Вы находитесь на странице: 1из 8


1, JANUARY 2010 193

Harmonic Impact on Distribution Transformer

No-Load Loss
Themistoklis D. Kefalas, Member, IEEE, and Antonios G. Kladas, Member, IEEE

AbstractThe losses in European Union distribution trans- V  Average rectified voltage.

formers are estimated at about 33 TW h/year, whereas reactive H Parameter of the hysteresis model.
power and harmonic losses add a further 5 TW h/year. The re-
duction of distribution transformer no-load loss is particularly im-
portant as the ratio of no-load to load losses is nearly three. In this I. I NTRODUCTION
paper, the no-load operation of wound-core transformers under
sinusoidal and distorted supply-voltage conditions is investigated.
For that purpose, a 2-D nonlinear transient finite-element analysis
taking into account hysteresis has been developed. The hysteresis
A CCORDING to Strategies for development and diffusion
of Energy Efficient Distribution Transformers (SEEDT),
the losses in European Union (EU-27) distribution transformers
model is based on a modified JilesAtherton representation, and are estimated at about 33 TW h/year. Furthermore, reactive
the proposed analysis is compared to experimental data. power and harmonic losses add a further 5 TW h/year. Thus,
Index TermsFinite element methods, industrial power system the total losses of distribution transformers in EU-27 might total
harmonics, magnetic anisotropy, magnetic cores, magnetic hys- to about 38 TW h/year. In comparison, the energy consump-
teresis, magnetic losses, nonlinear magnetics, power transformers, tion of Athens subway (ATTIKO METRO S.A.) in 2006 was
soft magnetic materials, transient analysis.
more than 400 times less, i.e., 0.09 TW h/year.
Transformer losses can be resolved into two components,
N OMENCLATURE namely, no-load and load losses. No-load loss occurs from the
, a Parameters of the Langevin function. energy required to retain the continuously varying magnetic
A Time derivative of the magnetic vector potential flux in the core, and it is invariant with load on the transformer.
vector. Load loss arises mainly from resistance losses in the conducting
H Parameter of the hysteresis model. material of the windings, and it is varying with the second
Az zth component of the magnetic vector potential. power of the load. One might argue that the load loss of a
Bp Peak flux density. typical distribution transformer is more important than the no-
D Potential-current coupling stiffness matrix. load loss. This stems from the fact that the load loss at rated
DF Deterioration factor. load is six times more than the no-load loss in the case of
f Frequency. transformers constructed of grain-oriented steel and 13 times
G Inductive damping matrix. more in the case of transformers constructed of amorphous
HAN Anhysteretic component of the magnetic field. steel [1]. Nevertheless, a distribution transformer is in circuit
HH Hysteretic component of the magnetic field. every day for 24 h and has a no-load loss load factor of 100%,
HHS Parameter of the hysteresis model. irrespective of the load delivered. The load loss, on the other
HHW Irreversible part of the hysteretic component. hand, will vary over a wide range, even for the same average
HRET Reversible part of the hysteretic component. load, since load losses are proportional to the square of load.
I Time derivative of the electric current vector. As a result, the continuous loss of energy to the magnetic core
Jz zth component of the current density. exceeds the loss of energy to the conductive material of the
L(AN ) Langevin function. transformer, even in the cases where the core is constructed of
MS Saturation magnetization. amorphous steel.
Mx , My Components of magnetization. Due to the loading characteristics of EU-27 distribution
N Number of turns of the excitation coil. transformers, the ratio of no-load to load losses is close to
S Lamination cross-sectional area. three. Consequently, the reduction of distribution transformer
no-load losses has recently become the top priority of many
countries [1].
Manuscript received February 28, 2009; revised August 5, 2009. First pub-
lished September 1, 2009; current version published December 11, 2009. This No-load loss is sensitive to the distortion of the supply-
work is part of the 03ED45 research project, implemented within the framework voltage waveform [2]. The deterioration of the supply-voltage
of the Reinforcement Program of Human Research Manpower (PENED) and waveform is caused by industrial and domestic loads, e.g.,
cofinanced by National and Community Funds (25% from the Greek Ministry
for DevelopmentGeneral Secretariat of Research and Technology and 75% switch-mode power supplies and fluorescent lamps that inject
from EUEuropean Social Fund). current harmonics in the power grid. Researchers developed
The authors are with the School of Electrical and Computer Engineer- passive networks for the reduction of the ac input-current
ing, National Technical University of Athens, 15780 Athens, Greece (e-mail:
thkefala@central.ntua.gr; kladasel@central.ntua.gr). total harmonic distortion (THD) [3], [4], techniques for the
Digital Object Identifier 10.1109/TIE.2009.2030207 minimization of undesired low-order harmonics at the input
0278-0046/$26.00 2010 IEEE

Authorized licensed use limited to: Themistoklis Kefalas. Downloaded on December 19, 2009 at 13:36 from IEEE Xplore. Restrictions apply.

current of asymmetrically loaded six-pulse rectifiers [5], and

acdc converters with low-line-current distortion [6]. More re-
cently, cascaded multilevel inverters were developed for effec-
tive control of the input-current and output-voltage waveform,
diminishing in this manner harmful harmonic components [7],
[8], and active power filters and static var compensators using
multilevel inverters for the compensation of load currents with
a high-harmonic impact [9].
This paper deals with the distorted supply-voltage wave-
forms of the power grid that arise from the aforesaid situations,
i.e., without multiple zero crossings (no minor loops) and only
odd higher harmonics [10], in order to evaluate in particular
the harmonic impact on wound-core distribution transformer
no-load operation. Transformers are classified, depending on
Fig. 1. Typical wound core.
their magnetic core type, stack core, or wound core. Many
researchers have published their results from investigations of
stacked-type transformer cores [11], [12]. As a counterpart,
there are a few published works in the case of wound-type trans-
former cores, and many issues still remain unclear since the
stack- and wound-core topologies are completely different. The
authors have developed recently numerical techniques based
on the finite-element (FE) method, specifically formulated for
the no-load loss and flux-distribution evaluation of wound-
core distribution transformers under sinusoidal supply-voltage
conditions [13], [14]. This paper extends the aforementioned
numerical technique in order to include the case of distorted
supply-voltage conditions commonly encountered in the power
grid. A transient analysis is necessary for determining the
harmonic impact on wound-core distribution transformer no-
load operation. The analysis developed herein integrates the
hysteresis phenomena in the FE method. The hysteresis model
that the authors used is based on previous formulations given
in [11] and [15][17]. Nevertheless, the hysteresis model and
the respective FE formulation have been modified, as previous
attempts were constrained to the modeling of stacked magnetic
cores [11], ferrite pot cores [16], and rectangular Epstein frames
[17]. The simulated results are compared to the experimental
results obtained by a no-load loss and local-flux-density mea- Fig. 2. Wound-core transformer. (a) One-phase core type. (b) One-phase shell
surement apparatus [18]. type. (c) Three-phase shell type.
The purpose of this paper is to point out to the reader
the importance of distribution transformer no-load loss, as the stack core and the wound core. A typical wound core, shown
well as the harmonic impact on transformer no-load opera- in Fig. 1, is composed of hundreds of long continuous strips
tion, and to provide transformer manufacturers a no-load loss of sheet steel. The main advantages include reduction of joints
analysis numerical tool with which they can evaluate the tech- and the use of the grain direction of the steel for the flux path.
nically and economically optimum distribution transformer, Wound cores are commonly used for distribution transformers
working under either sinusoidal or distorted supply-voltage with ratings that range from a few kVA to a few MVA. One-
conditions. and three-phase wound-core distribution transformers are con-
structed by using one or more wound cores assembled about a
preformed electrical winding coil, as shown in Fig 2.
In order to transmit and distribute electrical energy over large
distances economically, it is necessary to minimize Joule losses The FE method has been used extensively by many re-
in the transmission lines by using a high voltage. The required searchers for the accurate field analysis of electromagnetic de-
increase and decrease of the voltage are performed by the vices. A 2-D FE analysis approach used for the analysis of 3-D
electrical transformer that, in its simplest form, consists of two magnetic components was developed in [19], an FE analysis to
coils of conductive wire wound around a magnetic core of soft validate a new winding design for planar integrated magnetics
iron. In practice, two types of magnetic core are used, namely, has been carried out in [20], and a coupled electrothermal

Authorized licensed use limited to: Themistoklis Kefalas. Downloaded on December 19, 2009 at 13:36 from IEEE Xplore. Restrictions apply.

numerical analysis of planar and medium-frequency power motion HRET . Thus, it follows that the magnetic energy of the
transformers was presented in [21] and [22], respectively. Also, hysteretic component HH is given by
the FE method was used for determining the transient behavior   
and short-circuit forces of stack-core power transformers in HH dB = HHW dB HRET dB. (2)
[12] and [23], respectively. The main advantage of the FE B B B
method over analytical methodologies like the one developed
in [24] for the behavior modeling of magnetizing currents in a The anhysteretic component HAN is obtained by solving
dcdc converter is the accurate computation of the local distri- (3), where MS is the saturation magnetization, L(AN ) is the
bution of field quantities such as flux density. This is necessary Langevin function of AN given by (4) and (5), and and a are
in the case of wound cores where it is experimentally verified the parameters of the Langevin function
that the flux-density distribution is nonuniform [13], [14].
The authors developed an FE methodology in [14] that is HAN = B/0 MS L(AN ) (3)
particularly formulated for wound cores in order to evaluate
accurately the no-load losses under sinusoidal excitation. The L(AN ) = coth(AN ) 1/AN (4)
method is based on the combination of the experimentally
evaluated local specific core losses with the FE-computed peak- AN = [HAN (1 ) + B(/0 )] /a. (5)
flux-density distribution of the wound core. The local specific The irreversible part HHW and the reversible part HRET
core losses are expressed as a function of peak flux density, of the hysteretic component HH are given by (6) and (7),
and the computation of the peak-flux-density distribution of the respectively, where ID is a variable that takes the +1 and 1
wound core is performed using the FE method. The accurate values when B > 0 and B < 0, respectively, and HHS , aH ,
representation of the wound core with low computational cost and H are the parameters of the hysteresis model
was achieved by considering the iron-laminated material as a
homogeneous medium and by developing an elliptic anisotropy HHW = HHS L(H ), H = (HH + ID HHS )/aH (6)
model that is specifically formulated for wound cores [14]. The
elliptic anisotropy model takes into account the directional de- HRET = ID H dHH /dB. (7)
pendence of the BH characteristic due to the iron laminations
The differential equation of hysteresis is obtained by substi-
and the grain orientation of the magnetic steel. The material
tuting (6) and (7) into (2) and isolating dHH /dB
modeling of the wound core with the proposed technique
results in significant computational effort reduction, not only dHH HHS L(H ) HH
= . (8)
in the case of 2-D FE analysis but also in the case of 3-D FE dB ID H
analysis [14].
In this paper, the numerical methodology developed in [14] The application of the Euler scheme to (8) yields the fol-
is extended by considering the time evolution of flux density. lowing nonlinear equation, from the solution of which the
For this to be achieved, a nonlinear transient FE analysis is hysteretic component HH is obtained:
developed, which integrates the hysteresis phenomena. In this
ID H HH = (HHS L(H ) HH ) B. (9)
manner, the technique developed in [14] can be applied not only
to sinusoidal supply-voltage cases but also to distorted supply-
voltage conditions.
In 2-D FE analysis, Poissons equation is solved, which is
IV. H YSTERESIS M ODEL a function of magnetic vector potential and reluctivity. When
hysteresis is taken into account, numerical difficulties arise due
The hysteresis model is similar to the JilesAtherton model,
to the discontinuity of reluctivity when B is equal to zero. The
but the independent variable is the flux density B instead of
aforesaid problem is tackled by considering the constitutive
the magnetic field H [15]. In this way, the hysteresis model
equation for ferromagnetic materials [11], [17]
can be integrated in a 2-D FE analysis, where B in each
element can be directly evaluated by the curl of magnetic vector B = 0 (H + M). (10)
potential A. Simplicity, computational efficiency, and ease of
implementation are its main advantages, in contrast to other Combining the aforementioned equation and Amperes law
hysteresis models like Preisach [11]. According to [15], H is yields the following equation for the 2-D problem:
partitioned into two components, namely, the anhysteretic HAN
and the hysteretic HH , as follows: v0 Az + (My /x Mx /y) + Jz = 0. (11)

Considering that the two components of magnetization

H = HAN + HH . (1) are a function of B [17], i.e., Mx = f (Bx , By ) and My =
f (Bx , By ), the differential of M is given by
Furthermore, HH is divided into two parts, i.e., one associ-     
ated with the irreversible part of domain-wall motion HHW , and dMx Mx /Bx Mx /By dBx
= . (12)
the other one associated with the reversible part of domain-wall dMy My /Bx My /By dBy

Authorized licensed use limited to: Themistoklis Kefalas. Downloaded on December 19, 2009 at 13:36 from IEEE Xplore. Restrictions apply.

By using (12) and the 2-D vector potential, the second left-
hand term of (11) is given by
My Mx My Az My Az
x y Bx x y By x x
Mx Az Mx Az
+ . (13)
Bx y y By y x

Also, the relationship between the differentials of magneti-

zation and flux density can be expressed by means of a tensor
[16], [17]. Using the Euler scheme to represent the differentials
Mx |M| cos sin Bx
= (14) Fig. 3. Experimental setup.
My |B| sin cos By

where is the rotation angle between M and B. By

multiplying and dividing the right-hand side of (14) with |B|,
it follows that
Mx Bx
= (15)
My By

where and are given by

M B |M B|
= = . (16)
|B|2 |B|2

From (12) and (15), it follows that

 Mx Mx    Fig. 4. Block diagram of the experimental setup.
Bx By
My My = . (17)


The problems solution is obtained by discretizing (18), The experimental setup used for evaluating the harmonic
which is derived by substituting (13) and (17) into (11) impact on wound-core no-load loss is shown in Fig. 3. The
block diagram of the experimental setup is shown in Fig. 4.
v0 A 23-turn excitation coil was supplied from a single- and
Az + Jz = 0. (18)
v0 three-phase supply via a variable transformer (Regulac, type
RQ 25-M, input: 240 V 50/60 Hz, output: 0275 V, 25 A
The coupled field-circuit global system of equations is given and Regulac, type RK8-G3M, input: 415 V 50/60 Hz, output:
by (19), where D is the potential-current coupling stiffness 0476 V, 8 A) in order to magnetize the wound cores with a
matrix, G is the inductive damping matrix, and A and I are the distorted supply-voltage waveform. A programmable ac and dc
time derivatives of the magnetic vector potential and electric power supply (California Instruments MX30) was used so as to
current vectors, respectively. To solve the time-dependent sys- magnetize the wound cores with a sinusoidal waveform.
tem (19), the Euler backward numerical integration technique The voltage across the excitation coil terminals was captured
is used, and the nonlinearities are taken into account by using using an active differential voltage probe. The specific voltage
the NewtonRaphson iterative scheme probe provides safe means of measuring floating potentials by
        converting the high input differential voltage ( 1400 V peak)
0 0 A S D A 0 into a low voltage ( 5 V). The 3-dB frequency of the voltage
+ = . (19)
G L I 0 R I V probe is 18 MHz, the rejection rate on common mode at 50 Hz
is 90 dB, the dc output offset voltage is 0.15 mV, and the
The important advantage of the proposed methodology is that attenuation factor is equal to 205.
it provides similar accuracy to the other techniques, existing in A current probe based on the Hall effect is used for capturing
the literature [25], while involving faster convergence due to the the no-load current. It provides a galvanic isolation between the
adoption of the modified JilesAtherton model. However, such primary circuit (high power) and the secondary circuit (elec-
techniques involve time-stepping procedures for waveform- tronic circuit), and very good linearity (< 0.2%). The current
distortion consideration and require important computation measurement range is 36 A peak, and the 1-dB frequency of
resources. the probe is 150 kHz.

Authorized licensed use limited to: Themistoklis Kefalas. Downloaded on December 19, 2009 at 13:36 from IEEE Xplore. Restrictions apply.

Fig. 6. Percentage of higher voltage harmonics plot of deformed line-to-line

and line-to-neutral supply voltages.
Fig. 5. Current and active differential voltage probes.
did not change much due to the search coils being introduced.
The voltage and current probes were designed and manufac-
The voltages induced in the search coils were captured by
tured by the authors at the Laboratory of Electrical Machines,
directly connecting the search coils into the NI6143 DAQ card
National Technical University of Athens. The circuit analysis
inputs. Necessary calculations and analysis of the captured data
of the probes was carried out by using OrCAD-PSpice. Both
were carried out by using LabVIEW software.
probes are shown in Fig. 5.
The output of the probes was connected to a noise-rejecting
shielded BNC I/O connector block (BNC-2110) via passive VII. R ESULTS AND D ISCUSSION
probes (TP6060). A noise-rejecting shielded cable (SHC68-68- Seven wound cores of different geometric parameters
EP) connects the data-acquisition (DAQ) device directly to the and core materials, i.e., a conventional grain-oriented (M4
BNC-2110 connector block. 0.27 mm) and a high-permeability (M-OH 0.27 mm) grain-
The DAQ device used was National Instruments (NI) 6143. oriented electrical steel, have been tested at peak flux densities
NI 6143 has eight differential channels, an ADC resolution of up to 1.8 T. Both line-to-line and line-to-neutral voltages were
16 bit, a maximum sampling rate of 250 000 samples/s per used during the no-load test in order to evaluate the effect of
channel analog input, and a 5 V analog-input-signal range. their different harmonic contents. The no-load loss and the
Finally, the DAQ device was placed into a PCI slot of a desktop spectral information of the voltage and no-load current, up
PC (processor: 1.8-GHz Intel Core 2 Duo, memory: 1-GB to the 29th harmonic component, were obtained by the VIs
DDR2). described in Section VI. The results are presented in terms of
Three virtual instruments (VIs) were created with the use of the deterioration factor (DF) that is defined as the ratio of the
LabVIEW software. The purpose of the first VI was the real- no-load loss due to a distorted voltage waveform to the no-load
time measurement of the excitation coil voltage and no-load loss due to a pure sinusoidal voltage, for the same peak flux
current waveforms, fast Fourier transform (FFT), and peak, density [10]. The peak flux density Bp is given by (20), where
rms, and THD values. The second VI was used for capturing V  is the average rectified voltage, f is the frequency, N is the
the voltage and no-load current waveforms and FFT, for two number of turns of the excitation coil, and S is the lamination
periods of the fundamental frequency, and for the maximum cross-sectional area
sampling rate of the DAQ device. The third VI was used for
manipulating the acquired voltage and current data in order to Bp = V /4f N S. (20)
compute the no-load loss.
The repeatability error of the experimental setup is 0.3%, The phase of the voltage harmonics relative to the fundamen-
whereas the absolute error is within 0.5%. Thus, the specific tal, as well as their amplitude, influences the no-load loss [10].
experimental setup permits the evaluation of the harmonic Figs. 6 and 7 show the harmonic contents of the distorted line-
impact on no-load loss, even in the cases of distorted supply- to-line and line-to-neutral supply voltages, which were used as
voltage waveforms of low THD. input for the nonlinear FE transient analysis.
The experimental setup was also used for local-flux-density One-half of the wound-core geometry is modeled due to
measurements in order to evaluate the FE-computed flux- symmetry. Fig. 8 shows the geometry and the respective flux-
density distribution of wound cores. Two-turn search coils density vector plot of one of the seven wound cores tested.
wound around the total width of the steel sheet were inserted in The output of the nonlinear FE transient analysis is the no-load
order to determine the peak-flux-density distribution along the current waveform. The mean power, i.e., no-load loss, is easily
limb, yoke, and corner of the wound cores. A 0.1-mm-diameter obtainable by the voltage and the theoretically determined no-
solderable enameled copper wire was used for the search coils. load current waveforms.
The overall loss of the core did not change noticeably after the The simulated and experimental no-load currents for si-
coils were wound in position, showing that the flux distribution nusoidal, distorted line-to-line, and distorted line-to-neutral

Authorized licensed use limited to: Themistoklis Kefalas. Downloaded on December 19, 2009 at 13:36 from IEEE Xplore. Restrictions apply.

Fig. 7. Phase-angle plot of deformed line-to-line and line-to-neutral supply Fig. 10. Wound-core simulated and experimental current waveforms under
voltages. deformed line-to-line voltage excitation.

Fig. 8. Wound-core geometry and flux-density vector plot (wound-core width:

190 mm, electrical steel: M-OH 0.27 mm).

Fig. 11. Wound-core simulated and experimental current waveforms under

deformed line-to-neutral voltage excitation.


Fig. 9. Wound-core simulated and experimental current waveforms under

sinusoidal voltage excitation.
A comparison of the harmonic content of the experimental
voltage excitations are shown in Figs. 911, respectively. The and simulated no-load currents under sinusoidal, deformed line-
peak flux density is 1.8 T, and f = 50 Hz. to-line, and deformed line-to-neutral supply-voltage waveforms
Table I presents a comparison between the experimental is shown in Figs. 1217. The simulated results are in relatively
and calculated no-load loss, DF, and rms values of no-load good agreement with the experimental ones. Nevertheless, in
current for the sinusoidal and distorted voltage waveforms the case of deformed line-to-line and line-to-neutral excitations,
examined. The deterioration percentages of no-load loss due to there is a large discrepancy in the experimental and simulated
distorted line-to-line and line-to-neutral voltages are 3.46% and phase angles of the 17th, 19th, and 23rd and the 19th, 25th,
1.51%, whereas the rms values of no-load current, an important and 27th harmonics, respectively. These differences explain
transformer design constraint, increase by 37.3% and 27%, the deviance between the experimental and simulated no-load
respectively. current waveforms shown in Figs. 10 and 11.

Authorized licensed use limited to: Themistoklis Kefalas. Downloaded on December 19, 2009 at 13:36 from IEEE Xplore. Restrictions apply.

Fig. 15. Phase-angle plot of experimental and simulated no-load currents

Fig. 12. Amplitude plot of experimental and simulated no-load currents under under deformed line-to-line supply-voltage excitation.
sinusoidal supply-voltage excitation.

Fig. 13. Phase-angle plot of experimental and simulated no-load currents Fig. 16. Amplitude plot of experimental and simulated no-load currents under
under sinusoidal supply-voltage excitation. deformed line-to-neutral supply-voltage excitation.

Fig. 14. Amplitude plot of experimental and simulated no-load currents under Fig. 17. Phase-angle plot of experimental and simulated no-load currents
deformed line-to-line supply-voltage excitation. under deformed line-to-neutral supply-voltage excitation.

Also, Figs. 1217 show the effect of the distorted supply- excitation. Nevertheless, this is not the case for the rest of the
voltage waveform on the no-load current harmonic content. A harmonics. Also, the impact of deformed line-to-line excitation
comparison of Figs. 12, 14, and 16 shows that the amplitude on no-load current is more significant than that of deformed
of the fundamental, third, and fifth harmonics of the no-load line-to-neutral excitation.
current under deformed-voltage-waveform excitation is larger To sum up, the theoretical and experimental results presented
than that of the no-load current under sinusoidal waveform in this paper clearly illustrate that the distorted supply-voltage

Authorized licensed use limited to: Themistoklis Kefalas. Downloaded on December 19, 2009 at 13:36 from IEEE Xplore. Restrictions apply.

waveform of the grid caused by nonlinear loads results in a [13] T. D. Kefalas, P. S. Georgilakis, A. G. Kladas, A. T. Souflaris, and
notable increase of wound-core distribution transformer no- D. G. Paparigas, Multiple grade lamination wound core: A novel tech-
nique for transformer iron loss minimization using simulated annealing
load loss and a significant increase of the magnetizing no- with restarts and an anisotropy model, IEEE Trans. Magn., vol. 44, no. 6,
load current. The deterioration of no-load loss and magnetizing pp. 10821085, Jun. 2008.
current depends not only on the amplitude of voltage harmonics [14] T. D. Kefalas and A. G. Kladas, Robust numerical analysis of wound
core distribution transformers, in Proc. ICEM, 2008, pp. 16.
but also on the phase of voltage harmonics relative to the [15] P. I. Koltermann, L. A. Righi, J. P. A. Bastos, R. Carlson, N. Sadowski,
fundamental. and N. J. Batistela, A modified Jiles method for hysteresis computa-
tion including minor loops, Phys. B, vol. 275, no. 13, pp. 233237,
Jan. 2000.
[16] L. A. Righi, P. I. Koltermann, N. Sadowski, J. P. A. Bastos, R. Carlson,
VIII. C ONCLUSION A. Kost, L. Jnicke, and D. Lederer, Non-linear magnetic field analysis
by FEM using Langevin function, IEEE Trans. Magn., vol. 36, no. 4,
This paper has described the significance of distribution pp. 12631266, Jul. 2000.
[17] P. I. Koltermann, J. P. A. Bastos, N. Sadowski, N. J. Batistela, A. Kost,
transformer no-load loss, as well as the harmonic impact on L. Jnicke, and K. Miethner, Nonlinear magnetic field model by FEM
no-load operation, and has introduced a methodology based on taking into account hysteresis characteristics with M B variables,
FE hysteresis model, enabling one to predict satisfactorily the IEEE Trans. Magn., vol. 38, no. 2, pp. 897900, Mar. 2002.
[18] Y. Han and Y.-F. Liu, A practical transformer core loss measurement
harmonic impact on wound-core distribution transformer no- scheme for high-frequency power converter, IEEE Trans. Ind. Electron.,
load operation. The originality of the proposed method relies vol. 55, no. 2, pp. 941948, Feb. 2008.
on the modified JilesAtherton hysteresis model adopted, ne- [19] R. Prieto, J. A. Cobos, O. Garcia, P. Alou, and J. Uceda, Study of 3-D
magnetic components by means of double 2-D methodology, IEEE
cessitating reduced execution time with respect to other similar Trans. Ind. Electron., vol. 50, no. 1, pp. 183192, Feb. 2003.
techniques. [20] S. Jian and V. Mehrotra, Orthogonal winding structures and design for
The methodology introduced in this paper, if implemented at planar integrated magnetics, IEEE Trans. Ind. Electron., vol. 55, no. 3,
pp. 14631469, Mar. 2008.
the design stage of distribution and power transformers, may [21] C. Buccella, C. Cecati, and F. de Monte, A coupled electrothermal
provide great services in reducing the no-load losses due to model for planar transformer temperature distribution computation,
IEEE Trans. Ind. Electron., vol. 55, no. 10, pp. 35833590, Oct. 2008.
distortion of the supply voltage. [22] I. Villar, U. Viscarret, I. Etxeberria-Otadui, and A. Rufer, Transient
Future research will focus on the modeling of eddy currents thermal model of a medium frequency power transformer, in Proc. IEEE
in the windings of the transformers, caused by current harmon- IECON, 2008, pp. 10331038.
[23] M. R. Feyzi and M. Sabahi, Finite element analyses of short circuit forces
ics, using a coupled thermalelectromagnetic analysis. in power transformers with asymmetric conditions, in Proc. ISIE, 2008,
pp. 576581.
[24] Y.-S. Lee, K. C. Wong, and C. K. Ng, Behavior modeling of magnetizing
currents in switch-mode DCDC converters, IEEE Trans. Ind. Electron.,
R EFERENCES vol. 47, no. 1, pp. 3644, Feb. 2000.
[1] P. Georgilakis, N. Hatziargyriou, and D. Paparigas, AI helps reduce trans- [25] P. J. Leonard, P. Marketos, A. J. Moses, and M. Lu, Iron losses under
former iron losses, IEEE Comput. Appl. Power, vol. 12, no. 4, pp. 4146, PWM excitation using a dynamic hysteresis model and finite element,
Oct. 1999. IEEE Trans. Magn., vol. 42, no. 4, pp. 907910, Apr. 2006.
[2] E. So, R. Arseneau, and E. Hanique, No-load loss measurements of
power transformers under distorted supply voltage waveform conditions,
IEEE Trans. Instrum. Meas., vol. 52, no. 2, pp. 429432, Apr. 2003. Themistoklis D. Kefalas (M09) was born in Greece
[3] P. Pejovic and Z. Janda, An improved current injection network for in 1977. He received the Electrical Engineering Ed-
three-phase high-power-factor rectifiers that apply the third harmonic ucator degree from the School of Pedagogical and
current injection, IEEE Trans. Ind. Electron., vol. 47, no. 2, pp. 497499, Technological Education, Athens, Greece, in 1999
Apr. 2000. and the Diploma and the Ph.D. degree in electrical
[4] C.-L. Chen and G.-K. Horng, A new passive 28-step current shaper engineering from the National Technical Univer-
for three-phase rectification, IEEE Trans. Ind. Electron., vol. 47, no. 6, sity of Athens, Athens, Greece, in 2005 and 2008,
pp. 12121219, Dec. 2000. respectively.
[5] C. Rech and J. R. Pinheiro, Line current harmonics reduction in multi- He is currently with the School of Electrical
pulse connection of asymmetrically loaded rectifiers, IEEE Trans. Ind. and Computer Engineering, National Technical
Electron., vol. 52, no. 3, pp. 640652, Jun. 2005. University of Athens. His research interests include
[6] A. K. S. Bhat and R. Venkatraman, A soft-switched full-bridge single- transformer and electric machine modeling and optimization.
stage AC-to-DC converter with low-line-current harmonic distortion,
Dr. Kefalas is a member of the Technical Chamber of Greece.
IEEE Trans. Ind. Electron., vol. 52, no. 4, pp. 11091116, Aug. 2005.
[7] P. Lezana, J. Rodriguez, and D. A. Oyarzun, Cascaded multilevel inverter
with regeneration capability and reduced number of switches, IEEE
Trans. Ind. Electron., vol. 55, no. 3, pp. 10591066, Mar. 2008. Antonios G. Kladas (S80A99M02) was born
[8] S. G. Song, F. S. Kang, and S. J. Park, Cascaded multilevel inverter in Greece in 1959. He received the Diploma in
employing three-phase transformers and single DC input, IEEE Trans. electrical engineering from the Aristotle Univer-
Ind. Electron., vol. 56, no. 6, pp. 20052014, Jun. 2009. sity of Thessaloniki, Thessaloniki, Greece, in 1982
[9] P. Flores, J. Dixon, M. Ortuzar, R. Carmi, P. Barriuso, and L. Moran, and the D.E.A. and Ph.D. degrees from Pierre and
Static var compensator and active power filter with power injection Marie Curie University (Paris 6), Paris, France, in
capability, using 27-level inverters and photovoltaic cells, IEEE Trans. 1983 and 1987, respectively.
Ind. Electron., vol. 56, no. 1, pp. 130138, Jan. 2009. He was an Associate Assistant with Pierre and
[10] A. J. Moses and G. H. Shirkoohi, Iron loss in non-oriented electrical Marie Curie University from 1984 to 1989. From
steels under distorted flux condition, IEEE Trans. Magn., vol. 23, no. 5, 1991 to 1996, he was with the Public Power Cor-
pp. 32173220, Sep. 1987. poration of Greece, where he was engaged in the
[11] H. L. Toms, R. G. Colclaser, and M. P. Krefta, Two-dimensional finite System Studies Department. Since 1996, he has been with the School of
element magnetic modeling for scalar hysteresis effects, IEEE Trans. Electrical and Computer Engineering, National Technical University of Athens,
Magn., vol. 37, no. 2, pp. 982988, Mar. 2001. Athens, Greece, where he is currently a Professor. His research interests include
[12] B. Kawkabani, G. Rosselet, and J. J. Simond, Combined transformer and electric machine modeling and design, as well as the analysis
analyticalnumerical approach for the modeling and analysis of of generating units by renewable energy sources and industrial drives.
three-phase transformers, in Proc. IEEE IECON, 2006, pp. 15211526. Dr. Kladas is a member of the Technical Chamber of Greece.

Authorized licensed use limited to: Themistoklis Kefalas. Downloaded on December 19, 2009 at 13:36 from IEEE Xplore. Restrictions apply.