434

IEEE TRANSACTIONS ON MAGNETICS, VOL. 42, NO. 3, MARCH 2006

Investigation of Very Fast Transient Overvoltage

Distribution in Taper Winding of Tesla Transformer
Peng Ying and Ruan Jiangjun
School of Electrical Engineering, Wuhan University, Wuhan, Hubei 430072, China
In order to study the very fast transient overvoltage (VFTO) distribution in the taper winding of a tesla transformer under high-frequency steep-fronted voltage surge, we built a distributed line model based on multiconductor transmission line (MTL) theory. We used
a new hybrid algorithm combining finite-element-method (FEM) and interpolation formulas to quickly evaluate the induction coeffiby utilizing some characteristics of the taper structure. The turn-to-ground and interturn voltage distributions can be
cient matrix
obtained by solving the telegraphists equations in the frequency domain. We measured the voltage distribution inside the taper winding
to find some ways to weaken the voltage oscillations. Here, we compare the results with numerical values.
Index TermsMulticonductor transmission line (MTL), taper winding, tesla transformer, very fast transient overvoltage (VFTO).

I. INTRODUCTION

N a nanosecond periodically pulsed generator, the magnitude of output pulse of a tesla transformers secondary
winding is even higher than several hundred kilovolts. The
reflected wave travels back and injects the secondary winding
with a high amplitude, steep wave front, and high repeating
frequency, which comprises high-order harmonics that can
cause internal resonance and lead to an extremely uneven
distribution of turn-to-ground voltage and interturn voltage.
This often results in partial discharge and interturn insulation
failure in the winding. So it is necessary to study the very fast
transient overvoltage (VFTO) in a tesla transformers secondary
winding and find some ways to even the voltage distribution,
eliminate or weaken the internal oscillation, and improve the
transformers endurance. The VFTOs depend not only on the
incident waves style, but also on the structure of winding.
Most of the related studies emphasize the wave propagation
in windings of ac machines and power transformers. Seldom
are done either on the simulation or experiment of transient
behavior of a tesla transformers taper winding. Mainly there
are three methods to simulate such a problem.
1) Lumped Circuit Model: When the windings length is far
less than the wavelength, each coil can be represented with a
, shunt capacilumped element, including self-inductance
tance
, resistance
, mutual inductance
, and series
capacitance
, whereas the mutual resistance and conductance are always neglected. Electromagnetic transients program
(EMTP) or frequency-domain method can achieve the exact response of the whole network after linking all these elements.
Initial works [1][5] showed its successful application in predicting VFTO or partial discharge in windings of power transformer and ac machines.
2) Multiconductor Transmission Line (MTL) Model: As
model 1) does not consider the waves traveling in each coil,
it is not accurate enough while the pulse has a shorter rise
or fall time or the coils are long enough to compare with the
wavelength. In the situation, each coil should be taken as a

Digital Object Identifier 10.1109/TMAG.2005.862759

distributed line. The voltage distribution along each coil can

be calculated by solving the telegraphists equations with
given boundary conditions. The most recent research [6][8]
on the waves propagation in windings used the MTL model
combined with the single transmission line (STL) model. In
particular, works by Petrarca [6], [8], Guardado [9], [10],
McLaren [11], [12], Shibuya [13], [14], and Cornick [15] were
of great significance in this field.
3) Full-Wave Solution: Although both 1) and 2) simplify the
field problem to a circuit problem, the measured and numerical values show the simplification is reasonable and sometimes accurate enough. Nonetheless, the direct field solution
is more attractive but hard to obtain due to the complexity of
model and lots of degrees of freedom. Miri [16][19] performed
transient electric field analysis in racetrack-shaped toroidal field
model for the International Thermonuclear Experimental Reactor using the finite-element method (FEM), but the process
was not given in detail, such as the governing equations, degrees
of freedom, boundary conditions, and so on.
Other related topics can also be found. For example, Yamashita [20] carried out a field-circuit method to calculate
transient electric field in transformer winding. A two-dimensional (2-D) axisymmetrical electric field finite-element (FE)
analysis is performed, applied with Dirichlet boundary values
obtained from an advanced lumped circuit network calculation.
Oyegoke [21] compared three approaches for predicting the
magnitude and distribution of the transient voltages in the stator
winding of an electric machine subjected to a steep-fronted
surge. For reference, [22][26] may give more suggestions on
similar problems, which will not be expatiated here.
The electrical parameters evaluation is rather important for
an equivalent circuit model. It greatly determines the accuracy
and efficiency of numerical models, such as the unit lengths
inductance matrix , induction coefficient matrix , and resistance matrix of the MTL model. (Neglecting conductance
matrix affects little to the results [5], [9].) Analytical formulas
or field calculations (based on electromagnetic energy) are normally used to extract these parameters [27][31]. The analytical
method is very limited for the formulas are from experience or
related to special windings with a simple geometry. The field
method may be more applicable for complex, various winding

0018-9464/$20.00 2006 IEEE

YING AND JIANGJUN: VFTO DISTRIBUTION IN TAPER WINDING OF TESLA TRANSFORMER

geometries, but for a large -conductor system, it is considerably time consuming.

One difficulty still exists, forcing us to make an important
assumption. The iron cores magnetic characteristic is hard to
describe under different high frequency. By measurements or
simulation, most of the authors [3], [5], [8], [9], [11][14], [21]
thought that the iron core behaves like a flux barrier at frequency of interest (110 MHz higher). Two reasons may explain
this phenomenon. One is the skin effect (skin depth less than
1 10 m) that causes the component of magnetic flux penetrating into the core is too small. The second is the transmission
line effect results that the current in adjacent turns may flow in
opposite direction which counteracts the total flux. Therefore,
the main flux loops not in the iron core, but in the air. The inductance coupling between coils is no more nonlinear when eliminating the effect of iron core, thus the winding system is a linear
system allowing the superposition theorem.
Also, voltage or electric field measurement is a hard task
that should be done to validate the computer models accuracy. Malewski [32] had developed a method consisting of
the time-domain measurement of a transient voltage between
the adjacent discs and disc to ground. High resolution, a fast
digital recorder, and an optically isolated probe were used in
measurement. Fujita [33] used an optical fiber transmission
system (Sony-Tektronix A6904S) to measure interturn voltages
directly. The results clearly showed a traveling wave generated
at coils end and propagating along the winding.
Considering all the above factors and the speciality of our
problemsteep wave front (ns), high repeating frequency, and
taper structure of winding, the lumped equivalent circuit will not
be accurate enough. If we divide the taper winding into several
coils and combine STL with MTL as [6], [13] did, without
taking into account the mutual inductances between coils, an
improper discontinuity of voltage distribution will occur at the
connection point of adjacent coils. So a MTL model of the tesla
transformers taper winding is built to properly model actual
electromagnetic coupling among turns. Direct FE analysis to
extract an arbitrary -conductor systems needs
steps of energy solution, which is unacceptable if is large. Fortunately, by analyzing of the tesla transformers sample taper
winding, we find its good characteristics (detailed in Section III)
that help us to build some analytical formulas depending on
s several elements obtained from a few steps of advanced
FE analysis. The quantitative level of energy solution steps is
greatly reduced from
to . Based on the flux barrier assumption, an equivalent relative permeability
(1 10 ) is
used so as to evaluate in static magnetic field calculation instead of high-frequency analysis. By fast Fourier transformation
(FFT), the pulse is represented by the sum of harmonics. So the
telegraphists equations are solved utilizing the modal analysis
in frequency domain and at last the results will be transferred
into the time domain, plotting the turn-to-ground and interturn
voltage distribution. Besides the simulation, turns voltage with
respect to ground are measured under several situations. The results show the severe voltage oscillations in windings and the
most severe unevenness of voltage distribution among coils appear mostly at head section of the winding whereas in the middle
and grounding section, the voltage oscillation seems to be well
constrained.

435

Fig. 1. Typical configuration of a tesla transformer. Its secondary winding is

wrapped on a taper-shaped framework.

Fig. 2.

MTL model of the taper winding.

II. MTL MODEL OF THE TAPER WINDING

A. Structure of a Tesla Transformer
The tesla transformer, capable of transmitting high-voltage
pulses, is embedded in numerous devices in the fields of radio,
television, telecommunications, medicine, high-voltage periodically pulsed generators, etc. It consists of five main parts: inner
iron core, outer iron, chamber, primary winding, and secondary
winding (taper-shaped). The chamber is filled with transformer
oil to ensure good insulation of winding. One end of the secondary winding is connected with the inner iron core while the
other end is connected to the ground. Fig. 1 shows a typical
structure of the 2-D axial section.
B. MTL Model
An -conductor uniform lossy transmission line system is
constructed by splitting the winding along its axial plane and
spreading turns into parallel lines in which the chamber wall
is treated as the ground plane. This modeling assumes that the
average perimeter of coils is far longer than the windings axial
length. Thus, by an initial capacitive coupling, the voltage distribution will be established instantaneously across the whole
winding. The lengths of adjacent lines are almost the same because of the windings small taper angle (6 12 ). As for any
two distant lines, though they are quite different in length, their
electromagnetic coupling is weak enough to be neglected, compared with two adjacent ones (illustrated in Section III). So the
taper winding in this case can be treated as an equal-length MTL
model shown in Fig. 2.
In Fig. 2,
, and are the head and tail ends voltages and currents of the th line, respectively. Arrows indicate
the positive direction of currents.
When every turn in the winding is represented as a transmission line, then the propagation phenomena in transformer

436

IEEE TRANSACTIONS ON MAGNETICS, VOL. 42, NO. 3, MARCH 2006

winding can be fully described by making use of the telegraphists equations in time domain
(1)
(2)
are the voltage and
In (1) and (2), respectively,
current vectors
.
and are square matrices
of the resistances, inductances, conductances, and induction coefficients in MTL model, whereas will be neglected in this
paper.

second feature enables that we need to fill a few elements in

only, which greatly reduce s half bandwidth and changes it
to a banded matrix. As
will not be used
in later calculation, in the example, we need only evaluate the
and it also presents a
linear increasing in middle part shown in Fig. 3(b). Depending
and
s features, a rapid accurate method to evaluate
on
is achieved by linear interpolation.
Build the actual FE model of the taper winding as in
; calculate
Fig. 3(a); calculate
of those turns that do not follow the feature mentioned
above, especially the turns at the head and tail end, e.g.,

III. ELECTRICAL PARAMETERS EVALUATION

A. Induction Coefficient Matrix
Based on the electromagnetic theory, total static electric enof an -conductor system in region can be given by
ergy

. There are only

FE solution steps in total and the middle turns
can
be interpolated by

(3)
(6)
where
are electric field intensity and electric flux density,
are the voltage to ground and charges of the th conductor.
is also written as

(7)

(4)
where
is a self-induction coefficient and
induction coefficient,
, and
.
Substituting in (3) with (4) gives

(8)

is a mutual

(5)

As the windings special structure, analytical formulas are

hard to list. So a normal procedure to evaluate by field method
is as follows.
First, apply a unit potential on the th conductor,

while the others are set to zero potential. Due to an approximate axisymmetrical structure of the model, a 2-D static
electric field FE analysis can calculate
. Then by (5),
the diagonal elements of are obtained.
Apply unit potentials on the th and th conductor,

while the other

are set to zero potential. As in step 1,
after the FE analysis, other elements of are obtained as
well by (5). (
is from 1 to
from
to ).
FE soFollowing the procedures above requires
lution steps to get , which is almost impossible for teslas
taper winding
. Some alternative methods should
be carried out. By analyzing
of a typical sample winding
[
, taper angle 15 , Fig. 3(a)], we can find some interesting characteristics that will help us overcome the difficulty. As shown in Fig. 3(b), first, except the end turns, the
of middle turns increase linearly; second, except the iron
core (conductor no. 1), the turns
decrease rapidly with the
distance between th and th conductor increasing, which may
be explained as the shielding effect of their interval turns. The

The modified method produces results that correspond well

to the direct method shown in Fig. 3(b). The formulas may be
different for other structures, depending on the taper angle and
linear section.
Attention should be paid that the framework of winding
cannot be neglected in FE analysis (relative permittivity
). The postprocess for
includes two simplifications.
First is the equivalent of iron cores capacitive effect on the
turns by multiplying
with a coefficient
whose value
should be adjusted according to the measurements. Second, as
the procedure above obtains coils
of the three-dimensional
model, it should be transferred into unit length value with
dividing the th turns perimeter and
dividing average
perimeter of the th, th turns.
B. Inductance Matrix
Due to the existence of the framework and iron core, the relationship between and does not match
(9)
where
is the velocity of wave propagation in uniform
medium.
So elements in are evaluated by the following definitions:
(10)
where
are the self-inductance and current of the th coil,
is the mutual inductance between the th and th coils,

YING AND JIANGJUN: VFTO DISTRIBUTION IN TAPER WINDING OF TESLA TRANSFORMER

Fig. 3.

K evaluation (a) Mesh view. (b) Comparison between two methods.

437

438

IEEE TRANSACTIONS ON MAGNETICS, VOL. 42, NO. 3, MARCH 2006

are phasor voltage and current vectors. The waves

where
angular frequency is , so the complex impedance and admittance matrices of the line, respectively, are
(14)
The common solutions of (12) and (13) have the forms
(15)
(16)
are incident and reflected voltage vector, and
where
is the characteristic admittance matrix. Based on modal analysis, it gives
(17)
The voltage and current vectors at head
ends satisfy [9]

and tail

(18)
Fig. 4. Typical magnetic flux.

(19)

represents the self flux linkages of th coil, and

is the
flux linkages of th coil produced by .
The exact behavior of iron core at high frequency is impossible to describe. So the shielding effect of iron core and conductors to the flux at the interested frequencies is equivalent by
setting a small relative permeability
when
performing a static magnetic field FE analysis. Its value is adjusted with reference to measurements. As shown in Fig. 4, the
chamber wall, iron core, and conductors together form flux
barriers to limit the flux path similar to high-frequency phenomenon, which results in the rapid decrease of mutual inductive coupling between turns. A static FE solution will obtain the
magnetic flux density by applying unit current density in the
th coil. Then by surface integrations of for each coil, along
with (10),
are evaluated. should also be transferred
into unit length value the same as .

(20)
where

where is an identity matrix.

According with the connections of turns and grounding
impedance , the boundary conditions can be written as

After applying these conditions in (20) and some technique

[9], it is possible to arrange (20) to yield

C. Resistance Matrix
As the turns are thin enough (cross section radius 0.2 mm),
the skin effect at high frequencies is approximately equivalent
by multiplying with a coefficient

(21)

(11)

is calculated with
and
is known as the
where
exciting source. Then
can be obtained. The head voltage of
each turn is eventually achieved by means of a single multiplication of
and the first column of inverse .
So first by decomposing the inject pulse into time harmonics
of different frequencies, with the solution method, we can obtain
the winding systems voltage distribution in frequency domain.
Then by superposition of Fourier series, the voltage distribution
in time domain can be obtained.

where
is the unit length resistance of the th coil, is the
conductivity of turns, and is the turns cross section area.
Without mutual resistance,
is a diagonal matrix. As mentioned before, the dielectric loss will not be taken into account.
IV. SOLUTION
Because of the symmetry of
forms can be written as

(1) and (2)s phasor

V. MEASUREMENTS
(12)

A. Model Description

(13)

The iron core is wrapped into a cylinder by laminated silicon

sheet that is hollow inside to be filled with transformer oil. Its

YING AND JIANGJUN: VFTO DISTRIBUTION IN TAPER WINDING OF TESLA TRANSFORMER

439

TABLE II
MEASURED RESULTS OF PRIME CONCERN

Fig. 5. Models description. (a) Iron cores dimension. (b) Sketch of the
models structure.
TABLE I
MODELS SIZES

size and the sketch of model for testing are shown in Fig. 5. The
winding will not be seen in the sketch.
The pulse generator can produce various types of pulse,
which inject from the connection point of windings head and
iron core. It is hard to set measuring points for every turn
because of close winding. And the spatial voltage distribution
is believed not to change suddenly within every ten turns. So
the probe of Tektronix TDS220 scope (bandwidth 100 MHz,
sampling rate 1 GS/s) is kept well connected with the measuring
points set every ten turns, where the insulation is well removed.
Table I lists all the testing combinations and Table II lists the
results of prime concern.
B. Analysis
The measured values clearly show the voltages at head
end of winding fluctuate more sharply than those at the
middle or ground section. The positions where the maximum
turn-to-ground voltage
or interturn voltage
appears will change with different taper angles and pulse types.
However, most of these positions are concentrated at the be-

ginning section of winding. For example, the worst case that

the maximum voltage to ground reaches 2.32 times of peak
source magnitude
happens at turn 120 of model 3 under
100 kHz sinusoidal wave. The heaviest interturn (between
two measuring points, ten-turn interval) electric stress appears
between turn 1 and 10 of model 1 under 1 MHz pulse exciting,
with a magnitude 1.8 times of
.
By comparison, the existence of iron core can limit the inside voltage oscillation to a great extent. The models structure
determines a natural resonant frequency . When the main frequency of inject wave differs much with , the internal resonance will not happen, such as 100 kHz sinusoidal waves case.
If is close to , like cases of 1 MHz sinusoidal or 100 kHz
pulse (duty cycle 0.5, rise time 20 ns) that consist of abundant high-order harmonics, severe oscillations happen in most
models.
An attempt has been carried out to lower the magnitude of
peak interturn voltage. A tapered aluminum sheet connected to
the ground is placed adhering to the inner wall of the framework, covering about 20 turns of model 3 where the maximum
interturn voltage appears under 100 kHz pulse exciting. The results show a decrease of
ratio from 2.32 to 2.29,
ratio from 1.38 to 1.04.

440

IEEE TRANSACTIONS ON MAGNETICS, VOL. 42, NO. 3, MARCH 2006

Fig. 6. Comparison between numerical and measured values. Voltage to ground of turn 10; 20; . . . ; 90.

A comparison between numerical and measured voltage

waveforms of model 1, 100 kHz pulse
is shown in Fig. 6, from which we can see
the oscillations occur at the two slope edges of the pulse. Although the waveforms of several turns have good agreements,
some errors still exist especially at the edges of pulse, which
are probably caused by three factors listed below.
The inaccuracy of and are due to the idealization of actual model: the effect of iron core on each turn is not the same,
so it is not accurate to model it by a single
. Furthermore, the
iron cores magnetic characteristic under high frequency may
not be so simple just using single . It could be nonlinear with
or frequency. The eddy current in the core may be another
reason. Limited by rough equipment, the measurement work
could easily be affected by EMI signals.
VI. CONCLUSION
The study of VFTO in a tesla transformers secondary
winding, which occurs during the wave traveling back that
causes steep-fronted surges, is important for insulation coordination. It is especially important to find the position that

maximum interturn voltage happens, which depends on the

configuration of the winding and wave style. In this paper,
the MTL model is used to study VFTO in the taper winding,
whose may be simplified as a banded matrix and its middle
section is linear to be evaluated by analytical formulas more
rapidly without loss of accuracy. An equivalent
to simulate
conductors flux barrier feature makes be evaluated with
static magnetic FE analysis. The electromagnetic coupling
between turns decreases quickly with their distance increasing
in this kind of structure. With the frequency-domain method,
the equations of MTL applied with proper boundary conditions
could be easily solved for obtaining the voltage distribution
across the winding. A simple platform for measurement is built
to study VFTO in windings under various conditions and is
helpful for finding effective ways to improve the uneven voltage
distribution. The results show severe voltage oscillations inside
the windings, especially under situations that the surges main
frequency
is close to the resonant frequency . So for
steep-fronted surges, the fluctuation occurs more easily and
the peak value of interturn voltage may reach several times of
the sources magnitude, which is rather dangerous to interturn
insulation. The numerical voltage waveforms with respect to

YING AND JIANGJUN: VFTO DISTRIBUTION IN TAPER WINDING OF TESLA TRANSFORMER

the time have good agreements but also show some deviation
from the measured ones. It indicates that some simplifications
and , but
of computer model may be reasonable, such as
not accurate enough. So more accurate methods for simulation
and measurement work are in our future plan.
REFERENCES
[1] W. Zanji, Modeling for simulating very fast transient in transformer
windings, in Proc. CSEE, vol. 16, China, Sep. 1996, pp. 299305.
[2] Y. Xuechang, W. Yongping, Q. Qingcheng, and S. Xinfu, Calculations
of transient voltage distribution and field strength in transformer windings taken account of losses, Transformer, vol. 35, pp. 710, May 1998.
[3] Z. D. Wang, P. A. Crossley, and K. J. Comick, A simulation model
for propagation of partial discharge pulses in transformers, in Proc. Int.
Conf. Power System Technology, vol. 1, Beijing, China, Aug. 1998, pp.
151155.
[4] A. Purkait, R. S. Barve, N. K. Deshmukh, and L. M. Sarda, Prediction
of impulse distribution in double section coils of stator windings of high
voltage motors, in Eleventh Int. Symp. High Voltage Engineering, vol.
1, Aug. 1999, pp. 299302.
[5] A. Narang, B. K. Gupta, E. P. Dick, and D. K. Sharma, Measurement
and analysis of surge distribution in motor stator windings, IEEE Trans.
Energy Convers., vol. 4, no. 1, pp. 126134, Mar. 1989.
[6] G. Lupo, C. Petrarca, M. Vitelli, and V. Tucci, Multiconductor transmission line analysis of steep-front surges in machine windings, IEEE
Trans. Dielectr. Electr. Insul., vol. 9, no. 3, pp. 467478, Jun. 2002.
[7] M. Popov, L. van der Sluis, G. C. Paap, and H. De Herdt, Computation
of very fast transient overvoltages in transformer windings, IEEE Trans.
Power Del., vol. 18, no. 4, pp. 12681274, Oct. 2003.
[8] C. Petrarca, A. Maffucci, V. Tucci, and M. Vitelli, Analysis of the
voltage distribution in a motor stator winding subjected to steep-fronted
surge voltages by means of a multiconductor lossy transmission line
model, IEEE Trans. Energy Convers., vol. 19, no. 1, pp. 717, Mar.
2004.
[9] J. L. Guardado and K. J. Cornick, A computer model for calculating
steep-fronted surge distribution in machine windings, IEEE Trans. Energy Convers., vol. 4, no. 1, pp. 95101, Mar. 1989.
[10] J. L. Guardado, V. Carrillo, and K. J. Cornick, Calculation of interturn
voltages in machine windings during switching transient measured on
terminals, IEEE Trans. Energy Convers., vol. 10, no. 1, pp. 8794, Mar.
1995.
[11] P. G. McLaren and M. H. Abdel-Rahman, Steep fronted surges applied
to large ac motors-effect of surge capacitor value and lead length, IEEE
Trans. Power Del., vol. 3, pp. 990997, Jul. 1988.
, Modeling of large ac motor coils for steep-fronted surge studies,
[12]
IEEE Trans. Ind. Appl., vol. 24, no. 3, pp. 422426, MayJun. 1988.
[13] Y. Shibuya, S. Fujita, and N. Hosokawa, Analysis of very fast transient
overvoltage in transformer winding, IEE Proc. Gener. Transm. Distrib.,
vol. 144, pp. 461468, Sep. 1997.
[14] Y. Shibuya, S. Fujita, and E. Tamaki, Analysis of very fast transients in
transformers, IEE Proc. Gener. Transm. Distrib., vol. 148, pp. 377383,
Sep. 2001.
[15] K. J. Cornick, J. L. Guardado, and P. Bekaert, Interturn voltages in machine windings evaluated from on-site test results and computer simulation, IEE Proc. B., vol. 139, pp. 238244, May 1992.
[16] A. M. Miri, N. A. Riegel, A. Ulbricht, and S. Fink, Finite element model
for the investigation of the transient oscillations in the ITER toroidal field
(TF) model coil, in Eleventh Int. Symp. High Voltage Engineering, vol.
2, U.K., Aug. 1999, pp. 4346.
[17] A. M. Miri, N. A. Riegel, and D. L. Lindenmeyer, Finite element and
network models for the computation of the transient potential and field
distribution in the winding system of high voltage power transformers,
in Proc. 6th Int. Conf. Optimization of Electrical and Electronic Equipments, vol. 1, Brasov, Romania, May 1998, pp. 9398.
[18] A. M. Miri, N. A. Riegel, and A. Kuhner, Finite element models for
the computation of the transient potential and field distribution in the
winding system of high voltage power transformers, in Eleventh Int.
Symp. High Voltage Engineering, vol. 2, U.K., Aug. 1999, pp. 3942.
[19] A. M. Miri, C. Meinecke, N. A. Riegel, A. Ulbricht, and S. Fink, Influence of the radial plates on the transient oscillations in the ITER toroidal
field (TF) model coil using a finite element approach, IEEE Trans. Appl.
Supercond., vol. 10, no. 1, pp. 596599, Mar. 2000.

441

[20] H. Yamashita, V. Cingoski, E. Nakamae, A. Namera, and H. Kitamura,

Design improvements on graded insulation of power transformers using
transient electric field analysis and visualization technique, IEEE Trans.
Energy Convers., vol. 14, no. 4, pp. 13791384, Dec. 1999.
[21] B. S. Oyegoke, A comparative analysis of methods for calculating the
transient voltage distribution within the stator winding of an electric machine subjected to steep-fronted surge, in Proc. Eighth Int. Conf. Electrical Machines and Drives, Sep. 1997, pp. 294298.
[22] J. Zhang et al., The design of a compact pulse transformer, in 12th
IEEE Int. Pulsed Power Conf., Dig. Tech. Papers, vol. 2, Jun. 1999, pp.
704707.
[23] V. Woivre, J. P. Arthaud, A. Ahmad, and N. Burais, Transient overvoltage study and model for shell-type power transformers, IEEE Trans.
Power Del., vol. 8, no. 1, pp. 212222, Jan. 1993.
[24] S. Islam and G. Ledwich, An equivalent circuit for calculation of interturn voltage distribution of stator windings in the presence of slot discharges, in Proc. 5th Int. Conf. Properties and Applications of Dielectric Materials, vol. 1, May 1997, pp. 507510.
[25] O. Honorati and E. Santini, New approach to the analysis of impulse
voltage distribution in transformer windings, IEE Proc., vol. 137, pp.
283290, Jul. 1990.
[26] Y. Tang, Analysis of steep-fronted voltage distribution and turn insulation failure in invertor-fed form-wound AC motor, IEEE Trans. Ind.
Appl., vol. 34, no. 5, pp. 10881096, Sep.Oct. 1998.
[27] L. M. Popovic, New method for calculation of series capacitance for
transient analysis of windings, in Proc. 9th Mediterranean Electrotechnical Conf., vol. 2, May 1998, pp. 10421046.
[28] V. Venegas, R. Escarela, R. Mota, E. Melgoza, and J. L. Guardado,
Calculation of electrical parameters for transient overvoltage studies on
electrical machines, in IEEE Int. Electric Machines and Drives Conf.,
vol. 3, Jun. 2003, pp. 19781982.
[29] P. Poulichet, F. Costa, and E. Laboure, High-frequency modeling of a
current transformer by finite-element simulation, IEEE Trans. Magn.,
vol. 39, no. 2, pp. 9981007, Mar. 2003.
[30] K. B. Tatis, A. G. Kladas, and J. A. Tegopoulos, Harmonic iron loss
determination in laminated iron cores by using a particular 3-D finiteelement model, IEEE Trans. Magn., vol. 40, no. 2, pp. 860863, Mar.
2004.
[31] O. Moreau, R. Michel, T. Chevalier, G. Meunier, M. Joan, and J. B. Delcroix, 3-D high frequency computation of transformer R; L parameters, IEEE Trans. Magn., vol. 41, no. 5, pp. 13641367, May 2005.
[32] R. Malewski, M. A. Franchek, and J. H. McWhirter, Experimental validation of a computer model simulating an impulse voltage distribution
in HV transformer windings, IEEE Trans. Power Del., vol. 9, no. 4, pp.
17891798, Oct. 1994.
[33] S. Fujita, N. Hosokawa, and Y. Shibuya, Experimental investigation
of high frequency voltage oscillation in transformer windings, IEEE
Trans. Power Del., vol. 13, no. 4, pp. 12011207, Oct. 1998.

Manuscript received September 9, 2005; revised November 22, 2005 (e-mail:

hoverpain@hotmail.com, ruan308@126.com).

Peng Ying was born in Wuhan, Hubei Province, China, on December 20, 1980.
He received the bachelors degree in electrical engineering from Huazhong University of Science and Technology in 2002. Currently, he is working toward the
Ph.D. degree in numerical analysis of electromagnetic field in the Institute of
Advanced Technology of Electrical Engineering, Wuhan University.
His major fields of interest are numerical method of electromagnetic field
calculation in engineering.

Ruan Jiangjun was born in Zhejiang, China, on June 25, 1968. He received the
B.S. and Ph.D. degrees in electric machine engineering from Huazhong University of Science & Technology (HUST) in 1990 and 1995, respectively, and
finished his post-doctoral research in 1998 from Wuhan University of Hydraulic
& Electric Engineering.
He is currently Professor of Wuhan University, Wuhan, Hubei Province,
China. His research interests include electromagnetic field numerical simulation, electromagnetic compatibility, and high-voltage engineering.