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

net/publication/318266536

Electromagnetic Finite Element Analysis of Electrical Steels Combinations in

Lamination Core Steps of Single-Phase Distribution Transformers

Conference Paper · November 2017

DOI: 10.1109/ROPEC.2017.8261585

CITATIONS READS

11 1,895

4 authors:

Salvador Magdaleno-Adame Themistoklis D. Kefalas

Salvador Consultant National Technical University of Athens
66 PUBLICATIONS 377 CITATIONS 53 PUBLICATIONS 699 CITATIONS

SEE PROFILE SEE PROFILE

Sigridt García Catalina Ines Rojas Perez

Universidad Michoacana de San Nicolás de Hidalgo Universidad de las Américas (Chile)
21 PUBLICATIONS 169 CITATIONS 17 PUBLICATIONS 55 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Themistoklis D. Kefalas on 16 February 2018.

The user has requested enhancement of the downloaded file.

 2017 IEEE International Autumn Meeting on Power, Electronics and Computing (ROPEC 2017). Ixtapa, Mexico
Electromagnetic Finite Element Analysis of
Electrical Steels Combinations in Lamination
Core Steps of Single-Phase Distribution
Transformers
Salvador Magdaleno-Adame Themistoklis D. Kefalas
Magnetic Engineering Hellenic Electricity Distribution
Department Network Operator HEDNO S.A.
Oersted Technology Athens, 10434, Greece
Oregon, 97055, USA t.kefalas@deddie.gr
smagdalenoa@hotmail.com
Abstract—This paper presents an electromagnetic finite
element (FE) analysis of combinations of electrical steels in the
lamination core steps of a real 6.3 MVA single-phase distribution
transformer. The magnetic core of this transformer has a
cruciform cross-section with lamination core steps. Two electrical
steels are combined in the lamination core steps of transformer: a
convectional grain oriented electrical steel (M-5) and laser-
scribed electrical steel (23ZDKH90). 3-D FE simulations are
performed to calculate the core losses (no-load losses) without
and with combinations of electrical steels. B-H curves and iron
loss curves of electrical steels are taken into account in the
numerical simulations. The core loss calculated in FE simulation
without combination of steels is compared with the core loss
measured in no-load laboratory tests. Numerical results show
that the combination of electrical steels in the lamination core
steps can reduce 5% the core losses in single-phase distribution
transformers with stacked magnetic cores. Finally, material costs
are estimated for the steel combinations in the magnetic core of
transformer.
Keywords—core loss; distribution transformer; electrical steel;
no-load loss; finite element; lamination core step
I. INTRODUCTION
The combinations of electrical steels in magnetic cores of
transformers are employed by manufacturers to reduce losses,
temperature, energy, and costs [1]-[16]. Actually in USA,
Mexico, Greece, Germany, Italy, Japan, and Poland some
transformer companies are combining electrical steels and high
permeability alloys in magnetic cores of transformers.
Furthermore, in the last twenty years, the combinations of
electrical steels and ferromagnetic materials have been studied
and applied in instrument, power, and distribution transformers
and new magnetic core topologies have been proposed to
combine electrical steels in distribution and power transformers
[2]-[16]. Some authors have analyzed the combination of grain
oriented electrical steels (GOESs) in magnetic cores of
distribution and power transformers to reduce core losses,
978-1-5386-0819-7/17/$31.00 ©2017 IEEE
Sigridt García-Martínez Carlos Perez-Rojas
Electrical Engineering Electrical Engineering
Department Department
U.M.S.N.H. U.M.S.N.H.
Morelia, 58030, Mexico Morelia, 58030, Mexico
sigridt@umich.mx cprojas@umich.mx
temperature, manufacturing costs, and material costs [2]-[4].
Other authors have analyzed combinations of GOESs in
magnetic wound cores of distribution transformers to reduce
losses [5]-[10]. Other authors have combined GOESs and non-
grain oriented electrical steels (NGOESs) in instrument
transformers, and other authors have combined GOESs and
some high permeability alloys in toroidal magnetic cores for
current transformers to reduce material costs and to improve
their measurement properties [11]-[15]. Finally, some authors
have combined GOESs and amorphous steels in magnetic
cores of distribution transformers to reduce losses and material
costs [16].
Furthermore, after reviewing the existing literature, the
authors note that there is not information related with
combinations of GOESs in the lamination core steps of
distribution transformers. J.C. Granfield in 1949 proposed to
combine GOESs and NGOESs in lamination core steps in
stacked cores of transformers, and reactors, see Fig. 1 [1]. He
proposed to use GOES laminations in the internal core steps
and NGOES laminations in the outer core steps to reduce
power losses [1]. In this paper the idea of J.C. Granfield is
utilized and analyzed but considering only GOESs with
different magnetization and iron loss (W/kg) properties and
with different lamination thickness in the core steps of a real
distribution transformer.
This paper presents a magnetic flux and core loss numerical
analysis of combinations of electrical steels in a magnetic core
of a real 6.3 MVA single-phase distribution transformer. The
magnetic core of this transformer has a cruciform cross-
section. A convectional GOES (M-5) and a laser-scribed
electrical steel (LSES) (23ZDKH90) are combined in the
lamination core steps of transformer. 3-D finite element (FE)
simulations are performed to compute the core losses and
magnetic flux distributions in the transformer magnetic core
with and without combinations of electrical steels.
Furthermore, the core loss of transformer without combinations
of electrical steels is measured in laboratory and compared
 2017 IEEE International Autumn Meeting on Power, Electronics and Computing (ROPEC 2017). Ixtapa, Mexico

with FE simulation. With this comparison, authors validate the

FE simulations in this paper. The magnetization properties and TABLE I.
iron loss properties of electrical steels are taken into account in TECHNICAL CHARACTERISTICS OF DISTRIBUTION TRANSFORMER
the nonlinear FE simulations. No. phases 1
Frequency 50 Hz
Impedance 7.83%
II. 3-D FINITE ELEMENT SIMULATIONS
Transformer rating 6.3 MVA
The distribution transformer is modeled in three dimensions High voltage (HV) winding 10.5 kV
(3-D). ANSYS Maxwell software is utilized to compute the Low voltage (LV) winding 710 V
losses in the magnetic core with and without combination of
electrical steels [17].
A quasi-static magnetic vector potential formulation is
utilized to solve 3-D eddy current problems given by:
§1 ·
∇ × ¨ (∇× A) ¸ = J e − jωσ A (1)
©μ ¹
where μ is the permeability (H/m), A is the magnetic vector
potential (Wb/m), Je is the current density (A/m2), Ȧ is the
angular frequency (rad/s), and ı is the electrical conductivity
(S/m). The finite element method (FEM) is utilized to solve
(1) in the domain of transformer model.
A. Exciting Windings Modeling
For 6.3 MVA single-phase distribution transformer only the
low voltage (LV) winding is used to magnetize the core of
transformer, the high voltage (HV) winding is omitted in order Fig. 1. 3-D model of magnetic core of 6.3 MVA distribution transformer.
to provide open circuit test conditions. The exciting winding is
made of aluminum with a relative permeability of 1 and an
electrical conductivity of 3.8×107 S/m. The excitation winding
has 18 turns. The power losses produced in the excitation
winding are omitted. Fig. 1 shows the 3-D distribution
transformer model and Table I shows the technical
specifications of distribution transformer.
B. Magnetic Core Modeling
The original and real magnetic core of the transformer is
built using the step-lap overlap technique. In the 3-D FE
simulations the ‘45°’ overlap core joints are not taken into
account in this paper. To reduce the computational complexity
inter-laminar layers are avoided and only the solid lamination
steps are considered in the FE simulations.
The real magnetic core is made of laminations of M-5
electrical steel [18]. This is a convectional GOES utilized in
the manufacturing process of distribution transformers. The Fig. 2. View of cruciform cross-section and lamination core steps of magnetic
core of distribution transformer.
M-5 steel laminations have a thickness of 0.30 mm, a mass
density of 7650 kg/m3, and an electrical conductivity of
The experimental B-H curve and the specific iron loss curve
1.96×106 S/m. Furthermore, 23ZDKH90 steel is a GOES
of M-5 and 23ZDKH90 steel at 50 Hz are used in the FE
treated using the laser scribing method [19], [20]. The laser
simulations [18], [19]. These curves are obtained from the
scribing process refines the magnetic domains, increases the
steel manufacturers using an Epstein frame. Fig. 3 shows the
magnetic permeability, and reduces the losses in the steel [20].
specific iron loss curves for M-5 and 23ZDKH90 steel at 50
The 23ZDKH90 steel laminations have a thickness of 0.23
Hz. The magnetic core of distribution transformer is designed
mm, a mass density of 7650 kg/m3, and an electrical
to operate at 1.63 T.
conductivity of 2×106 S/m [19]. In the distribution
transformer, the electrical steels M-5 and 23ZDKH90 are
combined in the lamination core steps of transformer. Fig. 2
shows a view of the lamination core steps of the cruciform
cross-section of distribution transformer.
 2017 IEEE International Autumn Meeting on Power, Electronics and Computing (ROPEC 2017). Ixtapa, Mexico

D. Finite Element Mesh of Transformer

Fig. 3. Experimental specific iron loss curves for M-5 and 23ZDKH90 steel
at 50 Hz [18], [19].
C. Core Losses in Distribution Transformer
The core losses in the distribution transformer are computed
by ANSYS Maxwell software utilizing a numerical dynamic
core loss model (DCLM) [21]. Utilizing this core loss
numerical model, one avoids the use of very small finite
elements in the magnetic core models of transformers or
electrical machines [21]. A hysteresis loss coefficient (kh) and
an eddy current loss coefficient (kc) are used to compute the
total core losses in magnetic cores of transformers. DCLM
uses the magnetic field distributions and the loss coefficients
to compute the losses in the core of distribution transformer
with and without combinations of steels. DCLM utilizes
equivalent elliptical loops to simulate the hysteresis cycles in
the core of transformer and electrical machines [21]. Utilizing
these hysteresis loops ANSYS Maxwell can calculate the
hysteresis losses in the core of distribution transformer [21].
Furthermore, kc coefficient is given by [21]:
A total of 113,000 finite elements are used in the
transformer model. Fig. 4 shows the FE mesh of the magnetic
core of the distribution transformer.
E. Nonlinear Time-Harmonic Analyses
Nonlinear time-harmonic analyses are carried out to
compute the core losses in the transformer core. An exciting
current Iexc = 8.3 A at 50 Hz is injected in the LV winding to
magnetize the magnetic core. The excitation current of this
transformer is measured in laboratory during the no-load test
and corresponds to 0.1% of the nominal current of
transformer.
F. Core Losses in Magnetic Core without Combination of
Electrical Steels
Fig. 5 shows the magnetic flux density distributions for core
of distribution transformer without combinations of electrical
steels.
Table III shows the core loss computed and measured for
distribution transformer without combinations of electrical
steels. A difference of 3.48% is calculated between the core
loss calculated in FE simulation and the core loss measured in
no-load test.

π 2σ s d s2
kc = (2)
6
where ds is the thickness of electrical steel laminations and ıs
is the electrical conductivity of electrical steel laminations.
The loss coefficients kh and kc for electrical steels in the
Fig. 4. Finite element mesh of distribution transformer.
magnetic core of the transformer are computed by ANSYS
Maxwell software utilizing a regression method, the specific
iron loss curves of electrical steels, and properties of electrical
steels.
Table II shows the hysteresis and eddy loss coefficients for
the M-5 and 23ZDKH90 steels at 50 Hz computed by ANSYS
Maxwell software.

TABLE II.
LOSS COEFFICIENTS UTILIZED TO COMPUTE CORE LOSSES IN MAGNETIC
CORE OF DISTRIBUTION TRANSFORMER
Loss M-5 23ZDKH90
Coefficient (W/m3) (W/m3)
kh 51.92 43.26
kc 0.29 0.17

Fig. 5. Magnetic flux density distribution in magnetic core of transformer with

M-5 steel.
 2017 IEEE International Autumn Meeting on Power, Electronics and Computing (ROPEC 2017). Ixtapa, Mexico

TABLE III. of 1.805 T is calculated in the combined core in the core steps
CORE LOSS IN MAGNETIC CORE WITHOUT ELECTRICAL STEEL with 23ZDKH90 steel.
COMBINATIONS IN DISTRIBUTION TRANSFORMER
Core Loss Core Loss Loss TABLE IV.
(FE analysis) (Measured) difference CORE LOSSES AND LOSS DIFFERENCE OF STEEL COMBINATIONS
Total core
4.15 kW 4.30 kW 3.48% Steel Core Loss Loss
loss
Combination (kW) difference
A1 4.15 −
A2 4.10 1.20%
G. Core Losses in Magnetic Core with Combinations of A3 4.03 2.89%
A4 3.95 4.82%
Electrical Steels
The different combinations of electrical steels in lamination
core steps of distribution transformer are shown in Fig. 6.

Fig. 7. Magnetic flux density distributions in lamination core steps.

Fig. 6. Electrical steel combinations in lamination core steps of transformer.

The larger volume of steel in the main and central core steps
contributes higher amount of losses. Hence, LSES is preferred
in the central steps and GOES is applied on the outer steps.
Figs. 7 and 8 show the magnetic flux density and core loss
distributions in the core steps of the distribution transformer for
the different combinations of steels. From these figures one
can see that the magnetic flux and loss distributions change
when one increases the amount of LSES in the core steps. The
increment of LSES in the inner core steps and the reduction of
GOES in the outer steps produce a reluctance change in the
magnetic core. One part of the magnetic flux prefers to
circulate in the inner core steps where the permeability is high
and the losses (W/kg) are smaller and another part of the
magnetic flux is forced to circulate in the outer core steps
where the permeability is lower and the losses (W/kg) are a
little higher. This imbalance of flux and losses produces a
reduction of the total core loss in the magnetic core of
distribution transformer. Fig. 8. Volumetric core loss distributions (W/m3) in lamination core steps.
Table IV shows the core losses calculated for the steel
combinations in the magnetic core of distribution transformer.
This table includes the loss difference between the steel III. CORE MATERIAL COST FOR STEELS COMBINATIONS
combinations in magnetic core and the magnetic core without Actually, LSESs are more expensive than convectional
steel combinations. The authors calculated the average GOESs. The average cost of GOESs is $2.6 USD/kg and the
magnetic flux density (Bavg) in the cruciform cross-section for average cost of LSESs is 3.6 USD/kg. Table V shows the
each combination of steels. A value Bavg = 1.72 T is calculated material costs for the different steel combinations made in the
for the steel combinations. Moreover, a maximum flux density core of transformer. From this table, one can see that the total
 2017 IEEE International Autumn Meeting on Power, Electronics and Computing (ROPEC 2017). Ixtapa, Mexico

cost of the magnetic core is increased when the amount of [3] A.J. Moses, S. Hamadeh, “Effects of mixing materials on losses and cost
LSES is increased in the lamination core steps. A maximum of operation of three-phase transformer cores,” Journal of Applied
Physics, vol. 64, no. 10, pp. 5379-5381, 1988.
cost increment of 39% is calculated for this distribution [4] D. Snell, A. Coombs, ‘‘The influence of core materials and mix on the
transformer. performance of a 100 kVA three phase transformer core,’’ J. Magn. &
TABLE V. Magn. Mater., vol. 254---255, pp. 349-351, 2003.
MATERIAL COSTS OF STEEL COMBINATIONS IN DISTRIBUTION TRANSFORMER [5] T.D. Kefalas, A.G. Kladas, “Mixed Si-Fe wound cores five legged
transformer: losses and flux distribution analysis,” IEEE Trans.
Steel M-5 23ZDKH90 Total Core cost
Magnetics, vol. 48, no. 4, pp.1609-1612, 2012.
Combination region cost region cost core cost increment
[6] T.D. Kefalas, A.G. Kladas, “Reduction of cost and losses of
($USD) ($USD) ($USD)
transformers by using composite magnetic cores,” 2014 International
A1 9769.15 − 9769.15 − Conference on Electrical Machines (ICEM), pp. 2257-2263, 2014.
A2 6569.56 4172.68 10742.25 9.96%
[7] T.D. Kefalas, A.G. Kladas, “Development of distribution transformers
A3 3427.73 8522.91 11950.64 22.33% assembled of composite wound cores,” IEEE Trans. Magnetics, vol. 48,
A4 547.20 13026.38 13573.58 38.94% no. 2, pp. 775-778, 2012.
[8] T.D. Kefalas, P.S. Georgilakis, A.G. Kladas, A.T. Souflaris, D.G.
Paparigas, “Multiple grade lamination wound core: a novel technique
Furthermore, the electrical steel combinations in the core of for transformer iron loss minimization using simulated annealing with
transformer don’t require investing on specialized machinery restarts and an anisotropy model,” IEEE Trans. Magnetics, vol. 44, no.
or changes on their existing production line. In addition, the 6, pp. 1082-1085, 2008.
labor costs are not increased during the manufacturing process [9] T. D. Kefalas, ‘‘Transformers made of composite magnetic cores: An
of these combined cores. The cutting and stacking processes innovative design approach,’’ Journal of Recent Patents on Electrical
Engineering, vol. 2, no. 1, pp. 1---12, 2009.
of laminations in the core are not modified during the [10] I. Hernandez, J.C. Olivares-Galvan, P.S. Georgilakis, J.M. Canedo, “A
manufacturing of transformer. Finally, the energy costs of the novel octagonal wound core for distribution transformers validated by
manufacturing process of these combined cores don’t increase. electromagnetic field analysis and comparison with convectional wound
core,” IEEE Trans. Magnetics, vol. 46, no. 5, pp. 1251-1258, 2010.
[11] R. Pytlech, “Mixed D-core with a distributed air gap as an alternative for
IV. CONCLUSIONS medium voltage instrument transformers,” IEEE Trans. Magnetics, vol.
The magnetic field distributions and core losses in a 46, no. 10, pp. 3816-3825, 2010.
[12] E. Lesniewska, R. Rajchert, ‘‘3D field---circuit analysis of measurement
magnetic core of a real 6.3 MVA single-phase distribution
properties of current transformers with axially and radially connected
transformer without and with combinations of electrical steels cores made of different magnetic materials,’’ Progress In
in the lamination core steps are analyzed utilizing 3-D FE Electromagnetics Research M, vol. 28, pp. 1---13, 2013.
simulations. [13] E. Lesniewska, R. Rajchert, “Application of the field-circuit method for
the computation of measurement properties of current transformers with
Authors concluded that the core losses in a single-phase cores consisting of different magnetic materials,” IEEE Trans.
distribution transformer can be reduced a maximum of 5% Magnetics, vol. 46, no. 10, pp. 3778-3782, 2010.
combining GOESs and LSESs in the lamination core steps of [14] J. Kwiczala, B. Kasperczyk, “Composite cores in current transformers,”
transformer. Magnetic saturation problems were not found in 16th International Conference on Composite Materials (ICCM-16), pp. 1-
5, Kyoto, Japan, 2007.
the electrical steels combinations in the magnetic core of [15] P. Sankaran, V. Jagadeesh Kumar, V. Jayashankar, “Error estimation of
transformer. In addition, authors concluded that the current transformers using single or composite cores of ferromagnetic
combinations of GOESs and LSESs increase the material costs materials,” IEEE Trans. Magnetics, vol. 26, no. 6, pp. 3086-3088, 1990.
of transformer in 39% but they don’t increase the labor and the [16] M. Mohan, P.K. Singh, “Distribution transformer with amorphous-
energy costs during the manufacturing process of transformer. CRGO core: an effort to reduce the cost of amorphous core distribution
transformer,” ARPN Journal of Engineering and Applied Sciences, vol.
The methodology presented in this paper can be applied to 7, no. 6, pp. 680-685, 2012.
combine electrical steels in power, distribution, and instrument [17] “User’s Guide – Maxwell 3D/ANSYS Maxwell v.15,” REV6.0 ed.
ANSYS Inc., Canonsburg, PA, USA, 2012.
transformers with stacked magnetic cores.
[18] CARLITE Grain oriented electrical steels M-3, M-4, M-5, M-6–
Authors are going to perform laboratory tests of magnetic product data bulletin, AKSteel, West Chester, Ohio, USA, 2012.
cores (prototypes) with different electrical steels combinations [19] Grain-Oriented Electrical Steels Sheets, Nippon Steel Corporation,
Tokyo, Japan, 2004.
to measure and to analyze the core loss, temperature, and [20] L. Lahn, C. Wang, A. Allwardt, T. Belgrand, J. Blaszkowski, “Improved
magnetic flux distributions. transformer noise behavior by optimized laser domain refinement at
ThyssenKrupp electrical steel,” IEEE Trans. Magnetics, vol. 48, no. 4,
pp. 1453-1456, 2012.
ACKNOWLEDGMENT [21] D. Lin, P. Zhou, W. N. Fu, Z. Badics, and Z. J. Cendes, “A dynamic
The authors also thank Sudheer Mokkapaty R&D core loss model for soft ferromagnetic and power ferrite materials in
Transformer Engineer at SBG Transformers for transformer transient finite element analysis,” IEEE Trans. Magnetics, Vol. 40, No.
2, pp. 1318–1321, 2004.
model data and for permitting publication of this work.

REFERENCES
[1] J.C. Granfield, Magnetic Core, US Patent 2 465 798, 1949.
[2] S. Magdaleno-Adame, E. Melgoza-Vazquez, J.C. Olivares-Galvan, R.
Escarela-Perez, “Loss reduction by combining electrical steels in the
core of power transformers,” Int. Trans. Electr. Energ. Syst., vol. 26, no.
8, pp. 1737-1751, 2016.

View publication stats