdf
[14] S. Tumanski and T. Bakon, “Measuring system for two-dimensional
testing of electrical steel,” J. Magn. Magn. Mater., vol. 25, no. 1, pp.
315–352, 2001. Yuriy V. Serdyuk received the M.Sc. and Ph.D. degrees in high-voltage engi-
[15] O. Stupakov, I. Tomàš, and J. Kadlecovà, “Optimization of single-yoke neering from the National Technical University of Ukraine “Kiev Polytechnic
magnetic testing by surface fields measurement,” J. Phys. D: Appl. Institute” Kiev, Ukraine, in 1986 and 1995, respectively.
Phys., vol. 39, pp. 248–254, Jan. 2006. From 1986, he worked as an engineer, and since 1991, as a researcher at the in-
[16] L. Dixon, “Deriving the equivalent electrical circuit from the magnetic stitute of Electrodynamics of the National Academy of Science of Ukraine Kiev.
device physical properties,” Texas Instruments, 2003 [Online]. Avail- In 1996–1997, he was a Research Engineer at the R & D HV Laboratory, ABB
able: http://focus.ti.com/lit/ml/slup198/slup198.pdf High Voltage Technologies Ltd, Zurich, Switzerland. Currently, he is an Asso-
[17] H. J. Williams, “Magnetic properties of single crystals of silicon iron,” ciate Professor in High Voltage Engineering, Department of Manufacturing and
Phys. Rev., vol. 52, pp. 747–751, Oct. 1937. Materials Technology, Chalmers University of Technology, Göteborg, Sweden.
[18] V. D. Solovei and Y. N. Dragoshanskii, “Magnetic anisotropy con-
stants of a Fe-3.8% Si single crystal,” Phys. Met. Metallogr., vol. 103,
no. 1, pp. 33–38, 2007.
[19] K. G. N. B. Abeywickrama, Y. V. Serdyuk, and S. M. Gubanski, “Ef-
fect of core magnetization on frequency response analysis of power Stanislaw M. Gubanski (M’89–SM’90–F’01) received the M.Sc. degree in
transformers (FRA),” IEEE Trans. Power Del., accepted for publica- high-voltage engineering and the Ph.D. degree in material science from the
tion. Technical University of Wroclaw, Poland, in 1973 and 1976, respectively.
[20] G. Bertotti, Hysteresis in Magnetism. San Diego, CA: Academic, He was a Research Fellow at the University College of North Wales, Bangor,
1998. U.K., from 1976 to 1977, and a Senior Lecturer at the Technical University of
[21] D. C. Crew, P. G. McCormik, and R. Street, “The interpretation of mag- Wroclaw, Wroclaw, Poland, from 1977 to 1988. Afterwards, he was Associate
netic viscosity,” J. Phys. D: Appl. Phys., vol. 29, no. 9, pp. 2313–2319, Professor at the Royal Institute of Technology, Stockholm, Sweden. Currently,
1996. he is Professor in High Voltage Engineering at the Department of Materials
[22] A. Lyberatos and R. W. Chantrell, “Fluctuation field of ferromagnetic and Manufacturing Technology, Chalmers University of Technology, Göteborg,
materials,” J. Phys.: Condens. Matter, vol. 9, no. 12, pp. 2623–2643, Sweden, and research leader of High Voltage Valley in Ludvika, Sweden. He
Apr. 1997. is Chair of the IEEE-DEIS Nominations Committee. He is also convener of the
CIGRE Task Force D1.01.14 “Dielectric Response Methods for Diagnostics of
Power Transformers.”

Manuscript received August 17, 2007; revised December 14, 2007.

Corresponding author: K. G. N. B. Abeywickrama (e-mail: nilanga.abeywick-
[email protected]).

K. G. Nilanga B. Abeywickrama (S’04) received the B.Sc. degree in electrical

and electronic engineering in 2001 and the M.Sc. degree in electrical power
and energy in 2003 from the University of Peradeniya, Sri Lanka, and the Lic.
Eng. degree in 2005 from Chalmers University of Technology (CTH), Göteborg,
Sweden, where he is currently pursuing the Ph.D. degree. His research interests
include high-frequency modeling, high-frequency measurements, and diagnosis
of power transformers.

View publication stats