This is the authors version of a work that was published in the IEEE Transactions on Power

Delivery journal in the following source:

Filipovic-Grcic, D.; Filipovic-Grcic, B.; Uglesic, I., "High-Frequency Model of the Power
Transformer Based on Frequency-Response Measurements," IEEE Transactions on Power
Delivery, vol. 30, no. 1, pp.34-42, Feb. 2015. DOI: 10.1109/TPWRD.2014.2327061

Notice: Changes introduced as a result of publishing processes such as copy-editing, formatting

and technical enhancement may not be reflected in this document. For a final version of this
work, please refer to the published source:

DOI: 10.1109/TPWRD.2014.2327061

URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6839055&isnumber=7017601
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON POWER DELIVERY 1

High-Frequency Model of the Power Transformer

Based on Frequency-Response Measurements
Dalibor Filipovi-Gri, Boidar Filipovi-Gri, Member, IEEE, and Ivo Uglei, Senior Member, IEEE

AbstractIn this paper, a new high-frequency black-box model described in [5][7]. Lumped and distributed parameter trans-
of the power transformer is presented for the calculation of former models have been used for the analysis of transferred
transferred lightning overvoltages. The model is based on fre- overvoltages [8][16]. It is shown that the lumped parameter
quency-response analysis measurements and the application of
four-terminal network theory. The computational verification of model can give adequate results for fast transients (up to 1
the presented model was successfully performed on a white-box MHz). However, for very fast transients (above 1 MHz), a
transformer model in Electromagnetic Transients Program-Re- distributed-parameter model provides better results.
structured Version software. The experimental verification was In [17][19] it is shown that a transformer can be modeled by
carried out on 130-kVA and 36-MVA three-phase power trans- considering turns as conductors of a multiconductor transmis-
formers. The presented calculation and measurement results
confirm the validity of the proposed model for full and chopped sion line while lightning overvoltages transferred through distri-
lightning impulse voltages. bution transformers are investigated in [20][22]. The analysis
of transferred overvoltages can be of importance at the design
Index TermsFast Fourier transform (FFT), four-terminal
network, frequency-response measurements, high-frequency
stage of winding insulation for the determination of dielectric
transformer model, transfer function, transferred lightning over- stresses and selection of the appropriate overvoltage protection.
voltages. Therefore, a high-frequency power transformer model for ac-
curate calculation of lightning overvoltages transferred through
the power transformer is preferred. The electromagnetic-field
I. INTRODUCTION approach [23], [24], based on the finite-element method (FEM),
is used to establish the elementary parameters of a lumped pa-
rameter transformer model. This approach requires construc-

L IGHTNING overvoltages propagating along transmis-

sion lines and entering substations are transferred from
the high-voltage (HV) winding of the power transformer to
tional details of the transformer and is, in general, complex and
highly case dependent. An important issue in the parameter de-
termination for high-frequency transformer models used for the
the low-voltage (LV) winding and vice-versa by inductive and calculation of internal overvoltages is that they require very de-
capacitive coupling. The capacitive transfer depends on the tailed information of the transformer geometrical configuration
surge capacitance and increases with the rate of rise of the [25] which is only available to manufacturers. However, if over-
applied overvoltage. The inductive transfer of the overvoltage voltages generated within the windings are not required and
depends on the flow of surge current in the HV winding and it only external overvoltages are of interest, parameters can be ob-
is less sensitive to the rate of rise of the applied voltage [1]. tained from transformer terminal measurements.
Transferred overvoltages can endanger winding insulation or For the calculation of external transferred overvoltages,
the insulation of equipment in a substation. Transferred light- black-box-type models are usually suitable since they do not
ning overvoltages may be particularly harmful for generator require information about the transformer geometry and be-
step-up transformers which have a large voltage ratio and trans- cause they can reproduce the transformer terminal behavior.
formers with an LV tertiary winding [2][4]. Therefore, a transient interaction between the transformer
So far, extensive research has been performed regarding and the network can be analyzed. A transformer model based
the investigation of high-frequency transients in power trans- on the measured admittance matrix of the transformer is de-
formers. High-frequency transformer models for the calcula- scribed in [26][29]. In this way, a general representation of
tion of internal overvoltages that occur inside windings are the transformer in a wide frequency range can be done by
applying vector fitting [30]. In [31], an accurate and simplified
Manuscript received October 31, 2013; revised February 09, 2014; accepted high-frequency model of the distribution transformer under
May 24, 2014. Paper no. TPWRD-01237-2013. lightning strokes is presented, which is based on experimental
D. Filipovi-Gri is with the KONAR Electrical Engineering Institute,
frequency-response analysis (FRA) measurements.
Zagreb 10000, Croatia (e-mail: [email protected]).
B. Filipovi-Gri and I. Uglei are with the University of Zagreb, Faculty of This paper presents a new high-frequency transformer model
Electrical Engineering and Computing, Zagreb 10000, Croatia (e-mail: bozidar. for the calculation of lightning overvoltages transferred through
[email protected]; [email protected]).
power transformers. The main advantage of the model is its sim-
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. plicity and the fact that it can be derived from widely used FRA
Digital Object Identifier 10.1109/TPWRD.2014.2327061 measurements.

0885-8977 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 IEEE TRANSACTIONS ON POWER DELIVERY

Fig. 1. Four-terminal network.

Fig. 3. FRA measurement between output terminals while input terminals are
short-circuited.

can be determined from the results of two FRA mea-

surements. In the first measurement shown in Fig. 2, frequency
Fig. 2. FRA measurement between input and output terminals. response is measured between the input terminal A to which
the voltage of variable frequency is applied and the output
terminal a grounded through the matching resistor .
II. HIGH-FREQUENCY TRANSFORMER MODEL The output current is calculated as

A. Model Description (3)

In order to calculate the transferred overvoltages, the transfer
function of the four-terminal network has to be determined. is derived from (1):
In the presented approach, the input voltage is expressed as a
sum of sine and cosine functions of varying frequency and the (4)
frequency-dependent network transfer function is formed from
the FRA results. The transferred overvoltage is calculated as a The ratio of output and input voltage for this case is
sum of products of input voltage and transfer function for each
frequency. FRA is widely used in the industry today, mainly (5)
to detect winding displacements after overcurrents and inter-
turn winding faults [32]. Therefore, it is highly convenient to
apply FRA results for high-frequency modeling of power trans- Ratio can be measured with frequency-response an-
formers. alyzers and the result is given in a form of magnitude and
phase angle
B. Determination of the Transfer Function From FRA Results
(6)
Commercially available frequency-response analyzers mea-
(7)
sure input voltage and output voltage with an output terminal
grounded through the matching resistor. The ratio of output
can be derived from (6) and (7) as
voltage and matching resistance gives output current, and
the ratio of input voltage and output current gives transfer
(8)
impedance. So the ransfer function cannot be determined from
a single measurement. However, it can be obtained from the (9)
results of two independent measurements. Fig. 1 shows a part
of a transformer represented by a four-terminal network. Input In the second measurement shown in Fig. 3, the FRA is mea-
voltage is applied to the HV winding and output voltage sured between the output terminals a-b while the input termi-
is transmitted to the LV winding. nals are short-circuited.
The relationship between input and output voltage is defined Expressions for the output voltage (10) and current (11) are
by the transference matrix as follows: derived from (1) for

(1) (10)

When measuring the transferred overvoltages, only the mea-

suring system is connected to output terminals. This approx- (11)
imately corresponds to the no-load condition and, therefore,
. Then, from (1), the voltage transfer function (2) is cal- The ratio of the output voltage and current gives the short-
culated circuit impedance

(2) (12)
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

FILIPOVI-GRI et al.: HIGH-FREQUENCY MODEL OF THE POWER TRANSFORMER 3

is the impedance between terminals a-b and can

also be expressed as

(13)
where is equal to ratio and can be de-
rived from FRA results in the same way as (8)-(9). Sub-
stituting in (5) with (12) gives

(14) Fig. 4. Recorded lightning waveform approximated by FFT up to 90 s and

extended with the base curve to its complete attenuation.
Transfer function is determined by (14)

(15)

Finally, substituting in (15) with (13) gives

(16)

C. Approximation of Input Lightning Impulse Voltage and

Calculation of Output Voltage
The input lightning impulse voltage waveform can be Fig. 5. Autotransformer representation for the lightning impulse test.
expressed as a sum of sine and cosine functions of varying fre-
quency
where is the amplitude and is the angle of transfer func-
tion th harmonic.
(17)

III. COMPARISON WITH EMTP-RV WHITE-BOX MODEL

where is the total number of harmonics considered;
is the fundamental frequency; and , are coefficients of The purpose of this section is to check the proposed approach
the Fourier series computed using the MATLAB built-in fast in controlled conditions. Prior to the experiment on a real trans-
Fourier transform (FFT) function. former, the calculation of the transferred overvoltage was car-
The lighting impulse voltage waveform is usually recorded ried out on a single-phase model of a three-phase autotrans-
up to around 100 s and during this period, voltage does not former 250 MVA, 400/231/22 kV. The model consists of se-
decay to zero value. If a waveform of such duration is expressed ries winding, parallel winding, regulation winding, and tertiary
by Fourier series, an error in approximation is introduced, and winding. The parameters shown in the Appendix for a simpli-
the frequency of the first harmonic is relatively high. As a con- fied Electromagnetic Transients Program-Restructured Version
sequence, errors are introduced in the calculation of transferred (EMTP-RV) model were provided by the manufacturer. The
overvoltages. Therefore, the recorded lightning impulse wave- transferred overvoltage across the regulation winding during the
form is approximated by the corresponding function and ex- lightning impulse test is analyzed. Voltage is applied on the ter-
tended until it decays nearly to zero value. tiary winding while the HV and medium-voltage (MV) termi-
Lightning impulse voltage is approximated from the last nals are grounded (Fig. 5).
recorded point to its almost complete attenuation with the base The autotransformer is represented by a four-terminal net-
curve whose construction is given in [33]. Fig. 4 shows an work ABCD according to the previously described procedure.
example of recorded waveform and its approximation. This The lightning impulse voltage 1.2/50 s, a double exponen-
approach gives a very accurate approximation of input voltage tial function in EMTP-RV, is applied between input terminals
even for waveforms with strong superimposed oscillations or A-B, and the transferred voltage is calculated at output ter-
chopped lightning-impulse voltages. minals C-D. Then, frequency response is calculated (using
After the impulse voltage is approximated and the the frequency scan option in EMTP-RV; 1 Hz,
transfer function is determined, the transferred over- 2 MHz, frequency scan step 1 Hz) between ter-
voltage can be calculated minals A-C, voltage is applied on A while C is grounded
through the 50- resistor. Afterwards, the response between ter-
minals C-D with A-B short-circuited is calculated, voltage
is applied on C, while D is grounded through the 50- re-
(18) sistor (Fig. 6).
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 IEEE TRANSACTIONS ON POWER DELIVERY

Fig. 6. Frequency response between terminals A-C and C-D. Fig. 10. Test setup for the measurement of transferred overvoltages.

Fig. 7. Comparison between the results of the proposed model and EMTP-RV.

Fig. 11. Test setup for the FRA measurement.

Fig. 8. Influence of on the calculated transferred overvoltage at terminals

C-D.

Fig. 12. Measured magnitude between terminals A-a, A-b, and A-c for
the Yd connection.

Fig. 9. Test circuit for the measurement of the transferred overvoltages in case
of the Yd connection.

Results of the EMTP-RV simulation and the proposed model

are given in Fig. 7 and show good agreement. EMTP simulation
results are represented by dashed curves. The lightning impulse
voltage is approximated by FFT (red curve), the transfer func-
tion is calculated from the results of frequency responses, and
Fig. 13. Measured phase angle between terminals A-a, A-b, and A-c for
the transferred overvoltage (black curve) is calculated by (18). the Yd connection.
The accuracy of the proposed model depends on and
upper limit frequency. For 2 MHz, satisfactory
approximation of lightning impulse voltage is achieved. In this Enlargement of results in larger deviations from the accu-
case, the calculated transferred overvoltage is very sensitive to rate result. Influence of is significant because the frequency
as shown in Fig. 8. response between terminals C-D changes rapidly at a very
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

FILIPOVI-GRI et al.: HIGH-FREQUENCY MODEL OF THE POWER TRANSFORMER 5

Fig. 14. Measured magnitude between terminals a-N, b-N, and c-N for Fig. 18. Transferred overvoltage on terminal b-N when a full lightning im-
the Yd connection. pulse voltage is applied to terminal A.

Fig. 15. Measured phase angle between terminals a-N, b-N, and c-N for Fig. 19. Transferred overvoltage on terminal c-N when a full lightning im-
the Yd connection. pulse voltage is applied to terminal A.

IV. EXPERIMENTAL VERIFICATIONMEASUREMENT AND

CALCULATION OF TRANSFERRED OVERVOLTAGES
The experimental verification was carried out on a
three-phase 130-kVA intermediate reconnectable power trans-
former. Measurements of FRA and transferred overvoltages
were performed for YNd1 and YNyn0 transformer connections.
Fig. 9 shows the test circuit arrangement for the measurement
of transferred overvoltages in the case of the Yd connection.
Full and chopped lightning impulse voltage with an amplitude
Fig. 16. Recorded full lightning impulse voltage approximated by FFT. of 300 V are applied on HV terminal A while transferred
overvoltages are measured between the LV terminals a, b,
c and neutral point N.
A recurrent surge generator, a low-voltage single-stage
equivalent of a high-voltage impulse generator, was used as
a lightning impulse voltage generator. The impulse voltages
were recorded using a digital oscilloscope (500 MHz, 1 GSam-
ples/s) connected to a PC. Fig. 10 shows the test setup for the
measurement of transferred overvoltages.
The transformer is observed as a four-terminal network with
input terminals A-N and output terminals a-N. In order
to obtain the transfer function between A-N and a-N, two
frequency responses have to be measured. The first response
Fig. 17. Transferred overvoltage on terminal a-N when a full lightning im- is measured between terminals A-a while terminal N
pulse voltage is applied to terminal A. and transformer tank are grounded the same way as when
measuring transferred overvoltages. The second response is
measured between a-N with A, N and the transformer
narrow frequency range around a resonant frequency of approx- tank short-circuited and grounded through a 50- matching
imately 21 kHz (Fig. 6). Nevertheless, the calculation results for resistor.
1 Hz confirm that the proposed method gives accurate re- Fig. 11 shows the test setup for the FRA measurement. FRA
sults for the simple transformer model shown in Fig. 5. measurements were performed within the frequency band 1 kHz
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 IEEE TRANSACTIONS ON POWER DELIVERY

Fig. 20. Recorded chopped lightning impulse voltage approximated by FFT.

Fig. 23. Transferred overvoltage on terminal c-N when chopped lightning

impulse voltage is applied to terminal A.

Fig. 24. Test circuit for the measurement of transferred overvoltages in case of
the Yy connection.

Fig. 21. Transferred overvoltage on terminal a-N when chopped lightning

impulse voltage is applied to terminal A.

Fig. 25. Transferred overvoltage on terminal a-n when full lightning impulse
voltage is applied to terminal A.

Fig. 22. Transferred overvoltage on terminal b-N when chopped lightning

impulse voltage is applied to terminal A.

to 2 MHz, 1 kHz, by the Doble SFRA M5200 instru-

ment. The accuracy guaranteed by the manufacturer for FRA
measurement is 1 dB down to 80 dB. Figs. 1215 show the
measured magnitude and phase angle of the frequency response
for the Yd connection.
Fig. 16 shows the recorded full lightning impulse waveform
Fig. 26. Transferred overvoltage on terminal b-n when full lightning impulse
applied on HV terminal A and its FFT approximation. voltage is applied to terminal A.
Although not shown in Fig. 16, FFT approximation lasts up
to 1 ms to match the FRA fundamental harmonic frequency of
1 kHz. Calculated transferred overvoltages are expressed as a the case of the Yy connection. Fig. 24 shows the test circuit ar-
percentage of input lightning impulse voltage amplitude (100% rangement for the measurement of transferred overvoltages.
corresponds to an amplitude of 300 V). The comparison be- The comparison between measured and calculated trans-
tween measured and calculated transferred overvoltages on the ferred overvoltages on the LV side (Figs. 2528) shows good
LV side (Figs. 1719) shows good agreement. Fig. 20 shows the agreement.
recorded chopped lightning impulse waveform applied on HV However, disagreements between measurements and calcu-
terminal A and its FFT approximation. Figs. 2123 confirm lation can be observed at lower frequencies in Figs. 19, 26, and
very good matching between the calculated and the measured 27. This is probably caused by high attenuation at lower fre-
transferred overvoltages. The similar analysis is performed in quencies ( less than 60 dB) as shown in Fig. 12.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

FILIPOVI-GRI et al.: HIGH-FREQUENCY MODEL OF THE POWER TRANSFORMER 7

Fig. 30. Measured magnitude between terminals A-a and a-N.

Fig. 27. Transferred overvoltage on terminal c-n when full lightning impulse
voltage is applied to terminal A.

Fig. 31. Measured phase angle between terminals A-a and a-N.

Fig. 28. Transferred overvoltage on terminal a-n when chopped lightning

impulse voltage is applied to terminal A.

Fig. 32. Transferred overvoltage on terminal a-N when full lightning impulse
voltage is applied to terminal A.

the measured magnitude and phase angle of the frequency

response.
Both full (1.13/54 s) and chopped ( s) lightning im-
pulse waveforms were applied on HV terminal A. The calcu-
lated transferred overvoltages on LV terminal a are expressed
as a percentage of input lightning impulse voltage amplitude
(100% corresponds to an amplitude of 300 V). The comparison
between measured and calculated transferred overvoltages on
the LV side (Figs. 32 and 33) shows good agreement.
Fig. 29. Test setup for the measurement of transferred overvoltages. The choice of frequency band parameters 1 kHz,
2 MHz, 1 kHz is based on preliminary FRA mea-
surements performed within the frequency band 100 Hz,
Additional experimental verification was carried out on a 6 MHz, 100 Hz. Analysis showed that approxi-
three-phase power transformer 96.7/10 kV, 36 MVA, YNd11. mated impulse voltages and calculated transferred overvoltages
Fig. 29 shows the test setup for the measurement of transferred change negligibly if and are increased to 1 kHz and
overvoltages. is decreased to 2 MHz. However, a more complex FRA
FRA measurements were performed within the frequency response will probably require lower , but it is unlikely that
band 1 kHz to 2 MHz, 1 kHz. Figs. 30 and 31 show the upper limit frequency will increase significantly.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8 IEEE TRANSACTIONS ON POWER DELIVERY

APPENDIX
The self and mutual inductance matrix of a three-phase
250-MVA, 400/231/22-kV autotransformer

(A1)

Matrix elements represent: a series winding which consists

of two windings connected in parallel ( and ), regulation
Fig. 33. Transferred overvoltage on terminal a-N when chopped lightning
impulse voltage is applied to terminal A.
winding , parallel winding , and tertiary winding .
Capacitance of the regulation winding to ground: 8.87 nF.

V. CONCLUSION REFERENCES
[1] Power Transformers Part 3. Insulation Levels, Dielectric Tests and
External Clearances in Air, IEC 60076-3, Int. Electrotech. Commis-
sion, Geneva, Switzerland, Mar. 2000, 2nd ed.
A new high-frequency black-box transformer model for [2] T. Ureda, T. Funabashi, T. Sugimoto, and L. Dub, A transformer
the calculation of lightning overvoltages transferred through model for transformer transfer voltage simulations, presented at the
IPST, Budapest, Hungary, Jun. 2024, 1999.
power transformers has been presented. The model is based [3] I. Uglei, Z. Maljkovi, and L. Kuterovac, Calculations of overvolt-
on a frequency-dependent voltage transfer function which is ages in the generator electrical circuit of a power station, presented at
determined from FRA measurements and the application of a the IPST, Budapest, Hungary, Jun. 2024, 1999.
[4] A. Morched, L. Marti, R. H. Brierly, and J. G. Lackey, Analysis of
four-terminal network theory. The transferred overvoltage is internal winding stresses in EHV generator step-up transformer fail-
calculated from the voltage transfer function and FFT approxi- ures, IEEE Trans. Power Del., vol. 11, no. 2, pp. 888894, Apr. 1996.
mation of input lightning impulse voltage. [5] M. Popov, L. van der Sluis, R. P. Smeets, and J. L. Roldan, Analysis of
very fast transients in layer-type transformer windings, IEEE Trans.
The computational verification of the presented model was Power Del., vol. 22, no. 1, pp. 238247, Jan. 2007.
successfully performed on a white-box model representing a [6] S. M. H. Hosseini, M. Vakilian, and G. B. Gharehpetian, Comparison
single phase of a three-phase 250-MVA autotransformer. The of transformer detailed models for fast and very fast transient studies,
IEEE Trans. Power Del., vol. 23, no. 2, pp. 733741, Apr. 2008.
results of the presented model and EMTP-RV simulations show [7] A. De, D. Debnath, and A. Chakrabarti, A study on the impact of
good agreement. It is shown that the accuracy of the model de- low-amplitude oscillatory switching transients on grid connected EHV
pends on the frequency scan step and upper limit frequency. The transformer windings in a longitudinal power supply system, IEEE
Trans. Power Del., vol. 24, no. 2, pp. 679686, Apr. 2009.
experimental verification of the presented model was carried out [8] R. C. Degeneff, W. J. McNutt, W. Neugebauer, J. Panek, M. E. Mc-
on 130-kVA and 36-MVA three-phase power transformers. The Callum, and C. C. Honey, Transformer response to system switching
comparison between the measured and calculated transferred voltage, IEEE Trans. Power App. Syst., vol. PAS-101, no. 6, pp.
14571470, Jun. 1982.
overvoltages has demonstrated that the proposed model accu- [9] R. C. Dugan, R. Gabrick, J. C. Wright, and K. V. Pattern, Validated
rately reproduces the high-frequency voltage transfer between techniques for modeling shell-form EHV transformers, IEEE Trans.
the windings when full or chopped lightning impulse voltage is Power Del., vol. 4, no. 2, pp. 10701078, Apr. 1989.
[10] P. G. Blanken, A lumped winding model for use in transformer models
applied. The main advantages and contributions of the presented for circuit simulation, IEEE Trans. Power Electron., vol. 16, no. 3, pp.
model are: 445460, May 2001.
transferred overvoltages can be quickly and accurately [11] A. Miki, T. Hosoya, and K. Okuyama, A calculation method for
impulse voltage distribution and transferred voltage in transformer
calculated; windings, IEEE Trans. Power App. Syst., vol. PAS-97, no. PAS-3,
its parameters are derived from FRA measurement results pp. 930939, May 1978.
obtained with commonly used equipment; [12] D. J. Wilcox and T. P. McHale, Modified theory of modal analysis for
the modeling of multiwinding transformers, Proc. Inst. Elect. Eng.,
it is stand alone and does not require EMTP-like tools; Transm. Distrib., vol. 139, no. 6, pp. 505512, Nov. 1992.
it could be of interest for power utilities when calcu- [13] K. Ragavan and L. Satish, An efficient method to compute transfer
lating lightning overvoltages transferred through power function of a transformer from its equivalent circuit, IEEE Trans.
Power Del., vol. 20, no. 2, pt. 1, pp. 780788, Apr. 2005.
transformers and performing insulation coordination [14] A. N. de Souza, M. G. Zago, O. R. Saavedra, C. C. Oba Ramos, and K.
studies, or to check the results of transferred overvoltage Ferraz, A computational tool to assist the analysis of the transformer
measurements; behavior related to lightning, Int. J. Elect. Power Energy Syst., vol.
33, no. 3, pp. 556561, Mar. 2011.
it could be used by transformer manufacturers to ad- [15] P. Mitra, A. De, and A. Chakrabarti, Resonant behavior of EHV
vise their clients about the level of surge-transferred transformer windings under system originated oscillatory transient
overvoltages. over voltages, Int. J. Elect. Power Energy Syst., vol. 33, no. 10, pp.
17601766, Dec. 2011.
The presented model is assumed to be linear since its exper- [16] M. H. Nazemi and G. B. Gharehpetian, Influence of mutual inductance
imental verification was carried out at low voltage. Further re- between HV and LV windings on transferred overvoltages, presented
search will focus on checking the model validity at HVs and at the XIVth ISH Conf., Beijing, China, Aug. 2529, 2005.
[17] M. Popov, L. van der Sluis, G. C. Paap, and H. de Herdt, Computa-
upgrading it in order to be applicable to time-domain simula- tion of very fast transient overvoltages in transformer windings, IEEE
tions. Trans. Power Del., vol. 18, no. 4, pp. 12681274, Oct. 2003.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

FILIPOVI-GRI et al.: HIGH-FREQUENCY MODEL OF THE POWER TRANSFORMER 9

[18] K. J. Cornick, B. Fillat, C. Kieny, and W. Mller, Distribution of very [33] High-Voltage Test Techniques Part 1: General Definitions and Test Re-
fast transient overvoltages in transformer windings, in CIGRE, Paris, quirements, IEC 60060-1, 2010, 3rd ed..
France, 1992.
[19] M. Popov, L. van der Sluis, and R. P. P. Smeets, Evaluation of surge
transferred overvoltages in distribution transformers, Elect. Power
Syst. Res., vol. 78, no. 3, pp. 441449, 2008.
[20] R. Dugan and S. D. Smith, Low-voltage-side current-surge phe- Dalibor Filipovi-Gri was born in Sinj, Croatia, in
nomena in single-phase distribution transformer systems, IEEE 1980. He received the Ph.D. degree in electrical en-
Trans. Power Del., vol. 3, no. 2, pp. 637647, Apr. 1988. gineering and computing from the University of Za-
[21] M. J. Manyahi and R. Thottappillil, Transfer of lightning transients greb, Zagreb, Croatia, in 2010.
through distribution transformers, in Proc. ICLP Conf., Cracow, Currently, he is the Head of the High-Voltage Lab-
Poland, Sep. 26, 2002, pp. 435440. oratory, Transformer Department,.Konar Electrical
[22] A. Borghetti, A. Morched, F. Napolitano, C. A. Nucci, and M. Engineering Institute, Zagreb. His areas of interest in-
Paolone, Lightning-induced overvoltages transferred through distri- clude high-voltage test and measuring techniques, in-
bution power transformers, IEEE Trans. Power Del., vol. 24, no. 1, strument and power transformers, and insulation sys-
pp. 360372, Jan. 2009. tems optimization.
[23] E. Bjerkan, High frequency modeling of power trans- Dr. Filipovi-Gri is a member of the technical
formersstresses and diagnostics, Ph.D. dissertation, Dept. Elect. committees TC E 38 Instrument Transformers and TC E 42 High Voltage Test
Power Eng., Norwegian Univ. Sci. Technol., Trondheim, Norway, Techniques.
2005.
[24] N. Y. Abed and O. A. Mohammed, Physics-based high-frequency
transformer modeling by finite elements, IEEE Trans. Magn., vol. 46,
no. 8, pp. 32493252, Aug. 2010.
[25] J. A. Martinez-Velasco, Power system transients: Parameter determi- Boidar Filipovi-Gri (S10M13) was born
nation, in Book. Boca Raton, FL, USA: CRC, 2010. in Sinj, Croatia, in 1983. He received the B.Sc.
[26] B. Gustavsen, Wide band modeling of power transformers, IEEE and Ph.D. degrees in electrical engineering and
Trans. Power Del., vol. 19, no. 1, pp. 414422, Jan. 2004. computing from the University of Zagreb, Zagreb,
[27] M. Tiberg, D. Bormann, B. Gustavsen, C. Heitz, G. Muset, J. Mah- Croatia, in 2007 and 2013, respectively.
seredjian, and P. Werle, Generic and automated simulation modeling Currently, he is with the Faculty of Electrical
based on measurements, presented at the IPST, Lyon, France, Jun. Engineering and Computing (Department of Energy
47, 2007. and Power Systems), University of Zagreb. He is the
[28] M. J. Manyahi and R. Thottappillil, Simplified model for estimation of Head of Quality in the High Voltage Laboratory. His
lightning induced transient transfer through distribution transformer, areas of interest include power system transients, in-
Int. J. Elect. Power Energy Syst., vol. 27, no. 4, pp. 241253, May 2005. sulation coordination, and high-voltage engineering.
[29] B. Gustavsen, Frequency-dependent modeling of power transformers Dr. Filipovi-Gri is a member of the CIGR Study Committees A3 High
with ungrounded windings, IEEE Trans. Power Del., vol. 19, no. 3, Voltage Equipment and C4 System Technical Performance.
pp. 13281334, Jul. 2004.
[30] B. Gustavsen and A. Semlyen, Application of vector fitting to state
equation representation of transformers for simulation of electromag-
netic transients, IEEE Trans. Power Del., vol. 13, no. 3, pp. 834842, Ivo Uglei (SM08) was born in Zagreb, Croatia, in
Jul. 1998. 1952. He received the Ph.D. degree in electrical en-
[31] N. A. Sabiha and M. Lehtonen, Lightning-induced overvoltages gineering and computing from the University of Za-
transmitted over distribution transformer with MV spark-gap opera- greb, Zagreb, Croatia, in 1988.
tionPart I: High-frequency transformer model, IEEE Trans. Power Currently, he is a Professor at the Faculty of Elec-
Del., vol. 25, no. 4, pp. 25652573, Oct. 2010. trical Engineering and Computing (Department of
[32] V. Behjat, A. Vahedi, A. Setayeshmehr, H. Borsi, and E. Gockenbach, Energy and Power Systems), University of Zagreb.
Sweep frequency response analysis for diagnosis of low level short He is the Head of the High-Voltage Laboratory at
circuit faults on the windings of power transformers: An experimental the Faculty of Electrical Engineering and Com-
study, Int.. J. Elect. Power Energy Syst., vol. 41, no. 1, pp. 7890, puting. His areas of interest include high-voltage
Nov. 2012. engineering and power transmission.