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

net/publication/296685178

Mathematical modeling of transformers based on the frequency response of

magnetizing parameters

Article  in  Ciencia and Engenharia/ Science and Engineering Journal · March 2016

DOI: 10.14393/19834071.2015.30311

CITATIONS READS

0 4,088

7 authors, including:

M. Florentino Tarso Vilela Ferreira

Universidade Federal de Campina Grande (UFCG) Universidade Federal de Sergipe
6 PUBLICATIONS   3 CITATIONS    50 PUBLICATIONS   200 CITATIONS   

SEE PROFILE SEE PROFILE

Edson G Costa Benedito Luciano

Universidade Federal de Campina Grande (UFCG) Universidade Federal de Campina Grande (UFCG)
179 PUBLICATIONS   913 CITATIONS    57 PUBLICATIONS   236 CITATIONS   

SEE PROFILE SEE PROFILE

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

Use of a real 345 kV transmission line as a pilot line for experimentation and development of new techniques to improve performance against atmospheric discharges -
experimental development View project

Sistema de Monitoramento Online de Para-raios de Óxi-do de Zinco Baseado na Medição da Corrente de Fuga View project

All content following this page was uploaded by Edson G Costa on 26 September 2016.

The user has requested enhancement of the downloaded file.

 Ciência & Engenharia (Science & Engineering Journal) ISSN 1983-4071
24 (2): 59 – 64, jul. – dez. 2015

MATHEMATICAL MODELING OF TRANSFORMERS BASED ON THE

FREQUENCY RESPONSE OF MAGNETIZING PARAMETERS
MODELAGEM MATEMÁTICA DE TRANSFORMADORES BASEADA NA RESPOSTA EM FREQUÊNCIA
DOS PARÂMETROS DE MAGNETIZAÇÃO

Jalberth Fernandes de Araújo1, Marcus Tulios Barros Florentino2, Tarso Vilela Ferreira3, Edson Guedes
da Costa4, Benedito Antonio Luciano5, George Rossany Soares de Lira6, Rafael Mendonça Rocha
Barros7
1
Mestre em Engenharia Elétrica, Universidade Federal de Campina Grande, PB, Faculdade Maurício de
Nassau, Campina Grande, PB. E-mail: jalberth.araujo@ee.ufcg.edu.br
2
Engenheiro Elétrico, Universidade Federal de Campina Grande, PB
E-mail: marcus.florentino@ee.ufcg.edu.br
3,5,6
Doutor em Engenharia Elétrica, Professor Adjunto na Universidade Federal de Campina Grande, PB
3
E-mail: tarso@dee.ufcg.edu.br
5
E-mail: benedito@dee.ufcg.edu.br
6
E-mail: george@dee.ufcg.edu.br
4
Doutor em Engenharia Elétrica, Professor Titular na Universidade Federal de Campina Grande, PB
E-mail: edson@dee.ufcg.edu.br
7
Graduando do Curso de Engenharia Elétrica na Universidade Federal de Campina Grande, PB
E-mail: rafael.barros@ee.ufcg.edu.br

ABSTRACT
A mathematical modeling procedure for single phase low power transformers is presented in this paper. It is based on the
frequency response of the shunt parameters, taken from the electric model. The open circuit test was performed for different
frequencies in four transformers, and the magnetizing parameter frequency responses were determined. It was verified that
the sum of exponential functions can be employed to represent the behavior of the magnetizing parameters as a function of
frequency, with notable accuracy. The results contribute to the improvement of mathematical models and circuit equivalents
of single phase transformers.
Keywords: Frequency response, magnetizing parameters, mathematical modeling, transformers.

RESUMO
Um procedimento de modelagem matemática para transformadores monofásicos de baixa potência é apresentado neste
artigo. Este procedimento é baseado na resposta em frequência dos parâmetros de derivação do modelo elétrico. O teste de
circuito aberto foi realizado para diferentes frequências em quatro transformadores, e a resposta em frequência dos
parâmetros de magnetização foi determinada. Foi verificado que a soma de funções exponenciais pode ser empregada para
representar o comportamento dos parâmetros de magnetização como uma função da frequência, com boa acurácia. Os
resultados contribuem para o melhoramento de modelos matemáticos e circuitos equivalentes de transformadores
monofásicos.
Palavras-chave: Resposta em frequência, parâmetros de magnetização, modelagem matemática, transformadores.

1 – INTRODUCTION based on physical principles and mathematical modeling

based on the classical theory of electrical circuits can be
The transformer is one of the most important equipment of employed. These methods can also be used to estimate and
electrical systems. Its working principle is based on to predict alterations in operational characteristics of a
electromagnetic induction due to the coupling between two transformer.
or more coils though the closed core, which is made of Techniques related to transformer modeling are
ferromagnetic material. The main function of the employed in several researches and their applications are
transformer is to transfer energy from the input circuit to diverse, such as: transformer design, in order to operate at
output circuit, increasing or decreasing the values of different frequencies, power quality, improvement of
voltage and current, keeping the frequency constant. Like mathematical models, simulation and prediction
any real equipment, transformers have losses arising from characteristics, as cited in the next paragraphs.
the electrical resistance of windings, the dispersion of Roshen (1991) studied the influence of the frequency
magnetic flux and the cyclical process of core on the modeling and design of transformers, since it affects
magnetization. the losses, the parameters, and the indexes of power quality
With that in mind, in order to analyze the transformer equipment. The author presented a method for core loss
and determine its model, methods of equivalent circuits calculation at high frequencies. It is observed, that the

59
 Jalberth Fernandes de Araújo, Marcus Tulios Barros Florentino, Tarso Vilela Ferreira, Edson Guedes da
Costa, Benedito Antonio Luciano, George Rossany Soares de Lira, Rafael Mendonça Rocha Barros

losses in magnetic circuits are important design In this work, the open circuit test was performed in
parameters, and in many high frequency models the core the transformers, applying a sinusoidal voltage with
magnetization is limited by their losses. variable frequency between 60 and 2,580 Hz. As the
Similarly, Petkov (1995) provided a model for frequency was swept, the voltage and excitation current
transformer design considering the influence of frequency, signals were measured and recorded. The voltages
which is considered the major contributor to the reduction (primary and secondary) and the current were the routine
of volume and mass of switching power sources. inputs for the magnetizing parameter determination in
Additionally, Duffy and Hurley (1995) performed studies different frequencies. Once the magnetizing parameter
of transformer operation at different frequencies with frequency responses were found, the analytical expressions
respect to the skin and proximity effects. These effects were determined using the Recursive Least Squares (RLS),
cause increase of losses in the conductors and changes on and its adherence to the experimental data was evaluated
the values of resistance and inductance for each frequency. by the coefficient of determination (R²).
The influence of the frequency can also be used as a
transformer diagnostic technique, such as in Abu-Siada 2 ‒ THEORETICAL FRAMEWORK
and Islam (2007), where frequency response analysis was
used to detect deformation in the windings. The theoretical foundation on which the method is based is
The electrical parameters of transformers are also the presented here and in the next sub-sections. Thus, the
focus of several studies, and are important for design information about the open circuit test, the recursive least
modeling, monitoring and simulation. Other studies focus squares method, the functions used for representing the
on the determination of the transformer electrical magnetizing parameters as a function of frequency and the
parameters as a function of frequency, which has a direct transformer modeling by equivalent circuits for different
impact on power quality and transformer operation. In excitation frequencies are discussed.
Meister and Oliveira (2009) the open circuit test was
performed at different excitation frequencies. The 2.1 Open circuit test
objective was to make a curve adjustment to determine
mathematical expressions that represent the magnetizing The open circuit test is performed to determine the
parameters as a function of frequency. They showed that a transformer excitation current and the magnetizing
third degree polynomial can be adjusted by the parameters, applying the rated voltage and frequency on
magnetizing reactance, and a polynomial of second degree the secondary side, while the primary side remains open.
represents the core loss resistance. The voltage must be sinusoidal (IEEE STD., 1995).
In Mitchell and Welsh (2013) a methodology for Sometimes the voltage contains harmonics, thus, a
obtaining initial estimates and constraints for several source that provides a less distorted voltage ensures a more
fundamental transformer parameters was verified. The reliable measurement. In this paper, sinusoidal voltages
paper aimed to show the importance of an accurate were applied on the transformers with different
transformer model for use in the study of power fundamental frequencies. The frequency range applied was
transformers, since the manufacturer’s proprietary between 60 Hz and 2,580 Hz.
restrictions will generally make access to the detailed
design of a transformer difficult. 2.2 Recursive Least Squares
Moreover, the exact real time identification of
transformer winding parameters is favorable for a new The recursive least squares is a technique that estimates,
transformer protection and winding deformation detection. iteratively, unknown parameters. The method monitors the
In Wang (2012), a new model for the calculation of estimation of parameters, reaching the expected value
transformer winding parameters was presented using the before the stopping criterion, which results in reduced
least squares method. The method was not affected by the computational effort (ASTROM and WITTENMARK,
operating conditions and the difference among the three 1990).
phase winding parameters, and had fast convergence and In (1), (2) and (3) a set of expressions which may be
high precision and stability. used to implement the algorithm are presented.
As shown in the last citations, modeling the influence
of frequency and magnetizing parameters is an issue α(k + 1) = α(k ) + K (k ) ⋅ [ yk +1 − ϕT (k + 1) ⋅ α(k )] , (1)
associated to the design, diagnosis, behavior analysis and
simulation of the equipment. K (k ) = P(k ) ⋅ ϕ(k + 1) ⋅ [λ + ϕT (k + 1) ⋅ P(k) ⋅ ϕ(k + 1)]−1 , (2)
Hence, this study was performed with the objective of
contributing to the development of mathematical and
[I − K (k ) ⋅ ϕT (k +1)]⋅ P(k ) ,
circuit equivalent models in low power single phase P(k +1) = (3)
λ
transformers, by analyzing the magnetizing parameter
frequency responses. The proposed mathematical models
Where: α is the unknown parameters vector; K is the gain
are analytical expressions that represent the magnetizing
parameters as a function of the frequency. The method for matrix; P is proportional to the covariance matrix of α ; y
designing the models is based on an algorithm that is represents measured values; I is the identity matrix; λ is
efficient and has low computational cost.

60 Ciência & Engenharia, v. 24, n. 2, p. 59 – 64, jul. – dez. 2015

 Mathematical modeling of transformers based on the frequency response of magnetizing parameters

the forgetting factor; and ϕ corresponds to the represents the active core loss; Xm is the magnetization
approximation function that represents the model. reactance; N is the transformation ratio; V1 and V2 are the
The forgetting factor used in this paper was 1, which primary and secondary voltages, respectively; Ie is the
represents the classic recursive estimation, wherein the excitation current; Cs and Ce corresponds to the
actual estimative is determined by the previous estimative. capacitances between windings and between windings and
ground. The excitation current is, in concept, separated into
2.3 Modeling for determination of analytical expressions two components: a magnetizing current and a core loss
current.
In the modeling for the determination of the analytical
expressions, two types of functions were used, the Figure 1 – Transformer equivalent circuits (SLEMON, 1994)
polynomial and the sum of exponential functions. for: (a) low frequencies, (b) medium frequencies and (c) high
frequencies
Furthermore, to evaluate the quality of the analytical
expressions that represent the magnetizing parameters as a
function of frequency, the coefficient of determination (R²)
was used.
The coefficient of determination (CAMERON and
WINDMEIJER, 1988) can be calculated by (4).

SSE , (a)
R2 = 1− (4)
SST

Where: R² is the coefficient of determination; SSE is the

summation of the squared differences of the model data
and the experimental data; and SST is the summation of the
squared differences between the model data and the
experimental data mean.
(b)
Additionally the analytical expressions that represent
the magnetizing parameters as a function of frequency
were determined using the functions considered in (5), (6)
and (7).

Y ( x) = a ⋅ x3 + b ⋅ x2 + c ⋅ x + d , (5)
2
Y ( x) = a ⋅ x + b ⋅ x + c, (6)
b⋅ x d ⋅x (c)
Y ( x) = a ⋅ e + c⋅e , (7)
It is known that in low frequencies (50 to 200 Hz) the
nonlinear effect of the magnetization branch is more
Where: Y is a function of x and a, b, c and d are parameters significant. In medium frequencies (200 to 1,500 Hz) this
unknown, which can be estimated by the RLS. Y represents effect is reduced. In high frequencies (above 1,500 Hz), the
the magnetizing parameters and x corresponds to the capacitance effects should be take into account
frequency. (FALCONE, 1979).
These functions were chosen to compare the Consequently, the equivalent circuit modeling was
polynomial modeling to the exponential sum modeling. used to study the transformer physical phenomena and they
were sufficient to assist in the representation of
2.4 Equivalent Circuits for Different Frequencies transformer behavior used in this work.
To characterize a transformer in a wide range of 3 ‒ MATERAL AND METHODS
frequencies, it is necessary to write the representative
equations of circuit parameters as a function of frequency. In this section, the characteristics of the transformer and
If the behavior of the equivalent circuit parameters as a details of the performed tests are presented.
function of frequency is known, several simplified circuits
can be modeled, and each one is valid for a determined 3.1 Material
frequency range (SLEMON, 1994) and (FALCONE,
1979). The transformers employed in this paper are characterized
Three equivalent circuits of a transformer, in Table 1.
characterizing the behavior in different frequencies, are The source used to feed the transformers has the
shown in Figure 1 (WANG, LEE and ODENDAAL, following specifications: zero to 400 V, 40 to 5,000 Hz and
2003). 3 kVA of maximum power output.
In Figure 1, Rs is the equivalent winding resistance The experimental platform confectioned to measure
referred to the primary side; Xs is the leakage reactance the voltage and current signals is presented in Figure 2.
referred to the primary side; Rm is the resistance which

Ciência & Engenharia, v. 24, n. 2, p. 59 – 64, jul. – dez. 2015 61

 Jalberth Fernandes de Araújo, Marcus Tulios Barros Florentino, Tarso Vilela Ferreira, Edson Guedes da
Costa, Benedito Antonio Luciano, George Rossany Soares de Lira, Rafael Mendonça Rocha Barros

Table 1 – Transformers characteristics Figure 3 – Block diagram that represents the experimental and
Rated Voltage Ratio computational procedure
Tag Core material
Power (VA) (V/V)
T1 1,000 220/110 Amourphous
T2 1,000 220/110 FeSi
T3 100 220/110 FeSi
T4 60 127/12 FeSi

Figure 2 – Experimental platform, containing: (a) computer that

controls the source, (b) source, (c) transformer and (d)
oscilloscope

(d)

(a)

(b) (c) 3 ‒ RESULTS AND DISCUSSIONS

With the objective of contributing to the development of

There are seen in Figure 2: (a) computer that controls mathematical and circuit equivalent models in low power
the source; (b) a programmable frequency source; (c) the single phase transformers, the magnetizing parameters
transformer; and (d) a four-channel digital oscilloscope frequency response was determined for the four
(sampling rate of 1 GS/s). On the experimental platform transformers, according to Sections 2.1, 2.2 and 2.3. In
other components were used, such as: voltage probe to Figures 4 to 7 the frequency response of the resistances
measure the voltage, a current sensor with ratio 100 mV/A (Rm) are shown, which represent the active core loss, and in
to measure the excitation current and a contact matrix to Figures 8 to 11 the frequency response of the
make the connections between the components magnetization reactances (Xm) are shown. In Figures 4 to
11 the curves determined by RLS are also presented, which
3.2 Experimental and computational procedure correspond to the analytical expressions that represent the
magnetizing parameters as a function of frequency.
The open circuit test was performed in the transformers,
applying a sinusoidal voltage with variable frequency from Figure 4 – Frequency response of the resistance which represents
60 to 2,580 Hz. the active core loss and approximation curves for T1
As the frequency changed, the voltage and excitation 70
current signals were measured and recorded in the
60
oscilloscope. The recorded signals were subsequently
Resistance (kilo Ohms)

processed by a dedicated Matlab® routine. The voltage and 50 Experimental

the current were the routine inputs for the magnetizing Polynomial
Exponential
parameter determination in different frequencies. Once the 40
frequency response of the magnetizing parameters was
defined, the analytical expressions were determined using 30
RLS, and its fidelity to the experimental data was
20
evaluated by R².
It is shown in Figure 3 a block diagram that represents 10
the experimental and computational procedure. 0 500 1000 1500 2000 2500
Frequency (Hz)
Figure 5 – Frequency response of the resistance which represents
the active core loss and approximation curves for T2
6
Resistance (kilo Ohms)

5
Experimental
Polynomial
Exponential
4

2
500 1000 1500 2000 2500
Frequency (Hz)

62 Ciência & Engenharia, v. 24, n. 2, p. 59 – 64, jul. – dez. 2015

 Mathematical modeling of transformers based on the frequency response of magnetizing parameters

Figure 6 – Frequency response of the resistance which represents Figure 10 – Frequency response of the magnetization reactance
the core loss and approximation curves for T3 and approximation curves for T3
70 2500
Experimental
60 Polynomial
2000
Resistance (kilo Ohms)

Reactance (kilo Ohms)

Exponential
50 Experimental
Polynomial 1500
40 Exponential
1000
30

20 500

10 0
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500
Frequency (Hz) Frequency (Hz)

Figure 7 – Frequency response of the resistance which represents Figure 11 – Frequency response of the magnetization reactance
the active core loss and approximation curves for T4 and approximation curves for T4
4500 35
Experimental
4000 30 Polynomial
Exponential

Reactance (kilo Ohms)

Resistance (Ohm)

3500 25
Experimental
3000 Polynomial
Exponential 20
2500
15
2000
10
1500
5
1000
0 500 1000 1500 2000 2500 0
Frequency (Hz) 0 500 1000 1500 2000 2500
Frequency (Hz)

Figure 8 – Frequency response of the magnetization reactance

and approximation curves for T1 In Figures 4 to 11 it was shown that the magnetizing
parameters increase with the frequency elevation. Thus, the
Experimental equivalent circuit modeling presented in Figure 1 fits the
60 Polynomial
presented results, since tin the Figure 1 the magnetizing
Reactance (kilo Ohms)

Exponential
parameters tend to those of the open circuit, showing an
increase with the frequency. The resistance that represents
40
the active core loss (Rm) increases with the frequency
elevation because of the decrease of current flow area.
Moreover, the magnetization reactance (Xm) increases with
20
the frequency elevation because of the proportional ratio
between the reactance and frequency.
0
Wang (2012) shows that the use of cores with a
0 500 1000 1500 2000 2500 smaller size can increase the capacitive effect. In this
Frequency (Hz)
manner, in Figure 11 a decrease of the magnetization
Figure 9 – Frequency response of the magnetization reactance reactance is presented at 2,580 Hz for T4. It occurred
and approximation curves for T2 because the capacitance effect in this transformer was
18 enough to decrease the value of the magnetization
Experimental
16 reactance, since the size of T4 is the smallest compared to
Polynomial
Exponential the other transformers. Once the magnetizing parameter
Reactance (kilo Ohms)

14
frequency response curves were defined, the analytical
12 expressions could be determined. The RLS was used to
10 determine the polynomial and exponential sum functions,
8 as seen in Figures 4 to 11.
In Figure 11 it can be observed that in the frequency
6
range between 60 and 1,250 Hz, the frequency response of
4 the magnetization reactance is better represented by
2 exponential sum functions. In the frequency range between
0 500 1000 1500 2000 2500
Frequency (Hz) 1,250 and 2,580 Hz, the polynomial functions represent
better the behavior of the magnetization reactance as a
function of frequency.

Ciência & Engenharia, v. 24, n. 2, p. 59 – 64, jul. – dez. 2015 63

 Jalberth Fernandes de Araújo, Marcus Tulios Barros Florentino, Tarso Vilela Ferreira, Edson Guedes da
Costa, Benedito Antonio Luciano, George Rossany Soares de Lira, Rafael Mendonça Rocha Barros

In Table 2 the analytical expressions that represent the CAMERON, A. C.; WINDMEIJER, F. A. G. An R-
magnetizing parameters as a function of frequency are squared measure of goodness of fit for some common
shown. The polynomial functions, as proposed in Meister nonlinear regression models. Journal of Econometrics,
and Oliveira (2009), were used to represent the 1988.
magnetizing parameters as a function of frequency, and the DUFFY, M. C.; HURLEY, W. G. High frequency effects
results were compared to exponential sum functions. in planar magnetics: inclusive of skin effect. Proceedings
26th Annual IEEE Power Electronics Specialists
Table 2 – Analytical expressions that represents the magnetizing Conference, p. 645-650, 1995.
parameters as a function of frequency http://dx.doi.org/10.1109/pesc.1995.474887.
Analytical Expression R² FALCONE, A. G. Eletromecânica: transformadores e
R mT 1 ( f ) = − 0 , 0113 ⋅ f 2 + 42 , 39 ⋅ f + 21230 0.9480 transdutores, conversão eletromecânica de energia,
RmT1( f ) = 46530⋅ e0,0001⋅ f − 36580⋅ e−0,0036⋅ f 0.9956 máquinas elétricas. São Paulo: Edgar Blücher, 1979.
R mT 2 ( f ) = − 0 , 0014 ⋅ f 2 + 6 , 256 ⋅ f + 2806 0.9620 IEEE Standard test code for dy-type distribution and
RmT 2 ( f ) = 6913⋅ e0,0001⋅ f − 5191⋅ e−0,0026⋅ f 0.9880
R mT 3 ( f ) = − 0 , 0093 ⋅ f 2 + 37 , 57 ⋅ f + 24230
power transformers. Institute of Electrical and
0.8692
R mT 3 ( f ) = 42820 ⋅ e 0 , 0001 ⋅ f − 44380 ⋅ e − 0 , 0079 ⋅ f 0.9740
Electronics Engineers ‒ IEEE, 1995. (IEEE Standard
R mT 4 ( f ) = − 0 , 0005 ⋅ f 2 + 2 , 419 ⋅ f + 1500 0.9511 C57.12.91-2001).
RmT 4 ( f ) = 2818⋅ e0,0001⋅ f − 2038⋅ e−0,0041⋅ f 0.9950 http://dx.doi.org/10.1109/IEEESTD.2012.6152116.
X mT 1 ( f ) = ( 2 ,821 x10 − 5 ) ⋅ f 3 − 0 ,1105 ⋅ f 2 + 161 ,3 ⋅ f + 13860 0.9403 MEISTER, D.; OLIVEIRA, M. A. G. The use of the least
X mT 1 ( f ) = 57780 ⋅ e 0 , 0004 ⋅ f − 133300 ⋅ e − 0 , 0133 ⋅ f 0.9907 Squares method to estimate the model parameters of a
X mT 2 ( f ) = ( 5 , 422 x 10 − 7 ) ⋅ f 3 − 0 , 0019 ⋅ f 2 + 7 , 259 ⋅ f + 2827 0.9945 transformer. Proceedings 2009 IEEE Electrical Power
X mT 2 ( f ) = 5791 ⋅ e 0 , 0004 ⋅ f − 3262 ⋅ e − 0 , 0022 ⋅ f 0.9951 Quality and Utilisation Conference, p. 1-6, 2009.
X mT 3 ( f ) = 0 , 0004 ⋅ f 3 − 1, 2 ⋅ f 2 + 934 , 2 ⋅ f − 89840 0.9840 http://dx.doi.org/10.1109/epqu.2009.5318853.
X mT 3 ( f ) = 39060 ⋅ e 0 , 0011 ⋅ f + 3 , 494 ⋅ e 0 , 0049 ⋅ f 0.9993 MITCHELL, S. D.; WELSH, J. S. Initial parameter
X mT 4 ( f ) = ( − 1, 252 x10 − 5 ) ⋅ f 3 + 0 , 0428 ⋅ f 2 − 23 ,84 ⋅ f + 5871 0.9187 estimates and constraints to support gray box modeling of
X mT 4 ( f ) = ( − 1, 501 x 10 6 ) ⋅ e 0 , 0020 ⋅ f + (1 , 504 x 10 6 ) ⋅ e 0 , 0020 ⋅ f 0.9666 power transformers. IEEE Transactions on Power
Delivery, v. 28, n. 4, p. 2411-2418, 2013.
It is important to observe that the estimated http://dx.doi.org/10.1109/tpwrd.2013.2259266.
parameters on the analytical expressions do not have any PETKOV, R. Design issues of the high-power high
physical meaning. frequency transformer. Proceedings 1995 IEEE Power
Electronics and Drive Systems Conference, v. 1, p. 401-
CONCLUSIONS 410, 1995. http://dx.doi.org/10.1109/peds.1995.404888.
ROSHEN, W. Ferrite core loss for power magnetic
In this paper a mathematical a model based on the components design. IEEE Transactions on Magnetics, v.
frequency response of the magnetizing parameters of a 27, n. 6, p. 4407-4415, 1991.
transformer was presented. The open circuit test was http://dx.doi.org/10.1109/20.278656.
performed in transformers, applying sinusoidal voltages SLEMON, G. R. Equipamentos magnetelétricos:
with variable frequency. transdutores, transformadores e máquinas. São Paulo:
The analytical expressions that model the magnetizing Livros Técnicos e Científicos, 1994.
parameters as a function of frequency were determined WANG, S.; LEE, F. C.; ODENDAAL, W. G. Single layer
employing the least square method. The R² of the iron core inductor model and its effect on boost PFC EMI
exponential sum functions are very close to 1, showing that noise. Proceedings of 2003 IEEE 34th Annual Power
the exponential sum functions model more satisfactorily Electronics Specialist Conference, v. 2, p. 847-852, 2003.
the magnetizing parameters behavior, when compared with http://dx.doi.org/10.1109/PESC.2003.1218167.
the polynomial functions. WANG, X. Research on the transformer winding
The results contribute to improve the mathematical parameters real-time calculation method based on UD
models of transformers, since that is achieved by using the decomposition. Proceedings of 2012 IEEE International
frequency response of magnetizing parameters. Moreover, Conference on Power Electronics, Drives and Energy
the results allow characterizing the transformers in a wide Systems (PEDES), p. 1-5, 2012.
band of frequencies, since used analytical expressions are a http://dx.doi.org/10.1109/PEDES.2012.6484269.
function of frequency.

REFERENCES

ABU-SIADA, S.; ISLAM, S. High frequency transformer

computer modeling. Proceedings of Australasian
Universities IEEE Power Engineering Conference,
Perth, WA, p. 1-5, 2007.
http://dx.doi.org/10.1109/AUPEC.2007.4548049.
ASTROM, K. J.; WITTENMARK, B. Computer
Controlled Systems. A Division of Simon & Schuster-
Englewood Cliffs, 1990.

64 Ciência & Engenharia, v. 24, n. 2, p. 59 – 64, jul. – dez. 2015

View publication stats