УДК 621.

314
NUMERICAL SIMULATION OF LOSSES AND HEATING IN THE CONSTRUCTIONAL
ELEMENTS OF TRANSFORMERS OF FERROMAGNETIC STEEL

V.F.Ivankov1, A.V.Basova1, I.V.Khimjuk2

1
− PJSC "Zaporizhtransformator",
Dnipropetrovske shose, 3, Zaporizhia, 69600, Ukraine.
2
− Institute of electrodynamics of NAS of Ukraine,
Peremohy, 56, Kyiv-57, 03680, Ukraine, e-mail: [email protected]

The application of the methods of calculation of surface and volume losses in the constructional elements of transform-
ers of ferromagnetic steel is considered. A brief description of the developed practical procedures of their calculation
using the ANSYS software is given. The testing results of the proposed procedures conducted on the known experimen-
tal studies TEAM21, as well as calculation samples of losses and heating in the constructional elements of a single-
phase autotype transformer with capacity of 500 MVA and reactor of ROM type with capacity of 80 MVAr, are pro-
vided. References 16, table 1, figures 16.
Keywords: powerful electrical transformer, losses, heating, three-dimensional modeling.

1. Introduction. Additional losses in the constructional elements of power transformers with significant
specific electromagnetic loads may lead to the unacceptable heating. That is why researchers and designers
pay constant attention to the issues of local losses and heating calculation, as well as the total losses in trans-
formers.
The experimental studies, using natural samples of the equipment and the corresponding to it physical
models, were conducted and continue to be conducted. The mathematical models and methods of calculation
of this problem class are constantly improved. Significant results have been achieved through the extensive
use of such invariant systems of numerical finite-element analysis (FEA) as ANSYS, Comsol, ELCUT,
ELEKTRA, FEMM, Infolytica, OPERA, etc. A systematic representation of the known publications on this
problem is given, for example, in the following books: [1, 4−6, 13, 15, 16].
The issues related to the definition of magnetic field in the constructional elements of ferromagnetic
steel with a pronounced nonlinear surface effect, are particularly complex when determining the local losses
and heating, as well as the total losses in transformers. The known methods of research and calculations de-
voted to these issues in the power transformer equipment can be roughly classified into two groups of meth-
ods: surface and volumetric losses.
The method of surface losses is based on the fact that in the massive conductive bodies with a pro-
nounced non-linear surface effect, the electromagnetic field is concentrated in a thin surface layer. [6]. In this
case, the conductive body shows up as an equivalent impedance surface, and the losses are determined by
using an input impedance (surface resistance) of the body and the value of the tangential component of mag-
netic field intensity on its surface [4, 5, 16].
However, the method of surface losses does not provide the calculations of the bodies of limited dimen-
sions. For example, in the power transformers they are yoke beams, pressing metal rings, and lifting plates
on the rods of magnetic system (MS). The losses calculation in the yoke beams is further complicated by the
fact that they often have horizontal shelves of non-magnetic steel, and vertical walls from a ferromagnetic
steel. Electric welding of such parts of the construction leads to the exchange of the eddy currents in their
volumes. Such can be observed when realizing the inserts of non-magnetic steel in ferromagnetic caps and
tank walls of transformers and reactors, that is used to reduce the local losses of the total magnetic fields of
windings and taps with significant currents [8].
Additional losses in the constructional elements of transformers of ferromagnetic steel can be deter-
mined numerically by using the method of volumetric losses [2, 4, 5, 9−11]. This method provides a way to
determine the field vectors and energy parameters directly in the volume of conductive bodies. The method
of volume losses has been investigated numerically, and is represented in several publications of the authors
of testing TEAM method (Testing Electromagnetic Analysis Methods) on physical models [9−11]. Numer-
ous studies of a family of TEAM21 problems using Infolytica software are also given on the website [12].
The research of the problems for two plates with a section of nonmagnetic steel, and two plates of ferromag-
netic steel are given in [14].

© Ivankov V.F., Basova A.V., Khimjuk I.V., 2014

ISSN 1607-7970. Техн. електродинаміка. 2014. № 3 19

 It should be noted that in the KEA ANSYS [7] software there are no standard procedures for calculating
the losses in the bodies of ferromagnetic steel with a nonlinear surface effect. Therefore, the additional de-
velopment and testing of these calculation procedures are needed when using ANSYS software.
The aim of this work is to represent the main methods and examples of losses calculation in the con-
structional elements of power transformer equipment of ferromagnetic constructional steel, and evaluate their
heating on the basis of the developed additional procedures of numerical simulation using ANSYS software.
The main approaches to the calculation of electromagnetic field of transformers.
Electromagnetic processes in the power transformer equipment are described by the system of Max-
well's equations relative to the vectors of intensity and induction of magnetic field H , B , and intensity and
induction of electric field E , D [2, 4, 5, 13, 15, 16]
∇ × H = j + ∂D / ∂t , ∇ × E = −∂ B / ∂ t , ∇ ⋅ D = ρ , ∇⋅ B = 0 . (1)
The ratio between the field vectors are determined by constitutive equations
j = jces + σ E , D = εE , B = μ H , (2)
where σ − electrical conductivity; μ − magnetic permeability; ε − dielectric permeability.
The bulk densities of electric charge ρ and current j are the sources of electromagnetic field, and are
connected with the equation ∇ ⋅ j + ∂ρ / ∂t = 0 .
In the ratios (2), the electrical conductivity σ and magnetic permeability μ for unisotropic mediums
are tensor quantities [4, 5, 13, 15, 16]. In many problems the nonlinear dependence of electrical conductivity
σ on temperature θ is taken. For ferromagnetic mediums the relation between B and H will be nonlinear
[4, 5, 6]. At the industrial frequencies the displacement currents are neglected, which leads to the known ra-
tios: ∂D / ∂t = 0 and ρ = 0 .
The harmonic change of the currents from equations (1) leads to the second-order differential equation
with respect to the complex amplitudes H and E [2,4,16]
∇ 2 H − k 2 H = −∇ × j , E = σ −1 ( ∇ × H − j ) , (3)
where k = − iωσμ − phase constant.
Similar equations also determine the relationship between the amplitudes E and H .
To reduce the number of equations (3), the auxiliary vector potential function A by definition B = ∇ × A
is entered. In this case, instead of two equations (3) we get a single equation for the potential A [2, 4, 16]
∇2 A − k 2 A = − μ j . (4)
In this case the field vectors are
H = μ −1∇ × A , E = iωA − iωk −2∇∇ ⋅ A . (5)
For the problems of determining the magnetic field of permanent currents k = 0 and equation (4) will
be Poisson's equation at controlled currents of external sources jces . For the problems of determining a vari-
able magnetic field k ≠ 0 and the equation (4) will be Helmholtz equation, at which the electromagnetic
field is generated by external sources as well as eddy currents in the conductive bodies j = jces + σ E .
There are also other problems solutions of electromagnetic fields with the use of other auxiliary poten-
tials [2, 4, 16]. To solve the boundary problems, the Maxwell's equations are complemented by known
boundary relations at the interface surface of mediums and boundary conditions at infinity. For electromag-
netic fields that depend on time, the initial conditions must be given as well [2, 4, 16]. In analytic solutions
when determining the field in the air near conducting medium with a strong surface effect, the so-called Le-
ontovich approximate boundary conditions of impedance type are used [2, 4, 16].
In this paper, in order to conduct electromagnetic calculations of transformer equipment, the methods of
stationary, harmonic and nonlinear harmonic analysis of ANSYS software [7] are used, and only necessary
explanations to the practical procedures developed by means of ANSYS for certain problems and corre-
sponding to them computational models and calculation methods are represented.
To achieve the aim of this work it is necessary to develop more specific procedures to calculate the energy
parameters in ferromagnetic bodies. The main algorithmic features of these procedures are set out below.

20 ISSN 1607-7970. Техн. електродинаміка. 2014. № 3

 The energy characteristics of electromagnetic field are expressed through dependencies [2, 4]
wem = we + wm , we = E ⋅ D / 2 , wm = B ⋅ H / 2 , pces = jces ⋅ E , p j = j ⋅ E , П = E × H . (6)
They determine the bulk wem density of electromagnetic energy as the sum of electric we and magnetic
wm components, bulk densities of loss power from eddy p j currents and extraneous currents pces , as well as
the energy flux density of electromagnetic field or Poynting vector П .
The features (6) are linked by Umov-Poynting theorem relative to surface density of energy power
transferred by electromagnetic field through the surface S in the direction to the outward normal n and
losses in the volume V
∂
∫ wem dv = ∫ pces dv − ∫ p j dv − ∫ ( E × H ) ⋅ nds .
∂t V
(7)
V V S
On the basis of (7), the relationship between the Poynting vector and power densities of electric losses
and magnetic energy in complex form is set, which in the absence of external currents in the constructional
elements of transformers can be written as follows [4]

−
1
[
2 ∫S
E × H *
]⋅ n ds =
1
2 V∫
j ⋅ E *
dv − i ω
μH ⋅H*
∫V 2 dv . (8)

The given equations allow to determine the similarity and differences of the methods for calculating surface
and volumetric losses. Common to both methods is the need to determine the electromagnetic field vectors. The
difference is that for the method of surface losses, the determination of the field vectors on the surface of bodies
under review, and for the method of volumetric losses − within their volume, is necessary and sufficient.
The properties of structural steels. When determining the vectors of field and energy parameters tak-
ing into account the constitutive equations (2), it is usually assumed that the electrical conductivity of the
constructional steels is a constant. Later, after the thermal calculation, this parameter can be updated. For
nonmagnetic steels the magnetic permeability is a constant, which is slightly different from the magnetic
permeability of air.
More complicated is the account of the nonlinear and hysteresis relationship between tension and induc-
tion in ferromagnetic steels. The known magnetic permeability along the normal induction curve is defined
as the locus of the vertices of hysteresis symmetric loops. At the cyclic magnetization of ferromagnetic mate-
rial from the demagnetized state, its magnetization occurs through the so-called ascending and descending
parts of the hysteresis loops.
For the important in engineering practice case of harmonic field change ω = 2π / T , a complex mag-
netic permeability μ 1 on the first harmonic is used.
A complex magnetic permeability μ 1 is determined by the complex amplitudes on the first harmonics
T T
2 2
B (t ) e iωt dt , H (t ) eiωt dt .
T ∫0 T ∫0
B1 and H1 [4 − 6] μ 1 = B1 / H 1 , B1 = H1 = (9)

When integrating (9), the values of the area of hysteresis loops that determine the value μ 1 , are deter-
mined numerically [4] or by planimetering [6] the area bounded by the descending and ascending loop
+ H1m
branches with vertex H m S ( H1m ) = ∫ ⎡⎣ BD ( H , H1m ) − B A ( H , H1m ) ⎤⎦ dH , (10)
− H1m

where S(H 1 m ) − square of hysteresis loop.

The graphic representations μ 1 and analytical descriptions of the dependence μ 1 ( H ) for different parts of
the magnetization curve are given in [4–6]. The introduction of μ 1 allow to trace the system of Maxwell equations
to the equations of first harmonic, which differ from the equations (1), (2) with respect to the complex amplitudes of
the field for a medium with constant parameters σ and μ only with μ for μ 1 ( H ) replacement.
The introduction of complex magnetic permeability μ 1 on the first harmonic enabled to solve several
problems of electromagnetic field propagation in ferromagnetic half-space with a surface effect, without and
taking into account the hysteresis loss [4, 6]. The obtained solutions allow to determine that at the "suffi-

ISSN 1607-7970. Техн. електродинаміка. 2014. № 3 21

ciently strong fields H that exceed the intensity value at which μ has a maximum" of active losses in fer-
romagnetic with hysteresis, with more losses in ferromagnetic with a constant value of magnetic permeabil-
ity of about 1.4 times [6]. The given ratio was also obtained by solving the problem using numerical method
of grids, the results of which are described in [16]. To exclude the adopted in [6] reduction in [4, 5], the so-
called equivalent magnetic permeability on the first harmonic μ Э was introduced. The analytical methods
for calculating equivalent magnetic characteristics of the known static magnetic characteristics, and experi-
mental methods for determining this characteristic are presented in [4, 5].
Fig. 1 shows the dependence of the modulus and argument of equivalent magnetic permeability on the
first harmonic on the current value of the amplitude of first har-
monic of the magnetic field intensity for the steel of Cт.3 mark
and for the frequently used constructional steel of 09G2C mark.
The method of surface losses, as mentioned above, deter-
mines the specific surface losses w in ferromagnetic half-space
as through the tangential components of the electric and mag-
netic fields intensity according to (7), (8) and through the known
[2, 4−6] determination of the conventional surface impedance
( )
through the field vectors on the first harmonic Z H1 = E1 / H 1
and the square of the magnetic field intensity
w=
1
2
( ) 1
Re E 1 H 1* = Re Z H 1 .
2
2
(11)
Fig. 1
Using the equivalent magnetic permeability, the surface
losses p S are expressed through the tangential component of the magnetic field intensity H τ on the first
harmonic and surface impedance Z [4, 5]
1
pS = Re Z ( H τ ) H τ2 , Re Z = 2 ωμ 0 μ ′э ( H ) σ −1 cos ( π / 4 − 0,5arg μ э ( H ) ) . (12)
2
To calculate specific surface losses, the values of modulus and argument of equivalent magnetic perme-
ability for the steel Ct. 3 and steel 09G2S provided in Fig. 1, can be used. The works [4, 5] also present the
experimental dependence of surface losses on magnetic field intensity for steels Cт.2, Cт.3. The obtained
values by (12) correspond to the conditions of massive conductive bodies. The shape and dimensions of the
experimental samples have a negligible effect [4, 5].
The analytical method of the surface losses is used to calculate the field losses and losses in the tanks of
transformers and electrical reactors [3, 4], as a closed surface of the tank to the maximum extent corresponds
to the condition of the flat conductive half-space. Following the [4] method, on the first stage (using meas-
urements, and calculations) the primary magnetic field H τ0 in the air along the contour coinciding with the
tank surface (without tank) is determined. In the second stage the field is calculated directly on the tank sur-
face using an influence coefficient of impedance type H τ = H τ0 kτ ( Z ) , including taking into account the
local placement of the magnetic screens on the tank surface [3]. At the same time the determining surface
impedance and losses (12) nonlinear dependence of magnetic permeability on the magnetic field intensity of
tank steel is installed using the method of successive refinements.
By numerical simulation in ANSYS, a computing model of transformer (reactor) with tank is formed. A
ferromagnetic constructional steel of the tank is characterized by basic magnetization curve and electrical
conductivity. By the method of nonlinear harmonic analysis, the tangential component of the magnetic field
intensity directly on the tank surface H τ is calculated. Developed by means of ANSYS additional procedure
implements the algorithm (12) for determining the local losses. The total losses in the tank are calculated by
integrating the local losses of its surface.
The method of volumetric losses according to (6) determines the losses as the total of ohmic losses on
the eddy currents and losses on alternating magnetization of constructional steel with a nonlinear magnetic
characteristic and with hysteresis [2, 4, 5, 9–11]
pv = p j + p h . (13)
Power density of electrical losses is pj = j ⋅ E . (14)

22 ISSN 1607-7970. Техн. електродинаміка. 2014. № 3

 The losses p h (Bm ) spent on alternating magnetization with a frequency f in the body with the density
ρ are proportional to the area of hysteresis loop S(Bm ) [2, 4 – 6]
ph ( Bm ) = f ρ−1S ( Bm ) . (15)
The area of hysteresis loop S(Bm ) is defined as in (10) on the amplitude of the magnetic field induction Bm
+ Bm
S ( Bm ) = ∫ ⎡⎣ H D ( B,Bm ) − H A ( B,Bm ) ⎤⎦ dB . (16)
− Bm
In particular, for steel Art.3 the family of static hysteresis
loops for a range of values Bm is graphically represented in [4].
With their use by the expressions (15) and (16) at a frequency of
alternating magnetization 50 Hz, the values of losses of alternat-
ing magnetization shown in Fig. 2 by points are obtained.
There's also a solid line, which represents almost identical losses
values for steel with the A3 designation from the work [11].
The use of the above-mentioned algorithms for calculating
the losses by means of ANSYS requires the performance of a
series of actions. The calculation is performed by using a
nonlinear harmonic analysis. The time dependence of the har-
Fig. 2 monic currents changes of the specified sources (windings,
bends) is given. Instantaneous values of losses in ferromagnetic bodies are calculated by integrating on the
total volume (13) of specific eddy and hysteresis losses, and in those construction parts which are of non-
magnetic steel, the losses are determined only on account of eddy currents. When reaching a steady-state
process, the non-linear functions of losses are averaged over a period of a given harmonic change of current.
These procedures are implemented by specially designed macros for ANSYS.
The testing of the developed procedures was performed on a family of TEAM21 problems [9 – 11].
As is known, for TEAM21 problems the experimental installation of transformer type, containing two identi-
cal windings of rectangular cross-section with oppositely directed currents and frequency supply of 50 Hz, is
used. At the outer surface of the windings there are plates of nonmagnetic (N) and/or ferromagnetic (M)
steels, including those with cross-sectional cut or direct-axis cuts.
The calculations results of the mentioned test models TEAM21 with the use of the developed procedures are
given in the Appendix of this paper. The obtained results confirm the accuracy of the developed practical procedures
of simulation of electromagnetic fields and losses in the nonmagnetic and ferromagnetic bodies in the ANSYS, and
the possibility of their application for the relevant calculations of transformer equipment.
Heating calculation. The basis for calculating the heating of the constructional elements of transform-
ers and reactors are the values of volume losses. Thermal processes in the power transformer equipment with
oil cooling are characterized by complex heat-exchange conditions of constructional elements with internal
oil coolant and ambient air. The processes of losses and heating resolution are paired into the force of inter-
dependence of thermal and electrical parameters. However, as previously mentioned, for the practice of en-
gineering calculations, the execution of sequential calculations with iterative refinement of necessary pa-
rameters such as electromagnetic and thermal problems is valid.
With a known bulk density of losses Q in order to determine the body temperature θ with a thermal
conductivity λ and heat-exchange coefficient α on the surface it is convenient to use the available proce-
dures FEA of the thermal problem modeling in the form of Poisson's equation and with the boundary condi-
tions on the surface with normal n
div(λgradθ ) = −Q , − λ ∂θ / ∂n = α(θ − θ0 ) , (17)
where θ0 − ambient temperature. If θ0 = 0 , then the problem determines the increase of body temperature
above the external medium.
In the practical calculations the heat-exchange coefficients α included in the boundary conditions (17)
in the oil and air medium are determined by engineering dependencies of the form α = mq n [1]. The sur-

ISSN 1607-7970. Техн. електродинаміка. 2014. № 3 23

 face density of the losses q can be both balanced, and/or on sites.
The coefficients m and n depend on the body position (vertical,
horizontal), type of cooling, parameters of the radiation and convec-
tion in the air [1].
Example of transformer calculation. Fig. 3 shows the comput-
ing model of single-phase autotype transformer with a tank capacity of
500 MVA, and Fig. 4 shows its active part. Three-dimensionality and
finiteness of the sizes of MS, windings, constructional elements, mag-
netic shielding system are taking into account.
For simplicity, MS is represented as a homogeneous body with-
out special features simulation in the parts of burden revision of
Fig. 3 Fig. 4 plates of electrotechnical steel. The rods and yokes are characterized
by basic magnetization curve of electrotechnical steel. Along the normal to the flat surface of plates MS the
magnetic permeability is determined by with consideration of the filling coefficient of steel, and is a constant.
The windings are represented with the actual distribution of magnetizing forces along their height in the
form of zones of equal density of ampere turns. In the justified cases, equilibrium distribution of magnetizing
forces is defined in the windings.
Upper and lower yoke beams have horizontal shelves of non-magnetic steel, and walls of ferromagnetic
steel. The lower shelf of the upper beam has protrusions for winding pressing. The lower beams have vertical
sections between the rod and side yokes.
Longitudinal and transverse shunts on the lower yoke beams are made of electrotechnical steel package.
On the tank surface vertical magnetic shunts are mounted (see Fig. 3). Magnetic and electrical properties of
computational models of shunts are described by magnetization curve of steel and its electrical conductivity.
When carrying out the calculation the symmetry property of construction relative to the vertical planes
along the longitudi-
nal and transverse
axes of MS is used.
The calculation is
performed according
to the method of
Fig. 5 Fig. 6 volumetric losses.
The pattern of
distribution of losses and temperature rises of parts of up-
per yoke beams above the ambient oil are shown in Fig. 5
and Fig. 6. The existence of notch in the protrusion of the
lower shelf ensures the concentration limitations of eddy
currents and losses in it..
Distribution of volumetric losses in the surface layer
Fig. 7 of lower yoke beams is shown in Fig. 7.
Lower loss rate can be seen on the part of lower yoke
beam under the side yoke. Major losses are released in the beam under the rod. The presence of the vertical
section of beam promotes the losses reduction.
The distribution of volumetric losses in the sur-
face layer of tank without shunts is shown in Fig. 8.
As shown in Fig. 9, the degree of local and total
losses in the tank decreases at the presence of
shunts. The obtained heat patterns of tank surface
temperatures at the heat tests of transformer showed
a close match of the results of calculations and
measurements.
Calculations also determined that the calculated
elementwise total additional losses in the construc-
tion of this transformer are close to the measured
values.
Fig. 8 Fig. 9

24 ISSN 1607-7970. Техн. електродинаміка. 2014. № 3

 Example of reactor calculation. The structures of electrical shunt reactors of ROM type with capacity
of 80, 110 and 120 MVAr with gaps in the core of MS are studied by means of developed methods. As an
example, Fig. 10 shows the calculating model of the 80 MVAr reactor. In the actual structure of reactor on
the longitudinal walls of the tank the vertical shunts are mounted. To compare calculations results by means
of the method of surface and bulk losses, the calculations are carried out for the model without magnetic
shunts. The distribution of surface losses is shown in Fig. 11. Fig. 12 shows the distribution of volumetric
losses in the surface layer of the tank walls. The calculations are qualitative and quantitative close to each
other according to both methods.
For the actual structure, the total losses are de-
termined in MS, in the tie rods of MS cores, in
windings, in yoke beams, in the beams of given re-
actors, which showed close fit between the results of
calculations and measurements.
Conclusion. The practical procedures for
calculating the parameters of electromagnetic field,
losses in the limited details of construction of com-
putational models of transformer equipment of
nonmagnetic and ferromagnetic steel, are developed
by means of ANSYS software tools. The methods
and software procedures verified on test experimen-
Fig. 10 Fig. 11 Fig. 12 tal and computation data of the known family of
TEAM21 problems are developed. The examples of
practical calculations of transformers and reactors show that the developed procedures provide necessary
accuracy of calculation of the local and total losses in the constructional elements, their heating, and are also
applicable to elementwise calculation of total additional losses in construction.
Appendix. Verification of numerical procedures in TEAM21problems.
This article presents numerical models for continuous non-magnetic plate (P21a-0), non-magnetic plates
with one, two and three longitudinal cuts (P21a-1, P21a-2, P21a-3), continuous ferromagnetic plate (P21-B),
two plates of nonmagnetic and ferromagnetic steels (P21b-2N, P21b-2M), a combination of three welded
plates: nonmagnetic-magnetic-nonmagnetic (P21b-NMN) and welded plates in a different order (P21b-
MNM). These
given models des-
ignation correspond
to the models taken
from works [9–11].
Fig. A.1 a) −
e) shows the ge-
ometry of the basic
computational
models.
a) P21a-2 b) P21a-3 c) P21b-2N, P21b-2M d) P21-B e) P21b-NMN, P21b-MNM
Calculation of
Fig. A.1 the magnetic field,
the bulk eddy
losses in the mod-
els containing only
non-magnetic plate
is formed in AN-
SYS using the
method of linear
harmonic analysis.
The distribution of
volumetric losses
max 0.18.105 max 0.18.105 max 0.14.105 max 0.87.104 max 0.10.105 (W/m3) in non-
a) P21a-0 b) P21a-1 c) P21a-2 d) P21a-3 e) P21b-2N magnetic plates of
Fig. A.2 different configura-

ISSN 1607-7970. Техн. електродинаміка. 2014. № 3 25

ferent configuration is shown in Fig. A.2 a) - e). Under the figures, the maximum values of losses given to
compare the results of models calculation instead of the usual scale in ANSYS are given.

max 0.62.105 max 0.14.106 max 0.63.105

max 0.23.106
a) P21-B b) P21-B c) P21b-2M d) P21b-2M
front surface back surface front surface back surface
Fig. A.3
In particular, the comparison of Fig. A.2 a) - d) shows a known fact of a significant reduction of losses in
non-magnetic plates (for example, on lifting plates of MS cores of transformers) by performing longitudinal cuts.
The calculation of magnetic field, volumetric eddy and hysteresis losses in the models with magnetic
plates is performed in ANSYS using nonlinear harmonic analysis and using practical procedures of calcula-
tion of volumetric eddy losses and hysteresis losses represented in the main part of this work. The distribu-
tion of volumetric losses (W/m3) in the surface layer for the magnetic plates is shown in Fig. A.3 a) - d). The
major part of the distributed losses is centered on the front surface facing to the windings, and on the back
surface of plates there is a current "wicking" on the small area at the edge of plates.
Calculation of the combined models of plates with magnetic and non-magnetic elements is also per-
formed by using a nonlinear harmonic analysis ANSYS. The distribution of volumetric losses on the front
and back surfaces of the plates is shown in Fig. A.4 a) - d).

max 0.17.106 max 0.37.105 max 0.18.106 max 0.16.105

b
a) P21-MNM b) P21-MNM c) P21 -NMN d) P21b-NMN
front surface back surface front surface back surface
Fig. A.4

Eddy current calculated losses ( Pe ), hysteresis ( Ph ), total PΣ , W Measured total Calculation error,
Model Fig. losses [9−11], W %
Nonmagnetic plate Ferromagnetic plate Total losses
Pe Pe Ph PΣ , Pизм ε
a
P21 -0 P.2 9.39 - - 9.39 9.17 2.4
P21a-1 P.2 3.18 - - 3.18 3.40 -6.5
P21a-2 P.2 1.67 - - 1.67 1.68 -0.6
P21a-3 P.2 1.04 - - 1.04 1.25 -16.8
P21b-2N P.2 1.34 - - 1.34 1.38 -2.9
P21b-2M P.3 - 6.69 2.16 8.85 9.34 -5.2
P21-B P.3 - 8.58 2.74 11.32 11.97 -5.4
P21b-MNM P.4 4.62 3.16 1.07 8.85 10.53 -16.0
P21b-NMN P.4 1.90 3.47 1.10 6.47 7.44 -13.0

26 ISSN 1607-7970. Техн. електродинаміка. 2014. № 3

 On the front surfaces of magnetic plates, the surface densities of volumetric losses are much greater than
the density of losses in the non-magnetic plates. On the back surfaces of welded plates, the surface densities
of losses in the plates of non-magnetic and magnetic steel are comparable with each other. The mentioned
features must be considered, for example, when calculating the losses in composite yoke beams, the parts of
which are welded together from the parts of different types of steel.
The comparison of the total measured and calculated values for the TEAM 21 problems is shown in Table.

1. Biki M.А. Design of power transformers. Calculation of basic parameters. – Moskva: Znak, 2013. – 612 p. (Rus)
2. Vainshtein L.A. Electromagnetic waves. – Мoskva: Sovetskoe Radio, 1957. – 581 p. (Rus)
3. Ivankov V.F. Calculation of the magnetic field, losses in the tanks of transformers and electric reactors // Pratsi Insty-
tutu elektrodynamiky Natsionalnoi Akademii Nauk Ukrainy. – 2005. – № 1 (10). – Pp. 91–104. (Rus)
4. Kravchenko А.N., Nizhnik L.P. Electromagnetic calculations in energetics. – Kyiv: Tekhnika, 1977. – 184 p. (Rus)
5. Leites L.V. Electromagnetic calculations of transformers and reactors. – Moskva: Energiia, 1981. – 392 p. (Rus)
6. Neiman L.R. Surface effect in ferromagnetic bodies. − Leningrad-Moskva: Gosenergoizdat, 1949. – 190 p. (Rus)
7. ANSYS software // http://www.ansys.com.
8. Basova A.V., Ivankov V.F., Khimiuk I.V. Losses, heating in tank covers of transformers // Tekhnichna elektrody-
namika. – 2013. – № 4. – С. 74–80.
9. Cheng Z., Takahashi N., Yang S., Fan C., Guo M., Liu L., Zhang J., Gao S. Eddy current and loss analysis of multi-
steel configuration and validation // IEEE Trans. on Magn. – 2007. – Vol. 43. – No. 4. – Pp. 1737–1740.
10. Yang S., Cheng Z., Yang Q., Zhu X. The Stray loss on magnetic shielding in power transformers // Proc. of the 6th
WSEAS International Conference on Applied Computer Science, Hangzhou, China. – 2007. – Pp. 420–425.
11. Cheng Z., Takahashi N., Forghani B. TEAM Problem 21 Family (V.2009), approved by the International Compu-
mag Society Board at Compumag-2009, Florianópolis, Brazil, http://www.compumag.org/jsite/team.
12. Infolytica software // http://www.infolytica.com.
13. Kulkarni S.V., Khaparde S.A. Transformer Engineering. Design and Practice. Marcel Dekker, Inc., New York
(USA) – Basel (Switcerland). – 2004. – 477 p.
14. Ostrenko M.V., Tarchutkin L.O., Andienko B.Yu. Power Transformers and Reactors Stay Losses and Temperatures Calculation
Using Coupled IEM and FEM Technique // Proceedings 2010 IEEE Region 8 International Conference on Computational
Technologies in Electrical and Electronics Engineering SIBIRCON-2010. – Juli 11-15, 2010. – Vol. II. – Pp. 717–721.
15. Robert M. Del Vecchio, Bertrand Poulin, Pierre T. Feghali, Dilipkumar M. Shah and Rajendra Ahuja. Transformer
design principles: with application to core-form transformers // Gordon and Breach Science Publishers. – 2001. – 599 p.
16. Turowski J. Obliczenia elektromagnetyczne elementow maszyn i urzadzen elektrycznych // Wydawnictwa
Naukowo–Techniczne, Warszawa. – 1982. – 321 s.

УДК 621.314
ЧИСЕЛЬНЕ МОДЕЛЮВАННЯ ВТРАТ І НАГРІВІВ В ЕЛЕМЕНТАХ КОНСТРУКЦІЇ
ТРАНСФОРМАТОРІВ З ФЕРОМАГНІТНОЇ СТАЛІ
В.Ф.Іванков1, A.В.Басова1, І.В.Хімюк2
1
- ПАТ «Запоріжтрансформатор», Дніпропетровське шосе, 3, Запоріжжя, 69600, Україна,
2
- Інститут електродинамики НАН України, пр. Перемоги, 56, Київ-57, 03680, Україна.
e-mail: [email protected]
В роботі розглянуто застосування методів розрахунку поверхневих і об’ємних втрат в елементах конструкції транс-
форматорів з феромагнітної сталі. Подано короткий опис розроблених практичних процедур їхнього розрахунку із засто-
суванням програмного забезпечення ANSYS. Наведено результати тестування запропонованих процедур на відомих експе-
риментальних дослідженнях ТЕАМ21, а також приклади розрахунку втрат і нагрівів в елементах конструкції однофаз-
ного автотрансформатора потужністю 500 МВА і реактора типу РОМ потужністю 80 МВАр. Бібл. 16, табл. 1, рис. 16.
Ключові слова: потужний трансформатор, втрати, нагрів, тривимірне моделювання.

ЧИСЛЕННОЕ МОДЕЛИРОВАНИЕ ПОТЕРЬ И НАГРЕВОВ В ЭЛЕМЕНТАХ КОНСТРУКЦИИ

ТРАНСФОРМАТОРОВ ИЗ ФЕРРОМАГНИТНОЙ СТАЛИ
В.Ф.Иванков1, A.В.Басова1, И.В.Химюк2
1
- ПАТ «Запорожтрансформатор», Днепропетровское шоссе, 3, Запорожье, 69600, Украина,
2
- Институт электродинамики НАН Украины, пр. Победы, 56, Киев-57, 03680, Украина.
e-mail: [email protected]
В работе рассмотрено применение методов расчета поверхностных и объемных потерь в элементах конструкции транс-
форматоров из ферромагнитной стали. Дано краткое описание разработанных практических процедур их расчета с приме-
нением программного обеспечения ANSYS. Приведены результаты тестирования предложенных процедур на известных экс-
периментальных исследованиях ТЕАМ21, а также примеры расчета потерь и нагревов в элементах конструкции однофазного
автотрансформатора мощностью 500 МВА и реактора типа РОМ мощностью 80 МВАр. Библ. 16, табл. 1, рис. 16.
Ключевые слова: мощный трансформатор, потери, нагрев, трехмерное моделирование.
Надійшла 15.01.2014

ISSN 1607-7970. Техн. електродинаміка. 2014. № 3 27